USING NATURAL GAMMA RAY TO CALCULATE POTASSIUM BUDGETS

Introduction

A primary objective of drilling Hole 801C during Leg 185 was to quantify the chemical alteration of Jurassic basement in the west Pacific in order to calculate geochemical fluxes into the Mariana subduction zone. Although important to geochemical mass balances, estimating the bulk compositions of altered basement is difficult because of the extremely heterogeneous nature of alteration domains in basalt. Consequently, most studies of basement alteration focus on styles and mechanisms of water-rock reaction and not absolute budgets. If basement alteration involved long homogeneous sections of rock, then chemical analyses of discrete samples may adequately represent the compositional variations in the core and can be easily summed. However, when the chemical composition of the basement varies on the scale of millimeter- to centimeter-sized veins, halos, or interpillow domains, then the analytical problem becomes extremely complicated.

The continuous collection of data on whole-round cores using the MST provides a potentially powerful complement to discrete sampling and analysis for estimating the bulk composition of long sections of core. In particular, the NGR emission data reflect the chemical composition of the cores and relate simply to the concentrations of K, Th, and U, the dominant, naturally occurring radioactive elements in rocks. Although this provides only a limited view of the total chemical variation in the core, K and U are two elements that are enriched and intimately associated with alteration zones in the upper oceanic crust. If the MST-NGR data can resolve variations in basement alteration, then this information will not only help to guide discrete sampling efforts for more comprehensive chemical analyses but also provide a clear tie to the downhole logs. In fact, the MST-NGR data are the only continuous core data collected in Hole 801C that can be compared directly to the logs. If the MST-NGR data can be calibrated to K, Th, or U concentrations, it would provide an excellent control on the bulk composition of the recovered core.

There are some difficulties, however, in using the MST-NGR data for quantitative analysis. The first is that there is currently no calibration for the NGR data; the output is in counts per second (cps) and not elemental concentrations. The second is that although multichannel spectral data are collected, the results are not integrated into K, Th, and U peaks separately but as the sum of the gamma-ray counts. Finally, although alteration domains may be enriched in K and U (>2 wt% K2O and >0.5 ppm U), the K, U, and Th abundances in MORB are extremely low (generally <0.2 wt% K2O and <0.3 ppm Th and U) and thus may be below the detection limit of the technique. To assess these problems, we conducted studies during Leg 185 to ascertain the effectiveness of the MST-NGR data in measuring the chemical composition of the basement cores. These were neither extensive nor systematic studies; for a more in-depth discussion of limitations, improvements, and applications of the ODP MST-NGR data, see Blum et al. (1997) and Hoppie et al. (1994).

Correspondence between MST-NGR Data and Features in the Core

The first consideration is whether the MST-NGR data relate to observable alteration domains in the core. Figure F91 shows a schematic lithologic section of Core 185-801C-14R, along with the MST-NGR and MST-magnetic susceptibility (MS) data. The MST data were measured every 10 cm for 20 s (NGR) and every 4 cm for 4 s (MS) (see "Physical Properties"). Also shown for comparison are the downhole gamma-ray logs and K2O measurements of discrete core samples by atomic absorption (see "Analysis of Potassium in Basalts by Atomic Absorption Spectrometry" in "Igneous and Sedimentary Geochemistry" in the "Explanatory Notes" chapter for methods) (Table T23). We chose this core for examination because NGR measurements show a simple pattern of low count rate regions with two large peaks midway through the core, and the variations span almost the full dynamic range of the tholeiite section in Hole 801C. These variations in NGR correspond closely to features in the core. The low count rates in the upper 2.3 m are in a minimally altered massive basalt flow, which we expect to have low MORB-like K, U, and Th values throughout. The low expected K abundances are confirmed by the low K2O atomic absorption (AA) values (0.074 wt%), and although U and Th data are not available at this time (but will be collected onshore), we expect them to be as low as <0.3 ppm Th and U, which are typical of the less altered basalt cored during Leg 129 (Castillo et al., 1992). Both NGR peaks correspond to celadonite-bearing interpillow siliceous sediment. Celadonite is the main K-rich alteration mineral, and in general, its presence correlates with high NGR counts, high K concentration and a blue-green color in the cores. The rest of the core consists of less altered basaltic pillows and flows, all with low NGR counts. MS is generally out of phase with NGR because alteration reduces MS, and siliceous interpillow material has low intrinsic MS. A similar trend to the MST-NGR is shown by the K2O core data, if we assume that the discrete K2O measurements are typical of the sampled lithologic units. Finally, the downhole gamma-ray log (from the triple-combo string) shows the same over-all pattern, albeit greatly smoothed, such that the two peaks merge into one large one. Thus, the MST-NGR data show a great deal of coherence with the measured K2O abundances, the gamma-ray logs, and the patent alteration and sedimentary features in the core.

Whereas the MST-NGR data appear to faithfully reflect the broad features of the core at the scale of meters, they also record features on the scale of tens of centimeters. Figure F92 shows the NGR data within a single section of core (Section 185-801C-14R-3), alongside a digital photo of the core. Again, the variations in the MST-NGR values closely follow the lithologic changes in the core, from low count rates in the basalt at the top of the section to high count rates in the green interpillow material at the bottom of the section. This section was also subsampled for AA K2O analyses, and the five analyses generally confirm the NGR variations. The AA analyses are on small samples (a few 100 mg or <1 cm3), and, thus, do not sample the core at the same length scale as the MST. The MST measurements are taken every 10 cm with a footprint of at least ±10 cm and possibly as great as ±14 cm from the point of measurement (Hoppie et al., 1994). To calibrate MST-NGR data at a similar sampling rate, the AA analyses were used to generate a synthetic data set at 10-cm intervals. This was accomplished by considering the length scale over which the AA sample was representative. For example, we assume the sample at 60 cm is typical of the unaltered basalt from 10 to 90 cm and that the interpillow sample at 70 cm only constitutes ~25% of the interval sampled by the MST from 60 to 80 cm, with the remainder being unaltered basalt. Table T24 gives the mixing proportions for each interval and resulting K2O estimates for the entire section. Although there are differences in detail, in general, the MST-NGR trend closely corresponds to the estimated K2O. The NGR shows a broader peak in the lower interpillow material and does not resolve small features such as the interpillow section at 70 cm. There are also decreases in the NGR counts, for example, at ~110 cm, which could result from decreases in core volume caused by rubbly intervals and spacers. Ideally, one should correct for volume changes, although Hoppie et al. (1994) found no significant decrease in count rate for as much as a 20% volume reduction in the core. Despite these effects, the first-order correspondence between the MST-NGR and K2O measurements is sufficient to provide a calibration.

Calibration of Multisensor Track-Natural Gamma Radiation Data

To provide better NGR data for calibration, we passed Section 185-801C-14R-3 through the MST several times to improve the counting statistics. Because this test was run after the core was split, the count rates were no longer directly comparable with original whole-round core data. Fortunately, there is an excellent relationship between the half-round and whole-round cores (Fig. F93A), with the slope being roughly 50%, in accord with the relationship between volume percent of core and count rate found by Hoppie et al. (1994). We passed the section three times under normal parameters (10-cm interval with 20-s counts at each interval) and twice with longer counting times (10-cm interval with 60-s counts). The percent standard deviation on the three 20-s measurements averages 5%, which is better than the precision on the AA analyses (~10%). Figure F92 shows the average of the 20-s and 60-s runs, which agree well with each other and the K2O data. These data form the basis of the calibration.

Ideally, the calibration should include homogeneous intervals of known K, U, and Th composition with the same matrix as Leg 185 Hole 801C cores (tholeiitic basalt), which could be measured with good precision on both the MST and by AA. Furthermore, it would be advantageous to be able to resolve the K, Th, and U contributions to the total gamma-ray data. Since neither of these conditions is met, such a calibration is necessarily imperfect. Homogeneous sections in the tholeiites exist only for the unaltered domains, and although there is good sample control at the low NGR range, count rates are low and precision is worse. On the other hand, high NGR regions in the tholeiites are generally small and heterogeneous and, therefore, difficult to characterize analytically. The alkali basalts at the top of Hole 801C could provide homogeneous, high NGR samples for calibration, but because they contain abundant K, U, and Th, they are not a good match for the tholeiites. In general, alteration of the tholeiites involves predominantly K enrichment, with sporadic and sometimes unrelated U enrichment and little net change in Th (Staudigel et al., 1995; Castillo et al., 1992). This relationship is also reflected in the downhole gamma-ray logs, which resolve K, U, and Th contributions (Fig. F70). In the Leg 185 section, most of the NGR signal is attributed to K variations, with U increasing in a few intervals, and Th low in all the sections analyzed. Shore-based U and Th analyses on the cores will test this interpretation. Thus, although not ideal, our calibration assumes that the MST-NGR variations are largely because of K variations and that the variations observed in AA data in the tholeiites in Section 185-801C-14R-3 can be used to form the basis for a calibration. We augmented the data from Section 185-801C-14R-3 with a few other key intervals (long, minimally altered regions of Cores 14R and 25R, which is the core with the lowest NGR signals in the hole).

Figure F93B shows the calibration obtained, using the data shown in Figure F92 and in Table T24, with all MST-NGR data corrected to whole-round counts. The R2 value on the calibration is very good (0.92), after removing a point at Sample 185-801C-14R-3, 80 cm, for which we do not have good control on the K2O concentration (there is no measurement nearby, and its proximity to the large interpillow zone may lead to significant alteration and concomitant increase in K). Uncertainties on the slope and intercept are shown by the gray region (see Fig. F93B). In theory, the intercept should be zero if blank subtraction was performed correctly. However, there was no true blank for most of the points in the calibration, which were calculated from the half-round-whole-round relationship in Figure F93A. The blank on the MST-NGR is measured by analyzing at the beginning of the leg a long, whole-round liner filled with distilled water, which had no K, Th, or U, and shielded random background gamma rays approximately the same way as the rock whole rounds. These blank measurements are not strictly valid for the half-round data; thus, the calibration was not forced through the origin. For the low count rates (<5 cps) that are typical for most of the hole, the intercept contributes the greatest uncertainty in the calibration, exceeding NGR counting statistics and AA analytical errors. Uncertainty in the intercept leads to an error of ~±0.1 wt% K2O.

Statistical detection limits are usually calculated from uncertainty in the background, but because we have no appropriate background measurements, we assessed working detection limits. It is clear from Figure F93B that even where many repeat measurements are available (as for Sections 185-801C-14R-4 and 14R-1 and Core 25R), it is impossible to resolve 0.1 wt% from the uncertainty in the intercept; thus, the detection limit is >0.1 wt% K2O. We can also assume Core 185-801C-25R represents near background NGR, at 0.08 wt% K2O (as in the core analysis). NGR values for Core 185-801C-25R vary from -0.45 to 3.2, averaging 1 ± 1 cps standard deviation. Thus, count rates up to 3 cps are probably indistinguishable from background, and the corresponding K2O (~0.35 wt%) may be considered a working detection limit for individual 20-s measurements. Increasing counting times on the MST and taking more whole-round background measurements would improve the NGR detection limits.

Results

The MST-NGR calibration can be used to obtain some useful estimates of the K2O content of individual sections (such as Section 185-801C-14R-3), individual cores (such as 14R) and the entire Hole 801C tholeiitic section (the calibration is probably not robust for the alkalic section, where significant U and Th also contribute to the gamma-ray counts). The average MST-NGR counts for Section 185-801C-14R-3 (Fig. F92; Table T24) give an average K2O of 0.87 wt%, with 7% relative standard deviation for the three standard (20 s) passes through the MST. This means that in a typical MST run, the precision on the NGR average could be ~7%. The absolute value of the MST-NGR K2O concentration is higher than the average calculated from the actual core samples analyzed by AA (0.8 wt%). The difference between the two should be minimal because this section was used to calibrate the MST. The difference probably results from the high NGR counts around 80 cm in the section, which were excluded from the calibration. For this section either the K2O is higher than estimated from the sparse AA data, in which case the AA average should be >0.8 wt%, or the interval contains significant U, which would lead to false high K2O in the NGR counts. Nonetheless, the agreement is still within the 0.1 wt% K2O error on the calibration, and the section is extremely enriched in K2O (0.8-0.87 wt%) with respect to the unaltered rock (<0.1 wt%). In fact, Section 185-801C-14R-3 is one of the more gamma-ray enriched sections in the hole (Fig. F94). The NGT logging data for K2O in the depth interval corresponding to Section 185-801C-14R-3 has higher K2O than either the MST or AA averages and may reflect a lack of recovery of potentially K-rich interpillow zones.

Comparison between the calibrated MST-NGR data and the AA/XRF core data for the entire Core 185-801C-14R shows even better agreement, both at 0.45 wt%. Again, the NGT log data has higher K2O values (0.53 wt%), which could reflect a recovery bias or a calibration shift in the logging data. In fact, at each scale, from section to core to entire hole, the logging data consistently give higher K2O values than the MST data. Assessing the calibration of the logging data will be an important postcruise activity, and the MST data will form a critical tie point between the log and core data.

Applying the MST calibration to all sections downcore reveals three zones of high K2O: one below the upper hydrothermal unit (II), one centered around the lower hydrothermal unit (V), and one coinciding with the large breccia unit (VII) in Core 185-801C-37R. The most K2O-enriched zone is the upper one, where values extend as high as 2 wt% in the deep green celadonite-rich rocks in Core 185-801C-5R. The lower two zones peak at ~1 wt% K2O. A minimum in K2O occurs at ~720 mbsf, where there is a major lithologic boundary between flow-dominated and pillow-dominated cooling units and abundant fresh glass. These periodic high and low K2O abundances downhole correspond to altered and less altered intervals, which contrast with a typical decrease downhole in K2O that has been observed at Sites 504, 417, and 765 (Alt et al., 1986; Staudigel et al., 1995; Gillis et al., 1992). Perhaps the pattern of alteration at Hole 801C, with focused hydrothermal alteration zones, is a feature more typical of fast-spreading oceanic crust.

The bulk K2O calculated from the MST-NGR data for the entire tholeiitic section is 0.31 ± 0.1 wt% (Table T25; Fig. F94). At face value, it would seem that the average for the entire site is below the working detection limit and, thus, unresolvable from zero. The working detection limit applies only to single measurements, however. Because the tholeiite average includes many significant measurements at high cps, which will exert the most influence on an average, its value has significance. One way to test this is to assume that all MST measurements below 3 cps are indistinguishable from the fresh rock value of 0.08 wt% K2O, whereas those above 3 cps are a significant measure of K2O. The average calculated this way is 0.33 wt% K2O, which is within error of the 0.31 wt% value determined from the simple average of the MST data.

Our estimate of the bulk K2O in the tholeiitic section from the MST data (0.31 wt%) as well as the bulk K2O from the logged interval (0.36 wt%) both exceed that calculated from the percentage of discrete alteration veins, halos, breccias, and interpillow material logged by the alteration team and a limited number of XRF analyses of these domains (0.13 wt% K2O; see "Alteration Geochemistry" and Table T10). The difference may lie in part in the assumption that everything that was not a discrete alteration feature was fresh rock and that 97% of the tholeiite section has 0.08 wt% K2O. This does not take into account any pervasive alteration of the rock; thus, this estimate is a minimum. Our MST estimate would require that the 97% of the core that did not contain patent alteration features has 0.27 wt% K2O. An average K2O of 0.31 wt% is lower than that calculated for Deep Sea Drilling Project (DSDP) Site 417 (0.56 wt%) (Staudigel et al., 1995). The technique laid out in this report could be used to calculate bulk K2O at the few other sites drilled deeply into basement (Holes 504B, 765C, 332), to start to form a better understanding of the controls on seafloor alteration fluxes.

NEXT