The Nanoscale World

Measuring noise

rated by 0 users
Answered (Not Verified) This post has 0 verified answers | 3 Replies | 2 Followers

posted on Thu, Jan 7 2010 9:03 PM

Can someone describe, or better still, give a reference to a standard method for determining noise levels in AFMs? In particular,

 I'm interested in a method applicable to compare noise levels across different microscopes,

so it should not be dependent on a particular scope's software.

All Replies

Answered (Not Verified) replied on Thu, Jul 21 2011 10:42 AM

You have 2 ways to measure the noise on a system:

1) Imaging mode: I suggest to use a superflat substrate like, mica, HOPG, silicon wafer, but not glass, and capture a 1x1 um image with the minimum possible z-limit and the gains as low as possible (specific to the tool you are using). The resolution can be 512x512 and the scan rate 0.5 Hz for tapping mode or 1 Hz for contact mode. The open "Nanoscope Analysis" and measure the roughness on the entire surface. Look at the RMS value. The height noise should be <0.2 nm.

2) Force mode. It's even faster. just capture a force curve on the same type of substrate and display it. The noise level should be <10pN with a cantilever suitable for force measurements on soft samples, which means with a spring constant <0.1 N/m.


  • | Post Points: 12
Top 25 Contributor
29 Posts
Points 319


Alex's answer is pretty much correct. What you might find is that there are some advanced controls in the different software that affect the results you get, such as bandwidth control, filters, etc. I actually asked this question of a few different vendors, and they tend to do it slightly differently, but Alex's answer is a good one. Make sure the sample is perfectly clean in the area you choose. I wrote this up into a "standard procedure" in Appendix B of my book.


  • | Post Points: 10
replied on Wed, Jul 27 2011 11:26 AM

I would like to qualify Alex answer a bit.


1. When you put your PIDs as low as possible you can basically get almost any noise "number" you want. Your PIDs should be set properly so that you can track the surface in the range you are scanning (or in a few nm scan range if you so choose to set your noise imaging range to "0" nm). On the sustrate side please make sure that there is no static charge present and/or take care to dissipate that by properly grounding and/or discharging the sample. This can be especially critical with samples like Si-wafers.Your scan rate will also act as a filter for data. This is of course not a bad thing but one has to pay attention to it. At the suggested 512pts with 0.5Hz one would move the tip each second over 512 pixels thus having a pixel frequency of 512Hz. Just reducing the number of pixels to 256 might give you better noise data as one has more averaging between pixels. If you really want to run a nice experiment you can scan at different pixel frequencies and plot the noise number vs the pixel frequency.


  • | Post Points: 10
Page 1 of 1 (4 items) | RSS
Copyright (c) 2011 Bruker Instruments