Epitaxial growth of Cu(100) and Pt(100) thin films on perovskite substrates

Thin Solid Films 496 (2006) 317 – 325 www.elsevier.com/locate/tsf

Epitaxial growth of Cu(100) and Pt(100) thin films on perovskite substrates Andrew J. Francis, Yan Cao, Paul A. Salvador * Department of Materials Science and Engineering, Carnegie Mellon University, Pittsburgh, PA 15213, USA Received 2 February 2005; received in revised form 17 August 2005; accepted 30 August 2005 Available online 7 October 2005

Abstract Pulsed laser deposition has been used to grow epitaxially oriented thin films of Cu and Pt on (100)-oriented SrTiO3 and LaAlO3 substrates. Xray diffraction results illustrated that purely epitaxial Cu(100) films could be obtained at temperatures as low as 100 -C on SrTiO3 and 300 -C on LaAlO3. In contrast, epitaxial (100)-oriented Pt films were attained on LaAlO3(100) only when deposited at 600 -C. Atomic force microscopy images showed that films deposited at higher temperatures consisted of 3D islands and that flat, layered films were obtained at the lowest deposition temperatures. Importantly, Cu films deposited at 100 -C on SrTiO3(100) were both purely (100)-oriented and morphologically flat. Pt and Cu films displaying both epitaxial growth and smooth surfaces could be obtained on LaAlO3(100) only by using a three-step deposition process. High-resolution transmission electron microscopy demonstrated an atomically sharp Cu/SrTiO3 interface. The crystalline and morphological features of Cu and Pt films are interpreted in terms of the thermodynamic and kinetic properties of these metals. D 2005 Elsevier B.V. All rights reserved. Keywords: Copper(80); Platinum(375); Physical vapor deposition(370); Epitaxy(124)

1. Introduction The growth of thin metal films is of considerable scientific and technological interest, since supported metals are currently in wide use as electrodes, interconnects, magnetic layers, and catalysts [1– 6]. For metal films to be of practical use, it is very important to have control over their crystalline properties (orientation, grain size) as well as morphological features (3D particles or flat layers). In many instances, these metal films must be integrated with ceramic or semiconducting substrate layers, but the general characteristics of metal/ceramic interfaces are unfortunately still not well understood [1]. Epitaxial growth of metal thin films on ceramic substrates can provide insight into interfacial bonding, surface quality, strain, and other factors that can affect the properties of metal thin films in multilayer structures. The chemical inertness of Pt, particularly its stability against oxide formation, has made it a popular choice for studying metal films grown on ceramic substrates. Pt thin films have been deposited epitaxially on a number of single crystal ceramic substrates with low-index orientations, including * Corresponding author. Fax: +1 412 268 7596. E-mail address: [email protected] (P.A. Salvador). 0040-6090/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2005.08.367

SrTiO3(100) [7– 9], (110) [10], and (111) [11], and MgO(100) [12 – 19], using a variety of deposition techniques. Cu is another attractive choice as a thin metal film because it is used in thin film form as electrical interlayers [2,3] and supported catalysts [20,21], and is quite inexpensive compared to Pt. A number of studies exist describing the epitaxial growth and interface structure of Cu on single crystal sapphire(0001) [22 – 25], Si(100) [26 – 32], and Si(111) [29 – 33] substrates, as well as textured growth on glass substrates [32 – 35]. However, very few detailed studies exist describing Cu heteroepitaxy on commonly used cubic ceramic substrates like MgO [36] or SrTiO3 [37,38]. In the present work, single-crystal, (100)-oriented cubic SrTiO3 and LaAlO3 were chosen as substrates to take advantage of their close structural matches with Cu and Pt and promote heteroepitaxial growth. The metals are both facecentered cubic (fcc) and the oxides each adopt the perovskite structure, which can be considered to be an fcc derivative. Furthermore, the lattice parameters of the materials are quite similar: Cu (a = 0.3615 nm) has a lattice mismatch of 8.0% with SrTiO3 (a = 0.3905 nm) and just 4.8% with LaAlO3 (a = 0.379 nm) (see Ref. [39]). Likewise, Pt (a = 0.3924 nm) has a lattice mismatch of just
0.5% with SrTiO3 and
3.4% with LaAlO3. These structural similarities between the metals and

A.J. Francis et al. / Thin Solid Films 496 (2006) 317 – 325

the (100)-oriented oxide substrates should promote the epitaxial growth of (100)-oriented films. Both the crystalline and morphological characteristics of epitaxial thin metal films grown on ceramic substrates exhibit a clear temperature dependence; higher temperatures promote crystalline epitaxy and 3D island growth, while lower temperatures lead to randomly oriented (or (111)-textured), flat films. This work will demonstrate that (100)-oriented Cu films having flat surface morphologies can be obtained at a single deposition temperature, and that multi-step processing methods allow for such growth in other metal/perovskite systems. We discuss the roles of the intrinsic properties of Pt and Cu (surface diffusion rate and surface energy anisotropy) and processing characteristics (deposition temperature and rate) in determining film microstructure and surface morphology. These factors can offer insight into the kinetic and thermodynamic processes taking place during metal deposition. An improved understanding of the relevant factors in metal/ ceramic heteroepitaxy will allow for the prediction of growth characteristics and the selection of appropriate metal and ceramic materials for use in thin film structures. 2. Experimental details Polished single crystal substrates of SrTiO3(100) and LaAlO3(100) (miscut angle < 0.5-) were obtained from Crystal GmbH. Prior to film growth, all substrates were ultrasonically cleaned in acetone, followed by ethanol, for 5 min each. Samples were then etched 4 min in a 3 : 1 HCl : HNO3 solution, rinsed with distilled water, dried with a heat gun, and annealed 1 h at 800 -C in 13 Pa flowing O2. This treatment procedure has previously been shown to be effective for producing highquality low-index SrTiO3 surfaces [9,11,40– 43]. Once prepared, substrates were attached to the heating surface with conductive silver paint and placed into the deposition chamber. Samples were heated to the desired deposition temperature in vacuum (< 0.1 Pa), maintained by a turbomolecular pump. The deposition atmosphere was established by lowering the turbopump speed and introducing flowing Ar (for Cu) or O2 (for Pt) gas into the chamber to attain a dynamic pressure of 1.3 Pa. Deposited metal films were cooled in a dynamic vacuum (< 0.1 Pa) to prevent postoxidation. Ar gas was selected for Cu growth to avoid oxidation during growth. Pt is non-reactive, and thus O2 was selected for its possible surfactant effects [44,45]. Pulsed laser deposition (PLD) was performed using a Compex 201 laser (Lambda Physik). The metal targets were 0.25 mm thick Cu and Pt foils, 99.99% purity from Alfa Aesar. The KrF laser (k = 248 nm, pulse duration = 20 ns) was operated at a rate of 3 Hz, with a laser energy density at the target of å8.0 J/cm2. The target-to-substrate distance was maintained at å60 mm in all experiments, and targets were rotated around their centers during the deposition to keep their surfaces fresh. Following a short 5 min ablation (with the substrate shielded), to clean the target surface, Cu or Pt metal was deposited for 60 min. Reflectometry scans [46] indicated that the Cu films were grown at a rate of 4.0 T 0.3  10
3 nm/pulse, consistent with

transmission electron microscopy (TEM) measurements. This rate is considerably slower than that previously observed in our system for Pt (å18  10
3 nm/pulse) [47]. The crystal structures of the resultant films were characterized by X-ray diffraction (XRD). Conventional 2h-h scans were carried out to determine the crystalline orientation normal to the surface. Cu and Pt films that were confirmed to be (100)oriented were further examined using a 3-circle diffractometer, which recorded 2h-h, /, and x scans to determine crystalline quality, lattice parameters, and orientation; in-plane epitaxy; and mosaic spread, respectively. Cross-sectional TEM specimens were prepared using standard techniques: grinding of glued sandwich slices, followed by dimple grinding and Ar+ ion-beam milling. The primary energy of the Ar+ ions was 3.5 keV with a beam current of å15 AA. Conventional TEM was carried out using a JEM-2000EX (JEOL) microscope, operated at an accelerating voltage of 200 kV. High-resolution imaging and analysis were performed using a FEI Tecnai F20 (field emission gun) operating at 200 kV and equipped with a Gatan Imaging Filter. Film surface morphologies were analyzed with atomic force microscopy (AFM) using an AutoProbe CP atomic force microscope (Park Scientific Instruments), fitted with a 5 Am scan head for optimal lateral resolution, and operated in contact mode. 3. Results 3.1. Epitaxy of metal thin films Cu films were deposited on SrTiO3(100) and LaAlO3(100) substrates at temperatures ranging from room temperature (RT) to 600 -C, with the goal of depositing purely (100)-oriented Cu films having flat surface morphologies. The thicknesses of all films were evaluated using X-ray reflectometry, and were found to be in the range of 40– 46 nm. XRD results for Cu films deposited on SrTiO3(100) substrates are presented in Fig. 1.

STO (200) Intensity (log scale, arb. units)

318

Cu (200)

STO (300)

STO (400)

Cu (400)

a

b

Kβ

c

d Cu (111) e

30

50

70

90

110

2 theta (º) Fig. 1. Normal 2h-h XRD patterns for Cu films grown on SrTiO3(100) substrates at (a) 600 -C, (b) 500 -C, (c) 300 -C, (d) 100 -C, and (e) RT.

A.J. Francis et al. / Thin Solid Films 496 (2006) 317 – 325

a

b

Cu ψ=45º 2θ=43.35º

Cu (200)

LAO (300)

LAO Cu (400) (400)

a

b

Kβ

c d

Cu (111)

e 30

50

70

90

110

2 theta (º) Fig. 3. Normal 2h-h XRD patterns for Cu films grown on LaAlO3(100) substrates at (a) 600 -C, (b) 500 -C, (c) 300 -C, (d) 100 -C, and (e) RT.

thick Pt films deposited on LaAlO3(100) substrates via PLD. XRD patterns taken from Pt/LaAlO3(100) samples deposited at various temperatures (Fig. 4) demonstrate a quite different behavior than those recorded on Cu samples. At the highest deposition temperature, 600 -C, nearly perfectly oriented Pt(100) films are obtained. However, Pt(111) is the dominant orientation at 500 -C and virtually no (100)-oriented crystallinity is observed at deposition temperatures of 400 -C and below. The in-plane epitaxy of Pt(100)/LaAlO3(100) is not cube-oncube (Fig. 5); instead, the Pt film is rotated 45- about its normal with respect to the LaAlO3(100) substrate: (100)Pt||(100)LaAlO3 : [011]Pt||[001]LaAlO3. There have been several published instances of similar 45- azimuthal rotations for metal – ceramic epitaxy [48,49], but these usually exist to match better the two lattices, which is not the case here. Pt does have a larger mismatch with LaAlO3 (
3.4%) than with

Intensity (log scale, arb. units)

Intensity (log scale, arb. units)

SrTiO3 ψ=45º 2θ=39.95º

LAO (200) Intensity (log scale, arb. units)

The sharp Cu(200) and (400) peaks in these patterns indicated that (100)-oriented Cu can be grown on SrTiO3 at temperatures as low as 100 -C. No other Cu orientations or copper oxide impurity phases were evidenced by XRD. Fig. 2 gives /-scans of the {111} planes of Cu and SrTiO3 recorded from the Cu(100) film deposited at 600 -C, and is representative of data collected from all of the (100)-oriented films. The results of all the /-scans collected confirm that the films grew with a ‘‘cube-on-cube’’ orientation relationship of (100)Cu||(100)SrTiO3: [001]Cu||[001]SrTiO3. The XRD patterns presented in Fig. 3 illustrate that good quality epitaxial Cu films can also be grown on LaAlO3 substrates under proper conditions, although not over as wide of a temperature range as on SrTiO3(100). Films grown on LaAlO3(100) at temperatures  300 -C were purely (100)oriented. As was the case for films deposited on SrTiO3(100) substrates, Cu films that did deposit in a (100)-oriented manner formed cube-on-cube epitaxial relationships with LaAlO3(100), again pointing to the presence of a relatively low-energy interface structure. The very small, narrow peaks observed at 2h å 43- for films grown above 300 -C correspond to the weak LaAlO3(200) Kh reflection; the larger, broader peaks found at that location for films deposited at temperatures  200 -C, however, can be indexed as Cu(111) peaks. The decreased overall diffracted intensity of films deposited on LaAlO3(100) at low temperatures indicates that they are polycrystalline. Although Cu has a more favorable lattice match with LaAlO3 (4.8%) than with SrTiO3 (8.0%), Cu films deposited on LaAlO3 are of lower quality than those on SrTiO3. Thus, strain is not the key factor in determining the epitaxial crystallinity of these metal films. Further work is necessary to determine whether it is the structural characteristics of our LaAlO3 crystals—twin boundaries and a slight distortion from cubic symmetry—that adversely affect Cu thin film epitaxy. In order to extend our studies of Pt on perovskites [9,11,47] and to provide a basis for a more thorough comparison of Cu and Pt deposition characteristics, we report here on å200 nm

319

LAO (400)

LAO (300)

LAO Pt (200) (200) Pt (111)

Pt (400)

a

Pt (222)

b Cu(111)

c 0

90

180

270

360

Phi (º) Fig. 2. XRD /-scans of the {111} planes of (a) SrTiO3 and (b) Cu, for a Cu film deposited at 600 -C on SrTiO3(100). Values for the angles 2h and w are indicated.

30

50

70

90

110

2 theta (º) Fig. 4. Normal 2h-h XRD patterns for Pt films grown on LaAlO3(100) substrates at (a) 600, (b) 500, and (c) 400 -C.

320

A.J. Francis et al. / Thin Solid Films 496 (2006) 317 – 325

Intensity (arb. units)

a

LaAlO3 ψ=45º 2θ=41.22º

a

b

100

20

nm

nm 0.5 µm

0 b

0

Pt ψ=45º 2θ=39.95º 90

180

270

c

d

10

3

nm

nm

360

Phi (º)

SrTiO3 (0.5%), but as discussed previously for the case of Cu, interfacial strain does not appear to be the deciding factor of epitaxial quality. The cause of the 45- in-plane rotation thus remains unclear at this time, and further investigation is necessary to determine the mechanism responsible for the somewhat unusual behavior of Cu and Pt thin films on LaAlO3(100). 3.2. Surface morphology AFM was used to investigate the effect of deposition temperature on film surface morphologies. The topographs presented in Figs. 6 and 7 clearly demonstrate a flattening out of Cu film surfaces as the deposition temperature is lowered, consistent with typical observations of metal films on ceramic substrates [9,12,13,15,47]. For Cu films deposited on SrTiO3(100) (Fig. 6), the morphology changes from thick,

a

b

60

25

nm

nm

0

0.5 µm

0

c

d

10

1.5

nm

nm

0.5 µm

0.5 µm

widely spaced islands at 600 -C, to flatter, more coalesced islands from 300 to 500 -C, down to almost perfectly flat films at 100 -C and below. Note the change in height scale (black/ white contrast) as the films become flatter. The root mean square (rms) roughness values for these films are presented in Table 1 and show a sharp temperature dependence; for comparison, substrate rms roughness values are < 0.15 nm. Cu films deposited on LaAlO3(100) substrates show the same qualitative trend of surface character as a function of temperature (Fig. 7), with 3D particles giving way to smooth layers at lower growth temperatures. However, the absolute rms roughness values are higher for Cu films deposited on LaAlO3 than on SrTiO3 (Table 1). Pt films deposited on LaAlO3(100) also show qualitatively similar microstructural behavior, as shown in Fig. 8. The transition from 3D islands to nearly flat surfaces occurs at significantly higher temperatures for Pt. A comparison of the surface scans with the diffraction results (Fig. 4) demonstrates that these temperature-related improvements in surface quality come at a cost of oriented crystallinity. The key result from these XRD and AFM analyses is that the Cu/SrTiO3(100) samples prepared at 100 -C are both epitaxial and have flat surfaces. In general, it is difficult to select substrate temperatures that allow for both of these features, since epitaxy is favored at high temperature and flat surfaces at low temperatures. Deposition of Cu on SrTiO3 at Table 1 Summary of AFM rms roughness values

0.5 µm

0.5 µm

0

0.5 µm

Fig. 7. AFM topographs of Cu films deposited on LaAlO3(100) substrates at (a) 600, (b) 500, (c) 300, and (d) 100 -C. See Table 1 for rms roughness values.

Temp (-C)

0

0.5 µm

0

Fig. 5. XRD /-scans of the {111} planes of (a) LaAlO3 and (b) Pt, for a Pt film deposited at 600 -C on LaAlO3(100). Values for the angles 2h and w are indicated.

0

0

Fig. 6. AFM topographs of Cu films deposited on SrTiO3(100) substrates at (a) 600, (b) 500, (c) 300, and (d) 100 -C. See Table 1 for rms roughness values.

600 500 400 300 100 3-step

rms roughness (nm) Cu on SrTiO3(100)

Cu on LaAlO3(100)

Pt on LaAlO3(100)

14.0 4.8 – 2.0 0.1 1.1

21.4 5.8 – 3.0 0.4 0.6

23.8 4.3 0.6

0.5

A.J. Francis et al. / Thin Solid Films 496 (2006) 317 – 325

a

b

c

80

20

10

nm

nm

nm

0.5 µm 0

321

0.5 µm 0

0.5 µm 0

Fig. 8. AFM topographs of Pt films deposited on LaAlO3(100) substrates at (a) 600, (b) 500, and (c) 400 -C. See Table 1 for rms roughness values.

3.3. Three-step processing To overcome the challenge of growing thin Cu and Pt films that are both epitaxially oriented and morphologically smooth, we have employed a three-step growth process originally developed by Wagner et al. to deposit Pd(100) thin films on SrTiO3(100) substrates [50]. First, epitaxial ‘‘seed’’ grains are grown for a short time at a high temperature previously determined to be suitable for (100) epitaxy. Next, the temperature is lowered to a value where full substrate coverage and flat surfaces are expected, though this metal will not initially be epitaxial. The third step is a 1 h post-annealing intended to crystallize flat, epitaxial films by the growth of the epitaxial seeds deposited in step one without loss of the flat surface morphology established in step two. This procedure has been successfully adapted by our group for fabrication of wellcrystallized epitaxial Pt(100) [9] and (621) [9,47] films having flat surfaces on SrTiO3 substrates. In this work, Cu films were grown using temperatures for the three steps of 300, 100, and 300 -C; corresponding Pt temperatures were 600, 250, and 600 -C. In both cases, seeding in step one was 5 min (900 pulses), deposition in step two lasted 55 min (10,800 pulses), and heating in step three took 1 h. The combination of XRD and AFM results presented in Figs. 9 and 10 illustrates that the three-step process can be used to grow flat, epitaxial Cu(100) and Pt(100) films. Fig. 9 shows that purely (100)-oriented Cu can be grown on SrTiO3(100) as well as LaAlO3(100), and that (100)-oriented Pt films can be grown on LaAlO3(100), each using a multistep procedure. Phi scans taken from these films are just like those recorded for single-step samples, meaning that the three-step Cu films are cube-on-cube and the three-step Pt films display a 45- in-plane rotation. These orientation relationships can be taken as proof that the three-step process operates in the intended fashion: the high-temperature seeds deposited in the first step grow during the third step, maintaining their epitaxial relationships, at the expense of the film deposited in the second step.

Furthermore, the surface morphologies of the epitaxial three-step films are relatively flat, as illustrated by the AFM images in Fig. 10 and rms roughness values in Table 1. The overall character of the Cu and Pt metal surfaces is smooth, although there are small pinholes or hillocks distributed on the surface. These surfaces are very similar to the best Pt(621) film surfaces previously achieved by the three-step processing technique [9]. These results demonstrate that Pt and Cu films on LaAlO3(100) can be grown having both relatively flat surfaces and (100) crystallinity, something which was not possible during single-step depositions. 3.4. Cu/SrTiO3(100) interfacial character Additional experiments were carried out to investigate the possible existence and effects of an interfacial copper oxide layer between film and substrate. A high-resolution TEM (HRTEM) cross-sectional image (Fig. 11a) of the Cu / SrTiO3(100) film grown at 300 -C clearly demonstrates that no second phase is present at the interface. Misfit dislocations between Cu and SrTiO3 (black circles) can be easily identified, again indicating a sharp interface between ceramic substrate and a fully relaxed metal film. A dislocation occurs, on average, for approximately each 13.5 unit cells of Cu. This is consistent with the dislocation density predicted by lattice mismatch. The corresponding electron diffraction pattern, taken along a [001] zone axis (Fig. 11b) confirms the results from XRD that

STO Cu (200) (200) Intensity (log scale, arb. units)

100 -C represents an overlap between regions of flat surface and crystalline epitaxy. Cu and Pt on LaAlO3 represent cases where the two structural features are exclusive for most combinations of deposition temperature and rate that have been studied. To obtain films having both of the desired characteristics for the other metal/ceramic combinations being studied in this work, we turn our attention to multi-step growth techniques.

STO (300)

STO (400)

a LAO (200)

LAO Cu (400) (400)

LAO (300)

Cu (200)

b Pt LAO (200) (200)

LAO (300)

Pt (400)

c 30

Cu (400)

50

70

90

LAO (400)

110

2 theta (º) Fig. 9. Normal 2h-h patterns for films grown via a three-step deposition procedure: (a) Cu/SrTiO3(100), (b) Cu/LaAlO3(100), and (c) Pt/LaAlO3(100).

322

A.J. Francis et al. / Thin Solid Films 496 (2006) 317 – 325

a

b

c

8

5

3

nm

nm

nm

0

0.5 µm

0

0.5 µm

0.5 µm

0

Fig. 10. AFM topographs of films deposited by 3-step processes: (a) Cu/SrTiO3(100), (b) Cu/LaAlO3(100), and (c) Pt/LaAlO3(100). See Table 1 for rms roughness values.

illustrated cube-on-cube epitaxy. The lack of an interfacial oxide is in agreement with previous studies of evaporated Cu films on SrTiO3(100), where ultraviolet photoelectron spectroscopy [37] and high-resolution electron energy loss spectroscopy [38] indicated that the interaction between Cu and SrTiO3(100) is weak and that no interfacial oxide is formed. Furthermore, an extra deposition was conducted to see how the presence of an interfacial oxide layer would affect the subsequent Cu growth. A Cu film was grown via PLD at 300 -C as described earlier, except that the initial 2 min of growth (å2 nm) took place in 1.3 Pa of O2 to intentionally oxidize the depositing metal. The crystallinity of thin film Cu is completely changed by the presence of even a small amount of copper oxide. A film grown entirely in Ar at this temperature displays sharp Cu(h00) peaks (Fig. 12a), as presented earlier, while poor crystalline epitaxy is observed for the film with the oxide layer (Fig. 12b). These diffraction scans illustrate clearly that

a Cu

epitaxial growth of Cu(100) films on ceramic substrates depends strongly on the character of the film/substrate interface and that the presence of an interfacial phase has an adverse effect on crystalline epitaxy. As the HR-TEM image shows, we are able to prevent oxidation of the Cu films and obtain a sharp epitaxial metal/ceramic interface. 4. Discussion We have demonstrated that both Cu and Pt films can be grown epitaxially on (100)-oriented perovskite substrates using PLD, and that their surface morphologies change from 3D islands to flat layers as deposition temperature is lowered. Although the temperature dependencies of film properties were qualitatively similar for the two metals, Cu films grew as epitaxial 3D islands at significantly lower temperatures than Pt films. An overlap between regimes of (100) epitaxy and flat surface morphologies was also observed for Cu/SrTiO3(100). These results allow us to provide some insight into the variables governing metal thin film growth on oxide substrates and explain the differences in the behavior of the two metals. In this section, we discuss how thermodynamic (surface energy) and kinetic (diffusion) factors intrinsic to the metals influence film characteristics. 4.1. Surface energy Growth of (100)-oriented fcc metals is challenging because close packed {111} planes have the lowest relative surface

SrTiO3

STO (200)

b

STO (200)

Cu (200)

STO (020)

Intensity (log scale, arb. units)

2 nm

30 Fig. 11. (a) Cross-sectional HR-TEM image of the Cu/SrTiO3(100) interface for a film deposited at 300 -C, and (b) corresponding electron diffraction pattern. Circles in (a) indicate misfit dislocations.

STO (300)

STO (400)

Cu (200) a

Cu (400)

Kβ

Cu (111) b

50

70

90

110

2 theta (º) Fig. 12. Normal 2h-h XRD patterns for Cu films grown on SrTiO3(100) substrates at 300 -C, (a) without and (b) with a thin layer deposited in oxygen.

A.J. Francis et al. / Thin Solid Films 496 (2006) 317 – 325

energy [51,52], and thus metals naturally prefer to grow with a (111) orientation. The difference, or anisotropy, between the surface energies of these competing orientations plays a key role in determining whether interfacially driven epitaxy can be achieved, and can help explain the disparities between Pt and Cu films. A recent report by Chatain et al. [53] on the equilibrium crystal shape (ECS) of Cu has shown that, compared to Pb and Au (the only fcc metals for which reliable ECS data is available), Cu has a relatively low surface energy anisotropy (Table 2, Refs. [54 – 56]). Quality experimental data concerning Pt’s surface energy anisotropy is not yet available, yet a number of calculated values have been published for both metals (Table 3). Although there is considerable spread between the absolute values of c (100), c (111), and even the anisotropy ratio c (100) / c (111) for the various reports, the general trend is clear: Cu has a considerably lower surface anisotropy than Pt [57 – 59]. These studies indicate that the usual preference shown by fcc metals to expose {111} surface facets is therefore not as pronounced for Cu as for other fcc metals, including Pt. The energetic preference for Cu to grow with a (111) orientation is relatively low, and the (100) epitaxial orientation is more easily stabilized by perovskite (100) surfaces. This is in good agreement with the observed crystalline orientations of deposited films in this work (Figs. 1 –4), wherein the (100)oriented Cu films are stable down to low temperatures. On the other hand, Pt, which has a relatively high surface energy anisotropy, grows with a (111) orientation for T < 500 -C. In this case, the interfacial stabilization is unable to compensate for Pt’s strong (111) energetic preference. In addition to the material-dependent differences, surface energy anisotropy is also dependent on the homologous temperature, T / T M, of a metal. As temperature is increased, the surface energy anisotropy is lowered (the lower limit being the fully isotropic liquid state). The energetic preference to expose {111} planes thus becomes lower as deposition temperature is increased, while the film/substrate interfacial energy is unaffected, promoting the epitaxial (100) orientation at higher temperature. This temperature-dependence of surface energy anisotropy helps to explain how it is possible to attain (100)-oriented Pt films at high temperatures, where anisotropy is low, but not at reduced temperatures. The concept of homologous temperature also reinforces the differences between the behavior of Pt and Cu. Films of Cu(100) are stable down to 100 -C (T / T M = 0.27) while Pt films transform to (111) orientations below 600 -C (T / T M = 0.42). This highlights the fact that the intrinsic low surface energy anisotropy of Cu, relative to Pt, facilitates growth of Cu(100) while Pt(100) is more difficult to achieve.

Cu Pb Au

Table 3 Calculated values of c(100) / c(111) for selected fcc metals (J/cm2) Cu

Ag

Pt

Ref.

1.09 1.15 1.31

1.14 1.22 1.30

1.15 1.25 1.38

[57] [58] [59]

The faster deposition rate in our PLD system for Pt also affects the energetics of the system because the supersaturation of Pt atoms during each pulse is higher than that for Cu. This kinetic rate factor increases the driving force for metal film nucleation on the substrate surface, for both epitaxial (100) and bulk textured (111) orientations. Since Pt has both a high supersaturation and a high surface energy anisotropy, there will be an increased probability for the formation of Pt(111) nuclei. Cu, on the other hand, is more energetically isotropic and is grown at a slower rate (low supersaturation), and thus nucleates only in the (100) fashion made thermodynamically possible by a low-energy film –substrate interface. Finally, Cu has a higher affinity for oxygen than Pt, which will also affect the film/substrate interfacial energy through interactions between the deposited metal and the oxide surface. Wagner et al. [60] demonstrated that, for molecular beam epitaxy growth on SrTiO3(100), metals having high oxygen affinities were more likely to grow (100)-oriented. The Cu(100)/SrTiO3(100) interface may gain some additional stability from interfacial bonding whereas Pt(100)/SrTiO3(100) will not, since Pt has in fact a slight repulsion to oxygen. As illustrated by the HR-TEM results (Section 3.4), the relatively high oxygen affinity of Cu does lead to a sharp film/substrate interface, yet avoids the formation of an oxide phase that could be deleterious to epitaxy. 4.2. Diffusion The thermodynamic preference of metal/ceramic systems is to leave exposed substrate surface and form 3D island-like morphologies. Even when predicted thermodynamically, however, 3D growth is not always observed for metal films. At low temperatures a ‘‘quasi-2D’’, or ‘‘2D island’’ growth mode [61] dominates, and flat metal films can be grown, as shown in this report and in previous studies [12,16,50]. It has been consistently observed in the literature [12 – 14,24] that the appearance of these morphological regimes can be attributed to changes in the ability of metal adatoms to diffuse across substrate surfaces. Although metal diffusion on ceramic surfaces is not wellquantified, an empirical relationship for fcc metal surface selfdiffusion for fcc metals was given by Gjostein [62]: DS ¼ 0:014expð
13TM =kB T Þ:

Table 2 Experimentally determined surface energy anisotropies c(100) / c(111)

Ref.

1.004 1.008 1.025 1.040

[53] [54] [55] [56]

323

ð2Þ

Fig. 13 shows plots of surface diffusion coefficient (D S) versus temperature for Pt and Cu, calculated using this approximation. Cu is expected to exhibit surface diffusion several orders of magnitude faster than Pt at any given temperature; thus, the transition from island growth to layered growth should occur at higher temperatures for Pt. This

324

A.J. Francis et al. / Thin Solid Films 496 (2006) 317 – 325

-5

Cu

log(Ds)

-10

Pt -15

-20 0

200

400

600

T (ºC) Fig. 13. Plot of estimated surface diffusion coefficients vs. reciprocal temperature for Cu(n), and Pt(O).

behavior is mapped reasonably well by comparing growth of the two metals on LaAlO3(100). The D S of Pt at 300 -C is near that of Cu at 100 -C (Fig. 13), and from the experimental data (Table 1), these temperatures are when the two metals become relatively flat (rms roughness < 1 nm). Even when 3D island growth is predicted by simple energetics, films will often coalesce above a critical thickness. The Cu and Pt films deposited at high temperatures in this work have not reached this stage, as large 3D islands remain separated by regions of exposed substrate area. The lowtemperature films do have flat surface morphologies, which is the case even for very thin samples. This indicates that the films grow either layer-by-layer or by coalescence of 2D islands during the first stages of growth. If very thin islands do coalesce, they do because limited diffusion at low temperatures prohibits coarsening and growth of large 3D particles (as with high-temperature Pt films) and exposure of bare substrate surface. Thus, surface diffusion is responsible for the smooth morphology regardless of the initial mechanism of metal growth at low temperature. During PLD, there is significant relaxation time between pulses, and one can calculate an approximate total distance d that an adatom may p travel ffiffiffiffiffiffiffi before the next plume of ablated species arrive as d ¼ DS t , where t is the time between pulses. Consider the calculated surface diffusion coefficients for Pt and Cu at 600 -C, which are D S = 3.5  10
9 cm2/s and 5.7  10
7 cm2/s, respectively. These lead to approximate diffusion distances for Pt and Cu of 0.4 and 4.4 Am, which is more than enough to reach any thermodynamically desired surface destination. At 100 -C, these distances are lowered drastically, to 0.01 and 5 nm, for Pt and Cu respectively. This means that Pt is virtually immobile at this temperature but, as we have seen, this temperature is sometimes suitable for Cu to exhibit oriented crystallinity but not 3D island formation. This diffusion distance of Cu at 100 -C (5 nm) represents the point where nearly flat surfaces are observed (Fig. 7). Pt atoms are able to move the same distance at approximately 300 -C, also corresponding to the point where flat surfaces are obtained [9]. It is also necessary to consider the significantly faster deposition rate of Pt (0.018 nm/pulse) relative to Cu (0.004

nm/pulse) in our system. The higher concentration of Pt atoms arriving with each pulse will increase the probability of clustering and slow the diffusion of Pt across the growth surface, further hindering Pt atomic movement relative to Cu. In summary, we have demonstrated that high-quality epitaxial thin films of Cu(100) can be grown on SrTiO3(100) and LaAlO3(100), and that Pt(100) films can also be obtained on LaAlO3(100). HR-TEM images demonstrate a sharp Cu/ SrTiO3(100) interface. For both metals, deposition at high temperature leads to growth of epitaxial 3D islands; as the temperature is lowered, flatter surface structures are attained. Cu(100) films are stable even at low deposition temperatures while Pt converts to a (111) orientation at a relatively high growth temperature. A key observation is that Cu films deposited on SrTiO3(100) at 100 -C are both epitaxially (100)-oriented and morphologically flat. The differences in the responses of the two metals to temperature changes can be understood by considering the kinetic and thermodynamic properties of Cu and Pt. The lower surface energy anisotropy, faster diffusion, lower T M, and slower deposition rate of Cu relative to Pt all make Cu films more likely to adopt the thermodynamically preferred structures, as observed here. For the other film/substrate pairs, where such an overlap was not possible [Cu or Pt on LaAlO3(100)], a three-step deposition procedure resulted in epitaxial, morphologically flat films. We believe that consideration of these film growth parameters permits a better understanding and prediction of the heteroepitaxial behavior of metal thin films on ceramic substrates. Acknowledgements AJF was supported by a National Defense Science and Engineering Graduate fellowship sponsored by the Office of the Deputy Under Secretary of Defense for Science and Technology and the Army Research Office. This work was also supported partially by the MRSEC program of the National Science Foundation under Award Number DMR-0079996. References [1] C.T. Campbell, Surf. Sci. Rep. 27 (1997) 1. [2] D.A. Glocker, S.I. Shah (Eds.), Handbook of Thin Film Process Technology, Institute of Physics Pub., Bristol, UK, 1995. [3] K. Niwa, Y. Kotaka, Y. Goto, Y. Imanaka, in: R.S.C. Smart, J. Nowotny (Eds.), Ceramic Interfaces: Properties and Applications, IOM Communications, London, 1998, p. 409. [4] G.R. Harp, Thin Films: Heteroepitaxial Systems, vol. 15, 1999, p. 117. [5] K.R. Coffey, M.A. Parker, J.K. Howard, IEEE Trans. Magn. 31 (1995) 2737. [6] F. Loffler, Surf. Coat. Technol. 132 (2000) 222. [7] B.S. Kwak, P.N. First, A. Erbil, B.J. Wilkens, J.D. Budai, M.F. Chisholm, L.A. Boatner, J. Appl. Phys. 72 (1992) 3735. [8] A.D. Polli, T. Wagner, T. Gemming, M. Ru¨hle, Surf. Sci. 448 (2000) 279. [9] A.J. Francis, P.A. Salvador, 106th Annual Meeting and Exposition of the American Ceramic Society, Indianapolis, U.S.A., April 2004, Ceramic Transactions, vol. 158, 2005, p. 37. [10] G.R. Harp, R.F.C. Farrow, R.F. Marks, J.E. Vazquez, J. Cryst. Growth 127 (1993) 627. [11] A. Asthagiri, C. Niederberger, A.J. Francis, L.M. Porter, P.A. Salvador, D.S. Sholl, Surf. Sci. 537 (2003) 134.

A.J. Francis et al. / Thin Solid Films 496 (2006) 317 – 325 [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38]

C. Gatel, P. Baules, E. Snoeck, J. Cryst. Growth 252 (2003) 424. K.H. Ahn, S. Baik, S.S. Kim, J. Mater. Res. 17 (2002) 2334. P.C. McIntyre, C.J. Maggiore, M. Nastasi, J. Appl. Phys. 77 (1995) 6201. J. Narayan, A. Tiwari, K. Jagannadham, O.W. Holland, Appl. Phys. Lett. 64 (1993) 2093. R.E. Leuchtner, D.B. Chrisey, J.S. Horwitz, K.S. Grabowski, Surf. Coat. Technol. 51 (1992) 476. J.F.M. Cillessen, R.M. Wolf, D.M. de Leeuw, Thin Solid Films 226 (1993) 53. M. Morcrette, A. Gutierrez-Llorente, W. Seiler, J. Perrie`re, A. Laurent, P. Barboux, J. Appl. Phys. 88 (2000) 5100. B.M. Lairson, M.R. Visokay, R. Sinclair, S. Hagstrom, B.M. Clemens, Appl. Phys. Lett. 61 (1992) 1390. F. Zaccheria, R. Psaro, N. Ravasio, Recent Res. Dev. Catal. 2 (2003) 23. J. Zhou, Q.-H. Xia, S.-C. Shen, S. Kawi, K. Hidajat, J. Catal. 225 (2004) 128. P.J. Møller, Q. Guo, Thin Solid Films 201 (1991) 267. G. Dehm, C. Scheu, M. Ru¨hle, R. Raj, Acta Mater. 46 (1998) 759. G. Dehm, M. Ru¨hle, G. Ding, R. Raj, Philos. Mag., B 71 (1995) 1111. C. Scheu, Interface Sci. 12 (2004) 127. I. Hashim, B. Park, H.A. Atwater, Appl. Phys. Lett. 63 (1993) 2833. C.S. Liu, L.J. Chen, Appl. Surf. Sci. 92 (1996) 84. F. Thie´ry, Y. Pauleau, L. Ortega, J. Vac. Sci. Technol., A 22 (2004) 30. T. Nakahara, S. Ohkura, F. Shoji, T. Hanawa, K. Oura, Nucl. Instrum. Methods Phys. Res., B 45 (1990) 467. C.-A. Chang, J. Appl. Phys. 67 (1990) 566. J.B. Lai, L.J. Chen, J. Appl. Phys. 87 (2000) 2237. T. Ohmi, T. Saito, M. Otsuki, T. Shibata, J. Electrochem. Soc. 138 (1991) 1089. T. Hanabusa, K. Kusaka, O. Sakata, Thin Solid Films 459 (2004) 245. J. Musil, A.J. Bell, M. Cepera, Czechoslov. J. Phys. 45 (1995) 249. R.J. Nika, P.M. Hall, IEEE Trans. Components Hybrids Manuf. Technol. CHMT-2 (1979) 412. D.M. Borsa, D.O. Boerma, Surf. Sci. 548 (2004) 95. J.E.T. Andersen, P.J. Møller, Thin Solid Films 186 (1990) 137. T. Conard, A.-C. Rousseau, L.M. Yu, J. Ghijsen, R. Sporken, R. Caudano, R.L. Johnson, Surf. Sci. 359 (1996) 82.

325

[39] G.S. Rohrer, Structure and Bonding in Crystalline Materials, Cambridge University Press, 2001. [40] T.-D. Doan, J.L. Giocondi, G.S. Rohrer, P.A. Salvador, J. Cryst. Growth 225 (2001) 178. [41] M. Kawasaki, K. Takahashi, T. Maeda, R. Tsuchiya, M. Shinohara, O. Ishiyama, T. Yonezawa, M. Yoshimoto, H. Koinuma, Science 266 (1994) 1540. [42] G. Koster, B.L. Kropman, G.J.H.M. Rjinders, D.H.A. Blank, H. Rogalla, Appl. Phys. Lett. 73 (1998) 2920. [43] T. Ohnishi, K. Shibuya, M. Lippmaa, D. Kobayashi, H. Kumigashira, M. Oshima, H. Koinuma, Appl. Phys. Lett. 85 (2004) 272. [44] T. Goto, J.R. Vargas, T. Harai, Mater. Trans. 40 (1999) 209. [45] M.H. Kim, B. Park, E. Yoon, D.-S. Lee, B. Park, H.-J. Woo, D.-I. Chun, J. Ha, J. Mater. Res. 14 (1999) 1255. [46] P.A. Salvador, T.-D. Doan, B. Mercey, B. Raveau, Chem. Mater. 10 (1998) 2592. [47] A.J. Francis, P.A. Salvador, J. Appl. Phys. 96 (2004) 2482. [48] P.C. Chow, R. Paniago, R. Forrest, S.C. Moss, S.S.P. Parkin, D. Cookson, Structure and Evolution of Surfaces, Boston, MA, USA, December 2 – 5, 1996, Mat. Res. Soc. Symp. Proc., vol. 440, 1997, p. 359. [49] A. Madan, X. Chu, S.A. Barnett, Appl. Phys. Lett. 68 (1996) 2198. [50] T. Wagner, G. Richter, M. Ru¨hle, J. Appl. Phys. 89 (2001) 2606. [51] C.L. Liu, J.M. Cohen, J.B. Adams, A.F. Voter, Surf. Sci. 253 (1991) 334. [52] L. Vitos, A.V. Ruban, H.L. Skriver, J. Kolla´r, Surf. Sci. 411 (1998) 186. [53] D. Chatain, V. Ghetta, P. Wynblatt, Interface Sci. 12 (2004) 7. [54] M. McLean, Acta Metall. 19 (1971) 387. [55] J.C. Heyraud, J.J. Metois, Surf. Sci. 128 (1983) 334. [56] Z. Wang, P. Wynblatt, Surf. Sci. 398 (1998) 259. [57] S.M. Foiles, M.I. Baskes, M.S. Daw, Phys. Rev., B 33 (1986) 7983. [58] T. Ning, Q. Yu, Y. Ye, Surf. Sci. 206 (1988) L857. [59] A.M. Rodrı´guez, G. Bozzolo, J. Ferrante, Surf. Sci. 289 (1993) 100. [60] T. Wagner, A.D. Polli, G. Richter, H. Stanzick, Z. Metallkd. 92 (2001) 701. [61] J.H. van der Merwe, Interface Sci. 1 (1993) 77. [62] N.A. Gjostein, in: J.J. Burke, N.L. Reed, V. Weiss (Eds.), Surfaces and Interfaces, Syracuse Univ. Press, 1967.