Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/230947479

A fresh look at the semiconductor bandgap using constant current data

Article  in  European Journal of Physics · June 2011


DOI: 10.1088/0143-0807/32/5/003

CITATIONS READS

2 977

2 authors, including:

R. O. Ocaya
University of the Free State
64 PUBLICATIONS   183 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

A new shape memory alloy film/p-Si solar light four quadrant detector for solar tracking applications View project

New semiconductor devices with emphasis on Schottky devices View project

All content following this page was uploaded by R. O. Ocaya on 27 May 2014.

The user has requested enhancement of the downloaded file.


Home Search Collections Journals About Contact us My IOPscience

A fresh look at the semiconductor bandgap using constant current data

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

2011 Eur. J. Phys. 32 1155

(http://iopscience.iop.org/0143-0807/32/5/003)

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

Download details:
IP Address: 146.182.9.9
The article was downloaded on 22/01/2013 at 08:11

Please note that terms and conditions apply.


IOP PUBLISHING EUROPEAN JOURNAL OF PHYSICS
Eur. J. Phys. 32 (2011) 1155–1161 doi:10.1088/0143-0807/32/5/003

A fresh look at the semiconductor


bandgap using constant current data
R O Ocaya1,3 and P V C Luhanga2
1 Department of Physics, University of the Free State, P Bag X13, Phuthaditjhaba 9866,
South Africa
2 Department of Physics, University of Botswana, P Bag UB 0022, Gaborone, Botswana

E-mail: ocayaro@qwa.uovs.ac.za and luhangap@mopipi.ub.bw

Received 22 March 2011, in final form 23 May 2011


Published 27 June 2011
Online at stacks.iop.org/EJP/32/1155

Abstract
It is shown that the well-known linear variation of p–n diode terminal voltage
with temperature at different fixed forward currents allows easy and accurate
determination of the semiconductor ideality factor and bandgap from only two
data points. This is possible if the temperature difference required to maintain
the same diode voltage drop can be measured. The results for silicon and
germanium bandgap energy using this approach are in excellent agreement
with the literature. The method therefore provides a fresh and original insight
into the derivation of the bandgap using data from a popular experiment. The
experiment is suitable for undergraduate laboratories on semiconductors.

1. Introduction

The measurement of the semiconductor bandgap by electrical methods has been a favourite
experiment in many undergraduate physics curricula for decades. Typically, the semiconductor
is in the form of an encapsulated germanium or silicon diode. A constant current is maintained
through the diode as its temperature is varied. The voltage developed across the diode is
measured and the bandgap is determined graphically from a plot of diode voltage versus
temperature. Several data points are needed to produce the best linear plot. In addition, the
experimenter typically assumes a value of the ideality factor [1–3]. The latter turns out to
be requisite for an accurate bandgap result. The method in this paper is novel because it
allows direct and accurate determination of both the ideality factor and the bandgap from
at most two data points. This makes it suitable for a moderately equipped experimenter.
The feedback we have received from colleagues familiar with the earlier methods has been
one of amazement at the relative simplicity of the method. Indeed, some have found the
underlying mathematical approach attractive that leads to a representation of the bandgap
3 Author to whom correspondence should be addressed.

0143-0807/11/051155+07$33.00 
c 2011 IOP Publishing Ltd Printed in the UK & the USA 1155
1156 R O Ocaya and P V C Luhanga

in terms of voltages and temperature ratios only. This paper underscores the importance of
looking anew at old equations for innovative approaches to extracting data. The experiment is
easily achievable in the undergraduate laboratory and will benefit students in furthering their
understanding of the semiconductor bandgap and its measurement. With careful design the
measurements can be automated using a microcontroller but this is left as a future student
project.

2. Theory

The current I through a p–n diode at applied voltage V and junction temperature T is given by
the equation
 Eg 
I = In exp − [exp(qV /nkT ) − 1], (1)
nkT
where In is a constant, k is Boltzmann’s constant, q is the electronic charge, n is a diode-
dependent ideality factor, T is the absolute temperature and Eg is the bandgap of the
semiconductor [4–6]. Under the condition that qV /nkT  1,
 Eg   qV 
I ≈ In exp − exp . (2)
nkT nkT
Alternatively,
 
Eg (0) nkT I
V = + ln , (3)
q q C1 T C 2
where C1 and C2 are constants [7]. The ideality factors reported in the literature range from
1.4 to 2.0 for silicon, and 1.7 to 3.0 for germanium [8]. The bandgap varies with temperature
[4, 9] and an uncertainty analysis of the bandgap between 60 and 80 ◦ C gives a bandgap
variation of about 0.001 eV for silicon. A smaller temperature variation is therefore desirable
when determining the bandgap. The uniqueness of this approach is in showing that if the
constant forward current through the diode is switched in a controlled manner, then it is
possible to estimate n and Eg using only measurements of the temperatures at which the new
constant current reasserts the set forward voltage as the temperature changes. Figure 1 shows
the typical curves for the 1N4148 diode with constant currents below 4 mA, for which self-
heating effects are negligible [5]. Now consider equation (1) and figure 2, which represents
two constant currents of figure 1. Suppose that the diode is heated to a high initial temperature
T0 beyond point A in the figure, whereupon the heat source is removed. With the forward
current set to I1 the diode then cools to T1. The forward voltage of the diode is then measured
as VF 1 (point B in figure 2), and the diode current is then switched to a new constant value
I2, where I2  I1 . As the diode continues to cool down under the new current I2 its terminal
voltage will eventually reach VF again, at which the measured temperature will be T2. These
measurements are sufficient to determine the ideality factor and bandgap. In the calculations
that follow the Boltzmann constant and electron charge used are respectively 1.3806 × 10−23
J C−1 and 1.6022 × 10−19 C.

2.1. Determination of the ideality factor

At a temperature T the use of two sequentially switched currents gives the data pair: (T, I1,
V1) and (T, I2, V2). Substituting the data pair into equation (3) allows the difference voltage
V (T ) = V1 (T ) − V2 (T ) to be readily found. Consequently,
qV (T )
n= . (4)
kT ln(I1 /I2 )
A fresh look at the semiconductor bandgap using constant current data 1157

Figure 1. Typical variation of forward diode voltage with temperature and current for the 1N4148
silicon diode. (Reproduced from [5].)

Figure 2. Depiction of the cooling and measurement trajectory (solid lines) under constant voltage
and currents.

Hence, using equation (4), the ideality factor can be estimated in the part of the measurement
trajectory labelled B→C, where the temperature is T1, and V (T1 ) = VF 1 − VF 2 .

2.2. Determination of the bandgap

The implication of constant current switching on bandgap calculation can be seen when the
two currents are expressed as a ratio, e.g.
I2 = aI1 ,
where 0 < a < 1. It follows that at the two temperature endpoints for which the diode voltage
is the same (i.e. VF), equation (1) can be written as
   
Eg qVF
I1 ≈ In exp − exp . (5)
nkT1 nkT1
and    
Eg qVF
aI1 ≈ In exp − exp . (6)
nkT2 nkT2
Dividing equations (5) and (6) and defining a new temperature T ∗ such that
1 1 1 T1 T2

= − , or T∗ = , (7)
T T2 T1 T 1 − T2
1158 R O Ocaya and P V C Luhanga

leads to
Eg = qVF − nkT ∗ ln a (in joules, J) (8)
or
nkT ∗
Eg = VF − ln a (in electron volts, eV). (9)
q
The bandgap can be estimated in the part of the measurement trajectory labelled C→D, where
a characteristic temperature T ∗ can be determined.

2.3. Alternative representation of the bandgap

An alternative representation of the bandgap arises when equation (4) (with T = T1 ) is


substituted into equation (9). That is

−qV (T1 ) kT ∗ ln a
Eg = VF − . (10)
kT1 ln a q
Substituting T ∗ from equation (7) gives
T2
Eg = VF + V (T1 ). (11)
T1 − T2
This result shows that the bandgap can easily be determined by measuring the voltage change
at two temperatures if the currents are switched in a known ratio at the two temperatures.
In other words it is not necessary to precisely know the values of the currents themselves
provided the switching is repeatable. Equations (4) and (9) may be used directly as the basis
for the experimental measurement of the ideality factor and bandgap.

3. Experimental details

3.1. The setup

The simplest approach submerses the diode in a constantly stirred hot water bath at boiling
point that is gradually cooled towards 0 ◦ C. The diode temperature is measured on any suitable
thermometer, preferably a type K thermocouple with 0.1 ◦ C resolution. Figure 3 is a suitable
adaptation of the two-transistor current source [6, 10]. The additional transistor Q3 switches
the load on the collector circuit of Q1 depending on the state of its base current. The output
current of the simple current mirror is given by
VA
I= , (12)
RC
where RC is the effective collector resistance of Q1, and the potential (VA) at point A is
nominally (VCC − VBE ) above ground. When point C is at 0 V (logic 0), Q3 is off and
therefore the collector of Q1 has an effective resistance RC equal to (RC1 + RC2 ). When point
C is at +5 V (logic 1), transistor Q3 is driven into full conduction and RC2 is effectively shorted
to ground. Point B is therefore at 0 V and the effective collector resistance is RC1 . Thus

VCC − VBE

VCC − VBE

I1 =
, or I2 =
. (13)
RC1
RC1 + RC2

C=1 C=0

Table 1 gives the transfer characteristics of the current source under digital control.
A fresh look at the semiconductor bandgap using constant current data 1159

Figure 3. A voltage of 0 or 5 V applied to point C makes the current either I1 or I2. See table 1.

Table 1. Transfer characteristics of the switchable current source. Note that RC1 = 39k and
RC2 = 8.2k.

Point C VA (V) VB (V) Output current (μA) Label

0V 4.33 2.3 92 I2
5V 4.31 0.0 115 I1

4. Results

Table 2 shows typical results obtained for both germanium and silicon using off-the-shelf
diodes. For germanium diodes, it was observed that constant currents lower than 100 μA
through the diode reduce the voltage–temperature linearity. Much higher currents should
also be avoided to limit diode self-heating. The temperatures and voltages were measured to
within 0.1 ◦ C and 0.001 V, respectively. The currents were measured to be within within 2%
when resistor tolerances of 5% or better were used. To obtain the worst-case, non-calculus
estimates of uncertainty in n and Eg, consider the following illustration using an arbitrary
function f = f (x). If the worst variation in x is δx, then the worst estimate of the function is
f1 = f (x ± δx) provided that δx is signed to maximize f = |f − f1 |. For example, using
equation (4) it can be seen that the uncertainty in n is due to the measurement uncertainty in
T1 and V (T1 ), and how accurately the currents I1 and I2 can be set. Consequently the worst
possible value of n is given by
q V (T ) + δV (T )
nworst = . (14)
k (T − δT ) ln[(I1 − δI1 )/(I2 + δI2 )]
Hence n = ±|n − nworst |. A similar estimate is possible for Eg . The calculated ideality
factors, bandgaps and uncertainties are shown in table 2. The average values were found to
be n = 2.415 ± 0.347 and Eg = 0.977 ± 0.165 for the OA91 diode, n = 1.857 ± 0.347 and
Eg = 1.185 ± 0.163 for the 1N4148 diode. Iterated calculations show that the uncertainty
in the voltage measurement is the most significant factor in determining the uncertainty in n
and Eg. When determining n the switching of diode current from I1 to a new steady value
I2 occurs well before the bath temperature changes appreciably from T1. The variations in
1160 R O Ocaya and P V C Luhanga

Table 2. Typical results for germanium (OA91 diode) and silicon (1N4148 diode).

VF 1 VF 2 T1 T2 Eg ± Eg
(V) (V) (◦ C) (◦ C) n ± n (eV)

OA91
0.078 0.067 50.0 44.6 1.787 ± 0.322 0.731 ± 0.149
0.101 0.088 40.0 34.9 2.110 ± 0.322 0.868 ± 0.160
0.128 0.114 30.0 25.1 2.386 ± 0.344 0.974 ± 0.166
0.158 0.143 20.0 15.3 2.609 ± 0.356 1.060 ± 0.171
0.190 0.175 10.0 5.4 2.775 ± 0.368 1.104 ± 0.170
0.207 0.192 5.0 0.5 2.825 ± 0.375 1.125 ± 0.172

1N4148
0.449 0.438 50.0 45.1 1.835 ± 0.323 1.190 ± 0.149
0.472 0.461 40.0 35.2 1.844 ± 0.333 1.185 ± 0.167
0.495 0.484 30.0 25.3 1.871 ± 0.344 1.187 ± 0.166
0.518 0.508 20.0 15.5 1.864 ± 0.356 1.192 ± 0.166
0.541 0.531 10.0 5.6 1.856 ± 0.368 1.181 ± 0.165
0.552 0.542 5.0 0.6 1.871 ± 0.375 1.174 ± 0.162

the measured ideality factor and bandgap due to fluctuations in the currents are small since
the currents contribute in the ratio. This minimizes the effect of bias circuit changes on the
calculations.

5. Conclusions

We have described a simple but insightful technique that allows the bandgap of a semiconductor
to be determined accurately over a narrow temperature range of about 5 ◦ C. Thus the
spread of the measured bandgap with temperature is limited. By switching the currents
in a nominally small but known ratio we show that bandgap measurement can be reduced to
simple measurements of voltage and temperature changes. The results obtained for germanium
using the OA91 diode show a maximum standard deviation of 17% in the bandgap and ideality
factor, while those for silicon using the 1N4148 diode show a maximum spread of only 0.8%.
The non-calculus uncertainty analysis shows that the accuracy of the results depends mostly
on the voltage measurements. Germanium point-contact diodes such as the OA91 exhibit high
nonlinearity in their voltage–temperature characteristic due to high internal resistance [11].
This restricts their useful range when they are used as temperature sensors and also accounts
for the high variance observed in the results for the OA91 diode in table 2. The experimental
determination of the bandgap using this method will likely produce the best results with silicon
diodes. The experiment is suitable for undergraduate laboratories on semiconductors. A close
inspection of the method will show that it is possible to automate the measurements using a
suitably programmed microcontroller, although this is left as scope for future work by our
students.

References
[1] Collings P J 1980 Simple measurement of the band gap in silicon and germanium Am. J. Phys. 48 197–9
[2] Fraser D A 1986 The Physics of Semiconductor Devices (Oxford Physics Series) 4th edn (Oxford: Clarendon)
A fresh look at the semiconductor bandgap using constant current data 1161

[3] Shockley W 1976 Electrons and Holes in Semiconductors (New York: Krieger)
[4] Van Zeghbroeck B 2004 Principles of Semiconductor Devices http://ece-www.colorado.edu/∼bart/book/
[5] Ocaya R O and Dejene F B 2007 Estimating p–n diode bulk parameters, bandgap energy and absolute zero by
a simple experiment Eur. J. Phys. 28 85–91
[6] Ocaya R O 2006 An experiment to profile the voltage, current and temperature behaviour of a p–n diode
Eur. J. Phys. 27 1–9
[7] Sze S M and Ng K K 2007 The Physics of Semiconductor Devices 3rd edn (New York: Wiley)
[8] Fischer C W 1982 Elementary technique to measure the energy band gap and diffusion potential of p–n junctions
Am. J. Phys. 50 1103–5
[9] Bludau W, Onton A and Heinke W 1974 Temperature dependence of the band gap of silicon
J. Appl. Phys. 45 1846–8
[10] Horowitz P and Hill W 1989 The Art of Electronics 2nd edn (Cambridge: Cambridge University Press)
[11] The Omega passive receiver http://www.neazoi.com/xtalpage/index.htm (last accessed May 2011)

View publication stats

You might also like