Your SlideShare is downloading. ×
!!Comparison of superplastic behavior in two 5083 aluminum alloys
Upcoming SlideShare
Loading in...5
×

Thanks for flagging this SlideShare!

Oops! An error has occurred.

×

Introducing the official SlideShare app

Stunning, full-screen experience for iPhone and Android

Text the download link to your phone

Standard text messaging rates apply

!!Comparison of superplastic behavior in two 5083 aluminum alloys

695
views

Published on

Published in: Business, Technology

0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total Views
695
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
14
Comments
0
Likes
0
Embeds 0
No embeds

Report content
Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
No notes for slide

Transcript

  • 1. Materials Science and Engineering A351 (2003) 228 Á/236 www.elsevier.com/locate/msea Comparison of superplastic behavior in two 5083 aluminum alloys R.M. Cleveland a, A.K. Ghosh a,*, J.R. Bradley b a Department of Materials Science and Engineering, The University of Michigan, 2300 Hayward Street, Ann Arbor, MI 48109-2136, USA b General Motors R&D and Planning, Warren, MI 48090-9055, USA Received 8 May 2002; received in revised form 28 October 2002Abstract Superplastic elongation is generally controlled by both strain-rate sensitivity and particles causing cavitation. The superplastictensile behavior of 5083 Al (nominally Al Á/4.6% Mg) alloys from two different sources has been examined. It is found that for nearlythe same grain size, one of the alloys (alloy A) has a slightly lower strain-rate sensitivity m . This alloy also contains a significantlylarger number of hard particles and inclusions. Because this alloy exhibits somewhat lower flow stress during constant strain-ratetests, a reduced tendency for cavity initiation was expected. However, the larger density of particles tends to increase cavitation byproviding more nucleation sites. Then the lower m value leads to more rapid cavity growth and interlinkage. Thus, while thepresence of fine grain size and high-angle grain boundaries are important for superplastic flow, this research shows that in similarlyprocessed alloys small changes in m values and variations in alloy chemistry that generate cavity-causing particles can have amedium to large effect on superplastic elongation.# 2002 Elsevier Science B.V. All rights reserved.Keywords: Cavity; Superplastic; Strain-rate sensitivity1. Introduction Superplastic flow is generally terminated by internal cavitation at pre-existing particles and by the growth Currently, there is a strong interest in superplastic- and the interlinkage of such cavities [5 Á/9]. Cavities tendforming technology for the fabrication of automotive to nucleate on second-phase particles or other high-sheet parts from 5000-series aluminum (i.e., Al Á/Mg energy boundaries. It is well known that damage canalloys). It is well known that fine grain size and the occur at the matrix Á/particle interface during cold rollingpresence of high-angle grain boundaries are important in alloys with hard particles. Recent evidence [10]requirements for good superplastic behavior. The degree suggests that such damage at the particle interface mayof superplasticity in engineering alloys is influenced by not be entirely eliminated by the recrystallizationgrain size and grain size distribution because of the role process and therefore could be a source of nucleationof grain boundary sliding in the deformation mechan- sites for cavities during subsequent superplastic forming.ism. Smaller grain size generally leads to lower flow Diffusion and plastic flow during tensile straining leadstresses, higher values of strain-rate sensitivity (m ), and to cavity growth and, as cavities become larger orgreater ductility [1 Á/3]. The distribution of grain sizes numerous enough to interlink, failure can occur. Gen-present in production alloys can also have an important erally, increasing strain rate is found to increase cavita-impact on the strain-rate range in which m is high (and tion in these alloys, while higher temperature is found to decrease cavitation. For example, cavitation was foundpeak ductility is achieved) [4]. Grain size is controlled in to be significantly reduced in a 5083 Al alloy by raisingthese alloys by adding dispersoid-forming elements to the temperature from 525 to 550 8C [11].pin grain boundaries. In addition to affecting the grain size and cavitation, variations in the alloy chemistry and microstructure of the selected alloy may also influence the achievable * Corresponding author. forming severity in parts. It is well known in industry E-mail address: akg@umich.edu (A.K. Ghosh). that the superplastic formability of 5083 Al sheet varies0921-5093/03/$ - see front matter # 2002 Elsevier Science B.V. All rights reserved.PII: S 0 9 2 1 - 5 0 9 3 ( 0 2 ) 0 0 8 4 8 - 1
  • 2. R.M. Cleveland et al. / Materials Science and Engineering A351 (2003) 228 Á/236 229considerably with source and processing. While many temperatures is often observed with increased levels oftrends in the superplastic behavior of 5083 alloys have Mg [12]. A higher level of Mg can also cause greaterbeen well documented, little attention in the literature strain hardening of the alloy during thermomechanicalhas been directed at the effect of small differences processing and is likely to produce internal damage.between sheets of nominally the same composition on While the intermetallic dispersoids formed by thethe superplastic formability of these alloys. addition of Fe and Mn are important for grain size In this study, two slightly different thermomechanical stability, a high level of Fe and Mn is known to produceprocessing methods were used to prepare samples from larger hard particles that can act as nucleation sites forone alloy chemistry and the behavior of this alloy was cavitation.compared with a commercially available 5083 Al sheet To check the potential for grain refinements of alloymaterial. These samples were examined to determine the A, two different thermomechanical treatments weregrain size and particulate size distributions in the alloys. examined. The first method (TMT-1) involved coldBoth these factors can influence superplastic ductility in rolling the plate (4.5 mm) down to a thin sheet (1.2definite ways [4,7]. The influence of these parameters on mm), a 73% reduction, with intermediate stress reliefthe superplastic mechanical behavior and fracture of the anneals of 15 min at 170 8C to prevent cracking. Thealloys was examined using constant strain-rate tensile sheet was then recrystallized at 500 8C for 30 min. Thetests. Such tests permit a comparison with the overall second method (TMT-2) was similar to the first exceptsuperplastic ductility. Additional tests were used to that the plate was given a solution heat treatmentdetermine the strain-rate sensitivity of flow stress and followed by an overaging step to precipitate Al3Mg2the progression of cavitation as a function of strain to particles. The particles are expected to generate a greaterdevelop an understanding of the mechanisms leading to amount of stored strain energy during the rolling step.failure. This in turn can lead to finer recrystallized grains. In TMT-2, the plate was solution heat treated at 500 8C for 30 min, quenched in water, and then aged at 230 8C for 20 h prior to rolling and final recrystallization.2. Experimental work 5083 Al alloys were procured from two different 2.1. Microstructuressources for investigation in this study. Alloy A wasproduced by ARCO aluminum and alloy B was The microstructures of alloy A in the as-received andproduced by SKY aluminum. Their compositions are TMT-2 conditions are shown in Fig. 1(a) and (b), andlisted in Table 1. Alloy A contains more Mg, Mn, and the average grain diameters of alloy A in the as-received,Fe than alloy B. Alloy A was received in 4.5 mm thick TMT-1, and TMT-2 conditions are listed in Table 2 (thehot-rolled plates with an elongated coarse grain struc- rolling direction of the sheet is horizontal and theture, whereas alloy B was received in cold-rolled and thickness direction is vertical). These samples were alsoannealed sheets (1.5 mm) with a fine grain structure. aged for 16 h at 120 8C, polished in colloidal silica, and More Mg in alloy A would not generally be regarded then etched in 10% phosphoric acid at 50 8C. The TMT-as detrimental to the magnitude of superplastic ductility 2 condition was found to produce the smallest grain sizeeven though more Al3Mg2 particles would be present in (5.6 mm) for alloy A, and was therefore chosen forthe alloy at room temperature. A higher amount of Mg tensile testing. The microstructure of alloy B obtained incan lead to higher strength in the alloy at room the sheet form is also shown in Fig. 1(c) for comparison.temperature. At superplastic temperatures, Mg-rich Alloy B was found to have an average grain diameter ofparticles are in solid solution and thus higher Mg 4.5 mm, which is slightly finer than alloy A in TMT-2content can produce enhanced diffusivity in the alloy condition. A histogram comparing the grain size dis-arising from the atomic size differences. Higher diffu- tributions of alloy A in the TMT-2 condition and alloysivity should in turn lead to lower flow stresses. In Al Á/ B is shown in Fig. 2. For each material, the areas ofMg alloys that exhibit solute pinning of dislocations at more than 500 grains were individually measured fromlow temperatures, lower flow stress at superplastic digital micrographs using NIH Image software and theTable 1Compositions of 5083 Al alloys by weight (remainder aluminum) Mg (%) Cr (%) Mn (%) Si (%) Fe (%) Cu (%) Zn (%) Ti (%) SourceA 4.75 0.09 0.85 0.12 0.26 0.08 0.032 0.013 ARCOB 4.36 0.086 0.68 0.13 0.06 0.011 0.001 0.001 SKY Remainder is aluminum.
  • 3. 230 R.M. Cleveland et al. / Materials Science and Engineering A351 (2003) 228 Á/236Fig. 1. Micrographs of the grain structures of alloys A and B. Alloy A is shown in the as-received condition (a) and after TMT-2 (b). The sheetstructure of alloy B is shown in (c).Table 2 similar except that alloy A is shifted about 1 mm towardGrain diameters of 5083 Al alloys (mm) the larger size compared with alloy B. As-polished micrographs of alloys A and B are shownA: as-received A: TMT-1 A: TMT-2 B in Fig. 3(a) and (b). A significant difference in the9.4 6.4 5.6 4.5 particulate size and volume fraction in the two alloys is evident. Clearly, alloy A has larger particles than alloy B. Such digital micrographs were analyzed using NIH Image to count and measure the particles. The resolu- tion of the micrographs allowed particles about 0.5 mm in diameter and larger to be counted. A minimum sample area of 0.04 mm2 was analyzed for each material. Measurements of particles from micrographs (Table 3) indicate that the volume fraction of particles in alloy B is much lower and that the average particle size is also smaller. Only particles of about 0.5 mm in diameter and larger could be detected by the optical method used. A histogram of the particle size distribution is shown in Fig. 4, where the relative frequency is determined by the number of particles in each diameter group divided by the total number of particles counted for each alloy. TheFig. 2. The grain size distribution in Al 5083 alloys A (in TMT-2 plot clearly shows that alloy A has larger particles thancondition) and B. Table 3 Measurements of particles in Al 5083 alloysdiameters calculated by assuming that the grains are pffiffiffiffiffiffiffiffiffiffiffi Alloy A Alloy Bspherical. The diameter is given by D 0 4A=p; where Area fraction of particles (%) 4.2 0.5A is the measured area in a two-dimensional section. Average particle diameter (mm) 2.9 1.3Relative frequency is the percent of the measured grains Standard deviation (mm) 2.3 0.9falling in each bin. The distributions are generally
  • 4. R.M. Cleveland et al. / Materials Science and Engineering A351 (2003) 228 Á/236 231 Fig. 5. SEM micrograph of particles in alloy A. The particles labeled ‘‘a’’ are found to be rich in iron and manganese. The particles labeled ‘‘b’’ are found to be rich in silicon and oxygen.Fig. 3. Micrographs showing the particulates in aluminum 5083 alloyA (a) and alloy B (b).alloy B. The compositions of the particles were exam-ined using X-ray energy-dispersive spectroscopy on aPhilips XL30 SEM. Fig. 5 is an SEM micrograph of Fig. 6. Photographs of samples after various test conditions.alloy A with labels indicating the two types of particlesFig. 4. The particle size distribution in Al 5083 alloys A and B. Therelative frequency is the number of particles in each diameter group Fig. 7. Stress Á/strain curves for both Al 5083 alloys tested at constantnormalized by the total number of particles counted for each alloy. strain rate at 500 and 550 8C.
  • 5. 232 R.M. Cleveland et al. / Materials Science and Engineering A351 (2003) 228 Á/236 Fig. 8. An example of a stress Á/strain curve used for the determination of m values.Fig. 9. Flow stress as a function of strain rate for alloy A in the TMT-2 condition (a, b) and alloy B (c, d).Plots (a, c) are for 500 8C and (b, d) are for550 8C. The flow stresses increase for higher strain levels.
  • 6. R.M. Cleveland et al. / Materials Science and Engineering A351 (2003) 228 Á/236 233found. The particles labeled ‘‘a’’ were found to be rich in tions were 270% at 550 8C and 310% at 550 8C. Theseiron and manganese, while those labeled ‘‘b’’ were found elongations represent a reasonable degree of super-to be rich in silicon and oxygen. plasticity. StressÁ/strain plots for these tensile tests are shown in2.2. Tensile tests Fig. 7. Alloy A exhibited lower flow stresses than alloy B at both temperatures*/a result unexpected on the basis Superplastic tensile samples were machined from of its slightly coarser grain size, but reasonable from thealloy A sheet in the TMT-2 condition and from the standpoint of higher diffusivity resulting from the higheralloy B sheet that was already in fine grain condition. Mg content. Alloy B failed at a higher elongation atThe tensile specimens had a gage width of 6.2 mm and 500 8C ( Â/270%), but both alloys failed at nearly thegage length of 12.7 mm. Samples of both Al 5083 alloys same final elongation at 550 8C (310 Á/340%). Micro-were tested at temperatures of 500 and 550 8C. The tests graphs of samples deformed at 500 8C showed that thewere conducted on a computer-controlled Instron 4505 average grain size increased to between 7 and 8 mm inwith a three-zone split furnace that wrapped around the alloy A (for strains of 0/0.5) and to between 9 and 10sample. Load was measured using a 1000 kN load cell mm in alloy B (for strains of 0/0.7), indicating similarand strain was calculated from cross-head displacement grain growth characteristics.and checked by final measurements of the samples. A To obtain a clear understanding of the above features,constant strain rate of 0.001 s (1 was maintained by the the strain-rate sensitivity of the alloy must be examined.method developed by Friedman and Ghosh [13]. Photo- The variation of flow stress as a function of strain rategraphs of the initial sample geometry and samples after and the corresponding strain-rate sensitivity of flowtesting are shown in Fig. 6. At this strain rate, the stress, m , were characterized using step strain-rate tests.average elongations for alloy A were 190% at 500 8C In these tests, a succession of strain-rate decrements andand 340% at 550 8C. For alloy B, the average elonga- increments were performed to determine steady-stateFig. 10. The effect of strain rate on strain-rate sensitivity index, m , for alloy A in TMT-2 condition (a, b) and alloy B (c, d). Plots (a, c) are for 500 8Cand (b, d) are for 550 8C.
  • 7. 234 R.M. Cleveland et al. / Materials Science and Engineering A351 (2003) 228 Á/236flow stress under different strain rates (and also over (m ), given by the slope of the log stress vs. log strain ratedifferent strain ranges). The scheme of these tests was curves from Fig. 9. The peak m values of the two alloysoptimized previously [2]. The stress Á/strain plot from in Fig. 10 are in the range 0.4 Á/0.65 for differentone such test is shown in Fig. 8. Similar tests were temperatures and generally increase with increasingperformed at 500 and 550 8C. The resulting values of test temperature. However, for alloy A, the peak valuesflow stress for each imposed strain rate are plotted in of m were about 0.47 at 500 8C and 0.54 at 550 8C andFig. 9(a) Á/(d) on a log Á/log scale for the two alloys. The the peak values of m for alloy B were consistentlygeneral shapes of these curves are sigmoidal as those higher, ranging from 0.52 at 500 8C to 0.64 at 550 8C.found in the previous studies [1 Á/3]. The flow stress is This difference in m values was repeatable and isnot simply a function of strain rate but is seen to consistent with the higher tensile elongation of alloy Bincrease somewhat at higher strain levels. This strain- at 500 8C. It should be noted that the grain sizeinduced hardening in superplastic materials has been distribution plot in Fig. 2 indicates that in the finerfound to be a result of concurrent grain growth in the grain size range (2 Á/4 mm), alloy B has considerablyalloy. Although the flow stress values for alloys A and B more grains present. As the model presented in [4]do not vary a great deal from each other, upon closer indicates, this is exactly the reason responsible for theexamination, alloy A is found to have somewhat lower slightly higher m value in alloy B. We believe that thisflow stress at higher strain rates, but higher flow stresses grain size distribution is a result of local variations inat lower strain rates compared with alloy B. stored cold work in alloy A with more Mg and more The above observation of the trend of flow stress particles that provide local areas of grain growth. Thevariation with strain rate is consistent with the results uniformity of composition in alloy B may result in ashown in Fig. 10, which shows strain-rate sensitivity more homogeneously distributed strain energy andFig. 11. Micrographs showing the progression of cavitation in the two alloys at 500 8C: alloy A (a, b) and alloy B (c, d). True strains at the positionon the sample, where each micrograph was taken, are indicated.
  • 8. R.M. Cleveland et al. / Materials Science and Engineering A351 (2003) 228 Á/236 235hence a finer grain sizes. Since m values are generally models of cavitation [14]. The volume fraction of cavitieshigher for smaller grain sizes [1,2], the decrease in m is nearly the same for the two alloys at low strains, asvalue with increasing strain observed in the plots of Fig. shown in Fig. 12(c). However, in alloy A, which has a10 is consistent with concurrent grain growth. At both larger number of more closely spaced cavities, the rapidthese temperatures, the m vs. strain rate curves are more growth in the cavities leading to cavity interlinkage andsharply peaked for alloy B than for alloy A. The broader failure begins at a lower strain.peaks in m for alloy A suggest that its formability is lesssensitive to strain rate, which may make it a morepractical alloy for production if the cavitation can be 3. Conclusionsminimized. The effect of small differences in microstructure and2.3. Cavitation study composition between sheets of nominally the same In order to further examine the differences betweenthe behavior of the two Al 5083 alloys, the tendency forthe formation of cavities was studied as a function ofstrain in the samples that were pulled to failure at500 8C and an overall strain rate of 0.001 s (1. Gridlines on the surface of the samples as well as width andthickness measurements were used to determine thestrain levels at several points along the sample length sothat cavitation could be measured at known strainlevels. Since alloy B failed at a larger strain, the selectedstrain levels were larger than for alloy A. Micrographsof the samples in the as-polished condition from theselected locations were examined. Several examples ofthe micrographs are shown in Fig. 11(a) Á/(d). In Fig.11(a) and (b), the cavities in alloy A are shown at truestrains of 0.51 and 1.02, respectively. Alloy B is shown attrue strains of 0.85 in (c) and 1.53 in (d). Alloy A is seenin the micrographs to have a larger number of cavities atthe depicted strain levels, but many of the cavities in thealloy A micrographs are relatively small. It is notsurprising that alloy A form more cavities since it hasmany more second-phase particles than alloy B onwhich cavities can nucleate. It is believed that the highfrequency of large particles in alloy A may havecontributed to the creation of damage during coldrolling, which ultimately influenced the levels of cavita-tion during superplastic testing. Micrographs such as those shown in Fig. 11 werequantitatively analyzed using NIH Image software todetermine the volume fraction and frequency of cavities.Fig. 12(a) depicts the measured number of cavities perunit area as a function of strain for the two alloys. Thevariations in the measurements lie within the boundinglines that describe the data trend. It is generally expectedthat the number of cavities would approach zero at zerostrain. (Micrographs of the grip region of the sampleconfirm that this is true.) Fig. 12(b) shows the averagecavity diameter as a function of strain. The averagecavity diameter in alloy A is somewhat smaller, i.e., ithas a greater number of smaller cavities at low strainuntil rapid growth of the cavities occurs near failure.This smaller average cavity size has previously been seenin alloys containing more particles and fits with existing Fig. 12. Measurements of cavitation for alloys A and B at 500 8C.
  • 9. 236 R.M. Cleveland et al. / Materials Science and Engineering A351 (2003) 228 Á/236chemistry has been investigated to understand the Acknowledgementsdifferences in their superplastic response. The two sheetshad similar grain sizes but different grain size distribu- The authors thank General Motors Research andtions, different strain-rate sensitivity values, and sig- Development Center for the support of this research.nificantly different particle content resulting from smallvariations in Mg, Mn, and Fe content. This researchshows that in similarly processed alloys with changes in Referencesm values, cavity-causing particles can have a measurable [1] C.H. Hamilton, A.K. Ghosh, J.A. Wert, Met. Forum 8 (1985)effect on superplastic elongation. 172. Alloy A with a slightly higher Mg level compared with [2] D.H. Bae, A.K. Ghosh, Acta Mater. 48 (2000) 1207.alloy B exhibited lower flow stress in constant strain- [3] A.K. Mukherjee, in: R.W. Cahn, P. Haasen, E.J. Kramer (Eds.), Materials Science and Technology, vol. 6, VCH, New York, 1993,rate tests, possibly because of enhancement in the p. 404.chemical diffusivity of the alloy. [4] A.K. Ghosh, R. Raj, Acta Metall. 29 (1981) 607. In spite of the lower flow stress for alloy A, the [5] T.G. Langdon, Met. Sci. 16 (1982) 175.tendency for cavity initiation is not reduced because of [6] C.C. Bampton, A.K. Ghosh, M.W. Mahoney, in: R. Pearce, L. Kelly (Eds.), Superplasticity in Aerospace */Aluminum, Ashfordits larger density of particles and lower strain-rate Press, Southampton, UK, 1985, p. 1.sensitivity value. [7] J. Pilling, N. Ridley, Res. Mech. 23 (1988) 31. The peak values of strain-rate sensitivity were lower [8] A.H. Chokshi, T.G. Langdon, J. Mater. Sci. 24 (1989) 143.for alloy A than for alloy B, which appears to result [9] N. Ridley, Z.C. Wang, Mater. Sci. Forum 170 Á/172 (1994) 177. [10] D.H. Bae, A.K. Ghosh, Mater. Sci. Eng. A 322 (2002) 233.from a lower fraction of fine grains. The m values [11] R. Verma, P.A. Friedman, A.K. Ghosh, Metall. Mater. Trans. Adecreased with increasing strain, as expected from 27 (1996) 1889.concurrent grain growth. Lower m values led to more [12] M. Furukawa, A. Utsunomiya, K. Matsubara, Z. Horita, T.G. Langdon, Acta Mater. 49 (2001) 3829.rapid cavity growth and interlinkage in alloy A. The [13] P.A. Friedman, A.K. Ghosh, Metall. Mater. Trans. A 27 (1996)combination of a large number of particles and lower m 3030.values in alloy A resulted in reduced ductility. [14] D.H. Bae, A.K. Ghosh, Acta Mater. 50 (2002) 1011.