PRECAMBRIAN GOLD: PERSPECTIVES FROM THE TOP AND …rruff.info/doclib/cm/vol31/CM31_917.pdfdu magmatisme alcalin et d'une pdriode soutenue de carbonatisation d Kirkland Lake. Une fois

9L7

The C anndian M ine ralo gi stVol. 3 1, pp. 917 -944 (1993)

PRECAMBRIAN GOLD:PERSPECTIVES FROM THE TOP AND BOTTOM OF SHEAR ZONES*

EIONM. CAMERON

Geological Sarvey of Canada, 601 Booth Street, Ottawa, Ontario KIA 0E8 andDepartment of Geology, University of Onawa, Ottana, Ontario KIN 6Ns

ABsrRAsr

Studies of the Kirkland Lake gold camp, along a narrow, brittle shear in the upper crust, and the Bamble bel! Norway, abroad shear zone from tle lower crust, test a hypothesis that deep shear-zones were a source for gold. At KirHand Lake, long-term displacement along a major strike-slip fault resulted in localized extension and thinning of the crust. This produced exten-sional basins at the surface, a broad deformation zone at depth, and upwelling of mantle. Partial melting and decarbonation ofthe upwelling mantle were represented at Kirkland Lake by alkaline magmatism and a long period of carbonatization. Whenexlension ceased, there was uplift, during which overpressured, CO2-bearing fluids, which carried gold from.depth, werefocused into the narrow, upper part of the shear zone. The Bamble shear belt is a lower crustal analogue of this extensionalterrane, characterized by a counterclockwise P - T - t metamorphic path. During extension, mafic magmacanied CO, into theshear belt; the CO2 may have been a cause for oxidation of its rocks during prograde metamorphism. There-was a long periodof isobaric sooling, when fluid movement was inhibited, because inclusions formed during peak r4e_tamorphism were under-pressured. During this period, internal buffering by the minepl assemblage caused oxidation of pynhotite to pyrite. Pyrrhotiteis a principal host of gold in mafic lower crust; its transformation to pyrite made the gold accessible to later fluids. Isobariccooling was followed by rapid, isothemral uplift. Fluid inclusions, containing CO2, now became overpressured, causing theirrelease. It was during this period that gold was likely extracted from the deep shear belL the rocks of which now contain only0.06 ofthe crustal abundance ofthis element. I

Keywords: Kirkland l,ake, Ontario, Bamble belt, Norway, gold, extensional tectonics, metamorphism, alkaline magmatism,mantle, oxidation, isobaric cooling, uplift.

Sovnaeru

Les 6tudes des gltes aurifbres du camp minier de Kirkland Lake, en Ontario, situ6s le long d'une dtroite zone cisaill6e cas-sante dans la cro0le suffrieure, ainsi que des roches de la ceinture de Bamble, en Norvdge, une zone cisaill6e plus diffrrse typ-ique de la cro0te inf€rieure, permettsnt tie vdrifier l'hypothdse que les zones cisaill6es profondes sont la source de l'or. AKirkland l,ake, les d6placements horizontaux soutenus le long d'une faille majeure ont provoqud une extension localisde et unamincissement de la cro0te. Il en r6sulta des bassins extensionnels d la surface, une zone de d6formation diffrrse en profondeur,et un bombement du manteau. l,a fusion partielle et la perte de gaz carbonique 1ib6r6 du rn rnteau ainsi soulevd sont i l'originedu magmatisme alcalin et d'une pdriode soutenue de carbonatisation d Kirkland Lake. Une fois la p6riode d'extension ter-minde, une phase fluide surcomprim6e, riche en COr, a transportd l'or des niveaux profonds lors d'un soulbvement rdgional,pour ensuite I'acheminer vers la partie 6troite sup6rieure de la zone cisaill6e. l,a ceinture de Bamble sert d'exemple du niveauinf6rieur de cette zone d'extension. Elle ddmontre un [ac6 P - T - t de l'dvolution m6tamorphique contraire au sens de l'hor-loge. Pendant l'extension, des magmas mafiques ont introduit le CO2 dans la zone cisaill6e, ce qui pounait bien expliquerI'oxydation des roches durant le m6tamorphisme prograde. Au cours d'une longue p6riode de refroidissement isobare, leddplacement de la phase fluide 6tait limit6, parce que les inclusions fluides form6es sous conditions de mdtamorphisme maxi-males sont sous-comprim6es. Pendant cette pdriode, un tampon interne des assemblages a men6 i l'oxydation de la pyrrhotiteen pyrite. Comme la pyrrhotite est l'hdte principal de 1'or dans la cro0te infdrieure mafique, sa transformation en pyrite renditI'or accessible aux fluides tardifs. Le refroidissement isobare fOt suivi par un souldvement isotherme rapide. [.es inclusions flu-ides, contenant du gaz carbonique, devirent surcomprim6es, et furent d6truites. C'est durant cette p6riode de d6gazage queI'or aurait probablement 6t6 extrait de cette zone cisaill6e profonde, dont les dchantillons ne rendent plus compte que de 670 del'or dans la cro00e terrestre.

(Traduit par la R6daction)

Mots-cl€s: Kirkland l,ake, Ontario, ceinture de Bamble, Norvbge, or, tectonique d'extension, m6tamorphisme, magmatisme2lealin, sranteau, oxydation, refroidissement isobare, soulOvement.

* Geological Survey of Canada contribution number 35992.

918 THE CANADIAN MINERALOGIST

INrnooucrrou

Large Precambrian gold deposits are invariablyassociated with major linear faults or shear zones. Therelationship is so apparent that understanding thegenesis of these deposits must depend on knowledgeof the processes, such as localized metamorphism andmagmatism, that occur along these structures. Shearzones serve as an important conduit for material trans-fer within the lithosphere, for both fluids and melts;younger fault zones of similar regional extent havebeen shown to cut the lithosphere from mantle to sur-face. The presence of rocks of mantle origin alongthese fault zones in close spatial association with somegold deposits suggests that the entire lithosphere maybe involved in the genesis of deposits (Libby et al.1990). Yet veins appear to postdate peak metamor-phism and magmatism. Possible explanations for thisenigma are not readily apparent at the level of expo-sure of the gold veins. This paper examines some offhe deeper processes tlat may be relevant.

The Kirkland Lake camp, in the southern Abitibigleenstone belt, is used as a model for the "top" of agold-productive shear-zone. At Kirkland Lake, virtual-ly all of the features tlat are associated with shear-zone-related gold are well exposed. The selected areafor the "bottom" of the shear zone is the ProterozoicBarnble shear belt, Norway. Major transcurrent shear-zones broaden with depth below the brittle - ductiletransition to as much as 40 km wide in the lower crust@ak et al. 1975, Hanmer 1988). Such is their volume,allied with strain-induced permeability, that they pro-vide a potential source for gold (Cameron 1989a, b).The Bamble shear belt, 30 km wide, was tectonicallyactive during metamorphism at granulite and upperamphibolite grade, when the presently exposed areawas at a depth of about 25 km (Lamb et al. 1986).Initial studies of the Bamble rocks (Cameron 1989a, c)showed tlem to be strongly depleted in gold and inother chalcophile elements enriched in gold deposits.

Panr I:Top op rm SIrn'AR Zo].IE

T\e 2.7 Ga Abitibi belt has produced more goldthan any other Archean greenstone belt. Much of thishas come from mines on and near two major east -west linear faults, the Porcupine - Destor and theLarder Lake - Cadillac (-LC) (Fig. 1). targe deposits,with production plus reserves in excess of50 t Au, arenot evenly distributed along these faults, but show astrong clustering near the western ends of the mainzones of deformation" at Timmins and at KirklandLake.

Gsolocv oF KIRKLAND LAI<E

Much of the geological knowledge of KirklandLake derives from the classic reports of Thomson(1950) and Thomson et al. (1950), which are thesource of much of the description below. The principalfeature of the district is the Kirkland Lake - LarderLake Fault zone (KLF), a branch of the LLC fault(Fie. 2). Mines within the KLF, mainly at KirklandLake and t arder Lake, have produced 1 1 10 t Au. TheKLF is 50 km long and up to 5 km wide and cuts abasaltic (geenstone) basement. It forms a "basin" con-taining interbedded sedimentary and alkaline volcanicrocks of the Timiskaming Group, which lie uncon-formably on the older basaltic rocks. This basin can besubdivided into three segments (Fig. 3). In the west,the Kirkland basin trends -067"; in the east, theLarder basin has a variable strike, which in the vicinityof the Kerr Addison mine is 060" to 070". These arelinked by the Dobie basin, which trends 106'.

Rocks of the Timiskaming Group, up to 3500 mthick, form a south-dipping homocline. The southernmargin is a fault contact against uplifted basaltic base-ment, whereas the northern margin is, in part, anunconformity of Timiskaming rocks on basement, andelsewhere a fault. Hyde (1980) identified a non-marine

, 50km ,

KIRKLANDLAKE

LAKE

FIc. l. Location of -ajor gold deposits (production plus reserves greater than 50 t Au) in the region of the Abitibi belt contain-ing the Porcupine - Destor and l,arder Lake - Cadillac faults.

series of conglomerate and sandstone from a braidedriver environment, thin to medium-bedded sandstone"siltstone, and argillite, deposited in a flood plain, andthick-bedded sandstone, which may represent eoliandunes. There is also a resedimented series: principally

PRECAMBRIAN GOLD AND SHEAR ZONES 9L9

1800 m of proximal turbidites with conglomerate,probably formed in a submarine fan. There are calc-alkaline to alkaline flows and pyroclastic rocks, thealkaline rocks including alkali olivine basalt, trachy-basalt, tristianite, hawaiite and mugearite (Basu et al.

MRDTR LAKT318 t AuKIRKTAND LAKE

723 t Au

Ftc. 2. Geology of the Larder Lake - Cadillac fault.

ARCHEAN

ffiffi

li;l;i;i;i;i;i;i;lli:i:::i:::::::::l

flFtc. 3. Geology of Kirkland Lake - Larder Lake fault zone (I(LF). LLF: Larder l,ake fault; I(LF: Kirkland Lake fault; UC:

Upper Canada mine; KA: Kerr Addison mine. Modified from Thomson (1950).

Syenlte and related lntruslona

Temlskamlng group

Malnly mstavolcanlo andmetaeedlmentary rocko

Glsologlcal boundary(detlned, aeeumed), .. /- - /

Basal Temlskamlng groupuncon to rm f t y . . . . . . . . . - . r l

Major lault, . ...^rvwvw

M a l o r g o l d d e p o s l t . . . . . . . . . . . . .


34',Fro. 4. A. Surface plan of composite syenitic intrusion that hosts most of the gold ore at Kirkland Lake (from Thomson et al.

1950). B. and b. Orientati6n of structures formed by sinistral transtension and by dextral fanspression; EF: extensionalfracture (from Sanderson & Marchini 1984).

1984).The district is notable for LllE-enriched syenitic

intrusions, which represent one of the earliest majoroccurrences of alkaline magmatism. These form acluster of round plutons south of the KLF, but withinthe KLF they are strongly elongate, parallel to themargins of the basin (Figs. 3, 4). Intrusions are maficto felsic in composition. For the Murdock Creek stock,Rowins et al. (7991) found alkali-feldspar syenite tobe the dominant rock-type, with melasyenite, mela-monzodiorite, meladiorite, hornblendite, and clinopy-roxenite also represented. Lamprophyre dykes cutTimiskaming rocks and the alkaline intrusions.

Greenstones underlying the Timiskaming Groupwere regionally metamorphosed at prehnite - pumpel-lyite to greenschist grade (Jolly 1978). Timiskamingrocks show only diagenetic assemblages, with clayminerals and white mica being the only secondaryminerals. Clasts within the Timiskaming conglomer-ates contain a prehnite - pumpellyite assemblage(Jolly 1978), implying that metamorphism of thegreenstones preceded deposition of the Timiskaming.

Al1 important gold deposits-'occur along steeply dip-ping faults, classified by Thomson (1950) as "strikefaults", that parallel the trend of the Kirkland andLarder basins. Seven mines, with a total production of723 t Au, are spaced along a four-km length of theKirkland Lake fault (Fig. 3). This fault strikes 067',dips steeply south, and has a reverse component, withthe south side moved upward as much as 460 m.

Strike faults created a series of parallel blocks, each ofwhich rose relative to the block to the north, with themajor displacements preceding the forrnation of gold-bearing veins. At Kirkland Lake, deposits are hostedby an elongate, steeply south-dipping intrusion ofmafic to felsic syenite cut by the Kirkland Lake fault.The long axis of the intrusion has the same orientationas the strike of this fault (Frg. A) and as the axis ofthe Ki*land basin. Deposits in the other basins lie onsteeply dipping faults subparallel to the Kirkland Lakefault. The Kerr Addison deposit (318 t Au) in theLarder basin is located along tle Larder Lake fault,which trends 060" - 070'. The only important depositin the Dobie basin, the Upper Canada mine (43 t Au),is associated with faults trending 073' to 080". Othersets of fault in the KLF trend north (cross faults), ornortheasterly (diagonal faults), the former beingmainly post-ore.

There was extensive carbonatization within theKLF, structurally controlled along strike and diagonalfaults. All rockso and particularly the basic units, werealtered into siliceous carbonate rock containing iron-rich dolomite, calciteo and green mica. Syenite intru-sions are affected, but also, in places, cut carbonatizedzones (Thomson 1950). This observation, togetherwith Hewitt's (1963) finding of pebbles of carbona-tized rock in upper beds of the Timiskarning, suggest along period of CO2 fluxing. In ore-bearing rocks, cat-bonatization was sequentially followed by pyritizationof wallrocks, precipitation of quartz, introduction of

PRECAMBRIAN GOLD AND SHEAR ZONES 921

GREENSTONE VOLCANISM

OROGENESIS, METAMORPHISM

STRIKE.SUP DEFORMATION

FAULTING

NMFKAMING DEPOSITION

ALKAUNE VOLCANISM

ALKAUNE PLIITONISM

CO, FLUXING (+S)

GOID VEINS

2730 Mo<t_+7ol

27cnl-l26a4 Tronstenslo! . Tronspreslc!

Normol Reverse

H2680 ]-{ 2677 Myr

zooo H2673 (omp.)

l- -+H

ARC VOLCANISMAND OROGENESIS

STRIKE-SLIP SEDIMEMATION,MAGMATISM AND OROGENESIS

Frc. 5. Sequence of geological events, Kirkland Lake - Larder Lake area. See text for sources of geochronological data.

calcite and, finally, deposition of sulfides, telluridesand gold. Values of 6l3C in carbonates from KirklandLake vary from -2.5 to 44 %o (Kenich et al. 1987), arange consistent with CO, of juvenile or mantle origin(Taylor 1986).

Relevant temporal relationships are summarized inFigure 5. These are constrained by high-precisionU-Pb dates on zircon and titanite. Corfu & Noble(1992) estimated that volcanism in the Abitibi green-stone belt occurred between 2i730 and 2700 Ma ago,and orogenesis, comprising folding, ttrusting, deposi-tion of turbidites and calc-alkaline plutonism, at about2i700 to 2685 Ma. The latter period is taken here to becoincident with regional metamorphism. Subsequentto these regional events were localized sedimentation,magmatism and orogenesis along the KLF.Timiskaming sedimentation was coincident with mag-matism within the KLF, which spans the interval 2680- 2677 Ma (Cortu et al. l99l), whereas alkaline mag-matism outside the KLF, represented by the Ottostock, was of a similar age at2680 + 1 Ma (Corfu etal. 1989). Corfu er al. (1991) emphasized tle closetemporal relationship between Timiskaming sedimen-tation and alkaline magmatism, although Timiskamingsedimentation may have started as early as 2685 Ma(Corfu er al. 1,991).In the discussion below, sedimen-tation and magmatism within the KLF are related tolocalized extension (transtension) along the KLF. Alamprophyre dyke that cross-cuts syenite, and thataccompanied later folding and thrusting of theTimiskaming rocks (Corfu et al. 1989), has been esti-mated at 2674 L 2Maby U-Pb dating of titanite(Wyman & Kenich 1989).

Attempts to directly date minerals from gold-bearing veins elsewhere in the Abitibi bel! includingmuscovite, rutile and scheelite, have given dates that

are substantially younger, by 50 Ma or more, than theyoungest intrusions (e.9., Bell et al. 1989). However,there is no consensus on whether these dates representthe time of introduction of gold. At Val d'Or, at theeastern end of the LLC fault, Wong et al. (199L) usedPb--U dating of rutile to estimate an age of 2599 1- 7Ma for alteration around gold-bearing veins, whereasregional greenschist metamorphism occurred at 2684+ 7 Ma. But Claou6-Long et al. (1990) estimatedgold-bearing veins from the same area at 2680 Ma,based on hydrothermal zircon. [n the Yilgarn Block,Australia, CTark et al. (1988) estimated from U-Pb inrutile that veins in the Victory mine were formed30 Ma later than the peak of metamorphism and theyoungest felsic plutons. No precise date is availablefor the formation of gold-bearing veins at KirklandLake. However, two different types of evidence sqg-gest that a substantial interval of time elapsed betweenemplacement of the host syenite and the veins. Thefust is described in the succeeding sections, where thesyenite is shown to have been intruded during a phaseof crustal extension and basin formation. whereas theveins were emplaced during a later phase of compres-sion and uplift. Second, Haftori (1993) found altaiteintimately associated with gold in the Kirkland Lakeveins to have a radiogenic Pb isotopic composition.The data are consistent with formation of the veinsafter the U-rich host syenite had generated significantradiogenic Pb, which was extracted by the hydrother-mal fluids.

RSI-.ATIoN oF [.ocALZED Smuvmvrerroll exlMacuerrsv ro SrRn<s-SLP Faurrwc

Strike-slip movements, where the velocity is suffi-ciently high and long-lived, may cause uplift or sub-

922 TIIE CANADIAN MINERALOGIST

sidence along faults. These effects occur where thefaults cross the horizontal trace of plate movement, orat overstepso or at fault terminations. Narrow orogenicbelts may be produced by a phase of transtension,leading to basin formalion and sedimentation, fol-lowed by transpression causing basins to be upliftedand folded @eading 1980). Extensional (srike-slip orpull-apart) basins along major strike-slip faults aresmall but deep. Subsidence is rapid. The dynamicenvironment is reflected in the sediment fill" withrapid change in facies as a principal characteristic.Basins above sea level are typically bordered by allu-vial fans that pass rapidly through fluvial deposits intolacustrine sediments. Marine basins are filled by sub-aqueous mass-gravity transport of coarse clastic mate-rial, deposited, in part, on submarine fans. During sub-sidence, normal faulting dominates, whereas duringtranspression, reverse and thrust faults are developed.In strike-slip orogenic belts, magmatic activity isalmost entirely confined to the transtensional phase,and metamorphism of the basinal sediments is feeble.

Volcanic rocks are found in a minority of strike-slipbasins. Where present, they are commonly of alkalinecomposition, such as in the Erzincan basin along theNorth Anatolian fault in Turkey (Hempton & Dunne1984) and the submarine Pantelleria Trough, betweenSicily and North Africa (Ben-Awaham et al. 1981).The sparing occwrence of volcanic rocks in strike-slipbasins, their restriction to the transtensive phase, andtheir tendency to alkaline compositions, are consistentwith the model of McKenzie & Bickle (1988) for gen-eration of basaltic magma by extension of the lithos-phere. Extension causes upwelling of mantle; solidusboundaries may be crossed, causing partial melting.Where there is major extension, as at plate margins,upwelling is greatest, and large volumes of tholeiiticbasalt are generated by extensive melting at shallowlevels. Where extension is less, upwelling is moremodest, and partial melting, if it occurs, leads to asmall fraction of melt, of alkaline composition, atgreater depth. The degree ofextension (p), defined asthe ratio of final to initial surface-area, required togenerate melt depends on several conditions, includingthe potential temperature (T") of the mantle. For a Toof 1480'C, melting, to proiluce alkali basalt, occursonly where p reaches 1.5 to 2.0. Strike-slip move-ments usually cause only moderate extension. Forbasins along the boundary between the North Americaand Pacific plates, with a high tansform-velocity of4 - 6 cmlyr, J.A. Andrews & M.S. Steckler (in Pihnan& Andrews 1985) found B to range up to about 1.6.This is near the lower bound required for melting.Thus most strike-slip basins will form in crust whereextension is less than that required for partial melting;these basins will not contain volcanic rocks.

The southern part of the Archean Superior Provinceis composed of east-trending, elongate belts (sub-provinces) of metavolcanic and granitic rocks (green-

stone belts) that altemate with metasedimentary belts.Major, linear faults or "breaks", such as the LLC,which also approximate an easterly trend, cross sub-provinces or form their boundaries. In part, these fea-tures are now related to terrane accretion, thrustingand strike-slip movement during oblique convergence(e.g., Percival & Williams 1989). In the YilgarnBlock, Australia, gold deposits are associated withfaults that extend for hundreds of kilometen, some ofwhich may be terrane boundaries, and which were thelocus for intrusions of lamprophyre and felsic por-phyry (Libby et aL 1990).

The Timiskaming Group at Kirkland Lake is inter-preted to have been deposited within deep, nalrowbasins formed during strike-stp extension. Its featuresshow close correspondence with those of youngersfiike-sLip mobile belts (Table 1). Alkaline magmatismis most reasonably interpreted to be the product of par-tial melting during the mantle upwelling that accom-panied thinning of the lithosphere. The Timiskamingis cut by elongate intrusions, such as the syenitic intru-sion hosting the ore at Kirkland Lake (Fig. 4A); elon-gation is parallel to the basin axes. The plutons areinterpreted to have been passively emplaced intoextensional fractures after the Timiskaming had beentilted to the south. but while the fault zone was stillundergoing extension. The approximately 067' orien-tation for the Kirkland and tarder basins and for theplutons passively intruded into extensional fracturessuggested to Cameron (1990) that these events were aconsequence of sinistral transtension along the east -west-trending LLC fault @g. 4B). Many of the struc-tural features that might have been associated withsinistral transtension have been obscured by laterdeformation. However, based on the geometry ofinternal magmatic foliation, distribution of igneousphases, and pluton shape, Cruden (1992) has shownthat the Murdock Creek syenitic stock (Fig. 3) wasemplaced into a dilational zone during sinistral defor-mation. In contrast" the round, alkaline intrusionssouth of the KLF were forcefully emplaced. An analo-gous contrast between elongate and round alkaline andcalc-alkaline granites occurs along the Ajjaj shearzone, Saudi Arabia (Davies 1982).

Extension may be interpreted to have occurred inthe south quadrant at the west termination of a faultundergoing sinistral displacement (Cameron 1990).Extension in this quadrant accounts for the major con-centration of alkaline plutons south of the fault zone.If the proposal of Leclair et al. (1993) is correct, thatthe LLC fault continues west of Kirkland Lake, butchanges strike to west-southwest, then the Kirklandand Larder basins are releasing bends along the LLC,where this fault crossed the direction of plate motion.The two explanations are not necessarily incompatr-ble; a major part of strike-slip displacement along theeastern part of the LLC fault could have been taken upbv extension around Kirkland Lake, with a lesser

amount expressed as displacement along a fault to thewest of Kirkland Lake.

Strike faults within the KLF" such as the KirklandLake fault, are reverse faults. With an orientation thatis the same as the basin and the principal axis of theelongate syenite intrusions, these are interpreted tohave resulted from reactivation during compression ofnormal faults that formed during the earlier extension-al phase.

Thus there is evidence for two distinct phases ofdeformation: first extension, as indicated by basin for-mation, sedimentation, and injection of alkalinemagma, and then compression, with reactivation ofearlier normal faults as reveme faults and the introduc-tion of gold-bearing fluids along these faults.Shortening of the area also is reflected in differencesin petrography for the alkaline and calc-alkalineigneous rocks near Kirkland Lake. Ldvesque et al.(1991) have shown tlat the round intrusions south ofthe KLF crystallized at depth: they are plutonic. Byconfast, intrusions within the KLF. and scarce intru-sions north of the fault zone, are hypabyssal and occurin association with similar rocks of extrusive origin.This implies relative uplift of the southern block sub-sequent to the extension-related magmatic activity,compatible with the hypothesis of reverse faultins

923

noted above.The orientation ofthe reverse faults that guided the

ore fluids is consistent with reactivation under dextraltranspression (Fig. C), although this need not be aunique solution. Previous studies along the -200 kmlength of the LLC fault provide evidence for bothsinistral and dextral movement. At Rouyn-Noranda,east of Kirkland Lake, Hlbert et al. (1984) identifiedtwo important deformations. They related the first,represented by west-northwest folds, to sinistralmovement along the Porcupine - Destor and LLCfaults. This was followed by folding along east-westaxes caused by compression normal to these faults.Along the eastern portion of the LLC fault, Robert(1989) found evidence for dextral transpression, coin-cident with the formation of important gold-bearingveins.

Colvine et al. (1988) suggested that Timiskamingsediments at Timmins may have been deposited in apull-apart basin. Extensive carbonatization in thiscamp suggests extension-related uplift of mantle, withdegassing. But the absence of significant volumes ofalkaline rock at Timmins may indicate a lower degreeof lithospheric extension or a stalling of melts at thebase of the crust. Mueller et al. (L991) considered thatsediments deposited within the Duparquet basin,

PRECAMBRTAN GOLD AND SHEAR ZONES

TABLE 1. COMPARISON OF CHARACTERISNCS OF STRIKE-SUP MOBILE BELTSAw|fi TIMISKAMING GROUP, KIRKTAND LAKE.IIRDER I-AKE FAULTZONE.

(a) Mainly after Reading (1980)

CFIARACTERISTICS OFSTRIKE€UP MOBILE

BELTS

TIM|SKAMING GROUP,KIRKLAND LSXE - I-ARDER

LAKE FAULTZONEr€t?s3[nalsedlmentstonr

MarlnsSedlmertadon:

Facles Change:

StraffgraphlcThlckn$s:

Metamorphlem:

Volcanlcm:

Faultng DuringTranstsnslon:

Fautung DurlngTransp]gsslon:

Allwial fans. fluvial andhorsfine deposits; oftenthick conglomerates.

Maes-gravity fansport ofcoargg dagtica to submarinefans, baain plains, slopeaprons.

Extseme lateralfacieschange.

Great relative to basin she.

LiUe or none.

Mosty abeefi wfrsrs present,ofbn alkaline.

Normal faul6 and extensionaltfactures.

ReveBe, ihrust

BEided river, floodplaln, andeollan deposits; conglomerateunlts to 1300 m.

Thick proimal i/rbidite,conglomerates; probablydepoeited in a submarine fan.

Rapld local facies change; e.9.,allwlal fanlflwial sediments toturbidites, all in less than 25 km.

3.5 km lhidcess in basin of 5 krwidh.

Less than sub-gresmchist

Alkallne fiove and pyroclasticdgposits.

Normal faults?; alkalinelnfusions In extensionalfraciur€s.

Reverse faulte. ThrustB?


Quebec, may have formed at a releasing bend duringdextral strike-slip along the Porcupine - Destor fault.Toogood & Hodgson (1985) suggested that the Dobiebasin (Fig. 3) formed as a pull-apart during dextraltranspression while sedimentation and volcanism weretaking place within other parts of the KLF. Theirscenario of 1) dextral transpressiono accompanied bysedimentation and volcanism and 2) a compressivephase, with beds becoming vertical before syeniticmagma was intruded, differs from the interpretationgiven here.

A younger example that replicates several featuresof the KLF is the 500-km long, 20-km wide NorthPyrenean transform zone (Vielzeuf & Kornprobst1984, Golberg & Leyreloup 1990). Along this faultzone, there was an init ial phase of extensionaltranscurrent deformation during Upper Cretaceoustime. Pull-apart basins developed, with alkaline mag-matism and metamorphism localized within the faultzone. This was followed by a phase of compression.Uplift of mantle along the fault zone is most insistent-ly shown by the emplacement of lherzolites, whichVielzeuf & Kornprobst (1984) attributed to the com-bined effects of extension and compression. Extensioncaused mantle to rise; during compression, slices ofmantle were tlrust upward.

Penrtr:Tru MaNn-e BELow KIRKLAND LAIc

In this account, two principal features of theKirkland Lake camp derive from the mantle: alkaline

m0 l lm l3mTEMPERATURE,'C

FIc. 6. Solidus curve for CO2 - H2o-bearing or carbonatedperidotite. After Meen (1987).

magmas and CO2 fluxing. Both can be related toupwelling of the mantle that resulted from localizedlithospheric extension and thinning along the LLCfault. For volatile-bearing (COz and H2O) or carbon-ated peridotite, carbonate is stable on the solidus atpressures greater than 17 kbar. Partial melts of carbon-ated peridotite, released at depth during moderateextension, will rise to intersect a solidus ledge near17 kbar (Fig. 6). There, following Meen (1987) andothers, melts will react to produce carbonate-free peri-dotite or pyroxenite and CO2-rich fluid. Repeatedimpingement of melts at the solidus ledge enrichesperidotite in incompatible elements, creating a metaso-matized zone" which can melt to an alkaline basaltmagma rich in these elements. These melts may carryCO2 to the crust.

The alkaline magmas were strongly enriched inlarge-ion lithophile elements (LILE) (Fig. 7) anddepleted in Nd. Ben Othman et al. (1990) estimated ae*6 of 2.5 for the Timiskaming volcanic suite. Hattori& Hart (1990), who analyzed samples of clino-pyroxene that are unaffected by hydrothermal fluids,found no difference in Sr and Nd isotope ratios forsyenite and lamprophyre, indicating a common sourcefor these rock types. The CTSrF6Sr)i values are low,0.7009 to 0.7016, and eN6 is high,0.8 to 3.0. Positivee*4 values indicate previous depletion by the extrac-tion of melts. Effichment of the previously depletedmantle in LILE occurred too close in time to theextraction of the alkaline melts to allow for generationofradiogenic isotope components. There are two pos-sibilities to explain the relative lack of radiogenicisotopes. The first, outlined in the previous paragraph,is that localized lithospheric extension along a majorfault caused mantle upwelling. This triggered LILE-and volatile-rich melts to rise within the mantle andcreate a metasomatized zone, which tlen became thesource of the alkaline melts. The source of the LILE-and volatile-rich melts may have been undepletedmantle, with possible addition of material from a slab.In this model, metasomatism and partial melting areparts of a single process tlat extended over only a fewMa. The second model, favored by Hattori & Hart(1990), is that LILE enrichment took place by materialdirectly introduced from a subducting slab. Thisrequires a cycle from generation of the slab rocks,through subduction, metasomatism and partial meltingthat is rapid, i.e., less than 20 Ma.

A feature that attracted Hattori & Hart to a fastrecycling model is that the LllE-enriched alkalinemagmas were not similarly enriched in high-field-strength elements ([IFSB), a feature that is widelytaken to be characteristic of arc volcanism related tosubducted slabs. Spidergram plots (Fig. 7) of rockunits comprising the Murdock Creek stock (Fig. 3)show troughs for the IIFSE, particularly M. The rea-sons why this may be related to processe$ other ttransubduction have been addressed by Rowins el al.

5 0 um:q

.-i{

75

l 0

e.co><uig2 t ,aLUe.o-

%AMPHIBOLE

PRECAMBRIAN GOLD AND SHEAR ZONES

B E T b K T a C oR b U N b L a s r

(1993). HFSE troughs are found in continental ultra-potassic rocks distant from subduction zones. Samplesfrom the Murdock Creek stock are classified wiihingroup-Il ultrapotassic mafic rocks of Foley et al.(1987), which are characteristic of continental riftzones, such as the East African rift. Depletion inHFSE (Ti-group) requires a Ti-rich residual phase;Rowins et al. (1993) favored titanite, in part becauseof the intrinsically oxidized nature of the MurdockCreek magmas.

Oxidation provides a link between the mantlesource-region of the alkaline magmas and tle gold-bearing veins. Cameron & Hattori (1987) notedseveral characteristics of the veins at Kirkland Lakethat indicate their deposition from oxidized fluids,which contrasted with the otherwise reduced nature ofthe Archean crust. This included the presence of sul-fate minerals, hematite in altered wallrocks, and strongsS depletion of pyrite intimately associated with thegold, indicative of isotopic fractionation betweensulfide and sulfate. Rowins et al. (1991) found that theMurdock Creek stock formed from an inherentlv oxi-dized magm4 that was derived from oxidized mantle.This is shown by clinopyroxene and biotite, whichretain a consistently low Fe/(Fe + Mg) and highF€'/(Fez+ + Fer+) ratios tbroughout the pluton's evo-lution, and by abundant, early-formed magnetite andtitanite. The link between oxidized ore-forming fluidsand oxidized magma caused Cameron & Hattori(1987) to suggest that the fluids were derived from themagma. Data showing that introduction of gold sub-

Ftc. 7. Chondrite-normalized incompatible element patterns (spidergram) for rock unitsof the Murdock Creek pluton (from Rowins et al. 1993). Rb, K and P are normalizedto terrestrial abundances; "n" is number of samples averaged.

925

Alkali-feldspar syenltsMelasyenite (n=4)Clinopyroxenlte (n=5)Melamonzodlorlts &Meladlorite (n=6)Hornblendlte (n=3)

(n=1 2)

sm HfZr Tl Y Y b

stantially postdated magmatism indicate that this inter-pretation is incorrect; any link was indirect. The latearrival of gold also makes it difficult to invoke a sce-nario for the extraction of gold from the mantle, or itsmagmatic products, since the principal processesaffecting the mantle, i.e., metasomatism, melt extrac-tion, and devolatilization, were concurrent withmagmatism. Rowins et al. (1993) reported an averageof 2 ppb Au for 13 rock samples from the MurdockCreek stock that are unaffected by alteration. Wyman& Kerrich (1989) reported an average of 3.9 ppb Aufor fresh shoshonitic larnprophyres from the CanadianShield that are associated with structures tlat hostgold. Both sets of data indicate neither enrichment nordepletion of gold in magmatic products of the lateArchean mantle below major gold deposits.

Mclnnes et al. (n prep.) and Mclnnes & Cameron(1994) reported on mantle xenocrysts and xenoliths inbasanites and alkali olivine basalts from the Tabar -Lihir - Tanga - Feni (TLTF) ensimatic island arc offPapua New Guinea. The arc contains one of theworld's major gold deposits in the Luise caldera onLihir island. The nodules contain alkali-rich alumi-nosilicate glass, which represent a melt that reactedwith the host xenocrysts and xenoliths. The glass con-tains daughter minerals: anhydrite, calcite, sodaliteand fluorapatite. Analyses indicate that the melt con-tained substantial dissolved H2O, [email protected], SOf-, and hada high.(O). Such a melt will tend to dissolve sulfidesand the gold that is contained within the sulfides whenreacting with mantle material. If gold was extracted

+xIAo-

!adE

q t(x)q

oxo

oog

E l oo€()!o0E


during mantle metasomatism and partial melting, thencanied in a volatile phase to the surface, its depositionin the upper crust would be early and coincident withvolcanism, as it is on Lihir. The late arrival of gold atKirkland Lake and many other major Archeandeposits presents apuzz)e, as does the relatively highstate of oxidation of mantle-generated magma and orefluids at Kirkland Lake. A possible explanation lies inthe understanding of processes operating within thedeep shear-zone that lies between the mantle and thedeposits, as considered in Part m.

Panr III:A Dnsp SmAR-ZoNE. THE BAvBLE Srnan BBlr

The deep portion of the shear zone below KirklandLake is inaccessible. In searching for an analogueexposed elsewhere, there are fwo important features toseek. The first is evidence for the passage of CO2. AtKirkland Lake, there was a long period of CO2 flux-

ing. Given its apparent origin in the mantle, the CO,must have passed through the deep portion of theshear zone. The second feature is evidence for oxida-tive metamorphism. If gold at Kirkland Lake wasderived from the deep shear-zone, the oxidized natureof the ore fluids may denote a similarly oxidizedsotuce.

Grolocv oF rHE BAMBLE SHsan Bnr

The 30-km-wide Bamble shear belt of southernNorway (Fig. 8) formed as a result of transcurrentmotion between fwo major crustal blocks (Falkum &Petersen 1980). Metamorphism at upper amphiboliteand granulite grades was concurrent with ductiledeformation; strong deformafion is reflected in tightfolding with steeply dipping foliation (Falkum &Petersen 1980). Touret (1971) found abundant CO2-rich fluid inclusions in the granulite-facies rocks.Lamb et al. (1986) estimated granulite metamorphism

[ !q .o.f? "ssz*.'**S"Y*t

-"r*" \t!

l.*4 *rynpli{W Trmlwe

Tvedeetmnd

?r" I

d"',.*s-,o'',\; ry&,,"n e

n" -ror tu** p',L**u

se

fc

I

t

, 4( ^Q,& Tm$fr\&gY

a/"&

f{fEtrY

it /I

NUORWAY ,f / .

Frc. 8. Bamble shear belt. Metamorphic zones, after Field et al. (1980): A is upper amphibolite, B is a transitional facies withonhoplroxene itr I minefly of metabasite intrusions, C is granulite, D is granulite with less homblende ard biotite rhanzone C, and orthogneiss depleted in LILE. Shaded area contains enderbitic and chamockitic rocks.


FIc. 9. Photomicrographs of: (A) composite grain of pyrite and chalcopyrite, both with rim of magnetite, upper-amphibolite-grade metabasite, Tvedestrand, Norway. (B) Grain of pyrrhotite partially oxidized At right to a mixture of pyrite and mag-netite, coronitic gabbro interior of hyperite dyke, Tvedestrand. (C) Composite grain of pyrite and chalcopyrite, both withrim of magnetite, retrograded granulite, Canisp shear zone, Scotland. @) Back-scattered electron image of ilmenite grain

. .'from upper-amphibolite-gade metabasite, Tvedestrand, Norway. Broad lamellae of ilmenite (dark grcy) and hematite(light gxer, with fine secondary lamellae of these minerals. Thin white lamellae of magnetite lie mainly within broadlamellae of hematite.

to have occurred at 800 + 60'C and 7.3 -r 0.5 kbar.These conditions differ little from T and P estimatesobtained from fluid inclusions in upper-amphibolite-facies rocks (Touret & Dietvorst 1983), which sug-gests that the change from upper amphibolite toglanulite occurred at constant T and P, the result offluid composition changing from HrO-rich to CO2-rich (Touret 1971). The presence ofMg-rich pyroxeneand hematite-rich ilmenite indicates prograde meta-morphism at high/(Or) (Cameron 1989c, Harlovr99D.

The earliest rocks in the Bamble belt are sediments,overlain by a mixed volcanic - sedimentary sequence.Mafic dykes and sheets (metabasites) of tholeiiticcomposition are abundant. Tonalitic - trondhjemiticmagma (tonalite gneiss) was intruded in the coastalbelt of highest grade around tle island of Trompy.Metamorphism of all these rocks spanned the interval1152-1095 Ma (Kullerud & Dahlgren 1991).Anatectic granitic rocks are present in the upper

amphibolite zone. Granulites on Trom@y are depletedin LILE @eld & Clough 1976, Cooper &Field 1977,Clough & Field 1980). Mafic dykes and sheets (hyper-ites) were intruded in several stages subsequent to themain metamorphic event; they were moffied to coro-nitic gabbros and then locally amphibol' i t ized.Subsequently, medium-grained granite. was intruded,and there was localized retrogression of granulites atgreenschist grade at about 1060 Ma (Field et al.1985).During retrogression, Au, As, Sb, Cu, sulfide andLILE were re-introduced into the previously depletedgranulites (Cameron 1993).

Golr rN THE BAMBr-E Snpan Brr.r

Sampling of rocks of varying lithology from acrossthe width of the shear belt showed them to be substan-tially depleted in gold relative to its crustal abundance,both in the granulite and upper amphibolite facies(Cameron 1989a, c). For example, metabasite intru-

928 TTIE CANADIAN MINERALOGIST

5

4

3

2

gxlJ-

3c'{o

CDo

oCIo

LATE MA6N FTITE.ILMENITTMetoboslter I

Toncllte: IMoflc Bond ln Quqrtzlte: I - o,l

ffilt

T- IO

6ex

,r00 500 600 700

Temperoture "C

Frc. 10. Upper arrow is estimate of path of metamorphic cooling for Bamble metabasites inflO) - T space. Conditions at thepeak of metamorphism are derived from the intersection of an isopleth band for primary ilmenite with an isopleth fororthopyroxene. Itrnenite isopleths were derived from the data for four metabasite samples (two granulite facies, two upperamphibolite facies) computed by a program of M.S. Ghiorso (see Ghiono & Sack l99l). The orthopyroxene dat4 repre-senting the mean of six samples, were applied to the empiiical formulation of Fonarev & Grafchikov (1984). The coolingpath is located along the pyrrnotite - pynte - magnetite (PPM) buffer by point estimates for samples of metabasite (twofrom granulite facies, two from upper amphibolite), a tonalite (glanulite) and a rnafic band in metaquartzite (granulite). Thepoint estimates were obtained from the composition of lamellae of magnetite and ilmenite in ilmenite grains. [Note thatusing program QUILF (Andersen et al, 1993), the estimated temperature at which the metabasite samples intersected theppM 6uner is significantly higher, averaging 560'Cl. The lower arow is an estimate of the path taken during amphiboliti-zation of coronitic gabbro, obtained from a hyperite dyke near Tvedestrand (see Cameron et al. 1'993 for details). The PPMbuffer is from Kishima (1989), the FMQ buffer, from Myers & Eugster (1983).

0

sions contain only 0.04 of the usual abundance of goldin mafic rocks, i.e., a mean of 0.17 ppb Au versus 4,Ippb (Cameron 1989c). Other chalcophile elementscommonly associated with gold in veins, such as Sband As, also are depleted. Granulite samples fromTrompy are depleted in Cu, S and Rb, but these arepresent at near-normal levels in samples from theupper amphibolite zone.

During crystallization of mafic rocks, gold mainlyenters the sulfide phase; the partition factor, sulfidemelt/silicate melt, has been estimated to be in therange 1,000 (Stone er al. L990) to 10,000 (Barnes1993). Release of gold during metamorphism of maficrocks requires modification of this phase. Pyrrhotite isthe principal sulfide mineral in unmetamorphosedmafic rocks. In the Bamble metabasites, both from thegranulite and upper amphibolite zones, pyrite is thedominant sulfide mineral, accompanied by chalco-pyrite; grains of both minerals have invariably beenreplaced around their margins (mantled) by magnetite

800 900

(Frg. 9A). Fyrrhotite is virtually absent in these rocks,but must have been present during peak metamor-phism at -800 'C, given that pyrite is not stable aboveabout 740'C (Kullerud & Yoder 1959). As will be dis-cussed below, transformation of pyrrhotite to pyriteprovided an opportunity for the release of gold.

High-grade metamorphism in the Bamble shear beltresulted in an unusually oxidized assemblage of min-erals, with hematite-rich ilmenite and Mg-rich ferro-magnesian silicates. A typical grain of ilmenite frommetabasite is shown in Figure 9D. At peak metamor-phism, this was a homogeneous grain of hematite-richilmenite. During cooling, the ilmenite - hematitesolvus was intersected, causing phase separation intobroad lamellae of iknenite and hematite. Further cool-ing resulted in the exsolution of successive genera-tions of fine lamellae of ilmenite in hematite andof hematite in ilmenite. Finally, there was reduction ofsome hematite to magnetite, with magnetite lamellaeoccurring preferentially within broad lamellae of


hematite (Fig. 9D). Other granulites that containhematite-rich ilmenite include those of Proterozoic agefrom Lofoten - Vestralen, Norway (Scblinger 1985), alower crustal shear-zone of Archean age from LabworHills, Uganda (Nixon et al. L973, Sandiford er a/.1987), and granulites retrograded to amphiboliteuplifted along the Red Sea Rift (Seyler & Bonatti1988)

It is possible to reconstruct the path taken in"(Ot -T space during cooling from the peak of metamor-phism (Camerot et al. 1993). The composition ofprimary ilrnenite was obtained by re-integrating theexsolved ilmenite, hematite and magnetite componentsusing image analysis to estimate their proportions andresults of electron-microprobe analyses of the threephases. In Figure 10, an isopleth band representing thereconstructed compositions of primary ilmenite inmetabasite samples is shown intersecting an isoplethfor orthopyroxene [note that this figure and othersshow {/(Or) values, which is /(Or) relative to theFMQ bufferl. The itnenite isopleth band is derivedfrom two samples of metabasite from the granulitefacies and two from the upper amphibolite facies. Fiveorthopyroxene compositions are from the granulite

facies, the other from the upper amphibolite facies, butthe results from both facies are s,imilar. The intersec-tion of the isopleths defines theflO) - T where oxidesand ferromagnesian silicates equilibrated during peakmetamorphism. This intersection brackets 800'C, inagreement with the independent est:mates cited abovefor the temperature of peak metamorphism.

During cooling from peak metamorphism, theilmenite - hematite solws was intersected at approxi-mately 740"C (Cameron et al. 1993). In grains ofilmenite, this resulted in the exsolution of broad lamel-lae of ilmenite and hematite. as described above.During continued cooling down the solvus, furthergenerations of these minerals were exsolved as fineIamellae. The formation of magnetite lamellae inilmenite grains by reduction of hematite occurredlater. Compositions of these magnetite lamellae andadjoining iknenite provide es :mates of theflO) and Tat which the phases reached equilibrium. These valuescluster along the pynhotite - pyrite - maguetite (PPM)buffer @9. 10). Thus the cooling path was subparallelto ttre FMQ buffer and intersected the PPM buffer at atemperature below 600'C. Magnetite - ilmenite pairsin ilrnenite grains from a sample of tonalite gneiss and

1

0.7

0.5

Ftc. 1 l. Spidergram plot for elements in the amphibolite marginal zone (pyrite-dominant) and in the coronitic gabbro intezone (pyrrhotite-dominant) of hlperite dyke 4O from near Tvedestrand. Data represent the mean of two samples forzone, normaliz€d to the mean composition of sixteen metabasite samples from Tromgy. The metabasites represent satthat are strongly depleted in chalcophile elements and LILE from the high-grade metamorphic event.

10

3

UJF-ao(L

oou.!Fct)mFuJ

o$

l,.rJ:<o

K Sr La Sm Tb Lu Ti P Co Zn As Au


from a thin mafic band in qvartzite, both from thegranulite facies, also plot close to the PPM buffer (Fig.10). Clustering ofthese points around the PPM buffer

can be accounted for if the conversion of pyrrhotite topyrite was a closed-iystem oxidation reaction, wherethe complementary reduction of hematite to magnetite

PRECAMBRIAN GOLD AND SHEAR ZONES 93r

FIc. 12..A. Grain of pyrrhotite @o) partially transformed to pynre (Pj) and magnerite (M0. From sample 4008, coronitic gab-bro-interior ofhyperite dyke 40, Bamble shear belt. B. Same grain as in A after spot analyses by ion microprobe; results of3nalls9s are shown in ppb gold. C. Grain of pyrrhotite @o) containing inclusions of chalcopyrite (Cp) and pentlandite (pn);Lewisian granulite sample 1604, Kylestrome. The pyrrhotite was partly converred to pyriie (py)wlttr void space (V), thelatter now filled by chlorite and calcite. Circles indicate the position of spots analyzed for goid by ion microprobe. Datashown in parts per billion Au.

served as the 02 donor:6 FeS + l2FqO3+ 3 FeS, + 9 FerO*.

The cooling path from 800'C, being subparallel to theFMQ buffer, is also indicative of a closed system, withthe path determined by the oxidation state of the min-erals formed during prograde metamorphism. Giventhe lower blocking temperature of oxide minerals rela-tive to silicates (Frost 1991), the cooling patl wasmainly determined by the composition of the oxides(Frost et al. 1988, Frost 1991). Thus the hematite-richnature of the primary iknenite was important in deter-mining the cooling path (Cameron et al. 1993); thehematite-rich ilmenite was a product of progrademetamorphism under oxidizing conditions. The reasonwhy prograde metamorphism was oxidizing in nature,an observation confirmed by Harlov (1992), is notclear. One possibility is that pervasive COr-bearingfluids acted as an oxidant. However, the petrologicalstudies required to find the causes of oxidative meta-morphism in the Bamble belt and similar terranes,such as Labwor and Lofoten - Vestralen" have vet tobe attempted.

Rocks of the Bamble belt metamorphosed at a highgrade represent the end stage ofprocesses that resultedin severe depletion of gold, to below its averagecrustal abundance. The nature and timins of these

processed can only be surmised. However, the coolingpath shown in Figure 10 provides a realistic mecha-nism, witl tle transformation of pyrrhotite to pyritemaking gold accessible to later fluids. Transformationcould only have taken place a substantial time afterpeak metamorphism, when the belt had cooled from800'C to below 600'C. After pyrite was formedo therewas partial dissolution of pyrite and chalcopyrite,which are mantled by magnetite Grg. 9A). This pro-vides a mechanism for the removal of Au+ in a meta-morphic fluid, complexed by dissolved sulfide(Cameron 1989a c).

EvtonNce FRoM HypERrrEs

Whereas interrnediate stages in the process for theextraction of gold are not found in the high-graderocks, evidence is preserved in the hyperites, whichwere intruded after peak metamorphism, then modi-fied to coronitic gabbros and partially amphibolitized.One dyke of hyperite examined shows a transitionfrom coronitic gabbro in the interior to foliated amFhi-bolite at the margin. Samples taken from the interiorzone, through an intermediate zone, to the amphibolitemargin show a progressive decline in contents of goldand related chalcophile elements, whereas other ele-


SECONDS

C'Fz= 10roO

6 l 0 'zl!F

Z .|00

ct)Fz 10"=oO

E ro'(tztltt-1 ...,

SECONDS

Frc. 13. A. Ion beam in-depth profile of concentrations through pyrrhotite from sample 4008. This 1nofile averages 330 ppbAu. B. Comparative p.ofiles showing similar (background) values for ELBA pyrite (5 ppb Au) and the pyrite reactionproduct shown in Figure 12.

ments remain relatively constant (Fig. 1l). Corres-ponding to these chemical changes is the replacementof a pyrrhotite-dominant sulfide assemblage in thegabbroic interior by a pyrite-dominant assemblage inthe marginal amphibolite. Some grains of pyrite in the

latter are mantled by magnetite, replicating the grainsfound ubiquitously in lithologies that experiencedpeak metamorphism.

To substantiate the role of the pynhotite + pyrite +magnetite transformation in the loss of gold, Cameron


& Chryssoulis (submitted) have measured the goldcontent of the three phases using an ion microprobe.Figure 12A shows a grain of pyrrhotite that is partlytransformed to pyrite and magnetite, from sample4008 in the interior of hyperite dyke 40. Figure 12Bshows the same grain after analysis by the ion micro-probe. The circular pits represent material eroded bythe ion beam during the analysis of individual spotsacross the gtain surface. At each spot, the ion beamprovides an in-depth profile of concentrations asmaterial is eroded. A typical profile for pyrrhotite(Fig. 13A), with broad and narrow oospikes", showsthat gold is not uniforrnly distributed as a solid-solu-tion impurity, but is present as colloidal-size particlesof gold or gold-rich material. For all spot analyses,profiles from the surface to a depth of 1.3 pm (-500seconds) were taken, and the results averaged. Thepresence of particulate gold effectively increases theanalytical sensitivity, since a narrow spike will beaveraged over the whole profile. For example, asingle-point spike to 100 ppb Au, which can readily bedetected, amounts to 7 ppb Au if averaged over15 points. The inhomogeneous distribution of the goldalso is shown by the variation in the results of spot

analyses across the pyrrhorite grain (Fig. l2B); elevenanalyses gave from 25 to 430 ppb Au, with a mean ofl42ppb.In total, forty-four spot analyses of pynhotitefrom the interior of the dyke averaged 136 ppb Au,and eleven analyses ofchalcopyrite averaged 125 ppb.These values compare well with 11 1 ppb Au reportedby Boyd et al. (1988) for a bulk sample ofpyrrhotite,chalcopyrite, pyrite and pentlandite from the Erteliendeposit, one of several deposits of Fe-Ni-Cu sulfidessegregated from hyperite magmas in southernNorway. Bulk analyses of sample 4008 gave, on aver-age, 0.8 ppb Au, which likely represents a decreasefrom the primary content, given the partial fiansforma-tion of pyrrhotite to pyrite. Sample 4001 from themargin of the amphibolite dyke contains 0.3 ppb Au,compared to an average of 0.16 ppb for 16 samplesfrom nearby metabasite dykes at Tvedestrand.

Six analyses were made of the pyrite portion of thegrain shown in Figure 12B, none of which detectedgold. Unlike pyrrhotite, there is a weak molecularbackground for pyrite (FeSr) from CsS;, with thesame mass as gold (Cs is from the primary ion beam).Background was established by analyses of ELBApyrite, containing 5.2 ppb Au. In-depth profiles of

TROMOYGRAN U L IT , .

o

a

o

o

a.

f,

COU R IEDYKES

L E W I S I A N' vvY

v

v

G RAN U LITE

o. l 0.3 0.5 o.t r .o 3 5

Au, ppbFrc. 14. Curves showing cumulative frequency distribution curves of Au (ppb) for Trompy granulite (n = 67), Lewisian

granulite (n = 33) aad Scourie dykes (n = 39) . Median for each disuibution shown by arrow.

99

90

70

50

30

t0

F

J

o

moo-

LU

=

J

:)O

934 ITIE CANADIAN MINERALOGIST

concentration for all analytical spots on pyrite arebackground values s imi lar to those in ELBA(Fig. 13B); what is most significanr is rhar profileslack the spikes that reveal particulate gold in the adja-cent pyrrhotite. The results indicate that essentially allgold is lost when pyrrhotite changes to pyrite. No gold(<10 ppb) was detected by subsequent analyses ofthemagnetite reaction-product shown on Figure 12.

The path taken in flOr) - T space by the hyperitedyke to reach the PPM buffer is difficult to define,because magnetite, that might be used with ilmenite toestimate this path, continued to re-equilibrate to lowtemperatures. However, reconstructed primary compo-sitlons of ilmenite (Cameron et al. 1993) indicate thatthe coronitic gabbro crystallized close to the FMQbuffer. Pyrite formed in the margin of the dyke underamphibolite-facies conditions. The approximate pathconstrained by these observations is shown in Figure10. It implies an increase inflO2) relative to the FMQbuffer, different from that of the metabasites, whichcooled subparallel to the FMQ buffer. Whereas thecooling path for the metabasites was internallybuffered by the assemblage of oxidized mineralsformed during peak metamorphism, the increase in{flOr) during amphibolitization of the margin of thehyperite dyke may have been externally buffered bythe fluid that caused amphibolitization. In effect, rhefluid would have equalizedflOr) conditions betweenthe dyke margin and the host rocks previously meta-morphosed at high grade under oxidizing conditions.

CoupanrsoN wmt rrn LewrsrAN CoMpLEX

The Lewisian granulite complex in northwesternScotland is one of the world's most strongly LILE-depleted exposures of lower crust. Lewisian gran-ulites, which are orthogneiss, subhorizontally bandedinto mafic material of tholeiitic affinity and felsicbands of tonalitic and trondjhemitic composition, haveonly 0.04 of the crustal abundance for U and 0.07 forTh (Weaver & Tamey 1984). Peak metamorphism forthe Lewisian occurred at2.66 Ga (Pidgeon & Bowes1972), 940-990'C and I 1 kbar (Cartwright &Barnicoat 1987). From about 2.6 to 2.3 Ga, there wasthe Inverian period of deformation and retrogressionby fluids introduced under amphibolite-facies condi-tions, which was most severe along several shear-belts. This was followed at 2.4 and 2.0 Ga by majorinjection of the Scourie mafic dykes.

Figure 14 shows comparative data on gold concen- DtscussroNtrations in granulite-facies rocks from Trom@y in theBamble belt, for Lewisian granulite from the Drumbeg At Kirkland Lake, the formation of large deposits- Kylestrome area, and for the Scourie dykes near ofgoldwasnottheproductofregionalmetamorphismKylestrome. The Trom@y suite consists of 16 samples or magmatism, but the result of localized crustalof metabasite and 5l of tonalite gneiss; the Lewisian extension, then uplift, along a major fault-zone. Thegranulites are represented by 33 samples of banded same broad tectonic pattern of extension followed byorthogneiss, and the Scourie dykes, by 39 samples of uplift can be recognized at a deeper level within thequartz dolerite. Lewisian granulite is depleted in gold Bamble shear belt. Metamorphism in an extensional

relative to its crustal abundance of about 4 ppb(Anderson 1983), but not as severely as the Tromgysamples. Like the Bamble rocks, the sulfides presentin the l,ewisian granulites are mainly pyrite with somechalcopyrite; both minerals commonly have a thin orpartial mantle of magnetite. Peak temperatures ofmetamorphism above 900"C for the Lewisian gran-ulites were well above the stability limit of pyrite.Although obvious zones ofretrogression were avoidedduring sampling, most Lewisian samples show someeffects. Oxide minerals were particularly sensitiveto the effects of retrogression; ilmenite was changed tofine-grained mixtures of magnetite and titanite orrutile. Changes affecting the oxide and sulfide miner-als are most severe within the shear zones. Figure 9Cshows a composite grain of pyrite and chalcopyritefrom the Canisp shear-zone with a well-developed rimof magnetite, similar to those of the Bamble rocks.

Pyrrhotite also is present in the kwisian granulites,but only in a few thin sections that lack al1 evidence ofretrogression, further suggesting that the formation ofpyrite was the result of the fluids that caused Inverianretrogression. These fluids were oxidizing in nature(Beach & Fyfe 1972, Beach & Tarney 1978, Sills1983). The Scourie dykes, which were emplaced afterInverian rerogression, but while the terrane was stilldeep in the crust, retain a pyrrhotite-dominant sulfidemineralogy, and a gold content that is significantlyhigher (average 1.5 ppb Au) than the pyrite-dominantTromoy granulite (mean 0.22 ppb Au) and Lewisiangranulite (mean 0.54 ppb Au).

Figure 12C shows a grain of pyrrhotite fromKylestrome with inclusions of chalcopyrite and pent-landite. The pynhotite has, in part, altered to pyrite.Ion-microprobe analyses (Cameron & Chryssoulis1993) of six spots in pyrrhotite range from 400 to9,200 ppb Au, averaging 2,200 ppb Au, substantiallyhigher than the pyrhotite from the Bamble hyperitedyke. Intensity profiles show that, like the Bamblepyrrhotite, gold is present as fine, irregularly dis-tributed particles. Two analyses of a chalcopyriteinclusion gave an average of 1,200 ppb Au; gold alsois present in this mineral as fine particles. In contrastto the sharp peaks in the intensity profi les forpyrrhotite, indicative of particulate gold, profiles frompyrite are relatively flat. Analyses of pyrite gave, onaverage, 224 ppb Au, an order of magnitude less thanin pyrrhotite.

PRECAMBRIAN GOLD AND SHEAR ZONES

Flc. 15. Conceptual model for north-south section of lithosphere a[ Ktkland Lake.

935

setting occurs at high T and low P, and is termedoodiastatlermal" by Robinson & Bevins (1989). Hightemperature derives from uprise of mantle and dec-tion of mafic melts into the base of the crust.Modeling indicates an counterclockwise P - T - t path(Robinson & Bevins 1989), a consequence of heatingbefore loading, rather than the clockwise path ofregional burial metamorphism, which results fromloading before heating (Bohlen 1987). After extensionstops, there is a substantial period of isobaric cooling(Sandiford & Powell 1986). The Bamble belt con-forms to this model; I(hle (1989) has established acounterclockwise P - T - t path, with isobaric coolingat about 8 kbar from 800 to about 540"C, followed byisothermal uplift. Major injection of mafic magmas(now represented by metabasites and hyperites)occurred before, during and after peak metamorphism.These magmas were of mainly tholeiitic nature, sug-gesting that extension was greater at Bamble than atKirkland Lake, where magmatism was mainly alka-line. At Kirkland Lake, the subgreenschist grade ofmetamorphism of the Timiskaming sequence is conso-nant with that for diastathermal metamorphism of anextensional basin filI (Robinson & Bevins 1989). TheLabwor Hills shear belt, Uganda, is an Archean ana-logue of the Bamble belt. Metamorphism appears tohave been more strongly oxidizing, near the hematite

- magnetite buffer, with hematite dominant overilmenite in intergrowths of these minerals (Sandifordet al. 1987). Peak metamorphism occurred at a hightemperature, 850-1000'C, which Sandifotd et al.(1987) attributed to heat from the mantle duringcrustal extension and thinning. This was followed byisobaric cooling at 7 -9 kbar.

An idealized lithospheric cross-section throughKirkland Lake (Fig. 15) shows a broad shear-zone atdepth, perhaps analogous to that of the Bamble shearbelt. Broad zones of deformation may be associatedwith both vertical (strike-slip) and low-angle (thrust ordetachment) faults. Knowledge of the processes ofdeformation along strike-slip faults has come fromfield observations, such as those of Bak et al. (1975)and Hanmer (1988), and from models based on rheo-logical properties of rocks, such as those of Yuen e/ al.(L978), Turcotte et al. (198O) and Lockett & Kusznir(1982).Initially, movement along a major fault is con-fined to a narow zone. Within this zone, frictionalheating increases temperature over ambient con-ditions, reducing the viscosity of the rocks and facili-tating ductile deformation. With time, frictional heatspreads out, lowering the viscosity within a broaderzone. This broadening increases with depth as theambient temperature increases. Since the velocity offault movement determines the frictional heat gener-

HuI

Cot

RLI{RLIN€ MCLTS

11CRBBONRT€D MEITS


KIRKIAND LAKE

TRANSPRESSION

IMRODUCTION OFGOLD lN CO"-H.oFLUIDS

UPLIFT

FAULT ZONE

IRANSTENSIONBASIN FORMATIONTIMISKAMING DEPOSIT

MANTLE UPWELLINGCARBONATIZATIONALI(ALINE MAGMATISM

BAMBLE SHEAR BELT

500 600 700 900

TEMPERATURE, OC

Ftc. 16. Pressure - temperature - time (P - T - t) path of Bamble shear belt (Kihle 1939) that is characteristic of a deep crustalresponse to extensional tectonics. Superimposed is isochore for CO2 inclusions formed during peak metamorphism.lnclusions will be at less than lithostatic pressue during isobaric cooling, but will become overpressured during uplift,causing the release of CO2. The diagram is plotted with pressue increasing downward, opposite to the usual practice; thusthe P - T - t path appears as clockwise. The top part of the diagram shows cornparative events at Kirkland Lake that arerelated to an upper cnrstal response to extensional tectonics.

2

4ao\<uie.)act)t!u.o-

8

10800400

ated, the rate of broadening is greatest for highvelocities of displacement. Broadening is facilitatedby the heating generated by uplift of manrle in anextensional terrane.

Rutter & Brodie (1985) observed that under condi-tions of medium- to high-grade static metamorphism,permeability is low, sufficient ro permit a fluid phaseto permeate only a few meters per Ma. Increasedpermeability during ductile deformation is caused bydilatancy related to intense intergranular and trans-

granular microcracking. Dilatant flow occurs underhigh differential stress. Dilation is also favored byhigh fluid pressure, notably at negative effective pres-sure. Experiments by Spien & Peach (1989) on defor-mation of rock salt in the dilatant field showed anincrease in permeability of 5 orders of magnitude overa range of strain from 0 to 5Ea.

CO2-rich high-density inclusions of peak metamor-phic origin in Bamble rocks (Touret l97l) are over-whelmingly predominant in the granulite terrane of

PRECAMBRIAN GOLD AND S}IEAR ZONES 937

Trom0y (Touret 1986). In the transition zone betweenthe granulite and amphibolite facies, H2O-rich inclu-sions are common, but COt-rich inclusions ale presentin the amphibolite facies near mafic intrusive rocks.Touret (1986) suggested that mafic magma was avector for introduction of CO2 of mantle origin. CO2-rich inclusions are common in many granulite terranesand are not necessarily derived from the mantle, butmay simply be the residue from dehydration by partialmelting (Fyfe 1973). An important topic for researchis to determine whether CO2 fluids played a part in thetransformation of Bamble rocks into a relatively oxi-dized assemblage during prograde metamorphism,which, in turn, determined the closed-system coolingpath that crossed the pyrrhotite - pynte phase bound-ary. This is of consequence in the Kirkland Lake arca,where a long period of CO2-fluxing in the KLF hasbeen alluded to in an earlier section. Rowins et c/.(1991) found that the magma for the Murdock Creekstock was "dry", consistent with dehydration by theevolution of a CO2 - H2O phase during ascent. ACO2-rich silicate melt will tend to evolve a CO2-H2Ophase on rising because of the limited solubility ofCO2 and its pressure dependence (Holloway 1976).

In Figure 16, events in the Bamble shear belt rele-vant to the loss of gold are related to the metamorphicP - T - t path of Kihle (1989). Based on the citedfluid-inclusion data, CO, fluids were present at thepeak metamorphic conditions. During isobaric coolingfrom 800'C to about 540'C" conditions were unfavor-able for fluid migration. Fluids rapped as inclusionsduring peak metamorphism would likely have beenpreserved, since lithostatic pressure during isobariccooling was gleater than the internal pressure of theinclusions (Touret 1992).Lack of fluid during thisphase is supported by the cooling path in flO2) - Tspace (Fig. 10) that was internally buffered by themineral assemblage. On crossing the PPM buffer, at atemperature of 600'C or below, oxidation of pyrrhotiteto pyrite also took place in a closed system, with thereaction being balanced by reduction of hematitewithin ilmenite grains. The transformation of primarypynhotite made gold available to extraction by a laterfluid phase.

The following phase of isothermal uplift providedconditions conducive to the out-migration of fluids.Inclusions preserved through isobaric cooling wouldhave been released when their internal pressure

aLL+

N

o\-

ooJ

oc)o

400 500 600 700 800 900

Frc. 17. Diagram showing paths taken infO2) - T space during cooling from metamorphic peak where the rocks are bufferedby oxide - oxide re-equilibration. "A" is an oxidized assemblage, similar to that of the Bamble or Labwor shear belts.Where iknenite is the only oxide, cooling will be along the isopleth for hematite in ilmenite. For a mag'netite-only assem-blage, cooling will be along the ulviispinel in magnetite isopleth. Mixtures of ilmenite and magnetite result in an intermedi-at€ path. However, ilmenite is well buffered with 0.30 mole fraction hematite, relative to 0.16 mole fraction ulvdspinel inmaguetite. Thus, except in rocks with magnetite in large excess over ilmenite, the cooling path is likely to cross into thefield of stability of pyrite. The reduced assemblage "B" is poorly buffered with respect to hematite and is unlikely to coolinto the pyrite field. Based on Frost et al. (1988) and Frost (1991). The graphite buffer at 3 kbar is from Frost & Chacko(1989); the pyrrhotite - plrite - magnetite buffer is from Kishima (1989). Oxide isopleths were computed by a program ofM.S. Ghiorso (see Ghiorso & Sack 1991), the FMQ buffer, from Myers & Eugster (1983).

4

2

0

-t

G/Pryry

938

exceeded the decreasing lithostatic pressure. Whereasthe change from pyrrhotite to pyrite during isobariccooling did not require a fluid oxidant, the subsequentpartial replacement of pyrite and chalcopyrite by mag-netite required both an oxidant and a fluid to removethe sulfide released by the reaction. Although there isno direct evidence as to when gold was removed fromthe Bamble belt, this fluid seems the most probableagent, with the sulfide acting as a ligand to iomplexAu+. The same fluid may have been responsible forthe partial amphibolitization of the hyperites. Finally,there was variable retrogression of the uplifted gran-ulite terrane under greenschist-facies conditions, withre-introduction of gold into previously depleted gran-ulite (Cameron 1993). The time from peak metamor-phism until the start ofuplift and possible loss ofgoldcan be approximated. Geochronological data citedearlier for the termination of peak metamorphism(-800"C) at 1095 Ma and for retrogression of gran-ulites under greenschist conditions (-350'C) at 1060Ma give an approximate cooling rate of 8 Ma per100'C. or 21 Ma to cool from 800 to 540"C.

Rapid uplift results in the compression ofisotherms; as a result, the brittle - ductile transitionmay rise to within 6 to 8 km of the surface (Holrn eral.1989). Thermal gradients are highest near the sur-face. Koons (1987) estimated that rock uplifted from25 km would have a 10'C&m gradient during the fust10 km of uplift, 20"Clkm for the next 10 km, and70'C/km for the final 5 km. The high thermal gradientin the upper crust favors convective flow of fluid(Koons & Craw 1991). Norris & Henley (1976)showed that in metamorphic belts with a thermalgradient greater than l2"Clkm, uplift will result in theexpansion of metamorphic watero with resultinghydraulic fracturing. This gives rise to overpressuredfluids at depth migrating upward to mix with con-vecting fluids above the brittle - ductile transition.Sibson et al. (1988) suggested that high-angle reversefaults acted as valves to release overpressured(supralithostatic) fluids from below into a hydrostaticregime, where gold was deposited. The ore at KirklandLake was largely deposited along a high-anglereverse fault during a major phase of uplift. AtKirkland Lake, fluids that scoured gold from the lowercrust were focused into the narrow, upper part of theshear zone.

An important conclusion of previous sections is thatconditions of deep crustal metamorphism should beconducive to the modification of sulfide minerals thatare the principal host of gold. In mafic rocks, whichare an important component of the lower crust, thesulfide fraction amounts to about 0.25 wt.%o. of whichpyrrhotite is the most important mineral. In both theLewisian granulites and the Bamble shear belt, it ispyrite, not pyrrhotite, that is now the dominant sulfidemineral. Ion-microprobe analyses of samples fromboth areas show that the metamorphic transformation

of pyrrhotite to pyrite resulted in a loss of gold by afactor of 10. This occurred well after peak metamor-phism, when the rocks had cooled below 600"C. In theBamble belt, the silicate minerals that forrned duringpeak metamorphism were retained in unmodified formduring cooling, whereas oxide and sulfide mineralscontinued to re-equilibrate long after the peak.

Modification of the minor sulfide fraction, withoutinvolvement of silicate phases, required less fluid thanotherwise and avoided dissolution of major compo-nents within the rock. The oxide minerals influencedthe modification of the sulfide fraction. Frost et al.(1988) and Frost (1991) suggested that thefiO) - Tpath taken after peak metamorphism is largely deter-mined by the nature and composition of the oxides.For rocks where ilmenite is the only oxide, the path ofcooling will follow the isopleth for ilmenite, whichcan result in a moderate increase in $O) (Fig. 17).Where magnetite is the only oxide, the path willfollow the appropriate isopleth for ulvdspinel in mag-netite, which results in a decrease in ffO) (Frg. 17).For mixtures of magnetite and ilmenite, an inter-mediate path is obtained. Along this intermediate path,the hematite content of ilmenite and the ulvdspinelcontent of magnetite both decrease, reflecting thebalanced redox reaction and the exchange of compo-nents between the two oxide minerals:

Fe2TiOa + Fe2O3 = FeTiO: + F%O4.The Bamble metabasites contain an oxide assem-

blage that is particularly favorable to a cooling paththat crosses the pyrrhotite - pyrite phase boundary. Inthese rocks, ilmenite is the dominant oxide phase(Cameron et al. 1993); ilmenite forrned during peakmetamorphism was well buffered with respect tohematite, with 0.30 mole fraction of the latter. Thereduced assemblage ooB"

@ig. 17) is poorly bufferedwith respect to hematite and is unlikely to enter thefield of stability of pyrite. It appears that many gran-ulite terranes took a path that decreased in {flO), tointersect the graphite saturation surface (Frost &Chacko 1989). Sedimentary sequences containinggraphite also will buffer the rocks to low{O). Thismay bear on the preferential concentration of majorArchean gold deposits in metavolcanic rather thanmetasedimentary belts.

For the Lewisian granulites, conversion of primarypyrrhotite to pyrite and tle associated loss of goldinvolved the introduction of oxidized fluids duringretrogression. The fluids not only modified the sulfidefraction, but caused change in the oxide and silicateminerals. The effects of the fluids were more pro-nounced along shear zones.

Ages for gold veins in the Yilgarn Block, Austra1i4fall within a tight interval of 2660 to 2630 Ma (Groveset al. 1990), which is similar to the ages cited abovefor veins in the Abitibi belt. The association ofdeposits with major faults, which, in some cases, havebeen shown to be terrane boundaries, suggests that


gold was related to a worldwide episode of terraneassembly at the end of Archean time. The majorstrike-slip movements required for terrane assemblyproduced broad zones of ductile deformation in theIower crust, to serye as the large-volume, permeablesource-regions for gold. The observation of Libby elal. (1990), of lamprophyre and porphyry intrusionsalong the major faults in the Yilgarn Block, suggeststhat localized extension, with upwelling and meltingof mantle, was widespread during terrane assembly.When assembly was completed, the extended areaswere uplifted. Permeability was then high, as fluidsbecame overpressured.

CoNCLUSIoNS

For a mafic lower crust, a precondition for extrac-tion of gold is modification of the sulfide fraction, theprincipal host for gold. Pyrrhotite is the main primarysulfide mineral and the stable form of iron sulfideduring high-grade metamorphism. The phase changeof pyrrhotite to pyrite, which occurs during metamor-phic cooling, and which requires a relatively highfiO), is required for the extraction of gold. Both forthe Bamble shear belt and Lewisian granulite, ion-microprobe analysis has shown a factor of ten loss ingold during the metamorphic transformation ofpynhotite to pyrite.

Prograde metamorphism of the Bamble shear beltunder conditions of tle upper amphibolite and gran-ulite facies produced a mineral assemblage that wasrelatively oxidized. During cooling from the metamor-phic peak, buffering by this assemblage, notably byhematite-rich ilmenite, took the rock into the stabiliryfield of pyrite. Silicate minerals retained their peakmetamorphic composition, contrasting with sulfideand oxide minerals, which continued to be reactive tobelow 600'C. Since the latter are only small fractionsof the total rock, gold was, in effect, selectivelyextracted from a large volume of mainly unreactiverock, requiring lesser amounts of fluid.

The Bamble shear belt is perhaps analogous to thelower crust below the Kirkland Lake gold camp. Bothat Bamble and Kirkland Lake. extensional tectonicsappeared to play a key role in the mobilization andhansfer of gold. Rocks metamorphosed at high gradein an extensional terrane rypically follow a counter-clockwise P - T - t path. Isobaric cooling is followedby near-isothermal uplift. Fluids present as inclusionsduring peak metamorphism will tend to be retained byhigh lithostatic pressure during isobaric cooling. Butwith uplift, and after phase change in the sulfideminerals, the fluids become overpressured and arereleased. This is the probable time when gold wasextracted, carried upward, and focused at the brittle -ductile transition. Fluids present, then expelled, fromthe deep crust were usually CO2-rich, but the CO2 maybe of diverse origin: CO, derived from magma of

mantle origin, and CO2 of crustal origin that accumu-lated as a consequence of dehydration reactions.

Kirkland Lake displays feafures of extensional tec-tonics at a higher level. Here, extension was a conse-quence of motion along a major, long-lived, linearfault, interpreted to be sftike-slip. Extension causedformation of basins, in which Timiskaming sedimentswere deposited, and upwelling of mantle, accompa-nied by CO2-degassing and melting to LllE-enrichedalkaline compositions. High heat flow, together withthe long-lived nature of displacement along the fault,would have favored tJre formation of a broad, ductileshear-zone in the lower crust that served as a sourcefor gold. Localized magmatism and metamorphismalong the fault zone occurred during extension. It waslater, during uplift" that gold was deposited alonghigh-angle reverse faults. The late arrival of gold inthe upper part of the shear zone is in conformity withthe evidence from the Bamble shear belt suggestingthat gold was extracted during rapid uplift followingisobaric cooling.

The Bamble scenario, for metamorphic oridation ofa large volume of lower crust, is most favorable forextraction of gold. If the primary composition of theBamble rocks was close to the crustal average of 4 ppbAu (Anderson 1983), a 1O-km-long segment of thedeep shear belt, 5 km deep and 30 km wide, couldhave released and focused 15,000 t Au into the uppegnarrow portion of the shear zone. This compares withthe 723 t Au produced from Kirkland Lake.Widespread metamorphic oxidation in the Bambledeep shear zone may relate to the finding (Cameron &Hattori 1987, Mikucki & Groves 1990) that several ofthe largest Archean gold deposits, including thosefrom Kirkland Lake. were formed from oxidized flu-ids. The demonstrated high mobility of gold in thedeep crust suggests that a variety of tectonic condi-tions (and thus genetic models) may be able to exploitits mobility.

Both in Australia and Canada, the formation of golddeposits in late Archean time occurred during a periodof terrane accretion that formed the cratons existingtoday. The major strike-slip movements implicit in ter-rane accretion produced the conditions required for thegeneration of gold deposits described here. These arebroad zones of deformation in the lower crust: local-ized extension gave rise to mantle upwelling withmagmatism, CO2-degassing and heating of the lowercrust, followed by rapid uplift, with release of storedfluids and the extraction and concentration of gold.

AcrNowr-poceN4nNrs

Work in the Kirkland Lake area has been carriedout with colleagues at the University of Ottawa: BillCarrigan, Keiko Hattori, Andr6 Lalonde, GuyL6vesque and Steve Rowins, and work in the Bamblebelt, with Stephen Chryssoulis, Ersen Cogulu and


John Stirling. The Bamble studies were made easierby the hospitality and cooperation of many Norwegiangeologists, particularly Ame Bj@rlykke, Jan Kihle andPeter Padget. The deep crustal studies would havebeen impossible but for the skill of Gwendy Hall, whodeveloped a highly sensitive method for the determi-nation of gold. Keiko Hattori, Neil Phillips, FrangoisRobert and Graham Wilson all reviewed the manu-script with care. I had previously received constructivecomments on interpretations of the tectonics of theAbitibi belt from Paul Hoffman and on the metamor-phic petrology of oxide and sulfide minerals from RonFrost. Robert Martin substantially improved the clarityof the text. I am indebted to all of these persons, as Iam to those who supported the studies at the GSC, andto NSERC, who helped fund the studies at theUniversity of Ottawa.

RnrsnsNcEs

ANDmsEN, D.J., Lnnslrv, D.H. & DAvDsoN, P.M. (1993):QTIILF - a PASCAL program to assess equilibria amongFe-Mg-Ti oxides, pyroxenes, olivine and quartz.Comput and Geosci. (in press).

ANornsoN, D.L. (1983): Chemical composition of the man-tle. J. Geophys. ftas. 888,41-52.

BAK, J., SORENSnN, K., Gnocor:r, J., KoRsrcARD, J.A., Nass,D. & WATTERSoN, J. (1975): Tectonic implications ofPrecambrian shear belts in western Grcenland. Nature2s4,566-569.

Banres, S.J. (1993): Partitioning of the platinum g oup ele-ments and gold between silicate and sulphide magmas inthe Munni Munni Complex, Austral ia. Geochim.Cosmochim Acta 57. 1,n 7 -1290,

Besu, A.R., Goopwnr, A.M. & TArsuMoro, M. (1984):Sm-Nd study of fuchean alkalic rocks from the Superiorhovince of the Canadian Sttreld. Eanh Planet. Sci. Izx.70.4046.

Beacn, A. & FvFE, W.S. (1972): Fluid transport and shearzones at Scourie, Sutherland: evidence of overtlrusting?Contrib. Mineral. Petrol. 36" 175-180.

- & Teruwv, J. (1978): Major and trace element patternsestablished during retrogressive metamorphism of gran-ulite-facies gneisses, NW Scotland. Precambrian Res.7,325-348.

Boll, K., ANcLtr{, C.D. & FnaNrrw, J.M. (1989): Sm-Ndand Rb-Sr isotope systematics of scheelites: possibleimplications for the age and genesis of vein-hosted golddeposits. Geology 17, 500-504.

BEN-AVRAHAM,2., Nun, A. & Celr-o, G. (1987): Activetranscurent fault system along the north African passivemaryln Te ctonophy s. l4l, 249 -260,

BnN Orrnaan, D., Amor, N.T., WHnE, W.M. & JocHUM,K.P. (1990): Geochemistry and age of Timiskaming alka-li volcanics and the Otto syenite stock, Abitibi, Ontario.Can. J. Eanh Sci. n. n04-l3ll.

Bortr EN, S. (1987): hessure - temperafue - time paths and atectonic model for the evolution of granulites. J. Geol. 95,617-632.

BoyD, R., BanNss, S.-J. & GRoNLIE, A. (1988): Noble metalgeochemistry of some Ni-Cu deposits in theSveconorwegian and Caledonian orogens in Norway /nGeo-Platinum 87 (H.M. Prichard, P.J. Potts, J.F.W.Bowles & S.J. Cribb, eds.). Elsevier, London (145-158).

CAMERoN, E.M. (1989a): Scouring of gold from the lowercrust. Geology 17, 26-29.

- (1989b): Derivation of gold by oxidative metamor-phism ofa deep ductile shear zone. 1. Conceptual model.J. Geochem. hpL3l, 135-147 .

- (1989c): Derivation of gold by oxidative metamor-phism of a deep ductile shear zone. 2. Evidence from theBamble belt, south Norway. J. Geochem. Expl.3l, I49-169.

- (1990): Alkaline magmatism at Kirkland Lake,Ontario: product of strike-slip orogenesis. Geol. Surv.C an. P ap. 90-lC, 261 -269.

- (1993): Reintroduction ofgold, other chalcophile ele-ments, and LILE during retrogression of depleted gtan-ulite, Tromgy, Norway. Lithos 29, 303-309.

- CocuLU, E.H. & Stnrnc, J. (1993): Mobilization ofgold in the deep crusl evidence from mafic intrusions inthe Bamble belt, Norway. Lithos 30, 15 1-166.

- & HATroRr, K. (1987): Archean gold mineralizationand oxidized hydrothermal flrtids. Econ. Geol.82,1177-I 1 9 1 .

Canrwnrcut, I. & Baruqconr, A.C. (1987): Petrology ofScourian supracrustal rocks and orthogneisses from Stoer,NW Scotland: implications for the geological evolutionof the Lewisian complex. ln The Evolution of theLewisian and Comparable Precambrian High GradeTerrains @.G. Park & J.Tamey, e.ds,) Geol. Soc. London,Spec. Publ. n,93-IM .

CLAoUE-LoNG, J.C., KING, R.W. & KERRICH, R. (1990):Archaean hydrothermal zircons in the Abitibi greenstonebelt: constraints on the timing of gold mineralization.Eanh Planet. Sci. Iztt.98. 109-128.

Cmrx, M.E., CanurcHarq D.M. & HoDcsoN, C.J. (1988):Metasomatic processes and T - XCO2 conditions of wall-rock alteration, Victory Gold mine, Kambalda, WesternAustralia. /n Bicentennial Gold 88. Extended Abstracts.Geol. Soc. Au.st., Abstr. 22, 230-234,

PRECAMBRIAN GOLD AND SI{EAR ZONES 941

CLoucH, P.w.L. & Fmo, D. (1980): Chemical variation inmetabasites from a hoterozoic amphibolite - granulitetransition zone, South Norway. Contrtb. Mineral. Petrol.73,277-286.

Colvn'n, A.C., FyoN, J.A., Heerunn, K.8., MARMoNT, S.,Slrrnr, P.M. & Tnoop, D.G. (1988): Archean lode golddeposits in Ontario. Ont. Geol. Surv., Misc. Pap.139.

Coorrn, D.C. & FEr-o, D. (1977): The chemistry and originsof Proterozoic, low-potash, high-iron charnockiticgneisses from Tromgy4 south Norway. Eanh Planet. Sci.Iztt.35. 105-1 15.

Conzu, F., JncrsoN, S.J. & SUTcLIFFE, R.H. (1991): U-Pbages and tectonic significance of late Archean alkalicmagmatism and nonmarine sedimentation; TimiskamingGroup, southem Abitibi Belt, Ontario. Can. J. Earth Sci.28.489-s03.

-, KRocH, T.E., KwoK, Y.Y. & JsnssN, L.S. (1989):U-Pb zircon geochronology in the southwestern Abitibigreenstone belt, Superior Province. Can. J. Eanh Sci.26,r747-1763.

- & Nost-e, S.R. (1992): Genesis of the southern Abitibigreenstone belt, Superior Province, Canada: evidencefrom zircon IIf isotope analyses using a single filamenttechnique. Geochim. Cosmachim Acta 56,2081-2097 .

CRUDEN, A.R. (1992): Syntectonic plutons in the Larder l,ake- Cadillac deforrnation zone: implicationi for the timingof late Archean deformation in the SW Abitibi belt.Lithoprobe Rep. 25, (Abitibi-Grenville Transect), 139-142.

DAvrEs, F.B. (1982): Pan-African granite intrusion inresponse to tectonic volume changes in a ductile shearzone from northern Saudi Arabia. ./. Geol, 90,467483.

FALKI,rN4, T. & PETERSEN, J.S. (1980): The Sveconorwegianorogenic belt, a case of late-Proterozoic plate-collision.GeoI. Rundsch. 69. 622-647 .

Frsr-o, D. & Ct-oucu, P.w.L. (1976): K/Rb ratios and meta-somatism in metabasites from a hecambrian amphibolite- granulite transition zone. J. GeoI. Soc. London 132,n7-288.

- DRURy, S.A. & CoopER, D.C. (1980): Rare-earth andLIL element fractionation in high-grade charnockiticgneisses, South Norway. Zhhos 13,281-289.

=_ Suenrv, P.C., Leun, R.C. & RArnutl, A. (1985):Geochemical evolution of the 1.6-1.5 Ga-old amphibolite- granulite facies terrain, Bamble Sector, Norway; dis-pelling the myth of Grenvillian high-grade reworking. .lnThe Deep Proterozoic Crust in the North AtlanticProvinces (A.C. Tobi & J.L.R. Touret, eds.). NATO Adv.Study Inst., Ser. C,259-290.

For;ay, S.F., VEl.rrunnl-l-r, G., GnmN, D.H. & TosceNt, L.(1987): The ultrapotassic rocks: characteristics, classifi-

cation, and constraints for petrogenetic models. Earth Sci.Rev. A.8l-134.

FoNAREV, V.I. & Gnarcsrov, A.A. (1984): Experimentaldata and calculations on the stability of the orthopyroxene+ clinopyroxene + magnetite + quartz association.Ge ochem" 2l(4), 46547 l.

FRosr, B.R. (1991): Stability of oxide minerals in metamor-phic rocks 1n Oxide Minerals: Petrologic and MagneticSignificance (D.H. Lindsley, ed.). Rev. Mineral.25,469-487.

- &, CHAcKo, T. (1989): The granulite uncertainty prin-ciple; limitations on thermobarometry it granulites. J.Geo|.97,435450.

Lnvpslrv, D.H. & ANDERSEN, D.J. (1988): Fe-Tioxide - silicate equilibria; assemblages with fayaliticolivine. Arz. Mineral. n. 7n -7 40.

FvFE, W.S. (1973): The granulite facies, partial melting andthe Archean crust. Phil. Trarc., R. Soc. ktndon, Ser. An3.45746r.

Guronso, M.S. & Sacr, R.O. (1991): Fe-Ti oxide geother-mometry: thermodynamic formulation and the estimationof intensive variables in silicic magmas. Contrib.Mineral. P etrol. 108, 485-510.

GoLBERc, J.M. & Lnvnsl-ow, A.F. (1990): High temperature- low pressute Cretaceous metamorphism related tocrustal thinning (eastern Nortl Plrenean Zone, France).Contrib. Mineral, P etrol. 104. 194-207.

GRovEs, D.I., Krox-RoBINsoN, C.M., Ho, S.E. & Rocr,N.M.S. (1990): An overview of Archaean lode-golddeposits. 1z Gold Deposits of the Archean Yilgam Block,Western Australia: Nature, Genesis and ExplorationGuides (S.E. Ho, D.I. Groves & J.M. Bennett, eds.).Univ. We st. Australia. Publ. 20, 2-18.

Har.nvren, S, (1988): Great Slave Lake shear zone, CanadianShield: reconstructed vertical profile of a crusta1-scalefault zone. Tectonophys. 149 "

245-2&.

Hanr,ov, D.E. (1992): Comparative oxygen barometry ingranulites, Bamble sector, SE Norway. J. Geol. lO0,447-4@.

HArroRr, K. (1993): Diverse metal sources of Atchean golddeposits; evidence ftom in situ lead-isotope analysis ofindividual grains of galena and altaite in the Ross andKirkland Lake deposits. Contrib. Mineral. Petrol. ll3,185-195.

- & Henr, S.R. (1990): Fast recycling model inferredfrom isotope and trace element study of pyroxenes andfeldspars in Timiskaming-type alkaline rocks, soutlemAbitibi greenstone&lt: Third Int. Archean Symp. (Perth)'Ext ende d Ab s tr. Vo 1.. 267 -269.


Hrrrarrou, M.R. & DuNr.ls, L.A. (1984): Sedimentarion ofpull-apart basins: active examples in eastem Turkey. "LGeol.92,5 13-530.

HEwrrr, D.F. (1963)l The Timiskaming Series of theKirkland t ake area. Can. Mineral.7.497-523.

Holr-owav, J.R. (1976): Fluids in rhe evolution of graniticmagmas: consequences of finite CO, solubility. Geol.Soc. Am. Bull. 87. 15 l3-15 18.

HoLM, D.K., NoRRrs, R.J. & Cnew, D. (1989): Brittle andductile deforniation in a zone of rapid uplift: centralSouthern Alps, New Zealand. Tectonics 8, 153-168.

HuBERr, C., Tnuorr, P. & Grir-rNes, L. (1984): Archeanwrench fault tectonics and structural evolution of theBlake River Group, Abitibi belt, Quebec. Can. J. EanhSci. 2l. 1024-1032.

Hvop, R.S. (1980): Sedimentary facies in the ArcheanTimiskaming Group, Abitibi greenstone belt, northeastemOntario, Canada. Precambr. Res, 12, 1 61 -195.

JoLLy, W.T. (1978): Metamorphic history of the ArcheanAbitibi belt. /r Metamorphism in the Canadian Shield(J.A. Fraser & W.W. Heywood, eds.). Geol. Surv, Can.,Pap. 78-10,63-78.

KrnnrcH, R., Fnyen, 8.J., l(n,rc, R.W., WTLLMoRE, L.M. &VaN HBns, E. (1987): Crustal outgassing and LILEenrichment in major lithosphere structures, ArcheanAbitibi greenstone belt evidence on the source reservoilfrom strontium and carbon isotope tracers. Contrib.M ineral. P etrol. 97. 156- 1 68.

KIrfl-e, J. (1989): Polymetamorphic Evolution of Cordierite-Bearing Metapelites of the Bamble-Sector, SouthernNorway. M.Sc. thesis, Univ. Oslo, Oslo, Norway.

KIsruua, N. (1989): A thermodynamic srudy on rhe pyrite -pyrrhotite - magnetite - water system at 300-500"C withrelevance to the fugacity/concentration quotient of aque-ous H2S. C e oc him. C o s mochim- Acta 53, 2143 -21 55.

Koot{s, P.O. (1987): Some thermal and mechanical conse-quences of rapid uplift: an example from the SouthemAlps, New Zealand. Earth Planet. Sci. Lett.86,307-319.

- & Cnaw, D. (1991): Gold mineralization as a conse-quence of continental collision: an example from theSouthern Alps, New Zealand. Earth Planet. Sci. Len.103. l-9.

KuLr-ERUD, G. & Yooan, H.S. (1959): Pyrite stability rela-tions in the Fe-S system. Ecoz. Geo\.54,533-572.

Ku l lenuo, L . & DauroneN, S .H. (1991) : Sm-Ndgeochronology of Sveconorwegian granulite facies min-eral assemblages in the Bamble Shear Belt, SouthNorway. Unpubl. manuscript.

Laur, R.C., Svellev, P.C. & Frnlo, D. (1986): P-T condi-tions for the Arendal granulites, southern Norway: impli-cations for the roles of P, T and CO2 in deep crustalLllE-depletion. J. Metamarph. Geol, 4, 143-160.

LECLATR, A.D., Ewsr, R.E. & Harront, K. (1993): Crustal-scale auriferous shear zones in the central SuperiorProvince, Canada. Geology 21, 399-402.

LEvEseuE, G.S., CeurnoN, E.M. & Ler-or,ur, A.E. (1991):Duality of magmatism along the Kirkland Lake'LarderLake fault zone, Ontario. Geol. Surv. Can., Pap, 9l-lC,l7-24.

Lmny, J.W., BAru-Ey, M.E., ErssxlosR, 8.N., Gnovrs, D.I.,HRoNsKy, J.M.A. & VEARNcoMBE, J.R. (1990): Craton-scale deformation zones. /lr Gold Deposits ofthe ArcheanYilgarn Block, Western Australia: Nature, Genesis andExploration Guides (S.E. Ho, D.L Groves & J.M.Bennett, eds.). Univ. West. Australia, PubL 20,30-37 .

Loci<rrr, J.M. & Kuszun, N.J. (1982): Ductile shear zones:some aspects of constant slip velocity and constant sheatstress models. Geophys. J. R. Astron. Soc. 69,477-494,

MCINNES, B.I.A. & CauEnoN, E.M. (1994): Carbonated,alkaline melts from a sub-arc environment: wedge sarrr-ples from the Tabar - Lihir - Tanga - Feni Arc, PapuaNew Guinea. Earth Planet. Sci. l*n. (in press).

McKeNzr,, D. & Btcru-n, M.J. (1988): The volume and com-position of melt generated by extension of the lithbs-phere. "/. Petrol. 29, 625-679.

MEEN, J.K. (1987): Mantle metasomatism and carbonatites;an experimental study of a complex relationship. Gaol.Soc. Am., Spec, Pap.2t5, 91-100.

Mrrucrr, E.J. & Gnoves, D.I. (1990): Mineralogical con-straints. 1n Gold Deposits of the Archean Yilgarn Block,Western Australia: Nature, Genesis and ExploratidnGuides (S.E. Ho, D.I. Groves & J.M. Bennett, eds.).Univ. West. Australia, Publ. 20, 212-220.

MunLLen, W., DoNalnsoN, J.A., DurnnsNe, D. &R@HELEAU, M. (1991): The Duparquet Formation: sedi-mentation in a late Archean successor basin, Abitibigreenstone belt, Quebec, Canada. Can. J. Eanh Sci.28,1394-1406.

MyERs, J. & Eucsren, H.P. (1983): The system Fe-Si-O:oxygen buffer calibrations to 1,500 K. Contrtb. Mineral,Petrol. 82,75-90.

NrxoN, P .H. , REEDMAN, A.J . & BuRNs, L .K . (1973) :Sapphirine-bearing granulites from Labwor, Uganda.M ine ral. Mag. 39, 420428.

Nonrus, R.J. & HsNLsv, R.W. (1976): Dewatering of a meta-morphic pile. Geology 4,333-336.

Prncrver, J.A. & Wrluevs, H.R. (1989): The late ArcheanQuetico accretionary complex, Superior Province,Canada. Geola gy 17, 23-25.


PTDGEoN, R.T. & Bowns, D.R. (1972): Zircon U-Pb ages ofgranulites from the cenftal region of the Lewisian, north-west Scotland. Geol. Mag. lW, 247 -258.

PnMAN, W.C., m & ANonews, J.A. (1985): Subsidence andthermal history of small pull-apart basins. In Strike-SlipDeformation, Basin Formation, and Sedimentation (K.T.Biddle & N. Christie-Blick, eds.). Soc. Econ. Geol.,Paleontol. & Mineral., Spec. Pub1.37,45-49.

Rsaonc, H.G. (1980): Characteristics and recognition ofstrile-slip fault systems. 1n Sedimentation in Oblique-Slip Mobile Zones (P.F. Ballance & H.G. Reading, eds.).Int. Assoc. Sedimentol., Spec. Publ.4,7-26.

RoBERT, F. (1989): lnternal structure of the Cadillac tectoniczone southeast of Val d'Or, Abitibi greenstone belt,Quebec. Can. J. Earth Sci. ?.6,2661-2675.

RoBNsoN, D. & BEvNs, R.E. (1989): Diastathermal (exten-sional) metamorphism at very low grades and possiblehigh grade analogaes. Earth Planet. Sci. Iztt.92, 81-88.

RowrNs, S.M., CAMERoN, E.M., LALoNDE, A.E. & ERNsr,R.E. (1993): Petrogenesis of the Late Archean syeniticMurdock creek pluton, Kirkland l,ake, Ontario: evidencefor an extensional tectonic setltng. Can. Mineral.3l,219-244.

-, LALoNDr, A.E. & CawnoN, E.M. (1991): Magmaticoxidation in the syenitic Murdock Creek intrusion,Kirkland Lake, Ontario: evidence from the ferromagle-sian silicates. "L Geol. 99, 395-414.

Rurrnn, E.H. & Bnoore,, K.H. (1985): The permeation ofwater into hydrating sheat zones. 1r MetamorphicReactions Kinetics, Textures, and Deformation (A.B.Thompson & D.C. Rubie, eds.). Springer Verlag, Berlin(242-250).

SaNpnnsoN, D.J . & Mnncu lNr , W.R.D. (1984) :Transpression. "/. Struct. Geol.6, M9458.

SeNorrono, M., Ne,al l , F.B. & PowELL, R. (1987):Metamorphic evolution of aluminous granulites fromLabwor Hills, Uganda. Contrib. Mineral. Petrol. 95,217-225.

- & PowELL, R. (1986): Deep crustal metamorphismduring continental extension; modern and ancient exam-ples. Earth Planet. Sci.14ft.79, 151-158.

Scu-rNcr,n, C.M. (1985): Magnetization of lower crust andinterpretation of regional magnetic anomalies; examplefrom Lofoten and Vesteralen, Norway. J. Geophys. Res.90 .1189-1201.

Srns, J.D. (1983): Mineralogical changes occurring duringthe retrogression of Archean gneisses from the Lewisiancomplex of NW Scotland. Lithos 16, 113-124.

Spmns, C.J. & Psecs, C.J. (1989): Development of dilatancyand permeability in rocks during creep. experiments onrocksalt as rock analog. Int. Geol. Congress 28th, Abstr.3. 162-163.

SroNE, W.E., CRocKEr, J.H. & Flrrr, M.E. (1990):Partitioning of palladium, iridium, platinum, and goldbetween sulf ide l iquid and basalt melt at 1200"C.Geochim" Cosmochim- Acta 54, 2341-2344.

Tavlon, B.E. (1986): Magmatic volatiles: isotopic variationof C, H, and S. 1lz Stable Isotopes in High TemperatureGeological Processes (J.W. Valley, H.P. Taylor, Jr. &J.R. O'Neil, eds.). Rev. Mineral. 16,185-225.

TEoMsoN, J.E. (1950): Geology of Teck Township and theKenogami Lake area, Kirkland t ake gold belt. Ont. Dep.Mines. Ann. Rep.57. l-53.

Cuanr-Ewooo, G.H., GRIFFIN, K., Hawlrr, J.E.,Hoprnqs, H., Meclxrosn, C.G., OcRIzLo, S.P., PERRY,O.S. & Wano, W. (1950): Geology of the main ore zoneat Kirkland Lake. Ont. Dep. Mines, Ann. Rep. 57, 54-188.

ToocooD, D.J. & Hoocsorv, C.J. (1985): A structural investi-gation between the Kirkland lake and Larder l,ake goldcamps. Ont. Geol. Sun., Misc. Pap. Ln,200-205.

TouREr, l. (1971): Le facies granulite en Norvbge m6ri-dronaTe. Lithos 4, 239 -249.

- (1986): Fluid inclusions in rocks from the lower conti-nental crust. 1n The Nature of the Lower ContinentalCrust (J.B. Dawson, D.A. Carswell, J. Hall & K.H.Wedepohl, eds.). Geol. Soc. htmdon, Spec. Publ. 24,161-172.

- (1992)t CO2 transfer between the upper mantle and theatmosphere: temporary storage in the lower continentalcrust. Terra Nova 4,87-98.

- & DrErvoRsr, P. (1983): Fluid inclusions in high-grade anatectic metamorphites. J. GeoI. Soc. hndanl40,635-649.

TuRcorrE, D.L., TAc, P.H. & Cooppn, R.F. (1980): A steadystate model for the distribution of stress and temperatureon the San Andreas Fault. "/. Geoplrys. Res.B,85,6224'6230.

sevmn, M. & BoNATrl, E. (1ggg): petrology of a gneiss - vIErzEUF, D. & KonNpnossr, J. (1984): Crustal splitting and

amphibolite lower crustal unit from Zabargad Island, Red the emplacement of Pyrenean lherzolites and granulites'

Sei. Tectonophys. lS0, l7i-207. -

Earth Platuet. Sci. Iza. 67,87-96.

SrBsoN, R.H., RoBERr, F. & Pounrx, K.H. (1988): High- WEAIIER, B.L. & Tanrrnv, J. (1984): Empirical approach toangle reverse faults, fluid pressure cycling, and mesother- estimating the composition of the continental crust.malgold-quartzdeposits. Geology16,551-555. Nature3l0,575-577.

944 THEcANADTANMTNERAIocIST

Worc, L., Davrs, D.W., Knocu, T.E. & RoBERr, F. (1991): YI,'BN, D.A., Frsnour, L., Scllrusmr, G. & Fnororvaux, C.U-Pb zircon and rutile geochronology of Archean green- (1978): Shear deformation zones along major transformstone formation and gold mineralization in the Val d'Or faults and subducting slabs. Geophys, J. R, Astron, Soc.Region, Quebec . Earth Planet. Sci. Izx. IM,325-336. 54, 93-119.

Wvrraer, D.A. & Krnrucn, R. (1989): Archean lamprophyredykes of the Superior Province, Canada: distribution,pefology, and geochemical characteristics. J. Geophys. Received JuIy 2, 1992, revised manuscript acceptedRes. 94,46674696. December 9. 1992.

Documents

PRECAMBRIAN GOLD: PERSPECTIVES FROM THE TOP AND …rruff.info/doclib/cm/vol31/CM31_917.pdfdu magmatisme alcalin et d'une pdriode soutenue de carbonatisation d Kirkland Lake. Une fois