# Answer to Question #6701 Submitted to "Ask the Experts"

*Category: Instrumentation and Measurements — Surveys and Measurements (SM)*

The following question was answered by an expert in the appropriate field:

In order to evaluate whether the bench is contaminated you (or
someone else) must first make measurements with an appropriate detector
to evaluate the presence of radioactivity. You describe data from wipe
samples, and such samples are valid in many cases for assessing
removable contamination. It is also necessary to evaluate fixed
contamination that may not be removable via wipes. If the radionuclides
that had been used on the bench are radionuclides that can be detected
with portable instrumentation, then such instrumentation should be used
to evaluate the presence of such radionuclides. If tritium (^{3}H)
was a radionuclide of concern, then it might even be necessary to take
some representative samples of the bench surface and to subject them to
analysis. Such analysis could be as simple as allowing the sample to
equilibrate in a liquid scintillation cocktail mix and counting in a
liquid scintillation counter.

Once you have obtained data for fixed and removable contamination, you
can assess the figures to try to determine whether contamination is
present. If your facility is going through an official decontamination
and decommissioning process, you will likely have to abide by specific
release criteria that are predicated upon dose limitations when the
property is disposed of. It does not sound as if this is your
situation, but rather as if you are simply trying to get rid of an
unwanted item. The simplest way to evaluate the possible contamination
levels is to use instrumentation that is adequate to measure the
radiations of possible concern and to compare the readings obtained to
the background count rates for the same instruments.

This appears to be where you are at with respect to the wipe
samples. Regarding such, it is indeed possible for the background
(blank) count rate to exceed the sample count rate even if the sample
has no contamination on it. This is a matter of statistics. If we use
Poisson statistics, the standard deviation in a single count, σ_{C}, is the square root of the count, and the standard deviation, σ_{R}, in a count rate, R, (obtained from a count, C, over a time interval, t) is the square root of the count divided by the time, C^{1/2}/t.

For example, if 30 counts were obtained in a three-minute background count, the standard deviation in the count would be 30^{1/2}
= 5.48, and the standard deviation in the count rate (R = 10 cpm) would
be 5.48/3 = 1.83 cpm. There is approximately a 95 percent probability
that any three-minute background count that we took (so long as the
background did not change) would fall within ± 2 σ_{C}
of the mean count. If 30 counts is a reasonable estimation of the mean
count, and 5.48 cpm is its standard deviation, this means that 95
percent of the repetitive counts would expectedly fall between 19
counts and 41 counts. Thus, if one were counting samples that had no
contamination, one would expect to see counts both above and below the
mean blank count on a regular basis.

The question then becomes: when is an observed count indicative of
the presence of net activity (i.e., contamination)? The question is
perhaps most cogent when the observed gross sample count is greater
than the blank count. (For purposes here, since you have specified
count rates, we will refer to count rates rather than counts.) A
generally acceptable approach to answering this question is to
determine what is called the critical level, L_{C}, which represents a net count rate such that any observed net sample count rate greater than L_{C}
will be assumed to be indicative of the presence of real activity. The
critical level, as defined originally by Currie (Currie LA. Limits for
qualitative detection and quantification determination. Analytical
Chemistry 40(3):587-593; 1968), is given by

L_{C} = k(R_{b}/T_{s} + R_{b}/T_{b})^{1/2},

where k is the number of standard deviations above the mean expected net count rate of zero when no activity is present at which the critical level is specified to lie;

In most health physics applications the value of k is specified as 1.645; for a normal distribution this would imply that the fraction 0.05 (5 percent) of the net counts would fall above the critical level even when no net activity was present in the sample. This level is often specified by what is called an alpha value; thus alpha, α, would be equal to 0.05. By accepting such a value, we are saying that we are willing to accept false positive results 5 percent of the time when evaluating samples that have no net activity. If we do then observe a net count rate above the critical level we would conclude that net activity is present. This can be used as a simple tool to decide, on a reasonable statistical basis, whether activity is present so that we can proceed with whatever action is appropriate.R

_{b}is the background (blank) count rate;T

_{b }is the background counting time, andT

_{s}is the sample counting time.

For the wipe data you have presented we don't have all the information to do exact calculations, but we can do a representative example. Let us assume that the blank count rate was 18.2 cpm and it was obtained by counting the blank for five minutes. Let us also assume that the gross sample count rate was the high value of 20.5 cpm that you gave and let us assume that it was based on a two-minute count. We can estimate the critical level as

L_{C} = 1.645(18.2/2 + 18.2/5)^{1/2} = 5.87 cpm.

The observed sample gross count rate of 20.5 cpm yields a net count
rate of 20.5 - 18.2 = 2.3 cpm. Since this value is less than the
critical level we would conclude that the wipe sample contained no net
activity.

It is important to note that the critical level can be decreased by
increasing the counting time for the sample and/or background. For
example, if we had increased the counting times for both the sample and
the background to 20 minutes we would have obtained a critical level of
2.22 cpm. The net count rate of 2.3 cpm would have just exceeded this
value, leading us to conclude that the sample did contain net activity.
The net count rate can be converted to activity by dividing the count
rate by the appropriate counting efficiency. The determination that net
activity may exist does not necessarily mean that a given item cannot
be disposed of, but such disposal must be consistent with stipulations
and requirements of your regulating group. A proper determination that
no activity exists is justification for disposing of the item as
uncontaminated. Naturally, when the net sample count rate is less than
the critical level, we would assume that no net activity is present in
the sample.

The description of the critical level used above is applicable to
samples that have been analyzed in a laboratory counting system in
which a count is obtained over a measured time interval. The count rate
is obtained by dividing the count by the counting time. If a ratemeter
is used to obtain count rate directly, as might be the case when making
measurements of fixed contamination with some portable instruments, the
approach to calculating the critical is somewhat different. We will not
discuss that here, but if that becomes an issue for you, please post
another question.

There are a number of other statistical methods and criteria that can
be used in making judgments about the presence of activity in
contamination samples. For example, the concepts of the lower limit of
detection and/or minimum detectable activity (MDA) are two related
quantities that are commonly used in decision making. These quantities
are calculated from the critical level and are larger than the critical
level. While they are often used in making decisions about the presence
of activity, I believe it is more appropriate to use the critical level
for such judgments.

You can obtain more information about these and other decision tools by
reviewing some of the questions that have been answered previously on
the Health Physics Society Ask the Experts Web site. If you go to the Experts' Answers page and type "statistics" in the Search box, you will get a list of questions and answers; questions 5958, 2018, 6175, 3644, and 2888, and perhaps others, include discussions and references that you might find helpful. Good luck.

George Chabot, PhD, CHP