The Newly Discovered Neoproterozoic Aillikite Occurrence in Vinoren (Southern Norway): Age, Geodynamic Position and Mineralogical Evidence of Diamond-Bearing Mantle Source

: During the period 750–600 Ma ago, prior to the final break-up of the supercontinent Rodinia, the crust of both the North American Craton and Baltica was intruded by significant amounts of rift-related magmas originating from the mantle. In the Proterozoic crust of Southern Norway, the 580 Ma old Fen carbonatite-ultramafic complex is a representative of this type of rocks. In this paper, we report the occurrence of an ultramafic lamprophyre dyke which possibly is linked to the Fen complex, although 40 Ar/ 39 Ar data from phenocrystic phlogopite from the dyke gave an age of 686 ± 9 Ma. The lamprophyre dyke was recently discovered in one of the Kongsberg silver mines at Vinoren, Norway. Whole rock geochemistry, geochronological and mineralogical data from the ultramafic lamprophyre dyke are presented aiming to elucidate its origin and possible geodynamic setting. From the whole-rock composition of the Vinoren dyke, the rock could be recognized as transitional between carbonatite and kimberlite-II (orangeite). From its diagnostic mineralogy, the rock is classified as aillikite. The compositions and xenocrystic nature of several of the major and accessory minerals from the Vinoren aillikite are characteristic for diamondiferous rocks (kimberlites/lamproites/UML): Phlogopite with kinoshitalite-rich rims, chromite-spinel-ulvöspinel series, Mg- and Mn-rich ilmenites, rutile and lucasite-(Ce). We suggest that the aillikite melt formed during partial melting of a MARID (mica-amphibole-rutile-ilmenite-diopside)-like source under CO 2 fluxing. The pre-rifting geodynamic setting of the Vinoren aillikite before the Rodinia supercontinent breakup suggests a relatively thick SCLM (Subcontinental Lithospheric Mantle) during this stage and might indicate a diamond-bearing source for the parental melt. This is in contrast to the about 100 Ma younger Fen complex, which were derived from a thin SCLM.


Introduction
Although ultramafic lamprophyres (UML) are volumetrically insignificant rocks, they may play a crucial role in the understanding of deep (mantle) melting events. UML form dyke swarms and rarely pipes commonly associated with continental extension, commencing during the initial stages of continental rifts evolution. UML often occurs together with alkaline mafic-ultramafic and carbonatitic intrusive complexes [1]. UML are classified as melanocratic rocks with abundant olivine and phlogopite macrocrysts and/or phenocrysts and can be subdivided into three rock types depending on a third essential mineral [2]. (1) Alnöits are melilite-bearing UML; (2) aillikites contain primary carbonate; and (3) damtjernites are nepheline-and/or alkali feldspar-bearing. Clinopyroxene and/or richteritic amphibole might be present in all three types, whereas spinel, ilmenite, rutile, perovskite, Ti-rich garnet, titanite, apatite are typical minor and accessory phases. UML show similarities to other volatile-rich rocks, such as kimberlites, lamproites and silicocarbonatites in terms of the occurrences and mineralogy. Nevertheless, some compositional differences between the rock types and their distinctly different geodynamic settings (rift-related for UML and stable cratonic for kimberlites and lamproites) suggest that they have different magma sources and petrogeneses. Similar to kimberlites and lamproites, UML may contain diamonds [3][4][5][6][7], indicating that the depth of magma generation for UML can be in excess of 130 km.
In this paper, the mineralogy, whole rock compositional data and the age of the recently discovered Vinoren UML dyke within the Kongsberg silver district, Kongsberg lithotectonic unit, Southern Norway, are presented. Based on the new data, the origin of the dyke and the geodynamic implications of the discovery will be discussed.

Geological Setting
The major part of the crust in Southern Norway is built up of Paleo-to Mesoproterozoic rocks that underwent multiphase reworking along the Fennoscandian margin during the Sveconorwegian Orogeny, between 1140 and 920 Ma ago [11][12][13]. This orogeny was one of several orogenic events worldwide that resulted in the formation of the supercontinent Rodinia, and it has been inferred to result from the collision between proto-Baltica and Amazonia (e.g., [14][15][16][17]). However, an accretionary and non-collisional model for the formation of the Sveconorwegian Orogeny has also been proposed [18,19]. The orogenic belt has been sub-divided in five orogen-parallel lithotectonic units, which are separated by major Sveconorwegian shear zones: The Eastern Segment, Idefjorden, Kongsberg, Bamble and Telemarkia units [20].
The Kongsberg silver district is situated within the Kongsberg lithotectonic unit and includes a variety of gneisses (1600-1400 Ma) and granitoids (1171-1146 Ma) [17,21]. The silver district is characterized by subvertical zones enriched in sulfides (predominantly pyrite and pyrrhotite), inferred to be of hydrothermal origin. These zones, which are called fahlbands (e.g., [22,23]), are up to 900 m wide and subparallel to the foliation of the surrounding lithologies. The fahlbands and the older lithologies are crosscut by E-W trending dolerite dikes, quartz veins and silver bearing calcite veins of Permian age [24][25][26]. Already in the early days, the miners realized that the silver mineralizations occur almost exclusively at the intersections of the calcite veins and the fahlbands (e.g., [27]). Neumann [28] referred to the mineralized veins as calcite-nickel-cobalt-arsenide-native silver veins. The veins vary from a few millimeters up to 0.5 m in thickness, although up to several meters thick zones have been observed [28]. In a recent study of the silver mineralizations, Kotková et al. [29] gave an update of the paragenetic sequence presented by Neumann [28].
The UML dyke reported here occurs in the Klausstollen adit, adjacent to the Ringnesgangen underground silver mine, S. Vinoren, which is located in the northernmost part of the Kongsberg silver district (Figure 1). The dyke strikes toward NE with a dip of approximately 35° toward NW ( Figure 1). The dyke, which is about 50 cm thick, is fractured and tectonized; however, significant parts appears to be undeformed (Figure 2a). In places, the contact between the dyke and the host-rock appears as an undeformed and sharp intrusive contact. Some of the fractures within the dyke are filled with calcite.
Accessory mineral identification and qualitative composition of grains and mineral inclusions less than 20-30 µm was performed using a LEO-1450 SEM (scanning electron microscope) (Carl Zeiss AG, Oberkochen, Germany) equipped with XFlash-5010 Bruker Nano GmbH EDS (energy-dispersive Xray spectroscopy). The system was operated at 20 kV acceleration voltage, 0.5 nA beam current, with 200 sec accumulation time.
Materials from minerals forming possible pseudomorphs after olivine close to points analyzed by microprobe were examined by the X-ray diffraction (XRD) method (Debye-Scherer) by means of an URS-1 (Bourevestnik JSC, Saint-Petersburg, Russia) operated at 40 kV and 16 mA with RKU-114.7 mm camera and FeKα-radiation.

Whole Rock Analyses
Whole rock compositions were obtained at the Kola Science Center in Apatity, Russia. Most of the major elements were determined by atomic absorption spectrophotometry; TiO2 by colorimetry; K2O, Na2O, Cu, Ni, Co, Cr, V, Rb, Cs, and Li by flame photometry; FeO and CO2 by titration (volumetric analysis); and F and Cl by potentiometry using an ion-selective electrode (for the full description of the methods, see [30]).

40 Ar/ 39 Ar Analyses
Fragment of phlogopite with diameter about 1 mm was hand-picked from one phenocrystic sample of the dyke rock, cleaned by ultrasonic bath and dried up at 40 °C. The mineral fragment was in cadmium foil. The grain was placed in a capsule made of 99.999% aluminum. The sample was irradiated for neutron activation at the CLICIT (cadmium-lined-in-core irradiation tube) facility at the Oregon State TRIGA reactor (OSTR), Oregon State University, Oregon, USA. To obtain the degree of neutron activation (J), the neutron flux monitoring mineral Fish Canyon Tuff sanidine (27.5 Ma [31,32]) was used. To correct possible interference of Ar isotopes produced by the reaction of K and Ca, crystals of K2SO4 and CaF2 were irradiated separately. Irradiation time was 4 h, and the fast neutron flux was 2.47 × 10 13 n/cm 2 /s. After irradiation, the sample was cooled down for one month and transported to the Ar/Ar laboratory at the University of Potsdam, Germany. The sample was analyzed with a Gantry Dual Wave laser ablation system by the stepwise heating method until total melting. The system work with a 50 W CO2 laser (wavelength of 10.6 µm), using a defocused continuous laser beam with a diameter of maximum 1500 µm during 1 min for heating and gas extraction. The released sample gas was exposed to the SAES getters and cold stainless trap cooled at −90 °C through the ethanol by electric cooler in order to purify the sample gas to pure Ar for 10 min in a closed ultra-high vacuum purification line. The pure argon gas was analyzed by a Micromass 5400 noble gas mass spectrometer with high sensitivity and ultra-low background. The spectrometer operates with an electron multiplier for very small amounts of gas. During the measurements, blanks were measured every third step. The software Mass Spec, designed by Dr. Alan Deino of Berkeley Geochonology Center, Berkeley, CA, USA was used for processing the data. The recommended atmospheric 40

Petrography and Mineral Compositions
In hand specimen, the UML rock is massive and characterized by anhedral phenocrysts of phlogopite (up to 1 cm in diameter) and calcite (up to 1 mm in diameter), and rounded aggregates of a serpentine-like mineral (up to 3 mm in diameter) in a fine-grained, grey groundmass ( Figure  2b The carbonate in the groundmass is represented by almost pure calcite with <0.1 wt.% MgO (Table 1). We infer that the mineral is primary as it forms triple-junction boundaries between intergrown grains (Figure 3f). Secondary calcite occurs in aggregates with serpentine-like minerals and is characterized by high SrO content (up to 2 wt.%).
Phlogopite occurs both as phenocrysts and as grains up to 1 mm in the groundmass ( Table 2). The phenocrystic phlogopite is homogenous, whereas two types of chemical zonation can be observed in the groundmass phlogopite. In back-scatter electron (BSE) images, the first type of zonation is characterized by a dark core and brighter rim of phlogopite ( Figure 3e). The bright rim typically shows higher BaO than the core. The second type of zonation is represented by a few µm thick bright rims in BSE images (Figures 3e and 4), reflecting elevated FeO and lower Al2O3 and MgO in the thin rims. The groundmass phlogopites are sometimes bent suggesting that the mineral already had formed when the magma was emplaced as a crystal mush.
The serpentine-like aggregates consist of a mixture of a mineral that is closer in composition to saponite than serpentine, and minor talc ( Table 3). The presence of saponite has been confirmed by XRD analysis. The formation of saponite after olivine and serpentine during low-temperature hydrothermal alteration has been reported from some kimberlite occurrences (e.g., in the Arkhangelsk province, [34]).
Ilmenite is present as the two solid solution series geikielite-ilmenite and ilmenite-pyrophanite. The first one occurs as ca. 200 µm rounded resorbed grains with titanite rims (Figure 5e). The composition of the grains varies from core to rim mainly in MgO (from 12 to 2 wt.%), FeO (from 31 to 42 wt.%) and MnO (from 0.4 to 3.9 wt.%) ( Table 5). The mineral is characterized by the presence of Al2O3 (0.44-0.57 wt.%), NiO (0.12 wt.%), Cr2O3 (up to 0.09 wt.%) and CaO (up to 0.13 wt.%). Ilmenite of similar Mg-rich composition is an indicative mineral for diamondiferous kimberlites. The compositional zonation revealed for ilmenite from the studied dyke is similar to that from Torngat UML. Ilmenite corresponding to the ilmenite-pyrophanite series (up to 16 wt.% of MnO) occurs as single 10-20 µm grains included in titanite (Figure 5f). Ilmenite compositions like this are characteristic for carbonatites.
Rutile is a relatively abundant accessory mineral, found in titanite in association with lucasite-(Ce) (Figure 5b,g,h). The replacement of rutile by titanite apparently took place during a late-magmatic carbonatization stage with high Ca-and REE-activities. Rutile is characterized by a moderate Nb2O5 content (0.4-0.6 wt.% (Table 5)) that is different from typical Nb-rich kimberlitic rutile. The associated lucasite-(Ce) belongs to the same stage and occurs as needles included in titanite. Lucasite-(Ce) is a characteristic mineral of diamondiferous lamproite, e.g., from Argyle, Western Australia [37]. Vinoren lucasite-(Ce) differs from the lamproitic mineral by elevated CaO (3.3-5.5 wt.%, Table 5).
Garnet is a secondary minor mineral formed as bud-shaped grains associated with saponite and in interstices between grains of phlogopite (Figure 6e,f). EMPA data (Table 3) indicates that the mineral is hydroandradite [Ca3Fe 3+ 2(SiO4)3-x(OH)4x] with low to moderate TiO2 content (0.3-1.2 wt.%), in contrast to the Ti-rich garnets that is characteristic for UML.
Zircon occurs as needles of about 20 µm long, assembled in subparallel aggregates ( Figure  6a,b). The skeletal form of zircon indicates rapid growth of the mineral.
A Ni-Fe-S mineral phase with the composition 30.8 wt.% S, 38.9 wt.% Ni, 27.1 wt.% Fe and 3.1 wt.% Co (possibly godlevskite: (Ni,Fe)9S8), which occurs as numerous rounded grains of 1-2 µm in diameter in the saponite-talc aggregates (Figure 6c), is inferred to be an alteration product after olivine. Secondary quartz occurring in the saponite-talc aggregates is also inferred to be an alteration product after olivine. Barite and strontianite form anhedral grains, 1-3 µm in diameter, occur as inclusions in calcite. Other accessory phases that were observed (pyrite, galena, chalcopyrite, sphalerite, pentlandite, and celestine) in couple with other sulfides and sulfates indicate a relatively high S activity during the formation of the studied dyke.     Note. bdl-below detection limit. Note. P-phenocryst; G-groundmass; na-not analyzed; bdl-below detection limit. Note. na-not analyzed; bdl-below detection limit. Note. bdl-below detection limit; host-main part of spinel grain. Note. na-not analyzed; bdl-below detection limit.

Figure 7.
Whole rock compositional field for ultramafic lamprophyre, kimberlite and melilitite rocks (after [38]). Blue Gray circles show data from this study for the Vinoren occurrence.

40 Ar/ 39 Ar Geochronology
Results and measurement conditions of 40 Ar/ 39 Ar analyses of Vinoren phlogopite are given in Table 7. Plateau was not obtained. But an arithmetic average age of 686 ± 9 Ma was calculated from the last 5 steps which show very similar ages (Figure 8a). The integrated 40 Ar/ 39 Ar age is 689 ± 3 Ma. The measured Ca/K ratios were very stable, indicating that phlogopite has not been affected by alteration or degassing processes. In the normal isotope correlation diagram in Figure 8b, the data yields an age of 679 ± 6 Ma.

Geochemical Constrains for Rock Affinity
From its diagnostic mineralogy (carbonate-rich, but nepheline-and/or alkali feldspar-and melilite-absent; see section 4.1) and whole rock geochemistry (low SiO2 and Al2O3, high TiO2, CO2, P2O5, Ba and Sr; see section 4.2), the rock is classified as aillikite. According to [2], aillikite is a carbonate-rich member of the UML group derived from a volatile-rich, potassic, SiO2-poor magma.
The affinity and a possible source of the studied rock can be constrained by comparative studies. The nearest UML occurrences of similar age and tectonic setting are from the Labrador-Greenland areas, which are the parts of NAC. Two aillikite occurrences in these areas, i.e., Aillik Bay and Torngat, were chosen for comparison as their parental magmas originated at different depths [5,35,39]. The Aillik Bay aillikites are diamond-free, whereas the Torngat rocks are diamond-bearing with accessory mineral and xenocryst assemblages indicating a deep source. The Vinoren rock shows similar contents of SiO2, Al2O3, K2O, CO2 and P2O5 as the Torngat aillikite, but lower MgO, Na2O and higher CaO (Figure 9). At the same time the studied aillikite is differing from the Aillik Bay rocks by most components. It has been proposed that the Torngat ailikite was related to partial melting of metasomatized mantle (assemblages similar to MARID = mica-amphibole-rutile-ilmenite-diopside xenoliths from kimberlites [40]) during CO2 fluxing [7]. MARID nodules and veins are highly enriched in volatiles and incompatible elements [41,42] and according to [43], they crystallize within the diamond stability field, i.e., >4 GPa. Although aillikites are rich in MgO and Ni, their low SiO2 content and high contents of alkalis and volatiles suggest that they cannot be produced by melting of pure mantle peridotite. Foley [44] suggested a vein-plus-wall-rock melting mechanism for the generation of lamproitic magma. Accordingly, potassic and hydrous lamproitic magma can be produced by remelting of phlogopite-richterite-clinopyroxene dominated veins accommodated in peridotite of subcontinental lithospheric mantle (SCLM). Later, Foley et al. [45] and Tappe et al. [39] developed a similar model for the generation of UML melts, using a phlogopite-carbonate vein assemblage with minor apatite and Ti-oxide. Their remelting can produce potassic, hybrid carbonate-ultramafic silicate magma batches corresponding to aillikite melts. This has not been directly demonstrated yet, but the process is confirmed by experimental data [43], and encouraged by proximity of diamond-bearing aillikite and model MARID (see Figure 9). Both phlogopite and K-richterite can be present in MARID assemblages. However, the extremely high K/Na of the Vinoren aillikite combined with its strongly Si-undersaturated character indicate a dominating role of phlogopite in the source, because melting of a richterite-dominated source would have given more Si-rich melts. The difference in Na and K composition between the natural products and model MARID-like material (Figure 9) can be explained by the extremely different proportions of amphibole and mica in MARID. The low MgO/CaO ratio (<1) of aillikite suggests that calcite is the dominating carbonate in the source. The high TiO2 content of aillikite (2.75 wt.%) cannot be explained by melting of Ti-rich phlogopite only, suggesting the presence of ilmenite and/or rutile in the source [46]. Figure 9. Major element oxide vs. SiO2 (wt.%) of the Vinoren aillikite (gray circles). Also shown are the compositional fields of the diamond-bearing Torngat aillikite [35] and the diamond-free Aillik Bay aillikite [39] in Labrador which are of similar ages as the Vinoren rock. The black box shows the experimental melt compositions produced from MARID-type material [43].

Mineralogical Constrains for Rock Genesis
Minerals belonging to the phlogopite, oxyspinel and ilmenite groups may give important information about the mechanisms responsible for the genesis of volatile rich ultramafic rocks.
The chemical zonation observed for the groundmass phlogopite shows high kinoshitalite and tetraferriphlogopite components along the rim of the mineral. Kinoshitalite-rich rims are characteristic of kimberlitic mica [47], while tetraferriphlogopite rims are typical of lamproitic mica [36]. The elevated BaO content in phlogopite from Vinoren (up to 2.3 wt.%) is much lower than what is observed from kimberlites, but higher than what is typical for phlogopite from aillikites. BaO content of 3.5 wt.% has been recognized in UML, including diamondiferous ones, from Australia [48,49]. The high TiO2 (4-7 wt.%) in phlogopite from Vinoren is distinctly different from phlogopite from kimberlites and orangeites, but close to the compositions of phlogopite from UML and lamproites ( Figure 10). Furthermore, the Al2O3 content in Vinoren phlogopite is different from high-Al kimberlitic phlogopite and low-Al orangeitic and lamproitic phlogopite. Phlogopite from orangeites and lamproites typically shows an evolutionary trend with an increase in Fe coupled with a decrease in Al toward pure tetraferriphlogopite. For phlogopite from the Vinoren rock, this trend is very weakly developed. In conclusion, phlogopite from Vinoren shows a hybrid character with some similarities to phlogopite from kimberlites and lamproites, but it is more similar to UML phlogopite, and it shows some affinity to MARID-like phlogopite ( Figure 10). . Compositional fields and evolutionary trends of phlogopite from kimberlites, orangeites, lamproites and lamprophyres are after [50]. MARID (mica-amphibole-rutile-ilmenite-diopside) compositional field is after [40] and [51]. Phlogopite compositions from Torngat ultramafic lamprophyres (UML) are from [35].
The compositional variations of ilmenite from Vinoren indicate a hybrid nature also of this mineral ( Figure 11). The Mg-rich core (up to 12 wt.%) is typical for kimberlitic ilmenite, while the more marginal part of the mineral is similar ilmenite from UML. The elevated MnO content (up to 3.9 wt.%) may be considered as a result of the reaction trend in kimberlitic ilmenite as shown in Figure 11 [47,52,53]. Moreover, similar Mn-rich ilmenites have been observed as inclusions in diamonds from Brazil [54,55].
Thus, phlogopite, ilmenite and spinel from the studied rock show compositions that suggest a hybrid and multistage origin of the rock. It is inferred that a primary melt originated from deep (kimberlitic) and possibly diamond-bearing mantle levels. Phlogopite compositions indicate that the melt originated from MARID-like source. During the ascend, the residual silicate melt with significant carbonate content was still reactive and resulted in the formation of ilmenite, manganilmenite and titanomagnetitic spinel at shallower (UML) mantle levels. Fe 2+ /(Fe 2+ + Mg) of spinel from the studied rock. The compositional fields for magnesian ulvöspinel/Cr-spinel from kimberlites (trend 1) and titanomagnetite from lamproites and UML (trend 2) are from [39] and [50].

Possible Geodynamic Setting of Vinoren Aillikite
The North Atlantic Craton of Rodinia is composed of Archean blocks surrounded by Paleoproterozoic mobile belts covering large areas in the Northeastern Quebec, Labrador and Western Greenland ( [15], and references therein). Widespread lithospheric thinning occurred throughout eastern NAC along the Laurentian margin during the Late Neoproterozoic [59][60][61][62], resulting in continental breakup and subsequent opening of the Iapetus Ocean at 600 Ma, which was associated with rift-related UML-carbonatite-kimberlite magmatism. In central Labrador, this episode of continental stretching is recorded by remnant graben structures forming the eastward continuation of the St. Lawrence valley rift system [63]. Although Baltica today is separated from Laurentia, the two continents probably shared a common drift history during the time interval 750-600 Ma.
Studies of Neoproterozoic sedimentary systems along the northwestern region of Baltica, and geochemical and geochronological studies of magmatic rocks in the same region, have been used to constrain the break-up of Rodinia [60,64,65]. Prior to the active rift-related drift at ca. 600-550 Ma [66,67], this margin was inferred to have faced Laurentia (e.g., [68][69][70]).
During this stage, with thin SCLM and shallow asthenosphere, several carbonatitic-ultramafic complexes formed, including the Fen Complex in South Norway [71,72], the Seiland Igneous Province in North Norway (e.g., [73]) and the Alnö Carbonatite Complex in Sweden [74,75]. The initiation of rifting along the Baltic margin is marked by the 650 Ma Egersund tholeiitic dykes (SW Norway) which probably were derived from a mantle plume [60]. The emplacement of the Vinoren aillikite pre-dates this event. This is in accordance with the concept of [76] suggesting that continental extension was going on from 750 to 530 Ma, but separated in two distinct phases: (1) At 750-680 Ma, and (2) at 615-550 Ma. The first phase marked a failed rifting event between Laurentia and Amazonia, while the second phase led to the final breakup of Rodinia and the opening of the Iapetus ocean. Our data show that the first phase was active also between Laurentia and Baltica. The geochemical and mineralogical data presented here suggest that the parental magma of the dyke originated under a relatively thick SCLM, and that the continental root might have reached the depth of diamond stability.