UCRL-ID-119658

Forensic Radiochemistry of
PUBLIC Site Inspection Samples

Kenton J. Moody
Nuclear Science Applications Section
Isotope Sciences Division
Chemistry & Materials Science

January 5,1995
.. "

This is an infonnal report intended primarily for internal or limited external

distribution. The opinions and conclusions stated are those of the author and may
or may not be those of the Laboratory.
Work perfonned under the auspices of the u.s. Department of Energy by the
Lawrence Livennore National Laboratory under Contract W-7405-Eng-48.

DlSTAtSUTION OF THIS DOCUMeNT IS UNLIMITED

\
 DISCLAIMER

This document was prepared as an account of work sponsored by an agency of

the United States Government. Neither the United States Government nor the
University of California nor any of their employees, makes any' warran!y,
express or implied, or assumes any legal liability or responsibllity for tlie
accuracy, completeness, or usefulness of any information, apparatus, product,
or process disclosed, or represents that its use would not infringe privately
owned rights. Reference herein to any specific commercial products, process,
or service by trade name, trademark, manufacturer, or otherwise, does not
necessarily constitute or imply its endorsement, recommendation, or favoring
by the United States Government or the University of California. The views
and opinions of authors expressed herein do not neCessarilr state or reflect
those of the United States Government or the University 0 California, and
shall not be used for advertising or product endorsement purposes.

This report has been reproduced

directly from the best available copy.
Available to DOE and DOE contractors from the
Office of Scientific and Technical Information
P.O. Box 62, Oak Ridge, TN 37831
Prices available from (615) 576-8401
Available to the public from the
National Technical fnformation Service
U.S. Department of Commerce
5285 Port Royal Rd.,
Springfield, VA 22161
 DISCLAIMER

Portions of this document may be illegible

in electronic image products. Images are
produced from best available original
document .

. -.,..-' ,,,:
 Forensic Radiochemistry ofPUBUC Site Inspection Samples
Kenton J. Moody
Isotope Sciences Division
Lawrence Livermore National Laboratory
Livermore, CA 94551

INTRODUCTION
For the past two years, the Isotope Sciences Division (formerly Nuclear
Chemistry) has been developing a program to extract forensic information from
samples of plutonium or highly-enriched uranium. In the case of Pu, it is
possible to determine the date of chemical separation, the date of its casting as
metal, the enrichment of the uranium starting material, the length and perhaps
other details (neutron spectrum and fluence) of reactor irradiation, the
reprocessing technique, and clues to the identity of a specific facility. For
enriched uranium, information is possible on the detailed timeline of material
production, including the date of enrichment, whether the plant feed was
formerly-irradiated uranium, the date of final purification; and facility-specific
clues. For more information, see "Dissolved Oralloy Standards and the Origin of
HEU", by Ken Moody, LLNL report UCRL-ID-117611, April 13, 1994.
It would be valuable to be able to apply the same techniques to materials of
opportunity associated with clandestine or illicit nuclear weapons production.
However, interpretation of the radionuclide contents of dilute samples containing
small or finely-divided amounts of Pu and U is more complicated than a similar
analysis of a bulk sample; the isotopics of the sample- may be associated more with
the plant waste stream than with product, and chemical effects of the
environment after dispersal of the sample alter its properties. Nevertheless, it
was decided to analyze interesting samples associated with the U.S. weapons
production facilities and determine how much forensic information was
contained therein. In support of this endeavor, samples from the recent PUBLIC
site inspections of two uranium enrichment facilities were obtained.
It is possible that the U.S. gaseous diffusion facilities at Portsmouth, OH and
Paducah, KY will be subject to international inspections for safeguards and
accountability purposes. The question arose whether an inspector with a covert
agenda could surreptitiously learn classified information from the analysis of
process debris found at these sites. To explore that possibility, the LLNL Forensic
Science Center (FSC) was asked to undertake "counter-forensic" sampling in
order to evaluate the potential magnitude of the risk to classified technology.
Project PUBLIC (Portsmouth/Paducah Uranium Barrier: Livermore
Investigative Campaign) was designed to apply state-of-the-art analytical
technologies to interrogate wipes and other samples from areas of the plants
where converters were disassembled or barrier was destroyed.

MASTER
- 1 - DtSTRU3UTlON OF THIS DOOUMENT IS UNL1Mmm
f"d
 While the radiochemical analysis given in this report does not relate directly to
the original purpose of PUBLIC, collateral radioanalytical chemistry provides
background information about the activities at the plants and the time scale of the
operations. Such knowledge could prove valuable in placing deductions about the
diffusion barrier within a context that might support the conclusions that are
drawn. As a result, this work also appears as a chapter in the final (classified)
PUBLIC report ("Project PUBLIC: Counter-forensic Investigation of U.S. Gaseous
Diffusion Plants", E. Randich et al.,1995; in process).

SAMPLE SELECTION
While the PUBLIC inspection team had no prior knowledge of the classified
aspects of the diffusion barrier, they did have the unclassified information that
the barrier material was formed by sintering a nickel powder. Thus, while
several samples were collected by scraping loose debris into plastic bottles with
spatulas, or by wiping interesting surfaces with moist filter paper, most of the
samples were collected magnetically. The team concentrated on collecting
specimens from confined or sheltered areas where debris could have collected
over a period of years.

A total of approximately forty samples were collected in the two site inspections.
These samples were returned to LLNL where they were subjected to an initial
pre screening (including a radiation survey, an X-ray fluorescence measurement,
and examination under an optical microscope) to identify potential barrier
materials. Samples containing significant amounts of nickel were divided into
two or more subsamples by segregating dissimilar debris components.

The radiochemical analyses are labor-intensive; it was therefore impossible to

even consider treating most of the 69 samples generated by PUBLIC.
Furthermore, the analyses are destructive, resulting in the total consumption of
the selected samples. Consequently, it was decided that no more than five
samples would be radiochemically analyzed, and that only one of these samples
would be a "pure" Ni specimen selected as candidate diffusion barrier material.
The chosen samples would have fairly high radioactivity contents, as measured
by initial sample prescreening. With these considerations, the specimens listed
in Table 1 were selected.

,\ , : 1. -,. I 1':·lj(", i, ·f

- 2 - -
Table 1. Samples Selected for Radiochemical Analyses, Portsmouth and Paducah
Site Inspections.

Sample LD. Weight Description

94-7-14 (PO-7) 7.0 g Ni flakes collected with a spatula
into a polyvial; from Building X-705
catwalk; Portsmouth.
94-7-17 (PO-9) lAg Magnetic flakes, both rust and Ni,
collected from the top of an I-beam
approximately 10 feet from the floor;
Building X-705, basement cutting area,
east wall; Portsmouth.
94-7-19 (PO-II) 3.3 g Magnetic grit, mostly Fe, collected
from groove in floor around railroad
tracks; Building X-705; Portsmouth.
94-7-38 (pK-4) 4.9 g Loose debris, perhaps some Ni
flakes, collected with spatula from
bracket on support E8, approximately
15 ft from floor; Building C-400;
Paducah.
94-7-45 (PK-7) 20.5 g Magnetic grit, with most of the Ni
removed, collected along concrete/
transite interface, about 8 feet from
floor; radioactive waste area, Building
C-400; Paducah.
Blank CX) 4.7 g Radiochemistry blank constructed from
2.93 g Fe203 + 1.73 g Ni20 3.xH2 0(77.0% Ni)
+ 0.055 g CaO.
Sample 94-7-14 was very rich in candidate barrier material, and was the source of
94-7-15, which was extensively analyzed and is described in detail in the PUBLIC
report. The Blank sample was constructed to check for cross contamination
among the samples during the analyses; it was treated with the same reagents at
the same time as the other samples, and should return completely negative
results for radionuclide content.

SAMPLE DISSOLUTION AND XRF RESULTS

As the first step, a sample must be totally dissolved. The dissolution must be
complete to avoid fractionating the elements from one another (which can result
in improper age determinations) and to yield a uniform sample that can be
quantitatively analyzed for gamma-ray emission and by XRF. As anticipated, the
dissolution of samples containing nickel, iron, dirt, paint flakes, plastic shards,
oil, broken glass, hair, and (in one case) a dead insect, can be formidable. The
following procedure was used on the PUBLIC samples:

- 3 -
The samples were poured into Erlenmeyer "reaction" flasks, and the original
sample containers were sequentially washed with water and with 8M HNO a to
ensure quantitative transfer. Samples were heated gently until reaction was
complete. A few drops of 9M HCI were added to each reaction flask, and the
samples were gently evaporated to moist residues. Several times over the next few
days, 9M HCI was added to the samples, which were then evaporated to moist
residues. 4M HCI was added to each sample, and the supernatant solutions were
filtered through Whatman #42 paper into fresh Erlenmeyer flasks which were to
serve as primary containers for the analytical solutions. Several times over the
next two days, 8M HNO a was added to the residues in the reaction flasks, which
were evaporated to moist deposits. Next, 2 ml of7.5M HCI0 4 and several ml of 8M
HNO a were added to each flask, and the resultant mixtures were evaporated until
the evolution of perchloric fumes was observed. 4M HCI was added to the
samples, and the supernatant solutions were filtered into the Erlenmeyer
primary containers. Several ml of 8M HNO a and 1 ml of cone HF were added to
the residues in the reaction flasks, which were then evaporated to moist deposits.
Three ml of 7.5M HCI04 were then added to each reaction flask, and the samples
were evaporated through perchloric fumes to dryness. The reaction flasks were
then heated vigorously with a Bunsen burner to volatilize SiF4, Several ml of 4M
Hel were added to each flask, and the resultant solutions were filtered into the
Erlenmeyer pots. The filter papers were washed with 4M Hel, and were removed
from the funnels and placed into the appropriate reaction flasks. Several ml of
8M HNO a and 2 ml of 7.5M HCI0 4 were added to each flask and the samples were
heated until the filter papers were destroyed with the evolution of copious
perchloric fumes. The contents of the flasks were quantitatively transferred to
centrifuge cones with 4M Hel, the solutions were centrifuged, and the supernates
carefully decanted into the primary containers.

Four of the six samples were completely dissolved, but several milligrams of
intractable white powder remained of samples 94-7-38 and 94-7-45 (the two
Paducah specimens). XRF showed that this material was mostly titanium, with
some minor zirconium, and was therefore probably residual paint chips. It was
decided to treat these residues separately from the bulk samples, under the
assumption that only a small fraction of the total radionuclide content was
entrained in the precipitates. The residues were transferred to Ni crucibles with
a minimum amount of dilute Hel and were evaporated to dryness. Several grams
of NaOH and 1 gram of Na202 were added to the crucibles, which were heated to
melt the mixtures. Twenty ml of water were added to the molten samples; after
cooling, volumes were doubled with 9M Hel and the supernatant liquids were
transferred to fresh Erlenmeyer flasks. Several ml of 4M HCI were added to the
residues in the crucibles; after heating for several minutes, the supernates were
decanted into the flasks containing the previous solutions. The residues were
heated to dryness, 0.5 g of (NH4)2S04 and 4 ml of cone H 2S0 4 were added to each
crucible, and the mixtures were heated for several hours. The resulting solutions
were diluted with 4M Hel and transferred quantitatively to the flasks containing
the previous supernates. Only a trivial amount of residue was left in the

-4 -
crucibles. The titanium solutions were analyzed with an abbreviated version of
the analytical procedure given below, and were found to contain less than 1% of
the Th, U, Np, and Pu determined in the main samples. Analytical results given
in Table 5 (see below) for samples 94-7-38 and 94-7-45 have been corrected for these
losses.
The final analytical working solutions are described in Table 2. Aliquots of 60
microliters of each solution were dried on I-inch diameter Whatman #541 filter
papers and subjected to XRF analysis. The results for major constituents, scaled
to the total solution volume, are also given in Table 2. In many cases, the sum of
the measured masses of these constituents is significantly lower than that of the
original sample; the reason is that XRF is insensitive to low-Z materials like C,
AI, and Si, particularly in the presence of significant amounts of CI from the
solvent. In the case of the Blank, however a direct comparison of the measured
XRF values with the known materials input gave good agreement (1940 mg Fe
measured by XRF vs. 2050 mg added, and 1240 mg Ni measured vs. 1330 added).
The explanation for the minor amounts of Cu and Zn measured in the blank is
their presence as contaminants in the reagent compounds used for solution
synthesis.

Table 2. Description of Dissolved Samples and XRF results. .XRF uncertainties

are ± 10%.
SamIlle I,D. Total Volume Color XRF Comllosition
94-7-14 70ml green 6550 mg Ni, 6 mg Cu.
94-7-17 75ml yellow 725 mg Fe, 225 mg Ni,
4mgCu, 3 mgZn, 6mgU.
94-7-19 100ml orange-tan 2220 mg Fe, 145 mg Ni,
68 mg Cu, 8 mg Zn,
4 mgPb, 14 mgU.
94-7-38 . 150ml orange 1500 mg Fe, 108 mg Ni,
13 mg Cu, 75 mg Zn,
16 mgPb, 32 mgU.
94-7-45 215ml dark brown 8600 mg Fe, 1680 mg Ni,
204 mg Cu, 337 mg Zn,
9 mg Pb, 183 mg U.
Blank 75ml olive 1940 mg Fe, 1240 mg Ni,
7 mg Cu, 1 mg Zn.

- 5 -
ALIQUOTING AND TRACING
Aliquots of each working solution were taken for different aspects of the
radiochemical analyses. The 60-microliter aliquots for XRF analysis were
indicated above, and other aliquots are listed in Table 3. Gamma-spectrometry
aliquots were transferred to plastic counting vials and sealed in Saran wrap.
This is a standard counting geometry for LLNL's Isotope Sciences counting
facility. All other samples were delivered into 40-ml centrifuge cones.
Unspiked samples were not traced with radionuclides. To "Spike Mix P"
solutions were added standard aliquots containing about 175 dpm 236pu, 30 dpm
232U (with 228Th and daughters in equilibrium), and 440 dpm 243Am (with 239Np
in equilibrium). To "Spike Mix N" solutions were added standard aliquots
containing about 44 dpm 243Am and 820 dpm 237Np (with 233Pa in equilibrium).
Comparison of measurements of both spiked and unspiked samples results in
chemical yield determinations and quantitative, absolute data for radionuclide
atoms per ml of original solution.

Table 3. Aliquot information for radiochemical analysis.

Sample J.D. Gamma-Spec Unspiked Spike Mix P Spike Mix N

94-7-14 25ml 15ml 10ml 10ml
94-7-17 25ml 20 ml 10ml 10ml
94-7-19 25ml 20 ml 20ml 20ml
94-7-38 25 ml 25 ml 25ml 25ml
94-7-45 25 ml 30 ml 25ml 25ml
Blank 25 ml 20 ml 10ml 10ml

GAMMA-SPECTROMETRY RESULTS

The gamma-spec aliquots were counted with large-volume germanium detectors

for emitted photons having energies between 50-2000 keY. Each data-acquisition
period was between 3 and 6 days in length. U sing detector calibration
parameters, gamma-ray peaks were identified with the proper energies and were
corrected for detector efficiency to yield resultant photon intensities. The only
radionuclides detected in the samples with this technique were 235U, 238U, and
237Np. The number of atoms thus determined in each gamma-spec sample are
listed in Table 4.

It was somewhat surprising that the 237Np contamination level was so high,
particularly in the Paducah samples, where the radioactivity due to decaying
237Np was comparable to that from uranium. From the radionuclide assay of the
Blank specimen, no significant cross contamination among the samples was
evident. The uranium measurements allow the calculation of enrichment

- 6 -
values, given in the last column of Table 4. All of the Portsmouth samples are
somewhat higher in 235U than is natural uranium (0.71%), while both the
Paducah samples are depleted in 235U.

The gamma-spec results are presented here separately, instead of combined with
the more precise results of the radio analytical chemistry procedure described
below, for two reasons: (1) The results were available before chemical processing
was begun and were the first clear indication that procedures developed for the
measurement of tracer-level actinide nuclides would need modification to
accommodate the multi-milligram quantities of uranium present in the
analytical samples; and (2) the data presented in Table 4 also give an indication of
the limitations of the precision of radio nuclide determinations via non-destructive
gamma-ray spectrometry.

Table 4. Results of Gamma-Spec Analysis of Working Solution Aliquots.

Sample U-238 atoms U-235 atoms Np-237 atoms U-235 enrich.

94-7-14 (9.05±2.00)E17 (1.75+O.07)E16 (1.29±0.27)E13 (1.90+0.43)%
94-7-17 (5.39+0.79)E18 (5.35+0.13)E16 (3.07±0.48)E13 (0.98+0.15)%
94-7-19 (1.10+0.14)E19 (3.67+O.06)E17 (3.23+0.41)%
94-7-38 (1.81±0.23)E19 (5.50±0.18)E16 (l.31±0.02)E15 (0.30+0.04)%
94-7-45 (6.92+0.80)E19 (2.46+O.05)E17 (2.99+0.04)E15 (0.35+0.04)%
Blank

RADIOANALYTICAL FORENSIC PROCEDURES

The analytical samples described in Table 3 were evaporated to dryness. Each

sample was then dissolved in 9M HCI and evaporated to dryness again to expel
residual HN0 3. Each sample was dissolved in 9M HCI with 0.5 ml of conc HI and
evaporated to dryness, then was dissolved in 9M HCI with 0.5 ml of conc HN0 3
and evaporated to dryness. This step was then repeated; the oxidation! reduction
cycle ensures that any added radiotracers are equilibrated with the sample.

Each sample was dissolved with 6M HCI, and Fe 3+ was extracted into hexone.
The hexone was then back-extracted with 6M HCI. These steps were repeated
with fresh hexone until the organic phase became colorless, indicating little iron
was left in the sample. The 6M HCI washes were recycled to the appropriate
samples.

To each fraction were added 2 drops of 10 mg/ml Fe3+ carrier, 1 drop of 10 mg/ml
La3+ carrier, and 1 drop of 10 mg/ml Al3+ carrier. The samples were evaporated
to dryness and dissolved in minimum volumes of 1M HCI. The solution volumes
were then doubled by the addition of cone NH4 0H, and the samples were digested
overnight in a warm-water bath. After centrifugation, the supernatant liquids
were decanted to waste, the precipitates were contacted with water, and 1 ml of

- 7 -
conc NH 4 0H was added to each. After digesting for an hour, the samples were
centrifuged, and the supernatant liquids were decanted to waste. The hydroxide
precipitates were dissolved in a minimum volume of 9M Hel, the solution
volumes were doubled with addition of conc Hel, a few drops of 8M HNO a were
added to each sample, and the solutions (absent most of the original Fe, Ni and
Cu) were stored until needed.

The analytical solutions were loaded onto Dowex-1x8 anion exchange columns, 1
cm in diameter by 10 cm long, and the eluents were collected as An fractions.
Subsequent column washes with 9M HCI were combined into these An fractions.
Plutonium (Pu) was then eluted from the columns with a warm solution
composed of 1 part by volume of conc HI to 12 parts of 10M HCI. Protactinium (Pa)
was then eluted with a solution of composition 9M in HCI and 0.02M in HF.
Neptunium (Np) was eluted next with a solution that was 4M in Hel and 0.1 Min
HF. Finally, uranium (U) was eluted with 0.5M HeI. All of the fractions were
evaporated to dryness prior to further processing.

Np fractions (from Spike Mix P samples only) were dissolved in 10M HCI with a
drop of nitric acid, heated gently, and loaded onto 4-mm-diameter Dowex-1x8
anion-exchange columns. The resin beds were washed with 10M HCI, then with
8M HNO a (until color from transition metals was gone), then with 9M HCI.
Residual Pu was then removed by washing the column with a warm solution of 1
part conc HI and 12 parts 10M HCI, followed by Pa removal with a solution 9M in
HCI and 0.02M in HF. Np was then eluted with a solution 4M in Hel and O.lM in
HF. The Np fractions were evaporated to dryness, dissolved in 8M HNO a, and
loaded on 2.5-mm-diameter Dowex-1x8 columns, which were washed sequentially
with solutions of 8M HNO a, 9M HCI, and warm 1-to-12 conc HI/10M HCI. Once
again, Np was eluted with a solution 4M in HCI and O.lM in HF, and the Np
fractions were evaporated to dryness. The Np fractions were dissolved in 8M
HNO a and loaded onto 2.5-mm-diameter Dowex-1x8 columns, which were washed
sequentially with 8M HNO a and 9M HCI. Np was eluted with 0.5M HCI, and the
solutions were evaporated to dryness.

Pa fractions "(from Spike Mix N samples only) were dissolved in 8M HNO a and
were loaded onto 6-mm-diameter Dowex-1x8 anion-exchange columns. The resin
beds were washed with 8M HNO a (until color from transition metals was gone),
then with 9M HeI. Pa was eluted with a solution 9M in HCI and 0.02M in HF.
The Pa fractions were evaporated to dryness, dissolved in 8M HNO a, and loaded
onto 2.5-mm-diameter Dowex-1x8 columns, which were then sequentially washed
with 8M HNO a, 9M HCI, and warm 1-to-12 conc HI/10M HCI. Pa was then eluted
with 9M HClIO.02M HF, after which the fractions were evaporated to dryness.

U fractions (from Spike Mix P and Unspiked samples only) were dissolved in 8M
HNO a and were loaded onto 4-mm-diameter Dowex-1x8 anion-exchange columns.
The resin beds were washed with a small volume of 8M HNO a, then with 9M HCI.

- 8 -
 Residual Pu was removed by washing the column with a warm solution of 1 part
cone HI and 12 parts 10M HCI, followed by residual Pa and Np removal with a
solution 4M in HCI and O.lM in HF. U fractions were eluted with O.5M HCI, and
were then evaporated to dryness. The U fractions were dissolved in 8M HNO a and
were loaded onto 4-mm-diameter Dowex-1x8 columns, which were then
sequentially washed with 8M HNO a, 9M HCI, and 4M HClIO.1M HF, after which
the U fractions were eluted with O.5M HCI and evaporated to dryness. U fractions
were then dissolved in 8M HNO a and were loaded onto 4-mm-diameter Dowex-1x8
columns, which were then washed with 8M HNO a and with 9M HCl. Final U
fractions were then eluted with O.5M HCI and evaporated to dryness.
Pu fractions (from all analytical samples) were dissolved in 8M HNO a• To each
fraction was added a drop of saturated NH4N0 2; the samples were heated briefly,
then loaded onto 4-mm-diameter Dowex-1x8 anion-exchange columns. The resin
beds were washed with 8M HNO a, then with 9M HCI, after which Pu fractions
were eluted with a warm solution of 1 part cone HI and 12 parts 10M HCl. The Pu
fractions were evaporated to dryness, dissolved in warm 8M HNO a with a drop of
saturated NH4N0 2, and the previous ion-exchange step was repeated using a 2.5-
mm-diameter Dowex-1x8 column. The resultant Pu fractions were evaporated to
dryness, dissolved in warm 8M HNO a (NH4N0 2), and loaded onto 2.5-mm-
diameter Dowex-1x8 columns, which were washed sequentially with copious 8M
HNO a U decontamination) and 9M HCl. Final Pu fractions were then
eluted with O.5M HCI and evaporated to dryness.
An fractions (from all analytical samples) were dissolved in 8M HNO a. To each
fraction was added a drop of saturated NH4N0 2; the samples were heated briefly,
then loaded onto 6-mm-diameter Dowex-1x8 anion-exchange columns. The resin
beds were washed with 8M HNO a; the load and wash solutions were collected as
An fractions (containing Am, Ac, and Ra). Thorium (Th) fractions were then
eluted from the columns with 9M HCI and evaporated to dryness (see below). An
.... fractions were evaporated to dryness several times with addition of 9M HCI (to
remove nitrate ion), after which they were dissolved in water and loaded onto 6-
mm-diameter Dowex-50x4 cation-exchange columns. It often took several water
additions followed by column loadings to completely dissolve the samples. The
resin beds were then washed with copious O.5M HCI and then with 1M HCI, after
which An fractions were eluted with 9M HCl. The An fractions were evaporated
to dryness, dissolved in water, and loaded onto 4-mm-diameter Dowex-50x4
columns, which were washed sequentially with O.5M HCI, 1M HCI, and 2M HCl.
At the first indication of 241Am alpha activity in the 2M HCI eluents, An fractions
were eluted with 6M HCl. After the An fractions were evaporated to dryness, the
last step was repeated with 2.5-mm-diameter Dowex-50x4 columns. The resultant
An fractions were evaporated to dryness. An fractions derived from the Spike Mix
P samples were further processed as follows: The fractions were dissolved in
saturated (nominally 13M) HCI and loaded onto 2.5-mm-diameter AG MP-50
cation-exchange columns. Americium (Am), radium (Ra), and

- 9 -
lanthanum/actinium (Ac) fractions were sequentially eluted from the columns
with saturated HCI, and were evaporated to dryness.

Th fractions (from An fractions derived from Spike Mix P and Unspiked samples)
were dissolved in 8M HNO a. To each sample was added a drop of saturated
NH 4N0 2; the samples were heated briefly and then loaded onto 4-mm-diameter
Dowex-lx8 anion-exchange columns. After column washes with 8M HNO a, Th
fractions were eluted with 9M HCI. The Th fractions were evaporated to dryness,
dissolved in 8M HNO a, and loaded onto 2.5-mm-diameter Dowex-lx8 columns,
which were then washed with 8M HNO a. Th fractions were eluted with 9M HCI
and evaporated to dryness. The previous ion-exchange step was repeated, and the
final Th fractions were evaporated to dryness.

Ra fractions from samples 94-7-38 and 94-7-45 contained 226Ra, as determined by

alpha spectroscopy (see below). Unfortunately, the chemical yield as determined
from short-lived 224Ra in the Spike Mix P samples was not valid; early
precipitation steps, that may have had poor efficiencies for Ra resulted in
intermediate samples containing high yields of the 228Th precursor nuclide,
which would have caused 224Ra to grow back into the samples. For the two
samples with indications of Ra, special analytical samples were constructed from
15 ml of working solution and three times the standard P Mix aliquot. After a
single oxidation/reduction cycle and several hexone extractions (described above),
the samples were evaporated to dryness, dissolved in. 9M HCI, and loaded on 8-
mm-diameter Dowex-lx8 anion exchange columns, which were then washed
with more 9M HCI. The load and wash solutions were collected as Ra fractions,
evaporated to dryness several times with the addition of HNO a, and were then
dissolved in 8M HNO a. These solutions were loaded onto 4-mm-diameter Dowex-
lx8 columns, which were then washed with more 8M HNO a (this is the ThlRa
separation time). The load and wash solutions were collected as the Ra fractions,
and evaporated to dryness several times with the addition of 6M HCI. The Ra
fractions were then treated to the same cation-exchange procedure outlined above
for the An fractions, including the AG MP-50 step. Final Ra fractions were
evaporated to dryness.
Final Ra, Ac, Th, Pa, Np, and Am samples were dissolved in minimum volumes
of 6M HCI and transferred to tungsten filaments where they were re-evaporated.
Counting samples were prepared by volatilizing the samples onto I-inch diameter
platinum disks through the rapid heating of the filaments. U fractions were
subdivided into several samples of different strengths; some of these were
volatilized onto platinum, while others were subjected to isotopic analysis by
inductively-coupled-plasma mass spectrometry (ICP-MS). Pu fractions from
Spike Mix P and Spike Mix N samples were volatilized onto platinum, as were all
Unspiked fractions except those from samples 94-7-38 and 94-7-45, which were
analyzed for isotopic composition by thermal-ionization mass spectrometry
(TIMS).

- 10 -
Np and Pa samples were mounted in standard-geometry gamma-spec holders
and were counted close to the faces of large-volume germanium photon detectors;
gamma-ray intensities were used to measure the activities of the 239Np and 233Pa
chemical-yield indicators. Gross alpha-decay rates of these samples were
measured with calibrated 2-pi gas-proportional counters in order to complete the
link to chemical yields. All Pt-backed samples were pulse-height analyzed for
alpha particle emission with surface-barrier or Frisch-grid gas-filled detectors.

In total, more than 130 samples were generated and counted in the course of these
radio analytical analyses.

RADIOANALYTICAL RESULTS

Results of the analyses of the six working solutions are given in Table 5, scaled to
the masses of the original samples. These data are derived from a combination of
alpha-particle, gamma-ray, and mass spectrometry analyses. Data are given in
atoms, except for 239pu and 240pu for those samples where Pu fractions were not
subjected to mass-spectrometric analysis. Since 239pu and 240pu are not
resolvable with alpha spectroscopy (due to their similar alpha-decay energies), the
sum of their activities (in disintegrations/minute) is given in the Table.
Assessments of 239pu content are derived .from 240puJ239pu mass ratios estimated
from 238PuI(239pu+ 240pu) activity ratios (see below). Values listed for 233U are
calculated from the 234U content and the relative amounts of their decay
daughters (229Th and 230Th, respectively); as will be seen below, this may not be
an unambiguous calculation.

The distribution of uranium isotopes in each sample is given in Table 6 in the

form of atom ratios to 238U content. These are substantially the same data given
in Table 5, but do not include the propagated error on chemical yield.

Observables in the plutonium fractions are given in Table 7. The

240
238PuI(239pu+ pu) activity ratios were obtained from alpha spectroscopy, while
the atom ratios are the results of thermal-ionization mass spectrometry. The data
for the higher-mass Pu isotopes in sample 94-7-38 were compromised by a
contamination in the instrument; even the 240pu/239 pu ratio for that sample
should be treated with suspicion.

- 11 -
Table 5. Results of the Analysis of the Heavy-Element Contents of Selected
PUBLIC Samples. Data are total atoms in each sample, decay corrected to day
260,1994. Also given are disintegrations per minute of the sum of 239pu and 240pu
in each sample. (Continued on next page.)

Nuclide 94-7-14 94-7-17 94-7-19

243Am
241Am (5.22 ± 1A3)E9 (1.60 ± 0.20)E9 (9.14 + 3.28)E8
242Pu 1.6E10
241Pu
240fu
239Pu -9E10 -2E11 -lE11
238Pu (4043 ± 0047)E7 (1.57 ± 0.21)E7 (4.58 + 0.77)E7
236Pu 1.4E5
237Np (2.96 ± 0.29)E13 (8046 ± 0.50)E13 (3.57 ± 1.93)E13
238U (3.23 ± 0.13)E18 (1.63 ± 0.05)E19 (4.23 + 0.07)E19
236U (2.24 ± 0.14)E14 (9.54 ± 0.67)E14 (1.75 ± 0.08)E15
235U (5.61 + 0.23)E16 (1.80 + 0.06)E17 (1.66 + 0.03)E18
234U . (3.36 ± 0.14)E14 (1.03 + 0.03)E15 (1.71 + 0.03)E16
233U < 1.3E11 1.5E11
232U < 1.5E7 (2.07 ± 1.79)E7 (5.86 + OAO)E8
231Pa <6.3E10 1.3E11 (5.27 + 0.31)E11
232tfu (3.59 ± 2.17)E15 (3.77 ± 2.09)E15 (2.08 + 1.29)E15
23DTh (1.54 ± 0.09)E12 (7.14 ± 0.38)E12 (5.33 + 0.28)E12
229Th <9.0E8 1.0E9
227Ac (4.05 ± 2.08)E7
226Ra

activity
239pu+24°Pu (7.10 + 0.20) (13.65 + 0.30) (8040 + 0.26)

- 12 -
Table 5. (continued)

Nuclide 94-7-38 94-7-45 Blank

243Am 1.6E9 <5.8E9
24 lAm (3.89 ± 0.06)E10 (1.23 ± 0.02)E11 <7.4E8
242Pu 1.5E11 (1.06 ± 0.11)E10
24lPu (3.79 + 0.15)E10
24°Pu (5.59 + 0.19)E11 (1.81 + 0.02)E12
239Pu (8.12 ± 0.10)E12 (2.79 + 0.03)E13
238Pu (9.70 ± 0.28)E8 (2.36 + 0.06)E9
236Pu
237Np (7.81 ± 0.08)E15 (2.56 ± 0.02)E16
238U (9.70 ± 0.33)E19 (5.25 ± 0.15)E20 <8.0E15
236U (3.85 ± 0.17)E15 (2.08 + 0.09)E16
235U (3.22 + 0.11)E17 (2.03 + 0.06)E18
234U (1.90 + 0.07)E15 (1.24 + 0.04)E16 <4.4E10
233U < 1.5E10 (1.77 + 0.73)E11
232U <9.6E7
23lpa (2.82 + 0.26)E11 (8.53 + 0.58)E11
23% (1.49 + 0.13)E17 (3.28 ± 0.26)E17 <3.0E15
230rrh (1.29 + 0.07)E14 (5.15 + 0.27)E14
229Th < 1.3E9 (1.16 + 0.44)E10
227Ac (1.24 ± 0.46)E8
226Ra (1.26 + 0.15)E10 (3.23 + 1.13)E10

activity:
239pu+24Opu (556.3 + 5.5) (1886 + 16)

- 13 -
Table 6. The Uranium Isotopics of PUBLIC Samples, Atom Ratios. Natural
uranium values are included for comparison.

Ratio 94-7-14 94-7-17 94-7-19

232Uj238U <5E-12 (1.3 ± 1.1 )E-12 (1.38±0.10)E-11
233Uj238U <4E-8 <9E-9 <6E-8
234Uj238U (1.040± 0.011)E-4 (6.30 ± 0.06 )E-S (4.034± 0.020)E-4
235Uj238U (1.738± 0.013)E-2 (1.108± 0.009)E-2 (3.927± 0.016)E-2
236Uj238U (6.92 + 0.29 )E-S (S.86 + 0.37 )E-S (4.14 ± 0.04 )E-S

Table 6. (continued)

Ratio 94-7-38 94-7-4S natural

232Uj238U O.
233Uj238U (3.4 ± 1.4 )E-10 O.
234Uj238U (1.960+ 0.013)E-S (2.3S8± 0.017)E-S S.47E-S
235Uj238U (3.320+ 0.021)E-3 (3.873+ 0.027)E-3 7.2SE-3
236Uj238U (3.97 ± 0.12 )E-S (3.96 ± O.lS )E-S O.

Table 7. Observables in PUBLIC Samples, Plutonium Fractions.

Activity Atom ratios

Sample 238PuI(239pu+24Opu) 240puJ239pu 241puJ239pu 242puJ239Pu

94-7-14 0.0936±0.0089
94-7-17 0.0173±0.0023
94-7-19 0.0816±0.0131
94-7-38 0.0262±0.0007 0.069 +0.002
94-7-4S 0.0188±0.0004 0.0648±0.0001 0.00136±O.000S 0.OO038±O.OOO04

Results for the radio analytic Blank are consistent with a negligible radionuclide
content, giving no evidence for significant cross-contamination among the
samples. The data in the last column of Table S also give a feeling for the
sensitivity of technique, of potential use in the planning for future site inspections.

- 14 -
DISCUSSION AND CONCLUSIONS
The basic assumption underlying all the data interpretation that follows is that,
unless otherwise indicated, the radionuclide content of a given sample is
characteristic of a specific location in the enrichment cascade, and that there was
a single release time associated with that sample. Contributions from multiple
releases of material at different cascade process times would compromise many
of the following conclusions.

Information about operations at the Paducah and Portsmouth plants is available

in the unclassified literature (e.g. "Recent Developments in Uranium
Enrichment", J. R. Merriman and M. Benedict, eds., AIChE Symposium Series
221, New York, 1982; and "Uranium Enrichment", S. Villani, ed., Springer-
Verlag, Berlin, 1979). This is the starting point of the ensuing forensic analysis:
The main function at both plants is the enrichment of uranium by means of
gaseous diffusion of warm UF6; however, other enrichment techniques have been
explored (e.g. gas-centrifuge), and might have contributed to the isotopic content
of specific samples, but not significantly to the general level of contamination at
these sites. Operations at both plants started in the mid-1950's and continue to the
present. During much of this time, the Paducah plant has produced uranium
with a low enrichment of 235U (1-5%), some of which has been used as feed stock
at the Portsmouth plant, which makes several different products at 235U
enrichments of up to 97%.
The last statement is confirmed by the data in Table 6: It is much more likely that
feed or tails materials would be represented in the site-collection samples than
would be product materials, both because of their relative masses and their
relative values. The Paducah samples (94-7-38 and 94-7-45) are distinctly depleted
in 235U relative to natural uranium (235U/238Unat = 7.25x10-3), but those from
Portsmouth are all distinctly enriched. The fact that each of the five samples
represents a different 235U enrichment is confirmation that these are plants that
are actively changing the isotopics of uranium and not just utilizing materials
modified at other facilities.
Most of the diffusion barrier in both plants was replaced between 1975 and 1983 in
a "Cascade Improvement Program (CIP)". Because of the enormous volume of
contaminated material handled during the ClP, it seems likely that most of the
site contamination, particularly in sheltered or out-of-the-way places, dates from
that time.
An examination of the data in Tables 5 and 6 shows that there is a significant
amount of 236U present in the samples. The mass-236 isotope is not a component
in natural uranium, arising only in the interaction of neutrons with 235U. This
proves that at least some of the cascade feed material had been irradiated in a
reactor, and that the plutonium and neptunium in the samples probably came
into the plants as contaminants in the uranium, rather than as unrelated
entities. Furthermore, the mass-240 content of the plutonium in the Paducah

- 15 -
samples (Table 7) is not what one would expect of material that had been used in
high-burn-up power applications, but is characteristic of the enrichments found
in standard weapons-grade materials. Therefore, even though we did not find
any direct evidence of highly-enriched uranium, the sample analyses clearly
indicate that the plants have some tie to the weapons complex.

The ORIGEN2 code (A. G. Croff, Nucl.Tech. 62, 335 (1983); ORNL RISC Computer
Code Collection, CCC-371, rev. 1991) was used to understand the isotopic
compositions of the PUBLIC samples. ORIGEN2 calculates the nuclide content of
nuclear fuels at specified burn-up intervals for several types of reactor. The
calculated correlations between the mass ratio 240puJ239pu and the activity ratio
238PuI(239pu+ 240 pu) for three standard reactor types and an external target
(Thermal) are given in figure 1. Reactors that are unlikely to have contributed to
weapons production in the 1960's (e.g. mixed-fuel, or liquid-metal-cooled) have not
been considered. A reactor power level of 37.5 MW thermal was assumed for the
calculations. The measured data from the Paducah samples fall in between the
curves given for CANDU and Boiling Water Reactors. 238pu is a fairly short-lived
nuclide (t l/2 = 88 years); since the samples are probably at least 15 years old
(midpoint of the CIP), allowance for radioactive decay occurring since the
material was discharged from a reactor would lead to a corrected value for
238PuI(239pu+ 240 pu) significantly closer to the BWR curve. Moreover, the mass
diffusion principle works on the Pu isotopes in PuF 6 as readily as it does on the U
isotopes in UF 6. The Paducah samples consisted of plant tails materials, depleted
in low-mass isotopes; consequently, a correction for the action of the diffusion
process would also lead to Pu data closer to the BWR curve (see Figure 2).

Feed material to the Paducah plant was probably not significantly enriched (or
depleted) in 235U. If the Pu component of the plant feed had been produced in a
reactor with enriched fuel, a much higher 240puJ239pu value would have resulted
from an irradiation that would have burned the 235U content to near-natural
levels. This implies that the production of weapons-grade Pu in the U.S. took
place in reactors using uranium fuel of near-natural composition. However, the
observables in the Paducah Pu samples do not match the ORIGEN2 calculations
for either of the natural-uranium-fueled standard reactor types, particularly after
making the above-mentioned corrections. It would be naive to conclude that Pu
production occurred in BW or PW Reactors with lower fuel enrichments though;
an ORIGEN2 calculation for a hypothetical (unrealistic) natural-fuel BWR
resulted in the curve given in figure 2, compared with the same calculation
depicted in figure 1 for a 3.2%-enrichment 235U BWR. Some of the difference
between the calculations results from the fact that much of the production of 238pu
is from capture of a neutron by 237Np; at medium-to-Iong irradiation times, 237Np
arises largely from sequential capture of two neutrons by 235U followed by beta
decay of 6.7-day 237U. An enrichment in the 235U content of the starting material
thereby results in an increase in production of 238pu.

- 16 -
Also shown in Figure 2 are measured Paducah plutonium observables (given as
circular data points). The result of correcting the Pu data for a 20-year decay
period and an enrichment (or depletion) of 33% of that defined by the 235U-content
of the uranium component of the samples (since the mass difference among the
Pu isotopes is only ±1 unit, while it is -3 in the uranium) is denoted by the
triangular points. While there is seeming agreement between the original data
and the "natural-uranium BWR" curve, modified Pu isotopics very closely match
those expected from an enriched-fuel BWR after realistic corrections are made for
decay and the effects of mass separation.
There is thus an apparent conflict in that a natural-fuel reactor is required by the
comparison of the analytical data with the literature, but the only reactor types in
the ORIGEN2 data library that fit the corrected measurements require enriched
fuel. However, another path to 238pu is through the interaction of high-energy
neutrons with 239pu via the 239Pu(n,2n) reaction. Production of the Pu isotopic
mix in the Paducah samples from natural uranium requires a harder neutron
spectrum than that generated by the reactors in the calculations, implying a less
efficient (higher Z) moderator than either light or heavy water. This means that
U.S. weapons-grade Pu production takes place in graphite-moderated reactors.
Unfortunately, the ORIGEN2 libraries available at LLNL do not contain a
graphite-moderated "Production Reactor" data set. Nevertheless, since the
production of most of the uranium and plutonium isotopes are through neutron-
capture reactions that are relatively spectrum insensitive, it seems likely that Pu
and U production in the hypothetical "natural-uranium BWR" would closely
approximate that resulting in the cascade feed material.
The isotopic composition of plutonium as a function of burn-up in a "natural-
uranium BWR" is given in figure 3, along with the production of 239pu relative to
the unburned 238U remaining in the fuel. The value of 239pu used in the
calculation of these ratios is the sum of the number of atoms of 239U, 239Np, and
239pu given in the code output at the appropriate irradiation length. Again, a
power level of 37.5 MW thermal is assumed in the calculation. The isotopic ratios
obtained from sample 94-7-45 (Table 7) are plotted as circular points at an
irradiation time of 33 days (1250 MW-days), which is defined by the 240puJ239pu
value. The value of 242puJ239pu for the sample falls very close to the appropriate
curve. 241pu is a short-lived nuclide (t l/2 = 14.4 years); a decay correction of 41
years is required to adjust the observed data to the 2411239 curve. This decay time
(i.e. average time of reactor discharge in 1953) is not inconsistent with the time
since last chemical separation calculated from the 241Am content of the sample
(see below).
Correction of the Pu isotope ratios for the action of the cascade, assuming a linear
enrichment with mass number as estimated from the 235U,t238U value for sample
94-7-45 (Table 6) results in the data plotted as triangular points at an irradiation
time of 28.5 days (1070 MW-days) in figure 3. The fit of the corrected value of
242puJ239pu to the calculated curve becomes better. The decay correction required

- 17 -
to make the adjusted value of 241puf239pu fit the calculated curve increases
slightly, to 42 years. The assumption of a linear relationship between enrichment
and isotope mass is probably reasonable for small enrichments, such as that
resulting in sample 94-7-45; for larger enrichments it is no longer appropriate
(see below).

Comparison of the plutonium data from the Paducah sample with the curves
given in figure 3 indicates that the accuracy of the calculation for the production
of plutonium isotopes in a pre-cascade reactor irradiation is good. Therefore, it
can be assumed that the change of uranium isotope ratios with irradiation time is
also well reproduced by the code. Uranium isotope ratios from the same
ORIGEN2 calculation are given in Figure 4, along with the production of 237N p
(including 6.7-day 237U) relative to residual 238U.

From Figure 3, the discharged reactor fuel that contributed to the Paducah feed
stock could be expected to have a 239puf238U atom ratio of approximately 10-3.
From Table 5, the observed 239puf238U is on the order of 10-7 in the Paducah
samples and on the order of 10-8 in the Portsmouth samples. If the assumption is
made that the source of the Portsmouth Pu is the Paducah product, and that the
mass fractionation of 239pu from 238U is negligible, then loss of Pu in the Paducah
cascade (or in transportation containers) from chemical fractionation is on the
order of 90%. This is not surprising, since volatile PuF 6 is much more reactive
than UF6' and is reduced to PuF4 by residual air (both 02 and N 2) and by small
amounts of UF4 contaminating the UF6. If the further assumption is made that
the irradiated uranium made up a large fraction of the Paducah "feed stock, then
the observed 239puf238U values are consistent with 10-5 to 10-4 of the weapons-
grade Pu product being lost to the uranium fraction in the pre-cascade fuel
reprocessing.

From Figure 4, the 237Np/238U value in the discharged reactor fuel was
approximately 10-5 . From Table 5, the observed 237Npf2 38U is on the order of 5x10-5
in the Paducah samples and on the order of 5x10- 6 in the Portsmouth samples. If
the same assumptions apply to 237Np as apply to 239pu, then loss of neptunium in
the Paducah cascade is also on the order of 90%; volatile NpF6 is intermediate in
chemical reactivity between UF 6 and PuF 6' and a significant cascade loss would
be expected. The fuel reprocessing was notably unsuccessful in removing 237N p
from the Paducah feed stock, leaving essentially all of it with the uranium. This
last observation reveals a detail about the post-processing conversion of product to
volatile hexafluorides: While UF 6 can be produced in high yield by a variety of
strong fluorinating reagents (e.g. BrF3' ClF3' SF4' etc.) from several different
uranium compounds at temperatures near 2200C, the formation of NpF 6 in high
yields requires the reaction of F 2 with lower Np fluorides at higher temperatures.
This means that the UF 6 production probably went through a UF4 intermediate.

- 18 -
Although the relative concentrations of Np and Pu do not provide a conclusive
signature, the reprocessing of spent fuel by means of a solvent extraction with
tributyl phosphate (the PUREX process) provides for an efficient recovery of
plutonium and a good decontamination from fission products, but leaves
neptunium with the uranium. It would be interesting to look for long-lived 1291 or
99Tc in the samples, both of which have fluorides that are volatile under the same
conditions as UF6' and whose contamination levels might provide further clues to
the reprocessing technique.

The 237Np and the residual Pu comprised a very small mass fraction, and only
about 1% each of the total alpha radioactivity, of the Paducah feed material.
However, the 237Np content approximately doubled the gamma-ray dose from the
material, requiring increased shielding, and the presence of Np and Pu in the
final product would require additional chemical clean-up for most non-reactor
applications. Rather than permanently contaminating the cascade and fouling
the plant product, it would have been preferable to clean the material thoroughly
before producing the UF6 feed. A possible explanation for use of this less-than-
ideal strategy: Once a UF 6-diffusion plant is in operation it cannot be stopped and
without enormous effort and" expense and a large lead time. Feed of UF6
must be continuous. It seems likely that the use of contaminated uranium in the
cascade was a symptom of a chronic uranium shortage, where stockpiles were
insufficient to support independent programs for the production of weapons-grade
Pu and highly-enriched uranium, and the input requirements of the enrichment "
cascades did not permit enough time for a lengthy decontamination of the feed
stock.
The Pu activity ratios given in Table 7 can be corrected for enrichment in the
cascade by assuming, once again, that the action of the cascade on the plutonium
isotopes is similar in magnitude to the enrichment of uranium isotopes, and that
the effect is roughly linear with mass. After each activity ratio was corrected for
an arbitrary decay period of 20 years, the 3.2%-enrichment 235U BWR curve in
Figure 2 was used to estimate the corresponding 240puJ239pu mass ratio. Both the
238pu and 240pu activities were corrected for the effect of the cascade on their
enrichment (defined by +33% of the 235U enrichment of the sample), the activity
ratio was recalculated, and Figure 2 was again used to estimate
This process was iterated until the value of 240puJ239pu converged.

The results of the calculation for each sample are given in Table 8. The atom ratio
240puJ239pu in the feed stock can be used with Figure 3 to determine the length of
the reactor irradiation resulting in the Pu isotopic mix, in units appropriate for
use with Figure 4. Both 240puJ239pu and 241puJ239pu (from Figure 3), corrected for
the effect of the cascade on the samples, are given in Table 8; however, the 241pu
data contains no correction for decay since reactor discharge. Considering the
approximations inherent in computing the data in Table 8, the resultant values

- 19 -

, "
for reactor irradiation length are remarkably consistent, yielding an average
exposure of(1100 ± 300) MW-days (thermal) for the five samples.

Table 8. Plutonium Isotope Production and Reactor Irradiation Length (assuming

37.5 MW thermal power). The 240pu/239Pufeed data is taken from the 3.2%-
enriched 235U BWR curve in Figure 2; the 238pu/(239pu+ 240 pu) and
240puJ239Pucascade data are corrected for mass fractionation in the cascade and a
20-year decay period; the 241pul239 pu data are taken from Figure 3, and are
corrected for the action of the cascade, but not for decay since reactor discharge.

238PuI(239PU+24OpU) atoms 240puJ239pu 241puf239pu Irradiation

Sam}2le activitys corr. feed cascade dischg s casco length
94-7-14 0.069 0.083 0.057 0.0083 43 days
94-7-17 0.017 0.036 0.031 0.0024 20 days
94-7-19 0.035 0.056 0.023 0.0018 28 days
94-7-38 0.039 0.060 0.073 0.0140 32 days
94-7-45 0.027 0.048 0.057 0.0081 25 daY§
29.6±8.7
The irradiation lengths given in Table 8 can be used with Figure 4 to estimate the
starting composition of the reactor-irradiated fraction of the cascade feed
resulting in the analytical samples. It should be noted that at these low burn-ups,
the atom ratios of 234U/238U and 235U/238U have not changed much from their
values in natural uranium. The apparent enrichment of each uranium isotope
relative to the cascade feed stock can be calculated from the data in Table 6:
Enrichment = (XUj238U)observei(XUj238U)figure 4' where X is the atomic number of
the isotope of interest. The enrichments of 234U, 235U, and 238U, relative to
238U feed for each of the analytical samples are plotted in Figure 5. The lines
connecting the points are drawn free-hand. It is observed that the enrichment as
a function of mass number is not even approximately linear for the Portsmouth
samples, casting considerable doubt on the calculations resulting in their entries
in Table 8. For the Paducah samples (used in most of the prior analyses) the
linear-enrichment assumption is reasonable.

For each of the five samples, the calculated enrichment of 236U is considerably
less than that expected from the free-hand curves in Figure 5. For samples 94-7-
17,94-7-38, and 94-7-45, the 236U-enrichment is approximately 30% of that expected
from the curves. For sample 94-7-14 the 236U-enrichment is 10% of expected, and
for sample 94-7-19 it is only 4% of expected. If it is assumed that the ORIGEN2
calculation of 236U production is accurate, then about 30% of the Paducah feed
stock was irradiated uranium and 70% was natural uranium. Sample 94-7-17
from Portsmouth exhibits the same isotopic fingerprint, thereby connecting it to
the Paducah cascade. The other two Portsmouth samples originated from
starting materials with a significantly smaller dilution of reactor fuel in natural

- 20 -
uranium. This implies that Portsmouth received its feed from more than one
source, and that the cascade was subdivided into different sections which received
variable proportions of the different materials. An alternative explanation would
be that the fraction of reactor-irradiated materials in the feed stock has changed
over the years and samples 94-7-14 and 94-7-19 represent materials from a
different operational era at the plant than do the other samples. The 235U
enrichments of 94-7-14 and 94-7-19 are higher than that in the other Portsmouth
sample; given that feed or tails materials are more likely to be released into the
environment than is product, these two samples may be representative of tails
from a separate cascade, that was used in the production of highly-enriched
uranium.
Samples can be age-dated using pairs of radionuclides that are linked by
radioactive decay. It must be assumed that there existed a time in the past when
the sample contained only parent activities, meaning that a chemical separation
was performed to remove daughters. A gaseous-diffusion enrichment plant
would appear to be an ideal facility for age determinations, since a likely
mechanism for radioactive release to the environment would be the escape of
volatile UF6 (and PuF6) into the air. Such a process would leave refractory Pa, Th,
Ac, Ra, .and Am daughters and grand-daughters behind.
Table 5 gives concentrations for several'pairs of nuclides that can be used to
calculate specimen ages, which are given in Table 9. The ages thus calculated
from the measurements of relative quantities of volatile and refractory nuclides
result in values of the age that are inconsistent and absurdly old. This implies
that a chemical separation has indeed occurred, but that the refractory daughter
nuclides have been dispersed after most of the volatile parent activities were
removed. For example, consider

- 21 -
Table 9. Pseudo-ages of the Analytical Samples, in years. The data are calculated
from the indicated nuclide pairs, assuming that at the time defining the age,
there was a complete absence of the radiogenic daughter. No corrections have
been made for the radionuclide content of soil.

daughtgr/:gargnt 94-7-14 94-7-11 94-7-19 94-1-38 94-1-45

230rrh I 234U 1640±160 2490±210 110+8 28600±3100 16200±1500
226Ra/234U 754±62 457±93
226Ra I 23Orrh* 11±2 7±3
231Pa I 235U 320±25 900±110 430.±40
227Ac/ 235U 1200±600 130.±40
227Ac I 231pa* 8±4

232Th/ 236U 1x108 6x107 3x107 2x108 1x108

241Am /241pu# 30.5+0.9

* Assumes that precursor (234U or 235U) is completely absent.

# Calculated from data in Table 5, no corrections.

the case of a large amount of UF 6 stored in a container. There is a constant

production of Th and Pa fluoride species. through radioactive decay of the U
isotopes; these daughter compounds are non-volatile and tend to deposit on the
walls of the container. If the UF 6 is pumped away (leaving a small residue), a
sample of the mixture of nuclides on the walls of the container would appear too
old: It would normally be assumed that the daughters arose from the long-term
decay of a small mass of parent rather than from the short-term decay of a large
(missing) mass of parent. Another way this mixture could obtain is through the
enrichment process itself; decay of uranium occurs throughout the enrichment
cascade, and ThF4 and PaF5 would tend to accumulate in the diffusion barrier.

If the assumption is made that the 230Th activity in each sample was produced
from a steady-state ingrowth of ten years (the proper order of magnitude), then
3.5x104 atoms of 234U were required to account for each atom of 230Th observed. If,
at the end of the accumulation period, all volatile UF 6 were removed from the
sample, the data in Table 5 can be used to estimate the loss in the process caused
by chemical reactions producing non-volatile uranium compounds (e.g. U0 2F 2)'
In the Paducah samples, approximately 0.05% of the uranium remained with its
Th daughter. In Portsmouth samples 94-7-14 and 94-7-17, 0.5% of the uranium
remained. Sample 94-7-19 retained 10% of its original uranium content; it seems
unlikely that the diffusion barrier would clog to this extent before it was changed
out, or that this kind of accumulation on the walls of a container would not be
recovered. Sample 94-7-19 is clearly more volatile in nature than are the other

- 22 -
four samples, and may have had contributions through both a mechanical
dispersal of daughters and a volatile gas release (see below).

The ages determined from the relative amounts of non-volatile daughter and
grand-daughter activities (230Th and 226Ra, or 231Pa and 227 Ac) are lower than
expected, possibly as the result of the presence of a background due to the
incorporation of dirt within the samples. The activity ratio of daughter/grand-
daughter is approximately 1.0 in soil, but it is much higher for samples in which
the daughter was recently produced. The 232Th content of soil can be used to
explain the inordinately long ages derived from 236U, and provides a means of
estimating the soil contribution to the radionuclide content of each sample. The
Earth's crust contains an average of 12 ppm 232Th. If the soil contributions to the
analytical samples contain thorium in the same proportion, Table 5 data can be
used to calculate the soil components given in Table 10. (Also given are the
original sample weights from Table 1). Soil contributed only a small fraction to
the masses of the Portsmouth samples, all of which were collected magnetically.
Paducah sample 94-7-38 was collected with a spatula, and only about a third of its
mass could be accounted for by XRF analysis (Table 2); it therefore was indeed
composed mostly of low-Z dirt. Paducah sample 94-7-45 was part of a larger
sample from which nickel flakes were segregated during prescreening. Even
though it was collected magnetically, it is not unreasonable to believe that half of
the residue was dirt entrained during collection.

The 238U content of soil is usually about 25% that of 232Th. On that basis, an
estimate of the soil contribution to the uranium in each sample is given in Table
10. Comparing these values with data in Table 5, it is concluded that soil does not
significantly perturb either the uranium content or the relative uranium isotopics
of any of the samples. If it is assumed that the radiogenic daughters of the soil
nuclides are in equilibrium with their parents, then corrections to Table 5 data for
231Pa, 230Th, 227Ac, and 226Ra can be calculated. These are also provided in Table
10. Any corrections to the 230Th and 231Pa data are small and would not be
significant given the precision of the measurements. The soil contribution to the
226Ra content of the Paducah samples is quite high, however, on the order of
100%; therefore the sample ages computed from the 226RaJ2 30Th pair (ca. 10 years)
are much too low, in agreement with expectations. The corrections to 227Ac are
not large enough to explain the ages obtained from the 227Acf231Pa pair. There is
clearly a problem with the 231Pa data.

If it is valid to assume that the 'mix of 231Pa and 230Th in each sample arose from
the decay of uranium of the same composition as that in the samples, then the
same (incorrect) ages should be

- 23 -
Table 10. Soil Contributions to the Radionuclide Contents of PUBLIC Samples,
calculated from the 232Th contents given in Table 5. Uranium content is assumed
to be 25% of the thorium content.
94-7-14 94-7-11 94-7-19 94-1-38 94-7-45
sample wt 7.0g l.4g 3.3g 4.9g 20.5g
soil content 0.12g 0.12g 0.07g 4.8g 10.5g
soil U atoms 9.1E14 9. 1E14 5.1E14 3.6E16 8.0E16
231Pa soil frac >0.5% >0.2% 0.03% 4.3% 3.2%
230trh soil frac 1.0% 0.2% 0.2% 0.5% 0.3%
227Ac soil frac >0.4% >23% 15%
226Ra soil frac 100% 85%

calculated from both the 230ThJ2 34U and 231paJ235U nuclide pairs. This is clearly
not the case. With the exception of sample 94-7-19, the 231Pa ages are significantly
shorter than the 230Th ages, indicating that 231Pa was removed from the samples
after release. Another way to look at this: The atom ratio 235U/234U in the
material whose decay resulted in the non-soil radionuclides in the samples can be
calculated from the uranium half lives and the 231pa/230Th atom ratio. These
data are given in Table 11, along with the direct measurement of 235U/234U
derived from Table 6. Again, with the exception of sample 94-7-19, there is less
231 Pa in the samples than there should be. One might consider a possible
explanation to be that the uranium composition of the samples is unrelated to the
daughter content; however, this could not explain either of the Paducah samples,
where the derived values of 235Uj234U are an order of magnitude lower than would
be produced by gaseous diffusion of the known starting materials.

An explanation for the loss of 231Pa is in its aqueous chemistry. PaF5 is soluble in
water (in (act, deliquescent), while AcF3 and ThF4 are not. The 94-7-14 and 94-7-17
samples from Portsmouth may have been moist, resulting in partial leaching of
231Pa. The 231Pa content of the Paducah samples is so severely reduced that it
seems likely that water was used in the process that resulted in the dispersal of
the material; both samples were collected well above floor level, indicating the
possible use of pressurized water. Since uranyl fluoride (U0 2F 2' the product of
the reaction ofUF6 with moist air) is also water soluble, it is likely that the process
uranium loss calculated above for Paducah samples is actually an order of
magnitude too low, and that Portsmouth and Paducah both suffered from process
losses on the order of 1%.

The mix of activities in sample 94-7-19 is difficult to explain. The relative

proportions of 231Pa and 230Th in the material are inconsistent with the uranium
isotopics, even with a correction for different solubilities. Much of the
radionuclide content of

- 24 -
 Table 11. Evidence of Chemical Activity in PUBLIC Samples After Release.
Calculated atom ratio 235Uj234U from Table 6 ("direct") and from relative amounts
of 231Pa and 230Th given·in Table 5 ("daughter"). Natural uranium 235U/234U =
132.

94-7-14 94-7-17 94-7-19 94-7-38 94-7-45

direct 167 ±2 176±2 97±1 169±2 164±2
daughter <56 290±20 6.3+0.7 4.8± OA

the sample may have been leached from other materials by water, which would
explain the elevated levels of uranium and 231Pa relative to other activities.

The only age in Table 9 that is consistent with the samples being generated during
the CIP is the 241Am/241pu value given for sample 94-7-45. If the 241Am
component of the sample was the result of its long-term accumulation in barrier
material as (partially) volatile PuF 6 flowed by, the computed age would be
significantly higher than the actual age. In fact, it is impossible to calculate an
age for probable-barrier samples 94-7-14 and 94-7-17 from the calculated 241pu
data in Table 8 because there is more measured 241Am in the samples than can be
attributed to full decay. However, Table 2 indicates that potential barrier material
(nickel) made up less than 10% of sample 94-7-45. It seems likely that tp.e
radionuclides in this specimen were introduced with debris components that
were not barrier-related. These might be fragments from the of a container
used for storage of tails material upon which a large fraction of the highly-
reactive PuF 6 had reduced to a non-volatile compound. This would result in an
age value that was only slightly too old. The midpoint in plant operations between
start up (-1955) and the CIP (-1979) is 1967, or an age of27 years.

SUMMARY

.. The radio nuclide contents of samples taken in the PUBLIC site inspections not
only reveal information about plant operations in Portsmouth and Paducah, but
also about the U.S. weapons complex as a whole:

Feed stock to the enrichment plants contained a significant amount of uranium

that had been irradiated in a natural-uranium-fueled graphite-moderated reactor
for the purpose of producing weapons-grade plutonium. Recovery of Pu from this
fuel was probably accomplished with the PUREX process, leaving about 10-3 % of
the plutonium and virtually all of the byproduct 237Np with the uranium. The use
of this incompletely decontaminated material in the cascade feed was probably
required by a uranium shortage occurring sometime after the plants started
operations. The UF6 cascade feed stock was produced through the action of free
fluorine on UF4 rather with other fluorinating agents.

- 25 -
The same isotopic signatures exist in the Paducah and Portsmouth feed stocks;
however, feed of a different signature was also used in Portsmouth, implying the
presence of more than one enrichment cascade. This separate cascade had tails
with a moderate enrichment in 235U, as would be expected from a process for the
production of highly-enriched (weapons-grade) uranium. Cascade losses of
uranium in both Portsmouth and Paducah were on the order of 1%; about 90% of
the plutonium and neptunium introduced with the Paducah feed stock were lost
in the cascade.

The site contamination at both plants took place by dispersion of non-volatile

activities rather than through leaks of volatile UF6 into the gas phase. In the
Portsmouth facility this took place mechanically, while in the Paducah facility
water was used. It proved to be difficult to age date any of the operations at the
plant; decay of 241pu relative to the other Pu isotopes and ingrowth of 241Am
indicate enrichment operations in Paducah in the 1950's and 1960's.

ACKNOWLEDGEMENTS

Thanks are due to P. Grant and R. Whipple of the Forensic Science Center for
collecting samples in the site inspections, and performing the sample
prescreening at LLNL. R. Nuckolls, A. Volpe, B. Ruiz, C. Velsko, and J. Wild
performed some of the analytical measurements on radiochemical fractions,
without which there would be little data to interpret. The hospitality and
helpfulness of the staffs of both the Portsmouth and Paducah facilities during the
site inspections is also gratefully acknowledged.

- 26 -
FIGURE CAPTIONS
1. Results of ORIGEN2 calculations for several reactor types, correlation
between plutonium observables: 238PuI(239pu+ 240pu) activity vs. 240pu/239 pu
mass.
2. Results of ORIGEN2 calculations for Boiling Water reactors, and the effect
of variable fuel enrichment on Pu observables. Circles are derived from data in
Table 7 for samples 94-7-38 and 94-7-45. After correction for the effects of 20 years
of radioactive decay (arbitrary) and mass fractionation in the cascade, adjusted
points are represented by triangles.

3. ORIGEN2 calculation of the relative production of the indicated plutonium

isotopes as a function of irradiation time at a power level of 37.5 MW (thermal) in
a hypothetical natural-uranium-fueled BWR. Also shown is the production of
239pu relative to residual 238U in the fuel. Circular data points are for sample 94-
7-45, from Table 7. Triangular points are these data corrected for mass
fractionation in the cascade. Arrows on 2411239 data indicate that corrections for
the decay of short-lived 241pu would adjust the data closer to the appropriate
curve. The data are plotted at irradiation times defined by the values of 240/239.
4. ORIGEN2 calculation of the relative production of the indicated uranium
isotopes (and 237Np) as a function of irradiation time at a power level of 37.5 MW
(thermal) in a hypothetical natural-uranium-fueled BWR. Computation
analogous to Figure 3.

5. Enrichment of uranium isotopes relative to cascade feed stock, defined as

(xUj238U)ob/(Xuj238U)feed' for the five PUBLIC samples. Curves are free-hand.

- 27 -
 10-1

Pressurized
Water
(3.20/0 235U)
»o,.....
-.
<
-,:
Boiling m
,.....
Water -.
(3.20/0 235U) ...o
N
eN
en
"'0
E.
eN
10-3 -C
CANDU s::::::
(natural U) +N
o-'='"
"'0
s::::::
Thermal "'-'
(natural U)

10-4

1.0 0.1
Mass Ratio , 240pUp39pU figure 1
- 28 -
 Boiling
Water
(3.2% 235U) 10-1

»
_.
....-.
<
10-2 '<
"""'I:
Boiling
Water
_.
....
m
o
(natural U) '" N
(,)
(X)

"'0
c:
(,)
(0

10-3
+N
-'="
o
"'0
c:
"-"
.. '

10-4

1 0.1
Mass ratio, 240puJ239pu figure 2
- 29 -
 1

10-1

10-2

o
3
:0
....
m
-.
o
10-3

10-4

figure 3
- 30 -
 10-2

10-3

10-4

»
r+
o
3
Q)
_.
r+
o
10-5

.'

10-6

_ ___I 10-7
1000 100 10
Irradiation length (days) figure 4
- 31 -
 10

94-7-19

- CD
,....
:l
'- 94-7-14
.....0
0
CO
CD
'-
.....0 94-7-17
CD
.-
.....CO> 1
- CD
'-
.....
C
CD
E
.c:
.-'-0 94-7-45
c
W
94-7-38

0.1 __ ____ ____ ______ ____ ____

234 235 236 237 238
Atomic mass figure 5
- 32 -