Abstract

Xenoliths from Engeln–Kempenich in the East Eifel volcanic field (Germany) comprise gabbroic to ultramafic cumulates, and meta-igneous and meta-sedimentary granulite- to amphibolite-facies lithologies. They provide evidence for Pleistocene heating and metasomatism of the lower continental crust by mafic magmas. The metamorphic xenoliths were divided into three types: (1) primitive type P, which are little affected by metasomatic replacement structures; (2) enriched type E1 defined by metasomatic replacement of primary pyroxene and garnet by pargasitic amphibole and biotite; (3) enriched type E2 defined by breakdown of hydrous phases. Type E rocks are geochemically related to type P and cumulate xenoliths by compositional trends. During modal metasomatism, type E rocks were oxidized. Type E1 rocks were typically enriched in Rb, Th, U, Nb, K, light rare earth elements (LREE) and Zr, and E2 enriched in Rb, Th, U, Nb, K, REE, Zr, Ti and Y, relative to type P rocks. Formation of the hydrous, chlorine-bearing phases amphibole and scapolite containing glass and fluid inclusions in the E1 rocks provides evidence for a water and Cl-bearing fluid phase coexisting with silicate melt. Accordingly, we calculated 10 mol % H2O back into the CO2-dominated fluid inclusions, in agreement with experimental data on the composition of a fluid phase coexisting with mafic alkaline melts at elevated pressure. Primary CO2-dominated fluid inclusions coexisting with glass inclusions in metamorphic corona phases and neoblasts, and in cumulate xenoliths, have overlapping densities. Fluid inclusion barometry using the corrected densities indicates that both cumulates and metamorphic xenoliths originated from the same depth at 22–25 km (650 ± 50 MPa). This is interpreted as being a main magma reservoir level within the upper part of the lower crust close to the Conrad discontinuity, where the xenoliths represent wall-rocks. The Conrad discontinuity separates an upper-crustal layer, consisting of preferentially ductile granodioritic and tonalitic gneisses, and more brittle lower-crustal mafic granulites. The brittle–ductile transition appears to be a preferred level of magma stagnation.

INTRODUCTION

Basaltic magmas contribute to the formation of lower continental crust by ‘underplating’ through intrusion and crystallization (e.g. Cox, 1983; Furlong & Fountain, 1986; Rudnick & Fountain, 1995). These basaltic intrusions interact with the lower crust by transfer of heat and matter, especially by metasomatism, which may be induced by fluid phases released from the intruded magmas (e.g. Newton et al., 1980; Touret, 1986; Katz, 1987). The components added to and/or removed from the lower crust, the nature of the transporting fluids and the relative importance of metasomatism during the evolution of the lower crust, are poorly constrained.

Xenoliths are fragments of the wall-rocks of the magma plumbing system and can provide important evidence for metasomatic interaction of the magma with the lower crust. The clearest evidence for metasomatism of lower-crustal rocks is the appearance of new phases in the xenoliths. Metasomatism involving development of new phases is referred to as ‘modal’ (Harte, 1983). Here we present a study of xenoliths with different degrees of modal metasomatism, sampled from a Pleistocene (<450 ka; Schmincke et al., 1990) phreatomagmatic nephelinitic tephra deposit of probably one single eruption of the East Eifel volcanic field (Engeln–Kempenich, Germany) (Fig. 1). We suggest a model in which metasomatism of the lower crust is due to the formation of mafic magma chambers. Barometry of fluid inclusions and mineral thermometry of cumulate and metamorphic xenoliths was used to locate the depth of a magma reservoir and to estimate the conditions of formation of the metasomatic phases.

Fig. 1.

Location of the Engeln–Kempenich locality and the Pleistocene East Eifel volcanic field. After Wörner et al. (1985).

Fig. 1.

Location of the Engeln–Kempenich locality and the Pleistocene East Eifel volcanic field. After Wörner et al. (1985).

GEOLOGICAL SETTING AND PREVIOUS STUDIES OF THE EIFEL CRUSTAL XENOLITHS

The East Eifel volcanic field is part of the Rhenish Massif, which has been undergoing uplift since the late Tertiary (Meyer et al., 1983), probably as a result of upwelling asthenosphere. The Pleistocene volcanism in the Eifel has been summarized by Schmincke et al. (1990). Xenolith studies of the upper and middle crust beneath the East Eifel volcanic field have been published by Wörner et al. (1982), Voll (1983), Wörner & Fricke (1984) and Mengel et al. (1991). The uppermost Eifel crust is characterized by folded Devonian sediments. The Devonian strata are underlain by greenschist-facies rocks and by medium- to high-grade amphibolite-facies rocks (e.g. staurolite schists, granitic to tonalitic gneisses). Okrusch et al. (1979), Voll (1983), Loock et al. (1990) and Mengel et al. (1991) have studied petrographically mafic garnet granulite xenoliths, which are considered to be part of the lower crust. Stosch & Lugmair (1984) and Stosch et al. (1986) found Sm–Nd isotopic evidence for metasomatic overprinting of these lower-crustal xenoliths. They suggested that the addition of a light rare earth element (LREE-)enriched metasomatic component had occurred more recently than Jurassic times. The source of the metasomatism was attributed to fluids from the upper mantle.

PHASE ASSOCIATIONS

The studied xenoliths from Engeln–Kempenich comprise [See Table 1; abbreviations of mineral names are from Bucher & Frey (1994).] All studied xenoliths from Engeln–Kempenich are well rounded. The crustal xenoliths have diameters up to 20 cm. Peridotite xenoliths are generally smaller, having diameters ≤3cm.

  1. gabbroic and clinopyroxenitic cumulates, which are probably comagmatic with the host magma;

  2. rare granulite-facies meta-sedimentary quartz-bearing garnet–sillimanite gneisses;

  3. mafic to ultramafic meta-igneous granulites, meta-clinopyroxenites and meta-hornblendites;

  4. rare peridotites (highly recrystallized spinel harzburgites and amphibole-bearing spinel harzburgites of the upper mantle), which are not considered here.

Table 1:

Phase associations of granulite-facies xenoliths, mineral temperatures (only for underlined samples; mineral rims, clinopyroxene–orthopyroxene solvus, garnet–biotite, appearance of early-textured glass inclusions and interstitial glasses)

Sample nos
 
T Wells
 
T BK
 
Phases
 
K5/48, K5/61, K5/74, 835 816     Rt ±Ttn Di Opx Prg, Hbl  Pl  ±Ap  
K5/76, K9/30 
K5/55, K5/65, K9/28, k9/32 781 664    Ilm   Di  Prg, Hbl  Pl    
K5/57, K5/71 868 773   ±Ti-Mag  Rt  Di  Prg, Hbl  Pl   ±liq1 
K5/51, K5/59, K5/68   Grt   ±Ilm Rt   Opx Prg, Hbl  Pl  ±Ap  
K5/31   Grt  Ti-Mag    Di  Prg  Pl    
K5/45, K5/56, K5/70, k5/72, 827, 846, 797, 690, 741, 650, Grt  ±Ti-Mag ±Ilm   Di Opx Prg, Hbl ±Bt Pl  ±Ap  
K5/73, K5/75, K5/77, K9/24, 844, 867, 805 734, 762, 695 
K9/35, K9/70 
K5/52, K5/54, K9/23b 800, 827 668, 761 Grt Hc-Mag ±Ti-Mag    Di Opx Prg, Hbl  Pl ±Scp1 ±Ap  
K5/64, K9/23a, K9/27 827, 1230 761, 1082 Grt Hc-Mag     Di Opx Prg, Hbl     ±liq1 
K5/47, K5/67 >s >s  Hc-Mag ±Ti-Mag    Di Opx Prg  Pl ±Scp1 Ap  
K9/29, K9/69 >s, >s >s, >s  ±Hc-Mag ±Ti-Mag    Di  Prg  Pl ±Scp1 Ap liq1 
K5/58, K9/21, K9/22, K9/31 >s, >s, >s >s, >s, >s  ±Hc-Mag ±Ti-Mag    Di  Prg   ±Scp1 ±Scp2 Ap liq1 
K9/32     Ti-Mag   Ol Di Opx Prg      
K9/33 >s >s   Ti-Mag   Ol Di  Prg     liq1 
Sample nos
 
T Wells
 
T BK
 
Phases
 
K5/48, K5/61, K5/74, 835 816     Rt ±Ttn Di Opx Prg, Hbl  Pl  ±Ap  
K5/76, K9/30 
K5/55, K5/65, K9/28, k9/32 781 664    Ilm   Di  Prg, Hbl  Pl    
K5/57, K5/71 868 773   ±Ti-Mag  Rt  Di  Prg, Hbl  Pl   ±liq1 
K5/51, K5/59, K5/68   Grt   ±Ilm Rt   Opx Prg, Hbl  Pl  ±Ap  
K5/31   Grt  Ti-Mag    Di  Prg  Pl    
K5/45, K5/56, K5/70, k5/72, 827, 846, 797, 690, 741, 650, Grt  ±Ti-Mag ±Ilm   Di Opx Prg, Hbl ±Bt Pl  ±Ap  
K5/73, K5/75, K5/77, K9/24, 844, 867, 805 734, 762, 695 
K9/35, K9/70 
K5/52, K5/54, K9/23b 800, 827 668, 761 Grt Hc-Mag ±Ti-Mag    Di Opx Prg, Hbl  Pl ±Scp1 ±Ap  
K5/64, K9/23a, K9/27 827, 1230 761, 1082 Grt Hc-Mag     Di Opx Prg, Hbl     ±liq1 
K5/47, K5/67 >s >s  Hc-Mag ±Ti-Mag    Di Opx Prg  Pl ±Scp1 Ap  
K9/29, K9/69 >s, >s >s, >s  ±Hc-Mag ±Ti-Mag    Di  Prg  Pl ±Scp1 Ap liq1 
K5/58, K9/21, K9/22, K9/31 >s, >s, >s >s, >s, >s  ±Hc-Mag ±Ti-Mag    Di  Prg   ±Scp1 ±Scp2 Ap liq1 
K9/32     Ti-Mag   Ol Di Opx Prg      
K9/33 >s >s   Ti-Mag   Ol Di  Prg     liq1 
Sample no. Bt–Grt Phase associations in Grt cores Corona Matrix Replacement of Grt + Bt by 
 thermometry        around           
 Mineral rims
 
       Grt
 
          
K4/10 792 (IM-A) Grt Hc1 Sil Qtz Pl1 Rt liq1 Bt Pl2 Qtz Rt Ilm Zrn liq3 Opx Hc2 Pl3 liq2 
 759 (IM-B) 
 747 (PL) 
 684 (B) 
Sample no. Bt–Grt Phase associations in Grt cores Corona Matrix Replacement of Grt + Bt by 
 thermometry        around           
 Mineral rims
 
       Grt
 
          
K4/10 792 (IM-A) Grt Hc1 Sil Qtz Pl1 Rt liq1 Bt Pl2 Qtz Rt Ilm Zrn liq3 Opx Hc2 Pl3 liq2 
 759 (IM-B) 
 747 (PL) 
 684 (B) 

The mafic and ultramafic phase assemblages contain additional pyrrhotite. Sample K9/23 is a spinel- and garnet-bearing, composite xenolith (the spinel is a hercynite–magnetite solid solution), consisting of a garnet–spinel websterite dyke (K9/23a) in a garnet–spinel pyriclasite (K9/23b). Temperatures derived from mineral core compositions are ∼20–80°C below rim temperatures. T Wells, clinopyroxene–orthopyroxene thermometry after Wells (1977); T BK, orthopyroxene–clinopyroxene thermometry of Brey & Köhler (1990). Garnet–biotite thermometry (mineral rims): IM-A, IM-B, Indares & Martignole (1985), models A and B; PL, Perchuk & Lavrent’eva (1983); B, Bhattacharya et al. (1992).

All studied xenoliths show no or low degrees of alteration. Glasses (<1 vol. %) occurring along grain boundaries and in interstitial pockets are fresh or only partly altered into sheet silicates or palagonite. Minerals are completely fresh, except olivine, which is partly or completely iddingsitized (sample K9/33).

The meta-igneous granulites are composed of clinopyroxene, pargasite or pargasitic hornblende, ± garnet, ± spinel (hercynite–spinel–magnetite solid solutions), ± orthopyroxene, ± biotite, ± plagioclase, ± scapolite, ± glass, ± magnetite, ± ilmenite, ± rutile, and composite granulite xenoliths with websterite dykelets. Meta-clinopyroxenites and meta-hornblendites are dominated by clinopyroxene and pargasitic amphibole, respectively, and have variable contents of olivine (two samples only), hercynitic spinel, scapolite, plagioclase, apatite, Ti-magnetite and glass. All studied xenoliths contain high-density CO2-dominated fluid inclusions. The spinel-bearing meta-igneous phase associations and the garnet–sillimanite gneisses have not been described previously.

Metasomatism is a texturally late-stage process and includes hydration and dehydration reactions. From their petrographic appearance, the xenoliths can be subdivided into enriched (metasomatized) and primitive types as follows:

  1. an enriched type E, with modal metasomatic overprinting, which can be subdivided into (1a) subtype E1 with replacement structures of pyroxene and garnet by hydrous phases amphibole and biotite, and (1b) subtype E2 defined by breakdown of hydrous phases;

  2. a primitive type P generally lacking type E replacement textures.

Cumulate xenoliths

The cumulate xenoliths comprise gabbros and phlogopite clinopyroxenites. We have chosen two representative rocks for detailed descriptions.

The first is a gabbroic heteradcumulate (K9/34), which has the bulk composition of a basalt (see Table 4, below), using the total alkalis vs silica diagram of Le Maitre et al. (1989). Cumulus phases crystallized in the order: Ti-Mag + Ap + Cpx, Ttn, Am (compositions are given in Table 2). They are medium grained (1–2 mm) and scattered without common orientation in coarse-grained (5 mm) potassic oligoclase (An31Ab56Or13), defining an ophitic texture. Clinopyroxene is zoned from pleochroic light green cores (En29Fs22Wo49) to darker greenish rims. Except for sub- to anhedral amphibole and plagioclase, minerals have a subhedral to euhedral morphology, with prismatic clinopyroxene and apatite, octahedral Ti-magnetite and sphenoidal titanite.

Table 2:

Mineral analyses (EMP, electron-microprobe; SYXRF, synchrotron-XRF microprobe)

Sample: K9/34
 
K9/25
 
Phase: GI in Interstitial Cpx Am Pl core Ti–Mag Tit core Ap core Interstitial Cpx core Bt core 
 Cpx
 
glass
 
core
 
core
 
 core
 
  glass
 
  
wt % EMP 
SiO2 53·71 55·13±0·27 45·41  38·71 62·56  29·10  38·91±0·22 44·66   36·47 
TiO2  2·52  1·59±0·03  2·25   4·50  0·05    9·52 34·74   2·88±0·09  2·65    5·40 
Al2O3 18·28 15·94±0·15  6·59  12·09 22·71    3·19  1·77  15·18±0·13  9·34   16·87 
CaO  3·74  6·46±0·11 21·60  11·45  5·22    0·00 27·85 55·30 11·88±0·61 23·71    0·15 
FeO*  5·98  6·30±0·16 12·20  15·35  0·40   79·11  1·42  0·24 10·58±0·48  6·32    8·27 
MgO  0·89  1·75±0·04  9·32  10·14  0·04    2·40  0·00  0·14  3·71±0·04 12·60   18·78 
MnO  0·32  0·35±0·05  0·51   0·58  0·00    1·27  0·12  0·01  0·31±0·01  0·07    0·04 
Na2 7·07  6·60±0·04  1·51   2·50  5·26   0·00   5·52±0·08  0·40    0·44 
K2 3·38  3·57±0·04  0·00   1·92  1·80   0·00   6·32±0·34  0·00    9·22 
P2O5  0·33  0·29±0·02      40·00  1·32±0·02     0·04 
 0·47  0·22±0·02    0·04     3·00  0·52±0·22     0·50 
Cl  0·25  0·21±0·02    0·05     0·21  0·38±0·01     0·04 
SO2*  0·40  0·31±0·03       0·93  0·44±0·02     0·00 
Sum 96·93 98·73 99·39  97·33 98·02   95·49 95·00 99·83 97·95 99·77   96·22 
Cuμg/gEMP  840 
Zn    150  2050      
Ni    460   660      
Ba 5320 
   En29 XMgAn31     En38 Phl80 
   Fs22 0·54 Ab56     Fs11 Ann20 
   Wo49  Or13     Wo51  
Sample: K9/34
 
K9/25
 
Phase: GI in Interstitial Cpx Am Pl core Ti–Mag Tit core Ap core Interstitial Cpx core Bt core 
 Cpx
 
glass
 
core
 
core
 
 core
 
  glass
 
  
wt % EMP 
SiO2 53·71 55·13±0·27 45·41  38·71 62·56  29·10  38·91±0·22 44·66   36·47 
TiO2  2·52  1·59±0·03  2·25   4·50  0·05    9·52 34·74   2·88±0·09  2·65    5·40 
Al2O3 18·28 15·94±0·15  6·59  12·09 22·71    3·19  1·77  15·18±0·13  9·34   16·87 
CaO  3·74  6·46±0·11 21·60  11·45  5·22    0·00 27·85 55·30 11·88±0·61 23·71    0·15 
FeO*  5·98  6·30±0·16 12·20  15·35  0·40   79·11  1·42  0·24 10·58±0·48  6·32    8·27 
MgO  0·89  1·75±0·04  9·32  10·14  0·04    2·40  0·00  0·14  3·71±0·04 12·60   18·78 
MnO  0·32  0·35±0·05  0·51   0·58  0·00    1·27  0·12  0·01  0·31±0·01  0·07    0·04 
Na2 7·07  6·60±0·04  1·51   2·50  5·26   0·00   5·52±0·08  0·40    0·44 
K2 3·38  3·57±0·04  0·00   1·92  1·80   0·00   6·32±0·34  0·00    9·22 
P2O5  0·33  0·29±0·02      40·00  1·32±0·02     0·04 
 0·47  0·22±0·02    0·04     3·00  0·52±0·22     0·50 
Cl  0·25  0·21±0·02    0·05     0·21  0·38±0·01     0·04 
SO2*  0·40  0·31±0·03       0·93  0·44±0·02     0·00 
Sum 96·93 98·73 99·39  97·33 98·02   95·49 95·00 99·83 97·95 99·77   96·22 
Cuμg/gEMP  840 
Zn    150  2050      
Ni    460   660      
Ba 5320 
   En29 XMgAn31     En38 Phl80 
   Fs22 0·54 Ab56     Fs11 Ann20 
   Wo49  Or13     Wo51  
Sample: K9/29
 
K5/71
 
Phase: Sp
 
Sp
 
Cpx
 
Am
 
Opx Opx Cpx Rt Am 
 Phase association: Sp + Cpx + Am + Core, Rim to Rim to  Rim, late- 
 vesicular glass replaces Cpx Opx  stage corona 
     Am
 
   around Opx
 
wt % EMP 
SiO2    0·0   0·0 50·55   39·33  53·47  54·11  52·90   0·0  46·03 
TiO2    0·14   0·03  0·33    1·80   0·10   0·05   0·24  98·36   0·02 
Al2O3   56·41  53·51  5·12   14·54   2·19   1·37   2·09   0·17  10·44 
CaO    0·18   0·11 20·77   10·46   0·35   0·67  23·02   0·61  11·59 
FeO*   27·15  32·02 10·70   16·88  15·68  15·10   5·04   0·16   8·52 
MgO   14·61  13·07 10·08    9·04  27·74  28·18  15·98   16·62 
MnO    0·36   0·36  0·26    0·27   0·38   0·38   0·12    0·14 
Na2   1·68    2·93   0·0   0·02   0·42    1·84 
K2   0·0    1·29   0·0   0·0   0·0    0·39 
     <0·01      <0·005 
Cl       0·05       0·057 
Sum   98·85  99·10 99·49   96·59  99·91  99·88  99·81  99·30  95·647 
Method EMP EMP  EMP, EMP, EMP, EMP, EMP EMP, 
    SYXRF SYXRF SYXRF SYXRF  SYXRF 
Cuμg/gEMP  99 
Zn  800 800   630 <50 <50 <50 <50 396 
Ni  100 600  1040  85 <50 <50 <50 693 
Cr 1400 700   140  453 295 495  
   En32Fs19 XMgEn75Fs24 En76Fs23 En45Fs8  XMg = 0·78 
   Wo48    0·49 Wo1 Wo1 Wo47   
Sample: K9/29
 
K5/71
 
Phase: Sp
 
Sp
 
Cpx
 
Am
 
Opx Opx Cpx Rt Am 
 Phase association: Sp + Cpx + Am + Core, Rim to Rim to  Rim, late- 
 vesicular glass replaces Cpx Opx  stage corona 
     Am
 
   around Opx
 
wt % EMP 
SiO2    0·0   0·0 50·55   39·33  53·47  54·11  52·90   0·0  46·03 
TiO2    0·14   0·03  0·33    1·80   0·10   0·05   0·24  98·36   0·02 
Al2O3   56·41  53·51  5·12   14·54   2·19   1·37   2·09   0·17  10·44 
CaO    0·18   0·11 20·77   10·46   0·35   0·67  23·02   0·61  11·59 
FeO*   27·15  32·02 10·70   16·88  15·68  15·10   5·04   0·16   8·52 
MgO   14·61  13·07 10·08    9·04  27·74  28·18  15·98   16·62 
MnO    0·36   0·36  0·26    0·27   0·38   0·38   0·12    0·14 
Na2   1·68    2·93   0·0   0·02   0·42    1·84 
K2   0·0    1·29   0·0   0·0   0·0    0·39 
     <0·01      <0·005 
Cl       0·05       0·057 
Sum   98·85  99·10 99·49   96·59  99·91  99·88  99·81  99·30  95·647 
Method EMP EMP  EMP, EMP, EMP, EMP, EMP EMP, 
    SYXRF SYXRF SYXRF SYXRF  SYXRF 
Cuμg/gEMP  99 
Zn  800 800   630 <50 <50 <50 <50 396 
Ni  100 600  1040  85 <50 <50 <50 693 
Cr 1400 700   140  453 295 495  
   En32Fs19 XMgEn75Fs24 En76Fs23 En45Fs8  XMg = 0·78 
   Wo48    0·49 Wo1 Wo1 Wo47   
Sample: K9/27
 
Mineral: Grt core
 
Grt rim
 
Hc
 
Pl
 
Opx
 
Cpx
 
Am 
 Corona around Grt
 
Rim to Px+Hc
 
wt % EMP 
SiO2   40·03  40·32  46·49 49·64  51·25 41·49 
TiO2    0·0   0·09    0·18  0·0  0·0   0·09  0·22 
Al2O3   22·50  22·81   57·58 34·11  7·27   4·14 16·04 
CaO    4·12   3·86    0·07 17·45  1·73  13·58 10·68 
FeO*   19·32  18·70   27·97  0·30 17·40  11·61 10·07 
MgO   13·12  13·74   12·52  0·02 22·45  17·84 14·52 
MnO    0·90   0·88    0·57  0·01  0·89   0·59  0·15 
Na2   0·0   0·0   1·45  0·0   0·16  2·58 
K2   0·0   0·0   0·02    0·29 
Sum   99·99 100·40   98·91 99·85 99·38  99·26 96·04 
Zn(μg/g EMP) 1200 700  300   600  
Ni    500   100  
Cr  400 100 1000     
 Prp47·8 Prp50·0  Ab13An87 En67Fs29 En52Fs19 XMg = 0·72 
 Alm39·5 Alm38·1  Or0 Wo4 Wo29  
 Grs10·6 Grs9·9      
 Sps1·9 Sps1·8      
 Adr0·2 And0·0      
Sample: K9/27
 
Mineral: Grt core
 
Grt rim
 
Hc
 
Pl
 
Opx
 
Cpx
 
Am 
 Corona around Grt
 
Rim to Px+Hc
 
wt % EMP 
SiO2   40·03  40·32  46·49 49·64  51·25 41·49 
TiO2    0·0   0·09    0·18  0·0  0·0   0·09  0·22 
Al2O3   22·50  22·81   57·58 34·11  7·27   4·14 16·04 
CaO    4·12   3·86    0·07 17·45  1·73  13·58 10·68 
FeO*   19·32  18·70   27·97  0·30 17·40  11·61 10·07 
MgO   13·12  13·74   12·52  0·02 22·45  17·84 14·52 
MnO    0·90   0·88    0·57  0·01  0·89   0·59  0·15 
Na2   0·0   0·0   1·45  0·0   0·16  2·58 
K2   0·0   0·0   0·02    0·29 
Sum   99·99 100·40   98·91 99·85 99·38  99·26 96·04 
Zn(μg/g EMP) 1200 700  300   600  
Ni    500   100  
Cr  400 100 1000     
 Prp47·8 Prp50·0  Ab13An87 En67Fs29 En52Fs19 XMg = 0·72 
 Alm39·5 Alm38·1  Or0 Wo4 Wo29  
 Grs10·6 Grs9·9      
 Sps1·9 Sps1·8      
 Adr0·2 And0·0      
Sample: K5/58
 
Phase: Scp 2 core, porphyroclast
 
Method: EMP,  SYXRF 
 CO2by   
 difference
 
  
wt %  μg/g  
SiO2  48·83 Cu   <1 
TiO2   0·07 Zn   22·2 
Al2O3  26·86 Ni   <1 
CaO  15·86 Cr   <1 
FeO*   0·26 Ga   15·8 
MgO   0·0 Br   28·6 
MnO   0·00 Rb   32·6 
Na2  4·02 Sr 8275·7 
K2  0·27   67·5 
Cl   0·421 Zr  104 
SO2*   0·594 Nb   19 
CO2*   2·815 Sn   <2 
Sum 100 Ba  264 
  La   26 
  Ce   52 
  Ta   <5 
  Pb    3 
 Mei68Ma32   
Sample: K5/58
 
Phase: Scp 2 core, porphyroclast
 
Method: EMP,  SYXRF 
 CO2by   
 difference
 
  
wt %  μg/g  
SiO2  48·83 Cu   <1 
TiO2   0·07 Zn   22·2 
Al2O3  26·86 Ni   <1 
CaO  15·86 Cr   <1 
FeO*   0·26 Ga   15·8 
MgO   0·0 Br   28·6 
MnO   0·00 Rb   32·6 
Na2  4·02 Sr 8275·7 
K2  0·27   67·5 
Cl   0·421 Zr  104 
SO2*   0·594 Nb   19 
CO2*   2·815 Sn   <2 
Sum 100 Ba  264 
  La   26 
  Ce   52 
  Ta   <5 
  Pb    3 
 Mei68Ma32   
Sample: K 4/10
 
Phase: Grt porphyro- Grt porphyro- Bt corona Rt porphyro- Ilm corona Pl neoblast 
 clast centre clast rim to Bt around Gt clast centre around Rt in matrix 
   porphyroclast
 
   
wt % 
SiO2  38·68  38·77  36·22  0·0  0·0  59·36 
TiO2   0·0   0·0   5·1 98·07 49·94   0·0 
Al2O3  21·63  21·78  16·47  0·26  0·53  26·02 
V2O3     0·84   
FeO*  27·85  27·91  15·35  0·57 43·91   0·19 
MnO   0·81   0·92   0·1  0·03  0·86   0·0 
MgO   8·27   9·11  13·01  0·0  2·77   0·0 
CaO   2·71   2·01   0·03  0·02  0·01   7·98 
Na2  0·0   0·0   0·05  0·0  0·0   6·43 
K2  0·0   0·0   9·64  0·0  0·0   0·54 
Sum 100·07 100·65  95·97 99·88 98·03 100·51 
μg/g 
Ni  37  43  <1 <1   
Cu  <1  <0·1  <1 55   
Zn 110 122 297 65   
Ga  <1  <0·1  29 <0·1   
Br  <0·1  <0·1  <0·1 <0·1   
Rb   1   6 199 <0·1   
Sr   2   4  23 <0·1   
316 607  <1 <1   
Zr  14  35  14 18   
Nb  <1  <1   2 35   
 Prp34·4 Prp34·1 Phl60   An39 
 Alm59·4 Alm59·5 Ann40   Ab57 
 Grs7·2 Grs5·2    Or3 
 Sps1·7 Sps2·0     
 Adr0·1 Adr0·2     
 Ti-Grt0·0 Ti–Grt0·0     
Sample: K 4/10
 
Phase: Grt porphyro- Grt porphyro- Bt corona Rt porphyro- Ilm corona Pl neoblast 
 clast centre clast rim to Bt around Gt clast centre around Rt in matrix 
   porphyroclast
 
   
wt % 
SiO2  38·68  38·77  36·22  0·0  0·0  59·36 
TiO2   0·0   0·0   5·1 98·07 49·94   0·0 
Al2O3  21·63  21·78  16·47  0·26  0·53  26·02 
V2O3     0·84   
FeO*  27·85  27·91  15·35  0·57 43·91   0·19 
MnO   0·81   0·92   0·1  0·03  0·86   0·0 
MgO   8·27   9·11  13·01  0·0  2·77   0·0 
CaO   2·71   2·01   0·03  0·02  0·01   7·98 
Na2  0·0   0·0   0·05  0·0  0·0   6·43 
K2  0·0   0·0   9·64  0·0  0·0   0·54 
Sum 100·07 100·65  95·97 99·88 98·03 100·51 
μg/g 
Ni  37  43  <1 <1   
Cu  <1  <0·1  <1 55   
Zn 110 122 297 65   
Ga  <1  <0·1  29 <0·1   
Br  <0·1  <0·1  <0·1 <0·1   
Rb   1   6 199 <0·1   
Sr   2   4  23 <0·1   
316 607  <1 <1   
Zr  14  35  14 18   
Nb  <1  <1   2 35   
 Prp34·4 Prp34·1 Phl60   An39 
 Alm59·4 Alm59·5 Ann40   Ab57 
 Grs7·2 Grs5·2    Or3 
 Sps1·7 Sps2·0     
 Adr0·1 Adr0·2     
 Ti-Grt0·0 Ti–Grt0·0     

Table 4:

Selected whole-rock analyses of major (wt %) and trace elements (μg/g)

Sample: K9/25 K9/34 K4/10 Average of K9/30 K5/31 K5/52 K5/71 K5/65 K5/64 K5/62 K9/27 K9/35 K5/70 K9/32 K9/23 K9/22 K9/21 
     11 samples              
Rock type: Cumulate Cumulate Grt–Sil- ‘Typical’ E1 E1 E1 E1 E1 E1 E1 E1 E2 E2 
   gneiss, E1
 
P, SD
 
              
wt % 
SiO2  41·68   47·00  45·00    53·54  54·9  48·7  50·21  48·03  47·24  42·00  43·34   45·23  47·01   46·97  46·12  41·18  43·19 
TiO2   3·51    2·35   1·31   1·01±0·52    0·36   0·8   0·5   0·48   1   0·31   3·33   0·37    2·06   0·71    0·68   1·325   1·735   3·37 
Al2O3  11·63   14·53  22·7     6·01  22·83  12·75  10·87  10·73  16·91  16·31  17·59   10·5  13·31    6·59  14·18  10·43   9·49 
Fe2O3   2·77    6·22   1·55   2·11±1·38    0·75   0·6   1·22   0·0   1·76   0·75   6·42   1·54    2·41   0·98    2·51   1·13   6·77   4·48 
MgO  12·79    4·64   4·8    22·41   1·91  12·61  14·96  13·47  11·78   7·66  14·23   14·37  13·86   19·13  14·32  10·36  13·61 
FeO   5·43    4·24  13·08     9·53   4·02   8·95   7·14   6·74   6·4   9·29   7·43   11·16   7·86    8·95   8·77   7·17   9·71 
MnO   0·10    0·23   0·39   0·15±0·02    0·22   0·09   0·18   0·16   0·36   0·12   0·21   0·13    0·21   0·17    0·20   0·14   0·27   0·18 
CaO  15·70   14·22   4·28     5·64   7·69  10·84  13·33  14·19  13·65  10·89  10·92   10·10  13·38   12·71  10·62  15·94  10·80 
Na2  0·74    3·35   2·69     0·95   4·82   1·62   1·12   1·82   1·13   1·93   2·36    1·60   1·18    0·94   1·84   2·01   2·20 
K2  2·76    0·65   0·61   0·13±0·06    0·08   0·35   0·13   0·06   0·25   0·05   0·10   0·24    0·30   0·57    0·19   0·14   0·46   0·64 
P2O5   0·15    0·73   0·23   0·36±0·15    0·09   0·15   0·26   0·34   0·23   0·19   0·14   0·16    0·23   0·26    0·14   0·12   0·79   0·33 
Sum  97·26   98·16  96·64    99·58  98·16  97·75  98·67  98·58  98·53  98·28  98·31   98·17  99·29   99·01  98·71  97·11  98·00 
XMg   0·81    0·66   0·40   0·6–0·8    0·81   0·46   0·72   0·79   0·78   0·77   0·60   0·77    0·70   0·76    0·79   0·74   0·72   0·71 
μg/g 
309  697  52   200  40 326  56 115 133 291  34   57  87   60  62 164  89 
Ga  20   21  22     8  23  14  15  11  14  22  12   15  14   11  13  20  18 
As   3         7  33       4  11    6   7    6 
Sc  37·2   20  69·9  33±6·3   24   36·3  39·4  44·7  30·1  37  11·7   25·8  41·4   46·3  29·1  42·4  29·5 
317  344 229 254±96  116  90 208 156 223 100 629  78  293 180  184 237 402 342 
Cr  62    3 220 280±214 1841  599 709 703 532  50 309 1005 295 1612 924 185 709 
Co  43·7   29  12·5  42±6·5   62·2  12  47·6  45·8  38·1  47·9  42  62   54·2  55·9   68·2  55·9  51·7  62·8 
Ni  75   27  21  90±36·5  492  33  90 184 121 164  67 338  366 184  332 339 128 480 
Cu 130·5   73   6·4  32±10·5   43·7  13  34  23·2  18·2  23·9 145   3·1    9·1   5·2   20  55  78·6  18·2 
Zn  95·5  104  48·4  78·6±25·5   74·2  50  97·6  50·3  55·7  44·7  99  39·7  123·4  54·6   71·3  63·5 130·4 134·3 
Rb   5    7   1±0    1     1   3   1    1    1   4    2   1   2   4 
Sr 488 1810 410 442±176  123 839 355 661 242 385 393 135  185  86  154 135 384 111 
 22·6   20 178·7  15·1±5·3    5·2   13·8  10·5  16·7   6·4  12   5·8   25·9  13·7   16  17·5  22  49·4 
Zr 223  387  23  38·9±16·1   11 705  34  21  37  13  28  13   42  42   68  40 185  47 
Nb  31  197  16   2±0·7    1    3   2   3   3    2   16   5    3   4 114  55 
Ba 242  560 442 137±53   76 381 107 280 105 133 148  91   80 107  183  38 221 117 
La  34·4  107  33·9   7±0·7    3·38   45   7·63   7·36   7·58   4·39    3·65   13·5   6·17   10·8   2·37  30·3  23·3 
Ce  84·2  174  54·5  16·3±1·3    6·39  49  18·2  15·6  18·7   7·4  36   6·9   37·7  14·5   25·2   6·9  94·6  73·2 
Pr  12·2   44   5·61     0·97    2·81   2·18   3·05   1·02    0·93    6·15   2·07    3·55   1·35  12·71  11·45 
Nd  50·5   96  20·0  10·5±1·1    3·97   12·2   9·2  14·5   5·1    4·26   28·7   9·31   15·8   7·71  53·6  53·5 
Sm   9·21    5·96     1·01    2·87   2·14   3·68   1·27    1·03    6·8   2·46    3·46   2·53   9·98  12·6 
Eu   2·55    2·41     0·35 101   0·79   0·68   1·11   0·56    0·51    1·96   0·75    0·99   0·95   2·74   3·1 
Gd   7·4   13·2     1·02    2·9   2·19   3·63   1·35    1·16    6·42   2·7    3·47   3·08   7·98  11·6 
Tb   1·04    3·4     0·18    0·46   0·35   0·6   0·22    0·19    0·98   0·46    0·54   0·53   1·08   1·83 
Dy   5·36   28·3     1·02    2·79   2·06   3·57   1·36    1·19    5·58   2·77    3·27   3·39   5·31  10·6 
Ho   0·96    6·91     0·22    0·56   0·42   0·7   0·27    0·25    1·03   0·55    0·63   0·69   0·92   1·97 
Er   2·52   22·8     0·61    1·65   1·16   1·93   0·73    0·67    2·83   1·53    1·81   1·96   2·4   5·45 
Tm   0·33    3·47     0·1    0·23   0·16   0·27   0·1    0·1    0·38   0·21    0·25   0·27   0·3   0·75 
Yb   2·11   24·8   1·2±0·3    0·63    1·61   1·12   1·69   0·62    0·64    2·47   1·32    1·62   1·75   1·88   4·82 
Lu   0·69    3·72     0·1    0·23   0·16   0·24   0·08    0·1    0·34   0·18    0·23   0·25   0·27   0·66 
Th   2·63    1·9   0·07±0·03    0·1    0·07   0·12   0·2   0·39    0·28    0·35   0·54    0·49   0·08   1·87   0·8 
  0·44    0·35   0·06±0·02    0·06    0·04   0·08   0·1   0·4    0·3    0·07   0·1    0·21   0·04   0·3   0·13 
Pb   1·75   11·3   2·1±0·3    1·86    2·37   1·85   2·98   1·12    2·51    1·07   1·24    2·95   0·5   1·2   1·71 
Mo     1       1   2     1    1   1    1   
Sample: K9/25 K9/34 K4/10 Average of K9/30 K5/31 K5/52 K5/71 K5/65 K5/64 K5/62 K9/27 K9/35 K5/70 K9/32 K9/23 K9/22 K9/21 
     11 samples              
Rock type: Cumulate Cumulate Grt–Sil- ‘Typical’ E1 E1 E1 E1 E1 E1 E1 E1 E2 E2 
   gneiss, E1
 
P, SD
 
              
wt % 
SiO2  41·68   47·00  45·00    53·54  54·9  48·7  50·21  48·03  47·24  42·00  43·34   45·23  47·01   46·97  46·12  41·18  43·19 
TiO2   3·51    2·35   1·31   1·01±0·52    0·36   0·8   0·5   0·48   1   0·31   3·33   0·37    2·06   0·71    0·68   1·325   1·735   3·37 
Al2O3  11·63   14·53  22·7     6·01  22·83  12·75  10·87  10·73  16·91  16·31  17·59   10·5  13·31    6·59  14·18  10·43   9·49 
Fe2O3   2·77    6·22   1·55   2·11±1·38    0·75   0·6   1·22   0·0   1·76   0·75   6·42   1·54    2·41   0·98    2·51   1·13   6·77   4·48 
MgO  12·79    4·64   4·8    22·41   1·91  12·61  14·96  13·47  11·78   7·66  14·23   14·37  13·86   19·13  14·32  10·36  13·61 
FeO   5·43    4·24  13·08     9·53   4·02   8·95   7·14   6·74   6·4   9·29   7·43   11·16   7·86    8·95   8·77   7·17   9·71 
MnO   0·10    0·23   0·39   0·15±0·02    0·22   0·09   0·18   0·16   0·36   0·12   0·21   0·13    0·21   0·17    0·20   0·14   0·27   0·18 
CaO  15·70   14·22   4·28     5·64   7·69  10·84  13·33  14·19  13·65  10·89  10·92   10·10  13·38   12·71  10·62  15·94  10·80 
Na2  0·74    3·35   2·69     0·95   4·82   1·62   1·12   1·82   1·13   1·93   2·36    1·60   1·18    0·94   1·84   2·01   2·20 
K2  2·76    0·65   0·61   0·13±0·06    0·08   0·35   0·13   0·06   0·25   0·05   0·10   0·24    0·30   0·57    0·19   0·14   0·46   0·64 
P2O5   0·15    0·73   0·23   0·36±0·15    0·09   0·15   0·26   0·34   0·23   0·19   0·14   0·16    0·23   0·26    0·14   0·12   0·79   0·33 
Sum  97·26   98·16  96·64    99·58  98·16  97·75  98·67  98·58  98·53  98·28  98·31   98·17  99·29   99·01  98·71  97·11  98·00 
XMg   0·81    0·66   0·40   0·6–0·8    0·81   0·46   0·72   0·79   0·78   0·77   0·60   0·77    0·70   0·76    0·79   0·74   0·72   0·71 
μg/g 
309  697  52   200  40 326  56 115 133 291  34   57  87   60  62 164  89 
Ga  20   21  22     8  23  14  15  11  14  22  12   15  14   11  13  20  18 
As   3         7  33       4  11    6   7    6 
Sc  37·2   20  69·9  33±6·3   24   36·3  39·4  44·7  30·1  37  11·7   25·8  41·4   46·3  29·1  42·4  29·5 
317  344 229 254±96  116  90 208 156 223 100 629  78  293 180  184 237 402 342 
Cr  62    3 220 280±214 1841  599 709 703 532  50 309 1005 295 1612 924 185 709 
Co  43·7   29  12·5  42±6·5   62·2  12  47·6  45·8  38·1  47·9  42  62   54·2  55·9   68·2  55·9  51·7  62·8 
Ni  75   27  21  90±36·5  492  33  90 184 121 164  67 338  366 184  332 339 128 480 
Cu 130·5   73   6·4  32±10·5   43·7  13  34  23·2  18·2  23·9 145   3·1    9·1   5·2   20  55  78·6  18·2 
Zn  95·5  104  48·4  78·6±25·5   74·2  50  97·6  50·3  55·7  44·7  99  39·7  123·4  54·6   71·3  63·5 130·4 134·3 
Rb   5    7   1±0    1     1   3   1    1    1   4    2   1   2   4 
Sr 488 1810 410 442±176  123 839 355 661 242 385 393 135  185  86  154 135 384 111 
 22·6   20 178·7  15·1±5·3    5·2   13·8  10·5  16·7   6·4  12   5·8   25·9  13·7   16  17·5  22  49·4 
Zr 223  387  23  38·9±16·1   11 705  34  21  37  13  28  13   42  42   68  40 185  47 
Nb  31  197  16   2±0·7    1    3   2   3   3    2   16   5    3   4 114  55 
Ba 242  560 442 137±53   76 381 107 280 105 133 148  91   80 107  183  38 221 117 
La  34·4  107  33·9   7±0·7    3·38   45   7·63   7·36   7·58   4·39    3·65   13·5   6·17   10·8   2·37  30·3  23·3 
Ce  84·2  174  54·5  16·3±1·3    6·39  49  18·2  15·6  18·7   7·4  36   6·9   37·7  14·5   25·2   6·9  94·6  73·2 
Pr  12·2   44   5·61     0·97    2·81   2·18   3·05   1·02    0·93    6·15   2·07    3·55   1·35  12·71  11·45 
Nd  50·5   96  20·0  10·5±1·1    3·97   12·2   9·2  14·5   5·1    4·26   28·7   9·31   15·8   7·71  53·6  53·5 
Sm   9·21    5·96     1·01    2·87   2·14   3·68   1·27    1·03    6·8   2·46    3·46   2·53   9·98  12·6 
Eu   2·55    2·41     0·35 101   0·79   0·68   1·11   0·56    0·51    1·96   0·75    0·99   0·95   2·74   3·1 
Gd   7·4   13·2     1·02    2·9   2·19   3·63   1·35    1·16    6·42   2·7    3·47   3·08   7·98  11·6 
Tb   1·04    3·4     0·18    0·46   0·35   0·6   0·22    0·19    0·98   0·46    0·54   0·53   1·08   1·83 
Dy   5·36   28·3     1·02    2·79   2·06   3·57   1·36    1·19    5·58   2·77    3·27   3·39   5·31  10·6 
Ho   0·96    6·91     0·22    0·56   0·42   0·7   0·27    0·25    1·03   0·55    0·63   0·69   0·92   1·97 
Er   2·52   22·8     0·61    1·65   1·16   1·93   0·73    0·67    2·83   1·53    1·81   1·96   2·4   5·45 
Tm   0·33    3·47     0·1    0·23   0·16   0·27   0·1    0·1    0·38   0·21    0·25   0·27   0·3   0·75 
Yb   2·11   24·8   1·2±0·3    0·63    1·61   1·12   1·69   0·62    0·64    2·47   1·32    1·62   1·75   1·88   4·82 
Lu   0·69    3·72     0·1    0·23   0·16   0·24   0·08    0·1    0·34   0·18    0·23   0·25   0·27   0·66 
Th   2·63    1·9   0·07±0·03    0·1    0·07   0·12   0·2   0·39    0·28    0·35   0·54    0·49   0·08   1·87   0·8 
  0·44    0·35   0·06±0·02    0·06    0·04   0·08   0·1   0·4    0·3    0·07   0·1    0·21   0·04   0·3   0·13 
Pb   1·75   11·3   2·1±0·3    1·86    2·37   1·85   2·98   1·12    2·51    1·07   1·24    2·95   0·5   1·2   1·71 
Mo     1       1   2     1    1   1    1   

Clinopyroxene, titanite and Ti-magnetite contain primary glass inclusions of dark to light brown colour (Fig. 2a). The glass inclusions define two compositional groups: (1) tephrite–basanite to basalt, and (2) phonotephrite and tephriphonolite [using the classification of Le Maitre et al. (1989)]. An analysis of the glass inclusion in K9/34 with the highest S, Cl and F concentrations is given in Table 2. The glass inclusions are associated with high-density CO2-fluid inclusions indicating a cogenetic origin. Most glass inclusions show little or no evidence of partial crystallization but frequently contain fluid vesicles (0–10% of inclusion volume).

Fig. 2.

Photomicrographs. (a) Gabbroic cumulate K9/34, showing clinopyroxene in phonotephritic interstitial glass. The clinopyroxene contains tephritic glass inclusions (GI). (b) Garnet–sillimanite gneiss K4/10, polarized light, showing two sillimanite inclusions in the core of a garnet porphyroclast. One sillimanite contains a spherical inclusion of hercynitic spinel (black). The garnet core also contains inclusions of quartz. (c) Garnet pyriclasite K5/75, a type E1 xenolith. Orthopyroxene and garnet are partially replaced by a corona of pargasite. The xenoliths have highly variable corona sizes. The blackish corona around garnet is probably due to heating after incorporation of the xenolith by the host magma. (d) Meta-pyroxene hornblendite K9/29, showing a type E1 reaction: pargasite replaces older clinopyroxene. Close to the grain boundary, high-density CO2 fluid, glass and amphibole inclusions coexist in clinopyroxene. (e) Same sample as in (d). Pure CO2 inclusions (FI, partly decrepitated; FTIR, no water detected) and cogenetic glass inclusions (GI), consisting of brownish, partially altered, OH-bearing glass and a large vesicle filled with CO2. (f) Sample K5/58, showing three primary, vesicular glass inclusions in a scapolite porphyroclast (Mei68Ma32) and replacement of clinopyroxene by pargasite. The vesicles in the glass inclusions are filled with high-density liquid CO2. The occurrence of primary glass inclusions indicates hypersolidus conditions during scapolite formation. (g) Sample K5/58, showing porphyroclastic recrystallization and replacement structures. Clinopyroxene (neoblasts) is replaced by pargasite (neoblasts), followed by late-stage formation of scapolite (scapolite2, homogeneous variety; porphyroclasts and neoblasts). (h) Sample K9/27, showing hercynitic spinel and plagioclase + orthopyroxene (light), which have completely replaced garnet. The matrix consists of pargasitic hornblende and minor clinopyroxene and plagioclase.

Fig. 2.

Photomicrographs. (a) Gabbroic cumulate K9/34, showing clinopyroxene in phonotephritic interstitial glass. The clinopyroxene contains tephritic glass inclusions (GI). (b) Garnet–sillimanite gneiss K4/10, polarized light, showing two sillimanite inclusions in the core of a garnet porphyroclast. One sillimanite contains a spherical inclusion of hercynitic spinel (black). The garnet core also contains inclusions of quartz. (c) Garnet pyriclasite K5/75, a type E1 xenolith. Orthopyroxene and garnet are partially replaced by a corona of pargasite. The xenoliths have highly variable corona sizes. The blackish corona around garnet is probably due to heating after incorporation of the xenolith by the host magma. (d) Meta-pyroxene hornblendite K9/29, showing a type E1 reaction: pargasite replaces older clinopyroxene. Close to the grain boundary, high-density CO2 fluid, glass and amphibole inclusions coexist in clinopyroxene. (e) Same sample as in (d). Pure CO2 inclusions (FI, partly decrepitated; FTIR, no water detected) and cogenetic glass inclusions (GI), consisting of brownish, partially altered, OH-bearing glass and a large vesicle filled with CO2. (f) Sample K5/58, showing three primary, vesicular glass inclusions in a scapolite porphyroclast (Mei68Ma32) and replacement of clinopyroxene by pargasite. The vesicles in the glass inclusions are filled with high-density liquid CO2. The occurrence of primary glass inclusions indicates hypersolidus conditions during scapolite formation. (g) Sample K5/58, showing porphyroclastic recrystallization and replacement structures. Clinopyroxene (neoblasts) is replaced by pargasite (neoblasts), followed by late-stage formation of scapolite (scapolite2, homogeneous variety; porphyroclasts and neoblasts). (h) Sample K9/27, showing hercynitic spinel and plagioclase + orthopyroxene (light), which have completely replaced garnet. The matrix consists of pargasitic hornblende and minor clinopyroxene and plagioclase.

The latest phase is interstitial, clear light brownish vesicular phonotephrite glass (Table 2, under K9/34). Vesicles in the glass are partly filled with sheet-silicates, variably enriched by Cu (up to 1 wt %). Fluorine is enriched in the glass adjacent to vesicles and cracks.

The second representative rock, ultramafic cumulate (K9/25), has a nephelinitic bulk composition (see Table 4, below). It consists of early crystallized cumulus clinopyroxene and minor late-stage phlogopite (Phl80Ann20), and vesicular foiditic interstitial glass (Table 2). Clinopyroxenes are zoned, with pleochroic light brownish–greenish cores (En38Fs11Wo51), and brownish outer rims. Cumulus phases are medium grained (average grain size 1–3 mm) and sub- to euhedral, defining a euhedral granular texture. Clinopyroxene has a prismatic, and phlogopite a tabular morphology, without preferred orientation.

Glass inclusions show strong conspicuous crystallization. The glass inclusions are apparently cogenetic with high-density CO2-fluid inclusions.

Meta-igneous xenoliths

The granulite xenoliths are dominated by meta-igneous, mafic to ultramafic phase associations and represent products of phase transitions of meta-gabbroic rocks within the granulite facies above ∼500 MPa. Table 1 shows the phase associations roughly in the order of increasing equilibration pressure and temperature. The petrography of some of the mafic xenolith types (garnet granulites) has been described earlier (Okrusch et al., 1979; Voll, 1983; Loock et al., 1990; Mengel et al., 1991).

Mafic granulites have granoblastic inequigranular to equigranular microstructures [for microstructural terms and definitions we follow Passchier & Trouw (1996)]. Plagioclase in mafic, equigranular granulite xenoliths shows undulose extinction, kinking, tapering deformation twins, bent twins and deformation bands. The rocks have suffered different degrees of recrystallization of plagioclase, scapolite, pyroxene and amphibole. Mantles of plagioclase neoblasts are common, with a sharp boundary around cores of old grains, whereby typical core-and-mantle structures have developed, characterized by a well-defined bimodal grain size distribution and by a large difference in grain size.

Modal abundances of volatile-bearing phases are highly variable; amphibole ranges from 5 to 95 vol. %, biotite between 0 and 5 vol. % and scapolite from 0 to 30 vol. %. Amphibole occurs as texturally apparently primary grains, in apparent equilibrium with similar-sized clinopyroxene, as coronas around pyroxene and garnet, and as texturally early grains partially replaced by clinopyroxene or orthopyroxene. Biotite occurs associated with pargasitic hornblende as coronas around garnet. Scapolite occurs (1) as an apparently texturally equilibrated early phase (scapolite1), forming equigranular clusters, with numerous inclusions of pyrrhotite or haematite and surrounded by coronas of Pl + Hem, and (2) as a porphyroclastically recrystallized phase (scapolite2) with centimetre-sized porphyroclasts and no corona formation. The porphyroclasts show highly undulose extinction and occur in two varieties: as an optically homogeneous phase, and with ubiquitous pyrrhotite inclusions. Scapolite is the most important CO32−, SO42−, Cl and Br phase. Fourier transform IR (FTIR) analysis of homogeneous scapolite2 in sample K5/58 provides evidence for significant amounts of HSO4 [interpretation of the spectra following Swayze & Clark (1990)]. The scapolite2 porphyroclasts have a remarkably constant major element, sulphur (Table 2, sample K5/58) and fluorine composition. Chlorine concentrations decrease about 40% from the core to the rim, e.g. from 0·47 to 0·28 wt %, and bromine shows a decrease of about 80%, e.g. from 30 to 6 μg/g (sample K5/58).

Meta-pyroxenites and meta-hornblendites are porphyroclastically recrystallized or show mosaic structures. Clinopyroxene and pargasitic amphibole neoblasts frequently contain glass inclusions and are in contact with vesicular glass occurring in pockets, indicating recrystallization under hypersolidus conditions. Idiomorphic clinopyroxene and pargasitic hornblende in contact with glass provide evidence for continued growth from the melt after incorporation of the xenoliths by the magma.

Some meta-clinopyroxenitic and meta-hornblenditic xenoliths have compositions similar to the cumulate xenoliths but are distinguished from them by the appearance of scapolite and by deformation structures such as porphyroclastic recrystallization, deformation twinning of plagioclase, and undulose extinction and kinking of olivine, clinopyroxene, pargasitic amphibole, plagioclase and scapolite.

Garnet–sillimanite gneisses

The garnet–sillimanite gneiss xenolith K4/10 is macroscopically characterized by alternating layers of garnet (≥1 cm; 50 vol. %) and plagioclase (≥0·5 cm), and minor amounts (≤10 vol. %) of sillimanite, quartz, green spinel (Spl–Hc–Mag solid solution), biotite, rutile, ilmenite and glass. Quartz is not in contact with spinel. Garnet (rim Prp34Alm59Grs5Adr0Spess2; core Prp32Alm62Grs4-Adr0Spess2), plagioclase (An36–39Ab61–57Or3), rutile and quartz are highly porphyroclastically recrystallized, and the microstructure is inequigranular polygonal to inequigranular interlobate [after Moore (1970), in Passchier & Trouw (1996)]. Biotite (Phl59–60Ann41–40) forms coronas around garnet. Garnet and biotite are separated by a younger phase assemblage of vesicular glass + fine-grained (0·5 mm) Hc + Opx + Pl. Rutile (1 vol. %) occurs as small inclusions in garnet, and in the matrix as single millimetre-sized prismatic porphyroclasts and as clusters of neoblasts, frequently associated and epitactically intergrown with zircon. Matrix rutile in contact with plagioclase neoblasts and glass is surrounded by coronas of ilmenite. In the garnet cores, inclusions of sillimanite, quartz and green spinel (hercynite–spinel–magnetite solid solutions) and inclusions of prismatic sillimanite with spherical inclusions of dark green spinel (hercynite–spinel–magnetite solid solution, Fig. 2b) provide evidence that spinel has been consumed by the process  

\[ \hbox{Hc} + \hbox{Qtz} \rightarrow \hbox{Grt} + \hbox{Sil} \]
either by isobaric cooling and/or by a pressure increase. Spherical zircon occurs as inclusions in garnet and in the matrix, indicating an earlier event of zircon corrosion.

Fluid and glass inclusions

Mode of occurrence

Several generations of fluid and glass inclusions occur abundantly in the xenoliths. The fluid inclusions are CO2 dominated and are common in clinopyroxene, orthopyroxene, garnet, plagioclase, hercynitic spinel, scapolite (scapolite1 and -2) and amphibole, and in garnet and plagioclase in the garnet–sillimanite gneiss K4/10. Inclusions can be divided into three groups: (1) texturally early inclusions in porphyroclasts, occurring singly or in groups, well removed from host grain boundaries, especially common in garnet and scapolite (fluid and glass inclusions, liq1 in Table 1), and to a lesser extent in amphibole, clinopyroxene and orthopyroxene; (2) texturally late fluid and glass inclusion trails in porphyroclasts (liq2), reaching or cross-cutting grain boundaries and also other inclusion trails; (3) primary inclusions (fluid and glass inclusions, liq1) occurring singly or in groups (a) in neoblasts of clinopyroxene, amphibole, plagioclase and scapolite, (b) in corona phases: amphibole and hercynitic spinel replacing garnet; clinopyroxene or orthopyroxene replacing amphibole.

In the garnet–sillimanite gneiss K4/10, CO2-fluid inclusions are cogenetic with early-textured glass inclusions (liq1) in garnet containing OH, H2O, CO2 and CO32− (analyses by FTIR). Similarly, CO2 inclusions in amphibole and scapolite in clinopyroxenites and hornblendites are paragenetic with melt inclusions (e.g. Fig. 2h). Primary inclusions in the gabbroic cumulate K9/34 occur in the cores of clinopyroxene and titanite idioblasts and are associated with glass inclusions.

Many fluid and glass inclusions show textural evidence of partial decrepitation (e.g. Hansteen et al., 1991). Sizes range from <2 to 40 μm; adjacent texturally early inclusions can have widely varying sizes. Many late fluid inclusions also contain brown to colourless glass in variable proportions.

Composition of fluid inclusions

Samples chosen for fluid inclusion analysis are cumulates and metamorphic xenoliths showing different degrees of modal metasomatism and deformation (Table 3).

Table 3:

Samples chosen for fluid inclusion barometry

Sample
 
Type
 
Degree of deformation
 
Phase association
 
K9/34   Glass + Pl + Cpx + Am + Ap + Tit + Ti-Mag 
Cumulate 
K5/71 No recrystallization, deformation  
  twinning of Plg Opx + Cpx + Am + Pl + Rt 
K5/54 No recrystallization, deformation  
  twinning of Pl Grt + Cpx + Opx + Pl + Am + Scp1 + Mag + Ilm 
K5/47 E1 Strong Pl recrystallization, core  
  and mantle structures Cpx + Opx + Am + Scp1 + Pl + Mag + Ilm + Tit + Po + glass 
K5/31 porphyroclastic recrystallization Grt + Opx + Bt + Pl + Mag + Ilm + Zirc 
K4/10 E1 porphyroclastic recrystallization Grt + Sil + Qtz + Pl + Rt + Hc + Pl + Opx + Bt + Ilm + glass 
K5/58 E1 porphyroclastic recrystallization Hc-Mag + Cpx + Am + Scp2 + Ap + glass 
K9/27 E1 mosaic textured Grt + Hc + Opx + Cpx + Am + Pl + Po + glass 
Sample
 
Type
 
Degree of deformation
 
Phase association
 
K9/34   Glass + Pl + Cpx + Am + Ap + Tit + Ti-Mag 
Cumulate 
K5/71 No recrystallization, deformation  
  twinning of Plg Opx + Cpx + Am + Pl + Rt 
K5/54 No recrystallization, deformation  
  twinning of Pl Grt + Cpx + Opx + Pl + Am + Scp1 + Mag + Ilm 
K5/47 E1 Strong Pl recrystallization, core  
  and mantle structures Cpx + Opx + Am + Scp1 + Pl + Mag + Ilm + Tit + Po + glass 
K5/31 porphyroclastic recrystallization Grt + Opx + Bt + Pl + Mag + Ilm + Zirc 
K4/10 E1 porphyroclastic recrystallization Grt + Sil + Qtz + Pl + Rt + Hc + Pl + Opx + Bt + Ilm + glass 
K5/58 E1 porphyroclastic recrystallization Hc-Mag + Cpx + Am + Scp2 + Ap + glass 
K9/27 E1 mosaic textured Grt + Hc + Opx + Cpx + Am + Pl + Po + glass 

All of the fluid inclusions froze to aggregates of solid CO2 and vapour when cooled to −65 to −100°C. Further cooling to about −190°C produced no visible phase changes, implying that CH4 or N2 can be present only in minor quantities, if at all (e.g. Thiéry et al., 1994). Heating of the inclusions from about −190°C caused the following three phase transitions: (1) initial melting (Ti) of CO2 crystals in the temperature interval −57·8 to −56·4°C (CO2 triple point at −56·6°C), in many cases coinciding with (2) final melting of CO2 (Tm) at −57·2 to −56·4°C, and (3) final homogenization of liquid + vapour (L + V) into liquid or vapour (Thl and Thv, respectively) at <31·1°C (the inclusions are of the microthermometric type H3; Van den Kerkhof, 1990).

Most inclusions show a well-defined triple point melting of CO2 at −56·6 ± 0·2°C and no melting interval, indicating essentially pure CO2, which has also been confirmed in some inclusions by FTIR analysis. About 25% of the inclusions, melting between −56·8 and −57·2°C, show a melting interval covering 0·1 or 0·2°C, possibly indicating minor amounts of additional components such as N2, CH4, CO or noble gases. Several inclusions show final CO2 melting at −57·2 to −57·8°C, preceded by melting intervals varying between 0·2 and 0·8°C, suggestive of significant but small amounts of components additional to CO2 (Van den Kerkhof, 1990). The additional minor components could again be N2, CH4, CO or noble gases in amounts less than ∼2–4 mol % (Burke & Lustenhouwer, 1987; Van den Kerkhof, 1990). Such minor components are, however, not significant for our thermobarometric calculations, and will be discussed elsewhere.

GEOTHERMOBAROMETRY

Mineral thermobarometry

P–T conditions related to ancient granulite metamorphism

The association of green spinel and quartz within the garnet cores of the garnet–sillimanite gneiss (sample K4/10) was probably paragenetic at an earlier stage and thus provides evidence for a high-temperature event with T > 800–1100°C. We derived a lower stability limit at 950 MPa of the paragenesis Grt + Qtz + Sil (molar bulk XMg = 0·35; Aranovich & Podlesskii, 1983) inserting temperatures from garnet–biotite thermometry (see below). The spinel-consuming reaction Spl + Qtz → Grt + Sil, preserved in the phase associations, indicates that the estimated pressure post-dates the high-temperature stage.

P–T conditions related to late-stage heating and metasomatism

Coexisting mineral pairs were analysed for major elements, to estimate late-stage temperatures. Results obtained from mineral rim compositions are shown in Table 1. Temperature uncertainties associated with deviations from equilibrium may be large, because PCO2PH2O, but are impossible to quantify at present.

Garnet–biotite thermometry (Perchuk & Lavrent’eva, 1983; Indares & Martignole, 1985; Bhattacharya et al., 1992) applied to garnet–sillimanite gneiss K4/10 (biotite coronas not in contact with garnet) indicates a late-stage temperature increase from garnet cores to rims between ∼20 and 70°C. This temperature increase may be partially reflected in coronas of vesicular glass + Hc + Opx + Pl, separating garnet and biotite. Temperatures above the solidus are recorded by texturally early and late glass inclusions in garnet.

We also applied the orthopyroxene–clinopyroxene solvus thermometers of Wells (1977) and Brey & Köhler (1990). Mineral rims of clinopyroxene- and orthopyroxene-bearing granulites yield temperatures of apparent last equilibration between ∼670 and 900°C [thermometer of Brey & Köhler (1990)] or 779 and 950°C [thermometer of Wells (1977)] (Fig. 3). Mineral rim temperatures of mafic granulites are up to 70°C higher than the core temperatures. Further, temperatures above the solidus are recorded by glass-bearing xenoliths. Glass occurs in two varieties: as pockets of vesicular interstitial glass in contact with Cpx ± Spl (Hc–Mag solid solution) ± Hbl (pargasitic hornblende) ± Pl; and as texturally early glass inclusions in scapolite2, clinopyroxene and plagioclase (neoblasts and porphyroclasts) where it is associated with early CO2-fluid inclusions. The phase association Spl + Cpx + Prg + Pl + liq1 provides evidence for temperatures between 900°C and 1030°C [XCO2 = CO2/(CO2 + H2O) = 0·5] according to melting experiments of Springer (1992) on the Kempenich granulite suite.

Fig. 3.

Mineral thermometry. Temperature distributions for clinopyroxene–orthopyroxene and garnet–biotite pairs. (See Table 1 for abbreviations.)

Fig. 3.

Mineral thermometry. Temperature distributions for clinopyroxene–orthopyroxene and garnet–biotite pairs. (See Table 1 for abbreviations.)

Type E2 xenoliths are defined by dehydration reactions (Fig. 2). On average, type E2 rocks record higher temperatures of last equilibration than the other xenoliths. An important type of amphibole-consuming and clinopyroxene-forming reaction is found in samples K9/31 and K9/22:  

\[ \hbox{Hbl} + \hbox{Cpx1} \rightarrow \hbox{Cpx2} + \hbox{Spl}(\hbox{Hc--Mag solid solution}) + \hbox{liq(vesicular glass)} \]
(Hbl is pargasitic hornblende) and can be attributed to a temperature rise above 1050°C. The glass occurs as glass inclusions and pockets of interstitial glass in contact with clinopyroxene2 and spinel.

Further pyroxene-forming reactions in meta-hornblendites (e.g. sample K9/27, Table 2; Fig. 2h) are also due to a temperature increase:  

\[ \eqalign{& \hbox{Grt} \pm \hbox{Cpx1} \rightarrow \hbox{Cpx2} + \hbox{Hc} + \hbox{Pl}\cr \noalign{\vskip5pt} & \hbox{Grt} \rightarrow \hbox{Opx2} + \hbox{Hc} + \hbox{Pl}.} \]
The reactions proceeded in the presence of a CO2-dominated fluid, as indicated by high-density CO2 inclusions in hercynitic spinel.

Suitable mineral barometers are not available for the mafic granulites, and can give only maximum pressure values; this is ∼950 MPa for the garnet pyriclasites (Loock et al., 1990).

The temperatures obtained from mineral thermometry, however, apparently correlate positively with the degree of recrystallization of plagioclase, and provide evidence that the recorded temperatures reflect the conditions of plagioclase recrystallization.

Late-stage coronas around garnet in some meta-igneous xenoliths consist of glass + crystals (sizes ≤1 μm; spinel, pyroxene?) and probably indicate short-time syn-eruptive heating and decompressive melting after incorporation of the xenoliths by the host magma.

Fluid inclusion thermobarometry

Composition of the magmatic fluids

In addition to the occurrence of CO2-fluid inclusions, the presence of a CO2-dominated fluid phase is evidenced by FTIR analyses of carbonate-rich homogeneous scapolite2 porphyroclasts [Sachs & Hansteen (1996), unpublished data]. We have no direct evidence for the presence of H2O as a major component in the fluid inclusions. In the E1-type rocks, however, the activity of H2O during overprinting with silicate melt was obviously high enough to stabilize the hydrous phases amphibole and biotite. But even texturally early inclusions in amphibole contain no detectable water, at variance with experimental results showing that diffusive hydrogen loss from fluid inclusions at high temperatures is probable (e.g. Bakker & Jansen, 1991). CO2 + H2O fluid inclusions can thus selectively release considerable amounts of their H2O content. For example, growth imperfections along healed cracks and lattice defects formed during porphyroclastic recrystallization provide possible routes for fluid transport. Bakker & Jansen (1991) demonstrated that secondary CO2+ H2O fluid inclusions in quartz originally containing 20 mol % CO2 increased their CO2 contents up to 54 mol % during 1 month when the external pressure was decreased from 200 to 100 MPa at a temperature of 850 K. Similarly, water is expected as an original component in the fluids that caused metasomatic overprinting of the xenoliths. Experiments on the degassing of CO2 and H2O from basanitic and nephelinitic melts, i.e. of compositions similar to the host melts of the xenoliths, show that the fluid phase at the expected pressure of about 650 MPa should have an XH2O = H2O/(CO2 + H2O) close to 0·1 (Dixon, 1997). Provided that the fluid inclusions in the xenoliths originated from mafic alkaline magmas, this line of evidence also strongly indicates that the fluid inclusions have lost significant amounts of H2O after entrapment.

Scapolite is a sensor of the activity of NaCl in the fluid phase (e.g. Ellis, 1978; Mora & Valley, 1989). Derived from the nearly constant eqivalent anorthite and S contents, and on the decrease of Cl from cores to rims of scapolite2, NaCl/(NaCl + H2O) of the coexisting fluid decreased during the growth of the porphyroclasts. This is accompanied by a similar but more pronounced decrease of bromine, which can be expected to have a geochemical behaviour similar to chlorine.

In summary, the CO2-dominated fluids in question must have contained some H2O and additionally significant but unknown amounts of chloride. The activity of chloride decreased during the metasomatic stage.

Fluid inclusion densities

To compensate for the expected water loss from the fluid inclusions, we calculated 10 mol % water back into the CO2-fluid inclusions using the assumption that the volumes of the inclusion cavities did not change after inclusion formation:  

\[ D_{\rm Corr} = D_{\rm Meas} + Mw_{{\rm H}_{2}{\rm O}}/9V'_{{\rm CO}_{2}} \]
where DCorr and DMeas are the corrected and measured inclusion densities, VCO2 is the measured molar volume of CO2 in the inclusion and MwH2O is the molecular weight of H2O.

The corrected densities are 4·5% higher than those for the corresponding pure CO2 inclusions. We believe that the corrected compositions and densities provide the best first-order estimate of the fluids originally trapped during inclusion formation. We use such corrected inclusion densities for the further thermobarometric considerations.

The homogenization temperatures and corresponding densities of fluid inclusions are shown in Fig. 4a and b. Our fluid inclusion data comprise both texturally early and late inclusions in each sample. The histograms can be subdivided into three groups, as follows.

Fig. 4.

Fluid inclusion data. Microthermometric measurements of CO2-dominated inclusions in various mineral phases. (a) Homogenization temperatures. The data represent homogenization temperatures of the phase transition liquid–gas to liquid [Th(l)]. (b) Densities calculated from homogenization temperatures assuming that the inclusions originally contained 10 mol % H2O. Stippled bars represent examples of formation temperatures of fluid inclusions in the metamorphic xenoliths, assuming they were trapped at a pressure of 650 MPa. A, B and C indicate group A, B and C inclusions.

Fig. 4.

Fluid inclusion data. Microthermometric measurements of CO2-dominated inclusions in various mineral phases. (a) Homogenization temperatures. The data represent homogenization temperatures of the phase transition liquid–gas to liquid [Th(l)]. (b) Densities calculated from homogenization temperatures assuming that the inclusions originally contained 10 mol % H2O. Stippled bars represent examples of formation temperatures of fluid inclusions in the metamorphic xenoliths, assuming they were trapped at a pressure of 650 MPa. A, B and C indicate group A, B and C inclusions.

Group A (samples K9/34, K5/54, K5/71) is characterized by a sharp density maximum corresponding within the analytical error to the broad density maximum for the cumulate sample K9/34 at 0·82–0·89 g/cm3. This density maximum for K9/34 includes both primary and secondary inclusions in clinopyroxene, where the secondary fluid inclusions coexisting with glass inclusions have the higher densities of 0·85–0·89 g/cm3. K5/71 additionally contains fluid inclusions with densities of 0·90–1·05 g/cm3. Primary group A inclusions occur in plagioclase, clinopyroxene, scapolite (Scp1) and in corona amphibole around garnet.

Group B (samples K5/47, K5/31, K5/58) shows a similar maximum of secondary inclusions as group A, but at slightly higher densities of 0·91–0·97 g/cm3. K5/47 additionally contains texturally early fluid inclusions with densities in the range 0·98–1·05 g/cm3. In sample K5/58 [Cpx + Am + Scp2 + Ap + Spl (Mag–Hc) + Po + liq (fluid + silicate melt)], the main density maximum of secondary inclusions is practically identical to densities of primary inclusions in amphibole neoblasts. The porphyroclast cores of K5/58 additionally contain inclusions with densities slightly higher than the inclusions in the neoblasts.

Group C consists of the garnet–sillimanite gneiss K4/10 and the spinel–garnet–pyroxene amphibolite K9/27. Compared with the densities of secondary inclusions in the cumulate sample K9/34, the density maximum of secondary and primary inclusions in amphibole and plagioclase show a strong shift towards higher densities of 0·99–1·02 g/cm3. Early inclusions in garnet cores of sample K4/10 have densities up to 1·12 g/cm3 (corresponding to 1·07 g/cm3 for the pure CO2 fluid) (Fig. 4b).

In some samples, a small additional maximum of partially decrepitated inclusions occurs at lower densities of between 0·55 and 0·75 g/cm3.

Fluid inclusion pressures and depths of origin

Isochores for fluid inclusions containing 90 mol % CO2 and 10 mol % H2O were calculated using the modified Redlich–Kwong equation of state of Kerrick & Jacobs (1981). The isochores calculated using such corrected densities yield pressures 14–19% higher for a given temperature in the range of interest as compared with isochores calculated for the pure CO2 fluids. Addition of up to 6 molal NaCl solutions to the CO2-dominated fluids instead of pure water would lead to slightly increased isochore pressures of between 0 and 1% relative (Brown & Lamb, 1989; Joyce & Holloway, 1993), which is insignificant for the thermobarometric results.

The density maximum for secondary fluid inclusions coexisting with glass inclusions in the cumulate K9/34 is 0·85–0·89 g/cm3. If we assume a temperature of 1200°C for the host magma, the secondary inclusions would have been trapped at a pressure of 650 ± 50 MPa, which we interpret to represent the pressure of a magma chamber at 22–25 km depth (Fig. 5). The most important feature of group A secondary fluid inclusions is that the density maxima overlap completely with the densities of inclusions in clinopyroxene of the cumulate K9/34. We therefore assume that all group A fluids had nearly the same composition and temperature during entrapment. Such high temperatures are supported by the observation that the fluid inclusions are usually associated with, apparently cogenetic, glass inclusions. The occurrence of primary group A fluid inclusions in late phases in the xenoliths (amphibole and hercynitic spinel replacing garnet and pyroxene; clinopyroxene and orthopyroxene replacing pargasite; scapolite) provides first-order evidence that the phases originated by fluid and silicate metasomatism from the magma chamber.

Fig. 5.

Pressure–temperature–time (P, T, t) paths derived from thermobarometry. The shaded bars indicate P, T conditions inferred for the magma reservoir and the xenoliths. Relative positions and thickness of the bars correspond to the frequency maxima of the homogenization temperatures, Th(l), in Fig. 4. E, early; L, late; P, primary; S, secondary inclusions. Isochores are for inclusions containing 90 mol % CO2 and 10 mol % H2O. TM (shaded area), estimated host magma temperature. EMD, range of most dense, early inclusions of all meta-igneous xenoliths. The EMD is considered to represent starting temperatures of the granulites before heating. The starting temperatures are similar to those obtained from mineral rim thermometry (light shaded vertical bar, Tx). K5/58 indicates homogenization temperatures of primary inclusions in neoblasts and of secondary inclusions in porphyroclasts. Partial decrepitation of many fluid inclusions occurred at 250–450 MPa.

Fig. 5.

Pressure–temperature–time (P, T, t) paths derived from thermobarometry. The shaded bars indicate P, T conditions inferred for the magma reservoir and the xenoliths. Relative positions and thickness of the bars correspond to the frequency maxima of the homogenization temperatures, Th(l), in Fig. 4. E, early; L, late; P, primary; S, secondary inclusions. Isochores are for inclusions containing 90 mol % CO2 and 10 mol % H2O. TM (shaded area), estimated host magma temperature. EMD, range of most dense, early inclusions of all meta-igneous xenoliths. The EMD is considered to represent starting temperatures of the granulites before heating. The starting temperatures are similar to those obtained from mineral rim thermometry (light shaded vertical bar, Tx). K5/58 indicates homogenization temperatures of primary inclusions in neoblasts and of secondary inclusions in porphyroclasts. Partial decrepitation of many fluid inclusions occurred at 250–450 MPa.

The coexistence of glass and fluid inclusions provides strong evidence for a common origin from a CO2-dominated silicate-rich fluid penetrating cracks that were opened towards the host magma. We suggest that the cracks were formed as a result of the intrusive process of the host magma, driven by the temperature difference and by viscous forces acting between the magma and the wall rock.

Isobaric heating

We use a combination of mineral-pair thermometry and fluid inclusion barometry to show that the density distribution of fluid inclusions in the metamorphic xenoliths can best be explained in terms of wall-rock reactions close to a magma chamber, i.e. by isobaric heating.

Using the assumption that the pressure was fixed at 650 MPa, the highest density of the texturally early inclusions within the samples K5/71, K5/47, K5/58 and K9/27 translates into isochore temperatures between 660 and 760°C (Fig. 4b). This is very similar to the temperature of 650°C obtained from Grt–Bt thermometry of mineral cores. The density maxima for group B and C fluid inclusions would in the isobaric heating case correspond to temperatures about 250–400°C lower, respectively, than the magma temperature. Similarly, fluid inclusions having densities even higher than the density maxima in samples K5/71, K5/47 and K5/58 can most simply be explained by assuming that the texturally late inclusions were formed at temperatures about 200–500°C higher than the early inclusions. Heating during neoblast formation is also reflected by mineral thermometry.

The small maximum of partially decrepitated inclusions in most samples at densities of between 0·55 and 0·75 g/cm3 can at best be explained as due to the syneruptive pressure release of the xenoliths.

Trapping mechanisms for the late fluid inclusions

The assumption that Group A fluids had nearly host magma temperature during the entrapment process does not necessarily implicate a similar temperature for the host crystals of the inclusions. In our case, the preservation of the primary density of the fluid is possible only if the time to isolate the fluid from its environment, Δt(isolation), is considerably shorter than that necessary to cool the fluid to the ambient temperature, Δt(equilibration), i.e. Δt(isolation) ≪ Δt(equilibration). Such a process is probably possible only within a small distance from the host magma, and if silicate melt is available in addition to fluid, to support the healing process by sealing as defined by Brenan (1991).

In this scenario, the temperature of the fluid decreases quickly after fluid migration into mineral cracks. Thus the densities of the inclusions depend on the rate of the crack-sealing process. When the sealing rate is fast, the fluid injected from the magma into the wall-rock will cool relatively little until complete isolation of the inclusions occurs. The resulting inclusions would thus have a lower density than those formed if the sealing rate was slow, at which the injected fluid would cool further before complete isolation of an inclusion from its environment. A final possibility is that the fluids can equilibrate thermally with their host crystals before sealing: Δt(isolation) ≥ Δt(equilibration).

This model does not require that the total rock was at temperatures close to that of the host magma during inclusion formation. Such high temperatures would lead to high degrees of melting for the mafic granulites, i.e. a melt fraction of more than ∼30 wt % (Springer, 1992), in strong contrast to the actually observed melt fraction of <1 vol. %.

The best illustration of this model is probably the ultramafic sample K5/58, where the lowest fluid densities overlap with those in the cumulate K9/34, and the displacement of the inclusion densities towards higher values can be explained as due to a temperature decrease of the fluids by about 280°C from 1200°C to ∼920°C, until isolation of the inclusions (Fig. 4b). Because the primary inclusions in amphibole neoblasts and the secondary inclusions in amphibole porphyroclasts in K5/58 indicate an identical entrapment temperature, we infer that the fluids in the cracks thermally equilibrated with their host crystals before sealing. In sample K5/58, porphyroclastic recrystallization was apparently contemporaneous with the formation of secondary inclusions.

The only possible exceptions to our model involving CO2–H2O-fluids are the early inclusions in Gt cores in sample K4/10. In this case, the assumption that the fluid had the same composition as the fluids of the host magma leads to unrealistically low temperatures of <600°C. Using the assumption of a pure CO2-fluid at the time of formation of the garnet cores under granulite-facies conditions, a temperature of ∼650°C is obtained. Both temperatures contradict our conclusion of a temperature T > 800–1100°C derived from the possible earlier paragenesis of Spl + Qtz associated with the inclusions. The early inclusions in garnet are therefore possibly relics of an ancient high-pressure stage of the rock.

Hornblenditic veinlets (thickness ∼100 μm to 1 mm) preserved in type E1 samples (K9/23, K5/63, K5/65, K5/71) indicate that a considerable amount of amphibole-forming matter has been transported along microfractures. The formation of the veinlets is possibly related to fragmentation of the wall-rocks, and thus in a wider sense to xenolith formation. Clinopyroxenes adjacent to the veinlets are partially replaced by pargasitic hornblende and contain numerous inclusions of pargasitic amphibole, which we interpret to have been formed simultaneously with the secondary glass inclusions, i.e. through healing of microcracks. The relationship between the formation of of amphibole inclusions and secondary glass and fluid inclusions is demonstrated by fluid and OH and H2O-bearing glass inclusions coexisting with amphibole inclusions in clinopyroxene, e.g. in sample K9/40 (Fig. 2d and e).

MAJOR AND TRACE ELEMENT GEOCHEMISTRY

Primitive (type P) granulites

Type P xenoliths, by definition, show little or no evidence of modal metasomatism. We thus consider them to reflect the granulite suite before the onset of modal metasomatism. Type P meta-igneous rocks show a trend in CIPW normative compositions from plagioclase–olivine websterite toward tholeiitic and alkaline gabbros and anorthosites. MgO correlates positively with Cr, Ni, Co, FeO, heavy rare earth elements (HREE), Y and Zn, and correlates negatively with Ga and Al (Fig. 6). We interpret the correlations as a protolithic cumulate trend involving olivine, spinel, clinopyroxene and plagioclase fractionation. The molar fractions of Mg, XMg = Mg/(Fe2+ + Mg), scatter from 0·46 to 0·81, with a frequency maximum at 0·75. The meta-igneous P-type rocks involve the following compositional features (Figs 6, 7 and 8): The proposed cumulate trends of the protoliths overlap with picritic and komatiitic compositions (e.g. Arndt & Nisbet, 1982) (Fig. 6).

  1. samples with low (sample K5/31: MgO = 1·9 wt %) to intermediate MgO content (up to 13 wt %), and anorthositic and olivine tholeiitic to quartz tholeiitic compositions; e.g. sample K5/52 with MgO = 12·61 wt %, XMg = 0·72, SiO2 = 48·7 wt %, K2O = 0·13 wt %, TiO2 = 0·50 wt %, CaO/Al2O3 = 0·85, chondrite-normalized ratios (La/Sm)n = 1·7 and (Gd/Yb)n = 1·5 (Fig. 7).

  2. MgO-rich members, e.g. the pyroxenitic sample K9/30 with XMg = 0·82, MgO = 22·41 wt % and (La/Sm)n = 0·6, (Gd/Yb)n = 1 (Fig. 7). Major element parameters are SiO2 ≤ 53·54 wt %, K2O < 0·9 wt %, TiO2 < 0·9 wt %, CaO/Al2O3 > 1.

Fig. 6.

Major and trace element variation of the xenoliths. The compositions partly overlap with picritic and komatiitic compositions. The P-type xenoliths with the lowest and highest MgO concentrations correspond to the most plagioclase- and pyroxene-rich granulites, K5/31 and K9/30, respectively.

Fig. 6.

Major and trace element variation of the xenoliths. The compositions partly overlap with picritic and komatiitic compositions. The P-type xenoliths with the lowest and highest MgO concentrations correspond to the most plagioclase- and pyroxene-rich granulites, K5/31 and K9/30, respectively.

Fig. 7.

Chondrite-normalized REE diagrams (Nakamura, 1974).

Fig. 7.

Chondrite-normalized REE diagrams (Nakamura, 1974).

Fig. 8.

Multielement diagram of representative type E and cumulate xenoliths. The sample concentrations are normalized by the average concentrations of type P xenoliths with 0·6 < XMg < 0·8.

Fig. 8.

Multielement diagram of representative type E and cumulate xenoliths. The sample concentrations are normalized by the average concentrations of type P xenoliths with 0·6 < XMg < 0·8.

Compared with N-type MORB (McCulloch & Gamble, 1991; Rollinson, 1995), meta-igneous type P xenoliths have trace element patterns characterized by higher Th, U and Pb concentrations, and by lower K, REE, Sr, P, Zr, Ti, Y and Sc.

Enriched (type E) xenoliths

The XMg of type E1 and E2 xenoliths have a frequency maximum at ∼0·7, which is similar to the type P rocks. The xenoliths are characterized by a well-defined correlation of Fe2O3/FeO with Fe2O3, and a weak correlation of Fe2O3/FeO with total Fe (Fig. 9). Fe2O3 does not correlate with MgO. The P-type samples are the least oxidized. The gabbroic cumulate K9/34 can be interpreted as a hypothetical endmember at high Fe2O3/FeO. Fe2O3 correlates strongly with V and Ti, providing evidence that Fe–Ti oxides are the most important Fe3+ phases and that alteration plays no significant role for the oxidation state of the xenoliths, which is also confirmed by optical microscopy and by electron microprobe analyses. In general, Fe–Ti oxides of E-type xenoliths are dominated by Ti-magnetite, whereas P-type xenoliths predominantly contain rutile or ilmenite. Ilmenite is partly surrounded by a corona of Ti-magnetite. This provides further evidence that oxidation of Fe2+ and not the addition of Fe3+-enriched matter controlled the oxidation state of the rocks. Oxidation could have been controlled by a CO2-rich fluid.

Fig. 9.

Selected major-element ratio plots. (a) E-type xenoliths partly have higher K2O/Na2O ratios than the type P rocks, and higher K2O concentrations. The gabbroic cumulate K9/34 is a possible endmember at high K2O/Na2O. (b) The xenoliths are characterized by a well-defined correlation of Fe2O3/FeO with Fe2O3, and a weak correlation of Fe2O3/FeO with total Fe. The P-type samples are the least oxidized. The gabbroic cumulate K9/34 can be interpreted as a hypothetical endmember at high Fe2O3/FeO.

Fig. 9.

Selected major-element ratio plots. (a) E-type xenoliths partly have higher K2O/Na2O ratios than the type P rocks, and higher K2O concentrations. The gabbroic cumulate K9/34 is a possible endmember at high K2O/Na2O. (b) The xenoliths are characterized by a well-defined correlation of Fe2O3/FeO with Fe2O3, and a weak correlation of Fe2O3/FeO with total Fe. The P-type samples are the least oxidized. The gabbroic cumulate K9/34 can be interpreted as a hypothetical endmember at high Fe2O3/FeO.

Our petrographic observations indicate that type E1 rocks are enriched in K, OH, F and Cl, and depleted in Si relative to type P rocks. The replacement of pyroxene by secondary amphibole (typical SiO2 contents of 39–46 wt %, compared with 49–54 wt % in pyroxene; Table 2 samples K5/71 and K9/29) is reflected by the bulk compositions of the E1 xenoliths partly having lower SiO2 contents and higher K2O/Na2O ratios than the type P rocks (Fig. 9), and by higher K2O concentrations (Fig. 6). The enrichment of scapolite implies enrichment of Cl, Br, S and CO32−, and also Rb, Nb, Sr, REE and Y as seen from synchrotron X-ray fluorescence (SYXRF) analyses (Table 2, sample K5/58). Apatite is strongly enriched in REE.

The replacement of amphibole by pyroxene defining type E2 rocks indicates a relative increase in Si, and a loss of K, halogens and OH.

As Fig. 7 shows, type E rocks appear to be enriched in REE as compared with P-type rocks. The REE compositions of type E2 rocks overlap with the host magma cumulates. To better visualize the typical compositional differences between the P- and E-type rocks, we use a multielement diagram in which the E-type rocks are normalized to the most typical composition of our P-type rocks (Table 4, Fig. 8), with mg-numbers 0·6 < XMg < 0·8.

The following distinct compositional trends can be observed. E1-type xenoliths show an enrichment in incompatible elements Rb, Th, U, Nb and K. Sr and P are relatively depleted. The more compatible elements Y, Yb, Cr, Mn, Co, Ni and Zn show no or little increase relative to average P-type compositions. Relative to type P rocks, E2-type xenoliths are enriched in Rb, Th, U, Nb, K, REE, Zr, Ti, Y and Fe3+, and are depleted in Sr.

The Grt–Sil gneiss K4/10 can be considered to be a E1-type xenolith because of the biotite coronas partially replacing Prp–Alm garnet. Chondrite-normalized REE concentrations in this meta-sediment (Fig. 7) are characterized by a marked concave-up pattern, indicating a relative enrichment of the LREE and the HREE + Y as compared with the middle rare earth elements (MREE) (Nd, Sm, Eu). The enrichment of Y and HREE probably reflects earlier melting events and restite formation. The enrichment of Zr, Y, Sr, Rb, V and Zn in rims relative to cores of the garnet porphyroclasts (SYXRF analyses, Table 2, sample K4/10) in K4/10 is probably cogenetic with E1 metasomatism and with the formation of biotite coronas around garnet.

DISCUSSION

Chemical evolution of the xenoliths

We want to test the hypothesis that the compositions of E-type xenoliths are genetically related to the host magma chamber. This can be tested by comparing element pairs having similar bulk distribution coefficients during protolith formation by partial melting (of peridotitic systems) and subsequent crystal fractionation, but will be affected by the presence of a metasomatic component originating from the host magma reservoir. Thus we chose elements that are characterized by positive or negative deviations from the average P-type composition and compared them with reference elements with similar bulk compatibility under fluid-absent conditions. Element pairs in question are Th–Ba, U–Nb, K–Nb, LREE–Pb, Zr–Ti, Zn–Y, Co–Y and Cu–Y. If the metamorphic xenoliths were affected by the host magma, we would expect a mixing line in bivariate plots, where the host magma cumulates (samples K9/25 and K9/34) should define an endmember composition. This can be expected because the cumulates contain abundant clinopyroxene, Fe–Ti oxides, amphibole, ± biotite and apatite, which are also the main constituents of the metamorphic xenoliths. Therefore melts and fluids released from the magma would tend to equilibrate at similar bulk partition coefficients with both the cumulate crystals and the mafic wall-rock phases. Element concentrations in the cumulates can therefore be expected to be most similar to those metamorphic xenoliths that had the most intense and longest contact with the host magma, probably the type E2 rocks, because their phase associations generally record higher temperatures of last equilibration than the other xenolith types.

The nearly constant ratios Y/Zn vs Zn (Fig. 10) and Y/Co vs Co of P-type and E-type xenoliths indicate that these ratios are independent of Zn and Co concentrations. Thus in the presence of the proposed metasomatic agent, the geochemical behaviour of Zn, Co and Y is very similar. The absence of a mixing line relating the Zn and Co contents of the cumulate xenoliths to the granulites indicates that the Y/Zn and Y/Co ratios are controlled by processes that are not related to the host magma of the xenoliths. The meta-sedimentary xenolith K4/10 defines a separate compositional group, which expresses the different protolith history. The representation of the ratio Y/Cu vs Cu, however (Fig. 10), shows a well-defined trend with a distinct negative slope and with an endmember composition coinciding with the cumulate xenoliths at high Cu contents. A similar trend would result if HREE, Ni, Co or Zn were used as reference elements. The interstitial glass of the cumulate xenolith K9/34 is characterized by very low Cu and Y contents (Fig. 10). Although some Cu may have been lost by syneruptive degassing at low pressure, as indicated by the precipitation of Cu-enriched (sample K9/34: 0·2–1 wt % Cu) sheet silicates at the walls of the vesicles, this confirms our hypothesis that these elements are much more abundant in the crystalline phases than in the silicate melt, such that the observed trends represent in situ compositions of the xenoliths.

Fig. 10.

Bivariate ratio plots of trace element concentrations (μg/g). The nearly constant ratios Y/Zn for P-type and E-type xenoliths indicate that these ratios are independent of Zn concentrations. The absence of a mixing line relating the Zn contents of the cumulate xenoliths to the granulites indicates that the Y/Zn ratios are controlled by processes that are not related to the host magma of the xenoliths. All other plots indicate compositional trends toward a hypothetical endmember coinciding with the compositions of cumulate xenoliths K9/34 and K9/25. The P-type xenoliths define the other endmember. Xenoliths with compositions most similar to the cumulates are of type E2. The metasomatism of the metamorphic rocks could thus have had the same source as the cumulates. In the K-representation sample K9/25 was omitted because K was significantly enriched as a result of crystallization of phlogopite. MG, phonotephritic matrix glass of the cumulate K9/34, analysed by SYXRF.

Fig. 10.

Bivariate ratio plots of trace element concentrations (μg/g). The nearly constant ratios Y/Zn for P-type and E-type xenoliths indicate that these ratios are independent of Zn concentrations. The absence of a mixing line relating the Zn contents of the cumulate xenoliths to the granulites indicates that the Y/Zn ratios are controlled by processes that are not related to the host magma of the xenoliths. All other plots indicate compositional trends toward a hypothetical endmember coinciding with the compositions of cumulate xenoliths K9/34 and K9/25. The P-type xenoliths define the other endmember. Xenoliths with compositions most similar to the cumulates are of type E2. The metasomatism of the metamorphic rocks could thus have had the same source as the cumulates. In the K-representation sample K9/25 was omitted because K was significantly enriched as a result of crystallization of phlogopite. MG, phonotephritic matrix glass of the cumulate K9/34, analysed by SYXRF.

Similar to the pair Y/Cu, the representations of the pairs LREE/Pb vs Pb, K/Nb vs Nb, Ba/Th vs Th, Ba/U vs U, Ti/Zr vs Zr (Fig. 10) and Ba/Rb vs Rb reveal well-defined trends with negative slopes, and endmember compositions coinciding with the cumulate xenoliths at high Rb, Th, U, Nb, LREE, Zr and Cu. The garnet–sillimanite gneiss K4/10 falling into these chemical trends in spite of a different protolith history provides further evidence that the element mobilization was controlled by the host magma.

The role of fluid phases during metasomatism of the wall-rocks

Our chemical data strongly indicate that E1 and E2 metasomatic processes are related to the compositions of the cumulate xenoliths and thus to the host magmas by mixing curves. We can distinghuish two types of chemical plots where (1) The P-, E1- and E2-type xenoliths are separated into distinct compositional fields, and (2) The P-type compositions overlap completely with E-type compositions.

Group 1 is represented by Nb, Zr, LREE, Th and U. P-type xenoliths have the highest Ba/U, Ba/Th, K/Nb, Pb/LREE and Ti/Zr ratios. E2-type xenoliths have the lowest ratios and have concentrations of U, Th, Nb, LREE and Zr similar to the host magma cumulates. The decrease of the ratios from P to E2 xenoliths indicates that the addition of high field strength elements (HFSE) U, Th, Nb, LREE and Zr was more important than the addition of large ion lithophile elements (LILE) and Ti during the progression from P to E1 and E2 metasomatism. Probably, elements were enriched in the granulites through interaction with a silicate melt.

Group 2 is represented by Cu, where the P-type compositions widely overlap with E-type compositions. This provides evidence for a Cu-enrichment process that is decoupled from silicate metasomatism. We therefore infer that significant amounts of Cu were preferentially added to the P-type xenoliths through the action of a fluid phase.

The observation that P-type xenoliths have the highest and E2 xenoliths the lowest ratios Pb/LREE, K/Nb, Rb/Ba is compatible with the hypothesis that the amount of melt added to the P-type rocks was minimal and that a chlorine-bearing fluid has added Cu to the P-type rocks. Such a fluid will be able also to dissolve and preferentially mobilize Pb, K and Rb, more than Ce, Nb and Ba, if no silicate melt is present (Keppler & Wyllie, 1991; Keppler, 1993, 1996; Kravchuk & Keppler, 1994). The presence of the Cl- and Br-rich phases scapolite (Table 2, sample K5/58), amphibole (Table 2, samples K5/71 and K9/29), apatite and biotite indicates that halogens were important during metasomatism.

An overprinting of the wall-rocks of the host magma chamber by a CO2- and halogen-bearing fluid is in agreement with the hypothesis of, for example, Frost & Frost (1987), who, on the basis of experimental data, suggested that additionally to CO2, fluids exsolved from a mafic magma in the lower crust will also be enriched in alkali chloride.

Cenozoic evolution of the lower continental crust

On the basis of barometry of CO2-dominated fluid inclusions, we can conclude that the cumulate xenoliths originate from the same depth as the metamorphic xenoliths. Assuming a magma temperature of ∼1200°C, the primocrysts of the cumulates crystallized within the pressure interval of 650 ± 50 MPa, corresponding to a former magma chamber at a depth of 22–25 km. The rarity and the small sizes of peridotite xenoliths at Kempenich can be interpreted as the result of settling and/or assimilation of mantle xenoliths in this or a deeper reservoir (e.g. Sachs & Stange, 1993).

The pronounced density maxima of fluid inclusions, giving constant pressure and overlapping temperature ranges for the granulite xenoliths, indicate that they are pieces of fragmented wall-rocks of the magma chamber. Apparent equilibration temperatures derived from clinopyroxene–orthopyroxene and garnet–biotite rim compositions coincide with temperatures inferred from microthermometry of texturally early fluid inclusions and thus probably reflect the temperature field around the intrusion.

This is supported by the appearance of compositional trends relating the granulites to the cumulate xenoliths, demonstrating that the granulites belong to a genetically uniform rock unit, which may be relatively small. We therefore suggest that similar trends in other mafic granulite xenolith suites might be indicative of an origin from a limited depth range, and that such xenoliths represent wall-rocks of a magma stagnation zone. By analogy, other lower-crustal xenolith suites also may represent wall-rocks from magma chambers, as suggested by Hansteen et al. (1998) for the Canary Islands and Iceland.

Neoblasts in porphyroclastically recrystallized ultramafic xenoliths, containing primary glass inclusions, provide evidence for ductile deformation under hypersolidus conditions, i.e. during formation of fluid and melt inclusions. We thus infer that recrystallization is contemporaneous with the formation of the host magma chamber, and possibly was caused by the stress field of the intruding magma, as suggested for lower-crustal gabbro complexes in the Ivrea zone (Sinigoi et al., 1994). During deformation, the growing neoblasts and cracks in the porphyroclasts took up fluids that originated from the magma chamber.

Our model agrees with a model of lower-crustal magma reservoirs in the Eifel postulated by Schmincke (1977) and Duda & Schmincke (1985), and is supported by geophysical observations (Fig. 11). Beneath Kempenich, the seismic P-wave velocities between the depth of ∼18 km and the Moho at ∼30 km are anomalously low at 6·25 km/s (Mechie et al., 1983; Raikes & Bonjer, 1983), indicating the occurrence of either felsic rocks or mafic rocks at high temperatures (Mengel et al., 1991).

Fig. 11.

Model of the Pleistocene lithosphere beneath the Kempenich–Engeln area. Present-day seismic P-wave velocity structures beneath and north of the East Eifel volcanic field, based on refraction seismic survey data and teleseismic tomography, are compared with inferred crustal compositions. The residuals give the time delay in percent of teleseismic signals relative to a reference model. The depth distribution of seismic energy inferred from hypocentre distribution of earthquakes (from Langer, 1990) indicates the position of a ductile region immediately above the Conrad discontinuity. The Conrad discontinuity separates an upper-crustal layer, consisting of preferentially ductile amphibolite-facies granodioritic and tonalitic gneisses, and more brittle lower-crustal granulites. The brittle–ductile transition appears to be a preferred level of magma stagnation.

Fig. 11.

Model of the Pleistocene lithosphere beneath the Kempenich–Engeln area. Present-day seismic P-wave velocity structures beneath and north of the East Eifel volcanic field, based on refraction seismic survey data and teleseismic tomography, are compared with inferred crustal compositions. The residuals give the time delay in percent of teleseismic signals relative to a reference model. The depth distribution of seismic energy inferred from hypocentre distribution of earthquakes (from Langer, 1990) indicates the position of a ductile region immediately above the Conrad discontinuity. The Conrad discontinuity separates an upper-crustal layer, consisting of preferentially ductile amphibolite-facies granodioritic and tonalitic gneisses, and more brittle lower-crustal granulites. The brittle–ductile transition appears to be a preferred level of magma stagnation.

P-wave velocities north of Kempenich (6·1–6·4 km/s) between 10 and 22 km depth probably reflect the velocity structure of the Eifel crust before the volcanism. The velocities probably indicate the presence of amphibolite-facies gneisses of granodioritic and tonalitic composition (Mengel et al., 1991). The velocity increase to 6·7 km/s at a depth of 22 km defines the Conrad discontinuity. Thus, the xenolith suite originates from the uppermost lower crust immediately below the Conrad discontinuity. Studies of hypocentre distributions of earthquakes north of the Kempenich area indicate that this region corresponds to an aseismic zone that extends between 16 and 24 km depth, meaning that the zone comprises the lowermost part of the amphibolite-facies gneiss layer. The aseismic zone underlies a zone of high earthquake activity, which characterizes the brittle upper crust. At a depth of 24–28 km, a second but smaller activity maximum indicates the presence of a further brittle layer in the lower crust (Langer, 1990). The brittle material could possibly be composed of mafic granulites similar to the granulite xenoliths.

The aseismic zone provides evidence that the rocks are preferentially plastically deformed, i.e. the shearing strength is controlled by the dynamic viscosity. A relative decrease of the shear strength could be the reason why the magmas intruded at this depth. The ductile–brittle transition is probably the most efficient barrier against magma ascent.

CONCLUSIONS

We derive the following conclusions from this study of granulite xenoliths:

  1. fluid inclusion barometry using histogram density maxima indicates that the granulite xenoliths from Engeln were incorporated into the host magma at similar levels to those at which the cumulate xenoliths crystallized, and thus represent wall-rocks of a magma reservoir within the Pleistocene crust at 22–25 km depth (650 ± 50 MPa).

  2. A depth of 22 km corresponds to the position of the Conrad discontinuity. The Conrad discontinuity separates an upper-crustal layer, consisting probably of preferentially ductile amphibolite-facies granodioritic and tonalitic gneisses, and more brittle lower-crustal granulites. The brittle–ductile transition appears to be a preferred level of magma stagnation.

  3. Fluid inclusions with densities higher than those in the cumulate xenoliths correspond to lower temperatures than the host magma. The occurrence of high-density fluid inclusions in several xenoliths thus provides evidence for in situ heating of between 150 and 400°C, possibly as a result of local heating induced by percolating fluids released from the magma chamber.

  4. By combining thermobarometry of fluid inclusions with geochemical and petrological investigations, we have shown that cumulate xenoliths may originate from the same reservoir as the metasomatic fluids that have overprinted the source rocks of the granulite xenoliths. Metasomatism is a texturally late-stage process and includes ‘amphibolite’-forming hydration reactions (enriched type E1: formation of amphibole and biotite) and ‘granulite’-forming dehydration by breakdown of volatile-bearing phases, in particular of amphibole (enriched type E2).

*Corresponding author. e-mail: psachs@geomar.de

APPENDIX

Methods

The samples were prepared as polished thin and doubly polished thick sections under petroleum without water contact for optical microscopy, scanning electron microscopy (SEM), electron microprobe analyses, ionprobe, microthermometry, FTIR and SYXRF microprobe.

Mineral analyses were performed on a Cameca SX-50 electron microprobe at GEOMAR applying the built-in PAP correction procedure (Pouchou & Pichoir, 1984). Analytical conditions included an acceleration voltage of 15 kV, a beam current of 8–20 nA, and counting times of between 20 and 60 s on peaks. A focused beam was used for olivine, pyroxenes and oxides, and a rastered beam of 1–40 μm2 for other phases. Natural and synthetic minerals were used as standards and monitors. Analytical accuracy is <0·3% for concentrations of >10 wt % and <5% for 0·1–10 wt %.

Central pieces of the samples were cut and selected for whole-rock analyses, avoiding cracks and rims affected by melt. Chemical analyses of rocks showing petrographic evidence of melt infiltration along grain boundaries were not considered. Samples for whole-rock analyses were crushed and powdered in agate ball-mills. Before analysis, the samples were dried at 110°C. Major and trace elements were determined by XRF on fused beads using an automated Philips PW1480 spectrometer. All analyses were performed with a Rh tube; calibration was performed using international geological reference samples. Trace elements were analysed by inductively coupled plasma-mass spectrometry (ICP-MS) on a VG Plasmaquad PQ1 at the Geological Institute of the Christian-Albrechts University in Kiel (Garbe-Schönberg, 1993). FeO has been repeatedly analysed by potentiometric titration. Precision was better than 15%. Total sulphur was analysed by an IR photometer, Rosemount CSA 5003. Analytical precision is better than 20%.

Trace element compositions in the phases were measured by SYXRF microprobe at the DORIS positron storage facility at HASYLAB/DESY in Hamburg. Experimental setup and quantification of the spectra have been described by Lechtenberg et al. (1996). Detection limits are 0·1–3 μg/g for atomic numbers 21 (Sc) to 26 (Co); 0·1–1 μg for 28 (Ni) to 60 (Nd); ∼5 μg/g for >60 (Nd). Analytical accuracy was checked by analysis of international standards (analytical signal Kα; Th and Pb: Lα; Sachs & Lechtenberg, 1997), and are better than 5% at concentrations <100 μg/g, and better than 15% at <10 μg/g.

Microthermometric measurements were performed on inclusions in clinopyroxene, amphibole, plagioclase, scapolite and garnet using Linkam® THM 600 and Fluid Inc.® heating–cooling stages, which were calibrated using SYNFLINC® synthetic fluid inclusion temperature standards. Accuracy and precision was estimated at ±0·2°C near the triple point of CO2 (−56·6°C), and at better than ±0·4°C at other temperatures. Isochores for the CO2-dominated inclusions were calculated using the computer program FLINCOR (Brown, 1989) utilizing the Kerrick & Jacobs (1981) equation of state for CO2. Densities of CO2 were derived from Angus et al. (1976). Compositions of fluid inclusions were qualitatively analysed by FTIR on a Bruker IFS 120 instrument at Bayerisches Geoinstitut in Bayreuth (H. Keppler). Detection limits are >100 μg/g.

Thanks are due to H.-U. Schmincke for initiating this project. Discussions with H.-U. Schmincke, K. Hoernle, A. Gurenko and E. Harms are gratefully acknowledged. Reviews by H. Downes, E.-R. Neumann, H.-G. Stosch and B. Upton helped improve the manuscript. ICP-MS analyses were performed by C. D. Garbe-Schönberg, and F. Lechtenberg is thanked for his support during SYXRF analyses. This research was funded by the Deutsche Forschungsgemeinschaft through Grants Schm 250/41-1/2 and Schm 250/47, and the Volkswagen Foundation (Grant I/68 581).

REFERENCES

Angus, S., Armstrong, B., de Reuck, K. M., Altunin, V. V., Gadetskii, O. G., Chapela, G. A. & Rowlinson, J. S. (
1976
). International Tables of the Fluid State, Vol. 3. Carbon Dioxide. Oxford: Pergamon,
385
pp.
Aranovich, L. Ya. & Podlesskii, K. K. (
1983
).
The Cordierite–Garnet–Sillimanite–Quartz equilibrium: experiments and applications.
  In: Saxena, K. (ed.) Kinetics and Equilibrium in Mineral Reactions. Berlin: Springer, pp.
173
–197.
Arndt, N. T. & Nisbet, E. G. (
1982
).
What is a komatiite?
  In: Arndt, N. T. & Nisbet, E. G. (eds) Komatiites. London: Allen and Unwin, pp.
19
–27.
Bakker, R. J. & Jansen, J. B. H. (
1991
).
Experimental post-entrapment water loss from synthetic CO2–H2O inclusions in natural quartz.
  Geochimica et Cosmochimica Acta
55
,
2215
–2230.
Bhattacharya, A., Mohanty, L., Maji, A., Sen, S. K. & Raith, M. (
1992
).
Non-ideal-mixing in the phlogopite–annite binary: constraints from experimental data on the Mg–Fe partitioning and a reformulation of the biotite–garnet geothermometer.
  Contributions to Mineralogy and Petrology
111
,
87
–93.
Brenan, J. (
1991
).
Development and maintainance of metamorphic permeability: implications for fluid transport.
  In: Kerrick, D. M. (ed.) Contact Metamorphism. Mineralogical Society of America, Reviews in Mineralogy
26
,
291
–319.
Brey, G. P. & Köhler, T. (
1990
).
Geothermobarometry in four-phase lherzolites II. New thermobarometers, and practical assessment of existing thermobarometers.
  Journal of Petrology
31
,
1353
–1378.
Brown, P. E. (
1989
).
FLINCOR: a fluid inclusion data reduction and exploration program (abstract).
  Second Biennial Pan-Am Conference on Fluid Inclusions, Program and Abstracts, p.
14
.
Brown, P. E. & Lamb, W. M. (
1989
)
PVT properties of fluids in the system H2O ± CO2 ± NaCl: new graphical presentations and implications for fluid inclusion studies.
  Geochimica et Cosmochimica Acta
53
,
1209
–1221.
Bucher, K. & Frey, M. (
1994
). Petrogenesis of Metamorphic Rocks, 6th edn. Berlin: Springer,
318
pp.
Burke, E. A. J. & Lustenhouwer, W. J. (
1987
).
The application of a multichannel laser Raman microprobe (Microdil 28) to the analyses of fluid inclusions.
  Chemical Geology
61
,
11
–17.
Cox, K. G. (
1993
).
The Karoo province of southern Africa: origin of trace element enrichment patterns.
  In: Hawkesworth, C. J. & Norry, M. J. (eds) Continental Basalts and Mantle Xenoliths. Nantwich: Shiva, pp.
139
–157.
Dixon, J. E. (
1997
).
Degassing of alkalic basalts.
  American Mineralogist
82
,
368
–378.
Duda, A. & Schmincke, H.-U. (
1985
).
Polybaric evolution of alkali basalts from the West Eifel: evidence from green-core clinopyroxenes.
  Contributions to Mineralogy and Petrology
91
,
340
–353.
Ellis, D. E. (
1978
).
Stability and phase equilibria of chloride and carbonate bearing scapolites at 750°C and 4000 bar.
  Geochimica et Cosmochimica Acta
42
,
1271
–1281.
Frost, B. R. & Frost, C. D. (
1987
).
CO2, melts and granulite metamorphism.
  Nature
327
,
503
–506.
Furlong, K. P. & Fountain, D. M. (
1986
).
Continental crustal underplating: thermal considerations and seismic–petrological consequences.
  Journal of Geophysical Research
91
(B8),
8285
–8294.
Garbe-Schönberg, C. D. (
1993
).
Simultaneous determination of thirty-seven trace elements in twenty-eight international rock standards by ICP-MS.
  Geostandards Newsletter
17
(1),
81
–97.
Hansteen, T. H., Andersen, T., Neumann, E.-R. & Jelsma, H. (
1991
).
Fluid and silicate glass inclusions in ultramafic and mafic xenoliths from Hierro, Canary Islands: implications for mantle metasomatism.
  Contributions to Mineralogy and Petrology
107
,
242
–254.
Hansteen, T. H., Klügel, A. & Schmincke, H.-U. (
1998
).
Multi-stage magma ascent beneath the Canary Islands: evidence from fluid inclusions.
  Contributions to Mineralogy and Petrology
132
,
48
–64.
Harte, B. (
1983
).
Mantle peridotites and processes—the kimberlite sample.
  In: Hawkesworth, C. J. & Norry, M. J. (eds) Continental Basalts and Mantle Xenoliths. Nantwich: Shiva, pp.
46
–91.
Indares, A. & Martignole, J. (
1985
).
Biotite–garnet geothermometry in the granulite facies: the influence of Ti and Al in biotite.
  American Mineralogist
70
,
272
–278.
Joyce, D. B. & Holloway, J. R. (
1993
)
An experimental determination of the thermodynamic properties of H2O–CO2–NaCl fluids at high pressures and temperatures.
  Geochimica et Cosmochimica Acta
57
,
733
–746.
Katz, M. B. (
1987
).
Graphite deposits of Sri Lanka: a consequence of granulite facies metamorphism.
  Mineralium Deposita
22
,
18
–25.
Keppler, H. (
1993
).
Influence of fluorine on the enrichment of high field strength trace elements in granitic rocks.
  Contributions to Mineralogy and Petrology
114
,
479
–488.
Keppler, H. (
1996
).
Constraints from partitioning experiments on the composition of subduction-zone fluids.
  Nature
380
,
237
–240.
Keppler, H. & Wyllie, P. J. (
1991
).
Partitioning of Cu, Sn, Mo, W, U, and Th between melt and aqueous fluid in the systems haplogranite–H2O–HCl and haplogranite–H2O–HF.
  Contributions to Mineralogy and Petrology
109
,
139
–150.
Kerrick, D. M. & Jacobs, G. K. (
1981
).
A modified Redlich–Kwong equation for H2O, CO2 and H2O–CO2 mixtures at elevated temperatures and pressures.
  American Journal of Science
281
,
735
–767.
Kravchuk, I. S. & Keppler, H. (
1994
).
Distribution of chloride between aqueous fluids and felsic melts at 2 kbar and 800°C.
  European Journal of Mineralogy
6
,
1
–11.
Langer, H. (
1990
).
Seismicity along the central segment of the EGT.
  In: Freeman, R., Giese, P. & Mueller, St. (eds) The European Geotraverse: Integrative Studies. Strasbourg: European Science Foundation, pp.
121
–129.
Lechtenberg, F., Garbe, S., Bauch, J., Dingwell, D. B., Freitag, J., Haller, M., Hansteen, T. H., Knöchel, A., Radtke, M., Romano, C., Sachs, P. M., Schmincke, H.-U. & Ullrich, H. J. (
1996
).
The X-ray fluorescence measurement place at beamline L of HASYLAB.
  Journal of Trace and Microprobe Techniques
14
(3),
561
–587.
Le Maitre, R. W., Bateman, P., Dudek, A., Keller, J., Lameyre, C. W., Le Bas, M. J., Sabine, P.A., Schmid, R., Sørensen, H., Streckeisen, A., Woolley, A. R. & Zanettin, B. (
1989
). A Classification of Igneous Rocks and Glossary of Terms. Oxford: Blackwell.
Loock, G., Stosch, H.-G. & Seck, H. A. (
1990
).
Granulite facies lower crustal xenoliths from the Eifel, West Germany: petrological and geochemical aspects.
  Contributions to Mineralogy and Petrology
105
,
25
–41.
McCulloch, M. T. & Gamble, J. A. (
1991
).
Geochemical and geodynamical constraints on subduction zone magmatism.
  Earth and Planetary Science Letters
102
,
358
–374.
Mechie, J., Prodehl, C. & Fuchs, K. (
1983
).
The long-range seismic refraction experiment in the Rhenish Massif.
  In: Fuchs, K., von Gehlen, K., Mälzer, H., Murawski, H. & Semmel, A. (eds) Plateau Uplift. The Rhenish Shield—a Case History. Berlin: Springer, pp.
260
–275.
Mengel, K., Sachs, P. M., Stosch, H. G., Wörner, G. & Loock, G. (
1991
).
Crustal xenoliths from Cenozoic volcanic fields of West Germany: implications for structure and composition of the continental crust.
  Tectonophysics
195
,
271
–289.
Meyer, W., Albers, H. J., Berners, H. P., von Gehlen, K., Glatthaar, D., Löhnertz, W., Pfeffer, K. J., Schnütgen, A., Wienecke, K. & Zakosek, H. (
1983
).
Pre-Quaternary uplift in the central part of the Rhenish massif.
  In: Fuchs, K., von Gehlen, K., Mälzer, H., Murawski, H. & Semmel, A. (eds) Plateau Uplift. The Rhenish Shield—a Case History. Berlin: Springer, pp.
39
–46.
Mora, C. I. & Valley, J. W. (
1989
).
Halogen-rich scapolite and biotite; implications of metamorphic fluid–rock interaction.
  American Mineralogist
74
,
721
–737.
Nakamura, N. (
1974
).
Determination of REE, Ba, Fe, Mg, Na and K in carbonaceous and ordinary chondrites.
  Geochimica et Cosmochimica Acta
38
,
757
–775.
Newton, R. C., Smith, J. V. & Windley, B. F. (
1980
).
Carbonic metamorphism, granulites and crustal growth.
  Nature
288
,
45
–50.
Okrusch, M., Schröder, B. & Schnütgen, A. (
1979
).
Granulite facies metabasite ejecta in the Laacher See area, Eifel, West Germany.
  Lithos
12
,
251
–270.
Passchier, C. W. & Trouw, R. A. J. (
1996
). Microtectonics. Berlin: Springer,
289
pp.
Perchuk, L. L. & Lavrent’eva, I. V. (
1983
).
Experimental investigation of exchange equilibria in the system cordierite–garnet–biotite.
  In: Saxena, K. (ed.) Kinetics and Equilibrium in Mineral Reactions. Berlin: Springer, pp.
199
–239.
Pouchou, J. L. & Pichoir, F. (
1984
).
A new model for quantitative X-ray microanalysis, Part I. Application to the analysis of homogeneous samples.
  Recherche Aerospaciale
3
,
13
–38.
Raikes, S. & Bonjer, K.-P. (
1983
).
Large-scale mantle metasomatism heterogeneity beneath the Rhenish Massif and its vicinity from teleseismic P-residuals measurements.
  In: Fuchs, K., von Gehlen, K., Mälzer, H., Murawski, H. & Semmel, A. (eds) Plateau Uplift. The Rhenish Shield—a Case History. Berlin: Springer, pp.
315
–331.
Rollinson, H. (
1995
). Using Geochemical Data: Evaluation, Presentation, Interpretation. Harlow, UK: Longman,
352
pp.
Rudnick, R. L. & Fountain, D. M. (
1995
).
Nature and composition of the continental crust: a lower crustal perspective.
  Reviews in Geophysics
33
,
267
–309.
Sachs, P. M. & Hansteen, T. (
1996
).
The geotherm and metasomatism in the lower crust beneath the E-Eifel volcanic field/Germany: an application of fast kinetics fluid inclusion barometry on granulite xenoliths.
  Berichte der Deutschen Mineralogischen Gesellschaft. Beihefte zum European Journal of Mineralogy
8
,
235
.
Sachs, P. M. & Lechtenberg, F. (
1997
).
Synchrotron X-ray fluorescence (SYXRF) analysis of the international standards SY-3, JB-2, JF-2, NIM-G and NIM-S.
  HASYLAB Annual Report. Hamburg: Hamburger Synchrotronstrahlungslabor HASYLAB am Deutschen Elektronensynchrotron DESY, pp.
985
–986.
Sachs, P. M. & Stange, S. (
1993
).
Fast assimilation of xenoliths in magmas.
  Journal of Geophysical Research
98
,
19741–19754
.
Schmincke, H.-U. (
1977
).
Eifel-Vulkanismus östlich des Gebietes Rieden-Mayen.
  Fortschritte der Mineralogie
55
,
1
–31.
Schmincke, H.-U., van den Bogaard, P. & Freundt, A. (
1990
). Quaternary Eifel Volcanism. IAVCEI International Volcanological Congress, Mainz, 1990. Witten: Pluto Press,
188
pp.
Sinigoi, S. S., Quick, J. E., Clemens-Knott, D., Mayer, A., Demarchi, G., Mazzuchelli, M., Negrini, L. & Rivalenti, G. (
1994
).
Chemical evolution of a large mafic intrusion in the lower crust, Ivrea–Verbano Zone, northern Italy.
  Journal of Geophysical Research
99
(B11),
21575–21590
.
Springer, W. (
1992
). Entstehung granitoider Magmen durch partielle Aufschmelzung basischer Unterkruste: eine experimentelle Studie. Ph.D. thesis, Universität Köln,
116
pp.
Stosch, H.-G. & Lugmair, G. W. (
1984
).
Evolution of the lower continental crust: granulite facies xenoliths from the Eifel, West Germany.
  Nature
311
,
368
–370.
Stosch, H.-G., Lugmair, G. W. & Seck, H. A. (
1986
).
Geochemistry of granulite-facies lower crustal xenoliths: implications for the geological history of the lower continental crust below the Eifel, West Germany.
  In: Dawson, J. B., Carswell, D. A., Hall, J. & Wedepohl, K. H. (eds) The Nature of the Lower Continental Crust. Geological Society, London, Special Publication
24
,
309
–317.
Swayze, G. A. & Clark, R. N. (
1990
).
Infrared spectra and crystal chemistry of scapolites: implications for Martian mineralogy.
  Journal of Geophysical Research
95
(B9),
14481–14495
.
Thiéry, R., van den Kerkhof, A. M. & Dubessy, J. (
1994
).
VX properties of CH4–CO2 and CO2–N2 fluid inclusions: modelling for T < 31°C and P < 400 bars.
  European Journal of Mineralogy
6
,
753
–771.
Touret, J. L. R. (
1986
).
CO2 transfer between the upper mantle and the atmosphere: temporary storage in the lower continental crust.
  Terra Nova
4
,
87
–98.
Van den Kerkhof, A. M. (
1990
).
Isochoric phase diagrams in the systems CO2–CH4 and CO2–N2: application to fluid inclusions.
  Geochimica et Cosmochimica Acta
54
,
621
–629.
Voll, G. (
1983
).
Crustal xenoliths and their evidence for crustal structure underneath the Eifel volcanic district.
  In: Fuchs, K., von Gehlen, K., Mälzer, H., Murawski, H. & Semmel, A. (eds) Plateau Uplift. The Rhenish Shield—a Case History. Berlin: Springer, pp.
260
–275.
Wells, P. R. A. (
1977
).
Pyroxene thermometry in simple and complex systems.
  Contributions to Mineralogy and Petrology
62
,
129
–139.
Wörner, G. & Fricke, A. (
1984
).
Fluid inclusions in corundum in a contact metamorphic xenolith of the Quaternary Wehr volcano (East Eifel, Germany).
  Neues Jahrbuch für Mineralogie, Monatshefte
1
,
39
–47.
Wörner, G., Schmincke, H.-U. & Schreyer, W. (
1982
).
Crustal xenoliths from the Quaternary Wehr volcano (East Eifel).
  Neues Jahrbuch für Mineralogie, Abhandlungen
144
,
29
–55.
Wörner, G., Staudigel, H. & Zindler, A. (
1985
)
Isotopic constraints on open system evolution of the Laacher See magma chamber (Eifel, W Germany).
  Earth and Planetary Science Letters
75
,
37
–49.