The solidification behaviour of AlMn melts

The solidification behaviour of AlMn melts

816 Journal of Non-Crystalline Solids 117/118 (1990) 816-819 North-Holland THE SOLIDIFICATION BEHAVIOUR OF AI-Mn MELTS J. KREIS, D.M. HERLACH, H. A...

278KB Sizes 7 Downloads 81 Views

816

Journal of Non-Crystalline Solids 117/118 (1990) 816-819 North-Holland

THE SOLIDIFICATION BEHAVIOUR OF AI-Mn MELTS

J. KREIS, D.M. HERLACH, H. ALEXANDER* and W. WEISS* Institut for Raumsimulation, DLR, D-5000 K61n 90, *Universit~f K61n, Abt. Metallphysik, D-5000 K61n 41, Federal Republic of Germany Small liquid droplets of AI~_×Mn× ( x = 8, 11.8, 17, 21.5 at.%), with diameters in the range 250/1m 1000/lm have been undercooled and solidified by containerless processing during free fall in a drop tube. Undercoolings larger than 150 K have been estimated. The solidified droplets show crystalline phases and metastable quasicrystalline T-phase. However, no quasicrystalline I-phase has been found. The analysis of the results indicates primary nucleation of I-phase which transforms into T-phase during recalescence.

1. INTRODUCTION Ouasicrystalline phases have been the subject of extensive study since the detection of a phase with 54old symmetry by Shechtman et al. ~. The two most important quasicrystalline phases are the icosahedral phase, which is quasiperiodic in three dimensions ~, and the decagonal or T-phase, which is quasiperiodic in two dimensions and periodic in the third 2. These phases can be produced in many systems by a large variety of techniques: solid state reactions, ion bombardment, electron beam heating and rapid quenching 3. Among these, rapid quenching from the liquid state is the most commonly used. Critical cooling rates for the formation of the icosahedral (I-) phase have been reported ranging from 500 Ks ~ to 10~ Ks ', depending on the preparation conditions 4'5. With a few exceptions 6, quasicrystalline phases have been found to be metastable. Clearly, an undercooling of the melt below the melting temperature of the metastable phase is a necessary precondition for its nucleation. An important parameter determining the nucleation threshold is the inter'facial tension between nucleus and melt, which has been described within the negentropic model by Spaepen 7 . This model is based on the assumption that the short range order in the interface is tetrahedral. However, as already pointed out by Frank B the short range order within the undercooled melt should be icosahedral in nature. This in turn implies that the interfacial energy will be smallest for icosahedral nuclei. 0022-3093/90/$03.50 (~) Elsevier Science Publishers B.V. (North-Holland)

Therefore, homogeneous nucleation of icosahedral phase should be possible at moderate undercoolings. In the present work, small dro01ets of undercooled liquid AI-Mn alloys have been solidified by containerless processing in a drop tube under purified He-atmosphere. In this way, heterogeneous container-wall-induced nucleation at low undercooling is completely eliminated and other nucleation processes can become active. Despite the fact that undercoolings have been achieved which are comparable with results by Mueller et al. 4 in a range of AI-Mn alloys and cooling rates have been obtained in the range 103- 104 Ks-', no I-phase is found in the drop tube processed samples. The structure of the AI92Mna and AI~82Mn~,~ droplets consists of a mixture of metasfable quasicrystalline T-phase and supersaturated aluminium (AI,,) solid solution. The analysis of our results indicates that the thermal history during recalescence plays an important role in the development of the metastable phases in AI-Mn alloys. 2. EXPERIMENTAL Alloy ingots of Air ~Mn× ( x = 8, 11.8, 17, 21.5 at.%) were prepared by melting the constituents, with a purity of 99.999 % (AI) and 99.97 % (Mn), in an arc furnace under a purified Ar atmosphere. After the drop tube was evacuated to a pressure below 10 ' mbar, it was refilled with purified He.

J. Kreis et al./Solidification behaviour of A1-Mn melts

Alloy samples were melted inductively in crucibles of glassy carbon and were dispersed as small droplets into the drop tube by forcing the melt through a nozzle ~. The solidified droplets were separated into size fractions by sieving with standard meshes. The microstructures were investigated by X-ray diffraction (CuK=radiation), optical and scanning electron microscopy (including EDX analysis) and differential scanninq calorimetry (DSC).

817

~" as~~n8

~--)t i 1

,J' /Q

' I

38

40

42

L

44

46 ,

2 e, DB3RIBE FIGURE 2 X-ray diffractometer (D ~300/~m)

FIGURE 1 Scanning electron micrograph of a section through an AI882Mn1~~ droplet of diamter D ~850 #m 3. RESULTS AND DISCUSSION Fig. 1 shows a scanning electron micrograph of a section through an AI882Mn, ~ droplet from the 0.8 0.9 mm size fraction. Two different phases are apparent; a very pronounced dendritic microstructure of a Mn-rich phase (14-22 at.% Mn) which is embedded within a supersaturated A I , matrix (~ 3 at.% Mn). Fig. 2 shows an X-ray diffractometer trace from Ale~2Mn,8 revealing the existence of supersaturated AI,s phase and metastable decagonal T-phase. Heating these samples within a DSC up to a temperature of T = 880 K (heating rate: 10 K rain -1) leads to the transformation: Alss + T - , AI6Mn + AI with an onset temperature T× ranging from 720 K to 780 K. T× increases systematically with droplet diameter (decreasing cooling rate). Fixing the

trace from

AIsg2Mn,8 droplets

cooling rate and increasing the melt superheating temperature leads to the same result. Combining low cooling rates with high superheating temperatures produces the most stable T-phase. The heat of transformation, calculated from the area below the transformation peak, is AHx-- ( - 53 + 3)Jg when related to the total mass of the sample. For all the compositions we looked at, no trace of I-phase can be found. This is in constrast to melt-spun ribbons of AI-Mn of similar compositions. In the case of AITesMn2,5, besides those from T-phase and supersaturated A I , additional X-ray peaks are found, which cannot be indexed unambiguously. They probably indicate the presence of AI4Mn '°. It is interesting to note that in this alloy single T-phase is obtained by splatting an undercooled drop onto a Cu cooling substrate at the bottom of the drop tube (cf. Fig. 3). As a first step in the analysis of the results we estimate the undercooling, the most important parameter in the nucleation of the respective phases. Assuming Newtonian conditions, the cooling rate of freely-falling droplets is given by:

i

6 [hm(T (p.Cp.D)

TRT) -F ~.(~(T4 - l'~T)]

(I)

where h,. is the heat transfer coefficient for forced convection from a sphere", p the density, TR'r the

J. Kreis et al. / Solidification behaviour of AI-Mn melts

818

conditions (ElectroHydroDynamic atomization) as in our drop tube. Thus, the question arises of why no I-phase is found in the solidified droplets. This can only be understood if we consider the thermal history of solidification during recalescence, when the release of the heat of crystallization leads to a

1".--)

ats, .4

1"

temperature rise. We propose the following solidification path. At an undercooling of AT > 150 K primary icosahedral phase is formed. During the subsequent temper-

38

I

I

|

4Q.

42

44

"nl

2 e . ~

FIGURE 3 X-ray diffractometer trace from AI78~Mn21~after splating the undercooled droplet onto a Cu-cooling substrate room temperature, F. the surface emissivity, Cp the h e a t capacity and o- the Stefan-Boltzmann constant.

For Ale82Mn, e, we take values of p = 3.26 gcm 3, TRT= 300 K, and ~. = 0.2. As an upper limit for Cp, w e use the value Cp = 1.6 J g ' K ~. Droplets which a r e fully liquid when they reach the bottom of the drop tube form splats. From the mass of these splats, we deduce that droplets _> 0.9 mm diameter do not nucleate during the period of free-fall. By solving the initial value problem (Eqn.(1)) for a total fall-time of 0.55 s we arrive at a minimum undercooling of 150 K. It is likely that the value for smaller droplets, subjected to higher cooling rates, will be greater than this. Such a large undercooling is reasonable when taking into account the experimental procedure ~2. Knapp and Follstaedt ~3 determined the melting temperatures of I- and T-phase to be 85 K and 20 K, respectively, below the liquidus temperature of stable Ala82MnI~a. The nucleation of icosahedral phase is, therefore, only possible at undercoolings (AT) larger than 85 K. At an undercooling of more than 150 K, homogeneous nucleation of I-phase in our droplets is very likely, particularly in view of the icosahedral short range order present in the undercooled liquid. As reported by Bendersky and Ridder ~4, in fact icosahedral phase appears to nucleate homogeneously under similar processing

ature rise due to recalescence I-phase transforms into T-phase. This assumption is supported by experiments in which it has been shown that decagonal T-phase can grow epitaxially onto icosahedral I-phase 15. In a further step, the T-phase can transform into stable AI6Mn until solidification is completed. However, increasing the cooling rate by quenching the undercooled melt onto a cold substrate may interrupt these phase transformations during recalescence. This leads to complete suppression of the formation of the stable crystalline phase as experimentally observed in the splat quenched AI785Mn2,5 sample. A further improvement of heat extraction e.g. by melt spinning ~5 or in the much smaller droplet sizes by EHD ~4suggests that even I-phase can be preserved. 4. SUMMARY Phase selection processes in containerlessly processed AI-Mn droplets have been investigated with respect to the formation criteria of metastable quasicrystalline I- and T-phase and stable crystalline state. The results suggest primary formation of icosahedral phase, due to its low activation barrier for nucleation, followed by transformation into T-phase and crystalline phases during recalescence. This sequence can be interrupted by increasing the heat extraction rate during solidification to produce single T-phase in the AIT~Mn2, s samples. Low cooling rates and high melt superheating temperatures stabilize the T-phase. ACKNOWLEDGEMENTS TWO of the authors (J.K. and D.M.H.) wish to thank Prof. B. Feuerbacher for continuous support and Prof. K. Urban for stimulating discussions. We are

J. Kreis et a l . / Solidification behaviour of AI-Mn melts

also indebted to R.F. Cochrane for carefully reading the manuscript. REFERENCES 1.

D. Shechtman, I. Blech, D. Gratias and J.W. Cahn, Phys. Rev. Lett. 53 (1984) 1951.

2.

L.A. Bendersky, Phys. Rev. Left. 55 (1985) 1461.

3.

C. Suryanarayana and H. Jones, Int. J. Rap. Solid. 3(1987) 253.

4.

B.A. Mueller, R.J. Schaefer and J.H. Perepezko, J. Mater. Res. 2 (1987) 809.

5.

R.J. Schaefer, L.A. Bendersky and F.S. Biancaniello, J. Physique 47 C3 (1986) 311.

6.

M.D. Ball and D.J. Lloyd, Scr. Metall. 19 (1985) 1065; A.P. Tsai, A. Inoue and T. Masumoto, Jpn. J. Appl. Phys. 26 (1987) / 1505; W. Ohashi and F. Spaepen, Nature 330 (1987) 555.

7.

F. Spaepen, Acta Metall. 23 (1975) 729.

8.

F.C. Frank, Proc. Roy. Soc. London, Ser. A 215 (1952) 43.

9.

F. Gillessen and D.M. Herlach, Mater. Sci. Eng. 97 (1988) 147.

10. R.J. Schaefer, L.A. Bendersky, D. Shechtman, W.J. Boettinger and F.S. Biancaniello, Metall. Trans. A 17 (1986) 2117. 11. R.B. Bird, W.E. Stewart and E.N Lightfoot, Transport Phenomena (John Wiley, New York, 1960). 12. D. Turnbull, private communication 13. J.A. Knapp and D.M. Follstaedt, Phys. Rev. Lett. 58 (1987) 2454. 14. L.A. Bendersky and S.D. Ridder, J. Mater. Res. 1 (1986) 405. 15. R.J. Schaefer and L.A. Bendersky, Scr. Metall. 20 (1986) 745.

819