PHYSICAL REVIEW B 101, 064416 (2020)

Quasi-two-dimensional magnon identification in antiferromagnetic

FePS3 via magneto-Raman spectroscopy
Amber McCreary,1 Jeffrey R. Simpson,1,2 Thuc T. Mai,1,3 Robert D. McMichael ,1 Jason E. Douglas,4 Nicholas Butch,5
Cindi Dennis,4 Rolando Valdés Aguilar ,3 and Angela R. Hight Walker 1,*
1
Nanoscale Device Characterization Division, Physical Measurement Laboratory,
National Institute of Standards and Technology, Gaithersburg, Maryland 20899, USA
2
Department of Physics, Astronomy, and Geosciences, Towson University, Towson, Maryland 21252, USA
3
Department of Physics, The Ohio State University, Columbus, Ohio 43210, USA
4
Materials Science and Engineering Division, Material Measurement Laboratory,
National Institute of Standards and Technology, Gaithersburg, Maryland 20899, USA
5
NIST Center for Neutron Research, National Institute of Standards and Technology, Gaithersburg, Maryland 20899, USA

(Received 2 August 2019; accepted 20 December 2019; published 18 February 2020)

Recently it was discovered that van der Waals bonded magnetic materials retain long range magnetic ordering
down to a single layer, opening many avenues in fundamental physics and potential applications of these
fascinating materials. One such material is FePS3 , a large spin (S = 2) Mott insulator where the Fe atoms form a
honeycomb lattice. In the bulk, FePS3 has been shown to be a quasi-two-dimensional-Ising antiferromagnet,
with additional features in the Raman spectra emerging below the Néel temperature (TN ) of approximately
120 K. Using magneto-Raman spectroscopy as an optical probe of magnetic structure, we show that one of
these Raman-active modes in the magnetically ordered state is actually a magnon with a frequency of ≈3.7 THz
(122 cm−1 ). Contrary to previous work, which interpreted this feature as a phonon, our Raman data shows the
expected frequency shifting and splitting of the antiferromagnetic magnon as a function of temperature and
magnetic field, respectively, where we determine the g factor to be ≈2. In addition, the symmetry behavior of
the magnon is studied by polarization-dependent Raman spectroscopy and explained using the magnetic point
group of FePS3 .

DOI: 10.1103/PhysRevB.101.064416

I. INTRODUCTION vices, where 2D magnets coupled with other technologically

relevant materials could realize unprecedented capabilities in
Since the isolation of monolayer graphene in 2004 [1],
fields such as spintronics [11,21,22].
there has been a surge of research into van der Waals layered
In early 2017, intrinsic ferromagnetism was observed down
materials, where the strong intralayer and weak interlayer
to the few-layer and monolayer limit in two different, lay-
coupling allows for isolation of layers that are only a few
ered materials with magnetic anisotropy: Cr2 Ge2 Te6 and CrI3
atoms thick. These materials exhibit a wide range of elec-
[23,24]. For CrI3 , it was also shown that the interlayer mag-
tronic properties, including semiconducting, metallic, insu-
netic ordering (i.e., ferromagnetic or antiferromagnetic stack-
lating, superconducting, and charge density waves, allowing
ing) was dependent on the number of layers [24], and could
for device architectures composed of solely two-dimensional
be controlled by an external electric field [14,25]. Transition
(2D) materials [2–4]. While significant research has been
metal phosphorus trisulfides X PS3 (X = Fe, Mn, Ni, etc.)
dedicated thus far to investigating the optical, mechanical,
are another class of van der Waals antiferromagnets that are
and electrical properties of 2D materials [3,5–9], exploring
being studied in the 2D limit. Interestingly, although FePS3
magnetism is still in its infancy, even though 2D magnetic
(TN ≈ 120 K) [26], MnPS3 (TN ≈ 78 K) [26], and NiPS3
materials provide a solid-state platform to experimentally
(TN ≈ 155 K) [26] are isostructural, they have different spin
access fundamental, low-dimensional physics [10,11]. Ad-
structures below the Néel temperature. The choice of transi-
ditionally, any 2D magnetic material would likely still pos-
tion metal results in varied magnetic phenomena, since the
sess the captivating properties of 2D materials, including
spins align antiferromagnetically within the layers in different
extremely large mechanical flexibility [12,13], efficient tuning
fashions, including Néel, zigzag, or stripe ordering [27,28].
of transport properties with an electric field [14–17], relative
FePS3 , which is a Mott insulator [29,30], is especially in-
ease of chemical modification [18,19], as well as the ability
triguing as a 2D Ising antiferromagnet on a honeycomb lattice
to create van der Waals stacked heterostructures [20]. These
[26,31]. In addition, long-distance magnon transport (several
myriad of tuning parameters could unlock opportunities for
micrometers) has been experimentally observed in MnPS3 ,
custom-engineered magnetoelectric and magneto-optical de-
demonstrating that these materials are viable candidates for
possible magnonic devices [22].
Raman spectroscopy, being nondestructive and highly sen-
* sitive to minute lattice perturbations, is a powerful technique
[email protected]

2469-9950/2020/101(6)/064416(10) 064416-1 ©2020 American Physical Society

AMBER MCCREARY et al. PHYSICAL REVIEW B 101, 064416 (2020)

to study various properties of quantum materials, including ture has been a source of controversy [31,48–51], where in
the effects of layer number [32,33], strain [34], defects/doping both proposed structures, the moments of the large spin Fe2+
[35], electron-phonon coupling [36,37], phase transitions ions (S = 2) are aligned normal to the a-b plane and ordered
[38], spin-phonon coupling [39], and magnetic excitations in ferromagnetic chains that couple antiferromagnetically. The
[40,41]. In addition, unlike other measurements that require proposed structures differed from each other, however, by a
bulk, large-area crystals, such as neutron diffraction, x-ray 120° rotation of the ferromagnetic chains, as well as whether
diffraction, or magnetic susceptibility, Raman spectroscopy the layers themselves are ferromagnetically or antiferromag-
can probe atomically thin flakes with diffraction-limited spa- netically stacked. Recent neutron scattering by Lançon et al.
tial resolution (e.g. micrometer spot size). The Raman spec- [51] of large, single crystals of FePS3 confirmed the magnetic
tra of bulk X PS3 materials have been studied previously structure consists of ferromagnetic chains oriented along the
[27,42,43], and only recently extended to samples in the a axis, with antiferromagnetically coupled layers along the c
monolayer limit for NiPS3 [44] and FePS3 [45,46]. In partic- axis, as shown in Figs. 1(b) and 1(c).
ular, FePS3 is an interesting candidate to study using Raman Using a superconducting quantum interference device
spectroscopy because, due to the antiferromagnetic alignment (SQUID) magnetometer, the magnetization of the FePS3 crys-
of the spins and the resulting increase in the unit cell, Brillouin tals (i.e., before exfoliation) was measured both parallel and
zone folding leads to new modes appearing below TN [45,46]. perpendicular to the a-b plane with an applied magnetic field
In this work, we show that one of the modes that ap- of 0.1 T. The magnetization versus temperature from 5 K
pears below TN , which was attributed to a phonon [45], to 300 K is shown in Fig. 1(d). Similar to earlier reports
is actually a quantized spin wave, i.e., a magnon. This [26,45,46], the strong anisotropy in FePS3 results in higher
antiferromagnetic magnon, with a frequency of 3.7 THz magnetization perpendicular to the a-b plane, which is along
(≈122 cm−1 ), softens with temperature more rapidly than the spin direction in the antiferromagnetic state. In a 2D
typical phonon modes and splits upon application of a mag- antiferromagnet, TN is defined to be the temperature where
netic field, as expected for antiferromagnetic magnons. The there is a maximum in the slope of the magnetization vs
frequency of the nondegenerate magnons depends linearly on temperature curve [26]. Thus, by taking the first derivative
magnetic field, with an effective g factor g ≈ 2. However, its of the magnetization with respect to temperature [Fig. 1(e)],
symmetry behavior with respect to the polarization vectors of TN is found to be ≈118 K. The broad maximum above TN in
the incoming and scattered light contradicts the long accepted the magnetization vs temperature for both parallel and perpen-
interpretation of the work of Fleury and Loudon [40], in that dicular orientations is typical for low-dimensional magnetic
Raman scattering of first-order magnon excitations involves systems and has been attributed to short-range, spin-spin
solely antisymmetric Raman tensors and can thus only be correlation [26].
observed in cross-polarized light configurations. Instead, we
observe the magnon in FePS3 in both parallel and cross con-
figurations, showing that the established magnon symmetry B. Raman Spectroscopy and Emergence of New Modes
rule lacks the generality that has been suggested by previous The majority of the peaks in the Raman spectrum
Raman literature. We explain the symmetry of the observed [Fig. 1(f)] can be ascribed to phonons. From the symmetry of
magnon using the magnetic point group with the inclusion bulk FePS3 , there are 30 zone-center phonons with irreducible
of complex tensor elements. Because of the small interlayer representations:  = 8Ag + 6Au + 7Bg + 9Bu , of which only
exchange coupling, and thus quasi-2D magnetic nature of bulk the Ag and Bg modes are Raman active. Figure 1(f) shows
FePS3 , the magnon observed herein is also expected to be the Raman spectra of bulk FePS3 (laser excitation wavelength
quasi-2D. To the best of our knowledge, this work represents λ = 514.5 nm) in the 180° backscattering geometry at T =
the first verification of a magnon in a quasi-2D magnet using 135 K and 15 K, which correspond to above and well below
magneto-Raman spectroscopy. Furthermore, this work high- the Néel temperature, respectively. Phonons from the  point
lights temperature-dependent, magneto-Raman spectroscopy (ω > 150 cm−1 ) are labeled by whether they are an Ag mode
as an important technique to explore properties of magnetic or a combination of Ag/Bg modes based on previous first
excitations, such as frequencies and lifetimes, and to aid in principles calculations as well as their measured orientation-
investigating quantum materials for next-generation devices. dependent frequency, since a peak that is a combination of
Ag/Bg modes will appear to shift in frequency as the light
polarization is rotated with respect to the crystal orientation
II. RESULTS AND DISCUSSIONS [46,52]. The phonon modes Ag2 /Bg2 , Ag3 , Ag4 /Bg4 , and Ag5
are mostly ascribed to vibrations of the (P2 S6 )4− unit, and
A. Initial Characterization: Sample Details and Magnetization have similar frequencies for the different X PS3 materials due
Large crystals of FePS3 were mechanically exfoliated us- to the lack of contribution from the metal atom [52]. While
ing adhesive tape [47] onto 300-nm SiO2 thermally grown on the Ag and Ag/Bg modes in the spectra are similar in terms
Si(100). Thicker flakes were used for measurements due to the of frequency and intensity when comparing above and below
more intense Raman scattering signal. Bulk FePS3 crystallizes the Néel temperature, significant changes are seen in the
in the monoclinic structure with the point group 2/m [48]. In modes below 150 cm−1 . Above the Néel temperature, there
the a-b plane, as is shown in Fig. 1(a), the Fe atoms form a is a broad, asymmetric peak “ψ” that exists even up to room
honeycomb lattice. In the center of the Fe honeycomb lattice temperature. At T = 15 K, however, there are four new modes
is a [P2 S6 ]4− unit, where the P-P dimers orient normal to the that appear in this frequency range, labeled ψ1 , ψ2 , ψ3 , and
surface and coordinate with six S atoms. The magnetic struc- ψ4 .

064416-2
QUASI-TWO-DIMENSIONAL MAGNON IDENTIFICATION … PHYSICAL REVIEW B 101, 064416 (2020)

FIG. 1. (a) Top view of the crystal structure of FePS3 , where the Fe atoms form a honeycomb lattice. The a and b axes are labeled. (b)
Magnetic structure of the Fe atoms in the a-b plane below TN . The spins are colinear and normal to the surface, where pink indicates spin
“up” (pointing out of the page) and green indicates spin “down” (pointing into the page). The ferromagnetic chains are aligned along the
a axis [48]. (c) Below TN , the layers are stacked antiferromagnetically, resulting in an increase in the unit cell (dashed outline) along the c
axis. (d) Magnetization (M) vs temperature (T) for a bulk crystal of FePS3 , where the applied magnetic field (B = 0.1 T) is parallel (orange)
or perpendicular (cyan) to the a-b plane. (e) dM/dT of (d) with an inflection point at TN ≈ 118 K. (f) Raman spectra above (T = 135 K) and
below (T = 15 K) TN .

The evolution of the Raman spectra of mode ψ into four work, we investigate the origin of ψ4 and unequivocally show
modes below TN has been previously observed [27,42,43], yet that it is a magnon and not a phonon.
there is uncertainty regarding the origin of each component.
It is known that antiferromagnetic ordering in FePS3 leads
to a doubling of the unit cell and subsequent zone folding, C. Temperature Dependence of Raman Modes
such that the M point in the paramagnetic Brillouin zone Figure 2(a) shows the temperature dependence of the
folds into the  point of the antiferromagnetically ordered Raman spectra from 110 K down to 40 K in steps of 10
Brillouin zone. From the phonon dispersion calculated by K. ψ4 is not observed until temperatures below approxi-
Wang et al. [46], ψ1 and ψ2 appear to be M point phonons mately 100 K, well below TN of ≈118 K. As the temperature
present below the Néel temperature due to magnetic-ordering decreases, ψ4 dramatically blueshifts (increases in Raman
induced zone folding. On the contrary, Lee et al. [45] attribute shift frequency), especially compared with the other modes,
these modes to very different origins. From density functional including ψ1 , ψ2 , ψ3 and the  point phonons. This effect
theory calculations, they suggest ψ2 is a  point phonon is illustrated in Fig. 2(b), which shows the frequencies of the
without magnetic ordering, while ψ3 and ψ4 have both  and different modes relative to their frequency at 15 K (ω − ω15K ),
M point components, and would thus appear above TN . Other including standard errors from fitting each peak with a Voigt
older publications also observed ψ1 -ψ4 , but only comment function. While the frequency changes of the other modes
on possible origins of ψ1 and ψ2 [43,53]. Thus, although only range from 0.01% to 0.57% (1–2 cm−1 shifts), as is typi-
the emergence of these modes can provide insight on the cal for anharmonic lattice effects, the frequency of ψ4 changes
transition to a magnetically ordered state in FePS3 , there is by as much as 6.2% (8 cm−1 shift) as the temperature is in-
still substantial debate as to the source of each mode. In this creased to 100 K. The strong shift in frequency with respect to

064416-3
AMBER MCCREARY et al. PHYSICAL REVIEW B 101, 064416 (2020)

temperature offers the first indication that this mode is at-

tributed to a magnon, as suggested by Sekine et al. for FePS3
[54]. Similar magnon temperature dependence has been re-
ported for three-dimensional (3D) antiferromagnets, such as
FeF2 [55,56], NiF2 [57], and MnF2 [58]. The assignment of
this mode to a magnon is further supported by recent inelastic
neutron scattering measurements that observed a  point
magnon at ≈15.1 meV (122 cm−1 ) [51,59], which coincides
in frequency to our observed mode ψ4 .

D. Magnons and Magneto-Raman Spectroscopy of FePS3

Due to the contrasting behavior of phonons and magnons
in an applied magnetic field, we utilize our unique magneto-
Raman capabilities to investigate the magnetic field depen-
dence of the modes in FePS3 to identify if they have magnetic
character. We detail our findings from the magnetic-field
dependent Raman spectra of FePS3 . Then, by discussing how
magnons behave under applied magnetic fields, we assign ψ4
as a Raman-active magnon.
An exfoliated flake of bulk FePS3 was cooled to T = 5 K,
after which a magnetic field was applied normal to the a-b
plane, i.e., parallel/antiparallel to the directions of the spins in
the antiferromagnetically ordered state. As seen in Fig. 3(a),
when the magnetic field is increased from 0 T to 9 T, ψ4 splits
into two modes, labeled as ψ4(1) and ψ4(2) , and the splitting in-
creases with the applied magnetic field. Figure 3(b) shows the
frequencies of ψ4(1) and ψ4(2) as a function of magnetic field,
where the splitting is symmetric for positive and negative
magnetic fields. From a linear fit, the extracted slopes for the
ψ4(1) and ψ4(2) branches are 0.93 ± 0.02 cm−1 /T and 0.94 ±
0.01 cm−1 /T, respectively. This splitting only occurs for ψ4 ,
where the other ψ modes and phonon modes remain as single
peaks. As discussed below, this behavior matches well with
the expected magnetic field dependence of antiferromagnetic
magnons.
Spin waves, which as quasiparticles are called magnons,
are collective excitations of the spins in magnetic materials.
The classical dynamics of the magnetization at the  point
in an antiferromagnet can be modeled following Keffer and
Kittel [60] as two interacting macrospins [Fig. 3(c)] repre-
senting the spin-up sublattice magnetization (M1 , pink) and
spin-down sublattice magnetization (M2 , green) that orient
in the +z and −z directions in equilibrium, respectively.
The excitation of a magnon causes the magnetization on
both sublattices to precess about their equilibrium directions.
Figures 3(d) and 3(e) show side and top view illustrations
of the two normal modes for M1 and M2 precession. The
frequencies of the normal modes are given by
ωk=0 = γ {(2HE + HA )HA }1/2 ± γ H0 , (1)
where HE is the exchange field, HA is the anisotropy field,
FIG. 2. (a) Temperature-dependent Raman spectra of FePS3 be- H0 is an applied field, and γ = gμ2πB μh̄ 0 is the gyromagnetic
low TN . The ψ4 mode appears below 100 K and its frequency ratio, which for free electrons (g ≈ 2.0023) [61] is approx-
dramatically increases with decreasing temperature. (b) Frequencies imately 28.025 GHz/T, or 0.9348 cm−1 /T. Without an ap-
of the various modes (relative to frequency at 15 K). While the plied magnetic field (H0 = 0), the frequencies of the magnon
other ψ modes and phonon modes only shift by 1–2 cm−1 , as is normal modes are degenerate (E = E0 ). However, an applied
expected by lattice anharmonic effects, the frequency of ψ4 shifts magnetic field H0 along z will cause a Zeeman splitting,
by as much as 8 cm−1 before it is no longer measurable above leading to two separate modes with E < E0 [Fig. 3(d)] and
100 K. E > E0 [Fig. 3(e)], both changing in magnetic field with a

064416-4
QUASI-TWO-DIMENSIONAL MAGNON IDENTIFICATION … PHYSICAL REVIEW B 101, 064416 (2020)

FIG. 3. (a) Magnetic-field dependent Raman of FePS3 at T = 5 K, showing the splitting of ψ4 into two components, ψ4(1) and ψ4(2) , where
the frequency of ψ4(1) (ψ4(2) ) decreases (increases) with increasing magnetic field. (b) Frequency vs magnetic field of ψ4(1) and ψ4(2) with the
slopes of the linear fits for the two branches. (c) If we consider two magnetization sublattices M1 (pink) and M2 (green) in FePS3 , in the ground
state M1 and M2 point in the +z and −z directions. (d), (e) The magnon spectrum includes two normal modes with net moments Mnet in the x,y
plane that precess in opposite directions. The degeneracy of these modes is broken in an applied field H0 ẑ with the energies of the modes in (d)
and (e), respectively, decreasing and increasing with H0 . In terms of this simple model, the modes in (d) and (e) would represent ψ4(1) and ψ4(2)
in (b), respectively.

slope of ±γ . From Eq. (1), a splitting of the mode with applied bulk FePS3 , the effective magnon g factor is g ≈ 1.99 ± 0.05.
magnetic field offers a conclusive method to determine if a From the full width at half maximum value of approximately
Raman mode is a magnon, as a magnetic field dependence 3 cm−1 , which includes instrument broadening, we estimate
is not expected for a typical phonon, absent magnetoelastic the lower bound of the magnon lifetime to be on the order of
effects (discussed below). 10 ps.
The behavior of the normal modes of magnons in a It should be noted that while it is possible that magneto-
magnetic field described above matches with the observed elastic effects could result in a phonon whose frequency
magnetic-field dependence of ψ4 , where the application of strongly shifts with temperature, the splitting of a phonon in
a magnetic field parallel to the spin direction leads to a a magnetic field due to symmetry lowering would require the
lifting of the degeneracy of the magnon frequency into two phonon to be degenerate. For FePS3 , only Ag and Bg modes are
components, ψ4(1) and ψ4(2) . The frequencies of the two Raman active, and both are singlets. Based on the splitting of
nondegenerate magnon modes obey a linear response as a ψ4 in a magnetic field, it cannot be a phonon. Thus, through
function of magnetic field, in agreement with Eq. (1), with magneto-Raman spectroscopy, we can unequivocally assign
the two slopes equal (within experimental uncertainty) to the mode ψ4 at ≈122 cm−1 as a magnon. The extremely
the slope γ = 0.9348 cm−1 /T for the free electron limit of weak magnon dispersion measured by neutron diffraction
g ≈ 2.0023. Using our experimental values of γ from the experiments [51] implies that this magnon we observed in
slopes of the linear fits in Fig. 3(b), we estimate that for bulk FePS3 herein is a quasi-2D magnon.

064416-5
AMBER MCCREARY et al. PHYSICAL REVIEW B 101, 064416 (2020)

E. Magnon Symmetry Behavior and Magnetic Point Group

Raman scattering from magnons in both ferromagnets and
antiferromagnets has been theoretically discussed by numer-
ous authors, with one of the most cited works by Fleury and
Loudon from 1968 [40]. In their work, the resulting polar-
ization selection rules of Raman scattering from first order
magnon excitations are purely antisymmetric with respect to
the polarization of the incoming (εi ) and scattered (εs ) light.
Thus, from their theory, a magnon mode can be observed only
in the perpendicular (or crossed) polarization configurations
(εi = εs ± 90◦ ), regardless of the orientation with respect to
the crystallographic axes. This theory has worked well to
describe a variety of ferromagnetic and antiferromagnetic
3D materials with Raman-observed magnons, including FeF2
[40], MnF2 [40], YFeO3 [62], and Cd1−x Mnx Te [63].
To probe the symmetry behavior of the magnon in FePS3 ,
εi , and εs were selected as shown in the inset of Fig. 4(a),
where the incoming light εi is at an arbitrary angle ϕ with
respect to the crystallographic b axis and the scattered light
εs is at a controlled angle θ with respect to εi . Figure 4(a)
shows that the magnon in FePS3 , can be observed in parallel
polarization (θ = 0◦ ) in addition to perpendicular polariza-
tion (θ = 90◦ ). When θ is varied between 0° and 360°, a
twofold symmetric pattern emerges where the magnon never
disappears, but instead has a fitted minimum at θ = 60◦
and maximum at θ = 150◦ , although these angle values are
dependent on ϕ.
The theory presented in Fleury and Loudon applies rigor-
ously to magnetic ions with orbital angular momentum ground
states of L = 0, or those with crystal-field quenching of the
ground state to L = 0 [41], as is the case for the well-studied
antiferromagnetic magnon example of FeF2 [64,65]. For ma-
terials that do not satisfy the L = 0 ground state condition,
such as FePS3 [66], the magnetic excitations are more difficult
to treat; the standard interpretation of the “antisymmetric”
nature of the magnon Raman tensor does not apply. Instead,
we use the symmetry properties of the magnetic point group
of FePS3 to understand the selection rules and polarization
dependence of the ψ4 magnon mode shown in Fig. 4.
When a material orders antiferromagnetically, the symme-
try generally lowers because the spin-up and spin-down sites
are no longer equivalent. In this regard, for the magnetic point
group, unitary operations are symmetry operations that pre-
serve the spin direction. Other operations that do not preserve
the spin direction may become allowed with the application FIG. 4. (a) Raman spectra (T = 10 K, B = 0 T) showing in-
of time reversal symmetry and are referred to as antiunitary tensity of magnon ψ4 in parallel (θ = 0◦ ) and cross (θ = 90◦ )
operations. The magnetic point group for bulk FePS3 actually polarization configurations at an arbitrary angle ϕ. Inset shows a
remains as 2/m, where the unit cell is doubled along the schematic defining angles θ and ϕ with respect to the incoming (εi )
c direction. The Raman tensors for this magnetic group are and scattered (εs ) light polarization and the a- and b- crystal axes. (b)
given by [67–69]: Polar plot of intensity of the magnon as a function of θ at a constant
angle ϕ (B = 0 T). The maximum peak intensity of ψ4 is normalized
⎛ ⎞ ⎛ ⎞ to unity, where the intensity scale ranges radially from 0 to 1.1.
B 0 D 0 F 0
Ag = ⎝ 0 C 0 ⎠, Bg = ⎝F 0 G⎠.
D 0 A 0 G 0 εs = (sin ϕ + θ , cos ϕ + θ , 0) are the incident and scattered
light polarization vectors, respectively.
The experimental 180° backscattering configuration incident It should be noted that if B, C, and F are purely real
on the honeycomb plane can access only the upper left numbers, the intensity profiles for both Ag and Bg modes as
2 × 2 block of these tensors. The Raman intensity is pro- a function of θ would necessarily have two nodes, i.e., two
portional to |εi · Ag (Bg ) · εs |2 , where εi = (sin ϕ, cos ϕ, 0) and angles where the intensity goes to zero, which is contrary

064416-6
QUASI-TWO-DIMENSIONAL MAGNON IDENTIFICATION … PHYSICAL REVIEW B 101, 064416 (2020)

to the polar plot we observed in Fig. 4(b). These nodes, verification of a quasi-2D magnon in a layered material using
however, disappear from the symmetry response when the magneto-Raman spectroscopy.
tensor elements are complex, which is the general case for
absorbing materials [70]: IV. EXPERIMENTAL DETAILS
   
B 0 0 F Bulk FePS3 crystals were purchased from 2D
Ag = , Bg = Semiconductors1 and then exfoliated via adhesive tape
0 CeδBC F 0
onto 300 nm SiO2 thermally grown on Si(100) [47]. Based
where δBC , is the relative complex phase factor of the real on optical contrast and the lack of the silicon mode observed
tensor elements B with respect to C. Because the com- through the exfoliated flake studied herein, we estimate the
plex phase factors are not relevant for the Bg modes, since thickness to be between 50 and 100 nm. Raman spectra were
the tensor elements are the same, the non-nodal nature of measured with the 514.5308-nm excitation wavelength of an
the experimental polar plot in FePS3 in Fig. 4(a) eliminates the Argon ion laser in the 180° backscattering configuration using
possibility that the magnon M is of purely Bg symmetry. a triple grating Raman spectrometer (Horiba JY T640001 ,
Instead, the polar plot in Fig. 4(a) can reproduced through 1800 mm−1 grating) coupled to a liquid-nitrogen-cooled CCD
multiple combinations of the B, C, BC , and the angle φ. detector. Polarization was selected and controlled using ultra
Thus, the symmetry response of the magnon ψ4 in FePS3 broadband polarizers and achromatic half wave plates. To per-
deviates strongly from the common interpretation of the work form temperature- and magnetic field dependent Raman spec-
by Fleury and Loudon [40], where magnons are assumed to be troscopy, the sample was placed into an attoDRY1000 cryostat
antisymmetric in nature for magnetic ions with L = 0 ground (attocube Inc.1 ), where the sample holder was pumped to
state. Our work shows that there is no reason, a priori, to ≈1 × 10−3 Pa (≈7 × 10−6 Torr), backfilled with helium gas,
expect that a magnon will appear only in cross-polarization and cooled. Micrometer-sized flakes were studied by focusing
configurations for these honeycomb-like structures, such as the laser with a white light camera onto the sample with a
FePS3 , α-RuCl3 [71], or CrI3 [72,73], without considering low-temperature, magnetic field compatible objective (50×,
the magnetic point group symmetry and associated Raman N.A. = 0.82) and xyz nanopositioners. Integration times
tensors. were approximately 12 min and the laser power was kept
In principle, one should be able to use the generators of below 300 µW to reduce local heating of the sample. Spectra
rotations (i.e., angular momenta Jx , Jy , and Jz ) generally given with applied magnetic field were corrected for Faraday
in the character table as guides for the magnon symmetry rotation in the objective using half wave plates external to the
behavior. For example, in FePS3 where the spins are oriented cryostat. Magnetization vs temperature measurements were
along the z direction (normal to a-b plane), the excitation of performed on a SQUID-based vibrating sample magnetometer
a magnon reduces the magnitude of the magnetization Mz , (SQUID-VSM; Quantum Design, Inc.1 ). The piece of the
which is then converted to the precession of the spins in the unexfoliated FePS3 crystal was mounted onto a quartz holder
x-y (a-b) plane, Mx and My , as depicted in Figs. 3(d) and 3(e). using GE varnish (LakeShore Cryotronics, Inc.1 ), and then
Thus, a magnon in FePS3 , which is a magnetization rotation measured under vacuum (<1 kPa). The magnetization was
in the x-y plane, can only have the same symmetry as Jx and measured under an applied magnetic field of μ0 H = 0.1 T
Jy , which are generators of rotations. from 5 K to 300 K and back down to 5 K. The temperature was
varied in steps of 5 K between 5 K and 300 K, 1 K between
30 K and 150 K, and then 5 K between 150 K and 300 K.
III. CONCLUSIONS
In conclusion, we examined the temperature- and ACKNOWLEDGMENTS
magnetic-field-dependent Raman spectra of bulk FePS3 ,
where new Raman-active modes appear in the antiferromag- A.M., T.T.M., and A.R.H.W. acknowledge the National
netically ordered state. One of these modes (ψ4 ), with a fre- Institute of Standards and Technology (NIST)/National Re-
quency of ≈122 cm−1 (15.1 meV), was previously assigned as search Council Postdoctoral Research Associateship Program
a phonon mode appearing due to zone folding. However, the and NIST-STRS (Scientific and Technical Research and Ser-
strong temperature shift (compared to other phonon modes), vices) for funding. Work at The Ohio State University was
splitting, and linear shifting of the mode with applied mag- supported by the Center for Emergent Materials, an NSF
netic field observed herein allows us to unequivocally assign MRSEC under Grant DMR-1420451. We would also like to
this mode as a magnon. We also investigated the magnon acknowledge Professors N. Drichko and Y.-M. Lu for insight-
symmetry behavior, which is not purely antisymmetric, and ful discussions and Hans Nembach for a careful read-through
explained its behavior using the magnetic point group of of the paper.
FePS3 (2/m). This work will aid in future studies of mag-
netic excitations in similar magnetic layered materials, such
as α-RuCl3 and CrI3 , and demonstrates temperature- and 1
Certain commercial equipment, instruments, or materials are
magnetic-field-dependent Raman spectroscopy as a technique identified in this manuscript in order to specify the experimental
to probe magnon phenomena in 2D materials. The quasi-2D procedure adequately. Such identification is not intended to imply
magnetic nature of bulk FePS3 , with weak interlayer exchange recommendation or endorsement by the National Institute of Stan-
coupling, indicates that the magnon in bulk FePS3 is also dards and Technology, nor is it intended to imply that the materials
quasi-2D. To the best of our knowledge, this is the first or equipment are necessarily the best available for the purpose.

064416-7
AMBER MCCREARY et al. PHYSICAL REVIEW B 101, 064416 (2020)

[1] K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Hight Walker, S. Barlow, S. R. Marder, C. A. Hacker, and
Zhang, S. V. Dubonos, I. V. Grigorieva, and A. A. Firsov, S. J. Pookpanratana, Controllable, wide-ranging n-doping and
Electric field effect in atomically thin carbon films, Science 306, p-doping of monolayer group 6 transition-metal disulfides and
666 (2004). diselenides, Adv. Mater. 30, 1802991 (2018).
[2] Q. H. Wang, K. Kalantar-Zadeh, A. Kis, J. N. Coleman, and M. [19] Y. Q. Guo, K. Xu, C. Z. Wu, J. Y. Zhao, and Y. Xie, Surface
S. Strano, Electronics and optoelectronics of two-dimensional chemical-modification for engineering the intrinsic physical
transition metal dichalcogenides, Nature Nanotechnol. 7, 699 properties of inorganic two-dimensional nanomaterials, Chem.
(2012). Soc. Rev. 44, 637 (2015).
[3] K. S. Novoselov, A. Mishchenko, A. Carvalho, and A. H. C. [20] M. Y. Li, C. H. Chen, Y. M. Shi, and L. J. Li, Heterostructures
Neto. 2D materials and van der Waals heterostructures, Science based on two-dimensional layered materials and their potential
353, aac9439 (2016). applications, Mater. Today 19, 322 (2016).
[4] D. S. Schulman, A. J. Arnold, A. Razavieh, J. Nasr, and S. Das, [21] A. V. Chumak, V. I. Vasyuchka, A. A. Serga, and B. Hillebrands,
The prospect of two-dimensional heterostructures: a review of Magnon Spintronics, Nat. Phys. 11, 453 (2015).
recent breakthroughs, IEEE Nanotechnol. Mag. 11, 6 (2017). [22] W. Xing, L. Qiu, X. Wang, Y. Yao, Y. Ma, R. Cai, S. Jia,
[5] Z. Lin, A. McCreary, N. Briggs, S. Subramanian, K. Zhang, Y. X. C. Xie, and W. Han, Magnon Transport in Quasi-Two-
Sun, X. Li, N. J. Borys, H. Yuan, and S. K. Fullteron-Shirey Dimensional Van Der Waals Antiferromagnets, Phys. Rev. X
et al., 2D materials advances: From large scale synthesis and 9, 011026 (2019).
controlled heterostructures to improved characterization tech- [23] C. Gong, L. Li, Z. Li, H. Ji, A. Stern, Y. Xia, T. Cao, W.
niques, defects and applications, 2D Mater. 3, 042001 (2016). Bao, C. Wang, and Y. Wang et al., Discovery of intrinsic
[6] K. F. Mak and J. Shan, Photonics and optoelectronics of 2D ferromagnetism in two-dimensional van der Waals crystals,
semiconductor transition metal dichalcogenides, Nat. Photon. Nature (London) 546, 265 (2017).
10, 216 (2016). [24] B. Huang, G. Clark, E. Navarro-Moratalla, D. Klein, R. Cheng,
[7] J. R. Schaibley, H. Yu, G. Clark, P. Rivera, J. S. Ross, K. L. K. Seyler, D. Zhong, E. Schmidgall, M. A. McGuire, and D.
Seyler, W. Yao, and X. Xu, Valleytronics in 2D materials, Nat. H. Cobden et al., Layer-dependent ferromagnetism in a van der
Rev. Mater. 1, 16055 (2016). Waals crystal down to the monolayer limit, Nature (London)
[8] S. Manzeli, D. Ovchinnikov, D. Pasquier, O. V. Yazyev, and A. 546, 270 (2017).
Kis. 2D Transition metal dichalcogenides, Nat. Rev. Mater. 2, [25] S. W. Jiang, L. Z. Li, Z. F. Wang, K. F. Mak, and J. Shan,
17033 (2017). Controlling magnetism in 2D CrI3 by electrostatic doping, Nat.
[9] D. Akinwande, C. J. Brennan, J. S. Bunch, P. Egberts, J. Nanotechnol. 13, 549 (2018).
R. Felts, H. Gao, R. Huang, J. -S. Kim, T. Li, and Y. Li [26] P. A. Joy and S. Vasudevan, Magnetism in the layered
et al., A review on mechanics and mechanical properties of transition-metal thiophosphates MPS3 (M = Mn, Fe, and Ni),
2D materials-graphene and beyond, Extreme Mech. Lett. 13, Phys. Rev. B 46, 5425 (1992).
42 (2017). [27] M. Balkanski, M. Jouanne, G. Ouvrard, and M. Scagliotti,
[10] K. S. Burch, D. Mandrus, and J. G. Park, Magnetism in two- Effects due to spin ordering in layered MPX3 compounds
dimensional van der Waals materials, Nature (London) 563, 47 revealed by inelastic light-scattering, J. Phys. C 20, 4397
(2018). (1987).
[11] C. Gong and X. Zhang, Two-dimensional magnetic crystals [28] B. L. Chittari, Y. Park, D. Lee, M. Han, A. H. MacDonald,
and emergent heterostructure devices, Science 363, eaav4450 E. Hwang, and J. Jung, Electronic and magnetic properties of
(2019). single-layer MPX3 metal phosphorous trichalcogenides, Phys.
[12] C. Lee, X. D. Wei, J. W. Kysar, and J. Hone, Measurement of the Rev. B 94, 184428 (2016).
elastic properties and antrinsic strength of monolayer graphene, [29] C. R. S. Haines, M. J. Coak, A. R. Wildes, G. I. Lampronti, C.
Science 321, 385 (2008). Liu, P. Nahai-Williamson, H. Hamidov, D. Daisenberger, and
[13] A. Castellanos-Gomez, M. Poot, G. A. Steele, H. S. J. van der S. S. Saxena, Pressure-induced electronic and structural phase
Zant, N. Agrait, and G. Rubio-Bollinger, Elastic properties of evolution in the van der waals compound FePS3 , Phys. Rev.
freely suspended MoS2 nanosheets, Adv. Mater. 24, 772 (2012). Lett. 121, 266801 (2018).
[14] B. Huang, G. Clark, D. R. Klein, D. MacNeill, E. Navarro- [30] M. Tsurubayashi, K. Kodama, M. Kano, K. Ishigaki, Y.
Moratalla, K. L. Seyler, N. Wilson, M. A. McGuire, D. H. Uwatoko, T. Watanabe, K. Takase, and T. Takano, Metal-
Cobden, and D. Xiao et al., Electrical control of 2D magnetism insulator transition in mott-insulator FePS3 , AIP Adv. 8, 101307
in bilayer CrI3 . Nat. Nanotechnol. 13, 544 (2018). (2018).
[15] S. W. Jiang, J. Shan, and K. F. Mak, Electric-field switching of [31] K. C. Rule, G. J. McIntyre, S. J. Kennedy, and T. J. Hicks,
two-dimensional van der Waals magnets, Nat. Mater. 17, 406 Single-crystal and powder neutron diffraction experiments on
(2018). FePS3 : search for the magnetic structure, Phys. Rev. B 76,
[16] J. S. Ross, P. Klement, A. M. Jones, N. J. Ghimire, J. Yan, D. 134402 (2007).
G. Mandrus, T. Taniguchi, K. Watanabe, K. Kitamura, and W. [32] X.-L. Li, X.-F. Qiao, W.-P. Han, Y. Lu, Q.-H. Tan, X.-L. Liu,
Yao et al., Electrically tunable excitonic light-emitting diodes and P.-H. Tan, Layer number identification of intrinsic and
based on monolayer WSE2 p-n junctions, Nat. Nanotechnol. 9, defective multilayered graphenes up to 100 layers by the Raman
268 (2014). mode intensity from substrates, Nanoscale 7, 8135 (2015).
[17] D. Akinwande, N. Petrone, and J. Hone, and Two-dimensional [33] L. B. Liang, J. Zhang, B. G. Sumpter, Q.-H. Tan, P.-H. Tan, and
flexible nanoelectronics, Nat. Commun. 5, 5678 (2014). V. Meunier, Low-frequency shear and layer-breathing modes in
[18] S. Zhang, H. M. Hill, K. Moudgil, C. A. Richter, A. R.

064416-8
QUASI-TWO-DIMENSIONAL MAGNON IDENTIFICATION … PHYSICAL REVIEW B 101, 064416 (2020)

raman scattering of two-dimensional materials, ACS Nano 11, study on MnPS3 and FePS3 , J. Phys. Soc. Jpn. 52, 3919
11777 (2017). (1983).
[34] T. M. G. Mohiuddin, A. Lombardo, R. R. Nair, A. Bonetti, [50] K. C. Rule, S. J. Kennedy, D. J. Goossens, A. M. Mulders, and
G. Savini, R. Jalil, N. Bonini, D. M. Basko, C. Galiotis, and T. J. Hicks, Contrasting antiferromagnetic order between FePS
N. Marzari et al., Uniaxial strain in graphene by Raman spec- and MnPS, Appl. Phys. A. 74, S811 (2002).
troscopy: G peak splitting, Gruneisen parameters, and sample [51] D. Lancon, H. C. Walker, E. Ressouche, B. Ouladdiaf, K. C.
orientation, Phys. Rev. B 79, 205433 (2009). Rule, G. J. McIntyre, T. J. Hicks, H. M. Ronnow, and A.
[35] R. Beams, L. G. Cancado, and L. Novotny, Raman character- R. Wildes, Magnetic structure and magnon dynamics of the
ization of defects and dopants in graphene, J. Phys.-Condens. quasi-two-dimensional antiferromagnet FePS3 , Phys. Rev. B
Mat. 27, 083002 (2015). 94, 214407 (2016).
[36] E. del Corro, A. Botello-Mendez, Y. Gillet, A. L. lias, H. [52] A. Hashemi, H. P. Komsa, M. Puska, and A. V. Krasheninnikov,
Terrones, S. Feng, C. Fantini, D. RHodes, N. Pradhan, and L. Vibrational properties of metal phosphorus trichalcogenides
Balicas et al., Atypical exciton-phonon interactions in WS2 and from first-principles calculations, J. Phys. Chem. C 121, 27207
WSe2 monolayers revealed by resonance raman spectroscopy, (2017).
Nano Lett. 16, 2363, (2016). [53] M. Scagliotti, M. Jouanne, M. Balkanski, and G. Ouvrard,
[37] A. McCreary, J. R. Simpson, Y. Wang, D. Rhodes, K. Fujisawa, Spin dependent phonon Raman-scattering in antiferromagnetic
L. Balicas, M. Dubey, V. H. Crespi, M. Terrones, and A. R. FePS3 layer-type compound, Solid State Commun. 54, 291
Hight Walker, Intricate resonant Raman response in anisotropic (1985).
ReS2 , Nano Lett. 17, 5897 (2017). [54] T. Sekine, M. Jouanne, C. Julien, and M. Balkanski, Light-
[38] K. N. Zhang, C. Bao, Q. Gu, X. Ren, H. Zhang, K. Deng, Y. scattering study of dynamic behavior of antiferromagnetic spins
Wu, Y. Li, J. Feng, and S. Zhou, Raman signatures of inversion in the layered magnetic semiconductor FePS3 , Phys. Rev. B 42,
symmetry breaking and structural phase transition in type-ii 8382 (1990).
Weyl semimetal MoTe2 , Nat. Commun. 7, 13552 (2016). [55] P. A. Fleury, S. P. S. Porto, L. E. Cheesman, and H. J.
[39] Y. Tian, M. J. Gray, H. W. Ji, R. J. Cava, and K. S. Burch, Guggenheim, Light Scattering by Spin Waves in FeF2 , Phys.
Magneto-elastic coupling in a potential ferromagnetic 2D Rev. Lett. 17, 84 (1966).
Atomic Crystal, 2D mater. 3, 025035 (2016). [56] R. C. Ohlmann and M. Tinkham, Antiferromagnetic resonance
[40] P. A. Fleury and R. Loudon, Scattering of light by one- and in FeF2 at far-infrared frequencies, Phys. Rev. 123, 425 (1961).
two-magnon excitations, Phys. Rev. 166, 514 (1968). [57] P. A. Fleury, Paramagnetic spin waves and correlation functions
[41] A. Sethi, T. Byrum, R. D. McAuliffe, S. L. Gleason, J. E. in NiF2 , Phys. Rev. 180, 591 (1969).
Slimak, D. P. Shoemaker, and S. L. Cooper, Magnons and mag- [58] F. M. Johnson and A. H. Nethercot, Antiferromagnetic Reso-
netodielectric effects in CoCr2 O4 : Raman scattering studies, nance in MnF2 , Phys. Rev. 114, 705 (1959).
Phys. Rev. B 95, 174413 (2017). [59] A. R. Wildes, K. C. Rule, R. I. Bewley, M. Enderle, and T. J.
[42] M. Scagliotti, M. Jouanne, M. Balkanski, G. Ouvrard, and Hicks, The Magnon dynamics and spin exchange parameters of
G. Benedek, Raman-scattering in antiferromagnetic FePS3 and FePS3 , J. Phys.-Condens Mat. 24, 416004 (2012).
FePSe3 Crystals, Phys. Rev. B 35, 7097 (1987). [60] F. Keffer and C. Kittel, Theory of antiferromagnetic resonance,
[43] M. Bernasconi, G. L. Marra, G. Benedek, L. Miglio, M. Phys. Rev. 85, 329 (1952).
Jouanne, C. Julien, M. Scagliotti, and M. Balkanski, Lattice- [61] P. J. Mohr, D. B. Newell, and B. N. Taylor, CODATA recom-
dynamics of layered MPX3 (M = Mn, Fe, Ni, Zn, X = S, Se) mended values of the fundamental physical constants: 2014,
Compounds, Phys. Rev. B 38, 12089 (1988). J. Phys. Chem. Ref. Data 45, 043102 (2016).
[44] C.-T. Kuo, M. Neumann, K. Balamurugan, H. J. Park, S. Kang, [62] R. M. White, R. J. Nemanich, and C. Tsang, Raman-scattering
H. W. Shiu, J. H. Kang, B. H. Hong, M. Han, T. W. Noh, and from magnons in rare-earth ortho-ferrites, J. Magn. Magn.
J.-G. Park, Exfoliation and Raman spectroscopic fingerprint of Mater. 15–18, 773 (1980).
few-layer NiPS3 van der Waals crystals, Sci. Rep. 6, 20904 [63] S. Venugopalan, A. Petrou, R. R. Galazka, A. K. Ramdas, and
(2016). S. Rodriguez, Raman-scattering by phonons and magnons in
[45] J.-U. Lee, S. Lee, S. H. Ryoo, S. Kang, T. Y. Kim, P. Kim, C.-H. semimagnetic semiconductors - Cd1−x Mnx Te, Phys. Rev. B 25,
Park, J.-G. Park, and H. Cheong, Ising-type magnetic ordering 2681 (1982).
in atomically thin FePS3 , Nano Lett. 16, 7433 (2016). [64] D. I. Khomskii, Transition Metal Compounds (Cambridge Uni-
[46] X. Z. Wang, K. Du, Y. Y. F. Liu, P. Hu, J. Zhang, Q. Zhang, versity Press, Cambridge, 2014).
M. H. S. Owen, X. Lu, C. K. Gan, and P. Sengupta, Raman [65] J. W. Stout and S. A. Reed, The crystal structure of MnF2 ,
spectroscopy of atomically thin two-dimensional magnetic iron FeF2 , CoF2 , NiF2 and ZnF2 , J. Am. Chem. Soc. 76, 5279
phosphorus trisulfide (FePS3 ) crystals, 2D Mater. 3, 031009 (1954).
(2016). [66] P. Jernberg, S. Bjarman, and R. Wappling, FePS: A first-order
[47] K. S. Novoselov, D. Jiang, F. Schedin, T. J. Booth, V. V. phase transition in a “2D” Ising antiferromagnet, J. Magn.
Khotkevich, S. V. Morozov, and A. K. Geim, Two-dimensional Magn. Mater. 46, 178 (1984).
atomic crystals, P. Natl. Acad. Sci. USA 102, 10451 [67] M. I. Aroyo, A. Kirov, C. Capillas, J. M. Perez-Mato, and H.
(2005). Wondratschek, Bilbao crystallographic server: II. Representa-
[48] G. Leflem, R. Brec, G. Ouvard, A. Louisy, and P. Segransan, tions of crystallographic point groups and space groups, Acta
Magnetic-interactions in the layer compounds MPX3 (M = Mn, Crystallogr. A 62, 115 (2006).
Fe, Ni; X = S, Se), J. Phys. Chem. Solids 43, 455 (1982). [68] M. I. Aroyo, J. M. Perez-Mato, C. Capillas, E. Kroumova, S.
[49] K. Kurosawa, S. Saito, and Y. Yamaguchi, Neutron-diffraction Ivantchev, G. Madariaga, A. Kirov, and H. Wondratschek, Bil-

064416-9
AMBER MCCREARY et al. PHYSICAL REVIEW B 101, 064416 (2020)

bao crystallographic server: I. Databases and crystallographic and R. V. Aguilar, Polarization-resolved Raman spectroscopy
computing programs, Z. Kristallogr. 221, 15 (2006). of α-RuCl3 and evidence of room-temperature two-dimensional
[69] M. I. Aroyo, J. M. Perez-Mato, D. Orobengoa, E. Tasci, G. de magnetic scattering, Phys. Rev. B 100, 134419 (2019).
la Flor, and A. Kirov, Crystallography online: Bilbao crystallo- [72] W. Jin, H. H. kim, Z. Ye, S. Li, P. Rezaie, F. Diaz, S. Siddiq, E.
graphic server, Bulg. Chem. Commun. 43, 183 (2011). Wauer, B. Yang, C. Li, and S. Tian et al., Raman fingerprint
[70] H. B. Ribeiro, M. A. Pimenta, C. J. S. de Matos, R. L. Moreira, of two terahertz spin wave branches in a two-dimensional
A. S. Rodin, J. D. Zapata, E. A. T. de Souza, and A. H. C. Neto, honeycomb Ising ferromagnet, Nat. Commun. 9, 5122 (2018).
Unusual angular dependence of the Raman response in black [73] A. McCreary, T. T. Mai, F. G. Utermohlen, J. R. Simpson, K.
phosphorus, ACS Nano 9, 4270 (2015). F. Garrity, X. Feng, D. Shcherbakov, Y. Zhu, J. Hu, D. Weber
[71] T. T. Mai, A. McCreary, P. Lampen-Kelley, N. Butch, J. R. et al., Distinct magneto-Raman signatures of spin-flip phase
Simpson, J.-Q. Yan, S. E. Nagler, D. Mandrus, A. R. H. Walker, transitions in CrI3 , arXiv:1910.01237.

064416-10