Giant step bunching in epitaxial SrRuO3 films on vicinal SrTiO3(001)

Thin Solid Films 495 (2006) 159 – 164 www.elsevier.com/locate/tsf

Giant step bunching in epitaxial SrRuO3 films on vicinal SrTiO3(001) F. Sa´nchez a,*, G. Herranz a, C. Ferrater b, M.V. Garcı´a-Cuenca b, M. Varela b, J. Fontcuberta a a

b

Institut de Cie`ncia de Materials de Barcelona, CSIC, Campus U.A.B., 08193 Bellaterra, Spain Dep. de Fı´sica Aplicada i O`ptica, Universitat de Barcelona, Diagonal 647, Barcelona 08028, Spain Available online 27 September 2005

Abstract We report here on the giant step bunching, with steps reaching 20 – 40 unit cells high, observed in SrRuO3 epitaxial films on vicinal SrTiO3(001). Bunching develops in films thicker than tens of nanometers, whereas three-dimensional (3D) islands were found in thinner films. We show that bunching forms when these islands coalesce. The growth of the 3D islands on substrates having varied miscut angles is investigated. In spite of the formation of bunching from initial nucleation of islands, the final morphology of the film surface is highly ordered as statistical analysis confirmed. The influence of processing parameters on the bunching morphology was investigated, and it has been demonstrated that lateral and vertical dimensions can be separately adjusted with a proper combined selection of film growth rate and substrate miscut angle. D 2005 Elsevier B.V. All rights reserved. Keywords: Step bunching; Ferromagnetic oxide; Film growth; Self-organization; Epitaxy

1. Introduction SrRuO3 (SRO) is ferromagnetic oxide below 160 K, and it is a conductor with a room temperature electrical resistivity of about 200 –300 AV cm [1]. It has an orthorhombic structure that can be indexed as a distorted perovskite having a lattice parameter of 0.393 nm. The high compatibility of its crystal structure with other perovskite oxides, its high thermal stability [2], and the high electrical conductivity makes SRO suitable as electrode, mainly in ferroelectric devices [3]. SRO films deposited on vicinal, (001) oriented, SrTiO3 (STO) substrates have the lowest electrical resistivity [4– 6]. The surface of these films develops a remarkable step bunching [7 – 9], which contrasts to the morphology of terraces and monolayer steps observed in films on low vicinal substrates [2,9 –12]. Bunching of steps implies an increased roughness of the surface, so its control is important, particularly when considering the use of SRO as metallic electrode in some applications. Moreover, surfaces with bunched steps are of interest as they could be used as templates for growth of low-dimensional structures [13]. Therefore, understanding and controlling the formation of bunching in SRO films is of the highest interest. * Corresponding author. Tel.: +34 935801853; fax: +34 935805729. E-mail address: [email protected] (F. Sa´nchez). 0040-6090/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2005.08.330

To date, bunching has been mainly studied in some detail, in semiconducting and metallic systems, being the number of monolayer steps in a bunch generally below ten [14 –16]. Although less usual, bunches with more than ten monolayer steps were also reported [17 – 20]. It is commonly accepted that when growth mode is step flow, bunching results from differences in step velocity. However, we show here that bunching in SRO does not form in such a one dimensional process, but from the coalescence of three-dimensional (3D) islands. Details of the bunching formation process are reported elsewhere [8]. Here we present statistical analysis of both, lateral (in-plane) and vertical (out-of-plane) dimensions in SRO films displaying bunched morphology. We have also investigated the influence of the STO miscut angle, growth rate and the 3D islands-grown at initial stage, on bunching formation. 2. Experimental SRO thin films were grown by pulsed laser deposition using a KrF excimer laser (248 nm wavelength and 34 ns pulse duration). The laser beam was focused, with a fluence around 2 J/cm2, on a stoichiometric target of SrRuO3 (SCI Engineered Materials). The target was rotated during the ablation process to reduce non-uniform erosion. Films of varied thickness were prepared on SrTiO3(001) substrates (CrysTec), misoriented

160

F. Sa´nchez et al. / Thin Solid Films 495 (2006) 159 – 164

towards the [100] direction by controlled vicinality h v up to ¨4-. The films were deposited under 10 Pa of pure oxygen and at a substrate temperature of 750 -C. Atomic force microscopy (AFM) working in tapping mode was used to characterize the surface morphology of substrates and films. AFM image processing and analysis was done using the WSxM software [21]. AFM height profiles and X-ray diffractometry were used to determine the vicinality of the substrates. Additional details of the deposition conditions, as well as on epitaxial nature, lattice strain, and magnetotransport properties can be found elsewhere [5,6,22]. Films here reported were grown at rates of ¨0.018 or ¨0.0042 nm per laser pulse as deduced from proper calibration using X-ray reflectivity. The nominal thickness (t) of the films ranges from 1.7 to 100 nm.

resolutions of the microscope, islands are not observed. Steps are parallel and, remarkably, they can be above 8 nm high. These films are epitaxial, with a (001) out-of-plane orientation (pseudocubic lattice, 1 unit cell (u.c.)¨0.4 nm). Therefore, bunches can include more than 20 monolayer steps. This is illustrated in the height profile displayed in Fig. 1b, where u.c. units are used in the height axis. Bunching with so many monolayer steps is unusual, and thus it can be properly defined as giant bunching. The surface of the substrate used does not show any similar pre-existing morphology. In the inset of Fig. 1b we include an AFM image (1 1 Am2 scan size, 2D view) of the h v¨2- substrate after annealing in air during 2 h at 800 -C, a temperature higher than the used to grow the films. There is a well defined morphology of steps, one or half u.c. high, running along a [100] direction and separating terraces around 10 nm wide. It implies that bunching in SRO films is not a replica of a similar topography in the substrate, and thus it develops during the epitaxial growth. The high size of bunching in SRO films is well appreciated when comparing the AFM image of the substrate with a 2D view of the film

3. Results and discussion Fig. 1 shows the morphology of a SRO film grown on a vicinal (h v  1-) STO substrate, displaying flat terraces separated by bunches of steps. Within the vertical and lateral 120

Height (u.c.)

(a)

80 200nm

(b)

40

0 0

200

400

600

800

1000

(c)

Normalized distribution

x ( nm)

200nm

(d)

1.0 0.8 0.6 0.4 0.2 0.0 0

50

100

150

Height (u.c.)

(e)

(f)

Events

30 20 10

1.0µm

0 0

100

200

300

400

Terrace width (nm) Fig. 1. Atomic force microscopy topographic images of a t = 100 nm film on h v¨2- STO. (a) 3D view (1 1 Am2 scan size) and, (b) height profile (height in SRO lattice unit cell units) along the line drawn in the image. Inset: topographic image of the substrate (1 1 Am2 scan size); (c) 2D view (1 1 Am2 scan size) and, (d) histogram of heights; (e) derivative view (5  5 Am2 scan size) and, (f) histogram of terrace widths.

F. Sa´nchez et al. / Thin Solid Films 495 (2006) 159 – 164

(Fig. 1c). Terraces in the film are around 200 nm wide, which corresponds to around 20 times the substrate terrace width. We also note that in spite of the outstanding increase of the terrace width, height of a bunched step is the same at different points along the edge of a terrace. It indicates that terraces have basically atomically flat surfaces. The uniformity of height along each step is statistically confirmed by the histogram plotted in Fig. 1d. The distribution shows discrete peaks, each of them including the points within the same terrace. Film morphology is shown at a larger scale in Fig. 1e, where a derivative filter was applied (to a 5  5 Am2 scan image) to enhance contrast at steps. Steps run parallel and extend several microns long, although occasionally step intersections are also observed. In spite of these defects, terraces are highly uniform as a statistical analysis of its width demonstrates (Fig. 1f). A Gaussian curve (solid line) fits accurately the data in the histogram. The average terrace width l 0 and the ratio Dl / l 0 (Dl the width of the distribution) are 178 nm and ¨0.35, respectively. Clearly, step bunching in this film is highly ordered in both vertical and lateral directions. Now we pay attention to the 3D islands that, as mentioned in the Introduction of this paper, constitute the first stage in the bunching formation. Fig. 2a shows the morphology of a t = 1.7 nm film on h v¨2- STO. From left to right there is a 3D view of the topography (1 1 Am2 scan size), the same image after applying a flooding filter [21], and the height profile along the drawn line. Islands have a characteristic triangular base contour and a wedge-like profile, with an edge of the base aligned to the [100] direction of the STO single crystal, which is also the direction of the substrate steps. Interestingly, the length of islands perpendicularly to substrate steps can be above 100 nm, so some of them extend through more than ten substrate terraces. We note the absence of spatial order which is, in fact, what is expected when nucleation is random (lateral

161

length of the 1 1 Am2 image corresponds to ¨100 substrate terraces). On the other hand, height profiles confirm that the surface of the islands is a (001) SRO crystal plane, in accordance with the observed orientation of the islands along a [100] direction. Indeed, monolayer steps can be found on the surface of the islands (for instance, there are two monolayer steps in the plotted profile). To the best of our knowledge, islands similar to these peculiar faceted SRO islands only have been found in the Co/Au(788) heteroepitaxial system [23], although its coalescence and later evolution was not reported. We reported that epitaxial SRO films on h v  0.5- miscut STO substrates do not develop bunching morphology [9– 11]. In contrast, there are terraces and monolayer steps (when h v is in the 0.1– 0.5- range) or multilayered mounds (when h v is below ¨0.1-) [11]. The morphology of terraces and monolayer steps was found to be formed after a growth mode transition, since at early stages, 3D islands were also observed [10]. Therefore, irrespectively of the h v substrate miscut angle, 3D islands are found at initial growth stages. However, morphology in thicker films depends critically on h v. The critical effect of h v could be already determined by differences existing in the initial 3D islands. Therefore, a comparison of the 3D islands that form on low- and high-h v substrates is in order. We show in Fig. 2b the morphology of a film having the same nominal thickness, t = 1.7 nm, but grown on h v¨0.04- STO. We also present, from left to right, a 3D view of the topography (1 1 Am2 scan size), the same image after applying a flooding filter [21], and the height profile along the drawn line. The substrate is almost singular, with steps about ¨500 nm apart. The filtered AFM image evidences that islands nucleated mainly at steps. Their size (area and height) is similar to that of islands on the h v¨2- STO. However, islands on the h v¨0.04- STO do not have the wedge-like shape that should be expected if their

3

Height (nm)

(a)

2 1 0

0

40

80

120

160

x (nm)

(b) Height (nm)

3 2 1 0

0

40

80

120 160

x (nm) Fig. 2. Atomic force microscopy topographic images: (a) t = 1.7 nm film on h v¨2- STO. From left to right: 3D view (1 1 Am2 scan size), filtered image, showing only the highest height points, and height profile along the line drawn in the filtered image; (b) t = 1.7 nm film on a h v¨0.04- substrate. From left to right: 3D view (1 1 Am2 scan size), filtered image, showing only the highest height points, and height profile along the drawn line in the filtered image.

162

F. Sa´nchez et al. / Thin Solid Films 495 (2006) 159 – 164

(b)

8

Height (nm)

(a)

4 2 0

1.0µm

(c)

6

0

50

100 150 200 250

x (nm) Fig. 3. Atomic force microscopy topographic images of a t = 1.7 nm film on h v¨4- STO. (a) 2D view (5  5 Am2 scan size); (b) 3D view (1 1 Am2 scan size); (c) solid line: height profile along the line drawn in (b). Dashed line: typical height profile in the t = 1.7 nm film on h v¨2- STO.

(a)

2

1 1.0µm

0

-1

(b)

C (nm2)

4

2 1.0µm

0

-2

(c)

8

4 1.0µm

0

-4 0.0

0.5

1.0

r (µm) Fig. 4. Atomic force microscopy topographic images (3D view, 5  5 Am2 scan size) of t = 75 nm films grown at a reduced rate of 0.0042 nm/pulse on (a) h v¨1STO; (b) h v¨1.5- STO; and (c) h v¨2- STO. Right panels: autocorrelation functions C(r) along the direction r perpendicular to the steps, and 2D autocorrelation functions (Inset). In (c), the autocorrelation function plotted in dashed line corresponds to the t = 100 nm film on h v¨2- STO grown at 0.018 nm/pulse (its morphology is shown in Fig. 1).

F. Sa´nchez et al. / Thin Solid Films 495 (2006) 159 – 164

lateral surface was (001) faceted. This can be appreciated from the topographic image, and is demonstrated with the height profile (note that on a ¨0.04- STO substrate, a 2 nm high wedge-like island should be around 3 Am long). Thus, although there is nucleation at steps on both substrates (h v¨0.04- and ¨2-), the extremely different terrace widths (¨500 and ¨10 nm) induces distinct growth and shape of the initial islands, and thus distinct final film morphology after coalescence. Nanometric films were also grown on highly vicinal crystals. Fig. 3 shows AFM topographic images of a film, also 1.7 nm thick, on a h v¨4- STO. There are 3D islands separated quite uniformly by a distance of the order of 100 nm, whereas in the substrate terraces were nominally only ¨5 nm apart. The higher resolution image (1 1 Am2 scan size) in Fig. 3b confirms that islands are wedge-like, having also a triangular base contour. In Fig. 3c (solid line) the height profile along the line drawn on the topographic image is compared with a typical profile (dashed line) taken perpendicularly to a wedge-like island in the t = 1.7 nm film on the h v¨2- STO. Interestingly, lateral size in both islands is similar, and a comparison of the respective topographic images indicates that this is a general trend. It implies that in spite of the important role that steps have on nucleation of islands and diffusion of adatoms, other parameters affecting the islands density can be more relevant, as growth temperature and adatom supersaturation. Indeed, we demonstrated [8,9] that bunching size can be tailored by adjusting the growth rate. To get insight on the effect of the miscut angle (which determines the terrace width) on the bunching morphology, we show AFM images (5  5 Am2 scan size) of t = 75 nm films grown on h v¨1- (Fig. 4a), ¨1.5- (Fig. 4b), and ¨2- (Fig. 4c) substrates at a deposition rate 0.0042 nm/pulse. A reduced grow-rate increases the size of the bunching, and bunches of around 40 monolayer steps are found. Inspection of images in Fig. 4, reveals that in spite of the significantly different substrate miscut angle, the AFM images do not show clear differences in the terrace width of the films. The average terrace width can be accurately determined from autocorrelation functions. The function is defined as C(r) = , where r is a shift from the center of the image, <. . .> means an average over all possible positions defined by r 0 and h(r 0) is the height at position r 0 relative to the mean height. Intensity profiles of the autocorrelation in the direction perpendicular to the steps are plotted at the right of each topographic image, and the 2D autocorrelation functions are shown at the insets. It is to be noted that clear oscillation in autocorrelation functions originates when topographic structures are separated very uniformly, and the position of the first maximum to the origin is a precise value of the average structure separation [24]. In the radial profiles of C(r) the oscillation is clear, and lateral sizes in the 500 –700 nm range are determined from the position of the maxima. Clearly, miscut angle has a low effect on the lateral periodicity (i.e., terrace width) and therefore allows controlling bunching height (increasing height with miscut angle). Moreover, at a fixed miscut angle, growth rate is a critical parameter. This is evidenced by comparing the topographic image in Fig. 4c with

163

the topography of the film grown on the same substrate at a rate of 0.018 nm/pulse (Fig. 1). A intensity profile of its autocorrelation function, perpendicularly to the steps, is plotted (dashed line) in Fig. 4c (right panel). The first maximum is at ¨180 nm, whereas that corresponding to the film grown at 0.0042 nm/pulse was at ¨550 nm. These results demonstrate that the bunching size (lateral and vertically) can be tuned with film growth rate and substrate miscut angle. 4. Summary In summary, we have presented a detailed statistical analysis of the surface morphology of epitaxial SRO films with giant bunching. We have found that the 3D islands appearing in nanometric films and that constitute the first stage in the bunching formation, have similar lateral size on substrates with varied miscut angle. They were compared with islands appearing on low-miscut angle substrates, and the differences have been established. We have demonstrated that bunching size can be laterally and vertically tailored by a combined selection of substrate miscut angle and film growth rate. Acknowledgements Financial support by the CICYT of the Spanish Government (projects MAT2005-5656 and NAN2004-9094) and FEDER is acknowledged. References [1] C.B. Eom, R.J. Cava, R.M. Fleming, J.M. Philips, R.B. van Dover, J.H. Marshall, J.W.P Hsu, J.J. Krajewski, W.F. Peck, Science 258 (1992) 1766. [2] H.N. Lee, H.M. Christen, M.F. Chisholm, C.M. Rouleau, D.H. Lowndes, Appl. Phys. Lett. 84 (2004) 4107. [3] D.H. Do, P.G. Evans, E.D. Isaacs, D.M. Kim, C.B. Eom, E.M. Dufresne, Nat. Mater. 3 (2004) 365. [4] L. Klein, Y. Kats, A.F. Marshall, J.W. Reiner, T.H. Geballe, M.R. Beasley, A. Kapitulnik, Phys. Rev. Lett. 84 (2000) 6090. [5] G. Herranz, B. Martı´nez, J. Fontcuberta, F. Sa´nchez, C. Ferrater, M.V. Garcı´a-Cuenca, M. Varela, Phys. Rev., B 67 (2003) 174423. [6] G. Herranz, F. Sa´nchez, J. Fontcuberta, M.V. Garcı´a-Cuenca, C. Ferrater, M. Varela, T. Angelova, A. Cros, A. Cantarero, Phys. Rev. B 71 (2005) 174411. [7] R.A. Rao, Q. Gan, C.B. Eom, Appl. Phys. Lett. 71 (1997) 1171. [8] F. Sa´nchez, G. Herranz, J. Fontcuberta, M.V. Garcı´a-Cuenca, C. Ferrater, M. Varela, Appl. Phys. Lett. (unpublished results). [9] F. Sa´nchez, U. Lu¨ders, G. Herranz, I.C. Infante, J. Fontcuberta, M.V. Garcı´a-Cuenca, C. Ferrater, M. Varela, Nanotechnology 16 (2005) S190. [10] F. Sa´nchez, M.V. Garcı´a-Cuenca, C. Ferrater, M. Varela, G. Herranz, B. Martı´nez, J. Fontcuberta, Appl. Phys. Lett. 83 (2003) 902. [11] F. Sa´nchez, G. Herranz, I.C. Infante, J. Fontcuberta, M.V. Garcı´a-Cuenca, C. Ferrater, M. Varela, Appl. Phys. Lett. 85 (2004) 1981. [12] J. Choi, C.B. Eom, G. Rijnders, H. Rogalla, D.H.A. Blank, Appl. Phys. Lett. 79 (2001) 1447. [13] E. Patella, A. Sgarlata, F. Arciprete, S. Nufris, P.D. Szkutnik, E. Placidi, M. Fanfoni, N. Motta, A. Balzarotti, J. Phys., Condens. Matter 16 (2004) S1503. [14] H.C. Jeong, E.D. Williams, Surf. Sci. Rep. 34 (1999) 171. [15] Z.M. Wang, L.L. Shultz, G.J. Salamo, Appl. Phys. Lett. 83 (2003) 1749. [16] S.H. Lee, G.B. Stringfellow, Appl. Phys. Lett. 73 (1998) 1703. [17] M. Kasu, N. Kobayashi, Appl. Phys. Lett. 62 (1993) 1262.

164

F. Sa´nchez et al. / Thin Solid Films 495 (2006) 159 – 164

[18] E.S. Fu, M.D. Johnson, D.J. Liu, J.D. Weeks, E.D. Williams, Phys. Rev. Lett. 77 (1996) 1091. [19] G. Patriarche, A. Ougazzaden, F. Glas, Appl. Phys. Lett. 76 (2000) 306. [20] N. Be´court, F. Peiro´, A. Cornet, J.R. Morante, P. Gorostiza, K. Michelakis, A. Georgakilas, Appl. Phys. Lett. 71 (1997) 2961. [21] WSxM*, http://www.nanotec.es.

[22] G. Herranz, B. Martı´nez, J. Fontcuberta, F. Sa´nchez, M.V. Garcı´a-Cuenca, C. Ferrater, M. Varela, Appl. Phys. Lett. 82 (2003) 85. [23] V. Repain, G. Baudot, H. Ellmer, S. Rousset, Mater. Sci. Eng., B, SolidState Mater. Adv. Technol. 96 (2002) 178. [24] M. Raible, S. Linz, P. Ha¨nggi, Phys. Rev., E 62 (2000) 1691.