Production of Al–Ti–B grain refining master alloys from Na2B4O7 and K2TiF6

Production of Al–Ti–B grain refining master alloys from Na2B4O7 and K2TiF6

Journal of Alloys and Compounds 458 (2008) 271–276 Production of Al–Ti–B grain refining master alloys from Na2B4O7 and K2TiF6 Y¨ucel Birol ∗ Material...

2MB Sizes 0 Downloads 1 Views

Journal of Alloys and Compounds 458 (2008) 271–276

Production of Al–Ti–B grain refining master alloys from Na2B4O7 and K2TiF6 Y¨ucel Birol ∗ Materials Institute, Marmara Research Center, TUBITAK, Gebze, Kocaeli, Turkey Received 15 March 2007; accepted 6 April 2007 Available online 8 April 2007

Abstract It is very desirable to replace the KBF4 salt in the popular “halide salt” process to reduce the volume of fluoride salts to be added to molten aluminium in the production of Al–Ti–B grain refiners. Being over 2 times richer in B, Na2 B4 O7 is a promising replacement for KBF4 , and is used in the present work to produce Al–Ti–B grain refiner master alloys. A fraction of the aluminide particles were entrapped in the spent salt giving a relatively lower Ti recovery when KBF4 was replaced by Na2 B4 O7 . The grain refining performance of the Al–Ti–B grain refiner alloy thus produced was nevertheless acceptable. The spent salt became too viscous with the oxides, aluminides and borides to be removed by decanting when Na2 B4 O7 ·5H2 O was used to supply boron. The viscous spent salt, entrained in the grain refiner alloy, did not only impair its performance, but also hurt the fluidity of the molten alloy and made pouring difficult. © 2007 Elsevier B.V. All rights reserved. Keywords: Metals; Casting; Al–Ti–B master alloy; Grain refinement

1. Introduction The aluminium industry relies on grain refiner master alloys from the Al–Ti–B ternary system to control the cast grain size [1,2] eversince boron additions were shown to remarkably improve grain refinement of aluminium by titanium at hypoperitectic concentrations [3]. Addition of Al–Ti–B master alloys to molten aluminium produces fine, equiaxed grains after solidification which otherwise tend to be coarse and columnar. Such a structure not only yields superior mechanical properties and surface quality but also provides alloy homogeneity and allows for good castability yielding a high casting output and reduced down stream processing costs. While numerous methods have been offered for the production of Al–Ti–B grain refiner alloys [4–18], that which involves adding a mixture of KBF4 –K2 TiF6 salts to molten aluminium, is the most popular [4]. The use of KBF4 in the popular “halide salt” route suffers several drawbacks. A large volume of particulate material must be added to molten aluminium since this salt is very lean in boron. Over 90 wt% of KBF4 produces intense fluoride emis-



Tel.: +90 262 6773084; fax: +90 262 6412309. E-mail address: [email protected]

0925-8388/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.jallcom.2007.04.036

sions which require a costly emission control system and a spent salt which has to be removed from the sytem by decanting. This is certainly very unfortunate as KBF4 is an expensive compound which adds greatly to the raw material costs. Hence, it is very desirable to replace this halide salt with other precursor compounds richer in boron. Several alternative boron sources were recently offered to replace KBF4 salt [19–26]. The present work was undertaken to find out if Na2 B4 O7 can replace KBF4 in the production of Al–5Ti–1B grain refiners. Na2 B4 O7 is over 2 times richer in boron with respect to KBF4 and is less expensive. Na2 B4 O7 ·5H2 O, by far the most commercially important boron compound, was also used as a boron source in the present work. 2. Experimental Three experimental Al–Ti–B grain refiner alloys were produced, on a 1 kg batch scale, by employing the process described in [16]. The intended nominal composition for these alloys was Al–5Ti–1B. The first experiment involved the addition into molten aluminium of premixed KBF4 and K2 TiF6 salts of commercial purity, to obtain a standard grain refiner (experiment 1 in Table 1). Boron was sourced entirely from Na2 B4 O7, instead of KBF4 in the following experiment (experiment 2). In the third experiment, the precursor boron compound was Na2 B4 O7 ·5H2 O. The use of Na2 B4 O7 and Na2 B4 O7 ·5H2 O have led to savings of 70 g and 39 g of particulate charge per kilogram of grain refiner alloy, respectively. The alloy melts thus obtained were held at 800 ◦ C for 30 min in an electric resistance furnace in order to avoid the mixing of the spent salt

272

Y. Birol / Journal of Alloys and Compounds 458 (2008) 271–276

Table 1 K2 TiF6 , KBF4 , Na2 B4 O7 and Na2 B4 O7 ·5H2 O additions (g) employed in the production of three experimental alloys to achive a Ti/B ratio of 5 Experiment/alloy

K2 TiF6

KBF4

Na2 B4 O7

Na2 B4 O7 ·5H2 O

Total

1 2 3

250.7

116.4 – –

– 46.5 –

– – 67.3

367.1 297.2 318.0

Table 2 Constituents produced in the grain refiner production experiments and the Ti recovery (wt%) Constituents

Experiment/alloy 1

Experiment/alloy 2

Experiment/alloy 3

Grain refiner alloy (wt%) Fluid salt removed by decanting (wt%) Mushy salt removed by skimming (wt%) Ti recovery (wt%)

64 21 15 99.4

58 14 28 70.0

52 – 48 49.0

in the melt. The molten salt was finally decanted; the viscous portion, if any, was skimmed off and the melt was stirred thoroughly to rejuvenate the settled particles before it was cast into a permanent mold. The three experimental master alloys thus produced were assessed for their grain refining performances using the test described in [16]. A key quality feature of any grain refiner with an impact on its performance is its Ti content, which, in turn, is dictated by the selection of precursor compounds in the present work. The grain refining performance tests were thus conducted at the same grain refiner addition level in order to elucidate the effect of different boron compounds on the Ti recovery. Aluminium ingot with a purity of 99.7% Al, weigthing 1000 g, was inoculated with 4 g of each of the experimetal alloys. This is the exact amount of an Al–5Ti–1B master alloy needed to bring the Ti concentration of 1 kg aluminium melt to 0.02%. This practice allows the Ti content of each alloy to impact the grain refining efficiency and to better judge the effect of different precursor compounds on the performance. The inoculated samples were sectioned 20 mm from the bottom surface. Standard metallographic procedures were employed to prepare the inoculated samples which were etched with 0.5% HF reagent and then photographed for visual assesment. The insoluble boride particles in the aluminium matrix and the constituents of the spent salt were identified with X-ray diffraction (XRD) analysis. XRD was conducted with Cu K␣ radiation at a scan rate of 0.5◦ min−1 for the metallic samples in order to improve the counting frequency.

Approximately two thirds of the initial charge was converted into a grain refiner alloy in the first experiment (Table 2). Considering the fraction of the halide salt elements that can be carried over to the grain refiner, an alloy recovery of 64% is certainly acceptable. What is more impressive with the optimum halide salt process, however, is the very high Ti recovery. Nearly the whole Ti of the K2 TiF6 salt ended up in the grain

3. Results and discussion The reaction between molten aluminium and the particulate KBF4 or Na2 B4 O7 and K2 TiF6 mixture yields Al–Ti–B grain refiner alloys with varying levels of Ti depending on the precursor boron compound used. The unused fraction of the particulate charge, on the other hand, is either decanted (fluid salt) and/or skimmed (mushy salt) if and when a fraction of the spent salt is too viscous to be removed by decanting. The amount of grain refiner alloy produced out of the total material charge, and the Ti content of the grain refiner alloy thus produced with respect to the intended Ti level are referred to as the alloy and Ti recoveries, respectively. Alloy recovery (%) =

weight of grain refiner alloy obtained weight of total charge (1)

Ti recovery (%) =

Ti in the grain refiner alloy (wt%) intended Ti (wt%)

(2)

Fig. 1. XRD spectra of spent salts in experiments: (a) 1, (b) 2 and (c) 3. () K2 NaAlF6 , () KAlF4 , (䊉) K3 AlF6 , () Al2 O3 , () Al3 Ti, () TiB2 , (♦) K2 Al2 Si3 O10 ·2H2 O.

Y. Birol / Journal of Alloys and Compounds 458 (2008) 271–276

refiner alloy yielding a Ti recovery over 99% when the optimized practice described in [16] was employed. A substantial portion of the spent salt (approximately one fifth of the total charge) was decanted shortly before casting the molten Al–Ti–B alloy into a permanent mold. The predominant constituents of the spent salt were identified by XRD to be KAlF4 and K3 AlF6 (Fig. 1a). The decanted salt was free from contamination and was thus a perfect fluxing agent. The microstructure of the grain refiner alloy and the constituents of the spent salt are perfectly accounted for

Fig. 2. Microstructures of grain refiner alloys (a) 1, (b) 2 and (c) 3.

273

by the following reaction between KBF4 , K2 TiF6 and molten aluminium, 2KBF4 + 3K2 TiF6 + 12Al → 2Al3 Ti + TiB2 + 5KAlF4 + K3 AlF6

(3)

A fraction of the spent salt was highly viscous and could be removed from the system only by skimming. This mushy portion, was found by XRD to contain a small amount of oxides in addition to the above fluorides. The Al–Ti–B alloy produced in experiment 1, exhibits very favorable microstructural features and offers a remarkable grain refining efficiency. This is attributed to the outstanding Ti recovery in alloy 1, reflected in the microstructure by a large number of soluble aluminide and insoluble boride particles dispersed in an ␣-Al matrix. The former are blocky, ranging in size from several to 20 ␮m while the boride particles are smaller than 1 ␮m,

Fig. 3. The XRD spectra of alloys (a) 1, (b) 2 and (c) 3. (♦) ␣-Al, () AlB2 , () TiB2 , () Al3 Ti.

274

Y. Birol / Journal of Alloys and Compounds 458 (2008) 271–276

are numerous, and often clustered (Fig. 2a). The majority of these borides were identified by XRD to be of the TiB2 variety (Fig. 3a). These microstructural features assure an outstanding grain refining performance. The average grain size of the sample cast 2 min after commercial purity aluminium melt was inoculated with alloy 1, is 100 ␮m and is better than that can be attained with most commercial grain refiner master alloys (Fig. 4a). The fine grain size was retained until 60 min after inoculation, with no sign of fading. Contact time of 60 min is longer than that encountered in most foundry practices. The predominance of sub-micron TiB2 particles and the blocky morphology and wide size range of Al3 Ti particles are credited for the exceptional performance of alloy 1. The reactions producing the aluminides and the borides in molten aluminium were found to be relatively sluggish in experiment 2 which employed Na2 B4 O7 as the precursor boron compound. The temperature increase in the melt upon the addition of Na2 B4 O7 –K2 TiF6 mixture was substantially less than that encountered with KBF4 –K2 TiF6 , the reaction of which with molten aluminium is known to be highly exothermic. The composition of the reaction product has also changed considerably (Table 2). The amount of spent salt decanted in experiment 2 was nearly one third less with respect to experiment 1, while the mushy fraction has almost doubled. The rest, estimated to be 58 wt% of the initial charge, was converted into an Al–Ti–B grain refiner alloy, impying a slight decrease in alloy recovery. The Ti recovery, on the other hand, was estimated to be 70%, implying a substantial drop in the amount of Ti retained in the grain refiner alloy. The increase in the viscous fraction of the spent salt brings to mind an increase in oxidation of aluminium and/or titanium by the oxygen-bearing boron compound in the particulate charge. The loss to oxidation may indeed be responsible for the lower Ti recovery in alloy 2, produced in exactly the same manner as

alloy 1, but with Na2 B4 O7 as the precursor boron compound. The spent salt in experiment 2, however, was shown by XRD to be dominated by K2 NaAlF6 but also contained KAlF4 , K3 AlF6 , Al2 O3 and Al3 Ti (Fig. 1b). Al2 O3 and Al3 Ti reflections were very weak and there was no evidence for TiO2 , suggesting that the decanted salt was essentially a halide salt. The K3 AlF6 of the spent salt in experiment 1 was largely replaced by K2 NaAlF6 upon the use of a Na-bearing compound. While the XRD spectrum of the mushy fraction was quite similar, Al2 O3 and Al3 Ti reflections were much stronger. The XRD analysis convincingly shows that a fraction of Al3 Ti particles are entrapped inside the spent salt layer which is enriched also with Al2 O3 . It is concluded, in view of the foregoing, that Al3 Ti particles retained in the spent salt are reponsible for the lower Ti recovery in alloy 2 rather than the oxidation of Ti. This is evidenced by the metallographic analysis which showed that Al3 Ti particles were indeed entrapped inside the spent salt layer in appreciable numbers (Fig. 5a). The reaction that leads to the formation of Al3 Ti in experiment 2 is believed to be the same as that encountered in the “halide salt” route [26,27]. 3K2 TiF6 + 13Al → 3Al3 Ti + 3KAlF4 + K3 AlF6

(4)

K2 TiF6 is reduced by molten aluminium, releasing Ti which dissolves in the melt before precipitating out as Al3 Ti particles as soon as its solubility limit is exceeded. Formation of TiB2 particles, on the other hand, may involve the reduction of Na2 B4 O7 by either Al3 Ti particles or solute Ti, before it precipitates out as Al3 Ti: Na2 B4 O7 + 2Al3 Ti → 2TiB2 + 2Al2 O3 + Na2 O + 2Al

(5)

Na2 B4 O7 + 5Ti → 2TiB2 + 3TiO2 + Na2 O

(6)

Fig. 4. The grain refining performance test results of alloys (a) 1, (b) 2 and (c) 3.

Y. Birol / Journal of Alloys and Compounds 458 (2008) 271–276

275

Evidence for reaction 7 comes from the XRD spectrum of the spent salt which shows reflections of both K2 NaAlF6 and Al2 O3 (Fig. 1b). The overall reaction leading to the formation of Al3 Ti and TiB2 particles, upon the addition into molten aluminium, of K2 TiF6 and Na2 B4 O7 mixture may thus be written as, 3Na2 B4 O7 + 9K2 TiF6 + 33Al → 3Al3 Ti + 6TiB2 + 6K2 NaAlF6 + 3KAlF4 + K3 AlF6 + 7Al2 O3

Fig. 5. Al3 Ti particles entrapped inside the spent salt generated during the production of (a) alloy 2 and (b) alloy 3 and salt entrained in the microstructure of alloy 3 (c).

Lack of evidence for TiO2 and Na2 O in the spent salt, however, implies that the formation of TiB2 particles relies, perhaps on the following reaction: 3Na2 B4 O7 + 6K2 TiF6 + 20Al → 6TiB2 + K2 NaAlF6 + 7Al2 O3

(7)

(8)

Reaction 8 provides a full account of the phases identified in the XRD analysis of the grain refiner alloy as well as of the spent salt. Evidence of AlB2 , in the XRD spectrum of the grain refiner alloy 2 (Fig. 3b), seems to suggest that some AlB2 also forms in experiment 2 possibly due to the missing Ti that cannot make it to the grain refiner alloy. A portion of the B charge left unreacted thus ends up reacting with molten aluminium forming AlB2 particles. The lower Ti recovery of alloy 2 is manifested also in the microstructural analysis. The aluminide particles are almost invariably of the blocky variety and are generally smaller than 20 ␮m as in alloy 1, but are fewer (Fig. 2b). The boride particles, which exhibit the typical boride morphology as in alloy 1, are also slightly less in number. A small increase is noted, however, in the fraction of the AlB2 variety with respect to alloy 1 (Fig. 3b). While not as good as that of alloy 1, the grain refining performance of alloy 2 is adequate with slightly coarser grains in the inoculated samples for the entire range of contact times (Fig. 4b). AlB2 particles which are known to be less efficient in terms of grain refinement with respect to TiB2 [28,29] together with the reduced number of aluminide particles, must be responsible for the relatively inferior performance of this experimental alloy. The changes noted upon replacing KBF4 with Na2 B4 O7 were even more prominent when Na2 B4 O7 ·5H2 O was used as the precursor boron compound (experiment 3). The spent salt was too viscous to be decanted and the entire salt, which was nearly half of the initial charge, had to be skimmed before casting (Table 2). While K2 NaAlF6 was shown by XRD to be the predominant constituent of the spent salt, KAlF4 , K3 AlF6 Al2 O3 , Al3 Ti and TiB2 were also identified (Fig. 1c). Reflections of Al2 O3 and Al3 Ti in the XRD spectrum of the spent salt were much stronger with respect to experiment 2. Al2 O3 , which is a by product of the reaction between Na2 B4 O7 ·5H2 O and molten aluminium and a higher population of entrapped Al3 Ti and TiB2 particles (Fig. 5b) apparently increased the viscosity of the spent salt to a point where removal by decanting was no longer possible. The microstructural features of alloy 3 further confirmed the foregoing (Fig. 2c). Experiment 3 which relied on Na2 B4 O7 ·5H2 O for boron supply has produced an Al–Ti–B grain refiner alloy with few Al3 Ti and TiB2 particles. The aluminide and boride particles are typical but are much less in number. A fraction of the borides is once again of the AlB2 variety and the aluminide and boride reflections are relatively weaker (Fig. 3c). A striking difference in alloy 3, however, is the salt inclusions dispersed in the microstructure (Fig. 5c). The

276

Y. Birol / Journal of Alloys and Compounds 458 (2008) 271–276

highly viscous spent salt was partly suspended in the melt, hurting the fluidity of the molten alloy and making pouring difficult. The grain refining performance of alloy 3 is markedly inferior with respect to those of alloys 1 and 2, with columnar grains around the edges of the inoculated samples only 5 min after the addition of the grain refiner alloy (Fig. 4c). This poor performance is primarily linked with the spent salt entrained in the grain refiner alloy which is known to wet the boride particles, impairing their potency as nucleation sites and also leading to their agglomeration [30]. The reduced number of aluminide and boride particles has also degraded the grain refining efficiency. 4. Summary The Al–5Ti–1B grain refiner alloy which relied on Na2 B4 O7 for the boron supply, offered an acceptable grain refining performance in spite of a relatively lower Ti recovery. This performance came, however, with a slight decrease in the alloy recovery which can be readily compensated with additional particulate charge. So, it is fair to conclude that Na2 B4 O7 is a viable alternative for KBF4 in the popular halide salt route. The spent salt became too viscous with the oxides, aluminides and borides to be removed by decanting when Na2 B4 O7 ·5H2 O was used to supply boron in the same process. Largely entrained in the grain refiner alloy, the viscous spent salt did not only impair its performance, but also hurt the fluidity of the molten alloy and made pouring difficult, leading to a very low alloy recovery. Acknowledgements O. C ¸ akır and F. Alageyik are thanked for their help in the experimental part of this work. References [1] B.S. Murty, S.A. Kori, M. Chakraborty, Int. Mater. Rev. 47 (2002) 3–29. [2] D.G. McCartney, Int. Mater. Rev. 34 (1989) 247.

[3] A. Cibula, J. Inst. Met. 76 (1949) 321–346. [4] P. Davies, J.L.F. Kellie, D.P. Patron, J.V. Wood, Metal Matrix Alloys, U.S. Patent No. 6,228,185 (2001). [5] P. Li, E.G. Kandalova, V.I. Nikitin, A.R. Luts, A.G. Makarenko, Y. Zhang, Mater. Lett. 57 (2003) 3694–3698. [6] G.K. Sigworth, Third Element Additions to Aluminum–Titanium Master Alloys, U.S. Patent No. 5,100,488 (1992). [7] L. Backerud, R. Kiusalaas, H. Klang, M. Vader, J. Noordegraaf, E.H.K. Nagelvoort, Method for production of master alloys for grain refining treatment of aluminum melts, U.S. Patent No. 5,104,616 (1992). [8] B.S. Murty, S.A. Kori, K. Venkateswarlu, R.R. Bhat, M. Charaborty, J. Mater. Process. Technol. 89–90 (1999) 152–158. [9] M.S. Lee, B.S. Terry, Mater Sci. Technol. 7 (1991) 608–612. [10] M.J. Jackson, I.D. Graham, J. Mater. Sci Lett. 13 (1994) 754–756. [11] M.S. Lee, B.S. Terry, P. Grieveson, Metall. Trans. B. 24B (1993) 955– 961. [12] I.G. Davies, J.M. Dennis, A. Hellawell, Metall. Trans. 1 (1970) 275–280. [13] I. Maxwell, A. Hellawell, Acta Metall. 23 (1975) 895–899. [14] K.A.Q. O’Reilly, B. Cantor, P.G. Enright, Scr. Metall. Mater. 28 (1993) 173–177. [15] M.G. Chu, Mater. Sci. Eng. A179–180 (1994) 669–675. [16] Y. Birol, J. Alloys Compd. 420 (2006) 71–76. [17] Y. Birol, J. Alloys Compd. 420 (2006) 207–212. [18] Y. Birol, J. Alloys Compd. 427 (2007) 142–147. [19] M. Alliot, J.C. Beguin, M. Moutach, J.C. Percheron, Mother Alloy of Aluminum, Titanium and Boron and Process for Fabrication, U.S. Patent No. 3,961,995 (1976). [20] D.K. Young, W.C. Setzer, F.P. Koch, R.A. Rapp, M.J. Pryor, N. Jarrett, Aluminum Base Alloy and Method for Preparing Same, U.S. Patent No. 5,415,708 (1995). [21] D.K. Young, W.C. Setzer, F.P. Koch, R.A. Rapp, M.J. Pryor, N. Jarrett, Aluminum Base Alloy, U.S. Patent No. 5,484,493 (1996). [22] A. Hardman, F.H. Hayes, Mater. Sci. Forum 217–222 (1996) 247–252. [23] T.S. Krishnan, P.K. Rajagopalan, B.R. Gund, J. Krishnan, D.K. Bose, J. Alloys Compd. 269 (1998) 138–140. [24] Q. Zhuxian, Y. Yaxin, Z. Mingjie, S.K. Grotheim, H. Kvande, Aluminium 64 (1988) 1254–1257. [25] C.S. Sivaramakrishnan, R. Kumar, Light Met. Age 10 (1987) 30–34. [26] Y. Birol, J. Alloys Compd. 443 (2007) 94–98. [27] N.E. Mahallawy, M.A. Taha, A.E.W. Jarfors, H. Fredriksson, J. Alloys Compd. 292 (1999) 221–229. [28] L. Arnberg, L. Backerud, H. Klang, Met. Technol. 9 (1982) 1–6. [29] P.S. Mohanty, J.E. Gruzleski, Acta Mater. 44 (1996) 3749–3760. [30] M.S. Lee, P. Grieveson, Scand. J. Metall. 32 (2002) 256–262.