Petrogenesis of the Morobe Granodiorite and their shoshonitic mafic microgranular enclaves in Maramuni arc, Papua New Guinea

2.1 Tectonic setting

The geotectonic setting and emplacement of the Morobe Granodiorite is the result of the convergence between the continental Australian and oceanic Pacific Plates at the Melanesian Trench (locally known as the Manus-Kilinailau Trench) in the Upper Oligocene that brought the Ontong Java Plateau to the subduction zone. The Ontong Java Plateau could not be subducted which lead to the blocking of the Melanesian Trench (i.e., choking of subduction along the Manus-Kilinailau Trench) between 26 and 20 Ma [17,18]. To accommodate continued convergence between the Australian and Pacific Plates, north-dipping subduction of the Australian Plate commenced beneath New Guinea at the Pocklington trough resulting in the widespread magmatic activity of the Maramuni Arc along the New Guinea Orogen [6,19,20,21] including emplacement of the Morobe Granodiorite in the Middle Miocene [22]. Alternatively, other views theorize the magmatic activity of the Maramumi Arc to have been the result of south-dipping subduction at the Trobiand Trough to the north of New Guinea [1,4,17,19].

Miocene igneous rocks including high-K volcanics and granite, granodiorite, and diorite intrusives are broadly distributed to the south of the Owen-Stanley and Ramu-Markham faults (Figure 1a) [23]. One of them, the Morobe Granodiorite [24] is located at the eastern New Guinea Orogen within the northern region of the Owen Stanley Terrane [25,26,27]. The Owen Stanley Metamorphics constitutes the bulk unit of this area and is the oldest rock of Cretaceous age (Figure 1b). The Morobe Granodiorite intrudes the Owen Stanley Metamorphics as two main batholiths outcropping in the Ekuti and Kuper Ranges exceeding 600 km² and as some smaller bodies in the Bulolo and Watut Valley. Several Pliocene plug-dome complexes, stocks, sills, and dykes collectively termed the Edie Porphyry intrude the Morobe Granodiorite and Owen Stanley Metamorphics, which are variably overlain by the Bulolo Volcanics and sedimentary rocks of the Otibanda Formation, both of Pliocene age [22,28].

Figure 1

(a) Distributions of the Mid-Miocene granitoids and mineral deposits in the Maramuni arc, Papua New Guinea. The red square in the figure indicates the area of the map in Figure 1b. RMFZ: Ramu Markham Fault Zone. OSFZ: Owen Stanley Fault Zone. Modified from Holm et al. [6]. (b) Geologic map of the Morobe Granodiorite. Yellow triangles show sample localities for the study. Dashed lines represent inferred faults. Modified from Bain et al. [35].

2.2 Previous studies of the Morobe Granodiorite

The Morobe Granodiorite comprises predominantly of granodiorite with mafic to felsic variants occurring around the intrusion margins and includes monzonite, diorite, tonalite, hornblendite, gabbro, and pegmatite [24,29,30]. Rebek [31] identified stock-like bodies and sills of granodiorite in the Wau–Edie Creek area and classified them into three types: a medium-grained hornblende diorite to diorite; a coarse-grained hornblende granodiorite; and fine-grained granodiorite. Recent Mapping by Saroa et al. [32] observed granodiorite exposures along the Baiune Road to contain inclusions of three types: mafic inclusions with feldspar macrocrysts, elongated metamorphic inclusions of the Owen Stanley Metamorphics and inclusions of chlorite-altered mafic rock. A net vein complex of mafic and felsic rock compositions along the Watut Road was documented as well as dolerite dykes with epidote veins and nodules that were also observed by Cussen [22] as an appinitic swarm of dykes [32].

The granodioritic intrusions crop out primarily as texturally and mineralogically homogeneous, massive, medium- to coarse-grained phases, but locally the phases range from porphyritic to equigranular. Petrography studies indicate that the granodiorite is holocrystalline and medium- to coarse-grained with a hypidomorphic granular texture and is composed primarily of plagioclase, quartz, biotite, and hornblende, with an accessory mineral assemblage of titanite, apatite, magnetite, and pyrite; chlorite, epidote, and clinozoisite and clay minerals are common in hydrothermally altered phases [22].

The crystallization age of the Morobe Granodiorite is stated in three reports. The first dating was done by Page [33] using both K–Ar and Rb–Sr methods. Biotite and hornblende K–Ar ages from the granodioritic intrusions in the Ekuti and Kuper Ranges yielded an age of 14–13 Ma, which is 2 million years younger than the smaller intrusions in the Bulolo and Watut Valleys. Rb–Sr results for three biotite and whole-rock samples confirmed the K–Ar ages. These results indicated a Middle Miocene age of emplacement for the Morobe Granodiorite [33]. The second dating, done by Cussen [22] using the K–Ar method on fresh biotite and hornblende from samples of the Morobe Granodiorite, comprises age range of 15.1–10.5 ± 0.3 Ma (i.e., Early-Middle Miocene). The most recent age dating by Bodorkos et al. [34] using the new Sensitive High-Resolution Ion Micro Probe U–Pb method on zircons from the granodiorite hosting mafic inclusions along the Baiune Road yielded a magmatic crystallization age of 13.62 ± 0.13 Ma.

Whole-rock chemistries from 16 samples of the Morobe Granodiorite, including a suite of appinitic rocks from the more mafic intrusions-dolerite dyke have been already obtained by Cussen [22]. The results indicate a mafic to felsic composition with 41.6–65.7 wt% SiO₂. The data from this suite of rocks plotted on an AFM diagram follow the well-defined magmatic fractionation curve of a calc-alkaline intrusive complex. The appinitic assemblages display low major element-oxide vs silica correlation trends and are characterized by high total iron, CaO, MgO, Sr/Rb ratios, and a Fe₂O₃/FeO ratio greater than one. Geochemical studies indicated that the appinitic suite contains at least some cumulate phases and they were interpreted as a fractionation product of the Morobe Granodiorite [22].

3.1 Field mapping and sampling

Two field programs were carried out for this project. Before each field program was conducted, the desktop study of the area was done, and a literature review was compiled. Based on the literature review, selected strategic field mapping targets were proposed to carry out as much area as possible due to the large extent of the granodiorite and the ruggedness of the terrain.

The first field program was conducted in September 2012 using helicopter support provided by the then Morobe Mining Joint Venture now known as Harmony Gold Exploration. This field program covered the interior of the granodiorite in the Kuper Range (sample No. 1000xx series in Figure 1b) whereas, the second field trip conducted in March 2017, targeted the intrusion margins (sample No. 1028xx series in Figure 1b). Over 50 rock samples were collected by chipping off preferably fresh and unaltered samples from the outcrops with rock hammers, depending on the granularity, texture, and composition of the granodiorite. Metasediments of the Owen Stanley Metamorphics were also sampled at the intrusion margins. Appropriate measurements were recorded for structural data such as fault planes, joint sets, contact trends, and mineral lineations.

3.2 Petrography

A total of 37 thin sections were prepared for petrography at the Department of International Resource Science, Akita University, Japan. Petrography was performed using a Nikon Eclipse LV 100 N polarising microscope + CCD color camera CS5270B for reflective and transmitted light observations for the 37 thin sections. The mineral modal composition was determined using a James Swift automatic point counter by attaching its stage to the polarising microscope. Between 1,000 and 2,000 points (maximum count) were counted on the point counter at stage intervals of 1 and 2. The results were plotted on Quartz-Alkali Feldspar-Plagioclase (QAP) ternary diagram.

Observation of microtextures and identification of accessory minerals for the MME and surrounding granodiorite are conducted using the JEOL JSM-IT 300 scanning electron microscope with the Oxford EDS detector (SEM-EDS) at the Department of International Resource Science, Akita University, Japan.

3.3 Whole-rock chemistry

Major and minor elemental compositions of 23 samples from the Morobe Granodiorite and surrounding metamorphics were measured by X-ray fluorescence (XRF) analysis using a Rigaku ZSX Primus II (Rh X-ray tube, 50 kV, 60 mA) at the Graduate School of International Resources, Akita University. The analyzed samples consist of 19 from the Morobe granodiorite and 4 from the Owen Stanley Metamorphics near the granodiorite body. Calibration curves of elements were prepared by analyzing 15 standard samples of GSJ geochemical reference samples.

Before crushing the rock samples, they were checked as much as possible to be fresh, unaltered, and free of any veins. The glass beads for the XRF analysis were prepared by fusing the mixture of 0.8 g sample powder with 4.0 g lithium borate flux (Li₂B₄O₇:LiBO₂ = 3.0 g:1.0 g) [36]. Each mixture was fused at 1,200°C for 8 min using a bead sampler.

3.4 Mineral chemistry

Mineral chemical compositions of hornblende, plagioclase, and biotite from the MME and surrounding granodiorite were measured using an electron probe microanalysis (EPMA: JXA-8800) combined with wavelength dispersive X-ray spectrometry installed at the Graduate School of International Resources, Akita University. The analyses were performed at an accelerating voltage of 15 kV and a beam current of 20 nA.

5.1 Granitoids

The Morobe Granodiorite is mainly composed of holocrystalline, medium- to coarse-grained granitoids. The rocks comprise mainly hornblende, biotite, plagioclase, alkali feldspar, and quartz with accessory minerals of titanite, apatite, magnetite, and ilmenite.

Hornblende occurs as a euhedral to subhedral, straight-edged, hexagonal to rectangular medium-grained (0.3–1 mm in diameter) crystal (Figure 3a and d). Biotite commonly occurs as euhedral to subhedral medium-grained (0.5–1.5 mm) crystals (Figure 3e) but also as interstitial phases in cumulus-textured granitoids poor in alkali-feldspar (Figure 3f). Chlorite partly replaces biotite. Plagioclase occurs as euhedral to subhedral, medium- to coarse-grained (0.5–2 mm in diameter) crystals (Figure 3a). Plagioclase commonly shows oscillatory zoning, and some crystals exhibit wavy zoning boundaries (Figure 3b). Alkali feldspar occurs as anhedral coarse-grained (2–7 mm) crystals surrounding finer-grained euhedral crystals of plagioclase, hornblende, and biotite-forming poikilitic texture (Figure 3c and e) or as an interstitial mineral in cumulate phases. The intergrowth texture of micro-perthite is common with fewer amounts of microcline. Myrmekite occurs between alkali feldspar and plagioclase. Quartz occurs as anhedral to subhedral, medium- to coarse-grained (0.5–3 mm) crystals (Figure 3e). Larger crystals display undulose extinction. Titanite is generally anhedral to subhedral; however, there are some euhedral diamond-shaped crystals of fine to medium grain size (Figure 3c). Titanite occurs either as a discreet mineral in association with hornblende and opaque minerals or as an interstitial mineral to hornblende. Apatite is euhedral, hexagonal, prismatic, very fine-grained and commonly occurs as inclusions in hornblende and biotite (Figure 3d). Magnetite is the dominant opaque oxide mineral which commonly occurs as a medium-grained mineral in association with hornblende and titanate or as fine-grained inclusions in plagioclase (Figure 3d). Ilmenite is rare and fine-grained. According to the mineral modal compositions, the granitoids from the Morobe Granodiorite are classified into Quartz diorite, Quartz Monzodiorite, Granodiorite, and Quartz Monzonite (Figure 4).

Figure 3

Photomicrographs of the Morobe Granodiorite. (a) Quartz monzodiorite (sample No. 100034). Euhedral plagioclase and hornblende are included in a large anhedral alkali feldspar. (b) Quartz diorite (sample No. 100070). Large plagioclase shows distinct oscillatory zoning. (c) A euhedral titanite crystal in a quartz monzodiorite (sample No. 100034). (d) Accessory apatite and titanite occur with magnetite and hornblende in a granodiorite (sample No. 100017). (e) Poikilitic texture with large alkali feldspar containing hornblende, plagioclase, biotite, and opaque minerals (sample No. 100034). (f) Cumulus texture mainly consists of hornblende and plagioclase. Anhedral alkali feldspar and biotite occur as interstitial phases (sample No. 100070). (g) Gabbronorite is composed of two pyroxenes, plagioclase and opaque minerals (sample No. 102817). Pl: plagioclase, Afs: alkali feldspar, Hbl: hornblende, Ttn: titanite, Mag: magnetite, Ap: apatite, Bt: biotite, Opx: orthopyroxene, Cpx: clinopyroxene, Opq: opaque minerals.

Figure 4

Modal composition of granitoids from the Morobe Granodiorite in the QAP diagram [37]. Numbers for representative samples with poikilitic (sample No. 100034) and cumulus textures (sample No. 100070) are shown in the diagram.

Based on the textural relationship and the difference in grain size among main constituent minerals, we can divide the textures of granitoids into two types as follows: (1) poikilitic texture, in which euhedral crystals (mainly hornblende and plagioclase) are included in an anhedral large alkali feldspar (Figure 3e), and (2) cumulate texture, in which euhedral crystals (mainly hornblende and plagioclase) are tightly packed and form frameworks each other (Figure 3f). The poikilitic-textured rocks are rich in modal alkali feldspar and have more monzonitic compositions (e.g. Sample no. 100034 in Figure 4). The cumulate-textured rocks are poor in alkali feldspar and have more dioritic compositions (e.g. Sample no. 100070 in Figure 4). Modes of quartz and alkali feldspar roughly increase and those of plagioclase and hornblende roughly decrease with increasing elevation. The texture also varies from poikilitic to cumulus roughly with decreasing elevation.

The marginal parts of the Morobe Granodiorite show fine-grained equigranular to porphyritic textures. The rocks are mainly composed of biotite, plagioclase, alkali feldspar, and quartz and do not contain hornblende.

5.2 Gabbro

We found two types of gabbros in the outcrops, the main dark-colored and plagioclase-rich parts. The dark-colored part has a mineral assemblage of clinopyroxene, orthopyroxene, and smaller amounts of hornblende classifying it as a hornblende pyroxenite. The plagioclase-rich part comprises clinopyroxene, orthopyroxene, plagioclase, and minor amounts of hornblende and biotite classifying it as hornblende-bearing gabbronorite (Figure 3g). Opaque minerals commonly exist in both parts.

5.3 MMEs

The MME constitutes of hornblende, biotite, plagioclase, alkali feldspar, magnetite, and ilmenite, with minor amounts of quartz, pyrite, chalcopyrite, apatite, titanite, zircon, and barite. Subhedral hornblende, biotite, and plagioclase occur as 50–200 µm-sized grains throughout the MME (Figure 5). Anhedral alkali feldspar and quartz commonly occur as interstitial phases. Small amounts of 1–2 mm-sized hornblende and plagioclase crystals are also found in the MME (Figure 5a and b). The mm-sized euhedral hornblende contains many biotite and plagioclase inclusions. Pyrite and chalcopyrite occur in association with magnetite. Near the boundary with host granodiorite, large alkali feldspar poikilitically encloses fine-grained subhedral hornblende, biotite, and plagioclase (Figure 5c). Away from the boundary, the mode of alkali feldspar decreases.

Figure 5

Representative textures of the MME (sample No. 102812b). (a) A 1 mm-size euhedral plagioclase in the fine-grained matrix (cross-polarized light). Alkali feldspar occurs as an interstitial phase. (b) Large euhedral hornblende (lower right) and biotite (upper right) occur in the fine-grained matrix (plane-polarized light). Many small biotite crystals are contained in the large hornblende. (c) Backscatter electron image of poikilitic texture in the MME. Euhedral plagioclase and hornblende are in contact with the barite-bearing domain. (d) Barite occurs with hematite, voids, and a small amount of pyrite. (e) Backscatter electron image and corresponding elemental maps of (f) S, (g) Fe, and (h) Ba of the barite-bearing domain in Figure 5c. Py: pyrite, Brt: barite, Hem: hematite.

There are two types of barite occurrences in the MME. The first type of barite is coexistent with hematite and voids, within a certain area (Figure 5d and e). A very small amount of pyrite is also found within this area. Pyrite commonly occurs surrounded by hematite and is not in direct contact with barite. Another type of barite is usually found as inclusions in hornblende or biotite (Figure 6a–c). Barite may occur solitarily in hornblende or biotite (Figure 6c), or in voids in hornblende or biotite (Figure 6a and b). Barite in negative-crystal-shaped voids is commonly prismatic with a grain size of 2–5 µm (Figure 6b). It is accompanied by Cl-bearing chlorite.

Figure 6

Occurrence of barite and native gold in the MME (sample No. 102812b). (a) Barite inclusions in hornblende. (b) Close-up image of barite inclusions in Figure 6a. Prismatic barite crystals occur in the void in hornblende. Note the negative-crystal shape of the void. (c) Barite is included in biotite in hornblende. (d) A native gold grain in hornblende. (e) A native gold grain in plagioclase. The barite-bearing domain at the bottom of the photo is the same as in Figure 5e. (f) Close-up of the yellow rectangle area in Figure 6e. Native gold occurs in the void associated with altered plagioclase.

In addition, we found native gold grains in plagioclase and hornblende crystals (Figure 6d–f). Most gold grains occur in the voids or along the crack in plagioclase and hornblende. The grain size of gold is about 1–2 µm. We found 18 grains of gold in a 3 cm² of MME. A very small number of gold grains may also be found in the host granodiorite near the MME.

9.1 Crystal fractionation in the Morobe Granodiorite

In the Harker diagrams of whole-rock chemical compositions, the compositional variations in granitoids from the inner parts of the Morobe Granodiorite display a series of linear trends (Figure 11). However, the composition of MME deviates from the linear trends in the Harker diagrams for K₂O and Ba. Therefore, it is difficult to explain the linear compositional trends of the granitoids by mixing between the MME and a more silicic magma. The whole-rock compositional variations in granitoids correlate with textural variations associated with modal variations in the major cumulus phases, hornblende, and plagioclase. For example, samples showing cumulus textures are poor in SiO₂ and K₂O and rich in MgO (e.g., sample 100070 in Figure 3f). On the other hand, samples exhibiting poikilitic textures are rich in SiO₂ and K₂O and poor in MgO (e.g., sample 100034 in Figure 3e). Thus, the linear compositional trends of granitoids in the Harker diagram for Na₂O, MgO, Al₂O₃, K₂O, etc., can be explained by both hornblende and plagioclase fractionation. Compositional variations in Y and Zr in granitoids also represent the fractionation of hornblende with large partition coefficients for these elements. The increase in Ba with increasing SiO₂ suggests that biotite fractionation is not important.

Figure 11

Whole-rock Harker diagrams of major and trace elements for the host granitoids and MME.

Samples from the marginal parts deviate toward higher and lower, respectively, from the trend of the inner parts for Na₂O and K₂O. The reason for this is currently unclear, but mixing between the magma represented by the marginal composition and the MME may have occurred.

9.2 Magma genesis of the Morobe Granodiorite

The whole-rock compositions of granitoids from the Morobe Granodiorite are mostly plotted within the compositional range of partial melts of amphibolites among the compositions of partial melts of various rocks obtained in the partial melting experiments by Patiño Douce [48] (Figure 12). The Magnesian characteristics of granitoids also suggest that the source materials of granitoids are mafic rocks (Figure 7d). Whalen and Hildebrand [49] recently proposed new geochemical discrimination diagrams for magmas that occur at destructive plate margins based on the study of granitoids in the North American Cordillera. They used Nb/Y-Nb + Y diagrams and others to distinguish between arc magmas generated by volatiles streaming from a subducting slab and slab failure magmas derived through deep mantle processes, including partial melting of the metabasaltic/metagabbroic upper portions of torn slab (Figure 13). Based on the discrimination diagrams, the most samples from the Morobe Granodiorite are classified as slab failure, not arc. The geochemical signature of slab failure magmatism of the Morobe Granodiorite refutes the conventional idea that igneous activity around 15 Ma in the Maramuni Arc occurred in association with subduction (e.g., Holm et al. [6], Ӧsterle et al. [15]). Hall [17], Hill and Hall [20], and Webb et al. [16] argued that the arc-continent collision at about 25 Ma had ended the northward subduction of the Solomon Sea Plate that continued until the Late Oligocene and the southwestward oblique subduction had initiated in the Early Miocene. Since the continental collision causes the slab failure (e.g., Wortel and Spakman [50]; Davies and von Blanckenburg [51]; Hildebrand and Bowring [52]; Garzanti et al. [53]), it is quite possible that the old slab of the Solomon Sea Plate was torn in the Miocene age after the cessation of subduction followed by the arc-continent collision (Figure 14a). The Morobe Granodiorite may have been derived from magma generated by the partial melting of the metabasaltic/metagabbroic upper portions of the torn slab.

Figure 12

Whole-rock composition of the Morobe Granodiorite in major element diagrams after Patiño Douce [48]. Colored areas show compositional fields of experimental melts produced by the partial melting of felsic pelites, graywackes, and amphibolites.

Figure 13

Geochemical diagrams to discriminate arc, slab failure, and A-type magmatism (after Whalen and Hildebrand [49]) for the Morobe Granodiorite.

Figure 14

(a) Cross-sectional tectonic model showing the formation mechanism of Morobe Granodiorite and MME in the Middle Miocene. The direction of subducting plate motion and distribution of metamorphic rocks during this period are derived from Hill and Hall [20]. (b) Rb/Sr versus Ba/Rb diagram [64] shows the presence of phlogopite in the mantle for the MME in the Morobe Granodiorite. SCLM: subcontinental lithospheric mantle, PM: primitive mantle [74].

9.3 Petrogenesis of the MMEs

Previous studies of MMEs in granitoids have led to the following models for origins of MMEs: (1) melting remnants (restites) of source material for the host granitoid magma [54], (2) xenoliths or fragments of country rocks [55], (3) cognate mafic cumulates crystallized from the magma that produced the host granitoids [56,58], and (4) the mingled products of magmas of different origins from the host granitoids [59,62,].

The MMEs do not have any metamorphic- or sedimentary-rock fabrics which should be found in restites or xenoliths originating from the country rocks. Therefore, the hypotheses that the MMEs are restites and xenoliths of country rocks are not suitable as explanations for the petrogenesis of MMEs in the Morobe Granodiorite.

Poikilitic and porphyritic textures and zoned structures of plagioclase in the MMEs from the Morobe Granodiorite indicate that the MMEs are crystallized from magma. The rounded shape suggests that the MME existed as magma in unconsolidated granodiorite. The MMEs do not exhibit cumulus textures but poikilitic or porphyritic textures. This also contradicts the cognate hypothesis. In addition, as mentioned earlier, the composition of the MME deviates significantly from the linear trend of the host granitoids in the Ba- and K₂O-SiO₂ Harker diagrams of the whole-rock compositions (Figure 11). Furthermore, while the crystallization pressures of hornblende rims in the MME and host granitoids should indicate the emplacement depth of the granitoids, some hornblende cores in the MME show higher pressures (Figure 8b). This suggests that the MME was derived from a depth deeper than the emplacement depth of the granitoids. These textural, geochemical, and mineralogical characteristics are not consistent with the hypothesis that the MMEs are cognate mafic cumulates from the granitoid-forming magma. Therefore, the MMEs are likely to be mingled products of magmas of different origins from the magma that produced the host granitoids.

Altherr et al. [63] pointed out that extremely high temperatures (≥1,100°C) and geothermal gradients are needed to produce mafic metaluminous low-silica (<56 wt%) melts by dehydration melting of metabasaltic compositions. They also suggested that such high temperatures and geothermal gradients are unlikely since they would imply almost complete melting of the lower and middle crusts. Therefore, the MME (SiO₂ = ∼54 wt%) in the Morobe granodiorite likely did not originate from the melting of mafic crust but from the mantle. In addition, it has a higher Rb/Sr ratio than the primitive mantle (Figure 14b). This indicates that the MME is derived from the partial melting of phlogopite-rich mantle sources [64]. Experiments at 1 GPa have also confirmed that the partial melting of phlogopite-bearing peridotites produces shoshonitic mafic magma [65].

Phlogopite-bearing peridotites are generally regarded as the products of interaction between peridotite and slab-derived sediment melts and/or fluids (e.g., Kepezhinskas et al. [66]; Brandon et al. [67]; Zanetti et al. [68]; Mallik et al. [69]; Shaarschmidt et al. [70]). The occurrence of barite inclusions in euhedral hornblende and biotite, early-formed phases in the MME, suggests that the barite crystallized in the fluid inclusion trapped by them. This means that the MME-forming magma was originally rich in Ba, Sr, and SO₄ ²⁻ and that these elements were partitioned into the fluid phase, which was saturated in the early stage of crystallization of the magma. Considering that barite is found primarily in seawater and seafloor sediments on modern Earth [71], the crystallization of barite in the MME suggests that the mantle that produced the shoshonitic MME-forming magma has been metasomatized by slab-derived sediments. The metasomatism by slab-derived sediments is also consistent with the hypothesis of the granitoid magma genesis by slab failure since it is entirely possible that partial melting of detached slab or release of fluids from the slab can occur in the mantle after slab failure (e.g. Freeburn et al. [72]) (Figure 14a).

The domains in which hematite, barite, and a small amount of pyrite coexist form frameworks with euhedral hornblende and plagioclase (Figures 5e and 6e), suggesting that the domains were produced by the decomposition of magmatic primary minerals. Since hematite forms at higher oxygen fugacities than pyrite (e.g., Hodkiewicz et al. [73]), it is unlikely that they formed at the same time. Therefore, a small amount of pyrite in the domains must be the reactant mineral. The formation of hematite and barite may be explained by the oxidation reaction as follows:

Pyrite can form barite and hematite when reacting with Ba-bearing oxidizing fluids. The aforementioned Ba-bearing fluid which was saturated in MME-forming magma would have been the reacting fluid. The abundant native gold grains in the MME indicate that the MME-forming magma was rich in Au. They are unlikely to have crystallized directly from the magma, and most Au in the magma should have been partitioned into sulfides. Native gold grains in cracks and voids in hornblende and plagioclase may have been formed by the release of Au from pyrite that decomposed in the oxidation reaction. MMEs in the granitoids may have played an important role in transporting Au from the mantle to the crust (Figure 14a).