- Email: [email protected]
Deposition and x-ray characterization of epitaxial thin films of LaAlO3
Thin Solid Films 550 (2014) 90–94
Contents lists available at ScienceDirect
Thin Solid Films journal homepage: www.elsevier.com/locate/tsf
Deposition and x-ray characterization of epitaxial thin films of LaAlO3 Henrik Hovde Sønsteby ⁎, Erik Østreng, Helmer Fjellvåg, Ola Nilsen University of Oslo, Department of Chemistry, Centre for Materials Science and Nanotechnology, P.O. Box 1033 Blindern, N-0315 Oslo, Norway
a r t i c l e
i n f o
Article history: Received 19 April 2013 Received in revised form 24 September 2013 Accepted 18 October 2013 Available online 31 October 2013 Keywords: Atomic layer deposition Lanthanum aluminate Synchrotron diffraction Interface effects
a b s t r a c t Highly epitaxial thin films of lanthanum aluminate (LaAlO3) have been obtained on strontium titanate (SrTiO3) substrates by means of atomic layer deposition using La(thd)3 (Hthd = 2,2,6,6-tetramethylhepta-3,5-dione), Al(CH3)3 and ozone as precursors. The system shows a near linear relationship between pulsed and deposited composition. Thin films with stoichiometric composition have been subject to thermal annealing at 650 °C under oxygen atmosphere, thereby achieving epitaxial films on Ti-O-terminated substrates of SrTiO3. The thin film||substrate epitaxial relationship is determined to be LaAlO3(100)|LaAlO3[100]||-SrTiO3(100)|SrTiO3[100] by use of synchrotron radiation. Selected films were also deposited on LaAlO3(100) to achieve homoepitaxy. This resulted in the observation of split peaks for high q-reflections, pointing towards slight differences in stoichiometry. For ultrathin films, Bragg satellites were observed around the specular reflections, coming from either surface- or interface reconstruction. © 2013 Elsevier B.V. All rights reserved.
1. Introduction The insulating lanthanum aluminate (LaAlO3, LAO) with perovskite type structure (R-3c) has, due to its high-κ dielectric and thermal properties, been thoroughly studied for use in microelectronics and as a buffer layer in the growth of other perovskite type oxides [1–3]. In recent years, more exotic properties have been discovered, such as superconductivity on the interface between LAO and strontium titanate (SrTiO3, STO) [4]. Recent data provide evidence of the coexistence of superconductivity and ferromagnetism at this interface [5], a rarely encountered phenomenon. These exotic effects seem to stem from a polarity discontinuity at the LAO/STO-interface due to the heterosystem maintaining the AO–BO2 stacking sequence, giving rise to a polar LaO+/TiO02 interface. This in turn induces charge transfer effects and a 2D-electron gas [6,7] that exhibit superconductivity below 300 mK [8]. Careful investigations of interfacial structures and accompanying exotic effects require perfect interfacial morphology. This is commonly achieved by growth of a thin film of one compound on a substrate of the other terminated by a single structural species. Perovskite substrate surfaces do not naturally become single species terminated by simple cutting, however, a chemical route for TiO2-termination of STO has been described by Kawasaki et al. [9]. The desired heterostructural interface can thus be obtained by depositing LAO on such a pre-treated substrate of STO. This has previously been attempted using pulsed laser deposition [10], molecular beam epitaxy [11,12] and sputtering [13] among other techniques. This study aims at providing an easily achievable route to the LaO+/ TiO02-interface by depositing a highly oriented thin film of LAO on a pre⁎ Corresponding author. E-mail address: [email protected] (H.H. Sønsteby). 0040-6090/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.tsf.2013.10.139
treated, TiO2 terminated STO substrate. Corresponding LAO films are further deposited on non-pretreated STO substrates for comparison. The films are deposited using atomic layer deposition (ALD), which is a chemical vapor deposition related technique, and using commonly available precursors. ALD utilizes a layer-by-layer self-limiting growth mechanism providing control of interfacial chemistry, conformality and film thickness at the (sub-)monolayer level. Deposition of ternary oxides by ALD is commonly achieved by sequentially introducing volatile gaseous metal-organic precursors separated by an oxide precursor usually consisting of water or ozone [14]. Deposition of polycrystalline LAO by ALD is previously reported by Nieminen et al. [15] using La(thd)3 (Hthd = 2,2,6,6-tetramethyl-3,5heptanedione) and Al(acac)3 (acac = acetylacetonate) as precursors. Attempts using TMA (TMA = Al(CH3)3, trimethylaluminium) and La(i-PrCp)3 (i-PrCp = isopropyl cyclopentadienyl) as precursors have been made to provide highly oriented thin films of stochiometric LAO on STO [16], although lacking saturative conditions for the La(i-PrCpd)3 precursor. In this work we provide evidence of ALD growth of LAO using the precursor combination TMA/O3 and La(thd)3/O3, giving rise to a clearly surface controlled stoichiometry and very high structural orientation on STO-substrates. The obtained film structures have been subjected to a detailed analysis using synchrotron radiation at the Swiss–Norwegian Beam Lines at ESRF, Grenoble. 2. Experimental Thin films were deposited in a commercial F-120 Sat ALD-reactor (ASM Microchemistry) using La(thd)3 (Multivalent), TMA (technical grade) and O3 as precursors. Ozone gas was produced by an OT-020 (Ozone Technology) generator fed with pure O2 (99.999%, AGA) and used at an approximate flow rate of 500 cm3 min− 1. La(thd)3
H.H. Sønsteby et al. / Thin Solid Films 550 (2014) 90–94
was sublimated at 185 °C, whereas TMA was kept at room temperature in an external bubbler. A carrier-gas flow of 300 cm3 min− 1 was supplied from a Nitrox UHPN 3001 generator producing 99.9995% inert gas. The films were deposited on 2.5 × 2.5 cm2 substrates of Si(100) and soda lime glass and on 1.0 × 1.0 cm2 substrates of SrTiO3(100) LaAlO3(100) and MgO(100). In addition, some SrTiO3(100) substrates were etched for 10 min in a buffered NH4F–HF solution (Sigma-Aldrich) to achieve Ti–O-termination as described by Kawasaki et al. [9]. All substrates were further cleaned by pulsing O3 prior to the depositions. A number of La:Al pulsing ratios were investigated while keeping the total amount of metal pulses as close as practically possible to 1000. The resulting films had thicknesses between 40 and 80 nm. Selected depositions totaling 100 metal pulses were performed for comparison and to facilitate interfacial studies. The deposition temperature was fixed to 250 °C, which is well within possible ALD-windows of the binary processes applied [17,18]. The basic pulse sequence was n*(3 s La(thd)3), 5 s purge, 3 s O3,, 5 s purge, m*(1 s TMA), 5 s purge, 3 s O3 and 5 s purge where n and m were varied to control the stoichiometry. The individual pulse and purge times were chosen with basis of prior optimization of these processes on the selected ALD reactor, and are well within saturation limits for ALD-growth. An order of pulse sequences facilitating both maximum and minimum mixing of the different metal sub cycles were applied to study the effects of surface chemistry on the thin film growth. Resulting thin films were annealed in oxygen atmosphere (99.9%, AGA) at 650 °C for 30 min after a 20 °C min−1 ramp rate to induce crystallization. The samples were rapidly cooled to room temperature after heating by placing them on a metal plate. The cooling process was estimated to take 5–10 s. The thin films deposited using 2000 and 100 total precursor cycles (corresponding to a thickness of approximately 130 and 7 nm) were used for structural characterization both as deposited and after annealing. A Bruker D8 Discover x-ray diffractometer equipped with a Göbel mirror providing parallel Cu-Kα radiation in conventional θ–2θ geometry was used for all home-lab XRD (x-ray diffraction) studies. A similar diffractometer equipped with a Ge220 monochromator crystal and a Göbel mirror was used for the x-ray reflectivity (XRR) measurements. Films of specific structural interest were thoroughly studied at BM01A of the Swiss–Norwegian Beam Lines at ESRF, using a KUMA six-axis diffractometer and a CCD area detector from Oxford Instruments [19]. This setup was used for reciprocal space mapping and peak analysis leading to information on structural integrity. The La/Al stoichiometry was determined using x-ray fluorescence (XRF) measurements (Phillips PW 1400 XRF) equipped with the standard-less UniQuant software, (Omega Data Systems, ver. 2, 1994). Spectroscopic ellipsometry (J.A. Woollam α-SE) was used for determination of film thickness by fitting the acquired data to a Cauchy function. Topographical studies were performed using atomic force microscopy (AFM, Park Systems XE-70), equipped with a PPP-CONSTCR tip (Nanosystems) in contact mode for optimum lateral resolution.
3. Results and discussion 3.1. Atomic layer deposition of the La–Al–O system The combined TMA/O3 and La(thd)3/O3 precursor systems were studied by initially combining the metal pulses in a manner to enhance mixing. The targeted 1:1 composition between lanthanum and aluminium was achieved for a pulsed ratio of 1:1 between the metal precursors. As can be observed in Fig. 1, the achieved, deposited composition shows a nearly linear relationship with the pulsed composition, in particular for pulsed compositions La/(La + Al) N 0.5. The growth rate of the deposited films varies in a non-linear manner between those of the binary end members (Fig. 1). A rapid change in
91
Fig. 1. Growth rate (black, left) and deposited composition (blue, right) as a function of the pulsed composition (La/(La + Al)). The black circle corresponds to a 5:2 pulsing ratio with minimal mixing in the supercycle.
growth rate around the 1:1 achieved composition points at a change in surface chemistry for this ratio. The near linear relationship between pulsed and deposited composition, and in particular the rapid decay in growth rate around the 1:1 composition, is only true for maximum mixing of the metal sub cycles. As an example of the growth rate dependency on mixing conditions, in a deposition sequence composed of 5 sequential La(thd)3 + O3 pulses, and 2 sequential TMA + O3 pulses, a growth rate of 64.3 pm/ metal pulse was observed for the resulting film. This is in strong contrast to the 37.1 pm/metal pulse observed for maximum mixing of the metal precursor pulses. In both cases the stoichiometries of the films were similar. Studies were undertaken for understanding the growth dynamics responsible for these observed deviations. The system proved to be highly complex, as the growth after a particular pulse seemed not only to be correlated to the preceding pulse, but also to the second and third preceding pulses. An analytical solution cannot be given, but the study provides evidence that both La(thd)3 and TMA experience better growth conditions on an Al–O-surface than on a La–O-surface. This is also evident from Fig. 1, where the growth rate on each side of the stoichiometric composition is almost equal to the growth rate of the binary oxides.
3.2. XRD characterization of as-deposited and annealed films The main focus was put on characterizing films containing LaAlO3 phase, obtained by a 1:1 pulsed ratio of La(thd)3 and TMA subcycles. This process was used to deposit films on substrates of Si(100), SrTiO3(100), HF-etched SrTiO3(100), MgO(100) and LaAlO3(100). All comparative characterizations on the different substrates were performed on films deposited during the same run. Note that the LaAlO3 unit cell is considered pseudocubic throughout this paper, and is not reduced to its proper rhombohedral cell. This is done to emphasize its structural relationship with SrTiO3. All films were proven to be x-ray amorphous as deposited. For the films deposited on LaAlO3(100), this was proven by means of x-ray reflectometry that identified a separate surface layer giving rise to Fresnel oscillations from the as-deposited films. Crystallization was imposed by subjecting the samples to thermal annealing at 650 °C for 30 min, see experimental section. LAO films deposited on Si(100)- and MgO(100) remained amorphous after annealing, while ordering was observed by Homelab x-ray diffraction for 130 nm films deposited on SrTiO3(100) substrates (Fig. 2). This may be explained by a smaller lattice mismatch of ~2% for the LAO/ STO-systems, compared to a N10% mismatch for the LAO/MgO-system.
92
H.H. Sønsteby et al. / Thin Solid Films 550 (2014) 90–94
Fig. 2. X-ray diffractograms showing SrTiO3 and LaAlO3 (100), (200) and (300) reflections. Annealed LaAlO3 thin films are ordered along the surface normal to the pseudocubic c′axis parallel to the SrTiO3 c-axis and normal to the film surface (red line non-etched substrates, blue line HF-etched substrates). As deposited films (black) do not show any signs of crystallinity for LaAlO3.
The LaAlO3 films exhibit a selective ordering in a pseudocubic-(100) relationship to the SrTiO3(100)-substrate. Ultrathin films (b7 nm) proved difficult to study in the home lab, but were shown to exhibit the same ordering by means of synchrotron x-ray diffraction. The influence of the initial substrate surface termination on the film crystallinity was studied by comparing the films deposited on HFetched SrTiO3(100) with films deposited on untreated substrates. The Home-lab XRD did not provide much evidence of better crystallinity on the treated surfaces, as shown in Fig. 2, however, such improvements were clearly seen using synchrotron radiation and an area detector. AFM shows that the HF-etched substrates have a step-like surface morphology, with a step size of approximately 4 Å (Fig. 3). The SrTiO3 unit cell is 3.90 Å, implicating that the substrate is single species terminated. As the solubility of SrO in HF is much larger than that of TiO2 it is believed that mostly SrO is etched by the buffered HF, leaving the substrate Ti–O-terminated. Atomic force microscopy underlines that the as-deposited films are inherently flat on all substrates (average roughness b 1 nm, Fig. 4a), indicating an amorphous character of the films. The low roughness was also confirmed by x-ray reflectivity measurements, to which a fit estimates a roughness of 0.9 nm. Annealed films on Si(100) and Mg(100) are virtually unchanged from the as deposited situation; however, films on STO and LAO substrates were roughened after annealing, due to crystallites of about 100 nm width in plane was formed (Fig. 4b). Selected films were analyzed in detail using synchrotron radiation to obtain information on the structural relationship between the substrate
Fig. 4. AFM topography recorded using contact mode for a) as deposited, and b) annealed thin films. The color scheme is equal for the two images.
and the film. The samples were mounted as if being a single crystal on a KUMA six-axis diffractometer equipped with a CCD area detector, allowing for study of a large volume in reciprocal space. An initial overview of the reciprocal space, termed a panoramic view, was gathered, resulting in a slice of the (hl0)-plane in the reciprocal space, Fig. 5. This panoramic view for the LaAlO3||SrTiO3-system reveals a small mismatch between the film and the substrate unit cells, and a structural relationship of LaAlO3(100)|LaAlO3[100]||-SrTiO3(100)|SrTiO3[100]. The panoramic view itself does not confirm any in-plane ordering, as it is produced from volume integration. However, an inspection of all reflections in the collected volume confirms the suggested in-plane order. The panoramic view serves not only as a confirmation of the epitaxial relationship between the film and the substrate, but provides the possibility to map single reflections with high resolution. In Fig. 6, a more detailed space map of the (100) reflections for LaAlO3 deposited on HFtreated and non HF-treated SrTiO3(100) is shown. A much more intense and well defined reflection can be observed for the film on the etched substrate. This underlines the role the substrate surface plays during the crystallization process. The large q-range obtained by using synchrotron radiation was further utilized to study the evolution of broadening for families of diffraction peaks. The pseudocubic (100), (200), (300) and (400) were studied in terms of symmetric broadening. By performing a Williamson–Hall analysis of these reflections (Fig. 7), a broadening normal to the surface corresponding to 130 nm crystallites was revealed. This matches remarkably well with the thickness obtained from ellipsometry and XRR. Since the crystallites are on average 130 nm long normal to the substrate surface, the film most likely consist of a “pillared” structure with crystallites traversing from the LaAlO3||SrTiO3-interface throughout the film to the outer surface. A similar peak analysis can be done to
Fig. 3. AFM image showing step like features of the substrate morphology. The red line corresponds to the red line in the height profile, showing a characteristic 4 Å step size.
H.H. Sønsteby et al. / Thin Solid Films 550 (2014) 90–94
Fig. 5. An (hl0)slice in reciprocal space for the LaAlO3||SrTiO3 system, showing the SrTiO3 and LaAlO3 (hk0)-reflections. The (h00)-reflections, earlier characterized by traditional symmetric θ-2θ-scans and shown in Fig. 2, are not visible due to a dead zone in the geometry for this measurement.
estimate the in-plane crystallite size. For thin films, however, this method is crude, and as the size may vary severely, the estimate is not necessarily viable. However, such an analysis provided a smallest in-plane crystallite size of 140 nm, relatively similar to what was observed by atomic force microscopy.
3.3. Structural investigation of homoepitaxy LAO films were deposited on LaAlO3(100) to achieve homoepitaxy. As previously stated, these thin films were amorphous as deposited, and crystallized upon thermal annealing. All investigations utilizing Home lab x-ray equipment revealed no more than what appeared as the bare substrate. However, synchrotron radiation experiments revealed additional reflections for high-q values stemming from the deposited 130 nm thin film (Fig. 8). The observed peak mismatch corresponds to a change in the pseudocubic lattice of +0.01 Å for the LAO thin film with respect to the LAO substrate. XRF-measurements on the LaAlO3 film deposited on silicon does not show a perfectly equiatomic cation ratio between La and Al when compared to a clean LaAlO3 substrate. The clean substrate shows a La/(La + Al)-ratio of 0.493 as compared to the 0.482 ratio seen on the films as deposited on Si. Thus, the small lattice shift may be a result of a slightly off-stoichiometric composition in the thin films. The shift is not believed to stem from other impurities, as this would inherently be observed as a broadening effect due to compositional inhomogenities,
93
Fig. 7. Williamson Hall analysis of the (n00) family of reflections for the LaAlO3||SrTiO3system. Black line corresponds to broadening normal to the c-axis, whereas the red line corresponds to in-plane broadening. The crystallite size estimate is calculated at the intersection between the linear fit of the broadening and the y-axis.
and not as well-defined separate reflections. The feature is apparent both for ~130 nm and ~7 nm films. 3.4. Crystal truncation rods and Bragg satellites The high resolution diffraction studies near the (n00)-reflections for the ultrathin films (~7 nm) reveal a repeating oscillating pattern (Fig. 9). Such satellites appear due to either a surface- or interface ordering which is different from the bulk crystal structure. More detailed studies are needed to confirm the origin of these oscillations, but reconstruction of the LaAlO3 cell at the LAO||STO-interface due to buckling of the metal to oxygen bonds is a well known phenomenon [20]. Thus, the possibility exists that the satellites stem from a reconstruction effect at the interface between the substrate and the off-stoichiometric thin film. The satellites may also be a result of surface reconstruction, a known phenomenon for the LaAlO3 (100) surface [21]. 4. Conclusion Highly ordered thin films of LaAlO3 have been deposited using La(thd)3, Al(CH3)3 and O3 as precursors, with epitaxial relationship LaAlO3(100)|LaAlO3[100]||-SrTiO3(100)|SrTiO3[100]. Using only O3 and no water as the oxidizing agent removes any issues with the formation of lanthanum hydroxide, hence increasing the stability of the growth.
Fig. 6. Comparison of LaAlO3 (100) reflections for (a) etched and (b) untreated SrTiO3.
94
H.H. Sønsteby et al. / Thin Solid Films 550 (2014) 90–94
Fig. 8. Comparison between (110) and (660) for the LaAlO3||LaAlO3 homoepitaxial system. High-q reflections such as the (660) reveal a shift corresponding to a +0.01 Å of the pseudocubic lattice.
Acknowledgments The beam-line staff at the Swiss–Norwegian Beam Lines at ESRF in Grenoble are acknowledged for their contribution and help throughout the process of structural characterization of these films by means of synchrotron XRD. References
Fig. 9. Representation of LaAlO3 (200) and surrounding Bragg satellites for a LaAlO3 thin film on a SrTiO3-substrate. Note: the satellites are only seen for very thin films (b10 nm), and are lost in the Bragg peak for thicker samples.
There is an apparent linear relationship between the pulsed and deposited (achieved) stoichiometry for films when assuring maximum mixing of the metal pulses, regardless of the large difference in the growth rate for the binary systems. This one-to-one ratio increases the metal cation mixing and helps facilitate epitaxial growth. The films become epitaxial after annealing, resulting in crystalline pillars traversing through the film, with a width in the 100–150 nm range as seen by AFM and x-ray characterization. Thin films of LaAlO3 on LaAlO3 show a slight lattice change of the film with respect to the substrate, possibly stemming from a slight non-stoichiometry. The ultrathin films (~7 nm) show repeating Bragg satellites, which are believed to stem from a reconstruction of the LaAlO3 surface, or of the substrate at the film interface.
[1] X.B. Lu, Z.G. Liu, X. Zhang, R. Huang, H.W. Zhou, X.P. Wang, N. Bich-Yen, J. Phys. C Solid State Phys. 36 (23) (2003) 3047. [2] B.-E. Park, H. Ishiwara, Appl. Phys. Lett. 82 (8) (2003) 1197. [3] B.-E. Park, H. Ishiwara, Ferroelectrics 293 (1) (2003) 145. [4] S. Gariglio, N. Reyren, A. Caviglia, J.M. Triscone, J. Phys. Condens. Matter 21 (2009) 164213. [5] L. Li, C. Richter, J. Mannhart, R. Ashoori, Nat. Phys. 7 (2011) 762. [6] M. Basletic, J.L. Maurice, C. Carrétéro, G. Herranz, O. Copie, M. Bibes, E. Jacquet, K. Bouzehouane, S. Fusil, A. Barthélémy, Nat. Mater. 7 (2008) 621. [7] A. Ohtomo, H. Hwang, Nature 427 (2004) 423. [8] S. Gariglio, N. Reyren, A.D. Caviglia, J.M. Triscone, J. Phys. Condens. Matter 21 (16) (2009) 164. [9] M. Kawasaki, K. Takahashi, T. Maeda, R. Tsuchiya, M. Shinohara, O. Ishiyama, T. Yonezawa, M. Yoshimoto, H. Koinuma, Science 266 (1994) 1540. [10] C. Aruta, S. Amoruso, R. Bruzzese, X. Wang, D. Maccariello, G.F. Miletto, d.U.U. Scotti, Appl. Phys. Lett. 97 (2010) 252105. [11] K.M. El, C. Merckling, G. Delhaye, L. Arzel, G. Grenet, E. Bergignat, G. Hollinger, Mater. Sci. Semicond. Process. 9 (2007) 954. [12] W.F. Xiang, H.B. Lu, Z.H. Chen, X.B. Lu, M. He, H. Tian, Y.L. Zhou, C.R. Li, X.L. Ma, J. Cryst. Growth 271 (2004) 165. [13] I.M. Dildar, D.B. Boltje, M.H.S. Hesselberth, J. Aarts, Q. Xu, H.W. Zandbergen, S. Harkema, Appl. Phys. Lett. 102 (12) (2013). [14] O. Nilsen, M. Lie, H. Fjellvåg, A. Kjekshus, in: M. Fanciulli, G. Scarel (Eds.), Rare Earth Oxide Thin Films, vol. 106, Springer Berlin Heidelberg, 2007. [15] M. Nieminen, T. Sajavaara, E. Rauhala, M. Putkonen, L. Niinistoe, J. Mater. Chem. 11 (2001) 2340. [16] N.M. Sbrockey, M. Luong, E.M. Gallo, J.D. Sloppy, G. Chen, C.R. Winkler, S.H. Johnson, M.L. Taheri, G.S. Tompa, J.E. Spanier, J. Electron. Mater. 41 (2012) 819. [17] M. Nieminen, M. Putkonen, L. Niinistö, Appl. Surf. Sci. 174 (2) (2001) 155. [18] D.N. Goldstein, J.A. McCormick, S.M. George, J. Phys. Chem. C 112 (49) (2008) 19530. [19] H.H. Sønsteby, D. Chernyshov, M. Getz, O. Nilsen, H. Fjellvag, J. Synchrotron Radiat. 20 (2013) 644. [20] M. Huijben, A. Brinkman, G. Koster, G. Rijnders, H. Hilgenkamp, D.H.A. Blank, Adv. Mater. 21 (17) (2009) 1665. [21] Z.L. Wang, A.J. Shapiro, Surf. Sci. 328 (1–2) (1995) 159.
Contents lists available at ScienceDirect
Thin Solid Films journal homepage: www.elsevier.com/locate/tsf
Deposition and x-ray characterization of epitaxial thin films of LaAlO3 Henrik Hovde Sønsteby ⁎, Erik Østreng, Helmer Fjellvåg, Ola Nilsen University of Oslo, Department of Chemistry, Centre for Materials Science and Nanotechnology, P.O. Box 1033 Blindern, N-0315 Oslo, Norway
a r t i c l e
i n f o
Article history: Received 19 April 2013 Received in revised form 24 September 2013 Accepted 18 October 2013 Available online 31 October 2013 Keywords: Atomic layer deposition Lanthanum aluminate Synchrotron diffraction Interface effects
a b s t r a c t Highly epitaxial thin films of lanthanum aluminate (LaAlO3) have been obtained on strontium titanate (SrTiO3) substrates by means of atomic layer deposition using La(thd)3 (Hthd = 2,2,6,6-tetramethylhepta-3,5-dione), Al(CH3)3 and ozone as precursors. The system shows a near linear relationship between pulsed and deposited composition. Thin films with stoichiometric composition have been subject to thermal annealing at 650 °C under oxygen atmosphere, thereby achieving epitaxial films on Ti-O-terminated substrates of SrTiO3. The thin film||substrate epitaxial relationship is determined to be LaAlO3(100)|LaAlO3[100]||-SrTiO3(100)|SrTiO3[100] by use of synchrotron radiation. Selected films were also deposited on LaAlO3(100) to achieve homoepitaxy. This resulted in the observation of split peaks for high q-reflections, pointing towards slight differences in stoichiometry. For ultrathin films, Bragg satellites were observed around the specular reflections, coming from either surface- or interface reconstruction. © 2013 Elsevier B.V. All rights reserved.
1. Introduction The insulating lanthanum aluminate (LaAlO3, LAO) with perovskite type structure (R-3c) has, due to its high-κ dielectric and thermal properties, been thoroughly studied for use in microelectronics and as a buffer layer in the growth of other perovskite type oxides [1–3]. In recent years, more exotic properties have been discovered, such as superconductivity on the interface between LAO and strontium titanate (SrTiO3, STO) [4]. Recent data provide evidence of the coexistence of superconductivity and ferromagnetism at this interface [5], a rarely encountered phenomenon. These exotic effects seem to stem from a polarity discontinuity at the LAO/STO-interface due to the heterosystem maintaining the AO–BO2 stacking sequence, giving rise to a polar LaO+/TiO02 interface. This in turn induces charge transfer effects and a 2D-electron gas [6,7] that exhibit superconductivity below 300 mK [8]. Careful investigations of interfacial structures and accompanying exotic effects require perfect interfacial morphology. This is commonly achieved by growth of a thin film of one compound on a substrate of the other terminated by a single structural species. Perovskite substrate surfaces do not naturally become single species terminated by simple cutting, however, a chemical route for TiO2-termination of STO has been described by Kawasaki et al. [9]. The desired heterostructural interface can thus be obtained by depositing LAO on such a pre-treated substrate of STO. This has previously been attempted using pulsed laser deposition [10], molecular beam epitaxy [11,12] and sputtering [13] among other techniques. This study aims at providing an easily achievable route to the LaO+/ TiO02-interface by depositing a highly oriented thin film of LAO on a pre⁎ Corresponding author. E-mail address: [email protected] (H.H. Sønsteby). 0040-6090/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.tsf.2013.10.139
treated, TiO2 terminated STO substrate. Corresponding LAO films are further deposited on non-pretreated STO substrates for comparison. The films are deposited using atomic layer deposition (ALD), which is a chemical vapor deposition related technique, and using commonly available precursors. ALD utilizes a layer-by-layer self-limiting growth mechanism providing control of interfacial chemistry, conformality and film thickness at the (sub-)monolayer level. Deposition of ternary oxides by ALD is commonly achieved by sequentially introducing volatile gaseous metal-organic precursors separated by an oxide precursor usually consisting of water or ozone [14]. Deposition of polycrystalline LAO by ALD is previously reported by Nieminen et al. [15] using La(thd)3 (Hthd = 2,2,6,6-tetramethyl-3,5heptanedione) and Al(acac)3 (acac = acetylacetonate) as precursors. Attempts using TMA (TMA = Al(CH3)3, trimethylaluminium) and La(i-PrCp)3 (i-PrCp = isopropyl cyclopentadienyl) as precursors have been made to provide highly oriented thin films of stochiometric LAO on STO [16], although lacking saturative conditions for the La(i-PrCpd)3 precursor. In this work we provide evidence of ALD growth of LAO using the precursor combination TMA/O3 and La(thd)3/O3, giving rise to a clearly surface controlled stoichiometry and very high structural orientation on STO-substrates. The obtained film structures have been subjected to a detailed analysis using synchrotron radiation at the Swiss–Norwegian Beam Lines at ESRF, Grenoble. 2. Experimental Thin films were deposited in a commercial F-120 Sat ALD-reactor (ASM Microchemistry) using La(thd)3 (Multivalent), TMA (technical grade) and O3 as precursors. Ozone gas was produced by an OT-020 (Ozone Technology) generator fed with pure O2 (99.999%, AGA) and used at an approximate flow rate of 500 cm3 min− 1. La(thd)3
H.H. Sønsteby et al. / Thin Solid Films 550 (2014) 90–94
was sublimated at 185 °C, whereas TMA was kept at room temperature in an external bubbler. A carrier-gas flow of 300 cm3 min− 1 was supplied from a Nitrox UHPN 3001 generator producing 99.9995% inert gas. The films were deposited on 2.5 × 2.5 cm2 substrates of Si(100) and soda lime glass and on 1.0 × 1.0 cm2 substrates of SrTiO3(100) LaAlO3(100) and MgO(100). In addition, some SrTiO3(100) substrates were etched for 10 min in a buffered NH4F–HF solution (Sigma-Aldrich) to achieve Ti–O-termination as described by Kawasaki et al. [9]. All substrates were further cleaned by pulsing O3 prior to the depositions. A number of La:Al pulsing ratios were investigated while keeping the total amount of metal pulses as close as practically possible to 1000. The resulting films had thicknesses between 40 and 80 nm. Selected depositions totaling 100 metal pulses were performed for comparison and to facilitate interfacial studies. The deposition temperature was fixed to 250 °C, which is well within possible ALD-windows of the binary processes applied [17,18]. The basic pulse sequence was n*(3 s La(thd)3), 5 s purge, 3 s O3,, 5 s purge, m*(1 s TMA), 5 s purge, 3 s O3 and 5 s purge where n and m were varied to control the stoichiometry. The individual pulse and purge times were chosen with basis of prior optimization of these processes on the selected ALD reactor, and are well within saturation limits for ALD-growth. An order of pulse sequences facilitating both maximum and minimum mixing of the different metal sub cycles were applied to study the effects of surface chemistry on the thin film growth. Resulting thin films were annealed in oxygen atmosphere (99.9%, AGA) at 650 °C for 30 min after a 20 °C min−1 ramp rate to induce crystallization. The samples were rapidly cooled to room temperature after heating by placing them on a metal plate. The cooling process was estimated to take 5–10 s. The thin films deposited using 2000 and 100 total precursor cycles (corresponding to a thickness of approximately 130 and 7 nm) were used for structural characterization both as deposited and after annealing. A Bruker D8 Discover x-ray diffractometer equipped with a Göbel mirror providing parallel Cu-Kα radiation in conventional θ–2θ geometry was used for all home-lab XRD (x-ray diffraction) studies. A similar diffractometer equipped with a Ge220 monochromator crystal and a Göbel mirror was used for the x-ray reflectivity (XRR) measurements. Films of specific structural interest were thoroughly studied at BM01A of the Swiss–Norwegian Beam Lines at ESRF, using a KUMA six-axis diffractometer and a CCD area detector from Oxford Instruments [19]. This setup was used for reciprocal space mapping and peak analysis leading to information on structural integrity. The La/Al stoichiometry was determined using x-ray fluorescence (XRF) measurements (Phillips PW 1400 XRF) equipped with the standard-less UniQuant software, (Omega Data Systems, ver. 2, 1994). Spectroscopic ellipsometry (J.A. Woollam α-SE) was used for determination of film thickness by fitting the acquired data to a Cauchy function. Topographical studies were performed using atomic force microscopy (AFM, Park Systems XE-70), equipped with a PPP-CONSTCR tip (Nanosystems) in contact mode for optimum lateral resolution.
3. Results and discussion 3.1. Atomic layer deposition of the La–Al–O system The combined TMA/O3 and La(thd)3/O3 precursor systems were studied by initially combining the metal pulses in a manner to enhance mixing. The targeted 1:1 composition between lanthanum and aluminium was achieved for a pulsed ratio of 1:1 between the metal precursors. As can be observed in Fig. 1, the achieved, deposited composition shows a nearly linear relationship with the pulsed composition, in particular for pulsed compositions La/(La + Al) N 0.5. The growth rate of the deposited films varies in a non-linear manner between those of the binary end members (Fig. 1). A rapid change in
91
Fig. 1. Growth rate (black, left) and deposited composition (blue, right) as a function of the pulsed composition (La/(La + Al)). The black circle corresponds to a 5:2 pulsing ratio with minimal mixing in the supercycle.
growth rate around the 1:1 achieved composition points at a change in surface chemistry for this ratio. The near linear relationship between pulsed and deposited composition, and in particular the rapid decay in growth rate around the 1:1 composition, is only true for maximum mixing of the metal sub cycles. As an example of the growth rate dependency on mixing conditions, in a deposition sequence composed of 5 sequential La(thd)3 + O3 pulses, and 2 sequential TMA + O3 pulses, a growth rate of 64.3 pm/ metal pulse was observed for the resulting film. This is in strong contrast to the 37.1 pm/metal pulse observed for maximum mixing of the metal precursor pulses. In both cases the stoichiometries of the films were similar. Studies were undertaken for understanding the growth dynamics responsible for these observed deviations. The system proved to be highly complex, as the growth after a particular pulse seemed not only to be correlated to the preceding pulse, but also to the second and third preceding pulses. An analytical solution cannot be given, but the study provides evidence that both La(thd)3 and TMA experience better growth conditions on an Al–O-surface than on a La–O-surface. This is also evident from Fig. 1, where the growth rate on each side of the stoichiometric composition is almost equal to the growth rate of the binary oxides.
3.2. XRD characterization of as-deposited and annealed films The main focus was put on characterizing films containing LaAlO3 phase, obtained by a 1:1 pulsed ratio of La(thd)3 and TMA subcycles. This process was used to deposit films on substrates of Si(100), SrTiO3(100), HF-etched SrTiO3(100), MgO(100) and LaAlO3(100). All comparative characterizations on the different substrates were performed on films deposited during the same run. Note that the LaAlO3 unit cell is considered pseudocubic throughout this paper, and is not reduced to its proper rhombohedral cell. This is done to emphasize its structural relationship with SrTiO3. All films were proven to be x-ray amorphous as deposited. For the films deposited on LaAlO3(100), this was proven by means of x-ray reflectometry that identified a separate surface layer giving rise to Fresnel oscillations from the as-deposited films. Crystallization was imposed by subjecting the samples to thermal annealing at 650 °C for 30 min, see experimental section. LAO films deposited on Si(100)- and MgO(100) remained amorphous after annealing, while ordering was observed by Homelab x-ray diffraction for 130 nm films deposited on SrTiO3(100) substrates (Fig. 2). This may be explained by a smaller lattice mismatch of ~2% for the LAO/ STO-systems, compared to a N10% mismatch for the LAO/MgO-system.
92
H.H. Sønsteby et al. / Thin Solid Films 550 (2014) 90–94
Fig. 2. X-ray diffractograms showing SrTiO3 and LaAlO3 (100), (200) and (300) reflections. Annealed LaAlO3 thin films are ordered along the surface normal to the pseudocubic c′axis parallel to the SrTiO3 c-axis and normal to the film surface (red line non-etched substrates, blue line HF-etched substrates). As deposited films (black) do not show any signs of crystallinity for LaAlO3.
The LaAlO3 films exhibit a selective ordering in a pseudocubic-(100) relationship to the SrTiO3(100)-substrate. Ultrathin films (b7 nm) proved difficult to study in the home lab, but were shown to exhibit the same ordering by means of synchrotron x-ray diffraction. The influence of the initial substrate surface termination on the film crystallinity was studied by comparing the films deposited on HFetched SrTiO3(100) with films deposited on untreated substrates. The Home-lab XRD did not provide much evidence of better crystallinity on the treated surfaces, as shown in Fig. 2, however, such improvements were clearly seen using synchrotron radiation and an area detector. AFM shows that the HF-etched substrates have a step-like surface morphology, with a step size of approximately 4 Å (Fig. 3). The SrTiO3 unit cell is 3.90 Å, implicating that the substrate is single species terminated. As the solubility of SrO in HF is much larger than that of TiO2 it is believed that mostly SrO is etched by the buffered HF, leaving the substrate Ti–O-terminated. Atomic force microscopy underlines that the as-deposited films are inherently flat on all substrates (average roughness b 1 nm, Fig. 4a), indicating an amorphous character of the films. The low roughness was also confirmed by x-ray reflectivity measurements, to which a fit estimates a roughness of 0.9 nm. Annealed films on Si(100) and Mg(100) are virtually unchanged from the as deposited situation; however, films on STO and LAO substrates were roughened after annealing, due to crystallites of about 100 nm width in plane was formed (Fig. 4b). Selected films were analyzed in detail using synchrotron radiation to obtain information on the structural relationship between the substrate
Fig. 4. AFM topography recorded using contact mode for a) as deposited, and b) annealed thin films. The color scheme is equal for the two images.
and the film. The samples were mounted as if being a single crystal on a KUMA six-axis diffractometer equipped with a CCD area detector, allowing for study of a large volume in reciprocal space. An initial overview of the reciprocal space, termed a panoramic view, was gathered, resulting in a slice of the (hl0)-plane in the reciprocal space, Fig. 5. This panoramic view for the LaAlO3||SrTiO3-system reveals a small mismatch between the film and the substrate unit cells, and a structural relationship of LaAlO3(100)|LaAlO3[100]||-SrTiO3(100)|SrTiO3[100]. The panoramic view itself does not confirm any in-plane ordering, as it is produced from volume integration. However, an inspection of all reflections in the collected volume confirms the suggested in-plane order. The panoramic view serves not only as a confirmation of the epitaxial relationship between the film and the substrate, but provides the possibility to map single reflections with high resolution. In Fig. 6, a more detailed space map of the (100) reflections for LaAlO3 deposited on HFtreated and non HF-treated SrTiO3(100) is shown. A much more intense and well defined reflection can be observed for the film on the etched substrate. This underlines the role the substrate surface plays during the crystallization process. The large q-range obtained by using synchrotron radiation was further utilized to study the evolution of broadening for families of diffraction peaks. The pseudocubic (100), (200), (300) and (400) were studied in terms of symmetric broadening. By performing a Williamson–Hall analysis of these reflections (Fig. 7), a broadening normal to the surface corresponding to 130 nm crystallites was revealed. This matches remarkably well with the thickness obtained from ellipsometry and XRR. Since the crystallites are on average 130 nm long normal to the substrate surface, the film most likely consist of a “pillared” structure with crystallites traversing from the LaAlO3||SrTiO3-interface throughout the film to the outer surface. A similar peak analysis can be done to
Fig. 3. AFM image showing step like features of the substrate morphology. The red line corresponds to the red line in the height profile, showing a characteristic 4 Å step size.
H.H. Sønsteby et al. / Thin Solid Films 550 (2014) 90–94
Fig. 5. An (hl0)slice in reciprocal space for the LaAlO3||SrTiO3 system, showing the SrTiO3 and LaAlO3 (hk0)-reflections. The (h00)-reflections, earlier characterized by traditional symmetric θ-2θ-scans and shown in Fig. 2, are not visible due to a dead zone in the geometry for this measurement.
estimate the in-plane crystallite size. For thin films, however, this method is crude, and as the size may vary severely, the estimate is not necessarily viable. However, such an analysis provided a smallest in-plane crystallite size of 140 nm, relatively similar to what was observed by atomic force microscopy.
3.3. Structural investigation of homoepitaxy LAO films were deposited on LaAlO3(100) to achieve homoepitaxy. As previously stated, these thin films were amorphous as deposited, and crystallized upon thermal annealing. All investigations utilizing Home lab x-ray equipment revealed no more than what appeared as the bare substrate. However, synchrotron radiation experiments revealed additional reflections for high-q values stemming from the deposited 130 nm thin film (Fig. 8). The observed peak mismatch corresponds to a change in the pseudocubic lattice of +0.01 Å for the LAO thin film with respect to the LAO substrate. XRF-measurements on the LaAlO3 film deposited on silicon does not show a perfectly equiatomic cation ratio between La and Al when compared to a clean LaAlO3 substrate. The clean substrate shows a La/(La + Al)-ratio of 0.493 as compared to the 0.482 ratio seen on the films as deposited on Si. Thus, the small lattice shift may be a result of a slightly off-stoichiometric composition in the thin films. The shift is not believed to stem from other impurities, as this would inherently be observed as a broadening effect due to compositional inhomogenities,
93
Fig. 7. Williamson Hall analysis of the (n00) family of reflections for the LaAlO3||SrTiO3system. Black line corresponds to broadening normal to the c-axis, whereas the red line corresponds to in-plane broadening. The crystallite size estimate is calculated at the intersection between the linear fit of the broadening and the y-axis.
and not as well-defined separate reflections. The feature is apparent both for ~130 nm and ~7 nm films. 3.4. Crystal truncation rods and Bragg satellites The high resolution diffraction studies near the (n00)-reflections for the ultrathin films (~7 nm) reveal a repeating oscillating pattern (Fig. 9). Such satellites appear due to either a surface- or interface ordering which is different from the bulk crystal structure. More detailed studies are needed to confirm the origin of these oscillations, but reconstruction of the LaAlO3 cell at the LAO||STO-interface due to buckling of the metal to oxygen bonds is a well known phenomenon [20]. Thus, the possibility exists that the satellites stem from a reconstruction effect at the interface between the substrate and the off-stoichiometric thin film. The satellites may also be a result of surface reconstruction, a known phenomenon for the LaAlO3 (100) surface [21]. 4. Conclusion Highly ordered thin films of LaAlO3 have been deposited using La(thd)3, Al(CH3)3 and O3 as precursors, with epitaxial relationship LaAlO3(100)|LaAlO3[100]||-SrTiO3(100)|SrTiO3[100]. Using only O3 and no water as the oxidizing agent removes any issues with the formation of lanthanum hydroxide, hence increasing the stability of the growth.
Fig. 6. Comparison of LaAlO3 (100) reflections for (a) etched and (b) untreated SrTiO3.
94
H.H. Sønsteby et al. / Thin Solid Films 550 (2014) 90–94
Fig. 8. Comparison between (110) and (660) for the LaAlO3||LaAlO3 homoepitaxial system. High-q reflections such as the (660) reveal a shift corresponding to a +0.01 Å of the pseudocubic lattice.
Acknowledgments The beam-line staff at the Swiss–Norwegian Beam Lines at ESRF in Grenoble are acknowledged for their contribution and help throughout the process of structural characterization of these films by means of synchrotron XRD. References
Fig. 9. Representation of LaAlO3 (200) and surrounding Bragg satellites for a LaAlO3 thin film on a SrTiO3-substrate. Note: the satellites are only seen for very thin films (b10 nm), and are lost in the Bragg peak for thicker samples.
There is an apparent linear relationship between the pulsed and deposited (achieved) stoichiometry for films when assuring maximum mixing of the metal pulses, regardless of the large difference in the growth rate for the binary systems. This one-to-one ratio increases the metal cation mixing and helps facilitate epitaxial growth. The films become epitaxial after annealing, resulting in crystalline pillars traversing through the film, with a width in the 100–150 nm range as seen by AFM and x-ray characterization. Thin films of LaAlO3 on LaAlO3 show a slight lattice change of the film with respect to the substrate, possibly stemming from a slight non-stoichiometry. The ultrathin films (~7 nm) show repeating Bragg satellites, which are believed to stem from a reconstruction of the LaAlO3 surface, or of the substrate at the film interface.
[1] X.B. Lu, Z.G. Liu, X. Zhang, R. Huang, H.W. Zhou, X.P. Wang, N. Bich-Yen, J. Phys. C Solid State Phys. 36 (23) (2003) 3047. [2] B.-E. Park, H. Ishiwara, Appl. Phys. Lett. 82 (8) (2003) 1197. [3] B.-E. Park, H. Ishiwara, Ferroelectrics 293 (1) (2003) 145. [4] S. Gariglio, N. Reyren, A. Caviglia, J.M. Triscone, J. Phys. Condens. Matter 21 (2009) 164213. [5] L. Li, C. Richter, J. Mannhart, R. Ashoori, Nat. Phys. 7 (2011) 762. [6] M. Basletic, J.L. Maurice, C. Carrétéro, G. Herranz, O. Copie, M. Bibes, E. Jacquet, K. Bouzehouane, S. Fusil, A. Barthélémy, Nat. Mater. 7 (2008) 621. [7] A. Ohtomo, H. Hwang, Nature 427 (2004) 423. [8] S. Gariglio, N. Reyren, A.D. Caviglia, J.M. Triscone, J. Phys. Condens. Matter 21 (16) (2009) 164. [9] M. Kawasaki, K. Takahashi, T. Maeda, R. Tsuchiya, M. Shinohara, O. Ishiyama, T. Yonezawa, M. Yoshimoto, H. Koinuma, Science 266 (1994) 1540. [10] C. Aruta, S. Amoruso, R. Bruzzese, X. Wang, D. Maccariello, G.F. Miletto, d.U.U. Scotti, Appl. Phys. Lett. 97 (2010) 252105. [11] K.M. El, C. Merckling, G. Delhaye, L. Arzel, G. Grenet, E. Bergignat, G. Hollinger, Mater. Sci. Semicond. Process. 9 (2007) 954. [12] W.F. Xiang, H.B. Lu, Z.H. Chen, X.B. Lu, M. He, H. Tian, Y.L. Zhou, C.R. Li, X.L. Ma, J. Cryst. Growth 271 (2004) 165. [13] I.M. Dildar, D.B. Boltje, M.H.S. Hesselberth, J. Aarts, Q. Xu, H.W. Zandbergen, S. Harkema, Appl. Phys. Lett. 102 (12) (2013). [14] O. Nilsen, M. Lie, H. Fjellvåg, A. Kjekshus, in: M. Fanciulli, G. Scarel (Eds.), Rare Earth Oxide Thin Films, vol. 106, Springer Berlin Heidelberg, 2007. [15] M. Nieminen, T. Sajavaara, E. Rauhala, M. Putkonen, L. Niinistoe, J. Mater. Chem. 11 (2001) 2340. [16] N.M. Sbrockey, M. Luong, E.M. Gallo, J.D. Sloppy, G. Chen, C.R. Winkler, S.H. Johnson, M.L. Taheri, G.S. Tompa, J.E. Spanier, J. Electron. Mater. 41 (2012) 819. [17] M. Nieminen, M. Putkonen, L. Niinistö, Appl. Surf. Sci. 174 (2) (2001) 155. [18] D.N. Goldstein, J.A. McCormick, S.M. George, J. Phys. Chem. C 112 (49) (2008) 19530. [19] H.H. Sønsteby, D. Chernyshov, M. Getz, O. Nilsen, H. Fjellvag, J. Synchrotron Radiat. 20 (2013) 644. [20] M. Huijben, A. Brinkman, G. Koster, G. Rijnders, H. Hilgenkamp, D.H.A. Blank, Adv. Mater. 21 (17) (2009) 1665. [21] Z.L. Wang, A.J. Shapiro, Surf. Sci. 328 (1–2) (1995) 159.