CA1302848C - Process for making homogeneous lithium niobate - Google Patents

Abstract

PROCESS FOR MAKING HOMOGENEOUS
LITHIUM NIOBATE

Abstract Lithium niobate crystals are made by pulling a crystal boule out of a melt of lithium niobate at high temperatures. Use of a congruent melt composition is highly advantageous because it yields uniform properties and composition throughout the crystalline boule. The generally accepted congruent melt composition is 48.6 mole percent lithium determined in 1971 from Curie temperature measurements. The invention is based on the discovery that the congruent melting composition is significantly different from that previously accepted and involves growth of lithium niobate crystals using this corrected congruent melting composition.

Description

~L3(~Z~8 PROCESS FOR MAKING HOMOGENEOUS
LITHIUM NIOBATE

Technical Field The invention is a process for making lithium niobate crystals and for making devices comprising lithium niobate crystals.
Background of the Invention Lithium niobate is used extensively in a variety of applications both for its unique electrical and dielectric properties and the comparative ease in fabricating large single crystals of high quality.
Typical uses are in surface acoustic wave devices and as substrates in integrated optical devices. A low dielectric constant combined with a large acoustic velocity makes lithium niobate attractive for surface acoustic wave devices~ and a moderately large electro-optic effect together with a high Curie temperature makes it attractive for various optical devices.
Single crystal lithium niobate is generally made by pulling from a melt of lithium niobate typically by the so-called Czochralski prccedure. Early work on the growth of lithium niobate showed that the stoichiometric composition does not melt congruently.
Congruent melting is where the composition of the melt and solid in equilibrium with the melt is the same.
The congruent melting composition is of considerable importance in the growth procedure for crystals like lithium niobate. At this melt composition, the composition of the pulled crystal does not change along its length, and a crystal of constant composition is obtained. Such a crystal has identical optical and acoustical properties throughout its length.

~32~Z84~

For this reason, considerable effort has been ~ade to determine the congruent melting composition of lithium niobate. Early work by Lerner et al in a paper entitled "Stoechiometrie des Monocristaux de Metaniobate de Lithium," Journal of Crystal Growth, 3, 4 (1968), pages 231-235, established solubility li~its for single phase lithium niobate at about 1200 degrees C. At this temperature, the lithium content may vary from 44.1 mole percent Li2o ~o 50.1 mole percent Li2o in a single pha~e LiNbO3, where 50.0 mole pa~cent is the stoichiometric composition.
A word about nomenclature would be useful at this point. Traditionally, and in this application, it is regarded that LiNbO3 is made up of 50 mole percent Li2o and 50 mole percent Nb2o5. Deviations from stoichiometry are expressed in deviations from 50 mole percent Li2o.
Lerner et al in the reference set forth above also carried out lattice parameter measurements and differential thermal analysis measurements in an effort to locate the phase boundaries in the phase diagram and the transition temperatures for different compositions.
These measurements also indicated that the congruent melting composition lay between 48 and 49 mole percent Li2o.
Because of the importance of the congruent melt co~position, further studies were carried out on the phase diagram of the lithium oxide, niobium oxide system. A particularly detailed study was carried out by Carruthers et al and reported in a paper entitled "Nonstoichiometry and Crystal Growth of Lithium Niobate," Journal of Applied Physics, 42, No. 5, April 1971, pages 1846-1851. They reviewed the literature on growth and properties of lithium niobate crystals and the effect of crystal composition on the properties (particularly optical properties) of lithium niobate crystals.

13(~348 Carruthers et al also carried out a series of measurements to establish the congruent melting composition. These measurements involved determining the Curie temperatures for a series of ceramic compositions and comparing these results with the Curie temperature obtained from single crystals grown from melts of known composition. By comparing the Curie temperature vs. composition behavior of hoth the ceramic samples and single crystals grown from the melt, the congruent melting composition was determined. This was found to be 48.6 mole percent Li2o. ~lso determined was the solid solution range at 1020 degrees C. This was found to be 46.0 to 50.0 mole percent Li2o. Data taken by these authors yielded other valuable information concerning the equilibrium behavior of lithium niobate including the Curie temperature as a function of composition close to the stoichiometric composition of lithium niobate, distribution coefficients between liquid and solid lithium niobate in equilibrium and the phase diagram of lithium oxide and niobium oxide near the stoichiometric composition of lithium niobate.
Strong support for the congruent melting composition was obtained from a number of scientific investigations published in the scientific literature.
For example, F. R. Nash et al in a paper entitled "Effect of Optical Inhomogeneities on Phase Matching in Nonlinear Crystals," Journal of Applied Physics, Vol. 41, No. 6 ~May 1970), pages 2564-2576, measured the phase matching condition for second harmonic generation for various pieces of lithium niobate along the pulled axis. This was related to variations in the extraordinary index of refraction of the crystals and variations attributed to composition variations along the length of the crystals.
By measuring the birefringence of crystals along the growth direction, they concluded that the congruent melt composition was 48.6 mole percent Li2o.

~3(~84~3 Similar measurements by R. L. Byer et al, described in a paper entitled "Growth of High-Quality LiNbO3 Crystals from the Congruent Melt," Journal of Applied Physlcs, Vol. 41, No. 6 (May 1970), pages 2320-2325. The Carruthers et al study together with the Nash and Byer works established that the congruent melting composition was 48.6 mole percent Li2o. A paper by K. Chow et al ("The Congruently Melting Composition of LiNbO3,~
Material Research Bulletin, Volume ~, pages 1067-1072, 1974) shows further evidence that 48.6 mole percent Li2o is the congruent melting composition of lithium niobate.
Because of the above-described investigations, it became well established that the congruent melting composition of lithium niobate was 48.6 mole percent Li2o and crystal growers and others involved in the fabrication of crystals used this composition in the fabrication of lithium niobate crystals.
Summary of the Invention The invention is based on the amazing discovery that the congruent melting composition is significantly different from that previously accepted and used. The invention is a process for growing lithium niobate crystal by pulling from the melt in which the melt composition is 48.45 mole percent Li2o.
Although accurate control of the melt composition is highly desirable to ensure uniform composition throughout the grown crystal, variations up to 0.08 mole percent are tolerable, but variations less than + 0.04 or even + 0.02 are much preferred. The inventive procedure yields excellent lithium niobate crystals with highly uniform, constant composition and with uniform, constant properties such as index of refraction, dielectric constant, Curie temperature, nonlinear optical properties, birefringence, etc. The invention includes optical devices comprising lithium niobate grown in accordance with the invention. Such devices have unusually good properties because of the 13(~84~3 known and uniform composition of the lithium niobate crystals and uniform optical properties. Particularly useful are optical integrated circuits with lithium niobate substrates.
In accordance with one aspect of the invention there is provided a process for fabricating a device comprising at least one lithium niobate crystal in which the lithium niobate crystal is made by solidifying said crystal from a melt in which the melt consists essentially of molten lithium niobate, said molten lithium niobate consisting of 48.45 + 0.08 mole percent Li2o, remainder Nb2Os.
In accordance with another aspect of the invention there is provided a device comprising a lithium niobate crystal in which the lithium niobate crystal is made by solidifying said crystal from a melt in which the melt consists essentially of molten lithium niobate, said molten lithium niobate consisting of 48.45 + 0.08 mole percent Li2o, remainder Nb2O5.
Brief Description of the Drawing FIG. 1 shows an apparatus for pulling a crystal from the melt;
FIG. 2 shows data in graphical form of the distribution coefficient between crystal and melt as a function of melt composition;
FIG. 3 shows diagrams of structures used in the fabrication of optical integrated circuits; and FIG. 4 shows an optical integrated circuit with lithium niobate substrate made in accordance with the invention.
Detailed Description The invention is a process for fabricating lithium niobate crystals from a melt of lithium and niobium oxides in which the concentration of lithium expressed as mole percent of Li2o is 48.45 + 0.08. Although the melt is generally composed of high purity oxides of lithium and niobium (for example, less than 0.1 mole percent impurity or even less than 0.05 or even 0.03 mole percent impurity), small amounts (typically up to l.0 mole percent) of other elements (besides ~if ~

`` ~3~28~8 5a lithium and niobium) may be present to adjust various properties such as Curie temperature, index of retraction, etc.
The invention applies to any process for making crystalline lithium niobate involving equilibrium between congruently melted lithium niobate and solid lithium niobate.
Generally, lithium niobate is made by a crystal pulling technique from molten lithium niobate. This crystal pulling technique is often called the Czochralski technique and involves introducing a seed of lithium niobate into the lithium niobate melt and slowly pulling it out usually with rotation. A paper by C.D. Brandle and A.J. Valentino entitled "Czochralski :t 3~2841g Growth of Rare Earth Gallium Garnets," Journal of Crystal Growth, 12 (1972), pages 3-8, describe~ a typical crystal pulling apparatus and process.
FIG. 1 shows a typical crystal pulling apparatus 10 useful in the practice of the invention.
The apparatus comprises heat source ~rf coils 11), platinum crucible 12 containing the lithium niobate melt 13, and the lithium niobate crystal 14 being pulled. The platinum crucible 12 is equipped with a platinum after heater 15 and platinum lid 16 to control temperature gradients in the pulling apparatus. There is also considerable heat insulation comprising Zr2 base 17, ZrO2 supports 18, ZrO2 disk 19, ZrO2 tubing 20, ZrO2 grog 21, ZrO2 lid 22, ZrO2 felt 23, platinum reflector 24, and quartz sleeve 25. The initial melt level 26 is shown which decreases as the lithium niobate crystal 14 is pulled out of the melt 13.
A rod 27 sticking through the top of the apparatus is attached to the seed 28, provides support for the lithium niobate crystal 14, and often provides for rotation of the crystal.
An important aspect of the invention is the determination of the congruent melting composition. The procedure for determining the congruent melt composition involves accurate, precise determination of the Curie temperature (TC) for various known compositions of the lithium niobate crystal. The Curie temperatures were determined using precision differential thermal analysis (DTA~ measurements and the crystal compositions by combining known amounts of analyzed raw materials.
Melts whose initial composition was established in this way were sampled before and after crystal growth. Since the top and bottom sections of the grown crystal boules were in equilibration with the melt, distribution coefficients between melt and crystal were also determined in this way.

~3~JZ8~8 A Perkin-Elmer DTA Model 1700 was u~ed in conjunction with a Fluke Model 2452 based data acquisition system to obtain the DTA data. These measurements were carried out in the temperature range of 1000-1200 degrees C. Heating and cooling rates were 20 degrees C per minute, and the measurements were carried out in a flowing nitrogen atmosphere. Aluminum oxide was used as a reference material, and platinum crucibles were used to contain the materials. The melting curve of gold was used to calibrate the apparatus.
Crystals used in this study were grown by the Czochralski technique. ~oth starting materials (Li2Co3 and Nb2o5) were 99.999 percent pure and obtained from Johnson Matthey, Inc. Prior to preparation of the charge, each of the constituent oxides was dried at 500 degrees C for a period of 12 to 14 hours. After drying, the starting powders were weighted and thoroughly mixed. All weights were accurate to + 0.01 grams. The mixed powder was then reacted at 1000 degrees C for 12-14 hours. Weight loss was checked after firing to ensure complete decomposition of the lithium carbonate. The reacted powder was then pressed into a pellet for loading into the crucible. In all cases, the final reacted, pressed powder pellet weighed 240.0 + 0.1 grams.
The pressed LiNbO3 pellet was then loaded into the furnace (rf heated) which is shown schematically in FIG. 1 and described above. The molten charge (240 grams) was contained in a 2 inch D x 2 inch Pt crucible. The growth parameters used for these tests are shown in Table I. After the charge had completely melted and prior to growth, the liquid was stirred for several minutes to ensure complete mixing. Earlier tests had shown that complete melting of the charge did not necessarily ensure complete liquid homogeneity.
After stirring, an initial liquid sample was obtained by ~3~2~348 quickly inserting and withdrawing from the liquid a polycrystalline A12o3 rod. This process was repeated two or three times in order to obtain approximately a 0.5 gra~ sample. This sample then served as a composition standard for Tc measurements-13t~;~848 g TABLE I

Crystal Growth Parameters Growth Direction "C" axis (0001) Pull Rate 0.15"/hr (0.38 cm/hr~
Rotation Rate 15 RPM
Atmosphere Air Crucible 2" OD x 2" Ht Pt Initial Charge 240.0 grams Typical Crystal Wt. 150-160 grams Crystal Length ~10 cm Crystal Diameter 20-21 mm ~3~T;21348 After growth, a final liquid sample was taken using a similar procedure as in the case of obtaining the initial liquid sample. However, in this case, the furnace was cooled to room temperature (to remove the crystal) and then the crucible and residual melt reheated. As in the initial sample case, the liquid was again stirred to ensure complete mixing prior to obtaîning a sample.
Single crystal samples were taken from the top and bottom of each boule. The initial crystal sample which was used for Tc determination was typically 0.5 mm thick and 20 mm in diameter and weighed approximately 0.7-1 gram. In all cases, g (the fraction of liquid solidified) for the initial crystal sample was <0.01.
The final cry~tal sample was similar in size and always had one surface which was the growth interface. The Tc of the top and bottom slices of a boule was determined by DTA.
The results of the Tc measurements cover the range from 47.0-49.0 mole percent Li2o in the liquid, with the majority of samples in the 48.0-49.0 mole percent range. A linear least squares fit of this data yields the following dependence of Curie temperature on composition:

Tc Z -637.30 + 36.70 C

or C = 17.37 + 0.02725 Tc where Tc is the Curie temperature in degrees C, and C is the Li2o content in mole percent of the sample. Close examination of the composition vs. Curie temperature data shows some nonlinearity as one approaches higher Li2o liquid compositions. A least squares fit of the data using the form Tc = a + bC + cC2 yields ~3~ 4~3 Tc = 9095.2 - 369.05 C + 4.228 c2 which gives a better fit than the linear model.
Using this relationship and measuring the Tc of the final liquid sample, the crystal top and the crystal bottom, the Li2o content of each sample was determined. If one assu~es that mixing in the liquid is complete and that diffusion in the solid is negligible, then for each sample pair (solid-liquid]:

ko = keff = Cl where ko(keff) is the distribution coefficient of Li2o while Cs and Cl are the concentration of Li2o in the solid and liquid, respectively.
The results for each of the various crystal growth tests are shown graphically in FIG. 2 and are summarized in Table II. FIG. 2 is a graph showing the distribution coefficient between crystal and melt as a function of melt composition. Table II shows the results for some samples with initial compositions between 47.00 and 49.00 mole percent. Based on these present result~, the congruent composition (ko = 1.000 for LiNbO3 is indicated to fall between 48.4 and 48.5 mole percent I,i2o.

~3t~2848 T~BLE II

CrYstal Growth Results Composition (Mol. % Li2o) Distribution Liquid Solid Coefficient ~k) 5 Initial Final Initial Final Initial Final 47.00 45.45 48.03 47.44 1.0219 1.0438 48.00 47.44 48.30 48.10 1.0063 1.0139 ~8.20 47.73 48.41 48.19 1.0044 l.0096 48.40 48.21 48.45 48.34 1.0010 1.0027 48.60 48.80 48.49 48.51 0.9977 0.9941 48.80 49.32 48.56 4B.61 0.9951 0.9856 49.00 49.86 48.55 ~8.70 0.9908 0.9767 In order to demonstrate a congruent melt composition, an initial melt which contained 48.45 mole percent Li2o was prepared and a crystal was grown. For this growth, the fraction of liquid crystallized was 0.72. DTA heating curves for the initial melt, final melt, crystal top, and crystal bottom were taken. The data are summarized in Table III. The Tc's for the four different samples when corrected for the Au calibration are identical within + 2 degrees C. Therefore, the congruent composition is 48.45 mole percent Li2o, and its Tc is 1137 degrees C.

~3~

TABLE I I I

~8.45 Mole Percent Li2o Sample Curie Temperature (degrees C) !

Initial Liquid 1138 Final Liquid 1137 Crystal Top 1140 Crystal Bottom 1137 The Tc of Li-rich LiNbO3 compositions cannot be determined by DTA because the solidus lies below the Tc, and the enthalpy of melting masks the smaller Tc heat change. To determine the composition at the LiNbO3-Li3NbO4 boundary, the measured weight gain for full lithiation is converted to mole percent Li2o and added to the initial crystal composition as determined by DTA. These measurements yielded the following lithium niobate compositions at the phase boundary between LiNbO3 and Li3NBO4: 49.96 + 0-03 mole percent Li2o at 1060 degrees C, 49.89 + 0.03 mole percent at 1100 degrees C and ~9.81 + 0.03 mole percent Li2o at 1150 degrees C.
; Because crystal slabs formed from boules grown from congruent melting composition have identical compositions, they are more easily processed into a variety of device elements. Particularly useful are substrates for optical integrated circuits where identical processing can be used to produce a large number of devices.
A typical procedure for making such an optical waveguide circuit is as follows. The optical waveguide is made by diffusing titanium into the lithium niobate substrate. The titanium pattern is made by a typical photoresist procedure. ~ photoresist layer is put down on a lithium niobate as shown in FIG. 3A. Photographic ~3(?;~l348 techniques are used to pattern the photoresist by opening holes where waveguide is to be fabricated (see FIG. 3B). Deposition of titanium is then carried out to cover the entire surface of t~ia substrate (FIG. 3~). On S removal of the photoresist material, a pattern of titanium metal is left on the surface (~IG. 3D). In order to form the integrated optical circuit, a diffusion process is carried out to diffuse the titanium metal into the lithium niobate substrate (FIG. 3E). To ensure that the composition and homogeneity of the lithium niobate substrate remains unchanged during the titanium diffusion, the diffusion process is preferably carried out in a closed container with the lithium niobate in close proximity to a powder bed whose lithium activity is the same as the lithium niobate substrate.
Generally, the temperature in the titanium diffusion process is governed by the requirements of the titanium diffusion process since the powder bed is present to maintain the composition and homogeneity of the lithium niobate substrate. The diffusion is typically carried out in the temperature range from 1050-1100 degrees C
for 1-10 hours.
A typical optical integrated circuit 40 is shown in FIG. 4. Here, the particular optical integrated circuit 40 is a directional coupler. This device is made up of a substrate 41 of lithium niobate grown from the melt in accordance with the invention.
Optical waveguide sections 42 and 43 are also evident and typically made by the diffusion process described above. These waveguide sections extend along the substrate and out the other end. Electrical electrodes 44 and 45 are also present to affect coupling between the optical waveguides. In one mode of operation, light is introduced into waveguide 43 and in the absence of applied voltage of the electrodes, exits waveguide 42. On applying a proper potential to electrode 45, through terminal 46 and resistor 47, the 13~Z~48 light remains in and exits waveguide 43. Various other integrated optical circuits can be made in accordance with the invention. The principal advantage of substrates made in accordance with the invention is that substrates with identical composition and identical properties (such as optical properties) are always obtained, and processing can be adjusted to optimize performance for all devices.

Claims (23)

1. In a process for fabricating a device comprising a substrate of a lithium niobate crystal, in which the lithium niobate crystal is made by solidifying said crystal from a melt consisting essentially of molten lithium niobate, and the crystal is grown by pulling from the melt, the improvement which comprises preparing the melt so that said molten lithium niobate includes 48.45 ? 0.04 mole percent Li2O, remainder being Nb2O5 and less than 0.1 mole percent of impurities if any present, and wherein the Curie temperature, Tc, of the crystal grown from the melt, at least at an upper and a lower part of the crystal, is equal to 1138 ? 2°C.

2. The process of claim 1, in which said Li2O is present in the melt in an amount of 48.45 ? 0.02 mole percent.

3. The process of claim 1, in which the composition of the melt at least prior to said pulling is checked by withdrawing a melt sample, solidifying the sample, and determining whether or not the Curie temperature, Tc, of the solidified sample is 1138 ? 2°C, said crystal growing step being carried out only if said Tc is equal to 1138 ? 2°C.

4. The process of claim 1, in which the molten lithium niobate is stirred prior to said pulling step so as to ensure complete mixing of said melt.

5. The process of claim 1, wherein said device is an optical waveguide, and in which process at least one waveguide pattern is formed in said substrate.

6. The process of claim 5, wherein said optical waveguide is produced by forming a titanium metal pattern on the lithium niobate substrate and diffusing titanium into the lithium niobate substrate.

7. The process of claim 1, in which said melt additionally includes up to 1 mole percent of intentional additives.

8. A process of growing lithium niobate crystals having congruent composition suitable for use as a substrate in a device, which comprises preparing a melt consisting essentially of lithium oxide (Li2O), niobium oxide (Nb2O5) and less than 0.1 mole percent of impurities, and growing a lithium niobate crystal by pulling the crystal from the melt, wherein the melt consists of 48.45 ? 0.04 mole percent lithium oxide, the remainder being niobium oxide and less than 0.1 mole percent impurities, and the Curie temperature, Tc, at least at an upper and a lower part of the crystal grown from the melt, is equal to 1138 ? 2°C.

9. The process of claim 8, in which the melt includes 48.45 ? 0.02 mole percent Li2O.

10. The process of claim 8, in which the composition of the melt at least prior to said pulling is checked by withdrawing a melt sample and determining whether or not the Curie temperature, Tc, of the solidified sample is 1138 ? 2°C, said growing step being carried out only if said Tc is equal to 1138 ? 2°C.

11. The process of claim 8, in which the melt is stirred prior to said pulling step so as to ensure complete mixing of said melt.

12. The process of claim 8, wherein said device is an optical waveguide having at least one waveguide pattern formed in the substrate.

13. The process of claim 12, wherein said at least one waveguide pattern is produced by forming a titanium metal pattern on the lithium niobate substrate and diffusing titanium into the lithium niobate substrate.

14. The process of claim 8, in which said melt additionally includes up to 1 mole percent of intentional additives.

15. The process of claim 8, in which said crystal is grown using a major portion of the melt.

16. The process of claim 15, in which said crystal is grown using a 0.72-th fraction of the melt.

17. The process of claim 1, in which said crystal is grown using a major portion of the melt.

18. The process of claim 17, in which said crystal is grown using a 0.72-th fraction of the melt.

19. A process for fabricating a device comprising an optical waveguide having at least one waveguide pattern formed in a substrate of a lithium niobate crystal in which the lithium niobate crystal is made by solidifying said crystal from a melt consisting essentially of molten lithium niobate, and the crystal is grown by pulling from the melt, wherein the melt is prepared so that said molten lithium niobate includes 48.45 ? 0.04 mole percent Li2O, remainder being Nb2O5 and less than 0.1 mole percent of impurities if any present, and wherein the Curie temperature, Tc, of the crystal grown from the melt, at least at an upper and a lower part of the crystal, is equal to 1138 ? 2°C.

20. The process of claim 19, in which said Li2O is present in the melt in an amount of 48.45 ? 0.02 mole percent.

21. The process of claim 19, in which the composition of the melt at least prior to said pulling is checked by withdrawing a melt sample, solidifying the sample, and determining whether or not the Curie temperature, Tc, of the solidified sample is 1138 ? 2°C, said crystal growing step being carried out only if said Tc is equal to 1138 ? 2°C.

22. The process of claim 19, in which the melt is stirred prior to said pulling step so as to ensure complete mixing of the melt.

23. The process of claim 19, in which said at least one waveguide pattern is produced by forming a titanium metal pattern on the lithium niobate substrate and diffusing titanium into the lithium niobate substrate.