Thermal stability and kinetics of ylide-borane complexes

Article abstract of DOI:10.1016/j.tca.2004.05.024

Complexes of dimethylsulfoxonium methylide (1) and organoboranes are crystalline for ylide·BH₃ (2), ylide·BPh₃ (3), ylide·B(C₆F₅)₃ (4), and ylide·BF₃ (5). These complexes undergo exothermic rearrangement by 1,2-migration upon heating to produce homologated organoboranes and dimethylsulfoxide. Non-isothermal kinetic analysis of the differential scanning calorimetry (DSC) data for ylide·BPh₃ (3) and ylide·B(C₆F₅)₃ (4) complexes was applied using the Flynn-Wall-Ozawa and Kissinger methods. The calculated apparent activation energy for the reaction of ylide·BPh₃ (3) yielded consistent results between the A_1.5 model (E_a=120 kJ mol^-1, A=4.79×10¹³ min^-1) and Kissinger method (E_a=129 kJ mol^-1, A=1.73×10¹⁷ min^-1). The analysis for the reaction of ylide·B(C ₆F₅)₃ (4) gave consistent results between R₂, R₃, and F₁ models with the average parameters, E_a=262 kJ mol^-1, A=3.33×10³³ min^-1. The Kissinger analysis for the reaction of ylide·B(C₆F₅)₃ (4) gave Arrhenius activation parameters (E_a=171 kJ mol^-1, A=7.80×10¹⁹ min^-1) that were higher than for the reaction of ylide·BPh₃ (3). The kinetic data revealed that the C₆F₅ electron-deficient group has a higher activation energy for 1,2-migration and a higher entropy of activation for 1,2-migration than the C₆H₅ group. HF/6-31G(d) ab initio calculations agree with the kinetic data.

Full text of DOI:10.1016/j.tca.2004.05.024

Thermochimica Acta 424 (2004) 149–155
Thermal stability and kinetics of ylide-borane complexes
Jonathan M. Stoddard, Kenneth J. Shea∗
Department of Chemistry, University of California, Irvine, CA 92612-2025, USA
Received 3 September 2003; received in revised form 21 May 2004; accepted 24 May 2004
Available online 19 July 2004
Abstract
Complexes of dimethylsulfoxonium methylide (1) and organoboranes are crystalline for ylide·BH3 (2), ylide·BPh3 (3), ylide·B(C6F5)3 (4),
and ylide·BF3 (5). These complexes undergo exothermic rearrangement by 1,2-migration upon heating to produce homologated organob-
oranes and dimethylsulfoxide. Non-isothermal kinetic analysis of the differential scanning calorimetry (DSC) data for ylide·BPh3 (3) and
ylide·B(C6F5)3 (4) complexes was applied using the Flynn–Wall–Ozawa and Kissinger methods. The calculated apparent activation energy for
the reaction of ylide·BPh3 (3) yielded consistent results between the A1.5 model (Ea = 120 kJ mol−1, A = 4.79 × 1013 min−1) and Kissinger
method (Ea = 129 kJ mol−1, A = 1.73 × 1017 min−1). The analysis for the reaction of ylide·B(C6F5)3 (4) gave consistent results between
R2, R3, and F1 models with the average parameters, Ea = 262 kJ mol−1, A = 3.33 × 1033 min−1. The Kissinger analysis for the reaction of
ylide·B(C6F5)3 (4) gave Arrhenius activation parameters (Ea = 171 kJ mol−1, A = 7.80 × 1019 min−1) that were higher than for the reaction
of ylide·BPh3 (3). The kinetic data revealed that the C6F5 electron-deficient group has a higher activation energy for 1,2-migration and a
higher entropy of activation for 1,2-migration than the C6H5 group. HF/6-31G(d) ab initio calculations agree with the kinetic data.
© 2004 Elsevier B.V. All rights reserved.
Keywords: DSC; Non-isothermal; Kissinger kinetics; Homologation
1. Introduction
thought that a suitable group may be developed that does not
undergo 1,2-migration, but allows the other two alkyl groups
to undergo 1,2-migration. This is termed a blocking group
when polyhomologation occurs selectively along the other
alkyl chains. Such a catalyst would allow the synthesis of
novel polymethylene architectures and copolymers. Second,
the rate of 1,2-phenyl migration from Ph3B, for example, is
directly related to the rate of initiation, kin. In “living” poly-
merizations, the rate of initiation must be greater than or
Trialkylboranes react with the ylide dimethylsulfoxo-
nium methylide (1) to produce homologated boranes and
dimethylsulfoxide (DMSO) (Scheme 1). In the presence of
excess ylide, repetitive homologation results in the forma-
tion of polymethylene [1]. The solution kinetics of triethylb-
orane with excess ylide was analyzed to obtain the reaction
order and rate of propagation [1]. The rate of initiation (ethyl
migration) of alkylboranes has been assumed to be identical
with the rate of propagation (n-alkyl migration). The rates
of initiation of other organoboranes such as Ph3B (phenyl
migration) and BH3 (hydride migration) are of interest.
There are two reasons for developing a method to mea-
sure the rates of initiation. First, different Lewis acids used
for the polyhomologation reaction exhibit significant dif-
ferences in reactivity of the borane-ylide complex. These
differences in reactivity arise due to differences in the acti-
vation energy of the group undergoing 1,2-migration. It was
equal to the rate of propagation, kprop. This criterion, kin
>
kprop, is necessary for the simultaneous start of the poly-
merization reaction that leads to a narrow molecular weight
distribution. However, the rates of initiation cannot be de-
termined by analyzing the kinetics with excess ylide. This
report involves the development of a method to measure the
rates of initiation directly from the ylide-borane complex.
The initiation of the polyhomologation reaction involves
the injection of a small amount of trialkylborane into an
excess amount of preheated ylide in toluene. Reaction tem-
peratures from 70 to 80 ◦C are often used to ensure the sol-
ubilization of the developing tris-polymethylene organob-
orane. Often, an exothermic reaction is observed upon
the addition of the initiator, as evidenced by the refluxing
∗
Corresponding author.
E-mail address: kjshea@uci.edu (K.J. Shea).
0040-6031/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.tca.2004.05.024

150
J.M. Stoddard, K.J. Shea / Thermochimica Acta 424 (2004) 149–155
O
H
k2
k1
S
H
CH3
CH3
Et3B
Et2BCH2Et
CH2S(O)Me2
+
+
DMSO
k-1
B
Et
Et
Et
Scheme 1. Sequence of homologation reaction.
solvent, toluene (bp = 110 ◦C). This exothermicity is likely
due to the enthalpic change associated with 1,2-migration.
The synthesis, isolation, and characterization [2] of the
complexes between dimethylsulfoxonium methylide (1) and
BH3, BPh3, BF3, and B(C6F5)3 have been previously re-
ported. In solution, ylide·BH3 (2) and ylide·BPh3 (3) un-
dergo spontaneous rearrangement via 1,2-migration at 0 ◦C,
whereas ylide·B(C6F5)3 (4) and ylide·BF3 (5) are unreac-
tive at room temperature. However, all these complexes can
be prepared as crystalline solids that are stable at room tem-
perature. It was thought that the 1,2-rearrangement of these
complexes may occur upon heating from either the solid
state or the melt to afford homologated reaction products.
Differential scanning calorimetry (DSC) was used to obtain
kinetic data for the solid-state reaction of complexes 3 and 4.
The Arrhenius activation parameters for solution-phase
reactions are obtained with the Van’t Hoff plot method
when the order of reaction is known. Reaction rates at
various temperatures are used to generate an Arrhenius
plot to obtain activation parameters. DSC non-isothermal
kinetics differs in that all temperatures are probed during
a single heating experiment. However, DSC data must be
analyzed by finding the appropriate topochemical model for
the reaction [3,4]. The appropriateness of this model can be
established graphically, by plotting log g(α)β versus 1/T,
and testing for the generation of linear plots. The variables,
α and β refer to the degree of conversion and heating rate
(◦C min−1), respectively, where g(α) is related to the model
of reaction (see below). From these linear plots, one can
calculate the Arrhenius activation parameters.
2.5, 2.0, 1.5, and 1.0 ◦C min−1 for complex 2 and a data
collection rate of 10 point s−1
.
2.1. Computational methods
The ylide·B(C6F5)3 and ylide·BPh3 ground-state com-
plexes were optimized at HF/3-21G(d) and HF/6-31G(d) us-
ing the gaussian-98 software package [6]. Transition state
calculations were carried out by optimizing the structures
with a constrained C1–C2 bond at 2.2 Å and C2–S bond
2.5 Å with HF/3-21G(d). This starting geometry was used
to locate the transition states by optimizing at HF/3-21G(d)
and HF/6-31G(d) without constraints.
3. Results
An overlay of the DSC scans for complexes 2–5 at a heat-
ing rate of 10 ◦C min−1 is shown in Fig. 1. All scans show
exothermic peaks. It was independently established that the
products from complexes 2–4 are homologated organobo-
ranes and dimethylsulfoxide. The rearrangement products
of complex 5 that would occur from 1,2-fluorine migration
could not be established. The exothermic reaction has been
assigned to the 1,2-migration that occurs in the solid state
for complexes 2–4. It is straightforward to obtain kinetic
data from DSC analyses with different heating rates. How-
ever, applying an appropriate model to obtain meaningful
activation energies of solid-state reactions is more difficult
[7]. Although the intent of this investigation is to obtain
the relative migratory aptitudes and activation energies for
the 1,2-migration step for complexes 2–4, Arrhenius acti-
vation energies determined in the solid state do not cor-
relate with those determined in homogeneous solution (or
the gas phase). For example, induction delay, autocatalysis,
and topochemical behavior is observed that may correspond
to various modes of nuclei formation and rate acceleration
[8,9]. More importantly, organic solids are often preceded
or accompanied by melting upon reaction and two differ-
ent rates of reaction, one from the melt, and one from the
solid may complicate the kinetic analysis [10]. The follow-
ing equation can be used to describe the kinetics of decom-
position for both solid-phase and liquid-phase reactions:
2. Experimental
The synthesis of ylide·BH3 (2), ylide·BPh3 (3), ylide·
B(C6F5)3 (4), and ylide·BF3 (5) as air-stable, crystalline
complexes was previously described [2]. DSC curves were
recorded on a DuPont 910 DSC instrument with Al2O3 as
an internal reference standard with crimped aluminum pans
(TA Instruments Part Nos. 900793.901 and 900794.901) un-
der an atmosphere of N2 (15 cm3 s−1 flow rate). Sample
masses between 8 and 10 mg of complexes 3–5 were used
with a punched pinhole in the top of the pan for venting. Due
to the extreme exothermicity of complex 2, sample weights
less than 1.5 mg and three punched pinholes in the sample
pan were necessary to mitigate violent decomposition [5].
Scans between 25 and 400 ◦C were obtained at scan rates of
10, 7.5, 5.0, and 2.5 ◦C min−1 for complexes 3–5 and 5.0,
ꢀ
ꢁ
dα
dt
E
R
= A exp −
f(α)
(1)
where α is the degree of conversion, f(α) is the kinetic
expression that depends on the reaction model [9], E

J.M. Stoddard, K.J. Shea / Thermochimica Acta 424 (2004) 149–155
151
–1
Fig. 1. DSC traces of ylide-borane complexes at 10 ◦C min and arbitrary y-offset.
the apparent activation energy (J mol−1), R the gas con-
stant (J mol−1 K−1), and A the Arrhenius frequency factor
(time−1).
possible. The peak width for the ylide·BH3 (2) complex
is narrow (30–60 s), even at heating rates of 1.0 ◦C min−1
.
Since the data are obtained at a rate of 10 points min−1, the
peak is described by 5–10 data points. In addition, there is an
ill-defined baseline for this complex as well (Fig. 1). These
two factors led to inaccurate integration of the exothermic
peak. Furthermore, upon reaction, significant amounts of
gaseous reaction products are produced that oftentimes pro-
pelled the sample pan off the instrument’s metal contacts.
At scan rates of 10 ◦C min−1, an audible pop was heard that
corresponded with the separation of the two crimped sample
pans, despite the presence of a large vent hole [5].
Diffusion models (D1, D2, D3), order of reaction models
(F1, F2), geometrical models (R2, R3), and random nucle-
ation models (A1.5, A2, A3) were all considered as f(α) for
complexes 2–5. The kinetic model function g(α) and the
relationship to temperature is expressed in a form that en-
ables a master curve to be generated from a plot of all the
non-isothermal data. In order to find the model that best-fits
the experimental data, we applied the isoconversional inte-
gral method developed by Flynn and Wall [11] and Ozawa
[12,13] with the Doyle approximation [14,15] in the form:
Heating a small amount of ylide·BH3 (2) in an open
micro-capillary to the onset of decomposition, for exam-
ple, emits a green flash as pyrophoric gases, presumably
methylborane, ignite in air while solid sample is thrusted
halfway up inside the capillary. A clear, colorless, oily
residue, presumably dimethylsulfoxide, is left behind along
the length of the capillary. The DSC accessories required
for gas-evolving samples were not available to us at the
time and we opted not to pursue optimizing the data acqui-
sition of this complex due to its rapid and very exothermic
(ꢀH = −230 kJ mol−1) reaction upon heating.
ꢂ
ꢀ
ꢁ
ꢃ
AE
E
log g(α)β = log
− 2.315 − 0.4567
(2)
R
RT
where
ꢄ
α
dα
g(α) =
(3)
f(a)
0
where β is the heating rate (◦C min−1) and T the absolute
temperature (K). This method is referred to as the Doyle
equation throughout this paper. For each ylide-borane com-
plex, a graph of log g(α)β versus 1/T was constructed over
0.1 < α < 0.95 for the composite DSC scans of all heat-
ing rates. The degree of conversion α was obtained from the
DSC trace by integration of the partial peak area at different
times. A linear plot with a high correlation coefficient (r) is
used as the criterion to find the best model for the decom-
position kinetics.
The plots of log g(α)β versus 1/T for ylide·BF3 (5) and
ylide·BH3 (2) complexes exhibited nonlinearity with every
model tested. The ylide·BF3 (5) complex exhibits multiple,
exothermic peaks characteristic of multiple transformations.
The accurate integration of these overlapping peaks was not
The data for the kinetic reactions of ylide·BPh3 (3) and
ylide·B(C6F5)3 (4) complexes generated a number of linear
plots and these two complexes will be discussed in detail.
The fraction remaining (1 − α) is easily obtained from the
integration of the DSC peak areas. The total heat released
(Hrxn) from the exothermic 1,2-migration and the partial
heat released at a particular time (Ht) are related to α, since
α = Ht/Hrxn. The graphs of fraction remaining (1 − α) ver-
sus temperature (◦C) for ylide·BPh3 (3) and ylide·B(C6F5)3
(4) complexes are shown in Fig. 2. These values of α and
T were used with Eq. (2) with each kinetic model [9] and a
regression line was obtained for the composite data at all β.

152
J.M. Stoddard, K.J. Shea / Thermochimica Acta 424 (2004) 149–155
1
(a)
(b)
0.8
0.6
0.4
0.2
0
A
B
A
C
D
B
C
D
45
65
85
105
125
145
Temperature /oC
Fig. 2. Non-isothermal plots of fraction remaining (1 − α) vs. temperature (◦C) for (a) ylide·BPh3 (3) and (b) ylide·B(C6F5)3 (4). Heating rate—curve
(A): 2.5 ◦C min−1; curve (B): 5.0 ◦C min−1; curve (C): 7.5 ◦C min−1; curve (D): 10 ◦C min−1
.
Table 1
Models and kinetic parameters for ylide·BPh3 (3) and ylide·B(C6F5)3 (4)
Complex
Model
Ea (kJ mol−1
)
A (min−1
)
R
Ylide·BPh3 (3)
A1.5
A2
120
93
1
4
5.60 × 1017
(
(
0.8%)
4.0%)
0.9944
0.9929
4.79 × 1013
Ylide·B(C6F5)3 (4)
R2
R3
F1
248
248
280
7
5
5
3.33 × 1030
1.64 × 1031
1.00 × 1035
(
(
(
2.0%)
2.0%)
2.1%)
0.9693
0.9691
0.9664
The correlation coefficient (r) and the Ea and A values ob-
tained from the equation of the best-fit lines are summarized
in Table 1. Fig. 3 shows a plot of the Ea values at various α
values for ylide·BPh3 (3) and ylide·B(C6F5)3 (4) complexes.
The Ea for these complexes is fairly constant over the range
of α used in the kinetic analyses, as expected [16].
The regression lines for R2 and R3 models fit well with
the composite DSC data (Fig. 4) and F1 (not shown) with
ylide·B(C6F5)3 (4) complex using scan rates of 5.0, 7.5, and
10 ◦C min−1. The 2.5 ◦C min−1 set deviated significantly
1.5
1
0.5
0
-0.5
-1
2.35
2.4
2.45
2.5
2.55
1000/T, K-1
R2 = 0.9396
(a)
250
(C6F5)3B-ylide
1.5
1
Ph3B-ylide
200
0.5
0
150
100
50
-0.5
-1
-1.5
2.35
2.4
2.45
2.5
2.55
0
0.2
0.4
0.6
0.8
1
1000/T, K-1
(b)
α
Fractional reaction ( )
Fig. 4. Regression lines for the composite DSC data for ylide·B(C6F5)3
(4) model: (a) R2; (b) R3.
Fig. 3. Plots of the activation energy vs. α as estimated by the Ozawa
method [12,13].

J.M. Stoddard, K.J. Shea / Thermochimica Acta 424 (2004) 149–155
153
Table 2
2
1
Activation parameters for the 1,2-migration of ylide·borane complexes
Ylide·B(C6F5)3 (4)
Ylide·BPh3 (3)
–1
b
a
Ea (kJ mol
)
171 ( 7)
129 ( 8)
A (min−1
)
4.68 × 1021
168 ( 7)
(
4.5%)
1.04 × 1019
127 ( 8)
( 7.0%)
0
‡
–1 c
)
ꢀH (kJ mol
‡
–1 d,f
ꢀS (J mol–1 K
)
127 ( 17)
130 ( 1.7)
76.6 ( 12)
104 ( 1.2)
-1
2.65
‡
–1 e,f
ꢀG (kJ mol
)
2.75
2.85
2.95
3.05
1000/T, K-1
a
Eq. (4).
(a)
b
Eq. (5).
c
d
e
f
‡
ꢀH = Ea − RT.
‡
1.6
ꢀS = R ln(Ah/ekT).
‡
‡
‡
ꢀG = ꢀH –T ꢀS .
T = 298 K.
0.8
0
tivation parameters to be similar for different kinetic models
in solid-state reactions [9].
-0.8
The exothermic peaks were often preceded by a falling
baseline, and establishing the limits for integration was
sometimes ambiguous. As a result there may be errors asso-
ciated with the integration of peak areas, which are used to
obtain α. In order to test the validity of previous calculations
(Table 1), the method of Kissinger [17] was used, which
does not require the extraction of α from the area under the
DSC peaks. By studying the shift in the peak maximum
Tm at different heating rates β the energy of activation Ea
may be obtained by using Eq. (4). The Kissinger method
uses only the peak maxima and does not rely on accurate
integration of the peak areas to obtain α. The Arrhenius
frequency parameter A was obtained using Eq. (5) [17].
2.65
2.75
2.85
2.95
3.05
1000/T, K-1
(b)
Fig. 5. Regression lines for the composite DSC data for ylide·BPh3 model:
(a) A1.5; (b) A2.
from the other data and was not included in the analysis since
it was rejected with the Q-test (90% confidence limit). The
slope and intercept of the regression line were used to cal-
culate the Arrhenius activation parameters, which are listed
in Table 1.
The regression lines for the A1.5 and A2 models fit well
with the composite DSC data for ylide·BPh3 (3) complex
using scan rates of 2.5, 5.0, 7.5, and 10 ◦C min−1 (Fig. 5). In
this case, the good data nearly coincide and it is difficult to
identify individual heating curves from the composite over-
lay (Fig. 5). The derived Arrhenius activation parameters
from these kinetic models are listed in Table 1. The calcu-
lated activation energies for the reaction of ylide·B(C6F5)3
were similar for models R2, R3, and F1. The calculated ac-
tivation energies for the reaction of ylide·BPh3 were similar
for A1.5 and A2. It is common for the derived Arrhenius ac-
d(ln(β/Tm2 ))
d(1/Tm)
Ea
R
= −
(4)
(5)
Ea
RT2m
A = β
eE /RT
a
The plots of ln(β/Tm2 ) versus 1/Tm for the reaction of
ylide·BPh3 (3) and ylide·B(C6F5)3 (4) complexes are shown
in Fig. 6 and yield the kinetic parameters, which are listed
in Table 2.
-13.2
-14
ylide•B(C F ) ( )
4
6
5 3
ylide•BPh3 ( )
3
-14.8
-15.6
2.4
2.6
2.8
3.0
1000/T, K-1
Fig. 6. Plots of ln(β/Tm2 ) vs. 1/Tm for the reaction of ylide·BPh3 (3) and ylide·B(C6F5)3 (4).

154
J.M. Stoddard, K.J. Shea / Thermochimica Acta 424 (2004) 149–155
CH3
S
O
C6F5
C6F5
CH3
S
C6F5
B
C6F5
C6F5
CH3
O
B
CH3
+ DMSO
B
CH2C6F5
C6F5
C6F5
C6F5
Scheme 2. Reaction of ylide·B(C6F5)3 (4) produces a DMSO organoborane complex.
4. Discussion
orane (C6F5)2BCH2(C6F5) (Scheme 2). The activation
parameters obtained with the Kissinger method may not be
affected as much, due to the exothermic complexation, since
α and peak areas are not determined using this method.
The energy of activation for the reaction of ylide·BPh3
(3) and ylide·B(C6F5)3 (4) are consistent with a solid
that undergoes transition upon heating above room tem-
perature (Ea > 105 kJ mol−1) [2]. The gas-phase ab ini-
tio HF/6-31G(d) calculations of activation energies for
ylide·Ph3B (Ea = 67 kJ mol−1) and ylide·B(C6F5)3 (Ea =
92 kJ mol−1) parallel the relative reactivity of the experi-
mentally derived energies of activation (see Fig. 7). The
The conversion plots of ylide·B(C6F5)3 (4) and ylide·BPh3
(3) (Fig. 2) differ in shape and may indicate differences in
models of topochemical reaction or other complications. For
example, the 1,2-migration of ylide·B(C6F5)3 (4) produces a
strong Lewis acid and DMSO, which are capable of combin-
ing to form an organoborane·DMSO complex (Scheme 2)
[2]. This complexation is expected to be exothermic and will
cause the conversion plots to be inaccurate due to errors in
integration. However, the activation energies obtained using
the model-free Flynn–Wall–Ozawa method [12,13] (Fig. 3)
are constant over the α considered and establish a basis for
using the data in subsequent kinetic analyses [18,19].
The kinetic analysis of the decomposition of ylide·BPh3
(3) with the Kissinger method (Ea = 129 kJ mol−1, A =
1.04 × 1019 min−1) and Doyle’s equation using the A1.5
model (Ea = 120 kJ mol−1, A = 5.60 × 1017 min−1) gave
consistent results for calculated values of both Ea and A.
The satisfactory agreement between these data demonstrates
that obtaining rates of initiation is amenable by this DSC
methodology.
‡
entropies of activation (ꢀS > 0) are consistent with a
unimolecular decomposition to two fragments [20]. The
Kissinger method was used to obtain the entropies of activa-
‡
tion for ylide·BPh3 (3) (ꢀS = 76.6 12 J mol−1 K−1) and
‡
ylide·B(C6F5)3 (4) (ꢀS = 127 17 J mol−1 K−1). Due
to lengthy computational times, the frequency calculations
for ylide·B(C6F5)3 (4) and ylide·BPh3 (3) transition states
were not performed at HF/6-31G(d) to obtain the entropies
directly. However, the transition state for ylide·B(C6F5)3
(4) exhibits both a shorter C1–C2 bond (2.240 Å) and C2–S
bond (2.400 Å) than the ylide·BPh3 (3) transition state
(C1–C2 = 2.250 Å, C2–S = 2.453 Å). This suggests that
the transition state for ylide·B(C6F5)3 (4) is more ordered
compared with ylide·BPh3 (3). In solution the ylide·BPh3
complex undergoes phenyl 1,2-migration at 23 ◦C while
the ylide·B(C6F5)3 complex is unreactive at 23 ◦C, these
differences also parallel the trends in the solid state.
The kinetic analysis of the decomposition of ylide·
B(C6F5)3 (4) with the Kissinger method (Ea = 171 kJ mol−1
A = 4.68 × 1021 min−1) and Doyle’s equation using the
R2, R3, and F1 models (Ea = 248–280 kJ mol−1, A =
3.33×1030to1.00 ×1035 min−1) deviated significantly. It is
believed that the conversion plots used for Doyle’s equation
contain significant error due to the exothermicity associated
with complexation of DMSO to the homologated organob-
,
Fig. 7. HF/6-31G(d) calculated transition-state structures for ylide·BPh3 (3) and ylide·B(C6F5)3 (4).

J.M. Stoddard, K.J. Shea / Thermochimica Acta 424 (2004) 149–155
155
[2] J.M. Stoddard, K.J. Shea, Organometallics 22 (2003) 1124.
[3] S. Vyazovkin, Int. Rev. Phys. Chem. 19 (2000) 45.
5. Conclusion
[4] J.D. Sewry, M.E. Brown, Thermochim. Acta 390 (2002)
217.
[5] Caution, the BH3-ylide complex has been prepared in 3 g quantities as
a white solid substance. The DSC (<1.5 mg) shows a very exothermic
transition above 55 ◦C and larger amounts of solid complex should
not be heated.
[6] M.J. Frisch, et al., gaussian 98, Revision A.11, Gaussian, Inc.,
Pittsburgh, PA, 2001.
[7] A.K. Galwey, Thermochim. Acta 399 (2003) 1.
[8] E. Davydov, A. Vorotnikov, G. Pariyskii, G. Zalkov, Kinetic Pecu-
liarities of Solid Phase Reactions, John Wiley & Sons, New York,
1998.
The Arrhenius activation parameters for the reaction of
ylide·Ph3B (3) and ylide·B(C6F5)3 (4) were obtained using
Doyle’s equation and Kissinger’s method. The kinetics of
1,2-migration best-fit with R2, F3, F1 models for ylide·Ph3B
(3), and A1.5 and A2 models for ylide·B(C6F5)3 (4). The two
methods provided similar kinetic parameters with an acti-
vation energy for the reaction of ylide·B(C6F5)3 (4) greater
than for the reaction of ylide·Ph3B (3). HF/6-31G(d) calcu-
lations substantiate the finding that C6F5 1,2-migration in
‡
complex 4 had a higher Ea and ꢀS than C6H5 1,2-migration
in complex 3.
[9] A.K. Galwey, M.E. Brown, Thermal Decomposition of Ionic Solids,
Elsevier, Amsterdam, 1999.
[10] C.E. Bawn, Chem. Solid State (1955) 254.
[11] J.H. Flynn, J. Wall, Nat. Bur. Stand. 70A (1966) 487.
[12] T. Ozawa, Bull. Chem. Soc. Jpn. 38 (1965) 1881.
[13] T. Ozawa, J. Therm. Anal. 2 (1970) 301.
[14] D. Doyle, J. Appl. Polym. Sci. 5 (1961) 285.
[15] B. Budrugeac, D. Homentcouschi, E. Segal, J. Therm. Anal. Cal. 63
(2001) 457.
Acknowledgements
This research was supported from a grant from the Na-
tional Sciences Foundation (CHE-9617475).
[16] S. Vyazovkin, New J. Chem. 24 (2000) 913.
[17] H. Kissinger, J. Res. Nat. Bureau Standards 57 (1956) 217.
[18] S. Vyazovkin, C.A. Wright, Thermochim. Acta 340/341 (1999)
53.
References
[19] S.V. Vyazovkin, J. Comput. Chem. 22 (2001) 178.
[20] T.H. Lowry, K.S. Richardson, Mechanism and Theory in Organic
Chemistry, 2nd ed., Harper & Row, New York, 1981.
[1] K.J. Shea, B.B. Busch, M.M. Paz, C.L. Staiger, J.M. Stoddard, J.R.
Walker, X. Zhou, H. Zhu, J. Am. Chem. Soc. 124 (2002) 3636.