Whole Rock Geochemistry

Transcription

1 5 Whole Rock Geochemistry 5.1. Geochemistry of Granites Previous geochemical assessments of the Lebowa Granite Suite have usually focused on differentiation characteristics of the stratigraphic components (McCarthy, 1977; Lenthall & Hunter, 1977; Kleeman & Twist, 1989; Bailie, 1997). This study was conducted with the aim of describing geochemical variations as a function of alteration and mineralisation in the context of IOCG deposits, and to distinguish these features from differentiation/crystal fractionation trends or weathering effects. It was further an objective to establish the potential of these alteration signatures for use as a tool in evaluating mineralisation. According to Hitzman et al. (1992), intense alteration occurs in the host rocks of IOCG deposits but the exact alteration mineralogy and intensity may vary considerably both within and between districts depending on the host lithology and depth of formation. In general, a trend exists from sodic alteration at deep levels (albite-rich), to potassic alteration at intermediate levels (feldspar-sericite), to sericitic alteration and silicification at shallow levels (sericite + quartz). There further appears to exist an intense Fe-metasomatism prevalent locally in the host rocks. The lowest level sodic alteration as described by Hitzman et al. (1992) was discussed petrographically in the previous chapter, although was found to be of limited extent. Geochemical evaluation of these same rocks did not yield appreciable albitic signatures. In hand specimen it may be difficult to identify albitic alteration directly, which usually manifests only as a relative hardening of affected rocks, presumably

2 due to the increased ordering of feldspar lattices. The absence of sodic alteration in the study area may be a reflection of the stratigraphic position with respect to the mineralisation; or that the products of early sodic alteration that have been overprinted by subsequent alteration. It may also be considered that the system, in general, may be too Na-poor for this alteration to be significant and thus somewhat unique with respect to other IOCG-type alteration. The granite sensu-stricto nature of the Lebowa Granite Suite is well documented (McCarthy, 1977; Kleeman & Twist, 1989; Bailie, 1997), having an A-type, intraplate, granitic signature. The common characteristics of various genetic types of granites are summarised in Table 5.1. The Bushveld Nebo granites appear to be chemically consistent east to west, across the outcrop of the Lebowa Granite Suite, having average values of alumina (~11.50 wt% Al 2 O 3 ), very high values of potassium and sodium (~5.00 wt% K 2 O; ~3.50 wt% Na 2 O), and extremely low values for calcium (~0.50 wt% CaO). Describing granites in terms of I-, S- & A-type (Table 5.1) makes some implication towards the parental magma. Generally, I-type granites are derived from partial melting of predominantly igneous rocks, producing granites of tonalitic to granodioritic composition, with S-type granites being derived from partial melting of predominantly sedimentary rocks producing granites of adamellitic to granitic composition. I-type granites also tend to be more oxidised and associated typically with porphyry Cu-Mo mineralisation, whereas S-type granites tend to be more reduced and typically associated with Sn-W mineralisation. A-type (anorogenic/post-tectonic) granites do not fit comfortably within this scheme and may exhibit characteristics of either of the other two groups. Barton (1996) presented a classification scheme that regarded granites as a continuum, and linked the granite composition and magmatic oxidation state to metal associations and ore deposit types (Figure 5.1)

3 Table 5.1. Characteristics of S-type, I-type, and A-type granites (Chappell & White, 1974). Granite Type Tectonic Environ Chemical Signature Typical Accessory Minerals S-Type Orogenic Metaluminous to strongly Muscovite, garnet, peraluminous, high cordierite, tourmaline 18 O/ 16 O, 87 Sr/ 86 Sr I-Type Orogenic Metaluminous Biotite, hornblende A-Type Anorogenic / rift-related Metaluminous to mildly peralkaline, Fe-enriched Fe-biotite, Na-amphibole, Na-pyroxene, hedenbergite, fayalite, titanite Figure 5.1. Generalised scheme that links granite compositions and magmatic oxidation state to metal associations and intrusion-related ore deposit types (modified after Barton, 1996). Metals shown in bold reflect the more important associations (taken from Robb, 2005)

4 5.2. Chemical Variation in the Bushveld Granites The fractionated character of the acid rocks of the Bushveld Complex is well documented (McCarthy, 1977; Kleeman & Twist, 1989; Bailie, 1997) with chemical variation demonstrable through a vertical section of the intrusion. The granites examined from the study area, however, appear to be derived from a similar stratigraphic level such that geochemical effects of crystal fractionation are expected to be muted. The Nebo granites in this study are therefore treated together as one population. This will be discussed further in the next section, below. The Nebo granite is the most abundant lithology and is considered to relate principally to the upper portions of the sheet, with Klipkloof and Bobbejaankop varieties occurring in lesser quantities. Major and trace element geochemistry presented here is suggested to reflect variations as a consequence of alteration, mineralisation, or both, with crystal fractionation having little influence. These are compared to a base population of unaltered/least-altered samples of Nebo granite, but may also include other finergrained varieties of granites. It may of course be strictly incorrect to refer to these as unaltered, as even the least altered samples have undergone some alteration during late-stage magmatic stages. It is, however, sufficient in these terms to distinguish background alteration from that directly related to the mineralisation event(s). Unaltered granites from literature sources have been used in some instances for comparison (Kleeman & Twist, 1989; Bailie, 1997). Geochemical analyses for selected background Nebo granites are presented in Table 5.2 with comparative analyses from fresh Nebo granites from other localities in the Bushveld. Selected analyses are also presented for differentiated granites and dominant alteration styles (Table 5.3). Complete analyses are presented in the Appendices of this text

5 Table 5.2. Geochemical analyses for least-altered Bushveld Nebo granites from Ruigtepoort area and comparisons with analyses of unaltered Nebo granites from elsewhere in the Bushveld Nebo Nebo Nebo Nebo Nebo Nebo Granite 1 Granite 2 BG BG BG BG BG BG GN-50 SiO TiO Al 2 O Fe 2 O FeO MnO MgO CaO Na 2 O K 2 O P 2 O L.O.I Total A/CNK R R Rb Sr Y Zr Nb Co nd nd nd nd nd nd 5 Ni 10 nd nd nd nd 11 7 Cu nd nd nd nd nd nd 6 Zn Ba La Ce Th Sn 12 6 U Au nd Qtz Ortho Alb Anor Source Data: Nebo granite, eastern Bushveld Complex. (Bailie, 1997). 2 Source Data: Nebo granite, Bushveld Complex. (Kleeman & Twist, 1989)

6 Table 5.3. Average geochemical analyses for differentiated granite types and altered Nebo granites from the Ruigtepoort area Bobbejaan 1 Klipkloof 1 Klipkloof Deuteric K-Metaso Sericitic Sil-Hem Chl GN-40 GK-23 BG BG BG BG BG BG SiO TiO Al 2 O Fe 2 O FeO MnO MgO CaO Na 2 O K 2 O P 2 O L.O.I Total A/CNK R R Rb Sr Y Zr Nb Co nd nd nd nd nd 91 Ni nd Cu Nd Nd nd 32 nd nd nd 4560 Zn nd nd nd 40 Ba La Ce Th Sn nd 49 nd U Au Qtz Ortho Alb Anor Source Data: Nebo granite, Bushveld Complex. (Kleeman & Twist, 1989)

7 Two general subsets have been identified in the samples, relative to the unaltered Nebo, and these are termed altered and mineralised samples. The altered subset is further divided according to the dominant alteration in the samples, although more than one style of alteration may be present. The altered groups are represented as deuteric alteration, K-metasomatism, sericitisation, silicification-hematisation and chloritisation. The mineralised groups have been classed upon their dominant mineral association, namely Fe-F, Cu-Au and Fe-REE. These mineralised associations broadly coincide with the following alteration assemblages: hematisation - (Fe-F); chloritisation (Cu-Au); silicification (Fe-REE)

8 5.3. Geochemistry of Weathering and Hydrothermal Alteration Weathering The expected weathering products of an average Nebo granite, with a basic mineralogical assemblage of K-feldspar, quartz, and biotite, are chlorite and sericite at preliminary stages of weathering, with kaolinite developing at more advanced stages of weathering. This process is represented by the following hydrolytic exchange reactions: (5-1) 3KAlSi 3 O 8 + 2H H 2 O KAl 3 Si 3 O 10 (OH) 2 + 6H 4 SiO 4 o +2K + K-feldspar muscovite dissolved silica (5-2) KAl 3 Si 3 O 10 (OH) 2 + 2H H 2 O 3Al 2 SiO 5 (OH) 4 + 2K + muscovite kaolinite (5-3) 2KAlSi 3 O 8 + 2H + + 9H 2 O Al 2 SiO 5 (OH) 4 + 4H 4 SiO 4 o +2K + Orthoclase kaolinite dissolved silica (5-4) 2Fe(OH) 3 Fe 2 O 3 + 3H 2 O or Fe(OH) 3 FeO(OH) + H 2 O hematite goethite The appearance of appreciable amounts of advanced weathering products such as kaolinite were not observed in any analysed samples, and is therefore considered to be of nominal importance in treating the geochemical data. Figure 5.2 is a trivariate plot of Al 2 O 3 -(CaO + Na 2 O)-K 2 O. Nesbitt and Young (1984; 1989) identified a weathering trend from an average granite composition to the illite composition, and an advanced weathering trend from illite to the kaolinite compositional end-member. Data for least altered Nebo samples plot near to the position for the average granite and occur in a tight cluster, suggesting agreement with fresh assemblages, differing perhaps only in modal compositions. It should be

9 recognised that the Bushveld granites are, in general, plagioclase poor, which may account, at least partially, for the offset of the least altered compositions from the presented average granite. It is therefore regarded that the least altered granite set are also relatively fresh. This offset might also reflect the inherent deuteric alteration of the Lebowa Granite Suite observed regionally (Crocker et al, 1988). The actual geochemical variation demonstrates a clear divergence from the weathering trend in the direction of the K 2 O end-member indicative of broad-scale K- metasomatism. Also important to note is the total absence of any sodic-calcic Al 2 O Kaolinite Alteration Type Nebo Granite Deuteric K-Metasomatism Sericitisation Silicification - Hematisation Chloritisation Kleeman & Twist, Advanced Weathering Trend 80 Illite Muscovite 60 Plag Average Granite Kfspar CaO + Na 2 O K 2 O Figure 5.2. Ternary of Al 2 O 3 -(CaO + Na 2 O)-K 2 O demonstrating common alteration trend of an average granitic composition. Data from this study follow a discrete trend which tends to the potassium end-member and reflects K-metasomatism of granitic host rocks (modified after Nesbitt and Young, 1984; 1989)

10 alteration which would be indicated by points plotted towards the CaO + Na 2 O endmember. The one sample found in this domain, corresponding to a silicic-hematitealtered granite, plots near the sodic-calcic end-member due to near total removal of aluminium and potassium from the whole rock determination. Hydrothermal Alteration A succession of alteration products is proposed as being developed as a response to hydrothermal alteration. Various metasomatic ion exchanges are expected, including sodic (Na + ), potassic (K + ), hydrolytic (H + ), and ferric (Fe 2+ /Fe 3+ ). These reactions were discussed in the previous chapter. Relative to unaltered Nebo samples, there does not appear to be any appreciable effects of sodic metasomatic enrichment in any samples shown here (Figure 5.2 above). This may be contrary to expected metasomatic products common in other IOCG deposits (Hitzman et al., 1992, Oliver et al., 2004). Albitisation, however, has been recorded elsewhere in the Bushveld granites (Crocker et al, 1988; Freeman, 1988). A strong enrichment towards the potassic end-member of increasingly altered samples is representative of potassium metasomatism accompanying mineralisation. The post-magmatic processes associated with K-metasomatism/potassic alteration include: (1) base-exchange reactions in feldspars, specifically Na for K, or, K for Na; (2) changes in the structural state of feldspars; (3) albitisation; (4) microclinisation; (5) growth of tri-octahedral micas (Pirajno, 1992). The products and characteristics of each of these processes may be significantly different, and any or none may accompany another process depending on various conditions. Potassium metasomatism is characteristically associated with the replacement of feldspar and

11 quartz by K-feldspar (microclinisation) or albite (albitisation). Biotite is another important product of potassic alteration and is commonly Fe-rich. The presence of potassic alteration in the high-temperature cores of porphyry and epithermal mineralising systems is well documented, with common assemblages of K-feldspar-biotite-quartz, K-feldspar-chlorite, K-feldspar-biotite-magnetite, and possible accessory phases of albite, sericite, anhydrite, apatite, and occasionally rutile (Pirajno, 1992). These general alteration assemblages are observed in many samples suggesting similar conditions for formation, although maintaining, perhaps, distinct genetic origins. The best examples of intense potassium metasomatism resulting in microclinisation/episyenitisation are represented on Figure 5.2 by samples plotting near the K-feldspar composition. Sericitisation of country rocks is abundant and is characterised by the growth of sericite as a replacement of feldspars and mafic phases, as seen in the previous section. These samples plot on the alumina-potassium abscissa, which chemically, is a response to a massive loss of both calcium and sodium, including a gain of potassium. Aluminium is largely unaffected. The hydration and hydrolysis processes responsible for weathering, discussed above, are similarly important in hydrothermal alteration, as seen in the equation below: (5-5) 3KAlSi 3 O 8 + 2H + KAl 3 Si 3 O 10 (OH) 2 + 6SiO 2 +2K + K-feldspar muscovite quartz Some of the sericitised samples tend towards the potassic end-member suggesting that the differing styles of alteration have not occurred discretely, and that some overprinting may have occurred

12 Normative values for quartz, orthoclase and albite (Ternary plot of Q-Ab-Or; Figure 5.3) produce a similar plot to that of Figure 5.2, with the silicification trend associated with alteration and mineralisation more clearly evident. Silicification is common in hydrothermal systems and in breccia pipes. The silicified samples represented on Figure 5.3 have been depleted in all end-member elements (CaO, Na 2 O, K 2 O, and Al 2 O 3 ) and enriched in SiO 2 (not plotted here). These samples generally have >80 wt% SiO 2, ~10 wt% Al 2 O 3, ~3 wt% K 2 O, and consist primarily of quartz and sericite. This alteration, in many cases, appears to be so pervasive that significantly different lithologies may exhibit similar assemblages and chemical properties. 0 Q 100 Alteration Type Nebo Granite Deuteric K-Metasomatism Sericitisation Silicification - Hematisation Chloritisation Kleeman & Twist, 1989 Silicification Sericitisation K-Metasomatism 80 Albitisation 20 Ab Or Figure 5.3. Ternary Q-Ab-Or demonstrating alteration trends in terms of normative values of quartz, albite plagioclase and K-feldspar

13 It should be noted, however, that none of the three mineralised subsets (Cu-Au, Fe-F & Fe-REE) retain significant feldspar, but concentrations of SiO 2 in many mineralised samples may be greatly enhanced. The source of the silica may either be from magmatic and/or hydrothermal solutions, or from alteration products. As seen in equation (5-5), silica is released during sericitisation of feldspar, which may be migrated away or act locally to silicify the affected rock. Silica may likewise be derived during potassic alteration, in a process resulting in the production of episyenitic granites. The removal of silica would expectedly have an associated volume loss. Figure 5.4 is a ternary plot of SiO 2 -(K 2 O + Na 2 O)-Fe 2 O 3 that is more functional in tracking the Fe-metasomatic characteristics, particularly associated with 0 SiO Alteration Type Nebo Granite Deuteric K-Metasomatism Sericitisation Silicification - Hematisation Chloritisation Kleeman & Twist, 1989 Least Altered Compositions K 2 O + Na 2 O Fe 2 O 3 Figure 5.4. Ternary SiO 2 -(K 2 O + Na 2 O)-Fe 2 O 3 demonstrating the strong ferric component that accompanies mineralisation

14 the mineralised subset. The granitic samples (altered subset) plot near the silica endmember, and become more silica-rich with alteration, as demonstrated previously. There appears to be little or no ferric component to any of these alteration styles. The iron-rich nature of the mineralised samples, however, is clearly demonstrated, with the iron contained in the Fe-chlorite of the chloritised samples (Cu-Au), and in hematite and magnetite (Fe-F) and (Fe-REE) of the hematised and hydrothermal breccias. Bivariate Major Element Analysis Major element fractionation trends are commonly evaluated using bivariate Harker diagrams, whereby major elements are plotted against SiO 2. All Nebo granite samples examined in this study are considered to be derived from similar levels in the Nebo sheet, such that the bulk of chemical variation displayed on these diagrams is considered be a consequence of alteration and mineralisation with crystal fractionation having little influence. The plot of potassium versus silica (Figure 5.5) demonstrates the most significant trends with respect to sericitisation and K-metasomatism, in particular, the strong potassium enrichment that accompanies these styles of alteration (c.f. Figure 5.2). The silicified-hematised rocks are depleted with respect to potassium and form a discrete field. The chlorite-altered samples exhibit a marked depletion of both silica and potassium relative to background Nebo compositions. The distribution of iron (Figure 5.6) demonstrates a strong enrichment of iron in both mineralised subsets of samples (chloritisation and hematisation), which seems to be in accord with what may be expected in IOCG deposits. The depletion of iron experienced by sericitised and silicified samples is generally accounted for by the loss of biotite to sericite from the original granitic assemblage. From this diagram it may

15 be suggested that Fe-metasomatism does not necessarily spatially correspond to K- metasomatism, as might be suggested by some authors (Oliver et al., 2004), indicated by the unchanged concentration of Fe within the K-metasomatised subset. The distribution of aluminium (Figure 5.7) fluctuates little with respect to alteration, and the depletion in samples that have been chloritised and silicified-hematised is likely a reflection of the absence of alumino-silicates such as muscovite-sericite or illite. Calcium generally decreases for all samples with respect to original Nebo compositions, which themselves have low initial Ca values (Figure 5.8). Only the chlorite-altered granite samples possess any appreciable amount of calcium due to the development of fluorite associated with mineralisation. The Na 2 O-SiO 2 Harker diagram (Figure 5.9) demonstrates a near total depletion of sodium in almost all samples with respect to the original Nebo composition. This severe loss of sodium may reflect metasomatic processes active in IOCG deposits. The commonly described sodic or sodic-calcic alteration associated with many IOCG systems is well documented (Sillitoe, 2003; Oliver et al., 2004), yet there exists little evidence of this alteration type in host lithologies of the study area. It may be possible that the signature of local sodic alteration has been overprinted by subsequent alteration and completely removed

16 10 Figure K 2 O (wt%) SiO 2 (wt%) 30 Figure Fe 2 O 3 (wt%) SiO 2 (wt%) Figure 5.5. K 2 O vs. SiO 2 plot for samples demonstrating variation in potassium with respect to different styles of alteration. Figure 5.6. Fe 2 O 3 vs. SiO 2 plot demonstrating variation in iron with respect to alteration

17 15 Figure Al 2 O 3 (wt%) SiO 2 (wt%) Figure CaO (wt%) 1 Crystal Fractionation Trend SiO 2 (wt%) Figure 5.7. Al 2 O 3 vs. SiO 2 plot demonstrating variation in aluminium with respect to alteration. Figure 5.8. CaO vs. SiO 2 plot demonstrating variation in calcium with respect to alteration

18 5 Figure Na 2 O (wt%) SiO 2 (wt%) Figure 5.9. Na 2 O vs. SiO 2 plot demonstrating variation in sodium with respect to alteration

19 Composition-Volume Relations in Altered Rocks of the Study Area Metasomatism of a least-altered parent can be expressed in terms of the constituents that are either lost or gained by the system. Gresens (1967) provided equations for calculation of losses and gains from chemical data relative to specific gravities of altered and unaltered equivalents, in order to determine the compositionvolume relations involved during metasomatism. According to Grant (1986), Gresens equation may be transformed into a linear expression between the concentration of a component in the altered rock and that in the original rock, which may be simultaneously solved for each component. Components that demonstrate neither loss nor gain in mass define an isocon (iso-concentration) denoted by a straight line through the origin of the graph. The slope of this line defines the mass change due to alteration, and deviation of data from this line defines the concentration change in the indicated component. The isocon diagrams prepared in this study (Figure 5.10) follow derivations of this calculation after methodology presented by Grant (1986), with scaling factors applied to each component individually, as indicated in Table 5.4. Table 5.4. Scaling factors used in construction of isocon diagrams. Component Scaling Factor Component Scaling Factor SiO Ba 0.05 TiO2 20 Rb 0.05 Al2O3 1 Sr 0.20 Fe2O3(t) 2.50 Y 0.10 MnO 100 Zr 0.05 MgO 30 Nb 0.20 CaO 10 La 0.10 Na2O 5 Ce 0.05 K2O 5 Th 0.50 P2O5 40 U 0.50 Zn 0.05 Cu 1 Ni

20 a) b) 5% Mass Loss 3% Mass Loss c) 6% Mass Gain d) 10% Mass Loss e) f) 31% Mass Loss 37% Mass Loss

21 Figure Isocon plots of averaged altered granite types with respect to averaged least altered granites. Thin line reproduced in plots (b)-(f) corresponds to the reference deuteric isocon in (a). a) Deuterically altered granites with good isocon for most components, with slight mass loss with respect to the least altered granites. b) K-metasomatic alteration with slight mass change and some concentration variation for components related to K-feldspar, in particular. c) Prolonged or intense K- metasomatism resulting in microclinised granite with relative mass gain. Potassium strongly enriched at the expense of sodium and calcium. d) Sericitic alteration with mass loss; aluminium expectedly involved in a one-for-one transformation from feldspar to sericite with no resultant concentration change. e) Silicic-hematitic alteration exhibiting large mass loss and broad component concentration variation. f) Chloritic alteration exhibiting large mass loss with significant increased concentrations of base metals, Rare Earths and iron. Best fits to the data, corresponding to the isocon, were calculated from the major element chemistry alone using the equation: (5-6) Y = bx The isocon presented in each case in Figure 5.11, as generated by the above equation, is corroborated by the coincidence of particularly immobile components with respect to the styles of alteration examined, specifically Al, Ti and Zr. For the purposes of this study, the isocons generated are considered to reflect acceptable estimations of actual mass changes. Changes in Si concentrations can be shown, for the most part, to vary according to mass changes, which need not necessarily require the addition or subtraction of silica to the system. The clear exception corresponds to the case of silicification associated with hematisation (Figure 5.10 (e)). The isocon plot illustrated in Figure 5.10 (a) shows very close correlation between the deuteric-altered granites and the least altered Nebo granites with a nominal mass change of -3%. The loss in Mg may be a result of chloritisation of biotites during deuteric alteration. Figure 5.10 (b) illustrates the concentration variations of K-metasomatised granites relative to least altered granites, where most noticeably, Na and Ca are becoming

22 significantly lost in favour of K; reflecting the microclinisation of albite within perthite. Most of the remaining components do not deviate much from the isocon line. A mass loss of 5% is inferred for K-metasomatised affected granites, which is only slightly more than that for deuteric altered granites. The components K, Ba, Ni, and Nb appear to be somewhat collinear, which may suggest that their concentration changes are related. A similar case could be made for the components Ca, Na, Ce, La and perhaps Sr, which are lost from the system. Concentration variations for microclinised granites (Figure 5.10 (c)) demonstrate extremely high gains of K at the expense of Na and Ca as in less intensely K- metasomatised granites. A mass gain of 6 % is inferred for microclinised granites. Component gains are observed for K, Ba, Y, Ni, Nb, Th & P; with losses observed for Na, Ca, Zr, Mg, Mn, and Zn. Concentrations of Al and Si appear to vary as a function of the volume change and not due to chemical addition or subtraction. The average sericitised granites (Figure 5.10 (d)) are calculated to have a mass loss of 10 % likely due to the conversion of feldspar to white micas. Many more components of the system appear to be mobile with gains observed for K, Ba, Mg, Rb, Ni, Nb, U, and Mn; losses observed for Na, Ca, Sr, Zn, La, Ce, and Th. The isocon for silicified-hematised granites (Figure 5.10 (e)) shows large gains in many components, notably, Si, Fe, Mg, Cu, Ni, Y, Sr, Ni, Ti, U, Mn and P, with losses observed for Na. The calculated loss in mass is ~37 %. The expected La-Ce gains that might be expected do not appear in the altered granite but appear to be confined to mineralised sedimentary xenoliths in the area which have been shown to have concentrations of LREE in excess of 1 wt%. The chloritised granite isocon (Figure 5.10 (f)) has a calculated mass loss of ~31 %. The concentrations of Si, Ba, Zr, Al, Ti, and Sr decrease relative to the least-altered granite but lie close to the chlorite-granite isocon. This suggests that the change in

23 concentration of these elements is proportional to the mass change of the rock and not related to the style of alteration. Gains are observed for Fe, Cu, Y, Ni, Mg, La, Ce, Mn, Ca, Nb, U, P and Zn, with losses observed for K, Na, and Rb. It is clear from this plot that mineralisation is intimately related, spatially, to chloritisation. Trace Element Geochemistry with respect to Crystal Fractionation Trace elements are commonly more selectively susceptible to magmatic and hydrothermal processes. They are particularly useful in characterising trends in crystal fractionation as shown by numerous authors; for example Hunter (1973), McCarthy (1977), Kleeman & Twist (1989) and Bailie (1997). The effect of crystal fractionation on chemical variations within the selected sample set is important to quantify before conducting any treatment with respect to alteration. Bailie (1997) demonstrated that the Bushveld granites were derived from a more fractionated source than the roof rocks with which it was coeval. He also noted normal fractionation trends internally within both of the Rashoop Granophyre Suite and the Lebowa Granite Suite, although observed trends in the Rooiberg Group were weak and Rb even decreased with height in the volcanic pile. The initial concentrations of Rb, Sr and Ba in granitic rocks are controlled by processes of fractional crystallisation and are dependant on their relative partition coefficients (Table 5.5). It is expected that concentrations of Sr and Ba decrease and Rb increase in residual fluids with increasing crystallisation. This translates to a relative increase in Rb upwards in the granite sheet with a reciprocal decrease in Ba and Sr. It is known that crystal-fluid partition coefficients may be significantly different to crystal-liquid values for the same mineral, such that it allows for the

24 Table 5.5. Mineral/melt partition coefficients for rhyolitic melts (after Rollinson, 1993). K-feldspar 1 Biotite 1 Muscovite 2 Quartz 1 Plagioclase 1 Rb Sr Ba Source Data: Nash & Crecraft (1985), rhyolites SiO wt% 2 Source Data: Icenhower & London (1995), Averaged data from experiments evaluation of deviations from the expected trends as being a function of geochemical modification due to alteration. The element pairs K-Rb, K-Ba, and Ca-Sr respectively have similar ionic properties (Kinnaird, 1987). As a consequence, Rb and Ba are usually substituted in potassic minerals such as K-feldspar and biotite, and Sr substituted in minerals such as plagioclase. Thus higher values are expected for Sr near the base of the sheet where cumulus plagioclase may be found, which decreases vertically owing to increasingly dominant potash nature of the Bushveld granites. Table 5.6 shows average values for Rb, Sr and Ba with respect to alteration and mineralisation for averaged granites of the study area. The table reflects the mineralogical transitions that occur as a consequence of alteration. Average Rb contents of Bushveld granites are generally very high, and the average Rb/Sr ratio exceptionally high in comparison to S- or I-type granites. Bivariate trace element plots of Rb vs. Sr or Rb vs. Ba tend to be the most sensitive to crystal fractionation. The plots presented in Figures 5.11 (a)-(b) indicate crystal fractionation trends as determined from data by Kleeman & Twist (1989) and

25 Table 5.6. Average values of selected trace elements for coarse-grained granites with respect to alteration and mineralisation. n Avg. Rb (ppm) Avg. Sr (ppm) Avg. Ba (ppm) Unaltered Nebo Altered Deuteric K-Metasomatism Sericitic Microclinisation Silicification Mineralised Silicification (REE) Hematisation (Fe-F) Chloritisation (Cu-Au) Qtz Hem Breccia proposed alteration trends as determined from geochemical analyses for granites of the study area. The alteration trend corresponding to K-metasomatism and sericitic alteration in Figure 5.11 (a) is considered to be distinct from that of the fractionation trend. The fields indicated for each of the alteration types increases along the given trend in accordance to increasing intensity of alteration. The similarity between the alteration trend and the fractionation trend is interpreted to be dependent on the primary mineral phase involved in both processes, namely the feldspars. With fractionation the trace element concentrations vary due to changing feldspar compositions during crystallisation; similarly, with alteration trace element concentrations may vary with K-metasomatism. Large ion lithophile element (L.I.L.E.) geochemistry can also be characterised by the relationship of source to magma, which can be modelled using melting vectors (Figure 5.12 (a-b)). According to Inger and Harris (1993), these vectors show that vapour-absent breakdown of muscovite results in an increase in Rb/Sr concomitant

26 a) 1000 Alteration Trend Crystal Fractionation Trend 100 Alteration Trend Sr (ppm) 10 1 b) Rb (ppm) Alteration Trend Alteration Trend 1000 Crystal Fractionation Trend Ba (ppm) Rb (ppm) Figure a) Bivariate plot of Rb vs. Sr with indicated crystal fractionation trends and alteration trends. b) Bivariate plot of Rb vs. Ba with indicated crystal fractionation trends and alteration trends

27 with a decrease in Sr and Ba that corresponds to the observed variation between the postulated and the final granite. This technique may be applicable in reverse with respect to crystal fractionation, but it remains to be seen if similar applicability exists with respect to alteration. Fractionation and alteration trends were constructed for both diagrams. It is interesting that the K-metasomatic-sericitic alteration trend is sub-parallel to the K- feldspar melting vector. It must be determined whether this trend reflects an original fractionation trend that has remained unaffected by alteration, or if the alteration products result in similar trends as those obtained from either melting or crystallisation. With potassic metasomatism and the resultant growth of new K-feldspar and biotite, an increase in Rb/Sr might be expected. The position of altered samples plotted on Figure 5.12 (a) concur with this prediction, and the accompanying Sr decrease may correspond to the reduction of plagioclase in the altered assemblage

28 100 Kfsp Plag Bi Alt C.f. 10 Rb/Sr Sr (ppm) 100 C.f. Alt Kfsp Plag Bio 10 Rb/Sr Ba (ppm) Figure LILE covariation in Ruigtepoort granites with indicated crystal fractionation trends and alteration trends. a) Bivariate plot of Rb/Sr vs Sr b) Bivariate plot of Rb/Sr vs Ba (after Inger and Harris, 1993)

29 Samples that have been sericitised also coincide with the K-feldspar trend, and suggest an augmentation of the K-metasomatism trend. High Rb/Sr values are expected for sericitised samples from the recognised fact that muscovite is able to host significant concentrations of Rb. The sericitised samples with the highest Rb/Sr ratios, however, represent a variety of original lithologies, including both Nebo and Klipkloof varieties. The fact that these samples have not retained the fractionation signatures of their original litho-type may suggest, superficially, that Rb and Sr have changed characteristically as a consequence of alteration and independent of their original lithologies. The higher Rb/Sr values for the sericitised samples therefore track the growth of new muscovite/sericite. This would essentially represent an equivalent corollary process to that described by Inger and Harris (1993), that is, the destruction of muscovite. This is not the case, however, for the mineralised samples, including those that have undergone silicification-hematisation. There exists a second trend relative to the unaltered Nebo samples that reflects a decrease in Rb only, with Sr remaining more or less constant. This dramatic loss of Rb must be attributed to some characteristic of the mineralising system. An equivalent plot of Rb/Sr vs. Ba (Figure 5.12 (b)) does not produce similar trends. What is reflected is a general decrease in Ba in the mineralised and silicified samples. The shotgun scatter produced on this diagram presumably reflects the fact that Ba is an important element involved in alteration and is highly mobile. The salient trends of these diagrams indicate the close association between Rb and K. The suggestion is that Rb and K occur largely in the same minerals, i.e. K-feldspar and muscovite; as identified earlier in discussing element pairs. Thus as the rocks become more K-metasomatised the proportion of K-feldspar increases, and similarly as the rocks become more silicified, the proportion of K-feldspar decreases

30 This chapter presented alteration characteristics in terms of their geochemical signatures. It was demonstrated that consistent trends exist with respect to alteration, in particular with respect to the mineralogical changes of feldspars, which may used as indicators of mineralisation. The next chapter will focus on the mineralisation as found on each of the farms, including their potential and relationships in terms of IOCG mineralisation. Deposits will be considered in the context of their petrography and geochemistry as established in this chapter and the previous chapter