4.3.1 Major element variations and rock classification

The concentrations (wt%) of the major oxides in Fobinso (this study), Esuajah (this study), Fobinso (Agbenyezi et al., 2020), Esuajah (Agbenyezi et al., 2020), diorite, granodiorite, and granite (Brako et al., 2020) vary in the following order: SiO2 (59.87–73.90), Al2O3 (12.00–22.90), MgO (0.03–7.82), CaO (1.40–7.74), Na2O (2.60–5.90), K2O (0.53–3.96), P2O5 (0.05–0.90), TiO2 (0.30–0.36), and FeOT (1.41–3.44) (Table 2).

The TAS diagram as suggested by Cox et al. (1979) is a plot, which is used to determine the type of plutonic rock. On this diagram (Figure 5), samples from the Esuajah and Fobinso pits of the Edikan Mine in the Kumasi Basin were compared with other granitoids elsewhere in the Kumasi Basin. It can be seen that C3 samples (Brako et al., 2020) fall in the diorite region, while almost all the samples from Fobinso (this study and Agbenyezi et al., 2020) and B2 (Brako et al., 2020) fall within the granodiorite region (Figure 5). Lastly, almost all A1 (Brako et al., 2020) and Esuajah (Agbenyezi et al., 2020) samples fall within the granite region.

On the Harker plots, there exists a good correlation between the major oxide and SiO2 (Figure 6) and it suggests that the rocks emanated from the same magma source, even though they have different spatial locations (Brako et al., 2020; Okobi et al., 2019). There is an increasing and later decreasing trend of Al2O3, K2O, and MgO (Figure 6a–c). This may be a result of fractional crystallization and/or assimilation of surrounding rocks that have lower concentrations of Al2O3, K2O, and MgO (Li et al., 2020). Minerals such as orthoclase, biotite, plagioclase, and anorthoclase (Figure 4), respectively, may have preferentially incorporated K, Mg, and Ca, separating them from the remaining melt. Na2O shows a progressively increasing trend (Figure 6d). This suggests that they were the last to crystallize from a melt or that the crustal assimilated rocks were rich in Na2O relative to the melt. Lastly, Figure 6e–h depicts a progressively decreasing trend, and this suggests that they were the first to crystallize out of the magma/melt or may have been contaminated with assimilated materials.

On the AFM diagram (Figure 7), most of the samples plot in the alkali–feldspar region, which explains the abundance of alkali–feldspar minerals in the petrography (Figure 4). Esuajah and Fobinso samples (this study and Agbenyezi et al., 2020) plot in the tholeiitic region, which implies a low abundance of K (Rollinson, 1993). On the other hand, all the samples from Brako et al. (2020) plot in the calc–alkaline region, indicating that they have a high amount of K. Some samples are plotting along the tholeiitic–calc–alkaline boundary. This transitional nature may be a result of magma mixing, assimilation, or fractional crystallization.

Samples from Esuajah and Fobinso (this study and Agbenyezi et al., 2020) and granodiorite (Brako et al., 2020) show peraluminous characteristics, while the remaining samples from Brako et al. (2020) show metaluminous characteristics (Figure 8). Fobinso (this study and Agbenyezi et al., 2020) and a few Esuajah samples (this study) plot in the S-type region, while the remaining samples plot in the I-type region.

4.3.2 Trace and rare-earth element variation: Implication for source characteristics

The source rock variation is determined using the trace and rare-earth element concentrations (Table 3). The concentrations (ppm) of the various trace elements in the study areas, respectively, are as follows: Fobinso- La (8.5–10.7), Yb (0.86–2.50), Sr (389–601), Hf (0.03–0.09), Rb (5.0–10.8), Y (1.57–3.32), Sm (0.25–6.01), V (6.0–9.0), Zr (1.1–1.7), Th (1.3–2.0), Nb (0.02–0.06), Ce (15.9–20.1), Nd (0.75–29.80), Ba (238–511), ∑REE (324.17), Average Eu/Eu* (1.08) and Esuajah- La (5.0–8.4), Yb (0.21–0.84), Sr (386–608), Hf (0.09–0.19), Rb (3.5–10.7), Y (1.58–3.98), Sm (0.22–3.15), V (3.0–6.0), Zr (1.7–5.1), Th (0.9–1.5), Nb (0.05–0.06), Ce (9.72–15.70), Nd (0.82–17.3), Ba (127–570) REE (173.16), and Average Eu/Eu* (1.05).

To appreciate the petrogenetic features and source rock characteristics of the studied granitoids, their trace element concentrations were normalized to the primitive mantle (McDonough & Sun, 1995) and REE chondrite (Boynton, 1984) concentrations (Figure 9; Table 4). In the normalized primitive mantle spider diagram (Figure 9a), most of the samples show similar distribution trends. They generally show enrichment in the large Ion lithophile elements (LILE) relative to the high field strength elements (HFSE). Negative Zr anomaly occurs in almost all the samples. This may be a result of the fractional crystallization of Zr-bearing minerals and/or the assimilation of Zr-poor-bearing materials, which may have diluted the Zr concentration in the magma (Rollinson, 1993).

From the chondrite-normalized REE plot (Figure 9b), it is clear that almost all the samples are enriched compared to the primitive chondrite. The light rare earth elements (LREEs) are relatively enriched as compared to both the middle rare earth elements (MREEs) and heavy rare earth elements (HREEs). This reflects the fractionation process during magmatic differentiation where minerals with high affinity for LREEs tend to preferentially incorporate them into their crystal structures (Adams et al., 2024). There exist two different types of Eu/Eu* in all the samples: almost all the granitic samples (Brako et al., 2020) and one sample each from Fobinso and Esuajah (this study) show strong negative Eu/Eu*, which reflects plagioclase fractionation. The remaining samples, on the other hand, show positive Eu/Eu*, indicating that a divalent Eu2+ was separated during the crystallization process and that Eu2+ predominates Eu3+.

4.3.3 Tectonic discrimination

On the R1[(4Si – 11(Na + K) – 2(Fe + Ti)] – R2[(Al + 2Mg + 6Ca)] diagram of Batchelor and Bowden (1985), almost all the samples plot in the syn-collisional region, with few samples from Brako et al. (2020) moving toward the late orogenic region (Figure 10a). This may suggest that the samples were formed in a tectonic setting related to collisional orogeny, where there is collision and convergence of continental crusts. According to Chappell and White (1974), samples that plot in the syn-collisional region are generally S-type in nature. This is because the magma before crystallization reacted with some of the crustal materials and hence, they may be anatectic in nature (Batchelor & Bowden, 1985; Okobi et al., 2019). While the remaining samples plot in the VAG zone, the granitic samples (Brako et al., 2020) plot in the syn-collisional region, which is suggestive of a subduction process (Adams et al., 2024).

4.4.1 Petrogenetic evolution

Kumasi Basin's diverse geochemical compositions indicate that the rocks were formed from various processes and emanated from several sources (Brako et al., 2020). The findings of this study, however, indicate that the studied granitoids came from a common magmatic source. This is evidenced by the strong linear correlation of the major oxides in the Harker diagrams (Figure 6) and the similarity in the distribution patterns observed in the spider diagrams—LILEs, HFSEs, and REEs (Figure 9).

The linear trend observed in the major oxides and trace elements can be attributed to the processes of magma differentiation and the fractional crystallization of plagioclase and biotite (Figure 4). On the other hand, the later decreasing trend of the major oxides on the Harker diagram and the transitional nature of some of the samples on the AFM diagram may reflect fractional crystallization and/or assimilation of the country rocks. The S-type signature signifies extensive crustal contamination of the melt during the formation of the rocks, while the I-type signature signifies limited or no crustal contamination with the melt (Chappell and White, 1992). The low K2O concentration against SiO2 (Figure 6) in diorite and samples from Esuajah (this study and Agbenyezi et al., 2020) buttresses the fact that there is minimal crustal contamination of the magma.

The coexistence of I-type mafic granites and S-type felsic granodiorites formed from the same magma source suggests a complex interplay of fractional crystallization and crustal assimilation in a subduction zone setting. The process begins with the generation of primary magma in the mantle due to the subduction of the oceanic plate under the continental plate. Fractionation took place as the magma ascended and continued to ascend to the surface; there was little or no assimilation of the surrounding crustal materials. The magma then solidifies into granites with limited crustal contamination and a lot of fractionated mafic materials, thereby assuming the I-type signature. The remaining magma within the magma chamber became relatively more felsic as the mafic content fractionated (Figure 11).

As the subduction process continued, a continental plate collided with the stable continental plate, which caused the magma generation in the magma chamber to stop. More mafic minerals were fractionated even as magma generation had stopped, thereby increasing the felsic content of the magma within the magma chamber. The relatively more felsic magma within the magma chamber then solidified into S-type felsic granodiorites after it ascended onto the surface with extensive crustal contamination. The model is summarized in the diagram below and it is consistent with the classic granite formation model of Petford et al. (2000).

4.4.2 Geodynamic setting of the area

According to significant geological, geophysical, and structural research, the Kumasi Basin is thought to have originated as a foreland basin that later rifted apart in a back-arc fashion to create several horst–graben structures (Brako et al., 2020; Chudusama et al., 2015; Tourigny et al., 2019). These structural features resulted in the formation of a basin-like feature that served as a deposition site for eroded volcaniclastic materials originating from the fault-bounded Birimian volcanic belts, specifically the Sefwi and Ashanti belts, thereby creating the appearance of a thick continental supracrustal unit. The morphology of the sheared contacts between the Kumasi Basin and those two fault-bounding volcanic belts may have been influenced by the rifting and pushing of the volcanic belts (Brako et al., 2020). According to Brako et al. (2020), the granitoids in the basin showed that Sr, Ba, and K were consistently enriched with significant depletions in HFSE.

Although most of the granodiorites plotted in the S-type region (Figure 8), their depletions in Nb suggest an active continental margin setting (Wang et al., 2018). All the samples plotted in the syn-collision field on the R1–R2 plot (Figure 10a). The idea that the granitoids in the Kumasi Basin primarily evolved in a syn-collisional environment is strongly supported by all geochemical data provided in this study. The distribution patterns of the granodiorite samples show high concentrations in LILE and LREE and low concentrations in HREE and HFSE.

4.5.1 Characteristics of gold deposits in the Kumasi Basin

4.5.2 Lithologies controlling gold mineralization in the Edikan Mine

Gold mineralization at the Edikan Gold Mine area is primarily hosted within the granite bodies, but also in classic “Ashanti-style” sediment-hosted shear zones (Agbenyezi et al., 2020). Major rock types within the Edikan Mine are the sheared metasediments, granites, and granodiorites. Even though they all contain gold, granodiorite is highly associated with gold within the area, followed by the granites. Sheared metasediments are less mineralized.

4.5.3 Structures controlling gold mineralization

Three distinct stages of veining (V1, V2, and V3) were also identified based on their cross-cutting and geometric relationships to other structures (Figure 3e–h). The second stage vein (V2) formed after the formation of the first stage vein (V1). The second stage vein (V2) is being cross-cut by the third stage vein (V3) (Figure 3d), which suggests that the formation of V2 was later followed by the third stage vein (V3). V3, which is the youngest among all the veins is highly mineralized. This is because it was co-emplaced with the gold mineralization.

4.5.4 Proposed simplified deposit genetic model

The Edikan Gold deposits developed in granitoids that intrude the metasedimentary basement (Figure 12). Like all other orogenic gold deposits, the Edikan Gold deposits were formed along subduction zone settings. It is proposed that subducted slab-derived fluids intersected deep crustal faults or shear zones. The subducted slab-derived fluids are composed of overpressured auriferous fluids, which ascended via the magma chamber to the surface. As these fluids interacted with the cold metasedimentary wall rocks, their physico-chemical characteristics changed, which resulted in the formation of a gold anomaly at the contact aureole (Figure 12). Some of the overpressured auriferous fluids passed through hydraulically fractured rock bodies along a brittle–ductile transition zone interacting with greywackes and graphitic schist, volcanic rocks, and some mafic lithologies in the area. This may have been the reason why some of the greywackes in the area appear to be mineralized. Mineralization in the metasedimentary lithologies occurred in the upper greenschist facies (Figure 12).