Document Sample
					Canadian Mineralogist
Vol. 19, pp. 607-618 (1981)

                          MINERALIZATION WATERFALL
          MAGMATICCu-Ni-PGE            AT

                                         S.E. TISCHLER
                    Frieda Copper Pty. Limited, Carpentaria Exploration Company,
                                   Madang, Papua New Guinea

                                   R. GRANT CAWTHORNI
               Geology Depdrtment, University of the Witwatersrand,Ian Smuts Avenue,
                                Iohannesburg2001, South Africa

                                        G.A. KINGSTON
                       Department of Mineral Exploitation. University of Carditt,
                                  Cardiff , Wales, United Kingdom

                                         S. MASKE
               GeologyDepartment,Universityof the lilitwatersrand,Ian SmutsAvenue,
                              lohannesburg2001, South Africa

                    ABSTRACT                            fond6e sur des critires g6ochimiquesmodernes'La
                                                        composition du magma ioitial i Insizwa serait
   The existence of Ni{u-PGg       6laslalizafion in    tbol6iitique ir haute teueur etr MgO; la s6gr6gation
the basal portion of, and the sediments underlying,     d'un liquide sulfur6 r€sulterait de la contamination
the Insizwa intrusion in the Transkei, southern         d'un tel magma par des shales siliceux. I*s don-
Africa, has been known for a long time. Here we         n6esisotopiquessur le soufre sont compatiblesavec
present a new interpretation, based on modern           une origine magmatique.
geochemical criteria, of the origin of the sulfides.
It is suggested that the composition of the parental                          (traduit Par la R6daction)
magma of the Insizwa intrusion is a high-MgO
tholeiite, and that the possible cause of sulfide       Mots-clis: complexe intrusif Insizwa, sulfures mag-
liquid segregation is the contamination of this mag-     matiques, nickel, cuiwe, 6l6mentsdu groupe du
ma by siliceous shales. The sulfur-isotope datp are      platine, Transkei.
consistent with a magmatic origin.

Keywords: Insizwa intrusion, magmatic sulfides,                          INrnopuctroN
  nickel, copper, platinum-group elements, Transkei.
                                                          A number of subhorizontal doleritic and gab-
                                                        broic sheetsexists in Pondoland, northern Trans'
                    Sorvluernn                          kei, southern Africa, as erosional remnants of
  On connait depuis longtemps l'existence d'une         a possibly continuous intrusion of Karroo dol-
zone min6ralis6e en Ni, Cu et 6l6ments du groupe        erites between Upper Permian Ecca sediments
du platine dans la partie inf6rieure du com,plexe       (grey sandstones and mudstones) and l-ower
Insizwa (franskei, Afrique du Sud) et dans les          Triassic Beaufort sediments (red sandstones and
unit6s s6dimentaires sousjacentes. Nous proposons       mudstones). One of the more extensiveof these
une interpr6tation nouvelle de I'origine des sulfures   is the Insizwa{ngeli-Tonti-Tabankulu       intru-
                                                        sion, which occurs near the border of the Trans-
                                                        kei with South Africa, 175 km southwest of
'r'The present paper is the manuscript of a con-
                                                        Durban (Fig. 1). It has been argued (Maske
tribution to the 1980 field conference of IGCP -        1966) that the present Pondoland area formed
project 161: "Magmatic Ni-sulfide deposits" in
                                                        the deepest part of the former Karroo basin,
Scandinavia; this contribution was presented by
the senior author on July 30, 1980 in Turku,
                                                        and that the Jurassic Karroo dolerites, which
Finland.                                                occur over an area of some 64,000 km2 in
                                                        southern Africa (Bruynzeel 1957), were em-
tto whom correspondence should be addressed.            placed under plutonic conditions. This resulted

608                                 THE CANADIAN MINERALOCTST

                                                         The base and top of the intrusion are not
                                                     planar but undulate slightly, resulting in thicken-
                                                     ing in the troughs and thinning over arches.
                                                     On a regional scale, the intrusive body is con-
                                                     cordant. However, Maske (1966) indicated the
                                                     presence of several transgressiverelationships;
                                                     in many places the dip of the contact attains
                                                     40" (Dowsett & Reid 1967), indicating local
                                                     discordanceswith the essentially flat-lying sedi-
                                                         Most workers have subdivided the intrusion
                                                     into a basal zone from about 100 to 13O m
                                                     thick. a central zone from 300 to 5O0 m thick
                                                     and a thin upp€r zone. The upper zone is com-
                                                     posed of quartz diorite and quartz monzonite
                                                     and, locally, micropegmatite, which show an
                                                     anastomosingrelationship to the hornfels in the
                                                     roof. These rapidly grade downward into hyper-
                                                     sthene gabbros of t}te central zone. At Waterfall
                                                     Gorge there are three olivine hypersthene gab-
                                                     bro bands overlain by quartz hypersthene
Flc. l. Generalized map of the Insizwa-Ingeli        gabbro, whereas in the Ingeli section there are
  Tonti*Tabankulu intrusions, showing the lower      two bands of each rock type. The basal zone
  picritic portions. Based on mapping of Scholtz     has been studied in more detail because of its
  (1937), Maske (1966) and Dowsett & Reid
  (1967).                                            sulfide mineralization; althoug it is largely pic-
                                                     ritic it has been subdivided into four units:
                                                     roctolite, basal picrite, basal olivine hyper-
                                                     sthene gabbro and a chill zone.
in gravitational differentiation producing basal,        The troctolite is not developed in the Water-
central and upper zones. Picrite and troctolite      fall Gorge section, but is 5O m thisk at Ingeli.
dominate the basal zone, hypersthene gabbro          The top of the basal zone is taken at the level
the central zone, and quartz diorite and quartz      where plagioclase increases markedly in abun-
monzonite the upper zone. Sections and mineral       dance to about 7OEo,This is associatedwith a
compositions are shown in Figure 2.                  marked decrease in the abundance of olivine.
   The basal zone at Waterfall Gorge (Fig. 3)        However, a small proportion of olivine is pres-
has been described in detail by Scholtz (1937);      ent above this level for about 20 m, forming
it is characterized by disseminationsof chalco-      the lower olivine hypersthene gabbro unit of
pyrite, pentlandite and pyrrhotite grains as well    the central zone. The picrite, composed of over
as cross-cutting veins and sheetsof massive sul-     6OVo olivine and about 100 m thick, is the
fide in association with acid phases.                dominant rock type of the basal zone. The
                                                     chill zone and basal olivine hypersthene gabbro
 Georocv oF THE Warnnrerr. Goncr Ansr,               together are only 20 m thick. However, it is
                                                     this lowest 20 m that contains the disseminated
   The sulfide mineralization is spatially con-      sulfide mineralization.
fined to the lower contact of the intrusive gab-         The textures of these rocks have been de-
broic sill; it is generally accepted that the two    scribed by Scholtz (1937), Bruynzeel (1957)
are genetically related (Scholtz 1937).              and Maske (1966) and, except for the finer
   The sill is dissected by the Umzintlava and       grained chilled margin, are fairly typical of
Umzimbubu river valleys. The Insizwa, Ingeli,        orthocumulate rocks. Layering is not well devel-
Tonti and Tabankulu massifs probably were            oped in the sequence, with the exception of
part of the same intrusion; if so, it has an areal   the troctolite unit, but textures and mineral and
extent of at least 1200 km! and a vertical thick-    whole-rock compositions are broadly consistent
nessof up to 900 m. Although lateral variations      with an origin by crystal settling and fractional
in the intrusion have been recorded (Fig. 2),        crystallization. I{owever, the reappearance of
it was probably emplaced in not more than            olivine in the central zone in tle Ingeli section
three very closely spaced intrusive events, and      led Maske (1966) to suggestthat there was an
hence representsa very major injection of mag-       addition of undifferentiated magma. Two sec-
ma-                                                  tions through the intrusion are shown in Figure
                          MAcMATTc Cu-Nr-pGE           MTNERALTZATToN. rNsrzwA
                                                                    AT                                609

    Woterfoll Gorge Seclion                                                    Ingeli Section
                             opx                                                  OI    P l o g opx
                            %En                                                   AoFo V"An "/" En
                             72                                                             ?a



           h:       73

                                                      Centrol olivine

                             79              -                                              73
           174                                  Lowerouortz
                             7l                       gobbro
                                             hypersthene                                    79

                    7C       69
                    66                                                                      7?
               74   66

               85   67

               89   53

                    99u"              taor   olvtno   ny9el$heng   oobbro

Fg'-?a Generalized profiles and. mineral compositions through the Waterfall
                                                                            Gorge section (Scholtz
  1937) and through the Ingeli section (Maske 1966).

 2. The mineral compositions are not entirelv            becomes more magnesian upward in the basal
definitive in this regard, nor do they show              zone at Waterfall Gorge, showing an iron-en-
the. range expected of a differentiating magma           richment trend thereafter, with a small reversal
body. The compositions of the piagioclase                at the central thin olivine-bearing horizon and
crystals do not show a significant gross trend          then another reversal at the upper olivine hyper-
of differentiation, although zoning of individual        thene gabbro. Above this level its composition
crystals is well developed (Maske 1966). For            remains constant. In the Ingeli section the min-
example, in the Waterfall Gorge section, the            eral shows a gradual trend from Enen to En-
feldspar becomes more anorthite-rich upward,            from bottom to top, with a reversal at the
from Anoo to Anze through the basal zone and            olivine-bearing horizon apparent in one profile
the lower half of the central zone. It shows a          but not in other profiles studied by Maske
typical trend decreasing in An content upward            (1966).
to Anoo, and then abruptly increases to Anzr                fn none. of the sections analyzed is there a
at the base of the thick upper olivine hyper-           convincing reversal in the olivine composition
sthene gabbro. Above this levels its compoiiiion        where it reappears in the middle of the central
remains nearly constant. In the Ingeli section,         zone. Again, in the Waterfall Gorge section,
the trend is similar except that the reversal at        the olivine becomes more magnesian upward
the olivine hypersthene gabbro horizon is in-           from Foeo to Foae over an interval of 60 m,
significant.                                            before it starts to follow a normal differentia-
   The trend is analogous for orthopyroxene. It         tion trend. Thus, although fractionation un-
610                                "THE CANADIAN MINERALOGIST

                                                                        OllVine hyperslhene
                                                                        Pl c r i t e
                                                                        Chill phose


                                                           EilIt        Monzonite
                                                           hl+lf              vein

                                                           I   Mossive
   U'                                                       tffi         Disseminoted
      c)                                                           GROUP
                                                            F-  Grrr
                                                            :            Shole
Flc. 3. Section through the basal zone at Waterfall Gorge, Insizwa, showing the sulfide distribution
  ter Scholtz 1937).

 doubtedly has taken place, there are features        by metamorphis.m. Very close to the contact,
 of the mineral compositions that are enigmatic.      hypersthene is present as irregular poikilo-
    The floor sediments. which are host to some       blasts, and a few grains of cordierite have been
 of the fracture-filled veins of sulfide, are sit-    reported (Scholtz 1937, p. 125). There are
 ceous shales. In immediate proximity to the          some calcareous nodules within the hornfels,
 floor of the intrusion they are hornfelsed, but      and these contain a small proportion of wollas-
 growth of metamorphic minerals is very re-           tonite, grossular and dioPside.
 stricted. The shales are composed of fine grained
 quartz and feldspar (partly kaolinized) in a                          Onn Mrxnnlr-ocY
 matrix of chlorite and sericite with some de-
 trital biotite. There are also a few larger flakes     Nickel sulfides were found at Waterfall
 of biotite that Scholtz (1937) viewed as formed      Gorge as early as 1865 by Rudlin (Kenyon
                                  MAcMATTc Cu-NI-PGE                    MTNERALIZATToNAT INsrzwA                           6ll
    1973).. Since 1930, various companies have
   intermittently carried out exploritory work;                               . An electron-microprobe investigation of the
                                                                            platinum-group minerals (pGMl        in the In-
   ho-wever, few quantitative data have beeu pub_                          sizwa ores from Waterfall Gorge has been
   lished.                                                                  initiated by one of us (G.A.K,). Analyses were
      therg are two principal ore rypes: (1) dis_                          performed on a four-channel Cambitdge In-
   seminations of chalcopyr,ite, pentlandite and                           strument Microscan III. pure elements and
  pyrrhotite,_mainly in basal hypersthene gabbro                           pyrite were used as standards at an operating
  and, to a lesser extent, in the chill and                                voltage of 20 kV and a specimen current of
  dykes in the surrouuding areas; ((2) massiie ore                         1O0 mA. Our work, particularly on the dissem-
  insheets and veins, both in the baiai hypersthene                        inated sulfide ore, has so far revealed the
  gabbro and adjacent hornfelsed sediments. al_                            additional occurrence of froodite pdBi, asso-
  ways in close spatial association with acid                              ciated with parkerite NLBirS, and irarsite IrAsS
  phases. The possible interrelations between                              with sperrylite PtAsz. Bismuthinite has also been
  various types of ore and host rock are illus-                            confirmed for the first time (Fig. 4d), together
  trated in
             rigrt" 3 (after scholtz lg37). Assay                          with a lead telluride (Fig. 4a) that has u="o-_
  values of the principal ore ty?es were given by                          position of Pb 66.4 and Te 33.0 wt. Vo. cor-
 Scholtz and are shown in fi6te t.                                        responding to the formula pbsTea. The iiffer-
     A detailed account of the mode of occurrence                         ence between the lead telluride and altaite
 and textural features of the base metal sulfides                         PbTe (Pb 61.9, Te 38.1 wL %) may be real
 in these-o-res given by Scholtz (1937), who                              or due to errors inherent in analyzing such
 reported that in both the massive and dissem_                            small grains (4 x 14 trr,m).However, the euhe-
 inated types of sulfide ore, the principal sul_                          dral prismatic habit suggests a nonisometric
 fide- minerals are pyrrhotite, peitlandite and                           structure, although no anisotropy could be seen
 chalcopyrite. pyrrhotite is, however. bv far the                         under crossed polars. Two analyses of parkerire,
 most dominant sulfide; it commonly d-isplav, u                           which was usually found intergrown with
 two-phase lamellar exsolution intergrowth of                             froodite (Fig. 4c), gave compositions of Ni
 troilite- and exagonalpyrrhotite. Scholtz e937                           25,O, 26.6%, Fe 2.1, 3.3Vo, Bi 6O.8, 6I.6Vo,
               l                                     )
 also identified minor amounts of pyrite, bra_                            Te 1.9, 2.4Vo and S 9.7, 9.3%, respectively,
 voite, valleriite and cubanite, togethei witfr rare                      approximating closely (NiFe) g(BiTe),S!.
 sphalerite, niccoline, millerite, 6ornite, chalco_
 cite and covellite. At least sixteen othei distincr
                                                                          The platinum-group minerals
minerals were found to occur generally as such
small grains as to render their identification                                Sperrylite PtAsz was previously reported by
uncertain. However, on the basis of micro-                                 Scholtz (L937) as a relatively common pGM
chemical tests on samples of minerals extracted                            in-the massive ore, where it invariably occurs as
f-rom the polished sections, Scholtz suggested                             idiomorphic crystals of up to 0.2 m; in length
that this list include a sulfarsenide of-iickel                            in contact with pyrrhotite, cubanite and parkir-
and iron, tetrahedrite,/freibergite, a nickel anti-                        ite. His identification was based on optica-l prop-
monide, a lead telluride, a tellurium_rich min-                           erties, with qualitative microchemicil tesG for
eral, native_bismuth, gold-silver--copper alloys,                         Pt and As. In the present preliminary polished-
galena, and a nickel sulfide and a platinum                               section study of the disseminated-ore, one
telluride that he named parkerite and iiggliite,                          euhedral rectangular grain of sperrylite 25 x 32
                                                                          /rm i? size was found; it is overgrown by chal-
                                                                          copyrite at the silicatq-sulfide contact          an
                                                                          interstitial   pyrrhotite-pentlandite--chalcopyrite
                            OF   FROITI
                                     INSIZIIA*                            aggregate (Fig. ab). An electron-micropiobe
                                                                          analysis gave the composition as pt 56.5, As
                                                                         42.3, Sb 1.4 and S O.2Vo,corresponding to the
            0.83           4.48          5.95           5.08
LU          0.60                        15.35           3.45             formula Pt(Aso.erSbo.oog.or)o.0.. ocJurrence
Co                        0.21
                                         0.37                            of "hard" PGM, such as sperrylite, braggite,
                                        0-t10             l6
                                      0.5-9.0            0.4             cooperite and laurite, growing inward from-the
PGE          l q
                                                                         silicate wall of the sulfide aggregatesis also a
No'                          7                                           characteristic feature of the Merensky Reef
I   Data from Scholtz (1937). L DisseninaredsutfiaelnoTii                (Kingston 1977, Yermaak & Hendriks 1976):
lle-lJp:rsthene sabbro: 2. eyrrhotite-iiifi-i;;i;;'Jie'ip;i:'
co_andareaveiase 2 ,*'pi!l-i,iil;
    s                                                                    this feature, together with the common euhedrai
               or                         1."'"ciiiliiiri!:"'
rich ore; 4. t4lxedmassiveoresi toiai weigtri irii!6a'io                 nature towards all the principal base metal
                                        z: nmber
analyzed by Johnson,l4atthey an
                                                                         sulfides, could be interpreted as indicating a
ilijl;;i,*ip*t"\il'iiil"ct"!"i"i'oillll";"nS                             very early stage of crystallization
61.2                                   THE CANADIAN MINERALOGIST

                                                     EgE1j ii
                                                     EAEijl; iat;;...;
                                                 fa"E+*;! t::i it;
                                                 +6dsa; i ;:;;i
                                                 ef;?ss! tt i: t;:!i
                                                 FesEha,,; i

                                                            t:t:1 1!l:l
                                                            |ltlt Ot!
                                                                :t l:t 1l

                                                                                                   contains flame'
Frc. 4a. photomicrograph of a prismatic crystal ol a lead telluride in chalcopyrite that
  like inclusions of mackinawit            s"ut" is lb g.m. 4b. Photomicrogf,aph showing a euhedral crystal of
  .po.yfit"      at a silicate-chut*pyrit"    boundary.' Irarsite occurs as a small darker, euhe{ral .corner over-
  growth of sperrylite. S"il i- 10 pm. 4c. Phitomicrograph showing me{ium grey .bireflecting.
  ir)                 a euneoJi gtui" [i *niie froodite G)             a marix of-chalcopyrite' A *ry.\ of gold
                                                                                                (c). is
  is white (Au). A o,ottf"A--Ouit gr€y area is a residue of carbon after analysis
   l0 um. ad.       photomicrograph rnor"loui an anhedral composite grarn of frooditre (f) and^bire-flecting bis-
   -rfririt"-iul-i"       ;;fi;'"f      pyrrhJtfte tlat displays a lamellar structure' Scale bar is 10 p.m.

    Froodite is considered to have an ideal end-                               Electron-microprobe analysis of three frood-
 member formula of PdBL and a general com-                                  ite grains approximately 5-1O p,m in size qave
 posirion of (Pd,Pt) (Bi,Te)z (cabri & Laflamme                             comlpositiorisof Pd 17.O, 12.8,-l4.8Vo; F"-?.9'
 iglO).In     the disseminatedsulfide ore of In-                            2.4, L.1/o; Ni 0.2, n.d., 0.37o; Bi 74.4' 78.9'
 sizwa, the mineral considered to be froodite is                            72.6V0 and S 0.2, n.d., O.ZVo,respectively, cor-
 normally intergrown with parkerite and, less                               responding total cation:anionratios of 1:1.83,
 commonly, with bismuthinite and native bis'                                 l:2.3 and 1:2.05, respectively,  with an average
 muth, forming small anhedral composite grains                              of 1:2.08. These rather unsatisfactory results
 less than 25 um in size dominantly in pyrrho-                              may be due in part to analytical error €aused
 tite and. in one case,in chalcopyrite(Fig. 4c, d).                          by the small size of the froodite grains. Several
 The mineral is bright, crearny white, with no                               uit"roptr to improve on these analyses iailed,
 detectable bireflectance (in contrast to the                                and it seems likely that there is insufficient
  darker parkerite, which displays a distinct bire-                          adjustment by the correction program-for this
 flectance, creamy white to greyish white with                               particular combination of elements. Ilowever,
  a bluish tint). The mineral is only weakly                                 lhe above results, although not conclusive, do
  anistropic under crossednicols, whereas parker-                            strongly suggest that the mineral is a ferroan
  ite displays a strong anisotropy. Froodite is                              variety of froodite.
  softer ihan parkerite and has a higber reflect-                               An Ir-As-S phase considered to be irarsite
  ance. It is most probable that these minute                                was found as an overgrowth on the corner of a
  composite grains in pyrrhotite resulted from                               sperrylite crystal in the disseminated sulfide
  exsolution.                                                                ore (Fig. 4b). Its properties correspond to
                         MAGMATTc Cu-NI-PGE       MTNERALTZATToNAT rNsrzwA                                            613
  those of irarsite (Uytenbogaardt & Burke l97l):       in understanding the genesis of the mineraliza-
  greyish white with a distinct bluish tint (com-       tion. It has long been known that there exists
  pared with white for sperrylite), reflectance        an excellent positive correlation between Ni
  lower tlan that of sperrylite and isotropic          and MgO contents of basic magmas. Eates &
  under crossed nicols. A qualitative electron_        Marsh (1979) argued that dolerites from the
  mic-roprobe analysis confirmed major Ir, As          central Karroo province with greater than lO%
  and S.                                               MgO were unlikely to represent true liquids,
     Niggliite PtSn has been described in the          but were enriched in MgO by disproportionately
  Inaslvg pyrrhotite ore from Waterfall Gorge          high olivine content, owing to differentiation
  by Cabri & Harris (1972\, who showed that            processes.However, it was argued by Cawtlorn
  the mineral named by Scholtz (1937) as nig-           (1980) that such changes did not occur in the
  gliite is not a platinum telluride but pt(Sn.        case of the Insizwa intrusion. As evidence of
  Sb,Bi). The mineral occurs as inclusionsup to        this, he cited the data of Scholtz (1937), who
  25 x 75 pm in size within insizwaite, parkirite      described the chill margin as containing only
  and pentlandite. Niggliite has not yet been found    O.9Vo olivine crystals even though the analysis
 in the disseminated-typeore.                          he gave showed 13.4% MgO. Thus, the peiro-
    Insizwaite PtBig was described for the first       graphic evidence clearly demonstrates thit this
 time by Cabri & Harris (1972) in a sample             composition is not enriched in olivine. Com-
 from the Waterfall Gorge section, donated iy         parative analytical data on the chill-zone rocks
 Professor Scholtz. They found insizwaite as          from the Insizwa intrusion and from other Kar-
 small rounded grains up to 70 x l2A nm in            roo dolerites are given in Table 2.
 size- enclosed in pentlandite, chalcopyriie and           The second feature that appears to be anoma-
 parkerite and enclosing niggliite. T[e averag"       lous relates to the composition of the olivine
 formula from five electron-microprobe analvses       and, to a lesser extent, of the other silicate
 was given as PtroBl.grSbo.o". This phase has not     minerals. Bruynzeel (1957) presentedresults of
 yet been found in the disseminated-type ore.         several analyses of olivine from the chilled
     Gold-silver-copper alloys that couid contnin     margin and the lower olivine-rich (picrite)
 minor amounts of the platinum-group elements,        zone. The compositions increased in their for-
 particularly palladium, were reporteA by Schottz     sterite content from about Foeo at the base to
 (1937) in polished sections and nonmagnetic          Foenin the middle of the zone, and then began
 concentrates prepared from the massive ore at        to decreasewith further increase in height. These
 W{erfal} Gorge. He recognized coppery red,           values show a larger range than the values of
pale, yellow and silver-white alloy-iyp6s; he         Foes-ao   reported by Eales (1980) for the pic-
 consideredthese to correspond to coppii, gold-       rites from Insizwa. However, he did not in-
 and -_silver-rich Au--Ag:{u alloys, reipectively.    dicate the stratigraphic range covered by his
Small specks of such alloys have been located         samples.Scholtz (1937) and Bruynzeel (1957)
in the disseminated ore (Fig. 4c) but require         also analyzed orthopyroxene, clinopyroxene and
further investigation.                                plagioclase compositions and found that these
    It is clear from the evidence of the manv         displayed a reversed differentiation trend from
unidentified phasesdescribedby Scholtz (lg37i,
and the many distinctly different unidentified
grains-still to be analyzed in the present study,
                                                       TABLE COMPOsITIONCHILLS
                                                           2.         OF      FROI,I
                                                                                  INSIZI,IA OTHER
                                                                                          AND   MRROO
that the platinum-group mineraiosv is mosr
probably as complex and extensiveai'that found
                                                      si02       52,00   54.05 49.18 51.71 54.73 5.t.96 0.57
in the. Merensky Reef. The authors' preliminary       Tt02                0.71   0.64
                                                                  0.60                   0.98  1.t8  0.90  0.06
investigation indicates that a differince prob-       A',l^0^    l?.81   t5,07 13.23 t5.31 15.64 t5.64     o.3l
ably exists between the pGM assemblagedf the          Fe^0^       1.84    2.'t3 l't.07* ll.2o* 0.87  o.g2  0.43
massive and disseminated tyIE$ of ore, and that       Fe0         6.92    5.94                 s.63  9.23
this can be related to a difference in genesis.A      tiln0       0,28    0.33   0.21    0..18 0.16  o,l7
detailed study of the mode of occurrence and          1,4S0      13,24    7,31 l2.8      6,83  6.18  7.55  o.5l
                                                      Ca0         7,65    9.00   9,76 10.84 9.20    .10.84
types of PGM in the various ore types is es-                                                               0.36
                                                      Na20       2.60     3.06   1.56    2.28  2,30  2.08  o.18
sential to any fufure exploitation oi these de-       Kzo        0,76    0.89    0.41    0.51  0.87  0.56  0..t9
posits.                                               1.0.1,      t.55    1.59 0.64      n.d.  n.d.  n.o.

                    GBNgsrs                           i:::iiiilrillr;i,"i
                                                                  i ii:linrgiiiiilF^iiiii:,:iri
                                                      from the Eastern Cape(Eales & Marsh1979): -s - Ctriii 6-i;;;e

  _A knowledge of the composition of the paren.
                                                      ;iil ii:     ;;3i;' I"[ii'
                                                                iilI"':il"'        3       i'{3'i,
                                                      e Kelo lylb): / - Standard deviation in average analysls 4
tal magma seems to be of crucial importance           (Eales & l4arsh 1979). *Total iron as Fe203.
614                                  THE CANADIAN MINERALOGIST

the base upward. Such reversed trends suggest            the analysis of Scholtz (1|937) is a true indica-
that supercooling and crystallization of anoma-          tion of magma composition. It is possible to
lous, lower-temperature mineral compositions             calculate theoretical compositions with MgO
may have occurred in the basal portion of the            contents comparable to typical Karroo dolerite
intrusion.                                               by removing varying proportions of olivine-from
     The reverse trend in olivine compositions           the chilled margin composition to see fi olivine
near the base may also relate to the presence            accumulation has taken place. According to
of sulfides. Campbell (1978) studied the Jim-            Eales & Marsh (1979), a typical dolerite com-
berlana intrusion, where disseminated sulfides           position would contain about 7Vo MgO. As
occur at the base. He found that where sul-              ihown in Table 2. l|Vo of. olivine of composi-
fides are present, the olivines are more magne-          tion Foeo has been removed from the chilled
sian than elsewhere, by up to 5 mol. Vo forsteF          margin composition of the Insizwa intrusion, so
ite. Although the effect described by Campbell           that the residual composition contains about
 (1978) may be important, it cannot be the               7Vo Mgp and hence would allegedly-be com-
entire reason for the discrepancy in mineral             parabdto a typical Karroo dolerite. The SiOa
composition. Maske (1966) presented analyses             NarO and KrO contents of this calculated com-
of olivine from the adjacent (and probably               position are now even more distinctly different
 originally contiguous) intrusion to Insizwa, the        ?rom typical Karroo dolerite; SiOa, NarO and
 Ingeli massif. He reported values of Fo* and            KrO aie higher than average dolerite by more
 a reversed differentiation trend from an essen-         than 4, I and 2 times, respectively, the calcrrlated
 tially unmineialized section (Fig. 2). Thus, the        standard deviations. Thus, the composition of
 more magnesian olivines do not always ocsur              the Insizwa intrusion given by Scholtz (1937)
 in association with sulfides.                            cannot be the result of olivine accumulation in
     The reversal in composition of the plagio-           a typical Karroo tholeiite composition, but
 clase, mentioned above, remains a problem even           polnti to the existence of a true liquid with
 if the formation of a sulfide phase were capa-           greater than l0% MgO.
 ble of removing iron from the silicate liquid
 and producing more magnesian mafic minerals.            FonlvrertoN oF AN IMMIScIBLE Sur-rmn LTQUTD
 As plagioclase contains no iron or magnesium,
 its composition should not be influenced by sul-            Silicate magmas contain low concentrations
 fide separation. The progressively more calcic          of Ni, Cu and S. However, if an immiscible
 nature of the plagioclase upward in the basal           sulfide liquid that scavenges Ni and Cu from
 zone (Fig. 2) is contrary to what is expected           the silicate magma forms, a high-grade ore
 for fractional crystallization conditions. Both         deposit may be produced (Maclean & Shimaza-
 Bruynzeel (1957) and Maske (1966) suggested             ki- 1976). There are at least three possible
  that this reversal may be due to more rapid            processes by which the separation of an im-
  cooling of the magma near the base of the in-          miscible sulfide may be induced: (l) Separa-
  trusion.                                               tion may be the result of fractional crystalliza-
     Thirdty, the chemical composition of the            tion. The silicate minerals crystallizing from a
  chilled margin of the Insizwa intrusion apPear$        magma contain no sulfur; thus the concen-
  to be atypical in comparison with average              tration of sulfur in the residual liquid will in-
  compositions of Karroo dolerite. The compo-            crease. As the solubility of sulfur in a silicate
  sition determined by Scholtz (1937) is given in        liquid decreases with decreasing temperature
  Table 2, where it is compared with other typical        (Haughton et al. t974), ultimately a sulfide
  sill compositions from the Karroo. It is apparent      phase will separate. (2) The magma may.react
  that as well as being unusuany high in Mgp,            wittr seaimentary country rocks that are rich in
  the Insizwa composition is enriched in SiO&            sulfur. Selective addition of sulfur may occur,
  NarO and K,o. It has been argued (e.9., by              causing the saturation limit to be exceeded and
  McBirney 1975) that, as a result of contamina-          inducing sulfide separation. (3) Siliceous coun-
  tion, fine grained sample from the margins of           try roJk may be assimilated into the basic
  intrusions may not rcpresent original magma             -ugrrlu. The iulfur solubility is lower i".3 tfi-
   compositions. Although this may be true for            c"our magma than in a basic liquid (Irvine
   large intrusions such as Skaergaard, it is unlikely     lg75).If a basic magma is contaminated by a
   that such a process would have changed the             siliceous melt from the country rocks, sulfide
   MgO content of the chill zone at Insizwa; fur-         saturation may result.
   thermore, any contamination from adjacent                  It is possible to test for these different mech-
   shales would have decreasedthe MgO content.            anisms by geochemical techniques. There is a
   Thus, we believe that the high MgO content in          well-known correlation between Ni/Cu ratios
                         MAcMATTc Cu-NI-PGE        MTNERALTZATToN rNsrzwA
                                                                AT                                     615
   and element abundances and degree of differ-            mas (Naldrett & Cabri t976): thus it is unlikelv
   entiation; if extensive differentiation has taken       that much fractionation occurred before the
   place prior to sulfide separation, the Ni/Cu            sulfide separated. The role of process I is
   ratio of the sulfide will be low. Juvenile mag_         therefore not tholght to be important in the
   matic sulfur has a restricted isotopic rati6,           formation of these ores.
   wh9r9g sedimentary sulfur genera[y has highly               The ratio of sulfur isotopes of sulfides from
   variable isotope ratios; hence su"h data irav           the Insizwa intrusion has been determined by
   indicate the source of the sulfur                       Jensen (1967). There is an extremelv smail
      Major- and trace-element compositions of             scatter,in the ratios, and they average'exactly
  Karroo dolerites are very differeni from those           the value expected for magmatic suttur. Un,
  of their sedimentary hosts, which are mainlv             fortunately, no sulfur from the adjacent sedi-
  sandstones and shales. Consequently, aoy ad_            ments was analyzed, for comparison, but these
  dition of these sediments to the basic magma             data suggest that the sulfur is more likelv to
  would produce a noticeable change in amounts            be magmatic. Even if the adiacent sediments
  of SiOc, NarO, K!O, Rb, Ba and Zr, Such an              had.a sulfur isotope ratio diffirent from mag-
  atypi-c{ silicate composition may indicate the          matic sulfur, such a difference would not com-
  p-ossibility of sulfide segregation resulting from      pletely disprove an origin by contamination.
  the operation of process 3, and a possiSle ex-          As shown below, the possibility exists that sul-
  ample exists. Walker & poldervaart (1941)               fide separation, and hence possible contamina-
  described numerous sedimentary inclusions in the        tion, commenced at depth, not at the present
  Hangaest sill in the Calvinia district of Cape          level of intrusion.
  Province and presented major-element analyses               There are several facts that tend to suDport
  of rocks showing the effects of contamination.         the third process. Scholtz (1937) trequently
  More- recently, Le Roux & Reid (197g) pre_             referred to the association of sulfide with biotit!
  sented trace-element data on the same sill and         and granitic fractions within the lower, picritic
  on an adjacent sill. They reported some differ-        portion of the sill. This led him to suggestthat
 encesin composition between the Hangnest and            the sulfide had actually segregated fiom the
 Blaukrans sills and noted that l ) Rb, Ba and Sr        end product of differentiation (the silica-rich
 are significantly enriched in the Hangnest sill,        residual liquid) of the basic magma. However,
 and 2) SiOr is also slightly higher. They re_           a-sdiscussedabove, such a differentiated liquid
 jected the mechanism of contamination as the
                                                         should contain little Ni and would be incapa-
 sole process for producing these differences be-        ble of producing the Ni-rich sulfide. Alterna-
 cause of the massive depletion in Ni in the             tively, this association of sulfide, biotite and
 Hangnest sill relative to the other sill. The           granite could be interpreted as supporting the
 Hangnest sill contains 3-5 ppm Ni at 7% MgO,    -Ni     hypothesisthat the addition of siliceous malerial
 whereas the Blaukrans sill has a normal                 changed the liquid composition such that it
 abundance of 1OG-12O      ppm at gZo MgO. They         produced a separate, immiscible sulfide phase.
 concluded that, in the absence of anv other            In support of this and as shown eartidi. the
 process, almost IOOVo contamination would be           composition of the chilled margin is enriched
needed to produce this difference. However.             in silica and alkalis.
if contamination caused the separation of an                 It is also of interest to note the analytical
immiscible sulfide phase, the Ni content might          data and conclusions of Eales (1980). He ana-
be drastically reduced by incorporation in ihe          lyzed several picrites from the basal zone at
sulfide phase, so that its deplelion in the re-         Waterfall Gorge and found that one sample
mainder of the sill would be expected.                  was anomalously enriched in incompatible ele_
     In conclusion, it is possible that the differ-     ments, compared with other cumulate rocks
ences in chemical composition between the               in the section. He suggestedthat this was the
Hangnest and Blaukrans sills may be the result          result of a back injection of highly differ_
of contamination-induced sulfide separation,            entiated liquid into the crystal mush. Tiris seems
possibly associated with some degree of frac-
                                                        similar to the model envisaged by Scholtz
tional crystallization.                                 (1937) for the granitic components and biotite
                                                       distributed sporadically throughout the picrite.
Application to the Insizwa intrusion                   It could also be interpreted as the reiult of
                                                       olivine accumulation into a heterogeneously     con-
  The Ni/Cu ratio of the ores from Insizwa             taminated portion of magma, which was eniiched
averages over unity, according to Scholtz              in incompatible elements owing to the assimila-
(1931).- This is slightly higher than typical
                                                       tion of country-rock shale (see Fig. 5).
values for ores associatedwith tholeiitic mae-              Small sulfide aggregates are present in the
616                                THE CANADIAN }VIINERALOGIST

              Frc. 5. Photomicrograph of the top of the chill zone showing olivine pheno-
                crysts surrounded by bronzite, with interstitial plagioclase. Note-the
                extremely resorbed nature of the olivine crystals A and B. The thre€
                small crystals of olivine marked C (and above) are in optical continurty
                and werl probably originally one grain. Such high degrees of corrosion
                suggest a more extensive reaction than is usual in an olivine tholeiite;
                thise crystals may be caught in the process of sinking into a siliceo's,
                contaminated magma. An analysis of this rock gave 15 ppm Rb aqd
                 148 ppm Sr, and hence a rathir hieh Rb and Rb/Sr ratio for a basic
                 rock, consistetrt with the hypothesis of contamination. Photograph covers
                 a width of 5 mm.

chilled margin (Scholtz 1937). These could not       drett 1978). In this regard, it is of interest
have formed after emplacsment of the sill, as        that the olivine that accumulated in the Ele-
cooling would have been too rapid at the mar-        phant's Head intrusion contains about 30@
gins to allow the sulfide liquid to s€parate and     bpm Ni (Fales 1980). The olivine in the pic-
sink in that zone. The presence of sulfide in        rites near the base of the Insizwa intrusion con-
the chill zone suggests that sulfide segtegation     tains from 1000 to 250O ppm Ni (Mitchel
commenced prior to emplacement of the sill.           1980). The reason for the variability of the Ni
If this is correct, the basic liquid was saturated   content of the olivines reported by Mitchell
with respect to sulfide prior to emplacement,         (1980) may be analogous to that proposed
and so the contamination, indicated by the high      by Campbell (1978) and discussed above. If
silica and alkalis, may have commenced at            the sulfide liquid forms only near the base of
depth. However, the heterogeneity observed by        the intrusion, it may not chemically process
Scholtz (1937) in the marginal zone indicates        the entire magma column but only deplete the
that contamination and mixing of the two sili-       lowermost layer in Ni and Cu. Olivine crystal-
cate liquids were still continuing after emplace-    lizing slightly higher up in the intrusion may
ment.                                                form from a liquid that has a more normal
    The Ni content of olivine may be informative     Ni content. However, the important conclusion
in geochemical models relating to sulfide for-       is that the olivine from near the base of the
mation. If a sulfide liquid segregates, it will      intrusion is variably and significantly depleted
 cause a decreasein the Ni content of the re'        in Ni compared with olivine from the Elephant's
maining silicate liquid. Subsequent crystalliza-      Head sill, as would be expected if a sulfide phase
 tion will produce olivine that is depleted in Ni    had segregated at an early stage before the
 compared with typical olivine of similar com-        olivine crystallized, not at a late stage as pro-
 position in terms of Mg and Fe (Duke & Nal-          posed by Scholtz (1937).
                       MAGMATIC    CU-NT-PGE     MINERALTZATION     AT INSIZWA                        6t7

                   Survruarv                          Dowsprr, J.S. & RBrp, N.T. (1967): An explora-
                                                        tion programme for nickel and copper in the
   It is possible to reassess abundant data on
                             the                        differentiated intrusives of East Griqualand and
the Insizwa intrusion in the light of modern            Pondoland. Trans. GeoL Joc. .S. Afr. 70, 67-79.
geochemical parameters. Unlike most Karroo
                                                      Dure, J.M. & Nelonem, A.J. (1978): A numerical
dolerite sills, the chilled margin of the Insizwa       model of the fractionation of olivine and molten
intrusion is a high-Mgo tholeiite, whose high           sulfide from komatiite magma. Earth Planet.
magnesium content is not the result of olivine          Sci. Lett, 39, 255-266,
accumulation. The higfo Ni/Cu ratio of the sul-
fides found near the base of the intrusion also       EALEs, H.V. (1980): Contrasted trace-elemetrt
reflects the existence of a more magnesian liquid.      variations in two Karroo cumulus complexes.
Howevern the higber-than-average SiOr, NazO             Chem. Geol. 29, 39-48.
and KrO contents of the chilled margin suggest
                                                              & Mnnsn, J C. (1979\: High-Me tholeiitic
that contamination may have taken place, a
                                                        rocks and their significance in lhe Karroo central
feature consistent with petrographic evidence.          province. S. Afr. J. Sci.75, 4OO-404.
It is thought that the sulfide phase was formed
because this contamination lowered the solubi-        fleucrrroN, D,R., RoBpB& P.L. & SrnrNen, B.J.
lity of sulfur in the basic magma. Contamination         (1974): Solubility of sulfur in mafic magmas.
would account for the presence of sulfides in            Econ. Geol. 69, 451-462.
the chill zone. Massive ore in association with
acid phases is attributed to further contamina-       IrvrNu, T.N. (1975): Crystallization sequences in
tion in sira. Sogregation of a seParate sulfide          the Muskox intrusion and otler layered intru-
 phase from the high-MgO tholeiitic magma prior          sions. [I. Origin of chromitite layers and similar
                                                         deposits of other magmatic orcs. Geochim. Cos'
 to emplac€ment may have significant implica-            mochim. Acta 39, 99t-lO2O.
tions for the economic viability of the Insizwa.
                                                      JsNseN, M.L. (1967): Sulfur isotopes and mineral
                                                        genesis. /n Geochemistry of Hydrotlermal  Ore
                                                        Deposits (H.L. Barnes, ed.). Holt, Rinehart &
                                                        Winston. New York.
    The paper benefited from the discussions
with Professor A.J. Naldrett and the criticism        KnNYoti, I. (L973): Final report on the Insizwa
of the members of the IGCP -       project 161          investigation carried out for African Selection
field cbnference. For expert help with regard           Trust Exploration (Pty.) Ltd., Arch. Geol. Surv.
to the microprobe    analyses, we are indebted          Trawkei.
to Mr. Peter Fisher, Cardiff. R.G.C. and S.M.
acknowledge financial support from C.S.I.R.'          KrNcsroN, G.A. (1977): The Mineralogy and'
South Africa.                                           Mode of Occurrence ol the Plattnoids and Asso-
                                                        ciated Minerals in the Merensky Horizon of the
                                                        Westem Bashveld. Ph. D. thesis, Univ. London.
                                                      Lr Roux, A.P. & Rero, D.L. (1978): Geochemistry
BnuvNzerr., D. (1957): A petrographic study of          of Karroo dolerite sills in the Calvinia district,
  the Waterfall Gorge profile at Insizwa. Annals        Western Cape Province, South Nrtca. Contt.
  Univ. Stellenbosc& 83, 484-534.                       Mineral. Petrology 66, 351-360.

CesRr, L.J. & Hmnrs D.C. (1972): The new min-         MACLEAN, WJI. & Srurvrazerr, H. (1976): The
  eral insizwaite (PtBis) and new data on nig-         partition of Co, Ni, Cu and Zn between sulfide
  gliite (PtSn). Mineral, Mag. 38, 79+800,             and silicate liquids. Econ. Geol. 7I'' lO49-1O57.

        & I-err.ervrMe, J.H.G. (1976)z The min-       MAsKE, S. (1966): The petrography of the Ingeli
  eralogy of the platinum-group elements from          mountain range, Annals Univ. Stellenbosch 41,
  some copper-nickel deposits of the Sudbury             r-107.
  area, Ontario. Econ. GeoI.7t, ll59-I195.
                                                      McBrRNEy, A.R. (1975): Differentiation of the
Ceurnrn,    I.H. (1978): Some problems with            Skaergaard intrusion. Natwe 259, 691-694.
  cumulus theory. Lithos 1-1^,
                                                      Mrrcnnr-r., A.A. (1980): The Extrusive and In-
CAwrHoRN, R.G. (1980): Hich MgO Karroo                  trusive Basaltic Rocks of the Molteno'tarnestown
  tholeiite and sulfide segregation in the Insizwa      Area, M.Sc. thesis, Rhodes Univ., Grahamstown,
  intrusion, Transkei. S. Alr. l. Sci. 76, 467-470,     South Africa.
618                                  THE   CANADIAN   MINERALOCIST

NALDRETT,  A.J. & Cennr, LJ. (1976): Ultramafic       VERMAAK, C.F. & HeNonxs, L.P. (1976): A
  and related mafic rocks: their classification and     review of the mineralogy of the Merensky Reef.
 genesis with special reference to the concentra-       witb special reference to trew data on the pre-
 tion of nickel sulfides and platinum-group ele-        cious metal mineralogy. Econ. Geol, 71, 1244-
 ments. Ecoa. Geol. 71, 1131-1158.                      1269.
Scnor,ra D.L. (1937): The magmatic nickeli-           Werxtn, F. & Pot-oer,vler.r, A. (1941): The
  ferous ore deposits of East Griqualand and           Hangnest dolerite sill, S.A. Geol. Mag. 78, 429-
  Pondoland. Trans. Geol, Joc. J. Afr. lg, g1-210.     450.
Uvrrxroclennr,     W. & Bunrs, E.A.J.      (1971):
  Tables for Microscopic ldentilication    of Ore     Received December 1980, revised manuscipt    ac-
  M inerals. Elsevier, Amsterdam,                       cepted September 1981,

Shared By: