Content uploaded by Qian Sun
Author content
All content in this area was uploaded by Qian Sun on Mar 29, 2014
Content may be subject to copyright.
Ultraviolet photorefractivity features in doped lithium niobate crystals
Haijun Qiao, Jingjun Xu,*Guoquan Zhang, Xinzheng Zhang, Qian Sun, and Guangyin Zhang
Photonics Center, College of Physics Science, Nankai University, Tianjin 300071, People’s Republic of China
and TEDA Applied Physics School, Nankai University, Tianjin 300457, People’s Republic of China
(Received 15 December 2003; revised manuscript received 22 April 2004; published 2 September 2004)
In this work we studied photorefractive effect of lithium niobate 共LiNbO3兲doped with Zn, In, and Na at
ultraviolet 共UV兲wavelength down to 351 nm. It is found that the UV photorefraction of LiNbO3doped with
Zn, In, or Na was enhanced significantly as compared to that of the nominally pure LiNbO3. Our results show
that the statement that the property of resistance against photorefraction in highly Zn and In doped LiNbO3is
correct only in visible light range. By contrary, these crystals exhibit as excellent photorefractive characteristics
in UV. We also find that there are doping concentration threshold values of Zn and In for the disappearance of
the light-induced lens-like effect both in visible and in UV, but such concentration threshold values are not
found for UV photorefraction within the highest doping concentrations we used. In highly Zn or In doped
LiNbO3crystals, diffusion dominates over photovoltaic effect and electrons are determined to be the dominant
charge carriers in UV photorefraction. The results are of interest to the study on the defect structure of LiNbO3.
Further investigation on this field is greatly urgent.
DOI: 10.1103/PhysRevB.70.094101 PACS number(s): 77.84.Dy, 42.70.Nq, 42.70.Ln
I. INTRODUCTION
Lithium niobate 共LiNbO3兲crystals remain as the promis-
ing versatile materials in optic, electro-optic and photonics
science and technology. They are widely used in optical dou-
bler and optical parametric oscillators, holographic data stor-
age, and even for acoustical memory as recently reported.1
Since researchers in Bell Laboratories observed “an
optically-induced inhomogeneity in the refractive index of
LiNbO3and other ferroelectrics”2in 1966, a lot of investiga-
tions have been done on this material. The topics ranged
from macro- to micro-mechanism, and from basic defect
structures to various applications.
The optically-induced damage and photorefractive effect
are two notable and important characters of LiNbO3. Both of
them mean that in this material the photo-induced change in
the refractive index is reversible even under low light inten-
sity. In application, lithium niobate crystals with enhanced
photorefraction or with optical-damage-resistance need dif-
ferent requirements, respectively. For example, lithium nio-
bate crystals with high optical-damage-resistance are adopted
for frequency doubling, especially for the promising PPLN
(periodically poled lithium niobate)used in quantum
information3and second-harmonic generation processing;4
on the other hand, crystals with high photorefractive sensi-
tivity and low light-induced noise scattering are needed for
holographic data storage. As is well known, photorefraction
of LiNbO3could be modified by doping the crystals with
appropriate impurities. It is believed that, as far as the effect
on photorefraction of LiNbO3is considered, dopants could
be divided into two groups: Dopants such as Fe, Co, Mn and
Cu tend to enhance the photorefraction, whereas those such
as Mg, Zn, and In tend to reduce the photorefraction, i.e., the
crystal shows optical-damage-resistance and photorefractive
resistance. Investigations have been done in this field for
several years, and presently it is commonly believed that the
strength of the photorefractive effect of LiNbO3is deter-
mined by the corresponding intrinsic or external defects of
the crystals. Although well-known conclusions have been
made and several theories are brought forward to support
them,5–7 many aspects still remain unclear so far, especially
the defect centers responding for the ultraviolet 共UV兲photo-
refraction in LiNbO3. The incomplete knowledge prevents
one from efficiently utilizing the material.
The UV photorefraction of nominally pure LiNbO3was
reported in 1992.8Later Laeri et al. made an review on UV
photorefraction in various ferroelectrics in 1995.9They re-
ported that the photorefraction of pure LiNbO3due to the
diffusion mechanism is enhanced in UV as compared to that
in visible, and the dominant charge carriers in UV are holes.
In 2000 Xu et al. reported an enhancement of UV photore-
fraction in highly magnesium doped LiNbO310 which, how-
ever, shows a perfect optical-damage-resistance in the visible
region. These phenomena were not fully understood up to
now. Since the UV photons can excite electrons or holes
from deeper levels rather than photons of visible region, in
this sense the UV photorefraction of LiNbO3provides an-
other window to look into the crystalline defect structure.
Moreover, this study would be helpful for manufacturing
good-property materials in application of the UV photore-
fractivity. In this paper, we carried out a comprehensive
study on the UV photorefraction of doped LiNbO3. Such a
study would be helpful to identify and to understand the UV
photorefractively sensitive defect centers in LiNbO3.
II. EXPERIMENTS DETAILS
Congruent pure LiNbO3crystals doped with several dif-
ferent concentrations of In and Na were grown by Czochral-
ski technique from the congruent melt. After being poled the
crystals were cut into Y-sheets whose thickness is 3.5 mm
and both XY-surfaces were polished. The Zn-doped LiNbO3
crystals were provided by Russian Academy of Science and
similar aftergrowth treatments were introduced to the
PHYSICAL REVIEW B 70, 094101 (2004)
1098-0121/2004/70(9)/094101(11)/$22.50 ©2004 The American Physical Society70 094101-1
samples. AY-sheet of congruent pure undoped LiNbO3crys-
tal was prepared for comparison. In Table I we listed the
samples used in our experiments and abbreviation of each
sample in the following statement is given as well. All the
doping concentrations listed in Table I were the concentra-
tions in the melt. The +c-axis direction was determined ac-
cording to the pyroelectric effect. Because the energy trans-
ferring direction was often taken as an indication of the type
of the dominant charge carriers type,11 the determination of
the +cdirection is of obvious importance. Firstly, we put two
sheets of c-cut LiNbO3crystals, of which one is congruent
pure and the other is 6.5 mol % Mg doped LiNbO3, into the
etching acid for 12 hours. It is well-known that the −cend is
easier etched than +cend, so the −cend becomes more
frosted after etching.12 Thus we can determine the −cdirec-
tion of these two “standard” pieces of crystal inerrably. Sec-
ondly, we heated the two crystals and then cooled them down
by putting them into charged printing ink. After photographic
fixing, we found that the determined −cend was not con-
taminated by ink, while the other end (+cend)was covered
by an ink film due to the pyroelectric effect and charge neu-
tralization effect. According to the above results, we checked
all samples, and the end covered with ink was determined to
be +cend of the crystal.
Among our samples, In:LiNbO3was predicted to be
optical-damage-resistant crystals13 and was confirmed in
1995,14 and the concentration threshold was suggested to be
about 5 mol % in the melt by the authors of Ref. 15. The
concentration thresholds for Zn and Mg were reported at
about 7 to 8 mol %16 and about 5 mol % for Mg.17,18 These
three kinds of optical-damage-resistant impurities are be-
lieved to affect the photorefractive properties of LiNbO3al-
most in the same way: They tend to enhance the photocon-
ductivity greatly, to induce a blue-shift the absorption edge,
and to enhance the resistance against optical-damage. As for
the Na-doped LiNbO3, few publications19,20 can be found. It
is probable that the LiNbO3:Na crystals are difficult to grow
due to the big difference in the radii of Li+共0.65 Å兲and Na+
共0.91 Å兲. Kong et al.19 reported that Na:LiNbO3crystal also
illustrates a decrease in photorefraction at 488 nm comparing
with the congruent one. However, it was also reported that an
increase in Na doping concentration was not accompanied by
a decrease instead of an increase in optical-damage-
resistance. Concerning the particularity of Na, we also per-
formed some experiments with these Na-doped samples. Al-
though Na seems not an effective optical-damage-resistance
impurity in LiNbO3, it may be interesting for investigating
defect structures in LiNbO3. The infrared absorption (char-
acterizing the OH−vibration)spectra of the samples listed in
Table I were measured. Among our samples, it is found that
the OH−absorption peak in infrared region moves from 3487
to 3508 cm−1 in sample CIn5, and moves from 3483 to
3530 cm−1 in both CZn7 and CZn9. This indicated that the
doping concentrations of samples CIn5, CZn7, and CZn9 are
above the threshold, while those in other samples are under
the threshold.
We used two-wave coupling scheme to study the UV pho-
torefraction of our samples. The whole experiment configu-
ration is shown in Fig. 1. Ar+laser operating at 351 nm was
selected as the working beam. We split the output laser beam
into three beams, and then intersected two of them with ex-
traordinary polarizations inside the samples and wrote a pho-
torefractive grating. We always kept the photorefractive grat-
ing vector K parallel to the c-axis. The third beam was
expanded and served as the uniform erasing beam incident at
off-Bragg angle when necessary. The formation and decay of
the gratings were monitored by a weak He-Ne laser beam
(wavelength at 632.8 nm)incident at the Bragg angle. Dur-
ing the measurement, we let the two writing beams and the
weak probe beam incident on the crystal at the same time. To
avoid a disturbance of the probe beam, we kept its intensity
as weak as possible, only about 1.0–1.4 mW/cm2. By this
means, we could observe the formation and decay of the
gratings by recording the diffracted beam intensity of the
probe beam, e.g., by the detector D3 shown in Fig. 1. The
diffraction efficiency
of the light-induced grating was mea-
sured by simply blocking one of the writing beams after
saturation. Here
is defined as
=共Id+Ii兲/Ii, where Idis the
diffracted beam intensity, and Iiis the transmitted beam in-
tensity. We measured the intensities of transmitted and dif-
fracted beams just after blocking one writing beam, in order
TABLE I. List of the lithium niobate crystals and their dopants
used in our experiment.
Abbreviation Dopant Concentration 共mol%兲
Dimensions
共x⫻y⫻c,mm兲
CLN None 0 10⫻3⫻10
CZn5a5.4 4⫻2⫻5
CZn7aZn 7.2 4⫻2⫻5
CZn9a9.0 7⫻2⫻4
CIn1 1.0 7⫻3.5⫻7
CIn3 In 3.0 5⫻3.5⫻5
CIn5 5.0 7⫻3.5⫻7
CNa1 1.0 7⫻3.5⫻6
CNa3 Na 3.0 7⫻3.5⫻6
aThe three crystals were grown in Institute of Crystallography of the
Russian Academy of Sciences.
FIG. 1. The experiment configuration for the UV photorefrac-
tion measurement. M: mirror, BS: beam splitter, BE: beam ex-
pander, C: crystal, D1–D4: detectors.
QIAO et al. PHYSICAL REVIEW B 70, 094101 (2004)
094101-2
to prevent the self-enhancement or self-depletion21,22 from
bringing up inaccuracy to the results. Considering the energy
transferring between the two writing beams, we did not
record the gratings with equal intensity writing beams, but
used a configuration with unequal intensities according to
Ref. 23. According to Kogelnik’s coupled wave theory,24 we
could calculate the amplitude of refractive index change ⌬n
from
=sin2关
⌬nd/共cos
兲兴, where
is the half intersec-
tion angle of the two writing beams. In the two-wave cou-
pling experiment, we turned off the probe and erasing beams
and only let the two writing beams illuminate the crystals at
the same time, and modulated the intensity ratio of the two
beams around 100:1. The two-wave coupling gain was ob-
tained through relation ⌫=共1/d兲ln关共IS
⬘IR兲/共ISIR
⬘兲兴, where the
subscripts Sand Rdenote signal and reference beams (here
the signal beam is the one with a low intensity in the follow-
ing description),IS
⬘,IR
⬘and IS,IRare the transmitted intensi-
ties of the two writing beams with and without coupling,
respectively, dis the thickness of the grating, e.g., the thick-
ness of the sample in our case. The photoconductivity
p共I兲
was estimated by fitting the intensity dependence of erasure
time constant
efor grating in terms of the relation
p
=0/
e, where 0is the vacuum dielectric constant and is
the relative dielectric constant. Here the erasure time con-
stant
eis defined as the time when the diffraction efficiency
decays to 1/eof its initial value. We also estimated the dark
conductivity
dby the relation
d=0/
d, where
dis the
decay time constant of the grating in darkness. The photore-
fractive sensitivity Sis an important parameter for the pho-
torefractive material and describes how much energy is
needed to produce a given refractive index change. Sis de-
fined as S=共1/Id兲共d冑
/dt兲兩t=0, where Iis the total recording
intensity. The dynamic range M/# is a parameter describing
the storage capacity of the photorefractive gratings, in other
words, it stands for how many gratings can be recorded in-
side a unit volume. M/# here is defined as M/#
=
e共d冑
/dt兲兩t=0. The term d冑
/dt兩t=0 in the two equations
above means the gradient of square root of diffraction effi-
ciency at the very beginning of the recording.
III. PHOTOREFRACTION OF DOPED LiNbO3
IN ULTRAVIOLET
From the overall experimental results in UV, the samples
exhibited UV photorefractive properties which are different
from those in visible. The UV photorefractive characteristics
of our samples are listed in Table II. The photorefractive
effect in these highly Zn and In doped crystals was enhanced
significantly in UV as compared to that of the pure ones,
whereas different influences were also observed for different
dopants in LiNbO3. In this section, we emphasize the de-
scription of the comparison of the photorefractive properties
in UV with those in visible and the comparison of photore-
fractive properties in doped crystals with those in pure one.
A. Zn doped congruent LiNbO3
Zinc is +2 valence dopant in LiNbO3. The photorefractive
effect of Zn:LiNbO3in visible is pretty weak, especially
when the doping concentration is over the threshold as in
CZn7 and CZn9. However, the Zn:LiNbO3crystals show
very strong UV photorefractive effect and are excellent ma-
terials for UV photorefraction.
From Table II, we note a drastic increase of the photocon-
ductivity in Zn:LiNbO3, for example, it is as high as 57.3
⫻10−12 cm/⍀·W in the sample CZn9. In the sample CZn9,
the response time is as short as 0.88 s at the total recording
intensity of 70.8 mW/cm2. With increase of Zn concentra-
tion, other photorefractive properties such as diffraction effi-
ciency, coupling gain, and recording sensitivity are also
greatly enhanced. The recording and optical erasing cycle is
shown in Fig. 2(b). A long time evolution of the recording
process is also given in Fig. 3(b), in which no remarkable
disturbance is found. From Fig. 2(b)we can see that the
gratings can be recorded and erased in Zn:LiNbO3in a very
short interval. The photorefractive sensitivities of
Zn:LiNbO3samples were measured and listed in Table II.
Comparing with that in CLN, the sensitivity is much im-
proved by the Zn doping. The sample CZn9 exceeds all other
samples and shows the largest UV photorefractivity, see
Table II. The dark decay processes of gratings are shown in
Fig. 4(b). It is obvious that, besides the photoconductivity
and diffraction efficiency, the dark conductivity also in-
creases with the increase of Zn concentration in the crystal.
Figure 5(a)shows the evolution of the transmitted signal
beams in Zn:LiNbO3samples during light amplification ex-
periments with an external crossing angle 2
=40°. A cou-
pling gain coefficient ⌫up to 21.7 cm−1 was observed in
CZn9. To our knowledge, this is the largest gain coefficient
TABLE II. The measured photorefractive characteristics of our samples.
Samples CLN CZn5 CZn7 CZn9 CIn1 CIn3 CIn5 CNa1 CNa3
Photoconductivity
ph 共⫻10−12 cm/ ⍀·W兲3.32 10.6 25.2 57.3 1.59 7.46 12.9 2.01 0.96
Diffraction efficiencya
共%兲9.05 16.9 22.3 25.3 10.1 15.9 17.7 7.5 18.1
Photorefractive response timeb
e共s兲12.4 1.97 1.01 0.88 13.9 3.06 1.68 11.3 22.2
Two-wave coupling gainc⌫共cm−1兲1.32 11.0 15.2 21.7 1.16 11.8 17.0 1.54 1.95
Photorefractive sensitivityaS共cm/J兲0.99 4.0 8.85 11.1 0.86 2.85 3.88 0.56 0.38
Dynamic rangeaM/# 0.14 0.11 0.12 0.14 0.26 0.19 0.15 0.11 0.17
aThe intersect angle in the air 2
=40°, corresponding to photorefractive grating period ⌳=0.5
m, two
recording intensity IS=121.7 mW/ cm2and IR=176.9 mW/cm2, respectively.
bThe response time constants were measured under the uniform UV illumination of 70.8 mW/cm2.
cThe intensity ratio between the reference and signal beam IR:ISwas 100:1.
ULTRAVIOLET PHOTOREFRACTIVITY FEATURES IN…PHYSICAL REVIEW B 70, 094101 (2004)
094101-3
in UV reported in LiNbO3crystals. Note that we could only
observe slight light amplification in CLN under the same
configuration. We found the light energy transfer unidirec-
tionally toward −c-axis in all Zn:LiNbO3samples in our
experiment. This result was reproducible in all the
Zn:LiNbO3samples. This indicates that diffusion is the
dominant mechanism and electrons are the dominant charge
carriers during the UV photorefractive processes in
Zn:LiNbO3. We noted that this conclusion is inconsistent
with the results reported earlier,25,26 where the holes were
supposed to be the dominant charge carriers in Zn:LiNbO3
with Zn concentration higher than 7.5 mol % under the illu-
mination of a 488 nm laser beam. The dependence of the
two-wave coupling gain coefficient ⌫on the grating period
⌳共⌳=/2 sin
兲was also measured and shown in Fig. 5(b).
As known, the relation between the two-wave coupling gain
coefficient and the beam-crossing angle is described by ⌫
=Asin
cos 2
in/共1+B−2sin2
兲cos
in,27 where 2
in and 2
are the internal and external beam-crossing angle, respec-
tively, A=
␥
eff
共8
2n3kBT/e2兲,B=共e/4
兲共Neff/0kBT兲1/2,
is electron-hole competition factor, Neff is the effective
charge density. By fitting the measured ⌳-dependence of ⌫
using the function given above, we obtain the effective trap
center concentrations Neff. The fitting curves are also shown
in Fig. 5(b). One should notice that high-order diffracted
beams appeared due to phase-mismatching with grating
spacing ⌳greater than 1.2
m. So those data were not taken
in the fitting. The effective trap density Neff is fitted to be
0.51⫻1016 cm−3 for the sample CZn5, 1.07⫻1016 cm−3 for
CZn7, and 2.81⫻1016 cm−3 for CZn9. It is clear that intro-
FIG. 2. Holographic recording
and optical erasure cycle with the
same intensity in our samples. The
intensities of two recording beams
are 121.7 and 176.9 mW/cm2
for (a)CLN, (b)Zn:LiNbO3,
and (c)In:LiNbO3, 120.2
and 152.5 mW/cm2for (d)
Na:LiNbO3, respectively, and the
external crossing angle 2
=40°.
The small arrows indicate the time
of decay beginning.
FIG. 3. Long time recording of
holograms recorded in our
samples, the external crossing
angle 2
=40°.
QIAO et al. PHYSICAL REVIEW B 70, 094101 (2004)
094101-4
ducing Zn into LiNbO3brings a noticeable increase in the
density of effective trap centers for photorefraction in UV.
We could also see from Table II that the sensitivity is
greatly improved by Zn doping. The sensitivity as high as
11.1 cm/J was achieved in the sample CZn9. The dynamic
range of Zn:LiNbO3, however, is not large because of the
short erasing time due to a very high photoconductivity.
Highly Zn doped LiNbO3crystals are suitable for the dy-
namic and real-time holographic application in UV.
In brief, the UV photorefractive effect of Zn:LiNbO3is
strong and totally different from its visible behaviors.
Zn:LiNbO3crystals show high photorefractive sensitivities
and are good photorefractive materials in UV. They also in-
dicate that there are some unknown defect structures in
LiNbO3crystals responsible for UV photorefraction.
B. In doped congruent LiNbO3
Indium is +3 valence dopant and is considered as the most
efficient dopant to suppress the photorefraction of LiNbO3in
visible due to the lowest threshold concentration. From the
experimental results, In:LiNbO3crystals also show an en-
hanced UV photorefraction, although not so high as that in
Zn:LiNbO3.
In general, the same measurements as for Zn:LiNbO3
were done to the In:LiNbO3samples. The results are also
shown in Table II. It is seen that the photoconductivity, the
diffraction efficiency, the two-wave coupling gain coeffi-
cient, and the photorefractive sensitivity also increase with
the increase of In concentration. As mentioned above, the In
doping concentration of sample CIn5 is above the threshold,
and those of CIn1 and CIn3 are below the threshold. We
FIG. 4. Dark decay of holo-
grams recorded in our samples,
only the weak He-Ne probing
beam is open.
FIG. 5. The light amplification
in our samples: (a),(c)Time de-
pendence of the amplified signal
beams ISin Zn:LiNbO3and
In:LiNbO3at incident angle 2
=40°, the intensity ratio between
the reference beam and signal
beam is 100:1; (b),(d)dependence
of two-wave coupling gain ⌫on
grating period ⌳in Zn:LiNbO3
and In:LiNbO3, and the solid
curves are the theoretical fitting
results.
ULTRAVIOLET PHOTOREFRACTIVITY FEATURES IN…PHYSICAL REVIEW B 70, 094101 (2004)
094101-5
notice that the photoconductivity and photorefractive sensi-
tivity of CIn5 are lower than those of CZn7 and CZn9, only
close to CZn5 whose Zn doping concentration is below the
threshold. Figure 2(c)shows the recording and erasing cycle
of the recorded grating and the long time recording process
is given in Fig. 3(c). As to the little perturbance of
during
the long time recording in all the samples, we considered that
it was perhaps caused by the disturbance of the environment,
or some competing mechanisms may exist. We also show in
Fig. 4(c)the dark decay process of the recorded grating.
Comparing Fig. 4(c)with Fig. 4(b), it is seen that photore-
fractive gratings in In:LiNbO3can be preserved much longer
than those in Zn:LiNbO3, indicating that the dark conductiv-
ity in In:LiNbO3is much lower than that in Zn:LiNbO3.We
notice that the dark decay time constant of grating in CLN is
even shorter than that in CZn1 but is approximate to that in
CZn3.
The In concentrations in sample CIn1 and sample CZn3
are below the threshold. We could not observe light amplifi-
cation in CIn1, whereas a ⌫=11.0 cm−1 was found in CZn3,
as shown in Fig. 5(c)which shows the time evolution of the
transmitted signal beam in the light amplification experi-
ment. Note that a large gain coefficient is also observed in
the sample CIn5. It seems that the concept of concentration
threshold for optical-damage-resistance in visible is no
longer valid in UV. During the two-wave coupling experi-
ments, energy always transferred towards −cdirection in
samples CIn3 and CIn5, which indicates that electrons are
dominant charge carriers in these crystals. The
⌳-dependence of ⌫of CIn3 and CIn5 are shown in Fig. 5(d)
and they are fitted by using the same function as that used in
the case for Zn:LiNbO3. The effective trap densities are fit-
ted to be 1.37⫻1016 cm−3 for CIn3 and 2.10⫻1016 cm−3 for
CIn5, respectively. Note that high-order diffraction appeared
again at large grating spacing and these data were not con-
sidered in the theoretical fitting.
We should remind here that the fitting function is a result
based on the pure diffusion mechanism. In the samples CLN
and CIn1 we observed fairly high diffraction efficiency but
negligible beam coupling. Consequently the UV photorefrac-
tion in these two crystals is probably based on photovoltaic
effect.
The UV photorefraction in highly In-doped LiNbO3is
also enhanced although it is not as effective as those in
highly Zn-doped LiNbO3. At the same time, with the increas-
ing of the In doping concentration, diffusion turns to domi-
nant mechanism over other mechanisms. We also notice that
the concept of concentration threshold for In in visible is no
longer valid for photorefraction in UV.
C. Na doped congruent LiNbO3
Sodium is +1 valence dopant. Na:LiNbO3crystals
showed different behaviors from the two formers in UV pho-
torefractive experiment. Two samples of CNa1 and CNa3
were used in our experiment. They were studied in visible by
Kong et al. in 199719 and it was reported that Na:LiNbO3
showed optical-damage-resistance or photorefractive resis-
tance compared to the congruent pure LiNbO3. However, the
ability to resist the opically-induced damage became weaker
with increasing Na concentration. So Kong et al. supposed
that Na is not a good optical-damage-resistant impurity.
Based on two-wave coupling configuration, we measured
the UV photorefractive properties of CNa1 and CNa3. The
diffraction efficiency was achieved with the two writing
beam intensities of 121.7 and 152.5 mW/cm2, respectively.
Figure 2(d)shows the typical grating recording and erasing
cycle. The temporal evolution of a long time scale is also
shown in Fig. 3(d). The saturated diffraction efficiency for
the sample CNa3 is nearly twice as that of CNa1. For the
long time recording, the diffraction efficiency did not vary
much after it reached the saturation. In general, the photo-
conductivity in Na:LiNbO3is lower than that in Zn:LiNbO3
or In:LiNbO3. From Table II, it is seen that the sample CNa3
exhibits a higher diffraction efficiency but lower photocon-
ductivity than CNa1. This phenomenon is different from that
in Zn:LiNbO3and In:LiNbO3, in which the increasing of
diffraction efficiency accompanies with the increasing pho-
toconductivity. The dark conductivity in Na-doped LiNbO3
is also very low, as can be inferred from the dark decay
curves in Fig. 4(d). The dark decay time constant of gratings
in CNa1 is approximately the same as that in CLN, while it
is much larger in CNa3. In the light amplification experi-
ment, we set the intensity ratio of two writing beams to be
1:1 and 100:1, respectively. In both cases the two-wave cou-
pling gain coefficient ⌫is negligibly small. So we suggested
that the main charge transport mechanism is the photovoltaic
effect but not diffusion in both CNa1 and CNa3. This con-
clusion is also valid for CLN as well. The calculated sensi-
tivity and dynamic range are also listed in Table II. It seems
that doping LiNbO3with Na does not improve the recording
sensitivity. Although the sample CNa3 has a fairly high dif-
fraction efficiency and a long dark decay time constant
which might be favorable for holographic storage, the low
sensitivity may be a barrier for such applications.
Na:LiNbO3crystals exhibit different UV photorefractive
properties from Zn:LiNbO3or In:LiNbO3. With the increase
of Na doping concentration, the saturated diffraction effi-
ciency increases while the photoconductivity decreases. On
the other hand, contribution from diffusion is negligible and
any light amplification cannot be observed during the two-
wave coupling. Such difference probably results from the +1
valence of sodium which is identical to that of lithium.
IV. LIGHT-INDUCED SCATTERING AND LENS-LIKE
EFFECT OF DOPED LiNbO3IN UV
Light-induced scattering (LIS)occurring in volume pho-
torefractive crystals is believed to be a charateristic of pho-
torefractive materials. LIS is explained as the amplified weak
scattered light due to the gratings recorded by the incident
light and its scattered lights in non-local response
medium,28,29 or due to a multi-wave mixing among the inci-
dent light and its scattered lights in local response medium.30
One often evaluates a photorefractive crystal according to its
LIS characteristics. However, for applications such as optical
storage and light amplification, LIS is a noise source that
induces serious deterioration. On the other hand, lens-like
QIAO et al. PHYSICAL REVIEW B 70, 094101 (2004)
094101-6
effect (LLE)in photorefractive crystals is slightly different
from LIS. Here LLE includes self-focusing or self-
defocusing effects in the crystal. LLE is often used to check
the ability of resistance against light-induced damage of a
material simply by a focused Gaussian laser beam31,32 and by
means of this method Kong et al. proved that the threshold
intensity for the appearance of optical damage in In:LiNbO3
of an In concentration of 5 mol % was two orders higher than
that of the congruent pure LiNbO3.15 Although both LIS and
LLE are due to the diffraction of the incident beam from
parasitic gratings recorded in the crystal, they have different
characteristics. However, one often thinks that LIS and LLE
are two accompanying phenomena, or even confuse them as
the same one. Indeed, both LIS and LLE are suppressed in
visible in LiNbO3crystals such as Mg:LiNbO3, Zn:LiNbO3
and In:LiNbO3when the impurity concentration is over the
threshold value. But this is not the case for LiNbO3in UV.
First, we introduced a focused extraordinarily polarized
laser beam and put the samples near to the rear focal plane of
the lens to observe the LLE. We could completely eliminate
LIS by moving the crystal near to rear focal plane of the lens
because of the “speckle size effect” for LIS.33 The profiles of
the light spots deteriorated after passing through all our
samples, as shown in Figs. 6(a),6(b), and 7(a), where the
intensity density of incident UV light was 23.9 kW/cm2.Itis
found from Figs. 6(a)and 6(b)that LLE is strong in CLN
and Na:LiNbO3, while it is suppressed in CZn5 and com-
pletely disappears in CZn7 and CZn9 with an available in-
tensity of 23.9 kW/cm2. The results shown in Fig. 7(a)are
similar to what Kong et al. obtained at 488 nm.15 The laser
beam passed through the crystal CIn5 without any distortion
while it was seriously deteriorated after passing the other two
In-doped crystals. We also did the same measurement on
three additional In-doped LiNbO3crystals which were also
provided by Russia Academy of Science. These three new
samples contain less In concentrations which are denoted in
Fig. 7(c). It is shown in Fig. 7(c)that the serious beam dis-
tortion occurs in these crystals. The less is the In concentra-
tion inside, the larger is the distortion. We measured the in-
tensity threshold of the LLE for our samples and the results
are listed in Table III. From the table we see the samples
CIn5, CZn7, and CZn9 have very high intensity thresholds
which are beyond total output intensity of the laser. Here we
may come to a conclusion that in UV LLE of LiNbO3can be
suppressed by Zn or In doping. This conclusion is in consis-
tent with that in visible, e.g., LLE of Zn or In doped LiNbO3
is suppressed both in visible and in UV. CNa1 and CNa3 are
somewhat different from others. The sample CNa3 shows a
lower threshold than CNa1. Na is not an effective dopant to
resist the optical damage as compared with Zn or In. This is
also valid in visible.19 As for as the LLE effect is considered
the threshold characters in UV and in visible are the same.
Secondly, we moved the focusing lens away and let the
extraordinarily polarized and collimated laser beam illumi-
nate the crystals directly. In this case LIS occurred, as shown
in Figs. 6(c)and 7(b). Among our samples, we could not
FIG. 6. (Color online)Pictures of LLE and LIS in CLN,
Na:LiNbO3and Zn:LiNbO3:(a)Beam deformation after passing
through CLN, CNa1, and CNa3 with focusing beams; (b)beam
deformation after passing through CZn5, CZn7, and CZn9 with
focusing beams; (c)LIS in CZn5, CZn7, and CZn9 with extra-
polarized laser beam.
FIG. 7. (Color online)Pictures of lens-like and light induced
scattering in In:LiNbO3:(a)Beam deformation after passing
through CIn1, CIn3, and CIn5 with focusing beam; (b)LIS in CIn1,
CIn3, and CIn5 with extra-polarized laser beam; (c)beam deforma-
tion after passing through In:LiNbO3crystals grown in Russia with
focusing beam.
TABLE III. The intensity threshold of the LLE at 351 nm in our samples.
Samples CLN CZn5 CZn7 CZn9 CIn1 CIn3 CIn5 CNa1 CNa3
Thresholda共kW/cm2兲0.09 17.9 ⬎23.9 ⬎23.9 0.16 10.9 ⬎23.9 0.13 0.078
aThe highest value is limited by the maximum output of our laser— 23.9 kW/cm2.
ULTRAVIOLET PHOTOREFRACTIVITY FEATURES IN…PHYSICAL REVIEW B 70, 094101 (2004)
094101-7
observe LIS in CLN, CNa1, CNa2, and CIn1. This is in
consistence with the results obtained in the light amplifica-
tion experiment because the two-wave coupling gain coeffi-
cients are negligibly small. No LIS could be observed in the
three additional In:LiNbO3samples either. In the rest four
samples, e.g., CZn7, CZn9, CIn3, and CIn5, strong asym-
metric scattering fanning toward −c-axis direction was ob-
served and it became stronger with the increase of Zn or In
concentrations in the crystal. Such results agree well with the
results of light amplification in Sec. III and Table II. We may
also draw a conclusion that the dominant charge carriers in
these samples are electrons. It is seen that the results of LIS
in UV are exactly contrary to those in visible.
V. DISCUSSION
It seems to be a well-known result that holes are the
dominant charge carriers in congruent pure LiNbO3in UV
photorefractive effect8,9 and in highly Zn doped LiNbO3in
visible.26 However, we notice the energy transferring direc-
tion is always toward the −c-axis direction during the two-
wave coupling experiments for the samples Zn:LiNbO3and
In:LiNbO3in which fairly large coupling gain coefficient ⌫
was observed. This indicates that in these samples diffusion
is the dominant charge transport mechanism and the light-
excited electrons are the dominant charge carriers. This is
inconsistent with what was reported by Jungen and Laeri.8,9
Here we summarize briefly the similarities and the differ-
ences between LLE and LIS in visible and those in UV. In
visible, both LLE and LIS are suppressed in Zn:LiNbO3and
In:LiNbO3with the doping concentration over their respec-
tive threshold. In CZn7, CZn9, and CIn5, both LLE and LIS
are negligible in visible. However, in UV LIS in them be-
comes stronger, although LLE is still suppressed as in the
case in visible. In other words, LIS in Zn or In doped
LiNbO3crystals has the contrary concentration dependence
in visible and in UV, while LLE has similar concentration
dependence. The threshold value of the doping concentration
is valid for LLE both in visible and in UV but no longer
valid for LIS in UV. But in Na:LiNbO3crystals, the results
are different. The laser beam distortion due to LLE does not
become better in CNa1 and CNa3 compared with that in
CLN, neither in visible nor in UV. LIS cannot be observed in
them either. In a word, Na:LiNbO3crystals exhibit the simi-
lar properties of LLE and LIS in visible and in UV, which is
not as the case of Zn or In doped crystals.
As is known, LLE and LIS are different behaviors result-
ing from the same origin—photorefractivity. Here we do not
consider thermal lensing for the following reason. Thermal
lensing originates from the thermal-optic effect. In general,
the response time of the thermal lensing is very short in the
order of 10−9 s. The response time of both LLE and LIS are
in the order of subsecond in our experiments. So we could
exclude the contribution from thermal lensing. Therefore,
both LLE and LIS result from the diffraction of the incident
beam on parasitic gratings recorded due to the photorefrac-
tive effect of the crystals. The difference is that LIS occurs
from gratings with different spatial frequencies (usually
high)recorded by the incident beam and coherent noise
beams, while in our case LLE is restricted only to gratings
with low frequencies recorded by different components of
the incident focusing beam. Therefore, LLE is observed only
at small angle from the direction of the transmitted beam,
and LIS is observed at a wide range of diffractive angles. In
our opinion, the essential difference between them is that
LLE is connected with the photovoltaic effect and LIS re-
sults from diffusion. LLE disappears in CZn7, CZn9, and
CIn5 because the photovoltaic effect decreases significantly
in them due to the increase of photoconductivity and dark
conductivity. At the same time, with a larger photoconduc-
tivity the crystal shows a higher sensitivity and a shorter
response time, which is desired for the real-time applica-
tions. By doping Zn or In the photoconductivity of LiNbO3
crystal is greatly enhanced especially when the doping con-
centration is over the threshold value. Whereas, the dopant
Na is not a good candidate to make a LiNbO3crystal with
high photoconductivity. What should be reminded of is that
we did not consider the variation of the Glass constant which
might be changed by doping LiNbO3with different dopants
of different concentrations.
On the other hand, with the decrease in the photovoltaic
field of Zn:LiNbO3and In: LiNbO3, the diffusion field domi-
nates over the photovoltaic field especially at large recording
angles. As is well known, diffusion leads to unidirectional
energy transferring via two-wave coupling. Therefore, in our
experiment unidirectional LIS in large angle range could be
observed in Zn:LiNbO3and In:LiNbO3samples except for
CIn1 (seen in Figs. 6 and 7). However, in CLN, CNa1,
CNa3, and CIn1, the photoconductivity is not high enough.
Comparing to the photovoltaic effect, the contribution from
diffusion is still neglectable so unidirectional LIS was not
observed in these four samples. But, the reason why diffu-
sion field should be enhanced with the increase of Zn or In
doping concentration still remains unclear to us.
According to the above results on LLE and LIS, Zn or In
doped LiNbO3crystals are able to find new applications in
UV. For instance, in highly Zn-doped LiNbO3crystals, one
can take advantage of the greatly enhanced photorefractive
sensitivity in UV but does not need to care about LLE of the
crystal any longer.
In general, when two coherent laser beams intersected in
the samples, amplitude gratings due to the light-induced ab-
sorption (LIA)and phase gratings due to the photorefractiv-
ity may be recorded. A characteristic of the photorefractive
effect, which distinguishes it from other mechanisms, is the
nonzero phase shift between the refractive-index grating and
the light interference pattern for holographic recording with-
out an external electric field. This phase shift would result in
a unidirectional energy transfer between the interacting
beams, while absorption gratings can only cause simulta-
neous changes in the intensities. We did the LIA experiment
according to the configuration of Ref. 34 and no obvious LIA
could be observed in all our samples. Furthermore, we esti-
mated the amplitude of the refractive index change ⌬nac-
cording to the relations
=sin2关
⌬nd/共cos
兲兴 and ⌫
=2共
⌬n/兲sin
, respectively (we estimate
to be
/2),in
those samples two-wave coupling gain coefficient ⌫could be
measured. We found that ⌬nwere approximately the same in
both cases. All the above results show that the amplitude of
QIAO et al. PHYSICAL REVIEW B 70, 094101 (2004)
094101-8
the absorption grating is negligibly small as compared to that
of the photorefractive gratings in our case. Therefore, in the
following discussions we consider only the photorefraction.
According to the results described above, we can see that
in the samples CLN, CIn1, CNa1 and CNa3, the photovol-
taic effect is the dominant contribution to the photorefractiv-
ity, while in CZn7, CZn9, and CIn5, diffusion played the key
role, and in CZn5 and CIn3, there is a competition between
the two mechanisms. Due to the strong influence from the
high-order diffraction at large grating spacings (small record-
ing angles), we cannot get direct evidences of competition
between these two mechanisms. In our opinion, no matter it
is a congruent pure crystal or the doped ones, the increasing
photoconductivity results directly in the decrease of the pho-
tovoltaic field and therefore the increase of contribution from
diffusion. According to the results of Ref. 9, with the same
light intensity, photoconductivity of LiNbO3in UV is larger
than that in visible. In the samples CLN, CIn1, CNa1, and
CNa3, the diffraction efficiency
is independent of the re-
cording angle or the grating spacing. This is a typical prop-
erty resulting from the photovoltaic space-charge field. When
the dopants like Mg, Zn, or In are introduced, photoconduc-
tivity in crystals is enhanced and photovoltaic field de-
creases, and diffusion becomes gradually the dominant
mechanism.
In comparison with that in visible, the samples show a
significant enhancement in UV photorefractivity. We think
that it is probably because the high energy of UV photons
can excite charge carriers from some new UV photorefrac-
tive centers. Then what are the new photorefractive centers?
According to the Li-vacancy model,35 in congruent pure
LiNbO3an intrinsic defect of NbLi (Nb-antisite)is compen-
sated by four VLi (Li-vacancy). As is believed that when
impurities like Mg, Zn, or In incorporate into LiNbO3, they
firstly repel the NbLi to the normal site and at the same time
form the new extrinsic defects of MgLi,Zn
Li,orIn
Li.Asa
matter of fact, the concentration threshold for optical-
damage-resistance in visible corresponds to the vanishing of
NbLi.36 In other words, when the doping concentration is
higher than the threshold, there is no NbLi any more in the
crystal. Considering that the UV photorefraction is enhanced
dramatically in samples CZn7, CZn9, and CIn5 whose dop-
ing concentrations are already higher than the threshold, the
new UV photorefractive centers should not be related to NbLi
centers. With regard to the extrinsic defects MgLi,Zn
Li,or
InLi, we do not think that they may act as the UV photore-
fractive centers because they are monovalence so that elec-
trons or holes are not able to be excited from or trapped on
them. In a word, the UV photorefractive centers should be
related to some defect structures other than intrinsic defects
of NbLi and extrinsic defects of MgLi,Zn
Li,orIn
Li.
In 2000, Lee et al. reported a UV-sensitive deep center in
near stoichiometric LiNbO3codoped with Tb and iron.37
This UV-sensitive center could be considered as an electron
donor/acceptor under the UV illumination. However, they
did not give other information on the microstructure of this
UV-sensitive center.
Recently Vikhnin et al. demonstrated an exciton structure
theoretically and experimentally in ABO3-type ferroelectric
oxides, and they named it as CTVE (charge transfer vibronic
excitons).38 They believe that CTVE is composed of a pair of
an electron polaron and a hole polaron and it can be treated
as a deep-level center with its energy level located just above
the top of the valence band. We measured the absorption
spectra for all the samples, and the spectra are shown in Fig.
8. Considering the possible different growing conditions in
two crystal-grown groups, we chose two congruent pure
LiNbO3, one was grown by us and the other was grown in
Russian Academy of Science, respectively, for a more rea-
sonable comparison of the absorption spectra. One sheet of
congruent pure LiNbO3with a thickness of 3 mm grown by
us was compared with the In:LiNbO3and Na:LiNbO3, and
FIG. 8. The absorption spectra
near the absorption edge of our
samples, the wavelength indicated
by the small arrows is our record-
ing wavelength—351 nm.
ULTRAVIOLET PHOTOREFRACTIVITY FEATURES IN…PHYSICAL REVIEW B 70, 094101 (2004)
094101-9
another sheet with a thickness of 1.2 mm grown in Russian
Academy of Science was compared with the Zn:LiNbO3.
That is the reason of different values of absorption coeffi-
cient for congruent pure LiNbO3in Fig. 8. The absorption
edge moves to shorter wavelengths with Zn-In doping, yet
the absorption coefficients at 351 nm are still larger than the
congruent pure one and the fact indicates an existence of
some deep-level centers. Anyway, only from Fig. 8 we can-
not verify the existence of CTVE structure in our samples.
Other mechanisms such as multi-centers models may also be
possible. Although many results are still unclear, the UV
photorefractive properties in LiNbO3give insight into to the
deep-level centers in the crystal. Further detailed investiga-
tions on the defect structure of LiNbO3are necessary to
clarify the UV photorefractivity.
Before the conclusion, we offer some comments on the
Na:LiNbO3crystals. Na:LiNbO3crystals exhibited different
UV photorefractive properties from Zn:LiNbO3and
In:LiNbO3. Such differences perhaps come from the differ-
ent influence of Na and Zn-In on defect structures of
LiNbO3. Na is a monovalence element, which does not need
a charge compensation when it enters Li-site. It may firstly
occupy Li-site and repel the NbLi to the normal Nb-site. The
defects of NbLi and VLi decrease sharply, and the absorption
edge has a blue-shift in CNa1. With the increase in Na dop-
ing concentration, no more NbLi or VLi are removed, and the
basic spatial arrangement of LiNbO3does not permit more
Na ions. Anyway, the situation in Na-doped LiNbO3crystals
is much more complicated than that in Zn and In doped
LiNbO3crystals. We do not have ideas on the explanation of
the abnormal behaviors of UV photorefraction of Na-doped
LiNbO3crystals so far.
VI. CONCLUSION
In conclusion, in this work we studied the UV photore-
fractive features in three types of doped lithium niobate crys-
tals and found that the properties of photorefractive effect in
UV are far different from that in visible. UV photorefraction
is enhanced in the so-called optical-damage-resistant crys-
tals, such as Zn and In doped LiNbO3, and their optical-
damage-resistance and the threshold for resistance are only
effective in visible. From another aspect, these crystals espe-
cially Zn:LiNbO3with fast response are excellent candidates
for UV photorefractive applications. Based on the character
of optical-damage-resistance in visible and a photorefraction
enhancement in UV, LiNbO3crystals doped with appropriate
impurities are probably an outstanding platform for the mi-
crostructure fabricated in UV and can be used in visible. This
work also leads us a new scope for defect structures in Zn
and In doped LiNbO3crystals, although the defects related to
light-induced charge carriers process in UV are not clear yet.
Investigation on this field is greatly needed for understanding
the famous and versatile optical material—LiNbO3.
ACKNOWLEDGMENTS
This work is sponsored by the “973” project
(G1999033004)of China, Tianjin Key Technique Project
(Grant Nos. 0031002311 and 003108911), The National
Natural Science Foundation of China (Grant Nos. 6010801
and 60308005), the Key Project of the National Natural Sci-
ence Foundation of China (Grant No. 10334010), the Excel-
lent Young Teachers Program of MOE, P.R.C. (Grant No.
2002-350), and the Special Foundation of Education Com-
mittee for Excellent Teachers. Thanks to Professor Romano
Rupp from Vienna University for his very helpful advice and
discussion. Thanks to Professor Tatyana Volk from Russian
Academy of Science for her LiNbO3crystals. Thanks to Pro-
fessor Jean Pierre Huignard and Professor Brigitte Loiseaux
in Thales Research and Technology (France)for helpful dis-
cussion.
*Electronic address: jjxu@nankai.edu.cn
1M. S. McPherson, I. Ostrovskii, and M. A. Breazeale, Phys. Rev.
Lett. 89, 115506 (2002).
2A. Ashkin, G. D. Boyd, J. M. Dziedzic, R. G. Smith, A. A.
Ballman, J. J. Levinstein, and K. Nassau, Appl. Phys. Lett. 9,72
(1966).
3S. Tanzilli, W. Tittel, H .D. Riedmatten, H. Zbinden, P. Baldi, M.
D. Micheli, D. B. Ostrowsky, and N. Gisin, Eur. Phys. J. D 18,
155 (2002).
4K. R. Parameswaran, J. R. Kurz, R. V. Roussev, and M. M. Fejer,
Opt. Lett. 27,43(2002).
5E. Krätzig, Ferroelectrics 21, 635 (1978).
6O. F. Schirmer, O. Thiemann, and M. Wöhlecke, J. Phys. Chem.
Solids 52, 185 (1991).
7T. R. Volk and M. Wöhlecke, Ferroelectr. Rev. 1, 195 (1998).
8R. Jungen, G. Angelow, F. Laeri, and C. Grabmaier, Appl. Phys.
A: Solids Surf. 55, 101 (1992).
9F. Laeri, R. Jungen, G. Angelow, U. Vietze, T. Engel, M. Würtz
and D. Hilgenberg, Appl. Phys. B: Lasers Opt. 61, 351 (1995).
10Jingjun Xu, Guangyin Zhang, Feifei Li, Xinzheng Zhang, Qian
Sun, Simin Liu, Feng Song, Yongfa Kong, Xiaojun Chen, Hai-
jun Qiao, Jianghong Yao, and Zhao Lijuan, Opt. Lett. 25, 129
(2000).
11 P. Günter and J. P. Huignard, Photorefractive Materials and Their
Applications I (Springer-Verlag, Berlin, 1989).
12A. Prokhorov and Y. S. Kuzminov, Physics and Chemistry of
Crystalline of Lithium Niobate (Adam Hilger, 1990).
13T. R. Volk and N. Rubinina, Ferroelectr., Lett. Sect. 109, 345
(1992).
14T. R. Volk, M. Wöhlecke, N. Rubinina, N. V. Razumovskii, F.
Jermann, C. Fischer, and R. Böwer, Appl. Phys. A: Mater. Sci.
Process. 60, 217 (1995).
15Y. Kong, J. Wen, and H. Wang, Appl. Phys. Lett. 66, 280 (1995).
16T. R. Volk, V. I. Pryalkin, and N. M. Rubinina, Opt. Lett. 15, 996
(1990).
17G. G. Zhong, J. Jian, and Z. K. Wu, Proceedings of the Eleventh
International Quantum Electronics Conference, IEEE, New
York, 631 (1980).
QIAO et al. PHYSICAL REVIEW B 70, 094101 (2004)
094101-10
18D. A. Bryan, R. Gerson, and H. E. Tomaschke, Appl. Phys. Lett.
44, 847 (1984).
19Y. F. Kong, J. K. Wen, and Z. F. Yang, Chin. Phys. Lett. 14, 629
(1997).
20Haijun Qiao, Jingjun Xu, Yongfa Kong, Qian Sun, and Guangyin
Zhang, OSA Trends in Optics and Photonics 62, 139 (2000).
21T. K. Gaylored, T. A. Rabson, F. K. Tittel, and C. R. Quick, J.
Appl. Phys. 44, 896 (1973).
22M. Segev, A. Kewitsch, A. Yariv, and G. A. Rakuljic, Appl. Phys.
Lett. 62, 907 (1993).
23Y. Liu, L. Liu, and C. Zhou, Opt. Lett. 25, 551 (2000).
24H. Kongelnic, Bell Syst. Tech. J. 48, 2909 (1969).
25H. Wang, J. Wen, J. Li, and H. Wang, Appl. Phys. Lett. 57, 344
(1990).
26J. Deng, J. Wen, Z. Wu, and H. Wang, Appl. Phys. Lett. 64, 2822
(1994).
27M. D. Ewbank, R. R. Neurgaondar, W. k. Cory, and J. Feinberg,
J. Appl. Phys. 62, 374 (1987).
28É. M. Avakyan, K. G. Belabaev, and S. G. Odulov, Sov. Phys.
Solid State 25, 1887 (1983).
29R. A. Rupp and F. W. Drees, Appl. Phys. B: Photophys. Laser
Chem. 39, 223 (1986).
30G. Zhang, G. Tian, S. Liu, J. Xu, G. Zhang, and Q. Sun, J. Opt.
Soc. Am. B 14, 2823 (1997).
31F. S. Chen, J. Appl. Phys. 40, 3389 (1969).
32F. Jermann, M. Simon, and E. Krätzig, J. Opt. Soc. Am. B 12,
2066 (1995).
33G. Y. Zhang, Q. X. Li, P. Pho, S. M. Liu, Z. K. Wu, and R. R.
Alfano, Appl. Opt. 25, 2955 (1986).
34S. Wevering, J. Imbrock, and E. Krätzig, J. Opt. Soc. Am. B 18,
472 (2001).
35N. Iyi, K. Kitamura, F. Izumi, J. K. Yamanoto, T. Hayashi, H.
Asano, and S. Kimura, J. Solid State Chem. 10, 340 (1992).
36G. Zhang, Y. Kong, and J. Xu, OSA Trends in Optics and Photo-
nics 62, 194 (2001).
37M. Lee, S. Takedawa, Y. Furukawa, K. Kitamura, and H. Hatano,
Phys. Rev. Lett. 84, 875 (2000).
38V. Vikhnin, R. Eglits, S. Kapphan, G. Borstel, and E. Kotomin,
Phys. Rev. B 65, 104304 (2002).
ULTRAVIOLET PHOTOREFRACTIVITY FEATURES IN…PHYSICAL REVIEW B 70, 094101 (2004)
094101-11
