Evolution of the ferromagnetic phase of ultrathin Fe films grown on GaAs(100)-4x6

Abstract

Epitaxial bcc Fe has been grown on GaAs(100)-(4×6) at room temperature and studied with in situ magneto-optical Kerr effect (MOKE), low-energy electron diffraction, and alternating gradient field magnetometry (AGFM). The magnetic properties at room temperature were found to proceed via three phases; a nonmagnetic phase for the first three and a half monolayers, a short-range-ordered superparamagnetic phase, and a ferromagnetic phase above about five monolayers. The thickness dependencies of the coercivity and MOKE intensity further suggested that the ferromagnetic phase is subdivided into three distinct regimes with different magnetic properties. A combination of the in situ MOKE and ex situ AGFM measurements shows that the entire Fe film is ferromagnetic with a bulklike moment after the onset of the ferromagnetism, in contrast with previous studies, in which magnetic dead layers or half-magnetization phases due to the intermixing of Fe and As were proposed. The results show that it is the growth morphology of the ultrathin films, rather than the diffusion of As, that plays the dominant role in determining the magnetic properties in this system.
© 1998 The American Physical Society.

Full text

Page 1

Evolution of the ferromagnetic phase of ultrathin Fe films grown on GaAs„100…-4?6
Y. B. Xu, E. T. M. Kernohan, D. J. Freeland, A. Ercole, M. Tselepi, and J. A. C. Bland
Cavendish Laboratory, University of Cambridge, Cambridge CB3 0HE, United Kingdom
?Received 10 December 1997?
Epitaxial bcc Fe has been grown on GaAs?100?-(4?6) at room temperature and studied with in situ
magneto-optical Kerr effect ?MOKE?, low-energy electron diffraction, and alternating gradient field magne-
tometry ?AGFM?. The magnetic properties at room temperature were found to proceed via three phases; a
nonmagnetic phase for the first three and a half monolayers, a short-range-ordered superparamagnetic phase,
and a ferromagnetic phase above about five monolayers. The thickness dependencies of the coercivity and
MOKE intensity further suggested that the ferromagnetic phase is subdivided into three distinct regimes with
different magnetic properties. A combination of the in situ MOKE and ex situ AGFM measurements shows that
the entire Fe film is ferromagnetic with a bulklike moment after the onset of the ferromagnetism, in contrast
with previous studies, in which magnetic dead layers or half-magnetization phases due to the intermixing of Fe
and As were proposed. The results show that it is the growth morphology of the ultrathin films, rather than the
diffusion of As, that plays the dominant role in determining the magnetic properties in this system.
?S0163-1829?98?04127-7?
INTRODUCTION
Fe on GaAs continues to be of interest as a model system
for the epitaxial growth of ferromagnetic metals ?FM? on
semiconductors. It has been shown previously by several
groups1–5that bcc Fe grows epitaxially on both the ?001? and
?011? surfaces of GaAs, due in part to the fact that the lattice
constant of bcc Fe (a0?2.866 Å) is almost exactly half that
of GaAs (a0?5.654 Å). Fe/GaAs is also of current interest
due to its potential for use in magnetoelectronic devices such
as FM spin injection pads.6,7Such spin-sensitive devices re-
quire well-defined and magnetic interface layers. However, a
strong reduction of the magnetization has previously been
found for Fe grown on GaAs.1The reduction of the Fe mo-
ment was attributed to the magnetically ‘‘dead’’ layers near
the interface, which would be detrimental to the spin-
dependent transmission and tunneling between the ferromag-
netic metal and the semiconductor substrate. Thus the inter-
face structure and magnetism is a key issue for current
research.
The magnetic hysteresis loops measured using in situ
magneto-optical Kerr effect ?MOKE? by Gester et al.8
showed that the ferromagnetic phase developed after about
15 Å (?10 ML) when Fe was grown on GaAs?001?-
(4?6) at 175 °C. Kneedler et al.5showed that the onset of
ferromagnetism occurred at 6 ML when Fe was grown on
both GaAs?001?-(2?4) and c(4?4) substrates. The mag-
netic dead layer in Fe/GaAs was attributed to the formation
of antiferromagnetic Fe2As microstructures at the interface
due to the As diffusion.1More recently, ex situ magnetic
measurements4using a superconducting quantum interfer-
ence device and alternating gradient field magnetometry
?AGFM? suggested the existence of a nearly half-magnetized
phase Fe3Ga2?xAsxat the interface instead of dead layers.
To prevent the formation of compounds at the Fe/GaAs in-
terface, S-passivated GaAs substrates have been exploited.9
These have 1 ML of bridge bonded sulphur that acts as a
surfactant to inhibit interdiffusion of As into the Fe over-
layer. The Fe films were found to be ferromagnetic after
about 4 ML of deposition.
In general, the magnetic properties of the first few mono-
layers are expected to be determined not only by the inter-
mixing at the interface, but also by the morphology of the
substrate and the deposited films. An interesting example is
the Co/Cu system. High quality layer-by-layer growth has
been achieved on both Cu?001? and Cu?111? substrates and
ferromagnetic hysteresis loops were observed at room tem-
perature for less than 2 ML of Co.10,11In contrast, the Co/
Cu?110? shows a three-dimensional ?3D? growth mode,12–14
possibly due to the corrugated Cu?110? surface. The onset of
the room-temperature ferromagnetism was found to be at
around 4.6 ML, when the islands began to coalesce.14Three-
dimensional growth ?Volmer-Weber mode? has been re-
ported on both Fe/GaAs?001? and ?011?.15–17A detailed low-
energy electron diffraction ?LEED? study further suggested
that a pyramidlike structure forms when Fe was grown on
GaAs?100?-(4?6),18similar to the pyramids observed in the
Fe/MgO system.19These ‘‘self-organized’’ structures are in-
teresting from the viewpoint of understanding the micromag-
netism of nanoclusters and the evolution of magnetic phases.
For example, superparamagnetic relaxation has been studied
for a 10-ML film of Fe grown on MgO?001?,20and related to
the particle size. Also, the nanoscale structure of pseudomor-
phic Fe?110? on W?110? was found to induce a rich variety
of new micromagnetic phenomena.21Technologically, these
nanostructures may have future applications in ultra-high-
density data storage.
Although the Fe/GaAs system has been extensively stud-
ied, the magnetic properties of the first few monolayers are
poorly understood, and there is still debate over whether
there are magnetically dead layers at the interface. In this
paper, the magnetic properties and structure of Fe grown on
GaAs?001?-(4?6) at room temperature have been studied.
A picture of the relationships between the Fe coverage, its
structure, and the magnetic phases has been proposed using
in situ MOKE and LEED and ex situ AGFM. The results
PHYSICAL REVIEW B1 JULY 1998-IIVOLUME 58, NUMBER 2
PRB 58
0163-1829/98/58?2?/890?7?/$15.00890© 1998 The American Physical Society

Page 2

suggest that there is no dead layer at the interface and that
the Fe shows a bulklike moment.
EXPERIMENTS
Fe films were grown on GaAs substrates in a molecular
beam epitaxy chamber using an e-beam evaporator. During
growth, the pressure was below 6?10?10mbar. The depo-
sition rate was monitored by a quartz microbalance that was
calibrated using reflection high-energy electron diffraction
oscillations. The Fe film was grown at ambient temperature
?35 °C? at a rate of approximately 1 ML per min. The sub-
strates used in this study are As capped GaAs?001? prepared
in another UHV chamber. A buffer layer (?0.5 ?m) of ho-
moepitaxial GaAs was grown on the commercial wafer to
provide the smoothest possible GaAs surface. The As cap
layer was then desorbed by annealing for metal film growth.
The As capping layer began to desorb at around 340 °C and
the substrate was further annealed to 550 °C for 1 h to obtain
a clean and ordered surface.
The surface structure of the substrate and the Fe films was
determined by means of LEED. Diffraction images were re-
corded from the phosphor screen using a conventional
charge-coupled-device camera. The magnetic properties of
the Fe films were studied using in situ MOKE. The MOKE
loops were collected during growth in the longitudinal geom-
etry using an electromagnet with a maximum field of 2 kOe,
and an intensity stabilized HeNe laser ?633 nm?.22The mag-
netization was measured ex situ using AGFM with sensitiv-
ity up to 10?6emu. The AGFM was calibrated with a
built-in coil and further checked against thick Fe and Ni
films.
RESULTS
Figure 1 shows the LEED patterns of ?a? the GaAs sub-
strate after As desorption, and ?b?–?f? after Fe deposition.
The LEED picture of the substrate shows a very clear
p(4?6) reconstruction, typical for Ga-rich surfaces.23This
clear and sharp LEED pattern for the reconstructed surface
indicates that the GaAs substrate surface is very flat and well
crystallized. Auger measurements show that the substrate is
free of O and C after As desorption. The LEED patterns were
monitored as Fe was deposited. No Fe LEED pattern was
observed for the first 4 ML deposited as shown in Fig. 1?b?.
After the deposition of 5 ML, faint LEED spots from the Fe
film appear. Clear LEED patterns were observed after the
deposition of 6 ML. The diffraction spots became broadened
at higher coverages as shown in Figs. 1?e? and 1?f?. The
LEED patterns show that Fe grows epitaxially on GaAs?001?
atroomtemperaturewith
Fe?001??100??GaAs?001??100?. The lack of Fe LEED pat-
terns for the first 4 ML indicates that the growth proceeds via
the three-dimensional Volmer-Weber growth mode as previ-
ously reported for higher temperature growth.1,17,18,24The
LEED pattern develops at a higher Fe coverage ?5 ML? than
that at higher growth temperature ?3 ML?.18This is consis-
tent with the previous finding that the optimum growth tem-
perature is around 170 °C.1,17,18
Figure 2 shows the development of the MOKE loops with
thickness. The magnetic field is applied along the ?01¯1? di-
theepitaxialrelationship
rection. No MOKE signal was observed from the substrate,
which showed that the magneto-optical Kerr effect of GaAs
is negligible for the applied field strength of up to 2 kOe. A
significant MOKE signal was first detected at a thickness of
3.5 ML, with the intensity linearly proportional to the ap-
plied magnetic field. With further Fe deposition the MOKE-
loop curves become s-shaped at 4.3 ML. The lack of hyster-
esis indicates that the ferromagnetic phase has not yet
developed. The magnetization curves indicate the presence
of either paramagnetism or superparamagnetism. The loop in
Fig. 2?e? clearly shows hysteresis, indicating the onset of the
ferromagnetic phase after 4.8 ML of Fe. Figures 2?f?–2?j?
show the hysteresis loops after the onset of the ferromagnetic
phase. These loops display the observed variation in the co-
ercivity with thickness, which is plotted in Fig. 3?b?.
We note that the hysteresis loops in Figs. 2?c?–2?e? show
an asymmetry under the transformation M→?M, H→
?H. This might be due to the second-order term in the
magneto-optical response. For example, if there is a trans-
verse magnetization component, it will give a contribution in
the longitudinal measurements because of the quadratic
FIG. 1. LEED patterns of ?a? the GaAs?001?-4?6 substrate af-
ter As desorption, 135 eV, and ?b?–?f? after Fe deposition, 120 eV.
PRB 58
891EVOLUTION OF THE FERROMAGNETIC PHASE OF . . .

Page 3

magneto-optical effect. A fuller explanation will require fur-
ther experimental and theoretical investigation, which is be-
yond the scope of this paper.
Figure 2 indicates that the magnetic easy axis is along the
?01¯1? direction rather than along the ?001? direction, the easy
axis of the bulk bcc Fe. This is due to a strong uniaxial
anisotropy. Although this uniaxial anisotropy has been ob-
served in several previous studies,1,5,8its origin remains an
open question. It might be due to the shape anisotropy, an-
isotropic strain relaxation, or the different nature of the
Fe-Ga and Fe-As bonds. The uniaxial anisotropy has also
been examined here, although systematic studies were not
attempted and we do not attempt to answer this interesting
question. Figure 3 shows the hysteresis loops of the Fe films
of ?a?, ?b? 5 ML and ?c?, ?d? 40 ML for the magnetic field
applied along the ?01¯1? and ?011? directions, respectively.
Figure 3 shows that the uniaxial anisotropy develops imme-
diately after the onset of the ferromagnetic phase at around 5
ML and persists up to 40 ML.
It has been shown that for ultrathin ferromagnetic films
the Kerr effect initially depends linearly on the thickness if
the magnetization is thickness independent.25Calculations of
the magneto-optic response of Fe films supported by GaAs
predict a near linear dependence up to at least 40 ML.26The
MOKE signal from the detector is proportional to the Kerr
effect, the intensity of the light, and the setting of the polar-
imeter. During in situ experiments that monitor the thickness
FIG. 2. In situ MOKE hysteresis loops for the Fe/GaAs?001?-
4?6 of different Fe thicknesses with the magnetic field applied
along the ?01¯1? direction.
FIG. 3. In situ MOKE hysteresis loops of ?a?, ?b? Fe?5 ML?/GaAs?001?, and ?c?, ?d? Fe?40 ML?/GaAs?001? for the magnetic field applied
along the ?01¯1? and ?011? directions, respectively.
892PRB 58
Y. B. XU et al.

Page 4

dependence of the MOKE intensity, care was taken not to
move either the sample or any of the optical components, so
eliminating the possibility of variations in intensity due to
changes in the optical alignment. The magnet was moved
away from the sample position during growth to avoid any
change of the deposition rate caused by the stray field. After
each deposition, the magnet was moved back for the mea-
surement while keeping the sample position unchanged.
The thickness dependence of the MOKE intensity is
shown in Fig. 4?a?. The empty and filled circles are the re-
sults before and after the onset of the ferromagnetic phase,
respectively. Figure 4?a? shows that the MOKE intensity in-
creases rapidly between 3.5 and 4.3 ML, just before the onset
of the ferromagnetism. Extrapolation of these points indi-
cates that the thickness of the nonmagnetic phase is about
3.2?0.2 ML. After the onset of ferromagnetism, the MOKE
signal is approximately linearly proportional to the thickness
as shown by the filled circles. Extrapolation of these solid
dots suggests that there are no magnetically dead layers, and
that the entire Fe film is ferromagnetic. The MOKE signal at
higher coverage ?above about 12 ML? shows a slightly re-
duced slope. The thickness dependence of the coercivity is
shown in Fig. 3?b?. The coercivities are rather small just after
the onset of the ferromagnetism. There is a sharp increase of
the coercivity around 5 ML. From about 6 to 10 ML, it is
almost constant and then increases slightly with further in-
creasing thickness.
Although the MOKE signal is proportional to the magne-
tization, it does not directly give the magnetic moment of the
sample. This was measured using AGFM. In carrying out
these measurements, the samples were capped with Au to
prevent oxidation. The thicknesses of the samples grown for
the AGFM measurements were chosen not to be very small
in order to minimize the effect of the Fe/Au interface and the
diamagnetic signal of the substrate. As indicated by the in
situ MOKE results in Fig. 4?a?, the magnetization does not
vary strongly with thickness after the onset of the ferromag-
netism. Figure 5 shows the magnetization hysteresis loops of
two samples; Fe?14 ML?/GaAs?001? and Fe?40 ML?/
GaAs?001?. The total magnetic moment from the 40 ML of
Fe is about 2.8 times bigger than that from the 14 ML, which
is in proportion to their thicknesses. The magnetization of
the films is 1.6?0.2?103emu/cm3, only slightly smaller
than that of the bulk bcc Fe?1.71?103emu/cm3?. The
AGFM measurements further show that the magnetization is
approximately thickness independent and the Fe films have a
bulklike moment.
DISCUSSION
These above results are of interest in the context of two
controversial questions concerning the basic magnetic prop-
erties of the ultrathin Fe films grown on GaAs. First, is there
any dead layer or half-magnetization phase near the Fe/GaAs
interface? The lack of magnetization for coverages less than
3.5 ML may be due to the intermixing of Fe with As and Ga
and the formation of nonferromagnetic compounds near the
interface region,1,8or it could be due to the formation of
clusters. As we mentioned in the introduction, the ferromag-
netic phase develops after more than 4 ML of deposition in
the Co/Cu?110? system due to the 3D growth.14Second, is
there local ferromagnetic ordering before the onset of the
ferromagnetic phase? The magnetization signal before the
onset of the ferromagnetism could, in principle, be due to
either a paramagnetic response or superparamagnetism.
FIG. 4. Thickness dependencies of the MOKE intensity and the
coercivity of Fe/GaAs?001?-4?6. The open dots are the results for
the superparamagnetic phase in an applied field of 2 kOe, and the
filled dots are the saturated MOKE intensity of the ferromagnetic
phase. The error bars ?not shown? are comparable with the size of
the data symbols.
FIG. 5.
GaAs?001? and Fe?40 ML?/GaAs?001? measured using AGFM with
the magnetic field applied along the ?01¯1? direction.
Magnetization hysteresis loops of Fe?14 ML?/
PRB 58
893EVOLUTION OF THE FERROMAGNETIC PHASE OF . . .

Page 5

In combination with the structural information obtained
from LEED, we propose that the correlation between the
coverage, morphology, and magnetic phases is as shown in
Fig. 6. The lack of the Fe LEED patterns suggests that the
films are not continuous below 4 ML and that clusters are
formed in the early stages of growth. Chambers et al.24
showed that Fe clusters with at least 3 ML height grew on
c(8?2) reconstructed GaAs?100? for coverage up to about 4
ML. This 3D growth mode of the Fe/GaAs system has been
confirmed by scanning tunnel microscope ?STM? images,16,17
though most of these STM studies mainly concentrated on
the submonolayer coverage range. The lack of magnetic sig-
nal for the first 3.5 ML might be due to the smaller initial
cluster size, which prevents the development of magnetic
ordering, or the ordering above room temperature. As more
Fe is deposited, the islands will grow and coalesce to form
bigger clusters. The exchange interaction within these clus-
ters becomes stronger and leads to internal ferromagnetic
ordering,27,28so giving rise to the well-known superparamag-
netic phase.29The lack of hysteresis is consistent with either
superparamagnetism,29or 2D paramagnetism.30However,
the s-shaped loops that were observed in this region are gen-
erally consistent with the Langevin function used to describe
the magnetization of superparamagnetic clusters.20,29,31Fit-
ting the curves of Figs. 2?c? and 2?d? within the range of
?1 kOe with a Langevin function, the average values of the
effective magnetic moment per cluster are obtained to be
(1.05?0.15)?104?Band (4.40?0.65)?104?B, respec-
tively, for the films of the coverage of 4 and 4.3 ML. Thi-
bado et al.17found that the average island width?length of 1
ML of Fe on GaAs?001?-2?4 is 35?90 Å2. Gu et al.32im-
aged a thick Fe film ?150 Å? on GaAs?001?-4?6 and found
that the film has islandlike undulations of about 10 Å height
and about 150 Å in diameter. Assuming average island sizes
of 100?100 Å2, and height 5 ML ?7.15 Å? for the coverage
of about 4 ML, the magnetic moment is 1.43?104?B,
which is comparable with the effective moments estimated
by fitting the magnetization curves using the Langevin func-
tion. Thus we can conclude that a superparamagnetic phase
develops in the thickness range 3.5–4.8 ML.
With further increase in the coverage, the islands coalesce
and long-range ferromagnetic ordering develops. The hyster-
esis loops after the onset of the ferromagnetic phase in Fig. 2
show that the films have a well-defined magnetic coercivity
and remnance ratio, indicating the behavior of a continuous
film. We should note that long-range ordering ?as well as the
first appearance of a LEED pattern? may possibly develop
before the complete coalescence of the islands due to inter-
particle interactions. A detailed combination of high-
resolution scanning tunneling microscopy and in situ MOKE
measurements is required to determine exactly the morphol-
ogy near to the transition.
The magnetic properties of the films after the onset of the
ferromagnetism show an interesting three-stage behavior.
The coercivities of the films are rather small just after the
onset of the ferromagnetism, and rise sharply up to 6 ML.
The coercivity then remains almost constant up to 10 ML
before increasing slightly with higher coverages. The sharp
increase of the coercivity is quite similar to a critical
behavior,33suggesting that thermal fluctuations are important
in the magnetization reversal process just after the onset of
the ferromagnetic phase. It has been shown by Schumann
and Bland34that the coercivity follows a power law Hc(d)
?(d/dc?1)?in the Co/Cu?100? system just after the onset
of the ferromagnetic phase. The further increase of the coer-
civity above about 10 ML may be due to a structural change.
It has been shown by Anderson et al.35that the epitaxial
quality of the Fe/GaAs?001? degraded after about 12 ML.
This is consistent with our LEED measurements ?Fig. 1?,
which show a broadening of the diffraction spots at higher
coverages, indicating a reduction of the film quality. It is also
interesting to note that the slope of the MOKE intensity de-
creases slightly in this region. Taken together, these effects
suggest that there is indeed a significant change in the struc-
tural and magnetic properties after about 10–12 ML.
The critical thickness ?4.8 ML? corresponding to the onset
of the ferromagnetic phase is much lower than that of the
films grown at higher temperature ?10 ML?,8and is compa-
rable with that of the films grown at room temperature on the
S-passivated GaAs substrates, where the onset of the ferro-
magnetic phase was found to be at about 4.0 ML.9The com-
bination of the MOKE and AGFM shows that the entire Fe
films studied here are ferromagnetic with a bulklike moment.
This is very different from the results of previous studies,
where ex situ magnetic measurements showed a magnetic
dead layer of about 16 ML.1More recently, this dead layer
was attributed to a half-magnetization phase near the
interface.4The thickness D of this half-magnetization phase
depends on the growth temperature T, D?10 ML for T
?50 °C and D?60 ML for T?200 °C.
These differences demonstrate the importance of the sub-
strate preparation and growth temperature. The sharp LEED
image in Fig. 1?a? showing the (4?6) surface reconstruction
indicates that the substrate used in this study is well ordered
and has a long coherence length. Such a clean and flat GaAs
surface would favor the Fe growth.
Ga-terminated surface. It might therefore be expected that
the interdiffusion of As into the Fe layer would be dimin-
ished in the samples grown here, especially since the sub-
strate was held at ambient temperature rather than 150–
175 °C. We have used Auger spectroscopy to monitor the
interdiffusion of As. The low-energy Fe and As peaks at 46
GaAs?001?-4?6 is a
FIG. 6. A picture of the correlation between the coverage, mor-
phology, and magnetic phases of Fe films on GaAs?001?-4?6 sub-
strate grown at the room temperature.
894PRB 58
Y. B. XU et al.

Page 6

eV ?Fe? and 35 eV ?As? have been measured since the energy
limit of our Auger system does not allow us to use the high-
energy As peaks above 1 kV. The probe depth for these
low-energy secondary electrons is very short ?7–8 Å,
?5 ML?,36and so using low-energy electrons makes the
technique more surface sensitive. The As peak is still present
after 40 ML of Fe deposited, showing the out-diffusion of As
into Fe.1However, the ratio of the Auger intensities of the
As peak and Fe peak was found to be almost constant at
about IAs/IFe?0.15?0.01 for thicknesses of 10, 20, and 40
ML. Considering the very short probing depth of the low-
energy Auger electrons, the constant ratio suggests that As
floats on the surface and does not react with Fe to form
nonmagnetic compounds. This is consistent with the bulklike
magnetic moment obtained from the magnetic measurements
and explains why the first three and half nonmagnetic layers
could become ferromagnetic at higher coverages. In view of
these results, the larger critical thickness for the onset of
ferromagnetism in previous studies8might therefore be due
to the reaction of As and Fe at higher growth temperatures.
CONCLUSION
We have studied the magnetic and structural properties of
epitaxial bcc Fe grown at room temperature on GaAs?001?-
(4?6) substrates. A superparamagnetic phase was observed
to develop within a narrow thickness range before the onset
of long-range ferromagnetic ordering. The critical thickness
of the ferromagnetic phase is much smaller than that of the
structures prepared at higher growth temperatures,8and com-
parable with that of the S-passivated substrate samples.9The
in situ MOKE and ex situ AGFM results show that the entire
film is ferromagnetic with a bulklike moment after the onset
of long-range ferromagnetism. These results support the
view that there is neither a magnetic dead layer nor a half-
magnetization phase at the interface. As a final point, it is
worth mentioning that the growth of ferromagnetic metals on
semiconductor substrates may offer an opportunity to study
the micromagnetism of nanostructures and the associated
critical phenomena of phase transitions, which have recently
attracted considerable attention on the ferromagnetic/
nonmagnetic ?Fe/W,21Co/Cu,37and Fe/Cu ?Ref. 38?? sys-
tems. The magnetically active nanoclusters of Fe on GaAs
may find applications, as the dipole fields from these meso-
magnets offer a natural way to generate magnetic fields for
nanoscale semiconductor devices,39,40and the high interface
moment is favorable for magnetoelectronic applications.
ACKNOWLEDGMENTS
We gratefully acknowledge the financial support of the
EPSRC, and ESPRIT ?EC?. We thank Dr. D. A. Ritchie and
R. Balsod for their help with this work.
1G. A. Prinz, G. T. Rado, and J. J. Krebs, J. Appl. Phys. 53, 2087
?1982?; J. J. Krebs, B. T. Jonker, and G. A. Prinz, ibid. 61, 2596
?1987?.
2J. M. Florczak and E. D. Dahlberg, Phys. Rev. B 44, 9338 ?1991?.
3C. Daboo, R. J. Hicken, E. Gu, M. Gester, S. J. Gray, D. E. P.
Eley, E. Ahmad, and J. A. C. Bland, Phys. Rev. B 51, 15 964
?1995?.
4A. Filipe, A. Schuhl, and P. Galtier, Appl. Phys. Lett. 70, 129
?1997?.
5E. M. Kneedler, B. T. Jonker, P. M. Thibado, R. J. Wagner, B. V.
Shanabrook, and L. J. Whitman, Phys. Rev. B 56, 8163 ?1997?.
6S. Datta and B. Das, Appl. Phys. Lett. 56, 665 ?1990?.
7G. A. Prinz, Phys. Today 48 ?4?, 58 ?1995?.
8M. Gester, C. Daboo, R. J. Hicken, S. J. Gray, A. Ercole, and J.
A. C. Bland, J. Appl. Phys. 80, 347 ?1996?.
9G. W. Anderson, M. C. Hanf, and P. R. Norton, Phys. Rev. Lett.
74, 2764 ?1995?.
10C. M. Schneider, P. Bressler, P. Schuster, J. J. De Miguel, and R.
Miranda, Phys. Rev. Lett. 64, 1059 ?1990?.
11D. Kerkmann, D. Pescia, and R. Allenspach, Phys. Rev. Lett. 68,
686 ?1992?.
12G. Harp, R. F. C. Farrow, D. Weller, T. A. Rabedeau, and R. F.
Marks, Phys. Rev. B 48, 17 538 ?1993?.
13J. Fassbender, G. Guntherodt, C. Mathieu, B. Hillebrands, R.
Jungblut, J. Kohlhepp, M. T. Johnson, D. J. Roberts, and G. A.
Gehring, Phys. Rev. B 57, 5870 ?1998?.
14S. Hope, Ph.D. thesis, Cambridge University, 1997.
15P. N. First, J. A. Stroscio, R. A. Dragoset, D. T. Pierce, and R. J.
Celotta, Phys. Rev. Lett. 63, 1416 ?1989?.
16H. Takeshita, H. Akinaga, M. Ehinger, Y. Suzuki, K. Ando, and
K. Tanaka, Jpn. J. Appl. Phys., Part 1 34, 1119 ?1995?.
17P. M. Thibado, E. Kneedler, B. T. Jonker, B. P. Bennertt, B. V.
Shanabrook, and L. J. Whitman, Phys. Rev. B 53, R10 481
?1996?.
18M. Gester, C. Daboo, S. J. Gray, and J. A. C. Bland, J. Magn.
Magn. Mater. 165, 242 ?1997?.
19K. Thurmer, R. Koch, M. Weber, and K. H. Rieder, Phys. Rev.
Lett. 75, 1767 ?1995?.
20Yongsup Park, S. Adenwalla, G. P. Felcher, and S. D. Bader,
Phys. Rev. B 52, 12 779 ?1995?.
21N. Weber, K. Wagner, H. J. Elmers, J. Hauschild, and U. Grad-
mann, Phys. Rev. B 55, 14 121 ?1997?.
22J. A. C. Bland, M. J. Padgett, R. J. Butcher, and M. Bett, J. Phys.
E 22, 308 ?1989?.
23R. Z. Bachrach, R. S. Bauer, P. Chiaradia, and G. V. Hanson, J.
Vac. Sci. Technol. 18, 797 ?1981?.
24S. A. Chambers, F. Xu, H. W. Chen, I. M. Vitomirov, S. B.
Anderson, and J. H. Weaver, Phys. Rev. B 34, 6605 ?1986?.
25S. D. Bader, J. Magn. Magn. Mater. 100, 440 ?1991?.
26A. C. Durst, Y. B. Xu, and J. A. C. Bland ?unpublished?.
27I. M. L. Billas, A. Chatelain, and W. A. de Heer, Science 265,
1682 ?1994?.
28J. Shi, S. Gider, K. Babcock, and D. D. Awschalom, Science 271,
937 ?1996?.
29C. P. Bean and J. D. Livingston, J. Appl. Phys. 30, 120 ?1959?.
30V. Pokrovskii, Adv. Phys. 28, 595 ?1979?.
31R. Skomski, D. Sander, J. Shen, and J. Kirschner, J. Appl. Phys.
81, 4710 ?1997?.
32E. Gu, J. A. C. Bland, C. Daboo, M. Gester, L. M. Brown, R.
Ploessl, and J. N. Chapman, Phys. Rev. B 51, 3596 ?1995?.
PRB 58
895EVOLUTION OF THE FERROMAGNETIC PHASE OF . . .

Page 7

33R. B. Stinchcombe, in Phase Transitions and Critical Phenom-
ena, edited by C. Domb and J. L. Lebowitz ?Academic, London,
1983?.
34F. O. Schumann and J. A. C. Bland, J. Appl. Phys. 73, 5945
?1993?.
35G. W. Anderson, M. C. Hanf, X. R. Qin, P. R. Norton, K. Myrtle,
and B. Heinrich, Surf. Sci. 346, 145 ?1996?.
36A. Zangwill, Physics at Surface ?Cambridge University Press,
Cambridge, England, 1988?.
37F. O. Schumann, M. E. Buckley, and J. A. C. Bland, Phys. Rev. B
50, 16 424 ?1995?.
38J. Shen, R. Skomski, M. Klaua, H. Jenniches, S. S. Manoharan,
and J. Kirschner, Phys. Rev. B 56, 2340 ?1997?.
39I. P. Smorchkova, N. Samarth, J. M. Kikkawa, and D. D. Aw-
schalom, Phys. Rev. Lett. 78, 3571 ?1997?.
40A. Nogaret, S. Carlton, B. L. Gallagher, P. C. Main, M. Henini,
R. Wirtz, R. Newbury, M. A. Howson, and S. P. Beaumont,
Phys. Rev. B 55, R16 037 ?1997?.
896PRB 58
Y. B. XU et al.