47302

Assignment 4

by
Clara Sachmann s203942
Ida Hvidsten s194096
Jasraunaq Singh s242662

27665 LapChip: design of Lab-on-a-chip systems

Technical University of Denmark
9. December 2024
Contents

1 Introduction 1

2 Theory and calculations 1

2.1 Dielectrophoresis (DEP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2.2 Particle manipulation by DEP force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

3 Impedance simulation 3
3.1 Set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
3.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

4 DEP simulation 8
4.1 Set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

5 Discussion and comparison 15

6 Conclusion 16

References 17

A The obtained signal when varying the beads’ position in the channel for diameter 8 and 12
µm. 18

B The calculated value for electric diameter and pulse amplitude relative difference when
varying the beads’ position in the channel for diameter 8 and 12 µm. 18
1 Introduction
Microfluidic impedance cytometry (MIC) and dielectrophoresis (DEP) are popular methods to achieve selective
particle separation, where the MIC acts a sensor and the DEP functions as an actuator. The MIC-sensor
characterize the particles and the DEP-actuator uses the characterization information to decide on how to
manipulate their trajectory to achieve particle sorting. This method is useful for particle sorting as it is label-
free, non-invasive, low-cost and require minimal sample processing [1, 2].
The article ”Real-time impedance-activated dielectrophoretic actuation for reconfigurable manipulation of single
flowing particles” [1] utilizes the two methods, MIC-sensing and DEP-actuation, in a coplanar electrode layout
for selective single-particle manipulation by implementing a control center. The control center is programmed
to do real-time processing, where it evaluates the signal from the MIC-sensor and characterize the particle
properties such as size and position. Next, it will instruct the DEP-actuator to achieve the required particle
separation for the particle with these specific properties. In the article, three actions are programmed for the
actuator; particle position swapping, size-based particle separation and sorting of a particle sequence. The
particles used for their experiments have a diameter 8, 10 and 12 µm.
In our report, we will simulate the experiments in COMSOL with the same properties and channel dimensions
used in the article. This will be done by simulating the MIC-sensor and DEP-actuator separately. However, we
will only focus on the first two actions; particle position swapping and size-based particle separation.

2 Theory and calculations

The structure and electric circuit for the impedance part consists of two sets of coplanar electrodes with
asymmetric wiring configuration shown on figure 1 [1]. The specific electric configuration results in two different
types of currents, cross current between the same-side electrodes and transversal currents between the opposite-
facing electrodes. This is sketched on figure 1 by the black dotted field lines.

Figure 1: Picture of the electric potential and field lines for the specific wiring configuration. The setup of the
of the simulation is explained in section 3.1.

The benefit of the two types of currents is that they characterize different aspects of a particle when it travels
through the channel and disturbs the field lines. The transversal current will mainly be influenced by the size
of the particle, much like front facing electrodes. This is because the position-dependence of the transversal
current from fringing of the non-uniform electric field is assumed to be negligible where the transversal current
dominates (see its dotted line on figure 1). Therefore, the particle’s lateral position shouldn’t influence the
transversal current contribution to the final measured current. Thus the obtained signal amplitude when the

1
particles crosses the transversal current will only reveal information about the particle size. The transversal
current contribution can be expressed by the average amplitude of the two peaks.
a1 + a2
aavg = (1)
2
From this value, an estimation of the bead diameter called the electrical diameter can be expressed because the
signal amplitude at low frequency is dependent on the particle volume, thus the particle size [2]. The electrical
diameter also accounts for the electronic circuitry by the constant G, which is found by experimental work and
influenced by several factors such as the electrode configuration, AC voltage, electrode double-layer capacitance,
buffer conductivity, etc. [2].
D = G|a|1/3 (2)
When a particle crosses and interferes with the cross currents, the lateral particle position will affect the cross
current contribution, as the cross current field lines differ along the width of the channel. The closer the particle
is to the channel wall when it interferes with the cross current, the bigger influence it will have on the measured
signal, much like regular co-planar electrodes. This can be seen on figure 1, where the cross current field lines
change across the width of the channel. This means that the cross current contribution reveals information
about the lateral placement of the particle and can be expressed by the difference in the maximum signal
amplitude of the two peaks from the detected signal, that is obtained when the particle crosses the two pairs of
opposite-facing electrodes.
a = a2 − a1 (3)
The ratio of the the two contributions, (1) and (3), is called the pulse amplitude relative difference and holds
information about the lateral position of the particle and its size [2].
a2 − a1
∆= a2 +a1 (4)
2

If the cross current contribution is negative, meaning the amplitude of the second peak signal is bigger than
the amplitude of the first peak, the particles is displaced closer to the top channel wall with the 1st ground
electrode (see figure 1). This will also result in a negative ∆-value. If the value is positive, the particle is
displaced closer to the bottom channel wall with the 2nd ground electrode. Therefore, the sign of the ∆ value
reveals information about the lateral off-centering of the particle [1].

2.1 Dielectrophoresis (DEP)

In this system, DEP is used for precise control of particle trajectories. It is implemented by generating cross-
electric field lines via multiple sets of coplanar electrodes with opposite polarities. This generates a non-uniform
electric field, which induces the motion of the polarized beads. The electric field generation necessary for DEP
is generated by applying voltage pairs V1 and V2 to electrodes on opposite sides of the microfluidic channel. If
the permittivity of the medium is larger than the permittivity of the particle, the particles experience negative
DEP. This means they get repelled from regions where the electric field is large, e.g. the channel walls where the
electrodes are placed. This will effect the location of the resulting equilibrium position, that can be controlled
across the channel width by manipulating the ratio of the voltage of the opposite-facing electrodes. The
actuating ratio determines the trajectory of the particle based on its properties. The time-averaged DEP force
on a spherical particle is given by:
FDEP = πϵm r3 Re[K(ω)]∇|Ey |2 (5)
Where ϵm is the permittivity of the medium, r is the particle radius, Re[K(ω)] is the real part of the Clausius-
Mossotti factor and ∇|E|2 is the gradient of the squared electric field magnitude in the y direction. Here, the
E-field is given as the applied electric field amplitude and not the root mean square value (RMS).
The Clausius-Mossotti factor in equation (5) determines the direction and strength of the DEP force and can
be calculated by the following equation.
ϵp − ϵm
K(ω) = (6)
ϵp + 2ϵm
Where ϵp and ϵm are the complex permittivity of the particle and medium. The sign and magnitude of
Re[K(ω)] determines the direction and strength of the DEP force. When Re[K(ω)] < 0, which occurs when the

2
permittivity of the medium is larger than the particle, the particles experience negative DEP and move towards
regions of lower electric field strength. When choosing to either use RMS or the peak electric field amplitude
values in the DEP force equation, the factor of 2 will be multiplied if RMS is used as the RMS value includes
an implicit averaging over the AC signal’s cycle. Using peak electric field values directly avoids this averaging,
leading to a more straightforward calculation of the instantaneous DEP force .

2.2 Particle manipulation by DEP force

The DEP force can be controlled and used to manipulate the particle trajectories and the equilibrium position
by changing the voltages of the opposite-facing electrodes, referred to as V1 and V2 . The possible outcomes are
listed below.

• V1 > V2 : Equilibrium position occurs in the bottom half of the channel (y < 0)
• V1 < V2 : Equilibrium position occurs in the top half of the channel (y > 0)
• V1 = V2 : Particle focusing along the channel axis

These configurations can be used to change the action of the system, where the actions will be referred to
as modes. In the article [1], three primary operational modes are demonstrated, each customized for specific
particle manipulation tasks.
The first mode, Particle Position Swapping, enables the displacement of particles entering the channel on
one side to the opposite side. The MIC-sensing region identifies the lateral position of the particle, represented
by the metric ∆. The DEP-actuation region then applies appropriate DEP voltages to shift the particle’s
trajectory. For instance, when ∆ < 0, the particle is initially located in the lower half of the channel and is
displaced to the upper half by setting V1 < V2 . Conversely, when ∆ > 0, the particle is in the upper half and
is displaced to the lower half by setting V1 > V2 . This results in track swapping.
The second mode, Size-Based Separation, sorts particles into size categories such as small, medium, and
large, based on their corrected electrical diameter D-corr. Each size category is assigned a specific equilibrium
track within the channel. Small particles, where D-corr < 9 µm, are directed to the center track. Medium
particles, for which 9 µm ≤ D-corr < 11 µm, are directed to the upper track. Large particles, with D-corr ≥ 11
µm, are directed to the lower track. By adjusting the applied voltages V1 and V2 , the system controls the
particle trajectories, achieving size-based sorting.
The third mode, Sequence Sorting, sort the particles based on a predefined sequence, such as alternating
sizes. The MIC-sensing region identifies the size of each incoming particle and the DEP-actuator displaces only
the particles matching the target sequence to the desired track. The other particles gets discarded to a second
track. As mentioned in the introduction section 1, this mode will not be investigated in this report.
Real-Time Signal Processing and Control
The integration of MIC-sensing and DEP-actuation is facilitated in the article [1] by a real-time control algo-
rithm, which operates through three distinct states. In the first state, Wait for New Event, the system monitors
the MIC-sensing region for the presence of a particle. Upon detection, it transitions to the second state, Peak
Detection, where the signal peaks are identified and the values of ∆ and D-corr are computed. In the final
state, Actuation, the system adjusts the voltages V1 and V2 based on the particle’s properties and the selected
operational mode.

3 Impedance simulation

3.1 Set-up
The impedance part of the sensor is simulated by use of an AC current with the Electric Currents physics and
Frequency Domain study at 3 MHz. The structure is modeled in 2D with use of shallow channel approximation
with height of 21.5 µm. The main channel has a length of 200 µm in the x-direction and width of 50 µm in
the y-direction. Two coplanar electrodes are facing each other, recessed 15 µm away from the main channel

3
walls, and have a width of 30 µm and a pitch of 60 µm. The ground electrodes are placed at the top left and
bottom right electrodes by a Terminal boundary condition with 0 V. The terminals are set at the bottom left
and top right electrodes with Terminal at 6 V. This is indicated on figure 2, where the grounds are marked blue
and the terminals are marked red. In figure 2, the bead diameter is 10 µm and it is placed at the coordinates
(x0 , y0 ) = (25, 25) µm.

Figure 2: Picture of geometry for the 2D model of the impedance part. The ground electrodes are marked red
and the terminal electrodes at 6 V are marked blue.

The material properties for the medium and beads are listed in table 1.

Relative permitivitty εr Conductivity σ [S/m]

Medium 80 0.9
Bead 2.5 6.7·10−4

Table 1: Table with the material parameters for the medium and the beads.

The mesh is a Fluid dynamics calibrated mesh with Extra fine element size. This mesh-size was deemed
sufficient, as simulations with Extremely fine element size gave the same data results.

Figure 3: Picture of the mesh with Extra fine element size for the 2D model of the impedance part.

The system is investigated by probing the current at the two ground electrodes and measure the difference of
the currents, Isignal = I2 − I1 . Here, I1 is the detected signal at the left electrode and I2 is the signal at the
right electrode. The current is measured by placing a Domain Point Probe at the middle of each of the two
ground electrodes and detect the absolute value of the current, abs(ec.I0).
Two different cases have been simulated:

• Investigation of the bead size d.

• Investigation of the position of the bead across the width of the channel, its y0 -coordinate.

4
For both investigations, the x0 -coordinate is changed by a parametric sweep to simulate the bead moving across
the length of the main channel from x0 = 25µm to x0 = 175µm with step size 2.5 µm.

3.2 Results
To investigate the influence of the bead-size, the bead diameter d is simulated for when d equals 8, 10 and 12
µm. The y0 -coordinate is fixed at 25 µm in the middle of the channel. The results are shown on figure 4 below.

Figure 4: Measured current difference at the two ground electrodes, where the influence of the bead size d is
investigated. The width-position of the beads is y0 = 25µm. Note, that r in the figure is the radius, r = d/2.

The graph shows, that bigger bead sizes result in larger signal. This is because a bigger volume will interfere
more with the electric field, which leads to a larger difference between the measured current the two electrodes
and thus a larger signal.
The influence of the bead position, its y0 -coordinate, is investigated by simulating the bead’s journey across the
length of the main channel, where the bead is placed at y0 = 7.5, 15, 25, 35, 42.5 µm. The bead size is fixed
at d = 10 µm. The plots for diameter 8 µm and 12 µm are shown in appendix A.

Figure 5: Measured current difference at the two ground electrodes, where the influence of the y0 -coordinate of
the bead is investigated. The bead size is d = 10µm for the exhibited figure. The plots for the other diameters
can be found in appendix A.

5
It can be seen, that the signal indeed is influenced by its y0 position. The closer the bead is to the electrodes
and the channel walls, y0 = 7.5 µm, y = 42.5 µm, the larger signal is detected. This shows, that it is important
to take the position into account when analyzing the results and trying to find the bead size.

• Investigation of the bead size d from figure 4

To investigate the bead size, the absolute maximum and minimum values for the peaks are extracted from the
dataset (which was plotted on figure 4) to calculate the relative amplitude a using equation (3), the electric
diameter D using equation (2) and the pulse amplitude relative difference ∆ using equation (4). The G value
used to calculate the electric diameter is set to 950 Aum
1/3 as given by the article [1]. The calculated values are

listed in table 2, where it can be seen that the ∆-values are close to zero. This is as expected as the particles are
fixed in the middle of the channel for these simulations where the off-centering of the position should be equal
to 0, which the ∆-value should reflect. It is also seen that the relative amplitude increases for larger diameter.
This is expected as bigger volume interfering with the electric field results in bigger signal. The calculated
electric diameter D does not match the exact diameter exactly, however the value is significantly different for
the chosen bead sizes, which shows that the sorting based on the bead size should be possible.

Diameter [um] a [A] D [um] ∆

8 8.51 · 10−7 9.00 −4.78 · 10−6
10 1.33 · 10−6 10.44 −3.13 · 10−6
12 1.90 · 10−6 11.77 −1.92 · 10−6

Table 2: Calculations for the simulations of different diameter. The data plotted on figure 4 is used for the
calculations.

• Investigation of the bead size d from figure 5

The same equations as before are used to calculate the electric diameter (2) and the pulse amplitude relative
difference (4) for the signals obtained on figure 5. These results are from the bead with d = 10µm that travels
across the width of the channel. The results can be seen in table 3. The relative amplitude a is largest for the
particles closest to the channel walls. The ∆-value is positive for particles near the bottom of the chamber and
negative for particles near the top of chamber. The electric diameter values is increasing near the channel walls
and lowest in the middle of the chamber with a value of 10.44 um, where the correct diameter value for the
beads would be 10 um.

Position [um] a [A] D [um] ∆

7.5 3.11 · 10−6 13.87 0.93
15 1.72 · 10−6 11.38 0.61
25 1.33 · 10−6 10.44 −3.13 · 10−6
35 1.72 · 10−6 11.38 -0.61
42.5 3.11 · 10−6 13.87 -0.93

Table 3: Calculations for different positions for bead size d = 10µm.

To take the y0 -position into account, a more accurate diameter can be calculated from the electrical diameter
D, referred to as the corrected diameter D-corr. To find this value, the pulse amplitude relative difference ∆
versus the electrical diameter D is plotted. This will reveal a relation between the two parameters, where the
accurate D-corr can be extracted. At first, only the bead with size d = 10µm is investigated. The results are
shown in figure 6.

6
 Figure 6: Data points plotted along with their fitted curve, where the curve-parameters are listed in table 4.

Looking at the data points in figure 6 it is clear that there is a parabolic relation between the ∆ value and the
electric diameter. This can be used to set up a fitting equation to find the parameters that reveal D-corr. The
equation used in the article [1] for the fitting curve and figure 6 is shown below.
h i
D = a · 1 + b · (∆ − c)2 (7)

Here, b and c are the calibration constants and a is D-corr. The calibration constant for the fitted curve on
plot 6 are listed in table 4 below.

Fitting parameters for d = 10 µm

a/D-corr (10.1 ± 0.3) µm
b 0.42 ± 0.06
c 1.41 · 10−5 ± 0.03
R2 -value 0.98

Table 4: The fitting parameters for the fitted curve written as equation (7), displayed on figure 6.

The extracted D-corr value is 10.1 µm with an uncertainty of 0.3 µm for the exact d-value of 10 µm. The
calibration parameters are found to be b = 0.42 ± 0.06 and c = 1.41 · 10−5 ± 0.03. However, note that parameter
c is close to zero and its uncertainty is larger than the value. This implies, that the chosen fit (7) used in the
article [1] is not the best fit for these data points. Instead, equation (8) is used as the fitting curve, where
calibration coefficient c is excluded. h i
D = a · 1 + b · ∆2 (8)

The new fit with the data points are shown on figure 7 along with the results for the other bead sizes. The data
used to calculate the data points for diameter 8 and 12 µm is plotted in appendix A and the calculated D- and
∆-values are listed in appendix B.

7
 Figure 7: Data points plotted along with their fitted curve, where the curve-parameters are listed in table 5.

The fitting parameters for the fitted curves are displayed in table 5.

Fitting parameters for d = 8 µm Fitting parameters for d = 10 µm Fitting parameters for d = 12 µm

a/D-corr (8.7 ± 0.2) µm (10.1 ± 0.3) µm (11.3 ± 0.4) µm
b 0.41 ± 0.05 0.42 ± 0.05 0.45 ± 0.07
R2 -value 0.98 0.98 0.98

Table 5: The fitting parameters for the fitted curves written as equation (8), displayed on figure 7.

The D-corr values found from the new fits are 8.7 ± 0.2 µm for the exact value d = 8 µm, 10.1 ± 0.3 µm for
d = 10 µm and 11.3 ± 0.4 µm for d = 12 µm. The D-corr value for d = 10 µm show very good result, as the
exact diameter is found within the uncertainty. However, the other D-corr values for d = 8, 12 µm only come
close to their exact values within the uncertainty.

4 DEP simulation

4.1 Set-up
The DEP part of the microfluidic system is set up with a 2D model geometry with a main channel length of
700 µm in the x-direction. Six lateral channels, each with an electrode, are extended from the main channel in
the y-direction and evenly spaced with a pitch of 60 µm. The geometry is shown below in figure 8 along with
the electric wiring configuration.

Figure 8: Picture of geometry for the 2D model of the DEP part with the electric wiring configuration.

The blank material feature is used to define the medium material with the custom variables as detailed below
in table 6.

8
 Relative Conductivity Density Dynamic Particle
permittivity εr σ [S/m] ρ [g/mL] viscosity η [mPa·s] diameter d [µm]
Medium 80 0.9 1.04 2.0
Bead 2.5 6.7·10−4 1.05 (8, 10, 12)

Table 6: Table with the material properties for the medium and the beads used in the DEP simulation. The
medium is described as a PBS buffer augmented with 13% sucrose with a density close to the beads [1]. The
medium density values is therefore an approximation.

The primary objective is to achieve the desired particle movement and thus the final particle position based
on the DEP force established due to the specific electrical configuration. This is achieved by implementing 3
physics and two different studies. The first study, called Study 1, includes a Stationary and Frequency Domain
step, operating at 0.4 MHz. The second study, Study 2, includes a Time Dependent step in the range of 0s to
0.5s with steps of 0.005s. Detailed descriptions of the physics are listed below.

Laminar Flow physics

The 2D model is simulated as 3D by using the shallow channel approximation feature with a channel
thickness of 21.5 µm. The inlet is placed at the furthermost left channel boundary and the outlet is placed
at the right boundary. At the inlet a fully developed flow is selected as a boundary condition with a flow
rate of 100 nL/min. The static pressure is set to 0 Pa at the outlet.
Electrostatics physics
An Out-Of-Plane thickness is implemented with a thickness of 21.5 µm. The electric wiring configuration
is set up by corresponding positive and negative voltage electrode pairs with the boundary condition
Terminal. The wiring configuration is illustrated on figure 8. 3 different actuation voltage pairs are
considered and investigated.
(i) (V1 , V2 ) = (1.5, 10.5) V
(ii) (V1 , V2 ) = (7.5, 7.5) V
(iii) (V1 , V2 ) = (10.5, 1.5) V
Particle Tracing for Fluid Flow physics
The particle properties for the beads, such as their density, diameter, relative permittivity and electrical
conductivity are inserted in the physics in the Particle Properties step. Their values are listed in table 6.
The inlet is set to the left boundary, where 15 particles will be released and the outlet is set to the right
boundary with the Freeze Wall condition. Dielectrophoretic Force and Drag Force are added and their
material properties are set to From material.

The mesh is defined as a Fluid dynamics calibrated mesh with normal element size. The mesh size at the
boundaries of the lateral channels is set to finer to ensure sufficient mesh near the electrodes for more accurate
simulation of the DEP force.

Figure 9: Picture of the mesh for the 2D model of the DEP part.

To investigate the trajectory of the released particles due to the DEP force, two different models are made:

• The DEP force for the different electron configurations is visualized by a surface plot of the magnitude of
the DEP force in the y-direction, which is the direction in which the particles will be pushed.

9
 • The trajectory of the particles is investigated by looking at their position at the beginning and at the end
of the simulation. This is visualized with 1D plot histograms, which depict the lateral positions of the
particles at the beginning of the simulation (t = 0s) and when the particles have reached the outlet wall
(t = 0.5s).

Both models are simulated for each of the three configurations of the voltage pairs.

4.2 Results

Equation (5) is used to visualize the DEP force with the following constant values: ϵm = 80, ϵ0 = 8.854 · 10−12
F/m, r = 5 um and Re[K(ω)] is set to -0.5 as done in the article [1]. The ∇|Ey |2 value using the amplitude of
the electric field depend on the position in the channel and proximity to the electrodes. This value depends on
the set up of the electrodes in the COMSOL simulation.
The amplitude of the electric field can be seen in figure 10, and shows where the amplitude of the E-field (its
y-component) is greatest. On figure 10a it is clear that the field is strongest at the channel wall at y+ -position
and close to zero at the wall at y− -position. Figure 10c display the opposite, where the field is strongest at the
wall at y− -position. Figure 10b shows the field, where four spots at the middle of the channel is close to zero, as
the electrodes at the channel walls have equal potential. The current runs across the channel and not between
the directly opposite electrodes, which creates these spots of zero E-field in the middle of the channel.

(a) E-field for the DEP force for (V1, V2)=(10.5, 1.5)V

(b) E-field for the DEP force for (V1, V2)=(7.5, 7.5)V

(c) E-field for the DEP force for (V1, V2)=(1.5, 10.5)V

Figure 10: E-fields displaying the DEP force for different electron configurations.

Using the electric field amplitude, it is possible to simulate the DEP force in the y-direction across the channel
and visualize the results in a 2D surface plot. If the DEP force is plotted with no editing, the surface plot will
look like figure 11a with the configuration of (V1 , V2 ) = (10.5, 1.5)V. Here, the DEP force in the channel looks

10
to be zero at the entire surface except for the corners of the electrodes at the channel walls in the y+ -part of
the channel. This is due to the magnitude of the colour bar scale that is of the magnitude 105 pN. For a more
descriptive surface plot, a fixed range can be applied to the colour bar so that any value that is above the limit
will be coloured as the limit value.
Two examples of this can be seen in figures 11b and 11c, where figure 11b is plotted with a colourbar with
limits of -60 pN to 60 pN as the article [1], while figure 11c is plotted with a colourbar with limits of -600 pN to
600 pN. This makes the DEP force look constant at the y+ -part of the channel and close to zero for the rest of
the channel. However, this is only visible after manipulation of the colourbar scaling. Figure 11c is the result
from the simulation that comes closest to figure 3a in the article [1], even though the plot in this report displays
larger magnitudes.
Using the colourbar scale of -600 pN to 600 pN, the three possible electrode configuration can be seen in figure 11,
where the 12a shows a negative DEP force at y+ -positions, the 12c shows a positive DEP force at y− -positions
and the 12b shows both a positive and a negative DEP force that meets in the middle of the channel.

(a) DEP force with no set range on the DEP force colour bar. The range is from −2 · 105 pN to
3.7 · 105 pN.

(b) DEP force using the colour bar range used in the article [1]. This bar is from -60 pN to 60
pN.

(c) DEP force using a range that replicate the images from the articles. This bar is from -600
pN to 600 pN.

Figure 11: The effect the limits of the colourbar when plotting the DEP force in the y-direction in the
channel. This is electron configuration (V1 , V2 ) = (10.5, 1.5)V.

11
 (a) Distribution of the y-component of the DEP force for (V1 , V2 ) = (10.5, 1.5)V.

(b) Distribution of the y-component of the DEP force for (V1 , V2 ) = (7.5, 7.5)V.

(c) Distribution of the y-component of the DEP force for (V1 , V2 ) = (1.5, 10.5)V.

Figure 12: Distributions of the y-component of the DEP force. In the article, the axes are switched compared
to this report. This means, that their results for FDEP,x is the same for FDEP,y in this report.

For a more quantitative measurement of the DEP force, the DEP force averages of 50 lines along the channel are
computed, where the lines range from x = 0 um to x = 500 um across the channel. The 50 lines are displaced
across the width of the channel from y = −25 µm to y = 25 µm with a line for every 1 um. In figure 13 an
illustration shows the lines that the average values are measured over. The data is extracted for each line from
COMSOL. The average values for the lines are calculated, resulting in one data point for every line. The total
of 50 lines equals 50 data points. This is repeated for each of the three configurations, resulting in a total of
three graphs, one for each configuration. The results are displayed in figure 14, where the x-value of the plot is
the average DEP force value and the y-value is the position.

Figure 13: Illustration of the position and length of the lines used for line average measurement.

In the figures 14 the left figure shows the full range of average DEP force values for the three configurations.
The black line shows the (V1 , V2 ) = (10.5, 1.5)V configuration, where the DEP force average is negative at the
channel wall in the y+ -part and close to zero at the rest of the channel. The opposite is true for configuration
(V1 , V2 ) = (1.5, 10.5)V (blue graph), that displays positive DEP force average for y− -values and close the zero
at the rest off the channel. The last line is configuration (V1 , V2 ) = (7.5, 7.5)V that shows symmetric positive
and negative DEP force average.

12
For a result that is closer to the graphs shown in the article [1], the x-axis is constrained to only show the DEP
force value between -600 pN to 600 pN. This gives a plot that is very similar to the plot in the article. The only
major difference is that their values are 10 times smaller and the data is plotted as a line, not as data points as
in figure 14.

(a) (b)

Figure 14: Plot displaying the average-value of DEP force computed for lines across the channel.

To examine the DEP-actuator, simulation with DEP was employed to visualize and study its effect on particle
positions. The start- and end-positions of particles as they moved across the lateral vertical axis were examined
using a histogram for different electron configurations. 15 particles of diameter 10 µm were uniformly distributed
across the vertical axis at the left side of the channel. It is observed, that at voltage pair (V1 , V2 ) = (1,5, 10.5)V,
the particles are pushed towards y+ -positions of the microfluidic channel, seen on figure 15a. For the voltage
pair (V1 , V2 ) = (7.5, 7.5)V, the particles are adjusted to the middle of the channel, as seen in figure 15b and
for voltage pair (V1 , V2 ) = (10.5, 1.5)V, the particles move to y+ -part of the channel, shown in figure 15c.
The modes of operation, which was explained in section 2.2, can be demonstrated with these prescribed voltage
pairs:

• Particle Position Swapping

Particle position swapping, i.e. the movement of a particle from track 1, y+ -part of the channel, to track 2,
y− -part of the channel, (or vice versa), occurs effectively for particles with diameter of 10 µm as observed
in the histograms on figure 15. If experiments were made using the mode operations as in the article,
when the value of ∆ is negative because the particle is positioned at y+ -coordinate, the microprocessor
actuates the voltage pair (10.5, 1.5)V. This pushes the particle to the lower half of the channel. A similar
pattern can be equipped for when the ∆-value is positive, where the voltage pair (1.5, 10.5)V will push
the particle to the upper half of the channel, switching its position.

13
 (a) Particle distribution histogram (1.5, 10.5)V. (b) Particle distribution histogram (7.5, 7.5)V.

(c) Particle distribution histogram (10.5, 1.5)V.

Figure 15: Particle distribution histograms for different voltage configurations, (V1 , V2 ). The simulated
particles have a diameter of 10 µm.

• Size-Based Particle Separation

For this operation, the particles are classified small, medium and large based on the electrical diameter
value. The study [1] employs optimized DEP actuation with slightly different voltage configurations
compared to previous simulation, such as (0, 10.5 V), (10.5, 0 V), and (7.5, 7.5 V). This creates larger
electric field gradients that effectively manipulate the particles as the voltage pairs (0, 10.5)V and (10.5,
0)V result in a stronger DEP force. This stronger force allows for more efficient and accurate particle
manipulation. Particle trajectories for sizes 10 µm and 12 µm were simulated with voltage pairs (0, 10.5)V
and (10.5, 0)V respectively. If this mode-operation was implemented in experiments, incoming particles
that have been characterized as medium size due to the computed D-corr value from the MIC-sensor, are
shifted to the upper half of the channel. This is seen on figure 16a for particles with diameter of 10 µm.
Whereas if the particles were characterized as large sized , the would be shifted to a different track in the
lower half, as seen on figure 16b with particles of diameter 12 µm. These simulations were made with 5
particles.

14
 (a) Particle distribution histogram (0, 10.5)V for (b) Particle distribution histogram (10.5, 0)V for
particle size 10 µm. particle size 12 µm.

Figure 16: Particle distribution histograms for different voltage configurations and bead size. The simulations
were made with 5 particles.

5 Discussion and comparison

It was seen in the results for the MIC-section in section 3.2, that it was possible to differentiate between the
sizes of the incoming particles and estimate their diameter. The values for D-corr had thresholds at 9 µm and
11 µm, which could be used to distinguish small, medium and larger particles as done in the article. However,
when comparing our simulation results with the exact diameter, it was only one of the three bead sizes where
the calculated D-corr values was found within the uncertainty. A reason for the error margin could be due to
the experimental G-factor used to calculate the electric diameter D. Since this factor was found experimentally
from calibration in article [1], it might not be a good representative for our simulated system. Another reason
could be because our simulations are modeled in 2D. This will not only simplify the geometry of the system
but in some cases also distort the model. E.g. the modeled beads are inserted by circles with the theoretical
diameter value. However, when the model is extended to 3D by shallow channel approximation, these circles
becomes cylinders, which is not a good representative shape for the beads. Another reason is that the electrodes
also gets extended with the shallow channel approximation. This means, that the edge boundaries where the
electrodes are defined becomes walls, and not plane electrodes on the bottom surface as done in the article [1].
When comparing the simulation with the experimental results from the article [1], it is not possible to compare
the calculated electric diameters, since the article never discloses these results. This is because they focus on
linking the MIC-sensor to the DEP-actuator, and therefore only uses the D-corr values for the DEP settings
instead of comparing with the calculated diameter with the exact value. Instead, it is possible to compare
the calibration factors b and c. In the article they got b = 0.36 and c = 0.03 from their experiments. In our
simulation, the values are found to b = 0.42 ± 0.06 and c = 1.41 · 10−5 ± 0.03. It can be seen, that the b-value
aligns with the article’s value within one uncertainty. However, the c-value does not correspond to their value,
and because of the uncertainty, the constant was later excluded from our results. To evaluate whether the
c-value should also be excluded for their fitting curve, it would be useful if they displayed the uncertainty for
their calibration factors or at least the correlation R2 value to estimate how well their fitting curve values align
with the data.
For the simulation of the DEP force, there could be multiple explanations for the discrepancy between the
magnitude of the force, that the article [1] finds during the simulation compared to the results from our work,
which had almost ten times higher magnitude.
The one of the more likely differences between the two simulations is the set up of the electrodes for the
simulation of the DEP. Firstly, they use a 3D set up, while we use a 2D set up with a height defined by the
shallow channel approximation. In their work it is unclear how the electrodes are set up in the 3D space, as
only a 2D sketch of the channel is shown. Either the electrodes are flat areas of the bottom of the channel or
the electrodes have a height, meaning that the electrodes are simulated as boxes at the side of the channel. If

15
the box electrode is used then the set up is very similar to ours where we have a wall electrode at the side of the
channel that will be extended upwards during the shallow channel approximation. If the electrodes are planes
at the bottom of the channel then it is relevant at which plane the simulation of the x-component of the DEP
force is simulated at. The results will differ according to whether this plane is positioned at the bottom of the
channel, at the same plane as the electrodes or whether it is placed at the middle height of the channel. If the
plane was positioned in the same plane as the electrodes, then that would be similar to our set up. However, if
the simulated planes is positioned in the middle of the channel, then it is likely that the computed force would
be less than the results from our 2D simulation.
The simulation depend on the equation of the DEP force, were the constants are kept the same as in their work.
The most unclear part of this equation is if they used RMS values for the electrodes or not, as the equation
for the DEP force given by the article shows a electric field measured in RMS. This leads to the question
whether the given potentials are also given in RMS-values. If the article uses RMS values, it would account for
a difference as the difference between peak amplitude value and RMS-value is √12 .
For the comparison of the image of the DEP force given in their article in their fig. 3, it is never mentioned
in the article whether they have set a range on the colourbar for the image to show a better DEP force image.
However, looking at their graph for average value DEP force (in their fig. 3b), it is clear that they have used
limits of -60 pN to 60 pN, which does not display the full extend of their data. This was not clearly stated in
the article.
Additionally, it can be noted from observations made from the histograms that particle size affects the final
position in size-based sorting, while the paper [1] successfully demonstrates and examines the various operations
related to particle separation and sorting, but does not explicitly talk about the relationship between particle
size and end position. The DEP force is directly proportional to the third power of the particle radius which
means that the larger particles experience a stronger DEP force. This makes it easier for the larger particles
to move to their designated tracks compared to smaller particles. This is not discussed in the article but
would affect how close to the channel walls each particle size would be pulled. Positional inaccuracies may
be introduced due to the overshooting of larger particles toward the walls of the microfluidic channel and the
undershooting of medium-sized particles closer to the middle of the channel, for the same voltage configuration.
Integrating simulations or experimental studies that examine these aspects could provide insights for optimizing
multi-way particle sorting systems.

6 Conclusion
This study has successfully demonstrated simulations of the experimental setup and simulations done in the
article Real-time impedance-activated dielectrophoretic actuation for reconfigurable manipulation of single flowing
particles [1] and reconfirmed some of the results. While our results largely align with the trends, theoretical
expectations and experimental results, minute discrepancies can be observed stemming from the non-standard
aspects of the experimental setup i.e. that is the values that depend on the empirical nature of the setup. This
calls for further standardized research on device calibration and particle dynamics to enhance the reliability
and scalability of such systems for real-world applications. The D-corr/a values and the fitting parameters
accompanying it are found along with the DEP force, that displayed results with a magnitude of 10 times
higher than what is mentioned in the article [1]. Despite these challenges, the system effectively sorted particles
based on their corrected electrical diameters, showcasing its potential for label-free, real-time applications in
diagnostics and materials science.

16
References
[1] Alexis Lefevre, Cristian Brandi, Adele De Ninno, Filippo Ruggiero, Enrico Verona, Michaël Gauthier, Paolo
Bisegna, Aude Bolopion, and Federica Caselli. Real-time impedance-activated dielectrophoretic actuation
for reconfigurable manipulation of single flowing particles. Lab Chip, Volume 24, Issue 24, Pages 5145-5154,
DOI: http: // dx. doi. org/ 10. 1039/ D4LC00622D , 2024.
[2] Riccardo Reale Federica Caselli, Adele De Ninno, Luca Businaro, and Paolo Bisegna. A novel wiring scheme
for standard chips enabling high-accuracy impedance cytometry. Sensors and Actuators B: Chemical, Volume
256, Pages 580-589, DOI: https: // doi. org/ 10. 1016/ j. snb. 2017. 10. 113 , 2018.

17
A The obtained signal when varying the beads’ position in the chan-
nel for diameter 8 and 12 µm.

B The calculated value for electric diameter and pulse amplitude

relative difference when varying the beads’ position in the channel
for diameter 8 and 12 µm.

Calculated values for d = 8 µm

Position [um] a [A] D [um] ∆
7.5 1.94 · 10−6 11.85 0.92
15 1.10 · 10−6 9.80 0.61
25 8.51 · 10−7 9.00 −4.78 · 10−6
35 1.10 · 10−6 9.80 -0.61
42.5 1.94 · 10−6 11.85 -0.92

Calculated values for d = 12 µm

Position [um] a [A] D [um] ∆
7.5 4.70 · 10−6 15.91 0.93
15 2.47 · 10−6 12.85 0.61
25 1.90 · 10−6 11.77 −1.92 · 10−6
35 2.47 · 10−6 12.85 -0.61
42.5 4.70 · 10−6 15.91 -0.93

18