Stress oscillations and surface alloy formation during the growth of FeMn on Cu„001…
Wei Pan,
1,2
Dirk Sander,
2
Minn-Tsong Lin,
1,
* and Ju
¨
rgen Kirschner
2
1
Department of Physics and Center for Nanostorage Research, National Taiwan University, Taipei106, Taiwan
2
Max-Planck-Institut fu ¨r Mikrostrukturphysik, Weinberg 2, D-06120 Halle, Germany
Received 16 May 2003; published 19 December 2003
In situ stress and medium-energy electron-diffraction MEED measurements have been performed simul-
taneously during the deposition of FeMn on Cu001. For a thickness above 5 layers, stress and MEED exhibit
coherent layer-by-layer oscillations with a period of one atomic layer, where the largest compressive stress
corresponds to the filled layer. In this thickness regime, the average stress is -0.590.02 GPa. From this, we
deduce the biaxial modulus of FeMn layers as 148 ( 5) GPa, which agrees well with the respective bulk
value. For a FeMn thickness below 1.5 layers, the resulting stress is qualitatively ascribed to the sum of the
individual stress contributions from Fe on Cu001 and Mn on Cu001.A c (2 2) low-energy electron
diffraction pattern in this thickness regime indicates the formation of a c (2 2) MnCu surface alloy in the
initial growth of FeMn on Cu001, which induces a compressive surface stress of -0.7 N/m for the initial
deposition of the FeMn alloy. This surface alloy formation leads to a Fe-rich FeMn alloy near the Cu interface.
This compositional change might modify the antiferromagnetic coupling of the 1:1 FeMn alloy.
DOI: 10.1103/PhysRevB.68.224419 PACS numbers: 75.70.-i, 68.35.Gy, 68.35.Ct, 68.55.-a
I. INTRODUCTION
Fe
x
Mn
1 -x
alloys are prototypes for antiferromagnetic ma-
terials with a high Ne
´
el temperature T
N
.
1
FeMn thin films
are widely used in a spin-valve structure in combination with
magnetoresistive materials for application in magnetic sen-
sors and magnetic data storage.
2
The lattice constant of the
Fe
x
Mn
1 -x
alloys us depends on the alloy composition x.
1
This allows to choose a composition x, which leads to a
moderate misfit between alloy film and substrate. For the 1:1
composition, i.e., x =0.5, the lattice constant of bulk FeMn is
3.629 Å.
1
This alloy has a face-centered cubic structure and
its deposition on Cu001 could lead to epitaxially ordered
films, as the misfit between FeMn and Cu is small and it
amounts to =( a
Cu
-a
FeMn
)/( a
FeMn
) =-0.4 %.
This paper focuses on combined stress measurements and
diffraction experiments to correlate atomic structure and re-
sulting stress in the thickness range from 0 to 18 layers of
FeMn on Cu001. We find that in contrast to a simplistic
growth model where FeMn alloy layers grow pseudomorphi-
cally strained on Cu001, the formation of a surface alloy
between Mn and Cu is clearly observed by the stress and
diffraction experiments. Only for an alloy thickness above 5
layers, we find experimental evidence from stress oscilla-
tions that the alloy grows in a layer-by-layer fashion. The
formation of the MnCu surface alloy has important conse-
quences for the magnetic properties of ultrathin FeMn films,
as due to the surface alloy formation, Mn atoms are incorpo-
rated into the Cu surface. This intermixing might affect the
antiferromagnetic coupling near the FeMn-Cu interface.
II. EXPERIMENT
The experiment was performed in an ultrahigh vacuum
chamber with a base pressure of 1 10
-10
mbar. A 124 m
thin Cu001 crystal width: 3 mm, length: 12 mm was used
as a substrate. The long edge of the crystal was oriented
along the 110 direction. It was clamped at one end along its
width, and the other end was free. This mounting allows us
to measure the stress during FeMn deposition by measuring
the change of curvature of the crystal during film growth.
3
The surface of the Cu001 was cleaned by Ar
+
sputtering
and followed by short annealing to 720 K 30 s by radiation
from a W shield, which was heated by a W filament. The
surface cleanliness was checked by Auger electron spectros-
copy AES and sharp diffraction spots of low-energy
electron-diffraction LEED indicated good crystalline order-
ing. The FeMn alloy was grown by codeposition from indi-
vidual Fe and Mn evaporators.
4,5
The purity of Mn and Fe is
99.5% and 99.99% at.%, respectively. The deposition rate
of the alloy was set to a 1:1 atomic ratio by adjusting the
evaporation rate of each evaporator, as confirmed by indi-
vidual MEED measurements and by AES.
6–9
The error for
the alloy composition is estimated to be 5%. Both evapo-
rators were adjusted to give a deposition rate close to 0.25
ML/min, which results in a growth rate of the FeMn alloy of
0.5 ML/min. Here ML stands for monolayer.
The stress measurement was carried out during film
growth at room temperature. The mechanical stress in the
alloy film
f
induces a change of curvature of the crystal
(1/R ), (
f
t
f
) = Yt
s
2
/6(1 - ) (1/R ), where t
f
and t
s
are the thicknesses of the alloy film FeMn and the substrate
Cu, respectively, Y and are the Young’s modulus and
Poisson ratio of Cu001, and R is the radius of the
curvature.
10,11
The slope of the curvature signal as a function
of film thickness gives the film stress
f
in gigapascal. The
curvature was obtained by a highly sensitive optical beam
deflection technique, which is schematically shown in Fig. 1.
Two laser beams are reflected from two points of the surface,
which are separated by a distance d, onto two position-
sensitive split-photodiode detectors. The detectors are
mounted at a distance L away from the substrate. Thus the
curvature (1/R ) is derived from the difference of the posi-
tion signal as (1/R ) =( P
1
- P
2
)/2dL , where P
1
and
P
2
are the changes of the position signals for the bottom
PHYSICAL REVIEW B 68, 224419 2003
0163-1829/2003/6822/2244195/$20.00 ©2003 The American Physical Society 68 224419-1