top of page

Contaminated Platinum

 

Working in a research center of a conglomerate is a mixed bag. You are tits on a bull as far as the operating divisions are concerned, just a burden assessed by corporate. Some may not even know you exist, apart from their resentful financial guys. And usually your own management has no concern with the factories in return. But that can change when problems come up that affect production and our boss wants to show relevance. With no resources or expertise officially dedicated to manufacturing engineering or similar support our center relies (more or less) on ad hoc groups of generalists—like three or four guys, often including me—to wave the flag occasionally and try to come up to speed. Sometimes it’s handholding, sometimes we commute to help. You might imagine how their engineers take to that. We show up from a place they have vaguely heard of because local management may not trust them to fix it. That’s a problem of perception and we almost always smooth it out. We do our best, but we seldom hit a home run. We usually make a good dent but follow-on relationships are infrequent after they get the bills.

 

However, once in a great while you get a chance to make a splash; this one took only a day after a phone call I got. They had asked for a metallurgist and I was the only one at the time; lucky for both of us I was able to give a quick answer. No genius here, just the one with any experience in high temperature gas-metal reactions I got in a previous job.

 

Think of an aging plant (lots of those in the company) hanging on by making replacement components for an obsolete product in a shrinking market, one red bottom line from closing. Then some bad luck. A new platinum foil assembly necessary to the process got contaminated “somehow” with I forget how many parts per million of chromium, a trivial amount but sufficient to poison a critical part of the product. Even after heating to just below the melting point in hydrogen the Cr wouldn’t budge. Surface treatments didn’t work, either and fabrication of a new assembly was weeks away.

 

The fix was simple and obvious:  heat the assembly in oxygen or air at close to the Pt melting temperature. Oxides of platinum aren’t normally stable—there would be no significant loss of material—but chromium has two common ones that are. Cr2O3 can form a tenacious scale that protects many high temperature alloys from attack in service, but here, if it formed at all, it couldn't be continuous and would convert to the higher oxide, CrO3 The neat part is that CrO3 is a volatile oxide at extreme temperatures, and annealing in air lets it form and then just bugger off. As Cr is lost the oxide the remaining contaminant diffuses down the shallow concentration gradient to the surface until the overall concentration is so low that the process stops. The near-surface concentration should be close to zero after sufficient time although reheating might be necessaryif enough stray Cr finds it way out. Behavior of CrO3 is outside the experience of most metallurgists, but well known to anyone who has worked in high-temperature gaseous corrosion. My guess of a couple of hours at temperature was just that, a guess, but it was more than sufficient. They stuck the assembly in an induction coil and that did it. Minutes might have worked but no one at the factory cared to look beyond the fix.

 

Just to make it look like it took more time than it really did I added a diffusion calculation to the report estimating how long it would take for Cr to diffuse from half-thickness to a final concentration of 1 ppm. Given that the diffusion data for Cr in Pt are really sketchy and initial (and even the final) conditions were vague I wasn't surprised not to be even remotely close to my guess, unless you consider two orders of magnitude "close".

 

I did get some street cred and two attaboys out of it, one in a letter from the factory to our management and another from my physicist boss, in his typical fashion: “I guess you know that metallurgy shit”. Thanks, I think.

bottom of page