Improved halide salt process to produce Al–B master alloys

Introduction

Transition metals impair the electrical conductivity of aluminium dramatically when solutionised in the aluminium matrix and must be removed to produce conductor grade aluminium. 1 1,2 This is accomplished by the addition of Al–B master alloys into the melt before casting. Transition metals, such as titanium, vanadium, chromium and zirconium, are readily precipitated by boron as their borides are more stable than those of aluminium.3

Al–B master alloys have recently been receiving some attention to replace Al–Ti–B master alloys, which fail to grain refine Al foundry alloys adequately.4 The high content of Si is responsible for the poor response of foundry alloys to grain refinement by Al–Ti–B master alloys,5^–7 which are the predominant industrial refiners for Al alloys.8^–12 The Al–B master alloys, on the other hand, take advantage of the high levels of Si, which enhance the nucleation potential of the AlB₂ particles claimed to be the nucleating agent in these alloys. 5 5,13 The superior performance of Al borides, which, in the absence of Si, cannot act as an efficient grain refiner, may be attributed to the dissolved Si in the foundry alloys.5 Al–B master alloys were recently shown to have grain refining potency also for commercial pure aluminium.14

While a novel powder metallurgy process was proposed recently,3 the industrial production of Al–B alloys involves direct addition of KBF₄ salt into molten aluminium.2 Boron of the halide salt is reduced by and then reacts with aluminium so as to form aluminium borides. Dispersing boride particles in the aluminium melt does not occur on its own and requires special measures. The majority of the aluminium borides thus formed are retained in the dross layer, while those that mix with the melt are often in the form of clusters and contaminated with salt residues. This contrasts the case of co-addition of K₂TiF₆ and KBF₄ salts into molten Al in the production of Al–Ti–B grain refiners, where the TiB₂ particles are dispersed relatively more easily. An attempt was made in the present work to improve the conventional halide salt practice in the manufacture of Al–B alloys so as to increase the B recovery and the quality of the final alloy and to optimise the revised manufacturing process.

Experimental

A series of Al–3 wt-B alloys were produced in the laboratory by reacting KBF₄ salt with molten aluminium. The starting materials were Al (99·7Al), containing 0·001Ti and <0·0002B, and KBF₄ salt of commercial purity. Production campaigns were carried out on a 1·0 kg batch scale. Aluminium metal was brought to 850°C using an electrical resistance furnace. KBF₄ salt was preheated to 110°C to remove its moisture before it was added to molten Al. Melt temperatures were recorded to follow the progress of the KBF₄–Al reaction. The melt was held at this temperature for up to 30 min. The spent salt or the dross generated at the melt surface was removed before half of the melt was cast into a permanent mould. The remaining half was cast after it was stirred thoroughly. Stirring was performed manually with a preheated graphite rod.

The above sequence was repeated in exactly the same manner but with KBF₄ mixed with various fluride based rafination and oxygen bearing exothermic drossing fluxes. The mixtures thus obtained were employed either in powder form or after they were pressed into tablets. The addition of the salt mixture to molten aluminium, whether in powder or tablet form, was performed gradually, to avoid excessive cooling of the melt and to retain the melt temperature above 800 C throughout the addition process.

The microstructures of the Al–B alloys thus obtained were examined after etching with a 0·5 HF solution using an Olympus BX51M model optical microscope. The X-ray diffraction (XRD) patterns were recorded with a Shimadzu XRD 6000 diffractometer equipped with Cu K_α radiation. The diffractometer was operated at very low scanning rates to improve the counting frequency. The composition of the experimental alloys were determined with an optical emission spectrometer after dilution with commercial purity aluminium for B content measurements.

Results and discussion

The KBF₄ salt melts almost immediately once it is added to molten aluminium and reacts with aluminium melt in the fully liquid state above 750°C. It is inferred from the melt temperature measurements that it takes several minutes for the reaction between KBF₄ and aluminium (1) to start. The temperature first drops owing to the enthalpy exchange between the relatively colder KBF₄ salt and the aluminium melt and then starts to increase once the exothermic KBF₄–Al reaction kicks off. This reaction takes place at the salt layer/melt interface, and both the borides and the K–Al–F salt thus formed accumulate at the melt surface. The latter is initially very fluid but thickens with time as it mixes with the surface oxides and the freshly generated borides.

The conventional practice of adding KBF₄ salt to molten Al outlined above involves a number of difficulties and fails to yield an Al–B alloy of acceptable quality. It produces a B lean Al–B alloy melt and a B rich mush, which is very viscous and is thus difficult to separate from or incorporate into the melt. The former reveals upon solidification few AlB₂ particles that are invariably coagulated with fluoride salt residues and distributed in the aluminium matrix in a heterogeneous fashion (Fig. 1a). The mushy part (Fig. 1b), on the other hand, is crowded with boride particles (Fig. 1c). It is interesting to note that the K–Al–F salt produced by reaction (1) does not help to separate the boride particles from the mushy fraction. This is in marked contrast to what happens when the K₂TiF₆ and KBF₄ salts are co-added to molten Al to produce Al–Ti–B alloys. The K–Al–F spent salts in the so called ‘halide salt process’ are readily decanted, while the spent salt in the present work collects the oxides and a significant fraction of the borides and transforms into a fluoride rich dross. Increasing the processing temperature to improve the fluidity of the spent salt did just the opposite since higher melt temperatures promoted oxidation. The majority of the AlB₂ particles produced by the Al–KBF₄ reaction are thus entrapped in the dross. A uniform dispersion of borides that are retained in the aluminium melt is highly unlikely even when vigorous stirring is employed. Stirring helps to improve B recovery but impairs the quality of the Al–B alloy due to heavy contamination with the spent salt residues (Table 1 and Fig. 2).

OPEN IN VIEWER

Figure 1

a Microstructure of Al–3B alloy produced by adding KBF₄ salt to molten aluminium at 850 °C; b macro photo and c microstructure of mushy part

OPEN IN VIEWER

Figure 2

Microstructure of Al–3B alloy produced by adding KBF₄ salt to molten aluminium at 850°C and mechanically stirred before casting into permanent mould

OPEN IN VIEWER

Table 1

B recoveries and degree of cleanliness in experimental Al–3B alloys prepared with different processes

Process		B recovery/	Cleanliness
Salt used	Stirring
KBF4	Without	<30	Low
	With	∼40	High
Premixed KBF4+Na3AlF6 at ratio of 2∶1	Without	∼70	High
	With	>90	High

The XRD analysis of the dross reveals reflections of a series of potassium aluminium fluorides (KAlF₄, K₃AlF₆ and KAl₄F₁₃), oxides (Al₂O₃ and KO₂) and aluminium borides (AlB₂ and AlB₁₂) (Fig. 3). The formation of the latter boride is not expected at this processing temperature. AlB₁₂ can form only when the B content in Al is 44·5 wt- according to the Al–B binary phase diagram.15 This could happen in the present case only upon substantial B segregation. The evidence for some AlB₁₂ in the dross further confirms the B enrichment in the mushy fraction during processing and the inevitable low B recovery in the final alloy.

OPEN IN VIEWER

Figure 3

X-ray diffraction analysis of dross generated shortly after KBF₄ salt is added to molten aluminium at 850°C

Several measures were taken to counteract the above mentioned problems associated with the conventional practice. A variety of foundry fluxes were employed to improve the B recovery in the final alloy via reclaiming the boride particles retained in the dross. Out of a series of fluoride bearing rafination and oxygen bearing exothermic drossing fluxes, cryolite (Na₃AlF₆) of commercial purity produced the best results. Hence, a revised practice was designed to maximise the B recovery and the cleanliness of the final alloy. The revised practice relied on the use of highly surface active Na₃AlF₆ to break up the oxide–boride agglomerates, which are glued together with the spent salt. The impact of Na₃AlF₆ addition was remarkable. Na₃AlF₆ helped to break-up the mushy part and recover the borides, which otherwise would end up in the dross. The dispersion of the recovered borides in the Al matrix was quite uniform (Fig. 4), and the B recovery was much improved (Table 1).

OPEN IN VIEWER

Figure 4

Microstructure of Al–3B alloy produced by adding KBF₄–Na₃AlF₆ salts premixed at ratio of 2∶1 to molten Al at 850°C

Several experiments were performed to optimise the Na₃AlF₆ addition practice. First, Na₃AlF₆ worked better when mixed with the KBF₄ salt before addition than when added subsequently. Addition levels ranging from Na₃AlF₆/KBF₄ of 1∶10 to 1∶1 on a weight basis were employed to identify the minimum amount of Na₃AlF₆ required to fully recover the boride particles. Na₃AlF₆ failed to achieve full recovery of the borides at an addition level of Na₃AlF₆/KBF₄ = 1∶4. Some mushy part was retained and could not be incorporated into the melt even with the help of vigorous stirring at this Na₃AlF₆ level. After several trials, it was concluded that the minimum addition level required for a trouble free processing and full B recovery is Na₃AlF₆/KBF₄ = 1∶2. This much Na₃AlF₆ sufficed to eliminate the mush formation and to recover the borides. Higher addition levels did not improve the situation and were thus judged to be unnecessary. The form of salt addition, powder versus tablets, was also investigated. Pressing the KBF₄ and Na₃AlF₆ salt mixture into tablets offered practical benefits but did not improve the boride recovery or the final alloy quality. In fact, the addition of the salt mixture in powder form promoted and accelerated the reaction between KBF₄ and Al and was thus favoured.

The use of Na₃AlF₆ mixed with KBF₄ salt eliminated the mush formation and improved the microstructural features of the Al–B alloy. The alloy thus processed was much cleaner with only occasional salt residues (Fig. 4). A fraction of the AlB₂ particles, however, were still agglomerated, suggesting that there is still room for improvement. A second revision was thus introduced in the conventional process to take care of the remaining salt residues and borides they glue into clusters. This was achieved through mechanical stirring, performed manually with a preheated graphite rod once the spent salt was decanted. No dross formation was noted, thanks possibly to a very fluid salt layer on the melt surface. Decanting the spent salt was straightforward and so was pouring the much more homogeneous Al–B alloy melt.

The microstructural features of the Al–3B alloy were improved further owing to mechanical stirring. The boride dispersion in the aluminium matrix was uniform with no evidence of salt residues (Fig. 5). The XRD analysis of the Al–B alloy thus produced shows reflections of the AlB₂ phase in addition to those of the aluminium matrix (Fig. 6). No evidence for the AlB₁₂ phase reported to form at these process temperatures by other investigators 2 16 2,16,17 was available. This confirms the results of the metallographic analysis, which shows only equiaxed AlB₂ platelets. The lack of AlB₁₂ particles in the Al–B alloy is welcomed as they are known to dissolve in aluminium melts relatively slower. The slow dissolution of borides is not an attractive feature in an Al–B alloy whether it is intended for refining the grain structure or for precipitating transition elements.

OPEN IN VIEWER

Figure 5

Microstructure of Al–3B alloy produced by adding KBF₄–Na₃AlF₆ salts premixed at ratio of 2∶1 to molten Al at 850°C and mechanically stirred before casting into permanent mould

OPEN IN VIEWER

Figure 6

X-ray diffraction analysis of Al–3B alloy produced by adding KBF₄–Na₃AlF₆ salts premixed at ratio of 2∶1 to molten Al at 850°C and mechanically stirred before casting into permanent mould

Having identified the use of Na₃AlF₆ salt and mechanical stirring to be the essential elements of the manufacturing process for a uniform dispersion AlB₂ particles and a sufficiently clean aluminium matrix, several additional experiments were performed for the optimisation of the reaction time and temperature. Different holding times were investigated to allow for the KBF₄ and Al reaction to reach completion once the salt mixture was added to molten Al. The alloy cast right after the salt mixture melted suffered a low B yield with fewer than expected AlB₂ particles. Microstructural and XRD analysis have shown that 5 min after the addition sufficed to achieve a complete reaction between the aluminium melt and KBF₄, and longer holding times did not produce an improvement in B recovery.

The salt addition, i.e. reaction, temperature is also of great technological interest. Temperatures between 750 and 950°C were investigated for the efficiency of the revised process. The fluidity of the Al–B alloy melt (actually slurry as it contains solid boride particles) was slightly degraded at 750°C, and the B recovery in the final alloy was relatively lower. Salt addition temperatures between 800 and 900°C produced more or less the same results. While the process was practically identical above 900°C, the microstructural features were considerably different at 950°C. There appeared to be two types of boride particles generated at 950°C, with low and high aspect ratios. The population of the latter increased with increasing addition temperatures (Fig. 7). However, the XRD spectrum of the Al–3B alloy produced at 950°C also revealed only the AlB₂ reflections in addition to those of the aluminium matrix with no evidence of the AlB₁₂ compound. This suggests that the present transition is merely a morphological, rather than a structural, one. This is in agreement with earlier studies that report two types of AlB₂ crystals with low and high aspect ratios.18^–22 Low aspect ratio AlB₂ is converted to high aspect ratio AlB₂ by heating above 950°C during the preparation of AlB₂ metal matrix composites,23 suggesting that the peritectic reaction occurs near 950°C.24 Both high and low aspect ratio AlB₂ particles were found to exhibit similar XRD patterns characteristic of single crystal AlB₂.24

OPEN IN VIEWER

Figure 7

High aspect ratio AlB₂ particles generated at a 925°C, b 950°C and c 975°C

Summary

Reacting KBF₄ salt with molten Al produces a B lean Al–B alloy melt and a B rich dross. A revised process that relies on mixing KBF₄ with Na₃AlF₆ and mechanical stirring of the melt is proposed in the present work. The highly surface active Na₃AlF₆ breaks up the oxide–boride particles glued together with the fluoride based spent salt, while mechanical stirring facilitates uniform dispersion of the freed borides. The microstructural features of the Al–3B alloy thus processed and the B recovery were improved in a marked fashion. The boride dispersion in the aluminium matrix was uniform with no evidence of salt residues inside the aluminium matrix. An improved salt addition practice to ensure full B recovery and an Al–B alloy of sufficient quality is thus claimed to comprise the following steps: melting commercial purity aluminium ingot, adding premixed KBF₄ and Na₃AlF₆ salts to molten aluminium at 850°C gradually to avoid excessive cooling of the melt, holding the melt for 5 min at 850°C to allow the reaction to reach completion, decanting the spent salt and throughly stirring the melt before casting into desired shape.