Effect of Grain Size and Mineral Mixing on Carbonate Absorption Features in the SWIR and TIR Wavelength Regions

Reflectance spectra of carbonate minerals in the shortwave infrared (SWIR) and thermal infrared (TIR) wavelength regions contain a number of diagnostic absorption features. The shape of these features depends on various physical and chemical parameters. To accurately identify carbonate minerals or rocks in pure and mixed form, it is necessary to analyze the effects of the parameters on spectral characteristics. In this study, we analyzed spectral absorption feature characteristics of calcite and dolomite in the SWIR (features at 2.3 and 2.5 μm) and TIR (features at 11.5 and 14 μm) wavelength regions, as a function of grain size and carbonate mineral mixtures. Results showed that varying grain sizes and mineral contents in the sample, influence reflectance values and absorption feature characteristics. Absorption band positions of pure and mixed calcite and dolomite in the SWIR and TIR regions for both features were displaced slightly as observed in previous studies. The band positions of calcite and dolomite varied relative to grain size only in the TIR region. These positions shifted to longer wavelengths for the feature at 11.5 μm and to shorter wavelengths for the feature at 14 μm from fine to coarse grain size. The band positions of calcite-dolomite mixtures in the SWIR and TIR regions were determined by the quantity of calcite and dolomite in the sample. These results can be applied for the identification of pure and mixed calcite and dolomite, as well as estimating the relative abundance of both minerals with different grain size and mineral mixtures in a OPEN ACCESS Remote Sens. 2012, 4 988 synthetic sample or rock. They can also be used as a preliminary proxy for assessing dolomitization patterns in carbonate rocks.


Introduction
Carbonate rocks, which are recognized as an important natural resource for construction materials and the cement industry [1,2], are sedimentary rocks that mostly consist of calcite (CaCO 3 ) and dolomite (CaMg(CO 3 ) 2 ) [1][2][3][4].These carbonate minerals in the form of limestones also have an economic interest in terms of petroleum geology, because their porosity is a potential storage reservoir for oil and natural gas [1,2,5,6].The development of porosity in limestones is affected by the sequence of diagenetic processes, which include dolomitization [1,5,6].The dolomitization process involves the replacement of calcite by dolomite in the rock when magnesium-rich water permeates through limestone [1,6,7].The process may result in an increase in the porosity of the rock by up to 12%, making it suitable as a natural reservoir for oil [5,6,8].The implications of dolomitization for oil exploration [9,10] and the importance of dolomite in making cement make it a favorable mineral to be investigated.
Besides the interest in oil and gas reservoir characterization, there is an interest in looking into calcite-dolomite mineralogy from an 'ore geology' perspective, as carbonates are important pathfinder and alteration minerals associated with calcic skarn deposits, low sulphidation epithermal deposits and porphyry Cu deposits [11,12].In addition, coral reefs are under worldwide decline as a result of climate change [13].Corals are among the species that contribute to carbon sequestration, the capture of carbon dioxide (CO 2 ) from the atmosphere, and act as natural carbon sinks.As they are largely built of carbonate material, more information on its spectral characteristics will help in better species differentiation and in mapping using remote sensing [14,15].
Carbonate minerals have diagnostic absorption features in the shortwave infrared (SWIR) and thermal infrared (TIR) regions due to vibrational processes of the carbonate ions (CO ) [16][17][18][19][20]. In general, carbonate minerals can be distinguished by the presence of two prominent spectral absorption features in the wavelength ranges around 2.50-2.55µm and 2.30-2.35µm in the SWIR [8,18,[20][21][22] and around 13.70-14.04µm and 11. 19-11.40µm in the TIR [17,19,23,24].These features can be used to identify pure and mixed calcite and dolomite in synthetic samples or carbonate rocks, because the absorption band position of calcite is located at a slightly longer wavelength than that of dolomite [8,21].Although the absorption band positions of these particular minerals have been observed by multiple researchers, the precise position of calcite and dolomite absorption bands in the SWIR [8,18,21,22] and TIR [19,[23][24][25][26] spectra has been recorded at different wavelengths.
There are several factors controlling the position of carbonate absorption bands at the atomic level: cation mass, cation radius, cation and anion valences, cation coordinate number, the gap between cation and anion, and site symmetry [27,28].Spectral absorption features of carbonate minerals in the infrared region are also influenced by physical and chemical parameters such as grain size [8,21,29], texture [29], packing or porosity [21], carbonate mineral content [8], and chemical composition (metal ion impurity) [8,18,27].Previous studies of the effect of particle size on carbonate absorption features in the SWIR revealed that differences in grain size change reflectance values and absorption feature depth [8,21,29], but absorption band position, width [8,21], and asymmetry of the features [8] are invariant to grain size.Position, depth, and asymmetry of carbonate absorption feature in the SWIR are displaced by changing the relative amounts of calcite and dolomite composing a sample [8].The presence of iron in the form of Fe 2+ in carbonate minerals reduces the reflectance value [8,27] and depth of the carbonate absorption feature in the SWIR [8].
Although the spectral absorption features of carbonate minerals in the SWIR wavelength region have been extensively studied [8,21,29], they have not been fully explored and observed in the TIR region, particularly with regard to the influences of grain size and carbonate mineral mixtures.In this study, we analyze the spectral absorption feature characteristics of calcite and dolomite in both the SWIR and TIR wavelength regions as a function of grain size and calcite-dolomite mixtures.To accurately identify carbonate minerals in pure and mixed forms, it is necessary to analyze the effects of those parameters that change spectral characteristics.

Sample Preparation
Synthetic samples of calcite and dolomite were prepared by crushing, sieving, mixing and packing.Rock samples of calcite from the Serra de Ficalho, Moura, Portugal and dolomite of chemical grade from the UK were separately pulverised using a jaws-crusher and a steel percussion mortar and pestle.The pulverized samples were dried in an oven overnight at 105 °C to evaporate water, and subsequently sieved into six grain size fractions of pure powdered minerals (<45 μm, 45-90 μm, 90-125 μm, 125-250 μm, 250-500 μm and >500 μm) using a Fritsch sieving machine with stainless steel sieves.Synthetic mixtures of calcite and dolomite were prepared by mixing weight percentages of powdered minerals ranging from pure calcite to pure dolomite with same grain size fractions.The samples used in this mixing process consisted of three different grain size fractions with a combination of some pure fractions, namely <45, 45-125, and 125-500 μm.The pure powdered calcite and dolomite were weighed on the basis of weight percentage of each mineral respectively, using a top-loading precision balance (Mettler PE360) with a reading precision of ±1 g.The total amount of material needed to prepare a synthetic sample was 25 g and the compositions of calcite-dolomite mixtures with different mineral contents for each grain size fractions were 75%-25%, 50%-50%, and 25%-75%.To obtain homogenous mixtures, samples were manually stirred in a porcelain mortar using a spoon-spatula for 5 min.For packing of the samples, the pure and mixed synthetic samples were placed separately in a cylindrical aluminum cup of 5 cm in diameter and 5 mm in depth, and pressed using a weight pressure of 1,090 g for 5 min.

FTIR Reflectance Spectra Measurement
A Bruker Vertex 70 FTIR Spectrometer was used to measure reflectance spectra of the synthetic samples in the SWIR (1.0-3.3 µm) and TIR (5.     by grain size.Coarse grain size fractions absorb more electromagnetic radiation penetrating to the grain surface than fine grain size associated with thickness of the grain or internal optical path of the grain [19,32] according to Lambert-Beer's Law, typical of transparent material [19], and roughness of the grain surface [29]. The presence of carbonate ion in a mineral or rock is characterized by several fundamental vibrational bands in the SWIR [17,18,20,21] and TIR [17,19,20] regions.The two prominent absorption features in SWIR reflectance spectra of calcite and dolomite (Figure 2) are caused by overtones and combination tones of internal vibrational processes of carbonate ions [17,18,20,21].In TIR spectra (Figure 4), the two diagnostic absorption features are generated by two vibrational bending modes of carbonate ions, which are out-of-plane bends for the feature at 11.5 µm and in-plane bends for the feature at 14 µm [17,19,20].
The position of absorption bands for both features in pure powdered calcite and dolomite spectra in the SWIR wavelength region appeared invariant to grain size fraction (Figure 3), confirming results reported by van der Meer [8] and Gaffey [21].Although the center of the absorption band position is not influenced by grain size, the band positions are centered at a slightly different wavelength, when compared to previous studies [8,18,21,22].Gaffey [21] revealed that using different spectroscopic instruments and methods to calculate band position may result in changing the center of the band position.In this study, we used an advance hyperspectral spectrometer with very high spectral resolution for reflectance spectra measurement, so that it could be the reason why our result is slightly different from others.The FWHM and asymmetry of the features at 2.3 and 2.5 µm in the SWIR region varied with grain size fraction (Figures 2 and 3), which is in contrast to what has been published in prior studies [8,21].Here it was stated that the width and asymmetry of absorption features, and the number of absorption bands are not changing or invariant with particle size [8,21].In addition, the FWHM for both absorption features is wider for dolomite than for calcite.
The band position of both features in the TIR wavelength region appeared to be dependent on grain size fraction of the sample (Figure 4).The center of the band position changed slightly at different wavelengths, confirming results of previous studies [19,[23][24][25][26]. Reflectance intensity of band shoulders on the left and right sides of both features at 11.5 and 14 µm decreased considerably more than for the band centers, when the grain size becomes coarser in the sample.Therefore, it influences spectral shapes and absorption feature characteristics.As grain size affects the shape of absorption features [19], it may also affect band position.The absorption band depth, FWHM, and asymmetry of the TIR carbonate features also varied with grain size, but the feature parameters were difficult to analyze given the weak nature of spectral features at larger grain sizes, especially between 125 and 500 µm grain sizes.

Effect of Mineral Mixing
Varying quantity of calcite and dolomite in the sample affects absorption band position in the SWIR and TIR wavelength regions.Work reported by van der Meer [8] for carbonate band positions in SWIR spectra of calcite-dolomite mixtures shows the same trends found in the course of this study.Gaffey [27] also reported that absorption band position in calcite and dolomite spectra displaces with changed chemical composition such as increased Mg content in calcite and Fe content in dolomite.However, the absorption band position displaced also with grain size, especially when the calcite content in the samples increases from 25% to 75% (Figure 7), although the band position did not shift in the SWIR spectra of pure powdered calcite and dolomite (Figure 3).This effect may be caused by the mixing of calcite and dolomite in the sample, which is dominated by one of the two carbonate minerals or "pilling effect" of the fine grains in the sample [18].The FWHM of the 2.5 µm feature in mixed mineral spectra is wider than the 2.3 µm feature (Figure 7), which is the same as the FWHM of pure calcite and dolomite spectral features (Figure 3), most likely due to stronger vibrational absorption at 2.5 µm [18].
The absorption band position of calcite-dolomite mixtures in the TIR was not only dependent on mineral contents, but also on grain size (Figure 9).Absorption features may be affected by grain size and multiple scattering effects [19].The absorption band depth, FWHM, and asymmetry of the features also varied with calcite content (Figure 8).These characteristics show the same pattern as pure calcite or dolomite feature characteristics (Figure 4), when the quantity of calcite-dolomite mixtures composing a sample is dominated by one of those minerals.It was also difficult to analyze due to less pronounced spectral features with increased grain size.

Mineralogical Mapping Applications
Results of this study indicate that absorption band positions of the two prominent carbonate features in the SWIR and TIR wavelength regions are distinctive of carbonate minerals with different grain size and mineral content (Figures 3, 5, 7 and 9).Previous research has revealed that the two prominent carbonate features in the SWIR [8,18,21,22] and TIR [17,19,23] regions are essential to distinguish between calcite and dolomite, as they contain the strongest vibrational absorption features of carbonate ions.Therefore, these features are useful for identifying pure and mixed calcite and dolomite as well as estimating the relative abundances of both minerals in a synthetic sample or rock.Previous research has demonstrated that distinctive spectral reflectance characteristics in the near infrared through laboratory experiment and remote sensing imagery can assist in identification of carbonate minerals and dolomitization patterns [9,10,43].Van der Meer [8] stated that estimation of the calcite-dolomite ratio from spectra could be done using diagnostic absorption features around 2.30-2.34µm, of which the exact position is dependent on the relative amounts of calcite and dolomite.Consequently, knowledge of these precise absorption band positions may be applied to analysis of hyperspectral remote sensing imagery for mapping carbonate minerals and dolomitization patterns in limestones.

Conclusions
Absorption feature characteristics and reflectance values of pure and mixed calcite and dolomite spectra in the SWIR and TIR wavelength regions were influenced by grain size and carbonate mineral content.Absorption band positions of both features in these wavelength regions were displaced slightly as observed in previous studies.The positions of SWIR calcite features at 2.340 µm and 2.537 µm and dolomite features at 2.323 µm and 2.515 µm were invariant to grain size.The FWHM and asymmetry of the calcite and dolomite spectral features at 2.3 and 2.5 µm in the SWIR region varied with grain size fraction, in contrast to what has been published in prior studies that the width and asymmetry of absorption features, and the number of absorption bands are not changing or invariant with particle size.In the TIR region, the positions depended on grain size and shifted to longer wavelength for the feature at 11.5 μm and to shorter wavelength for the feature at 14 μm from fine to coarse grain size fractions.The positions of both features of calcite-dolomite mixtures in the SWIR and TIR wavelength regions were determined by the calcite or dolomite content composing the sample.The positions of the features in the SWIR region were centered within the wavelength range of 2.323-2.340μm and 2.515-2.537μm.In the TIR region, however, the absorption band positions of the calcite-dolomite mixtures not only depended on mineral content, but also on grain size in the sample.Knowledge of these detailed feature characteristics can increase the accuracy of spectral identifications of pure and mixed calcite and dolomite with different grain size and carbonate mineral content, which is of particular use for assessments of dolomitization patterns in limestones.
This study has contributed to understanding the effects of grain size and mineral mixtures on the spectral absorption feature characteristics in the SWIR and TIR wavelength regions, which will help in improving mineral identification on the earth's surface using hyperspectral remote sensing.
Figure a func shown Figure size fr (a) fea Figure 8 each grain s both feature decreases w calcite-dolom ange of 125