Coupling of Audio Signals into AFM Images

yakzephyrAI and Robotics

Nov 24, 2013 (3 years and 8 months ago)

67 views
























Senior Project
-

Electrical

Engineering
-

2012

Coupling of Audio Signals

into AFM Images

Matthew Manning

Advisor: Prof.
Catravas

Atomic force microscopy is capable of

imaging samples with nanometer scale features.
However, given the sensitive nature of
the system, the measurements are easily influenced
by external
vibrations. This project investigated the system response of an
AFM to audio excitations. Images were produced in the presence of audible noise, and frequency recovery was attempted via
digital signal processing methods. The impulse response of the system was measured for three operating modes of the AFM:
attractive, repulsive and intermittent contact modes. The results reflect the linearity/nonlinearity of the system and can p
rov
ide
insight into the imaging process.

Results
:


Signal Delivery:









Impulses, single frequencies and music signals
delivered to system in different operating modes


Speakers rated to 48Hz utilized due the to low
sampling rate of AFM (
Fs

= 1024s/sec)


Model MULTI75AI probes used for imaging; useful
for imaging in different AFM modes


Imaged on LEXAN Polycarbonate samples for
minimization of noise caused by sample surface




Within Noise Shield

Actively Imaging



Impulse responses in repulsive, attractive and intermittent
contact mode are shown at the right.



Each mode of operation produced a different impulse
response. The shortest response was found in repulsive
mode; the longest in intermittent contact mode. This is
qualitatively consistent with the non
-
linear characteristics of
the force
-
distance curve.



Convolution of incident sinusoids excitations with the
measured impulse response produced results that resembled
the measured output, yet contained noticeable differences.
Differences were also observed in the images (shown far right)
obtained in the three modes for a music signal excitation. This
implies that a “small signal”
linearized

model was not valid for
the experimental parameters chosen.



It is of interest for an AFM user to know when a “small signal”
model is or is not valid when interpreting imaging results. The
approach described here has the potential to provide such
information.


Corrupted Image

x(t)

h
(t)

y
(t)



Improved filter design


Improved system characterization (higher sampling rate, large
range frequency sweeps)


Implementing lithography methods to mimic ambient noise
induced error, used as a physical writing method






Prof. Palmyra
Catravas


Union ECE Department


Armin Knoll, IBM Zurich


Nanofabrication Group, IBM Zurich






Single frequency excitation signals produced sinusoidal artifacts in the
images; Several digital filters were developed to minimize
effects of aliasing and reduce noise



Reference images were used for background noise
reduction



System was characterized by measuring


impulse response in different modes:


Attractive, Repulse and Intermittent



To further test hypotheses of
linearity/nonlinearity, convolution was utilized
to derive theoretical results for comparison



Results from complex music signals analyzed
with same methodology


Impulse
-
Like Incident Signal

System Response
:

A)
Repulsive

B)
Attractive

C)
Intermittent

Data Analysis
:

Design
:

Acknowledgements:

Abstract:

Impulse Response
-

Repulsive

Impulse Response
-

Attractive

Impulse Response
-

Intermittent

Music Signal
-

Repulsive

Music Signal
-

Intermittent

Music Signal
-

Attractive

Future

Work: