raman spectroscopy for fluid inclusion a

Upload: joel-pedrosa

Post on 06-Jul-2018

215 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    1/21

    This article appeared in a journal published by Elsevier. The attached

    copy is furnished to the author for internal non-commercial research

    and education use, including for instruction at the authors institution

    and sharing with colleagues.

    Other uses, including reproduction and distribution, or selling or

    licensing copies, or posting to personal, institutional or third partywebsites are prohibited.

    In most cases authors are permitted to post their version of the

    article (e.g. in Word or Tex form) to their personal website or

    institutional repository. Authors requiring further information

    regarding Elsevier’s archiving and manuscript policies are

    encouraged to visit:

    http://www.elsevier.com/copyright

    http://www.elsevier.com/copyrighthttp://www.elsevier.com/copyright

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    2/21

    Author's personal copy

    Raman spectroscopy for  uid inclusion analysis

    Maria Luce Frezzotti  a,b,⁎, Francesca Tecce  b, Alessio Casagli  a

    a Dipartimento Scienze della Terra, Università di Siena, Via Laterina 8, 53100 Siena, Italyb Istituto Geologia Ambientale e Geoingegneria - CNR, c/o Dipartimento Scienze della Terra, Università  “ La Sapienza” , P.le Aldo Moro 5, 00185 Roma, Italy

    a b s t r a c ta r t i c l e i n f o

     Article history:Received 7 June 2011Accepted 18 September 2011Available online 25 September 2011

    Keywords:

    Raman spectroscopyFluid inclusionsGeological  uidsRaman spectra database

    Raman spectroscopy is a versatile non-destructive technique for  uid inclusion analysis, with a wide  eld of applications ranging from qualitative detection of solid, liquid and gaseous components to identication of polyatomic ions in solution. Raman technique is commonly used to calculate the density of CO 2  uids, thechemistry of aqueous   uids, and the molar proportions of gaseous mixtures present as inclusions. Ramanspectroscopy has been applied to measure the  pH  range and oxidation state of  uids. The main advantagesof this technique are the minimal sample preparation and the high versatility. Present review summarizesthe recent developments of Raman spectroscopy in uid inclusions research to provide support for laboratoryanalyses.

    © 2011 Elsevier B.V. All rights reserved.

    Contents

    1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

    2. Fundamentals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33. Methods of analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44. Gaseous  uids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

    4.1. CO2  uids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64.2. Gaseous mixtures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

    5. Aqueous uids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85.1. Analyses of solutes: monoatomic ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85.2. Analyses of solutes: polyatomic ions and molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

    6. Identication of mineral phases: a catalog of reference Raman spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156.1. Native elements, halides, oxides and suldes (Table 2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156.2. Carbonates (Table 3). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156.3. Sulfates, phosphates, and borates (Tables 4 and 5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166.4. Silicates (Tables 6 and 7) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

    7. Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

    1. Introduction

    Fluid inclusions (Fig. 1) represent the only  rst-hand informationon   uids in the Earth's interior (e.g.,  Roedder, 1984; Wilkinson,2001). They are acknowledged in an enormous range of lithologies(e.g., hydrothermal ore deposits, metamorphic rocks, igneous rocks,and geothermal systems), and pressure and temperature conditions.

    Fluid inclusions are generally small closed volumes (i.e.,  b50 μ m in di-ameter; Fig. 1), in which pressure and temperature are interdepen-dent variables. Both are related by the equation of state of theenclosed uid, resulting in a nearly linear relation in the  P –T  space(isochore). Therefore, a key requirement for research and applica-tions is the ability to characterize   uid composition and density.These two properties are usually obtained by petrographic and micro-thermometric methods (Poty et al., 1976).

    Raman spectroscopy is the non-destructive technique which bet-ter characterizes liquid and gaseous compounds, solid phases, and

    solute species in 

    uid inclusions. One of the main advantages is that

     Journal of Geochemical Exploration 112 (2012) 1–20

    ⁎  Corresponding author. Tel.: +39 0577 233929; fax: +39 0577 233938.E-mail addresses: [email protected] (M.L. Frezzotti),

    [email protected] (F. Tecce), [email protected] (A. Casagli).

    0375-6742/$ –  see front matter © 2011 Elsevier B.V. All rights reserved.doi:10.1016/j.gexplo.2011.09.009

    Contents lists available at  SciVerse ScienceDirect

     Journal of Geochemical Exploration

     j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / j g e o e x p

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    3/21

    Author's personal copy

    it allows the chemical and structural characterization of samples assmall as 1 μ m in diameter, a resolution not possible by conventionalpetrography, microthermometry, and other spectroscopic methods(e.g., infrared spectroscopy). Raman spectroscopy has become a con-ventional method in   uid inclusion research starting from the 70's(Burke and Lustenhouwer, 1987; Dhamelincourt et al., 1979; Dubessy

    et al., 1982, 1989; Guilhaumou, 1982; Pasteris et al. 1986, 1988;Rosasco et al., 1975; Seitz et al., 1987). The continuing interest and

    Fig. 1. Photomicrographs of  uid inclusions: a) primary H2O  uid inclusions aligned following chevron halite bands, evaporite from Vitravo diapir, Crotone, Italy. b) Primary H2Ouid inclusions in anhydrite from a geothermal well (2410 m depth), Sabatini Volcanic District, Italy. c) Plane of liquid-rich and vapor-rich H 2O  uid inclusions in sanidine fromsyenite, Sabatini Volcanic District, Italy. d) H2O  uid inclusion containing calcite and anhydrite daughter minerals (same provenance as in c). e) Tri-phase H2O–CO2 (L 1+L 2+G)uid inclusions from an Alpine quartz vein, Binn, Switzerland. f) CO2  uid inclusions in orthopyroxene, peridotite from Italy.

    Fig. 2. Energy level scheme for elastic (Rayleigh) and inelastic (Raman) scattering atthe frequency of the light source (νl), and Raman and Rayleigh spectra. The molecularvibration of the analyzed sample is of frequency  νm.   Fig. 3. Schematic diagram of a Raman spectrometer.

    2   M.L. Frezzotti et al. / Journal of Geochemical Exploration 112 (2012) 1– 20

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    4/21

    Author's personal copy

    the importance of this technique is demonstrated by the number of publications and of review papers in this research  eld (e.g., Burke,1994, 2001; Burruss, 2003; McMillan et al., 1996; Nasdala et al.,2004).

    Present review gives an introduction to Raman spectroscopy forthe analysis of geological  uids trapped as inclusions. Our approach

    is instructional and we focus on selected examples from the literatureand from our laboratory experience, but only as far as concerning theroutine analysis. The theoretical and experimental treatment of thisspectroscopy is on a basic level, and more advanced approaches,such as high-pressure and/or temperature and cryoscopic Ramanmeasurements of   uid inclusions are not discussed in detail. As arst step toward the use of Raman spectroscopy for the study of geo-logical   uids, we provide a catalog of reference spectra for mainphases that can be present in  uid inclusions.

    2. Fundamentals

    Raman spectroscopy is based on inelastic scattering of light by

    matter in its solid, liquid, or gas state. Monochromatic light scatteredby matter contains radiations with frequencies different from theexciting light. This effect, predicted by Smekal (1923), was demon-strated by Raman (1928), and named after him. The discovery of anew optical scattering phenomenon won him the Nobel prize inphysics in 1930. In several liquids Raman observed scattered light,which had energy greater than the incoming light (Raman anti-Stokes, see below). The observation of an increase in energy con-vinced him that he was in presence of a new light-scattering effect,since energy decreasing light-scattering, such as  uorescence, wasalready known at that time (Raman and Krishnan, 1928). Landshergand Mandelstam (1928a,b) also found this effect independently andalmost simultaneously in Moscow.

    A straightforward way to explain the Raman scattering of light isby quantum mechanical model, which considers the interaction of photons with molecules in terms of energy-transfer mechanisms(cf., Colthup et al., 1975; Karr, 1975, and references therein). A mol-ecule has different vibrational energy levels, the ground state  n = 0,and the excited states   n =1, n =2, n =3 etc., which are separated

    by a quantum of energy  ΔE = hν m, where  h  is the Plank's constantand ν m is the frequency of the molecular vibration. The incident vis-ible light (λ=400–750 nm) with energy   ν l induces transitions tovirtual vibrational energy levels in molecules. A virtual level is notan actual energy level of the molecule and it is generated whenlight photons interact with the molecule, raising its energy. This vir-tual level is unstable, and light is instantaneously released as scat-tered radiation.

    Returning to theinitialstateoccurs by emitting light of frequencyν l,   ν l−ν m, and   ν l+ν m. The concept is illustrated in   Fig. 2. TheRayleigh or elastic scattering occurs when the transition starts andnishes at the same vibrational energy level without loss of energy(i.e., no frequency change;ν l). Inelastic scattering (Raman effect) in-duces a change to lower (ν l−ν m) and higher (ν l+ν m) frequencies

    in scattered light, which are known as Stokes and anti-Stokes lines,with   ν m representing a fundamental rotational, vibrational or latticefrequency of the molecule. Rayleigh scattering can account for thewide majority of light scattered by molecules, being the Raman ef-fect extremely weak  –  in the order of some 10−6–10−8 of incidentphotons  –  and variable, as the intensity of the Raman scattering isproportional to the fourth power of the frequency of the incidentlight.

    Raman spectroscopy is the measurement of the photons arisingfrom inelastic (Raman) scattering of light. A Raman spectrum is theplot of light intensity expressed as arbitrary units, or counts, versusthe frequency of scattered light (i.e., Raman vibrational modes) infrequency units (wavenumbers   ˜ ν= νc =

     1λ

      in cm−1, where  c   is the

    d   Rutile

    4000

    180

    Rt

    Rt

    I

    -1

    C

    C

    a b

    c e   CalciteDiamond

    DmdCc

    300 600 900 1200 1500   cm

    Fig. 4. Raman spectral images of daughter mineral distribution in an aqueous  uid inclusion. a) Optical microphotograph of analyzed  uid inclusion in garnet from ultra-high pres-

    sure metamorphic rocks, western Italian Alps, reporting the grid of single point measurements. b) Single point Raman spectrum showing the selected wavenumber intervals fordaughter mineral mapping [diamond (red), rutile (blue), and calcite (green)]. c, d, and e) Spectral images of diamond, rutile, and calcite distribution in the  uid inclusion. Thecolor intensity of the mineral phases (from black to white) reects the increase in the intensity of the Raman band. The aqueous  uid in the inclusion has no signicant Ramansignal in the investigated region, and thus does not interfere with the measurement; modied from Frezzotti et al. (2011).

    3M.L. Frezzotti et al. / Journal of Geochemical Exploration 112 (2012) 1– 20

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    5/21

    Author's personal copy

    velocity of light; Fig. 2). Typically, only Stokes Raman scattered fre-quencies are presented since they have the same energy but areabout 10 times more frequent than their anti-Stokes counterparts.The Rayleigh scattered frequency (i.e., light-source wavenumber)lies at 0 cm−1 and Raman frequencies are expressed as relative wave-numbers, or Raman shifts. On this scale, frequencies correspond to

    the energy levels of different molecular vibrations and are indepen-dent from the wavelength of the light source: a mode at 464 cm−1

    will occur whether the light source wavelength is 514.5 or 632.8 nm.

    A spectrum comprises one or more bands which reect the vibra-tional energies of the molecules within the analyzed sample; thesein turn are related to the nature of the bonding. Main molecular vibra-tions include stretching and bending modes, stretching frequenciesbeing generally higher than bending frequencies. In order fora normalmode of vibration to be Raman active, it should produce a change in

    the polarizability of the molecule. The“selectionrules” for Raman scat-tering depend on: 1) the creation of an induced dipole in the molecule(polarization); 2) the modication of the dipole by a molecular vibra-tion; 3) the successive scattering of a photon from the modied dipole(McMillan and Hess, 1988, and references therein). As a thumbnailrule, those molecules which are not easily polarized are poor Ramanscatterers. One example is H2O which has a strong dipole momentbut electrons are not easily polarized and Raman scattering is weak.

    3. Methods of analysis

    The basic instrumental set up requires a monochromatic lightsource, generally a laser, focused on a sample (solid, liquid, or gas-eous); the light is scattered, collected at a 90° or 180° angle, and ana-

    lyzed by a detector (Fig. 3). The 

    rst dispersive Raman spectrometershad the sun or a mercury lamp as the exciting source, a prism mono-chromator as the light disperser, and a photographic  lm as detector(Colthup et al., 1975; Kohlrausch, 1943). In modern commercial in-struments, polarized laser light sources in the UV, visible, and IR areused to excite molecular samples, because of the high intensity andnarrow bandwidth of wavelengths that are emitted (monochromatic-ity), and multi-channel charge-coupled devices (CCD) are generallyused as detectors. Their combination, together with notchholographiclters to eliminate the Rayleigh line, results in more intense Ramanscatter, with considerably reduced measuring time in obtaining high

     Table 1

    Main Raman vibrations (cm−1) of major gaseous species and of solutes in aqueousuids.

    Gasses Main vibrations Ref.

    COS 857 1SO2   s 1151 2

    w 524CO2 Fermi doublet s 1285 3vs 1388

    13CO2   w 1370O2   1555 1CO 2143 1N2   2331 4H2S 2611 1C3H8   2890 1CH4   vs 2917 5

    w 3020C2H6   2954 1NH3   3336 1H2   vs 4156 6

    w 4126w 4143w 4161w 1032w 586w 354

    H2O vapor vs 3657–3756 7w 1595

    Solutes Main vibrations Ref.

    Si(OH)40 750–800 8, 9

    Si2O(OH)60 590–680 8, 9

    ClO4− vs 928 10

    w 645w 460

    SO42− vs 980 10

    w 620w 450

    NO3− vs 1049 10

    w 690

    w 1355HSO4

    − vs 1050 11w 890

    HCO3− vs 1017 12

    m 1360CO3

    2− vs 1064 12w 684m 1380

    CO2 in solution vs 1384 12m 1276

    HS− and H2S 2570–2590 11NH4

    + vs 3040 13sh 2870

    B(OH)30 vs 877 14

    w 495H2O liquid vs 2750–3900* 15 a,b

    w 1630

    vs = very strong; m = medium; w = weak; sh = shoulder; * Broad bands of severalhundreds of cm−1; 1 Burke, 2001; 2 Herzberg, 1945; 3 Rosso and Bodnar, 1995; 4Herzberg, 1950; 5 Brunsgaard-Hansen et al., 2002; 6 Dubessy et al., 1988; 7 Fraleyet al., 1969; 8  Zotov and Keppler, 2000; 9   Hunt et al., 2011; 10   Ross, 1972; 11Dubessy et al., 1992; 12  Davis and Oliver, 1972; 13  Schmidt and Watenphul, 2010;14 Schmidt et al., 2005; 15 a,b Walrafen, 1964, 1967. Ref. = References. Underlinedvibrations indicate most intense Raman modes.

    Fig. 5. Raman spectra and relative wavenumbers of most common gaseousuid speciesin  uid inclusions. Note that the hypothetical CO2 Raman band at 1340 cm

    −1 is reallytwo bands at 1285 and 1388 cm−1, see text.

    4   M.L. Frezzotti et al. / Journal of Geochemical Exploration 112 (2012) 1– 20

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    6/21

    Author's personal copy

    signal to noise spectra (i.e., tens of seconds), and low detection limits.A detailed description of the different instrumental set up can befound in scientic Journals (e.g., Vibrational Spectroscopy, Elsevier;and Journal of Raman Spectroscopy, Wiley) and in the web at the

    pages of single manufacturing companies.Fluid inclusion analysis is based on the samefundamental principle:

    the laser excites the molecules to generate scattering. Raman

    microspectrometers are the common analytical set up, where the exci-tation of the sample and collection of the scattered light at a 180° angle(backscattering) are achieved using a ordinary optical microscope fo-cused within singleuid inclusions by means of high-magnication ob-

     jectives (50× or 100×). Instruments offer perfect visualization of thesubsurface of samples and of the laser spot, which makes easy the

    choice of the appropriate inclusion to be analyzed. The volume of theanalyzed sample (spot size) depends mostly on the numerical aperture(N.A.) of the objective, and on the excitation wavelength. As an exam-ple, fora 514.5 nm excitationsourceand a 100× magnicationobjectivewith N.A.=0.9, the spot size is 1 ×1 ×5 μ m3.

    Thick double-polished sections are easily studied and require nospecial preparation. Fluid inclusions can be studied down to 1 μ m di-ameter in situ, where microstructures are preserved and the differentpopulations of  uid inclusions can be discriminated. This is possiblebecause of the confocal arrangement of the optical pathway which al-lows a good spatial resolution perpendicular to the optical axis, aswell as along the optical axis of the microscope (depth) (see, Nasdalaet al., 1996, 2004). However, the depth resolution degrades with in-creasing optical penetration depth, therefore it is better to analyze

    uid inclusions not deeper than 30 μ m within a sample.The choice of laser wavelength inuences the performance of thespectrometer. The characteristics of each laser are different, so thatno laser may be ideal for every   uid inclusion analysis. In general,the optical power of the laser line and the ef ciency of Raman CCDdetectors tend to increase with decreasing wavelength. However,the cost of the laser, the likelihood of  uorescence (see below), andthe risk of sample heating increase as well. The most popular choicesare: (1) the green light Ar ion (λ=514.5 nm) water- or air-cooled;(2) the blue light Ar ion (λ=488 nm) air-cooled; and (3) the redlight He\Ne (λ=632.8 nm).

    Raman microspectrometers can be equipped with a programma-ble  x– y  microscope stage which allows sample areas to be mappedin the same way as with EDS and WDS microprobes. Single spot spec-tra are collected by multiple steps within a grid pattern, as illustratedin Fig. 4a. Each analyzed point contains the information of a wholespectrum (Fig. 4b). Generated Raman maps are chemical or structuralimages where integrated areas of single bands or band ratios, charac-teristics for the presence of a certain chemical species in a compositesample, are illustrated (Figs. 4c, d, e). The x– y resolution in a map de-pends on the distance between the single measuring points, while thedepth resolution along  z   is determined by the confocal instrumentsettings (see above). The best resolution is achieved by setting thedistance between two measuring points smaller than the laser spotsize (“oversampling”). By increasing the distance between twospots, the spatial resolution decreases, but larger areas can be ana-lyzed in a shorter time. Spatially resolved Raman spectra can beused to identify the distribution of uid or mineral species within sin-gle  uid inclusions (Frezzotti et al., 2011; Korsakov et al., 2011).

    Fluorescence and the presence of overlapping bands from hostmineral are possible competing effects during analysis, since theyoften overpower and conceal the weak Raman features from theuid inclusions. Fluorescence generally appears as a very broad back-ground, often much more intense than the Raman scattering. This ef-fect may commonly arise from epoxy used to embed or polish therock sections and can be easily eliminated using non-uorescent ep-oxies and/or cleaning the sample. However,  uorescence can also beemitted by  uids contained in inclusions (e.g., hydrocarbons) or bythe surrounding host mineral (e.g., Fe-bearing minerals). These lastcases are much more dif cult to cope with. Increasing the wavelengthof the light source is a way of overcoming  uorescence: red or near-infrared lower lasers (λ=630–1060 nm) should not, in principle,give rise to  uorescence (Carey, 1999). Another practical method to

    mitigate a uorescent background consists in repeating spectral accu-mulations for several times in order to bleach out this effect by pro-tracted exposure to laser light (photo-bleaching).

    Fig. 6. Raman spectroscopy applied to CO2 density measurement. a) Main spectral fea-tures of CO2  uids, which consist of the two bands of the Fermi doublet, bounded bythe hot bands. The distance between the Fermi doublet (Δ) depends on  uid density.

    b) Superdense CO2  uid inclusions (d  N1.178 g/cm3

    ) spectral features, including: i) in-creased  Δ  (≥106 cm−1), ii) shifting of bands to lower wavenumbers, iii) increasedband intensity ratio, iv) broadened band bases, and v)  attened   hot bands  (van denKerkhof and Olsen, 1990); analyzed  uid inclusions are in pyroxenes from peridotitexenoliths, Hawaii; modied from Frezzotti and Peccerillo (2007). c) CO2 density as afunction of  Δ (cm−1), as derived from the equations of: 1) Rosso and Bodnar (1995),2) Kawakami et al. (2003), 3) Yamamoto and Kagi (2006), 4) Song et al. (2009), 5)Fall et al. (2011), and 6) Wang et al. (2011). The inset shows that the maximum differ-ence in CO2 densities derived from the different equations is about 0.1 g/cm

    3; redrawnand modied from Wang et al. (2011).

    5M.L. Frezzotti et al. / Journal of Geochemical Exploration 112 (2012) 1– 20

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    7/21

    Author's personal copy

    Interpretation of spectra of crystalline phases is often complicated,due to the fact that Raman scattering intensity depends upon latticeorientation. Consequently, variations of band intensity ratios shouldbe taken into account in the analysis of most minerals. Knowledgeof the orientation of main crystallographic axes, and/or repetition of analysis after 90° rotation to get random orientations is helpful inmineral identication (Nasdala et al., 2004). In addition, due to latticegeometries, some minerals are very weak Raman scatterers. Unfortu-nately, among these there are major chloride species (e.g., NaCl, KCl,and CaCl2), which represent relevant constituents of aqueous   uidinclusions.

    The intensity of the Raman scattering can vary by many order of magnitudes depending on the nature of the molecules. Detectionlimits for single components within a single uid inclusion dependon several contributing factors, including   uid inclusions size andgeometry (i.e., number of molecules of the analyzed constituent),nature of the other constituents in   uid inclusions, and analytical

    conditions (e.g., intensity of the laser light, depth of the inclusionin the analyzed sample, etc.). Several approaches can be used, andthey will be discussed in the following sections.

    4. Gaseous  uids

    A custom application of Raman spectroscopy to   uid inclusionanalysis is the qualitative identication of major gaseous uid compo-nents. The characterizing Raman bands for most important geologicaluids are reported in Table 1 and Fig. 5. Most gasses show a singlesymmetric stretching strong band, whose wavenumber is traditional-ly reported at ambient   P –T   conditions, since a progressive slightwavenumber downshift is known to occur with increasing  uid den-sity (Burke, 2001; van den Kerkhof, 1988b).

    Early work on  uid inclusions allowed to recognize CO2, CH4, andN2  as relevant geological  uids (e.g., Dubessy et al., 1989; Frezzottiet al., 1992; Touret, 2001; van den Kerkhof, 1988a,b, 1990). H2S,

    COS, SO2, CO, H2, NH3   and O2   have also been detected in appre-ciable amounts in some   uids (Bény et al., 1982; Ferrando et al.,2010; Frezzotti et al., 2002; Giuliani et al., 2003; Grishina et al.,1992; Peretti et al., 1992; Siemann and Ellendorff, 2001; Tsunogaeand Dubessy, 2009). Identication of hydrocarbons heavier thanCH4   is also possible (e.g.,   Guilhaumou, 1982; Hrstka et al. 2011;Makhoukhi et al., 2003; Munz, 2001; Orange et al., 1996; Pironon,1993; Pironon and Barrès, 1990; Potter et al., 2004; Rossetti andTecce, 2008; Schubert et al., 2007; Weseł ucha-Birczyńska et al.,2010), although uorescence often does not allow conventional anal-ysis (see e.g., Pironon et al., 1998).

    4.1. CO 2 uids

    The Raman spectrum of molecular CO2 shows two strong bands at1285 and 1388 cm−1, and two symmetrical weak bands below 1285

    and above 1388 cm−1, the so-called hot bands  (Colthup et al., 1975;Dhamelincourt et al., 1979; Dubessy et al., 1999; Rosasco et al.,1975; Rosso and Bodnar, 1995; van den Kerkhof and Olsen, 1990).The two sharp bands appear because of a resonance effect, proposedby  Fermi (1931)  in order to explain the doublet structure in theregion of CO2   symmetric stretching vibration. A small peak at1370 cm−1 is the   13CO2.

    Fig. 6a and b shows examples of spectra of CO2   uid inclusionshaving different densities. The distance between the Fermi doublet(Δ, in cm−1) is proportional to  uid density (Garrabos et al., 1980;van den Kerkhof, 1988b; Wang and Wright, 1973). Several equations(e.g., Fall et al., 2011; Kawakami et al., 2003; Rosso and Bodnar, 1995;Song et al., 2009; Wang et al., 2011; Yamamoto and Kagi, 2006 ) havebeen proposed to calculate the density (d) of pure CO2 uid inclusions

    based on the distance between the Fermi doublet  Δ   (Fig. 6c). CO2density can be determined in the range from 0.1 to 1.24 g/cm 3 withan accuracy better than 5% (Wang et al., 2011).

    Fig. 7. Quantitative Raman analysis of H2 and CH4 contained in the gas bubble of an aqueous  uid inclusion in vesuvianite from rodingites, western Italian Alps (Ferrando et al.,2010). Relative mole% of H2 and CH4 in the gas bubble is calculated with Eq.  (1) based on band area integration, and considering the relative Raman cross sections (σ ) and the in-strumental ef ciency (ζ) at the wavenumbers of H2 and CH4. (σ  of CH4 is 3.5 times higher than of H2; Burke, 2001).

    6   M.L. Frezzotti et al. / Journal of Geochemical Exploration 112 (2012) 1– 20

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    8/21

    Author's personal copy

    We observe a very good agreement between density data derived

    from Raman spectroscopy and from microthermometry, also for CO2uids containing minor amounts of other gaseous species (i.e.,b5 mol% CH4   or N2;   Frezzotti and Peccerillo, 2007). These twomethods are complementary for the characterization of  uid inclu-sioncomposition and densities. Although the precision of microther-mometricmeasurements is higher, the Raman densimeter permitstoanalyze very small  uid inclusions (b5 μ m in diameter), and/or lowdensity  uids.

    The relative intensities of the   13CO2   and the associated  12CO2

    band (Fig. 6a) have been used to calculate the carbon isotope ratiosin single   uid inclusions. The development of Raman as a mass-spectroscopy, however, is still at a very early stage of development;reported   δ13C determinations have uncertainties  ≥20‰  (Arakawaet al., 2007; Dhamelincourt et al., 1979), andconsent only to discrim-

    inate between inorganic and organic CO2 at best. This is due to thedif culty in controlling all parameters inuencing intensity of scattering, probably including a dependence of   13CO2   and

      12CO2

    band intensities on   uid density. Nevertheless, Raman mass-spec-troscopy remains a particularly attractive prospective since it couldpermit to analyze samples several order of magnitude smaller thangenerally used by mass-spectrometry.

    4.2. Gaseous mixtures

    When  uid inclusions consist of mixtures of two or more gas spe-cies, the relative molar fractions of the end-members can be calculat-ed. The prerequisite to quantitative Raman analysis is the knowledgeof two essential parameters (cf., Burke, 2001): (1) the Raman scatter-ing cross-section, which indicates the activity of a certain gas compo-nent in a mixture (Schrötter and Klöckner, 1979); and (2) thevariation of the instrumental ef ciency at the different wavenumbersfor a specic excitation wavelength. The  rst parameter is dependenton the laser excitation wavelength. A list of major gas species cross-sections for the 632.8 nm red light (e.g., He\Ne laser source), the514.5 nm green light (Ar-ion laser source), and the 488 nm bluelight (Ar-ion laser source) is reported in Burke (2001). The second pa-rameter requires an empirical calibration for each Raman microspect-

    rometer, by measuring synthetic or natural gas-mixture standards of known composition and density (Beeskow et al., 2005; Chou et al.,1990; van den Kerkhof, 1988b).

    The molar fraction ( X ) of end-member components in a gas mix-ture can be obtained using the following equation (Beeskow et al.,2005; Burke, 2001; Dubessy et al., 1989; Morizet et al., 2009; Nasdalaet al., 2004; Wopenka and Pasteris, 1986, 1987):

     X a ¼

     Aaσ a   ζa

    ∑   Aiσ i   ζi

    ð1Þ

    where   X a, Aa,σ a   and   ζ a, are the molar fraction, the band area, theRaman cross-section and the instrumental ef ciency for gas a, respec-tively, whileΣ Ai,σ i, and ζ i represents the sum of values for all gas spe-cies in the   uid inclusion. In order to get reliable quantitativeanalyses, no change in the analytical conditions should be made dur-ing measurements (i.e., laser intensity, focus, number of accumula-tions, and accumulation time). Accuracy of analyses is reportedbetter than 5% (Pasteris et al., 1988; van den Kerkhof, 1988b). Notethat when CO2 uids are involved, the sum of the two bands formingthe Fermi doublet should be used (Dubessy et al., 1989).

    In Fig. 7 is reported for example an aqueous uid inclusion con-tained in vesuvianite from vein in rodingite from Bellecombe, ItalianWestern Alps (Ferrando et al., 2010). In the gas bubble, bands of CH4and H2 have been obtained using an Ar-ion laser (λ=514.5 nm) asthe excitation source. The integrated measurements of the single

    gas Raman band area ( A) are reported along with the relativecross-sections (σ ) of H2  and CH4   and the instrumental ef ciency(ζ ) of the Raman spectrometer at 2917 and at 4156 cm−1. UsingEq. (1), the resulting composition of the gas phase in the Alpine in-clusion is equal to 82 mol% H2 and 18 mol% CH4.

    In more complex gaseous–aqueous  uid mixtures, the quantita-tive analysis of the different components is much more dif cultand often requires measurements at high temperatures. Empiricalequations for (semi)quantitative analyses of H2O-CH4±NaCl andH2O-CO2±NaCl systems have been proposed based on relative bandareas in spectra (e.g.,  Azbej et al., 2007; Guillaume et al., 2003; Luet al., 2007). In these complex  uid mixtures, analysis should includedetection of gasses dissolved in water (e.g., CO2  or CH4), and thecharacterization of clathrate hydrates (ice-like compounds formed

    from CO2, CH4, or N2 and water under low-T  and high-P  conditions;Azbej et al., 2007; Dubessy et al., 2001; Fall et al., 2011; Orangeet al., 1996; Pironon et al., 1991) .

    Fig. 8. Raman spectra of water contained in  uid inclusions, presenting examples for:a) low-salinity (b1 NaCl wt.%) liquid water, b) high-salinity (20 NaCl wt.%) liquidwater, and c) optically-hidden water in a CO2-rich   uid inclusion, peridotite fromEthiopia.

    7M.L. Frezzotti et al. / Journal of Geochemical Exploration 112 (2012) 1– 20

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    9/21

    Author's personal copy

    5. Aqueous  uids

    The Raman modes of water consist of two main O\H stretchingmodes at 3657 and 3756 cm−1 and one very weak H\O\H bendingmode at 1595 cm−1 (Carey and Korenowski, 1998; Fraley et al.,1969). However, the Raman spectrum of liquid H2O consists of sev-eral large overlapping bands in the OH stretching region from 2750to 3900 cm−1 (Fig. 8a and b), and of a weak bending mode at~1630 cm−1 (Walrafen, 1964, 1967). Reduced to minimum terms,such spectral complexity results from the strong interactions of asingle water molecule with the neighboring molecules, formingintermolecular O\H\O bridging networks (Hare and Sorensen

    1992; Sun, 2009).The characteristics of the Raman spectrum of water have been

    used to prove the presence of H2O in small CO2   uid inclusions(b5–10 μ m in size;  Frezzotti and Peccerillo, 2007; Frezzotti et al.,2010; Hidas et al., 2010). Here, a water  lm of a thickness of 0.2 μ mwrapping the CO2  uid cannot be identied with optical techniques,although it may correspond to as much as 10–20 mol% of H2O. A de-tailed description of the method can be found in   Dubessy et al.(1992) and in McMillan et al. (1996). One example is illustrated inFig. 8c from CO2   uid inclusions in peridotites from Ethiopia. Thedominant spectral features of optically unnoticed water are the vibra-tional bands at 3658 and 3750 cm−1 characteristic of OH− stretchingvibrations for isolated molecules of H2O (i.e., lack of signicant Hintermolecular bonding).

    Raman spectroscopy allows determination of the appropriatewater content of melt inclusion glass in minerals of granites and peg-matites (e.g., Behrens et al., 2006; Chabiron et al., 2004; Di Muro et al.,

    2006; Severs et al. 2006; Thomas, 2000; Thomas and Davidson, 2006;Thomas et al., 2008b; Zajacz et al., 2005). Note that, during cooling of a natural water-rich melt inclusions, often SiO2 is deposited on the in-clusion wall and makes an apparent aqueous   uid inclusion fromwhat was primary a melt inclusion (Thomas et al. 2011a).

    5.1. Analyses of solutes: monoatomic ions

    Qualitative and (semi)quantitative Raman analysis of water-richuid inclusions typically focuses on determination of solutes.Monoatomic charged cations, such as Na+, K+, Ca2+, and Mg2+

    have too weak Raman spectra to be analyzed in  uid inclusions. A

    way to obtain spectra is by nucleation of salt-hydrates at low tem-peratures, but this requires the combination of the Raman micro-spectrometer with a   uid inclusion cooling stage. Spectra arereported for all major salt-hydrates, such as NaCl·2H2O, FeCl3·6H2O,CaCl2·6H2O, MgCl2·12H2O, KCl·MgCl2·6H2O, FeCl2·6H2O, LiCl·5H2O(Bakker, 2004; Baumgartner and Bakker, 2009, 2010; Derome et al.,2007; Dubessy et al., 1982, 1992; Samson and Walker, 2000; Schiffries,1990).

    Chlorine ions have the power of breaking certain hydrogen bondsin aqueous solutions. The variation of OH stretching bands induced bydifferent Cl concentrations in aqueous  uid inclusions (Fig. 8a and b)has been intensively investigated with different approaches. Semi-quantitative estimation of the salt content in aqueous  uid inclusionsrequires development of a specic calibration for each spectrometer

    and it is complementary to measurements of phase transitions atlow temperatures by microthermometry (e.g., eutectic and nal melt-ing temperatures).

    Fig. 9. Raman spectroscopy applied to solute analysis in aqueous  uids. a) Band of SO42− ions in a  uid inclusion in feldspar from syenite, Sabatini volcanic district, Italy. b) Bands of 

    native sulfur in a  uid inclusion in orthopyroxene from peridotite, Italy. c) Bands of CO32− and HCO3

    − ions in a  uid inclusion from ultra-high pressure metamorphic rocks, westernItalian Alps. d) Bands of Si(OH)4

    0, and deprotonated H4-nSiO4n− monomers in a  uid inclusion from metamorphic rocks from western Italian Alps. The Raman modes of anhydrite

    (Anh),quartz (Qtz), and Mg-calcite (MgCc)daughter minerals are also shown. Raman bands ofhost minerals are marked with asterisks. c and d: modied from Frezzotti et al., 2011.

    8   M.L. Frezzotti et al. / Journal of Geochemical Exploration 112 (2012) 1– 20

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    10/21

    Author's personal copy

    Mernagh and Wilde (1989) proposed a formula to calculate NaClwt.%, with a relative error of 15%:

    NaClwt:% ¼  α 2  Y − X ð Þ X  þ Y 

      2−   X =Y I   3400cm−1ð ÞI   3200cm−1ð Þ

    0B@1CA−β   ð2Þ

    where   X   is equal to the integral of the OH− band from 2800 to3300 cm−1, Y  is equal to the integral from 3300 to 3800 cm−1, I  is theintensity at the specied wavenumbers, and  α  and  β   are regressionparameter specic for each spectrometer (cf.,  McMillan et al., 1996).

    The idea behind Eq. (2) was to link theshape of the two halves formingtheOH stretching band to theamount of Cl− in solution. More recently,calibration curves were expanded also to LiCl, KCl, MgCl2, CaCl2, and to

     Table 2

    Main Raman vibrations (cm−1) of selected native elements, halides, suldes, oxides and hydroxides.

    Native elements, halides and suldes Main vibrations Ref.

    DiamondC 

    1332 [1]

    Graphite

    1355 1580 2

    SulfurS 8

    mw 157 m 220 s 462 3w 187 w 246 w 437

    Arsenic As

    mw 220 4w 225vs 253

    HaliteNaCl

    358 [1]

    SylviteKCl

    vw 291 [1]vw 213

    FluoriteCaF  2

    m 322 vw 641 [1]

    CryoliteNa 3(AlF 6 )

    vw 485 m 555 mw 620 5

    ElpasoliteK  2NaAlF 6 

    135 326 559 1009 6387

    PyriteFeS  2

    w 342 vs 428 7s 377

    MarcasiteFeS  2

    vs 324 7s 387

    ChalcopyriteCuFeS  2

    vs 293 w 322 7w 352w 378

    CovelliteCuS 

    vw 263 vs 471 7

    Blende ZnS 

    w 218 w 300 w 419 w 639 [1]w 274 w 310 w 669

    vs 349Galena

    PbS 

    vs 136 m 270 [1]

    Oxides and hydroxides Main vibrations Ref.

    RutileTiO 2

    w 139 m 238 vs 444 vs 609 w 920 [1]m 696

    AnataseTiO 2

    vs 143 w 395 w 514 mw 638 8vw 195

    BrookiteTiO 2

    s 127 s 247 s 318 w 412 mw 645 9vs 150 w 366

    SpinelMgAl 2O4

    w 313 vs 408 mw 666 w 76810

    MagnetiteFe 2+Fe 2

     3+O4

    w 193 w 306 s 538 vs 668 11

    Hematite s 223 vs 409 m 609 vs 1313 12Fe 2O 3   vs 290 w 498Ilmenite

    FeTiO 3

    w 232 mw 373 vs 685 13

    Gibbsite Al(OH) 3

    mw 242 m 322 vs 538 w 979 14m 255 vs 380 vs 569

    Diaspore AlO(OH)

    w 331 vs 448 3

    Corundum

     Al 2O 3

    mw 378 vs 417 m 644 w 750 15

    Goethiteα-FeO(OH)

    w 242 vs 389 m 547 mw 681 12mw 299

    vs = very strong; s = strong; m = medium;mw = medium weak; w = weak;vw = very weak; [1] = Raman SpectraDatabase,Siena Geouids Lab (http://www.dst.unisi.it/geouids/raman/spectrum_frame.htm); 2 Wopenkaand Pasteris, 1993; 3 Giulianiet al.,2003; 4 Thomas and Davidson, 2010; 5 Nazmutdinov et al.,2010; 6 R.Thomas, pers.comm.;7 MernaghandTrudu,1993; 8 Clark et al.,2007; 9 Yanqing et al.,2000; 10 Slotznick andShim, 2008; 11 Shebanova andLazor,2003; 12 Kuebleret al.,2006; 13 Rullet al., 2007; 14 Ruanet al.,2001; 15 Xuet al., 1995; Ref. = References.

    9M.L. Frezzotti et al. / Journal of Geochemical Exploration 112 (2012) 1– 20

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    11/21

    Author's personal copy

     Table 3

    Main Raman vibrations (cm−1) of selected carbonates.

    Carbonates CO32− vibrations Ref.

    T   ν4   ν2   ν1   ν3

    CalciteCaCO 3

    s 284 mw 711 vs 1085 vw 1435 [1]mw 156

    AragoniteCaCO 3

    s 154 w 704 vw 854 vs 1085 vw 1463 [1]mw 206

    VateriteCaCO 3

    m 301 w 740 vs 1090 vw 1465 2sh 118 w 750 s 1074

    Mg-Calcite(Ca,Mg)CO 3

    s 281 mw 714 vs 1087 vw 1438 [1]mw 155

    MagnesiteMgCO 3

    s 329 w 738 vs 1094 w 1444 3mw 212

    DolomiteCaMg(CO 3) 2

    s 299 w 725 vs 1097 vw 1443 [1]ms 176

    NatriteNa 2CO 3

    w 698 vs 1078 w 1428 4

    K-CarbonateK  2CO 3

    s 141 m 697 vs 1064 m 1405 5m 192 sh 1043

    ZabuyeliteLi 2CO 3

    mw 96 w 712 vs 1091 w 1459 4

    Siderite

    (Fe,Mg)CO 3

    m 301 sh 738 vs 1090 vw 1442 [1]

    w 194Rhodochrosite

    MnCO 3

    s 289 mw 718 vs 1087 vw 1416 [1]mw 185

    StrontianiteSrCO 3

    mw 149 w 700 vs 1073 vw 1450 [1]mw 183 vw 1057sh 250

    WitheriteBaCO 3

    s 136 w 692 vs 1059 w 1420 [1]m 152w 227

    CerussitePbCO 3

    s 150 m 682 m-w 839 vs 1056 s 1378 [1]mw 180sh 215

    Smithsonite ZnCO 3

    m 303 w 731 vs 1093 mw 1408 [1]mw 196

    NahcoliteNaHCO 3

    mw 688 vs 1048 w 1432 [1]m 1271

    KaliciniteKHCO 3

    w 635 vs 1028 4w 673 mw 1277

    Hydrated carbonates CO32− vibrations OH− Ref.

    T   ν4   ν1   ν3

    MalachiteCu 2(OH) 2(CO 3)

    vs 154 w 721 sh 1098 vs 1492 vs 3468 [1];vs 178 mw 3386 6vs 434ms 272ms 537

    AzuriteCu 3(OH) 2(CO 3) 2

    s 397 vs 1095 vw 1457 vs 3453 [1];m 246 sh 937 6mw 170mw 279

    ArtiniteMg  2(OH) 2(CO 3)· 3H  2O

    s 147 w 704 vs 1094 vs 3593 7s 173 s 3229 8w 472 s 3030

    Hydromagnesite

    Mg 5(CO 3)4(OH) 2·4H  2O

    m 184 vs 1119 sh 1487 n.a. 7

    m 202m 232

    DypingiteMg 5[(OH)(CO 3) 2]  2·5H  2O

    mw 203 w 727 vs 1122 mw 1447 vs 3648 8mw 249 mw 1092 m 3421w 311 mw 3515w 434

    DawsoniteNaAl(CO 3)(OH) 2

    ms 189 vs 1091 mw 1505 vs 3282 9m 260 w 1065 m 3250mw 587

    ThermonatriteNa 2(CO 3)·H  2O

    s 156 vs 1062 sh 1432 n.a. [1]m 185w 230

    TronaNa 3H(CO 3) 2· 2H  2O

    mw 140 vs 1060 w 1430 n.a. [1]mw 185w 225

    GaylussiteNa 2Ca(CO 3) 2·5H  2O

    s 164 w 723 vs 1071 vs 2944 [1];sh 265 s 3334 10

    ν1=Symmetric stretching vibration;  ν2 = Out-of-plane bending vibration;  ν3 = Antisymmetric stretching vibration;  ν4 = In-plane bending vibration; T = Translational latticemodes; OH−= OH stretching vibrations; vs = very strong; s = strong; ms = medium strong; m = medium; mw = medium weak; w = weak; vw = very weak; sh = shoulder;[1] = Raman Spectra Database, Siena Geouids Lab (http://www.dst.unisi.it/geouids/raman/spectrum_frame.htm). 2 Carteret et al., 2009; 3 Gillet, 1993; 4 Thomas et al., 2011a,b;5 Koura et al., 1996; 6 Frost et al., 2002; 7 Edwards et al., 2005; 8 Frost et al., 2008; 9 Frost and Bouzaid, 2007; 10 Frost and Dickfos, 2007; n.a. = not analyzed; Ref. = References.

    10   M.L. Frezzotti et al. / Journal of Geochemical Exploration 112 (2012) 1– 20

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    12/21

    Author's personal copy

     Table 4

    Main Raman vibrations (cm−1) of selected sulfates.

    Sulfates SO42− vibrations Ref.

    ν2   ν4   ν1   ν3

    AnhydriteCaSO4

    mw 430 w 611 vs 1018 mw 1131 2mw 500 w 629

    w 676Mg-sulfate

    MgSO4

    ms 451 s 608 vs 1023 ms 1136 3ms 475 vw 681 s 1053 w 1220ms 499 vw 697 vw 1256

    ThenarditeNa 2SO4

    w 452 w 621 vs 994 w 1103 [1]w 469 w 632 w 1153

    GlauberiteNa 2Ca(SO4) 2

    w 485 m 618 vs 1002 m 1106 [1]m 644 m 1140

    BurkeiteNa6 (CO 3)(SO4) 2

    mw 451 mw 620 vs 994 4w 474 mw 633 m 1065*

    mw 644Sulfohalite

    Na6 (F,Cl)(SO4) 2

    m 471 m 634 vs 1002 m 1125 [1]

    ArcaniteK  2SO4

    mw 457 mw 622 vs 983 w 1093 5w 1109w 1145

    Aphthitalite

    (K, Na) 3Na(SO4) 2

    m 457 s 619 vs 984 m 1104 [1]

    mw 447 mw 1093Celestine

    SrSO4

    m 452 ms 656 vs 1000 ms 1156 [1]vw 627 mw 1190

    BariteBaSO4

    s 461 w 617 vs 988 w 1143 [1]

    AnglesitePbSO4

    mw 438 w 608 vs 978 w 1160 [1]mw 450 vw 641 vw 1068

    Hydrated sulfates SO42− vibrations OH− Ref.

    ν2   ν4   ν1   ν3

    GypsumCaSO4· 2H  2O

    s 494 w 621 vs 1008 w 1142 vs 3405 [1];m 414 mw 3491 6

    EpsomiteMgSO4·7H  2O

    mw 447 vw 612 vs 984 vw 1061 vs 3303 3vw 1095 s 3425vw 1134

    ExahydriteMgSO4·6H  2O

    w 445 vw 610 vs 984 w 1146 vs 3428 3w 466 vw 1085 m 3258

    PentahydriteMgSO4·5H  2O

    m 447 vw 602 vs 1005 vw 1106 vs 3391 3vw 371 vw 1159 vs 3343

    m 3553m 3494m 3289

    StarkeyiteMgSO4·4H  2O

    vw 401 vw 565 vs 1000 w 1156 vs 3427 3vw 462 vw 616 vw 1086 s 3481

    vw 664 vw 1116 m 3558vw 1186 m 3331

    SanderiteMgSO4· 2H  2O

    m 447 w 597 vs 1034 m 1164 vs 3446 3w 492 w 630 m 3539

    KieseriteMgSO4·H  2O

    m 436 m 629 vs 1046 mw 1117 vs 3297 3w 481 w 1215

    K-AlumKAl(SO4) 2·12H  2O

    mw 455 mw 614 vs 989 mw 1130 vs 3396 7w 442 s 974 w 1104 m 3072

    Alunite

    KAl 3[(OH) 3(SO4)]  2

    mw 509 mw 654 vs 1026 mw 1190 vs 3509 [1];

    w 485 w 1079 vs 3482 8Syngenite

    K  2Ca(SO4) 2·H  2Omw 474 w 642 vs 983 w 1142 vs 3301 [1];w 494 w 662 s 1007 w 1168 s 3378 9

    GörgeyiteK  2Ca5(SO4)6 ·H  2O

    m 480 m 631 vs 1013 w 1108 vs 3525 10w 433 w 595 vs 1005 w 1115 m 3580w 440 w 602 w 1085 w 1162w 457 w 654

    MirabiliteNa 2SO4·10H  20

    m 458 mw 627 vs 989 w 1129 vs 3506 11m 3340

    CesaniteNa 3Ca 2(OH)(SO4) 3

    vs 448 vs 626 vs 1004 sh 1104 n.a. [1]mw 474 mw 647

    ν1 = Symmetric stretching vibration; ν2 = Out-of-plane bending vibration;  ν3 = Anti-symmetric stretching vibration; ν4 = In-plane bending vibration; * = Symmetric stretchingvibration of CO3   group. Peak intensities as in   Table 3. [1] = Raman Spectra Database, Siena Geouids Lab (http://www.dst.unisi.it/geouids/raman/spectrum_frame.htm);2 Thompson et al., 2005; 3  Wang et al., 2006; 4  Korsakov et al., 2009; 5  Montero and Schmolz, 1974; 6  Kloprogge and Frost, 2000; 7  Barashkov et al., 2004; 8  Frost et al., 2006;9 Kloprogge et al., 2002; 10 Kloprogge et al., 2004; 11 Hamilton and Menzies, 2010; n.a. = not analyzed; Ref. = References.

    11M.L. Frezzotti et al. / Journal of Geochemical Exploration 112 (2012) 1– 20

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    13/21

    Author's personal copy

    other more complex salt systems (Dubessy et al., 2002; Sun et al.,2010). The methods described above are all similar, they only differin the selected bands of water in the OH-stretching region taken asstandards.

    5.2. Analyses of solutes: polyatomic ions and molecules

    Polyatomic charged anions have Raman spectra characterizedby the presence of one or more bands (Table 1). Band area and

    intensity, although proportional to the solute concentration, cannotbe linearly transformed into absolute concentrations, since these areconsiderably inuenced also by measurement conditions (e.g., laserpower, optical arrangement, etc.;  McMillan et al., 1996; Nasdalaet al., 2004). Semi-quantitative analysis of polyatomic solutes inuid inclusions has been in some cases possible based on relative

    band-intensity ratios, using selected bands of water as standard.The application of intensity ratios eliminates the inuence of mea-surement conditions. Note that during analyses high laser power

     Table 5

    Main Raman vibrations (cm−1) of selected phosphates and borates.

    Phosphates PO43− vibrations Ref.

    ν2   ν4   ν1   ν3

    ApatiteCa5(PO4) 3(OH,F,Cl)

    w 428 w 578 vs 960 w 1026 [1]w 446 w 588 w 1040

    FluorapatiteCa5(PO4) 3F 

    mw 432 m 592 vs 965 m 1053 2w 449 w 608 mw 1081

    mw 581 w 1042Chlorapatite

    Ca5(PO4) 3Cl

    w 430 w 581 vs 963 mw 1039 3w 1127

    HerderiteCaBePO4 (F,OH)

    584 983 1005 4595

    Triplite(Mn,Fe,Mg,Ca) 2(PO4)(F,OH)

    425 610 980 1034 4;5

    Berlinite AlPO4

    437 1111 1229 4;461 5

    AmblygoniteLiAl(PO4)F 

    601 1011 4644

    LacroixiteNaAl(PO4)F 

    609 1001 4623

    Na-phosphateNa 3PO4

    391 524 910 5482 544 942

    993Lazulite

    (Mg,Fe)Al 2(PO4) 2(OH) 2

    344 611 1059 1100 5630 1136741

    Xenotime(Y,Yb)PO4

    485 642 998 1394 51056

    Monazite(La,Ce,Nd,Th)PO4

    m 466 m 620 vs 987 mw 1054 6

    Borates Main vibrations Ref.

    Metaboric acidHBO 2 (monoclinic)

    428 518 782 5475 533

    Metaboric acidHBO 2 (orthorhombic)

    401 595 809 5415475

    Li-metaborate

    LiBO 2

    713 1419 4

    SassoliteH  3BO 3

    w 500 vs 880 7

    HambergiteBe 2BO 3(OH,F)

    vs 153 w 992 7

    Na-tetraborateNa 2B4O7 ·10H  2O

    385 461 576 756 852 948 1036 5

    Li-tetraborateLi 2B4O7 ·5H  2O

    391 446 543 772 845 1028 5493 896 1097

    1352Borax

    Na 2B4O5(OH)4·8H  2O344 405 571 776 943 5

    463 997Ca–Mg-hexaborates

    CaB6 O10, MgB6 O10 with 4 to 7.5 H  2O

    634 852 953 4638 855 964641 861

    HydroboraciteCaMgB6 O11·6H  2O

    181 322 524 606 753 837 5257 383 548 876

    398 564Cs-RamaniteCsB5O8·4H  2O

    m 98 vs 548 m 768 m 907 8mw 293

    Rb-RamaniteRbB5O8·4H  2O

    mw 101 w 508 w 765 w 914 8vs 554

    ν1  = Symmetric stretching vibration;  ν2 = In-plane bending vibration;  ν3 = Antisymmetric stretching vibration;  ν4  = Out-of-plane bending vibration. Peak intensities as inTable 3. [1] = Raman Spectra Database, Siena Geouids Lab (http://www.dst.unisi.it/geouids/raman/spectrum_frame.htm); 2  Penel et al., 1997; 3 Kuebler et al., 2006; 4  Rickerset al., 2006; 5 R.Thomas, pers. comm.; 6  Silva et al., 2006; 7  Thomas and Davidson, 2010; 8  Thomas et al., 2008a,b; Ref. = References.

    12   M.L. Frezzotti et al. / Journal of Geochemical Exploration 112 (2012) 1– 20

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    14/21

    Author's personal copy

    could result in heating the inclusion uid with consequent possiblechanges in the speciation of ions.

    The study of the speciation of sulfur in aqueous solution to deter-mine the redox potential (H2S/SO4

    2−) and   pH   range (SO42−/HSO4

    −;HS−/H2S) of geological  uids represents one of the  rst applicationsof Raman microspectroscopy to  uid inclusion research (Boiron et al.1999; Dubessy et al., 1983, 1992, 2002; Rosasco and Roedder, 1979).Sulfate ions give rise to a main S\O stretching band at ~980 cm−1

    (Fig. 9a) and to two additional weak bands around 620, and450 cm−1 (Table 1; Ross, 1972; Schmidt, 2009). Only the 980 cm−1

    band is generally strong enough to be observed in   uid inclusions,and has very low detection limits (0.01–0.05 mol/kg; Dubessy et al.1982, 1983; Rosasco and Roedder, 1979). Bisulfate ions (HSO4

    −) canbe identied by their main S\O and S\OH stretching modes at

    ~1050 and 890 cm−1, respectively (Table 1). Hydrogen sulde (H2S0and HS−) is characterized by S\H stretching modes in the 2570–2590 cm−1 range.

    The carbonate ion CO32− fundamental stretching mode is expectedat 1064 cm−1. Other less intense bands at ~1380, and 684 cm−1 maybe observed in concentrated solutions. HCO3

    − has a very strong C\OHstretching mode at ~1017 cm−1, and a less intense C\O stretchingmode at ~1360 cm−1 (Table 1). Raman studies of carbonates andbicarbonates in solution were initiated by  Davis and Oliver (1972)and Dubessy et al. (1982), although these ions were not detected inuid inclusions at that time. Absence was attributed mainly totheir low Raman scattering compared, for example, to that of sulfateions, and to their relatively low solubility in geological uids(cf., Burke,2001; Dubessy et al., 1982; McMillan et al., 1996). More recently,there has been increasing Raman evidence for signicant HCO3

    (aq)

    and CO32−

    (aq)  in   uid inclusions (Fig. 9c) mainly from pegmatites,ore deposits, and high pressure metamorphic rocks (Frezzotti et al.,

    2011; Hrstka et al., 2011; Thomas et al., 2006, 2009a,b, 2011a; Xieet al., 2009).CO3

    2−(aq) concentrationsas low as 0.36 wt.% can be mea-

    sured using a modied technique by Sun and Qin (2011) (R. Thomas,

     Table 6

    Main Raman vibrations (cm−1) of selected orthosilicates and tectosilicates.

    Orthosilicates Main vibrations Ref.

    Forsterite(Mg 0.9,Fe0.1) 2SiO4

    227 303 423 548 608 824 921 [1]856 964882

    PyropeMg  3 Al 2Si 3O12211 364 563 650 871 902 1066 3928

    AlmandineFe 3

     2+ Al 2Si 3O12

    170 216 323 500 863 916 1038 3342 556 897370

    SpessartineMn 3 Al 2Si 3O12

    175 221 321 500 630 849 905 1029 3350 552 879

    GrossularCa 3 Al 2Si 3O12

    181 247 373 420 550 827 1007 3280 848

    880Uvarovite

    Ca 3Cr  2Si 3O12

    176 242 370 509 828 3272 526 894

    590Andradite

    Ca 3(Fe 3+, Ti) 2Si 3O12

    174 236 325 452 516 816 995 3370 494 574 842

    874Kyanite

     Al 2SiO5

    302 405 562 669 952 4325 419360 437386 486

    Sillimanite Al 2SiO5

    142 235 310 456 597 708 874 907 1127 [2]964

    Andalusite Al 2SiO5

    293 323 453 553 719 834 920 1065 [2]361 992 1111

    Zircon ZrSiO4

    202 356 438 974 1008 [2]225

    Tectosilicates Main vibrations Ref.

    OrthoclaseKAlSi 3O8

    157 284 458 514 751 814 967 1035 1137 [1]177 477 583 1062197

    MicroclineKAlSi 3O8

    159 263 455 514 651 749 813 1007 1128 [1]178 267 475 1142199 286

    SanidineKAlSi 3O8

    163 284 462 514 767 813 1123 [1]475

    AlbiteNaAlSi 3O8

    183 210 457 508 764 816 977 1032 [2];5292 480 1098

    QuartzSiO 2

    128 206 356 402 520 608 807 1066 1161 6265 464 698

    485Coesite

    SiO 2

    116 204 326 427 521 785 815 1036 1144 6151 269 355 466 837 1065 1164176

    CristobaliteSiO 2

    114 230 420 792 1075 [2]273286

    [1] = Raman Spectra Database, Siena Geouids Lab (http://www.dst.unisi.it/geouids/raman/spectrum_frame.htm); [2] = Raman Spectra Database Lyon (http://www.ens-lyon.fr/LST/Raman). 3 Kolesov and Geiger, 1998; 4  Mernagh and Liu, 1991; 5 Sendova et al., 2005; 6 Palmeri et al., 2009. Ref. = References.

    13M.L. Frezzotti et al. / Journal of Geochemical Exploration 112 (2012) 1– 20

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    15/21

    Author's personal copy

    pers. comm.). Higher carbonate concentrations can be determinedeasily. These results are of particular interest since they suggest that al-kaline aqueous solutions may represent relevant geological uids.

    Raman spectroscopy is a powerful technique to study the speciationof silica in aqueous  uids at different P –T  and pH  conditions (e.g., Huntet al., 2011; Newton and Manning, 2003, 2008; Zotov and Keppler,2000, 2002). In neutral solutions, SiO2 dissolves predominantly as neu-tral monomers (Si(OH)4

    0) and dimers (Si2O(OH)60) under most crustal

    and upper mantle   P –T   conditions. Si(OH)40(aq) can be identied by a

    Raman band in the 750–800 cm−1 region (Table 1). In alkaline uids,increasing dissociation of monomers and dimers in deprotonated spe-cies (e.g., SiO(OH)3

    −, Si2O2(OH)5−) yields additional Raman bands in

    the 950–1100 cm−1 region, as shown in Fig. 9d.B(OH)3

    0 is the predominant boron species in aqueous uids over awide range of  P –T – pH  conditions. The Raman spectrum of B(OH)3

    0(aq)

    shows a strong band at 877 cm−1 and an additional weaker bandat 495 cm−1 (Table 1; Janda and Heller, 1979; Schmidt et al., 2005).A method of determining the B(OH)3

    0(aq)   concentration in   uid

    Fig. 10. Raman spectra of carbon phases in

    uid inclusions; a) Diamond in a CO2 

    uid inclusion from peridotites, Hawaii; modi

    ed from Frezzotti and Peccerillo (2007). b) Graphitein a CO2  uid inclusion from peridotites, Italy. Excitation light source: Ar ion laser (λ=514.5 nm). G_G-band, or order band; D_D-band, or disorder band. Note that the Ramanwavenumber of the D-band decreases with increasing wavelength of the excitation light source: for example using a He –Ne laser light (λ=632.8 nm), the graphite D-band isexpected at about 1330 cm−1.

     Table 7

    Main Raman vibrations (cm−1) of selected phyllosilicates and inosilicates, both single and double chains.

    Phyllosilicates Main vibrations OH− Ref.

    BiotiteK  2(Mg,Fe

     2+)6-4(Fe 3+,Al,Ti)0-2(Si6-5 Al 2-3O 20)(OH,F)4

    178 549 679 717 3658 2767 3680

    Muscovite

    KAl4(Si6  Al 2O 20)(OH,F)4

    178 216 385 407 639 702 914 1117 3627 [1]

    197 261 754 957PhlogopiteK  2(Mg,Fe

     2+)6 (Si6  Al 2O 20)(OH,F)4

    192 279 331 680 792 1038 3673 3715 [1]372 1096

    ParagoniteNa 2 Al4(Si6  Al 2O 20)(OH)4

    203 413 647 708 1062 3631 [1]218 465 756272

    TalcMg 6 (Si8O 20)(OH)4

    113 295 335 434 678 786 1018 3677 3196 366 793 1055

    Clinochlore(Mg, Fe 2+)5 Al(OH)8(AlSi 3O10)

    104 358 548 679 3477 3605 4198 3647

    3679Chrysotile

    Mg  3Si 2O5(OH)4

    231 345 620 1105 3657 3703 5;389 692 3718 6

    3745Antigorite

    (Mg,Fe 2+) 3(OH)4Si 2O5

    230 375 520 683 1044 3658 3709 5;3687 3729 6

    3774Lizardite

    Mg  3(OH)4Si 2O5

    233 350 510 630 1096 3708 5;388 690 3723 6

    Inosilicates Main vibrations Ref.

    EnstatiteMgSiO 3

    237 343 414 664 1011 [1]382 684

    DiopsideCaMgSi 2O6 

    320 662 1009 7389

    Hornblende(Na,K)0-1Ca 2(Mg, Fe

     2+,Fe 3+,Al)5(Si6-7  Al 2-1O 22)(OH,F) 2

    224 667 1040 [1]

    PargasiteNaCa 2Mg 4 Al 3Si6 O 22(OH) 2

    229 665 1017 8

    OH− = OH stretching vibrations; [1] Raman Spectra Database, Siena (http://www.dst.unisi.it/geouids/raman/ spectrum_frame.htm); 2 Kuebler et al., 2006; 3 Fumagalli et al.,2001; 4  Kleppe et al., 2003; 5  Rinaudo et al., 2003; 6  Auzende et al., 2004; 7 Thompson et al., 2005; 8  Downs, 2006. Ref. = References.

    14   M.L. Frezzotti et al. / Journal of Geochemical Exploration 112 (2012) 1– 20

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    16/21

    Author's personal copy

    inclusions has been presented by  Thomas (2002), with a minimumdetection limit of 0.050 wt.%. Nitrate and phosphate ions have notyet been reported in   uid inclusions, while NaOH(aq)  and LiOH(aq)can be present in some ore-forming  uids (Thomas et al., 2011b)

    6. Identication of mineral phases: a catalog of reference

    Raman spectra

    Fluid inclusions may contain mineral phases, which form by dif-ferent processes, including direct  uid precipitation (daughter min-erals) and reaction of   uid contained within inclusions with thehost mineral (step-daughter minerals) (Fig. 1; Roedder, 1984). Min-

    erals including or included within uid inclusions can be readily iden-tied by comparison of their spectral   ngerprints with referencespectra.

    A catalog of about 140 spectra of minerals which are of interest inuid inclusion research is presented in Tables 2–7, as a supplement tothe web Raman mineral library available at:  http://www.dst.unisi.it/geouids/raman/spectrum_frame.htm. Each table reports mineralname andformula, a list of themain Raman modes observed, and refer-ences. Main vibrations are reported using the ν  notation in scattering

    geometries, where the symmetric stretching vibration (ν1) representsthe strongest Raman mode. Reference spectra catalog also includesselected gas and solute species that were discussed above and listedin Table 1. All measured spectra correspond well to spectra reportedin literature. Relatively pure phases and/or phases contained withinuid inclusions were measured on a Horiba (Jobin Yvon) Labram spec-trometer at the University of Siena, using a water-cooled Ar ion laser(λ=514.5 nm) as the excitation source. Present catalog intends to pro-vide a rst library dedicated to  uid inclusion research.

    6.1. Native elements, halides, oxides and sul des (Table 2)

    Carbon is by far the strongest Raman scatterer and the most stud-ied phase by Raman spectroscopy. In C\O±H uid mixtures, precip-

    itation of C (graphite, or diamond at higher pressures) re

    ects adecrease in   f O2   buffer conditions in the   uid–rock system (e.g.,redox reactions), often induced by a change in P  and/or T . The processhas been studied and modeled in natural and synthetic uid inclu-sions by various authors (e.g., Frezzotti et al., 1994; Huizenga, 2001;Luque et al., 1998, 2009; Sterner and Bodnar 1984; van den Kerkhof et al., 1991).   Fig. 10  reports the spectra of diamond and graphitedetected within  uid inclusions. Diamond is characterized by a verystrong mode at 1332 cm−1 (sp3 bonds; Table 2). Well-crystallizedgraphite shows one intense bands at 1580 cm−1 (sp2 bonds; so-called G-band or order band). In microcrystalline graphite and disor-dered carbon, presence of defects gives rise to an additional band at1350 cm−1 (D-band or disorder band; excitation light source at514 nm), which increases in intensity with increasing disorder,and to an upshift to 1600 cm−1 of the G-band (e.g.,  Wopenka andPasteris, 1993 and references therein).

    The area ratio of the order–disorder bands has been proved to rep-resent a reliable geothermometer in natural graphite (i.e., increasingdisorder at decreasing temperature;  Beyssac et al., 2002; Wopenkaand Pasteris, 1993). However, caution should be used in applyingthe order–disorder geothermometer to graphite contained withinuid inclusions. The crystallinity of graphite precipitated from  uidsdoes not show large variations and it is generally rather high  –  evenat moderate temperatures – unlike what observed in natural graphite(Cesare and Maineri, 1999; Luque et al. 1998, 2009).

    The solubility of uncharged molecules of S in water is appreciable,and S 8

    0 in  uid inclusions (Fig. 9b) has been recognized by the domi-nant broad bands at 462 (S\S stretching) and 220 cm−1 (S\S\Sbending). Additional minor bands may occur at 153, 187, 246, and

    437 cm−1 (Giuliani et al., 2003). Spectra of chlorides (e.g., haliteand sylvite) have not been reported from  uid inclusions. The prob-lem with halides is that they are extremely weak Raman scatterers:one exception is represented by   uorides (Table 2;  Burruss et al.,1992; Rickers et al., 2006). Raman bands of most common oxideand hydroxide minerals are listed in  Table 2. The three polymorphsof TiO2 are also reported, although only rutile has been observed inuid inclusions (Frezzotti et al., 2007).

    6.2. Carbonates (Table 3)

    Carbonates are common phases in  uid inclusions, and a recentexample of Raman identication of multiple carbonates in  uid in-clusions in pegmatites is reported in Thomas et al. (2011a). Raman

    vibrational modes are dependent on the main carbonate groups,modied by the interactions with the bonded mineral lattice. CO3

    2−

    exhibits three main distinct internal vibrational modes over the

    Fig. 11. Comparison of the Raman spectra of calcite, dolomite, and magnesite in the in-terval 0–1600 cm−1. Main CO3

    2− group vibrations are illustrated.   ν1   = Symmetricstretching vibration; ν3 = Antisymmetric stretching vibration;  ν4 = In-plane bendingvibration; T = Translational lattice modes. Calcite, skarn from Vulsini volcanic district,Italy. Dolomite, eclogite from Sulu, China. Magnesite, peridotite from Baldissero, south-ern Italian Alps.

    15M.L. Frezzotti et al. / Journal of Geochemical Exploration 112 (2012) 1– 20

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    17/21

    Author's personal copy

    range 400–1400 cm−1. Generally, strong Raman modes appeararound 1050–1100 cm−1 due to the symmetric stretching vibration(ν1) of the carbonate group, while weaker (around 20 time less in-tense) Raman bands near 700 cm−1 and 1400 cm−1 are due to thein-plane bending mode (ν4) and the antisymmetric stretch (ν3) of CO3, respectively. Lattice modes show Raman shifts below400 cm−1. As shown in Fig. 11, close similarities exist in the Ramanmodes of the CO3 group between different carbonate minerals. Howev-er, signicantdifferences are evident in the positions of their respectivelattice modes over the range 100–350 cm−1 (T in Fig. 11 and Table 3):for example bands of CaCO3 (156 and 284 cm

    −1), CaMg(CO3)2 (176,299 cm−1), and MgCO3 (212 and 329 cm

    −1) are distinct and identi-able without dif culty.

    Raman spectroscopy is well suited to distinguish among the poly-morphs calcite, aragonite and vaterite (Table 3). Calcite has mainRaman modes at 1085 (ν1), 1450 (ν3), and 712 cm

    −1 (ν4). Aragonitehas the main vibrational mode at 1085 cm (ν1), and weak vibrationsat 1463 (ν3) and 704 cm

    −1 (ν4), and an additional very weak bandat 854 cm−1 (ν2). In vaterite, the main vibration mode (ν1) forms adoublet at 1074 and 1090 cm−1. A doublet is also present at 740 and750 cm−1 (ν4). The most intense lattice Raman modes are at 284,206, and 301 cm−1 for calcite, aragonite and vaterite, respectively.Mg-calcite shows a slight upshift of the main stretching band to1087 cm−1 and has a broader band base than pure calcite (Burke,2001). In hydrated (i.e., hydrous and OH-bearing) carbonates, the OHstretching vibrations give rise to additional broad Raman bands located

    between 3000 and 3700 cm−1 (Table 3).

    6.3. Sulfates, phosphates, and borates (Tables 4 and 5)

    Sulfates and phosphates are strong Raman scatterers. The Ramanbands of these minerals are due to the vibrations within SO 4 and PO4tetrahedra. Differences among spectra listed in Tables 4 and 5 resultfrom the nature of metals within the main molecular unit, from thebond strength between the main molecular units and neighboringatoms, and from the different degrees of distortion of the main mo-lecular unit in the mineral lattice (cf., Nasdala et al., 2004, and refer-ences therein).

    In sulfates, the strongest Raman band due to the symmetricstretching vibration (ν1), of SO4 tetrahedra is at about 1000 cm

    −1, at

    lower wavenumbers than that of CO3 groups: 1018 cm−1 for anhy-drite, 994 cm−1 for thenardite, and 1008 cm−1 for gypsum(Table 4). Additional weaker bands over the ranges 400–500 cm−1,

    600–700 cm−1, and 1100–1200 cm−1 are due to the in-plane (ν2)and the out-of-plane (ν4) bending modes, and to the asymmetricstretching of SO4   tetrahedra. The Raman bands of hydrated sulfatesare closely related to those of the sulfate ion in aqueous solution(i.e., 980, 620, and 450 cm−1; Table 1), and show a progressive shifttoward higher wavenumbers with decreasing of the hydration state(cf., hydrous magnesium sulfate list in Table 4; Wang et al., 2006).

    In borates, the distribution of the main Raman bands is mainly de-pendent on the mineral structure and on the type of borate ion (i.e.,boron–oxygen ratio, charge, and hydroxyl groups present); vibration-al modes are observed in the regions: 490–670, 690–800, 820–910,and 950–1040 cm−1 (Table 5). Borates in  uid inclusions have beeninvestigated by   Williams and Taylor (1996), Peretyazhko et al.(2000), Thomas (2002), and Rickers et al. (2006).

    6.4. Silicates (Tables 6 and 7 )

    Silicate minerals are critically important to   uid inclusion re-search: quartz, olivine, pyroxenes and garnet represent very commonhost phases for   uid inclusions, and their Raman bands should befully characterized before analyzing   uid inclusions. In addition,they can be found as daughter mineral phases in  uid inclusions, be-cause of high silica solubility in aqueous   uids at most crustal andupper mantle P –T  conditions (e.g., Manning, 2004).

    Compared to carbonates and sulfates, silicate minerals areweaker Raman scatterers, due to the low polarizability of the Si\O

    bonds. Silicates having different chemical composition or/and struc-ture are discriminated from their spectral features. Fig. 12 comparesmain vibration regions for the different classes of silicates. In ortho-silicates, Raman bands are determined by the vibration modes of theisolated SiO4 tetrahedra, similarly to what observed in sulfates andphosphates. Olivine and garnet show the main stretching modes of SiO4 group in the 800–1050 cm

    −1 region (Table 6). A very good cor-relation has been shown between the wavenumbers of the SiO 4main bands and cation substitution (e.g., Mg/Fe+Mg in olivine)which permits the determination of the chemical composition of these minerals (e.g., Guyot et al. 1986).

    In inosilicates and phyllosilicates, where tetrahedra are to someextent connected, bands generated by the vibration modes of theSi\Ob\Si bonds (Ob  = bridging oxygen) dominate the spectra.

    Spectra of pyroxenes show main Si\O bending and stretchingmodes over the 600–700 and the 900–1050 cm−1 regions, respec-tively (Fig. 12;   Wang et al., 2000). Clinopyroxene (diopside-

    Fig. 12. Raman vibrational mode regions for major silicate classes. Main vibrational regions of borates, phosphates, sulfates, and carbonates are reported for comparison.

    16   M.L. Frezzotti et al. / Journal of Geochemical Exploration 112 (2012) 1– 20

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    18/21

    Author's personal copy

    hedembergite series) and orthopyroxene (enstatite–ferrosilite se-ries) can be easily distinguished by the number of bands in the600–700 cm−1 region: orthopyroxene has two bands, while clino-pyroxene has only a single band (cf.,  Table 7). Most amphibolesand phyllosilicates are very weak Raman scatterers and reconnais-sance within   uid inclusion is often dif cult. As shown in Fig. 13,

    however, the position and the shape of the more intense OH stretch-ing band(s) can be used to distinguish among minerals containinghydroxyl groups.

    Tectosilicate spectra are dominated by vibrations of Si and Oatoms within the framework structures of fully linked tetrahedra(McMillan and Hess, 1990). Main bands, which occur over the range380–530 cm−1 (Fig. 12), are due Si\O\Si symmetric stretchingand bending modes. The Raman frequencies of main modes show arelationship with the size of rings made by tetrahedra (Sharmaet al., 1983): four-membered ring structures, such as feldspars andcoesite, have main modes above 500 cm−1, whereas structures with

    six-membered rings, such as quartz, tridymite, cristobalite, and neph-eline have main modes in the 380–480 cm−1 region (Table 6).

    7. Concluding remarks

    Raman analysis of  uid inclusions permits to qualitatively detect

    or identify gaseous and liquid phases, as well as enclosed (or enclos-ing) minerals. In some cases, quantitative analyses are possible (e.g.,relative mole% in gas mixtures, and solute concentration in aqueousuids). Major advantages of Raman spectroscopy are the minimalsample preparation, and the excellent volume resolution:uid inclu-sionsas smallas thelaser spot size (1–2 μ m) can be precisely locatedand analyzed within double polished thick sections. In addition,Raman is a non-destructive technique, meaning that there is noneed to decrepitate  uid inclusions.

    Fluorescence, that can cover the Raman spectrum, represents themost signicant disadvantage during analysis, and the risk of  uores-cence must be always considered when selecting uid samples to an-alyze (e.g., hydrocarbons). Another signicant disadvantage is theabsence of adequate libraries of reference spectra. This last inconve-

    nience is in part remedied by the compilation of a small spectral li-brary dedicated to  uid inclusion research, presented in our reviewpaper.

    Raman spectroscopy has been used to successfully analyze  uidinclusions with an increasing number of publications through theyears. No other technique can analyze liquid, gas and solid constitu-ents in   uid inclusions. Incorporating this exclusive method withevolving new technologies (e.g., spectral imaging) provides a brightfuture for this  “old” technique in the analysis of geological  uids.

     Acknowledgments

    Present research was in part supported by the PRIN 2008-BYTF98.We acknowledge helpful reviews by R. Thomas and an anonymous

    reviewer of an earlier version of the manuscript. We are grateful tothe Museo di Mineralogia of the University of Rome   “La Sapienza”and to the Museo di Mineralogia of the University of Siena for provid-ing several mineral samples for Raman analysis. Raman facilities inSiena were provided by PNRA, the Italian research program forAntarctica.

    References

    Arakawa, M., Yamamoto, J., Kagi, H., 2007. Selected daughter mineral phases and theirRaman spectral data. Developing micro-Raman mass spectrometry for measuringcarbon isotopic composition of carbon dioxide. Applied Spectroscopy 61, 701–705.

    Auzende, A.L., Daniel, I., Reynard, B., Lemaire, C., Guyot, F., 2004. High-pressure behaviourof serpentine minerals: a Raman spectroscopic study. Physics and Chemistry of Minerals 31, 269–277.

    Azbej, T., Severs, M.J., Rusk, B.G., Bodnar, R.J., 2007. In situ quantitative analysis of indi-vidual H2O–CO2  uid inclusions by laser Raman spectroscopy. Chemical Geology237, 255–263.

    Bakker, R.J., 2004. Raman spectra of uid and crystal mixtures in the system H2O, H2O–NaCl and H2O–MgCl2 at low temperatures: applications to  uid inclusion research.The Canadian Mineralogist 42, 1283–1314.

    Barashkov, M.V., Komyak, A.I., Shashkov, S.N., 2004. Vibrational spectra and structureof potassium alum KAl(SO4)2 ·12[(H2O)x(D2O)1-x]. Journal of Applied Spectroscopy71, 328–333.

    Baumgartner, M., Bakker, R.J., 2009. Raman spectroscopy of pure H2O and NaCl–H2Ocontaining synthetic   uid inclusions in quartz -a study of polarization effects.Mineralogy and Petrology 95, 1–15.

    Baumgartner, M., Bakker, R.J., 2010. Raman spectra of ice and salt hydrates in syntheticuid inclusions. Chemical Geology 275, 58–66.

    Beeskow, B., Rankin, A.H., Murphy, P.J., Treloar, P.J., 2005. Mixed CH4–CO2  uid inclu-sions in quartz from the South Wales Coaleld as suitable natural calibration stan-dards for microthermometry and Raman spectroscopy. Chemical Geology 223,3–15.

    Behrens, H., Roux, J., Neuville, D.R., Siemann, M., 2006. Quantication of dissolved H2Oin silicate glasses using confocal microRaman spectroscopy. Chemical Geology 229,

    96–112.Bény, C., Guilhaumou, N., Touray, J.C., 1982. Native-sulphur-bearing  uid inclusions inthe CO2–H2S–H2O–S system  – microthermometry and Raman microprobe (MOLE)analysis  –  thermochemical interpretations. Chemical Geology 37, 113–127.

    Fig. 13. Comparison of O\H stretching modes for selected phyllosilicates analyzed in

    uid inclusions. Phlogopite, peridotite from Italy. Muscovite, quartzite from Sulu,China. Paragonite, quartzite from Sulu, China. Talc, peridotite from Italy. Clinochlore,peridotite from Ethiopia. In clinochlore OH spectrum, the additional vibration at3565 cm−1 indicates excess of Al, or presence of a humite phase (Frost et al., 2007).

    17M.L. Frezzotti et al. / Journal of Geochemical Exploration 112 (2012) 1– 20

  • 8/18/2019 Raman Spectroscopy for Fluid Inclusion A

    19/21

    Author's personal copy

    Beyssac, O., Rouzaud, J.N., Goffé, B., Brunet, F., Chopin, C., 2002. Graphitization in a high-pressure, low-temperature metamorphic gradient: a Raman microspectroscopy andHRTEM study. Contributions to Mineralogy and Petrology 143, 19–31.

    Boiron, M.C., Moissette, A., Cathelineau, M., Banks, D., Monnin, C., Dubessy, J., 1999.Detailed determination of palaeouid chemistry: an integrated study of sulphate-volatile richbrines and aquo-carbonicuids in quartzveinsfromOuroFino (Brazil).Chemical Geology 154, 179–192.

    Brunsgaard-Hansen, S., Berg, R.W., Stenby, E.H., 2002. How to determine the pressure

    of a methane-containing gas mixture by means of two weak Raman bands, v3and 2v2. Journal of Raman Spectroscopy 33, 160–164.Burke, E.A.J., 1994. Raman microspectrometry of uid inclusions. In: De Vivo, B., Frezzotti,

    M.L. (Eds.), Fluid Inclusions in Minerals: Method and Applications. Short Course of Working Group (IMA)   “Inclusions in Minerals”, Virginia Polytechnic Institute andState University, pp. 25–44.

    Burke, E.A.J., 2001. Raman microspectrometry of  uid inclusions. Lithos 55, 139–158.Burke, E.A.J., Lustenhouwer, W.J., 1987. The application of a multichannel laser Raman

    microprobe (Microdril-28) to the analysis of uid inclusions. Chemical Geology 61,11–17.

    Burruss, R.C., 2003. Raman spectroscopy of uid inclusions. In: Samson, I., Anderson, A.,Marshall, D. (Eds.), Fluid inclusions: analysis and interpretation: MineralogicalAssociation of Canada, short course series, 32, pp. 279–289.

    Burruss, R.C., Ging, T.G., Eppinger, R.G., Samson, I.M., 1992. Laser-excited  uorescenceof rare-earth elements in  uorite: initial observations with a laser Raman micro-probe. Geochimica et Cosmochimica Acta 56, 2713–2723.

    Carey, P.R., 1999. Raman spectroscopy, the sleeping giant in structural biology, awakes. Journal of Biological Chemistry 274, 26625–26628.

    Carey, D.M., Korenowski, G.M., 1998. Measurement of the Raman spectrum of liquid

    water. Journal of Chemical Physics 108, 2669–2675.Carteret, C., Dandeu, A., Moussaoui, S., Muhr, H., Humbert, B., Plasari, E., 2009. Polymor-

    phism studiedby lattice phonon Raman spectroscopy and statisticalmixture analysismethod. Application to calcium carbonate polymorphs during batch crystallization.Crystal Growth & Design 9, 807–812.

    Cesare, B.,Maineri,C., 1999.Fluid-present anatexis of metapelites atEl Joazo (SESpain):constraints from Raman spectroscopy of graphite. Contributions to Mineralogy andPetrology 135, 41–52.

    Chabiron, A., Pironon, J., Massare, D., 2004. Characterization of water in synthetic rhy-olitic glasses and natural melt inclusions by Raman spectroscopy. Contribution toMineralogy and Petrology 146, 485–492.

    Chou, I.M., Pasteris, J.D., Seitz, J.C., 1990. High-density volatiles in the system C–O–H–Nfor the calibration of a laser Raman microprobe. Geochimica et Cosmochimica Acta54, 535–543.

    Clark, R.J.H., Wang, Q., Correia, A., 2007. Can the Raman spectrum of anatase in artworkand archaeology be used for dating purposes? Identication by Raman microscopyof anatase in decorative coatings on Neolithic (Yangshao) pottery from Henan,China. Journal of Archaeological Science 34, 1787–1793.

    Colthup, N.B., Daly, L.H., Wiberly, S.E., 1975. Introduction to Infrared and Raman Spec-troscopy, 2nd edition. Academic Press, New York. 523 pp.

    Davis, A.R., Oliver, B.G., 1972. A vibrational-spectroscopic study of the species presentin the CO2–H2O system. Journal of Solution Chemistry 1, 329–339.

    Derome, D., Cathelineau, M., Fabre, C., Boiron, M.C., Banks, D., Lhomme, T., Cuney, M.,2007. Paleo-uid composition determined from individual   uid inclusions byRaman and LIBS: application to mid-proterozoic evaporitic Na–Ca brines (AlligatorRivers Uranium Field, northern territories Australia). Chemical Geology 237,240–254.

    Dhamelincourt, J.M., Beny, C., Dubessy, J., Poty, B., 1979. Analyse d'inclusions  uides àla microsonde MOLE à effet Raman. Bulletin de Minéralogie 102, 600–610.

    Di Muro, A., Giordano, D., Villemant, B., Montagnac, G., Scaillet, B., Romano, C., 2006.Inuence of composition and thermal history of volcanic glasses on water con-tent as determined by micro-Raman spectrometry. Applied Geochemistry 21,802–812.

    Downs, R.T., 2006. The RUFF Project: an integrated study of the chemistry, crystallog-raphy, Raman and infrared spectroscopy of minerals. Program and Abstracts of the 19th General Meeting of the International Mineralogical Association in Kobe,

     Japan, O03-13.

    Dubessy, J., Audeoud, D., Wilkins, R., Kosztolanyi, C., 1982. The use of the Raman Micro-probe Mole in the determination of the electrolytes dissolved in the aqueous phaseof  uid inclusions. Chemical Geology 37, 137–150.

    Dubessy, J., Geisler, D., Kosztolany, C., Vernet, M., 1983. The determination of sulphatein  uid inclusions using the MOLE Raman microprobe. Application to a Keuperhalite and geochemical consequences. Geochimica et Cosmochimica Acta 47, 1–10.

    Dubessy, J., Pagel, M., Bény, J.-M., Christensen, H., Hickel, B., Kosztolanyi, C., Poty, B.,1988. Radiolysis evidenced by H2–O2- and H2-bearing uid inclusions in three ura-nium deposits. Geochimica Cosmochimica Acta 52, 1155–1167.

    Dubessy, J., Poty, B., Ramboz, C., 1989. Advances in C–O–H–N–S   uid geochemistrybased on micro-Raman spectrometric analysis of uid inclusions. European Journalof Mineralogy 1, 517–534.

    Dubessy, J., Boiron, M.C., Moissette, A., Monnin, C., Stretenskaya, N., 1992. Determina-tion of water, hydrates and pH in  uid inclusions by micro-Raman spectrometry.European Journal of Mineralogy 4, 885–894.

    Dubessy, J., Moissette, A., Bakker, R.J., Frantz, J.D., Zhang, Y.G., 1999. High temperatureRaman spectroscopic studyof H2O–CO2–CH4 mixturesin synthetic uid inclusions:rst insights on molecular interactions and analytical implications. European Journalof Mineralogy 11, 23–32.

    Dubessy, J., Buschaert, S., Lamb, W., Pirono n, J., Thiery, R., 2001. Methane-bearing aque-ous  uid inclusions: Raman analysis, thermodynamic modeling and application topetroleum basins. Chemical Geology 173, 193–205.

    Dubessy, J.,L'Homme,T., Boiron, M.C., Rull, F.,2002. Determination ofchlorinity in aque-ous  uids using Raman spectroscopy of the stretching band of water at room tem-perature: application to  uid inclusions. Applied Spectroscopy 56, 99–106.

    Edwards, H.G.M., Jorge Villar, S.E., Jehlicka, J., Munshi, T., 2005. FT-Raman spectroscopicstudy of calcium-rich and magnesium-rich carbonate minerals. SpectrochimicaActa Part A 61, 2273–2280.

    Fall, A., Tattitch, B., Bodnar, R.J., 2011. Combined microthermometric and Raman spec-troscopic technique to determine the salinity of H2O–CO2–NaCl   uid inclusions

    based on clathrate melting. Geochimica et Cosmochimica Acta 75, 951–964.Fermi, E., 1931. Über den Ramaneffekt des Kohlendioxyds. Zeitschrift für Physik 71,250–259.

    Ferrando, S., Frezzotti, M.L., Orione, P., Conte, R.C., Compagnoni, R., 2010. Late-Alpinerodingitization in the Bellecombe meta-ophiolites (Aosta Valley, Italian WesternAlps): evidence from mineral assemblages and serpentinization-derived H2-bearingbrine. International Geology Review 52, 1220–1243.

    Fraley, P.E., Rao, K.N., Jones, L.H., 1969. High resolution infrared spectra of water vapor:ν1 and  ν3 of H2

    18O. Journal of Molecular Spectroscopy 29, 312–347.Frezzotti, M.L., Peccerillo, A., 2007. Diamond-bearing COHS  uids in the mantle beneath

    Hawaii. Earth and Planetary Science Letters 262, 273–283.Frezzotti, M.L., Burke, E.A.J., De Vivo, B., Stefanini, B., Villa, I.M., 1992. Mantle  uids

    in pyroxenite nodules from Salt Lake Crater (Oahu, Hawaii). European Journal of Mineralogy 4, 1137–1153.

    Frezzotti, M.L., Di Vincenzo, G., Ghezzo, C., Burke, E.A.J., 1994. Evidence of magmatic CO2-rich  uids in peraluminous graphite-bearing leucogranites from Deep Freeze Range(northern Victoria Land, Antarctica). Contributions to Mineralogy and Petrology117, 111–123.

    Frezzotti, M.L., Andersen, T., Neumann, E.R., Simonsen, S.L., 2002. Carbonatite melt-CO2

    uid inclusions in mantle xenoliths fromTenerife, Canary Islands:a story of trapping,immiscibility and  uid–rock interaction in the upper mantle. Lithos 64, 77–96.

    Frezzotti, M.L., Ferrando, S., Dallai, L., Compagnoni, R., 2007. Intermediate alkali-alumino-silicate aqueous solutions released by deeply subducted continentalcrust:   uid evolution in UHP OH-rich Topaz-Kyanite Quartzites from Donghai(Sulu, China). Journal of Petrology 48, 1219–1241.

    Frezzotti, M.L., Ferrando, S., Peccerillo, A., Petrelli, M., Tecce, F., Perucchi, A., 2010.Chlorine-rich metasomatic H2O–CO2  uids in amphibole-bearing peridotites fromInjibara (Lake Tana region, Ethiopian plateau): nature and evolution of v