Microstructure evolution of 15 wt% boron carbide/aluminum composites during liquid-stirring process

Abstract

Microstructure evolution of 15 wt% boron carbide particle reinforced aluminum matrix composites (B₄C/Al composites) with titanium addition during liquid-stirring process was dynamically characterized in this paper. B₄C particles were rapidly dispersed under the mechanical stirring. Many B₄C clusters were formed in the melt before 20 min, but gradually scattered in matrix beyond 20 min, owing to further reactive wetting through interface reaction in addition to stirring. After rapid improvement, distribution uniformity slowly approached to completely uniform distribution during 20–55 min, even better than random distribution at 55 min. Interface reaction produced Al₃BC, TiB₂, and AlB₂ by B₄C erosion and Al₃Ti decomposition; however, AlB₂ only precipitated in matrix after long time stirring. The growth of TiB₂ transformed from a fine layer to discretely coarse crystals on the B₄C surface. Reaction mechanism and relationship between reactive wetting and particle dispersion were discussed.

Keywords

Metal–matrix composites boron carbide particle dispersion interface reaction reactive wetting

Introduction

Boron carbide, known for its high hardness (9.5+ in Mohs scale), low density (2.52 g/cm³), good thermal stability (melting point 2427℃), high chemical, and wear resistance, is a promising covalently bonded ceramic material used as reinforcement in Al matrix.^1–3 In recent years, B₄C/Al composites have been widely applied in the nuclear industry due to the high ability of B¹⁰ to capture neutrons (3850 barns).^4,5 B₄C/Al composite of 15 wt% is one of the neutron absorber materials used for spent fuel storage and transportation application.⁶ Among all the manufacturing methods, liquid-stirring casting is a practical technique to produce B₄C/Al composites since it is cost-saving and offers a wide selection of materials and processing conditions, which introduces reinforced particles into molten metal through a vortex created by a mechanical impeller.^7,8 Two aspects of particle distribution and interfacial reaction have to be seriously considered during liquid-stirring process of B₄C/Al composites,⁹ which directly determine preparation success and mechanical property of B₄C/Al composites.

Particle distribution greatly influences the mechanical properties of metal matrix composites such as fracture strength and plastic deformation. A clustered distribution is easier for crack propagation than uniform one because of the constrained deformation in clustered matrix.¹⁰ Damage accumulation preferentially occurs in clustered regions ahead of a propagating crack tip.¹¹ Higher ductility is exhibited by composites with more uniform particle distribution, especially when particle distribution becomes more regular than that of random one.¹² Mechanical property of B₄C/Al composites is also dependent on interfacial microstructure as the load transfers to reinforcements through interface.¹³ Interfacial reaction layer thinner than one micron is thought to be beneficial to mechanical property since it usually enhance both chemical and mechanical bonding between particle and matrix.¹⁴ Therefore, in order to obtain uniform distribution of B₄C particles with few clusters and well bonded interface, it is important to find out the real evolution of particle dispersion and interfacial microstructure in the B₄C/Al composites.

However, besides poor wetting by liquid Al, B₄C strongly reacts with molten Al to produce Al₃BC and AlB₂ phases, resulting in rapid quantity increase of total secondary phases in B₄C/Al melt.^2,15,16 Increased particle quantity deteriorates the fluidity of melt and brings difficulty in followed casting process.¹⁷ To suppress the interfacial reaction, active element Ti is chosen to add into melt using K–Al–Ti–F flux or Al–Ti master alloy.^18,19 Additive Ti will participate in interface reaction and form a protective layer surrounding B₄C particles, hindering further reaction and promoting wetting. Although researches have focused on the interface reaction in the past, few investigated on how interface varied at the stirring stage. It is practical to figure out the physical and chemical process of interface reaction. On the other hand, since they simultaneously take place during stirring process, associated effect of mechanical stirring and reactive wetting on particle dispersion is expected to be existence. Information should be obtained to reveal the change of B₄C distribution and interface, as well as their interaction during melt stirring process of B₄C/Al composites.

This article aims to study microstructure evolution of B₄C/Al composites during dynamically liquid-stirring stage, trying to obtain information about practical manufacturing process and reveal the relationship between B₄C particles dispersion and interfacial reaction. Observation of interface and quantitative characterization of B₄C particles distribution were carried out to understand the stirring process of B₄C/Al composites by liquid-stirring casting technology.

Experimental procedure

To prepare the B₄C/Al composites, commercially pure Al and Al–Ti (5 wt%) master alloy were first melt in a stirring-casting furnace under vacuum condition (20 Pa). The temperature was kept at 730℃ after long time holding. B₄C powders with average diameter D[4,3] = 24.937 µm and span of size distribution λ = 1.34 were added into the melt.²⁰ Masses of all materials were designed for B₄C/Al composites consisting of 15 wt% B₄C and 2.5 wt% Ti. Then melt was agitated by a mechanical stirrer to incorporate B₄C particles. The stirring speed was constant at 500 r/min. A transition period about 5 min was necessary for total entrance of B₄C particles and formation of stable vortex, after which samples were dynamically captured at the same position of surface part of the melt. Sampling time is listed in Table 1. The whole stirring process was under protective atmosphere of argon. Molten samples directly cooled down to room temperature in air for following characterization.

Table 1.

Sampling time of specimens during stirring process.

Sample number	1	2	3	4	5	6	7	8
Time (min)	2	5	10	15	20	30	40	55

Solidified samples with size 3 × 10 × 10 mm³ were polished by both ion beam (IB-09020CP, JEOL) and mechanical polishing methods. Mechanical polishing was down to 1µm by diamond suspension. A scanning electron microscopy (SEM, TESCAN MIRA 3 LMH), in back scattered electron (BSE) mode, equipped with Energy Dispersive Spectrometer (EDS), was used to observe the interfacial microstructure and particle distribution. Software Image-Pro Plus 6.0 was used to quantify volume fraction of B₄C particles in matrix, which was statistically measured on at least 10 BSE pictures of each sample under 500 × magnification. X-ray diffraction test (Dlmax 2500) was performed on solid and powder samples to identify reaction products. Powder sample was obtained by dissolving solid sample in dilute hydrochloric acid solution (1.2 mol/L) to eliminate the diffraction of Al matrix.

To characterize uniformity of particles distribution, Dirichlet tessellation, introduced in Murphy et al.¹² and Boselli et al.,²¹ was carried out on BSE pictures of B₄C/Al composites. Basically, Dirichlet tessellation involves constructing an enclosed cell around the centroid of each particle, which contains the region of space that is closer to the centroid of that particle than to any other. Then, one can use the statistics data of cell areas to characterize the distribution uniformity of B₄C particles. Two parameters of coefficient of standard deviation of cell areas and clustering parameter (P) are derived from Dirichlet tessellation. Clustering parameter P is defined as follows

P = V_{1} / V_{2}

(1)

where V₁ is the variance of cell areas of real distribution and V₂ is the variation of cell areas of corresponding random distribution.

Results and discussion

Figure 1 presents the distribution of B₄C particles during dynamically melt-stirring process. With increase of stirring time, the quantity of B₄C particles (black) gradually increases while that of Al₃Ti particles (white) inversely decreases in the captured samples. Al₃Ti completely disappears when stirring time is up to 40 min, demonstrating the occurrence of continuous reaction. Besides, B₄C and Al₃Ti do not contact with each other, occupying different regions in Al matrix, respectively. This phenomenon implies that liquid diffusion takes place for the decomposition of Al₃Ti. It is impressed that B₄C particles tend to form clusters at the early stage of liquid stirring. Especially when stirring time is less than 15 min, aggregation of B₄C particles is the main distribution state in Al matrix. As stirring time increases, the evolution of particles distribution comes out that more B₄C clusters arise in the melt before 20 min. These particle clusters gradually scatter from 20 to 40 min, and are finally dispersed to individual particle at 55 min. Statistical results of volume fraction of B₄C and Al₃Ti particles are presented in Figure 2. Volume fraction of B₄C particles in the captured samples rapidly increases at early stirring stage, reaching 16.5 vol.% at 20 min, while total quantity varies slightly and fluctuates around the designed volume fraction (15.96 vol.%) during last 20–55 min. However, the content of Al₃Ti nonlinearly decreases as stirring time increases, illustrating inhomogeneous decomposition of Al₃Ti. Required 20 min is important for the fabrication of 15 wt% B₄C/Al composites: the formation of many B₄C clusters and rapid dispersion of B₄C particles take place before 20 min; B₄C clusters are gradually broken and volume fraction is balanced beyond 20 min.

Figure 1.

Distribution of B₄C particles in the melt at different stirring time.

Figure 2.

Variation of volume fraction of B₄C (a) and Al₃Ti (b) verses stirring time in the captured samples.

Figure 3 shows the change of distribution uniformity of B₄C particles with stirring time. As is presented in Figure 3(a), distribution uniformity rapidly increases with time since the decrease of standard deviation coefficient of cell areas in Dirichlet tessellation, which is close to linear relationship before 20 min. Under this homogenization rate, it is expected that complete distribution uniformity of B₄C particles can be realized in a reasonably short time. However, a transition of homogenization rate comes up when stirring time is around 20 min. A much slower process of further dispersion of B₄C particles lasts to the final time, which implies that the effect of high-speed stirring is weakened during this period. We also compare the distribution of B₄C particles with randomly uniform distribution with equal particles number in Figure 3(b). The change of clustering parameter P is similar with that of standard deviation coefficient in Figure 3(a). It is worth noting that P approaches 1 when stirring time reaches 20 min. But it is less than 1 only after 55-min stirring, which illustrates that long holding time is needed to obtain uniform distribution better than the random one.

Figure 3.

Change of distribution uniformity of B₄C particles with stirring time: (a) standard deviation coefficient of cell areas and (b) clustering parameter P.

A simple diagram is shown in Figure 4 to understand the change of B₄C distribution. Solid particles suspended in the stirring melt are subjected to various forces or mechanisms such as sedimentation due to gravity and buoyancy (Stocks law), surface tension, vortex flow, and solidification process. However, due to the small density difference between B₄C particle and Al melt and fast solidification speed, the effect of sedimentation and solidification on particles distribution could be ignored in current discussion.^15,22,23 In fact, it is generally accepted that the mechanical stirring with high power output mainly promotes the dispersion of solid particles in a stirring vessel.^24–27 Strongly axial and radial flows were observed in this experiment, crushing B₄C particles through shearing stress,²⁴ as shown the shearing band in Figure 4(a). Besides, the trailing vortex created by stirring impeller with high turbulence energy effectively makes particle clusters broken and dispersed.²⁵ The combined effect of shearing and turbulent flow contributes to not only rapid dispersion of B₄C particles but also uniform distribution in a reasonable time. However, surface tension should also be taken into account during dispersion process of B₄C particles. The wettability of B₄C by liquid Al is poor with an starting wetting angle over 145° and still more than 110° after 100-min holding at even 900℃.² Nonwetting results in strong exclusion of B₄C particles by Al melt. Although mechanical stirring makes B₄C clusters broken and dispersed, large surface tension will spontaneously push B₄C particles into clusters again to decrease contact area, as shown in Figure 4(b). Therefore, many B₄C clusters prefer to exist in the melt at limited stirring time. Even though stirring time reaches 40 min, there still have some clusters in the matrix, seen in Figure 1 (40 min). However, completely uniform distribution of B₄C particles is finally obtained, even better than random distribution at 55 min, which illustrates that the wetting is effectively improved. Thus, corresponding evolution of interface through interfacial reaction should be focused on.

Figure 4.

Image of shearing band and B₄C clusters (a) and schematic diagram of stresses on B₄C cluster in the stirring vessel (b).

Phase identification results of selected samples with stirring time 5 min, 20 min, and 55 min are shown in Figure 5. It is obvious that the phases of B₄C/Al composites become complex as time increases. When stirring time is 5 min, only the diffraction peaks of Al, Al₃Ti, and slight B₄C are observed, whereas new phases Al₃BC and TiB₂ are obviously detected after 55-min stirring. The relative intensity of Al₃Ti decreases and that of B₄C and TiB₂ inversely increases along with time. Different change of diffraction intensity of Al₃Ti and TiB₂ illustrates that severe reaction occurs in the Al–B₄C–Ti melt, which features for the transfer of Ti element from Al₃Ti to TiB₂ along with time. SEM images of reaction products in B₄C/Al composites are shown in Figure 6. A protective coating of TiB₂ is already covered on all B₄C surface even though reaction time is short at 5 min, as shown in Figure 6(a). Intensity of Ti element in corresponding EDS map becomes abundant as time increases. Since the increasing quantity of TiB₂, a few TiB₂ crystals even fall into the matrix after long time holding, as shown in Figure 6(c). When time reaches 55 min, a gray phase with larger size also precipitates in the matrix in addition to TiB₂. To identify this phase, the result of slow-scanning XRD test in range 60.5°–62.5° is shown in Figure 7. Fitting curve agrees well with original diffraction spectrum. A clear peak shoulder is detected and identified as the diffraction of AlB₂, which has a similar lattice structure to TiB₂.¹⁷ Diffraction spectrum can be divided into TiB₂, boron carbide, and AlB₂, respectively. Thus, interfacial reaction not only leads to the formation of TiB₂ and Al₃BC but also AlB₂ if reaction time is sufficient. We should notice that all Al₃Ti has been completely decomposed when AlB₂ is formed.

Figure 5.

XRD spectrums of solid samples with different stirring times.

Figure 6.

Images of B₄C/Al composites at different time: (a) 5 min, (b) 20 min, and (c) 55 min and the insets are corresponding EDS maps of Ti element.

Figure 7.

Refined XRD spectrums of powder sample with 55-min holding after peaks fitting and dividing treatments.

Figure 8 presents the typical interface microstructure of B₄C/Al composites after stirring for (a) 5 min, (b) 20 min, and (c) 55 min. It is clear that the growth of TiB₂ transfers from a fine layer to discretely coarse crystals at the interface of B₄C/Al composites. This fine layer has been already generated at 5 min. However, it is not continuous, implying a nonuniform interfacial reaction at the early stage. Besides, the surface morphology of B₄C particle becomes much rough in Figure 8(a), similar to Vialia’s results without Ti addition by powder metallurgy.¹⁵ As time increases, the interfacial layer of fine TiB₂ becomes continuous and grows up into single crystals at some discrete interface spots. When reaction time reaches 55 min, those TiB₂ crystals with large size are likely to fall off from B₄C surface due to the mismatch stress or mechanical stirring. The quantity of rough TiB₂ crystals also obviously increases with time. A sublayer of Al₃BC between the fine TiB₂ layer and B₄C particle should be carefully paid attention, which is hardly observed at short time 5 min but becomes thicker and thicker as time increases. However, the thickness of Al₃BC sublayer is not constant that some points penetrate deeper into the B₄C particle. It is interesting to note that Al₃BC and TiB₂ form an ordered interface microstructure: Al₃BC is close to B₄C particle while TiB₂ approaches to Al matrix.

Figure 8.

High magnification images of interface microstructure in B₄C/Al composites with different stirring time: (a) 5 min, (b) 20 min, and (c) 55 min.

Considering above observations in Figures 6 and 8, a diffusion-controlled reaction mechanism is proposed as follows in the Al–B₄C–Ti system.¹⁵_’¹⁹ Due to the reaction environment of bulky Al melt, B₄C particles kinetically prefer to be eroded by liquid Al, generating Al₃BC and B atoms.²⁸ Then Al₃BC begins to nucleate and grows at interface while B atoms diffuse outward into liquid Al, seen in equation (2), where B is dilute solute in the liquid Al (the same below).

3 Al (l) + B_{4} C = {Al}_{3} BC + 3 \underline{B}

(2)

Once the concentration of B atoms in liquid Al exceeds its maximal solubility (extremely low, 0.0547 wt% at 730℃),²⁹ TiB₂ spontaneously precipitates in the melt. TiB₂ tends to nucleate on B₄C surface and grows into fine grain layer, immediately acting as a protective layer to prevent further attack on B₄C particles by Al melt, seen in the following equation

\underline{Ti} + 2 \underline{B} = {TiB}_{2}

(3)

Then interfacial reaction will be changed to strictly solid-diffusion controlled process and the growth of TiB₂ is suppressed, as shown the decreasing decomposition rate of Al₃Ti in Figure 2(b). Due to the lower growth dynamic, TiB₂ has to grow into coarse crystals instead of overall growth of fine layer. TiB₂ crystals constantly grow depending on the consumption of Ti atoms dissolved in liquid Al. The phase diagram for Al-rich corner of Al–Ti alloy is shown in Figure 9. Since the preparation temperature 730℃ of B₄C/Al composites is higher than peritectic temperature 665℃, there exists a chemical equilibrium between solid Al₃Ti and Al–Ti melt in the solid–liquid phase zone. Due to the decrease of Ti solubility, Al₃Ti is decomposed in equation (4) to keep the chemical equilibrium.

{Al}_{3} Ti = 3 Al (l) + \underline{Ti}

(4)

Therefore, we have observed the gradual decrease of Al₃Ti in Figure 2. Al₃BC sublayer also becomes thick with time, as observed in Figure 8(c). However, once Al₃Ti is totally decomposed, AlB₂ will discretely precipitate in the matrix, as shown in Figure 6(c).

Al (l) + 2 \underline{B} = {AlB}_{2}

(5)

Thermodynamically consider the reactions described above. The thermochemical data at 1000 K used in this study are given in Table 2.³⁰ The Gibbs free energy of different products can be achieved through the following procedure illustrated in equations (6) to (10) to change the standard states.^28,29,31

Ti (s) = \underline{Ti} Δ G_{6} = - 97166 - 13.624 T, J / mol

(6)

B (s) = \underline{B} Δ G_{7} = 14746 - 16.162 T, J / mol

(7)

Ti (s) + 2 B (s) = {TiB}_{2} Δ G_{8} = - 326647 + 18.471 T, J / mol

(8)

Al (l) + 2 B (s) = {AlB}_{2} Δ G_{9} = - 165201 + 22.339 T, J / mol

(9)

3 Al (l) + Ti (s) = {Al}_{3} Ti Δ G_{10} = - 181020 + 60.225 T, J / mol

(10)

Then, for reactions (3) to (5)

Figure 9.

The aluminum-rich side of the Al–Ti phase diagram.²⁹

Table 2.

Standard heats of formation and entropies at 1000 K.³⁰

Phase (1000 K)	$Δ H_{f}^{\circ}$ (kJ/mol)	S^° (J/mol K)
TiB₂	−326.647	104.394
AlB₂	−165.201	109.288
Al₃Ti	−181.020	225.393
Al(l)	–	73.595
B(s)	–	29.016
Ti(s)	–	64.833

ΔG₃ = ΔG₈−2ΔG₇−ΔG₆ = −258973 + 64.419 T, J/mol

−ΔG₄ = ΔG₁₀−ΔG₆ = −83854 + 73.849 T, J/mol

ΔG₅ = ΔG₉−2ΔG₇ = −194693 + 54.663 T, J/mol.

Figure 10 presents the calculation results of above equations. It is obvious that the formation energy of TiB₂ is more negative than that of AlB₂. Thermodynamically, it is understandable that TiB₂ is preferable to be formed in the melt unless Al₃Ti is used up. Therefore, there is a precipitation order of TiB₂ and AlB₂ in this Al–B₄C–Ti system, as shown in Figure 6. Besides, the formation free energy of Al₃Ti close to zero indicates its possibility of decomposition or formation since the equilibrium with Al melt. As there has no thermodynamics data about the newly reported phase Al₃BC in thermodynamic manuals or other references,³² the free energy change of reaction (1) is difficult to calculate, but the microanalysis results of well-organized interface microstructure shown in Figure 8 still confirm that reaction (1) preferably takes place. Due to the over-reaction with long time stirring, stirring time should be carefully controlled to obtain fine interface microstructure. But this is difficult considering the necessary time for uniform distribution. Other parameters such as Ti content and temperature may be accordingly adjusted to help achieving expected interface microstructure.

Figure 10.

Gibbs free energy of formation of various reaction products.

Since the formation of TiB₂ layer on B₄C particles, the dispersion process is also influenced. TiB₂ with more metallic properties has better wetting capacity with Al melt, which can be completely wetted by liquid Al in a reasonably short time (about 1 h).³³ However, we should note that the dispersion process is not homogeneous. Many B₄C particles lying in clusters have less chance to sufficiently contact with Al melt. A necessary stirring time is required for all particles to participate in reactive wetting. Thus a mutual promotion mechanism of B₄C dispersion can be concluded: mechanical stirring promotes dispersion of B₄C aggregation to contact with melt and following interfacial reaction improves wetting to maintain the dispersion state, which together achieve rapid increase of distribution uniformity of B₄C particles. This process lasts about 20 min, as shown in Figure 3(a). However, further improvement of B₄C distribution is difficult beyond 20 min, as the B₄C clusters shown in Figure 1 (20 min to 40 min). Long holding time is necessary to achieve spreading of liquid Al on B₄C surface and following completely uniform distribution. Completely uniform distribution is finally obtained after 55-min holding in our experiment, which is close to D.A. Weirauch’s results of 1 h through sessile drop test.³³ Therefore, during the stage of 20–55 min, further dispersion of B₄C particles mainly relies on improved wetting, considering the decrease of homogenization speed in Figure 3(a). We have pointed that stirring time should be limited to avoid over-reaction, whereas long time is necessary to achieve uniform distribution of B₄C particles. This contradiction requires a balanced time to be taken for the fabrication of B₄C/Al composites by melt stirring method. For the 15 wt% B₄C/Al composites, a moderate stirring time about 20–30 min may be suitable according to our results.

Conclusion

Rapid dispersion of B₄C particles in the melt is achieved before 20 min, but further improvement of distribution uniformity close to random needs a long holding time to obtain the spreading of liquid Al on B₄C surface. The dispersion of B₄C particles changes from clusters to discrete distribution as time increases, which attributes to improved wetting through interfacial reaction in addition to high speed stirring.

Interface reaction is strongly dependent on stirring time. Al₃BC, TiB₂, and AlB₂ are produced, of which TiB₂ and sublayer Al₃BC are orderly located at interface while AlB₂ precipitates in matrix only after Al₃Ti disappears. The growth of TiB₂ transforms from a fine layer to coarse crystals as time increases.

A diffusion-controlled reaction mechanism is proposed including erosion of B₄C by Al melt, formation of Al₃BC and fine grain layer of TiB₂, balanced decomposition of Al₃Ti, growth of coarse TiB₂ crystals and precipitation of AlB₂ at last. A balanced stirring time is better to be taken, considering both interface reaction and particle dispersion, which is about 20–30 min for the fabrication of 15 wt% B₄C/Al composites.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by the China Nuclear Power Engineering Co., Ltd. under grant no. 2013966003.

References

Thevenot

. Boron carbide - A comprehensive review. J Eur Ceram Soc 1990; 6: 205–225.

Halverson

Pyzik

Aksay

. Processing of boron carbide-aluminum composites. J Am Ceram Soc 1989; 72: 775–780.

Gangolu

Rao

Prabhu

. Effects of temperature and strain rate on compressive flow behavior of aluminum-boron carbide composites. J Compos Mater 2014; 48: 1313–1321.

Bonnet

Rohr

Chen

. Use of Alcan's Al-B₄C metal matrix composites as neutron absorber material in TN International's transportation and storage casks. Packag Transp Storage Secur Radioact Mater 2009; 20: 98–102.

Zhang

Wang

. The design, fabrication and properties of B4C/Al neutron absorbers. J Nucl Mater 2013; 437: 350–358.

Handbook of neutron absorber materials for spent fuel transportation and strorage application: 2009 edition, Palo Alto, CA: EPRI, 2009.

Shorowordi

Laoui

Haseeb

ASMA

. Microstructure and interface characteristics of B₄C, SiC and Al₂O₃ reinforced Al matrix composites: A comparative study. J Mater Process Technol 2003; 142: 738–743.

Lloyd

. Particle-reinforced aluminum and magnesium matrix composites. Int Mater Rev 1994; 39: 1–23.

Hashim

Looney

Hashmi

. Metal matrix composites: Production by the stir casting method. J Mater Process Technol 1999; 93: 1–7.

10.

Lloyd

. Aspects of fracture in particulate reinforced matrix composite. Acta Metall Mater 1991; 39: 59–71.

11.

Lewandowski

Liu

Hunt

. Effects of matrix microstructure and particle distribution on fracture of an aluminum metal matrix composite. Mater Sci Eng A 1989; 107: 241–255.

12.

Murphy

Howard

Clyne

. Characterisation of severity of particle clustering and its effect on fracture of particulate MMCs. Mater Sci Technol 1998; 14: 959–968.

13.

Luo

Song

Zhang

. Interfacial microstructure in a B₄C/Al composite fabricated by pressureless infiltration. Metall Mater Trans A 2012; 43: 281–293.

14.

Kennedy

. The microstructure and mechanical properties of Al-Si-B₄C metal matrix composites. J Mater Sci 2002; 37: 317–323.

15.

Viala

Bouix

Gonzalez

. Chemical reactivity of aluminium with boron carbide. J Mater Sci 1997; 32: 4559–4573.

16.

Lin

Shen

Qiu

. Wetting of polycrystalline B₄C by molten Al at 1173-1473K. Scripta Mater 2009; 60: 960–963.

17.

Zhang

Chen

Charette

. Fluidity and microstructure of an Al–10% B₄C composite. J Mater Sci 2009; 44: 492–501.

18.

Kennedy

Brampton

. The reactive wetting and incorporation of B₄C particles into molten aluminium. Scripta Mater 2001; 44: 1077–1082.

19.

Zhang

Fortin

Charette

. Effect of titanium on microstructure and fluidity of Al–B₄C composites. J Mater Sci 2011; 46: 3176–3185.

20.

Garcia-Cordovilla

Louis

Narciso

. Pressure infiltration of packed ceramic particulates by liquid metals. Acta Mater 1999; 47: 4461–4479.

21.

Boselli

Pitcher

Gregson

. Secondary phase distribution analysis via finite body tessellation. J Microsc 1999; 195: 104–112.

22.

Assael

Kakosimos

Banish

. Reference data for the density and viscosity of liquid aluminum and liquid iron. J Phys Chem Ref Data 2006; 35: 285–300.

23.

Lloyd

Lagace

Mcleod

. Microstructural aspects of aluminum silicon-carbide particulate composites produced by a casting method. Mater Sci Eng A 1989; 107: 73–80.

24.

Ali

Yuan

Dickey

. Liquid dispersion mechanisms in agitated tanks: Part 1. Pitched blade turbine. Chem Eng Commun 1981; 10: 205–213.

25.

Schafer

Yianneskis

Wachter

. Trailing vortices around a 45 degrees pitched-blade impeller. AIChE J 1998; 44: 1233–1246.

26.

Derksen

. Numerical simulation of solids suspension in a stirred tank. AIChE J 2003; 49: 2700–2714.

27.

Yapici

Karasozen

Schaefer

. Numerical investigation of the effect of the Rushton type turbine design factors on agitated tank flow characteristics. Chem Eng Process 2008; 47: 1346–1355.

28.

Zhang

Lee

Cho

. Effect of the Ti/B₄C mole ratio on the reaction products and reaction mechanism in an Al-Ti-B₄C powder mixture. Mater Chem Phys 2014; 147: 925–933.

29.

Sigworth

. The grain refining of aluminum and phase relationships in the Al-Ti-B system. Metall Trans A 1984; 15: 277–282.

30.

Barin

. Thermochemical data of pure substances, Beijing: Science Press, 2003.

31.

Rapp

Zheng

. Thermodynamic consideration of grain-refinement of aluminum-alloys by titanium and carbon. Metall Trans A 1991; 22: 3071–3075.

32.

Wenjie

Pengting

Tong

. Transformation from Al₃BC phase to doped TiB₂ or TiC particles in Al-Ti melts. J Alloy Compd 2013; 561: 48–53.

33.

Weirauch

Jr Krafick

Ackart

. The wettability of titanium diboride by molten aluminum drops. J Mater Sci 2005; 40: 2301–2306.