Journal of Crystal Growth 250 (2003) 251–255

Optical properties of Czochralski grown rare-earth garnetsingle crystals in solid solution

Hideo Kimura*, Akimitsu Miyazaki

Materials Engineering Laboratory, National Institute for Materials Science, 1-2-1 Sengen, Tsukuba, Ibaraki 305-0047, Japan

Abstract

Garnet single crystals in solid solution were grown by the Czochralski method. Optical properties were estimated. In

the transmission spectrum, small difference was observed between Dy3Ga5O12 and Dy3Al5O12. Shorter cut-off

frequency on the transparency limit decreased with the decrease of the lattice parameter. New results which revealed

that both refractive index and dielectric constant increased with the increase of their lattice parameters depending on an

ionic radius of rare-earth ions were obtained. Garnet single crystals in solid solution hardened by the solid solution

hardening.

r 2002 Elsevier Science B.V. All rights reserved.

PACS: 81.10.Fq; 78.20.Ci; S10.15

Keywords: A2. Growth from melt; A2. Czochralski method; B1. Rare earth compounds

1. Introduction

Rare-earth garnet single crystals are promisingfor optical and magnetic applications. We haveinvestigated the use of garnet single crystals asmagnetic materials for magnetic refrigerators [1].We grew many garnet single crystals in solidsolution [1], and many new properties have beenreported such as decrease of magnetic phasetransition temperature [1] and increase of thermalconductivity [2]. However, their optical propertieshave hardly been investigated. Up to the present,the refractive index of the rare-earth garnet hasbeen reported to be independent of rare-earth ions

[3]. It is well known that the most properties of therare-earth garnet are dependent on the latticeparameters, particularly the ionic radius of rare-earth ions. For example, the melting pointincreases with the decrease of the lattice parameterof small rare-earth ions [2].

In the present work, we attempted to growsingle crystals of rare-earth garnet by the Czo-chralski (Cz) method, and measured a latticeparameter and an optical transmission spectrum.Furthermore, we discussed a shorter cut-offfrequency as a transparency limit on the singlecrystals relating the composition and the latticeparameter. In addition, hardness, which wasimportant for handling of the crystal, was esti-mated. The grown single crystals were a series ofGa garnet such as Dy3Ga5O12 (DGG), Gd3Ga5O12

(GGG) and Y3Ga5O12 (YGG), a series of Al

*Corresponding author. Tel.: +81-298-59-2437; fax: +81-

298-59-2401.

E-mail address: kimura.hideo@nims.go.jp (H. Kimura).

0022-0248/03/$ - see front matter r 2002 Elsevier Science B.V. All rights reserved.

doi:10.1016/S0022-0248(02)02256-X

garnet such as Dy3Al5O12 (DAG) and Y3Al5O12

(YAG), and a series of solid solution such asGd3(GaAl)5O12 (GGAG), Dy3(GaAl)5O12 (DGAG),(GdY)3Ga5O12 (GYGG), (GdDy)3Ga5O12 (GDGG),(DyY)3Ga5O12 (DYGG), (GdY)3Al5O12 (GYAG),(GdDy)3Al5O12 (GDAG) and (DyY)3Al5O12 (DYAG).

2. Experimental procedures

Garnet single crystals were grown by the Czmethod using RF power supply. Crystals in uni-component were DGG, GGG, YGG, DAG andYAG. Crystals in solid solution were GGAG(GGG-GAG), DGAG (DGG-DAG), GYGG(GGG-YGG), GDGG (GGG-DGG), DYGG(DGG-YGG), GYAG (GAG-YAG), GDAG(GAG-DAG) and DYAG (DAG-YAG). In thisexperiment, single crystal of Gd3Al5O12 (GAG)could not be grown by the Cz method due to itsincongruent melting. The Ga garnet single crystalsin uni-component were grown with congruentcomposition, and the Al garnet single crystals inuni-component were grown with stoichiometriccomposition. The garnet single crystals in solidsolution were grown based on the stoichiometriccomposition. An Ir crucible of 47 mm diameterand 48.5 mm depth was used. Thus grown singlecrystals were 15–20 mm diameter and 30–40 mmlength. All crystals were grown under N2+2% O2

or Ar+1% O2 atmosphere for the Ga garnetsingle crystals and N2 or Ar atmosphere for the Algarnet single crystals. Garnet single crystals, whichcontained both Ga and Al, were grown underN2+1–2% O2 atmosphere. The use of variousatmospheres is attributed to the use of different Czgrowth furnaces, and does not affect the presentresults. An atmosphere with O2 was required tosuppress the evaporation of Ga from the melt onthe garnet containing Ga.

The lattice parameter as a cubic system wascalculated using the data of the powder X-raydiffraction. The optical transmission spectrum wasmeasured by a spectrophotometer using Halogenand Hydrogen lamps in a wavelength from 210 to800 nm at room temperature. The spectrumobtained with no sample was used as the reference.Samples were of wafer shape of 5–15 mm diameter

and 0.8–1.0 mm thickness cut perpendicular to thegrowth direction of the /1 1 1S; and were polishedas mirror surface. The refractive index wasmeasured by the prism-coupling method [4]radiated by He–Ne laser (633 nm) using a prismmade of TiO2 crystal, which had large refractiveindex. This method was more precise than theconventional Abbe method. The dielectric con-stant was measured using an LCR meter by meansof an electrode noncontact method in a frequencyof 1–100 kHz. By this method, the dielectricconstant of fused SiO2 was measured to be 3.78,which is similar to the generally believed value of3.8. The hardness was measured by the micro-Vickers hardness tester with a weight of 100 g.

3. Results and discussion

Fig. 1 shows the relationship between the latticeparameter and the content. The garnet describedby {C}3[A]2(D)3O12 has three sites for cations, i.e.,dodecahedral {24c} site, octahedral [16a] site andtetrahedral (24d) site. The ionic radii of the rare-earth ions were: 0.1053 nm for Gd, 0.1027 nm for

Fig 1. Relationship between lattice parameter, a; and content,

x; on garnet single crystals in uni-component and in solid

solution.

H. Kimura, A. Miyazaki / Journal of Crystal Growth 250 (2003) 251–255252

Dy and 0.1019 nm for Y at {24c} site, 0.0620 nmfor Ga and 0.0535 nm for Al at [16a] site [5]. TheVegard’s rule in solid solution was applied. Thusthe ions were substituted at each site roughly.Details of the lattice parameter on the garnetsingle crystals were discussed in Ref. [6].

Fig. 2 shows the optical transmission spectrumon DGG, DGG-DAG and DAG in relative scale.

The transmission spectrum on DAG and DGGwas different in some point. Absorption at 786.5and 788 nm was observed on DAG, but not onDGG. The absorption was also not observed onDGG-DAG [Dy3(Ga0.4Al0.6)5O12]. Furthermore,the absorption was observed on DAG at 257 nm,meanwhile the absorption was not observedon DGG due to a shorter cut-off frequency.

Fig 2. Optical transmission spectrum on Dy3Ga5O12, Dy3(Ga0.4Al0.6)5O12 and Dy3Al5O12 in relative scale. Arrow marks on

Dy3Al5O12 show absorption at 786.5 and 788 nm (black arrow), and 257 nm (white arrow).

H. Kimura, A. Miyazaki / Journal of Crystal Growth 250 (2003) 251–255 253

DAG-GAG in solid solutions as compared withDGG-GGG in solid solutions were not differentbesides the cut-off frequency. No absorption wasobserved on YGG and YAG. GGG and YGG aredifferent on absorption at 274, 307 and 312 nm byan affect of Gd3+ ion. Other garnet single crystalswere not observed as extraordinary absorption.Main absorption on the present garnet singlecrystals was caused by Dy3+ ion. The garnetcontaining Dy3+ ion was still transparent enoughfrom 500 to 740 nm for an optical waveguideapplications [3].

Fig. 3 shows the relationship between the short-er cut-off frequency of the transparency limit onthe ultra-violent region and the lattice parameter.The shorter cut-off frequency was mostly depen-dent on the lattice parameter. The dependence wasdecreasing with the decrease of the lattice para-meter, i.e., the decreasing of the ionic radius ofrare-earth ions.

Fig. 4 shows the relationship between therefractive index and the lattice parameter. Analmost linear relationship was obtained in thepresent scale. The refractive index below 1.86 onthe Al garnet single crystals could not be measuredusing the present prism made of TiO2 crystal. Therefractive index increased with increase of the

lattice parameter. Tien et al. show the relationshipas independent of the lattice parameter on the Gaand the Al garnet at 1520 nm [3] as shown inFig. 2. The results of Tien et al. indicate that therefractive index of the rare-earth garnet crystalschanges only by the Ga or Al content independentof the rare-earth ions, i.e., the refractive index isconstant on every Ga and Al garnet. It was difficultto understand the lattice parameter dependence ofthe refractive index because the refractive index onthe Al garnet single crystals could not be measured,thus we also measured the dielectric constant.

Fig. 5 shows the relationship between the di-electric constant and the lattice parameter at10 kHz. Almost linear relationship was obtainedin the present scale. The relationship was similar tothat on the refractive index. Frequency depen-dence was hardly observed. The lattice parameterdependence of the refractive index is not clearbecause the data on the Al garnet single crystals ismeager, but that of the dielectric constant is clear.Thus, we conclude that the refractive index and thedielectric constant are dependent on the latticeparameter, i.e., on the ionic radius of rare-earthions too.

Fig 3. Relationship between shorter cut-off frequency, fc; and

lattice parameter, a; on garnet single crystals in uni-component

and in solid solution. Dotted lines show trends of fc decreasing

with the decrease of the lattice parameter.

Fig 4. Relationship between refractive index, n; and lattice

parameter, a; on single crystals which contained Ga only, or

both Ga and Al, together with the results of Tien et al. at

1520 nm [3] (Dotted line marked as R3Ga5O12 and R3Al5O12, R:

Rare-earth ions).

H. Kimura, A. Miyazaki / Journal of Crystal Growth 250 (2003) 251–255254

Fig. 6 shows the relationship between the Vick-ers hardness and the lattice parameter. Thehardness increased by the substitution of therare-earth ions, Ga or Al ion, i.e., the solidsolution hardening. The maximum value wasobtained nearly at 1:1 composition in solidsolution. The garnet single crystals in solidsolution have 1.3–1.5 times larger hardness thanthat in uni-component. The hardening is useful forthe applications, but sometimes becomes a causeof a crack during cooling, cutting or polishing.

4. Summary

Garnet single crystals in solid solution weregrown by the Czochralski method. Small differ-ence was observed on the transmission spectrumbetween Dy3Ga5O12 and Dy3Al5O12. Most differ-ence on the transmission spectrum of the presentgarnet single crystals was caused by Dy3+ ion.Shorter cut-off frequency on the ultra-violentregion for the transparency limit decreased withdecrease in the lattice parameter. Both therefractive index and the dielectric constant weredependent on the lattice parameter, i.e., the ionicradius of rare-earth ions. The garnet single crystalsin solid solution hardened by solid solutionhardening.

References

[1] H. Kimura, T. Numazawa, M. Sato, Curr. Top. Cryst.

Growth Res. 1 (1994) 329.

[2] H. Kimura, T. Numazawa, M. Sato, J. Jpn. Assoc. Cryst.

Growth 23 (1996) 296.

[3] P.K. Tien, R.J. Martin, S.L. Blank, S.H. Wemple, L.J.

Varnerin, Appl. Phys. Lett. 21 (1972) 207.

[4] H. Onodera, I. Awai, J. Ikenoue, Appl. Opt. 22 (1983) 1194.

[5] R.D. Shannon, Acta Cryst. A 32 (1976) 751.

[6] H. Kimura, T. Numazawa, M. Sato, J. Mater. Sci. Lett. 13

(1994) 1164.

Fig 5. Relationship between dielectric constant, er; and lattice

parameter, a; on single crystals which contained Ga or Al only,

or both Ga and Al.

Fig. 6. Relationship between Vickers hardness, Hv, and con-

tent, x, on garnet single crystals in uni-component and in solid

solution.

H. Kimura, A. Miyazaki / Journal of Crystal Growth 250 (2003) 251–255 255