On the nature of the kimberlites of the Eastern Azov area, Ukraine: Isotopic (U-Pb, Hf zircon and Sr, Nd, Pb rock data) evidence for a depleted mantle origin and subsequent crustal contamination

Shumlyanskyy L.V.*, Billström K.**, Tsymbal S.N.*, Bogdanova S.V.***
*N.P. Semenenka Institute of Geochemistry, Mineralogy and Ore Formation, Kyiv, Ukraine; **Naturhistoriska Riksmuseet, Stockholm, Sweden; ***Lund University, Lund, Sweden
lshumlyanskyy@yahoo.com

The geological position and structure of the kimberlite bodies of the Azov area are described in detail in (Yutkina et al, 2004). Four kimberlite pipes (Petrovska, Nadia, Yuzhna, Novolaspinska) and two dikes – Yuzhna and Novolaspinska (Tsymbal et al., 2007) are known from the zone of conjunction of the Azov part of the Ukrainian shield with the Donetsk folded belt. The ages of these rocks were previously determined by various methods. Stratigraphically, at least one kimberlite body is seen to cut Middle Devonian sediments and be overlain by Late Devonian effusives, and basically all of them carry xenoliths of Devonian sedimentary rocks. Hence, a minimum age of emplacement of these rocks is close to 360 Ma (end of the Devonian period), and some of them, if not all, have maximum ages close to ca. 400 Ma (beginning of the Middle Devonian). Yutkina et al. (2004) determined absolute ages of the kimberlites using the Rb-Sr method on whole-rock samples and on phlogopite fractions. Their ages of the Novolaspinska and Yuzhna kimberlite pipes are 383,3 ± 3,8 and 384,7 ± 3,9 Ma, respectively. Tsymbal et al. (2007) quote similar Rb-Sr phlogopite ages obtained by F. Kruger and J. Holtzhausen; falling in a range between 380 to 391 Ma from the Novolaspinska pipe, and between 383 and 386 Ma for the Yuzhna pipe. However, these authors also mentioned much older ages for certain phlogopite crystals, ranging up to 423 Ma for the Yuzhna pipe and 465 Ma for the Novolaspinska pipe. In addition, Tsymbal et al. (2007) cite results of U-Pb dating of 11 grains of zircon separated from the Novolaspinska kimberlite pipe and dike. The ages obtained for the pipe fall in the range 417-425 Ma, whereas the dike-hosted zircons indicate ages between 420 and 425 Ma.

We have dated three large (about 1-2 mm each) zircon grains separated from two core sections that transect Novolaspinska kimberlite pipe and dike, using the SIMS technique. Each grain was sub-divided into smaller fragments which were analyzed individually. The following results were obtained: Novolaspinska pipe – sample 396/9 depth 17 m - 409,6 ± 7,2 Ma, sample 396/9, depth 29 m - concordant ages at 410,0 ± 4,6 Ma (two points) and 375,8 ± 5,9 Ma (one point); Novolaspinska dike – sample 393/16, weighted average of 4 concordant results: 389,2 ± 5,2 Ma, one point yielded an almost concordant age at 359,3 ± 4,2 Ma.
It is not trivial to interpret the significant of available age results as there is an age spread beyond the analytical uncertainties and apparent internal discrepancies. The Rb-Sr system is more susceptible to secondary disturbances and the reported two-point isochrones may not always produce robust age information, in particular if crustal contamination was a significant process, but also U-Pb zircon data indicate a considerable age range. Thus, a main question is if there was only one, or possibly several pulses of kimberlitic magmatism? On the basis of U-Pb zircon data, we propose that there was at least two separate magmatic pulses. The first of these took place at 410 Ma, as also supported by ages for the older generation of phlogopites. This event was followed by a second one at around 385 Ma, which fits well with the majority of the Rb-Sr phologpite results and with geological evidence. A main argument for the suggested dual type of magmatism is the nature of the investigated zircons. Their morphology and size are consistent with an origin from a mafic source and are not reminding of zircons typical from crustal rocks elsewhere in the Ukrainian shield. Thus, it is not likely that the studied zircons are xenocrysts, but it still needs to be explained why splitted fragments of single crystals yield contrasting ages. Possibly, this can be given an explanation if the younger kimberlitic pulse both produced newly formed zircon grains as well as forming over-growths on previously formed, approximately 30 Ma older zircons. We suppose, that kimberlitic magmas got halted during ascent and zircon forms over a more 30 Ma time period before all magmas came to a rest at 385 Ma.

In the present work also Hf isotopic compositions were obtained from zircons separated from samples 396/9 depth 19 and 29 m, and sample 393/16. All the results are very homogeneous; a weighted average of 13 determinations of εHf₃₈₅ is 7,1 ± 0,2. We argue that the uniform Hf data support the previous view that U-Pb zircon data are not affected by crustal assimilation. That is, if significant inheritance of core material or new growth of rims occurred as a result of magma-wall rock interaction this would have shown up as much more variable zircon Hf isotope signatures than detected as the rocks in the early Proterozoic basement (and the Devonian xenoliths) are likely to be characterized by highly negative εHf₃₈₅values_.

The isotopic composition of Sr, Nd and Pb in kimberlites of the Azov area was studied by Yutkina et al. (2004). These authors analyzed 3 samples of kimberlites and reached the following conclusions: isotopic compositions of Sr and Nd (εSr from 0,7 to 29, εNd from 1,9 to -0,1) indicate that their source was located in a mantle of BSE type. Moreover, the Pb isotopic composition points to the contamination of initial melts by crustal material.

We have studied the isotopic compositions of Sr, Nd and Pb in two samples of kimberlites and one peridotite xenolith (Table 1). Measurements were carried out in the isotopic lab of the Swedish Museum of Natural History, Stockholm.

Table 1. Isotopic composition of Sr, Nd and Pb in peridotite xenolith and kimberlite.

Sample	⁸⁷Sr/⁸⁶Sr	⁸⁷Rb/⁸⁶Sr	εSr₃₈₅	¹⁴³Nd/¹⁴⁴Nd	¹⁴⁷Sm/¹⁴⁴Nd	εNd₃₈₅	²⁰⁶Pb/²⁰⁴Pb	²⁰⁷Pb/²⁰⁴Pb	²⁰⁸Pb/²⁰⁴Pb
1	0,714846±10	0,2307	132,9	0,512363±14	0,0680	0,9	23,105	16,021	47,542
2	0,707321±12	0,2478	24,7	0,512044±11	0,0812	-6,0	19,750	15,727	41,074
3	0,712611±8	0,3827	89,4	0,511981±13	0,0697	-6,7	22,634	16,093	46,121

Samples: 1 - peridotite xenolith, borehole 219/24, depth 81,3 m; 2 and 3 - two samples of kimberlite, borehole 209/9, depth 53,0 m.

Our results significantly extend the range of Sr and Nd isotopic compositions. In particular, our kimberlite data display significantly lower εNd₃₈₅ values than were previously obtained. Besides, the peridotite xenolith and one of the kimberlite samples exhibit very radiogenic Sr isotopic composition. The Pb data support the isotopic heterogeneity pointed out by Yutkina et al. (2004) and it is clear that there was no single, isotopically uniform source that characterizes the different magmas. Having said this, it is obvious that Sr, Nd and Pb cannot be used to derive useful age information.

Nd isotopic composition of the peridotite xenolith is close to that reported by Yutkina et al. (2004) for the least contaminated kimberlites. Petrologically speaking, peridotite would probably approximate the source rock that upon partial melting produced kimberlite melts. However, it is highly questionable if any Nd, Pb or Sr isotope data for the analyzed whole rock suite, including the peridotite, really represent an uncontaminated melt. Actually, considering that zircon is a refractory mineral and that Hf would not be easily mobilized in zircon, the Hf isotope compositions of zircons from kimberlites would likely put reliable constraints on the ultimate magma source. When inspecting the Hf data, we argue for a depleted mantle source, i.e. not a BSE type of source. Following this, the variable Sr, Nd and Pb isotope compositions for rocks that clearly are not compatible with a homogenous, Devonian magma source, would rather be consistent with the hypothesis that a depleted kimberlite melt assimilated crustal components during magma ascent. This is reinforced by the very radiogenic composition of Sr in the peridotite xenolith, as there are no indications for post-crystallisation events that could have altered the isotope systematics. The latter is probably true despite the fact that back-calculated Pb isotope data for one of the kimberlites results in geologically impossible ratios, but a possible explanation for this behaviour is erroneous chemical data used for the ²³⁸U/²⁰⁴Pb calculation.

Hence, considering isotope-geochemical data on kimberlites of the Azov area one can draw the following conclusions:

- Emplacement of kimberlitic melts occurred at about 385 Ma, at the border of Middle and Late Devonian. However, some of the material present in kimberlites had crystallized much earlier, at 425-410 Ma. Perhaps, this period corresponds to the time of "mantle metasomatism" - enrichment of lithospheric mantle by easily fusible material.

- Relatively little contaminated samples of kimberlite display relatively elevated εNd values (up to 1.9), while εSr is characterized by moderately positive values. At the same time, zircons extracted from kimberlites are characterized by high positive values of εHf. Samples of contaminated kimberlite reveal low εNd values (as low as -6), while εSr increases to +90 ÷ +130. Obviously, the primary source of kimberlitic melts was situated in a moderately depleted lithospheric mantle that is probably represented by peridotite xenoliths. It seems unlikely that the substance that caused the mantle metasomatism was very different in their isotopic characteristics from peridotites. In any case, the Hf isotopic composition in zircons that probably crystallized from "metasomatic" substance evidences about its depleted source.

- The main contaminant of kimberlite melts was presented obviously by crustal (upper-crustal?) material as can be concluded from the Sr and Pb isotopic composition.

Yutkina E.V., Kononov V.A., Bogatikov O.A,, Kniazkov A.P., Kozar N.A., Ovchinnikova G.V., Levski L.K. Kimberlites of the Eastern Azov (Ukraine) and geochemical characteristics of their sources / / Petrology, 2004, v. 12, № 2, pp. 159-175.

Tsymbal S.N., Kremenetsky A.A., Strekozov S.N., Bondarenko V.A. Age of kimberlites of the Azov block of the Ukrainian shield accordingly to geological and isotopic data // Alkaline magmatism of the Earth and its ore-bearing. Proceedings of the international meeting, Donetsk, 10-16 September 2007 - pp. 248-250.