Structural control in porous/compact multilayer systems grown by magnetron sputtering

1 Structural Control in Porous/Compact Multilayer Systems Grown by Magnetron Sputtering A. Garcia-Valenzuela1, C. Lopez-Santos1, R. Alvarez1,2*, V. Rico1, J. Cotrino1,3, A.R. Gonzalez- Elipe1, A. Palmero1* 1 Instituto de Ciencia de Materiales de Sevilla (CSIC-US), Americo Vespucio, 49. 41092 Seville, Spain 2 Departamento de Física Aplicada I. Universidad de Sevilla. Virgen de África, 7. 41011 Seville, Spain 3 Departamento de Física Atómica, Molecular y Nuclear. Universidad de Sevilla. Avda. Reina Mercedes s/n. 41012 Seville, Spain Abstract.- In this work we analyze a phenomenon that takes place when growing magnetron sputtered porous/compact multilayer systems by alternating the oblique angle and the classical configuration geometries. We show that the compact layers develop numerous fissures rooted in the porous structures of the film below, in a phenomenon that amplifies when increasing the number of stacked layers. We demonstrate that these fissures emerge during growth due to the high roughness of the porous layers and the coarsening of a discontinuous interfacial region. To minimize this phenomenon, we have grown thin interlayers between porous and compact films under the impingement of energetic plasma ions, responsible for smoothing out the interfaces and inhibiting the formation of structural fissures. This method has been tested in practical situations for compact TiO2/porous SiO2 multilayer systems, although it can be extrapolated to other materials and conditions. Keywords: Oblique Angle Deposition; Magnetron Sputtering; Plasma-Assisted Deposition; Multilayers; TiO2; SiO2 * Email: [email protected]; alberto.palm[email protected] 2 I. Introduction Physical Vapor Deposition at oblique angles (OAD) is well known for promoting the growth of porous thin films with unique morphological features, e.g. high specific surface [1,2], tunable density [3,4] or high pore connectivity [5,6]. These features make these films ideal candidates for applications in optical devices (due to their higher transparency [7]), plasmonics [8], electrical anisotropy [9], sensors [10,11], electrodes for solar cells [12] or biomedicine [13], among others. From a microstructural point of view, these films are formed by numerous well-separated tilted nanocolumnar structures that originate due to the oblique incidence of deposition species on a substrate, i.e., under conditions that enhance surface shadowing mechanisms during growth [14,15]. Even though the OAD geometry has been known since the beginning of the 20th century [14], only recently has it been systematically explored by plasma-assisted magnetron sputtering (MS), widening its possibilities [14, 16, 17, 18]. Relevant controllable quantities in this case are the tilt angle of the substrate with respect to the target [19] and the plasma gas pressure [20], both of them determining the momentum distribution and angle of incidence of the deposition species on the substrate. In fact, it is known that only at low pressures and for tilt angles of the substrate above ~70º, a nanocolumnar microstructure may develop [19]. In all the aforementioned applications, the existence of a high void volume associated to the numerous pores open to the surface is essential for the good performance of OAD thin films. Yet, this type of porosity might introduce undesired collateral effects in multilayer systems, e.g., in photonic crystals or Bragg reflectors, where a strict periodicity between layers and sharp interfaces are required [5, 21]. In this case, the large intercolumnar voids that characterize these films render a relatively high surface roughness [22], that might incidentally affect the growth of any layer on top and, hence, the morphology and nanostructure of the whole multilayer system. In the literature, most strategies devoted to smooth out rough surfaces rely either on a postprocessing treatment of the material [23, 24] or the impingement of energetic species, namely ions, during the growth of the film [25, 26, 27]. While the former is not recommended for the synthesis of multilayer stacks (it is much more efficient if the deposition takes place in one single 3 run), the latter is well-known for promoting sputtering and mass transport processes that flatten the film surface [28]. Yet, ions may pose some drawbacks such as film densification and the decrease of the film porosity. For instance, in ref. [29], we analyzed the growth of SiO2 thin films by magnetron sputtering in the classical (non-oblique) configuration finding that, when the impingement rate of energetic negative oxygen ions was increased, the film morphology evolved from one characterized by sponge-like and vertically-aligned patterns to another characterized by an homogeneous and compact microstructure. Moreover, we reported a decrease in porosity of about one order of magnitude between these two situations, which would indeed represent a clear handicap in many applications relying on the porous nature of these films. To our knowledge, no in situ strategies have been yet specifically implemented to smooth out OAD thin films while maintaining most of its porosity, a procedure that would be of great relevance to achieve good structural control in porous multilayer structures. In this paper, we report a remarkable and undesired phenomenon that takes place when piling up porous and compact films in multilayer stacks grown by MS. In this type of systems, we have found that compact layers develop numerous structural fissures rooted in the intercolumnar voids of the porous layers below. We also demonstrate that this phenomenon amplifies for increasing number of layers, leading to the loss of structural control over the whole system. In order to inhibit the formation of these fissures, we have developed a simple strategy that relies on the introduction of thin and homogeneous interlayers grown under the impingement of plasma ions between the porous and compact films. Although the results presented in this paper have been primarily tested for porous SiO2/compact TiO2 multilayer systems at low temperatures, i.e., stacked structures that are typically employed to grow one dimensional photonic crystals, we also give some general insights to extrapolate our results to other cases and situations. 4 II. Experimental Setup Films were deposited in a cylindrical vacuum reactor (base pressure before deposition was below 5x10-4 Pa) equipped with two magnetron heads (Gencoa Ltd, Liverpool, UK) with silicon and titanium planar targets (diameter of 76.2 mm and 3 mm thick) as illustrated in Figure 1. Films were deposited onto 1x1 cm2 planar (100) silicon substrates mounted on a rotatable holder 7 cm and 11.5 cm apart from the silicon and titanium targets, respectively. Same plasma conditions explored in ref. [29] were implemented: a pulsed unipolar power supply (Pinnacle Plus, Advanced Energy®, USA) was employed, operated in a pulsed DC regime at a frequency of 80 KHz and a duty cycle of 40%, maintaining an electromagnetic power of 300 W. A mixture of Ar (purity 99.995 %) and O2 (purity 99.995 %) was used for the depositions, with an argon partial pressure of 0.2 Pa. The substrate tilt angle with respect to each target was set at 0° when growing TiO2 compact layers and 85º for the SiO2 porous films, respectively. Unless stated otherwise, the oxygen partial pressure to grow the SiO2 thin films was 0.02 Pa, i.e. same conditions under which we demonstrated that ion impingement from the plasma was not influencing the nanostructural development of the films [29]. For some particular experiments, an energy-resolved mass spectrometer (EQP 500, Hiden Analytical Ltd) was also placed at the location of the films to assess the amount of ions from the plasma impinging on the film, as well as their energy. The spectrometer had an orifice with a diameter of 0.05 mm, while the measurements were carried out in the energy range from 0 to 400 eV for TiO2 and SiO2 films. The surface roughness of the deposited layers was characterized with a Nanotec AFM microscope supplied with a Dulcinea electronics. The microscope was working in tapping mode by using high frequency levers. Au-covered SiN tips (Olympus) with a diameter of 40 nm were used. The resonance frequency of the tip was 20 kHz and the spring constant 0.09 Nm−1. The obtained surface maps were processed by means of the WSxM software [30]. Moreover, Field Emission Scanning Electron Microscopy (FESEM) pictures were recorded for each film using a Hitachi S4800 microsocope at the Instituto de Ciencia de Materiales de Sevilla (CSIC-US, Seville, Spain). 5 III. Results and Discussion Microstructural propagation phenomenon Figure 2a (left image) depicts a cross-sectional SEM image of a compact SiO2/compact TiO2 bilayer (monolayer thickness around ~300 nm), which was grown as a reference system. As expected, both layers show typical compact nanostructures and a rather sharp interface. Figure 2b (left image) shows the SEM image of a porous SiO2/compact TiO2 bilayer system with similar thickness. Remarkably, and even though the TiO2 thin films in both figures were grown under equivalent conditions, the reported images evidence clear morphological differences: while the bilayer structure in Figure 2a (left) is rather homogeneous, that in figure 2b (left) depicts numerous structural fissures. This difference is more evident in figures 2a-b (right images), which have been obtained after applying a border detection and contrast enhancement software to the images on the left. There, it is apparent the existence of structural fissures in the compact TiO2 film that root in the open porous structures of the SiO2 layer below. We dub this phenomenon microstructural propagation, an effect representing a serious shortcoming when piling up porous and compact layers in multistack structures. Indeed, a progressive loss of interface planarity and microstructural integrity are evidenced in the SEM micrograph in Figure 2c (left and right images) corresponding to a porous SiO2/compact TiO2 eleven-layers multistack structure. Formation of an interfacial accommodation interlayer To understand the origin of the microstructural propagation phenomenon we analyze the correlation between surface features of the SiO2 porous film and the structural fissures in the TiO2 compact layer. In figure 3a-b we show the top SEM images of compact and porous SiO2 films, respectively. Unlike the highly smooth surface of the compact film in figure 3a, the surface of the porous layer in figure 3b developed numerous mounds associated to the tips of the nanocolumns. In fact, AFM topographic profiles of these films (shown as insets in the images) indicate that the porous layer possesses much higher height variations than the compact one, a result that suggests that the origin of the structural fissures in TiO2 is linked to the existence of large height variations 6 on the surface it grows upon. To prove this connection, we have employed a well-known and tested growth model already presented in refs. [19, 31]. This model considers atomic shadowing as the main nanostructuration mechanism, in agreement with the Thornton’s Structure Zone Model [32, 33] that describes the competition between shadowing processes and thermally activated migrations. In figures 4a and 4b (left) we present the simulated compact SiO2/compact TiO2 and porous SiO2/compact TiO2 bilayer structures, which are in good agreement with the experimental images in figures 2a and 2b. Moreover, the series of snapshots presented in figure 4a (right) describing the first stages of growth of the compact TiO2 layer onto a compact SiO2 layer reveal the formation of a continuous interfacial accommodation layer that rapidly smooths out the underlying patterns and promotes a compact homogeneous growth. Similarly, the set of snapshots in figure 4b (right) corresponding the growth of the compact TiO2 film onto the porous SiO2 layer shows that the deposition does not only occur on top of the nanocolumns but also inside the voids (pores), just below the average surface level. As a result of this partial conformal growth, the interfacial region presents clear discontinuities at the pore entrances. According to these simulations it appears that the coarsening of this discontinuous interfacial region generates numerous compact domains whose borders can be interpreted as structural fissures. Growth of a continuous accommodation interlayer on porous films Based on the analysis above, it seems that a requirement to preclude the microstructural propagation phenomenon is the growth of a continuous interfacial region to promote the development of a uniform TiO2 film on top. From an atomistic point of view, this cannot be accomplished whenever the film nanostructuration is controlled by surface shadowing processes: only by inducing atomic migration phenomena could the isolated domains formed at the entrances of the pores be connected. A first possibility to trigger these processes relies on thermally activated mechanisms (e.g., by heating up the sample during growth [34, 35]). However, the Thornton’s Structure Zone Model indicates that the condition Tf /Tm > 0.3 must be fulfilled to ensure the dominance of surface diffusion, where Tf is the film temperature during growth and Tm 7 the melting temperature of the deposited material. For TiO2, this condition implies Tf >700 K, a rather high value that would restrain its practical implementation for many applications. An alternative to thermal activation in our conditions is the introduction of ion-induced mobility processes [25-27]. In fact, under MS conditions, ions from the magnetron plasma might play a critical role provided that they transfer enough energy to the growing film [36-38]. It has been mentioned in the introduction that under non-oblique (classical) conditions and at high oxygen fluxes the growth of SiO2 thin films by MS is strongly influenced by the impingement of energetic negative oxygen ions. These oxygen ions form at the silicon target surface and are accelerated towards the film with energies in the order of few hundred eVs, without applying any substrate bias [29]). Figure 5 shows the energy spectrum of negative oxygen ions measured by mass spectrometry when growing SiO2 films in the classical configuration with a high oxygen partial pressure (0.25 Pa, i.e. under conditions where negative ions were reported to effectively densify the film [29]). For comparison purposes, this figure includes the energy spectrum of the negative oxygen ions generated when growing a compact TiO2 thin film. In agreement with ref. [29], a high energy peak at around 300 eV was found in the former case, whereas in the latter no trace of high energy negative oxygen ions was detected (the peaks at lower energies in the two cases correspond to ions at the plasma potential). A hypothesis of the present work is that a thin and compact SiO2 interlayer can be grown under the impingement of negative oxygen ions on top of the SiO2 porous surface, and that this interlayer promotes a continuous interfacial region when growing the compact TiO2 film. Keeping this purpose in mind, we have deposited several compact SiO2 layers with different thicknesses on top of a porous SiO2 layer. The cross-sectional images of these films in Figure 6a-e show no trace of structural fissures, suggesting that, effectively, ioninduced atomic mobility processes dominate over surface shadowing mechanisms, leading to the formation of a compact and continuous layer. Minimizing the microstructural propagation phenomenon among layers Once established the growth conditions of a continuous SiO2 layer, the next step is the determination of its minimum thickness to promote the growth of a homogeneous and compact 8 TiO2 layer on top. For this purpose, we have studied the evolution of the surface roughness of this layer as a function of its thickness. According to the Dynamic Scaling Theory [39-41], the growth of a film can be described by a specific power law of quantities such as surface roughness, r, correlation length or grain size with the film thickness, d. Herein, we will characterize the growth by the relation 𝑟 ∝ 𝑑𝛽, where β is the so-called growth exponent [42]. Figure 7 shows the evolution of surface roughness with thickness for a SiO2 porous layer (solid upper curve) deposited on a flat substrate. From these curves it is possible to derive 𝛽~0.8. Figure 7 also shows the evolution of the surface roughness when a SiO2 compact layer is deposited on top of a porous film with thickness ~280 nm. It appears that when the compact layer thickness increases the surface roughness decreases until reaching a total thickness of 350 nm, a point from which thickness increases again but now according to a different power law characterized by 𝛽~0.6. Remarkably, and according to ref. [43], this latter value corresponds to a compact SiO2 growth, which implies that under negative oxygen ion bombardment an interfacial layer thickness of ~25% of the porous layer below (~70 nm/280 nm) suffices to recover a compact growth mode. As a final step in this investigation, we analyze whether the strategy developed above is effective to minimize the microstructural propagation phenomenon. The cross-sectional SEM image in figure 8a (left) of a porous SiO2/compact TiO2 bilayer with a continuous SiO2 interlayer clearly shows the absence of structural fissures. This is quite evident regarding the right column image, obtained after applying a border detection and contrast enhancement software. Finally, and as a final proof for the efficiency of the proposed strategy, in figure 8b (left) we show a multilayer structure similar to that depicted in figure 2c, but growing a ~30nm SiO2 continuous interlayer on top of each ~120nm porous SiO2 structure. There, it is clear that the microstructure of all monolayers is similar and that the structural propagation phenomenon is inhibited, confirming the validity of the proposed strategy. As a final remark, it is noteworthy that the proposed method relies on the growth of homogeneous and thin interlayers between porous and compact films by inducing atomic migration mechanisms. In our case, we have achieved this goal by promoting a relatively high energy ion impingement during the growth of a compact SiO2 interlayer. Yet, other approaches could also 9 be possible to achieve the same scope, e.g. heating up the film substrate during growth or electrically biasing the film to attract positive ions. Although we have not attempted these other approaches, they are, in principle, valid and could be implemented depending on the characteristics of the multilayer system. IV. Conclusions In this paper we have demonstrated the existence of an undesired effect, dubbed microstructural propagation phenomenon, which appears when porous layers grown at oblique angles are piled up with compact ones in multilayer systems. In these cases, the porous structure seems to promote the appearance of structural fissures throughout the compact films, in a process that becomes amplified when increasing the number of layers. Based on fundamental experiments and the results of a well-tested growth model, we have concluded that this phenomenon is caused by the existence of a discontinuous interfacial region that, upon coarsening, develops into numerous domains, whose borders can be appreciated as structural fissures in the compact layer. In order to minimize the propagation phenomenon, we propose depositing a continuous thin interfacial layer on top of each porous surface, previous to the deposition of the compact film. This strategy has been tested in a practical situation by depositing a periodic multilayer stack with porous SiO2/compact TiO2 bilayer structure achieving well-defined periodic interfaces and a good replication of the morphology from the first to the last layer. These results demonstrate the validity of the developed approach to achieve good morphological control on the multilayer structure. Acknowledgements Authors acknowledge the University of Seville (VPPI-US), the Junta de Andalucía (TEP8067 and TEP5283) and the Ministry of Economy and Competitiveness of Spain (MAT2013-42900-P, MAT2013-40852-R, MAT2016-79866-R, MAT2015-69035-REDC) for financial support. 16 Figure 2 17 Figure 3 18 Figure 4 19 Figure 5 20 Figure 6 21 Figure 7 22 Figure 8