Silicone membrane casting
Once the silicone membranes are released from the PET substrate and are freestanding on a frame (at the end of step 2.2), their thickness can be measured, for example by transmission interferometry. Figure 6 shows the thickness homogeneity of a silicone layer across the width of the 200 mm PET substrate for 3 different effective gap heights (50, 100 and 150 µm) at a casting speed of 1 mm/sec (note that because the applicator is wider than the PET substrate, the feet of the applicator rest on the vacuum and not on the PET substrate itself, as can be seen in Figure 4A. The effective gap between the applicator and the substrate is therefore equal to the applicator height minus the thickness of the PET substrate. For example a PET substrate of 125 µm and an applicator height of 225 µm, as used in the protocol, leads to an effective gap of 100 µm). For the 50 µm effective gap height, there is a clear height difference between the left and right side of the silicone layer. This is because the height of the applicator must be set manually on the left and right side, and some error is unavoidable. However with careful setting of the applicator, we generally obtain membranes with a thickness standard deviation of less than 1 µm, which is the case for the 100 µm effective gap height (σ=0.81 µm). When the applicator height becomes too large, waviness starts to appear on the membrane, caused by the evaporation of the solvent in the silicone mixture, as is visible in the membrane cast with an effective gap of 150 µm (Figure 6).
The relation between the obtained dry film thickness and the applicator height depends on the silicone mixture and the speed of casting. The silicone mixture used in this article consists of a 2-parts silicone, and a solvent to decrease the viscosity of the mixture. As the solvent evaporates from the membrane before curing, an estimate of the film thickness can be obtained by multiplying the effective gap height by the volume fraction of solids in the silicone mixture. However, there are dynamic effects at the trailing of the applicator, leading the creation of a meniscus and a thinner thickness than expected. The relation between the gap height and the resulting dry membrane thickness depends on the casting speed, applicator height, and by the applicator shape. Figure 7 shows the results of an experiment where membranes were cast at different speeds and heights to show how these parameters affect dry film thickness. It can be seen that casting at high speeds leads to thinner membranes, and that the effect of speed becomes more pronounced as the gap height increases.
Actuation Performance
The actuator fabricated here is characterized by measuring the outside diameter of the cog-like electrode as a function of applied voltage. A camera on a fixed stand is used to take pictures of the actuator as the voltage is increased. The images are analyzed with an image processing script (Vision, National Instruments) to quantify the expansion of the actuator. This was done by fitting a circle to the outside perimeter of the cog like electrode (Figure 8). The increase in diameter of the circle from the relaxed state is presented as diametral stretch (i.e., the actuated diameter divided by the diameter of the actuator when relaxed). The results of two separate actuators of identical thickness (34.5 µm) are shown in Figure 8. Both devices perform similarly with diametral stretch of 10% at an actuation voltage of 4 kV.
The response speed of the actuator was measured by applying a 2 Hz square signal of 3 kV, leading to a strain of about 4%. The expansion of the actuator was filmed with a high-speed camera with a time resolution of 0.25 msec. A rising edge was captured, with 200 frames (50 msec) before the voltage trigger, and 200 frames after. The images where then analyzed to extract the time-dependent deformation (Figure 9). The rise time (time needed to reach 90% of the final deformation) is 3.75 msec, and there is no observable viscoelastic creep before and after the voltage step, unlike what is observed when acrylic elastomers are used as membranes, for which rise times of several hundreds of seconds are usually observed12.
Application of the process flow to other devices
The actuator manufactured in this article demonstrates our fabrication process, as well as the basic functioning principle of a DEA with an increase in surface area of the electrodes upon application of a voltage, and is therefore a good illustration for this tutorial. However, this actuator has no specific purpose other than demonstrating the actuation principle of a DEA. Nevertheless, the process presented here is very versatile and can be used to manufacture a wide variety of soft transducers aimed at specific applications. We present here a few selected examples of applications that we developed based on actuators fabricated using the presented methodology.
Soft bio-inspired tunable lenses have been fabricated (Figure 10A). These are capable of changing focal length by 20% in less than 200 µsec9. The device can be actuated for more than 400 million cycles without noticeable decrease in actuation performance, which shows that the combination of adequate materials and good fabrication processes result in DEAs with fast response speeds and long lifetimes. Lenses of similar geometry but made using the widely used commercial acrylic elastomer VHB have a bandwidth more than 3 orders of magnitude smaller9.
Patterning the compliant electrodes with pad printing allows making very precisely-defined electrodes, thus enabling the fabrication of independent small-scale electrodes on the same membrane. This is for example demonstrated through the fabrication of a DEA-based rotary motor comprising three electrically independent electrodes (Figure 10B). The axis and proof mass at the center of the motor can spin at 1,500 rpm13. The motor concept has been pushed further to show that pad printing can also produce reliable actuators. A self-commutating rolling robot was built to run laps along a circular track (Figure 10C). The robot travelled more than 25 km at an average speed of 15 cm/sec13.
Other applications that have been produced with the present process (or slight variations thereof) include deformable cell culture systems14, Dielectric elastomer generators15, multi-segment soft grippers16, or tunable mm-wave radio frequency phase shifters17.

Figure 1. Basic principle of dielectric elastomer actuators. Top: (1A) In its most simple form, a DEA consists of a soft elastomer membrane sandwiched between two compliant electrodes. (1B) When a DC voltage is applied between the electrodes, the electrostatic charges brought on the electrodes create a compressive stress that squeezes the membrane, leading to a thickness reduction and a surface expansion. Bottom: (2A) the actuator described in the protocol consists of a membrane stretched on a frame. Circular electrodes are on either side of the membrane with extensions to the border of the membrane to allow for electrical connections. The active area is the zone where the two electrodes overlap, i.e., the circle at the center. (2B) When a voltage is applied, the electrostatic force compresses the membrane. This causes a decrease of the membrane thickness in the active area, and an increase of the surface of the electrode. Because the membrane is prestretched, the passive zone around the electrode relaxes to accommodate the expansion of the central active region. Please click here to view a larger version of this figure.

Figure 2. Demonstrator actuator fabricated in this protocol. Left: finished device comprising a stretched silicone membrane fixed on a frame, a pair of compliant electrode patterned on both sides of the membrane, and electrical connections. Right: composite picture showing the rest state (black) and activated state (cyan). A 10% increase in the diameter of the structure is observed with 4 kV applied across the electrodes. Please click here to view a larger version of this figure.

Figure 3. Exploded view of the actuator. The different components that form the actuator manufactured in the video. The membrane holder retains the prestretched silicone membrane and is used to manipulate the membrane during the electrode printing step. Once the electrodes are cured, the actuator frame is inserted inside the membrane holder and provides both a structural frame to hold the actuator, and an electrical contact to the bottom electrode. Once the membrane is fixed to the actuator frame, the membrane holder can be removed. Please click here to view a larger version of this figure.

Figure 4. Overview of fabrication process. (A) Casting of silicone membranes using an automatic film applicator coater. (B) Laser cutting of cured silicone membrane and prestretch supports. (C) Placing of silicone membrane on prestretch support. (D) Releasing of silicone membrane from PET substrate by dissolving of PAA sacrificial layer in hot water. (E) Cutting of prestretch support sections linking the fingers. (F) Prestretch and adhering of membrane holder to the membrane surface. (G) Cliché filled with conductive ink. (H) Laser etched electrode aligner, inset figure shows example of a well-aligned electrode.(I) Silicone membrane with stamped electrode. (J) Finished device. Please click here to view a larger version of this figure.

Figure 5. Operating principle of the membrane prestretcher. (A) Several metal fingers are attached to a plastic annulus and are constrained to move in a linear (radial) fashion along their length. The annulus is constrained to move circumferentially. The plastic annulus has several curved slots machined into it, into which the metal pins of the fingers reside. Radius of a circle bounding the edge of the fingers is R1. (B) The prestretcher annulus is rotated counter-clockwise, the fingers translate simultaneously, increasing the radius of the circle bounding the finger edges from R1 to R2. Please click here to view a larger version of this figure.

Figure 6. Thickness homogeneity of the casted silicone layers. Thickness measurement of the cured silicone membrane across the width of the 200 mm PET substrate, for three different gap settings of the applicator. The casting speed is 1 mm/sec. Please click here to view a larger version of this figure.

Figure 7. Dry film thickness as a function of casting parameters. Dry film thickness obtained for different applicator heights and speed for a silicone-solvent mixture with 62% solid content by volume. A higher speed leads to thinner membranes for equal applicator settings, and the influence of speed increases with increasing membrane thickness. Please click here to view a larger version of this figure.

Figure 8. Actuation of the demonstrator. Outside diametral stretch as a function of the applied voltage for two devices with a thickness (after prestretch) of 34.5 µm. An increase of diameter of about 10% is observed at the maximal applied voltage. Please click here to view a larger version of this figure.

Figure 9. Strain response to a voltage step input. A square, 3 kV 2 Hz signal is applied to the device, generating a strain of about 4% (see Figure 8). The area expansion is observed with a high speed camera at 4,000 frame per second. It takes less than 4 msec for the actuator to reach 90% of its final dimension. Before and after the transition, the dimension of the actuator remain stable and do not show viscoelastic creep. Please click here to view a larger version of this figure.

Figure 10. Dielectric elastomer actuators made with the presented process flow. Three examples of dielectric elastomer actuators made by following the methodology described in this document. (A) Fast and soft tunable lens capable of changing its focal length by 20% in less than 200 µsec. (B) Rotary elastomer micro-motor capable of spinning at 1,500 rpm. (C) Self-commutating rolling robot. Please click here to view a larger version of this figure.