Materials Letters 84 (2012) 16–19

Contents lists available at SciVerse ScienceDirect

Materials Letters

0167-57

http://d

n Corr

E-m

journal homepage: www.elsevier.com/locate/matlet

Weak ferroelectricity of potassium niobate K4Nb6O17 singlecrystal grown by pulling down technique

Hideo Kimura n, Rumi Tanahashi, Hongyang Zhao, Qiwen Yao

National Institute for Materials Science, Tsukuba, Ibaraki 305 0047, Japan

a r t i c l e i n f o

Article history:

Received 23 April 2012

Accepted 9 June 2012Available online 23 June 2012

Keywords:

Fiber crystal

Dielectric constant

Thermal analysis

Ferroelectrics

Piezoelectrics

7X/$ - see front matter & 2012 Elsevier B.V.

x.doi.org/10.1016/j.matlet.2012.06.052

esponding author. Tel.: þ81 29 859 2437; fax

ail address: kimura.hideo@nims.go.jp (H. Kim

a b s t r a c t

Although potassium niobate K4Nb6O17 single crystal is easy to absorb moisture due to the intercalation

of water molecules in the crystal, single crystals can be grown in fiber shape using an original pulling

down method. K4Nb6O17 single crystals are easy to break down in plate shape by cleavage which is in

turn affected by moisture. The dielectric constant and electric polarization measurements show that

K4Nb6O17 single crystal is with weak ferroelectricity.

& 2012 Elsevier B.V. All rights reserved.

1. Introduction

Potassium niobate KNbO3 single crystal (KN) is well known forits ferroelectric and piezoelectric properties [1–4]. Bulk KN can begrown from the melt in spite of its incongruent melting and twicesolid-state phase transformations [1]. During the KN crystalgrowth process, another crystal phase of K4Nb6O17 was some-times formed as a secondary phase caused by the potassiumevaporation in the KN melt [5]. The K4Nb6O17 crystal has alsobeen studied from the viewpoint of crystal structure [6,7] andcrystal growth [8,9]. The K4Nb6O17 crystal has an orthorhombicstructure of which its lattice parameters are a¼0.7816, b¼3.312and c¼0.6480 nm, and with space group P21nb (33) [7]. There is adisadvantage with the K4Nb6O17 crystal; it is easy to absorbmoisture (hydrated) to become K4Nb6O17 �3H2O [7,10]. From theviewpoint of electric properties, the K4Nb6O17 crystal wasreported as a ferroelectric nonlinear material [4], but its electricproperties have not been subjected to much study so far.

When the electric properties of KN crystal are measured, weare afraid to measure those of K4Nb6O17 crystal at the same time.So it is necessary to know electric properties of K4Nb6O17 crystalin order to know those of KN single crystal clearly. Of course,many defects will be expected in the K4Nb6O17 crystal. We believesingle crystals have less defects than ceramics.

In the present work, the growth of potassium niobate K4Nb6O17

single crystals, and characterization on the electric properties, such

All rights reserved.

: þ81 29 859 2401.

ura).

as dielectric constants, ferroelectric polarization (P–E) behavior, anda ferroelectric domain, will be reported.

2. Experimental procedures

A floating zone pulling down (Fz-PD) crystal growth method[11] was employed using a Pt tube with 4 mm diameter as the meltfeeder for the starting materials. A keen slant cut Pt wire with0.5 mm in diameter was used instead of seed crystal. This systemwas installed with conventional double ellipsoidal mirror halogenlamps and Pt tube. The potassium oxide and niobium oxide werestoichiometrically mixed as starting materials. Crystals were grownin Ar gas flow condition. The Pt wire’s pulling-down rate was10 mm/h without rotation. Schematics of crystal growth methodwere shown in our previous paper [12]. By using the conventionaldouble ellipsoidal mirror furnace, the crystals with big enough sizefor properties characterization can be grown.

For the characterization of these samples, cleavage crystalswere used as described later. Sample size was approximately2�2�0.2 mm3. The characterizations were carried out by X-raydiffraction (XRD) (Rigaku RINT2500) and Thermogravimetric–Differential thermal analysis (TG–DTA) using Shimadzu (DTG-60H). Ag paste was used for electrode fabrication. Dielectricconstant and loss were measured from 10 to 100 kHz by heatingup to 500 1C and cooling down process using an Agilent LCR meter(4984A). The ferroelectric hysteresis P–E loop was measured atroom temperature in silicone oil using an aixACCT (Easy Check300) ferroelectric tester and a Matsusada high-voltage amplifier.Ferroelectric domain was observed using a SII Piezo-responseForce Microscope (PFM: Nanocute).

Fig. 1. K4Nb6O17 crystals grown by the Fz-PD method: (a) As-grown crystal and (b) cleavage crystals. Scale: 10 mm.

Fig. 3. TG–DTA curves during heating and cooling: (a) first measurement at

heating and cooling and (b) second measurement at heating.

Fig. 4. Temperature dependence of dielectric constant, e, at heating and cooling

from 10 to 100 kHz: (a) first measurement, (b1) second measurement and (b2)

that of dielectric loss, D, at second measurement.

Fig. 2. X-ray diffraction pattern of K4Nb6O17 crystal: (a) Crystal powder and (b)

cleavage plane.

H. Kimura et al. / Materials Letters 84 (2012) 16–19 17

3. Results and discussion

K4Nb6O17 crystals were grown as shown in Fig. 1(a). Single-phase crystals were obtained with 1–2 mm in diameter and5–10 mm in length, in fiber-shape. Crystal’s grown directionwas basically in the c-axis (the shortest axis in the orthorhombicsystem). The grown crystals are reasonably stable in dry condi-tion. However, they were broken down to small pieces when

cut under conventional water-cooling cutting equipment, i.e., agrown crystal is easy to break down by cleavage in moisture asshown in Fig. 1(b). When cleaved crystal was observed undercross nicols, twin image was observed. We expect many otherlattice defects in the K4Nb6O17 crystal, but we do not care suchdefects so much in the present work because it is important toknow the electric properties.

The powder XRD pattern on the K4Nb6O17 crystal was obtainedafter milling. The powder single-phase XRD is shown in Fig. 2(a).Here, the index is based on Ref. [7]. All the peaks can be indexedas the K4Nb6O17 phase. The reason of large background signal atlow angle is due to the X-ray detector of D/teX by Rigaku. Themoisture absorption effect was not observed in the XRD measure-ment of K4Nb6O17 crystal. Fig. 2(b) shows the XRD pattern on thecleavage plane of K4Nb6O17 crystal in Fig. 1(b). The cleavage planeis determined to be (1 1 0) plane. This plane is not coinciding witha habitual plane of flux growth of (0 1 0) plane [8].

The TG–DTA curves are shown in Fig. 3. Fig. 3(a) shows the firstmeasurement during heating and cooling. Fig. 3(b) shows the secondmeasurement during heating just after the first measurement. Three

Fig. 5. (a) P–E hysteresis loops on cleavage K4Nb6O17 crystal. Inserted numbers are applied voltage and (b) PFM image.

H. Kimura et al. / Materials Letters 84 (2012) 16–1918

peaks on DTA and weight loss on TGA were observed around 70, 170and 250 1C at the first measurement. A small peak was observedaround 170 1C on the second measurement. The peaks around 70 and170 1C in Fig. 3(a) were caused by the releasing of the absorbedmoisture [14]. A third peak around 250 1C in the first measurement iscaused by a structure change from K4Nb6O17 �3H2O to K4Nb6O17 aswill be discussed later again. After the first measurement, someabsorbed moisture was still remained in the crystal. So a second peakwas still observed around 170 1C in Fig. 3(b). In other reports,inconsistent DTA measurements were reported on the K4Nb6O17

and the K4Nb6O17 �3H2O [13,14]. Amini and Mirzaee reported thatthe peak was observed at 50–150 1C on the K4Nb6O17 fabricated byhydrothermal method [15]. On the other hand, Nassau et al. reportedthat the peak was observed only on the K4Nb6O17 �3H2O but not onthe K4Nb6O17 in the second measurement that heated up to 1200 1C[14]. So such thermal properties are moisture dependent [13]. Weconcluded that there is no peak on the K4Nb6O17.

The temperature dependence of dielectric constant is shown inFig. 4 measured from 10 to 100 kHz. In the past reports up to200 1C, large peak was observed around 170 1C [14]. However, apeak was observed only in the first measurement as shown inFig. 4(a). A small step was observed around 250 1C. This step wasalso observed on the dielectric loss measurement. When thesecond measurement was carried out after the first measurement,the peak was no longer observed as shown in Fig. 4(b-1).According to the TG-DTA results in Fig. 3, the peak around170 1C is caused by the K4Nb6O17 �3H2O but not the K4Nb6O17

because no peak was observed at the second measurement asshown in Fig. 4(b-1). The step around 250 1C was caused by thestructure change from the K4Nb6O17 �3H2O to the K4Nb6O17 dueto a weight loss as observed in Fig. 3(a). In a past report [14], apeak around 170 1C was reported on both the K4Nb6O17 andK4Nb6O17 �3H2O crystals, but the peak on the K4Nb6O17 is smallerthan the one observed on the K4Nb6O17 �3H2O. Thus a peak wasdifficult to observe in the second measurement. We consider thatthe K4Nb6O17 was easily changed to the K4Nb6O17 �3H2O uponabsorbing moisture. In the past report [14], heating was carriedout up to 200 1C, meanwhile it is up to 500 1C in the present work.Higher temperature (more than 200 1C) is required to release themoisture completely. Thus it was concluded that the peak around170 1C was caused by the K4Nb6O17 �3H2O phase but not theK4Nb6O17 phase. Dielectric loss was small enough less than 0.1 asshown in Fig. 4(b-2).

The P–E hysteresis loop was shown in Fig. 5(a) on the cleavedK4Nb6O17 crystal (see in Fig. 2(c)). Newly cleaved surface was used

here. The applied voltage was in the range of 0.5 to 2 kV. Above 2 kV,the crystal was broken electrically. The moisture effect was notstudied. This result is for the K4Nb6O17 phase but not theK4Nb6O17 �3H2O phase, because the sample was used after theprocess of the 500 1C heating up treatment (as a result moisturewas completely released from the crystal). The polarization wasweak less than 1 mC/cm2 for a ferroelectric material. ConventionalKN single crystal has 20 mC/cm2 value at least [16]. Thus K4Nb6O17

crystal does not affect the ferroelectric properties of KN so much,even formed at the same time during the crystal growth. Theferroelectric–paraelectric phase transitions cannot be detected inthe dielectric constant and the TG–DTA results. Large leak current isexpected in spite of small dielectric loss in Fig. 4(b-2).

The PFM image under an out-of plane mode was shown inFig. 5(b) on the K4Nb6O17 crystal surface. Random ferroelectricdomain structure was observed. Thus the K4Nb6O17 crystal hasshown weak ferroelectricity. However, the Curie point of K4Nb6O17

crystal is not sure yet to be determined.

4. Summary

Potassium niobate K4Nb6O17 single crystals were grown usingan original pulling down method. Crystals were with transparentlook, but easy to break down as cleavage with (1 1 0) plane. Weakferroelectricity was observed. Since the crystals were easy toabsorb moisture, some peaks were observed on the dielectricconstant measurement and the thermo-gravimetric–differentialanalysis shows that the moisture affected the thermal properties.

Acknowledgments

Part of this work was supported by Grants from the JSPSGrant-in-Aid for Scientific Research (C) (21605012) and the JSPSFellowship (P09608 and P10796).

References

[1] Shirane G, Newnham R, Perinsky R. Phys Rev 1954;96:581–8.[2] Nakamura K, Tokiwa T, Kawamura YJ. Appl Phys 2002;91:9272–6.[3] Kestigian M, Leipziger FD, Carter JR, Garabedian FG. J Am Ceram Soc

1966;49:517.[4] Hirano H, Fukuda T. Jpn J Appl Phys 1968;7:1413.[5] Reisman A, Holtzberg F. J Am Chem Soc 1955;77:2115–9.[6] Gasperin M, Le Bihan MT. J Solid State Chem 1982;43:346–53.[7] Lundberg M, Sundberg M. J Solid State Chem 1986;63:216–30.

H. Kimura et al. / Materials Letters 84 (2012) 16–19 19

[8] Teshima K, Niina Y, Yubuta K, Suzuki T, Ishizawa N, Shishido T, Oishi S. Eur JInorg Chem 2007:4687–92.

[9] Zhang GK, He FS, Zou X, Gong J, Zhang H. J Phys Chem Solids 2008;69:1471–4.[10] Bizeto MA, Leroux F, Shiguihara AL, Temperini MLA, Sala O, Constantino VRL.

J Phys Chem Solids 2010;71:560–4.[11] Kimura H, Jia X, Shoji K, Sakai R, Katsumata T. J Cryst Growth 2000;212:

364–7.

[12] Kimura H, Miyazaki A, Maiwa K, Cheng ZX, Kannan CV. Opt Mater2007;30:198–200.

[13] Bizeto MA, Constantino VRL. Mater Res Bull 2004;39:1729–36.[14] Nassau K, Shiever JW, Bernstein JL. J Electrochem Soc 1969;116:348–53.[15] Amini MM, Mirzaee M. Ceram Int 2009;35:2367–72.[16] Kimura H, Tanahashi R, Zhao HY, Maiwa K. Cryst Res Technol 2011;46:37–40.