The electroplating process is costly in both raw materials and time investment.
On top of that, the yield is quite low: a 30% yield is typical.
We used Optimet sensors to both increase the yield and save hours in production.
Before I explain how, let’s backtrack a bit:
What is electroplating?
Electroplating is the process of applying voltages to a very specific pattern on a gold-plated disk (or nickel-plated, or whatever). This voltage pattern causes a thickening of the gold in that specific location that the voltage is applied.
Over time, this can be used to create small mechanical components with very fine features (tens of microns).
When I say over time – I mean over time. The components “thickening” process takes about 30 μm / hour.
What is it good for?
You can already probably imagine no one is going to use this method to produce a car, or anything large. This is a very slow process, especially for larger objects.
That means we’re dealing with small components (hundreds of microns thick) that can’t be produced any other way – usually since the parts require very high precision.
How does the electroplating process work?
I’m not an expert, but here are the basics of the electroplating process.
We start with a batch of silicon wafers. Then each wafer is coated with gold (or another metal) in a thin, uniform coating.
Then the true magic happens: Voltages are applied in a special pattern, and we sit back, grab some popcorn, and wait.
The parts are programmed to be a little thicker than necessary, since we can always polish them down, but of course, we can’t polish them “up.”
When the electroplating is done, we can use a point-touch sensor to measure a few points. Based on that measurement, we know how much to polish the wafer.
After the polishing is done, we can separate the individual components from the wafer. (Note: once they’re separated, there’s no going back.)
Finally, each individual part is measured against the production requirements. Usually about 30% pass. The rest are scrapped.
Issues with this process
Let’s take a look at the points of this production process that could be more efficient:
How can Optimet sensors help?
Let’s add Optimet sensors to the equation and see how much happier this situation looks.
First, the time issue:
One big reason for the spare is that the base wafer, after being thinly coated with metal is not quite flat. It has a slight, but significant curvature. That means that parts in the middle will be shorter than parts at the edges (after polishing evenly throughout).
We used a non-contact sensor to scan the thinly-coated wafer (at 3 kHz). This gives us a map of the terrain we’re dealing with. Before building components, we can use that very same electroplating technology to even out the wafer. Simply apply voltage to the edges of the wafer where the plating is too thin.
Once we have a base that is sufficiently flat, we can begin the true electroplating process. But because we’ve removed the biggest source of non-uniformity, we only need to add an extra ~30 or 50 microns instead of 100.
I don’t need to tell you how quickly time turns into money in the factory.
But that’s just the icing on the cake. The real breakthrough is in improving the yield.
Measuring and flattening the basic wafer helps on this front also. Working off of a flat disk makes it much easier to get the components to pass spec.
Another inspection is performed with Optimet’s unique collinear sensors after the build (but pre-polish) to determine exactly where there are problematic spots. This often indicates a non-uniformity in the electroplating machine or process.
This can be pinpointed and adjusted for the rest of the wafers in the batch.
There will always be a few stubborn parts that refuse to come out as planned. But using these methods should bring the 30% yield much higher.
For more information contact Shmulik Barzilay: Shmulik.Barzilay@optimet.com
Finally, here are some scans of the gold-plated wafer, courtesy of the TrueMap software developed by our partner TrueGage: