Hébert Labs

Electroplating Through Holes and Vias

For several years I used a conductive epoxy product called ProConduct®, proprietary to LPKF, to "plate" my through holes and vias. And while I was only doing two-layer boards this product was acceptable, for prototype work at least.

Why the caveat? Why are holes plated with conductive epoxy only acceptable for prototype boards? Because the resultant barrels aren't substantial enough to withstand even a little reworking. This is not a disparagement of LPKF, or ProConduct®. It is the nature of any conductive epoxy. The epoxy barrels dislodge from the hole walls easily, and once dislodged cannot be replaced.

And now that I'm making four-layer boards, in addition to durability concerns, conductive epoxy simply isn't reliable enough either.

The epoxy is applied to the hole wall by vacuum. The paste is smeared over the drilled board while on a vacuum table. The airflow draws the epoxy through the hole, leaving a coat deposited on the hole wall. Unfortunately, this process becomes less reliable with higher aspect ratios of diameter to depth. It's an air flow issue.

So with the need to produce a 4-layer board it was time to expand my PWB fabrication capabilities to include electroplating.

Happily, as it happened, I already had a power supply that generates enough current (20 ADC), and that's pretty much the most expensive piece of hardware required. Unfortunately, that was nowhere near all I would need. It turns out that there's far more to plating through holes and vias than just dipping them in electrolyte.

Preparing the PWB for Electroplating

There is more to plating a through hole than simply dipping the panel in electrolyte and applying a current. Before plating begins you must first desmear, and then activate the holes.

 

Desmearing

 

The first step toward plating a through hole is called desmearing. Imagine a cross-section of the through hole you want to plate. Visible on the hole wall, like strata in a canyon wall, are the edges of copper layers separated by expanses of FR-4. Ultimately you want to electrically deposit a layer of copper over the entire wall to provide continuity between all of those copper layers. That means the copper must actually make physical contact with the edges of those copper layers.

Unfortunately, when you drilled that hole, the bit that cut away the unwanted material generated significant heat from friction. One consequence of that heat is that it will soften the resin in the FR-4 layers. And once that resin is softened, the motion of the bit will smear it over the wall of the hole, creating a dielectric layer covering the edges of the copper.

So before you plate the hole you must remove that unwanted layer of dielectric in a process called desmearing.

In production facilities desmearing is done with harsh and caustic chemical baths. I, on the other hand, am not trying to produce large quantities of PWBs, nor am I fond of harsh and caustic chemicals (when I can avoid them at least). So I came up with an alternative to chemical desmearing that I call mechanical desmearing.

I drill the holes twice. It's as simple as that. I drill the first hole with a bit that is one size smaller than the final hole diameter, and then I redrill the hole to it's final diameter with the correctly sized bit.

When the first hole is drilled, the entire length of the cutting edge is removing material, and producing heat. But when the hole is redrilled only the very tip of the cutting edge is removing material. This involves so little friction that while the last bit of material is removed, along with any smear from the first operation, insufficient heat is produced to generate any new smear and the finished hole is left clean for plating.

You can demonstrate this more intuitively, for yourself, with standard bits and a power drill. In your shop, drill a hole in a piece of scrap material, and remember how much effort it took to cut that hole. Then, redrill the same hole with a bit one size larger and you will find it is relatively effortless. Machinists use this technique for critically dimensioned holes, where the heat of the original operation causes thermal expansion (enlargement) of the tool. It is the basis for reaming, and it works well for desmearing too.

 

Activating the Hole

 

Desmearing the hole will ensure that the electroplating will make contact with the inner layers, but not if you can't plate the wall of the hole. And since the FR-4 is not conductive (by design) it will not plate.

In order for copper to be electrically deposited on the hole wall,  the hole wall must first be made conductive. This process is called activation.

Once again, and with all due diligence and discrimination (there really is a lot of unreliable information out there, so browser beware), the internet came to my rescue. Rather than spending untold months developing my own chemistry, I found a highly effective recipe for a very reliable and effective activator solution at the web site Instructables.com.

It's a solution of

  • deionized (or distilled) water,

  • copper sulfate (root killer),

  • calcium hypophosphite (Ca(H2PO2)2),

  • 25% ammonium hydroxide and

  • liquid soap.

It's applied via dipping and pyrolysis (baking), and you can get all the details at the link in the preceding paragraph.

You may have to put some effort into getting the calcium hypophosphite. I don't know why. I ordered mine from an industrial chemical supplier. It's not illegal to own, make, buy or sell. Apparently it's just not something in high demand in a consumer market.

Also, the 25% solution ammonium hydroxide is not the ammonia you can buy in the cleaning aisle of Wal Mart. Oh, that stuff is ammonium hydroxide, but it's only about a 2% or 4% solution. I got the 25% solution through Amazon.com.

The copper sulfate is readily available at any hardware store. It's sold as root killer. Read the label. It should be 100% (or very nearly 100%) copper sulfate. You will recognize it as bright royal blue crystals.

For deionized water, I installed a 5 stage reverse osmosis deionized water filtration system. They are readily available at many places, including some big box home improvement stores. I got mine from an aquarium store (they're used in salt water aquariums).

Of course, I bought a DI filter system because I have a need for deionized water from time to time. If you don't, all you really need for this application is distilled water. If you have a pressure cooker, some tubing, a pot with ice and a little patience you can make your own. Or you can go down to Wal Mart and buy a gallon jug of distilled water.

Don't quote me on this, but I'm pretty sure the principle function of the liquid soap has to do with its surfactant properties, to help disperse the solution over the intended surfaces. Nonetheless, as far as I'm concerned, the principle function of the liquid soap is to make the smell bearable. I only use this stuff under a vent hood. If you don't have a vent hood, take it outside. Like I said earlier, 25% ammonium hydroxide is not the stuff you buy in the cleaning aisle.

Still, it is by far worth every effort necessary to put this solution together. Once done, you will have the ability to activate more holes than you'll likely drill in a lifetime, and for a fraction of the cost of any alternatives I found.

I would add one final point on the matter. Before you activate the hole walls, micro etch the panel. This will do two things.

First, it will greatly enhance the whetting of the holes with the activator solution. Literally, it will make the difference of having to dip the board multiple times and needing to dip it only once.

Second, micro etching the board will greatly enhance the adhesion of the electrodeposited copper to the existing copper. Without micro etching, the electrodeposited copper will exhibit a tendency to flake off of the underlying copper surface. This will be a significant issue if you machine, as opposed to chemically etch, your circuit features.

For details on micro etching, see the report "PWB Lamination Press."

Electroplating

To the outsider, electroplating chemistries can seem like so many alchemist's elixirs, made with voodoo and black spells. Of course, they're not. They're actually precisely controlled chemical solutions of electrolytes and ions, with maybe a catalyst thrown in here and there, each intended to a very specific purpose. It's just that most of those purposes are neither yours nor mine.

My purpose, and presumably yours, is to deposit a conductive layer on the wall of a hole in a panel of FR-4. I don't really care if it's bright or dull, red or brown, glossy or satin. I just want it to be conductive, or more precisely to have the same electrical characteristics as the materials I'm connecting. Toward that end I found yet another internet resource in Dalmar Electroplating.

In particular, I use the product they call Copper Electroforming Solution C-1. It works at room temperature, so it doesn't require heating. All it requires are copper electrodes flanking the work piece (the panel you're plating) on each side, and 0.1 ADC for each in2 of plating area. For my 9" x 12" panels, with between 0.5" and 0.75" revealed above the plating solution, that translates to about 20 ADC.

I have found that in order to get uniform and even plating over the surface of the panel, the flanking electrodes should be the same size as, and at least 3 or 4 inches away from, the panel. This allows the E field to become homogenous and normal (orthogonal) to the panel surfaces.

Using this configuration I have determined that copper deposits at a rate of very nearly 0.002" per hour. A 0.0236" hole was reduced to an i.d. of 0.0197" in 1 hour, which translates into a 0.0039" reduction in i.d., or a deposition thickness of 0.00195".

I did have difficulty initially, getting all of the holes to plate, specifically the smaller diameter holes. Analysis of the failures suggested that the problem was caused by air bubbles trapped in the holes.

Research suggested that mechanically knocking the board would dislodge the bubbles, but I could find no convenient way to implement such an approach in my tank. Instead, I tried agitating the plating solution with ultrasonic vibration units to dislodge the trapped air bubbles. It did not provide satisfactory results.

What I have found to be effective is removing the air bubbles from the holes before putting the board in the plating solution. I lay the panel flat in a tray of deionized water. It must be deionized so as not to contaminate the plating solution. I then lift the panel to the surface of the water and abruptly press it down. "Abruptly" doesn't have to be so hard as to slosh water out of the tray. It should be just forcefully enough to push water through the holes. When done properly you can see swells of water form over each hole as you press it down. This is an effective way to ensure that all of the holes are free of trapped air.

Once all of the holes are ready, lift the board by the edge to a vertical orientation and move it directly and immediately into the plating solution. Do not give it time for the water in the holes to begin draining. Once it is in the tank, I move it back and forth pushing against the broad surfaces a couple of times to displace the deionized water in the holes with plating solution. Using this technique I have achieved something approaching 100% effectiveness.

Reverse Pulse Plating

To date I've only done four-layer boards. They're about 0.062" thick. Since the smallest holes I use are 0.5 mm, or about 0.020". This produces an aspect ratio of approximately 3. Such holes can be nicely plated with a straight DC plating potential.

But, if I were to make a 6 layer panel, it would result in the same diameter holes having an aspect ratio of about 5, and that would necessitate a technique called reverse pulse plating. For more on the subject, see the Side Note in the right hand column of this page.

In order to be prepared for such an eventuality I developed a reverse pulse plating controller. I've not used it yet, but I have it should the need arise.

Reverse Pulse Plating Controller SchematicThe plating current (provided by an external power supply not shown) is passed through a DPDT relay off-board (also not shown). The contacts are wired such the plating current passes normally through the plating tank when the coil of the off-board relay is not energized. But when the off-board relay coil is energized with 120 Vac the contacts reverse the polarity of the current to the plating tank. JP3 connects to the plating tank electrodes and LEDs D1 and D2 provide visual confirmation of the polarity of the current.

JP1 is connected to normal mains power, 120 Vac at 60 Hz. JP2 connects that 120 Vac power to the coil of the off-board relay via K1, a SPST relay. It is through K1 that the 120 Vac power, and consequently the off-board relay and thus the polarity of the plating current, is controlled. When K1 is energized, the plating current is reversed. When K1 is un-energized the plating current is normal.

K1 is controlled via Q1, a common 2222 NPN transistor. U1 is a TLC556, a dual 555 timer configured in series, The sequence can be turned off by means of S2, which leaves the plating current biased normally. Once S2 is closed the 555 sequence can be initiated by means of S1, During the period of U1A, the first 555 timer, Q1 is unbiased (in cutoff) leaving the plating current biased normally. During the period of U1B Q1 is biased to saturation energizing K1 and ultimately reversing the polarity of the plating current.

The period of U1A is varied by means of R1, and the period of U1B is varied via R2. In this way the total period and duty cycle is controlled.

 

Oscilloscope measurement of Reverse Pulse Plating Controller performance

 

I built the circuit as depicted and tested it by letting it run overnight. I set the periods of U1A and U1B to approximately 10 s and 3 s respectively (approximately a 30% duty cycle). The resultant output, as measured on JP3, is shown above. The circuit ran overnight without incident, and the output was not measurably changed the next morning (> 12 hours).

Reverse Pulse Plating

 

Reverse pulse plating is an electroplating process that was introduced to the PWB industry specifically to facilitate the plating of through holes and vias with high aspect ratios (holes that are much deeper than their diameter). Reverse pulse plating allows the fabricator to control (i.e. eliminate) a phenomenon called "dog boning."

Dog boning

Electroplating builds up a metallic layer by depositing metal ions on a conductive surface. The surface to be plated is made an electrode, with the opposite electrode being a source of the metallic ions. Those ions are then moved through a solution by electromotive force.

The potential of that force (the first derivative of the electromotive force field) is most concentrated at sharp corners (at discontinuities of the derivative), such as at the ends of a through hole or via. Also, that potential is least dense in the middle of those holes where it is most shielded from the opposite electrode. The net result is that metal deposits build up most rapidly at the sharp corners, and most slowly in the center of the holes. (See the left example in the drawing above).

In some cases, the hole's aspect ratio can be so great that the ends plate closed before the center plates at all, leaving the circuit open. This was a problem in dire need of solution.

It turns out that the same mechanism which deposits metal ions will, if the polarity is reversed, remove those same ions. This process is called electro-cleaning.

By periodically reversing the polarity of the electroplating potential, metal ions can be removed from the plating surfaces of the PWB. And they will be removed more from those areas where the potential is densest, and less where it is most rarified, exactly where needed.

By controlling the duty cycle of the pulse reversal, even high aspect ratio holes can be plated with very uniform barrel (plated linings) thickness. (See the right example in the drawing above.)