Journal of Petrology Volume 42 Number 5 Pages 1019-1041 2001
© Oxford University Press 2001
Energy-Constrained Open-System Magmatic Processes II: Application of Energy-Constrained Assimilation–Fractional Crystallization (EC-AFC) Model to Magmatic Systems
WENDY A. BOHRSON,* and
FRANK J. SPERA
INSTITUTE FOR CRUSTAL STUDIES AND DEPARTMENT OF GEOLOGICAL SCIENCES, UNIVERSITY OF CALIFORNIA, SANTA BARBARA, CA 93106, USA
Received
January 10, 2000;
Revised typescript accepted
August 17, 2000
 |
ABSTRACT
|
|---|
Evidence for open-system magmatic processes is abundant in igneous
rocks from most tectonic settings and with ages spanning most
of geologic time. Accurately documenting these processes is
critical for understanding magma reservoir dynamics, including
the processes that lead to compositional diversity in igneous
rocks, and for deciphering the thermochemical evolution of the
crust and mantle. Quantitative models describing open-system
processes such as assimilation–fractional crystallization
(AFC) have provided significant insight into all of these, but,
nevertheless, suffer from several serious deficiencies. Foremost
among these are the absence of energy conservation and the lack
of consideration of country rock partial melting. For a magma
body undergoing AFC, a new quantitative model,
Energy-
Constrained
Assimilation
Fractional
Crystallization (EC-AFC), self-consistently
balances energy, species and mass while also tracking compositional
variations generated in anatectic melt as country rock undergoes
partial melting. EC-AFC represents a significant improvement
to existing AFC models for several reasons. First, the inclusion
of energy conservation provides a direct and crucial link between
thermal parameters and volcanological or geological data. Second,
unlike ‘classical’ AFC that models mass and chemical
properties only, EC-AFC models mass, chemical and thermal properties
of a magma body, thus allowing the energetics of the open-system
magma reservoir to be linked to the geochemical evolution. Third,
compared with ‘classical’ AFC models, EC-AFC geochemical
trends are distinct, exhibiting non-monotonic behaviors that
are directly linked to the effects of energy conservation and
country rock partial melting. Comparison of EC-AFC trends with
data from natural systems indicates that some of the criteria
currently used to demonstrate the efficacy of AFC require modification.
Finally, comparison of ‘classical’ AFC and EC-AFC
results for data from well-documented volcanic centers demonstrates
that EC-AFC does a superior job of tracking the compositional
trends, provides a plausible physical context for the process
of AFC, and allows geologically relevant predictions to be made
about particular magmatic systems.
KEY WORDS: assimilation–fractional crystallization; geochemical model; isotope; magma chamber; trace element
|
INTRODUCTION
|
|---|
Geochemical, petrologic and petrographic data for igneous rocks
provide definitive evidence for the occurrence of open-system
processes (e.g. assimilation of anatectic melt, magma recharge,
interaction between magma and fluids or magma and hydrothermally
altered country rock) in evolving magma bodies. Evidence for
such processes has been documented in igneous rocks from most
tectonic settings and with ages spanning most of geologic time
(Sparks, 1986
). Understanding open-system processes is therefore
critical for evaluating magma chamber processes and the thermochemical
evolution of the mantle and crust. The significant role that
open-system processes play in generating the compositional diversity
evident in igneous rocks has been recognized for many years
(e.g. Bowen, 1922
a, 1922
b, 1928
), and numerous quantitative
treatments have been developed describing the geochemical consequences
of these processes (e.g. Allègre & Minster, 1978
;
Taylor, 1980
; DePaolo, 1981
; O’Hara & Mathews, 1981
;
Albarède, 1995
). During the last 30 years, significant
progress has been made in our understanding of the dynamics
of open-system behavior primarily because of improvements in
analytical instrumentation, recognition of the critical information
provided by trace elements and radiogenic and stable isotopes,
and appreciation of the importance of integrating the thermal
and chemical characteristics of the magma body with the geologic
history of the associated complex.
Despite these great strides, notable deficiencies exist in quantitative treatments of open-system processes. Foremost among these are the lack of a comprehensive accounting of species and mass conservation that is self-consistently coupled to energy conservation and lack of consideration of assimilant compositional variations that result from partial melting of country rock. These deficiencies are especially acute when one notes that the interaction of magma with its host rock is essentially a thermal process. We suggest that geochemical trends resulting from application of species and mass balance equations may be flawed and may lead to incorrect conclusions about the petrogenetic histories of some igneous rock suites. In recognition of this, we have developed a model of mass, species and energy conservation for a magma body undergoing open-system evolution. In a companion paper (Spera & Bohrson, 2001), we present the conceptual framework of an energy-constrained open-system model and describe the mathematics and assumptions of the Energy-Constrained Assimilation Fractional Crystallization (EC-AFC) formulation (available at http://magma.geol.ucsb.edu/research/recharge.html). In this paper, using results of selected EC-AFC simulations, we demonstrate that the EC-AFC formulation is a notable improvement over existing models for several reasons. First, the inclusion of energy conservation provides a direct link between thermal parameters (e.g. initial liquidus and solidus temperatures of magma and country rock) and volcanological and geological data (e.g. depth of the magma reservoir, magma eruption temperature); such a link is critical but generally lacking in current geochemical models. We also show that, compared with ‘classical’ AFC models (herein defined as models based on only mass and species conservation), distinct geochemical trends may emerge for EC-AFC. On the basis of a comparison of results from EC-AFC and ‘classical’ AFC, we show that some of the generalizations made about the process of AFC are in need of modification. As a consequence, some of the broad-scale conclusions about the petrogenetic histories of particular igneous suites may be inaccurate. Finally, because the value of any theoretical model lies in its ability to describe natural systems, in the last part of this paper, we evaluate three published datasets and demonstrate that EC-AFC results do a better job of modeling the observed geochemical trends and of constraining the physicochemical processes associated with AFC. The overarching goal of incorporating energy conservation and country rock partial melting into a quantitative description of AFC is to provide a more realistic characterization of magmatic processes by explicitly coupling thermal and chemical properties of a magma body. By examining the magma body and country rock as a composite system governed by physicochemical principles, a more comprehensive understanding of the dynamics of magma plumbing systems and the origins of chemical diversity of magmas will emerge.
|
RESULTS OF SELECTED EC-AFC SIMULATIONS
|
|---|
Here, we present selected results of EC-AFC simulations to illustrate
physical (e.g. temperature of magma body and country rock, mass
of magma body) and chemical characteristics (trace element and
isotopic signatures) of two magma body–country rock systems
as well as highlight some of the differences between ‘classical’
AFC and EC-AFC results. Incorporated into this analysis are
results of sensitivity tests that illustrate examples of potential
compositional variability introduced by independently varying
particular thermal or chemical input parameters. Although a
complete sensitivity analysis is not presented, any and all
permutations of interest can be easily and rapidly handled by
the EC-AFC program. The two composite systems are a ‘standard’
upper-crustal case and a ‘standard’ lower-crustal
case (Table
1). The upper-crustal example is illustrative of
basaltic magma intruded at a liquidus of 1280°C into upper
crust of ambient temperature 300°C (depth
10 km). The composition
of the crust is roughly granitic, with a liquidus temperature
of 1000°C, and the implied water content of anatectic melt
is a few per cent by mass. The local solidus temperature,
Ts,
is 900°C. The geochemical parameters correspond to values
typical of upper crust (e.g. Taylor & McLennan, 1985
). The
‘standard’ lower-crustal case represents intrusion
of more primitive basaltic magma with a liquidus of 1320°C
into lower crust of ambient temperature 600°C (depth
20
km). The crust is mafic, with a liquidus temperature of 1100°C.
The local solidus temperature is 950°C. The geochemical
parameters correspond to typical trace element and isotopic
values for lower crust (e.g. Taylor & McLennan, 1985
). For
the purposes of comparison,
Teq = 980°C for both cases.
In both, melt productivity is a linear function of temperature
for magma and country rock; that is,
fm(
Tm) = (
T -
Ts)/(
Tl,m -
Ts) and
fa(
Ta) = (
T -
Ts)/(
Tl,a -
Ts) [for explanation of
nomenclature, see table 1 of Spera & Bohrson (2001)
]. Isobaric
specific heats, heat of fusion and heat of crystallization for
magma and country rock were computed from data in tables 2 and
3 of Spera & Bohrson (2001)
.
Physical characteristics of the magma body–country rock system
The EC-AFC formulation permits an accounting of thermal and mass characteristics during the AFC ‘event’, including the temperatures of the magma body and country rock, the masses of melt and cumulates in the magma body, and the mass of anatectic melt assimilated into the magma body. Such parameters are critical for developing an accurate understanding of the dynamics of magma reservoirs. For the ‘standard’ upper-crustal case, the trajectory in Tm-Ta coordinates is shown in Fig. 1a. Sensible and latent heat liberated by the magma as it cools and crystallizes goes into sensible heating of the country rock. For each degree drop in melt temperature (Tm), the average temperature of country rock (Ta) rises by 3·5°C. During this stage of the AFC ‘event’, on average, anatectic melt is not present because Ta < Ts, and the geochemical evolution of the magma body is driven by fractional crystallization. For Ta < Ts, the fraction of melt (Mm) decreases as cumulates form. Unlike ‘classical’ models in which energy conservation is ignored, however, Mm is not a monotonically decreasing function of Tm. At Ta > Ts, because of the addition of anatectic melt, Mm can increase as Tm continues to drop towards Teq (Fig. 1a). Also, at Ta > Ts, Ta rises more slowly because of the enthalpy requirements of anatexis (Fig. 1a). The mass of anatectic melt delivered to the magma body (Ma*) and mass of cumulates removed by fractional crystallization (Mc) are portrayed in Fig. 1b. When Ta < Ts (no anatectic melt), the ratio of the mass of anatectic melt to the mass of cumulates (Ma*/Mc) is zero but increases to 0·7 as country rock heats up to Teq (Fig. 1c). This variation should be noted. It should be recalled that in ‘classical’ AFC, Ma*/Mc (the parameter r) is a constant. Trends for the ‘standard’ lower-crustal case (Fig. 1a–c) are qualitatively similar to those of the upper-crustal case, although the ratio Ma*/Mc has a smaller range (0–0·35) throughout the AFC ‘event’, reflecting both the higher solidus and the greater fusion enthalpy of lower compared with upper crust. One additional example is illustrated where all parameters are the same as those of the ‘standard’ upper-crustal case except that the melt production curves for both magma and country rock are taken as nonlinear functions of temperature (referred to as ‘nonlinear’ upper-crustal case; Fig. 2). The distinct T–melt fraction relations introduce differences in the mass of anatectic melt assimilated and the magma–country rock temperature profile (Fig. 1a–c). In addition, as noted below, distinct differences in composition also result.
View larger version (19K):
[in this window]
[in a new window]
|
Fig. 2. Melt productivity (fraction of melt–temperature relations) for magma body (fm) and country rock (fa) for ‘nonlinear’ upper-crustal case. (Note that the x-axis is the magma body temperature.)
|
|
Isotope–element and element–element trends
Geochemical trends unlike those associated with ‘classical’ AFC are well illustrated by examining Sr concentration ([Sr]) vs 87Sr/86Sr trends for three EC-AFC cases (Fig. 3a). The first and second are the ‘standard’ upper-crustal case and the ‘nonlinear’ upper-crustal case, both of which model Sr as compatible in magma and country rock (Table 1a; Dm, Da for both = 1·5). The third illustrates the ‘standard’ lower-crustal case, where country rock Sr is modeled to behave incompatibly (Table 1b; Da = 0·05). Figure 3a shows that in all cases, [Sr] initially decreases whereas no change in 87Sr/86Sr occurs, leading to a flat trajectory in element–isotope space. This is because no anatectic melt forms as country rock heats up from its initial temperature (300°C, 600°C) to its solidus (900°C, 950°C); during this stage, the geochemical evolution of the magma body is characterized by only fractional crystallization. Although assimilation has not affected the geochemistry of the magma body, heat is being transferred from the magma body to country rock. Thus, although ‘classical’ AFC solutions would never associate this part of the path with AFC, it is an integral part of the process. Once country rock reaches its solidus, partial melt is generated and mixed into the magma body. When partial melting begins, in the upper-crustal cases, [Sr] of contaminated magma continues to decrease while 87Sr/86Sr becomes more radiogenic, consistent with the compatible nature of Sr in both country rock and magma and with incorporation of more radiogenic assimilant. It should be noted that although the patterns for the ‘standard’ and ‘nonlinear’ upper-crustal cases are similar, Fig. 3a reveals that for the ‘nonlinear’ case, 87Sr/86Sr and [Sr] are equal to 0·7200 and 310 ppm at Teq = 980°C, whereas for the ‘standard’ case, the values are 0·7135 and 318 ppm, respectively. These differences underscore the strong dependence of geochemical path on parameters included in the EC-AFC formulation; temperature–melt fraction relations as well as other thermodynamic parameters influence the extent and characteristics of magma contamination. Thus, one of the single most important conclusions of this work is that the energetics of petrologic processes is a profound constraint that must be applied to study complex AFC phenomena; AFC without self-consistent application of energy conservation is like an automobile without an engine.
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 3. (a) Comparison of Sr (ppm) vs 87Sr/86Sr results for EC-AFC models of ‘standard’ (•) and ‘nonlinear’ ( ) upper-crustal and ‘standard’ ( ) lower-crustal cases and ‘classical’ upper-crustal ( ) and lower-crustal ( ) cases. EC-AFC parameters in Table 1. For EC-AFC trends, arrows illustrate direction of falling Tm, and each symbol represents a normalized temperature increment of 0·02 (30°C drop in Tm for all cases). For ‘classical’ AFC trends, symbols represent a fraction of melt (F) increment of 0·1 but are terminated at F = 0·05. Where appropriate, parameters for ‘classical’ AFC trends are the same as those of EC-AFC. For the upper-crustal case, r = 0·33 and for the lower-crustal case, r = 0·17, which are average Ma*/Mc. (b) Concentration of Sr in anatectic melt generated from country rock undergoing fractional melting, where Sr behaves incompatibly (Da = 0·05; ‘standard’ lower-crustal case), and Sra° is 230 ppm. Sr concentration shown for fa(Ta) up to 1·00. For ‘standard’ lower-crustal case, fa(Ta) at Teq is 0·21.
|
|
For the ‘standard’ lower-crustal case, at
Ta >
Ts, [Sr] increases while
87Sr/
86Sr increases because during
fractional melting of an incompatible element, the smallest
(lowest temperature) melt fractions have very high Sr concentrations
(Fig. 3b). Thus, because of the highly incompatible behavior
of Sr in the lower-crustal case, more Sr is being added to the
contaminated magma by assimilation than is being removed by
fractional crystallization. Therefore, both [Sr] and
87Sr/
86Sr
increase. By
10% melting of country rock (
F = 0·10),
[Sr] in anatectic melt is less than that in the magma body itself
(
880 ppm Sr in contaminated magma body when country rock is
melted 10%), and, therefore, the Sr concentration in the contaminated
magma decreases in response to fractional crystallization and
the ‘dilution’ effect of adding low-Sr anatectic
melt. At this stage, because so little assimilant Sr is being
contributed to contaminated magma,
87Sr/
86Sr changes very little.
This part of the modeled EC-AFC process is reflected in the
relatively flat trajectory where
87Sr/
86Sr is
0·709 and
[Sr] decreases from
880 ppm to
530 ppm (Fig. 3a).
Despite the lack of variation in 87Sr/86Sr, it is important to keep in mind that assimilation is still continuing. Because such a trajectory cannot be modeled by ‘classical’ AFC, the lack of variation in some isotope ratios may lead to the misinterpretation that the associated magmas are the products of dominantly closed-system fractional crystallization. For example, some silicic volcanic suites that have isotope signatures that show little variation with indices of differentiation have been described as the products of dominantly fractional crystallization (i.e. low rates of assimilation, Wark, 1991; Grunder, 1992). The critical point is that some EC-AFC simulations predict only minor variations in some isotope ratios during the later stages of the AFC process when the total mass of assimilated material reaches its maximum, and suites of differentiated rocks with such signatures should be evaluated with this possibility in mind. Because of the abundance of O in most magmas, 18O/16O typically does not exhibit a flat trajectory at this stage in the AFC ‘event’, provided there is a contrast between the oxygen isotope signatures of magma and country rock; for this reason, characterizing oxygen isotopes in some suites of igneous rocks is particularly important.
Model results calculated using equations from DePaolo (1981) highlight the remarkable and very significant differences between ‘classical’ AFC and EC-AFC (Fig. 3a). Where appropriate, input parameters are the same, and r is estimated based on an average Ma*/Mc. Among the most important differences are the lack of a flat Sr–87Sr/86Sr trajectory at 0·7035 and the lack of the non-monotonic trajectory illustrated by the lower-crustal EC-AFC model.
Another interesting pattern in Sr isotope–[Sr] space is revealed by examining cases where Sr in both magma and country rock behaves compatibly, but [Sr] in the country rock is similar to or greater than that in the uncontaminated, unfractionated magma. Three cases are shown, where ‘EC-AFC, upper’ represents the ‘standard’ upper-crustal case, and ‘EC-AFC, 700’ and ‘EC-AFC, 1000’ are the ‘standard’ upper-crustal case but with the initial country rock Sr concentration, Sra°, equal to 700 ppm and 1000 ppm, respectively. Once wallrock partial melting begins, [Sr] in the ‘standard’ upper-crustal case decreases as magma temperature drops (Fig. 4a). [Sr] in ‘EC-AFC, 700’ initially decreases slightly but then increases. The changes in concentration reflect the effect that fractional melting has on a compatible element (Fig. 4b). At small degrees of partial melting, relatively little Sr is liberated from country rock into the anatectic melt. As the degree of melting increases, more Sr is partitioned into anatectic melt, until eventually, in ‘EC-AFC, 700’, proportionally more is being added by assimilation than is being removed by fractional crystallization. In ‘EC-AFC, 1000’, [Sr] increases as Tm drops, reflecting the effect of both fractional melting and the high Sra°. In this example, the high Sra° allows the contaminated magma 87Sr/86Sr to achieve relatively radiogenic values at fairly high concentrations of Sr, despite a large amount of fractional crystallization that has occurred (Mc 0·79 at Teq for all three cases). For similar input parameters, positive correlation between 87Sr/86Sr and [Sr] cannot be produced by ‘classical’ AFC models (Fig. 4a) because partial melting of country rock is not accommodated.
View larger version (17K):
[in this window]
[in a new window]
|
Fig. 4. (a) Comparison of Sr (ppm) vs 87Sr/86Sr results for EC-AFC and ‘classical’ AFC models of upper-crustal contamination, where Sra° varies. Arrows same as in Fig. 3a. (b) Concentration of Sr in anatectic melt generated from country rock undergoing fractional melting, where Sr behaves compatibly (Da = 1·5) and Sra° varies. Sr concentration shown for fa(Ta) up to 0·95. For ‘standard’ upper-crustal case, fa(Ta) at Teq is 0·86.
|
|
The behavior of an element, such as Nd, that may be incompatible
in both magma and country rock is illustrated in Fig.
5, which
depicts results for the ‘standard’ and ‘nonlinear’
upper-crustal and ‘standard’ lower-crustal cases.
For all, the Nd concentration ([Nd]) initially increases whereas
143Nd/
144Nd remains unchanged, reflecting heating of country
rock to
Ts. When anatexis begins, [Nd] increases as
143Nd/
144Nd
decreases, consistent with its incompatible behavior in both
magma and country rock and addition of anatectic melt that is
characterized by a less radiogenic Nd isotope signature. The
lower-crustal case has a more protracted period of wallrock
heating (and fractional crystallization only) because of its
higher
Ts, and therefore it has higher [Nd] when anatexis commences.
Fractional melting eventually depletes Nd from the country rock
residue. Therefore, anatectic melt has very little Nd, and its
concentration in the contaminated magma decreases despite Nd
behaving incompatibly during fractional crystallization. In
the case of both upper-crustal examples, the relatively flat
element–isotope trajectory at low
143Nd/
144Nd is due to
the small amount of assimilant Nd that is being added to contaminated
magma (Fig.
5); in the lower-crustal case,
Teq is reached before
this phenomenon occurs. It should be noted, again, that the
‘standard’ and the ‘nonlinear’ upper-crustal
cases exhibit different [Nd]–
143Nd/
144Nd characteristics.
Also, the non-monotonic trends are not generated by ‘classical’
AFC models (Fig.
5).
View larger version (27K):
[in this window]
[in a new window]
|
Fig. 5. Comparison of Nd (ppm) vs 143Nd/144Nd results for EC-AFC models of ‘standard’ and ‘nonlinear’ upper-crustal and ‘standard’ lower-crustal cases and ‘classical’ upper- and lower-crustal cases. Symbols and arrows same as in Fig. 3a.
|
|
The complex geochemical trends illustrated above are fingerprints of the EC-AFC process, reflecting the constraint of energy conservation and the effects of partial melting. In particular, the lack of variation in isotopic ratio of a trace element as its concentration varies is a tell-tale signal of an EC-AFC process and contrasts with ‘classical’ AFC trends where monotonic trends are typical. The distinctions in EC-AFC vs ‘classical’ AFC trends (Figs 3a, 4a and 5) underscore an important point. In some studies, AFC has been rejected because observed data trends are not consistent with modeled ‘classical’ AFC trends (e.g. Taylor, 1980; Carlson et al., 1981; Lum et al., 1989; Kempton et al., 1991; Blichert-Toft et al., 1992). It is critical to appreciate that the ability to determine whether AFC has occurred depends—at least in part—on the formulation used to quantify the process. A concrete example of this involves the use of 1/Sr vs 87Sr/86Sr plots. Because a monotonic 1/Sr vs 87Sr/86Sr trend is predicted by ‘classical’ AFC models, its absence in data trends from particular volcanic centers has been interpreted as evidence against AFC (e.g. Lum et al., 1989; Fitton et al., 1991; Kempton et al., 1991). It is clear, however, that such a trend will not result in the cases illustrated in Figs 3a and 4a, and yet these clearly are data trends that reflect one possible mode of AFC. Another example of model-dependent results involves the interpretation that the most differentiated sample is also the most contaminated sample. In ‘classical’ AFC, the most differentiated sample is typically regarded as that with the highest concentration of a particular incompatible trace element and that has undergone the greatest amount of fractional crystallization. The extent of contamination is most commonly assessed with radiogenic or stable isotope ratios, with the most contaminated being that sample with an isotope signature most like that of country rock. For AFC to be considered as an important process, many workers require that a good correlation between these two exist (e.g. Carlson et al., 1981; Marsh, 1989; Chazot & Bertrand, 1993; Mason et al., 1996). This filter has developed primarily because solutions to ‘classical’ AFC equations yield monotonic changes in incompatible trace element concentration vs isotope ratio. EC-AFC demonstrates that this monotonic correlation is not always present. For all EC-AFC simulations, magma at Teq has assimilated the largest mass of anatectic melt, but as illustrated in Fig. 5, magma with the highest [Nd] has not assimilated the most anatectic melt. The element–element patterns shown in Fig. 6a and 6b further emphasize this point. Magma that has assimilated the largest mass of anatectic melt does not have the highest concentration of Th or Nd (both modeled as incompatible in magma and country rock). On the basis of a comparison with trends calculated using ‘classical’ AFC equations (Fig. 6), the non-monotonic EC-AFC trends shown for Nd–Sr and Nd–Th would either not be interpreted as AFC trends or several different contaminants would be invoked to explain the range of abundances. Thus, conclusions regarding the importance of AFC based on the absence of simple correlation between degree of differentiation and amount of contamination may not be correct; indeed, the assumption that the sample with the highest concentration of incompatible element is the most differentiated (e.g. has undergone the most fractional crystallization) is not necessarily correct. The critical point is that although comparison between data and quantitative models forms a backbone of geochemical and petrological studies, the examples described here illustrate that potential misinterpretations may arise in model-dependent studies. In view of the very distinct differences that can occur between EC-AFC and ‘classical’ AFC geochemical trends, some case studies in which the process of AFC has been dismissed may merit re-examination.
Mass balance arguments
Mass balance arguments have been used to suggest that relatively radiogenic Sr isotope signatures of basalts cannot result from the process of AFC because addition of the mass of (differentiated) assimilant required to generate relatively radiogenic isotope signatures would yield magmas that are no longer basaltic. This argument has been used in interpretations of the origin(s) of extension-related basalts in the Western USA (Lum et al., 1989; Kempton et al., 1991). For the ‘standard’ lower-crustal case, with the addition of a mass of assimilant that is 10% of the total mass of original magma (Ma*/Mo 
0·1), 87Sr/86Sr climbs to >0·708 from its initial value of 0·7035 and [Sr] increases to 880 ppm (Fig. 7a and b). Depending on the composition of the assimilant, the addition of such a small mass may not significantly change the major element composition of the magma body. For example, addition of 10% assimilant with 70 wt % SiO2 would only change SiO2 of a basalt with 50 wt % by roughly 2 wt %. Certainly, a magma with 52 wt % and a Sr concentration of 880 ppm would be considered ‘basaltic’. A ‘classical’ AFC calculation using the same input parameters, where appropriate, requires an r of 0·45 to achieve 87Sr/86Sr of 0·7084 at a similar degree of crystallization (Mc 0·8). Compared with the EC-AFC results, ‘classical’ AFC requires addition of a factor of 3·5 more assimilant by mass. Although the effect on composition is strongly dependent on input parameters, this EC-AFC result indicates that relatively radiogenic Sr isotope signatures can be achieved through addition of small amounts of crustal assimilant that may not drastically alter the major element composition of the magma body. Mass balance arguments based on ‘classical’ AFC results may overestimate the required amount of assimilant and thereby lead to incorrect conclusions regarding the origin of ‘enriched’ signatures in basalts.
View larger version (14K):
[in this window]
[in a new window]
|
Fig. 7. (a) Comparison of Ma*/Mo vs Sr (ppm) results; (b) Ma*/Mo vs 87Sr/86Sr results for EC-AFC models of ‘standard’ upper-crustal and lower-crustal cases. Symbols and arrows same as in Fig. 3a.
|
|
Arguments have also been made about the ability of the process
of AFC to affect magmas characterized by high Sr concentrations
(Lum
et al., 1989
; Kempton
et al., 1991
) or other trace elements
(i.e. magmas with high abundances of incompatible trace element
may be considered insensitive to AFC; e.g. Davidson
et al.,
1990
). Figure 8a and b illustrates results of three upper-crustal
simulations; ‘EC-AFC, upper’ is the ‘standard’
upper-crustal model whereas ‘EC-AFC, 233’ and ‘EC-AFC,
175’ represent the ‘standard’ upper-crustal
model but with Sr
a° of 233 ppm and 175 ppm, respectively.
These values were chosen to reflect Sr
m°/Sr
a° of 2/1,
3/1 and 4/1, respectively. Modeled with these parameters, with
less than 20% addition of assimilant relative to the original
mass of magma (
Ma*/
Mo 0·2), the contaminated magma is
characterized by [Sr] that exceeds 300 ppm and by
87Sr/
86Sr
that is greater than
0·7055 (Fig. 8a and b). For the
‘standard’ lower-crustal case, where Sr is incompatible
in the country rock and Sr
m°/Sr
a° is 2/1, 3/1 or 4/1,
87Sr/
86Sr of contaminated magma is
0·708 and [Sr] >
700 ppm for just 10% addition of assimilant (Fig. 8c and d).
Thus, despite relatively high Sr
m°/Sr
a°, contaminated
magma Sr isotope signatures can be profoundly changed with addition
of a relatively small mass (
20% of total mass of original magma)
of assimilant.
View larger version (34K):
[in this window]
[in a new window]
|
Fig. 8. (a) Comparison of Ma*/Mo vs Sr (ppm) results for EC-AFC models of ‘standard’ upper-crustal contamination where the initial concentration of Sr in the country rock, Sra°, varies. Concentrations were chosen to reflect Srm°/Sra° of 2/1 (upper), 3/1 (233 ppm), 4/1 (175 ppm). Srm° is constant at 700 ppm. (b) Comparison of Ma*/Mo vs 87Sr/86Sr results for EC-AFC models of ‘standard’ upper-crustal contamination. Different trends reflect variations in Srm°/Sra°. Sr isotope ratios of magma and country rock are constant at 0·7035 and 0·7220, respectively. (c) Comparison of Ma*/Mo vs Sr (ppm) results for EC-AFC models of ‘standard’ lower-crustal contamination where Sra° varies. Concentrations were chosen to reflect Srm°/Sra° of 2/1 (350 ppm), 3/1 (lower), 4/1 (175 ppm). Srm° is constant at 700 ppm. (d) Comparison of Ma*/Mo vs 87Sr/86Sr results for EC-AFC models of ‘standard’ lower-crustal contamination. Different trends reflect variations in Srm°/Sra°. Sr isotope ratios of magma and country rock are constant at 0·7035 and 0·7100, respectively. Symbols and arrows same as in Fig. 3a.
|
|
Figure
9 shows that there is not necessarily a simple correlation
between mass of anatectic melt assimilated (
Ma*/
Mo) and the
geochemical signature of the magma body. This is in distinct
contrast to ‘classical’ AFC results that predict
that, all other parameters being equal, the more mass assimilated
(higher
r), the more crust-like the isotope ratios will be (e.g.
Figs 3a and 5a). All other parameters being equal in the ‘standard’
upper-crustal case (
Teq = 980°C), different
Teq (950°C,
920°C, Fig.
9) yield differences in the mass of anatectic
melt assimilated. The 920°C case involves the largest mass
of country rock,
Ma°, but the smallest proportion of assimilant,
Ma*/
Mo. This is because a
Teq of 920°C is closest to
Ts,
and therefore, at
Teq, fa(
Ta) is the smallest of the three examples.
Energy heats up a large mass of country rock but only melts
it to a small degree. In contrast, the 980°C case has the
smallest
Ma°, the largest
fa(
T), and the largest
Ma*/
Mo.
Figure 9b shows that the 920°C example produces magma with
the most contaminated Sr isotope signature (
0·7160),
whereas the 980°C example is characterized by magma with
the least contaminated Sr isotope signature (
0·7140).
For the 920°C case, assimilation initiates at the lowest
Tm because
Ma° is the largest. When country rock partial
melting commences, magma in the 920°C example has undergone
the most fractional crystallization, and as a consequence, [Sr]
in the magma body is lowest of the three cases (
250 ppm, Fig.
9a). Country rock Sr makes up a greater proportion of Sr in
the contaminated magma body, and therefore,
87Sr/
86Sr is the
most radiogenic. Nd exhibits different behavior. The 920°C
case yields magma with the highest [Nd] (Fig. 9c) whereas the
most contaminated (crustal) Nd isotope signature is the 950°C
case (Fig. 9d). The explanation for this lies in the complex
interplay between the incompatible behavior of Nd in both magma
and country rock and the effects of fractional melting.
View larger version (32K):
[in this window]
[in a new window]
|
Fig. 9. (a) Comparison of Ma*/Mo vs Sr (ppm) results; (b) Ma*/Mo vs 87Sr/86Sr results; (c) Ma*/Mo vs Nd (ppm) results; (d) Ma*/Mo vs 143Nd/144Nd results for EC-AFC models of ‘standard’ upper-crustal case, where Teq varies. Symbols and arrows same as in Fig. 3a.
|
|
Variations in compositionally dependent thermal parameters
It is well known that liquidus and solidus temperatures, specific heats, and heats of crystallization and fusion are broadly dependent on composition. Changes in any of these parameters have consequences on EC-AFC chemical trends. For example, for the ‘standard’ upper-crustal case where all other parameters are fixed, raising the liquidus temperature of country rock by 200°C (i.e. Tl,a = 1200°C, more mafic country rock) decreases the proportion of anatectic melt that is incorporated (Ma*/Mo) from 0·55 to 0·21. More energy is required to reach the solidus temperature and initiate anatexis because the total mass of country rock (Ma°) involved in the AFC ‘event’ is larger, which leads to a smaller Ma*/Mo. In this case, compared with the ‘standard’ upper-crustal case, at Teq, the magma body displays a less contaminated Sr isotope signature and a similar Nd isotope signature (Fig. 10a and b). In contrast, raising the liquidus and initial temperatures of uncontaminated magma (Tl,m, Tm° = 1500°C), which is equivalent to intruding a more mafic magma, yields a higher Ma*/Mo (Fig. 10a and b). More heat is generated by the magma body as it cools and crystallizes, and therefore more heat is available for anatexis, which leads to a larger Ma*/Mo (0·85). Compared with the ‘standard’ upper-crustal case, both Sr and Nd isotopes are more crust-like.
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 10. (a) Comparison of Ma*/Mo vs 87Sr/86Sr results; (b) Ma*/Mo vs 143Nd/144Nd results for EC-AFC models of ‘standard’ upper-crustal case, where Tl,a and Tl,m, Tm° vary. Symbols and arrows same as in Fig. 3a.
|
|
Changing the solidus temperature also influences the chemical
evolution of the magma body. Figure
11 illustrates results of
two cases, the ‘standard’ upper-crustal case and
a case in which all other parameters are equal but the solidus
is reduced to 800°C. For this comparison, the lower solidus
allows assimilation to begin at a higher
Tm. Despite assimilating
very similar total proportions of anatectic melt (‘standard’
Ma*/
Mm° = 0·55 vs 0·54 for the 800°C case)
the ‘standard’ case ultimately yields magma characterized
by more crustal isotope signatures. For
Ts = 800°C, because
assimilation initiates at a higher
Tm, less fractional crystallization
has occurred. [Sr] is therefore higher when assimilation commences,
and because of this, at
Teq, magma is characterized by a less
radiogenic Sr isotope signature (Fig. 11a). Compared with the
‘standard’ case, [Nd] in the 800°C case is less
enriched when assimilation commences (Fig. 11b), and this allows
the ‘dilution’ effect to occur at a higher
Tm. Therefore,
143Nd/
144Nd becomes relatively constant at a less crustal value
compared with the ‘standard’ upper-crustal case.
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 11. (a) Comparison of Sr (ppm) vs 87Sr/86Sr results; (b) Nd (ppm) vs 143Nd/144Nd results for EC-AFC models of ‘standard’ upper-crustal case, where Ts varies. Symbols and arrows same as in Fig. 3a.
|
|
Variations in parameters that reflect initial thermal conditions of country rock
The initial temperature of the country rock (Ta°), in part, reflects the ambient geothermal gradient (Spera & Bohrson, 2001). Figure 12 illustrates results of two cases, the ‘standard’ upper-crustal case and the same but with the Ta° elevated to 600°C. A possible scenario to explain this higher temperature would be regional heating of upper crust as a result of mafic underplating. We stress that the 600°C case discussed here is different from the ‘standard’ lower-crustal case where thermal and compositional parameters have been selected to reflect assimilation of more mafic lower crust. The proportion of anatectic melt that is assimilated is greater in the 600°C case (Ma*/Mo 0·85) than in the ‘standard’ case (Ma*/Mo 0·55). Because less heat is required to elevate country rock to Ts, assimilation commences at a higher Tm. The larger Ma*/Mo associated with the 600°C case, compared with the ‘standard’ case, yields more radiogenic 87Sr/86Sr (Fig. 12a) and less radiogenic (more crustal) 143Nd/144Nd (Fig. 12b). The 600°C case also provides evidence that ratios of mass of assimilated melt to mass of cumulates removed (equivalent to r, DePaolo, 1981) can exceed unity: at Teq, Ma*/Mc for the 600°C case is 1·1. On the basis of energy conservation, there is no justification for arbitrarily limiting this parameter to less than unity.
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 12. (a) Comparison of Sr (ppm) vs 87Sr/86Sr results; (b) Nd (ppm) vs 143Nd/144Nd results for EC-AFC models of ‘standard’ upper-crustal case where Ta° varies. Symbols and arrows same as in Fig. 3a.
|
|
Summary
The results presented in this section clearly demonstrate two critical points about EC-AFC: (1) thermochemical parameters that characterize magma bodies are complex and interconnected; inclusion of energetic constraints in modeling the geochemistry of rocks influenced by complex AFC processes is therefore an absolute necessity. (2) The incorporation of energy conservation and country rock anatexis into a formulation of AFC yields geochemical trends that can be very distinct from those predicted by ‘classical’ AFC; therefore, some of the generalizations made about the process of AFC merit re-evaluation, which, in turns indicates that some datasets may need to be re-examined.
|
APPLICATION OF EC-AFC TO NATURAL SYSTEMS
|
|---|
A critical test of the applicability of the EC-AFC formulation
is its ability to explain data from natural systems. Below,
results of EC-AFC simulations are presented for geochemical
data from three volcanic systems that are well documented with
respect to their eruptive histories and petrographic and geochemical
characteristics. Each of the associated magmatic systems was
affected by AFC, and particular data trends for each display
non-monotonic trace element–isotope behavior. Of particular
note in each dataset is the Sr vs
87Sr/
86Sr trend. From our
comparison of ‘classical’ vs EC-AFC modeling results,
we conclude that, in general, EC-AFC does a superior job of
modeling the compositional evolution, and it allows geologically
relevant predictions to be made about the magmatic systems.
Mafic to intermediate volcanic rocks from the Long Valley caldera and Devil’s Postpile National Monument, eastern California
Abundant mafic to silicic volcanism characterizes the region near Long Valley caldera, the site of caldera collapse associated with the eruption of the 0·73 Ma Bishop Tuff. Of specific interest in this region is a suite of Quaternary (0·151–0·415 Ma; Vogel et al., 1994) basalt to trachyandesite volcanic rocks exposed in the west moat of Long Valley caldera and in Devil’s Postpile National Monument, which is located several kilometers to the west of the caldera. Vogel et al. (1994) analyzed continuous core from the Inyo-4 drill hole, located in the west moat, for major and trace elements. Cousens (1996) analyzed surface samples from the west moat and Devil’s Postpile area for major and trace elements. In addition, Cousens analyzed the surface and drill hole samples for Sr, Nd, and Pb isotopes. Typically, samples from both suites are phyric (1–20 vol. %), with plagioclase > olivine > clinopyroxene (Cousens, 1996). SiO2 abundances range from 48 to 58 wt %, and MgO ranges from 2·6 to 7·5 wt %. Drill core samples, which define a smooth trend of more differentiated compositions up-section, are interpreted to be products of periodic eruption of a continuously evolving magma body (Vogel et al., 1994). Vogel et al. (1994) suggested that differentiation of the oldest, most mafic lavas was dominated by fractional crystallization whereas the chemical evolution of younger, more evolved lavas was controlled by assimilation. Detailed trace element and isotopic work by Cousens (1996) supports this interpretation. Sr isotope ratios (0·70591–0·70635) increase with increasing SiO2 abundance, and the stratigraphically lowest samples in the Inyo-4 drill core have the least radiogenic Sr isotope signatures. ‘Classical’ AFC models suggest that country rock similar in composition to locally sampled Sierran granitoids was assimilated during fractional crystallization of mafic magma (Cousens, 1996). However, trace element and isotopic variations among the moat–Postpile lavas cannot be explained by a single ‘classical’ AFC model. On the basis of ‘classical’ AFC results presented by Cousens (1996), multiple trends reflecting relatively large variations in r (0·25–2) are required, and Cousens also suggested involvement of compositionally distinct parental magmas and chemically heterogeneous country rock.
One of the more interesting element–isotope trends for the moat–Postpile lavas is [Sr]–87Sr/86Sr (Fig. 13). The data form a three-part trend in which the most mafic samples cluster at the least radiogenic Sr isotope values (0·70595) but range from 750 to 875 ppm Sr, intermediate composition samples exhibit increasing [Sr] with increasing 87Sr/86Sr, and the most evolved samples show a relatively small range of 87Sr/86Sr at lower [Sr]. This non-monotonic trend cannot be modeled by a single ‘classical’ AFC trend. The entire range of compositions requires very different conditions, including different r values (e.g. 0·2–0·8) and distinctly different mineral–melt partition coefficients for Sr (e.g. 0·33–1·5; Fig. 13). Consistent with the conclusion of Cousens (1996), the observation that plagioclase is abundant in most of the samples suggests that Sr behaves compatibly during crystal fractionation; this indicates that the trend where DSr = 0·33 is not realistic. Using ‘classical’ AFC equations, it is therefore impossible to generate the positive [Sr]–87Sr/86Sr trend evident among some of the samples.
View larger version (26K):
[in this window]
[in a new window]
|
Fig. 13. Sr vs 87Sr/86Sr data (filled gray circles) and results of EC-AFC and ‘classical’ AFC simulations for mafic–intermediate volcanic rocks from the Long Valley caldera and Devil’s Postpile National Monument, eastern California (Vogel et al., 1994; Cousens, 1996). EC-AFC parameters in Table 2. Arrows illustrate direction of falling Tm, and each symbol represents a normalized temperature increment of 0·02 (30°C drop in Tm). ‘Classical’ AFC parameters: upper—Srm° = 800 ppm, Sra° = 250 ppm, 87Sr/86Srm° = 0·70596, 87Sr/86Sra° = 0·70660, DSr = 1·5, r = 0·2; lower—Srm° = 800 ppm, Sra° = 250 ppm, 87Sr/86Srm° = 0·70596, 87Sr/86Sra° = 0·70660, DSr = 0·33, r = 0·8. For ‘classical’ AFC trends, symbols represent a fraction of melt (F) increment of 0·1 but are terminated at F = 0·05.
|
|
Reasonable input parameters (Table 2) yield an EC-AFC trend that is strikingly similar to that formed by the moat–Postpile data; the model suggests that AFC occurred in the middle crust, where partial melts of the less silicic roots of Sierran plutons may have contaminated the magma body. Several specific aspects of the EC-AFC results are noteworthy. First, the suggestion of Vogel et al. (1994) that the oldest lavas seem to be related dominantly by fractional crystallization, an observation that is confirmed by the Sr isotope data, is supported by the EC-AFC model. In EC-AFC, this fractionation-dominated stage represents heating of country rock from Ta° to Ts. Because Sr appears to have behaved compatibly in the magma body during fractional crystallization (Cousens, 1996), the [Sr]–87Sr/86Sr trajectory is consistent with Sr behaving incompatibly during country rock partial melting. The modeled country rock Sr isotope ratio is similar to that reported by Cousens (1996) for local Sierran granitoids, although the modeled initial and liquidus temperatures of the assimilant (550°C and 1100°C, respectively) suggest that the country rock may be the deeper, less silicic (granodioritic to dioritic) roots of a Sierran-type pluton. Second, major element modeling (Vogel et al., 1994) suggests that some of the youngest samples from the Inyo-4 drill hole experienced significant assimilation. Despite this, these samples exhibit a relatively narrow range of 87Sr/86Sr (0·70621–0·70638). A rhyodacite sampled in the Devil’s Postpile area (Cousens, 1996) also has 87Sr/86Sr in this range (0·70636, Fig. 13). Such small variation in Sr isotopes among the most evolved samples, despite evidence for assimilation, is difficult to reconcile with ‘classical’ AFC models because [Sr]–87Sr/86Sr trajectories typically do not flatten out (Fig. 13). The flat trajectory at 87Sr/86Sr of 0·70632 potentially reflects the dilution effect that causes little change in 87Sr/86Sr to occur. Finally, because the magmatic system associated with the moat–Postpile samples is hypothesized to be a single, continuously evolving chamber and the relative chronology of eruption of the lavas is so well known (Vogel et al., 1994), this case study provides an excellent test of the potential application of EC-AFC. Consistent with observations, the model results predict that the first magmas to be erupted will be the least evolved and least contaminated whereas the last magmas to be erupted will be the most evolved and most contaminated. Because this prediction as well as the others discussed above are consistent with observed data, we suggest that this EC-AFC model more closely describes the conditions under which AFC occurred. We also stress that the model results make specific predictions about the system (e.g. the composition and thermal characteristics of country rock) that can be tested with further examination of geochemical and geological data.
Columbia River Basalts
Between 6 and 17 Ma, the Columbia River Basalts (CRB) erupted in what are now the states of Oregon, Washington, and Idaho, USA. Because of their total volume (173 000 km3), tholeiitic character, and rapid eruption rates (>91% of the volume erupted in <2 my; Hooper & Hawkesworth, 1993), the CRB are considered a small but typical example of a continental flood basalt (Hooper & Hawkesworth, 1993). On the basis of the recently revised stratigraphic nomenclature of Hooper & Hawkesworth (1992), the CRB are divided into the subgroups Clarkston and Saddle Mountains Basalts. The Clarkston Basalt, which comprises >97% of the volume of the CRB and was erupted between 14·5 and 17·2 Ma, includes the Imnaha, Grande Ronde, Eckler Mountain, and Wanapum Formations. The Saddle Mountain Basalt includes volumetrically insignificant (<1·5%) basalts that range in age from <14·5 to 6 Ma.
Most CRB are dominantly aphyric, but it is likely that plagioclase, olivine and pyroxene were liquidus phases, with plagioclase being the dominant constituent of the cumulate assemblage. The lavas are classified as tholeiitic basalt, although SiO2 and MgO range from 48 to 57 wt % and from 2·5 to 7·9 wt %, respectively. Sr (0·7034–0·7145) and Nd (0·5121–0·5130) isotope ratios vary considerably among all formations of the CRB, although the volumetrically significant Imnaha and Grande Ronde Formations exhibit more limited ranges (Sr isotopes: 0·7035–0·7058; Nd isotopes: 0·51261–0·51304; Carlson et al., 1981; Hooper & Hawkesworth, 1993). Of particular note is the observation that some CRB have Nd isotope ratios similar to that of present-day chondrite (DePaolo & Wasserburg, 1976a, 1976b), although these samples are among the more fractionated of the CRB (Carlson et al., 1981).
The observation that some CRB samples, as well as a limited number of samples from other continental flood basalts, have 
Nd = 0 has led to the suggestion that the source of continental flood basalts was primordial undifferentiated mantle (DePaolo & Wasserburg, 1976a, 1976b). Carlson et al. (1981) disputed this conclusion, citing the isotopic heterogeneity evident in the CRB, an apparent correlation between geographic location and isotope signature, and the observation that those samples with Nd = 0 are among the most differentiated. Carlson et al. (1981) suggested that involvement of continental crust was responsible for some of the isotopic variation observed in the CRB. Although controversy ensued about the petrogenesis of the CRB and other continental flood basalts, the weight of the evidence seems to favor involvement of a crustal component to explain some of the isotopic heterogeneity. By comparing ‘classical’ AFC modeling results with their data, Carlson et al. (1981) concluded that AFC explains some of the CRB data, and the most likely assimilant is sialic continental crust, perhaps of Precambrian age. Specifically, the Grande Ronde basalts, which represent the dominant volume of the CRB, are described by AFC, whereas the lack of ‘coupled trace element–isotopic variation’ (Carlson et al., 1981) precludes an origin by assimilation of evolved crust during fractional crystallization for the Imnaha basalts.
The Sr vs 87Sr/86Sr trend for the Grande Ronde and Imnaha basalts, which comprise 91 vol. % of the CRB, may shed light on the origin of some of the isotopic heterogeneity in these formations. Data from some Imnaha basalts (Carlson et al., 1981; Hooper & Hawkesworth, 1993) exhibit a trend of relatively constant 87Sr/86Sr over a range of [Sr] from 400 to 260 ppm. For basalts of both the Grande Ronde and Imnaha, more radiogenic 87Sr/86Sr (0·704–0·705) is coupled with higher [Sr] (250–375 ppm), although the most radiogenic samples (0·7055) do not have the highest [Sr] (320 ppm) (Fig. 14). Carlson et al. (1981) required multiple ‘classical’ AFC trends and particularly noted the inability of ‘classical’ AFC to explain the Imnaha samples. Reasonable EC-AFC parameters (Table 3) yield Sr–87Sr/86Sr trends that encompass most of the range displayed by the Imnaha and Grande Ronde basalts and suggest that the associated magma body underwent AFC in the middle to lower crust. The assimilant was probably intermediate to mafic crust. The EC-AFC results can explain the relatively flat element–isotope trend as well as the increasing [Sr] with increasing 87Sr/86Sr in the CRB data (Fig. 14). The positive Sr–87Sr/86Sr trend evident in these CRB samples cannot be explained by a ‘classical’ AFC model unless Sr is modeled to behave incompatibly during fractional crystallization (e.g. AFC2 trend, Fig. 14). Because evidence exists to the contrary (i.e. plagioclase as a dominant phase, Carlson et al., 1981), we emphasize that ‘classical’ AFC models are totally inadequate to describe these trends. In the EC-AFC model, Sr is modeled as compatible in the magma, consistent with the dominance of plagioclase as a liquidus phase (Carlson et al., 1981), whereas Sr is modeled as incompatible during melting of country rock. This may be consistent with lower-crustal temperatures (Ta° = 600°C) as well as an assimilant liquidus temperature (Tl,a = 1150°C) that is indicative of intermediate to mafic composition. In addition, the Sr isotope ratio required for EC-AFC assimilant is much less radiogenic than that modeled by Carlson et al. (1981); the lower Rb/Sr of less silicic rock would lead to less radiogenic time-integrated 87Sr/86Sr. No single EC-AFC model can explain all of the Grande Ronde and Imnaha data (Fig. 14, Table 3), suggesting that the country rock associated with CRB magma chambers is heterogeneous in both its elemental and isotopic character. Similarly, the isotope ratios of the least differentiated basalts are also heterogeneous, suggesting that either the parental magmas were compositionally variable or that the least differentiated samples experienced some open-system processing that resulted in heterogeneous Sr isotopic signatures. Although the total mass of assimilant involved is 44% of the total volume of original magma, simulations show that most of the contamination occurred with inclusion of only 20% of the total volume of original magma. Thus, for the more extreme compositions, relatively large amounts of assimilant are required, whereas the dominant proportion of the Grande Ronde and Imnaha basalts can be explained by assimilation of small masses of lower crust.
View larger version (30K):
[in this window]
[in a new window]
|
Fig. 14. Sr vs 87Sr/86Sr data (Grande Ronde basalts, grey circles; Imnaha basalts, ; Carlson et al., 1981; Hooper & Hawkesworth, 1993) and results of EC-AFC and ‘classical’ AFC simulations for Columbia River Basalts. EC-AFC parameters for CRB1, CRB2, and CRB3 in Table 3. Arrows illustrate direction of falling Tm, and each symbol represents a normalized temperature increment of 0·02 (30°C drop in Tm). ‘Classical’ AFC parameters are similar to those of Carlson et al. (1981): AFC1—Srm° = 250 ppm, Sra° = 50 ppm, 87Sr/86Srm° = 0·7035, 87Sr/86Sra° = 0·7600, DSr = 0·2, r = 0·25; AFC2—Srm° = 400 ppm, Sra° = 50 ppm, 87Sr/86Srm° =0·7035, 87Sr/86Sra° = 0·723, DSr = 1, r = 0·25. For ‘classical’ AFC trends, symbols represent a fraction of melt (F) increment of 0·1.
|
|
Early to Middle Tertiary intermediate to silicic composition volcanic rocks, eastern Nevada
Dominantly intermediate to silicic composition, early to middle Tertiary volcanic rocks exposed in eastern Nevada are part of widespread, calc-alkaline volcanism that was associated with large magnitude extension (Gans & Miller, 1983; Gans et al., 1989). On the basis of work in eastern Nevada and western Utah, Gans et al. (1989) and Feeley & Grunder (1991) identified three stratigraphic and lithologic groups of rocks: early, middle and late. The early group, which is the most compositionally diverse, varies from basaltic andesite to rhyolite. For basaltic andesite to rhyodacite early group rocks, Grunder (1992) identified three distinct subsets: (1) fine-grained basaltic andesites to andesites that have few or no plagioclase phenocrysts; (2) basaltic andesites to rhyodacites that have disequilibrium textures indicative of magma mixing; (3) andesites and dacites that have abundant plagioclase. Dacite to rhyolites of the middle group are hornblende–biotite-bearing lavas that show disequilibrium textures similar to those identified in the early group or textures that suggest they have achieved equilibrium. In general, compositional trends of middle group rocks are similar to the trends of early group rocks that exhibit evidence of magma mixing. Compositions of late group rocks are similar to those of middle group rocks, but are mainly biotite-bearing, crystal-poor, and exhibit equilibrium textures. In general, compared with early group rocks, middle and late group rocks are characterized by more radiogenic Sr isotopes, less radiogenic Nd isotopes, and exhibit less isotopic diversity.
To explain the isotopic and chemical diversity and the complex textures of the early group, Grunder (1992) proposed a two-stage AFC process. The fine-grained mafic rocks exhibit increasing [Sr] with increasing 87Sr/86Sr and increasing [Nd] with decreasing 143Nd/144Nd (Fig. 15a and b), characteristics that are consistent with AFC. ‘Classical’ AFC modeling suggests involvement of large masses of assimilant (r = 0·8) that is rich in Nd but poor in Sr (AFC1, Grunder, 1992). On the basis of the relatively low Sr concentrations in the assimilant, Grunder (1992)
TOP
ABSTRACT
INTRODUCTION
m); (b) Tm vs mass of anatectic melt added to the magma body (