Chemical Peculiarities of Quartz from Peralkaline Granitoids

Abstract

Quartz from four typical but contrasting peralkaline quartz-saturated granite systems (Khan Bogd and Khalzan Buregte plutons (Mongolia), Ivigtut stock (Greenland), Europa and Madeira plutons (Pitinga magmatic province, Brazil)) was analyzed using LA-ICP-MS to define the range of selected trace element content and trends in their evolution and to compare this content with published data from granitoids of other geochemical types. The evaluation of about 1100 analyses found the studied trace elements mostly in ranges <0.01–18 ppm Li (median 2.41 ppm), 1.2–77 ppm Ti (median 8.2 ppm), 8.3–163 ppm Al (median 42 ppm) and 0.05–5.7 ppm Ge (median 0.98 ppm) (in all cases 5% of the lowest and 5% of the highest values were omitted). Quartz from geochemically less evolved riebeckite-bearing granite plutons shows no Ti/Ge fractionation and displays either a positive Ti–Al correlation or no Ti–Al correlation. More fractionated and potentially mineralized peralkaline magmatic systems were formed within two distinct magmatic episodes: quartz from the older phases is relatively Ti-rich and evolved via Ti decrease with no possible Ge enrichment, while quartz from younger phases is Ti-poor from the beginning and has the ability of enrichment in Al and Ge. Relative enrichment in Al and increase in Ge/Ti value of quartz can serve as a supporting method for the identification of potentially ore-bearing magmatic systems.

1. Introduction

Quartz is, after feldspars, the second most abundant mineral in the Earth’s crust [1]. It is present in a large proportion of igneous, metamorphic and sedimentary rocks, and it is the dominant mineral in most hydrothermal accumulations. Knowledge of its chemistry is therefore important for a number of geological disciplines and in particular for understanding the conditions of formation of deposits of many mineral raw materials. Effective study of quartz chemistry in geological practice was made possible by the introduction of laser ablation inductively coupled mass spectrometry (LA-ICP-MS) into routine geological operations 20 years ago. In a relatively short period of time, a large number of quartz analyses have been carried out, and a number of studies have been published, including research on quartz from barren and mineralized granites [2,3,4,5,6,7,8,9,10], rhyolites [11,12,13,14], pegmatites [15,16,17,18,19] and hydrothermal deposits [20,21,22,23,24,25,26], but also from metamorphic rocks [27] and carbonatites [28]. Quartz trace element chemistry was also utilized for the study of the provenance of clastic sediments [29]. In contrast, quartz from quartz-bearing peralkaline rocks (sensu Shand [30]) has so far been studied only exceptionally [31]. We decided to at least partially fill this knowledge gap and studied quartz from four typical but different examples of peralkaline granitoids.

Studied plutons represent different styles of peralkaline quartz-saturated magmatism: 1. large strongly peralkaline granite pluton with numerous post-granitic dikes (Khan Bogd), 2. multiphase peralkaline pluton evolved in direction of subaluminous rare-metal granites (Khalzan Buregte), 3. small extremely F-rich peralkaline stock with strong metasomatism at Ivigtut, and 4. magmatic province with nearly contemporaneous but geologically distinct strongly peralkaline granite (Pitinga-Europa) and mildly peralkaline rare-metal granite (Pitinga-Madeira). In all these systems, we tried to obtain data from all important stages of magmatic evolution.

Our goals were 1. to define the range of selected trace element content in quartz of peralkaline rocks and trends in their evolution; 2. to compare these data with published data for other geochemical types of granitoids.

Using the LA-ICP-MS method, we monitored a wide range of elements, but the content of many of them was found to be influenced to an unknown extent by an incidental capture of fluid inclusions (Na, K, B, Rb, Sr) or nano-inclusions of other mineral phases (Na, K and Al in feldspars, REE in phosphates, HFSE in oxides like rutile or columbite). Here, we publish and discuss the content of elements Li, Al, Ti and Ge, which, according to a broad consensus of authors [1,2,15,25] provide the most reliable petrogenetic information.

2. Geology

This study comprises four geologically different but in all cases well-documented igneous systems. These systems deeply differ in the mode of their origin, style of fractionation, and trace-element geochemical specialization; nevertheless, all are of peralkaline character and—in most facies—saturated in quartz. This should allow one to constrain the principal chemical features of quartz crystallizing from peralkaline magmas of different types, as well as quartz from related metasomatites.

The Khan Bogd pluton, Early Permian in age (292–272 Ma, [32]), is located in the Ömnögovi district in Southern Mongolia, about 500 km south of Ulaanbaatar city. It represents the largest intrusion (1500 km²) within frame of the Gobi–Tien Shan zone of peralkaline magmatism. The pluton consists of two main parts (Supplementary Figure S1): The relatively larger circular western part (30 km diameter) is made of locally porphyritic coarse-grained aegirine–arfvedsonite granite with a minor content of zirconosilicates. The smaller and rather younger eastern part (15 × 13 km) is formed by fine-grained aegirine granite. A countless amount of ring dikes are typically namely for the western body. Their composition varies from layered aplite–pegmatites to ekerites (fine-grained alkalifeldspar granite with porphyritic arfvedsonite). Some of the dikes are strongly metasomatized, forming a so-called roof complex. From the chemical point of view, in a direction from the porphyritic pericontact to the equigranular central facies of the western body (samples 1–4 in

Table 1. Content of selected trace elements in quartz from particular rock facies (means and medians in ppm).

	N	Li Mean	Li Median	Al Mean	Al Median	Ti Mean	Ti Median	Ge Mean	Ge Median
Khan Bogd, sample 1	42	3.15	2.21	37.39	27.80	31.34	23.63	0.78	0.80
Khan Bogd, sample 2	39	2.57	2.52	36.46	29.47	33.05	30.99	0.76	0.72
Khan Bogd, sample 3	38	5.69	5.14	40.52	38.24	62.14	61.36	0.93	0.95
Khan Bogd, sample 4	38	1.99	1.38	41.06	32.25	30.53	23.27	0.75	0.78
Khan Bogd ekerite	45	0.89	0.25	26.29	13.97	19.55	8.48	0.82	0.70
Khan Bogd roof complex	23	9.31	9.34	46.45	43.10	81.78	83.10	0.71	0.75
Khalzan Buregte, phase 2	145	2.03	0.90	41.70	34.61	21.40	7.20	0.90	0.75
Khalzan Buregte, phase 3	67	1.61	1.06	44.38	31.17	12.46	9.02	1.17	0.92
Khalzan Buregte, phase 5	63	2.15	0.83	46.72	29.97	5.52	3.11	0.53	0.39
Khalzan Buregte, phase 7	77	11.81	13.44	122.82	115.14	6.34	4.89	2.01	2.06
Ivigtut hypersolvus granite	39	0.15	0.11	20.50	16.91	63.34	64.37	0.64	0.62
Ivigtut subsolvus granite and pegmatite	102	6.81	4.58	81.30	54.90	6.06	4.15	1.72	1.78
Ivigtut UST zircon-rich dike	23	3.26	2.65	35.40	34.50	2.06	2.03	3.76	3.84
Ivigtut cryolite body	69	14.58	6.93	329.00	97.17	2.11	1.58	4.42	3.63
Ivigtut metasomatite	26	4.52	2.12	269.00	45.20	1.91	1.80	1.99	1.82
Europa riebeckite granite	71	4.02	2.91	54.40	52.20	37.33	33.60	0.81	0.82
Madeira albite granite	73	14.00	10.20	96.80	92.40	7.40	4.47	3.88	3.94
Madeira cryolite pegmatite	41	1.87	1.06	56.30	46.62	1.98	1.41	8.23	8.30

), the content of SiO₂, Na₂O and Zr increases (72→74 wt%, 4.6→4.8 wt%, 860→1130 ppm, respectively), while the content of Al₂O₃ and K₂O decreases (12.9→10.4 wt%, 5.2→5.0 wt%, respectively). These trends are accompanied by a decrease in the alkalinity index (A/NK = Al₂O₃/(K₂O + Na₂O) (molar), according to [30]) from 0.98 to 0.78. Dikes of ekerites contain 73.6 wt% SiO₂, 10.1 wt% Al₂O₃, 6.2 wt% Na₂O and 3.7 wt% K₂O on average; A/NK = 0.71 [33]. The metasomatized dike rocks in particular are enriched in Zr, Nb, Y and REE up to economically interesting concentrations.

We analyzed four samples of the main intrusive phase, two samples of ekerite and one sample of a metasomatized roof-complex dike.

The Khalzan Buregte pluton (also Khalzan Buregtey or Khaldzan Burgedei, ca. 100 km², 395–391 Ma, [34,35]), is located in Western Mongolia, to the north of the town of Chovd. Kovalenko et al. [36] described seven peralkaline evolutionary phases here (Supplementary Figure S2): (1 + 2) large intrusions of nordmarkite (i.e., alkaline quartz-bearing syenite) and alkaline granite, forming a substantial part of the pluton; (3) dikes of ekerite (i.e., peralkaline alkalifeldspar granite with riebeckite); (4) dikes of pantellerite (i.e., peralkaline rhyolite); (5) a stock (0.85 km²) of peralkaline rare-metal granite (RMG) with pericontact pegmatoidal facies; (6) syenite dikes; and (7) a stock (0.05 km²) of miarolitic slightly peralkaline rare-metal amazonite granite. Based on a Nd isotope study, Kovalenko et al. [36] interpreted phases 1–4 and 5–7 to be a product of the fractionation of two alkaline magmas having different sources (εNd = 5.8–7.6 vs. 4.3, respectively). Both RMG phases are poor in Al (mostly 7–11 wt% Al₂O₃) but rich in F (mostly 0.5–3 wt%), Zr, REE and Nb (means at 1.39 wt% Zr, 0.31 wt% REE and 991 ppm Nb in the RMG phase 5 vs. 1.71 wt% Zr, 0.37 wt% REE and 1128 ppm Nb in RMG phase 7 [36,37]. Later, Nb enrichment in metasomatized parts of phase 2 was found by [38]. The alkalinity index varied between 0.75–1.11 and 0.54–1.01, respectively). Besides quartz and alkali feldspars (smoky quartz and amazonite in phase 7), the RMGs contain common fluorite and alkali amphiboles. Zircon and Nb minerals pyrochlore and columbite locally reached economically interesting concentrations [38].

Altogether 13 samples of quartz from phases 2, 3, 5 and 7 were studied.

The Ivigtut granite stock (1248 ± 25 Ma, [39]) is as a member of the Gardar alkaline igneous province [40]. It is located at the sea coast in Southwestern Greenland and is known worldwide due to its associated cryolite deposit, which has been completely exploited by now (Supplementary Figure S3). Slightly peralkaline to subaluminous granite (A/NK = 0.98–1.08) intruded an Archean gneiss forming pipe-like body 300 m across (ca 0.1 km²). Contact of the intrusion is highlighted by a thin zone of intrusive breccia; another “bunk” breccia forms an independent pipe-like body nearby. From the surface down, granite changes from a hypersolvus to a subsolvus texture at a depth of 300–700 m. The granite contains 68–75 wt% SiO₂, 12–14 wt% Al₂O₃, 3.7–5.1 wt% Na₂O and 4.0–5.2 wt% K₂O [41]. Different types of alteration, including albitization, sericitization and greisenization, were documented around the former cryolite body. The source of metasomatizing fluid/melt (?) transporting F, Na and CO₂ was interpreted by [42] in the lithospheric mantle enriched via subduction. The studied granite samples are composed of quartz, alkali feldspars, edenitic amphibole and biotite. Cryolite, zircon, fluorite and siderite are common minor phases. Pegmatite dikes were found near the upper border of the cryolite body, while dikes of granophyre and zircon-rich aplopegmatite dikes were found in the exocontact. Eleven quartz samples representing different facies of granite, pegmatite, zircon-rich dikes, metasomatites and the cryolite body were analyzed.

The Pitinga magmatic province, located in the central Amazon Craton, Brazil, is composed of several nearly contemporary peraluminous (Aqua Boa pluton), subaluminous (biotite–amphibole and biotite granites of the Madeira pluton) to peralkaline granite bodies (Europa pluton and core albite granite of the Madeira pluton, Supplementary Figure S4). The Europa pluton (1829 ± 1 [43]) forms a ring-shaped (12 km in diameter) homogeneous body of peralkaline hypersolvus riebeckite granite (74–77 wt% SiO₂, 3.8–4.2 wt% Na₂O, 4.5–4.9 wt% K₂O, A/NK = 0.99–1.00 [42]). The Madeira albite granite (1817–1820 Ma, [43]) forms a cupola-like body (~3 km²) in the center of the Madeira pluton and is intercalated with cryolite pegmatites and massive cryolite pods [44]. The relatively unaltered “core facies” is peralkaline and medium-grained, composed of quartz, alkali feldspars, riebeckite, cryolite and some Li-Fe mica. It contains 68–72 wt% SiO₂, 12.3–13.8 wt% Al₂O₃, 5.6–7.0 wt% Na₂O, 3.8–4.7 wt% K₂O and 0.6–3.2 wt% F [45]. Alkalinity index A/NK varies between 0.8 and 1.0. Pegmatite bodies consist of quartz, cryolite, relicts of alkali feldspar and some common sulfides. Their chemical composition in cryolite-rich domains shows unusual values of <10 wt% SiO₂ and >30 wt% Na₂O and >40 wt% F; A/NK decreased to 0.3 [46].

Three quartz samples were studied from the Madeira core albite granite, one from Madeira pegmatite and two from the Europa granite within a detailed study on the Pitinga magmatic province [31].

3. Analytical Methods

3.1. Cathodoluminescence (CL)

Prior to the LA-ICP-MS analyses, panchromatic CL images of quartz from polished thin sections were obtained using the Jeol JXA-8230 electron microscope (JEOL Ltd., Tokyo, Japan) with a tungsten filament electron source. The images were obtained using a single-channel panchromatic CL detector–photomultiplier tube with a wavelength range of 200–900 nm and a working distance of 11 mm. The accelerating voltage and current were set to 15 kV and 5–20 nA (according to the intensity of cathodoluminescence). The scanning speed ranged from 100 to 1000 μs/pixel depending on the CL emissivity of the imaged area.

3.2. Trace Elements in Quartz

The content of trace elements in the quartz samples (polished thin sections) was determined using an LA-ICP-MS consisting of a quadrupole-based ICP-MS (Agilent 7900, Agilent Technologies Inc., Santa Clara, CA, USA) connected to the 193 nm ArF* excimer laser ablation system Analyte Excite + (Teledyne CETAC Technologies, Omaha, NE, USA) equipped with a 2-vol Cell HelEx II. The ablated material was carried by He flow (0.5 and 0.3 L min⁻¹) and mixed with Ar (~1 L min⁻¹) prior to entering the ICP mass spectrometer. The sample surface of individual spots was ablated for 60 s per spot by a 50 μm diameter laser beam with a fluence of 12.7 J cm⁻², 10 Hz repetition rate and 60 s washout time. The monitored isotopes were as follows: ⁷Li⁺, ⁹Be⁺, ^{10, 11}B⁺, ²³Na⁺, ²⁷Al⁺, ^{28, 29}Si⁺, ³¹P⁺, ³⁹K⁺, ⁴⁵Sc⁺, ^{47, 49}Ti⁺, ⁵⁵Mn⁺, ^{56, 57}Fe⁺, ^{69, 71}Ga⁺, ^{72, 73}Ge⁺, ⁸⁵Rb⁺, ^{86, 88}Sr⁺, ⁹⁰Zr⁺, ⁹³Nb⁺, ^{116, 117, 118, 119}Sn⁺, ¹²¹Sb⁺, ¹⁴⁰Ce⁺, ^{172, 173}Yb⁺, ^{177, 178}Hf⁺, ¹⁸¹Ta⁺, ²³²Th⁺, and ²³⁸U⁺. The ICP-MS was tuned using SRM NIST 612 with respect to the sensitivity and minimum doubly charged ions (Ce²⁺/Ce⁺ < 5%), oxide formation (²⁴⁸ThO⁺/²³²Th⁺ < 0.3%) and mass response ²³⁸U⁺/²³²Th⁺ ~1. Potential interferences were minimized via a collision cell (He 1 ml min⁻¹). The elemental content was calibrated using artificial glass standards SRM NIST 610 and 612 and Si as the internal reference element after baseline correction and integration of the peak area. The average detection limits under the operating conditions were 0.73 ppm Li, 1.03 ppm Al, 0.36 ppm Ti and 0.55 ppm Ge. To minimize undesirable hits of hidden mineral or fluid inclusions, we controlled the target area in a transmitted light mode and deleted ca. 5% of outliers.

4. Results

4.1. Zoning of Quartz Crystals in Cathodoluminescence (CL)

The intensity and shape of CL activity in the studied quartz crystals and grains are highly variable but generally low compared to the surrounding minerals.

At Khan Bogd, the quartz of the fresh coarse-grained granite facies has a distinct but inconsistent zonal CL structure. Grains with a “standard” zoning, i.e., relatively light cores and dark rims (Figure 1a), are closely associated with grains of the opposite color in the same thin section. There is no uniform relationship between CL intensity and the content of any of the elements studied. Quartz from slightly metasomatically altered granite and from the dike rocks is not CL-active.

Quartz from Khalzan Buregte often shows two evolutionary stages, with a marked difference in CL between the often euhedral core and usually anhedral rim. Grains with light oscillatory-zoned cores and dark homogeneous rims are more common than grains with an opposite zoning trend. Nevertheless, quartz grains with irregular, patchy or no CL are also common. The latest phase 7 of RMG contains grains with slightly CL-active euhedral cores and dark black rims penetrating into the surrounding groundmass (Figure 1b).

In the Ivigtut granite stock, only hypersolvus granite contains CL-active quartz in the form of rounded grains with numerous late healing structures (Figure 1c). Other quartz varieties (subsolvus granite, pegmatite, dike rocks and hydrothermal bodies) show no luminescence.

Among quartz varieties from the Pitinga magmatic province, Brazil, quartz from the riebeckite granite from the Europa pluton usually forms anhedral grains with a homogeneous to patchy CL without any zoning but with numerous veinlets and domains of late CL-black healing quartz (Figure 1d). Quartz from the Madeira albite granite is usually CL-free, and only one of the studied samples contains CL-active grains with usually rounded black cores with relatively thin light rims (Figure 1e).

4.2. Chemical Composition of Quartz

We performed ca. 1100 spot LA-ICP-MS analyses. After having eliminated spots with a noticeably high content of any of the monitored elements, i.e., probably occasional hits of mineral inclusions hidden under the surface of the measured sample, we herein present the results of 926 spot analyses of quartz from peralkaline granitoids and, in the case of Ivigtut, also from associated metasomatites. Despite the careful effort to review primary analytical results, it cannot be ruled out that some other above-average values are also influenced by hidden mineral inclusions (MIs). For this reason, the median values of all elements have a higher informative value than the mean values (Table 1).

Ti, Al, Li and Ge are the trace elements with the highest potential of petrogenetic information. As has been repeatedly confirmed, Ti content usually decreases and Al and Ge content increase during magmatic fractionation. Ti decreases due to the early crystallization of Ti oxides, while the increase in Ge can be explained by its low compatibility in most crystallizing minerals, and thus by a steady increase in the Ge content in the magma [47]. The higher water and Li content in the late magma may explain the easier replacement of Si in the quartz lattice by Al according to the equations Si⁴⁺↔Al³⁺ + H⁺ and Si⁴⁺↔Al³⁺ + Li⁺. Conversely, the increase in Al content in late quartz makes it possible to enrich it with Li up to the atomic ratio Li/Al = 1/1 [1,2,17,48,49].

Therefore, increasing values of Al/Ti and Ge/Ti ratios fairly reliably indicate an increasing degree of fractionation of the parent magma [3,6,9,14,18]. General evolutionary trends of these elements and differences among the studied magmatic systems are visualized in Figure 2; a more detailed view into the heterogeneity of individual plutons is presented in Figure 3, Figure 4 and Figure 5 and in Table 1. There, the content of individual studied samples (Khan Bogd), evolutionary stages (Khalzan Buregte) or facies (Ivigtut, Europa and Madeira) is presented separately.

The content of titanium at individual analytical spots varied significantly from <1 ppm to 180 ppm (Figure 2a), but this range is much smaller (3.11–83.10 ppm) when comparing the median values of parental rock types. The highest Ti values are found in most rock types from Khan Bogd (medians 19.55–62.14 ppm with a relatively wide variability within a single sample, Figure 3a) with a maximum in a metasomatized dike (median 81.78 ppm with only a narrow variability) and in the Europa pluton (13–70 ppm Ti, median 33.60 ppm Ti, Figure 3d), while values below 5 ppm dominate the more fractionated rocks of the Khalzan Buregte and Madeira plutons. At Ivigtut, quartz from fresh less fractionated hypersolvus granite (40–105 ppm (Figure 3c), median 63.34 ppm Ti) differs significantly from quartz from altered granites and pegmatites (median 6.06 ppm Ti) and all other rock varieties (2 ppm Ti). Extreme Ti variability is displayed by quartz from granites of phase 2 at Khalzan Buregte (median 7.20 ppm Ti), with values up to 180 ppm in the rims of some grains (Figure 3b). Similarly, in the Madeira albite granite, the relatively scarce high-Ti values (20–77 ppm at a median of only 7.47 ppm Ti) were found in the rims of some euhedral quartz crystals (compare Figure 1e and Figure 6c).

The content of aluminum in individual spots varied mostly between 10 and 200 ppm Al (Figure 2a), with values up to 450 ppm Al in granites of phase 7 at Khalzan Buregte (Figure 3b) and 1550 ppm in metasomatites at Ivigtut (Figure 3c). In terms of medians, our set is more homogeneous (medians of individual samples range mostly 30–60 ppm), and only granite of phase 7 at Khalzan Buregte exceeds the value of 100 ppm Al (Table 1). Our data indicate a certain antagonism of higher Al and Ti content at Khalzan Buregte, which is caused by the evolution from Ti-rich and Al-poor quartz from early phases to Al-enriched Ti-poor quartz of late intrusive phases (Figure 3b). At Ivigtut (Figure 3c), the early intrusive phases are also Ti-rich and Al-poor, changing to a Ti-poor and Al slightly enriched late intrusive phase and finishing with Ti-poor Al-rich metasomatites. The great majority of analyses of aegirine–arfvedsonite granite from Khan Bogd give a positive Al vs. Ti correlation (Figure 3a), and no correlation was found in the case of the Europa granite (Figure 3d). An increase in Al during fractionation, often proposed in the literature [1,3,6,18], is visible at Khalzan Buregte (a shift between comagmatic phases 5 and 7, Figure 3b) while no Al increase occurs between albite granite and the associated pegmatite in the Madeira pluton (Figure 3d).

Another common trace element in quartz is lithium. Its content generally varied between <1 and 25 ppm (Figure 2b), with only individual spots up to 75 ppm in quartz from the Madeira albite granite (Figure 4d). Medians mostly varied between <1 ppm and 7 ppm; exceptions are the metasomatized dike from Khan Bogd (9.3 ppm), Madeira albite granite (10.2 ppm) and granite of phase 7 from Khalzan Buregte (13.4 ppm Li). Al and Li enter the quartz crystal lattice following the equation Li + Al↔Si, so an atomic Li/Al value of about 1 should be generally expected (2, 7, 9, 17), but Li/Al values substantially lower than 1 prevail in the referred dataset. This means that a substantial portion of aluminum enters the quartz lattice in a different way, probably following the equation Al + OH↔Si + O.

Germanium is chemically similar to silicon [47,50] and, as such, possesses a common trace constituent of quartz. Its content in most types of studied rocks varied between 0.1 and 10.5 ppm Ge (Figure 2c) but quartz from all riebeckite- or arfvedsonite-bearing granites contained less than 2 ppm Ge (Figure 5a,b,d), and a content above 6 ppm was found only in quartz from the Madeira cryolite pegmatite (Figure 5d). The medians varied from <1 ppm in all rock types from Khan Bogd to 3.8 ppm in a UST dike from Ivigtut, 3.9 ppm in the Madeira albite granite and 8.2 ppm in the Madeira cryolite pegmatite; the increase in Ge content during magmatic fractionation is therefore convincing. Relatively high Ge values were also found in quartz from the Ivigtut cryolite body (median 3.63 ppm), while quartz from metasomatites in its surroundings contained only 1.82 ppm Ge (Figure 5c).

4.3. Chemical Zoning of Quartz Grains

Considering the generally applicable rule of a decreasing Ti content and increasing Al and Ge content during magmatic evolution via fractionation [2,3,6,9,14,18], fully confirmed by our results (Figure 2), we should expect a similar evolution at the scale of individual zoned crystals. But the situation for individual crystals is ambiguous. Similar to the contrasting CL patterns (Figure 1a vs. Figure 1e), crystals with opposite trace element trends can be found (Figure 6a vs. Figure 6c), often in close proximity, within a single thin section. The “normal” zoning is highlighted by a Ti decrease (and possibly Al, Ge increase) outwards, which results in an outward decrease in CL intensity, conformably with the general evolution of magmatic quartz and fractionated magma as a whole. The explanation of the opposite trend, i.e., an increase in Ti content in crystal rims, is a complex one. In volcanic rocks, Ti content can increase either due to a temperature increase after replenishment of the reservoir with new hot magma or due to a pressure drop during an eruption [11,12,13]. In granites, abrupt changes in pressure and temperature are unlikely, but an increase in Ti content can still occur, as evidenced by examples from Madeira (Figure 6c) and Khalzan Buregte (Figure 6d). Here we can assume that the increase in Ti activity and its ability to enter the quartz lattice is supported by the gradual increase in the content of volatile elements in the magma.

A strong positive correlation between Al and Ti in Khan Bogd granite, indicated already by Figure 3a, is also confirmed within the zoning of individual crystals (Figure 6a,b). Li and Ge correlated generally positively with Ti in quartz from Khan Bogd but varied chaotically at other localities (Figure 6c,d).

5. Discussion

5.1. Evolution of Individual Magmatic Systems

Concentrations of Ti, Al and Ge and their correlations offer useful information about the evolution of magmatic systems (Figure 3, Figure 4 and Figure 5). At Khan Bogd, a closer look indicates a general positive correlation between Ti and Al (Figure 3a), and a chaotic distribution of Ti vs. Ge values (Figure 5a) in quartz from all granite samples from the main intrusive phase. This indicates that during crystallization of the main body, no distinct fractionation of elemental pairs Ti-Al and Ti-Ge occurred. While, according to Huang and Audétat [51], Ti concentrations in quartz depend strongly on the quartz growth rate, the joint increase in Al and Ti content can be interpreted as a manifestation of accelerated quartz crystallization during the cooling of the magma after intrusion. The older suite of Khalzan Buregte (samples from the phase 2 and 3) evolved via Ti depletion, while neither Al nor Ge was enriched. Thus, the temperature during crystallization decreased, but the magma was unable to fractionate the trace elements effectively. In the case of Ivigtut, a major episode of trace-element enrichment was connected with magmatic–hydrothermal transition from the oldest Ti-rich hypersolvus granite to late UST-textured dikes and metasomatites. In contrast, Khalzan Buregtey phases 5 + 7, and Madeira albite granite + cryolite pegmatite represent already strongly evolved melts, initially poor in Ti, capable of effectively fractionating trace elements, in this case represented by Ge.

5.2. Comparison of Quartz from Peralkaline Granitoids with Quartz of Other Granite Types

The basic chemical difference between peralkaline granitoids and other types of rare-metal granites (subaluminous A-type and peraluminous S-type) is in the aluminum content, i.e., the content of “free” aluminum that is not bound into the feldspar structure. The low activity of Al in peralkaline melts is also reflected in its limited entry into quartz (Figure 7a).

While quartz from peraluminous S-type granites usually contains 285–1090 ppm Al (median 448 ppm), and quartz from subaluminous A-type granites and rhyolites contains 51–342 ppm Al (median 160 ppm), magmatic quartz from peralkaline granitoids commonly contains only 8–163 ppm Al (median 42 ppm Al). All the mentioned data intervals represent 90% of analyses after omitting the 5% lowest and 5% highest values [7,8,9,11,31,52,53]. The scarce higher values from Ivigtut represent associated metasomatites and these from the younger suite of Khalzan Buregte indicate a transition to the pegmatitic stage.

Common values of Ti in quartz (90% of measured values) fall within the intervals of <0.05–102 ppm, 0.18–90.3 ppm and 1.26–77 ppm; the medians are 15.2 ppm, 6.72 ppm and 8.23 ppm in S-type granites, A-type granites and rhyolites and peralkaline granitoids, respectively, i.e., differences among rock types are, in the case of Ti, minimal (Figure 7a).

Because Li enters the quartz lattice practically exclusively as an interstitial charge-balancing element following the equation Al³⁺ + Li⁺ = Si⁴⁺, the maximum Li enrichment is limited by the Al content: max. 174 ppm Li in quartz from peraluminous S-type granites, max. 65 ppm Li in subaluminous A-type granites and not more than 73 ppm Li in quartz from peralkaline granitoids. The common ranges (90% of values) here are 8–123, 0.2–36 and <0.01–18.2 ppm Li, respectively (Figure 7b).

Ge becomes enriched in relatively late stages of magmatic fractionation and reaches similar maximal values in quartz from all the studied rock types: 13.5 ppm in S-type granites, 7.5 ppm Ge in A-type granites and 10.6 ppm Ge in peralkaline granitoids (Madeira pluton). The common ranges of Ge are 0.6–5.8 ppm, 0.6–3.6 ppm and 0.05–5.7 ppm in S-type, A-type and peralkaline granitoids, respectively. Even higher values, up to 11 ppm Ge, were found in quartz from the Madeira cryolite pegmatite, but similar values have also been reported in the literature from peraluminous pegmatites (max. 18 ppm Ge [16]). Overall, a negative Ti-Ge correlation well illustrates pronounced magmatic fractionation (Figure 7c).

6. Conclusions

The main chemical features of quartz from peralkaline granitoids can be summarized as follows:

• Quartz from peralkaline granitoids is statistically poorer in Al and Li (medians 42 and 2.4 ppm, respectively) in comparison with quartz from peraluminous S-type granites (medians 448 and 37 ppm, respectively) and subaluminous A-type granites (medians 160 and 15 ppm, respectively). Content of Ti and Ge is similar in all three groups of granitoids (medians in peralkaline quartz 8.2 and 0.98 ppm, respectively).
• Quartz from large homogeneous bodies of riebeckite-bearing granitoids (Khan Bogd, older suite of Khalzan Buregte, Europa) shows no Ti/Ge fractionation and displays either a positive Ti–Al correlation (Khan Bogd) or no Ti–Al correlation.
• Quartz from more evolved late intrusive phases (5–7th phase of Khalzan Buregte, late granite/pegmatite Ivigtut) and the strongly fractionated Madeira pluton is characterized by a moderate enrichment in Al and strong increase in the Ge/Ti ratio (Figure 3 and Figure 5), i.e., features already well-known from rare-metal granites of both A- and S-type affiliation [3,11,16,17,18,49]. This can be used as an auxiliary indicator in exploration work for granite-related mineral deposits.
• A statistically significant dataset confirmed that quartz from the three basic geochemical types of granitoids differs significantly in Al and Li content, while the Ti and Ge content are similar. However, even in the case of Al and Li, there is a wide interval of overlap in the content across all rock types. This limits the practical applicability for analysis of quartz’s provenance [29,54].