Introduction

Recently, attention has been directed towards the development of new and different severe plastic deformation (SPD) techniques that may be used to fabricate ultrafine-grained materials with grain sizes in the submicrometer and the nanometer range (Ref 1-4). These techniques include equal-channel angular pressing (ECAP) (Ref 5-8), high-pressure torsion (HPT) (Ref 3, 9, 10), torsion-extrusion (TE) (Ref 11-14), combined tension-torsion (CTT) (Ref 15, 16) and elliptical cross-sectioned spiral equal-channel extrusion (ECSEE) (Ref 17-21). All of these procedures are capable of introducing large plastic straining and significant microstructural refinement in bulk crystalline solids.

It has been shown over the last two decades that a unique set of physical and mechanical properties of nanostructured materials originates mainly from two reasons: (i) size effects and (ii) the presence of high-density interfaces in the material interior (Ref 22). However, an earned grain size and nanostructure depend on the utilized SPD methods, processing regimes, typical chemical composition, and initial microstructure of a material (Ref 7). For processing regimes, Mishra et al. (Ref 23) studied the microstructural evolution and mechanical response of copper subjected to ECAP under different processing routes, and found that the structure evolved from elongated dislocation cells (DCs) to subgrains and equiaxed grains with a diameter of ~200-500 nm. Orlov et al. (Ref 11, 24, 25) investigated the microstructural evolution of high-purity Al at the room temperature using TE. The results showed that subgrains after first TE pass owned a size of ~1.6 μm and the size remained basically the same regardless of an increase in TE passes. Haghdadi et al. (Ref 26) utilized the microhardness measurements to study the effect of pass numbers and deformation routes on hardness of commercially pure aluminum during accumulative roll bonding (ABE) processing. Wang’s works (Ref 18-20) showed that severe plastic torsion straining could usually produce ultrafine-grained structure with a mean grain size of 200-300 nm.

As one of the novel SPD methods, ECSEE can achieve a heavy simple shear plastic deformation by combining extrusion-torsion deformation and be capable of producing fully dense samples with an ultrafined grain size. The principle of ECSEE process is shown in Fig. 1, where shear strain accumulates within the workpiece without changing the cross-sectional area.

Fig. 1
figure 1

Schematic diagram of the principle of ECSEE (Ref 20)

Full size image

Wang (Ref 18, 20) has investigated the grain refinement process of pure copper subjected to shear strain during ECSEE. However, other materials received little further attention either in microstructural evolution or in mechanical properties using the technology. Therefore, this study conducts an in-depth investigation for the microhardness distribution and microstructural evolution of pure Al during ECSEE. Pure Al is chosen for this study because the results of pure Al are well documented for other SPD processes such as ECAP, HPT, and TE.

Simulation and Experiment

Cylindrical specimens with 10 mm in diameter and 25 mm in length were machined from the as-received 99.99%-purity aluminum billets produced by ingot casting. They were annealed at 553 K for 2 h and then cooled down at a rate of 298 K/h in order to introduce fairly homogeneous and equiaxed microstructures. The average grain size is about 120 μm, as measured by the linear intercept method.

The ECSEE tests were carried out on the YA32-315 hydraulic press with the extrusion speed of 5 mm/min at 25 °C with lard lubrication (Ref 18). The extrusion die was a set as follows: D 1 = 10 mm, L 1 = 7 mm, L 2 = 10 mm, L 3 = 7 mm, ϕ = 120°, and m = 1.55 (m was the ratio of major-axis length to minor-axis length of the elliptical cross section) (Ref 21). The ECSEE die was made of H13 steel with a hardness of 55 RC, and a backpressure was applied to the billets by the ejector pins. With reference to the structural shape and geometry, equivalent strain of ~1 was introduced for a single pass (Ref 21). This ECSEE process was repeated up to six passes, coming to a total average strain of ~6.0. This level of strain was selected because an UFG structure was attained by ECAP (Ref 27-32) and HPT (Ref 3).

Since the cross-sectional shape of ECSEE-ed samples was elliptical rather than round, we defined three orthogonal directions: extrusion direction (ED) parallel to the longitudinal axis, normal direction (ND) parallel to the short axes, and transverse direction (TD) parallel to long axes. Figure 2 shows the representative positions for Vickers microhardness (HV) tests using an Akashi MVK-E3 testing machine under a 100-g load applied for 15 s (Ref 33). Hardness measurements were made with an interval of 1 mm to avoid any interfering effect of adjacent indentations.

Fig. 2
figure 2

Schematic illustration of candidate positions for Vickers microhardness (HV) tests

Full size image

The schematic diagram of the representative locations for TEM observation of ECSEE-ed samples is shown in Fig. 3. Microstructural observation is conducted by means of a Philips Tecnai-20 transmission electron microscope with an accelerating voltage of 200 kV, the specimens for TEM analysis were well prepared with a twin-jet polisher, and a solution of 66% methanol and 33% nitric acid was used for etching.

Fig. 3
figure 3

TEM candidate positions of pure aluminum after ECSEE

Full size image

Results

Equivalent Strain Distribution

Finite element simulations were used to predict the billet deformation in the different passes separately. Figure 4 shows the equivalent strain distribution on the selected positions of specimens after different passes ECSEE. After six-pass deformation, the strain value at the edge is about six, which agrees well with the analytic solution according to the Ref 21. Strain distribution on three planes is not uniform, decreasing from the edge to center. However, strain gradient gradually decreases with an increase in deformation passes, behaving as indicated by the gradually expanding green color areas seen in Fig. 4(d) to (f). The strain gradient is introduced by the torsion force during ECSEE. Similar strain gradient is more significant for HPT.

Fig. 4
figure 4

Strain distribution on the selected positions of specimens after (a) 1-pass, (b) 2-pass, (c) 3-pass, (d) 4-pass, (e) 5-pass, (f) 6-pass

Full size image

In contrast, the distribution characteristics are slightly different among these three planes. ND-ED plane owns the minimum strain gradient while TD-ND plane does the maximal one, and TD-ED plane possesses the in-between one. The differences are due to nonsymmetric deformation during ECSEE. Besides, the cross section of sample after deformation is not strictly rounded considering the volume loss and deformation damage in simulation settings.

Microhardness Distribution

Based on the correlation between the hardness and the internal deformation, hardness measurement acts as a simple and expedient procedure for reaching conclusions on the deformation characteristics and the degree of internal homogeneity within the processed work-pieces (Ref 14).

Variations of HV taken from positions A, B, and C after different ECSEE cycles are shown in Fig. 5, where green points represent the measured data. The initial microhardness on the A, B, and C planes before ECSEE is 27 HV. There is a significant increase in hardness after ECSEE; for example, after 1 pass of ECSEE (Fig. 5(a)), the Vickers hardness increases to almost twice that of initial microhardness. However, the differences between adjacent extrusion cycles become smaller, predicting the hardness saturation on TD-ND. It is important to note that the hardness values in the ND-ED and TD-ED planes are slightly lower than in the TD-ND plane. Vickers hardness in the post-processed material is inhomogeneously distributed within both the transversal and longitudinal sections, as it is minimal (41 HV) in the vicinity of the billet axis and maximal at the edge areas (55 HV) in Fig. 5(a). Similarly, inhomogeneity can be found on the TD-ED and TD-ND planes. However, the hardness gradient reduces from the central to edge areas with the increasing number of ECSEE cycles; for example, the smaller discrepancies for 6-pass ECSEE billets in comparison with other pass deformation, which illustrates that a more uniform plastic deformation could be realized by the increasing number of ECSEE cycles. This phenomenon is consistent with the simulated strain distribution. This tendency is the same for all the planes. Kim et al. (Ref 34) have divided the pressed work-pieces into three zones (front, steady, and end zones), where a lesser shear deformation zone occurred at the outer part of the work-piece due to the faster flow of the outer part in comparison with the inner part within the main deformation zone.

Fig. 5
figure 5

Microhardness distribution on the selected positions: (a) A (ND-ED), (b) B (TD-ED), (c) C (TD-ND)

Full size image

Microstructural Evolution

Figure 6 shows typical TEM images of the candidate positions (Fig. 3) subjected to different ECSEE cycles, where P and C represent candidate positions and extrusion cycles, respectively. For example, P m C n represents that the position m is subjected to n-cycle ECSEE deformation. For P 1 C 1, the original grains are clearly elongated to form shear bands (SBs) with the width of 0.5-1.2 μm as denoted by red stars. When 2-pass deformation is imposed on P 1 during ECSEE, namely P 1 C 2, SBs are separated by dislocation into a large number of subgrains (SGs) or DCs, expressed by red arrows, in which a higher density of dislocations exist on the cells walls but lower density in the cell. Most of the dislocations organize dislocation walls and networks, while some are isolated. The SGs are also surrounded by a higher density of dislocations, but the dislocation density in the grain-subgrain structural elements is rather lower. On the other hand, only one single slip system is activated in some inner grains (P 1 C 2), described by white dotted lines, which suggests that these larger grains undergo a low level of deformation. For 4-cycle deformation, the cell structures become much denser in the direction of extrusion. It should be noted that the GBs in P 1 C 6 also have configurations of typical nonequilibrium GBs with a grain size of ~0.4 μm.

Fig. 6
figure 6figure 6

TEM micrographs of candidate positions after different ECSEE passes

Full size image

Similar to position 1, position 2 undergoes the similar refinement process from SBs or DCs to eventually equiaxed grains with equilibrium GBs. However, some evident differences from position 1 are summarized. First, a lower level of grain refinement exists in position 2; for example, the initial SB with the width of ~0.4 μm is retained after the second-cycle deformation (P 2 C 1 and P 2 C 2). Second, dislocation density is reduced compared to P 1 C 4; for example, only small numbers of SBs and DCs can be found in P 2 C 4.

An evident distinction between P 2 C 1 and P 1 C 1 is that P 2 C 1 has one single SB with a sharp, straight, and narrow grain boundary. Mishra (Ref 27) has maintained the SB is completely free from dislocations. The subsequent refined SGs or DCs are surrounded by dislocations within grains (P 2 C 4 and P 2 C 6). Another typical feature is the extinction contours existing in the interior of the grains on position 2, which is attributed to the presence of internal stresses (Ref 35).

Apart from some general features such as GBs, SBs, and DCs during deformation, position 3 possesses several distinctions from positions 1 and 2. In terms of P 3 C 1, initial coarse grain is broken into two grains, and the GBs are curved with the anchoring of dislocation. The curved boundaries and the tilt GBs images contain one single central light band and two adjacent dark broadened bands (P 3 C 1). This observation is very similar to the results reported in Ref 36. Different from the curved grain boundary of P 3 C 1, P 3 C 2 possesses straight and narrow grain boundaries described by white dotted line, which is believed to be in an equilibrium state and are high-angle grain boundaries (HAGBs), formed by continuous dynamic recovery under large strain (Ref 37). P 3 C 4 has a smooth grain boundary that is caused by the coordinated movement of adjacent grains. However, these adjacent grains situate in different elastic stress fields characterized by the strong spreading of thickness extinction contours near the nonequilibrium boundaries. Therefore, it can be easily found that the refined SGs from P 3 C 4 are subjected to uniform elastic stresses. Inspection of P 3 C 6 reveals that there is a grain with a low density of dislocations as marked “A” while other grains contain many dislocations.

Position 4 has the heaviest refinement due to its position on ellipse major axis direction subjected to the heaviest straining. Different from the dislocation morphologies in the above three positions, dislocations motion is more violent in the long strip band. Dislocations pile as dislocation walls to cause the transverse brittle fracture of GBs. Besides, shrinking behavior, referred by white arrows, occurs in some shear zones due to its irregular geometry inducing the stress concentration (P 4 C 1). With further deformation, dislocations are inclined to gather on grain boundaries to cause grain boundaries becoming severely bended and fuzzy (P 4 C 2). The elliptical dislocation edges, surrounded by white stars shapes, may be attributed to the elliptical structure in the die cavity that promotes local grains to produce the corresponding flow. On the contrary, grains from P 4 C 4 and P 4 C 6 consist of the low dislocation density (LDD) zones compared to those from P 4 C 2. Eventually, the grains after six cycles are made up of the grain-subgrain, nonequiaxed, and equiaxed structures, varying in size from 0.2 to 0.5 μm (P 4 C 6).

Conclusions

  1. (1)

    Strain distribution is not uniform, decreasing from the edge to center on the orthogonal planes, and the strain gradient gradually decreases with the increase in deformation passes. Vickers hardness distribution is similar to the strain distribution. Besides, the hardness values in the ND-ED and TD-ED planes are slightly lower than those in the TD-ND plane. The saturated hardness and more uniform hardness distribution can be achieved by increasing extrusion cycles.

  2. (2)

    According to TEM observations, a grain refinement process is proposed for coarse microstructure broken into an ultrafine-grained structure undergoing the formation of the SBs, subgrains, high-angle misorientation grain boundaries, and equiaxed structures. However, different positions own different degrees of grain refinement, grain boundary styles, and dislocation morphology. TEM investigations are in good agreement with the microhardness test results.

Based on the significant improvement in hardness and the reduction in the grain size, ECSEE as one of the SPD techniques has a bright prospect in the future in producing bulk ultrafine or nanostructured materials.