Content uploaded by Alex Meshik
Author content
All content in this area was uploaded by Alex Meshik on Feb 05, 2021
Content may be subject to copyright.
Record of Cycling Operation of the Natural Nuclear Reactor
in the Oklo/Okelobondo A rea in Gabon
A. P. Meshik, C. M. Hohenberg, and O.V. Pravdivtseva
Physics Depar tment, Washington University, St. Louis, Missouri 63130, USA
(Received 13 May 2004; published 27 October 200 4)
Using selective laser extract ion technique combined with sensitive ion-counting mass spect rometr y,
we have analyzed the isotopic str ucture of fission noble gases in U-free La-Ce-Sr-Ca aluminous hydroxy
phosphate associated with the 2 billion yr old Oklo natural nuclea r reactor. In addition to elevated
abundances of fission-produced Zr, Ce, and Sr, we discovered high (up to 0:03 cm3STP=g) concen-
trations of fission Xe and Kr, the largest ever observed in any nat ural material. The specific isotopic
structure of xenon in this minera l defines a cycling operation for the reactor with 30-min active pulses
separated by 2.5 h dormant periods. Thus, nature not only created conditions for self-susta ined nuclear
chain reactions, but also provided clues on how to retain nuclea r wastes, including fission Xe and Kr,
and prevent uncontrolled runaway chain reaction.
DOI: 10.1103/PhysRevLet t.93.182302 PACS numbers: 28.50.–k, 28.41.Kw, 28.41.My, 91.90.+p
A natural nuclear chain reaction was predicted by
Kuroda [1] 20 years before the remnants of the natural
reactor were actually discovered [2 –4]. So far, 16 indi-
vidual reactor zones have been found in the Oklo/
Okelobondo area in Gabon. Aside from being a fascinat-
ing natural phenomenon, the occurrence of self-
sustaining natural nuclear reactors has several impor tant
implications, ranging from the verification of variability
of the long-term fundamental physical constants [5,6] to
storage of nuclea r wastes in geological environments [7].
Many elements extracted from the reactor material still
car ry clear isotopic signatures of 235Uand 239Pu fission
and neutron capture reactions. Isotopic compositions of
these elements allowed for reconstruction of the effective
neutron fluence (up to 1021 n=cm2), the amount 235U
consumed (>5 tons), the energy released (15 GW yr).
Also, using fission products of 24 000 yr 239Pu, an esti-
mate was made of the effective duration of this nuclear
fission chain reaction (150 000 yr). The average power,
therefore, was only about 100 kW, equivalent to a small
research reactor. The fact that natural reactors did not
explode and dissipate t hemselves right after they went
critical was evidently due to some self-regulation mecha-
nism providing a negative feedback. It is not clear, how-
ever, whether the reactor was operated continuously or
in pulses. This would depend on the mechanism of self-
regulation, and/or the time constant for the negative
feedback which prevented a runaway chain reaction.
One proposed mechanism was related to the burning up
of highly neutron absorbing impurities, such as rare ea rth
isotopes or boron, both of which have been detected in
Oklo [8]. As the strong absorbers were burned up at one
edge of the active reactor zone and uranium was burned at
the other, the active zone perhaps shifted along the U
vein, like a flame over a wet log. Therefore, different
par ts of the natural reactor could have operated at differ-
ent times [8]. Another potential self-regulation mecha-
nism could have involved water, which acts as a neutron
moderator [9]. As the temperature of the reactor in-
creased, all unbounded water was converted into steam.
This would reduce the neutron thermalization and shut
down the chain reaction.The chain reaction could resume
only when the reactor cooled down and the water concen-
tration increased again. But until recently, there was no
strong evidence in favor of any of these mechanisms.
Amazingly, isotopically anomalous xenon we found in
Oklo Al phosphate carries the fingerprint of a specific
cycling operation with a time scale, which suggests that
the self-regulation must indeed involve water.
The material from the nat ural nuclea r reactor was
acquired from a Pixie drill with a double swindler in
drill hole S2 in the SD.37 gallery on the east face of
reactor zone 13. It consisted mainly of massive lustrous
uranium oxide grains with numerous 0.1–0.5 mm-sized
La-Ce-Sr-Ca aluminous hydroxyl phosphate inclusions
[10]. A polished slice (34mm,1mm thick) was
prepared from this sample and placed into vacuum ex-
traction cell where heavy noble gases were extracted using
a slightly defocused beam from acoustically Q-switched
Nd-YAG laser (this extraction technique described in de-
tails in [11,12]). A typical diameter of the extraction
crater was about 25 mm —far smaller than the investi-
gated m ineral grai ns — ensur ing m ineral specific analy-
sis. All stable Xe and Kr isotopes (except 78 Kr) from 28
individual extraction spots on U-bearing minerals and 13
spots on Al phosphates have been analyzed using a high
transmission ion-counting mass spectrometer [13]. In all
experiments, the amount of extracted gases was sufficient
for precise measurement of their isotopic compositions.
Va rious U oxides contained from 105to 103cm3
STP=gof 136Xe (Fig. 1), while U-free alumophosphates
had even more fission Xe, up to 0:03 cm3STP=g,the
highest Xe concentration ever found in natural material.
Evidently, fission Xe migrated from the U-bearing phase,
VOLUME 93, NUMBER 18 PHYSICAL REVIEW LETTERS week ending
29 OCTOBER 2004
182302-1 0031-9007=04=93(18)=182302(4)$22.50 2004 The A merican Physical Societ y 182302-1
where it has been produced, to the adjacent Al phosphate.
86Kr=136 Xe ratios tend to be lower in minerals, which
have less fission gases (Fig. 1), providing further evidence
for losses of fission products from U oxides. Less than 1%
of all fission Xe (as calculated from 235 Uburnout) is
retained in the U-rich phase, with a great fraction of
lost Xe apparently recaptured by Al phosphates. It was
found earlier that Oklo Al phosphate is enriched in fission
Zr, Ce, and Sr, while adjacent uraninite is depleted in
those elements [10]. T his unique abi lity of A l phosphate t o
capture fission products may be useful in man-made re-
actors and long-term nuclear waste storage.
The second interesting finding in Al phosphate was
130Xe excess (Fig. 2). This isotope is not produced by
fission since it is shielded by stable 130 Te in the fission
chain. 130Xe is a likely product of the reaction
129In; 130 I!130Xe, previously observed only in Oklo
uraninites from Zones 2and 3[14,15]. The enhancement
of 130Xe in Al phosphates is much higher than in U-rich
phases (Fig. 2), suggesting that 129 Ihas been displaced
from its parent uraninite into adjacent U-free Al phos-
phate during exposure to the thermal neutrons. The shift
in 130Xe=129 Xe ratio allows us to estimate the effective
neutron fluence of 1:71021 n=cm2— more than was
previously determined in the sa me reactor zone (1:0
1021 n=cm2and 0:78 1021 n=cm2) [16,17], suggesting a
possibility that Al phosphate may attract fission 129 I.
The most rema rkable Xe anomalies were observed in
the heavy isotopes (Fig. 3). Xenon in U phase is a rela-
tively normal mixture of fission products of 235 Uand
239Pu (by thermal neutrons) and 238U(by fast neutrons).
Spontaneous fission of 238U, dominant in common rocks,
is negligible in Oklo samples. However, Al phosphate
once again carries the most extreme anomalies, which
are impossible to explain in terms of the mixing of
known fissile nuclei and n-capture reactions. The appa rent
feature of Xe in Al phosphate seems to be a deficit of
136Xe, the end product of the shortest fission chain. The
only -active precursor of 136 Xe is 86 s 136I;so,afterthe
onset of the fission chain reaction, 136 Xe appears first and
hence has more chance of being lost before the other Xe
isotopes start to accumulate. This, in itself, suggests a
cycling operation of the natural reactor. As the tempera-
ture rises during a pulse, diffusion of volatile Xe in Al
phosphate accelerates. During dormant periods, the tem-
perature returns to normal, slowing down the diffusion.
However, a sole deficit of 136Xe cannot explain experi-
mentally observed Xe isotopic anomalies in Al phosphate
(dotted lines, Fig. 3). Evidently, a more complex process
was responsible for the transformation of the relatively
normal fission Xe in U-rich phase into the anomalous Xe
observed in Al phosphate. Such processes must generate
isotopes in the following proportions: 131 Xe=134Xe 3:4,
132Xe=134 Xe 7:0,and129Xe=134Xe 0:95 (slopes of
solid lines, Fig. 3).
Tellurium is known to be the most retentive fission
product in Oklo reactors [18]. Measured yields of fission
Te isotopes precisely match the fission product yield curve
[19]. This implies that Te -active precursors, such as
2:8m 132mSb,23 m 131 Sb,4:4h 129Sb,2:4m 129 Sn,
6:9m129mSn, also retained well in the reactor material.
We assume that fission isotopes of iodine, including the
long-lived 129 I, were retentive as well, otherwise we
would not find the excess of 130Xe produced by 129 I
neutron capture. In addition, numerous observations of
129Iand 129Xe in meteorites clearly demonstrate that
iodine is much more retentive than xenon (e.g., [20]).
Therefore, during a reactor pulse, the radioactive fission
tellurium and iodine migrate from U oxide into Al phos-
FIG. 2. Excess 130Xe (fission shielded) provides a mean for
an estimation of neutron fluence from the slope of the dashed
line. Data corrected to account for atmospheric contamination.
Negative 130Xe=136 Xe values are due to slight overcorrections.
FIG. 1. Fission 136Xe and 86Kr in U oxides (open ci rcles) and
Al phosphates (solid squares) after minor correction to account
for atmospheric contamination using 128 Xe and 82Kr. Xe con-
centrations were ca lculated from measured amounts of 136Xe
and the amount of degassed material which was estimated from
the specific density and geometry of the extraction crater.
Fission 136Xe concentration and 86Kr=136 Xe ratios in U-rich
phases tend to be lower, suggesting a migration of fission
products to Al phosphate.
VOLUME 93, NUMBER 18 PHYSICAL REVIEW LETTERS week ending
29 OCTOBER 2004
182302-2 182302-2
phate, where they subsequently decay to Xe. The latter,
however, is volatile and is retained in Al phosphate only
when the reactor cools down between operational pulses.
There is only one apparent problem: why is not the Xe
produced in the first pulse given off at the next one?
This problem can be resolved considering the conditions
at which Al phosphate has been formed. The high con-
centration of short-lived intermediate fission products in
Al phosphate without significant quantities of uranium
implies that it precipitated during the operation of the
Oklo reactor. Hydrothermal experiments demonstrate that
Al phosphate grows fast at relatively low temperatures
(270–300 C) [21]. After being captured by growing
Al phosphate, fission products decay into Xe which re-
mains imprisoned in Al phosphate because of its frame-
work crystalline structure [22] (similar to the cagelike
structure of zeolite). Then Xe can be released from the
alumophosphate only after destroying its cr ysta lline
structure, which requires temperatures higher than those
during the operational pulse of the Oklo reactor.
To calculate the evolution of isotope ratios of accumu-
lating Xe precursors during and after the active pulses of
the fission chain reaction, we considered only those fis-
sion fragments which have a fission yield greater than
0.1% and/or a half-life shorter than 1 min. We also
assumed that the three major fissile nuclei in the Oklo
reactor —235 Uth,239 Puth (thermal neutron fission) and
238Ufn (fast neutron fission) —have relative contributions
of 75%, 7%, and 18%, accordingly. These numbers were
determined [16] using Ru, Pd, Nd, Sm, and Gd isotopic
compositions measured in the same reactor zone where
our sa mple ca me from. Then , using known i ndividual and
cumulative fission yields [23] for 235Uth,239Puth,and
238Ufn, we calculated independent and cumulative yields
for each fission fragment relevant to Xe production in our
reactor zone. Finally, cumulative accumulation of Sn, Sb,
Te, and I in each isoba ric chain was computed and plotted
in the form of isotope ratios on Fig. 4.
Evidently, the calculated isotopic ratios are changing
with time, and the final Xe composition will depend on
how long the operational pulses last (d) and when the Al
phosphate cools down enough to retain Xe (p). We tried to
vary these two free parameters dand puntil all three
measured Xe ratios (determined from Fig. 3 and shown as
gray horizontal lines on Fig. 4) matched the calculated
isotope ratios. This turned out to be impossible. However,
if we considered only two ratios h131i=h134iand h132i=
h134i, there is one single solution d30 m and p
2:5h. And there is no solution for h129i=h134icombined
with eit her one of the two others. To match all three ratios
the value of 129 Xe=134Xe needs to be adjusted from the
measured 0.95 to about 1.5 (light gray line on Fig. 4). This
can be done assuming that 37% of 129Ihas been lost by A l
phosphate subsequent to the termination of the reactor
2 Ga ago, which is not unreasonable. 129Ihas a 16
106yr half-life, several orders of magnitude longer than
all other Xe precursors, is chemically active, forms water
soluble compounds and, therefore, has a chance of being
par tia lly leaked out from Al phosphate in the aqueous
environment of the reactor. Indeed, there is clear evidence
for 129Imigration from uranium deposits [24].
Interestingly enough, the 30 min pulses of natural
nuclear reactor activity and 2:5h dormant periods re-
corded in the Oklo Al phosphate resemble a typical geyser
operation. Similar time scales suggest similar processes.
This simila rit y suggests that 0.5 h after the onset of the
chain reaction, unbounded water was converted to steam,
decreasing the thermal neutron flux and making the
reactor subcritical. It took at least 2:5h for the reactor
to cool down until fission Xe began to retain. Then the
water returned to the reactor zone, providing neutron
moderation and once again establishing a self-sustaining
chain.
FIG. 3. Xe isotopic composition in U oxides and
Al phosphates are relat ed by a process that shifted points along
the solid lines. Dotted lines illustrate a sole deficit of 136 Xe,
which cannot explain the Xe anomalies in Al phosphate.
Migration of all isotopes in each isobaric chain must be
considered. Also shown are Xe components produced by the
three potential progenitors (235 Uth,239 Puth,238 Ufn).
VOLUME 93, NUMBER 18 PHYSICAL REVIEW LETTERS week ending
29 OCTOBER 2004
182302-3 182302-3
It is fascinating that Xe in Al phosphate measured
today provides us with such pristine timing records for
a natural reactor operated 2 billion yr ago.
We are grateful to Donald Bogard and the late Paul
Kuroda, with whom the idea of cycling operation of Oklo
reactor was discussed. A precious sa mple from Zone 13
was kindly provided by Maurice Pagel (GREGU, France).
This work was suppor ted by NASA (Grant No. NAG5-
12776).
[1] P. K. Kuroda, J. Chem. Phys. 25, 781 (1956).
[2] M. Neuilly et al., C. R. Acad. Sci. Paris 275, 1847 (1972).
[3] R. E. Bodu, H. Bounzigues, N. Morin, and J. P.
Pfiffelmann, C. R. Acad. Sci. Paris 275, D 1731 (1972).
[4] G. A. Cowan, Sci. Am. 235, 37 (1976).
[5] A. I. Shlyak hter, Nat ure (London) 264, 340 (1976).
[6] Y. Fujii et al., Nucl. Phys. B573, 377 (2000).
[7] R. D. Walton and G. A. Cowan, Le Phenomene D’Oklo,
IAEA-SM-204, Vienna, 499, 1975.
[8] R. Naudet , IAEA-SM-204/41 589, 1975.
[9] Yu. V. Petrov, Sov. Phys. Usp. 20, 937 (1977).
[10] Yu. Dym kov, Ph. Hollinger, M. Pagel, A. Gorshkov, and
A. Artykh ina, Miner. Depos. 32, 617 (1997).
[11] R. H. Nichols, Jr., K. Kehm, and C. M. Hohenberg, Adv.
Anal. Geochem. 2, 119 (1995).
[12] A. P. Meshik, K. Kehm, and C. M. Hohenberg, Geochim.
Cosmochi m. Acta 64, 1651 (2000).
[13] C. M. Hohenberg, Rev. Sci. Instr um. 51, 1075 (1980).
[14] R. J. Drozd, C. M. Hohenberg, and C. J. Morgan, Ea rth
Planet. Sci. Lett. 23, 28 (1974).
[15] A. P. Meshik, Ph.D. thesis, Vernadsk y Institute, USSR
Academy of Science, 1988.
[16] H. Hidaka and Ph. Hollinger, Geochim. Cosmochim.
Acta 62, 89 (1998).
[17] Ph. Hollinger, Les Nouvelles Zones de Reaction d’Oklo:
Datation U-Pb et Caracterisation in-situ des Products de
Fission a L’Analyseur Ionique, Rapport d’advancement.
Contract No. CCE-CEA: Oklo –Analoques Naturels,
1991.
[18] H. Hidaka, T. Konishi, and A. Masuda, Geochem. J. 26,
227 (1992).
[19] D. Curtis, T. Benja min, A. Gancarz, R. Loss, K. Rosman
and J. DeLaeter, Appl. Geochem. 4, 49 (1989).
[20] C. M. Hohenberg, O.V. Pravdivtseva, and A. P. Meshik,
Geochim. Cosmochim. Acta 68, 4745 (2004).
[21] E. A. Aldushina, V. S. Balitsky, A.V. Chicagov, O.V.
Samok hvalova, a nd G.V. Bondarenko, Exp. Geosci. 7,
82 (1999).
[22] O.V. Yakubovich and O. K. Mel’nikov, Crystallog raphy
14, 663 (1996).
[23] T. R. England and B. F. Rider, LA-UR-94-3106,
ENDF-349.
[24] J. Fabryka-Martin, S. N. Davis, D. Elmore , and P.W.
Kubik, Geochim. Cosmochim. Acta 53, 1817 (1989).
FIG. 4. T he calculat ed evolution of isotopic composition of
intermediate fission products Sb, Te, and I, which hold up the
production of stable Xe isotopes. Bold black lines correspond to
the active period of the reactor, with the numbers indicat ing the
duration of that period. Dashed lines illustrate free decay of Xe
precursors. The gray horizontal lines show the compositions
required to ma ke Xe in Al phosphate from Xe in U oxides (as
inferred by the slope of the solid lines observed in Fig. 3). The
light gray line represents adjusted h129i=h134iratio assuming
37% losses of 129I.
VOLUME 93, NUMBER 18 PHYSICAL REVIEW LETTERS week ending
29 OCTOBER 2004
182302-4 182302-4
