Pulsed laser deposition of SrRuO 3 thin-ﬁlms: The role of the pulse repetition rate

SrRuO 3 thin-ﬁlms were deposited with different pulse repetition rates, f dep , epitaxially on vicinal SrTiO 3 substrates by means of pulsed laser deposition. The measurement of several physical properties (e.g., composition by means of X-ray photoelectron spectroscopy, the out-of-plane lattice parameter, the electric conductivity, and the Curie temperature) consistently reveals that an increase in laser repetition rate results in an increase in ruthenium deﬁciency in the ﬁlms. By the same token, it is shown that when using low repetition rates, approaching a nearly stoichiometric cation ratio in SrRuO 3 becomes feasible. Based on these results, we propose a mechanism to explain the widely observed Ru deﬁciency of SrRuO 3 thin-ﬁlms. Our ﬁndings demand these theoretical considerations to be based on kinetic rather than widely employed thermodynamic arguments. © 2016 Author(s). All article content, except where oth-erwise noted, is licensed under a Creative Commons license Thin ﬁlms are the most frequently investigated sample form of the metallic perovskite oxide SrRuO 3 . 1 This dominance is mainly due to the low lattice mismatch of SrRuO 3 with SrTiO 3 substrates and to the fact that SrRuO 3 has become the standard electrode material in

SrRuO 3 thin-films were deposited with different pulse repetition rates, f dep , epitaxially on vicinal SrTiO 3 substrates by means of pulsed laser deposition. The measurement of several physical properties (e.g., composition by means of X-ray photoelectron spectroscopy, the out-of-plane lattice parameter, the electric conductivity, and the Curie temperature) consistently reveals that an increase in laser repetition rate results in an increase in ruthenium deficiency in the films. By the same token, it is shown that when using low repetition rates, approaching a nearly stoichiometric cation ratio in SrRuO 3 becomes feasible. Based on these results, we propose a mechanism to explain the widely observed Ru deficiency of SrRuO 3 thin-films. Our findings demand these theoretical considerations to be based on kinetic rather than widely employed thermodynamic arguments. © 2016 Author(s). All  Thin films are the most frequently investigated sample form of the metallic perovskite oxide SrRuO 3 . 1 This dominance is mainly due to the low lattice mismatch of SrRuO 3 with SrTiO 3 substrates and to the fact that SrRuO 3 has become the standard electrode material in a variety of all-oxide devices, particularly for the application in novel data-storage concepts. 2-7 Fueled by the desire to better understand the basic physical properties of SrRuO 3 as well as to optimize device performance, researchers have developed a mature understanding of thin-film growth modes-particularly for pulsed laser deposition (PLD). 1,8 Here the influence of the laser fluence, 9 deposition temperature, 9,10 and oxygen partial pressure 11 is of pivotal importance. A major remaining challenge for the PLD growth of SrRuO 3 is Ru-loss during deposition. This loss is commonly attributed to the high vapor pressure of various Ru x O y species under typical deposition conditions and believed to prohibit the stoichiometric (1:1:3) growth of SrRuO 3 thin-films. 12,13 This argument is based on thermodynamics. The major principle of PLD, however, is the prevalence of kinetics over thermodynamics. 14,15 In this study, we reexamine the issue of Ru deficiency in PLD-grown thin-films of SrRuO 3 and the question of thermodynamics versus kinetics. Specifically, we show that deliberate control of deposition kinetics -i.e., variation of the pulse repetition rate -is sufficient to systematically vary the Sr:Ru ratio of PLD deposited thin-films and in this manner influence their structural as well as electric and magnetic properties.
Thin-film samples of SrRuO 3 were grown on TiO 2 -terminated, nominally undoped SrTiO 3 substrates (Crystec GmbH, Berlin, Germany) by means of PLD. The SrTiO 3 substrates were pre-treated according to the standard procedure, 16 in order to obtain the TiO 2 -terminated, vicinal surface structure with terrace widths ranging from 200 nm to 450 nm.
SrRuO 3 was ablated in an oxygen atmosphere of pO 2 = 0.133 mbar (p base ≤ 5 × 10 −5 mbar) from a SrRuO 3 -target (Ø ≈ 2.5 cm, 99.9% purity of raw materials, Praxair S.T. Technology, Inc., Indianapolis, USA) rotating with 20 rpm by means of a KrF-eximer laser (LPX305, Lambda Physik, Germany, λ = 248 nm, pulse duration 30 ns) operated in constant energy mode at repetition rates ranging from f dep = 2 Hz to f dep = 18 Hz. Before entering the deposition chamber, the laser light passed a physical aperture, which determines the spot area and eliminates effects resulting from a potentially varying beam divergence. 17 The laser pulse energy at the deposition chamber was adjusted by means of a beam attenuator before each deposition to E pulse = 105 mJ. The laser was then focused into an area of approximately 5 mm × 2 mm on the target. The transmittance of the chamber viewport is T ≈ 0.3; hence a laser fluence of approximately 0.3 J cm 2 is obtained. The substrate target distance in the PLD chamber (Neocera, Inc., College Park, Maryland, USA) was fixed at d S-T = 7.5 cm and the heater-temperature was set to T dep = 873 K.
After deposition, samples were cooled down to room temperature within two hours in an oxygen atmosphere of pO 2 = 500 mbar. Typical film thicknesses were in the range of d film = (40-60) nm, as determined by means of X-ray reflectometry (XRR) measurements. The deposition rate was constant for all films and in the order of 6 pm/pulse.
In order to confirm epitaxial relations of SrRuO 3 films with the substrate, X-ray diffraction (XRD) measurements were performed (X'Pert PRO MRD, PANalytical B.V., Almeno, Netherlands). The surface morphology of SrRuO 3 thin films was examined with an atomic force microscope (AFM, ULTRA Objective Pico Station, SiS GmbH, Herzogenrath, Germany) operated in tapping mode with a Si cantilever (tip radius approximately 20 nm). Conductivity measurements were performed with a physical property measurement system (PPMS) (Quantum Design INC, San Diego, USA) employing samples in four-terminal Hall-bar geometry. X-ray photoelectron spectroscopy (XPS) measurements were conducted with a PHI 5000 Versa Probe (Physical Electronics, Inc., Chanhassen, USA) with Al K α X-ray illumination, a pass energy of 29.35 eV, and a take-off angle of 45 • to determine the Sr:Ru ratio. Electron neutralization was employed in order to ensure consistent measurement conditions with the insulating SrO reference sample. The data were quantified and the SrO component was fitted using CasaXPS1 18 in the same fashion as previously reported: Sr 3d spectra similarly to those shown in Ref. 19 were obtained. By comparison with respective reference spectra, the Sr 3d peaks were either assigned to SrO/SrCO 3 or SrRuO 3 . As all measurements-including one of the reference sampleswere conducted ex situ, carbonate formation is unavoidable. This allows us only to distinguish reliably between the SrO/SrCO 3 and SrRuO 3 peaks. Since SrO is not homogeneously distributed 19 in the film, it represents rather an impurity than an integral component. For this reason, only the intensity of the Sr 3d SrRuO 3 was employed to calculate the Sr:Ru ratio of the differently deposited films.
The results are presented in Fig. 1(a): samples deposited with the highest repetition rate possess the lowest ruthenium content and thus more ruthenium vacancies. Furthermore, the overall fraction of SrO increases as the ruthenium content decreases (Fig. 1). In particular, deposition at low repetition rates results in films with stoichiometric Sr:Ru ratio (Sr/(Sr + Ru) = 0.5). This somewhat questions the wide belief that deposition of stoichiometric 1:1:3 SrRuO 3 is generally infeasible by means of PLD. 12 As a strong dependence on the film behaviour with the terrace width has been reported, 20 we plotted the Sr:Ru ratio versus the terrace width as well ( Fig. 1(b)). In this case, however, no clear trend could be observed.
A further conspicuity of our results is the deviation of the target cation ratio from the expected ideal value of Sr/(Sr + Ru) = 0.5. This might be due to the preparation conditions of the target itself, as the loss of volatile RuO x is a known issue in this context as well. 21 Another potential reason for the observed difference from ideal behaviour is that the information depth of XPS is limited to first few nm's of the sample. As our target was constantly exposed to low-pressure conditions in the PLD chamber, its surface and the bulk composition might slightly deviate from each other.
The same holds for the first few nm's of the films. Hence do the values depicted in Fig. 1 not necessarily represent the true ratios in the bulk of the film.
Nonetheless, we consider the observed trend in Ru non-stoichiometry to reliably reflect the differences between the samples. One reason for this assumption is that the same trends were observed for XPS analysis with various photoemission angles and therefore a different depth of origin of the analyzed electrons.
As correlations between Ru non-stoichiometry and structural parameters as well as electric and magnetic properties of SrRuO 3 have been extensively studied, further validation of our hypothesis can be obtained from structural analysis and conductivity measurements.
Reciprocal space mapping (RSM) around the SrTiO 3 (103) and the SrRuO 3 (332)/(420) (both reflexes are given, as in-plane twinning cannot be excluded 22 ) peaks confirms that epitaxial relationship between film and substrate still holds for the samples deposited at the highest repetition rate (cf. Fig. 2(a)). Due to the in-plane clamping of SrRuO 3 to the SrTiO 3 substrate, 2Θ − ω-scans probe the out-of-plane lattice parameter thus the strain-state of the film.
Measurements reveal that the out-of-plane lattice parameters range from d (220) = c pc = 3.95 Å to c pc = 4.03 Å. Here c pc is the pseudocubic out-of-plane lattice parameter of SrRuO 3 . 23 Values for c pc and = (c pc /a 0pc − 1) are plotted in Fig. 2 as a function of f dep . (For calculations the pseudocubic SrRuO 3 lattice constant of a 0pc = 3.93 Å is used.) 24 It is evident that concomitantly to the increased Ru deficiency an increase in repetition rate f dep results in more tensile strain in the out-of-plane direction (Fig. 2(b)). The same trend of ruthenium non-stoichiometry with out-of-plane lattice parameters was observed in numerous previous studies. 12,21,25 Temperature dependent measurements of the electric resistivity in the range of 2 ≤ T /K ≤ 300 were performed and the data are shown in Fig. 3. At first sight the different resistivity curves ρ(T ) exhibit a similar shape over the whole temperature range, with only a shift towards higher resistivities for high frequencies being observed. This shift can be regarded as one indicator of the Ru vacancy concentration, as SrRuO 3 samples with the lowest Ru deficit usually possess the highest conductivity. 12 Slight differences in crystal quality between the samples might also influence the resistivity. 26 quality of thin films revealed, however, no significant differences between the samples. Therefore, another indicator of the Ru vacancy concentration-the Curie temperature T C -was investigated. T C is extracted from the position of the discontinuity in ∂ ρ/∂T . 28 As shown in Fig. 3(b), the ferromagnetic transition shifts down to lower temperatures as f dep increases. Again, this observation is in accordance with the assumption of a decrease in Ru content with increasing repetition rate of the different films. 12,21,[29][30][31][32][33] In brief, we can conclude that despite the drawbacks of the individual methods, measurements of the surface composition, structural and electrical properties of as-deposited SrRuO 3 thin films build a coherent picture that the increase in the laser repetition rate results in a decrease in Ru content in these films.
Atomic force micrographs reveal that all films considered in this study exhibit atomically flat morphologies similar to the vicinal step-structure of the SrTiO 3 substrate. Deposition at high frequencies, however, resulted in a more and more facetted step-structure than when depositing at low frequencies (compare Figs. 4(a) and 4(b)).
With an increase in deposition repetition rate f dep , a decrease in the thin-film's ruthenium content is observed (see Fig. 1(a)). Parameters known to result in systematic variations of the cation offstoichiometry, such as laser fluence, 9,34,35 deposition pressure, 11,36 and heater temperature, 10 can be discarded as origin of this decrease, as they remained constant. The terrace width of the vicinal SrTiO 3 substrate seems also not to affect the film's stoichiometry (see Fig. 1(b)). Moreover the thermodynamic instability of RuO x 13 species alone cannot result in the observed behaviour, since a faster deposition should kinetically suppress ruthenium loss and result in observation of the opposite tendencies.
For these reasons it is concluded that differences in the film's nucleation and growth processes on the substrate are decisive in leading to the differences in stoichiometry. This can only be the case if not entire "building-blocks" of SrRuO 3 , but also the individual components, SrO and RuO 2 , diffuse independently of the substrate surface. This view is supported by molecular dynamic studies on thin-film growth of SrTiO 3 37 confirming that different adatom species have different surface mobilities. In our case, at low repetition rates pure step-flow growth is dominant. This implies that the lifetime of adatoms is shorter than the time between two pulses and adatoms barely interact with each other on the terraces. Consequently the film is formed only at the step-edges, resulting in propagation of the step-edges with ongoing deposition. Due to the fact that adatoms barely interact with each other, the probability of volatile RuO y forming is low. The same holds for the probability of SrO adatoms forming above the critical size nuclei of SrO. This leads to films with a comparably smoothly edged step-structure and ideal stoichiometry (see Figs. 1 and 4).
With increasing repetition rate the amount of deposit reaching the substrate in a certain time interval is increased and so is the probability of island nucleation. 8 These islands are then engulfed by newly incoming species during the next pulse, which results in more facetted step-edges ( Fig. 4(b)); a similar phenomenon was observed upon variation of the substrate's vicinal angle by Rijnders. 20 The differences in stoichiometry observed here have to stem from this increased random interaction of RuO 2 and SrO adatom species. As the probability of adatom interaction increases with repetition rate, the probability of volatile RuO y species forming is enhanced. The most likely reaction leading to Ru loss is the disproportionation of RuO 2 , 38 2RuO 2 (s) → Ru(s) + RuO 4 (g). (1) The emergent elemental Ru will be immediately reoxidised, as oxygen is provided in superfluous amounts by the oxygen background gas. Alternatively, oxygen may come from the substrate, which is oxidised separately in the subsequent annealing step. 39 In a similar fashion as volatile ruthenium species are formed, the thermodynamically stable nuclei of SrO constitute and remain on the steps (cf. Fig. 1). This does not mean that at high frequencies, SrO and stoichiometric SrRuO 3 are deposited. The bulk part of deposit arriving on the substrate will still form SrRuO 3 , only with a much larger amount of Ru vacancies as indicated by our structural and electrical investigation. Both proposed deposition mechanisms are contrasted in a simplifying schematic fashion in Figs. 4(a ) and 4(b ). The observation of two growth mechanisms does not necessarily mean that the vicinal step-structure is no longer preserved. As the AFM micrographs in Fig. 4 show, at first glance films still grow in a mode that could pass as step-flow. 20 Furthermore the results indicate that there is no abrupt change in film's growth characteristics. It is rather a smooth transition between "pure" step-flow and "pure" island growth that is accompanied by a similarly smooth enhancement of the Ru off-stoichiometry.
Commonly, the transition between the two growth regimes is marked by the lifetime of adatoms on the step-edges τ life becoming comparable to the time between two laser pulses (1/f dep ), τ life ≈ 1/f dep . At the highest laser repetition rate used, the transition point between the two regimes is approached closest. This allows for a rough estimate of the apparent adatom surface diffusion coefficient D * ad according to D * ad ≈ L step /2 f dep , with L step being the length of a step-edge. 8, 40 Taking L step = 350 nm and f dep = 18 Hz results in a value for the adatom diffusion coefficient of D * ad ≈ 1.4 × 10 6 nm 2 /s. Considering that not exclusively adatoms, but also small clusters thereof diffuse on the step edges during film growth, the estimated value for D * ad only represents a lower limit of the adatom diffusivity. Nonetheless, this rough estimate agrees remarkably well with what can be extrapolated from a more dedicated study 8 of adatom diffusion kinetics for the here used temperature, D * ad ≈ 2.1 × 10 6 nm 2 /s. The fact that SrRuO 3 films on SrTiO 3 have been grown with excellent stoichiometric properties at temperatures up to T dep = 1273 K (pO 2 ≈ 0.1 mbar) 9,10 provides further support for the hypotheses that the kinetics of stable 41 RuO 2 adatoms being a crucial parameter: If thermodynamically unstable RuO 4 /RuO 3 13 adatoms are significantly contributed to the growth of SrRuO 3 , the films should exhibit a far higher Ru deficiency. All in all, the common wisdom that lowering the deposition temperatures results in diminished Ru loss thus seems of questionable help when trying to avoid Ru deficiency. In fact, it has the opposite effect. This is because RuO 2 remains longer on the terraces so that the probability of adatom interaction (and thus the reaction Eq. 1 to happen) is increased.
In this contribution, we showed that a variation of the repetition rate in the PLD process is sufficient to systematically vary the cation stoichiometry and concomitantly structural, electrical, and magnetic properties of SrRuO 3 thin-films. The main reason for the observed Ru deficit was identified to be a change in the adatom behavior with an increasing pulse repetition rate. More