The Inﬂuences of Stirring on the Recrystallization of Ammonium Perrhenate

Featured Application: This study providesa theoretical basisand practical experiencefor optimizing the recrystallization process of ammonium perrhenate and the design of a recrystallization reactor. Abstract: Ammonium perrhenate is widely used in alloy manufacturing, powder processing, the catalytic industry, and other ﬁelds. Recrystallization can improve the speciﬁc surface area of ammonium perrhenate, reduce its particle size, and improve its particle size distribution uniformity. Therefore, recrystallized ammonium perrhenate can obtain better application beneﬁts in the above ﬁelds. Stirring is an important factor that a ﬀ ects the recrystallization of ammonium perrhenate, and this paper systematically analyzes the inﬂuence of the stirring paddle types and stirring intensities on ammonium perrhenate during the homogeneous recrystallization process, ultimately revealing the relationship between the growth rate of ammonium perrhenate and the stirring process. Particle image velocimetry physical simulation results showed that the ﬂow ﬁeld in the reactor was more evenly distributed when using the disc turbine impeller, and a relatively uniform velocity liquid ﬂow area was formed in the whole reactor, while the low-velocity liquid ﬂow area was smaller. Therefore, this information, combined with SEM test results, suggests that under the same recrystallization time and stirring intensity, the stirring e ﬀ ect of a disc turbine impeller is more suitable than a propelling propeller and an Intermig impeller for the recrystallization process of ammonium perrhenate. Moreover, the XRD patterns and SEM analysis showed that if the agglomeration in the systems was too strong or too weak, the growths of the (101) crystal plane and (112) crystal plane were restrained, which caused an attenuation in the growth rates along the crystallographic directions that were orthogonal to the crystal faces. Finally, the reduction experiments show that the recrystallization of ammonium perrhenate could improve the phase parameters of rhenium powders.


Introduction
As an important strategic rare metal, Re is widely used in the metallurgy, chemical, alloy manufacturing and aerospace industries [1,2]. At present, the preparation technologies for Re include

Analytical Methods
XRD detection: The light tube type was a Cu target, ceramic X light tube. λ = 0.15406 nm, the scan range was 10~90 degrees, and the scanning speed was 2 degrees/min.
Particle size distribution detection: The test medium was ethanol, the optical model was Mie, and the distribution type was volume distribution.
The parameters of SEM: (1) Low power imaging: The electron acceleration voltage was 20.0 KV, the working distance was 9.5 mm, and the magnification was 100 times. (2) High power imaging: The electron acceleration voltage was 20.0 KV, the working distance was 9.5 mm, and the magnification was 800 times.
Physical simulation analysis method: (1) Design of water model: In this study, a 1 L self-designed recrystallization condensation reactor was taken as the prototype, and the water model was established based on the similarity principle (geometric similarity). The water model and prototype met the following conditions: N2 = N1(T1/T2) × (Formula (1)), T1 was the inner diameter of the water model, T2 was the inner diameter of the prototype reactor, N1 was the stirring intensity (rpm) of the water model, and N2 was the stirring intensity (rpm) of the prototype reactor. X was the amplification index. X = 1 means that the linear stirring speeds of geometrically similar reactors were the same [17]. The inner diameter of the recrystallization condensation reactor was 140 mm. According to the principle of geometric similarity, the water model of the recrystallization condensation reactor was established. The inner diameter of the water model was 240 mm. The volume ratio of the water model to the prototype is 5:1.

Analytical Methods
XRD detection: The light tube type was a Cu target, ceramic X light tube. λ = 0.15406 nm, the scan range was 10~90 degrees, and the scanning speed was 2 degrees/min.
Particle size distribution detection: The test medium was ethanol, the optical model was Mie, and the distribution type was volume distribution.
The parameters of SEM: (1) Low power imaging: The electron acceleration voltage was 20.0 KV, the working distance was 9.5 mm, and the magnification was 100 times. (2) High power imaging: The electron acceleration voltage was 20.0 KV, the working distance was 9.5 mm, and the magnification was 800 times.
Physical simulation analysis method: (1) Design of water model: In this study, a 1 L self-designed recrystallization condensation reactor was taken as the prototype, and the water model was established based on the similarity principle (geometric similarity). The water model and prototype met the following conditions: N 2 = N 1 (T 1 /T 2 ) × (Formula (1)), T 1 was the inner diameter of the water model, T 2 was the inner diameter of the prototype reactor, N 1 was the stirring intensity (rpm) of the water model, and N 2 was the stirring intensity (rpm) of the prototype reactor. X was the amplification index. X = 1 means that the linear stirring speeds of geometrically similar reactors were the same [17]. The inner diameter of the recrystallization condensation reactor was 140 mm. According to the principle of geometric similarity, the water model of the recrystallization condensation reactor was established. The inner diameter of the water model was 240 mm. The volume ratio of the water model to the prototype is 5:1. (2) Procedures of physical simulation: According to Formula (1), when the stirring speed of the recrystallization condensation reactor was 200 rpm, the stirring speed of the water model was about 120 rpm. The right amount of tracer particles (polystyrene) was added into the water, and then the particle image velocimetry two dimensional particle velocity meter was turned on. After the computer received the image data, Tecplot.360.v11.0.1 software was used for processing to get the vector images.

Experimental Procedure
(1) The sample preparation of ammonium perrhenate: The NH 4 ReO 4 raw material was dissolved in deionized water to form a supersaturated solution at room temperature (25 • C). The NH 4 ReO 4 solution was then evaporated to the RE-2000A rotary evaporator (oil bath temperature 120 • C) and formed the supersaturated solution at a high temperature [18]. The hot supersaturated solution was passed into the recrystallization condensation reactor, the stirring strength of the reactor was adjusted to cool during crystallization, the recrystallization time was 3 h, the recrystallized solid-liquid mixtures were filtered by the vacuum filter extractor, the solid samples were obtained and dried prior in the oven to being measured, and the oven temperature was 60 • C. The ammonium perrhenate samples in this paper were prepared at the stirring strengths of 100, 150, 200 and 250 rpm. (2) Sample preparation of rhenium powders: In this paper, non-recrystallized ammonium perrhenate (as shown in Figure 1) and recrystallized ammonium perrhenate at a 200 rpm (the disc turbine impeller was used) stirring strength were used as raw materials to prepare the rhenium powder samples. The two kinds of ammonium perrhenate particles (60 g) were reduced with the same reduced time (3 h) in the high-temperature vacuum sintering furnace, and the reduced samples were obtained. The hydrogen flow rate was 500 mL/min. The heating rate was 10 • C/min.

The Influences of Paddle Types
Three kinds of stirring paddles that are commonly used in industry were studied in this study, which were the disc turbine impeller, the propelling propeller, and the Intermig impeller. The diagrams of the stirring paddles are shown in Figure 2. Under the same recrystallization time and agitation intensity (200 rpm), SEM images of the NH 4 ReO 4 crystals prepared in the prototype reactor with different stirring paddle types are shown in Figure 3. It can be seen that the morphologies of most crystal particles that were prepared by the disc turbine impeller were regular, the crystal surfaces were smooth, and a few crystal particles had rough surfaces, indicating that most crystal particle development was relatively complete. The crystal particle with a high magnification was the representative of the complete growth of the crystal particles. However, the morphologies of most crystal particles prepared by the propelling propeller and the Intermig impeller had uneven surface topographies, indicating that the crystals were not fully developed. The crystal particles with a high magnification were the representative of the incomplete growth crystal particles. This suggests that under the same operating conditions, the stirring effect of the disc turbine impeller is more suitable for the homogeneous recrystallization process.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 4 of 11 (2) Procedures of physical simulation: According to Formula (1), when the stirring speed of the recrystallization condensation reactor was 200 rpm, the stirring speed of the water model was about 120 rpm. The right amount of tracer particles (polystyrene) was added into the water, and then the particle image velocimetry two dimensional particle velocity meter was turned on. After the computer received the image data, Tecplot.360.v11.0.1 software was used for processing to get the vector images.

Experimental Procedure
(1) The sample preparation of ammonium perrhenate: The NH4ReO4 raw material was dissolved in deionized water to form a supersaturated solution at room temperature (25 °C). The NH4ReO4 solution was then evaporated to the RE-2000A rotary evaporator (oil bath temperature 120 °C) and formed the supersaturated solution at a high temperature [18]. The hot supersaturated solution was passed into the recrystallization condensation reactor, the stirring strength of the reactor was adjusted to cool during crystallization, the recrystallization time was 3 h, the recrystallized solid-liquid mixtures were filtered by the vacuum filter extractor, the solid samples were obtained and dried prior in the oven to being measured, and the oven temperature was 60 °C. The ammonium perrhenate samples in this paper were prepared at the stirring strengths of 100, 150, 200 and 250 rpm. (2) Sample preparation of rhenium powders: In this paper, non-recrystallized ammonium perrhenate (as shown in Figure 1) and recrystallized ammonium perrhenate at a 200 rpm (the disc turbine impeller was used) stirring strength were used as raw materials to prepare the rhenium powder samples. The two kinds of ammonium perrhenate particles (60 g) were reduced with the same reduced time (3 h) in the high-temperature vacuum sintering furnace, and the reduced samples were obtained. The hydrogen flow rate was 500 mL/min. The heating rate was 10 °C/min.

The Influences of Paddle Types
Three kinds of stirring paddles that are commonly used in industry were studied in this study, which were the disc turbine impeller, the propelling propeller, and the Intermig impeller. The diagrams of the stirring paddles are shown in Figure 2. Under the same recrystallization time and agitation intensity (200 rpm), SEM images of the NH4ReO4 crystals prepared in the prototype reactor with different stirring paddle types are shown in Figure 3. It can be seen that the morphologies of most crystal particles that were prepared by the disc turbine impeller were regular, the crystal surfaces were smooth, and a few crystal particles had rough surfaces, indicating that most crystal particle development was relatively complete. The crystal particle with a high magnification was the representative of the complete growth of the crystal particles. However, the morphologies of most crystal particles prepared by the propelling propeller and the Intermig impeller had uneven surface topographies, indicating that the crystals were not fully developed. The crystal particles with a high magnification were the representative of the incomplete growth crystal particles. This suggests that under the same operating conditions, the stirring effect of the disc turbine impeller is more suitable for the homogeneous recrystallization process.

The Influences of Stirring Intensities
An ammonium perrhenate crystal is tetragonal 4/m (isostructural with the scheelite CaWO4), and it is a typical form of monometric crystals [19]. The crystals were in solution, and the growth process of the crystals ended when the solute was precipitated and dissolved in equilibrium [20]. To inspect the effect of the agitation intensities on the crystal morphologies, the SEM micrographs of the NH4ReO4 samples with different agitation intensities (the disc turbine impeller was used) are shown in Figure 4. When the agitation intensity was 100 rpm, many irregularly shaped crystals with different sizes were formed when the crystalline system was in equilibrium; the surface structures of these crystals were rough and uneven, as shown in Figure 4a. Crystals with irregular morphologies are generally considered to be underdeveloped [21,22]. This suggests that the strong crystal agglomeration effect led to incomplete crystal development during equilibrium with the dissolved solute. When the agitation intensity was 150 rpm, the macroscopic morphologies of these crystals were mostly tetragonal crystals, and the surfaces of these crystals were defective and smooth, as shown in Figure 4b. It can be seen that the crystal agglomeration effect was subdued at high stirring speeds, the uniformity of the crystals increased, the crystal morphologies became regular, and, when the crystal system was in equilibrium, the morphologies of the crystals were more regular. When the agitation intensity was 200 rpm, the NH4ReO4 particles formed tetragonal crystals with a similar morphologies and sizes, and the surfaces of these crystals were neat and smooth, as shown in Figure 4c. This suggests that when the crystal agglomeration effect was further weakened, the morphologies of the crystals were further improved when in equilibrium with the crystal system. When the agitation intensity was 250 rpm, the macroscopic morphologies of these crystals were mostly tetragonal crystals. However, there were small crystals that could be found, and the surfaces of these crystals were defective and smooth, as shown in Figure 4d. This suggests that when the crystal agglomeration effect was excessively weakened, the integrity of crystal morphologies deteriorated when in equilibrium with the crystal system, and the crystal surface morphologies had obvious defects.
The specific surfaces and D50s (median diameter) of the NH4ReO4 samples with different agitation intensities are given in Table 1. It can be seen that as the stirring speeds increased, the specific surfaces and the D50 decreased. When the stirring speeds were higher than 200 rpm, the specific surface began to decrease and the D50 began to increase. The purpose of this work was to reduce the size of ammonium perrhenate and enhance its specific surface area. Therefore, it was determined that 200 rpm was an ideal recrystallization condition. Therefore, the stirring speeds above 200 rpm had no significance.

The Influences of Stirring Intensities
An ammonium perrhenate crystal is tetragonal 4/m (isostructural with the scheelite CaWO4), and it is a typical form of monometric crystals [19]. The crystals were in solution, and the growth process of the crystals ended when the solute was precipitated and dissolved in equilibrium [20].
To inspect the effect of the agitation intensities on the crystal morphologies, the SEM micrographs of the NH 4 ReO 4 samples with different agitation intensities (the disc turbine impeller was used) are shown in Figure 4. When the agitation intensity was 100 rpm, many irregularly shaped crystals with different sizes were formed when the crystalline system was in equilibrium; the surface structures of these crystals were rough and uneven, as shown in Figure 4a. Crystals with irregular morphologies are generally considered to be underdeveloped [21,22]. This suggests that the strong crystal agglomeration effect led to incomplete crystal development during equilibrium with the dissolved solute. When the agitation intensity was 150 rpm, the macroscopic morphologies of these crystals were mostly tetragonal crystals, and the surfaces of these crystals were defective and smooth, as shown in Figure 4b. It can be seen that the crystal agglomeration effect was subdued at high stirring speeds, the uniformity of the crystals increased, the crystal morphologies became regular, and, when the crystal system was in equilibrium, the morphologies of the crystals were more regular. When the agitation intensity was 200 rpm, the NH 4 ReO 4 particles formed tetragonal crystals with a similar morphologies and sizes, and the surfaces of these crystals were neat and smooth, as shown in Figure 4c. This suggests that when the crystal agglomeration effect was further weakened, the morphologies of the crystals were further improved when in equilibrium with the crystal system. When the agitation intensity was 250 rpm, the macroscopic morphologies of these crystals were mostly tetragonal crystals. However, there were small crystals that could be found, and the surfaces of these crystals were defective and smooth, as shown in Figure 4d. This suggests that when the crystal agglomeration effect was excessively weakened, the integrity of crystal morphologies deteriorated when in equilibrium with the crystal system, and the crystal surface morphologies had obvious defects.
The specific surfaces and D 50 s (median diameter) of the NH 4 ReO 4 samples with different agitation intensities are given in Table 1. It can be seen that as the stirring speeds increased, the specific surfaces and the D 50 decreased. When the stirring speeds were higher than 200 rpm, the specific surface began to decrease and the D 50 began to increase. The purpose of this work was to reduce the size of ammonium perrhenate and enhance its specific surface area. Therefore, it was determined that 200 rpm was an ideal recrystallization condition. Therefore, the stirring speeds above 200 rpm had no significance.
In order to research the relationship between the difference of crystal morphologies and the selectivity of crystal growth directions, XRD analyses were performed. The XRD patterns of the samples at different stirring speeds and the same crystallization time (2 h) are shown in Figure 5, and the XRD information is given in Table 2.  In order to research the relationship between the difference of crystal morphologies and the selectivity of crystal growth directions, XRD analyses were performed. The XRD patterns of the samples at different stirring speeds and the same crystallization time (2 h) are shown in Figure 5, and the XRD information is given in Table 2.     In order to research the relationship between the difference of crystal morphologies and the selectivity of crystal growth directions, XRD analyses were performed. The XRD patterns of the samples at different stirring speeds and the same crystallization time (2 h) are shown in Figure 5, and the XRD information is given in Table 2.   The diffraction peaks were from the NH 4 ReO 4 , and other crystalline species were not observed. These diffraction peaks had different intensities. When the stirring intensity was 200 rpm, the diffraction peaks had the strongest intensities. This indicated that when the stirring intensity was 200 rpm, the selective growth rates of the crystal planes corresponding to the characteristic peaks were higher than that of other samples. In this paper, the selectivity of the crystal growth directions on the (101) crystal planes and (112) crystal planes at different stirring speeds could be calculated by the Scherrer formula, where D = Kλ/(B cos θ); the results are given in Table 3. Table 3 shows that the growth rates of the (101) crystal planes and (112) crystal planes were promoted with increasing stirring speeds (100~200 rpm). However, the growth rates of the (101) crystal planes and (112) crystal planes were subdued at the excessive stirring speed (250 rpm).

Reduction Experimental
In this study, the recrystallized NH 4 ReO 4 crystals were prepared at 200 rpm, and the NH 4 ReO 4 crystals without recrystallization (NH 4 ReO 4 (99.99%, Re ≥ 69.4%) Halin Chemical Co. LTD, Weifang, China.) were reduced by H 2 at high temperatures. The full reduction temperature of the recrystallized NH 4 ReO 4 crystals was at 700 • C, while the full reduction temperature of the NH 4 ReO 4 without recrystallization crystals was at 1000 • C. Moreover, the rhenium powders that were prepared by recrystallized NH 4 ReO 4 crystals had a very even size distribution, as shown in Figure 6b. However, the rhenium powders that were prepared by the NH 4 ReO 4 crystals without recrystallization had an uneven particle size distribution, as shown in Figure 6a. The rhenium powders that were prepared by recrystallized NH 4 ReO 4 crystals had smaller D 50 s and larger specific surfaces, as shown in Table 4. This shows that the Re powders that were prepared by recrystallized ammonium rhenate had a more uniform particle size distribution, a smaller D 50 , and a larger specific surface area. prepared by recrystallized NH4ReO4 crystals had a very even size distribution, as shown in Figure  6b. However, the rhenium powders that were prepared by the NH4ReO4 crystals without recrystallization had an uneven particle size distribution, as shown in Figure 6a. The rhenium powders that were prepared by recrystallized NH4ReO4 crystals had smaller D50s and larger specific surfaces, as shown in Table 4. This shows that the Re powders that were prepared by recrystallized ammonium rhenate had a more uniform particle size distribution, a smaller D50, and a larger specific surface area. a Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 11 Figure 6. Particle size distribution of the rhenium samples. a-Sample was prepared with the NH4ReO4 crystals without recrystallization. b-Sample was prepared by the recrystallization NH4ReO4 crystals at 200 rpm. The red line represents particle size (µm), and the blue line represents the accumulation.

Discussion
In order to study the effects of agitator types on recrystallization, PIV physical simulation technique was used to simulate the flow field distribution in the reactor. The velocity vector cloud diagrams of the flow field distribution with different agitator types at the same agitator speed and agitator position are shown in Figure 7. It can be seen that the flow field in the reactor was more evenly distributed when using the disc turbine impeller, and a relatively uniform velocity liquid flow area was formed in the whole reactor while the low-velocity liquid flow area was smaller, as shown in Figure 7a. When the Intermig impeller was used, the high-speed liquid flow area was mainly distributed at both sides of the agitator shaft and at the bottom of the reactor. However, the other areas of the reactor belonged to the low-speed liquid flow zone, and the flow field was not evenly distributed, as shown in Figure 7b. When the propelling propeller was used, two high-speed circulating liquid flow zones were formed on both sides of the agitator, and the high-speed liquid flow zone was mainly distributed on both sides of the bottom of the reactor. However, the other areas of the reactor belonged to the low-speed liquid flow zone, and the flow field was not evenly distributed, as shown in Figure 7c. In conclusion, the distribution of flow field in the reactor was more uniform when the disc turbine impeller was used, and the relatively low-speed flow area of the reactor was smaller. Combined with the SEM test results (Figure 3), this suggests that under the uniform flow field, the large particles collided and broke into the small particles of uniform size, and then these particles became nuclei again and grew into complete crystals under uniform flow. Therefore, the stirring effect of the disc turbine impeller is more suitable for the homogeneous recrystallization process. b Figure 6. Particle size distribution of the rhenium samples. a-Sample was prepared with the NH 4 ReO 4 crystals without recrystallization. b-Sample was prepared by the recrystallization NH 4 ReO 4 crystals at 200 rpm. The red line represents particle size (µm), and the blue line represents the accumulation.

Discussion
In order to study the effects of agitator types on recrystallization, PIV physical simulation technique was used to simulate the flow field distribution in the reactor. The velocity vector cloud diagrams of the flow field distribution with different agitator types at the same agitator speed and agitator position are shown in Figure 7. It can be seen that the flow field in the reactor was more evenly distributed when using the disc turbine impeller, and a relatively uniform velocity liquid flow area was formed in the whole reactor while the low-velocity liquid flow area was smaller, as shown in Figure 7a. When the Intermig impeller was used, the high-speed liquid flow area was mainly distributed at both sides of the agitator shaft and at the bottom of the reactor. However, the other areas of the reactor belonged to the low-speed liquid flow zone, and the flow field was not evenly distributed, as shown in Figure 7b. When the propelling propeller was used, two high-speed circulating liquid flow zones were formed on both sides of the agitator, and the high-speed liquid flow zone was mainly distributed on both sides of the bottom of the reactor. However, the other areas of the reactor belonged to the low-speed liquid flow zone, and the flow field was not evenly distributed, as shown in Figure 7c. In conclusion, the distribution of flow field in the reactor was more uniform when the disc turbine impeller was used, and the relatively low-speed flow area of the reactor was smaller. Combined with the SEM test results (Figure 3), this suggests that under the uniform flow field, the large particles collided and broke into the small particles of uniform size, and then these particles became nuclei again and grew into complete crystals under uniform flow. Therefore, the stirring effect of the disc turbine impeller is more suitable for the homogeneous recrystallization process.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 9 of 11 Figure 7. The velocity vector cloud diagrams of flow field distribution with different agitator types (a-disc turbine impeller; b-Intermig impeller; and c-propelling propeller).
As discussed above, the recrystallization of ammonium perrhenate was accompanied by agglomeration, and appropriately increasing the stirring intensity reduced the agglomeration effect. The SEM and XRD analyses showed that an appropriate reduction of the agglomeration effect was conducive to increasing the growth rates of the ammonium perrhenate crystals. When the stirring intensity was 200 rpm, the growth rates of the (101) and (112) crystal planes corresponding to the characteristic peaks of the crystals were relatively fast, as shown in Table 3. Therefore, the crystal development was more complete at the same crystallization time. However, when the stirring intensity was 250 rpm, the agglomeration of the system was excessively reduced; the growth rates of the (101) and (112) crystal planes corresponding to the characteristic peaks of the crystals were subdued, resulting in a slow and incomplete crystal development. This suggested that the induction period for nucleation was reduced and the formation of crystal nuclei and growth of crystals reached an ideal equilibrium state when an appropriate reduction in the agglomeration was obtained. The high stirring intensity also made the large sedimentary particles collide, transition into the secondary crystallization stage, and maintain a uniform grain size in the crystal. However, the excessive stirring intensity enhanced the collisions between the particles, and the agglomeration was excessively weakened. Therefore, the excessive stirring intensity broke the balance for the nucleation process; the generated crystal nuclei collided and broke up under the excessive stirring intensity, resulting in a decrease of the number of crystal nuclei and an increase of the grain size, ultimately leading to an unsatisfactory crystal development. Moreover, the experimental reduction showed that the recrystallized ammonium perrhenate had a better mass and heat transfer efficiency, and it was more conducive to obtain a rhenium powder with more uniform particle size distribution and a larger specific surface area during hydrogen reduction.

Conclusions
In this study, the process control of the homogeneous recrystallization of ammonium perrhenate was investigated. According to the results of the present research, the conclusions are as follows: (1) The distribution of flow field in the reactor was more uniform and the relatively low-speed flow area of the reactor was smaller when the disc turbine impeller was used, and the disc turbine impeller was more suitable for the ammonium perrhenate recrystallization process. (2) The proper reduction of the agglomeration effect was conducive to increasing the growth rates of the ammonium perrhenate crystals, and the growth rates of the (101) crystal planes and (112) crystal planes were promoted with increasing stirring speeds (100~200 rpm). However, the growth rates of the (101) crystal planes and (112) crystal planes were subdued at the excessive stirring speed (250 rpm). (3) The recrystallized ammonium perrhenate had a better mass and heat transfer efficiency, and it was more conducive to obtain the rhenium powder with a more uniform particle size distribution and a larger specific surface area during hydrogen reduction. As discussed above, the recrystallization of ammonium perrhenate was accompanied by agglomeration, and appropriately increasing the stirring intensity reduced the agglomeration effect. The SEM and XRD analyses showed that an appropriate reduction of the agglomeration effect was conducive to increasing the growth rates of the ammonium perrhenate crystals. When the stirring intensity was 200 rpm, the growth rates of the (101) and (112) crystal planes corresponding to the characteristic peaks of the crystals were relatively fast, as shown in Table 3. Therefore, the crystal development was more complete at the same crystallization time. However, when the stirring intensity was 250 rpm, the agglomeration of the system was excessively reduced; the growth rates of the (101) and (112) crystal planes corresponding to the characteristic peaks of the crystals were subdued, resulting in a slow and incomplete crystal development. This suggested that the induction period for nucleation was reduced and the formation of crystal nuclei and growth of crystals reached an ideal equilibrium state when an appropriate reduction in the agglomeration was obtained. The high stirring intensity also made the large sedimentary particles collide, transition into the secondary crystallization stage, and maintain a uniform grain size in the crystal. However, the excessive stirring intensity enhanced the collisions between the particles, and the agglomeration was excessively weakened. Therefore, the excessive stirring intensity broke the balance for the nucleation process; the generated crystal nuclei collided and broke up under the excessive stirring intensity, resulting in a decrease of the number of crystal nuclei and an increase of the grain size, ultimately leading to an unsatisfactory crystal development. Moreover, the experimental reduction showed that the recrystallized ammonium perrhenate had a better mass and heat transfer efficiency, and it was more conducive to obtain a rhenium powder with more uniform particle size distribution and a larger specific surface area during hydrogen reduction.

Conclusions
In this study, the process control of the homogeneous recrystallization of ammonium perrhenate was investigated. According to the results of the present research, the conclusions are as follows: (1) The distribution of flow field in the reactor was more uniform and the relatively low-speed flow area of the reactor was smaller when the disc turbine impeller was used, and the disc turbine impeller was more suitable for the ammonium perrhenate recrystallization process. (2) The proper reduction of the agglomeration effect was conducive to increasing the growth rates of the ammonium perrhenate crystals, and the growth rates of the (101) crystal planes and (112) crystal planes were promoted with increasing stirring speeds (100~200 rpm). However, the growth rates of the (101) crystal planes and (112) crystal planes were subdued at the excessive stirring speed (250 rpm). (3) The recrystallized ammonium perrhenate had a better mass and heat transfer efficiency, and it was more conducive to obtain the rhenium powder with a more uniform particle size distribution and a larger specific surface area during hydrogen reduction.