So recently I melted down some scrap zinc ingots I'd made from recycled boat anodes. I wanted aluminium in the alloy to get something similar to ZAMAC which has bearing/bush alloy properties but as soft as possible so ZAMAC 3 with up to 4% Aluminium would be great but higher levels would be fine as I wasn't adding any copper that higher levels of ZAMAC have to make them harder (5-15% aluminium, 1% copper). I melted down about 10Kg of my zinc ingots and added 500 grams of clean aluminium extrusion as the most common 6063 is 98.7% aluminium. The aluminium melted like toffee in hot water as I'd hit the aluminium with a wire wheel to remove the oxide layer and the uncleaned patches were last to dissolve in the melt. Now I'd had a zinc ingot sample from a later batch tested at 98%+ zinc and one percent iron (from the steel crucible) so I wanted to test removing iron from zinc by adding aluminium as I'd read a paper where the aluminium reacts chemically with the iron and forms a floating solid (at molten zinc temps) in zinc that can be skimmed off. The paper then described adding ammonium chloride (sal ammoniac) to react with the aluminium and form a volatile aluminium chloride which comes out of the zinc as a vapour to leave pure zinc behind. I did get some lumpy solid dross that I generously skimmed off and collected and made several zinc castings. A few days later I had the brain wave of getting them XRF tested at my scrap metal dealer to see how I went: The rough machined sample casting had no aluminium whatsoever but had picked up some lead 0.5% and tin 0.5% , the spare unused source zinc ingot tested at 10% percent aluminium (possible to get aluminium from some anode types) and the dross zinc had somehow picked up 6% tungsten (after testing three patches top and bottom). Now unless I accidentally became a wizard (from not getting laid) with alchemical abilities: aluminium into tungsten, there is something wrong with these XRF results as you can't form tungsten (W) out of nowhere and besides, the dross skimmings weigh less than the 500g aluminium added to the mix, yet there's none in the castings. I was most careful in watching the order of the the sample testing: ingot, casting and finally dross which you can see from the dates and sample numbers on the results. Interestingly the scrap metal dealer fails to see any problem with me pulling tungsten out of thin air and disappearing aluminium at the same time. Zinc source ingot, (should have 1% iron) note the date and sample number: The rough machined zinc casting (should have aluminium) with the surface removed: The dross sample (should have aluminium), tested four days later:
I worked at a scrap yard for 8 years with an XRF, they are only as good as the calibration, most scrap yards will calibrate to a Known Aluminum Source, it is mounted in the charging station by default, others are available , from that they are very accurate on aluminum , not so much on other major alloys, think of the metals as major groups, zinc, copper, Fe, Al, also, if your first test after calibration is on a copper alloy, the gun will be thrown off, for any other elemental groups , basically if you are going to go shoot copper alloys, that is all you do, then recalibrate if you need to switch to aluminum alloys to a scrap yard anything under a percent, is a rounding error an XRF will not read dross accurately at all ever! mostly all you will get is major elements, you know there is H and O2 in the dross, but there is no reading at all an XRF needs a clean polished surface to get REALLY accurate readings, which is impossible on dross, and I have definitely seen reading much like you are showing, also sometimes the guns seem to try to match an alloy if its CLOSE, which is what the first test seems to have done to me , if it had found ANY copper it would have spit out ZA 12 V/r HT1
So you'd need a skilled operator who knows how to get the most out of the machine, thanks for confirming my suspicions about their accuracy. The owner of the XRF machine said he had it serviced recently, so maybe a new Xray source, he had what looked like a stainless steel coin as the calibration target. He put the samples on the plastic case for the XRF and there were extra targets and batteries inside the case that could throw off the readings. Not only does it show tungsten it's completely missing the aluminium that was added, it's not in the dross and not in the casting, despite half a kilo being added to the melt. That seems to back up Henry's experience with having to calibrate it often and some metals shifting the calibration.
servicing , most likely a replacement internal battery, or a cleaning , ours had a schedule every other year I think, the only time you got worried (this is scrap yard rule) you calibrated the machine, then tested the calibration disk, and of course it is a known standard Ours was 6061, if you got anything other then exact match it went to the calibration V/r HT1
I had the same problem when getting cast iron tested to see what the carbon and silicon levels were. The first cast iron I had tested was when I used mild steel to change it into cast iron. That was ten years ago and everything was what it should be except for high copper content. Mild steel does not have a high copper content and the ferrosilicon, carbon I added does not have copper in them so that puzzled me for many years as to why that happened. Nine months ago another test was made with the same company using two different samples of scrap cast iron to see how much carbon loss resulted when I remelted them. One sample showed a large carbon loss when remelted and the other sample showed a large carbon gain when remelted. Sulfur and silicon contents had similar results, some were up or down after remelting. The one consistent result was that the sample from ten years ago had copper in it and the sample from nine months ago had copper in it and I never added copper to any of those samples. When I took the samples to that company nine months ago he said that cast cast iron was a difficult metal to test. I know that cast iron is tested by pouring molten cast iron into a test cup with a thermocouple to see the cooling curve and a computer shows the carbon and silicon content in it. This method is too expensive for the home foundry. I also had a aluminium heat sink tested to see what was in it, most metals were in trace amounts but it had a large amount of iron in it. All these samples were expensive to get tested and with the crazy results I think it is a waste of time getting other samples tested.
THe foundry that I have used for metoring has a mass spectrometer that they use every day to analyze their melts. Maybe a foundry near you would have mercy on you and do the testing. All it takes is a minute or two and a clean surface to which a spark is applied and the resulting micro-cloud analyzed. Too bad those machines are so danged expensive. Denis
XRF can't read Carbon so it guesses. They have put in some electronics to help but it's still an educated guess. Each element has 2 spikes on the XRF chart and sometimes the machine thinks the second spike is a different element like Tungsten. Handhelds are only as good as the operator and calibration. You would need a known alloy that is close to what you have to be sure your getting a close reading. Thermo ARL Model 3460 OE Spectrometer, Al Base | eBay Great machines, I run one for 17 years. I used to run all kinds of samples for people but can't do it anymore.
Yep going back to that company is a waste of time. All the iron foundries that were close to me have closed down a long time ago and that is the reason why I pour my own iron castings.
