Journal of Physics: Condensed

Matter

ACCEPTED MANUSCRIPT

Coexistence of multiple magnetic interactions in oxygen-deficient V2O5

nanoparticles
To cite this article before publication: Tathagata Sarkar et al 2024 J. Phys.: Condens. Matter in press https://doi.org/10.1088/1361-648X/ad76f2

Manuscript version: Accepted Manuscript

Accepted Manuscript is “the version of the article accepted for publication including all changes made as a result of the peer review process,
and which may also include the addition to the article by IOP Publishing of a header, an article ID, a cover sheet and/or an ‘Accepted
Manuscript’ watermark, but excluding any other editing, typesetting or other changes made by IOP Publishing and/or its licensors”

This Accepted Manuscript is © 2024 IOP Publishing Ltd. All rights, including for text and data mining, AI training, and similar
technologies, are reserved..

During the embargo period (the 12 month period from the publication of the Version of Record of this article), the Accepted Manuscript is fully
protected by copyright and cannot be reused or reposted elsewhere.

As the Version of Record of this article is going to be / has been published on a subscription basis, this Accepted Manuscript will be available for
reuse under a CC BY-NC-ND 3.0 licence after the 12 month embargo period.

After the embargo period, everyone is permitted to use copy and redistribute this article for non-commercial purposes only, provided that they
adhere to all the terms of the licence https://creativecommons.org/licences/by-nc-nd/3.0

Although reasonable endeavours have been taken to obtain all necessary permissions from third parties to include their copyrighted content
within this article, their full citation and copyright line may not be present in this Accepted Manuscript version. Before using any content from this
article, please refer to the Version of Record on IOPscience once published for full citation and copyright details, as permissions may be required.
All third party content is fully copyright protected, unless specifically stated otherwise in the figure caption in the Version of Record.

View the article online for updates and enhancements.

This content was downloaded from IP address 14.139.190.82 on 06/09/2024 at 10:38

Page 1 of 32 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2

1
2
3 REFERENCES 1
4

pt
5 Coexistence of Multiple Magnetic Interactions in Oxygen-
6
7 Deficient V2O5 Nanoparticles
8
9
Tathagata Sarkar1, Soumya Biswas1, Sonali Kakkar2, A. V. Raghu1, S.

cri
10
11 D. Kaushik3, Chandan Bera2, Vinayak B. Kamble1,*
12 1School of Physics, Indian Institute of Science Education and Research Thiruvananthapuram,
13 India 695551.
14 2Institute of Nano Science and Technology, Habitat Center, Phase-X, Mohali, Punjab 160062,
15
India.
16

us
3UGC – DAE Consortium for Scientific Research Mumbai Centre, Bhabha Atomic Research
17
18 Centre, Mumbai-400085, India
19
20 E-mail: [email protected]
21
22
23
24
25
26
27
28
an
Abstract. This paper reports on the spin glass-like coexistence of competing magnetic orders
in oxygen-deficient V2O5 nanoparticles with a broad size distribution. X-ray photoelectron
spectroscopy yields the surface chemical stoichiometry of nearly V2O4.65 due to significant
defect density. Temperature-dependent electrical conductivity and thermopower
measurements demonstrate a polaronic conduction mechanism with a hopping energy of about
dM
0.112 eV. The V2O5-δ sample exhibits strong field as well as temperature-dependent magnetic
29
behaviour when measured with a SQUID magnetometer, showing positive magnetic
30
31 susceptibility across the temperature range of 2-350 K. Field-cooled and zero-field-cooled data
32 indicate hysteresis, suggesting glassy behaviour. The formation of small polarons due to
33 oxygen vacancy defects, compensated by V4+ charge defects, results in Magneto-Electronic
34 Phase Separation (MEPS) and various magnetic exchanges, as predicted by first-principle
35 calculations. This is evidenced by the strong hybridisation of V orbitals in the vicinity of
36 vacant oxygen site. An increase in V4+ defects shows an antiferromagnetic (AFM) component.
37
The magnetic diversity in undoped V2O4.9 originates from defect density and their random
38
distribution, leading to MEPS. This involves localised spins in polarons and ferromagnetic
pte

39
40 (FM) clusters on a paramagnetic (PM) background, while V4+ dimers induce AFM interactions.
41 Electron paramagnetic resonance spectra measured at different temperatures indicate a
42 dominant paramagnetic signal at a g-value of 1.97 due to oxygen defects, with a broad FM
43 resonance-like hump. Both signals diminish with increasing temperature. Neutron diffraction
44 data rules out long-range magnetic ordering, reflecting the composition as V2O4.886. Despite
45
the FM hysteresis, no long-range order is observed in neutron diffraction data, consistent with
46
the polaron cluster-like FM with MEPS nature. This detailed study shall advance the
ce

47
48 understanding of the diverse magnetic behaviour observed in undoped non-magnetic systems.
49
50
51
52
Keywords: defect magnetism, layered oxide, V2O5, spin-glass, room temperature
Ac

53 ferromagnetism
54
55
56
57
58
59
60
 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2 Page 2 of 32

1
2
3 REFERENCES 2
4

pt
5 1. Introduction
6
7 The origin of ferromagnetism in otherwise non-magnetic (d0) materials has been an
8
9 intriguing question for about a few decades.[1–8] Various reports have shown the existence

cri
10 of unexpected ferromagnetism (FM) or even other less ordered types of magnetism such as
11 anti-ferromagnetism (AFM), paramagnetism (PM) and superparamagnetism (SPM) etc, in
12
13 simple as well as complex oxides, nitrides which are otherwise expected to be diamagnetic
14 (DM). Primafacie the origin of the unusual magnetism was ascribed to magnetic ion dopants
15
16
in host oxides, but later they were identified as vacancy defects in lattice which contribute

us
17 reasonably long-range order to give ferromagnetic moment even in undoped samples at room
18 temperature[2, 5, 9–12]. Subsequently, the observations of defect-induced magnetism were
19
20 reported in other non-oxide systems [13–17] as well as recently in graphene and its
21 analogues[18–21]. However, the net concentration and distribution of these defects
22
23
24
25
26
27
28
down the origin of this phenomenon. an
(vacancies) is contingent upon the processing condition, hence diverse nature of magnetic
properties was reported for the same system in the literature which made it difficult to pin

Efforts were made by various researchers to induce magnetic moments or enhance the
existing moment in non-magnetic oxides through doping by transition metal ions. Those have
dM
29 been explained through the Bound Magnetic Polaron (BMP) model where magnetic exchange
30
31 happens between two polaronic sites through transition metal ions[22–24]. However, this
32 could not be applied to undoped non-magnetic oxides since there were no magnetic impurities
33
to mediate. Nevertheless, there seems to be a larger consensus on the defect at one of the
34
35 constituent sites being responsible for this unusual magnetism in undoped systems. The
36 transition metal oxides may show various sub-stoichiometric magnetic phases and thus
37
38 several magnetic properties are depicted due to variable d-orbital occupancy. This results in
pte

39 a wide variety of magnetic signatures in oxides of the same element. Nevertheless, the most
40
stable transition metal oxides (TMO), having empty d orbitals (d0, S=0), are expected to show
41
42 bulk diamagnetism because of their zero net spins e.g. vanadium pentoxide (V2O5), titanium
43 dioxide (TiO2) etc. Among these, V2O5 due to its narrow band gap (∼2.2 eV) has attracted
44
45 significant interest for visible absorption as opposed to other optically transparent wide
46 bandgap oxides ZnO, SnO2, TiO2, which have been extensively studied for dilute magnetic
ce

47
semiconductors[6, 9, 12, 22, 25–27]. Moreover, V2O5 has a layered crystal structure and hence
48
49 can be exfoliated into atomically thin layers. This has far-reaching implications for 2D
50 materials-based functionalities. Quite a few researchers have reported the observation of
51
52 magnetic interactions in the vanadium oxide (V2O5) system. However, like other oxide
Ac

53 systems, the magnetic behaviours observed are quite diverse in this case too.[28– 35]
54 Moreover, these studies have reported a wide range of micro-structures of V2O5, such as
55
56 nanotubes, nano-lamellas, nanowires, nanoparticles etc.
57
58
59
60
Page 3 of 32 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2

1
2
3 REFERENCES 3
4

pt
5 Mesoscopic materials form an integral part of condensed matter systems and
6 polydispersity is inherent to real materials that cannot be avoided in most generic cases even
7
8 including the growth of single crystals where a size distribution is inevitable. Measurements
9 performed on such single crystals are representative of the entire batch of samples and caution

cri
10 should be exercised while reporting the phenomena with relevant statistical variations.
11
12 Nevertheless, such statistical variations are small when reporting macroscopic or bulk
13 properties where the magnitude is large enough. In the case of results about non-trivial origins
14
15
such as noise, inhomogeneities, crystalline imperfections etc, the statistical variations become
16 overwhelmingly large. Nevertheless, their macroscopic average consistently behaves in a

us
17 certain fashion whose magnitude may vary from sample to sample and depending on different
18
19 processing conditions[34] To have a complete understanding of what is the breadth of the
20 different interactions possible, it requires marking the limiting cases at extremes of completely
21
homogenous vis-à-vis an inhomogeneous systems. A system which contains large
22
23
24
25
26
27
28
an
inhomogeneities is likely to exhibit all possible phenomena which are otherwise reported in
diverse studies with a limited degree of inhomogeneity in different compartments. [36–38]
Investigation of the defect-induced magnetism in otherwise non-magnetic oxides has led
to new questions and new possible applications. [25, 26, 39] Besides, recently exfoliated two-
dM
dimensional layered solids showing different physical attributes are being explored for new
29
30 physics and devices [40, 41]. However, making magnetic measurements in individual 2D
31 layers is challenging. Therefore, bulk solids like V2O5 which have layered structures are
32
33
interesting systems to investigate for this dual significance. This being the case, we have
34 explored the magnetic properties of V2O5 nanoparticles, which are bulk diamagnetic.
35 The experimental results were analysed in the light of first principle calculations which
36
37 support the observed nature of diverse magnetic interactions. Although the BMP model is
38 developed for magnetic impurity doped semiconductors, we find that it also applies to systems
pte

39
which are self-doped with magnetic ions due to charge nonstoichiometry. Further, for the first
40
41 time, the direct estimation of electric polaron properties has been made in the same system in
42 addition to magnetic measurements. The results of this work demonstrate that there exists a
43
44 mix of various magnetic interactions in the nanoparticles of V2O5, which are field as well as
45 temperature-dependent. Thus, it may address the origin of different magnetic phases reported
46
in V2O5 for different processing conditions and dimensionalities.
ce

47
48 The manifestation of inhomogeneous magnetic properties due to short-range phase
49 inhomogeneities is referred to as magnetoelectronic phase separation[42, 43]. Usually, it is
50
51 observed that upon carrier injection certain systems show FM clusters in non-FM matrix
52 leading to complex magnetic behaviour as demonstrated by several systems. The examples
Ac

53 include cobaltites (La1−xSrxCoO3[44]), manganites (Nd0.5CaxSr0.5−xMnO3[45],

54
55 La0.7Ca0.23Sr0.07MnO3 [46]) etc. Notably, the transport properties of these systems have a
56 commonality of the polaron hopping mechanism. Similarly, in oxygen-deficient V2O5 one can
57
58
imagine electrons doped by oxygen vacancy dopants and their polaron formation. In such
59 cases, the magnetic as well as the transport properties are governed by the site occupancy-
60
 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2 Page 4 of 32

1
2
3 REFERENCES 4
4

pt
5 induced randomness of the dopant (defect in this case) vis-a-vis the background spin
6 distribution. Thus, we observe a possible spin or cluster glass system wherein competing
7
8 magnetic interactions are observed having their origin in electronic localisation-inducing
9 magnetoelectronic phase-separated clusters. [36, 47]

cri
10
11
12 2. Experimental details
13
14 2.1. Synthesis and characterisation of nanoparticles
15
16 The precursor, NH4VO3, in powder form was heated in an alumina crucible at 500oC for about

us
17
18 3 hours (heating rate 10oC/min.) under an ambient atmosphere, in a tube furnace. By 500 oC
19 the thermal decomposition reaction was complete and pure V2O5 was left behind in the
20
21
crucible. NH3 as well as H2O are evolved as gaseous byproducts. The
22 details of thermochemical reactions can be found elsewhere.[48, 49]
23
24
25
26
27
28
an
The X-ray diffraction patterns were recorded on a Bruker D8 Advance diffractometer
with Cu Kα source having a wavelength of 1.5418 Å. The patterns were recorded in the range
20o - 80o, with a step size of 0.02o. TEM images were obtained on an FEI transmission electron
microscope with an accelerating voltage of 300 keV. The magnetic measurements have been
dM
29 performed using a very sensitive Superconducting Quantum Interference Device (SQUID)
30 magnetometer (Quantum Design, model MPMS3) in the temperature range of 2 to 350 K. The
31
32 resistivity was measured in a home-built setup using four probe van der Pauw method on a
33 pellet sample of 8 mm diameter and 2 mm thickness.
34
35
36 2.2. Neutron Diffraction
37
38 To further comprehend structural characterizations and potential long-range magnetic
pte

39 ordering in the title compound, the sample was subjected to neutron diffraction (ND) at room
40
temperature and low temperature (2K). A powder diffractometer (PD-3) at the National
41
42 Facility for Neutron Beam Research (NFNBR), Dhruva reactor, BARC Mumbai(India) was
43 utilized. This Multi-PSD (position sensitive detector) based focusing crystal diffractometer
44
45 was installed by UGC-DAE CSR Mumbai Centre for ND study [50]. The powder sample was
46 filled in a cylindrical sample holder which was then placed at the beamline in a Close Cycle
ce

47 Refrigerator based cryostat (M/s. Cryogenics Industries, UK) for room temperature and low
48
49 temperature (2 K) ND experiment. The monochromatized neutron beam (wavelength 1.48 Å)
50 was utilized for the neutron diffraction experiment and observed neutron diffraction data was
51
52
refined by the Rietveld method. FullProf suite software was utilized for the data analysis [51].
Ac

53
54
55
56 2.3. Computational details
57
58 First-principle calculations using Density Functional Theory (DFT) were performed with the
59
60
Projector Augmented-Wave (PAW) method as implemented in the Vienna Ab-initio
Page 5 of 32 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2

1
2
3 REFERENCES 5
4

pt
5 Simulation Package (VASP) [52–54]. The Generalized Gradient Approximation (GGA) with
6 the Perdew-Burke-Ernzerhof (PBE) exchange-correlation function was used [55]. The
7
8 electronic and magnetic properties of V2O5 were investigated for three cases: pure V2O5, V2O5
9 with 10% oxygen vacancies, and oxygen-deficient V2O5 with 30% of V 5+ ions substituted by

cri
10
V 4+ ions. The DFT+U method was employed to account for the strong on-site Coulombic
11
12 interactions between the localized 3d electrons of the transition metal V, using Dudarev’s
13 approach [56], where only the effective U (U-J) is relevant. Fixed values of 3 eV and 0.9 eV
14
15 were used for U and J, respectively, for the V atoms. Structure optimization was carried out
16 with a gamma-centered k-point mesh for supercell calculations. A kinetic energy cut-off of

us
17 400 eV was used for the plane-wave basis set of 4 × 2 × 6. Lattice constants and atomic
18
19 coordinates were fully relaxed with a convergence criterion for a total energy of 10−3 eV. Self-
20 consistent charge density calculations were performed with the same k-mesh. Spin-polarized
21
22
calculations were conducted with VASP to study the magnetic properties.
23
24
25
26
27
28
an
dM
29
30
31
32
33
34
35
36
37
38
pte

39
40
41
42 Figure 1. (a)The layered crystal structure of V2O5 has weakly bonded sheets like ladders. (b)
43 The unit cell of V2O5 and its possible three equivalent sites of oxygen in a VO5 square
44 pyramids. Thus, the three types of oxygen vacancies could arise in the V2O5 lattice. (c) apical,
45 (d) bridging and (e)doubly degenerate positions (Colour code: Pink: vacancy site, blue:
46 directly bonded to vacancy site, black: weak hydrogen bonded to vacancy site. (f) X-ray
ce

47 diffraction pattern recorded for V2O5 sample, (g) its magnified view of low angle region
48
showing no impurities.
49
50
51
52
Ac

53
54
55
56
57
58
59
60
 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2 Page 6 of 32

1
2
3 REFERENCES 6
4

pt
5 3. Results
6
7 3.1. Structure, size and crystallinity of V2O5 nanoparticles
8
9 Vanadium pentoxide has a layered structure wherein parallel sheets of atoms exist along the

cri
10
11 b-axis as shown in Fig 1(a). It has an orthorhombic structure with space group Pmmn and
12 lattice parameters of a = 11.51 Å, b=3.56 Å, and c =4.33 Å as shown in Fig 1(b). As shown
13
here, each VO5 square pyramid is slightly distorted with apical oxygen marked as O1, while
14
15 the basal oxygen ions can be identified as having one bridging oxygen (O2) and two having
16 equivalent positions (O3). Ngamwongwan et.al.[57] studied the energetic formation of

us
17
18 vacancy defects at each of these sites and the corresponding effect on the band structure. It
19 was demonstrated that the formation of a neutral vacancy at any site acts as a double ionised
20 donor with these electrons localisation of vanadium ions. Also, it was shown that out of these
21
22 possible sites, vacancy at O1 is a major defect for V2O5 synthesised at atmospheric conditions.
23
24
25
26
27
28
an
Besides, the higher the temperature of annealing or formation, the higher the oxygen defect
concentration and subsequently the carrier concentration of the solid. Nevertheless, having an
oxygen vacancy at any of the sites causes further lattice distortion in VO5 square pyramids.
Besides, the weak van der Waals bonds between the two layers (shown by broken bonds in
dM
29 Fig 1(c-e)) also get affected in adjacent layers due to vacancies.
30 The XRD pattern collected for V2O5 is shown in Fig 1(f). The obtained data has been
31
32
analysed using the Rietveld refinement profile fitting method. The full range of the XRD
33 pattern shows all reflections of the orthorhombic structure (JCPDS no. 600767). The Rietveld
34 refinement parameters including lattice parameters, occupancy etc have been summarised in
35
36 Table S1 in the Supplementary Materials section. Fig 1(g) shows the magnified view of the
37 low angle region of the XRD pattern, which indicates the absence of impurity phases in the
38
sample, seen to the best of resolution. Further, the crystallite size calculated using Scherrer
pte

39
40 formula is estimated to be about 50 nm. The size and crystallinity of the particles have been
41 further verified with Transmission Electron Microscopy (TEM). The images obtained have
42
43
been shown in Fig 2(a). It may be observed from the TEM image that the typical size of the
44 particles seen is nearly 50 to 100 nm, as shown in Fig 2(b). As the XRD shows the lower
45 threshold of the crystallites, these observations are in agreement with each other. Moreover,
46
there is a greater degree of agglomeration within the particles as the thermal decomposition
ce

47
48 process is more akin to producing agglomerates as opposed to solution-based methods where
49
nuclei grow into well-separated particles[58]. The lattice fringes obtained in high-resolution
50
51 TEM have been analysed using Fast Fourier Transform (FFT) and the spacing observed
52 corresponds to (010) planes of V2O5, as marked in Fig. 2(c) and (d). Fringe separation has
Ac

53
54 been measured using FFT of the fringes as shown in Fig. 2(d). The spacing of 0.36 nm is
55 observed in V2O5. The selected area electron diffraction (SAED) pattern (shown in Fig 2(e))
56
was obtained and indexed corresponding to the V2O5 planes, such as (200), (110), (011),
57
58 (400), (600), (240) and (170). The elemental scan using energy dispersive spectroscopy (EDS)
59
60
Page 7 of 32 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2

1
2
3 REFERENCES 7
4

pt
5 is shown in Fig 2(f) and it shows peaks of only V and O which marks the purity of the sample.
6 Therefore, no other impurity phases or elements have been observed in the sample.
7
8
9

cri
10
11
12
13
14
15
16

us
17
18
19
20
21
22
23
24
25
26
27
28
an
dM
29
30
31
32
33 Figure 2. Transmission Electron Microscope (TEM) images of the particles at (a) low and (b)
34 high magnifications. (c) High-resolution TEM (HRTEM) image with (d) its Fast Fourier
35 Transform (FFT) for calculating the lattice spacing of 0.35 nm corresponding to the (101)
36 spacing of V2O5. (e) The SAED pattern collected on the sample (f) Energy Dispersive X-ray
37
Spectrum (EDS) showing only vanadium and oxygen (Si signal from substrate).
38
pte

39
40
41 The X-ray and UV photoelectron studies of V2O5 were performed and the data reported
42 in our earlier work[49] is presented for a ready reference. The O 1s peak shows asymmetry
43
44 with a higher binding energy shoulder ascribed to O ions in the vicinity of oxygen defects in
45 the crystal[9]. The most probable defects are the oxygen vacancies in the V2O5 lattice, which
46 act as donors and impart an n-type conductivity to wide bandgap oxides[13, 59]. Since XPS
ce

47
48 is a surface-sensitive technique, these oxygen vacancies are mainly on the surface due to
49 abrupt surface discontinuity in the periodicity of the lattice. Nevertheless, their relative
50
51
fraction concerning regular O 1s is significant. The XPS is a local as well as superficial probe
52 (with a sampling depth of <10 nm. The same is about 1 µm for EDS, however it is far less
Ac

53 accurate). XPS gives a (local) accurate surface quantification of the composition. It may be
54
55 only used as a reference. Nevertheless, the sample surface was cleaned by ion milling before
56 collecting the spectrum. These oxygen vacancies are warranted by the thermodynamic
57
stability of the lattice and their formation energy is much lower [57], which justifies their
58
59 dominant nature as compared to V5+ metal ion vacancies. The UV photoelectron spectroscopy
60
 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2 Page 8 of 32

1
2
3 REFERENCES 8
4

pt
5 (UPS) also shows the presence of a small mid-gap peak, about 1 eV below the VBM,
6 ascribable to the presence of reduced oxidation states of V i.e. V4+[13]. The V spectrum shows
7
8 strong non-stoichiometry of V ions in 4+ and 5+ states. The resolved peaks have been ascribed
9 to chemical shifts due to the occurrence of V4+ substitution at V5+ sites. The presence of V4+

cri
10
is warranted for charge neutrality, balancing the oxygen vacancies, i. e. lack of negatively
11
12 charged ions.
13 Table 1. Quantification results from the X-ray photoelectron spectroscopy data.
14
15
V2O5
16

us
17 Element V O
18 Estimated fraction 2 4.63
19
20 Estimated formula V2O4.63
21 V4+:V5+ 39.9: 60.1
22
23
24
25
26
27
28
O2-:VO

3.2. What do transport measurements reveal?

an 60.5: 39.5
dM
29 The electrical resistivity (ρ) vs temperature (T) data for the V2O5 sample is obtained using a
30 4-probe van der Pauw geometry, in the temperature range 160 to 600 K. The data reveals a
31
32 semiconducting nature of the sample as the resistivity drops with increasing temperature, as
33 shown in Fig 3(a). However, when the Arrhenius plot is obtained i.e. ln ρ vs 1/T, the
34
activation energy of the sample turns out to be very low 0.11 eV. Such a low value of thermal
35
36 activation energy for a large bandgap material (∼ 2 eV, See supplementary material Fig. S1))
37 cannot justify the thermally activated behaviour in band transport. In Fig 3(a and b) the sudden
38
break/discontinuity in the resistivity data at 300 K, as shown in the plot. This is just a
pte

39
40 measurement artefact which resulted from the change of the experimental set-up used while
41 going from the low-temperature data (T < 300 K) to the high-temperature data (T > 300 K).
42
43 Two different setups were used for the resistivity measurements on the same sample pellet
44 because there is no single setup for measuring resistivity across this entire temperature range.
45
46
A custom-made close-cycle cryostat based low temperature set-up was used below 300 K
ce

47 while another custom made-heater based high temperature set-up was used from 300 K to 600
48 K.
49
50 This apparent discrepancy and the exact nature of transport of V2O5 has been ascribed to
51 the low-lying impurity levels and is well documented in the literature [60, 61]. The charge
52
imbalance resulting from oxygen vacancies in V2O5 is compensated by the excess electrons
Ac

53
54 located on neighbouring vanadium atoms, reducing them from V5+(d0) to V4+ (d1). This is
55 directly substantiated by the XPS data which indicates the significant presence of V4+ in the
56
57 given sample, as reported in our earlier work [49]. Nevertheless, XPS being a surface-
58 sensitive technique, those are most likely the surface defects formed as a result of surface
59
oxygen vacancies. However, those are also likely to be present in the bulk of the nanoparticles,
60
Page 9 of 32 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2

1
2
3 REFERENCES 9
4

pt
5 although in reduced concentrations balancing the net defect density. The localised charge on
6 V4+ deforms the lattice in its vicinity and leads to the formation of a bound state of lattice
7
8 distortion and a trapped electron, namely a polaron. These polarons can carry both current as
9 well as heat as they move within the solid. The transport properties of the V2O5 nanoparticles

cri
10
are probed as a function of P doping and the same is reported in our earlier work [49].
11
12
13
14
15
16

us
17
18
19
20
21
22
23
24
25
26
27
28
an
dM
29
30
31
32
33
34
35
36
37
38
pte

39
40
41
42
43
44
45
46
ce

47
48
Figure 3. (a) Resistivity vs temperature plot of V 2O5 pellet and the fit using (b) polaronic
49
model. (c) Seebeck coefficient vs temperature plot of V 2O5 pellet (the dashed line is a guide
50
51 for the eye) and (d) the fit using the polaronic model. (e) ln σT1/2 vs 1/T1/4 plot from room
52 temperature to high temperature and (f) ln σT1/2 vs 1/T from room temperature to low
Ac

53 temperature. Measured values of (g) carrier mobility and (h) carrier concentrations using hall
54 measurements.
55
56
57 Table 2. The estimated activation energies and their constituents.
58
59
60
 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2 Page 10 of 32

1 Data slope value energy (eV)

2 Resistivity 3581.43 ∆E = 0.308
3 REFERENCES 10
4 Seebeck 0.098 ∆Eg = 0.196

pt
5 −
∆Ep = ∆E− ∆Eg 0.112
6
7
8
9 The temperature dependence of DC resistivity through the polaron hopping channel is

cri
10
given by,
11
Δ𝐸
12 𝜌 = 𝐴 𝑇 3/2 exp (𝑘 𝑇) (1)
13 𝐵

14 where ∆E is the activation energy of the transport. Fig 3(b) shows that ln(ρT−3/2) vs 1/T is in
15 very good agreement with the above equation, confirming the polaronic transport and yielding
16
∆E = 0.308 eV. Further, ∆E is constituted of ∆Ep and ∆Eg. Where ∆Eg is the energy required

us
17
18 to put the charge into the polaronic site, while ∆Ep is the energy required to extract the charge
19 from one polaronic site and propagate it to the next. Since all the polarons are assumed to
20
have the same energies, ∆Eg can be independently ascertained from the Seebeck coefficient
21
22 (S) vs temperature data by fitting it with the equation,
23
24
25
26
27
28
𝑆=
an
𝑘𝐵
𝑒
exp (
Δ𝐸𝑔
2𝑘𝐵 𝑇
) (2)
Thus, the variation of the Seebeck coefficient with temperature can be used to extract the
value of Eg. The Seebeck coefficient (S) vs temperature (T) measurement is shown in Fig.
3(c). The S vs 1/T data shown in Fig 3(d) fits very well with this expression, confirming the
dM
29
presence of thermal activation of charges into the polaronic sites and yielding ∆Eg = 0.196eV.
30
31 These values are summarised in Table 2. Thus, the polaronic hopping energy calculated in
32 this case is ∆Ep = 0.112 eV which is in good agreement with literature for vanadium
33
34 pentoxide[57, 60] and other complex oxides which also have ∆Ep = 0.1 eV.
35 The Heike’s formula for S is given by [62, 63]
36
37 𝑘𝐵 𝑐
𝑆(𝑇 → ∞) = 𝑒
ln (1−𝑐) (3)
38
pte

39
where c is the ratio of the number of reduced transition metal ions (V4+) to the total
40
41 number of standard oxidation state metal ions (V5+)in the transition metal oxide. The negative
42 sign of the Seebeck coefficient implies n-type carriers.
43
44 The Seebeck values obtained from experiments are unusually high considering V4+ ion
45 constituents estimated from XPS as well as Eq. 3. The estimation of V4+ ions gives c at about
46
0.01 from S vs T. (see Fig. S2 in supporting information). Here we do understand that XPS
ce

47
48 being a surface-sensitive technique, gives a surface estimation which could be different from
49 the bulk. In nanoparticles, most defects reside on the surface due to uncompensated bonds and
50
51 surface states. This implies that despite a large number of defects, only a small fraction
52 participates in conduction.
Ac

53 Now, the polaron hopping conduction, as described by Greaves[62], actually consists of

54
55 two contributors - the optical modes (σopt) and the acoustic modes (σac), so that
56 𝜎 = 𝜎𝑜𝑝𝑡 + 𝜎𝑎𝑐 (4)
57
58 At higher temperatures, only the optical modes contribute to the conductivity while it is
59 only the acoustic modes that are responsible for conductivity at lower temperatures. However,
60
Page 11 of 32 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2

1
2
3 REFERENCES 11
4

pt
5 in the intermediate temperature range, contributions from both modes exist. Fig.3(e) shows
6 the plot of log σT1/2 vs 1/T in the high-temperature range. In Fig. 3(f), log σT1/2 has been
7
8 plotted against 1/T1/n+1; where n represents dimensionality. Here, n=3 fitting is shown for
9 intermediate temperatures, however, in the low-temperature range, n=1 gives the best fit

cri
10
implying Effros–Shklovskii variable range hopping [49]. In fact, in the very low-temperature
11
12 data both the plots show good linear fit. A closer inspection of the high-temperature data (Fig.
13 3(e)) shows that the data deviates from linearity at around 490 K. Above this temperature,
14
15 acoustic modes are no longer significant. A similar deviation from linearity is observed in the
16 low-temperature data (Fig. 3(f)), at around 190 K. Below this temperature all the optical

us
17 modes are quenched. In the intermediate temperature range (190 K < T < 490 K), both modes
18
19 are present and contribute to the conductivity. All these observations comply with the model
20 proposed by Greaves in [62] and the key temperature values obtained are in good agreement
21
22
with these kinds of glasses.
23
24
25
26
27
28
an
Thus, in the system of V2O5 having significant defect density the transport occurs
through equivalent site hopping within kT of EF at low temperatures, assisted by acoustic
phonon (representing bound polarons). This is characterised by a large hopping energy. While
at higher temperatures the slope of EF becomes flat across several kT, leading to band-like
dM
29
conduction between one site to the other having equivalent energies due to significant lattice
30 distortion (representing free polarons). This also leads to saturation of thermopower due to no
31 net change in carrier potential energies during hoping at equivalent sites. The same may be
32
33 observed in the carrier mobility and carrier concentration data shown in Fig. 3(g) and (h)
34 respectively. The sample shows very low carrier mobility and lower carrier concentrations
35
below room temperature (300K). The carrier concentration quickly escalates through thermal
36
37 activation, while the mobility gets adversely affected due to a sudden rise in free carriers.
38
pte

39
40 3.3. Do undoped V2O5 nanoparticles exhibit magnetic order?
41
42 The magnetic measurements of the V2O5 sample are carried out on a sensitive SQUID
43 magnetometer. The background data of the magnetometer without a sample is shown in Fig
44
45
S3 in the supplementary materials section. Fig. 4 shows all the temperature-dependent
46 magnetic measurement data obtained for V2O5. The M-T measurements are done over the
ce

47 temperature range 2 - 350 K, in three different bias fields, 20, 100 and 1000 Oe. The Zero
48
49 Field Cooled (ZFC) curves were obtained by cooling the sample in the absence of a magnetic
50 field, up to 2 K. Subsequently, a small bias field was applied and measured in the warming
51
cycle. Field-cooled (FC) curves were obtained by taking measurements while cooling the
52
Ac

53 sample down to 2 K, under a constant bias field. The sample holder the background has been
54 subtracted from all the sample data.
55
56
57
58
59
60
 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2 Page 12 of 32

1
2
3 REFERENCES 12
4

pt
5
6
7
8
9

cri
10
11
12
13
14
15
16

us
17
18
19
20
21
22
23
24
25
26
27
28
an
dM
29
30
31
32 Figure 4. (Colour online) (a), (b) M-T (FC-ZFC) at bias fields 100 and 1000 Oe respectively.
33 (c) The DC magnetic susceptibility (χ) vs temperature at 100 and 1000 Oe of undoped V 2O5
34
nanoparticles. The regions having paramagnetic dominance can be corroborated by comparing
35
36 them with the T−1 line shown for ready reference. The peak in ZFC near 200 K is marked by
37 an arrow. (d) The Curie-Weiss plots for the two field values. The negative intercepts denote
38 antiferromagnetic ordering in the undoped V2O5 nanoparticles.
pte

39
40
41 The M-T data resembles cases where there is a coexistence of paramagnetic (PM), and
42
43 ferromagnetic (FM) components within the same material. The FC data at 100 and 1000 Oe
44 shown in Fig. 4 (a) and (b) respectively, have a similar nature, consisting of a rising trend
45 when cooled from room temperature (300 K), a plateau in the intermediate temperature range
46
and again a sharp rise below 10 K. The FC data shows a rise in the moment in a Curie-Weiss
ce

47
48 (CW) fashion followed by a broad maximum as well as an upturn. These observations are
49
similar to those observed due to the presence of short-range interactions in a glassy system[64,
50
51 65]. Towards the lowest end on the temperature scale, a PM-like exponential tail is seen which
52 signifies the PM component of the system. At low temperatures, the sudden fall in moment
Ac

53
54 with temperature is reminiscent of perfect PM behaviour which may be attributed to
55 superparamagnetism (SPM) given the size of the V2O5 nanoparticles (<100 nm). Similar
56
variations of PM with FM and glass-like FC-ZFC are found in a number of non-magnetic
57
58 oxide nanoparticles [9, 22, 46, 66–68] The irreversibility of the ZFC and FC curves reveals a
59 spin-glass nature of the system. The gap begins to close at higher fields. That might be due to
60
Page 13 of 32 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2

1
2
3 REFERENCES 13
4

pt
5 the dominant PM contribution at higher fields which rises linearly with the field as opposed
6 to FM that saturates. Such behaviour has been previously reported for manganites by Pillai et
7
8 al. [45] which also associates such behaviour with the inhomogeneous mixing of FM-AFM
9 phases. The ZFC-FC data measured at 20 Oe bias field is shown in Fig. 5 wherein the curve

cri
10 is strikingly different in comparison to those of high field data. Nevertheless, when observed
11
12 in 20 Oe data it may be noted that as the field decreases (at low temperatures), M-T gradually
13 changes slope with the field. At 1000 Oe it is exponentially decreasing with a steep negative
14
15
slope, while the slope becomes less negative for 100 Oe and it becomes positive for 20 Oe.
16 This may be interpreted as the suppression of the PM contribution as the field decreases below

us
17 100 Oe. The presence of the AFM component is further accentuated in the slowly falling
18
19 nature of the ZFC curve, where two AFM components are seen, one at 270 K and the other at
20 80 K[45]. See Fig. S4 in the supplementary material section for a comparison of all the M-T
21
data.
22
23
24
25
26
27
28
an
dM
29
30
31
32
33
34
35
36
37
38
pte

39
40 Figure 5. M-T (FC-ZFC) at bias fields 20 Oe bias field for undoped V2O5 nanoparticles.
41
42
43
The DC magnetic susceptibilities are evaluated at given magnetic fields, and the same
44
45 are plotted in Fig. 4(c). The gradual change in trend with temperature, as mentioned above, is
46 observed in the case of magnetic susceptibilities (χ), shown in Fig 4(c). The temperature
ce

47
48 dependence systematically reduces the exponent as H increases from 100 to 1000 Oe as shown
49 with green dashed lines. The departure between FC-ZFC is maximum for 100 Oe field, which
50 is intermediate of low (20 Oe shown in Fig. 5) and high (1000 Oe) fields, demarcated in this
51
52 study, where the susceptibility of the FM centres is gradually surpassed by that of the PM
Ac

53 centres and going through a maximum glassy nature in between.

54
55
𝐶
56 𝜒 = 𝑇−𝜃 (5)
57 𝐶𝑊

58
59
A Curie-Weiss (CW) fit (Eq. 5) is attempted on the high-temperature region of
60
 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2 Page 14 of 32

1
2
3 REFERENCES 14
4

pt
5 the 1/χ vs T plot, as can be seen in Fig. 4(d), to ascertain the nature of the various magnetic
6 phases predicted in the system. However, the CW extrapolation lines in the high-temperature
7
8 region, yield negative intercepts on the temperature axis indicating a possible
9 antiferromagnetic (AFM) ordering. Further, the value of CW temperature (θCW ) decreases as

cri
10
the applied field increases. This could be extrapolated as θCW → 0 as H increases (purely PM
11
12 at high fields). Nevertheless, the apparent Neel temperature is nearly the same irrespective of
13 field i.e. 270 K. Similar AFM ordering has been observed by Dhoundiyal et.al. [35] and
14
15 Dreifus et.al. [69]. However, the Neel temperatures reported are quite different (50 K [35] and
16 80 K [69] respectively). These AFM interactions may result from V4+ ions, which has been

us
17
reported in the literature[70, 71]. However, as discussed earlier, the concentration and relative
18
19 proximity of V4+ ions depend on that of the oxygen vacancies which are predominantly
20 governed by the processing conditions as well as microstructures, which are different in these
21
22
cases. A strong deviation of CW fit for low field values indicates the presence of short-range
23
24
25
26
27
28
AFM matrix and resulting complexities.
an
clusters of FM in AFM or vice versa in glassy systems [64, 65]. Thus, the deviation from CW
fit observed for low field (20 and 100 Oe) points towards such short-range FM clusters in the
dM
29
The effective moment is deduced from the Curie constant using the Eq 6,
30
31 3𝑘 𝐶
32
𝜇𝑒𝑓𝑓 = √ 𝑁𝜇𝐵 (6)
𝐵
33
34 where N is Avagadro’s number, µB is Bohr magneton and C is the curie constant deduced
35 from the CW plot whose value has been found to change with the field. However, as can be
36 seen in Fig 4(b), the 1000 Oe field data shows clear signature of linear behaviour at high
37
38 temperature, considering that Curie constant value of 0.31 K.emu/mole(of V).Oe which gives
an effective moment of 1.56 µB which is close to 1.72µB, a value obtained by a single unpaired
pte

39
40
41
electron (S=1/2). The comparison of Langevin and Brillouin function fitting to the 2 K M-H
42 data is shown in Fig. S6 and S7 in the supplementary material.
43 The observations from the M-T data are corroborated by the M-H isotherms, obtained at
44
45 three different temperatures, 2, 100, and 300 K, as shown in Fig 6(a), (c) and (e) respectively,
46 and their corresponding magnified images in Fig 6(b), (d) and (f) respectively. The high-
ce

47
temperature isotherms are more or less like FM, with a prominent hysteresis loop and
48
49 saturation, overlaid with a small positive slope, which might come from both PM and AFM
50 components in the system. However, at 2 K there could be likely an SPM component that
51
52
completely overshadows the FM loop, as expected from the M-T data. Nevertheless, FM has
Ac

53 been found to persist from 2 K to 300 K, as depicted from the centred hysteresis loops in Fig
54 6. Thus, a mixed contribution from various FM-AFM-PM states is observed each of them
55
56 evolves independently with temperature. The respective contributions have been attributed to
57 various origins discussed in subsequent sections.
58
59
60
Page 15 of 32 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2

1
2
3 REFERENCES 15
4

pt
5
6
7
8
9

cri
10
11
12
13
14
15
16

us
17
18
19
20
21
22
23
24
25
26
27
28
an
dM
29
30
31
32
33
34
35
36
37
38
pte

39
40
41
42
43
44 Figure 6. Colour online) (a), (b), (c) M-H isotherms V2O5 at 300, 100 and 2 K respectively.
45
(d), (e), and (f) are the magnified M-H isotherms to show the presence of hysteresis at all
46
temperatures.
ce

47
48
49
50 Here FM dominates at low fields (depicted by hysteresis loops) and high temperatures,
51 while PM dominates at low temperatures (< 10 K) and high fields (as seen by linear rise of
52
Ac

53 M-H). At the intersection i.e. low temperature, low fields the system behaves like
54 superparamagnetic single-domain particles. However, a system at a low-temperature PM
55 phase turning into an FM at a high temperature is unlikely hence all the components must
56
57 have a coexistence where the measurements manifest their behaviour depending on the
58 relative strength of magnetic susceptibilities. There could be a wider PM background with
59
60
 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2 Page 16 of 32

1
2
3 REFERENCES 16
4

pt
5 short-range clusters of FM and AFM which reveal depending on temperature and field and
6 also convert to PM when heated beyond 300 K.
7
8 There is a hysteresis but no saturation is observed till the highest field is applied (7 Tesla).
9 The presence of hysteresis in an M-H isothermal curve without saturation may come from

cri
10
multiple factors like two or more competing phases. The M-T data in the low-temperature
11
12 range also shows a steep T-0.16 and T-0.3 dependence for 100 Oe and 1000 Oe data respectively
13 (See Fig. 4(b)) while M-H depicts a hysteresis at all temperatures. Here, because of the wider
14
15 size distribution of the nanoparticles some of them which are near the lower bound of the
16 distribution may exhibit SPM. (The experimental signatures of likely SPM are compared with

us
17 the reports in the literature [72] and presented in the supplementary information section E,
18
19 Fig. S5) Such smaller particles may have much lower curie temperature than the larger size
20 counterparts. The dependency of Curie temperature on size is given by eq [73],
21 3𝐿
22 𝑇𝑐 (𝐷) = 𝑇𝑐 (∞) (1 − 2𝐷) (7)
23
24
25
26
27
28
an
where D is the size of the nanoparticles, L is the thickness of the monolayer, and TC(D)
and TC(inf) stand for the Curie temperature of nanoscale and bulk systems respectively.
Therefore, it may be possible that the smaller particles may be exhibiting that phase transition
to PM much before the larger counterparts.
dM
29
30
31
32
33
34
35
36
37
38
pte

39
40
41
42
43
44
45
46
ce

47
48
49
50
51
52 Figure 7. EPR spectra of V2O5 nanoparticles were obtained at four different temperatures viz.
Ac

53 130, 150, 200 and 298 K. The inset shows the signal height vs temperature of the 1.97 g-value
54 signal.
55
56
57
58 To confirm the nature of magnetic interactions, Electron paramagnetic resonance (EPR)
59 spectroscopy was performed till 130 K the lowest possible temperature attainable in the given
60
Page 17 of 32 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2

1
2
3 REFERENCES 17
4

pt
5 system. We have collected EPR spectra at four different temperatures, 130, 150, 200 and 300
6 K as shown in Fig 7. The spectrum consists of a sharp single line (S=1/2) at a g-value of 1.97
7
8 with details of hyper-fine features (corresponding to spin 7/2) barely visible appearing
9 superimposed on the single sharp line. As reported for EPR spectra of single crystals of V2O5

cri
10
samples, the 1.97 signal is attributed to oxygen vacancy-induced localised spins. [74, 75]. The
11
12 signal gets sharper when cooled as shown in the inset figure of signal height vs temperature,
13 depicting a stronger paramagnetic localisation at low temperatures. This is in good agreement
14
15
with that of the magnetic measurement studies, that show the dominant paramagnetic slopes
16 at low temperatures in M-H data. In our earlier work [49], we have also shown that the feature

us
17 at a g value of about 2.4 the hump observed may be attributed to the ferromagnetic exchange
18
19 present in the system. The broad FMR signal appears to be superimposed with the sharp PM
20 signal. The earlier work demonstrated changes in the dominant contribution of PM with
21
increasing P doping in EPR as well as magnetic measurements due to localisation effects as a
22
23
24
25
26
27
28
investigated.
an
result of P doping. The signal shows a systematic change in temperature that may be further
dM
29
30
31
32
33
34
35
36
37
38
pte

39
40
41
42
43
44
45
46
ce

47
48
49
50
51 Figure 8. Powder neutron diffraction data for V2O5 nanoparticles were obtained at (a) 290 K
52
Ac

and (b) 2 K respectively with no additional reflections seen at 2 K corresponding to ordering

53
54 other than structural.
55
56
57
58
59
60
 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2 Page 18 of 32

1
2
3 REFERENCES 18
4

pt
5 To further confirm the nature of magnetic order, neutron diffraction was performed on
6 V2O5 powder samples at 290 and 2 K to compare. The same is shown in Fig 8 above at 290
7
8 K and 2 K. It was fitted with space group Pmn21 (s. g. No. 31). Table S2 in supplementary
9 information shows the refined data from 290 and 2 K neutron diffraction, the atomic position,

cri
10
goodness of fitting parameters and cell parameter are also depicted in the table. Since
11
12 vanadium is an incoherent scatterer, the signal-to-noise ratio is a bit noisy, which is reflected
13 in a slightly higher χ2 value. As neutron scattering length of the element is one of the factor in
14
15 structure factor calculation, which in turn help in identifying the concentration of light
16 elements in the compound. By utilizing this property of the neutron, we observed that by

us
17 introducing the deficiency in oxygen, the goodness of the fitting parameter improved
18
19 significantly and we could find the chemical formula comes out to be around V2O4.886.
20
21
22
23
24
25
26
27
28
an
dM
29
30
31
32
33
34
35
36
37
38
pte

39
40
41
42
43
44 Figure 9. (a)-(c) Total density of states calculated with PBE functional for the cases of pure
45 (a) V2O5, (b) V2O4.9 i.e. with O-vacancy. The comparison of DOS of oxygen-deficient V2O4.9
46 with (c) 10% and (d) 30 % charged defects having the FM and AFM configurations of the
ce

47
spins. Charge density difference plot between magnetic and non-magnetic state for ( e & f)
48
49 V2O5 with O-vacancy, and (g & h) V2O4.9 crystal with oxygen defects as well as V4+ ions for
50 two planes containing the vacancy site. (Colour bar: shows positive charge density difference;
51 higher charge density for magnetic state by red colour, while negative charge density
52 difference higher charge density for nonmagnetic state is indicated by the rest of the colours.
Ac

53 The comparison of (i) measured and (j) calculated DOS for valence band region with that of
54 oxygen-deficient V2O5.
55
56
57 By comparing raw data as well as refined data, it is very evident that the appearance of
58
59
a new peak in the pattern down to 2 K, could not be ascertained. We also did not observe an
60
Page 19 of 32 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2

1
2
3 REFERENCES 19
4

pt
5 enhancement in any preexisting peak, the signature of magnetism (antiferromagnetic or
6 ferromagnetic) could not be seen. The splitting at the 39.8o peak is more pronounced due to
7
8 shrinking in the cell which has led to shifting of uneven 2θ value thus improving the resolution
9 in the peak.

cri
10
11
12
13 3.4. Predictions from first principle calculations
14
15 To estimate the origin of different magnetic contributions, the first principle calculations
16

us
17 are performed using a supercell of vanadium pentoxide lattice stoichiometric V2O5. Further,
18 an oxygen vacancy is created in this supercell. The stoichiometric V2O5 is known to be a
19 diamagnetic semiconductor [76, 77], which is also verified in our study from electronic and
20
21 magnetic calculations. Spin-polarised calculations result in zero value of magnetic moment
22
23
24
25
26
27
28
an
on all the atoms indicating the diamagnetic state of V2O5. The total density of states (DOS)
plot as shown in Fig. 9(a) exhibits semiconducting properties with a band gap of ∼ 1.9 eV
between the valence and conduction band calculated using PBE functional (shown in blue
band). The HSE06 functions were also performed and those resulted in overestimation of the
band gap to 3.6 eV. (see Fig. S8 in Supplementary Materials section). Nevertheless, the PBE
dM
29 results are in good agreement with those reported earlier.[78] Here, PBE stands for an
30
exchange-correlation functional, while HSE06 stands for a hybrid functional classes of DFT
31
32 (using GGA).
33 Table 3. (Colour online) Total magnetic moment (µB/vacancy) and energy difference ∆E
34 (eV/vacancy) between magnetic and non-magnetic states for different cases of V2O5
35 considered.
36
37 Total magnetic ∆E
38
pte

39 moment per vacancy

40
41
(µB/vacancy) (eV/vacancy)
42 Pure V2O5 0 -
43
44 V2O5 with O-vacancy 1.775 -0.788 (FM)
45 O-deficient V2O5 with -0.382 (AFM) -
46 0.818 0.383(FM)
10% charged defect
ce

47
48 O-deficient V2O5 with -3.277(AFM) -
49 3.959
30% charged defect 3.273(FM)
50
51
52
Ac

53 Electronic and magnetic properties calculations were done of V2O5 by introducing 10%
54 oxygen vacancies in the perfect crystal structure. Spin-polarised calculations indicate a
55
56 magnetic moment induced on a vanadium atom bonded to the site of oxygen vacancy as well
57 as a small magnetic moment induced on oxygen atoms nearby to the vacancy site giving a net
58 magnetic moment of 1.775 µB per unit cell per vacancy. Total energy calculations to determine
59
60
 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2 Page 20 of 32

1
2
3 REFERENCES 20
4

pt
5 ground-state magnetic ordering stabilise the ferromagnetic solution. The energy difference
6 ∆E (in eV) between magnetic and non-magnetic states as well as total magnetic moment (in
7
8 µB ) values per vacancy are listed in Table 3. The electronic spin-polarized DOS plot in the
9 ferromagnetic state shown in Fig. 9(b) with PBE functionals, clearly indicates that the band

cri
10
gap value for V2O5 with O-vacancy estimated at 1.9 eV is closer to the experimentally
11
12 determined value of 2.2 eV. In the case of HSE calculations, the value obtained is also large
13 (greater than 3 eV).
14
15 The electronic and magnetic properties for the case of O-deficient V2O5 in which 10% and
16 30% fraction of V 5+ ions have been replaced with V 4+ ions (as charge compensatory defect)

us
17
were investigated. The spin-polarised calculations stipulate that the induced magnetic moment
18
19 on the V atom bonded to the vacancy site due to oxygen vacancies is reduced when V 5+ ions
20 are replaced with V 4+ ions. Also, moments on V atoms align antiferromagnetically as well as
21
22 small magnetic moment induced on O atoms as depicted by the total magnetic moment value
23
24
25
26
27
28
an
0.818 and 3.959 µB for 10% and 30% fraction of V 5+ ions respectively as shown in Table 3.
Further, as shown in Fig. 9(c) and (d), the estimated DOS for the case of O-deficient V2O5
with 10% and 30% charged defect reveals metallic behaviour with fermi level trapped inside
the valence band. HSE calculations yield similar behaviour to the case of PBE calculations.
dM
29 There is a very small change in FM and AFM DOS for both charged defect cases due to the
30 robust local magnetic configuration of the V atom bonded to the site of O vacancy with the
31 magnetic configuration of the system [79]. However, It should be noted that the DOS of
32
33 charged defect may be less accurate due to non-localised extra electron in the system. We also
34 performed spin-polarised calculations for the 10% charged defect in O-deficient V2O5, in
35
36
which case, the ferromagnetic ordering was found to have the lowest energy. The
37 corresponding energy values are listed in Table 3. Adding charged defects only helps in
38 explaining AFM contribution i.e. AFM increases as V4+ goes from 10% to 30 %. The moment
pte

39
40 value obtained for only the oxygen-deficient case is 1.77 µB which is close to a single electron
41 (S=1/2) as estimated from the Curie constant value.
42
43
The charge density difference plots obtained from theoretical calculations of magnetic
44 and non-magnetic states for V2O5 with O-vacancies are shown in Fig. 9(e) and (f). Similar
45 plots for the V2O4.9 crystal with O-vacancies as well as V 4+ ions, for two planes containing
46
the vacancy sites are shown in Fig. 9(g) and (h). The calculated DOS using PBS for crystal
ce

47
48 having oxygen vacancies has been compared with that of UV photoelectron spectrum (UPS)
49
which gives local DOS below the Fermi level. It may be observed from Fig. 9(i and j), that
50
51 there is a very good correlation observed in that of calculated and experimental spectra which
52 denotes the experimental validity of results. It is clear that despite the presence of V4+ surface
Ac

53
54 defects, the electronic band structure does not concur with its simulated band structure for
55 measured V4+ defect densities. Hence the contribution of such V 4+ defect is minimal in
56 deciding the electronic energies. Besides, the amount of charged defect contributing (10 or 30
57
58 %) cannot be large here as by finite temperatures the thermal activation helps bound polarons
59 to become free, resulting in more localised electrons[80]. It has also been seen from the
60
Page 21 of 32 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2

1
2
3 REFERENCES 21
4

pt
5 transport measurements that temperature dependence of conductivity becomes band-like
6 (1/T). Whereas, the thermopower measurements and Heike's formula tell us that the effective
7
8 V 4+ /V 5+ ratio should be one in a thousand or lower. (See supplementary material Fig. S2).
9 Thus, only a small fraction of charged defects leads to polaronic states. Others may contribute

cri
10
to the paramagnetic background. Therefore, none of those DOS for charged defects matched
11
12 the observed UPS. The electronic band structure is governed by metal d and O 2p characters
13 of the bands and the mid-gap states arise as a result of oxygen deficiencies. These mid-gap
14
15 states are characteristics of polarons in V2O5 (like other polaronic solids) [80].
16

us
17
18
19
20
21
22
23
24
25
26
27
28
an
dM
29
30
31
32
33
34
35
36
37
38
pte

39
40
41
42
43
44 Figure 10. The schematic diagram shows the various possible magnetic interactions in an
45 undoped V2O5 lattice having oxygen vacancies (charged or neutral) and compensating V 4+
46 defects, together having a bound state called polaron. These can be isolated or overlapping
ce

47 polarons depending on inter-polaron distance and hence may result in paramagnetic or

48 ferromagnetic interaction. The pair of V 4+ ions (dimers) could also couple
49 antiferromagnetically within the same or neighbouring strands of the ladder. Please note that
50
the figure is merely for the better explanation of different kinds of defects that may occur in
51
V2O5 lattice and their interactions which may lead to polarons as well as magnetic exchanges
52
Ac

53 of various kinds. Due to limitations of space and better clarity, all these have been drawn in a
54 very small space of V2O5 lattice. This should not be confused with the actual defect densities
55 which exist in the experiment, nor those used in simulations for modelling the system.
56
57
58
59
60
 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2 Page 22 of 32

1
2
3 REFERENCES 22
4

pt
5 4. Discussion
6
7 Although bulk V2O5 is diamagnetic, experimental magnetic measurements conducted on the
8
9 oxygen-deficient nanoparticles showed a large variation in the observations ranging from

cri
10 ferromagnetism[28, 30, 33], antiferromagnetism[69] and to a mixture of both[29, 31]. While
11
first principles calculations performed by Xiao et.al.[32] predicted that the V2O5 system may
12
13 have vacancy-induced ferromagnetism if the x-value lies within 0.19 to 0.45 or below 0.13.
14 However, through 0 < x < 0.5, it shall also exhibit an antiferromagnetic coupling (AFM-1
15
16 shown in Fig 10) because of antiparallel coupling of V4+ spins on adjacent ladders, (often

us
17 referred to as dimers[29]). Besides, at the x > 0.25, another AFM coupling (AFM-2 shown in
18
Fig 10) may arise due to anti-parallel coupling of V4+ spins on adjacent rungs of the ladder.
19
20 For low vacancy concentrations (and low V4+ defects) the electronic structure remains similar
21 to that of bulk V2O5−x, while for large oxygen vacancy concentrations, the electronic structure
22
23
24
25
26
27
28
an
may exhibit a significant change as also confirmed by our calculations. Thus, we envisage
that the electronic structure of our sample is likely to be similar to that of oxygen-deficient
V2O5−x with a trace amount of V4+ defects present. This can be readily confirmed upon
comparison of the UPS spectrum of our sample with the calculated DOS in Fig. 9(i-j), which
shows very good agreement. Besides, the x value for our sample (∼0.35), as calculated from
dM
29
30 the XPS data, legitimises the presence of a combination of both FM and AFM phases in it
31 (both AFM-1 as well as AFM-2). This is seen as the AFM to PM transition at about 280 K in
32
the Curie Weiss plot shown in Fig. 4(d). Here V4+ introduces the AFM interaction and
33
34 increases with increasing its amount as seen from the calculations.
35 Bhandari et.al. [81] investigated the effect of electron doping on the electronic and
36
37 magnetic properties of V2O5. Here they employed several strategies to dope the system with
38 electrons, such as varying the effective atomic numbers and intercalation of alkali metal
pte

39 within the layers of V2O5. Similar electron doping occurs when the system is doped with
40
41 oxygen vacancies such as in this case. Wherein the generalised Heisenberg Hamiltonian used
42 is of the form,
43
44
𝐻 = ∑𝑖≠𝑗 𝐽𝑖𝑗 𝒆𝒊 . 𝒆𝒋 (7)
45 Counting the spin on each non-equivalent vanadium atom of the two parallel rungs twice.
46 The interactions J1, J2, J3 were identified among the nearest neighbour vanadium ions and the
ce

47
48 net spin was found to be 0.5 µB on each vanadium ion due to the excess of electrons distributed
49 over all the vanadium atoms effectively rather than electron localisation on the defect. The
50
energy difference calculated between FM and AFM shows that below a particular electron
51
52 concentration of 0.88 electrons per vanadium atom, strong FM is shown by the sample, while
Ac

53 on exceeding this it tends to show AFM dominance. The maximum exchange coupling of J1
54
55 ∼ 44.8 meV is observed between two parallel rugs of the ladder with an electron delocalised
56 over each of them. i.e. same electron is shared by two vanadium atoms of the same rug of the
57 latter (making unit cell) (See Fig 10) and hence having the same spin. Whereas the j-values
58
59 of the other two nonequivalent sites predicted are J2 = -4.5 meV and J3 = -2.1 meV. Thus,
60
Page 23 of 32 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2

1
2
3 REFERENCES 23
4

pt
5 quoting from Bhandari et. al. [81] The nature of both interactions is antiferromagnetic. They
6 reduce significantly as distance increases. This is an indication of neighbouring chains
7
8 coupling antiferromagnetically and the within chain ordering is also antiferromagnetic. This
9 has been marked in Fig 10 as AFM1 and AFM2.

cri
10
It is perhaps wiser to examine the polaronic signature observed in the electrical transport
11
12 studies and the concurrently observed magnetic behaviour, due to their common origin from
13 an oxygen vacancy paired with a V4+ defect. An oxygen vacancy can be neutral (VO) singly
14
15 charged (𝑉𝑜+ ) or doubly charged (𝑉𝑜++ ) making shallow and deep donors. Most of the existing
16 literature has overlooked or only marginally discussed this under the umbrella of bound

us
17 magnetic polarons (BMP), without any reference to their electrical contributions. The problem
18
19 of polaronic conduction and its energetics has been a question of interest for a long time.
20 Sanchez et.al.[61] in their paper published in 1982 showed a remarkable correlation between
21
22
DC electrical conductivity and Electron Spin Resonance (ESR) experiments, to establish the
23
24
25
26
27
28
an
mechanism of polaronic conduction in the V2O5 system. The study revealed that there exists
a coupling between the unpaired electron of V4+ and an associated oxygen defect. The electron
may be localized on equivalent V5+ ions in the vicinity of the defect (as shown by the
blue/brown shaded circle in Fig 10). Nevertheless, this polaron is a small polaron which could
dM
29
be localised near the defect (i.e. bound polaron) or it could be localised (free polaron) when
30 thermally activated with a sufficient activation energy of about 0.2 eV, leading to increased
31 conductivity. Thus, the electrical conductivity quickly escalates as the polaron hopping is
32
33 enabled with sufficient thermal activation. The activation energies mentioned by Sanchez
34 et.al.[61] are in very good agreement with those calculated in this study.
35
We estimated stoichiometry from neutron diffraction via refinement. The modelled
36
37 composition of V2O4.886 is also very close to V2O4.9 used in DFT modelling whose band
38 structure matched with that of DOS estimated using UPS data. This system can be considered
pte

39
40 as equivalent to a d0 semiconductor, self-doped with trace transition metal (V4+) impurities
41 similar to those described in reference [23]. As mentioned earlier, at low temperatures, the
42 polarons are bound and the unpaired electrons are localised near the defects. These being
43
44 small polarons, they are reasonably near, yet isolated from each other. The calculated radius
45 of a small polaron (1.84 Å) is a good indicator of this. Isolated polarons are distributed
46
throughout space and can contribute to the paramagnetic background, which decreases
ce

47
48 sharply with temperature. The bounded electrons, confined in polaronic states in their vicinity,
49 may show a direct ferromagnetic exchange, referred to as polaron-polaron interaction.
50
51 However, their separation distances may play a key role in deciding the dominant interactions
52 (See Fig 10). The background semiconductor matrix may meanwhile have a carrier-carrier
Ac

53 antiferromagnetic interaction from ions.

54
55 As predicted by Durst et.al.[23], the susceptibility variation shown in Fig. S8, in the
56 temperature of interest, shows a dual magnetisation mechanism i.e. low field involving
57
58
polaron-polaron interaction which could induce direct ferromagnetic exchange. Whereas high
59
60
 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2 Page 24 of 32

1
2
3 REFERENCES 24
4

pt
5 field isolated polarons and among the magnetic ions, if any. While the linear slope at a high
6 field (> 1000 Oe) results from paramagnetic spin alignments.
7
8 Owing to the nano-sized nature of the particles additional effects may be expected, such
9 as superparamagnetism from the core of particles (domains), as observed here. Besides, the

cri
10
intergranular dipoles may contribute to the paramagnetic background. Further, due to the
11
12 orientational anisotropy of the particles, the relative orientation of the easy axis with the field
13 shall have a precision contributing to spin-flip distribution (abrupt along the easy axis and
14
15 gradual perpendicular to the easy axis).
16 Subsequently, at high temperatures, as the polarons become free by thermal activation,

us
17 long-range correlations are enabled. These polaron-polaron correlations observed at low
18
19 temperatures and low fields are disturbed, resulting in gradually vanishing magnetism at high
20 temperatures. Nevertheless, at high temperatures, a weak ferromagnetic hysteresis at 300 K,
21
22
suggests the existence of short-range FM clusters, which could be primarily due to the dilute
23
24
25
26
27
28
magnetism.

an
In our other works[49], it has been shown that doping phosphorus at the vanadium site
leads to a reduction in the hysteresis loop and introduces a positive slope in M-H at high fields,
implying dominating PM behaviour. Besides, in the same work, it is demonstrated that the
dM
PM nature is manifested in higher M-H slopes and sharper EPR signals with increasing P
29
30 doping.
31 Thus, we have a tentative picture of the competing processes that are present in the
32
33 system and how they contribute to the magnetic data. Fig. 10 is a pictorial representation of
34 that. There are V4+ ions besides V5+ ions in the material. The O-vacancies capture electrons in
35 them and they hybridise with the d-orbital of the nearby V4+ ions to form polarons. Polarons
36
37 can be thought of as large bubbles containing spins, spanning several formula units (FU).
38 Hence isolated polarons will have a paramagnetic moment as seen in EPR spectra. When they
pte

39
40
come close enough, they can overlap and give rise to FM or AFM in clusters since no long-
41 range order is observed. As in the sample, all these polarons and V4+ ions are distributed at
42 random, and depending on the spacing between them, RKKY interaction decides if the
43
44 coupling will be FM or AFM. Hence there is an inherent randomness and competing
45 interactions within the system and it is conducive to spin-glass behaviour, which might be the
46
reason for the split in the FC-ZFC curves. For confirmation, AC susceptibility measurements
ce

47
48 are required.
49
50
51 5. Conclusion
52
Ac

53 The study of the magnetic properties of otherwise bulk diamagnetic vanadium pentoxide
54 reveals the significance of oxygen vacancies and vanadium ion charge stoichiometry in
55
56 imparting magnetic moment to the sample. The electrical transport studies reveal a polaronic
57 hopping mechanism due to the presence of oxygen vacancies and charged vanadium defects.
58
Electrical resistivity and Seebeck coefficient show the polaron hopping mechanism with a
59
60
Page 25 of 32 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2

1
2
3 REFERENCES 25
4

pt
5 hopping energy of 0.112 eV. The radius of the polarons is estimated to be 1.8 Å. Polaron
6 binding energy of 0.287 eV confirms the possibility of long-range hopping at sufficiently high
7
8 temperatures causing a transition from bound to free polarons, thereby improving the
9 electrical conductivity at higher temperatures. Electronic and magnetic calculations for three

cri
10
cases of pure V2O5, V2O5 with 10% O-vacancy, and O-deficient V2O5 with 10 and 30% V5+
11
12 ions replaced with V4+ ions have been studied theoretically. Whereas pure V2O5 is a
13 diamagnetic semiconductor, V2O5 with 10% O-vacancy is a ferromagnetic semiconductor. On
14
15 introducing charged defects, V2O5 with 10% O-vacancy is found to be an antiferromagnetic
16 metal. Thus, it is proposed that vanadium-charged defects are in dilute limit which induced

us
17
antiferromagnetism within dimers but no changes in the band structure. Experimental
18
19 observations are in good agreement with the predictions of first-principles calculations. The
20 findings are concurred with techniques like EPR and neutron diffraction at different
21
22
temperatures. It shows a ferromagnetic ordering for low temperatures which is attributed to
23
24
25
26
27
28
an
polaron-polaron interactions at low fields and isolated spin alignments at high fields.
Nevertheless, there also exists antiferromagnetic contributions to the magnetic signals, due to
vanadium dimers across the ladders in the crystal structure of V2O5. For that further analysis
and time-dependent measurements are required, like AC susceptibility.
dM
29
30 Acknowledgements
31
32 The authors are grateful to the institute SQUID magnetometer facility of IISER
33
34
Thiruvananthapuram for the magnetic measurements Dr. Deepshikha Jaiswal-Nagar, Dr
35 Tuhin S. Maity, Mr Waseem for their help and Support during the magnetic measurements
36 and Hall effect Measurement systems (HEMS) for resistivity, mobility and carrier
37
38 concentration measurement at low temperatures. The funding received from the Science and
pte

39 Engineering Board (SERB), Government of India for the research grant (EEQ/2018/000769,
40
EEQ/2022/001016) supporting this work, is gratefully acknowledged. The authors express
41
42 their gratitude to Mr. Akshay S. Kamble from UGC CSR Mumbai for his help in the
43 refinement of neutron diffraction data.
44
45
46 Data availability statement
ce

47
48 The data that support the findings of this study are available upon request from the authors.
49
50
51
52 References
Ac

53
54
55 [1] JMD Coey. d0 ferromagnetism. Solid State Sciences, 7(6):660–667, 2005.
56 [2] JMD Coey, P Stamenov, RD Gunning, M Venkatesan, and K Paul. Ferromagnetism in
57
58 defect-ridden oxides and related materials. New Journal of Physics, 12(5):053025, 2010.
59
60
 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2 Page 26 of 32

1
2
3 REFERENCES 26
4

pt
5 [3] Jacek K Furdyna. Diluted magnetic semiconductors. Journal of Applied Physics, 64(4):
6 R29–R64, 1988.
7
8 [4] Nguyen Hoa Hong, Joe Sakai, Nathalie Poirot, and Virginie Brizé. Room-temperature
9 ferromagnetism observed in undoped semiconducting and insulating oxide thin films.

cri
10
11
Physical Review B, 73(13):132404, 2006.
12 [5] Nguyen Hoa Hong, Joe Sakai, Ngo Thu Huong, Nathalie Poirot, and Antoine Ruyter.
13
14
Role of defects in tuning ferromagnetism in diluted magnetic oxide thin films. Physical
15 Review B, 72(4):045336, 2005.
16

us
[6] Nguyen Hoa Hong, Joe Sakai, and Francois Gervais. Magnetism due to oxygen
17
18 vacancies and/or defects in undoped semiconducting and insulating oxide thin films.
19 Journal of magnetism and magnetic materials, 316(2):214–217, 2007.
20
21 [7] Malcolm P. Kennett, Mona Berciu, and R. N. Bhatt. Two-component approach for
22
23
24
25
26
27
28
Feb 2002.
an
thermodynamic properties in diluted magnetic semiconductors. Phys. Rev. B, 65:115308,

[8] PA Wolff, RN Bhatt, and AC Durst. Polaron-polaron interactions in diluted magnetic

semiconductors. Journal of Applied Physics, 79(8):5196–5198, 1996.
dM
29 [9] Vinayak B Kamble, SV Bhat, and AM Umarji. Investigating thermal stability of
30 structural defects and its effect on d0 ferromagnetism in undoped sno2. Journal of
31
Applied Physics, 113(24):244307, 2013.
32
33 [10] David W Abraham, Martin M Frank, and Supratik Guha. Absence of magnetism in
34 hafnium oxide films. Applied Physics Letters, 87(25):252502, 2005.
35
36 [11] V Fernandes, RJO Mossanek, P Schio, JJ Klein, AJA De Oliveira, WA Ortiz, N Mattoso,
37
J Varalda, WH Schreiner, M Abbate, et al. Dilute-defect magnetism: Origin of
38
magnetism in nanocrystalline CeO2. Physical Review B, 80(3):035202, 2009.
pte

39
40 [12] A. Sundaresan, R. Bhargavi, N. Rangarajan, U. Siddesh, and C. N. R. Rao.
41
42 Ferromagnetism as a universal feature of nanoparticles of the otherwise nonmagnetic
43 oxides. Phys. Rev. B, 74:161306, Oct 2006.
44
45 [13] Qi Wang, Mathew Brier, Siddharth Joshi, Ajinkya Puntambekar, and Vidhya
46 Chakrapani. Defect-induced burstein-moss shift in reduced V2O5 nanostructures. Phys
ce

47
48 Rev B, 94(24):245305, 2016.
49 [14] You Lai, Liqin Ke, Jiaqiang Yan, Ross D. McDonald, and Robert J. McQueeney. Defect-
50
51 driven ferrimagnetism and hidden magnetization in MnBi2Te4. Phys. Rev. B,
52 103:184429, May 2021.
Ac

53
54
[15] Matheus P. Lima, L. Cabral, Emanuela Margapoti, Suddhasatta Mahapatra, Juarez L. F.
55 Da Silva, Fabian Hartmann, Sven Höfling, Gilmar E. Marques, and Victor Lopez-
56 Richard. Defect-induced magnetism in ii-vi quantum dots. Phys. Rev. B, 99:014424, Jan
57
58 2019.
59
60
Page 27 of 32 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2

1
2
3 REFERENCES 27
4

pt
5 [16] Mathias Bersweiler, Evelyn Pratami Sinaga, Inma Peral, Nozomu Adachi, Philipp
6 Bender, Nina-Juliane Steinke, Elliot Paul Gilbert, Yoshikazu Todaka, Andreas Michels,
7
8 and Yojiro Oba. Revealing defect-induced spin disorder in nanocrystalline Ni. Phys. Rev.
9 Materials, 5:044409, Apr 2021.

cri
10
11 [17] Suhas Nahas, Biplab Sanyal, and Annica M. Black-Schaffer. Spontaneous
12 ferromagnetism and finite surface energy gap in the topological insulator Bi2Se3 from
13
surface BiSe antisite defects. Phys. Rev. B, 102:140407, Oct 2020.
14
15 [18] Oleg V. Yazyev and Lothar Helm. Defect-induced magnetism in graphene. Phys. Rev.
16

us
B, 75:125408, Mar 2007.
17
18 [19] Elton JG Santos, Andrés Ayuela, and Daniel Sánchez-Portal. Universal magnetic
19
properties of sp3-type defects in covalently functionalized graphene. New Journal of
20
21 Physics, 14(4):043022, 2012.
22
23
24
25
26
27
28
an
[20] Ahmet Avsar, Cheol-Yeon Cheon, Michele Pizzochero, Mukesh Tripathi, Alberto
Ciarrocchi, Oleg V Yazyev, and Andras Kis. Probing magnetism in atomically thin
semiconducting PtSe2. Nature communications, 11(1):1–7, 2020.
[21] Antal A Koos, Peter Vancso, Marton Szendro, Gergely Dobrik, David Antognini Silva,
Zakhar I Popov, Pavel B Sorokin, Luc Henrard, Chanyong Hwang, Laszlo P Biro, et al.
dM
29
Influence of native defects on the electronic and magnetic properties of cvd grown MoSe2
30
31 single layers. The Journal of Physical Chemistry C, 123(40):24855–24864, 2019.
32
[22] Kusuma URS, SV Bhat, and Vinayak Kamble. On exceeding the solubility limit of Cr+3
33
34 dopants in SnO2 nanoparticles based dilute magnetic semiconductors. Journal of Applied
35 Physics, 123(16):161518, 2018.
36
37 [23] Adam C. Durst, R. N. Bhatt, and P. A. Wolff. Bound magnetic polaron interactions in
38 insulating doped diluted magnetic semiconductors. Phys. Rev. B, 65:235205, Jun 2002.
pte

39
40 [24] Jian-Yun Zheng, Shan-Hu Bao, Yan-Hong Lv, and Ping Jin. Activation and
41 enhancement of room-temperature ferromagnetism in Cu-doped anatase TiO2 films by
42
43
bound magnetic polaron and oxygen defects. ACS applied materials & interfaces,
44 6(24):22243–22249, 2014.
45
46
[25] Pablo David Esquinazi, Wolfram Hergert, Markus Stiller, Lukas Botsch, Hendrik
ce

47 Ohldag, Daniel Spemann, Martin Hoffmann, Waheed A Adeagbo, Angelika Chassé,

48 Sanjeev K Nayak, et al. Defect-induced magnetism in nonmagnetic oxides: Basic
49
50 principles, experimental evidence, and possible devices with zno and TiO2. physica
51 status solidi (b), 257(7):1900623, 2020.
52
Ac

53 [26] Shuai Ning, Peng Zhan, Qian Xie, Weipeng Wang, and Zhengjun Zhang. Defectsdriven
54 ferromagnetism in undoped dilute magnetic oxides: a review. Journal of Materials
55 Science & Technology, 31(10):969–978, 2015.
56
57
58
59
60
 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2 Page 28 of 32

1
2
3 REFERENCES 28
4

pt
5 [27] Pratima Borah and Manos PC Kalita. Synthesis and elucidation of structural, optical,
6 vibrational and magnetic properties of mn-doped V2O5 dilute magnetic semiconductor
7
8 nanocrystals. Physica B: Condensed Matter, 670:415346, 2023.
9 [28] Manas R Parida, C Vijayan, Chandra Sekhar Rout, CS Suchand Sandeep, Reji Philip,

cri
10
11 and PC Deshmukh. Room temperature ferromagnetism and optical limiting in V2O5
12 nanoflowers synthesized by a novel method. The Journal of Physical Chemistry C,
13
115(1):112–117, 2011.
14
15 [29] E. Vavilova, I. Hellmann, V. Kataev, C. Täschner, B. Büchner, and R. Klingeler.
16

us
Magnetic properties of vanadium oxide nanotubes probed by static magnetization and
17
51V nmr. Phys. Rev. B, 73:144417, Apr 2006.
18
19
20
[30] L Krusin-Elbaum, DM Newns, H Zeng, V Derycke, JZ Sun, and R Sandstrom. Room-
21 temperature ferromagnetic nanotubes controlled by electron or hole doping.
22
23
24
25
26
27
28
Nature, 431(7009):672–676, 2004.

an
[31] SV Demishev, AL Chernobrovkin, VV Glushkov, AV Grigorieva, EA Goodilin, H Ohta,
S Okubo, M Fujisawa, T Sakurai, NE Sluchanko, et al. Magnetic properties of vanadium
oxide nanotubes and nanolayers. Physical Review B, 84(9):094426, 2011.
[32] ZR Xiao and GY Guo. Structural, electronic and magnetic properties of V2O5−x: An ab
dM
29
30 initio study. The Journal of chemical physics, 130(21):214704, 2009.
31
32 [33] AB Cezar, IL Graff, J Varalda, WH Schreiner, and DH Mosca. Oxygen-vacancyinduced
33 room-temperature magnetization in lamellar V2O5 thin films. Journal of Applied Physics,
34
35 116(16):163904, 2014.
36 [34] Sergio Correal, Daniel Hernández-Gómez, Andrea Steffania Esquivel, Alexander
37
38 Cardona-Rodríguez, Andreas Reiber, Yenny Hernandez, Rafael GonzálezHernández,
pte

39 and Juan Gabriel Ramírez. Tuning electronic and magnetic properties through disorder
40 in V2O5 nanoparticles. Scientific Reports, 13(1):6752, 2023.
41
42 [35] Hemlata Dhoundiyal, Pintu Das, and Mukesh C. Bhatnagar. Electrical and magnetic
43
properties of V2O5 microstructure formed by self-assembled nanorods. Physica B:
44
45 Condensed Matter, 603:412622, 2021.
46 [36] John A Mydosh. Spin glasses: an experimental introduction. CRC Press, 1993.
ce

47
48 [37] Per Nordblad. Competing interaction in magnets: the root of ordered disorder or only
49
frustration? Physica Scripta, 88(5):058301, 2013.
50
51 [38] JC Gómez Sal, N Marcano, JI Espeso, J Hernández Velasco, and B Farago. Neutron
52
studies in inhomogeneous magnetic systems. Physica B: Condensed Matter, 385:372–
Ac

53
54 374, 2006.
55 [39] Tao Xu, Takahiro Shimada, Masataka Mori, Gen Fujimoto, Jie Wang, and
56
57 Takayuki Kitamura. Defect engineering for nontrivial multiferroic orders in SrTio3.
58 Phys. Rev. Materials, 4:124405, Dec 2020.
59
60
Page 29 of 32 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2

1
2
3 REFERENCES 29
4

pt
5 [40] Andrey M Tokmachev, Dmitry V Averyanov, Oleg E Parfenov, Alexander N
6 Taldenkov, Igor A Karateev, Ivan S Sokolov, Oleg A Kondratev, and Vyacheslav G
7
8 Storchak. Emerging two-dimensional ferromagnetism in silicene materials. Nature
9 Communications, 9(1):1672, 2018.

cri
10
11 [41] Anakha Anson, Dipanjana Mondal, Varsha Biswas, Kusuma Urs MB, and Vinayak
12 Kamble. Suppressed polaronic conductivity induced sensor response enhancement in mo
13 doped V2O5 nanowires. Journal of Applied Physics, 133(19), 2023.
14
15 [42] M Yu Kagan, KI Kugel, and AL Rakhmanov. Electronic phase separation: Recent
16

us
progress in the old problem. Physics Reports, 916:1–105, 2021.
17
18 [43] Tian Miao, Lina Deng, Wenting Yang, Jinyang Ni, Changlin Zheng, Joanne Etheridge,
19
Shasha Wang, Hao Liu, Hanxuan Lin, Yang Yu, et al. Direct experimental evidence of
20
21 physical origin of electronic phase separation in manganites. Proceedings of the National
22
23
24
25
26
27
28
an
Academy of Sciences, 117(13):7090–7094, 2020.
[44] A Podlesnyak, Georg Ehlers, M Frontzek, AS Sefat, Albert Furrer, Th Strässle, Ekaterina
Pomjakushina, Kazimierz Conder, F Demmel, and DI Khomskii. Effect of carrier doping
on the formation and collapse of magnetic polarons in lightly hole-doped la1−xsrxcoo3.
Physical Review B, 83(13):134430, 2011.
dM
29
30 [45] S Savitha Pillai, G Rangarajan, NP Raju, AJ Epstein, and PN Santhosh.
31
Coexistence of ferromagnetic and antiferromagnetic phases in nd0.5caxsr0.5−xmno3
32
33 manganites. Journal of Physics: Condensed Matter, 19(49):496221, 2007.
34
35 [46] Luis Andrés Burrola-Gándara, L Vázquez-Zubiate, DM Carrillo-Flores, and JT Elizalde-
36
Galindo. Spin glass behaviour in la0.7ca0.23sr0.07mno3 nanofibers obtained by
37
38 electrospinning. Journal of Superconductivity and Novel Magnetism, 32(8):2501–2508,
pte

39 2019.
40
41 [47] PS Anil Kumar, PA Joy, and SK Date. Origin of the cluster-glass-like magnetic
42 properties of the ferromagnetic system. Journal of Physics: Condensed Matter,
43
44 10(29):L487, 1998.
45 [48] KM Shafeeq, VP Athira, CH Kishor, and PM Aneesh. Structural and optical properties
46
of V2O5 nanostructures grown by thermal decomposition technique. Applied Physics A,
ce

47
48 126(8):1–6, 2020.
49
50
[49] Tathagata Sarkar, Saptak Majumder, Soumya Biswas, Sona S Rose, and Vinayak
51 Kamble. Engineering lattice oxygen defects and polaronic transport in vanadium
52 pentoxide via isovalent phosphorus doping. Applied Physics Letters, 123(22), 2023.
Ac

53
54 [50] V Siruguri, PD Babu, M Gupta, AV Pimpale, and PS Goyal. A high-resolution powder
55 diffractometer using focusing optics. Pramana, 71:1197–1202, 2008.
56
57 [51] Juan Rodríguez-Carvajal. Recent advances in magnetic structure determination by
58 neutron powder diffraction. Physica B: Condensed Matter, 192(1-2):55–69, 1993.
59
60
 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2 Page 30 of 32

1
2
3 REFERENCES 30
4

pt
5 [52] P. E. Blöchl. Projector augmented-wave method. Phys. Rev. B, 50:17953–17979, Dec
6 1994.
7
8 [53] G. Kresse and D. Joubert. From ultrasoft pseudopotentials to the projector augmented-
9 wave method. Phys. Rev. B, 59:1758–1775, Jan 1999.

cri
10
11 [54] G. Kresse and J. Furthmüller. Efficient iterative schemes for ab initio total-energy
12 calculations using a plane-wave basis set. Phys. Rev. B, 54:11169–11186, Oct 1996.
13
14 [55] John P. Perdew, Kieron Burke, and Matthias Ernzerhof. Generalized gradient
15 approximation made simple. Phys. Rev. Lett., 77:3865–3868, Oct 1996.
16

us
17 [56] S. L. Dudarev, G. A. Botton, S. Y. Savrasov, C. J. Humphreys, and A. P.
18 Sutton. Electron-energy-loss spectra and the structural stability of nickel oxide: An
19
20 LSDA+U study. Phys. Rev. B, 57:1505–1509, Jan 1998.
21 [57] Lappawat Ngamwongwan, Ittipon Fongkaew, Sirichok Jungthawan, Pussana Hirunsit,
22
23
24
25
26
27
28
an
Sukit Limpijumnong, and Suwit Suthirakun. Electronic and thermodynamic properties
of native point defects in v2o5: A first-principles study.
Physical Chemistry Chemical Physics, 23(19):11374–11387, 2021.
[58] Yasser K Abdel-Monem, Sanaa M Emam, and Hager MY Okda. Solid state thermal
dM
29 decomposition synthesis of CuO nanoparticles from coordinated pyrazolopyridine as
30 novel precursors. Journal of Materials Science: Materials in Electronics, 28(3):2923–
31
32 2934, 2017.
33 [59] J Buckeridge, Charles Richard A Catlow, MR Farrow, Andrew J Logsdail, DO Scanlon,
34
35 TW Keal, P Sherwood, SM Woodley, AA Sokol, and A Walsh. Deep vs shallow nature
36 of oxygen vacancies and consequent n-type carrier concentrations in transparent
37 conducting oxides. Physical Review Materials, 2(5):054604, 2018.
38
[60] B. Sipos, M. Duchamp, A. Magrez, L. Forró, N. Barišić, A. Kis, J. W. Seo, F. Bieri, F.
pte

39
40 Krumeich, R. Nesper, and G. R. Patzke. Mechanical and electronic properties of
41
42 vanadium oxide nanotubes. Journal of Applied Physics, 105(7):074317, 2009.
43 [61] C Sanchez, M Henry, JC Grenet, and J Livage. Free and bound polarons in vanadium
44
45 pentoxide. Journal of Physics C: Solid State Physics, 15(35):7133, 1982.
46
[62] GN Greaves. Small polaron conduction in V2O5-P2O5 glasses. Journal of NonCrystalline
ce

47
48 Solids, 11(5):427–446, 1973.
49
[63] P Kounavis, A Vomvas, E Mytilineou, M Roilos, and L Murawski. Thermopower,
50
51 conductivity and the hall effect in V2O5 gels. Journal of Physics C: Solid State Physics,
52 21(5):967, 1988.
Ac

53
54 [64] K. V. Rao, M. Fähnle, E. Figueroa, O. Beckman, and L. Hedman. Deviations from curie-
55 weiss behavior in spin-glasses. Phys. Rev. B, 27:3104–3107, Mar 1983.
56
57 [65] Jianchao Lin, Peng Tong, Dapeng Cui, Cheng Yang, Jie Yang, Shuai Lin, Bosen Wang,
58 Wei Tong, Lei Zhang, Youming Zou, et al. Unusual ferromagnetic critical behaviour
59
60
Page 31 of 32 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2

1
2
3 REFERENCES 31
4

pt
5 owing to short-range antiferromagnetic correlations in antiperovskite cu1−xnmn3+x (0.1
6 ≤ x ≤ 0.4). Scientific reports, 5(1):1–7, 2015.
7
8
9
[66] G Narsinga Rao, YD Yao, and JW Chen. Superparamagnetic behaviour of

cri
10 antiferromagnetic cuo nanoparticles. IEEE Transactions on magnetics, 41(10):3409–
11 3411, 2005.
12
13 [67] Fengzhen Huang, Zhijun Wang, Xiaomei Lu, Junting Zhang, Kangli Min, Weiwei Lin,
14 Ruixia Ti, TingTing Xu, Ju He, Chen Yue, et al. Peculiar magnetism of bifeo3
15
16 nanoparticles with size approaching the period of the spiral spin structure. Scientific

us
17 reports, 3(1):1–7, 2013.
18
19 [68] Eric Vincent and Vincent Dupuis. Spin glasses: experimental signatures and salient
20 outcomes. In Frustrated Materials and Ferroic Glasses, pages 31–56. Springer, 2018.
21
22 [69] D Dreifus, MPF Godoy, AC Rabelo, AD Rodrigues, YG Gobato, PC Camargo, EC
23
24
25
26
27
28
48(44):445002, 2015.
an
Pereira, and AJA de Oliveira. Antiferromagnetism induced by oxygen vacancies in V2O5
polycrystals synthesized by the pechini method. Journal of Physics D: Applied Physics,

[70] Prashanta K. Mukharjee, K. M. Ranjith, B. Koo, J. Sichelschmidt, M. Baenitz, Y.

dM
29 Skourski, Y. Inagaki, Y. Furukawa, A. A. Tsirlin, and R. Nath. Bose-einstein
30 condensation of triplons close to the quantum critical point in the quasi-onedimensional
31
32 spin- antiferromagnet navopo4. Phys. Rev. B, 100:144433, Oct 2019.
33 [71] M. E. Saleta, H. E. Troiani, S. Ribeiro Guevara, G. Ruano, R. D. Sánchez, M. Malta, and
34
35 R. M. Torresi. Strong reduction of V4+ amount in vanadium oxide/hexadecylamine
36 nanotubes by doping with Co2+ and Ni2+ ions: Electron paramagnetic resonance and
37
magnetic studies. Journal of Applied Physics, 109(9):093914, 2011.
38
pte

39 [72] Amit K Guria, Koushik Dey, Suresh Sarkar, Biplab K Patra, Saurav Giri, and Narayan
40 Pradhan. Tuning the growth pattern in 2d confinement regime of sm2o3 and the
41
42 emerging room temperature unusual superparamagnetism. Scientific Reports, 4(1):6514,
43 2014.
44
45 [73] Ling-fei Cao, XIE Dan, Ming-xing Guo, HS Park, and T Fujita. Size and shape effects
46 on curie temperature of ferromagnetic nanoparticles. Transactions of Nonferrous Metals
ce

47
Society of China, 17(6):1451–1455, 2007.
48
49 [74] D Ballutaud, C R’kha, C Sanchez, and J Livage. Esr of p2o5-doped V2O5 single crystals.
50
physica status solidi (a), 66(1):271–276, 1981.
51
52 [75] E Gillis and EEPR Boesman. Epr-studies of V2O5 single crystals. i. defect centres in pure,
Ac

53
54
non-stoichiometric vanadium pentoxide. physica status solidi (b), 14(2):337– 347, 1966.
55 [76] V. Eyert and K.-H. Höck. Electronic structure of V2O5: Role of octahedral deformations.
56
57
Phys. Rev. B, 57:12727–12737, May 1998.
58
59
60
 AUTHOR SUBMITTED MANUSCRIPT - JPCM-123375.R2 Page 32 of 32

1
2
3 REFERENCES 32
4

pt
5 [77] A. Chakrabarti, K. Hermann, R. Druzinic, M. Witko, F. Wagner, and M. Petersen.
6 Geometric and electronic structure of vanadium pentoxide: A density functional bulk
7
8 and surface study. Phys. Rev. B, 59:10583–10590, Apr 1999.
9 [78] Top Khac Le, Manil Kang, and Sok Won Kim. A review on the optical characterization

cri
10
11 of V2O5 micro-nanostructures. Ceramics International, 45(13):15781–15798, 2019.
12 [79] LM Sandratskii. Exchange interactions in (znmn) se: Lda and lda+ u calculations.
13
14 Physical Review B, 68(22):224432, 2003.
15 [80] Cesare Franchini, Michele Reticcioli, Martin Setvin, and Ulrike Diebold. Polarons in
16

us
17
materials. Nature Reviews Materials, 6(7):560–586, 2021.
18 [81] Churna Bhandari and Walter R. L. Lambrecht. Electronic and magnetic properties of
19
20
electron-doped v2o5 and nav2o5. Phys. Rev. B, 92:125133, Sep 2015.
21 [82] VN Bogomolov and DN Mirlin. Optical absorption by polarons in rutile (TiO2) single
22
23
24
25
26
27
28
an
crystals. physica status solidi (b), 27(1):443–453, 1968.
dM
29
30
31
32
33
34
35
36
37
38
pte

39
40
41
42
43
44
45
46
ce

47
48
49
50
51
52
Ac

53
54
55
56
57
58
59
60