Hi Bill,

I've been on vacation, so just picked up on this. And, frankly, I only  
read a few responses, including yours.

Just wanted to mention that I found increases in Y and Lu on the rims  
of some Himalayan garnets (Kohn, 2009, GCA). I don't know whether  
you'd have any interest in modeling them, but if so I can track down  
the data and send to you. Or we could talk at GSA if you're going.

Best,

Matt

On Aug 7, 2010, at 4:08 PM, William Carlson wrote:

> All:
>
> Reliable measurements of diffusion rates in garnet exist for both  
> the common divalent cations (Mg, Fe, Mn, and Ca) and for some REEs.   
> The former are quite comprehensive – allowing evaluation across a  
> wide range of T, P, oxygen fugacity, and garnet composition, all of  
> which are important controls.  The latter are less comprehensive,  
> somewhat less consistent, and more difficult to extrapolate/ 
> interpolate.  But together these data probably explain Alexander's  
> observations, and most of the other observations offered by  
> contributors to this thread.
>
> (1) We should start by recognizing the elegant and definitive  
> experiments of Jiba Ganguly and Sumit Chakraborty and collaborators  
> (Chakraborty & Ganguly 1992 CMP 111:74; Ganguly et al. 1998 CMP  
> 131:171).  These data, and additional results from 5 other  
> investigations spanning a very wide range of P, T, and fO2, can all  
> be reconciled to one another when the effects of garnet composition  
> are taken into account: see Carlson 2006 AmMin 91:1.  From these  
> data, diffusional relaxation of major elements in garnet can be  
> calculated with considerable rigor for any specified temperature- 
> time history.  It's a multi-component calculation, and therefore not  
> an easy one, but it is certainly tractable, and examples are easy to  
> find.
>
> (2)  Data on diffusivity of selected REEs in garnet are available in  
> Ganguly et al. 1998 Science 281:805, Van Orman et al. 2002 CMP  
> 142:416, and in Tirone et al. 2005 GCA 69:2385.
>
> (3)  To this I can add my own results (in prep., I regret to say)  
> from numerical simulations of the development of stranded diffusion  
> profiles for REEs in partially resorbed garnets in the aureole of  
> the Makhavinekh Lake Pluton (MLP), Labrador. (These are the same  
> crystals for which equivalent simulations and data for Mg, Fe, Mn,  
> and Ca diffusion were reported in Carlson 2006.)  Elevated  
> concentrations of MREEs and HREEs (Gd-Lu) are measured by LA-ICPMS  
> as rims of relict garnets are approached, over ranges of ca. 200  
> microns (hot inner aureole, peak T near 900 deg C) to ca. 50 microns  
> (central aureole, peak T near 850 deg C).  Decreases in  
> concentrations of Sm, Nd, and Eu are measured by SIMS over the same  
> ranges of distance.  The difference reflects partitioning, during  
> resorption, with the product assemblage of crd+opx+mnz.  Profiles  
> are fit exceedingly well by numerical models of coupled resorption  
> and diffusion.  I'm still finalizing the numbers, but in general,  
> diffusivities measured on the same crystal show that REE diffusion  
> is slower by ca. 1.5 to 2.5 orders of magnitude than for divalent  
> ions (i.e., diffusion coefficients are smaller by about two log10  
> units).
>
> (4)  So with reference to various points raised in the thread: There  
> are a host of possible complications, not the least of which is  
> garnet composition, but I would maintain that … (a) Common reports  
> of homogenization of Mg, Fe, Mn, and to a lesser extent Ca in  
> millimeter-scale garnets at uppermost amphibolite facies conditions  
> or above  (Bruce, Bob,  and many others) is entirely consistent with  
> measured diffusion rates, presuming 'geological' time at those  
> temperatures (time spans of millions of years).  (b) Appreciable  
> diffusion of REEs requires markedly higher temperatures or markedly  
> longer times. Diffusion across length scales of only several tens of  
> microns or less is to be expected over million-year time frames at  
> temperatures near 800 deg C. So with respect to Alexander's  
> observations -- given that steep initial zoning of REEs should be  
> expected from strong partitioning into garnet during growth -- the  
> persistence of "bell-shaped" profiles (or other zoning features) for  
> REEs in crystals of millimeter scale is not surprising, even in  
> crystals for which zoning of divalent cations has been homogenized.   
> (c)  Nigel's attribution to diffusion of observed enrichments of  
> HREEs in garnet rims during resorption is quite justified,  and  
> matches the features seen in garnets from the MLP aureole.

**************************************************
Dr. Matthew J. Kohn, Professor
Department of Geosciences
Boise State University
1910 University Dr.; MS1535
Boise, ID 83725-1535
[log in to unmask]
phone: (208)-426-2757   fax: (208)-426-4061
**************************************************