Home & Garden

A natural example of superdense CO2 inclusions: Microthermometry and Raman analysis

A natural example of superdense CO2 inclusions: Microthermometry and Raman analysis
of 7
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  Cwchimica a Cosmochimica Ada Vol. 54, pp. 895-901 Copyright 0 1990 Pergamon Press plc. Printed in U.S.A. 0016-7037/90/ 3.00 + .oO LETTER A natural example of superdense CO2 inclusions: Microthermometry and Raman analysis A.M. VANDENKERKHOF andS.N. OLSEN* ‘Institut voor Aardwetenschappen, Vrije Universiteit, 108 1 HV Amsterdam, Netherlands *Department of Earth and Planetary Sciences, The Johns Hopkins University, Baltimore, MD 2 12 18, USA zyxwvutsrqponmlkji (Received November 13, 1989; accepted in revised form February 14, 1990) zyxwvutsrqponmlkjihgfedcbaZ Abstract The presence of extremely dense CO2 fluid inclusions in migmatites from the Colorado Front Range has been confirmed by Raman analyses. These inclusions show metastable homogenization of liquid and vapor (L + V --t L) at temperatures below the triple point of pure CO*. Raman analyses failed to show the presence of N2, CH4, H , or any other components found in the past in COz-rich inclusions with such low homogenization temperatures. We conclude that the high density resulted from a decrease in inclusion-cavity volume during isobaric cooling of less dense inclusions. INTRODUCTION REPORTSOFCO~INCLUSIONS which homogenize zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA elow he triple point are found in several published results of micro- thermometric studies of fluid inclusions by a heating-freezing stage (e.g., COOLEN, 1980; SWANENBERG, 1980). Such ho- mogenization (L + V -) L) is a metastable phase transition in the case of pure CO2 and indicates an extremely high den- sity for the fluid: the internal pressure of such an inclusion at room temperature is on the order of 2 kb. The apparent high densities have led to the conclusion that these fluids have been trapped at extreme depths, corresponding generally to pressures greater than 10 kb. Recent Raman microprobe analyses of some of these previously reported highdensity CO2 inclusions have shown that the inclusions do not contain pure CO2 but mixtures of CO*-N2 or CO&H4 with densities lower than that of CO2 at the triple point (1.178 g/cm’). This is the case for inclusions from the Furua Granulite Complex in Tanzania (KERKHOF, 1988b) and from the Rogaland area in southwestern Norway (KERKHOF et al., 1990). The phase behaviors of these fluid mixtures are very similar to those of high-density CO2 (TOURET and KERKHOF, 1986), and mi- crothermometry alone does not provide definitive evidence for high-density CO2 fluid. To date, all inclusions with ho- mogenization temperature (Th) below the triple point for pure CO*, when investigated by the combined techniques of microthermometry and Raman analysis, have been found to contain fluid mixtures with lower densities than srcinally reported. Herein, we present the first case of truly dense CO2 inclusions with metastable homogenization temperatures lower than the triple point. STUDY SAMPLES CO2 inclusions in this study are contained in matrix quartz in migmatites from the Colorado Front Range, USA. The migmatites are Proterozoic in age (Rb/Sr whole-rock age = - 1700 Ma) and have been metamorphosed to amphibolite grade (see OLSEN, 1987, 1988, for more details). Leucosomes (leucocratic lenses and layers) in the migmatites consist 895 mainly of quartz, microcline, and plagioclase. Paleosomes (host rocks) are biotite-quartz-feldspar gneisses. METHODS AND RESULTS Microthermometry The most commonly used technique for determination of densities in non-aqueous inclusions is a microthermometric study of the phase behavior at temperatures between - - 180 and +35”C. Most measurements in this study were made with a USGS-type heating-freezing stage modified by Fluid Inc. A Chaixmeca stage was also used in the range -100 to -40°C because it gave a slightly better optical resolution, especially for the phase transition S + L + L. The inclusions in this study are small (5- 15 pm in size), and phase transitions are difficult to observe. The heating rate around the temper- ature of CO2 melting was approximately 0.1 “C/s. The stages were calibrated at the COZ triple point (at -56.6”C) by using known pure CO2 inclusions of low density. Most of CO* inclusions in the Front Range samples are monophase at room temperature (OLSEN, 1987), but a vapor phase nucleates in all inclusions during cooling to about -100°C. Some supercooling is necessary for freezing any fluid. The inclusions in this study froze between about -85 and -lOO”C, or 30 to 40°C below the triple point of CO*; but they melt within a few degrees of the triple point upon subsequent heating (see below). This interval between the freezing and melting temperatures allows us to observe ho- mogenization below the equilibrium triple point: homoge- nization temperatures can be measured after cooling without freezing the inclusion. Homogenization temperatures of approximately 1300 in- clusions have been measured (OLSEN, 1987). Of these, four showed homogenization below the CO2 triple point at -56.6”C. Three were found in quartz in a leucosome, one in a paleosome. A homogenization temperature between the triple point and the temperature at which the inclusion freezes is denoted as metastable homogenization temperature (Thm)  89h 4. M. van den Kerkhofand S. N. Olsen because the assemblage liquid + vapor is metastable in this with liquid), but some ofthe solid remains. This transition range. Thm’s for the inclusions in this study range from -66 is denoted as partial homogenization to liquid (Ths L). to -61°C. In order to observe stable phase transitions, the The solid phase can easily go undetected at this stage inclusions must first be frozen. Upon heating the frozen in- because of the very small difference between the refractive clusions, the following features are observed (Fig. I ): indices of the liquid and solid. 1) 2) 3) (4) A frozen inclusion contains solid CO* and a vapor phase. The filling degree for the solid is approximately 0.8; the solid occupies about 80% of the inclusion volume. A liquid phase forms at about -56.6”C resulting in three phases: solid, liquid, and vapor. This transition is denoted as initial melting by formation of liquid (Ti L). Both solid and vapor phases rapidly shrink during a very narrow temperature interval after the first appearance of melt. At a few tenths of a degree above the initial melting, the vapor bubble dissolves into the liquid (homogenizing -60 r ‘t I -65 lo-3T-7-125-l r~ ~~~ ~~ table ii L ’ 0 -6LB The final phase transition takes place by dissolution of the solid into liquid. This homogenization has been de- noted as “sublimation” to the liquid phase (Ts L), rather than melting, by zyxwvutsrqponmlkjihgfedcbaZYXWVUTS ERKHOF 1988a: 1990) because in a general case the transition can be either S + V + V or S + L + L. The two transitions for C02-CH4-N2 inclu- sions cannot in general be distinguished by a microscopic observation because of the very small difference in re- fractive index between vapor and liquid. The only ex- ception is the case of pure CO2 inclusions for which the distinction is possible. For the inclusions in this study, -7 meta stable 1 4 FIG. I. Stable and metastable phase transitions in “superdense” CO2 inclusions observed on warming from 5 -85 to -40°C. Inclusions show a stable phase sequence, S + V +S+L+V-+S+L+L,whenwarmedafiercooling to - -100°C (temperatures at which they freeze), but met&stable homogenization L + V -) L when warmed after cooling to - -85°C without being frozen. The metastable homogenization is at a lower temperature than that of melting (see text).  Superdense CO2 nclusions in migmatites 897 Table 1. Microthermometric results Sample Thm TiL°C ThsL”C TsL”C 1 Q-3T-7-a0 -66.2 -66.6 -66.2 -62.1 1 O-3T-7-a6-2 -60.7 -66.4 7’ 1 O-3T-7-a6-1 -64.6 -66.7 -62.6 IO-3T-1 O-b7 -61.8 -66.6 -54.1 ‘Ultflcult to observe Thm=metarlable homogenization: L+V + L TIL ~Inltial meltlng: V+S V+S+L ThrL=parHaI homopenlzatlon to liquid: V+L+S L+S TsL ~subllmation(/melting): S+L e L or S+V V all final transitions are S + L --) L so that the transition is referred to as melting/sublimation, or simply as melting, below. The melting/sublimation in these inclusions was observed at -52 to - -54’C (Table 1). These temper- atures are significantly higher than the triple point of COZ Thus, the following phase transitions were observed during the stable path of a microthermometric analysis from - - 100 to -4O’C: S + V + S + L + V + S + L + L. These inclusions are Type S3 in the classification proposed by KERKHOF (1988a; 1990). “S” stands for the final phase tran- sition, “sublimation.” “ 3” stands for the total number of phase transitions. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA P T phase diagrams The observed phase transitions in these inclusions, which we believe contain extremely dense CO2 fluid, can be ex- plained by P-T phase diagrams (e.g., SWANENBERG, 1980; ROEDDER, 1984, p. 376-377). Figure 2 diagrams CO* in the region of the triple point constructed using the data of ANGUS et al. (1976) and shows stable L + V, S + V, S + L, and metastable L + V curves together with some isochores. P-T P bar 600 paths during cooling and subsequent heating are illustrated for an inclusion with the molar volume of 36.5 cm3/mol (density = 1.2 1 gjcm3), which is representative of the densities for the inclusions in this study. The inclusion in our example is in the liquid field at room temperature. Upon cooling, the inclusion follows a stable isochore from the liquid field into the stability field of solid (through point S) along the metastable extension of the liquid- field isochore because an inclusion must be supercooled be- fore it freezes. A solid generally needs a greater amount of supercooling to nucleate than does a bubble. In these inclu- sions, solid nucleates at 30 to 40°C supercooling, but a bubble at only 10 to 15°C. A bubble therefore is always the first to nucleate even though homogenization temperatures are lower than the triple point. Upon nucleation of a bubble at Bl, the P-T conditions for the fluid in the inclusion will change to B2 on the metastable extension of the L + V curve to reflect the new two-phase equilibrium. Note that a part of the meta- stable extension of the isochore in the region of bubble nu- cleation is at negative pressures down to approximately - 100 bars. Liquid is at a negative pressure when a failure to nucleate a bubble stretches it metastably to occupy a volume larger than that at equilibrium (ROEDDER, 1984, p. 298). After nucleation of a bubble, two different paths for further phase transition are possible depending on whether the in- clusion is heated or cooled. If heating is started at B2, meta- stable homogenization will occur at H. Upon further heating, the inclusion follows the metastable extension of the liquid- field isochore through the solid field and then the stable iso- chore through the liquid field. On the other hand, if cooling is continued at B2, the in- clusion will follow the metastable L + V curve until the solid I I I I I I I -90 -60 -70 -60 -50 -40 zyxwvutsrqponmlkjihgfedcbaZYXWVU Y FIG. 2. Phase relationship in the CO2 system in the region around the triple point (T) showing, as thick lines, the sublimation, boiling, and melting curves (S + V, L + V, S + L) and selected isochores. Stable isochores are shown as thinner solid lines and metastable extensions as dashed curves. Numbers on isochores are molar volumes in cm3/mol. The sublimation curve and the metastable extension of the L -t V curves are shown in more detail in the inset. The phase transition illustrated with letters and dotted lines is for an inclusion with a molar volume of 36.5 cm3/mol (see text for further discussion).  898 A. M. van den Kerkhof and S. N. Olsen CO2 nucleates at F2 (as determined experimentally), resulting in an instantaneous solidification of the liquid phase with a sudden but slight drop in internal pressure (to zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCB l ). Heating from this point results only in stable phases. The solid and vapor phases remain stable up to the triple point (T) where a liquid phase forms (initial melting, Ti L). The presence of three phases constrains P and T at the triple point until one of the phases completely disappears, in this case the vapor phase (partial homogenization to the liquid phase, Ths L). Upon further heating, dissolution of the solid phase proceeds along the S + L curve until it is completed at S. In our ex- ample, the molar volume of the liquid phase between T and S decreases from 37.347 to 36.5 cm3/mol, which corresponds to I. 178 and 1.2 1 g/cm3, respectively, in density. Therefore, observing the solid phase becomes increasingly difficult during “dissolution” as the density of liquid approaches that of solid (1.56 g/cm3). Melting temperature of pure CO2 in the absence of a vapor phase must be higher than the CO* triple point, as observed. Such phase behaviors for high-density COZ in- clusions have been discussed by zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA WANENBERC; (1980), but are not substantiated because his inclusions turned out not to contain pure COZ (KERKHOF, 1988a: KERKHOF et al., 1990). Comparison wilh phase behavior in CO*-N2 inclusions The phase behavior described above for high-density COZ is similar to that of some inclusions reported from the Furua Granulite Complex, Tanzania (COOLEN, 1980: KERKHOF, 1988b). Raman analyses, however, show that the Furua in- clusions (which are Type S3, as are the inclusions in this study) contain CO*-N2 fluids with 25 to 30 mol% Nz and densities of approximately 0.95 g/cm3. The density corre- sponds to a molar volume of 40 cm3/mol (compare Fig. 2). These Type-S3, C02-Nz inclusions have Ti L = -65”C, Ths L = -63”C, Ts L = -60 to -61 “C for stable phase transitions, and Thm L = -63°C for metastable ones. Their homogenization temperatures are close to or slightly below those of melting. Although the phase behaviors of high-density COZ and lower-density CO*-N2 inclusions are similar, they are not identical. A careful observation therefore can help distinguish the two types. The most important differences can be summarized as fol- lows: 1) 2) (3) (4) Initial melting in CO*-Nz inclusions (=Ti L) occurs at lower temperatures (_ -65°C for the Furua inclusions) than for pure CO2 (-56.6”C). Final melting of high-density COz in the absence of vapor (=Ts L) takes place at higher temperatures than the triple point of COZ (-56.6’C). Filling degrees of solid, or the volume solid occupies in frozen inclusions, are lower in C02-N2 inclusions than in COZ inclusions. Type-S3 inclusions are rare in C02-N2 inclusions. More common types are H3, Sl, S2, and S4 (see KERKHOF, 1988b). Inclusions which show only one transition (S + L + L: Ts L) in the temperature range -180 to +35”C were also found in the samples from the Furua Granulite Complex (COOLEN, 1980; KERKHOF, 1988b). These inclusions, denoted as Type Sl, can also be mistaken for high-density CO2 in- clusions. They contain CO>-N2 mixtures with higher N2 con- tents (-30 to 40 mol%) than S3 inclusions. Final melting temperatures in the absence of vapor (Ts L) range from --6 1 to -59°C: those of metastable homogenization (Thm) range from -74 to -64”C, with the lowest temperatures corre- sponding to critical homogenization. The other types of C02- Nz and C02-CH4 inclusions showing melting or sublimation (S2 and S4) contain three phases at low temperatures (- - 180°C) and can be easily distinguished from high-den- sity CO2 inclusions. Raman ana s The purity of the CO* inclusions in this study was checked by Raman analysis using a Microdil-28 Raman microspec- trometer (BURKE and LUSTENHOUWR, 1988) equipped with a multichannel detector. The inclusions were checked for the presence of Nz, CH4, and H2S, but no component other than CO2 could be detected. The Raman spectrum of COZ consists of two peaks, referred to as the Fermi diad, with a higher- intensity peak (2~) at 1388 and a lower-intensity one (Y,) at 1285 cm-’ at 1 atm (Fig. 3). Small peaks located at the sides of the main peaks are hot bands, which represent interaction frequencies (v, + u2 and ul + 3~~) due to the thermal energy of the molecules (HOWARD-LOCK nd STOICHEW, 197 1). In- ternal pressures of CO1 inclusions with a density of 1.21 g/ cm3 should be on the order of 2 kb at 25°C. The Raman spectrum of a high-density CO2 inclusion is compared to those of lower density inclusions in Fig. 3. The following changes in the Raman spectrum reflect higher pressures in the inclusions (GARRABOS t al. 1980; BERTRAN, 1983): I ) I t FIG. 3. Raman spectra of C@ with four different densities (from the lowermost to the uucertnost s.pecUum d = 1.2, 0.8, 0.7, and ~0. I, respectively). The-lowermost spectrum (d = - 1.2 g/cm3) was obtained from one of the Type-S3 inclusions of Table 1 (lO-3T-7- a5-a). The Raman spectra for other “superdense” inclusions in Table I a& very similar to the one shown. The other three spectra in Fig. 3 are for “normal” H3 inclusions. Spectra of hig&er-density luids are characterized by a shift to lower wavenumbers, an increased ratio in intensity of 2v2 o v2 peak, and suppression of the two hot bands (see text for further discussion).  Superdense CO2 nclusions in migmatites 899 a shift in the positions of both peaks to lower frequencies (for pressures lower than 6 kbar); (2) decreases in peak widths; (3) an increase in the frequency separation of the two peaks (A = zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA ~ - v, ; and (4) an increase in peak intensity ratio (R = 12,/I,,). In addition, we also observed decreases in inten- sities of the hot bands (this study). The peak positions of monophase CO2 inclusions measured at 20°C are different for the gas (d < 0.19 g/cm3) and the liquid (d > 0.77 gJcm3): the 21~2 eak is located at 1388 to 1387 cm-’ for the gas and 1385 to 1384 cm-’ for the liquid; the V, peak is located at 1285 to 1284 cm-’ for the gas and 128 1 to 1279 cm-’ for the liquid. There is a 2 cm-’ difference between the peak positions for the vapor and liquid in two- phase inclusions with total densities between 0.19 and 0.77 g/cm3 (BURKE and zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA ERKHOF npubl. data). Peak shifts within the gas and liquid regions are small compared to the accuracy in measurement of Raman lines in the range of the Fermi diadtihe accuracy in peak positions is +0.8 cm-’ for the instrument used. The instrument has been calibrated with a neon lamp and regularly checked by measuring the Raman peak of diamond at 1332 cm-‘, located between the two CO1 peaks. The position of the V, peak is shifted more than the 2vZ peak with increasing density so that the frequency sepa- ration (A = 2v2 - v,) also increases with density. The param- eter has been used as an indicator of densities in fluid inclu- sions (PASTERIS and WANAMAKER, 1988). Low-density gas inclusions have A - 103 cm-‘, while the high-density inclu- sions (d - 1.2) of this study have A - 106.5 cm-‘-which is in acceptable agreement with the calculated value of - 105.9 cm-’ (after BERTRAN, 1983). According to DUBESSY 1989X the sum of the relative Ra- man cross-sections for the two peaks remains constant. The relative peak intensities (R = 12,&J of the Fermi diad, on the other hand, gradually change from about 1.6 for low den- sities to 1.95 for the density of the triple point (GARRABOS et al., 1980). The values of R for the present inclusions with even higher densities than that of the triple point 1.178 cm3/ mol) are 1.95 and 2.0, consistent with those of GARRABOS et al. (1980) within the error. The accuracy of these values is no better than +O.l with the instrument used. The suppression of the low-intensity hot bands occurs in high- density fluids at constant temperature due to more restricted vibrational freedom of the molecules. We conclude that our observations from natural inclusions are in accordance with the experimental data of GARRABOS et al. (1980) and BER- TRAN 1983). DISCUSSION The extremely dense CO2 inclusions of this study belong to the group of CO2 inclusions considered to be the earliest inclusions in the Front Range migmatites. These early inclu- sions have negative crystal shapes and occur as either isolated inclusions, in small clusters (two to 10 inclusions), or in small intragranular healed fractures (up to about 100 inclusions), but never in a healed fracture that crosscuts a grain boundary. They occur more commonly in leucosomes (leucocratic layers and lenses) than in paleosomes (host rocks), and their rep- resentative isochores are consistent with the P-T conditions of migmatization estimated from the mineralogic data. These observations led OLSEN 1987) to conclude that these early CO* inclusions had formed during the migmatization episode. The isochores for the extremely dense inclusions in this study, however, are not consistent with the P-T conditions of mig- matization (Fig. 4). Nevertheless, these “superdense” inclu- sions do not belong to a special group of inclusions but to a general population of synmigmatization inclusions, because: (1) all of the inclusions have a similar mode of occurrence; (2) there is a continuum in homogenization temperature of the early CO* inclusions, including “superdense” inclusions (from +30 to -64’C; OLSEN, 1987, Fig. 4); (3) each of the superdense inclusions occurs in a cluster together with less dense inclusions; and (4) this study has shown that the su- perdense inclusions contain pure COZ, as less dense inclu- sions do. The srcin of the high-density CO2 inclusions in the Front Range rocks has been discussed by OLSEN (1988). The fol- lowing is a summary of her conclusions: 1) 2) (3) The high-density inclusions are not relics of an earlier high-pressure metamorphism which have escaped de- crepitation, because there is no correlation between the size and density of inclusions (OLSEN, 1988, Fig. 5) and because high-density inclusions are rare in the zones of migmatites with the least signs of deformation and melt- ing (paleosomes and quartz inclusions within feldspar grains). The superdense and other high-density inclusions (with Th < - -30”) must have been modified after their for- mation, because inclusions with extremely high-density CO2 tend to be associated with micro-shear zones in these rocks and because high-density inclusions are associated with plagioclase which has been heavily altered by hy- drothermal activity during a late event in leucosome for- mation. A change in composition of the fluid-for example, by a loss of water-is not likely to have occurred (see OLSEN, 1988, for further discussion). Olsen also postulated that the inclusions were modified to smaller volumes (= higher densities) during nearly isobaric cooling. If the host rock for an inclusion is cooled isobarically, the cooling path must intersect COZ isochores with increas- ingly higher densities (Fig. 4). If the volume of an inclusion can be equilibrated to the new P-T conditions (by volume decrease), a higher density for the inclusion would result. Such reequilibrations in density have been demonstrated ex- perimentally for synthetic aqueous inclusions by F&HER and BOULLIER 1984) and STERNER and BODNAR 1989). The observed wide variation in Th or density variation (see OLSEN, 1988, especially Figs. 3 and 6) indicated selective increases in density. Olsen suggested hydrolytic weakening of quartz (e.g., BLACK and CHRISTIE, 1984) as the factor for the selectiveness because of the association of high-density inclusions with hydrothermally altered plagioclase. The pres- ence of water in quartz immediately adjacent to an inclusion could have caused localized hydrolytic weakening and facil- itated the process of a volume decrease during deformation. There is no apparent correlation between the presence of aqueous inclusions and high-density CO* inclusions in these rocks, but this is not inconsistent with the model because
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks