Microstructures indicating GBS-limited creep

Several microstructural features can be interpreted as providing evidence for GBS in the ice. These microstructural features include: (i) elongated blocky grains (“brick-wall pattern”, Faria et al., 2009) (Figure 4.1 and 6.3), (ii) grain boundaries aligned parallel to the shear plane (Figure 4.1 and 6.3), (iii) SGBs aligned to the grain boundary network (Drury and Humphreys, 1988; Faria et al., 2009; Figure 6.3), (iv) rotation axes of low angle grain boundaries not related to host crystallography (e.g. Fliervoet and White, 1995; Bestmann and Prior, 2003; Chapter 3). The reflective LM image in Figure 6.3 contains microstructural features that can be interpreted as evidence for GBS. This sample is from a depth where high strain rates were predicted using the grain size sensitive composite flow law of Goldsby and Kohlstedt (1997, 2001) in Chapter 4. Another exemplary depth region (2103 m) for which high strain rates were predicted by the flow law modelling study of Chapter 4 was investigated using transmitted LM imaging and cryo-EBSD is shown in Figure 6.4.

The strong c-axes alignment of the grains in the EBSD map (Figure 6.4d), and generally strong c-axes alignment in the glacial ice of the NEEM ice core ((Eichler et al.,

2013; Montagnat et al., 2014; Figure 1.8a and 6.5), shows that strain is mainly produced by intracrystalline dislocation glide (e.g. Azuma and Higashi, 1985; Alley, 1992; Azuma, 1994). The alignment of a-axes (Figure 6.4d and 6.5) indicates that basal slip along the easy slip system of <12̅10>/(0001) is the dominant slip system in this part of the NEEM ice core (Kamb, 1961; Hondoh, 2000). The alignment of a-axes also indicates that plane strain deformation, such as simple shear, at this depth is dominant.

In the example shown in Figure 6.4, the rotation axis of SGB 2 is parallel to the c-axis and the trace of this boundary is normal to the basal plane. This SGB is therefore classified as an N[c] type SGB (Miyamoto et al., 2005, 2011; Weikusat et al., 2011a, b;

Chapter 2 and 3) and was likely formed by an array of edge dislocations with Burgers vector b=(1/3)<12̅10> gliding on a non-basal plane (Figure 3.2 number 2a). SGB 3 on the other hand is closest to a P[c] boundary but deviations of the trace by about 15°-20° and the rotation axis of about 15° suggests a more general SGB type (Chapter 3). Another

interesting feature of this EBSD map is that SGB 4 and 5 and SGB 8 and 9 have the same rotation axis, trace and misorientation angle. The exact rotation axis of SGB 9 is uncertain since the IPF contains only 6 pixels. However, the rotation axis of this SGB is, just like SGB 8, is close to <12̅10>. All four SGBs intersect with SGB 3 at the same angle with an offset of about 30-40 µm between SGB 4 and 5 and between SGB 8 and 9, respectively.

Sliding along newly formed SGB 3 could have disconnected SGB 4 from SGB 5 and SGB 8 from SGB 9 along this boundary. It is possible that both SGB pairs were continuous SGBs that became displaced by sliding along SGB 3 in a dextral shear sense. A sliding boundary would indicate that SGB 3 is actually a grain boundary instead of a SGB, which, based on its misorientation angle (5.5°-6.0°), agrees with the SGB-grain boundary critical angle of 5° found by Weikusat et al. (2011b) and Montagnat et al. (2015). The strain incompatibly between grain I and II and grain II and III, as a results of sliding along SGB 3, likely resulted in the formation of the bulge (white arrow Figure 6.4b) and to the boundary 11 and 12 and could have been responsible for the formation of the small grain between grain II and III.

The rotation axis of SGB 3 could be explained by the activation of multiple slip systems with different rotation axes contributing to the formation, as was proposed for the

‘general’ rotation axes of high angle SGBs (4.0°-6.5°) in the glacial ice (Chapter 3).

However, for this particular example, such a mixed boundary type cannot explain the offset of SGB 4 and 5 and SGB 8 and 9 along SGB 3. The development of a general rotation axes of boundaries that have been formed by subgrain rotation and initiated sliding along that boundary was first proposed by Fliervoet and White (1995) and convincingly shown in natural rocks by Bestmann and Prior (2003). Sliding along newly formed SGB 3 could have resulted in the development of the general rotation axis of SGB 3. With respect to the results from flow law modelling, which predict high strain rates produced by GBS-limited creep at this depth (Chapter 4), the observation of the general rotation axis of SGB 3 could be attributed to enforced SGB formation in grain 1 to enable local GBS.

Sliding along grain boundaries in fine grained bands could also explain the aligned subgrain boundaries observed in Figure 6.3 and SGB 11 and 12 in Figure 6.4. If sliding is

allowed along aligned grain boundaries, then a grain blocking this alignment will experience an accumulation of stress and strain at the triple junction where the aligned grain boundaries meet with the blocking grain. Therefore, the dislocation density in the blocking grain will increase as a result of the strain incompatibility (e.g. Drury and Humphreys, 1988; Bons and Jessel, 1999; Faria et al., 2009). When these dislocations recover into a SGB, the SGB will originate at the location of highest strain incompatibility.

With progressive sliding along the grain boundaries ahead of the blocking grain, the SGB eventually rotates to become a grain boundary and can contribute to sliding along the aligned grain boundaries. This process produces SGBs with a preferred orientation not controlled by the intrinsic slip systems of ice but by the external (local) stress configuration.

These enforced SGBs are produced in blocking grains and could explain the small grain size and flattening in the fine grained bands with aligned grain boundaries compared to the neighbouring coarser grained bands (Figure 6.3). Since these aligned SGBs are not controlled by the crystallography of the host crystal but by strain incompatibility with surrounding grains, it is expected that these SGBs will not have a rotation axis and/or a trace that is controlled by the (intrinsic) slip systems of the host grain. In Chapter 3 it was shown that the glacial ice in the NEEM ice core contains a significant number of high angle SGBs (4.0°-6.5°) with a general rotation axis and/or an oblique trace. SGB 3 in Figure 6.4c is such a boundary with a general rotation axis and oblique trace. SGB 11 and 12 in Figure 6.4 could also have been formed by this process as these SGBs are aligned with the local grain boundary network, although their rotation axes are not ‘general’ rotation axes like SGB 3. Part of the high angle SGBs (4.0°-6.5°) in the glacial ice with a general rotation axis and/or trace could have been formed in a similar way to SGB 3 in Figure 6.4c. These types of boundaries were not found in the Holocene ice where GBS is less likely to occur due to the coarser grain size and interlocking grains (Chapter 4).

It is often stated that a grain size sensitive deformation mechanism such as GBS-limited creep destroys CPO (e.g. Duval et al., 2000; Duval and Montagnat, 2002; Bestmann and Prior, 2003). However, in the glacial ice of the NEEM ice core, the c-axes are strongly clustered into a single maximum CPO (Eichler et al., 2013; Montagnat et al., 2014).

Furthermore, the a-axes are also aligned in this part of the glacial ice (Figure 6.4d and 6.5).

This suggests that basal slip and GBS operate simultaneously at this depth, a mechanism which was proposed by Goldsby and Kohlstedt (1997, 2001). Furthermore, SGB 2 in Figure 6.4c is a N[c] type boundary and likely formed by an array of edge dislocations with Burgers vector b=(1/3)<12̅10> gliding on a non-basal plane. This could be interpreted as GBS, basal slip and non-basal slip operating simultaneously within the same grain.

Premelting is expected to enhance grain boundary processes, like GBS (e.g.

Barnes et al., 1971; Schulson and Duval, 2009). It is interesting to note that the in-situ temperature of the samples shown in Figure 6.3 (255K) and Figure 6.4 (258K) is below the expected temperature threshold for the onset of premelting in ice of 262K (Chapter 5). The high impurity content in the glacial ice (Kuramoto et al., 2011; Eichler et al., 2017) could

lead to premelting occurring at lower temperatures in these samples, which enhances grain boundary processes such as GBS.

During their ice deformation experiments Goldsby and Kohlstedt (1997; 2001) argued for a deformation mechanism where GBS is the accommodating mechanism for basal slip (from here on called GBS-limited creep), and thus GBS may contribute only slightly to total deformation. In case of limited contribution of GBS to the bulk strain rate, intragranular slip accommodated by GBS can strengthen instead of weaken the CPO as was shown by the modelling studies of Zhang et al. (1994). Aligned grain boundaries, similar to the grain boundary alignment in Figure 6.3, were also observed during the deformation experiments on aluminium-magnesium alloys of Drury and Humphreys (1986, 1988).

During these experiments, strain was predominantly produced by intergranular dislocation creep, with only 10-15% of strain being produced by GBS. This shows that, even though the contribution to bulk strain is low, GBS can have a significant effect on microstructural development. A strengthening of CPO caused by the accommodation of basal slip by GBS could explain the exceptionally high CPO eigenvalue of 0.97 in the very fine grained sample in Figure 6.3. Furthermore, GBS-limited creep strengthening CPO could explain why the CPO is often stronger in the fine grained glacial regions than in the coarser grained interglacial regions, as has been shown in many different polar ice cores (e.g. Paterson, 1991; DiPrinzio et al., 2005; Durand et al., 2009, Montagnat et al., 2012; Eichler et al., 2013; Faria et al., 2014b; Fitzpatrick et al., 2014; Weikusat et al., 2017).