Although there is evidence for periodic geological perturbations driven by regular or semi-regular extra-terrestrial bombardment, the production of Earth's continental crust is generally regarded as a function of planetary differentiation driven by internal processes. We report time series analysis of the Hf isotopic composition of zircon grains from the North Atlantic and Pilbara cratons, the archetypes of Archean plate tectonic and non-plate tectonic settings, respectively. An ~170–200 m.y. frequency is recognized in both cratons that matches the transit of the solar system through the galactic spiral arms, where the density of stars is high. An increase in stellar density is consistent with an enhanced rate of Earth bombardment by comets, the larger of which would have initiated crustal nuclei production via impact-driven decompression melting of the mantle. Hence, the production and preservation of continental crust on the early Earth may have been fundamentally influenced by exogenous processes. A test of this model using oxygen isotopes in zircon from the Pilbara craton reveals correlations between crust with anomalously light isotopic signatures and exit from the Perseus spiral arm and entry into the Norma spiral arm, the latter of which matches the known age of terrestrial spherule beds. Our data support bolide impact, which promoted the growth of crustal nuclei, on solar system transit into and out of the galactic spiral arms.

Earth is unique among the known planets in having continents, whose formation has fundamentally influenced the composition of the mantle, hydrosphere, atmosphere, and biosphere. Cycles in the production of continental crust have long been recognized (Condie, 1998) and generally ascribed to the quasi-periodic aggregation and dispersal of Earth's continental crust as part of the supercontinent cycle (Murphy and Nance, 2003). However, such cyclicity is also evident in some of Earth's most ancient rocks that formed during the Archean (4000–2500 Ma) and Hadean (>4000 Ma) eons, when plate tectonics may not have operated. Frequencies of many geological processes become more challenging to decipher in the early Earth given that the rock record becomes increasingly fragmentary with age.

One mechanism to assess the cyclicity of crust production and understand its drivers is through the isotopic records of rocks and constituent minerals (Puetz and Condie, 2019). Zircon U–Pb–Lu–Hf isotopic data sets are particularly well suited to this endeavor as they retain a time-encoded proxy for the degree of source fractionation. We investigated the cyclicity in the addition of new mantle-derived (juvenile) crust and its subsequent reworking through the Hf isotopic record of dated zircon grains from the Archean North Atlantic craton (NAC) in West Greenland and the Pilbara craton in Western Australia. We complement this Hf data set with consideration of oxygen isotope compositions of dated Pilbara zircon (Johnson et al., 2022).

The NAC is dominated by Mesoarchean (3200–2800 Ma) felsic (silica-rich) gneisses and ultramafic complexes, along with younger supracrustal sequences and late-tectonic felsic to mafic intrusions (Friend and Nutman, 2019). It is commonly considered to have formed through horizontal (subduction–accretion) processes and to provide evidence for the operation of plate tectonics (a mobile-lid regime) by that stage in Earth history (Garde et al., 2020). In contrast, the older part of the Pilbara craton in Western Australia consists of variably metamorphosed Paleoarchean to Mesoarchean ultramafic to felsic supracrustal successions surrounding 3600–3300 Ma (Smithies et al., 2009) granite domes and is generally regarded to have formed through vertical (non-plate tectonic) processes in a single-plate (deformable stagnant lid) regime (Smithies et al., 2019).

Not all cratons were initiated at the same time, nor did they evolve at the same rate. Consequently, rather than using a global data set, we examined the isotopic time series on a per-craton basis. We fitted quantiles (25th, 50th, and 75th) to the NAC and Pilbara data sets using a moving 25 Ma bin. We interpreted the 75th quantile to chart a more significant contribution from juvenile magmatism and the 25th quantile to mark a greater degree of crustal recycling, whereas the 50th quantile was interpreted to reflect the average evolution. To further evaluate the zircon Hf time series, we used various spectral analysis approaches including periodograms and continuous wavelet transforms (CWT) (see the Supplemental Material1).

The 50th quantile of the NAC Hf data tracks along a 176Lu/177Hf ratio of 0.01 until 3200 Ma, when average values deviate toward more radiogenic signatures (Fig. 1A). A periodogram highlights frequency bands with >95% significance at ~198 m.y., ~113 m.y., several at ~74 m.y., at 54 m.y., and a band at ~680 m.y. that is above the noise model (Fig. 2A). Wavelet analysis of the various quantile time series revealed broadly similar frequencies, including a quasi-continuous band at ~200 m.y. (Figs. 1B1D), and a ~170 m.y. frequency considering only the >3200 Ma segment (see the Supplemental Material). All quantiles show more frequency structure in the pre-3200 Ma component of the record, with longer wavelengths dominating the post-3200 Ma segment.

In contrast, the Pilbara Hf time series (3800–2825 Ma) initially tracks values close to the chondritic uniform reservoir (CHUR), then becomes more variable with some super-chondritic values after 3500 Ma (Fig. 1E). Spectral analysis of the 50th quartile fit highlights frequency bands at >95% significance at ~114 m.y., ~80 m.y., and ~55 m.y., and bands at ~415 m.y. and ~168 m.y. that are above the noise model (Fig. 2B). Wavelet analysis of the various quantile time series have comparable bands with a continuous frequency at ~170 m.y. (Figs. 1F1H). The frequency response of the Pilbara zircon Hf timeseries is similar to the CWT pattern in the pre-3200 Ma NAC.

Various periodicities are thought to be superimposed onto the Earth system through processes acting across varying temporal and spatial scales, including planetary, solar system, and even galactic. These include a ~800 m.y. resonance between tidal and free oscillations of the core, a 500–300 m.y. supercontinent cycle, an ~200 m.y. galactic year, and an ~30 m.y. impact cycle (Rampino et al., 2019). Other long-period geological cycles have been proposed from studies of large igneous provinces (LIPs) and mantle plumes (Prokoph et al., 2013).

The zircon Hf time series analyses for both the NAC and Pilbara craton show frequency components in the ~170–200 m.y. range, which are similar to those reported from timeseries analysis of LIPs. These components are interpreted as a proxy for mantle plumes (Ernst and Buchan, 2002; Fig. 2C). However, the Milky Way galaxy appears to have four major spiral arms, which implies spiral arm passages also occur approximately every 170–200 m.y. (Rampino, 1997) as both arm and solar system orbit the galactic center at different rates. During this passage the solar system will transit through dense interstellar clouds and be subject to variable galactic tides (Fig. 3A). Perturbations of the Oort cloud (Fig. 3B) due to oscillations of the solar system around its galactic plane, and interactions between spiral arms with other areas of enhanced star formation during galactic transit, have been directly linked by some to the flux of meteorite impacts on Earth (Goncharov and Orlov, 2003).

Earth is subject to impacts from near-Earth objects (NEOs), which are primarily derived from the main asteroid belt, and comets (Ivanov, 2008). While the former are inferred to result in much more frequent impacts (Granvik et al., 2018), the latter would release two orders of magnitude more energy into the crust on collision of a comparably sized impactor (Nuth et al., 2018). More energetic impacts excavate a greater volume of material, while impact melt production is a function of energy and momentum (Schmidt and Housen, 1987). Comets would be the most likely impactor to carry a resolvable, low-frequency periodic signal as they are most susceptible to perturbations from outside of the solar system (Rampino, 1997). Independent of impact energies, NEOs responsible for >20-km-diameter craters occur every 0.75 m.y., which is too high a frequency for them to be the dominant signal retained in the Hf isotope record. The longer forcing frequency, coupled with higher impact velocities of comets, favor detection in the isotopic record of crust production, despite NEOs as the expected dominant impactors.

Impact-generated melt production is also affected by the target temperature, which would be greater for an impact of any particular size on a warmer young Earth with a steep thermal gradient and high internal temperatures (Potter et al., 2012). We posit that melts from energetic comet impacts would form buoyant crustal nuclei that, in turn, would support crustal growth through hosting the products of later differentiation. This process must have been independent of the prevailing geodynamic settings if the NAC and Pilbara craton represent subduction versus stagnant lid tectonics. Similar frequencies in their zircon Hf isotope time series are preserved, suggesting a common process. Additionally, the zircon Hf isotopic record tracks a frequency throughout the entire Archean that is similar to that advocated for Palaeozoic, impact-induced extinctions (Shoemaker, 1998). Arguably, the extension of this frequency into deep time points to a large-scale and relatively constant driver over most, if not all, of our planet's history.

Models of early Earth crust generation emphasize the melting of hydrated basaltic rocks to produce tonalite–trondhjemite–granodiorites (TTGs). Smithies et al. (2021) suggested that TTG magmas could be generated both via a bottom-up process facilitated by mantle-derived water and later, in the Paleoarchean, via a top-down process initiated by the sinking of greenstones. Models of impacts on the early Earth suggest that large, felsic, shallow impact melt pools were produced (Grieve et al., 2006). Hence, impacting critically fits into this formative history through providing buoyant seeds that acted as nucleation points for later TTGs generated via either mantle or greenstone extraction, simply as a function of having greater preservation potential.

To better understand any link between the periodicities identified in the zircon Hf isotope time series and external forcing, we developed a model for the motion of the solar system within the Milky Way galaxy that we used to estimate mass distribution relative to our solar system. We then compared this to the compiled zircon oxygen record of the Pilbara craton (Johnson et al., 2022; Figs. 4A and 4B).

Previous models have assumed a solar system orbit, relative to the spiral arms (i.e., galactic period), of 752 m.y., based on the average time between superchrons, the periods during which Earth's magnetic field remained stable (Gillman and Erenler, 2019). The best-fit galactic period to our data is 748 m.y., which results in a duration of 187 m.y. between passages through the spiral arms.

Oxygen isotopes in zircon crystals formed within and inherited by igneous rocks in the Pilbara craton define a secular trend that further constrains crust production (Smithies et al., 2021; Johnson et al., 2022). Zircon oxygen isotopes start with, on average, δ18O lighter than mantle values prior to 3400 Ma. Between 3400 Ma and 3100 Ma, zircon δ18O values are mainly within the mantle field, but thereafter they extend to heavy δ18O values consistent with widespread reworking of supracrustal material (Fig. 4A). A continuous ~200 m.y. frequency is evident in CWT analysis of a locally weighted scatterplot smoothing (LOWESS) fit to the zircon oxygen data (Fig. 4B), with departures to light oxygen isotope values in zircon at ca. 3560 Ma and ca. 3430 Ma (Fig. 4A).

Zircon with isotopically light oxygen is commonly associated with volcanic calderas and shallow extensional systems. Impacts can also establish light signatures through the generation of both large volumes of hydrothermally altered crust and widespread post-impact shallow melt pools (Grieve et al., 2006). Giant impacts may also drive decompression melting of the mantle (Shibaike et al., 2016). Notably, while the record of preserved crust provides very little material (other than zircon) to relate to the ca. 3560 Ma light oxygen isotope departure, the ca. 3430 Ma excursion corresponds to the age of spherule beds in Australia and South Africa that provide direct evidence of large impacts (Byerly et al., 2002). Similar frequencies may also be expected for plume-driven models of early continental growth (Reimink et al., 2014); however, the close temporal association with spherule beds favors an impact driver (Johnson et al., 2022). The ca. 3560 Ma light zircon oxygen signature occurs on the predicted exit from the Perseus arm, whereas the younger ca. 3430 Ma excursion corresponds to entry into the Norma arm (Fig. 3A).

Perturbations of Oort cloud comets have been linked to oscillations in the galactic tide or nearby passing stars (Levison et al., 2004). The motion of the Sun modulates the strength of these perturbations as they are influenced by the local stellar density. However, the observed anisotropy in the perihelia of long-period comets does not support galactic tide oscillations as the sole source of such perturbations (Delsemme, 1987), and single encounters with a nearby star will not cause significant perturbation of the Oort cloud unless they pass very close (<0.5 pc; Torres et al., 2019). The higher densities of stars in spiral arms increase the probability of multiple stellar encounters, which also increases across the time spent in the arm and may explain impacts around the exit of spiral arms.

A proposed mechanism for the clustering of impacts when entering a spiral arm may be perturbations of the Oort cloud induced by giant molecular clouds (GMC). Encounters with GMCs are known to cause disturbance and, in extreme cases, cause comets to be lost to interstellar space (Jakubík and Neslusan, 2008). Compression of gas passing across a galactic arm causes the formation of a GMC, with subsequent gravitational collapse initiating star formation (Dobbs et al., 2014). As GMCs have lifetimes (10–50 m.y.) less than the duration of passage through a spiral arm, they are unlikely to perturb the Oort cloud during its exit of a spiral arm.

Long-period variations in the flux of comet impacts to Earth likely influenced crust production through a variety of mechanisms. On the early Earth, giant impacts may have been the trigger for production of Earth's first crustal nuclei via impact-induced melting and magmatic differentiation (Grieve, 1980). Younger impacts may have enhanced or modified processes of crust formation and destruction operating in different modes and at different times and locations. Clearly the role of impacts in crust production will have changed over time owing to the exponential decline in average projectile size and particle density in the early solar system (Marchi et al., 2014), but also as the dominant tectonic mode on Earth transitioned into widespread subduction, as is evident in the increasing average crustal residence in Pilbara magmas after 3200 Ma (Fig. 4B).

Isostatic adjustments triggered by giant meteorite impacts (Fig. 3C) are predicted to result in extensive decompression melting of the asthenosphere to produce thick, plateau-like, mafic–ultramafic pre-cratonic nuclei (Jones et al., 2003). Meteorite impacts also cause intense fracturing and brecciation of bedrock (Riller et al., 2018) and facilitate the development of prolonged hydrothermal circulation (Kring et al., 2020; Fig. 3D). Such circulation promotes enhanced fluid–rock interaction, including significant shifts in δ18O values, and supports partial melting, both in the immediate aftermath of impacting and through the development of a fertile melt source long after the event (Johnson et al., 2022). Once initial mantle melt segregation occurred, it would be inevitable that such impact-induced buoyant felsic seeds would have had a lower propensity to be recycled than their dense mafic counterparts. While endogenous processes (e.g., plate tectonics) have been important in establishing and maintaining Earth's major geochemical reservoirs, the wider galactic environment may have seeded cratonic nuclei via impacting that ultimately grew to become the continental crust on which we live.

H. Smithies and M. Prause publish with the permission of the Geological Survey of Western Australia. We thank Martin Schmieder, Bill Bottke, and two anonymous reviewers for detailed comments that greatly improved this contribution, Christian Koeberl and Nick Gardiner for comments on earlier versions, and Urs Schaltegger for his willingness to support scientific discussion of potentially contentious ideas and for facilitating a very constructive review process.

1Supplemental Material. Details on the analytical methodology and a tabulated compilation of the data used in this work (Tables S1–S7). Please visit to access the supplemental material, and contact with any questions.
Gold Open Access: This paper is published under the terms of the CC-BY license.