You might have a look in the Firestone's Tables

I suppose it is thermal neutron activation? The energy of 161.9 keV can be used for determination of Se in biological samples (77mSe, 17.45 s half-life) when irradiated in Cd-capsule (epithermal mode). I wonder if Se could be detectable in soil samples (usually very rich in Al resulting in high 28Al activity, 2.24 min half-life = high Compton scattering). I will share this question and ask my experienced colleagues...

Thank you for your helpful reply and links. I am still puzzled however. See the attached chart showing data. Counts were collected after pulse activation with acquisition times of 3 sec x10, 10 sec x10, 30 sec x10, 100 sec x10. The chart shows all the peaks found, except for the Al28 escape peaks.

Perhaps you are seeing Bromine-78, which decays by electron-capture to Selenium-78 with 6.45 minute half-life, releasing a 161.9 keV gamma ray. One possible activation pathway could be, you may have Bromine-79 (51% elemental abundance) in your fly ash. If your neutron source is not fully thermalized, you can knock neutrons off the Bromine-79 by the spallation (N,2N) nuclear reaction, forming Bromine-78 with 6.45 minute half-life. Since your signal is dominated by the 1778 keV gamma rays from the Aluminum-28, you are seeing the Bromine-78 signal on top of a large Compton background from the Al-28. It is likely that for the first 8 minutes, the random (statistical) fluctuation in the Al-28 Compton background at your 161.9 keV energy bin in your HPGe detector (or silicon detector?) is larger than the Br-78 peak. Therefore when you use smart software (or "eyeball method") to subtract the Compton background at 161.9 keV, you see nothing but noise at that energy bin until the Al-28 signal has decayed by one order of magnitude. As the Al-28 signal decays, the Compton background decays also, so a subtraction of the difference (Br-78 signal) - (Al-28 Compton background at 161.9 keV) yields a Br-78 signal that appears to be "building in" or increasing over time. The apparent increase in Br-78 signal may be an artifact of your background subtraction. From 10 minutes onward, the Al-28 signal is weak enough that its background fluctuations are smaller than the Br-78 peak you are monitoring. At this point you see the Br-78 peak decaying with a time constant near 5 or 6 minutes, but it does not look like a clean exponential decay because you still have some artifact of the Al-28 background subtraction going on. That would be my guess.

You can probably model this in Excel or Matlab if you want to check my hunch. The isotope data on Br-78 are shown here

http://www-nds.iaea.org/relnsd/vcharthtml/VChartHTML.html

and as an earlier writer pointed out, you can look up the gamma ray energy of the Br-78 decay here

http://www.nndc.bnl.gov/nudat2/indx_adopted.jsp

-Ken

Hmm... this is a discouraging result. I went to

http://www-nds.iaea.org/exfor/endf.htm

and I looked up the cross-section for

n + Br79 -> Br78 + 2n

It turns out the cross-section is vanishingly small for neutron energies below 10 MeV. So my hypothesis above is only feasible if you are using a 14 MeV D-T neutron source or an accelerator neutron source over 10 MeV. Just keep that in mind.

Kenneth, Thank you for your answers. I now have a different explanation, that you might like to consider for me. I believe the peak is actually 77mSe 17.45s that is excited by inelastic (g,g') reaction from the1778 keV photos from 28Al in my samples. It is difficult to simulate the observed build and decay of the activity, but part of the problem may well be due to the high compton background that you mention. In this case one would expect to see the same effect in any sample containing a high Al fraction, together with Se. After looking very hard for all possible routes, I am now convinced that 77Se(g,g')77mSe is the only likely one.

John, It is especially challenging to match your data because you appear to have two time-constants: A "build-in" with a time constant between 6 and 10 minutes, and then a decay with a slightly shorter time constant (4 to 6 minutes?). The obvious thing to look for is a nuclear decay with 6 to 10 minute time constant followed by another nuclear decay with 4 to 6 minute time constant, with the second decay somehow involving release of 161.8 keV gamma rays. I searched up and down the table of nuclides near any nuclide associated with a known 161.8 keV gamma ray, and nowhere on the table of nuclides is there a known chain meeting that requirement: a 6 to 10 minute decay time, followed by a 4 to 6 minute decay path involving 161.8 keV.

This is why I hypothesized the "background subtraction" artifact. There are two or three nuclear decays that yield 161.8 keV with 4 to 6 minute half-life; if any one of those paths could be reached by neutron bombardment of stable isotopes, then it would only be necessary to hypothesize that the "build-in" time constant is actually an artifact of a decaying Compton background instead of part of the same decay chain. This increases the number of possible isotope combinations to nearly infinite (ANY Compton background would do) but in practical terms the 28Al at 1778 so dominates your data that it is the most likely source of any Compton background in your data, and its 2.2-minute time constant is roughly the right order of magnitude to account for a background-subtraction artifact that could fade away in about 8 minutes (four half-lives).

My concern about your explanation using 77mSe (17.45 seconds) is that the 17 seconds is too short to provide the 5-minute decay curve that you see. Are you then proposing that BOTH time constants in your data (the 8 minute build-in and the 5-minute decay) are background-artifact time-constants and not actual nuclear decay constants? I guess that is plausible. If you can show experimentally that the effect happens in a pure selenium/aluminum powder mixture but does NOT occur in a pure bromine/aluminum powder mixture, I guess that would clearly rule out my bromine hypothesis and lend strong support to your hypothesis.

Ken,

If the isotope is being continuously produced by the gamma rays from 28Al, then it would be in secular equilibrium and show the 2.24 min half life. It seems to be close enough. The build-up, on the other hand, must be an artifact as you previously suggested.

John,

I've thought about your hypothesis pretty hard and I cannot make it work. Here are my basic premises and simple conclusions; maybe you can find a loophole in my description that makes it work for your experimental data:

(1) The Al-28 signal is decaying from over 1000 counts per second down to 1 ct/s over a period of about 1300 seconds (this is from your data plot), with half-life 130 seconds.

(2) The Compton background at any particular energy, for example the Compton background from your Al-28 peak as seen in your HPGe detector as a baseline underneath the Se77m peak, will be proportional to the Al-28 peak height. Call that proportionality constant "C1". So at time t=0, you get 1000 counts per second in the Al-28 peak, and you also get (C1 x 1000) counts per second underneath your Se77m peak.

(3) Over time the Al-28 peak drops exponentially, and the Compton background count drops at the same rate. So for example at time t = 500 seconds, the Al-28 peak only has 100 counts per second, and the Compton background underneath your Se77m peak would at t = 500 then be (C1 x 100) counts per second.

(4) If the Compton background were a perfectly flat line, you would have no trouble subtracting it with perfect accuracy, and you would never miss a Se77m count. But backgrounds are not flat because these are random events. So when you plot the Compton background on an X-Y chart (X = energy bin, Y = counts per second), you see what looks like a fuzzy baseline. The mid-height of the fuzzy baseline is your best guess at the true Compton background rate, and the vertical fuzz is random statistical fluctuation proportional to SQRT(N) where N is the number of counts in any particular bin. When you subtract your best estimate of the baseline from your Se77m count, your subtraction error may be off then by as much as (C1 x SQRT(N)). Putting that another way, you cannot see your Se77m peak by eye in all that "fuzzy noise" if the Compton background statistical fluctuation (C1 x SQRT(N)) is greater than the number of counts in your Se77m peak.

(5) Therefore, if the Se77m peak has a LONGER half-life than the Al-28, you could see the peak gradually appear above the background noise as the Al-28 Compton background fluctuations drop according to

Compton_Noise = c1 X SQRT( N1 exp [- t / Tau28])

Se77m = N2 x exp[-t / Tau77],

where t = time in seconds,

N2 = initial (unseen) count rate of Se77m peak (lost in the initial noise),

N1= initial count rate for Al-28 peak

C1 x N1 = initial count rate for Al-28 Compton background under Se77m peak

Tau28 = decay time constant for Al-28

Tau77 = decay time constant for Se77m (or whatever it is).

In the case where Tau77 > Tau28, and N2 < c1 x sqrt(N1),

the signal Se77m will start out lost below the noise threshold, will eventually rise above the noise, and then will decay away back toward zero counts per second.

(6) But if, as you hypothesize, the Se77m signal with 17 second half-life is being excited by the Al-28 radiation, and therefore the Se77m is decaying away in equilibrium with the Al-28, then the situation looks like this:

Noise = c1 x SQRT(N1 exp[ -t / Tau28] )

Se77m = N2 x exp[ -t / Tau28 ]

with the same time constant Tau28 in both exponentials, I cannot find any set of parameters where the Se77m peak (if that's what it is) starts out lower than the noise, then rises above the noise, and then drops toward zero later.

(7) It seems to me that you need at least two different time constants to get the signal that you observe at 161.8 keV, even if you use the plausible hypothesis of a "baseline subtraction artifact": one shorter time-constant over which some part of the baseline decays away, revealing your signal, and then a longer time-constant over which your signal decays.

This just occurred to me: are there other peaks besides the ones you indicated on your graph? The Scandium-46m peak is just way too short I think to give you the baseline artifact model you are looking for, but if you had just ONE more large peak in there, with a decay time constant shorter than Al-28 and long enough to obscure your 161.8 keV line for the first 400 seconds, that would do it...

Good luck! This is what real experimental data is like, isn't it? It reminds me of why my experimental physics doctoral project took 7 years to finish...

-Ken

OK after thinking about this WAY too much, I realize that you'd have to look carefully at the raw spectral data and the fitting method. If there is a "background subtraction artifact" it does not have to be directly proportional to the statistical noise fluctuation on your Al-28 Compton background. If it is not directly proportional (if the artifact is somehow nonlinear), then you could easily have a situation where you background subtraction works badly for the first 400 seconds at very high count rates, and then starts to work well from then on, resulting in a purely mathematical "build-in" (no nuclear time-constant at all, just a function of counts or some algorithmic thingy) followed by the expected physical decay of your signal. I've probably been over-thinking this. Happy gamma spectroscopying!

Kenneth,

Well thank you for thinking so hard about it.

One thing you might have missed it that my counting periods are not all the same length. This affects the statistics for peaks appearing above background. I start with 10 3-sec counts (starting ~2 seconds after a TRIGA reactor pulse, that also triggers the rabbit transfer), then 10 10-sec counts, then 10 30-sec counts, then 10 100-sec counts. After that the geometry is changed to put the sample closer to the detector and then it continues with 10 300-sec counts, 10 1000-sec counts and 10 3000-sec counts. 70 spectra for each sample! Another sample of the same material was irradiated for an hour stead-state and counted for up to 10 days.

At earliest times there was N-16 activity contributing to the Compton counts, but off-scale for the photopeak.

This was a pretty nice set up that I had, but regrettably it no longer exists and the reactor is shut down. These were some of the last data taken at the University of Arizona Research Reactor.

Best Regards,

John

Ah! Yes, perhaps changing the counting periods and the sample/detector geometry could introduce non-proportional Compton background artifacts. That would indeed be difficult to model... Also, since your neutron source was a TRIGA reactor it seems VERY unlikely that you had enough > 10 MeV neutrons to form a measurable quantity of Bromine-78, as I had earlier speculated.

Hopefully your data will still yield some useful information about the Selenium concentration in spite of the complexities of the background subtraction.

-Ken