How a Fab Shop Qualifies GTP APT, AMT and Tungsten Oxide

The number that decides whether a batch of tungsten carbide cuts clean is not the price per kilo. It is 0.003%. That is the impurity ceiling GTP puts on its ammonium paratungstate, and the day I stopped treating it as a line on a certificate and started treating it as a process limit, my sinter runs quit drifting.

For a shop specifying tungsten feedstock, that 0.003% ceiling moves your sintering yield more than the ore source or the price band does. What follows is the route one shop lead actually walks — first phone call, first test bar, six-month scrap report.

Day One Starts With the Ore Path

Day one is a phone call, not a purchase order. Ask where the batch came from. GTP runs its tungsten chemicals — ammonium paratungstate, ammonium metatungstate and tungsten oxide — from either secondary raw material or virgin tungsten ore concentrates such as wolframite, hubnerite and scheelite. Get that answer in writing and staple it to the purchasing card for the lot.

It matters in the furnace, not on paper. Virgin wolframite and recycled feed differ in sodium level and grain morphology, and both show up weeks later as a change in reduction rate. Same hydrogen dew point, same push rate, and a higher-sodium feed reduces slower and leaves you a finer powder than the press was set for. Nothing on the purchase order warns you. The ore path does.

Two answers count as complete. The supplier names a mineral or a region, or says outright that the feed is secondary. If what comes back is 'tungsten concentrate, various sources,' you have no ore path, and you cannot compare this lot against the one still sitting in your crib.

Put the ore path on the receiving tag, not just the purchasing card. The person who opens the drum on the floor is the one who catches a colour shift in the oxide two weeks later, and a tag that says wolframite tells them the shift is normal for that feed instead of a reason to shut the line down.

Ask about the last three lots, not just this one. A supplier who can name the ore path for the current batch but goes quiet on the previous two is telling you the path is not tracked in their system either, and then your trial run becomes the tracking system.

By Week One, the 0.003% Number Sets Your Limits

By the end of the first week the number that matters is on the APT certificate: impurity levels below 0.003%. Treat it as a grain-size budget, not a badge.

Residual sodium, molybdenum, iron and potassium are what drive abnormal grain growth in sintered carbide. A few parts per million of one of them and you get a scatter of oversized grains inside an otherwise fine structure. Those grains are exactly where a finishing insert chips first, and no grinding puts that edge back.

Pull the per-element sheet and compare it element by element against the lot you are running now. Sodium against sodium, molybdenum against molybdenum. A single summary figure tells you nothing unless it is batch-tied and broken down per element. Otherwise you are buying a number, not a limit.

Here is what the ceiling is worth in money. One wasted sinter run — a hundred kilos of powder, a day of furnace time, a scrapped lot of inserts — costs more than the price gap between a certified lot and a cheap one. The 0.003% line is how you avoid paying for that run twice.

Your First Month Runs the Conversion Trial

The first month is a conversion trial, and the order carries the result. Dissolve the APT and look at the solution, because that is your first read on how clean the feed really is. Calcine to tungsten oxide. Reduce to tungsten metal powder. Carburize to tungsten carbide. Press a test bar and sinter it. Five steps, that order, and you have something you can cut and measure instead of argue about.

GTP sells APT as the source for high-purity tungsten oxides, tungsten metal powders and carbides, and as a laboratory reagent. If your shop needs only the oxide, for a spray coating say, you can stop after step two. AMT is the better fit when you want more tungsten into solution per litre, which is why the metatungstate route turns up in catalyst and electrode work.

Between calcining and reduction, hold the boat temperature on the exact ramp you will use in production. Not a lab ramp. That production ramp sets the Fisher size of the powder, and no certificate has anything to say about it. Same chemistry, different ramp, different particle size, different behaviour in the press.

Two checks before you carburize. The oxide colour has to match the spec sheet, and the powder tap density has to match it too. If either is off, stop and re-run the reduction. Push a bad oxide into the carburizer and the next honest reading you get is a sintered bar, three steps too late to save the lot.

Cut and measure that test bar the day it comes out of the furnace, not the following week. Hardness, density, and a polished section under the microscope for the oversized-grain check. The bar is your only cheap look at whether the chemistry you paid a premium for made it all the way through the chain.

Three Months In, the Batch Sheet Beats the Brochure

Three months in you have three or four batch sheets and something better than any datasheet: your own comparison. Line up ore path, per-element impurity, oxide colour, Fisher size and sintered hardness side by side for every lot you ran.

Every hardness drift I have chased tracked back to the impurity column or the ore-path column. It never tracked back to price. That comparison is what lets you tell a supplier what you need instead of taking what is on the shelf.

Change one variable at a time. If a new lot swings from virgin feed to secondary, re-run the whole month-one trial — dissolve, calcine, reduce, carburize, sinter — instead of trusting the certificate from the previous lot. That certificate was written for a different feedstock.

Keep the sheets in a binder by material, not by date. When a customer asks why a tool ran short, the binder answers in about a minute: which lot, which ore path, which hardness number, which tool-life count. Without it the same question costs a week of guessing.

Six Months Out, Scrap Rate Is the Judge

Six months out you stop reading certificates and start reading your own numbers — scrap rate, tool life, and the chipping count on the finishing insert. A feed change shows up there long before it shows up on a purchase order.

If scrap starts creeping, open the ore-path column before you call the machine shop. The furnace and the press are the usual suspects and usually innocent. A feed change is quieter and moves the same numbers.

Put the scrap number and the incoming impurity sheet on the same page every month. Two numbers, one page, and the trend shows up long before a customer complaint does. That page is the whole qualification, written down.

What confirms you did it right: the incoming impurity sheet, the trial-bar hardness and the tool-life figure all agree with each other. And your reorder spec names the material — APT or tungsten oxide, with the element that matters called out — instead of the words 'tungsten chemical.' That is the difference between ordering a feedstock and ordering a grade band.

Go back to that day-one call. Six months on, the answer to 'where did the ore come from' is sitting in a scrap number. Chipping flat and hardness in band, and the ore path was right and the 0.003% ceiling is doing its job. If it drifted, you know which column to open first and which furnace step to run again, and the next supplier call writes itself.