Content uploaded by Frans Rietmeijer
Author content
All content in this area was uploaded by Frans Rietmeijer on Mar 17, 2016
Content may be subject to copyright.
THE ASTROPHYSICAL JOURNAL, 527:395È404, 1999 December 10
1999. The American Astronomical Society. All rights reserved. Printed in U.S.A.(
METASTABLE EUTECTIC CONDENSATION IN A Mg-Fe-SiO- VAPOR: ANALOGS TOH2-O2
CIRCUMSTELLAR DUST
FRANS J. M. RIETMEIJER
Institute of Meteoritics, Department of Earth and Planetary Sciences, University of New Mexico, Albuquerque, NM 87131; fransjmr=unm.edu
JOSEPH A. NUTH III
Astrochemistry Branch, Code 691, NASA Goddard Space Flight Center, Greenbelt, MD 20771; uljan=lepvax.gsfc.nasa.gov
AND
JAMES M. KARNER
Department of Earth and Planetary Sciences, University of New Mexico, Albuquerque, NM 87131
Received 1999 April 30; accepted1999 July 26
ABSTRACT
Experimental studies of gas-to-solid condensation in a vapor reveal that thisFe-Mg-SiO-H2-O2
process yields only solids with magnesiosilica and ferrosilica compositions(MgO.SiO2) (Fe-oxide.SiO2)
that coincide with metastable eutectics in the and binary phase diagramsMgO-SiO2(FeO/Fe2O3)-SiO2
plus simple metal oxides (MgO, and FeO or No solids form with mixed Mg-Fe-O com-SiO2,Fe
2O3).
positions during condensation nor is there evidence for the formation of ferromagnesiosilica,
solids. The experimental evidence demonstrates that condensation of multicomponentMgO.FeyOx.SiO2
vapors yields only a limited number of metastable solids of well-deÐned composition. These results have
interesting consequences for models of grain formation in circumstellar outÑows, for predictions concern-
ing the chemical and mineralogical composition of presolar silicates, and for the composition of conden-
sates formed in protostellar systems.
Subject headings:circumstellar matter È dust, extinction È ISM : abundances È molecular processes È
solar system: formation
1.INTRODUCTION
Dust grains are observed to form in the winds of oxygen-
rich asymptotic giant branch (AGB) stars, and it has been
known for more than three decades that the infrared spectra
of such materials are dominated by amorphous silicates
(Woolf & Ney 1969). The exact nature of such condensates
has been a matter of considerable debate owing to the
broad nature of the dominant 10 and 20 km features
observed in most stellar sources and the difficulty of Ðtting
such features with pure mineral components. Spectroscopic
evidence obtained by the Infrared Space Observatory (ISO)
has dramatically changed this situation (Waelkins et al.
1996; Waters et al. 1996) with the discovery of narrow emis-
sion features at longer wavelengths due to the presence of
crystalline phases, i.e., silicate minerals.
Two di†erent aspects of these observations are intriguing.
First, the crystalline olivine and enstatite observed by ISO
are nearly pure magnesium silicate grains. No evidence for
the presence of iron-rich silicates has been obtained to date,
and the observations indicate that the crystalline grains so
far observed contain little, if any, iron. Second, observations
of crystalline silicates have been conÐned to only the highest
mass loss rate systems (Cami et al. 1998). In general, low
mass loss rate stars produce only amorphous silicates. Even
high mass loss rate systems produce primarily amorphous
grains: less than D10% of the dust in such outÑows is in the
form of crystalline minerals (Tielens et al. 1998). In contrast
to the expectation that more-crystalline materials would be
observed in the highest temperature shells where thermal
annealing could produce crystalline minerals from initially
amorphous condensates, the observations indicate that
almost the opposite could be true. Cooler shells appear to
contain higher fractions of crystalline grains (Tielens et al.
1998).
Sogawa & Kozasa (1999) have demonstrated that it is
possible to produce crystalline silicate grains in high mass
loss rate AGB outÑows by the heterogeneous condensation
of silicate mantles onto previously condensed alumina cores
followed by thermal annealing in the outÑow. Their model
cannot account for the production of nearly pure magne-
sium silicate minerals. Heterogeneous condensation of a
magnesium-iron-silica vapor onto preexisting alumina sur-
faces should trap at least some of the iron that would then
become incorporated into the crystalline silicate phases
formed by annealing in the outÑow. Similarly, Hallenbeck,
Nuth, & Daukantes (1998) and Hallenbeck, Nuth, &
Nelson (2000) have demonstrated that amorphous magne-
sium silicate grains anneal on much more rapid timescales
than do amorphous iron silicates. However, until the
experiments reported in this paper, there had been no
reason to postulate the existence of separate populations of
iron-rich and magnesium-rich silicates in a typical AGB
outÑow.
Theoretical studies of grain formation are ultimately
based on the attainment of thermodynamic equilibrium and
consider only the formation of stoichiometric crystalline
silicate minerals as part of a fractional condensation
sequence (Grossman & Larimer 1974; Lattimer & Gross-
man 1978; Gail 1998). Even models based on nucleation
theory (e.g., Kozasa & Sogawa 1997; 1998) represent a judi-
cious combination of kinetics and chemical thermodyna-
mics. Previous vapor phase condensation experiments
o†ered strong arguments against the likelihood of attaining
equilibrium mineral assemblages even under controlled
conditions in the terrestrial laboratory (e.g., De 1979; Donn
1979; Nuth 1996). Yet, equilibrium models remain central
to our intuitive understanding of circumstellar dust forma-
tion.
395
396 RIETMEIJER, NUTH, & KARNER Vol. 527
To some extent this situation persists even more strongly
in the meteoritics community. In the past, analytical tech-
niques were unable to probe meteorites and experimentally
produced dust analogs at the nanometer scale. Adequate
analytical techniques are now available (Sutton 1994). Since
1974 March interplanetary dust particles (IDPs) have been
routinely collected in the EarthÏs lower stratosphere
(Brownlee, Tomandl, & Hodge 1976; Warren & Zolensky
1994). These particles are a new type of extraterrestrial
material that is available for laboratory studies in addition
to the meteorites (Mackinnon et al. 1982). The mineralogy
and texture (Mackinnon & Rietmeijer 1987), and the high
carbon (Thomas, Keller, & McKay 1996) and volatile
element (Flynn et al. 1996) abundances, suggest that the
IDPs are the least modiÐed materials in the solar system.
This zodiacal dust includes a sizeable fraction of debris
from Oort cloud and Kuiper belt comets as well as dust
from carbon- and ice-rich asteroids in the outer asteroid
belt (Brownlee 1994). Such materials are our best hope of
attaining a relatively unbiased sample of solids from the
protosolar nebula.
2.PRINCIPAL COMPONENTS AND SYNTHETIC ANALOGS
UltraÐne-grained chondritic aggregate IDPs (5È25 kmin
size) have a matrix with variable proportions of embedded
Mg, Fe-olivine, (Ca, Mg, Fe)-pyroxene, Ni-free and low-Ni
pyrrhotite, pentlandite, refractory Al,Ti-oxides, silicates and
Fe-oxide grains, several micrometers in size (Rietmeijer
1998a). This matrix consists of three principal components
(PCs) that are similar in size and composition to dust near
the nucleus of comet Halley (Table 1). The carbon-free
PCs are (1) coarse-grained smectite dehydroxylate,
(Mg, Fe) units with Fe/(Fe]Mg) ( fe)\0È0.36
6Si8O22,
(element ratio), and (2) ultraÐne-grained units with a
serpentine dehydroxylate, (Mg, Fe) composition
3Si2O7,
and fe \0.36È0.83. The minerals and texture of ultraÐne-
grained PCs are similar to the ferromagnesiosilica
(Brownlee, Joswiak, & Bradley 1999) units of glass with
embedded metals and sulÐdes (GEMS) (Bradley 1994;
Bradley, Humecki, & Germani 1992). The chemical com-
positions of these PCs occur with rational atomic propor-
tions but do not match the stoichiometric ratios of
crystalline solids in which the three-dimensional arrange-
ments of tetrahedra control the crystallochemical pro-SiO4
portions of metals in the crystal lattice. The high degree of
compositional order is one of the most remarkable proper-
ties of chondritic aggregate IDPs, but the origin of crys-
talline phases in these IDPs remains undetermined.
Although the original dust accretion textures in these IDPs
do not appear to have been signiÐcantly modiÐed, such
appearances might be deceiving. The observed properties of
the PCs are the ultimate record of many possible processes:
(1) aggregation of unaltered interstellar dust, (2) conden-
sation in the solar nebula, (3) parent body alteration, (4)
thermal modiÐcation of the dust during solar system
sojourn, and (5) perihelion thermal processing in comet
nuclei. Atmospheric entry Ñash-heating of IDPs may seri-
ously modify their pre-entry properties, albeit in a recog-
nizable manner (Rietmeijer 1996).
Given the current level of understanding of chondritic
IDPs, ferromagnesiosilica PCs may represent surviving pre-
solar dust that accreted 4.56 Gyrs ago. This dust experi-
enced thermal modiÐcation at some time during its long
history. Its main presolar characteristic is most likely to be
its bulk composition rather than its mineralogy (Rietmeijer
1998a). Thus, the chemical composition of such dust may be
the only constraint on the origin of this material. It is not
evident whether originally amorphous ferromagnesiosilica
TABLE 1
PRINCIPAL COMPONENTS IN CHONDRITIC AGGREGATE IDPSaAND COMET P/HALLEYb
Principal Components in Chondritic Aggregate IDPs Dust Particles in Comet P/Halley
Carbonaceous Units
Refractory hydrocarbons and amorphous carbon units that are often fused in CHON particles (carbon, hydrogen, oxygen, nitrogen)
contiguous patches and clumps of vesicular, poorly graphitized and
pregraphitic carbons. Units are 400È4000 nm in diameter.
Carbon-bearing, Ferromagnesiosilica Units
UltraÐne (2 to D50 nm) platy Fe,Mg-olivine, Mg, Fe-pyroxene, Fe, Ni-sulÐde, cMixedcparticles (silicates plus carbon)
iron oxide, and metallic iron in a refractory hydrocarbon and amorphous
carbon matrix. The units contain minor Al, Ca, Cr, Mn and Ni, and traces of
phosphorous and zinc. Units are D100 nm in diameter.
Ferromagnesiosilica Polyphase Units
(1) Coarse-grained (10È410 nm) units that consist of Mg, Fe-olivine, Mg, Fe Silicate [(Mg, Fe, Si)O] particles
pyroxene and an amorphous Al-bearing silica material ^Ca,Mg and Fe. The
units have a bimodal size distribution with mean diameters of 530 and 1140
nm, (2a) ultraÐne-grained (\50 nm) units with an amorphous matrix containing
Mg, Fe-olivine, Mg, Fe-pyroxene, Fe, Ni-sulÐdes and magnetite grains. Units
are 125 nm to 1000 nm in diameter, and (2b) magnesiosilica units (100È1000
nm in diameter) of glass with embedded metal (kamacite) and sulÐdes.
NOTE.ÈThe minerals in PCs are secondary phases that formed in situ in the amorphous units during Ñash heating when IDPs decelerate in the
EarthÏs atmosphere between D100È80 km altitude, parent body alteration, preaccretionary thermal processes (irradiation), or a combination of
these processes.
aFrom Rietmeijer 1998a, and references therein.
bFrom Fomenkova et al. 1992.
No. 1, 1999 METASTABLE EUTECTIC CONDENSATION 397
FIG. 1. Transmission electron micrograph of magnetite in a Fe-Mg-SiO- condensed vapor. The di†raction maxima (arrows) in this orientationH2-O2
(straight brackets) conÐrm its single-crystal nature (inset). The gray background in all TEM images is the epoxy wherein the sample was embedded for
ultramicrotome sectioning.
PCs are circumstellar condensates, materials condensed in
the solar nebula or the result of thermal processing of cir-
cumstellar dust (Nuth, Hallenbeck, & Rietmeijer 1999a).
Experimental studies can address this question by exploring
the processes of dust formation and evolution through
analog condensation and thermal annealing experiments
(Hallenbeck et al. 1998; Nuth et al. 1999b) in the laboratory.
Circumstellar dust analogs produced in the laboratory by
vapor-phase nucleation from Al-SiO- Mg-SiO-H2-O2,H
2-
and Fe-SiO- vapors show several general proper-O2H2-O2
ties:
1. Metal oxide-silica grains, ranging in size from mostly a
few nanometers to 150 nm in diameter, form interconnected
necklaces in Ñu†y smokes.
2. Grains in necklaces typically show evidence for surface
free energyÈdriven coagulation and grain growth.
3. The Ñu†y texture o†ers restricted contact area for
chemical exchange between the grains via solid state di†u-
sion; the smokes are therefore chemically inactive during
thermal annealing until their texture has collapsed at tem-
peratures close to the melting point.
4. The grains include oxides with a stoichiometric
mineral composition (Rietmeijer 1992) and non-
stoichiometric mixed-metal grains.oxideÈSiO2
5. Postcondensation autoannealing induces readjust-
ments in condensed solids when the thermal energy of the
grain is still high. Such reactions occur before condensed
grains reach the quench temperature of the surrounding
gas. The adjustments include homogeneous phase decom-
position in nonstoichiometric mixed grains and crys-
tallization of pure metal oxides.
6. Invariably, the individual, amorphous mixed metal-
oxide/silica grains have well-deÐned composition distribu-
tions that match those of the metastable eutectics in the
(Rietmeijer & Karner 1999), (NuthAl2O3-SiO2MgO-SiO2
et al. 1999b; Rietmeijer 1998b) and FeO/Fe2O3-SiO2
398 RIETMEIJER, NUTH, & KARNER Vol. 527
FIG. 2. Transmission electron micrograph of ““ FeSiO ÏÏ necklaces in a condensed Fe-Mg-SiO- vapor.H2-O2
FIG. 3. Histograms of the distribution of individual solid grain com-
positions in a condensed Fe-Mg-SiO- smoke.H2-O2
(Rietmeijer, Nuth, & Karner 1999) equilibrium phase dia-
grams.
Observation of both experimentally produced and
natural condensation products that display metastable
eutectic chemical compositions are the best proof yet that
nonequilibrium vapor-phase condensation is not a chaotic
event but rather proceeds to produce solids with predictable
chemical order. The resulting amorphous solids resemble
dissipative structures, states of organization wherein non-
equilibrium becomes a source of order (Prigogine 1979).
During nucleation in a cooling vapor, the Ðrst stable nuclei
(molecules?) that form become seeds for grain growth when
the remaining vapor condenses. Given sufficient activation
energy and time, the condensed solids will arrange them-
selves into the least energetic physical and chemical con-
Ðguration until they equilibrate with the ambient
environment. However, kinetic factors might favor the for-
mation of metastable high-energy states if the vapor con-
denses rapidly.
In this paper we report the results of a vapor-phase con-
densation experiment from a Fe-Mg-SiO- vapor.H2-O2
Instead of widely variable Mg/Fe/Si ratios scattered about
the mean composition of the condensate we found only
individual grains of compositionally ordered ““ MgSiOÏÏ and
““ FeSiOÏÏ solids in addition to pure end-member oxides. We
predict that circumstellar outÑows contain only a limited
number of compositionally distinct condensed grain types
with relative proportions that may vary among di†erent
No. 1, 1999 METASTABLE EUTECTIC CONDENSATION 399
FIG. 4. Transmission electron micrograph of periclase (MgO) grains in a Fe-Mg-SiO- smoke. The di†raction maxima (arrows) in this orientationH2-O2
(straight brackets) conÐrm they are single crystals (inset).
stars. We submit that coarse-grained ferromagnesiosilica
PCs in aggregate IDPs may be common circumstellar or
protostellar condensates formed by postaccretion annealing
of condensed dust aggregates.
3.EXPERIMENTAL PROCEDURE
The samples were produced by nonequilibrium gas-to-
solid condensation in a Mg-Fe-SiO- vapor using theH2-O2
Condensation Flow Apparatus described by Nuth et al.
(1988; 1999b). In this apparatus there is no standard set of
Ñow, temperature, or pressure settings, but the following
conditions are all at least somewhat typical of an average
experiment. The total pressure in the system is D100 torr,
while the temperature in the furnace is D1000 K. The
hydrogen Ñow rate is D1000 standard cubic centimeters per
minute (sccm). The helium Ñow through the iron-carbonyl
is on the order of 500 sccm. We estimate that the iron
carbonyl concentration is D10% of the total Ñow of helium.
Magnesium metal is placed into a graphite boat inside the
furnace. At the nominal 1000 K temperature of the furnace,
magnesium metal has a vapor pressure of D5 torr. The
oxygen Ñow rate always equals or exceeds that of silane
but is usually within a factor of 2 of the silane setting.(SiH4)
The silane Ñow rate is typically D100 sccm. As a result the
hydrogen to oxygen ratio is always greater than 5, as some
amount of hydrogen is also contributed by silane. The
typical oxygen Ñow is sufficient to just balance the hydrogen
contributed from the silane Ñow to produce water and Si0
without e†ecting the bulk Ñow of hydrogen. In this system
the oxygen fugacity during condensation is a dynamic
quantity that is dependent on reaction rates, rather than a
thermodynamic quantity. The actual oxygen fugacity
during condensation is conducive to the formation of iron
oxides instead of Fe0, to the formation of silicon oxides
instead of Si0, and to the formation of magnesium oxides
rather than to Mg0. The exact bulk composition of the
condensing gas is unknown but will be very close to the
bulk composition of all solid grains in the resulting smoke.
This composition can be obtained analytically (see below).
400 RIETMEIJER, NUTH, & KARNER Vol. 527
FIG. 5. Ternary diagram MgO-FeO- (oxide wt%) with the chemi-SiO2
cal compositions of gas to solid condensed solids from a Fe-Mg-SiO-
vapor with metastable ““ MgSiOÏÏ and ““ FeSiOÏÏ solids matching theH2-O2
metastable eutectics. The ““ average bulk solid ÏÏ composition (dot)isa
proxy of the gas phase composition that was probably somewhat less
SiO-rich.
Ultrathin sections of the smoke sample were prepared in
a fashion similar to that employed in the analysis of IDPs
(Bradley & Brownlee 1986). Mineralogical characterization
of individual constituents in this smoke followed the same
procedures used in studies of aggregate IDPs (Rietmeijer
1998a). The sample was analyzed by analytical and trans-
mission electron microscopy (AEM, TEM) with a 0.2 nm
spatial resolution in the TEM viewing mode and an analyti-
cal probe size of 10È20 nm in diameter for in situ determi-
nation of the chemical composition of individual grains
greater than 10 nm in diameter. Rietmeijer & Karner (1999)
provided details of the analytical procedures and data
reduction techniques, and we refer the reader to this paper.
Artifacts in grain compositions could arise from overlap-
ping grains smaller than the section thickness (80È100 nm).
This potential artifact was not a problem in this AEM study
because the exact location of each chemical analysis was
recorded on electron micrographs that were used to evalu-
ate the possible ““ contamination ÏÏ of individual analyses
prior to data reduction. Also, prior to AEM analysis each
grain was viewed using a ““ through-focus ÏÏ technique to
determine the presence of additional grains along the path
of the analyzing electron beam. For the record we note that
mineral alteration due to autoannealing following sample
heating in the electron beam is readily recognizable
(Rietmeijer et al. 1999) and did not e†ect our analyses.
4.RESULTS
Tridymite, a high-temperature silica polymorph, was
observed to form spherical grains (D10 nm in diameter)
that occur in aggregates and compact clusters up to D200
nm in size. There is evidence for coarsening of these con-
densed grains to form D30 nm-sized subhedral grains.
These grains include Fe-bearing silica grains with up to 12
weight percent (wt%) FeO, or 11 wt% Rare largeFe2O3.
(D165 nm in diameter) subspherical grains of amorphous
silica were also found. Iron oxide grains (D34 nm in
average diameter) are interspersed in the silica aggregates
FIG. 6. phase diagram extended to the nominal quenchMgO-SiO2
temperature at 500¡ C with the observed metastable eutectic magnesiosil-
ica grain compositions (open squares) for amorphous serpentine (serp) and
smectite (SM) dehydroxylates and at D85 wt% MgO.
and clusters. They are rounded single-crystal grains of mag-
netite or (c-uptoD350 nm in(Fe3O4) maghe
mite Fe2O3)
diameter with compositions that are D85È100 wt% FeO
(Fig. 1). Mixed ““ FeSiO ÏÏ grains form both rounded and
subhedral grains 10È30 nm in diameter (mean \17 nm)
arranged in strands (Fig. 2). They contain between 22È30
wt% FeO (mean \26 wt% FeO), or 24È33 wt% Fe2O3
(Fig. 3). There is a hint of a fourth group of mixed FeSiO
grains with compositions of between 70È75 wt% FeO. Rare
periclase (MgO) crystals (90È100 wt% MgO) are observed
to occur isolated from the remaining smoke material (Fig.
No. 1, 1999 METASTABLE EUTECTIC CONDENSATION 401
FIG. 7. Revised phase diagram at (2A) and 6/4 (2B) showing the metastable eutectic points matching the observedFeO/Fe2O3ÈSiO2FeO/Fe2O3\8/2
ferrosilica compositions (open squares) produced by nonequilibrium gas to solid condensation. The positions of the high-silica eutectic point shifts as a
function of variable We made no attempt to control the oxygen fugacity, during condensation but and indicate oxidizingFeO/Fe2O3.fO2,c-Fe2O3Fe3O4
conditions, atm (500¡ C) and atm (1000¡ C) (see Lindsley 1976) A third metastable eutectic occurs as an iron serpentinefO2[10~18 fO2[10~6
dehydroxylate, Its formation requires values that promote FeO formation only. The condensed grains, 70È75 wt% FeO (see Fig. 5) in the(Fe2`)3Si2O7.fO2
Fe-Mg-SiO- vapor match this unique metastable eutectic. (Reproduced from Rietmeijer et al. 1999 ; copyright 1999 by the Royal Society of Chemistry.)H2-O2
4). These sub- to euhedral grains range from 70 to 140 nm in
size. Mixed ““ MgSiO ÏÏ grains occur as clusters of rounded
grains, 40È140 nm in diameter (mean \78 nm), with an
average of 43 wt% MgO. These listed compositions are
averages of well-deÐned Gaussian populations of individual
solid grains in this sample. The data summarized in the
ternary diagram MgO-FeO- show: silica grains, MgOSiO2
and Fe-oxide grains with limited solid solution of silica, and
clusters of mixed ““ MgSiO ÏÏ and ““ FeSiO ÏÏ grains (Fig. 5).
Note that there were no individual grains found in this
sample with compositions on the ““ interior ÏÏ of the ternary
diagram and none that displayed the ““ average ÏÏ composi-
tion of the original condensing vapor.
5.DISCUSSION
Gas-to-solid condensation in the Mg-Fe-SiO-H2-O2
vapor (Fig. 5) mimicked the condensation behavior
observed separately in the andMgO-SiO2FeO/Fe2O3-
systems: this is rather remarkable in that no mixedSiO2
magnesium-iron silicate grains were observed at all. The
grain morphology and size, mineralogy and compositional
clusters of grains condensed in this Mg-Fe-SiO-H2-O2
vapor closely resemble those produced by condensation in
the and systems. Grains inMgO-SiO2FeO/Fe2O3-SiO2
the former had four peaks at D0 wt% MgO (silica), at 28
wt% MgO, smectite dehydroxylate and 48(Mg6Si8O22),
wt% MgO, serpentine dehydroxylate with(Mg3Si2O7),
Mg/(Mg ]Si) (element) ratios of 0.4 and 0.55, respectively,
and 95È100 wt% MgO (periclase) (Fig. 6). Four composi-
tional peaks in the system also occurredFeO/Fe2O3-SiO2
at 0, 15, 30, and 88 wt% FeO (Fig 7). These peaks match the
metastable eutectic points and end member compositions in
both equilibrium phase diagrams. The preferred metastable
eutectic compositions are constrained by the liquidus topol-
ogy of the and phase dia-MgO-SiO2FeO/Fe2O3-SiO2
grams. However, condensation in the ternary vapor did not
produce any solids with preferred com-MgO-FeO/Fe2O3
positions (Fig. 5), which suggests the absence of metastable
eutectic points in the equilibrium phase diagram that would
lead to the formation of this predictable phenomenon.
Indeed, the phase relations in the andMgO-Fe2O3
MgO-FeO phase diagrams (Phillips, Somiya, & Muan
1961; Muan 1958) show complete magnesiowu
stite
[(Mg, Fe)O] or olivine [(Mg, Fe) solid solution at
2SiO4]
higher silica activity. There is no opportunity to form meta-
stable eutectic solids of mixed Mg-Fe composition, which is
consistent with the observations described here. We note
that the Mg-Fe-Si oxide system has considerable cosmo-
chemical interest because it describes not only the PC and
chondritic IDP bulk compositions (Rietmeijer 1998a) but
also the compositions of the meteorite matrix and of the
Ðne-grained rims on chondrules in carbonaceous and
402 RIETMEIJER, NUTH, & KARNER Vol. 527
FIG. 8. Mg-Fe-Si (element wt%) diagram is used to compare major
element compositions of IDPs, and matrix of undi†erentiated carbon-
aceous and ordinary chondrites. Metastable cotectic-mixing lines (dashed
lines) connect the compositions of the amorphous, metastable eutectic con-
densed solids (dots). These compositions are (1) smectite (Sm-d) and serpen-
tine (S-d) dehydroxylates and two low-FeO smectite-like dehydoxylates
(see Figs 6 and 7), (2) Si-bearing Fe- and Mg-oxides, and (3) silica grains
(solid square). Fe2`,Fe3`symbolizes variations in the Fe2`/Fe3`ratios
(legend to Fig. 7). Average fe (Av.; 0.23) and maximum (Max.) fe and three
bulk compositions of these PCs ( Ðlled diamonds) are indicated. A metasta-
ble cotectic mixing line (dashed lines) connects the S-d and greenalite (G-d)
(open circle) metastable eutectics. It is located in between the stoichiometric
olivine (Fo-Fa) and pyroxene (En-Fs) lines.
unequilibrated ordinary chondrites (Zolensky & McSween
1988).
Our data support the hypothesis that nonequilibrium gas
to solid condensation at considerable supercooling of a
Fe-Mg-SiO- gas will produce amorphous grains withH2-O2
unique mixed ““ MgSiO ÏÏ and ““ FeSiO ÏÏ compositions
deÐned by metastable eutectics, together with almost pure
Si-oxide (tridymite), Fe-oxide (c- and Mg-oxideFe2O3)
(periclase) (Fig. 8). The compositions of these mixed grains
do not represent thermodynamic equilibrium that predicts
stoichiometric pure Mg-silicates but not the formation of
iron oxides. Instead iron is predicted to condense as metal-
lic iron that will become oxidized when the temperature has
fallen below D325¡ C. Equilibrium condensates in our
experiment would have been magnesium silicates, iron
metal, and free silica. Instead we Ðnd metastable eutectic
magnesiosilica and ferrosilica condensates. Some of the
simple metal oxides underwent a rapid, amorphous-to-crys-
talline transition without chemical readjustment by solid
state di†usion, during autoannealing immediately after con-
densation and before thermal equilibration with the sur-
rounding gas. The metastable nature of these grains was
highlighted by the observation that thermal annealing of
polycrystalline tridymite globules quickly brought about
solid state amorphization. This transformation proceeded
when amorphous silica nucleated at the surface and grain
boundaries below the tridymite melting temperature but
above its glass transition temperature (Rietmeijer et al.
1999). It appears that when sufficient energy is available to
activate postcondensation modiÐcations all of the con-
densed solids will ultimately exist as amorphous materials.
Condensation.ÈThe Ðnding that nonequilibrium gas-to-
solid condensation of a Fe-Mg-SiO- vapor does notH2-O2
produce solids with mixed Mg/Fe/Si composition may be
critical to understanding the nature of circumstellar dust.
We submit that circumstellar dust grains originally con-
dense as amorphous solids with predictable metastable
eutectic ““ MgSiO ÏÏ and ““ FeSiO ÏÏ compositions plus a small
fraction of metal oxide grains. Considering the cosmic
abundances of the elements, these experiments indicate that
the most common circumstellar condensates include only a
limited number of compositionally well-deÐned grains.
Because there are only a limited number of primary conden-
sates, dust aggregates can form only along metastable
cotectic mixing lines linking these primary metastable
eutectics (Fig. 8). Postaccretion thermal annealing of simple,
binary, or ternary dust aggregatesMgO-FeO/Fe2O3-SiO2
will produce mixed Mg-Fe- grains, but these materialsSiOx
will still be compositionally unique solids.
Postcondensation modiÐcation.ÈUnique aggregate com-
positions are determined by the relative proportions of con-
densed dust grains on metastable cotectic mixing lines in
the Mg-Fe-Si (element wt%) diagram (Fig. 8). The aggre-
gate compositions along these mixing lines, and, more par-
ticularly, the compositions at the intersections of these lines,
match the smectite dehydroxylate, (Mg, Fe) com-
6Si8O22,
positions of coarse-grained PCs in aggregate IDPs (Fig. 8).
By virtue of their amorphous nature, the condensed dust
contains a signiÐcant amount of internal, i.e., chemical
(Clayton 1980), energy that can sustain spontaneous,
kinetically controlled exothermic reactions of low activa-
tion energy. It seems likely that condensed dust aggregates
will be able to undergo spontaneous exothermic reactions
following some energetic event (grain collision, UV photon
absorption) and become chemically homogeneous amorp-
hous solids. The fe \0È0.36 compositions of coarse-
grained PCs are determined by the condensed ““ FeSiO ÏÏ
dust compositions and not by a ““ primary ÏÏ solid solution
mechanism established during condensation from the
vapor. Such a mechanism is prohibited by the phase
relationships in the and MgO-FeO phase dia-MgO-Fe2O3
grams. During thermal annealing smectite dehydroxylate
will undergo amorphous phase decomposition into
serpentine dehydroxylate plus amorphous silica, viz.,
(Mg, Fe) whereby fe
6Si8O22 F2(Mg, Fe)3Si2O7]4SiO2,
remains constant. Coarse-grained olivine [(Mg, Fe)2SiO4]
and pyroxene [(Mg, Fe) (Fig. 9) with identical fe will
2Si2O6]
form according to 2(Mg, Fe)3Si2O7F2(Mg, Fe)2SiO4]
(Mg, Fe) and will yield an olivine to pyroxene ratio
2Si2O6,
of 2 to 1. Chemical analyses of the coarse-grained (10È410
nm) PCs in aggregate IDPs showed identical fe for the co-
occurring olivine and pyroxene in each unit (Rietmeijer
1997). Although it is not clear when these minerals formed,
Ñash heating simulation experiments (Joswiak & Brownlee
1998) suggest that they could form during deceleration of
IDPs in the EarthÏs atmosphere. This point underscores our
notion that the bulk composition of the ferromagnesiosilica
units is the primary critical property of these grains when
one looks for an explanation of their genesis. The minerals
actually observed in collected aggregate IDPs could easily
result from secondary alteration processes.
6.CONCLUSIONS
The gas-to-solid condensation experiments reported in
this paper showed the formation of amorphous, mixed-
No. 1, 1999 METASTABLE EUTECTIC CONDENSATION 403
FIG. 9. Transmission electron micrograph of a coarse-grained ferromagnesiosilica PC in chondritic aggregate IDP L2011A9 showing the coexisting
olivine (ol) and pyroxene (pyr) crystals along with an amorphous ““ silicaÏÏ phase. After crystallization of the originally amorphous PC any Al and Ca that was
present will reside in the nonstoichiometric ““ silicaÏÏ phase (Reproduced from Rietmeijer 1998a, courtesy of the Mineralogical Society of America.)
oxide solids with predictable metastable eutectic composi-
tions. These results imply that the composition of grains
condensed in circumstellar outÑows will include only solids
with a very limited number of mixed ““ MgSiO ÏÏ and
““ FeSiO ÏÏ compositions. In particular, no mixed
““ MgFeSiO,ÏÏ i.e., ferromagnesiosilica, grains will form as
primary condensates in such outÑows. The experimental
observations are consistent with ISO observations, indicat-
ing that the crystalline mineral fraction of high mass loss
rate AGB dust consists almost exclusively of pure magne-
sium silicate minerals. No evidence of mixed ““ MgFeSiO ÏÏ
minerals has yet been observed. Our experimental results
predict that no evidence for mixed ““ MgFeSiO ÏÏ conden-
sates will be found in such outÑows.
Our experiments also have implications for the composi-
tion of grains formed via condensation in protostellar
nebulae and, in particular, for the composition of primitive
grains derived from comets in our own solar system, the
aggregate IDPs. The coarse-grained, mixed oxide ferromag-
nesiosilica units in aggregate IDPs could not have formed
directly by condensation but must have been formed follow-
ing thermal processing of condensed dust aggregates. The
high free energy of the metastable amorphous dust in the
aggregates combined with the large surface free energy of
nanometer-size grains probably facilitated formation of the
compact amorphous units following the input of a small
amount of ““ activation energy.ÏÏ There is no constraint on
when this activation event might have occurred, and the
minerals in the units in aggregate IDPs might therefore
have formed during entry heating in the terrestrial atmo-
sphere. The original aggregate compositions are deÐned by
metastable cotectic mixing lines between the primary con-
404 RIETMEIJER, NUTH, & KARNER
densates. This again predicts that only limited variations in
the chemical compositions of these nonchondritic units will
be observed in aggregate IDPs and this prediction is consis-
tent with the available observational evidence.
We thank an anonymous referee for a constructive dis-
cussion. This work was performed in the Electron
Microbeam Analyses Facility at UNM, where Fleur-
Rietmeijer Engelsman provided technical support. Robert
N. Nelson and Susan L. Hallenbeck assisted with the con-
densation experiments. Part of this work was done in
partial fulÐllment of the requirements of the degree of
Master of Science at UNM (J. M. K.). This research was
supported by NASA grants NAGW-3646 and NAG5-4441.
REFERENCES
Bradley, J. P. 1994, Science, 265, 925
Bradley, J. P., & Brownlee, D. E. 1986, Science, 231, 1542
Bradley, J. P., Humecki, H. J., & Germani, M. S. 1992, ApJ, 394, 643
Brownlee, D. E. 1994, in AIP Conf. Proc. 310, Analysis of Interplanetary
Dust, ed. M. E. Zolensky, T. L. Wilson, F. J. M. Rietmeijer, & G. J.
Flynn (New York: AIP), 5
Brownlee, D. E., Joswiak, D. J., & Bradley, J. P. 1999, Lunar and Planetary
Science Conf. 30, CD-ROM, abstract 2031 (Houston: LPI)
Brownlee, D. E., Tomandl, D., & Hodge, P. W. 1976, in Interplanetary
Dust and the Zodiacal Light, ed. H. Elsasser & H. Fechtig (New York:
Springer:), 279
Cami, J., de Jong, T., Justtanont, K., Yamamura, I., & Waters, L. B. F. M.
1998, Ap&SS, 255, 339
Clayton, D. D. 1980, ApJ, 239, L37
De, B. R. 1979, Ap&SS, 65, 19
Donn, B. 1979, Ap&SS, 65, 167
Flynn, G. J. et al. 1996, in ASP Conf. Ser. 104, Physics, Chemistry and
Dynamics of Interplanetary Dust, ed. B. A. S. Gustafson & M. S. Hanner
(San Francisco: ASP), 291
Fomenkova, M. N., Kerridge, J. F., Marti, K., & McFadden, L.-A. 1992,
Science, 258, 266
Gail, H.-P. 1998, A&A, 332, 1099
Grossman, L., & Larimer, J. W. 1974, Rev. Geophys. Space Phys., 1, 71
Hallenbeck, S. L., Nuth, J. A., III, & Daukantes, P. L. 1998, Icarus, 131, 198
Hallenbeck, S. L., Nuth, J. A., III, & Nelson, R. N. 2000, ApJ, in press
Joswiak, D. J., & Brownlee, D. E. 1998, Lunar and Planetary Science Conf.
29, CD-ROM, abstract 1929 (Houston: LPI)
Kozasa, T., & Sogawa, H. 1997, Ap&SS, 251, 165
ÈÈÈ. 1998, Ap&SS, 255, 437
Lattimer, J. M., & Grossman, L. 1978, Moon Planets, 19, 169
Lindsley, D. H. 1976, in Oxide Minerals, ed. D. Rumble, III, (Washington,
DC: Mineralogical Society of America), L-61
Mackinnon, I. D. R., McKay, D. S., Nace, G., & Isaacs, A. M. 1982, Proc.
13th Lunar Planet. Sci. Conf. (J. Geophys. Res. Suppl., No. 87, A413
Mackinnon, I. D. R., & Rietmeijer, F. J. M. 1987, Rev. Geophys., 25,
1527
Muan, A. 1958, Am. J. Sci., 256, 171
Nuth, J. A., III. 1996, in The Cosmic Dust Connection, ed. J. M. Greenberg
(Dordrecht: Kluwer), 205
Nuth, J. A., III, Hallenbeck, S. L., & Rietmeijer, F. J. M. 1999a, Laboratory
Studies of Silicate Smokes: Analog Studies of Circumstellar Materials. J,
Geophys. Res., in press
Nuth, J. A., III, Hallenbeck, S. L., & Rietmeijer, F. J. M. 1999b, in Labor-
atory Astrophysics and Space Research, ed. P. Ehrenfreund, K. Kra†t,
H. Kochan, & V. Pirronello (Dordrecht: Kluwer), 143
Nuth, J. A., III, Nelson, R. N., Moore, M., & Donn, B. 1988, in Experi-
ments on Cosmic Dust Analogues, ed. E. Bussoletti, C. Fusco, & G.
Longo (Dordrecht: Kluwer), 191
Phillips, B., Somiya, S. & Muan, A. 1961, J. Am. Ceram. Soc., 44, 167
Prigogine, I. 1979, Ap&SS, 65, 371
Rietmeijer, F. J. M. 1992, ApJ, 400, L39
ÈÈÈ. 1996, Meteoritics Planet Sci., 31, 237
ÈÈÈ. 1997, Lunar and Planetary Science Conf. 28, CD-ROM, abstract
1173 (Houston: LPI)
ÈÈÈ. 1998a, in Planetary Materials, ed. J. J. Papike (Washington, DC:
Mineralogical Society of America), 2-1
ÈÈÈ. 1998b, Lunar. Planet. Sci. Conf. 29, CD-ROM abstract 1150
(Houston: LPI)
Rietmeijer, F. J. M., & Karner, J. M. 1999, J. Chem. Phys., 110, 4554
Rietmeijer, F. J. M., Nuth, J. A., III, & Karner, J. M. 1999, Phys. Chem.
Chem. Phys., 1, 1511
Sogawa, H., & Kozasa, T. 1999, ApJ, 516, L33
Sutton, S. R. 1994, in AIP Conf. Proc. 310, Analysis of Interplanetary Dust,
ed. M. E. Zolensky, T. L. Wilson, F. J. M. Rietmeijer, & G. J. Flynn (New
York: AIP), 145
Thomas, K. L., Keller, L. P. & McKay, D. S. 1996, in ASP Conf. Ser. 104,
Physics, Chemistry, and Dynamics of Interplanetary Dust, ed. B. A. S.
Gustafson & M. S. Hanner (San Francisco: ASP), 283
Tielens, A. G. G. M., Waters, L. B. F. M., Molster, F. J., & Justtanont, K.
1998, Ap&SS, 255, 415
Waelkens, C., et al. 1996, A&A, 315, L245
Warren, J. & Zolensky, M. E. 1994, in AIP Conf. Proc. 310, Analysis of
Interplanetary Dust, ed. M. E. Zolensky, T. L. Wilson, F. J. M. Rietmei-
jer & G. J. Flynn (New York: AIP), 245
Waters, L. B. F. M., et al. 1996, A&A, 315, L361
Woolf, N. J. & Ney, E. P. 1969, ApJ, 155, L181
Zolensky, M. E., & McSween, Jr., H. Y. 1988, in Meteorites and the Early
Solar System, ed. J. F. Kerridge & M. S. Matthews (Tucson: Univ.
Arizona Press), 114