JPH0324438B2 - - Google Patents

Description

[Detailed description of the invention]

[Technical field of the invention] This invention relates to surface acoustic waves (hereinafter abbreviated as SAW).
This article relates to the manufacturing method of lithium tantalate single crystal, which is useful as a device substrate. In particular, lithium tantalate (lithium tantalate), which is used as a substrate for narrow-band SAW filters and SAW resonators, where compositional fluctuations in the crystal substrate have a sensitive effect on device yield. (hereinafter abbreviated as LT)
This invention relates to a method for producing a single crystal. [Technical background of the invention and its problems] The LT single crystal manufacturing method is usually the Czyochralski method, which is a method in which raw materials are filled in a crucible, heated and melted, and pulled up while rotating using a seed crystal. Manufactured. It is generally well known that as the raw material composition in this case, it is better to use a congruent composition, which is defined in terms of crystal engineering so that the melt composition and crystal composition match, as shown in FIG. 1. However, the true congruent composition is ±1%
It is difficult to determine exactly what follows.
In particular, in the case of chemically stable LT, it is difficult to decompose it into Li ₂ O and Ta ₂ O ₅ and conduct chemical analysis with high precision, and in the past, many researchers estimated LT from the raw material composition.
The congruent composition of Li/Ta molar ratio is 0.95~
have reported that it is between 0.96 (RLBarns & J.
R.Carruthers: J.Appl Cryst. 3 (1970) 395)
(Miyazawa; Tsuken Practical Report 21 (1972) 1742). Therefore, LT single crystal production to date has been carried out using raw materials having the above molar composition. However, there has been a problem in that there are large variations in device manufacturing yield when SAW devices are fabricated using the manufactured single crystal substrates. In particular, narrowband SAW filters and SAWs whose crystal composition has a sensitive effect on SAW characteristics.
Manufacturing yields for devices such as resonators have varied significantly. In other words, in the case of LT single crystal production, a predetermined weight of LT crystal powder with the above molar composition ratio is filled into a noble metal growth crucible, and crystal growth is performed using the normal rotational pulling method. The sintered powder or grown crystal mass is filled with the weight of the crystal, and the pulling growth is performed again. Therefore, if the raw material composition differs from the true congruent composition, the crystal composition will gradually change as the number of consecutive pulls increases, and the SAW characteristics fluctuations within and between crystal lots will increase, which will reduce the device manufacturing yield. This is thought to cause a decline in Convenient methods for reducing such fluctuations include limiting the number of consecutive pullings to a few times, and reducing the solidification rate, that is, the ratio of the pulled crystal weight to the charge weight. However, these methods not only reduce the productivity of single crystal production, but also reduce the life of the crucible.
This is not preferable because it increases the amount of furnace components such as refractory heat insulators used. [Object of the Invention] The present invention has been made based on the above-mentioned problems, and provides a method for manufacturing LT single crystals with a predetermined crystal composition with good reproducibility by reducing compositional fluctuations within and between crystal lots. The purpose is to provide [Summary of the Invention] In the process of completing the invention, the inventors first determined that the crystal composition and the crystal Curie temperature (hereinafter abbreviated as _Tc ) were
Assuming that there is a one-to-one correspondence near the congruent composition, we investigated in detail the correlation between the initial melt composition (secondary sintering raw material composition) and T _c fluctuation in each pulled crystal. From the results, the true congruent composition can be determined by Li/
The Ta molar ratio is estimated to be 0.9342, and the raw material composition ratio is finally 0.937±0.004, taking into account the actual fluctuations in the single crystal manufacturing process. Raw materials for single crystal growth such as LT powder or single crystal lumps. The present invention was completed based on the discovery that the range of variation in crystal composition can be reduced by using the above method. In other words, according to the present invention, the fluctuation range of the crystal composition is within 0.935± in Li/Ta molar ratio both within a crystal lot and between crystal lots.
0.006, about 1/2 of the conventional fluctuation range (0.948±0.01)
As shown below, LT single crystals can be produced with good reproducibility. Next, the above content will be explained in detail. Now let the congruent composition Li ₂ O molar concentration be C _c ,
The initial melt composition (secondary sintering raw material composition) Li ₂ O molar concentration is C _p and the excess Li ₂ O molar concentration than the congruent composition in the same melt is C _p ′, and the solidification rate g of the first pulling is Crystal composition Li ₂ O molar concentration of crystal tail C _s
() and the excess Li ₂ O concentration as C _s ′(),
If the Li ₂ O/LiTaO ₃ segregation coefficient is K, the following relational expression holds true from the crystal composition variation theoretical expression. C _s ′()=K×C _p ′()×(1-g) ^K-1 ×{(
1-g) ^K×(I-1) +(1-g) ^K×(I-2) ×g+…+(1-g) ^K ×g+g}…(1) C _s ()=C _c ×( 1−C _s ′())+C _s ′()…(2
) while the crystal composition C _s () and the same Killie temperature T _c ()
Assuming that a linear equation like equation (3) holds between C _s () = K ₂ ×T _c () + K ₃ …(3), then from equations (1), (2), and (3), T _c (2)−T _c (1)=A×{C _p −C _c }…(4) T _c (1)=B×{C _p −C _c }+T _cp …(5). Here, A, B, and T _cp are each expressed by the following formulas. A=K/K ₂ × (1-g) ^K-1 × {(1-g) ^K − (1-G)} …(6) B=K/K ₂ × (1-g) ^K-1 … (7) T _cp = C _c − K ₃ /K ₂ …(8) Crystal tail (g =
0.5) is measured, and the difference between the crystal tail T _c of the first and second pulling is T _c (2) − T _c
The correlation between (1) and the raw material composition _Cp is plotted in Figure 2. In addition, the correlation between T _c (1) and C _p is plotted in Figure 3. As is clear from Figures 2 and 3, (4), (5)
The proportional relationship of T _c (2)−T _c (1)∝C _p and T _c (1)∝C _p predicted by the formula is almost established. From this figure, the true congruent composition Li ₂ O molar concentration was estimated to be 0.4830 (Li/Ta molar ratio 0.9342). Also, by solving equations (6), (7), and (8) simultaneously using the linear slopes in Figures 2 and 3, K ₂ and K ₃ in equation (3) are each 4.2725×10 ^-4 ，
0.22464, near the congruent composition.
A correlation as shown in FIG. 4 was obtained as the relationship between C _s and T _c . Therefore, theoretically, the Li/Ta molar ratio
It would be fine to use a raw material composition of 0.9342, but
In reality, there is Li ₂ O scattering during crystal pulling, so
From experiments under various conditions, the raw material composition was determined to be Li/Ta.
It has been found that if the molar ratio is kept within 0.937±0.004, the fluctuation in crystal composition becomes smaller. FIG. 6 shows the experimental results. [Effects of the Invention] This invention stabilizes crystal composition fluctuation within 0.935±0.006 in terms of Li/Ta molar ratio.
The manufacturing yield of narrowband SAW filters and SAW resonators, which are sensitive to SAW characteristic fluctuations, is significantly improved. Furthermore, whereas conventional crystal compositions varied, we conducted preliminary tests before device manufacturing to evaluate the surface wave velocity of each crystal lot, and accordingly selected and used several types of masks prepared in advance. However, with this invention, the crystal composition fluctuation width has been reduced, and the surface wave velocity fluctuation has been reduced to ±0.5 m/
s or less, the above-mentioned preliminary test evaluation is not necessary, and the process can be simplified and the number of masks can be reduced, which greatly improves the productivity of the device manufacturing process. Example 1 A predetermined amount of lithium carbonate and tantalum pentoxide were combined and mixed together so that the Li/Ta molar ratio was 0.941 as the sintering raw material composition, and 13 kg of the powder raw material was heated and calcined at 1550°C, and 13 kg was filled into a crucible made of platinum rhodium. Then, a 6.5 kg X-axis crystal with a diameter of 100 mm and a length of 90 mm was created using the Czyochralski method, which is a normal high-frequency heating method.
For the second time, 6.5 kg of the above-mentioned sintering raw material was additionally charged into the crucible, and 6.5 kg of the same crystal was pulled up and produced, and thereafter 12 crystals were continuously produced using the same additional charging method.
The produced crystals were subjected to substrate processing in a conventional manner, and the T _c distribution was measured at five points on each crystal head and tail wafer. The results are summarized in Table 1. For comparison, Table 1 also shows measurement data of T _c distribution in each crystal using a conventional raw material with a sintering composition of Li/Ta = 0.955. As is clear from Table 1
When using raw materials with a composition ratio of 0.955, inside the crystal lot,
While the T _c variation between lots is a maximum of 12.0°C,
When the raw material with a composition ratio of 0.941 based on this invention was used, the maximum T _c fluctuation was 3.0°C, which was about 1/4 smaller. The relationship between the surface wave velocity and T _c has been empirically obtained as shown in Figure 5. According to this figure, the relationship between the surface wave velocity and Tc is
When the T _c fluctuation is ±1.5℃, the surface wave velocity fluctuation width is ±0.5m/
Within s, it is actually used for SAW resonators such as VTRs.
When a 91.25MH ₂ RF modulator was prototyped using one type of mask, the total yield was 93.5 in the case of 12 crystals of series 1 in Table 1 using raw materials with a molar ratio of 0.941.
%, but in the case of 9 series 2 crystals, there were many lots where the same mask could not achieve the specified resonant frequency, and in the end, 4 types of masks were selected and used in preliminary tests, and the total Yield is 88.3
It was %. Furthermore, as is clear from Table 1, in Series 2, the range of composition fluctuation increases with the number of times of pulling, and cracks occur in the crystal if the number of pulls exceeds 10, whereas in Series 1, even after 12 times of pulling, the range of composition fluctuation increases by 1. It was almost the same as the previous time, and no cracks occurred.

【table】

[Table] Example 2 Crystal preparation was performed 10 times in a row in the same manner as in Example 1 using a molar ratio of 0.933 as the sintering raw material composition. The T _c distribution of each crystal at that time is summarized in Table 2. For comparison, the results in the case of using raw materials at a molar ratio of 0.931 (Series 4) are also shown. In the case of 0.931, the crystal tail T _c fluctuates in the direction of decrease with the number of pulling, and the maximum T _c within a lot and between lots is 8.5℃.
In contrast, in the case of raw materials with a molar ratio of 0.933, a slightly similar trend is observed, but the maximum fluctuation range is
It was 3.0℃. Even though the 0.933 molar ratio is almost a congruent composition, there is a slight tendency for T _c to decrease, which is thought to be due to Li ₂ O scattering during crystal pulling.

[Table] Example 3 Sintering raw material composition is Li/Ta molar ratio of 0.9175,
0.9225, 0.931, 0.933, 0.937, 0.939, 0.941,
When the same crystal formation experiment as in Example 1 was carried out 9 times in a row with different values of 0.945, 0.950, and 0.955, the maximum fluctuation range of T _c in each series was plotted against the sintering raw material composition, as shown in Figure 6. The results were obtained. From the diagram
In order to keep the T _c maximum fluctuation width within 3℃ (two-surface wave velocity within ±0.5m/s), the raw material composition must be 0.933
It is clear that setting between ~0.941 is good. All of the above examples use sintered powder as the raw material, but the LiTaO ₃ raw material with the above molar ratio may be, for example, a cracked crystal product pulled by the above method or a wafer product recovered from the SAW device manufacturing process. But of course it's a good thing. As is clear from the results of the above examples, according to the present invention, even if continuous pulling is performed 10 times or more, there is little compositional variation and the crystal does not crack, so the productivity of single crystal production has been greatly improved.

[Brief explanation of drawings]

Figure 1 is a correlation diagram of Li ₂ O−Ta ₂ O ₅ binary system, Figure 2
Figure 3 shows the experimentally obtained △T _c = T _c (2)−T _c
(1) Correlation diagram between ∝C _p and T _c ∝C _p , Figure 4 is a correlation diagram between crystal composition and crystal Killie temperature, Figure 5 is a correlation diagram between experimentally determined crystal Killie temperature and surface wave velocity, FIG. 6 is a correlation diagram between the raw material composition Li/Ta molar ratio and the maximum width of intracrystal T _c variation.

Claims (1)

[Claims] 1. When growing a lithium tantalate single crystal by the pulling method, a lithium tantalate single crystal growth raw material with a Li/Ta molar ratio of 0.937±0.004 is used as a raw material for lithium tantalate single crystal growth. A method for producing lithium tantalate single crystal.