Methodological approach

The subsequent section presents a general overview of the methods applied during this study.

Detailed descriptions of analytical procedures are given in the following chapters (2 ̶ 6) of this thesis.

Framework petrography (‘Light mineral’ analysis)

Petrographic investigations of thin sections have a long tradition and are a basic tool in sedimentary research (Weltje and von Eynatten, 2004). By definition, framework grains (‘light minerals’) have a specific density < 2.89 g/cm³ (Boenigk, 1983), and commonly represent more than 90% of the total content of sands and sandstones. Even though the examination of framework components is rather time-consuming and results often less informative compared to other methods, e.g., heavy mineral or single grain analysis, the qualitative and quantitative analysis of framework grains can give important information on the origin and history of clastic sediments. Common applications include the classification of sediments in ternary diagrams (QFL, QmFLt, e.g., McBride, 1963, Garzanti, 2016), determination of the tectonic setting (arc, orogeny, craton, recycled) of the sediment source (e.g., Dickinson, 1985; Dickinson and Suczek, 1979), and reconstruction of the diagenetic history (Gaupp, 1996). Detailed investigations of the lithic grains allow more specific characterization of the provenance. Ideally, the texture and mineral parageneses of rock fragments give first-hand information on the source rock (low-grade metamorphic, high-grade metamorphic, sedimentary) and might even allow reconstruction of e.g., metamorphic temperature-pressure-time paths. Due to the decomposition of rock fragments in finer grain size fractions, this, of course, requires grain sizes big enough to still carry the full suite of coexisting minerals, which usually is not the case in grain size fractions normally used for microscopic investigations. Quantification of the modal composition is routinely achieved by counting a representative amount of grains per thin section, usually performed by using an electronic point-counter with mechanical stage.

Heavy mineral studies

Heavy minerals are useful indicators for source rock determination, because, unlike quartz and feldspar, their composition and parageneses strongly depends on the chemistry and tectono-stratigraphic level of their protosource. Per definition, heavy minerals have a specific density of > 2.85 g/cm³ (Boenigk, 1983). Although some heavy minerals (e.g., amphibole, pyroxene) are rock-forming

in magmatic or metamorphic rocks, in clastic sediments they usually occur as accessories with proportions less than 1%. In many cases, the heavy mineral association in sediments does not simply reflect the original assemblage of the provenance, but might be modified during several external processes, described in the section above (1.5.1), whereas hydraulics and burial diagenesis are believed to be the most crucial controlling factors (Morton and Hallsworth, 1999, 2007). Chemical dissolution during weathering or diagenesis may lead to loss of instable minerals (e.g., olivine, pyroxene) and enrichment of chemically and mechanically (ultra)stable components, such as zircon, tourmaline and rutile, during the sedimentary cycle (Morton, 1985; Velbel, 2007). Mineral sorting during transport and dispersal may lead to segregation of different mineral types or even within the same mineral group due to a different hydraulic behavior. Generally, changes in the provenance signal in overprinted sediments can be obtained by: (i) following the occurrence of diagnostic stable minerals (e.g., spinel, e.g., von Eynatten and Gaupp, 1999), (ii) comparing the ratio of two stable minerals (e.g., apatite/tourmaline or garnet/zircon; Morton and Hallsworth, 1994), and (iii) examining changes in the chemistry of single grains of a certain mineral group (= varietal studies; e.g., garnet, rutile; e.g., Morton, 1985; Zack et al., 2004 a, b). The effect of physical sorting that causes deposition of minerals with the same size and shape in different size fractions due to differences in their hydraulic behavior can be minimized by analyzing a specific grain size interval (Morton and Hallsworth, 1994).

Chemical analysis of heavy minerals (garnet and rutile)

Single-grain chemical analysis by microbeam techniques allows investigating the major and trace element composition of minerals from a certain mineral group and the variability among these grains (Mange and Morton, 2007). Because every rock suite has specific elemental characteristics that are also inherited by the minerals that originate from this source, distinct elemental signatures or variations within the group can be used to trace back their protosource. Chemical analysis can be applied to a variety of minerals, such as feldspar (e.g., Trevena and Nash, 1981), pyroxene (e.g., Krawinkel et al., 1999; Pinto et al., 2004), tourmaline (e.g., Henry and Guidotti, 1985; Morton et al., 2005), or opaque minerals (e.g., Basu and Molinaroli, 1989, 1991; Amini and Anketell, 2015). The most often used constituents are those that occur in a variety of rock suites and are widespread in siliciclastic rocks. In this thesis, garnet and rutile are chosen for microprobe investigations; both minerals have been proven as reliable index minerals in multiple provenance studies.

Garnet

Due to its many compositional variations with six principal end-member compositions (pyrope, almandine, spessartine, grossular, andradite, uvarovite), garnet is a widely used indicator mineral in provenance studies, that can give distinct information about the source area of siliciclastic rocks (e.g., Morton, 1985, 1987; Aubrecht et al. 2009). The chemistry of garnet depends both on the composition of the parental rock and on the P-T conditions during formation. Besides its chemical variability, other advantages that make garnet a suitable fingerprinting mineral is its relative stability during long distance fluvial-deltaic transport and its resistance against chemical modification during diagenesis or low-grade metamorphism (Morton, 1984; Hutchison and Oliver, 1998). However, garnet responds sensitive to acidic environments, in which it might undergo secondary alteration processes, like chemical modification until total dissolution (Morton, 1984). Dissolution of garnet in East African modern river sands has been described by Andò et al. (2012) and Garzanti et al. (2013), who found that garnet weathers out much faster than hornblende in wet and humid equatorial environments. The stability of garnet depends on the chemical composition with the most Ca-rich species becoming instable first (Morton, 1987).

Figure 1-11. Schematized work flow of this thesis.

The general chemical formula of the garnet group is X3Y2Si3O12. Possible substitutions for X are Fe²⁺, Ca²⁺, Mg²⁺, and Mn²⁺, and the Y position can be occupied by either Al³⁺, Fe³⁺, and Cr³⁺. The most common garnet species are almandine (Fe3Al2Si3O12), pyrope (Mg3Al2Si3O12), spessartine (Mn3Al2Si3O12), grossular (Ca3Al2Si3O12), andradite (Ca3(Fe,Ti)2Si3O12), and uvarovite (Ca3Cr2Si3O12).

In nature, garnet with a chemical composition corresponding to any end-member is rare. They are usually a solid solution of these principal end-member compositions in highly diverse proportions (Deer et al. 1992). Garnet is characteristic for a wide range of metamorphic rocks, but may also originate from granites, pegmatites, acid volcanic rocks, skarns and mantle rocks (Mange and Maurer, 1992).

Rutile chemistry and thermometry

For many reasons rutile became an important indicator mineral in sedimentary provenance studies (summary in Meinhold, 2010). Rutile crystallizes in a wide range of rock suites (predominantly in greenschist- to granulite-facies metamorphic rocks) and, because of its chemical and mechanical stability during weathering, transport and burial diagenesis (e.g. Morton and Hallsworth, 1999), rutile is a common accessory mineral even in very mature clastic sediments and sedimentary rocks, in which many diagnostic, but unstable minerals, are already lost.

Rutile is the high temperature polymorph of TiO2. It is not only a major source for titanium dioxide, but it also incorporates a large variety of trace elements, such as Al, Cr, Fe, Hf, Mo, Nb, Sb, Sn, Ta, Th, U, V, W, Zr, which substitute for Ti in the crystal lattice (e.g., Deer et al., 1992; Zack et al., 2002).

Variations in trace elements are dependent on the composition of the source lithology and might therefore be used as fingerprint of chemical and physical conditions in the host rock during rutile formation. Previous studies on the application of detrital rutile geochemistry by, e.g., Zack et al.

(2002), Zack et al. (2004a, b), Watson et al. (2006), Tomkins et al. (2007), Triebold et al. (2007, 2012), Meinhold et al. (2008) led to two principal conclusions: (1) the content of Nb and Cr allows to differentiate between the two main sources for rutile - metapelitic and metamafic rocks, and (2) the incorporation of zircon in the rutile lattice gives information about the maximum metamorphic temperatures during rutile crystallization, i.e., rutile serves as a geothermometer.

Cr-Nb chemistry of rutile

On the basis of Cr and Nb contents, the two major host lithologies for rutile, metamafic (e.g., mafic granulite, eclogite) and metapelitic (e.g., felsic granulite, paragneiss, mica-shist) rocks, can be distinguished (Zack et al., 2002; Zack et al., 2004b; Triebold et al., 2007, 2012; Meinhold et al., 2008).

Zack et al. (2002) were the first to describe that the Ti/Cr and Ti/Nb ratio in the source rock is also mirrored by associated rutile. Based on their studies, metapelitic rocks contain high Nb, but low Cr contents, while, both, low Cr and low Nb values or low Nb, but high Cr concentrations, respectively, are characteristic for metamafic rocks. Discrimination fields using the minimum and maximum Nb and Cr values were introduced. In recent studies, several modifications of the discrimination line between the fields for metapelitic and metamafic rutile have been proposed (Triebold et al., 2007;

Meinhold et al., 2008). The latest known refinement was presented by Triebold et al. (2012).

Zr-in-rutile thermometry

Zr-in-rutile thermometry bases on the assumption that the incorporation of zirconium (Zr) into rutile is temperature dependent (Zack et al., 2004b; Watson et al., 2006; Tomkins et al., 2007), with the requirement that zircon and quartz were coexisting phases during rutile formation (Zack et al., 2004b).

The first to introduce rutile as a geothermometer were Zack et al. (2004b), who proposed a temperature calculation that bases on the empirical study of rutile from 31 natural metamorphic rock samples covering a wide range of temperature conditions. Further studies on Zr-in-rutile thermometry (Watson et al., 2006; Ferry and Watson, 2007; Tomkins et al., 2007) are based on experimental data.

Watson et al. (2006) suggested that pressure conditions during metamorphosis play a key role on the Zr content of rutile, and Ferry and Watson (2007) emphasize the importance of SiO2 activity. The latest refinement of the thermometer calculation was given by Tomkins et al. (2007), who introduced three equations that do not only take pressure, but also the polymorph type of coexisting quartz into account. However, when dealing with detrital rutile, pressure conditions during metamorphosis of the source rock are commonly unknown. Therefore, in studies lacking any information about pressure, a default setting (10 kbar, α-quartz) should be applied (Triebold et al., 2012).

Bulk rock geochemistry

Geochemical analyses of the whole rock composition serve as a tool to discriminate, classify and categorize rocks according to their major and trace element inventory. In contrast to other methods, e.g., petrographic investigations, whole rock geochemistry has the advantage that it can be applied to all grain sizes and that a large number of variables can be processed within a relatively short period of time (Weltje and von Eynatten, 2004). However, geochemical analyses do not distinguish between detrital and authigenic minerals, nor do they give any information about mineral parageneses or textures of grains, which might be relevant for the interpretation of the source rock/area. Nevertheless, when combined with other techniques, particularly optical observations, bulk rock geochemistry can give important insights into the evolution of sediments.

The main controlling factor on the geochemical composition of sediments is the geochemical character of the parental rock. Different lithologies and tectonic settings show systematic mineralogical and thus chemical compositional variations, e.g., some elements are sensitive indicators of either felsic (e.g., Th, U, Zr, Hf, Nb, Ta, Y) or mafic rocks (e.g., Ni, Co, Cr, V, Cu; Bhatia and Crook, 1986; Cullers, 2000; Lee, 2002). However, secondary processes, like weathering, diagenesis, sorting and metamorphism, can affect and modify the initial chemical signature. Chemical weathering results in selective mobilization and leaching of elements, leading to either a loss or enrichment of certain elements in the sediment. Especially, alkaline and alkaline-earth elements might be highly fractionated due to their mobility during weathering and diagenesis. Mobile elements with small ionic radii (Ca, Na, Sr) are selectively removed from the weathered system, while elements with large ionic radii, like Cs, Ba and Rb, are assumed to remain within the sediment as they are adsorbed to secondary (clay) minerals (Nesbitt et al., 1980). Fractionation of sedimentary grains by means of their size, shape, density, mechanic stability etc. leads to enrichment of certain elements in different grain size fractions.

Coarser grain size classes are characterized by increased Si due to a higher content of quartz, whereas other major elements, such as Al, Mg and Fe remain more abundant in the mud fraction, because of their affiliation to clay minerals, phyllosilicates and oxyhydroxides (Garzanti et al., 2011). The highest concentration of trace elements and REE is present in clay minerals and heavy minerals and therefore depends on their distribution (Rollinson, 1993). Zr and Hf which are both highly associated with zircon are enriched in the finest tail of the sand mode, where ultradense zircon usually accumulates (Garzanti et al., 2011). After deposition, chemical modifications are mostly dependent on the pH value

and the redox potential of the depositional environment. Some element ratios, like Fe²⁺/Fe³⁺, V/Cr and Ni/Co can provide information about oxidative or reductive conditions (Mingram, 1995). Generally, elements, such as La, Ce, Nd, Y, Th, Zr, Hf, Nb, Ti and Sc are described to be most suitable for provenance reconstruction (Bathia and Crook, 1986), because they are relatively immobile within the sedimentary cycle and are directly transported into the sediment (McLennan et al., 1983). According to McLennan and Taylor (1991), the most reliable constituents are the REE, Th and Sc, because they are not influenced by any modification processes, whereas Zr, Hf and Ti are stronger influenced by heavy mineral fractionation.

Zircon U-Pb geochronology

In the last decades, rapid isotope geochronology by laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) became one of the principle tools in unravelling the geological history of sediments and their source regions. This is because this method provides precise age determinations of large populations from a single sediment sample in a relatively short time period. High-U minerals that can be utilized for this method include zircon, monazite, apatite, xenotime, titanite, rutile, baddeleyite, allanite, and perovskite. Of all of these geochronometers, in situ U-Pb dating of zircon is, however, widely applied as the method of choice, likely due to its many advantages (Corfu et al., 2003). Zircon is present in a variety of lithologies, and is ubiquitous in sediments, because of its high resistance to physical and chemical weathering, and metamorphic processes. Due to its robustness against external factors, zircon crystals behave like timecapsules carrying the history of their proto-sources.

Furthermore, zircon has a high closure temperature of approximately 900 °C (Pilot et al., 1998), presenting a valuable measure for the timing of magma crystallization. Thus, the age information obtained for a single grain can be linked to a certain crystallization event in the source area, i.e., the age directly reflects the age of a sediment’s protosource. Besides provenance and palaeogeographic reconstructions, most common applications of geochronological methods are stratigraphic correlations, to assess the timing of magmatic and metamorphic events, and to infer the maximal age of sediment deposition (summarized in Gehrels, 2014). Sedimentary rocks usually contain several components from multiple crystallization cycles as result of (1) various magmatic cycles within the same locality, (2) the adjacent occurrence of rocks with different ages as consequence of tectonic processes, (3) the mixing of detritus from various localities during transport, and/or (4) the mixing of detritus from primary source(s) with older recycled sedimentary rocks (Thomas, 2011). Especially, the latter process provides a challenge in provenance studies, because zircon can be reworked during multiple sedimentary cycles due to its durability.

U-Th-Pb geochronology bases on the fact that multiple parent isotopes decay to different stable isotopes of Pb, each with a different half-life (Fig. 1-12). The element lead has four naturally occurring stable isotopes, ²⁰⁴Pb, ²⁰⁶Pb, ²⁰⁷Pb, and ²⁰⁸Pb, of which the latter three are produced through the independent radiogenic decay of ²³⁸U, ²³⁵U, and ²³²Th, respectively. The isotopic decay does not lead directly to Pb, but traverses several alpha and beta decays during which a series of intermediate daughter isotopes are created before reaching the stable isotope of Pb.

The age of a zircon crystal is calculated by the ratio of the initial isotopes and their daughter products.

For each of the three independent decay systems, an isochron equation is expressed:

Ideally, all three equations result in similar ages. In this case, the obtained age data are concordant, and data interpretation can be seen as reliable. In reality, however, the three independent decay systems usually show deviation between calculated ages, i.e., the obtained ages are discordant.

Commonly, discordancy of zircon ages results from two main processes: (1) the impact of inheritance, i.e., complex zircon grains have several age domains due to multiple individual growth phases, and (2) disturbance of the U-Pb system after crystallization that leads to loss of the relatively mobile Pb when in contact with hydrothermal fluids (Gehrels, 2014). The decision whether to include or exclude discordant analyses for data interpretation is still controversially discussed between geochronologists and largely depends on the aim of the study (Gehrels, 2014).

Figure 1-12. Radioactive decay chain of the U–Th–Pb system. Parent isotopes and stable daughter isotopes of Pb are outlined in red. α = alpha particle, β = beta particle, Q = energy released during the decay (after Schoene, 2013).