RESEARCH ARTICLE | NOVEMBER 21 2022

Impact of interfaces on magnetic properties of Gdx(Fe90Co10)1−x

alloys 
Jean-Loïs Bello ; Daniel Lacour ; Sylvie Migot; Jaafar Ghanbaja; Stéphane Mangin ; Michel Hehn 

Appl. Phys. Lett. 121, 212402 (2022)

https://doi.org/10.1063/5.0125011

CrossMark

 
View Export
Online Citation

Articles You May Be Interested In

Spin-orbit torques in ferrimagnetic GdFeCo alloys

Appl. Phys. Lett. (September 2016)

Magnetoelectric response of AlN/[(Fe90Co10)78Si12B10 + Terfenol-D] composite films

J. Appl. Phys. (January 2014)

Improvement of soft magnetism of Fe90Co10 sputtered films by addition of N and Ta

Journal of Applied Physics (April 1996)

17 October 2023 09:47:40

 Applied Physics Letters ARTICLE scitation.org/journal/apl

Impact of interfaces on magnetic properties

of Gdx(Fe90Co10)1x alloys
Cite as: Appl. Phys. Lett. 121, 212402 (2022); doi: 10.1063/5.0125011
Submitted: 9 September 2022 . Accepted: 7 November 2022 .
Published Online: 21 November 2022

Jean-Lo€ıs Bello, Daniel Lacour,  phane Mangin,

Sylvie Migot, Jaafar Ghanbaja, Ste and Michel Hehna)

AFFILIATIONS
 de Lorraine, CNRS, IJL, F-54000 Nancy, France
Universite

a)
Author to whom correspondence should be addressed: michel.hehn@univ-lorraine.fr

ABSTRACT
A 5 nm thick ferrimagnetic film made of amorphous rare-earth transition-metal alloys Gdx(Fe90Co10)1-x was grown by physical vapor deposi-
tion. Its magnetic properties (coercivity, perpendicular magnetic anisotropy, and compensation composition at room temperature) were
investigated for various buffer and capping layers in contact with a ferrimagnetic thin film. While Gdx(Fe90Co10)1-x appears to be amorphous
for all the samples, it appears that (111) textured Cu is the best material to promote perpendicular magnetization. The large compensation
composition change as a function of the magnetic film interface at room temperature is analyzed in terms of polarizability of the surrounding

17 October 2023 09:47:40

buffer and capping materials.
Published under an exclusive license by AIP Publishing. https://doi.org/10.1063/5.0125011

Amorphous rare-earth transition-metal (RE-TM) alloys are a RF Ar plasma at a pressure of 102 mbar for 5 min. During deposition,
very interesting class of materials since their magnetic properties can the sample holder is rotated at several tens of revolutions per minute to
be finely tuned by their chemical composition and temperature.1,2 ensure uniform layer thickness. The 5 nm thick Gdx(Fe90Co10)1x layer
They show large magneto-optical and spintronic effects and are useful was co-sputtered from highly pure Gd, Fe, and Co targets. The compo-
to realize devices to propagate magnetic textures or build THz oscilla- sition was controlled by adjusting the power of each target and was cal-
tors.3 One material of particular interest is the GdFeCo ferrimagnetic ibrated before deposition. The atomic composition was varied
alloy. In this material, the magnetic moment of the rare-earth sublat- systematically in a range of 22% < x < 33%. In all multilayers, a buffer
tice (Gd) is antiferromagnetically exchange coupled to one of the of tantalum (Ta) is sputtered before the magnetic layer promotes layer
transition-metals (FeCo).5 The resulting net magnetization and coer- adhesion, block diffusion of Si and O2 from the substrate, and erase
civity can easily be tuned by changing the temperature or the RE con- any texture that could come from the substrate. Also, a platinum (Pt)
centration.6 Magnetization compensation, i.e., zero net magnetization, film is added over as a capping layer to prevent against oxidation. Y
can be achieved for specific composition or temperature. and Z have been varied by choosing Cu (copper), Pt, Ta, or Ir (irid-
Recently, this material has attracted much interest thanks to its ium). For each sample, we measured the magnetic hysteresis loop at
light and current induced switching properties.4,7,8 While the bulk room temperature with a Kerr laser setup in the polar geometry. The
properties have been largely reported in the literature,6 its recent use as effective magnetic anisotropy will result from the competition between
ultrathin films led to a variability of properties: variability of composi- the magnetic anisotropies and the shape anisotropy related to the
tion for room temperature compensation or in-plane/out-of-plane demagnetization field of the thin film. From the shape of the loop, the
magnetization stabilization for a given chemical composition. In this direction of the effective magnetic anisotropy is extracted: out of
paper, we study those properties, keeping the thickness of GdFeCo the plane (OOP) if the loop is square with hysteresis and in the plane
fixed to 5 nm and changing both the concentration of Gd and the (IP) if the loop is linear. From the loops, we also determine the coercive
materials in contact with GdFeCo. We show that the variability of fields as well as the saturation fields. The former is expected to diverge
properties can be related to the buffer and capping layers used in the at the magnetization compensation. Since our Kerr setup mainly
multilayer while GdFeCo keeps its amorphous state. probes the TM sub-lattice, it is also straightforward to deduce which
SiO2//Ta(5 nm)/Y(ynm)/Gdx(Fe90Co10)1x(5 nm)/Z(znm)/Pt(5 nm) sub-lattice dominates the magnetization of the sample as a function
heterostructures were grown by magnetron sputtering. Before the mul- of x. Finally, due to the low saturation magnetization and reduced
tilayer is deposited, the surface of the substrate surface is etched by a thickness of the Gdx(Fe90Co10)1-x alloy thin film, the magnetic moment

Appl. Phys. Lett. 121, 212402 (2022); doi: 10.1063/5.0125011 121, 212402-1
Published under an exclusive license by AIP Publishing
 Applied Physics Letters ARTICLE scitation.org/journal/apl

and so the magnetic anisotropies could not be quantified, and the mag- those magnetic response is the low anisotropy of the magnetic layer
netic signal of the substrate is too large. and the presence of DMI (Dzyaloshinskii–Moriya Interaction) at the
Figure 1(a) reports the evolution of the coercive field of SiO2// Gdx(Fe90Co10)1-x/Pt interface. Inserting Ta at this interface changes
Ta(5 nm)/Y(5 nm)/Gdx(Fe90Co10)1-x(5 nm)/Z(5 nm)/Pt(5 nm) multi- drastically the magnetic response. First, square hysteresis loops with the
layers with various Y and Z interfaces. Only samples that exhibit effec- field applied along the direction perpendicular to the film could be
tive perpendicular magnetic anisotropy (PMA) are reported, and observed for two compositions on both sides of the compensation.
samples with in-plane magnetization in the range 22% < x < 33% are Second, the composition for compensation moved from 29.2% to 24%
not shown in the graph. The vertical lines represent the change (Fig. 1).
between CoFe-rich samples and Gd-rich composition, respectively, on Adding Cu at the Gdx(Fe90Co10)1-x interfaces has a major impact
the left and right sides of the lines, for each couple of interfaces. The on the magnetic properties. Comparing to two Ta interfaces, inserting
analysis of the Kerr hysteresis loops provides the compositional ranges Cu either at the lower interface [Fig. 1(a), orange circle], at the upper
in which the alloy is TM or RE dominant. A visual representation of interface [Fig. 1(a), blue circle], or at both interfaces [Fig. 1(a), red cir-
TM or RE dominance is reported in Fig. 1(b). An obvious and clear cle] improves the PMA. In the latter case, PMA is maintained for a
correlation with the coercive field is observed. very wide range of composition from x ¼ 20.2% to x ¼ 30.4%.
Starting with SiO2//Ta(5 nm)/Gdx(Fe90Co10)1-x(5 nm)/Pt(5 nm), Removing one interface leads to a reduction in the window of x for
no PMA could be demonstrated for x < 24.4% or x > 29.2% since only which PMA exists. As a result, Cu appears to be the best candidate to
hysteresis loops’ characteristics of in plane effective magnetic anisot- promote PMA. By adding one or two Cu interfaces, we can clearly
ropy could be observed (Fig. 2). For 24.4% < x < 29.2%, a curvy loops observe that the concentration for magnetic compensation shifts to a
characteristic of multi-domains’ states with PMA is observed. At higher concentration of Gd.
x ¼ 29.2%, the sign of the loop changes indicates that the composition For all other samples studied in this paper, a Cu interface is
of magnetic compensation has been passed. A possible explanation of retained to promote PMA. The other interface was modified to check

17 October 2023 09:47:40

FIG. 1. (a) Coercive field as a function
of the Gd concentration for 5 nm thick
GdFeCo layers with different interfaces.
(b) Summary of the magnetic behavior
as a function of the bottom interface
material (left column) and top interface
material (right column): red—CoFe domi-
nant, green—Gd dominant, white in
plane. The blue points are the calculated
interface magnetic polarizations consid-
ering both interfaces.

Appl. Phys. Lett. 121, 212402 (2022); doi: 10.1063/5.0125011 121, 212402-2
Published under an exclusive license by AIP Publishing
 Applied Physics Letters ARTICLE scitation.org/journal/apl

interface, the induced magnetic moment on the Pt atoms has been

estimated to 0.3 lB,15 and at the Co/Pt interface, 0.14 lB per Pt atoms
have been reported.16,17 The same applies for TM/Ir interfaces where
moment is also induced. The polarization of the Ir interface atoms is
estimated at 0.3 lB with Co18 and 0.2 lB with Fe.19 Finally, the polari-
zation of the d-band of the Cu atoms is also observed either in Co/Cu
multilayers of 0.05 lB20,21 or in Fe/Cu multilayers up to 0.09 lB.21
The induced moments in FeCu alloys and in CoCu alloys are similar
and around 0.13 lB.20–22
Considering that the TM composition of the studied ferrimagnet
is roughly 90% of Fe and 10% of Co, we can estimate the polarization
of Cu, Ir, and Pt by the FeCo sub-lattice from the literature. The esti-
mations are reported in Fig. 1(b), blue color circle. The evolution of
the compensation composition follows the evolution of the magnetic
induced moment. Indeed, when Pt or Ir is directly on top of the
GdFeCo grown on a Cu layer, the induced moments are maximum
among our studied stacks; consequently, additional moment parallel
to the FeCo moment has to be taken into account. It is then necessary
to add more Gd (x  29%) compared to the other stacks to reach the
magnetic compensation point. For Ta/Ir/GdFeCo/Cu/Pt, the compen-
sation is at x  26.6%. This can be explained by a lower intermixing
inducing a lower moment at the Ir/GdFeCo interface than at the
GdFeCo/Ir interface (see EELS spectra of both multilayers presented
in Fig. 3). Since the induced polarization in Cu is much lower, the
amount of Gd to be added is also much lower. Finally, for Ta interfa-
ces, no induced polarization was reported, which explains why less Gd

17 October 2023 09:47:40

is needed (x  23.9%) to achieve compensation. On the contrary, a
CoFeB dead layer has been reported in the literature in Ta/CoFeB bi-
layers.24 Considering that Gd is not magnetic at room temperature
FIG. 2. Hysteresis loops for Ta/GdFeCo/Pt samples at different compositions mea-
and gets magnetic in contact with the CoFe subnetwork, the compen-
sured by Kerr magnetometry.
sation composition will not be dependent on this interfacial ferrimag-
netic dead layer.
its impact on the magnetic properties. Thus, we tested the insertion of The next step is to elucidate the origin of the magnetic anisotropy
Pt or Ir. For example, the use of Cu is mandatory for the Ta/GdFeCo/ in amorphous GdFeCo ferrimagnetics and to explain why, for the
Pt stacking: except for x below 22.2%, all samples have PMA [Fig. 1(a), same chemical composition, some multilayers show in-plane magneti-
light green circle]. Replacing Pt with Ir either at the lower interface zation and others out-of-plane.
[Fig. 1(a), purple circle] or at the upper interface [Fig. 1(a), green cir- Obviously, as we have seen in the previous paragraph, the mag-
cle] induces enough PMA to maintain out-of-plane magnetization. netic polarization of the interfaces implies an increase in the saturation
Furthermore, we observe that the interfaces not only play a role magnetization and, therefore, the demagnetization field. This explains
on the effective anisotropy but also on the composition for which the why among two samples both having a Cu interface, Ta/Cu/
magnetic compensation occurs. In this study, the shift of composition
of compensation goes up to 6%. Considering the compensation com-
position with the lowest concentration of Gd (Ta/GdFeCo/Ta), we first
notice the need to add more Gd to obtain compensation when Cu is
added on top of a Ta buffer. The same behavior is observed for Cu on
top of the ferrimagnet. Second, the maximum shift occurs with a copper
seed layer and when GdFeCo is directly in contact with Ir and/or Pt.
In the following, we will see why Cu plays a major role in multi-
layer stacks to obtain PMA and why the composition of the magnetic
compensation is modified by the materials in contact with GdFeCo.
Because the exchange interaction between conduction electrons is
large in Pt, it can exhibit giant magnetic moment induced by magnetic
impurities.9 The proximity of ferromagnetic layers is known to induce
indirect ferromagnetic order due to long range oscillatory magnetic
exchange interaction called as the Ruderman–Kittel–Kasuya–Yosida
(RKKY) interaction. This induces magnetization in Pt either through FIG. 3. STEM-EELS elemental maps on Cu/GdFeCo/Ir multilayer (top) and on Ir/
the Co/Pt interfaces or through the Fe/Pt interfaces.10–14 At the Fe/Pt GdFeCo/Cu multilayer. Purple: Ir; yellow: Gd; red: Fe; green: Co; blue: Cu.

Appl. Phys. Lett. 121, 212402 (2022); doi: 10.1063/5.0125011 121, 212402-3
Published under an exclusive license by AIP Publishing
 Applied Physics Letters ARTICLE scitation.org/journal/apl

Gd22.2FeCo/Pt is magnetized in-plane while Ta/Gd22.2FeCo/Cu/Pt is

magnetized out of plane. However, within this framework, we cannot
explain other major changes, especially in the case of Ta at each inter-
face where the PMA is particularly small. In the Ta/CoFeB/MgO sys-
tem, the anisotropy was then correlated with the CoFeB dead layer at
the Ta interface. We again must assume that the interfaces play a
major role in the anisotropy.
It is well known that after a few nanometers, copper as well as
iridium acquire a (111) texture when deposited over an amorphous Ta
layer.23 It is also known that the copper texture has an impact on the
perpendicular anisotropy in Fe and Co thin layers.25,26 This led us to
study the magnetic properties as a function of the Cu thickness.
Hysteresis loops of Ta(5 nm)/Cu(y nm)/Gd23.4FeCo(5 nm)/Pt(5 nm)
with y varying from 1 nm to 4 nm are reported in Fig. 4. As can be
seen, the Cu thickness has a direct impact on the PMA. For 1 nm of
Cu between the Ta and the GdFeCo layers, the magnetization lies FIG. 5. HRTEM micrograph of Ta/Cu/Gd24.3FeCo/Ir/Pt with the FFT of the GdFeCo
clearly in the sample plane with a saturation field around 3200 Oe. For and Cu layers.
2 nm, the magnetization remains in-plane but the saturation field is
now reduced to about 2400 Oe. At 3 nm Cu, the cycle shows a slight with high-resolution transmission electron microscopy (HRTEM) pre-
opening, and the saturation field drops to 1200 Oe. At 4 nm, the mag- sented in Fig. 5. The amorphous characteristic of both Ta and GdFeCo
netization is perpendicular to the plane. The field-dependent magneti- layers is confirmed. The Fourier transform (FFT) of the GdFeCo layer
zation cycle is well open. Its shape is typical of the appearance of shows a diffuse ring pattern characteristic of an amorphous structure.
magnetic domains such as bubbles or skyrmions at a low field. As pre- The Cu layer appears as being textured with a growth direction along
sented previously, for 5 nm of Cu, the magnetization is perpendicular [111]. The chemical intermixing and roughness seem to be weak at the
to the plane at all fields. So, while Cu certainly acquires a strong (111) Cu/GdFeCo interface. We also investigated the reverse order stack
texture when the thickness increases, the perpendicular saturation field Sub/Ta/Ir/Gd24.3FeCo/Cu/Pt (not shown). GdFeCo is also clearly

17 October 2023 09:47:40

decreases and PMA increases. Nevertheless, at this level, the impact on amorphous as shown by the FFT that displays again a diffuse ring. As
the crystalline properties GdFeCo (reported as amorphous) was still a result, in both samples, no induced texture could be observed in
unknown. Therefore, a local analysis was required to further investi- GdFeCo. On the other hand, the Cu layer, either under or on top of
gate the details of the local crystallographic structure. GdFeCo, is (111) textured, and the interface with GdFeCo is straight
Transmission electron microscopy (TEM) was carried out without inter-mixing. In comparison to the other materials studied, Ir
(Fig. 5). Samples were prepared by focused ion beam (FIB) to mill a also acquires a (111) texture but the interfaces are more intermixed
cross section. We first observed a Sub/Ta/Cu/Gd24.3FeCo/Ir/Pt sample and Ta is fully amorphous. Since there is no visible change in the
GdFeCo layer, texture and interface mixing appear to be the key ingre-
dients of PMA. From the TEM images, it was, however, not possible to
highlight any texture that could be induced at the GdFeCo interfaces
and gradually lost away from the interfaces, but such texture could
explain the origin of interfacial anisotropy.
In conclusion, in thin our ultra-thin amorphous rare-earth tran-
sition-metal alloys Gdx(Fe90Co10)1-x, both compensation composition
and anisotropy are found to be strongly depend on the interfaces.
While Gdx(Fe90Co10)1-x appears to be amorphous in all combinations
of materials and independently on x, (111) textured Cu layers emerge
as the best material to promote PMA. In addition, we have shown that
the composition for room temperature compensation depends on the
magnetic polarizability of the materials at the interface.

This work was supported partly by the French PIA Project

“Lorraine Universite d’Excellence,” Reference No. ANR-15-IDEX-
04-LUE, by the “FEDER-FSE Lorraine et Massif Vosges
2014–2020,” a European Union Program, and by the Grand Est
Region and ICEEL.

AUTHOR DECLARATIONS
Conflict of Interest
FIG. 4. Kerr hysteresis loops for Ta(5 nm)/Cu(y nm)/Gd23.4FeCo(5 nm)/Pt(5 nm)
with 1 nm <¼ y <¼ 5 nm. The authors have no conflicts to disclose.

Appl. Phys. Lett. 121, 212402 (2022); doi: 10.1063/5.0125011 121, 212402-4
Published under an exclusive license by AIP Publishing
 Applied Physics Letters ARTICLE scitation.org/journal/apl

8
Author Contributions D. Cespedes-Berrocal, H. Damas, S. Petit-Watelot, D. Maccariello, P. Tang, A.
Arriola-Cordova, P. Vallobra, Y. Xu, J. L. Bello, E. Martin, S. Migot, J.
Jean-Lo€ıs Bello: Formal analysis (equal); Investigation (equal); Ghanbaja, S. Zhang, M. Hehn, S. Mangin, C. Panagopoulos, V. Cros, A. Fert,
Methodology (equal); Validation (equal). Daniel Lacour: Formal anal- and J. C. Rojas-Sanchez, Adv. Mater. 33, 2007047 (2021).
9
ysis (equal); Investigation (equal); Writing – review & editing (equal). A. I. Larkin and V. I. Melnikov, Sov. Phys. JETP. 34, 656 (1972).
10
Sylvie Migot: Formal analysis (equal); Investigation (equal). Jaafar S. S. Parkin, Phys. Rev. Lett. 67, 3598 (1991).
11
Ghanbaja: Formal analysis (equal); Investigation (equal). Stephane K. Le Dang, P. Veillet, C. Chappert, R. Farrow, R. Marks, D. Weller, and A.
Cebollada, Phys. Rev. B. 50, 200 (1994).
Mangin: Funding acquisition (equal); Project administration (equal); 12
A. Simopoulos, E. Devlin, A. Kostikas, A. Jankowski, M. Croft, and T.
Supervision (equal). Michel Hehn: Conceptualization (equal); Formal Tsakalakos, Phys. Rev. B. 54, 9931 (1996).
13
analysis (equal); Investigation (equal); Methodology (equal); J. Knepper and F. Yang, Phys. Rev. B. 71, 224403 (2005).
14
Supervision (equal); Validation (equal); Writing – original draft (equal). F. Meier, S. Lounis, J. Wiebe, L. Zhou, S. Heers, P. Mavropoulos, P. H.
Dederichs, S. Bl€ugel, and R. Wiesendanger, Phys. Rev. B. 83, 075407 (2011).
15
R. Wu, L. Chen, and N. Kioussis, J. Appl. Phys. 79, 4783 (1996).
16
DATA AVAILABILITY T. McGuire, J. Aboaf, and E. Klokholm, J. Appl. Phys. 55, 1951 (1984).
17
A. Mukhopadhyay, S. K. Vayalil, D. Graulich, I. Ahamed, S. Francoual, A.
The data that support the findings of this study are available Kashyap, T. Kuschel, and P. A. Kumar, Phys. Rev. B. 102, 144435 (2020).
18
from the corresponding author upon reasonable request. M. Perini, S. Meyer, B. Dupe, S. V. Malottki, A. Kubetzka, K. V. Bergmann, R.
Wiesendanger, and S. Heinze, Phys. Rev. B. 97, 184425 (2018).
19
F. Wilhelm, P. Poulopoulos, H. Wende, A. Scherz, K. Baberschke, M.
REFERENCES
Angelakeris, N. Flevaris, and A. Rogalev, Phys. Rev. Lett. 87, 207202
1
P. Hansen, S. Klahn, C. Clausen, G. Much, and K. Witter, J. Appl. Phys. 69, (2001).
3194 (1991). 20
M. Samant, J. St€ohr, S. Parkin, G. Held, B. Hermsmeier, F. Herman, M. van
2
P. Hansen, C. Clausen, G. Much, M. Rosenkranz, and K. Witter, J. Appl. Phys. Schilfgaarde, L.-C. Duda, D. Mancini, N. Wassdahl et al., Phys. Rev. Lett. 72,
66, 756 (1989). 1112 (1994).
3 21
B. Dieny, I. L. Prejbeanu, K. Garello, P. Gambardella, P. Freitas, R. Lehndorff, G. Held, M. Samant, J. St€ ohr, B. Hermsmeier, M. Schilfgaarde, and R.
W. Raberg, U. Ebels, S. O. Demokritov, J. Akerman et al., Nat. Electron. 3, 446 Nakajimas, Z. Phys. B Condens. Matter. 100, 335 (1997).
22
(2020). W. Kuch, M. Salvietti, X. Gao, M. Klaua, J. Barthel, C. V. Mohan, and J.
4
Y. Xu, M. Deb, G. Malinowski, M. Hehn, W. Zhao, and S. Mangin, Adv. Kirschner, J. Appl. Phys. 83, 7019 (1998).
23
Mater. 29, 1703474 (2017). I. L. Castro, V. P. Nascimento, E. C. Passamani, A. Y. Takeuchi, C. Larica, M.
5
A. Pogorily, E. Shypil, and C. Alexander, J. Magn. Magn. Mater. 286, 493 Tafur, and F. Pelegrini, J. Appl. Phys. 113, 203903 (2013).

17 October 2023 09:47:40

24
(2005). S. Shen, D.-S. Lee, C.-W. Cheng, W.-J. Chan, and G. Chern, IEEE Trans. Mag.
6
P. Hansen, J. Magn. Magn. Mater. 83, 6 (1990). 55, 1 (2019).
7 25
J. Wei, B. Zhang, M. Hehn, W. Zhang, G. Malinowski, Y. Xu, W. Zhao, and S. F. Huang, G. J. Mankey, and R. F. Willis, J. Appl. Phys. 75, 6406 (1994).
26
Mangin, Phys. Rev. Appl. 15, 054065 (2021). M. T. Kief and W. F. Egelhoff, J. Appl. Phys. 73, 6195 (1993).

Appl. Phys. Lett. 121, 212402 (2022); doi: 10.1063/5.0125011 121, 212402-5
Published under an exclusive license by AIP Publishing