Chemical zoning and crystal size distribution of garnet in a quartzo-feldspathic gneiss from the Lützow-Holm Complex at Skallen, East Antarctica: Implication for geothermometry

Abstract

Studies of chemical compositions and grain size of garnet in a quartzo-feldspathic gneiss from the Lützow-Holm Complex at Skallen, East Antarctica, form the basis of an interpretation of timing of garnet homogenization. The gneiss contains only garnet as mafic minerals except sporadic biotite inclusions in the garnet. The garnet represents approximately constant Mg/(Fe + Mg) within a grain. The values of small grains are higher than those of large grains, while the grossular content at their periphery is similar. The grain size shows lognormal distribution. These features indicate that the grains were mutually not in equilibrium and were poorly annealed, which implies insufficient grain-boundary diffusion. The consideration based on a qualitative phase diagram suggests that the Mg/(Fe + Mg) of the small grains and that of the large grains were homogenized before and after cessation of the grain-boundary diffusion, respectively. The small grains preserve the composition that was in equilibrium with biotite. In contrast, the large grains represent their bulk compositions of chemical zoning profiles that were formed during fractional growth. The homogeneous interior of large grains is, therefore, not always appropriate for geothermometry. The highest Mg/(Fe + Mg) among small grains would be suitable when the grain-boundary diffusion ceased before homogenization. The erroneous use of large grains causes depression of the estimated temperature by 60-80 °C, even though the compositional difference corresponds to only ∼ 10 °C in the phase diagram.

INTRODUCTION

Garnet in low- to medium-grade metamorphic rocks commonly shows concentric chemical zoning patterns with an outward increase in Mg/(Fe + Mg), denoted hereafter as #Mg (e.g., Atherton, 1968). In contrast, it is composed of a homogeneous interior with or without rim that shows an outward decrease in #Mg in high-grade metamorphic rocks (e.g., Grant and Weiblen, 1971). The transition of the compositional patterns has been recognized even within a single metamorphic region (Tracy et al., 1976; Yardley, 1977; Dempster, 1985; Ikeda, 1993). These features have been interpreted as elimination of growth zoning of outward increase in #Mg due to intracrystalline diffusion and subsequent modification at the rim due to exchange and garnet-consuming reactions during retrograde metamorphism (e.g., Spear, 1993; Yardley and Warren, 2021). This scenario assumes that surface equilibrium was maintained during growth of the garnet as well as its retrograde modification. The homogeneous interior has been considered, therefore, to preserve the composition at the peak metamorphism (e.g., Lasaga, 1983; Spear, 1991).

However, use of such a homogeneous interior for geothermometry does not always provide consistent temperatures with other methods. This is clear in the Lützow-Holm Complex (LHC) in East Antarctica when we use the garnet-biotite geothermometer for pelitic and quartzo-feldspathic lithologies. In Skallevikshalsen, the garnet-biotite thermometer yields a wide temperature range with the lowest estimate of 770 °C (Yoshimura et al., 2008), whereas the Ti-in-zircon thermometer estimates the peak temperatures of 820-850 °C (Kawakami et al., 2016). In Skallen, the garnet-biotite thermometer mostly shows 650 to 750 °C (Suzuki, 1983; Yoshida and Aikawa, 1983; Motoyoshi, 1986; Santosh and Yoshida, 1992). In contrast, the thermometer based on the partitioning of carbon isotopes between calcite and graphite (Satish-Kumar and Wada, 2000) and that of Zr content in rutile (Suzuki and Kawakami, 2019) provide the equilibrium temperatures of 850 and 900 °C, respectively. In Akarui Point, the garnet-biotite thermometer yields 770-790 °C using the periphery of the garnet that exhibits an outward increase in #Mg (Kawakami et al., 2008). In contrast, the pseudosectioning method (Iwamura et al., 2013) and the two-feldspar thermometer applied to mesoperthite (Nakamura et al., 2013) estimate 900-920 and 825-900 °C, respectively. These examples clarify that the garnet-biotite geothermometer applied to pelitic and quartzo-feldspathic lithologies estimates lower temperatures than the other methods regardless of zoning patterns of garnet. This discrepancy has not been so far well explained.

This study describes a garnet-bearing quartzo-feldspathic gneiss from the LHC at Skallen. It contains only garnet as mafic minerals, which enables to examine homogenization process of garnet without retrograde modification. This study reveals that the garnet grains have different homogeneous #Mg values from each other. The considerations using a qualitative phase diagram conclude that the large grains do not preserve equilibrium #Mg at high temperatures. This may be one of the possible reasons for the above discrepancy.

GEOLOGICAL SETTING

The Lützow-Holm Complex is a Cambrian orogenic belt showing metamorphic grade from amphibolite to granulite facies, and is divided into the amphibolite-facies area, transitional area and granulite-facies area (index map of Fig. 1) (Hiroi et al., 1983, 1987; Shiraishi et al., 2003). The complex shows clockwise pressure-temperature paths that include nearly isothermal decompression at high pressures (Motoyoshi et al., 1985; Kawasaki et al., 1993). The thermobaric structure of the LHC is currently revisited based on the thermometry using solubility of minor elements and the pseudosectioning method (e.g., Iwamura et al., 2013; Suzuki and Kawakami, 2019; Takamura et al., 2020). These studies reveal that Akarui Point and Tenmondai Rock in the transitional area also experienced similar high temperatures to the granulite-facies area.

Figure 1. Geological map of Skallen simplified after Osanai et al. (2004), showing the locality of the studied sample. The index map represents the metamorphic zonation of the Lützow-Holm Complex, East Antarctica, after Hiroi et al. (2019).

Skallen is one of the exposures in the granulite-facies area and composed mainly of metamorphic rocks with various lithologies including pelitic, quartzo-feldspathic, calcareous and mafic, together with minor pre-, syn-, and post- metamorphic intrusive rocks (Yoshida et al., 1976; Osanai et al., 2004). Thin alternation of the above four lithologies comprises the main constituent, which is denoted as layered gneiss in Figure 1 as used by Hokada and Motoyoshi (2006). The metamorphic rocks show general trend of ENE-WSW to ESE-WNW and are characterized by ENE-WSW to E-W trending open to gentle upright folds. They are geographically divided into the northern, central and southern area by E-W trending thrusts or shear zones (Osanai et al., 2004). The northern and the southern area contain calcareous gneiss regionally, which is distinct from the central area. In the central area, the layered gneiss is dominant on both the northern side and the southern side of the central mafic gneiss (Fig. 1). The layered gneiss would continue to Skallevikshalsen, 1 km west of Skallen, as judged from the general trend of the metamorphic rocks. It contains the garnet-sillimanite gneiss that has been investigated by Kawakami and Motoyoshi (2004), Kawakami and Hokada (2010), and Kawakami et al. (2016).

The metamorphic pressures have been estimated by the combination of conventional geothermobarometers at 410-650 MPa at 820 °C (Yoshida and Aikawa, 1983), 630 MPa at 730 °C (Suzuki, 1983), and 770-1080 MPa at 810 °C (Motoyoshi, 1986). However, recent studies estimate higher temperatures of 850-900 °C, as described before. Therefore, it would be required to reexamine the pressure conditions. Motoyoshi et al. (1985) recognize the inclusions of kyanite and staurolite in garnet of the garnet-sillimanite quartzo-feldspathic gneisses, suggesting that prograde transition from kyanite-stable to sillimanite-stable conditions. Satish-Kumar et al. (2006) detect CO₂-rich fluid infiltration at 600 °C during retrograde metamorphism. Hokada and Motoyoshi (2006) recognize monazite showing ages of 650-580 Ma that are significantly older than 550-520 Ma of SHRIMP zircon ages for the LHC (Shiraishi et al., 2003).

The investigated quartzo-feldspathic gneiss (I-241) was obtained from an eastern part in the central area of Osanai et al. (2004) (69°40′02′′S, 39°26′23′′E in Fig. 1), the same locality as Goto and Ikeda (2008) described. In the outcrop, the prevailing quartzo-feldspathic gneiss comprises compositional layering several decimeters thick, which stems from different modal abundance of visible garnet (Fig. 2).

Figure 2. Photograph showing the mode of occurrence of the layered quartzo-feldspathic gneisses. The layer examined in this study is colored in red.

ANALYTICAL WORK

The studied sample was obtained from a single random layer of quartzo-feldspathic gneiss ∼ 10 cm thick (Fig. 2). The analytical procedures are as below. To examine chemical zoning of garnet, we picked grains in various sizes from the sample using a thick needle of 2.5 mm diameter. Each grain was separately mounted in a vessel filled with resin. It was scraped off so that the grain showed its maximum cross-section area, and its surface was polished. We also prepared ordinary thin sections to describe microstructures and determine chemical compositions of constituent minerals including garnet.

The chemical compositions were determined by a scanning electron microscope (JEOL 5800LV) combined with an energy dispersive analytical system (Link ISIS) at Kyushu University. A 20 kV accelerating voltage, a 0.5 nA beam current and a 100 s livetime were employed. The element standards used in this study are calcium fluoride for Ca and oxides for Mg, Al, Si, and metals for Mn and Fe. The analytical precision 1σ is within 5% for elements when the oxide content exceeds 1 wt%.

The ferric estimates in garnet based on stoichiometry show that a significant number of analyses (∼ 15%) provide negative Fe³⁺ and most of the analyses show Fe³⁺/(Fe²⁺ + Fe³⁺) less than 0.05. These features do not correlate with the grain size or the analyzed position of each garnet. Therefore, this study treats total iron in garnet as divalent to avoid subjective treatment.

The sizes of garnet grains were determined using digitalized photographs on the surface of the layer that we determined the chemical compositions of the constituent minerals. Counting the pixels of each grain provides its area. Its size is defined as the diameter of a circle of the equivalent area.

The histogram of the above grain size was stereologically converted to the 3-D size distribution based on the method of Saltykov (1967), which is briefly shown below. The grain sizes measured on the layer surface do not represent the maximum of each grain, i.e., true size, because the surface traverses an arbitrary position of the grain. Conversely, grains will appear smaller on the surface than their true sizes. The 2-D sizes of a single grain can be described as a probability function. Therefore, the number of grains within a size interval in 2-D histogram is obtained as accumulation of the probability functions for larger grains than this size bin. The inversion of this relation enables to estimate 3-D crystal size distribution (CSD).

RESULTS

Petrography

The sample is composed mainly of garnet, plagioclase (An 35-45), K-feldspar (Or 90) and quartz, together with minor sillimanite, ilmenite, rutile and zircon, and rare biotite. The inclusion minerals in garnet are quartz, plagioclase (An 35-50), sillimanite, biotite, ilmenite, rutile, and zircon.

Garnet grains show spatially homogeneous occurrence without significant clustering (Fig. 3). They represent convex and rather euhedral shape (Figs. 4a and 4b) and commonly contain quartz and plagioclase as large as several millimeters, in which quartz grains are partially surrounded by a thin film of plagioclase (Fig. 4a). Some garnet grains include fine-grained elongated quartz grains that locally represent sigmoidal trails being continuous through the garnet grains (Fig. 4b). Rare sillimanite grains with needle-like shape are also recognized so far at the peripheral part of large garnet grains (Fig. 4c).

Figure 3. Photograph of the studied sample (no. I-241).

Figure 4. Photomicrographs showing the mode of occurrence of constituent minerals. (a) Garnet including coarse-grained quartz and plagioclase. (b) Sigmoidal trace of inclusions composed mainly of fine-grained quartz. (c) Tiny sillimanite inclusions near the periphery of garnet. (d) Euhedral sillimanite grain in the matrix. (e) Shape preferred orientation of quartz and plagioclase in the matrix. (f) Retrograde replacement of the matrix biotite by chlorite. Abbreviations: Grt, garnet; Pl, plagioclase; Qtz, quartz; Sil, sillimanite.

Sillimanite in the matrix is, in contrast, coarse-grained and shows euhedral shape (Fig. 4d). Quartz and plagioclase in the matrix show significant variation in size and show shape-preferred orientation (Fig. 4e). Plagioclase grains both in garnet and in the matrix locally undergo retrograde alteration, recognized as dusty domain with composition of nearly pure albite or presence of fine-grained muscovite. Rare biotite grains both in garnet and in the matrix are largely replaced by chlorite (Fig. 4f). Rutile in the matrix is locally surrounded by sphene. Some cracks in garnet are filled with chlorite.

Chemical compositions of garnet

The molar fractions of spessartine and grossular are less than 0.013 and 0.077, respectively. The analytical errors shown before cause uncertainties in the fraction of the endmembers and #Mg of 0.01 and 0.007, respectively. The garnet of this study can therefore be treated as ternary Ca-Fe-Mg solid solution. Representative compositions are shown in Table 1.

Table 1. Representative garnet analyses (12 oxygen)

Grain	C-grt1	C-grt2	C-grt3X	C-grt4	C-grt5	C-grt6	C-grt7	C-grt8	C-grt9	C-grt10	C-grt11b
Analysis	138	10	54	13	14	21	27*	a-9	38	20	13
Distance from Periphery (mm)	6.23	0.70	5.04	1.25	1.20	3.60	5.45	0.80	8.09	8.30	5.00
SiO2	38.55	38.44	38.99	38.43	38.49	38.68	37.99	38.60	38.74	38.72	38.99
Al2O3	21.55	21.78	21.84	21.56	21.65	22.02	21.58	21.74	21.62	22.16	22.00
FeO	31.59	31.05	29.83	30.68	30.36	29.36	29.98	30.58	30.65	31.35	31.38
MnO	0.44	0.42	0.43	0.39	0.47	0.49	0.49	0.40	0.43	0.35	0.42
MgO	6.89	7.22	7.68	6.88	7.90	7.50	7.13	7.95	7.08	7.27	7.65
CaO	1.85	1.53	2.21	2.62	1.41	2.62	1.94	1.21	2.08	1.32	1.47
Total	100.87	100.44	100.98	100.56	100.28	100.67	99.11	100.48	100.60	101.17	101.91
Si	2.999	2.993	3.005	2.994	2.993	2.990	2.993	2.994	3.009	2.990	2.991
Al	1.976	1.999	1.984	1.980	1.984	2.006	2.004	1.988	1.979	2.017	1.989
Fe	2.055	2.022	1.922	1.999	1.974	1.898	1.975	1.984	1.991	2.025	2.013
Mn	0.029	0.028	0.028	0.026	0.031	0.032	0.033	0.026	0.028	0.023	0.027
Mg	0.799	0.838	0.882	0.799	0.916	0.864	0.837	0.919	0.820	0.837	0.875
Ca	0.154	0.128	0.182	0.219	0.117	0.217	0.164	0.101	0.173	0.109	0.121
Total cation	8.012	8.008	8.003	8.017	8.015	8.007	8.006	8.012	8.000	8.001	8.016

The zoning profiles of 11 grains that traverse their centers are shown in Figure 5, arranged in the order of ascending grain size. The grossular represents the largest variation among the three components within a grain, which is still small as 0.05 at the most (C-grt6 in Fig. 5). This variation is compensated by almandine and pyrope that show similar patterns of profiles, resulting in that each grain shows approximately constant #Mg mostly within a range of 0.02 (Fig. 5). The four small grains show similar values of #Mg that are larger than the other large grains.

Figure 5. Compositional profiles with respect to the content of almandine, pyrope and grossular together with #Mg [= Mg/(Fe + Mg)] of 11 grains cut through their center. The grains are arranged in the order of increasing their size that is measured as the diameter of a circle of the equivalent area, as shown at the bottom of the figure (mm).

Most of the grains have minimum grossular content of ∼ 0.03 at their periphery (Fig. 5). In contrast, the grossular content at the center varies and does not correlate with the grain size. The six grains larger than 5.9 mm show similar zoning patterns, where the maximum grossular content appears at the intermediate part between the center and the periphery. They show approximately symmetric patterns with respect to the geometrical centers. In contrast, the five grains smaller than 5.1 mm represent different profiles from each other. Some are symmetric and the others are asymmetric. The latter grains may have their nuclei at out of geometrical center due to anisotropic growth.

Figure 6 shows the values of mean #Mg against the sizes of the grains. The sizes of the grains cut through their centers (solid symbols) are larger than those of the grains in thin sections (open symbols) at any #Mg. This feature can be well explained geometrically such that the true sizes of the grains should be larger than the sizes in thin sections.

Figure 6. Plot of mean #Mg [= Mg/(Fe + Mg)] of the garnet grains shown in Figure 5 versus their size (solid symbols). The grains from ordinary thin sections are also shown as open symbols.

Grain size distribution of garnet

332 grains were recognized in total on the layer surface of 150 cm², representing the number density of 2.19 cm⁻². The modal abundance of garnet and the mean grain size are 31.5% and 3.31 mm, respectively. These values are within the ranges from five layers of the same outcrop in Figure 2, as described by Goto and Ikeda (2008).

Figure 7a shows a histogram of the 332 grains with a size interval of 1 mm. It shows positive skewness, i.e., a long tail on the coarse-grained side. A small peak may be recognized in the size bin of 8-9 mm in Figure 7a. This is derived from only one grain more than the next size bin on the fine-grained side, which is negligible.

Figure 7. Results of grain size measurement. (a) Histogram of the grain size measured on the surface of the layer. (b) Estimated 3-D CSD. The red line shows the best-fitted lognormal distribution.

The size distribution in Figure 7a is converted to 3-D CSD, using the method shown before. The 3-D distribution seems to be recognized as either lognormal or semilogarithmic linear. The difference between the two recognitions will appear within the smallest size bin of 0-1 mm. We confirm that the histogram with a size interval of 0.1 mm shows monotonic increase of the frequency in the size bins from 0-0.1 to 0.5-0.6 mm. The chi-squared test also indicates that the distribution can be regarded as lognormal with the maximum frequency at 0.553 mm (Fig. 7b).

DISCUSSION

Insufficient grain-boundary diffusion

The presence of biotite, sillimanite, and quartz as inclusions in garnet and the common occurrence of K-feldspar in the matrix are indicative of the following reaction that is responsible for growth of the garnet,   

\begin{align} &\text{biotite} + \text{sillimanite} + \text{quartz} \\&\quad= \text{garnet} + \text{K-feldspar} + \text{melt or aqueous fluid} \end{align} (1).

Taking the above solidus condition of the metamorphism in Skallen into account, melt is more likely than aqueous fluid. The concave and rather euhedral shape of the garnet suggests its growth in surface equilibrium with other reactant minerals rather than its disequilibrium growth under supersaturated conditions. The inclusion trails showing sigmoidal shape are continuous through the grain (Fig. 4b), as described before. This means continuous growth of garnet during a single stage of deformation without significant coalescence.

The virtual absence of biotite in the matrix indicates that the biotite was completely consumed during prograde metamorphism and was not produced during retrograde metamorphism. The absence of retrograde biotite suggests absence of melt and aqueous fluid during the retrograde metamorphism because they are necessary to produce biotite at the expense of garnet.

The lognormal CSD (Fig. 7b) implies paucity of the process of annealing or Ostwald ripening that modifies any CSD to show negative skewness (e.g., Cashman and Ferry, 1988; Miyazaki, 1991). In addition, preservation of the shape preferred orientation of quartz and plagioclase (Fig. 4e) suggests that the recovery process was also incomplete. These features are indicative of a condition of insufficient grain-boundary diffusion because these processes require migration of components between the grains. This condition would be derived from the absence of melt and aqueous fluid.

The feature of different compositions in different grains (Fig. 6) means that the grains were chemically isolated without mutual equilibrium after their growth, which is also indicative of the condition of insufficient grain-boundary diffusion. This condition exhibits a remarkable contrast with the surface equilibrium during the garnet growth. This transition is likely synchronous to the disappearance of biotite in the rock. That is, the melt produced by reaction (1) was removed completely when biotite was entirely consumed.

Chemical heterogeneity within and between garnet grains

The chemical heterogeneity in a grain is not so intense but is significant, which correlates with the distance from the periphery (Fig. 5). In the ternary Ca-Fe-Mg solid solution, Ca is compensated by the total amount of Fe and Mg. Considering the diffusion coefficient of Ca that is smaller than those of Fe and Mg in the fifth order of magnitude (Carlson, 2006), the grossular profiles are more likely to preserve the growth zoning even which would also be modified to some extent by intracrystalline diffusion. The garnet-forming reaction (1) predicts that the #Mg of garnet increases during its growth. However, it is almost constant within the grains (Fig. 5). This suggests that Fe and Mg were homogenized in the crystallographic site of eightfold coordination even though Ca in the site preserved the growth zoning. The feature of the same grossular content at the periphery of most of the grains (Fig. 5) indicates that these grains finished growing at the same time when biotite disappeared.

It is still puzzling that the homogeneous #Mg values differ between grains in different sizes (Fig. 6). Provided that the surface equilibrium was maintained during homogenization of each grain, the resultant homogeneous #Mg values of all the grains should be the same. The present feature, therefore, suggests that the surface equilibrium already ceased by the end of the homogenization at the latest.

Recall a schematic pseudo-binary phase diagram of reaction (1) at constant pressure (Fig. 8). Biotite is the only mafic mineral present before garnet formation, indicating that the bulk #Mg of the rock, X₀, is identical to that of biotite, B₀, which produces garnet with the #Mg of G₀ at T = T₀ (Fig. 8). There are two extreme processes during the garnet growth. One is equilibrium growth where garnet and biotite are homogenized step by step, and the homogeneous garnet (G₁), showing identical #Mg to X₀, is formed when biotite disappears at T₁ (Fig. 8). The other is fractional growth in surface equilibrium with homogeneous biotite, where garnet forms growth zoning and its interior is out of chemical equilibrium. The effective bulk #Mg of the rock is close to the #Mg of biotite at any temperature, and both biotite and the periphery of garnet are in equilibrium until #Mg = 1.0 at T₂. The geological process would be an intermediate one between the two due to limited intracrystalline diffusion in garnet. Therefore, garnet would grow even at a higher temperature than T₁.

Figure 8. Schematic pseudo-binary phase diagram for reaction (1) at constant pressure. G_i and B_i denote #Mg [= Mg/(Fe + Mg)] of the coexisting garnet and biotite at temperature T_i, respectively. X₀ and G_bulk represent bulk #Mg of the rock and that of garnet, respectively. A two-way arrow shows a possible range of G_bulk in the presence of biotite at T₁.

Firstly, we suppose that the system contains a single grain of zoned garnet in surface equilibrium with homogeneous biotite at T_H (>T₁) (Fig. 8). The effective bulk #Mg of the rock exceeds X₀ because it locates between G_H and B_H that represent the equilibrium #Mg of garnet and biotite, respectively. The difference between the effective bulk #Mg and X₀ is compensated by the #Mg that is stored in the interior of the zoned garnet. The bulk #Mg of the zoned garnet, G_bulk, at this time is lower than X₀ even though G_H at the periphery of the garnet is higher, because the rock (X₀) is composed of garnet (G_bulk) and biotite (B_H) (Fig. 8). During the intracrystalline diffusion of Fe and Mg, G_bulk tends to approach G_H as long as biotite (B_H) is present. This compositional change requires irreversible progress of reaction (1) under the isophysical condition, as pointed out by Spear (1988). The change in G_bulk is therefore associated with further growth of the garnet grain at the expense of biotite, which would be quantitatively described as,   

\begin{equation} \text{garnet ($G_{\textit{bulk}}$)} + \text{biotite ($B_{\text{H}}$)} = \text{garnet ($G_{\text{H}}$)} \end{equation} (2).

However, it is stressed that G_bulk cannot reach G_H because an imaginary rock composed of homogeneous garnet (G_H) and biotite (B_H) should have #Mg different from the initial bulk #Mg of X₀ (Fig. 8). In other words, reaction (2) terminates when biotite (B_H) disappears before G_bulk reaches G_H, and the G_bulk at the time should be identical to G₁ because the only mafic mineral is garnet with the #Mg of G_bulk. Therefore, the garnet with the bulk #Mg of G₁ is still not homogeneous, the periphery of which shows G_H. After that, the garnet will be homogenized isochemically with the bulk #Mg remaining at G₁.

Secondly, we expand the argument to the system that contains multiple grains of garnet. The bulk #Mg of garnet, G_bulk, at T_H is, in turn, composed of each bulk #Mg of the multiple grains. The #Mg at the periphery of all the grains were G_H because they were in equilibrium with biotite (B_H) (t₂ in Fig. 9). During homogenization in the presence of biotite, every periphery of the grains keeps G_H while each bulk #Mg, G_mean, tends to approach G_H at the expense of biotite (B_H) by reaction (2). This brings the growth of garnet to some extent, as mentioned above, which may be recognized more clearly in smaller grains (t₃ in Fig. 9). The grain-boundary diffusion ceases when biotite disappears, as discussed above. As each grain is thereafter chemically isolated and continues to be homogenized, the #Mg at the periphery does not remain G_H anymore.

Figure 9. Time-stepped illustration showing formation of different homogeneous compositions between a small (the upper) and a large (the lower) grain, based on the phase diagram of Figure 8. The small grain and the large grain are homogenized before and after disappearance of biotite, respectively. The size of biotite represents its modal abundance qualitatively. The steps include nucleation of garnet (t₁), growth with increasing temperature from T₀ to T_H in Figure 8 (t₁-t₂), complete homogenization of the small grain (t₃), disappearance of biotite (t₄) and complete homogenization of the large grain in the absence of biotite (t₅). Note that the grains grow even at constant temperature T_H by reaction (2) during homogenization process, as shown by two broken lines on the small grain between t₂ and t₃. The relevant consumption of biotite is demonstrated by the decrease of its size.

It is likely that small grains are chemically homogenized earlier than large grains because their volume to be homogenized is smaller. This predicts that the #Mg of the small grains is homogenized to G_H in the presence of biotite (t₃ in Fig. 9). In contrast, the large grains are not yet chemically homogenized completely. Provided that the large grains gain the same amount of components as the small grains by reaction (2), this amount would be small compared with the total amount stored within the interior of the large grains. Thus, the values of bulk #Mg of the large grains at the disappearance of biotite would not be largely different from those of bulk #Mg of the original growth zoning (t₄ in Fig. 9). The #Mg of the large grains will be afterward homogenized in the isolated system (t₅ in Fig. 9).

Based on these arguments, we consider that the average of the similar mean #Mg values of the four small grains in this study, 0.316, represents G_H. In contrast, the values of #Mg of the seven large grains, distinctly lower than 0.316, are regarded as bulk #Mg of their growth zoning. Considering that the bulk #Mg of the whole grains is G₁ and that of the small grains show G_H, the bulk #Mg of at least some large grains should be smaller than G₁. The values of the bulk #Mg of the zoned grains seem, in general, not necessarily similar to each other, and they are able to show a range between G₀ and G_H at the most. Nevertheless, the large grains show similar mean #Mg values between 0.287 and 0.298, the approximate average of which is 0.29 (Fig. 6). This suggests that the bulk #Mg of the multiple grains, which is G₁ and hence identical to B₀ and X₀, is larger than but close to 0.29.

Bulk #Mg of zoned grain

The similarity in the bulk #Mg of zoned grain can be examined as follows. Considering a single concentrically zoned grain in the binary system, its bulk #Mg, G_mean, can be described as:   

\begin{align} &\frac{4}{3}\pi G_{0}R_{0}{}^{3} + \int\nolimits_{R_{0}}^{R_{\text{H}}}4\pi\text{#Mg}R^{2}dR \\&\quad= \frac{4}{3}\pi G_{\textit{mean}}R_{\text{H}}{}^{3} \end{align} (3),

where R₀ and R_H denote the radius of the spherical grain at T₀ and T_H in Figure 8, respectively. The first and second term of the left-hand side represent the total amount of #Mg inside and outside of the nucleus with radius R₀, respectively. We employ a growth model proposed by Eberl et al. (1998) that accounts for lognormal CSD. This model requires that the grains nucleate simultaneously and grow at a rate proportional to their size shown as:   

\begin{equation} \frac{dR}{dt} = \beta R \end{equation} (4),

where β is a number that varies slightly and randomly from grain to grain. The rate of temperature increase is assumed to be constant,   

\begin{equation} \frac{dT}{dt} = \alpha \end{equation} (5).

Though this assumption is not so far evaluated, the value of α may be obtained, if necessary, as the temperature range during grain growth divided by the duration time because Eq. (3) does not deal with any diffusion process. The #Mg at the periphery where surface equilibrium is maintained changes with T, and its rate is assumed to be constant.   

\begin{equation} \frac{d\text{#Mg}}{dT} = \varepsilon \end{equation} (6).

It is valid within any #Mg range of ∼ 0.2 according to the thermodynamic calculation of Holland and Powell (1990, 1998) together with ideal mixing models for garnet and biotite.

Substituting Eqs. (4) to (6) into Eq. (3) and integrating yield:   

\begin{equation} G_{\textit{mean}} = G_{\text{H}} - \frac{\alpha \varepsilon}{3\beta}\left\{1 - \left(\frac{R_{0}}{R_{\text{H}}}\right)^{3} \right\} \end{equation} (7).

Equation (7) shows that G_mean descends from G_H with increasing grain size, 2R_H. The value of G_H is regarded as 0.316, as discussed before. The critical size of nucleation, in general, decreases with the increase of degree of supersaturation, which is 0.001 mm at the temperature overstepping of 20 °C (Miyazaki, 2015). Assuming 2R₀ of 0.001 mm yields virtually constant G_mean irrespective of grain size (the blue and the red line in Fig. 10). The values of G_mean increase with increasing 2R₀ for small grains but are still constant for the grains larger than 5 mm even when 2R₀ is 1.3 mm, slightly smaller than the smallest grain in Figure 6 (the yellow and the green line in Fig. 10). This consideration confirms the interpretation that the similar #Mg values of the seven large grains are regarded as bulk #Mg of their growth zoning. The growth model of Eberl et al. (1998) explains not only the lognormal CSD but also the approximately constant G_mean of the large grains.

Figure 10. Results of Eq. (7) showing G_mean (bulk #Mg of a grain) as a function of grain size, 2R_H. The #Mg value of G_H is set as 0.316, the average of the mean #Mg of the four small grains in Figure 6. The solid symbols are the same in Figure 6.

Evaluation of the simplification

Our arguments are based on the isobaric phase diagram of Figure 8. The used feature is that garnet grows with increasing its #Mg at the expense of biotite. The reaction (1) shows a Clapeyron slope of ∼ 4 MPa/°C in both the Fe- and the Mg-endmember system, which is much larger than the slope of kyanite-sillimanite transition. Therefore, the used feature is still valid in the clockwise pressure-temperature path proposed in the LHC that includes the transition from kyanite to sillimanite (e.g., Motoyoshi et al., 1985).

We also assume a continuous growth process in Figure 8, which is supported by the continuity in both the inclusion trails and the CSD. The former feature suggests, strictly speaking, that the grains grew in a single stage of deformation. In contrast, Kawakami and Hokada (2010) recognize discontinuity of phosphorous content in garnet from garnet-sillimanite gneiss in the neighboring exposure, Skallevikshalsen. The inner part contains prograde relic minerals (Kawakami and Motoyoshi, 2004) and monazite that shows older ages of 650-580 Ma than the matrix monazite (Kawakami et al., 2016). The outer part is interpreted as crystallization from melt during retrograde stage (Kawakami and Hokada, 2010). These features suggest that the garnet grew in two stages by different reactions at possibly different ages. The lithological similarity and the geographical proximity suggest that the present rock could experience the similar history.

The bimodal mean #Mg divided by grain size (Fig. 6) might seem consistent with the interpretation of the two-stage growth. While the similar grossular content at the periphery of most of the grains indicates that they all grew until the temperature reached T_H in Figure 8 irrespective of their #Mg and size, the argument based on Figure 8 does not exclude the two-stage growth associated with some temperature descent during the interval. The #Mg of the pre-existing grains would be homogenized partially or completely during the interval and revive to grow when the temperature exceeds the previous maximum. The zoning profile at t₂ in Figure 9 will be modified to show a somewhat homogeneous core with zoned rim, leading to the higher bulk #Mg of the grain. The size of such pre-existing grains can also be regarded as 2R₀ in Eq. (3), and the present argument still survives when 2R₀ is 1.3 mm, as discussed above. Furthermore, we obtain the same conclusion even if another reaction is responsible for the subsequent growth. That is, the phase diagram of Figure 8 and the growth law of Eq. (4) can be applied to the subsequent reaction.

Implication for geothermometry

The important feature inferred in this study is that the grain-boundary diffusion ceased before completion of garnet homogenization. This cessation does not always require the disappearance of biotite. In Skallen, Satish-Kumar et al. (2006) detected multiple stages of fluid filtration of which composition was controlled by the host lithology during retrograde metamorphism. Ikeda (2004) recognized local formation of retrograde biotite. The lognormal CSD of garnet is also recognized in other layers of the same outcrop of this study (Goto and Ikeda, 2008). Yoshimura et al. (2008) consider that the sillimanite and garnet in the similar lithology to this study were the restitic products, which requires the effective leaching of melt out of the rocks. Furthermore, the corona microstructures are common in the LHC (e.g., Hiroi et al., 1983; Iwamura et al., 2013; Mori and Ikeda, 2018). These features are indicative of intermittent or insufficient activity of fluid and/or melt. Therefore, the condition of insufficient grain-boundary diffusion would be common after peak metamorphism in the whole LHC.

A serious problem arises in geothermometry if the #Mg of some grains is not completely homogenized at the cessation of grain-boundary diffusion. After the cessation, the grains are chemically isolated. The continued intracrystalline diffusion will change the composition at any position of each of the grains without changing its bulk composition. There is no place that preserves the equilibrium composition under any peak or retrograde condition. For geothermometry, we need to identify the grains of which #Mg was homogenized before the cessation of grain-boundary diffusion.

As an example, the difference in #Mg between the small and the large grains in this study seems not so large. Treating the average of #Mg in the small grains of 0.316 as G_H and that in the large grains of 0.29 as G₁ (hence, B₀), respectively, the thermodynamic calculation shown before provides temperatures T₀, T₁, and T_H as 658, 822, and 836 °C, respectively, at a nominal pressure of 700 MPa. The temperature range for the fractional growth is 178 °C, which is larger than that for the equilibrium growth only by 14 °C. In the pseudosectioning method, this difference corresponds to the difference in the stability of garnet-biotite assemblage between the equilibrium and the fractional model. The evaluation using some natural examples shows that this difference is as small as 30 °C (Evans, 2004). In contrast, the erroneous use of the large grains for geothermometry causes serious differences in the estimated temperatures. We firstly estimate the #Mg value of biotite as 0.612, which coexists with the garnet that shows #Mg of 0.316 (G_H), using the geothermometer of Ferry and Spear (1978) and assuming a nominal temperature and pressure of 800 °C and 700 MPa, respectively. Employing this biotite and a large garnet grain (C-grt1: #Mg of 0.288) for the same thermometer yields a lower temperature by 67 °C. This depression varies 60-80 °C in the range of 750-900 °C at 500-900 MPa, which is never negligible.

The above consideration suggests that the use of large grains for geothermometry is not always appropriate especially under the condition of insufficient grain-boundary diffusion, which provides the erroneous estimates lower than the equilibrium temperature by 60-80 °C. This would be responsible in part for the discrepancy in the temperature estimations in the LHC. The estimation using the highest #Mg of small grains would be suitable when the grain-boundary diffusion ceased before homogenization.

It should be noted that the depression of estimated temperatures is recognized mainly in the garnet-biotite geothermometer for pelitic and quartzo-feldspathic gneisses. The garnet-clinopyroxene thermometer and the garnet-orthopyroxene geothermometer for mafic gneisses provide consistent temperatures with other methods. For example, they provide 850-860 and 800-810 °C in Tenmondai Rock and Sudare Rock, respectively, which are almost the same as the estimates using the pseudosectioning method (Takamura et al., 2020). In Rundvåkshetta, the garnet-orthopyroxene geothermometer (Motoyoshi and Ishikawa, 1997) yields temperatures that agree with the experimentally determined stability of minerals (Kawasaki et al., 2011). These features would imply that the #Mg of garnet in mafic gneisses was homogenized before the grain-boundary diffusion ceased. The lithological difference may cause the difference in timing of garnet homogenization with respect to the cessation of grain-boundary diffusion.

CONCLUSIONS

The garnet grains in a quartzo-feldspathic gneiss from Skallen represent homogeneous #Mg that varies from grain to grain. They show lognormal CSD. The former feature is ascribed to the difference in timing of the completion of homogenization relative to the disappearance of biotite. The small grains preserve the equilibrium #Mg at the peak temperature whereas the large grains show the bulk #Mg values of their growth zoning profiles. A growth model that accounts for the lognormal CSD also explains the similarity between the #Mg values of the large grains. The erroneous use of the large grains for the geothermometry causes the serious depression of the estimated temperatures by 60-80 °C. This would be responsible in part for the discrepancy between the garnet-biotite geothermometry and the other methods in the LHC.

ACKNOWLEDGMENTS

The sample examined in this study was taken during the 44th Science Program of the Japanese Antarctic Research Expedition (JARE) in 2002-2003. It was supported by the National Institute of Polar Research (NIPR) under MEXT. We are grateful to all crew members of the icebreaker Shirase and all members of JARE, especially T. Kawakami, Y. Kawano, and T. Kawasaki, for their support in performing field research. We would like to thank K. Miyazaki and an anonymous reviewer for critical suggestions that significantly improved the manuscript, and appreciate T. Hokada for editorial handling.

REFERENCES

•  Atherton,  M.P. (1968) The variation in garnet, biotite and chlorite composition in medium grade pelitic rocks from the Dalradian, Scotland, with particular reference to the zonation in garnet. Contributions to Mineralogy and Petrology, 18, 347-371.
•  Carlson,  W.D. (2006) Rates of Fe, Mg, Mn, and Ca diffusion in garnet. American Mineralogist, 91, 1-11.
•  Cashman,  K.V. and  Ferry,  J.M. (1988) Crystal size distribution (CSD) in rocks and the kinetics and dynamics of crystallization III metamorphic crystallization. Contributions to Mineralogy and Petrology, 99, 401-415.
•  Dempster,  T.J. (1985) Garnet zoning and metamorphism of the Barrovian type area, Scotland. Contributions to Mineralogy and Petrology, 89, 30-38.
•  Eberl,  D.D.,  Drits,  V.A. and  Srodon,  J. (1998) Deducing growth mechanisms for minerals from the shapes of crystal size distributions. American Journal of Science, 298, 499-533.
•  Evans,  T.P. (2004) A method for calculating effective bulk composition modification due to crystal fractionation in garnet-bearing schist: implications for isopleth thermobarometry. Journal of Metamorphic Geology, 22, 547-557.
•  Ferry,  J.M. and  Spear,  F.S. (1978) Experimental calibration of the partitioning of Fe and Mg between biotite and garnet. Contributions to Mineralogy and Petrology, 66, 113-117.
•  Goto,  S. and  Ikeda,  T. (2008) Crystal size distribution of garnet in quartzo-feldspathic gneisses from the Lützow-Holm Complex at Skallen, East Antarctica. In Geodynamic Evolution of East Antarctica: A Key to the East-West Gondwana Connection (Satish-Kumar, M., Motoyoshi, Y., Osanai, Y., Hiroi, Y. and Shiraishi, K. Eds.). Geological Society, London, Special Publications, 308, 391-399.
•  Grant,  J.A. and  Weiblen,  P.W. (1971) Retrograde zoning in garnet near the second sillimanite isograd. American Journal of Science, 270, 281-296.
•  Hiroi,  Y.,  Shiraishi,  K.,  Yanai,  K. and  Kizaki,  K. (1983) Aluminum silicates in the Prince Olav and Soya Coasts, East Antarctica. Memoirs of National Institute of Polar Research Special Issue, 28, 115-131.
•  Hiroi,  Y.,  Shiraishi,  K.,  Motoyoshi,  Y. and  Katsushima,  T. (1987) Progressive metamorphism of calc-silicate rocks from the Prince Olav and Soya Coasts, East Antarctica. Proceedings of the National Institute of Polar Research Symposium on Antarctic Geosciences, 1, 73-97.
•  Hiroi,  Y.,  Hokada,  T.,  Kato,  M.,  Yanagi,  A., et al (2019) Felsite-nanogranite inclusions and three Al₂SiO₅ polymorphs in same garnet in ultrahigh-temperature granulites from Rundvågshetta, Lützow-Holm Complex, East Antarctica. Journal of Mineralogical and Petrological Sciences, 114, 60-78.
•  Hokada,  T. and  Motoyoshi,  Y. (2006) Electron microprobe technique for U-Th-Pb and REE chemistry of monazite, and its implications for pre, peak- and post-metamorphic events of the Lützow-Holm Complex and the Napier Complex, East Antarctica. Polar Geoscience, 19, 118-151.
•  Holland,  T.J.B. and  Powell,  R. (1990) An enlarged and updated internally consistent thermodynamic dataset with uncertainties and correlations: the system K₂O-Na₂O-CaO-MgO-MnO-FeO-Fe₂O₃-Al₂O₃-TiO₂-SiO₂-C-H₂-O₂. Journal of Metamorphic Geology, 8, 89-124.
•  Holland,  T.J.B. and  Powell,  R. (1998) An internally consistent thermodynamic data set for phases of petrological interest. Journal of Metamorphic Geology, 16, 309-343.
•  Ikeda,  T. (1993) Compositional zoning patterns of garnet during prograde metamorphism from the Yanai district, Ryoke metamorphic belt, southwest Japan. Lithos, 30, 109-121.
•  Ikeda,  T. (2004) Garnet-biotite geothermometry of a pelitic gneiss from the Lützow-Holm Complex in Skallen, East Antarctica: Constraints on retrograde metamorphism. Polar Geoscience, 17, 45-56.
•  Iwamura,  S.,  Tsunogae,  T.,  Kato,  M.,  Koizumi,  T. and  Dunkley,  D.J. (2013) Petrology and phase equilibrium modeling of spinel-sapphirine-bearing mafic granulite from Akarui Point, Lützow-Holm Complex, East Antarctica: Implications for the P-T path. Journal of Mineralogical and Petrological Sciences, 108, 345-350.
•  Kawakami,  T. and  Motoyoshi,  Y. (2004) Timing of attainment of spinel + quartz coexistence in the garnet-sillimanite leucogneiss from Skallevikshalsen, Lützow-Holm Complex, East Antarctica. Journal of Mineralogical and Petrological Sciences, 99, 311-319.
•  Kawakami,  T.,  Grew,  E.S.,  Motoyoshi,  Y.,  Shearer,  C.K., et al (2008) Kornerupine sensu stricto associated with mafic and ultramafic rocks in the Lützow-Holm Complex at Akarui Point, East Antarctica: what is the source of boron? In Geodynamic Evolution of East Antarctica: A Key to the East-West Gondwana Connection (Satish-Kumar, M., Motoyoshi, Y., Osanai, Y., Hiroi, Y. and Shiraishi, K. Eds.). Geological Society, London, Special Publications, 308, 351-375.
•  Kawakami,  T. and  Hokada,  T. (2010) Linking P-T path with development of discontinuous phosphorus zoning in garnet during high-temperature metamorphism - an example from Lützow-Holm Complex, East Antarctica. Journal of Mineralogical and Petrological Sciences, 105, 175-186.
•  Kawakami,  T.,  Hokada,  T.,  Sakata,  S. and  Hirata,  T. (2016) Possible polymetamorphism and brine infiltration recorded in the garnet-sillimanite gneiss, Skallevikshalsen, Lützow-Holm Complex, East Antarctica. Journal of Mineralogical and Petrological Sciences, 111, 129-143.
•  Kawasaki,  T.,  Ishikawa,  M. and  Motoyoshi,  Y. (1993) A preliminary report on cordierite-bearing assemblages from Rundvågshetta, Lützow-Holm Bay, East Antarctica: evidence for a decompressional P-T path? Proceedings of the National Institute of Polar Research Symposium on Antarctic Geosciences, 6, 47-56.
•  Kawasaki,  T.,  Nakano,  N. and  Osanai,  Y. (2011) Osumilite and a spinel + quartz association in garnet-sillimanite gneiss from Rundvågshetta, Lützow-Holm Complex, East Antarctica. Gondwana Research, 19, 430-445.
• Lasaga, A.T. (1983) Geospeedometry: an extension of geothermometry. In Kinetics and Equilibrium in Mineral Reactions (Saxena, S.K. Ed.), Springer-Verlag, New York, 81-114.
•  Miyazaki,  K. (1991) Ostwald ripening of garnet in high P/T metamorphic rocks. Contributions to Mineralogy and Petrology, 108, 118-128.
•  Miyazaki,  K. (2015) Diffusion-controlled growth and degree of disequilibrium of garnet porphyroblasts: is diffusion-controlled growth of porphyroblasts common? Progress in Earth and Planetary Science, 2, 25.
•  Mori,  Y. and  Ikeda,  T. (2018) Formation of triple-layer coronas between corundum and hornblende from the Lützow-Holm Complex at Akarui Point, East Antarctica. Journal of Mineralogical and Petrological Sciences, 113, 68-81.
• Motoyoshi, Y. (1986) Prograde and progressive metamorphism of the granulite facies Lützow-Holm Bay region, East Antarctica. pp. 238, Doctoral Thesis, Hokkaido University.
•  Motoyoshi,  Y.,  Matsubara,  S.,  Matsueda,  H. and  Matsumoto,  Y. (1985) Garnet-sillimanite gneisses from the Lützow-Holm Bay region, East Antarctica. Memoirs of National Institute of Polar Research Special Issue, 37, 82-94.
• Motoyoshi, Y. and Ishikawa, M. (1997) Metamorphic and structural evolution of granulites from Rundvågshetta, Lützow-Holm Bay, East Antarctica. In The Antarctic Region: Geological Evolution and Processes (Ricci, C.A. Ed.). Terra Antarctica Publication, Siena, 65-72.
•  Nakamura,  A.,  Kitamura,  M. and  Kawakami,  T. (2013) Microstructural records of multiple retrograde local H₂O supplement in the pelitic gneiss, Lützow-Holm Complex at Akarui Point, East Antarctica. Mineralogy and Petrology, 108, 177-186.
• Osanai, Y., Toyoshima, T., Owada, M., Tsunogae, T., et al. (2004) Geological map of Skallen, Antarctica (Revised Edition). Antarctic Geological Map Series, Sheet 39 (with explanatory text 23p.). National Institute of Polar Research, Tokyo.
• Saltykov, S.A. (1967) The determination of the size distribution of particles in an opaque material from a measurement of the size distribution of their sections. In Stereology (Elias, H. Ed.). Proceedings of the Second International Congress for Stereology, Springer, New York, 163-173.
•  Santosh,  M. and  Yoshida,  M. (1992) A petrologic and fluid inclusion study of charnockites from the Lützow-Holm Bay region, East Antarctica: Evidence for fluid-rich metamorphism in the lower crust. Lithos, 29, 107-126.
•  Satish-Kumar,  M. and  Wada,  H. (2000) Carbon isotopic equilibrium between calcite and graphite in Skallen Marbles, East Antarctica: evidence for the preservation of peak metamorphic temperatures. Chemical Geology, 166, 173-182.
•  Satish-Kumar,  M.,  Hermann,  J.,  Tsunogae,  T. and  Osanai,  Y. (2006) Carbonation of Cl-rich scapolite boudins in Skallen, East Antarctica: evidence for changing fluid condition in the continental crust. Journal of Metamorphic Geology, 24, 241-261.
•  Shiraishi,  K.,  Hokada,  T.,  Fanning,  C.M.,  Misawa,  K., et al (2003) Timing of thermal events in eastern Dronning Maud Land, East Antarctica. Polar Geoscience, 16, 76-99.
•  Spear,  F. (1988) Metamorphic fractional crystallization and internal metasomatism by diffusional homogenization of zoned garnets. Contributions to Mineralogy and Petrology, 99, 507-517.
•  Spear,  F. (1991) On the interpretation of peak metamorphic temperature in light of garnet diffusion during cooling. Journal of Metamorphic Geology, 9, 379-388.
• Spear, F. (1993) Metamorphic Phase Equilibria and Pressure-Temperature-Time paths. Mineralogical Society of America Monograph. MSA, Washington, D. C.
•  Suzuki,  K. and  Kawakami,  T. (2019) Metamorphic pressure-temperature conditions of the Lützow-Holm Complex of East Antarctica deduced from Zr-in-rutile geothermometer and Al₂SiO₅ minerals enclosed in garnet. Journal of Mineralogical and Petrological Sciences, 114, 267-279.
•  Suzuki,  M. (1983) Preliminary note on the metamorphic conditions around Lützow-Holm Bay, East Antarctica. Memoirs of National Institute of Polar Research Special Issue, 28, 132-143.
•  Takamura,  Y.,  Tsunogae,  T. and  Tsutsumi,  Y. (2020) U-Pb geochronology and REE geochemistry of zircons in mafic granulites from Lützow-Holm complex, East Antarctica: Implications for the timing and P-T path of post-peak exhumation and Antarctica-Sri Lanka correlation. Precambrian Research, 348, 105850.
•  Tracy,  R.J.,  Robinson,  P. and  Thompson,  A.B. (1976) Garnet composition and zoning in the determination of temperature and pressure of metamorphism, central Massachusetts. American Mineralogist, 61, 762-775.
•  Yardley,  B.W.D. (1977) An empirical study of diffusion in garnet. American Mineralogist, 62, 793-800.
• Yardley, B.W.D. and Warren, C. (2021) An Introduction to Metamorphic Petrology. Second edition, Cambridge University Press.
• Yoshida, M., Yoshida, Y., Ando, H., Ishikawa, T. and Tatsumi, T. (1976) Geological map of Skallen, Antarctica. Antarctic Geological Map Series, Sheet 9 (with explanatory text 16p.). National Institute of Polar Research, Tokyo.
•  Yoshida,  M. and  Aikawa,  N. (1983) Petrography of a discordant metabasite from Skallen, Lützow-Holmbukta, East Antarctica. Memoirs of National Institute of Polar Research Special Issue, 28, 144-165.
•  Yoshimura,  Y.,  Motoyoshi,  Y. and  Miyamoto,  T. (2008) Sapphirine + quartz association in garnet: Implication for ultrahigh-temperature metamorphism in Rundvågshetta, Lützow-Holm Complex, East Antarctica. In Geodynamic Evolution of East Antarctica: A Key to the East-West Gondwana Connection (Satish-Kumar, M., Motoyoshi, Y., Osanai, Y., Hiroi, Y. and Shiraishi, K. Eds.). Geological Society, London, Special Publications, 308, 377-390.