Fluid Flow in Chondritic Parent Bodies: Deciphering the Compositions of Planetesimals

See allHide authors and affiliations

Science  12 Nov 1999:
Vol. 286, Issue 5443, pp. 1331-1335
DOI: 10.1126/science.286.5443.1331


Alteration of the Allende meteorite caused shifts in oxygen isotope ratios along a single mass fractionation line. If alteration was caused by aqueous fluid, the pattern of oxygen isotope fractionation can be explained only by flow of reactive water down a temperature gradient. Down-temperature flow of aqueous fluid within planetesimals is sufficient to explain the mineralogical and oxygen isotopic diversity among CV, CM, and CI carbonaceous chondrites and displacement of the terrestrial planets from the primordial slope 1.00 line on the oxygen three-isotope plot.

Carbonaceous chondrites comprise seven distinct groups of primitive meteorites. The groups are distinguished on the basis of mineralogy, bulk elemental concentrations, and size and proportions of constituents such as chondrules and calcium-aluminum–rich inclusions (CAIs) (1). Each group is characterized by distinctive oxygen isotope ratios (2). The diversity in mineralogy and oxygen isotopic compositions is spanned by the CV, CM, and CI groups (3) and has been attributed to interactions among different primordial oxygen reservoirs on distinctive parent bodies with different geological histories (4). Here we show that reaction between rock and flowing water inside a carbonaceous chondrite parent body could have produced zones that resemble CV, CM, and CI meteorites in mineralogy and oxygen isotope ratios.

Studies of the Allende CV3 carbonaceous chondrite using the ultraviolet laser microprobe show that increases in 17O/16O and 18O/16O at constant Δ17O on an oxygen three-isotope plot (5) (Fig. 1) are associated with alteration. The alteration is identified by localized enrichments in Fe, Cl, and Na and by growth of secondary minerals (6). Alteration occurred within several million years of chondrule and CAI formation about 4.5 billion years ago (7) and has been attributed either to reactions between vapor and solids in the early solar nebula (8) or to reactions between rock and liquid water at low temperatures within parent bodies that may have resembled some present-day asteroids (9).

Figure 1

Oxygen three-isotope plot showing the relative positions of the terrestrial mass fractionation curve (TMF), the slope 1.00 line (Y&R), and the Allende mass fractionation curve (AMF) defined by ultraviolet laser ablation analyses of an Allende CAI and two chondrules. The two lines for the AMF delimit physically plausible slopes between 0.51 and 0.53. Also shown are the compositions of the Allende whole rock (WR) (22) and bulk matrix (M) (2). Chondrule and CAI laser ablation analyses containing identifiable mineral grains with distinct Δ17O relative to the bulk of the object are excluded for clarity.

The preponderance of evidence is that water had higher Δ17O values than did coexisting anhydrous silicates in the early solar system (10). Laser ablation analyses showing that altered and unaltered components fall on a single mass fractionation curve (Fig. 1) suggest that the Δ17O of the aqueous fluid (or any other reactant) responsible for the alteration was changed from its original value to the rock value by exchange of oxygen with the rock.

The exchange of oxygen between rock and a motionless aqueous fluid cannot explain the data in Fig. 1 because, when the amount of fluid is sufficiently small that its Δ17O is controlled entirely by the rock, it has too little oxygen to change rock δ17O and δ18O. This conundrum is quantified by means of the commonly used expression for the mass balance of oxygen between reacting rock (r) and stagnant water (w)Embedded Image(1)In Eq. 1, N is the number of oxygens composing the reacting water or rock; δ is the δ18O or δ17O for the indicated phase after reaction; δ0 is δ18O or δ17O for the indicated phase before reaction; and Δ is the difference, δr − δw, between rock and water at equilibrium. The left side of Eq. 1 is referred to as the water-rock ratio. Because Eq. 1 applies to both δ17O and δ18O, invariant rock Δ17O during the exchange of oxygen between rock and water with different initial Δ17O values is only possible in the limit, where the amount of oxygen composing the water relative to the amount of oxygen composing the rock is effectively zeroEmbedded Image(2) Equations 1 and 2 show that reaction between static water and rock could not have shifted rock δ18O at fixed Δ17O as indicated by the data in Fig. 1 unless the water and rock had identical Δ17O before exchange.

Equation 1 and the restrictions it imposes on changes in rock δ18O and Δ17O no longer apply if the aqueous fluid flowed during isotopic exchange. Instead, mass balance during flow is satisfied by the expressionEmbedded Image(3)were α is the equilibrium rock/fluid isotope ratio fractionation factor (essentially 1),J17,18Ois a time-integrated flux of oxygen composing the fluid (as traced by 17O or 18O), ∇T is the gradient in temperature, and δr0 represents the initial δ17O or δ18O for the rock at a given position. Equation 3 describes the situation where pore-filling fluid equilibrates with the host rock at each position along a fluid flow path. At every location, the fluid Δ17O is controlled by the rock as long as the rock's capacity for exchange persists (11). Rock controls the fluid Δ17O because the pore volume is generally less than the volume of the host rock, and so the bulk of the oxygen resides in the rock. Despite rock-controlled Δ17O, Eq. 3 shows that shifts in rock17O/16O and 18O/16O are still possible when bothJ17,18Oand ∇T are nonzero. For water flowing through rock composed primarily of silicate minerals, (∂δw/∂T) is positive and an increase in δr requires flow in the direction of decreasing temperature (Eq. 3); for positiveJ17,18Oand (δr − δr0) the sign of ∇T must be negative. The increase in δr with alteration in Allende (Fig. 1) means that if the alteration occurred by reaction with liquid water, it did so in a hydrological system in which flow was from regions of higher temperature toward regions of lower temperature. Expressions such as Eq. 3 can lead to first-order constraints on the nature of fluid flow in a meteorite parent body, but they cannot capture the complexity that arises where fluid flux and temperature change with time along the flow path.

For this reason, we constructed finite difference models for the thermal, isotopic, and mineralogical evolution of a generic carbonaceous chondrite parent body composed of 20% water ice, 10% void space, and 70% silicate by volume (12). For simplicity, the silicate rock was assumed to be pure forsterite (Mg2SiO4). Carbon was included in the calculations because the occurrence of carbonate minerals in veins constitutes clear evidence for aqueous fluid flow, and because modeling carbonate formation allows us to compare our results with well-known differences in oxygen isotope ratios between carbonates and other minerals as a means of validation.

The components required to simulate reactions between magnesian silicate, water, and carbon can be represented by MgO, SiO2, H2O, and CO2. Aqueous alteration of carbonaceous chondrites resulted in formation of phyllosilicate minerals, as evidenced most clearly by the CI and CM groups (3). Prevalent among these hydrous silicates is saponite, a trioctahedral smectite. In the model system, we used talc [Mg3Si4O10(OH)2] as the MgO-SiO2-H2O-CO2 analog for saponite and magnesite (MgCO3) as the carbonate phase. Among forsterite, talc, magnesite, and a mixed aqueous fluid phase composed of H2O and CO2 there is one independent reactionEmbedded Image Embedded Image(4)The progress of this reaction serves as an analog for the mineralogical effects of aqueous alteration in our calculations. The phyllosilicates in CM and CI rocks include serpentine as well as saponite (13). Because the fractionation of oxygen isotopes between serpentine and talc is large (12), we converted our model talc isotopic compositions into mixtures of 50% talc and 50% serpentine (on an oxygen basis) using published fractionation factors to compare our results with the CI and CM data.

Results reported here are for a fictive parent body with a spherical radius of 25 km. Although we ignored Al in the chemical system as a way of simplifying the thermodynamic calculations,26Al was used as a heat source (12) with an initial 26Al/27Al consistent with formation of the body about 3 million years after formation of CAIs found in Allende (14). The fictive parent body was allowed to heat up from a temperature of 170 K (imposed by radiative heating from the sun) by decay of 26Al. Melting of water ice above 273 K caused down-temperature flow of liquid water. Liquid was permitted to pass through rock below 273 K to simulate percolation of unfrozen water (15). The driving force for the down-temperature flow in a microgravity environment (g = 0.01 N/kg in this case) would likely be dominated by capillary action rather than by body forces at temperatures below the boiling point (16). We made no attempt to model the driving force in detail. Instead we allowed the vapor pressure of H2O relative to the vacuum of space to drive the flow. The maximum water flux in the present model is a factor of 10 less than maximum cometary water fluxes at 1 astronomical unit (17). Mineralogical and isotopic reactions between the flowing liquid water and the rock were kinetically controlled (12).

We assumed that all condensed oxygen in the early solar nebula lay on the primordial slope 1.00 array on an oxygen three-isotope plot (18). The initial 17O/16O and18O/16O for the rock were taken as the intersection between the slope 1.00 array and the Allende mass fractionation curve defined by laser ablation measurements of Allende. Initial 17O/16O and18O/16O for the water ice in our calculations were defined by the intersection of the slope 1.00 array and the mass fractionation curve passing through the hydrous phases of the Orgeuil CI carbonaceous chondrite. Our choice of initial water17O/16O and 18O/16O is consistent with magnetite oxygen isotope ratios in Orgeuil and other chondrites (19).

The initial CO2 concentration of the water ice melt is constrained by forward progress of the reaction in Eq. 4. The reaction proceeds from left to right when the mole fraction of CO2 in the fluid is greater than the equilibrium value at the appropriate conditions in the model chemical system (the equilibrium value is <1 × 10−4 in this case). The rate of reaction was maximized by using a CO2 mole fraction of 0.2 in the calculations presented here. The precise starting value for CO2 concentration is not crucial for what follows as long as it is greater than the equilibrium value.

Flow of reactive aqueous fluid through the fictive parent body produces two mineralogically distinctive zones (Fig. 2). Toward the center of the body, upstream in the radial flow system, conversion of anhydrous silicate to hydrous silicate and carbonate is extensive. Downstream toward the outer surface, there is little or no alteration. The two regions are separated by a sharp discontinuity, or front, and by a zone of maximum alteration adjacent to the front (Fig. 2). Formation of zones separated by a front and an alteration maximum were found to be salient features of the solutions regardless of the exact values for initial26Al/27Al, porosity, and ice content.

Figure 2

Finite difference calculations portraying the temperature (in degrees kelvin), liquid flux [cubic meters of H2O per square meter per second], mineralogical alteration (progress of Eq. 4, ξ, in mole percent), and rock Δ17O evolution of a carbonaceous chondrite parent body composed of 20% by volume water ice over a period of 470,000 years. The ordinate for each plot is time in millions of years (My) after attainment of a temperature of 273 K at the core of the body.

Carbonate and hydrous minerals constitute <10% molar of the rocks even in the most highly altered zone adjacent to the reaction front in Fig. 2. The values for mole percent alteration shown in Fig. 2are averages for each model volume element. In real bodies, heterogeneous flow of fluid could have concentrated the <10% alteration into discrete, highly altered areas separated by alteration-free areas. Percentages of alteration minerals are in any case restricted by the finite supply of moving water when the volume of ice is comparable to the volume of pore space available in carbonaceous chondrites (20). The implication is that CI carbonaceous chondrites composed primarily of phyllosilicates, oxides, and carbonates must represent localized zones of aqueous alteration on the CI parent body (or bodies) and that such alteration was concentrated near the center of the body rather than near its surface.

The spatial pattern of isotopic alteration mimics the mineralogical pattern (Fig. 2). Minerals upstream of the alteration front in the interior of the model body have Δ17O approaching that of the original water ice, because the capacity of the rock to exchange oxygen isotopes with the fluid has been exhausted. Minerals downstream in the outer portion of the body retain the Δ17O of the original rock.

The finite difference calculations confirm the first-order predictions from Eq. 3. Downstream of the alteration front in the outer part of the fictive parent body, where ∇T is negative and fluid flux is persistent (radii >17 km, Fig. 2), the range of anhydrous mineral oxygen isotope ratios caused by exchange with water is analogous to the data for Allende (Fig. 3). The small volumes of new phyllosilicate and carbonate minerals (<1% by volume) produced in the outer portion of the body have isotopic compositions that are also on the rock mass fractionation curve and have δ18O values ranging up to 40 per mil (not shown in Fig. 3).

Figure 3

Comparison between the three-isotope compositions of mineral phases from the model parent body (black symbols); whole-rock and mineral compositions from the Allende CV, Murchison CM, and Orgueil CI carbonaceous chondrite meteorites (gray symbols) (22); and laser ablation analyses of Allende (6). Meteorite whole-rock compositions are represented by the large triangles. Model phyllosilicates are mixtures of talc and serpentine from the inner 17 km of the fictive body and represent highly altered rocks. Model forsterites come from the outer 8 km of the body and make up more than 99% of the minerals by volume in this largely unaltered region. Mineral compositions were sampled at 500-m intervals from the respective zones. The fluid mass fractionation curve is shown in gray. The Allende mass fractionation curve (which is also the model rock fractionation curve), the terrestrial mass fractionation curve, and the slope 1.00 line are in black. CI magnetites are consistent with the initial water ice used in the model (right-hand black star).

Upstream of the front within the interior of the model body (radii <17 km), hydrous silicates and carbonates are abundant in comparison to the small amounts of these minerals produced nearer to the surface. The δ17O and δ18O values of these new minerals approach the mass fractionation curve defined by water ice (Fig. 3). Forsterite is also driven toward the fluid Δ17O in this region (not shown in Fig. 3) but does not experience the shifts toward higher δ18O values along slope-1/2 lines as is the case downstream, because ∇T is small here. The pattern of oxygen isotope variability within the interior of our model parent body (radii <17 km) is similar to that defined by mineral separates from the Murchison CM and Orgueil CI carbonaceous chondrites (Fig. 3).

Figure 3 shows that the mineralogical and oxygen isotopic variability among the CV, CM, and CI groups of carbonaceous chondrites can be explained by a single process of fluid-rock reaction in parent bodies. This result is relatively insensitive to the initial isotopic value of the water ice (21). All that is required is one uniform oxygen reservoir for the bulk rock (Δ17O ∼ −3 per mil) and one uniform reservoir for the aqueous fluid (Δ17O ≥ 0.5 per mil).

Fluid-rock reactions within planetesimals may have been an important process before the final assembly of planets in the solar system. The loss of isotopically light oxygen from rock resulting from exchange with escaping water would have led to an increase in planetesimal δ17O and δ18O values (Fig. 4). In the model presented here, 25% of the water ice is lost from the body. The increase in rock δ18O is about 3 per mil, and the attendant increase in δ17O results in a slight increase in Δ17O of several tenths per mil. Subtraction of comparable effects from the bulk oxygen isotopic compositions of Earth and Mars (the martian meteorites) places the oxygen composition of progenitor planetesimals for the terrestrial planets on or near the slope 1.00 array (Fig. 4), removing the necessity for distinct primordial oxygen reservoirs other than the slope 1.00 array.

Figure 4

Three-isotope plot showing the evolution of terrestrial (⊕), martian (♂), and carbonaceous chondrite (CC) progenitor planetesimals resulting from aqueous alteration. The magnitude of shifts in δ17O and δ18O resulting from the finite difference modeling presented here is shown in the lower left corner. Similar shifts place the compositions of Earth and Mars (the martian meteorites) near the slope 1.00 line. The isotopic composition of primordial water ice was apparently above the terrestrial mass fractionation curve (TMF).

  • * To whom correspondence should be addressed. E-mail: ed.young{at}


Stay Connected to Science

Navigate This Article