1. Evolution of oxygen fugacity with crystallization in the Bjerkreim-Sokndal layered intrusion (Rogaland, Norway) J-C DUCHESNE, B CHARLIER & J VANDER AUWERA Department of Geology, University of Liège, Bat. B20, B-4000 Sart Tilman (Belgium) Fig 1: The Rogaland anorthosite province showing the various units and particularly the BKSK intrusion with its two main lobes. Fig 2: Cumulate stratigraphy in the Bjerkreim lobe of the BKSK intrusion. The layered series is divided in several megacyclic units (MCU) subdivided into a sequence of zones (a-f), defined by the presence or absence of certain index minerals (from Meyer et al. 2002). Fig 3: Calculation of T and fO 2 values close to liquidus using the QUILF algorithm (Andersen et al. 1994) and Lattard et al. (2005) approach in the BKSK intrusion. Fig 4: Variation of the V 2 O 3 content of ilmenite vs. the stratigraphic succession in the Bjerkreim lobe. Fig 5: DV mt/ilm vs. (Eu 2+ /Eu 3+ ) plag in the Grader deposit, the Fedorivka layered intrusion and the Bjerkreim lobe of BKSK. The dashed line is the overall correlation (with r = 0.84). The plain line refers to the Bjerkreim lobe. Note that sample 57 greatly influences the slope of the linear relationship. Fig 6: DVmt/ilm vs. V2O3 in ilmenite. The variation in the Sokndal lobe appears different from that in the Bjerkreim lobe (see text). When possible (see Fig. 3), the calculated liquidus value of DV is indicated by an arrow. The Bjerkreim-Sokndal layered intrusion (BKSK, Fig. 1) is made up of a Layered Series of cumulates, organized in several macrocyclic units (MCU) (Fig. 2). MCU IV comprises a complete series of rock types with leucotroctolites passing upwards to olivine-free norites and gabbronorites. This unit passes to a thin Transition Zone TZ (in which olivine reappears), which is itself topped by mangeritic and quartz mangeritic units. The QUILF approach The QUILF algorithm (Andersen et al., 1993) obtained in BKSK (Fig. 3) clearly give temperatures (ca. 760°C) which are lower than the expected liquidus temperatures - e.g. in quartz mangerite zircon saturation indicate temperatures in the 880°- 920°C range (Duchesne and Wilmart, 1997). The only plausible T- fO 2 conditions are those calculated for the leucotroctolite in MCU IVb which show  FMQ = 1.3. Interestingly a minimum value  FMQ= 0.54 ±0.4 is obtained following Sauerzapf et al. (2008) and assuming a temperature of 1160°C in MCUIVa (ilmenite being the only Fe-Ti oxide, there is no subsolidus readjustment with magnetite!). Nevertheless, at the recorded equilibrium temperatures, a slight decrease in fO2 is observed (1.2 log units of  FMQ) followed by an increase of 0.2 log units in mangerite and quartz mangerite. The (Eu2+/Eu3+)plag variation The Eu 2+ /Eu 3+ ratio is not directly measurable by analytical methods at the level of concentrations in most rocks and minerals. An approximate method has been suggested by Philpotts (1970). Sr having the same charge and nearly the same ionic radius as Eu 2+ is used as a proxy for Eu 2+ after correction according to the Lattice Strain Model of Blundy and Wood (1994). Total Eu is measured in pairs of plagioclase and apatite In BKSK the (Eu 2+ /Eu 3+ ) plag has been calculated in MCU IIIe, IVe, IVf and TZ for which REE data are available (Charlier, 2001; Roelandts and Duchesne, 1979). It varies from 26 at the base of MCUIVe to 162 in the Transition Zone. The Vanadium behaviour in BKSK The V 2 O 3 content in ilmenite from BKSK shows interesting variation in the stratigraphic succession (Fig. 4): (i) in IA, ilmenite is not a liquidus mineral (but crystallized from the trapped liquid). It slightly differs from that in IB where it is a liquidus mineral; (ii) similar contents appear in IB, IIc, IIIc, IVa and IVc, where ilmenite is the only oxide mineral (no magnetite). The bulk DV cum/liq thus remains close to unit and this leads to a DV ilm/melt = 8, following the cotectic proportions of Duchesne and Charlier (2005); (iii) in the Bjerkreim lobe, there is a continuous decrease of V contents from IVc to the TZ, through IVe and IVf; (iv) IVc appears as an “accident” in the evolution; (v) in the TZ and overlying units the V contents remains low and does not vary much. Variation of DV mt/ilm with (Eu 2+ /Eu 3+ ) plag and with fO 2 The relative proportions of the different V n+ ionic species V 3+, V 4+ and V 5+ are sensitive to fO 2 variations. Toplis and Corgne (2002) have shown that V content in magnetite can be used as an oxybarometer provided the V content of the parental magma is known. Duchesne et al. (2007) have shown that the partition coefficient of (V 3+ + V 4+ + V 5+ ) between magnetite and ilmenite (DV mt/ilm ) varies with fO 2. Large variations of DV mt/ilm are indeed observed between two Fe-Ti ore bodies related to anorthosite massifs. The first is the Grader deposit (Havre Saint Pierre, Québec) (Charlier et al., 2008) in which fO 2 is estimated at  FMQ = ca. 1.5 log units (Lattard et al., 2005). The second ore body is the Fedorivka layered intrusion (Korosten plutonic complex, Ukraine (Duchesne et al., 2006). The fO 2 is estimated following the QUILF method and varies from  FMQ +0.7 to -1.4 log units. DV mt/Ilm measured in Grader, Fedorivka and BKSK (after correction for subsolidus re- equilibration) are plotted against (Eu 2+ /Eu 3+ ) plag on Fig. 5. The overall correlation is a rough linear relationship (r = 0.84, Fig. 5). Each massif nervertheless gives somewhat different trends, Fedorivka and BKSK showing an increase of both parameters with fO 2.. A gross evaluation of the fO 2 variation in BKSK would be ca.  FMQ = 1 at the base of MCUIV to  FMQ = 0 in the transition zone TZ. Variation of DV mt/ilm in the BKSK intrusion DV mt/ilm vs. the V content in ilmenite is plotted in Fig. 6. The ilmenite content can be taken as a proxy for the degree of crystallization of the intrusion, as shown in Fig. 4. Fig. 6 shows that an overall increase of DV mt/ilm from values ca. 5 to values >10 in the MCU IVf is followed by a clear decrease in the TZ and the acidic upper part. Where the QUILF approach has been used to reconstruct the “initial” compositions of the oxides,. it can be seen that, as a first approximation, the effect of re-equilibration can be neglected (arrow). A significant difference in the evolution appears between the two lobes of the intrusion (Fig. 1). This would mean that the evolution of fO 2 was not homogeneous in the whole magma chamber, possibly due to different enclosing rocks (anorthosites in the Sokndal lobe, migmatitic gneisses in the Bjerkreim lobe). The decrease of the DV mt/ilm value from the top of MCU IVf to the quartz mangeritic unit also suggests an increase in fO 2 though it might be partly due to the temperature decrease and to the changing composition of the melt. A new influx of acidic magma on top of the layered series has been demonstrated by (Duchesne and Wilmart, 1997). This influx of a magma at a higher fO 2 than the residing melt might have risen the fO 2 in the melt constituting the quartz mangerites and also in the underlying mangerite and TZ cumulates. Conclusions The fO 2 evolution in BKSK is more complex than the classical decrease characteristics of a close system crystallization. Several evidence show that fO 2 increases in the upper part of the intrusion. The DV mt/ilm appears correlated to (Eu 2+ /Eu 3+ ) plag and is thus a potential oxybarometer. Our results tend to show that it is less sensitive to subsolidus re-equilibration than the major elements in the Fe-Ti oxide minerals. Experimental data and other case studies are needed to strengthen this empirical approach.