Sorry, you need to enable JavaScript to visit this website.

Air Filter Alpha Spectrometry Report

Air Filter Alpha Spectrometry Report

Posted: 7/31/2011 21:05

Back to Main UCB Air and Water Sampling Page

Back to Air Sampling Results Page


BRAWM took an air filter sample in mid-April and performed alpha spectrometry on the air filter. We did not detect any Uranium, Plutonium, or Americium-241, and our calculated detection limits indicate safe levels. Our limits are much higher than the limits set by the EPA in their testing for Uranium and Plutonium. If you would like to skip directly to our limits, please click here.


Radioactivity comes in three primary forms: alpha, beta, and gamma decay. A radioactive nucleus will undergo one or more of these decays, releasing particles with characteristic energies.

The fission product nuclei that are the subject of BRAWM's measurements are beta emitters with associated gamma rays. Beta particles are high energy electrons that do not have discrete energies — they are released in a continuum with an "endpoint" energy that depends on the nucleus. Gamma rays are photons ("particles of light") and have discrete energies. These discrete energies are fingerprints for a specific nucleus, and detecting gamma-rays is what BRAWM specializes in. For example, Cesium-137 has a strong gamma-ray "line" at 662 keV, and that is what we look for in our spectra.

What about alpha decay? Alpha particles are helium nuclei (two protons and two neutrons), and they are very similar to gamma rays in that they are released at specific energies that are fingerprints of a given nucleus. For example, Uranium-238 has two strong alpha lines at 4.151 and 4.198 MeV. Since many nuclei might be alpha emitters but not strong gamma emitters (e.g., U-238 and Pu-239), one must use an alpha detector to detect them.

Alpha spectroscopy is difficult, even when compared to gamma spectroscopy. The primary difference is that since alpha particles have electric charge (+2), even tiny amounts of material between the nucleus and the detector will cause energy loss or complete stopping of the alpha particle. So instead of nice, discrete lines you will find smeared out blobs in your spectrum — or even nothing at all. The more the alpha lines are smeared and blocked, the worse your ability to detect and identify alpha emitters is. Gamma-rays do not suffer from this problem.

Experimental Setup

We placed a piece of filter paper (3M Filtrete) over the nozzle of the shop vac used for our air measurements for three days, April 16–19, 2011. We filtered a total of approximately 5,500,000 liters of air (5,500 cubic meters). At the end of the collection time, the filter paper had turned black from the air particles it filtered from the air.

Within minutes, we took the filter paper and placed it in front of and nearly in contact with a silicon PIN alpha detector. The detector and sample were housed in a sealed chamber that we evacuated so that the air would not block the alpha particles. The chamber was covered with an opaque black cloth so that the ambient light in the room would not create noise in the silicon detector.

Natural Alpha Background from Radon decay products

Within minutes, there were four prominent alpha lines between 5 and 9 MeV. We identified these as coming from Polonium-210 (5.3 MeV), Polonium-212 (8.8 MeV), Polonium-214 (7.7 MeV), and Bismuth-212 (6.1 MeV), all naturally-occurring isotopes that are decay products of Radon gas (specifically, the two isotopes Rn-222 and Rn-220). It is not surprising to see them. Polonium-214 was the brightest line, but it was only visible in the spectrum during the first few hours. This is easily explained by its position in the decay chain of Radon-222 — it is "fed" by the decays of the beta-emitting nuclides Lead-214 (27 minutes) and Bismuth-214 (20 minutes), and it has an extremely short half-life itself (164 microseconds). So Po-214 is very bright for the first hour but disappears when the decay chain is exhausted.

This plot shows the spectrum from the first three hours, overlaid with a spectral model for the four isotopes:

This next plot shows the spectrum for 3 days after the Po-214 has decayed away. The three isotopes Po-210, Bi-212, and Po-212 remain:

After the first 3 days, the Bi-212 and Po-212 entirely disappear. This makes sense because they are on the Rn-220 decay chain after the beta-emitting isotope Lead-212 (10.6 hours), so after a few half-lives of Pb-212 this decay chain is exhausted.

What remains is Polonium-210 (138 days), which is "fed" by another Radon decay product, Lead-210 (22.3 years). So Po-210 remains for a very long time. In fact, its activity has been steadily increasing during our test as more and more Pb-210 decays into Po-210. Here is what the spectrum looks like from three days after the start until presently:

Searching for Uranium and Plutonium Signatures

Plutonium, Uranium, and Americium isotopes emit alpha particles at well-known energies, all in the vicinity of the 5.3 MeV line from Po-210:

Isotope Alpha Energies
U-238 4.1–4.2 MeV
U-234 4.7–4.8 MeV
U-235 4.3–4.4 MeV
Pu-238 5.4–5.5 MeV
Pu-239 5.1–5.2 MeV
Am-241 5.4–5.5 MeV

So the basic idea is to look at the alpha spectrum at those locations and see if there is any excess from those isotopes. A line, if present, should have the same shape as the alpha lines from the naturally-occurring isotopes — i.e., "triangle" shapes with the rightmost edge at the alpha line energy.

I fit the Po-210 spectrum with a model (red line in the previous figure), and subtracted the model from the observed spectrum. It is in this subtracted spectrum that the signatures of the isotopes of interest would lurk. Here is the subtracted spectrum, with black lines denoting the "1 sigma" uncertainties for each bin:

The subtracted spectrum has no clear signatures of any excess counts. These data appear to be random fluctuations around zero. So now we must go about determining what the detection limit (MDA) of this test is.

Calculating the Limits

To infer what this non-detection means, we calculated an upper limit on the total activity concentration for these isotopes.

One very important concept is efficiency. Efficiency is the fraction of all emitted alpha particles that are detected by the detector. There are various independent components to it, which I have conservatively estimated as follows:

Efficiency Estimate Reasoning
Geometric 3.9% This is the ratio of the area of the detector to the area of the filter paper.
Left/Right 50% Half of all alphas will travel away from the detector, and half will travel towards it.
Escape from filter 0.33% This is where the largest uncertainty lies. This was estimated by assuming a 1 micron range for alpha particles in organic material, and noting that the total thickness of the filter paper is 150 microns, with a course mesh covering about 50% of the fine filter paper.
Detection 100% Essentially all alphas that strike the detector should be detected.
Total 0.0064%  

Similar to the efficiency is what I will call the "spectral form factor." To calculate the detection limits, we'll select a certain region of the spectrum, such as 3.2–4.2 MeV for U-238. Since an alpha line will spread out beyond this range, we need to know what fraction should fall in that selected range. Based on the Po-210 line shape, a window 1 MeV wide should yield good sensitivity, but only about 50% of counts would be registered. We will use 1 MeV windows and a spectral form factor of 50%.

Here is some math showing how minimum detectable activity (MDA) is calculated:

Number of Alphas from Source =
Activity Concentration (Bq/L)
    × Air Volume (L)
    × Counting Time (sec)
    × Efficiency (%)
    × Spectral form factor (%)
Activity Concentration (Bq/L) =
Number of Alphas from Source

Air Volume (L) × Counting Time (sec) × Efficiency (%) × Spectral form factor (%)
Background Noise = SquareRoot[ Number of Alphas in Background ]
Minimum Detectable Activity (Bq/L)
2 × Background Noise

Air Volume (L) × Counting Time (sec) × Efficiency (%) × Spectral form factor (%)

For our test, here are the relevant data:

Air Volume: 5,500,000 L
Counting Time: 7,259,912 sec
Efficiency: 0.0064% (estimated)

Here are our estimated Minimum Detectable Activities for each isotope of interest:

Isotope Range of spectrum investigated Number of alphas in background MDA (Bq/L) MDA (aCi/m3) Years of breathing the air to equal dose from one plane flight (5 millirem)
U-238 3.2–4.2 MeV 3,036 < 8.6E-8 Bq/L < 2,300 aCi/m3 > 2.6 years
U-234 3.8–4.8 MeV 4,354 < 1.0E-7 Bq/L < 2,800 aCi/m3 > 1.9 years
U-235 3.4–4.4 MeV 3,445 < 9.2E-8 Bq/L < 2,500 aCi/m3 > 2.4 years
Pu-238 4.5–5.5 MeV 5,047 < 1.1E-7 Bq/L < 3,000 aCi/m3 > 0.7 years
Pu-239 4.2–5.2 MeV 5,588 < 1.2E-7 Bq/L < 3,200 aCi/m3 > 0.6 years
Am-241 4.5–5.5 MeV 5,047 < 1.1E-7 Bq/L < 3,000 aCi/m3 > 0.7 years

Comparison with EPA results

While our limits are comparable to the limits we have set for fission product isotopes from Japan, the EPA reached much lower limits. The EPA performed several tests for Pu and U in the first few weeks after the Fukushima disaster. They issued a report on their findings. In San Francisco, there was a detection of U-238, but no detections of U-234, U-235, Pu-238, or Pu-239. Here are the data, accessed via the EPA query search. MDA was estimated by taking twice the uncertainty (CSU):

Isotope Result (aCi/m3) MDA (aCi/m3)
U-238 14.1±6.3 12.6
U-234 ND 10.8
U-235 ND   8.8
Pu-238 ND   7.0
Pu-239 ND 11.6

So our limits are about 200–300 times higher than the EPA limits. Even though we spent a longer time collecting data, the signal-to-noise is just too poor for us to reach similar limits.

Because of the difficulties inherent in measuring alpha particles from an air filter, the standard procedure is to perform some chemistry on the filter to extract and concentrate the actinides (Uranium, Plutonium, Americium, and others). These elements are then electroplated onto a metal film and placed in front of a detector. Because there is no intervening filter material, these samples have much sharper lines and greater sensitivity (i.e., lower limits) can be achieved.

Since uranium and plutonium have long half-lives, we may consider performing other analyses on our filter sample that could be more sensitive.

More Information

If you would like to know more about alpha spectroscopy, here are a few online references: