Plate tectonics is the framework through which we understand the large-scale Phanerozoic history of Earth. The question of when and how plate tectonics began remains the subject of debate, in no small part because through subduction, plate tectonics destroys much of the evidence of its earlier activity. Estimates of the onset of plate tectonics vary from the Hadean (Hopkins et al., 2008), to the Archean (Brown, 2006), to the Neoproterozoic (Stern 2005, 2008). There is no rock record from the Hadean, and only a limited rock record from the Archean. Thus, it is unlikely that we will determine whether any deformation recorded during this time period was part of a globally connected plate boundary system or a regional, transient event.

Spherule beds are preserved within Archean age rocks in the Barberton greenstone belt, South Africa (Lowe and Byerly, 1986; Lowe et al., 1989) and the Pilbara craton, Australia (Glikson et al., 2016). Archean spherule beds formed from the distal ejecta of large bolide impacts. These beds contain important clues regarding their formation—the thickness of the beds can be used to estimate the size of the impactor (Johnson and Melosh, 2012). Lowe et al. (2014) described four additional spherule beds and placed their formation at the same time as the first major episode of orogeny and crustal deformation in the Barberton greenstone belt (3.26–3.23 Ga). Lowe et al. further suggested that these impacts may have been the trigger that initiated the modern plate tectonic regime. A new contribution by O’Neill et al. (2020, page 174 in this issue) uses the characteristics of these recently described spherule beds to constrain the size and velocity of the impactors that formed them, extending the Archean impact record. They then use the Archean impact record as input to geodynamic models to test Lowe et al.’s hypothesis that these impacts could have initiated a modern style of plate tectonics.

Lowe et al. (2014) were not the first to postulate that the Archean greenstone belts record plate tectonic activity. There are multiple lines of evidence that plate tectonics may have been operating in the Archean, including apparent polar wander curves (O’Neill et al., 2007), felsic volcanism consistent with melting of a water-rich source, and isotopic systematics similar to modern-day arcs (Hugh Smithies et al., 2018; O’Neill et al., 2018). The absence of clearly identified fold-and-thrust belts, tectonic mélanges, or ophiolites in the Archaean rock record casts doubts on the subduction interpretation (e.g., Stern, 2005; Moyen and van Hunen, 2012).

Geodynamic calculations have become an important hypothesis-testing tool when combined with the Precambrian geological record (c.f., van Hunen and Moyen, 2012: O’Neill et al., 2018; Stern and Gerya, 2018). These models are based on basic laws of physics, including the conservation of mass and energy, as well as Newton’s second law, sometimes misleadingly described as the conservation of momentum. To reduce the number of variables and create a set of equations that can be solved, a set of constitutive equations are required. For mantle convection, stress and strain rate are related through the viscosity. The viscosity of silicate minerals depends on temperature, pressure, composition, stress, grain size, water content, and history (c.f., King, 2016). Our understanding of viscosity is limited, in part due to the challenge of measuring the viscosity of silicate minerals at high pressures and temperatures, and the reality that the strain-rates achievable in such laboratory measurements must be extrapolated by orders of magnitude to mantle conditions. Geodynamic calculations are built upon solid physical principles; the calculations are limited in so far as our understanding of mantle viscosity is limited, and the appropriateness of the initial and boundary condition choices.

Several modes of surface behavior are recognized in geodynamic models. In stagnant-lid convection, the lithosphere is immobile with surface heat flow limited by conduction. In mobile-lid convection, the lithosphere is part of the convecting system, cooling as it advects along the surface, and sinking back into the warmer mantle. All other factors being equal, a stagnant-lid planet will have a hotter mantle than a mobile-lid planet. The transition from stagnant-lid to mobile-lid tectonics in geodynamic modeling has enriched our understanding of plate tectonics on Earth. As the mantle becomes hotter than at present day, many models show that plate-like behavior becomes episodic, with alternating periods of mobile-lid and stagnant-lid behavior (c.f., van Hunen and Moyen, 2012). An implication of these models is that evidence for plate tectonics may appear and disappear in the geological record, and subduction may repeatedly fail (O’Neill et al., 2018).

The primary force driving plate tectonics is the negative buoyancy in subducted slabs (Forsyth and Uyeda, 1975). It is unclear what additional processes could produce the large forces necessary to initiate subduction on a pre–plate tectonic planet. To identify and test candidate processes, researchers have modeled the arrival of large plumes under the lithosphere (Gerya et al., 2015), magmatic weakening and volcanic loading (Moore and Webb, 2013; Nakagawa and Tackley, 2014), and bolide impacts (O’Neill et al., 2017). In many cases, subduction is transient, with the mantle reverting to a stagnant-lid state after a relatively brief time interval.

In a previous study, O’Neill et al. (2017) examined the effect of bolide impacts on a Hadean Earth, finding that the thermal anomalies produced by extremely large impacting bolides (>∼700 km in diameter) induce mantle upwellings that are capable of driving transient subduction events. These transient events terminate because the hotter Hadean mantle had a lower viscosity than the present-day mantle, reducing the coupling between the mantle and lithosphere (O’Neill et al., 2007), and the higher temperatures weakened the core of the subducting slab, resulting in necking and breakoff that reduces the slab pull force on the plate (van Hunen and Moyen, 2012).

The present O’Neill et al. study (2020) addresses whether the estimated size and frequency of Mesoarchean impacts could have initiated subduction, and whether these events could have developed into a globally connected plate boundary network that continued without interruption to the present. There are several significant differences between the Hadean and Archean environments that will directly impact the geodynamic modeling. The Archean mantle should be cooler than the Hadean mantle (c.f., Christensen, 1985) and there should be smaller, less-frequent bolide impacts in the Archean when compared to the Hadean (Bottke et al. 2012). Thus, it is not possible to rescale the results of O’Neill et al.’s (2017) Hadean Earth study to address impact-induced subduction in the Archean.

The two studies by O’Neill et al. (2017, 2020) are among the first studies to model the role of bolide impacts on planet-wide tectonic behavior applied to Earth. Both studies concluded that, if initiated, subduction would be a transient event, lasting only tens of millions of years, and leaving open the question of when plate tectonics (as the globally connected plate boundary network that we recognize today) took hold. The O’Neill et al. (2017, 2020) results are at odds with the results of Foley et al. (2014). The reason for the discrepancy may be that O’Neill et al. used a yield-stress formulation while Foley et al. (2014) used a formulation where the plate boundaries are weakened by grain-size reduction (Bercovici and Ricard, 2014).

Foley (2018) suggested that the difference between the two formulations is related to how they respond to changes in lithospheric stress. In yield-stress formulations, when the lithospheric stress decreases, the stress level at the boundaries may no longer exceed the yield stress, resulting in lithospheric stagnation. In contrast, when lithosphere stress drops with increasing mantle temperature or heat production rate, the deformational work, which drives grain-size reduction, increases. Thus, in grain-size reduction formulations, the ability to form weak plate boundaries is not impeded by early Earth thermal conditions. Both formulations are based on sound physical principles. Yet it is not clear how each method applies to the complex environment of subduction zones. Our understanding of subduction initiation is limited by our understanding of the process by which a stagnant lithosphere begins to deform.

Some may wonder whether appealing to a bolide impact as a trigger for subduction initiation is truly necessary. It is clear from the other inner Solar System bodies (Mercury, Venus, Mars, and the Moon) that bolide impacts were a significant process in the inner Solar System during the Hadean and Archean (Bottke et al., 2012). Thus, bolide impacts played a significant role in the Hadean and Archean on Earth. Geodynamicists assumed that because Earth’s mantle is convecting well above the critical Rayleigh number, over the course of Earth history the initial state of the mantle will be long forgotten. Weller and Lenardic (2012) showed that models with identical parameters starting from different initial conditions may end up in different states (e.g., stagnant-lid versus mobile-lid convection). When and whether a planet exhibits plate tectonics are likely functions of the initial state and history of the planet.

When plate tectonics began has a fundamental control on the thermal evolution of Earth, because planets lose heat more effectively with a mobile lid than a stagnant lid. Thus, the earlier plate tectonics began, the further Earth has cooled from its initial state. The hypothesis that plate tectonics is necessary to create the kind of stable surface environment necessary for a habitable planet is of interest to the exoplanet community (Kasting and Catling, 2003).