Photoferrotrophic Bacteria Initiated Plate Tectonics in the Neoarchean

Plate tectonics distinguishes Earth from the other terrestrial planets but its initiation mechanism and onset time are debated. We propose plate tectonics was initiated by the deposition of magnetite‐rich banded iron formations (BIFs) through biogeochemical iron cycling in Neoarchean oceans. In the photic zone of proto‐continental margins, photoferrotrophic bacteria efficiently oxidized the dissolved Fe(II) and induced massive precipitation of ferric oxyhydroxide, which would rapidly react with Fe(II)‐rich hydrothermal fluids from coeval vigorous volcanism in Neoarchean oceans to produce magnetite‐rich BIFs. Mechanical models demonstrate that the localization of high‐density BIF deposition near proto‐continents induces collapse of the lithosphere and can initiate the earliest subduction. The peak deposition of BIFs in 2.75–2.40 Ga provides a time constraint on the inception of plate tectonics.

low-density (∼2,400 kg/m 3 ) silicate sediments cannot rupture a mature passive margin (Cloetingh et al., 1982). Here, we argue that the banded iron formation (BIF), an early Precambrian iron-rich sedimentary rock, could lead to the requisite conditions through a combination of their high density and sequence thicknesses (Konhauser et al., 2002;Li et al., 2017;Robbins et al., 2019) (Figure S1 in Supporting Information S1). The BIF deposition peaked during 2.75-2.40 Ga and was distributed globally on the Neoarchean continental margins (Bekker et al., 2010;Konhauser et al., 2017), but their potential as the driving force of subduction initiation has not been evaluated.
BIFs are Precambrian marine deposits composed of alternating iron-oxide and chert bands (Klein, 2005). It is widely believed that the oxidation of Fe(II) in the highly anoxic Archean oceans is dominated by anoxygenic photoferrotrophy, although oxygenic photosynthesis may also contribute Thompson et al., 2019). With the nutrient input from the proto-continent (Hao et al., 2020), photoferrotrophs thrived in the photic zone on the proto-continental margins, which have the highest contemporary primary productivity (Kappler et al., 2005;Konhauser et al., 2002;Thompson et al., 2019;Wollast, 2002). By harnessing the sunlight energy, the photoferrotrophic bacteria enzymatically oxidized Fe(II) to ferrihydrite (∼Fe(OH) 3 ) and fixed CO2 into biomass (Thompson et al., 2019) (Figure 1). Upon diagenetic solidification, ferrihydrite dehydrated and transformed to either hematite or magnetite in BIFs, with different mineralization pathways and condensed states (Li et al., 2017;Sun et al., 2015). Moreover, the iron oxides in Archean BIFs are mainly magnetite, and the hematite content has gradually exceeded the magnetite content since the Proterozoic (James, 1992;Li et al., 2017). The hematite in BIFs was directly dehydrated from its ferrihydrite precursor, maintaining an ultrafine crystal size (3-5 nm) and a high specific surface area (∼200 m 2 /g) so that it remained fluffy, similar to its oxyhydroxide precursor when settled on the seafloor (Sun et al., 2015). On the other hand, the active submarine volcanism and the released hydrothermal fluids in the Neoarchean promoted the formation of magnetite. These high Fe(II)-containing hydrothermal fluids would react with ferrihydrite on the seafloor to rapidly produce magnetite (Halevy et al., 2017;Li et al., 2017) (Figure 1). In contrast, magnetite nanoparticles make dense aggregation by magnetic interactions (Rebodos & Vikesland, 2010) (Figure S2 in Supporting Information S1). The magnetite would accumulate with silica to form thick BIFs on proto-continental margins. Also, these dense BIFs typically have a density 20%-40% higher than the lithosphere due to their high iron content (Konhauser et al., 2002;Thompson et al., 2019). Through their negative buoyancy, such BIF-hosted continental margins eventually develop into deep sedimentary basins (Figure 1), such as the Hamersley Basin in Western Australia (Trendall, 1983). In addition to their dense state and high density, another important observation is the peak deposition of BIFs (2.75-2.40 Ga) (Bekker et al., 2010;Klein, 2005;Konhauser et al., 2002). Many large BIF-hosted basins (also called Superior-type BIFs) developed in this period have been identified (Bekker et al., 2010). They are found to be deposited in near-shore shelf environments and localized on continental margin basins Trendall, 1983), which implies a widespread distribution of BIFs on global continental margins. Moreover, the depositional scales of BIFs were large and several giant BIFs in this period (2.75-2.40 Ga) were reported, such as the Hamersley Basin in Western Australia and the Transvaal Supergroup in South Africa. These two BIFs alone cover an area of over 10 5 km 2 and their thickness can reach kilometers (Robbins et al., 2019;Trendall, 1983).
During the peak deposition of BIFs, a depositional rate of ∼1 km∕Myr has been suggested by geochronological studies (Konhauser et al., 2002). This is equivalent to an iron depositional rate of ∼45 mol ⋅ m −2 ⋅ yr −1 for the ironrich BIFs with Fe wt% = 54.6% and = 4600 kg∕m 3 (Konhauser et al., 2002), which is close to the estimation given by recent laboratory experiments (Kappler et al., 2005;Thompson et al., 2019). Assuming all the Fe(III) in BIFs is the oxidation product of photoferrotrophy and using the estimated iron-oxidation rate of the photoferrotrophic bacteria (1 − 3 × 10 −11 mol ⋅ cell −1 ⋅ yr −1 ) (Konhauser et al., 2002;Maisch et al., 2019), a cell density of 3 − 9 × 10 4 cells∕cm 3 over a 50-m-deep photic zone is adequate to sustain the above-mentioned BIF depositional rates. Such a cell density should be feasible since modern observations find that the bacterial population in the photic zone is typically on the order of 10 6 cells∕cm 3 (Konhauser et al., 2002).
The high depositional rate (∼1 km∕Myr ) implies that thick magnetite-rich BIFs would accumulate along continental margins within a few million years. The loading from these heavy sediments could be large enough to induce the rupture and subduction of the lithosphere along proto-continental margins. Based on these observations, we propose that the peak deposition of the heavy magnetite-rich BIFs in 2.75-2.40 Ga might have initiated the earliest subduction, thus indicating the origin of plate tectonics.

Geodynamic Models
We formulated 2-D thermomechanical models with viscoelastoplastic rock rheology to test the hypothesis that BIF deposition initiated subduction in the Neoarchean. Our reference model ( Figure 2) consists of a 100-km-thick proto-continental plate (left side, 1,100 km wide) and a 100-km-thick oceanic plate (right side, 2,100 km wide). Presumably, crustal thickness variations before plate tectonics would have been gradual as they would have partially relaxed over millions of years. We assumed that before plate tectonics that there were regions of proto-continent composed of crust like tonalite-trondhjemite-granodiorite rocks with a density of ∼2,900 kg/m 3 (Merriman et al., 2013). Consequently, a transition (from x = 1,000 km to x = 1,100 km) between the proto-continent and the oceanic domain is set ( Figure 2a). The crustal thickness of the proto-continent and the oceanic plate is 30 and 15 km, respectively. A 40-km-thick "sticky air" layer above the lithosphere approximates the free surface (Crameri et al., 2012). The initial temperature field of the proto-continental and oceanic lithosphere is linearly distributed between 0 • C and 1,550 • C , while the sublithospheric temperature is homogeneous (1,550 • C ) and 200 • C higher than the modern value (Rey et al., 2014) (Figure 2a). The width of the BIF-depositional area 0 beneath the proto-continental margin is nominally set to 250 km (i.e., from x = 1,100 km to x = 1,350 km in Figure 2a), which is comparable to the basin width on modern mature passive margins (Cloetingh et al., 1982). BIF material is deposited on proto-continental margins at specified depositional rates ( Figure 2). In this study, we assume that the photic zone on the Neoarchean continental margins dominated contemporary primary productivity, consistent with modern observations (Wollast, 2002). In the model, photoferrotrophic bacteria need to take up nutrients transported off the proto-continental margin (Hao et al., 2020) and produce BIF sediments. Their accumulation causes the localization of the modeled BIFs at the proto-continental margin. We established a 1-D ocean-mixing model (see Supporting Information S1) to determine the spatial distribution of the depositional rate ( ) (Figure 2g). With a specific deposition rate ( ) over a time Δ , BIF sediment of thickness, ( ) , is placed on top of the solid surface replacing the sticky air material: where Δ is the time interval to deposit BIFs. With a non-uniform depositional rate, a horizontally averaged value ave is used when comparing models. The yield stress of the marine sediments is believed to be low and we use the value determined from previous studies (Davis & von Huene, 1987) in the model (Table S1 in Supporting Information S1). Other details about the model can be found in Supporting Information S1.

Results of the Reference Model Case
For the reference model ( Figure 2 and Figure S3 in Supporting Information S1), enough BIF sediments accumulate within ∼9 Myrs to rupture the lithosphere when the peak BIF depositional rate is maintained at 1 km/Myr. Initially, the sedimentary loading is limited to moderate plate deflections ( Figure 2b). However, as the sediments are weak, they flow horizontally, increasing the downward deflection of the lithosphere, which in turn leads to more localization of the high-density BIFs. The thickness of the sediments rapidly evolves away from the profile imposed by the ocean-mixing model (Figure 2g) to one with a more Gaussian shape (Figure 2b) and eventually one which is even more localized (Figure 2c). As the thickness of BIF sediments increases, the load eventually exceeds the yield strength of the lithosphere and the margin collapses (Figure 2c). After several million years, a retreating subduction slab starts at the position of plate rupture (Figure 2d). In the computations, when the lithosphere breaks and sinks into the mantle, BIF deposition shuts down and the horizontal-average BIF thickness at this time is recorded as the critical thickness (see Supporting Information S1 for details), which represents the minimum BIF thickness required to break the lithosphere.

Required BIF Thickness for Subduction Initiation With Different Mechanical Parameters
Compared with the reference model (Figure 2), thicker BIF sediment is required to induce subduction when either the yield stress or the thickness of lithosphere is increased (Figures 3a and 3b). When the maximum yield stress of the lithosphere increases from 150 to 600 MPa, the critical thickness of the BIF to rupture a 100-km-thick lithosphere varies from ∼5 to ∼12 km (Figure 3a). When the yield stress of the lithosphere is 300 MPa as recently estimated (Rey et al., 2014) and the lithospheric thickness increases from 75 to 125 km, the critical thickness changes from ∼5 to ∼11 km (Figure 3b). The thickness of the proto-continental crust can significantly influence the buoyancy and strength of the lithosphere, but its value before plate tectonics is poorly constrained. We change the thickness from 20 to 40 km and the corresponding critical thickness of BIFs varies from ∼12 to ∼5 km ( Figure 3c). Different densities of BIF from 4,600 kg/m 3 (iron-rich BIF, Fe wt% ≈ 55%) to 3,800 kg/m 3 (Fe wt% ≈ 35%) are tested, and the related critical thickness varies from ∼9 to ∼14 km (Figure 3d).
On modern passive margins, the thickness of the low-density silicate sediments can reach 15-20 km ( Figure  S4a in Supporting Information S1). However, with the deposition of BIFs, such a thickness is sufficient to induce subduction. In fact, the negative buoyancy of BIFs would drive it to subside continuously and deepen the basin (Trendall, 1983), so the upper bound of sedimentary thickness should be higher. Based on the computations and observed characteristics of BIFs in the Neoarchean (including its high density and depositional rate), magnetite-rich BIFs can trigger subduction over only several million years (∼10 Myrs).

Sensitivity Test for Other Parameters
The outcomes may also be affected by the degree to which deposition is localized, the sharpness of the ocean-continent transition, the sediment strength, and the time interval of deposition. When we change 0 (depositional scale) from 150 to 300 km, the average depositional rate ave from 0.2 to 2 km/Myr, the width of the transitional area between proto-continent and ocean from 0 to 100 km, the friction coefficient of BIF from 0.0 to 0.1, and the time interval to deposit BIFs from 0.2 to 1 Myr, computations show that the critical thickness of BIFs is not substantially influenced ( Figures S5a-S5e in Supporting Information S1).

The Biological-Induced Origin of Plate Tectonics
The computations show that subduction initiation induced by magnetite-rich BIFs is physically plausible, and with a distinctive biomineralization genesis for BIFs implies a biologically induced origin of plate tectonics. As one of the earliest energy metabolisms on Earth, the prosperity of photoferrotrophy presets the global tectonic environments. The photoferrotrophic bacteria used Fe(II) source in the anoxic seawater and efficiently metabolized it to acquire the necessary energy. Their thriving in the Neoarchean (Figure 4a) led to the massive deposition of ferrihydrite on the proto-continental margins. Meanwhile, vigorous submarine volcanism in the Neoarchean (Isley & Abbott, 1999) should have significantly increased the supply of hydrothermal Fe(II) to the photic zone, which has reacted with the ferrihydrite directly to make dense magnetite deposition on the seafloor at the proto-continental margins (Halevy et al., 2017;Li et al., 2017). The synchronous appearance of the anoxic oceanic environment (Kump & Barley, 2007), the high Fe(II)-oxidizing rate from thriving photoferrotrophs (Konhauser et al., 2002), and the abundant hydrothermal Fe(II) alternations in the Neoarchean (Isley & Abbott, 1999) satisfied the conditions to deposit massive magnetite-rich BIFs rapidly (Halevy et al., 2017;Li et al., 2017) (Figure 4b). We propose that these high-density sediments on the proto-continental margins broke the lithosphere, induced the earliest subduction, and initiated plate tectonics.
The BIFs that subducted presumably would have sunk into the deeper mantle; the transport of BIFs into the lower mantle has previously been suggested as a mechanism to form the ultralow-velocity zones (ULVZs) at the core-mantle boundary (Dobson & Brodholt, 2005). Interestingly, ULVZs are generally hundreds of kilometers wide, tens of kilometers high, and thought to be composed of high density, Fe-rich material (Bower et al., 2011), not unlike the BIF sedimentary packages required to trigger subduction initiation.
The distinctive biogeochemical formation of magnetite gradually shut down after the Neoarchean. The oxidation of the ocean since the early Paleoproterozoic (Figure 4c) has significantly depressed the dissolved Fe(II) in the seawater, reducing the contribution of photoferrotrophy to primary production and ferric oxide deposition (Li et al., 2017). After the Great Oxidation Event (GOE) (Figure 4c), the emerging oxygenic photosynthesis gradually surpassed the anaerobic photoferrotrophy (Figure 4a) so that the BIF deposition was significantly decreased (Bekker et al., 2010) (Figure 4b) and contained less magnetite (James, 1992;Li et al., 2017). On modern continental margins, the marine photic zone overlaps subduction zones, a similar scene we propose at the onset time of the global plate tectonics in the Neoarchean ( Figure S4b in Supporting Information S1). Following the Neoarchean, the low-density silicate deposits replaced the heavy BIFs, and sediment-induced subduction initiation shut down by which time new subduction zones could continuously formed by mechanisms now seen in the plate tectonic system (Gurnis, 2023;Stern & Gerya, 2018).

The Onset Time of Plate Tectonics
Our model implies a timing for the start of global plate tectonics. Despite long-standing discussion on this timing (Cawood et al., 2018;Condie & Kröner, 2008;Korenaga, 2013) and possible tectonic styles (T. Gerya, 2019;Sobolev & Brown, 2019;Stern, 2020;Yao et al., 2021), most studies accept that the onset of modern-type plate tectonics is coeval with the onset of the earliest subduction (T. V. Gerya et al., 2015;Rey et al., 2014). Based on our model, the earliest subduction started on global continental margins and was triggered by the rapid BIF deposition during 2.75-2.40 Ga, which is well supported by many contemporary plate-tectonic indicators and geological events. During the transition from the Neoarchean to Paleoproterozoic, volcanism recorded by large igneous provinces (LIPs) gradually shifted from submarine to subaerial (Kump & Barley, 2007) (Figure 4d). Since submarine volcanism is more reductive than subaerial volcanism (Kump & Barley, 2007), the preponderance of submarine LIPs in the Archean is consistent with the anoxic environment at that time and the related submarine hydrothermal alteration could provide Fe(II) for the rapid mineralization of magnetite during peak BIF deposition (Li et al., 2017). Once plate tectonics began, the frequent subduction of the oceanic lithosphere would make submarine LIPs preferentially destroyed and short-lived (Kump & Barley, 2007). Therefore, the decrease of submarine-volcano proportion since the Neoarchean-Paleoproterozoic transition (Figure 4d) probably corresponds to the time when BIF first induced the subduction of the oceanic plates. The existence of subduction is supported by the first identification of the Archean paired metamorphic units which are spatially and temporally linked (2.54-2.50 Ga) in the granite-greenstone belt   (Figure 4e). With the establishment of subduction zones, large-scale horizontal plate movement would start and is consistent with recent paleomagnetic reconstructions suggesting such movements (>5,000 km) between cratonic blocks already existed by ca. 2.7-2.4 Ga (Cawood et al., 2018) (Figure 4e). The crustal-growth peaked in the Neoarchean (2.7-2.5 Ga) (Condie et al., 2009) (Figure 4e) also supports our scenario. Arc magmatism resulting from the earliest subduction would strongly promote the formation of continental crust and provides a reasonable interpretation for the 7 of 10 rapid crustal growth during this period (Condie et al., 2009), although some numerical studies suggest that subduction with hotter Archean mantle may be inefficient in producing continental crust (Perchuk et al., 2019).
Convergent boundaries such as subduction zones are not the only proxy of plate tectonics, as divergent boundary records such as mid-ocean ridges (MORs) are proxies as well. The appearance of the oldest divergent-boundary  (Bekker et al., 2010). Most IFs before 2.4 Ga belong to BIF (Bekker et al., 2010). The sediments in 2.3-0.8 Ga and 0.8-0.56 Ga are dominated by the granular iron formation (GIF) and the Rapitan-type iron formation (RIF), respectively (Bekker et al., 2010). (c) Evolution models of atmospheric oxygen content suggested by Kump (2008) (purple) and Lyons et al. (2014) (orange). records in the Neoarchean (Cawood et al., 2018;Condie & Kröner, 2008) (Figure 4e) supports the inferred start of plate tectonics. Such records can be traced back to 2.75-2.5 Ga in several Archean cratons (Cawood et al., 2018). The appearance of nascent MORs means that more Fe(II)-rich fluids would be available for magnetite mineralization. Because of the higher elevation of these oceanic ridges compared with oceanic basins, the released hydrothermal Fe(II) fluids would reach the upper ocean layer close to the BIF-deposition depth more easily (Li et al., 2017), facilitating the rapid and massive deposition of magnetite-rich BIFs.

Conclusions
In this model of plate tectonics initiation, we find the extraordinary role of photoferrotrophic bacteria on early Earth. They effectively oxidize the dissolved ferrous iron and aggregate the resulting ferric oxides onto proto-continental margins, which were turned to dense magnetite by further reaction of Fe(II) from the submarine hydrothermal fluids. These high-density banded iron sediments could break the proto-continental margins to start subduction and kick-started plate tectonics. Their peak deposition in 2.75-2.40 Ga constrains the time plate tectonics started. This unique biogeochemical cycle with the biomineralization of amorphous ferrihydrite and the magnetization by the addition of hydrothermal Fe(II) fluids could only happen on Earth and only once during this period. The model implies that prosperous photoferrotrophy in the Neoarchean is the reason why Earth is the only planet with plate tectonics. In this sense, life has played an even more important role in uniquely shifting the tectonic style of Earth to a plate tectonic system.