Content uploaded by Vittorio Degiorgio
Author content
All content in this area was uploaded by Vittorio Degiorgio
Content may be subject to copyright.
Structural and optical properties of zirconium doped lithium
niobate crystals
N. Argiolas,1M. Bazzan,1M. V. Ciampolillo,1P. Pozzobon,1C. Sada,1,a兲L. Saoner,1
A. M. Zaltron,1L. Bacci,1P. Minzioni,2G. Nava,2J. Parravicini,2W. Yan,2I. Cristiani,2and
V. Degiorgio
1Department of Physics and CNISM, University of Padova, Via Marzolo 8, 35131 Padova, Italy
2Department of Electronics and CNISM, University of Pavia, Via Ferrata 1, 27100 Pavia, Italy
共Received 18 May 2010; accepted 3 September 2010; published online 4 November 2010兲
Zirconium doped lithium niobate is a promising candidate as a substrate for nonlinear optical
applications, since it does not suffer from the so-called “optical damage.” In order to optimize this
aspect, the proper Zr concentration has be used, hence the precise determination of the so-called
“threshold concentration,” i.e., the concentration above which the photorefractive effect is markedly
reduced, is of great importance. In this work, we prepared by Czochralski growth a series of
Zr-doped lithium niobate crystals with various Zr content and studied them using structural
共high-resolution x-ray diffraction兲and optical 共birefringence兲measurements as a function of the
dopant content in the melt. Both the approaches pointed out a marked change in the crystal
characteristics for a Zr concentration between 1.5 and 2 mol %, a value which is identified as the
threshold concentration. © 2010 American Institute of Physics.关doi:10.1063/1.3499275兴
I. INTRODUCTION
Wavelength conversion using parametric processes is
crucial for a large range of photonic applications including
ultrafast all-optical processing and generation of new visible
frequencies. Recent years have seen much research focused
on the cascading of two second-order processes in periodi-
cally poled ferroelectric-crystal waveguides,1and the nonlin-
ear optical process of four-wave mixing in highly nonlinear
fibers2and silicon waveguides.3Lithium niobate 共LiNbO3,
LN in the following兲is a well-known ferroelectric material
widely exploited in nonlinear optics. In particular, thanks to
the periodic poling technique and via the quasiphase match-
ing method, LN allows the fulfilling of the phase matching
condition in a wide wavelength range. In particular, by com-
bining the realization of waveguides with the use of periodi-
cally poled lithium niobate 共PPLN兲substrates, interaction
lengths in excess of 8 cm have been demonstrated.4
The main problem of LN, actually limiting the PPLN
exploitation in commercial components and devices, is the
semipermanent change in the crystal refractive indices that a
light beam may induce on the material, a fact known as
photorefractive effect.5This effect 共sometimes called also
“optical damage”兲occurs because a high-intensity inhomo-
geneous illumination with visible wavelengths is able to in-
duce an electric space charge distribution that modifies the
material refractive index through the electro-optic effect.
This has detrimental effects, on one hand on the coherence of
the pump and second harmonic beams, and on the other on
the beam scattering during propagation. This problem is par-
ticularly severe in optical waveguides, because of the high
power density, and also because the technology for wave-
guide fabrication might increase the sensitivity of lithium
niobate to the photorefractive effect.
One method to overcome these limitations is to use con-
gruent LN 共cLN兲crystals doped with elements able to reduce
the photorefractive response of the material. Since a key role
in the photorefractive process is played by the presence, in
the cLN crystal, of Li-sites occupied by Nb ions, the aim of
doping is that of removing these native defects by incorpo-
rating the dopant ion at the Li-site in competition with NbLi.5
In this field it is customary to introduce the concept of a
threshold concentration, defined as the minimum doping
concentration required to strongly reduce photorefractivity.
The threshold concentration can be interpreted as the dopant
concentration required for removing all the NbLi sites. In this
scenario, the possibility of suppressing the optical damage
while preserving the optical transparency was first demon-
strated in Mg:cLN crystals,6for which further studies7indi-
cated a stability to laser intensities up to 100 MW/cm2. Sub-
sequent studies considered several other dopants to be
incorporated, each one characterized by a specific threshold
concentration. A simple charge compensation analysis shows
that, in the case of divalent ions, such as Mg or Zn, the
threshold concentration for cLN should be about 5.5 mol %.
Since the growth of Mg:cLN crystals of very good optical
quality is difficult, several groups have investigated the pos-
sibility of reducing the threshold concentration by doping
cLN with trivalent8or tetravalent ions.9In particular, for tet-
ravalent ions, such as Hf and Zr, the charge compensation
approach predicts a threshold concentration equal to half of
the Nb excess concentration, that is:
cHf
ⴱ=0.5共关Nb兴–关Li兴兲.10,11 Typically, the Nb excess is in the
range between 3 and 4 mol %, and thus cHf
ⴱis predicted to be,
in the crystal, in the range between 1.5 and 2 mol %. It
should also be noted that the precise value of the Nb excess
in the grown crystal might depend on the dopant concentra-
tion, as shown in Ref. 12 for the case of Mg:LN. An impor-
tant point to be considered is that the dopant concentration
a兲Electronic mail: cinzia.sada@unipd.it.
JOURNAL OF APPLIED PHYSICS 108, 093508 共2010兲
0021-8979/2010/108共9兲/093508/5/$30.00 © 2010 American Institute of Physics108, 093508-1
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
inside the crystal does not coincide in most cases with the
concentration in the melt, as the so-called segregation coef-
ficient is different from one for most dopant ions, including
Hf.10
The experiments performed with Hf:cLN crystals show a
rather complex picture. Some crystal properties, such as the
optical birefringence or the phase matching temperature for
second harmonic generation, when plotted as functions of the
doping concentration, show a “threshold” at HfO2concentra-
tions slightly above 2 mol %.13 Earlier measurements of
green-light induced birefringence suggested that cHf
ⴱwas
around 4 mol %,9but subsequent studies of the induced bi-
refringence versus the intensity Iof the light beam indicated
that the doping concentration at which the photorefractive
change saturates, instead of growing linearly with I,is
around 2.5 mol %.14
As an alternative to Hf, it was recently reported that the
incorporation of zirconium 共Zr兲in LiNbO3crystals increases
the optical damage resistance up to 20 MW/cm2, presenting
a low threshold concentration 共2 mol %兲, and a segregation
coefficient close to one.15 This latter aspect is important for
the crystal growth process, since it makes the preparation of
homogeneously doped crystals with high optical quality
easier with respect to LN crystals doped with Mg, Zn, Sc, In,
and Hf. Moreover Zr incorporation has the additional advan-
tage of lowering the coercive field 共one third of that of the
pure cLN兲, a very promising aspect for PPLN realization.
Contradictory data are reported in the literature,15,16 regard-
ing the identification of the Zr threshold concentration, thus
not allowing a clear understanding of the Zr incorporation
mechanism, and of its impact on the crystal properties.
Therefore, although the results of Ref. 15 clearly suggest that
Zr might represent an excellent alternative for obtaining a
LN substrate with higher optical damage resistance, a sys-
tematic study of the structural and optical properties is man-
datory.
In this work, we focus our attention on correlating the
structural and the optical properties of Zr-doped LN crystals
共Zr:LN in the following兲grown by the Czochralski technique
with a Zr content in the range 0–3 mol %. We present mea-
surements of the lattice parameters and of the crystal bire-
fringence as functions of zirconium concentration, cZr. Our
results indicate that the basal lattice parameter 共perpendicular
to the optical axis兲is linearly dependent on Zr concentration,
while the one parallel to the polar axis exhibits a minimum
around 1.8 mol %. We also find that the crystal birefringence
curve presents a kink around 2 mol %. A possible model
describing the Zr incorporation is presented and the results
are discussed in this framework.
II. MATERIALS AND METHODS
Zirconium doped lithium niobate 共Zr:LN兲crystals were
grown by the Czochralski technique at the University of Pa-
dova. The growth direction was along the z-axis with a pull-
ing rate equal to 3 mm/h and with a rotation rate close to 30
rpm. The ZrO2content is in the range 0–3 mol %. The grown
crystal boules were poled at high temperature in air atmo-
sphere in order to achieve a single domain state through the
whole sample volume. By exploiting the x-ray diffraction
technique, they were oriented and cut in slices with the major
surface perpendicular to the y-axis, with a tolerance better
than 0.4°. The crystals were successively polished to achieve
optical quality surfaces, by standard procedures using a Log-
itech PM5 lapping machine.
The lattice parameter measurements were performed us-
ing a Philips MRD high-resolution x-ray diffractometer. The
system was operated using an x-ray sealed tube with copper
anode. The primary beam was conditioned by a parabolic
multilayer mirror combined with a Bartels four-bounce
monochromator. The resulting beam was characterized by a
wavelength =0.154 056 nm and a spectral purity ⌬/of
about 10−5. The beam divergence was equal to 3.9
⫻10−3 degrees. The beam impinged on the sample which
was mounted on a goniometer system with high angular ac-
curacy on the scan direction 共better than 10−3 degrees on
absolute large-angle measurements兲. The sample was hosted
by an Eulerian cradle which could be used to set any desired
Bragg-planes family into the scattering condition. The dif-
fracted beam was collected through a Bartels three-bounce
collimator into a Xe proportional counter, which was
mounted on a second high accuracy goniometer, coaxial with
the one holding the sample. The temperature of the measure-
ment chamber was controlled and set to be equal to
25.0⫾0.1 °C for all the measurements.
The crystal birefringence, ⌬n=ne−no, was measured as
a function of Zr concentration by a method described in de-
tail in Refs. 13 and 17. The setup consisted of a simple
polarization interferometer: the crystal under study was illu-
minated by a collimated beam propagating perpendicularly
to the optical axis of the crystal. The beam, emitted by a
broadband source, is polarized at 45° with respect to crystal
axis, so that the two equal-amplitude ordinary and extraordi-
nary components accumulated different phase-delays during
propagation along the crystal. The output radiation went
through a second polarizer, also oriented at 45° with respect
to crystal axis, and the transmitted radiation was collected by
an optical fiber and brought to an optical spectrum analyzer
共OSA兲. As discussed in Ref. 13, the transfer function of the
setup is a periodic function of the wavelength, the maxima
corresponding to the spectral components that have accumu-
lated a phase difference multiple of 2
during crystal propa-
gation, whereas the minima refer to wavelengths for which
the phase difference is an odd multiple of
. By taking two
wavelengths Aand Bcorresponding to two consecutive
transmission maxima and calling Lthe crystal length, the
crystal birefringence can be calculated from the relation
⌬n=AB
L共A−B兲.共1兲
If the spectral width of the used radiation is sufficiently
large, it is possible to observe the presence of several peaks
on the received spectrum, and from their position it is pos-
sible to evaluate the crystal birefringence with good preci-
sion. In our experiments we used, as a broadband radiation
source, the amplified spontaneous emission emitted by an
Erbium-doped fiber amplifier, covering the wavelength range
between 1520 and 1570 nm, and we polarized the radiation
093508-2 Argiolas et al. J. Appl. Phys. 108, 093508 共2010兲
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
using two Glen–Thompson polarizers. The optical spectrum
has been recorded by an OSA with high sensitivity
共⫺70 dB m兲, and good spectral resolution 共0.01 nm兲.
III. RESULTS
In order to derive the values of the lattice parameters
from the high-resolution x-ray measurements, the Fewster
method18 was used. Since the LN lattice belongs to the trigo-
nal system and can be described in the hexagonal coordinates
system, it is fully characterized by the length of the basal and
uniaxial lattice parameters, indicated as aand c, respectively.
In order to determine these parameters with high accuracy,
we measured the interplanar spacing dh,k,lfor seven indepen-
dent Bragg reflections. The resulting values were corrected
for dynamical effects18 共which may lead to a systematic un-
derestimation of the interplanar lattice spacing for low angle
reflections兲and fitted using the following relation, valid for
the hexagonal system
dh,k,l=
冉
4
3
h2+hk +k2
a2+l2
c2
冊
−1/2
.共2兲
This procedure gives the lattice parameters aand cfor each
sample considered in this work. The experimental uncer-
tainty is very low, due to the high resolution of the measure-
ments, and is estimated using multivariate regression analy-
sis and standard error propagation formulae. Moreover, as a
check of the sample quality, we measured for several reflec-
tions the width of the Bragg peaks and compared them with
the theoretical value expected for a perfect crystal, calculated
according to the dynamical theory of x-ray scattering18 and
corrected for instrumental broadening.
All the samples showed intense Bragg peaks near the
expected angles. The measured width of the Bragg peaks is
in agreement with the value predicted for the perfect crystal,
confirming the good structural quality of the samples. In par-
ticular the dopant incorporation did not introduce any mea-
surable compositional stress, as expected from the good op-
tical quality of the grown samples. The measured lattice
parameters are reported, as functions of the Zr content in the
melt, in Figs. 1共a兲and 1共b兲, respectively.
Due to the high angular accuracy of the instrument used
and the control of the temperature in the measure chamber
these measurements are characterized by a high accuracy,18
with typical experimental uncertainties of the order of
10−4 Å on the absolute determination of both a and c. The
main problem in data analysis can come from the nominal
concentration of the samples, which is not known directly
and may in principle be different with respect to the one of
the melt. However, according to Ref. 15 this should not be
the case, owing to the segregation coefficient of Zr in lithium
niobate which is close to one.
The aparameter exhibits an almost linear increase with
the dopant concentration, which may be fitted to a straight
line, a共cZr兲=a0+kcZr, where cZr is the Zr content in the melt
expressed in mol %, and ais the length of the basal lattice
parameter expressed in Å, giving the following result:
a0=共5.151 81 ⫾0.000 08兲Å,
k=共0.002 05 ⫾0.000 03兲Å.
The value found for the intercept a0is compatible with the
published lattice parameter of congruent lithium niobate
关a=共5.148⫾0.002兲Å兴. The adjusted r-square value of the
fit is 0.989, supporting the claim that the segregation coeffi-
cient is close to one. In fact if this was not the case, the Zr
content in the sample would depend on the portion of the
boule from which it was extracted, and a significant scatter-
ing in the experimental points should appear. The linear re-
lation can therefore be used to estimate the Zr content in a
crystal of unknown composition.
The cparameter, on the other hand, exhibits a com-
pletely different behavior, showing a nonmonotonous trend
with a minimum in the range between 1.5 and 2 mol %. As
discussed above, we can rule out the possibility of a mis-
taken sample composition, very different from the one of the
melt It is worth noting that a similar behavior of the two
lattice parameters was reported also for LN crystals doped
with Zinc,19 another photorefractive resistant ion. In that case
the minimum was found at a ZnO concentration of about 2.5
mol %.
As far as the optical properties are concerned, the mea-
sured birefringence is plotted in Fig. 2. The error bars 共⬍2
⫻10−4兲reported in Fig. 2were derived considering the larg-
est difference encountered between two measurements on the
same sample, considering different measurement positions.
The obtained results indicate a very good uniformity of the
FIG. 1. Measured values for lattice parameters a共panel a兲and c共panel b兲as a function of the Zr content in the melt. Note that for some concentrations error
bars have nearly the same width of the graphic symbol of the experimental point.
093508-3 Argiolas et al. J. Appl. Phys. 108, 093508 共2010兲
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
samples and a high reproducibility of the data. It can be seen
that the crystal birefringence is almost constant up to a 2
mol % dopant concentration.
However, this trend drastically changes when a larger
concentration is used, and an almost linear increase in bire-
fringence is observed. It should be noted that Eq. 共1兲is exact
only if ⌬nis independent of in the considered wavelength
range. The -dependence of ⌬nwould introduce an offset
between the obtained value and the real birefringence; with-
out yielding to any significant modification in the curve
showing the fit.
As discussed in Ref. 20 for the case of Mg:cLN, the
ordinary refractive index nois expected to be only slightly
dependent on the dopant concentration, so that the birefrin-
gence change should be essentially due to the variation in the
extraordinary index ne.
IV. DISCUSSION
From a structural point of view, we may tentatively ex-
plain the effect of Zr doping for concentrations below and
above 1.5 mol % with the disappearance of the NbLi antisite
defects. In fact it is known that the cparameter tends to
decrease when the material composition moves toward sto-
ichiometry, where the NbLi concentration is in principle
zero.21 It is generally accepted that photorefractive resistant
ions are able to reduce the NbLi concentration5by occupying
the Li-site in competition with Nb atoms. Above a certain
threshold concentration, the occurrence of NbLi defects be-
comes very unlikely, and the way the dopant atoms are in-
corporated in LN lattice changes, giving rise to a different
kind of lattice deformation.5,19,22
In the scientific literature slightly different theoretical
values of the threshold concentration are proposed, depend-
ing on the method employed for its assessment. In the case
of tetravalent dopants the charge compensation method pre-
dicts a value of the threshold concentration equal to half of
the Nb excess concentration, that is: cZr
ⴱ=0.5共关Nb兴–关Li兴兲.10
Typically, the Nb excess is in the range between 3 and 4
mol %, and thus cZr
ⴱis predicted to be, in the crystal, in the
range between 1.5 and 2 mol %. Anyway, it should also be
taken into account that the precise value of the Nb excess
depends also on the dopant concentration, as shown in Ref.
20 for the case of Mg:LN. A different method, based on a
chemical bond analysis for the evaluation of the global in-
stability index,11,23 sets the threshold value for Zr in the crys-
tal to 1.7 mol %. We can conclude that the measured value of
cZr
ⴱ, which is very close to the value reported formerly in Ref.
15, is consistent with the theoretical predictions.
When the concentration is increased above cZr
ⴱ, there
may be two possibilities for charge compensation: the forma-
tion of niobium vacancies or the incorporation of Zr at the
Niobium site. Since in general the formation of vacancies is
accompanied by a diminution of the unit cell volume, our
data suggest that above the threshold, Zr ions begin to be
incorporated at the Nb site, similarly to the situation depicted
for Mg doping5,21 and Zn doping.24 On the other hand, opti-
cal measurements show a change in birefringence at a dopant
concentration slightly above the one pointed out by structural
measurement. Also in this case, we can ascribe this specific
behavior to the fact that the dopant ions incorporated during
crystal growth will lie in distinct positions of the lattice de-
pending whether its concentration is below or above thresh-
old, dramatically influencing the dielectric response of the
material. This suggests that structural measurements are able
to probe the appearance of ZrNb defects before they induce a
significant birefringence change, a fact that, if confirmed,
indicates x-rays diffraction as a powerful technique to probe
the presence of this change in site occupation. Work is in
progress to clarify this aspect by exploiting experimental
techniques able to directly investigate the Zr site location
such as the proton induced x-ray emission or x-ray standing
waves techniques.
V. CONCLUSIONS
We have presented an optical and structural characteriza-
tion of Zr:LN crystals showing that the threshold concentra-
tion cZr
ⴱfor the complete removal of NbLi sites is less than 2
mol %, i.e., less then half with respect to cMg
ⴱ. This fact, in
combination with a close to one segregation coefficient al-
lowed for the growth of high optical quality crystals. As it is
known, the cancellation of NbLi defects is a key point in the
realization of photorefractive resistant crystals so that we
may expect that the studied samples exhibit good resistance
against optical damage: photorefractive characterization is
currently being performed and will be object of a forthcom-
ing paper. The overall picture is that Zr:LN crystals are very
promising candidates for the realization of efficient cascaded
wavelength converters working at room temperature.
ACKNOWLEDGMENTS
This work has been supported by the Fondazione Ca.Ri-
.Pa.Ro 共Fondazione Cassa di Risparmio di Padova e Rovigo,
Italia兲by financing the Excellence Project 2008-2009 “Inte-
grated visible frequency converter based on doped periodi-
cally poled lithium niobate crystals with enhanced optical
damage resistance,” and by Fondazione CARIPLO Rif.
2007.5193
1C. Langrock, S. Kumar, J. E. McGeehan, A. E. Willner, and M. M. Fejer,
J. Lightwave Technol. 24, 2579 共2006兲.
2M. N. A. S. Bhuiyan, M. Matsuura, H. N. Tan, and N. Kishi, Opt. Express
18, 2467 共2010兲.
FIG. 2. 共Color online兲Measured birefringence: the dashed line is only a
guide for the eye and does not correspond to a specific concentration depen-
dence of birefringence.
093508-4 Argiolas et al. J. Appl. Phys. 108, 093508 共2010兲
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
3A. C. Turner-Foster, M. A. Foster, R. Salem, A. L. Gaeta, and M. Lipson,
Opt. Express 18,1904共2010兲.
4D. Caccioli, A. Paoletti, A. Schiffini, A. Galtarossa, P. Griggio, G. Loren-
zetto, P. Minzioni, S. Cascelli, M. Guglielmucci, L. Lattanzi, F. Matera, G.
M. T. Beleffi, V. Quiring, W. Sohler, H. Suche, S. Vehovc, and M. Vidmar,
IEEE J. Sel. Top. Quantum Electron. 10,356共2004兲.
5T. Volk and M. Wöhlecke, Lithium Niobate: Defects, Photorefraction and
Ferroelectric Switching, Springer Series in Material Science 115
共Springer-Verlag, Berlin, 2008兲.
6G. G. Zhong, J. Jin, and Z. K. Wu, J. Opt. Soc. Am. 70, 631 共1980兲.
7D. A. Bryan, R. R. Rice, R. Gerson, H. E. Tomaschke, K. L. Sweeney, and
L. E. Halliburton, Opt. Eng. 24,138共1985兲.
8J. K. Yamamoto, K. Kitamura, N. Iyi, S. Kimura, Y. Furukawa, and M.
Sato, Appl. Phys. Lett. 61, 2156 共1992兲.
9L. Razzari, P. Minzioni, I. Cristiani, V. Degiorgio, and E. P. Kokanyan,
Appl. Phys. Lett. 86, 131914 共2005兲.
10S. Li, S. Liu, Y. Kong, D. Deng, G. Gao, Y. Li, H. Gao, L. Zhang, Z.
Hang, S. Chen, and J. Xu, J. Phys.: Condens. Matter 18,3527共2006兲.
11X. He and D. Xue, Opt. Commun. 265, 537 共2006兲.
12N. Iyi, K. Kitamura, Y. Yajima, and S. Kimura, J. Solid State Chem. 118 ,
148 共1995兲.
13P. Minzioni, I. Cristiani, J. Yu, J. Parravicini, E. P. Kokanyan, and V.
Degiorgio, Opt. Express 15, 14171 共2007兲.
14P. Minzioni, I. Cristiani, V. Degiorgio, and E. P. Kokanyan, J. Appl. Phys.
101, 116105 共2007兲.
15Y. Kong, S. Liu, Y. Zhao, H. Liu, S. Chen, and J. Xu, Appl. Phys. Lett. 91,
081908 共2007兲.
16L. Sun, F. Guo, Q. Lv, H. Yu, H. Li, W. Cai, Y. Xu, and L. Zhao, Cryst.
Res. Technol. 42, 1117 共2007兲.
17F. Bréhat, B. Wyncke, and A. M. Benoit, J. Phys. D: Appl. Phys. 32,227
共1999兲.
18P. F. Fewster, X-Rays Scattering from Semiconductors, 2nd ed. 共Imperial
College Press, London, 2003兲.
19F. Abdi, M. Aillerie, M. D. Fontana, P. Bourson, T. Volk, B. Maximov, S.
Sulyanov, N. Rubinina, and M. Wöhlecke, Appl. Phys. B: Lasers Opt. 68,
795 共1999兲.
20U. Schlarb and K. Betzler, Phys. Rev. B 50,751共1994兲.
21G. Malovichko, O. Cerclier, J. Estienne, V. Grachev, E. Kokanyan, and C.
Boulesteix, J. Phys. Chem. Solids 56, 1285 共1995兲.
22Á. Peter, K. Polgár, L. Kovács, and K. Lengyel, J. Cryst. Growth 284,149
共2005兲.
23D. Xue and S. Zhang, Physica B 262,78共1999兲.
24S. Sulyanov, B. Maximov, T. Volk, H. Boysen, J. Schneider, N. Rubinina,
and T. Hansen, Appl. Phys. A: Mater. Sci. Process. 74, s1031 共2002兲.
093508-5 Argiolas et al. J. Appl. Phys. 108, 093508 共2010兲
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp