«UNIVERSITY OF CALIFORNIA Santa Barbara A Micro/Nano-Fabricated Gecko-Inspired Reversible Adhesive A Dissertation submitted in partial satisfaction of ...»
The oxidized wafers were then coated with Shipley SPR 220-7 photoresist (PR).
Prior to resist application, the wafers were soaked HMDS (hexamethyldisiloxane).
The HMDS acts as a wetting promoter for the PR on the silicon dioxide wafers by forming a thin layer. The compound is a type of ampiphile that binds nicely to the oxide surface and allows the PR to evenly wet the surface. After soaking the wafers with HMDS for 30 seconds, the wafer was spun at a speed of 3,500 RPM (3.5 kRPM). A large puddle of SPR 220-7 PR was then immediately placed in the center of the wafer, approximately 4 cm in diameter, and the wafer was spun again at 3.5 kRPM for 30 seconds. Here it was found useful to ramp the acceleration of the spinner, to avoid edge bounce and to reduce the required amount of PR for complete wafer coverage. The wafer was then removed from the chuck and placed on a hot plate at 95°C for 60 seconds. (This process should apply an approximately 7 μm thick layer of PR.) The PR was next patterned using projection lithography. Wafers were placed in the GCA AutoStep 200 i-line wafer stepper on the 100 mm wafer chuck. The desired mask was then loaded and aligned. However, since this is a single step process, alignment is not a critical issue. The i-line stepper was then programmed with a given step size and exposure time. An exposure time of 3.75 seconds was used. After exposure the wafers were allowed 10 minutes for the photoreaction mechanism to complete before any further processing was performed. Premature placement of the wafers on a hotplate, for a post exposure bake, would cause a strange reaction in the PR. However, in this work, a post exposure bake was not used and processing proceeded directly to development. Wafers were developed in MF-701 developer. Generally, for this portion of work, developing proceeded until it was visually apparent that the unwanted resist had been washed away. However, as rule of thumb, a developing time of 90” is sufficient (no harm was seen in extending this time to 120”). After developing, the wafers were hard-baked at 110°C for 90”.
The PR pattern was then transferred into the underlying silicon dioxide. Initially the pattern was transferred into oxide using the #3 reactive ion etcher in the nanofabrication facility. Etching of 2 μm of oxide would take over an hour and a half, with cleaning of the chamber required in between. Fortunately, the cleanroom acquired a Panasonic E640 inductively coupled plasma (ICP) system. Using this system it was possible to reduce the etch time to 14 minutes, while at the same time yielding near perfect sidewall verticality. After the PR pattern was transferred into the oxide, the PR was removed using an ultrasonic acetone bath, followed by isoproponal and a DI water rinse.
The crux of this particular process was then to extend the oxide pattern into the silicon and undercut the silicon dioxide structures by performing an extended release etch. The 4 inch wafer could be placed into the Plasma Therm DRIE ICP etch tool in the nanofabrication facility. The tool is dedicated to running the Bosch™ process.
The Bosch™ process is a method for etching high-aspect-ratio features into silicon.
The method is to cycle between an aggressive etch step, with SF6 plasma, and a passivation step using polymeric deposition, C4F8 plasma. Although the SF6 etch is an isotropic etchant, the directionality from the ion bombardment in the chamber gives slight preferentiality to removal of material normal to the ion bombardment.
The passivation step, however, has no directionality and evenly coats all surfaces.
Thus, by cycling the two, a net removal occurs in the vertical direction. By tailoring the etch and passivation times, more vertical and smoother sidewalls can be achieved, or faster and more aggressive etches can be performed. For this work the standard process was run to create the initial trenches. However, after the trenches were created, the passivation step was turned off and a sustained SF6 isotropic etch performed. By tailoring the duration of the isotropic etch, it was possible to control the amount of undercutting of the silicon dioxide platforms until only a single slender silicon pillar was left supporting the platforms, Fig. III-1. It was also found that the shape of the platforms ultimately affected the pillar shape, see chapter IV, opening up the possibility for anisotropic pillar shapes and mechanical properties.
B. Multi-Scale Integrated Structures The silicon dioxide platforms opened up a very interesting 3-dimensional design space in microfabrication. However, the structures needed an application to drive the research. It was about this time that the gecko adhesive was starting to become of interest in the scientific and engineering communities. As discussed in the following chapters, it was clear that all the research aimed at mimicking the system was focused on the nanostructures, and NOT on the hierarchical structure. This hierarchical structure included flexible microstructures, Fig. II-2. After initial testing of the SHARPS structures, no adhesion was observed. Clearly, just as the nanostructures would require integration with microstructures to mimic the system, the converse was also true. Thus it was clear that the silicon dioxide platforms would need to be coated with a nanostructured surface.
Creating nanostructured surfaces presented quite a challenge, after all this was (and is) what most of the research towards mimicking the system was focused. Now the challenge was to create these structures on flexible platforms. A variety of techniques were tried, including: titania grass on the platforms (too stiff), electrodeposition in porous alumina and polycarbonate matrices, nanosphere lithography, and metal thin-film destabilization, to name a few of the more promising attempts. After a substantial amount of work, and failed attempts, the solution turned out to be quite serendipitous.
Figure III-3 Electron micrograph of the organic looking polymeric nanorods, ‘organorods,’ scale bar 10 μm.
Fabrication of the Multi Scale Integrated Structures (MICS) followed the same process flow of the SHARPS, except for one major change. Just as with the SHARPS, silicon dioxide platforms were formed by etching trenches and undercutting the platforms. However, after platform formation (without having removed the photoresist) an additional reactive ion etch was performed in the Panasonic ICP tool. The wafers were loaded into the etch chamber and exposed to an oxygen plasma at 40 sccm, 1 Pa and 300 W bias for 10 minutes. After which the photoresist was transformed into arrays of vertically aligned nanorods, Fig. III-3.
The nanorods were not perfectly cylindrical and were more organic looking nanorods, thus dubbed ‘organorods’ – much to the chagrin of organic chemists all around the world. In fact, conferring with polymer experts, with experience with this photoresist, the growth mechanism is not exactly certain. The proposed one, discussed in chapter VI, is that the dielectric polymer, when placed in an electric field gradient, is acted on by a force in the normal direction to the wafer. In addition, the oxygen plasma serves to reduce the surface tension of the resist (the stabilizing mechanism), allowing the field gradient force to dominate and for nanorods to grow in the vertical direction. It is also presumed that the reactivity of the oxygen deteriorates the nanorods, giving them the less uniform structure.
Figure III-4 Electron micrographs of SHARPS structures coated with organorods. Scale bars are 20 μm and 5 μm, left and right respectively.
Post organorod growth modification was achieved by fluorocarbon deposition using the passivation step in the PlasmaTherm dedicated Bosch process tool.
Running a 9 second deposition would deposit roughly 30 nm of fluorocarbon on the surface of the organorods. The hydrophobic coating switched the organorod surface from hydrophilic to highly-hydrophobic with a contact angle of 154°.
Figure III-5 (Top) Optical images of a water droplet on untreated (left) and treated (right) organorod surfaces. (Bottom) Corresponding electron micrographs of the structure, scale bars 2 μm.
C. Nickel Multi-Scale Integrated Structures Comparing the β-Keratin, E ~ 5 GPa, system found in the gecko to the silicon one, E ~ 150 GPa, presented here. There is a glaring difference in the material properties. So one would ask, is the silicon the right material system to use? Just as the gecko has relatively few structural materials to choose from (β-Keratin is used for everything from the sticky pads to the eye spectacle) there are few engineering materials that have the processing capabilities for creating functional microstructures at the micro and nano-scale. The processibility of silicon allows freedom in the structures that can be created and what other materials can be incorporated. The ultimate goal of this research was not to create another tape, or glue, but instead to create a new type of adhesive with controllable stickiness. The ability to switch adhesion on and off is what allows the gecko to run up, down and across walls. In order to accomplish this, the silicon system above was extended to include activated platforms capable of reversible conformational changes. To achieve this, the silicon platforms were replaced with ferromagnetic nickel paddles. This required an entire new process to be developed.
Silicon (100) wafers were prepped with HMDS, allowing the wafers to soak for 15 seconds and then spinning at 4 kRPM for 15 seconds. Following this, an image reversal photoresist (AZ 5214-IR) was spun onto the wafers at 4 kRPM for 30 seconds. The wafer was then placed on a hotplate at 105°C for 60”.
Photolithography was performed in the Karl-Suss MA-BA-6 mask/ bond aligner (MA6) with an exposure time of 5”. Proceeding exposure the wafers were baked at 115°C for 120” before a 60” flood exposure was performed in the MA6. The wafers were then developed in MF-701 developer for 120”. After a 120” dionized water rinse, the wafers were dried and descumed in an oxygen plasma (the Plasma Etching Systems PE-IIA) for 60” – to better promote metal adhesion in the subsequent step.
Next nickel was deposited into the vias created in the PR. Electrodeposited nickel on silicon develops large amounts of stress. This stress can cause delamination of the PR/nickel during the evaporation, causing the protective resist to peel up and nickel to deposit in unwanted areas. The maximum thickness of evaporated nickel was found to be 200 nm. For the evaporated nickel devices reported here, thicknesses were maintained at 150 nm. After nickel evaporation across the entire wafer in the CHA Muti-Wafer Metal Evaporator (electron beam evaporator #4 in the nanofabrication facility) the photoresist was removed, lifting off nickel not deposited through the vias. This was achieved through an ultrasonic acetone bath for a time of 10 minutes (till the resist was completely lifted off).
For thicker nickel structures, an electrochemical deposition from solution technique was implemented. Electrodeposited nickel does not develop stress during deposition. However, in order to plate nickel, it was first necessary to plate the silicon wafer with a highly conductive seed layer. This conductive layer made it possible to transfer charge from the potentiostat probe across the entire wafer. Thus, before the photoresist was patterned, gold or platinum was evaporated across the entire bare silicon wafer at thicknesses of 25 or 50 nm, respectively. After seed layer deposition, a 25 nm layer of PECVD silicon dioxide was deposited to inhibit nickel undergrowth during electrodeposition. Then the wafer was patterned resist as above. Next the wafer was placed in a nickel electrodeposition bath composed of 200 g/L nickel sulfate, 5 g/L nickel chloride, 25 g/L boric acid, and 3 g/L saccharin (to sweeten the results). To facilitate uniform deposition across the wafer, and save money, a counter electrode was fabricated by coating a 100 mm silicon wafer with 150 nm of platinum. The two wafers were then placed facing each other in a wafer storage box at a distance of 5 slots apart. The bath was circulated by placing the wafer box on top of a magnetic stirring plate and a magnetic stir bar was placed in the bottom of the container. Using galvanostatic deposition, the current was regulated; values varied from 40 – 100 mA, and electrodeposition proceeded for 15minutes. The thickness of the deposited nickel was ultimately limited by the depth of the vias through the photoresist, determined by the original thickness of the photoresist. After deposition, the wafers were placed in an ultrasonic acetone bath, removing the photoresist and leaving nickel structures atop a thin layer of gold or platinum. Next the gold or platinum was removed. The gold was removed using a proprietary wet etch.
Although the electrodeposited nickel did not have the stress issues of evaporated nickel, achieving thickness across the entire 100 mm wafer proved to be non-trivial. The addition of wetting agents and mixing aided the uniformity; ~30% variance across 100 mm. However, achieving the precise uniformity of the evaporated nickel is unlikely, even in an industrial environment.
Following nickel deposition and definition on the silicon surface, another layer of photoresist was patterned on the surface in direct correlation with the nickel structures. Just as in the SHARPS and MICS processes, a 7 μm layer of photoresist (SPR 220-7) was spun on at 3.5 kRPM. The resist was baked at 95°C for 60” and patterned using contact lithography with an exposure of 20 seconds in the Suss MA6 contact alignment system. After exposure, the wafers were developed (MF-701 developer) for 90”, rinsed with deionized water and blown dry with nitrogen.
For the thicker electrodeposited nickel samples, similar deep etching and releasing as with the SHARPS and MICS could be performed. Etch times, as with all this work, varied greatly depending on the geometry of the structures. Typically a deep etch time of 25 minutes was used and extended SF6 release etches varied from 15 minutes to 45 minutes, longer times were required for higher density structures. Proceeding platform formation, the photoresist was then transformed into organorods by placing the samples in the Panasonic ICP with oxygen plasma as previously described. Additional processing could then be executed to coat the structures with a fluorocarbon. Typically the fluorocarbon coating was found to enhance adhesion.