The Nanoscale World

How to measure a deep pit?

rated by 0 users
Answered (Verified) This post has 2 verified answers | 8 Replies | 4 Followers

Top 50 Contributor
15 Posts
Points 177
simon posted on Tue, Oct 5 2010 11:35 AM

I am using Dimension 3100 to measure a deep pit (100-1000 nm deep), but in the scan direction bumps are always appeared adjacent to the pit. The bumps are not real surface morphology only appeared in the scan direction. I reduced the scan rate to 0.2 Hz, but the bumps were still there.

Why are the bumps generated? Could you please teach me how to eliminate or reduce these bumps?

I have a picture in "My files" measured in the following conditions, I don't know how to post it here.

Dimension 3100; Tapping mode; scan size: 10 um; scan rate: 0.4 Hz; Integral gain: 0.4; Proportional gain: 1.2; Amplitude setpoint: 1.3V; Sample: SiC

Thank you very much.

  • | Post Points: 15

Answered (Verified) Verified Answer

Top 25 Contributor
30 Posts
Points 340
Verified by simon

Simon- Thanks for sending the file.  The artifact is not in the raw data before it is modified with "flatten."  One way to introduce an artifact like this is to modify the image with a first order flatten using a "stop box" that is too small to exclude the entire pit from the flatten.  If the entire pit is not excluded from the flatten each scan line that contain the non-blocked pit will be offset up as a result of the line by line modification of the image with "flatten."  Is it possible that this is what is happening (see below).  Let me know if this helps.  Thanks, Mike Maybrun

Image you provided that shows artifact.

 

An artifact similar to the image above can be produced by an incomplete "stop box" used for the flatten routine.  The box does not completely contain the pit.

 

Here is a 3d view that shows the artifact as a result of the stop box being too small.

Completely covering the pit with the stop box will result in no artifact.

 

3d view shows the artifact is not in the image.

  • | Post Points: 13
Top 25 Contributor
30 Posts
Points 340
Verified by simon

Simon, certainly not using a stop box at all would produce the same artifact to a greater extent.  Flatten removes offset (0 order), tilt (1st order) or bow (2nd order) out of each individual fast scan line in the image.  It is this line-by-line treatment of the image that is the reason for the artifact.  Scan lines that contain the pit have a lower average z value than the scan lines of the flat surface surrounding the pit.  Since the average z offset is subtracted out of each scan line the result is an upward shift in the scan lines containing the pit.  To prevent the artifact you can use the stop box to exclude features from the line by line correction.   When using flatten you should draw these boxes around high and low features in the image before executing the flatten.  Flatten is particularly useful for removing bright streaks across an image in the fast scan axis that are the result of different average z scanner values from scan line to scan line.  Flatten is not always the best choice.  You can also modify the image with 0, 1st, or 2nd order plane fit.  The difference is that plane fit removes offset, tilt, and bow based on average values of the image as a whole; not line by line.  It does not produce the type of artifact you saw.  On the other hand, it does not remove z offset between individual scan lines.  I suggest becoming comfortable with both methods and when they are best used. 

The feedback setting you are using look reasonable.   For the 1um deep pit I would keep the scan rate around 0.15-.25 Hz.  For the 100nm pit you might get away with 0.6-1.0 Hz.  To optimize the settings you need to watch the error signal and try to minimize it.  Higher gains will allow you to scan faster but at some point the z scanner will oscillate.  Lower amplitude set point will cause the tip to track better but will also accelerate tip wear.  Scanning slower will generally improve the tracking and allow you to operate with a higher amplitude setpoint (lower imaging force).

Hope that helps,

Sincerely, Mike Maybrun

 

 

  • | Post Points: 13

All Replies

Top 25 Contributor
30 Posts
Points 340

Simon- It could be tip artifact or a spike in the error signal but I really need to see the image.  Please send me the file directly.  Nanoscope data file would be more helpful than jpg.  Send to mmaybrun@veeco.com.  Thanks, Mike Maybrun

  • | Post Points: 10
Top 25 Contributor
30 Posts
Points 340
Verified by simon

Simon- Thanks for sending the file.  The artifact is not in the raw data before it is modified with "flatten."  One way to introduce an artifact like this is to modify the image with a first order flatten using a "stop box" that is too small to exclude the entire pit from the flatten.  If the entire pit is not excluded from the flatten each scan line that contain the non-blocked pit will be offset up as a result of the line by line modification of the image with "flatten."  Is it possible that this is what is happening (see below).  Let me know if this helps.  Thanks, Mike Maybrun

Image you provided that shows artifact.

 

An artifact similar to the image above can be produced by an incomplete "stop box" used for the flatten routine.  The box does not completely contain the pit.

 

Here is a 3d view that shows the artifact as a result of the stop box being too small.

Completely covering the pit with the stop box will result in no artifact.

 

3d view shows the artifact is not in the image.

  • | Post Points: 13
Top 50 Contributor
15 Posts
Points 177
Answered (Not Verified) simon replied on Tue, Oct 5 2010 4:48 PM
Suggested by Mike Maybrun

Hi Mike,

Really appreciate your help. I can get good image now. I have another two questions about this measurement.

First, I never used "stop box" with "flatten" in my previous analysis. I just clicked "flatten" then "execute". What's the difference of these two types of "flatten" and in which condition should I use "stop box"?

Second, Do you have any suggestions on parameter settings to get reliable results when measuring such kind of pit? (For example: scan rate, integral gain, proportional gain and amplitude setpoint)

Best

  • | Post Points: 12
Top 25 Contributor
30 Posts
Points 340
Verified by simon

Simon, certainly not using a stop box at all would produce the same artifact to a greater extent.  Flatten removes offset (0 order), tilt (1st order) or bow (2nd order) out of each individual fast scan line in the image.  It is this line-by-line treatment of the image that is the reason for the artifact.  Scan lines that contain the pit have a lower average z value than the scan lines of the flat surface surrounding the pit.  Since the average z offset is subtracted out of each scan line the result is an upward shift in the scan lines containing the pit.  To prevent the artifact you can use the stop box to exclude features from the line by line correction.   When using flatten you should draw these boxes around high and low features in the image before executing the flatten.  Flatten is particularly useful for removing bright streaks across an image in the fast scan axis that are the result of different average z scanner values from scan line to scan line.  Flatten is not always the best choice.  You can also modify the image with 0, 1st, or 2nd order plane fit.  The difference is that plane fit removes offset, tilt, and bow based on average values of the image as a whole; not line by line.  It does not produce the type of artifact you saw.  On the other hand, it does not remove z offset between individual scan lines.  I suggest becoming comfortable with both methods and when they are best used. 

The feedback setting you are using look reasonable.   For the 1um deep pit I would keep the scan rate around 0.15-.25 Hz.  For the 100nm pit you might get away with 0.6-1.0 Hz.  To optimize the settings you need to watch the error signal and try to minimize it.  Higher gains will allow you to scan faster but at some point the z scanner will oscillate.  Lower amplitude set point will cause the tip to track better but will also accelerate tip wear.  Scanning slower will generally improve the tracking and allow you to operate with a higher amplitude setpoint (lower imaging force).

Hope that helps,

Sincerely, Mike Maybrun

 

 

  • | Post Points: 13
Top 10 Contributor
288 Posts
Points 3,905
Bruker Employee

I would also recommend that you download NanoScope Analysis. There is an extensive, clear, and convenient online help for all of these features. Steve

Nanoscope Analysis v120r1sr3.pdf

  • | Post Points: 12
Top 50 Contributor
15 Posts
Points 177
simon replied on Wed, Oct 6 2010 12:33 PM

Thank you so much, Mike and Steve.Smile

 

  • | Post Points: 10
Top 500 Contributor
4 Posts
Points 44
Burden replied on Wed, Nov 10 2010 11:26 AM

Hello,

This question and thread of discussion is intersting.  We are trying to image something similar in our lab (a deep hole) with a Catalyst system.  I am trying to learn as much as I can about the experience of other users with this type of sample.

I am wondering if there is a way to see the image (and the edge effect) referenced in this thread?  This would help me assess if our images are affected by similar artifacts. 

I am new to the Nano World online community and so I may not be fully aware of the fetures provided in this online tool.  Did you wind up getting your image posted?  Or did you just send an example along via email?  Can you assist?

Thanks. 

Dan Burden

Wheaton College (IL)

  • | Post Points: 12
Top 10 Contributor
288 Posts
Points 3,905
Bruker Employee

Hi Dan, hope all is well.

Here is a post on how to upload images to the fourm: http://nanoscaleworld.bruker-axs.com/nanoscaleworld/forums/t/526.aspx

It looks like Adam and Bede are addressing your technical question at: http://nanoscaleworld.bruker-axs.com/nanoscaleworld/forums/t/522.aspx

Thanks for posting,
Steve

  • | Post Points: 10
Page 1 of 1 (9 items) | RSS
Copyright (c) 2011 Bruker Instruments