IEBL weekly progress report
Student Name: Renjie Chen
Project I: In situ TEM Study of Ni-InGaAs Reactions
Week of 09/02/2013 – 09/08/2013
1. TEM sample preparation
In our last discussion, I faced some problems with the Cl-ICP attacked Ni lines. The Ni lines were defined first, followed by HSQ writing. However, after Cl-ICP etch and consequent HF dip, the Ni lines became porous and were even totally removed in some places.
Figure 1: Ni lines were attacked by Cl-ICP and HF dip. Ni lines became porous in some places (a) and were totally removed in some other places (b).
We pointed out two attempts to solve this problem: 1) write the HSQ layer fully covering the Ni lines for protection; 2) For the already damaged Ni lines, write another layer Ni on top of them.
Because the EBL system was still down and new samples could not be prepared, I was using the NPGS-EBL system to write another Ni layer on top of two old samples:
Sample I: with backside Ni/SiO2/HfO2 layers
Sample II: already got the Ni/SiO2 removed from backside
For sample I, after writing the new Ni layer and removing the Ni and SiO2 in HF/HNO3/H2O (1/5/50) solution, the Ni lines are again been attacked and became porous (Fig 2), probably because the solution went through some holes or cracks on the HfO2 to the front side even though the front side is protected by PMMA.
Figure 2: Top surface of sample I after Ni/SiO2 removal from the backside.
Since there’re still some remaining Ni, so I did the RTA anneal (250’C 2.5h) for this sample. However, there’s no diffusion was found as our previous observation. Another 20min anneal at 300’C was performed, and there’s still no diffusion found. I think it’s maybe because the HSQ or PMMA residuals are not fully cleaned underneath the Ni layer.
Figure 4: No diffusion was found after 250’C 2.5h RTA anneal (a)-(c) and another 20min 300’C anneal (d)-(e).
For sample II, the Ni/SiO2 was totally removed from the backside, and the atomic stacking of InGaAs and HfO2 can be seem from TEM.
Figure 5: TEM images of sample II after Ni/SiO2 removal from the backside.
Then another Ni layer was carefully written on this sample. After lift-off step, the fins were cracked with the very thin HfO2 layer (10nm) probably due to the wrapping of Ni film during lift-off.
Figure 6: Sample II after another Ni layer writing and lift-off.
I then tried 2.5h 250’C annealing on this sample, it still seems that no diffusion had happened.
Figure 7: Sample II after 250’C 2.5h annealing.
I haven’t figure out the reason why there’s no diffusion in this sample yet, I’m planning to fabricate some more new samples when the EBL is back to condition.
2. Ni-InGaAs Bonding:
Last time we also discussed to try the Ni-InGaAs bonding on TEM grid, because sometimes the Ni cannot be removed completely from the backside. If using the Ni-InGaAs bonding on the TEM grid, there’ll be no Ni on the backside except some post-deposited SiO2 as supporting layer, which can be easily removed by HF. However, it’s not so successful as the edge of the window is easily attacked by the HCl etching.
Figure 8: Ni-InGaAs bonding on the TEM grid after InP removal.
Hence, currently it seems only the Ni silicide bonding works well for this process. In order to remove Ni completely from the backside, I did the test on the planar film. It seems that with various etching recipe, diluted HNO3/HF solution is most efficient, because the beneath Ti adhesion layer is fast dissolved in HF. So I now use the HF/HNO3/H2O (1/5/50) solution, and it can also remove the SiO2 from the backside, which saves a F-ICP step.
Plan for Next Week:
- Once the EBL is back to condition, prepare more TEM sample for test.
- Follow Daisy to learn writing Yoontae’s samples before she leaves.