DOI: 10.1021/cg1010709
Unconventional Growth Mechanism in Optical Traveling Solvent Floating Zone Growth of Large β-CuNb2O6 Single Crystals
2011, Vol. 11 154–157
A. T. M. Nazmul Islam,*,† O. Pieper,† Bella Lake,†,‡ and K. Siemensmeyer† † Helmholtz-Zentrum Berlin f€ ur Materialien und Energie, Hahn-Meitner Platz 1, 14109 Berlin, ur Festk€ orperphysik, Technische Universit€ at Berlin, Hardenbergstrasse 36, German, and ‡Institut f€ 10623 Berlin, Germany
Received August 16, 2010; Revised Manuscript Received November 10, 2010
ABSTRACT: A large single crystal (5 mm 4 mm 15 mm) of CuNb2O6 has been grown in a four-mirror-type optical floatingzone furnace by the traveling solvent floating zone (TSFZ) method using a composition having excess CuO as flux. The as-grown single crystal was found to be phase-pure orthorhombic β-CuNb2O6, excellent in quality and with a spontaneous growth direction along the crystallographic b-axis. A novel growth mechanism as observed for β-CuNb2O6 in the floating-zone machine is illustrated, and characteristics of the magnetic susceptibility and magnetization along different crystallographic axes are also shown.
Introduction CuNb2O6 belongs to the series of binary niobate ceramics with the formula M2þNb2O6, where M2þ can be Ca, Zn, Mg, Co, Ni, Mn, Cu, Cd, and Fe having an orthorhombic columbite structure.1 There is increasing interest in these materials for use in a diverse range of applications such as in dielectric resonators as a high-performance, low-cost microwave dielectric ceramic.1,2 CuNb2O6 can also potentially be used as a gas sensor because its resistivity changes considerably on exposure to reducing gases.3 Different columbite niobates have been investigated as cathodes for lithium batteries4 and for their strong mechanoluminescent characteristics.5 CuNb2O6, with quasi-one-dimensional alternating chains of magnetic Cu2þ ions (spin = 1/2) is also interesting for fundamental study of quantum magnetism since exotic properties at low temperatures, such as spin-gap, spin-singlet ground state, and free and bound spinon excitations, are expected in such systems. This material has two polymorphs. The structure of both polymorphs R- and β-CuNb2O6 consists of edgesharing NbO6 octahedra. However, in β-CuNb2O6 with an orthorhombic unit cell, the zigzag spin-1/2 chains run along the c-axis, whereas in monoclinic R-CuNb2O6, the Cu2þ chains lie along the a-axis.6 Both phases have been investigated by muon spin relaxation,6 powder diffraction, NMR,7 inelastic neutron scattering,8,9 dc magnetic susceptibilities, etc. The results indicate that the R-phase has ferromagnetic (F)-antiferromagnetic (AF) alternating chains and is characterized by a spin singlet ground state with a spin gap of ∼20 K between the singlet and triplet excited states. On the other hand, there is AF ordering with TN ≈ 7.5 K in the β-phase.6,10 In order to study, the dynamic magnetic properties of this system in more detail, high-quality single crystals are essential. There are reports on growth of single crystals of other niobates such as M2þNb2O6, M2þ = Mg, Zn, or Ba from flux11 and M2þ =Ni, Co, Fe, or Mn by the optical floating-zone technique.12 Small single crystals could be grown with a substantial amount (10 mol %) of impurity of Zn substituting for Cu in *To whom correspondence should be addressed. Tel: þ49 (0)30 8062 42724. E-mail:
[email protected]. pubs.acs.org/crystal
Published on Web 12/01/2010
CuNb2O6.9 However, there is no report on growth of phasepure CuNb2O6 single crystals until now. The other niobates M2þNb2O6 (M2þ=Ni, Co, Fe, Mn), which are all congruently melting compounds, could be grown from high temperature melts of stoichiometric composition. In contrast, CuNb2O6 is an incongruently melting compound13 and hence posed a greater challenge. A suitable self-flux is required as a solvent for a traveling solvent floating zone growth of this material. Here we report the first growth of large single crystals of β-CuNb2O6 in an optical floating-zone (FZ) furnace using excess CuO as solvent. We observe that the growth stage looks different from a conventional traveling-solvent floatingzone growth. A new growth mechanism was proposed and illustrated here for growth of CuNb2O6 single crystals in the FZ machine. Experimental Procedures Initially a conventional traveling-solvent floating-zone (TSFZ) growth was planned to produce CuNb2O6 single crystals in an optical floating-zone machine based on the CuO-Nb2O5 binary phase diagram13 as described below. This technique as later modified is discussed in the Results and Discussion section. For feed rod preparation, high-purity powders of Nb2O5 (Puratronic 99.9985%, Alfa Aesar) and CuO (Puratronic 99.995%, Alfa Aesar) were mixed thoroughly in the 1:1 molar ratio. After mixing, powder of stoichiometric composition was calcinated in an alumina crucible in air at 700 °C for 24 h. Another batch of powder with excess CuO having a composition of 40 mol % Nb2O5-60 mol % CuO was prepared in the same way by solid state reaction for use as a solvent. To prepare a feed rod, both powders were pulverized, and then a small amount (less than a gram) of solvent powder was packed into a cylindrical rubber tube making a length of about 6-8 mm. On top of this, powder of stoichiometric composition was packed in the tube making the total length of ∼7-8 cm. The rubber tube with powder was then pressed hydrostatically up to 3000 bar in a cold isostatic pressure (CIP) machine. We found that a feed rod of stoichiometric composition shatters when sintered at temperatures above 950 °C. So the feed rod was then sintered at 850 °C in air for 12 h. A cylindrical rod with diameter about 6 mm and length 7-8 cm was prepared by this process. Growth was carried out in an optical image furnace (CSI FZ-T10000-H-VI-VP, Crystal Systems, Inc., Japan) equipped with four r 2010 American Chemical Society
Article
Crystal Growth & Design, Vol. 11, No. 1, 2011
155
Figure 2. Scanning electron microscope image of the cross-section of the as-grown material. A quantitative EDX analysis on different parts of the material is shown in the inset.
Figure 1. (a) Snap-shot of the growth state in the floating-zone machine. (b) As-grown material. 150 W tungsten halide lamps focused by four ellipsoidal mirrors. The feed rod was suspended from the upper shaft using nickel wire, while another small feed rod was fixed to the lower shaft to support the melt. Crystal growths were performed in flowing air at a typical growth rate of about 1-3 mm/h. A seed crystal from a previous growth was used for subsequent crystal growths to avoid random nucleation and thus obtain one large single crystal. After the growth, different parts of the material were ground, and X-ray diffraction (Huber Guinier G670) was performed on the powders using Cu KR1 radiation (λ=1.5406 A˚). The powder samples were mounted on a flat sample holder, oscillating continuously in the horizontal plane to minimize preferred orientation effects, and measured for 16 h each. The powder patterns were refined by the Rietveld method to determine phase purity, structure, etc. A piece of single crystal was also polished and checked with energy-dispersive X-ray analysis (EDX) integrated to a scanning electron microscope (SEM) for any residual phase, grain boundaries, or inclusions. Single crystals of CuNb2O6 were also checked with both X-ray and neutron Laue diffraction. In addition, temperature-dependent susceptibility and field-dependent magnetization measurements were performed in a SQUID magnetometer on a CuNb2O6 single crystal with fields along all three crystallographic axes.
Results and Discussion Figure 1a shows a snapshot during growth in a floatingzone furnace. The material had very low surface tension after melting. For this reason, the feed rod had to be moved downward continuously at 3 mm/h while the mirrors are moving upward at 3 mm/hour, in order to maintain an uninterrupted and stable growth. We observed that the molten zone lost its volume in a short time after starting the growth. Due to low surface tension of the melt, the solvent flowed downward instead of attaching onto the feed rod. There was also no penetration of solvent into the feed rod by capillary effect. So, when the CuO-excess solvent was lost from the molten zone, the growth had to stop. This situation is completely in contrast to other materials grown by the TSFZ method using excess CuO in solvent such as the high-Tc superconductors, La2-xMxCuO4 (M = Sr, Ba), Nd1-xCexCuO4.14,15 In order to compensate for the rapid loss of excess CuO from the zone, in the subsequent growth, we prepared a feed
rod (L = ∼10-12 cm, D = 6-7 mm) of uniform composition having excess CuO throughout its length. The crystal growth was performed at rate of 3 mm/h, along with the feed-rod feeding rate of -3 mm/h. The optimum composition of the feed rod was found to be 43 mol % Nb2O5-57 mol % CuO. The as-grown cylindrical shaped material was about 8 mm in diameter and 30 mm in length as shown in Figure 1b. A slice of the crystal was cut perpendicular to the growth direction, polished mechanically, and checked with a polarized optical microscope and SEM with an EDX. Near the surface, the as-grown material seems to have a mixed phase with many grain boundaries (Figure 2), whereas about half a millimeter inside the surface, the crystal was found to be free of any grain boundaries or inclusions of secondary phase. A quantitative analysis on the two areas shows that Cu concentration is significantly higher in the areas near the surface (Figure 2, inset). X-ray powder diffraction patterns taken with powders obtained separately by grinding a piece of material from the surface and a piece from inside are shown in Figure 3 as red dots. The powder patterns have been refined by the Rietveld method using the FULLPROF software.16 The peak profiles were modeled with a pseudo-Voigt function convoluted with asymmetry due to axial divergence. The black line indicates the calculated data, and the green bars indicate the calculated Bragg positions. For the powder made from the inner part of the crystal, an excellent fit of the data (RB = 0.0715, Rp = 0.0736, Rwp = 0.0997) was obtained as shown in Figure 3a. The crystal was found to be single-phase orthorhombic β-CuNb2O6, without any detectable impurity phase within the accuracy of the measurement. The refined structural parameters are as follows: space group Pbcn, a = 14.1198(1) A˚, b=5.61480(4) A˚, c = 5.12910(4) A˚, R = β = γ = 90°; in good agreement with previous X-ray diffraction measurements.17 In contrast, the powder ground from outer surface of the crystal was found to contain a considerable amount (∼34%) of an impurity phase as revealed by additional peaks indicated by the arrows in the inset of Figure 3b. The lower lying green bars represent the Bragg positions for the impurity phase, which was unambiguously identified as Cu3Nb2O8. The surface with the mixed phase could be cleaved off using a surgical blade, and a clear shiny surface of crystalline CuNb2O6 was exposed as shown in Figure 4a. An X-ray Laue photograph taken in the back reflection geometry with the beam perpendicular to the cleaved surface is shown in Figure 4b. The single crystallinity of the material was also
156
Crystal Growth & Design, Vol. 11, No. 1, 2011
Figure 3. X-ray powder diffraction patterns (red dots) of crushed materials from (a) single crystal inside and (b) surface of the material grown. The black line indicates the calculated data, the blue line represents the difference between data and calculation, and the green bars indicate the calculated Bragg positions. For the surface material (b), an additional Cu2Nb3O8 impurity phase was refined as indicated by the arrows in the inset. The lower lying green bars represent the Bragg positions of the impurity.
Islam et al.
Figure 5. Illustration of growth mechanism of β-CuNb2O6 single crystal in a floating-zone machine.
Figure 4. (a) A CuNb2O6 single crystal showing cleaved surface with metallic luster after removal of surface impurity. (b) X-ray Laue diffraction pattern taken from the cleaved surface shown with simulated spots (red dots). (c) Neutron Laue backscattering diffraction photograph confirms that the whole material (5 mm 4 mm 15 mm) is one single crystal.
confirmed with the automatic neutron Laue diffractometer known as Orient Express18 at the Institur Laue-Langevin, Grenoble, France, as shown in Figure 4c. The orientation of the crystal determined from the Laue diffraction patterns is shown in Figure 4a and reveals that the β-CuNb2O6 crystal grows naturally along the crystallographic b-axis. The growth mechanism of CuNb2O6 single crystals from feed rod with excess CuO as illustrated in Figure 5 is as follows: Step 1. Melting;Lower tip of the feed rod melts in the infrared light focused zone. Step 2. Crystallization;After seeding of melt on the sintered rod attached to the lower shaft, the mirror-lamp system is moved upward. CuNb2O6 starts crystallizing on top of the sintered rod. Step 3. Change of composition in molten zone;As the CuNb2O6 crystal continues to grow, the composition of the molten zone shifts toward even higher CuO content.
Figure 6. (a) Temperature dependence of susceptibility of CuNb2O6 single crystal with field applied parallel to different crystallographic axes. (b) Field dependence of magnetization of CuNb2O6 single crystal with fields up to 5 T along different crystallographic axes.
Step 4. Reaction in molten zone;When the composition of CuO in the molten zone is sufficiently high, excess CuO reacts with CuNb2O6 to form some amount of Cu3Nb2O8. Cu3Nb2O8 has a lower melting point than CuNb2O6 and at similar temperatures has an even lower surface tension than CuNb2O6. So, Cu3Nb2O8 flows down the edges of the CuNb2O6 crystal and solidifies around the CuNb2O6 crystal at a lower position.Downward flow of Cu3Nb2O8 removes excess CuO from the melt, and the composition of the molten-zone changes back to the initial state in step 2. Consequently the CuNb2O6 crystal continues to grow on
Article
Crystal Growth & Design, Vol. 11, No. 1, 2011
top of the previous crystal and then steps 3 and 4 follow. With this mechanism, we were successful in growing one large (L=15 mm, D=5 mm) single crystal of CuNb2O6 covered by a thick shell (∼0.5 mm) of Cu3Nb2O8 on the cylindrical surface. We have measured the temperature dependence of the magnetic susceptibility of the single crystal samples with a field of 0.1 T applied along the a, b, and c axes as shown in Figure 6a. The susceptibilities above 35 K follow CurieWeiss behavior independent of orientation with a Curie temperature of 31 K in agreement with previous susceptibility measurements on powder samples. Below this temperature, χ continues to increase, reaching a maximum at 18 K, and then falls rapidly with decreasing temperature. Below the Neel temperature at ∼7 K, the susceptibility along a continues to drop indicating that the spins point along this direction, while the susceptibility along b and c becomes constant (see inset of Figure 6a). The susceptibility is typical of that of a lowdimensional antiferromagnet and the large difference between the Curie-Weiss temperature and the Neel temperature indicates the suppression of order due to quantum fluctuations. Figure 6b shows the magnetization of the CuNb2O6 single crystal at 5 K for fields up to 5 T applied along a, b, and c axes. We observe that the magnetization along the b and c axes is linear in nature, whereas the magnetization along the a axis has anomalies around 1.5 and -1.5 T, which are probably due to a spin flop transition in the material, thus confirming that the spins order along a. Altogether these measurements reveal that β-CuNb2O6 is a candidate material for the study of quantum magnetism. Conclusion We have grown large single crystals of β-CuNb2O6 in an optical floating-zone furnace, where defying conventional wisdom of TSFZ growths a feed rod having a nonstoichiometric composition with excess CuO throughout its length was used. We found that growth of a large single crystal from such a feed rod was possible due to an unusual growth mechanism as illustrated in this work. The single crystal was
157
found to be of highest quality as characterized by polarized optical microscopy, EDX, X-ray powder diffraction, singlecrystal neutron and X-ray Laue diffraction measurements, dc magnetic susceptibility, and magnetization. Acknowledgment. The authors thank D. Argyriou for help with the lab facilities. The authors also thank M.-H. LemeeCailleau and B. Ouladdiaf for help with the Neutron Laue instrument, Orient Express at ILL, R. Marx for help with the XRD measurements, and H. Kropf for the SEM and EDX measurements.
References (1) Pullar, R. C. J. Am. Ceram. Soc. 2009, 92 (3), 563. (2) Pullar, R. C.; Breeze, J. D.; Alford, N. McN. J. Am. Ceram. Soc. 2005, 88, 2466. (3) Biswas, S. K.; Pramanik, P. Sensors Actuators B 2008, 133, 449. (4) Martinez-de la, C. A.; Alcaraz, N. L.; Fuentes, A. F.; TorresMartinez, L. M. J. Power Sources 1999, 81, 255. (5) Sakuraba, D.; Toda, K.; Uematsu, K.; Sato, M. Key Eng. Mater. 2004, 269, 99. (6) Krishnamurthy, V. V.; Nagamine, K.; Nishiyama, K.; Ishikawa, M.; Yamaguchi, M.; Watanabe, I.; Ishikawa, T.; Das, T. P. Phys Rev. B 2003, 68, No. 014401. (7) Fukamachi, T.; Kobayasi, Y.; Kanada, M.; Kasai, M.; Yasui, Y.; Sato, M. J. Phys. Soc. Jpn. 1998, 67, 2107. (8) Kodama, K.; Harashina, H.; Sasaki, S.; Kanada, M.; Kato, M.; Sato, M.; Kakurai, K.; Nishi, M. J. Phys. Chem. Solids 1999, 60, 1129. (9) Kodama, K.; Harashina, H.; Sasaki, S.; Kanada, M.; Kato, M.; Sato, M.; Kakurai, K.; Nishi, M. J. Phys. Soc. Jpn. 1999, 68, 237. (10) Mitsuda, S.; Miyamoto, J.; Kobayashi, S.; Miyatani, K.; Tanaka, T. J. Phys. Soc. Jpn. 1998, 67, 1060. (11) Greenblatt, M.; Wanklyn, B. M.; Garrard, B. J. J. Cryst. Growth 1982, 58, 463. (12) Prabhakaran, D.; Wondre, F. R.; Boothroyd, A. T. J. Cryst. Growth 2003, 250, 72. (13) Sirotinkin, V. P.; Drozdova, N. M. Russ. J. Inorg. Chem. 1992, 37, 1334. (14) Tanaka, I.; Yamane, K.; Kojima, H. J. Cryst. Growth 1989, 96, 711. (15) Kojima, H.; Watanabe, T.; Komai, N.; Tanaka, I. Mol. Cryst. Liq. Cryst. 1990, 184, 69. (16) Rodriguez-Carvajal, J. Physica B 1993, 192, 55. (17) Norwig, J.; Weitzel, H.; Paulus, H.; Lautenschlager, G.; RodriguezCarvajal, J.; Fuess, H. J. Solid State Chem. 1995, 115, 476. (18) Ouladdiaf, B.; Archer, J.; McIntyre, G. J.; Hewat, A. W.; Brau, D.; York, S. Physica B 2006, 385-386, 1052.