2013 - Kohn - TOG-2nd-edition ZONING WITHIN MINERALS

4.
7 Geochemical Zoning in Metamorphic Minerals

MJ Kohn, Boise State University, Boise, ID, USA
ã 2014 Elsevier Ltd. All rights reserved.
This article is a revision of the previous edition article by M J Kohn, volume 3, pp. 229–261, © 2003, Elsevier Ltd.
4.7.1 Introduction 250

4.7.2 Major Elements 250
4.7.2.1 Rayleigh Distillation 250
4.7.2.2 Equilibrium Partitioning and Growth Zoning Models 251
4.7.2.2.1 P–T paths 253
4.7.2.3 Diffusion 254
4.7.2.3.1 Diffusional flattening of Mn growth profiles in garnet 255
4.7.2.3.2 Geospeedometry 256
4.7.2.4 Combination of Retrograde Diffusional Exchange and Reaction 257
4.7.2.5 Thermobarometric Implications 259
4.7.2.6 Kinetically Limited Transport within the Rock Matrix 260
4.7.2.7 Dissolution–Reprecipitation 261
4.7.3 Stable Isotopes 262
4.7.3.1 Growth Zoning 262
4.7.4 Trace Elements 264
4.7.4.4 Kinetically Limited Transport within the Rock Matrix 267
4.7.4.5 Oscillatory Zoning, Sector Zoning, and Textural Mimicry 268
4.7.5 Radiogenic Isotopes (Age Variability) 270
4.7.6 Case Study: Fall Mountain, New Hampshire 274
4.7.7 Discussion and Conclusions 275
Acknowledgment 277
References 277
Symbols f Fraction (of species i remaining in a rock)

4S2(x) Position-dependent term in closure temperature fO2, fH2O Fugacity of oxygen, water
calculation G Gibbs free energy
A Geometric constant in closure temperature H Enthalpy
calculations J Chemical flux rate
a Activity (in thermodynamic calculations) or KD Distribution coefficient
characteristic distance (in diffusion calculations) Keq Equilibrium constant
A, B, C Constants in stable isotope fractionation mi Number of moles of chemical component i in a
expressions rock
c Concentration Mk Number of moles of the mineral k in a rock
c* Normalized concentration nki, j Number of moles of chemical component i in the
C Radius phase component j in the mineral k
C* Normalized radius Nk, Nsys Number of moles of a chemical species in the
CRi, 0 Initial concentration of species i in rock phase k, system
CMi Concentration of species i in mineral M P Pressure
rC Concentration gradient R Gas constant
D Diffusivity or diffusion coefficient s Cooling rate
D* Self-diffusion coefficient S Entropy
D0 Preexponential constant t Time
E Activation energy T Temperature^_frp_firwid=80.1
Treatise on Geochemistry 2nd Edition http://dx.doi.org/10.1016/B978-0-08-095975-7.00307-7 249

Tc Closure temperature ai Fractionation factor

Tc(x) Position-dependent closure temperature ai–j Fractionation factor between i and j
V Volume (in thermodynamic calculations) or dij Kronecker delta (1 if i ¼ j; 0 if i 6¼ j)
velocity (in crystal growth calculations) Dij Difference between stable isotope d-values of i
W Weight fraction and j
x Distance or fractional distance « Exponential of Euler’s constant
X Mole fraction gi Activity coefficient of i
z Cation charge n Stoichiometric coefficient
4.7.1 Introduction in four sections: major elements, stable isotopes, trace ele-
ments, and radiogenic isotopes. Most theory has been devel-
Rocks encode the sum of Earth processes that affected them oped for major elements, which is why it is presented first, but
during their ‘lifetimes,’ and most geologic research seeks to the other types of zoning are becoming increasingly more stud-
invert that information to refine our understanding of those ied, and deserve separate, subequal sections.
processes. Metamorphic rocks record not just a single peak Analytical techniques are not explicitly described, and it is
pressure–temperature (P–T ) condition, cooling rate, assem- assumed that the reader is at least moderately familiar with
blage, or texture, but rather have undergone evolving histories electron microprobe, ion microprobe, and laser ablation
of changes in P and T, mineral abundances, rim compositions inductively-coupled plasma mass spectrometry (LA-ICP-MS),
and textures, acting over the metamorphic lifespan of a rock, in as well as X-ray mapping versus spot analysis for electron mi-
response to heat flow, stress and strain, and inter- and intra- croprobe studies. Most of these methods are described in the
granular movement of material. The greatest advances in Treatise Volume on Analytical Methods (McDonough, Ed.) to
understanding metamorphic rocks have recognized that meta- which the interested reader is referred. Mineral abbreviations
morphism is a continuum and have presented data and models are after Kretz (1983).
that directly address the continuum processes. Of the many
approaches for investigating and interpreting metamorphic
rocks, the characterization and quantitative modeling of geo- 4.7.2 Major Elements
chemical zoning in metamorphic minerals plays a central role. 4.7.2.1 Rayleigh Distillation
Geochemical zoning is uniquely useful because it quasi-
continuously records these metamorphic processes. Rayleigh distillation describes the partitioning of a chemical
Several excellent reviews have already described chemical species (element, molecular species, isotope, etc.) between two
zoning in metamorphic minerals (e.g., Chakraborty and reservoirs during a distillation process; that is, preferential
Ganguly, 1991; Loomis, 1983; Spear, 1993; Tracy, 1982). transfer of material from one reservoir to the other (Rayleigh,
This review differs in three ways: (1) It does not compile a 1896). If a geochemical species is especially compatible within
catalog of all the different minerals that exhibit zoning, and a mineral, then, as the mineral grows, the concentration of
what elements are zoned. Whereas some minerals may be that species in the matrix and at the rim of the mineral will
more obviously zoned than others, all minerals must be decrease. The rock matrix and the growing crystal are the
zoned in some element or isotope at some scale, and it is reservoirs from which and into which the species is being
simply a matter of time before that zoning is described. (2) It distilled. The rock matrix is assumed to be compositionally
emphasizes quantitative theoretical models of zoning, with homogeneous. If the mineral undergoes perfect fractional crys-
one or two of the best examples from nature. It also discusses tallization, so that it preserves a perfect compositional record
selected examples of zoning that are particularly relevant to of this process as it grows, then compositions will conform to
other endeavors, or hold special promise for future research. the equation (Hoefs, 1997)
(3) It presents a large body of information on trace element, R ai 1
i ¼ ai Ci, 0 f
CM [1]
stable isotope, and radiogenic isotope zoning; these latter re-
search areas have grown significantly since the original publi- where CM i is the concentration of species i in the mineral, ai is
cation of this review in 2003 (Kohn, 2003), and changes from the fractionation factor (partition coefficient) for species i in
Kohn (2003) mainly reflect this. the mineral relative to the rock, CRi, 0 is the initial concentration
Geochemists overwhelmingly favor garnet for zoning stud- of species i in the rock matrix, and f is the fraction of species i
ies, in part because it is indeed commonly zoned in major and remaining in the rock. For ai >1 (i.e., the species partitions into
trace elements, as well as stable and radiogenic isotopes, but the mineral), the instantaneous concentration of the species
also because its chemistry, geochemical partitioning character, in the mineral will exponentially decrease, asymptotically
and physical shape are readily modeled theoretically. Other approaching 0. A linear traverse across such a grain should be
minerals must also be zoned, but are not so uniformly amena- characteristically ‘bell-shaped.’ For ai <1 (i.e., the species par-
ble to study or modeling. Consequently, this review focuses on titions into the matrix), the instantaneous concentration will
garnet, but emphasizes theory and principles rather than simple increase, because mineral growth depletes the matrix in other
documentation, so the reader should be able to extrapolate to components, increasing the concentration of the less preferred
other minerals and chemical systems. The results are presented species.
Geochemical Zoning in Metamorphic Minerals 251
Hollister (1966, 1969) first proposed that Mn zoning in Hollister’s work was critical to later interpretations because
garnets could be understood in terms of Rayleigh distillation. Rayleigh distillation implies perfect preservation of composi-
He recast eqn [1] in terms of weight percent MnO in the garnet tions; that is, compositions were not modified during or after
(CGrt R
Mn ) and in the Mn-bearing minerals of the rock (CMn;0 ), and mineral growth by processes such as diffusion, dissolution–
the weight fraction of garnet (WGrt) relative to the other reprecipitation, recrystallization, etc. Therefore, ever since Holl-
Mn-bearing minerals in the rock (Wo): ister’s classic paper, ‘bell-shaped’ Mn profiles in garnets have
been cited as prima facie evidence for preservation of growth
WGrt aMn 1 zoning. Although bell-shaped profiles are expected for Rayleigh
CGrt
Mn ¼ a Mn C R
Mn;o 1 [2]
Wo distillation, they are not proof that a mineral faithfully retains its
He then showed that the Mn zoning profiles in garnets from original composition, because compositions may change during
Kwoiek, British Columbia, could be fitted with Rayleigh distilla- or after growth, yet still exhibit a general trend of decreasing Mn
tion models for aMn in the range 15–40 (Figure 1). Although Mn from core to rim. Furthermore, Rayleigh distillation sensu stricto
in garnet is the type example of Rayleigh distillation, other min- has a single value for a. Partitioning of elements depends to a first
erals that strongly partition an element can also show qualita- order on temperature (T ), and because changes in T will ther-
tively similar chemical depletions from cores to rims (Figure 2). modynamically drive mineral growth, T and a are likely to
change as the mineral grows. In fact, the best fit to the Kwoiek
profiles has a continuously decreasing value of aMn from core to
7.0 Analyzed point rim, which Hollister (1969) ascribed to an increase in T as the
garnets grew. That is, compositional patterns in minerals (e.g.,
Model with
Figures 1 and 2) result from a combination of contributing
6.0 a Mn = 23
processes, which is better termed ‘fractional crystallization’ rather
than ‘Rayleigh distillation.’
5.0
(wt%) MnO
4.0
4.7.2.2 Equilibrium Partitioning and Growth Zoning Models
Most generally, the matrix of a metamorphic rock is not a well-
3.0
stirred homogeneous material, but instead is better described
as multiple, chemically distinct minerals that simultaneously
2.0 change mode and composition during reaction. Furthermore,
if growth is driven by a change in P or T, then element parti-
tioning changes as the mineral grows. Modeling these effects
1.0 requires assessing the changes in abundance and composition
of each mineral in a rock, as well as any changes in P or T
0.0 attending mineral growth. This was first accomplished in a
400 300 200 100 0 100 200 300 400 fully quantitative fashion by Spear (1988), using a differential
Radius (μm) thermodynamic technique he dubbed the Gibbs’ method
Figure 1 Plot of wt% MnO versus radial distance in garnets from (Spear et al., 1982). In this approach, changes in mineral
Kwoiek, British Columbia (Hollister, 1966). MnO can be fitted with a composition and mode are determined from a starting set of
Rayleigh distillation model, with aMn 23. Reproduced from Spear FS compositions, modes, and P–T conditions, as a function of the
(1993) Metamorphic Phase Equilibria and Pressure–Temperature–Time thermodynamic properties of the minerals (entropy, volume,
Paths. Washington, DC: Mineralogical Society of America. and mixing parameters), compositions, and changes in P and T.
Th Ce Th Ce
8.0 8.0
27.0 27.0
100 mm 100 mm
25.0 25.0
2.0 2.0
Figure 2 X-ray maps of monazite from the Great Smoky Mountains, North Carolina, showing cores with high Th and slightly lower Ce, and rims with
low Th and slightly higher Ce, as expected from Rayleigh distillation of Th during monazite growth. Concentrations are in wt% of the oxide. Reproduced
from Kohn MJ and Malloy MA (2004) Formation of monazite via prograde metamorphic reactions among common silicates: Implications for age
determinations. Geochimica et Cosmochimica Acta 68: 101–113.
Abundant publications refined these models for interpreting phase component (j) of each mineral (k), respectively, Mk is the
geochemical zoning patterns (e.g., Kohn, 1993a; Menard and number of moles of the mineral k in the rock (i.e., proportional
Spear, 1996; Spear, 1988, 2010; Spear and Florence, 1992; to its mode), Xj,k is the mole fraction of the phase component
Spear and Markussen, 1997; Spear and Pyle, 2010; Spear j in the mineral k, and the summation is over all minerals in
and Selverstone, 1983; Spear et al., 1990b, 1995, 1999; Wang the rock. This equation simply reflects the transfer of material
and Spear, 1991; Young, 1993). An ‘integrated’ form of this as accomplished via changes to mineral compositions (first-
approach (Spear, 1993) was implemented by Powell et al. summation in dXs) and modes (second summation in dMs).
(1998) as an extension of their development of internally Combining equations of type [6] with the equilibrium and
consistent thermodynamic data for metamorphic minerals stoichiometric constraints yields a set of equations with a
(Evans et al., 2010; Holland and Powell, 1985, 1990, 1998; linear algebraic and thermodynamic variance of only 2, a result
Holland et al., 1998; Powell and Holland, 1985, 1988; Powell known as Duhem’s theorem (Spear, 1988). Duhem’s theorem
et al., 1998; White et al., 2000, 2001, 2007), and is central to implies that, after specifying the starting conditions (P, T, X,
the construction of ‘pseudosections,’ or bulk composition- and M), one can, for example, specify a DP and DT and predict
specific phase diagrams. changes in mineral composition and mode with thermody-
In thermodynamically based models, it is critical to account namic and mass balance rigor. This permits contouring of
for the P- and T-dependence of the partitioning of elements P–T space with mineral compositions and modes, to see graph-
among minerals, as well as the mass balance of reactions among ically how compositions and abundances change with chang-
minerals. A set of equations corresponds to each of these re- ing P–T conditions (Figure 3; e.g., Evans et al., 2010; Gaidies
quirements. The principles have been detailed elsewhere (Spear, et al., 2008; Kelsey and Powell, 2011; Kohn, 1993a; Spear,
1993), and the equations are simply presented here. 1993, 2010; Spear and Markussen, 1997; Spear and Pyle,
The equations that describe element partitioning are the 2010; Spear and Selverstone, 1983; Spear et al., 1982, 1990b).
differentials of the thermodynamic expression of equilibrium Strictly speaking, a zoning profile that is developed during
for each independent reaction: growth depends on the P–T path of the rock, because fractional
crystallization is path-dependent. Therefore, contour diagrams
dG ¼
0 or pseudosections are completely accurate only for equilibrium
@ ln Keq
¼ R ln Keq DS þ RT dT crystallization models in which minerals are compositionally
@T X @ ln Keq
homogeneous. Whereas such diagrams may not predict chem-
@ ln Keq
þ DV þ RT dP þ RT dXj [3] ical zoning precisely, they nonetheless reveal compositional
@P j
@Xj and modal trends that help interpret metamorphic mineral
growth and composition change. To a first order, Rayleigh
where DS and DV are the changes in molar entropy and volume
distillation would appear to be a dominant process controlling
for each reaction, R is the gas constant, dT, dP, and dXj are the
XSps and XPrp because neither shows maxima or minima over
changes in temperature, pressure, and mole fraction of the
the contoured region (Figure 3). However, XAlm shows obvious
phase component j, and
non-Rayleigh behavior, even though Fe is strongly preferred by
Y garnet. In fact, the trends in XAlm, XSps, and XPrp are described
Keq ¼ ð aj Þnj [4]
well in terms of Fe, Mg, and Mn phase equilibria. The end-
j
member reactions that form garnet from chlorite þ quartz
where aj and nj are the activity and the stoichiometric coeffi- occur at low, intermediate, and high Ts for the Mn, Fe, and
cient of the phase component j involved in the reaction. Each Mg end-members, respectively (Wang and Spear, 1991). Thus,
mineral is also subject to a stoichiometric constraint, because with increasing T, phase equilibria demand that Mn continu-
the sum of mole fractions must equal 1, which in differential ally decreases, Mg continually increases, and Fe shows a max-
form is imum in the vicinity of the Fe end-member reaction. These
X diagrams illustrate that Rayleigh distillation is not so useful for
dXj ¼ 0 [5] explaining Fe–Mg–Mn systematics, even though there are
j
major differences in partitioning of these elements between
The variance of the combined set of equations of type [3] and garnet and the matrix minerals. Instead, knowledge of reaction
[5] equals the thermodynamic degrees of freedom (as deter- locations in the Fe–Mg–Mn system better explains the overall
mined from the number of chemical components plus two (for trends. More realistic models of intracrystalline zoning com-
P and T ) minus the number of minerals). This set of equations bine thermodynamic equilibrium and mass balance with frac-
involves only intensive variables and is important for inferring tional crystallization (Spear, 1988, 1993), although predicted
P–T paths from chemical zoning. zoning profiles are necessarily path-dependent. An example
It is often convenient to assume a closed chemical system, involving an isobaric temperature increase (Figure 4) illus-
which imposes one mass balance constraint for each chemical trates how fractional crystallization profiles deviate from equi-
component. In differential form, each equation is librium models. Most importantly, Mn decreases faster than
! predicted by equilibrium models, causing Fe and Mg to in-
X X X X crease faster. The actual zoning patterns reflect both phase
dmi ¼ 0 ¼ Mk nki, j dXj, k þ nki, j Xj, k dMk [6]
equilibria and chemical distillation.
k j k j
Although no theoretical model can capture the idiosyncra-
where mi and nki;j
are the number of moles of each chemical sies of every rock, there are now hundreds of examples of
component i in the rock (e.g., SiO2, Al2O3, etc.) and in each garnets whose chemical zoning follows the basic patterns
10
ite
an e
Ky anit
lim
8 Sil
XPyrope XAlmandine
6
P (kbar)
0.219
0.179
0.259
0.139
0.299
0.099
0.68
0.73
0.78
0.059
0.68
0.58
0.039
0.48
4
2 Sill
im
And anite
alu
site
(a) 0 (b)
10
XSpessartine Moles garnet

6
P (kbar)
0.007
0.017
0.01
0.027
0.067
0.15
0.10
0.20
0.05
0.03
0.147
0.227
0
4
(c) 0 (d)
400 500 600 700 800 400 500 600 700 800
T (⬚C) T (⬚C)
Figure 3 Contours of garnet composition for the assemblage garnet þ biotite þ chlorite þ plagioclase þ quartz þ muscovite þ H2O as modeled in
the system MnO–Na2O–CaO–K2O–FeO–MgO–Al2O3–SiO2–H2O. Black dot shows reference P–T conditions. Horizontal arrow corresponds to isobaric
P–T path used for modeling garnet compositions in Figure 4. Note systematic increase in XPrp and decrease in XSps with increasing T, but maximum
exhibited by XAlm at 550 C. Reproduced from Spear FS (1993) Metamorphic Phase Equilibria and Pressure–Temperature–Time Paths.
Washington, DC: Mineralogical Society of America.
0.5 1.0 depicted in Figure 4 of decreasing Mn, increasing Mg, and

Fractional
crystal. decreasing Fe/(Fe þ Mg) from core to rim. This implies that
Fe/(Fe+Mg) Equilibrium thermodynamic equilibrium and fractional crystallization are
0.4 crystal. 0.9
plausible drivers of chemical zoning, and that both thermody-
XAlm, Fe/(Fe+Mg)
XPrp, XSps, XGrs
namics and mass balance must be accounted for to explain

0.3 XAlm 0.8 compositional zoning patterns quantitatively. The occurrence
XSps of profiles that have a Rayleigh-like appearance in one or more
0.2 0.7 elements (Figure 1) demonstrates that fractional crystallization
XGrs XPrp has occurred, but not that compositions are perfectly preserved,
0.1 0.6 or that simple Rayleigh distillation models are accurate.
4.7.2.2.1 P–T paths

0 0.5
Quantitative modeling of chemical zoning in metamorphic
Core Rim
garnets allows retrieval of metamorphic P–T paths. If the chem-
Figure 4 Plot of garnet composition versus radius for garnet ical zoning encoded in a garnet is a reflection of the changes in
growth between 500 and 577 C at 6 kbar. Assemblage and P or T that drove garnet growth, then in principle one can
chemical system are the same as in Figure 3. Dotted line shows invert the zoning profile to infer what those P and T changes
trace of equilibrium compositions, which result from phase were. That is, if DP and DT cause DXs, one should be able to use
equilibrium effects. Solid line shows fractional crystallization model.
the DXs that are preserved as chemical zoning to infer the
Similarity of curves for all components (not just Mn) indicates that
original DP and DT. This approach is computationally simple
phase equilibria strongly control garnet chemical compositional
changes, but significant differences between curves, especially for Mn, if mass balance constraints are imposed. Duhem’s theorem
indicate that fractional crystallization is also important. Reproduced requires a variance of only two, so zoning in only two compo-
from Spear FS (1993) Metamorphic Phase Equilibria and nents of the garnet (two DXs) is necessary to retrieve DT
Pressure–Temperature–Time Paths. Washington, DC: Mineralogical and DP. However, it is difficult to verify that the mass of a
Society of America. chemical component has remained constant. For example,
the whole-rock Na2O and CaO contents depend on bulk pla- Subsequent updates to mixing models and thermodynamic data
gioclase content and composition. However, if plagioclase (Berman, 1988; Holland and Powell, 1998; Pattison et al.,
cores are nonreactive, the effective bulk composition that the 2002), especially for water, suggest a larger change in DT of
garnet ‘sees’ at it grows may be substantially depleted in Na2O 75 C, although differences among calculations using the re-
and/or CaO compared to the whole rock. Plagioclase non- cent thermodynamic databases and models are now impercep-
reactivity affects Ca mass balance, which influences XGrs trends tible. Most importantly, none of these modifications affects
in garnet (Spear et al., 1990b). This problem may be evident in Selverstone et al.’s original conclusion that the garnet grew
plagioclase compositional zoning, but is difficult to model during exhumation with heating (Figure 5(b)).
quantitatively. Consequently, most workers use only the equa- The recovery of the Tauern Window P–T path (Selverstone
tions that involve intensive variables (dP, dT, and dXs). The et al., 1984) was revolutionary because the geodynamics com-
linear algebraic variance of the set of equations is then equiv- munity had predicted that exhumation with heating should be
alent to the thermodynamic variance, which ordinarily means the natural result of overthickening, relaxation of isotherms, and
that changes to several mole fractions must be measured to exhumation in collisional orogens (England and Richardson,
resolve dP and dT. In the case of typical garnet-grade metape- 1977). Indeed, England and Thompson (1984) published a
lites, the thermodynamic variance is four, so four independent now-classic thermal modeling study the same year as Selverstone
dXs must be measured. Only three of the major mole fractions et al. (1984), investigating numerically the P–T implications
in garnet are independent, so changes to the mole fraction of of overthickening during orogenesis. The convergent series of
another mineral component would need to be measured to publications in petrology and geodynamics (England and
model a P–T path. The calculated path can be very sensitive to Thompson, 1984; Selverstone et al., 1984; Spear and Selver-
the choice of the fourth dX (Kohn, 1993b). For example, dXAnn stone, 1983; Spear et al., 1984; Thompson and England, 1984)
and dXPhl in biotite are poor choices numerically because they inseparably linked these two fields, so that, today, most meta-
do not resolve changes in pressure, in contrast to dXAn in morphic petrologists justify P–T path research on its ability to
plagioclase, which does. inform geodynamic and tectonic processes.
The first P–T path determined by inversion of a zoning
profile was from the Tauern Window, Austria (Figure 5;
Selverstone et al., 1984). The rock has an unusual, low-
4.7.2.3 Diffusion
variance assemblage, so only two independent dXs were
needed to retrieve a path. Two logical pairings are XAlm–XGrs Diffusion is a thermally activated process, whereby a chemical
and XSps–XGrs, because XSps and XAlm are both sensitive to tem- or isotopic species moves down a chemical potential gradient
perature (e.g., Figure 3), whereas XGrs is moderately sensitive to (usually from high to low concentration), at a rate that is
pressure. The path calculations based on different components dependent on the diffusion coefficient D. This applies to the
show very similar P–T evolution – exhumation by 3 kbar progressive smoothing of growth-zoning in a mineral as a
(1 bar ¼ 105 Pa) with heating of 20 C (Figure 5(b); open rock heats and to diffusive fluxes into or out of the surface
symbols) – suggesting realization of the model assumptions; of a mineral as processes in the matrix alter the rim composi-
that is, chemical equilibrium was attained among minerals, tion relative to the mineral interior. Several recent reviews
and garnet faithfully retained a record of changing composition. of diffusive processes in Earth materials are provided by
12
Profile I Profile II Profile I
0.70 XAlm and XGrs
10
Profile I
XAlm
0.65 XAlm
XSps and XGrs
0.60 8 Profile II
XGrs XAlm and XGrs
0.15
P (kbar)
XGrs
0.10
6 Profile II, recalculated
0.10 XSps XAlm and XGrs
XSps
ite
Kyan ite
0.05 an
4 Sillim
0.15 Sillima
XPrp
nite
Kyanite Andalu
ite site
Andalus
XPrp 0.10 2
Rim Core Rim
0 1 2 3 4 5 0
400 450 500 550 600
(a) Distance (mm)
(b) T (⬚C)
Figure 5 (a) Compositional zoning profile collected across a single garnet from sample FH-1M, Tauern Window, Austria. (b) P–T path calculations
for sample FH-1M, from the Tauern Window, Austria. The mineral assemblage in this rock has an unusually low variance (2) which permits a P–T path to
be inferred solely from garnet compositional changes. Paths labeled I and II (open symbols) correspond to different halves of the zoning profile.
Recalculation of the original path by using improved thermodynamic data corroborates DP, but suggests a larger DT (closed symbols). Modified from
Selverstone J, Spear FS, Franz G, and Morteani G (1984) High pressure metamorphism in the SW Tauern window, Austria: P–T paths from
hornblende–kyanite–staurolite schists. Journal of Petrology 25: 501–531; Spear FS (1993) Metamorphic Phase Equilibria and
Pressure–Temperature–Time Paths. Washington, DC: Mineralogical Society of America. Aluminosilicate diagram from Holdaway MJ (1971) Stability of
andalusite and the aluminum silicate phase diagram. American Journal of Science 271: 97–131.
Chakraborty (2008), Ganguly (2002, 2010), Watson and where i and j are the interdiffusing species, Xi and Xj are the
Baxter (2007), and Zhang and Cherniak (2010). mole fractions of components i and j, respectively, and gi is the
The diffusion coefficient is formally defined as the propor- activity coefficient of component i. Thus, most generally, cat-
tionality constant between flux rate (J) and concentration ion diffusion calculations must account for the T, P, fO2, and X
gradient (rC): dependencies of the diffusion coefficient, including the mixing
properties of the mineral. Few if any applications actually
J ¼ DrC [7]
account explicitly for all these effects, and, instead, simplifica-
and is an exponential function of temperature: tions are made, for example, assuming that solutions are ideal,
that fO2 is buffered, or (occasionally) that D*s are approxi-
D ¼ D0 eE=RT [8] mately equal (which eliminates cross-terms in eqn [9]). In-
deed, Chakraborty and Ganguly (1991, 1992) suggested
where D0 is a preexponential constant, E is the activation
using an effective binary diffusion coefficient with a single D0
energy, R is the gas constant, and T is absolute temperature.
and E that describes interdiffusion of two species over the
For cations, E generally depends weakly on pressure and fO2
composition range in the sample studied. In part, these sim-
(for minerals that contain Fe). Commonly, the P-dependence
plifications result because even the best diffusion experiments,
is expressed as an activation volume contribution to E, and
when extrapolated to metamorphic conditions, yield uncer-
results are adjusted to a particular fO2 buffer, e.g., the graphite–O2
tainties in D of 1–2 orders of magnitude, so errors arising
buffer (Chakraborty and Ganguly, 1991, 1992; Chakraborty
from the assumptions are relatively small. Furthermore, many
and Rubie, 1996; Ganguly et al., 1998a, 2010; Loomis, 1978a,
diffusion calculations are more realistically limited by uncer-
1978b; Loomis et al., 1985). These corrections are especially
tainties in how initial concentrations or changes to rim
important for reconciling experimental data that are collected
compositions are assigned, not by model simplifications.
at different P and fO2.
Compilations of experimental and natural data provide stan-
The diffusion coefficient can also depend on composition,
dard self-diffusion expressions for Mg2þ, Fe2þ, Mn2þ, and Ca2þ
giving rise to three different types of diffusion coefficients (e.g.,
in garnet (Carlson, 2006; Ganguly et al., 1998a).
see Chakraborty and Ganguly, 1991; Zhang, 2010): (1) Tracer
diffusion describes movement of an element at infinite dilu-
tion (e.g., trace Fe in ‘pure’ forsterite); (2) Self-diffusion de-
4.7.2.3.1 Diffusional flattening of Mn growth profiles
scribes movement of an element in the absence of a chemical
in garnet
potential gradient (e.g., Fe in pure fayalite), as determined via
If an increase in temperature drives garnet growth, then frac-
measurement of isotope diffusion rates; (3) Interdiffusion, or
tional crystallization produces a core-to-rim Mn decrease.
chemical diffusion, describes movement of one element in
However, increasing temperature also increases diffusion
exchange for another (e.g., Fe–Mg interdiffusion in a forsterite–
rates, and the overall Mn pattern and composition of the rim
fayalite solid solution). Tracer and self-diffusion do not cause
region will change as Mn diffusively equilibrates. Basically,
appreciable chemical compositional changes. Consequently,
fractional crystallization competes with the core-to-rim diffu-
geologic applications to major and minor element zoning
sive flux to respectively decrease and increase Mn at the rim.
ubiquitously involve interdiffusion. Generally, however, no
At sufficiently high temperature, garnet compositions are ho-
single interdiffusion coefficient exists in a multicomponent
mogenized, which can be plotted in different portions of a field
mineral. Rather, for an n-component mineral, the concentra-
area and, in principle, related to the temperature–time history.
tion gradient (r C) and chemical flux rate (J) are (n 1) vectors
Several studies have documented progressively flatter Mn pro-
related via an (n 1) by (n 1) diffusion coefficient matrix
files in garnets at progressively higher metamorphic grades
(Lasaga, 1979). Each term in r C and J refers to a chemically
(e.g., Anderson and Olimpio, 1977; Carlson and Schwarze,
independent component, and the reduced dimensionality is
1997; Dempster, 1985; Woodsworth, 1977; Yardley, 1977).
because the flux rate of one component can be expressed in
Investigation of polymetamorphic rocks of the Llano Uplift
terms of the other (n 1) components. Lasaga (1979) de-
(Carlson, 2002; Carlson and Schwarze, 1997) illustrates both
scribed how to derive the diffusion coefficient matrix from
diffusional flattening for high-Mn cores and near-rim diffusion
self-diffusion coefficients, if the thermodynamic mixing prop-
profiles imposed via thermal overprinting and mineral resorp-
erties of the different components are known. Assuming ideal
tion. For garnet crystals with radii of 0.5–1 mm, the degree of
solid solutions, each component in the matrix (Dij) is
2 3 Mn-flattening depends on inferred peak metamorphic condi-
* h i tions and grain size (Figure 6). Samples that were collected
6 D C i i j 7
z z
Dij ¼ D*i dij 4P i * 5 D*
j D *
n [9] directly adjacent to plutons have completely flattened core
Dk Ck z2k profiles, and garnets of different sizes in a single rock exhibit
k
flatter Mn profiles in smaller grains (Figure 6). Both results are
where zi is the charge on the cation i, dij ¼ 1 if i ¼ j and dij ¼ 0 if explained by growth of garnets with Mn-enriched cores that
i 6¼ j, and the D*s are the self-diffusion coefficients. More com- were subsequently flattened as temperature increased, either
plex equations apply for nonideal solid solutions. Some geo- during the first metamorphic event for samples far from plu-
logic problems can be reduced essentially to binary exchange, tons, or during a contact metamorphic overprint for samples at
e.g., for Fe–Mg interdiffusion in silicates. For a binary nonideal pluton margins. The increases in Mn at the rims of many of the
solution, the interdiffusion coefficient expression is garnets result from post-peak metamorphic garnet resorption
" #" # and consequent repartitioning of Mn among matrix minerals
D*i D*j @ ln gi
Dij ¼ 1þ [10] and the garnet rim. Rim zoning impacts many applications,
Xi D*i þ Xj D*j @ ln Xi P, T
including ‘geospeedometry’ (the use of diffusion profiles for
5.0 12.0 12.0

10.0 700 ⬚C 700 ⬚C
585 ⬚C 700 ⬚C
MnO (wt%)
MnO (wt%)
3.0 8.0
10.0 10.0
1.0 6.0
8.0 8.0
–1.0 0.0 1.0 –1.0 0.0 1.0 –0.65 0.0 0.65 –0.85 0 0.85
Distance (mm from garnet center) Distance (mm from garnet center)
5.0 5.0 12.0

700 ⬚C
750 ⬚C Next to pluton
MnO (wt%)
MnO (wt%)
3.0 3.0
10.0
1.0 1.0
8.0
–1.0 0.0 1.0 –1.0 0.0 1.0 –1.45 0.0 1.45
(a) Distance (mm from garnet center) (b) Distance (mm from garnet center)
Figure 6 (a) Plots of garnet MnO content (wt%) versus distance across central sections of crystals from different localities in the Llano Uplift
polymetamorphic block, Texas. Garnets show increased core–rim compositional profile flattening with increasing temperature (either peak conditions
for prograde event or proximity to plutons during static overprinting event). (b) Garnet MnO content versus distance across central sections of
crystals of different sizes from a single locality. Garnets show increased core–rim compositional profile flattening with decreasing grain size. All data are
consistent with diffusive modification of an earlier formed growth profile. Modified from Carlson W and Schwarze E (1997) Petrological significance
of prograde homogenization of growth zoning in garnet: An example from the Llano Uplift. Journal of Metamorphic Geology 15: 631–644; Carlson WD
(2002) Scales of disequilibrium and rates of equilibration during metamorphism. American Mineralogist 87: 185–204.
estimating cooling rates), and thermobarometry (the determi- where a is the characteristic distance of the diffusing grain (e.g.,
nation of metamorphic P–T conditions). the grain radius). These substitutions transform eqn [11] to:
@c @2c
0 ¼ [16]
4.7.2.3.2 Geospeedometry @tr @x0r 2
Lasaga (1983) outlined a technique that he dubbed In principle, one must solve eqn [16] for all minerals in the
‘geospeedometry’ for inverting chemical diffusion profiles on rock that are diffusionally limited from equilibrating. How-
the rims of minerals to infer cooling rates. The basic diffusion ever, many applications involve Fe–Mg exchange between gar-
equation is net and biotite. Biotite’s nearly ubiquitous Fe/Mg homogeneity
@c @2c implies that it diffuses sufficiently quickly that one must solve
¼ D ð tÞ 2 [11] for only the diffusion profile in the garnet.
@t @x
Equation [16] can be solved for g, and hence s, subject to
where c is the concentration of the species of interest (e.g., Mg
the constraint of equilibrium partitioning on the rim of the
in garnet), D is the diffusion coefficient, t is time, and x is the
garnet due to exchange with another mineral:
distance. The difficulty in solving this equation is that D is a

function of T (eqn [8]), which changes with time. However, DH s
KD ð tÞ ¼ KD, init exp 2 t [17]
eqn [11] can be made analytically tractable, if linear cooling is RTinit
assumed, through a series of transformations (Lasaga, 1983;
where KD is the distribution coefficient for the exchange (e.g.,
see also Lindström et al., 1991). First, Lasaga (1983) intro-
for Fe/Mg between garnet and biotite), and DH is the enthalpy
duced a compressed time variable (t0 ):
of reaction. Lasaga’s approach has been improved subse-
0 1
t ¼ ð 1 egt Þ [12] quently both theoretically and computationally (Ganguly and
g Tirone, 1999; Ganguly et al., 2000; Jaoul and Sautter, 1999;
where t is time, and g is given by Lasaga and Jiang, 1995; Lindström et al., 1991).
Two fundamental interpretational cautions are, however,
Es
g¼ 2 [13] warranted. First, the diffusion profile really results from an
RTinit integrated diffusional history (e.g., Ganguly et al., 2000)
In the above, E is the activation energy, s is the cooling rate
D ð T0 Þ
(‘speed’ of cooling), R is the gas constant, and Tinit is the initial g ¼ Rt [18]
temperature. He then nondimensionalized time (t0 ) and dis- 0 D ð tÞdt
tance (xr) as
which corresponds to an infinite number of cooling paths.
0 Dinit 0 Linear cooling (i.e., eqn [13]) is only one solution of eqn
tr ¼ 2 t [14]
a [18]. Thus, one must interpret the results within the context
0 x of the most likely thermal history, for example, linear cooling
xr ¼ [15]
a versus steadily increasing or decreasing rates. Second, Lasaga’s
0.3 0.3 0.8

G1 Mg
XSps, XGrs, XPrp

Mg/(Fe+Mg)
0.2 0.2 0.7
G2
XAlm
Fe
0.1 0.1 0.6
Ca
Mn
0.0 0.0 0.5
0 40 80 120 0 100 200
(a) Distance (μm) (b) Distance (μm)
Figure 7 Composition versus distance profiles for high-T garnets from Ganguly et al. (2000). (a) Mg/(Mg þ Fe) ratio in garnet versus distance from the
rim in contact with biotite, showing smoothly decreasing values towards rim. Squares and circles are for profiles measured in two different garnets
(here denoted G1 and G2), and the difference is interpreted to result because edges of garnets are not necessarily perpendicular to the plane of the
section. The two profiles can be reconciled if profile from G2 is rotated 46 relative to profile from G1. Profiles can be fitted with a diffusion profile with
cooling rates of 20 C per Ma, assuming diffusion coefficients (Ds) from Ganguly et al. (1998a). Note that different Ds and uncertainties in D can
yield widely different cooling rates. (b) All garnet components are zoned in the outer rim of garnets, suggesting diffusion is not the only process affecting
rim garnet compositions.
transformation of variables essentially converts the problem order of magnitude at T 800 C, this directly translates into
into a description of the position-dependent closure tempera- an error in the retrieved cooling rate. That is, if D is not known
ture Tc(x). If the matrix remains compositionally homoge- to a factor of 10, neither is s (Lindström et al., 1991).
neous (e.g., very small and very large amounts of garnet and
biotite, respectively), then a general analytical solution exists
(Dodson, 1986; see also McDougall and Harrison, 1999):
4.7.2.4 Combination of Retrograde Diffusional Exchange
E=R and Reaction
Tc ð xÞ ¼ [19]
eRTc2 D0 =a2
ln þ 4S2 ðxÞ Interpretation of near-rim composition profiles depends criti-
Es
cally on the nature of retrograde reactions (Kohn and Spear,
where e is the exponential of Euler’s constant, 4S2(x) is a 2000; Robinson, 1991; Spear, 1993; Spear and Florence, 1992;
position-dependent term, x is the fractional distance from the Figure 8), which fall into two different classes. (1) Exchange
mineral center, and other terms are as described above. Closure reactions (ERs) involve the exchange of two elements between
temperatures (and hence compositions) depend weakly on two minerals, for example, Fe–Mg exchange between garnet and
cooling rate because s appears functionally in a logarithm. biotite (almandine þ phlogopite ¼ anniteþ pyrope). Retrograde
The insensitivity of the composition profile to s leads to large exchange reactions do not significantly change mineral modes,
uncertainties in retrieved s. This problem is further com- and cause divergence of mineral compositions. (2) Net-transfer
pounded by the large uncertainties in D because of extrapola- reactions (NTRs) involve production and consumption of min-
tion of experimental data to metamorphic conditions. For erals; that is, there is a net transfer of material among mineral
many problems, the bounds on the cooling rate may differ by reservoirs. Unlike ERs, NTRs cause mineral compositions to shift
orders of magnitude (Lindström et al., 1991), and it may prove in the same direction. Thus, a decrease in Mg and increase in Fe
difficult to refine estimates. towards the rim of a garnet (e.g., Figures 6 and 7) could result
Ganguly et al. (2000) provide an example of geospeedo- from retrograde Fe–Mg exchange between garnet and biotite (an
metric calculations for rocks of the Sikkim Himalayas (NE ReER), or retrograde reaction among several minerals, such that
India). They especially note that composition profiles should they all shift towards more Fe-rich compositions (an ReNTR;
be collected perpendicular to the grain edge, whereas edges Figure 8). The physical difference between them is that an ReER
may ‘dip’ with respect to the plane of the section, thus requir- does not change the position of mineral rims perceptibly,
ing correction of the diffusion profile for the crystal boundary whereas an ReNTR causes mineral resorption or growth. There-
orientation (Figure 7). However, for correctly oriented grains fore, it is critical for any geospeedometry study either to verify
one may solve for g 2.1 1014 in these rocks. One could petrologically that ReNTRs have not occurred, or to characterize
then use eqn [13] to infer s 20 C per Ma, but a uniform the amount by which the boundary of the mineral has moved.
cooling rate may not be applicable over the entire temperature Mn profiles provide one means of identifying operation of
range during which Fe–Mg exchange is evident. Instead, ReNTRs. Because garnet partitions Mn so strongly, any growth of
Ganguly et al. (2000) used the product a2g/D(T0) as a con- garnet will cause Mn to decrease towards the rim, whereas any
straint in a one-dimensional thermal model linking cooling resorption will cause it to increase (Figure 8). Therefore, Mn
with exhumation. From this, they inferred accelerated cooling increases on rims are prima facie evidence for dissolution of
from an initial rate of 15 C per Ma at 800 C, to >100 C per garnet via ReNTRs, whereas a flat or decreasing Mn concentra-
Ma by 450 C. The retrieved cooling rate depends directly on tion at the rim indicates only ReERs. This is evident in diffusion
the diffusion coefficient, and because experimental uncer- profiles in garnet around biotite inclusions, where Fe and Mg are
tainties in D for divalent cations in garnet are well over an zoned, but Mn is not (e.g., see Spear and Parrish, 1996).
High-T retrograde exchange (ReER) Rim

t• B1
Al2O3
t3
G2 t2
G1 B2
Fe/(Fe+Mg)
t1
t0 G2
Garnet
Biotite t0
G2 G1
G1 t1
t
t3 2
B1 B2 B2 t•
FeO MgO
(a) Distance
Rim
High-T retrograde net transfer (ReNTR)
Al2O3
B3
t•
t3
B3
G3 t2
G1 t1 G2
Fe/(Fe+Mg)
t0
G1 Garnet
G3 B3 t• t3 G1
t2
B3 B1 t1
Biotite t0
FeO MgO
(b) Distance
Low-T retrograde exchange (ReER) vs. net transfer (ReNTR)
ReNTR biotite Garnet ReER biotite

Trough
Fe Fe
(Fe + Mg)
(Fe + Mg)
Final Bt (Fe + Mg) (Fe + Mg)
No
Fe
Fe
Orig. Bt Troughs Orig. Bt
Trough
Final Bt
XSps XSps
ReNTR Core ReER

(c) Rim Rim
Figure 8 Diagrams illustrating the change in Fe/(Fe þ Mg) for garnet (G) and biotite (B) during retrograde reactions (Kohn and Spear, 2000; Spear,
1993). (a, b) High-grade, diffusionally homogenized garnets. G1 and B1 are peak compositions; G2 and B2 are retrograde compositions. (a) Effects of
retrograde exchange reactions (ReERs). On an Al2O3–FeO–MgO (AFM) diagram, garnet and biotite compositions diverge with decreasing temperature.
The compositions corresponding most closely to peak conditions are the garnet with highest Mg/Fe and the biotite with lowest Mg/Fe. Because of slow
DFe–Mg in garnet, but fast DFe–Mg in biotite, biotite remains homogeneous while a diffusion profile is established at the garnet rim. Pairing of garnet
core (G1) with matrix biotite (B2) yields a temperature below the peak. (b) Effects of retrograde net transfer reactions (ReNTRs). On an AFM diagram,
garnet rim and biotite compositions move towards higher Fe/(Fe þ Mg) during retrograde garnet dissolution. The compositions corresponding most
closely to peak conditions are the garnet and biotite with highest Mg/Fe. However, the original biotite may not be present, and pairing of matrix biotite
(B3) with garnet core (G1) yields too high a temperature. (c) Diagrams for lower-grade conditions illustrating the change to Fe/(Fe þ Mg) and Mn in
garnet, and Fe/(Fe þ Mg) in biotite during retrograde reactions. Dashed line shows original profile at peak conditions. General decrease in XSps towards
rim is result of prograde growth. ReERs (right side) cause an increase in garnet Fe/(Fe þ Mg) at the rim, which forms a compositional trough near
the grain edge (right side). No change in Mn occurs. Pairing of garnet trough composition with matrix biotite yields a temperature below the peak
because garnet and biotite have shifted to higher and lower Fe/(Fe þ Mg) respectively. ReNTRs (left side) cause an increase in both Fe/(Fe þ Mg) and
Mn in garnet, which forms compositional troughs near the grain edge in both profiles (left side). Pairing of garnet trough composition with matrix
biotite may yield a temperature that is too high because both garnet and biotite shift to higher Fe/(Fe þ Mg).
Carlson (2002) ascribed composition profiles in garnet application of geospeedometry to the Fe–Mg profiles would
rims from the Llano Uplift, Texas, to both ReERs and ReNTRs. yield an apparent cooling rate that is too fast. An estimate of
These rocks are unusual because spectacular retrograde coronas the amount of garnet resorption from the size of the coronas
permit identification of the original rim locations (Figure 9). permits a rigorous accounting of mass fluxes (particularly Mn),
Composition profiles show increases in Fe and decreases in Mg and fitting of the composition profiles. By accounting quanti-
towards garnet rims, as expected for diffusional reequilibration tatively for both ReERs and ReNTRs, Carlson (2002, 2006)
of the garnet rim during cooling (Figure 10). However, the refined the relative diffusion rates for Ca, Fe, Mg, and Mn,
coronal textures and the increase in Mn towards the garnet rims which, in turn, permitted reconciliation of the observed pro-
(Figure 10) both unequivocally indicate operation of one or files with the independently constrained cooling rate.
more garnet-consuming ReNTRs, so that the rim of the garnet Most samples lack coronas, and resorption may only
has not remained stationary. If this fact were not recognized, be evident either from rounding or embayed margins
(i.e., texturally) or from near-rim Mn increases (i.e., geochem- 4.7.2.5 Thermobarometric Implications
ically). Without direct markers of original rim locations, the
Thermobarometry, or the calculation of P–T conditions from
extent of reaction is difficult to characterize quantitatively.
mineral compositions, is not based on mineral zoning, nor
Therefore, Carlson’s work cautions against application of geos-
does it directly inform processes that cause geochemical zon-
peedometry except in special cases where the garnet rim has not
ing. Nonetheless, P–T calculations in metamorphic rocks usu-
moved, or the amount of garnet resorption is independently
ally require selecting compositions from minerals that are
determinable. In the case of the garnets from Sikkim analyzed
compositionally heterogeneous. How one selects composi-
by Ganguly et al. (2000), the small increases in Mn towards
tions from a zoned mineral can strongly influence retrieved
garnet rims (Figure 7(b)) imply that garnet has dissolved some-
P and T, and that issue warrants discussion.
where in the rock (albeit possibly not at those particular rims).
Normally one calculates P–T conditions based on the
If dissolution has shortened the near-rim profiles, then calcu-
T-dependence of a cation partitioning equilibrium (e.g., Fe–Mg
lated cooling rates assuming no resorption will be too fast.
exchange between garnet and biotite) and the P-dependence of
a net-transfer equilibrium (e.g., anorthite ¼ grossularþ alumino-
Re silicate þ quartz). Most barometers have moderate slopes, so the
calculated P depends moderately on T. The basic problem lies in
lic
choosing the best compositions from which one calculates T

tr
im
(Robinson, 1991; Tracy et al., 1976; see also Kohn and Spear,
Garnet 2000; Spear, 1991, 1993, ch. 17; Spear and Florence, 1992;
Figure 8). A higher Mg/Fe ratio in garnet or Fe/Mg ratio in
biotite will result in a higher computed T. If only ReERs have
occurred, then compositions have diverged and all calculated Ts
Corona will be below the peak. Therefore, the highest T will be the best
estimate of peak conditions and, of all compositions expressed
in a rock, will correspond to the highest Mg/Fe in garnet and
Figure 9 Photomicrograph of spectacularly well-developed corona highest Fe/Mg in biotite. By contrast, if ReNTRs have occurred,
around relict garnet from Llano Uplift, Texas. The coronas permit
then compositions of each mineral have shifted simultaneously.
identification of the original rim position of the garnet (assuming
isovolumetric replacement), which in turn allows quantification of the
Consumption of (Fe-rich) garnet causes biotite and the garnet
amount of garnet that was resorbed during cooling. Reproduced from rim to shift to more Fe-rich compositions. If the Fe-enriched
Carlson W and Schwarze E (1997) Petrological significance of prograde matrix biotite is then paired with relict Mg-rich garnet, temper-
homogenization of growth zoning in garnet: An example from the Llano atures exceeding peak conditions can be calculated (Kohn and
Uplift. Journal of Metamorphic Geology 15: 631–644. Spear, 2000; Spear, 1991; Spear and Parrish, 1996). Commonly,
0.0010 0.0010
Mn Mn
0.0005
0.0000 0.0000
0.0130
Fe Fe
0.0130
0.0125
Moles cm–3
0.0120 0.0120
0.0115 0.0110
Mg 0.0090
0.0080
0.0075 0.0080
Mg
0.0070
0.0070
0.0065
0.05 0.15 0.25 0.35 0.00 0.02 0.04 0.06

Distance from core (mm) Distance from core (mm)
Figure 10 Zoning profiles measured at the outer edges of garnets from rocks with much less pronounced coronas show diffusional patterns over the
outer 50 mm. However, the coronal texture and increase in Mn towards the rims unequivocally indicate operation of a retrograde net-transfer
reaction, which must have consumed part of the diffusion profile as it was growing. Thus, the apparent extent of diffusion is much lower than if the rim
had been static. These profiles can be fit to infer relative rates of cation diffusion. Reproduced from Carlson WD (2002) Scales of disequilibrium and
rates of equilibration during metamorphism. American Mineralogist 87: 185–204.
individual matrix biotite grains are rather homogeneous com- for an assumed cooling rate at peak conditions (i.e., s at T ¼ T0
positionally, and although different biotite grains sometimes and DT0), the minimum grain diameter a that preserves core
have different compositions depending on proximity to a garnet compositions can be calculated.
(Figure 8(a); Kohn and Spear, 2000; Spear and Parrish, 1996),
the main difficulties lie in selecting garnet compositions and
4.7.2.6 Kinetically Limited Transport within the Rock Matrix
assessing how much biotite compositions could have changed.
In high-grade garnets, the highest Mg/Fe (lowest Fe/(Feþ Mg)) Compositional zoning in a mineral depends on rates of crystal
ordinarily occurs in the core (Figures 7 and 8). In lower grade growth, ‘nutrient’ supply, and ‘waste’ removal. In the previous
garnets, the lowest Fe/(Fe þ Mg) occurs in an Fe/(Feþ Mg) discussions, it was implicitly assumed that matrix composi-
trough, whose location reflects competition between prograde tions are homogeneous and that equilibrium among grain
growth of garnet, which causes Fe/(Fe þ Mg) to decrease, and edges occurs on scales that are large compared to individual
ReERs or ReNTRs, which commonly cause Fe/(Fe þ Mg) to in- crystals. Whereas this may be true, limited transport of some
crease (Figure 8(c)). Similar troughs are observed in Mn for the elements through the matrix can profoundly impact chemical
same reasons (Figure 8(c)). zoning. If diffusion of nutrients limits the rate of crystal growth
Within this context, Kohn and Spear (2000) developed a (diffusion control), then a simple relationship exists for the
semiquantitative ‘ReNTR insurance policy.’ Because spacing between compositional contours (c) versus radius (C),
the increases in Mn towards garnet rims predominantly result as normalized to the largest grain in the rock (Kretz, 1974).
from garnet resorption, the excess Mn in the rim region rela- A different rate-limiting mechanism, such as adsorption and
tive to the lowest Mn reflects the amount of garnet that was addition of chemical species to the growing crystal surface
resorbed. The Mn-based estimate of this resorbed volume (interface control), yields a different normalized radius–rate
permits resurrection of the original composition of the matrix (c* vs. C*) relationship. This concept has been developed
biotite prior to the ReNTR. This revised biotite composition extensively by Carlson and coworkers (Carlson, 1989, 1991;
can then be paired with the highest Mg/Fe composition garnet Carlson et al., 1995; Chernoff and Carlson, 1997, 1999;
to better estimate peak P–T conditions prior to garnet resorp- Denison and Carlson, 1997; Denison et al., 1997), who have
tion. This approach is analogous to Carlson’s inference of shown that c* versus C* diagrams commonly conform to
material fluxes and resulting ReNTR correction of mineral diffusion control (Figure 11). It is important to recognize
compositions (Carlson 2002), except that his calculations that a c* versus C* plot assumes different diffusivities for
are texturally based whereas Kohn and Spear’s are based on different elements. Specifically, some element must be homo-
X-ray maps of Mn distributions. Application of ReNTR insur- geneous over the sampled region of the rock, and so can be
ance to most rocks in the author’s research collection indi- used to monitor the growth rate of different crystals as a
cates relatively small corrections: 10–20 C, compared to function of crystal size; that is, serves as a geochemical proxy
uncorrected compositions. However in some cases, correc- for time (Carlson, 1989; Kretz, 1974), whereas diffusive trans-
tions can be as large as 200 C (Kohn and Spear, 2000). port of another element limits growth (Carlson et al., 1995).
Virtually all garnets exhibit retrograde Mn increases towards For example, slow Al diffusion could cause diffusion-limited
their rims, so ReNTR insurance should be adopted for most growth, whereas the fast diffusion of Mn provides a time
rocks, as it permits identification and remedial petrologic marker. The success of diffusion control in explaining many
action, if necessary. Although both Carlson (2002) and radius–rate data apparently supports this paradigm.
Kohn and Spear (2000) developed their approaches for am- Other zoning studies further suggest radically different
phibolite-facies ReNTRs, the implications can be generalized intergranular diffusivities for different elements, and a strong
to all metamorphic facies in which diffusion and net transfer dependence on the behavior of a particular element with tem-
reactions can affect mineral compositions during cooling perature. For example, X-ray maps collected by Yang and Rivers
(Spear and Florence, 1992). (2001) show that some elements in garnet behave systemati-
Although the highest Mg/Fe in a garnet typically represents cally, for example, Mn, whereas others have patterns that
the highest temperature composition, Ganguly and Tirone mimic the original texture of the rock, e.g., Cr (see also Kohn,
(1999) derived expressions that describe the degree to which 2004; Martin, 2009). If intergranular diffusion of Mn is fast,
the core of an initially homogeneous mineral is affected by then the rim of the growing garnet always has the same Mn
diffusion during cooling; that is, whether a high Mg/Fe core content, and grains zone systematically. But if intergranular
can retain its peak-T composition. The parameter M is impor- diffusion of Cr is extremely slow, then local mineralogical
tant to this discussion: differences in Cr content due to textural variability in mica or
oxide abundances directly control local supply of Cr to differ-
DT0 T 2
M¼ [20] ent edges of the growing garnet, and grains develop heteroge-
sa2 E=R
neous zoning. At low grade, even Mn zoning can be quite
where DT0 is the diffusivity at peak temperature T0, and s is the heterogeneous (Daniel and Spear, 1998; Hirsch et al., 2003;
cooling rate at the temperature T. In deriving eqn [19], Dodson Spear and Daniel, 1998, 2001), which presumably reflects
(1986) assumed M1, so that the core of the mineral is extremely slow intergranular diffusion and incomplete matrix
initially homogenized by diffusion. By contrast, Ganguly and homogenization, causing locally heterogeneous supply of Mn.
Tirone (1999) consider cases in which a crystal is initially Rates of intergranular chemical transport are extremely im-
homogenized by some unspecified mechanism, and M <1. portant because they govern the length scales over which equi-
They show that for spherical minerals, the core of a mineral librium occurs. For example, the generally systematic behavior
is essentially unaffected by diffusion for values of M 0.1. Thus of Mn zoning in most garnets (albeit not garnet cores studied
by Daniel and Spear, 1998) implies equilibrium can be as- for trace elements, which are believed to have transport behav-
sumed over several grain diameters, and that equilibrium ther- ior more like Cr and less like Mn, and for accessory minerals,
modynamics can be readily applied to a whole rock. However, whose occurrences are controlled by trace elements.
the heterogeneous zoning of Cr implies that it is dominated by
local effects, and hence is much less easily modeled. These
differences in element mobility become increasingly important
4.7.2.7 Dissolution–Reprecipitation
Dissolution–reprecipitation may be defined in two ways: either
the dissolution of mineral grains in one part of a rock, and
7
reprecipitation of compositionally different grains or zones in
FeO
6 another part of the rock; or the simultaneous replacement of a
MgO
MnO Thermally grain or zone within a grain by the same mineral with a
different composition. These definition differences are not
c* (normalized rate)
5 accelerated
diffusional emphasized here. Dissolution–reprecipitation is ultimately
control
4 caused by extremely slow cation diffusivities, which prevent
changes to mineral composition demanded by changes to P, T,
3 or fluid composition. Few examples for garnet are known
l
ro outside of polymetamorphic rocks (e.g., Karabinos, 1984; Pollok
o nt
2 o nc et al., 2008; see also Martin et al., 2011; and Section 4.7.3.3),
fusi
al dif
Isotherm yet plagioclase commonly exhibits this type of zoning. Anor-
1
Interface control thite and albite end-members are linked via a coupled sub-
stitution (CaIVAlNa1Si1), so diffusive reequilibration of XAn
0
1.0 0.8 0.6 0.4 0.2 0.0 requires exchange of tetrahedral Al and Si, which is extremely
C* (normalized radius) slow (Grove et al., 1984). Therefore, the ubiquitously observed
zoning in XAn commonly results from dissolution of old grains
Figure 11 Plot of normalized growth rate (c*), as determined from
chemical gradients, versus normalized radius (C*). Normalization is and reprecipitation of new grains or overgrowths (Figures 12
relative to composition gradients and radius of the largest crystal in a and 13).
rock. Interface control implies that c* will be constant for all radii, In low-Ca bulk compositions, which are typical of metape-
whereas diffusion control implies an increasing c* with decreasing C*. lites, plagioclase compositional change is intimately tied to
Thermally accelerated diffusion control (Carlson, 1989) shifts the garnet growth and consumption because of Ca mass balance
diffusion-control curve to the right and permits c* values less than 1.0. in the rock (Figure 14). Garnet and plagioclase are commonly
If garnets grow over a larger range of T, then they should plot farther to the only Ca-bearing silicates in a metapelite, so growth or
right of isothermal diffusion control. Data from a pair of garnets are consumption of Ca-bearing garnet consumes or produces the
better described by thermally accelerated diffusion control than by either
anorthite component of plagioclase. Plagioclase cannot adjust
interface control or isothermal diffusion control. Modified from Carlson
composition diffusionally, so thermodynamics drives grain
WD (1989) The significance of intergranular diffusion to the mechanism
and kinetics of porphyroblast crystallization. Contributions to Mineralogy dissolution and reprecipitation. Such direct modal and com-
and Petrology 103: 1–24; Denison C and Carlson WD (1997) positional links among minerals must be accounted for in
Three-dimensional quantitative textural analysis of metamorphic rocks interpreting chemical zoning patterns. Even if the overall
using high-resolution computed X-ray tomography: Part II. Application to mode and composition of reactive plagioclase changes sys-
natural samples. Journal of Metamorphic Geology 15: 45–57. tematically, local dissolution or precipitation could occur
Garnet grows Continuous Continuous Discontinuous

dissolution precipitation precipitation
An40
An40 An40 An40
Time 1
An40 An40 An35

An35 An40
An40
Time 2 An35
An40
An35
An35
An40 An40
An40 An30
Time 3 An35
(a) An30 (b) (c) (d) An30
Figure 12 Sketch of plagioclase zoning patterns and compositional trends. Plagioclase cannot change composition significantly via diffusion and
so must dissolve and reprecipitate to maintain equilibrium with other minerals in a rock. (a) Growth of garnet drives plagioclase compositional changes.
Local textural and chemical variations can cause disparate zoning patterns in different plagioclase grains, which can be reflected in compositional
differences among plagioclase grains at the same radial distance from the garnet core. (b) Plagioclase dissolves continuously. (c) Plagioclase
precipitates continuously. (d) Plagioclase precipitates discontinuously. Reproduced from Spear FS (1993) Metamorphic Phase Equilibria and
Pressure–Temperature–Time Paths. Washington, DC: Mineralogical Society of America.
Pl 10
0.33
8 +Grt
–Pl
0.26 dM(plagioclase) = 0
P (kbar)
Fractional
Pl
–Grt crystallization
+Pl
Discontinuity 4
30 % M
10% M
0% Mic
20
5% Mica
% ic
M a
0.09
ica
0.15
ica
2
a
dM(garnet) = 0
Grt
0
400 500 600 700 800
Ca T (˚C)
Figure 13 X-ray map of Ca in garnet gneiss from the Greater Himalayan Figure 14 Modal isopleths of plagioclase in the assemblage
Sequence, central Nepal, showing compositional systematics in garnet þ biotite þ chlorite þ plagioclase þ quartz þ white mica
plagioclase (upper grains). A garnet grain (lower left, with inclusions) is (muscovite) þ H2O as modeled in the system MnO–Na2O–CaO–K2O–
also present. White numbers are mole fraction of grossular and anorthite. FeO–MgO–Al2O3–SiO2–H2O, assuming fractional crystallization of
Upper left grain of plagioclase is quasi-continuously zoned towards garnet. Stability fields of the aluminosilicate polymorphs are shown for
higher XAn (see Figure 12(c)). Upper right grain has continuous zoning in reference as straight lines. Different curves for plagioclase correspond to
one region and a compositionally distinct high XAn overgrowth on one different abundances of white mica, which affect plagioclase abundances
side (see Figure 12(d)). Retrograde consumption of garnet released Ca because of the paragonite component’s influence on Na mass balance.
to the matrix, stabilizing plagioclase with higher XAn, and causing both Black dot shows reference P–T conditions. Black arrows show that garnet
continuous and discontinuous growth of these plagioclase grains. growth generally causes a decrease in plagioclase abundance, whereas
Scale bar is 500 mm. Reproduced from Kohn MJ, Catlos EJ, Ryerson FJ, garnet consumption causes an increase in plagioclase abundance.
and Harrison TM (2001) Pressure–temperature–time path discontinuity Changes in XAn broadly follow plagioclase modal abundances, so garnet
in the Main Central thrust zone, central Nepal. Geology 29: 571–574. growth generally causes a decrease in XAn, whereas garnet consumption
generally causes an increase in XAn. Reproduced from Spear FS, Kohn
MJ, Florence F, and Menard T (1990) A model for garnet and plagioclase
continuously or sporadically, leading to different textures in growth in pelitic schists: Implications for thermobarometry and P–T path
different grains (Figures 12 and 13). Obviously, compositional determinations. Journal of Metamorphic Geology 8: 683–696.
zoning in one grain can be perplexing or even misleading
unless X-ray maps or backscattered electron images of many 1000 ln aij d18 O ð iÞ d18 O ð jÞ ¼ Dij [22]
grains are collected and the overall sense of zoning compared
Partitioning is almost solely dependent on absolute
with theoretical expectations. In general, if a mineral contains
temperature:
an essential constituent that is shared by other minerals in the
rock, and that constituent has an extremely slow intracrystal- Aij Bij
Dij ¼ þ þ Cij [23]
line diffusivity, then chemical or modal changes in the other T2 T
minerals will drive dissolution–reprecipitation.
where A, B, and C are constants. For relative partitioning of
oxygen among many silicates and oxides, B and C are nearly
zero and can be ignored, but for fluids, all terms must be used
4.7.3 Stable Isotopes
(e.g., see summary by Chacko et al., 2001). Pressure depen-
dencies are strongly subsidiary to temperature and are similar
There are fewer studies of stable isotope zoning in metamor-
in most minerals (Polyakov and Kharlashina, 1994), so that
phic minerals because measurement is relatively difficult. This
mineral–mineral Ds are nearly P-independent. Equation [23]
section will focus on oxygen, because there are more studies
indicates that mineral isotopic compositions converge with
and because oxygen transport is more readily modeled. Growth
increasing temperature.
zoning, diffusion zoning, and dissolution–reprecipitation are
best documented.
Isotopic compositions are usually expressed in permil no-
tation, relative to a standard, such as standard mean ocean 4.7.3.1 Growth Zoning
water (SMOW): Many orthosilicates have slow oxygen diffusion rates and
can readily develop growth zoning, whereas many sheet and
Ri RSMOW
d18 OðiÞ ¼ 1000 [21] tectosilicates have fast diffusion rates and instead control the
RSMOW
oxygen isotope mass balance of a rock. A rigorous theoretical
where Ri and RSMOW are the ratios of 18O to 16O in the material approach for predicting changes in isotopic compositions in
i and SMOW, respectively. Partitioning between two materials minerals was derived independently (Kohn, 1993a; Young,
i and j is usually expressed in terms of a fractionation factor 1993) by incorporating a set of oxygen isotope partitioning
aij, where equilibria and mass balance equations into the differential
thermodynamics method. These differential isotopic equations the whole rock, and mineral compositions converge with in-
maintain partitioning equilibrium: creasing temperature (eqn [23]). However, for nearly discon-
tinuous reactions, such as
2Aij Bij
dDij ¼ þ 2 dT [24]
T 3 T Garnet þ Chlorite þ Muscovite
¼ Biotite þ Staurolite þ Quartz þ H2 O [26]
(ignoring pressure-dependencies) and oxygen isotope mass
balance mineral modes change so rapidly that isotopic mass balance
(eqn [25]) controls compositions, as indicated by the coinci-
X
n X
n
dd18 Osys Nsys ¼ Mk Nk dd18 Ok þ Nk dd18 Ok dMk ¼ 0 [25] dence of compositional and modal slopes for the staurolite-in
k¼1 k¼1 reaction. However, such reactions also rapidly exhaust a reac-
tant, so their isotopic influence is ordinarily rather small.
where d18Osys is the whole-rock oxygen isotopic composition,
Models of growth zoning can be constructed by incorporating
Nsys is the total number of moles of oxygen in the system, Mk
fractional crystallization. Generally, in a closed system, the
and Nk are the number of moles and the number of moles of
strongest prograde isotope zoning will result from continuous
oxygen in the kth mineral, and d18Ok is the d18O of the kth
reactions that occur over a large temperature range. Even so,
mineral. For a rock with n minerals, there are n new variables
predicted growth zoning is subdued, and resolution of the
(the d18Oks) and n new equations (n 1 in the form of eqn
small changes in isotopic composition indicated by the models
[24], and one in the form of eqn [25]). Thus, the total linear
requires an analytically and spatially precise technique. Such
algebraic variance of the combined system of thermodynamic,
profiles could not be routinely measured until the advent of
mass balance, and isotope partitioning equations remains two,
miniaturized laser fluorination systems (Sharp, 1990, 1992)
and changes in only two variables (e.g., dP and dT ) need to be
and multicollector, large-radius ion microprobes (e.g., Fayek
specified to model changes in the d18Oks.
et al., 2002; Page et al., 2007; Vielzeuf et al., 2005a).
When mineral modes change slowly as a result of continu-
Oxygen isotope zoning in garnet was documented by the
ous reaction over a large temperature range, the temperature
early 1990s (Chamberlain and Conrad, 1991, 1993; Jamtveit
dependence of isotopic partitioning controls compositions
and Hervig, 1994; Young and Rumble, 1993), but Kohn et al.
(Figure 15). For example, garnet in a model metapelite that
(1993) described the first isotopic zoning profiles that clearly
contains biotite and chlorite simply increases in d18O with
conformed with independent predictions of growth models.
increasing temperature because it has a lower d18O value than
The observed monotonic 0.5% increase in d18O from core to
rim (Figure 16) is consistent with prograde growth over the
75 C temperature interval that had been inferred from major
10 element zoning. Staurolite from the same sample had no re-
d18OGrt Isopleths
0.0
+0.1
+0.2
solvable zoning in its interior, which again was consistent with

+0.3
Grt+Bt+Chl
closed-system models. The internal consistency between the
+0.4
8
+0.5
theoretical models and observed zoning in this study implies

+0.6
Grt+Bt+St
that isotopic partitioning was maintained during prograde
6 Grt+Bt+ metamorphism, and that the rock was closed to infiltration
P (kbar)
+0.4
+0.5
+0.3
St+Alsi
+0.2
1.2
4
+0.1
1.0 Alsi-in St-in

(600 ⬚C) (575 ⬚C) 5.0
0.8
Garnet
Δδ18O
Chl-out Independent
0.6
0.0
Grt+Bt+Chl+St a
d18O (‰) V-SMOW
2 theoretical
4.5
0.0
0.4
a⬘
.2
500 ⬚C St-out model

-0
0.2
(611 ⬚C)
0.0
Core Rim
0 4.0
400 500 600 700 800
T (⬚C)
3.5
18
Figure 15 Isopleths of garnet d O in a model metapelite, showing = ±2σ
nearly P-invariant orientations in a garnet þ biotite þ chlorite and
garnet þ biotite þ staurolite assemblages and slopes nearly 3.0
0 1 2 3 4 5 6 7 8
parallel to reaction boundaries (dotted isopleths) in a
Rim Core Rim
garnet þ biotite þ chlorite þ staurolite assemblage. Stability fields of the (a) (a⬘)
Distance (mm)
aluminosilicate polymorphs shown for reference as straight lines.
Absolute d18O values are arbitrary, and are zeroed at a particular isopleth Figure 16 Oxygen isotope compositions of a growth-zoned garnet from
to illustrate compositional differences within each assemblage field. Tierra del Fuego, Chile, showing a general 0.5% increase in d18O
Inset shows isotopic zoning predicted for a garnet that grows isobarically from core to rim. This zoning is consistent with independent calculations
from 500 to 611 C at 6 kbar, showing development of zoning and how of oxygen isotope growth zoning in a closed chemical and isotopic
mineral reactions affect zoning profile. Dotted lines with arrows show system. Two-sigma analytical uncertainty shown as a vertical bar.
compositional trajectory of the rim, which is not preserved because of Reproduced from Kohn MJ, Valley JW, Elsenheimer D, and Spicuzza MJ
garnet dissolution and subsequent regrowth. Reproduced from Kohn MJ (1993) Oxygen isotope zoning in garnet and staurolite: Evidence for
(1993) Modeling of prograde mineral d18O changes in metamorphic closed system mineral growth during regional metamorphism.
systems. Contributions to Mineralogy and Petrology 113: 24–39. American Mineralogist 78: 988–1001.
by a fluid of isotopically distinct composition. Conversely, one garnets. These dioritic migmatites physically and chemically
of the main objectives of measuring oxygen isotope zoning interacted with mafic magmas, ultimately reaching peak tem-
profiles is to identify open system processes by seeking out peratures of 860 C (Vielzeuf, 1996) and establishing unusual
core–rim isotopic trends that are inexplicable by closed-system chemical discontinuities which then diffusionally relaxed.
models. Of significance, phosphorous exhibits a major step between
core and rim, apparently preserving a change in matrix bulk
composition and defining the core–rim chemical boundary.
4.7.3.2 Diffusion
By contrast, smooth gradients in oxygen isotopes and major
Different minerals have radically different oxygen diffusivities elements are interpreted to reflect diffusional smoothing
(e.g., see review by Cole and Chakraborty, 2001), so there is (Figure 17). Interestingly, the diffusivity of oxygen appears
potentially a wealth of petrologic information encoded in the comparable to that of Ca. Because oxygen diffusion rates in
stable isotopic products of diffusion. The effects of diffusive garnets are approximately two orders of magnitude slower than
exchange on bulk mineral compositions are widely studied, Mg, Mn, and Fe (Carlson, 2006; Coghlan, 1990; Ganguly et al.,
mostly an outgrowth of the seminal work of Giletti (1986). 1998a), Ca may diffuse much more slowly than other divalent
He used the closure temperature concept of Dodson (1973) cations (Kohn, 2004; Spear and Kohn, 1996; Vielzeuf et al.,
together with the temperature dependence of isotope parti- 2007).
tioning and oxygen isotope mass balance to describe the
changes to bulk mineral composition that occur in a rock
during cooling. To a first order, his model is accurate for 4.7.3.3 Dissolution–Reprecipitation
slowly cooled rocks. However, the analytical expression of
Hot fluids readily dissolve and transport rocks. If such fluids
Dodson (1973) requires a uniform matrix reservoir, whereas
are isotopically distinct, then stable isotopes can help delineate
the matrix adjusts its partitioning character as minerals succes-
fluid–rock interaction. Isotopic processes at deeper levels are
sively close to isotope exchange. Therefore, Eiler et al. (1992,
likely to be rather subtle, because temperatures are higher,
1993) modified Giletti’s model, accounting for modes, grain
timescales are longer, and different oxygen reservoirs may not
sizes, and diffusive profiles. The Eiler et al. models are more
be very distinct isotopically. Therefore, contact metamorphic
rigorous, explain more observations, especially from mineral-
systems are commonly presented as examples of fluid–rock
ogically uncommon rocks, and predict isotopic zoning profiles
interaction and dissolution–reprecipitation, principally focus-
in each mineral. Jenkin et al. (1994) independently modeled
ing on quartz and calcite, which can be imaged for trace
and explored mode and grain size effects using a similar
elements using cathodoluminescence.
approach.
In contrast to several studies of hydrothermal systems,
An important feature of oxygen diffusion rates is a very
Martin et al. (2011) argue that garnet can dissolve and repre-
strong dependence on water fugacity fH2O (for a detailed
cipitate even in regional metamorphic settings (Figures 18
discussion, see Cole and Chakraborty, 2001, p. 126–155),
and 19). They analyzed amphibolite-facies metapelitic garnets
although there are different explanations as to why higher
from Greece that show unusual fluid inclusion clouding in
fH2O causes radically higher Ds (e.g., Doremus, 1998, 1999;
garnet rims (Figure 18) and convoluted yet sharp core–rim
Elphick and Graham, 1988; Graham and Elphick, 1990;
boundaries, as defined both texturally and chemically
McConnell, 1995; Zhang et al., 1991). Thus, the fH2O history
(Figure 19). Some garnets show large embayments and fluid
strongly influences diffusion rates and isotope compositions
inclusion clouding along zones interpreted as relict fractures.
and should be taken into account explicitly in diffusion
Increased porosity and sharp core–rim boundaries are both
models (Kohn, 1999). Conversely, if grain size, grain geome-
characteristic of simultaneous dissolution–reprecipitation
try, and cooling rate are independently characterized, then
(Putnis, 2002). Composition profiles show lower d18O values
differences in isotopic composition or zoning patterns can
for garnet rims versus cores and dramatic linear changes in
potentially reveal variations in fH2O which are otherwise diffi-
major elements towards garnet rims; these patterns are incon-
cult to ascertain from mineralogy alone (e.g., Edwards and
sistent with either closed-system behavior or diffusional ree-
Valley, 1998).
quilibration (Kohn, 1993a; Kohn and Spear, 2000; Spear et al.,
There are few published oxygen isotope diffusion profiles in
1990b). Martin et al. argue that original garnets underwent
natural samples, mainly due to analytical difficulties. The
dissolution–reprecipitation, resulting in increased porosity,
length scale of a natural diffusion profile is commonly on the
now reflected in distinctive fluid inclusion clouds and steep
order of several micrometers to 100 mm, and the magnitude
gradients in major elements and oxygen isotopes.
of the isotopic change is small (<1%) except in the outer few
micrometers. Thus, extreme spatial and/or analytical precision
is required to demonstrate that an isotopic profile exists,
let alone that it is the product of diffusion. Laser fluorination 4.7.4 Trace Elements
is analytically the most precise and accurate (0.1%) method
by which to determine a mineral’s d18O, but has relatively Trace element zoning is probably ubiquitous in metamorphic
coarse spatial resolution (250 mm); ion microprobe analysis minerals. Pioneering analysis by Hickmott and coworkers
is spatially superior (10 mm), but analytically inferior established trace element zoning as a sensitive monitor of
(0.2–0.4%). P–T histories (Hickmott and Spear, 1992; Hickmott et al.,
Granulite facies garnets from Spain (Vielzeuf et al., 2005b; 1987), changes in major mineral assemblages (Hickmott and
Figure 17) show compelling evidence for oxygen diffusion in Spear, 1992), abrupt breakdown of trace element-rich minerals
CaO 900
Zone II Zone I
% 800
Interface
Garnet boundary
4.8
700
P2O5 (ppm)
2.8
600
500
400
Discon-
300 tinuity
200
100
(c) Core
Zone II Zone I
14
13
12
Grs (mol%)
2000 mm (a) 11 Dt = 1.6e–8
10 Dt = 1.6e–7
9 Dt = 1.6e–9
11.4 8
Interface
9.5
7
10.3 6
(d) Core
10.6 9.9
10.2
Ib
Interface
12.6 11.6 IIa 14
13.2 13.2
d18OSMOW (‰)
12.7 13.9 13
13.5
Ia 12
13.9 10.1 Dt = 3.2e–9
13.0
Dt = 3.2e–8
13.8 10.1 11
12.0
13.7 13.7
12.8 11.2 10 Dt = 3.2e–9
500 mm 13.8
13.7
(b)
12.8 (e)
0 500 1000 1500 2000 2500

Distance from rim to core (mm)
Figure 17 Compositions of complexly zoned granulite facies garnets from Spain showing evidence for oxygen isotope diffusion in garnet. (a) X-ray
map of CaO shows low Ca cores (darker) and high Ca rims (lighter) in garnets. White box shows region of (b). (b) X-ray map of P2O5, showing
abrupt compositional shift between core and rim and locations of ion microprobe spots for oxygen isotope analyses. (c–e) Composition versus distance
for P2O5, mole percent grossular, and d18O, showing abrupt shift in phosphorus content and smooth changes in grossular and d18O. Gray lines
show best fit diffusion profiles, and illustrate comparability of Ca and O diffusion rates. Values for Dt (the diffusion coefficient times time) are in square
meters. Reproduced from Vielzeuf D, Veschambre M, and Brunet F (2005) Oxygen isotope heterogeneities and diffusion profile in composite
metamorphic–magmatic garnets from the Pyrenees. American Mineralogist 90: 463–472.
(Hickmott and Shimizu, 1990), and infiltration of fluids

4.7.4.1 Growth Zoning
(Hickmott et al., 1992; see also Hervig and Peacock, 1989).
However, the need for an ion microprobe to make such mea- Rayleigh distillation models have been developed for geochro-
surements likely deterred most petrologists. Renewed interest nologically useful trace elements such as Lu–Hf, Sm–Nd, and
has been spurred by the recognition that many trace elements Rb–Sr (Kohn, 2009; Lapen et al., 2003), but do not address
can be investigated with the electron microprobe (Spear and changes to mineral assemblages and trace element partition-
Kohn, 1996) and LA-ICP-MS (e.g., Bea et al., 1996) and by ing. More rigorous thermodynamically based petrogenetic
increased petrologic and geochronologic focus on accessory models for trace elements and accessory minerals have only
minerals whose stabilities depend on trace elements. recently been developed, focusing on Y in garnet and monazite
Unlike major elements, which can homogenize at high T (Spear and Pyle, 2010) and monazite, allanite, and zircon
via simple interdiffusion and exchange with matrix minerals, stabilities (Kelsey and Powell, 2011; Kelsey et al., 2008;
trace elements often have different charges compared to the Spear, 2010). Spear and Pyle’s work (Figure 20) illustrates
major elements for which they substitute, which is expected to changes in garnet Y content in a prograde metamorphic se-
impede diffusion. For example, to maintain charge balance, quence, and the demise of xenotime as an accessory mineral.
the PIV–SiIV, NaVIII–MgVIII, YVIII–MgVIII, and TiVI–AlVI ex- Garnet is a common target of Y zoning studies because it
changes all require coupled substitution of other cations on readily accommodates Y as ‘YAG’ (Y3Al5O12), where substitu-
other sites, such as AlIV–SiIV. This fact may permit preservation tion of Y3þ for divalent cations in the cubic site is charge-
of trace element zoning in minerals that otherwise lack obvi- balanced by substitution of tretrahedral Al3þ for Si4þ. In gen-
ous major element zoning. eral, garnet maintains high Y contents while xenotime is
et al., 2006; Yang and Pattison, 2006). Equilibrium models,

while a useful benchmark, may not be as widely applicable for
Outer rim
Pl trace elements as for major elements or stable isotopes.
4.7.4.2 Diffusion
Core
LA-ICP-MS analysis now resolves diffusional reequilibration of
trace elements in garnets. One example from high-grade rocks
Inner rim in central Nepal (Kohn, 2009) shows quasi-bell-shaped distri-
butions in Y and Lu, but with near-rim troughs and increases
(a) 100 mm on the rims, just as is commonly observed in Mn (Figure 22).
The compatibility of Y, Lu, and Mn in garnet implies that all
Pl elements will behave similarly during growth, with high con-
Pl centrations in cores and lower concentrations towards rims.
Outer rim Thus the troughs in Lu and Y are interpreted similarly to those
in Mn; that is, they reflect high-temperature dissolution with
A
Core backdiffusion towards garnet cores. Recent comparison of dif-
fusion lengths for major and trace elements suggests that triva-
lent cations diffuse roughly two orders of magnitude more
B slowly than divalent cations (Carlson, 2010), consistent with
experiments on rare-earth element (REE) diffusion rates
Inner rim (Coghlan, 1990; Ganguly et al., 1998b; Tirone et al., 2005;
100 mm Van Orman et al., 2002).
(b)
Embayments Outer rim Trace elements can provide a sensitive monitor of dissolution
and reprecipitation or regrowth. As summarized by Yang and
Inner rim Rivers (2002) (Figure 23), if an element is compatible in a
mineral, then resorption will increase the concentration of the
element at the rim. Subsequent regrowth will cause depletion,
leading to an annulus in that element (see also Hickmott et al.,
Core 1987; Kohn et al., 1997; Pyle and Spear, 1999; Spear et al.,
1990a). For an incompatible element, resorption will cause a
decrease in concentration at the mineral rim. Subsequent
regrowth will produce an increase, leading to a moat. The
development and preservation of an annulus or moat then de-
pends on the partition coefficient (the degree of compatibility),
the amount of material dissolved and then regrown, and trace
(c) Plagioclase element diffusivities. Importantly, if diffusivities are extremely
slow, the step on the inner side of the annuli and moats will be
Figure 18 Images of garnet from Ikaria, Greece. (a) Clouding in garnet extremely sharp, whereas if diffusion is nonnegligible, then
rims results from abundant fluid inclusions. (b) Backscattered electron backdiffusion towards the mineral core will smooth the profile.
image shows location of composition traverse and ion microprobe spots
A staurolite-grade garnet from New Hampshire exhibits a
for oxygen isotope analysis. Line A–B is location of major element
Y-annulus in garnet (Pyle and Spear, 1999; Figure 24). Most
traverse in Figure 19. (c) Calcium composition map shows convoluted
boundary between core and rim. Reproduced from Martin LAJ, Ballevre garnets in staurolite-bearing rocks are predicted to have origi-
M, Boulvais P, Halfpenny A, Vanderhaeghe O, Duchene S, and Deloule E nally grown in an assemblage containing garnet þ chlorite þ
(2011) Garnet re-equilibration by coupled dissolution–reprecipitation: biotite (Spear et al., 1990b). At the staurolite in isograd, garnet
Evidence from textural, major element and oxygen isotope zoning of and chlorite are consumed via the staurolite in reaction (eqn
‘cloudy’ garnet. Journal of Metamorphic Geology 29: 213–231. [26]). After chlorite is completely consumed, garnet can regrow
via continuous staurolite and biotite breakdown. This provides
present, then drops to low values as xenotime gives way to a thermodynamic driving force to produce the observed
monazite. Models of Zr distributions in migmatitic rocks Y-annulus. Interestingly, the annulus is not completely sharp
(Figure 21) show dependencies of garnet Zr content on P–T at the inner boundary, but instead is rounded on a scale of 10s
conditions and mineral assemblages (Kelsey and Powell, 2011). to 100 mm. This suggests that Y diffusivity cannot be several
It is emphasized, however, that trace element zoning in many orders of magnitude slower than divalent cations. Alterna-
garnets is strongly affected by other processes, including rates of tively, similar annuli in other garnets have been interpreted
intergranular diffusion and destabilization of trace element-rich to record breakdown of accessory and minor minerals, espe-
accessory minerals (e.g., Chernoff and Carlson, 1999; Corrie cially epidote-group minerals (Chernoff and Carlson, 1999;
and Kohn, 2008; Hickmott and Spear, 1992; Hickmott et al., Corrie and Kohn, 2008; King et al., 2004; Konrad-Schmolke
1987; King et al., 2004; Konrad-Schmolke et al., 2008; Skora et al., 2008; Yang and Pattison, 2006).
80 25
Inner rim Core Inner rim
75 20
Outer rim
15 11
70
10
10 9.5 9.7 9.6
65 9.3 9.4 9.4
5 9.2 9.5
9 8.5
Almandine Grossular 8.7
60 0
8 7.7 7.6
13 12
7 7.4
11
Mg# 8 d18O
6
9 0 50 100 150 200 250 300 350 400
4 Profile (mm)
7 Theoretical δ18O value of garnet (6.7 ± 0.4‰)
Pyrope
Spessartine in equilibrium with matrix quartz at 520 °C
5 0
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
A Profile (mm) B Profile (mm)
Figure 19 Compositional traverse across garnet from Ikaria, Greece, shows distinct core versus rim zoning patterns for major elements (in mole
percent) and d18O (in %, V-SMOW). Garnet rim d18O approaches the expected value determined from matrix quartz. Reproduced from Martin LAJ,
Ballevre M, Boulvais P, Halfpenny A, Vanderhaeghe O, Duchene S, and Deloule E (2011) Garnet re-equilibration by coupled dissolution–reprecipitation:
Evidence from textural, major element and oxygen isotope zoning of ‘cloudy’ garnet. Journal of Metamorphic Geology 29: 213–231.
12
+Qtz 1.5
2
+Pl 2.5 XYAG(Grt, ‰)
+ Mnz 3
+ Ap
10
4
0.016
5 Grt Bt
8 Grt Bt Ms 0.014
Chl Ms Fl
6
Fl Grt Bt
0.012
Pl-out
8
St Ms
P (kbar)
Grt Bt
3 Fl
6 10 AlSi Kfs
2 0.010
XYAG in garnet
15 2 4 L
20 1 10 –Xno 1
o
3 +Xn 0.008
2
2.5
4 5
Grt Bt 3
Grt Bt
o
AlSi Ms Fl
o
0.006
–Xn
+Xn
3 Crd Kfs
Alsi-in, St-out
St-in, Chl-out
2
Qtz-o
2 L
Bt Chl
Xno-out
Grt Bt
Grt-in
Ms 0.004
ut
Fl Crd Ms
2 2 Grt Bt
–X no
Fl
+X
no
Crd Kfs
1.5 Fl 0.002
0 0.010
400 500 600 700 800
(b) 1 2 3 4
(a) T (°C)
Figure 20 (a) Pseudosection of XYAG in garnet in pelitic assemblages. All models include plagioclase, quartz, and apatite, as well as muscovite,
K-feldspar, or melt (depending on P–T conditions). Fl ¼ H2O fluid; L ¼ silicate liquid. Xenotime is stable only at relatively low P. XYAG contours are
strongly dependent on major mineral assemblages. Gray line is the modeled path with garnet fractional crystallization. Numbered reactions are:
1 ¼ garnet-in, 2 ¼ xenotime-out, 3 ¼ staurolite-in and chlorite-out, 4 ¼ aluminosilicate-in and staurolite-out. (b) Modeled composition of garnet,
assuming fractional crystallization along the P–T path in (a), showing high XYAG contents while xenotime is present, and rapid drop to low XYAG contents
after crossing xenotime-out. Reproduced from Spear FS and Pyle JM (2010) Theoretical modeling of monazite growth in a low-Ca metapelite. Chemical
Geology 273: 111–119.
2002) and, in some instances, ascribed to limited transport of

4.7.4.4 Kinetically Limited Transport within the Rock Matrix
elements through the rock. For example, garnets from Picuris
Trace element annuli and spikes in garnets have been found quartzites have Yb, Y, P, Ti, and Sc annuli or moats that are
over a range of metamorphic grades (Chernoff and Carlson, spatially coincident with Ca spikes (Chernoff and Carlson,
1999; Lanzirotti, 1995; Pyle and Spear, 1999; Pyle et al., 2001; 1999; Figure 25). Kinetically limited intergranular transport
Skora et al., 2006; Spear and Kohn, 1996; Yang and Rivers, of Ca, Yb, Y, P, Ti, and Sc is expected to produce chemically and
12 ms bt
12
–6 –6
6
grt pl kfs
10 7 10
10 –
11 11 21 29
liq qtz zrn x 14 x
ms bt (v = 3) ky
12 36 –6
x
ms grt pl liq sil 10 0
10 x1
24
qtz zrn 7 14 21 29
bt
grt pl (v=4) grt als pl kfs 48
9 liq qtz zrn grt 9 7 14 21 29
–6
36
36
H 2O (v=4) sil 0
8
qtz bt grt als grt pl liq
8
14
21 x1
zrn pl kfs sil qtz 28 60 36
Pressure (kbar)
Pressure (kbar)
pl liq (v=6)
rn
liq qtz zrn qtz z 43 43
zz
qt il
7 rn 7 21 28
liq t s
(v=3) 35
ms bt sil
pl H2O qtz zrn
pl s b grt sil crd pl kfs liq qtz zrn grt crd pl

m
liq qtz 6 43
0 –6
6 grt crd pl 21 35
fs (v=6) 28 42
pl k zrn
liq qtz zrn
grt crd pl ksp liq qtz zrn
sil
x1
(v = 5)
5 bt liq qtz bt grt (v = 4) 5 27 35
42
42
crd pl
72
kfs liq liq
4 qtz zrn grt opx
crd pl pl
rd 7) 4 27 34
kfs liq qtz
zrn 5) xc =
(v = op ( v
rn
3ansil bt crd pl
kfs liq qz 3
d opx crd pl l liq
qtz zrn kfs liq dp zrn 2
2 bt crd pl xc iq
qtz zrn op pl l crd pl liq
kfs H2O rd 6) (v = 8)
p x c (v =
1 qtz zrn o 1
opx crd pl liq
H2 O zrn Garnet Zr content
opx crd pl kfs H2O qtz zrn
650 700 750 800 850 900 950 650 700 750 800 850 900 950
(a) Temperature (⬚C) (b) Temperature (⬚C)
Figure 21 Pseudosection of migmatitic pelitic assemblages in NCKFMASH–Zr system, with 200 ppm Zr. (a) Assemblage stability fields – lighter
shading indicates lower variance assemblages. Dotted and dashed lines bound the region of stable melt þ zircon; liq ¼ liquid, als ¼ aluminosilicate.
(b) Garnet Zr content in migmatitic pelitic assemblages, showing dependence on assemblage as well as temperature. Numbers along contours are mole
fraction of Zr component in garnet; small numbers and dots show Zr concentration in garnet in ppm. Reproduced from Kelsey DE and Powell R (2011)
Progress in linking accessory mineral growth and breakdown to major mineral evolution in metamorphic rocks: A thermodynamic approach in the
Na2O–CaO–K2O–FeO–MgO–Al2O3–SiO2–H2O–TiO2–ZrO2 system. Journal of Metamorphic Geology 29: 151–166.
(a) 0.08 minerallogically distinct zones that are concentric about each
(b)
growing porphyroblast (Chernoff and Carlson, 1997, 1999).
0.06
As the zones grow outward through time, they begin to im-
XSps pinge upon each other, possibly causing abrupt changes to
0.04
local mineralogy. If a trace phosphate or other mineral within
0.02 one of these zones suddenly reacted out, then many elements,
Mn 0.00 1.25 mm including P, Ca, and REEs, might suddenly change local con-
800 80
centration, which would be recorded as a spike or step in
(c) 31a
Yttrium garnet compositions. Petrologic interpretations may be further
refined by the shape of the profiles (Figure 23), which ele-
600 60 ments are enriched, and whether element enrichments versus
depletions are systematic with respect to garnet compatibility
Lu, Hf (ppm)
(Yang and Rivers, 2002). For example, sudden enrichment in

Y (ppm)
400 40 LREEs could reflect breakdown of apatite, allanite, or mona-

zite; enrichment in MREEs could reflect breakdown of epidote;
Lutetium Trough and enrichment of HREEs could reflect breakdown of xenotime
Trough
or zircon (Corrie and Kohn, 2008; King et al., 2004; Konrad-
200 20
Schmolke et al., 2008; Yang and Rivers, 2002). Identification of
which reactions occurred in a particular rock requires careful
Hafnium characterization of textures, zoning patterns, and inclusion
0 0 suites.
0.0 0.5 1.0 1.5 2.0 2.5
Distance (mm)
Figure 22 (a) Mn X-ray map, (b) rim to core Mn profile, and
(c) rim-to-rim Y, Yb, and Hf profile for upper amphibolite-facies 4.7.4.5 Oscillatory Zoning, Sector Zoning,
metapelitic garnets from the Nepal Himalayas. Mn, Y, and Yb show high and Textural Mimicry
concentrations in core, dropping towards lower concentrations outward
Individual annuli are discussed in the context of garnet growth
(i.e., quasi-bell-shaped profiles), and small increases towards rims,
forming near-rim troughs. The similarity of trace element and Mn profiles and dissolution in Section 4.7.4.3. Whereas this mechanism
suggests that the near-rim increases result from garnet dissolution and may apply to one or two annuli, some garnets exhibit numer-
backdiffusion of Yb and Y. Modified from Kohn MJ (2009) Models of ous compositional oscillations, and other explanations have
garnet differential geochronology. Geochimica et Cosmochimica Acta been offered to explain them. Such garnets sometimes exhibit
73: 170–182 and unpublished data of the author. sector zoning and chemical patterns that appear to mimic
Com Annulus
pa
tib
le
e lem
en
ts
Concentration
Original
profile
nts
leme
mpa tible e
Inco
Moat
Dissolved Original
rim rim Final rim
Growth
Dissolution
Core Regrowth
Distance
3200
Figure 23 Schematic plot of concentration versus distance for
compatible and incompatible elements in a mineral that is resorbed and
Consumption
2800
regrown. Solid line shows final profile. Dashed line shows original profile
for compatible element, prior to dissolution. If trace element zoning is
limit
2400
Petrographic rim
radially distributed, this process would lead to an annulus and moat in
compatible and incompatible elements, respectively. Reproduced from
Yang P and Rivers T (2002) The origin of Mn and Y annuli in garnet and 2000
the thermal dependence of P in garnet and Y in apatite in calc-pelite
and pelite, Gagnon terrane, western Labrador. Geological Materials ppm Y 1600
Research 4, http://gmr.minsocam.org/contents.html.
1200
matrix textures, as in one case from the central Nepal Himalayas 800
(Figure 26; Kohn, 2004), so all three are discussed in this section.
Oscillatory zoning in crystals is normally modeled in refer- 400
ence to reaction overstepping of igneous feldspars in an under- Core
cooled magma (L’Heureux and Fowler, 1996). Depletion of the 0
surrounding melt in feldspar components increases local 0.0 0.5 1.0
undercooling and crystal growth rates. The accelerating crystal- Distance (mm)
lization front eventually overrides the local depletion zone,
Figure 24 X-ray map of Y-zoning in a staurolite-grade garnet,
restoring contact with undepleted melt and reinitiating the
illustrating a Y annulus, possibly caused by prograde dissolution of
process. Although garnet-forming reactions are also over- garnet at the staurolite in isograd, followed by regrowth. Composition
stepped, heating, not cooling, drives garnet growth, and deple- versus distance plot illustrates trend expected conceptually from
tion of the surrounding matrix in garnet-compatible elements Figure 23. Reproduced from Pyle JM and Spear FS (1999) Yttrium
decreases overstepping. Thus the igneous model for oscillatory zoning in garnet: Coupling of major and accessory phases during
zoning cannot be applied to metamorphic garnet growth in a metamorphic reactions. Geological Materials Research 1, http://gmr.
heating environment (Kohn, 2004). minsocam.org/ContentsLink.html.
Instead, some regular oscillating process is commonly in-
voked. Possibilities include repeated oscillations in pressure,
temperature, or fluid infiltration (Garcia-Casco et al., 2002; Y, Zr, and Yb are shown). Garnet-compatible and incompatible
Shore and Fowler, 1996), or in rates of thrusting, causing elements are generally, albeit not ubiquitously, anticorrelated,
stepwise, rather than smooth, increases in P and T (Kohn, so high-concentration annuli in Y and Yb are commonly
2004). These processes result in regularly varying changes in matched by troughs in Sc and Zr (although Sc may be compat-
garnet growth rates. If the transport of an element in the matrix ible in garnet in some assemblages, e.g., Hickmott and Shimizu,
is kinetically limited, then phases of rapid growth will produce 1990; Hickmott and Spear, 1992; Hickmott et al., 1987). Simi-
local depletions in garnet-compatible elements; during slow larly, Ca and P are anticorrelated. The variations in Ca and P are
growth periods, the matrix homogenizes, raising local con- far subtler than for Y and Yb. At least a dozen oscillations are
centrations of garnet-compatible elements (Carlson, 2002; recorded from core to rim, and possibly as many as twenty.
Chernoff and Carlson, 1999). The reverse behavior occurs for Compositional mimicry of matrix textures is also inter-
garnet-incompatible elements – they are locally enriched dur- preted in terms of kinetically limited transport in the matrix,
ing rapid growth and depleted during slow growth. For the where it is assumed that transport is so slow that the garnet
Himalayan garnet (Figure 26), obvious multiple oscillations simply inherits a local grain’s trace element composition. For
occur in Ca and Y. A trace element zoning profile collected in example, yttrium zoning patterns in garnet can mimic the
2006 (Figure 26(d)) shows large oscillations in all trace ele- texture of a micaceous matrix (Kohn, 2004; Martin, 2009;
ments having concentrations above 0.1 ppm (here only P, Sc, Yang and Rivers, 2001); these garnets presumably overgrew
Ca P 4.7.5 Radiogenic Isotopes (Age Variability)
Radiogenic isotopes are commonly measured in metamorphic

minerals to obtain mineral isochron ages. However, because of
sample size requirements, most isotopic measurements are
made on bulk mineral separates, and core versus rim isotopic
variability is rarely investigated. Nonetheless, some inferences
regarding isotopic trends are obtainable for measured concen-
tration variations in parent isotopes, or from direct measure-
ment of two or more isotopic compositions from large crystals
Y Yb that exhibit systematic zoning in other elements.
4.7.5.1 Growth Zoning

Because most radiogenic isotopes and their parent isotopes are
trace elements with large ionic radii, charge balance and phys-
ical arguments imply that they should have slow diffusivities.
Some of the most extensively studied systems include Rb–Sr,
Sm–Nd, and Lu–Hf in garnet and U–Th–Pb in monazite.
Experimental studies indeed indicate slow diffusion rates,
Figure 25 X-ray maps of trace elements and Ca from garnet-bearing
and hence high potential to retain original growth composi-
quartzite, showing spatially coincident, pronounced spikes. The location
of these spikes is systematically different in different-sized garnets, tions (Cherniak et al., 2000; Coghlan, 1990; Ganguly et al.,
relative to Fe–Mg–Mn systematics (Chernoff and Carlson, 1997, 1999). 1998b; Smith and Giletti, 1997; Tirone et al., 2005; Van
Diffusionally produced haloes around growing crystals can cause sudden Orman et al., 2002). Growth zoning in parent isotopes is some-
stabilization or destabilization of minerals (Johnson and Carlson, 1990). times obvious and highly resolvable (e.g., Th in monazite,
The coincidence of spikes in many trace elements as well as Ca is Figure 2; Lu in garnet, Figure 22), and relatively detailed
interpreted to reflect modal changes in a mineral such as apatite or theoretical models have been developed for garnets (Kohn,
allanite. A common diffusional mechanism for intergranular diffusion 2009; Lapen et al., 2003; Skora et al., 2006). Yet, low concen-
of trace elements and Ca may also be required, possibly by trations of daughter isotopes require relatively large amounts of
complexing with a common, slow-diffusing ‘carrier’ element.
material for analyses, resulting in coarse spatial resolution
Reproduced from Chernoff CB and Carlson WD (1999) Trace element
of growth profiles and crystal growth durations; pioneering
zoning as a record of chemical disequilibrium during garnet growth.
Geology 27: 555–558. work using the Rb–Sr system (Christensen et al., 1989, 1994)
provided only a few analyses between core and rim.
Neodymium isotopes in garnet also allow investigation of
the rates and duration of porphyroblast growth. Garnet
grains with high and low Y contents too rapidly to allow strongly prefers Sm to Nd, leading to 147Sm/144Nd ratios in
homogenization of matrix compositions. In the Nepal garnet excess of 3.0 in some crystals, or age resolution as low as several
(Figure 26), the yttrium map clearly shows mica mimicry in hundred thousand years ago. Slow REE diffusion rates allow
the core and in one high-Y band. garnet to preserve isotope compositions at high temperature
Sector zoning has long been observed in metasomatic and (Burton et al., 1995; Coghlan, 1990; Ganguly et al., 1998b;
skarn garnets (Akizuki, 1984), but is rare in regional metamor- Tirone et al., 2005; van Orman et al., 2002). Recent work by
phic garnets. Crystallographically distinct faces on a crystal are Pollington and Baxter (2010) illustrates well the power of
expected to have different compositions because partitioning radiometric zoning analysis. A shear zone at Stillup Tal,
depends on which planes of atoms are exposed. For example, Austrian Alps, hosts large garnets that grew during hydrothermal
the {112} crystal face in garnets evidently concentrates Ca and alteration of a granodiorite (Selverstone et al., 1991). Pollington
Y relative to the more common {011} face (Figure 26). Pres- and Baxter analyzed concentric rings from a large 6-cm-
ervation of these differences as sector zoning, however, re- diameter garnet from the center of the shear zone, as well as
quires that growth of crystal faces is rapid compared to matrix compositions. Each garnet analysis provides an age
intracrystalline diffusion (Watson, 2004; Watson and Liang, (Figure 27), and the distribution of ages allows calculation of
1995) and can be expressed in terms of the growth Peclet the rate and duration of garnet growth and evaluation of garnet
number, equal to Vl/D, where V is the face advance velocity, l growth mechanisms (Figure 28). Such results are important for
is the half-length of the growth boundary layer, and D is several reasons. First, garnets from this outcrop are known to
diffusivity. Sector zoning can occur when Vl/D>1. Typically, l have grown during 16 km of exhumation, so the 7.5 0.5 Ma
is taken as half the unit cell length, or 5 Å for garnet. For the duration of growth allows calculation of exhumation rates:
Nepal garnet, V was estimated at 4 mm per Ma from regional 2.1 0.15 mm year1. Second, these results indicate that single
petrologic and chronologic data, and l was assumed to be 5 Å. shear zones may be active over many millions of years. Third,
Adopting these values yields a D for Ca of 1021 cm2 s1 at these data constrain processes of garnet growth, showing broad
550 C, which compares well with experimental data (Ganguly consistency with a linear increase in volume with time, but
et al., 1998a; Vielzeuf et al., 2007). Alternatively, one could with important pulses of garnet growth at 26.8 Ma (garnet
assume a value for D and calculate V. core) and 24.6 Ma (near the garnet profile midpoint).
Calcium Yttrium Manganese

2}
{1 {11
{011} 12 {011}
}
{01
1}
?
{112}
1 mm
(a) (b) (c)
100 000
10 000
Y Ca
Concentration (ppm)
1000
Yb P
100
10
Sc
1.0 Zr
0.1
0 0.3 0.6 0.9 1.2 1.5 1.8 2.1 2.4 2.7
(d) Distance (mm)
Figure 26 Compositions of lower amphibolite-facies metapelitic garnet from the Nepal Himalayas; white line is location of trace element traverse.
(a, b) X-ray maps of Ca and Y show compositional oscillations, sector zoning (arrows), and compositional mimicry of matrix mica textures. (c) X-ray
map of Mn shows generally smooth zoning and single broad oscillation towards rim. Reproduced from Kohn MJ (2004) Oscillatory- and sector-zoned
garnets record cyclic (?) rapid thrusting in central Nepal. Geochemistry, Geophysics, Geosystems 5, Q12014, http://dx.doi.org/10.1029/2004GC000737.
(d) Composition versus distance showing multiple compositional oscillations, large variations in trace element compositions, and generally inverse
behavior of garnet-compatible (Y, Yb) versus incompatible (Sc, Zr) elements (unpublished data of the author).
4.7.5.2 Diffusion
of Lu produces a bell-shaped profile, high-temperature relaxa-
The principles of diffusional zoning in radiogenic isotopes tion of growth profiles will drive Lu/Hf ratios towards lower
follow many of those for major elements and stable isotopes. values in the core of the garnet. Conversely, on the rim of the
Just as for stable isotopes, diffusion profiles in radiogenic iso- garnet, dissolution will increase Lu, and backdiffusion towards
topes are measured with greater difficulty than growth zoning, the garnet core will drive Lu/Hf ratios towards higher values
and, instead, bulk analysis of different coexisting minerals has (Figures 22 and 29). Assuming that cores versus rims can
been used to assess the magnitude of diffusional reequilibra- somehow be separated and analyzed independently, both pro-
tion (Giletti and Casserly, 1994; Jenkin, 1997; Jenkin, et al., cesses will rotate Lu–Hf isochrons towards steeper slopes and
1995, 2001). Similarly, core versus rim isotopic compositions older apparent ages.
of garnet coronas adjacent to orthopyroxene and plagioclase Because of sample size restrictions, studies of garnets have
have been used to infer closure temperatures for the Sm–Nd typical spatial resolutions of millimeters, whereas high U and
and Rb–Sr systems relative to U–Pb (Burton et al., 1995). Th contents potentially allow fine-scale resolution of diffusion
The effects of diffusional reequilibration on garnet iso- profiles in accessory minerals. Indeed, Grove and Harrison
chrons was recently modeled for the Lu–Hf system (Kohn, (1999) showed that diffusional zoning of 208Pb in monazite
2009; Figure 29), which is particularly amenable to theoretical could be measured with high spatial and analytical resolution
treatment because of zoning similarities between Lu and Mn, via ion microprobe. Because monazite has such a high concen-
and relatively well-understood Mn systematics in garnets. tration of Th, sufficient 208Pb is produced for relatively straight-
Assuming that Lu3þ diffuses significantly faster than Hf4þ forward measurement (Harrison et al., 1995). Grove and
(Ganguly, 2010; Ganguly et al., 2010) and that growth zoning Harrison (1999) analyzed the cores of crystals using sectioned,
0.5131 28
0.5129 26
143Nd/144Nd
Age (Ma)
0.5127 24
22
0.5125
Measured garnets
Measured matrix 20 (a)
0.5123
0 1 2 3
50 Core Rim
Radius (cm)
40
30 28
Δζ(105)
Rapid growth
20 26
Age (Ma)
Rapid growth
10 24
0 22
–10
0 1 2 3 4 5 20 (b)
147Sm/144Nd
0 20 40 60 80
Core Volume (cm3) Rim
Figure 27 Isochron diagram for different concentric zones in a single
garnet from Stillup Tal, Austria, showing consistent decrease in slope Figure 28 Ages of different zones within a garnet plotted as a function
(age) from core to rim. Top panel shows conventional 143Nd/144Nd of distance. (a) Age versus radius showing inconsistency with linear
versus 147Sm/144Nd isochron diagram. Bottom panel shows differences growth. (b) Age versus volume, showing general consistency with linear
in isotope ratios relative to rim composition, emphasizing the spread of growth, but with important rapid growth episodes, as indicated by
ages. Reproduced from Pollington AD and Baxter EF (2010) High minimal changes in age, near garnet core and middle. Reproduced from
resolution Sm–Nd garnet geochronology reveals the uneven pace of Pollington AD and Baxter EF (2010) High resolution Sm–Nd garnet
tectonometamorphic processes. Earth and Planetary Science Letters geochronology reveals the uneven pace of tectonometamorphic
293: 63–71. processes. Earth and Planetary Science Letters 293: 63–71.
polished grains and the outer few micrometers of other crys- Core Unmodified
tals, using depth profiling. In depth profiling, grains with good [Lu] Relaxed
crystal faces are pressed into a soft mounting medium. The
176Lu
primary ion beam is used to sputter a crater into the face of a
177Hf
crystal, and isotopic compositions are collected continuously.
Rim
These compositions are periodically averaged and, knowing (a)
the sputter rate (from the depth of the crater and total sputter
time), the depth corresponding to each of these compositions
Δ176Hf*/177Hf
can be determined. Depth profiling into the face of a crystal

from the Greater Himalayan Sequence, Nepal, yielded a mono-
tonically increasing apparent age (Figure 30(a)), which Grove
and Harrison inverted to obtain a temperature–time history
(Figure 30(b)). The retrieved cooling path is generally compat-
ible with other chronologic data and reveals details of the cool- (b)
ing history otherwise difficult to resolve.
176
In principle, diffusion zoning profiles for radiogenic iso- Lu/177Hf
topes are inherently superior to major element profiles at
Figure 29 Theoretical shift of Lu–Hf ages resulting from differential
resolving cooling histories. Each point on a composition pro-
diffusion of Lu relative to Hf. (a) Concentration profile across
file fundamentally reflects temperature for major elements via garnet showing that core Lu decreases, but dissolution causes rim Lu to
cation exchange thermometry with matrix minerals versus time increase. (b) Garnet compositions shift towards steep slope and
for radiogenic isotopes via the radioactive decay equation. The increased apparent age and apparent initial Hf composition decreases.
distance of that point from the rim defines time for major Reproduced from Kohn MJ (2009) Models of garnet differential
elements versus temperature for radiogenic elements (specifi- geochronology. Geochimica et Cosmochimica Acta 73: 170–182.
cally Tc(x), eqn [19]). Retrieval of time from zoning profiles
depends directly on temperature, cooling rate, and diffusion
parameters, whereas Tc(x) depends on their logarithm. Thus
cooling rates inferred from major elements are much more Accessory minerals commonly contain high concentrations of
uncertain than those inferred from radiogenic isotopes. radioactive elements and are a common target of radiogenic
isotope measurements. Specific examples include U in zircon, specifically P (apatite), REE (allanite, monazite, xenotime), Zr
apatite, titanite, monazite, xenotime, and allanite; and Th in (zircon), and Ti (titanite). Trace elements also occur in the
monazite and allanite. Each of these minerals is stabilized in major minerals (particularly P, Zr, and Ti), so accessory min-
a rock via a single element or suite of related elements, erals participate directly in major mineral reactions (Corrie and
Kohn, 2008; Ferry, 2000; Kelsey and Powell, 2011; Kelsey et al.,
2008; Kohn and Malloy, 2004; Pyle and Spear, 1999, 2000,
15 800
Bulk
2003; Pyle et al., 2001; Spear, 2010; Spear and Pyle, 2002,
Core = 12.6 ± 0.7 Ma monazite 2010; Wing et al., 2003). However, the slow diffusion of
Th–Pb age (Ma)
Temperature (˚C)
10 600 many constitutive elements in accessory minerals, as verified
experimentally (Cherniak, 2010), means that dissolution–
reprecipitation must be a common mechanism for changing
5 400 compositions and that different zones within single grains
Monazite should have different elemental and isotopic compositions.
depth profile
Biotite Monazite ubiquitously exhibits this type of behavior.
0 200
0 0.5 1.0 1.5 0 3 6 9 12 15 Backscattered electron images and Y, Th, and U X-ray maps
(a) Distance (mm) (b) Time (Ma)
nearly always reveal complex zoning (e.g., DeWolf et al., 1993;
Figure 30 (a) Age variation versus depth for an ion microprobe depth Kohn et al., 2004, 2005; Parrish, 1990; Pyle and Spear, 2000,
profile collected from the face of a Himalayan monazite, showing 2003; Pyle et al., 2001, 2005; Townsend et al., 2000; Williams
apparent diffusive loss of 208Pb towards the rim, and consequent and Jercinovic, 2002; Williams et al., 1999; Zhu and O’Nions,
increase in age towards the core. Solid curve is best fit to the data. 1999; Zhu et al., 1997; Figures 31 and 32), and, most impor-
Dashed lines show implied compositional trends for bounds on cooling tantly, several studies now demonstrate significant age differ-
history. Vertical bars are 2s analytical errors. Distance is measured
ences among these chemically distinct domains (Figures 31
from the surface. (b) Inferred temperature–time history, based on
and 32). Extreme compositional and age heterogeneity implies
composition profile in (a). Solid line is best fit to data, dashed lines are
bounds. White boxes are temperature–time points based on peak that analysis of a bulk mineral separate or even a single grain is
metamorphic conditions and 40Ar/39Ar cooling age from biotite. Size of not very useful geochronologically, because zones of different
boxes reflects age and temperature uncertainties. Reproduced from origins and ages are then averaged. By analogy, geochronolo-
Grove M and Harrison TM (1999) Monazite Th–Pb age depth profiling. gists have long since recognized the domainal behavior of
Geology 27: 487–490. zircon and its ages, and we cannot now imagine analyzing
(a) Th Mα (b) U Mβ
50 mm
(c) Age (d)

1200
1000
800
Age (Ma)
600
400
1.05 Ga
200
400 Ma
0 2 4 6 8 10 12 14 16 18
Analysis number
Figure 31 (a, b) Th and U X-ray maps, (c) age map, and (d) age histogram for monazite grain from the Western Gneiss Region, Norway, as measured
via the electron microprobe, showing correlation of ages (and implied isotopic composition) with some chemically delineated zones within grains.
Reproduced from Williams ML, Jercinovic MJ, and Terry MP (1999) Age mapping and dating of monazite on the electron microprobe: Deconvoluting
multistage tectonic histories. Geology 27: 1023–1026.
60 Ma
56 Ma
21 Ma
48 Ma
43 Ma
61 Ma
59 Ma
54 Ma 50 mm 50 mm
Figure 32 Backscattered electron image and ages for complexly zoned monazite crystals from contact metamorphosed Ireteba granite, Nevada.
Crystallization age of the pluton is 65 Ma. Subsequent infiltration of fluids perhaps as young as 16 Ma apparently dissolved and reprecipitated new
monazite in an extremely complex fashion. Th–Pb ages were determined via ion microprobe, with typical 2s uncertainties of a few percent.
Reproduced from Townsend KJ, Miller CF, D’Andrea JL, Ayers JC, Harrison TM, and Coath CD (2000) Low temperature replacement of monazite in the
Ireteba Granite, southern Nevada: Geochronological implications. Chemical Geology 172: 95–112.
bulk zircon separates, or even single whole grains, without first reached sufficiently high temperature to homogenize Mn, Fe,
characterizing their internal chemistry via backscattered elec- and Mg by diffusion (Spear et al., 1990a, 1995). The existence
tron imaging or cathodoluminescence. Conversely, there is a of Grt1 is suggested by patchy zoning in Ca in garnet cores
direct link between major mineral reactions and accessory (Figure 33), by phase equilibrium arguments (Spear et al.,
mineral abundance and chemistry (Corrie and Kohn, 2008; 1995), and by trace element compositions in some relict
Ferry, 2000; Kelsey and Powell, 2011; Kelsey et al., 2008; Kohn cores (Pyle and Spear, 1999). Grt3 grew during partial melting,
and Malloy, 2004; Pyle and Spear, 1999, 2000; Pyle et al., which is indicated both macroscopically in migmatitic textures
2005; Rubatto, 2002; Spear, 2010; Spear and Pyle, 2010; and by the occurrence of trace element discontinuities towards
Wing et al., 2003), so different zones within an accessory the rims of some garnets (Figure 33). Most importantly, the
mineral grain can potentially be linked to different reactions. abrupt increase in Cr (Figure 33) is diagnostic of a mica
The most fruitful research will be to link chemical zones in breakdown reaction, specifically the muscovite þ plagioclase þ
accessory minerals with metamorphic reactions, determine quartz dehydration–melting reaction. Although this reaction
ages for those zones, and deduce the timing of mineral does not produce or consume garnet, it increases Cr contents
reactions and hence overall mineralogical evolution of meta- of matrix phases, and further heating produces garnet with
morphic rocks (Harrison et al., 2002; Spear and Pyle, 2002). increasing Cr contents (Figure 33) via continuous biotite
dehydration–melting (Kohn et al., 1997; Spear and Kohn,
1996; Spear et al., 1999). The occurrence of these melting
4.7.6 Case Study: Fall Mountain, New Hampshire reactions after earlier low-P metamorphism, rather than dehy-
dration reactions that produce K-feldspar, implies an increase
This section describes rocks from the Fall Mountain nappe, in pressure. Because garnet cores are unzoned in d18O
southwestern New Hampshire, to illustrate how different (Figure 35), this loading and garnet core growth apparently
approaches described above can be used to unravel the meta- occurred nearly isothermally.
morphic and tectonic history of a rock. Garnets in these rocks On the retrograde path, cooling and melt crystallization
have been analyzed for intracrystalline zoning in major ele- produced biotite at the expense of garnet. This ReNTR con-
ments, trace elements, and stable isotopes (Kohn et al., 1997; sumed most Grt3, causing an increase in Mn concentration at
Pyle and Spear, 1999, 2000; Spear and Kohn, 1996; Spear et al., the rims of garnets that backdiffused towards garnet cores.
1990a). Each type of zoning reveals important details about At the muscovite dehydration–melting reaction, muscovite
thermal evolution, tectonism (loading), and mineral reactions. again crystallized (the melt fully crystallized) and, in the result-
Major and trace elements show evidence for at least four and ing solid-state assemblage, garnet started to grow again with
possibly five generations of garnet growth, whereas oxygen cooling, to produce Grt4. Garnet growth was accompanied by a
isotope profiles monitor fluid infiltration or open system be- decrease in Mn content towards the rim, and differential ther-
havior (Figures 33–35). All generations of garnets, the impor- modynamic modeling of this zoning indicates growth during
tant reactions that produce or consume garnet, and oxygen isobaric cooling (Spear et al., 1990a). The combination of
isotope trends are summarized in Figure 36. Grt3 consumption followed by Grt4 growth resulted in the
Garnet cores (Grt1 þ Grt2) are believed to have grown at low pronounced Mn ‘hump’ (Kohn et al., 1997; Spear et al.,
pressure (andalusite-field), but flat profiles in Mn, Fe, and Mg 1990a), just as annuli in trace elements can result from disso-
(although not Ca; Figures 33 and 34) indicate that rocks lution followed by regrowth (Figure 23). The decrease in d18O
XSps FM
XGrs
0.94
0.08 0.07
0.84
0.03
0.04
Cr Sc
1700
500
100 200
Ca Mn
Figure 33 X-ray maps of Mn, Ca, Fe/(Fe þ Mg) (‘FM’), Cr, and Sc in garnets from Fall Mountain, New Hampshire. White box indicates area of major
element maps (upper panels). Different generations of growth can be linked to prograde reaction history to understand mineralogical changes attending
metamorphism and anatexis, and overall P–T evolution. Black line in lower Mn panel shows location of composition traverse plotted in Figure 34.
Reproduced from Kohn MJ, Spear FS, and Valley JW (1997) Dehydration melting and fluid recycling during metamorphism: Rangeley Formation,
New Hampshire, USA. Journal of Petrology 38: 1255–1277.
towards garnet rims is simply the result of garnet growth dur- mineralogical, and tectonic evolution than any one zoning
ing cooling (Figure 35). High-Ca garnet overgrowths (Grt5) are method. Two particularly interesting results of this work are
believed to reflect late-stage fluid infiltration, as emplacement the following: (1) Fluids were ‘recycled’ during high-grade
of the nappe caused dewatering of structurally lower metape- metamorphism from hydrous minerals to dispersed melt
lites. The d18O compositions of quartz, muscovite, and biotite pockets during heating and then back to hydrous minerals
in different rocks provide additional evidence for low-T fluid again during cooling (Kohn et al., 1997). (2) Accessory mineral
infiltration (Kohn et al., 1997). abundance and trace element compositions of major silicates
One interesting feature of this work is that no one zoning are very strongly influenced by melting reactions. This is per-
study could possibly have resolved all of the details of the haps not surprising given the high phosphorus solubilities in
mineralogical and P–T evolution. Specifically, (a) major ele- most felsic melts.
ments indicate general P–T conditions, high temperatures
(compositional homogenization), retrograde resorption fol-
lowed by growth during isobaric cooling, and late-stage 4.7.7 Discussion and Conclusions
overgrowths; (b) trace elements indicate the importance of
dehydration–melting reactions; and (c) stable isotopes indi- It is perhaps self-evident from this review that the most pow-
cate an isothermal pressure increase, cooling during Grt4 erful application of geochemical zoning involves combining
growth, and a general absence of open system behavior theoretical models (both forward and inverse) with detailed
throughout most of the rocks’ history. Clearly, a combination geochemical measurements. Theory readily accommodates
of zoning studies provides a much better picture of the P–T, only end-member processes, however, and recent research
0.5 1.0 10
Grt1+2
St-out
Grt5
Grt4
St-in
11.9‰
12.3‰
12.7‰
0.4 Fe/(Fe+Mg)
0.9 8
X
XPrp, XSps, XGrs
Alm,
fs+ tz
lt
me
Q
Als s+Pl+
0.3 0.8
Fe/(Fe+Mg)
6
+K
P (kbar)
M
Alm Grt4
Mn-trend results from
0.2 Grt3 consumption
0.7 Grt5
Sps
4 I1 Grt3
Grt2
I2
H2
0.1 Prp 0.6
O-
sa
z
Grt1 (?)
Qt
tur
Grs 2
s+
tz
a
+Q O
+M
te
Pl H 2 pe
0.5
d
0.0 O s+ fs+
hl
C 2 M K lite
t+ t+H Als+ sol
0.00 0.30 0.60 0.90 1.20 1.50 Gr B
s+
idus
Al
Rim Rim 0
Distance (mm) 400 500 600 700 800
Figure 34 Electron microprobe compositional traverse across garnet in T (⬚C)
Figure 33, showing relatively flat core, pronounced Mn ‘humps,’ and Figure 36 P–T diagram showing important reactions, isopleths of
high-Ca rims. Vertical lines separate different garnet generations, as oxygen isotope compositions (thin vertical lines), and inferred P–T path,
inferred from compositional differences. Third-generation garnet is not as deduced from the major element, trace element, and oxygen isotope
present, but is evident from Cr-zoning in other garnets (Figure 33). zoning in garnet porphyroblasts. The parallelogram at the end of the P–T
Reproduced from Kohn MJ, Spear FS, and Valley JW (1997) Dehydration path is the thermobarometrically derived estimate of final P–T conditions.
melting and fluid recycling during metamorphism: Rangeley Formation, Stability fields of the aluminosilicate polymorphs shown for reference as
New Hampshire, USA. Journal of Petrology 38: 1255–1277. unlabeled straight lines. I1 and I2 are invariant points relevant to the
reaction history of the rock. Reproduced from Kohn MJ, Spear FS, and
Valley JW (1997) Dehydration melting and fluid recycling during
metamorphism: Rangeley Formation, New Hampshire, USA. Journal of
= ± 1σ 600–650 ⬚C, 4–5 kbar
12.5 Petrology 38: 1255–1277.
d 18O (‰, V-SMOW)
12.0
contact metamorphic rocks (Lasaga, 1986, 1989). Thus, some
Qtz
elements may equilibrate on the scale of a thin section (e.g., Fe,
11.5 Grt Mg, Mn in high-grade rocks), whereas others may not (e.g.,
REEs). Deciphering these effects from the geochemical record
first requires understanding the basic phase equilibria of a rock
475 ⬚C, Grt
Grt
Qtz
11.0
and how elements partition and repartition in an equilibrium
4–5 kbar Qtz
0 mm 5
sense as mineral abundances and species change. Equilibrium
0.0 0.5 1.0 1.5 2.0 2.5 processes do explain some chemical behavior quite well, and
Rim Core inversion of composition profiles can yield information on
Distance (mm) P–T evolution, cooling rates, etc. Conversely, trace elements
Figure 35 Oxygen isotope profile across garnet porphyroblasts show local concentration heterogeneities, indicating poor
showing homogeneous core, and pronounced decrease in d18O towards equilibration, at least over length scales of millimeters. In
rims. The homogeneous core implies growth at relatively constant isotope studies, equilibrium partitioning is difficult to verify,
temperature, whereas decreasing d18O is interpreted to result from and is commonly simply assumed. Any behavior outside of
growth during cooling. Dashed line is theoretical prediction assuming an equilibrium models is then assigned to open system effects.
absence of infiltration by an isotopically distinct fluid. Inset shows a The fastest growing areas in geochemical zoning analysis
sketch of the garnets analyzed. Horizontal lines are saw cuts used to
are in trace and radiogenic elements, in part driven by a pro-
subsample garnets. Short vertical lines show size of each subsample.
pensity of minerals to retain growth zoning in trace elements,
Reproduced from Kohn MJ, Spear FS, and Valley JW (1997) Dehydration
melting and fluid recycling during metamorphism: Rangeley Formation, but also because of geochronologic interests. One highlight of
New Hampshire, USA. Journal of Petrology 38: 1255–1277. recent work is the realization that all minerals, not just acces-
sories, control trace element behavior. Accessory mineral com-
positions (and therefore preserved ages) are undoubtedly yet
more clearly reveals that minerals do not exhibit end-member complexly related to major mineral reactions. The advent of
equilibrium or disequilibrium behavior with respect to ele- geochronologic microanalysis of inclusions and zones in
ment partitioning and geochemical zoning. It has long been grains has inspired a new generation (at least) of petrologic
recognized that minerals will not grow without overstepping research. However, assigning chronologic significance to an
a reaction (e.g., Lasaga, 1986, 1989). The degree to which a accessory mineral grain or even a single zone within it requires
reaction is overstepped and compositions are equilibrated identifying the mineral reaction(s) responsible for its growth.
within a rock volume clearly depends on many variables, and In metamorphic rocks whose petrologic and reaction histories
reaction overstepping can be quite small for some elements in are unknown, geochronologic interpretation is perilous or, in
regional metamorphic rocks, or quite profound, especially for some cases, pointless.
Several areas deserve additional study, especially in kinetics Carlson WD (1991) Competitive diffusion-controlled growth of porphyroblasts.
and in the phase equilibria of accessory minerals. Whereas Mineralogical Magazine 55: 317–330.
Carlson WD (2002) Scales of disequilibrium and rates of equilibration during
some minerals and elements likely equilibrate rapidly because
metamorphism. American Mineralogist 87: 185–204.
of fast inter- and intragranular diffusion rates (e.g., oxygen), for Carlson WD (2006) Rates of Fe, Mg, Mn, and Ca diffusion in garnet. American
other elements there is simply little information (e.g., Sr, Nd, Mineralogist 91: 1–11.
etc.). If different elements do not equilibrate on the scale of a Carlson WD (2010) Diffusion of REEs and other trivalent cations in garnet: New data
thin section, then different regions could have different isoto- on rates and mechanism. EOS Transactions of the American Geophysical Union
V44A-03.
pic compositions. Additionally, there is no systematic informa- Carlson WD, Denison C, Ketcham RA, and Boyle AP (1995) Controls on the nucleation
tion on the effects of water fugacity on cation diffusion rates. and growth of porphyroblasts: Kinetics from natural textures and numerical models.
There could be major differences in diffusive behavior during Geological Journal 30: 207–225.
heating, when water is being liberated and fH2O is high, versus Carlson W and Schwarze E (1997) Petrological significance of prograde
homogenization of growth zoning in garnet: An example from the Llano Uplift.
during cooling, when water is being consumed and fH2O
Journal of Metamorphic Geology 15: 631–644.
rapidly decreases. The links among major and trace mineral Chacko T, Cole DR, and Horita J (2001) Equilibrium oxygen, hydrogen and carbon
reactions, modal abundances, and compositions are also par- isotope fractionation factors applicable to geologic systems. Reviews in Mineralogy
amount for understanding metamorphic geochemistry. There 43: 1–81.
are several examples of chemical zoning, such as annuli, which Chakraborty S (2008) Diffusion in solid silicates: A tool to track timescales of processes
comes of age. Annual Review of Earth and Planetary Sciences 36: 153–190.
are extremely commonly observed in trace elements, but ex- Chakraborty S and Ganguly J (1991) Compositional zoning and cation diffusion in
tremely rare in major elements. Some annuli can be explained garnets. In: Ganguly J (ed.) Diffusion, Atomic Ordering, and Mass Transport:
by equilibrium and mass balance processes, and some cannot. Selected Topics in Geochemistry, pp. 120–175. New York: Springer.
A significant effort is required to explore the systematics of Chakraborty S and Ganguly J (1992) Cation diffusion in aluminosilicate garnets:
Experimental determination in spessartine–almandine diffusion couples, evaluation
accessory mineral chemistry and its relation to major mineral
of effective binary, diffusion coefficients, and applications. Contributions to
reactions and abundances. Mineralogy and Petrology 111: 74–86.
Finally, new technologies commonly yield new insights. Chakraborty S and Rubie DC (1996) Mg tracer diffusion in aluminosilicate garnets at
The increasing availability of LA-ICP-MS and ion microprobes, 750–850 C, 1 atm. and 1300 C, 8.5 GPa. Contributions to Mineralogy and
as well as more general application of electron microprobes for Petrology 122: 406–414.
Chamberlain CP and Conrad ME (1991) Oxygen isotope zoning in garnet. Science
mapping and analyzing minor and trace elements, means that 254: 403–406.
petrologists have a substantially larger suite of elements and Chamberlain CP and Conrad ME (1993) Oxygen isotope zoning in garnet: A record of
isotopes at their disposal than ever before. Zoning patterns in volatile transport. Geochimica et Cosmochimica Acta 57: 2613–2629.
trace elements and isotopes will undoubtedly reveal insights Cherniak DJ (2010) Diffusion in accessory minerals: Zircon, titanite, apatite, monazite
and xenotime. Reviews in Mineralogy and Geochemistry 72: 827–869.
into growth processes, diffusion, kinetics, etc. Nonetheless,
Cherniak DJ, Watson EB, Harrison TM, and Grove M (2000) Pb diffusion in monazite:
interpretation of new data will require a full understanding of A progress report on a combined RBS/SIMS study. EOS Transactions of the
the many processes, equilibrium or otherwise, that lead to American Geophysical Union 81: S25.
geochemical zoning in metamorphic minerals. Chernoff CB and Carlson WD (1997) Disequilibrium for Ca during growth of pelitic
garnet. Journal of Metamorphic Geology 15: 421–438.
Chernoff CB and Carlson WD (1999) Trace element zoning as a record of chemical
Acknowledgment disequilibrium during garnet growth. Geology 27: 555–558.
Christensen JN, Rosenfeld JL, and DePaolo DJ (1989) Rates of tectonometamorphic
processes from rubidium and strontium isotopes in garnet. Science
Special thanks are due to Frank Spear, Mark Harrison, and 244: 1465–1469.
Bruce Watson for many years’ patient instruction in petrology Christensen JN, Selverstone J, Rosenfeld JL, and DePaolo DJ (1994) Correlation of
and diffusion. Frank Spear, Bill Carlson, Dave Kelsey, Daniel Rb–Sr geochronology of garnet growth histories from different structural levels
within the Tauern Window, Eastern Alps. Contributions to Mineralogy and Petrology
Vielzeuf, Laure Martin, Anthony Pollington, and Ethan Baxter
118: 1–12.
graciously provided original photographs and figures; Coghlan RAN (1990) Studies in Diffusional Transport: Grain Boundary Transport of
R. Rudnick and B. Wing provided incisive reviews. This work Oxygen in Feldspars, Diffusion of Oxygen, Strontium, and the REEs in Garnet, and
was funded by NSF grants EAR 0073803 and EAR 0810242. Thermal Histories of Granitic Intrusions in South-Central Maine Using Oxygen
Isotopes. PhD Thesis, Brown University.
Cole DR and Chakraborty S (2001) Rates and mechanisms of isotopic exchange.
Reviews in Mineralogy 43: 83–223.
References Corrie SL and Kohn MJ (2008) Trace element distributions in silicates during prograde
metamorphic reactions: Implications for monazite formation. Journal of
Akizuki M (1984) Origin of optical variations in grossular-andradite garnet. American Metamorphic Geology 26: 451–464.
Mineralogist 69: 328–338. Daniel CG and Spear FS (1998) Three-dimensional patterns of garnet nucleation and
Anderson DE and Olimpio JC (1977) Progressive homogenization of metamorphic growth. Geology 26: 503–506.
garnets, South Morar, Scotland: Evidence for volume diffusion. The Canadian Dempster TJ (1985) Garnet zoning and metamorphism of the Barrovian type area,
Mineralogist 15: 205–216. Scotland. Contributions to Mineralogy and Petrology 89: 30–38.
Bea F, Montero P, Stroh A, and Baasner J (1996) Microanalysis of minerals by an Denison C and Carlson WD (1997) Three-dimensional quantitative textural analysis of
Excimer UV-LA-ICP-MS system. Chemical Geology 133: 145–156. metamorphic rocks using high-resolution computed X-ray tomography: Part II.
Berman RG (1988) Internally-consistent thermodynamic data for minerals in the system Application to natural samples. Journal of Metamorphic Geology 15: 45–57.
Na2O–K2O–CaO–MgO–FeO–Fe2O3–Al2O3–SiO2–TiO2–H2O–CO2. Journal of Denison C, Carlson WD, and Ketcham RA (1997) Three-dimensional quantitative
Petrology 29: 445–522. textural analysis of metamorphic rocks using high-resolution computed X-ray
Burton KW, Kohn MJ, Cohen AS, and O’Nions RK (1995) The relative diffusion of Pb, tomography: Part I. Methods and techniques. Journal of Metamorphic Geology
Nd, Sr and O in garnet. Earth and Planetary Science Letters 133: 199–211. 15: 29–44.
Carlson WD (1989) The significance of intergranular diffusion to the mechanism and DeWolf CP, Belshaw NS, and O’Nions RK (1993) A metamorphic history from
kinetics of porphyroblast crystallization. Contributions to Mineralogy and Petrology micron-scale 207Pb/206Pb chronometry of Archean monazite. Earth and Planetary
103: 1–24. Science Letters 120: 207–220.
Dodson MH (1973) Closure temperature in cooling geochronological and petrological Grove M and Harrison TM (1999) Monazite Th–Pb age depth profiling. Geology
systems. Contributions to Mineralogy and Petrology 40: 259–274. 27: 487–490.
Dodson MH (1986) Closure profiles in cooling systems. Materials Science Forum Harrison TM, Catlos EJ, and Montel J-M (2002) U–Th–Pb dating of phosphate
7: 145–154. minerals. Reviews in Mineralogy 48: 523–558.
Doremus RH (1998) Diffusion of water and oxygen in quartz: Reaction–diffusion model. Harrison TM, McKeegan KD, and LeFort P (1995) Detection of inherited monazite in the
Earth and Planetary Science Letters 163: 43–51. Manaslu leucogranite by 208Pb/232Th ion microprobe dating: Crystallization age and
Doremus RH (1999) Diffusion of water in crystalline and glassy oxides: tectonic significance. Earth and Planetary Science Letters 133: 271–282.
Diffusion–reaction model. Journal of Materials Research 14: 3754–3758. Hervig RL and Peacock SM (1989) Implications of trace element zoning in deformed
Edwards KJ and Valley JW (1998) Oxygen isotope diffusion and zoning in diopside: quartz from the Santa Catalina mylonite zone. Journal of Geology 97: 343–350.
The importance of water fugacity during cooling. Geochimica et Cosmochimica Acta Hickmott DD and Shimizu N (1990) Trace element zoning in garnet from the Kwoiek
62: 2265–2277. Area, British Columbia: Disequilibrium partitioning during garnet growth?
Eiler JM, Baumgartner LP, and Valley JW (1992) Intercrystalline stable isotope Contributions to Mineralogy and Petrology 104: 619–630.
diffusion: A fast grain boundary model. Contributions to Mineralogy and Petrology Hickmott DD, Shimizu N, Spear FS, and Selverstone J (1987) Trace element zoning in a
112: 543–557. metamorphic garnet. Geology 15: 573–576.
Eiler JM, Valley JW, and Baumgartner LP (1993) A new look at stable isotope Hickmott DD, Sorensen SS, and Rogers PSZ (1992) Metasomatism in a subduction
thermometry. Geochimica et Cosmochimica Acta 57: 2571–2583. complex: Constraints from microanalysis of trace elements in minerals from garnet
Elphick SC and Graham CM (1988) The effect of hydrogen on oxygen diffusion in amphibolite from the Catalina Schist. Geology 20: 347–350.
quartz: Evidence for fast proton transients? Nature 335: 243–245. Hickmott DD and Spear FS (1992) Major- and trace-element zoning in garnets from
England PC and Richardson SW (1977) The influence of erosion upon the mineral calcareous pelites in the NW Shelburne Falls Quadrangle, Massachusetts: Garnet
facies of rocks from different metamorphic environments. Journal of the Geological growth histories in retrograded rocks. Journal of Petrology 33: 965–1005.
Society 134: 201–213. Hirsch D, Prior D, and Carlson W (2003) An overgrowth model to explain multiple,
England PC and Thompson AB (1984) Pressure–temperature–time paths of regional dispersed high-Mn regions in the cores of garnet porphyroblasts. American
metamorphism, Part I: Heat transfer during the evolution of regions of thickened Mineralogist 88: 131–141.
continental crust. Journal of Petrology 25: 894–928. Hoefs J (1997) Stable Isotope Geochemistry. New York: Springer.
Evans K, Powell R, and Holland T (2010) Internally consistent data for sulphur-bearing Holdaway MJ (1971) Stability of andalusite and the aluminum silicate phase diagram.
phases and application to the construction of pseudosections for mafic greenschist American Journal of Science 271: 97–131.
facies rocks in Na2O–CaO–K2O–FeO–MgO–Al2O3–SiO2–CO2–O–S–H2O. Holland T, Baker J, and Powell R (1998) Mixing properties and activity–composition
Journal of Metamorphic Geology 28: 667–687. relationships of chlorites in the system MgO–FeO–Al2O3–SiO2–H2O. European
Fayek M, Harrison TM, Ewing RC, Grove M, and Coath CD (2002) O and Pb isotopic Journal of Mineralogy 10: 395–406.
analyses of uranium minerals by ion microprobe and U–Pb ages from the Cigar Holland TJB and Powell R (1985) An internally consistent thermodynamic dataset with
Lake deposit. Chemical Geology 185: 205–225. uncertainties and correlations: 2. Data and results. Journal of Metamorphic Geology
Ferry JM (2000) Patterns of mineral occurrence in metamorphic rocks. American 3: 343–370.
Mineralogist 85: 1573–1588. Holland TJB and Powell R (1990) An enlarged and updated internally consistent
Gaidies F, De Capitani C, and Abart R (2008) THERIA_G: A software program to thermodynamic dataset with uncertainties and correlations: The system
numerically model prograde garnet growth. Contributions to Mineralogy and K2O–Na2O–CaO–MgO–MnO–FeO–Fe2O3–Al2O3–TiO2–SiO2–C–H2–O2. Journal of
Petrology 155: 657–671. Metamorphic Geology 8: 89–124.
Ganguly J (2002) Diffusion kinetics in minerals: Principles and applications to Holland TJB and Powell R (1998) An internally consistent thermodynamic data set for
tectono-metamorphic processes. EMU Notes in Mineralogy 4: 271–309. phases of petrological interest. Journal of Metamorphic Geology 16: 309–343.
Ganguly J (2010) Cation diffusion kinetics in aluminosilicate garnets and geological Hollister LS (1966) Garnet zoning: An interpretation based on the Rayleigh fractionation
applications. Reviews in Mineralogy and Geochemistry 72: 559–601. model. Science 154: 1647–1651.
Ganguly J, Block E, and Ito M (2010) Experimental diffusion kinetics of Hollister LS (1969) Contact metamorphism in the Kwoiek Area of British Columbia:
geochronological systems and interpretations of mineral ages in terrestrial rocks An end member of the metamorphic process. Geological Society of America Bulletin
and meteorites. Geological Society of America Abstracts with Programs 42: 340. 80: 2465–2494.
Paper no. 135-2. Jamtveit B and Hervig RL (1994) Constraints on transport and kinetics in hydrothermal
Ganguly J, Cheng W, and Chakraborty S (1998a) Cation diffusion in aluminosilicate systems from zoned garnet crystals. Science 263: 505–508.
garnets: Experimental determination in pyrope-almandine diffusion couples. Jaoul O and Sautter V (1999) A new approach to geospeedometry based on the
Contributions to Mineralogy and Petrology 131: 171–180. “compensation law”. Physics of the Earth and Planetary Interiors 110: 95–114.
Ganguly J, Tirone M, and Hervig RL (1998b) Diffusion kinetics of samarium and Jenkin GRT (1997) Do cooling paths derived from mica Rb–Sr data reflect true cooling
neodymium in garnet, and a method for determining cooling rates of rocks. Science paths? Geology 25: 907–910.
281: 805–807. Jenkin GRT, Ellam RM, Rogers G, and Stuart FM (2001) An investigation of closure
Ganguly J, Dasgupta S, Cheng W, and Neogi S (2000) Exhumation history of a temperature of the biotite Rb–Sr system: The importance of cation exchange.
section of the Sikkim Himalayas, India: Records in the metamorphic mineral Geochimica et Cosmochimica Acta 65: 1141–1160.
equilibria and compositional zoning of garnet. Earth and Planetary Science Letters Jenkin GRT, Farrow CM, Fallick AE, and Higgins D (1994) Oxygen isotope exchange
183: 471–486. and closure temperatures in cooling rocks. Journal of Metamorphic Geology
Ganguly J and Tirone M (1999) Diffusion closure temperature and age of a mineral with 12: 221–235.
arbitrary extent of diffusion: Theoretical formulation and applications. Earth and Jenkin GRT, Rogers G, Fallick AE, and Farrow CM (1995) Rb–Sr closure temperatures
Planetary Science Letters 170: 131–140. in bi-mineralic rocks: A mode effect and test for different diffusion models.
Garcia-Casco A, Torres-Roldan RL, Millan G, Monie P, and Schneider J (2002) Chemical Geology 122: 227–240.
Oscillatory zoning in eclogitic garnet and amphibole, Northern Serpentinite Johnson CD and Carlson WD (1990) The origin of olivine–plagioclase coronas in
Melange, Cuba: A record of tectonic instability during subduction? Journal of metagabbros from the Adirondack Mountains, NY. Journal of Metamorphic Geology
Metamorphic Geology 20: 581–598. 8: 697–717.
Giletti BJ (1986) Diffusion effects on oxygen isotope temperatures of slowly Karabinos P (1984) Polymetamorphic garnet zoning from southeastern Vermont.
cooled igneous and metamorphic rocks. Earth and Planetary Science Letters American Journal of Science 284: 1008–1025.
77: 218–228. Kelsey DE, Clark C, and Hand M (2008) Thermobarometry modelling of zircon and
Giletti BJ and Casserly JED (1994) Strontium diffusion kinetics in plagioclase feldspars. monazite growth in melt-bearing systems: Example using model metapelitic and
Geochimica et Cosmochimica Acta 58: 3785–3793. metapsammitic granulites. Journal of Metamorphic Geology 26: 199–212.
Graham CM and Elphick SC (1990) Some experimental constraints on the role of Kelsey DE and Powell R (2011) Progress in linking accessory mineral growth and
hydrogen in oxygen and hydrogen diffusion and Al-Si interdiffusion in silicates. In: breakdown to major mineral evolution in metamorphic rocks: A thermodynamic
Ganguly J (ed.) Diffusion, Atomic Ordering, and Mass Transport, Selected Topics in approach in the Na2O–CaO–K2O–FeO–MgO–Al2O3–SiO2–H2O–TiO2–ZrO2 system.
Geochemistry, pp. 248–285. New York: Springer. Journal of Metamorphic Geology 29: 151–166.
Grove TL, Baker MB, and Kinzler RJ (1984) Coupled CaAl–NaSi diffusion in plagioclase King RL, Bebout GE, Kobayashi K, Nakamura E, and Van der Klauw SNGC (2004)
feldspar: Experiments and applications to cooling rate speedometry. Geochimica et Ultrahigh-pressure metabasaltic garnets as probes into deep subduction
Cosmochimica Acta 48: 2113–2121. zone chemical cycling. Geochemistry, Geophysics, Geosystems 5: 17.
Kohn MJ (1993a) Modeling of prograde mineral d18O changes in metamorphic Martin AJ (2009) Sub-millimeter heterogeneity of yttrium and chromium during growth
systems. Contributions to Mineralogy and Petrology 113: 24–39. of semi-pelitic garnet. Journal of Petrology 50: 1713–1727.
Kohn MJ (1993b) Uncertainties in differential thermodynamic (Gibbs Method) P–T Martin LAJ, Ballevre M, Boulvais P, et al. (2011) Garnet re-equilibration by
paths. Contributions to Mineralogy and Petrology 113: 249–261. coupled dissolution–reprecipitation: Evidence from textural, major element and
Kohn MJ (1999) Why most “dry” rocks should cool “wet”. American Mineralogist oxygen isotope zoning of ’cloudy’ garnet. Journal of Metamorphic Geology
84: 570–580. 29: 213–231.
Kohn MJ (2003) Geochemical zoning in metamorphic minerals. In: Rudnick RL (ed.) McConnell JDC (1995) The role of water in oxygen isotope exchange in quartz.
Treatise on Geochemistry, vol. 3, pp. 229–261. Oxford: Elsevier. Earth and Planetary Science Letters 136: 97–107.
Kohn MJ (2004) Oscillatory- and sector-zoned garnets record cyclic (?) rapid thrusting McDougall I and Harrison TM (1999) Geochronology and thermochronology by the
40
in central Nepal. Geochemistry, Geophysics, Geosystems 5: Q12014. http://dx.doi. Ar/39Ar method, 2nd edn. New York: Oxford University Press.
org/10.1029/2004GC000737. Menard T and Spear FS (1996) Interpretation of plagioclase zoning in calcic pelitic
Kohn MJ (2009) Models of garnet differential geochronology. Geochimica et schist, south Strafford, Vermont, and the effects on thermobarometry. The Canadian
Cosmochimica Acta 73: 170–182. Mineralogist 34: 133–146.
Kohn MJ, Catlos EJ, Ryerson FJ, and Harrison TM (2001) Pressure–temperature–time Page FZ, Ushikubo T, Kita NT, Riciputi LR, and Valley JW (2007) High-precision oxygen
path discontinuity in the Main Central thrust zone, central Nepal. Geology isotope analysis of picogram samples reveals 2 mm gradients and slow diffusion in
29: 571–574. zircon. American Mineralogist 92: 1772–1775.
Kohn MJ and Malloy MA (2004) Formation of monazite via prograde metamorphic Parrish RR (1990) U–Pb dating of monazite and its application to geological problems.
reactions among common silicates: Implications for age determinations. Canadian Journal of Earth Sciences 27: 1431–1450.
Geochimica et Cosmochimica Acta 68: 101–113. Pattison DRM, Spear FS, Debuhr CL, Cheney JT, and Guidotti CV (2002)
Kohn MJ and Spear FS (2000) Retrograde net transfer reaction insurance for Thermodynamic modeling of the reaction muscovite þ cordierite ! Al2SiO5 þ
pressure–temperature estimates. Geology 28: 1127–1130. biotite þ garnet þ H2O: Constraints from natural assemblages and implications
Kohn MJ, Spear FS, and Valley JW (1997) Dehydration melting and fluid recycling for the metapelitic petrogenetic grid. Journal of Metamorphic Geology 20: 99–118.
during metamorphism: Rangeley Formation, New Hampshire, USA. Journal of Pollington AD and Baxter EF (2010) High resolution Sm–Nd garnet geochronology
Petrology 38: 1255–1277. reveals the uneven pace of tectonometamorphic processes. Earth and Planetary
Kohn MJ, Valley JW, Elsenheimer D, and Spicuzza MJ (1993) Oxygen isotope zoning in Science Letters 293: 63–71.
garnet and staurolite: Evidence for closed system mineral growth during regional Pollok K, Lloyd GE, Austrheim H, and Putnis A (2008) Complex replacement patterns in
metamorphism. American Mineralogist 78: 988–1001. garnets from Bergen Arcs eclogites: A combined EBSD and analytical TEM study.
Kohn MJ, Wieland MS, Parkinson CD, and Upreti BN (2004) Miocene faulting at plate Chemie Erde 68: 177–191.
tectonic velocity in the Himalaya of central Nepal. Earth and Planetary Science Polyakov VB and Kharlashina NN (1994) Effect of pressure on equilibrium isotopic
Letters 228: 299–310. fractionation. Geochimica et Cosmochimica Acta 58: 4739–4750.
Kohn MJ, Wieland MS, Parkinson CD, and Upreti BN (2005) Five generations of Powell R and Holland TJB (1985) An internally consistent thermodynamic dataset with
monazite in Langtang gneisses: Implications for chronology of the Himalayan uncertainties and correlations: 1. Methods and a worked example. Journal of
metamorphic core. Journal of Metamorphic Geology 23: 399–406. Metamorphic Geology 3: 327–342.
Konrad-Schmolke M, Zack T, O’Brien PJ, and Jacob DE (2008) Combined Powell R and Holland TJB (1988) An internally consistent thermodynamic dataset
thermodynamic and rare earth element modelling of garnet growth during with uncertainties and correlations: 3. Applications to geobarometry,
subduction: Examples from ultrahigh-pressure eclogite of the Western Gneiss worked examples and a computer program. Journal of Metamorphic Geology
Region, Norway. Earth and Planetary Science Letters 272: 488–498. 6: 173–204.
Kretz R (1974) Some models for the rate of crystallization of garnet in metamorphic Powell R, Holland T, and Worley B (1998) Calculating phase diagrams involving solid
rocks. Lithos 7: 123–131. solutions via non-linear equations, with examples using THERMOCALC. Journal of
Kretz R (1983) Symbols for rock-forming minerals. American Mineralogist 68: 277–279. Metamorphic Geology 16: 577–588.
Lanzirotti A (1995) Yttrium zoning in metamorphic garnets. Geochimica et Putnis A (2002) Mineral replacement reactions: From macroscopic observations to
Cosmochimica Acta 59: 4105–4110. microscopic mechanisms. Mining Magazine 66: 689–708.
Lapen TJ, Johnson CM, Baumgartner LP, Mahlen NJ, Beard BL, and Amato JM (2003) Pyle JM and Spear FS (1999) Yttrium zoning in garnet: Coupling of major and
Burial rates during prograde metamorphism of an ultra-high-pressure terrane: accessory phases during metamorphic reactions. Geological Materials Research
An example from Lago di Cignana, western Alps, Italy. Earth and Planetary Science 1: http://gmr.minsocam.org/ContentsLink.html.
Letters 215: 57–72. Pyle JM and Spear FS (2000) An empirical garnet (YAG) xenotime thermometer.
Lasaga AC (1979) Multicomponent exchange and diffusion in silicates. Geochimica et Contributions to Mineralogy and Petrology 138: 51–58.
Cosmochimica Acta 43: 455–469. Pyle JM and Spear FS (2003) Four generations of accessory-phase growth in
Lasaga AC (1983) Geospeedometry: An extension of geothermometry. In: Saxena SK (ed.) low-pressure migmatites from SW New Hampshire. American Mineralogist
Kinetics and Equilibrium in Mineral Reactions, pp. 81–114. New York: Springer. 88: 338–351.
Lasaga AC (1986) Metamorphic reaction rates laws and development of isograds. Pyle JM, Spear FS, Cheney JT, and Layne G (2005) Monazite ages in the Chesham
Mineralogical Magazine 50: 359–373. Pond Nappe, SW New Hampshire, U. S. A.: Implications for assembly of central New
Lasaga AC (1989) Fluid flow and chemical reaction kinetics in metamorphic systems: England thrust sheets. American Mineralogist 90: 592–606.
A new simple model. Earth and Planetary Science Letters 94: 417–424. Pyle JM, Spear FS, Rudnick RL, and McDonough WF (2001)
Lasaga AC and Jiang J (1995) Thermal history of rocks: P–T–t paths for Monazite–xenotime–garnet equilibrium in metapelites and a new monazite–garnet
geospeedometry, petrologic data, and inverse theory techniques. American Journal thermometer. Journal of Petrology 42: 2083–2107.
of Science 295: 697–741. Rayleigh JWS (1896) Theoretical considerations respecting the separation of gases by
L’Heureux I and Fowler AD (1996) Isothermal constitutive undercooling as a model for diffusion and similar processes. Philosophical Magazine 42: 493.
oscillatory zoning in plagioclase. The Canadian Mineralogist 34: 1137–1147. Robinson P (1991) The eye of the petrographer, the mind of the petrologist. American
Lindström R, Viitanen M, Juhanoja J, and Holtta P (1991) Geospeedometry of metamorphic Mineralogist 76: 1781–1810.
rocks: Examples in the Rantasalmi–Sulkava and Kiuruvesi areas, eastern Finland: Rubatto D (2002) Zircon trace element geochemistry: Partitioning with garnet and link
Biotite–garnet diffusion couples. Journal of Metamorphic Geology 9: 181–190. between U–Pb ages and metamorphism. Chemical Geology 184: 123–138.
Loomis TP (1978a) Multicomponent diffusion in garnet: I. Formulation of isothermal Selverstone J, Morteani G, and Staude JM (1991) Fluid channelling during ductile
models. American Journal of Science 278: 1099–1118. shearing: Transformation of granodiorite into aluminous schist in the Tauern
Loomis TP (1978b) Multicomponent diffusion in garnet: II. Comparison of models with Window, Eastern Alps. Journal of Metamorphic Geology 9: 419–431.
natural data. American Journal of Science 278: 1119–1137. Selverstone J, Spear FS, Franz G, and Morteani G (1984) High pressure metamorphism
Loomis TP (1983) Compositional zoning of crystals: A record of growth and reaction in the SW Tauern Window, Austria: P–T paths from hornblende–kyanite–staurolite
history. In: Saxena SK (ed.) Kinetics and Equilibrium in Mineral Reactions. schists. Journal of Petrology 25: 501–531.
Advances in Physical Geochemistry, vol. 3, pp. 1–60. New York: Springer. Sharp ZD (1990) A laser-based microanalytical method for the in situ determination of
Loomis TP, Ganguly J, and Elphick SC (1985) Experimental determinations of oxygen isotope ratios of silicates and oxides. Geochimica et Cosmochimica Acta
cation diffusivities in aluminosilicate garnets. II. Multicomponent simulation 54: 1353–1357.
and tracer diffusion coefficients. Contributions to Mineralogy and Petrology Sharp ZD (1992) In situ laser microprobe techniques for stable isotope analysis.
90: 45–51. Chemical Geology 101: 3–19.
Shore M and Fowler AD (1996) Oscillatory zoning in minerals: A common de Recherches Geologiques et Mminieres – Instituto Geologico y Minero de Espana
phenomenon. The Canadian Mineralogist 34: 1111–1126. 568–576.
Skora S, Baumgartner LP, Mahlen NJ, Johnson CM, Pilet S, and Hellebrand E (2006) Vielzeuf D, Baronnet A, and Perchuk AL (2007) Calcium diffusivity in alumino-silicate
Diffusion-limited REE uptake by eclogite garnets and its consequences for Lu–Hf and garnets: An experimental and ATEM study. Contributions to Mineralogy and
Sm–Nd geochronology. Contributions to Mineralogy and Petrology 152: 703–720. Petrology 154: 153–170.
Smith HA and Giletti BJ (1997) Lead diffusion in monazite. Geochimica et Vielzeuf D, Champenois M, and Valley JW (2005a) SIMS analyses of oxygen isotopes:
Cosmochimica Acta 61: 1047–1055. Matrix effects in Fe–Mg–Ca garnets. Chemical Geology 223: 208–226.
Spear FS (1988) The Gibbs method and Duhem’s theorem: The quantitative Vielzeuf D, Veschambre M, and Brunet F (2005b) Oxygen isotope heterogeneities and
relationships among P, T, chemical potential, phase composition and reaction diffusion profile in composite metamorphic–magmatic garnets from the Pyrenees.
progress in igneous and metamorphic systems. Contributions to Mineralogy and American Mineralogist 90: 463–472.
Petrology 99: 249–256. Wang P and Spear FS (1991) A field and theoretical analysis of garnet þ chlorite þ
Spear FS (1991) On the interpretation of peak metamorphic temperatures in light of chloritoid þ biotite assemblages from the tri-state (MA, CT, NY) area, USA.
garnet diffusion during cooling. Journal of Metamorphic Geology 9: 379–388. Contributions to Mineralogy and Petrology 106: 217–235.
Spear FS (1993) Metamorphic Phase Equilibria and Pressure–Temperature–Time Watson EB (2004) A conceptual model for near-surface kinetic controls on the
Paths. Washington, DC: Mineralogical Society of America. trace-element and stable isotope composition of abiogenic calcite crystals.
Spear FS (2010) Monazite–allanite phase relations in metapelites. Chemical Geology Geochimica et Cosmochimica Acta 68: 1473–1488.
279: 55–62. Watson EB and Baxter EF (2007) Diffusion in solid-Earth systems. Earth and Planetary
Spear FS and Daniel CG (1998) 3-dimensional imaging of garnet porphyroblast sizes Science Letters 253: 307–327.
and chemical zoning: Nucleation and growth history in the garnet zone. Watson EB and Liang Y (1995) A simple model for sector zoning in slowly
Geological Materials Research 1: http://gmr.minsocam.org/ContentsLink.html. grown crystals; implications for growth rate and lattice diffusion, with
Spear FS and Daniel CG (2001) Diffusion control of garnet growth, Harpswell Neck, emphasis on accessory minerals in crustal rocks. American Mineralogist
Maine, USA. Journal of Metamorphic Geology 19: 179–195. 80: 1179–1187.
Spear FS, Ferry JM, and Rumble D (1982) Analytical formulation of phase equilibria: White RW, Powell R, and Holland T (2000) The effect of TiO2 and Fe2O3 on metapelitic
The Gibbs method. Reviews in Mineralogy 10: 105–152. assemblages at greenschist and amphibolite facies conditions: Mineral equilibria
Spear FS and Florence FP (1992) Thermobarometry in granulites: Pitfalls and new calculations in the system K2O–FeO–MgO–Al2O3–SiO2–H2O–TiO2–Fe2O3.
approaches. Journal of Precambrian Research 55: 209–241. Journal of Metamorphic Geology 18: 497–511.
Spear FS, Hickmott DD, and Selverstone J (1990a) Metamorphic consequences of White RW, Powell R, and Holland T (2001) Calculation of partial melting equilibria in
thrust emplacement, Fall Mountain, New Hampshire. Geological Society of America the system Na2O–CaO–K2O–FeO–MgO–Al2O3–SiO2–H2O. Journal of Metamorphic
Bulletin 102: 1344–1360. Geology 19: 139–153.
Spear FS, Kohn MJ, Florence F, and Menard T (1990b) A model for garnet and White RW, Powell R, and Holland T (2007) Progress relating to calculation of
plagioclase growth in pelitic schists: Implications for thermobarometry and P–T partial melting equilibria for metapelites. Journal of Metamorphic Geology
path determinations. Journal of Metamorphic Geology 8: 683–696. 25: 511–527.
Spear FS and Kohn MJ (1996) Trace element zoning in garnet as a monitor of crustal Williams ML and Jercinovic MJ (2002) Microprobe monazite geochronology: Putting
melting. Geology 24: 1099–1102. absolute time into microstructural analysis. Journal of Structural Geology
Spear FS, Kohn MJ, and Cheney JT (1999) P–T paths from anatectic pelites. 24: 1013–1028.
Contributions to Mineralogy and Petrology 134: 17–32. Williams ML, Jercinovic MJ, and Terry MP (1999) Age mapping and dating of monazite
Spear FS, Kohn MJ, and Paetzold S (1995) Petrology of the regional sillimanite zone, on the electron microprobe: Deconvoluting multistage tectonic histories. Geology
west-central New Hampshire, USA, with implications for the development of inverted 27: 1023–1026.
isograds. American Mineralogist 80: 361–376. Wing BN, Ferry JM, and Harrison TM (2003) Prograde destruction and formation of
Spear FS and Markussen JC (1997) Mineral zoning, P–T–X–M phase relations, and monazite and allanite during contact and regional metamorphism of pelites:
metamorphic evolution of some Adirondack granulites, New York. Journal of Petrology and geochronology. Contributions to Mineralogy and Petrology 145:
Petrology 38: 757–783. 228–250.
Spear FS and Parrish RR (1996) Petrology and cooling rates of the Valhalla Complex, Woodsworth GJ (1977) Homogenization of zoned garnets from pelitic schists.
British Columbia, Canada. Journal of Petrology 37: 733–765. The Canadian Mineralogist 15: 230–242.
Spear FS and Pyle JM (2002) Apatite, monazite and xenotime in metamorphic rocks. Yang PS and Pattison D (2006) Genesis of monazite and Y zoning in garnet from the
Reviews in Mineralogy 48: 293–335. Black Hills, South Dakota. Lithos 88: 233–253.
Spear FS and Pyle JM (2010) Theoretical modeling of monazite growth in a low-Ca Yang P and Rivers T (2001) Chromium and manganese zoning in pelitic garnet and
metapelite. Chemical Geology 273: 111–119. kyanite: Spiral, overprint, and oscillatory (?) zoning patterns and the role of growth
Spear FS and Selverstone J (1983) Quantitative P–T paths from zoned minerals: Theory rate. Journal of Metamorphic Geology 19: 455–474.
and tectonic applications. Contributions to Mineralogy and Petrology 83: 348–357. Yang P and Rivers T (2002) The origin of Mn and Y annuli in garnet and the thermal
Spear FS, Selverstone J, Hickmott D, Crowley P, and Hodges KV (1984) P–T paths from dependence of P in garnet and Y in apatite in calc-pelite and pelite, Gagnon terrane,
garnet zoning: A new technique for deciphering tectonic processes in crystalline western Labrador. Geological Materials Research 4: http://gmr.minsocam.org/
terranes. Geology 12: 87–90. contents.html.
Thompson AB and England PC (1984) Pressure–temperature–time paths of regional Yardley BWD (1977) An empirical study of diffusion in garnet. American Mineralogist
metamorphism II: Their influence and interpretation using mineral assemblages in 62: 793–800.
metamorphic rocks. Journal of Petrology 25: 929–955. Young ED (1993) On the 18O/16O record of reaction progress in open and closed
Tirone M, Ganguly J, Dohmen R, Langenhorst F, Hervig R, and Becker H-W (2005) Rare metamorphic systems. Earth and Planetary Science Letters 117: 147–167.
earth diffusion kinetics in garnet: Experimental studies and applications. Young ED and Rumble D (1993) The origin of correlated variations in in-situ 18O/16O
Geochimica et Cosmochimica Acta 69: 2385–2398. and elemental concentrations in metamorphic garnet from southeastern Vermont,
Townsend KJ, Miller CF, D’Andrea JL, Ayers JC, Harrison TM, and Coath CD (2000) USA. Geochimica et Cosmochimica Acta 57: 2585–2597.
Low temperature replacement of monazite in the Ireteba Granite, southern Nevada: Zhang Y (2010) Diffusion in minerals and melts: Theoretical background. Reviews in
Geochronological implications. Chemical Geology 172: 95–112. Mineralogy and Geochemistry 72: 5–59.
Tracy RJ (1982) Compositional zoning and inclusions in metamorphic minerals. Zhang Y and Cherniak DJ (2010) Diffusion in Minerals and Melts. Washington DC:
Reviews in Mineralogy 10: 355–397. Mineralogical Society of America.
Tracy RJ, Robinson P, and Thompson AB (1976) Garnet composition and zoning in the Zhang Y, Stolper EM, and Wasserburg GJ (1991) Diffusion of a multi-species
determination of temperature and pressure of metamorphism, central component and its role in oxygen and water transport in silicates. Earth and
Massachusetts. American Mineralogist 61: 762–775. Planetary Science Letters 103: 228–240.
van Orman JA, Grove TL, Shimizu N, and Layne GD (2002) Rare earth element diffusion Zhu XK and O’Nions RK (1999) Zonation of monazite in metamorphic rocks and its
in a natural pyrope single crystal at 2.8 GPa. Contributions to Mineralogy and implications for high temperature thermochronology: A case study from the
Petrology 142: 416–424. Lewisian terrain. Earth and Planetary Science Letters 171: 209–220.
Vielzeuf D (1996) Les massifs nord-pyrénéens à soubassement granulitique et la croûte Zhu XK, O’Nions RK, Belshaw NS, and Gibb AJ (1997) Significance of in situ SIMS
hercynienne des Pyrénées: Une synthèse. In: Barnolas A and Chiron JC (eds.) chronometry of zoned monazite from the Lewisian granulites, Northwest Scotland.
Synthèse Géologique et Géophysique des Pyrénées, pp. 502–521. Madrid: Bureau Chemical Geology 135: 35–53.

2013 - Kohn - TOG-2nd-edition ZONING WITHIN MINERALS

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2013 - Kohn - TOG-2nd-edition ZONING WITHIN MINERALS

Uploaded by

Copyright:

Available Formats

4.

7 Geochemical Zoning in Metamorphic Minerals

4.7.1 Introduction 250

Symbols f Fraction (of species i remaining in a rock)

Treatise on Geochemistry 2nd Edition http://dx.doi.org/10.1016/B978-0-08-095975-7.00307-7 249

Tc Closure temperature ai Fractionation factor

XSpessartine Moles garnet

0.5 1.0 depicted in Figure 4 of decreasing Mn, increasing Mg, and

namics and mass balance must be accounted for to explain

4.7.2.2.1 P–T paths

5.0 12.0 12.0

5.0 5.0 12.0

0.3 0.3 0.8

XSps, XGrs, XPrp

High-T retrograde exchange (ReER) Rim

ReNTR biotite Garnet ReER biotite

ReNTR Core ReER

choosing the best compositions from which one calculates T

0.05 0.15 0.25 0.35 0.00 0.02 0.04 0.06

Garnet grows Continuous Continuous Discontinuous

An40 An40 An35

solvable zoning in its interior, which again was consistent with

theoretical models and observed zoning in this study implies

1.0 Alsi-in St-in

500 ⬚C St-out model

0 500 1000 1500 2000 2500

(Hickmott and Shimizu, 1990), and infiltration of fluids

et al., 2006; Yang and Pattison, 2006). Equilibrium models,

2002) and, in some instances, ascribed to limited transport of

pl s b grt sil crd pl kfs liq qtz zrn grt crd pl

(Yang and Rivers, 2002). For example, sudden enrichment in

400 40 LREEs could reflect breakdown of apatite, allanite, or mona-

Ca P 4.7.5 Radiogenic Isotopes (Age Variability)

Radiogenic isotopes are commonly measured in metamorphic

4.7.5.1 Growth Zoning

Calcium Yttrium Manganese

can be determined. Depth profiling into the face of a crystal

(c) Age (d)

You might also like