ASM International 1059-9495/$19.00
JMEPEG (2016) 25:2432–2444 DOI: 10.1007/s11665-016-2076-6
Precipitation Hardening and Statistical Modeling of the Aging Parameters and Alloy Compositions in Al-Cu-Mg-Ag Alloys A.M. Al-Obaisi, E.A. El-Danaf, A.E. Ragab, and M.S. Soliman (Submitted January 20, 2016; in revised form April 11, 2016; published online April 25, 2016) The addition of Ag to Al-Cu-Mg systems has been proposed to replace the existing high-strength 2xxx and 7xxx Al alloys. The aged Al-Cu-Mg-Ag alloys exhibited promising properties, due to special type of precipitates named X, which cooperate with other precipitates to enhance the mechanical properties significantly. In the present investigation, the effect of changing percentages of alloying elements, aging time, and aging temperature on the hardness values was studied based on a factorial design. According to this design of experiments (DOE)—23 factorial design, eight alloys were cast and hot rolled, where (Cu, Mg, and Ag) were added to aluminum with two different levels for each alloying element. These alloys were aged at different temperatures (160, 190, and 220 °C) over a wide range of time intervals from 10 min. to 64 h. The resulting hardness data were used as an input for Minitab software to model and relate the process variables with hardness through a regression analysis. Modifying the alloying elementsÕ weight percentages to the high level enhanced the hardness of the alloy with about 40% as compared to the alloy containing the low level of all alloying elements. Through analysis of variance (ANOVA), it was figured out that altering the fraction of Cu had the greatest effect on the hardness values with a contribution of about 49%. Also, second-level interaction terms had about 21% of impact on the hardness values. Aging time, quadratic terms, and third-level interaction terms had almost the same level of influence on hardness values (about 10% contribution). Furthermore, the results have shown that small addition of Mg and Ag was enough to improve the mechanical properties of the alloy significantly. The statistical model formulated interpreted about 80% of the variation in hardness values. Keywords
aging, aluminum alloys, analysis of variance, design of experiments, precipitation hardening
1. Introduction Lightweight materials have played a significant role in the advance and development of many aviation applications. Speed, long range, as well as operational cost are the obvious factors in determining the aircraft performance (Ref 1, 2). Aluminum (Al) and its alloys are still the dominant materials for aircraft structure. For example, Concorde (Supersonic transport) fuselage material was the AA2618 Al alloy (Al2.2%Cu-1.5%Mg-1%Fe-1%Ni-0.2%Si) (Ref 3). After the famous problem with the Concorde aircraft, some aviation centers that are concerned with high-speed civil transport proposed the use of new Al alloys, Al-Cu-Mg-Ag alloys, to replace the one that had been used (Ref 3, 4). Adding silver (Ag) even with a very small amount to Al-Cu-Mg alloys has shown promising results in terms of room-temperature
A.M. Al-Obaisi, Mechanical Engineering Department, King Saud University, Riyadh, Saudi Arabia; and Mechanical Engineering Department, King Abdulaziz University, Jeddah, Saudi Arabia; E.A. El-Danaf and M.S. Soliman, Mechanical Engineering Department, King Saud University, Riyadh, Saudi Arabia; A.E. Ragab, Industrial Engineering Department, King Saud University, Riyadh, Saudi Arabia. Contact e-mail:
[email protected].
2432—Volume 25(6) June 2016
strength and toughness combinations. These notable improvements in mechanical properties of precipitation-hardened Al-CuMg-Ag alloys originated from a special new phase named X that forms as thin platelet precipitates on {111}a planes (a is an Albased solid solution) and having either hexagonal or orthorhombic shape (Ref 5-7). Since {111}a are slip planes in a Al-based solid solution alloys, precipitation of X in these alloys tends to improve resistance to dislocation slip and improve mechanical properties (Ref 8). The addition of Mg and Ag resulted in superior mechanical properties in the Al-Cu-Mg-Ag alloys compared with Al-Cu-Mg free of Ag and Al-Cu-Ag free of Mg (Ref 9). In Al-Cu-Mg systems with high weight ratio between Cu and Mg (e.g., 10-20), it was found that a small inclusion of Mg (0.3 wt.%) stimulated a group of strengthening intermetallic phases that include hÕ intermetallic phase (Al2Cu) with tetragonal shape and the S and SÕ intermetallic phases (Al2CuMg) with orthorhombic shape (Ref 10, 11). It was reported that hÕ and S phases have plate-like and lath/rod shapes, respectively (Ref 10). However, adding Ag to aged Al-Cu alloys free of Mg did not exhibit an apparent effect on the system. In contrast, small introduction of silver (0.1 wt.%) to Al-Cu-Mg with high weight percent ratio between Cu and Mg enhanced the creation of the X precipitates. This X phase was not seen in Al-Cu-Ag free of Mg. Thus, it was concluded that the existence of this phase was a consequence of a collaborated impact of Mg and Ag (Ref 6). A recent study (Ref 7) reported the formation of X in Al-Cu-Mg free of Ag, noting that Mg was intrinsic for the formation of X precipitates and Ag assisted the formation of X precipitates kinetically at the expense of hÕ phases. TEM micrographs of h¢, h, X, and S precipitates within aged Al-Cu-
Journal of Materials Engineering and Performance
Mg-Ag alloys and the effect of the aging time on the thickness of precipitates are presented in previous work (Ref 7, 10, 12). Until now, the existence of a precursor phase for the X precipitate such as Mg3Ag has not been reported in the asquenched alloy. Also, co-clustering of Ag and Mg has not been evidenced in the as-quenched alloy (Ref 13). After aging for 15 s at 180 C, co-clustering of Ag and Mg atoms occurs in AlCu-Mg-Ag alloy (Ref 14). However, at this stage, the coclusters did not include copper atoms. Cu atoms migrate to the sites of these co-clusters after aging for 30 s at 180 C and extremely small precipitates including Ag, Mg, and Cu atoms could be detected by TEM. These precipitates are believed to be a precursor to the formation of the X phase (Ref 14). The dimension of the precipitates in the Ag-containing alloy was smaller and the volume fraction was higher than that in the Ag-free alloys. This may be the reason for the novel mechanical properties observed for these alloys (Ref 15). Existence of three such precipitates h¢, X, and S¢ in an Al alloy is unique and can be attributed to the high level of age hardening achieved due to the existence of solute atoms such as Cu, Mg, and Ag (Ref 16, 17). The small size and high volume fraction of hÕ and X precipitates in the Ag-containing alloy improved the tensile strength and yield strength, especially at high temperatures. However, it was also responsible for the decrease in elongation (Ref 15). An increment in strength of 182 MPa was noted from the asquenched to peak-aged conditions at 180 C after 10 h for Al3.97Cu-1.07Li-0.43Mg-0.42Ag-0.15Zr-0.11Ti. This value of strength was maintained constant for up to 40 h of aging (Ref 9). It was, also, noted that Al-Cu-Ag alloy consumes longer time of about 40 h to realize the peak aging condition (Ref 9). This alloy exhibited age strengthening which was much lower compared with Al-Cu-Mg alloy. The combined addition of magnesium and silver was obviously seen as a necessary parameter for realizing the high strength of Al-Cu-Mg-Ag alloy. Precipitations of X phase, which are the source of the high thermal stability of Al-Cu-Mg-Ag alloys, were extensively studied from various perspectives including composition, structure, size, and stability range. Researchers used several techniques to study such precipitates, mostly transmission electron microscopy (TEM), time-of-flight atom probe field ion microscopy (APFIM), selected area electron diffraction pattern (SAED), collected beam electron diffraction (CBED), microbeam electron diffraction (MBED), and convergent beam electron diffraction (CBED) (Ref 12-15, 17-20). Most researchers agreed that the composition of X is Al2Cu (Ref 12, 13), and also deduced that Ag and Mg were detected at the matrix/X interface (Ref 12, 13, 19). Hutchinson et al. (Ref 19) stated that the X precipitates are stable if the alloy is aged below 250 C and also studied the size of X precipitates at different temperatures and aging times extensively and concluded that there is no change in size for aging at 200 C and the thickness is about 5.5 nm. Ringer et al. (Ref 12) deduced that X precipitates have plate-like orthorhombic structure and that h¢ precipitates are more stable at higher temperature (300 C) compared with X phase. Precipitation hardening assumes various forms (under-aged, peak-aged, and over-aged), and each has its specific type, size, and distribution of precipitates and can be expected to exhibit different behavior during plastic deformation. During the isothermal aging for solution-treated Al alloys, the hardness displays an increase until the peak-aged condition and finally a decrease in hardness is noted as the sample becomes over-aged.
Journal of Materials Engineering and Performance
In this over-aged condition, precipitates are largely bypassed by dislocations through the Orowan mechanism and the low strength is attributed to the poor solid solution hardening as the majority of the solutes present in the alloy are precipitated as relatively large non-coherent particles. The higher the temperature, the faster the response to achieve peak-aged condition. However, the peak-aged hardness decreases with increasing aging temperature (Ref 21). From an industrial point of view, it is interesting to note that at high temperatures the artificial aging response is quite rapid. It is now obvious that varying percentages of alloying elements (Cu, Mg, and Ag) in an Al alloy, aging temperatures as well as aging incubation durations can give rise to a distinct distribution of mechanical properties. This is what motivated the current work. In this study, based on a design of experiments (DOE) full factorial design, eight Al alloys were prepared with two levels of three alloying elements (Cu, Mg, and Ag). Then a comprehensive aging study, at three temperatures and nine aging durations for each temperature, was performed to fully characterize the aging characteristics of these alloys. The focus of the current work is to fully explore the age hardening capability of these, relatively new Al alloys, since most of previous work investigated the aging at a single combination of temperature and time for a single alloy composition. From the above discussion of microstructure and precipitating phases, it seems obvious that many previous investigations were focused on defining these phases with different techniques, but to date the full documentation of hardness over a wide range of alloy compositions, aging temperatures, and durations is still missing.
2. Methodology 2.1 Material Preparation Based on the factorial design, three alloying elements (Cu, Mg, and Ag) were added to Al with two levels. Therefore, eight alloys with the intended different compositions shown in Table 1 were chill cast in a steel mold. The cast ingots had the dimensions of 100 9 40 9 15 mm. These alloys were homogenized at 540 C for 24 h. The elemental analysis of the eight alloys was carried out using arc and spark excitation and is shown in Table 2. The eight alloys were processed by hot rolling at 450 C to reduce the thickness up to 80%. It can be concluded, from the chemical analysis presented in Table 2, that the maximum variation in the composition of each element does not exceed ±8% from its intended value.
Table 1 Intended compositions of the designed eight Al alloys Alloy no. 1 2 3 4 5 6 7 8
wt.% Cu
wt.% Mg
wt.% Ag
3.0 5.0 3.0 5.0 3.0 5.0 3.0 5.0
0.5 0.5 1.0 1.0 0.5 0.5 1.0 1.0
0.3 0.3 0.3 0.3 0.6 0.6 0.6 0.6
Volume 25(6) June 2016—2433
Table 2 Chemical analysis of the eight Al alloys (wt.%) Alloy no.
Cu
Mg
Ag
Cu/Mg ratio
1 2 3 4 5 6 7 8
3.09 5.29 3.24 4.98 3.15 5.22 3.24 5.11
0.469 0.461 0.949 0.951 0.466 0.448 0.939 0.959
0.313 0.301 0.309 0.306 0.617 0.611 0.634 0.614
6.5 11.5 3.4 5.2 6.8 11.6 3.45 5.32
Table 3 Factors and their levels in the experiment factorial design Factor
The percentage of other alloying elements: Si £ 0.05, Fe £ 0.2, Ni £ 0.03, Cr £ 0.056, Zn £ 0.031, Ti £ 0.016, and Mn £ 0.006. Chemical analysis was conducted by SPECTROMAX spark analyzer
Each rolled alloy was cut into small pieces with dimensions 10 9 10 9 3 mm to be used in the aging study. Alloys with high weight percent of Cu (5 wt.%) were solution treated at 540 C and those with low weight percent of Cu (3 wt.%) were solution treated at 500 C. After solution treatment, the samples were quenched. Subsequently, they were aged in a salt bath, composed of 50% KNO3 and NaNO2, at three different temperatures (160, 190, and 220 C), and for nine different time intervals (10 and 30 min, 1, 2, 4, 8, 16, 32, and 64 h), followed by water quenching.
2.2 Material Testing The Vickers micro-hardness values were measured, for all aged samples, using a Buehler MICROMET hardness tester at a load of 200 g and the reported value is the average of eight readings. Optical microscopy was conducted, on alloys 1 and 8, on an Olympus light microscope. The samples were ground and polished and then chemically etched using KellerÕs reagent. The line intercept method was employed to estimate the average grain size. Room-temperature quasi-static tensile experiments were performed on alloys 1 and 8, at a strain rate of 103 s1. The samples were machined in accordance with ASTM: E8, from the hot-rolled billet and heat treated to peak-aged condition. These experiments were performed using an electro-mechanical 3385 Instron material testing machine. The experiments were performed in the displacement-controlled mode to obtain a constant engineering strain rate throughout the deformation process. Fractography was conducted on the fractured surface using a 6610 LV JEOL scanning electron microscope. Energydispersive spectroscopy (EDS) was conducted using Oxford Instruments system.
2.3 Design of Experiments (DOE) Experiments were conducted according to a full factorial design as the nature of interactions between the studied factors is not known. Full factorial, although a comprehensive and resource-consuming design, allows users to depict all types of interactions within the experimental data. The design included three factors (Cu%, Mg%, and Ag%) with two levels each, one factor (aging temperature) with three levels, and one factor (aging time) with nine levels. 216 experiments were needed for the full factorial design. Table 3 summarizes the
2434—Volume 25(6) June 2016
A (wt.% of Cu) B (wt.% of Mg) C (wt.% of Ag) D (Temp, C) E (Time, h)
Levels
Values
2 2 2 3 9
3, 5 0.5, 1.0 0.3, 0.6 160, 190, 220 0.167, 0.5, 1, 2, 4, 8, 16, 32, 64
factors and their levels in the experiments. The reasons behind selecting the aforementioned compositions and aging parameters are to investigate the effect of alloying elements (Cu, Mg, and Ag) on the age hardening capability of Al-Cu-Mg-Ag alloys while limiting the number of cast alloys to eight (23 factorial design). Most of the previous work either investigated the aging at a single combination of temperature and time for a single alloy composition, or a combination of two temperatures with a limited set of time intervals, again for a single alloy composition. Therefore, there is no previous work that fully documented the range of hardness that can be attained by aging these alloys. The objective of the current study is to give a full documentation of the range of hardness that can be obtained with a limited number of tests (8 alloys). Since the work is relatively time demanding (casting, hot rolling followed by aging study), we started with 23 factorial design to limit the number of alloys (full experiments will require 33 = 27 alloys). The aging temperatures were selected to cover the range of temperature used to age similar alloys in previous work. The nine time intervals were needed to fully characterize the age hardening characteristics to scan the full acceptable time range to be able to depict the change in hardness where under-aged, peak-aged, and over-aged conditions could be clearly distinguished for each alloy and at each temperature.
3. Results and Discussion 3.1 Aging Study Figure 1 to 8 present the variation of hardness versus aging time for all eight alloys at three different aging temperatures. These figures demonstrate that the hardness values increase with time, showing under-aged behavior, reaching the peakaged condition, and finally drop, exhibiting an over-aged behavior. Lower aging temperature (160 C) needed more time to reach the peak hardness (16-32 h), and the over-aged behavior was hardly noticed in the time range investigated for some alloys. At 220 C, the peak hardness was reached rapidly and fast softening took place. Table 4 presents the peak hardness values for the three aging temperatures and the corresponding aging time. Worth to mention is that the reported values are the average of eight readings on each sample. The peak hardness values were found to be inversely proportional to the aging temperature, and the higher hardness value came with the lower temperature, as apparent in Table 4. Aging at 190 C demonstrated good values of hardness with a reasonable time that fit industry needs. Taking alloy (1) as a reference point, and comparing to it alloys (2), (3), and (5) can give an indication to the individual effect of increasing
Journal of Materials Engineering and Performance
140
Hardeness (Hv)
120 100 80 220 °C 60
190 °C
40
160 °C
20 0 0.1
1
10
100
Time (hours)
Fig. 1
Micro-hardness values of alloy 1 (3Cu, 0.5Mg, and 0.3Ag)
200 180
Hardeness (Hv)
160 140 120 100
220 °C
80
190 °C
60
160 °C
40 20 0 0. 1
1
10
100
Time (hours)
Fig. 2
Micro-hardness values of alloy 2 (5Cu, 0.5Mg, and 0.3Ag)
160 140 Hardeness (Hv)
120 100 80
220 °C
60
190 °C
40
160 °C
20 0 0 .1
1
10
100
Time (hours)
Fig. 3
Micro-hardness values of alloy 3 (3Cu, 1Mg, and 0.3Ag)
the fraction of Cu, Mg, and Ag, respectively. The peak hardness values of alloy 2 (higher level of Cu), alloy 3 (higher level of Mg), and alloy 5 (higher lever of Ag) were HV 167, 132, and 127, respectively, at 190 C aging temperature. The peak hardness of alloy 1 aged at 190 C (low level of all alloying elements) was 123. The effect of high level of Cu was calculated roughly as 167123 123 100 ¼ 36%. The improvement in hardness values was about 7 and 3%, for increasing the level of Mg and Ag, respectively. The impact of having high level of Cu and Mg (alloy 4; peak hardness of 163 at 190 C), high level of Cu and Ag (alloy 6; peak
Journal of Materials Engineering and Performance
hardness of 173 at 190 C), high level of Mg and Ag (alloy 7; peak hardness of 150 at 190 C), and high level of the three alloying elements Cu, Mg, and Ag (alloy 8; peak hardness of 172 at 190 C) can be seen in the enhancement in hardness values as 32, 41, 22, and 40 in percent, respectively. For individually increasing each alloying element, Cu seems to be the most significant in increasing the hardness value due to aging (36%). The increase in Cu and Ag resulted in the highest increase in hardness (41%) that is almost similar to the level of enhancement achieved by increasing the fraction of all alloying elements (40%).
Volume 25(6) June 2016—2435
200 180
Hardeness (Hv)
160 140 120 100
220 °C
80
190 °C
60
160 °C
40 20 0 0.1
1
10
100
Time (hours)
Fig. 4
Micro-hardness values of alloy 4 (5Cu, 1Mg, and 0.3Ag) 140
Hardeness (Hv)
120 100 80 220 °C 60
190 °C
40
160 °C
20 0 0.1
1
10
100
Time (hours)
Fig. 5
Micro-hardness values of alloy 5 (3Cu, 0.5Mg, and 0.6Ag)
200 180
Hardeness (Hv)
160 140 120 100
220 °C
80
190 °C
60
160 °C
40 20 0 0.1
1
10
100
Time (hours)
Fig. 6
Micro-hardness values of alloy 6 (4Cu, 0.5Mg, and 0.6Ag)
It would be worthwhile to evaluate the performance of the currently prepared and processed Al-Cu-Mg-Ag alloys by comparing them with other existing aerospace aluminum alloys. For this comparison, the peak hardness values of the alloys 2, 4, 6, and 8 that exhibited the highest peak hardness values at aging temperatures of 160 and 190 C will be considered for this comparison (check the values in Table 4). AA2024 alloy is commonly used in aircraft structure, especially in wing and fuselage structures under tension. The weight percent of Cu and Mg in AA2024, is about 4.3 and 1.3 wt.% respectively. Sha et al. (Ref 22) showed that the peak hardness of such aged alloy at 170 C is about HV 150 and the
2436—Volume 25(6) June 2016
duration needed to reach this peak value is 80 h. In contrast, the peak hardness values of alloys 2, 4, 6, and 8, in the current work, for both aging temperatures at 160 and 190 C are higher than that exhibited by AA2024. Besides, much less aging durations are needed to reach to the peak hardness in the current work compared to the 80 h reported by Sha et al. (Ref 22). AA6061 is also used in aircraft structure. Tan and Said (Ref 23) studied the aging of this alloy under different aging temperatures for time intervals of up to 10 h. It was reported that the highest peak value was 122 if the alloy was aged at 175 C for 10 h. This value is much lower compared with alloys 2, 4, 6, and 8 for both aging temperatures of 160 and
Journal of Materials Engineering and Performance
180 160 Hardeness (Hv)
140 120 100
220 °C
80 190 °C
60
160 °C
40 20 0 0 .1
1
10
100
Time (hours)
Micro-hardness values of Alloy 7 (3Cu, 1Mg, and 0.6Ag)
Fig. 7
200 180
Hardeness (Hv)
160 140 120 100
220 °C
80
190 °C
60
160 °C
40 20 0 0.1
1
10
100
Time (hours)
Micro-hardness values of alloy 8 (5Cu, 1Mg, and 0.6Ag)
Fig. 8
Table 4 Peak Hardness of all alloys with its time at different temperatures 160 °C Temperature Alloy Alloy Alloy Alloy Alloy Alloy Alloy Alloy
1 2 3 4 5 6 7 8
190 °C
220 °C
Peak hardness (Hv)
Time (h)
Peak hardness (Hv)
Time (h)
Peak hardness (Hv)
Time
127 173 144 179 128 176 159 181
32 32 64 16 16 16 32 16
123 167 132 163 127 173 150 172
8 2 2 4 8 4 2 1
99 153 132 145 119 161 146 161
30 min 1h 30 min 1h 1h 10 min 10 min 30 min
190 C. Another Al alloy that is used in aircraft structure is the non-age-hardenable Al 5052. Kwon et al (Ref 24) reported the hardness value of 79 for such alloy. On the other hand, alloys 1 to 8, in the present study, can achieve higher hardness values regardless of aging temperature and time. AA7xxx is highly used in aerospace and aviation applications. Rout et al (Ref 25) investigated AA 7017, where it was reported that the peak hardness value was 153 if the alloy was aged at 120 C for 25 h. Moreover, the peak hardness values of alloys 2, 4, 6, and 8 for both aging temperatures at 160 and 190 C are higher than the depicted hardness for alloy AA7017. Fang et al (Ref 26) added Li to 7xxx (Al-5.17%Zn-1.22%Mg1.74%Cu-1.01%Li) and aged the alloy at 120 and 160 C. The
Journal of Materials Engineering and Performance
stated hardness values were HV 215 and 203, respectively. Also, the duration needed to reach these values were 35 and 12 h, respectively. Hardness values for both aging times and aging temperatures were higher than the obtained values in all current Al-Cu-Mg-Ag alloys. Finally, it can be deduced that the peak hardness of alloys 2, 4, 6, and 8 has larger values than most of the mentioned aerospace alloys, except for AA7xxx with Li addition. Direct observation of Table 4 reveals that maximum peak hardness values are achieved with alloys 2, 4, 6, and 8 for the three aging temperatures. It is worth mentioning that two of these alloys (alloys 2 and 6) exhibited the maximum Cu/Mg ratio, around 11.6.
Volume 25(6) June 2016—2437
Optical microscopy was conducted on alloys 1 and 8. The grains are equiaxed with an average grain size of about 250 lm, for both alloys, using the line intercept method. This microstructure can be seen in Fig. 9. To reveal the internal microstructure of alloys 1 and 8, tensile testing was conducted and the fractured surface was studied. Figure 10 shows the engineering stress-strain response for the two alloys. Alloy 8 exhibited a yield and a tensile strength of about 350 and 449 MPa, respectively. The corresponding values in alloy 1 are 250 and 348 MPa, respectively. Both alloys reached almost the same elongation to failure of about 15%. A detailed fractography study was conducted on the fractured surface of the two samples using SEM, at different magnifications. Figure 11 represents the SEM secondary electron images of the fractured surface for alloy 1. Figure 12 shows the EDS spectrum taken at a particle which is almost 1 lm in size. Figure 13 presents the fractography for alloy 8 at different magnifications. Figure 14 shows the EDS spectrum taken at a particle that was almost 1 lm in size. It is clear that both alloys exhibited a mixed fracture mode in a sense that intergranular fracture suggesting crack propagation along grain boundaries as well as dimples suggesting ductile fracture is evidenced for the two alloys. The particle analyzed for alloy 1 exhibited clusters of Al, Cu, and Mg atoms, which indicates the possible existence of a coarsened S phase precipitate. The particle analyzed for alloy 8 exhibited clusters of Cu and Al suggesting that it is basically a coarsened equilibrium h precipitate. Figure 15 presents an element map of an area on the fractured surface of alloy 8. The map points toward clusters of Cu, Mg, and Al mainly with lesser Ag content. This again could be an indication to the transformation of X to S equilibrium precipitates. It has been established (Ref 27) that during the aging of these alloys the sequence of formation of metastable and stable precipitates is as follows:
relative stability of the X phase with respect to that of the competing S¢ phase (Ref 27). This is important for achieving good mechanical properties in these alloys.
3.3 Statistical Analysis A general linear model was built in Minitab 17 to model the results of the experiments based on the factors and levels illustrated in Table 3. For the first run, the model included all terms (factors and interactions of all levels), and then insignificant terms (P value >0.05) were removed from the model one by one starting with the term that has the highest P value unless it is a part of higher level interaction term. Also, the term E3 was removed since it had a very high co-linearity with E2 and E. The analysis of variance is shown in Table 5 and summarized in Table 6. Each studied parameter showed a significant effect on the hardness by itself and/or by interaction with other terms. Detailed examination of the P values and adjusted sum of squares (Adj SS) spot more light on the results. Factor (E), aging time, is insignificant in its linear form, but significant in its quadratic form. E also has a strong interaction with aging temperature (D). The explanatory contribution of factors to the model is represented by the adjusted sum of squares. Apparently, factor
500
Engineering Stress (MPa)
3.2 Microstructure and Tensile Testing
450 400 350 300 250 200 150
Alloy 8 - PeakAged
100
Alloy 1 - Peak aged
50 0
GP zones ! h00 ! h0 þ X ! h0 þ S0 ! S þ h;
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
Engineering Strain
where GP zones are clusters of Cu atoms on {100}a planes which form in the early aging process. It is also well established that high Cu/Mg ratios in these alloys tend to enhance
Fig. 9
Fig. 10 Engineering stress-strain response for alloys 1 and 8 heat treated to peak-aged condition
Microstructure of alloy 1 (a) and alloy 8 (b)
2438—Volume 25(6) June 2016
Journal of Materials Engineering and Performance
Fig. 11 Secondary electron image of the fractured surface generated by tensile testing of alloy 1 in peak-aged condition at different magnifications
Fig. 12 EDS spectrum from a point analysis taken at a particle from the fractured surface of alloy 1 (Mg: 0.37 wt.%, Cu: 3.57 wt.% and Al: 96.06 wt.%)
Journal of Materials Engineering and Performance
Volume 25(6) June 2016—2439
Fig. 13 Secondary electron image of the fractured surface generated by tensile testing of alloy 8 in peak-aged condition at different magnifications
Fig. 14
EDS spectrum from a point analysis taken at a particle from the fractured surface of alloy 8 (Al: 77.19 wt.% and Cu: 22.81 wt.%)
A (wt.% of Cu) has the greatest contribution to the data variation, about 42%. The significance of third levels of interactions does not allow meaningful interpretation of the main effects and second-level interactions.
2440—Volume 25(6) June 2016
Model adequacy measures as the coefficient of determination (R2, adjusted R2, and predicted R2) show high levels (above 80%) which proves the adequacy of the model. With respect to adjusted R2, the model explains 84% of the
Journal of Materials Engineering and Performance
Fig. 15 Element mapping for an area on the fractured surface of alloy 8, showing primarily clusters of Cu and Mg atoms and to a lesser extent Ag atoms
Table 5 Analysis of variance Source
DF
Adj SS
Adj MS
F value
P value
A B C D E D*D E*E A*B A*C A*D A*E B*C B*D B*E C*D C*E D*E A*C*D B*C*D B*E*E C*E*E D*D*E D*E*E Error Total
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 192 215
54364 1312 1661 10919 193 1815 1475 6298 19 121 204 309 277 683 0 1729 15420 746 523 390 949 569 5704 18100 129741
54363.6 1312.2 1660.7 10918.9 192.8 1815.2 1475 6297.9 19 121.5 203.5 309.3 276.7 682.6 0.2 1728.6 15419.8 746.2 522.7 390.3 949.1 569.4 5704.3 94.3
576.68 13.92 17.62 115.83 2.05 19.26 15.65 66.81 0.2 1.29 2.16 3.28 2.93 7.24 0 18.34 163.57 7.92 5.54 4.14 10.07 6.04 60.51
0 0 0 0 0.154 0 0 0 0.654 0.258 0.143 0.072 0.088 0.008 0.96 0 0 0.005 0.02 0.043 0.002 0.015 0
Journal of Materials Engineering and Performance
Volume 25(6) June 2016—2441
variation in the data. The 82.89% value of the predicted R2 shows that the model is not over-fit and has a good predictability. Figure 16 shows the model residual plots. Apparently, the residuals show no pattern against the fitted value or against the run order. Random distribution of the residuals is clear. The plot shows also the residual normal probability plot and histogram. The residuals look normal except few residuals with very high negative values. 3.3.1 Regression. To simplify the regression equation, terms with low contribution ((adj. SS/ Total SS)< 0.01) to the data variation were removed one by one while assuring that the coefficient of determination does not change much. The model summary after removing low-contribution terms is illustrated in Table 7. The final result is
Table 6 Model summary S 9.70927
R2
R2 (adj)
R2 (pred)
86.05%
84.38%
82.89%
Hardness ¼ 314:7 þ 32:06A þ 108:0B þ 45:05C þ 2:792D þ 10:73E 0:00683D2 0:1235E 2 21:60A B 0:05409D E þ 0:000610D E 2 : Figure 17 shows the optimization plot for hardness. From the figure, it is clear that the hardness value increases with increasing weight fraction of Cu, Mg, and Ag and decreasing aging temperature and at an intermediate aging time. The optimum factor values to achieve maximum hardness were calculated as follows: Cu wt.% = 5, Mg wt.% = 1, Ag wt.% = 0.6, aging temperature = 160 C, and aging time = 22.73 h. With these settings, maximum hardness is expected to be 178.2 Hv. Based on this finding, a new cast of alloy 8 was made and processed under almost similar conditions of hot rolling. Aging was then conducted using the optimum values of aging temperature (160 C) and time (23 h) and the average hardness value was found to be 170 ± 5 using eight readings, which is very close to the value deduced earlier from the ANOVA analysis. The analysis of variance (ANOVA) was used to study the significance and weight of each term in the fitted model. From
Residual Plots for Hardness Norm al P robability P lot
Vers u s F its
99.9
20
90
Residual
Percent
99
50 10
0
-20
1 0.1
-30
-15
0
15
30
90
120
150
Residual
Fitted Value
H is togram
Vers u s Order
180
20
Residual
Frequency
30 20 10 0
0
-20
-30.0
-22.5
-15.0
-7.5
0.0
7.5
15.0
22.5
1
20
40
60
Residual
Fig. 16
80
100 120 140 160 180 200
Observation Order
Residual plots for hardness
Table 7 Model summary after removing low-contribution terms S 10.71
2442—Volume 25(6) June 2016
R2
R2 (adj)
R2 (pred)
81.89%
81.01%
80.04%
Journal of Materials Engineering and Performance
A 5.0 [5.0] 3.0
Optimal High D: 0.9760 Cur Predict Low
B 1.0 [1.0] 0.50
C 0.60 [0.60] 0.30
D 220.0 [160.0] 160.0
E 64.0 [22.7342] 0.1670
Hardness Maximum y = 178.2172 d = 0.97602
Fig. 17
Optimization plot for hardness
Cu wt. %
8%
Hold Values B 0.75 C 0.45 D 190
Mg Wt. % Ag Wt. % 1 60
21% Aging temp. 49% H ardness
Quadarc terms
1 20
Second lever interacon terms 9%
14 0
third level interacon terms
60 40
1 00 4
10%
Cu wt. %
2% 1%
Fig. 18
0 5
Fig. 20 Hardness vs. aging time and wt.% of Cu
Relative contribution of ANOVA model terms
Hold Values A 4 B 0.75 C 0.45
160
Hardness
Aging Time
20
3
140
Hold Values A 4 C 0.45 D 190 14 0
H ardness 1 30
1 20
6 60 120
1 00 60 160
40
180
20
200
Aging Temp. ( oC)
Fig. 19
40
220 0
0
20
0 0.50 0.75
M g wt . %
Aging Time
0 1.00
Aging Time (H ours)
Fig. 21 Hardness vs. aging time and wt.% of Mg
Hardness vs. aging time and temperature
ANOVA results, the sum of squares of each term divided by the total sum of squares represents a rough estimation of the relative contribution of each term (Ref 28). These relative contributions are illustrated in Fig. 18. It is obvious that Cu composition is the main affecting parameter in terms of controlling the alloy hardness. Figure 19 to 22 show the response surface plots of hardness against two factors while fixing the other three factors at their mean values. Figure 19 shows the variation of hardness with aging temperature and time, while the Cu, Mg, and Ag are at their mean values. The figure captures the fact that lower aging temperatures are associated with higher peak hardness values and vice versa. The effect of aging time cannot be depicted
Journal of Materials Engineering and Performance
clearly due to the non-linear nature of its effect. Figure 20 shows the variation of hardness with Cu fraction and aging time at an aging temperature of 190 C, while Mg and Ag are at their mean values. Obviously, the hardness reaches a peak value at the highest fraction of Cu and at an intermediate aging duration of about 20 h. Figure 21 and 22 present the variation of hardness with Mg and Ag fraction, respectively, and aging time, at an aging temperature of 190 C. The peak hardness was reached at the maximum fraction of alloying elements and at about 20 h aging time. The two figures showing the effect of Mg and Ag along with aging time exhibit a wavy structure due to the significant interaction between these alloying elements and aging time, as illustrated in Table 5. On the other hand, the waviness is much less for Fig. 19 where the effect of Cu is
Volume 25(6) June 2016—2443
Hold Values A 4 B 0.75 D 190
15 0 1 40
H ardness 130 60 40
12 0 20
0.3
0 .4
0.5
Ag w t . %
Fig. 22
Aging Time
0 0.6
Hardness vs. aging time and wt.% of Ag
displayed. This is due to the fact that there is no significant interaction between Cu fraction and aging time.
4. Conclusions Three alloying elements, namely Cu, Mg, and Ag, are added between two levels to 99.9% commercial pure aluminum, using chill casting, followed by homogenization, and hot rolling to about 80% rolling reduction. The aging characteristics for the processed materials are documented for three aging temperatures and nine different time intervals. A statistical model was formulated by regression analysis that relates process parameters to the resulting peak hardness. It was concluded that the individual effect of Cu weight percent was the most significant in impacting a variation in hardness values. The statistical model developed had a predictability that explains about 80% of the variation within the measured hardness values. Also the model showed that the optimum process variables were 5 wt.% of Cu, 1 wt.% of Mg, 0.6 wt.% of Ag, 160 C, and 22.7342 h, to achieve a maximum peak hardness of 178.22 Hv. The exhibited properties of Al-Cu-Mg-Ag systems make them potential candidates to be used, or even to replace existing aerospace aluminum alloys, in aviation applications.
Acknowledgment This work was supported by the Research Center, College of Engineering, Deanship of Scientific Research, King Saud University.
References 1. J.-P. Immarigeon, R.T. Holt, A.K. Koul, L. Zhao, W. Wallace, and J.C. Beddoes, Lightweight Materials for Aircraft Applications, Mater. Charact., 1995, 35(1), p 41–67 2. E.A. Starke, Jr, and J.T. Staley, Application of Modern Aluminum Alloys to Aircraft, Prog. Aerosp. Sci., 1996, 32(2–3), p 131–172 3. I.J. Polmear, G. Pons, H. Octor, C. Sanchez, A.J. Morton, W. Borbidge, and S. Rogers, After Concorde: Evaluation of an Al-Cu-Mg-Ag Alloy for Use in the Proposed European SST, Mater. Sci. Forum, 1996, 217– 222, p 1759–1764 4. C.L. Lach and M.S. Domack, Characterization of Al-Cu-Mg-Ag Alloy RX226-T8 Plate, NASA. 2003 NASA/TM-2003-212639
2444—Volume 25(6) June 2016
5. K.M. Knowles and W.M. Stobbs, The Structure of 111 Age-Hardening Precipitates in Al-Cu-Mg-Ag Alloys, Acta Crystallographica Sect. B, 1988, 44(3), p 207–227 6. B.C. Muddle and I.J. Polmear, The Precipitate X Phase in Al-Cu-MgAg Alloys, Acta Metallurgica, 1989, 37(3), p 777–789 7. A. Garg, Y.C. Chang, and J.M. Howe, Precipitation of the X Phase in an Al-4.0Cu-0.5Mg Alloy, Scripta Metallurgica et Materialia, 1990, 24(4), p 677–680 8. M. Grujicic, G. Arakere, C.-F. Yen, and B.A. Cheeseman, Computational Investigation of Hardness Evolution During Friction-Stir Welding of AA5083 and AA2139 Aluminum Alloys, J. Mater. Eng. Perform., 2011, 20(7), p 1097–1108 9. B.-P. Huang and Z.-Q. Zheng, Independent and Combined Roles of trace Mg and Ag Additions in Properties Precipitation Process and Precipitation Kinetics of Al-Cu-Li-(Mg)-(Ag)-Zr-Ti alloy, Acta Materialia, 1998, 46(12), p 4381–4393 10. R.J. Chester and I.J. Polmear, Tem Investigation of Precipitates In AiCu-Mg-Ag And Ai-Cu-Mg Alloys, Micron, 1980, II, p 311–312 11. R.K. Wyss and R.E. Sanders, Microstructure-Property Relationship in a 2xxx Aluminum Alloy with Mg Addition, Metallurgical Transactions A, 1988, 19(10), p 2523–2530 12. S.P. Ringer, W. Yeung, B.C. Muddle, and I.J. Polmear, Precipitate Stability in Al-Cu-Mg-Ag Alloys Aged at High Temperatures, Acta Metallurgica et Materialia, 1994, 42(5), p 1715–1725 13. K. Hono, N. Sano, S.S. Babu, R. Okano, and T. Sakurai, Atom Probe Study of the Precipitation Process in Al-Cu-Mg-Ag alloys, Acta Metallurgica et Materialia, 1993, 41(3), p 829–838 14. K. Hono, T. Sakurai, and I.J. Polmear, Pre-Precipitate Clustering in an Al-Cu-Mg-Ag Alloy, Scripta Metallurgica et Materialia, 1994, 30(6), p 695–700 15. M. Song, K. Chen, and L. Huang, Effects of Ag Addition on Mechanical Properties and Microstructures of Al-8Cu-0.5Mg Alloy, Trans. Nonferr. Met. Soc. China, 2006, 16(4), p 766–771 16. I.J. Polmear and R.J. Chester, Abnormal Age Hardening in an Al-CuMg Alloy Containing Silver and Lithium, Scripta Metallurgica, 1989, 23(7), p 1213–1217 17. R.N. Lumley and I.J. Polmear, The Effect of Long Term Creep Exposure on the Microstructure and Properties of an Underaged Al-CuMg-Ag alloy, Scripta Materialia, 2004, 50(9), p 1227–1231 18. D.H. Xiao, J.N. Wang, D.Y. Ding, and H.L. Yang, Effect of Rare Earth Ce Addition on the Microstructure and Mechanical Properties of an AlCu-Mg-Ag Alloy, J. Alloys Compd., 2003, 352(1–2), p 84–88 19. C.R. Hutchinson, X. Fan, S.J. Pennycook, and G.J. Shiflet, On the Origin of the High Coarsening Resistance of X Plates in Al-Cu-Mg-Ag Alloys, Acta Materialia, 2001, 49(14), p 2827–2841 20. B.M. Gable, G.J. Shiflet, and E.A. Starke, Jr., Alloy Development for the Enhanced Stability of X Precipitates in Al-Cu-Mg-Ag Alloys, Metall. Mater. Trans. A, 2006, 37(4), p 1091–1105 21. E.A. El-Danaf and T.M. El-Hossainy, Different Stress States Deformation of AA6082 Subjected to Different Artificially Aged Conditions, Adv. Mater. Res., 2010, 83–86, p 421–428 22. G. Sha, R.K.W. Marceau, X. Gao, B.C. Muddle, and S.P. Ringer, Nanostructure of Aluminium Alloy 2024: Segregation, Clustering and Precipitation Processes, Acta Materialia, 2011, 59(4), p 1659–1670 23. Chee Fai Tan and R. Mohamad, Said, Effect of Hardness Test on Precipitation Hardening Aluminium Alloy 6061-T6, Chiang Mai J. Sci., 2009, 36(3), p 276–286 24. K.W.O.N. Yong-Jai, S.H.I.M. Seong-Beom, and P.A.R.K. Dong-Hwan, Friction stir welding of 5052 aluminum alloy plates, Trans. Nonferr. Met. Soc China, 2009, 19(1), p s23–s27 25. P.K. Rout, M.M. Ghosh, and K.S. Ghosh, Influence of Aging Treatments on Alterations of Microstructural Features and Stress Corrosion Cracking Behavior of an Al-Zn-Mg Alloy, J. Mater. Eng. Perform., 2015, 24(7), p 2792–2805 26. W. Fang, L. Jinshan, H. Rui, and K. Hongchao, Influence of 1.0 wt.% Li on Precipitates in Al-Zn-Mg-Cu Alloy, Chin. J. Aeronaut., 2008, 21(6), p 565–570 27. A. Cho and B. Be`s, Damage Tolerance Capability of an Al-Cu-Mg-Ag Alloy (2139), Mater. Sci. Forum, 2006, 519–521, p 603–608 28. S.E. Urreta, F. Louchet, and A. Ghilarducci, Fracture Behavior of an Al-Mg-Si Industrial Alloy, Mater. Sci. Eng. A, 2001, 302(2), p 300–307
Journal of Materials Engineering and Performance