292 IEEE TRANSACTIONS ON ELECTRON

DEVICES,
VOL. ED-29, NO. 2 , FEBRUARY 1982

Electron and Hole Mobilities in Silicon as a Function

of Concentration and Temperature

Abstract-An analytical expression has been derived for the electron effect of electron-electron scattering in the case of electron
and hole mobility in siliconbased on bothexperimentaldata and mobility and hole-hole scattering in the case of hole mobility
modified Brooks-Herring theory of mobility. The resulting expression
was incorporated empirically for dopant densities greater than
allows one to obtain electron and hole mobility as a function of con-
centrationup to -lozo in an extended and continuoustempera- 2 X 10l6 cm-3 following theapproach of Li andThurber
ture range (250-500 K) within *13 percent of the reported experimen- [9], [ 101. For higher concentrations, experimental values of
tal values. mobility were takenfromtheliterature. These mobility
values, both calculated andexperimental(henceforth called
INTRODUCTION experimental for the sake of distinguishing them from those
calculated from the empirical relation to be derived) were then
T HE ELECTRON and hole mobilities in silicon as a func-
tion of dopant
importantparametersfor
concentration
and
temperature
device design and analysis. These
are fitted into thefollowing expression for the mobility:

mobilities have been measured [ 11-[8] at room temperature

for concentrations up to 4 X lo2’ ~ m - However,
~ , very little
data are available [6],[8]at higher temperaturesandfor which is similar to the oneused by Caughey and Thomas [ 141.
higher concentrations. Based on classical theory, Li and Here pminis the minimum mobility value expected, po is the
Thurber [ 9 ] ,[ 101 have calculated mobilities for concentrations difference between the maximum and minimum mobility ex-
upto lo” cm-3 andtemperaturesupto 500 K and have pected, Nrefis a reference concentration, while a is an expo-
found that the calculated values agree well with the experi- nentialfactorthatcontrolsthe slope around N = Nref. All
mental values reported in theliterature. For concentrations four parameters are taken to be constant at a particular tem-
higher than lo” cm-3 the classical theory, however, fails to perature.To define these parametersat afixed temperature
predictcorrect values-it gives values ofmobilitymuch we have used an optimization technique [ 151 which minimizes
lowerthan thosemeasured experimentally. Varioustheories the error
[8], [ 11]-[ 131 have been developed to explain the anomaly
between the theoretical and experimental values, but they are
too complicated to be used for device analysis and design work
where an analytical expression is more appropriate. where pi represents the experimental mobility while pj repre-
In this paper, we present a single empirical relationship for
sents themobilitycalculated by (1) pertainingtothejth
mobility as a function of temperatureandconcentration
concentration N . In thisway, four parameters in (1) can be
which can be used for temperatures up to 500 K and concen-
calculated atdifferent
temperatures.
Thus knowing the
tration up to about 10’’ ~ m - ~ .
temperature variation of these parameters we arrive at analyti-
MOBILITY FORMULATION
METHOD cal expressions, for both electron and hole mobilities similar
to that of (1).
The procedure adopted to arrive at the analytical expression
follows.The values ofmobilityforconcentrationupto RESULTSAND DISCUSSIONS
5 X 1OI8 cmm3 were calculated by appropriatelycombining
A . Electron Mobility
thelatticeandionizedimpurityscattering mobilities. The
The experimental data of Lang [3] and Norton et al. [4] for
Manuscript receivedAugust 16, 1981; revised September 28,1981. lightly dopedn-type silicon inthetemperature range 200-
This workwas supported inpartunder a ResearchGrant from SERI, 500 K were used to calculate the lattice scattering mobility
Golden, CO. yL. It is found that their data can be fitted very well to the
N. D. Arora is with the Department of Electrical Engineering, North
Carolina State University, Raleigh, NC 27650, on leave from the Solid following expression:
State Physics Laboratory, Delhi 110007, India.
J. R. Hauser is with the Department of Electrical Engineering, North pL = 8.56 X lo8’T-2.33 (3 )
Carolina State University, Raleigh, NC 27650.
D. J. Roulston is withtheDepartment of ElectricalEngineering, where T i s the temperature in Kelvins.
University of Waterloo, Waterloo,Ont., Canada N2L 3G1. The ionized impurityscatteringmobility p I is calculated

0018-9383/82/0200-0292$00.75 0 1982 IEEE

ARORA e t al.: ELECTRON AND HOLE MOBILITIES IN SILICON 293

using a modified Brooks-Herring formulation [9] which takes 1000

intoaccounttheanisotrophicscatteringeffectsand can be CURVE TEMPERATURE (OK)

written as # I 200
# 2 300
7.3 X 1017T312 2 800
# 3 400
500
"= NIG(b)
where NI is the number of ionized impurity atoms andG(b) is
a function given by
b
G(b)= In (b t 1) - -
b t l
where

n'
and n' = n [ 2 - (n/N)] which assumes the acceptor concentra-
tion to be zero, n being electron density per cubic centimeter.
After taking into account electron-electron scattering effects
on p~ and p~ as is done by Li and Thurber [9], the two mobili-
ties are combined according to themixedscatteringfor-
2oo
I O 0 1016

Fig. 1. Theelectron
1017 10'6

DOPANT CWCENTRATION ( crn")

1019 10'0

mobility in phosphorous-doped silicon as a

function of concentration for four temperatures 200, 300, 400, and
mula [ 161 500 K. The continuous lines are experimental values of the mobility
(see text) while the dotted lines are the corresponding values calcu-
lated using (8).

where x2 = 6pL/pI, and Ci (x), Si (x) are cosine and sine inte- TABLE I
BESTFITTING PARAMETERS OBTAINED FROM THE MOBILITY DATA
grals of x, respectively. Normally pL and P I are combined (SEE TEXT)
using Mathiessens' rule for the sake of simplicity, but this is
known to give incorrect results [8], [ 161. Thuselectron
mobility can be calculated at any temperature using (3)-(7),
for concentration up to 5 X 10" ~ m - For ~ . higher concentra-
tions, experimental value of mobilities have been taken from Electrons
theliterature. However, at these concentrations (degenerate 200
3103.6 109.7 0.503 0.947
condition) the mobility depends upon the nature of the dopant 300 88.3 1241.8 1.295 0.891
[6], [8]. For phosphorus dopant, the mobility value is 10-15
400 73.76 644.3 2.46 0.823
percent higher than the corresponding value for arsenic dopant
500 66.2 364.0 4.29 0.031
[6], [8]. In what follows, we have taken experimental values
of themobilityforphosphorusdopant. Although there is a Holes
large spread (20-25 percent) in the mobility values reported in
300 54.3 106.9 2.35 0.88
the literature, [ 51-[ 81, [ 121, [ 171 there is consistency in the
observation-the mobility becomes almost constant indepen-
dent of concentrationforconcentration greater thanabout is thenfittedinto (1) using theoptimizationtechnique as
2 X 10'' cmA3 and decreases slowly with increasing tempera- discussed earlier. The values of the four parameters thus ob-
ture.Inourcalculations, we have takenmobility value for tained are shown in Table I. As can be seen from this table all
concentrations greater than 2 X~-I O.l S cm-3 from
~~ ~

- .
experimental four parameters vary with temperature. It is to be noted that
data of Mousty e t al. [5] and Finetti e t aZ. [ 121.for room the values of the four constants at 300 K are very close to
temperature and below as their data seem to be fairly accurate those obtained by Baccarani et al. [7] based on measurements
being within 2 percentoftheexperimentalerror.Fortem- of electron mobility in phosphorus-doped silicon. Taking into
peratures above 300 K, themobility is calculated fromthe account these temperature variations oftheparameters, we
resistivity data of Chapman et aZ. [6],for heavily doped arrive atthe followingempiricalrelation fortheelectron
phosphorus,and used inthepresentcalculations, taking mobility as a hnction of temperature and concentration:
Mousty e t al. [SI mobility value at room temperature as the
base value. These mobility values for four values of tempera- 7.4 X lo8 T-2.33
pe = 88 Ti0.57t
ture 200, 300, 400, and. 500 K are shown as a continuous line 1 t [N/(1.26 X 1017 T,2.4)]0.88
in Fig. 1. Near an electron concentration of 5 X 10l8 ~ m - ~ ,
the data are smoothed out to match between the theoretical (8)
andexperimental values-the maximum difference observed where T, = T/300. The values of electron mobility calculated
was about 7 percent at 5 X 10" ~ m - ~Each . mobility curve from (8) are shown in Fig. 1 as dotted lines for the four tem-
 IEEE TRANSACTIONS ON ELECTRONDEVICES,VOL. ED-29, NO. 2 , FEBRUARY 1982

continuous line fordifferentconcentrations.Thepoints are

experimental values of resistivity taken from theliterature
[ 51, [6], [9], [ 121. It can be seen that the theoretically calcu-
lated resistivities are within8 percent of those measured
experimentallyin thetemperature range 200-500 K except
fortheconcentration range 4-8 X 1OI8 cm-3 where the
deviationbecomes26 percent at 200 K. l%e reason for this
high deviation in this dopant range is because the experimen-
tallymeasured resistivity changes its behavior [ 5 ] , [6], [9],
[ 121, [ 171 from decreasing resistivity, with decrease of tem-
perature, to increasing resistivity through a constant value.
This abnormal behavior of resistivity at lower temperatures is
not taken into account fully in the analyticalexpression(8)
formobility which is used to calculate resistivity from (9).
J
0 n
" IxIo'a
0 This also explains the reason for a larger error in the equation
at lower temperatures.
A
4
A
- 4.4 x I 0 "
B. Hole Mobility
The data given by Li [ 101 fortheholelattice scattering
mobilityinthetemperature range 150-400 K can be fitted
into the following expression:
=
.C~L 1.58 X lo8 T-2*23. (1 0)
I I I I The hole ionizedscattering mobility using themodified
200 Joo 400 500
Brooks-Herring formulation can be written as
TEMPERATURE (OK)

Fig. 2. Resistivity as a functionof temperature for different phosphorus- 5.6 X lOI7 T3I2
(continuous line)andboron- (dotted line) doped silicon slices of
different concentrations. The points are experimental data (0-Li and "= NIG(b)
Thurber, A-Finetti et d.,and O-Chapman e t d.).
where G(b) is given by (5)
2.5 x 1015 T Z
TABLE I1
b=
ROOTMEANSQUARE DEVIATIONS
AND MAXIMUM DEVIATIONS
BETWEEN
P'
THE EXPERIMENTAL
MOBILITY
AND THAT OBTAINEDFROM (8) and
(ELECTRON
MOBILITY)
OR (13) (HOLEMOBILITY)

T('.K) Root
Maximum
mean
square deviations Deviation
Root
Haximum
mean
square deviation deviation
p ' = p (2- ):
(percent) (percent) (percent) (percent)
p is the hole density ( ~ m - ~and) the donor density is assumed
Electrons Roles to be zero.
Again taking into account hole-hole scattering effects on p L
12.17 250 7.03
and as done by Li [ l o ] and combining thetwo mobilities
300 4.06 8.62 4.43 8.9
according to (7), hole mobilities are obtained for concentra-
LOO 4.67 9.6 5.4 13.2 tions up to 2 X 10l8 ~ m - ~For. higher concentrations, experi-
6.13 500 mental values of holemobility have been taken.Thehole
mQbilities thus obtained at room temperature are fitted into
(l), again resulting in four parameters whose values are given
peratures. As can be seen, the dashed curves are quite close to in Table I. These parameters for hole mobility could not be
the continuous line. In fact, as reported in Table 11, the root calculated at higher temperature because of the lack of hole
mean squaredeviation S (2) is less than 7 percent while the mobilitydata available athigh temperaturesandconcentra-
maximum deviation is less than 12 percent in the temperature tions. However, since the three parameters pmin,Nref,and a
range 250-500 K. Although for the sake of clarity Fig. 1 is of (1) are mainly governed by pI and since factors affecting the
drawnforconcentrations greater than10l6 (8) is valid temperature dependence of pI are the same for electrons and
concentrations
for in the range ~ m - ~ . holes (see (4) and(1 l)), these temperature variations have
To further check the validity of (8), the resistivity of n-type been taken here to be the same for electrons and holes. Thus
samples were calculated from the expression we arrive at the following expressions for the hole mobility as
1 a function of temperature and concentration:
p=-.
4NPe
In (9) pe is substitutedfrom (8). The resistivity thuscom- 1.36 X lo8 T-2.23
t (13)
puted as a functionoftemperature is shownin Fig. 2 as a 1 -t [N/(2.35 X l O I 7 TZa4)]0.88 *
ARORA e t al.: ELECTRON AND HOLE MOBILITIES IN SILICON
The maximum deviation i.n the mobility calculated from (13)
and those calculated from(7) and (10)-(12) for concentrations
less than 2 X 10l8 cm-3 are given in Table 11. As in the case
of n-type samples, (13) was used to calculate the resistivity of
p-type boron-doped samples and compared with experimental
value of resistivities reported in the literature [ 6 ] , [ l o ] . The
dashedlinein Fig. 2are the resistivity calculated using (9)
where p e is replaced by fib of (13) while the points are the
correspondingexperimentalpoints. As can be seen,theo-
retically calculated resistivities are within 10 percent of these
measured experimentally except in the resistivity range 5-9 X
cmV3 where resistivity shows anabnormal behavior as
in the case of n-type samples.This further proves the point
that the temperature variation of the parameters pmin, Nref,
and a are essentially the Same for both electrons and holes.
Thus it can be seen that (13) represents hole mobility varia-
tions with temperature and concentration within an error of
less than 14 percent in thetemperature range 200-400 K.
This seems reasonable taking into consideration the spread in
the mobility values reported in the literature.
CONCLUSIONS
In conclusion, an analyticalexpression has been given for
electron and hole mobility in silicon as a function of concen-
tration and temperature. The resulting expression allows one
to obtain electron and hole mobility as a function of concen-
trationin an extendedandcontinuoustemperature range
within 13 percent of reported experimental values.
REFERENCES
[ 11 F. J. Morin and J. P. Maita,“Electrical properties of silicon
containing arsenic and boron,” Phys. Rev., vol. 96, no. 4, pp.
28-35, Oct. 1954.
[2] H. Brooks,“Theoryofelectricalproperties of germanium and
silicon,” in Advances in Electronics and Electron Physics, L.
295
Marton, Ed., vol. 7. New York: Academic, 1955, pp. 85-182.
[31 D. Long, “Scattering of conduction electrons by lattice vibrations
in silicon,” Phys. Rev., vol. 120, no. 6,pp. 2026-2032, Dec.
1960.
[41 P. Norton, T. Braggins, and H. Levinstein, “Impurity and lattice
scatteringparameters as determinedfrom Hall and mobility
analysis inn-type silicon,” Phys. Rev., vol. B8, no. 12, pp.
5632-5653, Dec. 1973.
[51 F. Mousty, P. Ostoja,and L. Passari, “Relationshipbetween
resistivity and phosphorous concentrations in silicon,” J. Appl.
Phys., vol. 45, no. 10, pp. 4576-4580, Oct. 1974.
P. W. Chapman, 0. N. Tufte, J . D. Zook, and D. Long, “Electri-
cal properties of heavily doped silicon,” J. Appl. Phys., vol. 34,
no. 11, pp. 3291-3295, Nov. 1963.
[71 G . Baccarani and P. Ostoja, “Electron mobility empirically related
to the phosphorous concentration,” Solid-state Electron., vol. 18,
no. 6, pp. 579-580, June 1975.
V. I. Fistul, Heavily Doped Semiconductors. New York: Plenum,
1969, p. 77.
S. S. Li and W. R. Thurber, “The dopant density and temperature
dependence of electron mobility and resistivity in n-type silicon,”
Solid-state Electron., vol. 20, no. 7, pp. 609-616, July 1977.
S. S. Li, “The dopant densityand temperature dependence of
hole mobility and resistivity in boron doped silicon,” Solid-state
Electron., vol. 21, no. 10, pp. 1109-111’7, Oct. 1978.
J. B. Krieger, J.Gruenebaum, and T. Meeks, “Resistivity of
n-type degenerately doped silicon at low temperature: Dielectric
screening effect,” Phys. Rev., vol. B9, no. 8, pp. 3627-3629,
Apr. 1974.
M. Finetti and A. M. Mazzoue, “Impurity effects on conduction
in heavily doped n-type silicon,” J. Appl. Phys., vol. 48, no. 11,
pp. 4597-4600, NOV.1977.
L. M. Scarfone and L. M. Richardson, “Electron mobilities based
on an exact numerical analysis of the dielectric-function-
dependent linearized Poisson’s equationfor the potential of
impurity ions in semiconductors,” Phys. Rev. B , vol. 22, no. 2,
pp. 982-990, July 1980.
1141 D. M. Caughey and R. F. Thomas, “Carrier mobility in silicon
empirically related to doping and field,”Proc. IEEE, vol. 5 5 , pp.
2192-2193, Dec. 1967.
R.Fletcher, Rep. R7125, AERE, Harwell, England, June 1972.
P. P. Debye and E. M. Conwell, “Electrical properties of n-type
germanium,” Phys. Rev., vol. 93, no. 4 , pp. 693-706, Feb. 1954.
C. Yamanouchi, K. Mizuguchi, and W. Sasaki, “Electric conduc-
tion in phosphorous doped silicon at low temperatures,”J. Phys.
SOC.Japan, vol. 22, no. 3, pp. 859-864, Mar. 1967.