Thermal hysteresis measurement of the VO2 dielectric function for its metal-insulator

transition by visible-IR ellipsometry

J. A. Ramirez-Rincon, C. L. Gomez-Heredia, A. Corvisier, J. Ordonez-Miranda, T. Girardeau, F. Paumier, C.
Champeaux, F. Dumas-Bouchiat, Y. Ezzahri, K. Joulain, O. Ares, and J. J. Alvarado-Gil

Citation: Journal of Applied Physics 124, 195102 (2018); doi: 10.1063/1.5049747

View online: https://doi.org/10.1063/1.5049747
View Table of Contents: http://aip.scitation.org/toc/jap/124/19
Published by the American Institute of Physics

Articles you may be interested in

Rare-earth high entropy alloys with hexagonal close-packed structure
Journal of Applied Physics 124, 195101 (2018); 10.1063/1.5051514

Identification of the shallow donor state in Sb doped ZnO by photoluminescence excitation spectroscopy
Journal of Applied Physics 124, 195701 (2018); 10.1063/1.5054337

Connecting post-pulsing electrical and microstructural features in GeTe-based inline phase change switches
Journal of Applied Physics 124, 195103 (2018); 10.1063/1.5031840

Direct evidence for the coexistence of nanoscale high-conduction and low-conduction phases in VO2 films
Applied Physics Letters 113, 173104 (2018); 10.1063/1.5032270

Temperature dependence of secondary electron emission: A new route to nanoscale temperature measurement
using scanning electron microscopy
Journal of Applied Physics 124, 195104 (2018); 10.1063/1.5050250

Efficient spin injection into graphene through trilayer hBN tunnel barriers
Journal of Applied Physics 124, 194301 (2018); 10.1063/1.5050874
 JOURNAL OF APPLIED PHYSICS 124, 195102 (2018)

Thermal hysteresis measurement of the VO2 dielectric function for its

metal-insulator transition by visible-IR ellipsometry
J. A. Ramirez-Rincon,1,2 C. L. Gomez-Heredia,1,2 A. Corvisier,2 J. Ordonez-Miranda,2
T. Girardeau,2 F. Paumier,2 C. Champeaux,3 F. Dumas-Bouchiat,3 Y. Ezzahri,2 K. Joulain,2
O. Ares,1 and J. J. Alvarado-Gil1,2
1
Departamento de Física Aplicada, Cinvestav-Unidad Mérida, Carretera Antigua a Progreso Km. 6,
97310 Mérida, Yucatán, México
2
Institut Pprime, CNRS, Université de Poitiers, ISAE-ENSMA, F-86962 Futuroscope Chasseneuil, France
3
CNRS, IRCER, UMR 7315, University of Limoges, F-87000 Limoges, France

(Received 25 July 2018; accepted 28 October 2018; published online 15 November 2018)
The real and imaginary parts of the dielectric function of VO2 thin films, deposited on r-plane
sapphire via pulsed laser deposition, are measured by means of visible-infrared ellipsometry for
wavelengths ranging from 0.4 to 15 μm and temperatures within its phase transition. For both the
insulator-to-metal (heating) and metal-to-insulator (cooling) transitions, it is shown that the two
ellipsometric signals exhibit three temperature-driven behaviors, which are well described by appro-
priate combinations of the Tauc-Lorentz, Gaussian, and Drude oscillator models. By fitting
Bruggeman’s effective medium model for the dielectric function to the corresponding measured
experimental values, using the volumetric fraction of the VO2 metallic domains as a fitting parameter
for different temperatures within the VO2 phase transition, we have found that this model is suitable for
describing the dielectric function in visible and near-infrared wavelengths (∼0.4 to ∼3.0 μm), but it
generally fails for longer infrared ones. Furthermore, the hysteresis loop of the VO2 emissivity averaged
over a relevant interval of wavelengths is determined and shown to vary from ∼0.49, in the insulator
phase, to ∼0.16, in the metallic one. These values, based on the VO2 dielectric function, are consistent
with previous measurements reported in the literature, and therefore, our measured data are
expected to be useful for describing the behavior of VO2 films involved in optical and radiative
applications. Published by AIP Publishing. https://doi.org/10.1063/1.5049747

I. INTRODUCTION
Vanadium dioxide (VO2) is a thermochromic material
widely studied over the last decades,1 due to its reversible
metal-insulator transition (MIT) at a relatively low tempera-
ture of TC ¼ 68  C.2 Around this transition critical tempera-
ture, VO2 presents a first order phase change from a
monoclinic (M1) to a tetragonal (R) structure.3,4 This struc-
tural transition modifies the insulator behavior of VO2 at low
temperatures (T  TC ) by a metallic one at high tempera-
tures (T  TC ),5 which results in sizable variations of its
optical,6,7 electrical,8,9 and thermal10,11 properties.
Several recent theoretical and experimental studies have
reported the suitability of VO2 thin films for technological
applications related to thermal energy management, such
as smart windows,12 infrared (IR) absorbers for sensing
process,13 IR reflectors for space devices,14 as well as the
modulation, amplification, and rectification of radiative heat
currents.15–17 All of these applications required the accurate
characterization of the VO2 optical performance in a given
spectral range. For instance, in high-temperature solar
energy applications, the reflectance and transmittance in the
visible (Vis) and near-infrared (NIR) spectra12,18 (from 0.3
to 2.5 μm) are of primary interest, while the emissivity for
wavelengths longer than 2.5 μm is the most suitable for
medium- and low-temperature energy applications.19 These
optical properties have been successfully measured for a
VO2 film + substrate system through convectional reflectance
spectroscopy in the Vis-IR spectrum.11,12 This methodology
requires a delicate analysis to single out the substrate contri-
bution, whose impact is expected to be strong in the insulat-
ing phase of VO2, because of its high transmittance in the IR
region.20 Vis-IR ellipsometry thus emerges as a reliable tech-
nique for the direct optical characterization of VO2 thin films
at a given temperature, by using a robust modeling that takes
into account the layered structures of samples, to properly
interpret the ellipsometric signals Psi (Ψ) and Delta (Δ).18
Nazari et al.21 reported ellipsometric measurements for
VO2 thin films grown on c-, m-, and r-plane sapphire, below
the NIR spectral range (0.2–2 μm), and showed that (1) the
main variations of the real and imaginary parts of the
dielectric function of VO2, between its insulating and
metallic states, occur for wavelengths longer than 0.5 μm
and 0.8 μm, respectively. (2) The transition temperatures of
the VO2 films deposited on m- and r-plane sapphire are
around 10 °C lower than the one determined for the film
grown on c-plane sapphire. This indicates that the substrate,
as well as the deposition process characteristics, can modify
the VO2 phase transition.22,23 Furthermore, the metal-insulator
variations of the refractive index (n) and extinction coefficient
(k) of about 50% and 97%, respectively, were measured by
means of Vis-NIR (0.6–2 μm) ellipsometry for a VO2 thin
film deposited on a silicon substrate.24 These significant
variations involve a change on the sample transmittance of
about 88% between room temperature (IR transparent) and

0021-8979/2018/124(19)/195102/7/$30.00 124, 195102-1 Published by AIP Publishing.

195102-2 Ramirez-Rincon et al.
70 °C (IR opaque). On the other hand, Qazilbash et al.25
reported the real part of the optical conductivity (i.e., the
imaginary part of the dielectric function) of a 100 nm-thick
film of VO2 grown on r-plane sapphire, across the MIT and
for wavelengths from 0.6 to above 15 μm. Based on the com-
bination of ellipsometric measurements and images of near-
field scattering, these latter authors showed that the surface
structure of the VO2 film changes with temperature, as a
result of the coexistence of the insulator and metallic phases
at temperatures T
TC . Taking into account that the imagi-
nary part of the optical conductivity (i.e., the real part of the
dielectric function) in the IR region still remains unexplored,
the complete measurement of the real and imaginary parts of
the VO2 dielectric function in a wide interval of visible and
IR wavelengths is required for temperatures within the MIT.
In this work, we measure the temperature dependence of
the ellipsometric signals Ψ and Δ of VO2 thin films deposited
on r-plane sapphire by means of pulsed laser deposition
(PLD). The real (εR ) and imaginary (εI ) parts of the VO2
dielectric function are determined in the spectral range of
0.4–15 μm, by fitting these experimental data to a combina-
tion of Tauc-Lorentz (TL), Gaussian (GOSC), and Drude
(DOSC) oscillator models, for both the heating and cooling
processes of VO2, within its MIT. Based on the εR and εI
experimental values and applying Fresnel’s equations and
Kirchhoff’s law, the hysteresis loops of the VO2 average
emissivity are found and compared with its effective values
(VO2 + sapphire substrate) reported in the literature.
II. MATERIALS AND METHODS
A. Sample preparation
VO2 thin films were deposited via the PLD technique,
which uses a pulsed high-power laser beam to evaporate
matter from a solid target. A KrF pulsed excimer laser (λ =
248 nm) with a pulsed width of 25 ns and a repetition rate of
25 Hz was employed to grow a VO2 film (120 nm) on r-plane
sapphire (0.5 mm) at the temperature of 600 °C and within an
ultra-high vacuum chamber with an oxygen pressure of 2.2
Pa. Our VO2 samples were grown by means of PLD because
FIG. 1. (a) Room temperature XRD pattern and (b) Raman spectra for the insulator (blue), insulator-metal (orange), and metallic (red) phases of the VO2 thin
film grown on r-plane sapphire.
J. Appl. Phys. 124, 195102 (2018)
of two reasons: First, this technique is well known for the
epitaxial growth of complex oxide films, due to the control
of high instantaneous flux of species in an oxidizing ambient
favorable for the formation of the desired phase.26–28 Second,
PLD is among the most convenient methods for growing
VO2 thin films with the highest contrast on their electrical
and optical properties across the MIT.29 The crystalline struc-
ture of a VO2 + r-plane sapphire sample was analyzed by
X-ray diffraction (XRD), whose room-temperature pattern is
shown in Fig. 1(a). This pattern was determined by a diffrac-
tometer D-8 advance operating with a Bragg-Brentano geom-
etry and CuKα1 radiation. According to the International
Centre for Diffraction Data (Card 04-003-2035), the XRD
peaks correspond to the monoclinic phase of VO2 and are
consistent with those found for a VO2 thin film deposited on
r-plane sapphire.23
The Raman spectra measured by means of a Confocal
Raman Witec Alpha300 spectrometer are shown in Fig. 1(b)
for three relevant different temperatures across the MIT.
In the insulating state (T  TC ), the spectrum exhibits nine
characteristic peaks (194, 225, 255, 309, 343, 390, 440,
500, and 620 cm−1) associated to the different vibration
modes of the VO2 monoclinic structure, as was previously
identified.14,21,24 This dielectric phase is mainly characterized
by the V-V mode at ∼620 cm−1 and the V-O modes at
194 cm−1 and 225 cm−1.22 As the VO2 sample temperature
increases (T
TC ), the intensity of the Raman peaks decreases
as a result of the appearance of the metallic domains within
the MIT. Finally, at higher temperatures (T  TC ) for
which the VO2 film reaches its metallic phase (tetragonal
structure), the Raman spectrum becomes featureless, as
expected for the Raman spectroscopy of metals.
B. Experimental setup
The ellipsometric characterization of our VO2 thin
film has been performed within its MIT and the spectral
range from 0.4 to 15 μm (incident angle 70°) in two
steps: First, the ellipsometric signals Ψ and Δ have been
measured from 0.4 to 1.7 μm using a M-2000 ellipsometer
195102-3 Ramirez-Rincon et al.
( J. A. Woollam Co., Inc.). For this experimental setup,
the sample temperature is measured with a K-thermocouple
(0.05 mm in diameter) connected to the thermocouple
monitor (2-Channel Handheld Digital Thermometer) and
controlled with a Peltier cell, which switches the applied
voltage in steps of 0.1 V every 5 min, leading to temperature
changes of about 2 °C in the range from 20 to 80 °C.
Second, the signals Ψ and Δ are measured in the complemen-
tary wavelength interval from 1.7 to 15 μm through the
IR-VASE Mark-II ellipsometer ( J. A. Woollam Co., Inc.)
equipped with a built-in cryostat (CRV-725V, Janis Research
Co.) that controls the sample temperature with a VASE
Ellipsometer Control Module (VB-400, J. A. Woollam Co.,
Inc., of sensitive 0.1 °C) in steps of 2 °C every 45 min, under
a pressure of 5 × 10−7 Torr. In both experimental setups, the
temperature has been measured on the heated surface rather
than on the VO2 one, which induces a temperature offset of
3 °C (10 °C) in Vis-NIR (IR) measurements compared to the
real temperature of the sample. These temperature shifts have
been taken into account throughout all our analysis.
C. Methodology
The ellipsometric signals Ψ and Δ measured for the VO2
sample in the spectral range from 0.4 to 15 μm, for different
temperatures within the MIT, have been analyzed by means
of the CompleteEASE™ data analysis software of J.A.
Woollam Co., Inc. The sample has been modeled as a three-
layer system involving the sapphire substrate, the VO2 thin
film, and the interface roughness. Taking into account that
the optical properties of sapphire are expected to remain
constant for the considered range of temperatures around the
VO2 MIT,30 the fitting process consists in assigning different
oscillator models (Tauc-Lorentz, Gaussian, and Drude) to the
VO2 layer and a Lorentzian oscillator to the substrate, which
enables to determine the spectral refractive index (n) and
extinction coefficient (k) of VO2 at different temperatures.
The real εR and imaginary εI parts of the complex dielectric
function
pffiffiffi ε have then been obtained by using the relation
ε ¼ n þ ik.31 More details about optical models and the
fitting parameters, used to obtain the dielectric function of
VO2, are shown in the supplementary material.
FIG. 2. Temperature dependence of the spectral ellipsometric signals (a) Ψ and (b) Δ of the VO2 + r-plane sapphire sample. The symbols þ, , and  stand for
the fitted data obtained using the Tauc-Lorentz (TL), Tauc-Lorentz + Gaussian (TL + GOSC), and Drude + Gaussian (DOCS + GOSC) oscillator models, respec-
tively. Dashed lines correspond to ellipsometric signals of the r-plane sapphire substrate at room temperature.
J. Appl. Phys. 124, 195102 (2018)
III. RESULTS
A. Dielectric function
The ellipsometric signals Ψ and Δ measured in the spec-
tral range from 0.4 to 15 μm for the VO2 thin film deposited
on r-plane sapphire are, respectively, shown in Figs. 2(a)
and 2(b) for three relevant temperatures. In the insulating
phase of VO2 (T  TC ), both Ψ and Δ exhibit a strong
dependence on the substrate properties (dashed lines), espe-
cially for wavelengths comparable to or higher than 2.0 μm,
where VO2 shows its maximum transmittance11,32 (i.e.,
εI ¼ 0). In this insulating phase, the VO2 dielectric function
is determined by fitting the experimental data (Ψ and Δ) with
the Tauc-Lorentz (TL) oscillator model, which has been suc-
cessfully applied in previous works.21,33,34 This fitting proce-
dure also enables to calculate the VO2 bandgap, whose value
for our sample is 0.35 eV, which is within the ranges reported
in the literature.21,25,33
As the sample temperature rises to values T
TC , the
spectral behavior of Ψ and Δ significantly changes due to the
metallic domains arising inside the insulating layer.7,32,33
This coexistence of the insulating and metallic phases
enables to describe the VO2 film as a composite material, for
which the fitting of both ellipsometric signals has been
carried out by including a Gaussian oscillator (GOSC) in the
Tauc-Lorentz (TL) model. The combination of these oscilla-
tors has been previously explored for numerous dielectric
compounds35,36 and has provided a suitable description of
the optical properties of solid crystalline samples in wide
ranges of wavelengths.37,38 Finally, at high temperatures
(T  TC ), the dielectric function has been determined by
combining a Gaussian oscillator with the Drude model
(DOSC + GOSC), as a consequence of the relatively low
optical properties of the metallic VO2 below 1.5 μm, com-
pared with those in the IR region (above 3 μm). The plasma
pffiffiffiffiffiffiffiffiffi
frequency ωp ¼ 1= ε0 ρτ ¼ 4:16 eV determined with the
resistivity (ρ = 2.86 Ω μm) and scattering time (τ = 0.98 fs) of
the VO2 thin film deposited on r-plane sapphire (see
Table S1 of the supplementary material) is comparable to the
one reported by Nazari et al.21 (4.0 eV) but higher than that
(2.62 eV) reported by Rajeswaran39 for a VO2 film deposited
on a glass substrate. This difference indicates that the
195102-4 Ramirez-Rincon et al.
substrate has a strong impact on the plasma frequency and
other properties of VO2 thin films, as shown in Table S2 of
the supplementary material. Even though we have also inves-
tigated other alternative models, such as those related with
Lorentzian oscillators,24 the best fitting of both signals Ψ and
Δ measured for our VO2 sample has been obtained by the
combination of the Tauc-Lorentz, Gaussian, and Drude oscil-
lators. The fitted spectra shown in Fig. 2 (and Fig. S3 in the
supplementary material) were obtained by minimizing the
mean square error of the fitting parameters and confining
the real and imaginary parts of the complex refractive index
to physically acceptable values (n > 0 and k > 0) throughout
the whole spectrum.
The real (εR ) and imaginary (εI ) parts of the dielectric
function in the spectral range from 0.4 to 15 μm are shown
in Figs. 3(a) and 3(c) and Figs. 3(b) and 3(d) for various
temperatures set during the heating and cooling processes
of the VO2 sample, respectively. In the insulator phase
(heating: 20 °C < T < 63 °C and cooling: 20 °C < T < 55 °C),
εR keeps near zero through the whole spectrum [see
Figs. 3(a) and 3(b)], as is the typical behavior of dielectric
materials, while εI  0 in the IR region [Figs. 3(c) and 3(d)],
as a result of the low reflectance of the VO2 film surface
at T  TC .40 On the other hand, at high temperatures
(T > 69 °C), the sample reaches the metallic phase character-
ized by negative values of εR and a monotonal increase of εI
with the NIR-IR wavelength, which is consistent with the
data reported by Qazilbash et al.25,33 for the real part of
FIG. 3. Temperature dependence of the real (εR ) and imaginary (εI ) parts of the dielectric function of the VO2 film, during its heating [(a) and (c)] and cooling
[(b) and (d)] processes, in the spectral range from 0.4 to 15 μm. Insets show the behavior of εR and εI in the shorter wavelength range (0.4–2.5) μm.
J. Appl. Phys. 124, 195102 (2018)
the optical conductivity and induces sizable changes on the
reflectance of VO2 films.37 At intermediate temperatures
(from 64 °C to 67 °C in the heating process and 65 °C to
57 °C in the cooling one), both εR and εI display a mixed
spectrum with a metallic-like behavior for wavelengths
smaller than 3 μm and 5 μm, respectively, and an insulator-
like phase around 10 μm. This latter behavior known as
anomalous absorptance13 is due to the interference processes
that appear at specific wavelengths, for which the substrate
exhibits high reflectivity and the optical constants of the VO2
film satisfy the condition n
k.41 In the present case, this
interference shows up during the predominant growth of
metallic domains inside the thin film (T
TC ), which
increases the values of εI (i.e., k = 0) around 10 μm, where
the sapphire substrate presents the highest IR reflectiv-
ity.11,41 According to the insets of Fig. 3, for visible wave-
lengths (from 0.4 to 0.8 μm), both the real and imaginary
parts of the dielectric function exhibit relatively small varia-
tions compared to those in the NIR region (above 1.5 μm)
for which the film behaves as a metal with εI  1 for
T  TC . These results are consistent with those reported
by Seal et al.42 and Nazari et al.21 for VO2 thin films
deposited on sapphire.
Figure 4 shows the temperature dependence of the real
(εR ) and imaginary (εI ) parts of the dielectric function at the
wavelength of 10 μm. Within the MIT, εR (εI ) decreases
(increases) during the heating process and increases
(decreases) for the cooling one following a different path,
195102-5 Ramirez-Rincon et al.
FIG. 4. Hysteresis loop of the real εR (red) and imaginary εI (blue) parts of
the VO2 dielectric function at λ = 10 μm for the heating (full symbols) and
cooling (empty symbols) processes.
which leads to a hysteresis loop. This behavior of εI has also
been found for the IR reflectance of VO2 films deposited on
sapphire,11,20 while that of εR has not been reported before,
to the best of our knowledge. The hysteresis curve of εR is
characterized by (i) negative values of εR for T  TC ,
which is characteristic of metals37 due to their usual high
extinction coefficient, and (ii) the presence of a maximum
related to the anomalous energy absorption of VO2, which
for the cooling process is higher than during the heating
one, as a consequence of the relatively longer coexistence
of both phases before returning to the insulator phase.
Furthermore, the transition temperature (average tempera-
ture of the peaks of @εR =@T ¼ 0 and @εI =@T ¼ 0 during
the heating and cooling processes23) of both εR and εI is
around TC  64:5  C, and the hysteresis width associated to
the MIT reversibility is ΔH  3:2  C. These values are con-
sistent with those recently determined for the VO2 inte-
grated emissitivity.43
B. Effective medium theory
Taking into account that the MIT of VO2 is driven by
the coexistence of metallic and insulating domains,3,7,25,33,44
FIG. 6. Comparison of the measured real εR (red) and imaginary εI (blue) parts of the VO2 dielectric function (full symbols) with their corresponding theoreti-
cal counterparts (stars) predicted by Eq. (1), as functions of the (a) wavelength and (b) temperature within the MIT. Calculations have been done with the
average volumetric metallic fraction shown in Fig. 5 for two representative wavelengths.
J. Appl. Phys. 124, 195102 (2018)
FIG. 5. Temperature evolution of the VO2 volumetric metallic fraction deter-
mined by means of Bruggeman’s model at 3 μm (circles) and 10 μm (stars)
for the heating (red) and cooling (blue) processes.
VO2 can be considered as a composite made up of metallic
(insulating) domains embedded in an insulating (metallic)
matrix,21 during its heating (cooling) process. The tempera-
ture evolution of these domains is characterized by the perco-
lation of the VO2 properties, which, for the case of the VO2
dielectric function ε, can be well described by Bruggeman’s
symmetrical model45 as follows:
εm (λ)  ε(λ) εi (λ)  ε(λ)
f þ (1  f ) ¼ 0, (1)
qεm (λ) þ (1  q)ε(λ) qεi (λ) þ (1  q)ε(λ)
where f and 1  f stand for the volumetric fractions of the
metallic and dielectric domains during the heating process,
respectively, εm and εi are the respective dielectric functions
of VO2 in its metallic and insulating phases, and q is the
depolarization factor defined by the average shape of
domains. Considering the dimensionality of our VO2 film
sample (thickness = 120 nm and area = 10 × 10 mm2), we
have taken q = 0.5 (flat domains mainly), as was considered
previously for this effective medium theory (EMT).46 For the
cooling process, the insulating domains show up inside the
metallic matrix, and therefore, the volume fractions f and 1  f
195102-6 Ramirez-Rincon et al.
must be interchanged in Eq. (1). By inserting the experimental
values of ε, εm , and εi shown in Fig. 3 into Eq. (1), the tem-
perature evolution of f has been determined and is shown in
Fig. 5 for two IR wavelengths, wherein VO2 exhibits the
largest differences between the insulating and metallic phases.
Note that for the heating process, f increases monotonously
from zero, in the insulating phase, to one, in the metallic
phase, as expected. At the critical temperature TC = 64 °C,
f  0:1 for both wavelengths. In the cooling process, on the
other hand, the metallic fraction remains around 0.8 until
66 °C and then linearly falls down to zero, leading to the
expected hysteresis loop. The similar behavior and values
of f for both IR wavelengths indicate that the volumetric
metallic fraction is nearly independent of the wavelength, as
established by Eq. (1). The percentage of metallicity in the
VO2 surface modifies the optical spectral properties mostly
in the IR region. Therefore, it is possible to associate these
changes directly to the volumetric metallic fraction, as shown
in Fig. S5 of the supplementary material, for two VO2
samples deposited on different substrates.
Figures 6(a) and 6(b) show the comparison of the mea-
sured real εR (red) and imaginary εI (blue) parts of the VO2
dielectric function with their corresponding theoretical coun-
terparts predicted by Eq. (1), as functions of the wavelength
and temperature, respectively. Note that the theoretical values
of εR are in good agreement (deviation smaller than 15%)
with the experimental data in the Vis-NIR region (0.4–3 μm),
while those of εI show a good description of their experimental
counterparts in the spectral range 0.4–5 μm. It is therefore
clear that Bruggeman’s EMT can describe well the spectral
behavior of the VO2 dielectric function for wavelengths
smaller than 3 μm mainly.
C. Average VO2 emissivity
The VO2 dielectric function shown in Fig. 3 is now
used to determine the reflectance r(λ, T) and transmit-
tance t(λ, T) of VO2 by means of Fresnel’s equations for
normal incidence.47,48 According to Kirchhoff’s law, the
spectral emittance is then given by 1  r(λ, T)  t(λ, T),
whose average value (ϵ) over the spectral wavelengths λ can
FIG. 7. (a) Spectral ratio of power emitted (REP) of a VO2 film and its (b) average emissivity ϵ obtained for its heating (red) and cooling (blue) processes
across the MIT. The spherical and square dots represent the values of ϵ determined by means of ellipsometry [Eq. (2)] and the TWRC technique.38
J. Appl. Phys. 124, 195102 (2018)
be calculated by46
Ð1
[1  r(λ)  t(λ)]Θ(λ, T)dλ
ϵ¼ 0
Ð1
0
Θ(λ, T)dλ
1 1 Ð
¼ [1  r(λ)  t(λ)]Θ(λ, T)dλ, (2)
σT 4 0
where Θ(λ, T) ¼ C1 λ5 =(eC2 =λT  1) is the blackbody spectral
radiance, C1 ¼ 1:48  1015 W m2 , C2 ¼ 1:43  102 m K,
and σ ¼ 5:67  108 W m2 K4 .
The spectral ratio of emitted power REP ¼ [1  r(λ)
t(λ)]Θ(λ, T)=σT 4 by the VO2 thin film is shown in Fig. 7(a)
for wavelengths from 2 to 20 μm. The spectral values above
15 μm have been obtained at each temperature by extrapola-
tion. In the insulating phase (T < 60 °C), REP increases with
temperature and its maximum changes of wavelength, as estab-
lished by Wien’s law. Within the MIT (T
68  C), r(λ, T)
increases, t(λ, T) decreases, and REP falls down until reaching
its metallic values. In addition, note that the main contribution
to the average emissivity ϵ arises from the spectral range from
3 to 15 μm, where the VO2 dielectric function has been mea-
sured. According to Fig. 7(b), the values of ϵ become indepen-
dent of temperature for the pure metallic and insulating
phases, as expected. Within the MIT, ϵ decreases during the
heating process and increases for the cooling one, following
a different path, which leads to a hysteresis loop similar
(TC ¼ 64:5  C, ΔH ¼ 3:1  C) to those found for the real
and imaginary parts of the dielectric function (see Fig. 4).
In the insulating phase, the value ϵ ¼ 0:49 is lower than
that measured by the thermal-wave resonant cavity (TWRC)
technique,40 while the metallic emissivity ϵ ¼ 0:16 agrees
rather well with the value measured with this later tech-
nique. This agreement in the metallic phase is reasonable,
given that the VO2 film is totally IR opaque at high temper-
ature, and hence, the substrate effect becomes negligible.
On the other hand, the difference in the insulating phase is
explained by the IR semi-transparency of the VO2 thin
film, and therefore by the sapphire substrate, which is con-
sidered by our ellipsometric measurements, but not by the
TWRC ones that considered the VO2 + substrate like a
single effective layer.43 The emissivity values shown in
195102-7 Ramirez-Rincon et al. J. Appl. Phys. 124, 195102 (2018)

Fig. 7(b) thus correspond to the VO2 thin film without the 12
S. Li, G. A. Niklasson, and C. G. Granqvist, J. Appl. Phys. 115, 053513
substrate effect. (2014).
13
M. A. Kats, R. Blanchard, S. Zhang, P. Genevet, C. Ko, S. Ramanathan,
and F. Carpasso, Phys. Rev. X 3, 041004 (2013).
14
IV. SUMMARY M. Benkahoul, M. Chaker, J. Margot, E. Haddad, R. Kruzelecky,
B. Wong, W. Jamroz, and P. Poinas, Sol. Energy Mater. Sol. Cells 95,
We have measured the real and imaginary parts of the 3504 (2011).
dielectric function of VO2 thin films deposited on r-plane
15
J. Ordonez-Miranda, Y. Ezzahri, J. Drevillon, and K. Joulain, J. Appl.
Phys. 119, 203105 (2016).
sapphire via pulsed laser deposition by means of visible-IR 16
K. Joulain, Y. Ezzahri, J. Drevillon, and P. Ben-Abdallah, Appl. Phys.
ellipsometry for wavelengths ranging from 0.4 to 15 μm and Lett. 106, 133505 (2015).
temperatures within its full phase transition. This first-ever 17
H. Prod’homme, J. Ordonez-Miranda, Y. Ezzahri, J. Drevillon, and
direct measurement of the VO2 dielectric function has been K. Joulain, J. Appl. Phys. 119, 194502 (2016).
18
I. G. Madida, A. Simo, B. Sone, A. Maity, J. B. Kana Kana, A. Gibaud,
achieved by combining the Tauc-Lorentz, Gaussian, and G. Merad, F. T. Thema, and M. Maaza, Sol. Energy 107, 758 (2014).
Drude oscillator models for describing the two ellipsometric 19
G. Leahu, R. Li Voti, C. Sibilia, and M. Bertolotti, Appl. Phys. Lett. 103,
signals in the insulating, insulating-metallic, and metallic 231114 (2013).
phases of VO2. By comparing the experimental data with the
20
H. S. Choi, J. S. Ahn, J. H. Jung, and T. W. Noh, Phys. Rev. B 54, 4621
(1996).
predictions of Bruggeman’s effective medium model, it has 21
M. Nazari, Y. Zhao, V. V. Kuryatkov, Z. Y. Fan, A. A. Bernussi, and
been shown that this model is suitable to describe the VO2 M. Holtz, Phys. Rev. B 87, 035142 (2013).
dielectric function for wavelengths smaller than 3 μm mainly. 22
V. Thèry, A. Boulle, A. Crunteanu, J. C. Orlianges, A. Beaumont,
Furthermore, we have also determined the VO2 emissivity R. Mayet, A. Mennai, F. Cosset, A. Bessaudou, and M. Fabert, Phys.
Rev. B 93, 184106 (2016).
averaged over a relevant interval of wavelengths and found 23
T. H. Yang, R. Aggarwal, A. Gupta, H. Zhou, R. J. Narayan, and
that its values exhibit a hysteresis loop and change a factor of J. Narayan, J. Appl. Phys. 107, 053514 (2010).
three between its insulating and metallic phases. 24
J. Houska, D. Kolenaty, J. Rezek, and J. Vlcek, Appl. Surf. Sci. 421, 529
(2017).
25
M. M. Qazilbash et al., Phys. Rev. B 79, 075107 (2009).
SUPPLEMENTARY MATERIAL 26
M. Lorenz, H. Hochmuth, D. Natusch, H. Börner, G. Lippold, K. Kreher,
and W. Schmitz, Appl. Phys. Lett. 68, 3332 (1996).
See supplementary material for a comparison of the 27
R. K. Singh and D. Kumar, Mater. Sci. Eng. R: Rep. 22, 113 (1998).
optical properties of a VO2 thin film deposited on r-plane 28
D. B. Chrisey and G. K. Hubler, Pulsed Laser Deposition of Thin Films
sapphire with those measured for another one deposited on (Wiley, New York, 1994).
c-plane sapphire.
29
J. Nag and R. Haglund, J. Phys.: Condens. Matter 20, 264016 (2008).
30
M. E. Thomas, R. I. Joseph, and J. Tropf, Appl. Opt. 27, 239 (1988).
31
C. M. Herzinger, B. Johs, W. A. McGahan, and A. Woollam, J. Appl.
ACKNOWLEDGMENTS Phys. 83, 3324 (1998).
32
L. Jiran, H. Ming, K. Qiang, L. Xiuqin, W. Xiaodong, L. Guike, and
This work has been partially supported by the Projects C. Hongda, Rare Metals 30, 247 (2011).
192 “Fronteras de la ciencia,” 251882 “Investigación M. M. Qazilbash et al., Science 318, 1750 (2007).
33
34
M. M. Qazilbash, A. A. Schafgans, K. S. Burch, S. J. Yun, B. G. Chae,
Científica Básica 2015,” 207450 “CeMIESol” within the
B. J. Kim, H. T. Kim, and D. N. Basov, Phys. Rev. B 77, 115121
strategic project No. 10, COSOL-pi, the Conacyt-Secretaría (2008).
35
de Energía-Sustentabilidad Energética Fund, and the Cellule S. D. Pop, K. Hinrichs, S. Wenmackers, C. Cobet, N. Esser, and
Énergie du CNRS through Grant No. 267745. J.A.R.R. and D. R. Zahn, in Ellipsometry of Functional Organic Surfaces and Films,
edited by K. Hinrichs and K. Eichhorn (Springer, Berlin, 2018).
C.L.G.H. thank Conacyt for their Ph.D. scholarships as well 36
D. De Sousa Meneses, G. Gruener, M. Mailki, and P. Echegut,
as the “Becas Mixtas” fund for supporting their stay at the J. Non-Cryst. Solids 351, 124 (2005).
37
Pprime Institute of the CNRS in France. The authors also H. G. Tompkins and E. A. Irene, Handbook of Ellipsometry (William
acknowledge the technical support of Hélène Grassin for her Andrew, New York, 2005).
38
J. Orava, T. Wagner, J. Sik, J. Prikryl, M. Fumar, and L. Benes, J. Appl.
assistance in performing the ellipsometric measurements. Phys. 104, 043523 (2008).
39
B. Rajeswaran, J. K. Pradhan, S. A. Ramakrishna, and A. M. Umarji,
1
F. J. Morin, Phys. Rev. Lett. 3, 34 (1959). J. Appl. Phys. 122, 163107 (2017).
2
M. Imada, A. Fujimori, and Y. Tokura, Rev. Mod. Phys. 70, 1039 (1998). 40
L. Kang, Y. Gao, H. Luo, Z. Chen, J. Du, and Z. Zhang, ACS Appl.
3
A. C. Jones, S. Berweger, J. Wei, D. Cobden, and M. B. Raschke, Mater. Interfaces 2, 135 (2011).
Nano Lett. 10, 5 (2010). 41
M. A. Kats, D. Sharma, J. Lin, P. Genevet, R. Blanchard, Z. Yang,
4
J. M. Atkins, S. Berweger, E. K. Chavez, M. B. Raschke, J. Cao, W. Fan, M. M. Qazilbash, D. N. Basov, S. Ramanathan, and F. Capasso,
and J. Wu, Phys. Rev. B 85, 0201101 (2012). Appl. Phys. Lett. 101, 221101 (2012).
5
M. Yang et al., Sci. Rep. 6, 23119 (2016). 42
K. Seal, A. Sharoni, J. M. Messman, B. S. Lokitz, R. W. Shaw,
6
M. Sullivan, T. V. Son, N. Beaudoin, and A. Hache, Opt. Commun. 356, I. K. Shuller, P. C. Snijders, and T. Z. Ward, Sci. Rep. 4, 6529 (2014).
43
395 (2015). C. L. Gomez-Heredia, J. A. Ramirez-Rincon, J. Ordonez-Miranda,
7
S. Kumar, J. P. Strachan, A. L. Kilcoyne, T. Tyliszczak, M. D. Pickett, O. Ares, J. J. Alvarado-Gil, C. Champeaux, F. Dumas-Bouchiat,
C. Santori, G. Gibson, and R. S. Williamns, Appl. Phys. Lett. 108, 073102 Y. Ezzahri, and K. Joulain, Sci. Rep. 8, 8479 (2018).
(2016). 44
D. Bruggemann, Ann. Phys. 24, 636 (1935).
8 45
X. Wu, F. Lai, L. Lin, Y. Li, L. Lin, Y. Qu, and Z. Huang, Appl. Surf. Sci. T. J. Huffman, D. J. Lahneman, S. L. Wang, T. Slusar, B. Kim, H. Kim,
255, 2840 (2008). and M. M. Qazilbash, Phys. Rev. B 97, 085146 (2018).
9
D. Kucharcsyyk and T. Niklewski, J. Appl. Crystallogr. 12, 370 (1979). 46
G. L. Carr and S. Perkowitz, Far-Infrared Properties of Inhomogeneous
10
D. W. Oh, C. Ko, S. Ramanathan, and D. G. Cahill, Appl. Phys. Lett. 96, Materials (Academic Press Inc., FL, 1985).
47
151906 (2010). J. D. Jackson, Classical Electrodynamics (John Wiley & Sons, Inc., NJ,
11
A. Crunteanu, M. Fabert, J. Cornette, M. Colas-Dutreilh, J. C. Orlianges, 2001).
A. Bessaudou, and F. Cosset, Proc. SPIE 9364, 93640J1 (2015). 48
D. M. Trotter and A. J. Siervers, Appl. Opt. 19, 711 (1980).