JOURNAL OF PETROLOGY VOLUME 50 NUMBER 3 PAGES 387^403 2009 doi:10.1093/petrology/egp004
Onset and Progression of Serpentinization and Magnetite Formation in Olivine-rich Troctolite from IODP Hole U1309D
JAMES S. BEARD1*,B.RONALDFROST2,PATRICIAFRYER3, ANDREW McCAIG4, ROGER SEARLE5, BENOIT ILDEFONSE6, PAVEL ZININ3 ANDSHIVK.SHARMA3
1VIRGINIA MUSEUM OF NATURAL HISTORY, 21 STARLING AVENUE, MARTINSVILLE, VA 24112, USA 2DEPARTMENT OF GEOLOGY AND GEOPHYSICS, UNIVERSITY OF WYOMING, LARAMIE, WY 82071, USA 3HAWAI’I INSTITUTE OF GEOPHYSICS AND PLANETOLOGY, UNIVERSITY OF HAWAI’I, HONOLULU, HI 96822, USA 4SCHOOL OF EARTH AND ENVIRONMENT, UNIVERSITY OF LEEDS, LEEDS LS2 9JT, UK 5DEPARTMENT OF GEOLOGICAL SCIENCES, UNIVERSITY OF DURHAM, DURHAM DH1 3LE, UK 6GE¤ OSCIENCES MONTPELLIER, CNRS, UNIVERSITE¤ MONTPELLIER 2, CC 60, 34095 MONTPELLIER CE¤ DEX 05, FRANCE
RECEIVED AUGUST 16, 2008; ACCEPTED JANUARY 14, 2009 ADVANCE ACCESS PUBLICATION FEBRUARY 19, 2009
Serpentinization of olivine-rich troctolite from core 227, Integrated silica to the system converted the Mg-component of the brucite to Ocean Drilling Program (IODP) Hole U1309D ranges from serpentine. Magnetite forms one or more distinct bands in the 510% to 490%. Two episodes of serpentinization are recognized. interior of the vein and is never in direct contact with relict olivine. The first, dominant in weakly serpentinized samples, is an approxi- A brucite^serpentine mixture, similar to that found in type 1 veins, mately isochemical (except for water) replacement of olivine (Fo84^ but with lizardite instead of antigorite, is commonly present at the 85) by a mixture of serpentine (antigorite, Mg-number 92) and bru- margins of type 2 veins (i.e. where they are in reaction contact with cite (amakinite-rich; Mg-number 65). The compositions of the relict olivine). We interpret type 2 veins as a steady-state system minerals in type 1 veins are a reflection of Fe^Mg exchange between where brucite continually forms at the olivine^vein contact and then olivine and the brucite þ serpentine formed during early serpentini- reacts out in the interior of the vein. This continual formation and 5 zation. The early serpentinite veins (type 1) are thin ( 0 05 mm), destruction of brucite imposes an exceptionally low aSiO2 on the irregular, and exploit pre-existing cracks in olivine. The presence system. Magnetite and olivine are never in contact in type 2 veins of antigorite suggests that early serpentinization occurred at (or anywhere else) because the olivine-out reaction yields ferroan T43008C. Type 1 veins reflect rock-dominated serpentinization, brucite and not magnetite.The desilication of serpentine in the type became isolated early in their history, and persist as relicts in all 2 veins is a reflection of the inherent instability of Fe-rich serpentine but the most altered samples. The main episode of serpentinization with respect to magnetite at low silica activity.Thus, the composition is manifested by through-going lizardite (average Mg-number 96)^ of serpentine in equilibrium with magnetite in serpentinites is a func- magnetite veins (type 2). Type 2 veins define an anastomosing tion of serpentine^magnetite and not serpentine^olivine equilbria. foliation, may be several millimeters in width and appear to exploit pre-existing, favorably oriented type 1 veins.Type2 veins reflect open- system serpentinization. Magnetite in these veins formed by KEY WORDS: serpentinite; magnetite; brucite; antigorite; lizardite; oxidation of the Fe in brucite and serpentine, whereas addition of ocean drilling;1309D
ß The Author 2009. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@ *Corresponding author. E-mail: [email protected] oxfordjournals.org JOURNAL OF PETROLOGY VOLUME 50 NUMBER 3 MARCH 2009
INTRODUCTION SETTING AND SAMPLES Some of the most extreme geochemical environments on IODP Hole U1309D was drilled during IODP Expeditions Earth are associated with the serpentinization of olivine. 304 and 305 (Blackman et al., 2006), and is located on the Serpentinization is associated with strongly reducing con- Atlantis Massif at 30810 12’N, 42 87 11’W on the western rift ditions that lead to the generation of H2, a fluid pH that flank of the Mid-Atlantic Ridge (Fig. 1a). The Hole pene- ranges from 3 (at high temperature) to 12.5 (at low T), trates 1415m of mostly (490%) gabbroic rocks including and among the lowest aSiO2 in terrestrial silicate systems olivine gabbro, gabbronorite, oxide gabbro, and troctolite, (Coleman, 1963; Barnes & O’Neil, 1969; Frost, 1985; along with lesser amounts of peridotite (a few tens of cen- Abrajano et al., 1988; Peretti et al., 1992; Muntener & timeters) and diabase. Recovery was 75%. The gabbroic Hermann, 1994; Charlou et al., 1998, 2002; Beard & rocks include some of the most pristine and primitive ever Hopkinson, 2000; Allen & Seyfried, 2003; Mottl et al., sampled in the ocean basins (Blackman et al., 2006; 2004; Palandri & Reed, 2004; Shervais et al., 2005; Frost & Ildefonse et al., 2006). Some troctolites are very rich in oli- Beard, 2007). As a consequence of this there are minerals vine, with modal olivine approaching 90%. In this study, (e.g. brucite, hydrogarnet) and mineral assemblages (e.g. we examine the serpentinization and associated alteration awaruite^andradite, which pairs a ferric iron garnet with of the olivine-rich troctolites from core U1309D 227R-3 a metallic iron alloy; e.g. Frost,1985) found in serpentinites (Fig. 1b). that occur almost nowhere else on Earth. Additionally, Cores U1309D 227R-2 and U1309D 227R-3 preserve a there is growing recognition that hydrothermal vents asso- zone of serpentinization in olivine-rich (80^90% olivine) ciated with serpentinizing rocks may have played a key troctolites where relatively unserpentinized (i.e.510% ser- role in the genesis of life in the early oceans (Sleep et al., pentine) troctolite progressively grades into almost com- 2004; Russell & Arndt, 2005). pletely serpentinized rock and grades out again to Most serpentinites contain magnetite. Frost & Beard relatively unserpentinized rock. We selected four samples (2007) argued that magnetite formation was a direct conse- from core U1309D 227R-3 that represent the sequence quence of the instability of ferroan serpentine at low aSiO2. from most to least altered (Fig. 1b). More recently, Evans (2008) has argued that magnetite forms because the Fe^Mg exchange between olivine and serpentine during serpentinization requires that the ser- PETROGRAPHY pentine be more magnesian than olivine. Magnetite is Overall important, not only for understanding serpentinite geo- chemistry (e.g. magnetite formation is largely responsible Four samples from a 70 cm length of core U1309D 227R-3 for the reducing conditions that characterize actively ser- (henceforth referred to as core 227R-3) were analyzed in 5 pentinizing systems), but also for understanding magneti- detail. Serpentinization ranges from 10% in the sample zation of serpentinite. In particular, Toft et al. (1990) have from 69^71cm to 30% in the sample from 41^43 cm to 4 suggested that the observed lag between the onset of ser- 70% in the sample from 14^16cm and to 90% in the pentinization (as indicated by a drop in density) and the sample from 2^4 cm (Figs 1b and 2a^d).The degree of ser- production of modal magnetite (as indicated by magnetic pentinization is inhomogeneous on the thin-section scale. susceptibility) is a consequence of a multi-stage serpentini- The estimates reflect an average value, but, particularly in the moderately serpentinized samples, there is grain-to- zation in which magnetite formation is delayed. Bach et al. grain variation in the degree of serpentinization. Overall, (2006) proposed a two-stage process involving early forma- we interpret the samples as representative of progressive tion of ferroan brucite and subsequent oxidation^silicifica- serpentinization of the olivine-rich troctolite. tion (during late fluid influx) of the brucite to magnetite plus serpentine that could account for the magnetic and density characteristics of serpentinized peridotites drilled Early intra-olivine veins (type 1) at Ocean Drilling Program (ODP) Site 1274. (early or incipient serpentinization) In this study we describe a sequence of serpentinized oli- These narrow (50 05 mm wide) and irregular veins form vine-rich (70^90% olivine) troctolites from Integrated a network within single olivine grains (Fig. 2a and b).They Ocean Drilling Program (IODP) Hole U1309D. Detailed are the commonest vein type in weakly serpentinized sam- examination of several samples from this core allows us to ples. Type 1 veins are brownish in color, usually free of constrain several aspects of the timing and progression of opaque phases (but see Fig. 2c), and rarely pass from one serpentinization. In particular, we demonstrate that mag- olivine grain to another nor into any surrounding phases netite in serpentinites forms via a two-stage process as a (Fig. 2a). The color and textural characteristics of these consequence of both oxidation^silication of brucite and veins are distinctive, and relict examples of type 1 veins oxidation^desilication of ferroan serpentine (e.g. Frost & have been identified in all but the most strongly serpenti- Beard, 2007). nized samples (Fig. 2d and e).
388 BEARD et al. SERPENTINIZATION IN THE TROCTOLITE
42 15'W 42 10'W 42 05'W 42 00'W 41 55'W (b)
-3000 -2000 cm alteration
(a) sample locations 30 15'N 0 x -3000 10 >50% -3000 x -2000 -4000
-4000 20 hole -20001309D ssif 30 10'N 30 20-50% 40 Atlantis-1000 ma x -2000 0 Lost City -4000 50 -200-3000 -4000 30 05'N -3000 MAR Unit 562, Dunitic troctolite 60 -4000 x <20% Atlantis 70 -4000 FZ 305-U1309D-227R-003 (top = 1094.93 mbsf) 80 -4000 30 00'N
Fig. 1. (a) Bathymetry of the Atlantis Massif region and the location of IODP Site 1309D. (b) Top 80 cm of core U1309D-227R-3W showing distribution of alteration (Ildefonse et al., 2006) and locations of samples (crosses) examined for this study. mbsf, meters below sea floor.
Late intergranular serpentine^magnetite equivalent to, but less well-developed than, the chlorite^ veins (type 2) (main stage serpentinization) tremolite reaction zones found in olivine gabbros through- In samples with410% serpentinization, most serpentine is out the U1309D core (Blackman et al., 2006; Frost et al., contained within intergranular, colorless serpentine veins 2008). Chlorite forms a continuous, thin ( 0 01 mm) r im (type 2 veins, Figs 2d^f). Type 2 veins range up to 1mm around plagioclase. The chlorite rims extend as veins (or more) thick. They always cut across type 1 veins along olivine grain boundaries and connect with reaction (Fig. 2d and e). Type 2 veins are commonly subparallel rims in nearby plagioclase grains. Tremolite forms fine- and may transect the width of a thin section or core, defin- grained needles that grow outward at high angles from ing a mesoscopic foliation (Fig. 2d and e). In most cases, the chlorite zone, forming acicular sprays in adjacent oli- relict type 1 veins are oriented at high angles to type 2 vine (Fig. 3a). veins (Fig. 2d and e). This suggests that type 2 veins In the more altered samples, plagioclase alteration is exploited a weakness provided by pre-existing type 1 closely associated with type 2 serpentine veins. Where veins. A single relict olivine (or olivine pseudomorph) type 2 veins intersect plagioclase, the feldspar is altered to may be traversed by several type 2 veins (Fig. 2d^f). The low-silica phases including prehnite and, in more altered parallelism of the type 2 veins yields an unusual mesh tex- samples, hydrogrossular (Fig. 2d and f). In the most ser- ture (O’Hanley,1996), with lensoidal fragments of relict oli- pentinized rocks, plagioclase may be completely pseudo- vine defining the mesh centers (Fig. 2e). Type 2 veins morphed by a complex mineral assemblage dominated by commonly traverse olivine^olivine grain boundaries with- chlorite and hydrogrossular (Figs 2f and 3b). out interruption. The mineralogy of the vein changes when plagioclase is traversed (Fig. 2e, see below). Unlike type 1 veins, type 2 veins are almost always associated with mag- RAMAN SPECTROSCOPY netite. Magnetite occurs as one or more bands located OF TYPE 1 AND 2 VEINS towards the center of the type 2 vein. These bands are To determine which serpentine minerals are present in the sometimes disposed symmetrically about the center of the core 227 samples, Raman spectra were collected from vein (Fig. 2f).Veins transitional between types 1 and 2 con- three samples ranging from weakly (sample 69^71) to tain a narrow, magnetite-bearing band in the center of an strongly (sample 14^16) serpentinized. We performed otherwise typical type 1 vein (Fig. 2c). micro-Raman analysis with a WiTec Alpha300R confocal Raman microscope using a green laser (532 nm), a 2 1 Progressive alteration of plagioclase mW excitation power, and a 0 5 s integration time. The In the least altered rocks, plagioclase alteration is limited microscope on the system was enhanced with a transmis- to thin plagioclase^olivine reaction zones. These zones are sion-light source to augment the reflected light (standard)
389 JOURNAL OF PETROLOGY VOLUME 50 NUMBER 3 MARCH 2009
(a) (b)
olivine
plag oliv
sserperp 1 0.1 mm 1 mm
(c) (d)
oliv
0.1 mm
serp2 1 mm plag
(e) (f) plag prehn serp2 hgr
oliv chl oliv serp2 serp1 serp2 serp2
1 mm 1 mm
Fig. 2. (a) Weakly serpentinized (510%) sample 227R-3, 69^71cm. Network of dark type 1 veins in olivine. The limited alteration of plagioclase and the overall lack of through-going veins and mesh development should be noted. (b) Close-up of type 1 (serpentine^brucite) vein, same sample. The absence of magnetite should be noted. (c) Same sample. Type 1 vein, transitional to type 2. Narrow central band of magnetite plus serpentine should be noted. (d) Sample 227R-3, 41^43 cm. Development of a magnetite-cored, subparallel, type 2 (serpentine^magnetite) vein set should be noted. Incipient mesh development, with elongate lenses of olivine in the mesh cores is evident. This sample, overall, is about 20^30% serpentinized. (e) Sample 227R-3, 14^16cm, 50^70% serpentinized. Type 2 veins predominate, but dark, relict type 1 veins occur in lensoidal olivine mesh cores. Alteration of plagioclase where intersected by type 2 veins should be noted. (f). Sample 227R-3, 2^4 cm, 490% serpentinized. Extensive areas of water-clear serpentine and linear magnetite bands occur, some of which, especially toward the bottom of the image, show symmetrical longitudinal structure. Plagioclase in this sample is extensively altered to a mixture of chlorite and hydrogrossular.
390 BEARD et al. SERPENTINIZATION IN THE TROCTOLITE
0.2 mm (a) (b)
chl
olivine amesite chl
serp1 amph plag
hgr chlorite 0.1 mm
amph chl
227r3w, 69-71 cm, Si map 227r3w, 2-4 cm, Al map
Fig. 3. (a) Si map of sample 227R-3, 69^71cm. Dark band surrounding plagioclase and extending along olivine grain boundaries to connect plagioclase grains is chlorite. Dark veinlets in olivine are type 1 veins. Bright needles extending into olivine are tremolite. (b) Al map of sample 227R-3, 2^4 cm. Gray-to-white area is a pseudomorph of chlorite and hydrogrossular after plagioclase. Late amesite vein is indicated at the top of the field of view.
(a) (b)
Fig. 4. Raman spectra from type 1 (a) and type 2 (b) veins. (a) Peaks at 530 and 1055 cm^1 occur only in antigorite. Other labeled peaks occur in all serpentine polytypes. (b) Except for the peak at 1072 cm^1, labeled peaks are typical of lizardite. Antigorite peaks, especially the 530 cm^1 peak, are missing. The 1072 cm^1 peak lies between the 1055 cm^1antigorite peak and the 1096 cm^1 lizardite peak, and may reflect the replace- ment of antigorite by lizardite. observation of samples, making it possible to distinguish Table 1). Raman data from samples 14^16cm and the brown type 1 veins containing iron-rich brucite from 41^43 cm indicate that the serpentine in the younger, type colorless to green type 2 serpentine veins. Signals from 2 veins is largely lizardite (Fig. 4b; Table 1). In some sam- the glass slides and epoxy mounting medium were ples, especially towards the margins of veins, the lizardite obtained beforehand as a blank analysis. We subtracted may be a replacement of antigorite (Fig. 4b). the interfering signal (and corrected for background aber- A weak brucite Raman signal is detectable in type 1 rations) using the Grams software package (Thermo veins, with an apparent concentration of brucite near the Fisher Scientific Inc., Waltham, MA). olivine^vein contact. Although brucite^serpentine inter- Spectra from type 1 veins in sample 69^71cm indicate growths are identified from chemical data at the margins that the serpentine in these veins is antigorite (Fig. 4a; of some type 2 veins (Table 2 and see below), the presence
391 JOURNAL OF PETROLOGY VOLUME 50 NUMBER 3 MARCH 2009
Table 1: Comparison of the Raman peak positions (cm^1) stoichiometric brucite (Fig. 5a). This relation means that the iron in the serpentine is mostly ferrous. If a substantial measured in type 1 and type 2 veins for this study and taken part of the iron were dioctohedral ferric iron, the regres- from the literature sion would not intersect the y-axis at brucite. The linear trend in Fig. 5b also indicates that the iron is not present 3þ 3þ Type 1 Type 2 Ant Anty Lzd Lzdy Chr Chrz as a Mg-cronstedtite [Mg2Fe (Fe Si)O5(OH)4] or cron- 2þ 3þ 3þ (edge) stedtite [Fe 2Fe (Fe Si)O5(OH)4] molecule. When the Si^Mg line is extrapolated to end-members, it appears that the serpentine end-member has Mg-number 92 whereas 228 230 230 235 233 238 231 231 the Mg-number of the brucite end-member is 64^65 350 350 345 (Fig. 5b). Type 1 veins are Al-poor (50 2 wt %) compared 381 383 375 377 388 393 389 388 with type 2 vein serpentines (Table 2). 431 432 The stoichiometry of the average of type 1 vein analyses 470 466 from the least serpentinized sample is very similar to that 530 520 510 of the average olivine from this sample (Fig. 5c). The 618 635 630 620 607 slightly more silicic character of the veins may reflect the 683 690 683 685 690 695 692 692 concentration of brucite near the olivine^vein contact (see 1055 1072 1044 1096 1105 1102 Raman data). Given that the veins are so narrow, microbeam analyses tend to be of the central areas of Ant, antigorite; Lzd, lizardite, Chr, chrysotile. Data source Rinaudo et al. (2003). veins, and hence brucite-rich areas may be under- yData source Auzende et al. (2004). represented. zData source Kloprogge et al. (1999). The approximate isochemical serpentinization reaction is
2 olivine þ 2H2O ¼ serpentine þ brucite: ð1Þ of brucite in type 2 veins could not be confirmed from the When this reaction is mass balanced using serpentine with Raman spectra. The overall poor quality of the brucite Mg-number 92 to reconstruct the average (Fo85) olivine signal in the U1309D samples may be caused by the pres- composition, the calculated brucite composition (Mg- ence of a significant iron (amakinite) component. number 64) is near that predicted by the extrapolation in Fig. 5b (Table 3; Fig. 5c). MINERALOGY Relict type 1 veins in the more strongly serpentinized samples have chemical and stoichiometric characteristics Methods like those seen in type 1 veins from the least serpentinized Minerals were analyzed by electron microprobe in the sample (Fig. 6; Table 2). Department of Geological Sciences at Virginia Tech. Analyses were carried out using a focused 3 mm beam at 15 kV and 20 nA, and standardized to natural oxide and Type 2 serpentine veins silicate standards. Most serpentine in type 2 veins is nearly stoichiometric (Fig. 6a). Mg-number averages 96 (Table 2). Al2O3 is sub- Primary mineralogy stantially higher than in the type 1 veins, averaging nearly The original mineralogy consists of 75^90% olivine (Fo84^85), 1wt % and rarely ranging as high as 3 6wt%.Alinthis 10^25% plagioclase (An77^82), 1^5% clinopyroxene serpentine appears to reflect a Tschermak’s substitution 2þ (En46^49Fs6^8Wo 43^48)and 1% Cr^Al^Mg^Fe spinel. towards amesite (Fig. 6b). Olivine and plagioclase are progressively altered during Unlike the material in type 1 veins, typical type 2 ser- serpentinization. Spinel is unaltered. Clinopyroxene is frac- pentine does not lie along a brucite^serpentine mixing tured where intersected by alteration veins, but chemically line. Instead, most of the variation in þ2 cations appears unaltered. Primary sulfides include pyrrhotite and pentlan- attributable to the Tschermak’s trend combined with an Fe- dite. Most primary pentlandite grains contain some oxide mixture, or perhaps more probably, a cronstedtite (2^5%) Cu, some contain substantial Co (up to 1 6%). (ferri-Tschermak’s) substitution (Fig. 6a^c; Evans, 2008). Also unlike type 1 veins, most of the variation in Mg con- Type 1 serpentine veins tent can be accommodated by serpentine solid solutions Microprobe analyses of type 1 serpentine veins indicate (e.g. with greenalite and/or Mg-cronstedtite; Fig. 6c). that the veins are submicron-scale mixtures consisting of Analyses from the margins of type 2 veins proximal to serpentine and Fe-rich brucite. The nature of this mixture relict olivine, however, have chemical characteristics simi- is evident from Fig. 5. Microprobe analyses lie along a lar to type 1 veins. In particular, they plot near or along straight line connecting stoichiometric serpentine and the brucite^serpentine mixing line (Fig. 6d).
392 BEARD et al. SERPENTINIZATION IN THE TROCTOLITE
Table 2: Mineral chemistry, 1309d, core 227 r3W
Olivine Serpeniteþbrucite Serpenite Prehnite
SD Type 1 SD Type 1 SD Type 2 SD Type 2 SD average SD 1s 69–71 cm 1s other 1s vein edge 1s average 1s 1s n: 13 29 10 9 73 8
SiO2 39 38 0 25 34 15 2 12 35 10 1 40 37 01 1 68 40 80 0 96 43 39 0 36
TiO2 0 00
Al2O3 0 07 0 04 0 07 0 09 0 37 0 25 0 96 0 65 24 83 0 48
Fe2O3 0 50 0 38
Cr2O3 MgO 45 39 0 25 38 39 1 11 39 54 0 86 40 10 0 93 40 14 0 76 0 03 0 02 MnO 0 24 0 04 0 14 0 05 0 08 0 04 0 06 0 03 0 05 0 03 0 09 0 05 FeO 14 64 0 26 10 88 1 83 9 44 1 64 6 92 1 45 3 12 0 88 CaO 26 50 0 56
Na2O 0 08 0 04
K2O 0 00 0 00 Cl 0 34 0 14 0 26 0 10 0 27 0 09 0 23 0 14 0 78
‘H2O’ Total 99 67 0 52 83 97 1 40 84 48 0 50 84 73 1 13 85 30 0 54 95 42 0 78 7 oxygen 7 oxygen 7 oxygen 7 oxygen 11 oxygen Si 0 992 0 003 1 772 0 070 1 791 0 042 1 847 0 063 1 959 0 028 2 987 0 013 Ti 0 000 Al 0 004 0 002 0 004 0 006 0 022 0 014 0 055 0 037 2 011 0 030 Fe3þ 0 026 0 020 Cr Mg 1 704 0 009 2 970 0 085 3 007 0 031 2 983 0 082 2 873 0 045 0 003 0 002 Mn 0 005 0 001 0 006 0 002 0 003 0 002 0 003 0 001 0 002 0 001 0 005 0 003 Fe2þ 0 308 0 005 0 472 0 089 0 403 0 078 0 289 0 063 0 125 0 037 Ca 1 954 0 041 Na 0 010 0 006 K 0 000 0 000 H/4 Sum 3 008 0 003 5 225 0 070 5 207 0 040 5 142 0 065 5 014 0 023 6 997 0 010 Mg-no. 84 70 386 32 288 22 091 21 695 81 2
(continued)
Cl in serpentine veins relict grains have little obvious relationship to any overall The distribution of type 1 veins in the least serpentinized vein structure (Fig. 7c). Serpentine analyses from the samples is faithfully reproduced by a map of Cl distribu- Cl-rich coronas and from the Cl-rich margins of type 2 tion (Fig. 7a). In the more serpentinized samples, Cl veins are similar to those from the ‘serpentine’ in type 1 behavior is more complex. Cl-rich late veins and Cl-rich veins. In particular, they plot along the brucite^serpentine chlorite occur in many of the strongly serpentinized sam- mixing line (Fig. 6d). ples. Some type 2 veins are complexly zoned in Cl. However, the typical zoning pattern has the margins of Prehnite the type 2 veins (i.e. the part of the vein next to olivine) In the least serpentinized sample (69^71cm), prehnite enriched in Cl, whereas the center of the vein is slightly (Table 2) occurs sparsely as thin blades along cleavage depleted in Cl (Fig.7b). In strongly serpentinized samples, planes in fractured plagioclase. It is common in moder- relict olivine grains may have a Cl-rich corona, even if the ately serpentinized rocks (41^43 cm and 14^16cm) where
393 JOURNAL OF PETROLOGY VOLUME 50 NUMBER 3 MARCH 2009
Table 2: Continued
Hydrogarnet Chlorite Chlorite, plagioclase replacement Amphibole
grossular SD andradite early rims SD other SD amesite Fe-rich other SD 1s 69–71 cm 1s 1s 1s n:16282133232
SiO2 33 40 2 36 34 47 35 00 1 45 33 06 1 41 21 16 26 07 27 01 2 57 57 98
TiO2 0 00 0 04
Al2O3 21 31 1 66 5 28 15 53 1 95 16 23 1 92 27 45 17 17 20 17 4 27 0 60
Fe2O3 4 52 3 18 23 85
Cr2O3 0 01 MgO 0 38 0 60 0 05 29 54 1 58 29 61 1 37 6 01 4 21 15 94 6 17 22 51 MnO 0 78 0 36 0 57 0 08 0 03 0 15 0 07 3 62 3 64 1 77 1 16 0 09 FeO 6 92 0 60 7 28 0 80 30 52 36 86 21 97 6 57 3 22 CaO 35 86 2 77 33 50 0 22 0 20 0 15 0 06 0 11 0 74 0 44 0 29 13 49
Na2O 0 28
K2O 0 01 Cl 0 04 0 02 0 08 0 08 0 05 0 08 0 16 0 23
‘H2O’ 3 75 1 99 2 28 Total 96 25 1 99 97 72 87 34 0 29 86 56 1 10 88 92 88 77 87 45 1 90 98 20 12 oxygen 12 oxygen 7 oxygen 7 oxygen 7 oxygen 7 oxygen 7 oxygen 23 oxygen Si 2 578 0 171 2 843 1 660 0 057 1 593 0 056 1 152 1 478 1 413 0 123 7 932 Ti 0 000 0 004 Al 1 948 0 143 0 515 0 873 0 113 0 926 0 111 1 771 1 153 1 250 0 264 0 097 Fe3þ 0 262 0 186 1 480 Cr 0 001 Mg 0 044 0 068 0 006 2 089 0 103 2 126 0 089 0 488 0 356 1 243 0 457 4 589 Mn 0 051 0 023 0 039 0 003 0 001 0 006 0 003 0 167 0 175 0 078 0 055 0 010 Fe2þ 0 275 0 025 0 293 0 034 1 390 1 748 0 961 0 318 0 368 Ca 0 233 0 011 0 010 0 008 0 003 0 007 0 045 0 025 0 017 1 977 Na 0 073 K 0 001 H/4 0 241 0 130 0 159 Sum 8 090 0 065 8 004 4 910 0 023 4 951 0 016 4 975 4 954 4 971 0 037 15 052 Mg-no. 88 41 387 91 626 016 956 417 492 6
H2O for hydrogarnet estimated from analytical sum. it typically occurs as a dusty alteration of fractured plagio- the two phases are generally separated by prehnite. clase where plagioclase grains are intersected by type 2 Hydrogen substitutes for between 5 and 30% of the sil- serpentine veins. In the most strongly serpentinized ica in the hydrogarnets (Fig. 8), which is fairly typical sample, dusty grayish and brownish masses of prehnite for serpentinites (Onuki et al., 1981; Beard & Hopkinson, replace plagioclase and separate relict plagioclase from 2000). Hydrogen in hydrogarnet is estimated from hydrogarnet. difference and subject to large error. Most analyses lie below the substitution curve in Fig. 8, suggesting that Hydrogarnet the microprobe totals for the hydrogarnets might Hydrogrossular (73^99% Gr) and hydroandradite (rare; be high. The presence of both hydrogrossular and hydro- 70^80% And) (Table 2) have been identified as replace- andradite in these samples is somewhat unusual ments of plagioclase in strongly serpentinized samples. and probably represents local and variable In samples containing both hydrogarnet and plagioclase, microenvironments.
394 BEARD et al. SERPENTINIZATION IN THE TROCTOLITE
brucite/ferrobrucite 7.0 Table 3: Mass balance (weight basis) for reaction (1) (a) 6.5 type 1 serpentine veins in olivine Calculated: 6.0 1309d, core 227r3w, 69-71cm alteration <10% 100olivine (Fo85) ¼ 82serpentine(Mg-no. 92) þ 18brucite(Mg-no. 64), sum 5.5 r2 ¼ 0 05 5.0 From stoichiometry: 100olivine (Fo85) ¼ 83serpentine (Mg-no. 92) þ 17brucite(Mg-no. 64) 4.5
4.0
3.5 (Table 2, Fig. 9a and b). These rims persist and thicken (Mg+Fe), cations/7oxygens throughout serpentinization. In the most serpentinized 3.0 0.0 0.5 1.0 1.5 antigorite samples, chlorite is part of the mineral assemblage that 2.0 replaces plagioclase. This chlorite is approximately clino- brucite chlore, with some analyses having elevated Al and trend- type 1 serpentine veins in olivine (b) ing toward amesite (Fig. 9a). Most notably, these chlorites 6 1309d, core 227r3w, 69-71cm alteration <10% are all substantially richer in Fe (and Mn) than the chlor- brucite, Mg# = 65 ite rims (Fig. 9c and d). A few analyses have Mg-numbers as low as 10 (Fig. 9c).The latest-formed chlorite approaches 4 Mg-antigorite Fe-amesite in composition (Fig. 9a). It occurs in a vein cut- ting across a plagioclase pseudomorph (Fig. 3b). Mg-cronstedtite 2 antigorite, Mg# 92 Tremolite Mg, cations/7oxygens The amphibole associated with the early formed chlorite cronstedtite rims is close to end-member tremolite (Table 2). In particu- 0 0.0 0.5 1.0 1.5 2.0 lar, it should be noted that is not a significant sink for Na derived from reactant plagioclase.
Oxides 3.8 (c) End-member magnetite is the only oxide formed during serpentinization. It occurs in type 2 serpentine veins and 3.6 is most abundant in strongly serpentinized samples. Magnetite also occurs as a replacement of primary sulfide 3.4 minerals. Sulfides and metals 3.2 average olivine Wairauite (FeCo) was found in samples 227R-3, 41^43 cm average type 1 "serpentine" (Mg+Fe), cations/7oxygens and 227R-3, 14^16 cm. Awaruite (Ni2^3Fe) was found only 3.0 in 227R-3,14^16 cm. These metals occur in association with 1.5 1.6 1.7 1.8 1.9 2.0 magnetite in type 2 serpentinite veins. They are never Si, cations/7oxygens found in type 1 veins. Native copper has been tentatively Fig. 5. Chemistry of type 1 veins. All data from 227R-3, 69^71cm. (a) identified in all samples. No other metals were found in Atomic Si vs Mg þ Fe. Analyses lie along a serpentine^brucite mixing 227R-3, 69^71cm or 227R-3, 2^4 cm. Unusual, unidentified line. (b) Mg vs Si. Projection of this line to end-members suggests a relatively Fe-rich brucite coexisting with a relatively Mg-rich serpen- low-sulfur sulfides were found in association with magne- tine. It should be noted that the variation cannot be explained by a tite in serpentine veins in sample 227R-3, 2^4 cm. These, cronstedtite substitution. (c) Close-up view of (a), showing the simi- like the metals, are extremely fine-grained and difficult to larity between the average olivine composition and the average com- analyze. They may be mixtures or intergrowths. Primary position of type 1 serpentine. sulfides are altered to valeriite (rare) and magnetite (common). Chlorite Chlorite in the core 227R-3 samples is always associated DISCUSSION with plagioclase. In the least altered samples, it occurs as Type 1 (antigorite^brucite) veins formed in a largely iso- thin rims separating olivine and plagioclase. This chlorite lated, rock-dominated system. Type 2 (magnesian lizar- is a magnesian (Mg-number 485) and silicic clinochlore dite^magnetite) veins reflect continuing serpentinization
395 JOURNAL OF PETROLOGY VOLUME 50 NUMBER 3 MARCH 2009
4.0 0.4 Serpentine, type 2 veins brucite trend amesite tr Serpentine, type 2 veins (a) all samples (b) all samples 0.3 end 3.5
0.2
3.0 cronstedtite/ 0.1
amesite trend Al, cations/7 oxygens (hydr)oxide trend (Mg+Fe), cations/7 oxygens
2.5 0.0 1.5 1.6 1.7 1.8 1.9 2.0 1.8 1.9 2.0
4.0 Mg-serpentine 3.0 (c) Serpentine, type 2 veins brucite trend all samples 3.8 (d) 2.8 3.6 2.6
3.4 2.4 relict type 1 veins
Mg, cations/7 oxygens Coronas around olivine 2.2 3.2 cronstedtite trend and vein margins (Mg+Fe), cations/7 oxygens Mg-cronstedtite 2.0 3.0 1.0 1.2 1.4 1.6 1.8 2.0 1.5 1.6 1.7 1.8 1.9 2.0 Si, cations/7 oxygens Si, cations/7 oxygens
Fig. 6. Chemistry of serpentine in type 2 veins. 2a^c include analyses of all type 2 veins from all samples, except for the margins and coronas plotted in 2d. (a) Atomic Si vs Mg þ Fe. (Compare especially with Fig. 5a.) Type 2 serpentine is close to stoichiometric composition. Most variation lies along a trend of decreasing R2þ cations, reflecting aTschermak’s substitution (amesite trend), combined with an oxide mixing or 2þ 3þ 3þ cronstedtite [Fe 2Fe (Fe Si)O5(OH)4] trend. (b) Al vs Si. This again shows that the amesite trend accounts for much of the variation seen in type 2 serpentines. (c) Mg vs Si. Unlike type 1 veins, most variation in type 2 veins occurs within the Mg range of serpentine and cronstedtite solid solutions. (d) Composition of relict type 1 veins and margins of type 2 veins. The relict type 1 veins have retained their chemical character- istics, even in strongly serpentinized rocks. Additionally, the margins of type 2 veins (i.e. the parts of the vein in contact with relict olivine) have a chemistry like type 1 veins. Specifically, they appear to represent, in large part, mixtures of brucite and serpentine.
in a fluid-dominated system. Type 2 veins reflect the The brucite^serpentine silica buffer lies at three log units steady-state production of brucite via olivine ¼ brucite þ below quartz at 2008C, among the lowest aSiO2 seen in any serpentine at the margins of the vein (i.e. at the olivine^ terrestrial silicate system. For comparison, the desilication fluid contact) coupled with the destruction of brucite in of plagioclase to grossular occurs less than one log unit the vein interior. Throughout, the presence of Fe-rich bru- below quartz (Frost & Beard, 2007). The Fe-brucite^mag- cite buffered silica and oxygen at the low values typical of netite oxygen buffer has not been experimentally docu- serpentinization via the reactions mented, but, when calculated, lies at or below iron^ magnetite, although it is apparently metastable with 6bruciteþ 4SiO ¼ 2serpentine þ H O ðfor silicaÞð2Þ 2 2 respect to that assemblage (Bethke, 2005). and Early serpentinization, type 1 veins 3FeðOHÞ ðFe bruciteÞþ1=2O2 ¼ magnetite þ 3H2O 2 Type 1 antigorite^brucite veins are, except for water and ð Þ 3 minor Cl, approximately isochemical replacements of or (equivalent) their host olivine. They represent what may be the extreme case of serpentinization in a rock-dominated system. ð Þ ð Þ¼ þ þ : 3FeOH 2 Fe brucite magnetite 2H2O H2 Several inferences can be drawn from the petrography, ð4Þ mineralogy, and chemistry of the type 1 veins. First, they
396 BEARD et al. SERPENTINIZATION IN THE TROCTOLITE
4 (a) >70% andradite oliv oliv >70% grossular tion) 3
garnet substitu 2 oliv
H /12 oxygen H=4(3-Si) (hydro 1 plag 0.2 mm
0 oliv 0.0 0.2 0.4 0.6 0.8 1.0 oliv 3-(Si /12 oxygen)
Fig. 8. H substitution in (hydro)garnet. 227r3w, 69-71cm, <10% serpentinization
were cut off from any significant external source of silica (b) or, indeed, any other component, except water. Second, oliv access of water to the system was limited. To a first order, this is reflected by the limited size of the veins themselves. The lack of material access to the system explains why type oliv 1 veins are so narrow (usually 0 005^0 02 mm; e.g. Fig. 2b)çthe material for vein growth was simply not oliv available beyond a certain point in their growth history. The high Cl content of serpentine in the type 1 veins oliv (Table 2) can be interpreted as reflecting crystallization in 0.2 mm the presence of an increasingly concentrated brine. Relict type 1 veins maintain their chemical integrity in strongly serpentinized rocks (Fig. 6d), suggesting that the chemical isolation is robust and can be long-lived. Finally, the pres- ence of antigorite requires that the early serpentinization 227r3w, 41-43cm, ~30% serpentinization occurred at relatively high temperature, at least 3008C and, conceivably, up to 5008C (Evans, 2004). (c) Evans (2008) has proposed a major role for Fe^Mg exchange in producing the mineral compositions observed during serpentinization, specifically, the composition of oliv serpentine in equilibrium with magnetite. We agree that this exchange is important during serpentinization, but propose, instead, that the manifestation of Fe^Mg exchange during serpentinization is the relatively Mg-rich serpentine coexisting with relatively Fe-rich brucite observed in type 1 veins. These observations are consistent with a system that, once it formed, was quickly isolated. Our preferred inter- oliv oliv pretation of type 1 veins is that they formed as water pene- 0.3 mm trated pre-existing (contraction?) cracks in olivine. Once hydration and consequent expansion began, the cracks 227r3w, 14-16 cm, ~70% serpentinization were quickly sealed. They remained sealed both during the later stages of hydration and long after hydration Fig. 7. Cl maps of three increasingly serpentinized samples. (a) 227R-3, ceased. In strongly serpentinized rocks, relict type 1 veins 69^71cm. The Cl map faithfully outlines the distribution of type 1 brucite^serpentine veins in this sample. (b) 227R-3, 41^43 cm. The are always at high angles to type 2 veins. This suggests that zoning in Cl seen in the large (type 2) veins, with high Cl concen- type 2 veins, which have a strong preferred orientation, trated at the margins of the veins in contact with olivine, should be noted. (c) 227R-3, 14^16cm. High-Cl coronas surrounding relict olivine in this strongly serpentinized sample. Most high Cl areas in (b) and (c) are brucite^serpentine mixtures resembling type 1 veins.
397 JOURNAL OF PETROLOGY VOLUME 50 NUMBER 3 MARCH 2009
amesite 2 3.0 latest amesite (b) late plag replacement (a) early chlorite-trem rims 2.5
1 clinochlore
Al / 7 oxygen 2.0 Mg+Fe+Mn / 7 oxygen
serpentine 0 1.5 1.0 1.2 1.4 1.6 1.8 2.0 1.0 1.2 1.4 1.6 1.8 2.0 Si / 7 oxygen Si / 7 oxygen
100 0.3 (c) (d) 80
0.2 60
40 0.1 Mn / 7 oxygen 100*Mg/(Mg+Fe)
20
0 0.0 1.0 1.2 1.4 1.6 1.8 2.0 012 Si / 7 oxygen Fe / 7 oxygen
Fig. 9. Chlorite chemistry. (a) Si vs Al. Compositions trend along a line connecting serpentine and amesite. The lowest Al chlorites occur in the early chlorite^tremolite zones. The most aluminous chlorites are the latest to form and occur only in the most extensively altered sample (227R-3, 2^4 cm). (b) R2þ cations vs Si, also showing the serpentine^amesite trend. (c) Mg-number vs Si. Early chlorite (e.g. Fig. 3a) is magnesian. Later chlorite, specifically that which pseudomorphs plagioclase, is Fe-rich, sometimes extremely so. (d) Mn vs Fe. These elements are strongly correlated. MnO contents of43% occur in some Fe-rich chlorites. developed by exploiting and expanding favorably oriented lizardite (Tables 1 and 2; Figs 6d and 7b, c); (3) magnetite type 1 veins. in type 2 veins tends to be concentrated in vein centers and is never in contact with olivine (Fig. 10a and b); (4) away Main stage serpentinization (type 2 veins) from olivine contacts, serpentine in type 2 veins is magne- and the formation of magnetite sian (Mg-number 96, on average), approximately stoichio- The structure and composition of type 2 serpentine veins metric, and free of brucite (Fig. 6a and c); (5) iron^nickel suggest that their formation can be modeled as a two- and iron^cobalt alloys occur exclusively in association with stage process. Although two principal reactions (olivine ¼ magnetite in type 2 veins, suggesting that the most redu- brucite þ serpentine and brucite ¼ serpentine þ magne- cing conditions were associated with magnetite formation. tite) can be recognized, it is important to note that the Implicit in this model is that the composition of magnesian reactions are going on simultaneously, with olivine react- serpentine in type 2 veins reflects magnetite^serpentine ing out at the vein edge and earlier-formed brucite reacting equilibrium and not the olivine^serpentine Fe^Mg out within the vein. Key observations are: (1) serpentine in exchange suggested by Evans (2008). type 2 veins is lizardite, suggesting that temperatures were The paragenesis of type 2 veins can be modeled as a lower than those experienced during type 1 vein formation; series of reactions beginning with the type 1 vein assem- (2) the margins and other parts of the type 2 veins in con- blage and ending with the type 2 assemblage. In Table 4, tact with relict olivine are brucite^serpentine mixtures the reaction and recrystallization of the type 1 assemblage similar to type 1 veins, except that the serpentine is is represented by the brucite-out reaction [Table 4,
398 BEARD et al. SERPENTINIZATION IN THE TROCTOLITE
Table 4: Formation of magnetite and serpentine in type 2 veins
Start: type 1 veins Bru64 Serp92 Serp100 Serp93 5 Serp96 SiO2(aq) Mt Notes 0 18 0 82
Reaction (1), brucite-out
0 18bru64 þ 0 09SiO2(aq) ¼ (0 18) 0 18 (0 09) 0 09 requires external
0 18serp100 þ 0 09mt silica Reaction (2), new serpentine forms
0 18serp100 þ 0 82serp92 ¼ (0 82) (0 18) 1 00
1 00serp93 5 Reaction (3), serpentine (Mg-no. 96) forms
1 00serp93 5 ¼ (1 00) 0 96 0 013 0 024 driven by low aSiO2
0 96serp96 þ 0 24mt þ 0 013SiO2(aq) Net reaction:
0 18bru64 þ 0 82serp92 þ 0 076SiO2 ¼ 0 96 (0 08) 0 11
0 96sSerp96 þ 0 11mt End: type 2 veins 0 90 0 10
Reactions (1) and (2) taken together represent the net result of brucite reacting out in this system. Reaction (3) reflects the adjustment of serpentine composition required to match observation in the core 227 rocks. The net reaction and reactions (1) and (3) are open system reactions that require net addition of silica, hence they do not sum to 1 00.
reaction (1)] and the consequent adjustment of the resul- & Beard, 2007). A consequence of magnetite formation is tant serpentine composition [Table 4, reaction (2)]. the imposition of exceptionally reducing conditions on the Brucite-out requires the oxidation of the Fe component of system. This is exemplified by the presence of iron alloys, the brucite to magnetite and the silicification of the Mg which occur exclusively in association with magnetite in component to serpentine. The overall serpentine composi- type 2 veins. This contrasts with type 1 veins, in particular, tion must then be adjusted to accommodate the Mg from which show no evidence of either externally derived silica brucite-out. These two reactions require the addition of or of particularly reducing conditions. silica. The immediate source of the extra silica required to In the core 227R-3 samples, magnetite is nowhere in convert brucite to serpentine is the fluid. Thus, by implica- contact with relict olivine (Fig. 10a and b), a common tion, the conversion of brucite to serpentine þ magnetite observation in serpentinized peridotites (Viti & Mellini, requires a high fluid flux. The ultimate source of the silica 1998; Oufi et al., 2002; Rumori et al., 2004; Bach et al., is a matter of speculation. One candidate is silica derived 2006). This is consistent with the interpretation that mag- from plagioclase-out reactions. netite is not forming directly from olivine, but rather from A problem arises in that, even assuming that all Fe in the early formed, brucite^serpentine assemblage. brucite goes to magnetite and all Mg into serpentine, the Furthermore, magnetite is not dispersed evenly throughout serpentine that is a product of the brucite-out reaction is the type 2 veins, but concentrated in one or more discrete less magnesian (Mg-number 93 5) than the average ser- bands that are either central to the vein or disposed sym- pentine in type 2 veins (Mg-number 96) (e.g. Tables 2 and metrically about the center of the vein (Fig. 10c). This dis- 4).Toremedy this, a desilication reaction of the type tribution of magnetite in vein cores is seen even in the smallest magnetite-bearing veins (Fig. 2c). The persistence Fe-serpentine ¼ magnetite þ 2SiO2 þ H2O þ H2 ð5Þ of a central zone of magnetite in larger veins suggests that must occur. It is important to note that the assemblage magnetite precipitation may be concentrated in vein cen- brucite þ serpentine is continuously forming from olivine ters by favorable nucleation on extant magnetite crystals. at the edges of type 2 veins throughout their growth his- Multiple, symmetric bands of magnetite in some veins tory. The desilication of Fe-serpentine is consistent with may form in response to multiple cycles of reaction or the silica activities imposed by the brucite^serpentine when the energetics of transport over distance exceed assemblage. It reflects the inherent instability of ferrous those of favorable nucleation on existing, but distant, sites. silicates in general with respect to magnetite at the silica The symmetry of the bands would seem to favor the activities imposed by the serpentine^brucite buffer (Frost former interpretation (Fig. 10c).
399 JOURNAL OF PETROLOGY VOLUME 50 NUMBER 3 MARCH 2009
(a) (b)
olivine olivine
0.4 mm
0.5 mm
(c) olivine
olivine
0.4 mm
Fig. 10. Magnetite in type 2 veins. (a) and (b) Fe maps of type 2 vein systems. It should be noted that magnetite is concentrated in vein centers and is never in contact with relict olivine. (c) Symmetrical bands of magnetite in a type 2 vein.
Figure 11 is a diagrammatic representation of the devel- mixtures undergoes transition from the antigorite charac- opment of type 2 veins. As fluid penetrates a favorably teristic of type 1 veins to the lizardite characteristic of oriented type 1 vein, brucite is destabilized and a magne- type 2. tite^serpentine assemblage forms at the fluid^rock inter- face (Fig. 11a and b). As the vein grows, olivine at its margins continues to react to brucite^serpentine, which, Alteration of plagioclase in turn, reacts with the fluid phase to form serpentine^ The earliest alteration seen in plagioclase (the chlorite^ magnetite, with magnetite nucleating on pre-existing, tremolite rims that form contemporaneously with type 1 older magnetite crystals now located in the center of the veins) appears to reflect the reaction growing vein (Fig. 11c). Multiple magnetite bands form as plagioclase þ olivine þ H2O þ aqueous silica ¼ the vein grows. As temperature drops during the course of ð6Þ serpentinization the serpentine in the brucite^serpentine chlorite þ tremolite:
400 BEARD et al. SERPENTINIZATION IN THE TROCTOLITE
(a)(b) (c) (d) young old young olivine olivine olivine olivine olivine olivine olivine olivine
serpentine + brucite type 2 serpentine magnetite
Fig. 11. Growth of type 2 veins. (See text for discussion.) (a) Early type 1 vein. (b) Initiation of magnetite growth in vein center. (c) Continued vein growth. Magnetite preferentially nucleates on pre-existing sites in the vein center. (d) Development of multiple magnetite bands. It should be noted that the vein grows at its edges and that brucite is continually formed at the vein edges and reacted out towards the vein center.
1.0 This desilication accompanies the main phase of serpen- tinization in the troctolites (type 2 veins) and occurs as a
consequence of the low aSiO2 imposed by the brucite^ serpentine assemblage. These reactions are consistent with the observed replacement of plagioclase by chlorite in the more serpentinized samples. The high Fe content of the late chlorite suggests that the Fe component of serpentine is preferentially consumed in chlorite-forming reactions. The most aluminous chlorites are amesites formed during the final stages of plagioclase alteration. This may reflect, at least in part, excess alumina released during formation of hydrogarnet in the final stages of desilication (Frost et al., 2008). All of the plagioclase desilication reactions are accom- density of unaltered olivine-rich troctolite panied by a loss of Na from the system. There is no local sink for Na in any of the serpentinized troctolites. Na 2.20 2.40 2.60 2.80 3.00 3.20 3.40 must be removed by the fluid phase, either out of the 3 density (gm/cm ) system altogether or, possibly, to be reprecipitated in Fig. 12. Grain density vs magnetic susceptibility plotted for the another part of the gabbroic section, perhaps as zeolites or U1309D troctolites (filled circle) and other oceanic and alpine serpen- albite (e.g. Blackman et al., 2006). tinites (Toft et al.,1990; Oufi et al., 2002; Bach et al., 2006).The olivine- rich troctolites follow a path close to that predicted for simultaneous magnetite formation and serpentinization (dashed line). Magnetic characteristics of the Hole 1309 troctolites A distinctive feature of the type 2 and larger type 1 veins in This reaction is very sensitive to aSiO .AttheaSiO pos- 2 2 the olivine-rich troctolite is that they are generally cored tulated for these rocks (i.e. log aSiO2 ¼ ^2 to ^4), the reac- tion is strongly overstepped and the products are favored at with magnetite. It appears texturally that magnetite any temperature below 6008C. formed very early in the serpentinization, after the forma- The bulk of plagioclase alteration in these samples tion of the type 1 veins. This is important because the pro- represents progressive desilication (Frost et al., 2008) via cesses by which magnetite forms has become a matter of the reactions considerable discussion (Frost & Beard, 2007; Evans, 2008; Frost et al. 2008). plagioclase þ serpentine þ H2O ¼ One way to characterize the relation between serpenti-
prehnite þ clinochlore þ SiO2 nization and the formation of magnetite is to plot the den- ð7Þ sity of a group serpentinized peridotites against their magnetic susceptibility (Fig. 12; Toft et al., 1990; Oufi et al., and 2002; Bach et al., 2006). On this diagram partially serpenti- nized serpentinites should form arrays that show an plagioclaseðor prehniteÞþserpentine þ H2O ¼ ð8Þ increase in susceptibility with a decrease in density. At hydrogrossular þ clinochlore=amesite þ SiO2: most localities partially serpentinized peridotites form
401 JOURNAL OF PETROLOGY VOLUME 50 NUMBER 3 MARCH 2009
arrays that show a distinct decrease in density before sus- increase in fluid/rock ratio and, as implied by the presence ceptibility increases. Toft et al. (1990) used this observation of lizardite, a lower temperature of serpentinization. to argue that the process of hydration is decoupled from The sequence of serpentinization reactions seen in the the processes that make magnetite. core 227 rocks is in full accord with that observed in ser- Bach et al. (2006) recognized a two-stage process of ser- pentinized peridotites at ODP Site 1274 (Bach et al., 2006). pentinization similar to that described here, with early In both cases, olivine reacts initially to form a mixture of magnetite-free mesh giving way to a magnetite-bearing brucite and serpentine in a rock-dominated system. As margin. They explained the relationship as reflecting a dif- larger amounts of fluid gain access to the serpentinizing ference in water^rock ratios. Early serpentinization caused rock, aqueous silica then reacts with the brucite to form isochemical hydration of olivine forming brucite and ser- magnetite and a more magnesian serpentine. Although pentine that are Fe-rich. Later open-system serpentiniza- the mantle olivine in Hole 1274 is substantially more mag- tion allowed access of silica from the outside (mostly nesian than the igneous olivine in U1309D (Mg-number 90 produced by hydration of Opx) causing the Fe-rich brucite vs Mg-number 85), the serpentine coexisting with magne- to react to magnetite þ serpentine. tite has an nearly identical composition (Mg-number 95 vs We have included in Fig. 12 data for the susceptibility Mg-number 96). In other words, the composition of ser- and density of olivine-rich troctolites from Hole U1309D. pentine coexisting with magnetite appears to be indifferent Because the olivine-rich troctolites have a primary density to the composition of the primary olivine in the rock. near that of peridotite, we contend that it is reasonable to Furthermore, the two-stage reaction that was postulated compare the olivine-rich troctolites on this diagram with by Toft et al. (1990), and is documented in Holes 1274 and other peridotitic rocks. The olivine-rich troctolites from U1309D, indicates that magnetite, in general, does not U1309D, like the serpentinites from ODP 395, 670 and form directly from olivine. This is consistent with petro- 920, plot at higher susceptibility for a given density than graphic observation at both sites (i.e. magnetite is not in alpine peridotites and lie close to the trend that would be direct contact with relict olivine).Taken together, this indi- followed if magnetite formed soon after serpentinization cates that final serpentine composition is dictated by mag- began (Toft et al., 1990).We contend that this trend reflects netite^serpentine, not olivine^serpentine, equilibria. This the fact that the type 1 veins occupy only a minor volume suggests that the desilication of ferrous serpentine is an of the rock, compared with the type 2 veins, which means important reaction in serpentinite petrogenesis. that in the 1309D rocks only minimal hydration occurred in a rock-dominated system before open-system hydration began. This is consistent with our observations that even ACKNOWLEDGEMENTS weakly serpentinized rocks from U1309D contain some The authors would like to thank the Joint Oceanographic magnetite (Fig. 2c). Institutions for allowing us to sail on Integrated Ocean Drilling Program (IODP) Expeditions 304 and 305 and providing us access to the samples. Thanks go also to the crew of the Resolution, the IODP staff and scientists and IMPLICATIONS AND the shipboard science parties of expeditions 304 and 305. CONCLUSIONS This work was supported by grant JOI 48299 to B.R.F., Early, type 1 veins consist of antigorite plus Fe-rich brucite, by grant JOI T305B27 to J.S.B., and by grant JOI T305B4 to P.F. This is SOEST Contribution 7608 and HIGP formed at high temperature, and occupy pre-existing Contribution 706. This project is based, in large part, cracks in the olivine. Mineral compositions in type 1 veins upon work supported by the National Science Foundation are controlled by Fe^Mg exchange between reactant oli- under a Cooperative Agreement (OCE-0431095) with the vine and product brucite þ serpentine. Later, type 2 veins Consortium for Ocean Leadership for a US Science are zoned and can be modeled by a two-stage process. The Support Program associated with the Integrated Ocean margins of the veins consist of brucite plus serpentine Drilling Program. Participation of R.S. in IODP formed directly by hydration of olivine. The centers of the Expedition 304 was funded by the Natural Environment veins (and the bulk of the serpentine) formed in a more Research Council’s UKIODP programme. This paper open system where silica was available from the ambient benefited greatly from constructive reviews by Wolfgang fluid, possibly with contributions from the alteration of Bach, Bernard Evans, John Ferry and editor Marjorie nearby plagioclase. These consist of lizardite plus magne- Wilson. tite that formed as a consequence of combined and oxida- tion plus desilication of Fe-rich serpentine. The two reactions probably occur simultaneously, with olivine-out REFERENCES occurring at the edge of the vein whereas brucite-out Abrajano,T.A., Sturchio, N. C., Bohlke, J. K., Lyon, G. L., Poreda, R. J. occurs within the vein proper. Type 2 veins record a large & Stevens, C. M. (1988). Methane^hydrogen seeps, Zambales
402 BEARD et al. SERPENTINIZATION IN THE TROCTOLITE
Ophiolite, Philippines: Deep or shallow origin? Chemical Geology 71, (2006). IODP Expeditions 304 & 305 characterize the lithology, 211^222. structure, and alteration of an oceanic core complex. Scientific Allen, D. E. & Seyfried, W. E. (2003). Compositional controls on vent Drilling 3, 4^11, doi:10.2204/iodp.sd.3.01.2006. fluids from ultramafic-hosted hydrothermal systems at midocean Kloprogge, J. T., Frost, R. L. & Rintoul, L. (1999). Single crystal ridges: An experimental study at 4008C, 500 bars. Geochimica et Raman microscopic study of the asbestos mineral chrysotile. Cosmochimica Acta 67,1531^1542. Physical Chemistry Chemical Physics 1, 2559^2564. Auzende, A. L., Daniel, I., Reynard, B., Lemaire, C. & Guyot, F. Mottl, M. J.,Wheat, C. G., Fryer, P., Gharib, J. & Martin, J. B. (2004). (2004). High-pressure behaviour of serpentine minerals: a Raman Chemistry of springs across the Mariana forearc shows progressive spectroscopic study. Physics and Chemistry of Minerals 31, 269^277. devolatilization of the subducting plate. Geochimica et Cosmochimica Bach, W., Paulick, H., Garrido, C. J., Ildefonse, B., Meurer, W. P. & Acta 68, 4915^4933. Humphris, S. E. (2006). Unraveling the sequence of serpentiniza- Muntener, O. & Hermann, J. (1994).Titanian andradite in a metapyr- tion reactions: petrography, mineral chemistry and petrophysics of oxenite layer from the Malenco ultramafics (Italy): implications for serpentinites from MAR 158N (ODP Leg 209, Site 1274). Geophysical Ti-mobility and low oxygen fugacity. Contributions to Mineralogy and Research Letters 33, L13306, doi: 10.1029/2006GLO25681. Petrology 116, 156^168. Barnes, I. & O’Neil, J. R. (1969). The relationship between fluids in O’Hanley,D.S.(1996).Serpentinites: Records of Tectonic and Petrological some fresh alpine-type ultramafics and possible modern serpentini- History. Oxford Monographs on Geology and Geophysics 34, 277 pp. zation. Geological Society of America Bulletin 80, 1947^1960. Onuki, H., Yoshida, T. & Nodachi, M. (1981). Notes on petrography Beard, J. S. & Hopkinson, L. (2000). A fossil, serpentinization-related and rock-forming mineralogy: X. Awaruite and other accessory hydrothermal vent, Ocean Drilling Program Leg 173, Site 1068 minerals coexisting with Ti-rich hydroandradite in metamor- (Iberia Abyssal Plain): Some aspects of mineral and fluid chemis- phosed ultramafics of the Sanbagawa belt. Journal of the Japanese try. Journal of Geophysical Research 105, 16527^16540. Association of Mineralogists, Petrologists, and Economic Geologists 76, Bethke, C. M. (2005). The Geochemists WorkbenchÕ, version 6.0, GWB 372^375. Essentials Guide. Urbana, IL: Hydrogeology Program, University of Oufi, O., Cannat, M. & Horen, H. (2002). Magnetic properties of Illinois, 76 pp. variably serpentinized abyssal peridotites. Journal of Geophysical Blackman, D. K., Ildefonse, B., John, B. E., Ohara, Y., Miller, D. J., Research 107, 2095^2114, doi:10.1029/2001JB000549. MacLeod, C. J. & the Expedition 304/305 Scientists (2006). Palandri, J. L. & Reed, M. H. (2004). Geochemical models of metaso- Proceedings of IODP, 304/305. College Station, TX: Integrated Ocean matism in ultramafic systems: Serpentinization, rodingitization, Drilling Program Management International, Inc., doi:10.2204/ and sea floor carbonate chimney precipitation. Geochimica et iodp.proc.304305.2006. Cosmochimica Acta 68, 1115^1133. Charlou, J. L., Fouquet, Y., Bougault, H., Donval, J. P., Etoubleau, J., Peretti, A., Dubessy, J., Mullis, J., Frost, B. R. & Trommsdorff, V. Jean-Baptiste,P.,Dapoigny,A.,Appriou,P.&Rona,P.A.(1998). (1992). Highly reducing conditions during Alpine metamorphism
Intense CH4 plumes generated by serpentinization of ultramafic of the Malenco peridotite (Sondrio, northern Italy) indicated by rocks at the intersection of the 15820’N fracture zone and the Mid- mineral paragenesis and H2 in fluid inclusions,. Contributions to Atlantic Ridge. Geochimica et Cosmochimica Acta 62, 2323^2333. Mineralogy and Petrology 112, 329^340. Charlou, J. L., Donval, J. P., Fouquet,Y., Jean-Baptiste, P. & Holm, N. Rinaudo, C., Gastaldi, D. & Belluso, E. (2003). Characterization of
(2002). Geochemistry of high H2 and CH4 vent fluids issuing from chrysotile, antigorite and lizardite by FT-Raman spectroscopy. ultramafic rocks at the Rainbow hydrothermal field, (36814’N, Canadian Mineralogist 41, 883^890. MAR). Chemical Geology 191, 345^359. Rumori, C., Mellini, M. & Viti, C. (2004). Oriented, non-topotactic Coleman, R. G. (1963). Serpentinites, Rodingites, and Tectonic Inclusions in olivine^serpentine replacement in mesh-textured serpentinized Alpine-type Mountain Chains. Geological Society of America, Special Papers peridotites. EuropeanJournal of Mineralogy 16, 731^741. 73. Russell, M. J. & Arndt, N. T. (2005). Geodynamic and metabolic Evans, B. W. (2004).The serpentinite multi-system revisited; chrysotile cycles in the Hadean. Biogeosciences 2, 97^111. is metastable. International Geology Review 46, 479^506. Shervais, J. W., Kolesar, P. & Andreasen, K. (2005). A field and che- Evans, B. W. (2008). Control of the products of serpentinization by the mical study of serpentinizationçStonyford, California: chemical 2þ Fe Mg^1 exchange potential of olivine and orthopyroxene. Journal flux and mass balance. International Geology Review 47,1^28. of Petrology 49, 1873^1887. Sleep, N. H., Meibom, A., Fridriksson, , Th., , Coleman, R. G. & Frost, B. R. (1985). On the stability of sulfides, oxides, and native Bird, D. K. (2004). H2-rich fluids from serpentinization: metals in serpentinite. Journal of Petrology 26,31^63. Geochemical and biotic implications. Proceedings of the National Frost, B. R. & Beard, J. S. (2007). On silica activity and serpentiniza- Academy of Sciences of the USA 101, 12818^12823. tion. Journal of Petrology 48, 1351^1368. Toft, P. B., Arkanai-Hamed, J. & Haggerty, S. E. (1990).The effects of Frost, B. R., Beard, J. S., McCaig, A. & Condliffe, E. (2008).The for- serpentinization on density and magnetic susceptibility: a petro- mation of micro-rodingites from IODP Hole U1309D: key to under- physical model. Physics of the Earth and Planetary Interiors 65, 137^157. standing the process of serpentinization. Journal of Petrology 49, Viti, C. & Mellini, M. (1998). Mesh textures and bastites in the Elba 1579^1588. retrograde serpentinites. European Journal of Mineralogy 10, Ildefonse, B., Blackman, D. K., John, B. E., Ohara, Y., Miller, D. J., 1341^1359. MacLeod, C. J. & the IODP Expeditions 304^305 Scientists,
403