EBSD Optimises pFIB of Steel

Introduction

Focused ion beam scanning electron microscopes (FIB-SEM) are fantastic machines, capable of preparing TEM lamellae, cross-sections and performing 3D analysis of samples. Due to the limited coherence at high current, gallium (Ga+) liquid metal ion sources (LMIS) can only mill samples to maximum dimensions of 10s of microns in all directions, and as such place a limit on what can be analysed. However, their younger sibling, the plasma FIB (pFIB) can mill much larger areas/volumes, but cannot be expected to work in the same way as Ga+-FIBs. One key difference is that the ion beam in pFIBs can achieve much higher currents (multiple µA) than can be achieved in a Ga+-FIB, but the beam is generally much wider even at low currents, resulting in a poorer spatial resolution. There are also differences in beam interactions – e.g. if aluminium is milled in a Ga+-FIB, Ga+ ions will implant at grain boundaries, whereas this is not the case in a pFIB.

When moving from a Ga+-FIB to a pFIB, one of the first things that it is tempting to do is generate much larger cross sections, as shown in figure 1. The figure shows a cross section carried out in the pFIB on a 316 steel. A 60 nA Xe+ beam was used to cut a 50 µm cleaning cross section (CCS), which gave an overall glancing depth of ~200 µm and a usable depth of ~60 µm – although this is partly a function of cut depth and the platinum (Pt) layer thickness.

Figure 1. pFIB cross section of austenitic steel

After creating a cross section, we can then proceed to perform some analysis on the sample. One such analysis that is particularly revealing is Electron Backscatter Diffraction (EBSD). This approach can give us information on the crystal structure of materials – an example of this is shown in figure 2. This data was collected without removing the sample from the microscope by making use of a pre-tilted sample holder. In figure 2, the EBSD data clearly shows that a phase transformation has occurred, with the top ~40 µm transforming from face centre cubic (FCC) to body centre cubic (BCC); the data also shows there is an orientation change. This is a very specific issue that effects austenitic steel polished by Xe+ and Ga+ ion beams. Studying the literature there are 3 plausible causes of this phase change:

1. Elemental effect of the beam in the matrix, due to ion implantation
2. Localised heating induced by the ion beam
3. Residual stress introduced by the polishing process – this model fits well as mechanical polishing is also observed to cause the FCC to BCC transformation

There are two possible ways to mitigate this transformation in stainless steel when milling in the pFIB - one being low kV milling; however, this is a very time consuming approach as the sputter yield is decreased. The alternative is to use XeF₂ aided milling, which has been shown to improve both the sputter yield and surface finish.

Figure 2. Images of pFIB cross section with a. band contrast, b. IPF, c. electron image and d. phase colour. With the bottom right phase colour image showing a clear change in phase, from the blue FCC to red BCC at the top of the crosssection due to the ion beam

Experimental

For this work, a test matrix was chosen which included literature-based implantation values as a benchmark but also higher currents to simulate typical 3D prep. This matrix is shown in figure 3 and is a matrix of probe current vs milling depth/dose. For each point in the matrix a 20 x 20 µm box was used for pFIB milling, using a 90% overlap and 1 µs dwell. Two test matrices were carried out, one as a benchmark to compare to and the other using XeF₂. Both tests were then analysed using EBSD, to determine the effects on the orientation and phase.

This work was carried out using a ThermoFisher Helios G4 pFIB with a Symmetry EBSD detector.

Results

Test matrix 1 can be seen in figure 3, and shows almost all the boxes having a phase transformation, which was observed by a colour change in the phase map, from blue (FCC) to red (BCC). This test matrix also shows that different orientations can effect the transformation. For the second test matrix, 1% XeF₂ flow was used, with the results being shown in figure 4. The results show that the XeF₂ mitigates the phase change with the majority of the boxes retaining their original FCC phase and orientation (most of the boxes remain blue with relatively little red seen in the D-phase image). However, at high currents the XeF₂ is flowing for such a short period of time, it appears to have little to no effect. This is best observed in the difference between working at 60 nA, and milling for 250 nm, where there is minimal phase change, whereas when milling for 10 nm, there is a complete phase change. This would indicate the flow time has an effect on the mitigation of the phase transformation, and there may also be some differences in grain susceptibility.

Figure 3. Test matrix 1 showing the effect of different milling parameters on grain structure with a. backscattered electron (BSE), b. band contrast, c. IPF and d. phase. The inset images show a high resolution of the corresponding highlighted test box shown in b.

Figure 4. Test matrix showing the effect of using 1% XeF2 flow on the grain structure with a. BSE, b. band contrast, c. IPF and d. phase

Testing the Solution

As shown previously with EBSD, XeF₂ aids in mitigating phase transformation of 316 steel. This solution can be tested on a cross-section prepared via pFIB. An initial test was carried out using 60 nA, with a 50 µm cut depth, a beam overlap of 85% and 1% XeF₂. The results can be seen in figure 5.

Figure 5. Images showing the initial cross-section test using 1% XeF2 and 85% beam overlap, showing (left) the electron image, (middle) the band contrast, and (right) the phase colour, with red being BCC and blue being FCC

The initial results show a partial phase transformation of the FCC to BCC (blue to red). This may suggest that the beam overlap is also influencing the phase transformation. In order to check this, a second cross-section test was carried out using a beam overlap of 25%, plus running 30 cycles of 250 nm thick, 50 µm deep cuts (this was carried out using the auto slice and view, ASV). The results of this second test can be seen in figure 6. The results show no phase transformation, even after 30 cycles of milling, which would suggest that both the XeF₂ and the beam overlap influence the phase transformation of the 316 steel.

Figure 6. Images showing the second cross-section test using 1% XeF2 and 25% beam overlap, showing (left) the electron image, (middle) the band contrast and phase colour after 1 cycle, and (right) the band contrast and phase colour after 30 cycles, with blue being FCC

Conclusion

The phase transformation discussed here can only be observed via EBSD, as it is a change in the crystal structure and not a chemical change, which could otherwise be observed in an electron image or via EDS. This is evident in figures 1, 2 and 5, where the electron images do not show any changes in the cross-section. However, with EBSD (as shown in figures 2 and 5), the phase transformation can be observed. This is also evident in figures 3 and 4, where both the BSE and band contrast images both show areas where something has occurred, but it is unclear what has happened. In these cases, the EBSD data (shown in the IPF and phase maps), show for each part of the matrix whether a phase change has occurred or not, via a change in colour. Figure 4 shows that both untransformed and transformed regions (shown in the IPF and phase maps) appear nearly identical in the BSE and band contrast images.

In conclusion, this work shows that using XeF₂ during milling will aid in mitigating the FCC to BCC phase transformation in 316 steel. The ion beam overlap also influences this phase transformation. However, the optimum flow rate and overlap are dependent on beam current, and size of the cut being made. Additionally, this process will require optimisation for different pFIB systems, as the flow and loop time will be dependent on gas injection system (GIS) needle position.

Acknowledgment

We would like to thank Dr Stuart Robertson, from the Loughborough Materials Characterisation Centre (LMCC) at Loughborough University, for the data and allowing us to write this application note with them.