The discovery of diamonds in metamorphic rocks of continental affinities occurred shortly after the discovery of coesite in similar rocks. These important discoveries led to a revolution in our understanding of the subduction and exhumation of continental materials and the establishment of a new discipline, ultra-high-pressure metamorphism (UHPM). After these discoveries more ultra-high pressure (UHP) minerals were found and dozens of UHPM terranes were established. Diamond is a remarkable material which is chemically inert and stable over geological timescales making it the perfect “geological container” where gas, fluid, and solid inclusions can be preserved. Moreover, its presence is indicative of specific pressure and temperature conditions which implie subduction to a minimum depth of ~120 km. The inclusions trapped inside these diamonds can shed light on the composition, redox state, and evolution of the fluids related to UHPM diamond formation while the carbon isotope ratios of the diamond itself can inform us about the source of carbon. The purpose of this review is to provide a brief history on the discovery of microdiamonds starting with work that began in the mid-1960s in Kokchetav massif, Kazakhstan, through to the present day. Particular attention will be on more recent micro-diamond discoveries in the last decade. The current state of understating of the mechanisms of UPHM diamond formation, the misidentification of microdiamonds due to contamination from sample extraction/preparation, and the future of UHPM diamond research are discussed. The paper also considers the controversial topic of the occurrence of lonsdaleite within the population of microdiamonds from UHPM terranes as well as recently reported ophiolite- and volcanic-hosted diamonds.

A multitude of geological and geodynamic events are recorded in collisional orogens where large-scale processes such as deep subduction of lithospheric plates, mantle convection, and mountain building occur. Continental collision is accompanied by the subduction of continental materials into the Earth’s deep interior. The exhumation of these materials back to the Earth’s surface is one of the largest natural processes that occur on the Earth. These rocks contain minerals formed under pressures that are higher than the pressures that occur within the crust and they provide a considerable amount of information about processes that occur in the Earth’s mantle. Evidence of mineral reactions and deformation, fluid and melt migrations, phase transformations, and geochemical recycling are recorded in these rocks. On a regional scale continental collision zones release “energy” through earthquakes and volcanic eruptions.

In the early stages of the theory of plate tectonics several concepts were established; one such concept was that the continental crust is buoyant and will always remain floating above the oceanic crust. Therefore, this early version of plate tectonic theory did not provide a mechanism for the subduction of continental lithosphere into the Earth’s deep interior. This paradigm was significantly challenged in 1984 with two discoveries of coesite, first in the Dora Maira Massif, Alps (Chopin 1984) and the second, published only three months later, in the Western Gneiss Region, Norway (Smith 1984). Soon after these exciting discoveries, graphite pseudomorphs after diamond in the Beni Bousera alpine peridotite massif (Pearson et al. 1989) and then true diamonds, “microdiamonds” were discovered in the Kokchetav massif, Kazakhstan (Sobolev and Shatsky 1990). These startling revelations ushered in an entirely new discipline, ultra-high-pressure metamorphism (UHPM).

Crustal rocks of that have experienced recrystallization within or above the coesite and diamond stability fields (≥ 2.7–4.0 GPa, ~700–1000 °C) implying subduction to a depth of ≥ 90–120 km are considered as UHPM rocks. The presence of coesite and/or diamond in metamorphic rocks of continental affinities has become the standard for defining UHPM terranes. UHPM research flourished due to the detailed studies of outcrops and mineral/rock thin-sections in laboratories with the aid of advanced state-of-art analytical and synchrotron assisted instruments and technologies. This led to the establishment of more than 20 coesiteand nine diamond-bearing localities (Fig. 1) (e.g., Ogasawara 2005; Dobrzhinetskaya 2012; Schertl and Sobolev 2013; Liou et al. 2014 and references therein). By combining all the results of this research, an integrated picture of the processes operating in the deep Earth at converging plate boundaries is realized. Moreover, a greater understanding of how crustal materials are recycled into the deep mantle environment and incorporated into large-scale mantle dynamics is also gained from these studies.

Diamond is an important index-mineral of UHPM events. Since diamond is chemically inert and stable over geological timescales it is the perfect “geological container” where gas, fluid, and solid inclusions can be preserved. These inclusions can provide detailed information about the conditions and chemistry of diamond forming media as well as mantle mineralogy and geochemistry. Moreover, diamond, which is often included in zircon, can be a precise geochronological indicator of the peak of UHP metamorphism. Aside from the inclusions, δ13C values of the diamond itself can provide information about the carbon reservoir from which it crystallized. The type of nitrogen aggregation in diamond can shed light on the residence time of diamond in the environment where the diamond crystallized.

There are many types of reviews published in international forums that discuss impact of UHP mineralogy and UHP metamorphism studies on refining thermobarometry and plate tectonic modelling (e.g., Ernst and Liou 1999, 2008; Chopin 2003; Liou et al. 2004, 2009, 2014; Green 2005; Gerya 2010; Beltrando et al. 2010; Green et al. 2010; Guillot et al. 2008; Schertl and Sobolev 2013; Shirey et al. 2013; Faryad and Cuthbert 2020). However, comprehensive reviews of the studies of UHPM diamonds as well as their detailed studies with modern state-of-the-art instruments and technologies are still scarce (e.g., Ogasawara 2005; Dobrzhinetskaya 2012).

The goal of this review is to provide an update of new diamond discoveries that led to the establishment of new UHPM terranes, and to summarize and critically analyze existing concepts of diamond formation in orogenic belts, ophiolites and volcanic sources. An additional objective is to specifically emphasize certain aspects of knowledge from studies of diamonds which have been missed before and/or less highlighted in current research projects. The review outlines new constraints from many well-known collisional orogens which will lead to the improvement of geotectonic modelling that is currently being developed for very deep subduction of continental rocks, and their exhumation.

The history of microdiamond discoveries in the Kokchetav massif, Kazakhstan

In the mid-1960s several detrital grains of diamond ranging in size from 7 to 200 mm were found in the Tertiary detrital deposits of titaniferous sands in the vicinity of the Kokchetav massif, northern Kazakhstan (Kashkarov and Polkanov 1964). The alleged source of the detrital material was determined to be the Precambrian metamorphic rocks of the Kokchetav massif (Essenov et al. 1968). By the 1970s extensive geological mapping and exploration for mineral resources began in Kazakhstan, led by A. A. Zaychkovsky from the Kokchetav Geological Survey. They collected nearly 10 m3 of rocks from the eclogite lenses enclosed in garnet–biotite gneisses that were exposed at the shore of Kumdy Kol lake, Akmola Region, Kazakhstan. After crushing the rocks, the heavy minerals separation and flotation techniques were applied and, several microdiamonds were recovered from the residua (Rozen et al. 1972). It was concluded that these diamonds had formed under unusual crustal metamorphic conditions (Rozen et al. 1979).

The first detailed geological and petrographic descriptions of microdiamonds found in situ together with their host minerals was reported in Russian publications by Sobolev and Shatsky (1987, 1988). However, these unusual diamonds remained unknown to Western scientists until Sobolev and Shatsky (1990) published their results in Nature. They provided a detailed description of diamond morphologies and presented credible and convincing images of diamonds in situ. They were the first who postulated that these diamonds had formed in regional metamorphic rocks of continental affinities during deep subduction. They emphasized that new models are needed to explain how crustal rocks can be subducted to depths greater than ~100–120 km and subsequently returned to the Earth’s surface with preserved mineralogical relicts of ultra-high-pressure metamorphic (UHPM) events.

Following the efforts of the international community, the Kokchetav diamond-bearing metamorphic terrane became the “classic locality” of ultra-high-pressure metamorphism (e.g., Ernst and Liou 1999; Ogasawara 2005; Dobrzhinetskaya 2012; Schertl and Sobolev 2013; Liou et al. 2014). The “classic UHP rocks” of Kokchetav massif include metamorphic rocks with sedimentary and volcanic protoliths. They consist of varying lithologies of felsic gneisses, calc-silicate rocks, schists, quartzites, marbles, eclogites and garnet pyroxenites. Microdiamonds ranging in size from ~10 to 100 µm (rare up to 150–300 µm) with an average size ~40–50 µm occur in all the above-mentioned lithologies (Fig. 2) in high concentrations (e.g., Dobrzhinetskaya 2012; Schertl and O’Brien 2013). According to Pechnikov and Kaminsky (2008) the Kokchetav microdiamond reserves are calculated to be as large as ~2.5 billion carats, with the average content of ~20 carats per metric tonne. The Kokchetav diamonds have varying morphologies such as rose-like and dendritic-like crystals (Fig. 3A), platy skeletal-like and shapeless single crystals (Fig. 3B,C), or polycrystalline aggregates (Fig. 2D), hopper-like cuboidal, or cuboidal and octahedral-like with truncated corners, and sometimes slightly rounded single crystals (Fig. 3E,F). They are often observed as inclusions in refractory minerals such as zircon and garnet where they remain well-preserved and protected from complete transformation to graphite during exhumation of the UHPM rocks accompanied by retrograde metamorphism. While zircon and garnet protect diamond inclusions from graphitization, the radiation damage (metamictization) of microdiamonds included in zircon is a common phenomenon. Several studies (e.g., Orwa et al. 2000; Shimizu and Ogasawara 2014) have reported that the radiation damage of the Kokchetav diamonds occurred due to α-particle emissions from radiactive decay of U and Th in host zircon. This damage can be qualitatively probed with Raman spectroscopy, the Raman mode of diamond in the core of zircon is broadened and shifted to lower wavenumbers relative to pristine diamond (Shimizu and Ogasawara 2014), additionally two Raman modes at 1490 cm−1 and 1630 cm−1 have been observed and attributed to the vacancy and the split interstital, respectively (e.g., Prawer et al. 2004).

The peak metamorphic conditions of the Kokchetav UHPM rocks are estimated by thermobarometry of the rock forming minerals to be P = ~4.5 GPa and T = 950–1000 °C (e.g., Ogasawara 2005). Studies of nanoinclusions in these diamonds suggest that the Kokchetav diamonds formed at a pressure range of 6–9 GPa, and a temperature range of 980–1200 °C (e.g., Dobrzhinetskaya et al. 2006a). The age of the UHPM event is Cambrian ~531 Ma whereas the protoliths of diamond-bearing rocks are as old as Precambrian ~2100–2700 Ma (Table 1).

Soon after Kazakhstan, similar microdiamonds (Fig. 4) were found in the Western Gneiss Region (WGR), Norway (Dobrzhinetskaya et al. 1995; van Roermund et al. 2002), the Erzgebirge Massif of Germany (Massonne 1999), in China in the metamorphic rocks of the Dabie mountains (Xu et al. 1992, 2003), and Qinling region (Yang et al. 2003), and the Kimi complex of the Greek Rhodope (Mposkos and Kostopoulos 2001, Perraki et al. 2006). Microdiamonds have also been discovered in UHP oceanic floor metasediments at Lago di Cignana, the western Alps, Italy (Frezzotti et al. 2011, 2014). The above-mentioned localities are unconditionally classified as UHPM terranes because aside from diamond other ultra-high-pressure minerals have been discovered there such as coesite, TiO2-II (rutile with α-PbO2 structure), majoritic garnet, Ca-Eskola-rich clinopyroxene (Ca-Eskola = Ca□Al2Si4O12 where □ represents a cation vacancy in the pyroxene structure, see Vogel 1966 and others). Since the detailed reviews on UHPM diamonds were published by Ogasawara (2005), Dobrzhinetskaya (2012) and Liou et al. (2014) several new diamond-bearing metamorphic rock localities have been reported.

Table 1 summarizes the occurrences of microdiamonds in metamorphic rocks starting from their earlier descriptions (e.g., Dobrzhinetskaya 2012) to present. Within them are new diamond localities reported from granulite and garnet peridotites from the northern part of the Bohemian massive (Kotková et al. 2011; Naemura et al. 2011; Perraki and Faryad 2014), detrital diamonds from Erzgebirge (Schönig et al. 2019), from felsic granulites of the Betic Rif Cordillera, NW Africa and SE Spain (Ruiz-Cruz and Sanz de Galdeano 2012, 2013, 2014), from Straumen area of Western Gneiss Region, Norway (Smith and Godard 2013), from gneisses and eclogites of the Tromsø Nappe of Caledonides (Janák et al. 2013), from gneisses of Seve Nappe of Caledonides, Sweden (Majka et al. 2014; Klonowska et al. 2017), from oceanic metasediments of Lago di Cignana, Western Alps, Italy (Frezzotti et al. 2011; 2014; Frezzotti 2019), from garnet–kyanite–quartz–feldspathic gneisses of Pohorje, Eastern Alps (Janák et al. 2015), from amphibolites of Quinling region, China (Wang et al. 2014), from West Greenland (Glassley et al. 2014), and from garnet megacryst in mélange of the Edough Massif, NE Algeria (Bruguier et al. 2017).

New microdiamond localities discovered between 2011 and 2020

České Středohoří, Strážnad Ohří, Plešovice and Gföhl areas of Bohemian massif, Central Europe. The first in situ microdiamonds were reported as inclusions in garnet from felsic gneisses exposed in vicinities of Erzgebirge, northwestern part of the Bohemian Massif, Germany (Massonne 1999). The Erzgebirge locality became the second classical example of an UHPM terrane supported by extensive studies and publications (e.g., Nasdala and Massonne 2000; Stöckhert et al. 2001, 2009; Massonne 2003; Massonne and Tu 2007; Massonne et al. 2007b; Dobrzhinetskaya et al. 2003, 2006b, 2013). Since 2011 several new microdiamonds localities (Table 1) in the northern part of the Bohemian massif have also been reported (Kotková et al. 2011; Perraki and Faryad 2014).

Kotková et al. (2011) unambiguously documented microdiamonds ranging in size from 5–30 µm in situ in thin sections (Fig. 5) prepared from drill cores of felsic granulites extracted from two boreholes at the České Středohoří Mountains and from an outcrop in Stráž nad Ohří (Czech Republic).

Microdiamonds from the felsic granulites of these new localities occur as inclusions in grt, zr, and ky included in garnet. The correctly performed polishing technique provided indisputable evidence that the microdiamonds are in situ (Fig. 5). The microphotographs show that diamond inclusions “protrude” above the host minerals (zircon and/or garnet) well-polished surface. Coesite inclusions in kyanite included in garnet were also found in felsic granulite (Kotková et al. 2011). The authors hypothesized that during subduction (>120–150 km) the K-feldspar- and qtz-rich continental crust had “amalgamated” with small fragments of mantle materials and the latter are now presented as sparse lenses of the garnet peridotite containing relicts of Cr-spinel.

Kotková et al. (2011) emphasized that the protolith of these newly discovered diamond-bearing felsic granulites was “granitic crustal” rocks, which is different from the protolith of the Erzgebirge locality in Germany where the first diamond from this locality was reported by Massonne (1999). The Erzgebirge diamonds occur in garnet–biotite–phengite–quartz–feldspathic gneisses with a metasedimentary protolith. However, all these rocks belong to the same rock-unit of the Variscan orogeny, 340–380 Ma (e.g., Kröner and Willner 1998; Massonne and O’Brien 2003). Faryad and Cuthbert (2020) considered these diamond-bearing felsic granulites as former UHPM rocks that were overprinted by granulite facies metamorphism after their exhumation to the crust. Therefore, further studies of K-Fsp-rich felsic granulites with a primary granitic protolith may provide the opportunity to discover other UHPM minerals such as K-cymrite and/or K-wadeite which were first synthesized in laboratory (e.g., Fasshauer et al. 1997; Davies and Harlow 2002). Later K-cymrite was discovered in North Qaidam UHP eclogite, Western China (Zhang et al. 2009).

Perraki and Faryad (2014) discovered diamond, coesite, and moissanite inside of polished zircon grains. The zircons were extracted from the heavy minerals fraction that was separated from the crushed kyanite-bearing Gföhl granulite samples (the Moldanubian Zone of the Bohemian massif). The Gföhl kyanite-bearing granulites contain lenses and boudins of eclogites, garnet peridotites, and garnet pyroxenites. The thermobarometric calculations suggest that UHP metamorphism occurred at P > 4 GPa and T = 680 °C followed by granulite facies metamorphism and partial melting at P < 2 GPa and T = 850–950 °C. Perraki and Faryad (2014) agreed with Kotková et al. (2011) that the diamonds crystallized during UHP metamorphism of the Variscan Orogeny.

One novel aspect of the study of Perraki and Faryad (2014) is that they found SiO2 (coesite) coexisting with diamond and moissanite (SiC). Moissanite formation requires extremely reducing conditions, e.g., 4.5–6 log units below the iron–wustite (IW) buffer, whereas diamond and SiO2 (coesite) may coexist at an oxygen fugacity close to the diamond + CO fluid buffer (CCO) and the fayalite–magnetite–quartz (FMQ) oxygen buffers (e.g., Mathez et al. 1995; Frost and Frost 2014). Additional studies of the coexisting diamond, coesite, and moissanite inclusions in zircons from Gföhl granulites should be a high priority for future investigations, since such a study would shed light on the redox conditions of fluid–rock interactions and microdiamond formation in UHPM terranes.

Microdiamonds in detrital garnets from Erzgebirge, Germany. After several decades of the intensive studies motivated by the first discoveries of coesite and microdiamonds in rocks of continental affinities, it was shown that UHP metamorphism is not a rare phenomenon. However, despite the widespread exposure of presumably UHPM rocks in the continent–continent collisional orogens the mineralogical evidence of ultra-high-pressure metamorphism remains scarce, since during exhumation UHP minerals are typically overprinted with lower-grade metamorphism and partial melting. Schönig et al. (2019) proposed to study the distribution and characteristics of UHPM rocks by analyzing detrital garnets that have accumulated in surface sediments.

They collected detrital garnets from loose sediments deposited near the Erzgebirge massif where diamonds were described as inclusions in garnets and zircons (e.g., Nasdala and Massonne 2000; Dobrzhinetskaya et al. 2003). The sediments consist of sand from the creeks around the Saidenbach reservoir that have been draining the Erzgebirge UHPM rocks for a long time. After heavy mineral separation and examination of the residua with optical microscopy about 700 grains of garnets were chosen for further research. Out of the 700 inclusion bearing garnets they found 26 garnets containing 46 inclusions of monomineralic coesite, and 22 garnets containing 41 diamond inclusions, and they examined them with Raman spectroscopy (Schönig et al. 2019). These results provide undisputable evidence that the processes of erosion and weathering of UHPM rocks can relocate UHP minerals to sediments scattered around the hard rock exposures. Therefore, this method may be used as an additional technique for discovering new UHPM terranes. Similar methodology has been successfully applied to single zircon grains that were separated from UHPM rocks and has been successfully used for more than a decade to identify microdiamonds and other UHPM mineral inclusions.

The Betic and Rif Cordilleras, NW Africa, and SE Spain. A new microdiamond-bearing locality was discovered within the felsic gneisses of granulite facies in the Betic–Rif arc (NW Africa) and in the SE of Spain (Ruiz-Cruz and Sanz de Galdeano 2012). This tectonically complicated domain belongs to the pre-Mediterranean Alpine orogenic zone formed during the Africa–Eurasia collision (e.g., Platt et al. 2013). It should be noted that almost two decades ago cubic graphite pseudomorphos after diamonds were discovered in peridotites of Beni Bushera massif (Morocco), situated within the Rif orogenic belt (Pearson et al. 1989). This discovery was the first report that mantle rocks can be tectonically emplaced into the crustal environment from a depth corresponding to the diamond stability field. However, unambiguous proof of this concept was emphasized by finding similar cubic graphite in the Ronda peridotite massif, Spain (Davis et al. 1993). They showed that the basal planes {0001} of graphite are parallel to the {111} of the octahedra which demonstrated distinctly that graphite is pseudomorph after diamond. Studies of carbon isotopes in cubic graphite of both Beni Bushera and Ronda massifs show that the precursor diamonds crystallized from a 13C-depleted reservoir most likely originating from subducted crust (Pearson et al. 1991). A decade later, El Attrasi et al. (2011) reported finding of the 0.5–2 μm size diamond crystals included in large graphite flakes extracted manually from a garnet pyroxenite layer of the Beni Bousera massif, thereby providing unequivocal confirmation that these rocks were at some point in the diamond stability field.

Ruiz-Cruz and Sanz de Galdeano (2012) reported findings of octahedral, cuboid and cubo-octahedral microdiamonds of < 5 × 10–15 µm size (Table 1) as inclusions in garnet, kyanite, and quartz with relicts of coesite from felsic granulites exposed in the Ceuta area of the Northern Rif, NW Africa. The authors have estimated the metamorphic conditions of the Ceuta granulites as P = 4.3 GPa and T = 1100 °C (corresponding to a depth of >150 km) and suggested that the diamond-bearing crustal rocks and peridotites experienced a similar UHP metamorphic condition. They also suggested that the UHP metamorphism occurred ~330 Ma ago followed by partial melting and migmatization at ~265 Ma (Ruiz-Cruz and Sanz de Galdeano 2013). Given that the diamond-bearing garnet from felsic gneiss contains apatite’ exsolution lamella and clusters, the authors calculated the phosphorus solubility in garnet and compared it with those obtained in high-pressure and high-temperature experiments (Konzett and Frost 2009; Konzett et al. 2012). Based on such a comparative analysis they concluded that the highest pressure reached during UHPM was 7 GPa, and temperature ~1150 °C (Ruiz-Cruz and Sanz de Galdeano 2013).

Straumen localiy, Western Gneiss Region, Norway. After the discovery of coesite inclusions in clinopyroxene occurring in the dolomite–eclogites from Grytting, Norway (Smith 1984) the Geological Survey of Norway (NGU) initiated a search in 1992 for possible microdiamonds in the WGR. Garnet–kyanite–phlogopite gneisses were collected at Fjørtoft island of WGR and crushed at the NGU laboratory. Several grains of microdiamonds (30–45 µm size) were separated from the heavy mineral concentrates extracted at the NGU chemical laboratory using a method of thermochemical decomposition (e.g., Dobrzhinetskaya et al. 1995). Later, microdiamonds were found in situ in thin sections prepared from hand-samples of garnet peridotite from Fjørtoft (van Roermund et al. 2002), and in Fe-Ti garnet peridotite from Svartberget (Vrijmoed et al. 2006, 2008).

In situ diamond has also been reported from two more localities in the central and northern parts of WGR, Norway (Smith and Godard 2013; Janák et al. 2013). Smith and Godard (2013) conducted Raman spectral point analysis and Raman mapping of the carbon-bearing inclusions in zircon from kyanite–phengite–coesite-bearing eclogites from the Straumen eclogite pod in the WGR, and they identified two graphitized diamonds. The authors noticed that the position of the 1st order Raman mode of the diamond inclusions varies from 1332 cm−1 (ambient value for single crystal diamond) down to as low as 1322 cm−1. They observed additional features in the Raman spectra of these diamonds (doublets with peaks at 1328 cm−1 and 1322 cm−1) which are like the Raman spectra of diamonds from the Kokchetav Massif. Given the Raman results and the in situ nature of the diamonds the geological origin of the Straumen diamonds is indisputable.

Tonsvika area, Tromsø nappe, the northern Norwegian Caledonides.Janák et al. (2013) documented microdiamonds (~10 µm) in situ in kyanite–garnet–biotite–gneisses exposed near Tonsvika, Tromsø nappe (452 Ma), in the northern part of the Norvegian Caledonides territory (Table 1). Microdiamonds were observed in thin sections with optical microscopy and confirmed with Raman spectroscopy. Microdiamonds occur as single crystals as well as multiphase inclusions of diamond + MgCO3 and/or diamond + CaCO3 in garnet. In addition to the diamond and diamond–carbonate inclusions, monocrystalline quartz with radial fractures was observed suggesting that this quartz is a product of decompression of coesite. Most of the Raman spectra collected from 21 microdiamonds in two petrographic thin sections have peaks that are observed between 1332 and 1330 cm−1 suggesting the presence of sp3-hybridized carbon bonding (diamond). The presence of G-bands (1580 cm−1) in Raman spectra indicate that sp2-hybridized carbon bonding (graphite) is also present. The thermodynamic calculations reported by Janák et al. (2013) suggest that diamond crystallization took place at P = 3.5 ± 0.5 GPa and T = 770 ± 50 °C. The microdiamond discovery in Tromsø nappe combined with the previous discoveries of coesite, microdiamonds, majoritic garnets in the neighboring Western Gneiss Region extends this UHPM terrane by >100–120 km.

Jämtland and Årescutan areas, the Sweden Caledonides. Diamond inclusions are documented in 2014 for the first time in porphyroblastic garnets from garnet–sillimanite–biotite gneisses in the Seve Nappe Complex of northern Jämtland, Swedish Caledonides (Majka et al. 2014). Diamonds ranging in size from 4 to 7 µm with spheroidal and bleb-like forms occur inside of a multiphase assemblage of carbonate, quartz, and rutile (Fig. 6). The presence of sp3 carbon bonding was confirmed with Raman spectroscopy, and scanning electron microscope (SEM) images acquired in secondary electron (SE) mode reveal negative relief of diamond inclusions with respect to the very flat, perfectly polished surface of the host garnet (Majka et al. 2014).

Klonowska et al. (2017) discovered another microdiamond locality approximately 50 km from the locality described by Majka et al. (2014). Microdiamonds (~5 µm size) are found in situ as single and composite (diamond + carbonate) inclusions in garnet from garnet–phengite–kyanite–rutile gneisses in the Åreskutan area of the Seve Nappe Complex. Raman modes were observed at 1328 cm−1 and 1331 cm−1 confirming the presence of diamond (Solin and Ramdas 1970). Moreover, besides these diamonds, the gneisses contain textural relicts of decompressed coesite, observed as polycrystalline quartz surrounded by radial cracks. Thermodynamic calculations suggest that diamond crystallization took place at P = 4.1–4.2 GPa and T = 830–840 °C (Klonowska et al. 2017). The mineral assemblages of the UHP metamorphism were later overprinted under amphibolite and granulite facies conditions. The age of the UHP metamorphism recorded in the Åreskutan diamond-bearing gneisses occurred before ~445 Ma (Klonowska et al. 2017).

Diamonds from Lago di Cignana, Italy, Western Alps. Although UHP metamorphism in the Western Alps was proposed in the mid 1980s when coesite first was discovered by Chopin (1984) and later by Reinecke (1991), diamonds were not found in the Western Alps until almost two decades later (Frezzotti et al. 2011, 2014; Frezzotti 2019). Frezzotti et al. (2011) discovered microdiamond inclusions ranging from < 2–30 μm in size in Mn-rich garnets from garnetite nodules and boudins enclosed within mica–schists and quartzites that are spatially associated with eclogite facies metabasites. The upper part of this metamorphosed ophiolitic section belongs to the Lago di Cignana tectonic unit of the Western Alps and appears to be the first recognized UHPM terrane which originated as a fragment of the Thetyan oceanic lithosphere. The tiny diamond inclusions are closely associated with COH fluid inclusions, and tiny inclusions of magnesite and SiO2 (quartz). There is no evidence that the quartz is a product of decompression of coesite during exhumation. Instead, the SiO2 clusters are products of lower temperature post–metamorphic reactions that occurred in adjacent layers of the mica-schists (Frezzotti et al. 2011, 2014).

Detailed studies of diamond and COH fluid enclosed in garnets using Raman spectroscopy and C and O isotope geochemistry revealed that organic carbon dissolved in hydrous fluids was transported to a depth ~110 km together with oceanic sediments during Alpine oceanic subduction about 35 Ma ago. Frezzotti et al. (2011, 2014) and Frezzotti (2019) concluded that during subduction the UHP metamorphism led to diamond crystallization from COH fluid at respectively high oxidation state [fO2 = 0–1.5 log units higher than the fayalite–magnetite–quartz (FMQ) buffer]. This UHP metamorphism took place at P ≥ 3.2 GPa and T ≤ 600 °C and is classified as a cold subduction metamorphism (e.g., Frezzotti 2019).

Diamonds from Pohorje (Slovenia) in the Austro-Alpine UHPM terrane, Eastern Alps. The Pohorje Mountains have been considered as an UHPM terrane since Janák et al. (2004, 2006) constrained the pressure and temperature (P ≥ 3.5 GPa and T =800–850 °C) for eclogite and peridotite enclosed in meta-carbonaceous rocks, although at that time no UHP index minerals had been known in this locality. Later, Janák et al. (2015) discovered ~10 µm microdiamond inclusions in garnets from metapelitic gneisses that consist of layers rich in biotite, white mica, garnet and kyanite alternating with felsic layers, veins and segregations of quartz and feldspar. Inside of the host-garnets diamond occurs as single cubes or cubo-octahedral crystals and as a part of composite inclusions where it is associated with moissanite (SiC) and fluid inclusions containing abundant CO2 and traces of CH4. Diamonds, associated minerals, and fluid inclusions were confirmed with Raman spectroscopy. The authors concluded that diamond formed at depth > 100 km during Late Cretaceous (ca. 92–95 Ma) subduction of the continental slab from reduced supercritical fluids that formed during UHP metamorphism of carbonaceous sediments. This diamond-forming media is similar to that reported by Perraki and Faryad (2014) in the Gföhl granulites of the Moldanubian Zone of the Bohemian massif, where they documented polyphase inclusions of diamond+moissanite+coesite in zircon.

More diamonds from Qinling region, China. Though the Central Orogenic belt (COB) of China is a well-established UHPM terrane where coesite (e.g., Okay et al. 1988) as well as quartz paramorphs possibly after stishovite (Liu et al. 2007, 2018) and majoritic garnet with lamella exsolution of ortho- and clinopyroxenes (Ye et al. 2000) are well-documented, diamonds in these rocks are scarce. There are two reports of diamond occurrences in the Dabie Mountains (Xu et al. 1992, 2003) and the Qinling region of China as inclusions in zircons (Yang et al. 2003). However, these reports were not confirmed by others for more than a decade.

Wang et al. (2014) documented more microdiamonds as inclusions in zircons that had been separated from amphibolite rocks from the North Qinling region. To date, all microdiamonds reported from the COB have only been confirmed with Raman spectroscopy. Advanced electron microscopy studies of these diamonds such as FIB and TEM have not yet been reported. Nevertheless, both FIB and TEM research on these unusual diamonds is very important since it would likely shed light onto the geochemical conditions under which these diamonds crystallized. Considering that very few microdiamonds have been documented in the UHPM rocks of the COB it is understandable that researchers may be skeptical of these reports. It is possible that the low concentration of diamonds in the COB may be explained by geochemical conditions that would impede diamond crystallization. Hypothetically, such conditions could be: (i) low concentration of carbon in the bulk rocks and/or in small amounts of fluid/hydrous melt that can penetrate subducting slabs during UHP metamorphism; (ii) perhaps the concentration of carbon was high enough to start diamond nucleation, but the oxygen fugacity was extremely high which would prevent diamond crystallization; (iii) or all the diamonds were replaced by graphite during slow exhumation of the UHPM rocks accompanied by low-pressure and high-temperature regional metamorphism.

Graphitized diamond in West Greenland UHPM terrane. A new UHPM terrane was documented by Glassley et al. (2014) within the Nagssugtoqidian Orogen (Proterozoic age, ca. 1.8 Ga) of West Greenland. The Orogen consists of undifferentiated metasedimentary/metavolcanic rocks and amphibolites. They include strongly deformed quartzo-feldspathic granitic and dioritic gneisses, garnet–sillimanite–graphite–schists, and amphibolites; all which were traditionally considered as upper amphibolite to granulite facies from regional metamorphism (e.g., Davidson 1979; Glassley and Sørensen 1980; Mengel 1983).

The mineralogical confirmation of different UHP events includes (i) the presence of relicts of orthopyroxene exsolved from a supersilicic garnet precursor, (ii) exsolution lamella of rutile from garnet and pyroxenes, (iii) exsolution lamella of magnetite from olivine, and (iv) unusual quartz needles exsolved from olivine—a hypothetical ringwoodite. Glassley et al. (2014) showed that the Raman spectra of three mm-sized carbon-bearing inclusions in garnet exhibit modes at 1335 cm–1, which unambiguously suggests the presence of sp3-carbon bonds, e.g., the diamond structure. Broad peaks at ~1420–1450 cm−1 which are consistent with hydrogenated carbon on the diamond surfaces and well-pronounced Raman bands of graphite at 1580 cm–1 suggest that diamond probably gradually transformed to graphite. The authors hypothesized that these diamonds were encapsulated in the garnet during their growth and were transformed to graphite during decompression accompanied by metamorphism to granulite facies. The UHP metamorphic conditions suggest a pressure of ~ 7 GPa and a temperature of ~975 °C.

The Western Mediterranean orogen—the Edough Massif, NE Algeria, new UHPM terrane. Several microdiamond crystals were discovered by Caby et al. (2014) as inclusions in garnet megacryst (≥ 5 cm) collected from a mélange zone exposed at the northern margin of Africa (the Edough Massif, NE Algeria). The garnet megacrysts were adjacent to actinolite-bearing and ultramafic boudins associated with tectonic fragments of marbles occurring within a major mylonite–ultramylonite band. Several cuboid-like diamond crystals ranging from 3–50 µm in size were identified with Raman spectroscopy (Caby et al. 2014). Later, a single diamond crystal approximately 50 µm in size intergrown with rutile enclosed in an almandine-rich garnet was studied in detail with the aid of SEM techniques (Bruguier et al. 2017). The secondary electron SEM image presented in Figure 3a from Bruguier et al. (2017) is unequivocal evidence that the diamond–rutile inclusion is indigenous (e.g., not from sample preparation or contamination) to the Edough massif. The image shows that the diamond is clearly “protruding above” the perfectly polished surface of the host mineral, which is very typical for a diamond–garnet interface. The contact zone between the diamond–garnet interface, appears intact which strongly suggests that the diamond is naturally occurring and not contamination from a cutting with a diamond-saw and/or polishing with diamond grit. Other garnets from the same area contain rutile, zircon, apatite, and plagioclase inclusions.

Thermodynamic calculations suggest that an UHP metamorphic event took place at P ≥ 3.6 GPa and T = 750 °C (Bruguier et al. 2017; Caby et al. 2014). Studies of the zircon inclusions suggest that the Alpine age of UHP metamorphism of the West Mediterranean Orogen is between 32.4 ± 3.3 Ma and 20.7 ± 2.3 Ma (Bruguier et al. 2017). This was followed by exhumation as slab rollback occurred in the Oligocene (Brun and Faccenna 2008; Tirel et al. 2013). Additional studies of this locality should find more evidence of UHPM, such as UHP minerals and/or traces of their decompression products and/or proof of retrograde metamorphism.

Diamonds from Nishisonogi unit, Nagasaki metamorphic complex, western Kyushu, Japan.Nishiyama et al. (2020) report the discovery of microdiamond aggregates in the matrix of a metapelite from the Nishisonogi unit, Nagasaki metamorphic complex, western Kyushu, Japan. The Nishisonogi unit represents a Cretaceous subduction complex which has been considered as an epidote–blueschist subfacies metamorphic unit, and the metapelite is a member of a serpentinite mélange in the Nishisonogi unit (Nishiyama et al. 2020). The glaucophane and phengite ages are reported to be 95–90 Ma using the 40Ar/39Ar method (Faure et al. 1988) and 85–60 Ma using the K–Ar method (Hattori and Shibata 1982). The aggregates of diamonds range in size from 10 to 50 μm and they are embedded in dolomite, phengite, and albite (see Fig. 4 in Nishiyama et al. 2020). The diamonds are not sticking out above the polished surface of the sample. However, it should be noted that they are surrounded by soft minerals. Using SiC and Al2O3 abrasives as the main polishing abrasives, one would expect to see the diamonds “standing above” the flat surface of the surrounding softer minerals (see Fig. 4 in Nishiyama et al. 2020). This is because of the hardness of diamond itself, and it would not be “eroded” or ‘flattened” by either SiC or Al2O3 abrasives since these abrasives are not as hard as diamond. Unfortunately, images where the diamonds are completely enclosed by the host minerals are not presented in this publication.

The authors characterized the microdiamonds with electron diffraction and Raman spectroscopy and carefully discussed both possibilities of metastable and stable conditions of diamond crystallization in metapelites of the epidote–blueschist subfacies from a Cretaceous subduction complex in Western Kyushu, Japan. They considered several lines of evidence for HP to UHPM conditions such as the presence of quartz pseudomorphs after coesite and high Si content in phengite. They speculate that almost all other HP–UHPM mineral assamblages have beed oblitered due to the retrograde metamorphism. Ultimately the authors hypothesized that the diamond crystallized from a COH fluid at P > 2.8 GPa and T = ~ 450 °C.

The authors specify that no diamond abrasives were used for sample polishing, and that ony Al2O3 and SiC were used. However, they did not specify if any diamond-bearing tools were used to extract or cut the samples prior to their polishing. Though Nishiyama et al. (2020) used FIB-TEM techniques they did not carefully examine the diamond–host mineral interfaces. In Figure 5a,b from Nishiyama et al. (2020) they show STEM images at different magnifications aquired from a FIB milled foil from the sample area contaning microdiamond aggregate. In Figure 5 plate c, Nishiyama et al. (2020) show a d-spacing = 2.06 Å (direction [111]) for the inner spot. Importantly, these data neither prove nor disprove that the diamond is indigeneous. To unambiguously classify the metapelites from a Cretaceous subduction complex in Western Kyushu, Japan as a new UHPM terrane more FIB-TEM research is required on both the existing samples and/or new samples from this locality, and studies of the diamond–host mineral interface are necessary to confirm that the diamond is indigenous.

Lonsdaleite from WGR, Norway and Kokchetav massif, Kazakhstan, and a problem with its identification

Since lonsdaleite, a hexagonal polymorph of diamond, has been reported to occur in UHPM terranes as well as in meteorites and impact craters, a short discussion is warranted. At General Electric Company Bundy and Kasper (1967) synthesized a wurtzite like polymorph of carbon at P = 13 GPa and T = 1000 °C, and they named it hexagonal diamond. In nature hexagonal diamond was first identified from the Canyon Diablo iron meteorite and Goalpara ureilite (e.g., Frondel and Marvin 1967; Hanneman et al. 1967). It was named lonsdaleite in honor of the distinguished crystallographer Professor Kathleen Lonsdale (Frondel and Marvin 1967). Following this initial report lonsdaleite has been found in many other meteorites and impact craters (e.g., Daulton et al. 1996; Clarke et al. 1981; Kvasnitsya et al. 2013; Nakmuta and Toh 2013; Nestola et al. 2020). It has been reported to occur in the following terrestrial impact craters: Ries Crater, Germany (e.g., Rost et al. 1978), Popigai impact structure (e.g., Masaitis et al. 1972; Koeberl et al. 1997; Murri et al. 2019), Sudbury impact structure, Ontario, Canada (e.g., Masaitis et al. 1997), and Lappajärvi structure, Finland (e.g., Masaitis et al. 1998). There is also one report of lonsdaleite occurrence in nodules from Pipe 50, Fuxian kimberlite field, China (Leung and Winston 2002). These discoveries led to the hypothesis that lonsdaleite forms during impact events and very high pressure and temperatures are needed. However, it has also been reported to occur in UHPM rocks which would require alternative hypothesis to explain their existence.

Lonsdaleite has been reported in WGR, Norway (Godard et al. 2003, 2004; Smith et al. 2004) and in Kumdy Kol, Kokchetav massif, Kazakhstan (Smith 2004; Smith et al. 2004, 2011; Dubinchuk et al. 2010; Shumilova et al. 2011). Since these localities have been proven to be UHPM terranes by numerous studies, which clearly showed that they did not form during impact events, alternative hypotheses are required to explain the existence of lonsdaleite in UHPM rocks. One hypothesis is that the formation of hexagonal polytypes such as lonsdaleite could be cumulative with the metamictization of diamond (Smith et al. 2011; Smith and Godard 2013). Another hypothesis, which will be discussed in greater detail below, is that they formed during an impact event with a comet (Tretiakova and Luykin 2016, 2017).

Raman spectroscopy has been one of the most common techniques for confirming the presence of natural lonsdaleite since it is non-destructive, and mm-sized samples can be studied in situ (Smith and Godard 2009; Shumilova et al. 2011; Goryainov et al. 2014). Lonsdaleite has a 2H structure (spacegroup P63/mmc) which has three theoretically predicted Raman active modes, the transverse optical oscillations of the E2g mode, and the longitudinal A1g and transverse E1g optical modes. Wu and Xu (1998) report the following theoretically predicted active modes E2g 1193 cm−1, A1g 1312 cm−1, and the E1g 1305 cm−1, while Denisov et al. (2011) predict them to be at 1221, 1280, and 1338 cm−1 respectively. The E2g mode is predicted to have very low intensity so the A1g and E1g modes are typically used to identify lonsdaleite. Since the predicted lonsdaleite modes occur near the cubic diamond Raman active T2g mode it should be noted that the peak position and width of the cubic diamond mode can vary significantly. As discussed by Nasdala et al. (2016) the peak position of the Raman active diamond mode can be downshifted and broadened by local heating through absorption of the laser (Herchen and Cappelli 1991), nanometer sized crystals (Lipp et al. 1997), substitution of 13C for 12C (Anthony and Banholzer 1992), incorporation of B-impurities (Pedroza-Montero et al. 2005) and accumulation of structural damage from nearby radioactive phases (Nasdala et al. 2013). Hence, one needs to be cautious when interpreting Raman spectra of natural diamonds since shifts in the cubic diamond Raman peak position and peak width could be misinterpreted as lonsdaleite.

Notably, there appears to be no consensus on the existence of lonsdaleite as a discrete mineral (e.g., Németh et al. 2014). Németh et al. (2014) point out that despite extensive efforts lonsdaleite has never been produced as a pure material and they show that defects in cubic diamond provide an explanation for the characteristic d-spacings and reflections reported for lonsdaleite. Natural lonsdaleite has been described as stacking disordered diamond where cubic and hexagonal sequences are interlaced in a complex way (e.g., Salzmann et al. 2015; Murri et al. 2019). Thus, while numerous studies of lonsdaleite have been reported from UHPM terranes, more studies of natural lonsdaleite using high-resolution techniques are needed to resolve this controversy.

Carbon and nitrogen isotopes, and nitrogen content in UHPM diamonds

Carbon isotopic composition (expressed as δ13C, see Stachel et al. 2022, this volume for definition) is a key parameter used to constrain the origin and formation environment of a diamond. Since UHPM diamonds formed in situ in subducted continental lithologies, their carbon isotope ratios should reveal the origin of carbon, and the isotope fractionation of the COH fluids from which UHPM diamond formed. These carbon isotope ratios record the formation conditions both before and during diamond crystallization.

Among the reports on carbon isotopic compositions, E-type (eclogitic) and P-type (peridotitic) diamonds (e.g., Harris et al. 1975; Cartigny 2005; Stachel et al. 2022, this volume) found in kimberlites clearly show sfome similarities; both types of diamonds have a peak at approximately −5‰, which is similar to mantle-derived carbon observed in mid-ocean ridge basalts (MORBs), carbonatites, and kimberlites (Table 2, Fig. 7). The E-type diamond carbon isotopes are also spread towards lighter carbon to −41‰ (De Stefano et al. 2009; Smart et al. 2011), suggesting their crustal origin (Fig. 7).

A typical feature of UHPM diamonds is their yellow color (e.g., Schertl and Sobolev 2013) due to the presence of isolated nitrogen, which is the main substitutional impurity in diamond. Nitrogen has an extra electron compared to carbon, and the incorporation of a few ppm of isolated nitrogen atoms and their accompanying electrons changes the energy band gap in diamond to a level equivalent to absorption of blue and violet lights, resulting in a yellow color. The concentration of nitrogen, which is the main substitutional diamond impurity, is also an important parameter when it is combined with δ13C to constrain the origin of metasomatic COH fluids and therefore diamond formation conditions (Table 2, Fig. 8). As noted by Cartigny et al. (2014), the behavior of nitrogen during diamond formation is unclear since nitrogen in diamond may be both compatible and incompatible (Smart et al. 2011; Palot 2013). On the contrary, if nitrogen compatibility is experimentally constrained, it would shed light on the composition, redox state, and evolution of the fluids related to UHPM diamond formation.

Nitrogen isotopic composition (expressed as δ15N, see Stachel et al. 2022, this volume for definition) also provide a clue to the origin of diamonds. Crustal nitrogen is enriched in 15N, by approximately +5‰ compared with the atmosphere (δ15N = 0‰), whereas mantle nitrogen is depleted in 15N by approximately −5‰. Being different to carbon, nitrogen isotopes are fractionated by only a few per mill during their incorporation into diamond, so nitrogen that is incorporated into diamond during crystallization is representative of the growth medium. Most E-type diamonds have a mantle-like δ15N, which is inconsistent with their origin from subducted organic carbon. On the other hand, UHPM diamonds systematically exhibit heavy δ15N values ranging from +2 to +12 ‰ (Cartigny et al. 2001, 2014), suggesting their sedimentary source because devolatilization during sediment subduction would preferentially release 14N, resulting in further enrichment of subducted material in 15N (Bebout and Fogel 1992; Busigny et al. 2003). For thorough discussion on the origin of carbon and nitrogen isotopic compositions of diamond, the reader is refered to Stachel et al. (2022, this volume).

Carbon isotopes characteristics of the Kokchetav diamonds. Carbon isotopic composition studies of UHPM diamonds are limited due to their mm-size. One approach involves separating and grouping tens of thousands of microdiamonds and processing them together so that there is enough CO2 to be analyzed with a conventional isotope-ratio mass spectrometer, this approach provides the average δ13C ratio. For instance, Cartigny et al. (2001) reported δ13C = −11 ‰ to −10 ‰ in diamonds from garnet–clinopyroxene rocks and nitrogen contents of 11,150 ppm, while diamonds from marbles of the Kokchetav massif, Kazakhstan have a δ13C = −10.2‰ to −8.5‰ and 2650 ppm of nitrogen (Table 2, Fig. 8). Based on possible carbon isotope fractionation between coexisting carbonate and diamond, Cartigny et al. (2001) estimated diamond crystallization conditions as T ≤ 700 °C and P = 3 GPa, e.g., slightly below the peak of UHPM, and concluded carbon and nitrogen stable isotopes strongly support their metasedimentrary origin. If Cartigny et al. (2001) data is combined with earlier δ13C bulk measurements of the Kokchetav diamonds, they fall in range of −10.2 ‰ to −26.9 ‰ (e.g., Lavrova et al. 1997; Chopin and Sobolev 1995; Ogasawara 2005).

Later, Imamura et al. (2013) using Secondary Ion Mass Spectrometry (SIMS) technique analyzed carbon isotopes in diamonds (10–25 μm in size) included in garnet from dolomite marble of the Kokchetav Massif. These diamonds were subdivided on to morphological types: S-type consisting of single-crystal core and polycrystalline rim, and R-type presented by single crystal with rugged surfaces. In S-type diamonds the rim exhibits lighter δ13C = −17 to −27‰, whereas the core is much heavier, with δ13C = −9 to −13‰. The R-type diamonds fall into narrow ranges of δ13C = −8 to −15‰, which are like those in the core of the S-type diamond. This suggests that the R-type diamonds probably formed at the same stage as the core of the S-type, whereas rim growth during the second stage did not occur, or occurred, very weakly in R-type microdiamonds. Therefore, carbon isotopic data corroborate a two-stage growth of microdiamond in the Kokchetav dolomite marbles (Imamura et al. 2013).

Interestingly, the δ13C value of dolomite from the host rocks ranges from −4 to −7‰, whereas the bulk dolomitic marble exhibits δ13C = −2‰ (Ohta et al. 2003). The marginal range of δ13C = −8 to −27‰ in S-type and R-type diamonds is different than that in mineral dolomite (δ13C = −4 to −7‰) and dolomitic marble host rock (δ13C = −2‰). It is difficult to attribute these differences to isotopic fractionation considering that the metamorphic temperature was very high (∼1000 °C; e.g., Ogasawara et al. 2000). Under such conditions, the diamonds in dolomitic marble would not have been in carbon isotopic equilibrium with the dolomite. There appears to be no good candidate material or process that could occur in the dolomitic marble that would explain the origin of the very light δ13C values between −17 and −27‰, measured in the rim of the S-type diamonds. Imamura et al. (2013) speculated that perhaps a fluid with a light carbon isotope ratio infiltrated into the dolomite marble from the surrounding gneisses, which might contain light carbon of organic origin. With regards to this explanation, it will be discussed below that the Erzgebirge microdiamonds from felsic gneiss have a similar light carbon, e.g., δ13C = −17 to −27‰ (Dobrzhinetskaya et al. 2010). However, Imamura et al. (2013) have not discussed these data because no carbon isotope measurements in diamonds from felsic gneisses of the Kokchetav massif have been reported.

Carbon isotope characteristics of the Erzgebirge diamonds.Massonne and Tu (2007b) used a bulk analysis approach to obtain the carbon isotope ratios of diamond inclusions in zircons from the Erzgebirge UHPM terrane, and they reported that lighter carbon was involved (δ13C = −24‰ to −33‰) in diamond formation where the precursor was graphite.

Dobrzhinetskaya et al. (2010) used SIMS to analyze the δ13C and nitrogen contents in ~5 to 10 μm diameter diamonds included in garnet from the garnet–quartz–feldspathic gneiss of the Erzgebirge massif (Fig. 9). They showed that the diamonds situated within the garnets closer to the garnet rim zone showed some δ13C variations up to 10‰, with average values ranging from −22 to −26‰ (Table 2, Fig. 9). Their nitrogen contents vary heterogeneously between each individual measurement and between all diamonds without any systematics, ranging from 240 to 4600 ppm. On the other hand, the diamonds situated closer to the central part of the zoned garnet exhibits δ13C = −17 to −19‰, with an average value of −17.8‰. The latter is heavier than another diamond in the intermediate zone of the same garnet (δ13C = −23‰) and the other diamonds situated closer to the outer zone of the garnet (Fig. 10). Nitrogen in the diamonds from intermediate zone varies from 100 to 1200 ppm, with an average value of 820 ppm; and average value of nitrogen in diamonds from outer garnet zone is 1,838 ppm. The nitrogen content data plotted versus δ13C show a weakly pronounced tendency to increase with the increase of lighter carbon isotope content in the diamonds (Fig. 8).

Based on relationships between the diamond inclusion in the garnet (Fig. 10), diamond formation can be constrained by the chemical zoning of the host garnet, and δ13C values of two diamonds. Dobrzhinetskaya et al. (2010) argued that there were two stages of diamond crystallization from a COH fluid rich in biogenic carbon and diverse in minor elements of crustal origin. The δ13C of the first stage of crystallization falls within the range of −17 to −19‰, while the second stage of crystallization is characterized by δ13C = −22 to −26‰. Both δ13C ranges strongly suggest that the carbon from which they crystallizedbelongs to the same biogenic reservoir. Graphite from the 3.7 Ga-old turbidite greywacke and slate from West Greenland have primary δ13C = −19‰, which is similar to the range of biologically reduced carbon, and it is within the range of reduced-carbon compositions found in most modern marine sediments (e.g., Rosing 1999). Dobrzhinetskaya et al. (2010) suggest that the 1st-stage of diamond crystallization was from a more reduced carbon than the second stage of crystallization. This work indicates that within Erzgebirge diamond populations there are two δ 13C signatures which are distinguished from each other not only by δ13C values but also by the positions of the diamond within the garnet where different zones in the garnets are related to recrystallization events during UHP metamorphism. It also suggests that the carbon-rich fluid that deeply infiltrated the subducted sediments was probably reduced during an earlier stage, producing diamonds with slightly heavier isotopes than the diamonds that crystallized during the second stage.

Additional carbon isotope analysis of UHPM diamonds from both Kokchetav and Erzgebirge UHPM terranes and any of the new UHPM localities at small scales (μm to nm) with SIMS are necessary to shed light on the evolution of carbon-bearing fluids in subducted continental lithologies before and during diamond formation.

Noble gas isotopes in diamonds from UHPM terranes

Noble gas isotopes in mantle-derived samples are key tracers of chemical heterogeneity of the Earth’s mantle and the origin of the atmosphere (e.g., Ozima and Podosek 2002; Porcelli and Ballentine 2002). Samples of MORBs (quenched glass of mafic minerals) and ocean island basalts (OIB) have provided the most comprehensive understanding of mantle noble gases. MORBs show relatively uniform 3He/4He ratios around 1.1 × 10−5 and a trend in 22Ne/22Ne–20Ne/22Ne ratios toward a composition more enriched in nucleogenic 21Ne. In contrast, OIB samples, which are derived from a deeper region of the mantle, exhibit higher 3He/4He ratios of up to 7 × 10−5 and less nucleogenic Ne isotope compositions (Honda et al. 1991; Trieloff et al. 2000; Graham 2002; Stuart et al. 2003). The latter characteristics are believed to be evidence for a mantle source in which primordial He, with 3He/4He of (1.4–4.6) × 10−4, and Ne (“solar” or “Ne–B” composition, Honda et al. 1991; Trieloff et al. 2000; Ballentine et al. 2005) has been less diluted by the addition of radiogenic 4He and nucleogenic 21Ne produced by decay of U- and Th-series elements. The primordial noble gas source is considered to have been isolated from whole mantle convection over geological timescales and sampled by upwelling mantle plumes (Jackson et al. 2010; Mukhopadhyay 2012). However, the reasons for the less radiogenic/nucleogenic character of the plume source are still under debate; it may be “undegassed” (e.g., Allègre et al. 1983; Kaneoka 1983; Kurz et al. 1982; Porcelli and Ballentine 2002; Porcelli and Elliott 2008; Mukhopadhyay 2012), or it may be depleted in U and Th by ancient melt extraction events (Parman 2007; Porcelli and Ballentine 2002; Jackson et al. 2010). The main advantage of analyzing noble gases contained in diamonds is their great potential to constrain the noble-gas state of the deep mantle because they are formed deeper than 150 km and act as chemically/mechanically stable “capsules”. Most noble gas studies are focused on kimberlitic diamonds (e.g., Basu et al. 2013; Timmerman et al. 2018).

“Unprecedentedly” high Helium-3 content in the Kokchetav diamonds: myth or reality? Although currently there are dozens of UHPM diamond-bearing terranes that are known, only a few noble gas studies of microdiamonds from the Kokchetav massif have been reported (Begemann 1994; Pleshakov and Shukolyukov 1994; Shukolyukov et al. 1993; Verchovsky et al. 1993; Sumino et al. 2011). Due to the small (micrometer) size of UHPM diamonds, noble gas analyses were conducted using several tens-of-thousands (or even a greater number) of microdiamonds that were separated from their host rocks/minerals by treatments with HF and oxidizing acids. Pleshakov and Shukolyukov (1994) and Shukolyukov et al. (1993) reported “unprecedentedly” high 3He/4He ratio during in vacuo stepwise heating of Kokchetav microdiamonds with a maximum value of 0.47 in the low temperate release, which was the highest He content known among terrestrial diamonds (Pleshakov and Shukolyukov 1994; Shukolyukov et al. 1993). Moreover, this high 3He/4He value exceeds the maximum values reported for any terrestrial mantle-derived sample of ≤ 7 × 10−5 (Stuart et al. 2003) and even the primordial 3He/4He value of (1.7–4.6) × 10−4 (Porcelli and Elliott 2008).

The most plausible source of the high 3He/4He ratio is nucleogenic 3He production via the 6Li(n, α)3H (β) 3He reaction (Kurz et al. 1987). However, this requires an anomalously high Li content of ~100 ppm either in the diamonds, or within a ~20 µm vicinity in the host rocks (Begemann 1994; Verchovsky et al. 1993). Moreover, the host rock would need to contain more than 1% of U (Th/U = 3.3 is assumed) to provide a neutron source sufficient to produce the levels of 3He observed (Begemann 1994; Verchovsky et al. 1993). An important feature of the Kokchetav diamonds was that high 3He/4He values were obtained from gas released over the relatively low temperature range of 200–1100 °C during in vacuo stepwise heating (Pleshakov and Shukolyukov 1994; Shukolyukov et al. 1993). These temperatures are significantly lower than the diamond graphitization temperature (over 1700 °C), and lower than the temperatures at which noble gases trapped in the diamond lattice are released. This suggests that the high 3He/4He component was not hosted by the diamond itself. Although Verchovsky et al. (1993) confirmed release of He with high 3He/4He (2.4 × 10−4) during in vacuo heating of the Kokchetav diamonds at 820 °C. However, He released during diamond combustion under a controlled oxygen atmosphere showed a rather low 3He/4He ratio of ≤ 3 × 10−6. Additionally, Verchovsky et al. (1993) showed that the amount of ‘excess’ 3He (as well as 3He/4He ratio) strongly depends on the grain-size of diamond fractions: the coarser the fraction the less (by orders of magnitude) the 3He concentration. These findings were interpreted to indicate that the high 3He/4He component may reside in Li-rich contaminants, which are quite resistant to treatments with HF and oxidizing acids during diamond separation from their host minerals/rocks, and therefore not trapped within the diamond itself (Begemann 1994; Verchovsky et al. 1993).

To clarify this “unprecedented” 3He/4He value Sumino et al. (2011) focused their attention on the presence of abundant nanometer-sized fluid inclusions in Kokchetav microdiamonds (Dobrzhinetskaya et al. 2001, 2005, 2006a), which could provide potential sites for preserving the noble gas signature of the diamond-forming COH fluid. Sumino et al. (2011) employed in vacuo sequential dynamic crushing extraction with which sample grains are mechanically crushed in vacuum in order to selectively extract noble gases from fluid/melt inclusions with significantly less noble gas release from the diamond lattice and any associated mineral inclusions (Kurz 1986; Kurz et al. 1987).

The crushing technique results were compared with those obtained from a conventional vacuum stepped heating method which releases noble gases from fluid and solid inclusions and the diamond lattice. According to Sumino et al. (2011) the diamond crushing process extracts most of the 3He, indicating that 3He occurs within inclusions trapped during the Kokchetav diamond formation. The inclusion-hosted 3He/4He of (3.3–6.5) × 10−5 is significantly higher than that of the MORB-source mantle (1.1 × 10−5), but close to the maximum value observed in OIBs (~7 × 10−5) containing noble gases enriched in a primordial component and delivered from the deep mantle by plumes. On the other hand, very low 3He/4He (5.0 × 10−8) obtained by heating of the crushed powder supports the presence of radiogenic 4He in the diamond lattice, resulting from α-particle implantation from U- and Th-series radioactive elements in the diamond-bearing minerals, which is consistent with the combustion results by Verchovsky et al. (1993). The results of Sumino et al. (2011) clearly indicate that the fluid inclusions in the Kokchetav diamonds preserve primordial He similar to those observed in OIBs and that abundant He with “unprecedentedly” high 3He/4He ratio reported by previous works (Shukolyukov et al. 1993; Pleshakov and Shukolyukov 1994) should be attributed to Li-rich contaminants resistant to separation treatments of the diamonds from their host rocks/minerals (Verchovsky et al. 1993; Begemann 1994).

Sumino et al. (2011) showed that isotope compositions of Ne (20Ne/22Ne and 21Ne/22Ne ratios) extracted from the Kokchetav diamonds during relatively low-temperature heating steps are on the trend defined by OIBs from deep-mantle-derived hotspots (e.g., Hawaii, Iceland, and Galapagos) which are enriched in primordial Ne (Honda et al. 1991; Trieloff et al. 2000; Kurz et al. 2009) but deviate from the trend defined by MORB data (Sarda et al. 1988) (Fig. 11). The Ne-isotope data indicate that deep-mantle-derived, primordial Ne is an intrinsic feature of the microdiamonds, but at higher temperatures during diamond graphitization, the Ne released shows a nucleogenic contribution. The Ne isotope data indicate that primordial Ne is an intrinsic feature of the microdiamonds, but at higher temperatures during diamond graphitization, the Ne released shows a nucleogenic contribution. The nucleogenic Ne resides in the diamond lattice as a result of implantation by recoiling Ne isotopes mainly produced during the reaction of α-particles (from U–Th decay) with oxygen and fluorine atoms in surrounding rocks (Kurz et al. 1987; Lal 1989; Verchovsky et al. 1993). The nucleogenic Ne composition released from the microdiamonds requires a very high F/O ratio of 0.02 (Fig. 11B) compared with the average crustal values of 0.0013 (Ballentine and Burnard 2002).

The observation that deep-mantle-derived, plume-like noble gases are contained in the Kokchetav microdiamonds paradoxically requires that deep-mantle-derived volatiles be present in the relatively shallow mantle wedge at 190–280 km depth. To solve this paradox, Sumino et al. (2011) suggest a few possible processes which are either capable of bringing deep-mantle-derived noble gases to the relatively shallow mantle: (i) a plume tapped from the lower mantle or core–mantle boundary; and/or (ii) a large-scale mantle convection cell which entrained a fragment of the deep mantle, and subsequently added it to the mantle wedge at depths of 190–280 km.

To date there is no geological evidence in the vicinity of the Kokchetav massif of mantle plume activity at the time of diamond formation (~530 Ma). A plume may have interacted with the subcontinental lithosphere which was beneath metasedimentary rocks before the Kokchetav slab was subducted and microdiamonds formed. To clarify the origin of the plume-like, deep-mantle-derived noble gases in the Kokchetav microdiamonds, more investigation is required particularly on UHPM diamonds from other localities. If metasomatism by plume-derived volatiles is essential for subduction and exhumation of continental crust as proposed by Seno and Rehman (2011), and the metasomatic volatiles play an important role in UHPM diamond formation, then plume-like noble gas is expected to be observed in UHPM diamonds from other localities.

It is generally agreed upon that during a continent–continent collision crustal rocks can be subducted to depths of >150 km, which leads to the formation of UHPM terranes, however readers may be surprised to learn that there are countless hypotheses for the origin of microdiamonds found in these terranes. The debates are primarily focused around the crystallization conditions of diamond, some of the hypotheses include (i) from carbon-rich supercritical fluids vs. hydrous silicate and/or carbonatite melts, (ii) oxidation state of the media (e.g., see reviews: Ogasawara 2005; Frezzotti et al. 2011, 2014; Dobrzhinetskaya 2012; Schertl and Sobolev 2013; Liou et al. 2014; Zhang et al. 2017; Frezzotti 2019), (iii) stable vs. metastable conditions (e.g., Pechnikov and Kaminsky 2008, 2011; Simakov 2010), and/or (iv) from meteorite impact (Tretiakova and Luykin 2016, 2017). These hypotheses will be discussed individually in greater detail below.

During the last decade there have been attempts to resolve some of these disagreements which have resulted in considerable progress in our understanding of the formation conditions of microdiamonds. Particularly advancements in high resolution instruments such as STEM, FIB, a ‘Structural & Chemical Analyser’ (SCA) probe and ‘inVia’ Raman spectrometer (a Renishaw trademark) combined with SEM system (e.g., O’Bannon et al. 2020), mm and sub-mm monochromatic and white beam synchrotron X-ray diffraction (e.g., Stan et al. 2020) and synchrotron infrared (IR) micro-spectroscopy (e.g., Dobrzhinetskaya et al. 2006b) have all made it possible to perform analyses at the nano-scale of solid and gas/liquid inclusions in microdiamonds. Table 3 presents summarized information on nanoscale inclusions detected in diamonds from UHPM terranes by different techniques.

As new instruments and technology are adopted by researchers it has led to many other important discoveries and increased our knowledge about the origins of microdiamond. Moreover, the integration of high-pressure devices such as multianvil presses and diamond anvil cells into synchrotron X-ray facilities has enabled a wide range of experiments to study UHPM diamond crystallization at varying pressures, temperatures, fO2, XH2O, organic/inorganic carbon and different bulk chemistry of the starting materials (e.g., Akaishi and Yamaoka 2000; Yamaoka et al. 2000; Kumar et al. 2001; Pal’yanov et al. 2002, 2013; Dobrzhinetskaya et al. 2004; Dobrzhinetskaya and Green 2007a; Zhang et al. 2011). These experiments have demonstrated that the kinetics of diamond crystallization varies depending on different experimental conditions such as hydrous or anhydrous, oxidation states, and if the pressure and temperature are high enough to stabilize sp3 carbon bonding.

UHPM diamonds crystallized from a multicomponent supercritical COH fluid or a hydrous silicate melt. Studies of nanometric fluid and solid inclusions in microdiamonds (Table 3) have revealed important information on the crystallization conditions of microdiamonds. Conventional infrared (IR) spectroscopic studies revealed the presence of molecular H2O and CO32− radical from Kokchetav microdiamonds occurred in garnet clinopyroxenite, calc-silicate, garnet–pyroxene–quartz rocks and gneisses (e.g., de Corte et al. 1998, 2002; Sitnikova and Shatsky 2009). Dobrzhinetskaya et al. (2006b) carried out synchrotron IR spectroscopic studies of the Erzgebirge diamonds separated from garnet–phengite–quartz–feldspathic gneisses using a thermochemical microwave digesting method. They collected synchrotron IR spectra from two diamond crystals (Fig. 12). Both spectra (Fig. 13) show the presence of nitrogen impurities, molecular H2O, OH hydroxyl radical, CO32− radical, and silicate inclusions in addition the C–C diamond phonon absorption bonds. Dobrzhinetskaya et al. (2006b) intentionally studied the same samples with synchrotron IR that had been previously studied using FIB-TEM techniques (Dobrzhinetskaya et al. 2003). Abundant nanometric solid crystalline inclusions were found in the diamond FIB-foils; some of the identified phases are: (i) SiO2-containing traces of K, P, Ti, Fe; (ii) inclusion of unknown stoichiometry which consists of three elements, K, P, and O, (iii) oxide PbxOy, (iv) carbonate (Pb2CO3), and (v) Al2SiO5. Due to the extremely small size of these inclusions and their random orientations, their lattice parameters could not be determined by selected-area electron diffraction (SAED), however, all of them yield Bragg diffraction spots confirming their crystalline nature.

Besides the presence of crystalline nano-inclusions a considerable amount of nanometric-size euhedral cavities representing traces of decrepit fluid were documented. Many of these cavities were associated with dislocations related to diamond growth and are evidence of diamond nucleation and growth that took place in a COH fluid-rich media (Dobrzhinetskaya et al. 2003). The presence of larger scale diamond-bearing multiphase inclusions (or “pockets”) inside of garnets and zircons (Fig. 14) observed in polished slides with optical microscopy and scanning electron microscopy corroborate perfectly with the nanoscale observations (Stöckhert et al. 2001; Dobrzhinetskaya et al. 2003, 2006a). Moreover, direct TEM observations of fluid-filled nanometric inclusion-bubbles preserved in FIB foil prepared from the Kokchetav diamond, and the serendipitous measurement of their contents immediately after bursting, showed that the fluid consists of C, H, O, Cl, S, Ca, Fe and K (Dobrzhinetskaya et al. 2005). This well-documented direct observation of residual fluid bubbles in microdiamonds unambiguously establishes their fluid growth medium.

Polycrystalline microdiamonds appear to be the best samples to study nano-fluid inclusions. There are several types of polycrystalline diamonds from the Kokchetav massif: (i) “spiral diamonds” which formed from a metal–sulfur–COH–silicate fluid (Hwang et al. 2003), (ii) S-type (star-shaped) and R-type (rugged face) diamonds that crystallized from a fluid during two stages (Ishida et al. 2003; Ogasawara 2005), and (iii) “overgrown” diamonds which crystallized from an aqueous-carbonate and aqueous-silicate fluid/melt (Sitnikova and Shatsky 2009). Polycrystalline diamond has also been described as daughter crystals associated with larger diamonds included in garnets from Erzgebirge felsic gneisses (Stöckhert et al. 2001, 2009). Stöckhert et al. (2009) concluded that the tiny daughter crystals precipitated from a supercritical COH + K-, Na-rich silicate fluid.

Dobrzhinetskaya et al. (2013) conducted SEM and FIB-assisted TEM studies of the Erzgebirge polycrystalline diamond inclusions in zircons (Fig.15). They studied five polycrystalline diamonds and found that they consist of 5–15 single crystals ranging in size from 0.3–5 µm with irregular sharp-angled grain boundaries (Fig. 16). Numerous triangular nanoscale voids situated along the diamond–diamond interfaces were filled with a COH fluid containing traces of Al, Co, F, V, Zn, Si, Cl, S, Ca, Mg, Fe, K in varying combinations and concentrations. The latter, if considered along with the “spider-like” dislocations, sharp-angled grain boundaries, and stacking faults that were observed in these diamonds, suggests that they rapidly nucleated and grew from a fluid media under an internal stress. These observations combined with previous results provide an additional constrain for the formation conditions of these UHPM microdiamonds. They likely crystallized from a supercritical fluid with a composition that was close to one of two end members, (1) a carbon-rich hydrous-silicic fluid and (2) a hydrous-saline fluid (Dobrzhinetskaya et al. 2006a, 2013; Jacob et al. 2014).

Zhang et al. (2017) also conducted studies of polycrystalline diamonds in zircons separated from quartz–feldspathic gneisses at Saidenbach water reservoir, Erzgebirge, Germany. Their conclusions confirm the studies by Dobrzhinetskaya et al. (2013), and unconditionally support the concept of UHPM diamond crystallization from a multicomponent COH fluid or COH fluid–hydrous silicate melt media.

Frezzotti et al. (2011) studied fluid inclusions associated with UHPM diamond formation in metamorphic schists from Lago di Cignana, Western Alps where the protolith of these rocks was likely oceanic sediments. Based on studies of the fluid geochemistry they concluded that the carbon was released from the subducting oceanic sediments at relatively shallow depths through the dissolution process (Frezzotti et al. 2011, 2014; Frezzotti 2019). Furthermore, their detailed investigation convincingly demonstrated that the deep subduction of oceanic sediments leads to a continuing saturation of the shallow fluid with carbon and precipitation of diamonds at UHPM conditions occurred. The Lago di Cignana diamond’s nucleation from a COH fluid at UHPM conditions is unambiguously supported by the presence of coesite inclusions located in the same garnet growth zones where the primary fluid inclusions are located (Frezzotti et al. 2011, 2014; Frezzotti 2019). This is also in good agreement with thermodynamic calculations reported by Sverjensky et al. (2014) who investigated the stability of COH fluids as a function of pressure, temperature, and oxidation state.

A literature survey on UHPM diamond crystallization reveals that sometimes discussions are oriented around “fluid vs. melt” or “fluid vs. hydrous melt.” Without clear a definition of how one should distinguish a silica-rich fluid from a hydrous silicate melt at high pressures and high temperatures, this controversy simply reflects a semantic issue. For example, experiments conducted by Bureau and Keppler (1999) suggest that there is a complete miscibility between silicate melt and water in most of the upper mantle, except at very shallow depths. In these experiments, the hydrous melt and silicate-bearing vapor coexisted separately at low T, whereas at higher T they mixed to become a single-phase supercritical fluid that was stable at P = 1.5–2.5 GPa. Experiments and direct studies of fluid inclusions in natural rocks conducted by others (e.g., Akaishi and Yamaoka 2000; Pal’yanov et al. 2002; Manning 2004; Ferrando et al. 2005; Frezzotti et al. 2007) also demonstrated that at P ≥ 4 GPa and T ≥ 800 °C there is no distinction between aqueous fluids and silicate melts because volatile-rich chemical systems are in a supercritical state.

A melt concept for Erzgebirge diamond crystallization was proposed by Massonne (2003) which was motivated by their studies of melting events in the bulk rocks of both Kokchetav and Erzgebirge diamond-bearing UHPM terranes which were constrained by zircon geochronology (Massonne et al. 2007a; Massonne and Fockenberg 2012). The cathodoluminescence images (Figs. 3 and 4 from Massonne et al. 2007a) clearly show that the cores of the zircons are typically eroded. The eroded zone of the zircon core is interpreted as being caused by a melting event. The second zone overgrew the eroded core and contains diamonds, while the third zone at the outer rim of the zircon grain does not contain any diamond inclusions. The authors report the following ages: metamorphic zircon core 337.0 ± 2.7 Ma (21 analyses); diamond-bearing zone of zircon 336.8 ± 2.8 Ma (23 analyses); and the outer rim zone of zircon 330.2 ± 5.8 Ma (12 analyses). Indeed, zircon ages combined with petrological observations could provide some constraints on the pressure–time history of these rocks, however there would be uncertaintnies in the pressure–time history that is derived from this approach. Moreover, the presence of an eroded core of zircon enveloped by diamond-bearing zircon zone does not provide any direct evidence that the diamonds originated from a melt; it can be interpreted as a metasomatic “erosion”. We believe that information about the origin of the diamonds is still “hidden” inside these diamonds since Massonne et al. (2007a) did not study their inclusions. Thus, the media which these diamonds crystallized from, whether it be a carbon-rich silicate melt, and/or a COH fluid or a multicomponent fluid is not well constrained. Additional studies using high resolution techniques (e.g., synchrotron IR spectroscopy and FIB-assisted STEM or TEM coupled with nanoSIMS) would provide better constrains on the melt vs. fluid origin of these unique microdiamonds enclosed in zoned zircons.

Korsakov et al. (2004) suggested that diamonds from the Kokchetav UHPM terrane should be classified as “intragranular” type A (diamonds included in garnet, clinopyroxene and zircon; e.g., high-pressure minerals) and “intergranular” type B (diamonds occurring within pseudomorphs after high-pressure minerals), and “intergranular” type C which occur within the granitic-like material with a quartz+feldspar composition. They observed that type C “intergranular” diamonds occur in high concentrations within the quartz–feldspathic material (e.g., they could be called “granitic melt injections” or “migmatitic material”) along the contact zone between the garnet–biotite–quartz and clinopyroxene–calcite layers. They proposed that these diamonds formed during the interaction of pelite–derived hydrous melt with a granitic composition with carbonate rocks during UHPM events. Korsakov et al. (2004) suggest that a granitic melt could act as a crystallization medium as well as a transport medium for type C “intergranular” diamonds. If one hypothesizes that three geologic events could have taken place, then three generations of diamonds could have formed, and they would have different ages, formation mechanisms, and media from which they crystallized.

To prove such a hypothesis, one should collect diamonds of each type along with: (i) geochronological data of diamond formation, (ii) type of inclusions preserved in diamonds (fluid, gas, solid matter) and their geochemical and crystal structure parameters, and (iii) morphology of diamonds. Until such data become available, the “three different type of diamonds” proposed by Korsakov et al. (2004) can more straightforwardly explained by one crystallization event that occurred during the same geological process (supercritical fluid/melt at the peak of the UHP metamorphism) which would imply that the diamonds all have the same age. Since diamond is chemically stable it may survive a retrograde metamorphic event, whereas other UHP minerals such as coesite, majorite, phengite, super Si-titanite, high Ca-Esk clinopyroxene and others can become re-equilibrated or replaced by their lower-pressure phases, e.g., pseudomorphs. Therefore, the same diamond (type A) which formed under peak UHP metamorphic conditions can be preserved within former UHPM minerals subjected to retrogressive metamorphism. However, in this case such a diamond would be classified type B according to Korsakov et al. (2004). If diamonds of type C are formed due to interaction between Si-rich and C-rich layers during their partial melting at UHPM conditions, such diamonds are expected to contain inclusions with chemistry that would support this origin hypothesis. Until such data are collected, the concepts of three types of diamonds (A, B and C) crystallizing during different UHPM events remain unproven.

Korsakov and Herman (2006) have also proposed that carbonate melt can be a medium where diamond crystallization occurred in the Kokchetav dolomite marbles. However, this hypothesis has been questioned by Imamura et al. (2013). They note that it is difficult to explain the extremely negative δ13C values by diamond crystallization from a carbonate melt, because strong isotopic fractionation is not expected between diamond and carbonate melt at high pressures and temperatures (Bottinga 1968; Reutsky et al. 2015).

Oxidation state during UHPM diamond crystallization. Another important aspect of diamond crystallization is the oxidation state of the diamond-forming media. Association of Erzgebirge polycrystalline diamonds with carbonate (CaCO3) inclusions suggests that they formed in an oxidizing environment close to the CCO buffer (Fig. 17) (e.g., Dobrzhinetskaya et al. 2013; Jacob et al. 2014). This is also supported by synchrotron infrared spectra of Erzgebirge diamond inclusions which show a well pronounced absorption band at 1430 cm−1, which corresponds to the CO32− radical (e.g., Dobrzhinetskaya et al. 2006b). Nanoinclusions of MgCO3 and CaCO3–aragonite were directly observed with TEM in diamonds from garnet–biotite gneiss and metacarbonate rocks of the Kokchetav massif (Dobrzhinetskaya et al. 2001, 2005, 2006a). This allows one to hypothesize that the Kokchetav diamonds also formed in an oxidizing environment close to the CCO buffer (Jacob et al. 2014).

Frezzotti et al. (2014) suggested that the Lago di Cignana (Western Alps) diamond nucleation began when the H2O-rich fluid reached the excess concentration of carbon required for spontaneous nucleation of diamond. They showed that at high pressures the interaction of rock buffered fO2 and the prograde P–T path may control the carbon saturation. The thermodynamic modeling (Frezzotti al. 2014) suggested that the diamond crystallized from COH fluids with a water concentration of about 0.992 < XH2O < 0.997, and that the fO2 was at the level required for EMOD (enstatite + magnesite = olivine + diamond) equilibrium (Fig. 17).

However, there are two cases where microdiamonds were found together with SiC as multiphase inclusions in zircon from Gföhl granulite samples of the Moldanubian Zone, Bohemian massif (Perraki and Faryad 2014) and in garnet from metapelitic gneisses of Pohorje, the Austro-Alpine UHPM terrane, Eastern Alps (Janák et al. 2015). Notably, Perraki and Faryad (2014) stated that they used SiC abrasives for one step of the polishing procedure, but to eliminate the possibility of contamination they removed the polish with an ultrasonic cleaner and polycrystalline-type diamond spray was used for the polishing of the thin section surface. Additionally, the occurrence of SiC and diamond grains beneath the polished surface was checked with both transmitted and reflected light microscopy.

Though Janák et al. (2015) hypothesized that diamond + moissanite inclusions in garnet from metasedimentary rocks of Pohorje were crystallized from a reduced COH fluid, they concluded that further detailed studies are needed to fully understand the oxidation state during their formation. Certainly, these unusual findings of diamond + SiC inclusions in garnets from UHPM terranes deserve additional close attention.

UHPM diamond formation under metastable conditions. Shortly after the discovery of diamonds in UHPM rocks, crystallization under metastable conditions was considered as an alternative to the classical concept of diamond formation in the Earth at P ~ 4.0–6.0 GPa and T ~ 900–1400 °C (Bundy 1980; Boyd et al. 1985).

Throughout the 1980s and into the mid of 1990s the concept of metastable diamond formation at low-pressure conditions of crustal metamorphism was proposed (e.g., Letnikov 1983; Lavrova 1991; Ekimova et al. 1992; Lavrova et al. 1995). Even at that time it was considered controversial to the new concept of diamond crystallization during UHP metamorphism of crustal rocks subducted to a depth ≥120 km (Sobolev and Shatsky 1988, 1990; Shatsky and Sobolev 1993; Shatsky et al. 1995). It has since been shown unambiguously that the Kokchetav massif is an UHPM terrane through the documentation of numerous UHPM mineral phases by hundreds of high-quality studies involving state-of-art and synchrotron assisted instruments (see reviews: Ernst and Liou 1999; Ogasawara 2005; Dobrzhinetskaya 2012; Schertl and Sobolev 2013; Liou et al. 2014). Despite all these convincing observations the concept of diamond crystallization outside of the diamond stability field has still occasionally appealed as a formation mechanism for diamonds found in UHPM terranes (e.g., Pechnikov and Kaminsky 2008, 2011; Simakov et al. 2008; Simakov 2010, 2015, 2018). Since this hypothesis has been invoked to describe the formation conditions of microdiamonds a short discussion about what is known from experiments on metastable diamond growth is warranted.

Pechnikov and Kaminsky (2008) focused their attention on the geometry of the “productive” diamond-bearing ore body using detailed mapping completed by the local Kokchetav Geological Survey, and they combined this information with their petrographic observations. They concluded that the diamonds were formed under crustal conditions (e.g., metastable) from crustal fluid–metasomatic processes that were active along the fault-zone superimposed on the Precambrian gneisses. They accepted the diamond bearing zircon age of 527–537 Ma (Claoué–Long et al. 1991; Jagoutz et al. 1990, 1991; Hermann et al. 2006) as the age of diamond formation and therefore the age of the fluid–metasomatic activities along the fault zone. The authors believe that the micrometer size of the diamonds, along with their isotopic signatures (δ13C = −8.9 ‰ through −27 ‰), support diamond formation under metastable crustal conditions. On the other hand, Pechnikov and Kaminsky (2008) note that their concept does not exclude the occurrence of episodes of UHP metamorphism and they accept the presence of coesite and other UHP minerals in the diamond-bearing rocks. However, their concept fails to explain when the UHP metamorphic episodes took place, and the authors did not discuss that fact that zircons containing diamonds and/or coesite, or pyroxene with high Ca-Eskola component and other UHP phases have the same age (e.g., Katayama et al. 2000, 2001a,b, 2002; Ogasawara 2005). Though the prevalent light carbon isotope characteristics of diamonds prove that such carbon has a crustal origin it does not reject the possibility of the subduction of crustal rocks to depths ≥ 120–150 km where the pressure and temperature are high enough to stabilize carbon in the form of diamond. Indeed, the arguments of Pechnikov and Kaminsky (2008) on the metastable origin of the Kokchetav diamonds are contrary to the conclusions reported by most other scientists who have collected and studied diamonds from this locality.

Simakov (2010) reports that they synthesized nano-diamonds from a water–alcohol solution which was the source of free carbon C, in a high-pressure reactor (volume ~500 cm3) at P = 0.1 GPa and T = 500 °C. They directly applied the results of this experiment to explain the origin of mm-sized diamonds from UHPM terrane of the Kokchetav massif without taking into consideration all of the other UHP minerals associated with these diamonds (e.g., Katayama et al. 2000, 2001b, 2002; Ogasawara et al. 2002; Ogasawara 2005).

Furthermore, Simakov (2015, 2018) also discussed the metastable formation of diamond from a COHN fluid and applied this idea to explain the origin of UHPM diamonds. They discussed that nitrogen interaction decreases the total energy of diamond nuclei formation and the equilibrated methane pressure stabilizes the diamond, and they confirmed this by the hydrothermal synthesis of nano-diamond from nitrogen bearing organics at metastable P–T conditions (Simakov et al. 2008). Synthesis of nano diamonds at high temperatures and/or ambient–low pressure (e.g., metastable) are well-known in materials sciences processes such as laser ablation, chemical vapor deposition (CVD), autoclave synthesis from supercritical fluids, ion irradiation of graphite and ultrasound cavitation (e.g., Krueger 2008; Derjaguin and Fedoseev 1994; Fedoseev et al. 1971) as well detonation techniques (e.g., Danilenko 2004), which requires extremely high pressures. However, before invoking any of the processes mentioned above to explain how diamonds form in UHPM terranes, would require a critical evaluation of all the existing data that have been collected by many research groups. Thermodynamic calculations for PT-conditions of diamond-bearing rocks and the presence of UHP minerals associated with diamond are important petrological evidence indicating that the diamonds from Kokchetav formed under UHPM conditions and their formation does not easily fit to the models/experiments proposed by Simakov (2010, 2015, 2018). Additionally, experiments conducted by other groups using a starting material that is close to the bulk chemistry of the Kokchetav diamond-bearing rocks, under both hydrous and anhydrous, and at varying oxidation states have produced run products that are a good fit to the data collected from the rocks, diamond itself and other associated minerals (e.g., Hong et al. 1999; Pal’yanov et al. 1999, 2000, 2013; Akaishi and Yamaoka 2000; Yamaoka et al. 2000; Kumar et al. 2001; Hwang et al. 2003; Dobrzhinetskaya et al. 2004; Dobrzhinetskaya and Green 2007a,b). Therefore, while the models/experiments proposed by Simakov (2010, 2015, 2018) do, indeed, explain diamond formation outside the diamond stability field, they cannot be applied to diamonds from the Kokchetav massif or for diamonds from other well established UHPM terranes.

It should be emphasized that if diamond formation under metastable conditions is proposed the following data are expected to be presented for consideration: (i) thorough geologic mapping of the area where the sample was collected to find any other UHPM indicator minerals other than diamond; (ii) unambiguous proof of the indigenous origin of the diamond, the diamond should be in situ so that the interface between the diamond and its host mineral can be examined in detail with high resolution techniques, and (iii) all available thermodynamic data has been reviewed.

Comet impact hypothesis for the origin of diamonds from Kumdy Kol Lake, Kokchetav Massif. Recently a new impact–cosmic–metasomatic genesis of the Kokchetav diamond deposit was proposed (Tretiakova and Lyukhin 2016, 2017). It was provoked by an impact event followed by metamorphism and alteration of the rocks. The authors note that the ring structure of Kumdy Kol Lake, where most of the diamond-bearing metamorphic rocks of the Kokchetav massif are situated, has the form and size (~4 km diameter) resembling a small impact crater. They suggest that the diamond-bearing terrane formed by comet impact under an oblique angle which generated shock wave compression with peak pressures reaching > 50 GPa.

They argue that there is evidence of impact metamorphism, (1) presence of UHPM minerals diamond–lonsdaleite, coesite, omphacite, (2) the delivery by comet moissanite (SiC) and graphite spherules as well as ”meteoritic matter” such as magnetite, hematite, iocite, troilite, ∝-Fe, Ni–Fe, (3) annealed metallic globules are observed having various fanciful forms (globules, small dump-bells, drops, spherules and so on) in host rock and rock-forming minerals, and (4) dislocation and birefringence in diamonds, planar structure in quartz, inclusions UHP minerals in rock forming minerals. The authors suggested that collision of huge velocity comet and the Earth generated rapid shock wave compression (pressure peak ˃ 50 GPa) forming diamond, coesite and other UHPM minerals which later together with so-called “meteoritic matter’ were subjected to regional metamorphism and metasomatism.

Additionally, they hypothesize that the “unprecedentedly” high 3He/4He isotopic ratios of the Kokchetav diamonds reported by Shukolyukov et al. (1993) are, indeed, due to the formation of these diamonds outside of the solar system. However, new measurements by Sumino et al. (2011) revealed that 3He/4He value of the Kokchetav diamonds are close to the values found for OIB, and that the extremely high 3He/4He ratio likely originated from Li-rich contaminants.

The “impact–cosmic–metasomatic” origin remains unproven based on the following evidence: (i) Sumino et al. (2011) have shown that the “unprecedentedly high” 3He/4He isotopic ratios are not likely a true feature of the Kokchetav diamonds, (ii) the authors have not presented unambiguous evidence of the claimed “oblique” impact (e.g., asymmetrical ejecta deposits), (iii) the authors do not propose that any of the other dozens of similar shaped lakes around Kumdy-Kol lake were caused by impact events, and (iv) and they do not report that the Kokchetav diamonds show unique moiré fringes which are well-known features observed in impact-formed diamonds (e.g., Langenhorst 2003; Ohfuji et al. 2015).

Ophiolite-hosted diamonds

Liou et al. (2014) provided a comprehensive review of the history of the discovery of ultra-high-pressure minerals in ophiolitic chromitites and peridotites. To avoid unnecessary repetition here, we focus our reader’s attention on the studies of diamonds and specifically on the methods used for confirmation of whether the diamond is indigenous or not. This topic is still highly controversial as is evidenced by several recent critical comments and replies on a few recent publications that have reported ophiolite-hosted diamonds. The crux of the matter is the indigenous nature of these reported diamonds. In this section we will discuss our vision and approach for how the community can reach a consensus about whether these diamonds are myth or reality.

To present ophiolite-hosted diamonds have been reported from more than a dozen geographic localities in the World (Fig. 18). Most of them have been found in the Tertiary Yarling-Zangbo suture zone of Tibet which formed during the India–Asia continental collision. The ophiolite complex situated along this suture zone is mainly composed of remnants of the Neo-Tethys ocean floor. Within them the Luobusa massif encompasses diamond-hosted chromitite and harzburgite (e.g., Fang and Bai 1981; Bai et al. 1993, 2001; Yang et al. 2007a), and the Purgan, Xigaze, Dongbo, Burang, Dangqiong, and Dongqiao massifs enclose diamond-hosted chromitites and peridotites (e.g., Yang et al. 2011, 2013, 2014; Lian and Yang 2019). Other worldwide ophiolite hosted diamond-bearing localities include podiform chromitites and/or peridotites in the Sartohai massif, NW China, Ray-Iz massif, Polar Ural, Russia (Yang et al. 2007b, 2015a); Hegenshan massif of Inner Mongolia, China (Huang et al. 2015); Mirdita massif, Albania (Wu et al. 2017, 2019; Xiong et al. 2017); Pasani-Karsanti massif, Turkey (Lian et al. 2017); Nidar ophiolite, India (Das et al. 2017); Myikyina massif, Myanmar (Chen et al. 2018), Tehuitzingo, Puebla State, southern Mexico where diamonds was documented in situ in chromitite-bearing serpentinites (Farré-de-Pablo et al. 2018), and diamond inclusions in magmatic olivine from ophiolite section of the Moa-Baracoa ophiolitic massif, eastern Cuba (Pujol-Solà et al. 2020).

Ophiolitic diamond formation hypotheses

Since the formation conditions for diamond and ophiolites are incompatible with each other several hypotheses have been put forward to explain their existence and some of them are very controversial (e.g., Arai 2013; Yang et al. 2014, 2015a,b; Zhou et al. 2014; McGowan et al. 2015; Robinson et al. 2015; Griffin et al. 2016; Rollinson 2016; Ballhaus et al. 2017; Dilek and Yang 2018; Lian and Yang 2019). We summarize them as three main hypotheses:

Localities of ophiolites-hosted diamonds formed at mantle depth

Diamonds from ophiolites of the Yarling-Zangbo suture zone, Tibet. The first diamonds were found in Tibet during the examination of the heavy mineral separates retrieved from mechanically crashed chromitites using optical microscopy (Fang and Bai 1981). The heavy minerals separation techniques have different protocols and procedures in different countries; however, all are used for the exploration of industrial mineral deposits, which in the case of chromitites are platinum group elements (PGE) metals and alloys. Fang and Bai (1981) reported that 20 diamonds were recovered from the Luobusa chromitite and hargburgite, and six diamonds from similar rocks collected from Donqiao massif of the Bangong-Nujiang ophiolite belt situated 500 km northwest from the Luobusa massif (Bai et al. 1989, 1993). These diamonds were characterized as colorless and transparent euhedral/octahedral-like morphology; their sizes fall in range of 100–500 µm, though some broken crystals of 200–250 µm and diamonds as large as 900–1000 µm were also reported (Bai et al. 1989). Other than diamonds, additional minerals which are “unexpected” to occur in chromitites, such as SiC, graphite, Ni–Fe alloys, and Cr+2- bearing chromium spinel were also documented in the heavy minerals separate.

These first reports of diamond findings in Tibetan chromite were eventually disproved by an international group of kimberlitic diamond experts who concluded that these diamonds “…are consistent with synthetic contamination introduced at some stage during the sampling or concentration of heavy minerals from these ultramafic bodies” (Taylor et al. 1995). After this report a new effort commenced to carefully process and separate heavy minerals from the Luobusa and other chromitite deposits of the Yarling-Zangbo suture zone, Tibet. The Institute of Geology, Chinese Academy of Geological Sciences, the Tibetan Geological Bureau, and international collaborators all participated in this effort. As a result, dozens of diamond crystals ranging from ~100–600 µm in size were documented again (Fig. 19) as well as more unusual new minerals including native Cr, Os–Ir–Ru alloys, Ir–Ni–Fe alloys, Pt–Fe alloys and others were recognized (e.g., see reviews Yang et al. 2007a,2014; Liou et al. 2014).

Additional explorations for ophiolitic diamonds in Tibetan and other worldwide ophiolitic terranes began; they used the same protocol, e.g., the collection of ~200 to 1000 kg of chromitites or peridotites, mechanical crushing and sieving of the rocks followed by heavy minerals separation including vibration, magnetic, heavy liquid floatation, and electrical conductivity techniques. Most Tibetan diamonds and diamonds from other ophiolitic belts were found in mineral separates processed in the Institute of Multipurpose Utilization of Mineral Resource, Zhengzhou, China. The diamonds retrieved from other Tibetan ophiolites (Purgan, Xigaze, Dongbo, Burang, Dangqiong, and Dongqiao massifs) share morphological similarities with those of Luobusa massif (Fig. 19), and they are always associated with of ultra-reduced minerals, metallic alloys and native elements (e.g., Liou et al. 2014; Yang et al. 2014; Xiong et al. 2017; Lian and Yang 2019).

Tibetan diamond in situ, unusual UHP minerals and highly reduced minerals and alloys. A small diamond of ~6 μm size was found as an inclusion in an Os–Ir alloy that was recovered from heavy mineral separates from Luobusa chromitite. This diamond was confirmed using Raman spectroscopy and the first order Raman mode was reported to be centered at 1332.2 cm–1 (Yang et al. 2007a). The position of this diamond in the thin section (polished without diamond abrasives) was observed with both optical microscopy and with SEM in secondary electron mode. Although the authors claim this satisfies the minimum requirements for the diamond to be considered in situ, the diamond was not observed below the surface of the host mineral before it was exposed to the surface by polishing. Alsothe samplewas not studied with FIB-TEM technique to confirm the nature of the Os–Ir-alloy–diamond interface and to find/analyze any inclusions that may be encapsulated inside this diamond (Yang et al. 2007a). Observation of the diamond below the surface of the host mineral and detailed analysis of the diamond host mineral interface with FIB-TEM would have unambiguously proved this diamond is indigenous. It has been suggested that the in situ Tibetan diamonds were likely formed at depths of 150–300 km (Yang et al. 2007a, 2014) or even deeper than 380 km (Xiong et al. 2018).

Detailed studies were conducted on another unusual recovered mineral fragment, an Fe–Ti alloy which was retrieved from the same mineral separates as the Os–Ir alloy containing diamond (Yang et al. 2007a). The fragment of Fe–Ti alloy had a small portion of a silicate rock which contained coesite after stishovite and kyanite suggesting that coesite after stishovite and diamond, might be incorporated into the chromitites in the deep upper mantle, or that they have an impact origin with a prevalence of evidence for the deep mantle origin (Yang et al. 2007a).

Later Dobrzhinetskaya et al. (2009,2014b) studies of the same sample using the state-of-the-art FIB-TEM analytical techniques showed that the coesite and kyanite contained mm-sized inclusions of native iron, cubic–BN (qingsongite), TiC (osbornite) and TiO2-II (rutile with α-PbS structure). The authors discussed two possible explanations that these data support: (i) the UHPM minerals have been encapsulated within chromitite at mantle depth, or (ii) they were formed by shock events due to a meteorite bombardment (Dobrzhinetskaya et al. 2009,2014b). Since the microstructures of UHPM minerals in the studied fragment are inconsistent with those formed due to shock metamorphism (e.g., Martini 1991; Stähle et al. 2008) the mantle origin of the fragment was deemed more plausible (Dobrzhinetskaya et al. 2009, 2014b). Dobrzhinetskaya et al. (2009, 2014b) have clearly informed the readers that the studied fragment of these metamorphic rocks, contained cubic-BN (quingsongite), TiC-osbornite, TiO2-II, coesite after stishovite and some high Si-Al-rich amorphous material, all recovered from these Tibetan chromitites by “crushed rocks–minerals separation” technique.

Although Tibetan chromitites contain more than a dozen different native elements, alloys, and minerals only a few of them may be classified as UHP minerals (diamond, coesite, TiO2 II). Most of the other reported minerals are highly reduced minerals (e.g., SiC-moissanite and FeO-wustite), native elements (e.g., Fe, Si, Ti, Cr), alloys (e.g., Os–Ir–Ru, Ni–Fe, Ir–Ni–Fe Pt–Fe), carbides [e.g., WC-qusongite, (Cr,Fe,Ni)9C4-yarlongite] and silicides (e.g., Fe5Si3-xifengite, TiFeSi2-zanboite) to mention a few (see reviews: Robinson et al. 2004; Liou et al. 2014; Lian and Yang 2019; Yang et al. 2019a). Importantly verification of their stability field at high pressures and high temperatures requires further experimental studies.

Diamonds in situ from Rai-Iz ophiolite, Polar Ural, Russia. New reports of ophiolitic diamonds and ultra-reduced minerals in other countries of Euro-Asian continent (Russia, Albania, Turkey, India, and Myanmar) confirm that they are widespread and that new concepts are required to explain their occurrences in ophiolite formations. Some interesting results came from recent studies of Nano-inclusions in ophiolitic diamonds from the Ray-Iz massif, which is an early Paleozoic ophiolitic belt in the Polar Urals, Russia. Yang et al. (2015a) recovered 1000 grains of loose diamonds (~200–500 µm in size) from heavy mineral separates processed from 1500 kg of Ray-Iz chromitite samples. They were processed at the same facilities as the Tibetan diamonds. To search for diamond in situ they cut 40 pieces of rock that were roughly 4 cm2 in area and encased them in epoxy. The samples were grinded down in 100 µm increments using a diamond-coated disk with a maximum grain size of 37 µm. Optical microscope images show that two diamonds remained enclosed in chromite (Fig. 20, plates A and C). On plate B of Fig. 20 (sample Y–5B–17–2) one can see a sub-euhedral diamond crystal that is ~ 300 µm in diameter surrounded by thin patch of amorphous carbon. Sample Y5B-16-2 (plate D) exhibits a well-shaped hexagonal platy-like diamond crystal with dimensions of ~300 × 400 µm surrounded by an irregular-shaped mass of amorphous carbon—all are inside of a host chromite (Yang et al. 2015a).

Though the authors concluded that both crystals are diamonds, they provided only one spectrum (e.g., plate E shows only one Raman spectrum with band at 1330.8 cm−1, see caption to Fig. 8, page 469, Yang et al. 2015a). Furthermore, the nature of the amorphous carbon remains unclear. The authors claim that it is amorphous based the absence of a Raman modes in the areas where elemental mapping showed that carbon was present (Yang et al. 2015a). However, even amorphous carbons with very little graphitic ordering show the presence of G- and D-like features in their Raman spectra (e.g., Ferrari and Robertson 2000). So, it remains a mystery as to why Yang et al. (2015a) did not observe any Raman peaks, they do not show these spectra and they do not report any instrument details from their Raman measurements (e.g., laser wavelength, spectrometer focal length, spectrometer resolution etc.). Moreover, presence of coesite inclusions in ophiolitic diamond is not an undisputable feature that can be used as a criterion to sort out if diamond is synthetic or indigenous. Milledge (1961) reported that they found coesite inclusions in one of the synthetic diamonds produced by the General Electric Company (G.E.C.) Laboratories in Schenectady county NY, USA. Therefore, careful characterization of sp3-carbon (diamond), sp2-carbon (graphite), and disordered or amorphous carbon should be provided in the future by Yang et al. (2015a) to prove unconditionally that the Ray-Iz diamonds are indigenous. They also need to show the differences between their so-called “amorphous carbon” and epoxy resin, which was used for this sample preparation.

Several loose diamonds were studied with FIB-TEM technique (Yang et al. 2015a,b). The results showed that the diamonds contain several ~2–10 µm inclusions of NiMnCo alloys, graphite and one inclusion of coesite. Further studies of inclusions in diamonds recovered from ophiolites are needed so that we can start to better understand the formation conditions of these diamonds.

Diamonds from the Skenderbeu chromitite massif of the Mirdita ophiolite, Albania.Wu et al. (2017) collected 600 kg of chromitite rocks from the Skenderbeu massif of the Mirdita ophiolite, Albania, the western part of Neo-Tethys ocean. After processing these samples at Institute of Multipurpose Utilization of Mineral Resources, Zhengzhou, China, the authors found 20 loose diamonds (100–200 µm in size) of octahedral, cubo-octahedral and irregular shapes (Fig. 21). Diamonds were confirmed with Raman spectroscopy and studied with TEM.

TEM studies of two FIB-prepared foils revealed that the Skenderbeu diamonds contain inclusions of NiMnCo alloy, traces of S and Cl, indicated escaped fluid, and Ca0.81Mn0.19 (SiO)3—which has a structure similar to wollastonite. They hypothesized that this phase may be a product of the decompression of a Ca-perovskite, although no undisputable data to support this interpretation was provided. The authors proposed that diamonds were formed in the deep mantle from a carbon-saturated NiMnCo-rich melt acquired from a subducted slab of oceanic crust and lithosphere (Wu et al. 2017, 2020).

Diamonds from Pozanti-Karsanti podiform chromitite, Turkey. The Pozanti-Karsanti ophilite complex occurs within the eastern Tauride belt, in southern Turkey, and consists of well-preserved sections of mantle peridotite, ultramafic and mafic cumulates, podiform chromitites, gabbros, sheeted dikes, and basalts (Dilek et al. 1999). Lian et al. (2017) reported that they recovered more than 100 grains of euhedral and irregular shaped diamonds (50–250 µm in size) from heavy minerals separates from the Pozanti-Karsanti podiform chromitite. The rocks were processed at the same facilities in Zhengzhou, China, as all other ophiolitic diamonds described above. The authors also found SiC and many “crustal” minerals such as zircon, rutile, hematite, magnetite, quartz, sulfides, chlorite, monazite, and also olivine, pyroxene, and an unusual octahedral silicate pseudomorph with a complex chemistry (Mg, Si, Al, Cr, Fe, O and Mg,Si,O). Lian et al. (2018a) showed that the diamonds contain nanometric inclusions of Ni–Mn–Co alloys and (Ca,Mn)SiO3 with triclinic symmetry similar to wollastonite. The authors have hypothesized that wollastonite may be a product of the decompression of (Ca,Mn)SiO3-perovskite, and concluded that the diamonds may have crystallized from carbon-rich fluids in the mantle at a depth of more than 250 km and that they were rapidly transported to a shallower depth by an asthenospheric melt (Lian et al. 2017,2018a,b). Interestingly their interpretation is similar to what was proposed by Wu et al. (2017) for Ca0.81Mn0.19(SiO)3 wollastonite inclusions in Skenderbeu diamonds from Albania.

Diamond in situ from Nadar ophiolites, India. The Nadar ophiolites are situated within the Indus suture zone, southeast Ladakh Himalaya, marking a collision zone between India and Asia. Das et al. (2017) discovered diamond in situ in a polished petrographic thin section of peridotite collected from the lower section of the Nadar ophiolites. Peridotite consists of olivine (50%), orthopyroxene (30%), clinopyroxene (10%), Cr-spinel (5%) and serpentine (5%). A diamond of ~50 µm in size is included in orthopyroxene; a Raman spectrum of the diamond exhibits a well-pronounced Raman band at 1326.7 cm−1 (characteristic of disordered sp3-carbon bonding) and broad feature centered at 1583.2 cm−1 (sp2 disordered graphite carbon bonding often referred to as the G-band). Some grains of olivine in this thin section contain numerous 2–5 µm size inclusions of graphite associated with fluid inclusions of C–H and H2. For those grains of olivine which exhibit well-oriented exsolution lamellae of Cr-spinel and blebs of α-Fe2O3, the β-Mg2SiO4 was inferred to be a precursor. Das et al. (2017) suggested that diamond crystallized from C–H fluids that were derived from the bottom of the upper mantle—top of the mantle transition zone (depth ~ 410 km). Although 1, 3 and 6 µm diamond grits were used for sample preparation, Das et al. (2017) deduced that their diamond inclusion size (~50 µm) exceeds the size of diamonds from the polishing source, and therefore the diamond inclusion in orthopyroxene from Nadar peridotite is indigenous.

Carbon isotopes, nitrogen content and nitrogen aggregation in UHPM ophiolitic diamonds

According to Xu et al. (2018) carbon isotope characteristics of ophiolitic diamonds from chromitites are comparable with a light carbon reservoir: δ13C = −29.0‰ to −16.9‰ (Luobusa chromitite) and δ13C = −28.7‰ to −18.3‰ (Ray-Iz, Polar Urals, Russia). The carbon isotope values of ophiolitic diamonds bear similar features as metamorphic diamonds from UHPM terranes, and they are significantly different from cratonic diamonds (Fig. 7). Nitrogen content in Luobusa diamonds varies from 152 to 428 ppm, whereas in Ray-Iz diamonds it varies from 108 to 499 ppm, and both are significantly lower than the nitrogen content reported in diamonds from UHPM terranes and cratonic settings (see Table 2 for comparison). Nitrogen isotope characteristics for Tibetan ophiolite-hosted diamonds ranges from δ15N = −28.4‰ to −18.8‰ (e.g., Xu et al. 2018; Lian et al. (2018a) studied carbon and nitrogen isotope values in diamonds from Posanti-Karsanti chromitites, Turkey. The Pozanti-Karzanti diamonds are characterized by δ13C = −28.4‰ to −18.8‰ and δ15N values ranging from −19.1‰ to 16.6‰ with nitrogen contents ranging from 7 to 541 ppm. All these parameters are similar to those from diamonds of Loubasa and Ray-Iz ophiolites (Table 2) and their strongly 13C-depleted more than likely originated from a subducted crustal carbon-reservoir.

Howell et al. (2015) studied nitrogen aggregation in Luobasa diamonds. They show that these diamonds contain nitrogen impurities in the form of singly substituted C centers (Type Ib diamond). According to Taylor et al. (1996) natural diamonds with similar nitrogen characteristics have a short residence time and they are rare in nature because isolated nitrogen aggregates to form nitrogen pairs (A centers–Type IaA diamond) and more aggregated nitrogen (B centers–Type IaB diamond). Type IaAB is most typical for cratonic diamonds which have long-term (~0.5 to 3 billion year) upper mantle residence, whereas Type Ib is a nitrogen aggregation characteristic, which is typical for synthetic diamond (Taylor et al. 1996). Thus, if ophiolitic diamonds are indigenous, their low aggregation state (Ib) would reflect a short residence time (millions of years’ time scale) at temperatures < 900 °C (e.g., Xu et al. 2018). However, a literature survey shows that Ib diamonds were found within their cratonic populations (e.g., Hainschwang et al. 2013; Smit et al. 2016; Smith et al. 2016; Timmerman et al. 2018). Therefore, the nitrogen aggregation parameters do not allow us to unambiguously classify ophiolitic diamonds as either synthetic or indigenous.

Litasov et al. (2019a,b) have suggested that all ophiolitic diamonds are due to anthropogenic contamination. The main arguments are that the composition of the metallic inclusions in type Ib cuboctahedral diamonds in ophiolitic chromitites and peridotites is similar to the composition (Ni:Mn:Co = 70:25:5) of the catalyst used in industrial diamond synthesis. Yang et al. (2020) commented on Litasov et al. (2019b) stating that they have ignored the petrologic context related to microdiamond occurrences and inclusion assemblages discovered in ophiolite-hosted diamonds. Yang et al. (2020) also claim that they ignored the wide occurrence of ophiolite-hosted diamonds discovered in different locations by different research groups. Litasov et al. (2020) replied to this comment and made a simple thermodynamic and kinetic argument against the co-existence of diamond, moissanite, native elements, corundum, wadsleyite, coesite, and quisongite with major ophiolite minerals, they re-emphasized that the metallic inclusions have a similar composition to the catalysts used in industrial diamond synthesis, and they argue that the in situ findings discussed by Yang et al. (2019a) do not meet the basic requirements for in situ mineral grains (they were not observed below the surface before being exposed).

Yang et al. (2019a) also emphasized that the diamonds from ophiolites fall within the size ranging from 100–300 and ~500 µm, and they, therefore, are much larger than those which are used in typical laboratories as an abrasive material. Moreover, they asserted that the ophiolitic diamonds have well-developed crystal faces, whereas any other diamonds from drill bits, or stone-cutting saws would be presented by broken fragments of the crystals.

Shallow diamonds from ophiolites

Healed-crack diamond inclusions in chromite from altered ophiolitic chromitite pods in Tehuitzingo serpentinite (Southern Mexico). Farré-de-Pablo et al. (2019) report in situ diamond inclusions in chromite from ophiolitic chromitite pods hosted in the Tehuitzingo serpentinite, southern Mexico. They observed diamonds of 1 to 8 µm in size, quartz, clinochlore, serpentine, and amorphous carbon appeared as “healing-crack inclusions” in chromite. The authors concluded that diamond was formed at a shallow depth at T = 515–670 °C from reduced COH fluids, which infiltrated in during the shallow hydration of chromitite pods enveloped by serpentinite.

Yang et al. (2019b) and Massonne (2019) submitted comments on the Farré-de-Pablo et al. (2018) report raising different concerns. Yang et al. (2019b) specifically discuss that in suggesting a shallow origin for ophiolite-hosted diamonds Farré-de-Pablo et al. (2019) ignore the presence of many other UHPM phases occurring as both inclusions in diamonds and as associated grains. Yang et al. (2019b) propose an alternative scenario in which the diamonds, along with a carbon-rich fluid and some silicate grains were incorporated into the chromite grain at the time it crystallized in the diamond stability field thus sealing the fracture. Massonne (2019) raised concerns that these diamonds are not indigenous, and they could be considered contamination from abrasives or diamond-bearing tools. Massonne (2019) point out that only show diamonds that have been exposed to the surface, and that the five exposed grains are clearly xenomorphic.

In their reply Farré-de-Pablo et al. (2019) address the concerns of both Yang et al. (2019b) and Massonne (2019). They first address the concerns of Massonne (2019) and stress that they provided many lines of evidence supporting the presence of natural diamonds in the Tehuitzingo chromitites. They discuss that diamonds formed from COH fluids may also exhibit irregular morphologies and they cite Dobrzhinetskaya et al. (2004). In their reply to Yang et al. (2019b) they point out that UHPM minerals other than diamond were not found in the Tehuitzingo chromitites as evidence that the diamonds formed under low P and T conditions. The reply from Farré-de-Pablo et al. (2019) has not unambiguously proved the indigenous origin of these diamonds nor has it clarified the crystallization conditions of these diamonds.

Notably, the submission of comments on published papers and replies to the comments indicates that articles are being read by the community and that the topics are controversial. Moreover, they facilitate discussion not only between the authors but also the readers. These controversies should be considered positive since they are one way that science moves forward.

From our standpoint, finding diamonds as healing–crack inclusions in chromite (Farré-de-Pablo et al. 2018) is extremely interesting, and it may expand our understanding of microdiamond formation. However, the most important information that would unambiguously prove that the Tehuitzingo diamonds are indigenous is still missing. The authors presented FIB-TEM images accompanied with EELS spectroscopy data confirming both sp3 carbon bonding (diamond) and amorphous carbon. Farré-de-Pablo et al. (2018, 2019) emphasized that the Tehuitzingo diamonds exhibit “cohesive” contacts with the host minerals and particularly with enveloping amorphous carbon which contains hollows interpreted as evidence of the escaped fluid. We scrutinized their FIB-TEM data (Fig. 2 in Farré-de-Pablo et al. 2019, and Supplemental Materials) and strongly believe that higher quality HRTEM images of diamond–amorphous carbon and diamond–chromite interfaces are necessary to prove or disprove the indigenous origin of this diamond. The authors should continue to do TEM studies (maybe with additional FIB milling) on the same foil to acquire HRTEM images accompanied by selected area electron diffraction (SAED) patterns of each phase. These data would provide a better understanding of the relationships between all solid phases (lattice fringes, d-spacings) and traces of fluid (e.g., as an example see Fig. 6 d in O’Bannon et al. 2020).

Diamonds inclusion in magmatic olivine from gabbro and chromitite of the Moa-Baracoa ophiolitic massif, eastern Cuba.Pujol-Solà et al. (2020) report that they found in situ diamond hosted by magmatic olivine occurring in gabbro and chromitite of an upper mantle section of ophiolite. The authors described 200–300 nm sized diamonds associated with low-temperature and low-pressure minerals such as lizardite, chrysotile and magnetite and they are hosted by trails of the healed cracks and CH4-rich fluid inclusions inside of olivine. The authors concluded that the observed diamonds were formed at a shallow depth from reduced carbon-rich fluid at P < 200 MPa and T < 350 °C during serpentinization of olivine in gabbro and chromitite. They also hypothesized that the presence of super-reduced phases together with metastable diamond is clear evidence of a widespread infiltration process of CH4-rich fluids in the oceanic lithosphere, and therefore the ultra-high-pressure origin for other ophiolitic diamonds can be dubious (Pujol-Solà et al. 2020).

Since these samples were prepared with abrasive diamond material, we believe that there would be fewer concerns of contamination if they provided more detailed data of the diamond–host minerals interface. Special attention should be focused on Fig. 2a,b,c of Pujol-Solà et al. (2020) where ~400 nm diamond inclusion is enveloped by “soft” serpentine material; the latter is situated between olivine and magnetite. The diamond–serpentine interface shown in Figure 2c of Pujol-Solà et al. (2020) does not provide clear information about the diamond–serpentine interface. HRTEM images are needed to observe lattice fringes of the diamond inclusion and its host serpentine. HRTEM images would be helpful to investigate the material between the diamond and serpentine which looks like a darker line surrounding the diamond (see their Fig. 2 c). It also would be helpful to inspect both the diamond and serpentine for any nanometric fluid or solid inclusions. The SAED analysis (Fig. 2c) and EELS spectrum (Fig. 2e) shown by Pujol-Solà et al. (2020) do confirm the presence of diamond, however, they do not prove or disprove that this diamond is indigenous.

Ophiolite-hosted diamonds and super-reduced minerals and alloys formed from natural lightning strikes

Ballhaus et al. (2017) proposed a new concept for ophiolitic diamond and associated ultra-reduced minerals formation due to natural lightning strikes. They conducted laboratory experiments in which basalt rock was subjected to electrical discharge (T~5700 °C) to simulate a natural cloud-to-ground lightning strike (Uman 1986). Ballhaus et al. (2017) report that the high-electrical energy discharge into basalt created a plasma followed by melting and rapid crystallization of Fe–Si alloys, W–Ti alloys, metallic Ti, SiC, amorphous carbon and spherulitic ejecta. Though no diamond was produced by these experiments, the authors believe that they produced potential precursors to diamond. Ballhaus et al. (2017) hypothesized that diamond could nucleate in the cores of fullerenes, when they contract during cooling similar to that described in earlier experiments reported by Banhart and Ayajan (1996). Furthermore, Ballhaus et al. (2017) explicated that coesite and stishovite could also be formed during discharging experiments, though these minerals were not found in the run products of their experiment. Ballhaus et al. (2017) concluded that all existing models of ophiolites formation should be scrutinized because the diamond (and some ultra-reduced phases) have dubious origin.

The lightning origin of diamonds and associated ultra-reduced minerals, alloys and metals hosted by ophiolites has been met with criticism. Griffin et al. (2018) disagreed with Ballhaus et al. (2017) hypothesis because diamonds from ophiolites exhibit characteristics which are not comparable with the environment created in experiments by the high-electrical energy discharge (equivalent of natural lightning).

Yang et al. (2018) argued that Ballhaus et al. (2017) experiments have never produced diamond with solid inclusions, whereas diamonds from ophiolites contain wide variety of UHPM and highly reduced mineral inclusions of SiC, NiMnCo and other alloys, Mn-rich olivine, garnet, Mn-oxides, chromite, coesite and fluids (Yang et al. 2015b; Wu et al. 2017; Moe et al. 2018). Additionally, Yang et al. (2018) asserted that the lightning process cannot produce so many diamonds in such geographically different places (China, Turkey, Russia, India, Albania to mention a few), and in the rocks which have been situated at the mantle depth during diamond crystallization.

Volcanic diamonds from Kamchatka, Russian Federation

A sensational discovery of microdiamonds occurring in lava and ash of the modern Tolbachik volcanic eruption of 2012–2013 (Kamchatka peninsula, Russian Federation) was reported by Karpov et al. (2014) and Gordeev et al. (2014). Later Galimov et al. (2016) proposed that carbon dispersed during the eruption could be the perfect source of carbon for these diamonds. They suggested “…that Tolbachik microdiamonds were formed as a result of cavitation during the rapid movement of volcanic fluid” (Galimov et al. 2016). According to this concept the ultra-high-pressure required for diamond formation is concentrated locally, e.g., inside of the carbon-rich fluid bubbles expelled by the volcano, while the external pressure remains not very high.

Further investigation of Tolbachik diamonds lead to conclusions that their geochemical and mineralogical data strongly support their origin from gases released during volcanic explosion (e.g., Galimov et al. 2020a; Galimov and Kaminsky 2021). They reported that Tolbachik diamonds are characterized by δ13C = −22 to −29‰, which are comparable with the δ13C values of kimberlitic diamonds, and their δ15N = −2.58 and −2.32‰ are similar to δ15N of volcanic gases, while atmospheric nitrogen has δ15N = 0‰ (Galimov et al. 2020a, b). The nanometric inclusions of Tolbachik diamonds are presented by Mn–Ni–Si alloys and silicides with compositions ranging from (Mn,Ni)4Si to (Mn,Ni)5Si2, Mn5Si2, and MnSi. These features resemble those in diamonds found in situ in ophiolitic chromitite and peridotite (e.g., Galimov et al. 2020a,b).

The Tolbachik diamonds existence as a product of a volcanic explosion was met with extreme criticism and suggestions that those diamonds and accompanied ultra-reduced minerals, alloys and metals are products of anthropogenic contamination (Litasov et al. 2019a,b, 2020; Pokhilenko et al. 2019). There is a series of Comments and Replies publications between those who considered Tolbachik diamonds (as well as their ophiolitic counterparts) as natural Earth materials (Galimov et al. 2020a,b; Galimov and Kaminsky 2021), and those who insisted that all of them are a result of anthropogenic contamination (e.g., Litasov et al. 2019a,b, 2020, 2021; Pokhilenko et al. 2019). It is clear from reading of these comments and replies that each group believes that their arguments are correct, and therefore they deny any disagreements.

Are diamonds from ophiolite and modern volcanic eruption myth or reality?

The narrative of discovery of both ophiolitic and volcanic diamonds to some degree reminds us of the story of ultra-high-pressure metamorphism. This was best described by Harry W. Green II as “a confused mixture of surprising, sometimes spectacular, discoveries and emotional reactions” (Green 2005, p.439). Green (2005) emphasized that science advances in a conservative fashion in a process where accurate ideas are added to the existing “library” of knowledge and inaccurate ideas are not. He also stressed that new ideas which attempt to shift the paradigm will be met with criticism proportional to how large a shift is needed (Green 2005). The example most relevant to the recent controversial reports of ophiolitic and volcanic diamonds, is the story of the discovery of microdiamonds in metamorphic rocks of continental affinities described at the beginning of this paper. First diamonds from Kokchetav massif, Kazakhstan became known to local geologists due to processing large volumes of the garnet–biotite gneisses using industrial scale methods for minerals separation (Rozen 1972; Rozen et al. 1979 and references therein). After Sobolev and Shatsky (1990) clearly demonstrated that diamonds from Kokchetav were indeed indigenous, Massonne (1999) reported the discovery of similar diamonds in metamorphic gneisses of Erzgebirge, Germany. This discovery was met with harsh criticism and denied at the 1999 Kimberlitic Conference (Green 2005), fortunately additional investigations were published soon after the conference supporting this new diamond find (Massonne 1999; Nasdala and Massonne 2000; Green 2005).

In general, the disbelief that diamond and other UHPM minerals could occur naturally in crustal metamorphic rocks was overcome by a step-by-step process which eventually disregarded the inaccurate observations and deductions. The process included major efforts of many scientific groups who organized an uncountable number of seminars and scientific schools associated with International meetings, where participants had the opportunity to see these “mysterious” diamonds in situ for themselves with the aid of an optical microscope. In parallel to these efforts more advanced analytical techniques were also being developed and presented to the scientific community. The resolution of SEM and FIB-TEM and the flux and brightness of synchrotron facilities continued to improve. As a result, there was a shift in the paradigm and diamond-bearing UHPM terrains are now well-documented and accepted by the community.

The main argument underlying the disbelief in ophiolitic and volcanic diamonds is that they have a lot of common characteristics with their synthetic counterparts which are widely used in industrial machines and tools. The increasing concerns arise because almost all worldwide ophiolite-hosted diamonds were separated at the same mineral processing facilities at the Institute of Multipurpose Utilization of Mineral Resources, Chinese Academy of Geological Sciences, Zhengzhou (e.g., Xu et al. 2009). There is, however, one encouraging report of an ophiolitic diamond that was extracted from the Luobasa chromitite by an individually developed method of minerals separation which excluded any possible anthropogenic/laboratory contamination (Howell et al. 2015; Griffin et al. 2016). This provides us evidence that some ophiolitic diamonds are indigenous.

We strongly believe that the indigenous origin of ophiolite-hosted diamonds can be unconditionally proven if diamond abrasives/tools are eliminated from every step of the samples preparation from extraction to polishing stages. The observations of the same diamond in a polished thin section surface should be conducted using a combination of techniques such as optical microscopy, Raman spectroscopy, and SEM, FIB-TEM confirmation of diamond–host mineral interface. The same approach should be applied to studies of volcanic diamonds. The critical samples should be shared between different research groups so that a wide range of analytical techniques can be utilized to study origin of these diamonds.

Once anthropogenic contamination can be ruled out, the community can move past this question and focus on more important aspects of these findings. This would include searching for natural processes that would be similar to the industrial diamond synthesis processes, as well as the formulation of new models that can adequately explain diamond growth in ophiolites and processes that can transport diamonds from elsewhere to the shallow mantle beneath spreading centers.

General precautions and examples

Since many diamonds from UHPM terranes as well as from ophiolites and volcanic sources are small ranging in size from ≤ 3–50 µm they can be misidentified with diamonds which were mechanically embedded into softer minerals during sample preparation (e.g., cutting, drilling, and/or polishing with diamond tools or abrasives). Recognition of possible contamination has been mentioned in the scientific literature as well as in numerous instruction manuals attached to polishing instruments. Microdiamond researchers are acutely aware that contamination from sample preparation can occur, and that distinguishing indigenous microdiamonds from diamonds embedded during sample preparation is not trivial.

Figure 22AF contains a series of images showing the morphology of diamond particles found in a typical abrasive powder and those embedded in diamond saws. Figure 22GH shows “loose” UHPM diamond crystals recovered from crushed rocks. If a diamond-bearing sample is properly polished (see detailed procedure in Dobrzhinetskaya et al. 2001), the diamonds will be elevated above the polished surface of host minerals (e.g., zircon and garnet) which is due to the difference in hardness between the diamond and host minerals (Fig. 23). Visual inspection of the images, especially those acquired with an optical microscope or SEM using low magnification, reveal that the morphology of diamonds from abrasive grits and natural diamonds are quite similar (Fig. 22). Therefore, the morphologies of the diamonds cannot always be used to distinguish between natural microdiamonds and diamonds from abrasive grits.

Raman spectroscopy: indigenous diamond vs. its synthetic counterpart

One should be cautious when using Raman spectroscopy as the sole diagnostic for identifying indigenous diamonds. Diamond has a strong first order Raman mode at 1332 cm−1, some abrasives show broadening of this mode as well as a strong shift of this mode to lower wavenumbers. However, Nasdala et al. (2016) show that the Raman parameters of diamond abrasives (collected from 11 different manufacturers) widely overlap with UHPM microdiamonds (Fig. 24AC, E, F) making it nearly impossible to distinguish between a genuine UHPM relict or an introduced artefact based on Raman spectroscopy alone.

The study of Menneken et al. (2007) is a prominent example of diamond contamination during sample polishing. They reported the astonishing discovery of the oldest microdiamonds on Earth as inclusions in zircons from the Jack Hills conglomerate, Australia (~ 4.2–3.0 Ga, Nemchin et al. 2008). They used Raman spectroscopy as their primary diagnostic technique, particularly they relied on the FWHM of the diamond Raman mode to distinguish their diamonds from the diamond-based abrasive used for their sample preparation. Indeed, in Fig. 2 from Menneken et al. (2007) the plots of the FWHM vs. Raman shift show that the Jack Hills microdiamonds have larger FWHM’s and are shifted to lower frequency than the abrasive diamonds used during their sample preparation. In addition to the Raman results Menneken et al. (2007) ruled out contamination from the diamond polishing powder based on (i) some of the diamonds are completely enclosed by zircon, and (ii) they argued that the size of the diamond inclusions are mostly larger than the size of the diamonds in the polishing materials.

However, several unreconciled discrepancies existed. One is that it does not matter if the zircons are considered as magmatic (Hopkins et al. 2008, 2010, 2012) or as metamorphic (Rasmussen et al. 2010, 2011, 2012) the oldest Jack Hills zircons have a shallow (<10 km) crustal origin. If diamond crystallized along with the zircon at these shallow depths, the diamonds would have formed outside the diamond stability field. Alternatively, the diamonds could have formed within the diamond stability field at a depth of ≥150 km and they were later transported to these shallow depths where they became incorporated into the zircons when they crystallized.

Another discrepancy is that other high-pressure minerals or evidence of their decompression have not been reported in the Jack Hills zircons, and that boundary between the diamond included in zircon is highly irregular and sharp, and even porous and filled with granular “debris” (Fig. 25A). Such “unusual” boundaries between two detached phases (diamond and zircon) look significantly different from the boundaries between an indigenous diamond and its host zircon. The contact zone in many samples generally shows that the diamond is intricately intergrown with its host zircon, which immediately makes it clear that such a diamond is indigenous (Fig. 25B and C).

Considering these inconsistences, Dobrzhinetskaya et al. (2014a) examined the exact same diamond bearing zircon samples which were kindly provided by Menneken et al. (2007) and Nemchin et al. (2008). The SEM and TEM studies of FIB-prepared foils (Fig. 26 A,B,C) concluded that the “Jack Hills diamonds” are not indigenous, and that they are a product of contamination from the industrial diamond abrasive used for sample preparation (Dobrzhinetskaya et al. 2014a). Dobrzhinetskaya et al. (2014a) clearly demonstrated that fragments of broken diamond’s ranging in size from ≤0.10–0.30 µm were mechanically embedded/inserted into surficial pores and cracks (Fig. 26 A), and they eroded softer minerals together with polishing dust and petropoxy resin all which occurred during sample preparation (Fig. 26 B,C).

Later, Menneken et al. (2014) used TEM and Raman spectroscopy on FIB cut sections of the embedded diamonds in their samples and they concluded that the embedded diamond particles originated from the diamond polishing powder. This study is evidence that diamonds that contaminated the sample during cutting and polishing can be misidentified as indigenous UHPM diamonds. In the case of the Jack Hills diamonds if this misidentification was not brought to the scientific forum, it could have had a profound impact on our understanding of the earliest phase of Earth’s history. Since it is now known that the “Jack Hills diamonds” are not indigenous (Dobrzhinetskaya et al. 2014a; Menneken et al. 2014, 2017) several new thorough systematic studies of synthetic diamonds from different manufacturers and microdiamonds from well-established UHPM terranes (Fig. 24) have been reported (Steger et al. 2013; Nasdala et al. 2016). Raman data from Jack Hills diamonds reported by Menneken et al. (2007) (Fig. 24 D) fit well with the Raman data from diamond-based abrasives and tools (Fig. 24AC), and they partly overlap with Raman data collected from UHPM diamonds (Fig. 24 E,F). As discussed above, Nasdala et al. (2016) demonstrated that Raman spectroscopy can be used as fast and nondestructive diagnostic method for diamond identification. However, they emphasize that Raman spectroscopy cannot unambiguously differentiate between natural diamonds and industrial diamond–abrasives. To avoid erroneous discoveries of new microdiamond localities one should avoid using Raman spectroscopic techniques as the only diagnostic for confirmation of the indigenous origin of the diamonds.

In an earlier chapter we discussed that microdiamond occurring together with SiC, moissanite (Perraki and Farayd 2014; Janák et al. 2015) would require ultra-reduced conditions. These samples are very important for future investigations since they imply a specific oxidation state of the diamond–moissanite inclusions in garnet. However, additional documentations are required that these SiC inclusions have indigenous origin. Nasdala et al. (2016) also note that the Raman spectra of geological moissanite and the Raman spectra of synthetic silicon carbide are not distinguishable (see their Fig. 4d). Consequently, the presence of moissanite in multiphase inclusions from UHPM assemblages should be confirmed in order to exclude any misinterpretation of contamination from SiC-based abrasives used for polishing materials.

How to prove that microdiamond is indigenous

In general, we believe that it should be a mandatory for acceptance of any new discovery of non-cratonic microdiamonds to demonstrate the nature of the diamond–host minerals interface to avoid misidentification with remnants of diamond-based polishing abrasives. In some cases, studies of the diamond–host mineral relationships with optical microscopy and their documentation with high resolution SE images from an SEM can be enough to confirm the indigenous nature of a microdiamond (e.g., Dobrzhinetskaya et al. 2001). For example, Figures 25 B and C present the secondary electron SEM images which demonstrate the intergrown features of zircon–diamond interface. Such relationships unconditionally confirm indigenous origin of the diamond inclusion in zircon. In other cases, it is necessary to use a TEM to investigate FIB-prepared foils that were cut through the diamond–host mineral interface (Fig. 26) (e.g., Dobrzhinetskaya et al. 2001, 2003, 2005, 2006a). As an example, see Figures 26 and 27 which demonstrate FIB-milled sample of the diamond–zircon interface. FIB-TEM techniques are the safest approach for avoiding the misidentification of industrial diamonds–abrasive as natural diamonds, especially if the diamond inclusions under question are smaller than 5 mm (e.g., Dobrzhinetskaya et al. 2003, 2006a).

Notably one robust approach for avoiding the misidentification of diamond and moissanite inclusions would be to not use diamond and SiC tools and abrasives during sample extraction and preparation, one could use corundum instead. Final finishing of the samples with colloidal SiO2 liquid always provides a great polishing effect and makes diamond inclusions more visible on the flat polished surface of the host minerals.

Studies of UHPM microdiamonds have furthered our knowledge and understanding of geochemical and mineralogical processes during the deep subduction of crustal material and its interaction with the Earth’s mantle. Despite their mm-size these diamonds have preserved solid, liquid, and gas inclusions over long time periods. These inclusions reflect pathways of the crustal rocks and organic carbon from the Earth’s surface to its interior and back to the Earth’s surface. The next decade of in situ microdiamond research may be focussed on the following:

  • Confirmation of the indigenous nature of microdiamonds in newly discovered UHPM, ophiolites and volcanic localities requires the documentation of the diamond–host mineral interface with the aid of the high-resolution secondary electron SEM imaging, and/or TEM imaging and crystal structure data obtained from FIB-prepared samples.

  • Studies of nanoscale inclusions in microdiamonds from different lithologies to improve our understanding of geochemical environments and PT conditions of diamond crystallization.

  • Raman studies of anomalies in microdiamonds which are interpreted as evidence of the “lonsdaleite structure” need to be investigated with FIB-TEM, and/or synchrotron assisted X-ray and spectroscopic techniques.

  • Studies of stable isotopes chemistry and noble gases from additional UHPM, ophiolite and volcanic diamond localities to increase our knowledge of processes that occur during continental crust–mantle interaction.

  • Coordination of laboratory-assisted studies on microdiamonds with general questions of Earth’s dynamics through studies of UHPM minerals assemblages associated with diamond formations.

  • Modeling and calculations of subduction and exhumation rates; and understanding of how much crustal materials have reached the “point-of-no-return” during deep subduction and remained in the Earth’s deep interior being intermixed with the mantle.

LD research was performed as a part of the International Lithosphere Program, Task Force IV, project “Fate of the subducted continental lithosphere: insight through analytical mineralogy and microstructures”, and were supported in different years by NSF grants EAR 0229666, EAR 0107118 and INT–EAR 0329596; Los Alamos National Laboratory through Science and Technology Base Programs (grant 9949); the Consortium for Materials Properties Research in Earth Sciences (COMPRES); Pacific Rim Program of the University of California, and the University of California Lab Fee Research Grant. EFO acknowledges that a portion of this work was performed under the auspices of the US Department of Energy by Lawrence Livermore National Laboratory under Contract No. DE–AC52–07NA27344. HS research was supported by KAKENHI Grant Numbers JP20740314, JP23340169, JP24109702, JP26287139, and JP15KK0150 from MEXT/JSPS, Japan. Helpful discussions of several aspects of UHPM microdiamonds origin with C.V. Stan, M. Janák, J. Kotková and K. Naemura are acknowledged. Fruitful advices from the peer–reviewers H–P. Shertl, T. Tsujimori, S.W. Faryad and G. Pearson and editors K.V. Smit and S.B. Shirey of this special volume are greatly appreciated.

Open access: Article available to all readers online.