Table T10 is a compilation of shipboard and shore-based geochemical analyses for samples from Legs 118 and 176. Shore-based data for Leg 118 samples (0-500.7 mbsf) were obtained at Justus-Liebig Universität, Giessen, Germany (labeled "G" in Table T10), whereas those for Leg 176 samples (500.7-1500.8 mbsf) were generated at the GeoForschungsZentrum, Potsdam, Germany (labeled "P" in Table T10). Additional shore-based data were obtained at the Katholieke Universiteit Leuven, Belgium, and these are labeled "L" in Table T10. Those samples labeled "GL" in Table T10 indicate that the rare earth element (REE), Hf, Ta, Th, and U, analyses were carried out at Leuven, whereas the remaining data are from Giessen. During Leg 176, 186 samples were analyzed aboard ship for major oxides and the trace elements V, Cr, Ni, Cu, Zn, Rb, Sr, Y, Zr, and Nb. Of these, 180 were from the Leg 176 core (labeled "S" in Table T10) and 6 were from the lower 50 m of Leg 118 core (labeled "H" in Table T10).
Samples taken for shipboard analysis generally weighed between 20 and 30 g, but a few larger samples were taken from very coarse grained rocks. After coarse crushing, the samples were ground in an agate mortar (Leg 118) or in a tungsten carbide shatterbox (Leg 176). The powders were dried at 110°C and then ignited at 1010°C for several hours to determine the loss on ignition. 600 mg of ignited rock powder was then mixed with a fusion flux consisting of 80 wt% lithium tetraborate and 20 wt% La2O3. This mixture was then melted in a platinum mold to form a glass bead for major element analysis. Trace element analyses were carried out on pressed powder pellets prepared from 5 g of rock powder dried at 110°C. All analyses were carried out using the shipboard ARL 8420 wavelength-dispersive X-ray fluorescence (XRF) equipment. Full details of the analytical procedures can be found in Shipboard Scientific Party (1989). For major oxides, analytical error is <0.6 relative percent except for Na2O which has an error of 3.8 relative percent. For trace elements, the analytical error is <1.5 relative percent except for Rb (3.6%), and V (2.0%). After the Leg 176 cruise, all the shipboard samples were analyzed for ferrous iron at Leuven using the redox titration method of Shafter (1996).
A total of 220 samples from Leg 118 were analyzed for major and trace elements at Giessen, and 43 samples from Leg 176 were analyzed in Potsdam. Major element oxides were analyzed by XRF on samples prepared as fused disks of lithium metaborate (sample-to-flux ratio of 1:4 and 1:6 for "G" and "P" samples, respectively). Philips PW 1400 and Siemens SRS 303 AS computerized X-ray spectrometers were used, and commercial Philips and Siemens "alphas" programs calculated the concentrations. Ferrous iron was determined by manganometric titration ("G") and potentiometric titration according to Wilson's method ("P"). H2O+ analyses were performed by coulometric Karl-Fischer titration after thermal decomposition of the samples at 1200°C ("G") and with a LECO analyzer RC-412 ("P"). CO2 was measured coulometrically following decomposition at 1200°C ("G") and using a LECO RC-412 analyzer ("P"). Sulfur analyses were performed using LECO sulfur determinators SC 132 ("G") and CS 225 ("P"). Fluorine was analyzed with a fluoride-sensitive electrode, following a pyrohydrolytic decomposition method (Erzinger and Puchelt, 1982).
Determinations of Ba, Co, Cr, Cu, Ga, Nb, Ni, Zn, Sr, Y, V, Zn, and Zr were conducted by XRF on pressed powder pellets. The rhodium Compton peak of the X-ray tube was used for matrix corrections. Numerous international reference rocks were used for calibration of the XRF methods.
The REE were determined by optical emission spectrometry (OES) using an inductively coupled argon plasma for excitation (ICP-OES), following Na2O2 fusion and a chromatographic separation and concentration procedure. A detailed description of the method is given by Zuleger and Erzinger (1988).
Previous studies in our laboratory (e.g., on Hole 504B rocks) revealed that chemical data sets from various laboratories differ significantly because of different analytical accuracy and precision (see Boström and Bach, 1995, for a discussion). Therefore, we carefully tested our data with the reference data from Govindaraju (1994) and with data from other laboratories to ensure proper analytical accuracy. The precision of major and trace element determination was tested by multiple measurements on selected samples. The accuracy and precision were checked by carrying the international reference rocks BHVO-1 ("G"), BM ("P"), and the Leg 140 interlaboratory standard Bas 140 ("G") (Sparks and Zuleger, 1995) as unknowns through the whole procedure. The chemical results are given in Table T11 along with the analytical errors and recommended concentration values.
Assuming the analytical error is close to twice the standard deviation, the precision for the major elements is better than 1.5 relative percent and the precision for the trace elements is better than 10 relative percent, mostly better than 5. Further details on analytical accuracy and precision of the data are given in Bach et al. (1996) and Zuleger et al. (1996, 1995).
The samples analyzed at Leuven for major and trace elements were crushed and ground in agate mortars. Thus, neither Co nor Ta was affected by contamination from tungsten carbide.
For determination of the major elements at Leuven, 100 mg of sample powder was fused with 0.5 g Li metaborate in a graphite crucible by heating in a muffle furnace. The fused glass was then dissolved in dilute nitric acid. The concentrations of the elements Si, Al, Ti, Fe, Mg, Ca, and P were measured using a direct-current plasma atomic emission spectrometer (DCP-AES) (Bankston et al., 1979). The alkali elements Na and K were analyzed by flame atomization atomic absorption spectrometry (AAS). The analytical precision of the major elements is estimated at 2%-3% (relative) for the elements Si, Al, Fe, Mg, Ca, and Na. Precision for the minor elements Ti, Mn, and K varies from 5% to 15% (relative) depending upon their concentration.
The trace elements Sc, Cr, Co, REE, Hf, Ta, Th, and U were determined by INAA using 800 mg of sample powder. Samples were irradiated in the Thetis reactor of Ghent University, Belgium. The induced gamma ray activities were measured at Leuven using extra-high-resolution small and large Ge detectors. For details of the procedure and of the accuracy and precision obtained, see Hertogen and Gijbels (1981) and Pedersen and Hertogen (1990). The samples were analyzed relative to a secondary in-house standard rock that has been repeatedly calibrated against international reference rocks. The analytical precision of the trace elements is variable; generally better than 5% (relative) for Sc, Cr, Co, Sm, and Eu, between 5% and 10% for La, Tb, Yb, and Hf, and between 5% and 25% for Ce, Nd, Lu, and Ta, depending on their concentration.
Table T10 summarizes the chemical compositions of the 451 rock samples analyzed for this study. The values for Rb and Nb (<5 ppm) and Ba (<20 ppm) were below the detection limit of the methods used and are not reported. In the XRF analyses, total iron is expressed as Fe2O3t. For all shore-based analyses, FeO was determined independently and is reported in Table T10. Two calculated values for the Mg# (100Mg+2/[Mg+2+Fe+2]) are also given in Table T10. Where FeO was determined independently, the analyzed value was used in the calculation. Where only Fe2O3t is available, Mg#s were calculated by taking 80% of the total iron as FeO. Shipboard calculations of Mg#s assumed FeO to be 85% of total iron, but we found that a value of 80% better replicates the Mg#s produced for samples in which FeO was determined independently.
Table T12 presents the average chemical composition of each lithologic unit based on the data in Table T10. Where data are missing for some samples (e.g., V), the averages were recalculated to take into account the missing values.
Downhole variations in bulk rock chemistry are presented in the figures described in the following sections to illustrate the chemical variability of the rocks in Hole 735B.
SiO2 is relatively constant downhole, averaging just over 50 wt% (Fig. F7). The largest variations occur in lithologic Unit IV, which consists of Fe-Ti oxide-rich gabbro, and Unit VI, which consists of interlayered olivine gabbro and troctolite. The other analyses that deviate significantly from the average reflect mostly alteration or the addition of silicic material in the sample. There is very little scatter of points in the lowest 500 m of the core where the rocks are very fresh.
Al2O3 shows much greater downhole variability than SiO2 but with a very slight increase in the average abundance with depth (Fig. F8). Again, Units IV and VI deviate the most from the average value because they have the most extreme compositions. Because plagioclase is the major Al2O3-bearing phase, the scatter in Al2O3 reflects small-scale variations in plagioclase abundance and composition. Also, plagioclase is susceptible to variable degrees of alteration, which affect the bulk-rock geochemistry.
Fe2O3t and TiO2 vary in a sympathetic fashion downhole, increasing markedly in Units IV and VI (Figs. F9, F10). To be consistent, the Fe2O3t plot includes both calculated (Fe2O3 + FeO x 1.111) and measured values. Positive spikes reflect primarily thin bands of oxide gabbro commonly associated with narrow shear zones. Note that the average values of TiO2 decrease very slightly downward in Unit V and from the top of Unit VII to the bottom of the hole.
MnO shows limited variation in the hole and varies sympathetically with Fe2O3t (Fig. F11). There is a significant increase in Units III and IV and a number of small spikes lower in the hole corresponding to Fe-Ti oxide-rich bands. MnO is relatively constant below the top of Unit VII, and like TiO2, the average MnO value shows a slight downward decrease from this point.
MgO values vary widely, both within and between individual lithologic units (Fig. F12). For example, in Unit II MgO contents range from 4 to ~23 wt% and in Unit X from 2 to ~23 wt%. These variations appear to reflect both lithologic character and degrees of alteration. The lowest average values are in Unit IV, composed largely of Fe-Ti oxide gabbro, and the highest average values are in the olivine gabbros and troctolites. There appear to be distinct compositional breaks in MgO content at ~1160 mbsf in Unit XI and at the upper and lower contacts of Unit IV.
Figures F13 and F14 are both plots of Mg#s. Figure F13 contains data only for those samples in which FeO was actually measured, whereas Figure F14 (Mg# = 80) presents Mg#s calculated by assigning 80% of the Fe2O3t as FeO. Mg#s reported in the shipboard report (Shipboard Scientific Party, 1999) were calculated based on the assumption that FeO was equal to 85% of Fe2O3t, but we find that our calculations correspond more closely to those in which FeO was determined independently. Using a value for FeO of 80%, rather than 85% of total iron, doesn't change the shape of the plot; it merely shifts everything uniformly to higher numbers. The relatively high proportion of Fe2O3 in these rocks probably reflects local secondary oxidation.
The Mg#s show considerable variation downhole commonly with distinct breaks at lithologic unit boundaries. For example, Unit XII shows a small upward decrease in Mg#s from ~82 at the base to 75 at the top with a few values even lower. At the base of Unit XI, Mg#s shift to slightly higher values and then decrease upward very much like those in Unit XII. Values in Unit X are more scattered but again are highest at the base and decrease somewhat upward, particularly near the top of the unit. The distinctly low values in all the units correspond to Fe-Ti oxide bands or to felsic units within the host rock. Units IX and X also show some distinctly low Mg#s, but no overall trends are apparent. Unit VII has only a few values, but they appear to decrease upward. A distinctly different pattern is apparent in Unit VI where there is an upward trend to higher numbers. Unit V is characterized by a tight cluster of values, which again show a marked upward decrease from ~84 at the base to 76 at the top. As expected, Unit IV, rich in Fe-Ti oxide gabbro, has the lowest average value in the hole and is characterized by sharp breaks in composition at both the upper and lower contacts. Unit III has two separate clusters of values, one at ~85 and the other at ~70. Again, there is a sharp break in composition at the boundary between Units II and III. Although there is some scatter in Unit II, most values cluster around 81-82 with a small decrease in the upper 15-20 m. Unit I has a cluster of values between ~63 and 72 and another smaller one between 80 and 82.
Overall, there is a small increase in Mg#s downward in the lower part of the hole, particularly in Units X, XI, and XII, as the rocks become more olivine rich with depth. The cyclic nature of the variations in Mg#s, particularly the upward decrease in some units, has been interpreted as reflecting fractionation of several distinct magma bodies (Shipboard Scientific Party, 1999; Dick, Natland, Miller, et al., 1999).
Both CaO and Na2O exhibit considerable scatter in the upper parts of the hole but become more uniform at depth (Figs. F15, F16). The abundance and composition of plagioclase are the main controls on CaO and Na2O, and the relative proportions are highly sensitive to zoning and alteration. As expected, the downhole trends are nearly mirror images of one another. This is particularly noticeable in Units II, III, IV, and V. In the lower part of the hole, from approximately the top of Unit VII to the base of Unit XII, there is a small increase in the average content of CaO and a corresponding decrease in Na2O.
K2O is relatively uniform in the hole with a very small downward decrease in abundance (Fig. F17). The K2O resides largely in alkali feldspar and mica, and the few relatively high values correspond to felsic patches which contain both minerals.
Apatite is the only significant phosphorous-bearing mineral in Hole 735B and its occurrence is closely related the Fe-Ti oxide gabbros. Background levels of P2O5 are very low (<0.1 wt%) (Fig. F18), and the few spikes correspond to oxide-rich bands.
H2O was measured on only 10 shipboard samples, so there are relatively few data available for the lower part of the hole (Fig. F19). In the upper part of the hole to ~400 mbsf, there is a steady decrease in the average water content, reflecting the downward decrease in metamorphism and alteration. The few high values in the lower parts of the hole reflect local zones of alteration.
As with H2O, most of the data for CO2 and S are from the upper 500 m of the hole. CO2 has a narrow range of composition, with only two samples having >1 wt% (Fig. F20). The actual amounts are probably somewhat higher because most of the samples were selected to avoid veins and highly altered zones. It is clear, however, that the background level is relatively low and constant at ~0.1 wt% (Fig. F20). The few higher values reflect the introduction of carbonate along small cracks and fractures. S contents in most of the rocks are also low (generally <0.1 wt%). However, S is significantly enriched in Unit IV, which consists of Fe-Ti oxide gabbro (Fig. F21). This accords with the common presence of minute amounts of pyrite and sulfides in the Fe-Ti oxide-rich gabbros. The other spikes in S content shown in Fig. F21 also correlate with Fe-Ti oxide-rich zones.
Vanadium has a relatively constant background concentration of ~150 ppm in gabbros and olivine gabbros but increases significantly in the Fe-Ti oxide gabbros. Thus, Unit IV has strong V enrichment with most values between 300 and 1300 ppm and one value above 2000 ppm (Fig. F22). Similar scattered high values in the other lithologic units correlate with thin bands of Fe-Ti oxide gabbro. As expected, V contents correlate well with variations in TiO2.
Ni resides mostly in olivine with only minor amounts in pentlandite. Thus, Ni contents vary significantly downhole with maximum values (900 ppm) in Unit VI (olivine gabbro and troctolite) and minimum values in Unit IV (Fe-Ti oxide gabbro). Elsewhere in the core, Ni contents generally lie between ~50 and 200 ppm but show significant scatter. A distinct break in Ni contents occurs at ~1240 mbsf in Unit XI and at several lithologic unit boundaries (Fig. F23).
In Hole 735B, Cr resides chiefly in clinopyroxene and thus varies significantly downhole. Cr contents are particularly high and variable in the upper 550 m (Units I-VI), ranging up to 1000 ppm (Fig. F24). In the lower part of the hole, Cr contents are mostly <400 ppm, although they still exhibit considerable variability. Unit XI exhibits an unusual pattern with relatively high and scattered values at the base and very low, tightly clustered values at the top (Fig. F24). There are distinct boundaries in Cr contents between several lithologic units, particularly between Units IV and V, IX and X, and X and XI.
Zn contents in Hole 735B rocks are generally low, with most values between 10 and 50 ppm. However, a number of higher values (up to 260 ppm) are present locally, and these correlate with zones of Fe-Ti oxide gabbro. This correlation is well illustrated in Unit IV, where Zn values range from 100 to 260 ppm (Fig. F25). The other Zn spikes likewise correlate with thin bands of Fe-Ti oxide gabbro scattered through the core. Values are very consistent in Unit V but show a slight downward decrease. A small increase occurs at the top of Unit VI, and the average Zn values increase to the base of Unit VIII. Below this level, the background level is relatively constant (~40 ppm) to a depth of 1090 mbsf within Unit X, at which point it decreases to ~30 ppm and then increases again to 60 ppm at 1150 mbsf. The average value then decreases regularly to ~20 ppm at the base of the hole.
Copper is relatively low in Units I and II (generally <70 ppm) and highly variable in Unit III (Fig. F26). There are no obvious trends in the remainder of the hole. Although individual values are quite variable, the average value is ~80 ppm.
As incompatible elements, Y and Zr have similar downhole trends (Figs. F27, F28). Background levels are low for both elements (10-15 ppm), and both show relatively little scatter. Units IV, VI, VII, IX, and X have increased values for both elements (up to 150 ppm Y and 600 ppm Zr). These units have somewhat higher percentages of felsic material containing rare crystals of zircon. There is also a significant spike in both elements in Unit XI at ~1040 mbsf, which correlates with a zone of Fe-Ti oxide gabbro.
Strontium varies widely downhole, from ~15 to 260 ppm, but exhibits no obvious trends (Fig. F29). It resides primarily in plagioclase and correlates reasonably well with CaO.
The downhole variation of Sm is shown in Figure F30. The Sm content is a good measure of the overall REE concentration levels. The sampling density within the upper part of Hole 735B is rather high, and the analyzed samples probably cover the whole concentration range.
The total variation of absolute REE contents is quite high and is accompanied by rather large fractionation of light (LREE) from heavy (HREE) rare earth elements. This is illustrated by the downhole variation of chondrite-normalized La/Yb ratios (Fig. F31). The majority of the samples have normalized La/Yb ratios less than unity, a feature that the gabbroic rocks inherited from their LREE-depleted parental melts.
Figure F32 shows the chondrite-normalized REE patterns of representative Leg 176 samples. This sample subset illustrates the overall variation of the REE content and the shapes of the REE patterns in Hole 735B rocks. The lowest REE concentrations occur in the troctolitic rocks, which are characterized by low abundances (due to a low fraction of trapped liquid component), relatively flat chondrite-normalized patterns (due to low clinopyroxene content), and pronounced positive Eu anomalies (due to selective uptake of Eu+2 in anorthite).
Variations of REE abundances in the gabbroic (senso stricto) rocks reflect the variable amounts of trapped intercumulus liquids. The trapped parental liquid, which does not have a significant Eu anomaly, smooths the positive Eu anomalies of plagioclase and the low La/Yb ratios typical of clinopyroxene.
A number of the Hole 735B samples show a pronounced overall enrichment of REE. The majority of these samples are enriched in a strongly fractionated, plagiogranitic residual component (melt or fluid). Because of the very high REE content, these residual liquids are important components in the REE mass balance of the gabbroic section, in spite of the low volumetric contribution these rocks make to the plutonic pile.
The overall REE geochemistry is summarized in Figure F33 (La/Yb vs. Yb). For more details on this type of diagram, see also Barling et al. (1997). The REE content of a large fraction of the samples can be explained by a three component mixing of cumulus plagioclase (high La/Yb and low HREE), cumulus clinopyroxene (low La/Yb and higher HREE), and relatively unfractionated trapped melt. Most of the samples seem to converge to a liquid composition similar to the composition of the diabase dikes cored in the upper part of Hole 735B (e.g., Samples 176-735B-23R-4, 66 cm, to 23R-5, 21 cm). This suggests that these diabases may be fairly representative of the parental magma of the Hole 735B plutonic pile and that the parental magmas were rather uniform in composition. The oxide-rich gabbros are mainly positioned in the lower right quadrant of the figure as expected from their formation from more evolved liquids having a higher Yb content.