How to Braze Tungsten Carbide – Surface Preparation
Treating Tungsten Carbide for Brazing
In order to get a good braze joint the braze alloy has to stick to both the tungsten carbide and the steel. With steel it is usually just a matter of having a clean surface that is absolutely free of greases and oils. Getting a piece of tungsten carbide to wet and braze well can be a much more difficult matter.
Tungsten carbide is made of tungsten powder, carbon powder (lamp black) and cobalt powder. This is the traditional formula. Modern carbide grades will have additives such as chrome to make the binder more corrosion resistant, grain growth inhibitors to make finer grains and other materials to make the carbide tougher or longer wearing.
My theory is that the number one cause of wetting problems with tungsten carbide is the presence of carbon on the surface of the tungsten carbide. Traditionally tungsten carbide has been sandblasted after sintering. This is to clean it up and remove any sand if it was sintered in sand trays, for example. I believe the sandblasting can remove some carbon and some oxides but I don’t think it does a good enough job all the time. Sandblasting can remove rust, paint and similar from an underlying surface. However it cannot remove greases and oils entirely. What happens is that the sand in the sandblasting picks up much of the greases and oils. However there is also a smearing effect by which the sand in the sandblasting operation distributes the oil and grease in an even layer over the part. There is also a tendency to drive the oil and grease into the underlying material as much as possible.
I have been working on making carbide wettable since 1981. What follows is my opinion on how I see the situation. In 30 years we have treated umpteen million pieces of carbide to improve the wetting and bonding properties.
The first experiment was to take a piece of carbide that would not wet. We then ground the surface with an ordinary wheel from a Sears bench grinder. We ground the surface a couple of times cleaning the wheel in between. Once we had ground the surface we had a piece of carbide that wet readily.
For number of reasons, physical grinding was not a suitable process for us to use in production. In that operation we were tinning (or pretinning) tungsten carbide saw tips for a couple cents each. There was no way we could afford to grind each saw tip and there was really no way we could practically grind a hundred thousand saw tips a day.
So I went looking for a chemical process. Don Anderson, who was head of cutting tool research and development at Weyerhaeuser, gave me a pamphlet on removing scale from steel. The steel process used a mixture of potassium permanganate and sodium hydroxide. They made a big bath and just ran the steel through it. When I tried this with carbide the improvement was marginal at best.
What I ended up doing was drastically changing the ratios of the two chemicals and altering the way the process was used. We now put the carbide in the treating vessel and dump dry powder on top of it. We then add enough water to make a sludge and heated it at a gentle boil for about 20 min. This leaves the tips with a dark brown coating which does not seem to interfere with the brazing. These tips will re-oxidize if the brazing cycle is too long or too slow.
What I think happens is that the KMnO4 (Potassium Permanganate) breaks down and does two things. I think it removes free carbon from the surface of the tungsten carbide. I also think it helps remove oxides. The sodium hydroxide obviously does a good job on removing all oils and greases. These are assumptions are somewhat confirmed by the fact that the process produces a gas that extinguishes a flame. I mentioned that we initially add enough water to make a sludge. At the start of the process more water is created. By the end of the 20 min. we may have lost water due to evaporation from the boiling.
I believe it is in this initial reaction involving the free carbon on the tungsten carbide with the potassium permanganate and sodium hydroxide that actually prepares the surface to make it much more wettable.
We have run that process for about 25 years now. It is relatively inexpensive, pretty safe, pretty easy and we have permission to dispose of the spent chemicals in the city sewer.
There is a commercial treatment process currently in operation. It is a patented process and I will be happy to provide a copy of the patent for anyone who cares to look at it. In essence the patent calls for treating the tungsten carbide parts in a 2200° salt bath followed by a quench in us in a hot, cyanide salt bath.
This has been described as causing the cobalt binder to migrate to the surface. It has also been described as removing the tungsten carbide from the surface which leaves the cobalt exposed.
Looking at the drawings on the patent, they looked to me like unwettable carbide which had been cleaned. They show a definite improvement in the wetting but it is not a perfectly wettable surface. In addition the test parts were heated with the tungsten carbide lying on top of the braze alloy.
The cyanide salt bath never made any sense to me. Why would you use a cyanide salt bath for anything unless you absolutely had to?
The original patent indicates that the inventor worked for a twist drill company and I believe that he tried running some tungsten carbide through a plating operation to see what would happen.
I thought that you could take the heat out of the equation and the salt baths. Instead you should be able to plate in aqueous solutions at reasonable temperatures.
The key to any plating operation is an absolutely, scrupulously clean surface.
Sandblasting does not yield a surface clean enough on a reliable basis. With plating, an interference layer of something such as oil or grease can greatly affect results even if it is no more than a molecule thick.
There are two theories to customer support. Plan A, once a customer has bought it, you don’t have any further responsibility. Plan B, you have the obligation to help your customers use your product successfully.
We have always been Plan B people so we got into research on how to prepare the steel for successful brazing in that we had nothing to do with the steel.
Working with the steel preparation made it extremely clear to us just how critically important cleanliness was to successful brazing. You can get steel clean enough with solvents but it is much more common that someone leaves a thin film on the steel. The same is true with wiping with a clean rag.
Cleanliness is absolutely essential to successful brazing and successful plating. This is hardly surprising since both processes are at their best when they create a chemical bond between the two metals.
We developed the cleaning process we had several criteria. Our operation is in the City of Tacoma which is extremely sensitive about environmental issues affecting drinking water. The last glacier that came through scraped off all the cap rock so anything that spills on the ground goes directly into the drinking water. In addition we wanted the standard attributes of low-cost, reliability, safety, etc.
We developed a specific electro-cleaning process. We use a small, tumbling barrel to hold the tungsten carbide parts. We run an electric current from a stainless steel plate to the tungsten carbide parts. (See more detail below.) This cleans the parts. It also generates fine grit in the bath and often generates a film on top of the bath. When we ran the process for an extended amount of time on a production basis we created a film that we believe was a polymer.
We found that by cleaning the parts in this manner and then reversing the current we could prepare the surface of tungsten carbide to wet successfully. We believe that this process separates the tungsten carbide grit from the cobalt matrix because the difference in electrical resistivity is so great. This process either prepares the tungsten carbide parts to wet completely or greatly improved the wettability.
A major issue in all this research was the variation in the parts we received. There is certainly variation among manufacturers but there can also be significant variation among batches from the same manufacturer. In effect, you are running a chemical process where there is the addition of an unknown chemical. That unknown chemical seems to very both in description and in quantity.
Designing a chemical process that is reliable on materials with unknown contaminants calls for an extremely robust, forgiving process.
The commercial treatment company has been dominant in the industry for decades and rightly so. When their process works right it does an excellent job. However their process does not always work right. In addition there seems to be very little flexibility in their throughput so that lead times can often run into months.
There’ve been many attempts over the years to create a successful tungsten carbide treatment process. Most have failed because the processes have not delivered consistently high quality.
Many of these attempts have been plating processes to the extent that plating tungsten carbide has gotten a very bad reputation. We have seen and worked with many of these processes. I believe that, universally, the cause of failure was a process control issue. Specifically I would point to the initial cleanliness of the parts.
However the commercial treatment company delivers parts with a certain look and that look has become the industry standard for reliable quality. A second indicator of quality, as perceived by the industry, is the presence of microscopic holes in the treated tip.
Because our two existing processes left the tips dark there was a hesitancy on the part of industry to except them.
The next step was to develop a successful process that made the tips look right.
As I read the commercial treatment company process in the patent it appeared to me more likely that it was, in effect, a plating process. I cannot imagine anyone, at any time developing a cyanide-based process unless it was absolutely essential to use cyanide. There is mention in the patent of an electrode in the cyanide tank.
We took the electro-cleaning process and added two more steps with rinse steps in between. The second step was an acid bath. It was originally a dilute hydrochloric bath. When I run the process for titanium-based products I had hydrochloric acid to the hydrochloric as it seems to improve effectiveness and reliability.
The third step is a cobalt plating process. It is a pretty standard plating process except for the cobalt. We used cobalt because that is what the industry wants. I think the process could be done equally well with equal or superior results using a nickel plating. In addition the nickel plating process would be much less expensive. Finally, if you found someone who was familiar with nickel plating to run the process, they would find a huge number of ancillary processes and suppliers to support them. It is extremely difficult to find anyone who knows anything about cobalt plating or who can supply materials for the process. We developed our own titration standard for the cobalt bath. This would not have had to be done with a nickel bath. There is also a huge body of literature on the problems with and methods of controlling nickel baths compared to little or nothing about cobalt baths.
As I write this, on November 18, 2011, this is where we stand. I had originally hoped to run the process here that the city of Tacoma will not allow it. I had hoped to license the process but no one really seems to be interested in licensing a process that is not running. I have licensed a couple versions of chemical treatment processes. There is a problem with this. As one gentleman told me, the process didn’t work the first couple times he tried. Then he followed my directions and it worked. This is much like giving a recipe to a friend. They make substitutions and then are unhappy with you because they don’t get the results you did.
The commercial treatment company seems to be having problems with their treatment process lately. My lead time for parts is extended several weeks if not a couple months. Some of the parts we are receiving do not wet well and we have to retreat them here. Some of the parts work extremely well but don’t seem to adhere very well when brazed.
I am posting this information in the hope that it will aid someone working on the same problem. I would really like to see the carbide manufacturers ship ready to braze parts directly from their plants.
Tuffco Process — Lab Instructions
Three Tanks from left to right
NaOH – 150 F Use 20 – 22 Amps – No More than 25 amps at 6 – 10 volts for 5– 6 minutes
Rinse Hot Water 2 minutes
HCl – 80 F for 5 minutes with no electricity
Rinse Cold Water 2 minutes
Cobalt – 100 F at 6 – 7 amps for 15 min. for small barrel 40% full
Rinse Cold Water 2 minutes
Preparing the Baths – Tuffco Process
Deionized water is best for solutions. Tap water usually works well enough for rinsing.
1. Sodium hydroxide (NaOH) 150 g / 1,000 ml Temperature 150° -160° F
A. pH cold 14 and pH hot 13
Put 1 Liter of water in the outside bath
Put 1.2 Liters of water in the inside bath
Use 4 Liters of water
At 800 grams of NaOH which is 880 ml.
Heat to 150° F
2. Hydrochloric acid (110) – 200 ml / 1,000 ml Temperature – (70° – 80° F) pH room temp 4
Put 1 Liter of water in the bath
Use 2 Liters of water
Add 900 ml of HCI
Add 100 ml of HF
Heat to 70 – 80° F
3. Cobalt Solution:
A. Metal concentration off test strip; 1,000 – strip is dark blue – maxed out; pH cold 6.1, pH hot 5.4 – 5.6
Put 800 ml of water in the bath
Use 3 Liters of water
Add 360 g of Cobalt Sulfate which is 400 ml. (Cobalt sulfate COSO4 is 120 g / 1,000 mi.)
Add 240 g of Ammonium Sulfate which is 225 ml. (Ammonium sulfate (NI14)2SO4 is 80 g / 1,000 ml.) Heat to 100° F
* * Check pH every day or more often depending on use
If pH lower than 5.4 add 3% solution of Sodium Hydroxide (3 g / 100 ml) if pH higher than 5.6 then carefully add concentrated sulfuric acid H2SO4
Making the cobalt bath solution
120 grams per liter of cobalt sulfate and 80 grams per liter of ammonium sulfate
Use distilled water and warm the water to 120 F – 130 F to aid dissolution. Do not mix these all at the same time Dissolve Cobalt sulfate first so you can make sure it gets all dissolved. When the cobalt sulfate is totally dissolved then add ammonium sulfate Use the same temperature of 120 – 130 F
As solution cools it will precipitate out and cake on the bottom. When you re-use Cobalt the solution make sure it is fully dissolved. Check the pH should be 5.4- 5.6 when warm
Titrating Cobalt solution
Preparing the cobalt solution for titration
Take 1,000 ml of distilled water and put it in an Erlenmeyer flask. Add 25 ml of the warm, well stirred and fully dissolved cobalt solution.
Stir or shake to fully mix
Take 100 ml of distilled water and add 10 ml of the previous solution (add 10 ml of the stirred solution made by mixing 1,000 ml distilled water with 25 ml of cobalt solution)
Slowly add Ammonium Hydroxide to the beaker until the solution turns yellow, yellow-orange / orange
As soon as it turns yellow – stop
Use Murixide Mix indicator
Add 3/4 of 1/8 of a teaspoon to the solution. The solution will turn from yellow to red
Use EDTA solution and fill the tube to the 25 ml mark
Titrate until you see a blue color
Tittrate very slowly
Multiply amount used by a factor off a chart
13 – 14 ml EDTA will cause a color change in good solution
Over 14 ml EDTA used add distilled water
Under 13 ml EDTA used add cobalt sulfate
Leave a Reply