Hello everyone,
I routinely do tomography on 90nm resin sections (HM20 lowicryl) of embedded Arabidopsis root tips collected on forever filmed slot grids.
With one block in particular I am having the following issue :
When I acquire images, the image seems distorted in one direction. Astigmatism comes to mind at first, but their is none. It's only when I decrease the integration time of the camera (4k x 4k eagle bottom mounted camera) that the image seems to go back to normal. Unfortunately the electron dose is also decreased thus embedding on the quality of the image (which I can compensate by focusing the beam more).
My second suspicion was that the sample was drifting but it appears not to be the case.
Have any of you already experienced something similar ? I am thinking of maybe excess charges accumulating on my grid. If you do, how have you fixed it, other than switching blocks (only have one in this condition and the preservation is sadly very good :-( )
Thanks,
Regards
Hi William,
have you had this problem only with one particular block. If you cut different sections of the same block do you get the same?
I suppose that you have tried to redo the reference image on your camera. Perhaps it needs to be repeated with more counts?
Are your sections on slot-grids?
If it is charging perhaps you can try to cover the sections with an additional carbon layer.
Does increasing settling time (illumination time) prior acquisition helps?
It would be helpful to see the distorted image.
Best wishes
Hey Petr, thanks for answering so quickly.
I think it is related to the block somehow because yesterday I acquired tilt series on different blocs and it went just fine. For the block in question, I have cut multiple sections a few days ago and yesterday as well : the first time I screened it prior GP coating, it looked beautiful, no problem. Then when I took it back for tomography after coating, I started having the problem on the SAME section. I should suspect the coating process too (with the gold particles).
Now my coating procedure doesn't seem to produce such artifacts usually so I really don't know what to think.
Concerning the reference image, I haven't redone it but again when I acquire tilt series on other sections from other blocs, no prob.
Yes my sections are on slot grids
I was thinking of glow-discharging them but indeed I could carbon coat them. I recall you mentioned that at the EMBO workshop this summer actually. I should carbon-coat my slot grids prior to collecting the sections right ? Could the process harm the former film ?
Increasing the settling time does nothing, tried it.
I took comparison images at different integration times, I will put them together tomorrow and attach it.
What I will try tomorrow is to acquire tilt series on non coated GPs.
FYI : I changed my GP formula. Before I used to coat with old GP-antibodies (5nm) but now I mix 20% BSA with a solution of 5nm fiducial gold (1:1 ratio), aliquot at 1mL and freeze them. It seems to do the trick.
It's a pain, I see exactly what I need (on the fluorescent screen) to finalise the last figure for my paper but I keep having problems with this same condition :/
Regards
I think that acquiring a tilt series without GP is a good idea! You can use patch-tracking to align it in imod. Hopefully that will help.
You do not have to carbon coat the sections but in this particular time it might help.
I would recommend to use protein-A-gold as fiducial markers. I use the one from EMS. I think that BSA should be good it should prevent clustering if it is a uncoated colloidal gold.
Yes, please send your pictures! Good luck with your publication!
Hi
I hate slot grids. But should work. While bars could blind some regions Quantifoil 200 mesh S 7/2 are great, so stable and reliable. Carbon bars are transparent, so you could acquire tomograms.
Probably you experienced some charging effects, coming from the resin or carbon foil. Beam tilt problems could also create weird images but then you could check just performing a FFTs of the image corners. They should present different astigmatism behaviours.
Hello, thanks for the replies.
Petr, I tested before and after carbon coating the grids. It's hard to tell the difference, if their is an increase in quality on the carbon coated grids, it is only very slight.
I would like to use PA-gold but the facility is reluctant for letting me use it for such purpose. And yes the BSA mixed with the fiducials definitely makes a change. Fiducial solution alone and you get enormous clusters in some spots and none in the rest of the sample. BSA gold spreads them nice and clean. I will send you the pictures in private.
Francisco, Why do you slot grids ? I actually think they are very good because of as you say, no region blinding by the copper bars. However they are very fragile and I need to take extra care, especially when coating them with the fiducials.
Having said that, I have never heard of the quantifoil grids, I will look into it, it's awesome if the bars are electro-lucent !
Concerning the charging effect I think that indeed that's what is happening. Especially since I tried turning down the voltage a notch (120 --> 100 kV), this distortion effect was strikingly not as strong. So it has to do with the physics, it's definitely not a technical issue with the software nor the camera. The resin in this block (although it's the same as usual) must induce electron accumulation more than usual.
I usually check all the direct alignements of the TEM before acquiring so the pivot points are well aligned, therefore I have no distortions having to do with the beam tilts .
What do you call "performing a FFT of the image corners" exactly ?
- Badges
- Science topic
- Similar topics
- Filing