Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Home Search Collections Journals About Contact us My IOPscience

Hall effect in a moving liquid

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2012 Eur. J. Phys. 33 115

(http://iopscience.iop.org/0143-0807/33/1/010)

View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 152.3.102.242
The article was downloaded on 07/04/2013 at 08:55

Please note that terms and conditions apply.


IOP PUBLISHING EUROPEAN JOURNAL OF PHYSICS
Eur. J. Phys. 33 (2012) 115–127 doi:10.1088/0143-0807/33/1/010

Hall effect in a moving liquid


Alberto Di Lieto, Alessia Giuliano,
Francesco Maccarrone and Giampiero Paffuti
Dipartimento di Fisica, Università di Pisa, Largo B Pontecorvo 3, 56127 Pisa, Italy

E-mail: dilieto@df.unipi.it

Received 23 September 2011, in final form 19 October 2011


Published 28 November 2011
Online at stacks.iop.org/EJP/33/115

Abstract
A simple experiment, suitable for performing in an undergraduate physics
laboratory, illustrates electromagnetic induction through the water entering into
a cylindrical rubber tube by detecting the voltage developed across the tube in
the direction transverse both to the flow velocity and to the magnetic field. The
apparatus is a very simple example of an electromagnetic flowmeter, a device
which is commonly used both in industrial and physiological techniques. The
phenomenology observed is similar to that of the Hall effect in the absence of
an electric current in the direction of motion of the carriers. The experimental
results show a dependence on the intensity of the magnetic field and on the
carrier velocity, in good agreement with the theory. Discussion of the system,
based on classical electromagnetism, indicates that the effect depends only on
the flow rate, and is independent both of the velocity profile and of the electrical
conductivity of the medium.
(Some figures in this article are in colour only in the electronic version)

1. Introduction

In 1879, the American Journal of Mathematics published a letter [1] of a student attending the
lectures of Professor Henry A Rowland at John Hopkins University, announcing the discovery
of a new electromagnetic effect. The student was Edwin Herbert Hall and the phenomenon was
observed in a thin metallic conductor carrying electric current, subjected to a static magnetic
field perpendicular to the direction of the current.
In those conditions a current was measured through the galvanometer connected to points
‘opposite each other on the edges of the strip’. The effect was eventually interpreted as
the onset of an electromotive force E perpendicular to the magnetic field B and in the last
paragraph of Hall’s paper the proportionality between E and the product of B and the ‘velocity
of the electricity’ v was suggested:
E ∝ vB. (1)
c 2012 IOP Publishing Ltd Printed in the UK & the USA
0143-0807/12/010115+13$33.00  115
116 A Di Lieto et al

It is quite surprising, as usually we believe that only the contemporary age is characterized
by the rapidity of knowledge circulation, that a few months after the publication of that paper
by the young student, a commentary (Bemerkung) was written by Boltzmann [2] on the
new effect, pointing out the importance of the phenomenon with respect to the possibility of
determining the velocity of particles of electricity (soon after scientists began to call them
electrons) carrying the electric current.
The interest in the new phenomenon, which almost immediately took the name of the
Hall effect grew in the following years and led E H Hall in 1905 to publish an academic
communication [3] with a list of the scientific papers inspired by his discovery. Among
the authors besides Boltzmann one finds the names of H A Lorentz, H Poincaré, W Nernst,
P Drude, G Donnan and many others.
The Hall effect soon began to be used for its extraordinary suitability for investigating
several aspects of electrical conduction in materials. In fact, the phenomenon is sensitive to
the sign of the current-carrying particles and gave a clear proof that in metals the current is
carried by negative charges.
Moreover, in a simple geometry, the value of E gives information on the density of the
carriers: for a material in the form of a plate with cross section A = l × t, through which
passes a given current I, carried by particles of charge q uniformly filling the volume of density
n, these simple relations are valid:
I = nqvA
I I B
n= = .
qvA qt E
Very recently, advanced techniques have been developed using the Hall effect to map
microscopic currents in 2D electronic conduction observed in heterostructures [4].
Particular attention in the first two decades after Hall’s discovery was devoted to the study
of the effect in liquid electrolytic solutions where it was difficult to detect the phenomenon,
because of the low mobility of large ions in a liquid matrix, and the consequent low drift
velocity v to which the transverse electromotive force is proportional. Moreover [5], when
the conduction involves oppositely charged carriers moving in opposite directions under the
action of the external electric potential, the transverse magnetic deflection is the same for both
species so that the transverse Hall field results from the net balance of the two contributions
and is lower than what may be observed in unipolar conduction. Besides the already cited
electrolytic solution, semiconductors and weakly ionized gases are typical systems through
which bipolar conduction is possible and such a form of the Hall effect is detectable.
In semiconductors both signs of the effect are observed, depending on the prevalence of
different mobilities (velocities) of electrons and holes, and on their different concentration in
extrinsic doped semiconductors. With the advent of semiconductor technology in the middle
of the 20th century, the Hall effect found large application in the production of current and
magnetic field sensors.
A different phenomenology is present when a moving fluid, liquid or gaseous, drags the
charged species within itself at the same velocity of the flow. To guarantee the neutrality of the
fluid, the density of opposite charges is the same and the electric current is zero. The magnetic
field tends to separate positive and negative particles in opposite directions. A condition of
equilibrium is restored when the transverse charge distribution creates an electric field which
neutralizes the magnetic force. In that situation an electromotive force is measurable in the
transverse direction and its magnitude is again proportional to the velocity of the fluid and
to the magnetic field as in (1). The history of this phenomenon goes back as far as modern
electromagnetism, since even Faraday, albeit unsuccessfully, tried to detect an electromotive
Hall effect in a moving liquid 117

force plunging metallic wires into the stream of the river Thames flowing in the Earth’s
magnetic field [6].
A detailed study of this effect was made by another PhD student, working in Manchester
University, which investigated the flux of a copper sulphate solution in a strong magnetic field.
The results of his experiments were discussed in a session of the London Physical Society in
1930 [7]1 .
In our opinion, it is a question of physical taste to distinguish neatly between the two
situations. Some would prefer to reserve the name Hall effect to the voltage, magnetically
induced crosswise to a non-zero current and call the phenomenon in moving fluids the Faraday
effect [8]. Others point out that the main difference is in the external forces (electric or pressure
force) moving the charged particles. The same question could also be raised, for example,
with reference to a rather uncommon thermomagnetic effect, the Nernst–von Ettingshausen
effect [9]2 which consists in the transverse voltage observed when a direct motion of charges
is induced by a thermal gradient (generally known as the Seebeck effect) in the presence of an
orthogonal magnetic field.
Keeping this in mind we intend the title of this paper, in which a simple experimental
demonstration of what we like to call the Hall effect in flowing water is given, with materials
available in a good undergraduate physics laboratory. The advantage with respect to the
obvious demonstration of the Hall effect, for example with semiconductor sensors, is that also
the velocity, one of the key parameters, is on the macroscopic scale.
The objection that the carrier density plays no role is not completely convincing, as this
parameter does not enter directly as part of the physical mechanism but rather indirectly, being
a parameter of the current itself.
The apparatus is simpler but similar to that used in an interesting analogous demonstration
published some years ago [10]. We confirm the results of that investigation pointing out some
details which give information on the parameters of the water flow, but in contrast with
the previous results we find no evidence that an electrolyte is necessary to make the effect
observable. The normal dissociation constant of water appears sufficient for providing a
density of charges suitable for detecting the transverse voltage.
As was pointed out in [11], the observed effect is exploited in commercially available
electromagnetic flowmeters. In fact, the phenomenon was proposed as a method for measuring
instantaneous blood flow in human arteries [12].

2. Experimental setup and results

A straight cylindrical rubber tube with an inner diameter of 0.8 cm is placed horizontally inside
a couple of permanent neodymium magnets in the form of disks. The distance between the
facing surfaces of the disks can be varied, starting from a minimum distance of 1.7 cm (limited
by the external diameter of the tube and by the magnet mount), by a pair of counter-rotating
screws and two leading shafts insuring the alignment of the disks. The magnetic field at
the centre of the tube reaches a maximum of about 0.4 T, and the uniformity of the field is
better than 1% over the entire volume of the tube, as from a calibration procedure3 and from
calculation based on a uniform magnetization of the disks.
1 The author, primarily known for his pioneering work on experimental meson physics at the Niels Bohr Institute
in Copenhagen in the early 1930s, acknowledges W S Bragg, being his professor at Manchester University and the
Fellow presenting the paper to the Society, and D R Hartree for his interest in the subject.
2 It is worthwhile to note that the paper of Nernst and von Ettingshausen is listed in the McKay paper [3] of 1906 as
being related to the Hall phenomenon.
3 The calibration of a magnetic field has been carried out by using the Hall sensor model A1302, placed at different
points corresponding to the region normally occupied by the tube.
118 A Di Lieto et al

Figure 1. Schematic view of the experiment. The reference axes used in the text are also indicated;
the x axis is assumed orthogonal to the figure.

Two facing electrodes are placed inside the tube, drilling opposite holes at right angles
to the axis of the tube and placing two cap screws with their heads in contact with the inner
surface of the tube; the resulting distance between their surfaces coincides with the inner
diameter (0.8 cm) of the tube. The electrodes are connected to the inputs of an A/D converter
of an acquisition board, and the voltage developed across them is measured with a suitable
LabVIEW program, that averages each 0.1 s the signal from the A/D converted at a sample
rate of 100 kS s−1.
The tube was oriented to maintain the axis of the electrodes, the flow direction and the
magnetic field mutually orthogonal, as in the sketch of the apparatus shown in figure 1.
The tube is connected to a hydraulic circuit including an electromechanical aquarium
pump giving a flow of up to ∼75 g s−1, a vessel for the recirculation of the liquid (tap water)
and a mechanical valve placed at the end of the tube pouring into the vessel. When the liquid
is in a stationary motion the valve’s closure causes a reduction of the flow rate which, in turn,
regulates the velocity in the constant diameter tube, in accordance with the conservation of
flux. Of course, the velocity profile in a given tube depends on the diameter of the tube. As
we show later, when the magnetic field is uniform the signal is to an excellent approximation
proportional to the overall flux so that it is not necessary to know the exact profile of the
stream.
The flow rate was measured, for a given valve position, weighting the mass M of liquid
coming out from the circuit in a fixed time t (10 s typical). The conservation of flux and
the stationariness of the stream guarantee that the volume flux Q is the same for all the cross
sections A of the circuit. If, for the sake of simplicity, we assume that the velocity is constant
over the transverse section of the flow, we are able to estimate the magnitude of the expected
effect. In fact, the procedure gives a measure of the flux in the part of the circuit exposed to
the magnetic field:
Q M
v= = ,
A Aρt
where ρ is the fluid density. In our apparatus A ≃ 0.50 cm2 and for the maximum flow rate
the average velocity reaches a value of the order of 1.5 m s−1, a value slightly larger than the
ones examined in [10]. For the greatest values of the magnetic field and the flow velocity,
Hall effect in a moving liquid 119

Figure 2. Detected signal across the tube with and without magnetic field. The distance of the
magnetic disks is fixed at 1.7 cm, corresponding to an intensity of the magnetic field of 0.38 T and
the flow rate is  = 60 g s−1. The pipe is moved in and out of the magnetic field in a time less
than 0.1 s, pushing and pulling manually the magnetic disk mount. The inversion of the sign of
the electromotive force is observed by reversing, by hand, the position of the magnet mounting.

the expected value of the potential difference is of the order of a few millivolts. To make a
comparison also with the experimental parameters of the paper [7], the main difference is the
intensity of the magnetic field which was in the previous work up to 0.95 T. The diameter
of the glass tube used by Williams was 1.075 cm and the velocity of the flux was at most
0.66 m s−1. Besides that, particular care in the manufacturing of electrodes was taken in order
not to disturb the flux and to reduce spurious effects such as contact potentials. Moreover,
the distance between the electrodes was varied by a screw movement, thus allowing a third
experimental parameter and to examine, at least in principle, the inhomogeneity of the velocity
across the tube.
In order to subtract the contribution to the voltage from spurious effects such as contact
potential, Williams registered the difference of the detected voltage when the direction of the
electromagnet current was inverted in a time not declared, but certainly influenced by the
inductance of the magnetic circuit. In our apparatus, both the velocity and the magnetic field
can be set to zero or to the maximum value for a given gap between the disks, in a time that is
short with respect to the integration time of the signal, either switching on and off the power
supply of the pump or moving away the magnet mounting. In both cases, a sudden variation
of the detected signal is registered. Figure 2 shows the time registration of the signal moving
by hands back and forth the magnet mounting in a time of the order of one-tenth of a second
when the water is in motion (pump is on). The residual electric potential detected at zero field,
due to static charges, varies on a very long time scale and that baseline is easily subtracted.
The amplitude of the variation of the measured voltage is constant and in the following we
take that value as the measure VH of the effect.
The same behaviour is observed—at a given magnetic field—switching on and off the
pump and varying the flow velocity from zero to the stationary value and vice versa in a time
120 A Di Lieto et al

Figure 3. Detected signal across the tube with and without water flow. The distance of the
magnetic disks is fixed at 1.7 cm, corresponding to an intensity of the magnetic field in the central
zone of 0.38 T. The flux is started and stopped by switching the pump of the hydraulic circuit on
and off.

of the order of 1 s. Figure 3 shows the variation of the signal VH under this experimental
procedure. These two observations assure us that the effect manifests itself only when both
the magnetic field and the fluid motion are present.
Following (1), a linear dependence of the signal on the velocity is expected. We set the
distance of the magnets, we varied the flow rate  acting on the valve of the hydraulic circuit
and we measured the water flow. An experimental uncertainty of 5% on the value of v is a
cautious estimate of our accuracy.
Figure 4 shows the experimental points (, VH ) at five distances between the magnets.
Each set of data has been fitted with a linear model, because (1) can be rewritten, according to
section 3, equation (9), as
VH = k(B) . (2)
The continuous curve represents the best fit; the relative precision on k(B) is typically
∼ 4 × 10−3 .
Figure 5 shows the experimental points (B, VH ) at five values of the flow rate. Even for
these data the linear model
VH = h() B (3)
has been used in the fit; because of the poorer precision of the flux rates, the relative precision
of h() is typically ∼1.5 × 10−2 .
A supplementary check can be carried with the parameters k(B) and h(), because they
must be proportional respectively to B and : the fit with the linear model for the data sets
(B, k(B)) and (, h()) is statistically acceptable, the relative precision is of the order of
a few percent and the two coefficients are coincident within the errors. An alternative (and
more correct) method to analyse the data treats VH as a bilinear function of the two variables B
Hall effect in a moving liquid 121

Figure 4. Detected signal registered at five distances of the magnetic disks as a function of the
flux rate. The straight lines are the best fits of (2) to each set of data. The ratio of the slopes is in
very good agreement with the ratio of the magnetic field intensities.

Figure 5. Detected signal registered at five values of the flux rate as a function of the magnetic
field. The straight lines are the best fits of (3) to each set of data. The ratio of the slopes is in very
good agreement with the ratio of the flux rates.
122 A Di Lieto et al

and : the fit is statistically well grounded, and by using a data set of 60 measurements
the parameter is estimated to have the numerical value 0.1597, with a relative precision of
better than 5 ×10−3 , in very good agreement with the expected value, as calculated from
equation (9).
In [7] and [10], the authors pointed out that the water used in the experiment must contain
added free charges and so a salt solution was used in both cases. In particular, [10] states
that ‘pure water produces no measurable electric field’. We checked this statement by using
as liquid not only tap water but also distilled and deionized water, and we found that in all
cases the effect is present [13]. In fact, it is difficult to consider pure water a non-conductive
fluid as it contains dissociated ions both as the result of the H + ↔ OH − equilibrium and for
the presence of dissociated gases such as carbon dioxide. The Hall effect at zero current is
not dependent on the concentration of the charged particles and the amplitude of the detected
signal, which depends on the input impedance of the voltmeter, is influenced only at the lowest
value of the charge density, as we discuss in the following section.

3. Discussion

In order to discuss some details of the experimental results, let us make reference to the
derivation of the electric potential given by Kolin in 1945 [14]4 .
At first, we assume that the magnetic field is uniform, and directed vertically along the z
axis. The motion of the fluid is described by a stationary velocity field v directed at each point
along the axis of the pipe (x axis), with a magnitude v(r), depending only on the distance r
from the axis of the tube, with 0  r  R, where R is the radius of the pipe.
The electrical conduction properties of the fluid are described by the conductivity σ ,
which depends on the density of free carriers and on their mobility. Due to the homogeneity
and isotropy of the medium, we consider σ to be constant over its volume. It could be changed
using an aqueous solution and varying the concentration of the ions of an electrolyte. In any
case, the conductivity is small with respect to metals and other strong conductors so that the
magnetic field of the electric currents induced in the medium can be neglected.
Inside the fluid, a current density J is induced by the forces originating by the electromotive
effects of both an electric field E and an external magnetic field B, according to the relation
which generalizes Ohm’s law:
J = σ (E + v ∧ B). (4)
Here and in the following, the small magnetic field produced by the induced currents is
neglected. In the medium, in the absence of sources or current sinks, div J = 0 and from (4)
div E = −div (v ∧ B). (5)
We see that, in the hypothesis of homogeneity, σ does not enter in the following, as we pointed
out at the end of the previous section, and the Hall effect at zero current is not dependent on
the density of charged particles. Moreover, since the magnetic field is static (∂B/∂t = 0), the
electric field is irrotational and is derived from a potential v:
E = −grad V ,
which, substituted in (5), gives
∇ 2 V = div (v ∧ B). (6)
4 It is worthwhile noting that A Kolin—eminent biophysicist at UCLA—completed his course in Physics at the
Technische Hochschule in Berlin, studying plasmas under the guidance of Gustav Hertz. In the cited paper, he
acknowledged his indebtedness to W E Lamb Jr giving another example of the interest that the phenomenon here
described constantly attracted from outstanding scientists.
Hall effect in a moving liquid 123

Following the calculation shown in the appendix, the potential difference between two opposite
points on the inner surface of the tube along the y axis is
4B R

V = rv(r) dr. (7)
R 0
By definition, the mass flux across the tube cross section is
 R
=ρ 2π r v(r) dr. (8)
0
The mass flux gives the value of the integral in (7), without any reference to the velocity profile
v(r) of the flow:
2B
V =  (9)
ρπ R
and, at fixed flux, the detected voltage V does not depend on the distribution of the velocity in
the pipe. Expressing the mass flux in terms of the average velocity v through
 = ρπ R 2 v̄.
one finds the well-known Hall’s formula:
V = Bd v. ¯
For the laminar flow of a viscous fluid into a cylindrical tube, the velocity profile is
v(r) = v0 (1 − r 2 /R 2 ) and the average velocity is v̄ = v0 /2, half of the maximum velocity v 0.
Of course, it follows from (A.7) that the radial derivative of the potential is larger where
v(r) is larger, so that for a velocity distribution v(r) having a maximum at the centre of the
pipe, as in the case of a laminar flow of a viscous fluid in a cylindrical tube, the difference of
potential grows faster in the middle of the tube, and is stationary near the tube wall, where
the velocity of the water approaches zero. The result is different if the velocity has a constant
value on average through the section of the pipe. This should be the case of turbulent flow,
if at each point the non-stationary velocity takes values with uniform average. In this case,
the gradient of the potential is constant and the difference of voltage grows linearly. Williams
demonstrated experimentally the described behaviour in the two regimes by moving one of
the electric probes across the diameter of the tube, as is shown in figures 3 and 4 of [7].
It is worth commenting on the independence of V on the conductivity of the fluid, as
in equation (9). In fact, the basic intrinsic parameter in the effect is the charge velocity
v. In a textbook explanation, the conductivity enters when the Hall voltage is expressed as
a function of the electrical current along the conductor, which is σ -dependent. In a fluid,
we have access directly to the basic observable v, the speed of the medium which drags
along the molecular charges. No net electrical current is present in the direction of the flow,
so that the σ -independence is manifest. Moreover, it is easy to see that the usual description
of the Hall effect in a metallic conductor (i.e. the formation of opposite charge distributions
at the sides of the current flow) corresponds to the perfect fluid case, where v is constant and
the only contribution to the charge density, proportional to div E, is due to the discontinuity
at the walls. These considerations clearly refer to the ‘open circuit’ potential difference. In a
real experiment the electrodes used to measure V form a closed circuit with the instrument:
in this case a current flows and the measured value for V in principle is affected by the value
of σ , as discussed below.
Due to the finite section of the magnetic disk, a more realistic description should take into
account the dishomogeneity of the magnetic field along y and the existence of a y component
of B. In this case, the right-hand side of (5), again with a velocity v directed along x, is
 
∂v ∂v ∂By ∂Bz
div (v ∧ B) = −Bz + By +v − . (10)
∂y ∂z ∂z ∂y
124 A Di Lieto et al

The last term is the x component of the rotor of the current density and it is zero, considering
the sources of the magnetic field outside the volume of the fluid, and always neglecting the
effect of induced currents, which—in the case of water—give corrections many orders of
magnitude lower. Moreover, the contribution of the first two terms is non-zero only if the
velocity is not uniform. The calculation of the contribution to the transverse voltage of the
non-uniformity of the magnetic field is beyond the scope of the paper. We limit ourselves to
state that an exact calculation for the geometry of our experiments gives discrepancies less
than one part in a thousand.
Considering the turbulent flow to be the result of a uniform drift along the axis of the tube
and a non-stationary random velocity which averages to zero, this regime should give the same
results as in the case of a uniform laminar flow. In particular, the first and the second terms in
(10) vanish, and the solution of the Laplace equation gives a potential linearly dependent on the
transverse coordinate, in accordance with the experimental results cited above. If we consider
the electron flux in a conductor as a stream of charge with a uniform cross distribution it is
clear from the discussion above that, under the present assumptions, even an inhomogeneous
magnetic field is not able to induce a space charge density distribution into the conductor. This
is in accordance with the usual presentation of the Hall voltage resulting from two opposite
charge sheets accumulated at the border of the metallic strip. The generated electric field
compensates the Lorentz force on the charge carriers. The analysis performed on the moving
fluid demonstrates that the description valid for metals applies only as a limiting case, where
the velocity is uniform over the cross section. For a laminar flow with velocity profile v(r)
we have a separation of opposite space charges into the volume of the fluid which generates
the electric field and the measured voltage difference. In fact, it is shown in the appendix that
the orthogonal electric field, and consequently the surface charge, is null on the surface of the
tube. For the sake of completeness, we show in figure 6 the graphical result of the solution of
(A.7), assuming that the radial component of the current density vanishes at the inner surface
of the tube. Since the velocity of the fluid in contact with the tube is zero, the radial component
of the electric field is also zero (see equation (4)).
It is worth noting that in figure 6 the vector field lines represent the current density and its
magnitude scales with the electrical conductivity of the fluid. On the other hand, the voltage
and, consequently, the electric field do not depend on the conductivity. With the exception
of the limit σ = 0 the transverse voltage depends, at a fixed geometry, only on the velocity
of the fluid and on the magnetic field. This is in agreement with the experimental results
of Kolin (see figure 4 in [14]) who registered the same value of the transverse value for a
change of four orders of magnitude in salt concentration. It is also in agreement with the
experiments presented in [10] where the concentration of the NaCl solution was varied by
an order of magnitude without any observable change in the signal. On the other hand, the
independence of the signal on the conductivity is in contrast with the statement made in [10]
that the effect is cancelled using pure water. Of course, pure water, i.e. water without any trace
of other elements, is far from a non-conductive medium. Dissociation equilibrium results in
a concentration of hydrogen ions and the hydroxyl group which is close to 10−7 M and the
specific resistivity σ −1 of deionized water is of the order of 105 m which is only three orders
of magnitude larger that that of tap water, and not infinite.
Now, let us consider the experimental consequences on the detected signal derived from
the use of a bad liquid conductor such as deionized water. The electrical measurement of
the voltage VH consists in the connection of an instrument such as a voltmeter to points
diametrically opposite in the transverse direction. Therefore the equivalent circuit is an ideal
electromotive force source, controlled by the flux rate and the magnetic field, which feeds the
series circuit of the fluid and the internal resistance of the instrument. When the latter is much
Hall effect in a moving liquid 125

Figure 6. Current density field lines drawn in the plane orthogonal to the axis of the pipe. The
dashed lines are equipotential lines of the electric field. The coordinates are expressed in units of
pipe radius.

larger than the resistance of the conducting liquid medium, the current across the instrument
can be neglected and a full voltage drop should be observed across the pipe. On the other hand,
if the resistance of the external portion of the circuit is comparable to that of the electrical path
into the volume of the fluid, the voltage is proportionally decreased. In fact, we have observed
a decrease of the order of 20% of the signal using deionized water with respect to ordinary tap
water. The input resistance of the voltmeter is very large ( 10 G). The resistance inside the
fluid depends on the cross section of the current coupled to the instrument and its cross section
S could be estimated to be as small as 1 mm2 , of the order of the thickness of the electrodes.
That resistance should be of the order of L/(σ S) ≃ 1G which is not negligible with respect
to the impedance of the voltmeter and this explains the order of magnitude of the reduction of
the measured voltage, when deionized water was employed.

4. Conclusions

The Hall effect is an extraordinarily rich topic at the point of contact between classical
and modern physics. Many technical applications of this effect such as the measurement
of magnetic fields or the realization of electromagnetic flowmeters and its exploitation as a
probe of microscopic structure of materials give strong support to the choice of presenting its
phenomenology at the undergraduate level. On the other hand, the experimental observation of
electromagnetic induction on macroscopic bodies in motion is very useful, in our opinion, for
giving a deeper insight into basic electromagnetism than that usually attained with formalized
problems. The apparatus presented in this paper is as simple as one expects for a physics
experiment to be proposed in a classroom and as inexpensive as one requires in order to realize
126 A Di Lieto et al

several specimens for the simultaneous conduction of the proofs by small groups of students.
The control of the relevant experimental parameters is very clear and good agreement with
the expected results is easily found. The interpretation of the results and the examination of
the classical literature also offer the possibility of considering the main parameters describing
the flow of a liquid in a pipe in different regimes. A valuable result which is not intuitive is
that the effect is also present for a liquid with very low conductivity and that the decrease of
the detected signal with such a medium is interpreted with the simple analysis of the voltage
measurement across a high impedance source.

Acknowledgments

The authors gratefully acknowledge the collaboration of Mr Stefano Orselli, Technician of the
Physics Department of Pisa University, in the realization of the apparatus, and Professor L
Lovitch for reading the manuscript.

Appendix. Derivation of equation (7)

Let us consider for simplicity that the magnetic field is uniform, directed along the z axis and
with intensity B and that the velocity of the fluid has magnitude v(r), depending only on the
distance r from the axis, and is directed parallel to the axis of the tube (x axis). We assume
that the fluid is not magnetic (relative magnetic permeability of water μr = 1 to one part
in 105 ). With these conditions equation (6) is
∂vx ∂v(r)
∇ 2 V = −Bz = −B cos θ, (A.1)
∂y ∂r
where θ is the azimuthal angle with the positive y axis. The physical boundary condition on
the inner surface of the tube (y = R) is that the velocity of the fluid in contact with the tube is
zero. This is the standard condition in the case of a laminar flow. Then, the current density at
the surface depends only on the electric field and the vanishing of the current divergence on
the tube implies that the orthogonal component of E should be zero:

∂V 
= 0.
∂r r=R
The other boundary condition consists in choosing the arbitrary value of V at the centre of
the tube to be V (0) = 0. The system is uniform under translations along the x axis and the
potential V is a function only of the polar coordinates in the yz plane. Using the Laplacian in
polar coordinates, (A.1) can be written as
∂ 2 V 1 ∂V 1 ∂ 2V ∂v(r)
2
+ + 2 2 = −B cos θ. (A.2)
∂r r ∂r r ∂θ ∂r
Now we seek a solution in the form V = F (r) cos θ . Substituting in (A.2) one finds the
ordinary differential equation:
1 1 d F
F̈ + Ḟ − 2 F = F̈ + = −B v̇ (A.3)
r r dr r
and the first integration is immediate:
F
Ḟ + = −Bv + C. (A.4)
r
Hall effect in a moving liquid 127

The integration constant C is determined by observing that at the boundary r = R, the velocity
v vanishes and the orthogonal electric field is proportional to the radial derivative of F so that
Ḟ (R) = 0. Consequently, C = F (R)/R and we rewrite the last equation:
1 d F (R)
(rF ) = −Bv + . (A.5)
r dr R
Integration of (A.5) gives
 r
F (R) r 2
rF = −B rv dr + + C2 . (A.6)
0 R 2
The condition F (0) = 0 requires C2 = 0 and the solution is
 r
1 F (R)
F (r) = − B rv dr + r (A.7)
r 0 2R
which, multiplied by cos θ , gives V (r, θ ).
By substituting r = R in (A.7) the value of F (R) is
B R

F (R) = −2 r v dr. (A.8)
R 0
We are interested in the difference of the potential VH on points at the inner surface of the
tube (r = R, θ = 0, π ), namely VH = V (R, π ) − V (R, 0). Substituting the computed value
we obtain
4B R 2B R
 
VH = rv dr = 2π r v dr. (A.9)
R 0 πR 0
The last integral is the flux rate (see equation (8)), divided by the uniform density ρ, so that
(A.9) is consistent with (9). The most interesting result is that the signal does not depend on
the effective velocity distribution v(r) but only on the integral flux.

References
[1] Hall E H 1879 On a new action of the magnet on electric currents Am. J. Math. 2 287
[2] Boltzmann L 1880 Bemerkung über eine neue Wirkung der Magnete auf elektrische Ströme Acad. Wiss., Wien,
Anz. Beibl. 4 408
Boltzmann L 1880 On gas-friction, and on the velocity of electricity in the electric current Phil. Mag. 9 307
(Engl. transl.)
[3] McKay T C 1906 Titles of articles on the Hall effect, with the titles of some articles on the related phenomena
since discovered by other investigators Proc. Am. Acad. Arts Sci. 41 385
[4] Baumgartner A, Ihn T, Ensslin K, Pappi G, Peeters F, Maranowski K and Gossard A C 2006 Classical Hall
effect in scanning gate experiments Phys. Rev. B 74 165426
[5] Donnan G 1897 Theory of the Hall effect in a binary electrolyte Proc. Phys. Soc. Lond. 16 244
[6] Faraday M 1832 Experimental researches in electricity Phil. Trans. R. Soc. Lond. 122 125
[7] Williams E J 1930 The induction of electromotive forces in a moving liquid by a magnetic field, and its
application to an investigation of the flow of liquids Proc. Phys. Soc. Lond. 42 466
[8] Driver H T S 1978 Comment on “The Hall effect in a flowing electrolyte” Am. J. Phys. 46 1275
[9] von Ettingshausen A and Nernst W 1886 Ueber das Auftreten elektromotorischer Kräfte in Metallplatten, welche
von einem Wärme strome durchflössen werden und sich in magnetischem Felde befinden Acad. Wiss., Wien.
Anz. 23 114
[10] Wright J J and van der Beken S 1972 The Hall effect in a flowing electrolyte Am. J. Phys. 40 245
[11] Geselowitz D B 1972 Comment on “The Hall effect in a flowing electrolyte” Am. J. Phys. 40 1183
[12] Jochim K E 1962 The development of the electromagnetic blood flowmeter IRE Trans. Bio-Med. Electron. 9 228
[13] Giuliano A 2010 Effetto Hall in un flusso di elettroliti in mezzo acquoso BSc Thesis Physics Department, Pisa
University
[14] Kolin A 1945 Alternating field induction flow meter of high sensitivity Rev. Sci. Instrum. 16 109

You might also like