Interpretation of cathodoluminescence patterns by the example of fluorapatites from Kondyor koswites (Aldan shield
are mounted on scanning electron microscopes and electron microprobes is usually
limited to visualization of zoning of mineral individuals. Attempts to make use
of minerals’ ability to emit in genetic reconstruction are few. The present study
provides an example of a methodical approach that yields a new understanding
of koswites formation within the Kondyor massif through analyzing fluorapatite
Cathodoluminescence (CL) is a recognized visualization technique for trace elements distribution in crystals as well as for further structural and chemical assessment of the areas of interest. Over recent years, the method became less widely used, mainly due to the complexity of pattern interpretation. As reported by many researchers , single crystal spectral intensities depend on a number of factors, not just the impurity concentration. Variations in fluorescence brightness within a single crystal can be explained by heterogeneous distribution of luminescent centers and optical laws of light propagation in a mineral grain .
Moreover, the direct interpretation of CL patterns is complicated by the dual nature of trace elements entry into the composition of a mineral. Final concentration of an impurity is conditioned by not only its concentration in the medium but also the capacity of the mineral, which, in turn, depends on quite a number of factors (sorption ability, diffusion, coating, etc.). Thus, it would be incorrect to claim that local variations in rare earth elements (REEs) within the sample only correspond to their concentration changes in the medium during the process of mineral formation (though such a connection should not be excluded).
An apatite sample from koswite-like magnetite-apatite-clinopyroxene metasomatites that cut dunitic rocks in the south-western part of the Kondyor massif has been chosen for the study. This particular part of the massif is very rich of magmatic and metasomatic formations of different ages. Veins and irregularly-shaped bodies of subalkaline rocks, nepheline syenites and their pegmatite varieties, natrolite-richterite and aegirine-cancrinite metasomatites as well as dykes of koswites and koswite-like rocks from 1 cm to 30–40 m thick were observed there. Koswite-like metasomatites follow the fracture pattern of host dunites and form stockworks. Late igneous intrusions had a thermal and hydrothermal effect on all the complexes, including vein-rocks. Differences in growth conditions among koswites and koswite-like metasomatites along with superimposed alterations have led to mineral and texture diversity of these rocks. Their granulometric composition may vary even within several centimeters. It comes as no surprise that there is a lack of scientific consensus on the age of Kondyor koswites. According to some authors [3,4], the results of radiometric dating of Kondyor koswites, assumed to prove their palaeozoic or mesozoic age, are more likely to indicate periods of their postcrystalization. We think that the hypothesis of rock “rejuvenation” may be confirmed or refuted by studying the ontogenesis of mineral individuals.
Sample no. 3-1, which was examined, consists mainly of clinopyroxene (diopside-augite), fluorapatite (20%), and magnetite (5%). It also contains sparse xenomorphic grains of chalcopyrite completely replaced with chrysocolla. As for apatite, it forms prismatic grains from 0.2 mm to 1 cm in size.
Chemical compositions were studied in the Laboratory of Mineral Analysis of the IGEM RAS by the JEOL JXA-8200 electron probe microanalyzer (Japan) with one energy-dispersive and five wavelength-dispersive spectrometers. Analyses were done at accelerating voltage of 20 kV, Faraday cylinder current of 20 nA, and beam diameter of 3 µm. Exposure times were as follows: 10 sec for the predominant elements (Ca, P); 20 sec for Si, Fe, Al, Ba, Sr, Mg, Cl, Na, and S; 30 sec for F; 60 sec for Ce, Th, Nd, and La; 100 sec - for Sm, Y, Pr, Mn, and Dy. To impart electrical conductivity to thin sections, they were sputter coated with thin carbon films. Manufacturer’s own software was used to determine matrix correction factors (ZAF factors). The following synthetic compounds and minerals were chosen as standards: F, Ca, P (fluorapatite), Si, Al (anorthite), Fe (schorlomite), Sm (SmP5O14), Y (Y2O3), Mn (spessartite), Ba (barite), Pr (PrP5O14), Ce (CeP5O14), Sr (SrSO4), Th (ThO2), Nd (NdP5O14), La (LaF3), Mg (olivine), Cl (Cu2Cl(OH)3), Na (jadeite), and S (SrSO4). Color cathodoluminescence images and characteristic spectra in particular spots were obtained with the Cameca MS-46 microanalyzer (France) supplemented by an original optical detector adapted from the Videoscan-285 high resolution camera (Russian Federation) (Table 1) and USB 2000+VIS-NIR-ES spectrometer (Ocean Optics, USA) (Table 2). Unlike serial manufactured detectors this technical equipment provides true color cathodoluminescence images for further mathematical processing. For CL registration the beam was scanned in raster mode at accelerating voltage of 20 kV, current of 20 nA, and exposure time of 239.7 sec. Optical spectrum was recorded under similar conditions except for the exposure, which was of 20 sec. Colorimetric analysis of CL images was done using the LumDesign-С 04.005 software package (Glaukon, Russian Federation).
RESULTS AND DISCUSSION
Digital CL images (CL patterns) were obtained for thin sections of 23 fluorapatite single crystals (Fig.1). All grains demonstrated weak blue-grey and yellow emission. Most crystals exhibited color zoning and uneven brightness, which made them look irregularly spotted.
The maximum number of zones that could be identified was three: 1) central zone, notable for the brightest yellow luminescence; 2) middle zone, predominantly dull blue; and 3) peripheral zone of somewhat intermediate luminescence, which could be further subdivided to blue and yellow areas.
Spotted appearance of CL images is due to local overlap of optical defects. The bright mixed-type luminescence of the overlaps shows a good agreement with theoretical models of light propagation in a mineral grain . The large number of optical defects within the crystals is confirmed by a redshift of the grain boundary reflex, which is an effect of Rayleigh scattering (longer wavelengths have an advantage of being less scattered as they repeatedly pass through light-scattering regions of the mineral). In order to reduce the influence of anomalous radiance of scattering regions on pattern interpretation, RGB shift analysis was performed (Fig.1). It allowed us to distinguish between the three zones of a monocrystal on the basis of primary colors proportions. Employing a CCD-detector with lineal gamma, one can expect its response for each color channel to correlate with the impurity content of the corresponding crystal zone.
Most of thin sections appeared unsuitable for further analysis due to their incompleteness (absent central zone, see Fig.1) or oblique cutting (Fig.1). Therefore, we selected regular hexagonal sections with no boundary reflexes ensuring that the crystals were cut perpendicularly to their main axes and close to the centers of symmetry (Fig.2). As can be seen from the colorimetric map in Fig.3, central and peripheral zones of one of the grains selected for further analysis emit medium- and long-wave visible radiation of comparable intensity.
A chromacity diagram (Fig.4) allows simplified description of the spectral composition of grain cathodoluminescence. In our case, there are three statistical fields (consistent with the three zones of the grains, see Fig.3) but only two resultants, which means that the full diversity of CL colors in fluorapatite under study is due to just two sets of impurities combined in different proportions.
RELATIONSHIP BETWEEN CATHODOLUMINESCENCE AND IMPURITY SITES
In order to identify CL colors with particular radiating sites, a continuous optical spectrum (Fig.5) was recorded at three points representing the three zones defined above (A, B, and C Fig.3).
It has been found that longer-wavelength luminescence (A and C in Figure 3) is accounted for by additional lines in the red part of the spectrum (595–601, 567–570, 650, and 700 nm). In accordance with principles of additive color mixing, these lines together with blue luminescence (350–500 nm), which prevails across the whole grain and does not vary significantly in terms of intensity, cause a richer bright yellow color of the resultant emission (blue + orange-red = yellow). Zone B in Fig.3 produces short-wave luminescence only and is, therefore, blue. Judging from literature , persistent short-wave emission (350–500 nm) may belong to Ce3+ cations. Indeed, micromapping of the thin sections (Fig.6) revealed a near uniform distribution of Ce3+ throughout the samples with local maxima in zones of short-wave luminescence, which allowed us to identify one with the other with a high degree of certainty.
To explain the nature of long-wave CL, we determined concentrations of possible impurities, namely: Al, Fe, Mn, Sr, Ba, Pr, Sm, Dy, Eu, Y, Th, U, Cl, Si, S [6, 7, 8, 9, 10]. Readings were collected in 3 µm steps along two intersecting scan lines running through the center of the grain (Fig.3). For all these points, brightness was also determined on the basis of backscattered electron (BSE) image gray-level scale. BSE-brightness, though not part of SI, is proportional to the average atomic mass and, therefore, can well be used in correlation analysis.
Pairwise comparisons were then drawn between those impurities, concentrations of which were above the lower detection limit, and a number of physical indices. The complete list is presented in Table 3.
By analogy with zones of blue luminescence, spectral maxima in yellow zones should belong to Mn2+, Pr3+, and/or Sm3+ [1, 6, 7, 8]. In this case, however, correlation analysis showed no relationship between the intensity of long-wave emission and concentrations of these particular elements (Table 3). Moreover, long-wave luminescence was detected in areas of pure fluorapatite (consists almost entirely of essential elements, such as P and Ca, and lacks impurities), which is, of course, paradoxical.
Intending not to leave some exotic activator undetected, we superimposed BSE and CL images of the same thin sections (Fig.7). Despite apparent similarities, the BSE-derived crystal zoning demonstrated a weak correlation with long-wave luminescence, thus, confirming the relationship between the latter and pure fluorapatite.
Principles of luminescence  allow to consider background levels of Sm3+, Pr3+ and Mn2+ sufficient for formation of highly radiant areas. At this rate, the absence of emission lines at long wavelengths in the middle zone of the crystal (see B in Fig.3) is likely to be due to different structural positions of these elements rather than low concentrations. Moreover, the phenomenon may be associated with a different physical state of the impurity site. For further insight into causes of CL co-activation/inactivation we decided to analyze those impurities that demonstrated strong negative correlation with long-wave portion of visible emission, i.e. Si and S, both common structural elements of apatite supergroup minerals (Table 3).
In Fig.8, there is a diagram of Na/S/Si atomic ratios in ‘complete’ grains. Also plotted is the average relative intensity of long-wave luminescence for each zone. Although S/Si ratio always tends to 1, we see no intermediate compositions, so far as their absolute content is concerned, which, perhaps, indicates a prohibitive/permissive role of these impurities on luminescence. Supposedly, Si and S, taken together, determine the contrast of CL zoning, but have no influence on the formation of a particular type of luminescence.
It is clear from the results of statistical analysis that the actual activator of long-wave CL is the proportion between additional structural elements and charge compensators engaged in heterovalent isomorphism: 2Ca2+→TR3++Me+. The univalent metal ion, which is encoded as Me+, is most likely to be Na+. This conclusion is drawn from the fact that Na+ is abundant, while the level of other univalent metal ions is below the lower detection limit. Statistical fields of Na+ absolute concentrations match those of Si, however, show an evident diversity of S/Na ratios as we go from the long-wave zone to the short-wave one (see Fig.8).
Now, what does TR3+ stand for? Of all impurities capable of long-wave emission, i.e. Sm3+, Pr3+, and Mn2+ , the absolute level of Sm3+ is the highest (see Table 3). The link between CL and charge compensator in the reaction of substitution of Ca2+ ions with trivalent REEs together with the presence of three luminescence peaks (595–601, 650, and 700 nm) strongly suggests Sm3+ to be responsible for the yellow part of the CL spectrum.
Taking into account the uniformity of Sm3+ distribution over the three zones of the crystal as well as the results of statistical analysis, we consider that there is a certain set of conditions necessary for activation of long-wave cathodoluminescence that can be formalized as follows: yellow CL occurs if Si:S:Na ratio is close to 1:1:1 and is very dependent on Na+ concentration. Even minor sodium deficiency prevents the emission in most cases with the exception of those sites that are bound to Ce3+ (and perhaps Dy3+, which may be responsible for the unidentified peak at 567–570 nm).
BETWEEN THE RESULTANT CL PATTERN
AND EVOLUTION OF GROWTH CONDITIONS
OF MONOCRYSTALLINE FLUORAPATITE
Cases of simultaneous presence of silicon, sodium, sulfur, and REEs in fluorapatite have been well described [9, 10, 11]. The feasibility of their concerted removal has been also shown . According to the experiment, self-cleaning of monocrystalline fluorapatite goes in water medium under 900єC and 1000 MPa and is associated with the formation of certain phases such as elestadite, britolite, anhydrite, and xenotime. Another study describes an alternative scheme, in which xenotime forms a separate phase as a result of charge decompensation due to the inability of fluorapatite (with an altered S/Si/Na ratio) to append Y3+ or TR3+ .
A similar process is very likely to take place in zones with long-wave emission lines. If we compare the composition of fluorapatite grains in blue and yellow CL zones, we will find that microimpurities contained in yellow zones can decompose into elestadite and tenardite molecules with no significant remainder of S, Si, or Na, while short-wave luminescence zones exhibit no particular regularities in microimpurity content (Fig.9). If we exclude a coincidence, we might draw an analogy with experimental data (keeping in mind, however, that independent partitions of elestadite, monazite, and tenardite have not yet been found).
Nevertheless, there is an indirect indication of monazite presence in the system under study: the three maxima of emission intensity seen in yellow zones of the crystals (see Fig.4, C) resemble Sm3+ fluorescence in monazite, not in apatite .
Moreover, the rear part of yellow zones is abundantly impregnated with fluid inclusions (Fig.10). One may suppose a connection between the latter and recrystallization processes. If so, the inclusions are likely to be a reservoir for repelled sulfates.
We find it possible to use the data obtained on impurities withdrawal from fluorapatite to explain other natural processes. The regularity of atomic ratios of trace elements, on the one hand, and withdrawal of the latter, on the other, may be regarded as indicative of temperature effects on preformed fluorapatite.
A generalized pattern of cathodoluminescence can be interpreted as a particular ontogenetic sequence presented in Fig.11.
The results of preliminary colorimetric analysis of CL images allow us to consider the process of formation of mineral individuals to be largely isochemical, i.e. to take place under a uniform set of chemical elements. Even so, over the period of crystal growth, physical parameters of the medium changed dramatically more than once. According to the model proposed here, fluorapatite kernels, formed in a relatively low-temperature environment, bear traces of exposure to high temperatures and partial dissolution (Fig.12). Presumably, the two outer zones of a single crystal, while yet unformed, experienced cyclic temperature shifts and, possibly, a consistent reduction in Ce. Apparently, the shifts were most pronounced during the final stages of ontogenesis. At that, ‘cold’ zones, known to be more regular and to exhibit yellow luminescence, sometimes bear traces of cataclasis with cementation of ‘hotter’ areas rich with Ce (see Fig.1). Taking into account the preserved native relative orientation of fluorapatite grains, one can state that at the end of crystal formation the temperature rose in a cyclic manner and the medium showed no evident anisotropy. Conversely, violation of symmetry of growth zoning indicates a significant trophic anisotropy during “cold”-period recrystallization (see Fig.1, 3). Single crystal kernels can be regarded as relicts left after partial dissolution of parent crystals.
Thus, the body of evidence suggests matter remobilization and partial recrystallization in a relatively closed chemical system.
The work described has convincingly demonstrated the potency of cathodoluminescence study in assessing the distribution of trace elements and reconstructing favorable conditions for single crystals formation.
We believe that integration of CL findings and data on chemical composition of minerals as well as comparisons with experimental models are able to take the interpretation of CL zoning to a higher level. The latter would open prospects for a new type of genetic reconstructions. The current use of CL as just a visualization method for the internal structure of single crystals in order to solve problems of geochronology does not fully reveal its might and should be reconsidered.
1. Goetze J. Potential of cathodoluminescence (CL) microscopy and spectroscopy for the analysis of minerals and materials // Anal.Bioanal.Chem. 2002. V. 374. P. 703–708.
2. Novikov I.A., Griboedova I.G. Optical pattern formation in cathodoluminescing monocrystalline grains. Tezisy dokladov 5-oy Rossiyskoy konferentsii po izotopnoy geokhronologii “Geokhronometricheskie isotopnye sistemy, metody ikh izucheniya, khronologiya geologicheskikh protsessov” (Proc. 5th Russ. Conf. “Geochronometric isotropic systems, investigation methods, chronology of geological processes”). Moscow. 2012. (In Russian). P. 255–257.
3. Karetnikov A.S. The problem of age definition of the Kondyor massif // Russian Journal of Pacific Geology. 2005. V. 27. №. 4. P. 46–83 (in Russian).
4. Karetnikov A.S. Paleomagnetism of ultramafics from the Konder Massif and its age assessment // Russian Journal of Pacific Geology. 2009. V. 28. №. 6. P. 23–42 (in Russian).
5. Novikov I.A., Griboedova I.G., Golovanova T.I. Optical basis of cathodoluminescence zoning in monocrystalline grains. Godichnoe sobranie RMO “Mineralogiya vo vsem prostranstve sego slova: Problemy ukrepleniya mineral’no-syr’evoy bazy i ratsional’nogo ispol’zovaniya mineral’nogo syr’ya” i Fedorovskaya sessiya 2012 (Proc. Annual Meeting of RMS “Mineralogy in the whole space of this word: Problems of enhancing and rational use of mineral resources” and Fedorov session 2012), Saint Petersburg, 2012. (In Russian). P. 376–377.
6. Gorobets B.S., Rogozhin A.A. Spectry luminestsentsii. Spravochnik (Luminescence spectra. Handbook). Publisher: All-Russian Research Institute of Mineral Resources. 2001. 316 p.
7. Roeder P., MacArthur D., Ma X.-P., Palmer G., Mariano A. Cathodoluminescence and microprobe study of rare-earth elements in apatite // American Mineralogist. 1987. V. 72. P. 801–811.
8. Barbarand J., Pagel M. Cathodoluminescence study of apatite crystals // American Mineralogist. 2001. V. 86. P. 473–484.
9. Gunawardane R.P. Solid solubility of apatites in silicophosphates and silicosullphates // J. Nath.Sci.Coun.Sri Lanka . 1993. V. 21(2). P. 243–252.
10. Pasero M., Kamf A., Ferralis C., Pekov I., Rakovan J., White T. Nomenclature of the apatite supergroup minerals // European Journal Mineral. 2010. V.22. P. 163–179.
11. Harlov D.E., Forster H.-J. and Schmidt C. High P-T experimental metasomatism of a fluorapatite with significant britholite and fluorellestadite components: implications for LREE mobility during granulite-facies metamorphism // Mineralogical Magazine. February 2003, V. 67 (1) P. 61–72.
12. Ziemann M., Forster H., Harlov D. and Frei D. Origin of fluorapatite–monazite assemblages in a metamorphosed, sillimanite-bearing pegmatoid, Reinbolt Hills, East Antarctica // Eur. J. Mineral. 2005. V. 17. P. 567–579.
 The sample courtesy of A.G. Mochalov, leading researcher at the Institute of Precambrian geology and geochronology, RAS
According to one of the possible coupled substitutions, charge imbalance in fluorapatite can be compensated by trivalent REEs built in the position of Ca2+ : Ca2+ + TR3+ = Si4+ + Na+, 2 Ca2+ = TR3+ + Na+
 In the system supposed, monazite and thenardite are analogues of xenotime and anhydrite .