Can opposites coexist?
Diamonds are more than precious gemstones – they are time capsules from the deep Earth. Trapped inside them are microscopic inclusions, tiny grains of minerals, fluids, and melts that reveal the conditions of Earth’s mantle at the time of diamond formation. These inclusions provide a unique glimpse into regions otherwise inaccessible, hundreds of kilometers beneath the surface.
One of the long-standing questions in Earth sciences is the redox state (oxygen fugacity) of the mantle. Redox conditions determine which minerals are stable, how carbon is stored (as carbonates, graphite, diamond, or carbides), and how magmas form and evolve. Typically, inclusions in diamonds reflect either oxidized conditions, with carbonates and carbonate-bearing melts, or reduced conditions, where sulfides, carbides, or metallic alloys are present. But what happens when both carbonate and metal appear together? Until recently, such coexistence seemed almost impossible, as metals and carbonates occupy opposite ends of the redox spectrum. And yet, this is exactly what we found.
Research approach and discovery
We studied inclusions ranging in size from micrometers to nanometers within two natural diamonds from Voorspoed, South Africa (Figure 1). Using infrared and Raman spectroscopy, electron microprobe analyses (EPMA), transmission electron microscopy (TEM), and synchrotron micro-diffraction, we identified both the mineral assemblage and chemical composition of these inclusions.
The results were unexpected. Alongside familiar silicate and oxide phases, we discovered a Ni-rich metallic phase and, remarkably, Ni-rich carbonate minerals within the same diamond. This was not just a curiosity of chemistry; the coexistence of metallic and carbonate inclusions challenges the conventional understanding of mantle redox conditions and raises fundamental questions about deep mantle processes.
In most natural settings, Ni-rich metals and carbonates cannot form together. Metals require a strongly reducing environment, whereas carbonates demand an oxidizing one. Finding both locked within the same diamond implies that the diamond must have grown in a dynamic environment where reduced and oxidized domains interacted. This suggests that redox boundaries in the mantle are not static but rather zones of active chemical exchange and reaction. Our diamonds appear to have captured a snapshot of such a redox frontier in action.
Hurdles, frustrations, and serendipity
The two diamonds we studied came from two separate Voorspoed diamond parcels, donated to us at different times by the De Beers Group. As someone who has worked on diamonds for some years now, I can only say: what are the odds? – We were lucky.
Yeal Kempe first began working on Voorspoed diamonds during the transition from her M.Sc. to PhD research. Early infrared analyses of the two diamonds revealed a peculiar signal of nitrogen and evidence of carbonate and CO2, whereas EPMA analyses detected nickel but little else in microinclusions. The results were confusing, not matching the other Voorspoed diamonds in the parcels or anything else familiar, and so the two odd diamonds were left aside. They sat in a drawer for almost a year, waiting quietly.
It was not until Oded Navon went on sabbatical and I arrived at the Hebrew University to co-supervise Yael that I first heard about these two strange diamonds. Honestly, I did not believe what I was hearing and asked Yael to repeat the analyses, this time while I watched. To my surprise, there they were: Ni-rich microinclusions.
Since it was nickel, we tried to detect a magnetic signal to learn something about its phase, but no signal was detected. Soon after, we turned to the TEM – a turning point in the study. We targeted a few microinclusions, which indeed revealed the presence of nickel with minor iron. The diffraction pattern corresponded to carbonate. But there was something else; each microinclusion was surrounded by numerous nanoinclusions of solid nitrogen. This was a surprise, but there was one more. At one apex of each solid nitrogen octahedron, we noticed a tiny, bright spot, only 4-8 nm in size (Figure 2). I remember us sitting there, Yael, Oded, Sergei Remennik, and I, throwing out ideas about what elements this phase could contain. Then I said “nickel”, which seems obvious in retrospect, but not at the time. The energy range was set, the detector tuned, and suddenly… the signal appeared: a Ni-rich phase with minor iron and no oxygen or sulfur. It was a moment of enlightenment.
Later, synchrotron X-ray analyses with Oliver Tschauner strengthened the identification of the different phases and confirmed that the Ni-rich nanophase is a metal alloy. We calculated pressure for the various phases, substantiated their depth of origin, and developed a conceptual metasomatic model that involves five of the six phases trapped in the diamond. At this stage, although the analytical evidence was compelling, no theoretical predictions or experimental observations could account for it, which was frustrating. That was also a main comment on the first version of the paper during review. This was the time when Tim Holland came aboard to help with thermodynamic calculations. It was real teamwork; now all pieces of the puzzle fit together, and we knew we had nailed it.
A window into deep mantle redox dynamics and open questions
Our results show that the mantle is more heterogeneous and likely more reactive than often assumed. The presence of both metallic and carbonate inclusions reflects a metasomatic reaction in which the host diamond itself is a product. The Ni-rich nature of both phases is striking. Metallic nickel is predicted only near the mantle–metal boundary, where the intersection of the mantle oxidation profile with the so-called nickel precipitation curve occurs at ~250 km. Importantly, our findings provide the first natural evidence for the existence of such a metallic phase, and for its involvement in mantle redox-freezing processes via interaction with carbonatitic-silicic melt from a subduction slab. In addition, our pressure constraints on the reaction and thermodynamic calculations suggest that the mantle–metal boundary depth may be deeper and variable than previously predicted (possibly between 280-470 km), depending on the composition of mantle peridotite and its Fe3+ content.
A subduction-related carbonatitic-silicic melt infiltrated reduced peridotitic mantle and transported oxidized carbon, alkalis, and other incompatible elements to depths well below the stability range of carbonates. Such enrichment and net oxidation may act as a precursor to the genesis of enriched alkalic melts such as kimberlites, lamprophyres, and some ocean island basalts, suggesting a broader role for carbonatitic-silicic metasomatism in shaping deep mantle chemistry.
Several questions remain, pointing to exciting paths for future research: Is the observed coexistence of metallic and carbonate inclusions a rare phenomenon, or an underappreciated feature of mantle chemistry? Do single inclusions in diamonds tell us about stable mantle conditions, or are they part of a reactive assemblage? Where is the mantle–metal boundary located? Is it a narrow, well-defined interface, or a more diffused, transition redox region where reduced and oxidized domains intermingle?
Diamonds and their microscopic inclusions continue to provide some of the clearest windows into Earth’s hidden mantle regions. By uncovering Ni-rich metallic and carbonate inclusions locked within diamonds, we have captured a rare glimpse of a metasomatic redox reaction deep in the mantle. These findings challenge idealized models of a homogeneous mantle and highlight the importance of dynamic reactive chemical environments. Future work, combining further inclusion analysis, high-pressure experiments, thermodynamic calculations, and geodynamic modeling, will deepen our understanding of how redox processes and carbonatitic metasomatism govern the evolution and chemical heterogeneity of our planet’s interior.