Correlation between matrix residual stress and composite yield strength in PM 6061Al–15 vol% SiCw

5
Correlation between matrix residual stress and composite yield strength in PM 6061Al–15 vol% SiC w Pedro Ferna ´ndez a , Ricardo Ferna ´ndez b,1 , Gaspar Gonza ´lez-Doncel b , Giovanni Bruno a, * a Institut Laue-Langevin, ILL, Diffraction Group, 6, Rue Jules Horowitz, BP156, F-38042 Grenoble Cedex 9, France b Department of Physical Metallurgy, Centro Nacional de Investigaciones Metalu ´ rgicas (CENIM), CSIC, Av. de Gregorio del Amo 8, E-28040 Madrid, Spain Received 30 July 2004; received in revised form 1 December 2004; accepted 2 December 2004 Available online 25 December 2004 Abstract Upon relieving residual stress (RS) by means of isothermal annealing, it is observed that the RS and the yield strength (YS) of a 6061Al alloy follow a linear relationship. In contrast, two regimes are observed for a SiC-reinforced composite: firstly the YS decreases at constant RS, then the RS relaxes at constant YS. Ó 2004 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Yield strength; Discontinuously reinforced metal matrix composite (DRMMC); Residual stress; Stress relief 1. Introduction Discontinuously reinforced metal matrix composites (DRMMCs), such as Al alloys reinforced by SiC, are being increasingly used in structural applications be- cause of their enhanced mechanical properties with re- spect to the corresponding unreinforced matrix. These properties are strongly dependent on the microstructure and highly influenced by the presence of residual stress (RS) [1–3]. The RS state of DRMMCs is more compli- cated than that of the unreinforced alloys because of the presence of both macroscopic-RS (derived from con- ventional metallurgical procedures, such as welding and quenching) and microscopic-RS; see Ref. [4] for their definition. The microscopic-RS arises from three main sources [5,6]: (i) the thermal mismatch, (ii) the elastic mismatch between matrix and ceramic reinforcement, (iii) the plastic deformation of the matrix. In the absence of plastic pre-straining and macro-RS, the average microscopic (thermal) RS is compressive in the rein- forcement and tensile in the matrix. Knowledge of the RS state in industrial components is important since it may have a great influence on their behaviour in service. Moreover, the component fabrica- tion implies the application of thermo-mechanical pro- cesses (extrusion, rolling, etc.) and treatments (surface peening, pre-straining, heat treatments, etc.). These are able to affect both the mechanical properties and the final RS state of alloys and DRMMCs [7,8]. It is known, for example, that when a composite mate- rial undergoes plasticity or is heat-treated, macro- and micro-RS relaxation occurs [6,9,10]. In aluminium matrix composites, the problem is even more complex, because the matrix alloys are age hardenable, i.e. the specific precipitation state determines the final 1359-6462/$ - see front matter Ó 2004 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.scriptamat.2004.12.007 * Corresponding author. Tel.: +33 476 207 812; fax: +33 476 207 648. E-mail address: [email protected] (G. Bruno). 1 Present address: INDO, S.A. Thin Film Research, R&D Depart- ment, Sta Eulalia 181 08902, LÕHospitalet de Llobregat (Barcelona), Spain. www.actamat-journals.com Scripta Materialia 52 (2005) 793–797

Transcript of Correlation between matrix residual stress and composite yield strength in PM 6061Al–15 vol% SiCw

Page 1: Correlation between matrix residual stress and composite yield strength in PM 6061Al–15 vol% SiCw

www.actamat-journals.com

Scripta Materialia 52 (2005) 793–797

Correlation between matrix residual stress and compositeyield strength in PM 6061Al–15 vol% SiCw

Pedro Fernandez a, Ricardo Fernandez b,1, Gaspar Gonzalez-Doncel b,Giovanni Bruno a,*

a Institut Laue-Langevin, ILL, Diffraction Group, 6, Rue Jules Horowitz, BP156, F-38042 Grenoble Cedex 9, Franceb Department of Physical Metallurgy, Centro Nacional de Investigaciones Metalurgicas (CENIM), CSIC, Av. de Gregorio del Amo 8,

E-28040 Madrid, Spain

Received 30 July 2004; received in revised form 1 December 2004; accepted 2 December 2004

Available online 25 December 2004

Abstract

Upon relieving residual stress (RS) by means of isothermal annealing, it is observed that the RS and the yield strength (YS) of a

6061Al alloy follow a linear relationship. In contrast, two regimes are observed for a SiC-reinforced composite: firstly the YS

decreases at constant RS, then the RS relaxes at constant YS.

� 2004 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Yield strength; Discontinuously reinforced metal matrix composite (DRMMC); Residual stress; Stress relief

1. Introduction

Discontinuously reinforced metal matrix composites

(DRMMCs), such as Al alloys reinforced by SiC, arebeing increasingly used in structural applications be-

cause of their enhanced mechanical properties with re-

spect to the corresponding unreinforced matrix. These

properties are strongly dependent on the microstructure

and highly influenced by the presence of residual stress

(RS) [1–3]. The RS state of DRMMCs is more compli-

cated than that of the unreinforced alloys because of

the presence of both macroscopic-RS (derived from con-ventional metallurgical procedures, such as welding and

quenching) and microscopic-RS; see Ref. [4] for their

1359-6462/$ - see front matter � 2004 Acta Materialia Inc. Published by El

doi:10.1016/j.scriptamat.2004.12.007

* Corresponding author. Tel.: +33 476 207 812; fax: +33 476 207

648.

E-mail address: [email protected] (G. Bruno).1 Present address: INDO, S.A. Thin Film Research, R&D Depart-

ment, Sta Eulalia 181 08902, L�Hospitalet de Llobregat (Barcelona),

Spain.

definition. The microscopic-RS arises from three main

sources [5,6]: (i) the thermal mismatch, (ii) the elastic

mismatch between matrix and ceramic reinforcement,

(iii) the plastic deformation of the matrix. In the absenceof plastic pre-straining and macro-RS, the average

microscopic (thermal) RS is compressive in the rein-

forcement and tensile in the matrix.

Knowledge of the RS state in industrial components

is important since it may have a great influence on their

behaviour in service. Moreover, the component fabrica-

tion implies the application of thermo-mechanical pro-

cesses (extrusion, rolling, etc.) and treatments (surfacepeening, pre-straining, heat treatments, etc.). These

are able to affect both the mechanical properties and

the final RS state of alloys and DRMMCs [7,8]. It is

known, for example, that when a composite mate-

rial undergoes plasticity or is heat-treated, macro- and

micro-RS relaxation occurs [6,9,10]. In aluminium

matrix composites, the problem is even more complex,

because the matrix alloys are age hardenable, i.e.the specific precipitation state determines the final

sevier Ltd. All rights reserved.

Page 2: Correlation between matrix residual stress and composite yield strength in PM 6061Al–15 vol% SiCw

794 P. Fernandez et al. / Scripta Materialia 52 (2005) 793–797

mechanical properties such as their yield strength. The

annealing processes designed to relax macro-stresses

in MMC components are frequently not optimized in

time and temperature. Ideally, one would apply stress-

relieving (heat) treatments, which do not undermine

the material performance. In reality, those treatmentsoften cause a reduction in the yield strength or deterio-

rate other mechanical properties. Understanding the

link between the mechanical properties and the RS

in aluminium matrix composites is therefore of great

scientific and technological importance.

The aim of this work was to correlate the relaxation

of the matrix RS in a 6061Al–15 vol% SiCw composite

and in its unreinforced 6061Al alloy with the variationof their yield strength (YS), and observe the conditions

under which the first is possible while avoiding the

second. Both variations were induced by means of heat

treatments, bringing the materials from a T6 (fully

hardened) condition, through several isothermal treat-

ments at 300 �C, to an over-aged condition, OA (100 h

treatment).

Fig. 1. Stress–strain curves for the composite (E219) and the unrein-

forced alloy (E220). Only the plastic regime is represented.

2. Experimental details

The materials studied, a 6061Al–15 vol%SiCw com-

posite and its 6061Al alloy matrix, were prepared by a

powder metallurgical (PM) route [9,11,12]. The unrein-

forced matrix and the composite were labelled E220

and E219, respectively. Both compacted blends were ex-truded at about 500 �C. The extrusion ratio was 27:1,

which implies 3.30 true strain. This severe size reduction

led to a highly textured matrix material (h111i and

h100i fibre texture components) and to some trend of

the SiCw, of about 2 lm average length and with an

average aspect ratio of about 4, to be aligned with the

extrusion axis.

The stress-relieving heat treatments were also tailoredto obtain different mechanical properties at each treat-

ment stage. The composite and the unreinforced alloy

were first brought to the fully hardened (T6) condition,

and then aged at 300 �C for different treatment times, up

to an over-annealed (OA) condition.

The T6 treatment consisted of a solution treatment at

520 �C for 90 min followed by water quenching plus

annealing at 146 �C for 16 h. The reference alloy hadto be annealed for 56 h to achieve a similar precipitation

state as the composite matrix. The longer annealing time

in the reinforced material is due to the accelerated age-

ing phenomenon [13].

Intermediate heat treatments consisted of holding the

samples at 300 �C for 0.5, 1, 2, 5 and 9 h. The OA con-

dition was achieved by annealing at 300 �C for 100 h

and furnace cooling. It has been already reported[9,14], that the macro-RS in the OA condition is essen-

tially relaxed (although not completely in the compo-

site). In the OA condition, the unreinforced alloy was

annealed for the same durations as the composite since

at this temperature the accelerated ageing phenomenon

is minimized.

Compression tests were carried out in a computer

controlled SERVOSIS (class 1) testing machine at astrain rate of 10�4 s�1. Cylindrical samples (13 mm high

with 6.5 mm diameter) were used for both compressive

tests and neutron diffraction (ND).

The RS in all precipitation states, from T6 to OA,

was studied by neutron diffraction, ND. The ND experi-

ments were carried out on the diffractometer D1A at the

ILL, Grenoble, France, and the calculation procedure to

evaluate stresses from diffraction data are fully reportedelsewhere, see Refs. [2,9,14] for further details. Here we

mention that the 311 peaks of both phases were

exploited. They are elastically (and plastically for the

Al phase) isotropic and relatively texture-insensitive

[15]. All measurements were carried out at the centre

of the samples, using a relatively large gauge volume

of 3 · 3 · 1 mm2.

3. Results and analysis

Fig. 1 shows two examples (after 0 and 0.5 h anneal-

ing at 300 �C following the T6 heat treatment) of stress–

strain curves for composite E219 and the alloy E220.

The variation of the YS is clearly visible for both mate-

rials. The YS of the composite and the unreinforcedmatrix (evaluated at 0.2% strain, see Ref. [2]) decreases

exponentially with annealing time according to the

expression:

r ¼ r1 þ r0e�t

s ð1Þ

Page 3: Correlation between matrix residual stress and composite yield strength in PM 6061Al–15 vol% SiCw

Fig. 2. Variation of the yield strength vs. annealing time of the

composite (E219) and the unreinforced alloy (E220). The asymptotic

value and the total variation are shown.

Fig. 3. The variation of the hydrostatic residual stress vs. annealing

time. The total (or equivalently, the macro) stress for the unreinforced

alloy (E220) and the total, macro-, and micro-RS for the composite

matrix (E219) are shown.

P. Fernandez et al. / Scripta Materialia 52 (2005) 793–797 795

where s is the relaxation time, r1 is the asymptotic valueand r0 is the total range of stress variation. Fig. 2 showsthe decay of the YS as a function of the treatment time.

Table 1 lists the values of s, r1 and r0 for both materi-

als. The strength of the composite is always higher than

that of the unreinforced matrix. As can also be seen,

although the rate of decrease is the very similar in both

materials (i.e., sE219YS sE220YS ), the magnitude of the YS

drop (r0) is significantly larger in the unreinforced alloythan in the composite.

A similar analysis was carried out on the RS. The rule

of mixtures (ROM) was applied to separate the macro

(index M) and micro (index m) stress from the total

(phase specific, index T) stress, according to [4,16]:

rTAl=SiC ¼ rM þ rm

Al=SiC

rM ¼ ð1� f ÞrTAl þ frT

SiC

(ð2Þ

The hydrostatic component of stress rH was calcu-

lated as the average between the axial, radial and hoop

components. Since ND measurements were carried out

at the centre of each sample, we could assume rrad =rhoop and get

rH ¼ rax þ 2rrad

3ð3Þ

The phase-specific RS, as calculated from the ND

strain measurements, are fully reported in Ref. [14].

Table 1

Values of the fitting parameters obtained by using Eq. (1) for yield

strength and (matrix) residual stresses in both materials

Material r0 (MPa) r1 (MPa) s (min)

E220 YS 262 ± 11 90 ± 5 18 ± 2

RS 167 ± 10 19 ± 5 25 ± 7

E219 YS 228 ± 9 185 ± 4 14 ± 2

Total RS 82 ± 9 105 ± 5 66 ± 18

Macro-RS 52 ± 10 42 ± 6 63 ± 30

Micro-RS 18 ± 2 70 ± 1 40 ± 15

The evolution of the hydrostatic matrix RS with ex situ

annealing at 300 �C is shown in Fig. 3 for both materi-

als. Both macro and micro-hydrostatic RS relax as afunction of the ageing time in both the E219 composite

matrix and the E220 unreinforced alloy. Their beha-

viour can be again described through an exponential de-

cay, Eq. (1). The corresponding fit parameters are listed

in Table 1. Most relevantly, Fig. 3 shows that the RS of

the composite and the unreinforced alloy are radically

different after the OA treatment. The RS in the compos-

ite relaxes towards a stable value (r1 105 MPa of thetotal-RS), whereas in the unreinforced alloy the RS vir-

tually disappears (r1 20 MPa), Table 1. For both

matrix and reinforcement, the axial deviatoric compo-

nent of the micro-RS does not relax (see Ref. [14] for

a detailed discussion), while the deviatoric macro-RS re-

laxes very rapidly. For the sake of clarity, they will not

be discussed in this work and in the following implicit

reference to matrix RS will be made. We note thatmicro- and macro-asymptotic RS do sum to the total

stress, within the error of the fit.

4. Discussion

The YS of the composite is always higher than that of

the unreinforced alloy due to the strengthening effect ofthe whisker, Figs. 1 and 2. This effect is based on

the higher density of dislocations in composites (the geo-

metrically necessary dislocations, GNDs, are always

pinned at the particle/matrix interface) and on the load

transfer phenomenon [6,11].

The annealing treatment activates diffusion processes

in the matrix, by which the precipitation state changes,

leading to changes in the mechanical properties of the6061Al alloy. In the 6061Al alloy, the precipitation se-

quence is b00 ! b 0 ! b (Mg2Si), of which only the latter

is stable. The precipitates grow, reducing the number of

Page 4: Correlation between matrix residual stress and composite yield strength in PM 6061Al–15 vol% SiCw

796 P. Fernandez et al. / Scripta Materialia 52 (2005) 793–797

particles, but increasing their average size. In this situa-

tion the dislocations can move more easily making plas-

tic flow easier and consequently reduce the YS. In fact,

the Orowan�s mechanism [17,18] is progressively less

effective while particles coalesce. In the composite there

is an additional effect of the reinforcement, which doesnot change with the heat treatment: the whiskers induce

a higher dislocation density. Therefore, the movement of

dislocation is hindered. The presence of the whiskers

does not influence the b precipitation kinetics [13] at

300 �C. This has two consequences: (i) the decrease of

the matrix and the composite YS proceeds at the same

rate, and (ii) the amount of stress decrease is larger for

the alloy, i.e. rE2200 > rE219

0 , Table 1.The (total or macro) hydrostatic RS (Fig. 3) of the

unreinforced alloy relaxes almost to zero in a very short

time, whereas that in the composite is only partially re-

duced in a longer time. This behaviour is very different

from that of the YS relaxation.

By separating the total stress of the composite in

macro and micro-stresses, according to the ROM (Eq.

(2)), we observe an exponential decay also of themicro-RS, but with a very small total variation. The

GNDs created during the cooling process cannot be

annihilated by any annealing treatment [14]. The macro-

RS shows a variation almost parallel to the total stress

(see Eq. (2)). As in the case of the YS, the drop of the

macro-RS in the composite is much smaller than that

in the unreinforced matrix. This is related to the more

stable dislocation structure of the composite and to itssmaller thermal expansion coefficient (CTE), caused by

the presence of the ceramic reinforcement. It is well

established that the RS increases with the CTE. In par-

allel to this, also the RS relaxation appears to increase

with the CTE.

In Fig. 4 the total (rT), macro-(rM) and micro-(rm)RS resulting after each treatment are plotted as a func-

Fig. 4. The correlation between residual stress and yield strength. The

total hydrostatic stress for the unreinforced alloy (E220), the total,

macro- and micro-stress for the composite matrix (E219) are repre-

sented vs. the respective yield strength.

tion of the corresponding YS. For the unreinforced

alloy there is a linear dependence between macro-RS

and YS. This is to be expected, because for E220 they

relax at a similar rate, see Table 1.

The RS vs. YS curves of E219 show, however, two

well-defined regions, In fact, they relax at very differentrates. The YS relaxes faster than the RS (Table 1). Ini-

tially, with short heat-treatment times, a variation of

the strength takes place, independent from the residual

stress. While the RS is still large, the yield strength drops

rapidly to its minimum value. Successively, the RS de-

creases while the yield strength has already relaxed

and remains basically constant. The result is a logarith-

mic-type variation, very different from the linear beha-viour observed for E220.

The decrease of E220 YS with annealing at 300 �C is

only due to the growth and coalescence of precipitates in

the matrix. The more the precipitates grow, the lower

the strength and the less inhibited the dislocations

motion [19]. The RS relaxes to practically zero while

the yield stress decreases, as it cannot be retained by

the microstructure. The dislocation motion continuesto relax the RS and this is the only mechanism acting

[17]. Interestingly, the straight line extrapolates to zero

RS for a finite value of YS (about 50 MPa). This could

possibly correspond to a very solute-poor alloy, tending

to pure aluminium.

In E219 the YS is also influenced by the presence of

the reinforcement, via the load transfer mechanism

and the higher dislocation density with respect toE220. This is why the range of YS variation (r0) is largerin E220: the GNDs at the particle/matrix interface form

back after each cooling process. This implies that while

in the alloy one can arbitrarily annihilate dislocations by

means of suitable heat treatments, in the composite we

would always get a minimum amount of them. This

means that the difference between the dislocation densi-

ties in E219 and E220, Dq, is such that

DqT6 < DqOA

The YS of E219, however, decreases approximately at

the same rate as in E220. This similarity can be attri-

buted to the fact that the precipitation kinetics at

300 �C is the same in the composite and in the alloy [13].

With short-time heat treatments, the effect of the pre-cipitates on the YS is immediate, while that of the whis-

kers is mainly to lock the RS via the GNDs. Therefore,

Fig. 4 shows that the slope of the RS vs. YS curve is

much smaller in E219 than in E220: the residual stress

relaxes more slowly than the yield strength. Then, after

relatively long treatment time, the slope increases and

the RS relaxes dramatically, while the YS is essentially

constant. This occurs until the RS reaches a minimum.In fact, there is an accumulation of points at the end

of each curve in E219. As mentioned before, a minimum

value of the RS is present in E219, because a certain

Page 5: Correlation between matrix residual stress and composite yield strength in PM 6061Al–15 vol% SiCw

P. Fernandez et al. / Scripta Materialia 52 (2005) 793–797 797

amount of dislocations are still locked and cannot move

and annihilate: they are the GNDs [2,14].

5. Conclusions

The relationship between the matrix residual stress

and the yield strength in PM 6061Al–15 vol%SiCw has

been studied. In order to achieve both stress relief and

microstructural changes, samples were treated to a fully

hardened (T6), to a severe over annealed (OA, at

300 �C) condition, and at different annealing times.

The most relevant conclusions are:

1. The presence of the reinforcement leads to higher YS

in the composite matrix than in the unreinforced

alloy, after every annealing treatment at 300 �C.2. The YS relaxation time (sYS) is very similar in the

unreinforced alloy and in the composite. This implies

that relaxation is mainly driven by the solid solution

precipitation kinetics in the matrix. The small differ-

ence between the relaxation times can be attributedto the more rapid recovery in the alloy than in the

composite.

3. The YS relaxation (r0) is more pronounced in the

unreinforced alloy than in the composite. Again, the

faster recovery (dislocation annihilation and redistri-

bution) in the alloy than in the composite is responsi-

ble for this.

4. The residual stress in the unreinforced alloy relaxesalmost completely with annealing. The RS and the

YS relaxations occur at a similar rate: a linear depen-

dence is found between them in the whole range of

stress variation. This implies that the precipitation

sequence (b00 ! b 0 ! b) is the rate-controlling processof RS relaxation in the 6061Al alloy. Upon anneal-

ing, the precipitates coalesce and the dislocations

are able to move more freely, relaxing the residualstress but decreasing the YS proportionally.

5. In the composite, the relaxation times of the YS and

the matrix RS are clearly different. In fact, both the

precipitation and the reinforcement determine the

value of the YS. Upon annealing, the hindrance to

dislocation motion represented by the precipitates

(Orowan�s mechanism) decreases drastically, but thecontribution of the reinforcement related mechanisms(the GNDs) remains. Schematically, there is a dou-

ble-linear dependence between YS and matrix RS.

Firstly, the yield strength decreases steeply while the

RS is still locked by the GNDs, then, when the YS

has decreased to its minimum, the RS relaxes, accu-

mulating at a minimum. This implies that it is not

possible to relieve the RS without substantially

decreasing the material performance, at least by

means of heat treatments at 300 �C.6. The macro-RS relaxes less and more slowly in the

composite than in the unreinforced alloy. This is

because the dislocations motion in the matrix is moreinhibited in the former. Moreover, part of the com-

posite macro-RS does not relax.

Acknowledgements

Financial support from the Spanish government is

acknowledged (MAT01-2085 of the MCYT, Spain).

The ILL, Grenoble, France is acknowledged for the

neutron beamtime on D1A.

References

[1] Clyne TW, Withers PJ. An introduction to metal matrix

composites. Cambridge University Press; 1993.

[2] Fernandez R, Bruno G, Gonzalez-Doncel G. Acta Mater

2004;52:5471–83.

[3] Prangnell PB, Stobbs WM, Withers PJ. Mater Sci Eng

1992;A159:51–63.

[4] Pintschovius L. In: Hutchings MT, Krawitz AD, editors. Mea-

surement of residual and applied stress using neutron diffraction

NATO Advanced Research Workshop. Dordrecht: Kluwer Aca-

demics; 1992. p. 115–30.

[5] Fitzpatrick ME, Hutchings MT, Withers PJ. Acta Mater

1997;45:4867–76.

[6] Fitzpatrick ME, Withers PJ, Baczmanski A, Hutchings MT, Levy

R, Ceretti M, et al. Acta Mater 2002;50:1031–40.

[7] Muller K, Bergmann JP, Bergmann HW. Z Metallkd

2001;92:253–60.

[8] Levy-Tubiana R, Baczmanski A, Lodini A. Mater Sci Eng A

2003;341:74–86.

[9] Fernandez R, Bruno G, Lin Peng R, Gonzalez-Doncel G. Mater

Sci Forum 2003;426–432:2193–8.

[10] Bruno G, Ceretti M, Girardin E, Giuliani A, Manescu A. Scr

Mater 2004;51:999–1004.

[11] Borrego A, Fernandez R, del Carmen Cristina Ma, Ibanez J,

Gonzalez-Doncel G. Comp Sci Tech 2002;62:731–42.

[12] Borrego A, Ibanez J, Lopez V, Lieblich M, Gonzalez-Doncel G.

Scr Mater 1996;34:471–6.

[13] Borrego A, Gonzalez-Doncel G. Mater Sci Eng 1998;A245:10–8.

[14] Bruno G, Fernandez R, Gonzalez-Doncel G. Mater Sci Eng

2004;A382:188–97.

[15] VAMAS TWA 20 standard, ISO/TTA 3, 2001.

[16] Winholtz RA. In: Hutchings MT, Krawitz AD, editors. Mea-

surement of residual and applied stress using neutron diffraction

NATO Advanced Research Workshop. Dordrecht: Kluwer Aca-

demics; 1992. p. 131–46.

[17] Cox HL. Brit J Appl Phys 1952;3:72.

[18] Haasen P. In: Cahn RW, Haasen P, editors. Physical metal-

lurgy. North-Holland; 1983. p. 1373–9.

[19] Kim H-J, Niinomi M. Mater Sci Eng 2000;A284:14–24.