Pull-Out Testing of Electrochemically Etched NiTi Shape Memory Alloy Wires in Shape Memory Alloy Hybrid Composites

This study presents an experimental characterization of the interface strength of the shape memory alloy hybrid composite (SMAHC) consisting of the two-way effect NiTi shape memory alloy (SMA) wire structured by electrochemical etching and the surrounding thermoset matrix. Mechanically induced, in situ thermally induced, and cyclic load-increase pull-out tests consistently reveal that SMAHC with structured SMA wires outperforms those with as-delivered SMA wires by a substantial factor of 2.6 – 2.7 in interfacial strength — the critical factor governing the overall performance and functionality of the composite. Analyzing the force – displacement curves from mechanically induced pull-out tests demonstrates that structuring the SMA wire surface leads to a signi ﬁ cantly increased elastic energy to initiate a crack and achieve complete failure. SMA wires with structured surfaces continue transferring load even after the initial interfacial failure due to the presence of intact


Introduction
Shape memory alloys (SMA) enable the conversion of thermal energy into mechanical energy. [1]When combined with fibre-reinforced composites with tailored mechanical properties, a novel class of smart materials known as shape memory alloys hybrid composites (SMAHC) emerges, showcasing outstanding application potential.Utilizing this combination in spaceefficient lightweight actuators offers a promising solution to address current challenges related to lightweight construction and resource efficiency. [2,3]A so-called bimorph bending actuator can be realized by embedding the SMA as unidirectional SMA wires into the polymer matrix of the composite with an offset to the neutral axis, allowing them to induce a bending moment in the overall structure when activated. [4]To ensure this functionality, the interface between the SMA wire and the polymer matrix is of crucial importance, as the force transfer between the SMA wire and the polymer matrix occurs within this contact area, which is often the limiting factor for the durability and the thermomechanical strength of the resulting SMAHC. [5]ull-out tests are conducted to evaluate the force transmission between an SMA wire and a polymer matrix.The interface failure in a pull-out experiment is caused by a combination of different influencing factors such as shear stress and strain accumulation, resulting in the superposition of several failure mechanisms. [6]Shear stress and normal stress are combined into a complex three-dimensional stress state along the interface, which is also influenced by the temperature-dependent This study presents an experimental characterization of the interface strength of the shape memory alloy hybrid composite (SMAHC) consisting of the two-way effect NiTi shape memory alloy (SMA) wire structured by electrochemical etching and the surrounding thermoset matrix.Mechanically induced, in situ thermally induced, and cyclic load-increase pull-out tests consistently reveal that SMAHC with structured SMA wires outperforms those with as-delivered SMA wires by a substantial factor of 2.6-2.7 in interfacial strengththe critical factor governing the overall performance and functionality of the composite.Analyzing the forcedisplacement curves from mechanically induced pull-out tests demonstrates that structuring the SMA wire surface leads to a significantly increased elastic energy to initiate a crack and achieve complete failure.SMA wires with structured surfaces continue transferring load even after the initial interfacial failure due to the presence of intact mechanical interlocking sites.The cyclic load-increase pullout test confirms that SMA wires with structured surfaces exhibit a significantly higher load-bearing capacity, withstanding more load levels and demonstrating greater force of the first failure.Furthermore, in thermally induced pull-out tests with structured SMA wires, a re-initiation of failure from the lower side follows an initial failure progression at the SMA wire entry point.10][11] However, in this test, the influence of temperature during activation of the SMA wire and the resulting nonlinear stress-strain characteristics during phase transformation of the SMA wire are not considered.In contrast, in an in situ thermally induced pull-out test, the SMA wire generates the pull-out force itself upon heating above its phase transition temperature.Despite its inherent complexity arising from the inhomogeneous contraction during thermally induced phase transition and, therefore unpredictable stress distribution along the surface of the activated SMA wire, [12] a thermally induced pull-out test provides a closer approximation of real operating conditions and constitutes an important step in optimizing SMAHCs with controlled and predictable durability.Temperature has been shown, along with the pull-out speed, to notably affect the interface behaviour between the SMA wire and the polymer matrix. [5,9]oreover, various authors have demonstrated the dependency of the force of the first interfacial failure on the microstructure of the SMA phase (martensitic or austenitic), which can be attributed to the different mechanical properties of the SMA wire, such as elastic modulus and transverse contraction, in each phase state. [7,13]Furthermore, it has been observed that at temperatures above the transformation temperature, the stressinduced martensitic transformation within the SMA wire is likely to occur at multiple points, influenced by the constraints imposed by the polymer matrix, thereby impacting the debonding mode between the SMA wire and the polymer matrix. [14]ifferent methods have already been explored to improve the interface quality and force transmission between the SMA wire and the polymer matrix.These methods include sandblasting, [5,15,16] laser structuring, [17] laser gas nitration, [18,19] and various chemical surface treatments [8,10,11,[20][21][22][23][24][25][26][27] of the SMA wire surface to enhance the mechanical or chemical bonding between the substituents.Our previous study introduced a highly promising method for surface structuring of SMA wires through selective electrochemical etching. [28]This technique allowed for the successful generation of a homogenous and reproducible surface structure without compromising the SMA wires's shape memory and mechanical properties.Mechanically induced pull-out tests showed an increase in the force of the first failure at the interface between SMA and the surrounding polymer matrix by more than three times.This improvement was mainly attributed to mechanical interlocking influencing the interfacial failure process during the pull-out test.
In the present study, we have taken the next critical step by subjecting electrochemically etched SMA wires to in situ thermally induced and cyclic pull-out tests.The structuring of the SMA wire surface, which promotes mechanical interlocking and homogeneous constraint of the SMA wire by the polymer matrix, might impact the phase transformation, the debonding mechanism, and the force of the first failure.Therefore, a thorough investigation is necessary to understand and examine these effects.Furthermore, to the best of our knowledge, interfacial studies on SMAHC have predominantly focused on the quasi-static mechanical characterization of the interface, while the durability of the interface still lacks sufficient understanding.
Surface roughness can act as a stressor [29] and it has been previously shown that increased surface roughness generated through conventional methods, such as using abrasive sandpaper, can negatively impact the fatigue life of an adhesive joint. [30]owever, the application of selective electrochemical etching, as shown in our previous research, [28] offers the advantage of removing crystal defects and impurities from the surface of the SMA wire, resulting in a thermodynamically and mechanically more stable surface.Conversely, the microstructure applied to the SMA wire surface is replicated as a negative imprint in the polymer matrix, potentially resulting in stress peaks that could lead to crack initiation and premature interfacial failure under cyclic loading.Therefore, our study aims to address the abovementioned research questions by conducting and comparing three types of pull-out tests: mechanically induced, in situ thermally induced, and cyclic load-increase pull-out tests.

SMA Wire
The two-way effect NiTi SMA wire Flexinol 90c DA-19-5-2 (DYNALLOY, Inc., Irvine, USA), with a nominal diameter of 0.5 mm, was used for the experimental observation.As the material under investigation is an intrinsic two-way effect material, it exhibits a different transition behaviour during the first heating after full mechanical prestraining.The load-free transition temperatures, as determined through dynamic mechanical thermal analysis (DMTA), were as follows: austenite start (As) at 76.6 °C (first cycle) and 58.8 °C (second cycle), austenite finish (Af ) at 82.8 °C (first cycle) and 77.4 °C (second cycle), martensite start (Ms) at 31.5 °C and martensite finish (Mf ) at À24.4 °C.The as-delivered SMA wires had a dark blue oxide layer on the surface from heat treatment during manufacturing.Before each experiment, the SMA wires underwent a heat treatment at 100 °C for 15 min to mitigate any residual stress that might have resulted from storage or the fabrication process.Subsequently, the SMA wires were allowed to cool to room temperature (RT) and return to the martensitic state.The thermomechanical behaviour of the electrochemically etched SMA wires is expected to remain unchanged compared to the as-delivered state, as demonstrated in the previous study. [28]

Matrix Material
The polymer matrix material used was the cold-curing epoxy resin Araldite LY 5052, combined with hardener Aradur 5052, obtained from Huntsman Corporation (The Woodlands, USA). 100 g resin was mixed manually for two minutes at RT with 38 g hardener.To ensure a void-free polymer matrix, the SMAHC was initially cured for two days in an autoclave (two bar at RT) and subsequently cured for five days under atmospheric pressure at RT. Table 1 gives an overview of the material data provided by the manufacturer, [31] including the calculated shear modulus according to Ehrenstein. [32]

Electrochemical Etching
Surface structuring was carried out using a custom-built electrochemical etching setup in galvanostatic mode, as described in previous work. [28]However, an SMA wire with a smaller diameter was used in this study.The lateral surface of the SMA wire ð2πrlÞ is etched with a minimal necessary etching current density, while all etching current flows through the SMA wire with a cross-sectional area πr 2 .Thus the current densities are higher for smaller SMA wire radius, leading to large Joule heating.The SMA wire itself, and especially the contact areas to the SMA wire, can thus be more easily overheated compared to thick wires.Therefore, the electrochemical cell is designed to allow cooling through the electrolyte of both the contact areas (from one side) and the part of the SMA wire, where the current flows.Thanks to the design of the cell, it was possible to perform etching on the 0.5 mm diameter SMA wires used in this work.
All samples had an etched length of 15 cm.Before etching, the SMA wire's ends outside the targeted etching area were polished with sandpaper to eliminate the dark blue surface oxide, ensuring improved electric conductivity.To achieve the desired structured surface, two sets of electric current pulses were applied, with a 10 s interval between them.The process began with an initial nucleation pulse at 10 A cm À2 , followed by five pulses at 3 A cm À2 .Each pulse lasted for 0.5 s, separated by a 1 s pause.After the second series, a 5 s pause was implemented, followed by a final step of a single pulse at 1 A cm À2 for 3 s.A schematic representation illustrating the combination of pulses employed in the process can be found in Figure S1, Supporting Information.

SEM and EDS Analysis
Microscopic images of the SMA wire surface were captured using a ZEISS ULTRA PLUS scanning electron microscope (SEM) equipped with the GEMINI column (Carl Zeiss Microscopy GmbH, Jena, Germany) at an acceleration voltage of 15 kV.The elemental composition analysis was performed using energy-dispersive spectroscopy (EDS) (Oxford Instruments plc, Abingdon, England) at an acceleration voltage of 20 kV.For the analysis of the pulled-out SMA wires, the samples were sputtered before imaging.

Micro-CT
Structured SMA wires have been imaged at the high-energy materials science (HEMS) beamline P07, which is operated by the Helmholtz-Zentrum, hereon, at the PETRA III storage ring at the Deutsches-Elektronen-Synchrotron (DESY, Hamburg, Germany).Standard absorption contrast micro-computed tomography (μCT) was used at a photon energy of 64 keV with 4001 projections acquired over 180°and an exposure time of 40 ms.An indirect detector system was used with a CdWO 4 scintillator, a five times optical magnification, and a CMOS camera developed by the cooperation of Hereon and the Karlsruhe Institute of Technology (KIT). [33]The effective pixel size was 2.55 μm, with a binning of two.Tomographic reconstruction was performed as described in the previous publication [34] and implemented [35] in MATLAB R2020a (The MathWorks, Inc.) using the ASTRA [36,37] toolbox for tomographic backprojection.Following tomographic reconstruction, the images were converted to 16-bit, denoised using an iterative non-local means filter, [38] and segmented using the trainable WEKA segmentation tool [39] in Fiji.The reconstructed volume of the SMA wires was aligned along the long axis using Avizo 2021.1 (FEI SAS, Thermo Scientific, France), and the cross-sectional area was computed for each image along the 2.55 mm (1000 images) height of each SMA wire, as well as the equivalent diameter of the as-delivered SMA wire. [40]The surface area of each SMA wire was determined using a custom MATLAB script to determine the face contact between the SMA wire and background voxels.

Pull-Out Test
In this work, three types of pull-out tests were conducted: the mechanically induced pull-out, in situ thermally induced pull-out, and cyclic load-increase pull-out test.

Pull-Out Test Sample Preparation
To ensure the precise alignment of the SMA wires in the polymer matrix in both surface conditions, as-delivered and structured, a custom-made mold was designed.It consists of a silicone part and a steel frame and is positioned within a steel frame that allows precise adjustment and tightening of the SMA wires, effectively minimizing any potential lateral stresses.The silicone part ensures stress-free demolding of the cured samples.Before embedding, the SMA wires were cleaned with isopropanol and heat treated as described in Section 2.1.The curing process was performed as described in Section 2.2.Subsequently, the pull-out samples were cut to a defined size of 50 Â 18 Â 9 mm 3 (H Â W Â D), in accordance with previously published works. [28,41]The embedded SMA wire length was 50 mm.

Pull-Out Test Setups
Mechanically Induced and In Situ Thermally Induced Pull-Out: The mechanically induced and in situ thermally induced pull-out tests were conducted with a universal testing machine utilizing a setup for tensile testing (Zwick RetroLine, Zwick-Roell, Germany) with a 10 kN load cell.The preload was set to 5 N.For all pull-out tests, custom-made sample holders were used, which allowed for the stress-free clamping of the polymer matrix.
A schematic drawing of the pull-out test and the sample holder can be seen in Figure 1.The load is applied to the upper free end of the SMA wire.At a certain load, the interface becomes unable to transfer the induced load from the SMA wire to the surrounding polymer matrix, leading to interfacial failure.The expected starting point for the interfacial failure is at the top of the pull-out sample due to free edge effects resulting in a high shear-stress-singularity. [42]Optical observation methods were utilized for all pull-out tests conducted.Stress optics were employed to accurately determine the moment of the first failure, as it provides precise identification of the onset of interface delamination, which cannot be achieved solely by analyzing the stress-strain curve. [5,15,28]A setup with linear polarized light, as described in the previous publication, [28] was chosen.The failure progression at the interface was captured using an ultra-high speed camera (Motion v2512, Vision Research, Wayne, USA), coupled with a motion tracking system (Stemmer Imaging AG, Puchheim, Germany).The frame rate of 1000 frames per second and a resolution of 768 Â 768 was used.
10][11] This approach ensures quasi-static stress at the interface and is state-of-the-art for the characterization of SMAHC.Using the same settings as in the above-mentioned studies makes a comparison of the results possible.The pull-out was conducted at a strain rate of 25% min À1 , corresponding to the SMA wires activation speed in a typical application. [5]The free end of the SMA wire was fixed with pneumatic clamps (2 bar) in the upper clamping jaw with a free length of 40 mm.
For the in situ thermally induced pull-out test, the upper jaw clamping of the SMA wire was made in custom-designed, electrically insulating clamps, which were placed over the pneumatic clamps to isolate the SMA wire from the test stand while still enabling current to be injected into the non-embedded end of the SMA wire.Based on the results of previous work, [5] a current of 3 A was selected for activation, leading to a strain rate of %25% min À1 .This choice was also made to ensure that the Joule heating of the SMA wire would be slow enough to observe the transformation of the embedded SMA wire and the associated pull-out due to phase change caused by activation.
Cyclic Load-Increase Pull-Out Test: The cyclic quasi-static loadincrease pull-out test (Figure 2) was performed using an electrodynamic testing device (ElectroPulsTM E1000, Instron GmbH, Darmstadt, Germany).A load cell suitable for a nominal force of AE710 N was utilized.The samples subjected to cyclic loading were prepared in the same manner as those used in the mechanically and in situ thermally induced pull-out tests.A test frequency of 2 Hz was selected to simulate a realistic application frequency for SMAHCs and to enable the observation with stress  optics during the test.The free length of the SMA wire was 40 mm and mechanically clamped in the upper roughened clamping jaw.The lower jaw featured a test frame, previously described for stress-free clamping of the polymer matrix.The test setup was designed to resemble that of the mechanically induced pull-out test closely.A test procedure similar to a load-increase test was used.A static load was applied to the sample with a superimposed cyclic load amplitude of AE20% of the static load for 1000 cycles.The procedure for increasing the load between consecutive load levels involved setting the maximum load attained in the preceding load stage as the load for the next 1000 cycles.This configuration ensured a proportional increase in the load level and the cyclic load amplitude.The initial load was set at 3 N with a cyclic load amplitude of AE0.6 N. Subsequently, the load was systematically elevated, following the described steps, until reaching a load of 55.47 N AE 11.09 N.However, the strain rate varied during the individual load stages due to the constant test frequency.The test was monitored with a stress optics setup similar to the other tests.However, a slower camera, type Panasonic Lumix DMC-GH2 (Japan), was used to reduce the amount of data.Pictures were taken only every 100 cycles, resulting in 10 pictures per load level, automatically triggered at maximum load.From the images, it was possible to trace the development and release of intrinsic stress, just as described earlier in the mechanically and in situ thermally induced pull-out.However, due to the reduced image acquisition frequency of one image per 100 cycles, it was only possible to determine the load at which the first interface failure occurred rather than the exact cycle when it happened.In the analysis, the maximum applied load (static load þ cyclic load amplitude) is considered when comparing the results.

Pull-Out Data Analysis
As outlined in Section 2.6.2,stress optics was utilized to determine the moment of the first interfacial failure and to visualize the failure progression.Synchronization of the recorded mechanical and stress optics data for the pull-out tests was realized with a self-written routine in Python 3.7 using numpy, matplotlib, pandas, xlrs, padoc, ipython, ipykernel, opencv.This allows for an automated compensation of the recording frequency of the universal testing device and the high-speed camera.At least five pull-out samples of each SMA wire configuration were tested, and the Namilov Outlier test was employed to identify possible outliers.The maximum transferable interfacial shear stress was calculated following Greszczuk's model, [43] with the cross-sectional area determined using micro-CT (for details, please see S4, Supporting Information).The analysis of the interfacial shear stress is confined to the mechanically induced and in situ thermally induced pull-out test.As for describing micromechanical cyclic interfacial failure, no analytical model has been developed so far.
The absorbed mechanical energy for both the first interfacial failure (G first failure ) and complete interfacial failure (G max ) was determined by calculating the area under the force-displacement curves obtained from the mechanically induced pull-out test according to the equations: Custom routines developed in MATLAB R2022b (The Math-Works Inc., USA) were employed to process the data and derive the average force-displacement curves.

Surface Structuring
Figure 3 displays the surface structure following electrochemical etching, revealing a uniform distribution of etch pits with interlocking structures across the entire etched SMA wire length.
The surface enlargement calculated from the micro-CT data is 1.4 AE 0.1, which agrees with previous results. [28,41,44]The EDS analysis (Figure S2, Supporting Information) confirms that the electrochemical etching is selective to Ni, as has been shown previously. [28]iTi SMA wires typically exhibit an oxide layer on the surface, with TiO 2 as the predominant oxide due to the segregation of Ti to the surface during SMA manufacturing heat treatments.The oxide layer's exact composition and thickness depend on the heat treatment temperature and duration. [45,46]Structuring a surface with an oxide layer through electrochemical etching presents a challenge, as metal dissolution can only occur after an electrical breakthrough in the oxide layer. [28]However, this process may lead to uneven etching, as can be exemplarily seen in Figure S3, Supporting Information.To overcome this problem, two current pulses at high current densities (nucleation pulses) were applied with a long pause of several seconds in between.The pause after the first nucleation pulse allowed the SMA wire to cool down and allowed the created etched pits to passivate to a sufficient extent.Consequently, with the second nucleation pulse, the oxide was more likely to be attacked than the already etched areas, resulting in the nucleation of a second series of etch pits.In order for the created etch pits to grow into depth instead of growing laterally, a series of short, lower current density pulses with short pauses in between were applied after each nucleation pulse. [28]The last single pulse at low current density further deepened some existing etch pits to create a hierarchical interlocking structure.

Pull-Out Test
In this article, two different properties of a two-way effect NiTi SMA wire were addressed (please refer to the Figure S5, Supporting Information, for an illustrating schematic): 1) Pseudoplasticity: due to the detwinning of the twinned lowtemperature martensite variant, the wire is able to exhibit a pseudoplastic strain ϵ pre of %5%, which fully recovers if the wire is heated above its austenite finish temperature, which is %83 °C for the first contraction.2) "Intrinsic two-way effect": the wire used in this study was trained by the manufacturer to exhibit a load-free intrinsic two-way effect, with a repeatable transformation strain ϵ tr of %À3%, if heated above its austenite finish temperature, which is slightly lower for the succeeding activation steps (%77 °C).
Thus, the mechanically induced pull-out experiment probes the first property, the pseudoplastic detwinning, while the in situ thermally induced pull-out experiment is controlled by the contraction due to the two-way effect.

Interface Strength
Mechanically Induced Versus In Situ Thermally Induced Pull-Out: The results of the pull-out experiments are presented in Figure 4, showing the comparison between the results of the SMA wires with structured and as-delivered surfaces.In the mechanically induced pull-out test, the force at the first failure for the SMA wires with structured surfaces measured 66 N AE 9 N, which is %2.6 times higher than that of the as-delivered SMA wires (25 N AE 2 N).The maximum transferrable shear stresses were 41.2 and 14.9 MPa for SMA wires with structured and as-delivered surfaces, respectively.In the in situ thermally induced pull-out test, the SMA wires with structured and as-delivered surfaces exhibited forces of the first failure of 98 N AE 5 N and 36 N AE 8 N, and corresponding maximum transferrable shear stresses of 26.6 and 9.3 MPa, respectively.Notably, the SMA wires with a structured surface outperformed the as-delivered SMA wires in the in situ thermally induced pullout test by a factor of 2.7.
The observed discrepancy between the results of the mechanically induced and in situ thermally induced pull-out tests can be attributed to the limitations imposed by the elongation at break of the polymer matrix material at room temperature on the interface strength. [5]In the case of in situ thermally induced pullout, the polymer matrix becomes more flexible and undergoes a larger strain.The increased deformability of the polymer matrix can more effectively distribute or store the applied load, resulting in higher forces required for interface failure initiation.The difference in the strain rate during the pull-out test, resulting from intrinsic or extrinsic failure mechanisms, also affects the force transfer. [9]fter interfacial failure, SEM images were taken from the structured surfaces of the SMA wires.Figure 5a reveals an SMA wire surface with the etch pits completely filled with polymer matrix residues, indicating a predominantly cohesive failure within the etch pits while exhibiting a more or less adhesive failure on the remaining surface of the SMA wire.Furthermore, a sharp fracture surface indicates no evidence of plastic flow or deformation of the polymer matrix during the pull-out process.Figure 5b presents the surface after the in situ thermally induced pull-out test, showing not only a substantial number of etch pits filled but also a portion of the SMA wire surface covered with polymer matrix residues that exhibit signs of plastic deformation.An abundant presence of polymer matrix residues on the SMA wire surface indicates a cohesive failure within the polymer matrix.
In a previous study, [5] it has been shown that hand sanding of the SMA wire surface increases the interface strength in mechanically induced pull-out tests, but it did not lead to a higher force of the first failure in in situ thermally induced pull-out tests compared to as-delivered SMA wires.However, for SMA wires structured via electrochemical etching, an increase in interface strength can be observed in both conducted pull-out tests.The improved performance can be mainly attributed to the unique shape of the etch pits, with vertical and omega-shaped wall geometries.This configuration facilitates mechanical interlocking and enhances load transfer, even when the polymer matrix softens due to increased temperature.As demonstrated in a recent study that explored two distinct surface structures created through electrochemical etching, the force of the first failure does not exhibit a linear relationship with the increase in the surface area of the SMA wire. [41]Thus, given the absence of chemical bonding between the SMA wire and the polymer matrix, surface enlargement through structuring appears to play a minor role in enhancing interfacial adhesion. [28,41,44]yclic Load-Increase Pull-Out Test: The load levels that the SMA wires with as-delivered surfaces and structured surfaces withstood are presented in Figure 6.As evident from the graph, SMA wires with structured surfaces exhibited increased loadbearing capacity, with the first interfacial failure occurring at 38 N AE 14 N, which is 2.7 times as high as that of SMA wires with as-delivered surfaces (14 N AE 1 N).Hence, SMA wires with structured surfaces exhibited a performance advantage over SMA wires with as-delivered surfaces by a factor similar to what was observed in both the mechanically induced and in situ thermally induced pull-out tests.However, the standard deviation for SMA wires with structured surfaces is higher.SEM images of the surface following the cyclic load-increase pull-out test resemble those obtained from the standard mechanically induced pull-out test, showing etch pits filled with polymer residues.Moreover, larger polymer residues are observed, not only within the etch pits but also covering the surface of the SMA wire between the pits.This observation suggests that, in specific regions, the failure occurred within the polymer matrix, with mechanical interlocking remaining intact and unable to release.These findings indicate that cyclic loading may have gradually weakened the polymer matrix due to fatigue.This fatigue-induced weakening is likely a contributing factor to the higher standard deviation observed in SMA wires with structured surfaces that underwent a greater number of load cycles.The cyclic loading acts as a stressor that magnifies a system's vulnerabilities, making it especially evident in this case that the polymer matrix appears to be the most susceptible component.

Failure Energy
Due to the localized change in length and diameter during pseudoplastic detwinning or phase transformation from martensite to austenite, an inhomogeneous stress distribution along the interface occurs. [47,48]In the case of the in situ thermally induced pull-out test, accurately predicting the stress distribution along the interface proves to be even more challenging.The thermo-mechanical coupling of the phase transition, together with the thermal conductivity of the surrounding matrix, adds another level of complexity to the theoretical description. [5,12]s the in situ thermally induced pull-out test in our setup only records a force-time curve, and the strain of the SMA wire is not measured, our results cannot be evaluated in terms of fracture mechanics.For the mechanically induced pull-out test, a simplified analysis of the fracture process is presented by comparing the energy values absorbed by the SMA wires with as-delivered and structured surfaces during the test.As anticipated, both energies, G first failure and G max , are higher for SMA wires with structured surfaces compared to SMA wires with as-delivered surfaces.For SMA wires with structured surfaces, G first failure is %8 times larger, while the contrast becomes especially pronounced for G max , which is %60 times larger.
Figure 8 depicts the force-displacement curve during the mechanically induced pull-out experiment.The force and displacement, at which the first interfacial failure occurs, are marked with a black dot for both specimens (as-delivered and structured).In the case of as-delivered SMA wires, the interfacial failure process is instantaneous, and the crack growth through the specimen consumes considerably less energy.The point of the first interfacial failure coincides well with a kink in the force-displacement curve.From this point on, the crack moves along the interface, resulting in a rather uneven force progression until the final failure (marked with an arrow), which then is followed by frictional sliding.For the structured specimen, the point of the first failure has to be determined from the stress optics analysis (shown in Section 3.2.3),where it is visible as a dark shadow, moving from the sample's edge inwards as the surface structuring of the SMA wire leads to a smooth and controlled progression of the crack.SMA wires with structured surfaces continue transferring load even after the initial interfacial failure much longer due to the presence of intact mechanical interlocking sites, acting as crack-arresting sites, leading to significantly larger fracture energy.Comparing the force-displacement curve of the composite with the one of the SMA wire alone (shown in the Figure S6, Supporting Information) it becomes visible, how the surrounding matrix affects the pseudoplastic deformation behaviour of the SMA.The subtle plateau observed in the characteristics of the SMA wire alone is masked by the relatively linear deformation of the polymer matrix.

Optical Stress Measurement
For the mechanically induced pull-out experiments on the structured SMA wires, the anticipated first debonding occurred at the entry point of the SMA wire, attributable to the highest interfacial shear stress.However, in the in situ thermally induced pull-out tests with the structured SMA wires, stress optic data reveals a distinct behaviour, as shown in Figure 9.The interfacial failure initiation was observed at the SMA wire entry point, but at a specific moment in the process (%97 N), the failure re-initiated from the opposite side of the sample.The interfacial failure subsequently progresses from both ends until the sample fails approximately in the middle.This phenomenon is unique to the in situ thermally induced pull-out tests using structured SMA wire  specimens, and it can be attributed to several possible causes discussed below.
The temporal evolution of the stress distribution can be divided into two distinct steps.It can be assumed, that the free length of the wire undergoes its transformation from martensite to austenite (moving from point C to point B in an illustrative schematic shown in Figure S5, Supporting Information) first, as the temperature increase will be the fastest, due to the fact, that no other material is in contact to the wire.This leads to the observed increase in the measured force in the stress-time diagram on the right from 0.7 N to 100.9 N corresponding to time point (1) to time point (3).Due to the shear lag effect, [5,42] this force leads to a stress concentration at the upper entry point of the SMA wire.In the second step, when also the embedded part of the SMA wire begins its phase transformation, the stress distribution along the SMA wire (free and embedded length) can no longer be explained as simply as before.However, in time point (2), corresponding to 45.3 N in the force vs. time diagram, a first interface failure at the upper entry point of the wire can be seen.Between time points (3) and ( 4), the embedded part of the wire begins to transform from martensite to austenite.The stress induced by transformation, along with the outer tensile stress from the free length of the wire leads to a further progression of the interface failure and, in the case of the less abrupt failure behaviour of the structured interface, also induces a second failure at the lower "exit" point of the wire.The slight decrease in the measured force (from 100.9 N to 97.2 N) can be attributed to a settling of the setup.Finally, at time point (5), the proceeding failure from the upper and the lower onset merge in the middle and the whole interface fails.The measured force drops rapidly.

Summary and Conclusion
This research investigated the potential of electrochemical etching for surface structuring of SMA wires, aiming to enhance their interface strength with the surrounding polymer matrix.In addition to the standard mechanically induced pull-out test, in situ thermally induced and cyclic load-increase pull-out tests were conducted to approximate real operating conditions in actuation devices.The main outcomes of the study are: 1) Successful application of electrochemical etching is demonstrated on commercially available two-way effect SMA NiTi wires with a nominal diameter of 0.5 mm and a surface oxide.2) SMA wires with structured surfaces consistently outperformed SMA wires with as-delivered surfaces in all three pull-out test types by more than 2.6 times.3) The highest force of the first failure (98 AE 5 N) was observed in the in situ thermally induced pull-out test using SMA wires with structured surfaces.The elevated temperature enhances the polymer matrix's flexibility, leading to greater strain and improved load distribution or storage within the matrix.The enhanced performance of SMA wires with structured surfaces can be attributed to mechanical interlocking sites characterized by unique etch pits with vertical and omega-shaped wall geometries, where the energy is effectively absorbed both elastically and plastically before the polymer matrix tears.4) In both the mechanically induced and cyclic load-increase pull-out tests, the interface strength is limited by the elongation at break of the polymer matrix material at room temperature.Cyclic loading exposes the system's weakest point, which, in this case, appears to be the polymer matrix, which fails cohesively in several areas.5) The analysis of the force-displacement curves from mechanically induced pull-out tests revealed that SMA wires with structured surfaces demanded notably higher energy inputs to initiate a crack and reach complete interfacial failure.In contrast to SMA wires with as-delivered surfaces, SMA wires with structured surfaces sustained load transfer even after the initial interfacial failure due to the presence of intact mechanical interlocking sites.6) In the in situ thermally induced pull-out tests using structured SMA wires, the first interfacial failure occurred at the SMA wire entry point, followed by a re-initiation of the interfacial failure progression from the opposite side of the specimen.This behaviour can be attributed to an accelerated phase transition in the free length part of the SMA wire and a less abrupt failure progression in the structured interface.
In conclusion, building on the results of three types of pull-out tests, electrochemical etching appears to be a promising technique for structuring the surface of SMA wires in SMAHC.Future research will involve fabricating actuation devices that incorporate SMA wires with structured surfaces to assess their performance in hybrid systems.

Figure 1 .
Figure 1.a) Photograph of the experimental setup for mechanically and in situ thermally induced pull-out tests with labelled components: A (upper isolated lamping jaw with electric connection for in situ thermally induced pull-out test), B (free length of the SMA wire), C (analyzer and polarizer), D (pull-out specimen), E (lower clamping jaw), and F (unloaded end of the SMA wire with electric connection for in situ thermally induced pull-out test).The light source, positioned behind the setup, and the camera, standing in front, are not visible in the photograph.The standard mechanically induced pull-out test involves applying a force directly to the non-embedded SMA wire end.In the in situ thermally induced pull-out test, the clamping jaws remain fixed, and the pull-out is induced by activating the SMA wire.The pull-out force is measured by the universal testing machine.The schematics b) illustrate the difference between the two test setups.

Figure 2 .
Figure 2. Laboratory test setup a), showing the electrodynamic testing device employed for conducting the cyclic load-increase pull-out test.The schematic representation b) illustrates the test principles of the cyclic load-increase pull-out test.

Figure 3 .
Figure 3. SEM micrographs of an electrochemically etched SMA wire surface, demonstrating homogeneous distribution of etch pits: a) structured surface at Â300 magnification, b) two pieces cut from the same sample.Left: area from the centre, right: area closer to the edge at Â100 magnification, c) etch pit with interlocking structure at Â4.42 K magnification.

Figure 4 .
Figure 4. Results for the force of the first interfacial failure obtained in mechanically induced, in situ thermally induced, and cyclic pull-out tests for SMA wires with as-delivered and structured surfaces.

Figure 5 .
Figure 5. SEM images of the SMA wire with structured surface after the a) mechanically induced, b) in situ thermally induced, and c) cyclic load-increase pull-out tests.The markings serve as examples, highlighting selected residues of the polymer matrix on the SMA wire surface.

Figure 6 .
Figure 6.Scheme of increasing static load and corresponding cyclic load amplitudes used in the cyclic load-increase pull-out test, with vertical lines marking the first interfacial failure.

Figure 7
compares the mechanical energy absorbed by the sample until the moment of the first failure (G first failure ) with the mechanical energy absorbed until complete failure occurred (G max ).

Figure 7 .
Figure 7.The energy required for the first and complete interfacial failure in the mechanically induced pull-out test, plotted on a logarithmic scale.

Figure 8 .
Figure 8. Force-displacement curves and the first interfacial failure obtained from the mechanically induced pull-out test: Average curves (highlighted in thick lines) and individual original curves (depicted in thin lines).The marginal discrepancy in slope observed among the average curves at the start of the experiment is likely an artefact resulting from the sliding effect.The inset provides an enlarged view of the average curve for as-delivered wires, highlighting areas where complete failure and frictional sliding are observed.

Figure 9 .
Figure 9.In situ thermally induced pull-out test using the SMA wire with structured surface embedded in the polymer matrix.Left: optical stress measurement with red boxes indicating the location of the maximum shear stress.Right: force versus time diagram highlighting corresponding points of the maximum shear stress.At point 4 (97.2N), a distinct re-initiation of failure emerges from the lower side of the specimen in the in-load direction.

Table 1 .
Material data of the epoxy matrix utilized.