The Design of A Ferrofluid Magnetic Pipette
The Design of A Ferrofluid Magnetic Pipette
The Design of A Ferrofluid Magnetic Pipette
1
In Press, IEEE Trans. Biomedical En-
gineering. September, 1996
Nancy Greivell
Center for Bioengineering
University of Washington
Seattle, WA 98195
* Corresponding author
2
Abstract
An electromagnetic pipette using a ferrofluid was designed to sample liquid volumes smaller
than 0.2 microliter. Submicroliter sample sizes are desirable for reducing the amount of costly
reagents and reducing sample requirement for large-scale analysis. The pipette consists of four
electromagnets arranged such that air-gaps are aligned to accommodate a tube. A light-hydrocar-
bon-based ferrofluid is contained in the tube and acts as a plunger. The position of the ferrofluid in
the tube was controlled to within 0.2 mm by combining adjacent air-gap magnetic fields. The
position of the ferrofluid as a function of time and magnetic pressure as a function of position was
measured in one electromagnet air-gap from the device. Maximum pressure measured was 770
Pascals which corresponds to a maximum velocity of 0.9 cm/s. The assembled pipette weighs
approximately 25 grams and it measures 4 cm long, 1 cm wide, and 3 cm high.
1. Introduction
The handling of biological liquids for chemical analysis is a time-consuming and tedious job
in biochemical laboratories. Much of the liquid handling is done on automated instruments by a
device consisting of a plunger in a bore operated by a mechanical linear actuator. The discrete
movement of the plunger by the linear actuator samples a discrete volume of reagents and/or bio-
logical fluid. These conventional pipettes can sample an aliquot of liquid as small as 0.2 to 0.5
microliters with about 3-5% degree of precision. Smaller precision pipetting is problematic with
these types of mechanical pipettes.
New technological demands in biochemical laboratories have motivated the search for methods
of handling volumes of liquid smaller than 0.2 microliters. This is particularly true in the research
of the human genome and of proteins expressed by the human genome where biochemists are
working to automate current methods for large-scale analysis. Minimizing reagent and sample re-
quirements for such large-scale analysis would help reduce the cost of analysis. The ferrofluid-
magnetic pipette was designed to address the problem of handling submicroliter liquid volumes
smaller than 0.2 microliter. This pipette uses the properties of ferrofluids and the principles of mag-
netism to sample such quantities of liquid.
Ferrofluids are stable colloidal suspensions of magnetite in a liquid medium [1-3]. Suspending
magnetite in a liquid imparts unusual properties to the liquid. Forces are developed in ferrofluids
in the presence of a magnetic field. Because of this property, ferrofluids have found their way into
a variety of applications such as rotating shaft seals, sensors, and drug delivery in humans [4].
The idea of the ferrofluid-magnetic pipette is similar to a pumping device proposed by C.W.
Miller [5]. Millers proposed pump uses of a series of electromagnets to generate a gradient mag-
netic field which moves the ferrofluid through a length of a tube. This idea was proposed but not
actually built. In the ferrofluid-magnetic pipette, a discrete volume of ferrofluid is placed in a small
tube and acts as a plunger similar to the spring-loaded plunger in conventional pipettes. To sample
a small amount of fluid, the array of electromagnets is used to actuate a position change of the fer-
rofluid. The resolution of ferrofluid positioning and the inner diameter of the tubing determines the
volume of liquid sampled by the device.
This paper presents the theory, design, and testing of the ferrofluid-magnetic pipette device. An
analysis of experimental data of the ferrofluid in a gradient magnetic field is discussed. The goal
of the project was to develop a reasonable electromagnet configuration and a method of controlling
4
2. Theoretical Calculations
The analysis of the ferrofluid motion in a gradient magnetic field was important for designing
the pipette, evaluating its performance, and developing a method of high-resolution position con-
trol. This analysis involved computing the velocity of the ferrofluid from pressure measurements
to numerically describe the action of the pipette. This section presents the equations used to com-
pute this velocity.
The velocity depends on the type of ferrofluid used in the pipette and its properties. Table 1
summarizes the properties of EMG 901 ferrofluid - a suspension of magnetite in a light hydrocar-
bon based oil purchased from Ferrofluidics Corporation. It was used in the pipette design and used
in pressure and position measurements.
To analyze the ferrofluid in the gradient magnetic field, assumptions are made to simplify the
computations. The ferrofluid is assumed to be viscous and incompressible; as long as the windings
are sufficiently cooled, fluid flow is also considered to be isothermal [1]. M and H were assumed
to be colinear [3]; the gravity component was assumed negligible in the direction of ferrofluid
motion.
The modified ferrohydrodynamic Bernoulli equation [1, 6] is expressed in terms of fluid veloc-
ity and simplified to the Hagen-Poiseuille equation to compute ferrofluid velocity for the pipette
configuration. The well-known Hagen-Poiseuille equation for flow in a tube of circular cross-sec-
tion is
2
pr
= --------------- (1)
8L
where p is the pressure drop across the fluid, is the viscosity, r is the inner radius of the tube,
L is the length of the ferrofluid in motion [7, 8]. For average fluid velocity in a small diameter
tubing under varying pressure, such as the ferrofluid in a gradient-field, a quasi-steady Poiseuille
flow was computed with the above equation [7]. p in equation (1) is
H2
p = M dH (2)
H1
where M is the magnetization of the ferrofluid, H is the magnetic field, and is permeability.
5
Equation (1) assumes the ferrofluid flow in the pipette tubing to be laminar with a constant
pressure drop across the length of the tube. Laminar flow was verified by computing the Reynolds
number for a smooth walled tubing of inner diameter of less than or equal to 2 mm. The resulting
Reynolds number was less than 5 using the density and viscosity from Table 1 and a velocity of 1
cm/s. This velocity was estimated from observing the ferrofluid motion. The magnitude of the Rey-
nolds number that delimits laminar from turbulent flow is approximately 2000 [8].
Equation (2) is just the area under the ferrofluid magnetization curve. A direct application of
this equation with an appropriate demagnetization factor results in higher pressures than measured
pressures [5, 9]. Therefore, pressure was not computed to predict the ferrofluid motion; instead, we
relied on experimental data with equation (2) to qualitatively understand the ferrofluid response to
an electromagnet configuration.
3. Experiments
There were two sets of data collected which were critical for the pipette design: Magnetic pres-
sure as a function of ferrofluid position in one electromagnet air-gap; and ferrofluid position as a
function of time. Measured pressure was used to predict ferrofluid velocity in the gradient magnet-
ic field and to guide the electromagnet placement and the driving current. Measured position was
important for evaluating the pipette performance and for controlling ferrofluid position.
The magnetic pressure was measured using the set-up shown in Figure 1. This set-up included
one electromagnet from the pipette device and the Tefzel tubing containing a 22 mm long plug
of ferrofluid. Current in the windings was controlled such that the applied mmF was 96 5 amp-
(insert Fig 1)
turns (A-T). At the start of the pressure measurements, the ferrofluid moved to its equilibrium po-
sition with the port open. This equilibrium position marked the point at which p was zero. Then,
pressure inside the tubing between the ferrofluid and the manometer was increased by injecting air
into the port. The position of the ferrofluid plug end with respect to the center of the airgap was
recorded with increments of p until the magnetic force could no longer hold the ferrofluid in the
air-gap of the core.
The position of the ferrofluid plug end with respect to the center of the airgap as a function of
time was obtained from a video tape of the ferrofluid in motion. To record the ferrofluid motion on
VHS, the set-up shown in Figure 1 was placed under a dissecting microscope with a millimeter rul-
er included for proper scaling. A CCD camera was attached to the microscope and current was
6
applied to the windings. The motion of the ferrofluid was captured on VHS tape. This recording
was digitized at 10 frames per second. Each frame was examined and the position of the ferrofluid
was measured to within 0.1 0.05 mm.
A similar set-up was used to measure the resolution of position control of the pipette device.
Two cores from the pipette device were held in place under the dissecting scope such that the spac-
ing between the two core gaps was 0.5 mm. The field polarities were opposite with respect to one
another. The position change of the ferrofluid was recorded with the camera mounted on the dis-
secting scope.
Prototype Design
The ferrofluid-magnetic pipette consists of 4 iron-toroid cores of rectangular cross-section each
wound with 160 turns of 30 AWG wire. The Delrin casing holds the iron-toroid cores and the
Tefzel tube contains the ferrofluid. The pipette assembly and its parts are shown in Figure 2. The
(insert Fig. 2)
Tefzel tube containing the ferrofluid is not shown. The toroid cores are arranged in an alternating
pattern to accommodate bulk of the windings.
The toroid cores were machined of cast iron which have a high permeability and low coercive
field [10]. The dimensions of the iron-cores are: inner diameter of 3.5 mm, outer diameter of 6.5
mm, width of 2.5 mm, and air-gap length of 1.8 mm. Each toroid core and winding generates a
gradient magnetic field in the 1.8 mm air-gap.
Tefzel tubing was chosen for its minimal wetting characteristic with the ferrofluid. The tube
inner diameter is 0.4 mm and the outer diameter is 1.6 mm. EMG 901 ferrofluid was chosen be-
cause of its high magnetic saturation.
The Delrin casing consists of 4 pieces (two end pieces and two center pieces) which hold the
toroid cores in the configuration shown in Figure 2. Delrin was chosen because of its nonmag-
netic property and because it is easy to machine. One end piece of the casing is larger to provide
a means of attaching a holder for one end of the Tefzel tubing. Both end pieces of the casing have
a 1.7 mm diameter hole to accommodate the Tefzel tubing and a hole for inserting a bolt to attach
7
the pipette to a robotic arm or sampling system. The center pieces are identical such that additional
electromagnets can be easily added to the pipette.
The parts of the pipette are easily assembled and held together with two 2-56 by 1 inch bolts.
Paper or plastic spacers of ~0.5 mm thick must be placed between the iron cores to prevent the
cores from shorting out the magnetic field of its neighboring core. The plastic casing hold the cores
such that the air-gaps are aligned and the tubing is securely held inside the air-gap of each toroid
core. To prepare the pipette for liquid sampling, a volume of ferrofluid is placed in the Tefzel
tubing by capillary action and the tubing is threaded through the hole provided on the end pieces
of the Delrin casing. The entire assembled pipette weighs approximately 25 grams.
The ends of the windings on the toroid-cores are connected to a current source which supplies
0.6 0.03 amps. This current level will produce an mmF of 96 5 A-T in the 160 turns of wire.
The current source can be driven by digital logic such that a logic 1 will provide 0.6 amps and a
logic 0 will provide negligible current. The actual current output of this circuit was 0.58 amps for
a logic 1 and 0.02 amps for a logic 0. At 0.6 amps, winding and iron-core temperatures rise to
about 130 and 140F respectively in 15 minutes. Extra cooling should be provided for currents over
0.7 amps.
Magnetic Pressure
Magnetic pressure on the ferrofluid in the pipette device was measured as described previously.
The distance between the high pressure end of the ferrofluid and the center of the air-gap was plot-
ted against the corresponding manometer reading, p . Three separate pressure measurements (insert Fig. 3)
were taken for each position. The average of measured pressures for each position are plotted in
Figure 3. The error bars indicate the highest and lowest pressures measured at each position and
the center of the air-gap is the 0-mm mark on the position axis.
In Figure 3, the portion of the curve with the highest slope is the region from the edge of the
core to 1.5 mm. This is the region where there is maximum mechanical stiffness of ferrofluid plug
movement. Based on this observation, the toroid cores were placed to within 0.5 mm of each other
to move the ferrofluid in the pipette device. The maximum magnetic pressure at 96 A-T is about
770 Pascals which is near the center of the air-gap; pressures dropped off in the regions of magnetic
fringing. The equilibrium position of the ferrofluid, at p = 0, was 4.5 mm from the center of the
gap.
8
Position Control
The pipette prototype was designed to handle submicroliter, or nanoliter, sizes of liquids. To
handle liquid volumes in this range, ferrofluid position change must be a few tenths of a millimeter.
There were two methods of operating the ferrofluid-magnetic pipette to achieve this resolution.
The first method involved sequentially energizing the electromagnets at about 2 Hertz, called
the sequential-pulsing scheme. Each electromagnet was energized in a specific sequence such
that a travelling wave of magnetic field was created down the length of the tube at a speed which
can be followed by the ferrofluid. The volume sampled by the pipette is expressed as
2
Vol = [ NW + ( N 1 ) S + 2F d F L ] --- ( ID ) (3)
4
for F L > 2F d + W ; where N is the number of electromagnets in the pipette, W is the length of the
air-gap along the axis ferrofluid motion in the air-gap (z-axis), S is the spacing between each elec-
tromagnet, F d is the length of the fringing field along the z axis on the ends of the electromagnet
array, F L is the length of the ferrofluid, and ID is the tubing inner diameter.
The length of the ferrofluid, F L , cannot exceed the entire length of the electromagnet array and
fringing field on the ends ( NW + ( N 1 ) S + 2F d otherwise, there would be no magnetic pressure
on the ferrofluid when it is centered on the electromagnet array. In addition, F L cannot be smaller
than W because the ferrofluid must be pulled through one air-gap such that the field generated by
a neighboring electromagnet can act on the ferrofluid with sufficient pressure. For W < F L <
( 2F d + W ), the ferrofluid will center itself according to the symmetry of the gradient-field in the
air-gap.
The sequential-pulsing scheme works provided the remnant flux in the core of the energized
electromagnet is eliminated prior to removing the current and energizing the next electromagnet.
The accuracy of the desired volume handled depended on the accuracy of predicting F L , ID , F d ,
W , and S . There was also some concern with the compliance of position control particularly when
the ferrofluid reached the ends of the electromagnet array.
The second method of controlling the ferrofluid position involved changing the gradient mag-
netic field along the axis of ferrofluid motion (z axis). Two adjacent electromagnets were arranged
such that the fields generated were opposite in polarity with respect to one another. When both
were switched on, the gradient field profile was altered due to the subtraction of the x-component
10
of magnetic field particularly in the region between the two adjacent air-gaps. This subtraction al-
tered the gradient field and thus altered the resulting magnetic pressure on the ferrofluid plug. (insert Fig. 6)
To illustrate the gradient magnetic field alteration, the magnitude of the x-component of the
magnetic field of two electromagnets was computed using the circular-arc straight-line model [10,
13] and plotted as a function of position in Figure 6. The resulting computation predicts that the
ferrofluid will stop in the region between the two air-gaps when p of equation (2) is zero, assum-
ing both ends of the ferrofluid plug are open to atmospheric pressure. Once the ferrofluid comes to
rest at p = 0, a decrease in the opposing magnetic field of core 2 will cause the ferrofluid to adjust
its position due to the change in gradient field distribution. Figure 7 illustrates the general idea of (insert Fig. 7)
this position control scheme.
To obtain an experimental dynamic analysis of the two-electromagnet arrangement, position
as a function of time was measured from a digitized VHS recording of the ferrofluid in the two-
electromagnet arrangement. This data is shown in Figure 8. As Figure 6 predicts and as shown in
(insert Fig. 8)
Figure 8, the ferrofluid stopped in the region between the two cores with both electromagnets on
and readjusted its position when core 2 was switched off.
The position adjustment of the ferrofluid was measured experimentally as described previous-
ly. The measurements were taken for 5 different current levels in core 2: 0.095 (15 A-T), 0.18 (29
A-T), 0.34 (54 A-T), 0.44 (70 A-T), and 0.58 amps (93 A-T). The stopping position of the ferro-
fluid (corresponding to C of Figure 7) was noted; the ferrofluid stopped to within 1.7 to 1.9 mm
from the center of the air-gap depending on the level of current applied to core 2. The position
change of the ferrofluid was measured to within 0.05 mm after removing the current in the wind-
ings of core 2. These results are presented in Table 2 with the corresponding current change. The
position change was converted to a corresponding volume change and also summarized in Table 2.
The experimental data in Table 2 shows the effect of the field change in core 2 on the position
of the ferrofluid. The remnant magnetic field in core 2 was not eliminated and therefore, the mea-
sured ferrofluid position change ranged from 0.1 to 0.5 mm. The advantage of this control scheme
over the sequential-pulsing method, was the improved stiffness of position control with the stiffest
mechanical position control obtained at approximately 0.2 to 0.3 amp current change in the wind-
ings of core 2.
The ferrofluid pipette was tested with the sequential-pulsing scheme using water as a sample.
This device was able to dispense 90 nl of liquid. Because of magnetic hysteresis in the electromag-
11
net cores, it was difficult to control the ferrofluid position to dispense smaller quantities of water.
To remove the remnant magnetic field, a backwards pulse of current should be applied to the elec-
tromagnet immediately after the energizing current is removed.
5. Conclusion
This paper presented the design of the ferrofluid-magnetic pipette. The pipette consists of a fer-
ro-fluid contained in Tefzel tubing and four electromagnets which generate the magnetic field to
control the ferrofluid position. The ferrofluid acts as the plunger as in a mechanical pipette and the
position change of the ferrofluid corresponds to a volume sampled by the pipette. A position
change of 0.1 to 0.5 mm was achieved by modulating the gradient-field in one electromagnet. The
simple design and lack of moving parts is an advantage for manufacturing and maintaining the
device.
The next step with the pipette design would be to design software implementing the control
scheme presented in the paper. In addition, this software could be made to implement a feed-back
type control system which would increase the accuracy of the position control. We envision that
the final ferrofluid-magnetic pipette will be used in large-scale analysis such as the Human Ge-
nome Project.
Acknowledgments
This work was part of a M.S.E.E. thesis by Nancy Greivell submitted to the Center for
Bioengineering at University of Washington. The research was supported by a grant from the
Washington Technology Center.
12
References
[1] Rosensweig, R.E., Ferrohydrodynamics, (New York, 1985).
[3] Barral, J., R. Bonnefille, M. Duret, et M. Kant, Revue De Physique Appliquee, 12 (1977)
1711.
[4] Raj, K., R. Moskowitz, Journal of Magnetism and Magnetic Materials, 85 (1990) 233.
[6] Neuringer, Joseph L. and Ronald E. Rosensweig, Phys. Fluids, 7 (1964) 1927.
[7] White, Frank M., Viscous Fluid Flow, Second Edition, (New York, 1991).
[8] Benedict, Robert P., Fundamentals of Pipe Flow, (New York, 1980).
[9] Miller, C.W., E.L. Resler, Jr., Phys. Fluids, 18 (1975) 1112.
[11] McTague, John P., The Journal of Chemical Physics, 51 (1969) 133.
[12] Mozgovoi, E. N., . Ya. Blum, and A. O. Tsebers, Magnetohydrodynamics, 9 (1973) 52.
[13] Hanselman, Duane C., Brushless Permanent-Magnet Motor Design, (New York,1994).
13
Figure Legends:
Figure 1. Setup for taking magnetic pressure measurements as a function of ferrofluid plug end
position with respect to the center of the air-gap. One end of the tube was connected to a water-
filled manometer. Air was injected in the port and position of the end of the ferrofluid was mea-
sured with respect to the center of the air-gap (indicated by 0 on the axis).
Figure 2. The assembly of the ferrofluid pipette. The 160 turns of 30 AWG wire is not shown for
clarity. The entire assembly is approximately 4 cm long by 1 cm wide by 3 cm high.
Figure 3. Magnetic pressure versus position of the ferrofluid plug end with respect to the center
of the air-gap. The center of the air-gap is indicated by 0 on the x-axis. Three pressure measure-
ments were taken at each position point. Error bars indicate the highest and lowest pressure mea-
sured. Average of the three measurements are plotted as dots. Total length of the ferrofluid plug in
the tubing was 22 mm.
Figure 4. A: Experimental position of the ferrofluid plug end with respect to the center of the air-
gap versus time. The center of the air-gap is indicated by 0 on the y-axis. B: Average velocity of
the ferrofluid versus time computed from A. Total length of the ferrofluid plug in the tubing was
approximately 20 mm.
Figure 5. Velocity versus position of the ferrofluid plug end with respect to the center of the gap.
Dashed line: velocities computed from static pressures and corrected for magnetic induced viscos-
ity. Solid line: velocities computed from position versus time data.
Figure 6. Estimated magnitude of the x-component of applied magnetic field in the airgap of two
adjacent electromagnets. Computations use the circular-arc straight-line approximation [10, 13].
The field in core gap 1 was energized at 96 A-T. The field in core 2 was varied from 8 A-T to
14
128 A-T. Each plot shows the computed field profile in both air-gaps as a result of the addition of
the two fields - core 1 is always energized with 96 A-T and core 2 is energized at different levels
for each plot.
Figure 7. The idea of the gradient-field modulation scheme for ferrofluid position control. In A,
core 1 and core 2 are energized simultaneously. The ferrofluid moves through the air-gap of core
1 in B and stops at a point in the space between the two air-gaps in C. At this point, the end of the
ferrofluid-magnetic pipette is placed in the biological liquid. Sampling occurs when the field in
core 2 is switched off and the ferrofluid readjusts its position in D. This position adjustment cor-
responds to a specific volume of liquid sampled.
Figure 8. Measured dynamics of the gradient-field modulation scheme using two electromagnets
The plot shows the end of the ferrofluid plug with respect to the center of the airgap versus time.
Dashed line: ferrofluid motion in the tubing with one electromagnet at 96 A-T. Solid line with
circles: ferrofluid motion in the tubing with two cores: core 1 at 96 A-T, and core 2 at 48 A-T - at
the point indicated, the ferrofluid moved to a new rest position when the current was switched off
in windings on core 2.
15
*These properties were provided by Dr. K. Raj of Ferrofluidics Corporation, Nashua, NH.
Figure 1
Tefzel tube
core to manometer
Patm Patm + p
0
port
EMG 901
ferrofluid
z
17
Figure 2
Figure 3
800
edge of core
700
600
pressure in pascals
p
500
400
300
200
100
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
position with respect to airgap center in mm
19
Figure 4
A
3
3
0 1 2 3 4 5 6
time in seconds
1
B 0.9
0.8
velocity of ferrofluid in cm/sec
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 1 2 3 4 5 6
time in seconds
20
Figure 5
toroid core
1.2
1
velocity in cm/sec
0.8
0.6
0.4
0.2
z
0
5 4 3 2 1 0 1 2 3 4 5
position in mm with respect to center of airgap
21
Figure 6
core 1 core 2
800
96 X
core 1
600
Z
400
magnetic field in Oersteds
200
8
16
0
32
48
200 ferrofluid
72
400
96
600
128 core 2
800
6 4 2 0 2 4 6 8 10 12
distance from center of core in mm
22
Figure 7
core 1 core 2
ferrofluid
to liquid sample
D
23
Figure 8
3
position with respect to center of air gap in mm
0
current in core 2
switched off
3
0 2 4 6 8 10 12
time in seconds