1.4 Background geography
2.2.6 Electron probe micro-analyzer (EPMA)
Major and minor element analysis of single glass shards and titanomagnetite phenocrysts were carried out using a JEOL JXA-840a electron microprobe equipped with a Princeton Gamma Tech Prism 2000 Si(Li) energy dispersive spectrometer (EDS) at Massey University. An accelerating voltage of 15 kV, an 800 pA beam-current, and a 100 seconds acquisition time were used. These conditions along with a defocused beam of 10-20 µm in diameter were used for glass shard analyses to minimize Na-loss. A focused beam (approximately 2-3 µm diameter) was used for titanomagnetite analyses. Titanomagnetite phenocrysts and/or those within glomerocrysts were analysed mainly from the 250-500 µm size fraction. The energy spectrum at the analytical conditions was regularly calibrated using ASTIMEX albite and olivine standards for glass shards, and spinel and rutile for the titanomagnetites. Glass is the most commonly analysed phase for tephrochronology, but titanomagnetite has also been successfully used at many andesitic and rhyolitic volcanoes as discussed in other parts of this thesis. At least 10 analyses per phase per tephra were obtained. Major elements are expressed as oxides and reported in weight per cent (wt%). Glass shard data were normalised to 100% and titanomagnetite data corrected for iron following the procedure of Carmichael (1966). The titanomagnetite and glass shard data (mean ±S.D.) is given in Appendix 2 and 3 (lake-and-peat tephras) and Appendix 5 and 6 (proximal-and-medial tephra units) and as complete data set within the electronic appendices.
Detection limits for electron microprobe analysis are given in Table 2.2. As part of an inter-laboratory EMPA project (64 participating laboratories), basaltic glass was repeatedly analysed and deviations from its accepted composition are listed in Table 2.2 (R. Sims, unpubl. data).
Table 2. 2 Detection limits of element oxides measured at the electron microprobe (University of Auckland) including the deviation from a reference glass composition.
1σ (wt%) 3σ (wt%)
International Round-Robin Inter-lab EPMA project (64 participating labs): Difference to reference values of refused basaltic glass composition
SiO2 0.11 0.32 -0.18 TiO2 0.08 0.23 0.12 Al2O3 0.06 0.18 -0.05 FeO 0.07 0.20 -0.04 MnO 0.07 0.20 -0.01 MgO 0.07 0.21 0.17 CaO 0.04 0.13 -0.02 Na2O 0.11 0.32 0.05 K2O 0.03 0.10 0.02 P2O5 0.07 0.20 0.01 SO3 0.06 0.19 Cl 0.03 0.08 Cr2O3 0.06 0.17 NiO 0.10 0.30
2.2.6.1 Limitations
EPMA is an effective analysis tool, but there are several factors that may influence measured compositions. Some of these include:
1) Instrument failure and instrumental drift, which is monitored for and corrected by repeated standard analyses.
2) Poor polish qualities can lead to a lack of sufficient points to analyse or coating of grains, which can be addressed by re-polishing and re-analysis.
3) Small grains, microlites/impurities within grains and incorrect positioning of the beam, which are usually noted in the analytical totals.
4) Andesitic glass shards are often highly vesicular, and the vesicles are thin- walled, so it can be difficult to find an area >10-20 µm to analyse by a defocused beam.
5) Syn- and post-eruption crystallisation of nanolitic Fe-oxides and microlitic plagioclase often lead to fine-scale heterogeneity of andesitic glass (Best, 2003; Platz et al., 2007). These are difficult to avoid in the large-scale beam areas required for glass analysis.
6) Andesitic tephra hydrates and alters rapidly to allophanic clay and other phases (Lowe, 1986; Alloway et al., 1995), which can influence the major element composition of volcanic glass, resulting in low SiO2 and high Al2O3 contents
(Neall, 1977).
7) Exsolved titanomagnetite grains (trellises of ilmenite and magnetite) can lead to heterogeneous titanomagnetite compositions (Turner et al., 2008a).
To avoid most of the listed issues above, EPMA analysis were constantly monitored and the resulting data carefully screened. To minimise the effects of glass and titanomagnetite contamination back-scatter electron imaging and the removal of hybrid data points using the technique of Platz et al. (2007) were applied.
2.2.6.2 Geochemical outliers
Some compositions clearly stood out from the main geochemical groupings and could not be accounted for by the analytical issues described above. These could reflect any of the following factors:
Chemical variation in the magmatic system
1) Assimilation/contamination of xenocrystic or autocrystic titanomagnetites from geochemically different magma reservoirs, the conduit wall, or dome structure as magmas rises and explosively erupts (Stewart et al., 1998).
2) Magma mingling and/or mixing as described by Turner et al. (2008b) based on contrasting titanomagnetite morphologies and Ti-variations in the same eruptive unit. Magma mingling is also observed as banded pumices and zoned phenocrysts (Platz et al., 2007) and in whole-rock major and trace analyses of the general Mt. Taranaki rock suite (Price et al., 1999).
3) Magma recharge episodes may inject geochemically different titanomagnetites from a deeper and hotter magma source to an already existing magma body. Turner et al. (2008b) propose that the upper andesitic magma system of Mt. Taranaki was recharged by magma with similar properties creating cryptic magma mixing results.
4) Titanomagnetite exsolution caused by variations in pressure/temperature and oxidation states in the upper conduit/dome system as described by Turner et al. (2008a). They suggest that magmas that ascended slowly have heterogeneous titanomagnetite compositions while those of magmas that ascended quickly are more homogeneous. Exsolution lamellae are enriched in TiO2 (up to 30 wt%)
compared to the titanomagnetite host (3-6 wt% TiO2). The scale and degree of
the oxidation-induced exsolution may vary greatly (e.g., Haggerty, 1991) and may not always be visible under the electron microprobe leading to heterogeneous compositions.
Contamination due to post-depositional processes and during sample handling
1) Post-depositional processes may cause mixing of tephras and addition of other material from wind or water-borne sources (e.g., Green and Lowe, 1985; de Fontaine et al., 2007). Mixing may also be caused by bioturbation of plant roots and aquatic animals (e.g., Payne and Gehrels, 2010; Payne et al., 2005).
2) Low sedimentation rates and short-intervals between eruptions may also result in co-deposition of tephra with little or no intervening sediment. Very thin, closely spaced tephras are also difficult to extract without cross-contamination. 3) Contamination of samples may also occur during drilling, as material is dragged
along the inner walls of the core-liner. Radiocarbon/tephra samples were thus only sampled from the centre of the sediment cores, without material contacting the liner.