INTRODUCTION

Determination of the physicochemical parameters of granite crystallization is important for understanding the evolution of felsic magmas and processes leading to the accumulation of ore matter. The study of multiphase intrusives with granites of various mineral or geochemical types is of special interest, because it allows the reconstruction of the evolution of magmatic melts, that can lead to the formation of potentially ore-bearing magmas with respect to trace and transit elements. The precise estimation of magmatic temperatures for granitoids (mostly, their alkaline types), however, is often difficult due to the peculiarities of their mineral composition. Zircon (Zrn) is one of the most popular indicator minerals of the temperature regime of the formation of granitoid plutons. Several approaches have been proposed to its use as a geothermometer: (1) a zircon-saturation index [1–4], (2) the Ti content of zircon [5, 6], and (3) the partition of Zr and Hf between zircon and the melt [7, 8]. Each approach has advantages and disadvantages. For example, approach 1 suggests that Zrn is one of the first subliquidus minerals, which is generally incorrect [2, 8]. Approach 2 requires a careful analysis of the very low Ti content in Zrn and mainly a quantitative account of TiO2 activity in the melt at the moment of crystallization of Zrn [5, 6]. Approach 3 implies that the rock contains no minerals (in addition to Zrn) that are able to fractionate Zr and Hf, which is also an atypical case. For example, the Zr/Hf ratio of amphibole and biotite (the most typical mafic minerals of granitoids) is 22–24 and 13, respectively [9], which strongly differs from the chondritic ratio of ~35–40 [10].

The problems related to Zrn as an indicator of the thermal regime of granitoids are considered in our paper for granites of the Adzh-Bogd pluton in southwestern Mongolia. Some genetic interpretations are proposed as a result of our work.

OBJECT OF STUDY

The Adzh-Bogd granitoid pluton is located in Transaltai Gobi within the Hercynides of the Central Asian fold belt (Fig. 1). This is a subisometric pluton ~20 km in diameter, which belongs to the Early Permian magmatic stage. The granitoids of the pluton intrude the Early Carboniferous calc-alkaline volcanic series and intrusive rocks of a wide range from gabbro to granites. The pluton is composed of rocks of two main intrusive phases, which are divided by a series of mafic dikes. The early phase, that composes the main intrusive body, includes medium- to small-grained two feldspar leucocratic granites. The rocks of the late phase occupy ~15% of the area of the pluton, are abundant in its central part, and also compose a satellite 1.5 km to the north. These are mostly coarse- to medium-grained alkali-feldspar granites. The two feldspar granites of the first intrusive phase contain phenocrysts of pertitic feldspar in a small-grained matrix of albite, K-feldspar, and quartz. The alkali-feldspar rocks of the second intrusive phase consist of coarse crystals of pertitic feldspar and quartz. The amount of mafic minerals in both rock types does not exceed 5%. These are mainly hornblende and biotite. The accessory minerals include zircon, apatite, ilmenite, and titanite. The U–Pb ID-TIMS age of zircon from alkali-feldspar granites of the northern satellite (Bum pluton) of the second intrusive phase is 294 ± 5 Ma [11].

Fig. 1.
figure 1

Schematic geological map of southwestern Mongolia. 1, Mesozoic–Cenozoic sedimentary cover; 2, Late Neoproterozoic–Early Paleozoic island-arc complexes of the Ozernaya Zone; 3, Early Paleozoic (?) metamorphic complexes of the Gobi–Tien Shan Zone; 4, Early Paleozoic metavolcanosedimentary complexes of Mongolian Altai; 5, Middle Paleozoic island-arc complexes of Transaltai Gobi; 6, Early Carboniferous marginal—continental volcanosedimentary complexes; 7, zones of the tectonic mélange; 8–10, granitoids: 8, pre-Permian calc-alkaline granites; 9, Permian calc-alkaline granites; 10, Carboniferous–Permian alkali-feldspar granites; 11, main faults. Inset: deciphering scheme of the Landsat space image-7 of the area of the Adzh-Bogd granitoid pluton with the position of the studied samples: 12, two feldspar granites of the early phase (GrI); 13, alkali-feldspar granites of the second phase (GrII).

Full size image

PETROCHEMICAL CHARACTERISTIC OF GRANITES

The granites of two types significantly differ in the chemical composition (Table 1). The SiO2 content of two feldspar granites of the first intrusive phase (samples 22, 24; hereinafter, GrI) is 71–74 wt % at an Al2O3 content of 14–15 wt % and total amount of alkalis of 8 wt % (the agpaitic coefficient Kа = (Na2O + K2O)/А12О3 (mole) is 0.7–0.8). The alkali-feldspar granites of the second intrusive phase (samples 20, 21; hereinafter, GrII) are characterized by a SiO2 content of 67–70 wt %, an Al2O3 content of 16–17 wt %, and the total amount of alkalis 11–12 wt % (Table 1) at the agpaitic coefficient Kа of 0.9. According to the classification diagram (Fig. 2) based on the Peacock calc-alkaline index, the composition of GrI and GrII corresponds to the calc-alkaline and alkaline series, respectively. The rocks strongly differ in the Zr and Hf contents: 131–135 ppm Zr and ~4 ppm Hf in GrI in contrast to 410–460 ppm Zr and 8–9 ppm Hf in GrII.

Table 1. Composition of granitoids of the Adzh-Bogd pluton and glasses of homogenized melt inclusions in zircon of alkali-feldspar granitoids
Full size table
Fig. 2.
figure 2

Petrochemical classification of granites of the Adzh-Bogd pluton according to the Peacock calc-alkaline index. (1) Alkali-feldspar granite of the second phase (GrII); (2) two feldspar granite of the early phase (GrI).

Full size image

COMPOSITION OF ZIRCON

Zircon from both granite types is semitransparent with an average size of 0.1–0.2 mm. The grains exhibit oscillatory and (locally) sectorial zonation (Fig. 3) evident of magmatic origin. All grains are heterogeneous in the HfO2 content and each studied sample contains zircons with different zonation trends. Sample 22 (GrI) contains Zrn of two types: (i) ellipsoid grains with a low HfO2 content and a more or less systematic increase in its content to the margins (0.85–1.28 wt %; Fig. 3e, R5) and (ii) poorly zoned grains with small (10–15 rel %) variations in the HfO2 content (Fig. 3f, R6). Morphologically and chemically similar weakly zoned grains were also found in GrI (1.03–1.30 wt % HfO2, sample 24). GrII (samples 20 and 21) also host zoned (Fig. 3b, R9; 3c, R3) and homogeneous (Fig. 3a, R2; Fig. 3d, R4) Zrn grains with respect to the HfO2 content. The HfO2 content of more or less homogeneous Zrn grains from GrII could be close to both the central and marginal domains of the zoned grains. Similar patterns are also observed for Zrn of GrI.

Fig. 3.
figure 3

Profiles of the HfO2 content of zircons of granitoids of the Adzh-Bogd pluton: a–d, alkali-feldspar granite of the second phase (GrII); e, f, two feldspar granite of the early phase (GrI).

Full size image

MELT INCLUSIONS IN ZIRCONS OF GRANITES

Zircon contains recrystallized melt inclusions up to 30 µm in size. The thermometric experiments with inclusions in Zrn from GrII were conducted upon gradual heating with a temperature step of 50°C in a muffle furnace at the Institute of Geology of Ore Deposits, Petrography, Mineralogy, and Geochemistry, Russian Academy of Sciences (IGEM RAS, Moscow, Russia). During thermometric experiments with melt inclusions it was shown that the inclusions fully homogenize at a temperature of 825–850°C. To prevent decrepitation of the melt inclusions, that is possible after reheating experiments at atmospheric pressure and locally leading to escape of volatiles from the inclusions, we also conducted a series of experiments with melt inclusions in Zrn from these granites in high-pressure gas apparatus at the Institute of Experimental Mineralogy, Russian Academy of Sciences (IEM RAS, Moscow, Russia), at a temperature of 900°C and a pressure of 4 kbar with an exposure of five days and further isobaric quenching (2–3 min). It is found that the glasses of the homogenized melt inclusions in Zrn from GrII has a lower (relative to the rock) content of the amount of alkalis (6.7–7.6 wt %), FeO (up to 0.2 wt %), TiO2 (0.16 wt %), CaO (up to 0.4 wt %), and MgO (up to 0.16 wt %) (Table 1). The SiO2 content of glasses of homogenized melt inclusions is 72.3–73.6 wt %, whereas that of the rock does not exceed 70 wt %. Both melt inclusions in zircon and granites GrII are enriched in K2O (Table 1). The deficient analytical totals of the glasses of melt inclusions suggest that they may contain significant amount of H2O, the content of water, however, is difficult to estimate from the difference between 100% and the oxide sum (e.g., [7]) due to the small size of inclusions. In addition, Zrn hosts crystalline inclusions syngenetic to melt inclusions: alkali feldspar, quartz, and fluorapatite. These data and the results of study of melt inclusions give a reason to believe that Zrn crystallized after or simultaneously with alkali feldspar, quartz, and fluorapatite.

TEMPERATURE OF ZIRCON CRYSTALLIZATION

Two methods among those mentioned in the Introduction were used to reconstruct the thermal evolution of granites: (1) the zircon-saturation temperature (TZrnsat [4, 5]) and (3) the Zr–Hf geothermometer (TZH [7]). The initial version of the Zr–Hf geothermometer [7] was based on published experimental data on the solubility of zircon and hafnon in silicate melts of similar composition. Our recent new experimental data (in preparation) on the solubility of these minerals in a wide temperature range (900–1500°C) and thte composition of the melt confirmed the similar activity coefficients of Zr and Hf in the melt and allowed significant refinement of the geothermometer equation:

$${{T}_{{ZH}}},{\text{K}} = 1968{\text{/}}(14.548 + \ln {{[{\text{Hf/Zr}}]}^{m}} - \ln {{[{\text{Hf}}]}^{{Zrn}}}).$$
(1)

In Eq. (1), the superscript indices m and Zrn mean melt (rock) and zircon, respectively, and the element contents (ppm) are given in brackets.

Zircon-Saturation Temperature

The temperatures estimated from the zircon-saturation index are shown in Table 2. The temperatures, which were calculated from the equations proposed in [2] and [4], for the GrII samples almost coincide and are similar to the homogenization temperatures of melt inclusions in Zrn of these samples. The temperatures for GrI granites were significantly lower and the difference between calculations after [2] and [4] is large, although they lie within the errors of both calibrations. It is likely that this difference is related to a small extrapolation error of the high-temperature experimental data, that were the basis for the equation [2], to the area of relatively low temperatures. Note also that the equation suggested in [4] includes the parameters related to pressure (P) and the H2O content of the melt (xH2O), that slightly increases the uncertainty of the temperature estimation. In calculations we used Р = 0.1 GPa and xH2O = 0.1, that approximately corresponds to 3 wt % H2O, i.e., a value similar to the maximum solubility of H2O in the granitic melt at this pressure.

Table 2. Estimated temperatures of crystallization of zircon in granitoids of the Adzh-Bogd pluton
Full size table

Zr–Hf Geothermometer

Whereas all studied Zrn grains are heterogeneous with respect to HfO2 content, the calculation results by Eq. (1) strongly depend on the choice of the area corresponding to the beginning of mineral crystallization. Most zircons from both granitoid types have evident zonation with an outward increasing Hf content, that is consistent with numerous data on zircon from various igneous complexes ([7, 8] and references therein). This feature is caused by the preferred fractionation of Zr from melts relative to Hf during mineral growth. This leads to the enrichment of melts and crystallizing minerals in Hf upon cooling. The temperatures of the beginning of crystallization of these zircons were estimated by Eq. (1) using the composition of the central zones of these crystals with a maximum Zr/Hf ratio [7] excluding the zircon grain from sample 20 (Fig. 3b). First, according to the internal structure and strongly asymmetric zonation, this is most likely a fragment of a larger crystal, and, second, the HfO2 content of its core is similar to that of cores of zircons from GrI indicating that these crystals are antecrysts [8]. Thus, upon the calculation of TZH for similar grains, we used the composition of internal zones just next to the inherited cores.

The results of temperature calculations by Eq. (1) (Table 2) are generally in agreement with estimations based on the zircon-saturation index. For GrI they are closer to temperatures calculated by the equation in [2], and those for GrII are slightly lower than by both calibrations [2, 4] and lower than the estimations based on the homogenization temperature of melt inclusions in zircon (~850°C), although the estimations by all methods are consistent within the errors.

DISCUSSION

The differences in composition of melt inclusions in zircons and host granites GrII (Table 1) (the lower amount of alkalis, FeO, TiO2, CaO, and MgO and higher SiO2 content relative to the rock) indicate the earlier or simultaneous crystallization of amphibole with zircon. The Na2O/K2O ratio of inclusions is also slightly lower than in GrII, which is most likely related to the early crystallization of feldspar. It is difficult to estimate quantitatively the influence of the composition of the melt on TZrnsat, because the small sizes of melt inclusions prevent reliable measurement of the Zr content. Simple qualitative calculations, however, show that this effect is minor: the early crystallization even of all amphibole in the rock (4–5 vol %) with an average Zr content of 53 ppm [9] could lead only to a small (~3 ppm) decrease in the Zr content of the melt in comparison with the initial composition. This conclusion is also applicable to the influence of early crystallization of feldspar, the average Zr content of which is approximately five times lower than that of amphibole [9]. The effect related to a small decrease in the bulk Zr content should be compensated by the increasing SiO2 content of the melt. These arguments are also true for the estimated TZH, because the crystallization of amphibole and plagioclase could not significantly affect the Zr/Hf ratio of the melt.

The bulk Zr content of GrI is lower by 3–4 times than that of GrII, thus the early crystallization relative to the Zr-rich minerals could potentially affect the estimated TZrnsat. These granites, however, contain no amphibole, and the crystallization of biotite (5–6 vol %) could not significantly affect the calculation results due to the very low Zr content in biotite (1–2 ppm [9]).

Our estimations (Table 2) thus indicate a significantly higher (by 100–120°C) temperature of the formation of late alkali-feldspar granites (GrII) relative to the early two feldspar granites (GrI). In addition, GrI and GrII belong to two distinct igneous series: calc-alkaline and alkaline, respectively. Due to this fact, GrII cannot thus be considered the products of differentiation of GrI. Although the formation stages of GrI and GrII could be divided by a very short geological period, they could significantly differ in the thermal regime. The much higher temperature of the formation of GrII requires an additional heat source, which could be related to the mantle-derived melts [13]. Their involvement in the formation of the Adzh-Bogd pluton is evident from the presence of mafic dikes, which divide the rocks of thte two intrusive phases. The high-temperature nature of (sub)-alkaline A-type granitoids was repeatedly emphasized in petrological works (e.g., [13, 14]). The contrasting temperature regime of almost coeval calc-alkaline and alkaline intrusive phases, however, has been identified in our work for the first time.

CONCLUSIONS

(1) The two feldspar granites of the early intrusive phase and alkali-feldspar granites of the late phase were studied for the reconstruction of the thermal evolution of the Adzh-Bogd granitoid pluton. Two main approaches were proposed for using the zircon as a geothermometer: the zircon-saturation temperature and the Zr–Hf geothermometer. Both approaches demonstrate a similar thermal history of the formation of granites of the pluton.

(2) According to the temperature estimation by the zircon-saturation index, the two feldspar granites of the early phase (GrI) formed at temperatures of 700–770°C, whereas the crystallization temperature of alkali-feldspar granites of the late intrusive phase (GrII) was 850°C.

(3) The calculated crystallization temperature of the two granite types of the Adzh-Bogd pluton by Zr–Hf geothermometer lies in a narrow range: 700–730°C for the early two feldspar granites and 810–820°C for the late alkali-feldspar granites.

(4) According to thermobarogeochemical studies of the melt inclusions in zircon from alkali feldspar granites of the late phase the temperature range of melt inclusion formation was determined. It corresponds to the temperatures of 825–850°C.

(5) All estimated crystallization  temperatures of granites of two intrusive phases of the Adzh-Bogd pluton indicate a much higher (by 100–120°C) formation temperature of alkali-feldspar granites (GrII) relative to the earlier two feldspar granites (GrI). The much higher temperature requires an additional heat source, which could be related to the mantle-derived melts. Their involvement in the formation of the Adzh-Bogd pluton is evident from the presence of mafic dikes, that divide the rocks of two intrusive phases. A contrasting temperature regime of almost coeval calc-alkaline and alkaline intrusive phases, however, has been identified in our work for the first time.