Precipitation of the metastable cubic Al3Zr-phase in subperitectic Al-Zr alloys

Precipitation of the metastable cubic Al3Zr-phase in subperitectic Al-Zr alloys

PRECIPITATION OF THE METASTABLE CUBIC Al&r-PHASE SUBPERITECTIC Al-25 ALLOYS* IN E. NESt The decomposition of Zr in Al-O.18 wt.% Zr alloys has been ...

1MB Sizes 66 Downloads 81 Views

PRECIPITATION

OF THE METASTABLE CUBIC Al&r-PHASE SUBPERITECTIC Al-25 ALLOYS*

IN

E. NESt The decomposition of Zr in Al-O.18 wt.% Zr alloys has been examined by transmission electron The precipitating phase was the metaetable cubic AlsZr, which, during annealing at 460°C, microscopy. was found to be stable within the investigated time ranges (maximum 700 hr). The lattice constant of the cubic AlaZr cell has been measured to 4.08 A. The precipitate nucleation was heterogeneous on dislocations and subgrain boundaries. The precipitation kinetics was diffusion controlled and from the growth rate of the precipitate particles the diffusivit,y of Zr in Al has been estimated to about 10-l’ cm*/sec. The morphology were mostly spherical, while also rod-shaped precipitates, oriented in (100) matrix directions and plates on { 100) matrix planes were observed. In addition more irregularly shaped particles or clusters of particles appeared. PRECIPITATION

DE

LA

PHASE CUBIQUE METASTABLE Al-Zr SOUSPERITECTIQUES

Al&

DANS

LES

ALLIAGES

La d6composition de Zr dans les alliages Al-0,18x Zr en poids a Bt6 &udi& par microscopic Qlectronique par transmission. La phase qui prpcipite est la phase Al,Zr cubique m&stabl& Les autaurs ont trouv6 que, au tours du recuit B 46O”C, cette phase est stable, au moins pour la durBe des exp&iences (max. 700 heures). La constante de r&eau de la cellule cubique A1,Zr a 6tB mesuree et la valeur obtenue est de 4,08 A. La nucl6ation du pr6cipit6 est h&rrog&ne sur les dislocations et les joints des sous-grains. La cin&ique de pr6cipitation est contr616e par la diffusivit6 de Zr Al a Bt6 Bvalube B lo-” cms/sec environ, B partir de la vitesse de croissance des particules de pr&ipit& La morphologie est le plus souvent sphbrique, bien que des p&ipit& en forme de baguettes orient&s le long des directions (100) de la matrice, et des plaquettes dans les plans {loo) de la matrice aient BtP observ& Bgalement. 11 est apparu, en outre, des particules de forme plus irr&uli&e ou des agglomCrat,s de particules. AUSSCHEIDUNG DER METASTABILEN KUBISCHEN AlsZr-PHASE SUBPERITEKTISCHEN Al-Zr-LEGIERUNGEN

IN

Die Entmischung einer Al-O,18 Gew. % Zr-Legierung wurde elektronenmikroskopiach untersucht. Es bildetcn sich Ausscheidungen der metastabilen kubischen Al*Zr-Phase, die bcim Anlessen bei 460°C in der kubischen den untersuchten Zeitperioden (maximal 700 h) stabil waren. Die Gitterkonstante Al,Zr-Zelle wurde zu 4,08 A bestimmt. Die Keimbildung der Ausscheidungen erfolgte heterogen an Versetzungen und Subkorngrenzen. Die Aus der Wachstumsgeschwindigkeit der AusscheiAusscheidungskinetik war durch Diffusion bestimmt. dungen wurde die Diffusionskonstante fiir die Diffusion von Zr in Al zu etwa lo-l1 cm*+ abgeschiitzt. Die Teilchen hatten vorwiegend Kugelgestalt; es wurden jedoch such Stiibchen in (lOO)-Matrixrichtung und Platten auf {lOO}-Matrixebenen beobaohtet. AuDerdem bildeten sich unrcgelm&lJig gcformte Ausscheidungen oder Cluster von Teilchen. 1. INTRODUCTION

For the high strength Al-Zn-Mg-Cu and Al-Zn-Mg alloys small additions of Zr have a profound influence on the properties of the alloys. It is well known that Zr has a desirable effect on recrystallization behaviour ;(l-S) Zr also has been reported to improve toughness, stress-corrosion resistince and quench sensitivity of aluminium alloys.(a*6*7) It has further been established that these improved metallurgical properties of the Zr-bearing alloys are due to a high density of precipitate particles of a metastable cubic Al,Zr phase.(4J?*sJ To achieve a more fundamental understanding of the role the Al,Zr precipitates play as to the metallurgical behaviour of the Zr-bearing alloys, several workers(4*5*9)have studied the decompositidn of the binary Al-Zr system using transmission electron microscopy. These studies, however, were all concerned with alloys of a relatively high Zr concentration Received August 31, 1971; revised October 12, 1971, t Central Institute for Industrial Research, Forskningsveien 1, Oslo 3, Norway. l

ACTA METALLURGICA,

VOL. 20, APRIL

1972

499

(0.5-1.1 wt. % Zr compared to 0.1542% Zr for commercial Zr-bearing alloys). The most characteristic aspects of the decomposition as reported in the literature were the formation of either a high density of coherent particles(5*Q)or large fan-shaped precipitate arrangements.‘4*5*9) The present work was initiated by the discovered absence of these features of the Zr-decomposition in an Al-Zr alloy of a low subperitectic Zr concentration (peritectic concentration 0.28 wt. %(l”J1)). Hence, aa the precipitation behaviour of Zr appeared to be different in less saturated Al-Zr solid solutions, a detailed transmission electron microscope investigation of the decomposition in terms of precipitation kinetics, precipitate nucleation, growth and particle distribution’ and morphology in Al-O.18 wt. % Zr alloys haa been performed. 2. EXPERIMENTAL

Two alloys were prepared; an Al-618 wt. % .Zr axid an Al-O.18 wt.% Zr-O.5 wt.% Cr. (Cr are also fre. quently added as a grain refining agent in aluminium alloys; One of the initial objectives of this work was

ACTA

600

METALLURGICA,

to study the effect the presence of Cr had on the decomposition of Zr in Al. Except for the presence of a low density of large Al&r particles in the AlZrCr alloy, the decomposition of Zr in terms of precipitation kinetics, particles distribution and morphology appeared to be the same in the two alloys. Therefore no further attention will be paid to the presence of Cr in one of the investigated alloys.) The melting and casting were performed similar to what has been reported by Ruym.(*) The main impurities in the as-cast alloys were: Fe = 0.05x, Si = 0.01x, Mg = 0.01%. Specimens, 2 mm thick, were spark cut from the 20 mm diameter castings. The specimens were annealed at 460% in a salt bath furnace for periods ranging from 24 hr up to 700 hr. Foils for electron microscope investigation were prepared by electro-polishing in a methanol-nitric acid solution in a standard way. The foils were examined in a Philips E.M. 300 electron microscope. 3. EXPER1lKEN’I’A.L

RESULTS

Long annealing times (more than 24 hr at 460%) were required before any precipitation could be detected. Even after annealing for several hundred hours the maximum linear dimention of the precipitate particles {except the rod-shaped pre~pita~s) were usually less than a 1000 .k. The precipitates were found to appear in a variety of shapes, with the nearly spherical precipitates being the most common,

VOL.

20,

1972

Fig. l(a). Long rod-shaped precipitates oriented in (100) matrix directions, frequently in colonies, were also observed [Fig. l(b)]. Diffraction patterns show that the precipitates have a simple cubic structure with the same orientation and nearly the same lattice constant as the aluminium matrix. A diffraction pattern from a foil in a [l lo] orientation is given in Fig. l(a). Both the precipitate structure and orientation relations~p are consistent with recent work on the decomposition of Al-Zr alloys,(4*aJ+*B) which report on the formation of a metastable cubic Al,Zr phase. Even in specimens annealed for nearly 700 hr at 460% no precipitates were observed that could be identified as the equ~ib~um tetragonal Al&r phase. 3.1 Kinetics of precipitation The growth of the precipitate particles was slow. In specimens annealed for about 100 hr at 460°C the precipitates appeared with an average radius of about 400 A onIy. The diagram in Fig. 2 shows the average precipitate radius vs. the square root of the annealing time. The values for the average precipitate radii have been obtained from regions in the foils which have about the same density of nearly spherical particles, As can be seen from Fig. 2 the growth appears to be a linear function of the square root of the annealing time for shorter annealing times.

Pm. 1. (a) Dark field image and diffraction p&tern from nearly spherical precipitates in a (1 IO) surface foil. (b) Rod-shaped precipitates in a (001) foil.

NES:

PRECIPITATIOE

I

IO

OF

I

I

20

30

Al,Zr

1

IN

SUBPERITECTIC

estimates the unit cell to be about 1 per cent bigger than that of aluminum, while Izumi and Oelschl&gel@) found the difference to be 0.45 per cent with the AlsZr cell as the largest. Izumi and Oelschiigel based this misfit value on the size of the coherent strain contrast image. Figure 3 shows a matrix (020) two-beam diffraction pattern where the transmitted beam has been deflected to allow the (OiO) precipitate reflection to pass down the centre of the column. The weaker spot next to the (OiO) precipitate reflection represents a (010) precipitate double reflection of the (020) matrix reflection.* The distance between these two spots relative to that of the two matrix reflections gives a direct measure of the precipitate lattice misfit. The misfit is defined as:

l ANNEALING TIME, HOURS l”2 FIU. 2. Precipitate radius vs. square root of annealing time at 460%.

A theoretical model for the diffusion controlled growth of a random distribution of precipitates has been proposed by Ham. (12) This theory predicts the following functional relation between particle size, r and annealing time, t. r (t) =

kfi

(3.1)

k = 2(c, - c,)~/~/c~, D is the diffirsivity of Zratoms in aluminium, c,, is the solute concentration at t = 0, c, is the solute solubility (in this case at 460%) and cl, is the concentration of solute in the AlsZr precipitates. The model requires that the solute-drained region surrounding each precipitate is small relative to the average distance 1 between the precipitates, that is, where

601

Al-Zr

6=2

a1 -

-

a2

where a, and a2 are the matrix and precipitate lattice constants, respectively. A series of diffraction pattern have been analysed and the result is 6 = 0.8 f 0.1 per cent. As the matrix lattice constant is a, = 4.05 A this gives for the metastable cubic AlsZr phase; a2 = 4.08 A. 3.3 Preci@tate distribution and morph&gy The precipitates are found to be inhomogeneously distributed and they appear with higher density along dislocation tangles and subgrain boundaries. Figure 4 shows a bright field micrographs where the foil has been tilted to bring the dislocations slightly out of contrast and the particles are seen to be aligned along the subgrain boundaries. This effect is even more

1 < z/i%. Ham shows that ifr(0) = 0, equation (3.1) is a good approximation for E(t)/c,, < #, where E(t)is the average solute concentration in the matrix at time t. The corresponding time interval is indicated by the fully drawn line in Fig. 2, as can be seen from the diagram this interval covers the linear part of the r(t) vs. &relationship. All of the parameters in the constant k [equation (3.1)] are known or can be found from an equlibrium phase diagram.(lO*ll) Thus, based on the linear relationship in Fig. 2, equation (3.1) can be solved with respect to the difisivity D. This gives an estimate for the diffisivity of Zr in aluminium at 460% equal to about 10-l’ cmz/sec. 3.2

The lattice parameter AlsZr pha..se

of the ncetastable cubic

The lattice constant for the metastable cubic Al,Zr cell has not been determined accurately. Ryum(4)

Fra. 3. (020) two-beam diffraction pattern with a (OIO) and a (010) precipitate reflection and double reflection. l As can easily be seen from Fig. 2, the two precipitate spots do not fall exactly on the axis through the matrix reflections. This angular deviation corresponds to a small precipitate rotation relative to the matrix of about 0.2”.

ACTA

602

METALLURGICA,

VOL.

20,

1972

Aa shown in p”ig. l(b) the rod-shaped p~ci~i~~s are oriented in (100) matrix directions. Rod lengths in specimens annealed for more than 350 hr are ranged from about 0.5 to about 2 pm and the rod thicknesses vary from 200 up to 300 A. A characteristic feature is the frequent appearance of closely spaced pairs of rods, Fig. 6, even 3 or 4 parallel and closely spaced rods have been observed. Another unresting observation is that the rods in the darkfield images (Fig. 6) appear to be discontinuous in the long direction

Fra. 4. Precipitates segregated on the subgrain bound&es, after 696 hr at 460°C.

pronounced in Fig. 5 where the dark field micrograph [Fig. 5(b)] clearly demonstrates the precipitate distribution. Fignre 5(b) also illustrates the variety in particle shapes ; the morphology varies from nearly spherical through irregularly shaped clusters to rodshaped particles.

FIG. 6. P~oipi~tion

FIG. 6. Closely spaoed pairs of rod-&ape pre0ipitet.w

on a grain boundary, after 144 hr st 460°C.

(a) Bright %eld image.

(b) Dark field image.



NES:

PRECIPITATION

OF

AlrZr

IN

.#f&.d guw

FIN. 7. Rods partly chopped up into rows of spherical particles, after 70 hr at 460°C.

SUBPERITECTIC &.. . . :,mpmen

Al-Zr

603

tB

asnicrograph where the precipitates marked A, B and C actually are clusters of precipitates with each cluster consisting of a set of orthogonal plates parallel to (100) matrix planes. The diameter of the plates are ranged from about 200 to 1OOOAand the thickness varies from about 20 tc about 40 A. The most common particle appearance, however, is in the form of more or less spherically shaped precipitates. The nearly spherical character of the particles marked A and B in Fig. 9 is demonstrated by the concentric extinction contours outlining the precipitate-matrix interface. Except for the larger precipitate sizes in foils prepared from specimens annealed for longer times, the general appearances of the precipitates as described above did not seem to change with annealing time, in the time range from about 25 up to about 700 hr. 3.4 Interfacial

dielocation structure

It was, both for the irregularily shaped clusters and the spherical particles (Fig. 9), difficult to distinguish between interfacial and matrix dislocations and to determine the dislocation Burgers vectors. For the rod-shaped precipitates, however, the interfacial dislocations could to some extent be analysed in terms of Burgers vectors and structural arrangements. Figure 10 shows (100) type precipitates in a (110) surface foil. The dislocation segments along the rod marked A, Fig. 10, exhibits a lath like structure, while the precipitates marked B, C and D appear as pairs of

FIG. 8. Precipitate clusters each consisting of a set of orthogonal platee parallel to (100) matrix planes.

of the rods. This observation may in part be due to the variation in the black-white contrast caused by the precipitate-matrix interfacial dislocations (Section 3.4) but as can be seen from Fig. 6 some of the rods actually consists of smaller isolated parts aligned in a row. This chopped up character is clearly demonstrated by the two tiny orthogonal “rods” in Fig. 7. The smallest spacing of the nearly spherical parts of these configurations is about 10 A only. As already mentioned the precipitates may appear in clusters which exhibit a very irregular shape as illustrated by the darkfield image in Fig. 5(b). Other clusters, however, are made up of smaller parts where the morphology of each part easily can be established.

FIG. 9. The spherical morphology is outlined by the concentric extinction contours.

ACTA

604

.--

-

FIG. 19. Rod.shaped

METALLVRQICA,

.,

precipitates with dislocations in contrast.

the

interfacid

VOL.

20,

1972

parallel rod8 where the interfactial di8locaCons seem to be pair8 of helices (B,C) or rows of coaxial dislocation 1OOpS (D). Figure 11 shows two (111) two-beam bright field image8 and a precipitate dark field reflection of two closely spaced rods oriented in the [OOl] matrix direction. The interfacial dislocation structure aa seen in Fig. 11(a) is interpreted a8 two sets of equally spaced coaxial loops enveloping each one of the two precipitate rods. The spacing of the loops is about 200 A. The somewhat obscured dislocation contrast in the [ill] reflection [Fig. 11(b)] is probably caused by local strain in the precipitate-matrix interface. It can, however, easily be seen from Fig. 11(b) that the rows of interfacial dislocation loop8 are out of contrast in the [ill] reflection. Inspection of the stereographic projection in Fig. 11 thus give8 the Burgers vector of the loops a8 either a/2 [loll or a/2 [Olr], both making an angle of 45” with the [OOl] direction of the rods.

I

a

FIG. 11. Interfme

dislocation structure

along two perellel ctnd closely spaced rods. images. (0) Dark field image.

(a, b) !Cwo.beam bright field

NES:

PRECIPITATIOX

OF

Al,Zr

As the rodshaped precipitates in addition to had&g an interfacial dislocation structure are showing a complete lack of matrix strain contrast, it can be concluded that these precipitates are fully incoherent. When coherency is completely lost, the spacing, d, of the interfacial dislocations is given by the Brooks formulao3~ d

=

lb1

s

where lb/ in this case is the magnitude of the Burgers vector in the direction of the rod and 6 is the misfit as given by equation (3.2). As the Burgers vector of the interfacial dislocation loops, Fig. 11, is oriented at an angle of 46” relative to the rod and d is about 200 A, this gives a misfit 6 of about 1 per cent. This is in good agreement with the 0.8 per cent misfit as measured from the diffraction patterns (Section 3.2). The Brooks formula [equation (3.3)] predicts the breakdown of coherency for a spherical precipitate when the particle diameter reaches a size of about 300 A. However, even in foils annealed for only short times and with the particle sizes ranged from 300 to 400 A, the characteristic strain contrast associated with a coherent spherical particle was only infrequently observed. For larger precipitates matrix strain contrast was not observed (Fig. 9). Ryum,t4) however, reports strong strain contrast from the spherical metast,able Al,Zr precipitates with diameter up to 1500 A. 4. DISCUSSION

The most characteristic aspects of the decomposition of Zr in Al-Zr alloys as reported in recent electron microscopy studies(4*5*9)are : the formation of either a high density of coherent spherical particles by a continuous precipitation process(5*9) or large precipitate arrangements of rod-shaped particles distributed into fan-shaped configurations.‘4~5J’) Contrary to these observations the present’ work shows that the metastable A1,Zr phase, in a subperitectic Al-Zr alloy, appear as incoherent particles segregated on dislocations and subgrain boundaries. Fan-shaped precipitates were not observed, while rod-shaped particles oriented in (100) matrix directions were present in the foils. The possibility of having the Zr-decomposition, as observed in the present investigation, influenced by the impurities, notably the relatively high iron content, needs to be considered as the presence of iron has been reported to influence the recrystallization properties of Al-Zr alloys.o4) On the decomposition of Zr in an Al-Zr solid solution, 3

IN

SUBPERITECTIC

Al-Zr

505

h&&ver, the tiork-by. Ryumt4) (iron content 0.05 %> the same as in the present alloy) and Sundberg et aLf5) (iron content <0.005%) show that the iron content do not appear to affect the Zr precipitation. Hence, the differences between the present observations and the previous works on Al-Zr alloys are suggested to be caused by a difference in the precipitation kinetics in the highly supersaturated Al-Zr alloys(4*5*9) and the present Al-O. 18 % Zr alloy. In line with this suggestion the fanshaped precipitates are interpreted as being caused by a discontinuous precipitation effect,05) while the present work shows the precipitation to be continuous and diffusion controlled. The observed precipitate distribution suggests a heterogeneous nucleation mechanism, with the dislocations and the subgrain boundaries acting as nucleation sites for the particles. As the electron microscope micrographs reveal no sign of a matrix strain contrast around the precipitates, even for the smallest observed precipitates. this suggests that the nuclei were incoherent. Thus the precipitation in this case may be explained in terms of the classical theory for nucleation on dislocations as given by Cahn.06) Ryumc4) explains the improved recrystallization properties of Al-Zr alloys in terms of the following mechanisms : (1) the particles lead to a more uniform distribution of dislocations during deformation and (2) have also a pinning effect on grain and subgrain boundaries during annealing. The precipitates observed by Ryum were all associated with a strong matrix strain field, while the present results show the particles to be fully incoherent. The recrystallization properties of the present alloys have not been investigated, however, the observed heterogeneous nucleation, causing a fine dispersion of small precipitates Fig. 4, may also have a desirable effect on the recrystallization behaviour. The subgrain structure in Fig. 4 appears to be very effectively pinned by the precipitates. A startling aspect of the precipitation of the metastable Al,Zr phase, as reported in the present paper, is the appearance of particles with widely different morpologies, i.e. (1) spherical particles, (2) plates on (100) planes and (3) rod-shaped particles in (100) directions. In addition irregularly shaped clusters, which may be interpreted as a mixture of all of the above mentioned morphologies, are frequently observed. As the cubic lattice of the precipitates is only slightly larger than the aluminium host lattice, and as the aluminium lattice is almost elastically isotropic, the formation of spherical particles should be expected. Even if the nearly spherical modification by far is the most frequently observed, the question why also

506

ACTA

MMETALLtTRGICA,

plate-shaped and rod-shaped precipitates are formed will now be considered. An attempt has been made by Izumi and Oelschl&gel(l” to explain the formation of the fan-shaped precipitates in terms of a precipitation effect similar to that reported by Hornbogen and Roth(ls) on the distribution of y’ prec~pita~s in Ni-Al-alloys. According to Hornbogen and Roth; if the mis-match between the precipitate and matrix lattices exceeds a certain critical value, then the precipitates arrange themselves in rows and the rows may coalesce to form rods. The preferential orientation of the rods is interpreted in terms of the elastic anisotropy of the matrix. This might seem as a plausible explanation for the observed rows of sphe~cal particles, Fig. 7. However, as the driving force causing the precipitate distribution observed by Hornbogen and Roth is supposed to be the coherent strain field of the particles, this mechanism does not seem to explain the appearance of rows of incoherent particles and incoherent rod-shaped precipitates. In addition the aluminium matrix is only weakly elasticly anisotropic compared to the X-alloys studied by Hornbogen and Roth. The precipitation is observed to be closely associated with the matrix dislocations and the results show strong interactions between matrix and interfacial dislocations. The dislocations probably also act as sources and sinks for the exchange of point’ defects that are necessary in order to accommodate the particle without generating severe mritrie strain. Therefore, a precipitate gro~~hldislocation climb interaction may be responsible for the deviations from a spherical

VOL.

20,

1972

morphology. Precipitation on a dislocation associated with helical climb may explain the appearances of long rod-shaped precipitates. ACKNOWLEDGEMENTS The author is deeply grateful to Dr. J. Gjsnnes for many stimulating discussions. for reading the manuscript and for being allowed to use the electron microscope at. the UniversiQ of Oslo? Inst,it,ute of Physics. My special thanks are also due to Dr. Chr. Simensen for supplying one of the investigated alloys. This work was done under the auspices of the Royal Norwegian Council for Scientific and lndustrial Research through the Central Institute for Industrial Research. REFERENCES 1. H. HUSCHEA and H. NOWOTKS. ~e~u~~z~~g~ l2, 6 (1956). 2. 3. 4. 5.

H. YAMADA. J. Japan Inat. .Gg& Bfela-fs 10,263 (1960). E. DI Russo, Alluminio-Nuova Me&K 87, 349 (1967). N. RYUM, Acta Met. 17, 269 (1969). hf. SUNDBER~. R. SUNDBERG and B. JACOBSO?;. ,Jewko,lt.

Annlr 155, 1 (1971). 6. E. Dx Russo, Alluminio-Nuovu Metall. 88,505 (1964). i. hi. CONSERVA, E. DI %.X%30&nd 0. CALONI, &fet&. !ha~i8.

2, 1227 (1971). 8. B. THKXDAL and R. SUNDBERO. J. I?&. &felaEe 97, 160 (1969). 9. 0. IZUMI and D. OELSCHLJ~QEL, 2. X&al& 60,845 (1960). 10. M. Hanrsm, Conathtion~ of Binary Alloya, p. 152. McGmwHill (1958). 11. Ii. R. VAN HORX, Aluminivm,

12. 13. If.

15. 10. ii. 16.

Vol. I, p. 881. Amrricsn Society for Metals (1967). F. S. HAM. J. Phv8. solids 6. 335 (1958). H. BROOKS,Met& Interjaces;p. 20: An&can Society for Mnt,als I lRK21. S. NISEIKAWA, N. NAGASHIMA,T. S. SAWAQIWII and S. KOBAYASHI, J. Lighf Met&, 18.617 (19681. E. NES and N. R