Communication

2D Ferromagnetism www.advmat.de

Chemically Exfoliated VSe2 Monolayers with Room-

Temperature Ferromagnetism
Wei Yu, Jing Li, Tun Seng Herng, Zishen Wang, Xiaoxu Zhao, Xiao Chi, Wei Fu,
Ibrahim Abdelwahab, Jun Zhou, Jiadong Dan, Zhongxin Chen, Zhi Chen, Zejun Li,
Jiong Lu, Stephen J. Pennycook, Yuan Ping Feng, Jun Ding, and Kian Ping Loh*

2D intrinsic ferromagnets with high

Among van der Waals layered ferromagnets, monolayer vanadium diselenide Curie temperature (Tc) are desirable for
(VSe2) stands out due to its robust ferromagnetism. However, the exfoliation spintronic applications. The atomic thick-
of monolayer VSe2 is challenging, not least because the monolayer flake is ness of van der Waals (vdW) crystals
extremely unstable in air. Using an electrochemical exfoliation approach with allow magnetic ordering to be electrically
controlled, which is useful in memory
organic cations as the intercalants, monolayer 1T-VSe2 flakes are successfully
applications.[1–3] At the fundamental level,
obtained from the bulk crystal at high yield. Thiol molecules are further ferromagnetic ordering at the 2D limit is a
introduced onto the VSe2 surface to passivate the exfoliated flakes, which fascinating topic because long range mag-
improves the air stability of the flakes for subsequent characterizations. netic order is predicted to be suppressed
Room-temperature ferromagnetism is confirmed on the exfoliated 2D VSe2 at finite temperature according to the
Mermin–Wagner theorem, however, recent
flakes using a superconducting quantum interference device (SQUID), X-ray
discoveries prove that there are excep-
magnetic circular dichroism (XMCD), and magnetic force microscopy (MFM), tions.[4] For instance, atomically thin CrI3
where the monolayer flake displays the strongest ferromagnetic properties. Se and Cr2Ge2Te6, were discovered as intrinsic
vacancies, which can be ubiquitous in such materials, also contribute to the ferromagnets by Xu et al. and Zhang
ferromagnetism of VSe2, although density functional theory (DFT) calculations et al., respectively.[5,6] This stimulates the
show that such effect can be minimized by physisorbed oxygen molecules or search for other members in this 2D fer-
romagnets family, which is soon joined
covalently bound thiol molecules.
by Fe3GeTe2 and VSe2.[7,8] Among them,
monolayer VSe2 (1T and 2H phase) is
notable for its high Tc, due to the strong
W. Yu, Dr. J. Li, Dr. X. Chi, Dr. W. Fu, I. Abdelwahab, Dr. Z. X. Chen,
electron coupling in 3d1 odd-electronic configuration of V4+.[9–13]
Dr. Z. Chen, Dr. Z. Li, Prof. J. Lu, Prof. K. P. Loh Besides ferromagnetism, enhanced charge density wave (CDW)
Department of Chemistry transition temperature, metal–insulator transition, and valley
National University of Singapore polarization have been demonstrated or predicted to exist in
3 Science Drive 3, Singapore 117543, Singapore monolayer VSe2 as well.[14–16] Recently, single-layer 1T-VSe2 was
E-mail: [email protected]
successfully grown by Batzill et al. using molecular beam epitaxy
Dr. J. Li, Z. Wang, Dr. W. Fu, I. Abdelwahab, Dr. Z. Li, Prof. J. Lu,
Prof. Y. P. Feng, Prof. K. P. Loh (MBE) on vdW substrate including highly-oriented pyrolytic
Centre for Advanced 2D Materials and Graphene Research Centre graphite (HOPG) and MoS2 and its robust room-temperature
National University of Singapore ferromagnetism was confirmed by magnetic measurements.[8]
Singapore 117546, Singapore The origin of the ferromagnetism in monolayer VSe2 is still
Dr. T. S. Herng, Dr. X. Zhao, J. Dan, Prof. S. J. Pennycook, Prof. J. Ding under debate, some theoretician has questioned if the ferro-
Department of Materials Science and Engineering
National University of Singapore
magnetism is intrinsic in nature since calculations suggest that
9 Engineering Drive 1, Singapore 117575, Singapore CDW distortion can remove the intrinsic ferromagnetic ground
Z. Wang, Dr. J. Zhou, Prof. Y. P. Feng state in VSe2.[17] Moreover, recent angle-resolved photoemission
Department of Physics (ARPES) studies on MBE grown monolayer VSe2 revealed the
National University of Singapore absence of ferromagnetic order as no exchange splitting was
2 Science Drive 3, Singapore 117551, Singapore observed.[15,18,19] These MBE grown VSe2 samples were grown
Dr. X. Chi on polycrystalline HOPG substrate, thus the films are coalesced
Singapore Synchrotron Light Source
National University of Singapore from micrometer-sized domains and contain a lot of grain
5 Research Link, Singapore 117603, Singapore boundaries and misoriented domains, which would induce
The ORCID identification number(s) for the author(s) of this article spin frustration. It is challenging to verify ferromagnetism in
can be found under https://doi.org/10.1002/adma.201903779. single crystalline VSe2 due to the difficulty to isolate monolayer
flake by micromechanical exfoliation; the theoretical interlayer
DOI: 10.1002/adma.201903779 distance of 1T-VSe2 (0.61 nm), as illustrated in Figure 1a, is

Adv. Mater. 2019, 1903779 1903779 (1 of 8) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 1. Illustration of the electrochemical exfoliation of 1T-VSe2. a) Top and side view of the atomic structure of layered 1T-VSe2 crystal (a = b = 3.35 Å,
c = 6.1 Å). b) Schematic illustration of the electrochemical setup for cathode exfoliation of VSe2 using tetrapropylammonium cation (TPA) as the interca-
lant. Inset: photograph showing a small piece of VSe2 crystal clamped by two titanium plates (left), visible expansion of VSe2 cathode with a fluffy shape
(right). c–e) OM images of drop-casted ultrathin VSe2 flakes on SiO2 substrates. Inset: VSe2 dispersion solution after manual shaking (left) and its
strong Tyndall effect (right). d) Zoomed-in OM images of the red dashed square in (d). e) Large-sized VSe2 monolayer flake up to 120 µm. f) Statistical
histograms for the thickness distributions of exfoliated ultrathin VSe2 flakes and (inset histogram) lateral size distributions of VSe2 monolayers.

narrower than MoS2 (0.65 nm).[14] Another more challenging dispersed well. Meanwhile, this high boiling point solvent also
problem is that monolayer VSe2 is prone to oxidation and any acts as a protection barrier for the unstable ultrathin VSe2 flakes,
ferromagnetism would vanish quickly. One way around this such that the electrochemical exfoliation can be performed in
may be to use solution phase exfoliation methods to produce air. After electrochemical intercalation, manual shaking of the
monolayer flake and passivate it immediately in situ in solu- expanded crystal yielded VSe2 dispersion (Figure 1d, inset) with
tion. Motivated by recent successful electrochemical exfoliation a large percentage (93.9%) of micrometer-sized, ultrathin flakes
of 2D black phosphorous using organic ammonium cations as (one to five layers). To facilitate manual-shaking exfoliation, the
the intercalants, we performed organic-solvent based electro- working voltage and concentration of electrolyte have to be opti-
chemical exfoliation of 1T-VSe2 in this work.[20–22] The results mized to yield a cotton-like shape of expanded VSe2 crystal, as
show that large sized ultrathin VSe2 flakes with well-defined can be seen in the inset in Figure 1b (see the Supporting Infor-
thickness and size distribution could be obtained, and passi- mation). The VSe2 dispersion shows a strong Tyndall effect,
vated from air oxidation by thiol molecules. Magnetic measure- indicating the uniform dispersion of ultrathin flakes of VSe2 in
ments were performed to verify room-temperature ferromag- PC (Figure 1d, inset). Figure 1c–e shows the optical microscopy
netism of these exfoliated ultrathin VSe2 flakes. (OM) images of the drop-casted VSe2 flakes on SiO2-coated sil-
The ambient two-electrode electrochemical setup using for icon substrate, where majority of them are ultrathin flakes and
the cathode exfoliation of VSe2 crystal is illustrated in Figure 1b. lateral size of monolayer is up to 120 µm. Figure 1f presents
An organic quaternary ammonium salt with four C3 alkyl chains, the statistical histograms of the thickness and monolayer size of
tetrapropylammonium (TPA) chloride, was selected as the inter- exfoliated VSe2, as determined using OM, atomic force micro­
calants for exfoliation. Compared to traditional metal ions (Li+, scopy (AFM) and transmission electron microscopy (TEM).
Na+), TPA does not introduce significant charges into the host More than 90% of the VSe2 flakes are between one and five
structure, and its large size provides higher expansion efficiency layers with an average lateral size of ≈40 µm. Therefore, we can
of the crystal once it is intercalated. The electrolyte solvent we conclude that electrochemical exfoliation method is more effec-
used is propylene carbonate (PC), a polar organic solvent with tive than micromechanical exfoliation methods in generating
high boiling point (242 °C), in which VSe2 nanosheets can be large sized, monolayer VSe2 flake with a much higher yield.

Adv. Mater. 2019, 1903779 1903779 (2 of 8) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 2. Structural characterizations of exfoliated VSe2 flakes. AFM topography images of a) stepped and b) folded monolayer/bilayer VSe2 flakes
with height profile along the dashed line in (a). c) AFM height profile along the dashed line in (b) indicating the real single-layer thickness. d) EDS
elemental analysis of exfoliated VSe2 flake. Inset: V-K edge and Se-L edge mapping images. e) Atomic-resolution STEM image of exfoliated VSe2 flake.
f) Zoomed-in atomic image of the region highlighted by the red dashed square in (e). Purple circled V atoms, yellow circled Se atoms (bottom Se
in dashed). g) Fast Fourier transform (FFT) pattern of the VSe2 flake. h) Intensity line profile along the white line in (e) showing the in-plane lattice
constant a. i) Layer-dependent Raman spectra of VSe2 from monolayer to bulk. Inset: Raman mapping image on A1g modes of a few-layer VSe2 flake.

To analyze the morphology of the exfoliated VSe2 monolayers, (Figure 2d). The chemical stoichiometry of exfoliated VSe2 has
AFM and scanning transmission electron micro­scopy (STEM) been verified as V:Se = 1:1.84 ± 0.02 by X-ray photoelectron spec-
were performed. A clean flake with step heights of 2.0 and troscopy (XPS) spectra (Figure S2a, Supporting Information),
2.9 nm is shown in the AFM image in Figure 2a, these are which has also been confirmed by TEM quantitative analysis for
attributed to monolayer and bilayer thick VSe2, respectively. In Se vacancy concentrations of monolayer (≈11.0%) and bilayer
fact, the theoretical interlayer distance of each atomic layer is flakes (≈12.2%).
0.61 nm, the much larger heights are due to organic residues To study the atomic structure of the exfoliated VSe2, high-
and substrate effects, which has also been found in other liquid resolution TEM was performed. As shown in Figure 2e,f,
exfoliated samples.[23] To prove that this is the case, a folded atomic-resolution STEM image of a few-layer VSe2 flake reveals
edge in a monolayer sample was measured by AFM (Figure 2b). the octahedral 1T phase. Fast Fourier transform (FFT) pattern
From the height profile, we can clearly see that the top folded of the flake shows a sixfold rotational symmetric diffraction,
layer has a thickness of 1.0 ± 0.1 nm while the bottom layer due to the single-crystalline hexagonal structure (Figure 2g).
is 2.0 nm thick, suggesting that trapped solvent can give an The in-plane lattice constant a is revealed to be 3.35 ± 0.01 Å
error margin of 1 nm to the thickness (Figure 2c). Monolayer (0.58/√3), which agrees well with the documented constant of
VSe2 has also been examined by atomic-resolution STEM as 1T-VSe2 (Figure 2h).[24] To confirm the structure and crystal-
shown in Figure S1 (Supporting Information), showing Se linity of the exfoliated VSe2, Raman spectra and mapping were
vacancies. Energy dispersive X-ray spectroscopic (EDS) analysis studied as shown in Figure 2i. For 1T-VSe2, the in-plane E2g
of the exfoliated few-layer VSe2 flake indicates its high purity and out-of-plane A1g vibration peaks occur at 135 and 206 cm−1,

Adv. Mater. 2019, 1903779 1903779 (3 of 8) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 3. Covalent passivation of exfoliated VSe2 flakes. a) Schematic illustration of the covalent passivation processes (S1–S3) on exfoliated VSe2
flakes. S1, Drop-casting VSe2 dispersion on a SiO2 substrate and vacuum drying; S2, Capping the VSe2/SiO2 substrate by pure organic FDT solution;
S3, Removing FDT solution and washing with acetone. b) Schematic illustration for the covalent passivation mechanism bonding on Se defects by
the thiol group. c) XPS V2p spectra of bare and passivated VSe2 flakes on HOPG substrate after ambient exposures for different lengths of time.
d) AFM topography images of bare and passivated monolayer/bilayer VSe2 flakes after ambient exposure for different lengths of time. Top: bare VSe2,
0–60 min; Bottom: passivated VSe2, 0–24 h. Inset (middle images): AFM height profiles of the bare and passivated VSe2 flakes corresponding to the
white dashed lines in left image.

respectively.[25] The Raman spectrum of bulk VSe2 crystal introduced 1H, 1H, 2H, 2H-perfluorodecanethiol (FDT) to
agrees very well with that of 1T phase with two intensive peaks passivate the exfoliated VSe2 flakes.[27–29] The choice of FDT
at 135.8 and 207.0 cm−1, and a good phase retention is observed was also motivated by the presence of hydrophobic fluorine
from bulk to monolayer. However, it was observed that mono­ groups in the molecular chain, which can serve as a barrier
layer VSe2 flake is easily damaged by laser irradiation, and the to moisture permeation. The procedure for the covalent pas-
Raman peaks decrease in intensity with laser irradiation time. sivation of the exfoliated VSe2 is illustrated in Figure 3a. FDT
We had verified that flake is highly prone to laser damage by solution was introduced onto the surface of the sample to fully
monitoring the rapidly laser-induced amorphization of mono­ cover it for 1 d, followed by washing with acetone to remove
layer VSe2 using STEM (Figure S3, Supporting Information). the physically adsorbed molecules. All steps were performed in
Bilayer VSe2 is much more stable than monolayer ones and a glove box to avoid the degradation of VSe2 during the passi-
exhibits obvious Raman peaks. For typical 2D materials, the vation period. After this treatment, FDT molecules were found
ratio of A1g/E2g is related to the layer thickness and different to be anchored on VSe2 surface by covalent bond. The XPS
degrees of interlayer coupling.[26] From bulk to bilayer VSe2, the F1s and S3s spectra in Figure S4a,b (Supporting Information)
A1g/E2g intensity ratio is dramatically decreased from 3.77 to show strong F and weak S signals indicating the formation of
1.05. The peak intensity map of the out-of-plane A1g mode on a a FDT adlayer on the sample surface. To exclude the possibility
bilayer VSe2 flake in Figure 2i (inset) indicating a homogenous of physically adsorbed molecules, XPS was used to monitor
thickness and uniform 1T structure with high crystallinity. the F1s after annealing the sample in ultrahigh vacuum
Similar to what was observed previously, we find that 2D (UHV, pressure <5 × 10−10 mbar) above the boiling point of
VSe2 is extremely unstable in air and easily oxidized to form FDT (355 K atm−1). As shown in Figure S4c (Supporting Infor-
vanadium oxides (V2O5). XPS results of the exfoliated VSe2 mation), after annealing at 400 K in UHV for 1 h (return to
ultrathin nanosheets reveal that the Se/V stoichiometry ratio base pressure), the intensity of F1s peak was only slightly
is around 1.84, suggesting the presence of Se vacancies, which decreased. When the annealing temperature was increased
may be the origin of the air instability as metallic atoms with to 500 K, the F1s peak is dramatically weakened, and it disap-
unpaired electrons are prone to oxidation. One way of passi- pears after annealing at 600 K for 1 h. These results proved the
vating defects due to Se vacancies is to covalently bind thiol formation of strong chemical bonds between FDT and VSe2, as
molecules to the V-atoms at the Se-vacancies, for this we illustrated in Figure 3b.

Adv. Mater. 2019, 1903779 1903779 (4 of 8) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

To evaluate the effectiveness of this passivation method (≈2.0 and 3.0 nm, respectively). The thickness of the passivated
against air-induced degradation, OM, XPS, and AFM topography monolayer flake is slightly larger than the bare one by around
were used to investigate the chemical state and morphology of 0.2 nm, and this height difference is due to the anchored FDT
the VSe2 flakes. From the OM images in Figure S5 (Supporting molecules on VSe2 basal plane. Degradation of the samples
Information), we can see that the color contrast of the can be judged from the formation of particles on the surface.
as-exfoliated VSe2 (bare VSe2) monolayer becomes lighter after For bare VSe2 sample, the monolayer is severely degraded after
30 min of air exposure, whereas this change occurs more slowly 30 min, whereas the bilayer flake is fully degraded after 1 h
on bare bilayer VSe2. In contrast, both monolayer and bilayer (Figure 3d, top). For the passivated VSe2 sample, monolayer
FDT-passivated VSe2 (Passivated VSe2) flakes show almost no appears unchanged from the initial sample after 5 h, and only
color change up to 5 h, and bilayer flake exhibits slight color slight surface degradation occurs in the bilayer flake after 24 h
change only after 24 h. The oxidative degradation of VSe2 to of air exposure (Figure 3d, bottom). Therefore, AFM results
VOx species causes the flake to be increasingly transparent, confirmed effective passivation on both monolayer and bilayer
this is confirmed by V2p XPS spectra (Figure 3c), which shows VSe2.
that the valence state of vanadium is increased from V4+ to V5+ Ultrathin VSe2 flakes, scattered on SiO2 substrate, were ana-
as judged by the +3.7 eV chemical shift of the V2p3/2 peak. lyzed for its magnetic properties by collecting the magnetization-
The V2p spectra of the passivated VSe2 sample after 5 h of air magnetic field (M–H) curves on a superconducting quantum
exposure are almost the same with initial one. After 12 h of interference device (SQUID). In the presence of an in-plane
air exposure, the proportions of the oxidized V5+ (2p3/2 orbital magnetic field, obvious magnetic hysteresis loops were
electrons), as seen from the deconvoluted XPS spectra for the observed at both 10 and 300 K, providing direct evidence for
bare and passivated VSe2, are 67.4% and 13.2%, respectively the ferromagnetism of the exfoliated VSe2 sample (Figure 4a).
(Figure S6a, Supporting Information). Thus, XPS data cor- The M–H curves of VSe2 bulk crystal and SiO2 substrate were
roborate the evidence shown by the optical images regarding also investigated to eliminate the contribution of substrate and
the improved air-stability of the FDT-passivated VSe2 flakes. bulk crystal, which were determined to be paramagnetic and
Following air exposure, AFM study on both bare and passivated diamagnetic, respectively (Figure S7a,b, Supporting Informa-
VSe2 monolayer/bilayer flakes was also performed, as shown tion). The temperature-dependent saturated magnetization (Ms)
in Figure 3d. The inserted height profiles show the thick- of VSe2 in Figure 4b indicates a high Tc up to 470 K, which
nesses of both monolayer and bilayer as we discussed above is much higher than that observed in other exfoliated 2D

Figure 4. Ferromagnetism of exfoliated VSe2 flakes. a) M–H hysteresis loop of bare VSe2 flakes on SiO2 substrate under an in-plane magnetic field
at 300 and 10 K. b) Temperature-dependent saturated magnetization (Ms) of bare VSe2 from 10 to 500 K. Inset: M–H curve for the sample at 470 K
indicating the loose of magnetization. c) M–H hysteresis loop of bare and passivated VSe2 flakes at 300 K. d) XMCD spectra on V L-edge of bare and
passivated VSe2 flakes on HOPG substrate at room temperature (µH = ±2 T). e) MFM phase image of a passivated VSe2 monolayer/bilayer/trilayer
flake, the MFM signal is reflected from the phase color bar increasing from red to blue. Inset: 3D topography image. f) Topography height and MFM
phase profiles of the MFM image corresponding to the white dashed line in panel (e).

Adv. Mater. 2019, 1903779 1903779 (5 of 8) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

ferromagnets, including CrI3, Cr2Ge2Te6, and Fe3GeTe2. Mean- the phase shift of bilayer and trilayer (−0.08°). This result is
while, the zero-field cooling magnetization presents a signifi- consistent with previously reported thickness-dependent ferro-
cant jump at around 130 K corresponding to a CDW transition magnetism on MBE grown VSe2 samples.[8]
temperature (TCDW) of the VSe2 ultrathin flakes, which is con- In previous studies on the ferromagnetism of 2D VSe2, the
sistent with the values reported on MBE monolayer exfoliated role of atomic vacancies was ignored. Here, a systematic inves-
thin VSe2 (Figure S7c, Supporting Information).[8,30] tigation on the correlation between Se vacancy and magnetiza-
Monolayer VSe2 is highly unstable in the air and is easily tion was performed. First, Se vacancies were generated in VSe2
oxidized to vanadium oxides (V2O5) which is a diamagnetic sample by thermally annealing in UHV (600 °C, overnight).
insulator at room temperature.[31] To minimize the degradation The M–H hysteresis loops of VSe2 before and after thermal
effect, the magnetization of the chemically passivated VSe2 was annealing is presented in Figure 5a, where a remarkably
also investigated using SQUID at 300 K. Due to uncertainty in enhanced ferromagnetism over that of the intrinsic baseline
the mass of monolayer flakes, the values reported here are nom- level is observed. The reduced intensity of the XPS Se3d spectra
inal values and sample amounts are kept as the same by volume (normalized by the intensity of V2p spectra) for the thermally
of solution to achieve reliable comparison of magnetizations. annealed sample indicates an increased Se vacancy concentra-
A much enhanced Ms of 1.3 × 10−4 emu was obtained for the tion with Se/V ratio decreased from 1.83 to 1.76 (Figure 5a,
passivated samples as compared to bare sample (0.7 × 10−4 emu) inset). Exfoliated ultrathin VSe2 flakes undergo slow degrada-
at a similar coercive field (Hc) (≈120 Oe) (Figure 4c). Element- tion in PC solvent, leading to the formation of Se vacancies.
specific X-ray magnetic circular dichroism (XMCD) was also AFM topography images reveal increasing defect concentration
conducted to further demonstrate the passivation enhanced with time; this has also been verified by XPS analysis, indi-
ferromagnetism of the exfoliated VSe2 (Figure 4d). The V cating a decreased Se/V ratio from 1.83 to 1.72 (Figure S8a,b,
L-edge of VSe2 was investigated under room temperature with Supporting Information). Corresponding ferromagnetic inten-
two opposite directions of horizontal magnetic fields relative to sities were monitored by SQUID at 300 K and the Ms value
the fixed photon helicity of µ+ and µ− (µH = ±2 T), and the dif- shows an increased intensity from 0.9 × 10−4 to 1.8 × 10−4 emu,
ference of µ+ − µ− is defined as the XMCD signal. There are indicating Se vacancy enhances ferromagnetism (Figure 5b,
clear XMCD signals (µ+ − µ− is nonzero) for both bare and red curve and Figure S8c, Supporting Information). The cor-
passivated VSe2 samples, supporting the evidences of room- relation of Se vacancy and magnetic moment of monolayer
temperature ferromagnetism on VSe2 flakes, and the intensity VSe2 has also been studied using density functional theory
enhanced significantly after chemical passivation, where the (DFT) calculation. A similar rising trend of magnetic moment
total magnetic moment (spin and orbital magnetic moments) with increasing vacancy concentrations is obtained (Figure 5b,
increased from 0.094 µB to 0.315 µB per V atom.[32] The meas- purple curve). The magnetic moment of intact VSe2 monolayer
ured magnetic moment of the chemically passivated flakes is (Se/V = 2.0) from our DFT calculation is around 1.2 µB per
in the same order as that theoretically predicted for 1T-VSe2 V atom, which is consistent with previous calculation results
monolayer (≈0.10 µB).[9–11] We are aware there are a few reports indicating intrinsic ferromagnetism.[11] We have also calcu-
on MBE grown VSe2 film where no exchange splitting was lated the trend under different Hubbard U values, a similar
observed in ARPES, thus revealing the absence of ferromag- trend of increasing magnetic moments with Se vacancy con-
netism.[15,18,19] These VSe2 samples were grown by MBE on centrations is predicted (Figure S9, Supporting Information).
polycrystalline HOPG, thus the film inevitably contained a lot In addition, DFT calculations reveal that the adsorption of O2
of grain boundaries and rotationally disordered domains, which or thiol molecules on the Se vacancies decrease the magnetiza-
would lead to spin frustration. In contrast, our exfoliated VSe2 tions of VSe2 monolayer to near the intrinsic level due to the
monolayers are large size single crystal domains peeled from consumption of unpaired electrons for V around Se vacancies
high quality VSe2 crystal and randomly distributed on a sub- (Figure 5c). It suggests that although defects can contribute
strate (see Figure 1c). to ferromagnetism, they are not robust in air. Moreover, thiol
The air-stability of the passivated VSe2 flakes makes it pos- adsorption shows higher magnetization than O2 that agrees
sible to study the magnetic properties of individual flake using well with above SQUID and XMCD results of passivation
ambient magnetic force microscopy (MFM); this is rarely per- enhanced magnetism, demonstrating the protection effect on
formed on 2D ferromagnetic materials due to their extreme air Se vacancies against air O2. Thus thiol passivation can reduce
instability. As opposed to SQUID or XMCD measurements that defect-induced magnetization without affecting intrinsic fer-
only give an average signal, MFM provides spatially resolved romagnetism, and its presence is in fact necessary to prevent
magnetic signal that allows us to correlate the ferromagnetic air-induced degradation of the samples.
signal with the layer thickness. As shown in Figure 4e,f, pas- Although the measured magnetism from our exfoliated
sivated monolayer (2.3 nm), bilayer (3.4 nm), and trilayer VSe2 material most probably has mixed contributions from
(4.4 nm) VSe2 flakes, were investigated. Their height differ- both intrinsic and defect-related ferromagnetism, DFT calcua-
ences can be distinguished from the stepped height profile in tions show that passivation by thiol molecules on Se vacancy
Figure 4f (top) corresponding to the white dash line in Figure 4e sites should quench defect-related magentism. The fact that
and the inserted 3D topography image in Figure 4e. From the we managed to observe robust ferromagnetism most strongly
MFM phase image, it is obvious that the monolayer exhibits a on thiol-passivated monolayer flakes suggest that the intrinsic
much higher magnetic signal (larger phase contrast) compared ferromagnetism provides the lower limit of the observed ferro-
to thicker flakes. The MFM phase profile in Figure 4f (bottom) magnetism, while increasing defect concentration will increase
reveals that the monolayer (−0.18°) shows more than double it to the upper limit.

Adv. Mater. 2019, 1903779 1903779 (6 of 8) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 5. Se vacancy enhanced ferromagnetism of exfoliated VSe2 flakes. a) M–H curves of pristine and thermal annealed VSe2 samples. Inset: the
corresponding XPS Se3d spectra after normalization by V2p spectra, indicating the increased Se vacancies after thermal annealing. b) The variation
trends of theoretical (purple) and experimental (red) magnetization versus Se/V ratio of increasing vacancy concentrations produced by increasing the
chemical degradation time in solution. c) Theoretically calculated magnetizations of intact (VSe2), defective (VSe1.75), thiol (VSe1.75+thiol), and oxygen
(VSe1.75+O2) adsorbed monolayer VSe2, and their corresponding atomic structures.

By performing electrochemical exfoliation of VSe2 crystal work was supported by the Agency for Science and Technology, Singapore,
using organic intercalants, large sized ultrathin 1T-VSe2 AME-IRG funding under the project “Scalable Growth of Ultrathin
flakes with thickness fewer than five layers can be obtained Ferroelectric Materials for Memory Technologies (A1983c0035).” The
authors thank Dr. Fangfang Zhao at Bruker Singapore Pte. Ltd. for the
with >90% yield, of which the monolayer flakes dominate MFM measurements.
(43%). Using a combination of SQUID, XMCD, and MFM,
we confirmed the presence of room-temperature ferromag-
netism in monolayer 1T-VSe2 flakes with a magnetic moment
≈0.3 µB per V atom as determined by XMCD. Although we Conflict of Interest
are not able to conclude if all the measured magnetic signals The authors declare no conflict of interest.
are intrinsic in nature, both our experimental and computa-
tional results confirm that the presence of Se vacancies can
enhance the measured signal, thus claims of intrinsic ferro-
magnetism in VSe2 has to be qualified by generating a defect-
Keywords
free sample. 2D ferromagnetism, chemical exfoliation, covalent passivation, vacancy
enhanced ferromagnetism, VSe2 monolayers

Received: June 14, 2019

Supporting Information Revised: July 26, 2019
Supporting Information is available from the Wiley Online Library or Published online:
from the author.

[1] N. Samarth, Nature 2017, 546, 216.

Acknowledgements [2] C. Gong, X. Zhang, Science 2019, 363, 6428.
W.Y. and J.L. contributed equally to this work. K.P.L. thanks MOE Tier 3 [3] K. S. Burch, D. Mandrus, J. G. Park, Nature 2018, 563, 7729.
grant 2D crystal quantum exciton photonics MOE2018-T3-1-005. This [4] N. D. Mermin, H. Wagner, Phys. Rev. Lett. 1966, 17, 1133.

Adv. Mater. 2019, 1903779 1903779 (7 of 8) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

[5] B. Huang, G. Clark, E. Navarro-Moratalla, D. R. Klein, R. Cheng, [19] P. K. J. Wong, W. Zhang, F. Bussolotti, X. Yin, T. S. Herng,
K. L. Seyler, D. Zhong, E. Schmidgall, M. A. McGuire, D. H. Cobden, L. Zhang, Y. L. Huang, G. Vinai, S. Krishnamurthi, D. W. Bukhvalov,
W. Yao, Nature 2017, 546, 270. Y. J. Zheng, Adv. Mater. 2019, 18, 1901185.
[6] C. Gong, L. Li, Z. Li, H. Ji, A. Stern, Y. Xia, T. Cao, W. Bao, C. Wang, [20] C. Wang, Q. He, U. Halim, Y. Liu, E. Zhu, Z. Lin, H. Xiao, X. Duan,
Y. Wang, Z. Q. Qiu, Nature 2017, 546, 265. Z. Feng, R. Cheng, N. O. Weiss, Nature 2018, 555, 231.
[7] Z. Fei, B. Huang, P. Malinowski, W. Wang, T. Song, J. Sanchez, [21] J. Li, C. Chen, S. Liu, J. Lu, W. P. Goh, H. Fang, Z. Qiu, B. Tian,
W. Yao, D. Xiao, X. Zhu, A. F. May, W. Wu, Nat. Mater. 2018, 17, Z. Chen, C. Yao, W. Liu, Chem. Mater. 2018, 30, 2742.
778. [22] S. Yang, K. Zhang, A. G. Ricciardulli, P. Zhang, Z. Liao, M. R. Lohe,
[8] M. Bonilla, S. Kolekar, Y. Ma, H. C. Diaz, V. Kalappattil, R. Das, E. Zschech, P. W. Blom, W. Pisula, K. Müllen, X. Feng, Angew.
T. Eggers, H. R. Gutierrez, M. H. Phan, M. Batzill, Nat. Nanotechnol. Chem., Int. Ed. 2018, 130, 4767.
2018, 13, 289. [23] D. Hanlon, C. Backes, E. Doherty, C. S. Cucinotta, N. C. Berner,
[9] Y. Ma, Y. Dai, M. Guo, C. Niu, Y. Zhu, B. Huang, ACS Nano 2012, C. Boland, K. Lee, A. Harvey, P. Lynch, Z. Gholamvand, S. Zhang,
6, 1695. Nat. Commun. 2015, 6, 8563.
[10] F. Li, K. Tu, Z. Chen, J. Phys. Chem. C 2014, 118, 21264. [24] M. Esters, R. G. Hennig, D. C. Johnson, Phys. Rev. B 2017, 96,
[11] Z. I. Popov, N. S. Mikhaleva, M. A. Visotin, A. A. Kuzubov, S. Entani, 235147.
H. Naramoto, S. Sakai, P. B. Sorokin, P. V. Avramov, Phys. Chem. [25] Z. Zhang, J. Niu, P. Yang, Y. Gong, Q. Ji, J. Shi, Q. Fang, S. Jiang,
Chem. Phys. 2016, 18, 33047. H. Li, X. Zhou, L. Gu, Adv. Mater. 2017, 29, 1702359.
[12] H. R. Fuh, B. Yan, S. C. Wu, C. Felser, C. R. Chang, New J. Phys. [26] H. Li, Q. Zhang, C. C. R. Yap, B. K. Tay, T. H. T. Edwin, A. Olivier,
2016, 18, 113038. D. Baillargeat, Adv. Funct. Mater. 2012, 22, 1385.
[13] E. Marco, R. G. Hennig, D. C. Johnson, Phys. Rev. B 2017, 96, 235147. [27] S. S. Chou, M. De, J. Kim, S. Byun, C. Dykstra, J. Yu, J. Huang,
[14] J. Yang, W. Wang, Y. Liu, H. Du, W. Ning, G. Zheng, C. Jin, Y. Han, V. P. Dravid, J. Am. Chem. Soc. 2013, 135, 4584.
N. Wang, Z. Yang, M. Tian, Appl. Phys. Lett. 2014, 105, 063109. [28] E. P. Nguyen, B. J. Carey, J. Z. Ou, J. van Embden, E. D. Gaspera,
[15] G. Duvjir, B. K. Choi, I. Jang, S. Ulstrup, S. Kang, T. Thi Ly, S. Kim, A. F. Chrimes, J. Spencer, S. M. Zhuiykov, K. Kalantar-zadeh,
Y. H. Choi, C. Jozwiak, A. Bostwick, E. Rotenberg, Nano Lett. 2018, T. Daeneke, Adv. Mater. 2015, 27, 6225.
18, 5432. [29] D. M. Sim, M. Kim, S. Yim, M. J. Choi, J. Choi, S. Yoo, Y. S. Jung,
[16] W. Y. Tong, S. J. Gong, X. Wan, C. G. Duan, Nat. Commun. 2016, 7, ACS Nano 2015, 9, 12115.
13612. [30] Á. Pásztor, A. Scarfato, C. Barreteau, E. Giannini, C. Renner,
[17] P. Coelho, K. Nguyen-Cong, M. Bonilla, S. K. Kolekar, M. H. Phan, 2D Mater. 2017, 4, 041005.
J. Avila, M. C. Asensio, I. I. Oleynik, M. Batzill, J. Phys. Chem. C [31] Z. R. Xiao, G. Y. Guo, P. H. Lee, H. S. Hsu, J. C. A. Huang, J. Phys.
2019, 123, 14089. Soc. Jpn. 2008, 77, 023706.
[18] J. Feng, D. Biswas, A. Rajan, M. D. Watson, F. Mazzola, O. J. Clark, [32] C. T. Chen, Y. U. Idzerda, H. J. Lin, N. V. Smith, G. Meigs,
K. Underwood, I. Markovic, M. McLaren, A. Hunter, D. M. Burn, E. Chaban, G. H. Ho, E. Pellegrin, F. Sette, Phys. Rev. Lett. 1995, 75,
Nano Lett. 2018, 18, 4493. 152.

Adv. Mater. 2019, 1903779 1903779 (8 of 8) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim