Gold deposition from heterogeneous aqueous, CO2-rich fluids: resolving the enigmatic and misleading hypothesis of deposition from CO2 only fluids.Kingsley Burlinson, December 2017
Burlinson Geochemical Services P/L, Darwin, Australia
Gold is typically transported as aqueous species, often in CO2
rich fluids, but some workers have found fluid inclusions in
quartz which contain only CO2 with no visible aqueous
phase. This has caused concern in how to explain the trapping of
such fluids and the associated deposits. One proposed hypothesis
is for gold transport as a CO2 complex in a super
critical CO2-only fluid, despite the absence of any
chemical justification. An even more serious problem is the lack
of any explanation for transport and deposition of the silica
which actually comprises the majority of the mineral deposit, gold
being merely a trace impurity in the quartz veins observed. The
actual explanation is that a parent CO2-rich fluid
unmixed into 2 immiscible fluids (2 phases), one dominantly liquid
water and the other predominantly gaseous CO2. Such a
heterogeneous fluid mixture can transport and deposit gold and
silica from the dominant aqueous phase, while appearing to be only
a gaseous CO2 fluid system. The hypothesis of gold
transport in CO2 fluids is an error based on the
mis-identification of heterogeneous fluid inclusion assemblages
and both the gold and silica are in fact transported only in the
aqueous phase of the fluid mixture.
The underlying assumption of fluid inclusion studies is that the
parent fluid was a homogeneous single phase fluid when trapped.
Upon cooling this fluid separates into multiple phases which can
be studied to determine the original trapping conditions. But the
original fluid composition cannot be determined if the
mineralizing fluid underwent phase separation, giving a
heterogeneous fluid mixture at the time the fluid inclusions were
trapped and it can be very difficult to recognize such
heterogeneous fluids during microthermometric studies. In the
P-T-XCO2 phase relationship of CO2
and water there is a large region of immiscibility and such fluids
unmix into binary mixtures. Incorrect conclusions can be made in
the study of these CO2 rich fluids because of failure
to fully understand the nature of these common, but difficult to
recognize, heterogeneous fluids. Heterogeneous trapping does not
necessarily trap both phases in proportion to their abundance and
can produce CO2 only fluid inclusions while
simultaneously failing to trap the coexisting aqueous phase.
Together with recent thermodynamic studies on the CO2-water
system and an understanding of fluid inclusion formation and
morphology it is now clear that the transport and deposition of Au
and silica occurs within an aqueous, CO2 rich,
heterogeneous, unmixed fluid and that both gold and silica are
transported by the aqueous phase. There is no need to invoke
unknown chemistry to transport Au and SiO2 in pure CO2
fluids. Nor are convoluted phase separation mechanisms or post
entrapment modification explanations required to produce CO2
only fluid inclusion assemblages.
Carbonic only fluid inclusion assemblages (FIAs) have been found at many gold deposits and often there seem to be no closely associated aqueous inclusions. (Schmidt Mumm et. al., 1997; Yao et. al., 2001; Willie and Klemd, 2004; Chi et. al., 2006; Klein and Fuzikawa, 2010). To explain such FIAs some authors propose post entrapment modification with diffusion or leakage of water out of inclusions through the host quartz. Other authors propose unmixing of a parent aqueous-CO2 fluid followed by separation of the CO2 and aqueous phases with trapping of CO2 fluid inclusions from the CO2-only fluid. Schmidt Mumm et. al. (1997) proposed that at gold deposits in the Ashanti belt, Ghana, water was never present in the mineralizing fluid and that Au, and of necessity also SiO2, were transported in a super critical pure CO2 phase, invoking unexplained chemical species. This has led some subsequent research (Chi et. al. 2006, Klein and Fuzikawa 2010, Goldfarb et. al. 2017) to cite this paper, which appears to give credence to this dubious hypothesis. But none of these proposals are really very satisfactory. The post-entrapment removal of water has been questioned because it does not explain the consistency of inclusions on a deposit scale. The separation of a pure CO2 fluid prior to trapping introduces an even more serious problem as it fails to explain how the essential host silica is deposited, while the transport of Au in carbonic-only inclusions requires unknown chemistry to transport the Au and also fails to explain the transport and deposition of the essential silica from such a fluid. In all these cases the authors have discounted the idea of heterogeneous trapping from two coexisting phases of an unmixed CO2-rich fluid because they could not find co-existing fluid inclusions of both phases. They assumed that both phases must be trapped in proportion to their abundance and failed to understand the disproportional trapping which is well known in heterogeneous fluids (Roedder, 1984b; Burlinson, 2013). This disproportionation can be so extreme that only one of the two phases present in the fluid mixture is recognizable in the resulting FIA. The inclusions of one phase may be absent or simply be so small that fluids within them cannot be observed. Such small inclusions are often classified as secondary and are ignored, but they may well be small primary inclusions and should be called indeterminate as they may actually be the second phase of a heterogeneous fluid binary pair. There is no reason why fluid inclusions of both phases must be the same size!
The fluid inclusion observations are better explained as the result of unmixing of a parent CO2 rich aqueous fluid into a binary pair of immiscible fluids, an almost pure CO2 fluid and an aqueous fluid with CO2, whose compositions can be deduced from the P-T-XCO2 phase diagram. Inclusions in such a heterogeneous fluid are not trapped in proportion to their abundance in the mixture and there is no need to see inclusions of both fluids. Absence of one of the phases, e.g. the aqueous phase, does not prove the absence of that phase, merely the absence of trapping or the inability to observe it. Variations in the trapping of the two phases can be explained by remembering that this is actually a silica deposit and understanding the mechanisms by which inclusions are initiated, filled with fluid and then sealed into the quartz. The Au and SiO2 are transported in the aqueous phase of the heterogeneous fluid mixture and there is no need for convoluted explanations of phase separation or post entrapment water leakage and no problem with silica transport, and definitely no need to propose unsubstantiated chemical species to transport either silica or Au as a CO2 complex.
CO2 -water immiscibility
In the phase diagram for CO2-water (Fig. 1) there is
an extensive region of immiscibility within which the fluid will
unmix into two distinct and separate phases, one almost pure CO2
and the other an aqueous dominated fluid with lesser CO2.
Depending on the deposit formation pressures, this immiscibilty
can occur at up to 300°C (in the absence of NaCl) and occurs
in almost all CO2 containing fluids below 275°C.
The presence of NaCl in the fluid dramatically increases the
immiscibility, increasing the temperatures at which CO2
immiscibility can occur.The studies which have found pure CO2
inclusions all report trapping temperatures mostly within this
immiscibility field and it is almost certain that they are
observing the CO2 component of a heterogeneous binary
fluid mixture. In particular, Schmidt Mumm et. al. (1997) also
found aqueous-CO2 inclusions upon which to make
temperature and pressure measurements for deposit formation. They
estimated various pressures from <1.7 kbar to 5.4 kbar, mostly
near 2.5 kbar and temperatures from 100°C to 325°C, mostly
below 275°C. In the phase diagram for CO2-water
(Fig. 1) it is seen that most of these temperatures and pressures
fall within the field of CO2-water immiscibility and
the authors should have been alert to the near certainty that
heterogeneous trapping had occurred. However, they stated that
selective trapping of gaseous inclusions after phase separation
(i.e. a heterogeneous fluid) does not explain the lack of aqueous
inclusions, and therefore assumed that heterogeneous trapping did
not occur. But this assumption that both phases of a heterogeneous
system should be trapped is incorrect as disproportionate trapping
from such fluids is normal and well known. (Roedder 1984b,
Burlinson 2013). Chi et. al. (2006) studied the Campbell-Red Lake
deposit in Ontario, Canada, and considered heterogeneous trapping
but were so concerned by the absence of aqueous inclusions that
they instead proposed fluid separation of a pure carbonic fluid
before inclusion trapping, ignoring the serious problem that this
fluid could not transport the essential silica. Yao et. al. (2001)
noticed this same absence of aqueous inclusions in deposits in the
Birimian terrane, Ghana. They identified heterogeneous trapping in
various mines which also had aqueous inclusions, but at the Sansu
mine they proposed post entrapment modifications to explain the
lack of aqueous inclusions.
Fig. 1 Phase diagram showing the immiscibility of CO2 and water with temperature. Different curves show pressure dependency.
The thermodynamics of the CO2-water system is quite complex but has been thoroughly described by Diamond (2001). Pichavant et. al. (1982) state that because of the extremely low mutual solubility of CO2 and NaCl, it can be expected that the miscibility gap(s) in the ternary system H2O-CO2-NaC1 will be drastically enlarged with regard to those in the systems H2O-NaC1 and H2O-CO2 . The addition of small amounts of NaCl to the H2O-CO2 system greatly increases the region of fluid immiscibility (Bodnar et. al. 1985). The extensive P-T-XCO2 immiscibility is a major challenge in fluid inclusion studies but is well known and all such CO2 rich fluids, particularly near pure CO2 fluids, should be considered as quite probably unmixed phases of a heterogeneous fluid system.
Transport of the host mineral phase (quartz)
The fluid must be capable of transporting the host mineral
phase, silica in this case, as these are in fact silica deposits.
The gold is merely an interesting impurity. Fluid inclusion
studies are actually observing the host mineral, in most cases
quartz and the inclusions are minute crystal imperfections in the
host quartz, from which we try to infer the fluid conditions which
transported and deposited the quartz. It is then inferred, based
on paragenetic observations, that these fluid conditions also
relate to the gold impurities which are of interest. Roedder
(1984a) discussed the enigma of inclusions of fluid having zero
solubility for the host and points out that the trapped fluid must
have been a dispersed phase within a second continuous phase, that
is, heterogeneous trapping. The host quartz phase exerts primary
control over the trapping of fluid inclusions. Depending on the
hydrothermal fluid conditions, degree of silica supersaturation
and many other factors the quartz may or may not trap inclusions
and also controls the size of inclusions. It is logically
incorrect to ignore the host phase mineral simply because it is
not of economic importance. There is no attempted explanation of,
or known mechanism to transport large quantities of silica in pure
CO2 fluid and the hypothesis that pure CO2
fluids can transport gold and silica is untenable.
Fluid inclusion morphology
The fluid inclusion morphology shown in the published photo by
Schmidt Mumm et. al. (1997) shows an abundance of round
(spherical) fluid inclusions (Fig 2). Such inclusions are formed
by the surface tension of a liquid phase enclosing a gas bubble.
This shape is not normal for fluid inclusions in quartz, where the
inclusion shape is controlled by the crystal structure of quartz.
Fig. 2 Original photograph of fluid inclusions from Schmidt Mumm et. al. 1997, with added arrows pointing to a few of the many spherical inclusions. Such spherically shaped inclusions are abnormal in quartz and are formed by trapping a gas bubble within a host liquid phase, proving the presence of a dominant liquid phase and heterogeneous trapping. (Modified after Schmidt Mumm et. al., 1997)
Fluid inclusion studies usually report that inclusions are of negative quartz crystal shape, or they can be of irregular shape if silica deposition was fast. Any subsequent recrystallization actually enhances the crystalline shape. A scanning electron microscope (SEM) image of an opened fluid inclusion is shown in Fig. 3 and numerous additional SEM photographs of opened fluid inclusions (Burlinson 2014b) show the strong negative crystalline shape control of inclusions.
Fig. 3 SEM image of an opened fluid inclusion in quartz. Note the angular shape of the inclusion, controlled by the crystal structure of the host quartz. (From Lawrence et. al. 2013)
The numerous spherical, CO2-only inclusions seen in
the photo in the original published paper are clear proof that
these inclusions formed as gas bubbles within a dominant liquid
(aqueous) phase from which the silica deposited around the gas
bubble. The importance of inclusion shape was completely
unrecognized in the original published research. But not all gas
filled inclusions are spherical and some studies (Chi et. al.
2006) show sub-rounded or somewhat angular gas filled inclusions.
The process of trapping an inclusion is not instantaneous and it
takes a long time for the enclosing silica host mineral to deposit
around the gas bubble until it is completely sealed. The silica
will preferentially deposit in locations controlled by the crystal
structure, but is blocked by the presence of the gas bubble.
Depending of the silica saturation level of the liquid, diffusion
rates, fluid flow rates and numerous other variables, both the
crystal structure and bubble interference will control deposition
and affect the overall shape of the inclusion. In addition, the
gas bubble may not remain in place depending on fluid turbulence
and bubble coalescence. Bubbles may move about or leave completely
to be later replaced by another bubble. This would allow some
crystal faces to partially develop before the final sealing of the
inclusion occurs, leading to a mixture of angular and rounded
shapes to the inclusion cavity. But the end result is still a pure
CO2 fluid inclusion which was trapped from a dominantly
Competing mechanisms to trap fluid inclusions
Inclusions are formed by two entirely different processes which results in disproportionate trapping of heterogeneous fluids. For homogeneous fluid trapping, fluid inclusions are the result of random crystal defects during the deposition of quartz. These defects grow into crystal cavities which seal over into fluid inclusions containing some of the homogeneous fluid. But fluid inclusions also form when foreign items, such as a heterogeneous gas bubble, force the quartz to deposit around them by interference with normal crystal growth. Conditions which favor the formation of inclusions by one of these mechanisms may simultaneously suppress inclusion formation by the other mechanism, which results in extreme disproportional trapping of the heterogeneous fluid phases. This can be mistaken for deposition from a single phase fluid as there may be almost no evidence of inclusions from the other phase. Quartz veins growing from slowly moving fluids will facilitate the attachment of bubbles and at low silica deposition rates, will also result in very few crystal defects. The resulting FIA will be dominated by the subordinate inhomogeneous gas phase with very few or no inclusions of the dominant aqueous phase from which the quartz is actually depositing. High flow rates and turbulence would have the opposite effect as bubbles are swept away and crystal defects become common. The resulting inclusion assemblage will consist of only the aqueous phase (in crystal defect type inclusions) with few or no inclusions of the inhomogeneous gas phase. To trap a gas bubble within a liquid the bubble must remain in place for some time. Small, fluid inclusion sized bubbles have very low buoyancy and can remain in place long enough to become fluid inclusions. The presence of natural surfactants may also facilitate bubble stabilization on the growing quartz surfaces, enhancing the trapping of gas bubbles. The resulting disproportionate trapping can easily be so extreme that only one of the phases of a heterogeneous system may be observed in the resulting FIA (Burlinson, 2013). Schmidt Mumm et. al. (1997) failed to understand this disproportionate trapping and actually assumed that failure to trap both phases proved that the fluid was homogeneous and therefore water was absent. This is the critical error in their work as absence of one of the fluid phases in a FIA (in this case, the aqueous inclusions) cannot be used to prove the absence of a second phase in the parent fluid. This lack of aqueous fluid trapping was the only evidence that Schmidt Mumm et. al. (1997) offered against heterogeneous trapping and is incorrect. Other authors have also wrongly interpreted the absence of one of the unmixed fluid phases in FIAs as evidence against heterogeneous trapping.
Many workers become entangled trying to explain the formation of
CO2 only inclusions and argue in favor of physical
phase separation of CO2 from the H2O (Willie and Klemd
2004, Chi et. al. 2006, Klein and Fuzikawa 2010). But this is
self-defeating as it is not possible to deposit silica from the
resulting pure CO2 fluid and hence no fluid inclusions
can be formed at all, much less pure CO2 inclusions.
Deposition of the mineral host phase (silica) is the primary
requirement of fluid inclusion formation, only then can concerns
about the inclusion contents be addressed. The pure CO2
inclusions must be formed within a silica transporting phase which
is most probably liquid water. This is easily accomplished by the
trapping of heterogeneous CO2 bubbles. The real issue
is to understand why aqueous inclusions seem to be absent, which
is merely disproportional trapping and is a function of numerous
variables in the hydrothermal fluid system. And aqueous inclusions
may not actually be absent, simply overlooked because they are too
small to be properly examined.
Identification of fluid immiscibility
Ramboz et. al. (1982) outlined three general criteria for the
identification of fluid immiscibility. These are: (1)
contemporaneous trapping of two different fluid types reflected in
fluid inclusion characteristics, (2) contrasting total
homogenization into both the liquid and vapor phases over the same
range of temperature and (3) the same entrapment pressure at the
homogenization temperature of both fluid inclusion types (i.e. if
one inclusion type decrepitates before homogenizing, the other
type must behave similarly).These criteria may apply to the
boiling water heterogeneous system, but they ignore the problem of
disproportional trapping which is common in the CO2-water
system and which may trap only one of the fluid phases. Using
these criteria would result in highly disproportional
heterogeneous systems being incorrectly classified as homogeneous.
In addition, the decrepitation criteria of item 3 does not apply
to CO2 rich fluids, which are known to have anomalously
low decrepitation temperatures caused by the CO2 and
explained by Burlinson (2005). Microscope observation procedures
used to identify boiling water immiscible systems are not directly
applicable to water-CO2 immiscible systems.
Boiling and immiscibility heterogeneous systems are different
The expected inclusion assemblages of heterogeneous fluid systems are frequently based upon the observations of heterogeneous boiling water systems. But Burlinson (2014a) has explained that heterogeneous trapping of water and steam in a boiling hydrothermal system behaves very differently to heterogeneous trapping of the immiscible phases in the CO2-water system. Bodnar et. al. (1985) provide a comprehensive review of fluid inclusion techniques for studying epithermal boiling fluid systems. They explain that boiling systems should trap inclusions with variable liquid/vapor ratios, representing the end-member liquid and vapor phases as well as inclusions with random mixtures of these phases. Plots of the liquid-vapor phase ratios, or the observed homogenization temperatures should show a bimodal distribution. But that discussion applies specifically to single-component heterogeneous systems (boiling aqueous systems). It is also difficult to apply as often the vapor phase inclusions are absent or their homogenization temperatures cannot be measured. In boiling aqueous fluid systems, silica deposition and fluid inclusion trapping and sealing are very rapid, so inclusions represent the complete range of phase mixtures as they are sealed before phases can migrate and agglomerate. And both interference type-inclusions and crystal defect-type inclusions are abundant in this rapidly deposited quartz. As an aqueous hydrothermal fluid boils, changing much of the liquid to vapor, the volume of the liquid is drastically reduced and all the solutes in the parent fluid, including silica, are concentrated into this reduced liquid volume. The silica concentration exceeds saturation, but cannot become supersaturated because of the great turbulence in the boiling fluid and so silica is deposited rapidly at a rate to match the rate of boiling. Under these conditions of rapid deposition, the quartz is full of crystal defects, which become fluid inclusions when they are sealed over by continuing quartz growth and interference-type inclusions are also trapped. The defect-type inclusions trap the parent fluid liquid phase, as well as mixtures of the liquid and gas phases. Gas filled inclusions are also trapped by interference with quartz deposition. Overall these boiling systems are dominated by defect-type inclusions which contain both phases from the parent fluid. Consequently the boiling fluid inclusion assemblage data could give bimodal data distributions. But there is no certainty that all boiling systems will show bimodal data distribution plots and the lack of such bimodal plots cannot prove the absence of boiling.
However, immiscible gas fluid systems (multi-component), which did not boil, do not deposit silica so quickly and this can give a very different appearance to the FIAs. This difference in the rate of silica deposition can lead to preferential trapping of one of the two phases present because the liquid rich inclusions occur as defect-type inclusions and the gas rich inclusions occur as interference-type inclusions which are trapped by different and independent mechanisms. Slow silica deposition will give very few crystal defects and therefore no way to trap the dominant liquid phase as inclusions. But gas bubbles will be efficiently trapped by interference with the deposition of quartz, which is forced to deposit around the bubbles, which then become pure gas filled inclusions. This fluid system is prone to extreme disproportional trapping. Consequently it is unlikely that plots of such immiscible gas FIA data would be bimodal, despite this being a true heterogeneous fluid system. It is quite possible that only fluid inclusions of the minor CO2 gas phase are trapped with no inclusions of the dominant aqueous phase, or that only aqueous inclusions are trapped as the gas phase was flushed through before it could be trapped in inclusions.
In the boiling system, the two phases are metastable, with large
thermal transfers of latent heat taking place, causing turbulence
and rapid silica deposition as the liquid volume is reduced. In
the immiscibility system the phases exist together in complete
stability with no thermal transfers. Conclusions or expectations
drawn from the boiling heterogeneous system cannot be directly
applied to the immiscibility heterogeneous system.
Thermodynamic experiments on super critical CO2
Weihua Liu et. al. (2015) conducted experiments to examine the
solubility of gold in super critical CO2. They state
that the results showed that gold solubility has a negative
correlation to the CO2 content in the fluids and that
H2O as a polarized molecule plays a more active role than the
un-polarized CO2 molecule in the fluids, and hydrated
chloride species are the main form for transporting gold in the CO2-H2O-HCl
system. The gold would have been transported in water rather than
CO2. That work has been summarized by Burlinson (2016)
and Mei et. al. (2017) have published a detailed paper of this
Summary & Conclusions
Quartz vein hosted gold deposits, including those in West Africa, are formed from aqueous hydrothermal fluids which had relatively high CO2 contents, but no more than mesothermal gold deposits in other gold provinces. The suggestion of Au transport in pure CO2 hydrothermal fluids made by Schmidt Mumm et. al. (1997) was based on a failure to understand heterogeneous trapping at temperatures and pressures within the immiscibility field of water-CO2 mixtures. None of their observations prove the absence of water and the spherical morphology of the inclusions in the photograph in their published paper actually proves that a dominant liquid phase, almost certainly water, was present. Hence the entire concept of transport of gold and silica in pure CO2 fluids has no factual basis. In addition, the many convoluted explanations in the literature to explain pure CO2 inclusions are unnecessary and actually compound the problem by failing to understand the requirement to deposit silica (or another host mineral phase). Such pure CO2 fluid inclusions are merely heterogeneous gas bubbles trapped within a dominant aqueous phase which deposited the host mineral. The absence of trapping of the aqueous phase is a result of disproportional trapping due to the complexities of initiation and growth of fluid inclusions in the host mineral phase. Identification of FIAs trapped from immiscible heterogeneous fluids is frequently erroneous because of the failure to understand inclusion deposition mechanisms, the necessity of transporting the host mineral, disproportional trapping or that heterogeneous fluids due to boiling are very different to unmixed immiscible fluids. Almost all CO2 containing fluids will be affected to some degree by the extensive immiscibility of CO2 and water, particularly if even small amounts of NaCl are present, which must be carefully considered when interpreting these fluid systems.
Bodnar, R.J., Reynolds, T.J. and Kuehn, C.A., 1985. Fluid
inclusions systematics in epithermal systems. In Geology and
geochemistry of epithermal systems, Berger, B.R. and Bethke, P.M.
eds. Reviews in Economic Geology, v 2. The Economic Geology
Burlinson, K. 2005. Why CO2-rich inclusions
decrepitate at low temperatures http://appliedminex.com/decrep/general/pt.htm
Burlinson, K., 2013. Disproportionate inclusion trapping from
heterogeneous fluids. http://www.appliedminex.com/viewpoint/disprop.htm
Burlinson, K., 2014a. Contrasting types of heterogeneous
hydrothermal fluids. http://www.appliedminex.com/viewpoint/hetfitype.htm
Burlinson, K., 2014b. Spherical fluid inclusion shape can
indicate heterogeneous trapping. http://www.appliedminex.com/viewpoint/fincmorph.htm
Burlinson, K., 2016. Thermodynamic studies show that Au is NOT
transported in pure CO2 fluids. http://www.appliedminex.com/viewpoint/au-notin-CO2.htm
Chi, G., Dubé, B. and Williamson, K., 2006. Formation of
the Campbell-Red Lake gold deposit by H2O poor, CO2-dominated
fluids. Mineralium Deposita v. 40 p. 726-741.
Diamond, L.W., 2001. Review of the systematics of CO2-H2O
fluid inclusions. Lithos v. 55, p. 69-99.
Goldfarb, Richard J., Anne-Sylvie André-Mayer, Simon M.
Jowitt and Gavin M. Mudd, 2017. West Africa: The World's premier
paleoproterozoic Gold province. Economic Geology, v. 112,
Klein, E.L. and Fuzikawa, K. 2010. Origin of the CO2-only
fluid inclusions in the palaeoproterozoic Carará
vein-quartz gold deposit, Ipitinga auriferous district, SE-Guiana
shield, Brazil: Implications for orogenic gold mineralization. Ore
geology reviews v. 37 p. 31-40.
Lawrence, D.M., Treloar, P.J., Rankin, A.H., Boyce, A. and
Harbridge, P. 2013. A Fluid Inclusion and Stable Isotope Study at
the Loulo Mining District, Mali, West Africa: Implications for
Multifluid Sources in the Generation of Orogenic Gold Deposits.
Economic Geology v. 108 p. 238.
Liu, W., Mei, Y., Brugger, J., Migdisov, A., and Williams-Jones,
A 2015. Speciation and solubility of gold in CO2-HCl-H2O
fluids: MD simulations and solubility experiments. Goldschmidt
2015 abstracts. Also at http://www.appliedminex.com/decrep/general/gold-CO2-liu.htm
Mei, M., Liu, W., Brugger, J., Migdisov, A.A. and Williams-Jones,
A.E. 2017. Hydration is the key for gold transport in CO2-HCl-H2O
vapor. ACS Earth and Space chemistry, v. 1, p. 368-375.
Pichavant, M., Ramboz, C., and Weisbrod, A. 1982. Fluid
immiscibility in natural processes: use and misuse of fluid
inclusion data (I). Chemical Geology, v. 37 p. 1-27.
Ramboz, C., Pichavant, M. and Weisbrod, A. 1982. Fluid
immiscibility in natural processes: use and misuse of fluid
inclusion data (II). Chemical Geology, v. 37 p. 29-58.
Roedder, E., 1984a. Fluid Inclusions. page 35. Reviews in
Mineralogy v 12, Mineralogical Society of America.
Roedder, E., 1984b. Fluid Inclusions. pages 29 & 34. Reviews
in Mineralogy v 12, Mineralogical Society of America.
Schmidt Mumm, A., Oberthür, T., Vetter, U. 1997. High CO2
content of fluid inclusions in gold mineralizations in the Ashanti
Belt, Ghana: a new category of ore forming fluids? Mineralium
Deposita, v. 32, p. 107-118.
Willie, S.E. and Klemd, R. 2004. Fluid inclusion studies of the
Abawso gold prospect, near the Ashanti Belt, Ghana. Mineralium
Deposita v. 39 p. 31-45.
Yao, Y., Murphy, P.J., and Robb, L.J. 2001. Fluid characteristics of granitoid-hosted gold deposits in the Birimian Terrane of Ghana: Fluid inclusion microthermometric and raman spectroscopic study. Economic Geology v. 96 p. 1611-1643.