Petrogenesis of Rinjani Post-1257-Caldera-Forming-Eruption Lava Flows

DOI:10.17014/ijog.3.2.107-126After the catastrophic 1257 caldera-forming eruption, a new chapter of Old Rinjani volcanic activity beganwith the appearance of Rombongan and Barujari Volcanoes within the caldera. However, no published petrogeneticstudy focuses mainly on these products. The Rombongan eruption in 1944 and Barujari eruptions in pre-1944, 1966,1994, 2004, and 2009 produced basaltic andesite pyroclastic materials and lava flows. A total of thirty-one sampleswere analyzed, including six samples for each period of eruption except from 2004 (only one sample). The sampleswere used for petrography, whole-rock geochemistry, and trace and rare earth element analyses. The Rombonganand Barujari lavas are composed of calc-alkaline and high K calc-alkaline porphyritic basaltic andesite. The magmashows narrow variation of SiO2 content that implies small changes during its generation. The magma that formedRombongan and Barujari lavas is island-arc alkaline basalt. Generally, data show that the rocks are enriched in LargeIon Lithophile Elements (LILE: K, Rb, Ba, Sr, and Ba) and depleted in High Field Strength Elements (HFSE: Y, Ti,and Nb) which are typically a suite from a subduction zone. The pattern shows a medium enrichment in Light REEand relatively depleted in Heavy REE. The processes are dominantly controlled by fractional crystallization andmagma mixing. All of the Barujari and Rombongan lavas would have been produced by the same source of magmawith little variation in composition caused by host rock filter process. New flux of magma would likely have occurredfrom pre-1944 until 2009 period that indicates slightly decrease and increase of SiO2 content. The Rombongan andBarujari lava generations show an arc magma differentiation trend.


Introduction
Rinjani volcano complex lies in Lombok Island, West Nusa Tenggara Province, Indonesia ( Figure 1). The Lombok Island is tectonically located in the west of eastern most Sunda Arc where Australian Continental Plate subductes beneath Eurasian Plate. The crustal thickness is about 20 km (Curray et al., 1977). The Benioff-Wadati zone lies about 164 km beneath Rinjani (Nasution et al., 2010). The Rinjani calc-alkaline suite probably originated by partial melting of the peridotite mantle wedge overlying the active part of Benioff Zone beneath Lombok with The Syn-Caldera stage relates to the catastrophic eruption that destroyed the Old Rinjani and led to the formation of the so-called Segara Anak Caldera. The tephra are pumice fall deposits, pumiceous pyroclastic-flow deposits, and debris-flow deposits. Nasution et al. (2004) argued that this catastrophic eruption occurred between AD 1210 and 1300 AD, whereas Lavigne et al. (2013) suggested that the eruption occurred between May and October 1257 AD, corresponding to the so-called "1258 mystery eruption" found in ice cores from both Greenland and Antarctica (Oppenheimer, 2003;Sigl et al., 2014Sigl et al., , 2015. The products are mainly characterized by dacite with SiO 2 content ranges from 62 to 63 wt % (Nasution et al., 2004).
The catastrophic eruption in 1257 AD (Lavigne et al., 2013) led to the formation of a 6 km x 7 km caldera. This makes Rinjani one of great calderaforming volcanoes in Indonesia. After this catastrophic caldera-forming eruption, there was an eruption from Young Rinjani that formed Segara Muncar Crater. Then, a new chapter of Rinjani volcanic activity began with the appearance of Rombongan and Barujari Volcanoes within the caldera (Rachmat, 2016) (Figures 2 and 3).  (Rachmat, 2016; not to scale).

Stage I (Pre-Caldera)
During the Pleistocene, the Old Rinjani or Samalas initially reach ~4000 m asl height. It was formed far before 12,000 years B.P.

Stage II (Pre-Caldera)
The Young Rinjani was formed in the eastern flank of the old Rinjani. between 11,940 ± 40 until 5990 ± 50 years B.P.
The Barujari concentrates the main postcaldera volcanic activity since the catastrophic event of 1257 AD. In 1944, the Rombongan Volcano appeared in the north of Barujari, but then its activity became extinct. Barujari erupted and spewed lava and scoria during its activities. The eruptions of Barujari Volcano consist of moderately explosive explosions and occasional lava flows (Komorowski et al., 2014). Barujari Volcano eruptions of 1884, 1904, 1906, 1909, and 1915 produced ash materials (Kusumadinata et al., 1979).
A recent study on Rombongan and Barujari lava flows shows the characteristic of basalticandesitic lava (Komorowski et al., 2014). The lava shows 53 -55 wt.% of SiO 2 content, i.e. a less evolved magma than the syn-caldera forming products with 61 -64 wt.% SiO 2 (Vidal et al., 2015). This less evolved characteristic of the Rombongan and Barujari lavas gives us informa-tion about the relatively true nature of magma source in this area. Hamilton (1974) discussed petrology and tectonic relationship in the Banda Arc Region. Volcanic rocks from the Banda arc islands are dominantly composed of mafic and intermediate calc-alkaline composition (andesite, basalt, and dacite) (Abbot and Chamalaun, 1981;van Bemmelen, 1949;Brouwer, 1942;Ehrat, 1928;Matrais, 1972;Neumann van Padang, 1951;Whitford et al., 1977). Rocks of the young volcanoes display an usual increase in the ratio of potassium to silicon with increasing distance to the Benioff zone beneath (Hatherton and Dickinson, 1969;Hutchison, 1976). Based on Foden and Varne (1980), the 87 Sr/ 86 Sr ratio from Rinjani is 0.70386 -0.70402 and among the lowest in Sunda Arc. This low 87 Sr/ 86 Sr ratio suggests that the crust beneath Rinjani is also among the thinnest in the Sunda Arc.

Petrogenetic Study of Rinjani Complex
Based on Foden and Varne (1980), Rinjani lava range varies from ankaramite and high-Al basalt to andesite and dacite. Futhermore, in 1981, Foden and Varne concluded that Rinjani  G calc-alkaline suite probably originated by partial melting of the peridotite mantle wedge overlying the active Benioff Zone beneath the Lombok Island. However, they also notified the existence of some variations in primary melts that represent a separate evolving line. This is particularly the case for the ankaramite and high-Al basalt suite.
Based on whole-rock geochemistry, Nakagawa et al. (2015) grouped the rock into two folders: (1) stratovolcano building and post-caldera group, and (2) low-activity and caldera-forming group. Rocks from the stratovolcano building group mainly consists of basaltic andesite with SiO 2 ranging from 44.8 to 63.7%, while those from post-caldera group are olivine-pyroxene andesites (SiO 2 ~55%). This distinction is based on distinct chemical trends in the SiO 2 diagram for major and trace elements. The ratio of incompatible elements and Sr isotope are also distinct. Based on this this criterion, Nakagawa et al. (2015) suggested that the dacitic magma from low-activity and caldera-forming group could not be produced by a crystalization differentation of basaltic magma, but by additional processes, such as crustal melting and/or assimilation-fractional crystallization processes.
Moreover, Nakagawa et al. (2015) also suggested that the magma from low-activity stages was not the preceding stage of syn-caldera stage in term of petrogenesis. This argument is based on the difference in Sr isotope ratio and many chemical trends of magma from both stages.
Petrologic and geochemical studies of Rinjani Volcano had been conducted by many authors. Most of these studies mainly focused on the activities before and during the Syn-Caldera stage (e.g. Nakagawa et al., 2015;Vidal et al., 2015). Another paper also discussed the petrogenetic in general term without correlation with Rinjani Activity stage (e.g. Foden and Varne, 1981). In this paper, petrogenesis of lava from Rinjani post-1257-caldera activity is mainly discussed.
The Rombongan and Barujari lava flows are a good subject for a petrogenetic study because: 1. There are no published research that focuses on these lava flows;

2.
Relative clear information about lava flow time frame will give us a good temporal variation analysis; 3. A study of those lava flows will provide an understanding about magmatic evolution after the caldera-forming eruption.

Sampling and Methods
The Rombongan eruption in 1944 and Barujari eruption in pre-1944Barujari eruption in pre- , 1966Barujari eruption in pre- , 1994Barujari eruption in pre- , 2004Barujari eruption in pre- , and 2009 produced basaltic andesite pyroclastic materials and lava flows (Nasution et al., 2010). Only the lava products from each period of eruption are collected as samples for analyses. A total of thirtyone samples were analyzed and each period of eruption is composed of six samples except from 2004 with only one sample ( Table 1).
The samples were used for petrography, whole-rock geochemistry, and Trace and Rare Earth Elements (REE) analyses (Table 1). The petrographic analysis was carried out using a Leica MPS 52 polarization microscope. The whole-rock geochemistry was determined at The Centre for Geological Survey (Geological Agency) Laboratory by X-ray fluorescence (XRF) analysis using ARL Advant XP+. Trace and Rare Earth Elements (REE) data were also determined at The Centre for Geological Survey (Geological Agency) Laboratory by Inductively Coupled Plasma-Mass Spectrometry (ICP-MS).

Petrography
Most of lavas are porphyritic basalt to porphyritic andesite. The phenocrysts volume ranges from 39 % to 66 %. The dominant phenocrysts are plagioclase and pyroxene for all lavas.    Results of whole-rock geochemistry analysis comprising major-, trace-, and rare-earth elements (REE) of Rinjani Post-caldera Lava are presented on Table 1.
On the basis of data plotted on the Le Bass diagram, products of Rinjani (Barujari and Rom-bongan Volcanoes) post-1257 caldera eruption are dominantly composed of basaltic-andesite type, except sample-1966 lava having an andesitic composition ( Figure 5). Furthermore, by plotting the data on Peccerillo and Taylor diagram (1976) (Figure 6), the rocks tend to fall under high K calc-alkaline affinity with minor calc-alkaline one. Therefore, the volcanic products of Rinjani can be classified as high K calc-alkaline to calcalkaline basaltic-andesite type.
The Harker variation diagram for major elements shown in Figure 7 a and c indicates positive relationships that increases in SiO 2 content tend to be associated with increases in K 2 O and Na 2 O Figure 5. Plotting total alkali silica (Na 2 O+K 2 O) vs. SiO 2 using Le Bass diagram (Le Bass, 1986;Rollinson, 1993;Manullang et al., 2015) showing rock classification of the Rombongan and Barujari Volcanoes. The sample which shows a notable value is Lava 1966-06, displaying very high SiO 2 content (>58.88 wt.%) relative to the others. This value is also supported by the trace and rare earth elements values, which are also relatively high in the same samples.

Trace Elements
The trace element data were plotted into a spider diagram using Wood's chondrite normalization ( Figure 8). Generally, data show that the rocks are enriched in Large Ion Lithophile Elements (LILE: K, Rb, Ba, Sr, and Ba) and depleted in High Field Strength Elements (HFSE: Y, Ti, and Nb) which is a typical suite from the subduction zone. Nb and P display steeply depleted values that show relatively nearly similar with Nb and P values from Mid Oceanic Ridge Basalt (MORB) and Oceanic Island Basalt (OIB).

Rare Earth Elements
REE were plotted into a spider diagram with Haskin chondrite normalization (Figure 9). Lavas from all periods show a relatively similar pattern. The pattern displays an enrichment in Light REE and relatively depleted in Heavy REE. Value of (La/Yb) N is between 3 and 4, which suggests a medium enrichment in Light REE. Sample of lava 2009-2 shows high Nm with 0.58 ppm, that is easily recognized in the REE pattern.
All of the samples show enrichment of Rb, K, Th, Ba, La, Ce, and Sr relative to those from Mid-Oceanic Ridge Basalt. The trend indicates a similar pattern with those from Oceanic Island Basalt and Upper Crust, but several elements relatively depleted, such as Nb (Figure 9).  Peccerillo and Taylor (1976) and Manullang et al. (2015).  Value of (Eu/Eu*) is 0.9, relating to a very low negative Eu anomaly. Hanson (1980) argued that the presence of feldspar would contribute to negative Eu anomaly in the melt, whereas the presence of pyroxene would contribute to positive Eu anomaly. This very low negative Eu anomaly is possible because of the co-presence of plagioclase and pyroxene during the melting processes or the minimum presence of both plagioclase and pyroxene.

Magma Mixing
Magma mixing is likely to have occurred during the generation of Barujari Lava. It can be demonstrated by the appearance of sieve texture, oscillatory zoned plagioclase, and resorption texture. The sieve texture is interpreted by some authors as the result of a mixing process (e.g. Nixon and Pearce, 1987).
The sieve texture coated by a relatively euhedral plagioclase can be found in 1994-6 lava (Figures 4b, 4d, 4f). The sieve texture is likely formed, at first as a consequence of mixing of two batches magma with different composition. As the crystallization continues after the mixing process, the growth of plagioclase also continued and coated the sieve-textured plagioclase.
The oscillatory zone plagioclase can be found in several samples (e.g. lava 1966-2, 2004-1, 2009-6). Oscillatory zoning usually occurs because of dynamic conditions of crystallization that lead to compositional changes. Another effect of new magma injection is, as stated before, resorption of already crystallized mineral in the magma chamber. The repeated injection of fresh basic magma into the chamber of differentiated magma could cause resorption of already crystallized plagioclase (Nixon and Pearce, 1987). The resorption texture can be found in 1966-1 lava where a plagioclase crystal was likely corroded by a hot magma.
The magma mixing process could also be implied by temporal variation within Barujari and Rombongan lavas. The MgO versus SiO 2 diagram shows that pre-1944 and 1994 lavas could possibly be a basic end-member and 1966 lava is the highest silica end-member. The 1944 lava lies between the two end members and could be a transition product. After the forming of the pre-1944 lava, it evolved to more silicic condition and produced basaltic andesite endmember lava (1944 and 1966). The 1994 lava is compositionally more basic and relatively similar to the pre-1944 lava. It could be an indication of new supply of magma after 1966 eruption that led to the formation of more basic 1994 lava.

Fractional Crystallization
Evidences of the occurrence of fractional crystallization process can be obtained from SiO 2 variation diagram and trace elements trend. Frost and Frost (2014) argued that suite of rocks related by fractional crystallization will form curved arrays, whereas those that related by magma mixing will form linear trend on the Harker diagram. The Harker diagram of TiO 2 and CaO versus SiO 2 of Rinjani post-caldera lava form curved arrays ( Figure 6). Therefore, it suggests that the fractional crystallization process likely occurred in the genesis of Rinjani post-caldera lava.
The relatively low Mg number (100*Mg/ Mg+Fe) (Table 1) suggested that lava from Barujari and Rombongan Volcanoes had undergone extensive modification (e.g. 1966(e.g. -6 lava : Mg number 19661966-2 = 0.16;1966-5 = 0.17;1966-6 = 0.16). These Mg numbers were lower than 0.28 which means the magma was not originated from the mantle. This argument is similar with the argument of Annen et al. (2006). This argument is also supported by Foden and Varne (1981) who argued that low Mg/Mg+Fe values and low Ni concentrations of many members of the Rinjani suite, particularly andesites and dacites, suggested that the mantle derived melts must have undergone an extensive modification as a result of fractional crystallization. Normal plagioclase zonings that occur in many basaltic andesitic of Barujari and Rombongan lavas also support the fractional crystallization processes. Frost and Frost (2014) argued that tholeiitic basalt and island-arc environment were similar in terms of major element composition, but they I J O G I J O G followed different differentiation trends. This can be illustrated using a plot of Fe tot /(Fe tot +Mg) versus SiO 2 ( Figure 10). categories, those are: 1) crust-level fractional crystallization of primary basaltic magma from partial melting of mantle peridote, 2) fractional crystallization of basaltic precursors accompanied by assimilation of crustal rocks, 3) mixing of basalt and melted sialic crust, 4) lower crustal melting, assimilation, storage, and homogenization, and 5) partial melting of subducted oceanic lithosphere.

Differentiational Trend
Trace elements patterns from all of the samples show a similar pattern. This would likely imply that the Barujari and Rombongan lavas were derived from the same source of magma. Very low (Eu/Eu*) anomaly (0.9) and the appearance of plagioclase and pyroxene in most of the samples would suggest that the magma have undergone plagioclase and pyroxene fractionation.
Partial melting of subducted oceanic lithosphere model would likely be impossible because of the relatively low contents of Mgnumber (<0.28). Annen et al. (2006) also argued that arc andesite with low Mg-number could not have been in a direct equilibrium with the mantle. Therefore, other processes would be involved during magma ascending.
High concentration of Al 2 O 3 (>16 wt %) and high content of Sr (> 228 ppm) suggest the rock was originated from the crust or there was a crust involvement during the generation of the rock. However, a relatively higher content of Y (> 16 ppm) also suggests another complex generation of the rock. High Al 2 O 3 contents suggest high content of plagioclase phenocryst (Crawford, 1987, in Hartono, 1997. Foden and Varne (1981) argued that the crustal involvement was not possible because of the thinness of the crust in the area. However, considering the spider diagram pattern that shows a relatively similar pattern with those form upper crusts, it would be possible that a small amount of crustal involvement had occurred during the generation. It would be possible that a density filter or "host rock filter" occur during the magma ascending and caused the low Mg content in the melt.  During differentiation, tholeiitic melts from oceanic environment become enriched in Fe relatives to Mg. This ferromagnesian silicates in the lavas become enriched in iron end member as differentiation progresses and late in the differentiation history the melts underwent silica enrichment. In contrast, arc magma suites show a strong enrichment in silica with increasing differentiation but ferromagnesian silicates show moderately increases in the Fe/(Fe+Mg) ratio. From this plot, it can be concluded that the Rombongan and Barujari lavas underwent an arc character differentiation where moderate ferromagnesian increase occurs.
The trend also implies that the differentiation trend moves toward the magnesian field.
It implies that iron enrichment was inhibited during the earlier stage of differentiation due to early crystallization of Fe-Ti oxides. This kind of differentiation trend leads to the presence of magnesian-rich pyroxene end member, such as diopside and hypherstene, in highest silica-end member rocks (e.g.1966 lava) in CIPW norm calculation (Table 2). Hartono (1997)

Conclusion
Based on the discussion above, it can be concluded that the Rombongan and Barujari lavas are composed of calc-alkaline and high K calc-alkaline porphyritic basaltic-andesite. Dominant phenocryst phases are plagioclase (andesine) and pyroxene (dominantly diopside and hypersthene). The magma shows a narrow variation of SiO 2 content that implies small changes. Magma that formed the Rombongan and Barujari lavas was an island-arc alkaline basalt. Generally, data show that the rocks are enriched in Large Ion Lithophile Elements I J O G I J O G (LILE: K, Rb, Ba, Sr, and Ba) and depleted in High Field Strength Elements (HFSE: Y, Ti, and Nb) which is a typical suite of a subduction zone. The pattern shows a medium enrichment in Light REE and relatively depleted in Heavy REE.
The generation processes are dominantly controlled by fractional crystallization and magma mixing processes. All of the Barujari and Rombongan lavas would be produced by the same source of magma with little variation in composition caused by a host rock filter process. New input of magma would likely have occurred after the 1966 eruption indicated by the formation of relatively basic lava. The arc magma differentiation trend to the magnesian field leads to the formation of magnesian-end-