Electrochemically Responsive 3D Nanoarchitectures

Responsive nanomaterials are being developed to create new unique functionalities such as switchable colors and adhesive properties or other programmable features in response to external stimuli. While many existing examples rely on changes in temperature, humidity, or pH, this study aims to explore an alternative approach relying on simple electric input signals. More specifically, 3D electrochromic architected microstructures are developed using carbon nanotube–Tin (Sn) composites that can be reconfigured by lithiating Sn with low power electric input (≈50 nanowatts). These microstructures have a continuous, regulated, and non‐volatile actuation determined by the extent of the electrochemical lithiation process. In addition, this proposed fabrication process relies only on batch lithographic techniques, enabling the parallel production of thousands of 3D microstructures. Structures with a 30–97% change in open‐end area upon actuation are demonstrated and the importance of geometric factors in the response and structural integrity of 3D architected microstructures during electrochemical actuation is highlighted.


Introduction
[11] While significant advances have been made in this field, it remains challenging to actuate these structures as this requires controlling the surrounding conditions (pH, temperature, etc).In addition, the fabrication of architected nanostructures presents notable complexities.So far, emerging micro and nanomanufacturing techniques such as DOI: 10.1002/adma.202304517
Herein, we address both challenges outlined above.First, we leverage an electrochemical alloying process that uses a simple current or voltage input to achieve a reversible, non-volatile, and controllable change in volume of Tin (Sn) as it alloys with Lithium (Li).This isotropic swelling of Sn is transformed into a directed motion through architected anisotropic carbon nanotube (CNT) composites (see further and Video S1, Supporting Information).Second, these structures are made by batch wafer processes including lithography, chemical vapor deposition (CVD) of CNTs and atomic layer deposition (ALD).Batch wafer processing offers scalable, cost-effective, and compatible fabrication techniques, providing distinct advantages for producing numerous microstructures and large area morphing micro-and nanostructures.Competing technologies such as E-beam or 2-photon lithography can deliver higher resolution at the cost of lower process throughput.In essence, the proposed actuation mechanism relies on conductive CNT networks that transport electrons to and from a Tin (Sn) coating to drive a lithiation and de-lithiation process.This in turn leads to volume changes of the Sn layer with up to 250% upon full lithiation to Li 22 Sn 5 (capacity of ≈993 mA h per gram of Sn), [29] which pushes the CNTs into a new configuration (see Figure 1a).Further, compared to previous work using Si alloying, [1] Sn allows for repeated actuation.This process works particularly well because CNTs have excellent tensile strength (100 GPa), high Young's modulus (1 TPa), [30] high fatigue resistance, [31] electrochemical stability, and remarkable electrical and thermal conductivity. [2]In this study, CNTs serve as a versatile 3D scaffold and current collector for the active material.Techniques to arrange CNTs in aligned 3D structures allow to precisely control the spatial arrangement of active materials and transform the uniform swelling of the active material into a directed actuation.Overall, this facilitates the creation of intricate and complex 3D responsive microstructures.[34] Furthermore, tin's high stiffness as compared to previously proposed gel based responsive materials [2] makes it effective at displacing the relatively stiff CNT network.Figure 1a depicts how at the microscale, vertically aligned CNT arranged in a cone-shaped structure can open and close as a result of this swelling process (the rationale behind the intricate wall design shown in Figure 1a will be discussed further on).Overall, the hybrid CNT-Sn microactuators created in this study display actuation up to 97% increase in open-end area.By controlling the process conditions, structures ranging from tens of micrometers to a millimeter in height have been demonstrated across large areas.These structures exhibit high cyclability of over 70 cycles, which represents important improvements compared to previous reports on reconfigurable microstructures and active surfaces.

Fabrication and Actuation of Sn Coated 3D CNT Microstructures
To create 3D CNT architectures, we first define catalyst patterns (1 nm of Fe on 9 nm of Al 2 O 3 ) via lithography on a Silicon (Si)wafers coated with a 40 nm molybdenum (Mo) layer as current collector (see Experimental Section and Figure S1a, Supporting Information).The CNTs are subsequently grown by CVD, [33,34] which extrudes the lithographically defined catalyst patterns into vertically aligned 3D CNT forests with heights ranging from microns to millimeter as demonstrated in Figure 1b,c.This fabrication method is compatible with wafer scale processing [35] and the CNT synthesis rate is fast compared to other microfabrication processes (≈100 μm min −1 ).Like most composite materials, the resulting 3D structures have anisotropic mechanical properties which are greater along the length of the CNTs fillers as compared to in the lateral direction. [33,36]The 3D CNT microstructures are then conformally coated with a ≈100 nm-thick layer of tin oxide (SnO 2 ) using ALD using a wafer scale tool (Figure 1b).This process was followed by reduction of SnO 2 to Sn metal in a reducing hydrogen atmosphere at 750 °C [37] (Figure 1b).Note that compared to physical vapor deposition methods that allow to deposit Sn metal directly, ALD is chosen here because it allows for material deposition inside the CNT structure.During the reduction of SnO 2 to Sn, the ALD coating shrinks and beads up (see Figure 1b), pulling the CNTs in a closer packing and transforming the prismatic forests to cone-shaped 3D microstructures.This transformation is governed by the spacing between the CNTs, the adhesion of the CNTs to the substrate and the amount of shrinkage of the SnO 2 layer during reduction.This process is somewhat similar to solvent based elastocapillary aggregation of CNTs reported previously. [33]All the microstructures can be manufactured in parallel as all fabrication steps are relying only on batch silicon wafer processes and the height of the structures can be adjusted by changing the CNT synthesis time (See Figure 1c).A chip, cut from a wafer with hundreds of microstructures on its surface is shown in Figure S1b, Supporting Information.
The 3D microstructures are actuated by lithiating Sn by applying a constant current (galvanostatic) lithiation of the actuators versus Li as the counter electrode in a LiPF 6 salt in carbonate solvent electrolyte (see Experimental Section for more details).The actuation process is controlled by the charges transferred to the actuator, which we achieve by tuning the current density and the duration of current application.Alternatively, if a pre-set actuation is desired, a current can be applied until a pre-set potential versus Li is achieved, for instance, in what follows, the structures are considered fully actuated when a cut-off voltage of 0.2 V versus Li is achieved.40] The actuation is first studied using ex situ scanning electron microscopy (SEM), and Figure 2 shows the actuation of different structures considered in this study.Because the structures are stuck to the substrate at their base, their radial expansion is most prominent at their free end (Figure 2).Further, the microstructures offer anisotropic properties with significant lateral actuation (≈30%−97% increase in open-end area) and negligible vertical actuation (≈1% elongation in height).

Geometrical Factors of 3D Microstructures
The first structures we considered are micro cylinders with monolithic walls (Figure 2a).However, these show excessive cracking of the wall even after only one actuation cycle.This may be due to a combination of the geometry as well as the ALD coating being non-uniform inside thick structures and being concentrated on the outside of the structure.Cracking can partially be addressed by decreasing the wall thickness but is not fully resolved (Figure S2, Supporting Information).Alternatively, we patterned the wall of the cylinders with elongated hexagonal lattices to allow for a more uniform ALD coating (Figure 2b).In addition, finite element analysis (FEA) was performed to study the effects of the wall geometry on the actuation performance of our 3D microstructures.These simulations take into account the swelling of the active battery material to predict the active structure change, details on the FEA simulation are provided in Supporting Information/ Experimental Section.The definitions of geometric parameters that were varied in this optimization process are listed in Table S1, Supporting Information, and the details of simulation conditions are explained in the Experimental Section and Supporting Information.The simulation results show that rather than making elongated hexagonal lattices in the cylinder wall, more compliant lozenge structures result in a larger diameter change (see Figure S3, Supporting Information).This trend is confirmed experimentally in Figure 2. Furthermore, the stress distribution analysis reveals that the microstructures with periodic lozenge lattices experience lower average stress on the opening end as compared to those with monolithic or slitted walls (elongated hexagonal lattices) upon lithiation (Figure 2; Figures S3 and S4, Supporting Information).
We also demonstrate that the actuation can be further controlled changing the inclination angle of the original cone structure.Our parametric sweep FEA study shows that, for microstructures with a constant ratio of height to base diameter, there is an optimal wall inclination angle that achieves maximum actuation (i.e., change in opening area of the cone).In the case of the 3D fabricated microstructures examined in this study (with a height to base diameter ratio of 115:200 μm), the optimal wall inclination angle was found to be ≈20°(Figures S5 and S6, Supporting Information).Initially, our analysis relied on the natural occurrence of wall inclination angles resulting from this process, however, this range of cone angles that can be fabricated using this approach are limited.To address this, we employed the method of elastocapillary aggregation [33] before the ALD coating, allowing us to explore steeper wall inclination angles as illustrated in Figure S7, Supporting Information.
For these experiments, we first densified the CNT forests by the condensation and evaporation of liquid (e.g., acetone).This process (schematically explained in Figure S7, Supporting Information) which has been called "capillary forming" [33,41] transforms the vertically aligned CNTs grown on the lithographically defined patterns into the intricate and robust 3D shapes by selfdirected capillary action. [33,41]After capillary forming, the final shapes of 3D microstructures are held by a combination of mechanical interlocking of the aligned CNTs because of their waviness as well as intermolecular attractive forces. [33,41]Figure S7, Supporting Information, shows different examples of capillary formed 3D microstructures.Methods for controlling the opening angle of CNT cones have been reported previously. [33]The Video S2, Supporting Information, illustrates the actuation of two side by side 3D microstructures with a ≈15°and ≈40°wall inclination angles.The 3D microstructures with a wall inclination angle of ≈15°showed a horizontal displacement (XY-displacement) of around 15 μm on the opening plane, which is equivalent to an increase of up to ≈41% in opening area.On the other hand, the microstructures with a wall inclination angle of ≈40°achieved a horizontal displacement of around 3 μm on the opening plane, which is equivalent to an increase of up to ≈33% in opening area.Our simulations show good agreement with the experimental data, with a displacement of 16.3 μm at a wall inclination angle of 15°a nd 5.5 μm at a wall inclination angle of 40°.

Shape and Color Cyclability/Reversibility of the 3D Microstructures
Next, we designed a costume in situ optical cell to observe and record the changes in configuration of the 3D microstructures during repeated actuation cycles (Figure S8, Supporting Information).Lithiation and delithiation were performed at a constant current density with cut-off voltages of 0.2 and 0.9 V.An in situ recorded video (Video S3, Supporting Information) shows the actuation of a 3D microstructure after two cycles (in addition to the formation cycle; Figure S9, Supporting Information) at C/5 rate.The corresponding voltage-capacity versus the actuation (% change in open end area)-capacity profiles of the 3D microstructure shows a close agreement between the electrochemical events and actuation (Figure 3a): In the discharge curve, the first plateau is the indication of transition from Sn to Li 2 Sn 5 followed by a small bump due to the transformation of Li 2 Sn 5 to LiSn, the next longer voltage plateau is the result of transition from LiSn to Li 2.5 Sn.The ending slope in the discharge profile represents the formation of Li 22 Sn 5 , which are consistent with previously reported electrochemical behavior of Sn in Li-ion batteries. [40,42]As this lithiation process takes place, the material volume increases and when the process is reversed, the material shrinks as the same electrochemical events take place during de-lithiation.
Progressive measurement of actuation reveals that the transition from LiSn to Li 2.5 Sn occurring between 0.45 and 0.35 V has the most significant contribution in the actuation of the reconfiguring 3D microstructures.Figure 3a shows both the cell voltage and actuation as a function of the amount of Li (capacity in mAh g −1 ) stored in the actuator.
In addition to controlling actuation by current, we also carried out voltage sweeps (cyclic voltammetry) at a rate of 0.1 mV s −1 between 0.05 and 1 V. Figure 3b shows that after the formation cycle, four cathodic (discharge, lithiation) peaks were observed: D 1 : 0.67-0.65V (formation of Li 2 Sn 5 ), D 2 :0.53 V (formation of LiSn), followed by the strongest peak D 3 :0.39-0.35V (formation Li 5 Sn 2 ) and last D 4 : 0.15-0.17V (formation of Li 22 Sn 5 ).Four peaks during the anodic process (charge, delithiation), at C 1 ≈ 0.33, C 2 ≈ 0.61 V, C 3 ≈ 0.72, and C 4 ≈ 0.79 V represent the reversible phase transitions corresponding to the cathodic peaks D 1 , D 2 , D 3 , and D 4 , respectively. [40,42]e observed that the electrochemical lithiation of Sn can provide sufficient force for the actuators to self-destruct, for instance by detaching themselves from the substrate (Video S4, Supporting Information).This can be addressed by reducing the amount of Sn deposited on the CNTs or by limiting the cut-off voltage.In this work, cut-off voltages of 0.2 V instead of 0.05 V were used to prevent excessive volumetric expansion when fully lithiating to Li 22 Sn 5 and the high stress this causes in the structure.As discussed above, the angle of the CNT cones can be modulated by capillary aggregation process, and this has a direct impact on the observed actuation.Interestingly, capillary aggregation also improves the adhesion of the CNT microstructures to the substrate, [33] which results in better cyclability.The capillary aggregated structures can be operated at lower cutoff voltages of 0.05 V without the actuators self-destructing as discussed above (Video S5, Supporting Information).The 3D microstructures lithiated with lower cutoff voltages of 0.05 V exhibited up to 97% actuation after formation cycle (% increase in opening area) and 93% reversibility after three cycles (see more quantitative details in Figure S10, Supporting Information).
Figure 3c illustrates the actuation (% change in open end area), stability and the operation at different current densities (test performed with cut-off voltages of 0.2 and 0.9 V).By increasing the rate to 2C, the degree of actuation is reduced to ≈20%, which is expected as battery materials are known to only partially lithiate at high current densities.Interestingly, Figure 3c shows a good agreement between the amount of lithiation (capacity) and actuation at different current densities.Video S1, Supporting Information, shows actuation of a 3D microstructure for nine consecutive cycles at three different C-rates.Lithiation and delithiation cycling at high rates, as observed in Video S1, Supporting Information, can induce the reconfiguration of microstructures in less than 12 min.While this may limit the application domains where these actuators will find use, adaptive surfaces often only need to adjust to daily cycles, such as for heat management.In such cases, actuation response times of only 12 to 24 h are needed.By controlling the current rate per mass of active material, we can adjust the speed of actuation.For instance, we observed an increase in actuation speed (change in open end area) from around 120 nm min −1 to around 960 nm/min when the rate was from C/5 to 2C (Videos S1 and S3, Supporting Information).
Finally, we studied the degradation mechanisms in our actuators.For these experiments, we carried out more than 70 actuation cycles, and carried out ex situ SEM imaging.After 70 cycles, the actuation % dropped by ≈18% (Figure 3c).SEM of the microstructures before and after 73 cycles shows that the geometry of the fully lithiated microstructures is preserved, but some cracks are starting to appear as a result of the repeated swelling and shrinking processes (Figure 3e,f and Figure S11, Supporting Information).We anticipate that microstructure failure is primarily caused by the expansion and contraction of the Sn layer which leads to cracks that grow over multiple cycles.Figure S11, Supporting Information, shows the actuator after one and after more than 70 charge discharge cycles and illustrates how both the CNT wall and active Sn particles crack during this process.
In addition, we found a strong correlation between electrochemical irreversibilities (measured as coulombic efficiency, which tracks charges that are lost in side-reactions [43] ) and the reversibility of the actuation stroke during each cycle (Figure 3d).During the first four cycles, the 3D microstructures experience large nonreversible deformations (up to ≈30%) which reduce to less than 90% after fifth cycle.The same trend is observed in the CE, which is low during the first cycles due to the formation of solid electrolyte interface and irreversible reactions or contact loss of the Sn particles.The close agreement of the CE and actuation reversibility is a logical consequence of the nature of the electrochemical processes that govern the actuation process.More detailed galvanostatic charge-discharge curves at different C-rates and cycle number are provided in Supporting Information (Figure S12, Supporting Information).
Furthermore, the electrochemical alloying reaction of Sn with Li is a spontaneous process, therefore when the actuator is "charged" or de-lithiated, it can spontaneously actuate or "discharge"/lithiate.In Video S7, Supporting Information, we demonstrate how the actuators can self-actuate without external power supply by shorting the actuator over a 20 kΩ resistor, but this current could also be used to drive a process rather than being dissipated in a resistor.The constant load discharge profile of this experiment is shown in Figure S13, Supporting Information.This offers new opportunities for self-deployable and large geometrical reconfigurable structures.
One of the dis-advantages of electrochemical actuators is that they need to operate in an electrolyte.While for some systems such as the study of artificial cilia this is not an issue, for others, this poses challenges that the field of electrochemical actuators has yet to resolve.In this paper, as a first example of a potential application, we focus on changes in optical properties of the structures caused by changes in color during Sn alloying and in appearance due to morphological changes during lithiation.This may on the long term enable optically adaptive surfaces which can function in glass packaged structures as the ones tested in this project.
Materials capable of simultaneous color change and actuation hold promise for active camouflage applications.Previously, both organic materials like metal-organic frameworks [44,45] and inorganic substances like TiO 2 /MXene, [46] V 2 O 5 , [47] and tungsten oxide [48] have been employed as active electrochromic layers.However, achieving both optical color and shape changes with a simple electrical signal has proven to be a challenge.For instance, previous research has shown that coating a nanolayer of electrochromic polymer onto the ligaments of nanoporous gold, resulting in reversible dimensional and color changes during electrochromic actuation. [49]In more recent research, the use of W 18 O 49 [48] and V 2 O 5 [47] nanowires demonstrated the dualresponsive color and shape actuation.While these advances are impressive, these dual-responsive electrochromic materials were packaged between transparent electrical electrodes and the actuation mode is limited to bending sheets whereas in this work, 3D microstructures are actuated, which can be arrayed to form structured surfaces.Figure 4 shows the color-changing behavior upon lithiation of our microstructures.The electrochromic behavior is most pronounced during the formation cycle, as evidenced by the substantial shift in the weighted intensity histograms of optical images taken during cycling (Figure 4b,c).Moreover, we have confirmed the reversibility of the electrochromic response after three cycles, as shown in Figure 4b.As demonstrated in Figure 4 and Videos S1, S2, and S3, Supporting Information, the structures reversibly switch in brightness as they are lithiated.Ultimately, these structures could allow for the development of electrochemically active surfaces with dual-responsive camouflagereconfigurable properties, where each 3D CNT structure would serve as a tunable pixel.

Conclusion and Outlook
This article presents new responsive nanomaterials that can be structured into complex architectures using scalable manufacturing processes.The actuation principle is based on the electrochemical lithiation of an active material, which by a simple current voltage This allows for a continuous position control rather than on-off switching and the structure is maintained at any given position when the input signal is stopped (non-volatile).[58][59] In addition, the active material used in this study (Sn) achieves expansions similar to polymers (≈250%), but because of its higher stiffness, we achieve large actuation than previously reported for CNT-gel composites. [10]In actuators with monolithic walls, we observe only small actuation (area change of 31%) and cracks forming in the actuator walls after only one actuation cycle.Using FEA, we demonstrate how changing the side wall from monolithic to a lattice can reduce stress and therefore cracks, and FEA was also used to optimize the overall design and achieve large strokes (up to area change of 97%) and can still actuate after 70 cycles despite some cracks forming in the side walls.
Despite the air-sensitivity of our electrochemically responsive system, we have shown some initial steps toward applications such as active tuning of optical properties, offering exciting potential for adaptive materials and responsive electrochromic architectures.While the actuation rate in our electrochemically responsive 3D nanoarchitectures may be slow, these structures offer unique reconfigurability and electrochromic features.
Furthermore, this work introduces electrochemically dualresponsive 3D microstructures that combine reversible reconfiguration and electrochromism.These 3D structures exhibit a unique bifunctionality, offering exciting potential for applications in adaptive materials and responsive electrochromic architectures.

Experimental Section
Growth of Patterned Vertically Aligned CNT Forests: Si wafers were first coated with 5 nm chromium (Cr) as an adhesion layer for deposition of 40 nm molybdenum (Mo) to create a conductive surface.The 3D CNT architectures were grown on the surface of Si wafers patterned with a catalyst layer of 9 nm Al 2 O 3 and 1 nm Fe.The catalyst was patterned via photolithography with AZ5214E photoresist.The Cr, Mo, and catalyst layers were deposited using Lesker e-Beam Evaporator.The CNT forests were grown via thermal CVD at atmospheric pressure and 800 °C with growing speed of ≈100 μm min −1 in a horizontal tube furnace in presence of C 2 H 4 /H 2 /He gases respectively flowing at 100/400/100 standard cubic centimeters per minute (SCCM).The grown CNT forests were cooled down for 5 min by moving the tube out of the heated region of the furnace while the tube atmosphere was maintained.
Fabrication of 3D Sn/CNT Microstructures: SnO 2 was conformally deposited by ALD (PICOSUN).Tetrakis(dimethylamino)tin (TDMASn) and H 2 O were used as precursors.Using N 2 as the carrier gas, TDMASn (at a flow rate of 150 SCCM) and H 2 O (at a flow rate of 120 SCCM) were sequentially pulsed into the deposition chamber (at 150 °C) for 600 and 100 ms, respectively.Following each precursor pulse, the chamber was purged for 10 s with 100 SCCM of N 2 .This process was repeated for 1500 cycles to reach ≈100 nm-thick layer of SnO 2 1 .The SnO 2 -coated microstructures were moved to the tube furnace under a reducing H 2 atmosphere (H 2 flows at 400 SCCM) at 750 °C [37] to transform SnO 2 to Sn metal.Quantifying the exact thickness of the active material (Sn) presented challenges due to the porous nature of CNT forests resulting in un-even SnO 2 depo-sition as well as de-wetting of the SnO 2 film during reduction of SnO 2 to Sn(0).While the authors attempted to increase the ALD thickness to more than 200 nm, this led to undesirable effects such as the detachment of microstructures from the substrate during reduction of SnO 2 to Sn (Figure S14, Supporting Information).An ALD thickness of 150 nm resulted in the formation of larger Sn beads instead of a more uniform coverage of CNT forests with Sn (Figure S14, Supporting Information).Large Sn beads on the surface of the actuators were not likely to contribute to the actuation of the overall CNT microstructures (and can introduce errors in the recorded electrochemical results).To ensure the integrity of these microstructures and measurement accuracy, a constant ALD thickness of ≈100 nm was maintained throughout the experiments.A representative sample was employed to measure the total Sn mass loading after ALD of SnO 2 and reduction to Sn using a microbalance (Mettler Toledo).
Electrochemical Analysis: Two types of CR2032 coin cells were modified and custom-designed; one with optical window for in situ virtualization of the 3D microstructures actuations (Figure S8a, Supporting Information) and one without optical window for cyclic voltammetry and postmortem imaging of the actuated 3D microstructures.The design used a supportive 1.2 mm-thick polypropylene washer which is attached by paraffin wax on top of the Mo-coated Si-substrate to create a sealed small cavity for electrolyte (Figure S8a,b, Supporting Information).This allows for use of very small amount electrolyte ≈10-15 μL which greatly reduces the side reactions and enhances the accuracy of the electrochemical analysis.The electrolyte used in this study consists of 1.3 m LiPF 6 in ethylene carbonate/ethyl methyl carbonate/diethyl carbonate (3/5/2, v/v/v), 10% fluoroethylene carbonate + 0.5% vinylene carbonate + 0.2% LiBF 4 (Targray).The 1 m LiPF 6 in ethylene carbonate/diethyl carbonate (50/50, v/v) (Sigma-Aldrich) exhibited a lower coulombic efficiency as shown in Figure S15, Supporting Information.
Lithium metal as a counter electrode and microporous polyethylene (Celgard 2325; PI-KEM) separator were used.The custom-designed coin cells were constructed inside Ar-filled glovebox.After cell assembly, the electrochemical testing was conducted using a BioLogic SP-150 Potentiostat or BCS 805 series battery cycler.
Image Collection and Processing: RGB color images of actuators were collected during cycling using an Olympus LC30 3.1-megapixel camera attached to a BX53M microscope in dark-field mode.A Märzhäuser TANGO stage was used alongside Olympus Stream Motion software to collect images during cycling, with autofocus carried out prior to each image.Actuators were monitored under constant illumination conditions and constant camera exposure and white balance for each cycle.Images were collected at the maximum camera resolution and saved without compression.Fiji (ImageJ) was used for image analysis.The active area of each CNT/Sn actuator was selected for pixel analysis, excluding the surrounding substrate.Weighted intensity histograms were extracted from RGB images using the formula, gray = 0.299red + 0.587green + 0.114blue.Linear intensity normalization and linear histogram stretching were performed on histograms using identical transformations from the same actuator or image stack to improve visualization and comparison.
Simulation: The FEA studies of the anisotropic expansion of the CNT-Sn composite cone-shape actuator were performed in COMSOL Multiphysics (version 6.0) using the linear elastic material-thermal expansion sub-node under Solid Mechanics modules to minimize complexity of the model and computation costs with a sufficiently accurate prediction of the relative degree of actuation.In the thermal expansion sub-node, the effective thermal strain of the structure is defined as In this model, the thermal strain was treated as a secant formula, and the coefficient of thermal expansion (CTE, ) was set as independent of the temperature variation.All the inputs (including the geometric parameters) required by COMSOL for solving the model are summarized in Table S3, Supporting Information.For the details of boundary conditions refer to Supporting Information (Figure S16, Supporting Information).

Figure 1 .
Figure 1.Actuation mechanism and wafer-scale fabrication of 3D microstructures: a) Schematic of the electrochemical alloying reaction of Li-Sn and resulting volumetric expansion at nano/microscale.This non-volatile and reversible process is controlled by simple electrical current.b) Schematic of the different steps in fabrication process of the CNT/Sn 3D microstructures at nano and micro scales (top rows) with their corresponding SEM images (lower rows).c) The SEM image of the 3D microstructures arrays fabricated by CVD growth of vertically aligned CNT forests with complex geometries from 2D patterns.(c, inset) CNT forests with varying heights ranging from microns to millimeter with growing speed of ≈100 μm min −1 .The SEM imaging is carried out at 45°tilt angle; scale bars: (b, top row), 50 μm; (b, lower row), 2 μm; (c) 200 μm; (c, inset), 100 μm.

Figure 2 .
Figure 2. Role of the geometrical factors in actuation of 3D microstructures: SEM images of 3D microstructures with various patterns in their walls before and after lithiations are compared with the FEA simulation of the overall 3D displacement.a-c) The 3D microstructures with periodic lozenge lattices (c) outperform those with the elongated hexagonal lattices (b) and monolithic cylindrical structures (a).d) Actuation (% change in open end area) measured after lithiation of 3D microstructures with various patterns in their walls.After lithiation, the representative wall thickness of lozenge lattices in the CNT/Sn microstructures shown in Figure 2c increases by ≈2.24 ± 0.26 μm.SEM imaging was carried out at 45°tilt angle.Scale bars: 50 μm; inset figure 10 μm.

Figure 3 .
Figure 3.In situ optical visualization and electrochemical analysis of 3D during actuations.a) Voltage-capacity profile measured progressive actuation (% change in open end area) of the 3D microstructure at corresponding capacity.Transition from 0.45 to 0.35 V corresponds to LiSn to Li 2.5 Sn which has the most significant contribution in the actuation of 3D microstructures.b) CV at a scanning rate of 0.1 mV s −1 between 0.05 and 1 V. c) C-rates performance of the 3D microstructures with corresponding degree of actuation (% change in open end area; cutoff voltages of 0.2 and 0.9 V). d) The correlation between the coulombic efficiency (CE) and reversibility of geometrical transformations upon lithiation in each cycle.e,f) SEM images of 3D microstructures (as synthesized and after more than 70 cycles).SEM imaging was carried out at 45°tilt angle.Scale bars: 50 μm; inset figure 10 μm.

Figure 4 .
Figure 4. Analysis of behavior of 3D microstructures during actuations.a) field micrographs of four neighboring actuators during the formation cycle displaying the electrochromic response upon lithiation of pristine Sn/CNT structures; b) weighted intensity histograms from RGB images of Actuator 1 in (a) before and after lithiation in the formation cycle; c) average modal pixel intensity values for the formation cycle of Actuators 1-4 shown in (a) under the same imaging and cycling conditions, where error bars represent standard deviation; d) composite image of a single actuator during the 3rd cycle showing reversible change in appearance upon cycling; and e) weighted intensity histograms from RGB images of the actuator in (d) displaying a clear difference in de-lithiated and lithiated intensity values and reversibility of this shift in intensity.Scale bars: 50 μm.