Plasmas and Acoustic Waves to Pattern the Nanostructure and Chemistry of Thin Films

Acta Materialia 255 (2023) 119058 Available online 1 June 2023 1359-6454/© 2023 The Authors. Published by Elsevier Ltd on behalf of Acta Materialia Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Plasmas and acoustic waves to pattern the nanostructure and chemistry of thin films V. Rico a , 1 , G.F. Regod´ on a , 2 , A. Garcia-Valenzuela a , 1 , A.M. Alcaide a , M. Oliva-Ramirez a , b , T.C. Rojas a , R. Alvarez a , c , F.J. Palomares d , A. Palmero a , * , A.R. Gonzalez-Elipe a , * a Instituto de Ciencia de Materiales de Sevilla (CSIC-Univ. Sevilla), Avda. Am´ erico Vespucio 49, Sevilla 41092, Spain b Departamento de Física At´ omica, Molecular y Nuclear, Universidad de Sevilla, Avda. Reina Mercedes, Sevilla 41012, Spain c Departamento de Física Aplicada I, Escuela Polit´ ecnica Superior, Universidad de Sevilla, c/Virgen de Africa 7, Sevilla 41011, Spain d Instituto de Ciencia de Materiales de Madrid (ICMM-CSIC). C/ Sor Juana In´ es de la Cruz 3, Campus de Cantoblanco, Madrid 28049 Spain ARTICLE INFO Keywords: Plasma-acoustic waves interaction Thin film patterning Nanostructure and chemistry Magnetron sputtering at oblique angle SiO 2 and SiO x thin films ABSTRACT In this work, piezoelectric AWs and plasmas have been brought together during the growth of a thin film as a novel methodology of plasma-assisted thin film structuration. The ensuing effects have been investigated on a model system where SiO 2 and SiO x (x<2) thin films have been deposited by magnetron sputtering at oblique angles (MS-OAD) on an electro-acoustically excited LiNbO 3 piezoelectric substrate under resonant conditions. The microstructure of the resulting films was 2D patterned and depicted submillimeter size intermingled zones with different optical characteristics, compositions (SiO 2 and SiO x ) and porosity, from highly porous to dense and compact regions. The 2D nanostructural pattern mimics the AW distribution and has been accounted for by means of a specific simulation model. It is concluded that the morphological and chemical film pattern replicates the distribution of polarization potential on the surface of the AW activated substrate immersed in the plasma. Moreover, we show that the main mechanism responsible for the appearance of domains with different morphology and chemical composition is the focused impingement of Ar + plasma ions on certain regions of the substrate. The general character of this patterning process, the underlying physics and its possibilities to tailor the composition and microstructure of dielectric thin film materials are discussed. 1. Introduction This work discloses the effect of activating the deposition of plasma thin films on piezoelectric substrates where acoustic waves (AWs) have been electroacoustically generated during growth. Separately, plasmas and AWs are pervasive in many technological fields and used for numerous applications. For instance, AWs and surface acoustic waves (SAWs) are employed for the analysis and handling of liquids and suspensions [1–3], cell and protein separation and sensing [4,5], the activation of different processes in materials and metamaterials [6–8], the patterning of bulk meso‑ and micro-structures [9] or for surface testing characterization [10]. A recent roadmap review on SAWs [11] reports other applications in a large variety of domains. On the other hand, plasmas constitute a ubiquitous tool for the synthesis and processing of materials, particularly for the growth of thin films [12]. Well-known techniques such as plasma enhanced chemical vapor deposition [13], plasma assisted atomic layer deposition [14] or magnetron sputtering (MS) [15,16] are extensively utilized both for scientific and industrial manufacturing applications [17,18]. However, merging AWs generated in piezoelectric materials and external plasmas for the deposition of thin films remains a practically unexplored topic, except for an early but not continued work about nickel deposition on high frequency SAW activated substrates [19] and a recent paper on the microstructure of TiO 2 thin films [20] that were magnetron sputtered at oblique angles (MS-OAD) [21–23]. However, neither a general explanation nor a critical prospect for applications have been explored so far in relation with the plasma thin film deposition onto AW activated substrates, even if control of composition, adjustment of stress or the modification of the nanostructure are still challenging questions for many plasma thin film manufacturing processes [24]. * Corresponding authors. E-mail addresses: [email protected] (A. Palmero), [email protected] (A.R. Gonzalez-Elipe). 1 These authors contributed equally to this work. 2 Present address: Departamento de Física, Universidad de C´ ordoba, C´ ordoba E-14,071, Spain. Contents lists available at ScienceDirect Acta Materialia journal homepage: www.elsevier.com/locate/actamat https://doi.org/10.1016/j.actamat.2023.119058 Received 26 January 2023; Received in revised form 29 May 2023; Accepted 31 May 2023 Acta Materialia 255 (2023) 119058 2 The scenario resulting from merging plasmas and AWs is complex and involves different physical mechanisms and interactions: i) the nanometer length mechanical oscillations of lattice atoms of the piezoelectric substrate due to its electroacoustical excitation; ii) an oscillating local polarization of the piezoelectric substrate accompanying the mechanical oscillations due to the direct piezoelectric effect; iii) the interaction of this substrate with a plasma and the formation of a modified plasma sheath over its surface; and iv) the activation of the adatoms giving rise to the formation of the thin film. This work focuses on the yet unexplored processes iii)-iv). Firstly, it demonstrates the formation in the film of a 2D pattern formed by submillimeter domains with different microstructure and chemical characteristics. Secondly, it describes the microstructure and chemistry found in these domains and correlates the differences with the effects of the interaction of the AW activated substrate with the surrounding plasma. The experiment selected to investigate these issues has consisted of the MS-OAD of SiO x thin films (x varying from 1.4 to 2 [23,25]) on a LiNbO 3 piezoelectric substrate where standing AWs were generated. In addition, the found phenomenology has been analyzed in the frame of a computer simulation model describing the AW-plasma interaction. We conclude that merging AW activation and plasma deposition offers a tunable methodology to control the nanostructure and chemistry of thin films at various length scales and hence opens the way to the development of alternative patterning procedures and new plasma thin film manufacturing methodologies. Other benefits in terms of stress release or the achievement of alternative thin film compositions at ambient temperatures are other envisaged possibilities of this combination of plasmas and AWs during thin film manufacturing. 1.1. Plasma-AW interaction: concepts and thin film patterning hypothesis To better understand the processes iii-iv outlined above a schematic description of the experiment is presented in Fig. 1. Although the physical phenomena investigated in this work would be of application for any kind of plasma thin films, the selected experiment consists of a MS-OAD deposition, a procedure that enables a direct, even visual, assessment of the thin film patterning. According to Fig. 1a, the MS deposition is carried out at an angle of 85◦between target and sample surface. In the absence of AW activation, this geometry gives rise to 2D homogeneous thin films formed by tilted nanocolumns that depict a high porosity [21]. This nanostructure is ideal to demonstrate the existence of localized densification phenomena during AW activation, although similar energizing processes are expected to occur for conventional normal deposition geometries where patterning would not be so easily visualized. Adjusting the oxygen pressure in the plasma enables the control of the SiOx stoichiometry (x ≤2) [23,25]. As illustrated by the inset, under the utilized deposition conditions a continuous and homogeneous plasma sheath forms all over the substrate [26]. Fig. 1b shows that the electroacoustic activation of the piezoelectric substrate induces standing AWs of the Lamb type. Meanwhile, Fig. 1c shows that during the experiment the AW activated piezoelectric substrate is placed in the plasma generated around the MS target. This ensures the development of a plasma sheath on the substrate surface as it happened without AW excitation (Fig. 1a). Main hypothesis of our model is that the plasma sheath is locally altered due to the effect of the polarization potential distribution generated at the electroacoustically excited piezoelectric substrate. To prove this hypothesis, we simulate the mechanical oscillations and variations Fig. 1. Schematic description of experiment and working hypothesis. (a) MS-OAD of a thin film on a flat substrate without AW activation. The obtained film is nanocolumnar and homogeneous along the whole substrate surface. The formed plasma sheath is homogeneous and Ar + ion impingement is smoothly distributed onto the whole substrate (inset). (b) Standing AWs generated in a flat piezoelectric substrate by electroacoustic excitation. The sinusoidal wave is a simplification of the complex plate wave generated in the piezoelectric substrate. (c) MSOAD of a thin film upon AW activation of the substrate. Nanocolumnar and compact domains with a different stoichiometry are obtained in this case. The plasma sheath is not homogeneous and Ar + ions become focused on specific domains of the substrate (inset). V. Rico et al. Acta Materialia 255 (2023) 119058 3 of the surface potential on the AW activated LiNbO 3 substrate and analyze how the appearance of a polarization pattern affects the plasma sheath. Following the simulation results in Section 3.3 we show that plasma sheath is not homogeneous upon AW interaction and that the Ar + plasma ions trajectories through the plasma sheath become focused on certain locations of the substrate (see inset in Fig. 1c). As a result, the deposited thin film becomes patterned both in density and composition, depicting submillimeter size domains characterized by, respectively, porous nanocolumnar and very dense microstructures. The oxidation degree also differs for each type of domain. It will be also shown that pattern distribution and domain size replicate the characteristics of the standing AWs generated in the piezoelectric substrate. 2. Experimental and model calculations 2.1. Thin film deposition and AW activation SiO 2 and SiO x thin films have been prepared by reactive MS in an OAD configuration [21,22], at an angle of 85◦between the target and substrate surface. A 3′′ target of Si located at approximately 7 cm from the substrate ensures the immersion of this latter within the plasma generated around the MS target. Ar (9.3 sccm) with some addition of oxygen was used as plasma gas. To control the stoichiometry of the SiO x thin films (i.e. x value from 1.4 to 2) the oxygen flow was adjusted between 1.25 and 3.5 sccm. The selected deposition conditions provided a growing rate of 12 nm/min in the OAD configuration without AW activation. Average film thicknesses between 200 and 300 nm have been selected for the experiments. A more detailed description of the experimental deposition conditions of SiO 2 and SiO x thin films is reported as supplementary material S1. It is noteworthy that the plasma gas (Ar+O 2 ) pressure was 0.15 Pa to ensure that sputtered atoms do not become randomized in their trajectory from target to substrate, a requirement to preserve the nanocolumnar microstructure of OAD thin films [21,22]. This value also permits to keep the condition ωτ <1 ( ω , frequency of the AC signal applied to the piezoelectric substrate, and τ the ion transit time through the sheath, a parameter dependent on the gas pressure), a requirement to ensure that the plasma sheath potential on the AW activated substrate follows the variable electric field polarization induced at its surface [26]. The relevance of this boundary condition will be further discussed in the Discussion section. The piezoelectric substrate was a plate (1 cm ×2 cm ×0.5 mm) of double-side polished LiNbO 3 single crystal wafer (128◦Y-cut) exposing the 1 1 1 2 crystallographic plane at the surface (Roditi International Corporation Ltd). Two metallic films or plates of 0.5 cm width and 1 cm length on the back side of the plate served as electrical contacts for the AW activation. Thin films prepared onto LiNbO 3 under AW activation will be designated as SiO x -AW (with x =1.4, 1.7 and 2). Reference thin films were deposited simultaneously on quartz and silicon substrates placed besides the AW activated piezoelectric plate. The sample holder utilized for these experiments has been described in Ref. [20]. Its design ensures a negligible damping of the stationary AWs generated in the piezoelectric plate substrate (plate AWs of Lamb type) and permits to adjust the deposition angle with a goniometer. A scheme of electronic connections and devices required for AW activation is reported as supplementary material S2. The piezoelectric plates were excited with a RF signal of 100 V peak to peak at the resonance frequency of the device previously determined recording a return loss spectrum (i.e., S 11 parameter) with a network vector analyzer (NVA). Resonance was found in the range between 3.4 and 3.5 MHz (see supplementary material S3) . 2.2. Thin film analysis Optical Reflectometry analysis was carried out with a focused light beam spectrometer (Bruker GmbH) in the range of 450–800 nm. Measurements with a focused 200 μ m spot were obtained moving the sample in x and y directions by ca. 200 μ m steps. For the SiO 2 thin films, refraction indices of examined areas were determined by Cauchy analysis of the reflectance spectra. Optical and topographic images of the surface of patterned films were recorded with a confocal interferometric microscope (Sensofar S neox), providing optical images with 10 ×10 mm 2 dimensions and interferometric surface profiles over an area of 1637 ×1320 μ m 2 . The microstructure of the films was determined by SEM and STEM (i. e., scanning and scanning transmission electron microscopies). Top view SEM images were taken in a Hitachi 4800 field emission instrument working at 2 keV. STEM combined with EDX/EELS (Energy dispersive X- ray analysis/Electron energy loss spectroscopy) has been used for the determination of the chemical state of silicon along lamellas taken from the dense and porous domains developed in the films. Experimental details are reported as supplementary material S4. The analysis of elemental composition and chemical state at the films surface has been carried out by X-ray photoelectron spectroscopy (XPS) with lateral resolution in a spectrometer equipped with a hemispherical electron energy analyzer (SPECS Phoibos 150 spectrometer), a 2D delayline detector (Surface Concept 2D-DLD) and Mg-K α radiation (1253.6 eV). Additional details about data acquisition and handling can be found in the supplementary material S5. 2.3. Model calculations A numerical model has been developed to describe the surface interactions developed at the AW activated substrate during plasma thin film deposition and growth. The model is divided into three coupled modules, which we briefly describe next. 2.3.1. Module 1 (M1). Excitation and propagation of the electro-acoustic wave in the piezoelectric material in vacuum Following the same approach as in Ref. [20] we have analyzed the propagation of the acoustic wave field on the LiNbO 3 substrate using the software COMSOL Multiphysics 6.0 (see additional details as supplementary material S6). In this way we determine the electric potential, the amplitude and the phase of the oscillation, as well as the surface charge distribution on the piezoelectric surface. The same voltage, geometry and electrode arrangement used for the experiments have been employed for the simulations. In M1, the system is considered to remain in a vacuum environment, i.e., in the absence of plasma. Simulation results were analyzed as a function of the frequency of the AC excitation signal applied to the substrate within the range [3.45 MHz,3.55 MHz]to find the frequency of resonance. The simulation of the propagation of the standing AW reproduced well the experimental pattern of domains characterized by a size similar to the wavelength of the AW, λ≅1mm, as determined in a previous work by means of a Doppler vibrometer [20]. 2.3.2. Module 2 (M2). Electric potential on the AW activated piezoelectric plate immersed in a plasma In M2 we consider how the presence of a plasma may affect the electric potential distribution on the piezoelectric surface obtained in M1. It is well-known that any surface immersed in a plasma receives the flux of fast plasma electrons, become charged and gives rise to the formation of the so-called plasma sheath [27]. In general, the particular value of the flux of plasma electrons towards a surface depends on the features of the plasma (electron temperature and density) and on the value of the electric potential of the surface. In our case, however, the electric potential on the piezoelectric surface oscillates according to the AW and, hence, we assume that the flux of plasma electrons reaching the surface may spatially change according to the value and distribution of the piezoelectric surface potential. To determine the time and spatial evolution of surface charge we use the following simplifications: i) the V. Rico et al. Acta Materialia 255 (2023) 119058 4 flux of plasma electrons instantly responds to the local variations of surface potential in space and time, and ii) the flux of plasma electrons at each location on the piezoelectric surface is only determined by the value of the electric potential at the same location. The first approximation implies that the time required for the plasma electrons to form the plasma sheath is much smaller than the period of the AW. The second approximation assumes that the electron flux at a given location over the surface only responds to the local value of the electric potential, i.e., that any electric field that might appear in the plane parallel to the piezoelectric surface is negligible. This means that the size of spatial variations of the potential on the piezoelectric surface must be long enough to be screened by the plasma or, similarly, that the Debye Length λD, which is the scale of the plasma sheath, must be much smaller than the AW wavelength (λ). In our case, this approximation holds because a Debye Length in the order of λD∼100 μ m can be estimated for the plasma in our experiment (i.e., electron densities around 10 16 cm −3 are Fig. 2. Patterning of a SiO 2 -AW thin film. (a) Confocal image (left) and 3D enlarged reproduction of a zone with hill features to highlight the surface topography of the films (right). The white marks are located on the back of the substrate and are used for signaling. (b) Series of top view SEM micrographs taken along a line connecting domains 1–2 of different aspect when examined at low SEM magnification scales. The distance from the origin of the arrow in domain 1 is indicated in the micrographs. (c) Optical photograph taken under conditions that maximize reflectivity (left), and small area reflectance spectra taken at points A, B and C (right). The squared stains in (c) corresponds to marks at the back of the (transparent) substrate. V. Rico et al. Acta Materialia 255 (2023) 119058 5 taken as typical for MS-pulsed DC), while λ∼1mm. Relying on these approximations we determine that plasma electrons accumulate preferentially on given zones of the surface of the piezoelectric substrate and that the electric potential distribution on the piezoelectric surface changes with respect to the calculated profile in vacuum. 2.3.3. Module 3(M3).- Impingement of positive plasma ions on the piezoelectric surface Once calculated the electric potential distribution on the piezoelectric surface in the presence of the plasma, in M3 we analyze how the positive plasma ions move through the plasma sheath and impinge on the substrate. A simplifying assumption has been adopted for the determination of plasma sheath potential, V(x,y,z;t), where x and y refer to the cartesian spatial directions in the plane of the piezoelectric surface, z to the perpendicular coordinate and t is time. If we dub Vp(x,y;t) the value of the electric potential on the piezoelectric surface in contact with the plasma (calculated in M2), we assume a typical Child-Langmuir behavior within the plasma sheath [26–29] and calculate V(x,y,z;t) according to: V(x,y,z;t) ∼ Vp(x,y;t)(ls−z ls)4/3 (1) where ls is the width of the plasma sheath, estimated as in Ref. [29]. Given a value of V(x,y,z;t), the trajectories of the different plasma ions are calculated by Monte Carlo simulation: each ion is initially placed at the onset of the plasma sheath (z=ls) at a random position (x,y)and time. A collisionless trajectory of the ion in the plasma sheath is then calculated using MATLAB until it reaches the piezoelectric surface (z=0)(see supplementary material S6 for details). The impact position as well as the ion energy are then stored for statistical purposes, after which the said ion is removed from the simulation and the process is repeated, up to a total number of 4192,304 ions. 3. Results 3.1. 2D patterning of SiO 2 -AW and SiOx-AW thin films A first experiment was carried out setting the MS deposition conditions to ensure the fabrication of thin films with composition SiO 2 . Fig. 2 shows a confocal microscopy image and the analysis of the surface topography of the film (Fig. 2a). A series of top view SEM micrographs taken along a segment connecting two adjacent domains (zones 1–2) with different visual aspect is included in Fig. 2b, while an optical image of the film surface obtained by tilting the sample to maximize contrast between domains is presented in Fig. 2c-left. Reflection spectra with a 200 µm beam resolution taken at regions A, B and C in the figure are also included in this panel (Fig. 2c)-right). Clearly, the optical (Fig. 2c-left) and confocal (Fig. 2a) images of the sample surface reveal the formation of aligned, well-contrasted, patches of different aspect, indicating that the optical properties (e.g. reflectivity) differ from one zone to the next. A distinct optical behavior is effectively confirmed by the reflectance spectra taken at adjacent zones (Fig. 2c)-right). These spectra depict interference oscillations of different periods and amplitudes as expected for transparent thin films of different thicknesses and refraction indexes. In concrete, the oscillations of spectrum B are characterized by a longer wavelength and a smaller amplitude than those of spectra A and C. According to effective medium theories, these spectral shapes indicate that the thin film is thicker but less dense and has a smaller refraction index in B [30]. A fitting analysis of these spectra is reported as supplementary material S7. Values ranging from 1.355 to 1.445 have been calculated for, respectively, the low and high refraction index domains. The 3D confocal amplified image reported in Fig. 2a-right shows that sample SiO 2 -AW is not flat but depicts a patterned 2D topography characterized by 0.5–1 mm big domains differing as much as 50 nm in height (note that the average film thickness determined on a reference silicon substrate is around 300 nm). The different microstructural characteristics of these zones are clearly appreciated in the set of normal SEM images presented in Fig. 2b showing an evolution from a flat surface termination typical of a compact film in domain 1 to the characteristic tip termination of the nanocolumnar microstructure of OAD thin films in domain 2. This lateral variation in microstructure contrasts with the homogeneous topography depicted by MS-OAD SiO 2 thin films deposited without AW activation. According to the top view and cross section SEM micrographs in the left panel in Fig. 3, these films depict the typical porous nanocolumnar microstructure of SiO 2 thin films deposited at oblique angles [23,25]. This nanocolumnar microstructure [21, 22] has been attributed to the shadowing effects that interfere the trajectories of silicon atoms arriving to the substrate surface under OAD conditions (note that for reactive MS conditions, species such as SiO may also reach the substrate [31]). We assume that the 2D patterned microstructure of the SiO 2 -AW thin films is linked to the AW excitation of the substrate: it appears that in certain domains the growing film experiences very effective densification processes capable to smear out the typical nanocolumnar structure of OAD thin films. We will show below that this densification pattern replicates the profile of the standing AWs formed on the piezoelectric substrate. Adjusting the oxygen amount in the plasma gas and the deposition angle [23] during reactive MS-OAD provides a way to prepare silicon suboxides thin films (SiO x , x ≤2, x referring to the O/Si ratio). Herein, we have selected OAD deposition conditions to obtain SiO 1.7 and SiO 1.4 thin films in the absence of AW substrate excitation. The x value has been determined by XPS and RBS (Rutherford Back Scattering) following a strategy already reported for this type of thin films [23] (see RBS spectra of SiO 2 and SiO 1.4 thin films in the supplementary material S8). The purpose of these experiments is to determine whether the 2D domains obtained under AW activation possess similar compositions or not. A first observation in this regard was that SiO x thin films deposited onto LiNbO 3 or fused silica substrates without AW activation depicted a homogenous brownish aspect with a distinct darkening intensity depending on film stoichiometry (the smaller the x value the darker the sample). A thorough analysis of the optical and electronic properties of this type of SiO x thin films can be found in previous publications [23, 32]. In addition, these films also depicted the typical nanocolumnar microstructure of OAD thin films (see middle and right panels in Fig. 3). Unlike the homogeneous distribution of nanostructure and chemical state found without AW activation, a thorough characterization of sample SiO 1.7 -AW revealed the formation of a 2D pattern, both in nanostructure and chemistry. Thus, the topography analysis in Fig. 4a reveals the formation of domains with different heights, while the SEM analysis of surface nanostructure in Fig. 4b confirms an evolution of surface termination from flat to tip terminated, similar to that depicted for sample SiO 2 -AW in Fig. 2. Very significantly, while the SiO 1.7 sample was homogeneously brown, the SiO 1.7 -AW film depicts brown and colorless domains (Fig. 4c)-left). Moreover, the small area reflectance spectra taken at zones with different visual aspect confirm a clear difference in optical properties for each type of domains. Thus, the optical reflection spectra of these thin films taken at points A and B in Fig. 4c- right show a clear evolution from spectra with rather intense amplitude oscillations in the brown zone to less intense and smaller period oscillations in the clearer zones. This distinct behavior confirms the formation of differently densified domains, with lower and higher refraction indexes at transparent and brownish zones, respectively (note that the real part of refraction index of colored thin films is higher than for transparent thin films of similar density). A similar patterned growth also occurred for SiO x -AW thin films with other compositions, as evidenced by the analysis of sample SiO 1.4 -AW reported in Fig. 5. Results in this figure confirmed that similar 2D patterning phenomena affected nanostructure and chemistry, although V. Rico et al. Acta Materialia 255 (2023) 119058 6 full oxidation up to complete transparency of the colorless /compact zones was not achieved in this case. 3.2. 2D compositional patterning in samples SiO x -AW The previous analysis of microstructure and optical properties of SiO x -AW thin films prove that they are heterogeneous both in microstructure and optical behavior. We have assumed that in the dark brown domains there is a SiO x stoichiometry with x<2, while the transparent colorless domains approach a SiO 2 composition. To confirm this assumption, we have carried out a thorough characterization analysis of samples SiO x -AW by laterally resolved XPS and TEM-EELS. Fig. 6 shows selected laterally resolved XPS spectra of sample SiO 1.7 - AW, obtained moving the observation slit in a direction perpendicular to the color distributed fringes of this sample (c.f. Fig. 6a). The intensity maps in Fig. 6b and c correspond to the Si 2p and O 1 s spectra averaged over an area of 1500 µm x 100 µm along a line scan of 4 mm. For the sample positions X1 and X2 marked as red and blue squares in Fig. 6a, inphase maximum/minimum intensity values are clearly observed for the intensities of the O1s and the Si 4+ component of the Si 2p spectra. This lateral evolution confirms the existence of areas with different contents of Si 4+ species and oxygen along the light/dark domains observed with bare eyes. To sustain this assessment, Fig. 6d shows the Si 2p averaged spectra taken for the red and blue slit areas. These spectra prove that Si 4+ species present a higher intensity at X2 (blue area) than at X1 (red area). A detailed fitting analysis of these spectra with chemically shifted components ascribed to the Si 4+ , Si 3+ , Si 2+ , Si 1+ and Si◦chemical states of silicon existing in SiO x [21,31–34] confirms this assessment. Repeating this analysis for spectra acquired for slit positions moved as indicated by the arrow in Fig. 6a, renders the line scan profiles in Fig. 6e where Si 3+ , Si 2+ , Si 1+ and Si◦contributions are grouped into a single line scan denoted as Si 3+,2+,1+,0 . This line appears in phase with those corresponding to the Si 4+ component and O1s spectral area, confirming the modulation of domains composition according to a pattern with a periodicity similar to that of the surface topography of the films. By the previous analysis, it is noteworthy that XPS is only surface sensitive and that the outmost zones of the films might be extra oxidized due to their exposure to air. Another remark is that compositional domains (c.f., Fig. 6b and c) are not sharp and that spectral contributions from adjacent domains cannot be discarded. Additional XPS evidence of the changes in composition along the domains in sample SiO 1.4 -AW is provided as supplementary materials S9 and S10 showing a complete set of characterization results exploiting all possibilities of the laterally resolved XPS technique. To get a more precise assessment of the microstructure of each type of domain and to confirm possible differences in the oxidation states of silicon, we carried out a systematic analysis by TEM/EELS complemented with EDX to evaluate the O/Si ratio in the interior of the films. This study also aims to unambiguously associate the oxygen rich zones evidenced by XPS to either the compact or nanocolumnar porous domains existing in SiO x -AW samples. Fig. 7 shows two STEM micrographs taken for lamellas retrieved from the dense and porous-nanocolumnar domains of sample SiO 1.7 -AW. These cross sections confirm the porous-nanocolumnar character of the brownish domains developed in this sample and the compact nanostructure of the clearer zones. Another evidence from the STEM images is that the thickness of the film in the porous zones is bigger than in the compact zones, in good agreement with the height differences determined by confocal microscopy in Fig. 4a (according to Fig. 7a and b, 230 nm for the former against 140 for the latter; note that the compact lamella was bended during manipulation). The EDX estimation of the O/ Si ratio in these two zones rendered average values of 1.95 and 1.6 that, within experimental error, approach the nominal SiO 2 and a SiO 1.7 stoichiometries. The EELS spectra recorded along the line profiles defined by the marked numbers in Fig. 7a and b provide additional information on the chemical state of silicon in the compact and porous zones. Thus, all Si L 2,3 spectra in Fig. 7c, recorded along a perpendicular direction to the surface in the compact zone lamella, depict a shape similar to that of the Si L 2,3 spectrum of a SiO 2 reference sample [33,35]. This attribution is confirmed by the shape and loss energy measured at the maximum of the plasmon structure at around 24 eV, which is also characteristic of SiO 2 [35]. A similar STEM/EELS analysis carried out for a SiO 2 compact thin film prepared by MS confirmed this attribution (see supplementary material S10). We can therefore confirm that the transparent zones in SiO 1.7 -AW thin films present a stoichiometry equal or close to SiO 2 , a result agreeing with the reported laterally resolved XPS spectra in Fig. 7c. The situation changed when analyzing a lamella retrieved from a porous nanocolumnar zone. The Si L 2,3 spectra in Fig. 7d, taken along a Fig. 3. Top view and cross-section SEM views of SiO 2 , SiO 1.7 and SiO 1.4 thin films (as indicated) deposited onto a silicon wafer under the same conditions that for the films deposited onto a LiNbO 3 substrate under AW activation. These thin films are homogeneous over the whole substrate and depict the typical tilted nanocolumnar microstructure of MS-OAD thin films. V. Rico et al. Acta Materialia 255 (2023) 119058 7 horizontal line in the positions signaled with numbers 1 to 4 in Fig. 7b, show a series of spectra characterized by an intense band at around 99 eV (marked with a star in the figure) and a band shoulder extending up to 101 eV and superimposed onto the typical spectrum of SiO 2 . This band can be attributed to Si 0 [35], as confirmed by comparison with the reference spectrum included in the figure. Some contribution of the Si 3+ -Si + oxidation states of silicon to the experimental spectral shape is also expected in the energy region up to 101 eV (see for example a previous study about the stoichiometry of SiOx nanocolumns prepared without AW activation [37]). Therefore, we can conclude that in the porous lamella, besides Si 4+ species, there are considerable concentrations of Si 0 , and Si n+ (n<4) species. A similar assessment can be deduced from the observation of the evolution of the broad plasmon structure in Fig. 7d, whose badly defined maximum appears at around 20 eV (i.e., close to the position of the plasmon band of Si 0 included as a reference for comparison [36]). A mixture of Si n+ species is typical of SiO x (x<2) thin films prepared by reactive MS-OAD [23,34] without AW activation. The detection of a Fig. 4. Patterning of SiO 1.7 -AW thin film. (a) Confocal image (left) and 3D enlarged reproduction of a zone with surface hill features to highlight the surface topography (right). (b) Series of top view SEM micrographs taken in normal geometry along a line connecting two zones 1–2 of different aspect when examined at low SEM magnification scales. The distances from the start of the arrow in 1 are indicated in the micrographs. (c) Optical photograph (left) and small area reflectance spectra taken at points A and B (right). V. Rico et al. Acta Materialia 255 (2023) 119058 8 similar chemical state in the porous domains of SiO x -AW thin films indicates that the same state is roughly preserved in these regions, while an almost complete oxidation of the silicon species to Si 4+ takes place in the adjacent compact zones. 3.3. Patterning phenomena: simulation of plasma interaction with AW activated substrates According to the inset in Fig. 1c, an implicit assumption of the working hypothesis outline in Section 1.1 is that, rather than AW induced mechanical effects, the focusing and acceleration of Ar + ions from the plasma is the most important factor contributing to the patterning of deposited films [20]. The two insets in Fig. 1a and c show schematically how we think this happens without and with the AW activation of the substrate. Herein, we confront the hypothesis with the results of a simulation of the impingement of Ar + ions onto the AW activated piezoelectric substrates. As outlined in Section 2.3, the simulation model proceeds in three steps, corresponding to modules M1-M3. Fig. 5. Patterning of SiO 1.4 -AW thin film. (a) Confocal image (left) and 3D enlarged reproduction of a zone with surface hill features to highlight the surface topography (right). (b) Series of SEM micrographs taken in normal geometry along a line connecting two zones 1–2 of different aspect when examined at low SEM magnification scales. The distances from the start of the arrow in 1 are indicated in the micrographs. (c) Optical photograph (left) and small area reflectance spectra taken at points A and B (right). V. Rico et al. Acta Materialia 255 (2023) 119058 9 Fig. 8a presents the COMSOL simulation in M1 of the AW generated in the piezoelectric substrate at a given moment (a dynamic description of the can be found in a video available as supplementary material S12) in the absence of plasma. The LiNbO 3 lattice displacements in the z direction has been represented in an enlarged scale for a better visualization (amplification factor of 10 5 ), while the associated variation in the surface potential is represented in a color scale from blue (minimum potential) to red (maximum potential). This simulation agrees well with the experimental characterization of surface oscillations by Doppler vibrometry reported in Ref. [20]. From this analysis we estimate a wavelength λ of the generated AWs in the order of 1 mm. Moreover, according to Fig. 8a a surface potential oscillation in each domain of ±7 V was also deduced. In M2 we determine the effects of the interaction of the variable surface potential with the plasma electrons during the formation of the plasma sheath. According to classical plasma sheath concepts [26–29], we consider that the fast arrival of plasma electrons at the electro-acoustically excited piezoelectric surface is responsible for an additional modulation of the surface voltage, as illustrated by the simplified sinusoidal scheme in Fig. 8b (see the transformation of the voltage variation in vacuum to that in the plasma in supplementary material S13). In this idealized 1D representation, the surface potential oscillations are negligible with a value close to zero in some domains (we will denote these zones as low amplitude regions LAR) while it oscillates between ±7 V in the other (denoted here as high amplitude regions HARs). The distance between a LAR and a HAR as represented in this schematic corresponds to the λ of the standing sinusoidal wave prior to the interaction with the plasma (see Fig. S13). The bluish colored shadow in Fig. 8b represents the accumulation of negative charge due to plasma electron impingement during the formation of the plasma sheath. It shows that negative charge preferentially concentrates on the HARs, while the LARs are depleted of charge. Using the surface potential and charge distribution maps as reference, in M3 we simulate with a Monte Carlo method the trajectories of Fig. 6. Laterally resolved XPS analysis of the surface state of sample SiO 1.7 -AW. (a) Optical image of the investigated sample showing regions with alternate light and dark domains. Scheme of the line scan procedure used for the measurements indicating the displacement direction of the sample (white arrow) and the projection of the entrance slit utilized for the analysis. Red, blue and white squares displayed in the scheme represent individual locations of the projected slit area where Si 2p and O 1 s spectra are measured. (b) and (c) Color intensity maps of Si2p and O 1 s maxima measured along a line scan of 4 mm for observation steps of 250 µm. (d) Si2p fitted photoelectron spectra taken at positions represented by the red and blue areas in the scheme (a) and maps (b/c). (e) Normalized line scan profiles of the intensities of the Si 4+ component and O 1 s signal compared with that of the sum of the other Si 2p fitting components representing the Si 3+ , Si 2+ , Si 1+ and Si 0 oxidation states of silicon. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) V. Rico et al.