Significance of continuous precipitation during creep of a powder mettallurgy aluminum alloy

MATERIALS SCIENCE & ENGINEERING ELSEVIER

A

Materials Science and EngineeringA216 (1996) 161-168

Significance of continuous precipitation during creep of a powder metallurgy aluminum alloy "

Lubo

i

Kloc 1,a, Emanuela Cerri a, Stefano Spigarelli a, Enrico Evangelista a, Terence G. Langdon b,*

~Deparmwnt of Mechanics, U~ziversityof Ancona, 1-6013i Ancona, Italy bDepartments of Materials Science and Mechanical Engineering, University of Southern California, Los Angeles, CA 90089-1453, USA

Received 14 February 1996; revised 24 April 1996

Abstract Experiments were conducted to evaluate the creep properties of a 2024 aluminum alloy fabricated by powder me@lurgy processing. The creep curves exhibit a minimum creep rate followed by an extended tertiary stage prior to failure. Using the values of the minimum creep rates, the apparent stress exponents are high and variable suggesting the presence of a threshold stress. Observations using transmission electron microscopy (TEM) reveal the occurrence of a continuoas precipitation of fine particles during the tests. Although the density of these particles is dependent upon the testing conditions, quantitative measurements show that their average size is of the order of ~ 60 nm under all conditions. A temperature compensated {ime is introduced to describe the evolution of fine particles during testing, and this permits the development of a relationship which can be used to estimate the density of fine particles under any selected conditions. Keywords: Aluminmn alloy; Creep; Particles; Powder metallurgy;Precipitation; Stress exponent; Threshold stress

1. Introduction Commercial A1 alloys produced by powder metallurgy techniques are widely used as the matrix alloys in discontinuous SiC-reinforced composites. To date, there have been several reports on the creep properties of these composites at elevated temperatures [1-8]. However, only limited data are available on the creep behavior of unreinforced aluminum alloys. For example, Krajewski et al. [9,10] compared the creep properties of unreinforced and reinforced 2219 A1 alloys and Park et al. [11] investigated the high temperature deformation of an unreinforced 6061PM (powder metallurgy) A1 alloy. The present paper describes an investigation of the creep behavior of an unreinforced 2024PM A1 alloy. This material is of interest because of earlier reports on the high temperature properties of 2024PM A1 corn• * Corresponding author, 1Permanent address: Institute of Physicsof Materials, Academyof Sciences of the Czech Republic, 616 62 Brno, Czech Republic.

posites [12,13]. As will be demonstrated, the results obtained in this investigation show that the creep data are influenced significantly by the occurrence of continuous precipitation during testing at elevated temperatures. Evidence for the occurrence of concurrent precipitation is not common in creep testing because, in principle at least, it is generally considered that any precipitation will be completed during the thermomechanical treatment preceding testing. Nevertheless, Blum and Reppich [14] gave a detailed discussion of the effect of concurrent precipitation during creep and the present results provide additional information on the evolution in microstructure during the creep of the 2024PM alloyl

2. Experimental material and procedures The experiments were conducted using a 2024PM A1 alloy produced by Atures (Novara, Italy). The alloy ,was fabricated by the cold compaction of powders into l~illets, with subse'quent sintering at 773 K, extrusion as rods at 723 K with a reduction ratio of approximately

162

L. Kloc et al./ Materials Science and Eng#~eering.4216 (1996) Idi-I68

Table i Chemical composition of 2024PM alloy (in wt.%)

,¢,,,,,

si

Fe

Cu

Mg

Mn

AI~O3

A1

0.24

0.26

4.42

1.47

0.56

0.13

balance

",~_ 25:1 and cooling in air to room temperature. Table 1 shows the chemical composition of the 2024PM alloy (in wt.%). The creep tests were conducted in air under conditions of constant load using cylindrical specimens machined parallel to the extrusion direction with gauge lengths of 25 mm and reduced diameters of 5 mm. Prior to testing, the specimens were solution treated for 5 h at a temperature of 673 K and then cooled to room temperature at a rate of ~ 60°C/h. Inspection by transmission electron microscopy (TEM) showed that the grains were reasonably equiaxed in the untested condition with an average size of ~ 3 /Lm. The tests were performed over a range of stresses from 25 to 50 MPa and at temperatures in the range from 523 to 603 K. During each test, the strains were continuously reco{ded using two linear voltage differential transducers, and the testing temperatures were maintained at the selected values to within +_ 2 K. The majority of specimens were pulled to failure but three of the slower tests were interrupted without failure to permit microstructural observations during testing. Specimens for examination by TEM were cut from within the gauge lengths of selected specimens after testing, and they were ground mechanically to a thickness of ---150 #m and then electrolytically polished to perforation at room ~:emperature using a solution of 33% H N O 3 in methanol. Observations by TEM showed that there was a continuous precipitation of fine particles during the creep testing, and this precipitation was especially significant during long-term testing at the highest temperature of 603 K. Measurements were taken directly from TEM photomicrographs in order to determine the density of fine particles, N,,, where N~ is defined a s np/A'c, where ~Zp is the measured number of fine particles within an area A and z is the foil thickness estimated from the extinction contours. Inspection showed that the shapes of the fine particles in the two dimensional sections varied from reasonably circular to elongated: in this report, the average particle size, d, is defined as the diameter of the reasonably circular particles and as the average of the measured widths and lengths for the elongated particles.

.

,'1|..

,,,.,-..

tt,

Fig. 1. Microstructure of the 2024PM alloy before creep testing.

3. Experimental results 3.I. Microsn'ucture prior to creep test#zg

Large primary particles were visible in the microstructure of the untested material after solution treatment: an example of the microstructure is given in Fig. 1. Inspection showed that these particles were present on the grain boundaries and within the grains. From X-ray analysis and diffraction patterns, the particles were identified primarily as A12Cu and AI2CuMg [15] with an average size of ,---450 nm and a volume density of the order of ~1.5x1018 m -3, In some grains there was also a dispersion of fine particles with diameters of < 200 nm: these particles were distributed inhomogeneously with a volume density estimated as ~(7.3+4.0) × 1018 m -3. 3.2. Characteristics of creel) behavior

Typical creep curves of strain, 6, versus time, t, are shown in Fig. 2 for an absolute testing temperature, T, of 548 K and applied stresses in the range from 30 to 50 MPa. These curves are representative of aU testing conditions and they show a very brief primary stage leading through a quasi-steady-state region into an extended tertiary stage prior to failure. The nature of the quasi-steady-state region is illustrated more clearly in Fig. 3 where the instantaneous creep rate, ~, is plotted against the normalized time of testing, t/tr~, where tr~ is the time to rupture, for two tests conducted at different

L. Kloc et al. / Materials Science ai~d Engineering A216 (I996) 161-168

i63

-3 0.04 ~

2024PM

14

It

o.03

/

11"

T= 48K • 30MPa

-4

•

-5

40MPa

0.02 0.01 0,00

-

~

0,0

0.2

0,4

0,6

,~. E .tO v

-6

o

-7

2024PM T [K]

* 523 ,548 • 573 • 603

-8

t [106s] -9

Fig. 2. Creep curves at a temperature of 548 K.

1.2

stress levels at a temperature of 603 K. Inspection shows that a reduction in stress leads to a longer fraction of the creep life in which the creep rate decreases or remains essentially constant. It is therefore apparent from Figs. 2,, and 3 that, unlike the creep of pure A1 or A1 solid solution alloys [16,17], there is no well-defined regon of steady-state flow in this alloy and instead it is necessary to analyze the creep data in terms of the occurrence of a minimum in the creep rate instead of a true steady-state condition. The total failure strains in these experiments ~yere generally ~ 5%. The recorded values of the minimum creep rate, km, are logarithmically plotted against the applied stress, o-, in Fig. 4 for the tests conducted at 523-603 K. It is clear from this plot that the values of the apparent stress exponent, n, defined as (8 in ,~/8 in o-), are variable over the range of conditions used in these experiments, and these values become very high at the lowest stress levels. This type of behavior is usually interpreted in terms of the presence of a threshold stress, and the threshold stress may arise from several different protO-S

2024PM

lilRiRiRiilml•

10-4

iBRil• 40MPa

10-5

i/£ i0_6 '~ 10-7

rI

il 1 ii • Iilllllli

emil 25MPa

IllllI

10-8 T=603K 10-9

I

0,0

0.2

I

I

0.4 "' 0.6

I

0.8

1.0

t/tR Fig. 3. Instantaneous creep rate versus fraction of the rupture time for two tests conducted at 603 K . '

1.3 1.4 1.5 t.6 log(o') [MPa]

t.7

t.8

Fig. 4. Minimum creep rate versus applied stress, showing the presence of a threshold stress.

cesses including the bowing of dislocations between particles [18,19], the back stress associated with the local climb of dislocations over particles [20,21] or the stress associated with detaching a dislocation from a particle which exerts an attractive interaction [22-24].

3.3. Microstructures after creep testing Several specimens were examined by TEM after creep testing and it was concluded that there was a continuous precipitation of fine particles during high temperature creep, with the precipitation becoming especially extensive in the tests conducted at the highest temperature for the longest times. Examples of this microstructural evolution are shown in Figs. 5-7 for testing temperatures of 548, 573 and 603 K, respectively. In Fig. 5, for a temperature of 548 K. the TEM photomicrographs show, at two different magnifications, the microstructure after testing to failure at 50 MPa for 11.2 h in (a) and (b) and the microstructure after interrupting the test without failure at 30 MPa after 2019 h at (c) and (d). Inspection shows that there is a dispersion of fine particles throughout the matrix after testing for 2019 h (Fig. 5@) and (d)) but there is less evidence of this dispersion after testing for only 11.2 h (Fig. 5(a) and (b)). As noted earlier, slow cooling from the annealing temperature led to Mg precipitation and the formation of A12CuMg particles. It seems likely that only limited Mg remains in solid solution so that a very long testing time is required at 548 K in order to complete the precipitation process. In Fig. 6, all of the TEM photomicrographs were recorded after failure when testing (a) at 50 MPa for 1.48 h, (b) at 40 MPa for 8.78 h and (c) at 30 MPa for 94 h. The photomicrographs in Fig. 7 were recorded (a)

164

L. Kloc et al./ Materials Science and Engineerbzg ,4216 (1996) 161-168

a

C,

d

Fig. 5. Microstructures after creep testing at 548 K (a) and (b) to failure at 50 MPa after 11.2 h and (c) and (d) at 30 MPa after interruption of the test at 2019 h.

after failure at 40 MPa in a test for ~ 1.58 × 103 s (0.44 h), (b) after interruption of a test without failure at 30 MPa after ~ 2.07 × 104 S (5.75 h) and (c) after failure at 25 M P a i n a t e s t for ~303 h. It is apparent from inspection of these photomicrographs, and especially from Fig. 7(c), that there is a significant increase in the density of very fine particles as a result of the creep testing at elevated temperatures. Measurements were taken to determine the distributions of sizes for the fine particles, having sizes up to 180 nm, precipitated during creep. An example is shown in Fig. 8 for tests conducted at 603 K under stresses of (a) 40, (b) 30 and (c) 25 MPa, respectively: these distributions correspond to the three microstructures shown in Fig. 7. The size distributions in Fig. 8 are plotted in the form of N~/NT, where Nzx represents the number of fine particles contained within each measurement increment of 20 nm and NT is the total number of fine particles. It is clear from Fig. 8 that the fine particles are reasonably stable with sizes which are generally < 140 nm and lying within the range of 40-80 nm. For the three experimental conditions shown in Fig. 8, the density of fine particles, Nv, increased from ~4.1 x 10is m -3 at 40 MPa to ~ 1.3 x 1019 m -3 and

6.7 x 1019 m -3 at stresses of 30 and 25 MPa, respectively. On the other hand, despite the increase in Nv with decreasing applied stress, the average dimensions of the fine particles at 603 K, obtained fi'om Fig. 8, were ~ 65, ~ 54 and ,-- 63 nm for the three stresses of 40, 30 and 25 MPa, respectively. Thus, these measurements show that the average particle size remains reasonably constant and of the order of ~ 60 nm under these testing conditions. In addition, similar average dimensions were recorded for the fine particles after testing at the other temperatures.

5.4. Quantitative measurements of the effect of continuous precipitation The preceding observations demonstrate that microstructural evolution through the precipitation of fine particles depends upon both the temperature of the test and the total time of testing. Tlfis suggests that the precipitation of fine particles may be represented by using the concept of a temperature compensated time, 0, where 0 is defined as t e x p ( - QIRT), where t is the total time of the test, Q is the appropriate activation energy associated with the precipitation process and R is the gas constant. Assuming, as a reasonable first

165

L. Kloc et al. / Materials Science and Enghzeer#zg A216 (1996) 161-I68

.,/ a

C

Fig. 6. Microstructuresafter creep testing to failureat 573 K: (a) at 50 MPa for 1.48 t~, (b) at 40 MPa for 8.78 h and (c) at 30 MPa for 94 h. approximation, that precipitation is controlled by lattice self-diffusion in aluminum, it follows that {2 143.4 kJ mol-~ [25]. Table 2 gives the experimental values of the density of fine particles, N~, as a function of the test conditions and Fig. 9 shows a plot of N~ against the temperature compensated time, 0. It is apparent from Fig. 9 that all of the datum points fall on or about a single line, thereby confirming the validity of using 0 as a normalizing parameter. "

4. Discussion

The 2024PM alloy used in this investigation exhibits a continuous precipitat{on of fine particles during high temperature creep, with the extent of the precipitation depending upon the testing temperature and the total time of testing. This precipitation contributes, at least in part, to the magnitffde of the threshold stresses, and therefore the high values'.of n, a~pparent in Fig. 4. Two important conclugidns ~ a y be reached from the measurements recorded in¢ the preceding section. First, although the density of fine particles changes depending upon the precise exp~rin~entaf conditions, the average size of the fine particles remains reasonably constant

and of the order of ~ 60 nm under all experimental conditions. Second, the evolution of fine particles within the matrix may be adequately described through the use of a temperature compensated time, as shown in Fig. 9. It is important to note that Fig. 9 was constructed by assuming that the activation energy associated with precipitation is equal to the value for lattice self-diffusion in pure aluminum. If it is assumed instead that the precipitation process takes place at nucleation sites associated with the presence of Mg solute atmospheres around the dislocations, the activation energy contained within the value of 8 is replaced by the value for diffusion of the Mg solute. However, the activation energy for diffusion of Mg in A1 is ,-~ 130.5 kJ tool- i [26] and the experimental measurements of Nv are not sufficiently accurate to permit any meaningful distinction between these two activation energies. It follows from the experimental data in Fig. 9 that, at least to a first approximation, the precipitation process obeys a relationship of the form 2(,, : N,o + ANv(t - e - 0/~')

(1)

where Nv0 is the initial density of fine particles, ANy is the total additional density produced by precipitation and t' is a time parameter.

L. KIoc et al. / Materials Science and Eng#zeering A216 (1996) 16I-i68

166

( b\

a

b

C

. ,q....

Fig. 7. Microstructures after creep testing at 603 K: (a) to failure at 40 MPa after 0.44 h, (b) at 30 MPa after interruption of the test at 5.75 h and (c) to failure at 25 MPa after ~ 303 h.

In order to fit Eq. (1) to the experimental data, it is first necessary to calculate the value of ANy after an infinite time by considering the total amount of alloying addition available for precipitation in the form of A12CuMg and A12Cu. This calculation is outlined in the Appendix and it leads to AN~ ~ 1.6 x 102o m -3. Then, using this value of AN,. and Eq. (1), a best fit to the experimental data gives Nvo-~ 7.8 x 1018 m -3 and t ' ~1.1 x 10 .6 s. The experimental datum points and the predicted curve are plotted on logarithmic coordinates in Fig. 10, and it is apparent that, despite significant scatter in the experimental data at short testing times, the agreement with the curve is generally good. The upper and lower horizontal broken lines in Fig. 10 represent the values calculated for AN,, and measured experimentally for Nv0, respectively. It is therefore concluded that Eq. (1) gives a very useful representation of the precipitation data because it provides .the possibility of estimating the density of fine particles in the 2024PM alloy under any selected creep conditions.

5. Summary and conclusions

(1) Creep tests were conducted at elevated temperatures on a 2024PM (powder metallurgy) A1 alloy. The creep curves exhibit a minimum creep rate and a subsequent extensive tertiary stage prior to failure. The apparent stress exponent is high and variable, suggesting the presence of a threshold stress. (2) Microstructural observations using TEM after creep testing showed that there was a continuous precipitation of fine particles during the tests. (3) Quantitative measurements reveal that the density of fine particles precipitated during creep depends upon the testing conditions. However, the average size of the particles is of the order of ,,-60 nm under all conditions. (4) The evolution of fine particles may be adequately described using the concept of a temperature compensated time, with the temperature dependence

167

L. Kloc et al. / Materials Science and Eng#zeering A216 (1996) 161-168 .4

2024PM T=603K e=40MPa .3

z•<.2 z

Table 2 Density of fine particles, N, as a function of test conditions T (K)

cr (MPa)

t (s)

Nv (× 1018m - 3 )

548 548 573 573 573 603 603 603

30 50 30 40 50 25 30 40

7.27 x 106a 4.03 x 104 3.38 x 105

27 _ 2.6 11 + 0.6 7.7 + 2.2 13 ± 5.5 10 _+3.7 67 + 10 13 + 3.2 4.1 _+1.8

3 . 1 6 x 10 4

5.33 x I03 1.09 x 10s 2.07 × 10 4a 1.58× 103

a Interrupted test.

0

40

a

80 120 d [nm]

160

.4 2024PM T=603K e=30MPa

related to the activation energy for lattice self-diffusion in aluminum. (5) A relationship is derived from the experimental measurements to describe the precipitation process, and this relationship permits an estimate of the density of fine particles under any selected experimental conditions.

.3

Acknowledgements ---<.2 z

We thank Dr Jerzy Stobrawa for assistance with the transmission electron microscopy.

.t

Appendix A: Procedure for estimating AN,, at t =

0

40

b.

80 ~.20 160 d [nm]

.4 2024PM T=603K .3

z

"-<.2 Z

.1

0 0 C

40

80 120 d [nm]

t60

Fig. 8. Size distributions of fine particles (< i80 nm) after testing at 603 K: (a) at 40 MPa; (b) at 30 MPa; and (c) at 25 MPa.

Inspection of Table 1 shows the presence, in wt.%, of 0.24 Si, 0.26 Fe, 1.47 Mg, 0.56 Mn and 0.13 A1203. It is anticipated that the material will contain large undissolved particles of (Mn,Fe)3SiAlI2. Since 0.102 and 0.0466 tool of Mn and Fe are present initially in 1000 g of the alloy, it is reasonable to conclude that a total of (0.102 + 0.0466)/3 ~-0.05 mol is present as intermetallic particles. It is assumed that, after an infinite time, all of the Mg is precipitated in the form of AlzCuMg particles with the remaining Cu precipitated as A12Cu. In practice, this precipitation occurs partly during processing of the alloy in the form of large particles and partly during heat treatment and creep testing as fine particles. With all of the Mg precipitated, and based on the chemical composition in Table 1, it follows that 1000 g of material will contain 0.605 mol of A12CuMg, 0.091 tool of AlzCu and 0.0128 mol of ALO3. Using the relevant atomic weights and specific volumes, it can be shown that 1000 g of the alloy contains, in the equilibrium condition, a volume fraction of 0.09 of precipitated phases, made up of 0.013 Mn3SiAI12, 0.006 Fe3SiAI~2, 0.059 A12CuMg, 0.007 A12Cu and 0.005 A1203, with the remaining volume (0.91) as A1.

I68

L. Kloe et al. / Materials Science and Engineering A216 (I996) I61-168

i02o

1021

2024P. M_M

102o /~'A

z"

2024PM T [K]

t

• 548 • 573

[] 603 •10 ~.8

-.1

r

,

I

1

i

0

.1

.2

.3

.4

calculated at equilibrium/I//

E 1019 ' - - ~

!

'nn

1018

T

1017 10-10

Fig. 9. Density of fine particles versus temperature compensated time, using an activation energy for lattice self-diffusion in aluminum.

Assuming, as a reasonable approximation, that the large precipitates are spheres with a diameter of 450 nm regardless of their composition, and using the measured initial density of large particles (1.5 × 1018 m - g ) , their volume fraction is estimated as -0.072, Therefore, at equilibrium after an infinite time, the volume fraction of fine particles is 0.09 - 0.072 ~- 0.018. If these fine particles are spheres with diameters of 60 nm, their number per unit volume is therefore estimated as ,-1.6 x 1020 m -3. This value of AN~ may be used to fit Eq. (1) to the data in Fig. 9.

References [1] T.G. Nieh, Metall. Trans., ILA (1984) 139. [2] T. Morimoto, T. Yamaoka, H. Lilholt and M. Taya, J. Eng. Mater. Tech., 110 (1988)70. [3] T.G. Nieh, K. Xia and T.G. Langdon, J. E1~g. Mater. Tech., II0 (1988) 77. [4] K,-T. Park, E.J. Lavernia and F.A. Mohamed, Acta Metall. Mater., 38 (1990) 2149. [5] A.B. Pandey, R.S. Mishra and Y.R. Mahajan, Scr. Memll. Mater., 24 (1990) 1565. [6] A.B. Pandey, R.S. Mishra and Y.R. Mahajan, Acta Metall. Mater., 40 (1992) 2045. [7] G. Gonz~ilez-Doncel and O.D. Sherby, Acta Metall. Maier., 4.1 (1993) 2797. [8] K.-T. Park and F.A. Mohamed, 3Ietall. Mater. Trans., 26A (1995) 3119. [9] P.E. Krajewski, J.E. Allison and J.W. Jones, MetalL Trans., 24,4 (1993) 2731.

" ~ N v o

[exp,]

[K]

• 548 • 573 • 603

.5

Temperaturecompensatedtime [10"s]

1

ANv

,

, ,i,,,,I

,

10-9

, ,i,,,d

,

10-8

,

,I,,,,I

10-7

,

, ,Iiill

10-6

e Is] Fig. 10. Variation of density of fine particles with temperature compensated time in a logarithmic plot, showing the upper limiting value calculated for/,N~ in Eq. (I) and the initial experimental value for N~o.

[10] P.E. Krajewski, J.E. Allison and J.W. Jones, in K. Upadhya (ed.), Processing, Fabrication, and Applications of Adt, al~ced Composites, American Society for Metals, Metals Park, OH, I993, p. 15I. [11] K.-T. Park, E.J. Lavernia and F.A. Mohamed, Acta Metall. Mater., 42 (I994) 667. [12] D. Webster, Metall. Trans., I3A (1982) 151i, [13] H.Y. Kim and S.H. Hong, Scr. Metall. Mater,, 30 (1994) 297. [14] W. Blum and B. Reppich, in B, Wilshire and R.W, Evans (eds.), Creep Behaviour of Co'stalline Solids, Pineridge Press, Swansea, 1985, p. 83. [15] I.J. Polmear, Light Alloys: Metallurgy of the Light Metals, American Society for Metals, Metals Park, OH, 1982, pp. 2627. [16] 1. Harper and LE. Dorn, Acta Metall., 5 (1957) 654. [17] P. Yavari, F.A. Mohamed and T.G. Langdon, :Ida MetaIl., 29 (1981) 1495. [18] E. Orowan, in M. Cohen (ed.), Dislocations in Metals, AIME, New York, 1954, p. 131. [I9] U.F. Kooks, Philos. Mag., 13 (1966) 541. [20] R.S.W. Shewfelt and L.M. Brown, Philos. Mag., 35 (1977) 945. [21] E. Arzt and M.F. Ashby, Scr. Metall., I6 (1982) 1285. [22] E. Arzt and D.S. Wilkinson, Acta Metall., 34 (1986) 1893. [23] E. Arzt and J. R6sler, Acta Melall., 36 (1988) 1053. [24] J. R6sler and E. Arzt, Acta Metall. Mater., 38 (1990) 671. [25] F.A. Mohamed and T.G. Langdon, Metall. Trans., 5 (1974) 2339. [26] S.J. Rothman, N.L. Peterson, L.J. Nowicki and L.C. Robinson, Pays, Stat. Sol. (b), 63 (1974) K29.