ANALYTICAL METHODS

Total carbon and organic carbon contents were determined for all the samples studied using a LECO WR-12 carbon determinator. Approximately 0.1 g of powdered sample was weighed in a ceramic crucible and oxidized at 1500°C for 55 s, and the evolved CO₂ gas was measured to obtain total carbon content (tot-C). For determination of organic carbon content (org-C), approximately 0.1 g of powdered sample was treated with 2-N HCl for 24 hr and then dried at 60°C within a permeable crucible for 2 days. Carbonate carbon content (carb-C) was calculated by tot-C minus org-C. The precision of measurement was better than ±0.05 wt%.

Semiquantitative analysis of bulk mineral composition was conducted for all the samples using a MAC Science MXP-3 X-ray diffractometer equipped with CuK tube and monochromator. Measurements were conducted at tube voltage of 40 kV and tube current of 20 mA with variable slit system that automatically controls a 25-mm beam width on the sample. Scanning speed was 4° 2/min, and data sampling step was 0.02° 2. A powdered sample was mounted on a glass holder and X-rayed from 2° to 40° 2. Before reading out the positions and heights of diagnostic reflections, two steps of data processing were applied. As a first step, original data were smoothed by 5-pt. averaging, which is equivalent to a smoothing window width of 0.1° 2. This process minimizes an error caused by noise. As a second step, a background-including amorphous hump was estimated by the background evaluation program, which uses a wider smoothing window with 30-pt. (equivalent to 6° 2) between 10° and 40° 2. Because the smectite peak is broad (~6° 2 in width), a smoothing window of 100-pt. (equivalent to 20° 2) was used between 2° and 10° 2. The background profile is subtracted from the 5-pt. smoothed profile to obtain the net peak intensities of crystalline minerals other than smectite. The background profile using a 100-pt. smoothing window was used for smectite.

Identification of minerals is based on the following diagnostic peaks; 7.2° for smectite, 8.8° for illite, 12.1° for chlorite + kaolinite, 26.6° for quartz, 27.8° for feldspars, 29.3° for calcite, and 32.9° for pyrite. Amorphous material, which includes poorly crystalline aluminosilicate, organic matter, and opal, is identified and semiquantified based on a broad hump between 16° and 32.5°. A reference sample was measured twice per day to correct for the drift of measurement condition. The peak intensities of the minerals were used to semiquantitatively determine mineral abundance. The reproducibility of the measurement is better than ±7% for quartz, ±10% for amorphous hump, ±15% for feldspars, ±20% for calcite, pyrite and smectite, and ±30% for illite and chlorite + kaolinite.

The concentrations of 10 major elements (SiO₂, TiO₂, Al₂O₃, Fe₂O₃, MnO, MgO, CaO, Na₂O, K₂O, and P₂O₅) were determined for all the samples by XRF analysis using a Rigaku 3270 spectrometer equipped with a Rh tube at the Ocean Research Institute, the University of Tokyo. The measurement was carried out on a fused glass bead at an acceleration voltage of 50 kV and a current of 50 mA. Powdered samples were dried at 110°C for more than 4 hr and then ignited at 1000°C for 6 hr to remove volatiles. Loss on ignition (LOI) was calculated from the weight loss. Approximately 0.4 g of ignited samples were mixed with ~4 g of Li₂B₄O₇ flux with the exact ratio of 1.000:10.00, and fused in a platinum crucible to make glass beads within 8 hr after ignition so as to avoid weight changes due to absorption of H₂O and CO₂. A calibration curve was constructed using 40 standard samples provided from the Geological Survey of Japan, the United States Geological Survey, and the National Bureau of Standards (Irino and Tada, 2000). The reproducibility (95% reliability) of measurement in relative scale is ±0.6% for SiO₂, ±0.8% for TiO₂, ±0.7% for Al₂O₃, ±0.7% for Fe₂O₃, ±1.4% for MnO, ±1.0% for MgO, ±0.8% for CaO, ±1.6% Na₂O, ±0.7% for K₂O, and ±1.2% for P₂O₅.

We analyzed grain-size distribution of 79 selected samples by a laser diffraction-scattering method using Horiba LA-920 Grain Size Analyzer equipped with tungsten and He-Ne laser light sources. An entire pretreated sample was introduced into the analyzer with ~300 cm³ of 0.2% sodium pyrophosphate solution, which aided dispersion of particles. Sample dispersion was circulated in a closed-tube system with in situ ultrasonification for 20 min before measurement. The result was transformed into volumetric grain-size distribution. Measurement was conducted from 0.02 to 2000 µm in diameter, with grain-size class of every log₁₀(µm) = 0.06 (total 85 classes).

Because the analyzed samples occasionally showed tri- or bimodal grain-size distribution, we resolved the grain-size distribution curve into two or three log normal distribution curves using a repetitive numerical curve fitting by the Igor^TM computer program. Relative error (1) of grain-size parameters for "fine" (~12 µm) and "coarse" (~57 µm) size fraction mode were ±5% for the mode position (log₁₀[µm]), ±13% for the amount of mode, and ±15% for the mode width (log₁₀[µm]). Those for "very fine" (~3 µm) mode were ±5% for the mode position (log₁₀[µm]), ±57% for the amount of mode, and ±16% for the mode width (log₁₀[µm]).