Characterization of a me-

chanical oscillator using

fiber interferometry
TN2935-P: Research Practicum
Giovanni Ferro, Lucas de Lannoy, Qadim Sarucco
Delft University of Technology
 Characterization of a
mechanical oscillator
using fiber interferometry
by

Giovanni Ferro, Lucas de Lannoy, Qadim

Sarucco

Instructor: Prof. Simon Groeblacher

Teaching Assistant: Gaia da Prato
Institution: Delft University of Technology
Place: Faculty of Applied Sciences, Delft
Project Duration: May, 2024 - September, 2024
 Abstract
In this research practicum, the mechanical motion of a piezo stack was characterized using a fiber-optic
interferometer. The experimental setup involved a Fabry-Pérot interferometer, where an optical fiber
was aligned perpendicularly to a reflective chip mounted on the piezo stack. Firstly an attempt was
made at determining the responsivity of the piezo stack was made by measuring the output at the fiber
at different DC-voltages. Next the cutoff frequency was determined to be 137.5 ± 2.5 Hz, based on the
analysis of peak counts in the measured signal over one period of the driving signal. In the second
part of the experiment the resonance frequencies of the Long Wide Cantilever (LWC) attached to the
chip were measured by analyzing the photodiode signal in the frequency domain using a spectrum
analyzer. These resonance frequencies were compared to theoretical predictions. Additionally the 0th
and 2nd mode of vibration of the cantilever were studied and mapped showing clear patterns of nodes
and anti-nodes.

i
 Inhoudsopgave
Abstract i
List of Figures iii
List of Tables iv
1 Introduction 1
2 Theory 2
2.1 Fabry-Pérot interferometer & Piezo stack . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2.2 Beam mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.3 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
3 Experimental Method 4
3.1 Characterisation of the optical cavity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.2 Detection of resonance motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
4 Results and Discussion 6
4.1 The responsivity and cutoff frequency of the piezo stack. . . . . . . . . . . . . . . . . . . 6
4.2 Frequency response of the cantilever . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
5 Conclusion 10
References 11
A Theoretically determined roots and resonance frequencies of the cantilever 12

ii
 List of Figures

2.1 A schematic view of the Fabry-Pérot interferometer, with length l between the two reflec-
tive surfaces [2]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

3.1 A schematic view of the setup along with a zoomed in scheme of the laser and the
cantilever chip. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

4.1 Plots of the data obtained from the oscilloscope and optical fiber for different DC-offset
voltages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
4.2 Plot of the input signal (orange) and measured signal from the optical fiber at 50 Hz. . . 7
4.3 Plot of the input signal (orange) and measured signal from the optical fiber at 135 Hz. . 7
4.4 Plot of the input signal (orange) and measured signal from the optical fiber at 140 Hz. . 7
4.5 Plot of the driven frequency response of the Long wide cantilever ranging from 1kHz to
1.5MHz. The dashed vertical lines are the theoretically determined resonance frequen-
cies of the cantilever. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4.6 Plot of the driven frequency response of the Long wide cantilever at the 0th mode of
vibration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4.7 Plot of the peaks of the driven frequency response of the Long wide cantilever at the 0th
mode of vibration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
4.8 Plot of the driven frequency response of the Long wide cantilever at the 2nd mode of
vibration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
4.9 Plot of the peaks of the driven frequency response of the Long wide cantilever at the 2nd
mode of vibration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

iii
 List of Tables

A.1 The first 6 resonance frequencies with their corresponding αn values of the cantilever
with Young’s Modulus E = 270 GPa (silicon nitride), density ρ = 3184 kg m−3 and dimen-
sions: L = 100 µm, w = 20 µm [4], h = 0.8 µm . . . . . . . . . . . . . . . . . . . . . . . . 12

iv
 1
Introduction
Piezo stacks are commonly used across various industries for different purposes, particularly for its
high accuracy and responsiveness. The piezo stack has many applications such as medical devices,
industrial automation processes, noise cancellation systems, autofocus mechanisms for camera’s and
also instruments that require high resolution movements. The stacked bonded layers of piezoelectric
ceramics has the ability to expand or contract in a predictable way when a voltage is applied to it, lead-
ing to a precise motion that can be measured. By applying an electrical signal that varies periodically,
the piezo then translates the electrical input into a periodic displacement. This allows there to be accu-
rate control in positioning, often in the micrometer range [1].

In this research project the goals is to characterize mechanical motion with the use of a fiber-optic
interferometer. The research consists out of two parts, the first part mainly consists of the characteriza-
tion of the fiber-optic interferometer and the displacement responsivity of the piezo stack for oscillating
voltages at different amplitudes and frequencies. This will be done by the use of a Fabry-Pérot inter-
ferometer, consisting out of an optical fibre with the tip perpendicularly pointed to a reflective chip. The
chip is mounted to the piezo stack and is classified as the moving mirror in this setup. In the second
part the focus is on detecting the resonance frequencies of two cantilevers attached to edge of the chip.
This will be done by measuring the strength of the photodiode signal in the frequency domain using a
spectrum analyzer.

In chapter 2 the theoretical concepts behind this research will be described in order to give some
basic understanding of the tools used for this research. In chapter 3 the used method and the ex-
perimental setup will be described, the 4rd chapter presents all the results and in the 5th chapter the
conclusion will be discussed.

1
 2
Theory
2.1. Fabry-Pérot interferometer & Piezo stack
A Fabry-Pérot interfermeter is an optical instrument that uses to partially reflective surfaces to cause
multiple reflections between the surfaces which is called a cavity [3]. A part of the incoming light beam
gets reflected whereas the transmitted beam gets reflected by the second surface (see figure: 2.1),
causing to have a phase difference which is given in the following equation.

δ = 4πλL (2.1)

By knowing the length L between the two surfaces (which is called the cavity length) and as well
wavelength λ the interference is constructive or destructive. For which L can be calculated for either
constructive or destructive interference with the following two equations.
1
Constructive interference: L = mλ (2.2)
2
( )
1 1
Destructive interference: L = m+ λ (2.3)
2 2
Where m = 0, 1, 2, ....

Figuur 2.1: A schematic view of the Fabry-Pérot interferometer, with length l between the two reflective surfaces [2].

In this experiment, the fixed mirror is the interface between tip of the optical fiber and the surrounding
air, the moving mirror is a cantilever chip and the distance between the fiber and the chip is L. The
incoming light beam either reflects back into the fiber or it propagates out of it. The light reflected by the
chip propagates out of phase back into the fiber. The interference caused by the Fabry-Pérot cavity is

2
2.2. Beam mechanics 3

then used to measure the mechanical of the piezostack, which is where the cantilever chip is mounted
on. A piezostack is an energy converting material, which specifically converts an electrical voltage
into motion [6]. The piezostack will for this reason have an oscillating type motion when an oscillating
voltage is applied to it, which then can be studied using the Fabry-Pérot interferometer.

2.2. Beam mechanics

The Euler-Bernoulli equation (eq. 2.4) which is dependent of a time and spacial part, can describe the
movement of a cantilever beam.

∂ 4 Z(x, t) ∂ 2 Z(x, t)
EIz + ρA (2.4)
∂x4 ∂t2
Where ρ is the density of the material, A is the area of the cross section of the device (for A = w · h),
E is the Young’s modulus and Iz is the area moment of inertia. The area moment of inertia can be
calculated with equation 2.5, where h is the height and w the width of the cantilever cross section.
∫ w ∫ h/2 ∫ w [ ]h/2
1 3 1 3
Iz = z 2 dzdy = z dy = h w (2.5)
0 −h/2 0 3 h/2 12

With the assumption that Z(x, t) = u(x)eikt is a solution of the Euler-Bernoulli differential equation, the
∂4u
solution of u(x) can be written as − k 4 u = 0. Combining this with equation 2.5 gives the following
∂x4
dispersion relation: √
hk 2 E
ω= (2.6)
2 3ρ
The general solution for u(x) is given by u(x) = a1 cosh(kx) + a2 sinh(kx) + a3 cos(kx) + a4 sin(kx)
along with some boundary conditions we find that the solution is given by 1 + cosh(αn )cos(αn ) = 0,
where αn = kn L and n is the mode of the resonance frequency. With the knowledge that ω = 2πf and
αn = kn L, the frequency can be expressed by equation 2.7.
√
hαn2 Eh3
f= (2.7)
4πL2 3ρA

2.3. Data Analysis

The responsivity of the piezostack must be determined using a linear relation between the amplitude
of the piezo and the peak voltage measured using the oscilloscope. The relation should be somethiing
as follows:

Apiezo = A0 + r · Vpp (2.8)

Where r is the responsivity, Vpp the peak-to-peak Voltage of the triangular driving signal and A0 the
amplitude when the driving voltage is zero.
Because it is not possible to measure the amplitude of oscillation of the piezostack, a theoretical
expression for Apiezo must be determined in order to perform a linear fit. Regrettably the necessary
formula could not be determined in this report, thus the fit could not be executed to determine the
responsivity.
 3
Experimental Method
3.1. Characterisation of the optical cavity
As described in chapter 2 the used interferometer in this experiment consists of an optical fibre, with the
perpendicularly pointed at the surface of a reflective chip which can form a Fabry-Pérot interferometer.
To characterize the optical cavity created by the tip of the optical fiber and the chip mounted on the
piezo stack, a schematic of the setup is given in figure 3.1.

Figuur 3.1: A schematic view of the setup along with a zoomed in scheme of the laser and the cantilever chip.

In this setup a laser diode generates light with a wavelength of λ = 635 nm which is coupled into an
optical fibre. This is then followed by a 50/50 beamsplitter, where half of the incoming light is blocked
by a light trap and the other half propagates towards the piezo stack with the chip on it. The reflected
light then propagates back in the fibre and is detected by a photodiode. The position of the fibre can be
adjusted manually by a positioning stage. The piezo stack is connected to a PI-E660 voltage amplifier
which provides the driving signal. The amplified signal comes from a TG315 function generator, for
which the output is a sawtooth wave at a frequency of 10 to 50 Hz and a peak to peak voltage of 5V.
Using a beamsplitter this same function is also connected to an oscilloscope. The other cable plugged
in the oscilloscope is the output of the photodiode. Then the amplitude and DC of the driving signal
and the position of the fibre are adjusted in order to get an optimal interference response. To determine
the cut off frequency multiple traces are taken between 10 and 200 Hz, in order to find the frequency
at which the output of the amplifier is reduced by a factor of 2.

3.2. Detection of resonance motion

In the second part of the experiment the position of the optical fibre is placed in front of the tip of the
cantilever. The output of the photodiode is then plugged into the the input channel of the spectrum
analyzer. With this same spectrum analyzer the device can be driven as well, which makes it possible
to measure a driven frequency response of the cantilever. To excite the cantilever in this part of the
experiment a dither-piezo is used, because piezo can operate at a wider range of frequencies. On the
spectrum analyser the response is graphed against the driving frequency and resonance frequencies
are shown as peaks in the spectrum. These same frequencies can be estimated with equation 2.7.
Since each resonance mode is a standing mechanical wave of the cantilever, there are nodes (points

4
3.2. Detection of resonance motion 5

of no displacement) and anti-nodes (points of maximum displacement) along the length of the cantilever.
Using the positioning stage along the length of the cantilever traces are taken to map the resonance
mode.
 4
Results and Discussion
4.1. The responsivity and cutoff frequency of the piezo stack
Below in figure 4.1 four plots, showing the data obtained from measuring the output of the optical fiber
and the driving voltage. The data has been scaled for easier reading. The dotted lines represent the
driving voltage, a triangular wave, and the continuous lines represent the output measured from the
optical fiber. In each plot it is observed that the peaks from the optical fiber are slightly shifted to the
right compared to the peaks of the driving voltage. Another notable observation is that there is a quick
dip right at the center of the peaks. The amplitude of the optical fiber changes throughout the period
of the driving signal as the piezo stack vibrates, moving closer to and farther from the tip of the optical
fiber cable. It is also noteworthy that there are several factors which could cause some distortions in
the measured output. The most important is the optical fiber itself, as not all light traveling through it
reaches the piezo stack as there is some internal reflection and refraction within the cable itself.

Figuur 4.1: Plots of the data obtained from the oscilloscope and optical fiber for different DC-offset voltages.

Using the data obtained and formula 2.8, the responsivity should be determined using a linear fit,
this would require the actual amplitude of the piezostack, which possibly could be determined using
the data given in figure 4.1.
When determining the cutoff frequency, the number of peaks per period were observed while the
frequency was changed from 50 Hz to 200 Hz. Figure ?? shows the starting point at 50Hz, around 12

6
4.1. The responsivity and cutoff frequency of the piezo stack 7

peaks are observed between the lowest point and highest point of the driving signal (orange line).

Figuur 4.2: Plot of the input signal (orange) and measured signal from the optical fiber at 50 Hz.

After changing the frequency, a point is approached where there are about 6 peaks between the
peak-to-peak region of the driving signal, this point is reached at a frequency between 135-140 Hz (see
figure 4.3 and 4.4. Thus the cutoff frequency is determined to be around 137.5 ± 2.5 Hz.

Figuur 4.3: Plot of the input signal (orange) and measured signal from the optical fiber at 135 Hz.

Figuur 4.4: Plot of the input signal (orange) and measured signal from the optical fiber at 140 Hz.
4.2. Frequency response of the cantilever 8

4.2. Frequency response of the cantilever

From the measurements done on the Long wide cantilever, the results given by figure 4.5 and 4.6 were
obtained.
Figure 4.5 shows the frequency response of the cantilever and the theoretically determined res-
onance frequencies (see appendix A). It is noteworthy that the peaks, while most are close, do not
perfectly align with the theoretical resonance frequencies. Several factors could contribute to the devi-
ations from the expected values. A possible origin for the deviations is the variation in the characteristics
of the cantilever used in formula 2.7 and possible imperfections in the setup of the cantilever such as
the way it is fixed at one end. One interesting factor is the value of the Young’s Modulus of the cantilever.
In this case a value of 270 GPa [5] was used but there could be large variations in this value due to
material properties and environmental effects. Another important observation is that there are some
extra noise-peaks between the resonance peaks. The noise most likely comes from the chip itself.

Figuur 4.5: Plot of the driven frequency response of the Long wide cantilever ranging from 1kHz to 1.5MHz. The dashed
vertical lines are the theoretically determined resonance frequencies of the cantilever.

Figure 4.6 shows the change in the 0th vibrational mode, measured over the whole length of the
cantilever. As expected, at 0 µm there is a node, followed by an anti-node between 20 µm and 120 µm,
then another node is reached at around 140 µm. This is then followed by another anti-node.

Figuur 4.6: Plot of the driven frequency response of the Long wide cantilever at the 0th mode of vibration.

The highest peaks of each length-interval are extracted and plotted in figure 4.7 for a more schematic
overview, this image clearly shows the described pattern.
4.2. Frequency response of the cantilever 9

Figuur 4.7: Plot of the peaks of the driven frequency response of the Long wide cantilever at the 0th mode of vibration.

The same procedure is done on the 2nd vibrational mode. Figure 4.8 shows the frequency response
over the entire length of the cantilever, again showing the nodes and subsequent anti-nodes. As done
previously, the peaks at each length are extracted and plotted once more in figure 4.9,

Figuur 4.8: Plot of the driven frequency response of the Long wide cantilever at the 2nd mode of vibration.

Figuur 4.9: Plot of the peaks of the driven frequency response of the Long wide cantilever at the 2nd mode of vibration.
 5
Conclusion
From the experiment done on the piezo stack, a suitable method of determining the responsivity is
known, but the amplitude of the piezostack remains unknown, making it impossible to determine the
responsivity. The cutoff frequency was determined to be around 137.5 ± 2.5 Hz as at this point the
number of peaks in the signal measured at the output per period of the driving signal was halved.
In the second experiment, looking at the cantilevers, the oscillations in the long wide cantilever have
been mapped at two different modes of vibration. These images show a clear pattern of nodes and anti-
nodes over the length of the cantilever. Another important conclusion is that the theoretical model used
to describe the vibration of the cantilever and to it’s resonance frequencies ignores some important
environmental effects and experimental characteristics which effect the cantilevers, in particular the
Young’s modulus thereof. Thus the model is too simplified to give an accurate description of the driven
oscillatory motion of the cantilever.

10
 References
[1] Piezo Direct. What Is A Piezo Actuator? | Manufacturer and Supplier for Piezo Actuators. URL:
https://piezodirect.com/piezo-actuators/ (visited on 06/20/2024).
[2] Bolderink G. and Besseling T. Using Interferometry to Characterise Piezoelectric Materials and
Evaluate Mechanical Properties of Microscopic Features. 2016.
[3] HyperPhysics. Fabry-Perot Interferometer. URL: http://hyperphysics.gsu.edu/hbase/phyopt/
fabry.html (visited on 06/28/2024).
[4] Nanoworld. Pyrex-Nitride Rectangular Silicon-Nitride Cantilever AFM Tip PNP-DB. 2024. URL:
https://www.nanoworld.com/pyrex- nitride- rectangular- silicon- nitride- cantilever-
afm-tip-pnp-db (visited on 08/01/2024).
[5] H. O. Pierson. LPCVD SiN (stoichiometric). n.d. URL: http://www.mit.edu/~6.777/matprops/
lpcvd_sin2.htm (visited on 08/01/2024).
[6] Xu Q. and Tam L.M. “Mechanical Design of Piezoelectric Energy Harvesters Generating Electricity
from Human Walking”. In: (2022), pp. 93–113. DOI: 10.1016/b978-0-12-823364-1.00001-8.

11
 Theoretically determined roots and
A
resonance frequencies of the cantilever

n αn fn (kHz)
0 1.875 119.0
1 4.694 745.8
2 7.855 2088.2
3 10.996 4092.1
4 14.137 6764.6
5 17.279 10105.1

Tabel A.1: The first 6 resonance frequencies with their corresponding αn values of the cantilever with Young’s Modulus E =
270 GPa (silicon nitride), density ρ = 3184 kg m−3 and dimensions: L = 100 µm, w = 20 µm [4], h = 0.8 µm

12