Thanks to all the members of my thesis committee for your guidance: Paul Asimow, John Eiler and Claire Bucholz. Of course, I will always cherish all the knowledge that is dropped in our weekly Petrology Reading Group.
Introduction
Commentary on “Effects of composition and temperature on sulfur speciation and solubility in silicate melts” by Nash et al. Rate-controlled cooling experiments on olivine-hosted melt inclusions: chemical diffusion and lava cooling rates.
Controlled cooling rate experiments on olivine-hosted melt inclusions: chemical diffusion and the cooling rates of lavas on
Hawai‘i and Mars
Sample selection
Experimental inclusions are labeled according to the H#X# scheme, where the first number corresponds to the experiment (e.g., H2X) and the second to the specific olivine grain (e.g., the Kīlauea Iki melt inclusions are petrographically similar to the Papakōlea samples, and the Y984059 melt inclusions are described in more detail in section 6.2.
Experimental procedures
During cooling from 1225 °C, the gas mixing ratio remained unchanged, resulting in a progressive decrease of the furnace ƒO2 to ∆FMQ. The dotted lines indicate the temperature in the oven after moving the MgO rod containing the experimental samples.
Sample analysis
Bright phases in the inclusions include both sulfides (panels b, c and d) and Cr-Al spinel (panels b and f). The natural inclusions (not shown here) look essentially the same as the experimental samples (e.g., see Fig. 1a), except for the higher density of cracks in the experimental olivine.
Natural olivine and glass inclusions (unheated)
For each inclusion, a bulk composition was calculated by approximating the inclusion as a sphere and numerically integrating each oxide profile over the inclusion. The compositional zoning in the natural unheated glassy melt inclusions is discussed in more detail in section 6.
Temperature-time paths of cooling experiments
The x-axis is radial distance in micrometers relative to the center of the glass inclusion. The best-fit linear cooling rates for each experiment are shown in Figure 2 and listed in Table 2 .
Textural description of experimental samples
Such sulfide bubbles are found in at least one inclusion in 80% of the experimental olivines and are often spatially associated with the inclusion wall and the vapor bubble (Fig. 3c). Three of the twelve unheated Papakōlea olivines have one or two visible sulfide bubbles in the melt inclusions; natural inclusions had fewer and larger sulfides than most experiments and do not appear to be associated with a steam bubble.
Compositional zonation in experimental inclusions
If we take the FeO*/MgO plateau of the homogenized inclusions and the olivine measured along the walls of the inclusions, we get KD,Fe-Mg = 0.298–. The width of the boundary layer varies from oxide to oxide, reflecting their diffusivity in the melt (Newcombe et al., 2014), and the boundary layers terminate at compositional plateaus.
Zoning in olivine
In some samples there is a clear break in the slope between the wide zone and the steep, narrow zone, consistent with the latter being formed by crystallization of olivine on the inclusion wall due to a disturbance of the olivine environment ( for example, abrupt cooling during eruption). . The far-field olivine reported in the text is consistent with microprobe analyzes taken >150 µm from the inclusion wall.
Model description
The model assumes a local equilibrium between olivine and the fluid at the inclusion wall (hereafter "boundary fluid") and imposes a temperature-dependent relationship between MgO in the boundary fluid and Fo90.6 olivine; for this ratio, Newcombe et al. 2014) used the results of the olivine dissolution experiments of Chen and Zhang (2008). The blue and black solid lines correspond to batch MELTS and fractional crystallization models of olivine, respectively, showing the predicted MgO content of the liquid homogenized at 1225 °C ( Table 2 ) with progressive olivine extraction.
Application of the modified model
For H4X and H9X, the fastest cooled experiments, the residuals are artificially low because most of the data points in the profile correspond to the MgO plateau and thus the difference between the data and the model, with the initial MgO set by the central plateau, is small (e.g. e.g. figure 12ab). Application of the inverse model for MgO diffusion described in Section 4 to experimentally cooled olivine-hosted melt inclusions. The red horizontal line corresponds to the initial MgO in the model calculation, corresponding to T = Tmax via equation (1) (section 4.2).
The blue curves correspond to the result of the best fit of the inverse model, where the cooling rate was the only adjustable parameter, and the corresponding cooling rates are listed below each profile along with the best linear fits to the experimental thermocouple data in Fig. 2.
Assessing the accuracy of cooling rates extracted from zoning profiles of olivine-hosted melt inclusions olivine-hosted melt inclusions
In the context of determining cooling rates from natural samples, which span many orders of magnitude (see Newcombe et al., 2014 and Section 6.1), the agreement between the model and experimental rates demonstrates the utility of treating diffusion across melt inclusions as ' using a geospeed meter. The experimental profiles can also fit the unmodified model presented in Newcombe et al. 2014 ), and the single-phase cooling rates recovered from their model are as accurate as those calculated with the updated model (typically ±10% relative and within a factor of two). The accurate cooling rates calculated with the unmodified model are despite the fact that the modeled Tmax underestimates the homogenization temperature by ~70 °C (Section 4.1).
The agreement between the known and the best cooling rates for the experimentally treated inclusions shows that for determining the cooling rate experienced during the last stages of the cooling history of an olivine-hosted melt inclusion, either the model presented here or that proposed by Newcombe used. et al. 2014) is accurate to within a few tens of percent.
Observations of CaO in the experiments
As described in section 3.2, evidence for multicomponent diffusion is widespread in natural (Newcombe et al., 2014, and this work) and experimentally cooled, olivine-hosted melt inclusions (this work), with the development of the unexpected 'reversed' is. For example, at intermediate cooling rates (Figure 8bc), there are local maxima in CaO near the opposite walls of the containment. However, further investigation of the extreme CaO behavior in olivine-hosted melt inclusions – first described in Newcombe et al. 2014) – is now possible due to its reproducibility and the systematic evolution of the CaO profile shape observed in our cooling experiments.
In addition, low-CaO profiles in melts near the olivine interface relative to the inclusion center show varying amounts of relaxation and inward displacement of local maxima at intermediate distances from the inclusion rim.
A simplified model of CaO diffusion in olivine-hosted melt inclusions In this section, we develop a simplified treatment that can explain semi-In this section, we develop a simplified treatment that can explain semi-
Imagine a single increase in olivine crystallization followed by isothermal relaxation of the increased CaO content in the olivine-depleted melt at the interface. The low XCaO near the inclusion wall for all profiles is due to the build-up of Na2O in the interfacial region during progressive olivine crystallization. The red horizontal line is the calculated initial MgO in the liquid before cooling to Tmax.
In addition to the Papakōlea samples, melt inclusions were measured in four olivines collected at Kīlauea Iki crater.
Mars
MELTS batch crystallization of olivine was used for the temperature-dependent MgO boundary condition in the interstitial fluid, starting from the average reconstructed fluid composition of Y980459 inclusions. The distribution of mean cooling rates (10 iterations perturbed by 1σ noise) calculated for each inclusion is 383 ± 43 °C/hr (n = 8). Cooling rates calculated from MgO diffusion in melt inclusions correspond to stage 3 of the proposed cooling history of Y980459, which produced the vitrophyre texture in the rock.
The observed zoning develops with time due to the competition between the continuous formation of a boundary layer in the melt at the inclusion wall and the diffusive exchange between this boundary layer and the interior of the inclusion. The three different MgO rods used in the experiments are shown on the right side of the figure. Sections S1, S2, and S3 of the supplementary material in Newcombe et al. 2014) provide a detailed description of the model (S1), a discussion of the model assumptions (S2), and sensitivity tests (S3).
The temperature-dependence and kinetics of electron transfer between S and Fe in silicate melts: perspectives from olivine-
Abstract
The relationship between Fe and S valence as a function of oxygen fugacity (ƒO2), temperature and cooling rate was investigated using natural olivine-hosted melt inclusions as experimental vessels. Experimental melt inclusions were equilibrated for 24 h at 1225 °C and at values of ƒO2 applied by the external H2-CO2. Fe3+/FeTotal measured in unheated melt inclusions from Papakōlea, Mauna Loa, Hawai'i varied from the experimental melt inclusions have Fe3+/FeTotal.
Although it is unclear whether the inclusions reached equilibrium with the ƒO2 of the furnace atmosphere, the discrepancies between the models and measurements of Fe3+/FeTotal are systematic and are interpreted as due to (i) the presence of sulfide nanolites in the analytical volume of the Fe
Introduction
A similar magnitude and direction of the effect of S on Fe3+/Fe2+ was also documented in high-pressure experiments described in an AGU abstract (Graz et al. 2006), but never published. The same experimental approach can be used to investigate the possibility that the Fe3+/Fe2+ ratios in natural glasses can change during cooling and are only “frozen” at the vitrification temperature. If this is the case, it greatly complicates the use of room-T measurements of Fe3+/Fe2+ ratios in glasses to derive information, such as ƒO2, about high-T magmatic fluids.
This information can be used to provide guidance on which natural glasses Fe3+/Fe2+ and S6+/S2- may have changed during cooling and to estimate the magnitude and direction of this effect.
Methods
- Experimental Approach
- Sample Analysis
The ambient temperature in the oven at the final position of the quench bar in experiment SF13 was 600 °C. The routine electron probe analysis of the melt inclusions was the same as that described in Chapter 1. The reduced results from the earlier session were described in Chapter 1. The spot size on the glass surfaces was ~10 µm.
However, the probability of the s→p electronic transition increases with oxidation state (Waldo et al. 1991), and thus the absorption cross section is not constant over the different energies measured in the XANES spectra.
Results
- Petrography of Sulfides in Experimental Melt Inclusions
- Chemical Compositions of Experimental Melt Inclusions .1 Major and Minor Element Compositions of Melt Inclusions .1 Major and Minor Element Compositions of Melt Inclusions
- Compositional Zoning in Cooled Melt Inclusions and Host Olivines After the 24 hr homogenization step, three cooling regimes were employed: (1)
- S, H, and C Contents of Melt Inclusions
Major and minor elemental compositions of the centers of experimental and natural olivine-hosted melt inclusions. FeO* wt% across experimental melt inclusions (red squares) and the surrounding host olivine (green squares). Listed in each panel is the Fo content of the olivines adjacent to the inclusions and in the far field 75 µm from the glass–olivine interface.
Because all the inclusions were hosted in the same olivine (Fo88.4), these differences likely reflect heterogeneities in the compositions of the originally trapped melt inclusions.
