Comprehensive nuclear materials 5 22 minerals and natural analogues

5.22

Minerals and Natural Analogues

G. R. Lumpkin
Australian Nuclear Science and Technology Organisation, Kirrawee, NSW, Australia

T. Geisler-Wierwille
Universita¨t Bonn, Bonn, Germany

Crown Copyright ß 2012 Published by Elsevier Ltd. All rights reserved.

5.22.1
5.22.2
5.22.3
5.22.3.1
5.22.3.2
5.22.3.3
5.22.3.4
5.22.3.5
5.22.3.6
5.22.3.7
5.22.3.8
5.22.4
5.22.4.1
5.22.4.2
5.22.4.3
5.22.4.4
5.22.5
5.22.5.1
5.22.5.2

5.22.5.3
5.22.5.4
5.22.6
5.22.6.1
5.22.6.2
5.22.6.3
5.22.6.4
5.22.7
References
Appendix:

Introduction
a-Decay Damage in Minerals
Oxides
Pyrochlore Group
Zirconolite
Brannerite
Perovskite
Baddeleyite
Crichtonite
ABO4 and AB2O6 Minerals (B ¼ Nb, Ta, and Ti)
Hollandite
Silicates
Zircon
Thorite
Titanite (Sphene)
Allanite
Phosphates
Monazite
Apatite Group

Kosnarite and Related NZP Materials
Xenotime
Ore Deposits: Analogs for Spent Fuel
Preamble
General Aspects of Uraninite Alteration in Natural Systems
Natural Fission Reactors in Gabon
Uranium Migration in the Koongarra Ore Deposit
Conclusions
List of Mineral Names and Compositions
Oxides
Silicates
Phosphates, Arsenates, Vanadates
Carbonates, Fluorocarbonates, Fluorides
Sulfides
Uranyl Minerals (Section 5.22.6)

Abbreviations
apfu
AES
AFM

Atoms per formula unit
Auger electron spectroscopy
Atomic force microscopy

BSE

DTA

564

565
565
565
570
574
575
578
578
579
580
581
581
585
585
586
587
587
588
590
590
591
591
591
592
593
594
595
599
599
599

599
599
600
600

Backscattered electron, an imaging
mode in SEM, sensitive to mean
atomic number
Differential thermal analysis

563


564

Minerals and Natural Analogues

EDX
EELS
EPMA

EXAFS
HLW
ICP-OES
IMF
IR

Ln
M–M
M–O

M–O–M
MOX
NZP
RDF
SEM
Synroc
Synroc-C
Synroc-F
TEM
TOF-SIMS
Urf6
vpfu
XANES
XPS
XRD

Energy dispersive X-ray analysis
Electron energy loss spectroscopy
Electron probe microanalysis,
quantitative X-ray analysis with
crystal spectrometers
Extended X-ray absorption fine
structure
High level waste
Inductively coupled plasma-optical
emission spectrometry
Inert matrix fuel
Infrared, a type of vibrational
spectroscopy complementary to
Raman spectroscopy

Lanthanide series elements, La–Lu
Metal–metal distance in a crystal or
amorphous material
Metal–oxygen distance in a crystal or
amorphous material
Metal–oxygen–metal angle
Mixed oxide fuel, composed of
uranium and plutonium oxide
Sodium zirconium phosphate
Radial distribution function, a way of
describing M–O and M–M distances
Scanning electron microscopy
Synthetic rock
Hollandite + perovskite + rutile +
zirconolite-based material for HLW
Pyrochlore + uraninite ceramic for
partially reprocessed fuel
Transmission electron microscopy
Time-of-flight-secondary ion mass
spectrometry
A uranyl group with six equatorial
oxygen atoms
Vacancies per formula unit
X-ray absorption near edge structure
X-ray photoelectron spectroscopy
X-ray diffraction

5.22.1 Introduction
The immobilization and long-term disposal of
nuclear waste is one of the greatest challenges that

modern society faces today. Various types of high level
waste (HLW) have been generated from nuclear
operations around the world; for example, spent fuel
from commercial nuclear power stations, liquid waste
from the reprocessing of spent fuel, and waste from the

production of nuclear weapons and weapons grade
plutonium resulting from nuclear disarmament
treaties between the United States and Russia. Some
plutonium, particularly in France, has been used
in mixed oxide fuel (MOX, composed of uranium
and plutonium oxide) in place of the standard uranium
oxide fuel. Previous US policy adopted the strategy of
a once-through fuel cycle followed by direct disposal
of the spent fuel; however, recent changes have seen
a move toward more effective use of uranium-based
fuels in programs that combine reprocessing, transmutation, and separations technology in advanced fuel
cycles (e.g., Generation IV nuclear power systems).
Borosilicate glass is the currently accepted waste
form of choice for many countries that reprocess
their commercial spent fuel (see Chapter 5.18,
Waste Glass), but there exists a significant fraction
of ‘legacy’ waste and other nuclear materials that are
very complex in physical form and chemical composition (e.g., the Na-, Al-, and Zr-rich waste stored in
tanks at sites in the United States). These complex
waste materials, together with impure plutonium, and
the separated fission products and actinides generated
from the various partitioning strategies may be better
suited for existing or new types of high-performance
crystalline waste forms or glass-ceramics (see Chapter

5.19, Ceramic Waste Forms). Some of these materials, for example, inert matrix fuels (IMFs), are being
designed for recycling of reactor-grade plutonium and
minor actinides in commercial power stations, followed by geological disposal, an attractive option that
does not generate new plutonium.1
As envisaged by G.J. McCarthy, A.E. Ringwood,
and others in the 1970s, there exist alternative crystalline waste forms that may be capable of providing
a much higher level of chemical durability than
borosilicate glass or directly disposed spent fuel.
Many of these materials have been extensively developed over the previous 20–25 years, while others are
relatively new. Materials such as tailored ceramics,2
the synthetic rock (Synroc) polyphase titanate waste
forms,3,4 and related special purpose waste forms are
reasonably well developed and have been the subject
of extensive leach testing and radiation damage studies. Pyrochlore is the major component of Synroc-F, a
polyphase ceramic designed for partially reprocessed
nuclear fuel5 and later appeared as the principal host
phase for excess weapons Pu and U in a crystalline
titanate ceramic form. Zirconolite has also been proposed as an ideal host phase for actinides due to a
combination of crystal chemical flexibility and very
high durability in aqueous fluids,6 and hollandite


Minerals and Natural Analogues

may provide an excellent host material for separated
long-lived radioactive Cs for similar reasons.7
Additional special purpose waste forms for actinides
include zircon,8 monazite,9 and zirconium-based materials having the fluorite, defect fluorite, or pyrochlore
structures.10,11 Except for zircon, none of these materials has been studied to the same extent as the titanate
waste forms. Nevertheless, monazite and Zr-based

materials are promising in view of their resistance to
amorphization and excellent chemical durability.
Nuclear waste form materials must meet several requirements in order to reach final consideration for use
in a repository, including a high level of durability in
aqueous fluids, crystal chemical flexibility allowing the
material to cope with variations in the composition of
the waste stream, reasonably high waste loadings, volume reduction, and reliable and cost-effective processing technologies. Information derived from minerals
can be used to assess all but the latter criterion. Furthermore, studies of U ore deposits are useful in the assessment of the performance of spent fuel, including
transport of U away from the repository.
The purpose of this chapter is to summarize the
performance of minerals in terms of their response to
a-decay damage and their interactions with natural
aqueous fluids in geological environments. For comparison, we also discuss some of the relevant literature
results involving accelerated radiation damage of synthetic compounds doped with short-lived actinides,
controlled laboratory experiments on the dissolution
of synthetic materials with and without short-lived
actinides, and dissolution of radiation-damaged natural samples. Following a brief introduction to uraninite
and its alteration products in natural systems, we conclude with an overview that uranium ore deposits as
analogs for spent fuel under repository conditions,
including aspects of the natural fission reactors in
Gabon and uranium migration around the Koongarra
ore body, Northern Territory, Australia.

5.22.2 a-Decay Damage in Minerals
In this section, we briefly summarize the effects of
a-decay damage on the structures of some of the
more important natural analogs. Here, it is important
to point out that a-decay involves two concurrent
processes, the release of a high energy ($4–5 MeV)
a-particle together with a low energy ($70–100 keV)

recoil atom. The process can be expressed in the
following general way:
A nþ
ZP

ðnÀ2Þþ 4
!AÀ4
þ2 He2þ
ZÀ2 R

½IŠ

565

In this expression, P is the parent isotope, R is the
recoiling daughter isotope, and the charged He atom
is the emitted a-particle. The symbols A, Z, and n
represent the mass, atomic number, and nuclear
charge, respectively. The massive recoil nucleus has
a range of 20–25 nm and typically displaces on the
order of 1000 atoms primarily by nuclear stopping
processes; whereas, the a-particle has a range of
about 10–15 mm and loses most of its energy through
electronic interactions before displacing on the order
of 100 atoms near the end of its track. The valence
states of the a-particle and recoil atom are rapidly
reconfigured in the solid to produce 4He and to return
the recoil atom to a more stable state. This charge
reconfiguration process may be complex in the case of
U that can exist as the U4+, U5+, and U6+ ions in solids

or in general if there are other elements present with
variable valence states such as the transition metals
(see Section 5.22.4.3).
The important parent isotopes in minerals are
238
U, 235U, and 232Th. These isotopes decay to the
stable isotopes 206Pb, 207Pb, and 208Pb through their
respective decay series. Based on each decay series,
the total a-decay dose D can be calculated using the
following equation:
D ¼ 8N238 ðel238 t À 1Þ þ 7N235 ðel235 t À 1Þ
þ 6N232 ðel232 t À 1Þ

½1Š

In this equation, t is the geological age, N represents
the present-day concentration of the parent isotope,
and l is the decay constant. This equation is strictly
applicable to samples wherein the isotopic composition
of the U has been determined. In situations where
only the Th and U elemental concentrations have
been determined, one may assume that the U isotopic
composition consists of 99.28% 238U and 0.72% 235U.
Alternatively, the second term in the equation may be
ignored without major consequence, as the associated
error is usually smaller than the uncertainty in the
geological age. In this chapter, we give all dose values
in units of 1016 a-decays per milligram (this is
because 1 Â 1016 a-decays per milligram is approximately equivalent to one displacement per atom in
minerals).

5.22.3 Oxides
5.22.3.1

Pyrochlore Group

Pyrochlore is an anion-deficient derivative of the
fluorite structure type with a doubled a cell


566

Minerals and Natural Analogues

parameter and change in space group from Fm3¯ m to
Fd3¯ m.12–14 Minerals of the pyrochlore group conform to the general formula A2ÀmB2X6ÀwY1ÀnÁpH2O,
where A represents cations in eightfold coordination,
B represents cations in sixfold coordination, and
X and Y are anion sites. The basic structural element
of pyrochlore is the framework of corner-sharing
octahedra. Within this framework, continuous tunnels
exist parallel to the h110i directions. Both the A-site
cations and Y-site anions are located in these tunnels.
In synthetic systems, some A-site cation exchange
capacity has been demonstrated in defect pyrochlores,
in which the values of m in the general formula can be
quite large.
Most natural pyrochlores form under magmatic
conditions in granitic pegmatites, nepheline syenite
pegmatites, and carbonatites, or late-stage veins associated with these rock types. The composition of

common pyrochlore usually approaches the stoichiometric form (Na, Ca, Ln, U)2(Nb, Ta, Ti)2O6(F, OH, O),
but the structure type is extremely flexible in
terms of the sheer number of elements that can
be incorporated and is particularly amenable to the
incorporation of actinides. Natural samples are known
to contain up to 30 wt% UO2, 9 wt% ThO2, and 16 wt%
Ln2O3, an important consideration for the issue of
nuclear criticality. However, as shown by the general
formula, the crystal chemistry of pyrochlore is complicated by the potential for vacancies at the A-, X-,
and Y-sites (m ¼ 0.0–1.7, w ¼ 0.0–0.7, and n ¼ 0.0–1.0)
and the incorporation of water molecules (p ¼ 0–2)
in the vacant tunnel sites. The total water content
of the natural defect pyrochlores may be as high as
10–15 wt% H2O (with speciation as both water molecules and OH groups).
In a little known but classic paper, Krivokoneva
and Sidorenko15 examined a suite of Russian pyrochlores using X-ray diffraction (XRD) methods. An
analysis of the line broadening showed that strain
increased from 0.0009 to 0.0035 as crystallite dimensions decreased from 100–120 nm down to 35–40 nm
in the initial stages of damage, a decrease to 15 nm
was observed in the latter stages of damage. These
authors also carried out an analysis of the radial
distribution function (RDF) of an amorphous sample
and showed that there was no long-range order present beyond the second coordination sphere. However,
peaks in the RDF representing the major M–O and
M–M distances showed that the fundamental structural units (e.g., the coordination polyhedra) still
existed in the amorphous state. Lumpkin and
Ewing16 also used XRD to determine both the

beginning (Di) of the crystalline–amorphous transformation and the critical amorphization dose (Dc) for a
large suite of pyrochlores from different localities.

They showed that the transformation zone increased
in dose as a function of the geological age of the
samples. Both dose curves are well-described by an
equation of the form:
Di;c ¼ D0 etK

½2Š

In this expression, D0 is the intercept dose for Di or
Dc and K is a rate constant. Analysis of the dose-age
data gives a value of D0 ¼ 1.4 Â 1016 a per milligram
for the amorphization dose curve and K ¼ 1.7 Â 10À9
yearÀ1.17 The Bragg peak intensities of a subset of
these samples were fitted to an equation of the form:
I =I0 ¼ eÀBD

½3Š

Here, I/I0 represents the total intensity of all
observable Bragg peaks divided by the total intensity
obtained from an undamaged sample of similar
composition and B is a constant related to the amount
of material damaged by each a-decay event. Equation
[3] gives an excellent fit to the data with
B ¼ 2.6 Â 10À16 mg per a-particle, corresponding to
an average cascade radius of 2.3 nm in which a maximum of 2600 atoms are displaced. An analysis of line
broadening in these samples showed that crystallite
dimensions decreased from about 500 to 15 nm with
increasing dose. Strain initially increased with dose
and reached a maximum of $0.003 before falling to

values below 0.0005 at higher dose levels, consistent
with a description of the crystalline–amorphous
transformation as a type of ‘percolation’ transition.18
With increasing a-decay dose, transmission electron
microscopy (TEM) images reveal mottled image
contrast due to strain, followed by the appearance of
local amorphous domains that increase in volume and
begin to overlap to produce larger amorphous areas
until they are connected throughout the material.
This is the first percolation transition. With further
increases in dose, the crystalline areas diminish in
volume until they become isolated, giving way to a
microstructure dominated by amorphous pyrochlore.16
This is the second percolation transition.
During the 1980s, Greegor and coworkers19–22
carried out several studies of the local structure and
bonding around Ti, Nb, Ta, and U atoms in pyrochlore using EXAFS–XANES (extended X-ray
absorption fine structure–X-ray absorption near
edge structure). Results of these studies demonstrated that the M–O coordination polyhedra of
amorphous pyrochlore exhibit reduced bond


Minerals and Natural Analogues

Zircon
Apatite
Pyrochlore
Zirconolite

14

12
DV/ V0 (%)

distances, reduced coordination number, and
increased distortion relative to the undamaged crystalline structure. Furthermore, there was no periodicity in evidence beyond the second coordination
sphere, with some disruption of the M–M distances.
From these studies, it was realized that only a slight
increase in the mean M–M distance was required in
order to explain the overall increase in volume
caused by a-decay damage and that this could be
facilitated by increased M–O–M angles. The thermal
behavior of radiation-damaged natural pyrochlore
was investigated using differential thermal analysis
(DTA) and XRD.23,24 Results of this study indicated
that the samples recrystallized in the range of
400–700  C, depending upon the composition and
degree of crystallinity. Measured values of the recrystallization energy are 125–200 J gÀ1 and are inversely
correlated with the level of crystallinity.
In the early to mid-1980s, Clinard and his colleagues conducted an extensive set of experiments
on ‘cubic zirconolite’ CaPuTi2O7 in which the Zr
is completely replaced by 238Pu (t1/2 ¼ 87.7 years).
This material is actually a pyrochlore compound
similar to synthetic CaUTi2O7. In their first publication, Clinard et al.25 reported that CaPuTi2O7 has a
total volume expansion of 4.7%, an ‘apparent’ lattice
volume expansion of 2.2%, and a critical amorphization dose of 0.3 Â 1016 a per milligram based on XRD
analysis. Further analysis of the data showed that
CaPuTi2O7 exhibits a bulk volume expansion of
5.4% at ambient temperature and becomes amorphous at a dose of 0.5 Â 1016 a per milligram based
on bulk swelling curves. For samples held at 302  C,
the bulk swelling saturates at 4.3%, and the material

becomes amorphous at a dose of 1 Â 1016 a per milligram as estimated from the swelling data. This is a
very significant result, as it indicates that the critical
dose for an experiment lasting $3 years at 302  C
is roughly equivalent to what nature produces in
107–109 years. When stored at a temperature of 602  C,
CaPuTi2O7 did not become amorphous; however, the
material showed a bulk expansion of 0.4% consistent
with accumulation of lattice point defects.26 In retrospect, this is a stunning result and represents the first
and only realistic ‘bracket’ for the critical temperature
for amorphization of a nuclear waste form material.
Also during the 1980s, Weber et al.27 investigated synthetic Gd2Ti2O7 doped with 3 wt%
244
Cm (t1/2 ¼ 18.1 years) and determined the amorphization dose of $0.4 Â 1016 a per milligram with
B ¼ 4.4 Â 10À16 mg per a-particle, a total volume
expansion of $5.1% at saturation (Figure 1), and

567

10
8
6
4
2

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Dose (1016 a per milligram)
Figure 1 Plot showing total volume expansion of
synthetic pyrochlore Gd2Ti2O7 doped with 244Cm,
zirconolite CaZrTi2O7 doped with 244Cm, apatite (e.g.,
britholite) CaNd4(SiO4)3O doped with 244Cm, and zircon
(ZrSiO4 doped with 238Pu) as a function of increasing
a-decay dose.

an increase in fracture toughness together with a
decrease in hardness and elastic modulus. Changes
in microstructure with increasing dose mimic the
results for natural pyrochlores described earlier.
Based on the results of DTA experiments, an activation energy of Ea ¼ 3.8 eV was determined for recrystallization of this pyrochlore. A recrystallization
temperature of 700–800  C was determined by isochronal annealing. The authors also performed leach
tests on single-phase Cm-doped Gd2Ti2O7 pyrochlore samples. In this work, the leach tests were
limited to annealed, fully crystalline and fully amorphous samples, and were exercised at 90  C in pure
water for 14 days. The experiments revealed weight
losses of 0.02% and 0.05% for the crystalline and
amorphous pyrochlore samples, respectively. The
results of this study also indicated that the leach
rate of Cm increased by a factor of 17 as a consequence of amorphization.
More recently, Strachan and coworkers28 investigated the effect of 238Pu on the structure of four

synthetic pyrochlore samples with variable amounts
of Al, Gd, Hf, and U. The results of detailed XRD and
bulk swelling measurements indicate that the critical
dose for amorphization is $(0.2–0.4) Â 1016 a per
milligram and is associated with a total volume
expansion of <6%. Based on changes in the cubic
lattice parameter with time, it appears that the unit


568

Minerals and Natural Analogues

cell expansion of the pyrochlore phase is on the order
of 2.9–4.7%. Using purpose-built equipment for flow
through dissolution tests, Strachan et al.28 examined
the behavior of 238Pu-doped pyrochlore at pH ¼ 2–12
and 85–90  C. They found very low release rates
based on Pu and to a lesser extent U, possibly due
to solubility controls on these elements. Experiments
carried out on amorphous and recrystallized samples
demonstrated very similar release rates for Gd at
pH ¼ 2 and 85  C. This is a very important conclusion of this major research project and, together
with the observation that the materials did not
develop cracks with increasing dose, lends substantial
credence to the use of these ceramics as nuclear waste
forms. Following a careful analysis of the effects
of flow rate and specimen surface area, forward dissolution rates of (0.7–1.3) Â 10À3 g mÀ2 dayÀ1 were
obtained for two different samples. Other experiments have been conducted in order to determine
the kinetics of U release from pyrochlore, (Ca,Gd,Ce,

Hf,U)2Ti2O7, but without the complicating effects of
short-lived actinides.29,30 Both studies report that the
pH dependence follows a shallow v-shaped pattern
with a minimum near neutral pH. The release rates
for U, converted from the limiting rate constants
given by Zhang et al.,30 range from 6 Â 10À7 to
7 Â 10À5 g mÀ2 dayÀ1 for all experimental conditions
(e.g., T ¼ 25–75  C and pH ¼ 2–12).
Geisler et al.31–33 performed hydrothermal experiments with a natural, crystalline Ta-based pyrochlore
(microlite) from Lueshe near Lake Kivu of the
Democratic Republic of Congo in pure water and
acidic solutions (pH ¼ 0) at 175 and 200  C. The
hydrothermal treatment in the acidic solutions causes
the partial replacement of the microlite by a new defect
pyrochlore that is characterized by a larger unit cell
volume, a large number of vacancies at the A-site
(A ¼ Ca, Na) and anion vacancies, by molecular
water, and possibly, OH groups. Analyses of the
experimental fluid further revealed that U was lost
to the solution. TEM investigations of the interface
between the new defect pyrochlore and the unreacted
microlite revealed a topotactic relationship between
both pyrochlore phases. Furthermore, the interface
between both phases was found to be sharp on the
nanoscale with a sharp, step-like decrease of the Ca
and Na content at the interface toward the defect
pyrochlore. Time-of-flight-secondary ion mass spectrometry (TOF-SIMS) and confocal micro-Raman
mapping of the defect pyrochlore produced in an
acidic solution that was enriched with 18O ($47.5 at.%)
revealed that the defect pyrochlore is strongly

enriched with 18O with a sharp 18O gradient to relict
unreacted areas. The authors suggested that the
replacement of microlite by a defect pyrochlore
occurs by a pseudomorphic reaction that involves
the dissolution of the pyrochlore parent accompanied
by the simultaneous reprecipitation of a defect pyrochlore at a moving dissolution–reprecipitation front;
a process that has been named interface-coupled
dissolution–reprecipitation process. It is noteworthy
that the treatment in pure water for 14 days at 175  C
did not produce reaction zones detectable by backscattered electron (BSE) imaging. However, significant spectral changes in powder infrared (IR) spectra
of the reaction product and the detection of Na and
Ca in the experimental solution indicated that the
microlite has also reacted in pure water. The experimental, chemical, and textural alteration features bear a
remarkable resemblance to those seen in naturally
altered microlite samples (see Figure 2 and subsequent
discussion).
Later, Po¨ml et al.34 studied the hydrothermal
alteration of a natural, heavily radiation-damaged
pyrochlore (betafite) from a rare earth pegmatite
from Lindvigskollen near Kragerø, South Norway
and a synthetic titanate-based pyrochlore ceramic
[(Ca0.76Ce0.75Gd0.23Hf0.21)Ti2O7] produced at the
Lawrence Livermore National Laboratory, USA.
The authors treated cuboids of both samples with
edge lengths of $3.3 mm in a 1 M HCl solution containing 43.5 at.% 18O at 250  C for 72 h. During the
experiments, both samples were transformed mainly
into rutile with subordinate anatase. The degree of
transformation was significantly higher for the natural
radiation-damaged pyrochlore; for example, 18O was

highly enriched with the reaction products of both
samples with a sharp gradient (on a micrometer scale)
toward the unreacted pyrochlore and no apparent
diffusion profile. The replacement reaction retained
even fine-scale morphological features typical for
pseudomorphs. Based on these observations, the
authors suggested that the dissolution of pyrochlore
is spatially and temporally coupled with the precipitation of stable (metastable) TiO2 phases at an
inwardly moving reaction front, a mechanism that is
essentially the same as proposed for the experimental
replacement of crystalline microlite by a defect pyrochlore as discussed in the previous paragraph. The
authors pointed out that their results produced under
relatively extreme batch-experimental conditions show
similarities with nature as well as with results derived
from experiments conducted under moderate conditions rather expected in a nuclear repository.


Minerals and Natural Analogues

(a)

(b)

20 mm Na+

BSE

(c)

(d)

Ca+

18O-/O-

0
Reaction rim / new pyrochlore phase
Unaltered core / untreated microlite
Altered natural microlite
Unaltered natural microlite

A
0.00

0.25

1.00

0.25

0.50

0.75

0.00
1.00

A+

0.75

0.50

(e)

0.25

1.00

O–5
(f)

0.00

0.3

1.00

0.75

0.25

1.00
0.00
A+2

0.00

0.2

y

0.50

0.50

0.75

0.1
+

0.25

0.75

0.50

x

569

0.25

0.50

0.75

0.00
1.00

F

Figure 2 Experimental hydrothermal aqueous alteration of a natural pyrochlore (e.g., Ta-rich variety known as microlite) in
highly acidic solution at 200  C. Backscattered electron image (a) indicates increased average atomic number in altered
(brighter) areas, due to selective loss of the light cations Na (b) and Ca (c), and enrichment in 18O (d). Triangular diagrams also
indicate that changes in cation (e) and anion (f) compositions are similar to those observed in nature.

Numerous investigations have demonstrated that
natural pyrochlores are susceptible to alteration via
reaction with aqueous fluids over a range of conditions
involving pressure, temperature, and fluid composition.
At higher temperatures ($300–650  C, <400 MPa) in
highly evolved late-stage magmatic fluids, Ca enrichment is commonly observed; whereas, the main effect of
alteration at moderate temperatures under hydrothermal conditions ($200–350  C, <200 MPa) is the loss
of Na and F, often combined with cation exchange for
Sr, Ba, REE, and Fe. Further removal of Na, F, Ca, and
O may occur in low temperature hydrothermal or
weathering environments, resulting in the maximum
numbers of A-site, Y-site, and X-site vacancies, maximum hydration levels, and more limited exchange large
cations such as K, Sr, Cs, Ba, Ce, and Pb in certain
environments.35–45
Ti-rich pyrochlore (betafite) from hydrothermal
veins (Figure 3) in the contact metamorphic zone
adjacent to the Adamello igneous massif in northern
Italy contain 29–34 wt% UO2 and are chemically the
closest natural analogs presently known for nuclear
waste forms. Electron microscopy and microanalytical work have revealed that these pyrochlore samples

Figure 3 Photograph showing dark-colored Ti-rich
hydrothermal veins in marble from the contact metamorphic

zone of the Adamello igneous massif in northern Italy. These
veins have transported U, Th, and lanthanide elements as
demonstrated by the presence of pyrochlore (e.g., Ti-rich type
known as betafite) and zirconolite. Width of image ¼ 10 cm.

have only suffered a minor late-stage hydration event
as evidenced by lower backscattered electron image
contrast around the rims of the grains (Figure 4).


570

Minerals and Natural Analogues

of the U-rich rim. During this alteration, which is the
result of exposure to tropical conditions, Na, Ca, and
F were removed from the pyrochlore, thus leading to
increased A-site vacancies (up to about 1.8 A-site vpfu).
The alteration also led to localized redistribution of
radiogenic Pb and to hydration, but U remained immobile. Although U loss has been documented in certain
geological environments, it appears that U is highly
stable on the A-site of the pyrochlore structure. As
natural pyrochlore can accommodate U6+ in significant
amounts, it is possible that the geometry of the A-site,
which is similar to the Urf6 topology in uranyl minerals
with two short bonds and six long equatorial bonds,49
plays a role in the geochemistry of pyrochlore.
Figure 4 Backscattered electron image of pyrochlore
(betafite) and zirconolite from Ti-rich veins, Adamello, Italy.
Pyrochlore occurs as overgrowths on highly zoned

zirconolite crystals. Note the slightly darker rim on the
pyrochlore, due to minor loss of Na, F, and increased
hydration. Zirconolite was unaffected by this alteration
event. Width of image ¼ 60 mm.

Results of this study demonstrate quantitative retention of U and Th for time periods of 40 Ma, even
though the crystals experienced cumulative total
a-decay doses of 3–4 Â 1016 a per milligram.46 In
two samples from Bancroft, Ontario, Canada, Lumpkin
and Ewing39 had previously concluded that the major
result of alteration was hydration, with only minor
changes in elemental composition, apart from the precipitation of galena due to mobility of radiogenic Pb. In
contrast to these examples, the Ti- and U-rich pyrochlores from granitic pegmatites in Madagascar exhibit
a range of alteration effects, including relatively hightemperature, postmagmatic hydrothermal processes,
and lower temperature alteration.39,47 If the Ca content falls below 0.2–0.3 apfu, these Ti-rich pyrochlores
may exhibit various levels of recrystallization to liandratite þ rutile (or anatase). In the most severe cases
documented, this may be accompanied by a loss of up
to 20–30% of the original amount of U and local
redistribution of the radiogenic Pb.
A recent study of a 440 Ma pyrochlore from
Mozambique provides qualitative information on
the effect of radiation damage on the alteration of
pyrochlore.48 These pyrochlore crystals exhibit a distinct growth zoning, characterized by a U-free core
and a U-rich rim (up to 17 wt% UO2). Following
uplift and cooling, groundwater penetrated these
fractured crystals and led to the deposition of clay
minerals along both fractures and cleavage planes.
This low-temperature process also led to chemical
alteration of the pyrochlore but only within the zone

5.22.3.2

Zirconolite

The structure of zirconolite14 is also considered to be
an anion-deficient derivative of the fluorite structure
type and can be viewed as a volumetrically condensed, layered version of pyrochlore with reduced
symmetry and several polytypic forms (monoclinic
2 M or 4 M, both with space group C2/c; orthorhombic 3O with space group Acam, hexagonal 3T with
space group P3121). The chemical composition of
zirconolite 2 M corresponds to CaZrTi2O7, but in
nature, it commonly deviates from this end-member
composition due to extensive substitution of Y, Ln,
Th, and U for Ca and Nb, Fe, and Mg for Ti.50,51 In
natural samples, Zr is subject to only limited substitution by other elements (e.g., minor amounts of Y,
Ln, U, and Ti). Experimental work has shown that
extensive substitution of REE, Th, and U generally
results in a polytypic phase transformation from
monoclinic 2 M to trigonal 3T or from monoclinic
2 M to monoclinic 4 M. Zirconolite is an important,
highly durable host phase for actinides and fission
products, with the ability to incorporate up to 24 wt%
UO2, 22 wt% ThO2, and 32 wt% Ln2O3 in natural
systems.
Lumpkin et al.23,24 discussed the results of an
extensive study of amorphous and annealed zirconolite
from Sri Lanka using a variety of methods, including
XRD, EXAFS–XANES, TEM, and DTA. Electron
diffraction and high-resolution TEM studies suggested that amorphous zirconolite lacked periodicity
beyond the second coordination sphere, consistent

with a random network model of the amorphous state.
EXAFS–XANES results provided more detailed
information for the Ti- and Ca-sites and indicated
that amorphous zirconolite lacked periodicity beyond
the first coordination sphere, with reduced M–O


Minerals and Natural Analogues

bond lengths, reduced coordination number, and
increased distortion of the Ti–O polyhedra (determined from a prominent pre-edge feature in the
XANES results). Farges et al.52 provided additional
results for the Zr-, Th-, and U-sites and reported
nearly identical coordination numbers and bond
lengths for the amorphous and annealed samples.
However, a significant increase in the range of Zr–O
and Th–O distances was observed, leading the authors
to conclude that slight variation of the M–O–M
angles can have a profound effect on long-range
periodicity and medium-range order. Later work
also confirmed the reduced coordination of Ti in
the amorphous samples, pointing specifically to a
fivefold coordination environment.53
Annealing studies of the amorphous zirconolites
from Sri Lanka were performed using DTA and showed
exotherms at $780  C due to recrystallization.23,24 For
two different samples, the energy release associated
with recrystallization was 43 and 48 J gÀ1. TEM
investigation of samples annealed at 1100  C showed
that the structure recovered to monoclinic zirconolite2 M, but the material was highly twinned and

contained some stacking faults and intergrowths of
other polytypes on (001). A few polycrystalline grains
were also observed, which gave electron diffraction
patterns consistent with the fluorite structure type.
From these observations, it was suggested that amorphous zirconolite initially recrystallizes to a disordered, defect fluorite structure.
A detailed study of highly zoned zirconolite samples from the contact metamorphic zones of the
Bergell intrusion ($30 Ma) on the Swiss–Italian border and the Adamello igneous complex ($40 Ma) in
northern Italy provided the first detailed results on
the crystalline–amorphous transformation in natural
zirconolites.54 The crystals from both localities contain a wide range of ThO2 and UO2 concentrations up
to a combined maximum of over 20 wt%, thus ensuring a substantial range of a-decay dose (see Figure 4).
Due to the small grain size, the two suites of samples
were characterized by analytical electron microscopy.
In particular, a series of TEM dark field images
revealed the percolation-like behavior with increasing
dose: (1) the appearance of mottled diffraction contrast
(0.08 Â 1016 a per milligram), (2) extensive development of amorphous domains in a crystalline matrix
(0.3–0.5 Â 1016 a per milligram), and (3) overlap of
collision cascades to produce larger amorphous areas
that eventually dominate the structure as crystalline
domains diminish in size to <10 nm (0.7–0.9 Â 1016
a per milligram).

571

A study of seven suites of zirconolite samples ranging in age from 16 to 2060 Ma and with a-decay doses
of 0.008 Â 1016 to 24 Â 1016 a per milligram has also
been reported.55,56 For each suite of samples, the
beginning of the crystalline–amorphous transformation (onset dose) was defined as the first appearance
of mottled diffraction contrast in bright field images.

The end of the transformation (critical amorphization dose) was defined by the complete disappearance
of Bragg diffraction spots, leaving only diffuse rings
at $3.0 and 1.8 A˚ in diffraction patterns. These dose
‘brackets’ were used to construct a plot of dose versus
age, revealing a pattern of upward curvature with
increasing age for both the onset dose and critical
dose. As in the previous work on pyrochlore, this
upward curvature was interpreted as evidence for
long-term annealing of isolated a-recoil collision
cascades back to the crystalline structure. The data
were fitted using eqn [2] in order to determine the
intercept dose and annealing rate constant. Curve fits
gave values of D0 ¼ 0.11 Â 1016 a per milligram and
K ¼ 1.0 Â 10À9 yearÀ1 for the onset dose curve; and
D0 ¼ 0.94 Â 1016 a per milligram and K ¼ 0.98 Â 10À9
yearÀ1 for the critical dose curve.
Due to potential dose rate effects and other factors, it is instructive to compare the work summarized earlier with actinide doping experiments. Weber
et al.27 published a detailed body of work on radiation
damage in CaZrTi2O7 doped with $3 wt% 244Cm.
Synthetic zirconolite becomes amorphous at a dose
of $0.5 Â 1016 a per milligram with a total volume
expansion of 6.0% at saturation (Figure 1) and
B ¼ 4 Â 10À16 mg per a-particle for the amount of
material damaged in the collision cascade. The crystalline–amorphous transformation in this material
also leads to an increase in fracture toughness and a
decrease in hardness and elastic modulus. DTA studies show that the fully amorphous material exhibits
exothermic reactions at 500–530 and 680–700  C
and releases about 13 and 114 J gÀ1 of stored energy,
respectively, in the two exotherms. Isochronal annealing indicated that the main phase of recrystallization
occurs at 500–700  C; however, the zirconolite initially

forms as a pseudocubic (rhombohedral) structure before
transformation to monoclinic zirconolite at about
900  C. As shown in Figure 5, this material exhibits
anisotropic lattice expansion with increasing dose
(data from Wald and Offermann57). The a lattice
parameter increases by 0.3% and exhibits saturation
behavior at higher doses, the b parameter increases
marginally by 0.1% at low dose and then remains
nearly constant, whereas the c cell dimension


572

Minerals and Natural Analogues

1.25
0.729

1.248

b (nm)

a (nm)

1.249

0.728

1.247
0.727

1.246

(a)

(b)

1.035

1.155

1.145

Vc (nm3)

c (nm)

1.03
1.15

1.02

1.14

(c)

1.025

1.015

(d)

0.5

1

1.5

2

2.5

Dose (1015 a per milligram)
Figure 5 Plots showing anisotropic lattice expansion in zirconolite (CaZrTi2O7) doped with 244Cm. As shown in (a) and (b),
the a and b parameters exhibit normal ‘saturation’ behavior and increase by 0.32% and 0.14%, respectively. The c cell
parameter does not show saturation and increases by 1.6% up to the maximum dose of the experiment. The total unit cell
volume expansion (d) is 2.0%.

continues to expand with dose and reaches a maximum value of 1.5% relative to the undamaged zirconolite. In a concurrent study of synthetic zirconolite
doped with 238Pu, Clinard et al.58 reported a total
volume expansion of 5.5% and a critical amorphization dose of $0.5 Â 1016 a per milligram.
In a recent study of radiation damage via actinide
doping experiments, Strachan and coworkers59 investigated the effect of 238Pu on the structure of three synthetic zirconolites that also contained variable amounts
of Al, Gd, Hf, and U but no Zr. Although these are
not pure single-phase specimens, detailed XRD and
bulk swelling measurements indicate that the critical
dose for amorphization is $(0.3–0.5) Â 1016 a per
milligram and is associated with a total volume expansion of $5%. By employing similar procedures to
those used in their previous study of 238Pu-doped
pyrochlore (Section 5.22.3.1), Strachan et al.59

showed that the forward dissolution rate of radiation-damaged zirconolite is 1.7 Â 10À3 g mÀ2 dayÀ1
at pH ¼ 2 and 90  C. The dissolution tests exhibited
very little dependence on pH, were not dependent on

the level of radiation damage, and no cracking was
observed in the zirconolite specimens.
The dissolution of synthetic zirconolite without
short-lived actinides has been determined as a function of pH using pure water in single pass flow through
tests at temperatures of 75  C and lower.29,30 These
authors have independently studied a Ce–Gd–Hf zirconolite containing about 16 wt% UO2 and the results
of the two studies are similar. Release rates determined
by Zhang et al.30 for Ti and U indicate that zirconolite
dissolves congruently after about 20 days following an
initial period where U is released at a somewhat faster
rate than Ti. The limiting rate constants are equivalent
to U release rates of 6.4 Â 10À7 to 1.3 Â 10À5 g mÀ2
dayÀ2 for zirconolite over the entire pH range of 2–12
and temperature range of 25–75  C. The dissolution
rate of zirconolite is characterized by a shallow
v-shaped pattern with a minimum near pH ¼ 8, similar to the results obtained for pyrochlore.
Isotopic age dating work by Oversby and Ringwood60
and electron microscopy studies by Ewing et al.61
have shown that natural zirconolite exhibits closed


Minerals and Natural Analogues

system behavior for U, Th, and Pb for up to 650 Ma
with little, if any, evidence for geochemical alteration.
More recently, Rasmussen and Fletcher62 proposed

that zirconolite may become the principal mineral for
age dating in mafic igneous rocks due to its ability to
retain radiogenic Pb. Their analysis of 1200 Ma dolerite
intrusive rocks from Western Australia demonstrated
that zirconolite returned the same 207Pb/206Pb age as
zircon and baddeleyite, but the zirconolite age was
much more precise (by factors of $3.3 and 13, respectively). Lumpkin et al.17 and Hart et al.63 have described
the alteration of amorphous zirconolite from the
2060 Ma carbonatite complex of Phalaborwa, South
Africa in somewhat greater detail. Electron microprobe analyses, element mapping, and backscattered
electron images demonstrate that the alteration is
localized along cracks and resulted in the incorporation of Si and loss of Ti, Ca, and Fe. However, in these
samples, the Ln, Y, Th, and U contents remained
relatively constant across the alteration zones. Radiogenic Pb appears to have been mobile and precipitated mainly within the altered areas as galena. In
carbonatites, zirconolite may be replaced along
cracks and within micron-sized domains by an
unidentified Ba–Ti–Zr–Nb–ACT silicate phase, suggesting that zirconolite may not be stable in the
presence of relatively low temperature hydrothermal
fluids enriched with aqueous silicate species.64,65
At higher temperature and pressure in magmatic,
hydrothermal, or metamorphic systems, zirconolite
may be altered by dissolution–reprecipitation or
replacement mechanisms. Giere´ and Williams66
have described zoned zirconolites from Adamello,
Italy, which exhibit corrosion and replacement by
a new generation of zirconolite together with loss
of Th and U to a hydrothermal fluid. Importantly,
a thermodynamic analysis of the mineral assemblages
was performed in this work and indicated that the
zirconolite crystallization and corrosion occurred at

500–600  C in a reducing hydrothermal fluid rich in
H2S, HCl, HF, and P and relatively poor in CO2. Pan67
has also described the breakdown of zirconolite to a
new mineral assemblage consisting of zircon, titanite,
and rutile in metamorphosed ferromagnesian silicate
rocks at Manitouwadge, Canada. This reaction can be
expressed as follows (modified slightly from Pan67):
CaZrTi2 O7 þ 2SiO2 ¼ ZrSiO4 þ CaTiSiO5 þ TiO2
½IIŠ
This reaction illustrates the potential instability of zirconolite at high temperature and pressure in silica-rich

573

systems; however, the phase relations of zirconolite and
other minerals in the system CaO–SiO2–TiO2–ZrO2–
H2O–CO2 remain elusive, especially at low temperature and pressure.
Malmstro¨m68 investigated the performance of
several zirconolite compositions under hydrothermal
conditions (150–700  C, 50–200 MPa) in fluids containing different concentrations of HCl, NaOH, H3PO4,
silicate, or carbonate, in addition to pure water. Starting
materials consisted of near end-member CaZrTi2O7,
together with single-phase samples doped with Nd, Al,
U, Ce, Gd, and Hf. The results of these experiments
demonstrate that zirconolite is most highly reactive in
the NaOH-bearing fluids, but temperatures in excess of
500  C are required to produce a continuous alteration
layer consisting of perovskite þ calzirtite at 50 MPa or
perovskite þ baddeleyite at 200 MPa. In the case of
HCl, similar temperatures are required to produce an
alteration layer consisting of rutile and anatase. Somewhat surprisingly, scanning electron microscopy (SEM)

observations revealed that the silicate and carbonate
fluids had no visible effect on the zirconolite surface
after experimental runs at 550  C and 50 MPa. Only
limited reaction was observed in pure water or H3PO4
fluids at the same temperature and pressure, with rutile
and monazite appearing as products on the surface.
Recently Po¨ml,69 experimentally investigated the
hydrothermal alteration of a crystalline 239Pu-doped
and an X-ray amorphous 238Pu-doped (D % 7 Â 1018
a-decay per gram) zirconolite ceramics with the
composition Ca0.87Pu0.13ZrTi1.74Al0.26O7. A disk of
each ceramic sample was treated in a Teflon# vessel
with 2 ml of 1 M HCl at 200  C for 3 days under
autogeneous pressure. The analyses of the experimental fluids by inductively coupled plasma-optical
emission spectrometry (ICP–OES) revealed that significantly higher Ca, Al, and Pu concentrations were
released into solution from the 238Pu-doped than
from the crystalline 239Pu-doped sample. Optical
and SEM investigations of the 239Pu-doped sample
after the experiment revealed no signs of alteration,
while the X-ray amorphous 238Pu-doped sample
showed strong alteration features even under the optical microscope. Parts of the disc were covered by a
‘carpet’ of TiO2 crystals. Energy dispersive X-ray
(EDX) analyses further showed that the uncovered
areas lost Ca, Pu, and Al and have a composition
close to ZrTiO4. Surface areas yielding the original
composition could not be found. Such an observation
indicates a diffusion-controlled leaching process from
the X-ray amorphous 238Pu-doped zirconolite. However, further research is necessary before any

Minerals and Natural Analogues

conclusion about the alteration mechanism can be
made. Regardless of what details constitute the alteration process, the comparative experiments clearly
demonstrated that self-irradiation damage has a
strong effect on the aqueous stability of zirconolite.
In addition, further hydrothermal experiments
were conducted with powders of different Ce-doped
zirconolite ceramics (XCe ¼ 0À0.225 apfu) at temperatures between 100 and 300  C, with different
surface-area-to-fluid volume ratios, and different
solution compositions (1 M HCl, 2 M NaCl, 1 M
NaOH, 35% H2O2, 1 M NH3, pure H2O). Experiments at temperatures >200  C were carried out in
silver and nickel reactors, while for those at lower
temperatures, Teflon vessels were used. The results
of the different experimental series can be summarized
as follows: (1) The alteration rate was insignificant for
all solutions other than the 1 M HCl solution. A 1 M
HCl solution was therefore used for all other experimental series. (2) Rutile (and anatase) and baddeleyite replaced the zirconolite grains to varying
degrees in the 1M HCl solution, that is, zirconolite
dissolution was found to be incongruent. (3) No clear
correlation between the Ce-doping level and the
degree of alteration could be observed. (4) The degree
of alteration increased only slightly with increasing
temperature. (5) The alteration rate was found to be
independent on the surface-to-volume ratio. (6) Ag
dissolved from the silver reactors dramatically increased
the reaction rate, while Ni from the Ni reactors reduced
the solubility of Ti and Zr in the HCl solution, indicating that background cations have a strong effect on the
alteration rate. In summary, considering that only the
1 M HCl solution caused any significant alteration

at temperatures between 100 and 300  C, crystalline
zirconolite proved to be extremely stable in aqueous
solutions.
5.22.3.3

Brannerite

The crystal structure of brannerite, ideally UTi2O6,
is based on a distorted array of hexagonal closepacked oxygens. The structure is monoclinic, space
group C2/m, and consists of layers of Ti octahedra
connected by columns of U octahedra.70,71 Natural
and synthetic brannerites can incorporate substantial
amounts of Ca, Ln, Th, and other elements. In both
cases, the incorporation of lower valence elements on
the A-site may be charge-balanced by partial oxidation of U4+ to U5+ and/or U6+ ions.72,73 Lumpkin
et al.74 have examined a small suite of brannerites
from different localities by SEM–EDX, showing

Calibration
B4 brannerite
B10 brannerite
B12 brannerite

0.7

M5/(M4 + M5)

574

0.68

0.66

0.64

4

4.5

5

5.5

6

Average U valence
Figure 6 This diagram shows the electron energy loss
spectroscopy branching ratios M5/(M4 + M5) obtained from
the U M4,5 spectra of three natural brannerite samples.
Calibration data indicate average valence states between
about 4.3 and 4.8, for example, significant amounts of
U5+ and/or U6+ are present.

that unaltered areas of the samples contain up to
7 wt% CaO, 8 wt% Ln2O3, 7 wt% PbO, and 15 wt%
ThO2, together with minor amounts of Al, Si, Mn, Fe,
Ni, and Nb. Using electron energy loss spectroscopy,73 measured the U–M4,5 spectra of three of
these samples and reported average U valence states
of 4.4–4.8 based on the individual M5/(M4 þ M5)
branching ratios (Figure 6). Amorphous brannerite

may contain nanocrystal inclusions of uraninite. All
natural brannerite samples with ages greater than
$20 Ma appear to be amorphous due to a-decay
damage. Electron diffraction patterns of relatively
unaltered areas of the brannerite samples typically
consist of two broad, diffuse rings characteristic of
amorphous materials. Unaltered natural brannerites
have average a-decay doses of (2–170) Â 1016 a per
milligram in part due to the very high U content.74
A partially crystalline brannerite from Binntal, Switzerland,75 allows the critical dose to be estimated
at $1 Â 1016 a per milligram, a value similar to geologically young pyrochlore and zirconolite samples.
Geochemical alteration of brannerite is common
and appears to increase in severity with geological
age, although the P–T–X conditions are poorly
understood.76 Alteration usually occurs around the
rim of the brannerite or along cracks and may involve
the formation secondary such as anatase, galena, and
thorite. As reported for pyrochlore and zirconolite,


Minerals and Natural Analogues

the galena may precipitate due to the combined
effects of radiogenic Pb migration and the presence
of S species in the aqueous fluid. Altered brannerite
typically loses U, and the concentration may fall
to $1 wt% UO2 in the most heavily altered areas.
The loss of U may be compensated in part by the
incorporation of up to 18 wt% SiO2 and 16 wt% FeO,
together with significant amounts of P2O5, As2O5,

and Al2O3 in certain examples. In some cases,
the associated rock or mineral matrix surrounding
the brannerite may be highly fractured, providing
pathways for the migration of U, evidenced by the
precipitation of secondary U minerals.74
Several important experimental studies of synthetic and natural brannerite have been conducted
and are summarized as follows: For temperatures of
20–50  C and pH values in the range of 2–12, Zhang
et al.30 reported limiting rate constants for U equivalent to dissolution rates of about 10À3 to 10À5 g mÀ2
dayÀ1. The lowest dissolution rate was obtained at
20  C and pH ¼ 5.6. However, the pH dependence
was examined in detail at 70  C, and the results show
a shallow v-shaped pattern similar to that of pyrochlore and zirconolite, although the release rates of
U from these two phases are 1–2 orders of magnitude
less than that of brannerite.
In further detailed work, Zhang et al.77 demonstrated that the U release rate is strongly dependent
on bicarbonate concentration and that bicarbonate
does not interact strongly with titanium, either on
the solid surface or in solution. The dissolution of
brannerite is incongruent and a preferential release
100

U/ Ti atomic

pH = 2
pH = 11

10

1

0.1

5

10

15
20
Time (days)

25

Figure 7 Experimental aqueous dissolution of synthetic
brannerite showing how the U/Ti atomic ratio in solution
changes with time in acidic (pH ¼ 2) and alkaline (pH ¼ 11)
solutions. The experiment at pH ¼ 2 mimics the U loss
observed in natural brannerite samples.

575

of uranium occurs at pH ¼ 2 (Figure 7). TEM examination after the experiment identified a relatively
small amount of secondary TiO2 (anatase Æ brookite)
containing variable amounts of U and trace amounts
of other elements. Overall, the dissolution of the
brannerite at pH ¼ 11 is nearly congruent, and
TEM examination after the experiment shows that
the sample develops large areas of an amorphous
secondary phase. X-ray photoelectron spectroscopy
(XPS) analyses indicate the existence of oxidized U5+

and U6+ species on specimens both before and after
leaching, and U6+ was the dominant component on
the specimen leached in the pH 11 solution.
Zhang et al.78 also conducted thermal annealing
studies of amorphous brannerite in Ar at 500, 700,
900, and 1100  C followed by batch dissolution
experiments in solution at pH ¼ 4 at 30  C. XRD
analysis indicates that the major recrystallization of
the sample occurs at 900–1100  C, but detailed TEM
examination revealed that partial recrystallization
started at 700  C. Analysis of the starting material
showed that it contained small amounts of Al and Si
due to alteration and nanocrystals of UO2+x within the
amorphous matrix. Interestingly, the authors found
that the dissolution rate increases with annealing temperature, and this is attributed to the growth of UO2+x
and to the formation of an aluminosilicate glass phase
at the highest temperature. Relative to synthetic
brannerite,30 the dissolution rate of the amorphous
brannerite is about one order of magnitude higher.
5.22.3.4

Perovskite

Perovskite is an ABX3 structure type based around
a framework of corner sharing, octahedral B-site
cations with B ¼ Ti, Fe, and Nb, together with
minor amounts of Mg, Al, Zr, and Ta, depending on
bulk rock chemistry and mineralogy. The large A-site
cations occupy the center of a large cavity formed by
8 B-site octahedra and are coordinated to 12 oxygens

in the ideal cubic structure. Most perovskites are
distorted via octahedral tilting and generally have
lower symmetry. In nature, only near end-member
SrTiO3 is cubic, most other compositions are orthorhombic.79 The major A-site cations in natural perovskites are Na, Ca, Sr, and REE, with minor
amounts of K, Ba, and U. The concentration of Th
is usually low, but may reach levels as high as 18 wt%
ThO2 in certain alkaline rocks.80,81
Only a limited amount of information is available
on radiation damage in natural perovskite. In one
study, TEM methods were used to examine the


Minerals and Natural Analogues

microstructure and electron scattering properties of a
large suite of samples having a range of geological
ages.55,56 Perovskite grains containing 1.4–3.1 wt%
ThO2 indicate that the beginning of the transformation
is at a dose of 0.3–0.6 Â 1016 a per milligram for
t ¼ 295–520 Ma. A preliminary analysis of the data
using eqn [2] gave D0 ¼ 0.19 Â 1016 a per milligram
and K ¼ 2.3 Â 10À9 yearÀ1. Perovskite from Bratthagen,
Norway, contains up to 6.0 wt% ThO2, equivalent to
a-decay doses of (0.8–1.2) Â 1016 a per milligram.
These grains are partially damaged based on the
appearance of a weak diffuse ring in electron diffraction patterns. During this study, it was discovered that
the perovskite crystals contain a hydrothermal alteration phase known as lucasite. Although lucasite has a
slightly higher Th content relative to the associated
perovskite, it is completely electron diffraction amorphous at a dose of (1.5–1.7) Â 1016 a per milligram,
suggesting a lower radiation tolerance than the host

perovskite. Th-rich perovskites from the Khibina
alkaline complex, Russia, have been studied by electron probe microanalysis (EPMA) and XRD.80,81 The
crystals are zoned and contain 2.3–18.5 wt% ThO2,
increasing from core to rim. XRD results indicate that
the cores (2.3–7.4 wt% ThO2) retain some crystallinity, but the rims (8.7–18.5 wt% ThO2) are completely
amorphous. These data are consistent with a critical
dose of $2 Â 1016 a per milligram based on the maximum and minimum Th contents of the cores and rims,
respectively.
There have also been very few laboratory studies
of radiation damage in perovskite by doping with
short-lived actinides. Mosley82 studied CmAlO3, in
which about 95% of the actinide content is 244Cm,
and determined that the material is amorphous by
XRD after 8 days of the experiment, from which we
estimate a dose of $0.15–0.2 Â 1016 a per milligram
(equivalent to $0.2–0.3 dpa). Although the critical
dose may be slightly higher than the X-ray value,
it is still substantially lower than the value estimated
for natural perovskite. Using Mosley’s lattice parameter versus time data, we have plotted the unit cell
volume expansion, DVc/Vc0, as a function of the
estimated dose (Figure 8). This plot indicates that
the lattice volume expansion may be as high as 8.8%
at saturation based on the curve fit; however, the last
data point indicates a lattice volume expansion
of $6% prior to amorphization. The latter value is
close to the total volume swelling observed in pyrochlore and zirconolite doped with 238Pu or 244Cm.
Perovskite commonly releases Ca in aqueous
fluids even at low temperature, breaking down to

0.06

0.05
DVc / V0

576

0.04
0.03
0.02
0.01
0.05
Dose (10

0.1
16

a per milligram)

Figure 8 Unit cell volume expansion of CmAlO3
plotted as a function of a-decay dose. The total
expansion at saturation is 8.8%; however, the highest
measured expansion is $6.0%. These data suggest
that the total macroscopic swelling could be in excess of
10 vol.%.

one or more polymorphs of TiO2 (generally anatase Æ
brookite or TiO2–B). This is well illustrated by the
alteration of perovskite to anatase, cerianite, monazite, and crandallite group minerals during severe
weathering of carbonatites in Brazil.83 Using electron
microscopy, Banfield and Veblen84 have proposed
that the perovskite–anatase reaction mechanism

involves topotactic inheritance of layers of the perovskite Ti–O framework. In hydrothermal systems,
Mitchell and Chakhmouradian80,81 have described
the alteration of perovskite to kassite, anatase, titanite,
calcite, and ilmenite in the presence of a CO2- and
SiO2-rich fluid phase at temperatures of 250–600  C
in alkaline rocks. These observations are consistent
with the thermodynamic properties of perovskite and
related minerals. Nesbitt et al.85 have shown that
perovskite is unstable with respect to titanite, rutile,
calcite, and quartz in many hydrothermal fluids and
groundwaters at 25–300  C.
Lumpkin et al.55,56 and Chakhmouradian et al.86
described the hydrothermal alteration of Na-bearing
perovskite (ideally Na0.5Ln0.5TiO3) in alkaline igneous rocks. The primary result of this alteration is
removal of Na from the original perovskite, producing lucasite, LnTi2O6Àx(OH,F)xÁH2O. Lucasite is isostructural with kassite, CaTi2O4(OH)2ÁH2O, another
alteration product of perovskite (CaTiO3). The simplified reaction relationships for perovskite, kassite,
and lucasite are shown subsequently:
2Hþ þ H2 O þ 2CaTiO3
¼ CaTi2 O4 ðOHÞ2 Á H2 O þ Ca2þ

½IIIŠ


Minerals and Natural Analogues

Hþ þ H2 O þ 2Na0:5 Ln0:5 TiO3
¼ LnTi2 O5 ðOHÞÁ H2 O þ Naþ

½IVŠ

In both cases, the replacement product is usually
reported as having a distinctly fibrous or prismatic
morphology, a feature that can be observed by SEM
(Figure 9). In the following paragraphs, we will discuss
some of the important experimental data on perovskite
dissolution.
Thermodynamic calculations and data for natural
groundwaters and hydrothermal fluids (up to 300  C)
revealed that perovskite is generally unstable with
respect to titanite, titanite þ quartz, rutile, or rutile
þ calcite.85 Measurements of the dissolution rates of
two natural perovskites and synthetic SrTiO3 and
BaTiO3 samples were obtained in pure water at
25–300  C, indicating that elemental release rates
are $10À1 to 10À3 g mÀ2 dayÀ1 for Ca, Sr, and for
Ba. Kamizono et al.87 have also examined Ce, Nd, and
Sr-doped CaZrO3 in acidic (HCl, pH ¼ 1) and near
neutral (deionized water, pH ¼ 5.6) solutions at
90  C. In the acidic solution, the dissolution rates of
the impurity elements were near 0.1 g mÀ2 dayÀ1;
whereas Ca and Zr were released at rates on the
order of 10À3 and 10À3 g mÀ2 dayÀ1, respectively.
Leach rates were about two orders of magnitude
lower in the experiment using deionized water.
6729

577

Surface analytical studies of synthetic perovskite
after leaching at 150–250  C in silica-saturated aqueous fluids were reported by Myhra et al.88 These

authors, using a combination of Auger electron spectroscopy (AES), SEM–EDX, and XPS techniques,
identified the presence of a surface reaction layer
ranging in thickness from a few monolayers to several
hundred nanometers, depending on the leaching conditions. The alteration layer was composed mainly of
crystalline TiO2, a thin siliceous layer, and possible
calcium carbonate or hydroxide species. Following
this work, Myhra et al.89 reported additional results
for CaTiO3 and BaTiO3 leached at 300  C in pure
water. They determined release rates after 14 days of
1.4 Â 10À2 g mÀ2 dayÀ1 for Ca and 3.3 Â 10À2 g mÀ2
dayÀ1 for Ba from the two samples, respectively.
Surface analytical work showed that Ca and Ba
were depleted to a depth of about 200 nm and that
oxygen was enriched near the surface, consistent with
the release of Ca and Ba to solution and the formation of a crystalline TiO2 layer.
The studies noted earlier were extended to lower
temperatures of 20–100  C by Pham and coworkers,90,91 who showed that near surface decreases in
the Ca/Ti ratio determined by XPS are accompanied
by the formation of an amorphous Ti-rich layer up to
10-nm thick, as observed by TEM. The authors proposed a base catalyzed hydrolysis and ion exchange
model to account for their observations, whereby
surface Ca2+ is released to solution via exchange
with H+ and Ti–O–Ti surface species are converted
to Ti–OH species via reaction with OHÀ and H2O.
The overall reaction can be written as follows90:
CaTiO3 þ ð6 À xÞHþ ¼ Ca2þ þ TiðOHÞxð4ÀxÞþ
þ ð3 À xÞH2 O

50 mm
Figure 9 Example of hydrothermal alteration of natural

perovksite in alkaline rocks from Bratthagen, Norway. This
backscattered electron image shows partial replacement of
a large perovskite (e.g., Na-rich variety known as loparite)
single crystal by a brighter, fibrous secondary phase,
lucasite.

½VŠ

This reaction indicates that the aqueous Ca2+/H+
ratio and the presence of Ti–OH surface species
control the dissolution of perovskite. The presence
and buildup of the amorphous Ti–OH surface film at
low temperatures may be due to kinetic factors, as
crystalline anatase or rutile are thermodynamically
favored at low temperatures.90
McGlinn et al.92 have investigated the pH dependence
of the release of Ca from two perovskite samples: endmember CaTiO3 and Ca0.78Sr0.04Nd0.18Ti0.82Al0.18O3.
Results of this study, performed at 90  C with pH
ranging from 2.1 to 12.9, demonstrated that the
Ca release rates generally decrease with increasing
pH. After 43 days of leaching, the Ca release rate
for the end-member perovskite decreased from


578

Minerals and Natural Analogues

8.9 Â 10À2 g mÀ2 dayÀ1 at pH ¼ 2.1 to 2.2 Â 10À3 g
mÀ2 dayÀ1 at pH ¼ 12.9. Similar results were obtained

for the doped perovskite. SEM imaging clearly showed
the development of ‘agglomerated, submicron, titanaceous particles’ on the perovskite surface after the dissolution experiments performed in acidic aqueous
solutions.
In a more recent study by Zhang et al.,93 thermally
annealed perovskite (CaTiO3) surfaces were characterized by SEM, TEM, XPS, and atomic force
microscopy (AFM) techniques before and after aqueous dissolution testing in deionized water at room
temperature, 90  C, and 150  C. Results of this work
demonstrated that, although mechanical damage
caused higher Ca release initially, it did not affect
the long-term Ca dissolution rate. However, the
removal of surface damage by annealing did lead to
the subsequent spatial ordering of the alteration
product, which was identified as anatase (TiO2) by
both X-ray and electron diffraction, on CaTiO3 surfaces after dissolution testing at 150  C. The effect of
Ca2+ in the leachant on the dissolution reaction of
perovskite at 150  C was also investigated, and the
results suggest that under repository conditions, the
release of Ca from perovskite is likely to be significantly slower if Ca2+ is present in groundwater.
5.22.3.5

Baddeleyite

The monoclinic form of ZrO2, baddeleyite is the only
natural analog for the proposed cubic zirconia waste
forms. A recent literature review by Lumpkin94
shows that, even though baddeleyite is rather widespread as a trace phase in natural systems, it has a
limited composition range of 87–99 wt% ZrO2, with
most of the remainder comprised of FeO, TiO2, and
HfO2. Natural baddeleyite has a distorted fluorite
structure due to the low concentrations of other

large cations such as Ca and REEs. The total concentration of Th and U is usually <2000 ppm by weight;
nevertheless, the cumulative a-decay dose is in the
range of (0.1–1.1) Â 1016 a per milligram for samples
with ages of 1100–1200 Ma. Natural samples of baddeleyite appear to remain crystalline even at the
highest observed a-decay dose levels. A case study
of baddeleyite from the Jacupiranga carbonatite complex of southern Brazil shows that the mineral can
incorporate up to 4.1 wt% Nb2O5 and 1.2 wt%
Ta2O5.94 Incorporation of Nb5+ and Ta5+ is partially
compensated by the incorporation of up to 0.4 wt%
MgO and 0.3 wt% FeO in a charge-balanced substitution of the form 3Zr ¼ 2(Nb, Ta) þ (Mg, Fe).

These samples are also resistant to hydrothermal
alteration which affected associated pyrochlore crystals, even though the baddeleyite crystals received
a-particle doses as high as (3–4.5) Â 1016 a per milligram along common grain boundaries with a U-rich
pyrochlore phase.94 These observations are consistent with results from hydrothermal experiments on
zirconolite ceramics, wherein zirconolite is often
replaced by baddeleyite.68,69
5.22.3.6

Crichtonite

Minerals of the crichtonite group conform to the general formula A1ÀxM21O38 and are generally found in
mafic–ultramafic and granitic igneous rocks. In these
minerals, A ¼ Na, Ca, Sr, Pb, and the larger Ln elements (e.g., La, Ce, Pr, and Nd) and M ¼ Ti with
variable amounts of Mg, Al, V, Cr, Mn, and Zr. They
are of some interest here due to the ability to incorporate substantial amounts of U in the mineral davidite.
Based on a review of the literature, Gong et al.95 proposed that this structure type may be an effective host
phase for a range of fission products and actinides in
high-level nuclear waste. One member of the group,
loveringite, commonly occurs in titanate-based nuclear

waste forms as a minor phase. Very little is known about
the radiation effects of this mineral; however, we report
here some previously unpublished data from our laboratory for samples (five davidite, one crichtonite).
SEM–EDX analyses demonstrate that the chemical composition varies considerably from sample to
sample, but individual samples are relatively uniform
in composition with only limited evidence for zoning.
The U content ranges from $0.2 to 9.5 wt% UO2
(0.02–0.65 apfu). The Th content is much lower, ranging from <0.1 to 1.3 wt% ThO2 (<0.01–0.09 apfu).
Maximum amounts of other notable cations include
3.7 wt% V2O3, 4.1 wt% Cr2O3, 2.5 wt% Y2O3, 5.6 wt%
La2O3, 6.0 wt% Ce2O3, 4.0 wt% MnO, 2.4 wt% ZnO,
2.7 wt% SrO, and 4.9 wt% PbO. Estimates of the
geological age are available for these samples, giving
a dose range of (0.3–42) Â 1016 a per milligram using
eqn [1]. At present, the critical amorphization dose
is poorly constrained, but is probably somewhere
between 0.5 Â 1016 and 2 Â 1016 a per milligram
based on our TEM observations. This is consistent
with recent data for a 270 Ma davidite sample from
Bektau-Ata, Kazakhstan, which is heavily damaged at a
dose of 1.5 Â 1016 a per milligram in a recent study by
Malczewski et al.96 Two of our samples showed evidence for alteration in the form of replacement by an
assemblage of rutile þ ilmenite or rutile þ titanite.


Minerals and Natural Analogues

579

5.22.3.7 ABO4 and AB2O6 Minerals

(B ¼ Nb, Ta, and Ti)
A number of other oxide minerals are known to contain substantial amounts of Th and U and are interesting from the point of view of the effect of structure
type and composition on response to a-decay damage.
Minerals in this category are principally fergusonite,
aeschynite, euxenite, samarskite, and to a lesser
extent, columbite–tantalite. Fergusonite, ideally
YNbO4, may contain substantial amounts of heavy
Ln, Th, and U on the Y-site together with Ti and Ta
on the Nb site.97 This rare element oxide mineral
is commonly reported to be amorphous as a result of
a-decay damage. Giere´ et al.97 have given a detailed
description of the chemistry and radiation damage of
fergusonite occurring in a 40 Ma granitic pegmatite
from Adamello, Italy. These authors used a combination of quantitative EPMA, TEM–EDX, and microRaman analyses to provide an upper limit on the
critical amorphization dose of $1 Â 1016 a per milligram for a sample characterized by TEM–EDX. This
is consistent with previous work on amorphous fergusonite from Norway, containing 4.9 wt% UO2,98
and having an estimated a-decay dose of $4 Â 1016
a per milligram based on an assumed age of 320 Ma.
By way of comparison, amorphous fergusonite from
the Rutherford pegmatite, Amelia, Virginia, has a
well-defined age of 289 Ma, and contains 2.2–4.7 wt%
ThO2 and 1.5–7.4 wt% UO2, giving a dose range of
(2–7) Â 1016 a per milligram.99 Fergusonite from this
locality is commonly altered along grain boundaries
and cracks, accompanied by loss of most of the Y and
Ln elements and uptake of Ta, Ca, Fe, and some Th
(Figure 10).
The orthorhombic AB2O6 oxides aeschynite,
euxenite, and polycrase (with idealized compositions
CeNbTiO6 or YNbTiO6) typically occur in granitic

pegmatites where they incorporate Y and a range
Ln series elements, together with Ta, Th, and U,
and small amounts of Ca and Fe. Ewing100 investigated a suite of AB2O6 oxide minerals and reported
all of them to be amorphous or ‘metamict’ due to
a-decay damage. Based on Ewing’s carefully measured compositions, we have estimated that these
samples received a-decay doses of (2–21) Â 1016 a
per milligram wherein the lowest dose provides a
good upper limit for amorphization of samples with
ages of $250 Ma. According to Ewing, the AB2O6
oxides are commonly altered by hydrothermal fluids
whereupon they show a consistent increase in the Ca
content together with OHÀ and H2O, generally at the

Figure 10 Backscattered electron image showing
hydrothermal alteration of natural fergusonite (left) and
monazite (right) from the Rutherford #2 granitic pegmatite,
Amelia County, Virginia. Fergusonite is heavily altered to
calciotantite + fersmite, thereby losing Y and heavy
lanthanides to the fluid phase. The adjacent monazite
crystal has also been partially altered during this process,
but in this case, the alteration involves minor chemical
exchange with the fluid phase. Width of image ¼ 0.2 mm.

expense of Y, Ln, Th, and U. At lower temperatures,
for example, during weathering processes, A-site
cations tend to be globally depleted. Due to the
complications imposed by alteration, radiation damage, and potential phase transformations between
AB2O6 minerals, thermal annealing is generally complex. The observations of Ewing and Ehlmann101
show that, in the simplest case, the aeschynite structure type is the first phase to form beginning
at $400  C, followed by transformation to the higher

temperature euxenite form at 700–750  C. However,
pyrochlore and rutile are commonly formed in many
samples. Lumpkin et al.23,24 examined a small group
of amorphous Nb–Ta–Ti oxide minerals with stoichiometries of ABO4 and AB2O6 and reported
recrystallization energies in the range of 40–85 J gÀ1.
Rare examples of columbite–tantalite, ideally
(Mn,Fe)(Nb,Ta)2O6, were reported to contain enough
U to induce amorphization (see Lumpkin102 and
references to previous work). These $1800 Ma samples are zoned in both the TiO2 (2.2–4.8 wt%) and
UO2 (0.2–2.6 wt%) contents and exhibit a critical
amorphization dose of 8 Â 1016 a per milligram for
specimens examined by electron microscopy. This
unusually high dose was attributed to long-term
annealing of a-recoil collision cascades back to the
original structure. We have recently found zoned
columbite–tantalite crystals from the High Peak mine,


580

Minerals and Natural Analogues

in the Elk Mountain district of northern New Mexico.
These zoned crystals have U-rich cores with up to
2.5 wt% UO2 and U-poor rims. The rims, however,
contain inclusions of amorphous AB2O6 oxides resulting in cracking due to differential swelling (Figure 11).
This example provides dramatic evidence for mechanical failure and subsequent geochemical alteration in
course-grained materials with inclusions rich in
U and Th. Columbite–tantalite is commonly altered
to a secondary assemblage consisting of pyrochlore

(NaCaTa2O6F), fersmite (CaNb2O6), or calciotantite
(CaTa4O11) during postmagmatic hydrothermal
activity, depending upon the activities of Naþ, Ca2þ,
Fe2þ, Hþ, and HF in the fluid medium (see Lumpkin
and Ewing38 and references therein).
EXAFS and XANES studies generally indicate
that the fully amorphous structures of these radiation-damaged minerals do not possess atomic periodicity beyond the second coordination environment
(M–M distances), and in some cases, there is evidence for ‘disruption’ of the second coordination
sphere. The first coordination sphere (M–O distances) remains intact albeit with minor changes in
the bond lengths and degree of distortion of the M–O
polyhedra. In the case of fergusonite, the Nb–O bond
lengths of two longer bonds in the crystalline phase
appear to decrease slightly in the amorphous material, consistent with a reduced preedge feature in the
Nb–K edge.103 Thus, the distorted 4 þ 2 coordination
geometry of fergusonite is somewhat ‘homogenized’

Figure 11 Backscattered electron image showing
cracking and geochemical alteration of U-rich amorphous
AB2O6 mineral inclusions in U-poor columbite–tantalite from
the High Peak pegmatite, Elk Mountain, New Mexico. The
cracking is clearly induced by differential swelling of the
inclusions and the host phase, allowing aqueous fluid to
migrate through the cracks. Width of image ¼ 0.5 mm.

by a-decay damage. Apart from the loss of and considerable disruption of the second coordination
sphere, Nakai et al.103 found little difference in the
mean bond length and distortion of the Nb site in
euxenite. This result differs somewhat from the analysis of the Ti-site geometry in amorphous and annealed
aeschynite and euxenite presented by Greegor et al.104
Although the disruption of the second coordination

sphere is a common feature of both studies, it was
found that amorphization produced a slight reduction
in both the mean Ti–O bond length and coordination
number of these minerals, primarily due to displacement of the longer Ti–O bonds.104
5.22.3.8

Hollandite

Although they do not contain actinides in natural or
synthetic systems, minerals of the hollandite group
are extremely important for the encapsulation of the
relatively short-lived heat-generating radionuclides
90
Sr and 137Cs and long-lived 135Cs. The crystal
structure of hollandite, A1.1–1.7B8O16, is similar to
that of rutile and consists of edge-sharing chains of
octahedra connected via corner sharing to form a
three-dimensional framework. Hollandite, however,
has two octahedral chains connected by edge sharing
instead of the single chain found in rutile, resulting in
a rather large 2 Â 2 tunnel capable of accommodating
large A-site cations like K, Rb, Cs, and Ba.105–107
These cations exhibit various ordering sequences over
the available tunnel sites, commonly resulting in superlattice peaks in X-ray or electron diffraction patterns.
The space group is typically I4/m or C2/m depending
upon the A/B cation radius ratio. Numerous synthetic
samples have been produced with Ti, Mn, Mo, noble
metals, or Sn as the most common major elements and
with Mg, Al, V, Fe, Co, Ni, Zn, and Sb, among others, as
minor elements on the B-site. The composition of

hollandite in titanate-based waste forms is generally
4þ
given as ðBax Csy Þ½ðTi; AlÞ3þ
2xþy Ti8À2xÀy ŠO16 in which
charge compensation for Ba and Cs is usually provided
by Al or Ti3þ.108
In natural samples, typical B-site cations are Ti, V,
Cr, Fe, Mn2þ, and especially tetravalent Mn4þ. It is of
interest to mention here one geological occurrence in
particular, the Mn-rich metamorphic rocks Le Coreaux, Belgium, where deep purple and violet metasedimentary layers are host to quartz veins that
contain Ba–Sr hollandite and other Mn minerals.109
These stable hollandites formed at a pressure of
1–2 kbar and temperature below 360–380  C, they


Minerals and Natural Analogues

contain 19–8 wt% BaO and 0.3–8 wt% SrO, and Sr is
the dominant cation in the tunnels at concentrations
above $6 wt% SrO. This suggests that hollandite
may also be considered as a potential host phase for
radioactive Sr in nuclear waste forms. The closest
natural analog for synthetic titanium hollandite is
the mineral priderite that occurs in Western Australia
and elsewhere and has a composition of approximately (Ba,K)1.2–1.6[Ti,Fe,Mg]8O16.110,111
Very little information exists on the geochemical
behavior of hollandite in natural systems, therefore,
in this section, we will briefly outline some of the
relevant experimental studies. Pham et al.112 carried
out experimental work on synthetic Ba-hollandite

doped with Cs and containing Al on the B-site for
charge balance. These authors suggested that, following the initial release of Cs and Ba from reactive
surface sites, the first few monolayers of the structure
rapidly dissolved due to the release of Al and consequent precipitation of Al–OH species, driving solution pH to lower values. However, the alteration
process was mediated via the formation of a continuous Al- and Ti-rich surface layer. Further evidence
for selective removal of Ba and enrichment of Al and
Ti on the surface of hollandite leached at 250–300  C
was presented by Myhra et al.108 These conclusions
were largely based on the incongruent release of Ba
(0.113 g mÀ2 dayÀ1), Al (6.6 Â 10À3 g mÀ2 dayÀ1), and
Ti (<8 Â 10À4 g mÀ2 dayÀ1) after 14 days of leaching,
combined with XPS analyses of the altered surfaces.
In a study combining dissolution experiments and
detailed characterization by electron microscopy,
Carter et al.7 demonstrated that the release rates of
Ba and Cs from hollandite are nearly identical,
whereas Al and Ti are below the detection limits at
90  C and only Al was detected at 150  C (however,
only by a factor of 2–3 above the detection limit).
This is similar to previous observations; however,
SEM work revealed the presence of nodular secondary phases on the surface of the hollandite at both
temperatures. This was confirmed by TEM, which
identified both Ti-rich and Al-rich nodules in a ratio
of about 10 to 1, respectively. Furthermore, XPS
analysis of the hollandite surfaces after the dissolution experiments indicated the presence of an Al-rich
layer but only for the samples used in the experiments at 150  C. These authors also examined the pH
dependence of the elemental release rates at 90  C,
finding that the release of Ba decreases linearly from
about 2 Â 10À3 g mÀ2 dayÀ1 at pH ¼ 2.5 down to
4 Â 10À4 g mÀ2 dayÀ1 at pH ¼ 12.9.

581

5.22.4 Silicates
5.22.4.1

Zircon

Zircon is classified as an ABO4 type orthosilicate (space
group I41/amd) due to the presence of isolated SiO4
tetrahedra, which constitute the B-site, but the structure actually consists of a framework of edge-sharing
silicate tetrahedra and eight coordinated A-sites.113
In nature, the composition of zircon often approaches
ideal ZrSiO4 and is a common accessory mineral
found in a variety of geological environments. Natural
zircon may contain trace amounts of Ca, Ln, Hf, Th,
and U on the A-site and P on the B-site. Some or all of
these elements may be enriched significantly in zircon
specimens from highly fractionated granitic rocks and
especially in granitic pegmatites. Maximum concentrations of $26 wt% Ln2O3, 10 wt% ThO2, 10 wt%
UO2, and 5 wt% CaO have been reported in zircon,
but these values are exceptional. Natural zircons may
also contain both OHÀ and H2O species, some of
which may be incorporated as a means of providing
local charge balance at radiation damage sites.114
It is of interest to note here that U-rich zircon,
containing up to 12.9 wt% U ($0.1 U per formula
unit) occurs in the silicate melt that formed during
the accident at the Chernobyl nuclear plant in
1986.115 This melt, often referred to as ‘Chernobyl

lava,’ resulted from partial melting of the nuclear fuel,
related structural materials, and other materials
dropped into the reactor area by helicopter. The
presence of zircon, together with (Zr,U)Ox and UOx
phases and globules of Fe(Cr,Ni) metal, proved to be
important in determining the sequence of events.
Based on a consideration of the phase relations,
Burakov et al.116 suggested that in the initial phase
of the accident, nuclear fuel interacted with Zr metal
as temperatures increased to $2500  C just prior
to the explosion, followed later on by silicate lava
flow and emplacement of the melt below the reactor at temperatures of <1700  C. Geisler et al.117
conducted a detailed electron microprobe and
micro-Raman spectroscopic investigation of the compositionally zoned Chernobyl zircons and reported
that the UO2 content ranges from 0.8 to 15.8 wt%,
equivalent to 0.005–0.115 U atoms per formula unit.
Because the compositions of these remarkable zircons are confined to the system Zr1ÀxUxSiO4, they
provide an important benchmark for the analysis of
observable Raman bands. To summarize briefly, the
frequencies of the SiO4 stretching modes decrease
by 0.67–0.75 cmÀ1 per formula unit of U as a direct


582

Minerals and Natural Analogues

result of increasing Si–O bond length with increasing
U content. Similar results are found for the lattice
modes, but the SiO4 bending modes remain relatively

constant in terms of the Raman frequency shift. Line
broadening is significant for the lattice modes due to
the ionic size difference between Zr4þ (rVIII ¼ 0.072 nm)
and U4þ (rVIII ¼ 1.00 nm) and the resulting microscopic strain fields induced by substitution of Zr by
U. A detailed analysis of the lattice vibrational modes
indicates that the microscopic strain is larger in the
(001) plane than along the c axis, consistent with the
structure of zircon.117
In a pioneering study of natural zircons from
several different localities, Hurley and Fairbairn118
demonstrated that the mineral experiences a transformation from the crystalline to the amorphous state
and they determined that the diffraction angle of the
(112) reflection decreased from 35.635 2y to 35.1
2y up to $0.4 Â 1016 a per milligram. The data were
found to follow an exponential function that related
the ‘fractional disorder’ to the a-activity, the number
of atoms displaced per a-decay event, and an annealing parameter. From the data, Hurley and Fairbairn
determined a value of B ¼ 2.3 Â 10À16 mg per
a-particle and calculated that 4500 atoms were displaced by each a-decay event. Closely following this
work, Holland and Gottfried119 published their classic paper on the density, refractive indices, and unit
cell parameters of zircon as a function of dose. Using
a suite of samples from Sri Lanka, they showed that
the density decreased systematically from <4.70
to $3.96 g cmÀ3 at a dose above 1.2 Â 1016 a per
milligram. These data showed that the density of
zircon decreased by 16%, the largest change of any
potential waste form material. Refractive indices also
decreased systematically and approached a single isotropic value of 1.81 over a similar dose range defined
by the density measurements. XRD work revealed
that the a and c cell parameters both increased rapidly,

but anisotropically, as a function of dose before leveling off at $0.6 Â 1016 a per milligram. Holland and
Gottfried concluded their paper with an analysis of
the fractions of crystalline zircon, an intermediate
phase, and amorphous zircon as a function of dose.
Analysis of the data for the crystalline fraction showed
that B ¼ 3.8 Â 10À16 mg per a-particle.
Murakami et al.120 showed that certain XRD peaks
of zircon samples from Sri Lanka could be separated
into Bragg and diffuse scattering components with
increasing dose. This procedure enabled a more
accurate determination of the anisotropic expansion
with increasing dose, giving expansions of 1.5% along

the a axis, 1.8% along the c axis, and a lattice volume
expansion of 4.7% (total volume expansion is $18%).
Further work, including density measurements and
TEM observations, delineated three stages of damage
in the natural zircon samples. At dose levels below
0.3 Â 1016 a per milligram, the damage is characterized by the accumulation of point defects, unit
cell expansion, and lattice distortion. Within an intermediate dose range of (0.3–0.8) Â 1016 a per milligram, there is a progressive overlap of a-recoil tracks
to produce larger amorphous domains with increasing dose. Above 0.8 Â 1016 a per milligram, the zircon
is completely X-ray and electron diffraction amorphous. Weber et al.121 have also suggested a long-term
annealing rate of $1 Â 10À9 yearÀ1 for zircon from
Sri Lanka, a value that is similar to that reported for
natural zirconolite.
Changes in the mechanical properties of zircon as
a function of dose are extremely important. This is
not surprising in view of the large total volume
expansion and anisotropic unit cell expansion documented earlier. Using a single specimen from Sri
Lanka, Chakoumakos et al.122 provided a dramatic

illustration of the fracture properties of a zoned zircon sample from Sri Lanka. Even though the total
ThO2 þ UO2 concentration only varied by about
0.4–0.5 wt% between the 5–400-mm thick growth
zones, the variation in dose was sufficient to cause
microfracturing of the more brittle, low dose zones.
The fractures were pinned in the high dose zones,
indicating an increase in fracture toughness for
these actinide-enriched layers. Following this work,
Chakoumakos et al.123 revisited the sample and examined in detail the changes in chemistry and mechanical properties using electron microprobe data and
a mechanical properties’ microprobe. Results of this
study demonstrated that the hardness and elastic
modulus of natural zircon decreased by 40% and
25%, respectively, for a-decay doses ranging from
0.3 Â 1016 to 1.0 Â 1016 a per milligram. Zircon samples from Sri Lanka were also used in an important
study of the energetics of radiation damage as a
function of dose.124 In this study, the results of temperature calorimetry revealed that the enthalpy of annealing at room temperature follows a sigmoidal trend
with increasing dose, reaching a saturation value above
0.5 Â 1016 a per milligram with DH ¼ À59 kJ molÀ1,
consistent with structural changes at the subnanometer
scale. The magnitude of this value exceeds DH for
tetragonal ZrO2 þ SiO2 glass by $18–59 kJ molÀ1 and
baddeleyite þ quartz by À33 kJ molÀ1, indicating that
these assemblages may form upon thermal annealing of


Minerals and Natural Analogues

fa ¼ 1 À eÀBD

½4Š

The best fit of eqn [4] to the data gives a value of
B ¼ 2.7 Â 10À16 mg per a-particle, consistent with
amorphous cascades having radii of $2.5 nm. Although
there has been some controversy surrounding
the damage accumulation model, a recent article by
Palenik et al.129 indicates that sensitive measurement
techniques, such as diffuse X-ray scattering, IR spectroscopy, and NMR spectroscopy, are capable of direct
determination of the amorphous fraction and provide
support for damage accumulation via the direct impact
model. Using available data, the components of volume
expansion are plotted in Figure 12, where we see that
expansion of the crystalline fraction (e.g., unit cell)
dominates at lower dose levels, but saturates at about
5 vol.%. Subsequent expansion is dominated by accumulation of amorphous domains.
The structure of the amorphous state in natural
zircon has been examined using EXAFS–XANES by
Farges and Calas130 who found that the average Zr–O
distance $0.1 A˚ less than that of crystalline zircon.
They were also able to determine that the coordination
number of Zr decreases from 8 to 7, indicating that
O atoms are displaced from the ZrO8 coordination
sphere during a-decay damage. Although long-range
periodicity is lost, the Zr–Zr distances are still observed

Total
Unit cell
Difference

15

DV/ V0 (%)

amorphous zircon (see McLaren et al.125 for a detailed
study and implications for age dating by ion microprobe techniques).
Significant progress has been made in the understanding of a-decay damage in zircon through a comparison of natural samples and synthetic specimens
doped with short-lived 238Pu.120,126,127 It was initially
reported that the critical dose for amorphization of
238
Pu-doped zircon is $1.0 Â 1016 a per milligram,
and a multiple cascade overlap model was proposed to explain the accumulation of amorphous
domains.126,127 Detailed analysis of the dose dependence of the crystalline fraction, derived from XRD
analysis of the Bragg peak intensities, gives a value
of B ¼ 5.8 Â 10À16 mg per a-particle for the amount of
material damaged per a-decay event.127 The volume
expansion is $16% (Figure 1), similar to the natural
zircons. Until recently, the fraction of amorphous
material as a function of dose has never been directly
measured. Rı´os et al.128 accomplished this by directly
measuring the amorphous fraction by careful determination of the diffuse scattering component in a suite of
zircon samples from Sri Lanka, wherein the dose
dependence of the amorphous fraction fa is shown to
follow a direct impact model of amorphization:

583

10

5

0.1

0.2
Dose

0.3

0.4

(1016

0.5

0.6

a per milligram)

0.7

Figure 12 Plot showing the components of
volume expansion in zircon doped with 238Pu as a
function of a-decay dose. Unit cell expansion dominates
at low dose, but saturates at about 5 vol.%. At higher
doses, the total macroscopic swelling is dominated
by volume expansion due to amorphous domains in the
material.

in amorphous zircon, but on average they appear to
decrease by $0.3 A˚ relative to crystalline zircon. Interestingly, these authors also found that Hf and Th retain
eightfold coordination in amorphous zircon, but

U appears to exist in a sixfold coordination geometry.
Later work on synthetic zircon samples doped with
trivalent 238Pu and 239Pu (t1/2 ¼ 2.42 Â 104 years)
revealed that the amorphous 238Pu-doped samples
retained a ‘distorted zircon structure and composition’
on the subnanometer scale after 18 years, equivalent to
D ¼ 2.8 Â 1016 a per milligram.131 Thermal annealing
of these samples in air revealed that the zircon structure is restored for T ! 1200  C, but with oxidation of
Pu3þ to Pu4þ, together with minor PuO2 formation.
Below this temperature, the samples recrystallize to a
mixture of ZrO2, SiO2, and PuO2, generally consistent
with the thermodynamic data of Ellsworth et al.124 The
239
Pu-doped samples remained crystalline after a
cumulative dose of 0.012 Â 1016 a per milligram due
to the longer half-life of 239Pu. Thermal annealing
of these samples at 1200  C in air also resulted in
oxidation of Pu3þ to Pu4þ, decreased lattice distortion,
and formation of some PuO2 in the ceramic.
Wayne and Sinha132 were among the first investigators to show that the cracks in zircon, caused by
radiation damage and differential swelling of different compositional zones, serve as pathways for


584

Minerals and Natural Analogues

migration of aqueous fluids. These fluids were able to
penetrate into the zircon crystals, resulting in preferential leaching of radiation-damaged zones at temperatures of 450–500  C during deformation of
the host rocks. Furthermore, Geisler et al.133 have

provided a detailed description of the hydrothermal
alteration of 619 Ma zircon from a posttectonic granite, Eastern Desert, Egypt, at temperatures on the
order of 100–200  C. The zircon crystals exhibit
oscillatory zoning and cracking arising from anisotropic, differential volume expansion. These cracks
provided pathways for fluid migration and chemical
exchange with the solid zircon, resulting in preferential alteration of the U and Th rich zones. Geisler
et al.133 have determined that the more heavily damaged, higher dose zones, are enriched with Al, Ca,
Mn, Fe, light Ln (e.g., La–Nd), and H2O species, and
have lost Zr, Si, and radiogenic Pb.
Due to the importance of zircon in geological age
dating, numerous experiments have been conducted
on both natural and synthetic samples. Pidgeon
et al.134 conducted hydrothermal experiments on
X-ray amorphous natural zircon from Sri Lanka containing 0.6 wt% U. They showed that up to 61% of
the original amount of radiogenic Pb was lost after
treatment with an aqueous solution of 2 M NaCl at
500  C and 100 MPa fluid pressure. Later studies
carried out by Sinha et al.135 show that crystalline
zircon loses 35–51% Pb and 30–47% U when treated
with 2 M NaCl at 600  C and 600 MPa. Under the
same conditions, partially damaged zircon lost $87%
and 52% of the original amounts of Pb and U, respectively. Experiments were also conducted using
the crystalline sample in a fluid with 2% HNO3
at 600 MPa and two different temperatures. In these
experiments, the zircon lost $13–25% Pb and
0.5–4.3% U at 300  C and 30–33% Pb and 5.7–12%
U at 600  C. A study by Rizvanova et al.136 demonstrated that Pb and U are released from amorphous zircon at temperatures as low as 200  C at
100–500 MPa in 2 M Na2CO3; however, crystalline
zircon required higher temperatures of at least
650  C to produce significant levels of Pb and U

loss in this aqueous system. At such a high temperature, crystalline zircon was replaced by baddeleyite in
the silica-undersaturated solution.
Using natural, radiation-damaged zircon samples
from Sri Lanka,133,137–140 conducted a series of hydrothermal experiments at temperatures between 75
and 650  C in 2 M AlCl3, 2 M CaCl2, 0.1 M HCl, 0.1
and 3 M KOH, pure water, and complex aqueous
solutions at pressures between $50 bar and 2.5 kbar.

The authors observed inward penetrating, irregular
alteration fronts, which resemble those found in natural zircon. The experimentally altered areas are
characterized by a lowered backscattered and an
increased cathodoluminescence intensity and sometimes show complex internal nonequilibrium textures. Their thickness was found to be dependent on
temperature as well as on the duration and pH of the
experiments. Nanosized baddeleyite could be identified in some reaction zones by TEM and IR spectroscopy. The experiments also documented the uptake
of hydrogen (mainly as OH) and of cations such as
Ca, Ba, Mg, and Al from the fluid, combined with
release of variable amounts of Zr, Si, Hf, Ln, Pb, Th,
and U from the altered zircon. The loss of trace
elements and the degree of structural recovery
(including nucleation and growth of new zircon
from the amorphous phase and the removal of defects
in the crystalline remnants) was found to be temperature-dependent. At experimental fluid temperatures
between 75 and 200  C, recrystallization of the amorphous phase was not activated and loss of REEs, U,
Th, and radiogenic Pb was severe, while at higher
fluid temperatures, limited loss of trace elements
have been observed. This observation was interpreted
to reflect a competition between the kinetics of longrange diffusion and ion exchange and the kinetics of
the short-range diffusion necessary for the structural
recovery processes, which significantly reduce the
molar volume of the reacted domains. This creates,

on the one hand, stress that is partly released by
fracturing and, on the other hand, a nanoporous
microstructure, as observed by TEM, providing
pathways for fast chemical exchange between the
reaction front and the fluid. Based on these results,
the authors postulated a ‘diffusion-reaction’ model
for the alteration of radiation-damaged zircon
whereby moving recovery/recrystallization fronts
are driven by the diffusion of hydrogen species into
the amorphous zircon structure.
In an effort to determine the effect of radiation
damage on the dissolution of zircon, Ewing et al.141
performed experiments using natural zircon samples
at 87  C in an aqueous solution containing 5 wt%
KHCO3. Results of this study indicate that the dissolution rate increases by nearly two orders of magnitude from 3 Â 10À8 up to 2 Â 10À6 g mÀ2 dayÀ1 for
a-decay doses up to 1.0 Â 1016 a per milligram, for
example, the zircon samples range from highly crystalline to completely amorphous. Helean et al.142
determined the forward dissolution rate of zircon at
120–250  C. Using the elemental release rate of Si as


Minerals and Natural Analogues

a guide, these authors found that the dissolution rate
of zircon increases from 1.7 Â 10À4 g mÀ2 dayÀ1 at
120  C to 4.1 Â 10À4 g mÀ2 dayÀ1 at 250  C.
The impact of self-irradiation damage in zircon on
its aqueous durability was also studied by Geisler
et al.143 These authors performed a hydrothermal
experiment in a 2 M CaCl2 solution at 600  C and

100 MPa with 16 variably radiation-damaged, that is,
amorphized natural zircon samples from Sri Lanka.
They found a dramatic increase in the alteration rate,
monitored by the penetration of Ca and a lowered
backscattered electron intensity from the reacted
areas, at two critical fractions of amorphous domains.
Molecular-dynamics (MD) simulations showed that
the recoil cascades consist of a core depleted in
matter surrounded by a densified and polymerized
boundary.144 Simulations of a second and third recoil
event further revealed that the low-density regions
form percolating regions inside the amorphous cascades.143 The existence of such a nonuniform amorphous structure was confirmed experimentally by
low-angle X-ray scattering experiments. MD simulations of multiple events further reveal that strongly
overlapping cascades produce connected regions of
depleted matter (e.g., low atomic density), which
likely serve as fast-diffusion pathways. The authors
suggested that the first dramatic increase in the alteration rate marks the first percolation point where
the amorphous domains form infinite clusters. However, at this stage, the polymerized boundary regions
still exist as barriers to diffusion and have to be
overcome by hydrolysis and hydration reactions during the alteration process. The second dramatic
increase is interpreted to be a direct consequence of
the formation of interconnected regions of depleted
matter, allowing invasion-like penetration of Ca and
water over macroscopic length scales.
Geisler et al.145 have performed a comparative
experiment with a radiation-damaged natural zircon
from Sri Lanka and a synthetic 238Pu-doped zircon
(4.7 wt% of 238Pu) in an acidic solution at 175  C.
Both zircon samples have suffered a similar degree of
radiation damage, as given by their degree of amorphization. XRD measurements of the experimental

run products revealed that during the hydrothermal
treatment, only the disordered crystalline remnants
recovered in the natural zircon, whereas in the
238
Pu-doped zircon, the amorphous phase strongly
recrystallized. Such a different alteration behavior
of natural and Pu-doped zircon suggests two
fundamentally different alteration mechanisms. The
authors postulated that the alteration of natural

585

radiation-damaged zircon with a low doping level
of U and Th is controlled by diffusion-reaction processes as discussed earlier, but that the high degree of
recrystallization observed in the 238Pu-doped zircon
doped is more compatible with the concept of an
interface-coupled dissolution–reprecipitation process, where congruent dissolution is assumed to be
spatially and temporally coupled to the reprecipitation of new zircon (poorer in Pu) at an inward moving
reaction interface.
5.22.4.2

Thorite

Thorite (ideally ThSiO4) is isostructural with zircon
and contains $17–19 wt% SiO2 and 55–75 wt%
ThO2, depending upon the age and content of radiogenic Pb. In some natural systems, thorite may contain
up to 4 wt% P2O5, 6 wt% CaO, 5 wt% ZrO2, 2 wt%
As2O5, 21 wt% Ln2O3, 25 wt% UO2, and 5–16 wt%
total H2O (see Farges and Calas,130 Foord et al.,146 and
Lumpkin and Chakoumakos147 and references therein).

Thorite samples from the Harding pegmatite, New
Mexico, exhibit extensive solid solution toward the
end-members, Ca0.5Th0.5PO4, Ca0.5Th0.5VO4, and to a
lesser extent, YPO4. Rare, yellow thorites were also
found with a significant Ca0.5U0.5SiO4 component in
which the U is postulated to be hexavalent.147 Geologically young thorite crystals with a well-defined age of
6–7 Ma have been reported to be completely amorphous,146 providing an upper limit on the critical dose
of 0.8 Â 1016 a per milligram, consistent with data for
natural zircon discussed in the previous section. However, Lumpkin and Chakoumakos147 found that the
P and V thorites from the Harding pegmatite retained
substantial crystallinity even after sustained doses of
40–120 Â 1016 a per milligram (Figure 13). Based on
a simple comparison of bond energies, they proposed
that thorites rich in P and V have a lower energy barrier
to recrystallization than samples that are closer to
ThSiO4 in composition. Earlier work has shown that
thorite is subject to extensive alteration in natural systems, generally resulting in hydration, loss of Si, and loss
of radiogenic Pb, often forming secondary galena in the
presence of S-bearing fluids.148,149
5.22.4.3

Titanite (Sphene)

Titanite, ideally CaTi(SiO4)O crystallizing in monoclinic space group C2/c, may incorporate minor
Na, lanthanides, and low levels of actinides on the
Ca site, together with Fe, Al, and Nb on the Ti site.
Additionally, significant amounts of F and OH may


586

Minerals and Natural Analogues

Figure 13 Optical micrograph showing the presence of
crystallinity in P and V-rich thorite from the Harding granitic
pegmatite, Taos County, New Mexico. This image, taken
with crossed polarizers, reveals crystalline domains as
brightly colored spheroidal regions, suggesting
recrystallization of radiation damage thorite over time.
Width of image ¼ 1 mm.

replace O on the anion site that is not bonded to Si.
The amounts of actinides incorporated in the structure are generally below 500 ppm Th and 3000 ppm U.
Due to limited solubility of lanthanides and actinides
in the structure, the titanite has seen limited use
in waste forms apart from the titanite-based glassceramics developed for Canadian waste.150 Heavily
damaged samples studied by Hawthorne et al.151 from
the Cardiff mine, Ontario, Canada, indicate that the
critical dose is somewhat higher than (0.3–0.4) Â 1016
a per milligram if we assume an age of 1000 Ma for
this locality. This is in good agreement with a major
study conducted by Vance and Metson,152 which
showed that the critical dose is $0.5 Â 1016 a per
milligram and that the crystalline–amorphous transformation results in a density decrease of $8%. General features of a-decay damage in titanite with
increasing dose include increasing thermal vibration
parameters of the cations and anions, increasing unit
cell volume up to about 3% in the most heavily
damaged samples, and possible reduction of some
Fe3þ to Fe2þ during electron transfer processes associated with the radioactive decay (see Section 5.22.2).
5.22.4.4

Allanite

Minerals of the epidote group conform to the general
structural formula A2M3(SiO4)(Si2O7)(O,F)(OH)
with A ¼ Ca, Sr, Pb, Mn, Th, Y, Ln, and U, and
M ¼ Al, Fe, Mn, Mg, Cr, and V. In this section,

we are mainly interested in the mineral allanite,
characterized by the presence of light Ln elements
and Th on the A-sites together with Al, Fe3þ, Fe2þ,
and Mg on the M-sites. Allanite typically occurs as an
accessory mineral in felsic igneous rocks, granitic
pegmatites, volcanic rocks, and metamorphic systems, among others. It has assumed some importance
as a tracer of geochemical processes and has proved
to be useful for geological age dating. Only limited
data are available with regard to radiation damage
effects in allanite; however, it is clear that the mineral
becomes optically isotropic due to a-decay damage.
The most consistent data sets also indicate that the
density decreases by 8.5–10.5%, giving a reasonable
indication that the volume expansion is well below
that of zircon.153,154
Janeczek and Eby153 investigated three samples
from different geological localities and provided
a detailed assessment of the composition, microstructure, and annealing behavior. Based on careful
EPMA, XRD, and TEM studies of samples with
reasonably established geological ages, it appears
from this work that the critical amorphization dose is
somewhat >0.5 Â 1016 a per milligram for two samples

from the Appalachian orogen (200–400 Ma). A third
sample from Arizona has an age of 1400 Ma and exhibits a slightly higher level of crystallinity in the bulk
XRD pattern. The critical amorphization dose of this
sample is slightly >5 Â 1016 a per milligram, suggesting that a long-term annealing mechanism may be
operative in allanite. Thermal recovery of allanite
occurs above 500  C and the mineral decomposes
above 850  C. Lattice parameter changes are anisotropic during recovery and the unit cell volume
decreases by $2% at a temperature of 800  C.
Allanite is subject to hydrothermal and low temperature alteration and breakdown to new phase
assemblages. It is commonly replaced at low temperatures by fluorocarbonate minerals (e.g., bastnaesite,
LnCO3F), clay minerals, and thorite. Breakdown during weathering to cerianite (CeO2), monazite, clays,
and goethite has also been reported.154 Wood and
Ricketts155 described in some detail the alteration of
igneous allanite by low temperature (100–200  C)
hydrothermal F and P containing fluids circulating
in the Casto pluton, Idaho. The alteration occurs
along the rims of crystals and along fractures and
is accompanied by some Th enrichment and loss of
Ln elements. The most severe alteration resulted in
breakdown of allanite to fluorite, monazite, and other
secondary phases. Figure 14 shows an example of
hydrothermal alteration of natural allanite from the


Minerals and Natural Analogues

Figure 14 Backscattered electron image at low
magnification showing progressive alteration of allanite from
the Rutherford #2 granitic pegmatite, Amelia County,
Virginia. The lower, lighter gray area is unaltered allanite.

Darker gray zones in the middle of the image represent
allanite–epidote–clinozoisite alteration, for example,
chemically altered allanite possibly of hydrothermal origin.
The heavily altered dark area in the upper part of the image
consists of a complex assemblage of silicate, lanthanide,
and iron minerals probably formed at lower temperature.
Width of image ¼ 1 mm.

Rutherford #2 granitic pegmatite, Amelia County,
Virginia. This image reveals an outer zone of progressive alteration of allanite to an assemblage of late stage
silicate, lanthanide, and iron minerals.

5.22.5 Phosphates
5.22.5.1

Monazite

Like zircon and thorite, monazite also has ABO4
stoichiometry, but the crystal structure is monoclinic
(space group P21/n) and consists of chains of alternating BO4 tetrahedra and AO9 polyhedral sites.156
These chains are cross-linked by edge sharing with
the AO9 polyhedra, effectively closing off open tunnels and creating a structure that is $10% more
dense than the zircon structure type. Orthophosphates with Nd, Pr, Ce, La, Am, or Pu on the
A-site adopt the monazite structure; whereas, those
with the heavier and smaller Ln elements Lu, Yb,
Tm, Er, Ho, and Y adopt the tetragonal xenotime
structure, which is isostructural with zircon. In natural systems, the formula of monazite is generally
given as (Ln,Th,U,Ca)(Si,P)O4, representing a
solid solution of the ideal end-members LnPO4,
Ca0.5Th0.5PO4, Ca0.5U0.5PO4, and ThSiO4. Natural

587

monazite may contain up to 16 wt% UO2 and
52 wt% ThO2,156,157 and has recently been found in
alkaline rocks containing over 8 wt% SrO, indicating
solid solution toward an end-member of the form
Sr0.5Th0.5PO4.42 Therefore, the mineral may be considered as a potential host phase for actinides and a
range of fission products, including Sr. Natural monazite remains crystalline even up to a-decay doses
approaching 7 Â 1016 a per milligram, a feature that
makes synthetic monazite-based materials very
attractive for nuclear waste encapsulation. The radiation stability of synthetic (La,Pu)PO4 and PuPO4
doped with 8.1 and 7.2 wt% 238Pu, respectively, has
been investigated by Burakov et al.158 These authors
discovered that (La,Pu)PO4 remains crystalline up to
a dose of $(0.2–0.3) Â 1016 a per milligram, albeit
with a decrease in the intensity of the measured XRD
peaks. In contrast to this result, the PuPO4 ceramic
sample is heavily damaged at a dose of only
$0.1 Â 1016 a per milligram, and exhibits substantial
volume swelling and cracking.
A number of studies have documented alteration
of monazite during interaction with various hydrothermal fluids.159–164 An important study was conducted by Mathieu et al.162 on natural monazite
occurring in Lower Proterozoic sandstones of the
Franceville basin, Gabon (see Section 5.22.6.3).
The results of this work demonstrate that monazite
was altered to a microcrystalline Th-silicate phase
by interaction with a low temperature (<200  C,
100 MPa) diagenetic brine (NaCl–CaCl2, with Li,
Br, and SO4), resulting in loss of light Ln elements

and U. Several monazite alteration mechanisms
have been identified, including chemical exchange,
dissolution–reprecipitation, dissolution and replacement by a different mineral, and, in rare cases, selective Th removal (see Figure 10). In their detailed
study, Poitrasson et al.160 documented that while the
light Ln are typically released from monazite, U, Y,
and the heavy Ln were retained during hydrothermal
alteration at temperatures of 260–340  C and in fluids
with salinities ranging from 3 to 18 wt% NaCl equivalent. In a study of monazite from the Steenkampskraal mine, South Africa, Read et al.164 showed that
the light Ln elements are retained in altered monazite, the heavy Ln and Y are being released and
precipitated locally as secondary phosphate minerals,
and U is released to the fluid phase and removed
from the system. In general, Th is typically less
mobile than the lanthanide and Y, and is concentrated
into Th-bearing alteration products.165,166 More
recently, Hetherington and Harlov167 demonstrated