Ab Initio Quantum Chemical and Experimental (Shock Tube) Studies of the Pyrolysis Kinetics of Acetonitrile

1054

J. Phys. Chem. A 1999, 103, 1054-1072

Ab Initio Quantum Chemical and Experimental (Shock Tube) Studies of the Pyrolysis

Kinetics of Acetonitrile

Karina Sendt, Emi Ikeda,^‡George B. Bacskay, and John C. Mackie*

School of Chemistry, UniVersity of Sydney, NSW 2006, Australia

ReceiVed: August 26, 1998; In Final Form: December 1, 1998

The pyrolysis kinetics of acetonitrile dilute in argon has been studied in the temperature range 1400-2100

K at an average pressure of 12 atm in single-pulse shock tube experiments. The principal products are HCN,

C₂H₂, CH₄, and H₂, while the minor products include HCCCN, H₂CCHCN, C₂H₄, and C₄H₂. The overall

kinetics is successfully simulated by an 87 step kinetic model that accurately accounts for the temperature

profiles of the major products and also provides an acceptable fit for the minor products. The thermochemistry

and rate parameters of a number of key reactions have been obtained by ab initio quantum chemical calculations

carried out at CASSCF, CASPT2, and Gaussian-2 levels of theory. Several distinct reaction pathways were

studied, whereby the geometries, vibrational frequencies, and energies of approximately 70 molecular species

representing reactants, products, intermediates, and transition states were computed. The pyrolysis of acetonitrile

is initiated by CH bond fission, forming a cyanomethyl radical. This reaction is the most sensitive one in the

kinetic model. On the basis of sensitivity analyses of the model as well as ab initio calculations, the heat of

formation of cyanomethyl has been revised as ∆_fH⁰₂₉₈(CH₂CN) ) 263 ( 9 kJ mol^-1. The limiting high-

pressure value of the corresponding rate constant, as obtained by ab initio variational transition state calculations,

is k_1∞) 1.2 × 10¹⁶exp(-413 kJ mol^-1/RT) s^-1, which is in good agreement with our extrapolated experimental

measurement. A number of the observed products, including HCCCN and H₂CCHCN, largely arise from the

decomposition of succinonitrile, a key intermediate, that forms by the recombination of two cyanomethyl

radicals.

Introduction

+ H, Lifshitz et al. did not carry out detailed kinetic modeling

of the pyrolysis. More recently, Ikeda and Mackie³presented a

detailed kinetic model for the fuel-rich oxidation of acetonitrile.

Their oxidation model contained an acetonitrile pyrolysis model

as a subset. This subset contained several pyrolysis reactions,

not sensitive under oxidative conditions, that warrant further

investigation. In a more recent study, Lifshitz, Tamburu, and

Carroll⁴presented a kinetic model for the ignition delay in

mixtures of acetonitrile and oxygen in argon.

The observation of significant amounts of H₂CCHCN and

HCCCN in the products is intriguing since Lifshitz et al.²did

not report any formation of the expected product of recombina-

tion of two CH₂CN radicals, viz., succinonitrile, NCCH₂CH₂-

CN, even at quite low reaction temperatures. Evidently, this

molecule should be an intermediate and it must decompose in

an unexpectedly facile way to produce acrylonitrile and cyano-

acetylene. Thus, because of the importance of the pyrolysis

mechanism in the understanding of oxidation of organonitriles,

a further investigation of the pyrolysis of acetonitrile and the

development of a detailed reaction mechanism is warranted.

Furthermore, a study of the electronic structure and reactions

of CH₂CN is of importance in its own right, since cyanomethyl

is the simplest nitrile analogue of the important CH₃radical

and it is of interest to understand the effect of CN substitution

on the electronic properties of a simple alkyl radical.

To reduce the emission of pollutant oxides of nitrogen (NO_x)

from the large-scale burning of coal and other nitrogen-

containing fuel, an important strategy is to use staged combus-

tion involving a primary reaction zone that is fuel-rich.¹Most

fuel-bound nitrogen in coals of varying rank is found in five-

membered pyrrolic ring structures and in six-membered pyridinic

structures. Fuel-rich combustion of these heterocyclic structures

takes place initially essentially as pyrolysis in which volatile

nitrogen-containing precursors of NO_xsuch as HCN and aceto-

nitrile are evolved and undergo complex oxidative processes

eventually to form NO and other products of oxidation. It is

through initial rapid pyrolysis of the volatile precursors that

radical species are produced for subsequent oxidation reactions.

Thus, a detailed understanding of the pyrolysis of a model pre-

cursor such as acetonitrile, CH₃CN, is of importance for unravel-

ling the subsequent oxidative reaction pathways leading to NO.

The pyrolysis of CH₃CN was investigated some years ago

by Lifshitz, Moran, and Bidani²using single pulse shock tube

techniques at temperatures between 1350 and 1950 K and at

pressures around 3 atm. They showed that HCN and acetylene

were principal products of the pyrolysis initiated through CH

fission to produce the cyanomethyl, CH₂CN, radical. Lifshitz

et al. also observed additional products including acrylonitrile,

H₂CCHCN, and cyanoacetylene, HCCCN, thought to arise from

combination of two CH₂CN radicals. Although they reported a

value for the rate constant for the reaction CH₃CN f CH₂CN

The present work involves an experimental shock tube kinetic

study of the pyrolysis kinetics of acetonitrile dilute in argon

over the temperature range 1400-2100 K at an average pressure

of 12 atm and residence times of the order of 0.8 ms, together

with an ab initio quantum chemical investigation of several key

elementary reactions in the pyrolysis. In particular, we report

* Corresponding author. E-mail: j.mackie@chem.usyd.edu.au. FAX:

+61-2-9351-3329.

^‡Present address: Department of Chemical Engineering, University of

Illinois at Chicago, Chicago IL 60607.

Published on Web 02/05/1999

Pyrolysis Kinetics of Acetonitrile

J. Phys. Chem. A, Vol. 103, No. 8, 1999 1055

ab initio derived rate constants for the initiation reaction and

for hydrogen atom addition to acetonitrile and a detailed

investigation of the succinonitrile potential energy surface and

of possible routes to acrylonitrile and cyanoacetylene. From the

experimental data and the ab initio calculations we derive a

detailed kinetic model of the pyrolysis of acetonitrile.

While this manuscript was under preparation, Lifshitz and

Tamburu⁵published a modeling study of their earlier acetonitrile

pyrolysis data.²The rate constants presented in their kinetic

model, in contrast with our work, were optimized to fit their

experimental product profiles, rather than derived on the basis

of ab initio calculations. Consequently, it is instructive to

compare several of our theoretically rate derived constants with

those presented in their model.

In the flow reactor studies, reactants and volatile products were

analyzed using the same gas chromatograph. In the CSTR

experiments condensable products were extracted from the cold

traps using dichloromethane and the extracts analyzed by GC.

Some scatter in the chromatograms of cyanoacetylene occurred.

Separation of this species from acetonitrile was troublesome at

low levels of CH₃CN because the acetonitrile peak eluted on

the leading edge of the cyanoacetylene peak. This led to

integration problems when the residual acetonitrile was low and

the yield of HCCCN high, causing some scatter in the recorded

cyanoacetylene yields. Light hydrocarbons were analyzed with

a Shimadzu GC-9A with an alumina column and FID. A sample

of the product mixture was taken in a gastight syringe for

hydrogen analysis using a Shimadzu GC-8A with TCD.

Product identification was made by means of GC/MS using

a Hewlett-Packard MS Engine also using a 5890II GC and HP1

column, and where possible, confirmed by retention time

matching with authentic samples. When these were not available,

the online NBS MS library NBS 54K was used.

Experimental Section

The principal reactor used in this study is the single-pulse

shock tube. The shock tube has been used previously to study

the pyrolyses of coal model compounds such as pyrrole⁶and

the butenenitriles.⁷The temperature range covered in the present

study was 1400-2100 K over the pressure range 10-14 atm

with a residence time behind the reflected shock front between

700 and 1000 µs.

Theoretical and Computational Methods

The potential energy surfaces (PES) of the reactions were

studied by using ab initio quantum chemical methods, computing

the geometries, energies, and vibrational frequencies of the

appropriate species that correspond to stationary points, viz.

equilibrium and transition state geometries. For those systems

where a single reference expansion of the wave function is valid,

the Gaussian-2 (G2) method¹⁰was used. For singlet biradical

species and a number of transition states that were found to

display an appreciable degree of near-degeneracy, the geometries

and vibrational frequencies were obtained using the complete

active space (CASSCF) method^11,12in conjunction with Dun-

ning’s¹³double-ú plus polarization functions (DZP) basis. A

given CASSCF calculation with nel active electrons in an active

space of nao orbitals will be denoted CASSCF(nel/nao). In this

work, depending on the system, these multiconfigurational SCF

calculations range from CASSCF(3/4) to CASSCF(10/10).

When applied to larger systems, such as the decomposition

reactions of succinonitrile and acrylonitrile, the G2 method was

modified so as to make these calculations feasible on our smaller

workstations. First, in the modified method, the ∆E(2df)

correction is made at the MP2 level of theory (instead of MP4).

Second, the MP2/6-311+G(3df,2p) energy is estimated as

Some experiments were also carried out in a fused silica jet-

stirred reactor that has been described previously.⁸The well-

stirred nature of this reactor was established in earlier studies

of the pyrolysis of cyclopropane.^8,9Samples such as acetonitrile

that are liquid at room temperature are injected by means of a

syringe pump into a vaporizing oven where the vapor is mixed

with a metered stream of argon from a mass flow controller.

The reaction mixture then flows into the jet-stirred reactor, which

is mounted inside an electrical furnace. When operated at

pressures around 100 Torr and a residence time of 1 s, the

reactor well approximates a completely stirred reactor (CSTR).

This reactor is equipped with tandem cold traps and is useful

for studying trace products of reaction. Reactions of acetonitrile

and succinonitrile dilute in argon have been studied in the CSTR

over the temperature ranges of 1240-1320 K for acetonitrile

and 670-970 K for succinonitrile, respectively. Succinonitrile

is a solid at room temperature but is readily soluble in

acetonitrile. Acetonitrile solutions of succinonitrile were injected

into the CSTR after vaporization and dilution of the vapor with

argon. Succinonitrile decomposed over a temperature range

below the onset of decomposition of acetonitrile itself; hence

the former can be studied without significant interference from

acetonitrile. A further advantage of acetonitrile as a solvent for

succinonitrile as opposed to, e.g., a hydrocarbon solvent is that

cyanomethyl radicals arising from CC fission in succinonitrile

will merely be regenerated if they undergo hydrogen abstraction

reactions with acetonitrile, thus minimizing secondary product

formation and enabling us to search for possible intramolecular

routes to acrylonitrile and HCN.

In the shock tube studies, reactant mixtures were prepared

dilute in argon. The acetonitrile concentration was 0.4% in

argon. Acetonitrile (Waters Associates, HPLC Grade, >99%)

was thoroughly degassed and used without further purification.

There was no detectable impurity by GC. In the CSTR studies,

the total reactant concentration (either pure acetonitrile or succo-

nitrile/acetonitrile mixture) in the vapor phase was 4% in argon.

Solutions of succinonitrile in acetonitrile were 13 mol %.

Analyses of the reactant mixture and products in the shock

tube experiments were carried out using an on-line Hewlett-

Packard 5890II GC, equipped with a HP1 capillary column and

detection by FID and a nitrogen phosphorus detector (NPD).

E(MP2/6-311+G(3df,2p)) ≈ E(MP2/6-311G(3df,2p)) +

E(MP2/6-311+G(2df,p)) - E(MP2/6-311G(2df,p)) (i)

When tested for a few of the smaller systems studied here, the

results obtained by this modified G2 theory were found to be

very close to the G2 results, the average difference in the

calculated energies being ∼3 kJ mol^-1

.

Given that the wave functions in the G2 method are based

on single reference expansions, G2 cannot be used to describe

singlet biradical systems. On the other hand, contrary to our

initial expectations that it would run into difficulties for

transition states with an appreciable degree of near-degeneracy,

G2 was found to perform well in the calculation of most

transition state energies.

In the study of the decomposition reactions of succinonitrile,

the energies were also computed at the CASPT2^14,15level of

theory (at CASSCF/DZP geometries) using the cc-pVDZ basis

set^16,17of Dunning. In these calculations the reference states

range from CASSCF(6/6) to CASSCF(10/10). For selected

pathways the energies were also calculated at the CASPT2 level

1056 J. Phys. Chem. A, Vol. 103, No. 8, 1999

Sendt et al.

of theory with a CASSCF(10/10) reference in conjunction with

the cc-pVTZ basis set,¹⁷but with the 4f and 3d polarization

functions removed from the heavy and H atoms, respectively,

resulting in a valence triple-ú basis with two sets of polarization

functions: VTZ2P.

In the case of simple bond fission reactions where there is

no energy barrier along the reaction coordinate, i.e., no saddle

point on the potential energy surface, the rate constants were

obtained using Variational Transition State Theory.²¹This entails

the computation of the rate constant along the reaction coor-

dinate and locating its minimum, which then defines the

transition state. For the CH bond fission reactions in our work

the walk along the reaction coordinate was performed by starting

with a geometry that is essentially that of the product, but with

a CH bond distance of 2.0 Å. Allowing for a series of Newton-

Raphson relaxations with a small trust radius, i.e., small steps,

so as to search for a saddle point, then results in a walk on the

potential energy surface that is effectively along the reaction

coordinate. At each step then the rate constant was calculated

at a range of temperatures, from which data the variational rate

constant at each temperature of interest is readily obtained. All

rate constants were calculated in the temperature range 1400-

2100 K and fitted to an Arrhenius form in order to obtain the

appropriate activation energies and frequency, viz. A factors,

in this temperature range.

Many of the reactions are expected to be in the falloff regime

at our average experimental pressure of 12 atm. For pressure

dependent reactions, RRKM calculations have been carried out

using the UNIMOL program.²²In this treatment, molecular

vibration frequencies have been considered to be harmonic. Two

of the external rotations are regarded as adiabatic with the third

rotational degree of freedom (corresponding to the K rotational

quantum number) considered to be active. Hindered rotors,

where identified, have been considered explicitly. Unless

otherwise specified, all falloff calculations have been made using

a constant value of ∆E_downof 500 cm^-1. Molecular parameters

for reactant and transition state have been obtained by ab initio

calculation.

Agreement between the G2 and CASPT2 relative energies

is generally found to be good for species with the same spin

state, although CASPT2 theory appears to have consistently

predicted the singlet/triplet separations in carbenes to be ∼30

kJ mol^-1larger than the corresponding G2 values. The results

of studies on smaller carbenes (:CH₂, :CF₂, and :CHF), where

the theoretical values could be compared with experiment, also

conform to this trend. The comparisons with experimental values

indicate, however, that G2 underestimates the relative stability

of triplet molecules by ∼10 kJ mol^-1while CASPT2 tends to

overestimate it by ∼20 kJ mol^-1. By contrast, multireference

configuration interaction (MRCI) calculations, which were also

carried out on these small molecules as a further check, using

the same basis set as in the CASPT2 studies, yielded results in

close agreement with experiment. Interestingly, however, MRCI

and CASPT2 gave very similar results in a calculation of the

lowest singlet f triplet vertical excitation energy of ethylene,

both underestimating it by ∼20 kJ mol^-1. By comparison, the

G2 value of 431 kJ mol^-1is an overestimate by 10 kJ mol^-1

.

A similar level of disagreement between CASPT2 and G2 to

that noted above was also found to occur when computing the

dissociation energy of a formally closed shell singlet state

molecule to doublet state fragments.

In the study of the decomposition reactions of acrylonitrile,

where in addition to G2, energies were also calculated at the

CASPT2 level of theory (at CASSCF geometries) with eight

active electrons in eight orbitals using the cc-pVDZ basis set,

the agreement between the G2 and CASPT2 relative energies

has been found to be generally poorer than for the succinonitrile.

For these reactions the CASPT2 results for intermediates,

transition states, and products relative to acrylonitrile are

systematically ∼15 kJ mol^-1higher than the G2 results. Using

the larger VTZ2P basis as well as the larger (10/10) active space,

the above discrepancy between CASPT2 and G2 has been

reduced to 11 kJ mol^-1. However, if we compare the CASPT2

and QCISD(T) energies obtained using the same VTZ2P basis,

the difference becomes just 6.6 kJ mol^-1. We may thus conclude

that much of the discrepancy between CASPT2/cc-pVDZ and

G2 noted above is the result of a combination of factors, such

as the obvious differences in basis sets as well as the choice of

active space in the CASPT2 calculations.

For reactions that might proceed via chemical activation (e.g.,

addition of H to acetonitrile and subsequent fission), the possible

stabilization of an intermediate adduct was investigated by

QRRK analysis.^23,24Barriers and high-pressure rate constants

were taken from our ab initio calculations. The QRRK method

uses a modified strong collision model for the energy transfer

parameter, â, and a single geometric mean frequency for the

adduct.²³

The ab initio computations were carried out using the

Gaussian-92,²⁵Gaussian-94,²⁶CADPAC 5.2,²⁷CADPAC 6.0,²⁸

SIRIUS,²⁹ABACUS,³⁰DALTON,³¹MOLCAS-2,³²MOLCAS-

4,³³and MOLPRO³⁴packages on DEC alpha and IBM RS2000

workstations.

Rate constants and Arrhenius rate parameters for the reactions

of interest were calculated using the appropriate computed

energies and partition functions according to Transition State

Theory.¹⁸The partition functions of the various species were

calculated from the ab initio data, using the standard formulas

for classical rigid rotors, translations, and with the exception

of molecules with low vibrational frequencies, quantum har-

monic oscillators.¹⁹(Thermal corrections to enthalpies, as well

as molecular entropies and specific heats were also obtained

from these partition functions.) Certain low-frequency modes,

when identified as torsional modes, were treated separately as

hindered rotors,¹⁹whereby their contribution to the partition

functions were calculated from the computed energy levels of

individual one-dimensional rotors. The latter were obtained by

solving the appropriate Schro¨dinger equation, using the general-

ized finite element method,²⁰for a one-dimensional rotor in the

presence of potential generated by a series of pointwise ab initio

energy calculations.

Results and Discussion

Product Distribution. The principal products of the shock

tube pyrolysis of acetonitrile are HCN, C₂H₂, CH₄, and H₂.

There are lesser amounts of HCCCN, H₂CCHCN, and C₄H₂.

At very low extents of decomposition, yields of CH₄and HCN

are approximately equal. Distributions of products were similar

to those reported by Lifshitz et al.²Small yields of cyanogen,

dicyanomethane, ethyl cyanide, cis- and trans-fumaronitrile,

dicyanoacetylene, C₂and C₃hydrocarbons, and benzene were

also detected in the present study. Profiles of the reactant and

major pyrolysis products are shown in Figure 1. We have been

unable to detect any succinonitrile in our shock tube pyrolyses

although we have observed succinonitrile in acetonitrile py-

rolysis in our jet-stirred reactor at lower temperatures (∼1300

K and 1 s residence time). The lack of detection in the shock

tube experiments results either because succinonitrile has low

volatility or because it was formed but decomposed very rapidly.

Pyrolysis Kinetics of Acetonitrile

J. Phys. Chem. A, Vol. 103, No. 8, 1999 1057

Figure 1. Temperature dependence of yield (%) of designated species in the pyrolysis of acetonitrile. Symbols ) data points, lines ) model

prediction.

However, because elemental mass balances of reactant and

product were generally achieved within 10% and there was no

systematic mass loss with increase in pyrolysis temperature, we

consider that a significant yield of succinonitrile could not have

been formed and, additionally, we have been able to detect the

rather involatile products cis- and trans-fumaronitrile, albeit in

low yield.

In the jet-stirred reactor at a temperature of 1300 K and

residence time of 1 s, acetonitrile underwent significant

decomposition. Major products were similar to those observed

in the shock tube experiments. Additionally, extracts from the

cold traps revealed the presence of trace amounts of dicyano-

methane, the fumaronitriles, dicyanoacetylene, and as mentioned

above, traces of succinonitrile, identified by GC/MS.

Succinonitrile vapor (in the presence of acetonitrile vapor)

started to decompose at 700 K (1 s residence time). At

Figure 2. Arrhenius plot for the rate constant for disappearance of

acetonitrile.

the rate constant k₁at higher temperatures. The participation of

temperatures below ∼900 K the only products detected were

secondary reactions would also lead to marked non-Arrhenius

behavior. Approximate Arrhenius parameters for the disappear-

HCN and acrylonitrile. Low Arrhenius parameters typical of

surface-enhanced reaction were measured for this reaction in

the CSTR. Above 900 K other products including ethyl cyanide

and the fumaronitriles were also detected.

ance of CH₃CN were A ) 10^14.6(0.3s^-1and E_a) 397 kJ mol^-1

,

estimated from the data in Figure 2 by considering only values

of ln k_disat values of 1/T > 5.7 × 10^-4K^-1

.

Kinetics of Disappearance of CH₃CN. Thermochemical

considerations (see below) lead to the conclusion that the only

feasible initiation reaction is CH fission to produce the cyano-

methyl radical:

Although the apparent rate constant should be less influenced

by secondary reactions over this range of temperatures, the

participation of these reactions is still likely to have a significant

influence of the rate of disappearance of acetonitrile. At best

we can only obtain from the experimental Arrhenius plot an

initial estimate of the Arrhenius parameters for k₁, the rate

constant for the decomposition reaction 1. When proper allow-

ance is made for secondary reactions in the kinetic modeling

to be described below, a more accurate rate constant for k₁will

be derived. It is expected that this value, for a pressure of ∼12

CH₃CN f H + CH₂CN

(1)

An Arrhenius plot for the rate constant for disappearance of

acetonitrile (behind the reflected shock front) is shown in Figure

2. This plot shows considerable curvature at values of 1/T <

5.7 × 10^-4K^-1. Part of this curvature results from falloff in

1058 J. Phys. Chem. A, Vol. 103, No. 8, 1999

Sendt et al.

kJ mol^-1as the ∆_fH ₂⁰₉₈of cyanomethyl. These values are

significantly higher than the previously reported value of 245

( 9 kJ mol^-1, obtained from pyrolysis kinetics of alkaneni-

triles.³⁷In recent times there have been claims³⁸that the

derivation of heats of formation from pyrolysis kinetics could

lead to systematic errors of the order of 8 kJ mol^-1. Indeed, a

higher value of 250 ( 8 kJ mol^-1has been obtained by Moran

et al.³⁹on the basis of the observed electron and proton affinities

of cyanomethyl and the ∆_fH ⁰₂₉₈of acetonitrile. Substituting our

computed value of 79.1 kJ mol^-1for the latter would yield 255

( 8 kJ mol^-1as ∆_fH ₂⁰₉₈(CH₂CN). However, on the basis of the

ab initio results our conservative estimate for ∆_fH ⁰₂₉₈(CH₂CN)

is 263 ( 9 kJ mol^-1, which overlaps the upper bounds of the

experimental values. The kinetic modeling described in a later

section leads to computed species profiles of reactant and all

major products that agree significantly better with experiment

if the higher value of ∆_fH ⁰₂₉₈(CH₂CN), viz. 263 kJ mol^-1, is

used.

The heat of reaction at 298 K for the first step is thus 405 kJ

mol^-1when computed on the basis of the theoretical heats of

formation of acetonitrile and cyanomethyl (and the experimental

heat of formation of atomic hydrogen). Direct calculation of

the energetics of the fission reaction yields a heat of reaction

of 410 kJ mol^-1, indicating that the various theoretical ap-

proaches are consistent and any error cancelations in the

isodesmic or isogyric schemes are small.

atm, would be in the falloff regime. However, before considering

the extrapolation of our rate data to the high-pressure limit, it

is instructive to consider the results of the ab initio calculations

on the various isomers of CH₂CN, and especially the enthalpy

of formation of cyanomethyl, which is the most stable C₂H₂N

isomer.

Computational Results

In the course of this work a total of ∼70 molecular species,

corresponding to reactants, products, intermediates, and transi-

tion states, were studied using ab initio methods. Apart from a

few small molecules such as hydrogen cyanide, methane,

acetylene, etc., no existing literature data of sufficient accuracy

could be found for these species, hence the need for the current

study on at least 65 of them. The computed MP2 and/or

CASSCF geometries of these 65 species are given in Table 1S

in the Supporting Information, while the Supporting Tables 2S-

9S contain the corresponding absolute and relative QCISD(T),

G2, and CASPT2 energies, respectively, as well as the computed

zero point energies. In the paper itself the reaction energies and

critical energies will be given in the form of molecular potential

energy diagrams in the appropriate sections. These diagrams

also show schematic structures of the various molecular species

along with a number (1-65) that had been assigned to each

distinct molecule. Unless indicated otherwise, the relative

energies quoted in these diagrams and used in the computation

of reaction energies and Arrhenius are from G2 calculations of

the electronic energies at MP2 or CASSCF geometries, as

indicated in Table 1S. All vibrational frequencies were scaled

by 0.8929, when computing zero point energies and other

thermodynamic quantities.

The rate constants for this fission reaction at a range of

temperatures were estimated using variational transition state

theory with a PES calculated at the CASSCF(6/6)/DZP level

of theory. The energy was scaled so as to make the CASSCF

dissociation energy consistent with the value of 405 kJ mol^-1

for the heat of reaction at 298 K. In the temperature range 1400-

2100K, the transition state (structure 3 in Table 1S) was found

to have a CH bond length of approximately 2.5 Å, which

contracts slightly as the temperature is increased. Such a

contraction has been also noted in the dissociation of methane.²¹

Using the calculated rate constants, which correspond to

transition states with different bond lengths at different tem-

peratures, yielded an A factor of 1.2 × 10¹⁶s^-1and an activation

energy of 411 kJ mol^-1when an Arrhenius fit of the rate

constants was carried out. The Arrhenius parameters used in

Thermochemistry of CH₃CN. The enthalpy of formation

of acetonitrile (1) was calculated from the following (isogyric

and isodesmic) pathways:

CH₃CN + H₂f CH₄+ HCN

CH₃CN + CH₄f C₂H₆+ HCN

From the computed G2 enthalpies of the reactants and products

(at 298 K), the heats of the above reactions, ∆_rH ⁰₂₉₈, were

obtained. Using the latter, in conjunction with the experimental

values for the ∆_fH ₂⁰₉₈of H₂, CH₄, HCN, and C₂H₆(taken from

the CHEMKIN thermodynamic database³⁵), the heat of forma-

the kinetic modeling are 8.0 × 10¹⁴s^-1and 397 kJ mol^-1

,

specific to the 12 atm pressure of the experiments. (The falloff

calculation of the rate constant is discussed in the section on

Kinetic Modeling.)

tion, ∆_fH ⁰₂₉₈, of acetonitrile was computed to be 79.1 kJ mol^-1

.

This is slightly higher than the value of 73.9 ( 0.8 kJ mol^-1

that had been previously derived on the basis of experimental

kinetic data.³⁶

C₂H₂N Potential Energy Surface. The C₂H₂N PES was

studied in order to ascertain whether there are any low lying

isomers of the cyanomethyl radical that may need to be

considered in the kinetic modeling, especially since the mass

spectrometric and thermochemical studies of Holmes and

Mayer⁴⁰suggested the possible presence of a low-energy cyclic

isomer. In our study five isomers were located as local minima

on the PES. Their structures and relative energies are sum-

marized in Table 1. Note, however, that the valence structures

of these isomers as drawn (as well as those of many other

species to be discussed later) are only meant to provide a

qualitative guide to the bonding. Thus, e.g., the Lewis structure

of the isocyanomethyl radical CH₂NC implies the presence of

formal charges on the cyanide C and N atoms, while the

molecular orbital (MO) calculations suggest a CN triple bond

between atoms that are effectively neutral.

On thermochemical grounds, the first step in the decomposi-

tion of acetonitrile is the cleavage of the CH bond, resulting in

the formation of the cyanomethyl radical and atomic hydrogen.

(Other possible initiation channels, such as fission to CH₃+

CN, elimination of HCN to form CH₂or elimination of H₂to

form HCCN, can be discounted because of their high heats of

reaction. Furthermore, the elimination reactions would also be

expected to have barriers in addition to their unfavorably high

reaction enthalpies.) The enthalpy of formation of the cyanom-

ethyl radical (2) was calculated by utilizing the G2 value of

∆_fH ⁰₂₉₈of the isodesmic reaction:

CH₃CN + CH₃f CH₄+ CH₂CN

The resulting ∆_fH ₂⁰₉₈of cyanomethyl is then 267 kJ mol^-1

,

As expected, cyanomethyl is the most stable isomer, viz.,

global minimum on the PES. In comparison with acetonitrile,

cyanomethyl was found to have a shorter CN bond as well as

when the G2 value is used for ∆_fH ⁰₂₉₈of acetonitrile. If the

above experimental value is used for the latter, we obtain 261

Pyrolysis Kinetics of Acetonitrile

J. Phys. Chem. A, Vol. 103, No. 8, 1999 1059

TABLE 1: Isomers of C₂H₂N: Structures, Energies

(Relative to Cyanomethyl), Symmetries, and Spectroscopic

States

Figure 3. Potential energy surface for reaction between atomic

hydrogen and acetonitrile. (Energies at 0 K given in kJ mol^-1.)

The PES for these reactions is shown in Figure 3. The attack

by a hydrogen atom at the nitrile carbon results in a fairly stable

intermediate (9). The transition state (8) associated with this

reaction has a critical energy of 33 kJ mol^-1at 0 K. It is

characterized by a long HC bond of length 1.74 Å and a CCN

angle of 154.8°. The CH₃C(H)N intermediate (9) has a CN bond

length of 1.226 Å, which is consistent with a CN double bond,

as are the bond angles of ∼120° around the nitrile C atom. The

unpaired electron is largely localized on the N atom. The

subsequent CC bond cleavage via a transition state (10), where

the CC distance is 2.147 Å, results in a methyl radical and

hydrogen cyanide as products. This second transition state was

found to lie 26 kJ mol^-1above the reactants at 0 K and the

a considerably shorter CC bond (1.412 Å), indicating that a

greater degree of π delocalization that includes the unpaired

electron is present in the radical. Isocyanomethyl and the ketene

iminyl radical (structures 4 and 5) are found to be the lowest

energy isomers of cyanomethyl. As one may expect, the energy

difference of 102 kJ mol^-1between CH₂NC and CH₂CN is

comparable with the corresponding G2 value of 98 kJ mol^-1

that was computed for the parent molecules CH₃NC and CH₃-

CN. The relatively high stability of the ketene iminyl radical

can be similarly attributed to the high degree of π bonding,

viz., delocalization. The length of the CC triple bond was found

to be 1.227 Å, slightly longer than the accepted standard value

for a triple bond, viz. 1.20 Å. However, the CN bond length

was found to be 1.272 Å, which is comparable with the

experimental and computed CdN distances in CH₂NH⁴¹that

range from 1.27 to 1.28 Å, while being only slightly longer

than the CN distance of 1.228 Å that had been computed for

ketene imine.⁴²This suggests that there is substantial π

delocalization across the CCN moiety with the participation of

the nonbonding electrons of nitrogen. The two cyclic structures

(6, 7) are, however, considerably less stable than the open chain

isomers, owing to the presence of σ strain as well as the weaker

π bonds. The situation is thus qualitatively different from that

of the cationic species, where a cyclic isomer (with two π

electrons) is formally aromatic and has been found to be the

most stable, both experimentally⁴⁰and computationally.⁴³The

above energetics suggest that isomerization reactions of the

cyanomethyl radical are unlikely to be important reaction

pathways.

overall reaction was found to be exothermic by 18 kJ mol^-1

.

The calculated A factor of 5.3 × 10¹⁴cm³mol^-1s^-1and

activation energy of 51 kJ mol^-1(in the temperature range

1400-2100 K) are the parameters that were computed for the

first (slow) step. The ab initio potential energy surface for

reaction 2a has been used in conjunction with a QRRK analysis

to show that under our experimental conditions the intermediate

adduct is not stabilized. Thus, in the development of the detailed

kinetic mechanism given in Table 2 (discussed in the section

on Kinetic Modeling), reaction 2 represents the overall addition

and fission process and the rate constant given in Table 2 is

the result of the QRRK analysis using the computed ab initio

data. The reactions where a hydrogen attack occurs at the

nitrogen atom were found to result in intermediates (trans and

cis), which are approximately 33 and 55 kJ mol^-1higher in

energy (at 0 K), respectively, than the intermediate (9) in the

carbon attack pathway discussed above.

Other reactions important in the mechanism involve abstrac-

tion by H and CH₃radicals, and rate constants for these reactions

shown in Table 2 have been also evaluated in the present study.

The cyanomethyl radical plays an important role in the

mechanism, along with CH₃, CN, and H, as evidenced by the

observed termination products that include ethyl cyanide,

dicyanomethane, and cyanogen.

Recombination Reaction of Cyanomethyl Radicals. In

agreement with previous work by Lifshitz et al.²we have

observed in our shock tube studies significant yields of

cyanoacetylene and acrylonitrile without observing any succi-

nonitrile, viz., the dimer of cyanomethyl, NCCH₂sCH₂CN.

Lifshitz et al. suggested that two CH₂CN radicals recombine to

form a C₄H₄N₂intermediate, which then decomposes to

acrylonitrile and HCN, which are among the observed products.

Thermochemical calculations, however, suggest that if this

After the initial submission of this manuscript a paper

appeared by Mayer et al.⁴⁴that includes an ab initio study of

the thermochemistry of CH₃CN and three isomers of CH₂CN

(structures 2, 4, and 7 of Table 1). Their computed heats of

formation of these molecules are in excellent agreement with

those reported in this work.

Decomposition Mechanism of CH₃CN. We expect that HCN

principally arises from addition of H atoms to the reactant and

subsequent fissioning of the adduct into HCN and CH₃

according to the complex process

H + CH₃CN f CH₃C(H)N f CH₃+ HCN

(2a)

1060 J. Phys. Chem. A, Vol. 103, No. 8, 1999

Sendt et al.

TABLE 2: Kinetic Model for Acetonitrile Pyrolysis

forward rate constant

reverse rate constant

reactions^a

log A^b

n

E_a/kJ mol^-1

log A

n

E_a/kJ mol^-1

ref

1

2

3

4

5

6

7

8

9

CH₃CN S CH₂CN + H

14.90

14.00

13.18

12.70

13.70

13.00

13.36

27.92

14.46

14.34

14.77

15.40

13.48

14.30

12.30

13.70

12.54

14.34

13.75

13.49

13.30

13.00

15.00

12.00

13.60

12.70

13.70

12.24

12.42

15.17

14.30

13.30

17.41

13.70

13.00

13.56

2.73

-0.26

2.81

13.70

5.69

68.1

14.49

16.78

12.30

14.00

5.61

13.60

13.48

12.30

16.00

16.62

18.00

16.96

13.80

13.00

14.00

50.00

15.00

15.48

13.80

14.00

12.00

13.00

15.00

0.00

-3.95

0.00

0.19

0.13

0.00

3.50

4.00

3.00

0.00

2.44

-13.22

0.00

-1.00

0.00

2.39

0.00

-1.00

-0.60

0.00

-10.5

0.00

397.5

40.5

41.8

29.3

8.4

41.8

0.0

184.9

366.1

260.0

334.7

400.4

20.9

41.8

20.9

8.4

20.9

230.1

198.7

153.1

0.0

12.96

12.65

11.85

12.79

12.80

12.95

16.71

0.00

-10.5

62.5

71.0

59.5

110.8

88.9

308.8

0.0

268.6

187.1

310.2

-0.2

57.5

93.6

58.5

118.2

28.3

22.8

33.9

3.8

PW^c,d

H + CH₃CN S CH₃+ HCN

H + CH₃CN S CH₂CN + H₂

CH₃+ CH₃CN S CH₂CN + CH₄

CN + CH₃CN S HCN + CH₂CN

CN + CH₃CN S CH₃+ C₂N₂

2CH₂CN S SUCC

PW/QRRK^c

PW

est^c

est

PW

PW/QRRK

PW/abin^c

PW/abin

est

PW/abin

est

2CH₂CN w HCN + H₂CCHCN

SUCC S H₂CCHCN + HCN^e

SUCC S ISUCC

12.84

14.39

13.10

13.49

12.19

13.04

12.43

12.84

12.58

13.58

14.86

13.14

16.10

15.85

12.61

12.80

15.39

10.98

12.53

12.56

11.37

13.84

13.39

12.57

15.92

15.11

12.02

12.81

16.71

14.88

15.15

15.24

12.92

15.07

17.28

13.20

15.15

14.31

12.16

15.05

14.41

14.56

11.74

16.35

15.33

14.95

15.38

16.56

12.19

13.59

11.35

14.51

14.06

13.69

14.40

13.11

13.34

13.64

12.73

12.96

13.63

13.87

13.63

13.77

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

ISUCC S H₂CCHCN + HCN

SUCC S NCCHCH₂CN + H

H + SUCC S H₂+ NCCHCH₂CN

H + SUCC S CH₂CH₂CN + HCN

CH₃+ SUCC S CH₄+ NCCHCH₂CN

CN + SUCC S HCN + NCCHCH₂CN

CH₂CN + SUCC S CH₃CN + NCCHCH₂CN

NCCHCH₂CN S H₂CCHCN + CN

NCCHCH₂CN S H + C₄H₂N₂

CH₂CH₂CN S H₂CCHCN + H

CH₃+ CH₂CN S C₂H₅CN

CN + CH₂CN S NCCH₂CN

CH₃CN S CH₃+ CN

HCCHCN S HCCCN + H

H + HCCHCN S H₂CCHCN

H + H₂CCHCN S HCN + C₂H₃

H + H₂CCHCN S HCCHCN + H₂

H₂CCHCN S HCCCN + H₂

H₂CCHCN S C₂H₂+ HCN

C₄H₄S 2C₂H₂

339.7

412.5

1.1

0.0

489.5

196.6

0.0

est

7

7.6

440.0

13.4

30.6

128.7

155.1

194.8

192.2

10.2

160.0

-60.2

6.4

156.0

360.4

99.6

120.7

75.9

7.5

33.5

318.0

345.2

364.0

166.0

331.8

404.0

33.5

11.3

0.0

21.8

34.7

32.2

242.7

8.9

715.9

32.9

0.0

83.7

0.0

3.6

0.0

PW

7

PW^d

59

C₄H₄S C₄H₂+ H₂

59

53

C₂H₅(+M) S C₂H₄+ H (+M)

C₂H₄+ M S C₂H₂+ H₂+ M

C₂H₄+ M S C₂H₃+ H + M

C₂H₄+ H S C₂H₃+ H₂

60

H + C₂H₂(+M) S C₂H₃(+M)

2CH₃(+M) S C₂H₆(+M)

61,57^g

53

C₂H₆+ H S H₂+ C₂H₅

54

53

est

54

62

est

54

est

54

52

54

52

est

est/QRRK

est

C₂H₆+ CH₃S CH₄+ C₂H₅

CH₃+ H₂S CH₄+ H

HCCHCN S C₂H₂+ CN

CN + H₂S HCN + H

116.8

-7.4

7.9

C₂N₂+ M S 2CN + M

H + C₂N₂S HCN + CN

CH₃+ H (+M) S CH₄(+M)

2CH₂CN S CH₃CN + NCCH(3)

C₂H₅+ H S 2CH₃

424.0

68.4

48.7

120.5

292.5

61.5

-63.2

46.5

-72.4

423.0

428.7

530.5

35.5

193.6

-1.8

57.7

86.4

1.0

H₂+ C₂H S C₂H₂+ H

C₂H₃+ H S C₂H₂+ H₂

C₂H + C₂H₂S C₄H₂+ H

0.0

2C₂H₂S C₄H₃+ H

192.0

249.8

447.7

0.0

C₄H₃+ M S C₄H₂+ H + M

C₂H₂+ M S C₂H + H + M

2H + M S H₂+ M

2H + H₂S 2H₂

CN + C₂H₃S H₂CCHCN

CH₂CH₂CN S C₂H₄+ CN

H₂CCHCN + H S CH₃CHCN

CH₃CHCN S CH₃+ NCCH(3)

H₂CCHCN + C₂H₃S HCCHCN + C₂H₄

2HCCHCN S H₂CCHCN + HCCCN

C₂H₅CN S H + CH₂CH₂CN

C₂H₅CN S H + CH₃CHCN

C₂H₅CN + H S H₂+ CH₂CH₂CN

C₂H₅CN + H S H₂+ CH₃CHCN

C₂H₅CN + H S HCN + C₂H₅

C₂H₅CN + CH₃S CH₄+ CH₂CH₂CN

C₂H₅CN + CH₃S CH₄+ CH₃CHCN

CH₂CN + C₂H₅CN S CH₃CN + CH₂CH₂CN

CH₂CN + C₂H₅CN S CH₃CN + CH₃CHCN

CH₂CN + C₂H₆S CH₃CN + C₂H₅

CH₂CN + M S H₂+ C₂N + M

251.0

19.7

376.6

33.5

284.5

410.0

384.9

29.3

27.6

41.8

33.5

96.2

117.2

41.8

418.4

0.4

57.4

80.2

83.5

62.6

87.1

95.1

140.6

40.8

86.8

est

Pyrolysis Kinetics of Acetonitrile

J. Phys. Chem. A, Vol. 103, No. 8, 1999 1061

TABLE 2 (Continued)

forward rate constant

reverse rate constant

reactions^a

log A^b

n

E_a/kJ mol^-1

log A

n

E_a/kJ mol^-1

ref

est^d

est

73

74

75

76

77

78

79

80

81

82

83

84

85

86

87

CH₂CN S H + NCCH(3)

2NCCH(3) S C₄H₂N₂

2C₂N S C₄N₂

CN + C₂H₂S HCCCN + H

C₄H₂N₂+ H S HCCHCN + HCN

SUCC S c-C₄H₄N₂

c-C₄H₄N₂w H₂CCHCN + HCN

2CH₂CN w HCN + H₂CCHCN^f

CN + C₂H₂S HCN + C₂H

C₄H₂N₂S HCCCN + HCN

CN + HCCHCN S C₄H₂N₂

HCN + M S H + CN + M

H + CH₂CN S NCCH(3) + H₂

CN + CH₂CN S NCCH(3) + HCN

CH₃+ C₂H₂S C₂H + CH₄

23.74

13.00

12.70

11.70

13.00

15.30

15.48

13.30

13.70

13.48

29.02

13.70

11.26

-3.09

0.00

-3.30

0.00

434.5

0.0

8.4

351.5

246.9

142.3

20.9

309.6

0.0

529.7

33.5

8.4

12.45

15.43

16.72

13.28

10.39

15.00

0.00

-32.7

589.9

653.8

46.1

36.3

181.2

0.0

11.3

148.5

482.5

-27.6

47.4

est

PW/abin

est

0.00

13.46

11.89

17.14

15.86

14.38

14.81

12.01

PW^d

est

52

est

95.5

-10.8

71.1

62

^aS denotes reversible reaction, w denotes unidirectional reaction. (M) denotes a reaction in the falloff region; falloff parameters are given in

the reference. ^bUnits for A are cm³mol^-1s^-1or s^-1as appropriate. ^cPW denotes a rate constant determined in the present work. QRRK denotes

that the value was obtained from QRRK analysis. est denotes a rate constant estimated in the present work. abin denotes a rate constant evaluated

by ab initio calculation. ^dValue specific to the pressure of 12 atm. ^eProceeds via 1,3 H-transfer. ^fProceeds on triplet succinonitrile potential energy

surface. ^gRRKM extrapolation by ref 57. SUCC: succinonitrile NCCH₂CH₂CN. C₄H₂N₂: cis- and trans-fumaronitrile. ISUCC: succinonitrile

isocyanide NCCH₂CH₂NC. c-C₄H₄N₂:

intermediate is succinonitrile, it should be a stable intermediate

whose energy is some 300 kJ mol^-1below that of two CH₂CN

radicals. According to our calculations, cyanomethyl radicals

can recombine by forming either a new CC bond giving

succinonitrile (structure 11) or a CN bonded isomer NCsCH₂s

ΝdCdCH₂(structure 12 in Table 1S). Preliminary calculations,

however, indicated that succinonitrile was the more stable isomer

by approximately 155 kJ mol^-1. Given this, as well as the fact

that the observed products contain three to four carbon chains,

as in succinonitrile, only reactions involving the latter molecule

were further studied. Reaction 7, corresponding to the recom-

bination of two cyanomethyl radicals to form succinonitrile, was

calculated to be 317 kJ mol^-1exothermic at 0 K. The lowest

triplet state of succinonitrile (structure 13 in Table 1S) was also

investigated so as to determine whether the decomposition

reactions of succinonitrile could occur on a triplet PES. As triplet

succinonitrile was found to have an energy (at 0 K) that is 135

kJ mol^-1above that of two cyanomethyl radicals, the corre-

sponding recombination reaction is unlikely to be important,

even if its critical energy (not calculated in this work) is not

significantly higher than the endothermicity. However, reaction

80, the direct reaction between two CH₂CN radicals on a triplet

surface to form H₂CCHCN + HCN, was included in the kinetic

model with an activation energy of 142 kJ mol^-1, which is based

on an approximate barrier height of 135 kJ mol^-1. The

sensitivity analysis that was carried out (see below) indicated

that this direct mechanism had no sensitivity over the entire

temperature range studied. The reaction involving the decom-

position of succinonitrile to form acrylonitrile (structure 19) and

hydrogen cyanide was studied by exploring several different

pathways.

detail by means of ab initio quantum chemical calculations. The

pathways studied fall into three broad classes: (a) unimolecular

rearrangement of succinonitrile followed by decomposition, (b)

CH bond fission followed by decomposition of the resulting

NC-CH₂-CH-CN radical, and (c) bimolecular abstraction of

CN by a H atom, followed by decomposition of the resulting

cyanoethyl radical.

1,3-Hydrogen Transfer Reaction. In this reaction scheme, the

results of which are summarized in Figure 4a, a 1,3 type

H-transfer takes place via a four-center transition state (14) to

yield a biradical intermediate (15, 16) followed by CC bond

cleavage to yield acrylonitrile and HCN. The first transition state

(14) is analogous to that calculated for the 1,3-H transfer in

propene,⁴⁵and indeed, the key geometric parameters of the two

species are comparable. The computed critical energy of the

H-transfer reaction is 352 kJ mol^-1at 0 K. The biradical species

that results has a singlet ground state (15) although the

corresponding triplet state (16) is quite close both structurally

and energetically to the singlet, provided the energies are

calculated at the same level of theory. Using CASPT2(10/10)/

VTZ2P the singlet-triplet splitting was calculated to be -18

kJ mol^-1. As the triplet intermediate (16) was found to be 303

kJ mol^-1above succinonitrile at 0 K, the reaction that yields

the singlet intermediate (15) is 285 kJ mol^-1endothermic at 0

K, when using the above estimate of singlet-triplet splitting.

If the energy of this reaction is calculated directly from the

CASPT2(10/10)/VTZ2P energies of both succinonitrile and the

intermediate (15), the first step is predicted to be 256 kJ mol^-1

endothermic. This difference arises largely from the discrepancy

between the G2 and CASPT2 energies for triplet systems, as

discussed in the section on Theoretical and Computational

Methods. The A factor and activation energies for the first step

Decomposition Reactions of Succinonitrile. The subsequent

reactions of succinonitrile that are expected to yield the observed

products acrylonitrile or cyanoacetylene must proceed via

sufficiently low energy pathways in order to account for the

significant quantities of these products as well as the fact that

no significant amount of residual succinonitrile has been

observed. Six different pathways were considered as chemically

reasonable, and these were subsequently investigated in some

were calculated to be 2.9 × 10¹⁴s^-1and 366 kJ mol^-1

,

respectively, in the temperature range 1400-2100 K.

The singlet biradical (15) can undergo CC bond cleavage to

produce HCN and ground-state acrylonitrile by passing through

a transition state (17), whose geometry was optimized at the

CASSCF(6/6) level. The CC distance of the breaking bond in

the transition state was found to be ∼1.8 Å. Given that the

1062 J. Phys. Chem. A, Vol. 103, No. 8, 1999

Sendt et al.

Figure 4. (a) Potential energy surface for decomposition of succinonitrile via 1,3-H migration. (Energies at 0 K given in kJ mol^-1.) (b) Potential

energy surface for decomposition of succinonitrile via 1,2-H migration. (Energies at 0 K given in kJ mol^-1.) (c) Potential energy surface for

decomposition of succinonitrile via an isocyanide intermediate. (Energies at 0 K given in kJ mol^-1.) (d) Potential energy surface for decomposition

of succinonitrile via concerted elimination of HCN. (Energies at 0 K given in kJ mol^-1.) (e) Potential energy surface for decomposition of succinonitrile

via CH bond fission. (Energies at 0 K given in kJ mol^-1.) (f) Potential energy surface for reaction between atomic hydrogen and succinonitrile.

(Energies at 0 K given in kJ mol^-1.)

3

CASSCF wave function was found to be dominated by a single

closed shell type configuration, the energy was initially com-

puted using the G2 method, which predicts that the transition

state (17) is 242 kJ mol^-1above succinonitrile, i.e., 43-14 kJ

mol^-1below the intermediate (15), depending on the method

used to estimate the energy of the singlet intermediate, sug-

gesting that the bond breaking process may be barrierless. Of

course, at the CASSCF(6/6) level (that had been used in the

geometry optimizations) there is a barrier, albeit a very small

one, with a height of 14 kJ mol^-1. Consequently, the barrier to

CC bond cleavage on the singlet surface is unlikely to be

significant. On the triplet surface, the transition state (18) that

corresponds to CC bond cleavage is significantly higher in

energy than the analogous singlet state and results in the

formation of triplet acrylonitrile (20) and singlet HCN.

but of C_ssymmetry, and thus the electronic state is A′′. The

CC bond adjacent to the CN was found to be slightly shorter

than in the singlet ground state, suggesting a comparable degree

of conjugation. Interestingly, the CC bond linking the HCCN

fragment to the methylene moiety was found to be only slightly

longer at 1.46-1.48 Å. This is comparable with the CC distance

of 1.456 Å in triplet ethylene (computed using the MP2/6-31G-

(d) method), where the dihedral angle between the methylenes

is 90°, but it is significantly shorter than the “normal” C-C

bond length of ∼1.53 Å, e.g., in ethane.⁴¹The explanation for

these short CC distances in these biradical species is in terms

of hyperconjugation, viz., the stabilizing interaction of CH σ

MO’s with the half occupied p_πorbitals on the adjacent carbon

atom(s). Singlet acrylonitrile and HCN were computed to be

105 kJ mol^-1above succinonitrile at 0 K. Although energetically

triplet acrylonitrile (20) and HCN were found to lie 262 kJ mol^-1

higher, this corresponds to an energy that is only 64 kJ mol^-1

above the triplet intermediate (16). Intersystem crossing to the

singlet surface is possible, either by spin-orbit coupling or via

Acrylonitrile in its singlet state is planar, and on the basis of

the computed bond lengths, such as the relatively short CC

single bonds of 1.44-1.45 Å, it appears that there is a small

degree of conjugation. The geometry in the triplet state is twisted

Pyrolysis Kinetics of Acetonitrile

J. Phys. Chem. A, Vol. 103, No. 8, 1999 1063

a collisional mechanism; hence the reaction on the triplet surface

could, in principle, be included in the kinetic model. However,

given that on energetic considerations the singlet pathway

appears to be substantially more favorable, the alternative

pathway on the triplet PES is not part of our kinetic model.

The kinetic parameters for reaction 9, viz., an A factor of 2.9

× 10¹⁴s^-1and an activation energy of 366 kJ mol^-1, were thus

derived on the basis of the 1,3-H transfer step (with a critical

energy of 352 kJ mol^-1) being rate determining.

than the transition state leading to its formation, although at

the CASSCF/DZP(6/6 and 10/10) levels of theory (which were

used to optimize the geometries) the transition state is clearly

higher in energy by 38-45 kJ mol^-1. This apparent contradiction

is due to different levels of theory being used for the geometry

optimization (CASSCF) and energy calculation (G2) for reac-

tions with small barriers. The G2 calculations predict the singlet

and triplet carbenes (structures 26 and 27) as being effectively

isoenergetic. Given the observation in the Theoretical and

Computational Methods section that G2 tends to underestimate

the singlet-triplet energy difference by ∼10 kJ mol^-1, a

reasonable estimate of the latter quantity in these cyano carbenes

1,2-Hydrogen Transfer Reaction. In this reaction scheme,

summarized in Figure 4b, a hydrogen atom migrates to the

adjacent carbon atom of the cyano group of succinonitrile,

followed by CC bond cleavage of the resulting cyclic or biradical

intermediates (structures 22 and 23 respectively), resulting in

HCN and 2-cyanoethylidene. A subsequent 1,2 hydrogen

migration in the carbene yields acrylonitrile. The first transition

state (structure 21) lies 337 kJ mol^-1above succinonitrile at 0

K and its geometry is similar to those observed in studies of

1,2-H migration in propene.⁴⁵Although the expected product

of the initial 1,2-H transfer is a biradical, such a structure was

found only for the triplet state (structure 23). A singlet biradical

structure could be obtained if it was constrained to have C_s

symmetry, but at the CASSCF level such a structure is a saddle

point with an imaginary frequency of ∼200 cm^-1corresponding

to an a′′ mode, indicating that the true equilibrium structure

has lower symmetry. On removing the symmetry constraint the

structure simply collapsed to a cyclic structure (structure 22)

with a formally closed shell electronic configuration. While the

bond lengths and bond angles of this cyclic intermediate were

found to be fairly insensitive to the method of optimization,

the difference between the MP2 and CASSCF optimized

dihedral angles is more substantial and this difference is possibly

the main reason for the difference between the G2 and CASPT2

relative energies. The A factor and activation energies for this

first step (reaction 78) were found to be 2.0 × 10¹⁵s^-1and

353 kJ mol^-1, respectively, in the temperature range 1400-

2100 K. The reverse reactions, viz., singlet intermediate to

succinonitrile and triplet intermediate to succinonitrile, were

found to have Arrhenius parameters of 1.6 × 10¹⁵s^-1and 181

kJ mol^-1(reaction 78 in Table 2) and 4.7 × 10¹³s^-1and 64 kJ

mol^-1, respectively. Only the reaction pathway on the singlet

PES was included in the modeling.

is ∼11 kJ mol^-1

.

To form singlet acrylonitrile, a 1,2-H migration in the singlet

carbene is needed. Two transition states were found for such a

migration. Structure 28 results when a hydrogen atom migrates

so as to be trans to the cyano group. The structure is such that

the migrating hydrogen is above the rest of the molecule, which

is approximately planar. The barrier to this migration is just 1

kJ mol^-1at 0 K. The forward and reverse A factors and

activation energies for this process in the temperature range

1400-2100 K were computed to be 3.6 × 10¹²s^-1, 8.5 × 10¹³

s^-1, 3.5 kJ mol^-1, and 340 kJ mol^-1, respectively. Structure 29

is the transition state that results when a hydrogen atom migrates

so as to be cis to the cyano group. This structure is qualitatively

similar to 28, apart from the obvious cis-trans difference, but

it is associated with a slightly higher barrier, the difference being

4 kJ mol^-1. The forward and reverse A factors and activation

energies in the temperature range 1400-2100 K are 3.3 × 10¹²

s^-1, 7.8 × 10¹³s^-1, 7.3 kJ mol^-1, and 344 kJ mol^-1

,

respectively. Analogous calculations on the ethylidene mol-

ecule⁴⁶have resulted in a critical energy of 3.3 kJ mol^-1for

the singlet pathway at 0 K. These results suggest that the

assumption that the singlet reaction goes to completion is valid.

Therefore, in the kinetic modeling, the singlet carbene was not

considered explicitly; instead, its decomposition to acrylonitrile

was considered to be sufficiently fast and essentially irreversible

so that reaction 79 was taken to correspond to the overall

unidirectional formation of acrylonitrile and HCN.

The cyclic intermediate (22) discussed above and the biradical

intermediate (15) of the 1,3-H migration pathway can, however,

interconvert by a simple hydrogen migration that occurs via a

three centered transition state (structure 30 in Table 1S).

Energetically, this transition state was found to be 428 kJ mol^-1

above succinonitrile and the A factor and activation energy for

the 22 f 15 conversion were computed to be 5.4 × 10¹⁴s^-1

and 272 kJ mol^-1, respectively, while for the reverse reaction

they are 1.2 × 10¹³s^-1and 167 kJ mol^-1in the temperature

range 1400-2100 K. This aspect of the potential energy surface

was not considered in the kinetic modeling.

The second step in this reaction, involving CC bond cleavage

to form HCN and a carbene was found to be considerably higher

in energy than the corresponding step in the 1,3-hydrogen

transfer pathway. The singlet transition state (structure 24) was

found to be 231 kJ mol^-1at 0 K above the singlet intermediate.

Although the transition state resembles the CC bond cleavage

of a singlet biradical system, geometry relaxation along the

reverse pathway did in fact yield the closed shell ring structure.

The triplet transition state (structure 25) was found to be 172

kJ mol^-1above the triplet intermediate at 0 K. The geometries

of the triplet and singlet transition states are quite similar, with

the exception of the dihedral angle that specifies the relative

orientation of the two fragments and, to a smaller extent, the

CCH angle around the carbenic carbon, which is larger in the

triplet, just as in methylene itself. The A factor and the activation

energy of the singlet pathway (reaction 79) are 2.7 × 10¹⁵s^-1

and 248 kJ mol^-1, respectively, in the temperature range 1400-

2100 K. For the triplet pathway, the corresponding values were

found to be 3.3 × 10¹⁵s^-1and 173 kJ mol^-1. Only the singlet

pathway was considered in the kinetic modeling.

Rearrangement to Isocyanide. The reaction pathway involving

an isocyanide isomer of succinonitrile as an intermediate is

summarized in Figure 4c. Succinonitrile was expected and

indeed found to be similar to acetonitrile in that it isomerizes

readily to form an isocyanide, which then rearranges, yielding

acrylonitrile and hydrogen cyanide as products. The transition

state for the isomerization step (structure 31) is very similar

both structurally and energetically to that computed for aceto-

nitrile.⁴²The critical energy for this reaction was calculated as

251 kJ mol^-1at 0 K, with the reaction being 85 kJ mol^-1

endothermic at this temperature. The A factor for the forward

reaction (10) in the temperature range 1400-2100 K is 2.0 ×

10¹⁴s^-1and the corresponding activation energy is 260 kJ

mol^-1. As each of the three species (11, 31, 32) in this first

In the case of the singlet carbene intermediate (structure 26)

the G2 energies are such that it appears 30 kJ mol^-1less stable

1064 J. Phys. Chem. A, Vol. 103, No. 8, 1999

Sendt et al.

reaction was found to have a low-frequency vibrational mode

corresponding to hindered rotation about the central CC bond,

the partition functions were also calculated using a hindered

rotor approximation. However, the resulting A factors for the

forward and reverse reactions hardly differed from those

obtained by the all-vibrations model.

vibrational frequency obtained for the hindered rotation in the

transition state is lower than the analogous frequency in

succinonitrile, one would expect that use of the hindered rotor

treatment would have a larger effect on the partition function

of the transition state than on that of succinonitrile. As expected,

the partition functions of both systems increased when the above

modes were treated as hindered rotors rather than harmonic

vibrations, but the change turned out to be larger for succino-

nitrile, which then resulted in a lower A factor. The explanation

of this unexpected trend is that while the rotational potential

energy curves of both systems have three minima between 0

and 2π, in the transition state structure one of these minima

occurs at too high an energy to contribute appreciably to the

partition function, especially at lower temperatures. It is

expected, however, that if the geometries were allowed to relax

in conjunction with the internal rotation, the effect noted above

would become less pronounced. The A factor and activation

energy used in the kinetic model are 2.5 × 10¹⁵s^-1and 400 kJ

mol^-1(reaction 12).

The second step in this pathway is a concerted elimination

of HCN that occurs via a five-centered transition state (structure

33). The geometry of this transition state is characterized by a

very long breaking CN bond of length ∼2.7 Å, as well as a

long forming CH bond of length ∼1.7 Å. The critical barrier

for this reaction was computed to be 345 kJ mol^-1at 0 K. The

resulting A factor and activation energy are 2.9 × 10¹⁵s^-1and

364 kJ mol^-1respectively, in the temperature range 1400-2100

K (reaction 11).

Concerted Elimination of Hydrogen Cyanide. The details of

the concerted elimination of hydrogen cyanide from succino-

nitrile are summarized in Figure 4d. This pathway is the

simultaneous 1,3-H transfer and CC bond cleavage, and as such,

it is associated with a single transition state (structure 34) that

is characterized by a very long forming CH bond of length ∼1.8

Å, a long breaking CC bond of length ∼2.5 Å, and a relatively

short breaking CH bond of length ∼1.247 Å. The HCN fragment

in this transition state is near-linear (with an angle of 172.6°)

and lies almost normal to the near-planar acrylonitrile fragment.

The critical energy at 0 K for this reaction was found to be 444

kJ mol^-1at 0 K. In the temperature range 1400-2100 K, the A

factor and activation energy were calculated to be 1.4 × 10¹⁶

s^-1and 463 kJ mol^-1, respectively. Preliminary modeling studies

indicated that this reaction was far too slow to have a significant

effect on the overall kinetics, and therefore, it has not been

included in the final kinetic model.

The radical thus produced (structure 36) can then decompose

by further bond fission, with the loss of either a cyanide radical

or of another H atom. The first of these reactions, yielding

acrylonitrile and cyanide is endothermic by 239 kJ mol^-1at 0

K. Characterizing the transition state for this reaction proved

elusive and we were unsuccessful in locating it using CASSCF

theory. Thus, in this case the SCF geometry (structure 37) is

used instead, since SCF found a saddle point on the PES. The

critical energy of this reaction was found to be 224 kJ mol^-1at

0 K. The energy of the transition state is actually lower than

that of the products, when computed at a higher level of theory,

which suggests that the barrier to the reverse reaction is likely

to be quite small. Calculations on the chemically similar ethyl

cyanide system (replacing -CN by -H), where both SCF and

CASSCF successfully located the transition state, suggest that

SCF theory is likely to underestimate by ∼0.2 Å the length of

the breaking CC bond, which was calculated to be 2.3 Å in

structure 37. Although the energetic effect of such a discrepancy

is essentially negligible, it does have a more significant effect

on the rotational constants, as well as on the computed

frequencies, and hence on the partition function. Thus, in the

ethyl cyanide reaction the A factor, calculated using the CASSCF

geometry and frequencies, is 1.8 times larger than that obtained

from the SCF geometry and frequencies. Nevertheless, in this

work the kinetic parameters for reaction 18 are those based on

the SCF structure and frequencies. The resulting A factor and

CH Bond Fission Reaction. This mechanism, summarized in

Figure 4e, is analogous to the CH bond fission in acetonitrile

and in fact the endothermicities of the two reactions at 0 K are

within 14 kJ mol^-1of each other. For a fission reaction such as

this, where the reverse reaction is barrierless, variational

transition state theory needs to be used in order to obtain kinetic

parameters. The CASSCF calculations, similar to those per-

formed on acetonitrile, were carried out using two different

choices of active spaces: (6/6) and (10/10). The transition state

structures that were located are effectively the same. The

breaking CH bond distance in the transition state (structure 35)

is ∼2.4 Å, nearly the same as for acetonitrile. The CH distance

does depend on the temperature to some extent; it decreases

with increasing temperature, as found for acetonitrile as well

as methane.²¹To obtain the appropriate critical energy, the

energies of reactant, transition state, and products were recal-

culated at the CASPT2(10/10)/cc-pVDZ level of theory and then

scaled to bring the dissociation energy in line with the

corresponding G2 energy. Since both succinonitrile and the

activation energy are 2.2 × 10¹⁴s^-1and 230 kJ mol^-1

,

respectively, in the temperature range 1400-2100 K.

With respect to the loss of H from radical 36, there are two

possible pathways leading to the formation of cis- or trans-

dicyanoethylene. Only the second of these was studied. The

transition state (structure 38) was found to have a CH bond

with a length of ∼1.9 Å, with the leaving H atom out of plane

with respect to the NCCCCN skeleton. The critical energy for

this step was calculated to be 184 kJ mol^-1with an endother-

micity of 160 kJ mol^-1at 0 K. The calculated A factor and

activation energies in the temperature range 1400-2100 K were

computed to be 5.6 × 10¹³s^-1and 199 kJ mol^-1, and these

parameters have been used in the kinetic modeling of reaction

19.

transition state possess low-frequency modes (86 and 47 cm^-1

,

respectively) that correspond to hindered rotation about the

central CC bond, the partition functions and hence rate constants

were computed by the explicit hindered rotor treatment, as

discussed in the Theoretical and Computational Methods section.

The resulting A factor and activation energy in the temperature

range 1400-2100 K are 1.4 × 10¹⁵s^-1and 393 kJ mol^-1

,

respectively, when using the CASSCF(6/6)/DZP geometries and

frequencies. The alternative all-vibrations model gave 2.5 ×

10¹⁵s^-1and 393 kJ mol^-1. Use of the CASSCF(10/10)/DZP

geometries and frequencies in the all-vibrations model results

in 7.2 × 10¹⁵s^-1and 404 kJ mol^-1, indicating a fairly high

level of sensitivity to the level of theory. Given that the

Loss of Cyanide Radical from Succinonitrile. Succinonitrile

could lose a cyanide radical, resulting in a cyanoethyl radical

(structure 41) as a product. This step was found to be 505 kJ

mol^-1endothermic at 0 K, and consequently, it is unlikely that

this pathway could play a significant role in the decomposition

Pyrolysis Kinetics of Acetonitrile

J. Phys. Chem. A, Vol. 103, No. 8, 1999 1065

kinetics of succinonitrile since it is considerably more energetic

than any of the other pathways. It is not included in the kinetic

model and has not been studied further.

cyanide group, with a simultaneous CC bond break, as implied

by the transition state (structure 43). In this transition state the

forming CH bond was found to be quite long (1.43 Å), while

the breaking CH bond was only slightly longer than the

equilibrium bond length (1.120 Å). The breaking CC bond,

however, was found to be 2.27 Å long, with the HCN being

near-linear. The critical barrier height for this step was found

to be 421 kJ mol^-1at 0 K, while the A factor and activation

energies in the temperature range 1400-2100 K were calculated

to be 5.9 × 10¹⁵s^-1and 442 kJ mol^-1. The second pathway

proceeds via a planar transition state (structure 44) that results

when the CN group in acrylonitrile swings toward the hydrogen

atom and forms a CH bond while the CC bond breaks. However,

this structure is actually a second-order saddle point, with one

of the imaginary frequencies corresponding to an out-of-plane

mode. Allowing structure 44 to become nonplanar did lead to

a first-order saddle point, but the resulting transition state

(structure 62) connects with a different reactant, indicating the

presence of another pathway, to be discussed later. The critical

barrier for this pathway via structure 44 at 0 K was found to be

410 kJ mol^-1, with an A factor and activation energy in the

temperature range of 1400-2100 K of 3.5 × 10¹³s^-1and 412

kJ mol^-1. Consequently, this is an unimportant reaction and it

is not included in the kinetic model. This first step in the

decomposition reaction of acrylonitrile, resulting in HCN and

vinylidine (structure 45), was found to be 345 kJ mol^-1endo-

thermic at 0 K. The following step is the well-known low-energy

rearrangement of vinylidine to acetylene,⁴⁶via a low-energy

transition state (structure 46). In this work, the barrier to

decomposition, computed using G2, was found to be just 2 kJ

mol^-1at 0 K. This result is consistent with those obtained

previously.⁴⁶

Bimolecular Reaction with Atomic Hydrogen. Atomic hy-

drogen could abstract a cyanide radical from succinonitrile in a

way that is similar to the reaction (6) between acetonitrile and

hydrogen. In this case the other product of the reaction is the

cyanoethyl radical (structure 40), which could then lose either

a hydrogen atom to form acrylonitrile or a cyanide radical to

form ethylene. These pathways are shown in Figure 4f. The

first step in this reaction pathway is expected to have a low

activation energy, as in the case of acetonitrile, and therefore it

is unlikely to be rate determining. At 0 K, the first step was

calculated to be 26 kJ mol^-1exothermic (cf. 18 kJ mol^-1for

the acetonitrile reaction). This first step was not studied further

computationally. The kinetic modeling parameters for the overall

reaction (14) are an A factor of 2.0 × 10¹⁴cm³mol^-1s^-1and

an activation energy of 42 kJ mol^-1. Loss of hydrogen from

cyanoethyl occurs via a transition state (structure 41) with a

breaking CH bond length of 1.65 Å, with the H atom above

the plane of the acrylonitrile fragment. The critical energy for

this reaction was computed to be 142 kJ mol ^-1at 0 K with the

corresponding A factor and activation energy being 3.1 × 10¹³

s^-1and 153 kJ mol^-1in the temperature range 1400-2100 K.

These values have been used in the kinetic modeling (reaction

20).

Loss of CN from cyanoethyl is associated with a transition

state (structure 42) with a very long breaking CC bond of ∼2.6

Å, with the cyanide being above the plane of the ethylene

fragment. The critical barrier for this reaction was found to be

197 kJ mol^-1at 0 K, with an A factor and activation energy of

1.8 × 10¹⁵s^-1and 206 kJ mol^-1, respectively, in the temperature

range 1400-2100 K. Although the energy of the transition state

was found to lie below that of the products, at the CASSCF

level of theory (that was used to optimize the geometry was

carried out), the structure is unambiguously identifiable as a

saddle point with an energy that is slightly higher than that of

the products. Thus, it appears that the reverse barrier is quite

low. Given that the computed value of 206 kJ mol^-1for the

activation energy of reaction 57 is below the computed or

experimental heats of reactions, in the kinetic model the

Elimination of H₂To Form CyanoVinylidene. The concerted

elimination of H₂from acrylonitrile occurs the same way as

the analogous process in ethylene,⁴⁶but now there are two

slightly different pathways corresponding to the cis or trans

orientation of the departing H₂in the transition states, as shown

in Figure 5b. The geometries of the transition states (structures

47 and 48) are very similar to that obtained for ethylene, while

the computed critical energies at 0 K, viz. 406 and 401 kJ mol^-1

,

respectively, are slightly higher than in ethylene. The resulting

A factors and activation energies in the temperature range 1400-

2100 K are 1.9 × 10¹⁵s^-1, 1.7 × 10¹⁵s^-1, 429 kJ mol^-1, and

422 kJ mol^-1, respectively. The cyanovinylidene intermediate

(structure 49) that results can then rearrange to form cy-

anoacetylene, which can occur via hydrogen migration, as in

vinylidene, or by a qualitatively similar migration of CN. The

critical energy for the hydrogen migration, via structure 50 as

transition state, was found to be 5 kJ mol^-1, nearly the same as

in vinylidene. Migration of CN occurs via a high-energy

transition state (structure 51), which is ∼120 kJ mol^-1higher

than the barrier to hydrogen migration. This pathway is therefore

unimportant in the rearrangement process. The cyanoacetylene

product (structure 52) is a conjugated molecule and the high

degree of π electron delocalization is manifested in the very

short, formally single CC bond of length 1.375 Å.

activation energy for this reaction was increased to 251 kJ mol^-1

.

Sensitivity analysis (see Kinetic Modeling section) revealed,

however, that reaction 57 was not a sensitive reaction.

Eliminations of H₂and HCN from Acrylonitrile. Acrylo-

nitrile, a cyano-substituted ethylene, can undergo decomposition

in an analogous manner to ethylene. Quantum chemical studies⁴⁶

on the decomposition of ethylene revealed two decomposition

pathways: first, loss of H₂to form vinylidene followed by

H-migration to form acetylene and, second, rearrangement to

ethylidene followed by loss of H₂. The vinylidene pathway was

found to be the lower energy route, and although a direct 1,2

elimination was searched for, no transition state corresponding

to such a process (with a single imaginary frequency) could be

found.⁴⁶In this work, loss of H₂and HCN from acrylonitrile

was studied via both the vinylidene and ethylidine types of

pathways, as well as pathways that involve hydrogen migration

followed by CC bond cleavage, which are similar to those

occurring in succinonitrile. In the study of the latter mechanisms

only the singlet pathways were considered.

Formation of 2-Cyanoethylidene. The interconversion of

acrylonitrile and 2-cyanoethylidene (structure 26) is analogous

to the interconversion of ethylene and ethylidine, although

depending on the relative orientation of the cyano moiety and

the migrating H atom, two slightly different pathways are

possible. 2-Cyanoethylidene in turn can eliminate either H₂or

HCN. The salient features of these pathways are given in Figure

5c. As for the analogous 1,2-H migration in ethylene followed

Concerted Elimination of HCN To Form Vinylidene. There

are two distinct ways in which the one-step, concerted elimina-

tion of HCN from acrylonitrile can occur, as shown in Figure

5a. The first of these corresponds to a H atom transfer to the

1066 J. Phys. Chem. A, Vol. 103, No. 8, 1999

Sendt et al.

Figure 5. (a) Potential energy surface for decomposition of acrylonitrile by concerted elimination of HCN to form vinylidene. (Energies at 0 K

given in kJ mol^-1.) (b) Potential energy surface for decomposition of acrylonitrile by elimination of H₂to form cyanovinylidene. (Energies at 0 K

given in kJ mol^-1.) (c) Potential energy surface for decomposition of acrylonitrile via 2-cyanoethylidene. (Energies at 0 K given in kJ mol^-1.) (d)

Potential energy surface for decomposition of acrylonitrile via 1-cyanoethylidene. (Energies at 0 K given in kJ mol^-1.) (e) Potential energy surface

for decomposition of acrylonitrile by nonconcerted loss of HCN via 1,2-H migration. (Energies at 0 K given in kJ mol^-1.) (f) Potential energy

surface for decomposition of acrylonitrile by nonconcerted loss of HCN via 1,3-H migration. (Energies at 0 K given in kJ mol^-1.)

by 1,1-H₂elimination, the second step is the more energetic

one. The structure of the transition state leading to loss of H₂

(structure 53) is very similar to that found in the ethylene

decomposition reaction. Although the cyanoacetylene fragment

in the transition state was found to be almost planar, with the

CCCH dihedral angle being 13.6°, the structure is quite asym-

metric, with the breaking CH bond lengths being 1.396 and

1.454 Å. The critical barrier for this process at 0 K was found

to be 145 kJ mol^-1, while the A factor and activation energies

in the temperature range 1400-2100 K are 4.0 × 10¹³s^-1and

158 kJ mol^-1respectively. The transition state leading to the

loss of HCN (structure 54) was found to correspond to a

higher critical energy, 197 kJ mol^-1at 0 K, with the A factor

and activation energy in the temperature range 1400-2100 K

being 1.9 × 10¹³s^-1and 206 kJ mol^-1, respectively. In this

structure the acetylene fragment was found to be near-planar,

while the departing HCN is quite bent, the bond angle being

145.8°.

Formation of 1-Cyanoethylidene. Acrylonitrile could also

form 1-cyanoethylidene (structure 56) as an intermediate, via a

process that takes place in the opposite direction to that discussed

in the previous section. The PES corresponding to this pathway

is summarized in Figure 5d. The transition state (structure 55)

leading to the formation of 1-cyanoethylidene has a significantly

higher energy (391 kJ mol^-1at 0 K) than the transition state

that led to 2-cyanoethylidene, even though the former carbene

is the more stable isomer, by 52 kJ mol^-1. Thus, in contrast

with the interconversion of acrylonitrile and 2-cyanoethylidene,

there is now a significant barrier to the reverse reaction. The

forward and reverse A factors and activation energies for this

Pyrolysis Kinetics of Acetonitrile

J. Phys. Chem. A, Vol. 103, No. 8, 1999 1067

process in the temperature range 1400-2100 K were found to

be 9.5 × 10¹³s^-1, 1.2 × 10¹³s^-1, 402 kJ mol^-1, and 120 kJ

mol^-1, respectively. 1-Cyanoethylidene can eliminate H₂, with

the concomitant formation of cyanoacetylene. The transition

state for this process (structure 57) was found to be of C_s

symmetry, with the breaking CH bonds being 1.295 Å, shorter

than in the analogous H₂elimination process from 2-cyanoeth-

ylidene. The critical barrier at 0 K for this reaction was found

to be 138 kJ mol^-1, while the A factor and activation energy in

the temperature range 1400-2100 K were computed to be 5.9

× 10¹³s^-1and 150 kJ mol^-1, respectively.

Nonconcerted Loss of HCN Via 1,3-H Migration. 1,3-H

migration in acrylonitrile, as a first step, defines another possible

pathway, as shown in Figure 5f, that also leads to the elimination

of HCN. This occurs via structure 63 as the transition state and

results in a biradical intermediate (structure 64). The transition

state was found to be planar with CH bond lengths of 1.28 and

1.86 Å, respectively, for the forming and breaking bonds. The

critical barrier for this step at 0 K was found to be 399 kJ mol^-1

,

which is 30 kJ mol^-1higher than the critical energy for the

1,2-H transfer discussed in the previous section. With respect

to stability, the biradical intermediate (64) (whose energy was

obtained on the basis of CASPT2(10/10)/VTZ2P calculations)

is comparable with its isomer (59) that is an intermediate of

the 1,2-H transfer step. The 1,3-H migration was found to be

345 kJ mol^-1endothermic at 0 K and the forward and reverse

A factors and activation energies in the temperature range 1400-

2100 K were found to be 1.4 × 10¹⁴s^-1, 1.5 × 10¹³s^-1, 414

kJ mol^-1, and 60 kJ mol^-1respectively. This biradical inter-

mediate can readily undergo CC bond cleavage to produce HCN

and acetylene. The transition state (structure 65) for this process

was found to have a breaking CC bond length of ∼1.7 Å. This

step was found to have a critical barrier of 9 kJ mol^-1at 0 K,

while the corresponding A factor and activation energy in the

temperature range 1400-2100 K were found to be 4.0 × 10¹⁴

s^-1and 22 kJ mol^-1respectively.

Nonconcerted Loss of HCN Via 1,2-H Migration. In this

reaction pathway, summarized in Figure 5e, the first step is the

formation of a biradical intermediate (structure 59), as a result

of a 1,2-H transfer. The transition state (structure 58) as well

as the reactant and product in this reaction was found to be

planar, which is somewhat unusual in a 1,2-H migration.

However, this reaction is somewhat unique in that all three

species are conjugated and the π electron delocalization is of

course optimal for planar structures. The critical energy of this

reaction was found to be 369 kJ mol^-1at 0 K, and this step

was found to be 361 kJ mol^-1endothermic. The energy of the

biradical intermediate (structure 59) relative to acrylonitrile was

calculated at the CASPT2(10/10)/VTZ2P level of theory. The

forward and reverse A factors and activation energies were found

to be 3.7 × 10¹⁴s^-1, 1.9 × 10¹³s^-1, 385 kJ mol^-1, and 14 kJ

mol^-1, respectively, in the temperature range 1400-2100 K.

The biradical intermediate was found to be able to cyclize (via

structure 60) to a fairly stable, closed shell molecule (structure

61), which then loses HCN. The transition state (structure 60)

leading to the cyclic intermediate is characterized by a very

long (forming) CN bond of length ∼2.3 Å and a correspondingly

large NCC angle of ∼113°. The cyclic, conjugated intermediate

(structure 61) is isoelectronic with methylene cyclopropene and

its high stability can be explained in terms of the large π

delocalization (resonance) energy that such a molecule is

expected to possess. The geometric parameters, viz., short CC

and CN “single” bonds and somewhat long CC and CN “double”

bonds are consistent with the high level of π delocalization.

The cyclization reaction at 0 K was calculated to be 148 kJ

mol^-1exothermic, with a critical barrier of 24 kJ mol^-1. The

forward and reverse A factors and activation energies for this

step in the temperature range 1400-2100 K were found to be

Summary of Computational Results. The key reactions

studied in this work by ab initio methods are the decomposition

of acetonitrile to yield cyanomethyl and H, the subsequent

reaction between H and acetonitrile (leading to the products

CH₃and HCN), the recombination of cyanomethyls to form

succinonitrile, the decomposition of the latter to yield acryloni-

trile and HCN, and finally, the decomposition reactions of

acrylonitrile to acetylene and HCN and to cyanoacetylene and

H₂, respectively.

A crucial aspect of the work has been the ab initio computa-

tion of the heat of formation of cyanomethyl as the accuracy of

the kinetic model critically depends on this quantity. The

reaction between H and acetonitrile are similarly an important

part of the kinetic model and the ab initio parameters are used

in the model without any adjustment.

Six distinct intramolecular reaction pathways were considered

in the study of the decomposition of succinonitrile, although

only four of these (1,3-H transfer, 1,2-H transfer, rearrangement

to isocyanide, and CH bond fission) were found to be suf-

ficiently important to be included in the kinetic model. The most

important of these is the CH bond fission reaction, viz., H loss,

which is the dominant pathway at the higher temperatures. The

bimolecular reaction between H and succinonitrile, where H

effectively plays the role of a catalyst, is included in the model,

although it is not a sensitive reaction. As discussed in the next

section, H abstraction by cyanomethyl is the most important

reaction at the lower temperatures that leads to the formation

of acrylonitrile and HCN.

5.5 × 10¹²s^-1, 1.8 × 10¹⁴s^-1, 21 kJ mol^-1, and 182 kJ mol^-1

,

respectively. Loss of HCN is then predicted to occur by a one-

step process, via transition state 62, whose geometry was

optimized using the CASSCF method. The structure is fairly

loose, with the HCN and vinylidene fragments separated by ∼2

Å, the lengths of the breaking CC and CN bonds being

approximately 1.9 and 2.4 Å, respectively. As in the case of a

similar bond fission reaction in succinonitrile (involving struc-

tures 22, 24, and 26), the structure of this transition state is

such that one may consider the possibility that it connects the

biradical structure 59 as reactant with HCN and vinylidene as

products on the reaction path. However, the geometry relaxations

along the reaction path (when starting at the transition state)

consistently converged to the cyclic structure. A more extensive

search failed to locate another transition state (that would

connect the biradical and the products); irrespective of the

starting structures, the optimizations invariably converged to

structure 62. The critical barrier for this step at 0 K was found

to be 150 kJ mol^-1, while the corresponding A factor and

activation energy in the temperature range 1400-2100 K were

calculated to be 2.2 × 10¹⁵s^-1and 165 kJ mol^-1, respectively.

Concerning the decomposition of acrylonitrile, six distinct

mechanisms were studied. These are concerted 1,1-elimination

of HCN or of H₂(resulting in acetylene and HCN and

cyanoacetylene and H₂, respectively), 1,2-H migrations (to form

two cyanoethylidene isomers that decompose to acetylene and

HCN and cyanoacetylene and H₂), and nonconcerted HCN loss

reactions via 1,2-H and 1,3-H migrations. The computed critical

energies for these pathways were found to be too high, however,

by ∼12 and ∼36 kJ mol^-1for the HCN and H₂elimination

reactions, respectively, to successfully model the decomposition

of acrylonitrile. As discussed in the next section, the rate

1068 J. Phys. Chem. A, Vol. 103, No. 8, 1999

Sendt et al.

TABLE 3: Thermochemical Parameters for the Acetonitrile System

C⁰_p/J K^-1mol^-1

∆_fH ⁰₂₉₈/kJ mol^-1

S⁰₂₉₈/J K^-1mol^-1

300

500

1000

1500

2000

ref

CH₃CN

73.9

263.4

198.6

272.6

381.8

183.5

245.4

338.7

409.5

380.6

226.7

480.2

604.7

244.8

250.6

328.9

331.3

337.0

272.9

286.7

299.9

280.5

236.2

290.8

234.2

239.0

52.4

54.5

98.0

93.4

95.9

63.8

71.7

84.7

66.6

64.8

46.1

44.6

69.4

67.7

131.1

127.3

125.8

87.6

97.2

112.8

83.0

78.1

90.8

55.1

50.2

98.3

86.2

112.3

95.4

120.6

100.0

211.0

213.1

193.4

144.5

165.6

170.5

122.7

105.4

165.2

74.3

36^b

CH₂CN

PW/abin^c

63^b

SUCC^a

178.5

178.2

166.5

120.9

136.9

147.9

105.8

93.7

200.4

201.7

184.6

136.6

156.9

163.0

116.2

101.4

154.3

71.6

ISUCC^a

NCCHCH₂CN

H₂CCHCN

CH₂CH₂CN

C₄H₂N₂

HCCHCN

HCCCN

CH₃CHCN

NCCH(3)

C₂N

PW/add^d

PW/add

7/add

64/abin

PW/add

65/abin

66

133.5

66.4

57.5

59.8

61.2

0

^aName as given in Table 2. S₂₉₈and C⁰_pvalues calculated statistically in the present work. ^cEvaluated by ab initio calculation in the present

b

work. ^dEstimated by group additivity methods in the present work.

parameters for these reactions, viz., (28) and (29), were instead

derived on the basis of experimental data.²

use of QRRK analysis, taking into account the barriers derived

by ab initio calculations as described above.

Kinetic Modeling. Kinetic modeling using the model given

in Table 2 was performed using the Sandia CHEMKIN code,⁴⁷

together with the shock tube code⁴⁸(modified to allow for cool-

ing by the reflected rarefaction wave) and the ordinary differ-

ential equation solver LSODE.⁴⁹Predicted decomposition and

product profiles are compared with experimental data in Figure

1. Rate sensitivity analysis has been carried out on the model

using the SENKIN⁵⁰code. Thermodynamic sensitivity analysis

was performed using the code HSENKIN developed by Muris

and Haynes.⁵¹Several of the key rate constants in the model

have been evaluated as discussed above from ab initio calcula-

tions. In general, reverse rate constants have been evaluated

using the equilibrium constants, which, in turn, have been calcu-

lated from the thermochemistry by standard methods. Thermo-

chemical parameters for species important in the acetonitrile

system are given in Table 3. For comparison purposes, the

reverse rate constants given in Table 2 have been fitted to the

simple Arrhenius form over the temperature range of 1400-

2100 K.

Comparison between the predictions of the kinetic model and

experiment is made in Figure 1. It may be seen that the model

gives an acceptable fit to the experimental species yields. The

model also predicts that only an indetectably low yield of

succinonitrile is formed at all temperatures. Rate sensitivity

analysis of the model shows that overwhelmingly the most

sensitive reaction for overall reaction rate and for all major

species is the initiation reaction 1. Initially, we used our

experimental value of k_disfor k₁. On account of the high

sensitivity of virtually all product species to k₁throughout the

entire temperature range, we have made minor changes to our

initially chosen value of k₁to best fit the product profiles. The

adjustment to k₁is within our experimental error but probably

reflects the contribution of secondary reactions to the disap-

pearance of acetonitrile. The value of k₁given in Table 2 is

specific to our pressures (∼12 atm). Reaction 2, the addition

of H and subsequent fission of HCN exhibits considerable

sensitivity for the overall reaction and for major products. Rate

sensitivity coefficients of the most sensitive reactions for the

major products are shown in Figure 6.

In addition to reactions described above, there are bimolecular

reactions, including abstraction, addition, and recombination

reactions whose rate constants in some cases have been

estimated by comparison with analogous reactions of known

rate constant. In general, reactions for which rate constants have

been estimated are not sensitive reactions in the model. The

model contains a subset of C₁, C₂, and some C₃and C₄species

reactions whose rate constants have been taken from the

GRIMech,⁵²Warnatz,⁵³and Miller and Bowman⁵⁴compilations.

At temperatures around 2000 K, the C-C bond fission reaction

(23) of acetonitrile shows some sensitivity and the model

contains an estimated rate constant k₂₃. This value, which is

specific to our pressure regime, was chosen to optimize

agreement between experimental and theoretical acetonitrile

profiles above 2000 K. Also in this temperature region we have

included decomposition reactions of CH₂CN radicals to produce

both CCN and triplet state NCCH(3) species. Rate constants

for these reactions, which would be expected to be in the falloff

regime, were estimated by RRKM methods (using program

UNIMOL) and the values of k₇₂and k₇₃thus obtained are

specific to our pressure of 12 atm. Apart from the metathesis

reaction (86) between CN and CH₂CN radicals, which produces

NCCH(3), none of the reactions involving CCN or NCCH(3)

is a sensitive reaction. Also, a direct chemiactivated reaction

between two CH₂CN radicals to form HCN and acrylonitrile,

reaction 8, was included with rate parameters obtained by the

The high-temperature tail of the acetonitrile profile agreed

less well with experiment when the literature value for ∆_fH ₂⁰₉₈

(CH₂CN) of 245 kJ mol^-1was used. Thermodynamic sensitivity

analysis (using HSENKIN) indicated that an increase in the

enthalpy of formation of this species would bring modeled and

experimental profiles into better agreement. The optimal agree-

ment occurred for ∆_fH ⁰₂₉₈(CH₂CN) ) 263 ( 9 kJ mol^-1, in

excellent agreement with the ab initio result.

As can be seen in Figure 1, both acrylonitrile and cy-

anoacetylene exhibit maxima in their yields at temperatures

around 1700 K. Kinetic modeling of these profiles is a challenge.

In the initial formulation of the kinetic model we adopted the

ab initio barriers for 1,1 elimination of H₂and HCN from

acrylonitrile and assumed a similar barrier for 1,1 elimination

of HCN from the fumaronitriles (reactions 28, 29, and 82,

respectively). These are all sensitive reactions, at least for

formation of acrylonitrile and cyanoacetylene. However, the ab

initio barriers gave rise to rate constants for these reactions that

were too small and led to calculated yields of these products

that were too low in comparison with experiment, and also led

to a high residual yield of fumaronitriles.

As argued in the previous section, the transition state for 1,1

elimination of H₂from acrylonitrile should be analogous to the

1,1 elimination from ethylene. Some years ago, Kiefer et al.⁵⁵

Pyrolysis Kinetics of Acetonitrile

J. Phys. Chem. A, Vol. 103, No. 8, 1999 1069

Figure 6. Variation with temperature of sensitivity coefficients for the designated species. Only the most sensitive reactions are shown. Reaction

numbers are as given in Table 2.

1070 J. Phys. Chem. A, Vol. 103, No. 8, 1999

Sendt et al.

carried out a RRKM extrapolation of kinetic data for ethylene

pyrolysis measured by three different laboratories, all at

pressures just above the low-pressure limit. All sets of data

extrapolated to a critical energy for 1,1 elimination of H₂from

ethylene of 356 ( 8 kJ mol^-1. This is some 36 kJ lower in

energy than the ab initio barrier height obtained by Jensen et

al.,⁴⁶which in turn agrees closely with our G2 result as well as

the earlier value of Raghavachari et al.⁵⁶We have now

extrapolated some unpublished kinetic data for ethylene de-

composition measured by Organ⁵⁷in this laboratory using two

techniquesssingle pulse shock tube product analysis and

infrared real time laser absorption spectroscopy of C₂H₄behind

reflected shocks, both studies at a pressure of 22 atm, just below

the high-pressure limit. Using the same RRKM parameters as

Kiefer et al.,⁵⁵our RRKM extrapolation leads to the same critical

energy of 356 ( 8 kJ mol^-1. There is clearly an anomaly

between the RRKM extrapolated result and the ab initio

calculation that we are unable to reconcile at present.

quenching, the reaction flux forming succinonitrile is 2.0 × 10^-5

mol cm^-3s^-1, arising from reaction 7, the recombination of

two CH₂CN radicals. The net reaction flux decomposing

succinonitrile is also 2.0 × 10^-5mol cm^-3s^-1, of which 90%

takes place via reaction 17, abstraction from succinonitrile by

CH₂CN. The 1,3-H transfer pathway (reaction 9) accounts for

3.0% of reaction flux. The remainder of the reaction flux

comprises the bimolecular reactions (13), (14), and (15) together

with the 1,2-H transfer pathway on the singlet surface (reaction

78), each of these four reactions contributing approximately

equally to decomposition of succinonitrile. Although the barrier

to isomerization of succinonitrile is the lowest of all the

rearrangements of this molecule, there is a negligible decom-

position flux passing through reactions 10 and 11. In essence,

succinonitrile and its isocyanide equilibrate rapidly but the

barrier for decomposition of the latter is too large. In the

rarefaction wave, recombination of CH₂CN radicals will con-

tinue whereas the destruction reactions will be essentially frozen

because of their appreciable activation energies.

We therefore believe that the same anomaly exists for the

analogous 1,1 eliminations from acrylonitrile and the fuma-

ronitriles. Therefore, to model our acrylonitrile and cyanoacety-

lene data we first adopted the experimental values given by

Lifshitz et al.²for a pressure of 3 atm and made an RRKM

extrapolation to our average pressure of 12 atm. However, the

values of k₂₈, k₂₉, and k₈₂so evaluated were too large so that

we have adjusted the respective activation energies to best fit

our profiles of cyanoacetylene and acrylonitrile. The rate

constants k₂₈, k₂₉, and k₈₂as shown in Table 2 are specific to

our pressure of 12 atm and are comparable with the literature

values⁵²for the analogous elimination of H₂from ethylene. It

should be pointed out that, at 1700 K, the model predicts that

the rate constant for decomposition of acrylonitrile must be about

10³s^-1to model the experimental profiles of acrylonitrile and

cyanoacetylene correctly. Summing all the elimination paths we

have identified above that produce H₂from acrylonitrile results

in an overall rate constant of ∼300 s^-1at 1700 K, while the

rate constant for the elimination of HCN is ∼500 s^-1. Thus the

discrepancy between theory and model rate constants could be

resolved if it is assumed that the former overestimated the

elimination barriers by about 12 kJ mol^-1. Such a discrepancy

is only about twice as large as the expected average deviation

between G2 and experiment, when G2 is applied to the

prediction of dissociation energies, ionization energies, electron

affinities, and proton affinities.⁵⁸In the absence of comparable

systematic studies on the performance of G2 when applied to

the computation of critical energy barriers, an error of 12 kJ

mol^-1may well be judged as acceptable and does not necessarily

imply major shortcomings in the kinetic model or the theoretical

methodology employed. Unfortunately, the same cannot be said

of the 1,1 elimination reaction in ethylene. A discrepancy of

36 kJ mol^-1in the critical energy appears to be too large to be

explained as due to the neglect of quantum mechanical

tunneling. Nevertheless, so as to be able to quantify its

contributions to the rate of H₂elimination, we have undertaken

a theoretical study of tunneling in this system (treating it as a

two-dimensional system, where the two degrees of freedom

effectively correspond to the motion of the hydrogens relative

to vinylidene, given the appropriate two-dimensional PES).

At temperatures above about 1800 K, decomposition of

succinonitrile is rapid and the model predicts a negligible yield

of this intermediate. At these higher temperatures the H fission

reaction (12) becomes an important mode of destruction of

succinonitrile.

Concerning the formation of acrylonitrile and cyanoacetylene,

the maximum yields of these occur at ∼1700 and ∼1750 K,

respectively. Of the total reaction flux forming H₂CCHCN, 42%

results from reaction 18, loss of CN from the radical produced

from succinonitrile by H loss. About 22% comes from the

reverse of reaction 58, i.e., H fission from CH₃CHCN, which

is in turn produced by H loss from ethyl cyanide. The remaining

formation flux comes approximately equally from reaction 8

(the direct chemiactivated reaction between two CH₂CN radi-

cals), reaction 9 (the 1,3-H transfer reaction of succinonitrile),

H loss from the CH₂CH₂CN radical from ethyl cyanide, and

reaction 79 involving the cyclic intermediate identified on the

singlet potential surface for 1,2-H transfer in succinonitrile. At

1700 K there is a small decomposition flux from acrylonitrile.

This takes place principally via reactions 28 and 29 which are

the 1,1 elimination reactions of acrylonitrile to produce HCCCN

+ H₂and C₂H₂+ HCN, respectively.

At 1700 K, nearly all reaction flux involving HCCCN leads

to its formation. About 85% of flux is via reaction 82, the 1,1

elimination of HCN from the fumaronitriles. Reaction 28

contributes about 12%. Thus, both acrylonitrile and cyanoacety-

lene originate largely from succinonitrile, albeit indirectly in

the case of the latter product, although ethyl cyanide is also a

precursor of the former product.

We now turn to a consideration of a falloff calculation of

the rate constant for fission of a hydrogen from acetonitrile

(reaction 1). Using the ab initio CASSCF geometry and

(unscaled) frequencies of acetonitrile and the transition state

with the ab initio value of the CH bond distance of 2.532 Å in

the transition state as calculated according to Variational

Transition State Theory, together with the critical energy of 388

kJ mol^-1(calculated at the G2 level of theory), we have carried

out a RRKM high-pressure limit calculation of k₁. Over the

temperature range of 1350-1750 K, the Arrhenius form k_1∞

)

1.2 × 10¹⁶exp(-413 kJ mol^-1/RT) s^-1holds well, although

The model predicts a maximum yield of succinonitrile of

1.1% (based on the initial reactant) at a temperature of 1625 K.

This low yield, together with the low volatility of succinonitrile

(BP 267 °C), would render it most unlikely to be detected in

the shock tube experiments. At 1625 K and at the end of the

residence time (885 µs) just prior to the rarefaction wave

the empirical expression 9.18 × 10¹²T^0.85exp(-401 kJ mol^-1

/

RT) gives an even better fit of the RRKM results. Our value of

k₁) 8 × 10¹⁴exp(-397 kJ mol^-1/RT) for an average pressure

of 12 atm was evaluated from CH₃CN profiles over the

temperature range of 1500-1750 K. Using a value of ∆E_down

Pyrolysis Kinetics of Acetonitrile

J. Phys. Chem. A, Vol. 103, No. 8, 1999 1071

of 800 cm^-1and the calculated k_1∞and critical energy, we can

substantially reproduce our 12 atm kinetic results by means of

an RRKM falloff calculation.

product in shock tube experiments is attributed to its low

calculated yield (max. ∼1% based on initial reactant). This,

together with its low volatility, precludes its observation among

the products of a single pulse shock tube. Succinonitrile has,

however, been detected in trace quantities in the products from

a completely stirred flow reactor. A detailed ab initio study of

the succinonitrile potential energy surface has been made to

search for possible intramolecular decomposition pathways of

low barrier to acrylonitrile and HCN. However, succinonitrile

decomposes principally via hydrogen loss. At the lower tem-

peratures of this study, this takes place by H-abstraction by

cyanomethyl. At higher temperatures, direct fission becomes

an important route. The NCCHCH₂CN radical so produced is

an important precursor of acrylonitrile but also leads to the

intermediate formation of the fumaronitriles, which, in turn, are

important for cyanoacetylene formation. Rate constants for these

reactions have been evaluated by ab initio calculation. Ethyl

cyanide is also a source of acrylonitrile. Several H₂and HCN

elimination pathways from acrylonitrile have been identified.

Lifshitz et al.²obtained a value for k₁of 6.17 × 10¹⁵exp-

(-404 kJ mol^-1/RT) s^-1from CH₃CN profiles over the

somewhat lower temperature range of 1350-1550 K at about

3 atm pressure. Their values of k₁are, within experimental error,

equal to the calculated high-pressure limit values over their range

of temperatures. For their slightly lower temperature range, it

might be expected that their rates would be somewhat less

affected by falloff. We have not, however, been able to

reproduce their result with a value of ∆E_downof 800 cm^-1or,

indeed, with a very much larger value of this parameter. We

cannot explain this discrepancy since we would expect that a

molecule as small as acetonitrile dilute in an argon bath gas at

pressures around 3 atm should show significant falloff in its

fission reaction.

In their recent modeling study, Lifshitz and Tamburu⁵report

the value of k₁) 1.0 × 10¹⁵exp(-389 kJ mol^-1/RT) s^-1

,

presumably for a pressure of 3 atm. This value, derived from

their ignition delay studies,⁴is, at 1500 K, lower than their earlier

one by about a factor of 2 but is still about a factor of 2 larger

than ours at the same temperature.

Acknowledgment. The financial support of the Australian

Research Council is acknowledged. K.S. gratefully acknowl-

edges the award of an Australian Postgraduate Research

Scholarship.

Turning to a comparison between our kinetic models for

acetonitrile pyrolysis, both models agree that the major route

to primary products HCN, H₂, and CH₄involves initiation by

CH fission, H addition and fission to CH₃+ HCN, and H and

CH₃abstraction reactions to form hydrogen and methane

(reactions 1-4, Table 2) with minor differences between models

in the rate constants for these reactions. The major difference

between the models lies in the fate of the C₄H₄N₂species arising

from recombination of two CH₂CN radicals. We have identified

this recombination product as succinonitrile, and our detailed

study of its reaction potential energy surface has enabled a better

understanding of the mechanism of its subsequent decomposition

pathways and of the formation of acrylonitrile and cyanoacety-

lene to be made. Lifshitz and Tamburu considered that the major

decomposition pathways of C₄H₄N₂were to form H₂CCHCN

+ HCN or C₂N₂+ C₂H₄. As discussed above, at high

temperatures, our ab initio studies show that H fission from

succinonitrile is a most important route at high temperatures

and we find no evidence for a route to cyanogen.

Supporting Information Available: Geometries and ener-

gies of all molecular species discussed in this work are available

in Tables 1S-9S. This information is available free of charge

via the Internet at http://pubs.acs.org.

References and Notes

(1) Bowman, C. T. 24th International Symposium on Combustion; The

Combustion Institute: Pittsburgh, PA, 1992; p 859.

(2) Lifshitz, A.; Moran, A.; Bidani, S. Int. J. Chem. Kinet. 1987, 19,

61.

(3) Ikeda, E.; Mackie, J. C. 26th International Symposium on Combus-

tion; The Combustion Institute: Pittsburgh, PA, 1996; p 597.

(4) Lifshitz, A.; Tamburu, C.; Carroll, H. F. Int. J. Chem. Kinet. 1997,

29, 839.

(5) Lifshitz, A.; Tamburu, C. Int. J. Chem. Kinet. 1998, 30, 341.

(6) Mackie, J. C.; Colket, M. B.; Nelson, P. F.; Esler, M. Int. J. Chem.

Kinet. 1991, 23, 733.

(7) Doughty, A.; Mackie, J. C. J. Phys. Chem. 1992, 96, 272.

(8) Ikeda, E.; Mackie, J. C. J. Anal. Appl. Pyrol. 1995, 34, 47.

(9) Ikeda, E. B. Sc. (Hons.) Thesis, University of Sydney, 1993.

(10) Curtiss, L. A.; Raghavachari, K.; Trucks, G. W.; Pople, J. A. J.

Chem. Phys. 1991, 94, 7221.

(11) Roos, B. O.; Taylor, P. R.; Siegbahn, P. E. S. Chem. Phys. 1980,

48, 157.

(12) Roos, B. O. In Ab initio Methods in Quantum Chemistry; Lawley,

K. P., Ed.; Wiley: Chichester, U.K., 1987; Vol. II, p 399.

(13) Dunning, T. H. J. Chem. Phys. 1970, 53, 2823.

(14) Andersson, K.; Malmqvist, P.-A° .; Roos, B. O.; Sadlej, A. J.;

Wolinski, K. J. Chem. Phys. 1990, 94, 5483.

Conclusions

The pyrolysis of acetonitrile over the temperature range of

1500-2100 K can be simulated by an 87 step reaction kinetic

model. The pyrolysis is initiated via hydrogen fission to form

the cyanomethyl, CH₂CN, radical. Extensive ab initio quantum

chemical calculations on the C₂H₂N potential energy surface

located five isomers of which the lowest lying is cyanomethyl.

On the basis of isodesmic reactions calculated at the G2 level

of theory, the value of ∆_fH ⁰₂₉₈(CH₂CN) was found to be 263

( 9 kJ mol^-1, about 20 kJ mol^-1higher than previously

reported. In the reaction model the initiation rate constant is

overwhelmingly the most sensitive for overall reaction and for

all products of significance. The limiting high-pressure value

(15) Andersson, K.; Malmqvist, P.-A° .; Roos, B. O. J. Chem. Phys. 1992,

96, 1282.

(16) Dunning, T. H. J. Chem. Phys. 1987, 90, 1007.

(17) Woon, D. E.; Dunning, T. H. J. Chem. Phys. 1990, 98, 1358.

(18) Steinfeld, J. I.; Francisco, J. S.; Hase, W. L. Chemical Kinetics

and Dynamics; Prentice Hall: Englewood Cliffs, NJ, 1989; p 308.

(19) McQuarrie, D. A. Statistical Mechanics; Harper & Row: New York,

1973; p 129.

(20) Nordholm, S.; Bacskay, G. Chem. Phys. Lett. 1976, 42, 253.

(21) Hase, W. L.; Mondro, S. L.; Duchovic, R. J.; Hirst, D. M. J. Am.

Chem. Soc. 1987, 109, 2916.

(22) Gilbert, R. G.; Smith, S. C.; Jordan, M. T. UNIMOL Program Suite

(Calculation of Falloff Curves for Unimolecular and Recombination

Reactions), 1993. Available from the authors: School of Chemistry,

University of Sydney, NSW 2006, Australia.

(23) Dean, A. M. J. Phys. Chem. 1985, 89, 4600.

(24) Westmoreland, P. R.; Howard, J. B.; Longwell, J. P.; Dean, A. M.

AIChE J. 1986, 32, 1971.

for this rate constant of k_1∞) 1.2 × 10¹⁶exp(-413 kJ mol^-1

/

RT) s^-1obtained by ab initio variational transition state

calculation agrees with our extrapolated experimental measure-

ment for an average pressure of 12 atm. The major nitrogen-

containing product is HCN, arising from H addition and

subsequent fission of the adduct. Rate constants for the addition

and fission reaction have been derived by ab initio calculation.

Recombination of two CH₂CN radicals leads to the formation

of succinonitrile as an intermediate. The nonobservation of this

1072 J. Phys. Chem. A, Vol. 103, No. 8, 1999

Sendt et al.

(25) Frisch, M. J.; Trucks, G. W.; Head-Gordon, M.; Gill, P. M. W.;

Wong, M. W.; Foresman, J. B.; Johnson, B. G.; Schlegel, H. B.; Robb, M.

A.; Replogle, E. S.; Gomperts, R.; Andres, J.; Rachavachari, L. K.; Binkley,

J. S.; Gonzalez, C.; Martin, R. L.; Fox, D. J.; Defrees, D. J.; Baker, J.;

Stewart, J. J. P.; Pople, J. A. Gaussian 92, Revision A; Gaussian, Inc.:

Pittsburgh, PA, 1992.

(26) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;

Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T.; Peterson, G. A.;

Montgomery, J. A.; Raghavachari, K.; Al-Laham, M. A.; Zakrzewski, V.

G.; Ortiz, J. V.; Foresman, J. B.; Cioslowski, J.; Stefanov, B. B.;

Nanayakkara, A.; Challacombe, M.; Peng, C. Y.; Ayala, P. Y.; Chen, W.;

Wong, M. W.; Andres, J. L.; Replogle, E. S.; Gomperts, R.; Martin, R. L.;

Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart, J. P.; Head-

Gordon, M.; Gonzales, C.; Pople, J. A. GAUSSIAN94, Revision E2;

Gaussian, Inc.: Pittsburgh, PA, 1994.

(27) Amos, R. D.; Rice, J. E. CADPAC 5.2: The Cambridge Analytic

Derivatives Package; University of Cambridge: Cambridge, U.K., 1992.

(28) CADPAC 6.0: The Cambridge Analytical Derivatives Package

Issue 6; University of Cambridge: Cambridge, U.K., 1995. A suite of

quantum chemistry programs developed by R. D. Amos with contributions

from I. L. Alberts, J. S. Andrews, S. M. Colwell, N. C. Handy, D. Jayatilaka,

P. J. Knowles, R. Kobayashi, K. E. Laidig, G. Laming, A. M. Lee, P. E.

Maslen, C. W. Murray, J. E. Rice, E. D. Simandiras, A. J. Stone, M.-D.

Su, and D. J. Tozer.

(41) Hehre, W. J.; Radom, L.; von R. Schleyer, P.; Pople, J. A. Ab Initio

Molecular Orbital Theory; Wiley: New York, 1986; p 156.

(42) Doughty, A.; Mackie, J. C.; Bacskay, G. B. J. Phys. Chem. 1994,

98, 13546.

(43) Swanton, D. J.; Bacskay, G. B.; Willett, G. D.; Hush, N. S. J. Mol.

Struct. (THEOCHEM) 1983, 91, 313.

(44) Mayer, P. M.; Taylor, M. S.; Wong, M. W.; Radom, L. J. Phys.

Chem. A 1998, 102, 7074.

(45) Bernardi, F.; Robb, M.; Schlegel, H. B.; Tonachini, G. J. Am. Chem.

Soc. 1984, 106, 1198.

(46) Jensen, J. H.; Morokuma, K.; Gordon, M. S. J. Chem. Phys. 1994,

100, 1981, and references therein.

(47) Kee, R. J.; Miller, J. A.; Jefferson, T. H. CHEMKIN: A General

Purpose, Problem Independent, Transportable FORTRAN Chemical Kinetics

Code Package. SANDIA National Laboratories Report SAND80-003; Sandia

National Laboratories: Los Alamos, NM March 1980.

(48) Mitchell, R. D.; Kee, R. J. A General Purpose Computer Code for

Predicting Chemical Kinetic Behavior behind Incident and Reflected Shocks.

SANDIA National Laboratories Report SAND82-8205; Sandia National

Laboratories: Los Alamos, NM March 1982.

(49) Hindmarsh, A. C. LSODE and LSODI: Two New Initial Value

Differential Equation Solvers. ACM Signum Newslett. 1980, 15, 4.

(50) Lutz, A. E.; Kee, R. J.; Miller, J. A. SENKIN: A FORTRAN

Program for Predicting Homogeneous Gas-Phase Chemical Kinetics with

Sensitivity Analysis. SANDIA National Laboratories Report SAND87-8248;

Sandia National Laboratories: Los Alamos, NM 1988.

(51) Muris, S.; Haynes, B. S. HSENKIN, Department of Chemical

Engineering, University of Sydney, NSW 2006, Australia.

(52) Bowman, C. T.; Hanson, R. K.; Davidson, D. F., Gardiner, W. C.;

Lissianski, V.; Smith, G. P.; Golden, D. M.; Frenklach, M.; Goldenberg,

M. http://www.me.berkeley.edu/gri_mech.

(53) Warnatz, J. In Combustion Chemistry; Gardiner, W. C., Jr., Ed.;

Springer: New York, 1984; p 197.

(54) Miller, J. A.; Bowman, C. T. Prog. Energy Combust. Sci. 1989,

15, 287.

(55) Kiefer, J. H.; Sidhu, S. S.; Kumaran, S. S.; Irdam, E. A. Chem.

Phys. Lett. 1989, 159, 32.

(56) Raghavachari, K.; Frisch, M. J.; Pople, J. A.; Schleyer, P. von R.

Chem. Phys. Lett. 1982, 85, 145.

(57) Organ, P. P. Ph.D. Thesis, University of Sydney, 1989.

(58) Curtiss, L. A.; Carpenter, J. E.; Raghavachari, K.; Pople, J. A. J.

Chem. Phys. 1992, 96, 9030.

(59) Kiefer, J. H.; Mitchell, K. I.; Kern, R. D.; Yong, J. N. J. Phys.

Chem. 1988, 92, 677.

(60) Weisssman, M. A.; Benson, S. W. J. Phys. Chem. 1988, 92, 4080.

(61) Payne, W. A.; Stief, L. J. J. Chem. Phys. 1976, 64, 1150.

(62) Mallard, W. G.; Westley, F.; Herron, J. T.; Hampson, R. G.; Frizzell,

D. H. NIST Chemical Kinetic Database, Version 6.0; National Institute of

Standards and Technology: Gaithersburg, MD, 1994.

(63) Halverson, F.; Stamm, R. F.; Whalen, J. J. J. Chem. Phys. 1948,

16, 808.

(64) Francisco, J. S.; Richardson, S. L. J. Chem. Phys. 1994, 101,

7707.

(65) Francisco, J. S. Chem. Phys. Lett. 1994, 230, 372.

(66) Burcat, A.; McBride, B. 1994 Ideal Gas Thermodynamic Data for

Combustion and Air Pollution Use. Technion Aerospace Eng. Rep. No.

697; Technion University: Haifa, Israel Dec 1993.

(29) SIRIUS is an MCSCF program written by H. J. A° . Jensen, H. Agren,

and J. Olsen.

(30) ABACUS is an MCSCF energy derivatives program written by T.

Helgaker, H. J. A° . Jensen, P. Jorgensen, J. Olsen, and P. R. Taylor.

(31) DALTON Release 1.0 1997 is an ab initio electronic structure

program, written by T. Helgaker, H. J. A° . Jensen, P. Jørgensen, J. Olsen,

K. Ruud, H. A° gren, T. Anderson, K. L. Bak, V. Bakken, O. Christiansen,

P. Dahle, E. K. Dalskov, T. Enevoldsen, B. Fernandez, H. Heiberg, H.

Hettema, D. Jonsson, S. Kirpekar, R. Kobayashi, H. Koch, K. V. Mikkelsen,

P. Norman, M. J. Packer, T. Saue, P. R. Taylor, and O. Vahtras.

(32) Andersson, K.; Blomberg, M. R. A.; Fu¨lscher, M. P.; Kello¨, V.;

Lindh, R.; Malmqvist, P.-A° .; Noga, J.; Olsen, J.; Roos, B. O.; Sadlej, A. J.;

Siegbahn, P. E. M.; Urban, M.; Widmark, P.-O. MOLCAS version 2;

University of Lund: Lund, Sweden, 1991.

(33) Andersson, K.; Blomberg, M. R. A.; Fu¨lscher, M. P.; Karlstro¨m,

G.; Lindh, R.; Malmqvist, P.-A° .; Neogra´dy, P.; Olsen, J.; Roos, B. O.; Sadlej,

A. J.; Schu¨tz, M.; Seijo, L.; Serrano-Andre´s, L.; Siegbahn, P. E. M.; P.-O.

Widmark, P.-O. MOLCAS Version 4; Lund University: Lund, Sweden,

1997.

(34) MOLPRO is a package of ab initio programs written by H.-J.

Werner and P. J. Knowles, with contributions from J. Almlo¨f, R. D. Amos,

M. J. O. Deegan, S. T. Elbert, C. Hampel, W. Meyer, K. Peterson, R. Pitzer,

A. J. Stone, P. R. Taylor, R. Lindh, M. E. Mura, and T. Thorsteinsson.

(35) Kee, R. J.; Rupley, F. M.; Miller, J. A. The Chemkin Thermody-

namic Data Base; Sandia Report SAND87-8215; Sandia National Labo-

ratories: Los Alamos, NM, 1987.

(36) An, X.-W.; Mansson, M. J. Chem. Thermodyn. 1983, 15, 287.

(37) King, K. D.; Goddard, R. D. Int. J. Chem. Kinet. 1975, 7, 837.

(38) Berkowitz, J.; Ellison, G. B.; Gutman, D. J. Phys. Chem. 1994,

98, 2744.

(39) Moran, S.; Ellis, H. B., Jr.; DeFrees, D. J.; McLean, A. D.; Ellison,

G. B. J. Am. Chem. Soc. 1987, 109, 5996.

(40) Holmes, J. L.; Mayer, P. M. J. Phys. Chem. 1995, 99, 1366.

Article Doi

Ab Initio Quantum Chemical and Experimental (Shock Tube) Studies of the Pyrolysis Kinetics of Acetonitrile

DOI: 10.1021/jp9835216

Source and publish data:

Authors:

Article abstract of DOI:10.1021/jp9835216

Full text of DOI:10.1021/jp9835216

Products guided by the article

R&D Labs maybe for 1070-71-9

Relevant to this article

Hot Product