Microstructural Evolution of Dy2O3-TiO2 Powder Mixtures during Ball Milling and Post-Milled Annealing

The microstructural evolution of Dy2O3-TiO2 powder mixtures during ball milling and post-milled annealing was investigated using XRD, SEM, TEM, and DSC. At high ball-milling rotation speeds, the mixtures were fined, homogenized, nanocrystallized, and later completely amorphized, and the transformation of Dy2O3 from the cubic to the monoclinic crystal structure was observed. The amorphous transformation resulted from monoclinic Dy2O3, not from cubic Dy2O3. However, at low ball-milling rotation speeds, the mixtures were only fined and homogenized. An intermediate phase with a similar crystal structure to that of cubic Dy2TiO5 was detected in the amorphous mixtures annealed from 800 to 1000 °C, which was a metastable phase that transformed to orthorhombic Dy2TiO5 when the annealing temperature was above 1050 °C. However, at the same annealing temperatures, pyrochlore Dy2Ti2O7 initially formed and subsequently reacted with the remaining Dy2O3 to form orthorhombic Dy2TiO5 in the homogenous mixtures. The evolutionary mechanism of powder mixtures during ball milling and subsequent annealing was analyzed.


Introduction
High-energy ball milling has been widely used to prepare various types of materials, such as supersaturated solid solutions, metastable crystalline materials [1], quasicrystal phases [2], nanostructured materials [3], and amorphous alloys [4]. The technology was initially used in place of blending and sintering at elevated temperatures to prepare ceramic-strengthened alloys [5,6]. A large amount of mechanical energy is transformed into intrinsic energy in the target materials, which induces the formation of numerous defects in the crystal structure, such as vacancies, interstitials, cavities and dislocations, which are always in a non-equilibrium state [7][8][9]. The defects and structural disorders will increase the mobility of atomic diffusion and induce chemical reactions amongst components that are not present under equilibrium conditions [10]. Therefore, based on its excellent characteristics, ball milling was used to prepare bulk Dy 2 TiO 5 , which can be used as a neutron absorber in control rods in nuclear power plants. Control rods are very important in both operating and accident conditions because the nucleon reactivity must be controlled in order to safely operate a nuclear reactor [11]. In fact, bulk Dy 2 TiO 5 prepared by ball milling and sintering has been used in Russian power plant water reactors, such as MIR and VVER-1000 RCCAs [12,13], because of the excellent nucleon characteristics of the element dysprosium, as natural dysprosium consists of five stable isotopes with high thermal neutron absorption cross sections. The decay products are Ho and Er, which are also able to absorb neutrons. All of the radionuclides have low gamma activity and short half-life periods. The absorption ↔ cubic [15], which have different physical properties and radiation resistance abilities. In fact, bulk Dy 2 TiO 5 in the cubic crystal structure has the lowest neutron irradiation swelling and highest irradiation resistance. Therefore, it is necessary to synthesize bulk Dy 2 TiO 5 in the cubic crystal structure. Jung [16] synthesized bulk Dy 2 TiO 5 with high purity and density using a polymer carrier chemical synthesis process, in which ethylene glycol was used as an organic carrier for metal cations. An amorphous phase was detected below 800 • C, and orthorhombic Dy 2 TiO 5 was observed after sintering for 1 h at 1300 • C, while little else was observed while sintering in the range of 800 to 1300 • C. Panneerselvam [17] used both solid-state synthesis and wet chemical synthesis to prepare Dy 2 TiO 5 . However, the effect of sintering temperature on the phase evolution needs further investigation. The sinterability of Dy 2 O 3 and TiO 2 with different molar ratios was determined for various ball-milling and sintering conditions [18]. Amit Sinha [19] reported the synthesis of bulk Dy 2 TiO 5 from mixtures of equimolar Dy 2 O 3 and TiO 2 powders in a two-step process: (I) pyrochlore Dy 2 Ti 2 O 7 was initially formed and (II) Dy 2 Ti 2 O 7 then reacted with the remaining Dy 2 O 3 to form orthorhombic Dy 2 TiO 5 . The powder mixtures used in sintering were simply mixed during ball milling. Garcia-Martinez [7] observed the experimental phenomena of the transformation of Dy 2 O 3 from cubic to monoclinic and the synthesis of a hexagonal high-temperature phase, reported as Dy 2 TiO 5 , in an equimolar Dy 2 O 3 -TiO 2 mixture during ball milling. Therefore, further investigation is needed into the evolutionary behavior of the microstructure under different ball-milling conditions and the effect of the state of the ball-milled powder on the sintering behavior. In the present work, the microstructural evolutionary behavior and corresponding reaction mechanism of Dy 2 O 3 -TiO 2 powder mixtures under two types of ball-milling parameters were investigated. The annealing behavior of the ball-milled mixtures was also examined.

Experiments
Powders of Dy 2 O 3 (cubic crystal structure) and TiO 2 (rutile crystal structure) with an average particle diameter of 5 µm and 50 nm, respectively, were used as raw materials. The raw powders of Dy 2 O 3 and TiO 2 were purchased from Beijing HWRK Chem Co., Ltd. (Beijing, China). The purity of the raw powders of both Dy 2 O 3 and TiO 2 was 99.9%. Ball milling of the molar fraction Dy 2 O 3 -50% TiO 2 (Dy 2 O 3 :TiO 2 = 1:1) powder mixtures was carried out on an SFM-1 high-energy planetary ball mill at room temperature. Stainless steel balls that were 5 mm in diameter were used as the milling media. The ball-to-powder mass ratio was 10:1, and the rotational speed was 200 rpm and 500 rpm. No more than one weight percent stearic acid was added to the powder mixtures as a process control agent to prevent excessive cold welding and aggregation amongst the powder particles. During ball milling, a 5-min stopping interval was used after milling for 55 min to prevent excess heat generation, which has an obvious effect on the ball-milling procedure. The powder mixtures used for microstructural analysis were extracted from the loose powders in the steel can, not from powders adhered to the surface of the stainless balls or the steel can wall, after ball milling for 4, 12, 24, 48, and 96 h.
After various milling times, a small amount of ball-milled powders taken from the container were characterized and analyzed by X-ray diffraction (XRD) on a Rigaku Ultima IV X-ray diffractometer (Rigaku, Tokyo, Japan) with Cu Kα radiation (λ = 0.1540598 nm) and transmission electron microscopy (TEM) on a JEM-2100 instrument (JEOL, Tokyo, Japan). Analysis was also carried out for ball-milled powder mixtures annealed at different temperatures.
The grain size was calculated using Suryanarayana and Grant Norton's formula [20]. where, K is a constant (with a value of 0.9); λ is the wavelength of the X-ray radiation; L and η are the grain size and internal strain, respectively; and θ is the Bragg angle. B r is the full width at half-maximum (FWHM) of the diffraction peak after instrumental correction and can be calculated from the following equation: where, B and B S are the FWHM of the broadened Bragg peaks and the standard sample's Bragg peaks, respectively. The ball-milled mixtures and annealed mixtures were first put in ethyl alcohol, and then adequately dispersed by ultrasonic vibration. A carbon-coated copper grid was used to collect the dispersed powders in the ethyl alcohol and then dried by ultraviolet lamp. After that, the prepared samples were observed by TEM. Differential scanning calorimetry (DSC) was used to analyze the thermal behavior of ball-milled powders at a 5 • C/min heating rate in argon atmosphere using a SAT 449C instrument (NETZSCH, Bavarian State, Germany ). The powder mixtures milled for 96 h were annealed at temperatures ranging from 700 to 1150 • C in a tube furnace under atmospheric conditions. The heating and cooling rates were both 5 • C/min.

Results and Discussion
The XRD patterns of Dy 2 O 3 -TiO 2 powder mixtures milled at 500 rpm and 200 rpm for different times are shown in Figure 1. The XRD results show that the crystal structure of the original Dy 2 O 3 phase and TiO 2 phase are cubic and rutile, respectively. At the condition of 500 rpm, the diffraction peaks of cubic Dy 2 O 3 and TiO 2 broadened significantly and reduced in intensity with increased milling time. The broadening of the X-ray diffraction peaks is associated with the refinement in grain size and lattice distortions. Meanwhile, the diffraction peaks of monoclinic Dy 2 O 3 can be observed in the X-ray patterns as indicated by the black inverted triangles in Figure 1a. Ball milling induces a Dy 2 O 3 phase transformation from the cubic to the monoclinic crystal structure. A broad, singular diffraction peak is also present that indicates the formation of the amorphous phase during ball milling. Interestingly, the amorphous peak is present at the location of the diffraction peak for the monoclinic Dy 2 O 3 phase, not at the location of the diffraction peak for the cubic Dy 2 O 3 phase, which indicates that the formed amorphous phase is derived from the monoclinic Dy 2 O 3 phase, not from the cubic Dy 2 O 3 phase. The transformation from cubic to monoclinic increases with increased milling time. After milling for 96 h, only the amorphous phase can be observed, which indicates that the monoclinic Dy 2 O 3 phase was fully converted to the amorphous phase. Additionally, this behavior indicates that no new compounds are synthesized during ball milling. Even if new compounds were formed in the milled powders, the amount is very low and does not reach the sensitivity range of the X-ray measurement. Therefore, in the present work, the evolution of Dy 2 O 3 -TiO 2 powder mixtures is as follows: ball milling first induces the transformation of Dy 2 O 3 from the cubic to the monoclinic crystal phase, then monoclinic Dy 2 O 3 undergoes amorphization, and finally the powder mixtures completely transform to the amorphous phase.
However, the ball-milling behavior of powder mixtures at 200 rpm is distinctly different from that at 500 rpm. The change of the diffraction peaks with increased milling time at 200 rpm is shown in Figure 1b. Although the diffraction peaks of cubic Dy 2 O 3 and TiO 2 are also broadened and reduced in intensity with increased milling time, the diffraction peaks of TiO 2 can be observed in the XRD spectrums and are not disappeared. After ball milling for 96 h, the intensity of diffraction peaks of Dy 2 O 3 and TiO 2 are also high. The powder mixtures are not changed completely to amorphization. According to the shape of XRD diffraction spectrums, the effect of ball milling on powder mixtures after milling for 96 h at 200 rpm is only similar to that after milling for 4 h at 500 rpm. Therefore, at low ball-milling rotation speeds, the powder mixtures are only fined and homogenized.
Our experimental results are different from the results of G. Garcia-Martinez [7]. In their research, ball milling induced a phase transformation in Dy 2 O 3 from cubic to monoclinic. However, a Dy 2 TiO 5 compound with a hexagonal crystal structure was formed simultaneously. The ball-milled powders consisted of mixed phases of hexagonal Dy 2 TiO 5 and monoclinic Dy 2 O 3 . The Dy 2 O 3 -TiO 2 powder mixtures did not completely transform to the amorphous phase, but instead produced the hexagonal Dy 2 TiO 5 phase. This difference can be attributed to the different ball-milling conditions used in this research, with special attention to the different ball-milling facilities. During ball milling, experimental parameters such as rotation speed, ball-milling time, ball-milling media, and the ball-to-powder mass ratio have an important influence on the ball-milled products even when using the same proportion and type of oxides. For example, Gajović reported that nanosized ZrTiO 4 formed in ZrO 2 -TiO 2 powder mixtures, whereas only amorphous mixtures were obtained during ball milling in Stubičar's work [21,22]. In addition, the polymorphic transformation of Ln 2 O 3 was also observed in Gd 2 O 3 -TiO 2 and Y 2 O 3 -2TiO 2 powder systems [23]. powder mixtures did not completely transform to the amorphous phase, but instead produced the hexagonal Dy2TiO5 phase. This difference can be attributed to the different ball-milling conditions used in this research, with special attention to the different ball-milling facilities. During ball milling, experimental parameters such as rotation speed, ball-milling time, ball-milling media, and the ballto-powder mass ratio have an important influence on the ball-milled products even when using the same proportion and type of oxides. For example, Gajović reported that nanosized ZrTiO4 formed in ZrO2-TiO2 powder mixtures, whereas only amorphous mixtures were obtained during ball milling in Stubičar's work [21,22]. In addition, the polymorphic transformation of Ln2O3 was also observed in Gd2O3-TiO2 and Y2O3-2TiO2 powder systems [23]. The variation of Dy2O3 grain size with ball-milling time at the rotational speeds of 500 rpm and 200 rpm is shown in Figure 2. Actually, the size of Dy2O3 grain was calculated for Dy2O3 with cubic structure, not for Dy2O3 with monoclinic structure, because ball milling induced a phase transformation in Dy2O3 from cubic to monoclinic and simultaneously from monoclinic to amorphous. It is difficult to calculate the grain size of Dy2O3 with monoclinic structure. It can be seen that ball milling results in a fast decrease of Dy2O3 grain size in the initial stage at both 500 rpm and 200 rpm. The refinement rate of crystallite size is roughly logarithmic with ball-milling time at 200 rpm. After 96 h of ball milling, the size of Dy2O3 grain is up to approximately 60 nm. However, at 500 rpm, the diffraction peaks of Dy2O3 with cubic structure are hardly observed in the XRD spectrum after milling for 24 h as shown in Figure 1a, especially, when the ball-milling time is over 48 h. Therefore, the size of Dy2O3 with cubic structure is calculated only before 12 h of ball-milling time. It can be seen that the grain size is quickly decreased. In addition, after same ball-milling time, the grain size of Dy2O3 phase at the 500 rpm is obviously smaller than that at the 200 rpm. The effect of ball milling on the grain refinement of powder mixtures at 500 rpm is significantly more intense than that at 200 rpm. The size of Dy2O3 grain in the powder mixtures after milling for 4 h at 500 rpm is about 52 nm, which is smaller than that after milling for 96 h at 200 rpm (approximately 60 nm). The variation of Dy 2 O 3 grain size with ball-milling time at the rotational speeds of 500 rpm and 200 rpm is shown in Figure 2. Actually, the size of Dy 2 O 3 grain was calculated for Dy 2 O 3 with cubic structure, not for Dy 2 O 3 with monoclinic structure, because ball milling induced a phase transformation in Dy 2 O 3 from cubic to monoclinic and simultaneously from monoclinic to amorphous. It is difficult to calculate the grain size of Dy 2 O 3 with monoclinic structure. It can be seen that ball milling results in a fast decrease of Dy 2 O 3 grain size in the initial stage at both 500 rpm and 200 rpm. The refinement rate of crystallite size is roughly logarithmic with ball-milling time at 200 rpm. After 96 h of ball milling, the size of Dy 2 O 3 grain is up to approximately 60 nm. However, at 500 rpm, the diffraction peaks of Dy 2 O 3 with cubic structure are hardly observed in the XRD spectrum after milling for 24 h as shown in Figure 1a, especially, when the ball-milling time is over 48 h. Therefore, the size of Dy 2 O 3 with cubic structure is calculated only before 12 h of ball-milling time. It can be seen that the grain size is quickly decreased. In addition, after same ball-milling time, the grain size of Dy 2 O 3 phase at the 500 rpm is obviously smaller than that at the 200 rpm. The effect of ball milling on the grain refinement of powder mixtures at 500 rpm is significantly more intense than that at 200 rpm. The size of Dy 2 O 3 grain in the powder mixtures after milling for 4 h at 500 rpm is about 52 nm, which is smaller than that after milling for 96 h at 200 rpm (approximately 60 nm).
The morphology evolution of Dy 2 O 3 -TiO 2 powder mixtures with increasing ball-milling time at 500 rpm is shown in Figure 3. Both TiO 2 and Dy 2 O 3 are brittle components, which are fragmented during ball milling and particle size reduces continuously as a consequence of the energy provided during ball milling. The morphology of large particles is changed significantly due to fracture, agglomeration, and deagglomeration processes. components from the variation of the elemental diffraction intensity. The brittle Dy 2 O 3 particles are fragmented by ball-powder-ball collisions, leading to a considerable reduction in the powder particle size and subsequent amorphization as milling time increases. After ball milling for 4 h, the size of particles decreases significantly. A large number of small size of Dy 2 O 3 particles in nanometer can be observed in the milled powders as shown in Figure 3c. The morphology of the powder mixtures is transformed to uniform, as shown in Figure 3c-g, where the ball-milling time ranges from 4 to 96 h. The morphologies demonstrate that the refining effects of the powder particles are proportional to the ball-milling time for the same rotational speed. After 96 h of ball milling, a large number of nanoparticles agglomerate to form a large-sized particle, as shown in Figure 3g. In addition, TiO 2 particles disappear after ball milling for 96 h, as shown in Figure 1a. The surfaces of the Dy 2 O 3 particles in Figure 3g are clean compared with those in Figure 3a. It can be concluded that the particle size in the powder mixtures is refined to the nanoscale after ball milling for 96 h. The line-scanning results of elemental Dy, Ti, and O in the characteristic position in Figure 3g is shown in Figure 3h and also inserted in Figure 3g The morphology evolution of Dy2O3-TiO2 powder mixtures with increasing ball-milling time at 500 rpm is shown in Figure 3. Both TiO2 and Dy2O3 are brittle components, which are fragmented during ball milling and particle size reduces continuously as a consequence of the energy provided during ball milling. The morphology of large particles is changed significantly due to fracture, agglomeration, and deagglomeration processes. The morphology of the original powder mixtures consists of large-sized Dy2O3 particles in micrometer size and small-sized TiO2 particles in nanometer. The shape of the powder particles is irregular. The line-scanning results of elemental Dy, Ti, and O in the characteristic position in Figure 3a are shown in Figure 3b and also inserted in Figure 3a. It can be seen that the small-sized particles are TiO2 component and the large-sized particles are Dy2O3 components from the variation of the elemental diffraction intensity. The brittle Dy2O3 particles are fragmented by ball-powder-ball collisions, leading to a considerable reduction in the powder particle size and subsequent amorphization as milling time increases. After ball milling for 4 h, the size of particles decreases significantly. A large number of small size of Dy2O3 particles in nanometer can be observed in the milled powders as shown in Figure 3c. The morphology of the powder mixtures is transformed to uniform, as shown in Figure 3c-g, where the ball-milling time ranges from 4 to 96 h. The morphologies demonstrate that the refining effects of the powder particles are proportional to the ball-milling time for the same rotational speed. After 96 h of ball milling, a large number of nanoparticles agglomerate to form a large-sized particle, as shown in Figure 3g. In addition, TiO2 particles disappear after ball milling for 96 h, as shown in Figure 1a. The surfaces of the Dy2O3 particles in Figure 3g are clean compared with those in Figure 3a. It can be concluded that the particle size in the powder mixtures is refined to the nanoscale after ball milling for 96 h. The line-scanning results of elemental Dy, Ti, and O in the characteristic position in Figure 3g is shown in Figure 3h and also inserted in Figure 3g, which indicates these elements are uniformly distributed in the ball-milled   Figure 4a, due to the high active surface energy created upon ball milling. The size of the original TiO2 particles is approximately 50 nm. From the XRD results, it can be observed that ball milling leads to TiO2 particle refinement, dissolution, and finally disappearance after 96 h. Therefore, the main particles presented  Figure 4a, due to the high active surface energy created upon ball milling. The size of the original TiO 2 particles is approximately 50 nm. From the XRD results, it can be observed that ball milling leads to TiO 2 particle refinement, dissolution, and finally disappearance after 96 h. Therefore, the main particles presented in the TEM image are Dy 2 O 3 . The bright zones near the edge of the powder particles are the thin areas where the electron beam penetrates. The dark areas in the images of the powder particles are the thick areas where the electron beam rarely penetrates. Indexing and analyzing the ring-shaped SAED pattern taken from the area denoted by the letter "A" indicates that the Dy 2 O 3 grains are already nanocrystalline. The diffraction spots coming from cubic Dy 2 O 3 , monoclinic Dy 2 O 3 , and TiO 2 grains are present in the SAED pattern in Figure 4b. The diffraction halo in the SAED pattern also indicates the formation of an amorphous phase during high-energy ball milling.  Monoclinic Ln2O3 is initially formed in the mechanical alloying of lanthanum titanate or dititanate. In Moreno's research, the formation of Gd2(Ti(1−y)Zry)2O7 pyrochlores occurred in the final step of ball milling starting from an amorphous matrix of Gd2O3, TiO2, and ZrO2 [23]. However, in the present work, after 96 h of ball milling, the monoclinic Dy2O3 phase could not be detected and was completely transformed to the amorphous phase. Moreover, dysprosium titanate also could not be detected. To investigate the sintering behavior of the ball-milled powder mixtures, DSC was carried out on the powder mixtures milled for 96 h at test temperatures ranging from 200 to 1200 °C. The DSC curve in Figure 5 shows one exothermic peak close to 880 °C and one endothermic peak close to 1145 °C. An exothermic peak in a DSC curve can generally be attributed to a transition from disordered to ordered, the recrystallization of original components from the amorphous phase or the formation of a new compound from the ball-milled amorphous powders. Therefore, subsequent Xray analysis of the ball-milled powders annealed at different temperatures is used to further analyze occurrences in the heating process in more detail. The powder mixtures transform completely to the amorphous state after ball milling for 96 h. Therefore, it seems feasible that the transition from disordered to ordered produced the exothermic peak in the DSC curve. In fact, only the amorphous peak is observed in the XRD pattern of the 96 h ball-milled powder after annealing for 24 h at 700 °C. No diffraction peaks for Dy2O3 or TiO2 are detected. Even with prolonged annealing time, the XRD results are the same. Therefore, the exothermic peak in the DSC curve should be related to the new phase generated from the amorphous mixtures. After ball milling for 96 h, the ball-milled powders agglomerate to form large particles with large thicknesses such that the electron beam rarely penetrates, showing as a dark color in the TEM image in Figure 4c. It is difficult to observe the microstructure of the agglomerated particles. The SAED pattern from the area marked with the letter "B" indicates that the powder mixtures are almost completely converted to the amorphous phase, although sporadic diffraction spots are also present in this pattern. The atom arrangement in the amorphous phase is disordered over long distances, but ordered over short distances. As grain size decreases, the number of atoms at the grain boundaries increases. The proportion of atoms in the crystal volume relative to the crystal boundary decreases. Schwarz and Koch noted that the formation of an amorphous phase in as-milled powders was similar to the amorphization that occurs during the isothermal annealing of crystalline metallic thin films [24]. In high-energy ball milling, the intense deformation accelerates interdiffusion, and the large defect density increases the free energy of the components in the mixture to form an amorphous product.
Monoclinic Ln 2 O 3 is initially formed in the mechanical alloying of lanthanum titanate or dititanate. In Moreno's research, the formation of Gd 2 (Ti (1−y) Zr y ) 2 O 7 pyrochlores occurred in the final step of ball  [23]. However, in the present work, after 96 h of ball milling, the monoclinic Dy 2 O 3 phase could not be detected and was completely transformed to the amorphous phase. Moreover, dysprosium titanate also could not be detected. To investigate the sintering behavior of the ball-milled powder mixtures, DSC was carried out on the powder mixtures milled for 96 h at test temperatures ranging from 200 to 1200 • C. The DSC curve in Figure 5 shows one exothermic peak close to 880 • C and one endothermic peak close to 1145 • C. An exothermic peak in a DSC curve can generally be attributed to a transition from disordered to ordered, the recrystallization of original components from the amorphous phase or the formation of a new compound from the ball-milled amorphous powders. Therefore, subsequent X-ray analysis of the ball-milled powders annealed at different temperatures is used to further analyze occurrences in the heating process in more detail. The powder mixtures transform completely to the amorphous state after ball milling for 96 h. Therefore, it seems feasible that the transition from disordered to ordered produced the exothermic peak in the DSC curve. In fact, only the amorphous peak is observed in the XRD pattern of the 96 h ball-milled powder after annealing for 24 h at 700 • C. No diffraction peaks for Dy 2 O 3 or TiO 2 are detected. Even with prolonged annealing time, the XRD results are the same. Therefore, the exothermic peak in the DSC curve should be related to the new phase generated from the amorphous mixtures. The powder mixtures milled for 96 h were annealed for 3 h at 800, 900, 1000, 1050, 1100, and 1150 °C. The XRD results of the annealed powder mixtures are shown in Figure 6a,b. Several diffraction peaks different from the diffraction peaks of Dy2O3 (cubic and monoclinic crystal structure) and TiO2 (rutile structure) are observed, which indicates new components with crystal structures generated from the amorphous mixtures. This experimental phenomenon of synthesizing new compounds is similar to Stubičar's research in which the orthorhombic ZrTiO4 phase was generated from the hightemperature annealing of amorphous mixtures formed from the ball milling of a ZrO2-TiO2 powder system [21] and consistent with Khor's results that zirconia was produced from annealing amorphous mixtures formed from the ball milling of an equimolar ZrSiO4 and Al2O3 powder system [25].
According to the XRD standard database, cubic Dy2TiO5 {111} presents at 2θ = 30.043°, hexagonal Dy2TiO5 {102} presents at 2θ = 32.411°, orthorhombic Dy2TiO5 {201} presents at 2θ = 29.554°, and pyrochlore Dy2Ti2O7 {111} presents at 2θ = 30.698°; the difference in the above diffraction angles is not very large. Three diffraction peaks are observed in the XRD patterns of the powder mixtures annealed for 3 h at 800, 900, and 1000 °C. The main diffraction peak representing the crystalline phase is at 2θ = 30.0°, as shown in Figure 6b. Therefore, the newly formed product in the powder system annealed between 800 °C and 1000 °C is not pyrochlore Dy2Ti2O7. This result is different than that found in Amit Sinha's research in which pyrochlore Dy2Ti2O7 was initially created in the formation of Dy2TiO5 [19]. In their research using an equimolar Dy2O3-TiO2 system, the chemical reaction of Dy2O3 and TiO2 initially formed pyrochlore Dy2Ti2O7, and then Dy2Ti2O7 reacted with the remaining The powder mixtures milled for 96 h were annealed for 3 h at 800, 900, 1000, 1050, 1100, and 1150 • C. The XRD results of the annealed powder mixtures are shown in Figure 6a,b. Several diffraction peaks different from the diffraction peaks of Dy 2 O 3 (cubic and monoclinic crystal structure) and TiO 2 (rutile structure) are observed, which indicates new components with crystal structures generated from the amorphous mixtures. This experimental phenomenon of synthesizing new compounds is similar to Stubičar's research in which the orthorhombic ZrTiO 4 phase was generated from the high-temperature annealing of amorphous mixtures formed from the ball milling of a ZrO 2 -TiO 2 powder system [21] and consistent with Khor's results that zirconia was produced from annealing amorphous mixtures formed from the ball milling of an equimolar ZrSiO 4 and Al 2 O 3 powder system [25].
According to the XRD standard database, cubic Dy 2 TiO 5 {111} presents at 2θ = 30.043 • , hexagonal Dy 2 TiO 5 {102} presents at 2θ = 32.411 • , orthorhombic Dy 2 TiO 5 {201} presents at 2θ = 29.554 • , and pyrochlore Dy 2 Ti 2 O 7 {111} presents at 2θ = 30.698 • ; the difference in the above diffraction angles is not very large. Three diffraction peaks are observed in the XRD patterns of the powder mixtures annealed for 3 h at 800, 900, and 1000 • C. The main diffraction peak representing the crystalline phase is at 2θ = 30.0 • , as shown in Figure 6b. Therefore, the newly formed product in the powder system annealed between 800 • C and 1000 • C is not pyrochlore Dy 2 Ti 2 O 7 . This result is different than that found in Amit Sinha's research in which pyrochlore Dy 2 Ti 2 O 7 was initially created in the formation of Dy 2 TiO 5 [19]. In their research using an equimolar Dy 2 O 3 -TiO 2 system, the chemical reaction of Dy 2 O 3 and TiO 2 initially formed pyrochlore Dy 2 Ti 2 O 7 , and then Dy 2 Ti 2 O 7 reacted with the remaining Dy 2 O 3 to form orthorhombic Dy 2 TiO 5 .
After annealing for 3 h at 1100 °C, orthorhombic Dy2TiO5 and a small amount of pyrochlore Dy2Ti2O7 and cubic Dy2O3 are detected; notably, cubic Dy2TiO5 is not observed in the ball-milled powder mixtures.
The change of the grain size of the main characteristic phase in the annealed powder mixtures with annealing temperature is shown in Figure 6c. According to the above results, the grain size of the intermediate phase and orthorhombic Dy2TiO5 are calculated using Suryanarayana and Grant Norton's formula. It can be seen that the grain size increases with increasing annealing temperature. The grain size of the intermediate phase is about 27 nm in the powder mixtures annealed for 3 h at 800 °C and is about 275 nm in the powder mixtures annealed for 3 h at 1050 °C. Because the intermediate phase transforms to orthorhombic Dy2TiO5 when the annealing temperature is above 1050 °C, the grain size of orthorhombic Dy2TiO5 is calculated at the annealing temperature ranging from 1050 °C to 1150 °C. The grain size of orthorhombic Dy2TiO5 is about 43 and 380 nm after annealing for 3 h at 1050 °C and 1150 °C, respectively. In addition, the pressure created during ball milling is not high enough to transform the crystal structure of Dy2TiO5 from orthorhombic to hexagonal or from hexagonal to cubic. The average pressure on the contact surface of two colliding mill balls is approximately 8.5 GPa [26], which is far below the 100 GPa needed to induce a pressure wave to cause the Gd2TiO5 phase transformation from the low-temperature orthorhombic phase to the high-temperature hexagonal phase [27]. Therefore, the intermediate phase is not produced by collision pressure. Further investigation is needed into the cause of the formation of the intermediate phase.
To further investigate the annealing behavior of the ball-milled powder mixtures, the powder mixtures milled for 96 h at 200 rpm were sintered for 3 h at 800, 900, 1000, 1050, 1100, and 1150 °C. The phase evolution of the annealed powder mixtures identified by XRD analysis is shown in Figure 7. Under these ball-milling conditions, the Dy2O3-TiO2 powder mixtures are homogenized, and the polymorphic transformation of Dy2O3 from cubic to monoclinic is not observed in Figure 7a. In the XRD pattern of the powder mixtures annealed for 3 h at 1000 °C, diffraction peaks for Dy2O3 and Dy2Ti2O7 phase are observed. However, the main diffraction peak for orthorhombic Dy2TiO5 is not detected. Under these conditions, the annealed powder mixtures are composed of cubic Dy2O3 and pyrochlore Dy2Ti2O7. The powder mixtures generate the orthorhombic Dy2TiO5 phase at 1050 °C and are composed of cubic Dy2O3, pyrochlore Dy2Ti2O7, and orthorhombic Dy2TiO5. The powder mixtures are almost completely transformed to orthorhombic Dy2TiO5 after annealing at 1150 °C for 3 h. The diffraction peak intensity of orthorhombic Dy2TiO5 gradually increases with increasing annealing temperature, and simultaneously, the diffraction peak intensity of Dy2O3 and Dy2Ti2O7 decreases with increasing annealing temperature. For powder mixtures ball milled for 96 h at 200 rpm, the evolutionary behavior at various annealing temperatures is consistent with the data presented in Ref. [19], in which Amit Sinha reported that the chemical reaction of Dy2O3 and TiO2 initially formed pyrochlore Dy2Ti2O7, and then Dy2Ti2O7 reacted with the remaining Dy2O3 to form orthorhombic According to the equilibrium phase diagram, the Dy 2 TiO 5 phase has three crystal structure types depending on the temperature: orthorhombic [15]. Transformation from the high-temperature phase to the low-temperature phase is an exothermic process. For example, the polymorphic transformation of Gd 2 TiO 5 from hexagonal to orthorhombic produced an exothermic peak in the DSC curve [7]. However, in the present work, there is only one exothermic peak in the DSC curve. Although the diffraction peaks in the XRD patterns match well with the diffraction peaks in the standard pattern of cubic Dy 2 TiO 5 , the generated phase should not be cubic Dy 2 TiO 5 because low-temperature phases of orthorhombic and hexagonal Dy 2 TiO 5 were not detected after annealing over temperatures ranging from 700 to 1000 • C for annealing times ranging from several minutes to 3 h; additionally, as mentioned above, the formation temperature of cubic Dy 2 TiO 5 is over 1680 • C. It is not possible to achieve such a high temperature during ball milling. Therefore, the high-temperature phase of cubic Dy 2 TiO 5 should not be produced. Instead, the generated phase is an intermediate phase that has a similar crystal structure to cubic Dy 2 TiO 5 and is a metastable state that phase transforms to orthorhombic Dy 2 TiO 5 when the annealing temperature is above 1050 • C. After annealing for 3 h at 1100 • C, orthorhombic Dy 2 TiO 5 and a small amount of pyrochlore Dy 2 Ti 2 O 7 and cubic Dy 2 O 3 are detected; notably, cubic Dy 2 TiO 5 is not observed in the ball-milled powder mixtures.
The change of the grain size of the main characteristic phase in the annealed powder mixtures with annealing temperature is shown in Figure 6c In addition, the pressure created during ball milling is not high enough to transform the crystal structure of Dy 2 TiO 5 from orthorhombic to hexagonal or from hexagonal to cubic. The average pressure on the contact surface of two colliding mill balls is approximately 8.5 GPa [26], which is far below the 100 GPa needed to induce a pressure wave to cause the Gd 2 TiO 5 phase transformation from the low-temperature orthorhombic phase to the high-temperature hexagonal phase [27]. Therefore, the intermediate phase is not produced by collision pressure. Further investigation is needed into the cause of the formation of the intermediate phase.
To further investigate the annealing behavior of the ball-milled powder mixtures, the powder mixtures milled for 96 h at 200 rpm were sintered for 3 h at 800, 900, 1000, 1050, 1100, and 1150 • C. The phase evolution of the annealed powder mixtures identified by XRD analysis is shown in Figure 7. Under these ball-milling conditions, the Dy 2 O 3 -TiO 2 powder mixtures are homogenized, and the polymorphic transformation of Dy 2 O 3 from cubic to monoclinic is not observed in Figure 7a. In the XRD pattern of the powder mixtures annealed for 3 h at 1000 • C, diffraction peaks for Dy 2 O 3 and Dy 2 Ti 2 O 7 phase are observed. However, the main diffraction peak for orthorhombic Dy 2 TiO 5 is not detected. Under these conditions, the annealed powder mixtures are composed of cubic Dy 2 O 3 and pyrochlore Dy 2 Ti 2 O 7 . The powder mixtures generate the orthorhombic Dy 2 TiO 5 phase at 1050 • C and are composed of cubic Dy 2 O 3 , pyrochlore Dy 2 Ti 2 O 7 , and orthorhombic Dy 2 TiO 5 . The powder mixtures are almost completely transformed to orthorhombic Dy 2 TiO 5 after annealing at 1150 • C for 3 h. The diffraction peak intensity of orthorhombic Dy 2 TiO 5 gradually increases with increasing annealing temperature, and simultaneously, the diffraction peak intensity of Dy 2 O 3 and Dy 2 Ti 2 O 7 decreases with increasing annealing temperature. For powder mixtures ball milled for 96 h at 200 rpm, the evolutionary behavior at various annealing temperatures is consistent with the data presented in Ref. [19], in which Amit Sinha reported that the chemical reaction of Dy 2 O 3 and TiO 2 initially formed pyrochlore Dy 2 Ti 2 O 7 , and then Dy 2 Ti 2 O 7 reacted with the remaining Dy 2 O 3 to form orthorhombic Dy 2 TiO 5 . However, this experimental phenomenon is different from that observed in the annealed powder mixtures milled for 96 h at 500 rpm, as shown in Figure 6, due to the initial conditions of the ball-milling mixtures.
Materials 2017, 10, 19 10 of 13 Dy2TiO5. However, this experimental phenomenon is different from that observed in the annealed powder mixtures milled for 96 h at 500 rpm, as shown in Figure 6, due to the initial conditions of the ball-milling mixtures. The change of the grain size of main characteristic phase in the annealed powder mixtures with annealing temperature is shown in Figure 7c. The grain size of the cubic Dy2O3, pyrochlore Dy2Ti2O7 and orthorhombic Dy2TiO5 are calculated using Suryanarayana and Grant Norton's formula. The grain size of cubic Dy2O3 and pyrochlore Dy2Ti2O7 increases with increasing annealing temperature. Because the orthorhombic Dy2TiO5 is detected in the powder mixtures annealed at 1050 °C for 3 h, the grain size of orthorhombic Dy2TiO5 is calculated when the annealing temperature is over 1050 °C. The grain size of orthorhombic Dy2TiO5 is about 38 nm and 395 nm after annealing 3 h at 1050 °C and 1150 °C, respectively.  Figure 8a,b shows the bright field TEM image and corresponding SAED pattern of the ballmilled Dy2O3-TiO2 powder mixtures annealed at 1000 °C for 3 h, respectively. The powder mixtures are previously milled for 96 h at 500 rpm. After annealing for 3 h, the grain size of powder mixtures is kept in nanometer scale, which can be supported by the corresponding SAED pattern taken from the region marked letter "A" in Figure 8a. The diffraction ring is a typical SAED pattern of nanocrystal materials. After analyzing and indexing the ring-shaped SAED pattern, it is indicated that this SAED pattern belongs to the intermediate Dy2TiO5 phase that has a similar crystal structure to cubic Dy2TiO5, which is in accord with the XRD results as shown in Figure 6. The small-sized powders agglomerate to form large particles with large thicknesses as shown in Figure 8a. The bright zones near the edge of the powder particles is the thin area where the electron beam penetrates. The dark area in the image of the powder particles is the thick area where the electron beam rarely penetrates. After annealing, the amorphous ball-milled powder mixtures are changed to the intermediate Dy2TiO5 phase with crystal structure. Figure 8c,d is the bright field TEM image and corresponding SAED pattern of the annealed Dy2O3-TiO2 powder mixtures that were previously milled for 96 h at 200 rpm. Indexing and analyzing the ring-shaped SAED pattern taken from the area The change of the grain size of main characteristic phase in the annealed powder mixtures with annealing temperature is shown in Figure 7c. The grain size of the cubic Dy 2 O 3 , pyrochlore Dy 2 Ti 2 O 7 and orthorhombic Dy 2 TiO 5 are calculated using Suryanarayana and Grant Norton's formula. The grain size of cubic Dy 2 O 3 and pyrochlore Dy 2 Ti 2 O 7 increases with increasing annealing temperature. Because the orthorhombic Dy 2 TiO 5 is detected in the powder mixtures annealed at 1050 • C for 3 h, the grain size of orthorhombic Dy 2 TiO 5 is calculated when the annealing temperature is over 1050 • C. The grain size of orthorhombic Dy 2 TiO 5 is about 38 nm and 395 nm after annealing 3 h at 1050 • C and 1150 • C, respectively. Figure 8a,b shows the bright field TEM image and corresponding SAED pattern of the ball-milled Dy 2 O 3 -TiO 2 powder mixtures annealed at 1000 • C for 3 h, respectively. The powder mixtures are previously milled for 96 h at 500 rpm. After annealing for 3 h, the grain size of powder mixtures is kept in nanometer scale, which can be supported by the corresponding SAED pattern taken from the region marked letter "A" in Figure 8a. The diffraction ring is a typical SAED pattern of nanocrystal materials. After analyzing and indexing the ring-shaped SAED pattern, it is indicated that this SAED pattern belongs to the intermediate Dy 2 TiO 5 phase that has a similar crystal structure to cubic Dy 2 TiO 5 , which is in accord with the XRD results as shown in Figure 6. The small-sized powders agglomerate to form large particles with large thicknesses as shown in Figure 8a. The bright zones near the edge of the powder particles is the thin area where the electron beam penetrates. The dark area in the image of the powder particles is the thick area where the electron beam rarely penetrates. After annealing, the amorphous ball-milled powder mixtures are changed to the intermediate Dy 2 TiO 5 phase with crystal structure. Figure 8c,d is the bright field TEM image and corresponding SAED pattern of the annealed Dy 2 O 3 -TiO 2 powder mixtures that were previously milled for 96 h at 200 rpm. Indexing and analyzing the ring-shaped SAED pattern taken from the area denoted by the letter "B" indicates that the particles are composed of cubic Dy 2 O 3 and pyrochlore Dy 2 Ti 2 O 7 , which is accord with the XRD results of the powder mixtures annealed at 1000 • C for 3 h as shown in Figure 7. This experimental result is different from that observed in the annealed powder mixtures milled for 96 h at 500 rpm.

Conclusions
The microstructural evolution of Dy2O3-TiO2 powder mixtures during ball milling and postmilled annealing was investigated using TEM, SEM, XRD, and DSC. The conclusions can be made as follows: 1. The ball-milling parameters had a great effect on ball milling and the subsequent annealing process. 2. At 500 rpm rotation speeds, the mixtures were fined, homogenized, nanocrystallized, and then completely amorphized, and the crystal structure of Dy2O3 was transformed from cubic to monoclinic. The amorphous transformation resulted from monoclinic Dy2O3, not from cubic Dy2O3. However, at 200 rpm rotation speeds, the Dy2O3-TiO2 powder mixtures were only homogenized, and the polymorphic transformation of Dy2O3 from cubic to monoclinic was not observed. Meanwhile, the powder mixtures did not transform to the amorphous phase. 3. The powder mixtures milled for 96 h at 500 rpm were annealed for 3 h at a temperature range of

Conclusions
The microstructural evolution of Dy 2 O 3 -TiO 2 powder mixtures during ball milling and post-milled annealing was investigated using TEM, SEM, XRD, and DSC. The conclusions can be made as follows: 1.
The ball-milling parameters had a great effect on ball milling and the subsequent annealing process.

2.
At 500 rpm rotation speeds, the mixtures were fined, homogenized, nanocrystallized, and then completely amorphized, and the crystal structure of Dy 2 O 3 was transformed from cubic to monoclinic. The amorphous transformation resulted from monoclinic Dy 2 O 3 , not from cubic Dy 2 O 3 . However, at 200 rpm rotation speeds, the Dy 2 O 3 -TiO 2 powder mixtures were only homogenized, and the polymorphic transformation of Dy 2 O 3 from cubic to monoclinic was not observed. Meanwhile, the powder mixtures did not transform to the amorphous phase. 3.
The powder mixtures milled for 96 h at 500 rpm were annealed for 3 h at a temperature range of 800 to 1000 • C. An intermediate phase with a crystal structure similar to that of cubic Dy 2 TiO 5 was synthesized, which was a metastable phase that transformed to orthorhombic Dy 2 TiO 5 when the annealing temperature was above 1050 • C. However, the powder mixtures milled for 96 h at 200 rpm did not transform to the amorphous phase. The annealing behavior showed that the chemical reaction of Dy 2 O 3 with TiO 2 initially formed pyrochlore Dy 2 Ti 2 O 7 , and then Dy 2 Ti 2 O 7 reacted with the remaining Dy 2 O 3 to form orthorhombic Dy 2 TiO 5 .