FORC principal component analysis of coarse-grained samples: whole rock and mineral separates from the Bushveld Complex, South Africa

Igneous rocks, particularly those that cooled slowly, may contain a range of magnetic phases, such as acicular/bladed exsolution in silicates as well as discrete grains. First-order reversal curve (FORC) analysis is a powerful tool to characterize such complex magnetic signals, but is not widely applied to coarse-grained intrusive rocks. To test the validity and applications of the method with such samples, we used principal component analysis of FORC distributions (FORC-PCA) to resolve magnetic "fingerprints" in samples of layered mafic rocks from the Bushveld Complex, South Africa. The gabbro-norites of the Main Zone of Bushveld contain plagioclase feldspars that are commonly clouded by abundant magnetite needles, and pyroxenes with oxide blades of variable size and compositions. The co-existing magnetite grains exhibit a variety of microstructures, including oxy-exsolution of ilmenite and spinodal decomposition of ulvöspinel. In addition to the whole-rock specimens measured for FORC-PCA, plagioclase and pyroxene mineral separates were isolated to compare to the endmember models.

FORC-PCA of the whole-rock specimens provides 4 endmember FORC diagrams, of which 2 are physically valid and interpreted as non-interacting single-domain (SD) and vortex state signals. The SD signal dominates the signature of the plagioclase mineral separates, indicating that this phase can be identified and quantified in whole-rock measurements. The remaining 2 endmembers in the whole-rock samples are not individually physical, but can be summed to generate a realistic multidomain (MD) FORC diagram. This MD signal is attributed to discrete magnetite grains with possible contributions from larger blades in pyroxene. Because the MD signal is composed of independently varying constituents, it may be possible to relate these variations to the concentration of these phases, or to changes in the coarseness of exsolution microstructures. The consistency of the endmember models with FORC distributions of mineral separates indicates that FORC-PCA provides quantitative results when applied to suites of related, coarse-grained specimens. It may therefore prove a valuable tool to screen for stable paleomagnetic recorders, and potentially provide evidence of local variations in petrogenetic conditions and processes.