Metallogenic model of the Hongqiling Cu–Ni sulfide intrusions, Central Asian Orogenic Belt: Insight from long-period magnetotellurics

3.1 Data acquisition and processing

Long-period MT data were collected along a north–south-trending section (251 km) from the North China Craton (NCC) to Song Liao Basin (SLB), which vertically cut across the east–west-directed accretion of the East Asian Pre-Pacific Province (Figure 2). This included 22 MT sites at intervals between 3 and 15 km.

Figure 2

Topography and main tectonic boundaries in the central Jilin area. The red rectangle outlines the Hongqiling Cu–Ni deposit. The black circles indicate the long-period MT site locations of the section that originates in the NCC and passes through the middle of the Zhang Guang Cai Ling (ZGCL), before extending to the SLB.

Long-period MT data were acquired using an Aether 200 instrument (Crystal Earth Co., Ltd., USA) with a collection time of no more less than 10 h and a period range of 0.001–1,000 s. Then, this time series are processed by a statistically robust algorithm [21]. The data from each MT site include five components: two horizontal electric field components (E _x and E _y ), two horizontal magnetic components (H _x and H _y ), and one vertical magnetic component (H _z ). Information regarding lithospheric geoelectric structure can be embodied by the apparent resistivities (ρ _a) and phases (φ) [22]. Figure 3 displays the apparent resistivity (R _xy or R _yx ) and impedance phase (φ _xy and φ _yx ) curves of partial sites and the error bars are corresponding to the measurement errors of the observed data. The MT skin depth is the depth at which the MT field attenuates to (1/e) of its amplitude at the surface of the Earth [23] and can be calculated by equation (1):

where δ, ρ, μ, ω, and f represent the skin depth, bulk resistivity, magnetic permeability, angular frequency, and frequency, respectively. In geological terrains with a high resistivity, the MT signal has a larger penetration depth [24,25]. In this study, the bulk resistivity was estimated to be about 300 Ω m, and the lowest frequency was estimated to be 0.003 Hz. The skin depth and effective detection depth were nearly 160 and 100 km, respectively.

Figure 3

Curves of partial long-period MT sites. (a–c) Observed resistivity curves for sites 1, 4, 17, and 20 from the transverse electric (TE) (R _xy ) and transverse magnetic ™ (R _yx ) model apparent resistivity curves. (a′–c′) The impedance phase curves of the TE (φ _xy ) and TM (φ _yx ) models.

3.2 Analysis and inversion

After data processing, the phase-impedance tensor decomposition is used to estimate the regional dimensions [22]. The pseudosection map of resistivity inversion in Figure 4 shows that the 2D deviations vary with frequency. The value of the 2D deviations is generally less than 0.3 over different frequencies [26]. Only a few sites in low frequencies exhibit large 2D deviation values. This result indicates that the long-period MT section is generally compliance with 2D inversion [27].

Figure 4

Pseudosection map. The 2D deviations vary with frequency.

The geoelectric strike can be identified by the rose diagram [22]. Data in the frequency bands 100–1 Hz showed geoelectric strikes in the range N50°E–N60°E and S30°E–S40°E. Data in the frequency bands 0.1–0.001 Hz showed geoelectric strikes in the range N60°E–N80°E and S10°E–S30°E (Figure 5). The NE-NNE regional geological strike directions correlate well with the geoelectric strike. Moreover, the distribution of Carboniferous to Permian subduction-related igneous rocks have been found to be generally related to the north–south-directed subduction that accompanied the east–west closure of the Paleo-Asian Ocean [2]. Thus, the inherent 90° ambiguity could be resolved well and the data were rotated into a S20°E direction [28].

Figure 5

Results of the tensor decomposition and rose diagram of the main electrical axes. The red wedges correspond to the two solutions for the strike direction.

The inversion was consistent with the 2D nonlinear conjugate gradient (NLCG) inversion in the TE and TM modes [29]. The inversion models were discretized by a grid size of 46 × 90 in the X and Y directions with an initial uniform half-space resistivity of 100 Ω m. To ascertain the optimum value for the root mean square (RMS) misfit and the roughness for the 2D NLCG inversion, a number of inversions were performed [25]. We finally choose the parameter τ = 10 for the regularization factor [22]. After 200 iterations, the model responses are well fitted to the observed data, and the preferred model is obtained with an average normalized RMS of 2.12 for all sites.

3.3 Model test

To verify that the preferred model was properly inversed, the long-period MT result was test by some forward modeling (Figure 6). First, the C1 and C2 anomalies are displaced by 100 and 300 Ω m in forward model 1 and model 2. Then, forward calculation was adopted to assess the impact. The average normalized RMS misfits increase from 2.32 to 3.24 in forward model 1, and this trend is more sharply in forward model 2.

Figure 6

Model assessments verify the sensitivity of MT data. In (a)–(c), the preferred inversion model result; the forward model 1 which displaced the C1 and C2 anomalies with the 100 Ω m; the forward model 2 which displaced the C1 and C2 anomalies with the 300 Ω m.

Figure 7 shows the results of 12 and 13 sites data fitting. The preferred inversion fitting curve (the purple) fits well with the observed data, while the modified model response curve (the green and red) cannot fit well under the period of 10 s. The forward model test results indicate that reliability of C1 and C1 anomalies in preferred inversion.

Figure 7

Site-by-site RMS misfits for the preferred model, the modified model 1 and the modified model 2. (a) The RMS of each site. In (b) and (c), the inversion results of the resistivity and phase curves. The red and blue triangles represent the observed Ryx and Rxy model data. The blue curve, the green curve, and the red curve represent the preferred inversion result, the model 1, and the model 2 results, respectively.

3.4 Interpretation of the MT profile result

The high-resolution lithospheric electrical properties were inversed using the 2D NLCG method and long-period MT data (Figure 8). The lithospheric geoelectric structure was found to be generally consistent with the major surface geological features. Based on the locations of regional faults and the similarities and differences in the inversion model, the long-period MT section was laterally divided into three segments: the NCC, ZGCL, and SLB [30] (Figure 8).

Figure 8

(a) Elevation of the long-period section. (b) 2D resistivity model obtained following NLCG inversion of the TM data. The gray curve represents the topographic relief, the black triangle represents the MT site position, and the black cross represents the inversion depth. The low-resistivity geological bodies are represented by C1, C2, C3, C4, C5, C6, and C7, while the high-resistivity geological bodies are represented by R1 and R2.

In the ZGCL segment, at sites 7–10, a sudden change in the resistivity was observed, where the value decreased by one order magnitude from 700 to 60 Ω m. According to the previous research, this structure is understood to be loose, fractured, and filled with some low-resistivity filling materials [31] (Figure 8). Moreover, seismic imaging revealed a transition zone between high- and low-velocity anomalies, which is consistent with the site of the long-period MT survey. Thus, we infer that this structure, exposed near site 7, dipped steeply to the north is the Dunhua–Mishan fault [30]. At sites 7–11 and sites 15–17, two low-resistivity anomalies (C1 and C3) were observed with the resistivity values of 10^1.8 and 10^1.6 Ω m in the upper crust (Figure 8). The petrophysical parameter of minerals such as the ore-bearing amphibolite pyroxenite, altered gabbro, and Cu–Ni sulfides are characterized by low-resistivity anomalies (Table 1). In addition, according to the deep seismic-reflection sections in the CAOB, the short and strong reflections represent multiple volcanisms [32]. Thus, we infer that the low-resistivity anomalies C1 and C3 may represent the relics of ancient shallow magma chambers emplaced in the Triassic.

Vertically, the lithospheric geoelectric structure could be generally segmented into three layers. The first layer with an resistivity value ranging from 10^2.5 to 10^4.1 Ω m (except for conductor C1) represents crust at a depth from 28 to 40 km [32]. The crustal structure beneath northeastern China as imaged by the Northeast China Extended Seismic (NECESS) Array receiver indicates that the Moho depth varies from 26.7 to 42.3 km from the NCC to the SLB [33]. Thus, we argue that the first geoelectric layer represents the Moho depth (Figure 8). In Section 3.1, we have calculated the skin depth was nearly 160 km. Thus, the second layer with a value of 10^2.0 to 10^2.5 Ω m at depths from 52 to 100 km represents the lithospheric mantle, and the third layer with a value of 10^1.8 to 10^2.0 Ω m represents upwelling asthenosphere. Geomagnetic three-component depth sounding and terrestrial heat-flow data acquired for the region between Changchun and Tonghua had similar trends to that in our data [34]. Furthermore, previous deep exploration investigations showed that the lithosphere–asthenosphere boundary (LAB) depths in this region are marked on the MT section [35] (Xiong et al., 2011, p. 31). In addition, considering the fact that the study area is located near the Solonker Suture and far away from the Jiayin–Mudanjiang Suture, we infer that its geoelectric structure should be closely related to the evolution of the Paleo-Asian Ocean rather than the evolution of the Paleo-Pacific Ocean. Moreover, Jurassic magmatism (185–158 Ma) in the Hongqiling deposit has been found to be dominated by granitoids with few mafic–ultramafic rocks, which indicates that the degree of upwelling of the asthenosphere and the reworking of the LAB were quite limited during the Late to Middle Mesozoic evolution [3].

4.1 Partial melting of the mantle

Ore-fertile magma formed by high degree of partial melting of the mantle generally has high Ni/Cu ratio and low Pd/Ir ratio [6]. In geophysics, minute quantities of partial melting can have a tremendous influence on the overall electrical performances [37]. Thus, resistivity can be used to estimate the volume of fluid/melt that is present in the mantle, although the geothermal model of the lithosphere requires simplification prior to calculating the degree of partial melting of the mantle. The geotemperature is 1,300 K at the depth of 50 km, and then, the geothermal gradient increment is 0.4 K/km [33] (Figure 9). The resistivity of olivine, which has a variable fluid content that ranges from 0 to 1,000 ppm, can be obtained using the Grades laboratory database [38]. The Grades laboratory database states that the resistivity of dry olivine with a olivine fluid content of 350 ppm ranges from 60 and 120 Ω m (conductivity of 0.0083–0.0167 S/m) (Figure 9). However, at the same depth, the long-period MT section had a resistivity of ∼50 Ω m (conductivity of 0.02 S/m) between sites 10 and 13, which had lower resistivity than that the olivine fluid content of 350 ppm (Figure 9). The preferred explanation is that Hongqiling ore-bearing mafic-ultramafic intrusions are generally produced by the partially melting of mantle.

Figure 9

Lithospheric mantle resistivity; at a depth of 50–400 km, and a fluid content of 350–1,000 ppm, as computed from laboratory results of Karato and Dai [36].

4.2 Mineralization model of Cu–Ni sulfide deposits

Long-period MT section successfully imaged the Dun-Mi structures which formed the magma conduit system for mantle-derived metalliferous magmas intrusion and fluids migration through the crust, as well as the shallow chamber (C1), the deep chamber chamber (C2), and the emplacement of the ore-bearing intrusions.

Positive ε_Nd(t) value (3–5) of Hongqiling mafic-ultramafic intrusions indicated that [6]. The long-period MT section imaged the partial melting magma intruded the middle crust and formed a shallow chamber (C1) (Figure 10a). Wu et al. reported that the extensive distribution of Late Triassic A-type granites was the result of the direct partial melting of the juvenile crust [39]. We note that upwelling in the asthenosphere can realize such high temperatures and that mafic–ultramafic intrusions are generally coeval with the A-type granites in this region [3]. Thus, we suggest that the deep, separated chamber may have assimilated the SiO₂-rich A₂-type granites that were formed by underplated mantle-derived magma and that this triggered the PGE sulfide segregation prior to emplacement. The PGE-depleted magma subsequently intruded into the shallow chamber (C1) and then underwent variable degrees of crustal contamination before generating the second-stage sulfide segregation [6] (Figure 10b). In summary, the mineralization process involved voluminous, undersaturated, sulfide-mafic magma passing through the large magma plumbing system, where it reached sulfur saturation due to the addition of additional crust materials and triggered sulfide segregation [3]. The lithospheric-scale Dun–Mi structure and its secondary structures control the magma conduit system in the study area (Figure 10c).

Figure 10

Simplified schematic representation of the tectono-magmatic evolution and mineralization in the Hongqiling ore-bearing intrusions. (a) Long-period MT section images of the low-resistivity bodies (C1 and C2). (b) Geological interpretation based on the long-period MT results. (c) The processing of multi-time immiscible segregations in the deep chamber.

The top surface of the low-resistivity body in the upper mantle was consisent with the LAB. The trend of gradual thinning in the first layer indicates that a rapid uplift event occurred in this region. The depth variation in the third layer reflects the various degrees of asthenospheric upwelling, which suggests an extensional geodynamic regime that was probably induced by delamination. The volume of mafic–ultramafic magma in the shallow and deep chambers (C1 and C2) was found to be much larger than the volume of surface ore-related intrusions. This indicates that the strong extensional geodynamics regime in the study area is associated with intensive mantle-derived magmatism.

Three potential geodynamic processes could have formed such a large volume of mafic–ultramafic magma and the associated huge magmatic Cu–Ni–PGE deposits: (i) a mantle plume [39]; (ii) slab breakoff [40]; and (iii) lithospheric delamination [3]. First, it is impossible that a mantle plume existed during the Triassic when the CAOB was strongly associated with the evolution of the Paleo-Asian Ocean. Second, unlike delamination, slab breakoff cannot trigger a significant thermal perturbation in the overriding lithosphere; thus, slab breakoff cannot account for the intensive magmatism or additional diverse nature of the magmatic observations [41,42,43]. Therefore, the generation of intensive mafic–ultramafic magmatism was triggered by the intensive delamination of the lithosphere. Besides, the long-period MT section from the south Inner Mongolia to the Bainaimiao provided an image of the residual lithosphere in the asthenosphere [13]. Thus, we inferred that the genesis of the Hongqiling Cu–Ni deposits is related to lithospheric delamination.