Journal of Molecular Structure (Theochem) 664-665 (2003) 135143

www.elsevier.com/locate/theochem

A simple program to determine the reaction rate

and thermodynamic properties of reacting system
Patrcia R.P. Barretoa,*, Alessandra F.A. Vilelab, Ricardo Garganob
a

Laboratorio Associado de Plasma-LAP, Instituto Nacional de Pesquisas Espaciais-INPE/MCT,

CP515 Sao Jose` dos Campos, SP, CEP 12247-970, Brazil
b
Instituto de Fsica, Universidade Braslia, CP04455, Braslia, DF, CEP 70919-970, Brazil
Received 14 April 2003; accepted 11 August 2003

Abstract
We developed a simple program to determine the reaction rate by using conventional transition state theory with the Wigner
transmission coefficient and also the thermodynamic properties of the species. The hydrogen abstraction in the reaction
NH3 H ! NH2 H2 is used as a model to demonstrate usage of the program. The rate constants have been computed for the
gas-phase chemical reaction over the temperature range of 200 4000 K. Our data are in a good agreement with the published
data for this reaction.
q 2003 Published by Elsevier B.V.
PACS: 82.20.Pm; 82.20.Db; 82.20.Wt; 82.30.Hk; 82.30.Lp; 51.30. i; 82.60.-s
Keywords: Kinetics; Thermal rate constants; Conventional transition state theory; Direct dynamics; Bimolecular abstraction;
Thermodynamic properties

1. Introduction
There has been considerable interest in recent
years, in the growth of boron nitride thin films. Like
carbon, boron nitride has different allotropes, the most
common of which are the hexagonal (hBN) and cubic
(cBN) phases. The hBN, although electrically insulating, has properties that are very similar to graphite
while the cBN has properties comparable to diamond.
At present, there is little understanding of the
chemical process which are involved in and which
* Corresponding author. Tel.: 55-12-3945-6694; fax: 55-123945-6710.
E-mail address: patricia@plasma.inpe.br (P.R.P. Barreto).
0166-1280/$ - see front matter q 2003 Published by Elsevier B.V.
doi:10.1016/j.theochem.2003.08.098

GAUSSIAN 98;

control the synthesis of either hexagonal or cubic

boron nitride from the vapor phase. Theoretical
research found in the literature includes thermodynamic equilibrium calculations for mixtures involving
B/F/N/H [1 5], B/Cl/N/H [4,5], and B/N/H [4], as
well as limited kinetics studies of the reactions
between BCl3 and NH3 [6,7].
Barreto et al. proposed a kinetic mechanism to
describe the growth of boron nitride films, for the
Ar/B/F/N/H system, using both CVD and arcjet
reactor [8 12]. The gas-phase mechanism includes
35 species and 1012 reactions and also extends
a previous mechanism that contained 26 species and
67 elementary reactions. Rate constants for 117
elementary reactions were obtained from published

136

P.R.P. Barreto et al. / Journal of Molecular Structure (Theochem) 664-665 (2003) 135143

experimental/theoretical data and those for the other

895 reactions should be estimated using Transition
State Theory (TST). As the number of reactions are
large, we developed a simple code to determine the
reaction rate using conventional TST with the Wigner
transmission coefficient, since we normally work at
high temperature, 1000 3000 K, in which the tunneling effects decreases.
To show the capability of our code, we applied it to
the well-known reaction NH3 H ! NH2 H2,
since it has been studies by several groups who
provided many theoretical [13 18] and experimental
results [19 22] in the temperature range of 500
2000 K. Comparing the results obtained using our
code with the published data, our data are in a good
agreement and now we can apply our code to the
reaction in a Ar/B/F/N/H system.
Following, we outline how the paper is organized.
In Section 2, we present the main characteristics of
our program. Our results and discussion are presented
in Section 3.

2. The program
Our program has basically two subroutines, one to
calculate the thermodynamic properties of the species
and the other one to determine the rate constants. The
input files are very easy to write down and the output
files are very simple to analyze, and they can be
exported, directly, to graphical software. The following sections will discuss the calculation of the reaction
rate using TST [23 26] and the determination of the
thermodynamic properties [23,25].
2.1. Transition state theory
The TST was developed in the 1930s and has since
formed a framework for much of the discussion of rate
processes. It is a model to determine the rate constants
based on an interaction potential between reactant and
products with a statistical representation of the
dynamics. In addition to the Born Oppenheimer
approximation, it uses three more assumptions: there
is a surface in phase space that divides into a reactant
region and a product region; the reactant equilibrium
is assumed to maintain a Boltzmann energy distribution and the transition state or activated complex is

assumed to have Boltzmann energy distribution

corresponding to the temperature of the reacting
system.
Considering a bimolecular reaction, such as
A BC ! AB C

where A, B or C are atoms or group of atoms, the

thermal rate constants is given by
!

kB T QX
VaG
kTST
2
exp 2
h QA QBC
RT
where QX ; QA and QBC are the partition functions of
the transition state, X (saddle point), and the reactant,
A and BC, respectively, kB Boltzmann constant, h
Planck constant, T temperature and R universal gas

constant, VaG is the barrier

VaG V 1ZPE

where V is the classical potential energy of the saddle

point measured from the overall zero of energy and
1ZPE is the harmonic zero-point energyZPE.
The partition functions, QA ; QBC and QX ; are
given by
Qj Qtrans Qrot Qvib Qelect ;
j A; BC; X

where Qtrans ; Qrot ; Qvib and Qelect mean the component

of the translation, rotation, vibration and electronic
partition functions.
The translation partition function is


2pmkB T 3=2
Qtrans
5
h2
where m gives the mass of the atom or group of atom.
The rotation partition function depends on the
molecule structure, and for a linear molecule, the
rotation partition function, Qrot-2D ; is
!
8p2 IkB T
6
Qrot-2D
se h 2
whereas for a nonlinear molecule, the rotation
partition function, Qrot-3D ; is
"
!3=2 #
p1=2 8p2 Im kB T
7
Qrot-3D
se
h2

P.R.P. Barreto et al. / Journal of Molecular Structure (Theochem) 664-665 (2003) 135143

where I and Im IA IB IC denote the moment of inertia

and the product of inertia, respectively, se is the
symmetry number of the molecule.
The vibrational partition function and ZPE, in the
harmonic approximation, are given by


Y
2hvi 2di
Qvib
1 2 exp
8
kB T
i
1ZPE

1 X
h di v i
2 i

where vi is the frequency, i 1; ; F and F is the

degree of freedom (F 3N 2 5 for linear molecule,
3N 2 6 for nonlinear molecule and F 2 1 for the
transition state) and di is the mode degeneracy.
The electronic partition function is given by


X
2e i
Qelect
gi exp
10
kB T
i
where gi is the degeneracy, and e i is the energy of
electronic state g measured from the overall zero of
energy.
We have also included the transmission coefficient,
kT; in Eq. (2)
W
kTST
T

kTkTST T

11

where the transmission coefficient is used to account

for the tunneling effect along the reaction coordinate.
We have decided to estimate the transmission
coefficient by using the Wigner correction [13,23,
26] instead of some semiclassical tunneling approximation [26]. The Wigner correction for tunneling
assumes a parabolic potential for the nuclear motion
near the transition state and therefore, cannot be
considered as an accurate correction. The Wigner
transmission coefficient is given by
2

1  "v 
kT 1
12


24  kB T 
where the imaginary frequency at the saddle point is
denoted by v :
For more precise rate constants calculation, the
variational TST should be used and also semiclassical
methods for the tunneling effect [26], yet in this case,
more information about the potential energy surface is
required, but it is not employed in the present work. A
useful way to verify the importance using

137

the variational transition state instead of conventional

transition state is the curvature of the reaction path.
A constraint on reaction path curvature is provided
by the skew angle [26]:

1=2
mA mC
b ArcCos
13
mA mB mB mC
where mA ; mB and mC are the masses of the A, B and
C moieties, respectively, for the schematic reaction
shown by Eq. (1). Large reaction path curvature is
often encountered in the tunneling region in system
with small skew angles, so the tunneling effects
should be smaller.
For a generic bimolecular reaction
kf

A BOC D

14

kr

where the rate constant in the forward direction, kf ;

determined by TST, may be conveniently written in
the Arrhenius form as


2Ea
n
kf AT exp
15
RT
where A is the pre-exponential factor, n is the
temperature power factor and Ea is the activation
energy.
The rate constants in the reverse direction, kr ; can
be determined using the forward constant rate and the
equilibrium constant, Kc ; as:
kr

kf
Kc

Remembering

Pns
p
Kc Kp
RT

16

17

and
DG 2RT ln Kp

18

so, we can rewrite the reverse reaction rate as


Pns

 

RT
DS
DH
exp
kr kf
exp 2
19
p
R
RT
P
where ns is the stoichiometric factor that is defined
to be positive for products and negative for reactants
and p is the pressure.

138

P.R.P. Barreto et al. / Journal of Molecular Structure (Theochem) 664-665 (2003) 135143

2.2. Thermodynamic properties

Gibbs free energy, can be obtained combining the

previous properties as:

To determine the reverse rate constants, Eq. (16), it

is necessary to know the thermodynamic properties
such as entropy and enthalpy for the reactants and
products.
The thermodynamic properties, internal energy,
entropy, and heat capacity, can be written in terms of
the partition functions as:


ln Q
E kB T
20
ln T v


ln Q
S kB ln Q kB
21
ln T v
!


ln Q
2 ln Q
cv kB
k
:
22
ln T v B ln T2 v

H ; E pV

23

G H 2 ST:

24

The partition functions are given by Eqs. (4) (10).

Other thermodynamic properties such as enthalpy,

Our code prints out the entropy and sensible enthalpy,

Hsen H 2 Df H 0 ; for all species, reactants, products
and saddle point, as a functions of temperature.

3. Results and discussion

To apply TST, we must know the geometries,
frequencies, and the potential energy for reactants and
saddle point. These properties are obtained from
accurate electronic structure calculation performed
using the GAUSSIAN 98 program [27]. First, we
determined the reactants and saddle point geometry
and frequencies using the second order Mller Plesset

Table 1
) and bond angle (8)) of all stationary points of test reaction NH3 H ! NH2 H2
Geometrical parameters (bond distance (A
Parameter

This worka

NH3
RNH
AHNH

1.0168 (1.0124g)
106.4004 (106.67g)

NH2
RNH
AHNH

1.0283 (1.024g)
103.3089 (103.4g)

H2
RHH
TS
RNH
RNH1
RH1 H2
AHNH
AHNH1
ANH1 H2
DHNH1 H2
a
b
c
d
e
f
g

Reference
Bb

Cc

1.007
107.4

1.012
106.1

1.01
106.1

1.023
102.7

1.02
102.7

0.734

0.73

0.7375 (0.7414g)
1.019
1.290
0.920
106.40
100.20
162.59
57.48

MP2/6-31G(d).
HF/SSANO [13].
MP2/6-31G(d) [14].
MP2/6-31* [15].
Corchado et al. [16]
MP2/6-31G(d,p) [17].
JANAF table [28].

1.015
1.247
0.944
100.2
162.6
53.8

1.023
1.305
0.869
103.5
99.1
158.5
52.7

Dd

1.023
1.305
0.869
103.5
99.1
158.5

Ee

Ff

1.013
1.286
0.861

177.7

1.02
1.30
0.87
103.5
99.1
158.5
52.7

P.R.P. Barreto et al. / Journal of Molecular Structure (Theochem) 664-665 (2003) 135143

perturbation theory and the triple-zeta plus polarization 6-31G(d). Furthermore, we calculated the
energies for all species, reactants and saddle point,
at the MP2/6-311 G(3df,3pd), MP4(SDTQ)/631G** and B3LYP/6-311 G(3df,3pd) level.
The geometries computed at MP2/6-31G(d) for
reactant, product and saddle point are shown in
Table 1. These geometries are also compared with
reference data. The interatomic distance and bond
angle absolute errors for the reactants and products
and 0.38, respectively,
are smaller than 0.0044 A
when compared with the experimental results of
and 1.08, respectively, for the other
0.0098 A
calculated results. While the interatomic distance
and bond angle absolute errors for the transition
and 4.788, respectively, exception
state are 0.058 A
for the angle ANH1 H2 ; where this error is 15.118,
between our data and the Corchado and EspinosaGarcia one [16]. Fig. 1 shows the optimized
geometry of the saddle point, and also the notation
used in Table 1.
The harmonic vibrational frequencies calculated
from ab initio methods are overestimated, in this case,
in about 8%, then we have used a scaling factor of
92%. The scaled frequencies are also compared with
published data, as shown in Table 2. The absolute
error among our calculated and experimental frequencies was less than 118 cm21, except for the HH
stretching, where this error was 235 cm21. Comparing
the absolute error of our calculated frequencies with
published calculated frequencies for the reactants and
products, we observe for the NH stretching, an error of
480 cm21, when for the other frequencies, this error
was less than 189 cm21. The absolute error in
frequency for the transition state was less than
435 cm21, except for the imaginary frequency,
where the minimum error was 181 cm21 for the
Henon data [13] and the maximum error was
1119 cm21 for the Corchado data [16].

Fig. 1. Atomic numbering of the reaction system for reaction

NH3 H ! NH2 H2.

139

Table 2
Harmonic vibrational frequencies (cm21) and zero-point energy
(kcal mol21) of reaction NH3 H ! NH2 H2
This worka

Reference
Bb

Cc

1146
1803(2)
3709
3841(2)

1067
1642(2)
3386
3535(2)

1067
1642(2)
3386
3535(2)

NH2
1501 (1497g)
3147 (3219g)
3259 (3301g)

1516
3307
3417

1516
3307
3417

H2
4170

4301

4301

NH3
1068 (950g)
1614(2) (1627g)
3224 (3337g)
3368(2) (3444g)

TS
622
652
1033
1310
1490
1538
3243
3366
2536i
a
b
c
d
e
f
g

695
726
1209
1408
1633
1684
3666
3761
2717i

683
694
1089
1249
1530
1973
3314
3425
1920i

Dd

723
732
1258
1485
1541
1731
3647
3750
2833i

Ee

563
598
1038
1327
1640
1902
3383
3446
1417i

Ff

683
694
1089
1249
1530
1973
3314
3425
1920i

MP2/6-31G(d).
HF/SSANO [13].
MP2/6-31G(d) [14].
MP2/6-31* [15].
Corchado et al. [16].
MP2/6-31G(d,p) [17].
JANAF Table [28].

3.1. Thermodynamic properties calculations

The first subroutine in our code calculates the
thermodynamic properties, particularly the entropy
and enthalpy, Eqs. (21) and (23). Fig. 2 compares
the calculated properties with the reference data
from the JANAF tables [28] for the reactants, NH3
and H, products, NH2 and H2 and, also, gives the
thermodynamic properties for the transition state
(TS NH4).
Compared with reference data [28], the maximum
absolute errors in our calculation and the reference
data are 4.71 kcal K21 mol21 for the NH3 entropy at

140

P.R.P. Barreto et al. / Journal of Molecular Structure (Theochem) 664-665 (2003) 135143

Fig. 2. Comparisons of thermodynamic properties, enthalpy and

entropy, obtained by our code (symbols) with experimental data
from JANAF table [28] (lines).

4000 K and 2.41 kcal mol21 for the NH3 enthalpy at

298.15 K. Fig. 3 shows the maximum, minimum and
average error in the entropy and enthalpy for each
species analyzed here.
3.2. Rate constants calculations
The total energies for reactants, products and saddle
point as well as the reaction enthalpy and potential
barrier calculated by different methods and basis set
are shown in Table 3, in which we applied the ZPE
correction in the reaction enthalpy and potential
barrier. We found a large range of values for the
potential barrier, 14.39 25.37 kcal mol21, for the
methods and basis set used, as well as, for the reaction
enthalpy, where this range is 0.12 8.68 kcal mol21.

Fig. 3. Absolute error for calculates thermodynamic properties,

enthalpy and entropy, obtained by our code with experimental data
from JANAF table [28].

The same behavior, for the potential barrier, is found

in the literature, range of 7.6 22.6 kcal mol21
[13 15,17].We would like to emphasize the calculation made in MP4(SDTQ)/6-31G**, where the
absolute error in the potential barrier was
0.1 kcal mol21 when compared with the Garrett et al.
[15] results for the same methods and basis set.
The skew angle in this case is 46.78, as it is a midvalue angle, we should expect same tunneling effects.
For reactions with intermediate/high-value of the
skew angle, the Wigner transmission coefficient is not
the best correction. On the other hand, we are
interested in a high temperature range, 1000
3000 K, where this problem is greatly reduced.
Fig. 4 compares the reaction rate obtained in this
work, considering conventional TST and also applying the transmission coefficient of Wigner (TST/W)
with data from the reference [13 22]. Several
reaction rates were calculated, each one considering

P.R.P. Barreto et al. / Journal of Molecular Structure (Theochem) 664-665 (2003) 135143

141

Table 3
Total energy Ea ; potential barrier and reaction enthalpy (kcal mol21) including ZPE correction, where A MP2/6-31g(d), B MP2/6311 g(d,p), C MP4/6-31g**, D B3LYP/6-311 g(3df,3pd)

NH3
H
NH2
H2
TS
Reaction enthalpy
Potential barrier

256.354211
20.4982329
255.6908562
21.1441408
256.8096988
7.82
25.37

256.4567673
20.4998179
255.7728291
21.1649448
256.9196000
8.68
21.76

256.4013892
20.4982329
255.7298852
21.1645717
256.8693051
0.12
17.58

256.4284264
20.4998098
255.7544436
21.1683161
256.9030052
0.32
14.39

B3LYP/6-311 G [13], V 7:6 kcal mol 21 ; CAS(3,3)/SANO [13], V 22:6 kcal mol 21 ; PMP4/6-311 G(d,p) [14],
V 15:7 kcal mol21; MP2/6-31* [15], V 21:59 kcal mol21; MP4/6-31** [15], V 17:68 kcal mol21; MP2/6-31G(d,p) [17]
V 10:8 kcal mol21.

the different energy calculated by the four methods

and basis set. One can see from Fig. 4 that the rate
constants obtained for MP4/6-31G** level both agree
with experimental and theoretical results, in the
temperature range, 500 2000 K.
Fig. 5 compares the reaction rate obtained in this
work considering the energy determined by MP4/631G** level, with the data from the POLYRATE 8.7
program [29], considering TST, TST/W, canonical

variational transition state theory (CVT) and also

CVT with zero curvature tunneling (CVT/ZCT). It is
possible to see that the POLYRATE curves from TST
and CVT are coincident. In the region of low
temperature, T , 500 K, the reaction rate obtained
by CVT/ZCT considerably differs from our calculation and also the TST and CVT calculation, since
the skew angle has a moderate-value, 46.78, it is
excepted that the tunneling effects is not well

Fig. 4. Comparisons of Arrhenius plot of k against the reciprocal temperature (K) in the range of 4004000 K for reactions NH3
H ! NH2 H2 obtained in this paper with reference data where C; D; H; J and K are theoretical data from Refs. [14,16,18,15,17], respectively,
and E; F; G and I are experimental results from Refs. [21,22,20,19], respectively.

142

P.R.P. Barreto et al. / Journal of Molecular Structure (Theochem) 664-665 (2003) 135143

Fig. 5. Comparisons of Arrhenius plot of k against the reciprocal temperature (K) in the range of 2004000 K for the reactions NH3
H ! NH2 H2 obtained in this paper with data from the POLYRATE program, considering TST, TST/W, TST/MEPSAG, CVT, CVT/ZCT.

determined by the Wigner transmission coefficient.

The ratio between the POLYRATE rate constant (TST
or TST/W) and ours (TST or TST/W) are 22 times at
500 K and drops to seven times at 800 K and drops
more while the temperature increases. It does not
differ considerably if we consider the TST or the TST/
W rate constants.

4. Conclusion
Our purpose was to develop a simple code to
determine the reaction rate for the B/F/H/N system
easily. We opted to write our own code instead of
using the available ones such as POLYRATE program,
for example, since in that way we can simplify the
inputs files and write output files that can be used
directly by some graphical software. Our code allows
us to determine several reaction rate at once since we
have one input file for all the reactants and one input
file for the transition state and one input file describing
which and how many reactions we want to analyze.
The analysis of reaction NH3 H ! NH2 H2
allowed us to check the capability of our code. Our
code reproduce the reaction rate from the reference
data and from the POLYRATE program in a high

temperature range, T . 500 K. Now we can extend

this methodology for other reactions and systems.

Acknowledgements
The authors are grateful to the Centro Nacional de
Computacao Cientfica de Alto Desempenho de
Campinas-SP (CENAPAD-SP) for the provision of
computational facilities. A.F.A. Vilela acknowledges
the Conselho Nacional de Desenvolvimento Cientfico e Tecnologico (CNPq) for the fellowship which
provided her the opportunity to perform this work.

References
[1] W. Kalss, R. Haubner, B. Lux, Diamond Relat. Mater. 7
(1998) 369.
[2] O. Dugne, S. Prouhet, A. Guette, R. Naslain, J. Alloy Compd
176 (1991) 187.
[3] H. Hannache, R. Naslain, C. Bernard, L. Heraud, in:
Proceedings of the Fourth European Conference on Chemical
Vapour Deposition, 1983, p. 305.
[4] S. Matsumoto, N. Nishida, K. Akashi, K. Sugai, J. Mater. Sci.
31 (1996) 713.
[5] H. Hannache, R. Naslain, C. Bernard, J. Less-Common Met.
95 (1983) 221.

P.R.P. Barreto et al. / Journal of Molecular Structure (Theochem) 664-665 (2003) 135143
[6] A.H. McDaniel, M.D. Allendorf, J. Phys. Chem. A 102 (1998)
7804.
[7] M.D. Allendorf, C.F. Melius, J. Phys. Chem. A 101 (1997)
2670.
[8] P. Barreto, M. Cappelli, S. Matsumoto, in: 2000 International
Congress on Plasma Physics, vol. 3, 2000, p. 904.
[9] P. Barreto, A.E. Kull, M. Cappelli, in: 68 Encontro Brasileiro
de Fsica dos Plasma, 2001, p. 30.
[10] P. Barreto, in: 68 Encontro Brasileiro de Fsica dos Plasma,
2001, p. 31.
[11] P.R.P. Barreto, M.A. Cappelli, Private communication,
2000.
[12] P.R.P. Barreto, Desenvolvimento de Mecanismo Cinetico
para o Crescimento de Nitreto de Boro, Relatorio de Pesquisa
INPE-8701-PRP/227, Instituto Nacional de Pesquisas Espaciais, Sao Jose dos Campos SP, 2002 (in Portuguese).
[13] E. Henon, F. Bohr, J. Mol. Struct. (Theochem) 531 (2000) 283.
[14] J. Espinosa-Garcia, J.C. Corchado, J. Chem. Phys. 101 (1994)
1333.
[15] B.C. Garrett, M.L. Koszykowski, C.F. Melius, M. Page,
J. Phys. Chem. 94 (1990) 7096.
[16] J.C. Corchado, J.E. Garcia, J. Chem. Phys. 106 (1997) 4013.
[17] J.E. Garcia, S. Toloza, J.C. Corchado, J. Phys. Chem. 98
(1994) 2337.
[18] P. Marshall, A. Fontijn, J. Phys. Chem. 91 (1987) 6297.

143

[19] T. Ko, P. Marshall, A. Fontijn, J. Phys. Chem. 94 (1990) 1401.

[20] J.V. Michael, J.W. Sutherland, R.B. Klemm, J. Phys. Chem.
90 (1986) 497.
[21] P. Marshall, A. Fontijn, J. Chem. Phys. 85 (1986) 2637.
[22] W. Hack, P. Rouveirolles, H.G. Wagner, J. Phys. Chem. 90
(1986) 2505.
[23] D. Rapp, Statistic Mechanics, Holt, Rinehart & Winston, New
York, 1972.
[24] M.J. Pilling, P.W. Seakins, Reaction Kinetics, second ed.,
Oxford Science, Oxford, 1995.
[25] D.M. Golden, Evaluation of chemical thermodynamics and
rate parameters for use in combustion modeling, in: W.
Bartok, A.F. Sarofim (Eds.), Fossil Fuel Combustion: A
Source Book, Wiley, New York, 1991, pp. 49, (Chapter 2).
[26] D.G. Truhlar, A.D. Isaacson, B.C. Garrett, Generalized
transition state theory, Theory of Chemical Reaction
Dynamics, vol. 4, CRC Press, Boca Raton, Florida, 1985,
(Chapter 2).
[27] M.J. Frisch, et al., GAUSSIAN 98, Revision a.9, Gaussian,
Pittsburgh, PA, 1998.
[28] J.M.W. Chase, J. Phys. Chem. Ref. Data (1998) (Monograph
9).
[29] J.C. Corchado, et al., PolyrateVersion 8.7, Department
of Chemistry and Supercomputer Institute, University of
Minnesota, Minneapolis, Minnesota, 2001, p. 55455.