JOURNAL OF PETROLOGY
VOLUME 51
NUMBER 4
PAGES 947^972
2010
doi:10.1093/petrology/egq008
The Petrogenesis of Volcaniclastic Komatiites in
the Barberton Greenstone Belt, South Africa: a
Textural and Geochemical Study
MELANIE THOMPSON STIEGLER1, , DONALD R. LOWE1 AND
GARY R. BYERLY2
1
DEPARTMENT OF GEOLOGICAL AND ENVIRONMENT SCIENCES, STANFORD UNIVERSITY, STANFORD, CA 94305, USA
2
DEPARTMENT OF GEOLOGY AND GEOPHYSICS, LOUISIANA STATE UNIVERSITY, BATON ROUGE, LA 70803, USA
RECEIVED MARCH 10, 2009; ACCEPTED FEBRUARY 23, 2010
ADVANCE ACCESS PUBLICATION MARCH 30, 2010
The Onverwacht Group of the 3·5^3·2 Ga Barberton greenstone
belt, South Africa contains multiple stratigraphic units that include
laterally extensive beds of komatiitic ash, accretionary lapilli, and lapilli. These units have been affected by pervasive silicification,
serpentinization, or, less commonly, carbonate metasomatism.
Silicification resulted in SiO2 þ K2O 4 85 wt % and depletion of
most other major and trace elements. Most of these tuffs have prominent high Hf/Hf* and Zr/Zr* (0·5^12), which cannot result from
normal magmatic processes but are due to the typically immobile
rare earth elements migrating during post-silicification fluid^rock
interaction. Similarly, their low Ce/Ce* values do not reflect
Archean surface redox conditions but the circulation of later oxidizing fluids. Despite this intense alteration, ratios of Al2O3 and
TiO2 remain uniform and coherent within single volcanic units.
These ratios indicate that most silicified tuffs are not petrogenetically
related to the underlying or overlying komatiitic flow rocks and that
each originated from either separate mantle sources or different partial melting conditions. Serpentinized tuffs retain komatiitic element
abundances but Al2O3 fails to define a tight linear array with the
demonstrably immobile elements Ti and Zr. We speculate that this
is due to post-depositional mixing of Al-depleted and Al-undepleted
tuff layers by aqueous currents. Excellent textural preservation of
the silicified tuffs shows they are characterized by a dearth of phenocrysts, low particle vesicularity and abundance of fine vitric ash,
suggesting the eruption and rapid quenching of superheated or
near-liquidus anhydrous magmas. Minor assimilation of hydrated
basaltic or ultramafic crust within the dry magma may have
enhanced the surface phreatomagmatic explosivity while still allowing the magma to rise close to an adiabatic ascent path. However,
*Corresponding author. Telephone: (650)8046423. Fax: (650)7250979.
E-mail: [email protected]
textural and geochemical evidence for such a process is scarce.
Temporal and compositional constraints show that the diversity in
the types of komatiites throughout the Onverwacht Group can be accounted for by variations in plume^mantle dynamics and that komatiitic tuffs were deposited during intervals of volcanism
characterized by low effusive eruptive volumes and/or low emplacement rates.
KEY WORDS:
Archean; greenstone belts; komatiite; petrogenesis; tuff
I N T RO D U C T I O N
Komatiites are volcanic rocks containing 418 wt % MgO
that provide an important window on early tectonics
and mantle evolution. They are thought to represent very
fluid magmas that were emplaced primarily as lava flows,
occasionally as shallow-level sills, and rarely explosively.
Jahn et al. (1982) divided komatiites into three main
geochemical classes based on their major and trace element compositions: (1) Al-depleted komatiites (ADK) that
have relatively low alumina contents (3·0^4·0 wt % at
25 wt % MgO), Al2O3/TiO2515, and depletions in the
heavy rare earth elements (HREE) (i.e. Gd/Ybn41); (2)
Al-undepleted komatiites (AUK) that have roughly primitive mantle Al2O3/TiO2 ratios (15^30) and flat REE patterns; (3) Al-enriched komatiites (AEK) that are
relatively enriched in Al (4·5^5·5 wt %) and have
Al2O3/TiO2430 and Gd/Ybn51. Varying Al contents are
ß The Author 2010. Published by Oxford University Press. All
rights reserved. For Permissions, please e-mail: journals.permissions@
oxfordjournals.org
JOURNAL OF PETROLOGY
VOLUME 51
not always the primary control on Al2O3/TiO2 ratios and
other classes of komatiite have since been identified
(Barnes & Often, 1990; Hanski et al., 2001; Sproule et al.,
2002).
In part because similar high-Mg magmas are not
erupted on the modern Earth, the melting conditions and
tectonic settings required to produce the geochemical
variations in komatiites are unknown. One model of komatiite petrogenesis involves 530^50% partial melting
(Herzberg, 1992) of an anhydrous mantle source producing
magmas with eruption temperatures between 1400 and
16008C (Nisbet et al., 1993). Such magmas would have
originated within mantle plumes and erupted to form
oceanic islands or plateaux. By contrast, it has been
inferred that some komatiites were derived from hydrous
melting of a shallow, depleted mantle fluxed by water
from a descending slab (Alle'gre, 1982; Parman et al., 2001).
This scenario allows for significantly lower mantle temperatures than for dry melting, as hydrous melting reduces
the liquidus temperature. A third model combines the two
tectonic regimes and proposes that komatiites were generated during the impingement of mantle plumes on subducting oceanic lithosphere (Hollings & Wyman, 1999).
Those who favor dry melting for the genesis of komatiites have cited a number of trace element partitioning
experiments as support for deep, high degrees of partial
melting. Ohtani et al. (1989) demonstrated that major geochemical differences between the three main compositional
types of komatiite can be explained by the fractionation of
majorite garnet during melting. Garnet retains Al, Cr, Sc,
V, Zr and Hf relative to the middle REE (MREE), and
the HREE so that any ultramafic melt derived from such
a source would be depleted in those elements. For garnet
to be stable in an ultramafic liquid and to realize a high
degree of melting, pressures and depths greater than those
found in dehydrating oceanic crust are required (Arndt,
2003). Experimental data indicate that komatiites could
be generated at partial melting pressures between 2 and
9 GPa (Herzberg & O’Hara, 1998). The incompatible element depletion of most komatiites is inconsistent with generation in a subduction-related setting, as fractional melting
strips the source of both water and incompatible elements
(Arndt et al., 1998). Any water added subsequently would
have been accompanied by the addition of other large ion
lithophile elements (LILE).
By contrast, the high dissolved water contents estimated
for some Barberton magmas (3^6 wt %: Parman et al.,
1997) as well as the 1·2 wt % of magmatic water calculated
for the igneous amphibole-bearing komatiites of Boston
Creek, Ontario (Stone et al., 1997) have been cited as
evidence in favor of a subduction origin for komatiitic
magmas. Additionally, although the petrogenetic relationship between komatiites and komatiitic basalts, which
have 12^18 wt % MgO, is unclear (Grove & Parman,
NUMBER 4
APRIL 2010
2004), some komatiitic basalts and modern arc-related
boninites have similarly high MgO, SiO2, and LILE contents and low TiO2 and Nb/La values (Grove & Parman,
2004).
Field and textural analyses provide additional constraints on komatiite petrogenesis. Olivine and minor chromite are the main liquidus phases in komatiites, occurring
as skeletal crystals in spinifex zones and as euhedral
grains in massive flows and cumulate regions (Arndt
et al., 1997). Crystal transport commonly took place in
cumulate zones within lava flows (Renner et al., 1994), but
many cumulus olivines show compositional evidence for
liquid fractionation and occasionally occur intergrown
with poikilitic chromite, indicating in situ crystallization
(Barnes, 1998). Equivocal petrogenetic evidence of magmatic volatiles includes locally abundant vesicles in lava
flows (Beresford et al., 2000; Dann, 2001) and vesicular
fragments and olivine and pyroxene phenocrysts in pyroclastic komatiites (Saverikko, 1985; Barley et al., 2000).
These features suggest gas exsolution (Stone et al., 1997;
Arndt et al., 1998; Barley et al., 2000), but the origin and
percentage of volatiles required for their formation is
poorly constrained.
In this study, we discuss the origins of layers of ultramafic pyroclastic debris contained within sedimentary chert
units and thick serpentinized komatiitic sequences in the
3·55^3·25 Ga Barberton greenstone belt (BGB), South
Africa. Previous studies have described the sedimentology
and physical volcanology of these komatiitic pyroclastic
units (Lowe, 1999a, 1999b; Thompson Stiegler et al., 2008,
2010). This study focuses on their geochemistry and textures to determine the effects of post-eruptive alteration,
the tuffs’ genetic relationship to Barberton komatiites, and
the role of volatiles in their genesis. We present possible
petrogenetic models and briefly explore the temporal evolution of komatiitic volcanism in the BGB.
GEOLOGIC A L S ET T I NG
The Barberton greenstone belt, South Africa, includes a
12^15 km thick succession of volcanic and sedimentary
rocks that represents one of the oldest known, relatively
fresh stratigraphic sequences on Earth. The belt is divided
into three main lithostratigraphic units (Fig. 1a). From
base to top, these are the 8^10 km thick Onverwacht
Group, composed largely of mafic and ultramafic volcanic
rocks and subordinate cherts and felsic volcanic rocks;
the 1·8 km thick terrigenous clastic and dacitic volcaniclastic strata of the Fig Tree Group; and the up to
3 km thick Moodies Group composed of siliciclastic conglomerate, sandstone, and siltstone (Lowe & Byerly, 1999,
2007).
South of the Inyoka Fault and north of the Komati Fault
(Fig. 1b), major structural and stratigraphic boundaries in
the greenstone belt, the Onverwacht Group is divided into
948
THOMPSON STIEGLER et al.
(a)
SOUTHERN FACIES
BARBERTON KOMATIITES
thought to be correlative with the upper part of the
Mendon Formation (Lowe & Byerly, 1999). Komatiitic
tuffs are found throughout the Onverwacht Group, occurring in nine stratigraphic units (e.g. M1c, M2c, etc.;
Fig. 1) that have petrological and geochemical signatures
consistent with reported komatiitic lavas. Pyroclastic units
are further divided based on their occurrence either south
or north of the Inyoka Fault (Thompson Stiegler et al.,
2008). South of the fault, komatiitic tuffs are present in the
Hooggenoeg, Kromberg, and Mendon Formations. Their
bulk compositions are dominated by SiO2, Al2O3, and
K2O, a consequence of pervasive metasomatism that led
early workers to interpret them as products of felsic magmatism (Viljoen & Viljoen, 1969; Lowe & Knauth, 1978;
Lanier & Lowe, 1982). Their ratios of immobile elements,
such as Ti, Al, Th, Nb, and Zr, however, are close to primitive mantle values and similar to those in komatiites
(Lowe, 1999b; Thompson Stiegler et al., 2008). These tuffs
overlie komatiites, komatiitic basalts and, in one case, tholeiitic basalts, and represent deposition during breaks in effusive komatiitic and basaltic volcanism. North of the
fault, in the Weltevreden Formation, the komatiitic tuffs
have not been affected by Si-metasomatism but have been
serpentinized. Serpentinization has preserved elemental
abundances consistent with an origin from komatiitic liquids, including high MgO (23^36 wt %), Ni and Cr contents, and low levels of incompatible elements (Thompson
Stiegler et al., 2008). Primitive mantle normalized REE
patterns are typical of komatiites. The tuffs occur interbedded with komatiite and komatiitic basalt flow rocks
within thick sequences that lack material of intermediate
to felsic composition.
NORTHERN FACIES
Inyoka Fault
K1
← 3416 Ma °
← 3445 Ma §
H5
H4
H3
← 3470 Ma^
0
Komatiite
H2
‡
3286 Ma
Komatiitic basalt
H1
#
Basalt
Felsic intrusive rocks
~
Felsic volcaniclastic rocks
Mafic volcaniclastic rocks
Mudstone & shale
Sandstone, siltstone, & congl.
← 3472 Ma
Komati Fm.
Kilometers
Onverwacht Group
Hooggenoeg Fm.
H6
5
Kilometers
K2
Fig Tree Group
†
← 3258 Ma (Fig Tree)
← 3298 Ma† (M3)
2
←
3334 Ma†
Onv. Gr.
Welt. Fm.
Kromberg
Moodies
Fig Tree
Mendon
K3
← 3481 Ma
Terrigenous & volcanic rocks
Chert
3--- Age in millions of years
0
References: # Armstrong et al. (1990); † Byerly et al. (1996); ^ Byerly et al. (2002);
~ Dann (2000); § de Wit et al. (1987); ° Kroner et al. (1991); ‡ Lahaye et al. (1995).
(b)
31˚E
SEDIMENTARY FORMATIONS
Moodies Group
Fig Tree Group
VOLCANIC FORMATIONS
Onverwacht Group
Weltevreden Fm.
Mendon Fm.
Kromberg Fm.
Hooggenoeg Fm.
Komati Fm.
Undifferentiated mafic
volcanic rocks
Inyoka Fault
S T R AT I G R A P H Y O F T H E
O N V E RWAC H T G RO U P A N D
DI ST R I BU T ION OF
P Y RO C L A S T I C KO M AT I I T E S
Komati Formation
Onverwacht Anticline
Komati Fault
26˚S
200 km
BGB
N
South Africa
0
10 km
20
Indian Ocean
Fig. 1. Generalized stratigraphic section (a) and geological map (b)
of the western half of the Barberton Greenstone Belt (modified from
Lowe & Byerly, 1999). See also references: Armstrong et al., 1990;
Byerly et al., 1996, 2002; Dann, 2000; de Wit et al., 1987; Kroner et al.,
1991; Lahaye et al., 1995.
four formations. From base to top, these are the Komati,
Hooggenoeg, Kromberg, and Mendon Formations. North
of the Inyoka Fault, the Weltevreden Formation is the only
formation recognized in the Onverwacht Group and it is
The 3·1km thick Komati Formation is composed of massive, spinifex-textured komatiite and komatiitic basalt lava
flows and massive, pillowed tholeiitic basalts (Viljoen &
Viljoen, 1969; Williams & Furnell, 1979; Dann, 2000). A
5^10 cm thick dacitic tuff in the lower Komati Formation,
U^Pb zircon dated at 3481 2 Ma, is the only known sedimentary unit in the formation (Dann, 2000).
Hooggenoeg Formation
The 3·9 km thick Hooggenoeg Formation has been divided
into six members, H1^H6 (Lowe & Byerly, 1999). The
oldest member, H1, is a regionally traceable unit of silicified komatiitic ash, accretionary lapilli, and carbonaceous
matter 1^5 m thick (Lanier & Lowe, 1982). This chert has
949
JOURNAL OF PETROLOGY
VOLUME 51
been dated at 3472 5 Ma (Armstrong et al., 1990). The
unit overlies heavily altered komatiitic basalts at the top
of the Komati Formation and underlies the 1·2^1·8 km
thick sequence of tholeiitic basalts of Member H2v. Black
and black-and-white banded cherts (H2c) and locally, silicified vesicular scoria cap these tholeiitic rocks. Member
H3v is composed of variolitic pillow basalts, komatiitic
basalts, and minor komatiites capped by a regionally extensive layer, 2^22 m thick, of silicified ash, graded accretionary lapilli, and minor black carbonaceous cherts
(H3c). H4v is composed mainly of komatiitic basalts at
the base and high-Mg pillow basalts at the top. The capping chert, H4c, is a discontinuous layer up to 6 m thick
containing silicified volcaniclastic debris and carbonaceous
matter. A bed of impact-produced spherules in the chert
has been dated to 3470 2 Ma (Byerly et al., 2002). The
youngest mafic member, H5, includes both massive and
pillowed tholeiitic basalts. The capping chert, H5c, is
1^4 m thick and consists of black chert overlain by multiple, thin layers of komatiitic ash and accretionary lapilli.
The uppermost member of the Hooggenoeg Formation,
H6, is a sequence of dacitic intrusive and volcaniclastic
rocks.
Kromberg Formation
The Kromberg Formation comprises up to 1·8 km of volcanic and sedimentary rocks (Lowe & Byerly, 1999). On
the west limb of the Onverwacht Anticline, the lowest
member, K1, consists of 150^350 m of black-and-white
banded chert called the Buck Reef Chert that locally includes at its base a 5^50 m thick sequence of silicified evaporites, tuffs, and clastic rocks (Lowe & Worrell, 1999). One
of these tuffs has been dated at 3416 5 Ma (Byerly et al.,
1996). On the east limb of the Onverwacht Anticline, K1
includes a basal section of ultramafic flows, carbonaceous
cherts, and mafic volcaniclastic rocks overlain by interbedded pillow basalts and cherty metasediments. The
overlying rocks of K2 consist of up to 1km of mafic lapilli
tuff, lapillistone, interbedded basaltic volcanic rocks, and
at the top thin layers of silicified ash and black chert
(Ransom et al., 1999). K3 is made up of 350^600 m of
pillow basalts, pillow breccias, and locally interstratified
komatiitic and basaltic lavas. It is capped by 15^25 m of
black and black-and-white banded chert named the
Footbridge Chert (Lowe & Byerly, 1999) that includes a
5 cm thick felsic tuff dated at 3334 3 Ma (Byerly et al.,
1996).
Mendon Formation
The Mendon Formation is a cyclic sequence of komatiitic
volcanic rocks separated by generally thin chert layers
(Byerly, 1999; Lowe & Byerly, 1999). The lowest 200^250 m
thick cycle, M1v, is composed primarily of a single massive
cumulate komatiite and, locally at the base and top of the
section, thin spinifex-bearing komatiites. The overlying
NUMBER 4
APRIL 2010
Msauli Chert (M1c) is a distinctive, regionally continuous
unit, 20^35 m thick, of silicified pyroclastic debris interbedded with carbonaceous chert (Lowe, 1999c; Lowe &
Byerly,1999). In parts of the BGB, the Msauli Chert is overlain by 20^50 m of black, banded, and ferruginous chert
succeeded by rocks of the Fig Tree Group, the base of
which has been dated at 3259 4 Ma (Kroner et al., 1991).
Elsewhere, the Msauli Chert is overlain by M2, which includes 100^150 m of komatiitic volcanic rocks (Byerly,
1999) succeeded by a distinctive, 2^10 m thick layer of silicified komatiitic ash, large accretionary lapilli, and microporphyritic lapilli tuff (M2c). The overlying M3 contains
over 40 m of komatiitic basalts and komatiites overlain by
51^2 m of black chert and silicified komatiitic ash and
accretionary lapilli. Several thin, felsic tuffs in this upper
chert have a mean age of 3298 3 Ma (Byerly et al.,
1996). Higher cycles of the Mendon Formation are similar
but none is known to include volcaniclastic layers.
Weltevreden Formation
The Weltevreden Formation contains the oldest rocks north
of the Inyoka Fault and has been dated by Nd isotopes at
3286 29 Ma (Lahaye et al., 1995). It includes up to 2 km
of serpentinized komatiitic tuffs and flow rocks (Lowe &
Byerly, 1999), layered peridotitic complexes (Anhaeusser,
1985), and rare black and banded cherts. Laterally and vertically discontinuous units of tuffaceous material, 2^60 m
thick, are interbedded between single differentiated
komatiite and komatiitic basalt flow rocks. In the uppermost part of the Weltevreden Formation, just below the
base of the overlying Fig Tree Group, multiple thin beds
of silicified ash and accretionary lapilli crop out locally
and are probably contained in a displaced fault block.
These represent the only silicified ash in the Weltevreden
Formation.
P E T RO G R A P H I C
C H A R AC T E R I S T I C S
South of the Inyoka Fault, silicification of most komatiitic
pyroclastic units has converted them to various types of
impure chert and made them relatively resistant to erosion, forming prominent outcrops that have distinctive
grey, pale green, and bluish grey colors. Early silicification
(Lowe, 1999b) has resulted in minimal compaction and
excellent preservation of most primary textures and sedimentary structures. Most particles have been altered to
either pure silica or to microcrystalline mosaics of quartz,
chlorite, and sericite. Some silica-filled grains consist of
microcrystalline quartz or show cavity-fill textures with
quartz domains coarsening towards the interior. Other
clast textures include cores of phyllosilicate surrounded by
variably thick layers of microcrystalline quartz; complex
intergrowths of silica and phyllosilicates; or, less commonly, a thin layer of very fine-grained phyllosilicates aligned
950
THOMPSON STIEGLER et al.
BARBERTON KOMATIITES
perpendicular to outer grain boundaries with quartz in the
center. The interstitial cement is mainly microcrystalline
quartz, with relict chalcedonic banding preserved in some
samples. The only indications of initial zeolite precipitation
are radial arrangements of phyllosilicates inside spherulites
in the matrix of tuffs in H3. In addition to the pyroclastic
deposits of K2, which show extensive carbonate cementation, silicified tuffs near intrusive bodies are commonly
recrystallized to the extent that original pyroclast boundaries are not easily resolvable.
Pyroclastic debris includes ash, abundant armored and
accretionary lapilli (Fig. 2a), and microphyric to porphyritic lapilli fragments. Lapilli and ash are the only grain
sizes present. Beds are mainly fine-grained, well-sorted,
and massive to normally graded. Layers of armored and
accretionary lapilli commonly grade up into fine-grained
ash and frequently exhibit current-structures at the top.
Cross-lamination and flat-lamination are widely preserved
in tuff beds, indicating that the volcanic particles were
often reworked by moving water (Lowe, 1999c).
Occasionally, reworked tuff layers contain detrital chert or
carbonaceous grains mixed in with the ash and accretionary lapilli.
Fine-grained ash beds are nearly aphyric. Pyroclasts are
blocky, elongate, or curvilinear and poorly to non-vesicular
(Fig. 2c). A few beds contain abundant bubble walls
shards (Fig. 2d), which have a wide range of vesicularities
(0^65 vol. %). Lithic clasts are scarce. Primary volcanic
quartz and feldspar are absent. Accessory chromites are
the only magmatic minerals preserved. Spinels are sparsely distributed but have been identified in all volcaniclastic
units except K2. They are present as fine (515^75 mm)
euhedral grains in tuff layers, lithic clasts (Fig. 2e), accretionary lapilli (Fig. 2f), microporphyritic lapilli, and
altered glass shards. Their outer margins are often
rimmed by magnetite overgrowths.
Silica-replaced crystal forms are rare except in the lapilli
tuffs in the Kromberg (K1, K2) and Mendon (M2)
Formations. In M2, the lapilli fragments contain hexagonal, long, narrow microphenocrysts (most 5100 mm)
that occasionally show inclusions, usually along central
grain axes (Fig. 3a). Their morphology indicates that they
were originally olivine and/or pyroxene. The few larger
phenocrysts (55% of crystals) exhibit a similar morphology and range from 200 to 410 mm across. A third population of crystals is small, needle-like (aspect ratios are up
to 16:1), and either randomly oriented (Fig. 3b and c) or
preferentially aligned within lapilli fragments. These
could represent fine olivines, pyroxenes, or feldspars. The
presence of fine-grained quench rims, the overall fine crystal size, and the scarcity of coarse euhedral crystals
derived from fractured cumulate material indicate that
these microphyric lapilli represent juvenile ejecta and not
debris from explosively fragmented komatiitic flow rock.
Lapilli tuffs in K1 contain both fine, acicular crystal forms
and equant, silica-replaced olivine phenocrysts up to
750 mm in diameter. In silicified sections of K2, pseudomorphed olivines range up to 1·2 mm (Fig. 3d) and 52%
of the crystals are under 100 mm.
North of the Inyoka Fault, komatiitic tuffs are serpentinized and dominated by fine-grained amphibole, serpentine, chlorite, and/or talc. In outcrop, they are grey to
nearly black on fresh surfaces and weather to a tan or
medium grey colour. In general, the low-temperature
alteration has not preserved the microscopic tuffaceous
fabrics found elsewhere; however, macroscopic structures,
such as fine flat laminations (Fig. 2b), cross-beds and soft
sediment deformation, are well developed. Interbedded
non-volcanogenic and non-komatiitic detritus (such as
chert grains, zircons, and quartz phenocrysts) has not
been identified. Scattered chromites are the only primary
komatiitic minerals. These spinels range from 30 to 115 mm
in diameter and are euhedral or possess ragged, scalloped
edges. Magnetite often occurs as replacement of original
chromite cores.
G E O C H E M I S T RY
Analytical methods
Geochemical analyses of BGB tuffs were performed by
X-ray fluorescence (XRF) at the Washington State
GeoAnalytical Laboratory, employing the standard methods described by Johnson et al. (1999). Trace element compositions were determined by inductively coupled plasma
mass spectroscopy (ICP-MS) at Washington State
University, following the procedure of Knaack et al. (1994).
Table 1 contains representative major and trace element
analyses of the komatiitic tuffs (see Electronic Appendix 1
for the complete dataset, which is available for downloading at http://www.petrology.oxfordjournals.org). Only
samples with less than 95 wt % SiO2 are discussed as the
low abundance of all other elements in rocks with greater
than 95% silica produces element ratios that are irregular
and unreliable.
Chromites were analyzed by JEOL 733 microprobes at
Stanford University and at Louisiana State University
(LSU) using a focused beam and an accelerating voltage
of 15 kV (Table 2). Spinels either were examined in situ in
thin section or were extracted from the base of massive
and normally graded beds through hydrofluoric acid dissolution and then mounted on glass slides. The quality of
analysis was monitored by the repeated evaluation of a
suite of Smithsonian standards, including Kakanui hornblende, Johnstown hypersthene, chromite, ilmenite, and
glasses GL37 and GL39 at LSU and olivine, rutile, hematite, spessartine garnet, chromite, and V- and Ni-metals at
Stanford University. Counting precision was 1^2% for
major elements and decreased to 10% for minor elements.
Corrections were made to eliminate Ti Kb interference on
951
JOURNAL OF PETROLOGY
(a)
VOLUME 51
NUMBER 4
APRIL 2010
(b)
Accretionary
lapilli
Flat laminated ash
Cross-laminated ash
2 cm
(c)
(d)
Vesicles
Blocky
pyroclasts
200 µm
500 µm
(e)
(f)
45 µm
20 µm
Cr-sp
Ol
Cr-sp
Lithic
500 µm
100 µm
Fig. 2. (a) Silicified cross-laminated ash overlain by a layer of accretionary lapilli, M1c; (b) serpentinized flat laminated tuff, Weltevreden
Formation. Plane-polarized light photomicrographs of silicified (c) accretionary lapilli, H1, and (d) poorly vesicular shards, H3c. Chromites
within (e) a chloritized grain, M2c and (f) silicified accretionary lapilli, M1c. The insets are reflected light photographs of the chromites; the
bright spots within the chromite cores are analytical pits created by the microprobe beam. Cr-sp, chrome spinel.
952
THOMPSON STIEGLER et al.
(a)
Crystals
BARBERTON KOMATIITES
(b)
Inclusion
Finegrained
rim
Randomly oriented
acicular crystals
Bubble
wall
150 μm
500 μm
(d)
(c)
Elongate
crystals
Crystal
Crystals
Bubble wall
200 μm
1 mm
Fig. 3. Photomicrographs of lapilli tuffs from unit K1c. (a) A coarse ash fragment containing former glassy inclusions within the cores of elongate, hexagonal crystals; (b) lapillus with a very fine-grained quench rim and an interior containing randomly oriented bladed crystals;
(c) clast with both euhedral and long acicular microphenocrysts. (d) A silicified lapillus from K2 with euhedral phenocrysts.
the V Ka peak, and Fe(II) and Fe(III) were calculated
from total Fe assuming stoichiometry (Droop, 1987). Four
chromites were analyzed at both universities to detect any
analytical biases between machines; all tests returned identical or near-identical results. This study aims at describing
the primary compositional variations, therefore any samples that were modified by significant metamorphism or
alteration were discarded according to the methods outlined by Barnes (1998).
Southern facies: silicified volcaniclastic
rocks
All tuff layers south of the Inyoka Fault have bulk compositions dominated by SiO2, Al2O3, and K2O and are
depleted in FeO, MgO, CaO, and Na2O, reflecting
post-depositional alteration (Hanor & Duchac, 1990).
Al2O3 and TiO2 values are variable but Al2O3/TiO2
ratios are constant within single volcaniclastic units
(Table 3). Relative to the REE and high field strength
elements (HFSE), all tuffs are enriched in Rb and, less
strongly, in Ba (Fig. 4). Tuffs also exhibit strong to zero depletions in Ce (Ce/Ce* ¼ 0·1^1·0) and very slight depletions to pronounced enrichments in Eu (Eu/Eu* ¼ 0·9^3·1).
Al-undepleted tuffs are the most abundant komatiitic
volcaniclastic rocks in the BGB. They are present in the
Hooggenoeg (H1, H3c) and Mendon (M1c) Formations.
They are primarily identified by Al2O3/TiO2 ratios between 16 and 32. Tuffs in both H1 and H3c have variable
trace element patterns and, with few exceptions, their
HREE are not consistent with olivine fractionation as
953
JOURNAL OF PETROLOGY
VOLUME 51
NUMBER 4
APRIL 2010
Table 1: Representative major and trace element abundances for BGB komatiitic tuffs
Location:
Sample:
Weltevreden Fm.
M3c
M2c
M1c
K2
K1c
H5c
H3c
H1
MSA
MSA
SAF
MSA
SAF
SAF
SAF
SAF
SAF
SAF
14-1
23-1
181-3
33-2
102-3
75-19
156-1
493-3
336-1
499-4
Latitude (S):
25849·9150
25850·0030
258540 0100
25854·3330
258540 1700
258570 0300
258550 5800
25856·20
258560 09·700
25858·470
Longitude (E):
30857·3200
30857·1230
318010 0700
30857·0230
308540 0200
318050 2600
308590 7·500
30850·20
308520 4400
30853·90
Un-normalized major element oxides (wt %)
SiO2
47·53
43·17
90·74
91·07
91·10
94·00
92·22
94·59
89·48
93·09
Al2O3
3·14
3·89
4·03
2·43
5·37
3·07
5·89
3·93
6·28
4·11
TiO2
0·34
0·13
0·42
0·85
0·20
0·51
0·13
0·40
0·20
0·16
FeO*
10·45
7·66
2·68
0·91
0·42
0·51
0·79
0·10
0·48
0·46
MnO
0·12
0·09
0·03
0·02
0·00
0·00
0·03
0·00
0·01
0·01
CaO
9·63
3·26
0·01
0·09
0·04
0·08
0·02
0·04
0·01
0·02
MgO
23·42
31·90
0·93
0·37
0·85
0·29
0·34
0·32
1·34
0·75
K2O
0·01
0·01
0·88
0·51
1·88
0·99
0·69
1·13
1·79
1·01
Na2O
0·28
0·09
0·03
0·02
0·15
0·08
0·15
0·04
0·07
0·10
P2O5
0·02
0·01
0·02
0·07
0·01
0·03
0·01
0·01
0·01
0·01
LOI
5·36
9·61
1·07
0·70
0·96
0·75
1·21
0·67
1·46
0·89
94·93
90·19
99·76
96·35
100·02
99·56
100·26
100·56
99·66
99·72
Total
Trace elements (ppm)
Ni
1008
1588
226
199
55
247
55
16
105
53
Cr
2219
2384
60
1101
329
939
395
135
276
358
La
1·06
0·27
5·89
3·79
0·74
1·09
0·67
9·50
0·24
0·08
Ce
2·96
0·81
2·88
10·03
1·43
2·47
0·96
6·24
0·34
0·14
Pr
0·49
0·14
1·51
1·46
0·23
0·37
0·12
1·65
0·05
0·02
Nd
2·60
0·76
6·92
6·42
1·13
1·78
0·62
6·14
0·20
0·10
Sm
0·91
0·29
1·62
1·57
0·36
0·50
0·25
1·29
0·04
0·04
Eu
0·36
0·13
0·53
0·54
0·13
0·17
0·15
0·39
0·03
0·02
Gd
1·12
0·44
1·77
1·73
0·43
0·56
0·44
1·13
0·04
0·07
Tb
0·20
0·09
0·28
0·30
0·08
0·09
0·10
0·18
0·01
0·01
Dy
1·21
0·64
1·76
1·73
0·54
0·57
0·73
1·10
0·06
0·07
Ho
0·23
0·14
0·38
0·32
0·12
0·11
0·17
0·24
0·02
0·02
Er
0·58
0·42
1·05
0·69
0·36
0·27
0·51
0·62
0·05
0·04
Tm
0·08
0·06
0·15
0·08
0·05
0·03
0·08
0·09
0·01
0·01
Yb
0·50
0·41
0·86
0·39
0·38
0·18
0·56
0·50
0·08
0·05
Lu
0·07
0·07
0·14
0·05
0·06
0·03
0·09
0·08
0·02
Ba
1
7
Th
0·08
0·03
0·23
0·58
0·07
0·32
0·02
0·30
0·06
Nb
0·77
0·18
1·61
5·55
0·52
3·14
0·21
2·11
0·49
0·39
Y
5·52
3·43
12·77
7·45
3·25
3·03
5·13
6·51
0·41
0·34
Hf
0·47
0·19
0·98
1·41
0·30
0·82
0·21
0·94
0·31
0·26
Ta
0·06
0·01
0·11
0·38
0·03
0·21
0·01
0·14
0·03
0·03
U
0·02
0·01
0·44
0·24
0·10
0·18
0·17
0·61
0·09
0·10
Pb
0·11
0·10
0·52
0·68
0·18
1·17
0·72
2·49
0·26
Rb
0·9
0·7
Cs
0·65
0·61
0·23
0·57
1·31
0·66
Sr
8
2
4
7
1
2
Sc
19·2
16·5
10·8
13·6
17·6
10·7
Zr
16
32
51
10
30
6
91
17·9
96
16·3
225
49·4
*Total iron as FeO.
LOI, loss on ignition.
954
30
23·7
86
20·8
0·79
276
23·5
0·30
85
50·0
0·97
0·01
31
0·04
0·02
26·6
0·13
15
3
13
15·5
5·9
16·6
15·5
10
12
7
33
2
THOMPSON STIEGLER et al.
BARBERTON KOMATIITES
Table 2: Representative electron microprobe analyses of chromites from BGB komatiitic tuffs
Location:
Sample:
Weltevreden Fm.
M3c
M2c
M1c
K1c
H5c
H3c
H1
MSA
MSA
SAF
MSA
SAF
SA
SAF
SAF
SAF
8-1
31-3*
181-3*
16-4*
10-12*
25-5*
147-1*
281-2
499-4y
SiO2
0·06
0·08
0·11
0·05
0·08
0·12
0·09
0·06
0·06
MgO
6·96
15·16
16·40
12·54
14·18
14·56
8·81
8·89
12·00
8·25
Al2O3
5·23
9·29
11·16
12·80
11·80
14·05
9·37
16·20
NiO
0·07
0·13
0·14
0·10
0·14
0·17
0·11
0·16
0·10
FeO
20·76
9·59
8·06
14·34
11·72
11·34
19·71
20·03
14·37
6·43
Fe2O3
13·53
4·27
3·61
4·64
5·94
3·27
8·04
5·34
MnO
0·57
0·28
0·23
0·34
0·30
0·19
0·42
0·28
0·34
Cr2O3
50·33
59·45
58·58
54·14
54·36
53·46
50·50
45·41
56·27
TiO2
0·06
0·06
0·09
0·17
0·21
0·16
0·76
0·23
0·30
V2O3
0·01
0·05
0·01
0·00
0·06
0·16
0·10
0·17
0·08
Total
97·58
98·36
98·40
99·12
98·80
97·31
97·82
96·77
98·20
Cr-no.
0·866
0·811
0·779
0·739
0·755
0·718
0·783
0·653
0·821
Fe-no.
0·626
0·262
0·216
0·391
0·317
0·304
0·557
0·558
0·402
*Average of two analyses per grain.
yAverage of three analyses per grain.
All analyses were obtained within grain cores. Cr-number ¼ Cr/(Cr þ Al) and Fe-number ¼ Fe2þ/(Fe2þ þ Mg). Electronic
Appendix 2 contains the full dataset.
Table 3: Summary geochemical data for komatiitic tuffs of the Onverwacht Group
Formation:
Hooggenoeg
Kromberg
Mendon
Weltevreden
Member:
H1
H3
H5
K1
K2
M1
M2
M3
Type
AUK
AUK
ADK
AEK
ADK
AUK
ADK
ADK
ADK
AUK
Al2O3/TiO2
24–27
16–18,
10–12
44–65
5–6
26–29
3–4
10–11
9–15
19–33
29–32
(La/Sm)n
1·2–4·5
0·9–3·8
4·3–7·7
1·7–3·6
0·7–1·4
0·8–2·5
1·5–2·1
2·3–7·0
0·7–1·0
0·6–1·1
(Gd/Yb)n
1·2–2·8
0·5–2·1
1·5–2·1
0·4–0·6
2·2–2·5
0·9–2·8
2·8–3·6
1·0–1·7
1·4–1·8
0·7–1·3
Hf/Hf*
2·2–9·5
1·2–9·4
0·8–2·1
1·3–10
1·0–2·2
0·6–2·3
0·5–1·1
0·6–7·2
0·7–0·9
0·6–1·4
Zr/Zr*
2·0–12
1·0–8·7
0·8–2·1
1·2–9·8
0·9–2·2
0·5–2·1
0·5–1·1
0·6–7·0
0·6–0·9
0·5–1·1
Eu/Eu*
1·0–1·5
1·0–3·5
1·0–1·2
1·0–2·0
1·0–1·1
0·9–1·3
1·0
1·0–1·2
0·5–1·2
0·7–1·3
Ce/Ce*
0·2–0·8
0·4–1·0
0·4–0·7
0·6–0·9
0·9–1·0
0·6–1·0
1·0
0·1–0·7
0·9–1·0
0·9–1·1
Trace elements are normalized to primitive mantle values of Palme & O’Neill (2004). Zr/Zr*, Hf/Hf*, Eu/Eu*, and Ce/Ce*
are calculated with Zr* and Hf* ¼ (Nd þ Sm)n/2, Eu* ¼ (Sm þ Gd)n/2, and Ce* ¼ (La þ Pr)n/2.
predicted by their Al2O3/TiO2 ratios (Fig. 5a). Except for
La, their REE contents are anomalously low compared
with the HFSE. In contrast, most AUK tuffs in M1c possess flat, primitive mantle-like REE patterns and Hf/Hf*
and Zr/Zr* ratios near unity.
Al-depleted komatiitic tuffs occur as thin deposits in
the Hooggenoeg (H5c) and upper Mendon (M3c)
Formations. They are characterized by Al2O3/TiO2 ratios
between 10 and 12, moderately fractionated Gd/Ybn, and
enriched LREE (La/Smn ¼ 2·3^7·7). We distinguish between these tuffs and those with very low Al2O3/TiO2
ratios (3^6). The latter are almost exclusively lapilli tuffs
and occur in the Kromberg (K2) and Mendon (M2)
Formations. They have fractionated REE patterns
(La/Ybn ¼1· 6^8·5) and are enriched in Ti, Zr, and Hf
compared with other komatiites.
955
JOURNAL OF PETROLOGY
NUMBER 4
KROMBERG FORMATION
100
APRIL 2010
MENDON FORMATION
100
MSA 34-2
10
10
1
1
K2v
0.1
0.1
K2
0.01
100
Primitive mantle normalized
Primitive mantle normalized
VOLUME 51
10
1
0.1
K1v
K1
0.01
Rb Th La Pr Hf Sm Ti Tb Ho Er Yb
Ba Nb Ce Nd Zr Eu Gd Dy Y Tm Lu
M3
0.01
100
10
1
M2v
0.1
M2
0.01
100
10
HOOGGENOEG FORMATION
100
1
10
H5v
0.1
1
0.01
Primitive mantle normalized
M1
Rb Th La Pr Hf Sm Ti Tb Ho Er Yb
Ba Nb Ce Nd Zr Eu Gd Dy Y Tm Lu
0.1
0.01
100
M1v
H5
10
1
0.1
0.01
100
H3
10
Komati Fm.
1
0.1
0.01
H1
Rb Th La Pr Hf Sm Ti Tb Ho Er Yb
Ba Nb Ce Nd Zr Eu Gd Dy Y Tm Lu
Fig. 4. Primitive mantle normalized trace element patterns of komatiitic tuffs from the southern part of the Onverwacht Group. The dashed
grey lines in each section are patterns for the underlying komatiites, komatiitic basalts, or in H5, tholeiitic basalts. Data normalized using the
primitive mantle values of Palme & O’Neill (2004).
956
THOMPSON STIEGLER et al.
60
Al2O3 /TiO 2
Majorite
50 accumulation
12.00
SILICIFIED KOMATIITIC TUFFS
Mendon Fm.:
M1,
M2,
M3
Kromberg Fm.:
K1,
K2
H5
H3,
Hooggenoeg Fm.: H1,
SILICIFIED TUFFS
M1,
M2,
M3
K1,
K2
H5
H1,
H3,
10.00
BGB KOMATIITES
ADK
AUK
AEK
Al2O3 (wt.%)
70
BARBERTON KOMATIITES
40
30
SERPENTINIZED
TUFFS
8.00
6.00
4.00
20
20%
2.00
10
Majorite 50%
fractionation
(a)
0
0.2
0.6
1.0
35
1.4
1.8 2.2
(Gd/Yb)n
ADK
2.6
AEK
AUK
0.00
80%
0
3.0
3.4
3.8
7000
10
20
30
40
AUK
ADK
50
60
70
60
70
AEK
6000
SERPENTINIZED KOMATIITIC TUFFS
30
5000
Ti (ppm)
Al 2O3 /TiO 2
25
MSA 5-3
20
10%
4000
3000
15
SAF
AF 378-6
3
10
2000
30%
SA 700-1
1000
5
0
50%
Majorite fractionation
(b)
0.6
0
0
0.8
1.0
1.2
1.4 1.6
(Gd/Yb)n
1.8
2.0
2.2 2.4
Fig. 5. Variation in Al2O3/TiO2 and (Gd/Yb)n ratios as a function of
olivine and majorite removal for (a) silicified and (b) serpentinized
tuffs in the Onverwacht Group compared with Barberton komatiitic
flow rocks. Olivine fractionation results in no deviation from primitive mantle abundances. Partition coefficients for olivine are from
Kennedy et al. (1993) and those for majorite are from Yurimoto &
Ohtani (1992). Komatiite data are from sources listed in Table 4.
The lower Kromberg Formation (K1) contains the only
Al-enriched komatiitic tuffs in the Onverwacht Group.
They are characterized by high Al2O3/TiO2 ratios
(44^65) and sub-chondritic Gd/Ybn (0·4^0·6). Their high
Al2O3/TiO2 and low Ti/Zr ratios result from low TiO2
abundances (Fig. 6). With one exception, normalized REE
abundances are significantly lower than the HFSE.
Northern facies: Weltevreden Formation
The serpentinized tuffs in the Weltevreden Formation have
silica contents near komatiitic levels (47^55 wt %) and
10
20
30
40
Al2O3 /TiO 2
50
Fig. 6. Variation of Al2O3 (top) and Ti (bottom) as a function of
Al2O3/TiO2 for komatiitic tuffs. Tuffs with the lowest Al2O3/TiO2
ratios (M2) have very high Ti and low Al2O3 contents. In contrast,
Al2O3 and Ti contents in Al-enriched komatiitic (AEK) tuffs deviate
only moderately from primitive mantle values. The shaded area
shows the range of Al-undepleted komatiites (AUK) and the dashed
lines represent primitive mantle values.
komatiitic abundances of other elements, such as Ca, Fe,
Mg, Cr, and Ni, that have been mobilized in tuffs south of
the Inyoka Fault. MgO defines a strong negative linear
correlation with the incompatible elements TiO2, CaO
(Fig. 7), and Zr. The correlation of Cr and MgO is not
linear but is consistent with models of komatiite crystallization in which Cr behaves moderately incompatibly
until chromite crystallizes when Cr starts to decrease
(Murck & Campbell, 1986). Y is moderately correlated
with the immobile element Ti, indicating partial Y mobilization. Na2O, K2O, MnO, Sc, V, Rb, Sr, and Ba show
scatter when plotted against MgO and TiO2. Al2O3 is
957
JOURNAL OF PETROLOGY
1.20
VOLUME 51
R = 0.90
NUMBER 4
24.00
SERPENTINIZED ROCKS
BGB komatiites
Weltevreden Fm. tuffs
R = 0.80
1.00
20.00
0.80
16.00
0.60
12.00
0.40
8.00
0.20
4.00
0.00
0.00
R = 0.49
4000
R = 0.42
2500
3500
2000
3000
2500
1500
2000
1000
1500
1000
Ni (ppm)
Cr (ppm)
CaO (wt. %)
TiO2(wt. %)
APRIL 2010
500
500
0
0
0
10
20
30
40
50 0
MgO (w t. %)
10
20
30
40
50
MgO (w t. %)
Fig. 7. MgO variation diagrams for tuffs of the Weltevreden Formation compared with Barberton komatiitic flow rocks. The correlation coefficient, R, refers to the komatiitic tuffs.
poorly correlated with MgO, Ni, TiO2, and most elements
that typically define tight linear arrays in komatiites
(Fig. 8).
Despite this unusual behavior of Al2O3, major and trace
element ratios appear to represent derivation from both
Al-depleted and Al-undepleted komatiites. Al-undepleted
tuffs have Al2O3/TiO2 ¼19^33, Gd/Ybn ¼ 0·7 ^1·3,
depleted LREE (Fig. 9c; Table 3), and low CaO/Al2O3
(1·0). Although three tuffs have low Zr/Zr* (0·5^0·6)
and variable Eu anomalies (Fig. 9b), all other
Al-undepleted tuffs lack Hf^Zr anomalies and show no enrichment or depletion in Eu or Ce. Al-depleted komatiitic
tuffs have low Al2O3/TiO2 (9^15), depleted HREE, high
CaO/Al2O3 (2·2), and distinct negative Zr and Hf
anomalies (Fig. 9c). The one silicified accretionary
lapilli-bearing tuff has enriched LREE (La/Smn ¼ 2·7),
fractionated REE, a negative Nb anomaly, and lower
than primitive mantle Al2O3/TiO2 (15). A small group of
serpentinized tuffs, at least one of which contains carbonate, have HREE signatures inconsistent with their Al2O3/
TiO2 ratios and lie well outside the range typical for olivine or majorite fractionation (Fig. 5b). These rocks have
prominent low Hf/Hf*, Zr/Zr*, and Ce/Ce* ratios, and
varying Eu anomalies (Fig. 9d). The chemical composition
of these rocks is probably not primary and they are
excluded from the present study.
Chromites
Analyzed chromite cores vary on average by 50·5% Cr3þ,
1^2% Al3þ, Fe2þ, and Mg2þ, and 4% Fe3þ, indicating
little to no modification by metasomatic processes.
Cr-numbers are high and range from 0·61 to 0·92. Such
Cr-rich spinels are typical of komatiites and modern boninites, reflecting crystallization from primitive, high-degree
partial melts. It has been shown that in komatiites
Cr-number tends to correlate with bulk-rock composition
(Barnes & Roeder, 2001). ADK, which have high ratios of
Cr to Al, have high Cr-numbers compared with AUK,
which have elevated concentrations of Al. AEK contain
relatively Cr-poor spinels. Figure 10 displays the variation
in Cr- and Fe-numbers in chromites found in tuffs compared with those in the Barberton komatiites. It shows
that despite possessing near-uniform bulk immobile element ratios, most tuff units contain a wide range of spinel
compositions. Spinels, for example, from M2c
(Cr-number 0·69^0·89), M3c (Cr-number 0·61^0·86), and
the Weltevreden Formation (Cr-number 0·65^0·92) plot in
the fields for all three major types of komatiite.
958
THOMPSON STIEGLER et al.
MgO (w t. %)
50
BARBERTON KOMATIITES
Komatiitic tuff beds have a mineralogy dominated
by microcrystalline quartz sericite chlorite dolomite
or, in the Weltevreden Formation, talc þ chlorite þ
amphibole þ serpentine.
(a)
40
30
Southern facies: silicification and glass
alteration textures
20
Nearly all BGB volcaniclastic units south of the Inyoka
Fault are now cherts as a result of early, nearly syndepositional silicification after the particles settled into
silica-saturated Archean seawaters (Lowe & Byerly, 1986;
Lowe, 1999b; Knauth & Lowe, 2003). Silicification is
nearly ubiquitous in Archean sequences and it is thought
that Archean waters contained excess silica because of the
absence of silica-secreting organisms (Siever, 1992). In the
BGB, it also has been proposed that the silica was derived
from hydrothermal fluids that moved upwards through
fracture and vent systems and circulated through seafloor
sediments (Paris et al., 1985; Duchac & Hanor, 1987).
However, the early cementation, which has preserved the
three-dimensional shapes of uncompacted organic particles and the original porosities of fall-deposited accretionary lapilli, points to silicification within a few meters of
the seafloor (Lowe, 1999b). The stratiform and laterally extensive nature of the cherts indicates silicification across
enormous areas at very shallow burial depths through
rock^seawater interactions.
Slow suspension settling of the volcanic particles into
silica-rich seawater may have promoted surface adsorption
and the formation of a volcaniclastic^siliceous ooze at the
sediment^water interface (Rouchon & Orberger, 2008).
Silicification involved both direct precipitation in pore
spaces and dissolution^replacement reactions within
grains. Multiple episodes of silica deposition are recorded
in botryoidal banding in the microquartz matrix and in
the presence of cross-cutting quartz veins and fractures
(Lowe, 1999a). Si-rich fluids also penetrated the top
5^50 m of the underlying volcanic piles, converting pillows
and flow rocks to cherts, but leaving the bulk of single
volcanic sequences unsilicified (Lowe & Byerly, 1986).
Glass alteration probably accompanied the formation of
this grain-supporting siliceous ooze (Rouchon &
Orberger, 2008). Glass alteration rates are poorly constrained but their inherent thermodynamic instability
means that vitric debris rapidly alters to clays in most
modern surface environments (Fisher & Schmincke, 1984;
Stronick & Schmincke, 2002). In laboratory settings, it has
been demonstrated that alteration proceeds through hydration, dissolution, and the neoformation of secondary
phases (Crovisier et al., 2003). Palagonite is the first stable
product of basaltic glass transformation and forms rinds
on vitric surfaces exposed to aqueous fluids (Stronick &
Schmincke, 2002). In the komatiitic tuffs, different grain
replacement textures reflect various types of palagonization and silicification. The phyllosilicate-rimmed grains
10
BGB komatiites
Weltevreden tuffs
0
3000
(b)
2500
Ni (ppm)
2000
1500
1000
500
0
0.60
10
(c)
0
TiO2 (wt. %)
0.50
iO 2
/T
O3
l
2
A
0.40
=2
30
0.30
0.20
0.10
Weltevreden komatiites only
0.00
0
2
4
6
Al 2O3(w t. %)
8
10
Fig. 8. Al2O3 variation diagrams for Weltevreden tuffs. (a) MgO; (b)
Ni (ppm); (c) TiO2.
A LT E R AT I O N
One of the major difficulties encountered in studying the
petrogenesis of the Barberton volcanic rocks is the extensive alteration and metasomatism that they have undergone. Both tuffs and flow units have been subjected to
extensive metasomatism and lower greenschist-grade
metamorphism at maximum temperatures slightly above
3008C (Xie et al., 1997; Tice et al., 2004). Formerly
olivine-rich rocks are now typically serpentine þ chlorite
þ magnetite talc actinolite magnesite (Byerly, 1999).
Fresh magmatic olivine and pyroxene are rarely preserved.
959
JOURNAL OF PETROLOGY
10
MSA 22-1
MC 9-2
MSA 26-1
MSA 13-1
VOLUME 51
MSA 38-4
MSA 37-1
SAF 378-7
1
AL-UNDEPLETED TUFFS
Primitive mantle normalized
Primitive mantle normalized
APRIL 2010
10
MSA 23-1
MSA 35-1
MC 5-7
MSA 10-1
MSA 8-1
1
0.1
NUMBER 4
(a)
10
AL-DEPLETED TUFFS
0.1
AL-UNDEPLETED TUFFS
with large REE/HFSE anomalies
(b)
100
CARBONATIZED & SILICIFED TUFFS
10
SiO 2 = 88 wt%
1
1
MSA 5-5
MSA 25-1
SAF 477-10
MC 7-2
0.1
Th
La
Nb
Pr
Ce
Hf
Nd
MSA 14-1
MSA 30-1
MSA 5-8
(c)
SAF 390-1
SA 700-1
MSA 5-3
0.1
Sm Gd Dy Y Tm
Lu
Zr
Eu Tb Ho Er Yb
Th
La
Nb
Pr
Ce
SAF 378-6
Hf
Nd
(d)
Sm Gd Dy Y Tm
Lu
Zr
Eu Tb Ho Er Yb
Fig. 9. Primitive mantle normalized trace element patterns of komatiitic tuffs from the Weltevreden Formation: (a) Al-undepleted tuffs (Al2O3/
TiO2 ¼19^33); (b) Al-undepleted tuffs with enriched LREE and low Hf/Hf* ratios compared with the other Al-undepleted tuffs;
(c) Al-depleted tuffs (Al2O3/TiO2 ¼ 9^15); (d) carbonated and silicified tuffs. Data normalized using the primitive mantle values of Palme &
O’Neill (2004).
probably represent incipient palagonization, dissolution of
the rest of the glass, and subsequent replacement by silica.
We speculate that grains with an outer layer of silica represent early Si-adsorption onto particle surfaces and then
varying degrees of palagonization and/or silicification of
the interior.
REE mobilization
The process of silicification requires large chemical exchanges, usually involving the mobilization and/or dilution
of most of the original element abundances. In Barberton,
silicification involved the conversion of komatiitic flows to
a stable quartz þ mica þ chlorite assemblage, enriching
the rocks in Si, K, Ba, and Rb (Hanor & Duchac, 1990)
but not affecting their immobile incompatible element
ratios. Similar compositional changes were observed in
silicified komatiitic tuffs where ratios of immobile elements, Al2O3/TiO2, Ti/Zr, and Zr/Th, remain close to komatiitic levels. Some of these elements were retained
in alteration resistant phases, such as rutile and the
phyllosilicates that formed during glass dissolution.
Many silica-replaced tuffs also exhibit prominent and bizarre trace element anomalies in the form of unusually
low REE/HFSE ratios (Zr/Zr* and Hf/Hf* 1).
It is unlikely that these Zr^Hf enriched trace element
patterns result from partial melting or fractionation processes because: (1) Al-undepleted komatiites are erupted
with near-chondritic ratios of the REE and HFSE as a
result of the low partition coefficients of these elements in
olivine; (2) although Zr, Hf, and the HREE can vary in
ADK and AEK as a result of partial melting as they partition preferentially into the mantle minerals majorite and
perovskite, these high-pressure phases have high partition
coefficients for the HFSE relative to the REE (Kato et al.,
1988; Ohtani et al., 1989). Melts fractionated from a source
that retained either phase will have subchondritic Hf/Hf*
and Zr/Zr* ratios and therefore cannot account for the extremely high Hf/Hf* ratios in the tuffs. Instead, the positive Hf^Zr anomalies appear to reflect the preferential
mobilization of the REE compared with the HFSE.
Normalized trace element diagrams show nearly constant
HFSE concentrations within a unit whereas the REE
960
THOMPSON STIEGLER et al.
BARBERTON KOMATIITES
70
1.00
ADK
SILICIFIED KOMATIITIC TUFFS
Mendon Fm.:
M1,
M2,
Kromberg Fm.:
K1,
K2
Hooggenoeg Fm.: H1,
H3,
60
AUK
Majorite
accumulation
Al2 O3 /TiO2
50
Cr / (Cr+Al)
0.80
M3
H5
BGB KOMATIITES
ADK
AUK
AEK
40
30
20
10 Majorite 20%
fractionation 50%
80%
0
0.60
0
AEK
Welt. Fm.
M2c
H3c
M1c
H1
K1c
0.40
0
0.25
0.50
0.75
20
30
40
Zr (ppm)
50
60
70
Fig. 11. Al2O3/TiO2 vs Zr in silicified tuffs compared to Barberton
komatiitic flow rocks.
M3c
H5c
10
1
Fe 2+/ (Fe 2++Mg)
Fig. 10. Chromite compositions in komatiitic tuffs compared with
categories of chromites from Barberton komatiitic lavas (Byerly,
1999). AEK, Al-enriched komatiites; AUK, Al-undepleted komatiites;
ADK, Al-depleted komatiites.
often vary widely (Fig. 4). Zr abundances in the tuffs are
similar to those of Barberton komatiites (Fig. 11), suggesting that Zr and Hf (which has an identical charge and
similar radius and therefore behaves nearly the same)
have not been significantly mobilized. Selective loss of the
REE to produce the positive HFSE anomalies also probably resulted in the Eu anomalies, as high Hf/Hf* tends
to correlate with high Eu/Eu*.
The mobility of the REE during post-depositional alteration is known to be strongly dependent on the fluid/rock
ratio (Bau, 1991), which for these tuffs presumably was
highest during and immediately after deposition on the
seafloor. Rouchon & Orberger (2008) calculated a minimum fluid/rock ratio of 1·5 106 to maintain a low pH
(5·5) during Si^K metasomatism of the Msauli Chert.
However, neither phase dilution effects owing to silica precipitation nor incipient glass alteration can account for
the extremely low REE/HFSE ratios. The REE fit poorly
in the quartz structure and there is no systematic correlation of REE concentrations with SiO2 contents. Most
tuffs in the Mendon Formation have SiO2 490 wt % but
possess near primitive mantle REE/HFSE ratios. Silicified
komatiitic lavas also lack REE depletions and their Hf^Zr
anomalies appear to reflect those of their serpentinized
equivalents. Therefore, although the microquartz matrix
commonly accounts for 30^50 vol. % of accretionary
lapilli-bearing tuffs, the precipitation of interstitial silica is
not considered the major factor in generating the prominent low REE/HFSE ratios.
In modern volcanic deposits, the mobility of the REE
during glass hydration is highly dependent on the adsorption capacity of secondary minerals, usually zeolites,
smectites, or other clays. Both clays and zeolites have high
partition coefficients for the REE (Berger, 1992) and neither can fractionate the HFSE from the REE. Although
there probably was minor loss of the REE during alteration, because some of the large glassy grains are completely replaced by quartz without any trace of
phyllosilicates, these grains represent a volumetrically
small portion of the deposits. Their complete silicareplacement did not produce the observed order of magnitude depletion in REE and early glass alteration was probably not a significant factor in generating the low REE/
HFSE ratios.
The non-primitive mantle Hf/Hf* and Zr/Zr* ratios
probably reflect post-silicification fluid^rock interactions;
the REE in komatiitic rocks have been known to be disturbed during metamorphic and metasomatic alteration
(Arndt et al., 1989; Tourpin et al., 1991; Lahaye et al., 1995).
REE systematics are disturbed the strongest in the
Hooggenoeg and lower Kromberg Formations, whereas
the majority of tuffs in the Mendon Formation have negligible Hf^Zr anomalies. This decreased REE mobility
with stratigraphic height may reflect greater metasomatism of older formations as a result of proximity to
961
JOURNAL OF PETROLOGY
VOLUME 51
magma intrusions. Knauth & Lowe (2003) have shown
that increased fluid^rock interaction associated with the
emplacement of nearby 3·46^3·43 Ga tonalite^trondhjemite^granodiorite (TTG) plutons reset the 18O of cherts
in the Hooggenoeg Formation. A separate event occurred
between 3·41 and 3·33 Ga that involved the intrusion of
olivine-rich dikes, which invaded rocks within and below
the lower Kromberg Formation and appear to be associated with the eruption of the mafic volcaniclastic rocks
in K2 (Ransom et al., 1999). Fluids associated with both of
these intrusive events may have mobilized and depleted
the REE in tuffs below the middle Kromberg Formation.
Later local thermal or tectonic events may have disturbed
the REE in parts of the Mendon Formation.
Ce anomalies
Understanding the origin of Ce anomalies in Precambrian
sediments has become an essential factor in establishing
local and regional redox conditions. Ce exists dominantly
as a trivalent ion but in the presence of dissolved oxygen it
precipitates out as CeO2 or Ce(OH)4 (Kato et al., 2006).
This redox-controlled removal results in Ce-depleted
fluids, exemplified by modern seawater (Piepgras &
Jacobsen, 1992). Negative Ce anomalies in Archean chemical precipitates, particularly in banded iron formation,
have been interpreted to indicate locally oxidizing waters
(Kato et al., 2006). In the BGB, however, Tice & Lowe
(2006) reported a lack of Ce anomalies across a range of
open-marine depositional environments in the Buck Reef
Chert. This evidence, combined with the presence of primary siderite, led them to conclude that marine waters
were anoxic during the deposition of the 300 m of black
and banded cherts.
In komatiitic tuffs, the existence of distinct negative Ce
anomalies throughout the Onverwacht Group indicates
the occurrence of redox-related Ce fractionation.
However, caution must be used when relating present-day
Ce concentrations in the tuffs to those that existed during
deposition, as redox-sensitive elements are susceptible to alteration during metamorphism. Authigenic uranium
(Ua ¼ U ^ Th/3) is a proxy that, when employed in conjunction with cerium, can assist in constraining the
timing of Ce oxidation. Under reducing conditions, U4þ
forms an insoluble oxide, uranite, whereas in oxidizing environments, U6þ is soluble and highly mobile. Th is not
redox sensitive under surface conditions and is not fractionated from U by geological processes (Wignall & Myers,
1988). Excess U is thought to be introduced into sediments
by the reduction of dissolved U6þ complexes, leading to
Ua40. High Ua also can result from increased solubility
at low pH where U6þ is stabilized by complexation with
carbonate phases (Casas et al., 1998).
In reducing waters and sediments, Ce and Ua will
retain magmatic values, exemplified by the serpentinized
and carbonatized tuffs. In contrast, most silicified tuffs in
NUMBER 4
APRIL 2010
the BGB are enriched in Ua relative to the primitive
mantle but do not necessarily have Ce anomalies. Instead
of reflecting redox conditions, this increased U mobility
could have been favored by the slightly acidic conditions
of Si^K metasomatism (Rouchon & Orberger, 2008). In
tuffs with both negative Ce anomalies and Hf/Hf* near
unity, Ua is enriched by up to an order of magnitude.
These low Ce/Ce* values are moderately proportional to
high La/Smn ratios. The addition of LREE to induce negative Ce anomalies has been found in other Barberton
metasedimentary rocks (Hayashi et al., 2004; Rouchon &
Orberger, 2008). Hayashi et al. (2004) employed La^Sm
isotope systematics to show that the development of Ce depletion in cherts of the Fig Tree Group occurred after 1100
Ma. We suggest that the Ce fractionation in the tuffs probably did not take place during their formation but was the
result of the introduction of oxidized, LREE- and
U-enriched fluids during metamorphism. These fluids are
distinct from those that mobilized the rest of the REE
because Ce depletions do not correlate with high Hf/Hf*
ratios.
Northern facies: serpentinization
In the Weltevreden Formation, the present tuff mineralogy
includes phases typical of the serpentinization of ultramafic rocks. Unlike in the southern part of the belt, the absence
of Si-metasomatism means that absolute element abundances could reflect magmatic compositions, but only if
post-depositional fluid^rock interactions did not cause
extensive chemical perturbations. The hydration of ultramafic rocks involves the formation of serpentine as well as
hydrous minerals such as chlorite and amphibole from
pre-existing anhydrous and less-hydrous phases
(O’Hanley, 1996). This commonly occurs below the sediment^water interface. Kareem (2005) used the texture
and chemistry of alteration phases to conclude that heated
seawater was the only fluid responsible for altering and serpentinizing the komatiites in the north^central part of the
Weltevreden Formation. De Ronde & Kamo (2000), however, noted the presence of carbonate in mafic rocks in the
Weltevreden Formation and concluded that, at least locally, there was later overprinting of greenschist-grade metamorphism by CO2-bearing fluids. In the south^central
part of the formation, where most of the tuffs were collected, the existence of secondary carbonate in some of
the interbedded serpentinized komatiites and a number of
thick carbonate-rich alteration zones, generally marking
faults, also implies that CO2-bearing fluids flowed through
these rocks.
Most AUK tuffs in the Weltevreden Formation have flat
to slightly enriched HREE, chondritic Hf/Hf* and Zr/
Zr*, and lack Eu and Ce anomalies, suggesting olivine
fractionation of near primitive mantle partial melts that
had been subject to insignificant trace element disturbance
by post-eruptive alteration. The negative Hf^Zr anomalies
962
THOMPSON STIEGLER et al.
0.60
BARBERTON KOMATIITES
komatiites, and oxygen isotope data from the flows indicate
that rocks of the Weltevreden Formation were altered at
lower temperatures (140^3108C) than the rest of the BGB
(Kareem, 2005).
Although the bulk-rock element compositions of most of
the tuffs appear to be consistent with either enhanced olivine or garnet in the source residue, the incoherent Al2O3
trends remain troubling. Komatiitic lavas in the
Weltevreden Formation have been subject to 50^100% serpentinization yet retain tight Al2O3 and TiO2 correlations.
This suggests that Al2O3 probably did not become mobile
during serpentine formation. Instead, it is possible that
the variation in Al2O3 contents results from post-eruptive
sedimentary reworking. The slight LREE depletion of the
tuffs precludes the incorporation of Al-rich, felsic material
by aqueous currents. The komatiitic abundances and
ratios and the absence of non-volcanogenic detritus suggest
that the range of Al2O3 contents may represent the detrital
mixing of Al-depleted and Al-undepleted komatiitic tuffs
in varying proportions. This is supported by the presence
of diverse Al2O3/TiO2 ratios along strike in at least two of
the tuff units. Tuffs with Al2O3/TiO2 30 probably represent the AUK end-member source, as they define a fairly
linear array with the interbedded, geochemically similar
AUK flow rocks (Fig. 8c).
Olivine
fractionation
Majorite
fractionation
20%
Hf (ppm)
0.40
30%
0.20
Olivine
accumulation
WELTEVREDEN FM.
AUK tuffs
AUK
ADK tuffs
ADK*
* Komati & Mendon Fms.
0.00
4.00
Olivine
fractionation
Nd (ppm)
3.00
Majorite
fractionation
SAF 378-7
MSA 38-4
2.00
30%
1.00
MSA 37-1
Olivine
accumulation
0.00
0.00
2.00
4.00
6.00
8.00
Al2O3 (wt.%)
10.00
12.00
Fig. 12. Al2O3 vs Hf and Nd in tuffs of the Weltevreden Formation
compared with Al-depleted komatiites (ADK) and Al-undepleted komatiites (AUK). Olivine and majorite fractional crystallization
trends are calculated from primitive mantle values. The tick marks
represent 10% increments of crystal fractionation. Partition coefficients for olivine are from Kennedy et al. (1993). Majorite partition coefficients for Al2O3 and Hf in komatiites are from Yurimoto &
Ohtani (1992) and those for Nd in tholeiites are from Fujimaki et al.
(1984).
in three of the AUK tuffs (Fig. 9b) are probably due to rock
interactions with fluids in which the HFSE were stable
relative to the REE. Their Hf concentrations are similar
to those of other AUK and can be modeled by fractionation of olivine (Fig. 12a) whereas their Nd concentrations
plot well above the trend for olivine fractionation
(Fig. 12b). These samples were collected within a few
meters of diabase dikes and exhibit variable Eu and/or Ce
anomalies suggesting that these rocks have been subjected
to late-stage contact alteration.
In the ADK tuffs, the combination of low Al2O3/TiO2,
high Ti/Y, and fractionated HREE is indicative of residual
garnet in the source. Many samples have Eu anomalies
suggesting minor post-depositional element mobilization.
The komatiitic abundances and ratios in both AUK and
ADK tuffs are consistent with the persistence of fresh
olivine, pyroxene and chromite in the interbedded
R E L AT I O N S H I P T O B A R B E RT O N
KO M AT I I T E S
In the previous section we presented evidence that multiple
events have altered the primary composition of komatiitic
tuffs in the BGB. In the southern Onverwacht Group,
early tuff^seawater interactions deposited silica in the
intergranular pore spaces and converted the glassy pyroclasts to a combination of phyllosilicates and quartz. Later
circulating fluids, possibly associated with magma intrusion, resulted in REE and Ce depletions and Eu enrichments. Despite this extensive alteration, single silicified
pyroclastic units retain distinctive Al2O3/TiO2 ratios,
which are crucial for determining mantle source conditions (Herzberg, 1992). Although Al2O3/TiO2 ratios are
most reliable when based on mole proportions projected
from olivine compositions (Arndt, 2008), their variance is
low within single tuff units (2s 2·5), except for in K1,
which has TiO2 near detection limits in some samples. We
argue that these tight ratios in units across the belt are
not coincidental and that the Al2O3/TiO2 ratios in silicified tuffs have petrogenetic significance. North of the
Inyoka Fault, serpentinization appears to have preserved
most primary element abundances; however, the poor
microtextural preservation limits our understanding of
the tuffs as either discrete eruptive units or reworked heterogeneous mixtures. Therefore, the relationship of the
963
JOURNAL OF PETROLOGY
VOLUME 51
NUMBER 4
APRIL 2010
Table 4: Komatiites and komatiitic basalts of the Onverwacht Group
Formation:
Sandspruita Komatib,c,d
Member:
Type
Hooggenoege,f,g
H2
ADK
Al2O3/TiO2 3–6
H3
H4
ADK, AUK tholeiite ADK, AUK AUK
8–12,
—
8–13, 26
H5
Krombergg
Mendone,g
K1
M1
M2
M3
AEK
ADK
ADK ADK, AUK, AEK AUK, AEK
K2
tholeiite AEK ADK
22–27 —
57
15–18
6, 10–12 90–110 9–10,
14–16
Weltevredenh,i
10
M4
10, 23, 35,
21–34, 60
42–47, 59
References: aViljoen & Viljoen (1969); bParman et al. (2003); cChavagnac (2004); dLahaye et al. (1995); eByerly (1999);
f
Lowe & Byerly (1999); gthis study; hKareem (2005); iCooper (2008).
tuffs to a particular magma source or to the interbedded
lavas is also uncertain.
A comparison of Al2O3/TiO2 ratios shows that many of
the silicified komatiitic tuffs are unrelated to the adjacent
komatiitic flow rocks (see Table 4 for a list of the types of
komatiites and komatiitic basalts). Only in the Kromberg
and upper Mendon (M3) Formations do komatiitic tuffs
and underlying flow rocks share major element characteristics (Fig. 13). These tuffs may represent terminal activity
at the same vent that produced the lavas. Tuffs in the
Hooggenoeg and lower Mendon (M1, M2) formations
have Al2O3/TiO2 ratios distinct from the underlying and
overlying flow rocks. Neither olivine nor pyroxene fractionation can account for these differences, indicating that the
tuffs are not simply more evolved and therefore more
volatile-rich versions of the lava flows. Instead, both tuffs
and flows must be derived from separate mantle sources
or from the same source but through different partial melting conditions.
COM POSI T IONA L R A NGE OF
C H RO M I T E S
Chromites present within the komatiitic tuffs reflect a combination of primary crystallization within primitive melts,
entrainment of lithic debris during eruption, surface erosion of flow rocks, and possibly assimilation during
magma ascent. Tuff units in H1, H5, K1, and part of the
Weltevreden Formation contain chromites with compositions compatible with their bulk-rock Al2O3/TiO2 ratios.
These spinels are very small and could represent quench
crystals but, in H5 at least, the chromites are encased
within cuspate shards. This requires the chromites to have
formed before the liquid was chilled to glass, indicating
that the spinels had crystallized prior to eruption.
Chromite typically appears on the liquidus after olivine,
but early in the fractionation of komatiitic liquids, producing cotectic olivine and chromite phenocrysts. The most
magnesian komatiitic liquids will not contain chromite
(Barnes, 1998), but the liquids that produced the tuffs did
not have to be significantly evolved for chromite to have
crystallized prior to magma ejection.
Most chromites in tuffs of the Mendon and Weltevreden
Formations did not crystallize from liquids in equilibrium
with their bulk-rock immobile element ratios. In komatiites, spinel Cr-numbers generally exhibit little inter-flow
variation. A wide range of Cr-numbers within a single
flow has been suggested to result from interactions with
intercumulus liquid or surrounding minerals during cooling (Barnes, 1998). Cooling rates in the tuffs would have
been too high to allow significant post-eruptive modification, and there would have been no trapped liquid available to react with the chromites. These variations are also
not attributed to post-depositional alteration, as spinels
from silicified, carbonatized, and serpentinized komatiitic
flow rocks in the BGB define robust, uniform compositional populations.
In the Mendon tuffs, most spinels have relatively low
Cr-numbers indicating crystallization from AEK
magmas. In M1, chromites present in cross-laminated ash
potentially originate from erosion of the underlying AEK
flow rocks. Other spinels are discrete grains encased
within accretionary lapilli, indicating that the chromites
were present in the vent area. These grains could represent
surface weathering debris that was incorporated into the
eruption, implying either subaerial exposure of AEK flow
rocks near the eruptive center or deposition of minor
chromite-bearing AEK tuffs around the vent.
Alternatively, these low Cr-number spinels could be xenocrysts that were incorporated into the rising magma from
underlying flow rocks or from a subsurface chromite-rich
cumulate layer that was a residue of crystallization from
primitive magmas. The lack of evidence that komatiitic
flow rocks entrained spinels during ascent, however, suggests that either these spinels were not derived from assimilation or that the magmas that generated these tuffs more
readily assimilated material prior to eruption. In M2 and
M3 tuffs, most low-Cr-number spinels occur within lithic
964
BARBERTON KOMATIITES
LITHOLOGY
H6
KROMBERG FM.
HOOGGENOEG FM.
THOMPSON STIEGLER et al.
H5c
H5
Lapilli tuff
Felsic volcanic
& intrusive rock
Tuff
Black chert
Komatiitic
flow rock
Tholeiitic
flow rock
GEOCHEMISTRY
Komatiitic volcaniclastic rock
Komatiitic flow rock
K3
Fig Tree Group
MENDON FM.
H4
H3c
H3
K2
M4
M3c
M3
M2c
TiO 2 (wt.%)
1.0
M2c
0.8
M2v
M3v
0.6
1
0.4
M2
H2
0.5
kilometers
M1c
M3c
0.2
M1v
0.0
Sill
Dikes & sills (K2)
0
2
4
6
8
10
Al2O3 (wt.%)
km
km
0.1
K1
0
0
H1
Komati Fm.
0
K1c
0
20
Al2O3 /TiO2
40
0
20
40
60
Al2O3 /TiO2
M1c
M1
0
20
40
60
80
100
Al2O3 /TiO2
Fig. 13. Stratigraphic variation of Al2O3/TiO2 ratios in komatiitic tuffs and flow rocks in the southern Onverwacht Group. The shaded area is
the range of Al-undepleted komatiites. The inset shows rocks of the Mendon Formation, which exemplify the typical tight correlation of
Al2O3 and TiO2 within single units.
clasts, suggesting that these may be pieces of M1v caught
up in the explosion. This implies that either the flow rocks
in the lower Mendon Formation were present in the vent
area or other Al-enriched komatiites were exposed.
Tuffs in the Weltevreden Formation exhibit the largest
range in spinel Cr-number, with over 90% of spinels
having Cr-number 40·80. None of these spinels appear to
be encased in a texturally different matrix, suggesting
that they are not parts of lithic fragments. Two of the
three beds that contain chromites are cross-bedded, implying that their spinels could be detrital. However, because
primary deposit microtextures have been obliterated by
alteration, it is difficult to know with certainty whether
the spinels were erupted with the melt or are admixed detrital spinels from various komatiitic sources.
C RY S TA L L I Z AT I O N ,
A S S I M I L AT I O N , A N D M A G M A
ASCENT
Textural features, such as phenocryst content and particle
vesicularity, and the major and trace element
geochemistry of the Barberton komatiitic tuffs place constraints on the role and origin of volatiles involved in their
genesis.
Phenocryst content and mantle melting
Before relating crystallization to subsurface processes, it is
imperative to establish that the crystal content of these
tuffs does not reflect syn- or post-eruptive transport. In
many large eruptions, phenocrysts, which have higher densities than all but the most Fe-rich glasses, will settle out
of laterally expanding eruption clouds faster than vitric
particles. This has been documented in the 1980 Mt. St.
Helens fall deposit, where the proportion of lithic fragments and crystals relative to pumice and glass decreases
with increasing distance from the vent (Carey &
Sigurdsson, 1982). In the Barberton komatiitic tuffs, the
coarsest vitric particles ( 2·6 g/cm3) are 50·2 cm in
diameter, meaning that any olivines ( 3·3 g/cm3) and
pyroxenes ( 3·4 g/cm3) of equivalent density would
need to be 20^25% smaller than the coarsest glass particles. However, such fine grains would still be detectable,
as the morphology of glass shards typically remains resolvable above 100 mm. The presence of chromites in all of the
965
VOLUME 51
fine-grained tuffs and rare silica-replaced hexagonal forms
in H1c, H5c, and K1c supports the inference that crystals
were able to be retained into distal areas. We argue that if
there were abundant olivines and pyroxenes in the eruptions, then more crystals should be evident in the deposits.
The dearth of phenocrysts in the fine-grained tuffs is
more probably a reflection of the crystal-poor nature of
the erupted liquid.
In the lapilli tuffs, the morphology and size of the crystals have implications for the timing of crystallization. On
the basis of experimental runs on mafic and ultramafic
melts, it has been demonstrated that highly elongate and
skeletal olivines preferentially crystallize during rapid
cooling or from strongly undercooled liquids (Donaldson,
1976; Faure et al., 2003). When cooling rates are slowed
down or with only moderate degrees of undercooling, polyhedral olivines tend to form. The fine, elongate microphenocrysts and trapped liquid inclusions common in the
lapilli units suggests rapid cooling rates. This implies that
most crystallization took place upon eruption or, for the
moderately skeletal olivines, possibly as magma flowed
along conduits in cold rocks close to the surface. The
sparse, large phenocrysts in the Kromberg lapilli tuffs
point to minor crystallization in an insulated environment
prior to eruption. The difference between the crystallization of abundant microphenocrysts in the lapilli tuffs and
the aphyric fine-grained tuffs may simply reflect a
grain-size control. The larger lapilli would have experienced slower cooling rates in their interiors compared
with ash-size particles, resulting in crystallization.
Water-fluxed mantle melting, such as that which occurs
above subduction zones, has been proposed for the origin
of the komatiites in Barberton and in other greenstone
belts (e.g. Parman et al., 2001). At low pressures, hydrous
magmas become saturated with respect to water and volatiles are released into a fluid, forcing crystallization because the liquidus is raised. Arndt et al. (1998) showed that
a hydrous komatiitic magma, originally with 25^30 wt %
MgO, would erupt as a porphyritic lava containing
22^25% olivine phenocrysts. Boninites, for example, are
water-rich (up to 7 wt % H2O), high MgO arc-magmas
that erupt with abundant phenocrysts (Ohnenstetter &
Brown, 1995). Their glassy, porphyritic textures are at odds
with the phenocryst-poor nature of the Barberton komatiitic flow rocks and tuffs. Recently, Dann & Grove (2007)
suggested that some Barberton komatiites erupted with significant phenocryst loads but, as a result of a combination
of laminar flow and crystal setting during surface transport, the phenocrysts became concentrated in the lower
portions of the flow whereas the upper margins became
deprived of phenocrysts. This allowed for spinifex crystals
to nucleate and grow in the upper, now aphyric liquid.
However, if abundant phenocrysts were erupted with the
magma, they would be preserved in the pyroclastic
NUMBER 4
APRIL 2010
0.35
0.30
0.25
TiO2 (wt.%)
JOURNAL OF PETROLOGY
0.20
R = 0.96
0.15
0.10
WELTEVREDEN FORMATION
0.05
AUK tuffs
ADK tuffs
Liquid Comp. 1
Liquid Comp. 2
0.00
20.00
25.00
30.00
MgO (wt.%)
35.00
40.00
Fig. 14. MgO vs TiO2 variation for tuffs of the Weltevreden
Formation. Al-undepleted komatiitic (AUK) tuffs have MgO contents
that plot between those of the estimated source compositions of the
interbedded AUK flow rocks. Liquid compositions (Comp.) 1 and 2
are from Cooper (2008) and Kareem (2005), respectively.
deposits; however, only the lapilli tuffs in K1 and K2 contain 41% phenocrysts.
The high MgO contents of the serpentinized tuffs support the eruption of primitive liquids that had not undergone extensive subsurface fractional crystallization.
Figure 14 shows that the AUK tuffs lie on olivine control
lines between the two primary liquid estimates of 26 wt
% and 35 wt % MgO determined by Cooper (2008)
and Kareem (2005), respectively. These estimates are
based on olivine compositions and bulk-rock olivine control lines calculated for the parent magmas of komatiitic
flow rocks in two exposures of the Weltevreden Formation.
Olivine-driven fractional crystallization accounts for the
range of MgO contents in both the tuffs and flow rocks.
Although olivine control lines that result from different
degrees of partial melting in the petrogenesis of the parental magmas of the tuffs cannot be distinguished from
those of low-pressure fractionation processes, small crystals
would be expected in the erupted liquid in the case of the
latter. If the AUK tuffs were derived from the same source
as the AUK lava flows, which seems probable, then the
mineralogy, such as the presence of fresh pigeonite, and
the geochemistry of the flow rocks indicate near anhydrous
crystallization conditions (Cooper, 2008).
If Barberton komatiitic tuff parental magmas were generated during anhydrous melting and were quenched as
nearly aphyric liquids, the magmas must have erupted
either at temperatures above their liquidus or as
crystal-poor melts just below their liquidus. Dry magmas
ascend along an adiabat and have the potential to generate
superheat (Blatt & Tracy, 1996). This promotes
966
THOMPSON STIEGLER et al.
BARBERTON KOMATIITES
depolymerization of the melt structure and reduction in
the number of crystal nuclei, possibly accounting for the
scarcity of phenocrysts in some komatiite chill zones (e.g.
Lesher & Groves, 1986) and the near absence of crystals in
the tuffs. Alternatively, the presence of phenocrysts in
some komatiites and in the lapilli tuffs indicates that some
melts lost heat during ascent and were erupted at temperatures probably not far below their liquidus.
Vesiculation and assimilation
Elsewhere (Thompson Stiegler et al., 2010) we have argued
that vigorous magma interaction with external water was
the major driving force behind explosive fragmentation of
these komatiitic melts. External magma^water contact
can occur at any point during the vesiculation history of
magmas (Houghton & Wilson, 1989) and the ubiquity of
vesicles in these deposits indicates that these were not
strictly dry, but contained some volatiles. Although it has
been proposed that komatiitic magmas had moderate
CO2 contents (40·6 wt %) (Anderson, 1995), CO2 solubility is low and it exsolves from most magmas earlier than
H2O. It also possesses a higher density relative to H2O at
the same P^T conditions and generates lower mass flow
rates in the conduit (Papale & Polacci, 1999), which would
potentially inhibit magma explosivity. Water is considered
the most likely volatile phase in these komatiitic magmas.
Using H2O/La ratios of 500^1000 for the modern upper
mantle, Arndt et al. (1998) calculated 0·045^0·2 wt %
H2O present in komatiitic magmas. In the BGB, serpentinized komatiitic tuffs have on average 0·3 and 0·9 ppm La
for AUK and ADK, respectively (normalized to 25 wt %
MgO), which corresponds to 0·015^0·09 wt % H2O.
These values are lower than those Arndt et al. (1998) determined for komatiitic flow rocks as a result of the minor
mobility of La in the tuffs (the correlation coefficient for
TiO2/La is 0·90). Mid-ocean ridge basalts (MORB) and
ocean island basalts (OIB) have similar water contents.
Depleted MORB typically have 0·12 wt % H2O
(Sobolev & Chaussidon, 1996) and Hawaiian high-Mg
OIB have 0·4 wt % H2O (Dixon & Clague, 2001).
Exsolution of this small amount of water can produce a significant amount of moderately vesicular to bubble-rich
material (Mastin et al., 2004), particularly if the magma
had a limited residence time at shallow crustal levels, as
suggested by the primitive nature of the komatiitic liquids,
giving it little chance to degas prior to eruption.
Additionally, volatile-poor magmas do not become saturated with water until they have risen to very shallow
levels, limiting the time available for exsolution-induced
crystallization. There were probably no shallow crustal
reservoirs for some komatiitic magmas, making the depth
of volatile exsolution an important factor in the degree of
crystallization.
Small amounts of water may have been incorporated
during ascent through the crust as the high temperature
of komatiitic magmas would result in effective heat transfer
to the surrounding rocks. However, the probable high
ascent rates and the absence of shallow magma chambers
limit the time available for melt^country rock interactions.
The late introduction of water into dry magmas has been
postulated as a source for the low water contents
(0·2^0·8 wt %) in some komatiite melt inclusions
(McDonough & Danyushevsky, 1995; Shimizu et al., 1997)
as well as the driving force behind pyroclastic komatiitic
eruptions (Arndt et al., 1998; Capdevila et al., 1999). The
addition of water will initially depress the liquidus and
promote the dissolution of pre-existing crystals. If a substantial amount of water is assimilated, at low pressures
the melt will exceed the saturation of water, crystallize
and undergo potentially violent degassing. However,
minor amounts of dissolved water (a few tenths of a per
cent) can be assimilated without causing crystallization if
the process is rapid and kinetic effects do not allow for
crystal nucleation.
Hydrated or serpentinized basalts and komatiites are the
most likely sources of water assimilated into the parent
magmas of the tuffs. Assimilation does not require melting
and the serpentinites would probably only dehydrate,
resulting in dissolved water in the magma and a
difficult-to-digest, olivine-rich residue. The high to moderate water contents of both serpentine minerals
(13 wt %) and hydrothermally altered basalt
(1·5^4 wt % in amphiboles) means that little assimilation is required to increase the volatile content of the
magma by a few tenths of a per cent. Importantly, these
altered mafic and ultramafic rocks could provide water
without significantly altering the magma’s immobile element composition (Arndt et al., 1998), as geochemical evidence for contamination by sediments or upper crustal
material is scarce. Except for two samples (MSA 27-4 and
SAF 390-1), the tuffs have high Th/Nb (0·09^0·16) indicating little to no addition of felsic material (Thompson
Stiegler et al., 2008). The LREE depletion of tuffs in the
Weltevreden Formation also precludes significant felsic
contamination. Using assimilation^fractional crystallization (AFC) models (r ¼ 0·5) for a komatiite in the
Weltevreden Formation that is considered to approximate
a primary liquid composition (Kareem, 2005), assimilation
of even small amounts of the upper crust (42% for AUK
tuffs and 44% for ADK tuffs) produces unreasonably
high La/Smn.
B A R B E RT O N KO M AT I I T I C
I N T E RVA L S
We concur with previous workers who have attributed the
origin of the Barberton komatiites to the activity of
mantle plumes. The textures of the tuffs do not support
the explosive exsolution of water gained through hydrous
967
JOURNAL OF PETROLOGY
VOLUME 51
melting in subduction zones as advocated by Parman et al.
(1997). Instead, these predominantly aphyric deposits suggest the rapid ascent of near-anhydrous magma and, at
most, very limited assimilation of hydrated rock prior to
eruption. The primary geochemical signatures of the tuffs
are consistent with previously described classes of komatiites, revealing a remarkable diversity of komatiitic compositions in the BGB. This diversity is coupled with the long
time-scales of komatiitic volcanism recorded in the three
major intervals dominated by komatiitic rocks in the
Onverwacht Group. (1) The Komati-Hooggenoeg period
lasted for 10 Myr, with much of this time probably
recorded in the deposition and alteration of the five major
inter-flow sedimentary units in the Hooggenoeg
Formation. The Komati Formation lacks sedimentary
interbeds and surface alteration zones that might imply
pauses in volcanism, suggesting that its 3·5 km of komatiitic lavas may have erupted in as little as 105^106 years. The
2·3^2·8 km sequence of mafic^ultramafic rocks in the
Hooggenoeg Formation probably represents a slightly
longer interval, perhaps 3^5 Myr. (2) The 80 Myr duration Kromberg Formation contains 1·8 km of tholeiitic
and komatiitic lavas; however, a significant portion of this
interval is probably represented by up to 400 m of interflow
sedimentary cherts. (3) The 40^70 Myr duration Mendon
Formation consists of at least five inter-flow sedimentary
chert units that record breaks in effusive volcanism. This
punctuated stratigraphy and geochronology is consistent
with 107 years or more of magmatism, reflecting a collective eruption rate up to 3^4 orders of magnitude lower
than that of the Komati Formation. However, it is possible
that the eruption rates of single magmatic cycles were as
high as that of the Komati Formation if deposition of the
interflow sedimentary layers and underlying flow-top alteration zones represent most of the time.
The temporal and compositional variation recorded in
the eruption and deposition of the komatiitic lava flows
and pyroclastic tuffs in each of these intervals (Fig. 13)
could be accounted for by variations in plume^mantle dynamics. One possible explanation for the magmatism
within a single interval includes the partial melting of a
single mantle plume at different depths (Campbell et al.,
1989). The Komati^Hooggeneog sequence could have
been produced by initial high degrees of batch melting to
generate the ADK, leaving a residue of olivine, orthopyroxene, and garnet (Arndt, 2003). Continued melting at
decreasing pressure would lead to the elimination of
garnet and the production of AUK magmas. The initial
and largest volumes of magma generated would be the
ADK in the Komati Formation; subsequently, minor
amounts of AUK were erupted at slightly lower rates in
the Hooggenoeg Formation. This process is analogous to
the short (1^5 Myr), high-volume magmatic pulses that
characterize large igneous provinces (Bryan & Ernst,
NUMBER 4
APRIL 2010
2008). The initial, high-volume igneous pulses are typically
attributed to the arrival of a plume head at the base of the
lithosphere (Campbell, 2007). The plume then produces
smaller volume melts at lower emplacement rates, allowing
sediment accumulation in the pauses between flow events
(e.g. Hooper et al., 2007) and potentially producing a
sequence similar to that in the Hooggenoeg Formation.
Other options that could generate the various komatiite
compositions and durations in the Onverwacht Group include fractional melting of a heterogeneous mantle plume
source and/or melting within multiple, compositionally
distinct plumes, as the 40^80 Myr duration of volcanism
in the Kromberg and Mendon Formations could represent
extremely long-lived plume systems at fixed locations. As
for the contrast in the thicknesses of the komatiitic units,
it is possible that these differences also reflect proximity to
volcanic centers rather than a fundamental control by
source dynamics. In this case, the thinner flows of the
Hooggenoeg and Mendon Formations might represent
accumulation sites farther removed from their vents than
those for the very thick accumulations in the Komati and
Weltevreden Formations.
CONC LUSIONS
Metasomatic alteration, in particular the silicification that
affects most of the fine-grained volcaniclastic debris in the
southern part of the Onverwacht Group, complicates
direct interpretation of the petrogenesis of the komatiitic
tuffs based on elemental abundances. The addition of
SiO2 has diluted other elements and metasomatism has
depleted FeO, MgO, CaO, Na2O, Cr, Ni, V, and Sc. Later
fluid^rock interactions mobilized the REE, resulting in
non-magmatic Hf/Hf*, Zr/Zr*, Eu/Eu*, and Ce/Ce*
values. Al, Ti, and the HFSE are considered to be immobile because of their uniformity within single tuff units,
giving their ratios petrogenetic significance for magma
source composition and dynamics. A wide spectrum of komatiite compositions has been identified based on these immobile elements, including ultra-Al-depleted, Al-depleted
(ADK), Al-undepleted (AUK), and Al-enriched (AEK)
types.
In the northern part of the BGB, the Weltevreden
Formation contains relatively fresh, serpentinized komatiitic tuffs with high MgO (23^34 wt %) contents.
Elemental abundances are consistent with those reported
for ADK and AUK with possible mixing of the two types
by aqueous currents. The presence of hydrous mineral
phases in the secondary assemblage suggests that an
H2O-rich fluid, probably seawater, initially circulated
through the entire volcanic pile, driving serpentinization.
Subsequently, spatially restricted CO2-rich fluids resulted
in precipitation of diagenetic carbonate and preferential
mobilization of the REE in a small number of tuffs.
968
THOMPSON STIEGLER et al.
BARBERTON KOMATIITES
Although hydrous melting in subduction zones has been
suggested as an effective way of introducing water to komatiitic magmas to promote explosive eruptions, we find
little evidence to support this process for the generation of
these tuffs. Their vitric-rich and crystal-poor nature constrains the amount of water in the erupted magma and
implies the rapid quenching of essentially aphyric liquids.
These dense komatiitic magmas may have interacted with
hydrated mafic^ultramafic crust during ascent, but in
such a limited amount that the melts were able to continue
to rise nearly adiabatically.
FUNDING
This work was supported by National Aeronautics and
Space Administration (NASA) Exobiology Program
(NAG5-13442, NNG04GM43G) to D.R.L., by the
University of California Los Angeles NASA Astrobiology
Institute to G.R.B. and D.R.L., and by a Geological
Society of America student grant to M.T.S.
AC K N O W L E D G E M E N T S
The authors are grateful to the Mpumalanga Parks Board,
especially Louis Loock (Regional Manager) and
Property Mokowena, for allowing access to the
Songimvelo Game Reserve, and to Sappi Limited and
Martin Van Rensburg for permission to access private
forest roads. We thank Adina Paytan for use of her laboratory and equipment for spinel separations. Reviews by
Nick Arndt, Steve Barnes and Don Francis, and comments
by the editor Gerhard Wo«rner improved the quality and
clarity of the manuscript.
S U P P L E M E N TA RY DATA
Supplementary data for this paper are available at Journal
of Petrology online.
R EF ER ENC ES
Alle'gre, C. J. (1982). Genesis of Archaean komatiites in a wet ultramafic subducted plate. In: Arndt, N. T. & Nisbet, E. G. (eds)
Komatiites. London: Allen & Unwin, pp. 495^500.
Anderson, A. T. (1995). CO2 and the eruptibility of picrite and komatiite. Lithos 34, 19^25.
Anhaeusser, C. R. (1985). Archean layered ultramafic complexes in
the Barberton Mountain Land, South Africa. In: Ayres, L. D.,
Thurston, P. C., Card, K. D. & Weber, W. S. (eds) Evolution of
Archean Supracrustal Sequences. Geological Association of Canada, Special
Paper 28, 281^301.
Armstrong, R. A., Compston, W., de Wit, M. & Williams, I. S. (1990).
The stratigraphy of the 3·5^3·2 Ga Barberton Greenstone Belt
revisited: a single zircon ion microprobe study. Earth and Planetary
Science Letters 101, 90^106.
Arndt, N. (2003). Komatiites, kimberlites, and boninites. Journal of
Geophysical Research 108, 2293^2304.
Arndt, N., Teixeira, N. A. & White, W. M. (1989). Bizarre geochemistry of komatiites from the Crixas greenstone belt, Brazil.
Contributions to Mineralogy and Petrology 101, 187^197.
Arndt, N., Albarede, F. & Nisbet, E. G. (1997). Mafic and ultramafic
magmatism. In: de Wit, M. & Ashwal, L. D. (eds) Greenstone Belts.
Oxford: Clarendon Press, pp. 233^254.
Arndt, N. T. (2008). Komatiites. Cambridge: Cambridge University
Press, 467 p.
Arndt, N. T., Albarede, F., Cheadle, M. J., Ginibre, C., Herzberg, C.,
Jenner, G. et al. (1998). Were komatiites wet? Geology 26, 739^742.
Barley, M. E., Kerrich, R., Reudavy, I. & Xie, Q. (2000). Late
Archaean Ti-rich, Al-depleted komatiites and komatiitic volcaniclastic rocks from the Murchison Terrane in Western Australia.
AustralianJournal of Earth Sciences 47, 873^883.
Barnes, S. J. (1998). Chromite in komatiites, 1. Magmatic controls
on crystallization and composition. Journal of Petrology 39,
1689^1720.
Barnes, S.-J. & Often, M. (1990). Ti-rich komatiites from Northern
Norway. Contributions to Mineralogy and Petrology 105, 42^54.
Barnes, S. J. & Roeder, P. J. (2001). The range of spinel compositions in
terrestrial mafic and ultramafic rocks. Journal of Petrology 42,
2279^2302.
Bau, M. (1991). Rare-earth element mobility during hydrothermal and
metamorphic fluid^rock interaction and the significance of the oxidation state of europium. Chemical Geology 93, 219^230.
Beresford, S., Cas, R., Lambert, D. D. & Stone, W. E. (2000). Vesicles
in thick komatiite lava flows, Kambalda, Western Australia.
Journal of the Geological Society, London 157, 11^14.
Berger, G. (1992). Distribution of trace elements between clays and
zeolites and aqueous solutions similar to seawater. Applications to
Geochemistry. Special Issue 1çGeochemistry of Radioactive Waste
Disposal 193^204.
Blatt, H. & Tracy, R. J. (1996). Petrology: Igneous, Sedimentary, and
Metamorphic. New York: W. H. Freeman, 529 p.
Bryan, S. E. & Ernst, R. E. (2008). Revised definition of Large
Igneous Provinces (LIPs). Earth-Science Reviews 86, 175^202.
Byerly, G. (1999). Komatiites of the Mendon Formation: Late-stage
ultramafic volcanism in the Barberton Greenstone Belt. In:
Lowe, D. R. & Byerly, G. R. (eds) Geologic Evolution of the Barberton
Greenstone Belt, South Africa. Geological Society of America, Special Papers
329, 189^212.
Byerly, G., Kroner, A., Lowe, D., Todt, W. & Walsh, M. (1996).
Prolonged magmatism and time constraints for sediment deposition in the early Archean Barberton greenstone belt: evidence
from the Upper Onverwacht and Fig Tree Groups. Precambrian
Research 78, 125^138.
Byerly, G., Lowe, D., Wooden, J. L. & Xie, X. (2002). An Archean
impact layer from the Pilbara and Kaapvaal Cratons. Science 297,
1325^1327.
Campbell, I. H. (2007). Testing the plume theory. Chemical Geology 241,
153^176.
Campbell, I. H., Griffiths, R. W. & Hill, R.I. (1989). Melting in an
Archean mantle plume: heads it’s basalts, tails it’s komatiites.
Nature 339, 697^699.
Capdevila, R., Arndt, N., Letendre, J. & Sauvages, J.-F. (1999).
Diamonds in volcaniclastic komatiite from French Guiana. Nature
399, 456^458.
Carey, S. N. & Sigurdsson, H. (1982). Influence of particle aggregation
on deposition of distal tephra from the May 18, 1980, eruption of
Mount St. Helens Volcano. Journal of Geophysical Research 87,
7061^7072.
Casas, I., de Pablo, J., Gime¤nez, J., Torrero, M. E., Bruno, J., Cera, E.
et al. (1998). The role of pe, pH, and carbonate on the solubility of
969
JOURNAL OF PETROLOGY
VOLUME 51
UO2 and uraninite under nominally reducing conditions.
Geochimica et Cosmochimica Acta 62, 2223^2231.
Chavagnac, V. (2004). A geochemical and Nd isotopic study of
Barberton komatiites (South Africa): implication for the Archean
mantle. Lithos 75, 253^281.
Cooper, M. (2008). Komatiitic flows of the Pioneer Ultramafic
Complex of the 3·3 Ga Weltevreden Formation, Barberton
Greenstone Belt, South Africa: Stratigraphy, petrology, and geochemistry. M.S. thesis, Louisiana State University, Baton Rouge,
167 p.
Crovisier, J. L., Advocat, T. & Dussossoy, J. L. (2003). Nature and role
of alteration gels formed on surface of ancient volcanic glasses (natural analogs of waste containment glasses). Journal of Nuclear
Materials 321, 91^109.
Dann, J. C. (2000). The 3·5 Ga Komati Formation, Barberton
Greenstone Belt, South Africa, Part I: New maps and magmatic
architecture. South AfricanJournal of Geology 103, 47^68.
Dann, J. C. (2001). Vesicular komatiites, 3·5-Ga Komati Formation,
Barberton Greenstone Belt, South Africa: inflation of submarine
lavas and origin of spinifex zones. Bulletin of Volcanology 63, 462^481.
Dann, J. C. & Grove, T. L. (2007). Volcanology of the Barberton
Greenstone Belt, South Africa: inflation and evolution of flow
fields. In: van Kranendonk, M. J., Smithies, R. H. & Bennett, V.
C. (eds) Earth’s Oldest Rocks. Amsterdam: Elsevier, pp. 527^570.
de Ronde, C. E. & Kamo, S. (2000). An Archaean arc^arc collisional
event: A short-lived (ca 3 Myr) episode, Weltevreden area,
Barberton Greenstone Belt, South Africa. Journal of African Earth
Sciences 30, 219^248.
de Wit, M., Hart, R. A. & Hart, R. J. (1987). The Jamestown Ophiolite
Complex, Barberton mountain belt: a section through 3·5 Ga oceanic crust. Journal of African Earth Sciences 6, 681^730.
Dixon, J. E. & Clague, D. A. (2001). Volatiles in basaltic glasses from
Loihi Seamount, Hawaii: Evidence for a relatively dry plume component. Journal of Petrology 42, 627^654.
Donaldson, C. H. (1976). An experimental investigation of
olivine morphology. Contributions to Mineralogy and Petrology 57,
187^213.
Droop, G. T. R. (1987). A general equation for estimating Fe3þ concentrations in ferromagnesian silicates and oxides from microprobe
analyses, using stoichiometric criteria. Mineralogical Magazine 51,
431^435.
Duchac, K. & Hanor, J. (1987). Origin and timing of the
metasomatic silicification of an early Archean komatiite sequence,
Barberton Mountain Land, South Africa. Precambrian Research 37,
125^146.
Faure, F., Trolliard, G., Nicollet, C. & Montel, J.-M. (2003). A developmental model of olivine morphology as a function of the cooling
rate and the degree of undercooling. Contributions to Mineralogy and
Petrology 145, 251^263.
Fisher, R. V. & Schmincke, H. U. (1984). Pyroclastic Rocks. Berlin:
Springer, 472 p.
Fujimaki, H., Tatsumoto, M. & Aoki, K.-i. (1984). Partition coefficients
of Hf, Zr, and REE between phenocrysts and groundmasses.
Journal of Geophysical Research 89, 662^672.
Grove, T. L. & Parman, S. W. (2004). Thermal evolution of the Earth
as recorded by komatiites. Earth and Planetary Science Letters 219,
173^187.
Hanor, J. & Duchac, K. (1990). Isovolumetric silicification of early
Archean komatiites: Geochemical mass balances and constraints
on origin. Journal of Geology 98, 863^877.
Hanski, E., Huhma, H., Rastas, P. & Kamanetsky, V. S. (2001). The
Palaeoproterozoic komatiite^picrite association of Finnish
Lapland. Journal of Petrology 42, 855^876.
NUMBER 4
APRIL 2010
Hayashi, T., Tanimizu, M. & Tanaka, T. (2004). Origin of negative Ce
anomalies in Barberton sedimentary rocks, deduced from La^Ce
and Sm^Nd isotope systematics. Precambrian Research 135, 345^357.
Herzberg, C. (1992). Depth and degree of melting of komatiites.
Journal of Geophysical Research 97, 4521^4540.
Herzberg, C. & O’Hara, M. J. (1998). Phase equilibrium constraints
on the origin of basalts, picrites, and komatiites. Earth-Science
Reviews 44, 39^79.
Hollings, P. & Wyman, D. (1999). Trace element and Sm^Nd systematics of volcanic and intrusive rocks from the 3 Ga Lumby Lake
Greenstone belt, Superior Province: evidence for Archean plume^
arc interaction. Lithos 46, 189^213.
Hooper, P. R., Camp, V. C., Reidel, S. P. & Ross, M. E. (2007). The
origin of the Columbia River flood basalt province: Plume versus
nonplume models. In: Foulger, G. R. & Jurdy, D. M. (eds) Plates,
Plumes, and Planetary Processes. Geological Society of America, Special
Papers 430, 635^668.
Houghton, B. F. & Wilson, C. J. N. (1989). A vesicularity index for pyroclastic deposits. Bulletin of Volcanology 51, 451^462.
Jahn, B.-m., Gruau, G. & Glikson, A. Y. (1982). Komatiites of the
Onverwacht Group, S. Africa: REE geochemistry, Sm/Nd age and
mantle evolution. Contributions to Mineralogy and Petrology 80, 25^40.
Johnson, D. M., Hooper, P. R. & Conrey, R. M. (1999). XRF analysis
of rocks and minerals for major and trace elements on a single low
dilution li-tetraborate fused bead. Joint Committee on Powder
Diffraction StandardsçInternational Centre for Diffraction Data 41,
843^867.
Kareem, K. (2005). Komatiites of the Weltevreden Formation,
Barberton Greenstone Belt, South Africa: implications for the
chemistry and temperature of the Archean mantle. Ph.D. thesis,
Louisiana State University, Baton Rouge, 233 p.
Kato, T., Ringwood, A. E. & Irifune, T. (1988). Constraints on element
partition coefficients between MgSiO3 perovskite and liquid determined by direct measurements. Earth and Planetary Science Letters
90, 65^68.
Kato,Y., Yamaguchi, K. E. & Ohmoto, H. (2006). Rare earth elements
in Precambrian banded iron formations: Secular changes of Ce
and Eu anomalies and evolution of atmospheric oxygen. In:
Kesler, S. E. & Ohmoto, H. (eds) Evolution of Early Earth’s
Atmosphere, Hydrosphere, and BiosphereçConstraints from Ore Deposits.
Geological Society of America, Memoirs 198, 269^289.
Kennedy, A. K., Lofgren, G. E. & Wasserburg, G. J. (1993). An experimental study of trace-element partitioning between olivine, orthopyroxene and melt in chondrulesçequilibrium values and kinetic
effects. Earth and Planetary Science Letters 115, 177^195.
Knaack, C., Cornelius, S. B. & Hooper, P. R. (1994). Trace Element
Analyses of Rocks and Minerals by ICP-MS. Pullman: Washington
State University.
Knauth, L. P. & Lowe, D. (2003). High Archean climatic temperature
inferred from oxygen isotope geochemistry of cherts in the 3·5 Ga
Swaziland Supergroup, South Africa. Geological Society of America
Bulletin 115, 566^580.
Kroner, A., Byerly, G. & Lowe, D. (1991). Geochronology of early
Archean granite^greenstone evolution in the Barberton Mountain
land, South Africa, based on precise dating by single zircon evaporation. Earth and Planetary Science Letters 103, 41^54.
Lahaye, Y., Arndt, N., Byerly, G., Chauvel, C., Fourcade, S. &
Gruau, G. (1995). The influence of alteration on trace-element
and Nd isotope composition of komatiites. Chemical Geology 126,
43^64.
Lanier, W. P. & Lowe, D. (1982). Sedimentology of the Middle Marker
(3·4 Ga), Onverwacht Group, Transvaal, South Africa. Precambrian
Research 18, 237^260.
970
THOMPSON STIEGLER et al.
BARBERTON KOMATIITES
Lesher, C. M. & Groves, D. I. (1986). Controls on the formation of
komatiite-associated
nickel^copper
sulfide
deposits.
In:
Friedrich, G. H. (ed.) Geology and Metallogeny of Copper Deposits.
Special Publication for the Society for Geology Applied to Mineral Deposits
4, 43^62.
Lowe, D. (1999a). Geologic evolution of the Barberton Greenstone
Belt and vicinity. In: Lowe, D. R. & Byerly, G. R. (eds) Geologic
Evolution of the Barberton Greenstone Belt, South Africa. Geological Society
of America, Special Papers 329, 287^312.
Lowe, D. (1999b). Petrology and sedimentology of cherts and related
silicified sedimentary rocks in the Swaziland Supergroup. In:
Lowe, D. R. & Byerly, G. R. (eds) Geologic Evolution of the Barberton
Greenstone Belt, South Africa. Geological Society of America, Special Papers
329, 83^114.
Lowe, D. (1999c). Shallow-water sedimentation of accretionary lapillibearing strata of the Msauli Chert: Evidence of explosive hydromagmatic komatiitic volcanism. In: Lowe, D. R. & Byerly, G. R.
(eds) Geologic Evolution of the Barberton Greenstone Belt, South Africa.
Geological Society of America, Special Papers 329, 213^232.
Lowe, D. & Byerly, G. (1986). Archaean flow-top alteration zones
formed initially in a low-temperature sulphate-rich environment.
Nature 324, 245^248.
Lowe, D. & Byerly, G. (1999). Stratigraphy of the west^central part
of the Barberton Greenstone Belt, South Africa. In: Lowe, D. R.
& Byerly, G. R. (eds) Geologic Evolution of the Barberton
Greenstone Belt, South Africa. Geological Society of America, Special Papers
329, 1^36.
Lowe, D. & Byerly, G. (2007). An overview of the geology of the
Barberton Greenstone Belt and vicinity: implications for early crustal development. In: van Kranendonk, M. J., Smithies, R. H. &
Bennett, V. C. (eds) Earth’s Earliest Rocks. Developments in
Precambrian Geology. Amsterdam: Elsevier, pp. 481^526.
Lowe, D. & Knauth, L. P. (1978). The oldest marine carbonate ooids
reinterpreted as volcanic accretionary lapilli, Onverwacht Group,
South Africa. Journal of Sedimentary Petrology 48, 709^722.
Lowe, D. & Worrell, G. F. (1999). Sedimentology, mineralogy, and
implications of silicified evaporites in the Kromberg Formation,
Barberton Greenstone Belt, South Africa. In: Lowe, D. R. &
Byerly, G. R. (eds) Geologic Evolution of the Barberton Greenstone
Belt, South Africa. Geological Society of America, Special Papers 329,
167^188.
Mastin, L. G., Christiansen, R. L., Thornber, C., Lowenstern, J. &
Beeson, M. (2004). What makes hydromagmatic eruptions violent?
Some insights from the Keanakako’i Ash, Kilauea Volcano,
Hawai’i. Journal of Volcanology and Geothermal Research 137, 15^31.
McDonough, W. F. & Danyushevsky, L. V. (1995). Water and sulfur
contents of melt inclusions from Archean komatiites. EOS
Transactions, American Geophysical Union 76, 266.
Murck, B. W. & Campbell, I. H. (1986). The effects of temperature,
oxygen fugacity and melt composition on the behaviour of chromium in basic and ultrabasic melts. Geochimica et Cosmochimica Acta
50, 1871^1887.
Nisbet, E. G., Cheadle, M., Arndt, N. & Bickle, M. J. (1993).
Constraining the potential temperature of the Archaean mantle:
A review of the evidence from komatiites. Lithos 30, 291^307.
O’Hanley, D. S. (1996). Serpentinites: Records of Tectonic and Petrological
History. Oxford: Oxford University Press, 277 p.
Ohnenstetter, D. & Brown, W. L. (1995). Compositional variation and
primary water contents of differentiated interstitial and included
glasses in boninites. Contributions to Mineralogy and Petrology 123,
117^137.
Ohtani, E., Kawabe, I., Moriyama, J. & Nagata, Y. (1989).
Partitioning of elements between majorite garnet and melt and
implications for petrogenesis of komatiite. Contributions to
Mineralogy and Petrology 103, 263^269.
Palme, H. & O’Neill, H. S. C. (2004). Cosmochemical estimates of
mantle composition. In: Holland, H. D. & Turekian, K. K. (eds)
Treatise on Geochemistry. Oxford: Elsevier, pp. 1^38.
Papale, P. & Polacci, M. (1999). Role of carbon dioxide in the
dynamics of magma ascent in explosive eruptions. Bulletin of
Volcanology 60, 583^594.
Paris, I., Stanistreet, I. G. & Hughes, M. J. (1985). Cherts of the
Barberton Greenstone Belt interpreted as products of submarine
exhalative activity. Journal of Geology 93, 111^129.
Parman, S. W., Dann, J. C., Grove, T. L. & de Wit, M. (1997).
Emplacement conditions of komatiite magmas from the 3·49 Ga
Komati Formation, Barberton Greenstone Belt, South Africa.
Earth and Planetary Science Letters 150, 303^323.
Parman, S. W., Grove, T. L. & Dann, J. C. (2001). The production of
Barberton komatiites in an Archean subduction zone. Geophysical
Research Letters 28, 2513^2516.
Parman, S. W., Shimizu, N., Grove, T. L. & Dann, J. C. (2003).
Constraints
on
the
pre-metamorphic
trace
element
composition of Barberton komatiites from ion probe analyses
of preserved clinopyroxene. Contributions to Mineralogy and Petrology
144, 383^396.
Piepgras, D. & Jacobsen, S. B. (1992). The behavior of rare earth elements in seawater: Precise determination of variations in the
North Pacific water column. Geochimica et Cosmochimica Acta 56,
1851^8162.
Ransom, B., Byerly, G. & Lowe, D. (1999). Subaqueous to subaerial
Archean ultramafic phreatomagmatic volcanism, Kromberg
Formation, Barberton Greenstone Belt, South Africa. In: Lowe, D.
R. & Byerly, G. R. (eds) Geologic Evolution of the Barberton Greenstone
Belt, South Africa. Geological Society of America, Special Papers 329,
151^166.
Renner, R., Nisbet, E. G., Cheadle, M. J., Arndt, N. T., Bickle, M. J.
& Cameron, W. E. (1994). Komatiite flows from the Reliance
Formation, Belingwe Belt, Zimbabwe: I. Petrography and mineralogy. Journal of Petrology 35, 361^400.
Rouchon, V. & Orberger, B. (2008). Origin and mechanisms of
K^Si-metasomatism of ca. 3·4^3·3 Ga volcaniclastic deposits and
implications for Archean seawater evolution: Examples from
cherts of Kittys Gap (Pilbara craton, Australia) and Msauli
(Barberton Greenstone Belt, South Africa). Precambrian Research
165, 169^189.
Saverikko, M. (1985). The pyroclastic komatiite complex at
Sattasvaara in Northern Finland. Bulletin of the Geologic Society of
Finland 57, 55^87.
Shimizu, K., Komiya, S., Maruyama, S. & Hirose, K. (1997). Water
content of melt inclusion in Cr-spinel of 2·7 Ga komatiite from
Belingwe Greenstone Belt, Zimbabwe. EOS Transactions, American
Geophysical Union 78, 750.
Siever, R. (1992). The silica cycle in the Precambrian. Geochimica et
Cosmochimica Acta 56, 3265^3272.
Sobolev, A. & Chaussidon, M. (1996). H2O concentrations in primary
melts from supra-subduction zones and mid-ocean ridges: implications for H2O storage and recycling in the mantle. Earth and
Planetary Science Letters 137, 45^55.
Sproule, R. A., Lesher, C. M., Ayer, J. A., Thurston, P. C. &
Herzberg, C. (2002). Spatial and temporal variations in the geochemistry of komatiites and komatiitic basalts in the Abitibi greenstone belt. Precambrian Research 115, 153^186.
Stone, W. E., Deloule, E., Larson, M. S. & Lesher, C. M. (1997).
Evidence for hydrous high-MgO melts in the Precambrian. Geology
25, 143^146.
971
JOURNAL OF PETROLOGY
VOLUME 51
Stronick, N. A. & Schmincke, H. U. (2002). Palagoniteça review.
International Journal of Earth Sciences 91, 680^697.
Thompson Stiegler, M., Lowe, D. R. & Byerly, G. R. (2008).
Abundant pyroclastic komatiitic volcanism in the 3·5^3·2 Ga
Barberton greenstone belt, South Africa. Geology 36, 779^782.
Thompson Stiegler, M., Lowe, D. R. & Byerly, G. R. (2010).
Fragmentation and dispersal of komatiitic pyroclasts in the 3·5^
3·2 Ga Onverwacht Group, Barberton greenstone belt, South
Africa. Geological Society of America Bulletin (in press).
Tice, M. M. & Lowe, D. (2006). Hydrogen-based carbon fixation in
the earliest known photosynthetic organisms. Geology 34, 37^40.
Tice, M. M., Bostick, B. C. & Lowe, D. (2004). Thermal history of the
3·5^3·2 Ga Onverwacht and Fig Tree Groups, Barberton greenstone belt, South Africa, inferred by Raman microspectroscopy of
carbonaceous material. Geology 32, 37^40.
Tourpin, S., Gruau, G., Blais, S. & Fourcade, S. (1991). Resetting of
REE, and Nd and Sr isotopes during carbonitization of a komatiite
flow from Finland. Chemical Geology 90, 15^29.
NUMBER 4
APRIL 2010
Viljoen, M. J. & Viljoen, R. P. (1969). The geology and geochemistry
of the lower ultramafic unit of the Onverwacht Group and a proposed new class of igneous rocks. Geological Society of South Africa,
Special Publication 2, 55^86.
Wignall, P. B. & Myers, K. J. (1988). Interpreting benthic oxygen
levels in mudrocks: A new approach. Geology 16, 452^455.
Williams, D. A. C. & Furnell, R. G. (1979). Reassessment of part
of the Barberton type area, South Africa. Precambrian Research 9,
325^347.
Xie, X., Byerly, G. & Ferrell, R. (1997). IIb trioctahedral chlorite from
the Barberton Greenstone Belt: crystal structure and rock composition constraints with implications for geothermometry.
Contributions to Mineralogy and Petrology 126, 275^291.
Yurimoto, H. & Ohtani, E. (1992). Element partitioning between
majorite and liquidça secondary ion mass-spectrometric study.
Geophysical Research Letters 19, 17^20.
972