An Investigation of Compressive Creep Aging Behavior of Al-Cu-Li Alloy Pre-Treated by Compressive Plastic Deformation and Artificial Aging

In this paper, the effects of compressive pre-deformation and successive pre-artificial aging on the compressive creep aging behavior and microstructure evolution of the Al-Cu-Li alloy have been studied. Severe hot deformation mainly occurs near the grain boundaries during the compressive creep initially, which steadily extends to the grain interior. After that, the T1 phases will obtain a low radius–thickness ratio. The secondary T1 phases in pre-deformed samples usually only nucleate on dislocation loops or Shockley incomplete dislocations induced by movable dislocations during creep, which are especially prevalent in low plastic pre-deformation. For all pre-deformed and pre-aged samples, two precipitation situations exist. When pre-deformation is low (3% and 6%), solute atoms (Cu and Li) can be consumed prematurely during pre-aging at 200 °C, with dispersed coherent Li-rich clusters in the matrix. Then, the pre-aged samples with low pre-deformation no longer have the ability to form secondary T1 phases in large quantities during subsequent creep. When dislocation entangles seriously to some extent, a large quantity of stacking faults, together with a “Suzuki atmosphere” containing Cu and Li, can provide the nucleation sites for the secondary T1 phase, even when pre-aged at 200 °C. The sample, pre-deformed by 9% and pre-aged at 200 °C, displays excellent dimensional stability during compressive creep because of the mutual reinforcement of entangled dislocations and pre-formed secondary T1 phases. In order to decrease the total creep strain, increasing the pre-deformation level is more effective than pre-aging.


Introduction
Al-Cu-Li alloys are extensively used for aerospace applications due to their high specific strength, good damage tolerance, and excellent property stability [1]. They can contain various precipitates such as θ -Al 2 Cu, δ -Al 3 Li, T 2 -Al 5 Li 3 Cu, T B -Al 7 Cu 4 Li, χ-Al 5 Cu 6 Li 2 , and T 1 -Al 2 CuLi [2][3][4]. The main strengthening precipitate in Al−Cu−Li alloys is a T 1 phase, which forms as semicoherent platelets on {111} Al planes and exhibits a hexagonal structure [5]. Plastic deformation prior to aging usually increases the density of fine strengthening precipitates through the introduction of dislocations acting as preferential heterogeneous matrix nucleation sites [6,7]. Kim et al. have found that the stretching treatment greatly accelerated the nucleation kinetics of the T 1 phase at the expense of S phases in Al-Li-Cu-Mg alloys [8], where there exists another precipitation sequence: α-(SSSS, Supersaturated Solid Solution) → clusters → GPB zone + GPB II zone/S → S /S (Al 2 CuMg). Gable et al. [9] have found that the size and quantity of the θ phase and the T 1 phase of an Al-Cu-Li-X alloy vary with the pre-deformation degree in the same aging treatment.

Experimental Procedure
The material investigated is a novel Al-Cu-Li cast alloy. Its actual chemical composition has been determined by ICP-AES (Inductively Coupled Plasma-Atomic Emission Spectroscopy, Model: Agilent 5110) and OES (Optical Emission Spectrometer, Model: SPEC-TROLAB), as listed in Table 1. The shape of samples cut from a cast billet has a square cross-section of 10 mm × 10 mm and a height of 15 mm, as illustrated in Figure 1a. The homogenization parameter is "460 • C/18 h → 525 • C/22 h → Air cooled" and solution treatment parameter is "530 • C/1.5 h → Water quenched" in a controllable resistance furnace. The pre-compression strains are 3%, 6%, and 9%, respectively, at a strain rate of 1 mm/min, as shown in Figure 2. The oil-bath resistance furnace is for additional preartificial aging, only for samples 3A#, 6A#, and 9A#. The product model for creep experiments is RDL-10, with the compression direction parallel to Z direction of the sample. The creep temperature should be controlled at 170 • C and a compressive stress of 200 MPa should be applied. The procedure of heat treatment process flow is shown through schematic diagram in Figure 1c, with the corresponding parameters listed in Table 2.         digital Vickers hardness tester under a load of 200 g for 15 s. The hardness value for each sample was averaged across all the successive indentations. Electron backscattered diffraction (EBSD) data were acquired on Z-Y section via Oxford NordlysMax3 system and processed by Channel 5 software using JEOL-JSM-7800F. KAM (Kernel average misorientation) maps represent the average misorientation (within 5 • ) of each pixel with respect to its surrounding pixels. Misorientations over a certain value should be discarded in order to exclude the misorientations associated with discrete sub-grains and GBs. Furthermore, TEM samples were observed by JEOL-JEM-F200 with an accelerating voltage of 200 kV.

Hardness Evolution
In Figure 3a, the hardness of sample 9A# is 128.37 HV 0.1 before creep, and it reaches the peak (142. 31  After creep, successive indentations were made at a distance of 0.5 mm apart along the central axes (Horizontal and vertical) on Z-Y section of each sample using a DHV-1000 digital Vickers hardness tester under a load of 200 g for 15 s. The hardness value for each sample was averaged across all the successive indentations. Electron backscattered diffraction (EBSD) data were acquired on Z-Y section via Oxford NordlysMax3 system and processed by Channel 5 software using JEOL-JSM-7800F. KAM (Kernel average misorientation) maps represent the average misorientation (within 5°) of each pixel with respect to its surrounding pixels. Misorientations over a certain value should be discarded in order to exclude the misorientations associated with discrete sub-grains and GBs. Furthermore, TEM samples were observed by JEOL-JEM-F200 with an accelerating voltage of 200 kV.

Hardness Evolution
In Figure 3a,

Local Misorientation
As seen in Figure 4a-c, the average local misorientation will increase due to pre-deformation. Furthermore, it seems dislocations are entangled seriously for sample 9#. Since the average local misorientation of solution-treated samples is 0.47°, dislocation proliferation starts to occur in sample 3# for its slightly higher value (0.58°). Pre-aging at 200 °C can promote recovery behavior. Here, the amplitude of the decrease in the average local misorientation between samples 3# and 3A#, 6# and 6A#, and 9# and 9A# is 0.12°, 0.10°, and 0.09°, respectively (Figure 4d-f). Since precipitation on dislocations during pre-artificial aging would retard recovery to a low level [24], the difference between samples 9# and 9A# before creep is smallest.

Local Misorientation
As seen in Figure 4a-c, the average local misorientation will increase due to predeformation. Furthermore, it seems dislocations are entangled seriously for sample 9#. Since the average local misorientation of solution-treated samples is 0.47 • , dislocation proliferation starts to occur in sample 3# for its slightly higher value (0.58 • ). Pre-aging at 200 • C can promote recovery behavior. Here, the amplitude of the decrease in the average local misorientation between samples 3# and 3A#, 6# and 6A#, and 9# and 9A# is 0.12 • , 0.10 • , and 0.09 • , respectively (Figure 4d-f). Since precipitation on dislocations during preartificial aging would retard recovery to a low level [24], the difference between samples 9# and 9A# before creep is smallest.  After the 12 h compressive creep test, the average local misorientation of all samples increases, as illustrated in Figure 5. Thus, the dynamic recovery and recrystallization have not consumed the dislocations induced by creep. Compared between Figures 4 and 5, on the one hand, the amplitudes are 0.41°, 0.26°, and 0.22° for samples 3#, 6#, and 9#, respectively. On the other hand, the amplitude is slightly smaller for samples 3A#, 6A#, and 9A# (0.40°, 0.27°, and 0.02°, respectively).

Texture and Schmidt Factor
Seen from the average value of the Schmidt factor only, the difference of each sample is not significant, as shown in Figure 6. However, the Schmidt factor distribution of sample 6A#-12 h is more concentrated, because the crystal orientation in the compression direction is in a single direction, namely, in [110]α, as shown in Figure 7. This easily leads to continuous deformation along a single slip band, resulting in shape instability (Figure 3b).

Texture and Schmidt Factor
Seen from the average value of the Schmidt factor only, the difference of each sample is not significant, as shown in Figure 6. However, the Schmidt factor distribution of sample 6A#-12 h is more concentrated, because the crystal orientation in the compression direction is in a single direction, namely, in [110]α, as shown in Figure 7. This easily leads to continuous deformation along a single slip band, resulting in shape instability (Figure 3b).

Texture and Schmidt Factor
Seen from the average value of the Schmidt factor only, the difference of each sample is not significant, as shown in Figure 6. However, the Schmidt factor distribution of sample 6A#-12 h is more concentrated, because the crystal orientation in the compression direction is in a single direction, namely, in [110] α , as shown in Figure 7. This easily leads to continuous deformation along a single slip band, resulting in shape instability (Figure 3b). Since other samples have more texture components, multiple slips can enhance the stability of the shape of these samples [25]. Since other samples have more texture components, multiple slips can enhance the stability of the shape of these samples [25].

EDS and Backscatter Image of SEM
As the strength of the GBs decreases during creep, the plastic deformation of the GBs occurs more easily. Normally, longer T1 phases will form on the GBs during artificial aging. However, the precipitates near the GBs are shorter than grain internal ones for sample 3# after an 8 h creep, as shown in Figure 8a. Therefore, violent collisions between dislocations and precipitates near the GBs occur. As shown by samples 6#-8h and 9#-8h in Figure  8b,c, the size and the distribution of precipitates seem similar, regardless of proximity to the GBs or in the grain interior. As shown by Figure 8d, some longer and thicker grain internal T1 phases exist after 8 h for sample 9A#. Since other samples have more texture components, multiple slips can enhance the stability of the shape of these samples [25].

EDS and Backscatter Image of SEM
As the strength of the GBs decreases during creep, the plastic deformation of the GBs occurs more easily. Normally, longer T1 phases will form on the GBs during artificial aging. However, the precipitates near the GBs are shorter than grain internal ones for sample 3# after an 8 h creep, as shown in Figure 8a. Therefore, violent collisions between dislocations and precipitates near the GBs occur. As shown by samples 6#-8h and 9#-8h in Figure  8b,c, the size and the distribution of precipitates seem similar, regardless of proximity to the GBs or in the grain interior. As shown by Figure 8d, some longer and thicker grain internal T1 phases exist after 8 h for sample 9A#.

EDS and Backscatter Image of SEM
As the strength of the GBs decreases during creep, the plastic deformation of the GBs occurs more easily. Normally, longer T 1 phases will form on the GBs during artificial aging. However, the precipitates near the GBs are shorter than grain internal ones for sample 3# after an 8 h creep, as shown in Figure 8a. Therefore, violent collisions between dislocations and precipitates near the GBs occur. As shown by samples 6#-8h and 9#-8h in Figure 8b,c, the size and the distribution of precipitates seem similar, regardless of proximity to the GBs or in the grain interior. As shown by Figure 8d, some longer and thicker grain internal T 1 phases exist after 8 h for sample 9A#.  Figure 9a-d shows the selected area electron diffraction pattern (SADP), bright-field (BF), and dark-field (DF) TEM micrographs captured along the <110>α zone axis for sample 3#-12h. The aperture used for DF is illustrated by a blue circle in Figure 9a. The main precipitates were T1, together with a small number of δ′/β′ and θ′ phases. Since the alloy has a low content of Mg, σ and S′/S phases are seldom seen. Further study reveals two kinds of T1 phases. One (Figure 9e), namely secondary T1, is formed dominantly during creep with few atomic layers thickness in <111>α directions, which nucleated around the δ′/β′ phases (Figure 9f) for the convenient supply of Li atoms [26]. The other one ( Figure  9g) originated from the insoluble initial coarse T1 phase after solution treatment, which will hinder dislocation movement with greater probability. A long thin T1 phase within the grain was also verified, as shown in Figure 9h,i.  Figure 9a-d shows the selected area electron diffraction pattern (SADP), bright-field (BF), and dark-field (DF) TEM micrographs captured along the <110> α zone axis for sample 3#-12h. The aperture used for DF is illustrated by a blue circle in Figure 9a. The main precipitates were T 1 , together with a small number of δ /β and θ phases. Since the alloy has a low content of Mg, σ and S /S phases are seldom seen. Further study reveals two kinds of T 1 phases. One (Figure 9e), namely secondary T 1 , is formed dominantly during creep with few atomic layers thickness in <111> α directions, which nucleated around the δ /β phases (Figure 9f) for the convenient supply of Li atoms [26]. The other one (Figure 9g) originated from the insoluble initial coarse T 1 phase after solution treatment, which will hinder dislocation movement with greater probability. A long thin T 1 phase within the grain was also verified, as shown in Figure 9h As shown in Figure 10a, the diffraction spots of the δ′/β′ phases for sample 6#-12h were very weak. The initial T1 phases appeared in rough lines along the {111}α habitual planes, both within the grain (Figure 10b) and near the GBs (Figure 10c), implying the initial T1 phases were twisted [27]. Sha et al. [28,29] have found that deformation has an influence on the orientation of the precipitate, alters the atomic configurations of the precipitate-matrix interface, and increases the misfit strain energy of the interface. Because of this lattice distortion, the interfacial energy between precipitate and matrix increases, leading to the achievement of high free energy and the re-dissolution of precipitates [30]. A part of an initial T1 phase changes its orientation gradually, and the rest suddenly changes to a fixed direction, which is indicated by blue lines in Figure 10d. Some initial T1 phases were fractured into several fragments, with the longitude direction deviating from <112>α direction (Figure 10e). Since the T1 phase was twisted, the distance between (0001)T1 layers was shorter than 0.935 nm (Figure 10f). The distribution of the secondary T1 phases in sample 6#-12h was different from sample 3#-12h. As shown in Figure 10g,h, the density of the secondary T1 phases on the (111 )α plane was higher than that of the (11 1)α plane. (e) Enlarged T 1 phases forming a circle, (f) enlarged δ /β phases with inset HRTEM, (g) helical dislocations together with thick T 1 phases observed by multi-beam BF electron microscope technique, (h) a single long T 1 phase, and (i) HRTEM of (h). Figure 10a, the diffraction spots of the δ /β phases for sample 6#-12h were very weak. The initial T 1 phases appeared in rough lines along the {111} α habitual planes, both within the grain (Figure 10b) and near the GBs (Figure 10c), implying the initial T 1 phases were twisted [27]. Sha et al. [28,29] have found that deformation has an influence on the orientation of the precipitate, alters the atomic configurations of the precipitate-matrix interface, and increases the misfit strain energy of the interface. Because of this lattice distortion, the interfacial energy between precipitate and matrix increases, leading to the achievement of high free energy and the re-dissolution of precipitates [30]. A part of an initial T 1 phase changes its orientation gradually, and the rest suddenly changes to a fixed direction, which is indicated by blue lines in Figure 10d. Some initial T 1 phases were fractured into several fragments, with the longitude direction deviating from <112> α direction (Figure 10e). Since the T 1 phase was twisted, the distance between (0001) T1 layers was shorter than 0.935 nm (Figure 10f). The distribution of the secondary T 1 phases in sample 6#-12h was different from sample 3#-12h. As shown in Figure 10g,h, the density of the secondary T 1 phases on the (111) α plane was higher than that of the (111) α plane. Additionally, the structure of the tip of the plate-shaped T 1 or θ phase was immature (Figure 10i-k). Additionally, the structure of the tip of the plate-shaped T1 or θ′ phase was immature (Figure 10i-k).  For sample 9#-12h, the reflections at the 1/2 g {220} α positions are caused by the θ and insoluble β -Al 3 Zr phases, and the slight streaks along the <200> α directions are related to θ phases viewed edge-on, as shown in Figure 11a. Compared among samples 3#-12h, 6#-12h, and 9#-12h, the contrast of the Al matrix in the BFs is uniform when the pre-deformation is large (Figure 11b,c). This means that either the δ -Al 3 Li phases or solute atom clusters have been exhausted to supply the growth of the T 1 phase. The β -Al 3 Zr and θ phases remained in Figure 11d-f. Additionally, the dislocations tangle into a network structure near the initial coarse T 1 phases (Figure 11g). It is obvious that the twisted initial T 1 phase has partially transformed into another phase during creep (Figure 11h). The secondary T 1 phases are verified in Figure 11i,j. For sample 9#-12h, the reflections at the 1/2 g {220}α positions are caused by the θ′ and insoluble β′-Al3Zr phases, and the slight streaks along the <200>α directions are related to θ′ phases viewed edge-on, as shown in Figure 11a. Compared among samples 3#-12h, 6#-12h, and 9#-12h, the contrast of the Al matrix in the BFs is uniform when the pre-deformation is large (Figure 11b,c). This means that either the δ′-Al3Li phases or solute atom clusters have been exhausted to supply the growth of the T1 phase. The β′-Al3Zr and θ′ phases remained in Figure 11d-f. Additionally, the dislocations tangle into a network structure near the initial coarse T1 phases (Figure 11g). It is obvious that the twisted initial T1 phase has partially transformed into another phase during creep (Figure 11h). The secondary T1 phases are verified in Figure 11i,j. When pre-aged at 200 °C for 2 h, precipitation is quite different after 12 h creep. As shown in Figure 12a,b, the secondary T1 phase in sample 3A#-12h appear in small When pre-aged at 200 • C for 2 h, precipitation is quite different after 12 h creep. As shown in Figure 12a,b, the secondary T 1 phase in sample 3A#-12h appear in small amounts. The β -Al 3 Zr phase still exists with the δ phase wrapping around it, as shown in Figure 10c. The δ /β composite appears as a fish-eye shape. The δ phase alone may no longer exist. The θ phase was hardly present. Instead, a large number of coherent clusters consisting of a few nanometers are dispersed in the matrix (Figure 10d) and display weak Z-contrast, especially when adjacent to the coarse initial T 1 phase (Figure 10c). Another new phase has formed from the inside of the T 1 phase, with the new phase and the Al matrix separated by the T 1 phase a few atomic layers thick (Figure 10e,g,h). Combined with Figure 10b,e, the new phase also contains a large quantity of Cu atoms. Additionally, the secondary T 1 phase appears as a thick plate with a small diameter (Figure 10f). amounts. The β′-Al3Zr phase still exists with the δ′ phase wrapping around it, as shown in Figure 10c. The δ′/β′ composite appears as a fish-eye shape. The δ′ phase alone may no longer exist. The θ′ phase was hardly present. Instead, a large number of coherent clusters consisting of a few nanometers are dispersed in the matrix (Figure 10d) and display weak Z-contrast, especially when adjacent to the coarse initial T1 phase (Figure 10c). Another new phase has formed from the inside of the T1 phase, with the new phase and the Al matrix separated by the T1 phase a few atomic layers thick (Figure 10e,g,h). Combined with Figure 10b,e, the new phase also contains a large quantity of Cu atoms. Additionally, the secondary T1 phase appears as a thick plate with a small diameter (Figure 10f). The precipitation constitution of sample 6A#-12h is similar to sample 3A#-12h (Figure 13). The secondary T1 phases lay on the (11 1)α plane rather than on the (111 )α plane. Additionally, some secondary T1 phases have been divided into sections of low radiusthickness ratio, with the tip or edge of the plates dissolved to some extent (Figure 13d). The precipitation constitution of sample 6A#-12h is similar to sample 3A#-12h ( Figure 13). The secondary T 1 phases lay on the (111) α plane rather than on the (111) α plane. Additionally, some secondary T 1 phases have been divided into sections of low radius -thickness ratio, with the tip or edge of the plates dissolved to some extent (Figure 13d).

As shown in
The sample 9A#-12h is unique. It consists of the initial T 1 phase, the secondary T 1 phase, β -Al 3 Zr, and atomic cluster. Although the existence of minor δ or θ phases cannot be ruled out, it does not influence the mechanical properties. Only the reflections of the T 1 phases and the matrix are visible (Figure 14a). The cluster viewed under the [100] α zone axis displays a low diffraction contrast in Figure 14c. The lattice distortion of the matrix near the interface of the cluster can be observed in Figure 14d. Compared with samples 3A#-12h and 6A#-12h, the density of the secondary T 1 phase in sample 9A#-12h is quite high (Figure 14g,h). The reflections at the 1/2 g {220} α positions in Figure 14e could be caused by an insoluble β phase which happened to be enclosed in the diffraction aperture. The mature and stable crystal structure of the secondary T 1 phase is identified in Figure 14i. The sample 9A#-12h is unique. It consists of the initial T1 phase, the secondary T1 phase, β′-Al3Zr, and atomic cluster. Although the existence of minor δ′ or θ′ phases cannot be ruled out, it does not influence the mechanical properties. Only the reflections of the T1 phases and the matrix are visible (Figure 14a). The cluster viewed under the [100]α zone axis displays a low diffraction contrast in Figure 14c. The lattice distortion of the matrix near the interface of the cluster can be observed in Figure 14d. Compared with samples 3A#-12h and 6A#-12h, the density of the secondary T1 phase in sample 9A#-12h is quite high (Figure 14g,h). The reflections at the 1/2 g {220}α positions in Figure 14e could be caused by an insoluble β′ phase which happened to be enclosed in the diffraction aperture. The mature and stable crystal structure of the secondary T1 phase is identified in Figure  14i.

Discussion
Normally, the T 1 phase can nucleate on: (1) the dislocation and stacking fault; (2) the (sub-)grain boundary; (3) vacancies and octahedral holes (secondary defects formed by vacancies); (4) the Solute cluster; (5) the GP area; and (6) the Dispersion phase [31]. For pre-deformed samples only (3#, 6#, and 9#), the dislocation density continues to increase during creep, and the T 1 is dominant. The deformation during creep near the GBs is more severe for sample 3#. Then, the initial T 1 phases continue to crash into sections or dissolve to become spherical, which reduces the interfacial energy and makes the system stable. When dislocations become entangled around GBs, the dislocation density increases in the grain interior, where the initial T 1 also becomes twisted or dissolved. The difference in hardness between samples 3#-12h and 9#-12h is very small; comparatively, the dislocation density of sample 9#-12h is higher. Thus, the contribution of precipitation strengthening of secondary T 1 is great for sample 3#-12h. The movable dislocations in the grain interior, rather than the entangled ones around the GBs, tend to form dislocation loops or helical dislocations that benefit the nucleation of secondary T 1 and the formation of the final closed quadrilateral structure. Additionally, due to the entangled dislocation network and faster solute diffusion, the transformation of the initial T 1 phases easily occurred. The situation of sample 6#-12h are between samples 3#-12h and 9#-12h. On the one hand, the entangled dislocation network can accelerate the dissolution of initial T 1 phases; on the other hand, the relatively low movable dislocation density is not beneficial for the formation of a compact structure consisting of four T1 phases, thereby widening phase spacing. This causes the abnormal phenomenon of the hardness of sample 6# fitting into the middle between samples 3# and 9# after just a few hours of creep to occur.
When it comes to samples 3A# and 6A#, the evolutionary path of microstructure has changed. Since the pre-aging temperature is relatively high (200 • C), vacancy clusters and dislocation loops cannot form easily [32][33][34], which impedes the formation of secondary T 1 . In addition, the solute atoms Cu and Li diffuse easily to join the initial coarse T 1 or other stable phases, such as T 2 , δ-AlLi, θ, and Al-Cu-Fe in the GBs. The consequence is that solute atoms (especially Cu) prepared for the precipitation of secondary T 1 are insufficient before creep aging. Although Li-vacancy clusters or δ phases can be continuously cut into smaller pieces by movable dislocation and dissolved steadily to provide Li atoms and the vacancy needed for the nucleation and growth of the T 1 phase during creep, Li atoms and "free" vacancies would capture each other quickly rather than join the formation of T 1 by long term diffusion. The evidence mentioned in Figures 12d and 14d seems to suggest a new phase, whose size is between the Li-vacancy cluster and the GP-Li reported by Ma et al. [35]. Additionally, the precipitation of GP-T 1 required a longer incubation time because of a higher activation energy due to the lower degree of Cu super-saturation in the solid solution. Finally, the hardness of samples 3A#-12h and 6A#-12h is quite low. For sample 3A#, the continuous plastic deformation in one direction can be coordinated or impeded by other slip systems due to multiple texture components. The reason for the size instability of sample 6A# may be the single texture component.
Nevertheless, sample 9A# still obtains sufficient strength even in an over-aged state after a 12 h creep. Hence, there must exist another mechanism for T 1 nucleation. The dispersive "Suzuki atmosphere" by segregation of Cu and Li atoms in stacking faults on {111}α planes in the nucleation stage of T 1 in a deformed Al-Cu-Li alloy has been reported [36][37][38]. The conjecture is, apart from GP-T 1 , vacancy-solute clusters, and dislocation loops (a/6<112>Shockley incomplete dislocation), the sample 9A# may have produced a large quantity of stacking faults before creep, making Li atoms and Cu atoms continuously aggregate to the nucleating T 1 phase. The creep rate in the early creep stage is highly related to the pre-deformation level, rather than to pre-aging. The decreasing rate of the height is higher for samples 3# and 3A#, and lowest for samples 9# and 9A#, initially. An interesting result is that the nucleated T 1 phase in sample 9A# before creep will not only hinder the recovery behavior greatly but also impede the creep deformation, especially in the grain interior, making the change in local misorientation between samples 9A#-0h and 9A#-12h just 0.02 • , compared to 0.22 • between samples 9#-0h and 9#-12h. Above all, the stacking faults and incomplete dislocation obtained from dislocation decomposition seem to play more important roles in the nucleation of the T 1 phase during compressive creep.

Conclusions
In this paper, the effects of compressive pre-deformation and successive pre-artificial aging on compressive creep aging behavior and microstructure evolution of Al-Cu-Li alloys have been characterized through a comparative analysis of six different states (3#, 6#, 9#, 3A#, 6A#, and 9A#). Based on the results obtained, we can summarize the following conclusions: 1.
The pre-deformed samples (3#, 6#, and 9#) after creep mainly consist of a coarse initial T 1 phase, a secondary T 1 phase with minor β /δ and θ phases. The dislocation loop or Shockley incomplete dislocation induced by movable dislocation may benefit the nucleation of the secondary T 1 phase, especially with moderately low plastic deformation, as was shown in sample 3#. The dislocation density contributes more to the hardness of sample 9#-12h and accelerates the aging process to reach the peak-aged state earlier.

2.
For pre-deformed and pre-aged samples (3A#, 6A#, and 9A#), there exist two precipitation situations. If pre-deformation is not enough to produce the stacking fault, the solute atoms Cu and Li would be consumed prematurely before creep, rather than forming the secondary T 1 phase during pre-aging at 200 • C, with dispersed coherent Li-rich clusters in the matrix. Then, the alloys (samples 3A# and 6A#) no longer have the ability to form a secondary T 1 phase in any quantity during subsequent creep. In other words, when the dislocation entangles seriously to some extent, a large quantity of stacking faults together with a "Suzuki atmosphere" containing Cu and Li can provide the nucleation sites for secondary T 1 again, even pre-aged at 200 • C.

3.
The creep rate in the early creep stage is highly related to pre-deformation level rather than pre-aging. The decreasing rate of the height is higher for samples 3# and 3A# initially, due to the high density of movable dislocation. The low radius-thickness ratio of the secondary T 1 phase near the GBs indicates that severe compressive creep deformation first occurs near the GBs, which extends to the grain interior steadily. An inappropriate pre-deformation and pre-aged treatment (sample 6A#) would lead to a single texture component, which results in successive slip bands and an unstable region. Sample 9A#, with proper pre-treatment, displays excellent dimensional stability during compressive creep. Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data that support the findings of this study are available from the corresponding author, upon reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest. We have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.