Journal of Petrology Volume 42 Number 12 Pages 2303-2331 2001
© Oxford University Press 2001
Geochemical Evolution of Akagi Volcano, NE Japan: Implications for Interaction Between Island-arc Magma and Lower Crust, and Generation of Isotopically Various Magmas
KATSURA KOBAYASHI,* and
EIZO NAKAMURA
PHEASANT MEMORIAL LABORATORY OF GEOCHEMISTRY AND COSMOCHEMISTRY, INSTITUTE FOR STUDY OF THE EARTH’S INTERIOR, OKAYAMA UNIVERSITY AT MISASA, TOTTORI, 682-0193, JAPAN
Received
April 30, 2000;
Revised typescript accepted
June 10, 2001
 |
ABSTRACT
|
|---|
Major and trace element, and Sr, Nd and Pb isotopic compositions
were determined for whole-rock samples from the ‘isotopically
anomalous’ Akagi volcano in the volcanic front of the
NE Japan arc. Sr and Nd isotopic compositions of phenocrysts
were also analyzed together with their major and trace element
compositions. Compared with the other volcanoes from the volcanic
front, the whole-rock isotope compositions of Akagi show highly
enriched characteristics;
87Sr/
86Sr = 0·7060–0·7088,
Nd = -0·40 to -8·6, and
208Pb/
204Pb = 38·4–38·8.
The rare earth element (REE) patterns are characterized by heavy
REE (HREE) depletions with U-shaped patterns from middle REE
(MREE) to HREE, suggesting that amphibole fractionation was
induced by a reaction between clinopyroxene and H
2O-rich magma
in the lower crust. The integrated isotope and trace elements
systematics, and tectonic structure beneath Akagi volcano, suggest
that lower-crustal assimilation by the H
2O-rich primary magma
could have been affected by the double subduction of Philippine
Sea and Pacific oceanic plates. This double subduction could
have supplied larger amounts of water to the magma source region
in the wedge mantle than in the case of a single subduction
zone. Significant differences in isotopic compositions are observed
between phenocrysts and the coexisting melts. Such isotopic
disequilibrium may have resulted from magma mixing between an
isotopically depleted aphyric and an enriched porphyritic magma
in a shallow magma chamber. The geochemical characteristics
of these end-member magmas were retained in the lower crust,
despite differing extents of lower-crustal assimilation by the
H
2O-rich magmas.
KEY WORDS: Akagi volcano; H2O-rich magma; isotopic disequilibrium; lower-crustal assimilation; magma mixing
|
INTRODUCTION
|
|---|
Island-arc volcanism has been considered to result from partial
melting of the wedge mantle induced by addition of slab-derived
fluid-rich materials in subduction zones. This general model
is supported by many geochemical studies employing major and
trace element compositions, and radiogenic isotope systematics
of arc volcanic ejecta (e.g. Nakamura
et al., 1985
; Woodhead
& Fraser, 1985
; Tatsumi
et al., 1986
; Ishikawa & Nakamura,
1994
; Miller
et al., 1994
; Ryan
et al., 1995
; Shibata &
Nakamura, 1997
). These previous works mainly focused on the
source characteristics in relation to the evolution of the mantle
wedge in subduction zones. The magmatic evolution of individual
volcanoes, after partial melting of the wedge mantle, and in
shallow magma chambers has received less attention. Isotope
and trace element geochemistry has typically been restricted
to basaltic volcanic rocks, not the felsic rocks, to avoid the
complexities of processes such as crystal fractionation, magma
mixing and crustal assimilation. It is, however, essential to
characterize the shallower magmatic processes along with the
source characteristics. Magmas erupted at the Earth’s
surface possess original source-related characteristics considerably
modified by post-melting physico-chemical processes that may
obscure the source signature.
In this study, Akagi volcano in the NE Japan arc was investigated because of its distinctive isotopic composition relative to neighboring volcanoes from the volcanic front of the NE Japan arc. This arc is one of the ideal localities for discussing island-arc magmatism, because the structure of the subduction zone is well defined by abundant seismological data (e.g. Utsu, 1974; Yoshii, 1979; Zhao et al., 1990; Zhao, 1992). Moreover, systematic across-arc variations of chemical and isotopic compositions were correlated with the depth of the Wadati–Benioff Zone (Notsu, 1983; Nakamura et al., 1985; Sakuyama & Nesbitt, 1986; Shibata & Nakamura, 1997). These across-arc systematics have been explained by continuous dehydration of the subducting oceanic slab resulting in a continuous decrease in the slab component in the source region with increasing depth to the subducted slab (Shibata & Nakamura, 1997). Volcanoes in the northern part of the volcanic front show a relatively narrow range of 87Sr/86Sr ratio (0·7038–0·7045) (Notsu, 1983; Shibata & Nakamura, 1997). However, the Sr isotope composition in the southern part of the volcanic front increases to the south and has the highest 87Sr/86Sr ratio of 
0·7087 at Akagi volcano (Notsu, 1983; Kersting et al., 1996). According to Gust et al. (1997), such systematic isotopic change along the volcanic front in the NE Japan arc could be explained by the contribution of subcontinental lithospheric mantle with some crustal contamination based on Sr–Nd–Pb isotope systematics and trace element geochemistry. They also suggested that the primary magmas of the volcanoes located south of the Tanakura Tectonic Line might have been involved with Indian-Ocean-type mantle, although they pointed out the possibility of lower-crustal contamination in the petrogenesis of the felsic volcanic rocks. However, their data are limited for individual volcanoes (only one to four samples without petrological description), making it difficult to discriminate samples that might have been involved with crustal contamination. It is, therefore, risky to discuss their source materials in the mantle beneath individual volcanic centers based on previous studies with insufficient petrological description of samples or a systematic dataset for each volcano.
From such a point of view, we investigated Akagi volcano, measuring major and trace element compositions and multi-isotope compositions including Sr, Nd and Pb isotopes for whole-rock samples, to comprehensively understand the magmatic processes involved in the formation of the isotopically most anomalous volcano in the volcanic front of the NE Japan arc, and to assess the mechanisms of along-arc variation with respect to tectonic setting. In addition, the trace element, and Sr and Nd isotope compositions of phenocryst minerals were determined to further understand the shallow magma chamber processes and thereby elucidate the genesis of andesite magma in subduction zones.
Geological and petrological features of Akagi volcano
Akagi volcano (36°33'N, 139°12'E) is located in the southern part of the volcanic front of the NE Japan arc (Fig. 1). This area is characterized by the presence of many Quaternary volcanoes (e.g. Haruna, Hotaka, Nikko-Shirane volcano), and is one of the most volcanically active regions in Japan. As illustrated in Fig. 1, an unusual structure exists in the mantle wedge beneath the volcanic area including Akagi volcano; that is, a double subduction zone, in which the Philippine Sea plate is subducted into the wedge mantle above the Pacific plate (Ishida, 1991). The tip of the Philippine Sea plate is subducted to depth of 90 km just under Akagi volcano, and it probably adheres closely to the subducting Pacific plate, of which the interface depth is 110 km. Moreover, the depth of the Moho under Akagi volcano is 39 km, some 4 km deeper than is typical beneath the volcanic front of the NE Japan arc (Zhao et al., 1990; Zhao, 1992). These unusual tectonic conditions probably contribute to the formation of the distinctive geochemical characteristics of Akagi volcano.
Although Akagi volcano is known to be mostly Quaternary in age, the lack of systematic radiometric ages precludes a clear understanding of the commencement of volcanic activity. On the basis of stratigraphical relationships between Akagi volcano (Moriya, 1968) and the neighboring Komochi volcano (Iizuka, 1996), the activity of Akagi volcano is thought to have started before 350 ka, and lasted intermittently to 30 ka. The latter age is based on the only available radiometric age, determined by the 14C method by Koga (1981).
The volcanic history of Akagi volcano is divided into three main stages based on the eruption style: the older stratovolcano formation stage (O) characterized by the eruption of andesitic lava flow and ejecta of scoria, the younger stratovolcano formation stage (Y) characterized by pyroclastics without lava flows, and the central cone formation stage (Cc) (Moriya, 1968). The older stratovolcano formation stage (O) is further divided into three substages based on the volcanic stratigraphy: the early (Oe), middle (Om) and late (Ol) substages. The main volcanic stages and substages are geologically distinguishable. However, the eruptive sequence of volcanic ejecta within each stage is unclear based on stratigraphy. The present volcanic flank is mainly formed of pyroclastics of the Y stage and of secondary deposits. Lava flows can be observed in the caldera, which formed after the activity of the Y stage, and near the caldera wall.
Sample locality and description
To investigate the magmatic processes of Akagi volcano, volcanic ejecta were collected based on the volcano stratigraphy established by Moriya (1968). The sampling localities are presented in Fig. 1. Samples were picked from fresh parts of outcrops to minimize alteration effects, and examined under the petrographic microscope to confirm the absence of significant secondary alteration.
Petrographic observations indicate that the Akagi rock samples are highly porphyritic, with phenocrysts mainly consisting of plagioclase, hypersthene, augite and opaque minerals (see Table 1). Some of the felsic samples with SiO2 contents exceeding 58 wt % in the Ol substage and Y and Cc stages contain phenocrysts of hornblende instead of augite, even without quartz phenocrysts. Olivine phenocrysts are absent in the studied samples except for a sample AK1303, which contains an olivine pseudomorph surrounded by orthopyroxene. The mineral assemblage in the groundmass of the Oe and Om substages is plagioclase + pigeonite ± augite ± hypersthene. In contrast, samples from the Ol substage and the Y and Cc stage samples do not contain groundmass pigeonite.
Sample preparation
In the preparation of whole-rock samples for major, trace element and isotope analyses, chunks free from surface alteration were picked. These unaltered chunks were washed with Milli-Q water using an ultrasonic bath, and dried. Then, they were ground to fine powders with grain sizes under 200 mesh using a silicon nitride mortar.
To separate phenocrysts, rock powders obtained using a disk mill were sieved to collect materials with grain size ranging from 100 to 200 mesh, as the average grain size of phenocrysts of plagioclase, orthopyroxene, and clinopyroxene or hornblende is 500 µm. The sieved fraction was then processed to purify the minerals using conventional heavy liquid methods followed by an isodynamic separator. The mineral separation was finally accomplished by hand-picking, and the purity is regarded as being better than 99%.
Analytical methods
All geochemical analyses were carried out at the Pheasant Memorial Laboratory (PML), Institute for Study of the Earth’s Interior, Okayama University at Misasa. Major element compositions of whole rocks were determined by X-ray fluorescence (XRF) using a Philips PW-2400 system; details of the analytical procedure are described elsewhere (Takei et al., in preparation). Trace element analyses of the whole rocks were performed by inductively coupled plasma mass spectrometry (ICP-MS) using a Yokogawa PMS 2000 instrument with a flow-injection method developed by Makishima & Nakamura (1997). The analytical reproducibility for trace element analyses of andesitic samples was <5%, and typically 3% (Makishima & Nakamura, 1997).
The analytical procedures, including the chemical separations and mass spectrometry utilized in this study, are from Yoshikawa & Nakamura (1993), Makishima & Nakamura (1991a, 1991b) and Koide & Nakamura (1990) for Sr, Nd and Pb isotope analysis, respectively. Mass spectrometry was carried out on Finnigan MAT 261 and MAT 262 instruments equipped with five Faraday cups, applying a static-multi-collection mode. Normalizing factors used to correct isotopic fractionation of Sr and Nd are 86Sr/88Sr = 0·1194 and 146Nd/144Nd = 0·7219, respectively. The Pb isotope analyses were carried out with a nearly fixed filament temperature of 1200°C. Measured ratios of reference materials were 87Sr/86Sr = 0·710233 ± 8 (2
m) for 100 ng of NIST SRM 987, 143Nd/144Nd = 0·511818 ± 13 (2m) for 30 ng of La Jolla, and 206Pb/204Pb = 16·940 ± 15 (2m), 207Pb/204Pb = 15·494 ± 16 (2m) and 208Pb/204Pb = 36·709 ± 38 (2m) for 100 ng of NIST SRM 981.
Major element compositions of phenocrysts were determined by electron probe microanalysis (EMPA), using a JEOL JXA-8800R instrument, at Misasa following the techniques of Iizuka (1996). Trace element analyses of clinopyroxene, orthopyroxene and plagioclase phenocrysts were carried out employing a Cameca ims 5f ion microprobe at PML, Misasa, following procedures described by Nakamura & Kushiro (1997, 1998). Standard materials used for the calibration of trace element compositions were clinopyroxene from mantle xenoliths for pyroxene and hornblende analyses, and labradorite megacrysts for plagioclase analyses. These standards have been well characterized in terms of homogeneity and concentrations by both ion microprobe and ICP-MS. Clinopyroxene phenocrysts in the thin sections were sputtered with an O- primary beam of 10 nA intensity, resulting in 10 µm beam diameter, and orthopyroxene and plagioclase with 15–20 nA intensity, resulting in 15 µm beam diameter. Positive secondary ions were collected by ion counting using an energy offset of -60 V for Si, Rb, Sr, Y and Zr, and of -45 V for other elements from 4500 V acceleration with an energy bandpass of ±10 V. These operational conditions resulted in (1-2) x 105 c.p.s. for 30Si secondary ion in the analysis of phenocrysts. Analytical reproducibilities (RSD 1%, n = 10) for trace elements in the clinopyroxene standard were typically 5%, except for Ba (60%), and for the labradorite standard also typically 5%, except for Zr, Nb, Sm and heavy rare earth elements (HREE) (20–30%).
|
RESULTS AND DISCUSSION
|
|---|
Major element compositions of whole-rock samples
Major and trace element compositions of the Akagi samples are
given in Table
2. Selected major element oxides are plotted
against SiO
2 content in Fig.
2, classifying the samples into
the five stages defined by Moriya (1968)
with reference to the
Nasu volcanic zone, which forms the volcanic front of NE Japan
(Kawano
et al., 1961
). As shown in Fig.
2, the range of SiO
2 contents of the Akagi samples is between 53·4 and 71·4
wt %, consistent with those of previous studies, indicating
that basaltic rocks have not been discovered at Akagi (Koga,
1984
; Yamaguchi, 1990
). TiO
2, Fe
2O
3, MgO and CaO show negative
trends against SiO
2, and these trends are essentially consistent
with those of the Nasu volcanic zone. Two different trends,
the relatively steeper trend formed by the samples in the Oe
and Om stages and that in the Ol, Y and Cc stages, can be recognized
in the Fe
2O
3 and MgO diagrams, as well as in the Na
2O diagram.
View larger version (32K):
[in this window]
[in a new window]
|
Fig. 2. Major element variation against SiO2 (wt %) content in Akagi samples. Dark and light gray shaded areas are the tholeiite and calc-alkali rock series of the Nasu Volcanic zone in NE Japan, respectively [data from Kawano et al. (1961)].
|
|
The Al2O3–SiO2 diagram shows a more complicated feature, with a positive trend for the earlier stages of Oe and Om, and a negative trend for the later stages of Ol, Y and Cc. The former positive trend may be explained by considerable plagioclase accumulation in a shallow magma chamber, as the deduced Al2O3 contents in the melts obtained by subtracting the phenocryst compositions from those of the whole rocks form a negative trend against SiO2 overlapping the samples from the later stages as shown in Fig. 3. The major element compositions, after subtraction of the phenocrysts in Adatara volcano (the neighboring volcano located in the volcanic front of NE Japan), are also shown based on the data given by Fujinawa (1988, 1990). The corrected Al2O3 concentrations for Akagi volcano indicate a similar negative trend to that of Adatara volcano; however, that of Akagi is still systematically higher than for Adatara at similar SiO2 contents. The relatively high content of Al2O3 in the Akagi samples is consistent with the results of Yamaguchi (1990). The differentiation trend of corrected CaO against SiO2 (Fig. 3), which is also influenced by plagioclase accumulation, shows similar negative trends with increasing SiO2 content. However, the corrected Na2O contents are scattered and the differentiation trends are less obvious.
Two distinct trends in the co-variation of FeO* (total Fe as FeO) and MgO of Akagi volcano can be divided into different rock series using the Miyashiro diagram (Fig. 4a; Miyashiro, 1974). In this diagram, the Oe and Om suites belong to the tholeiitic rock series, and the Ol, Y and Cc suites to the calc-alkaline rock series. On the other hand, although the Oe and Om suites are slightly less depleted in FeO* than the Ol, Y and Cc suites, the entire Akagi volcanic suite follows a calc-alkaline differentiation trend in the AFM diagram (Fig. 4b). These major element characteristics of Akagi volcano cannot be explained by a simple magmatic differentiation process, because there appear to be considerable gaps in the SiO2 and total alkali contents in the transition from the Om to the Ol suites (Figs
2 and 4a). It may be, therefore, necessary to invoke more complex processes, such as magma mixing and crustal contamination associated with crystal fractionation. Moreover, the difference in the eruption style, which is the basis for the discrimination of the stages, does not correspond to the characteristics of the major element compositions. It may thus be more suitable to discuss the geochemical evolution of Akagi volcano without distinctions between stages and substages. Here, we discriminate simply between samples from the five stages (Oe, Om, Ol, Y, Cc), which are distinguishable by geological discontinuities.
Trace element compositions
Trace element compositions of whole rocks are listed in Table 1, and are plotted in primitive mantle (PM)-normalized diagrams in Fig. 5. The contents of incompatible elements are enriched compared with most basaltic rocks of the volcanic front of the NE Japan arc (Shibata & Nakamura, 1997). However, the less incompatible elements such as the HREE are less enriched. These trace element characteristics cannot be explained by simple fractional crystallization from a common primary basaltic magma in the volcanic front using the phenocryst assemblage in the Akagi volcanic rocks. Such a fractionation process should increase not only the most incompatible trace elements but also the HREE. The behavior of REE in the Akagi samples is discussed below in more detail in a section describing their isotopic compositions. With progress within a volcanic stage, the contents of highly incompatible elements become gradually higher. However, no correlation is observed between trace elements and SiO2 contents in the Om and Ol stages. Akagi volcanic rocks show positive spikes of Sr and Pb, and remarkably negative spikes for Nb and Ta, consistent with the general features of trace elements in island-arc volcanic rocks (e.g. Wood et al., 1979; Perfit et al., 1979; Sakuyama & Nesbitt, 1986; Shibata & Nakamura, 1997; Woodhead et al., 1998).
Sr–Nd–Pb isotope systematics of Akagi volcano
The whole-rock Sr, Nd and Pb isotopic compositions are listed in Table 3 and are plotted in Fig. 6 along with the reference variations of volcanic rocks from the NE Japan arc, lower-crustal materials, oceanic sediments and MORBs. The 87Sr/86Sr ratios of the Akagi samples vary widely from 0·70603 to 0·70879, consistent with the data reported by Notsu et al. (1985), and are significantly higher than those of other typical volcanoes in the northern part of Tanakura Tectonic Line in the volcanic front of the NE Japan arc, which have an average 87Sr/86Sr ratio of 0·7045 (Notsu, 1983; Gust et al., 1997; Shibata & Nakamura, 1997). The Nd isotopic compositions of Akagi range from -0·4 to -8·6 in Nd, and are much lower than those of other volcanoes in the northern part of the volcanic front with Nd of 3–10 (Gust et al., 1997; Shibata & Nakamura, 1997). The 206Pb/204Pb, 207Pb/204Pb and 208Pb/204Pb ratios vary in the range of 18·21–18·44, 15·56–15·64 and 38·44–38·79, respectively (Fig. 6b and c). These isotope characteristics of Akagi, therefore, are regarded as ‘unusual’ in the NE Japan arc, confirming the original observations based on Sr isotopic compositions by Notsu et al. (1985).
View larger version (42K):
[in this window]
[in a new window]
|
Fig. 6. Nd–Sr–Pb isotope systematics of the Akagi samples. Symbols are as in Fig. 2. Nd isotope ratios are normalized to CHUR (143Nd/144Nd = 0·512638). The large open circle represents the isotopic composition of the inferred primary magma of the NE Japan arc, and the thick continuous line extending from the open circle indicates the across-arc variation in the NE Japan arc (Shibata & Nakamura, 1991, 1997). The fields ‘N’ and ‘S’ represent the along-arc variations of the magmatic suites from north and south of the Tanakura Tectonic Line (NTTL and STTL), respectively (Kersting et al., 1996; Gust et al., 1997). x, isotopic composition of JG-1, the granitic rock standard material of the Geological Survey of Japan (Nohda & Wasserburg, 1981; Koide & Nakamura, 1990). The range of isotopic variation of MORB, oceanic sediments including those on the Philippine Sea plate and the lower crust are also shown. Data source for MORB: Cohen et al. (1980), Dupre & Allègre (1980), White & Hofmann (1982), Le Roex et al. (1983); oceanic sediments: O’Nions et al. (1978), White & Patchett (1984), Woodhead & Fraser (1985), White & Dupre (1986), Ben Othman et al. (1989), Cousens et al. (1994), Shimoda et al. (1998); lower crust: Zartman & Doe (1981), Rudnick et al. (1986), Zartman & Haines (1988), Stoltz & Davies (1989), Kempton et al. (1990), Rudnick & Goldstein (1990), Downes et al. (1991).
|
|
In the Sr and Nd isotope diagram (Fig. 6a), the Akagi samples have a linear trend with an extremely isotopically enriched signature compared with those of other ‘typical’ volcanoes in the volcanic front of the NE Japan arc, which have simple isotopic across-arc variations (Shibata & Nakamura, 1997). These isotopic compositions are widely varied not only over the entire history of the volcano but also with each volcanic stage of Akagi volcano. An extrapolation of the Akagi trend to the depleted direction nearly intersects the isotopic trend of the NE Japan arc at a value typical of volcanoes on the front. As is shown in Fig. 6b and c, the Pb isotopic compositions of the Akagi samples form clusters that are clearly distinct from the isotopic trend of the NE Japan arc (Shibata & Nakamura, 1997). The 206Pb/204Pb ratios are lower, whereas the 207Pb/204Pb and 208Pb/204Pb ratios are higher than those of the ‘typical’ volcanoes on the front. In the Nd–206Pb/204Pb diagram (Fig. 6d), the Akagi data also define a linear trend clearly different from the tendency in the isotope variations of other primary magmas in the northern part of the NE Japan arc, and which extrapolates to the values characteristic of the volcanic front.
The above isotope systematics suggest that the Akagi volcanic rocks were formed by two-component mixing of an isotopically depleted and an enriched end-member. Intersections between the ‘typical’ trend of the NE Japan arc and the extrapolations of the Akagi trends in the Sr–Nd and Nd–Pb diagrams indicate that the depleted end-member is similar to the primary magma in the volcanic front of the NE Japan arc, which was defined by Shibata & Nakamura (1997). It is, however, still difficult to identify the source of the enriched end-member using these isotope systematics.
On the basis of the existence of an enriched component deduced from the isotope systematics and the tectonic setting of the Akagi volcano, the following candidates for the enriched end-member can be postulated: (1) oceanic sediments; (2) upper crust; (3) lower crust. Isotopic variations of oceanic sediments in the NW Pacific and lower-crustal materials are shown in Fig. 6. The isotopic composition of JG-1, the granitic standard of the Geological Survey of Japan, which outcrops near Akagi volcano and is considered to be the upper-crustal basement of Akagi, is shown in the diagrams. It should be noted that the JG-1 composition lies within the field for oceanic sediments (data from Nohda & Wasserberg, 1981; Koide & Nakamura, 1990), therefore the upper-crustal field is not shown in Fig. 6. The isotopic characteristics of lower-crust materials under the NE Japan arc have not been well understood. The lower crust under the arc might be considered as a fragment of continental material and/or a piled-up sequence of young metamorphic rocks originally derived from the arc magmas (Arculus & Johnson, 1981). In the latter case, the isotopic characteristics should be similar to those of the arc volcanic rocks, as a result of less time-integrated isotopic evolution. We prefer, therefore, that the isotopic characteristics of the lower-crustal materials are similar to ‘continental’ type lower crust in Fig. 6.
In the Sr–Nd diagram (Fig. 6a), the isotopic compositions of oceanic sediments and lower crust overlap, making it difficult to unambiguously identify an enriched end-member. However, the Pb–Pb and Nd–Pb diagrams clearly discriminate between lower crust and oceanic sediments and/or upper crust. On the basis of Fig. 6b–d, it appears that oceanic sediments and/or upper crust are not possible candidates for the enriched end-member. In practice, the isotopic compositions of sediments, which occur on the Philippine Sea plate, cannot be the enriched end-member for Akagi volcano. Consequently, it is most likely that the enriched end-member involved in magma formation of Akagi volcano is lower-crustal material, although Notsu et al. (1985) proposed a sediment component derived from the subducting Philippine Sea plate as the enriched end-member in Akagi volcano based on 87Sr/86Sr and 
18O data. The complete Sr, Nd and Pb isotopic dataset from the along-arc volcanoes by Kersting et al. (1996) and Gust et al. (1997) is also shown in Fig. 6. The isotopic trends defined by the volcanoes located to the south of the Tanakura Tectonic Line (STTL: Akagi, Nikkoshirane, Nantai, Takahara, Nasu) essentially have the same direction as those of the primary magmas in the volcanic front of the NE Japan arc (Fig. 6). Therefore, it is likely that the depleted end-member for the source of the volcanoes from the STTL also is similar to the MORB-type wedge mantle metasomatized by the fluid derived from dehydration of the subducted slab as proposed by Shibata & Nakamura (1997) for the primary magma source of the NE Japan arc. Moreover, the Nd–Pb diagram (Fig. 6d) for the STTL group clearly indicates the involvement of lower-crustal materials as an enriched end-member, as well as at Akagi volcano. It is, therefore, not necessary to introduce a unique mantle source such as Indian Ocean-type mantle beneath the volcanoes from the STTL to explain the isotope systematics of the STTL group based on the dataset given in this study and in previous studies (Kersting et al., 1996; Gust et al., 1997).
Major and trace element compositions of phenocrysts
Seven samples from the Ol stage, which is characterized by large variations in whole-rock major element and isotopic composition (SiO2 = 55·8–63·5 wt %, 87Sr/86Sr = 0·70683–0·70879), were selected to investigate detailed magmatic processes based on the chemical and isotopic compositions of phenocrysts.
Phenocrysts in Akagi volcano rocks are always zoned. The mg-numbers [defined as 100 x Mg/(Mg + Fe2+)] of cores in orthopyroxene and clinopyroxene are 75–65 and 80–73, respectively. Reverse zoning is commonly observed in the pyroxene of samples AK1102, AK0910, AK1010 and AK1108. The an-number in the cores of plagioclase phenocrysts does not have a bimodal signature, which has been regarded as evidence of magma mixing according to Sakuyama (1981). Hornblende phenocrysts occur in only the most siliceous sample in this stage. The hornblende phenocrysts have a zonal structure with decreasing mg-number and edenite component from core to rim, and a pseudomorph of hornblende phenocrysts, consisting of fine-grained plagioclase + orthopyroxene + clinopyroxene + opaque minerals, also occurs in five siliceous samples listed in Table 1.
Trace element compositions of orthopyroxene, clinopyroxene and plagioclase phenocrysts in the Ol stage are given in Table 4 and their primitive mantle-normalized patterns are shown in Figs
7 and 8. The trace element patterns of pyroxenes show clear negative Sr and Nb spikes. Experimental determinations of the Kd for Sr between pyroxenes and silicate melt show no Sr depletion compared with the adjacent elements Pr and Nd (Green, 1994). The Sr depletion, therefore, indicates that the pyroxenes crystallized from Sr-depleted melts, although Sr enrichment is observed in whole-rock analyses of the Ol samples (see Fig. 5). Such Sr enrichment could be expected as a result of the accumulation of plagioclase, which is characterized by a strong Sr positive spike (Fig. 8). The influence of plagioclase accumulation is expected based on the major element compositions of the whole-rock samples (see Fig. 3). The positive Sr spikes are removed by consideration of phenocryst subtraction; for example, Sr/Sr* value [Sr* is defined as (Pr + Nd)/2 on the primitive mantle-normalized value] of the melt part in AK-A decreases from 2·0 to 0·7.
View larger version (47K):
[in this window]
[in a new window]
|
Fig. 7. PM-normalized trace element patterns of clinopyroxene and orthopyroxene phenocrysts in Ol stage samples. For the AK1108 sample, hornblende data are represented instead of clinopyroxene, because there are no clinopyroxene phenocrysts in this sample. Open and filled symbols indicate the rim and core of phenocrysts, respectively.
|
|
View larger version (40K):
[in this window]
[in a new window]
|
Fig. 8. PM-normalized trace element patterns of plagioclase phenocrysts in Ol stage samples. Open and filled symbols indicate the rim and core of phenocrysts, respectively.
|
|
The Nb depletions in the phenocrysts can be attributed to the whole-rock compositions that are unique to island-arc volcanic rocks (see also Fig. 5). The Zr depletion in clinopyroxene and the slightly positive Zr spike in orthopyroxene may be simply explained by the difference in Kd [i.e. KdHFSE/KdLILE,LREE > 1 for orthopyroxene, KdHFSE/KdLILE,LREE < 1 for clinopyroxene (where HFSE indicates high field strength elements, LILE indicates large ion lithophile elements and LREE are light REE); Green, 1994]. The Y depletion in the plagioclase phenocrysts may be a characteristic feature of plagioclase, because such depletion is always observed in the plagioclases used for our secondary ion mass spectrometry standards (also precisely characterized by ICP-MS). In samples AK1010 and AK1108, the incompatible elements in the orthopyroxene rims are significantly depleted compared with those of the core. This is consistent with reverse zoning (based on mg-number) in the pyroxene of these samples.
Sr and Nd isotope compositions of phenocrysts in the Ol stage
The isotopic compositions of Sr and Nd for phenocrysts in the Ol stage are given in Table 5 and plotted in a Sr–Nd isotope diagram in Fig. 9. In this diagram, the isotopic compositions of the melts are calculated using the Sr and Nd isotopic and elemental compositions of the whole rocks, and the modal abundance, densities, isotopic and elemental compositions of the phenocrysts. The results are given in Table 6. The isotopic compositions of the phenocrysts tend to have more enriched compositions, especially for Sr, exceeding the analytical uncertainties, than the calculated compositions of the melts. In other words, the phenocrysts are isotopically in disequilibrium with the surrounding melt.
View larger version (13K):
[in this window]
[in a new window]
|
Fig. 9. Sr and Nd isotope systematics of phenocryst minerals in Ol stage samples. Data from the same sample are connected by tie-lines. The bulk-rock and calculated melt isotopic compositions (see text) are also shown.
|
|
The uncertainty of calculated melt isotopic compositions mainly depends on the measured modal abundance of the phenocrysts, especially plagioclase, which is a main sink for Sr. The statistical error of the plagioclase modal abundance is estimated to be 3%, based on the counting statistics. This uncertainty affects the 87Sr/86Sr ratio of melts up to ±0·0002 for AK-A and AK1012, but the uncertainties in the others are less than ±0·0001. In contrast, the calculated Nd isotopic compositions of the melts are not significantly affected by the uncertainty in the modal abundance. Heterogeneity in the sample also causes the error in the estimation of the Sr and Nd isotopic composition of the melt. However, the distributions of phenocrysts are homogeneous in the samples from the Ol stage. Therefore, the isotopically depleted characteristics of the calculated melts compared with the phenocrysts are apparently significant except in sample AK1010. In AK1010, the Sr isotopic composition of the melt is identical to that of the phenocrysts within the analytical uncertainties. However, the Nd isotopic composition of the plagioclase is significantly lower than that of the calculated melt.
|
INTERACTION BETWEEN ISLAND-ARC MAGMA AND LOWER CRUST
|
|---|
In Fig.
10, chondrite-normalized REE patterns of Akagi volcano
samples are compared with those for Adatara volcano (Fujinawa,
1992
), which is a neighboring volcano in the volcanic front
of the NE Japan arc and has similar petrological and mineralogical
features to those of Akagi. This comparison shows that Akagi
samples have (1) significantly lower HREE abundance, (2) significantly
higher LREE/HREE ratios, and (3) U-shaped REE patterns between
the middle REE (MREE) and HREE. The REE patterns from Adatara
volcano were explained by crystal fractionation (Fujinawa, 1992
).
However, the REE characteristics of Akagi volcano cannot be
formed by such a simple fractionation involving removal of minerals
such as plagioclase, orthopyroxene, clinopyroxene and opaque
minerals occurring as phenocrysts in the Akagi samples. Because
the mineral–melt partition coefficients of the REE are
much smaller than one, fractional crystallization in a shallow
magma chamber would increase the total REE with LREE enrichment,
as is observed in Adatara volcano. This mechanism, therefore,
cannot explain characteristics (1) and (2) of Akagi’s
REE compositions.
On the basis of a high-pressure melting experiment involving a basaltic composition, Forden & Green (1992) suggested that the liquidus phase is clinopyroxene under high-H2O conditions in the lower-crustal environment, and olivine is the liquidus phase at lower pressure conditions. When the temperature of the melts drops, clinopyroxene becomes unstable, and reacts with the melt to form amphibole, resulting in the evolved melt being more siliceous and decreasing the H2O content of the magma itself. The lower-crustal granulites commonly contain clinopyroxene; ascending H2O-rich magma from the mantle, therefore, might react with the clinopyroxene and fractionate amphibole as a residual phase. The MREE are preferentially partitioned into amphibole, with mineral–melt partition coefficients greater than one (Adam & Green, 1994; Sisson, 1994; Witt-Eickschen & Harte, 1994). Consequently, the evolved magma acquires a U-shaped REE pattern because of the buffering effect of residual amphibole. Partial melting experiments on lower-crustal materials under high-pressure conditions indicate that the resultant liquid has a felsic composition caused by the persistence of mafic minerals in the residue (e.g. Beard & Lofgren, 1991; Beard et al., 1993, 1994). Although amphibole fractionation processes in shallow magma chambers could also explain the U-shaped REE patterns, the occurrence of hornblende phenocrysts is rare and only very small amounts occur in felsic samples in the later volcanic stages (Table 1). This process is, therefore, unlikely to explain the overall U-shaped signature retained in the Akagi volcanic rocks.
If the interaction between H2O-rich magma and lower-crustal material is a major process controlling the REE patterns of Akagi volcano, a process of ‘assimilation–fractional crystallization’ (AFC, DePaolo, 1981) under lower-crustal conditions might be invoked. To examine the assimilation process in the lower crust, the AFC model (DePaolo, 1981) was applied to the isotope and trace element systematics in the Akagi volcano. In this model calculation, a primary magma deduced from ‘normal’ primary magma on the northern part of the volcanic front in the NE Japan arc, defined by Shibata & Nakamura (1997), is assumed as an ‘assimilator’. It is, however, difficult to determine the other parameters such as isotope and element ratios of assimilated material (hereafter denoted as ‘assimilant’) under Akagi volcano, and bulk partition coefficients between the melt and the assimilant. Thus, we fixed these parameters, and only evaluated the assimilation rate (r) as given by
where
Ma(
t) and
Mc(
t)
represent a rate (mass/unit time) of wallrock assimilation and
a rate of fractionation of crystallizing phases, respectively.
Thus, the assimilation rate implies the rate of contribution
of masses controlled by the assimilation and the crystal fractionation
to the total mass of magma. Furthermore, the assimilation rate
is an essential parameter to evaluate the temperature conditions
at which the AFC takes place. The averages of element ratios
of lower-crustal materials compiled by Rudnick & Fountain
(1995)
are used for the assimilant. The isotopic compositions
of the assimilant are obtained by the extrapolation of the linear
correlations in the multi-isotope systematics of Akagi volcano.
The bulk partition coefficients between magma and the assimilant
were obtained by using the modal abundance of residual phases
consisting of plagioclase, amphibole, clinopyroxene and orthopyroxene,
as determined by hydrous melting experiments at lower-crustal
conditions by Beard & Lofgren (1991)
. The REE signature
of Akagi volcanic rocks indicates that amphibole fractionation
was involved in the evolution of magma under lower-crustal conditions
as discussed in the previous sections. All parameters used in
the AFC calculations are compiled in Table
7.
Representative results from the AFC model calculations are presented together with data for Akagi, Adatara and other volcanoes from NE Japan in Fig. 11a–d. There appear to be correlations between isotope and element ratios in the arc data, although the correlation is poor for Pb isotopes, probably as a result of the small variation of Pb isotope compositions and the small differences between those of the lower crust and the primary magma. These correlations are obviously different from those of basaltic rocks elsewhere in the NE Japan arc and at Adatara volcano, for which the source heterogeneity and crystal fractionation processes have caused variations in isotope and elemental ratios without crustal assimilation (Shibata & Nakamura 1997). As shown in Fig. 11, most of the Akagi data plot near the mixing lines with an assimilation rate of 0·9, although those of the earliest and final stages tend to plot above the mixing lines with assimilation rates <0·9. This observation indicates that the extent of crustal assimilation relative to the crystallization in the middle stage is larger than in the earliest and latest stages of Akagi volcano. In the middle stage, therefore, the enrichment of incompatible elements together with isotope variations may largely have resulted from significant lower-crustal assimilation by primary magma similar to ‘normal’ basaltic magmas in the northern part of the volcanic front in the NE Japan arc.
When the assimilation rate is nearly 1·0, the latent heat of crystallization and the heat of fusion for the assimilation process are thermally balanced, approximately. Such a circumstance is likely to be achieved under lower-crustal conditions, because the geotherm in the lower crust is such that temperatures are 600–700°C, making it feasible to reach the solidus temperature of the assimilant by intrusion of magma and with less heat loss. On the other hand, the latent heat of crystallization could be mostly consumed by raising the temperature of the wallrock to its solidus, when assimilation occurs under shallower crustal conditions. Consequently, this results in an assimilation rate smaller than that occurring under lower-crustal conditions. From the above discussion based on the AFC model calculations, it may be concluded that the assimilation occurred primarily in the lower-crustal region, resulting in the unique isotope characteristics of the magmas of Akagi volcano.
The seismologically determined Moho depth gradually increases from the northern part of NE Japan along the volcanic front, and it shows the maximum value beneath Akagi volcano (38 km, 1·3 GPa), which is 4 km deeper than the typical depth beneath the volcanic front of the NE Japan arc (Zhao et al., 1992). Hildreth & Moorbath (1988) suggested that chemical interaction between magma and crust is closely related to the crustal thickness in the Andean arc. The thicker crust might effectively increase the assimilation rate, because the higher temperatures in the bottom of the crust mean that it more easily attains its solidus temperature. Consequently, volcanoes such as Akagi located on thicker crust might experience a larger effect of assimilation of lower crust on their trace element and isotope compositions.
|
EXISTENCE OF A WATER-RICH MAGMA
|
|---|
The above discussion based on isotope and trace element systematics
requires a primary magma sufficiently enriched in H
2O to stabilize
amphibole in the lower crust. It is, therefore, essential to
understand the mechanism whereby an H
2O-rich source region for
the primary magma in the mantle wedge beneath Akagi volcano
could form.
On the basis of petrological observations, quartz phenocrysts generally occur in felsic volcanic rocks without hydrous phenocrysts in the volcanic front of the NE Japan arc (Sakuyama, 1979). However, quartz does not appear as a phenocryst in Akagi samples, even in the felsic samples. Sakuyama (1979, 1983a, 1983b) determined the H2O content of magmas based on the crystallization sequence of phenocrysts. Following his definition, the Akagi volcanic rocks belong to Type II–III, which allows magmas to contain 3–4 wt % of H2O at 0·5 GPa. This suggests that the H2O content of the Akagi magmas is considerably higher than that of the Type I volcanoes, typical of the volcanic front in NE Japan, which contain <3 wt % of H2O.
Yamaguchi (1990) inferred that the groundmass in the Akagi volcanic rocks has a significantly higher Al2O3 content than that of the adjacent Hotaka volcano. This is because the plagioclase stability field becomes narrower relative to the diopside stability field with increasing H2O content in the magma. This observation is also consistent with the hypothesis that the primary magma of Akagi is enriched in H2O relative to other volcanoes in the volcanic front in NE Japan.
The peculiar petrological and geochemical characteristics of Akagi volcano discussed above might be explained by the development of the unusual tectonic setting beneath Akagi volcanic area, where the Philippine Sea plate overlaps onto the Pacific plate. Such a tectonic setting might lead to a significantly larger amount of water in the magma source region than that expected from subduction of a single Pacific plate.
|
EVOLUTION OF MAGMA IN THE Ol STAGE
|
|---|
Recently, magma mixing at Akagi has been discussed, on the basis
of the major element composition of phenocrysts and their textures,
in the pumice from the Y stage, by Horio & Umino (1995)
and Umino & Horio (1998)
. Those workers have suggested that
high-
T, less fractionated magma was periodically injected into
a low-
T, silicic, mushy magma in a shallow magma chamber. However,
previous studies have not addressed the origin of these two
different magmas because of the limited geochemical information
available, especially isotope compositions, for Akagi volcano.
On the basis of the major element compositions of pyroxenes, the temperatures of the magmas in the Ol stage were determined by using the QUILF program of Andersen et al. (1993). Average compositions of the rims of clinopyroxene and orthopyroxene were used for the temperature estimation. Ferric–ferrous ratios were calculated to satisfy the stoichiometry of the pyroxenes. The samples investigated from the Ol stage can be divided into high-T (1000–1200°C) and low-T (<1000°C) groups, and tend to have a negative correlation between calculated temperature and Sr isotopic composition (Table 8 and Fig. 12). This indicates that such a variation in magma temperatures is not a consequence of the simple cooling process in the shallow magma chamber.
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 12. Magmatic temperature vs Sr isotopic composition of whole rock, clinopyroxene and orthopyroxene phenocrysts in the Ol stage. The magmatic temperature is determined using the QUILF program (Andersen et al., 1993).
|
|
In the Ol samples, melts are isotopically more depleted than the coexisting phenocrysts (Fig. 9) as discussed in the previous section. Such isotopic disequilibrium observed could be explained only by a mechanical mixing between the isotopically enriched magma equilibrated with the phenocrysts and the relatively depleted aphyric magma. Mass-balance calculations, based on the trace element composition and modal abundance of the phenocrysts, indicate that the Sr/Nd ratio in the glass of the phenocryst-rich rock is lower than that in the aphyric-primary magma as a result of the equilibration of phenocrysts, especially plagioclase. If the aphyric-primary magma has a more depleted isotopic signature than the phenocryst-rich magma, the mixing curve forms a convex hyperbola in the Sr and Nd isotope diagram (Fig. 13a). That is, the Sr isotopic compositions of glasses are characteristically more depleted than those of the coexisting phenocrysts. However, the difference in Nd isotopic composition between them is smaller than the difference in Sr isotopic composition. These observations, together with model mixing, may suggest that the involvement of depleted aphyric magma mixed with phenocryst-rich magma was relatively small, and that isotopic re-equilibration among the phenocrysts and the mixed melt has not been achieved.
View larger version (25K):
[in this window]
[in a new window]
|
Fig. 13. (a) Schematic illustration, in a Sr–Nd isotope diagram, of two-component mixing (after DePaolo & Wasserburg, 1979). K, Sr/Nd ratio in component A relative to component B; F, fraction of component A in product magma. (b) Model calculation of mixing between the aphyric, isotopically depleted melt (Pm) and the isotopically enriched melts (Fm1, Fm2). The chemical parameters are summarized in Table 9. The numbers beside each curve represent the weight fraction of Pm in the mixed melt. The mixing trajectories are concave hyperbolae, and the K values of Curve 1 and Curve 2 are 3·8 and 3·4, respectively. A relatively small amount of injection of the isotopically depleted primitive magma (Pm) can produce the depleted melt part for Sr isotopes (Mm1 and Mm2) as a result of the mixing.
|
|
To examine the above argument further, a model calculation was
carried out using the samples with remarkable isotope disequilibrium
in Sr between melt and phenocrysts (AK1012 and AK1102) representing
the high-
T and the low-
T group, respectively. The results are
presented in Fig.
13b, based on the parameters given in Table
9. In this model, the isotope and trace element compositions
of the primary aphyric magma are assumed to be those of the
primary basaltic magma (a depleted end-member, Pm) in the volcanic
front of NE Japan (Shibata & Nakamura, 1997
). The isotopic
compositions of two end-members (Fm1 and Fm2) are represented
by the values of the most enriched phenocrysts in AK1012 and
AK1102, assuming that the phenocrysts in the enriched end-members
were isotopically equilibrated with their host melts, when the
phenocrysts crystallized. The Sr and Nd abundances in Fm1 and
Fm2 were obtained by the subtraction of phenocryst compositions
from the whole rock of the most mafic sample AK0910 in the Ol
stage. In this model, clinopyroxene and opaque phenocrysts are
not considered. Because the
Kd values of Sr and Nd for clinopyroxene
are almost identical (Green, 1994
), and because of the extremely
low concentrations of these elements in the opaque minerals,
very lit
TOP
ABSTRACT
INTRODUCTION
