Two series of experiments were performed at atmospheric pressure to evaluate the dependence of cooling rate on the mineralogy and texture of lava from Site 989. The first series was conducted at isothermal conditions to bracket the liquidus temperature and the crystallization sequence for equilibrium conditions. The details of the liquid line of descent and the phase compositions for the run products are presented in a companion paper (Thy et al., Chap. 9, this volume). The second series of experiments were performed under controlled cooling conditions, where the sample was held above the liquidus, cooled at a linear rate to a specified temperature, and quenched.

Experiments were conducted in a Deltech DT31VT vertical quenching furnace, modified to accept various gas mixtures to control the redox state of the sample. In this study, oxygen fugacity (f_O2) of the sample atmosphere was controlled by a low volume stream (~0.1 mL/s) of mixed CO-CO₂. The f_O2 imposed by the gas mixture was measured before and after each run using a solid-state ZrO₂-CaO electrolyte cell calibrated at the Ni-NiO and Fe-FeO buffers. Oxygen fugacity was specified to correspond to the fayalite-magnetite-quartz (FMQ) buffer at the melting temperature in the case of the equilibrium experiments, and at the dwell temperature for the cooling rate experiments. In the latter experiments, no adjustments were made to the gas mixture to maintain f_O2 at the FMQ buffer as the temperature was lowered. The temperature was regulated via a standard feedback loop between the Pt/Rh control thermocouple and a Eurotherm 818 programmable controller. The sample temperature was monitored by a Pt/Pt₉₀Rh₁₀ (S-type) thermocouple calibrated against the melting point of gold (1064ºC) and located just above the sample hanger.

Starting material was a representative 5 g of Unit 1 (163-989B-10R-7, 55-59 cm) (Table 1). The sample was crushed in a tungsten-carbide shatterbox and ground by hand under acetone in an agate mortar. The average grain size of the powdered sample is ~10 µm. Aliquots of 50 mg of dry powder were compressed into small pellets and fused to Fe-alloyed Pt wire loops. Platinum-iron alloy loops were fabricated by iron electroplating pre-formed 0.01-cm-diameter Pt wire with 10 wt% Fe, followed by annealing in a N₂ atmosphere at 1200ºC for 12 to 15 hr before sample mounting (Grove, 1981). The Pt-Fe alloy loops were successful in minimizing iron loss from the sample, while the large sample size and low rate of gas flow ensure minimal loss of Na during the experiment.

Equilibrium experiments (i.e., constant temperature) were held at the target temperature for 8 to 27 hr (Table 2), and quenched in air by rapidly extracting the sample from the furnace. Controlled cooling experiments were held 10º-12ºC above the liquidus temperature for ~1 hr before initiating cooling to the target quench temperature (Table 3). The linear cooling rate was varied from 10º to 2000ºC/hr. Quench temperatures ranged from 1150º to 1000ºC. Some of the cooling rate experiments conducted at the highest cooling rates were dropped to the bottom of the furnace and quenched into a small beaker of water. No significant quench crystal growth was observed in any of the samples, either when manually removed or drop quenched.

After quenching, individual charges were mounted and polished for examination. All samples were examined using reflected-light microscopy, backscattered electron (BSE) microscopy, and wave-length dispersive spectrometry (WDS). WDS data for equilibrium experiments are given in Thy et al. (Chap. 9, this volume). The experimental run conditions and phase assemblage for the equilibrium and cooling rate experiments are summarized in Table 2 and Table 3. Phase proportions for equilibrium experiments estimated by mass balance by Thy et al. (Chap. 9, this volume) are provided in Table 2. Table 3 lists visual estimates of the phase proportions for the controlled cooling experiments. A distinction is made in Table 3 between optically clear glass (quenched melt) and the microcrystalline matrix of plagioclase + pyroxene ± olivine found in low-quench-temperature runs.

Textural analyses of both natural samples and experimental run products were conducted using a JEOL 6300 secondary electron microscopy. BSE images were collected using a 10-keV accelerating voltage, 15-mm working distance, and 5-na beam current. Semiquantitative characterization of textures was accomplished by measuring crystal widths directly from BSE images. More quantitative approaches were unsuccessful because of the extreme textural heterogeneity of the experimental run products and to the extent of groundmass alteration of the natural samples.