By now everyone has seen the latest IAEA Iran report, GOV/2012/23. Among other bits that have drawn attention is paragraph 28, concerning swipe samples taken at the subterranean enrichment facility outside of Qom in mid-February. These samples “showed the presence of particles with enrichment levels of up to 27% U-235,” more than the level that’s supposed to be produced there. That’s highly enriched uranium, in fact.

Iran has told the IAEA that “the production of such particles ‘above the target value’ may happen for technical reasons beyond the operator’s control.” While that answer clearly isn’t too satisfying, is it plausible? Could be. I don’t pretend to know what actually happened that resulted in the presence of 27% enriched HEU where it shouldn’t be. But a set of possibilities that don’t involve an attempt at cheating can be discerned, and one or another of them may be reasonably likely. (Caution: wonkish.)

For starters, hidden in footnote 16 of the same report is a reminder of a previous episode of over-enrichment in Iran. It concerns the main enrichment hall at the Natanz facility, which isn’t supposed to produce more than 5% enriched LEU: “A small number of particles from samples taken in the cascade area continue to be found with enrichment levels of between 5% and 7.4% U-235… the Agency assesses that these results refer to a known technical phenomenon associated with the start-up of centrifuge cascades.”

That’d be the “time transient” described in Houston Wood and Stephanos Tongelidis, “Gas Centrifuge Cascade Study for Maximum Assays During Start-Up,” in Proceedings of the 47th Annual Meeting of the Institute for Nuclear Materials Management (2006). In plain English: UF6 gas is fed into a centrifuge cascade slowly at first. Because the centrifuges are doing their usual work on an smallish volume of gas during this time, that initial volume gets over-enriched. Before very long, though, it gets blended down to the target level by the introduction of the rest of the feed gas into the cascade. Thus the “transient” label.

Reflux Redux

The same phenomenon would occur if the operator slowed the feed rate for any reason at any time, not just at startup. The generic term for the condition of over-enrichment due to slow feed is “reflux.” An ACW commenter also observed back in Oct. 2010 that if Stuxnet accelerated the centrifuges beyond their normal speed, the results would look the same as a slowdown of feed.

Left unanswered is why traces would have escaped the cascade during a reflux episode. Normally, traces exit when opening the cascade, dumping its contents, withdrawing samples from it, or switching out a product container. Why do any of these things during the time transient or any other reflux episode? In that sense, the presence of the HEU traces isn’t quite “beyond the operator’s control,” but it seems to be a matter of carelessness. That’s a little surprising for a facility designed to operate just shy of the politically sensitive HEU threshold. Or perhaps not. As noted above, it’s not the first case of over-enriched uranium traces escaping an Iranian centrifuge cascade.

In this instance, Albright, Stricker, and Walrond of ISIS attribute the over-enrichment to an operator error involving the enlarged cascades at Qom; if the original feed rate used for 15-stage cascades was initially used with the new, 17-stage cascades, it would wind up being too slow, resulting in an “overshoot” of the intended product level. (More centrifuges, more separative work, not enough UF6 gas, basically. Call it reflux by default.) In this scenario, which ISIS considers “likely,” the operators might have discovered their mistake when withdrawing samples, which would have spread the suspect traces. Whoops!

I haven’t done the math, but it makes a certain sense. Then again, Iran started using enlarged, 174-centrifuge cascades to produce “up to 20%” enriched uranium at Qom on Dec. 14, 2011. Would the first suspect traces only have emerged a full two months later, on Feb. 15, 2012? No samples taken, containers switched, etc., before then? What’s more, Iran had already been trying out a few 174-machine cascades at Natanz. You’d think they would have worked out these sorts of issues. I suppose one never knows.

Unfortunately, there are always darker possibilities. Like cross-contamination from another, still-clandestine enrichment plant. But 27% enriched HEU is close enough to the “up to 20%” enriched material expected at Qom that this explanation isn’t exactly one that leaps out and whacks you over the head.

Long story short: don’t panic just yet.

Postscript. By now, this issue may have been overshadowed by the hardening of Iran’s negotiating position. In the best case, it will turn out that nuclear diplomacy is also subject to over-enrichment and time transients. Let’s hope for the best…