Effects of Manufacturing Parameters in Planar Flow Casting Process on Ribbon Formation and Puddle Evolution of Fe–Si–B Alloy

Abstract

The effects of manufacturing parameters in the planar flow casting process on the ribbon formation and the puddle stability of Fe₇₈Si₉B₁₃ alloy are investigated experimentally. The ribbon morphology, surface quality, and puddle geometry are examined at different conditions and the transient evolution processes of puddle for molten metal passing through a rectangular nozzle are observed. The successful operability window for the production of Fe₇₈Si₉B₁₃ ribbon is established and it is found the scope is different from that of Al-based alloy. The trend lines of the alloys on the plane of Reynolds number versus Weber number corresponding to obtain a successful ribbon are established. The ribbon thickness is found to vary with the applied pressure across the crucible and the wheel speed to the power of 0.45 and −0.9, respectively. The formation of small air pockets could be enhanced by increasing the applied pressure difference and wheel speed, or by decreasing the nozzle-wheel gap and the jetting temperature.

1. Introduction

Rapidly solidified Fe–Si–B ribbons have outstanding soft magnetic properties and have been widely used in transformers and electric motors as iron cores to promote their conversion efficiency and performance.¹⁾ There are two major processes for continuous casting of rapidly quenched amorphous ribbons which are chill-block melt spinning (CBMS)²⁾ and planar-flow melt spinning (PFMS),³⁾ respectively. The latter planar flow casting process is the most popular one among various rapid solidification processing methods in both academic and industrial communities due to its capability of mass production for wide, thin, and uniform ribbons. The typical system apparatus as shown in Fig. 1 consists of a chill wheel, wheel driving system, wheel cooling system, crucible system, argon gas pressure system, and temperature control system. The applied argon pressure forces the molten metal in the crucible to flow through a rectangular nozzle onto the chill wheel, forming a puddle that is sustained by the surface tension between the nozzle and wheel. The wheel cools and solidifies the molten metal which could be pulled continuously out from the beneath portion of the puddle to collect solidified ribbon at the same flow rate with the melt ejecting onto the wheel.

Fig. 1.

Schematic of the planar flow casting process and blow-up of the puddle region. (Online version in color.)

Amorphous ribbons with high surface quality are the prerequisite for using them in transformers and electric motor cores.¹⁾ The quality of the ribbon produced by the PFMS process depends heavily on the shape and stability of the puddle formed between the nozzle and the copper chill wheel.⁴⁾ Many studies have found that the fluid dynamics and heat transfer characteristics dominate the stability of the puddle significantly.^5,6) One can obtain a continuous ribbon with good quality only when a dynamic equilibrium is established in the puddle. Several parameters in the planar flow casting process may influence the surface quality, crystalline structure, and soft magnetic property of the ribbon such as the applied pressure difference, wheel speed, jetting temperature, nozzle-wheel gap, and width of nozzle slot. The material properties also play important roles in the process such as the metal density, viscosity, surface tension, and thermal conductivity. The operating parameters must be inside the triangular window of the operability diagram for the successful production of ribbon.⁴⁾ However, such a scope is still not available for Fe-based alloy.⁷⁾

The experimental investigation of the planar flow casting process is difficult due to the severe conditions like high wheel speed, high jetting temperature, small dimensions of the active zone, and short duration of the process. Therefore, numerical modeling based on computational fluid dynamics (CFD) has been widely used to simulate the casting process.^6,8) Although the fluid flow and temperature distribution in the puddle have been predicted numerically,^6,8,9) experimental studies regarding the variation of puddle geometry with operating parameters in the casting process are still limited. Some investigations have used a high-speed camera together with back light technique to observe the transient behavior of the puddle.^10,11) These works were carried out for Al-based alloy. Recently, many reports have presented the results of Fe-based alloys for the high-speed images of the puddle at different parameters^7,12) and the surface quality of the ribbons under different wheel speeds.¹³⁾ However, a detailed parametric study for the ribbon formation and puddle evolution of Fe-based alloy is still absent.

The present study investigates the effect of manufacturing parameters in the planar flow casting process on the ribbon formation and the puddle stability of Fe₇₈Si₉B₁₃ alloy. Experiments were conducted to examine the ribbon morphology, surface quality, and puddle image. Moreover, the transient behavior of puddle formation for molten metal passing through a rectangular nozzle was also observed. Results provide more complete understanding for the characteristics of ribbon formation of Fe-based alloy.

2. Experimental Procedure

A laboratory-scale melt spinner was employed to analyze the effects of processing parameters on the rapidly solidified Fe–Si–B ribbons. The single roller melt spinning apparatus is shown in Fig. 2. The system mainly consists of the chill wheel, wheel driving system, crucible system, gas pressure system, and temperature control system. The major function of the chill wheel is to quench the molten metal rapidly in order to produce amorphous or microcrystalline metal. The chill wheel is made of Cu-based alloy because of its high thermal conductivity. The wheel driving system may dictate the wheel speed by controlling the rotation speed of the motor. The crucible system contains a reservoir in which the metal melts and delivers the melt through the nozzle onto the rotating chill wheel. Argon gas is used in the gas pressure system to provide ejection pressure in the crucible system and force the molten metal to flow through the nozzle slit. It may be controlled to exert a constant pressure difference on the melt during the casting process. The temperature of melt was controlled by the temperature control system using the high-frequency-wave coil.

Fig. 2.

Photograph of the apparatus of planar flow casting process. (Online version in color.)

The composition of the alloy used in the present study is Fe₇₈Si₉B₁₃. For each run of casting, an appropriate power was set for the high-frequency-wave heater to melt an alloy bar of 20 grams in the crucible. The melting process generally lasts about 3 minutes. The nozzle screwed into the bottom of the crucible is in the form of a rectangular slit with a breadth of 0.4 mm and length of 17 mm. The chill wheel was polished each time before performing experiment. The melted alloy was ejected through the nozzle onto the rotating chill wheel of 250 mm diameter and 40 mm width. Accordingly, a 17-mm-wide ribbon could be fabricated continuously. The system parameters such as the applied pressure difference (ΔP) between the crucible and the ambient gas, wheel speed (U), nozzle-wheel gap (G), and jetting temperature (T_j) are adjusted systematically to explore their effects on the ribbon quality. The details of experimental cases are given in Table 1. The ribbon thicknesses are measured at 60 different points along the ribbon within the length of 3 m and then the mean thickness (T) is determined. The ribbon structure was characterized by the x-ray diffraction (XRD) technique and the surface quality on the side contacting the chill wheel is examined by the optical microscope micrographs. The evolution of puddle formation during melt spinning process is recorded using a high-speed imaging system at the speed of 4800 frames/s.

Table 1. The experimental cases and resultant ribbon thicknesses and structures.

Run no.	ΔP (kPa)	U (m/s)	G (mm)	Tj (°C)	T (μm)	Structure
1	20	25	0.1	1280	25.3±0.7	Amorphous
2	20	25	0.15	1280	27.9±0.6	Amorphous
3	20	25	0.2	1280	Failure	–
4	20	17.5	0.1	1280	Failure	–
5	20	20	0.1	1280	30.2±0.7	Amorphous
6	20	22.5	0.1	1280	27.1±0.7	Amorphous
7	20	27.5	0.1	1280	24.3±0.8	Amorphous
8	20	30	0.1	1280	Failure	–
9	25	25	0.1	1280	28.6±0.8	Amorphous
10	30	25	0.1	1280	32.0±0.7	Amorphous
11	35	25	0.1	1280	33.8±0.8	Amorphous
12	20	25	0.1	1330	27.6±0.8	P.A.
13	20	25	0.1	1380	29.7±0.7	P.A.
14	20	25	0.1	1430	32.2±1.0	P.C.

P.A.=primarily amorphous; P.C.=partially crystalline.

3. Results and Discussion

3.1. Operability Window and Ribbon Thickness

The interaction of all parameters in the planar flow casting process can be understood through the plot of Weber number (We) versus the dimensionless pressure parameter ΔP/(2σ/G) as shown in Fig. 3 for the successful production of ribbon. The present experimental data are compared with two previous works^14,15) in which the results were published in tabular forms. Figure 3 shows the distribution of Fe-based and Al-based alloys in the operability window. It reveals that the value of the dimensionless pressure parameter is between 0.5 and 1 and the Weber number is between 80 and 150 for the successful production of continuous Fe₇₈Si₉B₁₃ ribbon without any defect. This result is different from that of Al-based alloy¹⁵⁾ in which the distribution of successful cases is within the zone with the dimensionless pressure parameter above unity.¹⁶⁾ Figure 4 shows the variation of Reynolds number (Re) versus Weber number for the Fe-based and Al-based alloys. For Al-based alloy, one can see that the values of Reynolds number corresponding to the successful cases of ribbon production are generally over 5000. This phenomenon is due to the relatively lower viscosity of Al-based alloy and more significant inertia effect which may enhance the flow rate of the melt passing through the nozzle. As a result, the upstream zone of the puddle will expand and be sustained stably and effectively when the dimensionless pressure parameter, ΔP/(2σ/G), is higher than one. In contrast to the result of Al-based alloy, the values of Re of Fe-based alloy are less than 5000. Because of the higher viscosity of Fe-based alloy, the upstream zone of the puddle is more difficult to expand which may cause the ribbons to be produced successfully with the dimensionless pressure parameter less than one. The trend line equation of the distributional scope of Al-based alloy is   

$$y=102.1x+7\mathrm{\hspace{0.17em} }535.8$$(1)

it can be rearranged and yield   

$$\frac{2\sigma }{\mu U}=102.1+7\mathrm{\hspace{0.17em} }535.8\left(\frac{2\sigma }{\rho U^{2}G}\right)$$(2)

where ρ=2300 kg/m³ is the density, σ=0.86 N/m the surface tension, and μ=0.001 kg/(ms) the viscosity for Al-based alloy. We can estimate the suitable value of wheel speed (U) for a specified nozzle-wheel gap (G) to obtain perfect ribbon. Similarly, the trend line equation for Fe-based alloy is derived from the present results and the work of Smith and Saletore¹⁴⁾ and leads to   

$$\frac{2\sigma }{\mu U}=11.9+411.4\left(\frac{2\sigma }{\rho U^{2}G}\right)$$(3)

where ρ=6817 kg/m³, σ=1.73 N/m and μ=0.012 kg/(ms) for Fe₇₈Si₉B₁₃ alloy. The typical suggested values of U with G are given in Table 2 for the manufacture of Fe-based and Al-based metal ribbons.

Fig. 3.

Distribution of the experimental cases of Fe-based and Al-based alloys in the operability window for the planar flow casting process. (Online version in color.)

Fig. 4.

The trend lines on the (Re, We)-plane for the planar flow casting process of Fe-based and Al-based alloys. (Online version in color.)

Table 2. Variations of suggested wheel speed U with nozzle-wheel gap G in the casting process for Fe-based and Al-based alloys.

Fe78Si9B13		Al-7%Si
G (mm)	U (m/s)	G (mm)	U (m/s)
0.15	17.8	0.8	9.8
0.20	20.0	0.9	11.5
0.25	21.1	1.0	12.4
0.30	21.7	1.1	13.0
0.35	22.2	1.2	13.4
0.40	22.5	1.3	13.8
		1.4	14.0
		1.5	14.3

Figure 5 shows the variation of the dimensionless thickness parameter T/G with Euler number (Eu=ΔP/(ρU²)) at assigned values of B/G for both Fe-based and Al-based alloys. It is evident that ribbon thickness increases with the pressure difference and decreases with the wheel speed. The ribbon thickness of Fe₇₈Si₉B₁₃ alloy is found to vary with the pressure difference ΔP to the power of 0.45 approximately, and with the wheel speed U to the power of −0.9. This result is in accordance with the prediction of the Bernoulli equation for PFMS process,¹⁷⁾ but it is different from the experimental data of Al-based alloy.¹⁵⁾ The result also indicates that the ribbon thickness generally increases with the width of nozzle slot B.

Fig. 5.

Variations of dimensionless thickness parameter (T/G) with Euler number (Eu) at assigned values of B/G for Fe-based and Al-based alloys. (Online version in color.)

3.2. Structure Characterization

All the present experimental cases are shown in Table 1. The result shows that the jetting temperature is an important factor which may affect the structural characterization of the ribbons. In the rapidly solidified process, the thickness of amorphous phase reduces with the jetting temperature.⁷⁾ Therefore, the high jetting temperature leads to the formation of crystalline structure. Figure 6 illustrates the typical XRD patterns for the air-side surface of the ribbon. The amorphous pattern is shown in Fig. 6(a) at the jetting temperature of 1280°C, which is superheated 100°C over the melting temperature. Figure 6(b) shows the primarily amorphous pattern at the jetting temperature of 1330°C and 1380°C. Figure 6(c) displays the partially crystalline pattern at the jetting temperature of 1430°C. The diffraction patterns for all ribbons show a broad amorphous halo at the approximate location of the α-Fe (110) peak (2θ=44.7°). There are no sharp peaks of crystalline structure over the range of the diffraction scans as shown in Fig. 6(a), which represents the short-range order but long-range disorder amorphous structure. Figure 6(b) reveals that the pattern is detected a slight crystalline peak at the approximate location of the α-Fe (200) peak (2θ=64.9°). There are few α-Fe (200) produced because this intensity is not strong enough. Figure 6(c) has a detectable crystalline peak at the approximate location of the α-Fe (200) peak. This pattern shows that many α-Fe (200) crystals are formed.

Fig. 6.

The typical XRD patterns of the ribbons obtained in the experiments. (a) amorphous (no. 1), (b) primarily amorphous (no. 12), and (c) partially crystalline (no. 14).

3.3. Ribbon Formation, Surface Quality and Puddle Geometry

Figure 7 shows the surface morphology of the ribbons, optical micrograph on the wheel-side ribbon surface, and the corresponding high-speed images of the puddle at different operating parameters. The jetting temperature is fixed constantly at 1280°C for all the cases. Figure 7(a) reveals that there are many fractures on the ribbon surface because the wheel speed is too high to give sufficient contact time for the melt and the wheel to complete solidification. The other unsuccessful case is shown in Fig. 7(g) that the bad edge quality appears on the ribbon. The main reason is the increment of the nozzle-wheel gap results in lower constriction for the puddle. The unstable puddle is easier to occur and accordingly influences the quality of the ribbon edges. Another failed case is shown in Fig. 7(m) that the ribbon appears to have serrated edges. This phenomenon is mainly due to the low wheel speed which may also reduce the production rate of the ribbon. However, the driving pressure is constant, so the melt may extend toward two sides of the nozzle which may result in the onset of flow instability on both sides of the puddle. Figures 7(d), 7(j), and 7(p) are the typical successful cases that the ribbons with smooth surfaces and edges could be produced continuously. Note that the thickness of the case 7(p) is greater than 30 μm while those of the cases 7(d) and 7(j) are less than 30 μm.

Fig. 7.

Surface morphology of the ribbons, optical micrograph on the wheel-side surface, and high-speed images of the puddle at assigned values of the applied pressure difference ΔP, wheel speed U, and nozzle-wheel gap G. (a)–(c) ΔP=20 kPa, U=30 m/s, G=0.1 mm; (d)–(f) ΔP=20 kPa, U=25 m/s, G=0.15 mm; (g)–(i) ΔP=20 kPa, U=25 m/s, G=0.2 mm; (j)–(l) ΔP=20 kPa, U=25 m/s, G=0.1 mm; (m)–(o) ΔP=20 kPa, U=17.5 m/s, G=0.1 mm; (p)–(r) ΔP=35 kPa, U=25 m/s, G=0.1 mm. (Online version in color.)

The optical micrographs on the wheel-side ribbon surface were observed for these six typical cases and it is found that the contact surfaces were all full of air pockets which were caused by the capture of air on the wheel-melt interface. The air pockets result in poor contact between the ribbon and chill wheel surface. Thus, the ribbon surface with more air pockets implies higher thermal contact resistance. The formation of air pockets is strongly influenced by the process parameters. An increase in the applied pressure difference and wheel speed, or a decrease in the nozzle-wheel gap and jetting temperature, will promote the formation of small air pockets. It can increase the contact area and improve the efficiency of heat conduction on the wheel-melt interface. Consequently, a higher cooling rate can be achieved. These observations are in agreement with previous studies.^14,18) The work of Hung and Fiedler¹⁸⁾ also indicated that an increase in the wheel speed can enlarge the contact area between the melt and the wheel, and thus promote the heat transfer efficiency on the wheel-melt interface.

Figure 7(c) reveals that an over-high wheel speed may result in the rupture of downstream meniscus of the puddle. This behavior may cause severe fluctuation in the downstream region of the puddle and produce many fractures on the ribbon surface. Figure 7(o) shows a longer puddle than that in Fig. 7(l). This phenomenon is mainly due to the lower wheel speed and cause a lower ribbon production rate. Simultaneously, the constantly applied pressure difference tends to lengthen the length of the puddle. This characteristic brings about the onset of instability on both sides of the puddle and makes the occurrence of serrated edges on the ribbon. Figures 7(f), 7(i), and 7(l) show that an increment of the nozzle-wheel gap may affect the flow of melt in the puddle and increase the downstream length of the puddle. The overlong downstream length may cause the puddle to be unstable and fracture the edges of the ribbon. The results illustrated in Figs. 7(l) and 7(r) show that an increase of ΔP increases the lengths of the puddle in both upstream and downstream directions relative to the nozzle. This effect not only lengthens the puddle and also increases the produced ribbon thickness.

3.4. Puddle Development and Evolution

The quality of ribbon produced by planar flow casting process depends heavily on the puddle stability during the manufacturing process. A high-speed imaging system has been used to record the images in each case. The typical results are shown in Fig. 8 for the transient evolution process of puddle after melt ejection. The operating condition is the run number 5 in Table 1 in which the value of the applied pressure difference (ΔP) is 20 kPa, wheel speed (U) 20 m/s, nozzle-wheel gap (G) 0.1 mm, width of nozzle slot (B) 0.4 mm, and jetting temperature (T_j) 1280°C. The thickness of produced ribbon is about 30 μm. By observing the images from Figs. 8(a) to 8(g), one can see that the melt emerges from the nozzle slit at 0.2 ms, and then touches the surface of the chill wheel at 0.8 ms. Subsequently, the melt is pulled downstream by the rotating chill wheel. The puddle is formed gradually until 5.0 ms and reaches steady state at about 34.2 ms.

Fig. 8.

The high-speed images for the transient evolution process of puddle after melt ejection. (Online version in color.)

The puddle maintains stable state till 107.5 ms. The images of Figs. 8(h) to 8(n) reveal the puddle variation from stable to unstable state. At the beginning, the fluctuation happens at the upstream meniscus of the puddle. It leads to the occurrence of periodic oscillation and causes the alteration of puddle length. The period of oscillation is about 3 ms. The main reason is probably due to the rise of wheel temperature which may reduce the heat conduction rate between the ribbon and chill wheel, and make the ribbon thickness decrease accordingly. However, under the condition of constant ΔP, this effect may lengthen the puddle and increase the contact area between the melt and the wheel. As a result, the heat transfer rate rises. Such an effect increases the ribbon thickness and also shortens the puddle length. The interaction between these two mechanisms causes the oscillation of the puddle. Finally, the upstream meniscus blows out and the melt sprays toward the upstream.

3.5. Effect of Jetting Temperature

The effect of jetting temperature could be indicated by the liquid Stefan number Ste_L which is defined as Ste_L=C_p(T_j−T_m)/ΔH_f, where C_p is the constant pressure specific heat, T_j the jetting temperature, T_m the melting temperature, and ΔH_f the latent heat of fusion. Figure 9 shows the variations of the dimensionless thickness parameter T/G and Reynolds number Re with the liquid Stefan number based on the experimental data of run numbers 1 and 12–14 as listed in Table 1. It can be observed that the ribbon thickness and Reynolds number increase monotonically with the jetting temperature. Because the viscosity of the melt decreases with increasing Ste_L, it is expectable that the Reynolds number also increases with Ste_L. This characteristic increases the flow rate of the melt passing through the nozzle and produces thicker ribbons as shown in the curve of T/G versus Ste_L. Figure 10 shows the optical micrographs of ribbons on the wheel-side contact surface at different jetting temperatures. The higher jetting temperature enhances the formation of large air pockets, which may reduce the contact surface area between the ribbon and chill wheel. Consequently, the rate of heat transfer on the melt-wheel interface may reduce and improve the formation of crystalline phase due to this effect. Some previous studies in the literature have suggested that the effect of surface tension is the dominant factor that dominates the puddle stability at high jetting temperature.^7,12) Because the surface tension reduces with increasing jetting temperature, the stability of puddle also will decrease and become more unstable at higher jetting temperature. As a result, it is easier to form large air pockets on the wheel-side ribbon surface.

Fig. 9.

Variations of T/G and Re with Ste_L for the ribbon. (Online version in color.)

Fig. 10.

Optical micrograph on the wheel-side ribbon surface at different jetting temperature. (a) 1280°C, (b) 1330°C, and (c) 1430°C. (Online version in color.)

4. Conclusions

Rapidly solidified Fe₇₈Si₉B₁₃ ribbons have been fabricated by using a single roller melt spinning apparatus in the present work. Different parameters in the planar flow casting process have been considered and adjusted systematically in the experiments to examine the effect of each parameter on the ribbon formation and puddle stability. The results are summarized as follows:

(1) The operability window for the successful production of Fe₇₈Si₉B₁₃ ribbon has been established. Its range distributes between 0.5 and 1 of the dimensionless pressure parameter ΔP/(2σ/G), and between 80 and 150 of the Weber number.

(2) An appropriate chill wheel speed corresponds to the nozzle-wheel gap can be inferred from the trend line on the (Re, We)-plane in order to produce ribbon successfully.

(3) The ribbon thickness varies with the applied pressure difference to the power of 0.45 and with the wheel speed to the power of −0.9 approximately. The result is quite close to the prediction of the Bernoulli equation for planar flow casting process that the thickness varies with the applied pressure difference to the power of 0.5 and with the wheel speed to the power of −1.0.

(4) The XRD patterns reveal that the crystalline phase is easier to appear at higher jetting temperature. The ribbon possesses completely amorphous phase at the jetting temperature of 1280°C, and partially crystalline phase at the jetting temperature of 1430°C. A higher jetting temperature also increases the ribbon thickness due to the reduction of melt viscosity.

(5) High wheel speed causes insufficient contact time for the melt to complete solidification. Accordingly, many fractures appear on the ribbon surface. On the contrary, low wheel speed tends to make the melt extend toward both sides of the nozzle. This makes the puddle unstable and produces serration on the ribbon edges.

(6) Optical micrographs on the wheel-side ribbon surface illustrates that an increase of the applied pressure difference or wheel speed, or a decrease of the nozzle-wheel gap or the jetting temperature may promote the formation of small air pockets on the ribbon surface. Therefore, the contact area at the melt-wheel interface increases which improves the heat transfer efficiency between the melt and chill wheel.

(7) The transient evolution process of melt puddle after ejection has been examined in details by high-speed images, including the initial development, stable configuration, and onset of instability. The result shows that a high wheel speed may cause the rupture of the downstream meniscus of the puddle, while a low wheel speed may induce the oscillation of the puddle. Both phenomena will fail the ribbon production.

Further studies focusing on the effect of surface roughness of chill wheel and thermostatic control of the wheel cooling system on the ribbon formation and puddle stability will be helpful to improve the system performance.

Acknowledgments

The support for this work from China Steel Corporation through the grant number 02T1D-RE003 is gratefully acknowledged.

References

• 1)   R.  Hasegawa: J. Magn. Magn. Mater., 215–216 (2000), 240.
• 2)   H. H.  Liebermann and  C. D.  Graham: IEEE Trans. Magn., 12 (1976), 921.
• 3)   M. C.  Narasimhan: US Patent 4142571, (1979).
• 4)   P. H.  Steen and  C.  Karcher: Annu. Rev. Fluid Mech., 29 (1997), 373.
• 5)   S. L.  Wu,  C. W.  Chen,  W. S.  Hwang and  C. C.  Yang: Appl. Math. Model., 16 (1992), 394.
• 6)   M.  Bussmann,  J.  Mostaghimi,  D. W.  Kirk and  J. W.  Graydon: Int. J. Heat Mass Tran., 45 (2002), 3997.
• 7)   M.  Srinivas,  B.  Majumdar,  G.  Phanikumar and  D.  Akhtar: Metall. Mater. Trans. B, 42B (2011), 370.
• 8)   H.  Liu,  W.  Chen,  S.  Qiu and  G.  Liu: Metall. Mater. Trans. B, 40B (2009), 411.
• 9)   H.  Liu,  W.  Chen and  G.  Liu: ISIJ Int., 49 (2009), 1895.
• 10)   P. D.  Wilde and  E. F.  Matthys: Mater. Sci. Eng., A150 (1992), 237.
• 11)   C. J.  Byrne,  E. A.  Theisen,  B. L.  Reed and  P. H.  Steen: Metall. Mater. Trans. B, 37B (2006), 445.
• 12)   B.  Majumdar,  M.  Sowjanya,  M.  Srinivas,  D. A.  Babu and  T. K. K.  Reddy: Trans. Indian Inst. Met., 65 (2012), 841.
• 13)   S.  Sohrabi,  H.  Arabi,  A.  Beitollahi and  R.  Gholamipour: J. Mater. Eng. Perform., 22 (2013), 2185.
• 14)   M. T.  Smith and  M.  Saletore: Rev. Sci. Instrum., 57 (1986), 1647.
• 15)   J. K.  Carpenter and  P. H.  Steen: J. Mater. Sci., 27 (1992), 215.
• 16)   C. J.  Byrne,  S. J.  Weinstein and  P. H.  Steen: Chem. Eng. Sci., 61 (2006), 8004.
• 17)   H.  Fiedler,  H.  Mühlbach and  G.  Stephani: J. Mater. Sci., 19 (1984), 3229.
• 18)   S. C.  Huang and  H. C.  Fiedler: Mater. Sci. Eng., 51 (1981), 39.