However, both of the favorable factors (i.e., high internal pressure and short diffusion distances) are missing in large TSV scale structures.
Even if this feature were only filled 10% with gas at its bottom, it would still take 20 minutes or more for that gas to be removed.
This procedure at first may appear to be appealing, but may suffer from two limitations.
First, for large deep vias, the diffusion distance for the gas may still be a significant limiting factor.
Second, since the amount of gas in the solution can never be less than zero, the magnitude of the driving force for dissolution is limited to be no more than approximately HxiP (P=1 atmosphere).
Feature depth is a significant limiting factor in all cases, with deep features having long dissolution times.
However, for a variety of reasons, including the hardware complexity of combining plating processes with vacuum processes, pre-wetting (including vacuum feature-backfilled pre-wetting) is often performed in a different cell, sub-cell, or module than the plating cell.
Furthermore, when using a pre-wetting fluid of a different composition than the plating bath, performing the pre-wetting process in the same module without enabling suitable mechanisms of removing and recovering excess entrained pre-wetting fluid that would be added to the plating solution would generally require mechanisms for mitigating, monitoring and/or otherwise correcting for plating solution modification over time.
This results in bubbles forming inside the vias.
While not wanting to be limited by a particular model or theory, a via bottom is a location of negative curvature, and it is believed that this location is a particularly susceptible to nucleating a bubble and releasing gas from the pre-wetting fluid.
The bubbles so formed can remain there after the pre-wetting process, which in turn can inhibit plating there and lead to associated defects.
If a separate pre-wetting chamber and apparatus are employed, but the pre-wetting fluid is not degassed, then intermittent and unreliable filling results may be observed.
Weeping can be a particularly difficult problem when dealing with salt containing pre-wetting fluids, because weeping salt laden fluids tend to dry and destroy the pores of the degassing device.
Also, reducing the temperature of the pre-wetting fluid reduces the rate of metal corrosion in the pre-wetting system.
Also, there are a limited number of practical mechanisms for applying hydrostatic pressure to the thin layer of pre-wetting fluid on a wafer.
This is because pressure can be transmitted to the bubble by a purely hydrostatic mechanism, and alternatively, application of pneumatic pressure will not quickly re-saturate the pre-wetting fluid around a bubble in a via with gas because of the relatively long diffusion distances involved.
Typically a 100 μm deep via might have a 25 μm diameter opening, so the assumption of a bubble rising in an infinite media is not correct, as wall flow-slip effects will increase the time.
If the wetting layer pulls back or coagulates from a previously wetted surface, then the attributes of the pre-wetting process are lost.
This de-wetting may cause the fluid to be drawn out from within any recessed features within the wafer substrate, possibly leading to gas being trapped within the feature on immersion into the plating bath.
Hydrophobic surfaces, particularly those that have completely de-wetted in some regions, have non-uniform fluid pre-wetting layer thickness over the wafer substrate.
In the case that the pre-wetting fluid in use has a different composition than the plating bath, the subsequent immersion of the pre-wetted wafer into the plating solution will not allow for a uniformly wetted surface if the pre-wetting fluid has not wetted the wafer properly.
This can lead to variation in feature filling behavior or the creation of various wafer surface defects, such as lines of entrapped bubbles, metal pits, metal thickness variations, or growth protrusions.
Furthermore, this transformation, specifically when occurring under vacuum and with a degassed pre-wetting fluid, leads to particularly favorable low defectivity when combined with the subsequent plating operation.
Furthermore, areas around state 3 that are in state 1 or 2 are wetted and are or will become hydrophilic, allowing fluid to flow freely and continuously over this surface and making the removal of the bubble or wetting of adjacent surfaces considerably more difficult.
Furthermore, bubble flushing is not 100% effective, and will often lead to bubble fragmentation, leaving a large number of smaller, hard to remove bubble behind.
The acid copper fluoroborate bath (mixture of copper and fluoroboric acid with boric acid), with its high solubility of copper and potential for high deposition rates, is also used, but has largely fallen out of favor and replaced by the methane sulphonate system (which also has high copper solubility), at least in part because of the tendency for the BF4− anion to decompose and form hazard