For a few days the headlines about Fukushima screamed that Unit 2 and Unit 4’s spent fuel pit “reignited” their nuclear reactions long after March 11th. As a nuke, these kinds of headlines make me cringe with disgust because they’re almost always false. The problem is that there is rarely enough original source information to analyze what is claimed. This instance is different.
The particular headline that a reader brought to my attention is here . After the sheer audacity of the headlines, the next thing that catches one’s attention is that the article is hosted by MIT. That should give it some creds, right? Didn’t work this time. Fortunately the writer cites his source, a paper by a Japanese professor named Tetsuo Matsui. His paper is here .
The Basis for the Claim
Two of the major fission products (the results of the uranium atoms splitting) are Cs-137 and I-131. The fractions of these two isotopes, that is the percentage of fissions that produce them are very well known. fI = 2.88 Ã— 10-2 and fCs = 6.22 Ã— 10-2 for the
cumulative thermal fission yields of U-235, taken from the IAEA data base.
I-131 has about an 8 day half-life so after the reactor has run at a certain power for at least 8 days, the amount of I that is produced equals the amount that decays. This is called secular equilibrium. On the other hand, Cs-137 with its approx 30 year half life continues to build in concentration for the length of time the fuel is in the reactor.
With a little fancy math detailed in the paper, one can compute the amount of I and Cs at the point of shutdown, what he calls point X, to a fair degree of accuracy. After point X, the ratio of I to Cs should decline at a fixed rate set by I-131’s 8 day half-life. Using data supplied by TEPCO he plots that ratio on a semi-log chart called Figure 1 in his paper and reproduced here. The red arrows and numbers are my addition.
Several things I don’t like about this plot. First, the data is noisy. the log scale compresses the noise toward the upper left but even then there is a huge outliers at “1” and another at “2”. I don’t like to see data systematically deviate from the norm as it does at “3”. That those points align with a much different slope than the rest alerts me to possible systematic measurement errors.
This is the first problem – unreliable data. Especially in the days immediately following March 11, the radio-chemlab must have been in a mighty struggle to handle the volume of samples to be counted and to devise ways of counting the extremely hot samples that were coming in. Preventing the hot samples from interfering with the measurement of lower level samples is another problem. This data is OK for its intended purpose – to show the extent of contamination at various points but it is too noisy to base such a controversial conclusion on.
This is the data that he bases his conclusion on, Figure 2 in the paper. In particular the grey dots above the red line. They represent samples taken from the drain sump of Unit 2. The unit 4 data are the pink dots which are all over the place and are worthless in this evaluation.
The principle of Occam’s Razor is that the simplest explanation that fits all the data is usually the correct one. Our good professor failed to apply that principle in several instances. First a critical assumption:
“If there is no strong chemical filtering effect in draining contaminated water from the reactor buildings, it would be difficult to understand the observed anomaly near the unit-2 reactor without assuming that a significant amount of fission products were produced at least 10 – 15 days after X-day.”
The problem with this assumption is that we know from TMI-2 that concrete has a strong preferential absorption for Cs over I. Further, we don’t know what compound form the Cs and I are in. Some will, of course be cesium iodide. But Cs is a very reactive element and would combine with other elements present in the water. Some of those elements include salt from the seawater. Cs would displace the Na in salt to form cesium chloride. That would attack the concrete and from an insoluble sludge. Thus with just a little mental exercise, one can think of several plausible “strong chemical filtering effects” that would preferentially reduce the Cs in the drain contents.
There are many other questions that arise simply because the professor failed at one of the most basic principles of science – be skeptical. Be skeptical of the data until proven valid and be skeptical of the results until validated by using a different method.
One such different method that could easily have been done would have been to calculate the number of fissions necessary to make the amount of excess iodine that he thinks he saw. I’m not going to try to compute that value for this blog post (I’d probably make a math mistake!) but I’d guess that the number of fissions would be in the 10E17 to 10E20 range.
That would not be a quiet reaction. Immense energy would be liberated causing violent boiling around the critical mass. The boiling would displace the water moderator with steam which would shut down the reaction. This process would repeat indefinitely until and unless the mass broke itself apart. The pulsations caused by the steam bubbles forming and collapsing would certainly have been audible outside the reactor building and a plume of steam would erupt.
Depending on how intense the fission reaction might be, this pulsating reaction might go on for days or even become a continuing process. Certainly not something that would go un-noticed.
In any event, doing the number of fissions calculation would have been a great cross-check on his I/Cs ratio calculation. Unfortunately he didn’t do that.
Let’s consider another faulty assumption – that the fission took place only in U-235. The fuel is primarily U-238, enriched from 2 to 3% with U-235. It is the U-235 that fissions with slow or thermal neutrons. Neutrons emitted from a fission are fast. They must be slowed before the U-235 reaction can take place. In this type of reactor, the cooling water performs this moderation. Remove the water and the reaction stops.
As the reactor operates some of the fast neutrons are absorbed by the U-238 and after a short decay chain become Pu-239. Pu-239 in turn can absorb a neutron and become Pu-240. Both isotopes are thermally fissionable. That is, they fission when a slow neutron strikes the nucleus. As the reactor runs, the concentration of Pu builds until at the point the fuel is ready to be removed during a refueling outage in excess of half the energy produced is from fissioning Pu.
The fraction of I and Cs produced from Pu is different than that produced by U-235. Therefore the assumption of only U-235 fission is incorrect. Unit 2’s core was an old core so many of the fissions would have been from other than U-235.
Another thing to consider is the lack of other fission products detected. A reaction large enough to make that much iodine would also make a large quantity of radioactive noble gases. None were detected. If you look here  you will see that the site is ringed with area radiation monitors. The readings for each day since the event are there for the downloading and study. I went back and looked at the data around the time period claimed for the criticality event and saw nothing other than the slow decay of fission products released by the initial event. This right here is a big nail in the coffin of the idea of a spontaneous criticality event.
Improbability of Spontaneous Fissions
This is the real biggie. A reactor designed in the free world is designed such that the nuclear reaction will only take place under precise conditions. The fuel rod spacing has to be just right. The water moderator has to be present. And of course, the control rods have to be withdrawn. Most nuclear engineers, at least the ones I’ve talked to, consider spontaneous criticality in a damaged core to be a very unlikely event. Almost impossible. Let’s look at some reasons.
As the center of the core melted, it made a mass of fuel pellets partially cemented together with molten fuel and control rods. The control rods in this type of plant contain boron carbide in powder form inside stainless steel tubes. These rods were fully inserted from the initial scram and so would be intimately mixed with the fuel. In the portions of the core where the temperature was great enough to melt the ceramic uranium oxide, it was also hot enough to melt the stainless steel control rods. Thus neutron absorbing boron carbide powder. The melting point of Boron carbide is higher than that of uranium oxide so it would be deposited as a powder throughout the “rubble bed” of fuel debris. This alone is enough to prevent criticality.
The molten and partially molten fuel forms a dense, though somewhat porous mass. This displaces most of the water that is necessary for the pile to sustain criticality. No water, no criticality.
There is evidence that the cores of all three reactors remained dry for several days after the earthquake. Again, no water, no criticality.
TEPCO personnel were adding neutron absorbing boron in the form of boric acid to the water being injected into the reactors. So when the cooling water finally reached the damaged cores, it contained sufficient boron to prevent criticality.
As I mentioned above, there was no detection of any other fission products above the quantities expected from the initial event.
No evidence other than that the one calculation made by the professor points to any possibility of a criticality event. The professor, probably wanting another publication for his resume, “pencil whipped” a few numbers, concluded that they looked OK and put the paper out there. But instead of sending it through the normal peer-review channels, he made it available to the media. Shades of Fleischmann and Pons of cold fusion infamy.
Until some other evidence of a criticality excursion presents itself – and none has to my knowledge – one must come to the conclusion that the professor is simply wrong in his conclusions.