Here is the problem. At magnifications above 5000x the images are fuzzy due to some kind of drift. At lower mags the drift is not noticeable. Polaroids do not show the drift. The drift rate is 100-500u per five minutes. I have the SEM manufarturer working on this too. So far I have not gotten acceptable results

Now for the questions. Has anyone had this problem on other image acquisition systems? Am I trying to do something that can't be done due to drift that all SEMs have to some degree?

Thanks for your answers in advance.
Gregory Rudomen
Greg@umic.umic.sunysb.edu

I'm responding to Greg Rudomen's posting about fuzzy digital images on his JEOL 5300.

Greg, can you give us some more details on your problem? It sounds like you are using a slow scan to do your frame averaging since you say it takes 1 min. 20 sec. to collect five frames.

Are you really seeing drift? Drift will generally be blurred in one direction; if the images are just not sharp, that may be another problem.

When you say the Polaroids do not show the drift, are these Polaroids of the analog or the digitized image? If the analog image is okay, something is happening during digitization.

Another item to check is the power and magnetic field situation in your lab. If the drift is more noticeable at low accelerating voltages, you may have a field problem.

Feel free to contact me directly about this. I'm using a JEOL 6300F with a Vision system, and it's no problem to obtain a clean 100K digital image.

Jim Stets
stetsjr@ttown.apci.com

Regarding the apparent drift problem in Greg Rudomen's images:

We have seen similar effects with a JEOL 6100 interfaced to a Voyager EDS/image acquisition system. The effect is also noticeable when using the image capture features of the SEM itself. Slow scans over a minute or two, whether for a Polaroid or for digitizing, do not show the problem, as the digitization occurs only once at each beam location and a small drift would be unnoticeable in terms of seeing a distorted image; however, for elemental mapping using single scans, with long dwell times resulting in acquisitions taking hours, the drift/distortion is extremely evident. Averaging, integrating or other algorithms which entail multiple scans show the diffuseness very strongly. Most frustrating is that the problem is sometimes worse during some sessions compared to others, with no obvious differences in specimens or operating conditions.

This problem was suggested by the SEM manufacturer to be one of charging or poor grounding in our specimens. Not likely, though, as we were observing Si and GaAs crystals when the problem was most noticeable in images. The problems were most notable with these because the magnifications were considerably higher than other routine work done here. Decreasing the beam energy to ca. 1-5 keV did not improve the problem, suggesting that charging plays no role.

The problem is as yet unresolved, but I wanted to add our observations. For the time being, I acquire images using slow scans only. Maybe it has something to do with imaging on every third Tuesday of months ending with the letter "y"...

My opinions only
Bob Keller
keller@arc1.mrd.bldrdoc.gov

I got caught with some drift the other day on an E-SEM. The image was drifting slowly in one direction. I half suspected charging, although the drift was only in one direction and varying the pressure and cutting back the beam did not help.

The problem was eventually tracked back to the modeling clay that I had used to support the irregularly shaped sample. I was not aware that it outgassed fast enough to change dimensions in the SEM. Now I know better.

Warren E. Straszheim
E-Mail: wes@ameslab.gov (or: wesaia@iastate.edu)

Dear Greg, Can you see the image drift at high mag. in TV-rate? Is this perhaps EM field ripple? or mechanical vibration? The image acquisition system can only acquire what the SEM puts out, so this drift will certainly wipe out the benefits of frame averaging. If it takes so long to acquire a frame-averaged image, why not just acquire one long frame? I have a passive system and usually take a 10 sec., 1024X768 single frame, but the image certainly shows up any drift of the SEM stage. This can always be traced to the specimen not being secured well enough. Sticky tabs are often guilty of this. If the SEM doesn't drift, the image is perfect to 50,000X.

One problem I have had in the past is bad vibration problems at 20 minutes past the hour on class days. This is when classes change and two hundred undergrad engineers tromp through the halls. People doing critical hi-res, low kV work know to do it after hours, when the building is quieter and the elevators aren't running.

Mary Mager
mager@pop.unixg.ubc.ca

Dear Greg,

Can you see the image drift at high mag. in TV-rate? Is this perhaps EM field ripple? or mechanical vibration? The image acquisition system can only acquire what the SEM puts out, so this drift will certainly wipe out the benefits of frame averaging. If it takes so long to acquire a frame-averaged image, why not just acquire one long frame? I have a passive system and usually take a 10 sec., 1024X768 single frame, but the image certainly shows up any drift of the SEM stage. This can always be traced to the specimen not being secured well enough. Sticky tabs are often guilty of this. If the SEM doesn't drift, the image is perfect to 50,000X.

One problem I have had in the past is bad vibration problems at 20 minutes past the hour on class days. This is when classes change and two hundred undergrad engineers tromp through the halls. People doing critical hi-res, low kV work know to do it after hours, when the building is quieter and the elevators aren't running.

Mary Mager
email: mager@unixg.ubc.ca

Gregory,

There are 3 kinds of averaging techniques used in SEM digitization: frame averaging, line averaging and pixel averaging. For most of "built-in" acquisition system, the frame capture is used, so only support frame averaging. The drawback for using frame averaging is to produce a fuzzy image as long as drafting exist. Pixel averaging is mostly suitable to overcome this situation.

Ya Chen
Email:YChen@macc.wisc.edu

If the image drifts significantly between frame times, pixel averaging such that the total acquisition time remains the same as for frame summing will just substitute spatial distortion for blurring. The image will be nice and crisp, but wrong. Suppose the drift is perfectly horizonal, left-to-right. The digital image will be sheared so that vertical edges slant at a diagonal.

Are you sure it's not already doing pixel averaging too? 16 seconds per frame sounds awfully long to me otherwise.

Unless there are other factors you haven't mentioned, I'm at a loss to explain a "drift" that only happens when the image is acquired electronically. If a Polaroid taken with the same sample and beam conditions shows no drift, the problem pretty much has to be with your imaging system.

Are you sure it's really a drift, that is, continuously in the same direction? You can get quite a bit of blurring from the effects of 60-hz power line noise coupling into the imaging signal. Frame averaging can mask the characteristic pattern of 60-hz interference, if the digital image collection is asynchronous with the line frequency. Check carefully for ground loops between your imaging system and the SEM.

Rick Mott
rick@pgt.com

From the various symptoms that have been described I think that Rick Mott's last suggested source of this problem is most likely to be correct. In conventional scanning the scan is usually synchronized to the line frequency to minimize distortions resulting from stray fields. When using an external scan generator for acquiring digitized images some of the scan speeds will be asynchronous with the line frequency. If there are stray fields in the vicinity of the SEM these will then cause some sort of image distortion on the screen, often showing as a uniformly jagged appearance of what should be straight edges. If one then averages a number of such frames it could lead to what has been described as a fuzzy image.

In order to establish if this is the problem it will be necessary to test for stray A/C fields around the SEM. These usually arise from sources such as large power supply cables running under the floor or in a nearby wall, but as Rick Mott has suggested, could simply be a problem of bad or wrongly routed earthing. If one doesn't have access to a proper magnetic field testing coil an old EM lens coil connected to an oscilloscope, although not calibrated, nevertheless does very well for comparing one area with another for frequency and magnitude of fields.

Good luck!
Robin Cross
eurc@giraffe.ru.ac.za