nmat1099PRINT 3/11/04 11:32 AM Page 221

LETTERS

Anomalous Hall effect governed by electron

doping in a room-temperature transparent
ferromagnetic semiconductor
HIDEMI TOYOSAKI1, TOMOTERU FUKUMURA*1, YASUHIRO YAMADA1, KIYOMI NAKAJIMA2,3,
TOYOHIRO CHIKYOW2,3, TETSUYA HASEGAWA3,4, HIDEOMI KOINUMA3,5 AND MASASHI KAWASAKI1,3
1
Institute for Materials Research,Tohoku University, Sendai 980-8577, Japan
2
National Institute for Materials Science,Tsukuba 305-0044, Japan
3
Combinatorial Materials Exploration and Technology,Tsukuba 305-0044, Japan
4
Department of Chemistry, University of Tokyo,Tokyo 113-0033, Japan
5
Materials and Structures Laboratory,Tokyo Institute of Technology,Yokohama 226-8503, Japan
*e-mail: fukumura@imr.tohoku.ac.jp

Published online: 21 March 2004; doi:10.1038/nmat1099

F
erromagnetic semiconductors are believed to be suitable for a Energy b Energy
future spintronics, because both charge and spin degrees of
freedom1,2 can be manipulated by external stimuli. One of the Down spin Up spin Down spin Up spin
most important characteristics of ferromagnetic semiconductors is the
anomalous Hall effect. This is because the ferromagnetically spin-
polarized carrier can be probed and controlled electrically, leading to EF EF
direct application for electronics. Control of the Curie temperature3
and magnetization direction4 by electronic field, and photo-induced
ferromagnetism5 have been performed successfully using the
anomalous Hall effect for group III-V ferromagnetic semiconductors.
In these cases, the operation temperature was much below room
temperature because of the limited Curie temperature of less than
Ferromagnetic Spin-orbit
160K (ref.6).Here,we report on the anomalous Hall effect governed by segregation interaction
electron doping in a room-temperature transparent ferromagnetic
semiconductor,rutile Ti1–xCoxO2–δ (of oxygen deficiency δ).This result Solid-soluted
magnetic ion
manifests the intrinsic nature of ferromagnetism in this compound, H
and represents the possible realization of transparent semiconductor Ti site
spintronics devices operable at room temperature. E
Electron path
Room-temperature ferromagnetic semiconductors are desired for
application to practical devices. Among various types of Electron
semiconductors, oxide semiconductors are one of the most promising
host compounds to realize a higher Curie temperature (TC).
Indeed, anatase and rutile phases of TiO2 doped with Co were found to
be ferromagnetic at room temperature7,8, and p-type ZnO doped with Figure 1 Schematics for extrinsic and intrinsic ferromagnetic sources in
Mn was theoretically predicted to be ferromagnetic above room- semiconductors doped with magnetic impurities. a,Ferromagnetic magnetization is
temperature9. Various oxide semiconductors have been reported to be caused by segregation of the ferromagnetic secondary phase (red arrows),regardless of
ferromagnetic10,11. However, the possibility of an extrinsic origin for the the non-ferromagnetic ordering of solid-soluted magnetic impurities (green arrows) and
ferromagnetism,such as ferromagnetic impurity segregation,could not the absence of the AHE (blue lines) due to no spin polarization of charge carriers (band
be ruled out in most cases. This is because the often-performed diagram in upper panel of a).b,The ferromagnetic ordering is caused by exchange
magnetization measurements using a magnetometer are unable to interaction between solid-soluted magnetic impurities (green arrows) and charge carriers
distinguish an intrinsic ferromagnetic signal from an extrinsic one, due (blue arrows).The spin-orbit interaction between ferromagnetically ordered impurity ions
to the characteristics of the method to probe all the magnetization from and charge carriers having ferromagnetic spin polarization (band diagram in upper panel of
both ferromagnetic sources (Fig. 1a). The intrinsic ferromagnetism is, b) generates an AHE (blue curves).

nature materials | VOL 3 | APRIL 2004 | www.nature.com/naturematerials 221

©2004 Nature Publishing Group

nmat1099PRINT 3/11/04 11:32 AM Page 222

LETTERS
a 102 Figure 2 Electric-transport properties for Ti1–xCoxO2–δ. a,Temperature dependence of
resistivity (ρxx) for Ti0.97Co0.03O2–δ films grown under different PO2.b,PO2 dependence of
conductivity (σxx) at 300 K and lattice constant along (101) direction (d (101)) for these
PO2 = 1.0 × 10–4 torr
101 films.c,Magnetoresistance normalized by ρxx in zero magnetic field (H ) at different
temperatures for the same films as in a.The applied magnetic field is perpendicular to the
film plane.
100
10–5 torr
ρxx (Ωcm)

10–6 torr
10–1 current and external magnetic field proportional to the magnetization,
that is attributed to asymmetric carrier scattering by magnetic
impurities in the presence of spin-orbit interaction (Fig. 1b).
10–2 10–7 torr
Hall resistivity (ρxy) in ferromagnets is generally expressed as
x = 0.03
ρxy = RoB + Rsµ0M (B: magnetic induction, µ0: magnetic permeability,
10–8 torr
M: magnetization, Ro: ordinary Hall coefficient, Rs: anomalous Hall
10–3
50 100 150 200 250 300 coefficient)12.The first term denotes the ordinary Hall effect (OHE) and
Temperature (K) the second term denotes AHE. AHE controlled by charge carriers in a
room-temperature ferromagnetic semiconductor would be strong
104 evidence of intrinsic ferromagnetism and also an important milestone
b towards the realization of semiconductor spintronics devices operable
300 K
0.254 at room temperature.
Ti1–xCoxO2–δ epitaxial thin films were grown by laser molecular-
102 beam epitaxy (see Methods). The electron concentration (n) was
0.252 controlled by varying the partial oxygen pressure (PO ) during growth.2

Atomically controlled film growth has been achieved, as confirmed by

σxx ( Ω –1 cm–1)

the intensity oscillation of reflection high-energy electron diffraction

d (101) (nm)

100
0.250 (RHEED),and the atomically flat surface morphology,which is suitable
for the fabrication of multilayer devices (Supplementary Information,
Fig. S1).Figure 2a shows the temperature dependence of resistivity (ρxx)
10–2
0.248
for Ti0.97Co0.03O2–δ films grown under various PO . By reducing PO , ρxx at
2 2

300 K can be controlled from 101 Ω cm to 10–3 Ω cm. Figure 2b shows

conductivity (σxx) at 300 K and the lattice constant along (101), d(101),
x = 0.03
as a function of PO . The gradual increase of the lattice constant with
2
10–4 0.246
10–8 10–7 10–6 10–5 10–4 decreasing PO represents a systematic increase of δ, resulting in
2

PO2 (torr) systematic doping of electrons to increase σxx.The value of ndetermined

from OHE ranges from 1018 cm–3 to 1022 cm–3 with a rather constant
mobility of the order of 10–1 cm2 V–1 s–1 (it is noted that higher n may
c 10–5 torr
contain the larger error because ρAHE is dominant over ρOHE in the ρxy
1.000 300 K
versus H curve). Figure 2c shows magnetoresistance normalized by ρxx
H 200 K in zero magnetic field at different temperatures for the same films as in
0.996
Fig. 2a. For each film, the magnetoresistance is negative and decreases
10–6 torr monotonically with increasing magnetic field and decreasing
1.000 300 K
temperature. No significant difference is observed in the behaviour of
200 K
magnetoresistance in these films.
ρxx (H )/ρxx (0 )

0.996 100 K The behaviour of ρxy strongly depends on n and Co content (x) as
1.000 10–7 torr 300 K described below. First, we describe the n dependence of ρxy. Figure 3a,b
200 K
shows magnetic-field dependence of ρxy at various temperatures for
0.996 Ti0.97Co0.03O2–δ films grown under PO = 10–7 torr and 10–6 torr,
2
100 K
respectively. As seen in Fig. 3a, ρxy increases rapidly up to 0.2 T with
10–8 torr 300 K increasing magnetic field, and gradually decreases linearly with further
1.000
200 K increasing magnetic field. This behaviour shows that the anomalous
0.996 x = 0.03 100 K part of ρxy (ρAHE), which is proportional to magnetization, is dominant
for lower magnetic field, whereas the ordinary part of ρxy (ρOHE), which
–8 –6 –4 –2 0 2 4 6 8
µ oH (T) is proportional to the inverse of nyielding the negative linear slope of ρxy,
overcomes ρAHE for higher magnetic field,as the magnetization saturates
and ρOHE keeps increasing linearly with increasing magnetic field.
As temperature decreases, n decreases and ρOHE becomes more
dominant.On the other hand, ρAHE cannot be seen in Fig. 3b because the
in a simple picture, caused by spin-polarized charge carriers that lower n gives the larger ρOHE. However, the contribution of ρAHE can be
mediate ferromagnetic exchange interaction between distant localized seen by subtracting ρOHE, wheras ρAHE cannot de detected for
spins of magnetic impurities.Hence,the ferromagnetic response caused Ti0.97Co0.03O2–δ films grown under PO = 10–5 torr and 10–4 torr (not
2

by charge carriers can be a direct manifestation of intrinsic shown), having even lower n ∼1019 cm–3 and ∼1018 cm–3 at 300 K,
ferromagnetism. A well-known ferromagnetic response of charge respectively. Secondly, we describe the x dependence of ρxy. Figure 3c,d
carriers in ferromagnetic semiconductors is the anomalous Hall effect shows magnetic-field dependence of ρxy at 300 K for Ti1–xCoxO2–δ with
(AHE)2—the emergence of voltage transverse to both the applied various x grown under PO = 10–7 torr and 10–6 torr, respectively. ρAHE is
2

222 nature materials | VOL 3 | APRIL 2004 | www.nature.com/naturematerials

©2004 Nature Publishing Group

nmat1099PRINT 3/11/04 11:32 AM Page 223

LETTERS

a 1
x = 0.03 c d
300 K 300 K
3 × 1020 cm–3 H 4
1.0
2
300 K
ρ (µΩ cm)

0.5 x = 0.10
0 0
0
2 × 1020 cm–3
200 K 0.0 –2
xy

5 × 1021 cm–3 x = 0.10

–4
4 × 1022 cm–3 –0.5 6

–1.0 4
–7 100 K
–1 10 torr 2

ρ (µΩ cm)

ρ (µΩ cm)
–2 –1 0 1 2 4 × 1021 cm–3
x = 0.05
µ o H (T) 0.5 0 0
2 × 1020 cm–3

xy

xy
0.0 –2
b 100
x = 0.05
x = 0.03 –4
7 × 1021 cm–3 –0.5 6
–6
4 × 1019 cm–3 4
3 × 1018 cm–3
ρ (µΩ cm)

0.5 2
300 K x = 0.03
0 0 0
2 × 1020 cm–3 0.0 2 × 1020 cm–3
xy

4 × 1022 cm–3 x = 0.03 –2

200 K
–0.5 –4
10–7 torr 10–6 torr
10–6 torr 100 K –6
–100 –2 –1 0 1 2 –2 –1 0 1 2
–2 –1 0 1 2
µ o H (T) µ o H (T)
µ o H (T)

Figure 3 Magnetic-field dependence of Hall resistivity (ρxy) for Ti1–xCoxO2–δ. ρxy versus magnetic field curves: a,b,At different temperatures for Ti0.97Co0.03O2–δ grown under
PO2 = 10–7 torr (a) and PO2 = 10-6 torr (b).c,d,At 300 K for Ti1–xCoxO2–δ with different x grown under PO2 = 10–7 torr (c) and 10-6 torr (d).The applied magnetic field is perpendicular to the
film plane.The electron concentration evaluated from the ordinary part of ρxy is labelled for each curve.

dominant,and increases with increasing xfor Ti1–xCoxO2–δ with higher n out the increase of ρAHE with x. It is noted that the scaling behaviour is
(PO = 10–7 torr, Fig. 3c), whereas ρOHE is dominant and ρAHE increases
2
maintained for wide range of σxx in the four orders of magnitude.
with increasing x for Ti1-xCoxO2-δ with lower n (PO = 10–6 torr, Fig. 3d).
2
This behaviour can be revealed because of the large change of
ρAHE cannot be detected for Ti0.99Co0.01O2–δ films even with higher n semiconducting σxx with δ and temperature for this compound as seen
grown under PO =10–7 torr (not shown).These results indicate that ρAHE
2

develops with increasing n and/or increasing x.

Recently, a quantitative theory for AHE in group III-V
ferromagnetic semiconductors has been proposed13.In this theory,Hall
100
conductivity (σxy ≡ ρxy/(ρxx2 + ρxy2)) is shown to be an essential measure
of the strength of AHE,where the anomalous part of σxy (σAHE) depends
10–1
on the contents of both carrier and spin,showing good coincidence with
experimental results14 for (Ga,Mn)As. We show the relationship
10–2
between σAHE and σxx (∝ n) for a series of Ti1–xCoxO2–δ as shown in Fig. 4,
x = 0.10 0.05 0.03
where ρAHE is evaluated by subtracting the linear background in ρxy
versus magnetic-field curves. As can be seen, the relationship shows a 10–3
scaling behaviour being σAHE ∝ σxxα with α ∼1.5–1.7 for the data with
σAHE ( Ω–1 cm–1)

300 K
different temperatures,x,and PO2.Sizable xdependence of σAHE is hardly 10–4
seen, contrary to that of ρAHE, because the increase of ρxx with x cancels 200 K
10–5

10–6 100 K
Figure 4 The relationship between the anomalous part of σxy (σAHE) and σxx for
Ti1–xCoxO2–δ. σAHE versus σxx plots for a series of Ti1–xCoxO2–δ with different Co contents 10–7
(x),measurement temperatures,and PO2.Red,green and blue plots correspond to the
data for x = 0.03,0.05 and 0.10,respectively.Rectangle,circle and triangle plots represent 10–8
the data for PO2 = 10–8 torr,PO2 = 10–7 torr,and PO2 = 10–6 torr,respectively.The data at
100 K,200 K and 300 K from the same sample are connected by lines.For each sample, σxx 10–9
increases with increasing temperature from 100 to 300 K as exemplified for 10–2 10–1 100 101 102 103 104
x = 0.03 with PO2 = 10–7 torr (red circles). σxx ( Ω–1 cm–1)

nature materials | VOL 3 | APRIL 2004 | www.nature.com/naturematerials 223

©2004 Nature Publishing Group

nmat1099PRINT 3/11/04 11:32 AM Page 224
LETTERS
in Fig. 2, whereas the change of σxx for ferromagnetic (Ga,Mn)As
is much smaller.
In the classical model of AHE for ferromagnetic metals, ρxy is
proportional to ρxx or ρxx2 (hence σxy is nearly proportional to σxx or σxx2)
are attributed to skew scattering or side jump,respectively12.The present
scaling law apparently sits between them. Taking into account the
relatively low mobility in this compound, the behaviour might be
understood as a consequence of the semiclassical nature of the charge-
carrier dynamics. It is noted, however, that σAHE is scaled only by σxx
irrespective of temperature, x, and oxygen deficiency to fall into a single
relationship extending over six orders of magnitude in σAHE. This fact
represents that the charge-carrier dynamics in the event of AHE is not
affected significantly by scattering by phonon, magnetic impurity and
non-magnetic impurity (that is, oxygen deficiency). Thus, the scaling
behaviour may imply a universal nature of the underlying physics of
AHE in this compound.The quantitative analysis of σAHE will be difficult
because even the band structure of TiO2 is still unclear at present.
However, this experimentally identified scaling law would stimulate
theoretical challenges to understand the band structure and AHE, and
these challenges would be of great value from the viewpoints of both
condensed matter physics and spintronics applications.
In conclusion, the scaling behaviour of AHE governed by σxx (or n)
ensures that ferromagnetically spin-polarized charge carriers play a
significant role in the emergence of ferromagnetism for this compound,
although detailed investigation will be needed both for experimental
and theoretical studies. Furthermore, the enhancement of σAHE over six
orders of magnitude by increasing σxx may directly lead to the possible
realization of semiconductor spintronics devices operable at room
temperature in such as field-effect devices.
METHODS
Rutile Ti1–xCoxO2–δ (x = 0.01, 0.03, 0.05, 0.1) epitaxial thin films were fabricated by using laser molecular-
beam epitaxy (MC-PLD, Pascal). The Ti1–xCoxO2–δ ceramics target was ablated by a KrF excimer laser
(Compex-102, Lambda Physik). The films with (101) orientation were grown on r-sapphire substrates at
400 °C. PO were varied from 1 × 10–8 torr to 1 × 10–4 torr to control the amount of δ. The reflection high-
2 Supplementary Information accompanies the paper on www.nature.com/naturematerials
energy electron diffraction (RHEED) intensity was monitored in situ during the film growth, and the
intensity oscillation could be observed during the initial stage of growth up to 5 nm thick
(Supplementary Information, Fig. S1). The film thicknesses were 70–120 nm. From X-ray diffraction
224
©2004 Nature Publishing Group
(Multiflex, from Rigaku), scanning electron microscopy (JSM-6700, JEOL), transmission electron
microscopy (H-9000NAR, Hitachi), and atomic force microscopy (SPI 3700, Seiko Instruments), neither
impurity phase nor segregation of secondary phase was observed. Photolithographically patterned Hall
bars (200 µm long × 60 µm wide) were used for transport measurements (PPMS 6000, Quantum
Design), where the magnetic field was applied along the out-of-plane direction. From the magnetization
measurements (MPMS-XL5HG, Quantum Design), the easy magnetization axis was verified to be along
the out-of-plane direction. The magnetic-field dependence of magnetization was coincident with that of
ρAHE including the negligibly small hysteresis (Supplementary Information, Fig. S2).
Received 11 December 2003; accepted 20 February 2004; published 21 March 2004.
References
1. Wolf, S. A. et al. Spintronics: a spin-based electronics vision for the future. Science 294, 1488–1495
(2001).
2. Ohno, H. Making nonmagnetic semiconductors ferromagnetic. Science 281, 951–956 (1998).
3. Ohno, H. et al. Electric-field control of ferromagnetism. Nature 408, 944–946 (2000).
4. Chiba, D., Yamanouchi, M., Matsukura, F. & Ohno, H. Electrical manipulation of magnetization
reversal in a ferromagnetic semiconductor. Science 301, 943–945 (2003).
5. Koshihara, S. et al. Ferromagnetic order induced by photogenerated carriers in magnetic III-V
semiconductor heterostructures of (In,Mn)As/GaSb. Phys. Rev. Lett. 78, 4617–4620 (1997).
6. Chiba, D., Takamura, K., Matsukura, F. & Ohno, H. Effect of low-temperature annealing on
(Ga,Mn)As trilayer structures. Appl. Phys. Lett. 82, 3020–3022 (2003).
7. Matsumoto, Y. et al. Room-temperature ferromagnetism in transparent transition metal-doped
titanium dioxide. Science 291, 854–856 (2001).
8. Matsumoto, Y. et al. Ferromagnetism in Co-doped TiO2 rutile thin films grown by laser molecular
beam epitaxy. Jpn J. Appl. Phys. 40, L1204–L1206 (2001).
9. Dietl, T., Ohno, H., Matsukura, F., Cibert, J. & Ferrand, D. Zener model description of ferromagnetism
in zinc-blende magnetic semiconductors. Science 287, 1019–1022 (2000).
10. Fukumura, T. et al. Exploration of oxide-based diluted magnetic semiconductors toward transparent
spintronics. Appl. Surf. Sci. 223, 62–67 (2004).
11. Prellier, W., Fouchet, A. & Mercey, B. Oxide-diluted magnetic semiconductors: a review of the
experimental status. J. Phys. Condens. Matter 15, R1583–R1601 (2003).
12. Chien, C. L. & Westgate, C. R. (eds) The Hall Effect and its Applications (Plenum, New York, 1980).
13. Jungwirth, T., Niu, Q. & MacDonald, A. H. Anomalous Hall effect in ferromagnetic semiconductors.
Phys. Rev. Lett. 88, 207208 (2002).
14. Jungwirth, T. et al. Dc-transport properties of ferromagnetic (Ga,Mn)As semiconductors. Appl. Phys.
Lett. 83, 320–322 (2003).
Acknowledgements
This work was supported by the Japanese Ministry of Education, Culture, Sports, Science and Technology
in Japan, Grant-in-Aid for Creative Scientific Research (14GS0204 an 13NP0201), NEDO International
Joint Research program (02BR3) and the Inamori Foundation.
Correspondence and requests for materials should be addressed to T. F.
Competing financial interests
The authors declare that they have no competing financial interests.
nature materials | VOL 3 | APRIL 2004 | www.nature.com/naturematerials