Crystal Engineering 5 (2002) 355364 www.elsevier.

com/locate/cryseng

Hexagonal rare earth (R=Eu-Dy) manganites: XRD study of epitaxially stabilized lms
nateur a, A.A. Bosak ab,, C. Dubourdieu a, J.-P. Se O.Yu. Gorbenko b, A.R. Kaul b
a

` res, France Laboratoire des Mate riaux et du Ge nie Physique, INPG, BP 46, 38402 St. Martin dHe b Department of Chemistry, Moscow State University, 119899 Moscow, Russia

Abstract Epitaxial thin lms of RMnO3 (R=Sm, Eu, Gd, Dy) were prepared using liquid injection MOCVD and studied by XRD. The formation of high-temperature hexagonal polymorphic structure of DyMnO3 and non-existing-as-bulk hexagonal EuMnO3, GdMnO3 and SmMnO3 (traces) was observed on ZrO2(Y2O3) (111) substrates at 900 C due to epitaxial stabilization. For the thickness exceeding the critical one the oriented stable perovskite form grew on the hexagonal phase. Parallel deposition on perovskite or non-coherent substrates gave only the stable perovskite phase, epitaxially grown in the former case or oriented in the latter. 2003 Elsevier Science Ltd. All rights reserved.
Keywords: MOCVD; Thin lms; Manganites; Hexagonal structure; Epitaxial stabilization

1. Introduction Stable hexagonal LuMnO3-type structures (space group P63cm) have been found for RMnO3 compounds where R is a trivalent rare earth cation with small ionic radius (Ho-Lu, Y, Sc). This structure can be described as a dense oxygen-ion packing (ABCACB) with Mn3+ ions having coordination number CN=5 (ve-fold trigonal bipyramidal coordination), and R3+ with CN=7 (seven-fold monocapped octahedral coordination) [1]. YMnO3 which belongs to this structural type is considered to be a potential ferroelectric material for electronic applications [2]. A metastable perovskite phase with space group Pnma can be obtained instead of

Corresponding author. Tel.: +33-(0)4-76-82-62-10; fax: +33-(0)4-76-82-63-94. E-mail address: Alexei.Bossak@inpg.fr (A.A. Bosak).

1463-0184/02/$ - see front matter 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S1463-0184(02)00047-3

356

A.A. Bosak et al. / Crystal Engineering 5 (2002) 355364

hexagonal RMnO3 by high pressure synthesis [3], by soft chemistry synthesis [4] or by the epitaxy on perovskite substrates [5]. The requirement of a high pressure can be easily understood taking into account the signicant decrease of the unit cell volume (89%) from hexagonal to perovskite structure [1], as shown in Fig. 1. When considering the 3d series, the stable hexagonal structure is only observed for RMnO3 compounds, but metastable hexagonal phases are known for some gallates and aluminates [4,9,10]. It is interesting to note that for both perovskite and hexagonal series a nearly linear relationship is observed between the renormalized cell volume and the Goldschmidt tolerance factor t as shown in Fig. 1 and there is no geometrical interdiction of hexagonal phase formation in an extended t area. It is worth noting that t remains a relevant parameter for the considered hexagonal structures. The formation criteria should be of energetic nature and correlated with the energy difference between octahedra and less symmetrical polyhedra of 3d ions (like trigonal bipyramide) [8]. For DyMnO3, which stable modication is perovskite, a hexagonal phase was obtained by quenching from high temperature (1600o C) [4]. A similar behavior was observed for YMnO3 (crystals of both types were grown from bismuth ux in nearly the same conditions). But for larger rare-earth ions like Gd3+ direct extrapolation gives a too high phase transition temperature (about 2800 C, which is well above the melting point of manganites). We supposed [5] that the most suitable way for the synthesis of metastable hexagonal manganites would be epitaxial stabilisation, as in the case of BaCu3O4 [11], RNiO3 [12], NdMn7O12 [13], LuCoO3 [14]. The ionic radii (CN=9) of R3+ for the studied systems and low-temperature as well as

Fig. 1. Dependence of the normalized unit cell volume per formula unit vs. tolerance factor (JCPDS data and [4,9,10]).

A.A. Bosak et al. / Crystal Engineering 5 (2002) 355364

357

Table 1 Ionic radii of R3+ (CN=9) [6,7]; low-temperature (LT) and high-temperature (HT) forms of the corresponding RMnO3 bulk compounds. The studied systems are marked by R3+ Nd Sm Eu Gd Tb Dy Ho ) rCN=9 (A 1.163 1.132 1.120 1.107 1.095 1.083 1.072 LT form perovskite perovskite perovskite perovskite perovskite perovskite hexagonal HT form perovskite perovskite perovskite perovskite perovskite hexagonal hexagonal

high-temperature stable forms of the corresponding RMnO3 bulk compounds are given in Table 1.

2. Experiment The calculations made using available data for stable hexagonal phases show that ZrO2(Y2O3) (111) has in-plane lattice parameters which are the closest to the parameters of the hypothetical hexagonal phases (RRDy) (Fig. 2) and excellent coincidence of oxygen crystallographic positions at the interface. The deposition runs were performed using a liquid injection MOCVD process [15,16]. The deposition tempera-

Fig. 2. Calculated lattice mismatch of RMnO3 hexagonal compounds with ZrO2(Y2O3) (111) (YSZ) substrate as a function of R3+ ionic radius the studied rare-earths are represented by closed symbols. The mismatch is given as M = 12ahex / 3aYSZ.

358

A.A. Bosak et al. / Crystal Engineering 5 (2002) 355364

ture was 900 C, the oxygen partial pressure was 0.33 kPa, and the total pressure was 0.67 kPa (the deposition rate was about 5 nm/min). Mn(thd)3 and R(thd)3, where thd=2,2,6,6-tetramethylheptane-3,5-dionate, were used as precursors (0.02 M solution in monoglyme, the molar ratio R/Mn in solution producing stoichiometric lms was about 1.4). Deposition of lms with approximate thickness 6070 nm was performed simultaneously on ZrO2(Y2O3) (111), ZrO2(Y2O3) (001), LaAlO3 (001) (pseudocubic notations), and MgO (001) substrates. The lms were characterized by X-ray diffraction (XRD) using a Siemens D5000 four-circle diffractometer (equipped with secondary graphite monochromator).

Fig. 3. /2 scans for GdMnO3 lms deposited on four different substrates: (a) (111) ZrO2(Y2O3), (b) (001) LaAlO3, (c) (001) MgO, (d) (001) ZrO2(Y2O3). Indexation is performed in hexagonal space group for (a) and orthorhombic for (b), (c), (d).

A.A. Bosak et al. / Crystal Engineering 5 (2002) 355364

359

3. Results and discussion 3.1. Perovskite lms on ZrO2(Y2O3) (001), MgO (001) and LaAlO3 (001) All the lms of SmMnO3, DyMnO3, GdMnO3 and EuMnO3 deposited on MgO and ZrO2(Y2O3) (001) are polycrystalline but strongly textured and exhibit a perovskite structure. The XRD patterns for GdMnO3 deposited on the four different substrates is given in Fig. 3. Films on LaAlO3 are epitaxial (cube-on-cube growth as seen in Fig. 4(a)) and consist also of perovskite only. X-ray /2 scans measured with tilted samples reveal superstructure reections of the perovskite lattice that are typical for the orthorhombic Pnma structure. The lattice mismatch on LaAlO3 for our series has a minimum for DyMnO3 and is much lower for (101) oriented domains, than for (010) domains. The lattice parameters were determined from a series of off-plane /2 scans. They are rather close to those of bulk (Fig. 4(b)), which is consistent with the fact that the large orthorhombic distortion of manganites cannot be suppressed by interface strain for our lms thickness, whereas the interface strain can be accommodated by an appropriate domain repartition. All the main features of epitaxial perovskite lms are in trend with the general tendencies for Pnma manganites of the small rare-earth ions evidenced in our previous work [5]. The following epitaxial relationships are obtained on LaAlO3: (010) RMnO3 / / (001) LaAlO3; [101] RMnO3 / / [100] LaAlO3.

Fig. 4. Epitaxial perovskite growth on LaAlO3 (001): (a) -scans for substrate and DyMnO3 orthorhombic lm on top of it; (b) lattice parameters of deposited perovskite phases (~70 nm) in comparison to the JCPDS data.

360

A.A. Bosak et al. / Crystal Engineering 5 (2002) 355364

3.2. Hexagonal lms on ZrO2(Y2O3) (111) The lms deposited on ZrO2(Y2O3) (111) substrates in the same deposition runs as those mentioned above reveal different diffraction patterns (Figs. 3 and 5). Observed reections (both symmetric (001) and asymmetric (111), (112)) are typical for the hexagonal LuMnO3-type structure. The hexagonal phase is c-oriented which was conrmed by -scans and pole gures. Epitaxial relations are established as follows: 0]hex.RMnO3 / / [11 0]ZrO2(Y2O3). (001)hex.RMnO3 / / (111)ZrO2(Y2O3); [11 Fig. 6 illustrates the case of DyMnO3 with the {112} pole gure and the sketch of epitaxy. It is observed (Fig. 5) that the hexagonal c parameter of the lms decreases with the increase of rare earth ionic radii, which can be explained by strain effect. The a parameter was measured from off-plane /2 scans. It appears that the lms are stretched in-plane, when comparing the a values to those extrapolated for bulk

Fig. 5.

/2 XRD patterns for RMnO3 lms deposited on ZrO2(Y2O3) (111) substrate.

A.A. Bosak et al. / Crystal Engineering 5 (2002) 355364

361

Fig. 6. Texture study of hexagonal DyMnO3 lm on ZrO2(Y2O3) (111): (a) pole gure of {112} zone of hexagonal DyMnO3; (b) scheme of epitaxy.

Fig. 7. Strained state of hexagonal RMnO3 lms: (a) unit cell parameters; (b) unit cell volume; open symbols represent literature data for bulk and closed symbols are for our experimental data.

samples (Fig. 7(a)). However, the unit cell volume remains close to that extrapolated from data for heavier rare earths (Fig. 7(b)). As shown in Fig. 8(a), the intensity of (001) reections decreases from Dy to Eu and nearly vanishes for Sm (the ngerprint of the hexagonal phase is its (002) peak well separated from any other) while, at the same time, the perovskite reections intensity increases. Actually, the DyMnO3 appears as a nearly pure hexagonal phase, whereas SmMnO3 appears as a nearly pure perovskite phase. The coexistence of oriented hexagonal and perovskite phases for R=Eu, Gd was proven by pole gures for corresponding zones of both structures. The perovskite phase growths as follows: ]ZrO2(Y2O3), (002) RMnO3 / / (111)ZrO2(Y2O3); [100] RMnO3 / / [112

362

A.A. Bosak et al. / Crystal Engineering 5 (2002) 355364

Fig. 8. Perovskite overgrowth on the hexagonal phase: (a) intensities of characteristic diffraction peaks of perovskite and hexagonal phases; (b) pole gure of {121}+{210} zones of perovskite SmMnO3; (c) calculated model of pole gure; (d) relative orientations of ZrO2(Y2O3) (111) substrate and RMnO3 perovskite.

] ZrO2(Y2O3). (200) RMnO3 / / (111) ZrO2(Y2O3); [001] RMnO3 / / [112 As an example, the experimental and model pole gures for the SmMnO3 perovskite are given in Fig. 8(b), Fig. 8(c) together with a sketch of epitaxy (Fig. 8(d)). Noteworthy, the increase of GdMnO3 lm thickness up to 300 nm results in the increase of the intensity of the perovskite reections, with no further increase of the intensity of the hexagonal phase reections. Thus, the hexagonal phase grows in the vicinity of the lm/substrate interface and the perovskite component forms in the upper part of the lm; 70 nm seems to be nearly the critical thickness for GdMnO3. This change of structure with lm thickness can be explained by the fact that the contribution of interface effects decreases as the thickness increases, thus being no more efcient for hexagonal phase stabilization. Following generalized thermodynamic estimation [11], when the energy gap between equilibrium and non-equilibrium states does not exceed 1015 kJ/mole, the latter can be stabilized by the free energy gained from diminishing the surface energy

A.A. Bosak et al. / Crystal Engineering 5 (2002) 355364

363

Fig. 9. Free energy of RMnO3 formation from R2O3 and Mn2O3 at 900 C (calculated using oxygen dissociation pressure data [17]).

term due to coherent interface formation. For our systems the stabilization energy can be estimated using the linear extrapolation of bulk RMnO3 formation energy dependencies vs. ionic radius (Fig. 9). It attains the maximum value of 10 kJ/mole for SmEu. Noteworthy, the extension of the region of epitaxial stabilization is nearly the same for the hexagonal phases as found for perovskite manganites [5]. We believe that this effect results in the appearance of high-temperature or even non-existing phases in our deposition conditions. As the contribution of the surface energy is inversely dependent on the thickness, the stabilizing effect can only be observed for the lms of limited thickness [11]. Already GdMnO3 thin lms do not crystallize completely in a pure hexagonal phase, the upper layer is indeed of perovskite structure. This tendency is even more pronounced for EuMnO3. 4. Conclusion We have succeeded in the preparation of epitaxial lms of hexagonal RMnO3 (R=Dy, Gd, Eu, Sm) by low-pressure MOCVD. For R=Gd, Eu, Sm these phases were not known before and apparently cannot exist as bulk. For Dy, the decrease of the hexagonal phase formation temperature is about 700 C (900 C for thin lm instead of 1600 C for the bulk phase). These phases can be formed with the hexagonal structure (instead of the stable perovskite one) because of its epitaxial stabilization on the oxide substrates with uorite structure of the proper orientation (111). Magnetic and ferroelectric properties of new phases are still to be investigated.

364

A.A. Bosak et al. / Crystal Engineering 5 (2002) 355364

References
[1] [2] [3] [4] [5] H.L. Yakel, W.C. Koehler, E.F. Bertaud, E.F. Forrat, Acta. Cryst. 16 (1963) 957. N. Fujimura, T. Ishida, T. Yoshimura, T. Ito, Appl. Phys. Lett 69 (1996) 1011. A. Waintal, J. Chenavas, Mat. Res. Bull. 2 (1967) 819. ` se University of Lyon, Lyon, France, 1969 G. Szabo, The A.A. Bosak, A.A. Kamenev, I.E. Graboy, S.V. Antonov, O.Yu. Gorbenko, A.R. Kaul, C. Dubourdieu, J.-P. Senateur, V.L. Svechnikov, H.W. Zandbergen, B. Holla nder, Thin Solid Films 400 (2001) 149153. R.D. Shannon, C.T. Prewitt, Acta. Cryst. B 25 (1969) 925. R.D. Shannon, Acta. Cryst. A 32 (1976) 751. L.A. Reznitskii, Inorganic Materials 20 (1984) 743. S. Geller, J.B. Jeffries, P.J. Curlander, Acta. Cryst. B31 (1975) 2770. ris, C.R. Acad. Sc. Paris 268 (1969) 1050. G. Sallavuard, G. Szabo, R.-A. Pa S.V. Samoylenkov, O.Yu. Gorbenko, I.E. Graboy, A.R. Kaul, H.W. Zandbergen, E. Connolly, Chem. Mater 11 (1999) 2417. M.A. Novojilov, O.Yu. Gorbenko, I.E. Graboy, A.R. Kaul, H.W. Zandbergen, Appl. Phys. Lett 76 (2000) 2041. A.A. Bosak, O.Yu. Gorbenko, A.R. Kaul, I.E. Graboy, C. Dubourdieu, J.-P. Senateur, H.W. Zandbergen, J. Magn. Magn. Mater 211 (2000) 61. M.A. Novojilov, A.R. Kaul, G.E. Adamov, I.E. Graboy, Thin Solid Films, submitted for publication. nateur, R. Madar, F. Weiss, O. Thomas, A. Abrutis, French patent FR 2707671 (1993), J.-P. Se European patent EP 730671 (1994), US patent US 945162 (1999). nateur, C. Dubourdieu, F. Weiss, M. Rosina, A. Abrutis, Advanced Materials for Optics and J.-P. Se Electronics 10 (2000) 155. T. Atsumi, T. Oghushi, N. Kamegashira, J. of Alloys and Compounds 238 (1996) 35.

[6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17]