Applied mineral exploration methods, hydrothermal fluids, baro-acoustic decrepitation, CO2 rich fluids
Newest Topics:
For the latest news, see the NEWEST TOPICS page.

Google is too dumb to let me put the list of news in this column and falsely claims that all my pages are self-duplicates.


Google's so-called "Artificial Intelligence" is an abuse of the concept of intelligence!

Heterogeneous fluid inclusion trapping:

Explaining the formation of CO2-only fluid inclusions

K. Burlinson      April 2018, November 2019


There has been much confusion in the literature over the observation of completely CO2 filled fluid inclusions within quartz, lacking any visible water phase. In many cases the authors have concluded that the inclusions must have formed from  a non-aqueous CO2 fluid. But this is nonsense because  everyone has forgotten that you must first have deposition of silica in order to trap a fluid inclusion and silica is not transported in or deposited from pure CO2 fluids. These CO2 filled inclusions can be formed by heterogeneous trapping from aqueous fluids; there is no need to propose unsubstantiated silica transport and deposition from non-aqueous CO2 fluids!

Because of the extensive immiscibility of CO2 and H2O fluids, low temperature trapping below 300 C and disproportional trapping, the fluid inclusion assemblages (FIAs) may leave no trace of either of the 2 fluid phases, which many authors fail to understand. An aqueous CO2 bearing hydrothermal fluid which is homogeneous at high temperatures will separate into a mixture of 2 immiscible fluids as the temperature reduces. At the trapping temperatures typically observed in hydrothermal gold deposits, even parent fluids with only a small content of CO2 will separate into a heterogeneous mixture of these immiscible fluids, one aqueous and the other carbonic. But depending on the rate of silica deposition and conditions within the fluid only one of the heterogeneous fluids may be trapped, leading to the false assumption of a homogeneous single phase fluid system. It is well known that inclusions trapped from a heterogeneous fluid do not trap the 2 (or more) (possibly immiscible) phases in proportion to their abundance.

By understanding the immiscibility of CO2-H2O fluids it is simple to explain the trapping of CO2-only fluid inclusions from a heterogeneous fluid system in which the aqueous phase transported the essential silica, and probably the elements of economic interest such as gold. The claims that gold and silica are transported in pure CO2 fluids are unsubstantiated and false.


This discussion uses the term "heterogeneous trapping" to refer to inclusion trapping from an aqueous CO2 bearing fluid which has become immiscible at the solvus and has separated into two fluids,  a conjugate-pair of fluids with compositions determined by the solvus curve. This results in two fluids, from which the individual fluid inclusions trapped would be comprised, in theory, of either or both of  the aqueous or the carbonic conjugate fluid.  I do not discuss mere mechanical mixtures of fluids of random composition as it is improbable that mechanical mixtures of immiscible fluids can be trapped within individual fluid inclusions as they would not remain stable enough for the long time period required to be encased in the slowly depositing silica.


  1. CO2 rich aqueous hydrothermal fluids frequently ex-solve into 2 conjugate fluids, one aqueous and one almost entirely CO2 because of immiscibility as the pressure, temperature and salinity varies.
  2. The CO2-rich fluid from which inclusions are trapped is frequently heterogeneous and should not be assumed to be homogeneous despite appearances in thin sections.
  3. The 2 conjugate fluids may not both be trapped because the gaseous and aqueous fluids are trapped as inclusions by completely different, mutually opposing mechanisms.
  4. This selective trapping is often so extreme that only one of the 2 fluids occurs in the observed fluid inclusion assemblages, leading to false assumptions of a homogeneous fluid.
  5. The criteria proposed by Ramboz et al. (1982) to identify heterogeneous trapping in observed assemblages can prove heterogeneous trapping but are too restrictive and do not disprove it.
  6. As CO2 ex-solves from the fluid, quartz solubility increases and fluid inclusion trapping ceases until quartz saturation is re-established at much lower temperatures.
  7. These complications cause serious misinterpretation of the original fluid composition when the interpretation fails to allow for selective and dis-proportional trapping of inclusions from heterogeneous fluids and assumes that heterogeneous fluids can be identified by thin section observations.


The diagram below shows the phase relations of a CO2 - H2O system with Temperature and Pressure. (XCO2 is the mole fraction of CO2) The solvus curves of the miscibility boundary are shown at 4 different pressures. The data is from Tdheide and Franck (1963). (Yes, the isobaric solvus curves do intersect and cross. The CO2-H2O system is complex and non-linear with pressure!) If there is any NaCl present, the immiscibility is greatly increased. Clearly immiscibility is a major issue in such fluids and will affect many gold bearing hydrothermal fluid systems, as they frequently contain CO2.

P T X-CO2 phase diagram solvus
        immiscible fluids

At a pressure of 1 Kbar, the fluid in the region shown in green is a homogeneous single phase. If we consider an initial fluid with just 0.1 mole fraction of CO2 at 350C, as it cools it will follow the path of the red line down until it intersects the solvus at about 260C. Here it will separate into two immiscible phases, one dominantly aqueous (liquid, as this is below the critical point temperature of water and at high pressure) and the other  a water saturated CO2 phase (a super-critical fluid because it is above the critical point temperature of CO2 and at high pressure). But inclusion trapping does not occur until silica saturation is reached, assumed in this example to be at 200C and shown as 2 red crosses in this diagram. Note that the solubility of silica increases as CO2 is ex-solved from the solution, which prevents the deposition of silica and formation of fluid inclusions until the temperature is considerably reduced (discussion below). From 260C down to 200C, both fluid phases evolve and coexist as a mixture in an open fluid system. The inclusions trapped at 200C will be of either or both of the 2 fluid compositions shown by the 2 red stars on the diagram. Intermediate compositions cannot occur because the 2 fluids are immiscible. (Trapping mixtures of these two immiscible fluids within the same tiny fluid inclusion is highly improbable.) Because the super-critical CO2-rich carbonic phase (which is gaseous and buoyant) is often lost and not trapped in inclusions, the original CO2 content of the fluid cannot be determined from the remaining aqueous inclusions trapped at 200C. In this example, after escape of the gas phase fluid, only the aqueous fluid will be trapped in fluid inclusions and it will seem that the parent fluid was homogeneous and had only a low CO2 content of about 0.05 mole fraction CO2. However this is incorrect because the CO2 rich fluid has been lost before trapping and is not present in any fluid inclusions.

Rarely, if the carbonic fluid bubbles are small and not flushed out of the system by turbulence in the system, the CO2 rich fluid may be trapped. These conditions may allow the carbonic bubbles to adhere to growing crystal surfaces which interferes with silica deposition and forms carbonic fluid inclusions as silica is forced to deposit around these bubbles. The bubbles would have to remain in place for a long time to be encased in silica. This would give CO2-only inclusions, despite actually forming from a heterogeneous, dominantly aqueous, fluid mixture. Such fluid inclusions would have CO2 contents of 90% and may well appear to be totally carbonic at room temperature because the volume of the condensed liquid water phase would be extremely small and exist merely as an invisible film on the host mineral walls of the inclusion. (A calculation of phase volumes confirming this is here.) These same depositional conditions may give rise to only few and/or only small aqueous fluid inclusions if silica growth is slow and forms well ordered quartz, lacking crystal defects. Such crystal defects are the precursors of the cavities which become fluid inclusions. Small fluid inclusions typically cannot be observed well enough to study them and are overlooked or assumed to be unimportant secondary inclusions. The failure to trap or to recognize these aqueous inclusions leads to the false assumption that only CO2 fluids were present. These competing mechanisms of fluid inclusion trapping act in mutual opposition and it cannot be assumed that both fluid phases should be present in FIAs, and the absence of either type of fluid inclusion in the sample does not prove the absence of that type of parent fluid. The misrepresentation of the parent fluid composition is merely a function of selective trapping of the heterogeneous fluid phases by mutually opposing trapping mechanisms.

The carbonic fluid inclusions may also suffer post-entrapment water removal. The carbonic inclusions typically have a very high internal pressure. The 0.9 X-CO2  inclusions trapped at 1 Kbar and 200 C  in the above example have a molar volume of about 50 cc/Mole (density  0.83 g/cc) and still have an internal pressure of more than 300 bar at 100 C and 1 bar room conditions. This pressure differential could facilitate preferential diffusion or migration of  H2O  out of the inclusion because the size of H2O molecules is much less than of CO2 molecules. This would result in more CO2  enriched, "water free" inclusions.

The apparent absence of aqueous inclusions in the sample and the absence of any visible water within carbonic fluid inclusions could be misinterpreted as deposition from a non-aqueous CO2 fluid. However as seen in this example, the parent fluid was in fact dominantly aqueous with a composition of 0.95 mole fraction H2O and 0.05 mole fraction CO2 at the trapping temperature. It would be quite wrong to infer formation from CO2-only fluids, despite the fact that only carbonic inclusions were observed in the sample. An aqueous phase must have been present to transport and deposit the quartz host mineral as silica  is not soluble in pure CO2 fluids, not even if they are super-critical. (Liu, W. et al. 2015)

If NaCl (or other salts) are present in the fluid, the CO2  immiscibility is greatly increased and this plot shows the significant effect of salt on the solvus, greatly increasing the immiscibility region.
P-T-X Co2 phase solvus diagram at
          2 Kbar for saline solutions

The effect of CO2 on silica solubility and inclusion formation

The solubility of silica in aqueous solution is inversely dependent on the concentration of CO2 in the fluid. Consequently as the fluid cools down  to the solvus temperature and partitions into two components, ex-solving a  CO2 rich phase, silica deposition will be interrupted or will not commence until considerably more cooling has occurred to compensate for the increased silica solubility in the now CO2 depleted aqueous phase. Many CO2 rich fluids will intersect the solvus and undergo immiscibility separation during cooling, becoming heterogeneous. There may be a complex fluid inclusion assemblage due to this hiatus of silica deposition, or just heterogeneous deposition from the aqueous phase after phase separation. The gaseous CO2 phase is likely to be lost due to its buoyancy and not trapped in fluid inclusions, so evidence of this phase separation is unlikely to be preserved in fluid inclusion assemblages.

This diagram shows the relationship between CO2 and quartz solubility with temperature in a 5 wt % NaCl solution. The data is replotted from Thomas Monecke, Jochen Monecke, and T. James Reynolds,  The co-influence of CO2 on the solubility of quartz in single-phase hydrothermal fluids: implications for the formation of stock-work veins in porphyry copper deposits.  Economic Geology, 2019, v. 114, no. 6, pp. 11951206.

coinfluence of CO2 on
          quartz solubility

This is calculated data  based on the density model of Akinfiev and Diamond (2009),  using the software package Loner AP of R.J. Bakker (, which is based on the thermodynamic equation of state of Duan et al. (1995).

The confusion over heterogeneous versus homogeneous trapping

By default, everyone assumes that inclusions are trapped from a homogeneous fluid, unless it is proven otherwise. The criteria outlined by Ramboz et al. (1982) is used to conclusively prove heterogeneous trapping. But failure to satisfy these criteria does NOT prove that the fluid must have been homogeneous.

 The main problem is that the criteria assume that inclusions trapped from heterogeneous systems will include both of the conjugate pair fluids. This requirement ignores the occurrence of disproportional trapping, which is normal in such fluids. Roedder has explained the importance of this issue which must not be ignored. The failure to see both conjugate fluids in the trapped inclusion assemblages is merely a function of selective trapping due to the 2 different and mutually opposing mechanisms which are involved in trapping either the aqueous or carbonic fluids (bubbles). This is explained here. Heterogeneous trapping is widespread in CO2-rich systems but rarely recognized!

The Ramboz criteria

The Ramboz et al. (1982) criteria are intended to be used to unambiguously identify heterogeneous trapping based solely on observations in thin section.

  1.  The two types of inclusions must occur in the same regions of the same sample, and there must be good evidence of their contemporaneous trapping: This assumes that both of the conjugate pair fluids are trapped, which is rarely the case. Selective trapping, which is common, may result in only one fluid being observed and this criteria cannot disprove heterogeneous trapping.
  2. The two types of inclusions must homogenize at the same temperature, or more realistically within the same range of temperature (because trapping is not an instantaneous and strictly isothermal-isobaric process):  Again, this assumes that both of the conjugate pair fluids are present. The absence of one of the conjugate fluids and inability to therefore measure its homogenization temperature may merely be selective trapping and cannot disprove heterogeneous trapping.
  3. Upon heating the pressure difference is generally not sufficient to allow the two types to decrepitate at very different temperatures (unless their size and shape are very different), so the pressures must reach the same value (trapping pressure) at homogenization temperature. Therefore, if one inclusion type decrepitates before homogenizing, the other type must behave similarly:  Again, this assumes that both of the conjugate pair fluids are present. Furthermore it is known that CO2 rich fluid inclusions decrepitate at anomalously low temperatures because the pressure in a gas-rich inclusion increases according to the gas law equation, PV=nRT. This will cause decrepitation long before the conjugate aqueous inclusion decrepitates, as I have explained here. This criteria is incorrect for CO2-rich inclusions even if both conjugate fluids are present.
These criteria may be useful in proving the existence of heterogeneous trapping, but its failure does not prove the converse, i.e. that the fluid was homogeneous.

The following P-T plot shows solvus curves for 3 levels of CO2 and shows the extensive immiscibility field, particularly so if even low levels of salt are present. The red solvus curves are for pure H2O with no salt, the purple curves show the solvus of saline fluids. Only limited data points are available to define the saline solvus curves. The solvus curves are derived from the data and 3D plot of P-T-X-CO2 shown here and from the salinity plot above.

Homogenization to a liquid phase inclusion happens close to the water liquidus, such as at the points shown by a black star. The actual trapping temperature of inclusions depends on the pressure and is along the isochore through the homogenization point, shown as pale blue lines. It is often assumed that trapping occurs from a homogeneous fluid, but in many cases trapping occurs below the solvus curve and is actually the aqueous phase of a separated immiscible fluid. Trapping in the dark shaded area is unquestionably heterogeneous. For fluids containing salt, trapping in the light shaded area may also be heterogeneous. The fluid may be  heterogeneous at pressures up to more than 1 Kbar, which is about 4 Km depth, lithostatic pressure. Many hydrothermal deposits form within this depth range and fluid salinity and the observed inclusions must be the aqueous fraction derived from a heterogeneous parent fluid. Even fluids which homogenize at temperatures as high as 300 C could be derived from a heterogeneous parent fluid.

PT-with co2 solvus

Many fluid inclusions from hydrothermal deposits have homogenization temperatures below 300 C, formed at depths less than 4 Km  and contained enough salt that they must have unmixed and be the result of heterogeneous trapping. The absence of inclusions of both of the conjugate fluids is merely the result of selective trapping because the buoyant gas phase bubble is swept away before it can be silicified in place. Such silicification is inhibited by the increase in quartz solubility as CO2 exsolves further suppressing  fluid inclusion trapping. The aqueous phase is trapped by a completely different mechanism to the gas phase which strongly favors selective trapping of the aqueous phase inclusion without trapping the conjugate fluid gas bubble.

Observations of the homogenization temperature and salinity are enough to indicate the occurrence of heterogeneous trapping, despite the absence of the related conjugate fluid. It is wrong to assume homogeneous trapping merely because no matching conjugate fluids are observed in thin section. CO2 immiscibility exists to pressures of more than 1 Kbar at depths exceeding 4 Km and occurs in many hydrothermal fluid systems.


The extensive immiscibility between water and CO2 is often ignored during fluid inclusion studies merely because it is difficult to identify the presence of these heterogeneous systems based solely on observations of fluid inclusion assemblages in thin sections of the samples. It is also wrong to assume that the CO2-H2O heterogeneous system is similar to the boiling water heterogeneous system. (The immiscible H2O and CO2 phases behave very differently to the miscible heterogeneous liquid-water / water-vapour system documented in the literature.) It is often assumed that inclusions trapped from heterogeneous fluid systems must trap both component fluids giving bimodal FIAs, but this is incorrect (long since explained by E. Roedder, 1984) because the carbonic fluid (a gas) is usually lost and not trapped in inclusions. It is also assumed there should be a continuum of compositions between the 2 end member aqueous and carbonic fluids. However that would require trapping of mixtures of the immiscible fluids within individual fluid inclusions, which is improbable. The often assumed (Ramboz) criteria for identification of heterogeneous trapping are inappropriate because they assume both conjugate fluids are trapped, which is not the case. The Ramboz criteria cannot disprove heterogeneous trapping and therefore cannot prove homogenous trapping. It is also essential to consider the transport and deposition of silica in order to seal up fluid inclusions, something which is ignored by almost all authors. Silica is not transported in non-polar CO2 fluid, not even if it is super-critical. And few authors consider the multiple mechanisms of fluid inclusion trapping which may act in mutual opposition to give extreme disproportional trapping from heterogeneous fluids.  The inverse relation between quartz solubility and CO2 content interrupts quartz deposition as CO2 exsolves, complicating the understanding of FIAs.

Many (perhaps most) CO2 rich hydrothermal fluids will undergo immiscible phase separation during deposition of quartz and formation of fluid inclusions and are hence the result of heterogeneous deposition. It is incorrect to assume such fluids are homogenous unless proved otherwise, and we should actually assume the inverse; that CO2 rich fluid inclusions are formed from heterogeneous fluids unless proven to be solely formed from a homogeneous fluid. The criteria proposed by Ramboz et al. (1982)   (and also discussed above) to prove heterogeneous trapping do not account for the occurrence of disproportional trapping which is often extreme in CO2 rich fluid systems.

 For CO2 rich fluids, heterogeneous inclusion trapping is the more common normality and the frequent unproven assumptions of deposition from homogeneous fluids are incorrect.

CO2-only fluid inclusions are rare, but may be trapped from heterogeneous aqueous, CO2-rich fluids which have undergone phase separation into a conjugate pair of fluids due to immiscibility. These fluids may be incorrectly recognized because of disproportional trapping of the separate phases. Such heterogeneous fluids are very common (although difficult to recognize in petrographic thin sections of samples) because there is an extensive Pressure - Temperature - Composition (P-T-X) region of immiscibility in the H2O-CO2-NaCl system which includes most of the P-T-XCO2 region in which hydrothermal gold deposits form. (Gold deposits are frequently associated with CO2 rich fluids.) Depending on the fluid conditions such as flow rate and turbulence and the rate of silica deposition, only the CO2 fluid may be trapped as fluid inclusions due to extremely selective trapping. It is counter-intuitive  to propose the formation of CO2-only inclusions from CO2-only fluid as such postulates fail to explain the essential deposition of the host mineral phase, silica.

A proposed method to help identify heterogeneous trapping of immiscible fluids is to observe the shape of the carbonic inclusions. Rounded inclusions are indicative of trapping as an immiscible gas bubble within a liquid host fluid, proving heterogeneous trapping, as discussed here.


Claire Ramboz, Michel Pichavant and Alain Weisbrod,  Fluid immiscibility in natural processes: Use and misuse of fluid inclusion data:  II Interpretation of fluid inclusion data in terms of immiscibility.
        Chemical Geology, 37 (1982) 29---48

Index to discussions of heterogeneous H2O-CO2 systems.