Web hosting Custom Email SiteBuilder

Iron Content and Oxidation State

The oxidation state of the rocks at Ok Tedi and of the magmas that crystallized to form these rocks are estimated in this section by reference to chemical analyses and to the presence of specific minerals in the modes of the rock samples.

Igneous rocks can be categorized as iron-poor or iron-rich by the ratio of their iron to magnesium oxide contents (Keith and others, 1991). These authors suggest that the amount of iron in a magma is more a function of the method by which a region of the crust or mantle is melted than a function of the composition of the source region. They attribute iron-poor magmas to hydrous melting and iron-rich magmas to adiabatic or thermal melting under anhydrous conditions. This topic will not be considered in detail in this dissertation, nonetheless, one of their diagrams is useful in characterizing the chemistry of the Ok Tedi intrusions. The iron contents of volcanic and plutonic rocks from the Star Mountains region are portrayed in Figure 79A, a binary plot on which the ratios of total iron (FeO^*) to MgO are plotted versus SiO₂ content. Curves drawn on this diagram separate fields of strongly iron-rich, moderately iron-rich, weakly iron-rich, moderately iron-poor, and strongly iron-poor rocks. Most samples from the Star Mountains plot near the boundary between weakly iron-rich and moderately iron-poor igneous rocks. Samples from the Ok Tedi Intrusive Complex are shown in Figure 79B using symbols that provide information on their texture and normative ferromagnesian mineral content. Most of those samples with diopside or diopside and hypersthene in their norms, the least-altered samples, fall in the field of weakly iron-rich rocks (area mineral content. Most of those samples with diopside or diopside and hypersthene in their norms, the least-altered samples, fall in the field of weakly iron-rich rocks (area enclosed by an ellipse in Figure 79B). Most of those with hypersthene as the only normative ferromagnesian mineral and with albite (or oligoclase) as their normative plagioclase feldspar, the potassically-altered rocks, are weakly iron-rich to strongly iron-poor (that is they are displaced in the direction of the arrow shown in Figure 79B). A few of the samples are moderately iron-rich; these have either andesine or oligoclase as their normative plagioclase feldspar. The deviation of iron content in altered rocks from the values of their unaltered equivalents can be attributed to redistribution of the metal during hydrothermal alteration and to dilution by increased SiO₂ and K₂O contents.

Two additional diagrams illustrating the iron content and the oxidation state of iron in rocks of the Ok Tedi Intrusive Complex are given in Figures 79C and D. Values of total iron (as Fe) are plotted against normative An content in Figure 79C. The three least-altered samples (areas a and b in Figure 79C) have similar total iron (about 3.25 percent). Total iron content remains approximately constant in almost all samples with anorthite content greater than about An₂₀ and with normative diopside or with both diopside and hypersthene. The variance of the iron content of these samples is relatively low. Samples with normative anorthite lower than An₂₀ and with hypersthene as the only normative ferromagnesian mineral are much more variable in iron content.

A plot of Fe₂O₃/FeO versus normative anorthite is given in Figure 79D. The ferric/ferrous ratios of samples with normative diopside and with andesine as the normative plagioclase feldspar are close to 2.6. This ratio is somewhat lower (~2.3) in samples with normative diopside and hypersthene and with oligoclase as the normative plagioclase feldspar. The variance in the ferric/ferrous iron ratio is relatively high in samples with anorthite content below about An₂₀ and with hypersthene as the only normative ferromagnesian mineral. The increased variance of Fe^* and Fe₂O₃/FeO at low anorthite contents also reflects the redistribution of iron and the dilution of the iron content by additions of SiO₂ and K₂O resulting from the metasomatic effects imposed on the rocks that have undergone potassic alteration.

Oxygen fugacity at the time of crystallization of the igneous rocks cannot be directly evaluated from the ferric/ferrous iron ratio of rock samples (Hall, 1996) because this ratio is dependant on bulk composition and is also affected by constituents other than iron and oxygen (Sack and Carmichael, 1980; Dickenson and Hess, 1986). Further, this ratio can be affected by subsolidus processes such as weathering which is usually an oxidation process and that generally leads to an increase in the ratio of ferric to ferrous iron. Instead, the presence of specific oxide or ferromagnesian silicate mineral assemblages are considered to more accurately reflect the oxidation state of the magmas which form igneous rocks than is the ratio of ferric to ferrous iron. The minerals (or mineral assemblages) that define the oxidation state of magmas are called buffers after their use in experimental petrology to control oxygen fugacities. The most common minerals used as buffers in petrologic experiments are hematite, magnetite, and wustite. These minerals are considered to define the range of oxygen fugacity at which magmas form because the majority of magmas are not sufficiently oxygen-deficient to contain wustite and not sufficiently oxygen-rich as to contain hematite (Hall, 1996). The stability relations of wustite, magnetite, and hematite are shown in Figure 80, which is modified from Buddington and Lindsley (1964). The reaction boundaries between quartz +magnetite and fayalite (Buddington and Lindsley, 1964) and between quartz + magnetite + sphene and ilmenite + hedenbergite (Dilles, 1987) are also given in Figure 80. These reactions are useful in constraining the oxygen fugacity of magmas that formed under conditions between the magnetite-hematite and magnetite-wustite boundary curves. The assemblage quartz + magnetite + sphene is common in unaltered samples with plutonic and volcanic textures from the Ok Tedi Intrusive Complex, and this assemblage defines the magmas from which these rocks were formed as relatively oxidized (Dilles, 1987; Wones, 1989).

Several of the minerals that are present in the massive ores and skarns at Ok Tedi reinforce the hypothesis that the intrusive and hydrothermal processes that formed the ore deposit were oxidizing in character. Epidote and andradite are minerals enriched in ferric iron and these are common in the endoskarns (and exoskarns?) at Ok Tedi. In addition magnetite, a relatively oxidized mineral, is the most abundant mineral in most of the massive ores.

The ratio of ferric to ferrous iron oxide can have a powerful influence on the nature and abundance of the normative minerals calculated for igneous rocks (Middlemost, 1989). Various authors have suggested different methods of assigning this ratio. Le Bas and others (1986) proposed that the Fe₂O₃/FeO ratio should be "taken as that assigned by the analyst." However, in chemical analyses where both oxides have not been determined as is commonly the case, they recommended that a method supplied by Le Maitre (1976) should be used to determine the ratio. Briefly, this method consists of adjusting the ratio of FeO to Fe₂O₃ to equal 0.93 - 0.0042 SiO₂ - 0.022 (Na₂O+ K₂O) for volcanic rocks and 0.88 - 0.0016 SiO₂ - 0.027 (Na₂O + K₂O) for plutonic rocks. Irvine and Baragar (1971) have suggested that the value of Fe₂O₃ should have an upper limit equal to the content of TiO₂ + 1.5 weight percent. Middlemost (1989) proposed a set of standard Fe₂O₃/FeO ratios for different volcanic rocks that he thought should be applicable to all normative mineral calculations (for example andesite and trachyandesite would be assigned the value of 0.35).

The normative minerals listed in Appendix 12 and the figures herein that are based on these minerals were calculated using the values of Fe₂O₃ and FeO that were supplied by Chemex Laboratories. Values for this ratio provided by direct chemical analyses for these iron oxides obtained in this study range from 0 to 8.6. However, calculations were also performed on several samples of least-altered rock using the ratio suggested by Middlemost (0.35). The principal effects of changing the iron ratio are as follows: (1) normative diopside, magnetite, quartz, and apatite become lower in abundance, (2) hematite and wollastonite are eliminated from the norm, and (3) hypersthene increases in abundance. The normative feldspars (An, Ab, and Or) are virtually unchanged in abundance. Although hematite has not been recognized in microscopic or microprobe analyses and in spite of the fact that the abundance of apatite in the norms that were calculated using standardized iron ratios more closely match the modal abundances of this mineral, these changes are considered objectionable because of their effect on normative pyroxene. Decreasing the diopside and wollastonite components in favor of an increase in hypersthene content would be particularly misleading because the pyroxene in unaltered Ok Tedi intrusive rock is, in fact, diopside as predicted from the normative mineralogy and because the composition of the pyroxene is an important factor in the characterization of the intrusions at Ok Tedi. In addition, the use of a standard value for the ratio of ferric to ferrous iron would camouflage the oxidized nature of the igneous rocks at Ok Tedi.

Volatiles

The amount of H₂O^- (surficial moisture) in the intrusive rocks from Ok Tedi ranges from 0.03 to 1.43 weight percent with an average of 0.34. H₂O⁺, which exists in mineral lattices as the H⁺ or OH^- ion and which is referred to as water of crystallizaion, ranges from 0.1 to 3.23 weight percent and averages 0.73. The amount of H₂O⁺ in the least-altered rocks from the Sydney, Kalgoorlie, and Ningi intrusions is less than 0.4 weight percent. Elevated H₂O⁺ contents in some samples of the highly-altered host rocks reflect the presence of relatively abundant hydrothermal micas and clays.

The sulfur content of the least-altered rocks of the Ok Tedi Intrusive Complex is generally less than 0.05 weight percent. A few samples that show little or no hydrothermal alteration are nonetheless crosscut by pyrite veinlets and these samples have elevated sulfur contents. The average sulfur content in more than 2600 samples from the strongly mineralized Fubilan Intrusion is 1.14 weight percent. The locations of the reverse-circulation rotary drill holes from which these samples were collected are given in Figure 56. A plot illustrating the range of sulfur values versus SiO₂ is given in Figure 82. The sulfur content of these samples varies from less than the limit of detection (0.01) to 10 weight percent. Relatively high sulfur values (greater than about .5 weight percent) are present throughout the range of silica contents but the highest values are in samples with about 65 weight percent SiO₂.

Analyses of fluorine in samples from the Ok Tedi Intrusive Complex range from a low of 240 ppm to 7100 ppm. The lowest values are in the least-altered rocks and those highly altered samples that contain little if any hydrothermal biotite, such as DDH 302-69. A map illustrating the distribution of fluorine values at Ok Tedi is given as Figure 83. Samples from the Sydney Intrusion contain relatively low levels of fluorine. Those from the Kalgoorlie and Ningi Intrusions contain both high and low values, reflecting the variable distribution of hydrothermal alteration in these areas. All but one sample from the Fubilan Intrusion contain moderate to high concentrations of fluorine. The distribution of fluorine values in Figure 83 clearly suggests that enrichment of this volatile element is an important factor in hydrothermal alteration and ore deposition at Ok Tedi.

Trace Elements

Rare Earth Elements (REE)

Johnson (1982) has described the chemical characteristics of andesites from Papua New Guinea and has included a figure with chondrite-normalized REE patterns of samples gathered from 10 locations where andesites had been identified. Their locations are shown in Figure 84. Two groups of andesite are distinguished and these correspond to two of three volcanic rock associations that have been identified in the arc and trench environments of Papua New Guinea. Group one andesites, present in the Highlands, Eastern Papua, Rabaul (New Britain), and Bougainville, are characterized by higher concentrations of alkalis and incompatible elements than andesites of group two. They show fractionated REE patterns in which light REE (LREE) are moderately or strongly enriched relative to heavy REE (HREE) as shown in Figure 85A, and high initial ⁸⁷Sr/⁸⁶Sr values from 0.7039-0.7049. Group two andesites are present in the volcanic arcs of the Bismarck Sea (excluding the Rabaul province) and are characterized by REE patterns that show little enrichment of LREE elements (Figure 85B) or may be LREE-depleted (Bamus). They are characterized by alkali and incompatible elements and initial ⁸⁷Sr/⁸⁶Sr ratios that are low (from 0.7034-0.7037) relative to those of group one. Hornblende phenocrysts are present in many andesites of group one but are uncommon in samples from group two. Biotite is present in some samples of group one andesites, particularly in trachybasalts and trachyandesites (Johnson, 1982). The third volcanic rock association in the region is a chain of islands east of New Ireland that is characterized by alkaline rocks. This association does not include andesites.

The chondrite-normalized REE contents of the least-altered intrusive rocks from the Sydney and Ningi Intrusions are listed in Appendix 13 and plotted in Figure 86A. These samples show strong enrichment of LREE to HREE with (La/Lu)_cn ratios of 14 to 17 and as such are similar to samples of Victory andesites (Eastern Papua) and Doma Peaks andesites (PNG Highlands) which have (La/Lu)_cn of 22 and 34 (Johnson, 1982). The patterns do not display Eu anomalies and therefore are similar to other rocks of andesitic composition (Henderson, 1984). The linear patterns of the four samples shown in Figure 86A form a tight bundle suggesting very little difference in the REE contents within or between the unaltered rocks of the Sydney and Ningi Intrusions.

The chondrite-normalized REE contents of altered intrusive rocks from the Ok Tedi Intrusive Complex are listed in Appendix 13 and plotted in Figure 86B. One sample of least-altered rock (DDH 340-166.1) is also included for comparison. All of the altered samples have normative anorthite contents that are less than An₆. All samples display the same LREE-enriched pattern as do their unaltered equivalents and all lack Eu anomalies. The REE patterns of all but one of these samples are depleted relative to DDH 340-166.1. The exception is JDD-94-07 which is similar in abundance to DDH 340-166.1 for most of the REE. The trend line representing sample DDH 334-216.2 shows the most displacement relative to DDH 340-166.1, even though this rock is not the most potassium-enriched or calcium-poor of the altered rocks shown in the diagram. These two factors suggest that although depletion of REE has occurred in the potassically-altered rocks at Ok Tedi, the degree of depletion cannot be simply related to the intensity of alteration.

Rubidium and Strontium

Rubidium and strontium are commonly present in trace to minor concentrations in intrusive rocks of intermediate to silicic composition. In samples from the Ok Tedi Intrusive Complex, rubidium ranges from 84 ppm in unaltered intrusive rocks to 280 ppm in samples with strong potassic alteration (Appendix 12). Values as low as 15 ppm characterize samples of endoskarn with significant quantities of calcite in their norms. Strontium ranges from 1200-1300 ppm in unaltered intrusive rock to as low as 230 ppm in samples with strong potassic alteration. The concentrations of these elements in samples with normative andesine, oligoclase, and albite are contrasted in Figure 87. In general the concentrations of these elements are consistent with and further reinforce interpretation of the trends of the major oxides. The behaviors of these elements reflect their position in the periodic table of elements. Potassium and rubidium are both group one elements and both are univalent. Calcium and strontium are both group two elements and both are divalent. Because they are similar in ionic radius and charge, rubidium commonly substitutes for potassium in the atomic structures of minerals, and strontium substitutes for calcium. Strontium is enriched in the least altered samples, which are also characterized by relatively high concentrations of calcium. Rubidium is enriched in samples having undergone strong potassium enrichment with potassic-alteration. Conversely, both strontium and calcium are strongly depleted in rocks that have undergone potassic alteration. The antipathetic relationship of K to Ca and Rb with respect to alteration-mineralization processes at Ok Tedi are expectable trends and consistent with those observed in porphyry systems elsewhere.

Gold and Copper

The elemental abundances of gold, in parts per million (ppm), and copper, in weight percent, of the samples collected for this study in 1994 are listed in Table 20. Part of the samples were analyzed by the Folomian laboratory at Ok Tedi by atomic absorption spectroscopy. All of these samples consist of diamond drill core. The core had earlier been split into two portions. One half of the core had been crushed and submitted for analyses; the other half had been archived. Hand specimens, typically 10 to 20 cm in length, were collected from the archive boxes and used for the thin-section and XRF analyses of this dissertation. The crushed rock submitted for gold and copper analyses at Folomian laboratory typically consisted of composites of two to three meters of core. The composite lengths for each of these samples are listed in Table 20. The metal values listed in Table 20 for these samples are representative of the intervals from which they were taken and are not necessarily representative of the hand samples used for thin-sections and major-oxide determinations. However, the hand samples were chosen to be representative of the intervals from which they were taken and, therefore, the composite value are considered to be suitable for presentation. The other samples in Table 20 were not analyzed at Folomian because they either came from unaltered drill core, which the geologic staff of the mine did not deem appropriate for analyses, or because they were collected as hand specimens in 1994. These samples were analyzed by ICP spectroscopy at Chemex Laboratories. They are considered to be representative of the hand specimens from which thin-sections and major-oxide determinations were taken.

Statistical distributions of major elements (or oxides) in rocks are characterized by histograms that are symmetrical about the mean values of sample populations. An example of such a normal distribution is given in Figure 88A. This plot illustrates the K₂O content of 2638 samples of reverse-circulation rotary drill cuttings from the Mt. Fubilan Intrusion. Curves derived by smoothing such histograms are bell-shaped, such as the one illustrated in Figure 88B, and they are considered to define normal distributions. The curves representing normal distributions can be converted to straight lines by plotting sample values versus cumulative probability values (Tennant and White, 1959; LePeltier, 1969; and Sinclair, 1976). A normal-probability plot for the K₂O values portrayed in Figure 88 is given in Figure 89. In contrast to the behavior of the major oxides, statistical plots of trace elements typically display skewed histograms that result in curves that are displaced towards low values (Ahrens, 1954, 1957). Histograms having non-symmetrical shapes can commonly be modeled mathematically using logarithmic distribution statistics. Such distributions are made symmetrical by plotting sample frequency versus the logarithms of sample values rather than the sample values themselves. An example of a histogram for a log-normal distribution in which cumulative sample frequency is plotted against the logarithms of sample values is given in Figure 90A and the curve derived from this histogram appears as Figure 90B. This plot is of the gold content in samples of 4533 blast hole cuttings from the Fubilan Intrusion. The curve resulting from smoothing the histogram is approximately bell-shaped and its log-probability plot (Figure 91) approximates a straight line.

Several factors can affect the symmetry of histograms and can cause deviations from straight lines on probability plots. These factors include: 1) the inclusion of excessively large numbers of high (or low) values in a data set, 2) the presence of more than one population of sample values, and 3) sample populations that are neither arithmetic or logarithmic. Average values and standard deviations of sample populations can be read directly from probability plots. The average values derived from normal-probability plots are referred to as arithmetic means and those from log-probability plots as geometric means.

The arithmetic and geometric means and their respective standard deviations of gold and copper values in the 1994 suite of samples together with similar statistics for 1000 samples of rotary drill cuttings, 2050 samples of drill core, and 4300 samples of blast hole cuttings are listed in Table 21. These statistics are graphically summarized in Figures 92-93.

The diamond drill core and hand samples collected in 1994 have an arithmetic mean of 0.30 ppm and a geometric mean of 0.19. About two percent have gold values larger than 0.9 ppm (or grams/tonne) and about one percent have values less than 0.015 ppm The mean values of this set of samples are similar to the means of gold in diamond drill core (0.33, 0.23) and about one-half of the means of rotary and blast hole drill cuttings (0,50-0.58, 0.33-0.36). The lower average values of the 1994 suite of samples and diamond drill core relative to those of rotary and blast drill cuttings are attributed to the fact that the former include substantial numbers of unaltered rocks from the Sydney, Kalgoorlie, and Ningi Intrusion whereas the latter are nearly all from the potassically-altered Fubilan Intrusion. The sample distributions of gold in rotary and diamond drill hole samples are portrayed on a log probability plot in Figure 92A. The gold values contained in samples of blast hole cuttings are summarized in Figure 92B. The data base of samples from blast hole drill cuttings includes details as to the presence, or absence, of sulfide minerals or limonites in the samples The average values and trends of sample sets containing sulfide minerals (S; chiefly pyrite, chalcopyrite, and chalcocite), limonite (X), and mixtures of sulfide minerals and limonite (M) are nearly identical (Figure 92B). The similarity of statistical values and trend lines for these samples on the log-probability plots suggest that the distribution of gold values has not been affected by supergene processes.

The arithmetic mean for copper in the 1994 suite of samples is 0.24 weight percent and the geometric mean is 0.14. About three percent of the samples have greater than one weight percent copper and approximately one percent have less than 0.015. The trends of copper content in diamond drill and rotary drill cuttings are approximately the same as one another in Figure 93A. The arithmetic and geometric means for the samples of diamond drill core are 0.40 and 0.21 and those of rotary drill cuttings are 0.32 and 0.15 weight percent. Data from samples of blast hole drill cuttings for samples containing sulfide minerals, limonite, and mixtures of sulfide minerals and limonite are plotted separately on log-probability diagrams in Figure 93B. These diagrams illustrate the fact that ore-grade copper is restricted to rock that contains sulfide minerals or mixtures of sulfides with limonite. The arithmetic and geometric means of copper in samples containing sulfide minerals with little if any limonite are 1.01 and 0.70 weight percent. The arithmetic and geometric means of samples containing both sulfide minerals and limonites are 0.72 and 0.45 weight percent. The rock samples with the highest grades come from areas where supergene chalcocite has replaced chalcopyrite and pyrite thus enriching the hypogene ore. Samples from rocks that have been so heavily oxidized as to destroy all sulfide minerals do not constitute an ore of copper. The arithmetic and geometric means for copper in "oxide" samples are 0.13 and 0.06 weight percent. Less than ten percent of oxide samples carry ore-grade concentrations of copper and most of this is contained in malachite, crysocolla, or cupiferous limonites. These minerals are not recovered in the flotation circuits at Ok Tedi and do not, therefore, constitute an economic asset.

Variations in the gold and copper contents of samples of rotary drill cuttings from the Fubilan intrusion with SiO₂ are illustrated in Figure 94. Samples that have SiO2 contents between about 55 and 70 weight percent consist entirely of altered intrusive rock whereas samples that have SiO₂ less than 55 weight percent have been contaminated with massive ores, limestone, or gossan and those with SiO₂ greater than 70 percent are contaminated by quartz veinlets. High values of gold and copper are present at nearly all levels of SiO₂ but are most plentiful in the range from about 60 to 70 weight percent SiO₂.