The Nanoscale World

Expected z-sensor ramp response PFM

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

Top 500 Contributor
2 Posts
Points 29
abayrak posted on Mon, Nov 21 2011 3:23 PM



I am performing Piezoresponse force microscopy (PFM) measurements using a NanoMan AFM with Nanoscope V controller and software version 7.30r1sr3.   I've been trying to determine the effective d33 coefficient of my ferroelectric thin films by using ramp measurements, however the response isn't quite what I expect.

I am expecting something like this when I take a ramp measurement of the z-sensor vs dc bias:

Expected Z-sensor

But instead, I am getting this result: (blue is approach, red is retract)

Z-sensor Result

Although the magnitude of the slope of either line provides the same d33 value, I am expecting the material to expand at some point.  For this measurement I am in contact mode with a z-limit to 1um, using MESP-RC (or SCM-PIT) tips and set my gain (integral/proportional) to zero.  I am also performing these measurements on parallel plate capacitors so the electric field is in the z-direction.

However, when I measure in piezoresponse mode, the ramp of the Amplitude/Phase vs Bias are similar to what I expect: (gain is enabled)

Amplitude Signal Phase Signal

(Note: These measurements were performed on PZT poly-crystalline thin films, but I have a similar response on Barium Titanate (BTO) and Barium Strontium Titanate (BST))


Would anyone be able to help me determine where I am going wrong with these measurements?





  • | Post Points: 12

All Replies

Top 25 Contributor
55 Posts
Points 831
Bruker Employee
Suggested by Ang Li

Dear Adrian,

Using ramping mode to determine d33 is not the preferred way, I recommend you use the Piezoresponse Mode.

Below is what why I think you are not getting what you expect with ramping:

  • You should not set gains to zero if you want z-piezo, thus z-sensor, to follow the mechanical response to the applied DC voltage.
  • The expected z_sensor should not be a "V" shape, but something as below:
  • From the what you got, there seems to be some drift in the z-sensor signal.

Anyway, the ac method used in Piezoresponse mode is far more advantageous over DC ramping. And if you really want to explore DC ramping, it is better off to look at vertical deflection rather than z-sensor at gains=0. Be aware, low frequency noise and drift can present a problem, and  the sensitivity will not be good enough for materials with samll d33. As you may know, in piezo-response mode, vertical deflection is multiplied by 16x to increase sensitivity.

Best regards


  • Filed under:
  • | Post Points: 12
Top 500 Contributor
2 Posts
Points 29
abayrak replied on Wed, Nov 30 2011 4:59 PM

Hello Chunzeng,

Thank you for your reply.  Originally I was expecting a very linear relationship with voltage but your z-sensor visualization is quite reasonable.  Would you expect this shape is due more to the material or the PFM actuator?  That is, if a ramp measurement is performed on a non-ferroelectric material would you still expect this shape since the actuator is comprised of PZT?

For the last two graphs above in my previous post (Amplitude, Phase), I performed a dc ramp measurement in piezoresponse mode in accordance with our nanoman PFM manual.  I believe the PFM is stimulating the ferroelectric material at the same time with an AC bias at a certain frequency (in kHz).  My concern with this method was that there would be kind of an interference when I tried to quantitatively determine d33.  For example, the maximum deflection induced in this measurement may be reduced if my AC drive bias is at a lower value (say 5Vac instead of 10Vac), although the dc bias sweeps across the same interval.

I suppose I could take the PFM Data channel, which has a hysteresis shape with the channel relationship Amplitude*sgn(Phase), and convert the y-axis to an effective d33 units using the equation

A(nm/V) = A (in mV) * deflection sensitivity (nm/mV) / ( vertical deflection gain (16) * Applied AC Bias)

which is similar to the figures in some author's papers.  I am not sure if this graph will have the same effective d33 values at different AC biases however.

Also, would the vertical 16x gain be the only gain that needs to be accounted for while taking quantitative measurements or do we have to take note of the integral and proportional gain values as well?

Thanks for the help,


  • | Post Points: 17
Not Ranked
1 Posts
Points 10

Hi Abayrak,

I am having the same kind of curves that you showed above and I would like to foolow the same way that you propose to plot d33 vs Bias Voltage. However, I did not understand it totally. Could you explain it to me again please? Are you combining the two graphs Amplitud and Phase ? 


Many thanks

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