Decoding how Earth's interior unleashes diamonds

During Earth history, diamonds have erupted in rhythm with the shuffling landmasses — but nobody really knows why. We discovered that the breakup of landmasses sets in motion a domino effect in Earth's interior that disrupts the roots of the continents, violently unleashing these precious stones.
Published in Earth & Environment
Decoding how Earth's interior unleashes diamonds
Like

How it started

Two decades ago, I was tasked with examining diamond mines in some of the planet's most hostile environments, from the sweltering heat of the Kalahari Desert to the bone-chilling cold of the Arctic tundra. These mines are scattered across the oldest and sturdiest parts of Earth's continents, known as the 'cratons'.  My PhD research focus was on kimberlites, a mysterious type of volcanic rock that carries diamonds from the Earth's interior to the surface of such cratons. My priority at the time was investigating the volcanic processes that occur near the surface; I was merely exploiting diamond mines to gain cross-sectional insight into the bowels of these strange volcanoes.

But there was always a nagging question in the back of my mind: how on Earth do those kimberlites get there in the first place? What makes them shoot up from deep within the planet after spending millions, or even billions, of years stashed away under the continents — sometimes with diamonds to boot?  It was the elephant in the room that no one seemed to have a good explanation for.

Diamond in kimberlite from Jwaneng Mine, Botswana (photo: Dr R.J. Brown)
Figure 1 | Diamond in kimberlite. Diamond in kimberlite from Jwaneng Mine, Botswana; the green minerals are olivine and the deep red mineral is garnet (photo: Dr R.J. Brown).

The problem

Most scientists agreed, however, that kimberlite eruptions occurred in sync with the supercontinent cycle, a recurring pattern of landmass formation and fragmentation that has occurred over billions of years of Earth's history. However, the exact mechanisms behind this relationship are debated: some researchers have argued that kimberlite magmas exploit the wounds in Earth's tectonic plates that open up during rifting, whilst others blamed mantle plumes — huge upwellings of hot magma from the core-mantle boundary nearly 3,000 kilometres beneath Earth's surface. 

Both ideas, however, are not without their problems. For one, the lower, thicker part of the tectonic plate — known as the lithosphere — is incredibly strong and stable, so fractures can't easily penetrate to allow magmas to flush through. Likewise, most kimberlites don't appear to exhibit the characteristic "flavour" of mantle plumes. Besides, plumes are so hot that they should stimulate intense melting on the underside of the continents, yet kimberlite formation typically involves very low degrees (less than one percent) of mantle melting.  If plumes play an important role, it's likely to be indirect: maybe they seed the upper mantle with super-deep diamonds, and/or warm the lithosphere enough that it can eventually rupture.

The solution

Fast forward nearly two decades. Despite a shift in research direction imposed by changes in the funding landscape, I remained preoccupied by the unanswered question that had vexed me during my earlier work. Fortunately, in the meantime I'd acquired a new set of tools — and expert collaborators — that would allow me and my team to tackle this thorny problem head-on.

With the input of T.H., we began applying our expertise in decoding the complex time lags in geological processes, knowing that they could yield substantial insight. It was perfect timing, because we'd just done some work on melding new machine learning tools we'd developed (initially to track the causes of manmade earthquakes) with sophisticated plate tectonic models. When we integrated these techniques, utilising an existing kimberlite database, we found that kimberlite volcanism peaks about thirty million years after continents break up — an apparently global phenomenon. We then made an unexpected discovery: the volcanism appears to migrate from the continental edges to the interiors over time and does so at a remarkably consistent rate across the continents.

Video 1 below is a data sonification of kimberlite eruptions spanning the past 240 hundred million years, since the supercontinent Pangaea began to split apart. 

Video 1: In this sonification, created by SYSTEM Sounds with our team's input, each kimberlite eruption is represented by a note, with the pitch of the note corresponding to the reconstructed latitude (that is, paleolatitude) of the eruption. Higher latitudes are associated with higher pitches. The longitude is reflected in the stereo position of the sound. The fragmentation rate of the tectonic plates is represented by sustained minor and major sounds, with darker minor sounds indicating plate merging and brighter major sounds indicating plate breakup. Additionally, the volume of crumbling rock sounds varies with the fragmentation rate, intensifying when the rate is high.

Excited by our findings, I emailed S.J. and S.B. separately to see if we could unpick what’s causing these signals, knowing that some dynamic process affecting the roots of cratons must play a significant role (S.J. and I used to work together at Trinity College Dublin, but I'd never met S.B.).  Both were intrigued by our results and got to work.  Because we were in the thick of the Covid pandemic, this process of discovery largely played out over Zoom calls and emails.

Realising the global nature of the kimberlite signal, we figured that the physical process triggering kimberlite volcanism must be fundamental, rather than a local quirk. We also recognised that the roots of continents don't abruptly end and transition to the mantle asthenosphere, but instead have a gradual contact zone known as a thermal boundary layer, which is about 35 kilometres thick. Could this layer be decoupling, driven by some thermal and/or compositional changes between the boundary layer and the asthenosphere? 

The next Zoom call was lively with excitement, as we all quickly realised that our different approaches were zeroing in on the same result.  S.J. had analytically shown that rifting creates a spatial and temporal order for convective removal of the roots of cratons, while S.B. had observed sequential instabilities migrating along the underside of the cratons using his advanced geodynamic models.

To our amazement, we found that the migration rates of these instabilities, derived using both approaches, closely matched the rates that T.H. and I had determined for the migration of kimberlite volcanism over time. This was the real eureka moment in the project. Like interdisciplinary science should be, our collective findings were stronger than the sum of their parts. They provided compelling evidence that a highly-organised pattern of mantle convection triggered by continental rifting was a key factor behind the signal we’d decoded in the kimberlites.

The thermal boundary layer, seen as the beige layer in Figure 2, is convectively removed (or simply put, stripped away), by Rayleigh-Taylor instabilities — a convective instability arising from the density contrast between colder lithosphere and hotter asthenosphere. Our models show that these instabilities, triggered by rifting, systematically sweep along the undersides of the cratons. In doing so, they remove the components stored in the thermal boundary layer and blend them with the hotter asthenosphere that wells up to replace the material that's stripped away (Fig. 2). Additional analytical models (shown in Fig. 3c of our paper) demonstrate that this process can generate low-degree, volatile-rich kimberlite melts and cause them to surge to Earth's surface. 

Schematic model of kimberlite formation
Figure 2 | Schematic model of kimberlite formation. (a) Simplified section through a craton showing the mechanical and thermal boundary layers. (b) Rifting generates a steep lithospheric gradient (labelled LAB) that gives rise to Rayleigh-Taylor instability. (c) Growth, migration, and detachment of the instability. (d) The process repeats: destabilization and convective removal of cratonic keel propagates inboard of the rift leading to migration of kimberlite volcanism toward cratonic interiors.

In full, our team tested this model using about ten different approaches. As we delved deeper, we uncovered further geochemical and geophysical evidence across multiple continents that bolstered the conceptual model (Fig. 2). The rhythmic spikes in kimberlite volcanism, which occur during and after the breakup of Earth's great supercontinents, can be attributed to a large-scale reorganisation of convective mantle tied to breakup.

Future directions and implications

In our study, we recognised a close similarity between the scale of kimberlite clusters and that of convective instabilities in the mantle. Future work should delve into the local and regional scale geochemical trends within kimberlite clusters at the surface. Analysing spatial geochemical patterns within such clusters may shed light on the variations in the characteristics of magmatism across the footprint of an instability, which may feature different degrees of melting that reflects the variable intensity of upwelling.

The implications of our research extend to diamond exploration as well. We've shown how a combination of paleogeographic, tectonic, and geochemical factors contribute to favourable conditions for kimberlite eruption. An intriguing question arises: do the variations in diamond-rich and diamond-poor varieties reflect the depletion of material during this magmatic/eruptive process?  Exploring this aspect will provide further insights into the dynamics of kimberlite magma generation and its relationship with diamond distribution — a major focus of my solo PhD work in the remote wilderness all those years ago. Back then, though, I couldn't have fathomed making sense out of these notoriously cryptic rocks, or the crucial role teamwork would play in the discovery.


Thomas Gernon is at the University of Southampton, based at the National Oceanography Centre, Southampton, United Kingdom.

Please sign in or register for FREE

If you are a registered user on Research Communities by Springer Nature, please sign in