[ RadSafe ] TFP and Background Correction

[George Stanford]

         Here's a footnote on Mangano's TFP paper.  The extract that 
James Salsman was kind enough to send us contains this passage:

"The spectrometer has special features so that the background
count-rate in the 400 to 1,000 channels is 2.25 plus or minus 0.02
counts per minute. The background has been counted for over 5,000
minutes so that the error associated with the background measurement
is about 1 percent ."

         It looks as though Mangano assumed that by determining the 
average background very accurately, he could make the uncertainty due 
to background correction arbitrarily small -- which is incorrect.
         With an average background of precisely 2.25 c/m, the 
average number of background counts in a 400-minute counting interval 
(which Mangano said he used) would be 900, with a standard deviation 
of  +/- 30, or 3.33%.  Adding the uncertainty in the average -- +/- 8 
counts (0.02 c/m) -- in quadrature, the variance in the background 
correction would be 900 + 64 =- 964, for a standard deviation of +/- 
31.05 counts, or 3.45%, rather than the 1% that Mangano quotes (if I 
interpret what he says correctly).
         In any event, such a mistake is only important when the 
counting rate from the sample is less than, say, 2 or 3 times the 
background rate.  In a quick scan of Mangano's paper, I didn't find 
the magnitude of the observed counting rates.  If the background 
component happened to dominate the count, the uncertainty due to the 
background correction was irreducibly very large.

         Just for the record.

                 George Stanford

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

At 01:39 PM 1/4/2006, James Salsman wrote:
The following is quoted from pp. 23-24 and 49 of:
   http://mtafund.org/prodlib/radiation_health/final_report.pdf

"Upon receipt of envelopes containing teeth, RPHP data managers assign
a unique control number to each tooth and logs it into a computerized
data base. Teeth are periodically sent in batches to REMS, Inc., a
radiochemistry laboratory in Waterloo, Ontario, Canada. Laboratory
personnel document Sr-90 concentrations by separately measuring Sr-90
activity (in picocuries) and calcium mass (in grams) in the teeth.
REMS personnel, headed by radiochemist Dr. Hari Sharma, are blinded
from any information about each tooth.

"A tooth is dried for 12 hours at 110 degrees centigrade, then ground
to a fine powder.  Approximately 0.1 gram of the powder is weighed in
a vial, then digested for a few hours with 0.5 milliliter of
concentrated nitric acid along with solutions containing 5 milligrams
of Sr2+ and 2 milligrams of Y3+ carriers at about 110 degrees
centigrade on a sand bath. The solution is not evaporated to dryness.
The digested powder is transferred to a centrifuge tube by rinsing
with tritium-free water. Carbonates of Sr, Y, and Ca are precipitated
by addition of a saturated solution of sodium carbonate, then
centrifuged. The carbonates are repeatedly washed with a dilute
solution of sodium carbonate to remove any coloration from the
precipitate. The precipitate is dissolved in hydrochloric acid, and
the pH is adjusted to 1.5 to 2 to make a volume of 2 milliliters, of
which 0.1 milliliter is set aside for the determination of calcium.
The remaining 1.9 milliliters are mixed with 9.1 milliliters of
scintillation cocktail Ultima Gold AB, supplied by Packard Bioscience
BV in a special vial for counting. A blank with appropriate amounts
of Ca2+, Sr2+, and Y3+ is prepared for recording the background.

"The activity in the vial with the dissolved tooth is counted four
times, 100 minutes each time, for a total of 400 minutes, with a
Wallac WDY 1220X Quantulus low-level scintillation spectrometer.
The spectrometer has special features so that the background
count-rate in the 400 to 1,000 channels is 2.25 plus or minus 0.02
counts per minute. The background has been counted for over 5,000
minutes so that the error associated with the background measurement
is about 1 percent . The overall uncertainty or one sigma associated
with the measurement of Sr-90 per gram of calcium is plus or minus
0.7 picocuries per gram of calcium.

"The efficiency of counting was established using a calibrated
solution of Sr-90/Y-90 obtained from the National Institute of
Standards and Technology, using the following procedure. The
calibrated solution is diluted in water containing a few milligrams
of Sr2+ solution, and the count-rate from an aliquot of the solution
is recorded in channel numbers ranging from 400 to 1,000 in order to
determine the counting efficiency for the beta particles emitted by
Sr-90 and Y-90. It is ensured that the Y-90 is in secular
equilibrium with its parent Sr-90 in the solution. The counting
efficiency was found to be 1.67 counts per decay of Sr-90 with 1.9
milliliters of Sr-90/Y-90 solution with 25 milligrams of Ca2+, 5
milligrams of Sr2+, 2 milligrams of Y3+, and 9.1 milliliters of the
scintillation cocktail.

"The calcium content was determined using a Varian A-A 1475 atomic
absorption spectrophotometer by flame spectroscopy at a wave length
of 422.7 nanometers, using acetylene plus air as fuel."

Please see also:
   http://dx.doi.org/10.1016/S0048-9697(03)00439-X

Sincerely,
James Salsman
_______________________________________________