It is not correct, you cannot read directly the y axis because segments are not parallel to x-axis. Instead, do as the I did in the picture (just a rough estimate i did in powerpoint :). The correct height is from parallel steps, I put the green line as the correct height and then just copied and rotated near the scale to obtain ~2.6 um

It is not correct, you cannot read directly the y axis because segments are not parallel to x-axis. Instead, do as the I did in the picture (just a rough estimate i did in powerpoint :). The correct height is from parallel steps, I put the green line as the correct height and then just copied and rotated near the scale to obtain ~2.6 um

Thank you Fernando Vallejos-Burgos 

But why this curve nature coming?

There is something wrong with a flattening procedure you applied. Two parts of the curve going up from left to right side should be horizontally oriented. What software do you use? With help of the software I use (former Digital Instrument, currently Bruker) it is easy to flatten such profile (in Manual Plane Fit Manual panel use Erase command). After such procedure application you should check how it affect the modified profile. The easiest way is to measure in the same conditions a known surface (e.g. grating for calibration purposes), apply the same flattening procedure, measure the step height and compare it with data given by the supplier of the grating. Good luck.

Thank you Andrzej Wawro

Actually I used Nanoscope Analysis 1.3 by Bruker. Here I attached my raw AFM image. The steps are not horizontally oriented.  I have tried Plane fit, Flatten like options , but still it look like same. 

Your raw data is good and corresponds to many experimental observations: Planes are not perfectly horizontally aligned in the microscope but instead the part of the sample you are observing has a little inclination. Also, there is some end effect in the boundary between steps. BTW, did you optimize all the control parameters of the tip before acquiring the image (i.e. is that the real image without noise generated by control settings?)

To measure the step size, if you follow what I told you, then you dont need to do more processing.

However, if you so first order flattening as data treatment then you should have the two segments parallel and will be even easier to measure.

Thanks again Fernando Vallejos-Burgos

Ya I have optimized all the control parameters (Integral & proportional gain, scan rate, tip velocity etc) of the tip before acquiring the image.

Yes it is the real image and there was no noise.

Yes you are right, whatever you told is correct way of measuring thickness. But why I am again asking is that it was my first try to measure thickness.  If there some additional care that we have to take care during these kind of measurements, I wanted to know more about that. My sample is PVDF thin film and I did it in tapping mode. And I dont even have much idea regarding the data processing. Mostly after taking the measurements I used to do a flatten and measure the roughness, Is this correct? And also is that necessary to do tuning of the tip before each measurements? So I expect more suggestions regarding this

We come up against this problem all the time. Your raw data looks fine, but the levelling process is trying to level out your step, hence inducing a steep looking slope. In Nanoscope Analysis perform a 1st order plane fit, but manually exclude one side of the step, at this stage by drawing an exclusion box around it. You will then be levelling the whole image to something you know is a flat surface. Once you have got it looking roughly level you can be more precise by using thresholding (in the same Nanoscope Analysis levelling function) to exclude everything but the flat portion you wish to level.

I would just add that when scanning over large areas with some AFMs there can be an artifact that leads to a curvature appearing on an actually flat surface.  After going through the procedure Simon Connell describes above you might experiment with a second order planefit (again with just one side of the step selected).

• Yes, I agree with John. Looking at your line profile there is a pronounced curvature, a 2nd order should deal with this nicely. Higher orders are available, but when excluding such large area from the flattening you will probably induce extra curvature in the "ignored" area, so in this case use the lowest order

Your thickness is very high for an AFM. Use Talysurf with scientific rigour and data analysis. AFM resolution is 0.6nm and hence in this case is equivalent to killing a mosquito with cannon.

Thank you all for your suggestions.....

you can minus ax+b from your data, where a is the  slope of your experiment data