- Email: [email protected]
Ternary system Li2O–K2O–Nb2O5
Journal of Alloys and Compounds 386 (2005) 246–252
Ternary system Li2 O–K2 O–Nb2 O5 Part II: Growth of stoichiometric lithium niobate ´ P´etera , K. Polg´ara , M. Ferriolb,∗ , L. P¨opplc , I. F¨oldv´aria , M. Cochezb , Z.S. Szallera A. a
Research Institute for Solid State Physics and Optics, Hungarian Academy of Sciences, P.O. Box 49, H-1525 Budapest, Hungary b Laboratoire de Chimie et Applications, E.A. No. 3471, Universit´ e de Metz, Rue Victor Demange, 57500 Saint-Avold, France c Department of Inorganic and Analytical Chemistry, E¨ otv¨os University Budapest, P.O. Box 32, H-1518 Budapest 112, Hungary Received 26 December 2003; accepted 30 April 2004
Abstract Growth conditions of stoichiometric or near-stoichiometric lithium niobate (sLN) from K2 O containing fluxes have been studied by investigating phase equilibria in the ternary system Li2 O–K2 O–Nb2 O5 . The crystallization area of LiNbO3 was confirmed by growth experiments on the g1 (LiNbO3 –K2 O), g2 (LiNbO3 –ternary eutectic liquid (Et : 45.0 ± 1.5 mol% Nb2 O5 , 26.0 ± 1.5 mol% K2 O and 29.0 ± 1.5 mol% Li2 O)) and g3 (LiNbO3 –KNbO3 ) joins. The yield and the composition of LiNbO3 single crystals were tested for a wide range of starting compositions. From these experiments, the starting compositions leading to stoichiometric lithium niobate have been determined. © 2004 Elsevier B.V. All rights reserved. Keywords: Crystal growth; X-ray diffraction; Nonlinear optics; Thermal analysis
1. Introduction In the binary system Li2 O–Nb2 O5 [1–4], lithium niobate LiNbO3 (LN) melts congruently with the composition of about 48.6 mol% Li2 O and single crystals can be grown from melt by the Czochralski technique. The lithium deficiency in the congruent crystals leads to intrinsic structural defects and affects many physical properties. Therefore, there exists a significant demand for stoichiometric or quasistoichiometric crystals having a perfect crystal lattice and improved properties. One of the most effective methods of growing stoichiometric LiNbO3 (sLN) single crystals is the high temperature top-seeded solution growth (HTSSG) technique from K2 O–Li2 O–Nb2 O5 fluxes [5]. In this method, the growth usually starts from a composition with Li/Nb ratio either 1 or ∗ Corresponding author. Tel.: +33-3-87-93-91-85; fax: +33-3-87-93-91-01. E-mail address: [email protected] (M. Ferriol).
0925-8388/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jallcom.2004.04.151
0.945 (ratio of the congruently melting lithium niobate) and K2 O added from 10 to 16 mol%. Recently, we have shown that the real composition of the HTSSG grown crystals depends to a large extent on the choice of the appropriate crystallization temperature and, therefore, of the starting melt composition [6]. Thus, the optimization of the flux and growth conditions requires an extended knowledge of the crystallization range of LiNbO3 in the Li2 O–K2 O–Nb2 O5 system. In previous works [7–8], we reported phase equilibrium studies on the ternary system Li2 O–K2 O–Nb2 O5 . Part I [8] of this work presents in detail the solid–liquid equilibria near the LiNbO3 compound. In this report, based on the results of these studies, some issues of the stoichiometric LN single crystal formation are treated: • Determination of the limits of LiNbO3 liquidus surface and investigation of the yield of LiNbO3 single crystal depending on the starting composition. • Outline of the area where stoichiometric or nearstoichiometric single crystalline LiNbO3 can be grown.
´ P´eter et al. / Journal of Alloys and Compounds 386 (2005) 246–252 A.
2. Experimental methods Crystals were grown by the HTTSSG method in a diameter controlled growth apparatus. The technical details of the material synthesis and crystal growth are given in refs. [5,7]. The starting materials used were Starck LN grade Nb2 O5 and Merck Suprapure K2 CO3 and Li2 CO3 pre-reacted in the solid phase. Generally the maximum amount of the LiNbO3 single crystal phase was pulled out, after which the growth process was continued and then, “new” phases were crystallized. Thus, from the yield of LiNbO3 , the mass balance of the crystallization can be obtained and the co-ordinates of the limits of LiNbO3 liquidus surface can be calculated and compared with those given by the phase diagram. The constituent phases were assessed by X-ray powder diffraction analysis with a Philips PW 1710 diffractometer using Cu K␣ radiation in the 2θ range of 0–80◦ . The overall composition of the new phases and remaining flux were determined by chemical analysis of the major components. In these experiments, the LiNbO3 crystals were pulled in 18 mm diameter along the 00.1 = Z-axis with a rate of 0.1–0.2 mm/h and rotation speed of 6–28 rpm. The real stoichiometry of the crystals and its evolution during the growth were determined by using a standardized spectroscopic test recording the UV absorption edge position of slices cut along the growth axis from the top to bottom part of the crystal [9].
3. Results and discussion In the ternary system, all compositions are expressed as molar percentages of Nb2 O5 (X) and K2 O (Y).
247
3.1. Assessment of the phase diagram by crystal growth experiments The polythermal projection of LiNbO3 crystallization field in the ternary system Li2 O–K2 O–Nb2 O5 is shown in Fig. 1 as presented in Part I [8]. Accordingly, the LiNbO3 liquidus surface is well determined above the LiNbO3 –KNbO3 join, but beyond it and for Nb2 O5 contents higher than 50–55mol%, the outlined equilibrium domains are only estimated. Thus, the LiNbO3 liquidus surface is limited by four monovariant lines: Liq + Li3 NbO4 + LiNbO3 , Liq + KNbO3 + LiNbO3 , Liq + “KLN solid solution” + LiNbO3 and Liq + LiNb3 O8 + LiNbO3 (the term “KLN solid solution” is used for the ternary tungsten bronze-type solid solution encountered beyond Nb2 O5 = 50 mol% and K2 O = 25 mol%). At 997 ◦ C, the ternary eutectic reaction occurs between liquid, LiNbO3 , -Li3 NbO4 and KNbO3 . The composition of the eutectic liquid (Et ) is 45.0 ± 1.5 mol% Nb2 O5 , 26.0 ± 1.5 mol% K2 O and 29.0 ± 1.5 mol% Li2 O. A quasi-peritectic reaction has also been identified at a temperature of about 1050–1055 ◦ C. The quasi-peritectic liquid (T) has a composition roughly equal to: 49 mol% Nb2 O5 , 25.5 mol% K2 O and 25.5 mol% Li2 O. Up to now, the single-crystal growth technique of sLN was experienced only along the LiNbO3 –K2 O (Y = 100–2X) join: i.e. from fluxes with starting composition of a [Li]/[Nb] ratio equal/or near to 1 and with various potassium oxide contents [5–10]. In the present work, growth experiments have been conducted from various fluxes listed in Table 1. This table summarizes growth results and the constituent phases identified in the grown boules as well. In these growth experiments, besides the starting compositions located on the LiNbO3 –K2 O line (labelled as g1 in Table 1 and in Fig. 1),
Fig. 1. Expanded view of the polythermal projection of LiNbO3 crystallization field (Et : LiNbO3 –Li3 NbO4 –KNbO3 ternary eutectic point, T: ternary quasiperitectic point, e1 : LiNbO3 –Li3 NbO4 binary eutectic point, e2 : LiNbO3 –LiNb3 O8 binary eutectic point), light grey area: crystallization region of fully stoichiometric lithium niobate (LN), line AB: quasi-binary limit (see text).
´ P´eter et al. / Journal of Alloys and Compounds 386 (2005) 246–252 A.
248
Table 1 Starting melt compositions used for growth experiments Growth label
K2 O (mol%)
Nb2 O5 (mol%)
Li2 O (mol%)
Remarks on the crystals and phases obtained
g1.1 g1.2 g2.1 g2.2
15.25 13.6 18 26
42.37 43.2 46.5 45
42.37 43.2 35.5 29
g2.3 g3.1
10 20
48 50
42 30
g3.2
15.66
50
34.34
Top: single crys.: bottom: “white” poly crys.: < LN + L3N> Whole boule: single crys.: Whole boule: single crys.: Top: single crys: “yellow” poly crys.: middle: “yellow” single crys.: bottom: “white” poly crys.: Whole boule: single crys.: Top: single crys.: “yellow” poly crys.: bottom: “yellow” single crys.: Whole boule: single crys.
LN: LiNbO3 , L3N: Li3 NbO4 , KN: KNbO3 , KLN: tungsten bronze type solid solution.
those of the LiNbO3 –Et and LiNbO3 –KNbO3 joins have also been used (respectively labelled as g2 and g3). To confirm the outlined phase relations, the “new” phases were assessed by X-ray diffraction analysis. The results of such analysis are shown in Fig. 2. In the growth experiments performed along the LiNbO3 – K2 O (g1) line, at the end of the growth of the LN crystal white poly-crystalline phases were formed. The X-ray powder diffraction patterns of this part reveal only the presence
of LiNbO3 and Li3 NbO4 phases (Fig. 2a, starting melt composition g1.1). The chemical analysis of the same sample gave the composition: 35.64 mol% Nb2 O5 , 64.36 mol% Li2 O and traces of K2 O (corresponding to: 40.3 mol% Li3 NbO4 and 59.7 mol% LiNbO3 ). These results are consistent with the evolution of the remaining liquid (after pulling out all the sLN crystal) along the univariant line e1 –Et , shown in Fig. 1, corresponding to the three-phase equilibrium: liquid + LiNbO3 + Li3 NbO4 and support the phase diagram [8]. For the crystals grown with starting composition located along the LiNbO3 –Et (g2) and the LiNbO3 –KNbO3 (g3) lines, following the LiNbO3 growth, a yellow “quasi-monocrystalline” phase was obtained. Pulling out the maximal amount of the LiNbO3 phase from the melt g2.1 (18 mol% K2 O, 46.5 mol% Nb2 O5 and 35.5 mol% Li2 O), the mass balance gave 24.8 mol% K2 O, 45.24 mol% Nb2 O5 and 29.96 mol% Li2 O for the residual composition. These co-ordinates are very close to the eutectic composition (Et ) obtained from the layout of the diagram in Fig. 1. The X-ray assessment of the residue is shown in Fig. 2b and it reveals the Li3 NbO4 , KNbO3 and the tungsten bronze (KLN solid solution) phases. This result was expected since the solid phases produced by the eutectic reaction (-Li3 NbO4 , KNbO3 and LiNbO3 ) undergo the quasi-peritectoid reaction at 966 ◦ C: < LiNbO3 > + < KNbO3 ><  − Li3 NbO4 > + < KLN solid solution >
Fig. 2. X-ray assessment of the LiNbO3 crystallization field by powder diffraction patterns of the samples obtained after pulling out the LiNbO3 phase: (a): polycrystalline part of g1.1, (b): residue of the growth of g2.1, (c–d): top and bottom of the yellow part formed during the growth of g2.2, (e): white polycrystalline part of g2.2 (: KLN solid solution, : KNbO3 , 䊉: Li3 NbO4 , : LiNbO3 ).
In the experiment g2.2 starting from the assumed eutectic liquid (Et ) composition, a cracked light-yellow boule was pulled out with a white polycrystalline phase at the bottom. The weight of yellow boule and that of the white phase corresponds to 29 and to 2 wt.% of the starting amount (white phase not entirely pulled out). The X-ray powder diffraction patterns of the sample taken just below the seed reveal the presence of LiNbO3 (very weak lines) besides the dominant lines of the KLN solid solution (Fig. 2c). This result is in concordance with a starting composition close to, but not yet identical with that of the eutectic liquid. So, a little of
´ P´eter et al. / Journal of Alloys and Compounds 386 (2005) 246–252 A.
crystallized LiNbO3 or LiNbO3 –KNbO3 mixture will be observed according to the exact position of the eutectic point towards the composition studied. In the samples, from the neck to the end of boule, only the lines of the KLN solid solution are observed (Fig. 2d). The diffraction pattern of that apparently “pure” KLN phase matches the lines of the tetragonal KLN single crystal pattern with lattice constant ratio a/c ∼3 [11]. The X-ray assessment of the white polycrystalline phase (Fig. 2e) from the bottom reveals -Li3 NbO4 and KLN solid solution as constituents. Once again, this was expected since the three-phase mixture (LiNbO3 , KNbO3 , -Li3 NbO4 ) produced by the eutectic reaction undergoes the quasi-peritectoid reaction at 966 ◦ C to give a mixture of KNbO3 , -Li3 NbO4 and KLN solid solution in concordance with the layout of our phase diagram. KNbO3 was very difficult to observe because its concentration was probably very low and its major lines, near 2θ = 31.5◦ , were swamped in the very intense lines of KLN. The fact that -Li3 NbO4 was not detected in the first yellow part and the color of the considered phase could be the result of an incomplete reaction path of the solid state reaction. Note that in the experiment g2.3, the pulled out amount of sLN crystal was less than the theoretical crystallization yield, since the growth was stopped before the eutectic liquid composition was reached. In the g3 growth experiments (i.e. LiNbO3 –KNbO3 vertical section) with starting melt composition g3.1 (20 mol% K2 O, 50 mol% Nb2 O5 and 30 mol% Li2 O), firstly a transparent lithium niobate crystal was obtained with a yield of 18.5%. Pulling further a yellow cracked crystal part was grown. The X-ray powder diffraction patterns of the samples taken from top and bottom part of the yellow phase are similar to those of g2.2 (shown in Fig 3a and b, respectively). Accordingly, at the top, the co-crystallization of LiNbO3 (as minor) and KLN as major phase is found and, at the bottom, only the KLN phase is observed. According to the phase diagram, the behavior of initial melts located on the g3 vertical section can be described as follows. Whereas sLN crystallizes, the liquid evolves on the liquidus surface near the line LiNbO3 –KNbO3 . When the entire amount of LiNbO3 is pulled out, the liquid reaches the univariant line Liq + LiNbO3 + KLN solid solution (for g3 solutions). Then, on further pulling, a mixture of KLN solid solution and LiNbO3 (in which KLN is the major phase) is crystallized. The co-crystallization of LiNbO3 and KLN solid solution continues until the liquid reaches the quasi-peritectic point T. Then after, the liquid undergoes the co-crystallization of LiNbO3 and KNbO3 until it reaches the eutectic point Et . Between T and Et points, on cooling, the crystallized phase undergoes the quasi-peritectoid reaction at 966 ◦ C. Thus, with room temperature X-ray diffraction measurement, according to the primary composition of the crystallized LiNbO3 + KNbO3 mixture, a mixture of -Li3 NbO4 , KNbO3 and KLN solid solution or LiNbO3 , -Li3 NbO4 and KLN solid solution will be observed. In all cases, a large amount of KLN solid solution is expected. The first part of the yellow phase cor-
249
Fig. 3. X-ray assessment of the LiNbO3 crystallization field by powder diffraction patterns of the samples obtained after pulling out the LiNbO3 phase: (a-b): top and bottom of the yellow part formed during the growth of g3.1, (c): residue of g3.1 (: KLN solid solution, : KNbO3 , 䊉: Li3 NbO4 , : LiNbO3 ).
responds well to a LiNbO3 –KLN solid solution mixture (as shown by the X-ray analysis of the top of g3.1 yellow part). As expected the proportion of LiNbO3 is small (Fig. 3a). At the bottom part, only KLN solid solution is observed which can be explained by the fact that the amounts of the other compounds are lower than their detection level. Only 30% of the available yellow phase of g3.1 was pulled signifying that the growth was stopped when the liquid was located between points T and Et . The X-ray diffraction pattern of the residue (Fig. 3c) reveals the presence of LiNbO3 and -Li3 NbO4 probably due to the position of the representative point of the remaining liquid towards the three-phase domains involved in the quasi-peritectoid reaction as explained above. The mass balance of g3.1 gave for the liquid composition attained after the entire growth of sLN: 25.0 mol% K2 O, 50.0 mol% Nb2 O5 and 24.9 mol% Li2 O, which is in good concordance with the composition deduced from the layout of the phase diagram (Fig. 1). From mass balance on the g2 growth experiments, the ternary eutectic composition could be estimated to: 44.5 ± 1.5 mol% Nb2 O5 , 28 ± 1.5 mol% K2 O and 27.5 ± 1.5 mol% Li2 O close to the one deduced from the graphical layout of the vertical sections investigated by DSC [8].
250
´ P´eter et al. / Journal of Alloys and Compounds 386 (2005) 246–252 A.
3.2. Evolution of the crystallization yield and stoichiometry
the point reached on the univariant line in the phase diagram, the mass m1 of one mole of the corresponding liquid is given by:
In a first step disregarding the starting crystallization temperature and the actual crystal stoichiometry, it is interesting to study the evolution of the crystallization yield of LiNbO3 as a function of the starting flux composition. In the following, it will be assumed that the LiNbO3 which crystallizes is always fully stoichiometric (Li/Nb = 1). If the co-ordinates of the point representing the starting melt composition in the ternary diagram are X0 and Y0 , the mass m0 of one mole of liquid is given by: X0 MNb2 O5 + Y0 MK2 O + (100 − X0 − Y0 )MLi2 O m0 = 100 (1) In this equation Mi is the molar weight of compound i. During the crystallization of LiNbO3 , the point representing the liquid in equilibrium with the solid moves on the liquidus surface towards lower temperatures and consequently,
m1 =
X1 MNb2 O5 + Y1 MK2 O + (100 − X1 − Y1 )MLi2 O 100
K2 O being practically insoluble in the crystallized solid phase, it can be supposed that its amount is the same in the starting and ending liquids. For the weight of solid obtained, this condition is expressed by: mS = m0 −
Y0 m1 Y1
and so:
Y0 X0 − X 1 Y1 MLi2 O Y0 + 100 − X0 − (100 − X1 ) 100 Y1
MNb2 O5 mS = 100
(3)
The crystallization yield R is then given by: × X1 + MLi2 O 100 − X0 − YY01 (100 − X1 )
MNb2 O5 X0 − YY01 mS R(%) = 100 × = 100 × m0 X0 × MNb2 O5 + Y0 × MK2 O + (100 − X0 − Y0 )MLi2 O
towards one of the univariant lines limiting the LiNbO3 liquidus surface. When the liquid composition reaches a point on this line, a mixture of two solid phases precipitates on further pulling (starting solution brought into a three-phase region) which leads to a non single-crystalline sample as experimentally observed. If X1 and Y1 are the co-ordinates of
(2)
(4)
(5)
The evolution of the calculated crystallization yield as a function of the Nb2 O5 content in the starting solution for different K2 O contents is shown in Fig. 4. For a given Nb2 O5 amount, the yield decreases as the K2 O concentration increases, and it can be seen that for a given K2 O content, the yield exhibits a maximum. Applied to the g1, g2 and g3 growth experiments, the yield Eq. (5) leads to the results
Fig. 4. Evolution of the calculated crystallization yield for different starting compositions.
´ P´eter et al. / Journal of Alloys and Compounds 386 (2005) 246–252 A.
251
Fig. 5. Evolution of the solidus composition for the g1, g2 and g3 growth experiments.
given in Table 2. The calculated yields are in good agreement with the experimental ones which confirms the layout of our phase diagram. In previous studies on the g1 line, we pointed out that below 13.8 mol% potassium oxide content of the flux, the obtained crystals are not yet stoichiometric and they grow with shifting composition along the growth axis. Therefore, on the LiNbO3 –K2 O line, stoichiometric crystals are only obtained for K2 O > 13.8 mol% and the respective starting crystallization temperatures is below 1114 ◦ C [6,7]. Thus, the growth temperature seems to be a key parameter, which determines the composition of the growing crystal. Table 2 gives the top and bottom crystal compositions deduced from the absorption edge measurements whereas, as K2 O is insoluble in LiNbO3 , Fig. 5 shows the evolution of crystal compositions during the growth along the g1, g2 and g3 lines in the plane of the binary system Li2 O–Nb2 O5 . It confirms that temperature is well a key parameter depending on the composition of the initial solution. From this plot, it is easy to see that the full stoichiometry (Li/Nb = 1) cannot be obtained in any way for initial solutions belonging to the g3 line. For initial solutions
Fig. 6. Scheme of the evolution of the quasi-binary limit AB in the polythermal phase diagram.
Table 2 Crystals composition and yields of crystallization Growth label
g1.1 g1.2 g2.1 g2.3 g3.1 g3.2 a
Crystal composition Li2 O (mol%) Top
Bottom
49.98 49.935 49.76 49.56 49.61 49.405
49.99 49.99 49.93 49.76 49.68 49.69
Growth stopped before all the lithium niobate was pulled out.
Experimental yield of LiNbO3 (%)
Calculated yield of LiNbO3 (%)
8.5 18.0 27.6 39.5a 18.5 34.47
9.8 20.0 30.0 60.7 18.4 35.0
252
´ P´eter et al. / Journal of Alloys and Compounds 386 (2005) 246–252 A.
can be obtained. All the growth experiments conducted in this work agree with this interpretation of the phase diagram. Moreover, the calculation of the yield along the upper limit of this domain (line AB) shows that the maximum yield of fully stoichiometric LN is obtained for the following initial solution: 43.5 mol% Nb2 O5 , 15.4 mol% K2 O and 41.1 mol% Li2 O (Fig. 7). 4. Conclusion Extended range of phase equilibria in the Li2 O–K2 O– Nb2 O5 ternary system was tested and used for the growth of LiNbO3 crystals. From the constructed phase diagram, it was possible to calculate the yield of growth of nearstoichiometric lithium niobate single crystals. Growth experiments were performed with starting compositions along the LiNbO3 –K2 O, LiNbO3 –Et and LiNbO3 –KNbO3 lines. The solidus lines were established down to 1000 ◦ C and allowed to determine the area corresponding exactly to the crystallization range of fully stoichiometric lithium niobate.
Acknowledgements
Fig. 7. Calculated maximum crystallization yield of fully stoichiometric lithium niobate along the quasi-binary limit AB.
on the g1 and g2 lines, the full stoichiometry will be reached below the threshold temperature of 1103 ± 10 ◦ C for g1 and 1031 ± 10 ◦ C for g2. It is particularly worthwhile to relate these results to the phase diagram. We can consider that below the threshold temperatures given by the plot of Fig. 5, the vertical section originating at the representative point of LiNbO3 (i.e.: 50 mol% Nb2 O5 and 50 mol% Li2 O) and passing through the crystallization area of LiNbO3 (as g1, g2 and g3 for example) becomes a quasi-binary join in the LiNbO3 neighborhood. The threshold is all the more increased because the considered section is close to the binary eutectic point e1 involving LiNbO3 and Li3 NbO4 . Fig. 6 schematizes this situation. The curve AB illustrates the evolution of threshold with temperature and composition. In a given section, the composition and temperature below which fully stoichiometric LN will be obtained is given by the intercept of this curve with the plane of the vertical section. In projection onto the plane of compositions, we obtain the layout given in Fig. 1 in which the light grey zone shows the region in which fully stoichiometric LN
We wish to thank Dr. M. T´oth for his help and advice in the X-ray analysis. This work was supported by the Hungarian Scientific Research Fund (OTKA) Nos. T034176, T034262 and by the European Union under Grant CMRC No. ICA1CT-2000-70029.
References [1] A. Reisman, F. Holtzberg, J. Am. Chem. Soc. 80 (1958) 6503. [2] P. Lerner, C. Legras, J.P. Dumas, J. Cryst. Growth 3/4 (1968) 231. [3] L.O. Svaasand, M. Eriksrud, G. Nakken, A.P. Grande, J. Cryst. Growth 22 (1974) 230. [4] A. Dakki, M. Ferriol, M.T. Cohen-Adad, Eur. J. Solid State Inorg. Chem. 33 (1996) 19. ´ P´eter, L. Kov´acs, G. Corradi, Z. Szaller, J. Cryst. [5] K. Polg´ar, A. Growth 177 (1997) 211. ´ P´eter, I. F¨oldv´ari, Opt. Mater. 19 (2002) 7. [6] K. Polg´ar, A. ´ P´eter, L. P¨oppl, M. Ferriol, I. F¨oldv´ari, J. Cryst. Growth [7] K. Polg´ar, A. 237–239 (2002) 682. ´ P´eter, J. Alloys [8] M. Cochez, M. Ferriol, L. P¨oppl, K. Polg´ar, A. Comp. 386 (2004) 238–245. [9] L. Kov´acs, G. Ruschhaupt, K. Polg´ar, G. Corradi, M. W¨ohlecke, Appl. Phys. Lett. 70 (1997) 2801. ´ P´eter, I. F¨oldv´ari, Zs. Szaller, J. Cryst. Growth 218 [10] K. Polg´ar, A. (2000) 327. [11] X.W. Xu, T.C. Chong, G.Y. Zhang, L. Li, P.F. Hu, H. Kumagai, M. Hirano, J. Cryst. Growth 211 (2000) 265.
Ternary system Li2 O–K2 O–Nb2 O5 Part II: Growth of stoichiometric lithium niobate ´ P´etera , K. Polg´ara , M. Ferriolb,∗ , L. P¨opplc , I. F¨oldv´aria , M. Cochezb , Z.S. Szallera A. a
Research Institute for Solid State Physics and Optics, Hungarian Academy of Sciences, P.O. Box 49, H-1525 Budapest, Hungary b Laboratoire de Chimie et Applications, E.A. No. 3471, Universit´ e de Metz, Rue Victor Demange, 57500 Saint-Avold, France c Department of Inorganic and Analytical Chemistry, E¨ otv¨os University Budapest, P.O. Box 32, H-1518 Budapest 112, Hungary Received 26 December 2003; accepted 30 April 2004
Abstract Growth conditions of stoichiometric or near-stoichiometric lithium niobate (sLN) from K2 O containing fluxes have been studied by investigating phase equilibria in the ternary system Li2 O–K2 O–Nb2 O5 . The crystallization area of LiNbO3 was confirmed by growth experiments on the g1 (LiNbO3 –K2 O), g2 (LiNbO3 –ternary eutectic liquid (Et : 45.0 ± 1.5 mol% Nb2 O5 , 26.0 ± 1.5 mol% K2 O and 29.0 ± 1.5 mol% Li2 O)) and g3 (LiNbO3 –KNbO3 ) joins. The yield and the composition of LiNbO3 single crystals were tested for a wide range of starting compositions. From these experiments, the starting compositions leading to stoichiometric lithium niobate have been determined. © 2004 Elsevier B.V. All rights reserved. Keywords: Crystal growth; X-ray diffraction; Nonlinear optics; Thermal analysis
1. Introduction In the binary system Li2 O–Nb2 O5 [1–4], lithium niobate LiNbO3 (LN) melts congruently with the composition of about 48.6 mol% Li2 O and single crystals can be grown from melt by the Czochralski technique. The lithium deficiency in the congruent crystals leads to intrinsic structural defects and affects many physical properties. Therefore, there exists a significant demand for stoichiometric or quasistoichiometric crystals having a perfect crystal lattice and improved properties. One of the most effective methods of growing stoichiometric LiNbO3 (sLN) single crystals is the high temperature top-seeded solution growth (HTSSG) technique from K2 O–Li2 O–Nb2 O5 fluxes [5]. In this method, the growth usually starts from a composition with Li/Nb ratio either 1 or ∗ Corresponding author. Tel.: +33-3-87-93-91-85; fax: +33-3-87-93-91-01. E-mail address: [email protected] (M. Ferriol).
0925-8388/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jallcom.2004.04.151
0.945 (ratio of the congruently melting lithium niobate) and K2 O added from 10 to 16 mol%. Recently, we have shown that the real composition of the HTSSG grown crystals depends to a large extent on the choice of the appropriate crystallization temperature and, therefore, of the starting melt composition [6]. Thus, the optimization of the flux and growth conditions requires an extended knowledge of the crystallization range of LiNbO3 in the Li2 O–K2 O–Nb2 O5 system. In previous works [7–8], we reported phase equilibrium studies on the ternary system Li2 O–K2 O–Nb2 O5 . Part I [8] of this work presents in detail the solid–liquid equilibria near the LiNbO3 compound. In this report, based on the results of these studies, some issues of the stoichiometric LN single crystal formation are treated: • Determination of the limits of LiNbO3 liquidus surface and investigation of the yield of LiNbO3 single crystal depending on the starting composition. • Outline of the area where stoichiometric or nearstoichiometric single crystalline LiNbO3 can be grown.
´ P´eter et al. / Journal of Alloys and Compounds 386 (2005) 246–252 A.
2. Experimental methods Crystals were grown by the HTTSSG method in a diameter controlled growth apparatus. The technical details of the material synthesis and crystal growth are given in refs. [5,7]. The starting materials used were Starck LN grade Nb2 O5 and Merck Suprapure K2 CO3 and Li2 CO3 pre-reacted in the solid phase. Generally the maximum amount of the LiNbO3 single crystal phase was pulled out, after which the growth process was continued and then, “new” phases were crystallized. Thus, from the yield of LiNbO3 , the mass balance of the crystallization can be obtained and the co-ordinates of the limits of LiNbO3 liquidus surface can be calculated and compared with those given by the phase diagram. The constituent phases were assessed by X-ray powder diffraction analysis with a Philips PW 1710 diffractometer using Cu K␣ radiation in the 2θ range of 0–80◦ . The overall composition of the new phases and remaining flux were determined by chemical analysis of the major components. In these experiments, the LiNbO3 crystals were pulled in 18 mm diameter along the 00.1 = Z-axis with a rate of 0.1–0.2 mm/h and rotation speed of 6–28 rpm. The real stoichiometry of the crystals and its evolution during the growth were determined by using a standardized spectroscopic test recording the UV absorption edge position of slices cut along the growth axis from the top to bottom part of the crystal [9].
3. Results and discussion In the ternary system, all compositions are expressed as molar percentages of Nb2 O5 (X) and K2 O (Y).
247
3.1. Assessment of the phase diagram by crystal growth experiments The polythermal projection of LiNbO3 crystallization field in the ternary system Li2 O–K2 O–Nb2 O5 is shown in Fig. 1 as presented in Part I [8]. Accordingly, the LiNbO3 liquidus surface is well determined above the LiNbO3 –KNbO3 join, but beyond it and for Nb2 O5 contents higher than 50–55mol%, the outlined equilibrium domains are only estimated. Thus, the LiNbO3 liquidus surface is limited by four monovariant lines: Liq + Li3 NbO4 + LiNbO3 , Liq + KNbO3 + LiNbO3 , Liq + “KLN solid solution” + LiNbO3 and Liq + LiNb3 O8 + LiNbO3 (the term “KLN solid solution” is used for the ternary tungsten bronze-type solid solution encountered beyond Nb2 O5 = 50 mol% and K2 O = 25 mol%). At 997 ◦ C, the ternary eutectic reaction occurs between liquid, LiNbO3 , -Li3 NbO4 and KNbO3 . The composition of the eutectic liquid (Et ) is 45.0 ± 1.5 mol% Nb2 O5 , 26.0 ± 1.5 mol% K2 O and 29.0 ± 1.5 mol% Li2 O. A quasi-peritectic reaction has also been identified at a temperature of about 1050–1055 ◦ C. The quasi-peritectic liquid (T) has a composition roughly equal to: 49 mol% Nb2 O5 , 25.5 mol% K2 O and 25.5 mol% Li2 O. Up to now, the single-crystal growth technique of sLN was experienced only along the LiNbO3 –K2 O (Y = 100–2X) join: i.e. from fluxes with starting composition of a [Li]/[Nb] ratio equal/or near to 1 and with various potassium oxide contents [5–10]. In the present work, growth experiments have been conducted from various fluxes listed in Table 1. This table summarizes growth results and the constituent phases identified in the grown boules as well. In these growth experiments, besides the starting compositions located on the LiNbO3 –K2 O line (labelled as g1 in Table 1 and in Fig. 1),
Fig. 1. Expanded view of the polythermal projection of LiNbO3 crystallization field (Et : LiNbO3 –Li3 NbO4 –KNbO3 ternary eutectic point, T: ternary quasiperitectic point, e1 : LiNbO3 –Li3 NbO4 binary eutectic point, e2 : LiNbO3 –LiNb3 O8 binary eutectic point), light grey area: crystallization region of fully stoichiometric lithium niobate (LN), line AB: quasi-binary limit (see text).
´ P´eter et al. / Journal of Alloys and Compounds 386 (2005) 246–252 A.
248
Table 1 Starting melt compositions used for growth experiments Growth label
K2 O (mol%)
Nb2 O5 (mol%)
Li2 O (mol%)
Remarks on the crystals and phases obtained
g1.1 g1.2 g2.1 g2.2
15.25 13.6 18 26
42.37 43.2 46.5 45
42.37 43.2 35.5 29
g2.3 g3.1
10 20
48 50
42 30
g3.2
15.66
50
34.34
Top: single crys.:
LN: LiNbO3 , L3N: Li3 NbO4 , KN: KNbO3 , KLN: tungsten bronze type solid solution.
those of the LiNbO3 –Et and LiNbO3 –KNbO3 joins have also been used (respectively labelled as g2 and g3). To confirm the outlined phase relations, the “new” phases were assessed by X-ray diffraction analysis. The results of such analysis are shown in Fig. 2. In the growth experiments performed along the LiNbO3 – K2 O (g1) line, at the end of the growth of the LN crystal white poly-crystalline phases were formed. The X-ray powder diffraction patterns of this part reveal only the presence
of LiNbO3 and Li3 NbO4 phases (Fig. 2a, starting melt composition g1.1). The chemical analysis of the same sample gave the composition: 35.64 mol% Nb2 O5 , 64.36 mol% Li2 O and traces of K2 O (corresponding to: 40.3 mol% Li3 NbO4 and 59.7 mol% LiNbO3 ). These results are consistent with the evolution of the remaining liquid (after pulling out all the sLN crystal) along the univariant line e1 –Et , shown in Fig. 1, corresponding to the three-phase equilibrium: liquid + LiNbO3 + Li3 NbO4 and support the phase diagram [8]. For the crystals grown with starting composition located along the LiNbO3 –Et (g2) and the LiNbO3 –KNbO3 (g3) lines, following the LiNbO3 growth, a yellow “quasi-monocrystalline” phase was obtained. Pulling out the maximal amount of the LiNbO3 phase from the melt g2.1 (18 mol% K2 O, 46.5 mol% Nb2 O5 and 35.5 mol% Li2 O), the mass balance gave 24.8 mol% K2 O, 45.24 mol% Nb2 O5 and 29.96 mol% Li2 O for the residual composition. These co-ordinates are very close to the eutectic composition (Et ) obtained from the layout of the diagram in Fig. 1. The X-ray assessment of the residue is shown in Fig. 2b and it reveals the Li3 NbO4 , KNbO3 and the tungsten bronze (KLN solid solution) phases. This result was expected since the solid phases produced by the eutectic reaction (-Li3 NbO4 , KNbO3 and LiNbO3 ) undergo the quasi-peritectoid reaction at 966 ◦ C: < LiNbO3 > + < KNbO3 ><  − Li3 NbO4 > + < KLN solid solution >
Fig. 2. X-ray assessment of the LiNbO3 crystallization field by powder diffraction patterns of the samples obtained after pulling out the LiNbO3 phase: (a): polycrystalline part of g1.1, (b): residue of the growth of g2.1, (c–d): top and bottom of the yellow part formed during the growth of g2.2, (e): white polycrystalline part of g2.2 (: KLN solid solution, : KNbO3 , 䊉: Li3 NbO4 , : LiNbO3 ).
In the experiment g2.2 starting from the assumed eutectic liquid (Et ) composition, a cracked light-yellow boule was pulled out with a white polycrystalline phase at the bottom. The weight of yellow boule and that of the white phase corresponds to 29 and to 2 wt.% of the starting amount (white phase not entirely pulled out). The X-ray powder diffraction patterns of the sample taken just below the seed reveal the presence of LiNbO3 (very weak lines) besides the dominant lines of the KLN solid solution (Fig. 2c). This result is in concordance with a starting composition close to, but not yet identical with that of the eutectic liquid. So, a little of
´ P´eter et al. / Journal of Alloys and Compounds 386 (2005) 246–252 A.
crystallized LiNbO3 or LiNbO3 –KNbO3 mixture will be observed according to the exact position of the eutectic point towards the composition studied. In the samples, from the neck to the end of boule, only the lines of the KLN solid solution are observed (Fig. 2d). The diffraction pattern of that apparently “pure” KLN phase matches the lines of the tetragonal KLN single crystal pattern with lattice constant ratio a/c ∼3 [11]. The X-ray assessment of the white polycrystalline phase (Fig. 2e) from the bottom reveals -Li3 NbO4 and KLN solid solution as constituents. Once again, this was expected since the three-phase mixture (LiNbO3 , KNbO3 , -Li3 NbO4 ) produced by the eutectic reaction undergoes the quasi-peritectoid reaction at 966 ◦ C to give a mixture of KNbO3 , -Li3 NbO4 and KLN solid solution in concordance with the layout of our phase diagram. KNbO3 was very difficult to observe because its concentration was probably very low and its major lines, near 2θ = 31.5◦ , were swamped in the very intense lines of KLN. The fact that -Li3 NbO4 was not detected in the first yellow part and the color of the considered phase could be the result of an incomplete reaction path of the solid state reaction. Note that in the experiment g2.3, the pulled out amount of sLN crystal was less than the theoretical crystallization yield, since the growth was stopped before the eutectic liquid composition was reached. In the g3 growth experiments (i.e. LiNbO3 –KNbO3 vertical section) with starting melt composition g3.1 (20 mol% K2 O, 50 mol% Nb2 O5 and 30 mol% Li2 O), firstly a transparent lithium niobate crystal was obtained with a yield of 18.5%. Pulling further a yellow cracked crystal part was grown. The X-ray powder diffraction patterns of the samples taken from top and bottom part of the yellow phase are similar to those of g2.2 (shown in Fig 3a and b, respectively). Accordingly, at the top, the co-crystallization of LiNbO3 (as minor) and KLN as major phase is found and, at the bottom, only the KLN phase is observed. According to the phase diagram, the behavior of initial melts located on the g3 vertical section can be described as follows. Whereas sLN crystallizes, the liquid evolves on the liquidus surface near the line LiNbO3 –KNbO3 . When the entire amount of LiNbO3 is pulled out, the liquid reaches the univariant line Liq + LiNbO3 + KLN solid solution (for g3 solutions). Then, on further pulling, a mixture of KLN solid solution and LiNbO3 (in which KLN is the major phase) is crystallized. The co-crystallization of LiNbO3 and KLN solid solution continues until the liquid reaches the quasi-peritectic point T. Then after, the liquid undergoes the co-crystallization of LiNbO3 and KNbO3 until it reaches the eutectic point Et . Between T and Et points, on cooling, the crystallized phase undergoes the quasi-peritectoid reaction at 966 ◦ C. Thus, with room temperature X-ray diffraction measurement, according to the primary composition of the crystallized LiNbO3 + KNbO3 mixture, a mixture of -Li3 NbO4 , KNbO3 and KLN solid solution or LiNbO3 , -Li3 NbO4 and KLN solid solution will be observed. In all cases, a large amount of KLN solid solution is expected. The first part of the yellow phase cor-
249
Fig. 3. X-ray assessment of the LiNbO3 crystallization field by powder diffraction patterns of the samples obtained after pulling out the LiNbO3 phase: (a-b): top and bottom of the yellow part formed during the growth of g3.1, (c): residue of g3.1 (: KLN solid solution, : KNbO3 , 䊉: Li3 NbO4 , : LiNbO3 ).
responds well to a LiNbO3 –KLN solid solution mixture (as shown by the X-ray analysis of the top of g3.1 yellow part). As expected the proportion of LiNbO3 is small (Fig. 3a). At the bottom part, only KLN solid solution is observed which can be explained by the fact that the amounts of the other compounds are lower than their detection level. Only 30% of the available yellow phase of g3.1 was pulled signifying that the growth was stopped when the liquid was located between points T and Et . The X-ray diffraction pattern of the residue (Fig. 3c) reveals the presence of LiNbO3 and -Li3 NbO4 probably due to the position of the representative point of the remaining liquid towards the three-phase domains involved in the quasi-peritectoid reaction as explained above. The mass balance of g3.1 gave for the liquid composition attained after the entire growth of sLN: 25.0 mol% K2 O, 50.0 mol% Nb2 O5 and 24.9 mol% Li2 O, which is in good concordance with the composition deduced from the layout of the phase diagram (Fig. 1). From mass balance on the g2 growth experiments, the ternary eutectic composition could be estimated to: 44.5 ± 1.5 mol% Nb2 O5 , 28 ± 1.5 mol% K2 O and 27.5 ± 1.5 mol% Li2 O close to the one deduced from the graphical layout of the vertical sections investigated by DSC [8].
250
´ P´eter et al. / Journal of Alloys and Compounds 386 (2005) 246–252 A.
3.2. Evolution of the crystallization yield and stoichiometry
the point reached on the univariant line in the phase diagram, the mass m1 of one mole of the corresponding liquid is given by:
In a first step disregarding the starting crystallization temperature and the actual crystal stoichiometry, it is interesting to study the evolution of the crystallization yield of LiNbO3 as a function of the starting flux composition. In the following, it will be assumed that the LiNbO3 which crystallizes is always fully stoichiometric (Li/Nb = 1). If the co-ordinates of the point representing the starting melt composition in the ternary diagram are X0 and Y0 , the mass m0 of one mole of liquid is given by: X0 MNb2 O5 + Y0 MK2 O + (100 − X0 − Y0 )MLi2 O m0 = 100 (1) In this equation Mi is the molar weight of compound i. During the crystallization of LiNbO3 , the point representing the liquid in equilibrium with the solid moves on the liquidus surface towards lower temperatures and consequently,
m1 =
X1 MNb2 O5 + Y1 MK2 O + (100 − X1 − Y1 )MLi2 O 100
K2 O being practically insoluble in the crystallized solid phase, it can be supposed that its amount is the same in the starting and ending liquids. For the weight of solid obtained, this condition is expressed by: mS = m0 −
Y0 m1 Y1
and so:
Y0 X0 − X 1 Y1 MLi2 O Y0 + 100 − X0 − (100 − X1 ) 100 Y1
MNb2 O5 mS = 100
(3)
The crystallization yield R is then given by: × X1 + MLi2 O 100 − X0 − YY01 (100 − X1 )
MNb2 O5 X0 − YY01 mS R(%) = 100 × = 100 × m0 X0 × MNb2 O5 + Y0 × MK2 O + (100 − X0 − Y0 )MLi2 O
towards one of the univariant lines limiting the LiNbO3 liquidus surface. When the liquid composition reaches a point on this line, a mixture of two solid phases precipitates on further pulling (starting solution brought into a three-phase region) which leads to a non single-crystalline sample as experimentally observed. If X1 and Y1 are the co-ordinates of
(2)
(4)
(5)
The evolution of the calculated crystallization yield as a function of the Nb2 O5 content in the starting solution for different K2 O contents is shown in Fig. 4. For a given Nb2 O5 amount, the yield decreases as the K2 O concentration increases, and it can be seen that for a given K2 O content, the yield exhibits a maximum. Applied to the g1, g2 and g3 growth experiments, the yield Eq. (5) leads to the results
Fig. 4. Evolution of the calculated crystallization yield for different starting compositions.
´ P´eter et al. / Journal of Alloys and Compounds 386 (2005) 246–252 A.
251
Fig. 5. Evolution of the solidus composition for the g1, g2 and g3 growth experiments.
given in Table 2. The calculated yields are in good agreement with the experimental ones which confirms the layout of our phase diagram. In previous studies on the g1 line, we pointed out that below 13.8 mol% potassium oxide content of the flux, the obtained crystals are not yet stoichiometric and they grow with shifting composition along the growth axis. Therefore, on the LiNbO3 –K2 O line, stoichiometric crystals are only obtained for K2 O > 13.8 mol% and the respective starting crystallization temperatures is below 1114 ◦ C [6,7]. Thus, the growth temperature seems to be a key parameter, which determines the composition of the growing crystal. Table 2 gives the top and bottom crystal compositions deduced from the absorption edge measurements whereas, as K2 O is insoluble in LiNbO3 , Fig. 5 shows the evolution of crystal compositions during the growth along the g1, g2 and g3 lines in the plane of the binary system Li2 O–Nb2 O5 . It confirms that temperature is well a key parameter depending on the composition of the initial solution. From this plot, it is easy to see that the full stoichiometry (Li/Nb = 1) cannot be obtained in any way for initial solutions belonging to the g3 line. For initial solutions
Fig. 6. Scheme of the evolution of the quasi-binary limit AB in the polythermal phase diagram.
Table 2 Crystals composition and yields of crystallization Growth label
g1.1 g1.2 g2.1 g2.3 g3.1 g3.2 a
Crystal composition Li2 O (mol%) Top
Bottom
49.98 49.935 49.76 49.56 49.61 49.405
49.99 49.99 49.93 49.76 49.68 49.69
Growth stopped before all the lithium niobate was pulled out.
Experimental yield of LiNbO3 (%)
Calculated yield of LiNbO3 (%)
8.5 18.0 27.6 39.5a 18.5 34.47
9.8 20.0 30.0 60.7 18.4 35.0
252
´ P´eter et al. / Journal of Alloys and Compounds 386 (2005) 246–252 A.
can be obtained. All the growth experiments conducted in this work agree with this interpretation of the phase diagram. Moreover, the calculation of the yield along the upper limit of this domain (line AB) shows that the maximum yield of fully stoichiometric LN is obtained for the following initial solution: 43.5 mol% Nb2 O5 , 15.4 mol% K2 O and 41.1 mol% Li2 O (Fig. 7). 4. Conclusion Extended range of phase equilibria in the Li2 O–K2 O– Nb2 O5 ternary system was tested and used for the growth of LiNbO3 crystals. From the constructed phase diagram, it was possible to calculate the yield of growth of nearstoichiometric lithium niobate single crystals. Growth experiments were performed with starting compositions along the LiNbO3 –K2 O, LiNbO3 –Et and LiNbO3 –KNbO3 lines. The solidus lines were established down to 1000 ◦ C and allowed to determine the area corresponding exactly to the crystallization range of fully stoichiometric lithium niobate.
Acknowledgements
Fig. 7. Calculated maximum crystallization yield of fully stoichiometric lithium niobate along the quasi-binary limit AB.
on the g1 and g2 lines, the full stoichiometry will be reached below the threshold temperature of 1103 ± 10 ◦ C for g1 and 1031 ± 10 ◦ C for g2. It is particularly worthwhile to relate these results to the phase diagram. We can consider that below the threshold temperatures given by the plot of Fig. 5, the vertical section originating at the representative point of LiNbO3 (i.e.: 50 mol% Nb2 O5 and 50 mol% Li2 O) and passing through the crystallization area of LiNbO3 (as g1, g2 and g3 for example) becomes a quasi-binary join in the LiNbO3 neighborhood. The threshold is all the more increased because the considered section is close to the binary eutectic point e1 involving LiNbO3 and Li3 NbO4 . Fig. 6 schematizes this situation. The curve AB illustrates the evolution of threshold with temperature and composition. In a given section, the composition and temperature below which fully stoichiometric LN will be obtained is given by the intercept of this curve with the plane of the vertical section. In projection onto the plane of compositions, we obtain the layout given in Fig. 1 in which the light grey zone shows the region in which fully stoichiometric LN
We wish to thank Dr. M. T´oth for his help and advice in the X-ray analysis. This work was supported by the Hungarian Scientific Research Fund (OTKA) Nos. T034176, T034262 and by the European Union under Grant CMRC No. ICA1CT-2000-70029.
References [1] A. Reisman, F. Holtzberg, J. Am. Chem. Soc. 80 (1958) 6503. [2] P. Lerner, C. Legras, J.P. Dumas, J. Cryst. Growth 3/4 (1968) 231. [3] L.O. Svaasand, M. Eriksrud, G. Nakken, A.P. Grande, J. Cryst. Growth 22 (1974) 230. [4] A. Dakki, M. Ferriol, M.T. Cohen-Adad, Eur. J. Solid State Inorg. Chem. 33 (1996) 19. ´ P´eter, L. Kov´acs, G. Corradi, Z. Szaller, J. Cryst. [5] K. Polg´ar, A. Growth 177 (1997) 211. ´ P´eter, I. F¨oldv´ari, Opt. Mater. 19 (2002) 7. [6] K. Polg´ar, A. ´ P´eter, L. P¨oppl, M. Ferriol, I. F¨oldv´ari, J. Cryst. Growth [7] K. Polg´ar, A. 237–239 (2002) 682. ´ P´eter, J. Alloys [8] M. Cochez, M. Ferriol, L. P¨oppl, K. Polg´ar, A. Comp. 386 (2004) 238–245. [9] L. Kov´acs, G. Ruschhaupt, K. Polg´ar, G. Corradi, M. W¨ohlecke, Appl. Phys. Lett. 70 (1997) 2801. ´ P´eter, I. F¨oldv´ari, Zs. Szaller, J. Cryst. Growth 218 [10] K. Polg´ar, A. (2000) 327. [11] X.W. Xu, T.C. Chong, G.Y. Zhang, L. Li, P.F. Hu, H. Kumagai, M. Hirano, J. Cryst. Growth 211 (2000) 265.