PHYSICAL REVIEW MATERIALS 9, 034409 (2025)

Control of ferrimagnetic compensation and perpendicular anisotropy

in Tbx Co(100−x) with H+ ion implantation
*
Robbie G. Hunt , Dmitrii Moldarev , Matías P. Grassi , Daniel Primetzhofer , and Gabriella Andersson
Department of Physics and Astronomy, Uppsala University, Box 516, SE-75120 Uppsala, Sweden

(Received 6 December 2024; accepted 27 February 2025; published 17 March 2025)

The tuning of magnetic properties through electrochemical loading of hydrogen has recently attracted
significant interest as a way to manipulate magnetic devices with electric fields. In this paper we investigate
quantitatively the magneto-ionic effect of hydrogen uptake on the magnetic properties of rare-earth transition-
metal alloy Tbx Co(100−x) in the composition range of x = 10−39 at.% using ion implantation. Using this
technique we are able to link changes in magnetic behavior to exact concentrations of hydrogen, isolated from
the movement of any other ions that would be a factor in electrochemical studies. The composition of the alloy
has been varied alongside the hydrogen dose to characterize the effect of progressive hydrogen loading on the full
range of x displaying out-of-plane magnetic anisotropy. We find large changes in two important properties: the
compensation composition and the Co-rich in-plane to out-of-plane magnetic anisotropy transition composition,
both of which move by 6 at.% toward higher Tb concentrations after hydrogen implantation. This shift in
composition does not increase with a larger dose. From the changes in magnetization we attribute the change in
compensation composition to a significant reduction of the moment on the Tb sublattice.

DOI: 10.1103/PhysRevMaterials.9.034409

I. INTRODUCTION where films on the order of tens of nanometers are modified

with an ionic species, charging times can take on the order of
The development of magnetoelectric materials has been
20 min to fully saturate the magneto-ionic response [16,17].
of significant interest for the development of new devices,
The response time can be improved down to a few millisec-
due to the promise of greater energy efficiency and unique
onds by taking advantage of interface-dependent properties
functionalities [1,2]. Few single-phase magnetoelectric mate-
[such as PMA] that can be modified with much smaller
rials exist, such as the room-temperature multiferroic BiFeO3
concentrations of the ionic species [18–20]. By using much
[3,4], and so a great deal of work is dedicated to indirect
smaller and more mobile lithium or hydrogen ions, it is possi-
means of achieving magnetoelectric coupling. These indi-
ble to reduce this response time down to microseconds while
rect approaches include devices based on interfacial strain,
retaining excellent cyclability commensurate with the needs
both static [5,6] and more recently dynamic [7,8] variations,
of device applications [21,22].
modification of the carrier densities in ferromagnetic semi-
Consideration should also be given to the circumstances
conductors [9] and transitions metals [10], and manipulation
under which experiments have so far been performed. For
via ionic motion [11].
oxygen- and nitrogen-based magneto-ionics much of the work
Magneto-ionic devices have recently attracted significant
done for these makes use of ionic liquids or liquid electrolytes
interest in the field of spintronics as a way to control
to amplify electric fields or source ions, and while this is suf-
magnetic properties through application of a voltage, in-
ficient for a proof of concept experiment it is not practical for
cluding perpendicular magnetic anisotropy (PMA) [12],
industrial application. Lithium has the enormous advantage of
Ruderman-Kittel-Kasuya-Yosida interlayer exchange [13],
well-studied solid-state electrolytes, and has also been used on
Dzyaloshinskii-Moriya interaction [14], and exchange bias
much thicker in-plane films from which it can be understood
[15]. These properties are all extremely relevant to device ap-
that the Li ions have excellent diffusivity in relevant materials
plications but are otherwise difficult to reversibly manipulate
[23]. However, a complicating factor is that lithium is not well
post-fabrication.
suited to complementary metal-oxide semiconductor (CMOS)
A significant downside to magneto-ionic devices is the
processes due to the creation of traps and defects in Si or SiO2
slow response time of ionic elements. In bulklike systems,
material [24].
In these respects, hydrogen-based systems present a good
alternative candidate ion system due to the reported charging
*
Contact author: gabriella.andersson@physics.uu.se times of only a few tens of microseconds, the development of
solid-state electrolytes, the high diffusivity of hydrogen, and
Published by the American Physical Society under the terms of the the large proven cyclability of devices [22].
Creative Commons Attribution 4.0 International license. Further Previous work in the literature using hydrogen as an ionic
distribution of this work must maintain attribution to the author(s) species has found significant success in the manipulation
and the published article’s title, journal citation, and DOI. Funded of both a Pt/Co system exhibiting perpendicular magnetic
by Bibsam. anisotropy [25], the modulation of interlayer RKKY coupling

2475-9953/2025/9(3)/034409(10) 034409-1 Published by the American Physical Society

ROBBIE G. HUNT et al. PHYSICAL REVIEW MATERIALS 9, 034409 (2025)

FIG. 1. Changes in the (a) GIXRD and (b) XRR patterns for a Tbx Co(100−x) sample after an implanted hydrogen dose of 1×1017 ions/cm2 .
Intensities for the implanted sample are offset for clarity.

[26], and the compensation behavior of rare-earth transition- to investigate the changes in PMA strength, compensation
metal (RE-TM) alloy GdCo [22]—demonstrating the wide behavior, and saturation magnetization. We investigate these
range of applicability. In the case of GdCo, there are sig- changes in response to two implanted doses of hydrogen ions.
nificant changes in the magnetic sublattice coupling leading We put forward an explanation for the reduction in Tb atomic
to a large reduction of the compensation temperature upon magnetic moment based on the sperimagnetic nature of these
hydrogen loading, allowing for effective control over the alloys, assuming that the Tb cone is altered by the introduction
net magnetization direction. Recently, in further work using of hydrogen and that the cone angle is increased.
an electrolyte source of hydrogen ions, it was found that
for Tbx Co(100−x) films with compensation temperatures be-
II. SAMPLE GROWTH
low room temperature it is possible to manipulate the PMA
strength to induce a voltage-controlled in-plane to out-of- Samples are grown using a DC magnetron sputtering tech-
plane transition [27]. nique in a UHV chamber with a base pressure of 6×10−10 Torr
However, in electrochemical systems there can be more or better. The thin films are deposited at an Ar pressure of 2.06
than one ionic species that moves in response to an elec- mTorr onto silicon substrates with a native oxide layer. Prior
tric field—including passivated oxygen ions. The polarized to deposition substrates are degassed at 200 °C for 40 min to
neutron reflectometry results of Sheffels et al. [28] on a clean the substrate surface. The TbCo layer is grown directly
Pt/Co/GdOx system revealed some of the complexities of onto the substrate by co-deposition, with the Co power varied
trying to understand the voltage-driven effects, with a compli- to obtain the desired composition and an Al layer is used as a
cated mix of oxide and hydroxide phases contributing to the cap to protect against oxidation.
reduction of PMA strength. The combination of multiple ionic The full sample structure for all films is Si/Tbx Co(100−x)
species makes it difficult to disentangle the effects of each (30 nm)/Al (8 nm), with a standard deviation of ±2 nm for
contributing species, which in turn makes it challenging to the TbCo thickness across all compositions. From previous
accurately model and predict future behavior through methods studies in the literature [29], this degree of variation in the
such as density functional theory. To simplify the system, it is thickness is unlikely to affect the magnetic properties as above
advantageous to consider a nonchemical loading mechanism 20 nm the coercivity, a key indicator of the coupling strength
such as ion implantation to study the effects of hydrogen and anisotropy, remains stable up to 100 nm.
loading on the magnetic behavior of a magnetic material. Film quality and structure are investigated through x-ray
In this work we investigate the hydrogen-induced changes measurements using a Cu Kα source with a wavelength of λ =
in the magnetic properties of Tbx Co(100−x) amorphous thin 1.54 Å. The amorphous character of the entire composition
film alloys by ion implantation. By using this technique we range is confirmed through grazing incidence x-ray diffraction
isolate the magneto-ionic effects to hydrogen ions alone, (GIXRD) with an incident angle of ω = 0.5◦ . For samples
where electrochemical means can have additional reactions across the entire composition range of interest there is a broad
involving hydroxide and oxide phases either forming or mi- amorphous peak but no crystalline peaks, shown in Fig. 1(a),
grating in response to the applied electric field. Like GdCo and so the samples are considered x-ray amorphous.
alloys, this material is ferrimagnetic and exhibits both a Layer thicknesses are quantified by fitting x-ray reflectom-
compensation composition at room temperature and a bulk- etry (XRR) data using GenX [30] and the Tb concentration,
like PMA with two transition compositions from in-plane x, of the Tbx Co(100−x) alloys is verified through Rutherford
anisotropy (IPA) to PMA with one in the Co-rich regime backscattering spectrometry (RBS) using 2 MeV He as the
(which we refer to as xT ) and one in the Tb-rich region. Using primary beam; the data are fit using SIMNRA [31]. An in-
ion beam analysis techniques we obtain precise knowledge dicative XRR profile is shown in Fig. 1(b) and demonstrates
of the exact alloy composition and hydrogen concentration that the films are smooth with Kiessig fringes up to 2θ = 7◦ .

034409-2
CONTROL OF FERRIMAGNETIC COMPENSATION AND … PHYSICAL REVIEW MATERIALS 9, 034409 (2025)

process. In their simulations, the damage profiles created by

these heavier ions are broad and introduce similar degrees of
damage throughout the RE-TM layer. From our simulations
we are also able to calculate the damage profile in Fig. 2(a)
and we find that our profile exhibits a peak toward the surface
of the film, which we expect to result in an increase in interfa-
cial roughness. Additionally, we expect much more significant
retention of H atoms as compared with the implantation of He
or Ne atoms.
We consider two nominal doses for implantation: 1×1017
and 2×1017 ions/cm2 of hydrogen atoms, which corresponds
to a ratio of approximately 1:2.5 (≈28.5 at.%) and and 2:2.5
(≈44 at.%) hydrogen atoms per metal atom based on the areal
densities obtained from RBS.
The potential structural changes caused by implantation are
probed by x-ray reflectometry and GIXRD after implantation.
From the GIXRD [Fig. 1(a)] we see no evidence of a change
in the amorphous peak or emergence of any crystalline phases
resulting from the implantation. From the reflectometry in
Fig. 1(b), we see that the H implantation does introduce
changes in electronic density, thickness, and surface rough-
ness (from approximately 5 to 10 Å in the surface layer)
which altogether leads to a decrease in intensity of the Kiessig
fringes in the reflectometry profile. In these films, the origin of
the PMA does not strongly depend on the interfacial coupling
between the Tbx Co100−x and Al layers and so we do not expect
the increase in roughness to significantly impact the PMA
strength and coercivity.
Although the implantation dose is the same, we can expect
FIG. 2. (a) Simulated implanted hydrogen ion distribution pro- that the final hydrogen concentration of each sample may vary
file and imlantation damage profile in a sample structure of
as a result of many factors. These factors include changes in
Al/Tb10 Co90 /Si for an ion energy of 2.5 keV and dose of
hydrogen solubility, differences in atomic density relating to
1×1017 ions/cm2 . (b) Measured hydrogen profile from nuclear re-
the change in Tbx Co100−x composition, and small variations
action analysis (NRA) of Tbx Co100−x samples after implantation of
1×1017 ions.
in sample quality during the growth process that will affect
hydrogen retention, such as the film thickness and porosity,
as such an essential element of this study is to measure the
retained concentration of hydrogen.
III. HYDROGEN IMPLANTATION
After implantation two ion-beam-based techniques are
Samples are implanted with 2.5 keV hydrogen at the used to measure the hydrogen content: nuclear reaction analy-
350 kV Danfysik Implanter at the Tandem Laboratory of sis (NRA), which is used to measure hydrogen depth profiles
Uppsala University [32]. The beam is raster scanned across with a depth resolution of a few nanometers, and elastic recoil
all samples on the plate to obtain a homogeneous distribution. detection analysis (ERDA) to measure the H concentration
On each sample plate a series of Tbx Co(100−x) samples with in the film and evaluate the uniformity of the implantation
varying compositions is included such that the dose received process across all samples. NRA was conducted using a 15 N
by each sample is the same. The implanted dose is measured beam with >6.385 MeV; the procedure is described in more
using a set of Faraday cups placed around the samples. detail elsewhere [35]. A primary beam of 2.1 MeV He+
Figure 2(a) shows the expected implantation profile for H+ ions was employed for ERDA measurements. Recoils were
ions with an energy of 2.5 keV simulated using the SRIM detected with a solid-state detector positioned at the forward
software [33]. The ion energy was chosen such that the peak of scattering angle of 40◦ , while the samples were rotated 70◦
the hydrogen profile is in the center of the TbCo film to max- with respect to the beam. For more experimental details we
imize the probability of obtaining a homogeneous hydrogen refer the reader to Ref. [36]. The hydrogen concentration has
profile. The calculated damage profile, represented in terms of been fit by assuming the majority of the ions are inside the
displacements per atom (DPA), shows that for this energy the Tbx Co(100−x) layer, as suggested by the NRA results.
most damage occurs at or just below the Al/TbCo interface. The obtained hydrogen concentration profiles from NRA
It is also important that we take into account the dam- in Fig. 2(b) show a flatter distribution of hydrogen atoms than
age done by the irradiation process. In a recent study by expected from SRIM simulations with a small bias toward the
Krupinski et al. [34] investigating irradiation of RE-TM films surface of the film for the fluence of 1×1017 ions/cm2 , indi-
by He and Ne species, significant changes in magnetic behav- cating that the hydrogen is able to diffuse through the film
ior were obtained purely by ion-induced damage, calculated as compared to the calculations of Fig. 2(a), where there is
as the displacement per atom (DPA) caused by the irradiation no diffusion. This diffusivity may be because the ions are not

034409-3
ROBBIE G. HUNT et al. PHYSICAL REVIEW MATERIALS 9, 034409 (2025)

calculate the atomic density of the film which is then con-

verted to the mass density by weighing the mass based on the
composition,
 
1 (100 − x) x
ρ= ρat. × MCo + ρat. × MTb , (1)
NA 100 100
where NA is Avogadro’s constant, ρ is the density in units
of g/cm3 , ρat. is the density in units of at./cm3 , and MCo
and MTb are the atomic masses of Co and Tb, respectively.
The as-deposited films have densities that are significantly
higher than the densities for either of the elemental species,
although not as high as the expected density for the inter-
metallic species TbCo2 . From this we can understand that
there is a Tb-composition dependence of the mass density
that exhibits a weak maximum in the region of x ≈ 24 at.%
at room temperature.
The relationship here between the as-deposited density of
the Tbx Co100−x layer and hydrogen solubility (the ratio of
hydrogen atoms to metal atoms, representative of the capacity
to retain) may then be the opposite of expected. A material
with a higher density of metal atoms would have a lower
concentration of the same number of hydrogen atoms, but
what we see is a weak trend of the hydrogen concentration
increasing with the density. However, this relationship is weak
and the most critical conclusion we can draw from these mea-
surements is that the hydrogen concentration is at saturation
for these films.

FIG. 3. (a) Composition dependence of hydrogen concentration, IV. MAGNETIC CHARACTERIZATION

as a percentage relative to the metal atoms, determined from ERDA Magnetic hysteresis loops at room temperature are per-
and NRA for implantation doses of 1×1017 and 2×1017 ions/cm2 . formed using a magneto-optic Kerr effect (MOKE) experi-
(b) Mass density derived from RBS and XRR measurements for the ment in the polar geometry with a laser wavelength of 670 nm.
as-deposited samples. From previous work in the literature by Khorsand et al. [37],
it is known that the Kerr effect for these ferrimagnetic ma-
strongly bound to the initial implantation sites. Moreover, the terials is not sensitive purely to the net magnetization, but is
profile might be biased by the irradiation-induced damage, as instead more sensitive to the magnetization on one particular
the implantation process may create vacancies that will act as sublattice dependent upon the wavelength of the probing laser.
local minima for the hydrogen atoms to migrate to, but this Based on this we know that a probe wavelength of 670 nm will
will be most pronounced at the Tbx Co100−x /Al interface and be sensitive to the Co sublattice and not the Tb sublattice. The
so the profile will not be solely determined by it. experimental setup is such that the angle between the polarizer
Equally of note is that the hydrogen concentrations vary and the analyzer remains fixed and so the change in orienta-
somewhat between the measured samples. Samples with the tion of the Co sublattice across the compensation composition
highest and lowest Tb concentrations show the smallest re- is clear. The inversion of the hysteresis polarity is used to help
tained hydrogen concentrations, while the samples around define the position of the room temperature compensation
x = 20 have a much higher hydrogen concentration, approxi- composition, xc , as shown in Fig. 4 together with the increase
mately 30 at.% compared to 20 at.%. in coercivity. Prior to implantation, all samples have square
To examine this in more detail we perform ERDA on hysteresis loops suggesting excellent uniaxial PMA with no
all samples from both doses. Figure 3 shows that there is significant in-plane contributions to the magnetic anisotropy.
no strong dependence of the retained dose on either the Tb Since all of the samples display PMA we take the position
concentration or the implanted hydrogen dose. For the lower of the Co-rich in-plane anisotropy to out-of-plane transition,
dose, there may be a small linear trend with Tb concentration xT , in the as-deposited state to be ≈10 at.%, based on work
but this is dominated much more by the scatter of the data reported in the literature for similar samples [38,39].
around the average value of 26.5 at.%. The films with the In agreement with previous results [29,40,41] for
highest (39 at.%) and lowest (11 at.%) terbium concentrations Tbx Co(100−x) alloys, the value of xc , at which point the Tb and
retain the lowest concentration of hydrogen, which suggests Co sublattice magnetizations perfectly compensate each other,
that at these doses the concentration of the RE element is not is found to be approximately 20.5 at.% in Fig. 5, indicated
contributing significantly to the retention. by the point at which the coercivity of the film diverges and
From the areal density (at./cm2 ) obtained from RBS data is accompanied by a reversal of hysteresis loop polarity as
and the thickness obtained from XRR data we are able to explained previously. A sample grown with this composition

034409-4
CONTROL OF FERRIMAGNETIC COMPENSATION AND … PHYSICAL REVIEW MATERIALS 9, 034409 (2025)

FIG. 4. MOKE hysteresis loops for as-deposited (black, left) and implanted (red, right) samples with a nominal dose of 1×1017 ions/cm2 .
The change in polarity from x=16 at.% to x=26 at.% corresponds to crossing the compensation composition.

had a coercive field in excess of 1.2 T and it was not possible addition to this, the value of xT increases by approximately
to measure the hysteresis behavior of the sample, in line with the same magnitude with the x = 11 and 13 at.% samples
what is expected at or extremely near to compensation, and falling in plane after implantation and the x = 16 and 18 at.%
so we take this composition to be the value of xc prior to samples displaying some tilted anisotropy with in-plane and
implantation. out-of-plane components. The reduced PMA strength in these
With the implanted hydrogen dose of 1×1017 ions/cm2 , samples is likely due to the implanted hydrogen disrupting the
there is an immediate shift of the compensation composition ratio of in-plane and out-of-plane Tb-Co, Tb-Tb, and Co-Co
toward a higher value of x by approximately 5–6 at.%. In pair correlations that are linked to the PMA in these materi-
als [42]. For the samples with IPA, we show the hysteresis
behavior measured in a longitudinal MOKE configuration in
Fig. 6(a). For the samples that lose PMA as a result of im-
plantation, the coercivity increases linearly with the value of
the Tb concentration.
We can initially understand the change in compensation
composition as a change in moment on the terbium atoms.
From the previously presented NRA data we know that the
hydrogen can diffuse after implantation and, based on the
high affinity of rare-earth materials and the poor affinity of
cobalt for hydride formation [43], we can presume that it
would be more preferable for the hydrogen to diffuse to the
Tb sites. From the conditions for compensation, we know that
the overall moment on the terbium atoms must be lower upon
introducing hydrogen and so a greater number of terbium
atoms is needed to reach a perfect balance between the two
sublattices. This explanation agrees well with the explanation
FIG. 5. Determination of compensation composition, xc , in the
proposed by Huang et al. [22], where a reduction in com-
as-deposited state (•) and after hydrogen implantation doses of pensation temperature of a GdCo layer was attributed to the
1×1017 ions/cm2 () and 2×1017 ions/cm2 (). Vertical lines indicate reduction in moment of the Gd and Co species upon hydrogen
the approximate position of xC , with the red dashed line showing loading driven by the structural changes in the Gd-Co bond
the position for both implantation doses. Shaded regions indicate lengths and the resulting reduction in the exchange coupling
compositions for which the films have in-plane anisotropy, with constants.
the left-hand side region marked as xT , and the lighter red region We now consider samples implanted with a dose of
being the change in the IPA region after an implantation dose of 2×1017 ions/cm2 . Here samples have been implanted on a
1×1017 ions/cm2 . more constrained range of x = 17−39 at.%, as samples below

034409-5
ROBBIE G. HUNT et al. PHYSICAL REVIEW MATERIALS 9, 034409 (2025)

change in the positions of either xT or xc , indicating that the

magnetic properties are not sensitive to the damage induced in
implantation for hydrogen at these energies or that there is no
significant damage to the amorphous structure. Turning to the
behavior of the x = 39 at.% samples that always have in-plane
anisotropy, Fig. 6(b) shows that there is a minimal increase in
coercivity when the dose is doubled. As we expect that the
damage profile will scale linearly with the dose, we conclude
that the ion-induced damage is not the leading cause of the
changes in xc and xT .
To examine this in more detail measurements of the net
uncompensated magnetization (Ms ) are taken using a vibrat-
ing sample magnetometer with the absolute magnetization
extracted from hysteresis loops. The net ferrimagnetic mag-
netization as a function of composition is shown in Fig. 7.
The composition-dependent curve can be related to the atomic
magnetic moments of the Co and Tb sublattices through two
models as shown previously by Suzuki et al. [44]. First,
through an initial naive model in which the atomic magnetic
moments of the Co and Tb atoms remain constant for all
compositions. In this model the net saturation magnetization
is calculated as
 
 mz,Co (100 − x) − mz,Tb x 
Ms (x) =  , (2)
vCo (100 − x) + vTb x 
where Ms is the net saturation magnetization of the ferri-
magnetic magnetization, mz,Co and mz,Tb are the atomic Co
and Tb magnetic moments in the z direction, and vCo and
vTb are the atomic volumes of Co and Tb taken to be the
same as in Ref. [44], that is, vCo = 1.10×10−29 m3 and vTb =
3.211×10−29 m3 . This model fits the compensation behavior
FIG. 6. In-plane magnetic hysteresis loops for (a) x = 10−16 reasonably well, but cannot fit the decrease in Ms at higher
at.% after a 1×1017 ions/cm2 dose and (b) x = 39 at.% under varying values of x far from compensation.
hydrogen doses. Insets show the coercivities from each hysteresis To encapsulate this decrease in magnetization we use a
loop. modified model in which the atomic magnetic moments vary
with increasing Tb composition, used initially by Jaccarino
and Walker [45] to describe magnetization of impurity species
x = 17 at.% are already expected to lose PMA. The lower in alloy materials. The composition dependence of the atomic
bound was chosen based on the shift in xT for a dose of moment is then described by
1×1017 ions/cm2 . From the previously presented ERDA data
in Fig. 3 we understand that there is no significant increase mz,i (x) = Pj (x)mz,i (x = 0), (3)
in hydrogen concentration as compared with the initial dose,
but from the increased dose we can expect a higher degree where i is the atomic species (Co or Tb), and Pj (x) is a
of ion-induced damage. Despite this, we see no significant probabilistic prefactor accounting for the change in atomic

FIG. 7. Saturation magnetization as a function of composition for the (a) as-deposited, (b) 1×1017 ions/cm2 , and (c) 2×1017 ions/cm2 H+
implantation doses. Black and green dashed lines represent best fits to a constant magnetization and environment-dependent magnetization
model, respectively (see text for more detail).

034409-6
CONTROL OF FERRIMAGNETIC COMPENSATION AND … PHYSICAL REVIEW MATERIALS 9, 034409 (2025)

(deg)
FIG. 8. (a) Schematic diagram of the sperimagnetic structure in TbCo alloys. (b) mTb (x) profiles with increasing hydrogen dose obtained
from the environment model fits in Fig. 7. (c) Resulting sperimagnetic cone angle assuming any reduction in the magnetic moment results
from changes to the sperimagnetic structure.

environment with increasing x, magnetic moment on the Tb sublattice—a reduction in the

N  
value of mTb (x=0) of 24.5% for a dose of 1×1017 ions/cm2
 N and a reduction of 25.1% for 2×1017 ions/cm2 . It is important
Pj (x) = (100 − x)k x N−k , (4)
k to note that because of the assumptions made for this fitting,
k= j
this represents the minimum reduction in the total moment
with j being the minimum number of nearest Co neighbors, on the Tb sublattice. While we presume that the Co magnetic
N the coordination
 number which for TbCo takes a maximum moment does not change significantly, if the Co sublattice also
of N = 12, and Nk is the binomial coefficient. reduces in magnetic moment then the corresponding reduction
For each dose we show in Fig. 7 fits to both of these in the Tb sublattice must be greater to achieve compensation
models. To simplify the fitting, we make the assumption that at the same composition observed here. Element-sensitive
the magnetic moment of Co at x = 0 is always the expected techniques such as x-ray magnetic circular dichroism would
atomic magnetic moment of 1.71 μB . For the as-deposited be required to distinguish this in greater detail.
samples this justification is simple as the saturation magneti- An important element to consider with this material is
zation at x = 0 should converge to the atomic moment of Co. the sperimagnetic structure—that is, the disordered aspect of
In the implanted samples, we turn to the x-ray magnetic circu- the alignment between the Tb and Co sublattices [46]. The
lar dichroism data of Huang et al. [22] obtained for GdCo, in sperimagnetic structure arises from the amorphous structure
which the change in Co atomic moment upon hydrogen load- of the film with each site having different local energy minima
ing is observed to be extremely small with differences only leading to deviations in the alignment of the magnetic mo-
becoming discernible when fitted models are extrapolated out ments which can be described as a fan or cone structure. The
to much higher temperatures than room temperature. Given Co-Co exchange remains the most dominant factor for Co
also that the value of xc shifts toward higher Tb concentrations atoms and so they align parallel and are typically described
it is also reasonable to assume that the largest changes in as having no conical structure [47]. For the Tb sites, the
magnetic moment occur on the Tb sublattice as opposed to Tb-Co exchange dominates and local fluctuations in compo-
the Co sublattice. sition have a large impact on the structure giving rise to the
Under these assumptions, the change in xc can be under- sperimagnetic cone structure, illustrated in Fig. 8(a).
stood to be result from a change in the extrapolated value of As a result, we can also consider changes in the atomic
mTb (x = 0) and that the implanted hydrogen ions behave con- magnetic moment of Tb as a change in the sperimagnetic
sistently across all samples regardless of change in rare-earth structure of the Tbx Co(100−x) film. From the changes in PMA
concentration. In previous work in the literature, it has been strength, we can understand that the hydrogen is disrupting
concluded in other RE-TM alloys that these changes could be the random single-ion anisotropy. This random anisotropy is
explained by the intercalation of hydrogen leading to larger also a leading factor in the existence of the sperimagnetism in
RE-TM bond lengths which reduces the magnitude of the these materials and so changes in the Tb atomic magnetic mo-
exchange coupling [22]. In the ab initio modeling presented ment could be understood through the lens of changes to the
in Ref. [22] for GdCo, hydrogen concentrations of 20 and sperimagnetic cone angle, as has been done in Ref. [48]. For
40 at.% were considered in a crystalline GdCo2 system and a filled cone structure the cone angle, θTb , can be calculated
predict a continuous degradation of the exchange coupling as by comparing the measured atomic magnetic moment, mTb , to
the concentration increases. Based on our NRA and ERDA the expected moment of a free terbium atom of μTb = 9.34μB
results, we consider that all samples are saturated with hydro- [49],
gen although the exact value of concentration varies possibly
mTb = μTb (1 + cos θTb )/2, (5)
due to the change in density. Between the two doses, the
most obvious change will be the increase in DPA which we which can be numerically determined based on the composi-
expect to scale linearly with the dose. Despite this, we find tional profiles of mTb obtained from the fits to the environment
that fitting to either model shows similar reductions in the model obtained previously and shown in Fig. 8(b). The

034409-7
ROBBIE G. HUNT et al. PHYSICAL REVIEW MATERIALS 9, 034409 (2025)

resulting obtained cone angle curves shown in Fig. 8(c) The origin of these effects is likely a result of the change
then follow the inverse of the mTb (x) curve, with the largest in metal pair correlations increasing in length due to the im-
changes at the smallest values of x and the hydrogen dose- planted hydrogen, but whether this results in a reduction of the
dependent curves converging for high values of x at which exchange coupling, a change in the sperimagnetic structure, or
point the atomic magnetic moment tends toward zero. The both requires further experiments to determine.
reduction in mTb in this model would imply that as a result These results will be important for future work focusing on
of hydrogen uptake there is an increase in the dispersion of devices that take advantage of recent advances in solid-state
the Tb sublattice magnetizations. proton-pump methods by providing quantitative information
It is, however, challenging to distinguish a reduction in the that relates the amount of hydrogen loaded with the change
measured magnetization due to an increased cone angle from in magnetic properties, as well as providing insight into ideal
a reduction in the Tb-Co exchange interaction. Measurement choices of composition.
of the cone angle requires characterization techniques that are
sensitive to the orthogonal components of magnetization, such ACKNOWLEDGMENTS
as neutron diffraction [50] or Mössbauer spectroscopy [51]
(although this would mostly be limited to alloys containing The authors would like to thank Johan Oscarsson and
Fe). Additionally, further characterization of the local atomic Mauricio Sortica at the Uppsala Tandem Laboratory for their
and electronic structural changes is needed to understand the help with ion implantations. R.G.H. acknowledges the sup-
change in moment fully. port of a Carl Trygger Foundation Postdoctoral Fellowship
(Grant No. CTS 22:2039, grant holder G.A.) and finan-
V. CONCLUSIONS cial support from the Wenner-Gren Foundations (Contract
No. RSh2024-0057). M.P.G. acknowledges support from the
In conclusion we have evaluated the effect of hydrogen ion Swedish Research Council (Project No. 2023-06359). Ac-
implantation with different doses on the magnetic properties celerator operation at the Ion Technology Center national
of amorphous Tbx Co(100−x) alloys for a range of alloy compo- infrastructure is supported by the Swedish Research Council
sitions. The effect of hydrogen loading in this way can clearly VR-RFI (Grant No. 2019_00191).
be seen as an effective shift in the compensation composition
and in the Co-rich in-plane to out-of-plane transition. From
DATA AVAILABILITY
our magnetometry, the change in compensation composition
can be linked to a reduction in the magnetic moment of the Tb Data are available from the corresponding author upon
sublattice primarily. reasonable request.

[1] S. Manipatruni, D. E. Nikonov, C.-C. Lin, T. A. Gosavi, H. perpendicular magnetic anisotropy by surface acoustic waves,
Liu, B. Prasad, Y.-L. Huang, E. Bonturim, R. Ramesh, and Appl. Phys. Lett. 120, 252402 (2022).
I. A. Young, Scalable energy-efficient magnetoelectric spin– [9] H. Ohno, D. Chiba, F. Matsukura, T. Omiya, E. Abe, T. Dietl, Y.
orbit logic, Nature (London) 565, 35 (2019). Ohno, and K. Ohtani, Electric-field control of ferromagnetism,
[2] A. Jaiswal, S. Roy, G. Srinivasan, and K. Roy, Proposal for Nature (London) 408, 944 (2000).
a leaky-integrate-fire spiking neuron based on magnetoelectric [10] M. Weisheit, S. FÃhler, A. Marty, Y. Souche, C. Poinsignon,
switching of ferromagnets, IEEE Trans. Electron Devices 64, and D. Givord, Electric field-induced modification of mag-
1818 (2017). netism in thin-film ferromagnets, Science 315, 349 (2007).
[3] C. Ederer and N. A. Spaldin, Weak ferromagnetism and mag- [11] C. Navarro-Senent, J. Fornell, E. Isarain-Chavez, A. Quintana,
netoelectric coupling in bismuth ferrite, Phys. Rev. B 71, E. Menendez, M. Foerster, L. Aballe, E. Weschke, J. Nogues, E.
060401(R) (2005). Pellicer et al., Large magnetoelectric effects in electrodeposited
[4] J. T. Heron, D. G. Schlom, and R. Ramesh, Electric field control nanoporous microdisks driven by effective surface charging
of magnetism using BiFeO3 -based heterostructures, Appl. Phys. and magneto-ionics, ACS Appl. Mater. Interfaces 10, 44897
Rev. 1, 021303 (2014). (2018).
[5] K. J. A. Franke, B. Van de Wiele, Y. Shirahata, S. J. [12] T. Bhatnagar-Schöffmann, A. Kovàcs, R. Pachat, D. Ourdani,
Hämäläinen, T. Taniyama, and S. van Dijken, Reversible A. Lamperti, M.-A. Syskaki, T. da Câmara Santa Clara Gomes,
electric-field-driven magnetic domain-wall motion, Phys. Rev. Y. Roussigné, S. Ono, J. Langer et al., Controlling inter-
X 5, 011010 (2015). face anisotropy in CoFeB/MgO/HfO2 using dusting layers and
[6] R. G. Hunt, K. J. A. Franke, P. M. Shepley, and T. A. Moore, magneto-ionic gating, Appl. Phys. Lett. 122, 042402 (2023).
Strain-coupled domains in BaTiO3 (111)-CoFeB heterostruc- [13] B. X. Tran, J.-H. Ha, W.-C. Choi, S. Yoon, T.-H. Kim, and J.-I.
tures, Phys. Rev. B 107, 014409 (2023). Hong, Field-free control and switching of perpendicular magne-
[7] I. S. Camara, J.-Y. Duquesne, A. Lemaître, C. Gourdon, and L. tization by voltage induced manipulation of RKKY interaction,
Thevenard, Field-free magnetization switching by an acoustic Appl. Phys. Lett. 124, 112409 (2024).
wave, Phys. Rev. Appl. 11, 014045 (2019). [14] L. Herrera Diez, Y. T. Liu, D. A. Gilbert, M. Belmeguenai,
[8] J. Shuai, M. Ali, L. Lopez-Diaz, J. E. Cunningham, and T. A. J. Vogel, S. Pizzini, E. Martinez, A. Lamperti, J. B.
Moore, Local anisotropy control of Pt/Co/Ir thin film with Mohammedi, A. Laborieux et al., Nonvolatile ionic modifica-

034409-8
CONTROL OF FERRIMAGNETIC COMPENSATION AND … PHYSICAL REVIEW MATERIALS 9, 034409 (2025)

tion of the Dzyaloshinskii-Moriya interaction, Phys. Rev. Appl. from in operando neutron reflectometry and secondary ion mass
12, 034005 (2019). spectrometry, Appl. Phys. Lett. 122, 022407 (2023).
[15] M. U. Hasan, A. E. Kossak, and G. S. D. Beach, Large exchange [29] A. Ciuciulkaite, K. Mishra, M. V. Moro, I.-A. Chioar, R. M.
bias enhancement and control of ferromagnetic energy land- Rowan-Robinson, S. Parchenko, A. Kleibert, B. Lindgren, G.
scape by solid-state hydrogen gating, Nat. Commun. 14, 8510 Andersson, C. S. Davies et al., Magnetic and all-optical switch-
(2023). ing properties of amorphous Tbx Co100−x alloys, Phys. Rev.
[16] J. de Rojas, A. Quintana, A. Lopeandía, J. Salguero, B. Muñiz, Mater. 4, 104418 (2020).
F. Ibrahim, M. Chshiev, A. Nicolenco, M. O. Liedke, M. [30] M. Björck and G. Andersson, GenX: An extensible X-ray re-
Butterling et al., Voltage-driven motion of nitrogen ions: A flectivity refinement program utilizing differential evolution,
new paradigm for magneto-ionics, Nat. Commun. 11, 5871 J. Appl. Cryst. 40, 1174 (2007).
(2020). [31] M. Mayer, SIMNRA, a simulation program for the analysis of
[17] J. de Rojas, A. Quintana, G. Rius, C. Stefani, N. Domingo, NRA, RBS and ERDA, AIP Conf. Proc. 475, 541 (1999).
J. L. Costa-Krämer, E. Menéndez, and J. Sort, Voltage control [32] P. Ström and D. Primetzhofer, Ion beam tools for nondestructive
of magnetism with magneto-ionic approaches: Beyond voltage- in-situ and in-operando composition analysis and modification
driven oxygen ion migration, Appl. Phys. Lett. 120, 070501 of materials at the Tandem Laboratory in Uppsala, J. Instrum.
(2022). 17, P04011 (2022).
[18] U. Bauer, L. Yao, A. J. Tan, P. Agrawal, S. Emori, H. L. Tuller, [33] J. F. Ziegler, M. D. Ziegler, and J. P. Biersack, SRIM—The
S. Van Dijken, and G. S. D. Beach, Magneto-ionic control of stopping and range of ions in matter (2010), Nucl. Instrum.
interfacial magnetism, Nat. Mater. 14, 174 (2015). Methods Phys. Res. Sect. B 268, 1818 (2010).
[19] B. Santos, S. Gallego, A. Mascaraque, K. F. McCarty, A. [34] M. Krupinski, J. Hintermayr, P. Sobieszczyk, and M. Albrecht,
Quesada, A. T. N’Diaye, A. K. Schmid, and J. de la Figuera, Control of magnetic properties in ferrimagnetic GdFe and TbFe
Hydrogen-induced reversible spin-reorientation transition and thin films by He+ and Ne+ irradiation, Phys. Rev. Mater. 5,
magnetic stripe domain phase in bilayer Co on Ru(0001), 024405 (2021).
Phys. Rev. B 85, 134409 (2012). [35] K. Komander, M. V. Moro, S. A. Droulias, J. Müggenburg,
[20] G. Chen, C. Ophus, A. Quintana, H. Kwon, C. Won, H. Ding, Y. G. K. Pálsson, T. Nyberg, D. Primetzhofer, and M. Wolff,
Wu, A. K. Schmid, and K. Liu, Reversible writing/deleting of Hydrogen site location in ultrathin vanadium layers by N-15
magnetic skyrmions through hydrogen adsorption/desorption, nuclear reaction analysis, Nucl. Instrum. Methods Phys. Res.
Nat. Commun. 13, 1350 (2022). Sect. B 455, 57 (2019).
[21] M. Ameziane, J. Huhtasalo, L. Flajšman, R. Mansell, and S. [36] E. Pitthan, P. Petersson, T. T. Tran, D. Moldarev, R. Kaur,
van Dijken, Solid-state lithium ion supercapacitor for voltage J. Shams-Latifi, P. Ström, M. Hans, M. Rubel, and D.
control of skyrmions, Nano Lett. 23, 3167 (2023). Primetzhofer, Thin films sputter-deposited from EUROFER97
[22] M. Huang, M. U. Hasan, K. Klyukin, D. Zhang, D. in argon and deuterium atmosphere: Material properties and
Lyu, P. Gargiani, M. Valvidares, S. Sheffels, A. Churikova, deuterium retention, Nucl. Mater. Energy 34, 101375 (2023).
F. Büttner et al., Voltage control of ferrimagnetic order [37] A. R. Khorsand, M. Savoini, A. Kirilyuk, A. V. Kimel, A.
and voltage-assisted writing of ferrimagnetic spin textures, Tsukamoto, A. Itoh, and T. Rasing, Element-specific probing of
Nat. Nanotechnol. 16, 981 (2021). ultrafast spin dynamics in multisublattice magnets with visible
[23] D. Pravarthana, B. Wang, Z. Mustafa, S. Agarwal, K. Pei, H. light, Phys. Rev. Lett. 110, 107205 (2013).
Yang, and R.-W. Li, Reversible control of magnetic anisotropy [38] M. Odagiri, H. Imaeda, A. Yagmur, Y. Kurokawa, S. Sumi, H.
and magnetization in amorphous Co40 Fe40 B20 thin films via all- Awano, and K. Tanabe, Coexistence of large anomalous Nernst
solid-state Li-ion redox capacitor, Phys. Rev. Appl. 12, 054065 effect and large coercive force in amorphous ferrimagnetic
(2019). TbCo alloy films, Appl. Phys. Lett. 124, 142403 (2024).
[24] M. G. Jani, L. E. Halliburton, and A. Halperin, Observation [39] K. A. Thórarinsdóttir, B. R. Thorbjarnardóttir, U. B. Arnalds,
of a simple lithium-associated electron trap in crystalline SiO2 , and F. Magnus, Competing interface and bulk anisotropies in
Phys. Rev. Lett. 56, 1392 (1986). Co-rich TbCo amorphous thin films, J. Phys.: Condens. Matter
[25] A. J. Tan, M. Huang, M. U. Hasan, K. Klyukin, D. Zhang, D. 35, 205802 (2023).
Lyu, P. Gargiani, M. Valvidares, S. Sheffels, A. Churikova, F. [40] A. Yagmur, S. Sumi, H. Awano, and K. Tanabe, Magnetization-
Büttner, J. Zehner, L. Caretta, K.-Y. Lee, J. Chang, J.-P. Wang, dependent inverse spin Hall effect in compensated ferrimagnet
K. Leistner, B. Yildiz, and G. S. D. Beach, Magneto-ionic con- TbCo alloys, Phys. Rev. B 103, 214408 (2021).
trol of magnetism using a solid-state proton pump, Nat. Mater. [41] S. Alebrand, M. Gottwald, M. Hehn, D. Steil, M. Cinchetti, D.
18, 35 (2019). Lacour, E. E. Fullerton, M. Aeschlimann, and S. Mangin, Light-
[26] A. E. Kossak, M. Huang, P. Reddy, D. Wolf, and G. S. D. induced magnetization reversal of high-anisotropy TbCo alloy
Beach, Voltage control of magnetic order in RKKY coupled films, Appl. Phys. Lett. 101, 162408 (2012).
multilayers, Sci. Adv. 9, eadd0548 (2023). [42] V. G. Harris, K. D. Aylesworth, B. N. Das, W. T. Elam, and N. C.
[27] Z. Xiao, R. Xie, F. Maccari, P. Klaßen, B. Eggert, D. Wang, Koon, Structural origins of magnetic anisotropy in sputtered
Y. Dai, R. Lizárraga, J. Lill, T. Helbig et al., Voltage-gated amorphous Tb-Fe films, Phys. Rev. Lett. 69, 1939 (1992).
90◦ switching of bulk perpendicular magnetic anisotropy in [43] H. M. Lee, The solubility of hydrogen in transition metals,
ferrimagnets, ACS Nano 19, 6021 (2025). Metall. Trans. A 7, 431 (1976).
[28] S. Sheffels, P. P. Balakrishnan, M. Huang, S. Muramoto, J. A. [44] D. H. Suzuki, M. Valvidares, P. Gargiani, M. Huang, A. E.
Borchers, J. A. Dura, A. J. Grutter, and G. S. D. Beach, In- Kossak, and G. S. D. Beach, Thickness and composition ef-
sight on hydrogen injection and GdOx/Co interface chemistry fects on atomic moments and magnetic compensation point in

034409-9
ROBBIE G. HUNT et al. PHYSICAL REVIEW MATERIALS 9, 034409 (2025)

rare-earth transition-metal thin films, Phys. Rev. B 107, 134430 the formation of atomic magnetic moments in amorphous films
(2023). RE-Co (RE=La, Gd, Tb), Phys. Solid State 57, 1142 (2015).
[45] V. Jaccarino and L. R. Walker, Discontinuous occurrence of [49] I. A. Makarochkin, E. V. Kudyukov, E. A. Stepanova, G. V.
localized moments in metals, Phys. Rev. Lett. 15, 258 (1965). Kurlyandskaya, V. O. Vas’ Kovskiy, and A. V. Svalov, Fea-
[46] J. J. Rhyne, Amorphous magnetic rare earth alloys, in Alloys tures of the sperimagnetic structure of TbCo-based multilayers,
and Intermetallics, Handbook on the Physics and Chemistry AIP Conf. Proc. 2313, 030057 (2007).
of Rare Earths (Elsevier, Amsterdam, 1979), Vol. 2, Chap. 16, [50] D. Givord, A. Liénard, J. Rebouillat, and J. Schweizer, Ex-
pp. 259–294. change interactions and random anisotropy in ErCo1.96 and
[47] S. Uchiyama, Magnetic properties of rare earth-cobalt amor- TbCo1.98 amorphous alloys, Physica B+C 130, 351 (1985).
phous films, Mater. Chem. Phys. 42, 38 (1995). [51] J. M. D. Coey, J. Chappert, J. P. Rebouillat, and T. S. Wang,
[48] V. O. Vas’kovskiy, O. A. Adanakova, K. G. Balymov, N. A. Magnetic structure of an amorphous rare-earth transition-metal
Kulesh, A. V. Svalov, and E. A. Stepanova, Specific features of alloy, Phys. Rev. Lett. 36, 1061 (1976).

034409-10