[3dem] Newbie question - Defocus mistmatch

[Bill Tivol]

On Oct 26, 2008, at 7:34 AM, rg2502 at columbia.edu wrote:

> I assume that you are using the low dose program on your tecnai when  
> you say that you use the series option.  One cause of the defocus  
> mismatch is that you are using a different magnification to focus  
> than the one you are using to take the exposure. The large  
> descrepancy in resulting defocus is likely related to the fact that  
> you do not have a parfocal lens series.  A way to check this is to  
> go to the magnification that you use to focus and check what the  
> objective lens current readout is.  Now change the magnification to  
> what you are using for exposure and check what the value is there.   
> My guess is that they differ by the amount that you observe.  There  
> is a step in the alignment procedure where you can correct for this  
> (right now I can not recall exactly where, but you should be able to  
> find it out from FEI).  The second reason that defocus settings can  
> go astray ,assuming the grid is flat and you are consistently at the  
> eucentric plane, is the calibration of the difference in lens value  
> as compared to what defocus it gives you.  This value is set in the  
> registry and generally is only set by the service engineer with help  
> from the factory.  I hope that helps.  These are the main reasons  
> for these inconsistencies that are related to the software.   
> Shixin's suggestions are related to the electron optics.  Good luck.
>
> Quoting shixinwang at micron.com:
>
>> The defocus value reading from the scope is normally not the real
>> distance along the optical axis. It is proportional to the Z value,  
>> but
>> the ratio is normally not 1.
>> You may run a test to roughly calibrate the defocus value against the
>> sample height Z (by moving sample up and down). Or you can use a  
>> tilted
>> flat sample: focus it a one location, move to another location (with
>> known distance from the first location), set the focus again. The  
>> real
>> height change is to be calculated from tilt angle and the distance
>> moved. This will give you the ratio to calibrate the focus value.
>>
>> Is the power spectrum a good way to calculate defocus value? I am not
>> sure about this. Theoretically, it should work for a thin amorphous
>> sample, to a certain defocus range. In your case, are you trying to  
>> use
>> it to measure the sample thickness? It won't work. Because it ignores
>> the thickness of the sample (correct me if I am wrong on this). The
>> defocus reading from the scope won't work either, because your sample
>> cannot be that thick.
>>
>> From: 3dem-bounces at ncmir.ucsd.edu [mailto:3dem- 
>> bounces at ncmir.ucsd.edu]
>> On Behalf Of Eduardo Sanz Garcia
>> Sent: Friday, October 24, 2008 6:17 PM
>> To: 3dem at ucsd.edu
>> Subject: [3dem] Newbie question - Defocus mistmatch
>>
>>
>> Dear 3DEM users,
>>
>> Has anyone observe a mismatch between the defocus values reported  
>> by the
>> electron microscope software and the real defocus values computed  
>> from
>> the powerspectrum?
>>
>> Example of a pair of micrographs:
>>
>> First micrograph (this one is the reference):
>>
>>
>> *	Reported defocus value by the microscope: -2 microns
>> *	Defocus value computed from the powerspectrum: -2.2 microns
>>
>> I use the "series" option to change the defocus and take the other
>> micrograph.
>>
>> Second micrograph:
>>
>>
>> *	Reported defocus value by the microscope: -4 microns
>>
>> *	Defocus value computed from the powerspectrum: -2.6 microns
>> (here is the problem, it should be closer to -4 microns).
>>
>> We use a Tecnai F30.
>> Is this a problem of  hysteresis?
>>
>> Thank you very much.
>
Dear Edwardo,
	How are you setting the defocus?  If you are finding exact focus by  
adjusting a live FFT until the central disk is maximally expanded,  
then resetting the defocus value to 0 and turning the defocus knob to  
-2 or -4 um, then I agree with Bob's explanation.  On the other hand,  
if you are using the autofocus function and either choosing 0 defocus  
and then resetting the focus knob to the desired defocus, or inputting  
that defocus into the autofocus function, there may well be a  
difference between what the auto function gives you and what the true  
value is.  If you choose 0 defocus, run autofocus, reset defocus to 0  
and rerun autofocus, you will usually find that the program will find  
a different value for focus, and if you continue to run auto focus the  
value found will converge after a few times to a consistent focus.  It  
is very possible to get inconsistencies between true focus and what  
the auto focus finds the first time that are of the magnitude you show  
above.  It is also possible that there is a systematic difference  
between what the auto focus gives and what the value should be, which  
can be a result of miscalibration or drift in the calibrations.  If  
your alignments and calibrations are all good, it can still be the  
case that rerunning auto focus will be necessary to reach a consistent  
value.
	Shixin, measuring the defocus from the power spectrum is the best way  
to find the defocus.  Of course, if the specimen is 0.5 um thick, the  
defocus at the top of the specimen will differ from that at the bottom  
by 0.5 um.  In the case of a thick specimen, the power spectrum will  
be a mixture of power spectra from various heights within the  
specimen--the FFTs add, then the power spectrum is the square of the  
sum of the amplitudes from the different heights, weighted by the  
amount of scattering at each height.
Yours,
Bill Tivol, PhD
EM Scientist
Ultrafast EM Facility
Noyes Laboratory, MC 127-72
California Institute of Technology
Pasadena CA 91125
(626) 395-8833
tivol at caltech.edu

-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mail.ncmir.ucsd.edu/mailman/private/3dem/attachments/20081027/5c26d278/attachment.html