2019 Volume 60 Issue 6 Pages 1044-1050
The effects of settling time up to 8 h on the quality of fluxless Mg–Nd–Zn–Zr melt preparation are discussed in terms of melt chemistry and inclusions’ structural properties. The content of Nd, Zn and Zr lie in the demanding scope throughout settling process. Neodymium content rises slightly at the settling time of 0.25 h, then decreases gradually up to 8 h. Zinc experiences a slight increase after 0.25 h, then levels off. Zirconium drops sharply at 0.25 h mainly due to the subsidence of undissolved zirconium (Zr) particles, then drops slightly at 1 h, and then stabilizes. The inclusions have various morphologies like rod-like, granular and clustered, and different types such as oxides, carbides, undissolved Zr particles, and etc. The effect of settling time on inclusions’ removal is remarkable during early settling stage, from 0 to 0.25 h, and is getting less stark with increasing time. The area fraction, average diameter and maximum diameter of inclusions all decrease rapidly at 0.25 h, then gradually, and finally reach a plateau region. The changes of minimum diameter and number density values are insignificant through whole process. The theoretical analysis on inclusions’ settling and rising behavior agrees with the experimental data.
Fig. 3 The settling distance as a function of settling time for various inclusions of different sizes, (a) 1 × 10−6, (b) 2 × 10−6, (c) 4 × 10−6, (d) 8 × 10−6, (e) 16 × 10−6 and (f) 32 × 10−6 m, in the melt at 1033 K. The dashed line indicates the sample depth, approximately 0.2 m below the melt surface.
Magnesium (Mg) alloys are the lightest metallic structural materials, and have been drawing more and more attention in different applications, as a result of the emphasis on the weight reduction in the industry. Mg–Nd–Zn–Zr alloy exhibits great application potentials not only in aerospace and automotive industries1–4) but also in biodegradable implant5,6) due to its good mechanical properties and high corrosion resistance. It is known that the melt cleanliness influences the performance of magnesium alloys components.7) Flux (consisting of MgCl2 and other cholrides and CaF28)) purification approach, i.e., flux refining and flux shielding, is commonly used to purify Mg melt in the industry. The melt of Mg–Nd–Zn–Zr tend to lose rare earth element neodymium (Nd) when utilizing flux purification,9) because Nd reacts with the magnesium chloride (MgCl2) in fluxes. In view of the loss of Nd and MgCl2, and other disadvantages of flux technology, fluxless technology (like gas shielding,10) filtration11) and gas bubbling12–15)) is gaining an increasing interest.15) Though fluxless Mg–Nd–Zn–Zr melt preparation avoids the drop of Nd, but the undesirable phenomena of the zirconium (Zr) loss, still exists; undissolved Zr particles in Mg–Nd–Zn–Zr melt are easy to settle to the melt bottom due the higher density compared to Mg melt, which is a characteristic of the melt of Zr alloying Mg alloys.16)
No matter what purification method is utilized, physical sedimentation is one simple, effective and necessary assistive technology. Obviously, the longer the settling time is, the more inclusions fall to the melt bottom, the cleaner the melt becomes. But, long time settling probably intensify the loss of major alloying elements in Mg–Nd–Zn–Zr melt, lead to the formation of new inclusions, increase production cost, increase production time and decrease production benefits. Therefore, in order to obtain a satisfactory Mg–Nd–Zn–Zr melt, it is necessary to consider carefully as many consequences as possible together when choosing time for settling process. Though a variety of researches3,6,17–20) have been carried out on different aspects like alloy composition design, heat treatment and process technology of Mg alloys, only a few9,21,22) concentrate on Mg melt preparation, very few work15) have been carried out on Mg–Nd–Zn–Zr melt preparation. Among all quality requirements on Mg melt preparation, stable and qualified chemistry and acceptable inclusion structural properties (like morphology, quantity, size and distribution) are recognized as two most significant factors. In this research, the effects of a series of settling times (0, 0.25, 1, 2, 4 and 8 h) on the quality of Mg–Nd–Zn–Zr melt employing fluxless purification technology are researched, with respect to the chemistry and inclusions of the melt.
The experiment was implemented in an industrial steal crucible of a depth of 1 m and an internal diameter of 0.5 m. A total of 240 kg raw materials, pure magnesium ingots (Mg > 99.80 mass%), pure zinc ingots (Zn > 99.95 mass%), Mg–30Nd mass% and Mg–25Zr mass% master alloy, were melt in the crucible, with a nominal composition of Mg–2.5Nd–0.6Zn–1.5Zr mass%. The melt was prepared by fluxless based refining, in which protective atmospheres (CO2 + 0.18 vol% SF6) and inert gas (Ar) bubbling were employed; the flow rate of CO2, SF6, and Ar are 5 × 10−4, 9 × 10−7 and 8.3 × 10−5 m3 s−1, respectively. SF6 is able to form a dense film containing Mg oxide (MgO) and Mg fluoride (MgF2) on the molten Mg surface, which prevents further oxidation and evaporation of Mg.23–25) Sulphur powder was used as flame retardants. After 0.25 h Ar bubbling, the melt was sampled at settling times 0, 0.25, 1, 2, 4 or 8 h. Master ingots were prepared from ∼0.2 m below the melt top, with a preheated boron nitride coated mild steel cone ladle (dimensions of cavity: ϕ0.04/0.05 × 0.05 m; outside dimension: ϕ0.09 × 0.07 m), then subjected to air cool. The samples were sectioned from the conical ingots, at the distance of ∼0.01 m above the bottom, for microstructure analysis. The chemical component of the sample is tested with inductively coupled plasma (ICP) and the inclusions in the central region of the section were analyzed utilizing optical microscope and scanning electron microscope.
The content values of alloying elements Nd, Zn and Zr in Mg–Nd–Zn–Zr melt during settling are displayed in Table 1. The nominal composition of the alloy is Mg–2.5Nd–0.6Zn–1.5Zr mass%, and the achieved content of Nd, Zn and Zr in the range of 2.0 to 2.8 mass%, 0.2 to 0.7 mass% and 0.4 to 1.0 mass%, respectively, will meet the actual industrial requirements.
Both Nd and Zn experience a slight increase between 0 and 0.25 h, 2.51 to 2.74 mass% for Nd and 0.65 to 0.74 mass% for Zn, mainly as a result of the sharp decline of Zr, in line with the law of conservation of mass. Zirconium drops from 1.23 mass% at 0 h to 0.58 mass% after 0.25 h settling, and such dramatic drop in a short time is consistent with the rapid subsidence phenomena of undissolved Zr particle observed in previous studies.15,16) The settling behavior of undissolved Zr particle and other forms of inclusions will be discussed later. Neodymium drops gradually after 0.25 h, to 2.27 mass% at final 8 h, probably due to the formation of stable Fe–Nd intermetallic in steel crucible. As a major alloying element for Mg alloy, Zinc is stable in the Mg melt,26) and its content varies from 0.70 to 0.74 mass% between 0.25 and 8 h. It can be noted that Zn values through the process are all larger than its nominal content, 0.6 mass%, which can be explained by the loss of Zr, Nd leading to the increase of Zn. The content of Zr shows a slight reduction from 0.25 to 1 h, then it levels off in the range of 0.53 and 0.55 mass% between 1 and 8 h. The loss of dissolved Zr in zirconium-refined magnesium alloys is studied in Ref. 21, which finds out that the phenomenon stems from the unique Zr-rich cores and is reversible by agitation when remelted in an iron-free environment while irreversible in steel crucibles due to the interference from iron. In addition, the content values of Nd, Zn and Zr all lie in the demanding scope even at the settling time as long as 8 h, which are 2.27, 0.73 and 0.53 mass% with a yielding rate of 90.8%, 121.7%, 35.3%, respectively.
3.2 Structural properties evolution of the inclusions during settling processSEM images and EDX analysis results of some typical inclusions before settling are displayed in Fig. 1 and Table 2, respectively. It can be seen that the inclusions exhibit various morphologies, such as long rod-like, short rod-like, granular and clustered. Based on the EDX results, O and C are two most common elements in the inclusions, indicating the Mg melt contain oxides and carbides inclusions. S element is found in some inclusions, probably because the sulphur powder used as flame retardants enter into the melt leading to the formation of MgS and MgSO4. Some inclusions have high content of Zr, implying these inclusions are mainly composed of undissolved Zr particles, which is typical in Zr alloying Mg alloys.16) The optical micrographs demonstrating the evolution of inclusions of as-cast Mg–Nd–Zn–Zr alloy through settling are shown in Fig. 2(a)–(f). Special attention has been paid to distinguish between inclusions and zirconium rich regions, and the method is introduced in details by Ref. 15. A great amount of inclusions can be observed in Fig. 2(a), with different morphologies. After 0.25 h settling, the Mg melt gets much cleaner and large inclusions are almost not observed, mainly as a result of inclusion sedimentation. As settling is proceeded, it can be seen from Fig. 2(b)–(f) that melt remain clean and no large inclusions are formed, indicating that a combination of sulphur hexafluoride (SF6) and carbon dioxide (CO2) utilized as cover gas in this research is effective to protect the melt from the formation of large inclusions.
SEM images of some typical inclusions before settling.
Optical micrographs of as-cast Mg–Nd–Zn–Zr alloy during settling process, (a) 0, (b) 0.25, (c) 1, (d) 2, (e) 4 and (f) 8 h, in polished condition without etch.
For each sample, inclusions within an area of approximately 1 × 10−4 m2 in the center region of the sample are observed. The calculated area fraction (ϕ), average diameter (Davg.), maximum diameter (Dmax.), minimum diameter (Dmin.) and number density (NA) of the inclusions are listed in Table 3. The area fraction is the ratio of the area of inclusions to the overall area; the diameter is the area equivalent diameter assuming a circular morphology for all inclusions; and the number density is the number of inclusions per area. The area fraction of inclusions in as cast sample before sedimentation is relatively high, 0.50%, then it drops rapidly to 0.22% after 0.25 h and slowly to 0.17% after 1 h, then begins to stabilize. The average diameter follows the same decline trend as area fraction. Davg. value drops from 7.77 × 10−6 m at 0 h, to 5.29 × 10−6 m at 0.25 h, to 4.42 × 10−6 m at 1 h, then reaches a plateau region between 4.21 × 10−6 and 4.85 × 10−6 m. Similar to area fraction and average diameter, maximum diameter also experiences a sharp decline after a settling time of 0.25 h, and the value decreases from 37.5 × 10−6 m before settling to 8.1 × 10−6 m after 0.25 h settling. Unlike area fraction and average diameter, Dmax. value continues to drop until 2 h with a gradually slowing down decline trend and becomes stable from 2 to 8 h, with the values ranging from 5.1 × 10−6 to 5.4 × 10−6 m. The above changes in area fraction, average diameter and maximum diameter imply the effect of settling on inclusions’ removal is remarkable for early settling stage and becomes progressively less obvious with increasing time. The minimum diameter and number density values exhibit little change through sedimentation process, the former vary from 1.3 × 10−6 to 1.8 × 10−6 m while the latter are between 10.0 × 107 m−2 and 11.1 × 107 m−2, indicating that settling procedure is not significant for small-sized inclusions. Based on the above, it can be concluded that large-sized inclusions exhibit obvious settling behavior, leading to the drop of the area fraction, while small-sized inclusions, whose number is much larger than the large-sized inclusions, exist through the whole settling process, thus the total number (or number density) of inclusions shows no obvious change. The effects of inclusions’ size and settling time on the melt quality will be discussed in more detail in the following text.
In this research, Stokes’ law27) is utilized to analyze the settling behavior of the inclusions in the melt, with the assumption that all the inclusions are of spherical morphology and undergoing gravitational settling, considering three types of forces (gravity, stokes drag force and buoyancy). The settling distance s as a function of settling time t can be described with the following equation:
\begin{equation} s = \frac{g(1 - \gamma)d_{i}^{2}}{18\gamma \nu}\left(t - \frac{d_{i}^{2}}{18\gamma \nu} \right) \end{equation} | (1) |
Inclusions including oxides (MgO, Nd2O3, ZnO and ZrO2) which are the most predominant non-metallic inclusions, undissolved Zr particles which are a characteristic of Zr alloying Mg alloys, MgS and MgSO4 as a result of the usage of sulphur powder are discussed in this section. The density values of MgO, Nd2O3, ZnO, ZrO2, Zr, MgS and MgSO4 are 3.58 × 103, 7.24 × 103, 5.61 × 103, 5.89 × 103, 6.52 × 103, 2.68 × 103 and 2.66 × 103 kg m−3, respectively.9,16,29) Employing eq. (1), the predicted settling behavior, settling distance vs. settling time, of different sizes (1 × 10−6, 2 × 10−6, 4 × 10−6, 8 × 10−6, 16 × 10−6 and 32 × 10−6 m) of various inclusion particles in the melt at 1033 K, are calculated and displayed in Fig. 3. The five hollow symbols of each line represent five sampling times, 0.25, 1, 2, 4 and 8 h. The dashed line indicates the sample depth, 0.2 m below the melt surface. Assuming that all inclusions drop from the surface, above the dashed line implies that the inclusions have fallen below the sampling depth while under the dashed line indicates that the inclusions are still above the depth.
The settling distance as a function of settling time for various inclusions of different sizes, (a) 1 × 10−6, (b) 2 × 10−6, (c) 4 × 10−6, (d) 8 × 10−6, (e) 16 × 10−6 and (f) 32 × 10−6 m, in the melt at 1033 K. The dashed line indicates the sample depth, approximately 0.2 m below the melt surface.
According to Fig. 3(a), when di = 1 × 10−6 m, all the lines are below 0.2 m even after 8 h settling, implying that the sedimentation procedure of as long as 8 h is meaningless when the inclusions are small. When di value goes up to 2 × 10−6, 4 × 10−6 or 8 × 10−6 m, Fig. 3(b)–(d), the density of the inclusions play the decisive role on settling, i.e., inclusions (like Nd2O3 and Zr) of large density tend to settle more rapidly and take less time to drop through the sample depth. After 8 h settling, it can be found that high-density inclusions of 2 × 10−6 m still exist in the melt while inclusions of 4 × 10−6 or 8 × 10−6 m should not be observed. When di continues to grow to 16 × 10−6 or 32 × 10−6 m, Fig. 3(e) and (f), only a short time (less than 0.25 h) is required for all the inclusions falling through 0.2 m depth. Such predicted results are basically in line with the experimental data, section 3.2. As seen in Fig. 2 and Table 2, large-sized inclusions are hardly observed after a short settling time, with a maximum diameter (Dmax.) value of 8.1 × 10−6 m at 0.25 h, while small-sized inclusions are always present, from 0 to 8 h. With large inclusions settled, the settling phenomena of inclusions becomes weak; correspondingly, the area fraction and average diameter values of inclusions level off after 1 h settling. It’s probably worth pointing out, however, there is incongruity between predicted and experimental results. Inclusions of a Dmax. value of 5.4 × 10−6 m can still be observed even at 8 h, though the predicted results show inclusions greater than 4 × 10−6 m should have fallen below the sample depth after such a long time, for example. Such incongruity results from multiple factors. For instance, the actual morphology of most inclusions is not a simple sphere; and the actual density values of the inclusions are variable, probably because inclusions could be gathered by several kinds of inclusions during gas bubble flotation,12,13) or because inclusions may exhibit a loose and porous structure.
3.4 The rising behavior of the bubbles carrying inclusions in the meltTechnically, there are three kinds of inclusions in the melt, the subsiding inclusions, the stationary inclusions, and the rising inclusions carried by the remaining Ar bubbles, at the onset of the settling procedure after 0.25 h bubbling (section 2). The sedimentation behavior leads to a reduction of inclusions while the rising behavior maintains the inclusions’ number in the samples in a short term. So, it is necessary to discuss the behavior of the bubbles carrying inclusions in the melt.
It should be reasonable to assume that the Ar bubble rising in stagnant Mg melt will reach a nearly-constant velocity quite rapidly after the injection, based on the observations on the velocity of gas bubbles rising in water in Ref. 30. Considering the high Reynolds number for the Ar bubbles rising in the melt, the classical Mondelson equation31) is utilized to calculate the terminal rising velocity:
\begin{equation} \upsilon_{b} = \sqrt{\frac{2\sigma}{d_{b}\rho_{m}} + \frac{\Delta \rho}{\rho_{m}}\frac{gd_{b}}{2}} \end{equation} | (2) |
The theoretical rising distance after 0.25 h settling as a function of Ar bubble diameter.
In practice, inclusions tend to stick to the rising Ar bubbles due to the driven force of the reduction of the system free energy. The attached inclusions will increase the overall density, narrow the density gap between the melt and the bubbles, thereby decreasing the rising velocity of the bubbles. Assuming N (N ≥ 1) inclusions are attached to one Ar bubble, the overall density of the bubble and the inclusions can be calculated with the following equation:
\begin{equation} \rho_{\textit{overall}} = \frac{N(d_{i}/d_{b})^{3}\rho_{i} + \rho_{b}}{N(d_{i}/d_{b})^{3} + 1} \end{equation} | (3) |
When the inclusions are assumed to be Zr particles of the largest size (Dmax. = 37.5 × 10−6 m) and the bubble diameter is assumed to be 3 × 10−3 m, the overall density is calculated to be increased by 0.8 to 78.5 percent when 1 to 100 Zr-inclusions are attached to the bubble. Therefore, the increased density is still far less than the melt density, and the bubbles carrying inclusions still maintain a high rising speed. Even the shortest settling time would be enough for the bubbles with inclusions rising to the melt surface. Thus, rising inclusions should not exist in the samples after 0.25 h settling. The theoretical analysis agrees with the experimental observations. Structural properties evolution of inclusions, such as the decreased area fraction and the almost constant number density at 0.25 h, Table 3, indicate that the removal of inclusions is dominated by the inclusions’ settling behavior, section 3.3, rather than the rising behavior.
The effects of settling time up to 8 h on the quality of fluxless Mg–Nd–Zn–Zr melt preparation with respect to chemistry of the melt and structural properties of inclusions were determined and analyzed with the following results.
The authors would like to acknowledge the support from National Natural Science Foundation of China (Grant no. 51875373). Yanyan Huang would also like to acknowledge the supports from the Fundamental Research Funds for the Central Universities (No. 2017SCU12021), the Aeronautical Science Foundation of China (No. 20160219001) and the China Postdoctoral Science Foundation (No. 2017M623026).