Nonstoichiometry and Crystal Growth of Lithium Niobate

J. R. Carruthers, G. E. Peterson, M. Grasso, and P. M. Bridenbaugh

Citation: Journal of Applied Physics 42, 1846 (1971); doi: 10.1063/1.1660455
View online: http://dx.doi.org/10.1063/1.1660455
View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/42/5?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Growth and characterization of Er-doped single crystal lithium niobate fibers
J. Appl. Phys. 104, 103114 (2008); 10.1063/1.3021292
Strongly sublinear growth of the photorefractive effect for increasing pump intensities in doped lithiumniobate crystals
J. Appl. Phys. 101, 116105 (2007); 10.1063/1.2739354
Domain growth kinetics in lithium niobate single crystals studied by piezoresponse force microscopy
Appl. Phys. Lett. 86, 012906 (2005); 10.1063/1.1845594
Growth and two-color holographic storage properties of Mn-doped lithium niobate crystals with varying Li
Nb ratio
J. Appl. Phys. 96, 5996 (2004); 10.1063/1.1812824
Nonstoichiometry and electrocoloration due to injection of Li + and O 2 ions into lithium niobate crystals
J. Appl. Phys. 88, 5687 (2000); 10.1063/1.1318367

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
14.139.220.33 On: Wed, 01 Jul 2015 14:34:49

JOURNAL OF APPLIED PHYSICS

VOLUME 42, NUMBER 5

APRIL 1971

N onstoichiometry and Crystal Growth of Lithium Niobate

J. R. CARRUTHERS, G. E. PETERSON, AND M. GRASSO
Bell Telephone Laboratories, Incorporated, Murray Hill, Nr:w Jersey 07974
AND

P.

M.

BRIDENBAUGH

Bell Telephone Laboratories, Incorporated, Holmdel, Nr:w Jersey 07733

(Received 1 July 1970; in final form 13 October 1970)
Deviations from stoichiometry occurring during the crystal growth of lithium niobate have been studied
by 93Nb NMR and by shifts in the ferroelectric transition temperature. The high-temperature phase equilibria of nonstoichiometric lithium niobate were accurately determined by these techniques together with
DTA and crystal-growth experiments. The congruently melting composition is located at 48.6 mole%
Li 20 and the solid-solution region at room temperature extends from 48% to 50% LhO. The implications
of growth at noncongruently melting compositions on the compositional uniformity of lithium niobate are
discussed in detail.

crystals grown by the Czochralski technique were found

to be nonstoichiometric. We then found that stoichioLithium niobate and lithium tantalate flux-grown metric crystals could be grown only from the flux or
crystals were first reported to be ferroelectric by from 58.0 mole% Li20 (off stoichiometric) melts.
Matthias and Remeika! in 1949 on the basis of hysteresis Similar shifts in stoichiometry have been found by
loops observed in the tantalate only. The phase equilibria Barns and Carruthersll for lithium tantalate on the
of the Li20-Nb20 5 system were reported by Reisman basis of precise lattice parameter, density, and Curieand Holtzberg2 in 1958. Their phase diagram indicated temperature measurements.
that LiNbOa melted congruently at 1253C. Large
Many unexplained optical phenomenon a have been
crystals of both lithium niobate and lithium tantalate observed in lithium niobate crystals. These were first
were grown from their melts by Ballman in 1964 using reported by Peterson et al. 12 and have been rediscovered
the Czochralski technique. 8 Subsequently many im- by almost everyone working with this material. Some of
portant properties of lithium niobate were investigated these effects are light emission upon heating crystals
by Nassau, Levinstein, and Loiacono. 4 ,5 In particular, through a temperature of 200C, residual birefringence
the Curie temperature was found to be very close to the parallel to the c axis and very large optical scattering in
melting point and several techniques were developed to unpoled crystals. A nuclear quadrupole resonance inproduce single-domain crystals.
vestigation of stoichiometric lithium niobate by
The first experiments performed to investigate the Schempp, Peterson, and Carruthers13 revealed no
influence of stoichiometric deviations on crystal evidence for a structural phase transition between 22
properties of ferroelectricswere those of Ballman, and 550 0 K. However a recent single-crystal NMR
Levinstein, Capio, and Brown on lithium tantalate. 6 study of nonstoichiometric lithium niobate by Peterson
Here the Curie temperature and birefringence were and Carruthers 14 has demonstrated the presence of new
both found to vary over a wide range of values for lithium and niobium sites. If these optical phenomenon a
changes in melt composition of up to 7-8 mole% are related to the defect nature of nonstoichiometric
(excess or deficit) Li20. Similar results were found by lithium niobate, then the high-temperature phase
Bergman et af.1 and also by Fay, Alford, and Dess8 on equilibria must be determined with more precision to
lithium niobate. Since the early phase diagram in- establish a starting point for future investigations.
dicated no solid-solution range for lithium niobate,
An understanding of the shifts in stoichiometry that
there was a need to reexamine this region more pre- occur on freezing is essential if some control over the
cisely. This information was provided by the recently crystal composition (and properties) is desired. The
published phase equilibrium investigation of Lerner, most important effect of variable composition in
Legras, and Dumas.9 These workers showed that there lithium niobate is the variation in phase match temwas a solubility or existence range from 48.0 to 50.0 perature for second-harmonic generation which occurs
mole% Li2 0 at room temperature and from 46.0 to 50.0 over the length of the crystal because of the normal
mole% Li20 at 1000C. Furthermore, x-ray lattice freezing distribution of composition. Other properties
parameter measurements on crystals and ceramic which may be affected are the electro-optic and piezospecimens indicated that the congruently melting electric coefficients. It has recently been confirmed by
composition was not at 50% Li2 0 but rather somewhere Nash, Boyd, Sargent, and Bridenbaugh!5 as well as
between 48% and 49% Li2 0. This shift was verified by Byer, Young, and Feigelson 16 that lithium niobate
Peterson and Carruthers lO using a new nuclear magnetic crystals grown at the congruently melting composition
resonance technique for investigating the shifts from (48.6% Li20) do not possess this variation. An associstoichiometry in lithium niobate. In fact, all early ated, undesirable effect of compositional variations is
1846
INTRODUCTION

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
14.139.220.33 On: Wed, 01 Jul 2015 14:34:49

1847

NONSTOICHIOMETRY OF LITHIUM NIOBATE

the change in extraordinary refractive index (and birefringence). These variations take two forms (Nash

et at. IS )

1200

r--r--r----,r----,---,---;,r;r---,----,--i

1180

CERAMIC OATA ______ _ ,./

(LiNbO,+1.0-" MQO)
--,.

tI

(i) A slow variation over the length of the crystal so

that one/ol~3X1O-4 cm-I ;
(ii) fluctuations of the order of one~lO-s over the
growth striations caused by growth-rate fluctuations.
Another undesirable form of compositional variation is
the facet effect which occurs in these materials because
of the high entropy of fusion (Jackson's a factorl7 for
c-axis growth in lithium niobate is a=4). The facet
effect gives rise to a radial variation in composition
across the crystal (see, for example, the recent work of
Cockayne, Chesswas, and Gasson l8 on garnets).
The main purpose of this paper is to establish more
precisely the high-temperature phase equilibria relevant
to the crystal growth of lithium niobate.
The general philosophy adopted in measuring nonstoichiometric shifts was based on the lack of availability of chemical analytical techniques with sufficient
precision. We therefore carefully prepared ceramic
specimens of known compositions for use in obtaining
calibration curves of Curie temperature and 93Nb NMR
derivative linewidth. These curves were then used to
determine the unknown compositions of crystals. It is
obvious that if the melt compositions of these crystals
are known then we can determine the location of the tie
lines between liquidus and solidus.
EXPERIMENTAL PROCEDURE
Specimen Preparation

Ceramic specimens were prepared from 99.99%

Li2COa (United Mineral) and 99.9% N~05 (CIBA).
Specimens were generally 1-3 g, carefully mixed in a
mortar and pestle, pressed at 10 000 psi and fired at
temperatures up to 50C below solidus temperatures,
where appropriate, for 12 h. This sequence was repeated to ensure complete reaction. Some LiNb0 3
specimens were also prepared with 1.0 mole% MgO to
compare with the results on early crystals grown with
MgO.4.S
Crystals were grown by the Czochralski technique
using both rf and resistance heated furnaces from melts
of different compositions. Growth rates of 0.2-0.5 in./h
and crystal rotation rates of 30 rpm were used. In order
to minimize normal freezing variations in stoichiometry,
no more than 15%-20% of the melt was pulled.
Measurement Techniques

DTA measurements were made on a Dupont 1600

Thermoanalyzer. X-ray diffraction measurements were
carried out on a Nonius-Guinier focusing camera using
monochromatic CuKa radiation.
Dielectric measurements to record the Curie temperature were performed on a General Radio 1673A

"

CRYSTAL DATA
(MELT COMPOSITION)

1000

44.0

45.0

46.0

47.0

48.0

49.0

50.0

51.0

52.0

53.0

MOLE PERCENT LllO

FIG. 1. Variation in Curie temperature with solid or liquid

composition for ceramics or crystals, respectively.

capacitance bridge with platinum electrodes at 1 kHz.

Ceramic specimens were pelletized at room temperature,
fired at the temperatures referred to previously, furnace
cooled to 1000C and removed to air cool. The pellets
were ground to 15-20-mil thickness and platinum paste
electrodes painted on both sides.
The details on the NMR apparatus and techniques
may be found in earlier work. 19 .20
EXPERIMENTAL RESULTS
The Curie temperatures as obtained from dielectric
measurements are shown in Fig. 1. The results for
crystals grown from known melt compositions have been
included from the work of Bergman et al.7 The ceramic
data (with experimental points shown), where the solid
composition is known, are seen to vary over the solidsolution range and to be constant outside this region.
The dielectric peaks for the ceramic lithium niobate
specimens were considerably broader than the crystals
and were subject to scatter. There are a number of
explanations for this scatter. The most important effect
in lithium niobate is the very large change in capacitance
and conductivity of the specimens at temperatures so
close to the melting point. Above 1140C, this change is
larger than that due to the ferroelectric transition in the
ceramic specimens so that the error in Tc is lOoC in
this region. In addition, the platinum electrodes are
quite soft and contact problems exist. The ceramic
pellets also develop fractures at these high temperatures.
Both of these problems introduce air gaps in the
electrical path which contribute disproportionately to
the total capacitance.
Additional complications are seen to arise in Fig. 1 at
both ends of the solid-solution range. From 45.0 to 46.0
mole% Li20, an extra peak is seen to occur which undoubtedly is associated with the dissolution and precipitation kinetics encountered while cycling through
the Curie temperature. The solvus line at 1020C is
nonvertical as will be seen later.
Ceramic data are included for specimens with 1.0
mole% MgO because many of the early crystals grown

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
14.139.220.33 On: Wed, 01 Jul 2015 14:34:49

1848

CARRUTHERS ET AL.

stoichiometric (50% Li20) specimens but rather low for

specimens near 48.0% Li20. Thus this method has been
25
used to establish tie lines on specimens containing
20
,48.6%-50.0% Li20 while the Curie-temperature method
has been used on specimens containing less than 48.6%
15
Li20. The NMR results are given in Table I for crystals
grown at 0.5 in./h and 30 rpm.
10
The phase relationships established between crystal
and melt compositions by the Curie-temperature and
NMR techniques are shown in Fig. 3. Data points have
~~6--4~7~~48~-4~~--~50~~5~1--~52~~5~3--~54~~55
been
omitted for clarity. The positions of the liquidus
MOLE PERCENT L.i 0
2
and isothermal eutectic lines were determined to
FIG. 2. Variation of half-width of low-field derivative 98Nb within soC by DTA. The solidus lines, which were not
NMR peak as a function of composition for lithium niobate detectable by DTA, were plotted from the information
ceramic specimens.
of Fig. 1 and Table 1. The ferroelectric-paraelectric
by Nassau and Loiacono,4,5 on which Tc measurements
were made, contained MgO. Nassau 21 reported that for
additions of !% MgO to the melt, the distribution coefficient for Mg was ko= 1.3. It can be seen that Tc is
increased uniformly by 22C for ceramic specimens
from 47.0 to 50.0% Li20. Thus this addition accounts
for the increased Tc range of 1177-119S oC reported
previously. The phase matching temperature is also
increased by MgO additions as will be discussed later.
The most important result obtained from the Curietemperature data of Fig. 1 is that the isothermal tie lines
representing solid-liquid concentration relationships
may be read directly across the solid-solution range.
This information allows the computation of effective
distribution coefficients for noncongruently melting
compositions. The degree to which these represent
equilibrium conditions is being investigated. Furthermore, it can be seen that there is one composition where
the ceramic and crystal Curie temperatures are equal.
This corresponds to the congruen tl y mel ting composi tion
and is 48.6% Li20 for lithium niobate. It is obvious that
if a crystal of uniform composition is desired, with
minimal striation, coring or normal-freeze variations,
then the congruently melting composition should be
selected. As previously mentioned, crystals of lithium
niobate have recently been grown at the 48.6% Li20
composition and have been shown to be very uniform
by various mapping techniques. 15 ,16
The powder NMR method reported previouslylO is
quite sensitive to stoichiometry in lithium niobate as
shown in Fig. 2. The signal-to-noise ratio is high for
TABLE 1. Isothermal tie lines in lithium niobate as determined
by 93Nb NMR linewidth.
Liquid concentration
(% LbO)

Solid concentration
(%Li,O)

50.0
52.7

54.0

48.95
49.45
49.65

58.0

50.00

1260
LIQUID

1240

1200
1180

I
I

U
'-

"'

1160

1140

1120

:5

1100

1080

"-

I
I

FERROELECTRIC

TRANSITION

1060

liNb03

1040

Li l Nb04

1020

1000
40

48

50

52

54

56

58

MOLE PERCENT Li 20

3. Phase diagram of lithium niobate as determined in this

paper. Experimental points have been omitted for clarity.

FIG.

transition, which was shown by Pankratz and King 22 to

be second order, was also not detectable by DTA.
However the ceramic Tc curve of Fig. 1 is plotted as the
dashed line. It can be seen that this curve meets the
solvus on the Nb20s-rich side at 46.0% Li20 and 1020C.
This agrees very closely with the x-ray results of Lerner,
Legras, and Dumas9 shown as horizontal bars at 1000
and 1170C. Guinier x-ray diffraction showed that the
room-temperature solid-solution range was 48.0-50.0
mole% Li 20. It is interesting to note that lithium
niobate grown from melts richer than 57% Li20 will
freeze directly in the ferroelectric state. The nature
of the small Curie-temperature peak at 45.0-46.0%
Li2 0 in Fig. 1 now becomes clear from its association
with the solvus line shown in Fig. 3.
During the CCl'urse of this work, we performed some
simple experiments to test for the extent of Li 20 losses
during sintering and crystal growth. Such losses were
troublesome in similar work with lithium tantalateY
Weight-loss measurements on specimens annealed for
extended periods of time revealed no detectable
volatilization. In addition these specimens showed no
measurable decreases in the Curie temperatures. We
have concluded that although Li 20 losses undoubtedly

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
14.139.220.33 On: Wed, 01 Jul 2015 14:34:49

NONSTOICHIOMETRY OF LITHIUM NIOBATE

existed, particularly for the samples near 50.0% Li20,

the resultant solid composition changes were below our
limits of detectability.

0.030

DISCUSSION

0.010

We will discuss the problems of deviations from

stoichiometry and crystal growth from noncongruently
melting compositions separately.
Stoichiometric Deviations

The location and extent of the existence regions of

compounds reveals some of their important thermodynamic aspects. Albers and Haas 23 have recently stated
that the existence region is wide when;
(i) The native disorder is large (the curvature of the
free-energy curve is small). Native disorder refers to the
positional interchange of cations anions, interstitialvacancy defects or thermally generated vacancies, all
under the constraint of electrical neutrality;
(ii) the free energy of formation of the compound is
high compared with that of the neighboring phases;
(iii) the neighboring phases have low solubility limits
for the components.

1849

0.020

o
~-0.010
I

~ - 0.020

tJ -0.030

-0.040

I:;

-0.050

it:

a:
iii _ 0.060
-0.070

MOLE PERCENT LI20 IN THE SOLID

FIG. 4. Variation of birefringence as a function of solid composition across the existence regions of lithium niobate and
lithium tantalate.

The shift of birefringence with composition in such

None of these explanations accounts for the relatively
wide existence regions found in lithium niobate and extensive solid-solution systems is of interest. Figure 4
lithium tantalate. The last two explanations are in- shows the experimental data of Bergman et aF for
adequate because they are not supported by the known lithium niobate and Ballman et al. 6 for lithium tantalate
phase diagram for lithium niobate. 2 However, closer plotted as a function of solid composition. The differstudy of the first criterion shows that it is too restrictive. ence in the absolute values between LiNb0 3 and
Although it is true that native disorder alone determines LiTa03 is probably associated with the more nearly
the curvature of the free-energy curve in simple binary octahedral geometry of the TaOe units as is suggested by
salts, the existence of complex anion or cation groups in NMR results on this solid solution series.27 However the
both the melt and the crystal may also provide a differences with composition in each material are simply
substantial contribution. This occurs because of the not understood at the present time.
relatively large contribution of the covalent bonding in
Crystal Growth
such ionic groups to the total lattice energy. The net
result is that charge compensation mechanisms may
The phase equilibria of Figs. 1 and 3 and Table I
occur with a relatively lower expenditure of energy. In serve as the basis for understanding some of the crystallithium niobate and tantalate, the NbO e and TaOe growth problems of lithium niobate. We have been
octahedra have been shown by Peterson and Briden- primarily concerned with compositional variations on
baugh 19 as well as Lines 24 to possess a substantial degree various scales throughout the crystal. Such composiof covalency. Hence these lattices can accommodate tional variations lead to undesirable birefringence varialithium vacancies at a small cost in energy, i.e., the tions through the relations shown in Fig. 4. It is seen
curvature of the free-energy curve (with composition)
that such variations in tJ.n are much larger for lithium
is small and the minimum is displaced towards the niobate than for lithium tantalate both because of the
Nb20,-rich region. A similar behavior has been observed larger absolute value of tJ.n as well as its greater conin the solid-solution range of barium sodium niobate 23 centration dependence.
which shows a large solubility in the Nb 20,-rich region
There are three major types of compositional variabut almost none in the BaO- or Na20-rich regions. The tions in these crystals, the normal freezing variation
location of the congruently melting composition is which results over the length of the crystal when the
primarily influenced by the location of the minimum in melt and crystal compositions are not equal, the
the free-energy curve of the solid phase. Thus in solids striations which occur due to fluctuating growth rates,
with complex cation or anion groups, the congruently and the radial variations which occur when the solid-melting composition may be expected to be always liquid interface possesses facets or the diffusion boundsubstantially displaced from the low-temperature ary layer across the interface is not uniform. These
stoichiometric composition. This shift is 1.4% Nb2 0" for problems should all disappear if the crystals are grown
lithium niobate, 1.0% Ta205 for lithium tan tala tell and from melts of the congruent composition and other
about 4% BaNb20 e for barium sodium niobate. 25 ,26
impurities are absent. However since other crystal

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
14.139.220.33 On: Wed, 01 Jul 2015 14:34:49

1850

CARRUTHERS ET AL.

CONGRUENTLY
MELTING
COMPOSITION

49.0

t-

",0
0",

~':i

Sat

3o i 4801---,(
C/)

TWO PHASE REGION AT

ROOM TEMPERATURE
UNDER EQUILIBRIUM
CONDITIONS

48.0

50.0

52.0

54.0

56.0

58.0

LIQUID

COMPOSITION
(mo "10 Li20)

FIG. 5. Variation of the solid composition as a function

of liquid composition across the existence region of lithium
niobate.

compositions may be desired, it is necessary to examine

the implications of growth at noncongruent compositions.
The compositional relationships which pertain to the
crystal growth are shown in Fig. 5, where C. and CL are
the solid and liquid concentrations in mole percent Li20
which are in "equilibrium" at a given temperature.
These data were obtained by the NMR and Curietemperature techniques from crystals grown generally
at 0.5 in.jh and 30 rpm. In this range the effective and
equilibrium distribution coefficients are expected to be
nearly equal; further experiments are in progress to
clarify the point. The distribution coefficient is calculated from this curve as the ratio of solid and liquid
compositions in excess of the congruently melting
composition as described by Hude, Jones, and Mullin. 28
The growth of stoichiometric crystals is of great
interest because the phase matching temperatures of
these crystals for the laser wavelengths of interest are
above the temperature at which the crystal is no longer
damaged. 29 The growth of stoichiometric crystals can be
accomplished from a melt composition of about 58.0
mole% Li20. Constitutional supercooling may occur
however unless precautions are taken. The minimum
temperature gradient in the liquid at the interface
(G L ) min which will avoid constitutional supercooling is
approximately:
(Gdmin= (-mRCdD)[r+ exp( -Ra/D) J-I,

where

o
D

liquidus slope
bulk liquid concentration (relative to 48.6%
Li20)
growth rate
diffusion boundary layer thickness at rotating
crystal interface
liquid diffusion coefficient
ko/ (1-k o).

Using values of o=5XlO-a cm and D=5XI0-5

cm2/sec, we obtain the results shown in Fig. 6 for
various melt compositions. It is obvious that very low
growth rates are required to avoid constitutional undercooling, especially for a low-temperature gradient. For
instance, with GL = 5C/ cm, we require R < 2.5 X 10-6
cm/sec (0.0036 in./h). Higher temperature gradients
will allow faster growth rates but will also introduce
striations due to irregular thermal convection and increase the susceptibility to cracking due to thermal
stress. The values of GL and R required to avoid constitutional supercooling during growth from Li 20-rich
melts are such that in practice many crystals grown
from such melts may have undergone cellular breakdown.
It is possible to estimate the magnitude of the
refractive index (or birefringence) changes across
compositional striations caused by growth-rate fluctuations. These calculations are based on the idea that
such compositional variations arise from stoichiometric
deviations and not from any impurities present in the
lithium niobate. It will be seen that the calculated
values agree with experiment and thus verify this
explanation for the origin of the striations. There are a
number of causes of fluctuating growth rate; vibrations,
melt temperature control, ambient temperature control,
thermal convection in the melt, and rotation of the
crystal through an asymmetrical temperature gradient.
Under good growth conditions, only the last two factors
contribute to unsteady growth. For typical growth
conditions, rotational striations are spaced 4-6-J.L apart
while thermal convection striations are 20-40-J.L apart.
The thermal convection temperature fluctuations
typically have a 1O-20-sec period. The birefringence
change ~(on) can be related to the temperature
fluctuation amplitude ~T by
~(on) =[~(on)/~C.J(~Csj~R) (~Rj~T)~T,

where C. is the solid composition and R the growth rate.

For LiNbO a, ~(on)j ~C.=O.Ol1/mole% LbO from Fig.
4. For time periods greater than o2jD=0.5 sec the

60"lo~%55%

200

54%

52%

180

E
u

....

160
140
120

.!: 100
E 80

-:,
!:!

50
40
20

10

20

30
40
50
R.10 5 (em/sec)

60

70

FIG. 6. Minimum liquid temperature gradient required to

avoid constitutional supercooling for various growth rates and
liquid compositions in lithium niobate.

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
14.139.220.33 On: Wed, 01 Jul 2015 14:34:49

NONSTOICHIOMETRY OF

second term may be obtained from the Burton, Prim,

and Slichter relation as

f1C.

= _

((a/D) (I/r) exp( -Ro/D)) C

L

[l+(l/r) exp(-Ro/D)J2
1.4% Li20 (corresponding to 50% Li 20 on the

f1R

For CL =
phase diagram) then f1C./f1R= -5.0 mole%/(cm/sec)
at R = 3.5 X 10-4 cm/ sec. The third term has been given
by Brice and Whiffen so as
f1R/f1T=KL/OTHf,

where KL is the liquid thermal conductivity, OT is the

thermal boundary layer thickness, and HI is the heat of
fusion. Using the values KL = 10-2 cal/ (cm secOC),
OT = 0.2 cm (for 30 rpm crystal rotation), and HJ = 527
cal/cms (from Pankratz and King 22 ), we obtain
f1R/f1T=10- 4 (cm/sec);oC.

Using these quantities, we obtain

f1(on) =S.SXlO-6 f1T(OC).

Thus growth from a stoichiometric melt in the presence

of temperature fluctuations of 1-10C amplitude and
periods of over 0.5 sec will cause birefringence variations
of 5.5 X lO- L S.SX 10-5. Higher frequency fluctuations
will give lower f1 (on) values. Birefringence variations
across striations have been measured by Nash et af.!;
and found to be in the predicted range. On this basis, the
reduction of the intensity of striation variations can be
achieved by growth in low-temperature gradient
furnaces. However, low-temperature gradients are not
favorable if constitutional supercooling must be
avoided so that some compromise is necessary.
An alternative solution to the problem of the lowphase matching temperature of congruent lithium niobate has recently been discovered by Bridenbaugh
et al. S! Beginning with the knowledge that MgO
additions raised the Curie temperature as shown in
Fig. 1, it was thought that the phase matching temperature would be similarly increased. This increase was
indeed observed along with The added advantage that
Mg2+ substitutes for Li+ at the congruent Li/Nb ratio
in such a way as to keep the phase-matching temperature virtually uniform over the length of the
crystal. Thus the spatial uniformity of the congruent
composition has been retained while allowing operation
above the damage-anneal temperature of lithium
niobate.
SUMMARY

Details of the nonstoichiometric shifts in composition

that occur during the crystal growth of lithium niobate
have been studied. The congruently melting composition

LITHIUM

NIOBATE

1851

is located at 48.6 mole% Li 20 and the solid-solution

region extends from 48% Li20 to 50% Li20. The
implications of growth at non congruently melting
compositions on the compositional uniformity of lithium
niobate have been discussed in some detail.
ACKNOWLEDGMENTS

We wish to thank F. R. Nash and R. A. Laudise for

many discussions of this work.
1 B. T. Matthias and J. P. Remeika, Phys. Rev. 76, 1886
(1949) .
2 A. Reisman and F. Holtzberg, J. Amer. Chern. Soc. 80, 6503
(1958) .
3 A. A. Ballman, J. Amer. Ceram. Soc. 48,112 (1965).
4 K. Nassau, H. J. Levinstein, and G. M. Loiacono, Appl.
Phys. Lett. 6, 228 (1965).
5 K. Nassau, H. J. Levinstein, and G. M. Loiacono, J. Phys.
Chern. Solids 27,983,989 (1966).
6 A. A. Ballman, H. J. Levinstein, C. D. Capio, and H. Brown,
J. Amer. Ceram. Soc. 50,657 (1967).
7 J. G. Bergman, A. Ashkin, A. A. Ballman, J. M. Dziedzic,
H. J. Levinstein, and R. G. Smith, Appl. Phys. Lett. 12, 92
(1968) .
8 H. Fay, W. J. Alford, and H. M. Dess, App!. Phys. Lett. 12,
89 (1968).
9 P. Lerner, C. Legras, and J. P. Dumas, J. Cryst. Growth 3,4,
231 (1968).
10 G. E. Peterson and J. R. Carruthers, J. Solid-State Chern. 1,
98 (1969).
11 R. L. Barns and J. R. Carruthers, J. App!. Cryst. 3, 395
(1970) .
12 G. E. Peterson, A. A. Ballman, P. V. Lenzo, and P. M.
Bridenbaugh, App!. Phys. Lett. 5, 62 (1964).
13 E. Schempp, G. E. Peterson, and J. R. Carruthers, J. Chern.
Phys. 53, 306 (1970).
14 G. E. Peterson and J. R. Carruthers (unpublished).
15 F. R. Nash, G. D.-Boyd, M. Sargent III, and P. M. Bridenbaugh, J. App!. Phys. 41, 2564 (1970).
16 R. L. Byer, J. F. Young, and R. S. Feigelson, J. App!. Phys.
41,2320 (1970).
17 K. A. Jackson, Progr. Solid-State Chern. 4, 53 (1967).
18 B. Cockayne, M. Chesswas, and D. B. Gasson, J Mater. Sci.
4,450 (1969).
19 G. E. Peterson and P. M. Bridenbaugh, J. Chern. Phys. 48,
3402 (1968).
20 G. E. Peterson, P. M. Bridenbaugh, and P. Green, J. Chern.
Phys. 46, 4009 (1967).
21 K. Nassau, in Ferroelectricity, edited by E. F. Weller (Elsevier,
Amsterdam, 1967), p. 259.
22 L. B. Pankratz and E. G. King, U.S. Bureau of Mines, Report
of Investigation No. 6862 (1966).
23 W. Albers and C. Haas, Philips Tech. Rev. 30, 83, 107, 132
(1969) .
24 M. E. Lines, Phys. Rev. (to be published).
25 J. R. Carruthers and M. Grasso, Mater. Res. Bull. 4, 413
(1969) .
26 R. L. Barns, J. Appl. Cryst. 1, 290 (1968).
27 G. E. Peterson, J. R. Carruthers, and A. Carnevale, J. Chern.
Phys. 53,2436 (1970).
28 D. T. J. Hurie, O. Jones, and J. B. Mullin, Solid-State
Electron. 3,317 (1961).
29 A. Ashkin, G. D. Boyd, J. M. Dziedzic, R. G. Smith, A. A.
Ballman, H. J. Levinstein, and K. Nassau, App!. Phys. Lett. 12,
186 (1968).
30 J. C. Brice and P. A. C. Whiffen, Brit. J. App!. Phys. 18, 581
(1967) .
31 P. M. Bridenbaugh, J. R. Carruthers, J. M. Dziedzic, and
F. R. Nash, App!. Phys. Lett. 17, 104 (1970).

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
14.139.220.33 On: Wed, 01 Jul 2015 14:34:49