Surfactant Effect of Oxygen Atoms on Epitaxial Growth of fcc Ultra
Materials Transactions, Vol. 43, No. 8 (2002) pp. 2143 to 2147
The Japan Institute of Metals
Surfactant Effect of Oxygen Atoms on Epitaxial Growth
of fcc Ultra Thin Films on Cu(001)
Masaki Onishi, Lin Li, Shingo Torii ∗ , Ayumu Kida, Masaaki Doi and Masaaki Matsui
Department of Crystalline Materials Science, Faculty of Engineering, Nagoya University, Nagoya 464-8603, Japan
Surface active agent (surfactant) effect of the adsorbed oxygen atoms on epitaxial growth of fcc ultra thin films (Co, γ -Fe, Cr) on Cu(001),
a Co/γ -Fe/Co trilayer on Cu(001) and a [Co/Cu]
√multilayer was investigated. The surfactant effect of oxygen atoms on the growth of fcc Co on
Cu(001)–O reconstructed structure of (2 2 × 2)R45◦ was found by the RHEED observation. The non-equilibrium fcc structure of Fe was
stable until 45 ML on the Cu(001)–O surface formed on Cu(001) single crystal, but 20 ML on the Cu(001)–O surface formed on Cu(001) buffer
layer deposited on MgO(001). The surfactant effect was not observed for the non-equilibrium fcc Cr. The surfactant effect√of oxygen
on the Co/γ -Fe/Co trilayer and [Co/Cu]20 multilayer were observed. The surface reconstruction structure of Cu(001)–O(2 2 × 2)R45◦ is
necessary for the oxygen surfactant effect on the epitaxial growths of fcc metals on Cu(001).
(Received March 15, 2002; Accepted June 21, 2002)
Keywords: ultra thin films, multilayer, oxygen surfactant, epitaxial growth, molecular beam epitaxy, reflection high energy electron
diffraction, magnetic properties, (GMR)
For recent developments of the high-density magnetic
recording, the large MR (Magneto-resistance) ratio in GMR
head is strongly required. Several studies have been reported
on the relationship between the flatness at interfaces of the
multilayer and the MR ratio.1–3) And it is known that the flat
interface controlled in the atomic level is effective to enhance
the MR ratio as shown in the reports, where the [Fe/Cr] multilayer prepared by the MBE method had fairly large MR ratio
as 116% at 4.2 K,4) and 220% at 1.5 K.5) The hetero-epitaxial
multilayer prepared by the MBE method could have the sharp
interface and show the large MR ratio.
On the other hand, the theoretical calculation suggests that
[Co/non-magnetic γ -Fe] multilayer shows larger MR ratio6)
than the [Fe/Cr] system. We already reported the preparation and magnetic properties of Co/γ -Fe/Co trilayer film.7)
The small MR ratio was, however, reported due to the difficulty of the preparation of high quality Co/γ -Fe/Co trilayer, where it was a problem that the γ -Fe is an unstable
phase at room temperature, and it grows up to only 11 ML
(1 ML = 0.180 nm) on the clean Co(001)8) or on the clean
Cu(001)9–11) single crystal substrate. However, the critical
thickness of γ -Fe depends on the surface roughness of the
substrate. In our previous study,7) in the case of growth on the
Cu buffer (100 nm) layer deposited on MgO(001), the critical
thickness was 7 ML, which was insufficient for studying the
large MR ratio.
Now, it is well known that the surfactant effect in the
growth of ultra thin film could improve the interface flatness of the multilayer12) and make the growth mode layerby-layer. The surfactant atoms are sometimes effective to
stabilize the non-equilibrium phase13) at room temperature.
Then, the surfactant effects of CO, C2 H2 , C2 H2 + O2 , and
CH4 gasses on the growth of γ -Fe were found by Kirilyuk
et al.14, 15) We also reported the surfactant effect of pure O2
gas on the γ -Fe growth and we were succeeded in the 45 ML
address: Fujitsu Co. Ltd.
in γ -Fe thickness.16) Before our report, the surfactant effect
of the oxygen was reported in the homo-epitaxial growth of
oxygen atoms constitute
Cu on Cu(001),17) where the 0.5
a reconstruction structure of (2 2 × 2)R45◦ with Cu atoms
on Cu(001) surface. The oxygen gas adsorption method is
useful to study the surfactant effect, because of no alloying
effect between ad-atom and oxygen atom.
In the present paper, we focus on the surfactant effect of
the oxygen gas adsorption system on the hetero-epitaxial ultra thin film whose constituents have the fcc structure. At
first, we report the surfactant effect of oxygen on fcc-Co, γ -Fe
and fcc-Cr epitaxial thin films on Cu(001). Then the epitaxial
growths of Co/γ -Fe/Co trilayer is investigated.
It is necessary for improvement of MR ratio of multilayer
to be investigated on the surfactant effect of oxygen atoms. In
the past study, it was reported that Pb atoms deposited on the
Cu(001) surface act as surfactant in the case of [Co/Cu] multilayer growth.12) However, there is a problem that Pb atoms
were left at interfaces between Co and Cu layers. So we
finally investigate surfactant effect of oxygen in the [Co/Cu]
2. Experimental Procedure
The ultra thin films and multilayers were prepared by the
molecular beam epitaxy (MBE) method. Reflection high
energy electron diffraction (RHEED) and X-ray diffraction
(XRD) were used for the structural analysis. The magnetization was measured by the conventional vibrating sample
magnetometer (VSM). The single crystals of Cu(001) and
MgO(001) were used as the substrates. The procedure of the
surface polishing of a Cu(001) single crystal substrate was
mechanical polishing by the diamond paste, electrical polishing and then AC sputtering at 20 W. Then the substrate was
annealed at 643 K for 5 h in the MBE chamber. In the case
of MgO(001) substrate, the 100 nm Cu buffer layer was deposited on the clean surface of a commercial MgO(001) crystal. After pumping up to an ultra high vacuum of 1.5×10−9 Pa
M. Onishi et al.
Fig. 1 RHEED pattern of the Cu(001)–O(2 2 × 2)R45◦ reconstruction surface after exposed to oxygen 1800 L at 500
az√ K (
imuth) (a) and a real space atomic arrangement of a missing row model for the reconstruction structure of Cu(001)–O(2 2× 2)R45◦
Fig. 2 RHEED pattern of the Co(001)–O( 2 × 2)R45◦ reconstruction
√ after 45 ML growth ( azimuth) (a) and a real space
atomic arrangement for the reconstruction structure of Co(001)–O( 2 × 2)R45◦ (b).
in the chamber, the sample was exposed to oxygen for about
20 min at a pressure of 2 × 10−4 Pa at 500 K.17) It corresponds
to 1800 L (1 Langmuir (L) = 1.33 × 10−4 Pa·s), which is
sufficient to cover the Cu(001) surface by 0.5 ML oxygen
The reconstruction structure Cu(001)–O(2 2 × 2)R45◦
reported by Yata et al.17) was confirmed by RHEED patterns. The deposition rates for Fe, Co, Cu and Cr were 0.14–
0.4 nm/min. The substrate temperature was 340 K. The final
Cu cap layer was deposited for protection from oxidization.
The base pressure for deposition was 1.5×10−9 Pa. The magnetization and the MR ratio were measured by VSM and by
the conventional four terminals method, respectively.
3. Results and Discussion
3.1 Co/Cu(001) and γ -Fe/Cu(001)
At First, the surfactant effect of oxygen atoms on equilibrium Co on Cu(001) substrate was investigated. Figure 1(a)
shows RHEED pattern of Cu(001) surface after 1800 L exposure by oxygen gas. The three streaks between the fundamental fcc Cu streaks are observed. The streaks indicate√
√ adsorbed reconstruction structure of Cu(001)–
O(2 2 × 2)R45◦ . The surface structure in real space is
shown in Fig. 1(b), where 0.5 ML oxygen atoms are constituting the reconstruction structure with Cu atoms (a missing row
model). Then, Co atoms were deposited on the reconstruction structure. Figure 2 shows a RHEED pattern (a) of the Co
√ and an atom arrangement model (b) of Co(001)–
O( 2× 2)R45◦ structure in the real space. The coverage of
oxygen for this structure corresponds to 0.5 ML,
√ is the
same as the oxygen coverage of Cu(001)–O(2 2 × 2)R45◦
reconstruction surface. Accordingly, it is concluded that all
the oxygen atoms on the Cu(001)–O surface floated up to the
surface of Co layer and the fcc structure of Co layer was stabilized by the oxygen atoms at room temperature. The maximum thickness of fcc Co layer checked in the present experiment was up to 100 ML (1 ML = 0.177 nm).
The result of the surfactant phenomenon of oxygen atoms
for γ -Fe layer grown on Cu(001)–O was same to that for the
present Co/Cu(001) system as reported already in Ref. 16),
where the fcc 45 ML γ -Fe was obtained but the structural
transformation to α-Fe took place after the deposition of
55 ML Fe. The RHEED pattern over 45 ML became gradually spotty and the in-plane lattice constant was increasing.
The result suggests that the martensite transformation takes
place partially in the film or proceeds gradually at the surface. The distinguished difference between the growths of
γ -Fe atoms on a clean Cu(001) surface and on a Cu–O surface was in the difference of surface reconstruction structures.
In the former case, the n × 1 reconstruction structure19) was
observed (the maximum γ -Fe thickness is 11 ML), indicating
the movement of Fe atoms along  direction from the face
centered symmetry at the surface.
√ latter case, the recon√ In the
struction structure of γ -Fe–O( 2 × 2)R45◦ was observed,
which shows no movement of Fe atoms. It should be noted
Surfactant Effect of Oxygen Atoms on Epitaxial Growth of fcc Ultra Thin Films on Cu(001)
Fig. 3 RHEED patterns for surface after 3 ML growth of Cr on the clean surface of Cu(001) (a) and the Cu(001)–O(2 2 × 2)R45◦
reconstruction surface (b).
that the different maximum thickness of γ -Fe in both cases
is attributed to the different surface reconstruction structures.
That is, the movement of Fe atoms to the stable α-Fe state is
Then we investigated critical thickness of γ -Fe deposited
on the Cu–O reconstruction structure formed on the Cu buffer
(100 nm) layer deposited on MgO(001). The thickness of
γ -Fe was 20 ML which is much smaller than the γ -Fe critical
thickness (45 ML) deposited on the same Cu–O reconstructed
surface formed on the Cu single crystal. It is suggested by the
experiment that the decrease of critical thickness was caused
by the difference of the surface roughness between the Cu single crystal and Cu buffer layer deposited on the commercial
MgO crystal, whose surface flatness is not well defined.
Next, the surfactant effect of oxygen on the growth of nonequilibrium fcc-Cr on Cu was investigated. The RHEED patterns for growths of Cr on a clean Cu surface and on Cu–O
reconstructed structure (see Fig. 1(a)) are shown in Fig. 3. In
the case of the deposition on the clean Cu(001) surface, the
fcc-Cr was grown up to 3 ML (Fig. 3(a)) and the structural
transformation to bcc-Cr took place after the deposition of
4 ML Cr. Figure 3(a) shows the co-existence of spotty streaks
and Kikuchi lines. These streaks were changed into spots by
the structural transformation to bcc-Cr. On the other hand,
in the case of the use of the oxygen atoms as surfactants,
√ √ intensity of super lattice lines from the Cu(001)–
O(2 2× 2)R45◦ reconstruction structure became gradually
weakened from the beginning of the growth. The super lattice
lines became invisible over 3 ML (Fig. 3(b)) and then the fundamental fcc streaks were changed into bcc spots. This result
means that the transformation to bcc structure was taken place
around 4 ML thickness when oxygen atoms were used. Thus,
it is concluded that the surfactant effect was not observed in
the case of fcc-Cr on Cu–O reconstructed structure.
The reason of no surfactant effect in the fcc-Cr case can be
explained in an atomic model shown in Fig. 4. The relation
between ad-atoms and oxygen atoms is drawn in the figure.
The oxygen atoms should float on the surface in order to decrease the surface energy (Fig. 4(a)). Nevertheless, the oxygen atoms are left behind in the under layer, if the adhesive
energy between oxygen atom and ad-atom is stronger than the
surface energy of the ad-atom layer (Fig. 4(b)). In Table 1,
: oxygen atom
: bond between oxygen
atom and ad-atom
: bond between ad-atom
Fig. 4 Model of oxygen atoms rising over the surface (a) and those left
behind in the under layer (b). The solid lines and doted line show bonds
between metal-oxygen atoms and metal-metal atoms bonds, respectively.
(a): the oxygen atoms float on the surface in order to decrease the surface
energy. (b): the oxygen atoms are left behind in the under layer, if the
adhesive energy between oxygen atom and ad-atom is stronger than the
Table 1 Binding energy between metal atom and oxygen (mole energy,
Q/J·mol−1 × 10−5 ).20)
where the chemical binding energies between oxygen atom
and various ad-atoms are listed,20) the binding energy of Cr
is larger than those of Fe, Co and Cu. Though the surfactant
effect relates to the physical cohesion, it is possible that the
tendency of the adhesive energy might be proportional to the
binding energy in Table 1.
3.3 Co/γ -Fe/Co/Cu/MgO(001)
According to the theoretical prediction, the [γ -Fe/Co] multilayer should show the largest MR ratio in the [TM/Co]
multilayer (TM = non magnetic 3d transition metal).6) In the
present study, the MgO(001) single crystal was used as a substrate, which is insulator to measure
√the MR ratio. In or√
der to obtain the Cu(001)–O(2 2 × 2)R45◦ surface structure, a Cu buffer layer (100 nm) was deposited on MgO(001),
the thickness of which was required to obtain the flat buffer
surface. Then, the buffer surface was exposed to the oxygen gas until 1800 L as mentioned above. In our previ-
M. Onishi et al.
Fig. 5 RHEED pattern with  azimuth of the Cu(001)–O(2 2× 2)R45◦ reconstruction
√structure of buffer layer surface after being
exposed to oxygen (a) and RHEED patterns with  azimuth of the Co(001)–O( 2 × 2)R45◦ surface for Co 1st layer (b), γ -Fe
layer (c) and Co 2nd layer (d).
ous investigation on the Co/γ -Fe/Co trilayer7) deposited on
a clean Cu(001) surface, it was reported that the γ -Fe in
the Co/γ -Fe/Co trilayer was so unstable that it could not
be grown over 7 ML (= 1.26 nm) in the case of using Cu
buffer layer (100 nm) deposited on MgO(001). We prepared a
Co(1 nm)/γ -Fe(1.8 nm)/Co(1 nm) trilayer on the Cu(100 nm)
buffer deposited on MgO(001) substrate, utilizing the surfactant effect of oxygen atoms for stabilize of Co/γ -Fe/Co trilayer. In Fig. 5, the RHEED patterns during deposition of
Co/γ -Fe/Co on Cu buffer layer are shown. The RHEED pattern for Cu surface
indicates the reconstruction structure of
Cu(001)–O(2 2 × 2)R45◦ (Fig. 5(a)). Every surfaces of
1st Co layer,
√ γ -Fe√layer, and 2nd Co layer indicate the Co (or
γ -Fe)–O ( 2 × 2)R45◦ reconstruction structure as shown
in Figs. 5(b), (c) and (d), respectively. This results are proofs
of surfactant effect of oxygen atoms in the deposition process
of Co/γ -Fe/Co.
The critical thickness of γ -Fe layer was 18 ML in the
present Co/γ -Fe/Co trilayer, which is almost the same value
to the γ -Fe thickness (20 ML) deposited on the Cu–O reconstructed surface using MgO crystal as mentioned in §3.1.
Now, in the present experiment, the surfactant effect of
oxygen atoms on the growth of the trilayer was confirmed.
The MR ratio of the trilayer sample was measured, but the
value of MR ratio was less than 1%. In the further investigation on the trilayer optimizations of non-magnetic γ -Fe
thickness, of the crystal orientation and of the buffer layer
thickness are needed for large MR ratio.
3.4 [Co/Cu]20 /MgO(001)
The surfactant effect in the preparation process
2)R45◦ surface was investigated. No investigation on the
surfactant effect of oxygen atoms for multilayer was reported
so far. In the present study, the MgO(001) single crystal was
used as a substrate. Figures 6(a) and (b) show a RHEED pattern of the 20th layered Co surface of [Co/Cu] multilayer and
the intensity profile for line A to B in Fig.
The profile indicates that the Co(001)–O( 2 × 2)R45◦ surface structure could be confirmed even on the last Co layer
surface in the multilayer though the intensity of the super lattice lines was gradually weakened as the superposition was
increased in the multilayer. Now it should be notified that the
oxygen surfactant effect was observed in the [Co/Cu] multilayer. Table 2 shows the results for the magnetization measurement and the X-ray diffraction. The saturation magnetization (Ms ), coercive force (Hc ) and artificial periodicity (D)
for the sample prepared by the oxygen surfactant effect are
almost same to the values of the sample made on a clean
Cu(001) buffer surface. Accordingly, the result suggests that
there is no oxidation effect on Co and Cu layers during deposition and most of oxygen atoms float on both Co and Cu
layers. However, the MR ratio of the multilayer was less than
1%. The reasons for the small MR ratio are considered as that
the Cu buffer layer was too thick to obtain the large MR ratio, which plays as a bypass of electric current and that the Cu
spacer layer thickness was not optimized in order to obtain the
antiferromagnetic coupling in the  direction. The investigations on these problems are necessary in future. It should
be noted that the present study is the first observation for the
clear surfactant effect of oxygen atoms on the “multilayer”
Surfactant Effect of Oxygen Atoms on Epitaxial Growth of fcc Ultra Thin Films on Cu(001)
Fig. 6 RHEED pattern of the Co layer surface of the 20th layer in [Co/Cu] multilayer (a) √
intensity profile recorded along
the A-B line in the figure (b). Super lattice lines in figure (b) are associated with Co–O( 2 × 2)R45◦ reconstruction structure.
Table 2 Magnetic properties and artificial periodicity of [Co/Cu] multilayers deposited on a clean Cu buffer surface and on a Cu(001)–O buffer
of Cu (001)
Ms /Wb·m·kg−1 × 10−4
1.83 ± 0.05
1.81 ± 0.05
Hc /A·m−1 × 104
Surfactant effect of adsorbed oxygen atoms on the growth
of fcc type hetero-epitaxial thin films were investigated by using MBE method. In the growths of Co, γ -Fe, Co/γ -Fe/Co
trilayer and [Co/Cu] multilayer,
√ the surface reconstruction
structure of Cu(001)–O(2 2 × 2)R45◦ is necessary for the
oxygen surfactant effect. The γ -Fe was stabilized until 45 ML
on the Cu(001)–O surface formed on Cu(001) single crystal,
but 20 ML on the Cu–O surface formed on Cu(001) buffer
layer on MgO(001). The surfactant effect for Co/γ -Fe/Co trilayer was observed. It should be noted that the present experiment is the first observation of the surfactant effect of oxygen
atoms on the growth of [Co/Cu]20 “multilayer”. In contrast,
the surfactant effect was not observed for the fcc-Cr because
of the large adhesive energy between Cr and oxygen atoms.
1) E. Efullerton, D. M. Kelly, J. Guimpel and I. K. Schuller: Phys. Rev.
Lett. 68 (1992) 859–862.
2) K. Takahashi, Y. Obi, Y. Mitani and H. Fujimori: J. Phys. Soc. Jpn. 61
3) K. Inomata: Jpn. Soc. Appl. Phys. 63 (1994) 1198–1209.
4) Y. Kamada: Ph. D. Thesis, Nagoya University (1997).
5) R. Schad, C. D. Potter, P. Belien, G. Verbanck, V. V. Moschchalkov and
Y. Bruyseraede: Appl. Phys. Lett. 64 (1994) 3500–3502.
6) H. Itoh, J. Inoue and S. Maekawa: Phys. Rev. B47 (1993) 5809–5818.
7) Y. Yamada, D. Suetsugu, B. Sadeh, M. Doi and M. Matsui: J. Magn.
Soc. Jpn. 22 (1998) 609–612.
8) W. L. O’Brien and B. P. Tonner: Phys. Rev. B52 (1995) 15332–15340.
9) A. Schatz and W. Keune: Surf. Sci. 440 (1999) L841–847.
10) J. Tomassen, B. Feldmann and M. Wuttig: Surf. Sci. 264 (1992) 406–
11) A. Kida: Ph. D. Thesis, Nagoya University (1996).
12) M. Kamiko, R. Furukawa, K.-Y. Kim, M. Iwatami and R. Yamamoto:
J. Magn. Magn. Mater. 198–199 (1999) 716–718.
13) J. Camarero, T. Graf, J. J. de Miguel and R. Miranda: Phys. Rev. Lett.
76 (1996) 4428–4431.
14) A. Kirilyuk, J. Giergiel, J. Shen and J. Kilschner: Phys. Rev. B52 (1995)
15) A. Kirilyuk, J. Giergiel, J. Shen, M. Staub and J. Kilschner: Phys. Rev.
B54 (1996) 1050–1063.
16) L. Li, A. Kida, M. Onishi and M. Matsui: Surf. Sci. 493 (2001) 120–
17) M. Yata, H. Rouch and K. Nakamura: Phys. Rev. B 56 (1997) 10579–
18) M. Wuttig, R. Franchy and H. Ibach: Surf. Sci. 213 (1989) 103–136.
19) A. Kida and M. Matsui: J. Phys. Soc. Jpn. 65 (1996) 1409–1412.
20) CRC Hand book of Chemistry and Physics, 80th Edition 9–51 (1999–