The display consists of two ITO coated glass plates, separated by 0.23 mm, with the droplets sandwiched between them. The glass is carved with the laser and broken to achieve the octagonal contour. On the bottom plate the 60 actuation electrodes are arranged, every third segment is connected with each other, so we have three sets of electrodes. On the top plate, the three segments of the counter electrodes are used to address the droplets. I made a first approach for the top plate with only two sets of electrodes, every second connected with each other. There for I tried many different shapes for the electrodes, to force the movement of the droplets in one direction. If I would have succeeded, I could have avoided the 20 conducting bridges over the dielectric – but I didn't. The whole design is made with LibreCAD and the .dxf file is imported into LaserCad. The structuring of the ITO coating by laser ablation is a very stable process. The isolation trenches are just a few micrometres wide!
The Bottom Plate
Next step is the application of the dielectric. The dielectric is a composite material. First a thin duroplastic foil is hot laminated and afterwards, after removing its thermoplastic cover layer with acetone, cured in the oven at 140°C. On top of this a very thin layer of silicone-modified alkyd resin is spin coated and subsequently cured at 100 °C in the oven. Last step is the spin coating with Fluoropel PFC1601V, to achieve hydrophobicity, which is again thermally cured. The blue watch dial on the dielectric is made of photosensitive epoxy, mixed with blue pigments and exposed with the laser. The 20 conducting bridges over the dielectric are drawn with conductive silver paint and finally sealed with silicone-modified alkyd resin. On the dielectric are two fiducials for alignment purposes with the top plate.
This dielectric is absolutely pin hole free and has an over all thickness of 5 micrometres.
The variation of the contact angle is determined by the Lippmann-Young equation:
cos θ (U) = cos θ0 + ε0 εr U2 : 2 γ d
θ0 is the contact angle without electric field
ε0 is the permittivity of vacuum
εr is the dielectric constant of the dielectric
γ is the liquid surface tension of the droplet's mediun
d is the thickness of the dielectric
U is the actuation voltage.
As it is the objective, to keep the actuation voltage as low as possible, to prevent the Fluoropel layer from charge trapping and degradation, low values for the liquid surface tension of the droplet and the thickness of the dielectric are needed. As liquid for the droplets I chose a mixture of ink (1/3) (unknown surface tension) and mono-ethylene glycol (2/3) with a surface tension of
48 mN/m (against air), what can be considered as low for a polar liquid (dipole moment 2.27 D). Mono-ethylene glycol was preferred over glycerol because of glycerol's much higher viscosity and higher surface tension of 65 mN/m. Bad experience was made with aqueous salt solutions due to their aggressive chloride ions. The thickness of the dielectric is 5 micrometres. I know, that the use of Parylene C as a dielectric with a usual thickness of ~ 1 micrometre would decrease the actuation voltage by 55% (assumed the same dielectric constant) but 1. I don't have access to the deposition process and 2. I would loose other advantages like the reflectivity. I don't know the value of the εr of my dielectric as it is a composite material but I suppose it to be > 3. I will make some future measurements for its determination.
The Top Plate
On the closeup of the top plate small round pillars can be seen, that have the task, to keep the droplets in their bistable positions. To produce them, the top plate is spin coated with a photosensitive epoxy with a viscosity of 7000 mPa·s (BEST KL6084). To further increase the viscosity, the resin is cooled down to 7°C before spin coating.
Subsequently follows the exposure process by laser and afterwards cleaning with ethanol and acetone. The inner pillars are 0.51 mm in diameter, the outer 0.70 mm.
Some trials were made with a complete superhydrophobic setup (top and bottom plate), but the result couldn't convince me: the movement of the droplets was choppy and after a while electrolysis started, probably due to a transitional state between Wenzel and Cassie–Baxter. I will continue with some experiments in the future!
Next trial was to limit the superhydrophobic coating to the top plate. The movement of the droplets worked sufficienly, but a new problem came up: I lost the influence of the counter electrodes for addressing the droplets. I think, it's because the hydrophobic coating in combination with a thin film of the filler liquid (“oil”) works as a small capacitor rather than a conductor in contrast to the thin fluoropolymer film, which is a poor insulator.
The solution was the selective, partial hydrophobic coating of trenches with Fluoropel in an superhydrophobic environment, covered with silica nanoparticles. These trenches are 0.5 mm in width, the outer droplet has a diameter of ~1.6 mm, the inner droplet of ~ 1.1 mm
On the closeups below, droplets and their corresponding contact angles can be seen on both surfaces. Please note, that the droplets consist of an (yellow) ink glycol mixture!
In case of water, the contact angle on the superhydrophobic surface would exceed 150°! It was impossible to place a water droplet because they rolled off with the slightest incline of the plate.
After fixation and alignment, the two plates are sealed, separated by a distance of 0.23 mm, with epoxy resin, except for a small opening. This opening is needed for the injection of the filler liquid and the droplets.
The filler liquid:
the liquid avoids the evaporation of the droplets and works as a lubrication that, reduces contact angle hysteresis. The medium has to be non polar and compatible with the other materials, be colorless and be a good insulator, of cause. The viscosity should be low, to allow fast movement of the droplets, as the medium has to be squeezed out beneath the droplets during wetting.
First liquid of choice was polydimethylsiloxane (“silicone oil”) with a viscosity of 20 mPa·s, later
5 mPa·s. Everything seemed ok. But after a couple of weeks, the droplets became smaller and darker: ingredients of the droplets seemed to diffuse into the silicone oil. The only feasible solution was to use a hydrocarbon instead. As n-dodecane was too expensive, I took some clear, odorless lamp oil from my shelf. I think, it's mix of n-alkanes with similar properties.
The problem is, that the droplets have a density of > 1 g/ml (glycol has 1.113 g/ml) and the n-alkanes have a density of ~ 0.75 g/ml. The silicone oil had a density of 0.92 g/ml, everything was almost in equilibrium. Now, with the low density of the medium, the droplets couldn't be held in place anymore, they slipped through the pillars by the force of gravity or other accelerations.
As a solution, I have increased the density of the medium by adding ~ 30% tetrachloromethane (density 1.594 g/ml), to make the droplets floating again. This solution seems to be working.
Update: The medium was replaced by a mixture of 50% silicone oil 20 mPa·s, 50% 5 mPa·s. This has been working without any problems for a year.
The injection of the droplets is a very difficile process, executed under the microscope, because a tiny amount of liquid has to be dosed and placed exactly through the narrow gap of 0.23 mm.
Great care has to be taken, not to damage the vulnerable surfaces.
For this purpose, I had to build a special droplet injector (details under “tools”).