For my experience, every time I prepare a TEM lamella from a bulk Al sample it starts bending when its thickness is below 100 nm.
It cannot be charging because there is electrical continuity through the Pt and the Omniprobe needle.
One reasonable explanation would be partial melting, but is that possible when using as little as 0.28 pA and 16 kV? Al has a melting point around 930 ºK..
I've made the same experience with several materials. I don't know the exact reason for sure, but I think it is simply the lower mechanical stability due to the decreasing thickness in combination with the handling of the sample with a still high-voltage Ga+ beam. You could try to 1) reduce the height of the lamella (just leave a bit more material at the front side during thinning) or 2) to reduce the voltage. A lower voltage is known to reduce the bending effect to my experience, though there is no guarantee. It's also documented in this paper: M. Schaffer, B. Schaffer, Q. Ramasse, Sample preparation for atomic-resolution STEM at low voltages by FIB, Ultramicroscopy, 114 (2012) 62-71. I hope this works!
Thanks Martin, I'll try both tricks, see what happens...
By the way, what I found even more surprising was that one way of decreasing bending was to stop acquiring a live electron image while millling!
When I did, the bending stopped, but as soon as I started again taking a live image the bending was back. Could it be some interactions between electrons and Ga ions?
Good reference Martin. Dear Alessandro, I don´t think the problem is the temperature increment. I feel that the key is the distribution of the stress in the thinner zone, it shouldn't be abrupt. You could attach the lamella in two points at V post of the grid, or If you attach it as a flag, the lamella should be milled with staircase shaped as Schaffer et al. recomend.
Regards,
Dear Jesùs y Alireza,
I also thought about strain relaxation, but I don't think this is the explanation for the following reason:
the samples I normally prepare consist of a bulk Si substrate on which various Ge/SiGe layers are grown; the total thickness of these layers do not exceed 4-500 nm and the lamella depth is normally a few microns, therefore it is difficult to imagine that relaxation effects can be so strong as tio affect the whole lamella.
Furthermore, the lamella tends to bend in the opposite direction with respect to the side being milled, in other words if you're milling the lower side (in the ion image) it bend upwards and viceversa.
This effect seems difficult to explain with relaxation!
One will always implant Ga into the sample during sputtering of the specimen and this will casue it to expand and bend away from the sputtering side. You could try with a more shallow insidence and lower kinetic energy to reduce the level of implantation.
In PIPS/CP machines the Ar sputtering is usually done with very shallow angle from both sides to minimize these effects.
Krister, you're right, but then why does it only happen for a few metals (Si/Ge alloys and Al, in my experience?)
I guess it is related to implantation depth and the sputtering yield, together with the overall stiffness of the specimen.
Atoms with lower mass than Ga, such as Al and Si, will be less efficient in stopping the Ga ions.
Hi,
Here is one promising method I use to avoid this bending problem while I am thinning down a TEM lamella (it works with Al alloys):
I do not mill the whole side of the lamella, I rather mill 4 patterns, two at each side, and leaving some distance between the patterns at the same side. I made a quick sketch to show you what I mean.
The numbers show the sequence of pattern milling. In this way you get slightly smaller thinned area for your TEM analysis from the whole lamella, but the lamella stays stable without bending. Of course when going from one pattern to the next one you have to tilt the FIB stage ~2.6 degrees. I hope this tip would help.
Cheers, Hisham
Very interesting Hisham; i will try it.. Although it seems a bit time consuming, but if ti solves the problem it`s worth the extra time!
Thanks a lot for your help
If you are just mounting the sample on one side to a TEM post and it's long ( say 25+ microns) and thinning down the entire protruding length below 100 nm then you will find that most materials will bend. There are a few reasons why this happens most have been mentioned already. Just mount it in the V of a omniprobe or similar grid attaching both ends. this will help considerably. If it still bends then when thinning don't mill the entire depth, leave a thicker bar running along the bottom for the full length. This will also work if mounted to a single post, even better to leave a bar up the side in the latter case as well.
Hi Hisham,
thanks so much for your sketch, I just saw it and it's really clear! I will try it, although I must say my question is more for the sake of knowledge than out of a real need, since I always managed to ge ta decent lamella in the end.. But still , it would be nice to suffer less!
Steven, thanks for the advice, I'll try it for sure!
- Badges
- Science method
- Similar topics
- Biological Science
- Histology
- Staining