Fabrication of Uniform Nanoscale Cavities via Silicon Direct Wafer Bonding

The overall goal of this procedure is to construct nanoscale enclosures for measuring the properties of fluids. This is accomplished by first rinsing chemically cleaned, patterned wafers thoroughly with deionized water to remove any particle contaminants that will hinder proper bonding. The wafers are spun slowly during this rinsing process.

In the second step, a cover is placed over the wafers and the assembly is spun to remove any residual dust. This is performed under an infrared lamp to dry the water as bonding cannot occur if there is any trapped water. After the spacers between the wafers are removed, the wafers are taken off the spinner in their carrier and placed in an arbor press where bonding is initiated symmetrically from the center.

The bonded wafers are then placed on a quartz vacuum chuck, and the air trapped in the microstructure formed between the bonded wafers is evacuated before the wafers are kneeled in a furnace to make the bonding permanent. Ultimately well-defined cavities between two silicon wafers are obtained. These cavities can then be used for confining fluids in experiments.

This method of using a bonded wafers with the well-defined separation and geometry allows us to study the behavioral liquids which are confined and hence are not in a thermodynamic limit. Generally, individuals who are new to this process will struggle because the bonding process is very sensitive to contaminants, such as residual resist or dust in the staging process. I had this idea for bonded wafers when I spent a sabbatical leave at Bell Laboratories in 1987, and since then, our work has been supported by the National Science Foundation.

To begin prepare the wafer in the clean room as described in the text protocol for bonding wafers. Cleanliness is paramount and steps should be taken to clean the wafers. First, rinse the wafers in deionized water, then prepare an acid bath of water, hydrogen peroxide, and hydrochloric acid.

Place the wafers in the 80 degrees Celsius acid bath for 15 minutes. With the pattern sides facing up. This step will eliminate any metallic contamination.

Remove wafers from the acid and rinse in a deionized water bath for five minutes. Next, prepare a base bath of water, hydrogen peroxide and ammonium hydroxide. Place the wafers in an 80 degrees Celsius base bath for 15 minutes.

With the pattern sides facing up. This step will eliminate any organic contamination. Then rinse the wafers in a deionized water bath for approximately 15 minutes.

Upon removing the wafers from the deionized water bath, they must remain clean in order for proper bonding to occur. To ensure this place the wafers with their pattern to etched sides facing each other on a Teflon chuck in a clean micro chamber, they're separated by approximately one millimeter Teflon tabs. Spray deionized water between the wafers while they spin slowly for approximately two minutes.

In order to remove any particle contamination, a film of water will be left between the wafers at this point, this prevents dust contamination prior to the next step. Cover the wafers with the clear acrylic lid and spin the wafers dry for approximately 30 minutes. At 3000 RPM, use a 250 watt infrared heat lamp to aid the drying process.

The rapid spinning will entrain any particle contaminants with the ejection of the water film. Before removing the lid over the wafers, remove the tabs, separating the wafers by rotating the lid. This will bring the wafers into light local contact while still in the micro clean chamber.

Now the wafers may be safely removed from the micro unclean chamber on their carrier. Do not pick up the wafers with tweezers at this point since this would initiate asymmetric bonding. Instead, transport the wafers with the use of the removable carrier onto the arbor Press.

Press the two wafers together using an arbor press in a fairly rigid and smooth Nerf ball. The Nerf ball is used to apply pressure to the wafers from the middle. Outward pressure applied this way allows trapped air to be pushed out as the bonding wave spreads from the center out.

Starting the bonding at the center minimizes the stresses which are built up as the wafers contour to each other. Check the bonding by looking for interference fringes using an infrared light source and detector with a one micrometer high pass filter interference. Fringes will appear if there is poor bonding as shown here.

If bonding is poor and there are non-uniformity, place the cell on an optical flat cover with filter paper to protect and cushion the top wafer and press the wafers together with wafer tongs, push debonded bubbles to either the middle where there is the filling hole or to the edges. Be careful when applying force near the edges to avoid cracking the top wafer. If the bonding irregularities persist or a dust particle is evident, split the wafers by wedging a razor blade between them and repeat the process from the beginning.

Up to this point, the bonding is reversible. The wafers can be rebounded at room temperature many times to get acceptable bonding. Since one of the wafers is cracked, the process needed to be repeated with new wafers.

After obtaining acceptable room temperature bonding proceed by kneeling stage the cell onto a quartz vacuum chuck, such that the filling hole is centered over the pumping hole in the chuck. The chuck is connected to a quartz pumping tube that extends outside the furnace to evacuate the cell. Prior to and during the annealing process.

A nitrogen trap is used to eliminate back streaming oil into the cell. Evacuating the cell causes a pressure of one atmosphere to be applied to the cell, which helps with bonding, pumping, prevents pressure buildup. If the furnace temperature is ramped up too quickly, the time it takes to significantly lower the pressure in the cell will depend on the geometry within the cell.

Avoid the growth of oxide on the outside of the cell by purging the furnace chamber with a non reacting gas, typically helium four so that no oxide is grown to allow for strains to have time to relax. It is important to ramp temperatures from 250 degrees Celsius to 1, 200 degrees Celsius over the course of approximately four hours. After staying at 1, 200 degrees Celsius for at least four hours, turn off the furnace.

Allow the system to cool to room temperature, analyze the cell once again using the infrared light source and detector. If a kneeling went well, the cell will look as good as or often better than when initially put in the furnace. If there are unacceptable fringes indicating poor bonding, the entire process must be repeated from the beginning.

After kneeling, the bonding is permanent and you can no longer split the waivers Properly. Bonded wafers will have no unbonded regions as shown here in infrared images attempting to split the wafers after a kneeling will cause the cell to break into pieces due to the strength of the bond. Often a kneeling improves the uniformity of the cell, especially if local unbonded regions are due to lack of flatness.

In the wafers here, the light spots and border are bonded areas. The center bright spot is the hole for filling this cell. The only unbonded region is near the border on the top left side of the image.

Since it is located beyond the edge of the oxide border and thus could not be filled up with liquid, this would not affect the use of this cell. There are multiple symptoms of poor bonding, which can manifest, however, the most common is having a trapped particle between the wafers. This will cause localized lack of bonding to occur and is visible via the appearance of interference.

Newton rings in the infrared image near the center of this cell where a square pattern of channels are formed. There's a pattern of several Newton rings. These cells would not be suitable for use.

Pressure was applied locally in an attempt to close the unbonded region. This is partly effective and there are fewer rings, but still there remains small in homogeneity. As a result, these wafers were split and the bonding process was restarted after bonding the wafers.

Fabry Perot interferometry is used to determine the local separation of the bonded structure by using the interference of transmitted light as it is multiply reflected by the parallel surfaces in this cell. Combined with the infrared imaging, these methods confirm the uniformity of the cell structure While attempting this procedure, it's important to remember that wafer flatness and cleanliness are essential for proper bonding. The development of this technique paved a way for researchers in a field of critical phenomenon and hydrodynamics to explore the role of dimensionality and confinement in in the behavior of these systems.