Sol-Gel Coating Process
SolGel coating is a way to apply simple yet efficient antireflection (AR) coatings. This is particularly useful for large, irregularly shaped or delicate components, such as iodine cells or the cross-disperser prisms that we use for our echelle spectrographs, that are difficult or expensive to coat dielectrically. In the past, SolGel coatings have been used in astronomical settings to coat extremely large lenses.
Two methods for applying SolGel exist: spin coating, where the optical surface rotates while the SolGel is poured on, and dip coating, where the substrate is pulled from the liquid. We have set up a facility at Yale that allows us to use both methods for coating circular as well as irregularly shaped optics up to approximately 12 inches across and 30 pounds.
We follow closely the procedure described by Stilburn and credited to Thomas, with three exceptions. One, we harden for two days, two, we spin at a rather fast rate (about 500 RPM), and three, we consistently find a shrink rate of about 20 percent. The recipe to produce 100 grams of 1.5% stock solution combines 93.2 grams anhydrous denatured ethyl alcohol, 5.2 grams TEOS, and 1.6 grams Ammonium hydroxide (30%). The ingredients are mixed, shaken, and allowed to stand at room temperature for three days while the reaction is completed.
Spin coating is a process commonly used to apply photoresist chemicals on lithography substrates. The layer thickness is determined by the viscosity of the liquid and its wetting properties, and the centrifugal force from the rotational motion (and hence the speed).
We use a simple DC motor and a variable DC power supply to adjust the speed. A table or specific mounts are directly attached to the motor shaft. We use speeds between 250 and 750 RPM to spin our substrates. A handheld rev meter allows us to check on the correct speed, and reproduce it. As we use the motor at low voltages and torques, the speed (at the same voltage) varies with temperature and the weight/inertia of the table and optics used. The motor shaft does not run particularly true. While we haven’t seen any detrimental effect on the coating properties, it would be desirable to use a rotary table with ball bearings, and attach the motor via a belt.
Figure 1. The dip-coating rig with an Iodine cell mounted on the rotation table. In front are the hand-held rev meter, and syringes with filters for Ethanol and SolGel.
We measured the layer thickness as a function of spinning speed, using a 70% concentrated SolGel solution. The following plot shows the reflectance curve for various speeds.
Edge effects prevent a uniform film thickness over the full aperture of the substrate. The change is abrupt, not gradual. For a 2 inch BK7 substrate, 70% stock solution and a speed of 500 RPM, the rim has a width of about 3 mm. Higher spinning speeds make for a narrower region of inhomogeneity.
Figure 2. A 2 inch test substrate spinning at 500 RPM after the SolGel has been applied.
The cleanliness of the surface while the coating is applied is critical. Any small contaminations in the form of dust particles, for example, produce a radially extended blemish in the coating. This is true for particles settling on the surface prior to coating, as well as particles floating in the SolGel solution.
Figure 3. A spin-coated 2 inch BK7 disc. One can clearly see the uncoated rim. While the surface is very homogeneous, two defects stand out, with the right one showing what is sometimes called a "comet." This is characteristic of a dust particle on the surface before the liquid SolGel was applied.
We had severe problems in producing a meticulous coating over apertures of 4 inch or larger. The use of a (relatively) clean room and a laminar airflow is the only effective way that we have found to prevent dust from settling on the surface before applying the SolGel. Further, the optical surface has to be cleaned directly before coating, preferably on the rotary table. One way to clean the substrate is to spray it with ethanol and blow the liquid off with a filtered compressed gas and nozzle, all while the substrate is rotating. We use a compressed nitrogen bottle as gas supply, and filter the gas directly before the air gun with a disposable in-line gas filter. The use of filtered ethanol is advisable. We also successfully used “First Contact”, coating the surface with the polymer and removing it directly before coating. Obviously, the surface must also be free of residue of any sort. We use standard cleaning procedures, with substrate cleaner and ethanol, and lint-free wipes.
For irregularly shaped optics, spin coating leads to obvious problems with balancing the rotary table as well as inhomogeneous film thickness resulting from a non-circular (interrupted) substrate surface. When spinning a non-circular substrate there is a tendency to see inhomogeneities form along the leading edges of the spinning surface, which are thought to be caused by turbulence in the air as a result of the edge passing through (Spin Coating for Rectangular Substrates, Luurtsema, Gregory A., Thesis, University of California, Berkeley, 1997).
Figure 4. An 8 inch diameter window, cut to 6 inch width. We tried to spin coat this substrate. In this picture, only the top surface is coated. In the upper right corner, and also slightly in the lower left corner, one can see the inhomogeneity from the drag effect.
Dip coating is a viable alternative to overcome these problems. We use a Newport z-stage as a crane to pull the optical elements from a tank filled with SolGel. The necessary amount of SolGel can become very large for large optics and, in an effort to reduce this, we used tanks with a cross-section matching that of the optics. Dip coating is our preferred method for coating large cross-disperser prisms, which also was the motivation to set up a SolGel facility. The largest prism coated so far was the CHIRON cross-disperser, with a height of 158mm and an aperture length of 290mm. The weight of the prism exceeded the lifting capability of our crane, so we used a steel cable and counterweights to offset the weight of the prism.
Figure 5. On the left, the dip coating crane is set up in the CTIO laboratory in Chile, with the CHIRON cross disperser prism. The triangular tank for the prism has 20mm of clearance around the sides, and is made of 1/2 inch plexiglas. On the right is the crane with the Echelle vacuum enclosure window. The crane is a Newport z-Stage with 6 inches of travel.
Figure 6. Reflections of a lamp on the prism surface before [left] and after [right] sol-gel coating show more light reflecting off the uncoated surface.
Dip coating presents different challenges compared to spin coating. The liquid, especially the surface, must be kept very clean to prevent contamination of the coating. Further, the motion must be very smooth and the liquid at rest to prevent splashing or waves, which leads to inhomogeneous coatings. The homogeneity that we achieve is good, but not as good as for round, spin-coated substrates. Also, edge effects prevent the coating of the lowest 15mm of the aperture.
Figure 7. The coated vacuum window in its holder. The substrate has a wedge, hence the two reflections. We measured a reflectivity of <0.3 percent at 543nm and perpendicular incidence.
Figure 8. A dip coating test on plate glass. The substrate is 4 x 4 inches. The edge effects of the dipping process, and the difference in reflectivity between the coated area on the bottom of the plate and the uncoated part on the top are clearly visible.
For dip coating, we lower the optics into the SolGel, without stopping along the way. This is important as surface waves of the liquid will coat some part of the glass not submersed, where the SolGel will dry instantly and lead to a line in the final coating. The piece is lowered such that the SolGel does not flow on top of the piece, to prevent liquid from dripping down the sides, ruining the coating. We use holders that do not obstruct the flow of the SolGel off the bottom of the substrate. This allows for easier cleaning and reduces the likelihood that the SolGel will form a bead around the lower edge of the optics.
The thickness of the coating is proportional to the speed at which the substrate is removed from the SolGel bath. The typical drawing speeds are on the order of 5mm/s for stock solution SolGel. Lower speeds with higher concentrations of SolGel and higher speeds with lower concentrations quickly lead to a high degrees of inhomogeneity in the coatings.
We successfully used glued, 1/2” thick Plexiglas for our tanks. It is important to make good, stress free glue joints, and leave the SolGel only as long as necessary in the tank, as the ammonia in the SolGel has the ability to dissolve Plexiglas.
The overall budget was just over $3,000, thanks to having a number of requirements already in-house. As a rough guide, we provide the budget estimates below. We had a motor, controller, clean room and miscellaneous supplies already available to us, which kept overall cost down.
|High purity tetraethyl orthosilicate (TEOS)||$200|
|Material used to construct tanks||$100|
This project was a collaborative effort involving AJ Riggs, Christian Schwab, Andrew Szymkowiak, and Brian Tennyson. We would like to thank Jim Stilburn, from the Herzberg Institute of Astrophysics, for his great advice and his SolGel handbook, and Drew Phillips and Joe Miller from UCO/Lick for useful discussions.
Stilburn, James R., High-efficiency sol-gel antireflection coatings for
astronomical optics, Proceedings of SPIE Vol. 4008 (2000)
Phillips, Andrew C., et al, Progress Toward High-Performance Reflective and Anti-reflection Coatings for Astronomical Optics, Proceedings of SPIE Vol. 7018 (2008)
Thomas, I.M., High laser damage threshold porous silica antireflective coating, Applied Optics, 25, No. 9, pp. 1481-1483 (1986)
Wikipedia - Sol-Gel page