Hello Vineeth,

the RMS roughness < 1 nm indicates very smooth surface. Deposition of metals on SiO2 substrates without seeding layer (Cr) generally lead to metal gain formation. Also some times the metal deposition may not be continuous. It is a bit surprising to achieve such a low surface roughness of < 0.3 nm. In AFM measurements we can consider 10% measurement error. Further the RMS roughness also depends upon the AFM scan quality. If possible share your AFM picture with me to take a look.

Also consider to do SEM analysis of your sample topography to observe the metal grain formation. If possible you can also do XRR or Ellipsometry for reliable surface roughness and film thickness measurements. Check the attached file.

Hello,

1) The instrumental noise should be known to you, either look at the device specs or you can measure it by yourself - just land and watch the fluctuations (on a scope inside the software) or scan without landing (both methods tell you something slightly different). For example the Solver NEXT I use has internal noise 0,04 nm in Z and 0,3 nm in X,Y according the manufacturer, but I have measured 0,05-6 nm. Also, you wrote correctly that digital resolution is necessary to know.

2) How large is your scanning area? Because calculations of Sq (RMS), Sa (Ra) can (usually do) depend on how large your area is (the values grow).

3) How many mathematical corrections do you use ? With the math included in current software you can make a flat surface pretty much out of anything (as my students continually prove to me :))

3) Don't forget about you tip size. I commonly use tips with radius approx. 6 nm. Unless you use some specialized (like DLC whiskers) probes you will not get much better. You always get convolution of the sample and tip.

My point: I consider the value possible (I do PLD depositions), but I would need to see the original image (possibly including phase or MAG signal). Then again, AFM is just one analysis, as suggested above, try (if you can) some other analysis such as SEM (under different angles).

Fully agree with all previous comments. 

What is the surface area of your sample ? if possible, perform more that one scan.

I may be wrong here, but can you do a cross sectional view of your prepared sample under SEM, to verify the thickness of the layers? 

Hi all i am attaching an AFM image of my sample here. SEM of the similar sample was no different, with no visible grains

Thanks to all you  for your suggestions.

Dear Vineeth,

the scan quality is not the best. Observing the image, I assume that the cantilever tip is not sharp. The tip curvature is blunt. Possibly the cantilever is also not installed properly. There are too many scan lines, which were flattened. I would say the roughness measurement

to conform the roughness you may perform X-ray reflectivity.

You could try a reference sample like single crystalline oxide/Si surface to rule out the instrumental error.  Tip radius is very very important.

Hi Vineeth,

if you want to verify the roughness, i suggest two things:

-Check for the tip radius by measuring a roughness control sample like etched titanium. Or try to measure the same area on your sample with several tips. If the tip is too blunt, you might not resolve all roughness features.

-measure a HOPG sample, freshly cleaved. There will be always areas that are atomically flat (typically areas of a few hundred Nanometers in width). Any deviation from that is your noise!

Regards, F.

There are alternative techniques to measure sub-nm surface roughness, although tapping mode AFM with a  fresh tip is an excellent method and would be my first choice.  Another way to do this is via glancing angle x-ray reflectivity.  While best performed with a high-power x-ray source (synchrotron or rotating anode), it can be performed using a good 4-circle x-ray diffractometer typically used for XRD, hence, it is widely available.  Fitting the low angle data will provide you with the surface roughness.  

5um range is huge; even if you use a very slow scanning speed, you would need to samples every nm in order to get a reliable result (so more than 5000 points per line)...

I would scan from 1um till 200nm using at least 512 points/line and a good overlap between back and forth lines, with a new tip and check how your roughness is changing. Look also at your peak to peak values, not only rms. Using XRD to compare roughness seems also like a pretty good idea.

Additionaly to previous comments about scan regime (less then 1micro, ...) you can try two AFM regimes: tapping and contact modes with different tips to be sure about your resalts.

If RMS = 26.287 nm, Ra= 21.128 nm, Img. Z-range= 204.61 nm, No. of samples = 512. How can we calculate error for RMS and Ra using this data?

My experience in measuring surface roughness at around 1 nm is with an AFM in tapping mode as introduced previously with new tip but also on an optical damped table and the sample with AFM inside a protection from any noise from air.