Mineralogical Magazine, February 2010, Vol. 74(1), pp.

111–126

Mechanism of formation of atoll garnet

during high-pressure metamorphism
S. W. FARYAD*, H. KLÁPOVÁ AND L. NOSÁL
Institute of Petrology and Structural Geology, Charles University, Albertov 6, 128 43 Praha 2, Czech Republic
[Received 22 August 2009; Accepted 16 February 2010]

ABSTR ACT

Atoll garnet has been found in metabasites and quartz- and mica-rich rocks that have experienced low-
to medium-temperature, high-pressure eclogite facies metamorphism in the Krušné Hory (Erzgebirge).
They occur in several localities but are restricted to thin, texturally distinct zones, even on a thin-
section scale. The mechanism of atoll garnet formation is documented by a series of micrographs and
compositional maps and profiles of atoll garnet in combination with textural relations to other phases in
the rocks. The core of full garnet or its relics in the atoll garnet have larger Ca and Fe, but smaller Mg
contents, compared with the thin rim (ring). In addition to quartz, Na-Ca amphibole and phengite, the
atoll cores are filled by a new garnet that has a composition similar to the outer rim. Formation of the
atoll garnet is interpreted as resulting from fluid infiltration and element exchange between the garnet
core and matrix, a process facilitated by a temperature increase during eclogite facies metamorphism.
In addition to fluid access, the primary textures, mainly grain size, were also effective for the atoll
garnet formation. Small grain fractions with thin rims were easily infiltrated by fluid, which used the
short distance for element exchange between core and matrix. The core garnet was gradually dissolved
and replaced by new garnet having the same crystallographic orientation as the rim or relics in the core.

K EY WORDS : atoll garnet, eclogite, fluid infiltration, Bohemian Massif.

Introduction
CHEMICAL zoning of metamorphic garnet is
regarded as a key to deciphering the P-T evolution
of metamorphic rocks. However, modification of
zoning is a complex process that depends not only
upon change in the P-T conditions but also on
fluid access and deformation. In addition to
volume diffusion, compositional zoning of
garnet, which is at least four orders of magnitude
faster than volume diffusion, can be changed by
diffusion along grain- or sub-grain boundaries and
fractures (Joesten, 1991; Whitney, 1991; Hames
and Menard, 1993; Florence and Spear, 1991). In
addition, the presence of sub-grains increases the
reactive surface area, thus enhancing solution-
precipitation processes.
* E-mail: faryad@natur.cuni.cz
DOI: 10.1180/minmag.2010.074.1.111
One of the much-debated processes in meta-
morphic petrology is the formation of atoll garnet,
which occurs in low- to medium-grade meta-
morphic rocks of differing composition. Selective
replacement of the garnet core along composi-
tional zones is considered a possible mechanism
for the formation of atoll garnet (Green, 1915;
Rast, 1965; Smellie, 1974; Cheng et al., 2007). In
this case, the substitution process would result in
compositional continuity from relict islands inside
the atoll to the atoll ring. The lack of
compositional zones in some atoll garnets has
been the reason for discounting this mechanism as
an explanation for atoll garnet (Atherton and
Edmunds, 1966; Cooper, 1972). Spiess et al.
(2001) and Dobbs et al. (2003) considered
formation of atoll garnet as resulting from
multiple nucleation and coalescence processes
that may envelope other phases. Coalescence
accompanied by resorption inside the garnet
during the early stage of garnet growth was

# 2010 The Mineralogical Society

 S. W. FARYAD ET AL.

assumed to be a possible mechanism for the Samples and methods

production of atoll-like forms by Săbău et al.
(2006). The presence of primary inclusions in
garnet as one of the main reactive sources for atoll
garnet formation is also discussed by Smellie
(1974).
In this paper, we describe atoll garnet occuring
in the low- to medium-temperature eclogite in the
Krušné Hory (Erzgebirge) at the border between
the Czech Republic and Germany (Fig. 1). In
addition to textural and compositional variations
of atoll and ‘whole’ (non-annular) garnets, the
evidence for element exchange between the core
and the rim of garnet is documented by a series of
backscatter electron (BSE) images and composi-
tional profiles. Mechanisms and processes leading
to the formation of atoll garnet in relation to the
rock fabrics and changes in the P-T conditions are
discussed.
The metabasites, mostly exposed on the Czech
side of the Krušné Hory are characterized by the
presence of prograde zoned garnet, omphacite and
minor amounts of paragonite and/or phengite. In
contrast to eclogites with a granulite facies
overprint in the Moldanubian zone (Medaris et
al., 1995; O’Brien and Vrána, 1995; Nakamura et
al., 2004; Faryad et al., 2006; Faryad, 2009), the
Krušné Hory eclogites possess a low-to medium
temperature signature and contain omphacite,
garnet, phengite, paragonite, and talc. They are
usually overprinted by medium- to low-pressure
mineral assemblages comprising sodic-calcic
amphibole, epidote, chlorite, and albite. Detailed
studies of these rocks, exposed near the villages of
Měděnec, Kovářská, and Domašı́n, and Meluzı́na
Hill (Fig. 1b), were made by Klápová (1990) and

FIG. 1. (a) A schematic geological map of the Bohemian Massif (after Franke, 2000). High-grade rock units in the
Moldanubian zone (Gföhl Unit) as well as in the Saxothuringian (Münchberg Massif, Saxonian Granulites) and West
Sudetes (Gory Sowie) regions are shown by dark shading. The white arrow shows the position of the studied area.
(b) (after Konopásek et al., 2001). Black and dark grey areas are eclogite and amphibolite respectively. The
locations of the studied samples near the villages of Kovářská (49a), Měděnec (k390, k384), Domašı́n (34b) and
Meluzı́na Hill (M) (ns3) are also shown.

112
 MECHANISM OF FORMATION OF ATOLL GARNET
Klápová et al. (1998), who estimated peak P-T
conditions of 2.5 GPa/600ºC. Massonne and Kopp
(2005) calculated higher P-T conditions of
2.7 GPa/720ºC for eclogite near the village of
Hammerunterwiesenthal, about 3 km west from
the study area.
Garnets in the eclogites usually show optical
and compositional zoning and have many small
inclusions, but most lack atoll structures.
Eclogites with atoll garnet were found in all
localities, but their relations to eclogites lacking
atoll garnet are not always clear. Only in the
Měděnec locality do they form a zone ~1 m thick
within eclogites lacking atoll garnet. The presence
of atoll garnet is not restricted to a specific rock
type, but occurs in metabasites, including quartz-
rich varieties, and probably represents a mixture
of mafic rocks and sediments.
Most samples with atoll garnet show strong
retrogression and are amphibole-rich. Five
samples, including one relatively unaltered
eclogite with atoll garnet, were chosen for this
study. All samples contain rutile, which is
rimmed, or partly replaced, by ilmenite. Sample
34b (Domašı́n) is a quartz-eclogite with ompha-
cite (30 vol.%), garnet (45 vol.%), quartz
(25 vol.%), and small amounts of rutile, amphi-
bole and albite. Omphacite is partially replaced by
an albite+amphibole symplectite. The remaining
four samples are rich in amphibole and have no
omphacite, or relics of it as inclusions in garnet.
Sample 49b (Kovářská) has more than 70 vol.%
amphibole with relict inclusions of atoll garnet.
Phengite was found as inclusions in the atoll
cores. Minor amounts of quartz and albite are also
present in this sample. Sample ns3 (Meluzı́na) is
characterized by thin layers of fine-grained,
garnet-rich rock with atoll garnet that alternates
with coarser-grained layers dominated by whole
garnet. Atoll garnets (0.1 0.5 mm in diameter)
occur mainly in leucocratic zones with quartz and
minor amphibole and talc. Whole garnets
(0.5 1.05 mm) contain inclusions of glauco-
phane, albite-amphibole symplectites and talc.
Samples k390 and k384 from Měděnec also
contain amphibole, quartz and phengite.
Some general features and textural relations of
the atoll garnet observed in most of the studied
samples are now summarized.
Typically, the more garnet is present, the
smaller are the crystals they form. When different
size varieties are compared, it is found that atolls
are developed mostly in the fine-grained varieties
(see Fig. 2a). The atoll garnets consist of a garnet
113
annulus surrounding one or more of quartz,
amphibole and phengite. If the atolls are occupied
by quartz, their interfaces are sharp and regular
and correlate with crystallographic planes of
garnet (Fig. 2b d). On the other hand, where
amphibole or mica overgrow the atoll, boundaries
with the garnet ring are irregular (Fig. 2a,e,f). In
sample 34b, omphacite occurs in the matrix and
within the atoll garnet (Fig. 2b,c). In amphiboli-
tized samples, amphibole surrounding atoll
garnet, and also occurring within the atolls, is
optically continuous (Fig. 2e).
Chemical analysis
Most minerals were analysed at the Institute for
Mineralogy and Petrology in Graz using a JEOL
6310 scanning electron microscope (SEM)
equipped with both wavelength- and energy-
dispersive spectrometers. Some garnet profiles
were obtained using a CamScan CS4 SEM at the
Institute of Petrology and Structural Geology in
Prague. Standards used for the analyses on both
microprobes were: jadeite for Na and Si, pyrope
for Mg and Al, adularia for K, andradite for Ca
and Fe, rutile for Ti and tephroite for Mn. Sodium
was measured by wavelength-dispersive spectro-
metry and the other elements by energy-
dispersive spectrometry. An accelerating potential
of 15 kV and a probe current of either 10 or 15 nA
and 20 s counting time on peak were used during
measurements of the elements both in Graz and in
Prague.
Data on crystallographic orientation for
different parts of atoll garnet crystals were
obtained by electron backscatter diffraction
(EBSD) using a CamScan CS4 SEM fitted with
a Nordif camera. The accelarating voltage of the
SEM for this work was 25 kV, and the EBSD
traces were indexed and plotted using the Channel
5 software from HKL Technology.
Results
Garnet
Backscattered electron images and compositional
maps (Figs 3, 4) in sample 34b, from whole and
atoll garnets respectively, indicate that garnet
cores (garnet I) and rims (garnet II) have different
compositions. The core I of whole garnet with
idiomorphic shapes in the BSE images has a large
Ca concentration (Gr 28 , Alm 62 , Py 6 , Sp 4 )
compared with its rim (II), which has small Ca,
large Mg and small Fe contents
 S. W. FARYAD ET AL.

FIG. 2. Photomicrographs of atoll garnet from eclogite with phengite and amphibole. (a) Contact between two zones
(layers) with coarse-grained whole and fine-grained atoll garnets; the zones are parallel to the foliation of the rock
(sample ns3). (b d) Quartz-rich variety, sample 34b. (b) Present atoll rims with island cores separated by
idioblastically shaped, polygonal quartz. The rock shows foliation defined by omphacite and quartz ribbons. (c and
d) Enlargements of the two boxed areas in (b). The arrow in (c) indicates an omphacite inclusion in the atoll core.
(e and f) The strongly amphibolitized variety (sample 49a) exhibiting atoll garnets with cores filled by amphibole or
quartz. (e) The amphibole crystal surrounding the atoll garnet shows the same optical orientation as the amphibole
inside the atoll garnet. (f) Shows an atoll garnet filled by idiomorphic quartz (upper left) and irregular amphibole
(lower right).

114
 MECHANISM OF FORMATION OF ATOLL GARNET

(Gr12Alm77Py9Sp2; Fig. 3a, a1; Table 1). Garnet I
shows prograde zoning with a decrease of Mn
towards the rim. Similar compositional zoning of
garnet can be seen in Fig. 3b, b1. However, the
centre of this garnet has two irregular-shaped
islands which have a compositional profile similar
to that of garnet II at the rim. Figure 3c shows a
whole garnet with a dark Ca-rich core and a
peninsula connected to the rim. The peninsula and
the rim have similar image contrast and, hence,
similar compositions (Fig. 3c, c1). Figure 4a
shows a more evolved stage of element exchange
which resulted in replacement of almost all the
core garnet I by garnet II of rim composition.
Only the inside edge of the atoll ring against
quartz shows a thin transitional zone which is a
remnant of the outermost part of garnet I.
Figure 4b shows the final stage, in which the old
core (garnet I) is almost totally replaced by garnet
II having a composition similar to that of the atoll
ring. Boundaries between the former core garnet
and the new rim are clearly visible by the
presence of quartz inclusions or by a narrow
compositional zone after garnet I. The presence of
numerous omphacite inclusions in the new garnet
is interesting in that it shows diffusion zoning in
contact with the host garnet II. Figure 4c e
indicates compositional maps of atoll garnet
filled by new garnet II in the core of garnet I.
Atoll garnet from the nearly bimineralic sample
(49a) (garnet+amphibole, Figs 5, 6) usually has at
least two compositional zones. Figure 5a shows
whole garnet I with a bell-shaped decrease of
spessartine content towards to the rim (Fig. 5a1).
This garnet is rimmed by a thin zone possessing
large pyrope and small almandine contents
(Fig. 5a2). The outer rim (garnet II) shows a
sharp change, with an increase in almandine and a
decrease in grossular contents. An increase in
spessartine content toward the rim is also evident.
The pyrope content is generally larger compared
with garnet I but smaller than that in the thin zone.
Interesting features of replacement progress were
observed by analysing whole garnet (see Fig. 6a),

TABLE 1. Selected microprobe analyses of atoll garnets.

Sample —————— 49a —————— ————— 34b ———————

Garnet I core I rim I-II II I core I rim I-II II

SiO2 38.76 38.58 39.83 39.18 37.78 38.06 37.70 37.68

TiO2 0.17 0.14 0.08 0.01 0.24 0.18 0.12 0.03
Al2O3 21.36 21.33 21.88 21.58 20.65 20.98 20.97 21.09
FeO 25.73 27.06 21.57 26.28 27.61 28.44 28.42 34.24
MnO 1.63 0.42 0.22 0.72 1.68 0.52 0.69 0.91
MgO 3.47 4.10 8.46 6.83 1.50 1.28 1.84 2.29
CaO 9.32 8.45 7.89 5.51 9.71 10.45 9.58 4.07
Total 100.44 100.08 99.93 100.11 99.17 99.91 99.32 100.31

Formulae based on 12 oxygens

Si 3.028 3.022 3.030 3.029 3.031 3.029 3.016 3.017
Ti 0.010 0.008 0.004 0.001 0.014 0.011 0.007 0.002
Al 1.967 1.968 1.962 1.967 1.952 1.988 1.977 2.003
Fe 1.681 1.772 1.372 1.699 1.852 1.910 1.901 2.304
Mn 0.108 0.028 0.014 0.047 0.114 0.035 0.047 0.062
Mg 0.404 0.479 0.960 0.787 0.179 0.153 0.219 0.275
Ca 0.780 0.709 0.643 0.457 0.835 0.900 0.821 0.352
S cations 7.978 7.986 7.985 7.987 7.977 8.026 7.988 8.015

Molecular proportions of garnet end members

alm 0.565 0.593 0.459 0.568 0.621 0.637 0.636 0.770
grs 0.262 0.237 0.215 0.153 0.280 0.300 0.275 0.117
py 0.136 0.160 0.321 0.263 0.060 0.051 0.073 0.092
sps 0.036 0.009 0.005 0.016 0.038 0.012 0.016 0.021

I, II are garnet types: I (core) and II (rim, atoll ring); I-II refers to the thin zone between core and rim.

115
 S. W. FARYAD ET AL.

which is also characterized by two compositional
zones having different Fe and Mg contents.
The close-up of garnet II shown in Fig. 6b
shows that the dark, low-Fe zone has microveins
with a similar image contrast to that of the
outermost rim of the garnet. These microveins
connect garnet core I with the rim, but the core
garnet shows a strong compositional gradient at the
points of contact between sub-grains with a large
Fe content. Compositional zoning of sub-grains
and microveins in garnet I change by substitution
of Fe and Mg, while Ca remains almost constant.
Compared with the narrow white veins in the dark
rim, sub-grains in the core are separated by wide
grey zones and channels with larger Mg contents.
Figure 4d shows garnet crystals with composi-
tional zones between garnet I and II and
replacement of garnet I core by amphibole. In
some cases, only rod-shaped relics of garnet II with
compositional zoning are preserved.
In addition to the preferential formation of atoll
garnet in small grains, a comparison of atoll and
whole garnets from the same sample indicates that
atoll garnet I has larger Ca and smaller Fe
contents (Fig. 7). The core-to-rim zoning
between the two garnets (I and II) is characterized
by a decrease of Ca and an increase of Mg, Fe and
Mg/(Mg + Fe). Some rim garnets II show, first an
increase of Fe, and then a decrease to the rim.
However, this trend is not always the case,

TABLE 2. Composition of omphacite and Na- and Na-Ca amphiboles in samples with the atoll garnets.

Mineral ————— Omphacite ————— – Amphibole –

Sample ——————— 34b ——————— —— 52 —— 49a
in grt* in grt{ m-c m-r in grt in grt m

SiO2 54.82 57.35 56.33 56.41 58.30 52.48 56.65

TiO2 0.31 0.08 0.16 0.26 0.04 0.16 0.12
Al2O3 9.29 14.80 8.77 9.04 11.60 10.16 5.64
FeO 10.49 2.30 10.22 10.22 6.99 8.85 6.62
MnO 0.12 0.09 0.02 0.07 0.04 0.10 0.07
MgO 6.02 5.81 6.01 5.60 11.31 13.02 16.04
CaO 10.00 10.75 11.21 11.07 1.94 7.28 8.23
Na2O 7.83 8.43 7.28 7.39 6.57 3.93 3.44
K2O 0.02 0.00 0.00 0.03 0.06 0.31 0.16
Total 98.90 99.61 100.00 100.09 96.84 96.29 96.97

Si 1.996 2.023 2.040 2.042 8.000 7.459 7.928

[4]
Al 0.004 0.000 0.000 0.000 0.000 0.541 0.072
Ti 0.009 0.002 0.004 0.007 0.004 0.017 0.013
[6]
Al 0.395 0.615 0.374 0.386 1.877 1.161 0.858
Fe3+ 0.146 0.000 0.048 0.037 0.000 0.000 0.000
Fe2+ 0.176 0.068 0.261 0.276 0.802 1.052 0.775
Mn 0.004 0.003 0.001 0.002 0.005 0.012 0.008
Mg 0.327 0.306 0.325 0.302 2.313 2.759 3.346
Ca 0.390 0.406 0.435 0.429 0.286 1.109 1.234
Na 0.553 0.577 0.511 0.518 1.747 1.083 0.933
K 0.001 0.000 0.000 0.001 0.010 0.056 0.029
Sum 4.001 4.000 3.999 4.000 15.043 15.248 15.196
M4
Na 1.714 0.892 0.766
A
(Na+K) 0.043 0.248 0.196

Aeg (mol.%) 15.0 0.0 5.0 4.0

Jd (mol.%) 40.0 58.0 43.0 44.0
XMg 0.65 0.82 0.55 0.52 0.74 0.72 0.81

* Omphacite inclusion in garnet (Fig. 2c)

{
Relict omphacite in new garnet (II) filling core of the atoll garnet (Fig. 4b), m = matrix, c = core and r = rim

116
 MECHANISM OF FORMATION OF ATOLL GARNET

FIG. 3. BSE images and compositional profiles from quartz-eclogite sample 34b. Images (a) (c) show whole garnets
with idiomorphic cores (dark, garnet I) along with compositional profiles (a1 c1). The dotted lines above the pyrope
profiles in a1 c1 represent XMg = Mg/(Mg+Fe) ratios in garnet. Note the two islands and a peninsula in the cores of
(b) and (c) have similar image contrast and compositions to that of the garnet rim II. The y axis in the compositional
profiles represents the fractional garnet end-member compositions.

117
 S. W. FARYAD ET AL.

FIG. 4. BSE images and compositional profiles of garnet with relics of garnet I (dark), replaced by garnet II (sample
34b). (a) Together with the compositional profile in a1, this shows an atoll garnet with the core replaced by new
garnet of similar composition to the rim and by quartz (Qtz), which follows the idiomorphic zone inside the atoll
ring. (b) Shows an atoll garnet with its core totally replaced by new garnet+quartz (black) with numerous relict
inclusions of omphacite (Cpx) and apatite (Ap). The y axis in the compositional profiles (a1) represents the fractional
garnet end-member compositions. (c e) BSE and compositional maps for Ca and Fe from an atoll garnet with a new
rim and an island within the core.

118
 MECHANISM OF FORMATION OF ATOLL GARNET

FIG. 5. BSE images and compositional profiles of garnet from an amphibolitized eclogite (Amp: amphibole;
sample 49a). (a) Shows whole garnet with compositional profiles (a1) from rim to rim. Note that the narrow dark
zone at the rim has small almandine and large pyrope contents (see profile a2). The y axis in the compositional
profiles represents the fractional garnet end-member compositions.

because the compositional zoning in garnet from
sample 34b is characterized by the Fe-poor and
Mg-rich zones.
Other minerals
Clinopyroxene from sample 34b is omphacitic
with a 43 58 mol.% jadeite component
(Table 2). It is almost homogeneous in composi-
tion. A small variation can be found in a single
crystal, in which the core shows relatively larger
amounts of Mg and hence XMg (core and rim =
0.55 and 0.52 respectively), but smaller amounts
of Na (0.50 0.51 a.p.f.u.) compared with the rim
(Table 2). Consistent with the increase of XMg at
the rim of garnet, compositional zoning of this
nature suggests an increase in temperature during
metamorphism. The Na-Ca clinopyroxenes
enclosed in garnet core I have a large
(15 mol.%) aegirine content, the Fe3+ estimate
being based on four cations and six oxygens.
Relict clinopyroxenes in new garnet, filling the
core of the atoll garnet, have large Al contents,
which could be the result of partial re-equilibra-
tion during the formation of garnet II.
Glaucophane was found as inclusions in whole
garnet from several samples which also contain
atoll garnet. Amphibole occurring in the matrix as
well as in the atoll core in sample 49a is a Na-Ca
amphibole of winchite-barroisite solid solution
with XMg = 0.6 0.8, where the B site is filled with
0.7 0.9 Na a.p.f.u. The Fe 3+ content in
amphibole was calculated on a formula based on
13 cations and 46 negative charges. In addition to
glaucophane, garnet from sample ns3 contains
inclusions of pargasitic amphibole and paragonite.
Talc, calcite, and chlorite were also observed in
the garnet core from this sample. Matrix phases
are winchite, paragonite, and retrograde chlorite
and plagioclase. The amphibole forming a
symplectite with albite is actinolitic hornblende.
Phengite analysed in sample k390 has Si =
3.35 a.p.f.u., and accessory epidote from sample
49a has the mol.% of XCzo = 100[(Al 2)/
(Altot.+Fetot. 2)] = 63%.
Crystallographic orientation of the atoll
garnets
Several atoll garnet grains in sample 34b were
examined by EBSD in order to quantify possible
angular deviation in the crystallographic axes
between atoll rims and islands or peninsula cores.
Three different varieties of cores and rims of the
atoll garnet were analysed (Fig. 8):
(1) island- or peninsula-shaped garnet in the
central part of the atolls;
(2) ‘more-or-less’ continuous and euhedral atoll
rings forming the edges of atolls;
(3) small, residual garnet inclusions in the
quartz between the atoll rings and island- or
peninsula-shaped garnet.

119
 S. W. FARYAD ET AL.

FIG. 6. BSE images and compositional profiles of garnet (sample 49a) showing microveins resulted from fluid
infiltration (a). The enlarged inset (b) exhibits narrow channels (veins) penetrating the rim of the garnet. The relics of
older core (lighter zones) occur as fragments dissolved by the new grey garnet. (c) indicates advanced stage of atoll
formation. The y axis in the compositional profiles represents the fractional garnet end-member compositions.

The original crystallographic orientations of the
crystal at sites of spot analyses in all three parts
among individual atoll garnets show no signifi-
cant variation (Fig. 8).
Discussion
The presence of microveins in the rim interior of
garnet I is good evidence for compositional
change due to fluid infiltration, and rules out a
model of multiple nucleation and coalescence
(Spiess et al., 2001; Dobbs et al., 2003). Electron
backscatter diffraction analyses (Fig. 8) showed
that the new garnet II, including ring, island and
peninsula in the core, have crystallographic
orientations matching that of the old garnet.
This observation argues against the multiple
nucleation and coalescence model and supports
overgrowing and surface-scale metasomatism
modifying the original garnet composition along
every pathway accessed by the metamorphic fluid,
even along cracks and some inclusion boundaries
inside the garnet. In what follows we will discuss
the major aspects of atoll garnet formation

120
 MECHANISM OF FORMATION OF ATOLL GARNET

FIG. 7. Composition of atoll and whole garnets from different samples in eclogites. Note that whole garnet has cores
with small Ca and large Fe contents compared with the atoll garnet from the same sample. Arrows indicate
compositional changes from core garnet I to ring garnet II. Garnet analyses from samples ns3a, k384, and k390 are
from D. Sidler, ETH Zürich (unpublished data).

examined in this study and outlined in the
introduction.
Mechanism of the atoll garnet formation by fluid
inf|ltration
Textural and compositional relations in the
samples studied show that atoll garnet formed by
replacement of the garnet I core which was zoned
according to the Rayleigh fractionation model and
formed during the prograde stage of eclogite facies
metamorphism. Elements released by the replace-
ment of garnet I were incorporated into later
garnet II. The importance of garnet fracturing
associated with a chemical modification of garnet
interiors during metamorphism has been addressed
by various researchers (e.g. Whitney et al., 1996;
Hames and Menard, 1993; Hwang et al., 2003;
Konrad-Schmolke et al., 2007). Selective compo-
sitional re-equilibration occurred along the sub-
grain boundaries and progressively modified the
garnet interiors with respect to Mg and in some
cases also Fe (Konrad-Schmolke et al., 2007). The
compositional resetting of mineral grains along
fractures and sub-grain boundaries occurs rapidly
as long as fluids are available, because chemical
potential gradients are maximized between the
matrix and grain interiors. In addition, the
presence of sub-grains markedly increases the
reactive surface area thereby enhancing solution-
precipitation processes.
In most cases in which atolls are filled by
quartz, there is a sharp demarcation at crystal
edges. If a crystal displayed original oscillatory
zoning, a discrete zone or zones would be
replaced preferentially, leaving the remainder of
the crystal intact, because a garnet/quartz inter-
face is thermodynamically much more stable
under metamorphic conditions than a garnet/
mica interface (Smellie, 1974; Waters, 2001).
The differential reactivity of these two interfaces
has been demonstrated by many studies of natural
and synthetic materials, in which discontinuous
nucleation and growth takes place across a
discrete interface (cf. Spry, 1969). If nucleation
is difficult, rod-shaped crystals may develop
(Christian, 1975).
In the presence of amphibole or mica, the atoll
garnet has irregular edges that are invariably
embayed or breached. This texture is probably

121
 S. W. FARYAD ET AL.

FIG. 8. Crystallographic orientation of the atoll garnet (sample 34b). (a) and (b) BSE images of the atoll garnet with
marked and numbered spots, where crystal axis orientation measurements were acquired. (c) and (d) Stereographic
projections of EBSD data from 13 and 12 spots marked in (a) and (b) respectively that correspond to the three <100>
directions in garnet. Solid squares represent the original orientation, with mutual angular differences less than 2º.
Open squares represent the rotated orientations of the fracture-divided domains (black arrows in (b)).

due to nucleation at a highly reactive site within
atoll garnet and gradual erosion of the core. In
many cases, primary inclusions of minerals that
react with the host garnet during changing
metamorphic conditions may play an important
role in the formation of atoll garnet (Smellie,
1974), but this is not the main formation process.
Replacement of earlier by later garnet
The gradual change of composition of garnet I
along veins and sub-grain boundaries suggests an
important role for fluid infiltration during the
formation of garnet II. Elements released by the
replacement of garnet I have been incorporated
into newly grown Mg-rich garnet rims. Prograde
resorption of garnet meant that Mn and Ca were
removed from the garnet core to contribute to the
formation of a ring and other phases, an
explanation confirmed by the large Mn content
of the atoll rim compared with the outermost rim
of garnet II. The outstanding feature of the atoll
garnet is its homogeneous central islands or
peninsulas, which have compositions similar to
those of the rims, implying growth under similar
conditions. The new central garnet was formed by
gradual compositional changes along the internal
boundary network as well as at the outermost rim
of the entire garnet grain (Fig. 6a). At the
beginning of this process, changes occur in the

122
 MECHANISM OF FORMATION OF ATOLL GARNET

FIG. 9. Inferred P-T conditions of atoll garnet in eclogites from the Krušné Hory, based on garnet/clinopyroxene
thermometry (Ravna, 2000; Ai, 1994). Maximum pressure conditions are based on data from Klápová et al. (1998)
and Massonne and Kopp (2005). The P-T path (grey line) is after Massonne and Kopp (2005). Numbers 1 3
correspond to the core (I, 1), the thin zone between the core and rim (2), and the rim or ring (II, 3) of garnet
formation. Metamorphic facies fields are after Okamoto and Maruyama (1999): BS blueschist, Law-Ec lawsonite-
eclogite, Ep-Ec epidote-eclogite, Amp-Ec amphibole-eclogite.

concentration of elements along narrow channels excluded, diffusion-induced dislocation migration

and subsequently along the internal boundaries,
which results in the modification of old garnet
core by a new garnet of rim composition. The
peninsular garnet may have acted as a main
channel for fluid infiltration. The final stage is
characterized by complete replacement of interior
garnet by a new garnet and/or another phase that
is in equilibrium with matrix minerals. The
breakdown of early garnet from inside, and
re-growth from outside inward, was confirmed
by Cheng et al. (2007).
Origin of veins
Most eclogites from the area studied show
foliation defined by a sub-parallel orientation of
omphacite and other eclogite-facies minerals.
However, both whole and atoll garnets are
idioblastic without significant elongation. While
a role for small-scale deformation fractures is not
and/or diffusion-induced recrystallization
processes (Hwang et al., 2003; Konrade-Smolke,
2007) appear to be important for the development
of vein and sub-grain textures. In such processes,
differential element fluxes occur along migrating
grain boundaries. A dissolution reaction can occur
most readily upon mineral surfaces where
intracrystalline defects, especially vacancies, are
concentrated. Previous studies have documented
preferential dissolution of a mineral in pits or
tubes that formed on surface defects (e.g. Blum et
al., 1990). Mineral growth can be promoted along
the edges of pits, even as the garnet is being
resorbed, because of its high surface energy and
resulting thermodynamic instability. Formation of
atoll garnet in small grains was probably due to its
very thin rim, which was easily penetrated by
fluid to exchange the necessary elements between
the matrix and the garnet core. Local equilibrium
during metasomatism (Thompson, 1959;

123
 S. W. FARYAD ET AL.
Korzhinskii, 1970; Hames and Menard, 1993)
could fractionate cations between garnet edges,
other minerals and the metamorphic fluid as
garnet dissolves and reprecipitates. If the equili-
brium was controlled by the fluid, then the
equilibrium composition of the reprecipitated
garnet should be the same along every surface
to which the fluid had access, irrespective of
which minerals are in contact with the garnet.
This metamorphic fluid could also be trapped
during metasomatism as pits are sealed by growth
along their edges. The garnet being resorbed had
relict, disequilibrium compositions, and the
equilibrium garnet compositions changed as a
result of the resorption. This interfacial meta-
somatism modified the original prograde garnet
zoning pattern along every pathway accessed by
the metamorphic fluid, even along cracks and
some inclusion boundaries within garnet.
Relative timing and metamorphic conditions of the atoll
garnet formation
From the compositions of the core I and rim II
garnets, it is clear that they formed under different
P-T conditions. As mentioned before, the source
material for garnet rim comes, not only from the
reactions between matrix phases, but also from
the garnet interior. Instability of the core garnet
and its subsequent breakdown occurred as the
result of reaction at higher temperature, which is
confirmed by prograde zoning of the full garnet
(garnet I) with inclusions of glaucophane and
aegirine-rich omphacite in the core (samples ns3
and 34b), implying a change from blueschist-
TABLE 3. Temperatures (ºC) calculated for garnet-clinopyroxene in sample 34b at 2 and 2.5 GPa.
– Pressure (GPa) – 2 2
Grt Cpx R Ai
I-r grt* 468\15
I-r m-c 544\11
I-r m-r 570\6
I-II grt* 527\9
I-II m-c 617\8
I-II m-r 649\5
II m-r 537\5
R = Ravna (2000); Ai = Ai (1994)
Garnets I, II, I-II are as described in Table 1.
* Omphacite enclosed in garnet (Fig. 2c).
m = Matrix omphacite.
124
eclogite to eclogite-facies conditions. Although
no lawsonite or its pseudomorphs were observed
in these eclogites, a nearly isobaric P-T path with
an increase of temperature from lawsonite (450ºC/
2.5 GPa) to epidote-eclogite facies conditions
(720ºC/2.7 GPa; Fig. 9) was proposed by
Massonne and Kopp (2005) based on thermo-
barometric calculations in phengite-bearing eclo-
gite. Such a P-T path, which indicates a rapid
change in temperature, probably reflects the
formation of thin zones between garnet I and II
with small Fe and large Mg contents (Figs 3 6).
The presence of omphacite in the matrix, as
well as in the atoll associated with garnet II,
suggests that atoll formation occurred during the
eclogite-facies stage before decompression. In the
absence of phengite or kyanite in sample 34b, we
could apply only garnet-clinopyroxene thermo-
metry. The results obtained using the calibrations
of Ravna (2000) and Ai (1994) are listed in
Table 3. The pressures of 2 and 2.5 GPa used in
these calculations are based on the data from
Klápová et al. (1998) and Massonne and Kopp
(2005). In general, the Ai (1994) calibration
provides a temperature 30 80º lower than that
of Ravna (2000), depending upon the pressure
used. Both methods show an increase in temp-
erature from the outermost rim of garnet I with
omphacite inclusions (420 470ºC/2 GPa) to the
narrow zone between garnet I and II (460 550ºC/
2.5 GPa). Calculations for the matrix omphacite
with a thin zone of garnet (I II) gave a higher
temperature (580 670ºC/2.5 GPa ) than that with
the ring garnet II (440 560ºC/2.5 GPa; see
Table 3). If we assume equilibrium of the garnet
2.5 2.5
R Ai
422\22 488\18 420\28
517\17 565\15 493\21
521\4 592\7 519\8
465\12 549\13 463\16
577\13 640\11 548\8
582\8 673\8 579\15
445\5 562\7 443\6
 MECHANISM OF FORMATION OF ATOLL GARNET
rim with matrix omphacite, these results suggest
that garnet growth continued after the peak of
metamorphism, at least in a rock of effectively
constant bulk composition. The fact that substan-
tial growth of garnet II happened at a late stage is
presumably related to the modification of the local
reactive bulk compositions caused by the fluid-
assisted enlargement of diffusion domains to
include previously inaccessible garnet cores.
The common occurrence of Na-Ca amphibole
of barrosite composition in the matrix with
inclusions of omphacite and garnet in eclogites
is good evidence for both pressure and temp-
erature decrease during exhumation of the rocks.
The retrograde P-T path crossing the amphibole
eclogite/blueschist facies boundary was probably
one of the reasons for preservation of eclogite
facies minerals and textures in the Krušné Hory
eclogite.
Acknowledgements
This work is a part of research projects
MSM0021620855 and P 210/10/0249. K.
Ettinger (Graz) kindly helped with microprobe
analyses. Giles Droop and Hans-Joachim
Massonne are thanked for careful and thorough
reviews of the manuscript. We also thank Mark
Welch for editorial handling of this paper.
References
Ai, Y. (1994) A revision of the garnet-clinopyroxene
Fe2+-Mg exchange geothermometer. Contributions
to Mineralogy and Petrology, 115, 467 473.
Atherton, M.P. and Edmunds, W.M. (1966) An electron
microprobe study of some zoned garnets from
metamorphic rocks. Earth and Planetary Science
Letters, 1, 185 193.
Blum, A.E., Yund, R.A. and Lasaga, A.C. (1990) The
effect of dislocation density on the dissolution rate of
quartz. Geochimica et Cosmochimica Acta, 54,
283 297.
Cheng, H., Nakamura, E., Kobayashi, K. and Zhou, Z.
(2007) Origin of atoll garnets in eclogites and
implications for the redistribution of trace elements
during slab exhumation in a continental subduction
zone. American Mineralogist, 92, 1119 1129.
Christian, J.W. (1975) The Theory of Transformations in
Metals and Alloys, Part 1: Equilibrium and General
Kinetic Theory, 2nd Edition. Pergamon, Oxford, UK.
Cooper, A.F. (1972) Progressive metamorphism of
metabasic rocks from the Haast Schist Group of
southern New Zealand. Journal of Petrology, 13,
457 492.
125
Dobbs, H.T., Peruzzo, L., Seno, F., Spiess, R. and Prior,
D.J. (2003) Unraveling the Schneeberg garnet
puzzle. A numerical model of multiple nucleation
and coalescence. Contributions to Mineralogy and
Petrology, 146, 1 9.
Faryad, S.W. (2009) The Kutná Hora Complex
(Moldanubian zone, Bohemian Massif): A composite
of crustal and mantle rocks subducted to HP/UHP
conditions. Lithos, 109, 193 208.
Faryad, S.W., Perraki, M. and Vrána, S. (2006) P-T
evolution and reaction textures in retrogressed
eclogites from Svetlik, the Moldanubian Zone
(Czech Republic). Mineralogy and Petrology, 88,
297 319.
Florence, F.P. and Spear, F.S. (1991) Effects of
diffusional modification of garnet growth zoning on
P T path calculations. Contributions to Mineralogy
and Petrology, 107, 487 500.
Franke, W. (2000) The mid-European segment of the
Variscides: Tectonostratigraphic units, terrane
boundaries, and plate tectonic evolution.
Pp. 35 61 in: Orogenic Processes: Quantification
and Modelling in the Variscan Belt. (W. Franke, V.
Haak, O. Oncken and D.Tanner, editors). Special
Publications, 179, The Geological Society, London.
Green, J.F.N. (1915) The garnets and streaky rocks of
the English Lake District. Mineralogical Magazine,
17, 207.
Hames, W.E. and Menard, T. (1993) Fluid-assisted
modification of garnet composition along rims,
cracks and mineral inclusion boundaries in samples
o f a m p h i b o l i t e f a c i e s s c h i s t s. A m e r i c a n
Mineralogist, 78, 338 344.
Hwang, S.L., Shen, P., Yui, T.F. and Chu, H.T. (2003)
On the mechanism of resorption zoning in
metamorphic garnet. Journal of Metamorphic
Geology, 21, 761 769.
Joesten, R. (1991) Grain boundary diffusion kinetics in
silicate and oxide minerals. Pp. 345 395 in:
Diffusion, Atomic Ordering and Mass Transport (J.
Ganguly, editor), Springer, Berlin.
Klápová, H. (1990) Eclogites of the Bohemian part of
the Saxothuringicum. Rozpravy Československé
akademie věd, (řada matematických a přı́rodnı́ch
věd), 100(5), 3 86. ISSN 0069-228X, Academia
Praha.
Klápová, H., Konopásek, J. and Schulmann, K. (1998)
Eclogites from the Czech part of the Erzgebirge:
Multi-stage metamorphic, and structural evolution.
Journal of the Geological Society, 155, 567 583.
Konopásek, J., Schulmann, K. and Lexa, O. (2001)
Structural evolution of the central part of the Krušné
Hory (Erzgebirge) Mountains in the Czech Republic
evidence for changing stress regime during
Variscan compression. Journal of Structural
Geology, 23, 1373 1392.
 S. W. FARYAD ET AL.
Konrad-Schmolke, M., O’Brien, P.J. and Heidelbach, F.
(2007) Compositional re-equilibration of garnet: the
importance of sub-grain boundaries. European
Journal of Mineralogy, 19, 431 438.
Korzhinskii, D.S. (1970) Theory of Metasomatic Zoning.
Clarendon. Oxford. UK, 162 pp.
Massonne, H-J. and Kopp, J. (2005) A low-variance
mineral assemblage with talc and phengite in an
eclogite from the Saxonian Erzgebirge, Central
Europe, and its P T evolution. Journal of
Petrology, 46, 355 375.
Medaris, G., Jelı́nek, E. and Mı́sař, Z. (1995) Czech
eclogites terrane settings and implications for
variscan tectonic evolution of the Bohemian Massif.
European Journal of Mineralogy, 7, 7 28.
Nakamura, D., Svojtka, M., Naemura, K. and Hirajima,
T. (2004) Very-high-pressure (>4 GPa) eclogite
associated with the Moldanubian Zone garnet
peridotite (Nové Dvory, Czech Republic). Journal
of Metamorphic Geology, 22, 593 603.
O’Brien, P. and Vrána, S. (1995) Eclogites with a short-
lived granulite facies overprint in the Moldanubian
Zone, Czech Republic: Petrology, geochemistry and
diffusion modelling of garnet zoning. Geologische
Rundschau, 84, 473 488.
Okamoto, K. and Maruyama, S. (1999) The high-
pressure stability limits of lawsonite in the
MORB+H2O system. American Mineralogist, 84,
362 373.
Rast, N. (1965) Nucleation and growth of metamorphic
minerals. Pp. 73 102 in: Controls of Metamorphism
(W.S. Pitcher and G.W. Flinn, editors). Oliver and
Boyd, Edinburgh.
Ravna, E.K. (2000) The garnet-clinopyroxene Fe2+-Mg
126
geothermometer: an updated calibration. Journal of
Metamorphic Geology, 18, 211 219.
Săbău, G., Negulescu, E. and Massonne, H.-J. (2006)
Chemical zonation and relative timing of growth
sections in garnets from eclogites of the Leaota
Massif, South Carpathians. Mineralogical Magazine,
70, 655 667.
Smellie, J.A.T. (1974) Formation of atoll garnets from
the aureole of the Ardara pluton, Co. Donegal,
Ireland. Mineralogical Magazine, 39, 878 888.
Spiess, R., Peruzzo, L., Prior, D.J. and Wheeler, J.
(2001) Development of garnet porphyroblasts by
multiple nucleation, coalescence and boundary
driven rotations. Journal of Metamorphic Geology,
19, 269 290.
Spry, A. (1969) Metamorphic Textures. Pergamon Press,
Oxford, UK, 350 pp.
Thompson, J.B. (1959) Local equilibrium in meta-
somatic processes. Pp. 340 361 in: Researches in
Geochemistry I (P.H. Abelson, editor). Wiley, New
York.
Waters, D.J. (2001) The significance of prograde and
retrograde quartz-bearing intergrowth microstruc-
tures in partially melted granulite-facies rocks.
Lithos, 56, 97 110.
Whitney, D.L. (1991) Calcium depletion halos and Fe-
Mn-Mg zoning around faceted plagioclase inclusions
in garnet from a high-grade pelitic gneiss. American
Mineralogist, 76. 493 501.
Whitney, D.L., Mechum, T.A., Dilek, Y. and Kuehner,
S.M. (1996) Modification of garnet by fluid
infiltration during regional metamorphism in garnet
through sillimanite-zone rocks, Dutchess County,
New York. American Mineralogist, 81, 96 705.