Thermal decomposition of 2,2-bis(difluoroamino) propane studied by FTIR spectrometry and quantum chemical calculations: The primary dissociation kinetics and the mechanism for decomposition of the (CH3)2CNF2 radical

Article abstract of DOI:10.1016/S0040-6031(01)00781-X

The kinetics of the thermal decomposition of 2,2-bis(difluoroamino) propane (BDFP) has been studied by pyrolysis/FTIR spectrometry at temperatures between 528 and 553 K using toluene as radical scavenger. The disappearance of BDFP was found to follow the first-order kinetics with the rate constant, k₁ = 10^16.0±0.7 exp[-(24200 ± 840)/T] s^-1, which agrees closely with the expression obtained by Fokin et al. [Dokl. Akad. Nauk. 332 (1993) 735], k₁ = 10^15.60 exp(-23600/T) s^-1. The measured large A-factor supports the earlier conclusion that the primary fragmentation process corresponds to the breaking of one of the two NF₂ groups. The measured activation energy is also consistent with the predicted first C-N bond dissociation energy, 44-48 kcal/mol, by the hybrid density-functional theory and that evaluated by variational RRKM calculations fitting the observed rate constant, 47.9 kcal/mol. The 2-difluoroamino propyl radical, (CH₃)₂CNF₂, was predicted to be thermally unstable, producing readily F atoms and HF molecules via the (CH₃)₂CFNF intermediate. The quantum-chemically predicted mechanisms for the fragmentation of (CH₃)₂C(NF₂)₂ and (CH₃)₂CNF₂ agree with the product distribution reported by Ross and coworkers.

Full text of DOI:10.1016/S0040-6031(01)00781-X

Thermochimica Acta 384 (2002) 101–111
Thermal decomposition of 2,2-bis(difluoroamino) propane studied
by FTIR spectrometry and quantum chemical calculations:
the primary dissociation kinetics and the mechanism for
decomposition of the (CH3)2CNF2 radical
J. Parka, D. Chakrabortya, S. Jamindara, W.S. Xiaa, M.C. Lina,*, C. Bedfordb
aDepartment of Chemistry, Emory University, Atlanta, GA 30322, USA
bWeapons Division, Naval Air Warfare Center, China Lake, CA, USA
Abstract
The kinetics of the thermal decomposition of 2,2-bis(difluoroamino) propane (BDFP) has been studied by pyrolysis/FTIR
spectrometry at temperatures between 528 and 553 K using toluene as radical scavenger. The disappearance of BDFP was
found to follow the first-order kinetics with the rate constant, k1 ¼ 1016:0Æ0:7 exp½Àð24200 Æ 840Þ=TŠ sÀ1, which agrees
closely with the expression obtained by Fokin et al. [Dokl. Akad. Nauk. 332 (1993) 735], k1 ¼ 1015:60 expðÀ23600=TÞ sÀ1
.
The measured large A-factor supports the earlier conclusion that the primary fragmentation process corresponds to the
breaking of one of the two NF2 groups. The measured activation energy is also consistent with the predicted first C–N bond
dissociation energy, 44–48 kcal/mol, by the hybrid density-functional theory and that evaluated by variational RRKM
calculations fitting the observed rate constant, 47.9 kcal/mol. The 2-difluoroamino propyl radical, (CH3)2CNF2, was predicted
to be thermally unstable, producing readily F atoms and HF molecules via the (CH3)2CFNF intermediate. The quantum-
chemically predicted mechanisms for the fragmentation of (CH3)2C(NF2)2 and (CH3)2CNF2 agree with the product distribution
reported by Ross and coworkers [J. Am. Chem. Soc. 94 (1972) 8776]. # 2002 Elsevier Science B.V. All rights reserved.
Keywords: Thermal decomposition; 2,2-Bis(difluoroamino) propane; Propellants
1. Introduction
Replacing some –NO2 groups by –NF2, the difluor-
oamino group, it is possible to improve both the
amount and the rate of energy release in energetic
material systems such as propellants.
Recently Lane and Politzer [1] discussed the repla-
cement of some –NO2 groups by –NF2. They con-
cluded that the compound’s density and the number of
moles of gaseous combustion products formed per
gram of material can be increased, thereby enhancing
propellant performance. Thus, in designing energetic
molecules (i.e. explosives and propellants), it can be
advantageous to include some difluoroamine groups
along with the usual nitro group.
The oxidizing nitro group increases the density
of propellants and explosives and promotes the for-
mation of very stable gaseous decomposition products
(N2, CO, CO2 and H2O) through fast, exothermic
redox reactions, resulting in a large energy release
and pressure buildup per unit volume of material.
* Corresponding author. Tel.: þ1-404-727-2825;
fax: þ404-727-6586.
E-mail address: chemmcl@emory.edu (M.C. Lin).
0040-6031/02/$ – see front matter # 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 0 4 0 - 6 0 3 1 ( 0 1 ) 0 0 7 8 1 - X

102
J. Park et al. / Thermochimica Acta 384 (2002) 101–111
In order to understand the complex mechanism for
suspension of 17.6 g of difluoroamine in 30% oleum.
After 2 h, the reaction mixture was treated with toluene
andthe organic layerwasseparated. Theorganicextract
was distilled at atmospheric pressure, yielding approxi-
mately a 1:1 mixture of BDFP: toluene, bp 80–85 8C.
The proton and fluorine NMR spectra were consistent
with previously reported values [7].
For general information, an excellent description of
the experimental procedure and the reaction set-up of
difluoroamination by Chapman et al. [8] is available.
Difluoroamine was generated by acidic hydrolysis of
N,N-difluorourea, prepared by direct fluorination of
aqueous urea. [9,10] Since difluoroamine under cer-
tain conditions can be an extremely sensitive and
detonable material, [11–13] adequate shielding should
be used to contain a possible detonation.
the thermal decomposition of a potential energetic
compound, 4,4-bis(difluoroamino)-1-nitropiperidine,
[2] we consider the thermal decomposition of a model
compound of 2,2-bis(difluoroamino) propane (BDFP)
in the present study. Earlier theoretical studies based
on some mono-difluoroamines [3] showed that HF
elimination through a four-centered transition state is
energetically more favorable than the initial rupture of
an N–F bond. Also, when all a-hydrogens are replaced
by methyl groups, N–F dissociation energy increases
slightly but undesirable destabilization due to HF
elimination can be avoided. On the other hand, C–
N bond dissociation energies in some related systems
(containing a-hydrogen) were observed to be
ꢀ65 kcal/mol [1,4]. Thus, in the present model com-
pound, BDFP, HF elimination might not be favorable
due to the absence of a-hydrogens; instead, C–N or N–
F bond cleavage might be the rate-controlling process.
Two earlier experimental studies [5,6] on the
decomposition of BDFP indeed indicated that the
initial products of the reaction measured at 723 and
1023 K under low-pressure conditions derived primar-
ily from the first C–NF2 bond scission with 45 kcal/
mol activation energy, [5] which is in close agreement
with the value, 47 kcal/mol, reported by Fokin et al.
[6] based on the BDFP decay rate determination at
lower temperatures. These results will be referred to
later in the discussion on the kinetics and mechanism
of the BDFP decomposition reaction.
2.2. Kinetic measurements
The thermal decomposition of BDFP in toluene was
carried out in a 270 ml quartz reactor placed in a well-
insulated double-walled cylindrical furnace. The reac-
tion temperature was measured using a thermocouple
placed at the center of the reactor and connected to a
temperature controller, which maintained the tempera-
ture within 1 K during the experiment. The thermo-
couple had been calibrated at the freezing point and
boiling point of water before use. The reactor was
directly connected to a gas-handling system and to a
FTIR sample tube. Four different mixtures with vary-
ing amounts of BDFP, toluene and H2O diluted with
argon were used in the present studies. The reactions
were carried out at nine different temperatures in the
range of 528–553 K at a constant total pressure of
760 Torr, in order to minimize surface effects. Both
unpyrolyzed and pyrolyzed mixtures were analyzed by
FTIR spectrometry (Mattson Instruments Polaris Spec-
trometer), using 7 in. long and 1 in. o.d. sample cell
sealed with KCl windows. The analysis cell is directly
attached to the reaction system through 3/4 in. o.d.
Teflon tubing. After expansion from the reactor at
760 Torr initial pressure, the sample pressure was
typically ꢀ300 Torr. Accordingly, all calibrations were
performed with ꢀ300 Torr standard mixtures contain-
ing various amounts of BDFP. Since original samples
came with toluene, calibration of toluene was also
performed in order to obtain the exact concentrations
of BDFP in the reaction mixtures. The strong infrared
Our present objective is to examine the mechanism
for the initial decomposition of the bis-difluoroamino
compound and to evaluate the rate constant for the
unimolecular decomposition process, aided by quan-
tum and statistical-theory calculations, in order to
shed some light on the mechanism for the thermal
decomposition of larger bis-difluoroaminonitramine
propellants.
2. Experimental details
2.1. Preparation of BDFP
BDFP was prepared at NAWL, China Lake, CA,
using the procedure similar to that of Baum [7] for the
difluoroamination of acetone. Acetone (4.5 g,
0.078 mol) was added drop-wise with stirring to a cold

J. Park et al. / Thermochimica Acta 384 (2002) 101–111
103
Table 1
The experimental conditions and measured first-order rate
constants for the unimolecular decomposition of BDFP at various
temperatures
T (K)
Mixturea
k1 (sÀ1
)
528
533
536
538
541
543
543
548
550
553
II
I
1.1 Â 10À4
1.8 Â 10À4
2.1 Â 10À4
2.3 Â 10À4
3.4 Â 10À4
4.3 Â 10À4
4.1 Â 10À4
5.5 Â 10À4
7.1 Â 10À4
8.6 Â 10À4
III
II
III
I
IV
II
III
I
a Mixture: (I), 0.1306% of BDFP and 0.1489% of toluene in
argon; (II), 0.0544% of BDFP and 2.442% of toluene in argon;
(III), 0.0464% of BDFP and 2.081% of toluene in argon; (IV),
0.0458% of BDFP, 2.054% of toluene and 1.315% H2O in argon.
Fig. 1. Typical FTIR spectra of pyrolyzed (a) and unpyrolyzed (b)
BDFP in mixture I. HF spectra above 3800 cmÀ1 were expanded
for better visibility.
bands at 985 and 729 cmÀ1 were used to calibrate the
concentrations of the BDFP and toluene, respectively.
major products in this thermal decomposition reac-
tion. Effort to identify other organic products is under-
way with the help of quantum chemical calculations.
Fig. 2 shows the decay of BDFP concentration and
the rise of SiF4 peak intensity. Under the experimental
conditions, the BDFP decay data show that the rate of
its decomposition follows the first-order rate law.
With the first-order kinetics for the decomposition
in the gas phase,
3. Results and discussion
3.1. Pyrolytic study
The thermal decomposition of BDFP has been
studied with various mixtures at nine different tem-
peratures in the range 528–553 K while maintaining a
constant total pressure of 760 Torr. Fig. 1 presents a
typical absorption spectral profile of pyrolyzed and
unpyrolyzed samples and Table 1 summarizes our
experimental conditions and results. Two strong peaks
at 1030 and 713 cmÀ1 are found as products from the
pyrolyzed spectra by FTIR analysis and identified as
SiF4 and HCN, respectively. Complex spectra in the
region above 3800 cmÀ1 confirmed [14] the formation
of HF in the present decomposition reaction. Forma-
tion of SiF4 may be caused by the reaction of HF and
SiO2 (from reactor quartz wall). Since H2O is formed
by the reaction of HF and SiO2, a mixture of 0.0458%
BDFP in toluene with excess H2O (1.3%) is also
pyrolyzed to study the effect of H2O on the initiation
of the decomposition reaction. The addition of 1.3%
H2O eliminated SiF4 through hydrolysis, but was
found to have no effect on the reactant decay rate
(see Table 1). Thus, HF and HCN are potentially the
k1
A ! B þ C
Fig. 2. Typical concentration profiles of BDFP at 533 K (~),
543 K (*), 553 K (&) and rise of peak intensity of SiF4 (*) at
543 K using mixture I.

104
J. Park et al. / Thermochimica Acta 384 (2002) 101–111
Fig. 4. Arrhenius plot for the rate constants obtained under various
reaction conditions using (*) mixture I, (~) mixture II, (&)
mixture III, and (5) mixture IV. Solid line, predicted value from
Variflex calculation with E0 ¼ 47:9 kcal/mol; dotted and dashed
lines, predicted values from Variflex calculations with
E0 ¼ 47:9 Æ 0:1 kcal/mol; dash-dotted line, Ref. [6].
Fig. 3. First-order decay plots of BDFP at 528 K (*), 538 K (&)
and 548 K (*) using mixture II.
d½AŠ
thermal decomposition is mainly controlled by the
homolytic cleavage of one of the C–N bonds leading
to the formation of (CH3)2C(NF2) and NF2 radicals.
Since there is no a-hydrogen in BDFP, direct HF
elimination is not feasible (vide infra). The experi-
mental activation energy (48 kcal/mol) is much lower
than the C–N bond dissociation energy of ꢀ65 kcal/
mol observed in some related alkyl mono-difluoroa-
mines [1,4]. These result suggest that the presence of
the second NF2 group in this model compound sig-
nificantly lowers the dissociation energy of the first C–
NF2 bond.
Summarizing, the experimental kinetic result
clearly favors the C–N bond breaking as primary
decomposition step although formation of HF and
HCN is strongly evident from the FTIR spectra of
pyrolyzed BDFP. Their origins will be addressed
later.
¼ Àk1½AŠ;
dt
ln ½AŠ ¼ ln ½A0Š À k1t;
k1, the first-order rate constant, can be obtained from
the slope of ln [A]t versus time plot.
Fig. 3 illustrates the first-order decay plots for the
three different temperatures with mixture II. Table 1
summarizes the initial reaction conditions and the
measured rate constants.
The rate constants are also graphically shown in
Fig. 4 for the temperature range 528–553 K. A least-
squares analysis of all the data based on the BDFP
decay gave
t
ꢀ
ꢁ
Àð24200 Æ 840Þ
k1 ¼ 1016:0Æ0:7 exp
sÀ1
T
This result agrees closely with the expression reported
by Fokin et al. [6] k1 ¼ 1015:60 expðÀ23600=TÞ sÀ1
obtained in the temperature range 493–533 K at BDFP
pressures between 20 and 150 Torr in terms of both the
A-factor and the activation energy. Their result is also
included in Fig. 4 for comparison. Our activation
energy, 48:1 Æ 1:7 kcal/mol, also agrees reasonably
with the value obtained at two temperatures (923 and
1023 K) under low-pressure conditions by Ross et al.
3.2. Quantum chemical calculations
To elucidate the mechanism for the decomposition
of BDFP, we performed ab initio MO and hybrid
density-functional calculations. The following three
major primary decomposition channels have been
identified.
[5] 45 kcal/mol. The high A-factor (ꢀ1016 À1) clearly
s
supports the conclusion reached earlier [5,6] that the
BDFP ! ðCH3Þ2CNF2 þ NF2
(I)

J. Park et al. / Thermochimica Acta 384 (2002) 101–111
105
BDFP ! CH3ðCH2ÞCNF2 þ HNF2
BDFP ! cÀCH3CðCH2ÞðNF2ÞNðFÞÀ
(II)
bond and formation of C–N p bond. This metastable
radical, (CH3)2C=N(F)–F, has an elongated N–F bond
˚
length of 2.04 A and is 39.1 kcal/mol higher in energy
(III)
Reaction channel (I) involves homolytic cleavage of
one of the C–NF2 bonds to form (CH3)2CNF2 and NF2
radicals. The strong electronegative character of F
atoms might have a tendency to delocalize the electron
from C-center and hence destabilize the radical. In our
preliminary study carried out at the B3LYP level with
three different basis sets given in Tables 2 and 3, the
(CH3)2CNF2 radical was found to be unstable. Full
optimization of the radical led to an elongation of N–F
than BDFP (minus NF2) at the B3LYP/6-31G(d,p)
level of theory. The ‘‘transition state’’ TS1 corre-
sponding to the formation of this metastable radical
had a barrier of 54.6 kcal/mol with respect to BDFP at
the same level of theory. A further single-point energy
calculation at B3LYP/6-311þG(3df,2p) lowers the
energy to 50.5 kcal/mol. The resultant (CH3)2C=NF
molecule is quite stable and its further decomposition
leading to the formation of either CH3F þ CH3CN or
Table 2
Moments of inertia and vibrational frequencies calculated at the B3LYP/6-31G(d,p) level of theory
Species
BDFP
Ii (amu)
nI
764, 1455, 1520
76, 109, 184, 198, 206, 289, 292, 326, 336, 370, 447, 506, 518, 556, 679, 774, 823,
907, 925, 965, 985, 1014, 1030, 1046, 1222, 1245, 1273, 1413, 1430, 1496, 1503,
1511, 1524, 3088, 3090, 3172, 3176, 3176, 3176
TS2
TS3
805, 1902, 2189
876, 1510, 1696
À549, 44, 71, 93, 118, 136, 189, 259, 313, 386, 414, 461, 517, 537, 573, 622, 737,
764, 884, 902, 947, 1009, 1016, 1049, 1058, 1215, 1306, 1401, 1425, 1452, 1480,
1490, 1514, 1720, 3035, 3116, 3119, 3164, 3211
À1433, 69, 104, 127, 158, 222, 237, 253, 306, 341, 365, 430, 452, 489, 545, 601,
649, 730, 820, 907, 922, 972, 993, 1003, 1051, 1096, 1202, 1232, 1262, 1391, 1474,
1486, 1498, 1515, 3080, 3145, 3170, 3187, 3236
(CH3)2CNF2
401, 531, 886
370, 626, 659
514, 1125, 1383
À34, 85, 89, 235, 287, 367, 418, 462, 502, 758, 853, 872, 964, 986, 996, 1076, 1266,
1299, 1410, 1425, 1479, 1493, 1495, 1508, 2989, 2994, 3099, 3101, 3156, 3158
CH3C(CH2)NF2
À167, 107, 255, 291, 379, 466, 580, 624, 759, 835, 870, 936, 961, 987, 1038, 1062,
1286, 1418, 1443, 1498, 1503, 1744, 3066, 3127, 3173, 3177, 3271
c–CH3C(CH2)(NF2)N(F)–
76, 148, 184, 251, 290, 328, 354, 472, 478, 522, 615, 717, 829, 862, 907, 932, 949,
1013, 1042, 1098, 1104, 1140, 1266, 1376, 1428, 1477, 1498, 1514, 3077, 3151,
3160, 3185, 3260
CH3C(CH2)NF
(CH3)2CNF
185, 446, 619
210, 450, 638
96, 243, 328, 384, 481, 536, 574, 795, 882, 970, 993, 1048, 1060, 1329, 1368, 1425,
1498, 1502, 1503, 3058, 3124, 3154, 3191, 3303
45, 189, 304, 328, 367, 471, 583, 830, 932, 961, 984, 1094, 1115, 1293, 1412, 1422,
1482, 1487, 1488, 1510, 1726, 3045, 3054, 3103, 3107, 3157, 3196
NF2
26, 154, 180
34, 168, 196
0, 3, 3
565, 980, 1119
HNF2
HF
492, 918, 1014, 1322, 1445, 3341
4085
(CH3)2C
43, 195, 217
218, 260, 428, 668, 749, 932, 1016, 1159, 1198, 1342, 1350, 1406, 1412, 1505,
1512, 2928, 2932, 2985, 2986, 3098, 3100
(CH3)2CH(NF2)
(CH3)2CH
393, 552, 878
48, 218, 243
104, 196, 230, 260, 347, 394, 457, 475, 563, 829, 902, 911, 950, 972, 1003, 1154,
1178, 1208, 1357, 1367, 1415, 1433, 1496, 1509, 1509, 1533, 3060, 3060, 3065,
3131, 3137, 3158, 3158
108, 125, 361, 410, 888, 942, 946, 1038, 1154, 1184, 1376, 1423, 1427, 1483, 1493,
1494, 1505, 2952, 2958, 3039, 3041, 3103, 3104, 3179

106
J. Park et al. / Thermochimica Acta 384 (2002) 101–111
Table 3
Calculated relative energies with ZPE corrections at different levels of theory based on the optimized geometries with B3LYP/6-31G(d,p) for
BDFP decomposition
Species
6-31G(d,p)
6-311G(d,p)
6-311þG(3df,2p)
BDFP
0.00
61.90
78.51
48.58
24.25
À4.40
0.00
59.40
0.00
57.13
TS2
TS3
76.84
46.50
75.55
43.40
NF2 þ (CH3)2CNF2
HNF2 þ CH3C(CH2)NF2
HF þ CH3C(CH2)(NF2)NF
18.95
14.23
À11.18
À17.23
cyclic–N=C(CH3)CH2– (2-methylazirineÞ þ HF needs
to overcome a substantial barrier. The calculated barrier
at the B3LYP/6-31G(d,p) level of theory was found to
be 81.5 and 69.6 kcal/mol, respectively.
the rate-limiting step for the unimolecular decomposi-
tion of BDFP. A variational RRKM calculation of rate
constant for this channel (pathway I) must be per-
formed to account for the experimental results as will
be discussed below. The formation of HF in the
reaction may be attributed to H abstraction by F atoms
and/or direct elimination from the (CH3)2CNF2 radi-
cal as will also be discussed later. HF can further
undergo reaction with the quartz wall of the reactor to
form SiF4 as mentioned above. The experimentally
observed HCN might result from the secondary
decomposition of the (CH3)2CNF2 radical, as also
concluded by Ross et al [5].
Although the (CH3)2CNF2 radical is unstable
according to the B3LYP method, it is stable by both
Hartree-Fock and MP2 calculations. The calculated
C–N bond dissociation energy for the first NF2 at the
B3LYP/6-311þG(3df,2p) level of theory using the
geometry optimized at MP2/6-31G(d,p) is 43.4 kcal/
mol, which is slightly lower than the values, 48.6 and
46.5 kcal/mol, computed with smaller basis sets 6-
31G(d,p) and 6-311G(d,p), respectively. These results
are in good agreement with the experimental activa-
tion energies, 45–48 kcal/mol mentioned above. A
further evaluation of the first C–N bond energy will
be made later in conjunction with the rate constant
calculation with a variational RRKM theory.
Reaction path (II) involves the elimination of HNF2
from BDFP through a four-member-ring transition state
(TS2). The barrier at TS2 is 57.1 kcal/mol with respect
to BDFP at the B3LYP/6-311þG(3df,2p)//B3LYP/6-
31G(d,p) level of theory. The substantially higher
barrier for this process clearly suggests that it cannot
be the primary decomposition pathway for BDFP.
A similarly high-energy barrier is associated with the
reaction path (III), the direct HF elimination from a
methyl group and the formation of a three-member-ring
cyclic intermediate. The calculated barrier at TS3 is
75.6 kcal/mol at the same B3LYP/6-311þG(3df,2p)//
B3LYP/6-31G(d,p) level. Thus, reaction path (III) is
alsounlikelytobeanimportantdecompositionpathway
for BDFP. The structures of BDFP, all the TSs and other
local minima, and the potential energy diagram are
shown in Figs. 5 and 6, respectively.
3.3. Varaitional RRKM calculation for k1
As the unimolecular decomposition of BDFP pro-
ducing NF2 and (CH3)2CNF2 does not have a well
defined transition state, we have calculated its rate
constant using the Variflex code of Klippentein and
coworkers [15–18]. The potential energy profile
required to describe the interaction of the separating
fragments was calculated by geometry optimization at
different fixed C–N separations in BDFP, covering
˚
from 1.6 to 3.4 A with an interval of 0.1 or 0.2 A
˚
with the smaller interval at the critical separation
˚
2.8–3.4 A. For each structure, we calculated 3NÀ7
vibrational frequencies projected out of the gradient
direction. The calculated potential energies at the
B3LYP/6-31G(d,p) level of theory were fitted to the
Morse potential with b ¼ 2:63 and D ¼ 48 kcal/mol
fortheE,J-resolvedRRKMcalculationwiththeVariflex
code [18]. The dissociation energy can be further
refined by comparing the predicted rate constant with
the experimental result as described below.
Our theoretical calculation supports the experimen-
tal observation of the C–N homolytic bond cleavage as
The rate constant for the decomposition of BDFP
predicted by the variational RRKM calculation is

J. Park et al. / Thermochimica Acta 384 (2002) 101–111
107
Fig. 5. B3LYP/6-31G(d,p) optimized geometries for the species involved in the decomposition of BDFP and (CH3)2CNF2. For simplicity only
important parameters are labeled.
shown in Fig. 4 for comparison with the experimen-
tally measured values. As the predicted value depends
very strongly on the dissociation energy, which is the
least reliable quantity predictable by the B3LYP
method (varying from 43.4 to 48.6 kcal/mol), we
adjusted the bond dissociation energy initially within
the 48 Æ 1 kcal/mol range. Final calculations allowed
us to bracket the energy within a rather small range of
47:9 Æ 0:1 kcal/mol, as illustrated in Fig. 4. The pre-
dicted values within the energy range fully encom-
passed the convoluted experimental errors. The solid
line presented in the figure allows us to cover a much

108
J. Park et al. / Thermochimica Acta 384 (2002) 101–111
Fig. 6. Calculated potential energy profiles with B3LYP/6-31G(d,p) for the decomposition of BDFP.
wider temperature range (500–2000 K) than can be
covered experimentally (493–553 K, including that of
Fokin et al. [6]; the theoretical result can be repre-
sented by k1 ¼ 1014:09 expðÀ21860=TÞ sÀ1. The sig-
nificant lowering of the apparent activation energy
(from 47.9 to 43.4 kcal/mol) and the associated A-
factor results from the effect of angular momentum
conservation (i.e. J-dependence) in the dissociation
process at higher temperatures.
barrier at TS10 (ni ¼ À825 cmÀ1). The radical can
also eliminate to give HF þ CH3(CH2)NF exothermi-
cally (DH ¼ À29 kcal/mol) with a small barrier of
16 kcal/mol via a five-centered transition state TS20
(ni ¼ À2338 cmÀ1). The small barriers for these
unimolecular decomposition reactions are responsible
for the thermal instability of the (CH3)2CNF2 radical.
On the other hand, (CH3)2CNF2 decomposes to give
F þ ðCH3Þ2CNF with an unreasonably large 62 kcal/
mol barrier (TS30, ni ¼ À316 cmÀ1) and 15 kcal/mol
exothermicity. It should be noted, however, that the
corresponding transition energies calculated by
B3LYP//MP2/3-21G(d,p) with larger basis sets are
all significantly lower as shown in Table 4.
The decomposition of (CH3)2CFNF giving HF þ
CH3ðCH2ÞCNF via a four-centered transition state
(TS40, ni ¼ À2238 cmÀ1) occurs with a 64 kcal/mol
barrieror,equivalently,15 kcal/molabove(CH3)2CNF2,
comparable to that of the five-centered TS20. Since
we were unable to find the transition states for the
decomposition or isomerization of the (CH3)2CFNF
giving F þ ðCH3Þ2CNF at the MP2/3-21G(d,p) level,
we obtained its transition state geometry at MP2/3-
21G(d) with additional single-point energy calcula-
tions using B3LYP with larger basis sets as before.
The result shows that (CH3)2CFNF decomposes more
3.4. Mechanism for (CH3)2CNF2 decomposition
The (CH3)2CNF2 radical is thermochemically
unstable. Since we were unable to find the transition
states for the decomposition and isomerization of the
(CH3)2CNF2 radical by B3LYP or MP2 calculations
with basis sets greater than 3-21G(d,p), we studied its
reaction mechanism with the geometry optimized at
MP2/3-21G(d,p), with additional single-point energy
calculations using B3LYP with larger basis sets. The
optimized geometry and potential energy profile com-
puted at the MP2/3-21G(d,p) level of theory are shown
in Figs. 7 and 8, respectively, and the predicted relative
energies are summarized in Table 4.
At the MP2/3-21G(d,p) level, (CH3)2CNF2 iso-
merizes to (CH3)2CFNF with a small (7 kcal/mol)

J. Park et al. / Thermochimica Acta 384 (2002) 101–111
109
Fig. 7. MP2/3-21G(d,p) optimized geometries for the species involved in the decomposition (CH3)2CNF2. For simplicity only important
parameters are labeled.
Table 4
Calculated relative energies with ZPE corrections at different levels of theory based on the optimized geometries with MP2/3-21G(d,p) for
(CH3)2CNF2 radical decomposition
Species
MP2/3-21G(d,p)
B3LYP/6-31G(d,p)
B3LYP/6-311G(d,p)
B3LYP/6-311þG(3df,2p)
(CH3)2CNF2
TS10
TS20
0.00
6.99
0.00
À6.93
5.87
0.00
À9.06
3.50
0.00
À9.78
3.51
15.99
TS30
TS40
TS50
62.08
20.96
5.66
17.85
19.64
15.48
3.39
3.14
a
À4.07
À48.66
78.71
À16.33
À47.23
69.00
0.82
À20.11
À47.47
66.83
À25.39
À47.43
71.28
(CH3)2CFNF
NF2 þ (CH3)2C
F þ (CH3)2CNF
HF þ CH3C(CH2)NF
À15.11
À3.39
À2.84
À29.37
À38.30
À45.25
À49.90
a The relative energies were calculated at different levels of theory based on the optimized geometries with MP2/3-21G(d).

110
J. Park et al. / Thermochimica Acta 384 (2002) 101–111
Fig. 8. Calculated potential energy profiles with MP2/3-21G(d,p) for the decomposition of (CH3)2CNF2.
readily to give F þ ðCH3Þ2CNF with a 44 kcal/mol
barrier via TS50 (ni ¼ À643 cmÀ1), which lies 4 kcal/
mol below (CH3)2CNF2 and 66 kcal/mol below TS30
producing the same products. It is worth noting that
a corresponding calculation for TS50 by B3LYP//MP2/
3-21G(d) shows no distinct saddle point with a higher
endothermicity of ꢀ45 kcal/mol (above (CH3)2CFNF).
These qualitative results show that (CH3)2CNF2
is thermally less stable than its isomer, (CH3)2CFNF;
their decomposition reactions produce F and HF
exothermically with respect to the former. The ready
formation of F and HF is consistent with the exp-
erimental observation of SiF4 as alluded to before.
The HCN absorption peak detected at 713 cmÀ1 in
the FTIR spectrum may derive from further decomposi-
tion of CH3C(CH2)NF producing CH2CH þ HFþ
HCN, for example. The reaction is estimated to
be exothermic by about 3 kcal/mol at the MP2/3-
21G(d) level of theory. In addition, HCN could also
be formed by the reaction of NF2 with CH3, which
couldbeproducedbythefragmentationof(CH3)2CFNF
[5].
3.5. The C–N bond energy in (CH3)2C(H)NF2
The C–N bond energies in several single NF2-sub-
stituted alkanes have been estimated to lie in the range
of 62–69 kcal/mol by Sanders and Lin [4] based on the
kinetics of NF2 þ olefin reactions. These values are
somewhat higher than those predicted by our B3LYP
calculations for (CH3)2C(H)NF2, 57.1, 55.8 and
54.7 kcal/mol, using 6-31G(d,p), 6-311G(d,p) and 6-
311 þ G(3df,2p) basis sets, respectively. Comparison
of these values with those given in Table 3 for the first
C–N bond predicted with the corresponding methods
suggests that the presence of the second NF2 group in
the same carbon atom weakens the first C–N bond by
about 10 kcal/mol, attributable to the strong electron
withdrawing effect of the second group.
4. Conclusion
The results of the thermal decomposition of BDFP
using the technique of pyrolysis/FTIR spectrometry
indicate for the first time that the breaking of one of the
C–NF2 bonds in this bis-difluoroamino compound is
the rate-limiting step. The following first-order rate
expression was obtained for the unimolecular decom-
position reaction:
The production of NF2 and (CH3)2C: diradical from
(CH3)2CNF2, however, was predicted to be very
endothermic (see Table 4), as one would expect. This
product channel is not expected to be significant to the
overall BDFP pyrolytic chemistry. The results of the
quantum chemical calculations as described qualita-
tively agree with the product distribution reported by
Ross et al. [5].
ꢀ
ꢁ
ð24200 Æ 840Þ
k1 ¼ 1016:0Æ0:7 exp À
sÀ1
T

J. Park et al. / Thermochimica Acta 384 (2002) 101–111
111
Over the narrow temperature range of 528–553 K. The
result agrees excellently with the value reported by
Fokin et al. [6], k1 ¼ 1015:60 expðÀ23600=TÞ sÀ1
obtained in the overlapping temperature range, 493–
533 K. A variational RRKM (Varilflex) calculation for
the rate constant, using the potential energy function
computedfortheinteractionoftheNF2and(CH3)2CNF2
fragments at the B3LYP/6-31G(d,p) level of theory,
provides a quantitative agreement with the experi-
mental rate constant as well as the first C–N bond
dissociationenergy,D(C–NF2Þ ¼ 47:9 Æ 0:1 kcal/mol.
The (CH3)2CNF2 radical has been shown to be
unstable; it can decompose readily without significant
barriers to give F atoms and HF molecules. The
theoretically predicted mechanisms for the decompo-
sition of (CH3)2C(NF2)2 and (CH3)2CNF2 agree rea-
sonably with the product distribution reported by Ross
et al. [5] based on their low-pressure, high-tempera-
ture pyrolysis data.
Dr. Richard S. Miller for his unwavering support over
many years.
References
[1] P. Lane, P. Politzer, Adv. Mol. Struct. Res. 3 (1997) 269.
[2] J.C. Oxley, J.L. Smith, J. Zhang, C. Bedford, J. Phys. Chem.
A. 105 (2001) 579.
[3] E.M. Grice, P. Lane, P. Politzer, J. Mol. Struct. 365 (1996) 89.
[4] W.A. Sanders, M.C. Lin, J. Chem. Soc. Faraday Trans. 83
(1987) 905.
[5] D.S. Ross, T. Mill, M.B. Hill, J. Am. Chem. Soc. 94 (1972)
8776.
[6] A.V. Fokin, V.N. Grebennikov, G.B. Manelis, G.M. Nazin,
Dokl. Akad. Nauk 332 (1993) 735.
[7] K.J. Baum, Am. Chem. Soc. 90 (1968) 7083.
[8] R.D. Chapman, M.F. Welker, C.B. Kreutzberger, J. Org.
Chem. 63 (1998) 1566.
[9] W.H. Graham, J.P. Freeman, J. Org. Chem. 34 (1969) 2589.
[10] A.V. Fokin, Yu.M. Kosyrev, V.A. Makarov, N.P. Novoselov,
Doklady Chem. (Eng. Transl.) 186 (1969) 350.
[11] E.A. Lawton, J.Q. Weber, J. Am. Chem. Soc. 85 (1963) 3595.
[12] C.O. Parker, J.P. Freeman, Inorg. Synth. 12 (1970) 307.
[13] K.O. Christe, R.D. Wilson, Inorg. Chem. 26 (1987) 920.
[14] M.T. Bowers, G.I. Kerley, W.H. Flygare, J. Chem. Phys. 45
(1966) 3399.
Acknowledgements
[15] S.J. Klippenstein, Chem. Phys. Letts. 170 (1990) 71.
[16] S.J. Klippenstein, J. Chem. Phys. 94 (1991) 6469.
[17] S.J. Klippenstein, J. Chem. Phys. 96 (1992) 367.
[18] S.J. Klippenstein, S.H. Robertson, A.F. Wagner, D.M.
Wardlaw, Faraday Disc. Chem. Soc. 102 (1995) 65.
The authors are grateful to the Office of Naval
Research for support of this work, under the direction
of Dr. J. Goldwasser through ONR grant no. N00014-
89-J-1949. We want to take this opportunity to thank