Anda di halaman 1dari 14

The Hall Effect in Semiconductors

Candidate no. 148658


December 14, 2016

Abstract
The Hall coefficient of an indium antimonide crystal was measured
using a constant magnetic field for a range of excitation currents, at
both room temperature and in a liquid nitrogen bath. InSb was found
to be an N type semiconductor, with Hall coefficient RH = (3.92
0.78)106 m3 c1 and carrier density n = (1.59 0.32)1024 m3 at 298K,
and at 77K RH = (2.11 0.07)102 m3 c1 with a charge carrier density
of n = (2.95 0.10)1020 m3 . The drift velocity of the electrons within
the semiconductor was found from these values as well as an intriguing
residual plot possibly showing unexpected thermodynamic effects.

Declaration
I declare that this report is entirely my own work, and that where supporting
information has been drawn from other sources, those sources are fully and
properly cited.
I am aware of the checklist of frequent errors made in Long Reports, which
has been made available to me via Study Direct. I understand that if I fail to
check my work adequately against that list I am liable to make simple mistakes
that could potentially result in a substantial reduction in marks.
I would particularly welcome feedback with regard to the following aspects
of my report:
(Interpretations of results, how well do my explanations describe the
data.)
(The explanations for interpretations of results and scientific background,
I want to improve my communication skills and ability to explain things
in the most understandable way.)

1 Introduction
When a charged particle moves through a magnetic field it experiences a force
given by [1]
F~ = q(E
~ + ~v B),
~ (1)

1
where ~v is the particles velocity, q is the charge of the particle, q = 1.6 1019 C
is the fundamental unit of charge, B ~ is the magnetic field and E ~ is the electric
field, this relation is known as the Lorentz force law, we can infer from the cross
product that the force on the charge will be perpendicular to the direction of
motion and will be proportional to both the magnitude of the velocity and the
magnetic field. Imagine now that there is a current moving through a conductor,
perpendicular to an applied magnetic field, we can see from equation 1 that the
particles making up the current will experience a force, with a direction given by
the sign of their charge, so the charges will be pushed to one side perpendicular
to the direction of the magnetic field, however as the charges build up they
repel each other, producing an electric field that opposes the collection of more
charges, very quickly the system reaches equilibrium where the Lorentz force
on the charges is balanced by the electrostatic repulsion, this charge imbalance
across the conductor creates a potential difference perpendicular to current flow,
this is known as the Hall voltage and what has just been described is the basics
of the Hall effect.

1.1 Brief Introduction to Semiconductor Physics

Figure 1: The position of the Fermi level in an insulator, the vertical axis gives
the energy level of the different bands, image from the 2nd year lab script.

A convenient way of modeling conduction in matter is through the band


theory. Conduction is just the movement of charge carriers within some mate-
rial, electrons in matter are bound to atomic nuclei and in some cases there are
also electrons free to move between different atoms, the outer bound electrons
are known as the valence electrons and the free ones make up what is known
as the conduction band. The Pauli exclusion principle states that [2] Fermions
(particles with non integer spin, such as electrons) must be in separate quantum
states, which include the particles spin and position. This makes the electrons
in matter form into bands of closely spaced energy levels as many electrons get

2
Figure 2: The position of the Fermi level in a conductor, image from the 2nd
year lab script.

as close to each other as they can without being in the same space, there can
be gaps within these bands where no electron can reside. The Fermi level is the
energy level of the electrons in their ground (minimum energy) state, its the
position of the Fermi level within band space that determines the conductive
properties of a material. If its within the valence band then the materials is an
insulator, the valence band is the collection of possible energy levels of electrons
within the outer shell of the atoms, these electrons are bound to the atoms so
the valence electrons cannot move throughout the material so it is an insulator
(provided the applied potential difference is not great enough for the electrons
to overcome the band gap). In a conductor the Fermi level is higher, within the
conduction band, and also the valence band and conduction band can overlap.
So even in their ground state some electrons are not bound to any particular
atom and are free to move throughout the material without any energy input
to raise them on the band diagrams, allowing conduction, these free electrons
are why metals make such good conductors.
Finally there are semiconductors, which have a smaller bandgap than insula-
tors or conductors, with an energy difference between the valence and conduction
band of only Eg = 0.1to0.3eV . Familiar examples of semiconducting materi-
als include silicon and gallium, semiconductors can be used to make diodes,
transistors and other solid state devices which have had a tremendous impact
on the modern world, there are very few things not drastically changed by the
introduction of the transistor, but its not just relevant to electrical engineer-
ing, there is also interest in the Hall effect as a way that angular momentum
can be transferred in protoplanetary clouds, leading to the creation of stars
and planets [3] among many other parts of science, so their study is of great
interest to physicists and electrical engineers. There are naturally occurring
semiconductors but the most effective ones are created by doping the material
i.e adding small amounts of impurities that contribute extra energy levels within

3
Figure 3: The position of the Fermi level in a P type semiconductor, image from
the 2nd year lab script.

Figure 4: The position of the Fermi level in an N type semiconductor, image


from the 2nd year lab script.

the forbidden region, shifting the height of the Fermi level.


There are two types of semiconductors, N type and P type. In a P type
semiconductor the Fermi level is just above the valence band see figure 3, so
with the application of a voltage across the material electrons are raised from
the valence band to fill the energy levels contributed by the dopant, this leaves a
net positive charge in the valence band, the electrons move around this and it is
often easier to consider it as a positively charged particle called a hole rather
than think about the motion of other electrons around an uncharged region,
this hole can move freely throughout the valence band as the electrons around
it dont need any extra energy as they stay in the same band, so the material
can conduct as if it contained positive charges.
In an N type semiconductor the conduction is not due to the movement
of holes but of electrons, the doping agent adds extra energy levels close to

4
the conduction band, electrons can be excited easily from these energy levels
to the conduction band as the Fermi level is so close to it, see figure 4, so in
an N type the majority charge carriers are electrons. As the probability of an
electron going above the Fermi level is related to its thermal energy there is
a correlation between the temperature and the charge carrier density, this is
especially true for intrinsic (naturally occurring semiconductors) rather than
extrinsic ones which are doped with impurities, at room temperatures most
semiconductors will contain both types of charge carriers.
The probability that an electron will cross the Fermi level is proportional
to its kinetic energy and therefore the temperature of the material [4], at high
temperatures its likely that the semiconductor contains conducting electrons
and holes, these of course are attracted to each other reducing the electric
field they generate, and the apparent magnitude of the Hall effect. In order to
investigate the effects of a dopant upon some semiconductor it is necessary to
simplify the situation, this is done by submersion in liquid nitrogen, which has
a boiling point of 77.36K, when a material undergoes a phase change it stays
at the same temperature throughout until the entire material is changed [4],
assuming heat can flow easily enough through the system, in other words as
long as heat in is not hugely greater than the maximum speed it can propagate
through the material. So when the sample is submerged in liquid nitrogen we
know the temperature very accurately as long as the nitrogen has not completely
evaporated away, the nitrogen boiling point is also dependent on the air pressure
but the temperature doesnt need to be known to that degree of accuracy for
this experiment so it was not taken into account. At lower temperatures fewer
electrons have enough thermal energy to be excited into the conduction band
or one of the extra energy levels in P doped semiconductors so there is a much
greater majority of the dominant charge carrier at low temperatures, this effect is
particularly significant in intrinsic or naturally occurring semiconductors, rather
than those that have been deliberately doped as there is fewer extra energy levels
within the band gap.
The experiments goal was to measure the effect of changing currents on an
indium antimonide sample within a magnetic field, with a range of currents going
through it. Then repeating while submerging the sample in liquid nitrogen,
to investigate the temperature dependence of semiconductivity. From these
measurements the majority charge carrier was found as well as the drift velocity
of these charge carriers and the number of them per unit volume.

2 Analysis
The diagram in figure 5 shows the forces acting on positive and negative charges
within a conductor with an applied magnetic field, B ~ is the applied magnetic
~ is the electric field produced by the build up of charges on either side
field, E
of the conductor, and VH is the voltage across the conductor, in electrostatic
conditions, which apply to this situation as the forces between the charges reach
an equilibrium very quickly E ~ = VH = VH [1] where w is the width and
w

5
Figure 5: Forces on moving charge carriers in a magnetic field, image from the
2nd year lab script

t is the thickness of the conductor, so we can measure the voltage parallel to


the current flow and from this find E~H . J is the current density going through
the conductor, and ~v is the drift velocity of the charge carriers. Notice how the
positive and negative charges go in different directions through the conductor,
from equation 1 we can see the direction of the force is given by both the sign of
the charge and the direction of the velocity so both charge carriers are pushed in
the same direction, this is important as charges with opposite signs attract, so in
the presence of both types of charge carriers there will be another electrostatic
force in the opposite direction to E~H that opposes E~H .
Once the system reaches equilibrium there are no forces on the charges gen-
erating the Hall voltage, so from equation 1 we get

q E~H + q~v B
~ =0 (2)

E~H = ~v B
~ (3)
The current density is given by J~ = qn~v where n is the charge carrier density,
if we define the Hall coefficient to be
1
RH = (4)
nq
and use the identity ~u ~v = ~v ~u we can write

E~H = RH B
~ J.
~ (5)

6
So from this relation we can find RH and from it both the sign and the
density of the charges within the conductor. using the relations for J and EH
previously stated we can express RH in terms of the measured parameters as
long as the magnetic field is perpendicular to the current flow, giving

~
RH B
VH = I. (6)
w
A linear fit was performed on the data, equation 6 was fitted to y=mx+c by
minimising the 2 , giving the gradient

~
RH B
m= (7)
w
which when rearranged gives
mw
RH = . (8)
B~

The uncertainties where propagated using the following formula


y 2 y
y(a, b)2 = (a ) + (b )2 (9)
a b
this relationship is easily extended for functions of more than two variables.
The uncertainty on RH was found to be
s
mw 2 mw 2 mwB ~
RH = ( ) +( ) + ( )2 (10)
~
B ~
B B~2

where the error on the gradient is given by the linear fitting algorithm used.
From RH we can use relationship 4 to find the charge carrier density within
the semiconductor
1
n= , (11)
RH q
with error given by
RH
n = | 2 q | (12)
RH
where we are taking the error on q to be zero as it is so small its insignificant
compared to the error on RH .
An interesting extension to the experiment is to use the definition of the
current density J~ = nq~v to find the drift or average velocity of the charge
carriers [2]. When J~ is perpendicular to B,
~ from equation 5 we get

E~H
J~ = = nq~v , (13)
RH B~
so

7
E~H VH
~v = = . (14)
~
RH Bnq ~
RH Bnqw
With the following uncertainty

r
VH VH 2 w 2 n 2 RH 2 B 2
v = ( ) +( ) +( ) +( ) +( ) . (15)
RH Bnqw VH w n RH B
There is a small vertical offset between the terminals used to measure the
Hall voltage, so they intersect the current flowing perpendicular to them, this
produces another voltage that has to be taken into account to find the Hall volt-
age, by rotating the InSb sample half a revolution with respect to the magnetic
field we can account for this offset, as the current flow is perpendicular to the
magnetic field and along the axis of rotation its in the same direction in both
configurations, but the direction of E~H flips so the voltage across the sample
is V1 = VR + VH , then after rotating the sample the sign of the Hall voltage
changes so V2 = VR VH , as VR is the same we can rearrange to get
V1 + V2
VH = , (16)
2
so by measuring the voltage across the terminals with both directions of
magnetic field the offset can be removed.

3 Experimental Procedure
In order to ascertain the Hall coefficient a permanent magnet with a known field
of 0.159 0.005T was used to produce the magnetic field across the sample, a
current controlled power supply was used to excite the sample with 5 to 50 mA,
measured with a digital multimeter, the sample itself was placed in an insulated
glass flask in order to contain the liquid nitrogen, and finally a sensitive digital
voltmeter was used to measure the voltage across the sample.
In figure 6 A is the dewar that holds the liquid nitrogen, B is an ammeter
used to measure the current through the sample, C is the current source, D is
the flask that holds the liquid nitrogen when its applied to the sample, E is the
position of the sample itself, F is the permanent magnet and G is a sensitive
voltmeter used to measure the voltage across the sample.
To simplify the analysis the experiment was performed with the current
flow perpendicular to the direction of the magnetic field, as EH = RH B ~ J~ =
RH BJsin() where is the angle between the vectors B ~ and J,~ so the maximum
Hall field will occur when sin() = 1, so when = /2rad, to find this position
15mA of current was passed through the sample and it was revolved in the
magnetic field until the voltage across the sample was maximized, at a value of
0.34mV , it was possible to change the angle of the sample without a measurable
change in voltage, so this could be a source of error, however if the change was

8
Figure 6: Equipment used to perform the experiment, image from study direct.

too small to measure it is taken into account by the error due to the finite
precision of the measuring device, which was just taken to be the least significant
digit of whatever scale was being used at the time. The voltage across the
sample was measured twice for each current, with the magnetic field in both
orientations, equation 16 was then used to find the Hall voltage.
Once the sample was set up data was taken for currents in the range 5mA to
50mA, going up in 5mA increments, the power dissipated by a resistive element
is P = IV , so there was resistive heating of the sample, which was proportional
to the current across it, to combat this readings were taken as quickly as possible,
a better way would have been to connect the sample to a thermal reservoir of a
known temperature that is large enough that the heat generated by the sample
has a negligible effect on the temperature of the reservoir, the liquid nitrogen
acted in this way for the cold experiment, the energy dissipated by the sample
just made the nitrogen boil faster. Upon initial placement of the semiconductor
in the liquid nitrogen there was violent boiling until the temperatures equalized,
and measurements could be taken.

4 Results
First a thermometer was used to measure the ambient room temperature, which
was found to be T = 298 1K, from here on Ill refer to this temperature as
300K for brevity and neglect the error as it is used as a name not as data, the
conclusions drawn depend on a temperature difference between the two sets of
measurements, not on its precise value.

9
Figure 7: Hall Voltage against Excitation Current at 300K

Figures 7 and 8 show the results of the experiment, there is a clear linear
relationship between current and voltage in both, although the cryogenic re-
sults look a lot better, this is partly reflected in the quality of fits, the 300K
data produced a 2 of 1.517 with a p value of 0.99, which is not acceptable to
confirm the relation if the convention that p between 0.05 and 0.95 is used, the
77K experiment gives a 2 of 4.48 which is quite high, higher than the 300K
result,which is puzzling as the fit visually looks worse at 300K, it does however
give a p value of 0.81, suggesting that we can reject the null hypothesis that
there is no correlation between VH and the current.
From the gradient of these fits values for the Hall coefficient where found
using equation 6, at 300K RH = (3.92 0.78)106 m3 C 1 giving a carrier
density of n = (1.59 0.32)1024 m3 . The 77k experiment gave RH = (2.11
0.07)102 m3 C 1 with a charge carrier density of n = (2.950.10)1020 m3 , so at
both temperatures the majority charge carriers are electrons as RH is negative,
indicating that InSb is an N type semiconductor, The higher carrier density at
room temperature shows the relation between the energy of the electrons and the
probability that they can rise into the conduction band, the room temperature
electrons are moving around more, thus more pass into the conduction band
raising the carrier density.
The drift velocities found using equation 14, for 300K at the maximum Hall
voltage magnitude of VHmax = (2.35 0.05)104 V the maximum drift velocity
is vmax = 9.9 2.9m/s, using the maximum Hall voltage at 77K which is
VHmax = (0.111 0.0005)103 V we get vmax = 4690 23m/s, showing that the
electrons are able to move much more easily through the sample when it is colder,
as the atoms that make it up are vibrating much less there are fewer collisions
slowing down the flow of charge, through a rather convoluted method we have
demonstrated that temperature is proportional to resistance. These results also
highlight the fascinating fact that although energy moves through conductors
at a significant portion of the speed of light [1] the electrons transmitting this

10
Figure 8: Hall Voltage against Excitation Current at 77k

energy have an average speed of only a few metres per second.


Figures 9 and 10 show the residuals, or the difference between the value
measured and the one predicted by the coefficients found in the linear fit. Ideally
they would be randomly distributed around and close to the center axis, showing
that the errors in the experiment are randomly distributed throughout the stated
uncertainty on each measurement. In the 300K graph the last data point, the
one corresponding to I = 50mA is the furthest by far, as this is the largest
current this suggests that at some current near this the resistive heating of the
sample starts to significantly change the properties of the semiconductor.
The 77K residuals look very odd, with a shape almost like part of an ex-
ponential curve bent around or a rotated quadratic function. This is probably
because when starting the experiment its possible that the sample had not
cooled fully, the rate of change of temperature is proportional to the temper-
ature difference [4], so as the temperature of the sample gets close to 77K the
heat transfer becomes very slow, so although it may have appeared that the
temperatures had equalised there was actually a small difference, as the cur-
rent increases the sample has to dissipate more power, if the readings are taken
quickly the nitrogen could absorb the heat slower than its produced, so the
sample heats up, and produces the curve, the apex on the right might be the
point where the resistive power going in becomes greater than the amount of
heat that can escape through the samples surface per second. It would be in-
teresting to repeat the experiment to investigate this further, but this time take
the measurements at a range of fixed time intervals, and also measure the volt-
age across the sample parallel to the current flow so the power into the sample
could be calculated and it might be possible to deduce the reason behind the

11
Figure 9: Residuals for 300K data

function seen in figure 10. In order to improve the other results obtained in
this experiment it would probably be enough to perform the experiment much
more slowly, so the temperatures equalise between each reading, maybe with
the addition of a heat sink to the sample, or maybe stir the liquid nitrogen to
increase the flow of heat from the InSb crystal into the liquid, so the effects of
the power dissipation become negligible.

5 Conclusion
The Hall coefficient was measured at both room and cryogenic temperature, with
negative RH each time, so the results suggest that the majority charge carriers
are electrons, so InSb is an N type semiconductor. The charge carrier density
was found to increase with temperature which suggests that the electrons higher
thermal energy increased the probability that they could move to the conduction
band. The Hall coefficient decreased by 4 orders of magnitude between the
cold and the hot experiments, which is quite significant, with a constant B ~
field equation 5 suggests RH is the proportionality constant between the Hall
electric field and the current density, so perhaps this decrease is because of the
increased likelihood of electrons being able to reach the extra bands in the band
gap at room temperature, so there was many more holes present than at 77K
and the attraction between these and the electrons decreased the size of E~H ,
and therefore RH .
The drift velocity of the electrons was found to be much higher at cryogenic
temperatures, in many materials resistance is proportional to temperature [5],
using Ohms law V = IR we can see that a smaller resistance leads to a higher
current for the same applied voltage, current is just the rate of change of charge,
so is proportional to how many electrons pass through in a given time, so if theres
higher current then the electrons must have a higher velocity to satisfy charge

12
Figure 10: Residuals for 77K data

conservation.
The biggest source of uncertainty in the experiment is probably the resistive
heating, the 300K results have pretty noticeably deviations from the trend line,
much more so than the 77K results, during the room temperature experiments
the sample was just in air, which is not a particularly effective dissipater of heat,
perhaps performing the experiment in a more controlled environment such as a
water bath could produce a lower 2 and a better p value, or maybe it would
be better to measure the sample temperature, and take this into account with
a more sophisticated model of semiconductor behavior than the one used here.
The deviations in the room temperature graph could also be because the Hall
voltage is lower at 300K so the signal to noise ratio is higher, a stronger magnetic
field or greater current density would produce a larger VH making the effects
of any noise less significant, alternatively the voltage could be measured with
no current so the noise levels could be measured and their amplitude could be
compared to VH to see if they could have had an effect.
Resistive heating also seems to have produced what is probably the most
interesting part of the experiment, which is the 77K residuals. If the experiment
where repeated seeing if the data could be fitted to some simple models of heat
flow into and out of the sample it might be possible to find out whether this is
the cause of the graphs strange curve, or perhaps its just an error in the analysis
or a statistical fluke, it would be nice to know which one it is.

Acknowledgements
I would like to thank my lab partner for all their help throughout the module,
I would also like to thank the ATs and lecturers for all the interesting things

13
theyve taught me as well as whoever has to build, setup and maintain the
experimental apparatus.

References
[1] David J.Griffiths, Introduction to Electrodynamics. Pearson, 4th Edition,
2013.
[2] Paul A. Tipler, Ralph A. Llewellyn, Modern Physics, 3rd Edition, 1978.
[3] Edward Liverts, Michael Mond, Arthur D. Chernin, The Hall instabil-
ity of weakly ionized, radially stratified, rotating disks, January 2007,
Astrophys.J.666:1226-1231,2007, DOI: 10.1086/520489.
[4] Enrico Fermi, Thermodynamics, Dover Publications Inc., 1st Edition, 1956.
[5] Paul Horowitz, Winfield Hill, The Art of Electronics, Cambridge University
Press, 3rd Edition, 1980.

14

Anda mungkin juga menyukai