Preamble

The symposium on Ultrafine-Grained Materials VII, which formed part of the TMS 141st Annual Meeting in Orlando, Florida, on March 11–15, 2012, was one of a series of symposia organized every 2 years within the TMS annual meetings. However, this meeting was significant because it marked the twenty-fifth anniversary since the first attempts were undertaken to evaluate the microstructures of materials processed through the application of severe plastic deformation (SPD). As reviewed elsewhere [1, 2], the first detailed scientific studies of SPD processes, and the first incorporating scientific rigor, were conducted by Bridgman [3, 4] in the 1930s and 1940s with classic experiments on the procedure now known as high-pressure torsion (HPT). Later, this was followed by the work of Segal et al. [5] in the 1980s in the development of the principles of equal-channel angular pressing (ECAP).

However, these early studies were devoted almost exclusively to the mechanical characteristics associated with the HPT and ECAP processes and it was only later, with the classic work of Valiev and his co-workers in Ufa, that modern microstructural analysis was used to evaluate the microstructures of the materials processed by SPD and to investigate their significance. This study began 25 years ago, in 1987, and it led very quickly to an extraordinary demonstration of low temperature superplasticity in an Al–Cu–Zr alloy processed by HPT and having an exceptionally small grain size of 0.3 μm. The report describing this study was submitted for publication on September 28, 1987, and the paper appeared in the scientific literature in 1988 [6]. This research, and the subsequent detailed descriptions of experimental results in western journals [7, 8], led to a widespread interest in the processing of metals using SPD techniques and the subsequent development of SPD facilities in many laboratories around the world. This area of research has grown over the last 25 years to the extent that the Ultrafine-Grained Materials VII symposium was easily the largest, in terms of the numbers of submitted abstracts, of the many symposia held at the 2012 TMS annual meeting. It should be noted that new developments in this area include recent reports describing the innovation potential of SPD processing and the opportunities for developing new approaches in SPD technology [9, 10].

It was recognized in the earliest research on SPD that the strain imposed on the disks in HPT processing varied from a maximum at the outer edge of the disk to zero strain at the disk center. This recognition led to several evaluations of the development of microhardness and microstructure across the planes of the HPT disks [11–18] but these measurements were taken exclusively on, or very close to, the disk surfaces and there were no parallel investigations to determine the degree of homogeneity that occurred within the thicknesses of the disks. Investigations into the development of homogeneity or heterogeneity in the axial directions of HPT disks have been undertaken only very recently [19–21] and these investigations, together with new experimental results, form the basis for the present report.

Introduction

Processing by HPT is generally undertaken by placing a thin disk between massive anvils, applying a high pressure and then rotating one of the anvils so that the disk is subjected to torsional straining. There have been some limited attempts to apply HPT processing to larger bulk samples, in the form of small cylinders [16, 22, 23], but most investigations of HPT processing involve the use of very thin disks.

In processing by HPT, it can be shown that the equivalent von Mises strain, εeq, imposed on the disk during torsional straining is given by a relationship of the form [24–26]:

$$ \varepsilon_{eq} = \frac{2\pi Nr}{h\sqrt 3 } $$
(1)

where N is the number of turns of the anvil, and r and h are the radius and height (or thickness) of the disk, respectively. It follows from Eq. (1) that the strain imposed in HPT varies with the position on the disk so that it is zero at the disk center where r = 0 and it reaches a maximum at the outer edge. This significant variation in strain suggests that the microstructures introduced by HPT are probably highly inhomogeneous and it is reasonable to anticipate that they will vary significantly as a function of the radial position within each disk.

Although Eq. (1) implies there will be gross inhomogeneities in the microstructures developed across the disks in HPT processing, early experiments using microhardness testing revealed a gradual evolution with increasing torsional straining towards a reasonable level of homogeneity throughout the disks [11–13]. It was shown also in these experiments that this evolution occurred more rapidly with increasing applied pressure, P, and/or increasing the numbers of revolutions, N, in the processing operation [11, 12]. The general evolution towards a hardness homogeneity within the planes of the disks was later confirmed in many experiments [27–39] and the evolution was explained theoretically using strain gradient plasticity modeling [40].

It follows from Eq. (1) that, although the strain may vary within the sectional plane of the disk, it remains constant along the axial direction within the thickness of each disk. Accordingly, it was only very recently that experiments were undertaken to make microstructural and microhardness observations in the axial direction perpendicular to the top and bottom surfaces [19–21, 41]. The present investigation was initiated to review the information available to date and to describe new results that provide a comprehensive description of the evolution of homogeneity or inhomogeneity within the through-thickness direction. Results for high-purity aluminum are described in the next section and data for the magnesium AZ31 alloy are given in the following section.

Experimental results for high-purity aluminum

Experiments were conducted on high-purity (99.99 wt%) aluminum and some of the results were published earlier [19]. The material was in the form of a rod with a diameter of 10 mm and it was annealed for 1 h at 773 K to give an initial grain size of ~1 mm. After annealing, the rod was sliced into disks with thicknesses of ~1.2 mm and then polished to final thicknesses of ~0.81 mm. These disks were processed by HPT at room temperature (RT) under quasi-constrained conditions [42–44] using an applied pressure of P = 6.0 GPa. The disks were processed for selected numbers of turns ranging from 1/4 to 20 revolutions and the values of the Vickers microhardness, Hv, were then recorded along the diameters of the disks. The measurements were taken at incremental distances of 0.3 mm with four individual values recorded around each separate point at distances from the selected point of 0.15 mm. These measurements were used to plot the variation of hardness across each disk together with the calculated 95 % error bars.

In addition, and in order to provide comprehensive information on the variations of hardness within the through-thickness of the disks, three separate disks were prepared for each condition of HPT processing. These three disks were used to take separate hardness measurements on either the central plane of each disk or on planes within ~200 μm of the upper or top surface and the lower or bottom surface after HPT processing, respectively. In the following descriptions of results, the locations for these three sets of measurements are designated upper, center, and lower, respectively.

For the disks processed through 1/4 and 1/2 turns, the hardness values were also recorded over the total planar sections at the upper, center, and lower positions with measurements taken following a rectilinear grid pattern with a separation of 0.3 mm between each separate point. These measurements were used to plot color-coded contour maps that provide a simple visual representation of the hardness distributions across the disk surfaces. The emphasis was placed on producing color-coded maps for the earliest stages of HPT processing because it has been shown that these early stages are critical in determining the overall behavior during HPT processing [34].

Figure 1 shows the hardness values recorded over the surfaces of the disks in the upper, center, and lower positions for HPT through 1/4 turn in Fig. 1a–c and through 1/2 turn in Fig. 1d–f. The results are presented as color-coded maps with the colors representing values of Hv in incremental steps of 5 as given by the color key on the right: in these maps the parameters X and Y are two arbitrarily selected but mutually perpendicular directions where the position (0,0) corresponds to the center of each disk. Inspection of these maps shows that the hardness distributions are very similar at the upper, center, and lower positions after 1/4 and 1/2 turns thereby confirming a general homogeneity in the through-thickness direction. Furthermore, the close agreement between each plane demonstrates that the HPT processing was successfully conducted without the occurrence of any significant slippage of the disks [45].

Fig. 1
figure 1

Color-coded contour maps showing the distributions of the Vickers microhardness values at the upper, the center and the lower positions of high-purity Al after processing by HPT through a–c 1/4 turn and d–f 1/2 turn [19] (Color figure online)

Full size image

A full set of hardness measurements is shown in Fig. 2 for disks processed through (a) 1/4, (b) 1/2, (c) 1, (d) 5, and (e) 20 turns where the lower lines at Hv ≈ 20 denote the annealed condition prior to HPT processing and the error bars show the 95 % confidence limits: an earlier report provided detailed hardness measurements in the unprocessed material and demonstrated there was no significant variation in hardness prior to processing by HPT [29]. For each plot, separate data are presented for measurements recorded on the upper, center, and lower planes in three different disks. From these measurements, it is concluded that there are no significant variations between the different sectional planes for disks of high-purity aluminum processed through a wide range of different numbers of turns ranging from 1/4 turn to 20 turns. It is also apparent that, although there are significant variations in the hardness values as a function of the positions on the disks in the early stages of processing, these variations are essentially lost for the disks taken through 5 turns and after 20 turns all three disks show excellent agreement and a high level of homogeneity that extends both across each plane and in the through-thickness direction.

Fig. 2
figure 2

Variation of the average Vickers microhardness recorded on the upper, center, and lower planes as a function of the distance from the center of the disk after processing by HPT using an applied pressure of 6.0 GPa through a 1/4, b 1/2, c 1, d 5, and e 20 turns

Full size image

The general variations in hardness which are visible in the early stages of processing in Figs 2a–c are consistent with expectations based on earlier results showing that high-purity aluminum exhibits high values of hardness in the central regions of disks processed through only 1 turn [27]. A detailed description of these variations in hardness in the early stages of HPT processing is beyond the scope of this short report but a description may be found elsewhere [19].

An alternative procedure for evaluating the through-thickness variations in hardness, and therefore in microstructure, is to plot color-coded contour maps of one-half of the cross-sections of disks processed under different conditions. These maps are shown in Fig. 3 and they cover five separate conditions from 1/4 turn (upper) to 20 turns (lower) with the center of each disk lying on the left. Inspection of these plots shows there are no significant variations in hardness in the axial directions in high-purity Al. Up to 1 turn it is apparent that there are regions of higher hardness but these regions extend throughout the disks from the upper to the lower surfaces. At 5 turns, there is a reasonable level of through-thickness homogeneity and after 20 turns there is an excellent homogeneity throughout the thickness.

Fig. 3
figure 3

Color-coded maps showing the microhardness distributions across the one-half cross-sections of high-purity Al disks after processing by HPT through 1/4, 1/2, 1, 5, and 20 turns (Color figure online)

Full size image

The development of homogeneity after 20 turns is consistent with earlier observations, using optical microscopy, where the flow patterns were recorded on the upper surfaces of disks of high-purity aluminum after different numbers of rotational turns [19].

Experimental results for the magnesium AZ31 alloy

Magnesium has a hexagonal crystal structure and therefore the numbers of active slip systems are limited. This means that the processing of Mg alloys is often difficult so that these and other similar metals are generally termed “difficult-to-work” materials [46].

The experiments were conducted using a commercial extruded AZ31 magnesium alloy containing 3 % Al and 1 % Zn. An earlier report documented the hardness values across an AZ31 disk in the unprocessed condition where the average hardness was Hv ≈ 55 and there was no significant variation in hardness along the disk diameter [37]. Prior to HPT processing, the grain size was measured as ~9.4 μm. Disks were prepared for HPT and again these disks were processed under quasi-constrained conditions [42–44] either at room temperature (296 K) or at an elevated temperature of 463 K. All processing was conducted under an applied pressure of 6.0 GPa and further details are given in earlier reports [20, 21]. Lines were scribed on the two surfaces of each disk prior to straining and these lines were used to check for the occurrence of any slippage [45]. Using this technique, it was confirmed that there was no disk slippage in these experiments. Separate disks were processed through 1/4, 1, and 5 turns and these disks were then sectioned perpendicular to the upper surfaces for microstructural observations and detailed hardness measurements.

The appearance of the vertical sections cut through the disks after processing at 296 K is shown in Fig. 4 for torsional strains through (a) 1/4 turn, (b) 1 turn, and (c) 5 turns, respectively [21]. Variations in the grain structures are visible in these vertical cross-sections and there is evidence for the occurrence of different amounts of deformation at different levels between the top and bottom surfaces. This suggests the development of heterogeneities within the microstructure after processing by HPT at room temperature.

Fig. 4
figure 4

Flow patterns visible on the one-half cross-sections of disks of the magnesium AZ31 alloy processed by HPT at 296 K for a 1/4, b 1, and c 5 turns [21]

Full size image

These heterogeneities are revealed more directly by plotting color-coded maps for the one-half cross-sections as shown in Fig. 5 where the maps correspond to 1/4, 1, and 5 turns after HPT processing at room temperature. These maps demonstrate the occurrence of remarkable heterogeneities during HPT processing with higher hardness values adjacent to the lower surface of the disk after 1 turn and with a further enhancement in this heterogeneity after 5 turns.

Fig. 5
figure 5

Color-coded maps showing the microhardness distributions across the one-half cross-sections of disks of the magnesium AZ31 alloy after processing by HPT at 296 K for 1/4, 1, and 5 turns (Color figure online)

Full size image

To check whether these results are also typical of AZ31 disks processed by HPT at higher temperatures, Fig. 6 shows a similar set of color-coded maps for disks processed by HPT at a temperature of 463 K [20]. As in Fig. 5, these maps delineate the development of higher hardness values adjacent to the lower surface of the disks after 1 and 5 turns with correspondingly lower values of the hardness adjacent to the upper surfaces. These results are consistent with Fig. 5 and they confirm the development of a high degrees of heterogeneity in the through-thickness directions of HPT disks of the magnesium AZ31 alloy.

Fig. 6
figure 6

Color-coded maps showing the microhardness distributions across the one-half cross-sections of disks of the magnesium AZ31 alloy after processing by HPT at 463 K for 1/4, 1, and 5 turns [20] (Color figure online)

Full size image

Discussion

The results from this investigation show there are significant differences in the levels of homogeneity achieved in two separate metals in HPT processing where these differences appear to depend upon the crystalline structure of the material.

Specifically, slip processes are easy in high-purity aluminum and the material exhibits excellent homogeneity in the through-thickness direction at all stages of the HPT processing. In the early stages of processing, up to 1 turn, there are variations in the Vickers microhardness at different positions on the planar sections of the disks but these variations are not reflected in the through-thickness direction where, as shown in Fig. 3, there are no significant variations between the upper and lower surfaces of each disk. Furthermore, the microhardness results for high-purity aluminum become essentially homogeneous after 5 turns, as shown in both Figs 2 and 3, and after 20 turns there is a very high level of homogeneity throughout the disk. The results are different for the magnesium AZ31 alloy where the numbers of slip systems are limited and significant microhardness heterogeneity is visible after 1 turn and after 5 turns when processing both at 296 K in Fig. 5 and 463 K in Fig. 6.

These results demonstrate that aluminum and magnesium represent two extremes in terms of the development of homogeneity in the microstructure and the microhardness during HPT processing. These differences between different materials must be considered in any future attempts to develop a universal plot for the hardness variations in pure metals processed by HPT [47].

Very recent experiments have revealed the development of unusual shearing patterns in materials subjected to HPT processing including the presence of local swirls and vortices that are indicative of the development of shear velocity gradients between adjacent positions within the disks [48–51]. However, the significance of these patterns is not yet known and therefore it is not possible at the present time to determine the significance of this effect in achieving homogeneous microstructures.

Summary and conclusions

  1. 1.

    Experiments were conducted on high-purity aluminum and a magnesium AZ31 alloy to evaluate the potential for achieving a reasonable level of homogeneity in the through-thickness direction of disks processed by high-pressure torsion.

  2. 2.

    For high-purity aluminum, the results show there are no variations in the microhardness in the through-thickness directions in any disks processed between 1/4 turn and 20 turns. After 5 turns of HPT there is a reasonable level of homogeneity throughout the disks and the overall level of homogeneity is increased after 20 turns.

  3. 3.

    For the magnesium AZ31 alloy, the numbers of slip systems are limited and the processed structures show extensive heterogeneity in the through-thickness directions after processing by HPT through 1 and 5 turns. In general, there is a tendency for high hardness values to occur adjacent to the lower surfaces of disks when the HPT processing is conducted either at 296 or 463 K.