Introduction

Garnet is an important constituent mineral in eclogite and other crustal, as well as upper mantle rocks (Baxter et al. 2013). The results of deformation experiments show that garnet can be deformed plastically at high-pressure and -temperature conditions, and its creep strength and brittle–ductile transition temperature are significantly higher than most of silicate minerals (Jin et al. 2001; Karato et al. 1995; Mei et al. 2010; Voegelé et al. 1998a; Wang and Ji 1999; Xu et al. 2013). Hence, the ductility of natural garnet provides information on the geodynamic processes in the lower crust and the upper mantle and, especially, in the subduction zones (Caddick and Kohn 2013; Ji and Martignole 1994; Voegelé et al. 1998b). However, as garnet is an isotropic mineral, it is difficult to examine its plastic deformation microstructures by optical microscope. Fortunately, the dislocation pattern of garnet can reflect its deformation conditions (Ando et al. 1993; Carstens 1969; Chen et al. 1996; Doukhan et al. 1994; Lü et al. 2004; Prior et al. 2000). To date, there are two commonly used methods to observe the dislocation microstructures of garnet: (1) transmission electron microscopy (TEM) for high-resolution analysis (Allen et al. 1987): TEM observations have some shortcomings on large area and in situ analysis of unevenly distributed dislocations (Prior et al. 2000; Voegelé et al. 1998b). (2) Etch pitting with hydrofluoric acid at the crystal surface (Carstens 1969, 1971; Voegelé et al. 1998a): this method is just suitable for Mg-rich garnet, which indicates that this etching reaction is controlled by chemical compositions of crystal (Ji and Martignole 1994). As the other method, the orientation contrast imaging with electron backscatter diffraction (EBSD) is effective for revealing subgrain boundaries, but it has not been proven to be useful in analyzing free dislocations (Massey et al. 2011; Smit et al. 2011). While the electron channeling contrast imaging (ECCI) with EBSD is a potentially powerful technique to analyze the free dislocations in garnet (Gutierrez-Urrutia and Raabe 2012; Kleinschrodt and McGrew 2000; Zhang et al. 2015), but it is an expensive and complex method. Therefore, the quantitative studies on the geodynamic implications of deformed garnets are still sparse (Ando et al. 1993; Chen et al. 1996; Lü et al. 2004). As the result, new method for observing the dislocation microstructures of garnet is needed.

The well-established method of oxidation decoration of dislocations in olivine (Jung et al. 2014; Karato 1987; Kohlstedt et al. 1976) encouraged us to explore a similar method applying the garnet grain. Furthermore, the crystal premelting, a process that melting preferentially occurs at the highly strained regions, such as dislocation lines and (sub)grain boundaries, below the thermodynamically calculated melting temperatures (Alsayed et al. 2005; Bartis 1977; Gómez et al. 2003; Hand and Dirks 1992; Mei and Lu 2007; Tumarkina et al. 2011), can be regarded as marks of mineral crystal defects (Levine et al. 2016). The previous studies found that the garnet end members of almandine and grossular would begin to decompose directly into melt with or without other crystalline phases present at ~ 800 °C under low-pressure conditions (Hsu 1968; Keesmann et al. 1971), even though their melting temperatures (Tm) are above 1000 °C (Karato et al. 1995). In this paper, we investigated the premelting phenomenon at the crystal defects of garnet at 1 atmosphere pressure and temperatures of 800–1000 °C, and propose that it is an effective new method for decorating the dislocation microstructures in garnet.

Sample description and experimental procedure

The garnet samples in this study were selected from one eclogite sample and one gem-quality megacrystal garnet. The eclogite sample Hcc4-1 was collected from the Shuanghe ultrahigh pressure (UHP) metamorphic terrane, which is part of the Dabie orogenic belt of China. There have been many detailed geological, petrological and geochemical studies about this area, which show that the age of peak metamorphism is 226–234 Ma at pressures of above 2.7 GPa and temperatures of 700–850 °C (Cong et al. 1995; Li et al. 2000; Liou et al. 1997; Liu et al. 2006; Okay 1993). The garnet from the eclogite sample is fresh and elongated substantially (Figs. 1a, 2a), which suggests that it underwent plastic deformation. The end members of almandine and grossular account for > 80 wt% of the studied garnet grains in which water content is up to ~ 2500 wt ppm H2O (Liu et al. 2016). The gem-quality megacrystal garnet Mgc1-1 was collected from Mozambique (Fig. 1b). We used both of our decoration method and TEM analysis to investigate the dislocation characteristics of garnet.

Fig. 1
figure 1

a Single-sided polished block of the eclogite sample Hcc4-1. b Double-sided polished section of the megacrystal garnet Mgc1-1

Full size image
Fig. 2
figure 2

a, b Photomicrographs (reflected light) of the same garnet grain’s polished surface in the eclogite block Hcc4-1 before and after cyclic heat treatment to 970 °C, respectively. The numerous dark decorated lines are observed only in b. Photomicrographs of the decorated lines in the same area of megacrystal garnet Mgc1-1 after cyclic heat treatment to 980 °C, taken by reflected light (c) and cross-polarized light (d)

Full size image

The experiments were carried out in the following steps: (1) the eclogite sample was cut into a ~ 4 × 3 × 1 cm3 block, which was highly polished on one side (Fig. 1a). The megacrystal garnet was prepared as a doubly polished, ~ 0.5-mm-thick section (Fig. 1b). (2) The microstructures on the polished side of the garnets were photographed by an optical microscope. (3) Since the decomposition temperature of almandine and grossular, which are the main end members of garnets from Shuanghe eclogite (Liu et al. 2016), is ~ 800 °C under low-pressure conditions (Hsu 1968; Keesmann et al. 1971), the prepared samples were heated using an open muffle furnace to 800 °C with a rate of 2°C/min. Then, the samples were cooled to room temperature just after reaching the setting temperatures. In addition, we also heated the eclogite sample Hcc4-1 in nitrogen atmosphere to check the effect of oxygen fugacity (\({f_{{{\text{O}}_{\text{2}}}}}\)) on the decorating phenomenon of garnet. (4) After the heat treatment at desired temperature, the polished surface was examined by an optical microscope. If the garnet grains do not show decorated lines, the target temperature of heat treatment will be increased by 10 °C. Such heating and cooling cycles would be continued until the decorated lines become easy to be identified under reflected light. (5) The decorated lines were investigated using a combination of SEM, LRS, and TEM at the State Key Laboratory of Geological Processes and Mineral Resources (GPMR).

The microstructures of heat-treated garnet grains were observed using a Quanta 450 FEG-SEM. The Raman spectra were obtained on a Renishaw RM-1000 LRS system. The 532-nm line of Ar+ laser at 20 mW was used for sample excitation (corresponding to 2 mW on the sample). Both the spectral resolution and the accuracy in the Raman shift were estimated to be ± 1 cm−1. The TEM samples were processed by a Gatan 656 dimple grinder, a Gatan-600 ion mill, and a JEOL-JEE4X vacuum evaporator for physical thinning, ion thinning and carbon coating, respectively. The chemical composition and microstructures of the decorated lines were determined by a Philips CM12 TEM, which was equipped with an EDAX PV9100 X-ray energy-dispersive spectroscope (EDS) and a Gatan 830 charge-coupled device (CCD) camera system. Observations were made at an accelerating voltage of 120 kV with a point-to-point resolution of 0.34 nm and a line resolution of 0.2 nm.

Experimental results

Decorated lines of garnet under the optical microscope

By comparing the photomicrographs taken by reflected light of each heat treatment experiment, we found that the first occurred temperatures of decorated lines on the polished surface of garnet grains are around 950 °C. As the heating temperature increases, the decorated lines become darker in color and wider in width and reach the best contrast with the matrix at 970–980 °C. As shown in Fig. 2b, when the heat treatment temperature reaches 970 °C, numerous dark decorated lines can be observed clearly on the polished surface of garnet grains in the eclogite block Hcc4-1 under reflected light. These decorated lines exhibit heterogeneous distribution characteristics, and most of them appear as small curves and loops distributing in the core area of the polished surface of garnet grains. In contrast, the decorated lines in the megacrystal garnet Mgc1-1 are short and straight, and their distribution is generally homogeneous (Fig. 2c, d).

Microstructures and compositions of the decorated lines

The microstructures and compositions of the decorated lines in the studied garnets were analyzed by SEM, LRS, and TEM. The decorated lines in the garnet grains are clear and bright in the backscattered electron image of SEM (Fig. 3a), indicating that their compositions are dominated by phases with high mean atomic number. The secondary electron image shows that the shape of these lines bulge out of the polished surface (Fig. 3b). This bulging nature could scatter the light, leading to a dark color appearance under the reflected light (Fig. 2b, c).

Fig. 3
figure 3

a Backscattered electron image and b secondary electron image of the decorated lines in the garnet from the eclogite sample Hcc4-1. The bright color and bulging nature of decorated lines strongly suggest that their compositions are dominated by the volume expanded phases with high mean atomic numbers

Full size image

As shown in Fig. 4 and Table 1, compared with the garnet before heat treatment, the Raman spectra indicate that there are some new phases generating in the heat-treated garnet grains. The peaks around 351, 502, 552 and 906 cm−1 in all spectra are assigned to a solid solution of almandine and grossular (Gillet et al. 1992). The Raman spectrum of decorated line on the polished surface of garnet heated in air (Fig. 4a) indicates that it is a mixture of iron oxides and garnet. The peaks at 224, 243, 291, 409, 502 and 615 cm−1 agree well with the reported data of hematite (De Faria and Lopes 2007; De Faria et al. 1997). The peak at 667 cm−1 is assigned to the A1g mode of magnetite (Shebanova and Lazor 2003; Thibeau et al. 1978). But the spectrum of decorated line in the interior of garnet after heated in air (Fig. 4b) shows that there is no hematite composition and the A1g mode peak of magnetite has shifted to 689 cm−1 (D’Ippolito et al. 2015; Verble 1974), which is similar to the spectrum of decorated line on the polished surface of garnet heated in nitrogen atmosphere (Fig. 4c).

Fig. 4
figure 4

Raman spectra of garnet in the eclogite sample Hcc4-1 before and after heat treatment. a, b The spectrum of decorated line on the polished surface and in the interior of garnet after heated in air, respectively. c The spectrum of decorated line on the polished surface of garnet after heated in nitrogen atmosphere. d The spectrum of garnet before heat treatment. Hem hematite, Mag magnetite and Grt garnet

Full size image
Table 1 Raman mode frequencies (cm−1) of garnet in the eclogite sample Hcc4-1 before and after heat treatment
Full size table

The dislocation microstructures in garnet within eclogite samples from the same studied area have been analyzed using TEM (Xie et al. 2018), and the results show that the free dislocations and subgrain boundaries (e.g., dislocation arrays, networks, and walls) are widely distributed in garnet grains. However, the present TEM study shows that the microstructures in the interior of heated garnet grains from the eclogite sample Hcc4-1 have been changed. We have not observed the typical free dislocation and subgrain boundary microstructures in the heated garnet grains, but only found numerous small decorated lines exhibiting as pseudomorph after dislocation and subgrain boundary textures. At the middle of decorated lines, numerous microcrystals are assembled as a narrow belt with accompanying gray belts on both sides (Fig. 5a). The selected area electron diffraction patterns and the X-ray energy-dispersive spectra (Fig. 5b) from these microcrystals and gray belts suggest that they are Al-bearing magnetite and Si- and Ca-rich amorphous phase (AP), respectively. The distribution features of these amorphous phases suggest that they are the quenched melt after heat treatment, which is the evidence of partial melting occurring in the heated garnet. Since our garnet was heated below its melting temperature, this partial melting should belong to the premelting phenomenon (Alsayed et al. 2005; Bartis 1977). The presence of melt with larger volume might be the reason why the shape of decorated lines bulges out of the polished surface (Fig. 3b).

Fig. 5
figure 5

a TEM micrograph of the decorated line in the interior of garnet from the eclogite sample Hcc4-1. b The selected area electron diffraction patterns and X-ray energy-dispersive spectra from the dotted encircled regions in a. The peaks of Cu in b are caused by the TEM specimen holder. Al-Mag Al-bearing magnetite and AP amorphous phase

Full size image

Dislocation density of the megacrystal garnet

Based on the distribution and morphology of the decorated lines in the megacrystal garnet Mgc1-1 (Fig. 2c, d), we infer that these lines are decorated dislocations. The density of these decorated dislocation lines, like that analyzed by TEM, can be simply estimated by the total number (N) of dislocations in the measured area (A) (Ando et al. 1993; Yamamoto et al. 2017). The estimated result shows that the dislocation density in Fig. 2c is 1.9 × 104 cm−2 (Table 2). On the other hand, since the true definition of dislocation density is the total length (L) of all dislocations in a unit volume (V) of a crystalline material, we calculated the sample volume corresponding to Fig. 2d based on its area and the thickness of this heated megacrystal garnet section and measured the total length of its decorated dislocations using the ImageJ image-process software (Jung and Karato 2001; Karato and Jung 2003). Based on this calculation method, we obtained a dislocation density of 1.7 × 104 cm−2 in Fig. 2d, which is almost equal to the value based on Fig. 2c. To compare the dislocation microstructures revealed by heat treatment, we also investigated the unheated part of megacrystal garnet Mgc1-1 by TEM. As shown in Fig. 6 and Table 2, only a small number of straight dislocation lines were observed in the TEM specimen yielding a dislocation density of 6.3 × 104 cm−2. Though this value is somewhat higher than that obtained from the optical photomicrographs, they are in the same order of magnitude.

Table 2 Dislocation densities in megacrystal garnet
Full size table
Fig. 6
figure 6

Dislocation microstructures in the unheated part of the megacrystal garnet

Full size image

Discussion

Premelting decoration mechanism

Based on the results of SEM, Raman, and TEM analyses, we confirm that the decorated lines observed on the polished surface of heated garnet grains are decorated dislocations or (sub)grain boundaries by premelting reaction. This dislocation decoration mechanism, which can be termed as premelting decoration (PMD), is different from that of oxidation decoration in olivine (Jung et al. 2014; Karato 1987; Kohlstedt et al. 1976).

As illustrated schematically in Fig. 7a, the premelting reaction at crystal defects in garnet under high temperatures can be divided into four stages: (1) before premelting, the crystal defects, such as dislocation lines, exist in garnet grains. (2) Premelting, at high temperatures, the premelting reaction preferentially initiates at the core of dislocations, which causes the crystal around dislocations to decompose directly into melt (Alsayed et al. 2005; Bartis 1977; Hand and Dirks 1992; Mei and Lu 2007; Tumarkina et al. 2011). (3) Nucleation, iron oxide minerals nucleate once the garnet premelts. As shown in Fig. 7b, the new products generated by premelting reaction are hematite, magnetite and melt on the polished surface of garnet grains after heated in air. While in the interior of these garnet grains, the new phases are Al-bearing magnetite and melt, which is similar to that on the polished surface of garnet heated in nitrogen atmosphere (Fig. 4). That means the Fe3+ content of the premelting products gradually decreases towards the interior of sample, which should be ascribed to the slow oxygen diffusivity (Hsu 1968; Meyers et al. 1980). The presence of Al in magnetite within the garnet interior may account for the shift of its A1g mode peak position from 667 to 689 cm−1 (Fig. 4a, b), since the atomic mass of Al3+ is lighter than Fe3+ (D’Ippolito et al. 2015). The new phases reported here differ somewhat from those reported by Hsu (1968), in which the reaction products are mainly composed of solid solution of magnetite + hercynite and quartz. This difference might be due to the fact that our experiments were carried out at one atmosphere pressure without dwelling at desired temperature, whereas those of Hsu (1968) were performed under high \({f_{{{\text{O}}_{\text{2}}}}}\), below 1 kbar and approximately 900 °C conditions for dozens of hours, which may lead to the incomplete crystallization of quartz. (4) Crystal growth, as temperature increases, the premelting melt belt becomes broader and the crystal size of iron oxide minerals grows bigger. These microcrystals and melts could be preserved and quenched, respectively, during the cooling process of heat treatment, which is the formation mechanism of dislocation pseudomorph textures (Fig. 5a).

Fig. 7
figure 7

Schematic diagram of the premelting reaction at crystal defects in garnet. a Four stages of the premelting reaction at crystal defects under high temperatures. b The distribution and composition of the premelting lines in garnet

Full size image

Reliability of the premelting decoration method

Previous studies suggest that the decomposition temperatures of silicate garnet range from 750 to 1200 °C, which could change with the factors of chemical composition, temperature, pressure, oxygen fugacity, water content, etc. (Aparicio et al. 2012, Hsu 1968; Keesmann et al. 1971). Therefore, it is very difficult to estimate the optimum temperature for premelting reaction of garnet. However, this is exactly the key point for using the PMD method to reveal dislocation microstructures in garnet. In the present study, the first target temperature of heat treatment is 800 °C, which is the decomposition temperature of the main end members of our garnet (Hsu 1968; Keesmann et al. 1971; Liu et al. 2016). As temperature increases, the premelting decoration lines first occurred at 950 °C and got the best contrast with the matrix at 970–980 °C. Therefore, it can be concluded that the heat treatment by stepwise increasing the target temperature from the decomposition temperature is the ideal procedure for revealing dislocations in garnet by PMD method.

Due to the heterogeneity of dislocation distribution, the larger the measured area is, the closer to the true value of dislocation density will be. In this study, the dislocation density of megacrystal garnet Mgc1-1obtained by PMD is similar to that analyzed by TEM (Table 2), which indicates that the PMD is a new reliable method for revealing dislocations in garnet. Since the measured area or volume using PMD is much larger than using TEM, we infer that the data obtained by PMD is closer to the true value of dislocation density in this sample. Compared with TEM and ECCI, the PMD method is relatively simple, inexpensive and rapid. However, this method, like etch pitting, is not suitable for analyzing the sample with high dislocation densities (> 108 cm−2) due the overlap of premelting lines.

Geological implications

The temperature for brittle–ductile transition of silicate garnet is above 850 °C (T > 0.65 Tm) in nature, which is so high that its dislocation pattern could reflect deformation information in the deep earth (Wang and Ji 1999). Thus, the dislocation pattern in deformed garnet is regarded as a high-temperature deformation indicator (Ji and Martignole 1994; Voegelé et al. 1998a) and a differential stress piezometer (Ando et al. 1993; Chen et al. 1996; Lü et al. 2004). The PMD method presented here could overcome the shortcomings of TEM, such as on large area and in situ analysis, and provide more reliable data on the dislocation density of garnet to relate its corresponding chemical and physical properties, which is important for understanding the plastic deformation mechanism of garnet and the associated geodynamic process (Chen et al. 1996; Karato et al. 1995; Voegelé et al. 1998b). Since the new dislocation density piezometer of olivine has been established by Karato and Jung (2003) based on the oxidation decoration method, we can use our PMD method to recalibrate the differential stress piezometer of garnet which is established by Ando et al. (1993).

Conclusions

The PMD method provides a new approach for observing dislocation microstructures in garnet by optical microscope. Compared with TEM, this approach greatly reduces the cost and time involved in examining large areas of dislocation microstructures in garnet. This method could be used to investigate the correlations between dislocation microstructure and grain size, water content and chemical composition, and it will help to establish a more practical dislocation density differential stress piezometer.