Osmium isotopes record a complex magmatic history during the early stages of formation of the North American Midcontinent Rift — Implications for rift initiation

Although the magmatism in the North American Midcontinent Rift System (MRS) is generally accepted to have formed via partial melting of the Keweenaw Plume (i.e., the plume that impinged on the MRS), the long duration of this magmatism and paleomagnetic evidence has led some authors to suggest a passive rifting model. Discerning between these two models (i.e., active vs. passive rifting) has been challenging, however, given that many of the earliest volcanic and intrusive rocks of the MRS have been variably contaminated by crustal material, masking the relative contributions of magmas derived from the plume and subcontinental lithospheric mantle (SCLM). Understanding these contributions, however, is critical to developing a genetic model for the early history of the MRS. Here, we assess these contributions by combining new ReOs isotope data of early MRS diabase sills (Nipigon, McIntrye, Inspiration sills), mafic–ultramafic sills (Shillabeer and Jackfish sills) and intrusions (Eva Kitto and Seagull), and subaerial volcanics (Wolfcamp and Coubran basalts) with previously determined SmNd and RbSr isotopes, and whole-rock geochemistry. The γOsi values of the early MRS rock suites range from slightly depleted (−10) to highly radiogenic (3857), with Initiation Stage melts (Eva Kitto and Seagull intrusions, Jackfish sill) exhibiting the smallest variability (4 to 50) and Early Stage melts (Nipigon sills, Wolfcamp basalt, and Thunder intrusion) exhibiting the greatest variability (−10 to 3857). This range of Os isotope compositions could not have resulted from variable degrees of partial melting, garnet retention in the mantle, or crystal fractionation as the values do not correlate with proxies for these processes (e.g., La/Sm, Gd/Yb, and MgO, respectively). Considering the variably elevated Th/La–Th/Nb and radiogenic εNdi–Sri values, it is possible that SCLM and/or crustal material contributed to the magmatic evolution of these rock suites. Based on mixing models in γOsi–εNdi space, the negative εNdi and slightly radiogenic γOsi of the mafic–ultramafic intrusions and sills indicates that they likely crystallized from hybrid magmas representing mixtures of enriched SCLM-derived melts with lesser plume-derived melts; these hybrid magmas were contaminated by limited amounts (<10%) of crustal material during emplacement. An SCLM source for these early MRS intrusions implies that rifting of the North American continent likely initiated passively. Given their negative to highly positive γOsi values, the diabase sills and subaerial lavas are suggested to have crystallized from melts derived from a depleted source region of the Keweenaw Plume and, with the exception of those that formed Wolfcamp basalt, were contaminated to variable degrees by crustal material. Although variable contamination can explain the negative εNdi values, it cannot independently explain the range of γOsi values because i) not all of the rock suites are contaminated (e.g., Wolfcamp basalt) and ii) the highly radiogenic γOsi values would require unrealistically high degrees of contamination. Rather, the variability likely also stems from segregation of sulfide liquid or crystallization of Os-bearing platinum-group minerals. This study demonstrates that the early MRS rock suites did not all crystallize from plume-derived melts as may be expected in a model whereby rifting initiated as a result of plume impingement. Rather, the earliest rock suites (>1109 Ma) crystallized from melts derived from the SCLM, indicating that rifting likely initiated passively.