Journal of Physics D:

Applied Physics

PAPER You may also like

- Global optimization for future gravitational
Optimization of pyroelectric figures of merit via wave detector sites
Yi-Ming Hu, Péter Raffai, László Gondán
magnesia doping in lithium tantalate single crystal et al.

- Material strategies to enhance the

performance of piezoelectric energy
To cite this article: Feng Tang et al 2018 J. Phys. D: Appl. Phys. 51 395101 harvesters based on lead-free materials
Ausrine Bartasyte, Giacomo Clementi,
Quentin Micard et al.

- Wrap-Gate CNT-MOSFET Based SRAM

Bit-Cell with Asymmetrical Ground Gating
View the article online for updates and enhancements. and Built-In Read-Assist Schemes for
Application in Limited-Energy
Environments
Abdolreza Darabi, Mohammad Reza
Salehi and Ebrahim Abiri

This content was downloaded from IP address 76.229.184.205 on 20/12/2024 at 00:45

 Journal of Physics D: Applied Physics

J. Phys. D: Appl. Phys. 51 (2018) 395101 (7pp) https://doi.org/10.1088/1361-6463/aad88d

Optimization of pyroelectric figures of merit

via magnesia doping in lithium tantalate
single crystal
Feng Tang1, Siwei Long1, Xin Yang1, Mingming Yang1, Jiliang Quan1,
Shaopeng Lin2, Decai Ma2, Yunzhong Zhu2 and Biao Wang1,2
1
State Key Laboratory of Optoelectronic Materials and Technologies/Institute of Optoelectronic and
Functional Composite Materials, School of Physics and Engineering, Sun Yat-sen University, Guangzhou
510275, People’s Republic of China
2
Sino French Institute of Nuclear Engineering and Technology, Sun Yat-sen University, Zhuhai 519082,
People’s Republic of China

E-mail: zhuyzh7@mail.sysu.edu.cn (Y Zhu) and wangbiao@mail.sysu.edu.cn (B Wang)

Received 31 January 2018, revised 27 July 2018

Accepted for publication 7 August 2018
Published 21 August 2018

Abstract
This paper reports a series of experiments and optimization based on pyroelectric figures of
merit (FoMs), which demonstrate that the proper doping of magnesia (MgO) can improve the
thermal performance of lithium tantalate (LiTaO3) in infrared sensors. Starting with MgO-
doped LiTaO3 grown by the Czochralski method, the heat capacity, pyroelectric coefficients
and dielectric parameters of these single crystals are then investigated systematically. By
considering the ultraviolet and infrared spectra, the variation of the electrical parameters and
the occupation mechanism of MgO doping could be explained. It is found that the dopant
increases the Curie temperature and broadens the operating temperature of pyroelectric
devices. Notably, compared with congruent LiTaO3, when the doping concentration reaches
5 mol%, the corresponding voltage responsivity and specific detectivity FoMs are enhanced by
ten and two times, respectively.
Keywords: lithium tantalate, figures of merit, pyroelectric coefficient, dielectric property,
Curie temperature

(Some figures may appear in colour only in the online journal)

1. Introduction method [10], possesses two advantages: convenient temper­

ature control and charge collection. Specifically, when the
Pyroelectric materials are widely investigated for thermal electrodes of the pyroelectric single crystal are linked in an
imaging and infrared sensors [1–5]. Here, it is reported that open circuit, a programmed temperature oscillation could
temperature changes undergone by a pyroelectric crystal lead induce current flow. The current stemming from the pyroelec­
to opposite surface charges accumulating along the polariza­ tric charge is given as I  =  pA dT/dt (where A is the electrode
tion direction that extends across the thickness of the crystal area and dT/dt is the rate of temperature change) [9].
[6]. This phenomenon is further quantified as the pyroelec­ The majority of existing literature related to pyroelectricity
tric coefficient p  =  ∂PS/∂T, which implies that the working focuses on improving the coefficient p of the specific mat­
temperature (T) could oscillate the spontaneous polarization erial. However, a high pyroelectric coefficient does not ensure
(PS) of pyroelectric materials [7]. In a standard pyroelectric excellent thermal performance [11]. In addition to the coef­
measurement system, the required temperature oscillation is ficient, a low dielectric loss tangent [12, 13] and low permit­
usually achieved via thermal radiation or conduction [8–10]. tivity [13, 14] are also required. A high Curie temperature
The thermal conduction method, also known as the dynamic and stable chemistry are also essential to obtain acceptable

1361-6463/18/395101+7$33.00 1 © 2018 IOP Publishing Ltd Printed in the UK

J. Phys. D: Appl. Phys. 51 (2018) 395101 F Tang et al

pyroelectric performance over a wide temperature range D8) was used to check the crystallinity [23] in several square
and diverse working environment. For instance, due to their millimeters of the wafer. As shown in figure 1(b), a single dif­
low Curie temperature, water solubility, chemical instability fraction peak with the full width at half maximum of 33.696
issues and environmental pollution problems, common pyro­ arcsec was obtained from the 5-LT wafer. The narrow and sym­
electric materials such as TGS, PVDF, PT and PZT [15–18] metric peak, located at 19.425°, suggested that the LT wafer
cannot meet the demanding requirements of complex working was a perfect single crystal. A Parttulab HDMS and an Agilent
environments. Hence, it is vital to find a robust pyroelectric E4980A precision LCR meter were used to measure the
material with both high-temperature stability and excellent dielectric constant and dielectric loss under different temper­
pyroelectric performance. atures (50 °C–350 °C) and continuously changing frequencies
To solve this problem, lithium tantalite (LiTaO3, LT) and (0.1–100 kHz). The –OH absorption spectra of Mg:LT wafers
lithium niobate (LiNbO3, LN) are considered [19]. Both were obtained at room temperature using a Fourier transform
LT and LN are known to have a high Curie temperature. infrared (IR) spectrometer. In addition, the dopant occupation
Specifically, compared with LN, LT possesses a relatively site was investigated by a UV–vis–NIR spectrophotometer
low Curie temperature (665 °C) but much higher pyroelec­ (Lambda 950) as well. Changes in the Curie temperatures
tric coefficient (1.7  ×  10−4 C m−2 K−1) [11]. To estimate the (TC) and specific heat capacity for different doping concentra­
practical performance of this pyroelectric material, pyroelec­ tions was monitored using differential scanning calorimetry
tric figures of merit (FoMs) can be referenced [11, 20, 21]. (DSC-204F1, Phoenix). In this work, the dynamic method
In infrared sensors, two types of FoMs are used to evaluate was used to measure the pyroelectric coefficient, using a test
the performance: voltage responsivity (FV) and specific detec­ platform presented in our previous work [22]. Specifically,
tivity (FD), the temper­atures were controlled by a Linkam T95 together
with a HFS600E stage, and the pyroelectric current across the
FV = p/CP εr ε0
(1)
single crystal was measured by a Keithley 6485 picoammeter
1/2 with the accuracy of 0.4%.
(2)
FD = p/CP (εr ε0 tan δ) ,
where Cp represents the volumetric specific heat capacity, εr 3. Results and discussion
is the relative dielectric constant, ε0 is the permittivity of free
space and tanδ is the dielectric loss [22]. The voltage responsivity and specific detectivity of pyro­
In this paper, seven doped LT single crystals are grown electric infrared sensors not only depend on the pyroelectric
from congruent melt with different Mg doping concentrations coefficient, but the dielectric constant, dielectric loss and heat
(0–6.0 mol%) by the Czochralski method (CZ). Infrared and capacity of the specific material. In this section, we will pro­
ultraviolet spectra were recorded to determine the occupation vide detailed illustrations of all relevant experimental results
site of the dopant. Then, based on the effect of Mg doping along with site occupation analysis based on the infrared and
on the spontaneous polarization and the dipole moment of LT ultraviolet spectra and the corresponding variation of sponta­
single crystals, the variation on the pyroelectric coefficient and neous polarization and electric dipole moment. Then, in the
permittivity could be explained. Combining the above results, last part of this section, we will evaluate dopant and temper­
we find that the pyroelectric FoMs are enhanced dramatically ature-based pyroelectric FoM results and provide quantitative
and the working temperature is also widened. insight into the improvement of metrics related to FoMs.

2. Materials and methods 3.1. Infrared and ultraviolet spectra

In our experiment, Li2CO3 (4N purity) and Ta2O5 (4N purity) Figure 2(a) shows the original IR transmission spectra
powders were mixed with a molar ratio of Li/Ta  =  0.951. (inset) and the collection of –OH stretching vibration [24]
MgO powders (5N purity) were added to obtain doping con­ of as-grown crystals with different MgO doping concen­
centrations of 0, 1, 2, 3, 4, 5 and 6 mol% (defined as CLT, trations. When the Mg2+ concentration is below 3 mol%,
1-LT, 2-LT, 3-LT, 4-LT, 5-LT and 6-LT). The raw powders the peak position gradually shifts to a longer wavelength.
were mixed and ball milled to achieve a uniform mixture. However, when it exceeds 4 mol%, the peak position shifts
Each mixture was heated to 1250 °C for 24 h in a muffle fur­ back slightly to a shorter wavelength. The fundamental
nace. Then, single crystals were grown by a CZ furnace (as optical absorption edge of ultraviolet (UV edge) spectra can
shown in figure 1(a)) in a nitrogen–oxygen gas mixture (98% also be used to establish the location of impurities in the
N2 and 2% O2). The growth direction was the z-axis [0 0 1], crystal lattice. Figure 2(b) shows the UV spectra and the col­
using rotation and pulling rates of 6–15 rpm and 0.5–1.0 mm lection of absorption edges (α  =  20 cm−1) for the seven as-
h−1, respectively. Then the as-grown crystals were annealed grown crystals. One can see that as the doping concentration
and poled to a single domain state. Finally, as shown in the is increased from 0 to 3 mol%, the absorption edge shifts
inset of figure 1(b), the crystals were cut into z-wafers, pol­ to a shorter wavelength, then returns sharply to 4 mol%.
ished on both sides, and Au electrodes were deposited by elec­ After that, as the doping concentration increases, again, the
tron beam evaporation (EB 450). X-ray rocking curve (Bruker absorption edge shifts back to a shorter wavelength.

2
J. Phys. D: Appl. Phys. 51 (2018) 395101 F Tang et al

Figure 1. (a) The CZ single crystal growth furnace. (b) The c-ray rocking curve of 5-LT wafer, and the full width at half maximum
(FWHM) is 33.696 arcsec. The inset presents the Au deposited LT wafers.

Figure 2. (a) The –OH absorption peak position as a function

of Mg concentration in Mg:LT crystals, and the original IR
transmission spectra of all samples are presented in the inset. Figure 3. (a) The relationship between the Curie temperature (TC)
(b) The UV absorption edge position of Mg:LT crystals as a and Mg doping concentration, the red line represents the Curie
function of Mg concentration, and the UV absorption edges of all temperature of Mg:SLT in [25]. (b) The specific heat capacities
samples are presented in the inset. (C) of Mg:LT are measured at 50 °C–250 °C. (a) Reprinted from
[25], Copyright (2006), with permission from Elsevier.
3.2. Curie temperature
ferroelectric phase is a prerequisite of the LT pyroelectric wafer.
Curie temperature (TC) is the temperature where the ferroelec­ Figure 3(a) shows the Curie temperature of all the samples. One
tric phase transforms to the nonpyroelectric paraelectric phase. can find that, in our Mg:LT single crystals, the Curie temperature
Since the pyroelectricity disappears in the paraelectric phase, and gradually increases with Mg doping. However, the dopant has
a high TC is required for most applications, then maintaining the little influence on the TC of the stoichiometric LT (SLT) [25].

3
J. Phys. D: Appl. Phys. 51 (2018) 395101 F Tang et al

3.3. Specific heat capacity

In order to evaluate the specific detectivity potential and the

voltage responsivity of a pyroelectric sensor, the specific heat
capacity is needed. As shown in figure 3(b), the heat capaci­
ties of all the samples are measured from 50 °C–250 °C. The
overall data tend to increase with the working temperature and
display a small dependence on dopant level. It should be noted
that the measured heat capacity is the mass specific heat (C)
for the single crystal. In calculating FoMs, the mass specific
heat capacity should be transformed into the volumetric spe­
cific heat capacity (CP  =  C  ×  ρ, where ρ  =  7.45 g cm−3 is the
density of the CLT crystal).

3.4. Pyroelectric coefficient

Figure 4(a) shows the pyroelectric coefficient of all the sam­

ples over a wide temperature range from 50 °C–250 °C. In
general, when the working temperature is far below the Curie
point, pyroelectric coefficients gradually rise with the temper­
ature. More interestingly, when the doping concentration is
below the threshold concentration of 4 mol%, the pyroelec­
tric coefficient decreases significantly from the undoped
crystal. However, when the MgO concentration approaches
4 mol%, the corresponding pyroelectric coefficient presents a
huge improvement. Furthermore, as the doping concentration
exceeds 4 mol%, the pyroelectric coefficient decreases again.

3.5. Dielectric property

The relative permittivities of seven samples in various temper­

ature ranges (from 50 °C–350 °C) and frequency ranges (from
0.1–100 kHz) are shown in figure 4(b). One can easily see
that the seven permittivity curves increase with temperature
and decrease with frequency. The dielectric drops with the
increase of doping concentration from CLT to 3-LT. However,
in 4-LT, the relative permittivity shows an obvious increase,
and then, as the doping concentrations reach 5 and 6 mol%,
the permittivity decreases sharply.
In the majority of research of FoMs, dielectric con­
stants are recorded at 1 kHz [11, 26]. Thus, in this paper, the
di­electric loss (tanδ) of all the samples is measured at 1 kHz.
As presented in figure 4(c), the tanδ curves of the seven sam­
ples are similar and are close to zero below 200 °C. However,
at high working temperature, they increase and separate.
Specifically, at high temperature, the dielectric loss decreases
with the rising MgO concentration (from 0 to 3 mol%), and Figure 4. (a) The pyroelectric coefficients of seven samples are
measured at 50 °C–250 °C. (b) The relative dielectric constants
then, reaches the highest value in 4-LT. After that, from
variation of seven samples under different temperatures
5 to 6 mol%, the dielectric loss decreases again. (50 °C–350 °C) and frequencies (0.1–100 kHz). (c) The dielectric
loss of all samples in various temperature at 1 kHz.

3.6. Discussion of electrical parameters ( V−

Li ) to maintain charge balance [28]. Specifically, in doped
samples, the substitution of Mg2+ for Li+ and Ta5+ will vary
A consideration of the MgO substitution process in CLT the spontaneous polarization in different ways, and signifi­
single crystal, as reflected in the IR and UV spectra in figure 2, cantly influence the pyroelectric coefficient [29]. Meanwhile,
could explain the variation of electrical parameters offered in high ionic polarizability will induce strong distortions of the
figure 4. According to the Li-vacancy model [27]: some Ta5+ oxygen octahedron and results in a large dipole moment [30],
ions occupy Li sites and form anti-site (Ta4+ Li ) defects, where which increases the dielectric susceptibility (χ) and dielectric
4+
one TaLi will require the introduction of four Li vacancies coefficient (εr  =  1  +  χ), and vice versa [31, 32].
4
J. Phys. D: Appl. Phys. 51 (2018) 395101 F Tang et al

In figure 2(a), the initial IR transmission peak (in 3485 cm−1)

can be attributed to the VLi–OH complex, which consists of
Li vacancies and H+ ions from H2O molecules in the air [33].
For the case of 0 to 3 mol% Mg-doping, the IR shift (to longer
wavelengths) could be explained by the weaker repelling force
4+ +
of Mg+ Li than TaLi to H ions. At the same time, the UV edge
in figure 2(b) is determined by the valence electron transition
energy from the 2p orbital of O2− to the 5 d orbital of Ta5+.
The order of polarizability (Ta5+  >  Mg2+  >  Li+) suggests
that the blueshift of UV edge could be due to the substitu­
tion for Ta5+, which broadens the band gap of valence electron
transition [34, 35]. Thus, an initial occupation process (where
Mg substitutes for the Ta at the Li site, Ta4+ +
Li → MgLi ) would
weaken the spontaneous polarization and the dipole moment
of oxygen octahedron (from TaLiO6 to MgLiO6 octahedron),
which would decrease the corresponding pyroelectric coeffi­
cient and dielectric susceptibility. Then, from 3 to 4 mol%,
the redshift of UV edge could be due to the substitution of Li+
ions (LiLi → Mg+ Li). Now, the stronger polarizability of the
substitution ion would enhance the spontaneous polarization.
In addition, the dipole moment of the oxygen octahedron is
enhanced from LiLiO6 to MgLiO6. These improvements result
in an increased pyroelectric coefficient and relative permit­
tivity. After that, as shown in figures 2(a) and (b), when the
doping concentration exceeds 4 mol%, the blueshift of UV
edge implies the extra Mg2+ starts to replace normal Ta sites,
and the IR shift (to shorter wavelength) is attributed to the
stronger attraction of MgTa–OH complexes to H+ ions. Again,
due to the polarizability difference between Mg2+ and Ta5+,
this substitution (TaTa → Mg3− Ta ) will reduce the spontaneous
polarization and dipole moment (from TaTaO6 to MgTaO6 octa­
hedron), and so will reduce the corresponding pyroelectric Figure 5. The voltage responsivity (FV) and specific detectivity
coefficient and relative permittivity. (FD) versus various doping concentrations under different
Comparing light and heavy doping processes, the permit­ temperatures.
tivity varies with different doping concentrations. This
could be explained by the huge difference between TaLiO6 affected dramatically by temperature and dopant, the FV
and TaTaO6 octahedrons in the perovskite-like LT structure. and FD behave differently. Specifically, in figure 5(a), FV
Specifically, along the z-axis, one TaLiO6 locates between two decreases weakly when the doping concentration is below
normal Ta ions. Under light doping, Mg occupation will break 3 mol%, then, it increases and reaches the highest value
the symmetry of TaTa–TaLi and form TaTa–MgLi. Obviously, in 5-LT. Also, due to the crucial effect of temperature on
the newly formed asymmetric structure enhances the dipole di­electric property and pyroelectric coefficient, the value of
moment. Considering the influence of both the oxygen octahe­ FV increases with temper­ature. In particular, at high temper­
dron and Ta/Li/Ta complex to Mg occupation, the reduction of ature, the FV (5 mol%) is improved more than ten times to
permittivity is limited. However, in heavy doping process, the CLT. Then, from figure 5(b), we see that the initial Mg doping
complex behaves differently. Since TaTaO6 is located between (ranges from 0–3 mol%) is detrimental to FD. However, the
Li ions, Mg occupation dramatically reduces the asymmetry FD of 4-LT and 5-LT are nearly double that of CLT. Finally, as
from LiLi–TaTa to LiLi–MgTa structure. Thus, in this case, the the doping concentration reaches 6 mol%, FD decreases again.
dopant decreases the permittivity significantly. This wave-shaped FD variation is mainly due to the compe­
tition between the dielectric and pyroelectric improvements.
3.7. Figures of merit
Table 1 shows the summary of the electrical parameters and
FoMs for various pyroelectric materials. Obviously, the out­
After measuring and analyzing the specific heat capacity, standing FV and FD for 5-LT represents remarkable potential
pyroelectric coefficient, relative permittivity and dielectric pyroelectric device application. For instance, in our work,
loss, finally, two types of FoMs (FV, FD) can be calculated the FV is 20 and 11 times greater than CLN and Zr:LN [22].
via equations (1) and (2). Figures 5(a) and (b) show the FV Compared with the newly produced Fe:KLTN single crystal
and FD of seven samples at different working temperatures, [36], the FV is enhanced 77 times. Moreover, even compared
respectively. Since pyroelectric and dielectric properties are with PVDF, SBN-50 and TGS [11], the FV of 5-LT shows

5
J. Phys. D: Appl. Phys. 51 (2018) 395101 F Tang et al

Table 1. Summary of electrical parameters and FoMs for kinds of pyroelectric materials. [11, 22, 36].

p (10−4 C Cp (J cm−3 FD (10−5

Material m−2 K−1) εr K−1) TC (°C) FV (m2 C−1) (m/F)1/2 Cm/J) Test condition
5 mol% Mg:LT 7.89 13 3.66 699.8 1.84 28.82 1 kHz, 200 °C
5 mol% Mg:LT 2.01 12 3.2 699.8 0.59 19 1 kHz, 50 °C
2 mol% Mg:LT 0.15 65 2.91 675.2 0.009 0.55 1 kHz, 50 °C
CLT 2.3 72 3.02 625.1 0.12 14 1 kHz, 50 °C
PVDF [11]a 0.27 12 2.43 80 0.1 0.88 10 Hz, 50 °C
SBN-50 [11]b 5.5 400 2.34 121 0.07 7.2 —
TGS [11]c 5.5 55 2.6 49 0.43 6.1 1 kHz, 35 °C
CLN [22]d 0.83 35 2.99 1140 0.09 2.8 1 kHz, 50 °C
Zr:LN [22]e 0.85 22 2.63 — 0.167 18.7 1 kHz, 50 °C
Fe:KLTN [36]f 6.65 — 2.7 97 0.024 0.85 100 kHz
a
Polyvinylidence fluoride.
b
Sr0.5Ba0.5Nb2O6, Strontium barium niobate.
c
Triglycine sulphate.
d
Congruent lithium niobate.
e
2 mol% zirconia doped congruent lithium niobate.
f
Fe-doped niobium-rich potassium lithium tantalate niobate.

18, 26 and 4 times improvement, respectively. Similarly, with References

respect to specific detectivity, when compared with traditional
pyroelectric materials such as PVDF, SBN-50 and TGS, the [1] Wang Z, Zhang J, Xu S, Wang L, Cao Z, Zhan P and Wang Z
FD is enhanced 33, 4 and 5 times [11]; compared with CLN 2007 J. Phys. D: Appl. Phys. 40 4482–4
and 2 mol% Zr:LN, the FD of Mg:LT shows ten and twofold [2] Fairley P D and Rutt H N 2000 J. Phys. D: Appl. Phys.
improvement [22]; and even 34 times more than Fe:KLTN 33 2837–52
[3] Zhang W L, Yu Y C, Luo W B, Shuai Y, Pan X Q, Wu Q Q
[36]. These exciting improvements of FoMs prove the great
and Wu C G 2016 Infrared Phys. Technol. 80 100–4
potential of Mg:LT single crystals in pyroelectric transducers. [4] Balakt A M, Shaw C P and Zhang Q 2016 J. Mater. Sci.:
Mater. Electron. 27 12947–54
[5] Dias C, Simon M, Quad R and Das-Gupta D K 1993 J. Phys.
4. Conclusions D: Appl. Phys. 26 106–10
[6] Liu S T and Long D 1978 Proc. IEEE 66 14–26
In conclusion, our work assesses the outstanding pyroelectric [7] Nye J F 1985 Physical Properties of Crystals (Oxford:
figures of merit of Mg:LT. The general tendency of permit­ Clarendon Press)
tivity and pyroelectric coefficient is revealed by the location [8] Siegfried B and Bernd P 1990 Ferroelectrics 106 393–8
[9] Hartley N P, Squire P T and Putley E H 1972 J. Phys. E: J. Sci.
analysis of MgO. Specifically, these intrinsic electrical param­ Instrum. 5 787–9
eters decrease with MgO doping from 0 to 3 mol%, but reach [10] Chynoweth A G 1956 J. Appl. Phys. 27 78–84
the highest value of 4 mol%, and then decrease again in more [11] Whatmore R W 1986 Rep. Prog. Phys. 49 1335–86
heavily doped samples. The 5 mol% Mg:LT single crystal [12] Suyal G and Setter N 2004 J. Eur. Ceram. Soc. 24 247–51
provides the best FoMs (FV and FD), which are enhanced ten [13] Lang S B and Ringgaard E 2012 Appl. Phys. A 107 631–8
[14] Suyal G, Seifert A and Setter N 2002 Appl. Phys. Lett.
and twofold compared to undoped LT. Moreover, the positive 81 1059–61
effect of doping on Curie temperature also contributes to a [15] Sinha N, Goel N and Singh B K 2012 J. Solid State Chem.
broadening of the LT thermal application. 190 180–85
[16] Abdullah K A, Batal M A, Hamdan R, Toni K, Zaraket J,
Aillerie M and Salame C 2017 Energy Proc. 119 545–55
Acknowledgments [17] Wu C G, Sun X Y, Meng J, Luo W B, Li P, Peng Q X, Luo S Y
and Shuai Y 2014 Infrared Phys. Technol. 63 69–73
This work was supported by the National Natural Science [18] Chopra S, Sharma S, Goel T C, Chatterjee R and
Mendiratta R G 2006 Comput. Mater. Sci. 37 134–40
Foundation of China (NSFC) (Nos. 11372361, 11302268,
[19] Nakamura K and Itagaki M 1994 Jpn. J. Appl. Phys.
11232015, 11472321, 11572355), the National Natural Sci­ 33 5404–6
ence Foundation of Guangdong Province (2017A030310426, [20] Chandramouli K and Koduri R 2009 J. Mater. Sci. 44 1793–9
2018A030313321), the Guangdong Science and Technology [21] Mangalam R V, Agar J C, Damodaran A R, Karthik J and
Project (2015B090927005) and the Fundamental Research Martin L W 2013 ACS Appl. Mater. Interfaces 5 13235–41
[22] Zhu Y Z, Lin S P, Zheng Y, Ma D C and Wang B 2016
Funds for the Central Universities (17lgpy37).
J. Mater. Sci. 51 3155–61
[23] Shibata Y, Kuze N, Matsui M, Kanal M and Kawai T 1996
Mat. Res. Soc. Symp. Proc. 401 273–8
ORCID iDs [24] Kovacs L et al 1993 J. Phys.: Condens. Matter 5 781–94
[25] Kumaragurubaran S, Takekawa S, Nakamura M and
Yunzhong Zhu https://orcid.org/0000-0002-5229-822X Kitamura K 2006 J. Cryst. Growth 292 332–6

6
J. Phys. D: Appl. Phys. 51 (2018) 395101 F Tang et al

[26] Ehre D and Cohen H 2013 Appl. Phys.Lett. 103 052901 [32] Kandpal B and Upadhyay T C 2016 IOSR J. Appl. Phys.
[27] Iyi N, Kitamura K, Izumi F, Yamamoto J K, Hayashi T, Asano H 8 53–9
and Kimura S 1992 J. Solid State Chem. 101 340–52 [33] Watanabe Y, Sota T, Suzuki K, Kitamura K and Kimura S
[28] Kang X, Liang L, Song W, Wang F, Sang Y and Liu H 2016 1995 J. Phys.: Condens. Matter 7 3627–35
CrystEngComm 18 8136–46 [34] Zhen X H, Zhao L C and Xu Y H 2003 Appl. Phys. B
[29] Xue D and Kitamura K 2003 J. Cryst. Growth 249 507–13 76 655–59
[30] Li L, Zhu M, Zhou K, Wei Q, Zheng M and Hou Y 2017 J. [35] Dai L, Yan Z, Jiao S, Xu C and Xu Y 2015 RSC Adv.
Appl. Phys. 122 204104 5 36385–9
[31] Betancourt J, Saavedra-Arias J J, Burton J D, Ishikawa Y, [36] Li Y, Li J, Zhou Z, Guo R and Bhalla A S 2013 Europhys.
Tsymbal E Y and Velev J P 2013 Phys. Rev. B 88 085418 Lett. 104 57008