Читать книгу Isotopic Constraints on Earth System Processes - Группа авторов - Страница 33

Richter et al. (2014b) carried out a number of experiments in which powdered spodumene (LiAlSi₂O₆), or Li₂SiO_3, was used as a source to diffuse lithium into Templeton augite or into Dekalb diopside at 900°C and oxygen fugacity ranging from log fO₂ = –17 to log fO₂ = –12. Fig. 1.10 shows the experimental design of these experiments. The major element composition of the pyroxene grains was sufficiently homogenous that the lithium measurements, which were made with a CAMECA 1270 ion microprobe at the Centre de Recherches Pétrographiques et Géochimiques in Nancy, France, did not require corrections for matrix effects.

The lithium concentration profiles and isotopic fractionations measured in the augite grains recovered from some of the Richter et al. (2014b) experiments were nothing like the smoothly varying profiles found in earlier experiments involving diffusion in silicate liquids (see Figs. 1.3 and 1.6). Fig. 1.11 shows the results from one such experiment, in which the lithium diffused into augite as a sharp step that propagated into the grain with time. The negative isotopic fractionation is extraordinarily large (~30‰) for only a factor of two difference in lithium concentration and it abruptly returns to the unfractionated value of the interior at the place where the lithium concentration abruptly drops down to the initial concentration in the grain. Dohmen et al. (2010) had previously reported similar step‐like profiles for lithium diffusing in olivine, which they explained in terms of lithium existing in two distinct sites, one as fast diffusing interstitial lithium and much slower diffusing lithium in octahedral metal sites. Exchange between these sites was mediated by vacancies. Richter et al. (2014b) adopted the Dohmen et al. (2010) two‐site model to generate model curves to fit of the lithium concentration and isotopic fractionation from experiments where lithium had diffused into augite. The interested reader will find a detailed description of the two‐site model for lithium diffusing in olivine in Dohmen et al. (2010) and its implementation for modeling both the concentration and isotopic fraction of lithium diffusing in pyroxene in Richter et al. (2014b). For present purposes the important result is that the isotopic fractionation shown in Fig. 1.11 was fit by a model calculation using a Dohmen‐type two‐site diffusion model with and β _Li=0.27 for the relative diffusion coefficients of the isotopes of the interstitial lithium.

Figure 1.10 Diagram showing the experimental design used to diffused lithium from a Li‐rich powder (spodumene in this figure) into an augite grain. The grain was surrounded by the lithium source in an evacuated silica ampule together with an oxygen fugacity buffer (Ni‐NiO, Fe‐FeO, or FeO‐Fe₂O₃) and then placed in a furnace at 900°C for a prescribed amount of time. Richter et al., 2014b

The data shown in Fig. 1.11 were so unexpected that Richter et al. (2014b) ran additional experiments with different durations, different orientations relative to the cleavage of the augite, and different fO₂. Most, but not all, of these experiments had results like those in Fig. 1.11. However, in some cases, as shown in Fig. 1.12, the lithium concentration profile and associated isotopic fractionation was much more like what one might have expected based on many previous the diffusion experiments in liquids and other solids. The main difference between results of experiment LiPx8 shown in Fig. 1.11 and those from LiPx22 shown in Fig. 1.12 was the oxygen fugacity, which was buffered at 10^–12 bars in experiment LiPx8 versus a value 10^–15 bars in experiment LiPx22. The model calculations that produced step‐like profiles required a larger concentration of vacancies compared to the lithium concentration at the boundary, while the calculations the resulted in smooth profiles like in Fig. 1.12 resulted when the initial concentration of vacancies was the same as or less than the lithium concentration at the grain edge. An important result is that regardless of the shape of the lithium concentration profile, the isotopic data can be fit using the same value of β _Li = 0.27.

Figure 1.11 The panel on the left uses black circles to show the lithium concentration (normalized to a value of one in the interior) measured along a traverse across a 700 μm augite grain from experiment LiPx8 from Richter et al. (2014b). Distance is normalized by the size of the grain. This experiment was annealed for 842 hours at 900°C and buffered by Ni–NiO (log fO₂ ~–12). The lithium concentration was measured along one line while the isotopic fractionation shown in the panel on the right was measure along two parallel lines. Comparing the isotopic data from the two lines provides a good measure of the uncertainty of the isotopic measurements. The lithium isotopic fractionation is reported as using the average of the ratio measured in the interior of the grain as the standard. The solid lines are the results of a model calculation based on lithium occupying two types of sites in the augite and β_Li = 0.27.

Figure from Richter et al. (2014b).

Figure 1.12 Lithium concentration and isotopic fractionation normalized by the average values in the interior portion of the grain that had not been affected by diffusion. These data are from experiment LiPx20 from Richter et al. (2014b) in which lithium diffused into the grain for 672 hours at 900°C with the oxygen fugacity buffered by wüstite–magnetite (log fO₂= –15). The data shown here were measured across only one boundary layer rather than across the entire grain. The isotopic data are shown with 2σ error bars derived from 20 or more repeated analyses at each spot. The model profile that fits the lithium concentration data used a two‐lithium species model with the initial vacancy concentration in the interior equal to the lithium concentration imposed as a boundary condition at the grain edge. The modeled profile of the isotopic fractionation associated with the diffusion of lithium into the grain shown by the black line was calculated assuming β_Li = 0.27.