The composition of hypersaline, iron-rich granitic fluids based on laser-ICP and Synchrotron-XRF microprobe analysis of individual fluid inclusions in topaz, Mole granite, eastern Australia

Geochimica

a Cosmochimicn

Ado

0016.7037/92/$3.00 + .Xl

Vol. 56, pp. 67-79

Copyright0 1992Pergamon Press plc.Printed in U.S.A.

The composition of hypersaline, iron-rich granitic fluids based on laser-ICP and Synchrotron-XRF microprobe analysis of individual fluid inclusions in topaz, Mole granite, eastern Australia* A. H. RANKIN,+ M. H. RAMSEY,' B. COLE&’ F. VAN LANGEVELDE,’and C. R. THOMAS~ ‘Department of Geology, Imperial College, London SW7 2BP, UK 2Department of Engineering, The University of Warwick, Coventry CV4 7AL, UK (Received September 10, 1990; accepted in revisedform October 30, 1991)

Abstract-High-temperature (>55O”C) hypersaline (>50 wt% salts) fluid inclusions, representative of the earliest hydrothermal fluids associated with the Sn-W-Cu-Pb-Zn-mineralised Mole granite of eastern Australia, are well developed in topaz from the Fielders Hill locality. Methods based on Inductively Coupled Plasma Emission Spectroscopy following laser ablation and on Synchrotron X-Ray Fluorescence microanalysis are described and applied to the semiquantitative point analysis of these inclusions. Crushleach analysis provides further information as well as highlighting the importance of point methods when several generations of inclusions are present. The laser-ICP results confirm the dominance of Fe, K, and Na in these early high-temperature fluids. The mean Fe:K:Na atomic ratios (0.95:0.79: 1.OO)are entirely in agreement with published experimental data on the composition of chloride brines in equilibrium with synthetic granite at magmatic temperatures and support the view that these fluids are direct products from a cooling granite magma. A number of trace and minor elements have also been detected in the inclusions. These include Ca, Mg, Li, B, Be, Ba, Sr, and several of the ore metals. Order of magnitude estimates of the ore metal contents of these fluids, based on combined XRF-microprobe and laser-ICP analysis, are in the percent range for Fe, Mn, and Zn, in the range from several hundred to several thousand ppm in the case of Sn, Cu, and Pb, and less than 600 ppm for MO and W. These results have important implications for ore genesis in granitic environments and point to the very high ore-carrying potential of high-temperature, hypersaline, chloride-rich brines exsolved from cooling granite magmas. This paper reports on the analysis of individual type 1 inclusions in topaz from this area primarily using laser ablation in association with Inductively Coupled Plasma Emission Spectroscopy (the laser-ICP method; THOMPSON and WALSH, 1983). A preliminary evaluation of the suitability of the method for qualitative fluid inclusion analysis has already been reported ( CHENERY and RANKIN, 1989). The present paper is a first attempt at obtaining semiquantitative data on ore metal contents of the inclusion fluids. The validity of the laser-ICP results for a range of ore metals is discussed in the light of preliminary Synchrotron-XRF microprobe analysis and scanning electron microscopy. Though FRANTZ et al. ( 1988) have already reported on the results of SynchrotronXRF microanalysis of synthetic fluid inclusions in quartz, the present paper is a first attempt at applying the method to natural fluid inclusions in topaz. Comparisons between the laser-ICP results and those obtained from bulk leachate analysis provides further information but highlights the importance of point methods of analysis when several generations of fluid inclusions are present, and when analyses are required on only one. Implications for ore-metal transport in magmatic hydrothermal fluids and for oxide and sulphide-facies metallogenesis in granitic environments are briefly discussed in the light of these data.

INTRODUCTION

debateintheliteratureover the relative importance of early, high-temperature, granitederived fluids and convectively driven, hydrothermal systems involving fluids from extraneous sources (meteoric and metamorphic waters, seawater, etc.). A knowledge of the composition of these early fluids, especially their ore metal contents, is particularly pertinent to our understanding of ore-forming processes in granitic environments. Limited information is available from experimental studies and theoretical considerations (e.g., EUGSTER, 1985; BRIMHALLand CRERAR, 1987). Analysis of individual fluid inclusions offers a more direct approach provided the inclusions are demonstrably early, large enough to analyse, and the analytical problems can be overcome ( ROEDDER, 1990). Relatively large (>30 cc)fluid inclusions with magmatic characteristics (high homogenisation temperatures and high salinities) are common in porphyry copper-molybdenum systems but are comparatively rare in granite-associated, Sn-W-base metal deposits ( ROEDDER, 1984). A notable exception is the Mole granite of eastern Australia where such inclusions, herewith termed “type 1”) are particularly well developed in an early topaz-quartz rock (“silexite”) associated with W-Sn-base metal mineralisation ( EADINGTON, 1983 ). THEREHASBEENCONSIDERABLE

* Presented at PACROFI III, Third Biennial Pan-American Conference on Research on Fluid Inclusions, held in Toronto, Canada, May 20-22, 1990.

GEOLOGICAL SETHNG MINERALISATION

+Present address: School of Geological Sciences, Kingston Polytechnic, Kingston upon Thames, Surrey KTl 2EE, UK.

AND ASSOCIATED OF SAMPLE STUDIED

Comprehensive reviews of the geology and mineralisation of the Mole granite are provided by EADINGTON ( 1983) and PLIMERand 67

A. H. Rankin et al.

68

KLEEMAN (1985). Unless otherwise stated the following summary is based largely on the accounts of these authors. The Mole granite forms a lenticular, sill-like mass about 650 km’ in an area close to the town of Torrington in eastern Australia. The granite forms part of the larger New England Batholith and was emplaced during early Permian to late Triassic times into a series of mudstones, siltstones, and acid to intermediate volcanic rocks of upper Permian to lower Triassic age (Fig. 1f. A large roof pendant, approximately 7 km in diameter, is the predominant host to an unusual quartz-topaz rock locally named “silexite.” The geology and genesis of these silexite bodies have been described and discussed by EADINGTON and NASWAR (1978) and KLEEMAN (1985). According to these authors the silexite forms small dyke- and sill-like bodies up to a few tens of metres wide, mainly within the roof pendant but also within the granite and associated with granite pegmatites. The rock comprises a medium-gang intergro~h of quartz and topaz. The topaz content varies from 18 to 27 ~01%and contains up to 5 ~01% miarotitic cavities. A variety of minor accessorary minerals have also been reported. These include micas, kaolinite (secondary) wolfmmite, tourmaline, zinnwaldite, beryl, bismuthinite, chalcopyrite. and molybdenite (EADINGTON and NASHAR, 1978; KLEEMAN, 1985; PLIMER and KLEEMAN, 1985). EADINGTON and NASHAR( 1978)note an increase in quartz content and replacement by white mica due to a late stage hydrothermal overprint. It is also apparent that the altered silexite contains the highest economic concentrations of coarse wolframite. A magmatic origin has been proposed for the genesis of the primary (i.e., unaltered) silexite bodies by EADINGTON and NASHAR ( 1978) and EADINGTON (1983). Evidence in support of this hypothesis is based partly on the observed field relationship and partly on the inte~retation that some of the solid inclusions in topaz are trapped silicate glasses (i.e., melts). Magmatic temperatures are also implied from the very high homogenisation temperatures reported from the associated high salinity aqueous fluid inclusions in (570 to over 620°C) which are far more abundant. More recently K.LEEMAN (1985) has proposed an alternative metasomatic or hydrothermal

origin based on field observations. The model of EADINGTON(1983), which invokes a magmatic origin with a localised, late-stage, hydrothermal overprinting, is favoured here. This is based on our own observations that later, lower temperature, lower salinity fluid inclusions are clearly superimposed on the earlier, primary high-temperature, high salinity fluid inclusions in topaz and associated quartz. A variety of tin, tungsten, bismuth, and Cu-PhZ~-s~phide deposits are known throu~out the granite and surrounding country rocks. Wolframite occurs principally as disseminations in hydrothe~aIiy altered (sericitised or ‘greisenised’) quartz-topaz rock, with accompanying minor Mo-Bi-Sn mineral&ion. Tin mineralisation occurs mainly as cassiterite within quartz-bearing lodes and stockworks typically within the surrounding country rocks. Some of these contain a prominent alteration assemblage comprising chlorite-sericite-quartz( t fluorite). Base metal minerali~tion is less common than Sn-W minerali~tion and also takes the form of quartz-bearing lodes containing sphalerite, galena and chalcopyrite. Paragenetically, these appear later than the Sn-W mineralisation (EADINGTON,1983). DESCRIPTION

OF SAMPLE STUDIED AND FLUID INCLUSIONS IN TOPAZ

Fluid inclusions were analysed in topaz from a medium to coarse-grained silexite from a small quarry at Fielders Hill in the Torrington district of New South Wales (Fig. 1). The topaz occurs as clear, well-formed euhedral to slightly elongated crystals averaging about 0.5 cm in size associated with abundant (approx. 7.5-80 ~01%) , medium-~ned granular quartz. The quartz is associated with very small amounts of very tine-grained sericite and kaohnite. Though the silexite from this locality is mineralised, the sample studied here appeared free from admixed metal sulphides and oxides. An Hitachi S2500 superprobe coupled with a Link Ana-

Mod&d from Kleeman. WI!

El

f

Coarse

granite

Silexite (~ua~z-Topaz)

Permian and Carboniferous sediment bodies

Permian vokxnics

Other granites

FIG. I. Geological sketch map of the Mole granite of eastern Australia showing the location of the Fielders Hill silexite.

Chemical compositions of fluid inclusions in topaz lytical Energy Dispersive System (EDS) was used to analyse the topaz from this locality. The window of the EDS detector was thin enough to detect F. The only elements detected, expressed as the mean of six analyses in weight percent, were MgO, 0.5%; A1203, 55.32%; SiOZ, 31.88%; and F, 19.99% (converting fluorine to oxygen equivalents at 8.42%, the total becomes 99.28%). Fluid inclusions in topaz from this locality are particularly large (averaging 80 microns) and abundant, and many are clearly primary in origin. In contrast, as also noted by EADINGTON (1983), the inclusions in associated quartz are much smaller and demonstrably secondary in origin. In topaz, apart from the rare solid (silicate melt) inclusions described and discussed in detail by EADINGTONand NASHAR ( 1978), three different compositional/morphological types of aqueous inclusions are also recognised. The optical characteristics, microthermometric properties, and daughter mineralogy of each of these inclusion types have been described in detail by EADINCTON( 1983). Type 1 inclusions are multiphase, multisolid inclusions. They comprise a small amount of liquid, a vapour bubble, and a large proportion of different daughter minerals. The inclusions typically form rounded or negative-crystal cavities and appear primary or pseudosecondary in origin. Their optical and microthermometric characteristics are summarised in Table 1 and are illustrated in Fig. 2. Type 2 inclusions are also multiphase but differ from type 1 inclusions in that they contain fewer daughter minerals (often only halite) and a smaller vapour bubble. Reported homogenisation temperatures ( EADINGTON, 1983, down to 340°C from his Fig. 3) are also much lower than those for type 1 inclusions. Type 2 inclusions also tend to be more irregularly shaped, more distinctly fracture bound than type 1 inclusions, and are therefore assigned a secondary origin. Type 3 inclusions are vapour-rich. They contain over 90% water vapour (no condensed gas phase on cooling) and only a small amount of liquid. Occasional daughter or captive mineral phases are present. These inclusions appear secondary or pseudosecondary in origin. Homogenisation temperature data are not available due to the difficulty in observing the precise point of homogenisation to the vapour phase. Point analysis using laser-ICP and Synchrotron-XRF methods were restricted to primary type 1 inclusions because these are believed to represent the earliest, pre-ore-mineralisation fluids associated with the Mole granite. Unfortunately, preliminary attempts at obtaining comparable laser-ICP data on type 2 and type 3 inclusions were unsuccessful because,

69

FIG. 2. Large( IO0pm) type 1 inclusion in topaz containing several daughter minerals as described in Table 1.

with the exception of Na, Al, and Si (the latter two from the host topaz), all other elements were too close to background to be meaningful. Further attempts were therefore abandoned. EXPERIMENTAL METHODS Scanning Electron Microscopy Qualitative data on the composition of daughter minerals exposed on the randomly broken surfaces oftopaz were obtained using a Link Analytical AN1000 energy dispersive X-ray analyser attached to a Jeo1733 superprobe following the procedures outlined by SHEPHERD et al. (1985). Crush-Leach Analysis Approximately 1 to 2 g of 1 to 3 mm-sized fragments of topaz crystals were carefully hand-picked, cleaned in Aristar dilute nitric acid overnight, and then rinsed several times in double distilled water (DDW). Care was taken to avoid composite grains with admixed quartz or other mineral contaminants, although a small amount ( <5 wt%) of quartz may, inadvertently, have been included. A clean agate pestle and mortar was used to gently crush the sample in 2.5 mL of DDW following the procedures described in SHEPHERDet al. ( 1985). The resulting leachate (the first crush solution) was transferred into a clean polythene test tube. A further 2 mL of DDW was added to the tube and the solution was centrifuged to separate small fragments of topaz. A 2 mL aliquot of the supernatant liquid was introduced via the standard nebuliser of an ARL 34000C Inductively Coupled Plasma Atomic Emmission Spectrometer (ICP-AES) equipped for the simultaneous analysis of up to 40 elements including Na, K, Li, Ca, Mg, Fe, Mn, S, B, Be, Sn, Pb, Zn, Cu, and W. A second more vigorous crush to a finer grain size was then carried out on the topaz remaining in the aqate mortar. The procedures described above were used to obtain an analysis on this second crush solution.

TABLEI. Summary of microthermometric and optical data for early multisolid inclusions in topaz (after EADINGTON,1983)

Laser-ICP Analysis

I) Homogenisation Temperature 2) Daughter mineral dissolution temperatures: Halite K, Fe, Na, Cl phase Fluoride-phase Unidentified transparent phase Opaques (2) 3) Salinity (Total Salt Content) 4) Mean density

Laser-ICP analysis was carried out using a commercial Carl Zeiss LMA IO laser microprobe equipped with a 1J Ruby laser which is focusscd through a high-quality, transmitted light microscope onto a pre-selected area of the sample. The sample is held within a specially constructed, perspex ablation chamber which sits on the microscope stage. A stream of Ar carrier gas is directed over the specimen, through the stage, and into the inlet of an ARL 34000C ICP-AES via a tenport Teflon valve ( RANKIN et al., 1990). The set-up is similar to that described by THOMPSONand WAL.SH( 1983) for the analysis of metals and alloys and is shown in Fig. 3. The major difference is that the

580 to r620”C 230°C 80 to 100°C 230°C 80 to 85°C 500°C 50 to 70 wt% 1.25 g/cc.

A. H. Rankin et al.

70

A simpiified method of data reduction based on that described by SPARKS( 1976) has been used to anaiyse the results and derive the mass ratios Zn:Fe, Cu:Fe, Pb:Fe, and Fe:Mn in each of the inclusions. The principles are outlined in Appendix 1. The choice of inclusion for analysis by XRF is affected by the nature of the samples and by the Daresbury microprobe. The original choice of excitation energy ( 15 keV) for the microprobe was decided in order to maxim& the ~nsiti~ty of the inst~ment for the detection of the full range of transition metals. This leads to a maximum sensitivity (see the values for X-ray fluorescence cross sections in Appendix Fig. A2) for the heavier transition metals and elements up to Br, but the sensitivity for heavy metals such as Pb and the lighter transition metals is still relatively high (the reduction factor Zn / Mn is about 2.5). The sensitivity for Ti and the lighter elements is, however, signifi~n~y low, and a lower energy excitation beam should ideally be used for the detection of these elements. Adoption in the topaz layer above the inclusion compounds this dependence of sensitivity on atomic number. Adequate counting statistics are still possible with reasonable counting times with inclusion depths up to about 100 p, giving a reduction in fluorescence output of less than an order of magnitude for Fe, Cu, Zn, and Pb and only marginally greater for Mn. The fluorescence from Ti and lighter elements is reduced by more than two orders of magnitude for inclusions of this depth (and by still more for absorption within the inclusion). and even with depths of 10 p the absorption of radiation is significant. Therefore, no attempt has been made to analyse quantitatively the concentration of the lighter elements, and the choice of inclusion was restricted by their suitability with regard to the detection of the elements Mn, Fe, Cu, and Zn. Typically, total counts for each element were in the range 5000 to 250,000, but for the deeper inclusion the minimum count was 570 for Mn and 63,ooO for Fe. The single inclusion analysed in the first polished wafer, was approximately 65 p below surface. The two inclusions examined in the second wafer were approximately 25 p and approximately 90 p below surface. All three inclusions were more or less rounded or ellipsoidal in shape with an average maximum diameter of about 80 CL.With this restriction the inclusions were chosen to be; ( 1) those which were clearly still intact with no signs of contamin~t leakage to surface, (2) of a reasonably regular ellipsoidal shape to minimise the errors associated with the simplified analytical technique specified in Appendix 1, and (3) not near any other inclusion so that results from the probe could be assigned definitely to the particular inclusion of measured depth.

r-----l SPECTROMETER

t /

COMPUTER

1

t Ablailon chamber

OPTICAL SWITCH

\

t

llluminat~on

FIG. 3. Schematic ~p~~ntation of the laser-ICP system showing the ablation chamber and pathway of the Ar carrier gas into the ICP.

present system is based on transmitted light microscopy and allows precise location and identification of particular inclusions below surface. By varying the “Q-switch” setting and focus of the laser beam, the size of the ablation craters in topaz could be controlled down to about 80 p, with penetration depths of the order 100 p. The flow rate of the carrier gas is carefully controlled so that only small fragments of material ablated from the specimen are transferred directly into the plasma (less than 20 p) for analysis. Nebulised aqueous solutions have proved to be reasonable calibration standards for the comparable decrepitation-linked ICP (D-ICP) technique for the analysis of element ratios in fluid inclusions (~0~~0~ and WALSH, 1983; RANKINet al., 1990). This calibration procedure has also been applied to the laser-ICP results reported here. As in the case of DICP analysis, it is only possible to determine relative concentrations of elements in the ablated material derived from the inclusions. These may be expressed as pseoudo-concentration units or as “equivalent parts per million” (eq. ppm) values, though it is more meaningful to present the results as element ratios relative to one of the major elements in the inclusions, either Na or K. Blank analyses were carried out periodically on inclusion-free areas of the topaz, and the data reported in the text have been corrected both for spectral line interferences and base line drift. Synchrotron X-R-F-microprobe Analysis A prototype X-Ray Fluorescence Micropro~, utilising the Synchrotron Radiation Source (SRS) at Daresbury, UK, was used to analyse three type 1 inclusions in topaz. The principle of XRF measurements using synchrotron radiation has been described by SPARKS ( 1980) The instrumentation used at Daresbury has been described by VAN LANGEVELDEet al. ( 1989, 1990). Monochromation and focussing of the X-radiation is performed in a single step with a high precision, eilipsoidaliy-cured Si( 111) crystal mounted on an aluminium support. A flux of 10’ photons/secof IS KeV (quasi-monochromatic) X-rays is focussed down to a spot about 20 p in diameter. Light elements, effectively in the range Si to Br, can be detected by the emission of K-lines, and heavier elements, including Pb, by the L-lines using a Li-drifted Si detector. The specimen is mounted at 45” to the exciting beam, to allow detection at the optimum angle of 90” thereby minimising scattered radiation, and can be traversed with respect to the beam on an X-Y-Z stage in steps of 1.25p. An integral microscope enables the selected inclusion to be identified and roughly positioned at the X-ray focus. A series of point measurements in a grid pattern are then made over an area spanning the inclusion (Fig. 4). The distance between the individual point measurements for the three inclusions investigated was typically 10 p.

PRESE~ATIO~

AND SUMMARY OF RESULTS

Crush-Leach Analysis

The results of crush-leach analysis are summarked in Table 2. These data are expressed as atomic proportions relative to Na (X:Na) to allow direct comparisons with previously pub-

Position.pm FIG. 4. Line profile across the centre of a type 1 inclusion in topaz (inset-inclusion number 1) showing the raw Synchrotron-XRF microprobe data for Fe, Zn, and Pb.

71

Chemical compositions of fluid inclusions in topaz

TABLE 2. Summary of present crush-leach data for two samples of Fielders Hill topaz (Top A and B) compared with some previously published results expressed as atomic ratios relative to sodium

TopA+B

Na K Ca Rb Fe Mn Zn M8 Li Sr Ba Pb cu ! Cl Cl/Total Cations

Top A 1st

Top A 2nd

Top B 1st

Top B 2nd

mean

sd

a

b

b

1.000 ,342 .06 I .008 .340 .078 ,024 bdl ,043 .0003 .0005 .0024 .0008 .046 bdl 2.168 .90

1.ooo .331 bdl .006 .356 .074 .024 bdl ,037 bdl bdl .0018 .0006 .025 bdl 2.227 .95

1.000 ,333 .lOl bdl ,242 .06 1 ,020 .030 .015 .0005 .0005 .OOlO .0003 ,020 bdl 1.878 .85

1.000 .396 .076 bdl ,274 ,057 .019 .019 .019 bld

1.ooo .350 .079 ,007 .303 ,068 .022 .025 .029 .0004

,031 >.020 1.001 ,054 .OlO ,003 >.0008 .014 >.OOOl

1.000 .206 .I40 ,003 .I25 ,034 .014 ,003 ,018 .0009

1.000 .65 .02 .18 .006 -

1.00 .66 .07 .16 <.014 -

bdl bdl bdl ,009 bdl 1.859 .72

.0005 .0017 .0006 ,025 bdl 2.033 .86

.0007 .0003 .016 -

.0006 1.746

-

-

-

-

2.500

2.381

I) 1st and 2nd. Refer to results of first and second crushes. 2) Top A + B mean. Refers to mean of Top A and Top B (1st + 2nd crushes). 3) Column sd. Standard deviation of Top A and Top B. 4) Column a. Mean values reported by BOTTRELLand YARDLEY(1988) for inclusions in Topaz from SW England. 5) Column b. Duplicate analyses reported by EAD~NGTON (1983) for inclusions in Topaz from Fielders Hill. 6) bdl. Below Detection Limit.

lished crush-leach data on topaz from this locality ( EADINGTON, 1983) and with some recent results from broadly similar, high salinity, high-temperature inclusions in topaz from the Sn-mineralised granites of southwest England ( BOTRELL and YARDLEY, 1988). The element ratios determined

for the first and second crush for each topaz are closely comparable. However, slight differences are noted between the two samples, especially in the Fe, Mn, and Li ratios relative to Na and the slightly elevated but variable calcium ratios in the second topaz sample. The results of EADINGTON ( 1983) show much lower Fe:Na ratios than the present analyses, probably reflecting differences in sample localities and the relative proportions of different inclusion types which they contain. Many of the minor element ratios reported by BOTRELLand YARDLEY( 1988) are similar to those reported for the Fielders Hill topaz samples. However, their K:Na and Fe:Na ratios are significantly lower. The present crush-leach data for the Fielders Hill topaz samples demonstrate the importance of K, Fe, Na, and Mn in the inclusion fluids. The results also point to the dominance of chloride over sulphate and borate, although the charge balance ratios, around 0.8 for the second sample of topaz, suggest that other anions, most likely fluoride ( EADINGTON, 1983 ), may also be present. SEM analysis of daughter phases in individual inclusions (Table 3) and the reported daughter mineralogy (Table 1) confirm the dominance of K, Fe, Mn, and Na chlorides in the inclusion fluids. Laser-ICP Laser-ICP analysis is literally a “hit or miss” technique, and it has proven very difficult to predict the likely outcome of a laser shot on a particular inclusion. Only large (60 to

120 p) daughter mineral-rich inclusions identified as type 1 inclusions were selected for analysis, and only those less than 100 ~1below surface were considered suitable for laser ablation. Analysis of type 2 and 3 inclusions was not possible because of the poor response for most of the elements. About half of the attempts to produce sizeable craters on the nearsurface, type 1 inclusions were unsuccessful. Most of these failures were due to the poor laser coupling with the inclusion resulting in little or no ablation. Others were judged unsuccessful either because the resulting crater was too large (>250 p) or the sample cleaved, releasing part of the inclusion contents into the fracture rather than up into the Ar carrier gas. Altogether 59 craters of adequate size and quality were produced on selected type 1 inclusions, penetration depths were of the order of 80 p which is at least equal to the distance below surface of most inclusions. The efficiency of laser ablation in releasing all the solid and liquid contents of the inclusions into the Ar carrier gas may be low or variable if some of the components, especially daughter minerals, are left behind in the inclusions. It is, therefore, important to confine acceptable results to those inclusions where ablation efficiency is at a maximum for quantitative analysis. Scanning electron microscopy has shown that even partially ablated inclusions are often remarkably “clean” which suggests that most of their contents have been released into the Ar carrier gas (Fig. 5). Matters are complicated if any of the ablated material (i.e., fragments of daughter minerals greater than about 20 ~1in larger inclusions) fails to enter the plasma, or if the ICP responses for elements are too close to backround in the case of smaller inclusions. In an attempt to alleviate these problems a semi-empirical approach has been adopted to restrict the data set to “ac-

12

A. H. Rankin

TABLE 3. Summary Phase morphology

of major elements

and relative size

Large, rounded cube Large rounded cube Large, sub-rounded rhombo-cuboid Small rectangular phase Small cuboid-rectangle Small rounded Irregular globules

phase

detected

et al.

by EPMA in daughter

Elements

detected

mineral

major and (minor)

Possible identity

but the random errors reduce approximately assymptotically as the K response increases. For individual inclusions the best estimate of their Fe:K ratios is where the boundary between acceptable and unacceptable values corresponds to a K concentration value of 0.5 eq. ppm. This boundary also corresponds to where the random error in the Fe:K ratios falls to 50%. Much the same trends are obtained by plotting other element ratios against K. The validity of this empirical approach may be illustrated with reference to the data presented in Fig. 7 where all values for Cu, Fe, and Sn which are above background levels for the ICP are ratioed against the corresponding K values. The X:K ratios for high K responses (above 0.5 ) show a very much tighter grouping than the wide spread of data for low K responses. Acceptable element ratios (X:K) corresponding to K levels above the cut-off value of 0.5 (a total of 18 inclusions) are summarised in Table 4. The major elements detected in the inclusions are K, Na, Fe, and Mn with occasional high values of S and Mg. Zinc appears above background in most of the analyses and is also an important component of the inclusion fluid. A range of other elements, including many of the ore metals have also been detected. Their low X:K ratios, however, suggest that they are not present in very high concentrations in the inclusion fluids ( < 1% ). The elements Mg, B, S, and Be show the largest variations in terms of their X:K ratios, covering at least an order of magnitude, and also the largest standard deviation relative to the mean value. Elements showing the least variation about their mean X:K ratios are the major elements Na, K, and Fe together with Zn and Pb.

3.4 /

2.6.

,g I2 3 Y i

t _+ ++ +‘,

,

+ +

,

,

,

,

,

1

i

2.2 -* ++ : +

+

-+++++++ t 1.6 -+++++ + + - ++t + 1.0 _B

++

0.4 1 0.1

FIG. 5. SEM photomicrograph of three ablation craters on the polished surface of topaz. The bottom crater shows minor debris surrounding the breached inclusion whereas the top left inclusion is “clean.” The top right crater has partially breached three inclusions with incomplete release of daughter minerals. In the latter case this would be deemed an “unsuccessful” attempt and the ICP data would be rejected (scale bar = 100 p).

in topaz

Halite Fluoride or Chloride of Fe? Complex Chloride Fe-Mn-chloride Chalcopyrite Sphalerite Precipitate from aqueous fluid

Na, Cl (Fe) Fe (Cl) Fe, Cl (K) Fe, Mn (Cl) Fe, Cu, S Zn, S K, Cl, Fe, Mn (Na)

ceptable” values-those significantly above background with a restricted spread of data points. This has been achieved by selecting one of the major elements (K), assigning a suitable cut-off value for the absolute ICP response for this element, and only considering element ratios for inclusions where K is above a predetermined fixed limit. Potassium is chosen in preference to the other elements because it is a major component of both the inclusion fluid and associated daughter minerals. The rationale behind this approach and for selecting the appropriate limit is shown in Fig. 6 where the K responses for the whole data set of 59 analyses are plotted against the Fe:K ratios. The spread of Fe:K ratios is very much greater for low K responses than for higher K responses. The mean Fe:K ratio of I .8 is more or less constant for all K responses

phases in inclusions

1

1 0.7

* +

++

+

+

+

*

+ + 1

1 1.3 K mm

1

1 1.9

1 2.5

3.1

(eppm)

FIG. 6. Laser-ICP results for K expressed as equivalent parts per million (eq. ppm) plotted against the corresponding Fe:K ratios for 59 inclusions in topaz. The dashed line marks the boundary between acceptable and unacceptable results corresponding to an Fe response of 0.5 eq. ppm.

13

Chemical compositions of fluid inclusions in topaz q

3.6

High K response

I

q

Low K response

'I'1

'I'

I

’

Many of the element ratios, notably Si and Al, were well above their detection limits for all ablated inclusions, but because these elements are principally derived from the topaz host, they have not been considered further. Checks on possible contamination from minor elements within the topaz lattice were carried out on ablated material released from inclusion-free areas of the topaz. All the elements listed in Table 4 were below their detection limits for these blanks.

I

Synchrotron-XRF

0

0.08

0.16

0.24

0.32

0.40

0.48

SniK

? 1.8 9

0

0.04

0.08

0.12

0.16

0.20

CuiK FIG. 7. Laser-ICP results for Sn:Fe and Cu:Fe ratios plotted against corresponding Fe:K ratios for individual inclusions in topaz. The ratios where the ICP response is low (co.5 eq. ppm) are considered unacceptable. Where the K responses are high (>0.5 eq. ppm) the Sn:Fe and Cu:Fe ratios form a much tighter grouping and the results are considered acceptable.

Microprobe Results

Spectral analysis of the data indicates the presence of a number of elements in all the three areas analysed. These include Si, Ar, Ti, Cr, Mn, Fe, Cu, Zn, Ga, Ge, Pb, Br, and possibly Ba. Calcium was unambiguously detected in only one sample. Fluorescence cross sections for elements lighter than Si are too small to make XRF microprobe analysis a suitable technique for such elements. Quantitative XRF analysis using the procedures outlined in the Appendix is restricted here to elements heavier than Ti. It is clear from profile analysis across the inclusions that only some of these elements are predominantly associated with the inclusions rather than the matrix (Fig. 4). Quantitative estimates of the elements detected in the inclusions are here restricted to the ore metals Fe, Mn, Cu, Zn, and Pb which have also been detected by laser-ICP analysis. Though Sn appears as a minor constituent of the inclusion based on laser-ICP results it cannot be determined at low concentrations by the XRF microprobe system currently used. Spectral line overlaps with Cu, Zn, and Ge preclude the detection of traces of W. Chromium and barium are too close to background levels to unequivocally assign to either host mineral or inclusion. Silicon, gallium, and germanium are universally

TABLE 4. Summary of acceptable Laser-ICP results, expressed as weight ratios standardised to potassium and mean atomic ratios relative to sodium. For comparison crush-leach data from Table 2 are also included

Laser

‘X’ K Zn Fe Na

Mg Ca Pb P Li Sn B cu Be Sr Ba

Number of analyses 18 16 18 18 I 10 17 4 I 6 11 I1 10 9 12 4 2

Element minimum 1.00 .08 .91 .42(*) .008 .23 .I1 .038 .040 .005 .023 .008 ,005 .0004 ,001 ,004

Weight ratio maximum

(X:K) mean

sd

1.00 .22 2.31 1.34 1.70 1.47

1.oo .18 1.73 .75 .39 <.lO .80 .83 .06 .06 .03 .04 .04 .02 ,006 .0015 ,007

.04 .35 .38 .62 .30 .55 .02 .02 .02 .02 .06 ,007 ,004 .oo 1 ,003

1.30 .ll .08 .07 .07 .22 .03 .Ol ,003 .009

(*)-Excluding one anomalous value for a secondary halite-rich inclusion at 0.2 1 (**)-These results are probably erroneous-see text. -W, Bi and MO below detection limit; X:K values all ~0.0 1. sd = standard deviation

-

X:Na At. mean .I9 ,008 .95

1.oo .49 1.07 .45 .80 ,009 .06 .13 .Ol .ll .Ol

.02 .0005 .0015

Crush leach X:Na At. mean .35 ,022 ,303 ,025 ,079 ,068 .0017 .029 ,025 .0006 .0004 .0005

A. H. Rankin et al.

74

recorded in all areas of analyses and are therefore identified as components of the host topaz rather than the inclusions. Although it is possible, using suitable calibration standards, to obtain absolute quantitative data on element concentrations in inclusions in quartz by XRF microprobe ( FFUNTZ et al., 1988), we have not yet attempted this in the case of topaz. All results are therefore reported as element weight ratios relative to Fe (Table 5 ) . On average, the proportions of Cu, Zn, and Pb are roughly the same, with X:Fe ratios around .07. However, it is important to note that the Cu:Fe ratio shows a much greater variability between samples than the other two base metals. The Mn to Fe ratio is consistently greater, by a factor of two to three, than the other ratios. Estimates of errors associated with the ratios by Synchrotron-XRF depend significantly on the depth of the inclusion analysed. Typically (for the Zn:Fe ratio) the uncertainty in the reported X:Fe for the most shallow inclusion is of the order of 25% and for the deepest inclusion of the order of 50%. For PbFe in the deepest inclusion, where errors are greatest, the uncertainty in the reported result could be as much as a factor of two. For comparison, the mean X:Fe ratios determined by laserICP are also presented in Table 5. Notable differences exist between these two sets of data and are discussed below. DISCUSSION

AND INTERPRETATION

OF RESULTS

Inspection of data presented in Tables 2 and 4 reveals discrepancies in the major element ratios determined by laserICP and crush-leach analysis. These differences are more readily apparent when the data are presented as mean values in graphical form (Fig. 8). Major Elements Both methods show that K, Fe, and Na are major components in the inclusion fluids though the Na:K ratios are lower by a factor of 2 for the crush-leach analyses. Calcium appears as a major component in the crush-leach analysis but only appears as a minor element in the laser-ICP results. The data for Zn and Fe (relative to K and Na) are in rea-

sonable agreement but the X:K ratios determined by laserICP for Mg, Mn, and S are greater by a factor of at least three than the comparable results from crush-leach analysis, In the latter three cases, the laser-ICP data set is small and the standard deviations are very high compared to the mean. This is especially true in the case of S which often shows unex~tedly high and sporadic values during analysis of inclusion fluids by the broadly comparable, bulk D-ICP method ( RANKIN et al., 1990). The cause of the wide variations in S levels in the analysis of inclusion fluids by both the D-ICP and laser-ICP is presently unclear. However, in parallel with the D-ICP method where S analysis is considered to be unreliable, we also consider the laser-ICP data for S to be spurious. The same might also apply to Mg because the high levels indicated by the laser-ICP results are not supported by SEM analysis of individual inclusions. High Mn levels on the other hand are supported by both the SEM data and also the Synchrotron results discussed below. The obvious explanation for the different ratios of K, Na, Fe, and Ca in the fluids is that the laser-ICP results are restricted to early high temperature fluids (type 1 inclusions) whereas the crush-leach results relate to mixtures of both early and late inclusions (types 1, 2, and 3). To test this hypothesis we have obtained a rough estimate of the volumetric pro~~ions of each inclusion type in the topaz sample. This was carried out under the microscope using conventional fluid inclusion wafers (about 300 Jo in thickness). Each inclusion greater than 5 p in size, in 80 randomly selected, fields-of-view (570 inclusions in total), was assigned to a particular type and size interval (S-50 pm, 5 l-100 pm, IOl150 pm). These data were then converted to volume estimates, based on mean radii and a model spheroidal inclusion shape, as follows: 40.1% type 1, 56.2% type 2, 3.7% type 3. Although we have insufficient laser-ICP data on the composition of individual type 2 and 3 inclusions to rigourously test this hypothesis, qualitative data on a limited number of type 2 inclusions shows that Na was the only element substantially above background. Furthermore, it is evident from the simpler daughter mineralogy (halite, predominantly) of the more abundant type 2 inclusions that they are likely to be more depleted in K and Fe, relative to Na, than the earlier

TABLE 5. Synchrotron-XRF microprobe results expressed as X:Fe weight ratios for three type i fluid inclusions in topaz compared with mean X:Fe weight ratios based on laser-ICP anaivses

Synchrotron-XRF

1 Inclusion number Depth (pm) 65 Zn:Fe Cu:Fe Pb:Fe Zn:Fe Sn:Fe

.048 .I00 .032 .156 -

2

3

23

90

.080 ,141 .087 .I55 -

.06 1 ,023 .086 ,244 -

Mean -

,063 .0@3 ,068 .I85 -

sd -

,016 ,060 .031 .051 -

Laser-ICP mean

Bias

I

,104 .012 .035 .462 .023

.041 .076 .033 ,277 -

0.73NS 0.61NS 3.13NS -

cv%-Coefficient of variation sd-Standard deviation of XRF microprobe results Bias-Refers to absolute difference between the mean of the XRF results and the mean of the laser-ICP results t-Refers to the value of Student’s ‘t’ test for statistical significance of Bias between XRF and laser-ICP results NS-Bias is NOT SIGNIFICANT at the 95% confidence level (t < 4.3).

1.48NS

Chemical compositions of fluid inclusions in topaz MAJOR ELEMENTS 2

n q

Laser-ICP Crush-Leach

0 .;;; I 01 .$ Y x

0 K

Zn

Fe

Na

Mg

Ca

Mn

S

Element

MINOR ELEMENTS 0.08

n q

Laser-ICP. Crush-Leach

n 0.06

15

showed that the Na, K, Ca, and Fe ratios of fluid in equilibrium with rocks of granitic composition are largely independent of f0, and salinity, and are primarily a function of temperature. At 400°C Na, K, and Ca are the dominant cations. At 500°C the levels of Ca fall off and the Fe concentration increases reaching a maximum at 600°C before falling off above this temperature. Figure 9 shows the relative proportions (atomic X:Na) of Na, K, Fe, and Ca in a buffered (hematite-magnetite) 1 N chloride solution equilibrated with a synthetic quartz monzonite at 1 kb as a function of temperature. Our choice in presenting data for the hematitemagnetite buffer is arbitrary. We do not imply that the inclusions were necessarily trapped under theseJO conditions, and it is important to note that choice of data buffered by NiNO or QFM would show virtually identical distributions of Na, K, Fe, and Ca as a function of temperature. The mean values for the Na, K, and Fe atomic ratios determined by laser-ICP are also plotted on this diagram and show a marked similarity to the composition of the chloride brine equilibrated at 600°C. In contrast, fluids equilibrated at much lower (400°C) and higher (700°C) temperatures show significantly lower Fe:Na ratios. Similarly, for Ca, there is no significant difference between the laser-ICP values for the Ca:Na ratios (below the detection limit of 0.07) and the experimental value for fluid equilibrated at 600°C (0.05). These observations are entirely in agreement with estimates of trapping temperatures for type 1 inclusions of around 5.50 to 650°C ( EADINGTON, 1983), attesting to the validity of the laser-ICP method at least on these samples and for these elements. Furthermore, the crush-leach data show lower K:Na and Fe: Na ratios and higher Ca:Na ratios which, according to the data of WHITNEY et al. ( 1985), would imply much lower average equilibration temperatures. This supports our view that the crush-leach results represent a composite analysis of both high- and low-temperature fluids in these topaz samples. Minor Elements

I.1

Sll

II

C‘u

k

Element FIG. 8. Comparison of X:K weight atomic ratios for major and minor elements of individual type 1 inclusions determined by laserICP and of the total fluid inclusion population in topaz based on bulk crush-leach analysis.

type 1 inclusions.

Type 3 inclusions are volumetrically insignificant in contributing to the crush-leach analyses. An independent assessment of the validity of the laser-ICP data (at least for the major elements) for early high temperature inclusions in topaz can be made with reference to the experimental data provided by WHITNEYet al. ( 1985 ) These

authors determined equilibrated with comparable to the bearing) over the

the composition of S-free chloride brines synthetic granitic compositions broadly Mole granite (biotite and/or magnetitetemperature range 400 to 700°C. They

Comparison of the mean X:K ratios for minor elements (Fig. 8) shows that those determined by laser-ICP are very much greater than the comparable ratios determined by crush-leach analysis. Partly, this may be explained, as above, by the composite nature of the crush-leach method. However, a further likely cause is that many of these minor elements may be located in small, insoluble daughter minerals which do not dissolve in the leachate solutions but are released during laser ablation. Experimental data on the minor element distribution in chloride brines comparable to the major element data of WHITNEY et al. ( 1985) are not available, and a comparable assessment of the validity of the laser-ICP for minor elements is not possible. However, the Synchrotron-XRF microprobe data does, at least, provide an independent estimate of certain of these minor element ratios. The mean X:Fe ratios (Table 5) for some elements (Mn and Zn) determined by laser-ICP are slightly higher by a factor of 1.5 and 2.5, respectively. In the case of Pb, the laser-ICP results appear to be a factor of two lower. However, using the t-test the difference is not statistically different (at the 95% confidence level). Within

A. H. Rankin

Whitney et al. (1985)

0

et al.

Hem-M! Buffer

0.8

._ z II: .o

0.6

E 2

0.4

0.2

Na

K

ca

FC

Element FIG. 9. Comparison of the mean laser-ICP results with the experimental data of WHITNEY et al. ( 1985) for a 1 N chloride brine equilibrated with synthetic granite at 1 kb, buffered by hematite-magnetite (H-M) at various temperatures (400, 500,600, and 700°C). Note the close comparability between the laser-ICP data and experimental data at 600°C.

the limits of experimental error, these differences are acceptable in view of the small number of inclusions analysed by the XRF microprobe. The Cu:Fe ratios show the widest variations between individual analyses and also the greatest disparity between the mean values determined by these two methods. The reasons for this are not entirely clear. However, SEM analysis has revealed the presence of chalcopyrite daughter (or captive?) minerals in the inclusions and a sporadic distribution or incomplete ablation of this phase could easily account for such variability. Estimation of Metal Concentrations

in the Inclusion Fluids

A rough estimate of the average metal contents of the early type 1 inclusions may be obtained from the mean element ratios reported in Tables 2 and 4. The estimates given in Table 6 are based on the assumptions that ( 1) the inclusions are essentially chloride brines and (2) the salinity or total dissolved salt content (TDS) is constant. The validity of the first assumption is apparent from the results of leachate and SEM analysis which show a predominance of chloride over sulphate and borate. Small amounts of undetected fluoride may be present, but the errors in the estimates provided in Table 6 are likely to be less than about 10% if chloride is taken to be the only anion. The second assumption is also valid (see Table l), and for the purposes of calculation a mean salinity of 60 wt% TDS has been assumed. The slight variations in salinity for these type 1 inclusions (50 to 70 wt% TDS) would introduce an error of not more than 20% in our estimates. It should be emphasized that the values reported in Table 6 are only mean values. The random error in the laser-ICP measurements is quantified as standard deviation (Table 4, column 5). Propagation of this error in to the average metal contents (Table 6) produced coefficients of variation typically between 20 and 50%. The variances of

both the XRF and laser-ICP data in Table 5 have been considered in the comparison of results to detect significant bias. Implications for Ore Genesis in Granitic Environments The extremely high concentrations of Fe, at around lo%, is one of the most notable features of the inclusions studied

TABLE 6. Estimates of the average metal contents of high temperature (approx. 6OO”Q hypersaline granite-associated fluids based on laser-ICP results. (Figures in parentheses refer to estimates based on XRF microprobe data) Element

wt% in solution

Fe K Na Mn

9.9 5.7 4.0 4.7 (1.8) 2.2 1.O (0.6) <0.6 0.34 (0.67) 0.23 0.17 0.11 (0.87)

Mg Zn Ca Pb Sn Li cu Ba Be Sr wt% metals wt% chloride Total

cv% 20

:; (28) 159 22 (25)

0.04 0.03

0.01

88746) 50 66 35 (68) 43 66 66

28.44 31.56 60.00

W, MO, Bi all less than 0.06 wt% B = 0.23 wt% P = 0.34 wt% S = indeterminate (range of S:K ratios too high) cv% is the coefficient of variation calculated from reulicate ICP results in Table 4. (sd * loo/mean)

laser-

Chemical

compositions

here but is entirely compatible with the experimental data of WHITNEY et al. (1985). Iron-bearing daughter phases, suggestive of Fe concentrations in the percent range, are quite typical of high-temperature, high salinity inclusions associated with many other Sn-W deposits worldwide (reviews by NAUMOV and SHAPENKO, 198 1; ROEDDER, 1984; KWAK et al., 1986). The present estimate is close to the maximum value of 9% recorded by KWAK et al. ( 1986) and is in close agreement with their empirical curve for Fe solubilities in high temperature brines as a function of salinity. The very high concentrations of Mn and Zn (at the percent level) and of Pb, Sn, and Cu (several thousand ppm) in the fluids clearly illustrates the very high ore-carrying potential ofthese high-temperature, chloride-rich brines in granitic environments. Experimental studies and thermodynamic predictions on the solubility of Sn, Mn, and Cu, and Pb and Zn in chloride brines, recently reviewed by EUCSTER( 1985 ) and EUGSTER and WILSON ( 1985), support the fluid inclusion evidence and demonstrate that such high concentrations may be attainable at the temperatures and chlorinities indicated (600°C and around 30 wt% or 0.9 molar/Kg of solution). Tungsten and molybdenum are also readily transported as chloride complexes at high temperatures (CANDELA and HOLLAND, 1984; MANNING, 1984). Both elements form significant mineral deposits associated with the Mole granite, and their lower concentrations than other metals in the inclusion fluids (~600 ppm) is somewhat unexpected. A possible explanation is that the lower MO and W contents may reflect differences in the partition coefficients for these metals between chloride brines and granitic melts (& = { cone. of metal in fluid} /cont. of metal in melt]) and the dependence of these parameters on the chlorinity of the brine. For Cu, Mn, and Zn, the Kd values are strongly dependent on salinity rising to 50 for brines where the chloride concentration is 6 molar/Kg of solution (HOLLAND, 1972; CANDELA and HOLLAND, 1984). In contrast, the published K4 values for MO and W are substantially lower (less than 5) and only show slight variations with changes in the chloride content of the brines (CANDELA and HOLLAND, 1984; MANNING, 1984). No data are available for Sn. The origin of these hypersaline, metalliferous brines is clearly linked in some way to the evolution ofthe Mole granite and associated silexite bodies. It is difficult to imagine how extraneous fluids, which have simply equilibrated with the granite shortly after it solidified, could have acquired such high salinities. The existence of vapour-rich inclusions in these topaz samples might imply that boiling had occurred at some stage, but such inclusions are often secondary and extreme boiling conditions would have to prevail to increase salinity by more than a few weight percent. Therefore, we favour a direct magmatic source for these early fluids which we believe exsolved directly from the cooling granitic magma. Direct precipitation of oxide-facies mineralisation (cassiterite and wolframite) is possible from this brine, but sulphide-facies, base metal mineralisation would require an independent source of S. EADINGTON( 1983) has noted a marked decrease in the temperature and salinity of the ore fluids associated with the Mole granite during the later base-metal sulphide stages compared with the earlier W-Sn stages and has sug-

of fluid inclusions

1-l

in topaz

gested that this implies mixing with cooler, more dilute fluids (groundwaters) which probably interacted with the surrounding country rocks. It is quite probable that these fluids and the surrounding sediments provided the reduced S necessary for the precipitation of the base metal sulphides. CONCLUSIONS Though there are several reports in the literature on the application of laser microprobes to the qualitative and semiquantitative analysis of fluid inclusions in minerals (reviewed by ROEDDER, 1990), the present study is one of the few attempts at providing quantitative and semiquantitative data on the ore metal contents of inclusion fluids. A method is described which limits the data set to “acceptable” values. Calibration of the system using synthetic fluid inclusions in quartz was attempted but proved problematical due to the small quantities of fluid in the small (usually less than 50 I*), flat inclusions currently available. Despite the uncertainties over the present calibration procedure using nebulised solutions, and possible sources of error caused by incomplete or inefficient transfer of material from the inclusion into the plasma, the results for many elements are in broad agreement with theoretical and experimental predictions of the likely compositions of high salinity fluids, exsolved from, and equilibrated with, granite at high temperatures. The very high concentrations of ore metals in the inclusion fluids attest to the ore-forming potential of granitic fluids associated with the Mole granite and strongly support a magmatic origin for ore metals at this locality. Very high-metal concentrations in the inclusion fluids are indicated by preliminary Synchrotron-XRF microprobe analysis despite the high variability of the analysis and resultant discrepancies between the results obtained by these two methods which have yet to be resolved. In the wider context, the results are of direct relevance to the current debate over the source of metals and ore-forming fluids associated with mineralised granites elsewhere in the world. Acknowledgments--We

thank the staff at The SERC Synchrotron Radiation facility at Daresbury and our colleagues at Imperial College (especially Dr. Jamie Wilkinson), the University of Warwick, and the Free University of Amsterdam for their help and encouragement. We also acknowledge the financial support provided by the NERC for studies on metal contents of fluid inclusions and SERC for the development of the XRF microprobe system. The samples of silexite were collected by AHR during a visit to Australia in 1982. I am most grateful to Dr. John Angus (Goldfields, Australia) who made this trip possible and for his support and hospitality during my visit, and to the mining and exploration geologists for their guidance in the field. Finally, we are grateful to the two anonymous reviewers for their comments and suggestions which helped clarify several ambiguities in the original manuscript.

Editorial handling: J. R. Bodnar REFERENCES BOTTRELL S. H. and YARDLEY B. W. D. ( 1988) The composition of a primary granite-derived ore fluid from SW England, determined by fluid inclusion analysis. Geochim. Cosmochim. Acta 52,

585-588.

78

A. H. Rankin et al.

BRIMHALLG. H. and CRERARD. A. ( 1987) Ore fluids: Magmatic to supergene. In Thermodynamic Modelling ofGeologicalMaterials: Minerals, Fluids and Melts (ed. I. S. E. CARMICHAELand H. P. EUGSTER);Reviews in Mineralogy 17, pp. 235-254. Mineral. Sot. Amer. CANDELAP. A. and HOLLANDH. D. ( 1984) The partitioning of copper and molybdenum between silicate melts and aqueous fluids. Geochim. Cosmochim. Acta 48, 373-380. CHENERYS. R. N. and RANKIN A. H. ( 1989) The use of a laser ablation microprobe attached to an ICP spectrometer for the elemental analysis of individual inclusions. 10th Meeting on European Current Research on Fluid Inclusions. Imperial College, London, Abstracts Vol., p. 2 1. EADINGTONP. J. ( 1983) A fluid inclusion investigation of ore formation in a tin-tungsten mineral&d granite, New England, New South Wales. Econ. Geol. 78. 1204- 122 1. EADINGTONP. J. and NASHARb. ( 1978) Evidence for the magmatic origin of quartz-topaz rocks from the New England batholith. Australia. Contrib. Mineral. Petrol. 67, 433-438. EUGSTERH. P. ( 1985) Granites and Hydrothermal ore deposits: A geochemical framework. Mineral. Mag. 49, 7-23. EUGSTERH. P. and WILSONG. A. ( 1985) Transport and deposition of ore-forming elements in hydrothermal systems associated with granites. In High Heat Production Granites, Hydrothermal Circulation and Ore Genesis; Inst. Mining Metail. (London) Spec. Publ., 87-98. FRANTZ J. D., MAO H. K., ZHANG Y. G., WV Y., THOMPSON A. C., UNDERWOOD J. H., GIAUQUER., JONESK. W., and RIVERS M. L. ( 1988) Analysis of fluid inclusions by X-ray fluorescence using synchrotron radiation. Chem. Geol. 69, 235-244. HOLLANDH. D. ( 1972) Granites, solutions and base metal deposits. Econ. Geol. 67,281-301. KLEEMANJ. D. ( 1985) Origin of disseminated wolframite-bearing quartz-topaz rock at Tonington, New South Wales, Australia. In High Heat Production Granites, Hydrothermal Circulation and Ore Genesis. Inst. Mining Metall. (London) Spec. Publ., 197-202. KRAUSEM. O., NESTORC. W., SPARKSC. J., and RICCIE. ( 1978) X-ray fluorescence cross sections for K and L X-rays of the elements. Oak Ridge National Laboratory Report ORNL-5399. KWAK T. A. P., BROWNW. M., ABEYSINGEP. B., and TAN T. H. ( 1986) Fe solubilities in very saline fluids: Their relationship to zoning in some ore deposits. Econ. Geol. 81, 447-465. VANLANCEVELDEF., BOWEND. K., TROS G. H. J., and VIS R. D. ( 1989) Non-imaging optics for photon probe microanalysis at the SRS, Daresbury (UK). Proc. XII Interl. Conf. on X-Ray Optics and Microanalysis, Cracow, Poland. VAN LANCEVELDEF., TROS G. H. J., BOWEND. K., and VIS R. D. ( 1990) The synchrotron radiation microprobe at the SRS, Daresbury. UK and its applications. Nucl. Instr. Meth. B49, 544-549. MANNINGD. A. C. ( 1984) Volatile control of tungsten partitioning in granitic melt-vapour systems. Inst. Mining Metall. Transacts. (Sect. B) 93, 185-189. NAUMOVV. B. and SHAPENKOV. V. ( 1981) Evidence from fluid inclusions on the iron concentrations in high temperature chloride solutions. Geochem. Interl. 17, 125- 13 I. PLIMERI. R. and KLEEMANJ. D. ( 1985) Mineralisation associated with the Mole granite, Australia. In High Heat Production Granites, Hydrothermal Circulation and Ore Genesis; Inst. Mining Metall. (London) Spec. Publ., 563-510. RANKIN A. H., RAMSEY M. H., HERRINGTON R. H., JONESE., COLES B., CHRISTOULA M. ( 1990) Current developments and applications of decrepitation-linked and laser-ablation-linked ICP techniques for the geochemical analysis of fluid inclusions in minerals. 8th IAGOD meeting-COFFI Session No. 3, Abstracts volume, p. A 19. Carleton University, Ottawa, Canada. ROEDDERE. ( 1984) Fluid Inclusions: Reviews in Mineralogy 12. Miner. Sot. Amer. ROEDDERE. ( 1990) Fluid inclusion analysis-Prologue and epilogue. Geochim. Cosmochim. Acta 54,495-507. SHEPHERDT. J., RANKINA. H., and ALDERTOND. H. M. ( 1985 ) A Practical Guide to Fluid Inclusion Studies. Blackie. SPARKSC. J. ( 1976) Quantitative X-ray flourescent analysis using

fundamental parameters. In Advances in X-ray analysis (ed. R. W. GOULD et al.), Vol. 19, pp. 19-52. Kendall Hunt. SPARKSC. J. ( 1980) X-Ray fluorescence microprobe for chemical analysis. In Synchrotron Radiation Research (ed. H. WINICKand S. D~NIACH), pp. 459-5 12. Plenum. THOMPSONM. and WALSHJ. N. ( 1983) A Handbook oflnductively Coupled Plasma Spectrometry. Blackie. WHITNEYJ. A., HEMLEYJ. J., and SIMONF. 0. ( 1985) The concentration of iron in chloride solutions equilibrated with synthetic granitic compositions: The sulfur-free system. Econ. Geol. SO, 444460. WILLIAMSK. L. ( 1987) Introduction to X-Ray Spectrometry. Allen and Unwin.

APPENDIX 1: METHOD FOR ESTIMATING ELEMENTAL MASS RATIOS FROM XRF MICROPROBE DATA Rather than attempting to allow for the complicated shape and internal structure of the inclusion, it has been assumed that the material within the inclusion is homogeneous and present as a constant thickness layer below a layer of topaz. The geometry is illustrated in Fig. Al. The equation for fluorescence intensity given by SPARKS( 1976) is I = DI&u,

[I - exp {(p, + p, sin a/sin /3)psT/sin a}] (/.&i+ 1, sin a/sin 0)

where D is the instrumental factor, including the detector geometry and detector efficiency; C, is the concentration of element z in mass of z per unit mass of sample; I, is the fluorescence intensity in counts per second from element z; pi is the mass absorption coefficient of the sample in cm2/g for the exciting radiation in the inclusion; ibis the mass absortion coefficient of the sample in cm2/g for the fluorescence radiation in the inclusion; I0 is the flux in counts per second of the exciting radiation at the top surface of the sample layer; 6, is the fluorescence cross-section in cm*/g for the exciting radiation; or is the density of the sample in gm/cm3; T is the thickness of the sample fluorescing layer in cm. Absorption in the surface topaz layer is significant. It can be included in the analysis by using the relation 1: = I, exp I -Ftp,dlsin

P 1,

where I: is the fluorescence intensity at the exit from the topaz layer; I, is the total excited fluorescence; F, is the mass absorption coefficient of the topaz for the fluorescence radiation; pI is the density of the topaz; d is the thickness of the topaz layer. Values for fluorescence cross sections are taken from KRAUSEet al. ( 1978) and are plotted for an exciting energy of 15 KeV in Fig. A2. Values for the mass absortion coefficients are calculated by summing the contribution from each element in the multielement layers using the elemental values given by WILLIAMS( 1987). Figure A3 shows the fraction of the fluorescent radiation transmitted through topaz as a function of the thickness of topaz.

fluorescent x-ray

p^

FIG. A 1.Geometric representation of the shape and internal structure of the inclusion.

Chemical compositions of fluid inclusions in topaz

19

I””

60 60

40

20

0 Cl

Ar

K

Cd SC

l7

V

Cr Mn

Fe

Co Ni

Cu

Zn

Ga

Ge

As Se

Br

Kr

Element FIG. A2. Fluorescence cross sections for elements detectable on the Daresbury microprobe (K-line).

0

20

SO

100

40

1;o

Thickness of topaz layer (microns)

FIG. A3. Absorption of fluorescence radiation from inclusion metals by topaz matrix (using measured topaz composition).

For the purposes of calculations the average density and salinity of the inclusions are taken as 1.25 gm/cm’ and 60 wt% total dissolved salts, respectively (Table 1) , and the average composition, based on

the present laser-ICP and analysis (see text), in wt% is Cl, 3 1.56; K, 5.76; Fe, 9.96; Zn, 0.96; Na, 4.32; Mg, 2.28; Mn, 4.56; “M,” 0.60; 0, 35.55; H, 4.44. “M” = sum of Cu, Pb, and Sn.