Mechanisms of de-icing by surface rayleigh and plate lamb acoustic waves

Mechanisms of De-icing by Surface Rayleigh and Plate Lamb Acoustic Waves Shilpi Pandey,* Jaime del Moral,* Stefan Jacob, Laura Montes, Jorge Gil-Rostra, Alejandro Frechilla, Atefeh Karimzadeh, Victor J. Rico, Raul Kanter, Niklas Kandelin, Carmen López-Santos, Heli Koivuluoto, Luis Angurel, Andreas Winkler,* Ana Borrás, and Agustin R. González-Elipe 1. Introduction Ice accretion on materials operating outdoors or in sub-zero environments is a critical issue that significantly impacts efficiency, maintenance, and security in various industries, including aviation, energy generation, and the functionality of sensor and camera windows and screens. To address this problem, two primary approaches have been proposed: 1) Passive anti-icing based on applying surface engineering processes to reduce or delay ice accretion; [1–5] very often, these approaches rely on superhydrophobic or liquid-infused surfaces. [6–9] 2) Active de-icing/anti-icing systems applying different strategies for ice removal or preventing its formation; recent advances in the field include nanotechnology-based melting by Joule heating, [10] using materials such as graphene, [11–13] or photothermal de-icing using plasmonic or magnetic nanoparticles. [14–18] An emergent de-icing approach involves the activation of surfaces by high-frequency acoustic waves (AW) with Acoustic waves (AW) have recently emerged as an energy-efficient ice-removal procedure compatible with functional and industrial-relevant substrates. However, critical aspects at fundamental and experimental levels have yet to be disclosed to optimize their operational conditions. Identifying the processes and mechanisms by which different types of AWs induce de-icing are some of these issues. Herein, using model LiNbO 3 systems and two types of interdigitated transducers, the e-icing and anti-icing efficiencies and mechanisms driven by Rayleigh surface acoustic waves (R-SAW) and Lamb waves with 120 and 510 μm wavelengths, respectively, are analyzed. Through the experimental analysis of de-icing and active anti-icing processes and the finite element simulation of the AW generation, propagation, and interaction with small ice aggregates, it is disclosed that Lamb waves are more favorable than R-SAWs to induce de-icing and/or prevent the freezing of small ice droplets. Prospects for applications of this study are supported by proof of concept experiments, including de-icing in an icing wind tunnel, demonstrating that Lamb waves can efficiently remove ice layers covering large LN substrates. Results indicate that the de-icing mechanism may differ for Lamb waves or R-SAWs and that the wavelength must be considered as an important parameter for controlling the efficiency. S. Pandey, [+] S. Jacob, A. Karimzadeh, A. Winkler Leibniz IFW Dresden, SAWLab Saxony Institute for Emerging Electronic Technologies (IET), Group “Acoustic Microsystems” Helmholtzstr. 20, 01069 Dresden, Germany E-mail: [email protected]; [email protected] The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/adem.202401820. [+] Present address: Materials Science and Environmental Engineering, Faculty of Engineering and Natural Sciences, Tampere University, 589, 33104 Tampere, Finland © 2024 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH. This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. [Correction added on 17 December 2024, after first online publication: The 9th author name has been corrected in this version.] DOI: 10.1002/adem.202401820 S. Pandey, Heinz-Nixdorf-Chair of Biomedical Electronics School of Computation, Information and Technology Technical University of Munich TranslaTUM, 80333 Munich, Germany J. del Moral, L. Montes, J. Gil-Rostra, V. J. Rico, C. López-Santos, A. Borrás, A. R. González-Elipe Nanotechnology on Surfaces and Plasma Lab Materials Science Institute of Seville Consejo Superior de Investigaciones Científicas (CSIC) Americo Vespucio 49, 41092 Sevilla, Spain E-mail: [email protected]ic.es S. Jacob German National Metrology Institute (PTB) Bundesallee 100, 38106 Braunschweig, Germany A. Frechilla, L. Angurel Instituto de Nanociencia y Materiales de Aragón (INMA) CSIC-Universidad de Zaragoza 50018 Zaragoza, Spain RESEARCH ARTICLE www.aem-journal.com Adv. Eng. Mater. 2024, 2401820 2401820 (1 of 16) © 2024 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH nano-scale amplitudes, a procedure compatible with industrial-rel- evant surfaces. In this context, a key distinction should be made between the localized application to surfaces of commercially available ultrasound generators [19,20] and the innovative concepts based on the incorporation of piezoelectric active supports or layers, as well as electrodes (lateral field excitation (LFE) or interdigitated transducers (IDTs)), as integral parts of the surface to be activated. [21–26] The present work fits within this second approach, dealing with the integration of surface or bulk acoustic waves (SAWs and BAWs) for de-icing and ice sensing. Among the articles addressing this topic, the work of Yang et al. (2021), which proposes a strategy to weaken ice adhesion using Rayleigh surface acoustic waves (R-SAW), [21] is particularly noteworthy. For this purpose, IDTs with different wavelengths (100, 200, 300, and 400 μm) were manufactured on a 5 μmthick ZnO piezoelectric thin film deposited on an aluminum substrate, producing SAWs with frequencies comprised between 7.38 and 27.84 MHz. The authors state that SAWs produce vibrations comparable to “nano-scale earthquakes”,which result in the generation of acoustic-heating effects, the development of micro-cracks, and the disturbance of ice nucleation through local vibrations and energy streaming into the formed liquid phase. The contribution of Joule heating effects directly stemming from ohmic losses in the IDTs cannot be discarded in these experiments since ice droplets were placed directly atop the IDTs. The same authors have recently claimed the possibility of monitoring ice formation using this type of SAWs. [22] Similarly, Zeng et al. [23] demonstrated a decrease in adhesion between ice and substrate due to interface heating and a reduction in electrostatic forces and mechanical interlocking on SAW-activated substrates. Del Moral et al. [24] and Jacob et al. [25] have recently extended the application of acoustic wave de-icing to transparent substrates and supports. In the first work, the authors demonstrated that extended electrodes on piezoelectric plates in an LFE configuration generate thickness shear mode - bulk acoustic waves (TSM - BAWs) in the 3–4 MHz range. These waves efficiently induced de-icing and a reduction of ice accretion (i.e., active-anti-icing effect), as well as a decrease in ice adhesion. A drawback of this approach is the necessity of a fast-electronic tuning of the excitation due to the very narrow bandwidth of BAW modes. The possibility of using the same device for ice detection under realistic operation conditions (i.e., experiments in icing wind tunnels (IWT)) and a profitable synergy to reduce power consumption upon the application of anti-icing coatings were also demonstrated in Del Moral et al. work. [24] On the contrary, Jacob et al. [25] demonstrated that R-SAW technology can be applied to large areas and transparent substrates beyond the centimeter scale, proving that R-SAWs can be used for de-icing industrially relevant surfaces. In that work, the authors also demonstrated that R-SAWs generated on piezoelectric plates and piezoelectric thin films induced the de-icing of glaze ice through a pure acoustic mechanism, discarding the melting through direct electrothermal (Joule) heating induced by ohmic losses in the IDTs on substrates with low thermal conductivity. Recently, the same authors demonstrated that a hybrid operational mode is possible by combining R-SAW for de-icing and BAW for sensing by carefully controlling the excitation mode of the two finger combs integrated into the IDTs. [26] Other relevant studies on the implementation of AW ice sensors and de-icing of rime ice have also been recently published. [27–30] Despite the significance of these contributions, de-icing with substrate-integrated AW systems faces substantial fundamental uncertainties that still hinder the implementation of this technology in real-world applications. Critical bottlenecks encompass the assessment of factors affecting the transmission of AW energy to ice or water present on the device surface, understanding the mechanisms behind ice-cracking and other processes contributing to ice removal without complete melting, and defining activation conditions to avoid ice formation under subzero temperatures, thereby enabling an efficient anti-icing mode. Additional features preventing an unequivocal comparison of efficiencies and de-icing mechanisms are the differences in the type of AWs, either SAWs, Lamb waves, or TSM-BAWs, [21–26] the effect of the substrate materials (e.g., transparent and black LN, ZnO thin films deposited on different substrates such as Al foils, glass and fused silica, the hydrophobic or hydrophilic surface modification of these substrates, etc.), or the plethora of conditions used in the experiments (type of ice, liquid water content (LWC) when working in IWTs, static and windy conditions, temperatures and relative humidity, water freezing or ice accretion, etc.). To shed some light on this twilight scenario, we herein systematically investigate the de-icing and active anti-icing activation of the same substrate with either R- SAW or Lamb waves. For this purpose, we use LiNbO 3 (LN) piezoelectric chips activated with IDTs operating at different driving frequencies. The selection of LN for preparing specific model systems to study de-icing with AWs exceeds this topic, as this piezoelectric material is widely utilized in other applications such as high-temperature sensors or optical waveguide tapered antennas. [31–33] Herein, we have electro-acoustically characterized a series of IDT configurations to optimize their performance, either for activating R-SAWs or Lamb waves. Then, for these two wave modes and various subzero temperatures, we performed de-icing and active anti-icing experiments of small ice aggregates and water droplets, monitoring the threshold power required in each case. Additionally, an analysis of the operational conditions of the chips using finite element model (FEM) simulations in COMSOL Multiphysics is used to model the generation of the R-SAW and Lamb wave modes and account for their interaction with ice. Finally, to validate these experimental and simulation results for Lamb waves, we have carried out some proof-of- concept de-icing experiments proving the suitability of the Lamb wave activation to remove large areas of accreted ice. The overall assessment of the de-icing of small ice aggregates and large ice layers with Lamb waves has permitted the proposal of a specific mechanism of de-icing using Lamb waves that differs from that already known for R-SAWs. [25,30] We are confident that all these findings will significantly contribute to establishing a rational pathway for exploiting AWs as the basis for a next-generation de-icing system. R. Kanter, N. Kandelin, H. Koivuluoto Materials Science and Environmental Engineering Faculty of Engineering and Natural Sciences Tampere University 589, 33104 Tampere, Finland www.advancedsciencenews.com www.aem-journal.com Adv. Eng. Mater. 2024, 2401820 2401820 (2 of 16) © 2024 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adem.202401820 by Readcube (Labtiva Inc.), Wiley Online Library on [19/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 2. Experimental Section 2.1. Fabrication and Electroacoustic Characterization of AW Chip Devices The selected piezoelectric material was a 0.5 mm thick black Lithium Niobate 128° YX wafer cut purchased from CSIMC- Freqcontrol, China. The 4 inch wafers were cleaned with acetone and IPA using ultrasonic cleaning for 5 min and O 2 plasma treatment at 100 W for 10 min. IDTs were patterned using photoresist spin-coating, a maskless laser writer (Heidelberg instruments MLA 100), and the lift-off technique. Ti (5 nm)/Al (295 nm) was selected for IDT metallization. Further, 100 nm SiO 2 , [34] was deposited by magnetron sputtering on top of the wafer. This layer acts as a passivation layer and protects the aluminum IDTs from corrosion. This SiO 2 thin film confers a hydrophilic character to the surface of prepared chips (wetting contact angle ≈35°). The contact pads were opened for electrical contact by a dry etching procedure. Figure S1 in the Supporting Information section (SI) shows the utilized mask design and a manufactured wafer aimed at fabricating IDTs with wavelengths varying from 120 to 510 μm and an aperture of 9.95 mm, including various designs with a different number of finger pairs. Chips operating at different wavelengths were diced into individual chips, as shown in Figure S2, Supporting Information. Keeping the same apertures and number of finger pairs, different IDTs layouts and sizes of diced LN plates were used for the chips utilized for proof-of- concept experiments, as explained in the corresponding section. Chip sizes of 15 20, 15 25, and 15 30 mm 2 have been prepared on the wafer, depending on wavelength. The devices with 120 and 510 μm wavelength (henceforth called 120 chip and 510 chip, respectively), used for the de-icing experiments of small ice aggregates as done in this work, were prepared on 15 20 and 15 30 mm 2 LN pieces, respectively. A 420 μm chip was also prepared for a comparative de-icing test with big ice aggregates. This ensured a minimum area of 15 10 mm 2 without IDTs for de-icing experiments. For the electroacoustic characterization of the chips as a function of finger pairs, on-wafer electrical measurements were carried out using a vector network analyzer (VNA, Keysight E5080B) and a wafer prober (Figure S3, Supporting Information). Before RF measurement, a viscous photoresist was applied between the IDTs to attenuate the waves and neglect reflections on neighboring IDTs, thus avoiding undesired effects by measuring the reflection coefficient (S 11 ) curves. 2.2. De-icing and Active Anti-icing Tests with Small Ice Aggregates De-icing and anti-icing experiments were carried out in a custommade icing chamber, where temperatures down to a minimum value of 20 °C can be controlled (see Figure S4, Supporting Information, showcasing a view of this chamber, the chip holder, and the water dosing system). An ad-hoc water dosing system consisting of heatable tubing inserted via the wall of the cooling chamber was employed to enable the dosing of droplets of bi-distilled water on the surface of the chips to produce iceparticles.Insidethechamber,thetubewaskeptata temperature of 2 °C to prevent water freezing inside the tube. The temperature of the substrate was controlled independently from that of the cooling chamber by a Peltier plate on which the holder and printed circuit board (PCB) were placed. Several thermocouples were mounted on the stage and the droplet hose to precisely control the temperature of the chip and water droplet dispenser. A USB camera with an attached C-mount objective was placed out of the cooling chamber to follow the evolution of ice aggregates and droplets. De-icing and active anti-icing experiments were conducted on chips with IDTs of 120 and 510 μm in this custom-built cooling chamber. The icing tests were performed with the best-suited configuration of available chips (according to the number of finger pairs) as determined by their electro-acoustical characterization. Experiments were carried out at three temperatures (5, 10, and 15 °C, as controlled on the Peltier plate) with the stage in a horizontal position. At these temperatures, the porosity and nature of ice formed upon accretion processes from the impact of supercooled water droplets change from glaze (at 5 °C) to mix (at 10 °C) to rime (at 15 °C). [35] However, under the experimental conditions of herein-reported experiments, the ice did not form by accretion but by cooling sessile water droplets. These conditions are prompt to generate little porosity. The stage temperature was continuously monitored during the experiments using a thermocouple mounted on an aluminum holder in direct contact with the Peltier plate. Once the temperature was stabilized, a 22 μL milliQ water droplet at þ2 °C was dispensed on the surface of the chips at a distance of ≈3 mm from the IDTs. After a while, the dispensed water droplet froze into a small aggregate of ice, and then the chip was electrically excited until the ice was completely melted. The minimum activation power (we denote this de-icing power as DP T , a magnitude that depends on temperature T) has been taken as a relevant parameter for comparing the de-icing efficiency of the 120 and 510 μm chips. For the active anti-icing experiments, the same volume of water was dripped on an electro-acoustically activated chip, and the minimum power required to prevent freezing (anti-icing power, AP T ) was taken as the relevant parameter for comparing efficiencies. In these experiments, the surface was permanently activated to avoid ice nucleation and the subsequent freezing of the deposited water droplet. The RF AW activation of the chips inside this cooling chamber was done by contacting the diced chips via a 50 Ωimpedance PCB provided with gold-coated spring-pin connectors as shown in Figure S2, Supporting Information together with the chip holder adapted to conduct the de-icing experiment. An RF supply, BSG F20 (BelektroniG GmbH, Germany), has been used for electroacoustic excitation, and the power supplied by the instrument has been taken as a reference parameter of the actual applied power. 2.3. Proof of Concept: Experiments for De-icing Large Ice Layers with Lamb Waves Two proof-of-concept experiments using Lamb waves have been carried out to determine the de-icing mechanisms of this type of wave applied to large-area ice layers or aggregates. www.advancedsciencenews.com www.aem-journal.com Adv. Eng. Mater. 2024, 2401820 2401820 (3 of 16) © 2024 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adem.202401820 by Readcube (Labtiva Inc.), Wiley Online Library on [19/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 2.3.1. Experiment 1 In the 510 μm chip used for this experiment, the IDT and connection pads were structured on the LN face that was not exposed to ice accretion. This means the 510 μm IDTs and PCB board were on the opposite side of the ice-formation surface. The connecting board was adequately protected to prevent shortage or malfunction with the IDT layout in the center of the LN plate and the connecting pads at each side of the chip (Figure S5, Supporting Information shows the chip, PCB, and protecting board). In this configuration, the connecting pad at the top side of the chip was the electrically polarized one, while that at the bottom was grounded. The icing area was equivalent to the 15 30 mm 2 chip size. Ice accretion was done in the IWT facility available at the TAU laboratory (detailed information about these Ice Laboratory facilities can be found in Koivuluoto et al. [36] ). A maximum wind speed of 25 m s 1 , with a liquid water content of 0.5 g m 3 at a temperature of 10 °C was used to accrete the ice in this experiment. The formation of a mixed ice type is expected from these experimental conditions. An ice film with a thickness between 0.5 and 1 mm, depending on the location on the chip surface, was accreted and kept in the wind tunnel chamber at 10 °C before being subjected to AW activation. 2.3.2. Experiment 2 The experiment was performed with a 510 μm chip of 15 30 mm 2 with the IDT structured on the ice-exposed side of the LN plate. In this case, a larger volume of water of ≈180 μL was deposited in ambient air on the chip surface outside the zone of IDTs. The water covered an area of ≈0.9 cm 2 , separated by a distance of ≈2 mm from the IDT. Then, the chip and connecting board were placed inside the cooling chamber, and the temperature decreased up to 15 °C to induce rime ice formation. Afterward, the chip and board were removed from the cooling chamber, and the AW de-icing was induced in ambient conditions immediately after connecting with the electronics (total time shorter than 10 s). The process was followed by placing a video camera outside. During this time, the sample temperature never surpassed 0 °C without AW activation. These conditions generated some frost on top of the surface of the chip due to water vapor condensation, as visible during the experiments. For comparative purposes, a similar experiment to the one done in Experiment 2 was also carried out with a 420 μm chip generating Rayleigh SAWs. For the experiments, the electronic control and activation systems were placed outside either the cooling chamber (experiment 2) or IWT (experiment 1, although in this case, the ambient temperature was 10 °C). Electronics consisted of a vector network analyzer (VNA) to determine the working frequency during and after ice accretion and a switch allowing to automatically connect the chip to a signal generator system consisting of a signal generator (Keysight33210A) and an amplifier (Mini-Circuits LZY-22þ). Further details of this electronic control and activation system can be found in Del Moral et al. [24] The reported de-icing experiments were carried out applying an RF signal at around 7.3 MHz with slight variations in the order tenths of kHz depending on the chip and peak-to-peak voltages of 53 V (experiment 1) and 70 V (experiment 2), equivalent to approximate power values of 1.7 and 4.8 W, respectively. 2.4. Finite Element Model (FEM) Calculations Finite element (FE) simulation of the 120 and 510 μm chips was conducted using COMSOL Multiphysics version 6.0. This simulation describes the generation and propagation of AWs and their interaction with ice, verifying the easiness of mechanical energy transmission through the piezoelectric LN and the ice interface. For this purpose, we have calculated the displacement field (u)in the substrate and the ice aggregate under the assumption that larger displacements in the ice aggregate mean a better transmission of the mechanical energy of the AW from the activated LN to ice. The ice aggregate was located at the same distance from the edge of the last finger of the IDT as in the experiments, trying to reproduce their actual conditions. To keep the computational costs within reasonable limits, the volume of the simulated droplet has been settled at 4 μL. Ice aggregates were assumed to present a hemispherical shape, as it typically happens after freezing water droplets deposited on surfaces. The AW vibrations of the LN plate with IDTs have been modeled using a series of electrodynamic and mechanic equations, as previously described by Fakhfouri et al. [37] Since R-SAW and Lamb waves on 128° X–Y LiNbO 3 are mainly polarized in the median plane, [38] a two-dimensional (2D) model was simulated using a generalized plane strain assumption. This assumption has been adopted to simulate both the 120 and 510 μm chips. The 0.5 mm thick substrate has been described with the elasticity, piezoelectric, and electric permittivity tensors of LN, [39] which were adequately rotated to account for the actual 128° X–Y crystal cut of the substrates used in the experiments. The ice was modeled as an isotropic linear-elastic material in the form of a semicircle geometry, using the ice properties provided by Victor and Whitworth, [40] corresponding to a compact type of ice similar to glaze ice. All this information, including the description of the characteristics of the IDTs, dimensions, applied voltage, time step used for the simulations and dimension, and other computational parameters, can be found in the SI section, SI6. Notably, the ice–substrate interface between the ice and the LN substrate has been simulated as flat, an assumption that fits well with the conditions of the real system formed by a single crystalline substrate with very little surface roughness. An ideal system consisting of an R-SAW with a long wavelength (510 μm) interacting with the ice aggregate was also simulated for comparative and discussion purposes. The parameters used for computing the 510 μm Lamb wave system were kept for this ideal system, except for the assumption of a finite thickness for the LN substrate. This semi-infinite thickness boundary condition reduces reflection of the AW at the bottom face of the plate to a minimum, that is, imposing a situation typical of the generation of R-SAWs. This simulated model will be designated in the text as 510 μm R-SAW to differentiate it from the simulated model for the 510 μm Lamb waves. Figure 1 presents a scheme with the definition of system coordinates (x,y,z) used for calculations. Since the COMSOL Multiphysics calculations have been carried out in two dimensions (2D), the relevant coordinates are xand z(c.f. Figure 1b). www.advancedsciencenews.com www.aem-journal.com Adv. Eng. Mater. 2024, 2401820 2401820 (4 of 16) © 2024 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adem.202401820 by Readcube (Labtiva Inc.), Wiley Online Library on [19/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License The former corresponds to the propagation direction of the wave, while the latter corresponds to the perpendicular to the LN plate. For calculations, figures, and data presentation, we have adopted the following definitions: 1) Volumetric displacements: the local, wave-induced displacement field undergone by the lattice of the LN plates. They are characterized by a lattice displacement vector u. The relevant magnitudes for data calculation and presentation are its modulus in 2D |u| and the two components u x and u z . 2) Surface displacements: the displacement field undergone by the outer surface of the plate, which in the icing experiment is in contact with the ice and is directly involved in the transmission of the mechanical energy to the ice. This definition simplifies the notation and remarks on the importance of power confinement on the surface of the chip. Surface displacements are characterized by a surface displacement vector, s. The relevant magnitudes to present data and for calculations are its modulus in 2D |s| and the two components s x and s z . The geometry of the whole system has been discretized with an unstructured triangular mesh, and the acoustic modes have been described with at least 10 elements per wavelength considering the following sound velocities: in LN substrate 3981 m s 1 , [39] in ice 1850 m s 1 . [40] The resulting grid had approximately three million elements (equivalent to ≈18.3 million degrees of freedom). Electrical excitation was modeled using an electric potential that follows the periodicity of the IDT fingers, which were located on the left half of the model geometry, neglecting the mechanical damping of the metal electrodes. To avoid undesired reflections of the AWs at the edges of the chips, the AWs were absorbed by incorporating perfectly matched layers (PML) as defined in the COMSOL software documentation. [41] The model was solved in both the frequency domain and the time domain using a frequency solver or a time-dependent solver, respectively. In the frequency domain, the IDTs were defined as an RF port where power is explicitly defined as an input parameter (1 W was arbitrarily selected for the calculations since simulations are linear and provide S 11 spectra and harmonic solutions of the displacement and electric fields, no dependence of results is expected on the used input power). Simulations were also done in the time domain to obtain realtime estimation of the vibroacoustic response of the LN plate and the LN/ice systems upon electroacoustic excitation. In this case, the electrical excitation is applied to the IDTs as a sinusoidal voltage signal with defined amplitude and working frequency as input parameters. Working frequencies were selected based on the frequency domain results following maximum power transmission criteria. To generate equivalent mechanical powers for the AWs in the simulation of the 120 and 510 μm chips, the selected voltage amplitudes in each case were different to compensate for the differences in electroacoustic coupling. Therefore, the resulting electric power for each wavelength was slightly different to ensure that the generated mechanical power was equivalent in all simulated models. 3. Results and Discussion 3.1. RF Characterization of AW Chips The radio-frequency (RF) electrical characterization of the chips was performed by measuring the S 11 (return loss) and |S 11 | 2 (reflection coefficient of power) parameters as a function of frequency. The measured resonance frequencies (frequency of the S 11 minima) closely match the calculated resonance frequencies of the chips, according to the well-known expression f¼vs λ(1) where f,vs, and λare, respectively, the working frequency, the wave velocity on 128° YX LN substrate (3981 m s 1 ) [39] and the wavelength, this latter defined by the IDT layout (i.e., the distance between fingers). The series of return loss spectra plotted as a function of frequency recorded for the different chips prepared in this work are reported as SI7. These spectra show that the S 11 minima deepen for a higher number of finger pairs. To minimize the reflected power, de-icing tests were carried out with 120 and 510 μm chips with the largest number of finger pairs tested during the electroacoustic evaluation, that is, 15 and 26, respectively. To estimate the power efficiency, the reflection coefficient of power |S 11 | 2 is the relevant parameter, which can be calculated from Equation (2) and (3): RLðdBÞ¼10 log10 Pi Pr  (2) jS11j2¼Pr Pi (3) where RL(dB) is the return loss of the chips and P r and P i are reflected and incident power, respectively. Figure 2 shows a series of plots of the reflection coefficient of power |S 11 | 2 versus frequency as recorded for the 120 and 510 μm chips with different numbers of finger pairs. A photograph of these two chips is also included in the figure. The full set of Figure 1. Schematic description of the geometry of chips with displacement field and axis definitions; a) 3D diagram of the chip; b) 2D diagram of the chips as used for the calculations. www.advancedsciencenews.com www.aem-journal.com Adv. Eng. Mater. 2024, 2401820 2401820 (5 of 16) © 2024 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adem.202401820 by Readcube (Labtiva Inc.), Wiley Online Library on [19/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License |S 11 | 2 curves recorded for all manufactured chips is shown in Figure S7, Supporting Information. To ensure the maximum efficiency of the devices, they should present |S 11 | 2 values close to zero to minimize the reflected power upon electrical activation. The plots in Figure 2 show that the reflected power is highly dependent on the number of finger pairs and decreases when its number increases. For the longest wavelength chip, that is, the 510 μm chip, power losses reached saturation at |S 11 | 2 values around 0.25 for 26 finger pairs. For the 120 μm chip, a |S 11 | 2 value smaller than 0.1 is reached with 15 finger pairs. In general, although the reflection coefficient of power could still slightly decrease by increasing the number of finger pairs, the benefits would be very small and likely compensated by the electrical power losses that also increase with this parameter. Therefore, we have selected 15 and 26 finger pairs for the 120 and 510 μm chips for de-icing and anti-icing experiments. This choice is a good compromise between low reflection and an acceptable free chip area for de-icing experiments (see below). These plots in Figure 2 also reveal a lowering in baseline for the 120 μm chip (ideally, |S 11 | 2 should equal one outside of the resonance peak), which is particularly noticeable for the 120 μm chip. This indicates the existence of some electrical power losses, which are not due to the resonance excitation of the crystal and are likely caused by factors such as parasitic capacitances in the IDT, ohmic resistance of the thin film metallization, or contact resistances. These parasitic losses are in the order of 10–20% for the characterized chips, a percentage of the incident power that will not contribute to the generation of AW vibrations through electromechanical coupling. The plots in Figure 2 also show that, as expected, these electrical power losses increase with the number of finger pairs. According to Figure 2, S7/S8, Supporting Information, the electroacoustic response of the 510 μm chip is different. The behavior of this chip is characterized by two resonance minima instead of one in the frequency region of interest. We attribute these two resonances to Lamb plate waves instead of Rayleigh Figure 2. Electrical characterization of AW chips: a1,b1) |S 11 | 2 versus frequency plots for fabricated AW devices with wavelengths (λ) of 120 μm (a1) and 510 μm (b1). Note the different span of the xscales used in each case. The arrow and ellipse highlight the progressive increase of power losses in the series of 120 μm devices as the number of finger pairs increases. Photographs of the 120 μm a2) and 510 μm b2) chip devices with the maximum number of finger pairs. www.advancedsciencenews.com www.aem-journal.com Adv. Eng. Mater. 2024, 2401820 2401820 (6 of 16) © 2024 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adem.202401820 by Readcube (Labtiva Inc.), Wiley Online Library on [19/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License waves. According to the well-established theory of SAW generation in piezoelectric plates, [42] an R-SAW can be regarded as the high-frequency limit of the superposition of symmetric and antisymmetric Lamb waves. These two counterparts start to separate for plate thicknesses of around 1.5 λ. From the |S 11 | 2 plots in Figure 2, it appears that when the wavelength of the AW approaches the thickness of the plate, the conditions for pure Rayleigh-SAW are no longer fulfilled. According to these considerations and Equation (1), the resonance at 7.36 MHz corresponds to the A0 Lamb wave mode, whereas the ≈8 MHz resonance must be attributed to the S0 Lamb wave mode. In the next section, we further support this attribution by the simulations obtained using COMSOL Multiphysics analysis of the AWs in the 510 μm chip. RF chips with a similar configuration but a wavelength larger than 510 μm behave similarly and generate two AWs at different frequencies (see Figure S8, Supporting Information for the |S 11 | 2 curves experimentally determined for a600μm chip, which is reported for comparison). Table 1 includes experimental and calculated (i.e., Equation (1)) resonance frequencies for the manufactured chips with the maximum number of finger pairs in each case, as well as the type of generated AW. 3.2. FEM Simulation of AW Simulation results of the AW generation and propagation in the 120 and 510 μm chips were carried out as described in the Experimental and Methods section. Figure 3a,b present selected results of this simulation analysis for each type of wave. Videos (V1_S10) and (V2_S11) provided as SI showcase the timedependent variation of volumetric displacements represented in the form of color maps with the same color code as for the snapshots in Figure 3a2,b2. The plots in Figure 3a1,b1 compare the experimental and calculated |S 11 | 2 spectra of the 120 and 510 μm chips. The concordance between curves obtained with experiments and simulations confirms that simulations reproduce the experimentally recorded signals, except for the electrical IDT power losses evidenced by the non-zero backgrounds of the experimental curves, a feature particularly noticeable for the 120 μm R-SAW. This proves that, except for this deviation attributable to ohmic losses and/or capacitive effects in the real chips not included in the analysis, the simulations for the two chips agree with their actual behavior. The analysis in Figure 3a2 of the volumetric displacements determined in the time domain for the 120 μm chip confirms that vibrations only affect the outer surface layers of the LN plate, as expected for a typical R-SAW type (see video (V2_S11), Supporting Information). This is evidenced by the color maps in this figure showcasing snapshots of the modulus (|u|) and u x and u z components of volumetric displacements of this wave that confirm the confinement in the outer surface regions of the plate. The plots in the same figure representing the value of |u| along the plate thickness as a function of the coordinate z show that the displacement field goes to zero in the bulk. Figure 3a3 presents the plots of surface displacement components s x ,s z , and the modulus of the displacement vector at the surface |s| along the direction of wave propagation (x). Notably, these parameters correspond to the AW-induced displacements in the plate outer plane of the chip plate, that is, the surface that will be in contact with the ice. Figure 3a3 reveals that maximum displacements along xand zare almost equivalent and are separated by a phase shift of 93°. This result means that the surface displacement field approaches a circular polarization, as shown by the plot in Figure 3c representing the component s x versus s z (note that the different circles in the plot refer to calculations at different times). The “in the time domain”simulation of the electroacoustic activation of the 510 μm chip in Figure 3b 1 gives rise to two well-defined and distinguishable |S 11 | 2 peaks matching the shape of the experimental curves, as well as the frequency values of the minima (Table 1). Meanwhile, the analysis in Figure 3b 1 demonstrates that volumetric displacements affect the whole plate thickness. The color plots in this figure show that the first peak at 7.4 MHz (note that there is a small difference with respect to the experimental minimum at 7.36 MHz) corresponds to vibrations with anti-symmetrical character and should be attributed to a A0 vibrational mode. Meanwhile, volumetric displacements corresponding to the second peak at 8.1 MHz have a symmetrical character (i.e., typical of an S0 mode, data not presented). Meanwhile, plots in this figure for |u| as a function of the coordinate zwithin the plate show that the displacement field does not vanish in its interior and extends from one face to the opposite of the LN plate. Figure 3b 3 shows that the magnitude of s x and s z components is similar, and there is a phase shift of 91º between them, that is, the surface displacements are circularly polarized. This is confirmed by the representation of s x versus s z in Figure 3c. When comparing the waves generated in the 120 and 510 μm chips, it is noteworthy that they present surface and plate character, respectively. Another significant difference is that the surface displacements in the 510 μm chip for the A0 mode are slightly smaller than for the 120 μm chip, and they present a certain modulation due to the superposition of a second contribution of much larger wavelength in the order of 4–5 mm. This second contribution is likely due to a weak standing wave mode with a larger wavelength that is superimposed on the main one (see Figure S9, Supporting Information). 3.3. De-icing Experiments of Small Ice Aggregates To verify the effect of the AW characteristics on the de-icing process, we have systematically compared a series of de-icing experiments carried out with the 120 and 510 μm chips. According to the previous simulations in Section 3.2, these chips generate Rayleigh SAW and Lamb waves, respectively. Table 1. Calculated and experimental frequencies for the best IDT configuration of the chips in terms of the number of finger pairs. Wavelength [μm] Number of finger pair Measured frequency [MHz] Calculated frequency [MHz] Mode 120 15 32.32 33.25 R-SAW 150 16 25.79 26.60 R-SAW 240 20 16.10 16.62 R-SAW 330 22 11.68 12.09 R-SAW 420 28 ≈9.20 9.50 R-SAW 510 26 ≈7.36 & 8.14 –Lamb wave www.advancedsciencenews.com www.aem-journal.com Adv. Eng. Mater. 2024, 2401820 2401820 (7 of 16) © 2024 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adem.202401820 by Readcube (Labtiva Inc.), Wiley Online Library on [19/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Experiments were done at temperatures of 5, 10, and 15 °C. A summary of the experimental conditions is shown in Table 2, including information about temperature, actual frequency, and incident power values (i.e., DPT values), as well as an estimation of the effective power actually available for AW excitation. The latter have been estimated after the correction of the experimental DP T values. A first correction considers that the electrical power that is not inserted into the LN plate in the form of AW electromechanical activation would be equivalent to 1- |S 11 | 2 . For instance, for the 120 μm chip with 15 finger pairs, it 0 0.1 0.2 0.3 0.4 0.5 0 0.4 0.8 1.2 1.6 t= 1/5T t= 2/5T t= 3/5T t= 3/5T t= T ) mn ( ||u|| z (mm) 0 0.1 0.2 0.3 0.4 0.5 0 0.4 0.8 1.2 1.6 t=1/5T t=2/5T t=3/5T t=4/5T t=T )mn( ||u|| z (mm) (b)(a) 28 30 32 34 36 0 0.2 0.4 0.6 0.8 1 Frequency (MHz) Experimental Simulation |11S| 2 32.5 MHz = 120 m 6 6.5 7 7.5 8 8.5 0 0.2 0.4 0.6 0.8 1 Frequency (MHz) Experimental Simulation | 11S | 2 7.4 MHz 8.1 MHz = 510 m -2 -1 0 1 2 -2 -1 0 1 2 510um BAW 120um SAW sz)m n ( s x (nm) (a2) (a3)(b 3) (a1)(b 1) (b2) 345 -1.5 -1 -0.5 0 0.5 1 1.5 )mn( x(mm) sx sz || s || 0.75 1 1.25 -1.5 -1 -0.5 0 0.5 1 1.5 sx sz || s || )mn( x(mm) Surface displacement field ( ) (c) Volumetric displacement field ( ) 9.6 x10 x (mm) )mm( z 6.4 x10 x (mm) 2.6 z (mm) x10 1.3 x (mm) z (mm) 9.6 9.6 x10 x10 6.46.4 x (mm) z (mm) 9.6 9.6 x10 1.31.3 Figure 3. Simulation of the AWs: a,b) simulations for the for the 120 μm (a) and 510 μm (7.4 MHz peak) (b) chips. a 1 ,b 1 ) simulation in the frequency domain of |S 11 | 2 .a 2 ,b 2 ) simulation in the time domain of the bulk displacements. Simulations include: i) color maps snapshots of the bulk displacements (modulus |u| and u x and u z components); ii) plots at various times of the modulus of the bulk displacement vector |u| along the coordinate z(thickness of the plate). a 3 ,b 3 ) plots the modulus |s| and s x and s z components. Note that the xscale is not the same for the two waves to properly compare the displacement fields. c) Plot of the s x versus s z components to assess the polarization of the displacement field at the surface in the two chips. www.advancedsciencenews.com www.aem-journal.com Adv. Eng. Mater. 2024, 2401820 2401820 (8 of 16) © 2024 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adem.202401820 by Readcube (Labtiva Inc.), Wiley Online Library on [19/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License appeared that approximately an 8% of the incident electrical power is not efficiently coupled to the device. As indicated in Section 3.1, parasitic electrical power losses associated with the IDTs should be summed up in this correction. Parasitic losses were important for the 120 μm chip with 15 finger pairs, which amounted to ≈20% of incident power (c.f., Figure 2). In total, a rough estimate of the electrical power actually converted into mechanical excitation with this chip might be in the order of 72% of the incident power (this percentage stems upon substrating the 20% of parasitic losses plus ≈8% due to the coupling issues). For the 510 μm chip, a similar analysis gives an estimated correcting factor of 69% (note that in this case ohmic or capacitive losses associated to the IDTs were smaller, c.f. Figure 2). Since the estimated power losses resulting from these correcting factors were rather similar for the 120 and 510 μm chips, comparative assessments of powers and efficiencies would be equivalent using either incident or effective power values. Note, however, that heating effects due to ohmic losses at the IDTs will likely be higher for the 120 μm chip. The de-icing experiments proceeded in the following way: once the chip temperature was stabilized in the cooling chamber, the system was kept under these conditions for at least 30 min for temperature homogenization. Then, a liquid droplet (þ2 °C, 22 μL) was delivered and frozen on the surface of the chips, forming a hemispherical ice aggregate defined by a circular perimeter of ≈4.5 mm. Then, an RF signal with the incident (DP T ) power reported in Table 2 was applied to generate AWs at the resonance frequencies of each chip. The reported values correspond to the minimum incident power required to induce melting. Both DP T and the effective AW power values in Table 2 confirm that melting on the 510 μm chip is more favorable than on the 120 μm chip. We tentatively attribute this difference to the distinct characteristics of the waves generated in each case, a feature that will be further discussed in the following sections. The melting sequence induced upon application of the reported powers in Table 2 was video-recorded for both chips. Figure 4 shows a series of snapshots representing dripping (Figure 4a1,b1), freezing (Figure 4a2,b2), partial melting Table 2. De-icing data for the 120 and 510 μm chips: temperature, operating frequencies, incident (DP T ), and effective AW powers after correction by the estimated losses. Chip [μm] Temperature [°C] Frequency [MHz] Incident power a) (DP T ) [W] error Effective power [W] 120 15 32.33 3.76 0.26 2.70 10 32.40 2.76 0.68 1.98 5 32.44 1.46 0.20 1.05 510 15 7.36 2.02 0.39 1.39 10 7.40 1.46 0.29 1.00 5 7.36 1.45 0.16 1.00 a) The uncertainty limits for the DP T values define the variation in the measurements in each case. Figure 4. Comparison of de-icing process on the a) 120 μm and b) 510 μm chips. A 22 μL droplet is delivered on the chips at –15 °C a1,b1); droplet is frozen; a2,b2): droplet starts melting, after chip activation with the DP T powers reported in Table 2 a3,b3); droplet remains liquid (active anti-icing effect) a4,b4). The behavior of the melted water droplet was different for the two chips: it spread for the 120 μm chip but was pushed toward the chip edge in the case of the 510 μm chip. The water droplet was dispensed at almost similar distance from the last finger of the IDTs (i.e., at about 3 mm). www.advancedsciencenews.com www.aem-journal.com Adv. Eng. Mater. 2024, 2401820 2401820 (9 of 16) © 2024 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adem.202401820 by Readcube (Labtiva Inc.), Wiley Online Library on [19/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License [22] X. Zeng, H. Ong, L. Haworth, Y. Lu, D. Yang, M. Rahmati, Q. Wu, H. Torun, J. Martin, X. Hou, X. Lv, W. Yuan, Y. He, Y. Fu, ACS Appl. Mater. Interfaces 2023,15, 35648. [23] X. Zeng, Z. Yan, Y. Lu, Y. Fu, X. Lv, W. Yuan, Y. He, Langmuir 2021,37, 11851. [24] J. del Moral, L. Montes, V. Rico-Gavira, C. López-Santos, S. Jacob, M. Oliva-Ramirez, J. Gil-Rostra, A. Fakhfouri, S. Pandey, M. Gonzalez delVal,J.Mora,P.García-Gallego,P.F.Ibá˜nez-Ibá˜nez, M.A.Rodríguez-Valverde,A.Winkler,A.Borrás,A.R.González-Elipe, Adv. Funct. Mater. 2023,33, 2209421. [25] S. Jacob, S. Pandey, J. Del Moral, A. Karimzadeh, J. Gil-Rostra, A. R. González-Elipe, A. Borrás, A. Winkler, Adv. Mater. Technol. 2023,8, 2300263. [26] A. Karimzadeh, U. Weissker, J. del Moral, A. Winkler, A. Borrás, A. R. González-Elipe, S. Jacob, Adv. Mater. Technol. 2024,9, 2301749. [27] V. I. Anisimkin, N. V. Voronova, Ultrasonics 2021,116, 106496. [28] P. Schulmeyer, M. Weihnacht, H. Schmidt, Sensors 2024,24, 2292. [29] V. Anisimkin, V. Kolesov, A. Kuznetsova, E. Shamsutdinova, I. Kuznetsova, Sensors 2021,21, 919. [30] D. Yang, L. Haworth, P. Agrawal, R. Tao, G. McHale, H. Torun, J. Martin, J. Luo, X. Hou, Y. Fu, Langmuir 2022,38, 11314. [31] M. Rashedul, H. Sarker, J. L. Silva, M. Castaneda, B. Wilburn, Y. Lin, N. Love, J. Intell. Mater. Syst. Struct. 2018,29, 938. [32] C. Zhaoa, W. Genga, X. Qiaoa, F. Xuea, J. Hea, G. Xuea, Y. Liua, H. Weia, K. Bia, Y. Lib, M. Xunb, X. Choua, Sens. Actuators: A Phys. 2022,333, 113230. [33] H. Lu, Y. Li, J. Zhang, Sensors 2021,21, 3672. [34] A. Winkler, S. Menzel, H. Schmidt, in Proc. of SPIE, vol. 7362, SPIE Europe Microtechnologies for the New Millennium, Dresden, Germany 2009, p. 73621Q. [35] G. Fortin, J. Perron, in 47th AIAA Aerospace Sciences Meeting Including The New Horizons Forum and Aerospace Exposition, Orlando, FL, January 2009, p. AIAA 2009-0274. [36] H. Koivuluoto, C. Stenroos, R. Ruohomaa, G. Bolelli, L. Lusvarghi, P. Vuoristo, in Proc. of the Int. Workshop on Atmospheric Icing of Structures (IWAIS), Uppsala, Sweden 2015, p. 28. [37] A. Fakhfouri, C. Devendran, T. Albrecht, D. Collins, A. Winkler, H. Schmidt, A. Neild, Lab Chip 2018,18, 2224. [38] D. Mandal, S. Banerjee, Sensors 2022,22, 820. [39] G. Kovacs, M. Anhorn, H. E. Engan, G. Visintini, C. C. W. Ruppel, IEEE Symp. Ultrason. 1990,1, 438. [40] V. F. Petrenko, R. W. Whitworth, in Physics of Ice, Oxford University Press, Oxford 2002. [41] Perfectly Matched Layers, https://doc.comsol.com/5.5/doc/com. comsol.help.aco/aco_ug_pressure.05.106.html (accessed: July 2024). [42] E. A. Ash, in Acoustic Surface Waves (Ed: A. A. Oliner), Springer, Berlin 1978. [43] P. Brunet, M. Baudoin, O. Bou Matar, F. Zoueshtiag, Phys. Rev. E 2010,81, 036315. [44] A. Wixforth, C. Strobl, C. Gauer, A. Toegl, E. J. Scriba, Z. V. Guttenberg, Anal. Bioanal. Chem. 2004,379, 982. [45] R. J. Shilton, V. Mattoli, M. Travagliati, M. Agostini, A. Desii, F. Beltram, M. Cecchini, Adv. Funct. Mater. 2015,25, 5895. [46] H. L. Ong, D. Yang, H. Chen, J. Zhou, L. Haworth, J. Zhang, D. Gibson, P. Agrawal, H. Torun, Q. Wu, X. Hou, Mater. Chem. Phys. 2023,304, 127842. [47] M. Baudoin, R. Chutani, F. Bretagnol, A. Peret, US 20240042969A1 2024, https://patents.google.com/patent/US20240042969A1/en?oq=. þUSþ2024%2f0042969þA1. www.advancedsciencenews.com www.aem-journal.com Adv. Eng. Mater. 2024, 2401820 2401820 (16 of 16) © 2024 The Author(s). Advanced Engineering Materials published by Wiley-VCH GmbH 15272648, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adem.202401820 by Readcube (Labtiva Inc.), Wiley Online Library on [19/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License