Micropipette-Based Fabrication of Free-Standing, Conducting Polymer Bilayer Actuators

developed in this work opens exciting opportunities to manufacture advanced artificial muscles and sophisticated soft microrobotics


Introduction
Electrically driven actuators, which are capable of transforming an electrical input into mechanical movements, are extremely useful tools in biomimetic applications, such as soft robotics and artificial muscles, as well as in manipulating biological entities, such as tissues or tiny organisms. [1] Among the different types of actuating materials, conducting polymers (CPs) have attracted widespread research attention due to their unique doping-based actuation mechanism and the special properties, such as the large actuating strain, low operational voltages, ability to hold the strain at open circuit, and the tunable actuation performance through the choice of dopants and applied potentials. [2] Importantly, CPs are soft and are amenable to microfabrication and patterning. [1b,3] The typical actuation mechanism of CPs is the volume change that results from the ionic migration in/out of CPs when they are chemically or electrochemically oxidized or reduced. [2c,d,e,f,g,4] In the last few decades, since the actuation property of CPs was firstly discovered by Baughman and his colleagues, [5] the research efforts of CP actuators have been mainly focused on improving the actuation strain and stress, [6] optimizing the structural design (linear, bilayer, or trilayer), [7] and also exploring the combination with other electrochemically active materials such as carbide-derived carbon (CDC), [7c] graphene, [8] carbon, and organic semiconductor nanotubes (CNTs and OSNTs) [4a,7a,9] .
The miniaturization of devices or functional units (e.g. sensors or electrode probes) is often among the key requirements for bioelectronics researches. [3a,10] Actuators that can be fabricated down to the microscale are finding widespread applications in interacting with multicell tissues, single cells, or even organelles, in the manufacturing of microrobots that can be operated in vivo and in creating complex and delicate biomimetic devices. [1c] However, despite these promising prospects, the fabrication techniques for CP microactuators are limited. [9,11] Due to the multilayer structural requirements, only 2D patterning methods, such as photolithography and laser engraving, have been used. [1b,11d,e,12] For example, Svennersten et al. fabricated micropatterned polypyrrole (PPy) linear actuators through photolithography and Electrically driven actuators have found widespread applications in biomimetics and soft robotics. Among different actuation materials, conducting polymers (CPs) have stood out due to their unique doping-based actuation mechanism. Fabricating actuators at the microscale is particularly important, not only in the manufacturing of delicate biomimetic/robotic devices but also in advanced microphysiological studies. However, the choice for microfabrication techniques is limited, with the reported CP microactuators being mainly planar. To overcome this issue, a new micropipette-based method is developed for the fabrication of free-standing 3D CP actuators. The two-layer actuator consists of a layer of PPy:CF 3 SO 3 , fabricated by localized electropolymerization, and a layer of PEDOT:PSS, fabricated by a "direct writing" technique. Due to the opposite contraction and expansion behavior of these two CPs, determined by the size of dopants, the electrically driven bending actuators have been demonstrated. This fabrication approach provides unprecedented capability for fabricating high aspect ratio microactuators: the 360° bending orientation of the actuators can be controlled by the relative position of the two layers. As a proof-of-principle, we demonstrate CP "microtweezers" and the manipulation of a PDMS microsphere. The technique developed in this work opens exciting opportunities to manufacture advanced artificial muscles and sophisticated soft microrobotics.
electropolymerization. [1c] These actuators were used for applying mechanical stimulations to individual epithelial cells via electrically driven poking or stretching. The intracellular Ca 2+ was monitored as the real-time response. An increased intracellular Ca 2+ level, caused by autocrine ATP signaling, could be recorded under the stimulation. [1c] Another example was demon strated by Maziz et al. in which microactuators of 45 µm width and 690 µm length were fabricated by photolithography and plasma etching. [11d] An interpenetrating polymer network (IPN) layer was used as the ion reservoir to enable actuation in air. The actuation frequency as high as 930 Hz was demonstrated due to the thin layer design. [11d] Indeed, photolithography is a highly useful and well-established microfabrication technique; however, the limitation to only 2D and low aspect ratio structures hinders its use in fabricating microactuators.
In our previous work, we demonstrated a meniscus-guided direct writing method for poly (3,4-ethylenedioxythiophene):po ly(styrenesulfonate) (PEDOT:PSS) micropillars. [13] The advantages of this technique include: 1) capability of fabricating free-standing, super-high aspect ratio CP structures, which is difficult to achieve with most of the other microfabrication techniques; 2) ability to use either "direct writing" of pre-made polymer inks or electropolymerization from a monomer ink, the latter being similar to reactive printing; 3) has a high fabrication resolution; 4) the flexibility and ease of customizing design.
Herein, we use the micropipette-based fabrication techniques of both direct writing of a CP and capillary-guided electropolymerization, to develop a new approach toward the fabrication of free-standing, high-aspect-ratio 3D CP bilayer actuators. The actuator consists of one layer of PEDOT, which is doped by the large dopant -PSS, and another layer of PPy, which is doped by the small dopant -CF 3 SO 3 − . Based on the opposite expansion/contraction behavior of the two layers caused by the different sizes of dopants, bending actuation was realized. We demonstrate actuating "micro-tweezers" and a microactuator manipulating a PDMS bead. This work brings exciting opportunities in the fabrication and applications of free-standing 3D microactuators.

Results and Discussion
The expansion and contraction behavior of CP actuators highly depends on the size of both dopants used during the electropolymerization and the ions in the electrolyte. [1b,d,7b] In the case of using smaller dopants (for electropolymerization) and smaller anions (in the electrolyte), such as Cl − , ClO 4 − , and PF 6 − , CPs expand in the oxidized state due to the ingress of the negatively charged counter anions (dopants). When such CPs are reduced, the counter anions are expelled out, thus the CPs shrink ( Figure 1A). [8] On the contrary, if the dopants (for electropolymerization) are large in sizes, such as dodecylbenzenesulfonate (DBS) or poly(styrenesulfonate) (PSS), they are trapped inside the CP matrix and become immobile. In this case, when the CP is reduced, these dopants cannot be expelled out, but rather counter-cations ingress into the CP to compensate the charge  [7b] This leads to expansion of the CP when it is reduced and contraction when it is oxidized ( Figure 1B). [7a] It is noteworthy that, even though the size of these mobile ions themselves (Cl − , ClO 4 − , PF 6 − , Na + , and Li + ) is small, they carry a solvation shell, which boosts the volumetric change of the CP actuators. [2e] Based on this opposite actuation behavior, which depends on the size of dopants, we designed our bilayer bending actuator as follows. The first actuating layer was PPy doped by a small dopant, CF 3 SO 3 − , while the second layer was PEDOT doped by a large dopant, PSS ( Figure 1C,D). The equations describing the ingress/egress of the ions in our actuation system are:

Oxidized Reduced
(1) CF 3 SO 3 − in methyl benzoate was chosen for the electropolymerization of PPy due to the large actuation strain and stress that have been reported for PPy with this combination of ion/ solvent. [14] In addition, CF 3 SO 3 − doped PPy showed advantageous mechanical properties over other dopants, such as PF 6 − , BF 4 − , ClO 4 − , and Cl − . [14] PEDOT:PSS is the most commonly used CP material, including its use as microelectrodes, due to its high conductivity [15] and being commercially available as an aqueous dispersion. Importantly, we have previously demonstrated the feasibility of PEDOT:PSS for direct writing of 3D PEDOT microstructures, [13b] which is the basis here for the fabrication of the microactuators.
In order to fabricate free-standing microactuators, the micropipette-based fabrication techniques previously explored by us, namely direct writing [13] and localized electropolymerization, [16] were employed. The schematic of the fabrication process is shown in Figure 1E-H.
The first step was the direct writing of the thin PEDOT:PSS pillars, which were 5 µm in diameter and 1500 µm in height, on Au substrates (Video S1, Supporting Information). The optical images showing the direct writing process, and the SEM images showing the printed PEDOT:PSS pillars, are presented in Figure 2 and Figure S4A,B (Supporting Information), respectively. These thin PEDOT pillars acted as the core electrodes in the following electropolymerization step.
As one end of the pillar was attached to the tip of the micropipette (through the meniscus) during writing, the vertical shape of the pillar was maintained during the drying. In addition, the solidified PEDOT:PSS possesses good mechanical properties, sufficient to support the weight of the pillar itself. The PEDOT:PSS pillars show good adhesion to the Au substrate, which we hypothesize is due to both the negative charge of the sulfonate group from PSS and the Van der Waals forces. [10c] Taking advantage of the above points, straight and vertically standing super-high-aspect ratio PEDOT:PSS pillars were fabricated, which is challenging for other fabrication techniques. Such a shape is critical for both the alignment of the capillary in the electropolymerization step and the direct writing of the second PEDOT:PSS layer discussed later. It is noteworthy that this method is capable of fabricating pillars of even higher aspect ratios of around 7 µm in diameter and 5000 µm in height, as we have reported previously. [13b] This creates more possibilities for the fabrication of CP actuators.
The second step in the fabrication of the microactuator is the electropolymerization of PPy:CF 3 SO 3 "sheath" that covers the fabricated thin PEDOT:PSS pillar (Video S2, Supporting Information). PEDOT:PSS is known to have a strong water absorption ability and can change its volume upon sorption and desorption of water. [17] As a result, once the PEDOT:PSS pillar was immersed into an aqueous electrolyte, it swelled and curved ( Figure S1, Supporting Information). Therefore, an organic solvent, methyl benzoate, which has been previously reported to promote the actuation strain of PPy:CF 3 SO 3 , [14a] was chosen in this work. In addition, compared to water, methyl benzoate is electrochemically stable in a wider potential range [18] and has lower evaporation rate, which was preferred.
A capillary-based setup was used for the electropolymerization ( Figure 1F). To be specific, the individual PEDOT:PSS pillars ( Figure 3A) were carefully aligned and immersed into the pyrrole monomer containing solution inside a capillary of 0.1 mm inner diameter ( Figure 3B). As mentioned above, the vertical shape of the PEDOT pillar was important for both a good alignment and visual observation (due to the refraction of light). A two electrodes setup was used with the PEDOT:PSS pillar as the working electrode (WE) and a Ag/AgCl wire inside the capillary as the counter/reference electrode. Importantly, a small gap of 20 µm was left between the bottom of the capillary and the substrate. This not only avoided the uncontrollable leaking of the solution once the pipette touched the substrate (a clear Au substrate with no PPy deposition is preferable for the fabrication of the second PEDOT:PSS layer), but also restricted the deposition of PPy to the PEDOT pillar only. The polymerization www.advmattechnol.de was accurately controlled by the total charge passed for electropolymerization of PPy. The current vs. time curve for the electropolymerization is shown in Figure S2 (Supporting Information). As shown in the optical images of Figure 3B,C, and SEM images in Figure S4C,D (Supporting Information), the diameter of the micropillar significantly increased after polymerization of PPy, from 5 to 12 µm. The gap between the PPy and the Au substrate is clearly visible in Figure 3D and Figure S4D (Supporting Information, SEM). The polymerized PPy:CF 3 SO 3 sheath on the PEDOT:PSS core acted as the first layer of the actuator.
The final step of the actuator fabrication is the direct writing of the second PEDOT:PSS layer (Video S3, Supporting Information). Different from fabricating the thin PEDOT:PSS core, a larger pipette of 50 µm tip inner diameter was used to match the size of the PPy layer. In addition, the PEDOT:PSS in this step formed a layer that adhered to one side of the PPy pillar. The high resolution, three-axis movement, controlled by the micropipette positioning system, enabled a precise alignment.
However, one issue associated with the fabrication of the second layer is the limitation on how close the pipette's tip can be brought to the base of a fabricated pillar (either approaching from the top or from the side). In other words, the tip of the pipette would touch the fabricated pillar before the tip could reach the desired location.
To overcome that issue, we relied on the exceptional mechanical property of PEDOT pillars (and especially the lower part of the initial PEDOT:PSS pillar that is connected to the Au substrate), where those pillars could be bent upon the mechanical force applied by the micropipette. This gave way for the tip of the micropipette and enabled the printing of the second PEDOT:PSS layer (Figure 4 and Figure S3 and Video S3, Supporting Information). Once the ink touched the bottom base of the first layer ( Figure 4B and Figure S3A, Supporting Information), the micropipette was raised up with a constant speed of 4 µm s −1 ( Figure 4C-E, Figure S3B,C, Supporting Information). Benefiting from the good hydrophilicity of Py:CF 3 SO 3 , the ink adhered well to the PPy layer during the printing ( Figure S3A,B, Supporting Information). As a result, upon drying, intimate contact ( Figure 4F) was established between the second layer of PEDOT:PSS and the PPy:CF 3 SO 3 layer, which was expected  to be critical for a good actuation performance. SEM images showing the bilayer structure of the actuator are presented in Figure 4F and Figure S4G (Supporting Information). The evaporation of solvent caused shrinkage of the PEDOT layer, and the fully dried actuator bent toward the PEDOT:PSS side ( Figure S4E,F, Supporting Information). In general, the thicker the second PEDOT:PSS layer was, the more bending it caused. However, once the actuator pillars were immersed into the electrolyte, the PEDOT:PSS side would partly swell back due to the hygroscopicity of PSS.
Before testing the actuation property of the bilayer pillars, cyclic voltammetry was used to characterize the oxidation and reduction properties of PEDOT:PSS and PPy:CF 3 SO 3 . Individual, pristine PEDOT:PSS pillars or PEDOT pillars with the PPy sheath were immersed into the electrolyte-containing capillary, while a small gap was maintained between the bottom of the capillary and the Au substrate. In this way, again, the interference from the Au electrode could be excluded without the need for an extra insulating layer. The cyclic voltammogram (CV) results are shown in Figure 5A. The oxidation peak of PEDOT/PPy pillar was at around +0.1 V (vs. Ag/AgCl), while the reduction peak was not that distinctive, ranging from −0.4 to −0.5 V. In terms of pristine PEDOT:PSS, no clear oxidation or reduction peaks could be seen. The different amplitude in redox currents is attributed to the different amounts of redox polymers in the pillars, and possibly, the difference in electroactive surface areas.
For the electrochemical actuation test, LiPF 6 aqueous solution was used. The relatively large ionic radius of PF 6 − , compared to BF 4 − , ClO 4 − , and Cl − , guaranteed a good actuation performance. [19] In addition, Li + cation's large hydration shell (larger than, e.g. of Na + or Cs + ), has been shown to be beneficial to the actuation of PEDOT:PSS. [2e] The strain of PPy:CF 3 SO 3 coated pillar (without the second PEDOT:PSS layer) was tested during cyclic voltammetry scanned (10 and 50 mV s −1 ) in the potential range from −0.4 to +0.8 V (vs. Ag/AgCl), with the results shown in Figure 5B. An average of 3-4% strain was measured for the PPy coated pillar, and no statistical difference could be seen between 10 and 50 mV s −1 scan rates. The linear actuation behavior of the PEDOT:PSS pillars with the PPy sheath is shown in Video S4 (Supporting Information).
For the actuation of the bilayer actuator, as expected, when a positive potential was applied, the PEDOT:PSS layer contracted while the PPy:CF 3 SO 3 layer expanded, leading to the bending toward the PEDOT:PSS side. On the contrary, when a negative potential was applied, the PEDOT:PSS layer expanded and the PPy:CF 3 SO 3 layer contracted, leading to the bending of the actuator toward the PPy side ( Figure 5G,H). Video S5 (Supporting Information) shows an example of the bending actuator.
A comparison of the bending displacement (measured as the change in the position of the actuating pillar's "head") against different scan rates (0.05-0.4 V s −1 ) is shown in Figure 5C. The position of each actuator's base was defined as "zero,"

www.advmattechnol.de
and bending left and right were defined as negative and positive, respectively. With lower scan rates, the actuator demonstrated a higher extent of displacement. For example, at 0.8 V, the displacement under 0.05 V s −1 scan rate was 212 µm, and decreased to 125, 81, and 25 µm for the scan rates of 0.1, 0.2, and 0.4 V s −1 , respectively. Under higher scan rates (0.2 and 0.4 V s −1 ), a "delay" in reaching the maximum actuation could be noticed, where the bending displacement kept on increasing when the potential was scanned back from 0.8 to 0.7 V. We suggest such a "delay" could be attributed to the slow solvatedions migration in the 'bulk', especially at high scan rates, that induce slow-moving oxidation front in the CPs. Under negative potentials, when the CPs were reduced, no significant difference could be seen with different scan rates.
We were unable to measure the generated actuation force due to the small size of the actuator and the pillar's high aspect ratio format. A cantilever set up, such as with a long and thin Au/Pt wire, could be a possible solution; however, that was not persuaded further in this work.
For the stability test, 100 repeated CV cycles (50 mV s −1 ) were performed. An example of the recorded displacement is shown in Figure 5D and the statistical data (n = 3) of the normalized displacement is shown in Figure 5E. The overall bending displacement gradually decreased with an increased number of cycles; at the 40 th and 100 th cycle, approximately 65% and 35% of the initial displacement remained, respectively. The stability data of these three individual bending actuators is included in Figure S5A (Supporting Information). There might be several possible reasons for the decreased actuation performance: 1) The opposite actuation behaviour of PEDOT:PSS and PPy:CF 3 SO 3 can result in the mechanical separation of these two actuating CP layers. Similar may happen with the PEDOT:PSS core and PPy sheath. 2) PPy:CF 3 SO 3 has been reported to generate relatively large strain and stresses when actuated. [6b,c] Such stress may stretch the PEDOT:PSS layer, which was in a soft hydrogel format, and cause unrecoverable deformation of the PEDOT:PSS. An example of the separation between CPs layers and the unrecovered deformation of the PEDOT:PSS layer is shown in Figure S5 (Supporting Information). 3) There could be an gradual ion exchange occurring in PPy, where larger CF 3 SO 3 − ions (0.270 nm [19] ) are progressively replaced by smaller water solvated PF 6 − (0.254nm [19] ) ions with more scanning cycles.
The use of a cross-linker (such as GOPS) in the PEDOT:PSS layer to enhance its mechanical stability, the use of an alternative, more elastomeric core material instead of PEDOT:PSS, and the use of the same solvent and electrolyte for the PPy polymerisation and actuation, are possible routes to improving the actuation stability.
The optical images in Figure 5F,G show the bending of the actuator under −0.4 and +0.8 V, respectively. The higher magnification images in the inset show the colour change of the PEDOT:PSS layer from dark blue to light blue when switched from reduced to oxidized state, demonstrating the successful oxidation and reduction of PEDOT:PSS. [20] It is noteworthy that instead of immersing the whole substrate into the electrolyte, these actuation tests were possible to carry out using an electrolyte-containing capillary (Video S6, Supporting Information). Actuators that were just taken out from the electrolyte could still operate in the air with the residue electrolyte (Video S6, Supporting Information). These bring more possibilities and flexibilities for the future applications of these bilayer actuators (e.g. fabrication of an extra layer of hydrogel electrolyte for actuators that need to be operational in the air [7a] ).
One significant advantage of the described fabrication technique is its high flexibility with respect to the design of the actuator. For example, as is shown in Figure 6 and Figure 1H, if the relative positions of the two layers change, the bending orientation of the actuators would change accordingly. In addition, a number of free-standing actuators of the same or different designs could be fabricated on the same Au-coated substrate, meaning only one electrical input could be able to drive all of the actuators simultaneously.
Making use of this principle, we demonstrated two pairs of electrically driven micro-tweezers, MT1 ( Figure 6A-C) and MT2 ( Figure 6D-F). The difference between them is that both the PPy:CF 3 SO 3 and PEDOT:PSS layers were thinner in MT1 than that of MT2 (refer to Experimental Section for more details). Their actuation performance was compared and is shown in Figure 6C,F. The MT1 demonstrated the bending displacement of −84 to 330 µm (414 µm in total) for the left arm and 169 µm to −281 µm (450 µm in total) for the right arm ( Figure 6C). MT2 showed an increased displacement of 135 µm to 880 µm (745 µm in total) for the left arm and 105 µm to −610 µm (715 µm in total) for the right arm ( Figure 6F, Video S7, Supporting Information). When inside an aqueous electrolyte, the PEDOT:PSS absorbs water and turns into a soft hydrogel format. Therefore, we hypothesize that the bending performance of the actuator was dominated by the actuation of the PPy layer. With thicker layers, a higher force could be generated by the PPy, leading to an extended stretch of the PEDOT:PSS core inside. As a result, a higher strain and higher bending displacement could be generated.
To demonstrate an application of the micro-tweezers, PDMS microspheres of ≈400 µm diameter have been fabricated as a model object. The schematic illustration of the fabrication setup is shown in Figure S6 (Supporting Information). The detailed experimental setup and the fabrication process of PDMS microspheres are included in the Experimental Section. The density of PDMS is lower than that of the electrolyte (1 m LiPF 6 ), but due to the adhesiveness of PDMS the microspheres adhered well to the glass substrate. The micro-tweezer was used to manipulate a PDMS microsphere. As is shown in Figure 7 and in Video S8 (Supporting Information), under the bending force of the actuator, the PDMS microsphere slightly rolled on the substrate ( Figure 7C,D, Supporting Information) and eventually detached to floating ( Figure 7E,F).
A griping of the PDMS microsphere was also attempted, but unfortunately, the griping force generated by the micro-tweezer was not sufficient for holding the microsphere in place, in addition to the nonideal alignment of the two arms. We hypothesize that the small force generated by tweezers is due to the small diameter and the high aspect ratio of the actuating pillars. This and the alignment issue could be addressed by using a three or four-arm setup with shorter arms; however, that was not attempted here. www.advmattechnol.de

Conclusion
In conclusion, a novel way of microfabrication of free-standing 3D CP bilayer microactuators is developed. The actuation mechanism is based on the opposite expansion/contraction behaviors of CPs with large and small dopants when driven with the same potential. The localized and high-resolution fabrication of both the PEDOT:PSS and PPy:CF 3 SO 3 was realized using a combination of our recently developed micropipette/capillary-based techniques, namely direct writing and localized electropolymerization printing in 3D. Importantly, this method is capable of fabricating super-high aspect ratio 3D microactuators with their bending orientation easily customizable. Taking advantage of this methodology, the electrically driven micro-tweezers have been demonstrated, with the maximum displacement of the arm reaching 745 µm under the potential difference between −0.4 to +0.8 V. The manipulating of PDMS microspheres with the fabricated actuators was demonstrated. The methodology developed here would not only address the limitations of the planar microactuator fabrication techniques (e.g. photolithography), but also create new opportunities in artificial muscles and microsoft robotics applications fields. Future studies could involve further optimization of the fabrication process, the improvement of actuation performance through optimization of dopants, for example, and the exploration of new applications, such as implantable microactuators inside blood vessels.
Direct Writing of PEDOT:PSS Micropillar: The "direct writing" of PEDOT:PSS micropillars was performed as previously reported by us. [13] In brief, the "ink" was prepared by mixing the PEDOT:PSS aqueous dispersion with dimethyl sulfoxide (DMSO) at a volume ratio of 20 (PEDOT:PSS) to 7 (DMSO). The ink was filled into a micropipette with a tip diameter of 20 µm, which was fabricated from single-barrel capillaries using the P-2000 laser puller (Sutter Instrument). After that, the micropipette was mounted onto a micropipette positioning system (Figure 8A,B) and was precisely controlled to approach the gold substrate. Upon establishing the contact between the meniscus of the ink and the substrate, the micropipette was raised upwards with a constant speed of 2 µm s −1 . This slow micropipette withdrawing process enabled the evaporation of solvents and led to the gradual solidifying of the CP micropillar. After reaching the set height of 1500 µm, a quick retreat of the micropipette terminated the printing.
Electropolymerization of the PPy:CF 3 SO 3 "Sheath": The localized electropolymerization of the PPy:CF 3 SO 3 sheath was carried out by dipping the PEDOT:PSS micropillar into a borosilicate capillary that contained 0.5 m Py and 0.75 m TBACF 3 SO 3 in methyl benzoate. A Ag/ AgCl wire was inserted into the capillary as the counter/reference www.advmattechnol.de electrode (CE/RE), while the PEDOT:PSS pillar acted as the working electrode (WE). A constant bias potential of 1.5 V was applied between the two electrodes for the electropolymerization of PPy:CF 3 SO 3 until a charge of 300 or 500 µC was reached. Importantly, a small gap of ≈20 µm was left between the bottom of the capillary and the substrate to confine the deposition of PPy to the PEDOT:PSS electrode only (rather than on the Au substrate). A significant increase in the diameter of the pillar could be noticed after the polymerization. This PPy:CF 3 SO 3 coated pillar acted as the first layer of the actuator.
Fabrication of the Second PEDOT:PSS Layer: The second PEDOT:PSS layer of the actuator was fabricated via direct writing of that layer next to the PPy:CF 3 SO 3 pillar, using a micropipette of 50 µm/70 µm inner diameter. Benefited from the good mechanical property (soft and elastic) of CPs, the electropolymerized pillar could be bent under the mechanical pushing force of the pipette's tip. This bending enabled the direct contact of the ink to the base of the pillar. Upon pulling the pipette upwards with a constant speed of 4 µm s −1 , the second PEDOT:PSS layer was formed.
The fabrication of the "micro-tweezers" was achieved by changing the relative position of two CP layers. The two microactuator arms of the micro-tweezers demonstrated in this work had their PPy:CF 3 SO 3 layers facing out and the PEDOT:PSS layers inwards. In other words, the PEDOT:PSS layers of the two arms were facing each other ( Figure 1H). Two pairs of micro-tweezers were fabricated and named as MT1 and MT2, respectively. For MT1, a 300 µC charge was used for the electropolymerization of PPy:CF 3 SO 3 , and a micropipette of 50 µm inner diameter was used for the direct writing of the second PEDOT:PSS layer. In the case of MT2, the charge used for electropolymerization was 500 µC, and tip inner diameter for fabricating the second PEDOT:PSS layer was 70 µm.
Electrochemical Characterization: To understand the oxidation and reduction properties of PEDOT:PSS and PPy:CF 3 SO 3 , the CVs of pristine PEDOT:PSS pillar and PPy:CF 3 SO 3 coated PEDOT:PSS pillar were measured in 1 m LiPF 6 . To obtain these CVs, the potential was scanned between −0.7 to 0.7 V (vs. Ag/AgCl) with a scan rate of 0.1 V s −1 . Similar to the electropolymerization setup, individual PEDOT:PSS or PPy:CF 3 SO 3 coated pillars were immersed into the electrolyte-containing capillary, while a small gap was maintained between the bottom of the capillary and the Au substrate. In this way, the interference from the Au electrode could be excluded without the need for an extra insulating layer.
Actuation Test: The schematic illustration of the actuation test setup is shown in Figure 8C. A copper wire was soldered to the Au substrate for the connection and both the Cu wire and the solder (tin/lead) were insulated. For the actuation test, the bilayer actuators, together with the Au substrate, were immersed into 1 m LiPF 6 aqueous solution. The cyclic voltammetry was performed in the potential range of −0.4 to +0.8 V (vs. Ag/AgCl), with different scan rates, to drive the actuator. A three electrodes setup was used with a Pt coil as the counter electrode and a Ag/AgCl (3 m NaCl) electrode as the reference. The actuation was quantified by measuring the displacement of the tip of the actuator, which was recorded by a CMOS camera placed in the horizontal direction and perpendicular to the bending plane of the actuator. The linear strain of PPy:CF 3 SO 3 coated pillars was measured by immersing the PPy pillars into an electrolyte (1 m LiPF 6 ) containing capillary and recording the linear elongation (Video S4, Supporting Information), while applying the scanned potential between −0.4 to +0.8 V (vs. Ag/ AgCl). The strain was calculated as (L − L 0 )/L 0 × 100%.
Fabrication of PDMS Microspheres: The fabrication of PDMS microspheres was inspired by an oil-in-water emulsion preparation method previously reported. [13] Different from the reported infusionbased fabrication using two syringe pumps (one for introducing PDMS and the other one for PVA solution), a simplified method has been developed here using a single withdrawing pump ( Figure S6, Supporting Information). In brief, a micropipette of 150 µm tip diameter, carrying the mixed PDMS ink (SYLGARD 184, ratio of PDMS base to curing  www.advmattechnol.de agent of 10:1 with bubbles removed), was aligned and inserted into a borosilicate capillary (0.58 mm inner diameter). The connection between the micropipette and capillary was immersed into a 3 wt% poly(vinyl alcohol) (PVA) aqueous bath. The other end of the capillary was connected to a syringe via a silicone tube. Once the syringe started withdrawing, the PDMS ink was sucked out from the micropipette, forming a droplet at the pipette's tip. In the meantime, the PVA solution was sucked into the capillary through the small gap between the tip of the pipette and the opening of the capillary. The shear force from the viscous PVA flow pushed the PDMS droplet down to the capillary, forming PDMS microspheres.
Manipulating of Individual PDMS Microsphere: A PDMS microsphere of ≈400 µm diameter was placed on a glass substrate before adding the electrolyte (1 m LiPF 6 ). The Au substrate (with the fabricated microtweezers on the substrate) was attached to a holder and placed upsidedown. The holder was mounted to the micropipette positioning system for controlling the movement of the Au electrode. Finally, the microtweezers were immersed into the electrolyte and aligned with the PDMS microsphere by the positioning system. Like the actuation test above, a three electrodes setup and Palmsens4 potentiostat were used to drive the actuation.
Statistical Analysis: The bending actuation of the bilayer actuator was quantified by measuring the displacement (D) of the actuator's tip. In the stability test, the displacements in the first scan cycle and subsequent cycles were defined as D 0 and D n , respectively. The stability of the bending actuation was normalized as (D n − D 0 )/D 0 × 100%. The presented data were calculated from three independent samples (n = 3).

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.