Petrography of the dike-gabbro transition at IODP Site 1256 (equatorial Pacific): The evolution of the granoblastic dikes



[1] The Ocean Drilling Program (ODP)/Integrated Ocean Drilling Program (IODP) three-leg campaign at Site 1256 (Leg 206, Expeditions 309 and 312) provides the first continuous in situ sampling of fast spread ocean crust from the extrusive lavas, through the sheeted dikes, and down into the uppermost gabbros (Cocos plate; East Pacific Rise; eastern equatorial Pacific). The lowest ∼60 m of the dikes above the gabbros were transformed to “granoblastic dikes” through a metamorphic overprint characterized by two-pyroxene domains formed under granulite-facies conditions. Equilibrium temperatures estimated by the two-pyroxene thermometer range between 930°C and 1050°C, implying that conditions within the granoblastic zone were appropriate for hydrous anatexis, with the potential to generate partial melts of trondhjemitic composition. The downhole evolution of the granoblastic overprint is expressed by systematic changes in texture, phase composition, and calculated equilibrium temperature, consistent with thermal metamorphism by a deeper heat source. Thermal modeling implies a long-lasting heat source located beneath the granoblastic dikes, providing thermal energy over several thousands of years. The most likely such source is a steady state, high-level axial magma chamber (AMC) located at the base of the sheeted dike section. We interpret the interval of granoblastic dikes as part of a dynamic conductive boundary overlying the AMC.

1. Introduction

[2] Ocean Drilling Program (ODP) Hole 1256D, was initiated by Leg 206 and continued by Expeditions 309 and 312 of Integrated Ocean Drilling Program (IODP). Site 1256 is located in the eastern equatorial Pacific on 15 Ma oceanic crust of the Cocos plate, that formed at the superfast spreading East Pacific Rise (220 mm/a full spreading rate). Hole 1256D penetrated the entire upper oceanic crust, passing through a ∼250-m thick sediment sequence, a ∼800-m thick lava series and a relatively thin, ∼350-m thick sheeted dike complex before finally extending ∼100 m into the uppermost gabbro which is composed of two individual bodies separated by a dike screen [Teagle et al., 2006]. The site was chosen to exploit the inverse relationship between spreading rate and the depth to axial low-velocity zones, thought to be magma chambers now frozen as gabbros, inferred from seismic experiments [Wilson et al., 2006]. During IODP Expedition 312 gabbro was first encountered at 1406.6 mbsf (meters below seafloor), within the depth range predicted by the extrapolation of the low-velocity zone depth versus spreading rate correlation for modern mid ocean ridges to the superfast spreading rate implied by magnetic survey data [Wilson et al., 2006]. Site 1256 has been proposed as a potential location for complete penetration of the ocean crust [Christie et al., 2006; Dick et al., 2006]. Initial drilling results from Site 1256, together with site maps and information on the geological setting, are given by Teagle et al. [2006, 2007]. An overview of the lithostratigraphic units of the core is given in Figure 1.

Figure 1.

Simplified lithostratigraphic column of the basement drilled at Site 1256D according to Wilson et al. [2006].

[3] The first complete penetration of the upper oceanic crust to reach the gabbroic section is necessarily an important reference section for the dike-gabbro transition of fast spreading ocean crust. In particular, the dike-gabbro transition corresponds in its structural level to the important boundary layer between the active magma system of the melt lens and the lower-temperature, convecting hydrothermal system within the sheeted dike section and above [e.g., Lister, 1974; Mével and Cannat, 1991], Hole 1256D provides the first in situ access for petrological and geochemical investigation of this geochemically important zone.

[4] This paper describes a detailed petrographic and microanalytical investigation focused on the lower dikes drilled during Expedition 312. The basic materials for this paper are the original shipboard thin sections, which were petrographically well investigated during the expedition [Teagle et al., 2006] and later analyzed by electron microprobe. In this paper, we present the individual petrographic and microanalytical results of the lowermost dikes directly above the gabbros which were transformed to the “granoblastic” dikes through a metamorphic overprint (Figure 1).

2. Methods

[5] Microscopic images presented in this paper are either from the shipboard image database of Expedition 312 [Teagle et al., 2006] or obtained onshore from the original shipboard thin sections. Sample names are shortened from the original ODP/IODP descriptions [see Teagle et al., 2006]. Details are given in Table 1.

Table 1. Samples From the Drill Core of Hole 1256D Used in This Study
Exp/LegaCoreScbTop (cm)Bot (cm)PieceDepth (mbsf)LithcUnitThin s. #dDescriptione
  • a

    Expedition and leg number. Data of samples from Legs 206 and 309 are published by Dziony et al. [2008].

  • b

    Section of the core.

  • c


  • d

    Shipboard thin section number.

  • e

    Abbreviated sample designation used in this study.

206927101a276.9pond1c 9_2_7_10
20613110911113351.4flow2 13_1_109_111
206322681a476.7flow8d 32_2_6_8
2063929101515.0flow13 39_2_9_10
206451929517553.4flow17 45_1_92_95
20649238401583.5flow19 49_2_38_40
206521571a600.8flow22 52_1_5_7
20657224261b647.1flow24a 57_2_24_26
309762041755.3flow27 76_2_0_4
30985376789b815.0flow31 85_3_76_78
30987263666832.7flow31 87_2_63_66
3091021707216a927.3flow34B 102_1_70_72
309118143485a1008.4flow40 118_1_43_48
30914717175151145.9dike51 147_1_71_75
3091541566014b1180.1dike54A 154_1_56_60
3091642115118191225.7dike59 164_2_115_118

[6] About 3500 single electron microprobe analyses (EPMA) were performed using a Cameca SX100 electron microprobe equipped with 5 spectrometers and an operating system “Peak sight.” All data were obtained using 15 kV acceleration potential, a static (fixed) beam, Kα emission from all elements, and the “PAP” matrix correction [Pouchou and Pichoir, 1991]. Most element concentrations were obtained with a beam current of 15 nA and accounting time of 20 to 120 s on peak and background. Averages are presented in Table 2. For comparison, some diagrams also include mineral compositions from 16 lavas and dikes of the upper section (Leg 206, Expedition 309) published by Dziony et al. [2008].

Table 2. Mineral Compositionsa
#/LitbTypcTex DomdPhaeAnfQualgNohSiO2TiO2Al2O3Cr2O3FeOMnONiOMgOCaONa2OK2OFClTotalMg#iAnj
  • a

    Dash (-), below limit of detection (see text for details); empty space, not analyzed; FeO = FeOtot; italics, one standard deviation.

  • b

    Sample and lithology; for complete description, see Table 1.

  • c

    Texture type according to Table 3.

  • d

    Textural domain: granbl, granoblastic; granbl dry, granoblastic “dry” paragenesis; granbl wet, granoblastic “wet” paragenesis; granbl ini, initial granoblastic domains within altered dikes.

  • e

    Phase: am, amphibole; cpx, clinopyroxene; il, ilmenite; mt, magnetite; opx, orthopyroxene; pl, plagioclase.

  • f

    Details of the analysis: co, core; ri, rim. For oxides: foc, focused analyzed; int, integral analyzed.

  • g

    Phase qualifier: hydro, secondary hydrothermal vein; lowest, compositions extremely low in Ti; poikbl, poikiloblastic; prim, primary magmatic; prim/mat, mix of primary magmatic and granoblastic compositions; prism, prismatic; pseudo, pseudomorph of magmatic clinopyroxene phenocryst; rel, primary magmatic relict; retro, formed by retrograde reaction. For details, see text.

  • h

    Number of analyses.

  • i

    MgO/(MgO + FeOtot)*100, molar.

  • j

    An content of the plagioclase, mol %.

Dikety3 cpxcoPrim1251.250.582.320.0610.200.19 15.9118.950.21-  99.6673.64 
       0.500. 0.661.340.03    4.48 
   cpxriprim450.290.752.02-14.550.29 13.5218.200.240.02  99.8862.34 
       0.310.070.11 2.170.05 1.400.910.020.02   5.92 
   plcoprim951.54 29.71 0.80  0.2114.003.760.03  100.05 67.17
       0.79 0.68 0.08  0.050.630.360.01    3.03
   plriprim653.52 28.31 0.91  0.1412.354.590.06  99.87 59.58
       2.03 1.39 0.13  0.101.631.020.02    8.54
   mtintprim50.5421.881.560.0566.903.35 0.100.46--  94.84  
       0.310.740.150.051.400.56 0.110.32       
Dikety4 cpxcoprim651.590.673.040.138.980.20 16.5819.370.22-  100.7876.70    1.12 
   cpxriprim750.700.842.99-10.760.21 15.3419.160.26-  100.2771.76 0.340.04 0.440.350.04    1.01 
   plcoprim952.91 29.00 0.77  0.1912.934.390.04  100.23 61.79
       0.59 0.38 0.13  0.030.390.300.01    2.28
   plriprim456.33 26.76 0.95  0.0610.035.970.08  100.19 47.91
       2.57 1.30 0.15  0.041.921.040.03    9.08
   mtintprim51.0122.710.580.0468.882.14 0.190.68--  96.24  
       0.462.750.150.032.810.44 0.060.38       
Dikety5 cpx prim5849.291.394.450.378.090.19 14.4420.970.30-  99.48176.10 
       0.550.190.510.120.400.04 0.300.330.03    1.10 
  granbl inicpx  853.240.060.44-7.530.27 14.5623.890.21-  100.2177.53 0.610.09 0.220.610.05    1.54 
   plcoprim351.54 29.80 0.80  0.1613.793.910.09  100.1 65.76
       0.17 0.08 0.05    1.34
   plriprim355.32 27.01 0.80  0.1810.465.710.15  99.619 49.89
       0.50 0.85 0.29  0.190.460.280.03    2.30
  granbl inipl matrix757.96 26.03 0.61  0.038.357.020.13  100.13 39.36
       2.21 1.57 0.07  0.021.700.990.10    8.12
  granbl iniam flaky449.061.055.900.2511.670.20 15.5212.111.110.05  96.9270.34 
       0.790.400.880.240.820.02 0.751.040.140.02   1.63 
  granbl iniilfoc 120.0446.31--49.002.99 0.050.20--  98.58  
       0.020.76  1.350.81 0.020.09       
  granbl inimtfoc 120.082.730.400.1589.290.19 -0.16--  92.99  
       0.020.980.120.161.370.10  0.11       
   mtintprim131.088.660.850.1082.770.45 0.360.64--  94.91 0.330.98       
Dikety5granbl inicpx  1151.600.431.21-9.910.27 14.6220.920.29-  99.2472.46 0.300.04 0.160.350.04    0.53 
  granbl inicpx lowest252.850.150.66-8.310.22 15.3122.910.25-  100.6676.67 
       0.410.060.06 0.990.05 0.300.230.04    2.48 
  granbl iniopx  652.500.290.62-20.440.52 23.071.58--  99.0066.81 
       0.300.020.03 0.150.05 0.140.04     0.11 
  granbl inipl matrix4954.92 27.83 0.94  0.0611.095.550.07  100.46 52.28
       1.10 0.64 0.13  0.180.830.520.02    4.15
  granbl iniam flaky2451.360.854.16-11.340.20-16.9811.860.850.060.170.1797.9972.74 
       0.990.180.64 0.940.04 0.690. 2.40 
  granbl iniilfoc 150.0345.360.030.0452.731.20 -0.09--  99.48  
       0.031.660.020.041.670.21  0.04       
  granbl inimtint 160.062.820.450.2088.770.11 0.050.09--  92.56  
       0.020.880. 0.030.04       
Dikety6granblcpx  2051.840.461.20-11.470.30 14.2320.680.27-  100.4668.87 0.300.04 0.190.310.04    0.57 
  granblcpx lowest153.220.090.51-8.800.30 14.9322.780.15-  100.8075.15 
  granblopx  3052.310.330.66-23.320.53 21.441.70--  100.3062.10 0.380.06 0.350.10     0.67 
  granblplcoprim/mat2053.98 28.32 0.89  0.0311.695.180.08  100.18 55.23
       0.85 0.49 0.12  0.020.620.370.03    3.03
  granblplrimatrix457.43 26.28 0.82  -9.296.310.12  100.24 44.56
       0.35 0.29 0.10    0.71
  granblam flaky352.280.453.390.0714.000.30 16.3210.240.470.05  97.5967.57 
       1.180.421.000.011.580.15 0.281.680.320.05   2.32 
  granblilint 110.0345.850.040.0951.741.12 0.050.10--  99.02  
       0.020.430.010.030.620.10 0.030.03       
  granblmtint -0.12--  92.65  
       0.021.430.080.051.820.09  0.02       
Dikety6granblcpx  653.120.100.53-8.820.29 14.6723.040.24-  100.8174.77 0.660.04 0.430.350.04    1.95 
  granblplcoprim/mat1753.57 28.40 0.94  0.0512.084.930.06  100.03 57.34
       0.64 0.53 0.11  0.020.440.260.02    2.14
  granblplrimatrix658.80 25.28 0.70  -7.867.240.13  100.01 37.22
       0.46 0.31 0.06   0.360.230.03    1.71
  granblam flaky1851.730.813.51-11.980.23-16.8111.570.740.090.130.1397.7371.45 
       0.680.240.49 0.600.03 0.440. 1.50 
  granblilint 250.0344.260.050.0753.151.06 0.070.08--  98.77 0.040.06       
  granblmtint 140.072.460.650.4089.210.12 0.080.10--  93.10 0.060.07       
Dikety7granblcpx rel951.390.532.340.179.140.28 15.0819.630.280.02  98.8674.68 
       0.990. 1.283.340.060.01   3.60 
  granblcpx  951.380.571.35-11.940.41 14.2919.420.27-  99.6368.10 0.490.05 0.170.460.02    1.04 
  granblam poikbl1048.661.285.48-12.680.25-16.3310.881.280.130.320.2497.5369.66 
       0.470.170.22 0.550.04 0.420. 1.32 
  granblam flaky651.040.784.16-11.730.25-16.6411.300.850.070.230.1397.1771.63 
       1.390.140.86 0.720.02 0.730. 2.11 
  granblpl matrix3153.94 28.06 0.93  0.0711.505.110.05  99.67 55.29
       0.43 0.43 0.17  0.120.350.180.01    1.58
  granblplcorel1951.02 29.96 0.79  0.0713.823.760.02  99.44 66.91
       0.29 0.18 0.06    0.78
  granblilfoc 7-44.620.05-51.091.55 -0.08--  97.39  
        0.570.02 0.320.07  0.05       
  granblmtint -0.06--  92.52  
       0.021.350.110.041.380.11  0.04       
Dikety8granblcpx  2051.490.591.41-9.640.24 14.9320.490.33-  99.1273.41    0.50 
  granblcpx cluster952.220.451.14-9.660.25 15.2120.550.32-  99.7973.75 
       0.350.130.25 0.540.03 0.210.610.03    1.02 
  granblcpx retro352.390.050.24-7.530.25 14.2323.790.25-  98.7377.13 0.840.06 0.290.690.13    2.35 
  granblopx  2353.120.310.67-19.140.43 24.161.550.04-  99.4069.23 0.420.04 0.360.130.01    0.75 
  veinopx cluster2052.880.340.72-19.590.43 23.841.680.04-  99.5268.45 0.300.04    0.48 
  granblplcomatrix4855.26 27.20 0.82  0.0410.295.890.07  99.58 48.92
       0.77 0.53 0.16  0.050.370.220.02    1.69
  granblplrimatrix757.96 25.75 0.67  -8.337.030.13  99.87 39.27
       0.51 0.37 0.01   0.370.280.03    2.00
  granblplcorel549.69 30.55 0.81  0.0514.593.300.02  99.01 70.91
       0.19 0.13 0.04    0.61
  granblam flaky1149.881. 
       0.780.190.420.010.420.03 0.330. 1.02 
  veinam  949.841.215.02-10.100.16-17.8111.211.700.120.910.1098.1975.86 
       0.470.160.25 0.260.02 0.66 
  granblam retro749.270.283.43-18.000.27-11.4011.840.490.05-0.1395.1652.84 
       2.390.050.43 2.230.05 1.890.630.100.01 0.07 6.70 
  granblilfoc 50.0634.580.040.3258.800.27 0.100.32--  94.51 0.080.07       
  granblmtfoc 50.071.720.650.7688.050.01 -0.21--  91.47  
       0.040.480.080.250.870.01  0.05       
Dikety8granbl drycpx  1051.920.571.45-10.530.33 15.1220.350.31-  100.5971.93 0.510.05 0.110.360.01    0.91 
  granbl drycpx rel250.620.443.230.3410.010.28 15.6517.530.530.02  98.6673.65 
       0.480. 0.051.410.150.01   2.79 
  granbl dryopx  1952.910.310.70-20.820.58 23.521.58--  100.4266.83 0.240.03 0.190.22     0.23 
  granbl drypl matrix2554.41 27.99 1.02  0.1711.185.410.06  100.24 53.14
       0.340.010.440.020.170.02 0.240.310.210.01    1.44
  granbl wetpl matrix2554.88 27.60 0.90  0.0510.765.650.08  99.91 51.05
       0.63 0.41 0.24  0.150.570.260.08    2.23
  granbl dryplcorel1749.93 30.71 0.93  0.0514.823.250.04  99.73 71.40
       0.25 0.18 0.04    0.61
  granbl wetam poikbl848.781.415.79 11.300.23 16.8911.231.520.190.400.1897.9272.72 
       0.750.130.48 0.580.02 0.430.430. 1.34 
  granbl wetam pseudo1649.281.485.690.0410.370.17 17.6311.361.810.21  98.0475.19 
       0.630.   0.38 
  veinam cluster649.701.255.16 10.670.19 17.1011.681.580.200.710.2098.4474.07 0.210.03 0.54 
  veinam hydro852.430.302.74 14.730.30 14.9011.990.400.04-0.0797.9164.14 2.450.08 0.01 6.82 
  granbl dryilint 100.0545.320.020.0551.511.11 0.050.30--  98.41  
       0.021.830.020.041.780.23 0.040.12       
  granbl wetilfoc 6-47.560.02-49.011.80 0.040.08--  98.52  
        0.570.02 0.580.16 0.020.02       
  granbl drymtint 100.051.890.700.4289.590.05 -0.18--  92.88  0.06       
  granbl wetmtint -0.12--  93.93  
       0.020.920.  0.04       
Dikety8granbl drycpx  351.380.521.26-10.000.35 14.5320.800.30-  99.1572.15 0.730.04 0.190.660.00    1.26 
  granbl dryopx  452.550.290.66-20.910.64 22.971.260.04-  99.3266.19 0.300.02    0.51 
  veinopx poikbl652.710.330.72-20.550.66 22.951.800.04-  99.7566.57 0.300.08    0.22 
  clusteropx prism952.310.350.72-20.550.60 23.141.830.05-  99.5566.75 
       0.390.020.03 0.250.04 0.330.070.02    0.46 
  granbl drypl matrix1053.68 28.25 1.14  0.0911.715.070.07  100.01 55.86
       0.500.030.410.010.140.03 0.170.430.290.01    2.21
  granbl wetpl matrix654.20 27.72 0.82  -11.055.580.07  99.44 52.05
       0.350.    1.05
  granbl dryplcorel549.94 30.77 0.78  0.1314.773.350.02  99.77 70.78
       0.460.    1.16
  granbl wetam flaky549.701.205.13-10.330.25-17.4511.631.270.150.610.1497.8775.07 
       0.340.110.18 0.080.03 0.400. 0.45 
  granbl wetilfoc 3-47.19-0.0650.451.41 -0.18--  99.29  
        2.00 0.051.870.01  0.01       
  granbl wetmtfoc 30.051.630.650.2689.200.05 -0.19--  92.03  
       0.010.730.  0.03       
Dikety8granbl drycpx cluster552.060.491.25-11.390.32 14.9119.700.28-  100.4070.00 0.110.02 0.090.340.02    0.17 
  granbl dryopx cluster952.680.340.70-21.510.54 22.671.830.03-  100.3065.26 0.310.04    0.37 
  granbl dryopx  1053.050.310.71-20.700.50 23.561.530.03-  100.3966.99 0.280.04    0.27 
  granbl wetpl matrix1055.01 27.96 0.71  -10.815.560.07  100.12 51.57
       1.35 0.89 0.12   0.960.560.01    4.71
  granbl drypl matrix855.48 27.63 0.84  -10.385.830.07  100.24 49.39
       0.81 0.43 0.12   0.560.340.01    2.76
  veinpl  1054.74 28.01 0.94  0.0310.675.620.09  100.11 50.91
       0.33 0.16 0.19  0.010.310.150.06    1.30
  granbl wetpl rel2749.96 30.81 0.89  0.0414.653.330.04  99.71 70.71
       0.40 0.28 0.05  0.010.330.170.01    1.45
  granbl wetam flaky949.171.145.42-11.160.20-17.1411.151.410.170.510.1797.6473.26 
       0.940.190.61 0.860.04 0.350.480. 1.81 
  veinam cluster2048.771.295.45-11.420.22-16.6111.301.450.410.570.2097.6772.17 
       0.900.170.49 0.520.04 0.320.360. 1.24 
  granbl dryilint 50.0444.790.03-53.521.10 0.080.06--  99.63 1.310.13 0.030.01       
  granbl drymtint 0.06---  94.72  
       0.030.840. 0.03        

3. Petrography of the Granoblastic Dikes

[7] Hole 1256D recovered the transition from low-temperature, relatively anhydrous alteration to high-temperature hydrothermal alteration in an intact section of oceanic crust. The alteration grade increases downward from the extrusives into the dikes, with low-temperature alteration phases in the lavas, progressing to chlorite and other greenschist minerals in the upper dikes (implying metamorphic temperatures > ∼250°C [Wilson et al., 2006]). Within the sheeted dikes, the grade of alteration increases downhole, with actinolite becoming more common than chlorite below 1300 mbsf and true hornblende with green brownish color occurring below 1350 mbsf, indicating hydrothermal temperatures approaching ∼400°C. The lowest ∼60 m of the sheeted dikes (∼1350 to 1407 mbsf) are partially to completely recrystallized, having developed the distinctive granoblastic textural features that led to their designation as “granoblastic dikes.”

3.1. Textural and Mineralogical Evolution of the Granoblastic Overprint

[8] In order to clearly document the changes in metamorphic overprint with depth in the sheeted dikes, especially within the narrow zone of the granoblastic dikes, we focused on the textural-metamorphic evolution of fine-grained samples of primary intersertal structure with progressive metamorphic overprint. The intersertal texture which is the most common texture of the basaltic rocks, is characterized by a framework of plagioclase laths, commonly associated with prismatic clinopyroxene, between which angular spaces are occupied by interstitial clinopyroxene, Fe-Ti oxide and glass, or their alteration products. In documenting the textural and mineralogical evolution of the lower dikes, Expedition 312 shipboard petrologists defined 7 textural types, assigning qualitative rank numbers [Teagle et al., 2006]. In this paper we add Type 8, which represents the highest grade of granoblastic overprint. The metamorphic-textural characteristics of the eight types are summarized in Table 3. Representative thin section images of selected textural types are shown in Figure 2. For each texture type (except type 1), we selected a reference thin section listed in Table 3. Type 1, representing the fully fresh rock including pristine interstitial glass, is not represented among the Hole 1256D thin sections. The downhole evolution of metamorphic grade is demonstrated in Figure 3.

Figure 2.

Downhole evolution of the granoblastic overprint documented in the textural and mineralogical changes within the intersertal texture. Representative metamorphic textures of type 3, 5, 7, and 8 are shown. For details, see text and Table 3. All images with plane-polarized light. Images are from the database of Expedition 312 from Teagle et al. [2006]. (a) Texture type 3: clinopyroxene modestly altered (<50%) by dusty brown fibrous masses (mostly actinolite); glass completely altered; fresh plagioclase filled in hollows by chlorite and dusty cryptocrystalline material; Fe-Ti oxides (titanomagnetite) are mostly primary; sample 173_2_6_10. (b) Texture type 5: clinopyroxene almost completely altered to fibrous aggregates of mostly actinolite and secondary oxides; hollows of plagioclase contain discrete crystals of oxides plus actinolite; first overgrowth of flaky brown green hornblende (top left); sample 192_1_11_13. (c) Texture type 7: granoblastic domains of secondary plagioclase, clinopyroxene, brown green hornblende (partly poikiloblastic), newly crystallized Fe-Ti oxide grains (both magnetite and ilmenite); hollows of plagioclase contain discrete crystals of clinopyroxene, hornblende, and Fe-Ti oxides; primary oxides no longer present; former primary prismatic clinopyroxene crystals can still be recognized by the disseminated oxide grains forming pseudomorphs together with hornblende (top left and bottom left); sample 198_1_45_49. (d) Texture type 8: granoblastic network of secondary plagioclase, clinopyroxene, orthopyroxene, and Fe-Ti oxide grains (both magnetite and ilmenite); characteristic of type 8 is the vein-like appearance of orthopyroxene (horizontal zone in the middle); the initial intersertal texture is still visible, but obscured; sample 207_1_ 10_15.

Figure 3.

Downhole evolution of metamorphic grade through the Expedition 312 sheeted dike section, shown by changes in texture of the intersertal-textured basalts. Textural types are defined in Table 2. For details, see text. Filled symbols correspond to those samples analyzed by EPMA. Also shown is the depth of the contact between the sheeted dikes and the uppermost gabbro intruded into them and the depths at which the first secondary clinopyroxene (cpx) and orthopyroxene (opx) were detected.

Table 3. Basalt Texture Types: Textural Development of the Metamorphic Overprinta
Texture TypebReference TSPrimary Igneous FeaturesSecondary Minerals
GlassCpxPlagioclase HollowsFe-Ti OxidesDusty Brown MaterialAmphibolePyroxene
  • a

    Variations as observed in microcrystalline to fine-grained rocks of intersertal to variolitic texture. Abbreviations: act, actinolite; hbl, hornblende; cpx, clinopyroxene; opx, orthopyroxene; TS, thin section.

  • b

    Representative images of selected textures are shown in Figure 2.

ty1not observed at Site 1256pristinefreshfreshall primarynonenonenone
ty213_1_109_111TS #109altered to clay mineralsfreshfreshall primarynonenonenone
ty3173_2_6_10 TS #3altered to chlorite and oxides<50% altered to dusty brown fibrous masseschlorite and dusty cryptocrystalline materialmostly primaryreplacing cpx act replacing cpx visible with EPMAnone
ty4178_1_31_34TS #17tiny fibrous masses and chlorite>50% altered to dusty brown fibrous massesdusty brown materialmostly primary, tiny secondary grains replacing cpxreplacing cpxvisible fibrous act masses replacing cpx and glassnone
ty5192_1_11_13 TS #32act/oxide overgrowths>90% altered to act and oxidesdiscrete tiny oxide and actinolitemostly primary, abundant tiny secondary grains in altered cpxrare, replacing cpxact replacing dusty brown material in altered cpx; First green, flaky hblfirst tiny granular cpx and opx in altered cpx only visible with EPMA
ty6196_1_32_33TS #36intense overgrowth of act, hbl, and oxide100% altered to act/hbl and tiny oxide grainsact/oxidesprimary grains, large secondary idioblastic grainsnonelarger, more prismatic act; green brown flaky hbltiny, anhedral, granular cpx and opx in altered cpx only visible with EPMA
ty7198_1_46_49TS #39fully overgrown by secondary phases100% altered to hbl, secondary cpx, opx and tiny oxidesact/oxides; secondary cpx, opx, and oxides in the granoblastic domainsno recognizable primary grains, increase in size of secondary grainsnoneflaky and poikiloblastic green brown hbl; prismatic actinolite still presentmicrogranular domains of cpx and opx; often associated with alteration of primary cpx
ty8205_1_10_14TS #49fully overgrown by secondary phasesonly rare visible as relics; altered to hbl, cpx, opx, oxidessecondary cpx, opx, and oxidessecondary anhedral grains within the granoblastic networknoneflaky and poikiloblastic brown green hbl; often forming veinsgranoblastic network of cpx, opx, plagioclase, and oxides; opx in large clusters and veins

3.1.1. Initial Granoblastic Stage

[9] Backscattered electron (BSE) imaging reveals that granular clinopyroxene, the key mineralogical indicator for the granoblastic dike lithology is first developed below ∼1325 mbsf, in dikes showing texture type 5. Texture type 5 is characterized by conversion of original altered, primary clinopyroxene, filled with oxide dust and actinolitic hornblende, to discrete clusters (granoblastic domains) of small granular clinopyroxene grains, often with associated hornblende, magnetite, ilmenite, and quartz (Figure 4). In rare cases, grains of orthopyroxene can be found within these aggregates. The compositions of the second-stage (prograde) metamorphic phases that form the granoblastic domains are significantly different from those of the primary igneous phases (see below). Plagioclase has a spotty, porous appearance in BSE images (Figures 2b, 2c, 4b, and 4d) reflecting a reaction from primary An-rich plagioclase to secondary, more albitic plagioclase. This texture is another characteristic feature of the granoblastic domains. During the initial granoblastic stage, preexisting μm-sized oxide particles formed during the original alteration of pyroxene (Figure 2c) coalesce to form larger grains (<50 μm in diameter) consisting of intergrown ilmenite and magnetite (Figure 4d). These granoblastic-phase oxides have textures and compositions different from those of the primary oxides (see below).

Figure 4.

Initial granoblastic stage within a dike of texture type 5. Sample 192_1_11_13. (a and b) Metamorphic aggregate of clinopyroxene (cpx), hornblende (hbl), ilmenite (il), and quartz (qz) surrounded by matrix plagioclase. (a) Optical microscope picture; plane-polarized light. (b) BSE image; the clinopyroxene is extremely low in TiO2 and Al2O3, thus clearly of metamorphic origin. The white spots within the aggregate are also ilmenite. Note that the surrounding plagioclase shows a spotty, porous appearance, reflecting a reaction from primary plagioclase rich in An content to secondary plagioclase richer in Ab, a characteristic feature of the granoblastic domains. (c) “Amphibolitic” patch within an initial granoblastic domain consisting of hornblende, plagioclase, ilmenite, and quartz. (d) Secondary oxide grains within a granoblastic domain with coexisting ilmenite and magnetite (with exsolution lamellae of ilmenite), well-suited for application to the 2-oxide thermo-oxybarometer.

[10] Texture type 6, is characterized by larger granoblastic aggregates that are visible under the optical microscope, and form continuous granoblastic networks. Primary matrix clinopyroxenes composition, rich in Al2O3 and Cr2O3-bearing, can no longer be detected.

3.1.2. Mature Granoblastic Stage

[11] In texture type 7, starting at a depth of ∼1350 mbsf, typical granoblastic domains have developed a microgranular, mosaic-like texture. In Type 7 samples, some domains contain obscure “ghost” outlines of former primary prismatic clinopyroxene crystals that are filled with oxide dust (Figure 2c). Texture type 8 is characterized by more or less continuous microgranular, granoblastic mosaics of clinopyroxene, orthopyroxene, plagioclase, hornblende, ilmenite and magnetite (Figure 2d). A general characteristic of the granoblastic overprint is that the primary magmatic precursor pattern, the intersertal texture, can still be recognized (Figures 2c and 2d). “Dry” and “Wet” Granoblastic Domains

[12] A characteristic feature of the mature granoblastic dikes is the coexistence of granoblastic domains with “dry” and “wet” parageneses, even within the same section (Figures 5a and 5b). A typical “dry” paragenesis is clinopyroxene-orthopyroxene-plagioclase-ilmenite-magnetite, a “wet” paragenesis is typically amphibolitic with the dominance of hornblende.

Figure 5.

BSE images of textural domains of the mature granoblastic stage within dike rocks of texture type 8. (a) Well-equilibrated domain of “dry” granoblastic crystallization; pl, opx, cpx, and oxides (to 90% ilmenites); opx and cpx in this image practically not distinguishable; pyroxenes show slight alteration to low-T amphibole (actinolite). (b) Domain of “wet” (amphibolitic) granoblastic crystallization of pl and hbl in the same thin section as Figure 5a; pl is strongly heterogeneous in composition (cores An ∼49 mol%, rims An ∼41 mol%). Note the cpx inclusions in the pl (arrow) showing the same composition as the cpx of the granoblastic “dry” domain in the same section. (c) Relictic pl phenocryst within a granoblastic network; composition of the relictic pl is typical for pl of fresh dikes (An 71). The rim of the relictic pl interfingering with the matrix pl shows the same compositions as the matrix pl; granoblastic matrix: cpx, opx, pl (some show An-richer cores), hbl, and two oxides. (d) “Dry” granoblastic network of cpx, opx, pl, and two oxides; pl show cores (black arrows) very rich in An (up to 73 mol%) interpreted as primary magmatic relics. (e) Poikiloblastic opx forming a vein enclosing plagioclase crystals within in granoblastic network of opx, cpx, pl, and two oxides; pl inclusions in opx show the same composition as the matrix plagioclase, implying that vein and matrix are in chemical equilibrium. (f) Relictic cpx phenocryst in a “dry” granoblastic matrix of pl, cpx, opx, and two oxides; core shows high Cr and Al contents like the cpx phenocrysts in the less altered dikes; rim is filled with oxides and shows the same composition as the matrix cpx with low Al contents and Cr below detection limit; a microprobe profile through this crystal is shown in Figure 6. Abbreviations: pl, plagioclase; cpx, clinopyroxene; opx, orthopyroxene, ox, oxide; hbl, hornblende. Samples: a, b, c, 203_1_10_14; d and e, 207_1_10_15; f, 205_1_10_15. Relics of Primary Magmatic Phases

[13] Relics of primary magmatic plagioclase and clinopyroxene are ubiquitous in the mature granoblastic domains. Rims of plagioclase relics typically interfinger with plagioclase of the granoblastic matrix (Figure 5c). Plagioclase of the granoblastic matrix often surrounds porous cores very rich in An, which are interpreted as relics of the primary matrix plagioclase (Figure 5d). Relics of primary clinopyroxene phenocrysts have been converted to secondary clinopyroxene (Figure 5f) in granoblastic “dry” domains, and to hornblende in “wet” domains. Poikiloblastic Orthopyroxene

[14] A characteristic feature of texture type 8 is the occurrence of mm-sized orthopyroxene clusters that are often arranged into vein-like zones. These zones are typically poikiloblastic (Figure 5e), sometimes monomineralic, and sometimes with interstitial plagioclase grains, identical in composition to those of the granoblastic matrix. These relationships preclude a primary magmatic origin for this type of orthopyroxene.

3.1.3. Retrograde Metamorphism and Late Hydrothermal Alteration

[15] Localized retrograde metamorphism of the granoblastic dikes is expressed by the development of two complementary, lower-temperature mineral assemblages. The first assemblage is characterized by poikiloblastic, pale green diopsidic clinopyroxene that is filled with opaque dust and coexists with calcite. The second assemblage is characterized by poikiloblastic albite blasts that coexist with actinolitic hornblende (Table 2).

[16] A final very late hydrothermal event stage also locally altered the granoblastic dikes. Hydrothermal alteration assemblages are characteristically present in veins and bands or as patches in the surrounding groundmass. They are characterized by greenschist facies assemblages including actinolite, albite, quartz, magnetite and sulfide. Often, hydrothermal actinolite or actinolitic hornblende forms pseudomorphs after original granoblastic hornblende and pyroxene while matrix plagioclase of the granoblastic stage is converted to very Ab-rich compositions.

4. Mineral Compositions

4.1. Clinopyroxene

[17] Primary clinopyroxenes in Hole 1256D lavas and upper dikes have high Al2O3 and Cr2O3 contents and are strongly zoned with respect to Al, Ti and Mg# [Dziony et al., 2008]. In contrast, the secondary clinopyroxenes in the granoblastic domains are characterized by higher CaO with low TiO2 and Al2O3, and Cr2O3 always below detection limit (Figure 6). Secondary clinopyroxenes of the early formed granoblastic domains are particularly low in TiO2 and Al2O3 (Table 2), becoming more homogeneous downhole, with TiO2 approaching primary values, while Al2O3 continues to be distinctly lower than primary values (Figure 6). The lowest TiO2 and Al2O3 contents are in retrograde poikiloblastic clinopyroxene (Figure 6).

Figure 6.

Selected compositional parameters for primary and secondary pyroxenes of lavas and dikes from Hole 1256D. Data points correspond to averages presented in Table 2. Data points with crosses as symbols correspond to clinopyroxene core compositions from fresh lavas and dikes from Leg 206 and Expedition 309 published by Dziony et al. [2008]; all other data are from Expedition 312 dikes. For these, texture types according to Table 3 are given. See text for details. (a) TiO2 versus Al2O3 in the clinopyroxenes. (b) Cr2O3 versus Mg# in the clinopyroxenes; symbols as in Figure 6a. (c) Plot in the pyroxene quadrilateral enstatite (En), ferrosilite (Fs), wollastonite (Wo); symbols for clinopyroxenes as in Figure 6a; circles for orthopyroxene; projection scheme according to Morimoto et al. [1988]. (d) Concentration profile for selected elements through a relictic clinopyroxene phenocryst shown in Figure 5f within a mature granoblastic dike.

[18] Relic primary clinopyroxenes in mature granoblastic dikes are compositionally similar to those of unaltered clinopyroxene of the upper dikes (Figure 6). In a concentration profile through a relic clinopyroxene phenocryst in a mature granoblastic dike (sample 205_1_10_14) (Figure 6d) Al2O3 and Cr2O3 are highest in the center of the crystal and decrease toward the rim to approach the clinopyroxene of the granoblastic matrix. The Mg#'s of secondary clinopyroxene are generally lower than those of the primary crystals (Figure 6). Since bulk rock Mg# is widely unaffected by the granoblastic overprint [Teagle et al., 2006], this observation implies that iron is retained in the rock as secondary magnetite, perhaps explaining the frequent occurrence of dense, very fine-grained opaques in altered pyroxene.

4.2. Orthopyroxene

[19] In contrast to clinopyroxene, the orthopyroxenes of the granoblastic dikes vary only slightly (Figure 7). Orthopyroxenes from different textural domains within the same thin section are compositionally identical within a single standard deviation (Table 2).

Figure 7.

TiO2 versus Al2O3 content for orthopyroxenes and clinopyroxenes from the granoblastic dikes from Hole 1256D. Data points correspond to averages presented in Table 2. Note that the orthopyroxene compositions cluster within a narrow field, while the clinopyroxenes show a considerable scatter. For comparison, the compositional variation of orthopyroxenes from the gabbros of Hole 1256D is also shown (unpublished data).

[20] Compositions of orthopyroxenes from the granoblastic dikes can clearly be distinguished from those of other Hole 1256D low-Ca pyroxenes. Pigeonites [Crispini et al., 2006; Dziony et al., 2008] are distinctly richer in CaO (4.4 wt% to 5.2 wt%; 5 samples [Dziony et al., 2008]) than orthopyroxenes of the granoblastic dikes (1.26 to 1.83 wt%), while orthopyroxenes from gabbro screens 1 and 2 are significantly richer both in Al and Ti (Figure 7).

4.3. Plagioclase

[21] Plagioclases of the granoblastic dikes vary in An content between 37 and 57 mol%, and are clearly distinct from primary plagioclases of the lavas and dikes which vary between 56 and 84 mol% [Dziony et al., 2008] (Figure 8). On the scale of a thin section, the compositional variability can be high (e.g., 39–71 mol% An in sample 203_1_10_14, Table 2), owing to the ubiquitous presence of relic, high-An primary matrix plagioclase and to the zoning to Ab-rich rim compositions in newly formed granoblastic plagioclase. Plagioclases of the granoblastic dikes can also be distinguished from primary plagioclase by their higher K and lower Mg contents (Figure 8).

Figure 8.

MgO versus An content for primary and secondary plagioclases of lavas and dikes from Hole 1256D. Data points correspond to averages presented in Table 2. Data points with crosses as symbols correspond to plagioclase core compositions from fresh lavas and dikes from Leg 206 and Expedition 309 published by Dziony et al. [2008]; all other data are from Expedition 312 dikes. For these, texture types according to Table 3 are given. See text for details.

[22] Within the granoblastic domains, both plagioclase inclusions in orthopyroxene and clinopyroxene, and pyroxene and amphibole inclusions within plagioclase are equivalent in composition to their matrix counterparts (Figures 5e and 5b, respectively). This implies that for the granoblastic paragenesis local chemical equilibrium was approached.

[23] For relic primary plagioclase phenocrysts (e.g., Figure 5c), however, high An contents are preserved, but Mg contents have been reset to the low values of the granoblastic matrix plagioclase (Figure 8). In a microprobe profile from the margin of a relict plagioclase phenocryst into an adjacent matrix plagioclase grain, An and TiO2 display strong concentration gradients, but Mg does not (Figure 9). This reflects significantly different diffusion behavior, which is discussed in section 4.2.

Figure 9.

Concentration profile for selected compositional parameters through the marginal part of a relictic plagioclase phenocryst and adjacent plagioclase grains of the granoblastic matrix (sample 205_1_10_14; texture type 8). The arrow in the BSE image marks the location of the profile (length: 32 μm). The whitish phases are hornblende, pyroxene, and oxide of the granoblastic matrix (including their alteration products). (a) An content. (b) K2O, TiO2 and MgO. Note that the shape of the profiles for the selected parameters is different (short profile for An content; long profile for TiO2; absent concentration gradient for MgO), reflecting the different diffusion behavior of the involved components. See text for details.

4.4. Amphibole

[24] All analyzed amphiboles of the granoblastic assemblages correspond to magnesiohornblende in composition, with one exception corresponding to actinolite (see section 5.3). In a F-versus-Cl diagram for discriminating between magmatic and hydrothermal/metamorphic amphiboles [Coogan et al., 2001], the amphiboles of the granoblastic assemblages plot outside the field of Coogan et al. [2001], with high F and Cl (up to 0.9 and 0.2 wt%, respectively) contents. While the high Cl contents are normally indicate a seawater-influence, high F contents imply that F-rich fluids were also involved during the formation of the granoblastic dikes. These are presumably late-stage magmatic fluids derived from the gabbros beneath. In contrast, both actinolitic amphibole from a crosscutting hydrothermal vein, and hornblende from a retrograde cluster (see section 2.1.3), have low F/Cl ratios, implying formation by later, seawater-derived hydrothermal fluids.

4.5. Oxides

[25] Oxides are present in the granoblastic dikes as granular magnetite and ilmenite solid solutions with significantly different compositions from the titanomagnetite that characterizes Hole 1256D fresh lavas and dikes. The magnetites of the granoblastic dikes differ from titanomagnetites of the fresh lavas and dikes in having low ulvospinel contents (8 to 18 mol% versus 61 to 74 mol%), and low Al2O3 contents (0.4 to 0.9 versus 0.9 to 2.3 wt%) [Dziony et al., 2008] (Figure 10). The ilmenite solid solutions of the granoblastic dikes also differ from primary, near-pure ilmenite of the lava pond in their high hematite contents [Dziony et al., 2008]. Finally, occasional large magnetite crystals that appear to be of primary origin are low in Al and ulvospinel component (Figure 10), implying that their compositions were reset to the temperature and redox conditions of the granoblastic overprint.

Figure 10.

Selected compositional parameters for primary and secondary Fe-Ti oxides of lavas and dikes from Hole 1256D. Data points correspond to averages presented in Table 2. Data points with crosses as symbols correspond to oxide compositions from fresh lavas and dikes from Leg 206 and Expedition 309 published by Dziony et al. [2008]; all other data are from Expedition 312 dikes. For these, texture types according to Table 3 are given. Projection scheme includes minor cations. See text for details. Figure 10a is a plot for magnetites and ilmenites in the FeO-FeO1.5-TiO2 ternary diagram; included are the lines for ilmenite-hematite and ulvospinel-magnetite solid solutions; symbols as in Figure 10b.

4.6. Downhole Mineral Evolution

[26] Although the bulk composition of the granoblastic dikes does not change with depth, there is a steady, downhole compositional evolution of most phases (Figure 11) with increasing metamorphic overprint. This evolution is, in part a record of temperature increase with depth and is consistent with geothermometry, as discussed in section 5.2, and with the hypothesis of Teagle et al. [2006] that the granoblastic overprint in the lower ∼ 60 m of the sheeted dikes results from prograde metamorphism. Like the silicates, the Fe-Ti oxides also show compositional trends downhole, e.g., expressed by increasing contents of Al2O3 and decreasing contents of ulvospinel component in magnetite. Of special interest are a clear downhole increase in F-content and the absence of a trend for Cl. This contrast is consistent with upward percolation of magmatic fluids derived from gabbroic intrusions below. Also of interest is a systematic decrease in clinopyroxene Mg# that is not compatible with a simple temperature increase with depth. This special feature of the reaction forming secondary pyroxene is discussed in the next section.

Figure 11.

Evolution of mineral compositions with depth within the granoblastic dikes. Data are from Table 2. Open symbols for clinopyroxenes from dikes with initial granoblastic overprint correspond to those analyses with exceptional low contents of Al2O3 and TiO2 excluded from the average composition (see Table 2 and text for details). Two filled symbols at the same depth are related to compositions for different textural domains within the same sample. The contact to the gabbro is at 1406.6 mbsf. Abbreviations for minerals as in Table 2.

5. Discussion

5.1. Granoblastic Parageneses and Reactions

[27] The most common mineral parageneses observed in the granoblastic dikes are as follows (for abbreviations, see Table 2): (1) CPX + PL + IL + MT, (2) CPX + OPX + PL + IL + MT, (3) OPX + PL + IL + MT, and (4) AM + CPX + PL + IL + MT.

[28] Taken as a whole, parageneses 1 to 4 can be regarded as products of a metamorphic reaction starting from the primary magmatic assemblage clinopyroxene–plagioclase–magnetite. In detail, the reactions leading to the granoblastic parageneses are very complex, involving the formation of interim phases, especially actinolitic hornblende and quartz. The preservation of these intermediate phases in some of the early formed granoblastic aggregates (texture type 5; Figure 4b) allows us to infer, by careful observation, details of the reaction pathways.

[29] The earliest stage of alteration involves the hydrous transformation of primary prismatic clinopyroxene into smaller secondary anhedral grains begins with alteration of primary clinopyroxene to fibrous actinolite plus extremely fine-grained (<5 μm) coexisting ilmenite and magnetite (sample 189_1_68_69 [Teagle et al., 2006, Figure F208A]), with compositions typical for the granoblastic dikes (Figure 10 and Table 2). In an advanced stage of initial granoblastic overprint, secondary anhedral clinopyroxene form discrete clusters that still incorporate smaller oxide grains (sample 194R_1_36_37 [Teagle et al., 2006, Figure F208B]). With increasing metamorphic overprint deeper in the section, the patches of isolated small anhedral secondary pyroxene grains evolve into equigranular frameworks with secondary plagioclase, amphibole, and oxides forming typical granoblastic networks (Figure 5). During this maturing of the granoblastic network, most of the smallest oxide grains coalesce to form larger grains. Some of the small micron-sized oxide grains remain as inclusions within the granular plagioclases and pyroxenes (e.g., Figures 5a, 5c, and 5f).

[30] During the maturing process, the ubiquitous, early formed actinolitic amphiboles may evolve along either “wet” or “dry” reaction paths. The resulting “wet” paragenesis is dominated by magnesiohornblende, while the more abundant “dry” parageneses may include any of the assemblages 1 to 3 listed above. Both “wet” and “dry” granoblastic domains are commonly observed in the same thin section (Table 2). The reasons for the patchiness are unclear. Most probable are locally varying fluid compositions.

[31] The secondary clinopyroxene of the granoblastic domains has generally lower Mg#'s relative to primary igneous clinopyroxene (Figure 6). Those secondary clinopyroxenes within initial granoblastic domains showing the lowest Al2O3 and TiO2 contents are always those with the highest Mg# (Table 2 and Figure 11). We assume that this is due to the actinolitic precursor which was obviously very rich in Mg#. This is indicated by the presence of significant amounts of secondary magnetite in these actinolitic precursor domains, implying that the actinolitic amphibole must have been higher in Mg# when balancing the composition back to primary clinopyroxene.

[32] The downhole evolution for Mg# in clinopyroxene clearly shows a negative correlation with depth (Figure 11), suggesting that during the maturing of the granoblastic domains more and more of the ubiquitous oxide inclusions are incorporated into the clinopyroxene. Overall, the evolution of the Mg# in the granoblastic phases seems to be very complex, also indicated by oxybarometry (see next section) revealing that the redox conditions during the granoblastic overprint were highly oxidizing in contrast to the primary magmatic formation. A consequence of this is the stabilizing of magnetite and the increase of Fe3+/Fe2+ in the mineral phases, which in turn influences the Mg# of the involved silicate minerals.

5.2. Constraints on Temperature and Redox Conditions of the Granoblastic Overprint

5.2.1. Constraints on Temperature

[33] We used four independent geothermometers to estimate equilibrium temperatures prevailing during the granoblastic overprint. The results are presented in Table 4. Although there are some reversals, temperatures increase with depth in all the models (Figure 12), although the scales of the estimated temperatures are surprisingly different.

Figure 12.

Evolution of temperature and redox conditions with depth within the granoblastic dikes. For details on the used geothermometers and oxybarometer, see text and Table 4. Data are from Table 4. Two symbols at the same depth are related to estimations for different textural domains within the same sample. The redox value for the primary magmatic rocks is derived from a sample from the lava pond where coexisting ilmenite and magnetite occur as published by Dziony et al. [2008]. The contact to the gabbro is at 1406.6 mbsf.

Table 4. Results of Geothermometry and Oxybarometrya
SampleLith/Textb2-Pyrc2-Pyr ErrdAmph-PlageTi-in-AmphfAna. Modeg2-OxidehΔNNOi
  • a

    All temperatures in °C; values for oxygen fugacity expressed in ΔNNO.

  • b

    Lithology/textural domain; abbreviations as in Table 2.

  • c

    The 2-pyroxene-thermometer according to Andersen et al. [1993].

  • d

    Errors of the 2-pyroxene-thermometer according to Andersen et al. [1993].

  • e

    Amphibole-plagioclase thermometer according to Holland and Blundy [1994].

  • f

    Ti-in-amphibole thermometer according to Ernst and Liu [1998].

  • g

    Mode of analysis for ilmenite/magnetite (foc, focused beam; int, integral); integral means that finest exsolution lamellae were observed; for such analyses, the following is valid: Treal < Tcalculated; ΔNNOreal > ΔNNOcalculated.

  • h

    Improved 2-oxide thermometer according to Sauerzapf et al. [2008]; values below 600°C are of larger error, indicated by a tilde (∼), since these estimations are not covered by the experimental database.

  • i

    Improved 2-oxide oxybarometer according to Sauerzapf et al. [2008].

  • j

    Analyses given by Dziony et al. [2008].

9_2_7_10jlava pond    foc/foc907−1.1
189_1_68_69granbl ini  708688foc/foc∼5902.8
192_1_11_13granbl ini935141741650foc/int6122.8
196_1_32_33granbl  745652int/int6082.9
198_1_46_49granbl  785658foc/int6212.6
203_1_10_14granbl wet  798713foc/foc6463.3
203_1_10_14granbl dry100635     
205_1_10_14granbl wet  854747foc/int∼5502.9
205_1_10_14granbl dry94458  int/int∼5703.1
207_1_10_15granbl wet  832714foc/foc∼5403.2
207_1_10_15granbl dry94551     
209_1_8_10granbl wet  841703   
209_1_8_10granbl dry104553  int/int7161.7
209_1_8_10vein  838729   

[34] The highest temperatures, calculated by the 2-pyroxene thermometer of Andersen et al. [1993], represent the highest grade of thermal imprint at temperatures from 940°C to 1050°C. The lowest temperatures from this thermometer are for coexisting pyroxenes in an initial granoblastic aggregate (e.g., Figure 4). This low temperature has a very high uncertainty (±141°C, Table 4) implying that the coexisting pyroxenes in the initial aggregates are far from equilibrium.

[35] Temperatures calculated from an amphibole-plagioclase thermometer [Holland and Blundy, 1994] are significantly lower, starting from 700° in the dikes with initial granoblastic overprint up to 850°C in the mature granoblastic dikes near the contact to the gabbro. This thermometer reveal the most consistent temperature-depth trend (Figure 12), although the Ti-in-amphibole thermometer [Ernst and Liu, 1998] calculates a similar trend, shifted to lower temperatures between 650 to 750°C.

[36] The temperatures derived from the 2-oxide geothermometer [Sauerzapf et al., 2008] are lowest of all, between 540 and 650°C for most oxide pairs. Only one sample directly above the contact to the gabbro has a higher temperature of 720°C. Since magnetites in the granoblastic dikes universally contain ilmenite exsolution lamellae, most likely the temperatures were reset by re-equilibration during cooling.

[37] Temperatures estimated with the same geothermometer for different textural domains within a single section are statistically indistinguishable (Table 4 and Figure 12). For example, the close match of the 2-pyroxene temperatures in sample 203_1_10_14, and the fact that the overall variations in pyroxene compositions in this sample are minimal, indicates general equilibrium conditions for the formation of the “dry” phase assemblages. A similar situation is indicated for typical “wet” parageneses (e.g., amphibole-plagioclase temperatures for sample 209_1_8_10).

[38] The wide range of temperatures yielded by the four thermometers is striking. Temperatures derived from coexisting pyroxenes in “dry” parageneses are 1–200°C higher than those derived from amphiboles in the “wet” parageneses. These different temperatures may, to some extent, reflect equilibria reached at different stages of a dynamic metamorphic process. For example, the high 2-pyroxene temperatures most likely record the peak of metamorphism, exceeding 1040°C in the “dry” parageneses. At subsequent, lower temperatures, hydrous fluid circulation was enabled, forming amphibole-bearing parageneses at temperatures between 700 and 850°C. The occurrence of poikiloblastic amphiboles in “wet” granoblastic domains enclosing granular clinopyroxenes of the same composition as those coexisting with orthopyroxene in the “dry” parageneses, supports this model (see next section and Figure 13b). Owing to the absence of water, the kinetics in the “dry” parageneses were apparently too slow for re-equilibration of the two pyroxenes at lower temperatures.

Figure 13.

Complex textural and mineralogical relationships in the granoblastic dikes demonstrated in sample 205_1_10_14. For all different textural domains shown here, average mineral compositions can be obtained from Table 2. Included are the results of thermo-oxybarometry (for details, see text and Table 4). All images with plane-polarized light. Except for Figure 13c, all images are from the database of Expedition 312 from Teagle et al. [2006]. (a) Thin section scan. Included are the different textural zones: wet, “wet” paragenesis; dry, “dry” paragenesis; HT-V, high-temperature vein, formed during the main stage of the granoblastic overprint; LT-V, low-temperature vein, formed after the granoblastic overprint by secondary hydrothermal activity. Insets correspond to the thin section photographs shown in Figures 13b–13f. (b) Example for a typical granoblastic “wet” domain with the paragenesis hornblende (brown green)–plagioclase-two oxides (±clinopyroxene). Note that the hornblende shows poikiloblastic growth. (c) Crossing of two veins: A early high-temperature vein (paragenesis: hornblende-plagioclase-two oxides) is cut by a secondary, low-temperature vein consisting of actinolitic hornblende plus albite. (d) Example for typical granoblastic “dry” domain with the paragenesis clinopyroxene-orthopyroxene-plagioclase-two oxides. (e) Relic of a primary magmatic clinopyroxene phenocryst within the granoblastic “dry” domain (associated with plagioclase phenocrysts). While the core still shows affinities to the primary magmatic compositions (high in Al and Cr), the rim shows the same composition as the clinopyroxene of the granoblastic matrix (low in Al; Cr below detection limit). The microstructural and chemical characteristics are similar to those shown in Figures 5f and 6d. (f) Relic of a primary magmatic clinopyroxene phenocryst within the granoblastic “wet” domain, converted to hornblende during the granoblastic overprint. The hornblende is mostly secondarily altered to actinolitic hornblende or actinolite.

[39] In spite of the complexities, the geothermometers generally record increasing temperatures with depth (Figure 12) consistent with a model of prograde metamorphism caused by a heat source below the sheeted dikes. The observation that different mineral pairs record different temperatures suggests that each may record a transient local equilibrium within a dynamically evolving prograde metamorphic event.

5.2.2. Incipient Hydrous Partial Melting

[40] Hydrous fluids liberated by the dehydration of altered dikes during the granoblastic overprint have the potential to trigger partial melting. According to Beard and Lofgren [1991] partial melting in metabasalts of MORB-type compositions starts at temperatures of 875°C at water-saturated conditions and a pressure of 100 MPa. Koepke et al. [2004] showed that oceanic gabbro starts to melt at 900°C at a pressure of 200 MPa provided the water activity is high. High water activities are probable, since the water solubility is low in silicate melts at the shallow pressures (∼50 MPa) prevailing at the base of the sheeted dikes.

[41] Melts generated by initial hydrous partial melting in MORB-type systems are typically trondhjemitic [e.g., Dixon-Spulber and Rutherford, 1983; Beard and Lofgren, 1991; Koepke et al., 2004, 2007; Berndt et al., 2005]. A 20 mm-wide, fine-grained vein of similar trondhjemitic composition intrudes the granoblastic dikes near the dike-gabbro boundary at 1404 mbsf in Hole 1256D [Teagle et al., 2006, Figure F227]. Felsic igneous rocks are also relatively abundant in the coarse sand-size material recovered in junk baskets during hole clearing operations at 1373 mbsf [Teagle et al., 2006]. These leucocratic rock fragments, which consist of plagioclase, quartz, and altered green hornblende [Teagle et al., 2006, Figure AF2], must have been derived from leucocratic intrusions that were not recovered in the core. These observations imply that limited hydrous partial melting did occur during the peak of the granoblastic overprint, locally forming hydrous trondhjemitic melts and complementary dehydrated granulitic residua.

[42] This situation is very similar situation to a contact aureole in the Troodos ophiolite, Cyprus [Gillis and Coogan, 2002]. Here, dike rocks at the base of the sheeted dike complex close to the underlying gabbro, were partially melted, forming leucocratic melts and residual pyroxene-hornblende hornfels.

5.2.3. Constraints on Redox Conditions

[43] Redox estimates based on the two-oxide oxybarometer of Sauerzapf et al. [2008] indicate highly oxidizing conditions for the granoblastic overprint, with a mean ΔNNO value of +2.8 (range +1.7 to +3.3, Table 4; NNO is the oxygen fugacity defined by the Ni-NiO buffer). These high redox values exceed normal magmatic values and contrast strongly with a single redox value of ΔNNO −1.1 determined for coexisting Ti oxides in a lava pond sample from Hole XXX [Dziony et al., 2008]. This much more reducing value is within the range typical for fresh MORB basalts from mid-ocean ridges [Christie et al., 1986; Bézos and Humler, 2005].

[44] The highly oxidizing conditions associated with the granoblastic overprint can be explained either: as an inheritance from the initial low-grade alteration of the dikes by oxidized seawater derived fluids [Teagle et al., 2006]; or as a result of high water activities during the granoblastic overprint. The dependence of oxygen fugacity on the water activity in hydrogen-buffered systems has been shown by Botcharnikov et al. [2005]. In this case, an increase from low water activity during the primary magmatic processes, to high water activity during the granoblastic overprint has the potential to shift the oxygen fugacity toward highly oxidizing conditions. Individual redox values for both “wet” and “dry” domains are more or less identical (Table 4). However, the corresponding temperatures are <600°C, implying that the re-equilibration took place at lower temperatures. Thus, the estimated redox values are obviously related to processes during a late hydrothermal circulation and cannot be used to constrain the redox conditions during the peak of metamorphism.

5.2.4. Duration of the Granoblastic Overprint Event

[45] Primary magmatic phenocrysts that survived the granoblastic overprint exhibit apparently diffusive concentration profiles at their boundaries with the granoblastic matrix (Figures 6d and 9a). If the metamorphic temperature, and the diffusivity of a given species are known, the duration of the granoblastic event can be calculated. For this calculation, we measured An content profiles in two plagioclase phenocrysts from a “wet” granoblastic domain in a single sample (205_1_10_14). Most profiles are not symmetrical and exhibit irregularities, probably owing to inhomogeneities caused by alteration, inclusions, and/or orientation effects. Two of the profiles (one is shown in Figure 9), however, show the typical shape of diffusion profiles resulting from exchange between plagioclase and matrix. Both profiles are excellently fitted by the diffusion equation [Crank, 1975] assuming constant diffusivity and an initially homogenous plagioclase (r2 = 0.989 and 0.994). To estimate the time required to develop the diffusion profile, we used the equation

equation image

where x is diffusion length, D is diffusion coefficient, and t is duration.

[46] Since the “wet” granoblastic domains equate to hydrous conditions, we used the NaSi-CaAl interdiffusion coefficient for water-saturated conditions [Liu and Yund, 1992]. We used an average temperature of 800°C, calculated for the “wet” granoblastic domain of the actual sample (Table 4). By extrapolation of the diffusion data of Liu and Yund [1992] to this temperature, we estimated for the NaSi-CaAl interdiffusion coefficient in plagioclase a value of D = 9.6*10−24 m2/s. This approach estimates the duration of the metamorphic overprint as 23,000 and 5,400 years, respectively.

[47] Despite the large difference between these values, and the errors due to the uncertainties in temperature and water activity, which in turn affect the NaSi-CaAl diffusivity in plagioclase drastically [e.g., Baschek and Johannes, 1995], these results may help to define the timescale of the metamorphic overprint, which is in the range of several thousand to 10,000 years. This rough estimate implies that a long-lasting heat source provided thermal energy from beneath the granoblastic dikes for at least several thousand years. This duration is consistent with the presence of a high-level axial magma chamber at the base of the sheeted dikes.

[48] Trace elements (K, Ti, Mg) were also included into the profile measurements, showing systematic features. For Ti and K characteristic concentration profiles were observed showing an opposite trend compared to An (Figure 9). This is due to the fact that these elements are typically enriched in the plagioclases of the granoblastic framework relative to the primary minerals. The recorded profiles for Ti and K are flatter compared to An, which is in accord with higher diffusivities for these elements. Concerning Mg, no concentration profile was observed (Figure 9). This is in accord with very fast Mg diffusion in plagioclase (F. Costa, personal communication, 2008), which obviously wiped out the primary concentration gradient illustrated in Figure 8.

5.3. Complexity of the Granoblastic Dikes

[49] The mature granoblastic dikes are characterized by complex, small-scale textural and mineralogical variability (e.g., Sample 205_1_10_14; Figure 13). On the scale of a thin section, several types of textural zones can be identified, including “wet” and “dry” granoblastic domains or matrix networks, relict magmatic phenocrysts, and crosscutting vein systems.

[50] The “dry” granoblastic mineral assemblage of paragenesis 2 (section 5.1) was formed under granulite-facies conditions, while the hornblende-bearing “wet” domains formed under hydrous amphibolite-facies conditions. In places, the hornblende poikiloblastically encloses clinopyroxene that is compositionally very similar to clinopyroxene of the “dry” granoblastic zones. This observation implies that the “wet” assemblage represents a later hydrous metamorphic stage. The temperature for this stage was 100 to 200°C lower than the peak represented by the “dry” granulitic paragenesis. In both domains, however, the two-oxide thermo-oxybarometer records very similar, highly oxidizing conditions at equilibration temperatures below 600°C (Table 4), implying that both domains re-equilibrated at distinctly lower temperatures compared to those of the peak of metamorphism.

[51] The metamorphic assemblages derived from primary clinopyroxene phenocrysts in the “wet” domains are quite different from those of the “dry” granoblastic domains. In the “dry” matrix, these were converted to secondary clinopyroxene (Figure 13e). The cores of such pseudomorphoses still show the compositional characteristics of magmatic phenocrysts (high Cr and Al contents, Figure 6d). In the granoblastic “wet” domain, former clinopyroxene glomerocrysts were transformed to hornblende (Figure 13f; analysis see Table 2), for which an equilibration temperature of ∼760°C was estimated. More or less identical temperatures were estimated for hornblendes of the granoblastic “wet” domain (Figure 13b), implying that these pseudomorphoses were formed during the hydrous amphibolite-facies stage of the metamorphic imprint.

[52] Crosscutting veins are of two different types (Figures 13a and 13c). Veins of the first type bear the mineral assemblage hornblende-plagioclase-ilmenite-magnetite, very similar to the surrounding “wet” granoblastic matrix; such veins show a continuous textural gradation into the matrix, and the mineral compositions are very similar. These observations suggest that veins of this type equilibrated with the matrix during the “wet” stage of the granoblastic metamorphism. On the basis of their style and orientation, we postulate that these veins formed during early hydrothermal alteration of the precursor dike rocks, prior to the metamorphic overprint. Veins of the second type are characterized by a lower-temperature assemblage dominated by actinolitic amphibole (Table 2) and albite. They crosscut both the high-temperature veins (Figure 13c) and the granoblastic matrix and are therefore of very late origin.

5.4. Formation of the Granoblastic Dike

[53] Several lines of evidence suggest that the sheeted dike swarm at Site 1256 had cooled significantly prior to the metamorphic event that led to formation of the granoblastic mineral assemblages. During, or after initial cooling, the dikes underwent typical hydrothermal alteration as seawater-derived fluids penetrated the brittle crust through microcracks and veins. Later, a heat source below the base of the sheeted dike section led to prograde metamorphism of these altered metabasalts to form the complex parageneses of the granoblastic dikes. The following observations provide evidence for initial cooling of the sheeted dikes followed by prograde metamorphism:

[54] 1. In mature granoblastic dikes, numerous relics of very fine-grained intersertal to variolitic textures, identical to those of the unaltered dikes, are preserved (Figures 2c and 2d). These textures can only be formed by effective quenching in a relatively cool, brittle environment (≤400°C [e.g., Mével and Cannat, 1991]). In this respect, the Hole 1256D granoblastic dikes are distinct from the fine-grained granulite-facies gabbronoritic “protodikes” of the sheeted dike root zone in the Oman ophiolite [Nicolas and Boudier, 1991; Nicolas et al., 2008]. These protodikes are generally interpreted as having crystallized at high temperatures on the boundary of a magma-rich section of the crust.

[55] 2. Our petrographic studies reveal that the evolution of the granoblastic dikes starts with very small, micron-sized aggregates of initial granoblastic domains, which develop in the greenschist-facies matrix. Since the matrix is without doubt the result of hydrothermal circulation, and since the new, first aggregates (as precursor for the granoblastic dikes) develop within this matrix, it is implied that the granoblastic dikes suffered a previous hydrothermal alteration prior to the granoblastic overprint. In some samples (e.g., 1256d_196r_1_32_33) it is clear that brownish magnesiohornblende of the granoblastic paragenesis develops from fine fibrous actinolite aggregates, and not vice versa.

[56] 3. Characteristic features of the granoblastic dikes are the presence of high-temperature veins and vein-like or patchy arrangements of orthopyroxene cluster. This correlates well with textures known from hydrothermal alteration in a brittle regime. This type of alteration typically produces locally highly variable textural and compositional inhomogeneities, such as small local veins and haloes. We infer that this type of heterogeneity acts as precursor for the similarly scaled, patchy distribution of multiple textural and compositional domains within the granoblastic dikes. High-temperature amphibolitic veins which are fully equilibrated within the granoblastic matrix provide additional evidence for early, lower-temperature alteration.

[57] 4. In Samples 194_1_36_37 and 209_1_8_10, fresh clusters of orthopyroxene and subordinate clinopyroxene incorporate single grains that are surrounded by fine coronas of sulfide granules. The pyroxene-sulfide assemblage is clearly grown during the granoblastic recrystallization event, and is not product of the later hydrothermal alteration process. Two-pyroxene thermometry implies equilibration temperatures >1000°C for the formation. Most probably, the orthopyroxene grew from alteration haloes which included sulfide minerals produced by the first hydrothermal alteration prior to the metamorphic event.

[58] 5. Relict plagioclase and clinopyroxene phenocrysts within the granoblastic dikes have been significantly affected by intracrystalline diffusion exchange during the granoblastic event. The estimated duration of these processes (section 4.2.4) implies that the heat source beneath the granoblastic dikes provided thermal energy over several thousand years.

[59] 6. In relic primary clinopyroxene phenocrysts of sample 205_1_10_14, numerous oxide inclusions occupy outer zones where the composition has been reset to that of secondary granoblastic clinopyroxene. Such inclusions are not present in fresh pyroxene phenocrysts in the upper dike section. They first appear in altered primary clinopyroxene during the initial phase of granoblastic formation and survived the granoblastic overprint (section 5.1) Thus, we interpret the small oxide inclusions in the outer parts of relict phenocrysts as remnants of a former alteration process, predating the granoblastic overprint.

[60] Of particular interest in the mature granoblastic dikes is the coexistence, in close proximity, of “dry, ”granulite-facies two-pyroxene assemblage, with “wet, ”hornblende-bearing domains. Our observations described in section 5.3 implies that the “wet” assemblage represents a later hydrous metamorphic stage, at temperatures which were 100 to 200°C lower than the peak represented by the “dry” granulitic paragenesis. Our preferred scenario is that the peak of metamorphism was reached under relatively dry conditions, which is also indicated by the lack of true high-temperatures amphiboles as pargasites/hastingsites which are stable up to 1000°C in MORB-type systems at water-saturated conditions [e.g., Koepke et al., 2004; Berndt et al., 2005]. This stage was followed by a retrograde reaction at 100 to 200°C lower, where a water-rich fluid phase invaded the dry system. Since large parts of the “dry” domains survived, it is implied that the invasion of the fluid phase was not pervasive. Eventually, this is due to only limited amounts of water. Possible other reasons could be: focused, channeled fluid flow, or strong variations in the permeability of the dike rock.

[61] Since the base of the granoblastic dikes is obviously not far away from the AMC, it seems possible that primary magmatic processes influenced the formation of granoblastic dikes. This is indicated by increasing F contents in amphiboles with depths, since F in amphibole is regarded as a typical primary magmatic feature [e.g., Coogan et al., 2001]. Since the observed amphiboles show a similar composition like those from trondhjemitic veins at Hess Deep, EPR [Natland and Dick, 1996] or ODP hole 735B, SWIR [Dick et al., 2002], it seems also possible that the “wet” granoblastic domains were influenced by percolating trondhjemitic melts. However, in that case the bulk rock compositions of the granoblastic dikes should show a trend to more andesitic or dacitic compositions, which is not the case [Teagle et al., 2006]. Moreover, we never found in the mature granoblastic dikes hornblende in coexistence with quartz, which would be expected if their formation was by crystallization from trondhjemitic melts.

[62] Another scenario for the formation of the granoblastic dikes would be that the dikes intruded directly into the very hot base of the sheeted dike complex where the fresh dikes were directly converted into granulite-facies 2-pyroxene-bearing hornfelses. Under such conditions, the initial cooling occurred under dry conditions. Later, after significant cooling to a level where high-temperature veining was possible (e.g., 700–800°C [Manning and MacLeod, 1996; Manning et al., 2000]) hydrothermal fluids had access to this zone producing the amphibole-bearing parageneses of the “wet” granoblastic matrix. However, this model does not fit with our key observation listed above (e.g., the general presence of relics of quench textures in the granoblastic dikes; the evolution of magnesiohornblende from actinolitic precursor in a greenschist-facies matrix).

5.5. Thermal Modeling

[63] Further detailed studies of diffusion profiles in mineral pairs used for thermometry will be required in order to better constrain the thermal history of the base of the sheeted dikes preserved in Hole 1256D. However, a few simple considerations may be used to guide future studies. The first relates to the relationship between the thermometric results from the granoblastic dikes and the presence of the narrow gabbro intrusions. Could cooling and crystallization in these intrusions have caused the heating recorded in the dikes? A simple thermal balance calculation for melt at 1200°C and solid basalt cooled to 350°C indicates that a laterally continuous sill of a given thickness can heat a similar thickness of roof-rock to a temperature of ∼1000°C. The physical properties of melt and solid for this calculation were taken from the compilation of Cannat et al. [2004]. However, transfer of heat from the cooling melt lens to the overlying rocks requires conduction or hydrothermal advection, both of which are likely to result in loss of heat to shallower levels in the crust. Simple one-dimensional conductive cooling models [Maclennan and Lovell, 2002] with a sill of 100 m thickness injected into a country rock initially at 350°C, do not produce a contact aureole of any notable thickness. Therefore the narrow (thin) intrusions penetrated by Hole 1256D could not have provided sufficient heat source to produce the observed granoblastic overprint. If, however, the initial country rock temperature were 700°C, temperatures of over 800°C could be developed in a zone ∼30 m above the intrusive boundary. This temperature distribution is similar to that defined by our amphibole-plagioclase thermometry, but apparently insufficient to account for the “dry,” two-pyroxene assemblages. Further constraints are therefore required on the temperature of the country rock prior to injection of the shallow sill. The assumption of a simple one-dimensional structure is also not necessarily supported by a single drill hole as it implicitly assumes that the gabbro intrusion is near-horizontal. If the hole fortuitously intersected a steep dike-gabbroic contact, a narrow thermal aureole could be observed over a significant depth interval.

[64] Taken together, these considerations strongly suggest that most of the heat responsible for the observed thermal evolution of the sheeted dikes originated from deeper in the ocean crust. If the crustal reheating occurred close to the axis, the heating duration of 103–104 years estimated from plagioclase diffusion profiles would correspond to the time of transit from the axis to the edge of a stable axial melt lens (typically <2 km at fast spreading ridges [Barth and Mutter, 1996]). If it is assumed that the temperature gradient preserved in the granoblastic dikes is representative of the thermal structure above the melt lens, then this gradient may be used to estimate the axial heat flux. The thermometric results displayed in Figure 12 are consistent with a maximum thermal gradient of approximately 3°C m−1. Assuming a conductivity of 2 W m−1 K−1, the conductive flux is 6 W m−2, and a total heat flux for a 1–2 km-wide magma chamber is 6–12 MW km−1 of ridge axis is available to the overlying axial hydrothermal system. This model does not involve latent heat of crystallization in the melt and therefore provides a minimum predicted heat supply. The flux estimate is also low relative to predictions a pure sheeted-sills accretion model at full spreading rates of 120 mm a−1 [Maclennan et al., 2004]. It seems likely that at least one of the assumptions involved in the simple calculation is incorrect and further work is clearly required.

5.6. Granoblastic Dikes as Part of the Conductive Boundary Layer

[65] Axial melt lenses beneath fast spreading ridges are probably overlain by a conductive boundary layer (CBL) tens of meters thick [e.g., Wilcock and Delaney, 1996; Chenevez and Nicolas, 1997; Gillis and Roberts, 1999; Gillis and Coogan, 2002; Coogan et al., 2003]. The only other detailed petrographic studies of rocks from within such zones are from the Troodos ophiolite on Cyprus, where a decameter-thick, strongly metamorphosed horizon affected by hydrous anatexis at the base of the sheeted dikes has been identified as CBL [Gillis and Roberts, 1999; Gillis and Coogan, 2002]. Some samples from this zone are very similar to the granoblastic dikes of Hole 1256D. From this and from the lithostratigraphic position of the granoblastic dike directly above the plutonic rocks, we conclude that the granoblastic metamorphism occurred within a CBL.

[66] On the basis of the results presented here, and the concept that axial melt lenses under fast spreading mid-ocean ridges may move up and down [Hooft et al., 1997; Lagabrielle and Cormier, 1999; Garel et al., 2002; Gillis, 2002; Gillis and Coogan, 2002; Karson et al., 2002; Coogan et al., 2003], we suggest the following two-stage model for a dynamic dike-gabbro transition (Figure 14):

Figure 14.

Cartoon depicting the dike-gabbro transition at IODP Site 1256D during different stages of evolution. (a) During an initial stage (t = t0), the axial magma chamber (AMC) which resides at the base of the sheeted dike complex (SD) is located in a deep position. A several decameter thick conductive boundary layer (CBL) overlies the AMC. Above the CBL, the dikes are hydrothermally altered (hy) by circulating seawater-derived fluids. (b) During a subsequent stage (t = t1), after an upward moving of the AMC (following replenishment, for example), a new CBL is formed, integrating the previously hydrothermally altered dikes, which are then thermally metamorphosed into the granoblastic dikes (CBL/GD). The strong thermal gradient in this horizon induces a decrease in the intensity of the metamorphic conditions away from the AMC (marked by the density of points in the stippled zone). In the lower part of the CBL/GD, the conditions for partial melting of altered dikes are reached, producing trondhjemitic veins (tv). The roof of the AMC is assimilated by stoping of granoblastic dikes, and fragments (fr) of granoblastic dikes are incorporated into the magma. (c) Cartoon illustrating the dike-gabbro transition at Site 1256D, as interpreted by the results of this paper. Included is the location of the drill core. The upper and the lower gabbro screens observed in the drill core are marked by gb1 and gb2, which are here regarded as dike-like intrusions derived from the AMC. Individual fragments of dikes (fr) were observed in gb2, but not in gb1. Here, characteristic ghost structures imply that granoblastic dikes were also assimilated, probably by incorporation of fragments of granoblastic dikes. See text for details. For clarity, the lowermost rock recovered from Hole 1256D, a basaltic dike that lacks granoblastic textures, which is interpreted to be a late dike that postdates the intrusion of the lower gabbro [Wilson et al., 2006], is not included in the cartoon. Figures 14a and 14b are not at scale; Figure 14c reproduces exactly the lithostratigraphy recorded in the drill core of hole 1256D as presented by Wilson et al. [2006].

[67] 1. During an initial stage (t = t0), the AMC is relatively deep. A several decameter thick CBL overlying the AMC separates two convecting systems: below, the melt lens filled with a basaltic magma, and above, the sheeted dike complex which is cooled by circulating seawater affecting the well-known hydrothermal alteration (Figure 14a).

[68] 2. During a subsequent stage (t = t1), the AMC migrates upward, perhaps as a consequence of replenishment, establishing a new steady state in a relative high position and a new CBL within the previously hydrothermally altered dikes. Within the new CBL, thermal metamorphism leads to formation of the granoblastic assemblages, with associated hydrous partial melting forming trondhjemitic veins within the deeper parts. At the roof of the AMC, assimilation proceeds mainly by stoping and fragments of granoblastic dikes are incorporated into the magma (Figure 14b). This scenario is very similar to that presented by Gillis and Coogan [2002] for the rooting zone of the sheeted dikes observed in the Troodos ophiolite.

[69] It is very likely that Hole 1256D is currently terminated above the base of the CBL, within a complex AMC roof zone, with massive gabbros representing the frozen AMC located meters to decameters deeper. Evidence for assimilation by stoping of granoblastic dikes is provided by the occurrence of partially resorbed, dike clasts within the lower gabbro [Teagle et al., 2006; Wilson et al., 2006]. Within these clasts high-grade granoblastic assemblages rich in orthopyroxene are very similar to those occurring in the very deepest granoblastic dikes, implying that the fragments correspond to stoped granoblastic dikes. The petrology and geochemistry of theses clasts is currently under investigation by S. Miyashita et al. (manuscript in preparation, 2008).

6. Conclusion

[70] The discovery of the granoblastic dikes in the drill core at Site 1256D enables the first direct insight into the root zone of the sheeted dikes above the AMC in fast spread ocean crust. In analogy to observations in the Troodos ophiolite on Cyprus [Gillis, 2002], we interpret the zone of granoblastic dikes as part of a dynamic CBL sandwiched between AMC and sheeted dikes. Downward moving of the AMC results in progressive hydration/alteration of the basal sheeted dykes by seawater circulation forming greenschist to lower amphibolite facies rocks. Upward moving of the AMC affects the basal dykes to undergo prograde dehydration reactions under the conditions of amphibolite to granulite facies conditions. As this high heat flux fades out, the magma-hydrothermal boundary deepens again resulting in transient cracking of the CBL which in turn enables the ingress of seawater-derived fluids provoking a “second” hydrothermal alteration.

[71] As thickness of the CBL we estimated ∼60 m from the distribution of granulite-facies two-pyroxene bearing domains in granoblastic dikes. As highest equilibrium temperature within the granoblastic dikes, we recorded 1045°C for a sample which was the structurally lowest in our collection. This temperature is ∼150°C lower than that expected in the narrow zone directly above the AMC where the convecting magma resides at a temperature of ∼1200°C. This observation, and the fact that the granoblastic dikes extend below the lower gabbro screen, implies that the original thickness of the CBL was even thicker than 60 m.


[72] We gratefully acknowledge the captain and shipboard crew of IODP Expedition 312 for their assistance in data collection at sea. We wish to thank the Scientific Party of IODP Expedition 312 for fruitful discussion during the cruise. This work benefited from fruitful discussions during field work in Oman with F. Boudier, A. Nicolas, B. Ildefonse, and L. France. We would also like to thank F. Boudier and J. Natland for helpful comments on the manuscript. Valuable editorial advice from V. Salters is acknowledged. This study used samples provided by the Integrated Ocean Drilling Program. Funding for this research was provided by grants from the Deutsche Forschungsgemeinschaft.