SerialEM Note: More About Z Height

Author

Chen Xu

Contact

<chen.xu@umassmed.edu>

Date Created

Dec 12, 2017

Last Updated

Apr 08, 2021

Abstract

This is the first of a series of SerialEM notes I have wanted to write for a while. An application of using SerialEM, even a simple one, could be very useful and “handy” for a SerialEM user. I try to give more explanation for what I did, rather than to just present plain lines of codes (yes, SerialEM scripting code) so that it can be helpful for a SerialEM user, especially a new comer to understand better how SerialEM works.

Quickly and accurately moving specimen to eucentric height is a frequently needed task. Everything is going to be easier if specimen is at eucentric height and objective lens at eucentric focus. I wrote a little document before how to use tilted-beam method to do this using SerialEM “SerialEM HowTo: Positioning Z”. In this note, I give you an improved version and hopefully it is easier to use and more robust too.

Background Information

SerialEM has built-in task function to do eucentricity using stage-tilt method. It is robust, but slower than beam-tilt method. Beam-tilt method is opposite to autofocus function. When specimen is at eucentric height and objective lens at eucentric focus, the tilted beams produce precise overlapped image pair. This is how autofocus works. This principle can also be used to adjust specimen height, because if we preset objective at eucentric focus, cross-correlaton between two tilted beam images also provides information how much specimen is away from eucentricity. Specifically, beam-tilt method used here does a few things:

• it sets scope objective lens to eucentric focus value

• it measures the defocus value for current specimen height using tilted-beam image pair,

• it then changes stage position to that reported value but in opposite direction,

• it moves stage to determined direction and distance,

• and it iterates until the reported defocus value is close enough to zero.

The beam-tilt method works very nicely most of time and it is pretty quick. However, there are couples of things making it less perfect. First, the signal becomes very weak when stage is already close eucentricity. We all know the contrast is the lowest when focus matches z height. We can use focus offset to increase the contrast, but non-linearity property causes some inaccuracy. The calibrated standard focus value could also change a little with time and scope condition. All these together makes it less robust.

When we use SerialEM Low-Dose mode, we often give large focus offset such as -200 microns to View area (I call it View beam) to make the View image good contrast. If we can use this large defocused View beam to obtain tilt-beam pairs for measuring defocus value accurately, that would be ideal.

Z_byV2 Function

The function code is below.

Function Z_byV2

Function Z_byV2 2 0 iter offset
Echo ===> Running Z_byV2 ...

# for defocus offset of V in Low Dose, save it
GoToLowDoseArea V
SaveFocus

## set objective lens
SetStandardFocus 0
ChangeFocus $offset

## Adjust Z
Loop $iter
   Autofocus -1 2
   ReportAutofocus
   If ABS $repVal1 < 0.3
      Echo Close enough, break...
      Break
   Endif
   Z = -1 * 0.72 * $repVal1
   Z = ROUND $Z 2
   MoveStage 0 0 $Z
   echo Z has moved --> $Z microns
EndLoop

## restore the defocus set in V originally
RestoreFocus
EndFunction

The real difference between this and previous version Z_byV is an additional line inserted after SetEucentricFocus:

SetStandardFocus 0
ChangeFocus $offset

These two lines will set objective lens to a specific strength; the large defocus offset also offers good contrast. A special thing here is to measure defocus value using a LD_View beam. The advantage of this command is to use high contrast image.

Autofocus -1 2

This function should be called in script like this way:

CallFunction MyFuncs::Z_byV2 3 -288.32

Obviously, the -288.32 is to pass to variable $offset in the function, and 3 to iteration variable $iter.

Now question is how to determine this offset value for accurate Z height for and under current scope condition.

Find the Offset Value using Script FindOffset

If we found the good “offset” value, it will be good for some time, at least this session. So this like a short term calibration. Here is how to find it:

• Adjust specimen to Eucentriciy, using FEI interface tool or SerialEM task function

• run script as below.

FindOffset

ScriptName FindOffset

# script to find proper offset value to run Z_byV2
# assume speciment is ON the eucentricity

## Eucentric Z
##
#Eucentricity 3
ReportStageXYZ
Z0 = $repVal3
#Z0 = 187.81

SetCameraArea V H
ReportUserSetting AutofocusBeamTilt BT
echo BT = $BT
SetUserSetting AutofocusBeamTilt 1.6

## now find the offset
# for initial offset, get a close value from current setting
ReportUserSetting LowDoseViewDefocus
offset = 0.72 * $repVal1   # or
# offset = -153            # some starting value from previous run

Loop 10
   CallFunction MyFuncs::Z_byV2 1 $offset
   ReportStageXYZ
   Z = $repVal3
   diffZ = $Z - $Z0
   echo $diffZ
   If  ABS $diffZ < 1
      offset = ROUND $offset 2
      echo >>> Found "offset" is $offset
      echo >>> run "Z_byZ2 $offset"
      Break
   Else
      offset = $offset + $diffZ
   Endif
EndLoop

NewArray X 1 11
NewArray Y 1 11

temp_offset = $offset - 10

Loop 11 i
   CallFunction MyFuncs::Z_byV2 1 $temp_offset
   ReportStageXYZ
   Z = $repVal3
   diffZ = $Z - $Z0
   Y[$i] = $diffZ
   X[$i] = $temp_offset
   temp_offset = $temp_offset + 2
EndLoop

LinearFitToVars X Y
echo $repVal1 $repVal2 $repVal3 $repVal4

real_offset = - $repVal3 / $repVal2
echo =====> $real_offset

SetUserSetting AutofocusBeamTilt  $BT
RestoreCameraSet

It uses function Z_byV2 to see which offset value to recover the Z height determined early by other method. It first find an offset value that recovers Z height within 1um(you can define 0.5), then it uses a fitting method to refine this value to make it more accurate. If this script runs and gives offset value as -153.51, then you should use the function with this value.

Note

This offset value changes when V beam size changes. Therefore, it makes sense to do this “calibration” of finding offset value after all the Low Dose area conditions are set and fixed. With the “good” offset value that gives good results, the program works very reliably, if the V beam doesn’t change. For example, on our Krios, the V beam (called Low Dose area V) illumination area stays the same, the script works very well.

CallFunction MyFuncs::Z_byV2 3 -153.51

It will move stage position to Eucentric Z height, almost magically!

Note about Damping Factor

You might have noticed I used 0.72 in the value of Z movement:

Z = -1 * 0.72 * $repVal1

This is to compensate the non-linear behavior of autofocus measurement, with the condition of large defocus offset used. For example, when the stage Z position is -100 microns off from the eucentric height, the autofocus measurement gives something like -136 microns. Therefore, using a proper damping factor (100 / 136 ~ 0.73 here) can make the Z movement more accurately to the target. Since this is a non-linear behavior, this damping factor changes with Z. For example, when Z is off very little, say 5 micron, the factor can be larger like 0.85. One would naturally try to find the curve so to use a more accurate damping factor value in interpolating fashion dynamically. However, if you think about backlash of stage movement, it is the best to avoid any overshoot. Using a single, slightly smaller value could help to keep stage move with backlash corrected when iterating a few times. 0.72 is found to be a good number in our situation.

What exactly the damping factor value should you use? I suggest you move your stage 200 microns away, and you calculate the the ratio of 200 to autofocus measurement value $repVal1 after ReportAutofocus ( damping factor = 200 / $repVal1 ) and use the result.

If setting correctly, even your stage is more than 150 microns away, calling the function with three rounds of iteration can bring the stage to eucentric height close to within 0.5 microns in a few seconds. Amazing to me.

Note

Since 4.0 Testing, eucentricity task using titled beam has been built in SerialEM task. Please see SerialEM helpfile Eucentricity by Focus Setup Dialog.