Forums
Bruker Media
Community Media
Bruker AFM Probes
SPM Digest
Application Notes
NanoTheater
Website
中文
Brochures & Datasheets
Publications
Probes Catalog
Events
Manuals & Documentation
Presentations
Guide to AFM Modes
News
Journal Club
Webinars & Video
Nanovations
Other
Hi,
am I right that the slope window size within the TW removal analysis is a measure for the number of data points taken to calulate the sample slope after the TW is removed? (e.g. if set to 10nm, it takes all privous data points within a 10nm range back and calculates the the current slope using these data points?)
Thank you for clarifying.
And in general is it better to choose TSE over TW for non symetrical samples?
Best
Robert
Hi Robert -
In general, yes you are correct. The easiest way to think about it is that the slope window size defines the length of a given scan line profile of the sample surface over which to determine which portion of the tip shape was interacting with the sample shape at that point, and thus to know which portion of the tip to deconvolve from the image. A smaller window size gives greater granularity on the tip shape extraction (TSE) but the deconvolution calculations will take longer and at some point you will always run into diminishing returns as you start to get into the noise of not only the measurement itself but also that of the tip shape.
Regarding the Tip Shape Extraction (TSE) v. Tip Width Removal (TWR), this again comes down to the amount of time that you have. TSE as you know can take up to an hour per image depending on the data density within the image, so if you are batch processing 100s of images, this could take a while. On the other hand though, TSE does provide TEM level profile accuracy, especially for process stressed artifacts (footing, standing waves, undercuts etc). For some good examples, check out the webinar:
Webinar Video - An Introduction to CD Metrology in the Semiconductor Industry
http://nanoscaleworld.bruker-axs.com/nanoscaleworld/media/p/499.aspx
There are some great examples of the power of TSE in this presentation. As for TWR, this is just an algebraic subtraction of the tip width from the image and thus will not improve resolution for top corner rounding and other tip-sample convolution effects. I would suggest using TWR+VEH (Vertical Edge Height) removal as this will also remove the VEH from the bottoms and tops of features.
To Summarize:
Cheers,
Sean
Hi Sean,
thanks for all the details.
By reading your summary at the end, another question came up.
If I want to do a repeatability study on non straight lines (e.g. spacers) where top, middle and bottom CD is of interest, it would be better to process the images with TW+VEH removal first. And then after that using linewidth analysis with tipwidth set to '0' and 'user specified'?! (so that the TW does not get substracted from the line again) Am I right in my thinking here? That all should not matter on 90° sidewalls.