Seasonal sea ice strongly modulates global climate and ecology through its effect on ocean albedo, ocean-atmosphere heat transfer and transfer of gases and particles (including nutrients) between the atmosphere and ocean. Small changes in the atmosphere/ocean/ice composition ratio may lead to significant changes in the nature of the sea-ice cover and its ecology. Investigations of mineral inclusions in sea ice are expected to provide insights which may help in climate studies. The presence of minerals in Antarctic ice will also provide significant information on past atmospheric compositions. The snowfall composition, buried over time, contains records of the climatic history in the ice, which might successfully be used in climate and environmental studies (Reference JouzelJouzel and others, 1989, Reference Jouzel, Lorius and Raynaud2006; Reference Legrand and MayewskiLegrand and Mayewski, 1997; Reference PetitPetit and others, 1999; Reference JouzelJouzel, 2003; EPICA Communiy Members, 2004)
The aim of this work is to report the occurrence of inclusions of the recently discovered mineral meridianiite in sea ice and Antarctic continental ice.
The MgSO4 ·11H2O crystal structure is triclinic with space group P-1 (no. 2). MgSO4 ·11H2O had previously been described as MgSO4 ·12H2O by Reference FritzscheFritzsche (1837). Through weight loss via dehydration, Fritzsche attempted to estimate the water content of salt, which led to an error by one water molecule. The MgSO4 ·11H2O crystal is colorless and needle-shaped with the parameters:
formula weight, FW = 318.55;
crystal size = 0.54 × 0.24 × 0.18 mm3;
length of the unit cell dimension along the crystallographic x axis, a = 6.725 48(7) Å;
length of the unit cell dimension along the crystallographic y axis, b = 6.779(14) Å;
length of the unit cell dimension along the crystallographic z axis, c = 17.290(5) Å;
angle between the y and z crystallographic axes, α = 88.255(1)°;
angle between the x and z crystallographic axes, β = 89.478(2)°;
angle between the x and y crystallographic axes, γ = 62.598(1)°;
unit cell volume, V = 699.54(3) Å3;
number of formula units in the crystallographic unit cell, Z = 2;
calculated density, D calc = 1.512 g cm−3; and
absorption coefficient, μ = 0.343 mm−1 (Reference Genceli, Lutz, Spek and WitkampGenceli and others, 2007).
On Earth, epsomite (MgSO4·7H2O) is present in, for example, crusts and efflorescences in coal or metal mines, limestone caves, oxidized zones of sulfide ore deposits, salt lakes and playas. In the laboratory, as well as in nature, MgSO4·11H2O crystallizes from a suspension around its eutectic point, i.e. concentrations between 17.3 and 21.4 wt% MgSO4 and temperatures between −3.9 and 1.8°C (Reference Genceli, Lutz, Spek and WitkampGenceli and others, 2007). If the temperature of the magnesium sulfate reservoir lies above 1.8°C, epsomite formation occurs instead of meridianiite.
Previously, Antarctic ice impurities were analysed using scanning electron microscopy and energy dispersive spectrometry (SEM/EDS) (Reference Baker and CullenBaker and Cullen, 2003; Reference Barnes and WolffBarnes and Wolff, 2004). Later, Reference Ohno, Igarashi and HondohOhno and other (2006) used micro-Raman spectroscopy to examine micro-inclusions in polar ice from Dome Fuji in order to determine the salt composition. It was found that the inclusions mainly consisted of sulfate salts with small amounts of other soluble salts and insoluble dust (Reference Ohno, Igarashi and HondohOhno and others, 2005, Reference Ohno, Igarashi and Hondoh2006).
We have now made detailed studies of inclusions in sea ice and in Antarctic core ice by micro-Raman spectroscopy.
Sea-ice samples were collected from Lake Saroma (Saromako), the third largest lake in Japan, located on the north-eastern shore of Hokkaido island. The lake is a semi-enclosed embayment connected with two openings to the Sea of Okhotsk. Most of the lake surface is covered with sea ice during winter (Reference Kawamura, Shirasawa, Ishikawa, Takatsuka, Daibou and LeppärantaKawamura and others, 2004).
The sea-ice samples were selected from a core sampled at 44°07’ N, 143°57’ E in March 2006: core No. 06.03.07-St.4. The average sampling-day temperatures at the surface and at sea level were −4.8 and −0.7°C, respectively. In this study, only the surface (top 1 cm) of the sea-ice core was analysed, as this layer was in direct contact with the air and was the coldest section. The lowest air and ice surface temperatures during March 2006 were −16.4 and −20.6°C, respectively, the average day temperature recorded was −3.9°C and the average ice surface temperature was −2.8°C (personal communication from T. Kawamura, 2007). The ice temperature was measured in situ immediately after extraction of the core, using calibrated Technol Seven, D617 thermistor probes (precision ±0.1°C) inserted into small, drilled holes with the exact diameter of the probe. After collection, the samples were stored at −15°C, at the Institute of Low Temperature Science, Hokkaido University.
During ice formation, the sea-water temperature near the ice decreases gradually while the remaining brine becomes increasingly saline. Since ice generally contains no salt in the ice body, the dissolved salts are rejected and remain in the adjacent sea water. However, a portion of the brine can become mechanically trapped in the ice matrix (Reference Wakatsuchi and KawamuraWakatsuchi and Kawamura, 1987). As cooling continues, different solid salts will crystallize from the entrapped brine at different temperatures (Reference Weeks and AckleyWeeks and Ackley, 1982). Crystallization of the salts from sea-ice brine at various temperatures was deduced from changes in brine composition and stability ranges of individual salts in the corresponding pure saltwater systems (Reference SinhaSinha, 1977). The solubility of one salt affects the solubility of the other.
As mentioned, the pure MgSO4-H2O aqueous system crystallizes in the MgSO4·11H2O form below +1.8°C and has a eutectic point at −3.9°C and 17.3 wt% MgSO4 concentration (Reference Genceli, Lutz, Spek and WitkampGenceli and others, 2007). Presence of impurities shifts the eutectic point to lower temperatures. Therefore, the formation path of MgSO4·11H2O from sea brine composition is not completely known, but it has been established here that inclusions of this salt were indeed present in the samples which were stored and measured at −15°C. Given that the lowest ice exposure air temperature during March 2006 was −16.4°C and the highest ice temperature was 0.2°C, it is reasonable to assume that the samples from the surface sea ice were kept intact during sampling and analysis.
Antarctic core ice
The ice core used in this study was collected from Dome Fuji station, East Antarctica (77°19’ S, 39°42′ E; 3810 m a.s.l.), near the summit of the east Dronning Maud Land plateau. The average snow temperature at 10 m depth is −58°C. For further information about the sampling, see Dome-F Deep Coring Group (1998). After collection, the samples were stored at −50°C, at the Institute of Low Temperature Science, Hokkaido University. In this work, samples from 362 m depth were analysed: core No. 03-124.
Preparation of ice samples
Preparation of synthetic MgSO4·11H2O has been extensively described by Reference Genceli, Lutz, Spek and WitkampGenceli and others (2007).
Samples of the top (surface) section of the sea ice and 362 m depth Antarctic ice samples were prepared by cutting with a bandsaw into 10 ×10 ×3 mm3 pieces. The upper and lower sides of the samples were then flattened (polished) with a microtome in a clean cold chamber at −15°C.
Determination of the chemical composition of the inclusions in both ice samples by direct methods was difficult to achieve due to the extremely small size of the crystals (a few μm) and their low material content and high solubilities in water. Therefore, cyro-micro-Raman spectroscopy was employed as the most powerful technique to identify the inclusions inside the ice samples. Spectra of synthetic MgSO4·11H2O were compared with the observed spectra of the inclusions.
A Jobin-Yvon T64000 triple monochromator equipped with a charge-coupled device (CCD) detector was used to obtain backscattered micro-Raman spectra. Prepared ice samples were placed into the cold chamber on an x-y translation stage of the microscope. Cooling was achieved indirectly by N2 gas circulation (at 1 bar) through the chamber, keeping the sample temperature at −15 ± 0.5°C (at 5% relative humidity) for sea ice and at −50°C (at <5% relative humidity) for Antarctic core ice. A laser beam (514.5 nm, 100 mW) was focused to a diameter of ∼1 μm on the specimen using a long-working-distance objective lens with 6 mm focal length (Mitutoyo, M Plan Apo 100×). The absolute frequency of the monochromator was calibrated with neon emission lines. The spectral resolution was ∼1 cm−1 in the range 100–4000 cm−1 with a detection time of 60 s. For this analysis, four pieces of sea ice and Antarctic ice-core samples were prepared. We measured 30–40 micro-inclusions per sample.
The inclusions in sea ice were typically 5–20 μm, while those in Antarctic core ice were typically 1–5 μm in diameter (Fig. 1). Most of the inclusions were embedded within the ice grains. The MgSO4·11H2O inclusions were identified by comparing the observed micro-Raman spectra with those of synthetic MgSO4·11H2O samples. Other inclusions of, for example, sulfate, nitrate and acid salts were also detected, but discussion of these is beyond the scope of this paper.
Micro-Raman spectra for low-frequency (100–2000 cm−1) and high-frequency (2000–4000 cm−1) ranges of synthetic MgSO4·11H2O, mirabilite and typical meridianiite inclusions in sea ice and Antarctic core ice are presented in Figures 2 and 3. As the inclusions were located in the ice matrix, the dominant ice bands could not be removed from the spectra. Thus, ice spectra are also presented in Figures 2 and 3 for comparison. Due to the small sizes of Antarctic core ice inclusions, it was not possible to measure the complete spectrum from a single inclusion. Therefore the higher-frequency range spectrum (for incorporated water in the lattice and for intracrystalline water bands) of the inclusions introduced in Figure 2 could not be collected and given in Figure 3.
The frequencies, intensities and assignment of the majority of the bands in the micro-Raman spectra of synthetic MgSO4·11H2O, sea-ice and Antarctic core-ice inclusions, ice and mirabilite are presented in Table 1 for comparison. The band identifications are based on literature data.
Micro-Raman spectroscopy of sea-ice inclusions
About 5% of the inclusions detected in sea ice were meridianiite. Significant similarities between the micro Raman spectra of the sea-ice inclusions and the synthetic MgSO4·11H2O crystals are observed in the SO4 spectral region. Bands in the synthetic MgSO4·11H2O crystals ν 1 mode (very strong symmetric stretching vibration) at 990 cm−1, ν 2 mode at 444 cm−1 and also a weak peak at 457 cm−1, ν 3 mode at 1116 and 1070 cm−1 and ν 4 mode at 619 cm−1 have also been observed in the sea-ice inclusions. Due to the water involvement in the hydrogen bonding, the stretching modes of crystal water in the lattice are manifested as a complex band in the 3000–3600 cm−1 region, with a maximum at 3393 cm−1 for synthetic MgSO4·11H2O crystals and 3400 cm−1 for sea-ice inclusions with several similar shoulders. The water bendings are observed at 1673 cm−1 both for synthetic MgSO4·11H2O crystals and sea-ice inclusions (Table 1). Note that in the micro-Raman spectra of sea-ice inclusion, extra water bands due to the presence of ice surrounding the inclusions are detected at 3265, 3141 and 2188 cm−1, which have also been observed for sea ice in close frequencies. The band of the O-H…O (sulfate) vibration located at 232 cm−1 for synthetic MgSO4·11H2O is not detectable in sea-ice inclusions as it is hidden by the presence of a strong ice-stretching band at 215 cm−1 with a resolved mode at 294 cm−1. The O–Mg–O band for sea-ice inclusions seems to be located at 188 cm−1, analogous to that for synthetic MgSO4·11H2O crystals.
In summary, the excellent micro-Raman spectra vibrational harmony between the synthetic MgSO4·11H2O and the inclusions of sea-ice samples proves the natural occurrence of meridianiite as a mineral in sea ice. The daughter minerals in inclusions of sea ice may have been formed through freezing of sea ice followed by concentration of brine captured in the ice and the meridianiite crystallization out of this brine. Reference Peterson, Nelson, Madu and ShurvellPeterson and others (2007) have also reported the natural occurrence of meridianiite, located in a tree trunk on the surface of a frozen pond in central British Columbia, Canada. Meridianiite has now been recognized as a valid mineral species by the International Mineralogical Association (IMA) Commission, where powder diffraction and physical properties data have been deposited. The crystal structure data of meridianiite have been presented by Reference Peterson and WangPeterson and Wang (2006), Reference Genceli, Lutz, Spek and WitkampGenceli and others (2007), and Reference Peterson, Nelson, Madu and ShurvellPeterson and others (2007).
Micro-Raman spectroscopy of Antarctic ice inclusions
Magnesium sulfate inclusions in Holocene ice-core samples from Dome Fuji were reported by Reference Ohno, Igarashi and HondohOhno and others (2005). We have now investigated these Antarctic ice-core samples using micro-Raman spectroscopy. About 3% of the inclusions detected in the Antarctic core ice were meridianiite. As seen in the typical spectra presented in Figure 2, the most significant peak is the SO4 2− associated symmetric stretching band ν 1 registered at 990 cm−1, and the sulfate modes 2 at 446 cm−1, ν 3 at 1117 cm−1 and ν 4 at 621 cm−1 are all in accordance with the literature data (Table 1). The 3 sulfate mode around 1070 cm−1 is not clear in Antarctic core-ice inclusions and the assignment of this band was not possible. As with the sea-ice inclusions, the strong ice-stretching band at 215 cm−1 with a resolved mode at 286 cm−1 hides the band of O-H…O (sulfate) vibration located at 232 cm−1.
The micro-Raman spectra for low vibrations of Antarctic core-ice inclusions compare well with the synthetic MgSO4·11H2O. This suggests that micro-inclusions of MgSO4 salt preserved inside Antarctic core ice almost certainly contain meridianiite. Meridianiite possibly formed as inclusions in Antarctic core ice by the reaction of acid-gas particles (H2SO4 and HNO3) with sea-salt aerosols or terrestrial dusts during their transport through the atmosphere as well as within the snowpack (Reference Röthlisberger, Hutterli, Sommer, Wolff and MulvaneyRothlisberger and others, 2000; Reference Ohno, Igarashi and HondohOhno and others, 2006).
Micro-Raman spectroscopy of mirabilite
By scanning sea-ice and Antarctic core-ice samples by micro-Raman spectroscopy, mirabilite (Na2SO4·10H2O) was found to be the most common mineral encountered in the inclusions. Figures 2 and 3 and Table 1 present the spectrum of synthetic mirabilite and the band assignments. In order to distinguish between inclusions of MgSO4·11H2O (denoted as MgSO4·12H2O by Reference Ohno, Igarashi and HondohOhno and others, 2005) and Na2SO4·10H2O in Holocene ice-core samples from Dome Fuji, Reference Ohno, Igarashi and HondohOhno and others (2005) measured the 300–1500 cm−1 wavelength range and used the symmetric stretching SO4 2− band (∼990 cm−1) as an indication of the inclusion composition. They distinguished the two salts with a slight shift (1 cm−1) of the main peak: 990 cm−1 for mirabilite and 989 cm−1 for MgSO4·11H2O. Since the band in the region of 990 cm−1 belongs to the symmetric ν 1 stretching mode of the sulfate ion contained in both salts, we consider it desirable to measure more bands if we are to distinguish between meridianiite and mirabilite inclusions.
In this work, mirabilite micro-Raman spectra were collected under atmospheric pressure in a cooling chamber at −15°C, providing the same conditions applied to meridianiite. The spectra of meridianiite and mirabilite harmonize at the SO4 2− symmetric stretching band at 990 cm−1. We can conclude that it is indeed useful to measure a wider spectrum range to distinguish between sulfate salts. Examples of discriminating bands of mirabilite are the υ2 mode at 456 cm−1, the υ3 mode at 1112 and 1127 cm−1 and the υ4 mode at 613 cm−1 (Murugan and others, 2000). Mirabilite’s O–H-band located at wavenumbers between 3100 and 3700 cm−1 has a maximum at 3441 cm−1 due to stretching water mode incorporated in the lattice (Reference Xu and SchweigerXu and Schweiger, 1999). Since detailed mirabilite Raman spectrum indexing is not readily available in the literature, not all peaks have been elaborated on in this work. However, by using our findings, it can be concluded that meridianiite has been identified in sea ice and Antarctic core ice.
The mineral meridianiite was found in sea-ice inclusions from Lake Saroma. The excellent micro-Raman spectra match between the Japanese sea-ice daughter crystals and synthetic MgSO4·11H2O crystals is presented as evidence of its existence.
Antarctic core-ice inclusions from Dome Fuji match well with the micro-Raman spectra for low vibrations of synthetic MgSO4·11H2O, pointing to the existence of meridianiite in Antarctic continental ice. The extremely small size of the inclusions (1–5 μm) was the greatest obstacle to measuring the whole micro-Raman spectrum from a single inclusion.
The Raman spectrum of MgSO4·11H2O (meridianiite) showed significant differences from that of Na2SO4·10H2O (mirabilite).
We thank E. Burke, Chairman of IMA Commission on New Minerals, for support and help during this work.