Running Neutron Activation Analysis on Hair Samples
Hair is actually a pretty useful matrix for trace element work. Unlike blood or urine, which reflect only recent exposure, hair grows in at about 1 cm per month and traps whatever's circulating in your system at the time. That means you can get a timeline if you section the strand properly. It also means contamination is a real problem, and I've spent more hours than I care to admit arguing with people who didn't understand how much handling their samples had already gone through. The basic procedure is straightforward enough. You collect a hair sample, usually from the posterior vertex, and you section it by proximity to the scalp to build a exposure history. Then you clean it to remove surface contamination. After that, you dry, homogenize, and package the sample in high-purity quartz or polyethylene vials. Irradiation happens in a nuclear reactor thermal neutron flux, typically somewhere between 10^12 and 10^14 neutrons per square centimeter per second. Gamma spectroscopy follows after a decay period that depends entirely on which isotopes you're chasing. Short-lived ones like sodium-24 need a different countdown than something like arsenic-76 or mercury-203.
Neutron Activation Analysis Hair: What You Actually Get Out of It
The strength of this approach is the multi-element capability. A single irradiation and measurement run can give you concentrations for fifteen to twenty elements without any digestion or chemical separation. That's one of the reasons labs still use it despite the fact that access to a research reactor is genuinely inconvenient. You don't need to prepare samples with acids. You don't need column separations. The matrix is mostly just carbon, hydrogen, nitrogen, and oxygen, which are transparent enough that your gamma peaks stand out clearly. What you don't get is infinite precision on everything. Detection limits vary wildly by element. Sodium and chlorine are easy because they produce strong gammas and don't suffer much from background. Trace metals like selenium or zinc sit in a harder range and their detection limits depend heavily on your detector efficiency and how long you can wait before the short-lived interferences die down. Lead is particularly tricky because the common lead isotopes aren't easily activated. You'll see results, but they're often near the edge of what's meaningful. If you need sub-ppb lead, I'd suggest inductively coupled plasma mass spectrometry instead. I should mention the sectioning problem because it comes up constantly. When people send me hair for analysis, they sometimes ask me to date the exposure to within a week based on a single centimeter segment. That's not realistic. Hair growth rates vary between individuals and even along the length of a single strand. Temperature, health, and geography all shift the rate. I usually tell people that one centimeter is approximately one month, give or take a week, and that's it. If they need tighter dating, they should look at other evidence, not expect the hair to volunteer precise calendar dates.
Another thing that catches people out is the cleaning protocol. You can't just wipe the hair. Surface contamination from shampoos, dyes, environmental dust, or even the person who pulled the sample is stubborn. The standard rinse sequence involves an external wash with a detergent solution, a deionized water rinse, an organic solvent wash like acetone or methanol to strip cosmetic residues, and another water rinse. The catch is that these solvents can also leach some elements from the outer cortex if you leave them in too long. I run the acetone step for about two minutes and no longer. Going longer noticeably depresses the measured concentration of softer elements like lithium and boron. Here's the edge case I mentioned. A few years ago a forensic lab sent me a batch of hair samples from a homicide investigation. The results came back showing elevated thallium in the distal portion but normal levels near the root. On paper this looked like poisoning months before death. I was about two sigma above the detection limit, which sounds dramatic until you realize that thallium has a pretty low activation cross section and the gamma peak at 337 keV sits right under a Compton continuum from the massive sodium signal. I re-irradiated a smaller aliquot with a longer count time and a different cooling period, and the thallium peak dissolved into noise. The initial reading was a statistical fluctuation. The subject died of natural causes. That one cost me a week of recalibration and a very awkward phone call with the detective. If you're setting up this kind of analysis yourself, or sending samples out, keep these practical points in mind. Use certified reference materials, ideally NIST hair SRM 797 if your lab accepts it. Run a blank along with every batch to check for cross-contamination in the packaging. Document the collection method because chain of custody matters more when you're working at trace levels. And don't send me samples that have been dyed, bleached, or permed recently without telling me, because the chemical treatment changes the porosity and yourCleaning will behave differently than expected.
Get the Full Details
The biggest limitation nobody talks about is reactor access. If your facility doesn't have a neutron source, you're either shipping samples elsewhere or using an alternative. Some labs use portable neutron generators, but the flux is orders of magnitude lower, which means longer irradiation times and poorer detection limits. For routine screening it might suffice. For the kind of work where you need to distinguish between background and a real anomaly, a research reactor is still the gold standard. There's no good substitute if you want the full multi-element picture. One more thing. When you report the results, include the section length and the estimated growth rate you assumed. I've seen too many reports that just say "hair analyzed" with no metadata. That makes the data useless for anyone trying to interpret it later. A concentration number without the context of when and how it was collected is just a number. It's not evidence until you attach the chain.
