Free radical chlorination of alkanes in supercritical carbon dioxide: The chlorine atom cage effect as a probe for enhanced cage effects in supercritical fluid solvents

Article abstract of DOI:10.1021/ja982289e

The chlorine atom cage effect was used as a highly sensitive probe for studying the effect of viscosity and the possible role of solvent clusters on cage lifetimes and reactivity for reactions carried out in supercritical fluid solvents. These experiments were conducted in supercritical carbon dioxide (SC-CO₂, 40 °C, at various pressures) with parallel experiments in conventional solvents and in the gas phase. The results of these experiments provide no indication of an enhanced cage effect near the critical point in SC-CO₂ solvent. The magnitude of the cage effect observed in SC-CO₂ at all pressures examined is well within what is anticipated on the basis of extrapolations from conventional solvents.

Full text of DOI:10.1021/ja982289e

VOLUME 120, NUMBER 46
NOVEMBER 25, 1998
© Copyright 1998 by the
American Chemical Society
Free Radical Chlorination of Alkanes in Supercritical Carbon
Dioxide: The Chlorine Atom Cage Effect as a Probe for Enhanced
Cage Effects in Supercritical Fluid Solvents
Beth Fletcher, N. Kamrudin Suleman,† and J. M. Tanko*
Contribution from the Department of Chemistry, Virginia Polytechnic Institute and State UniVersity,
Blacksburg, Virginia 24061-0212
ReceiVed June 30, 1998
Abstract: The chlorine atom cage effect was used as a highly sensitive probe for studying the effect of viscosity
and the possible role of solvent clusters on cage lifetimes and reactivity for reactions carried out in supercritical
fluid solvents. These experiments were conducted in supercritical carbon dioxide (SC-CO2, 40 °C, at various
pressures) with parallel experiments in conventional solvents and in the gas phase. The results of these
experiments provide no indication of an enhanced cage effect near the critical point in SC-CO2 solvent. The
magnitude of the cage effect observed in SC-CO2 at all pressures examined is well within what is anticipated
on the basis of extrapolations from conventional solvents.
Introduction
Moreover, because of the unique nature of SCF’s, there may
be additional factors which influence chemical reactivity in a
manner not possible in “conventional” solvents. Several studies
have demonstrated that, for SCF’s as solvents, the local solVent
density about a solute is often enhanced at pressures just above
the critical pressure (solvent/solute “clustering”)7-18 and that
cluster formation may influence chemical reactivity.19-25 In-
creased solvent density (and implicitly increased viscosity) may
The supercritical state is achieved when a substance is taken
above its critical temperature and pressure. The bulk properties
of a supercritical fluid (SCF) are intermediate between those
of a gas and a liquid. Important solvent properties of SCF’s
(e.g., dielectric constant, solubility parameter, and viscosity) can
be altered by manipulation of temperature and pressure,1,2
providing a potential means to control the behavior (kinetics)
of some chemical reactions.3-6
(6) Eckert, C. A.; Knutson, B. L.; Debenedetti, P. G. Nature 1996, 383,
313-318.
(7) Kim, S.; Johnston, K. P. AIChE J. 1987, 33, 1603-1611. Kim, S.;
Johnston, K. P. Ind. Eng. Chem. Res. 1987, 26, 1206-1213. Johnston, K.
P.; McFann, G. J.; Peck, D. G.; Lemert, R. M. Fluid Phase Equilib. 1989,
52, 337-346.
(8) Kajimoto, O.; Futakami, M.; Kobayashi, T.; Yamasaki, K. J. Phys.
Chem. 1988, 92, 1347.
(9) Debenedetti, P. G.; Petsche, I. B.; Mohamed, R. S. Fluid Phase
Equilib. 1989, 52, 347-356.
(10) Brennecke, J. F.; Tomasko, D. L.; Peshkin, J.; Eckert, C. A. Ind.
Eng. Chem. Res. 1990, 29, 1682-1690.
(11) Betts, T. A.; Zagrobelny, J.; Bright, F. V. J. Am. Chem. Soc. 1992,
114, 8163-8171. Zagrobelny, J.; Betts, T. A.; Bright, F. V. J. Am. Chem.
Soc. 1992, 114, 5249-5257. Heitz, M. P.; Bright, F. V. J. Phys. Chem.
1996, 100, 6889-6897.
† Visiting Research Scientist from the Department of Chemistry,
University of Guam, Mangilao, Guam 96923.
(1) McHugh, M.; Krukonis, V. Supercritical Fluid Extraction. Principles
and Practice; Butterworth: Boston, MA, 1986; pp 1-11.
(2) Johnston, K. P. In Supercritical Fluid Science and Technology,
Johnston, K. P., Penninger, J. M. L., Eds.; American Chemical Society:
Washington, DC, 1989; pp 1-12.
(3) Subramantam, B.; McHugh, M. A. Ind. Eng. Chem. Process Des.
DeV. 1988, 25, 1-12.
(4) Wu, B. C.; Paspek, S. C.; Klein, M. T.; LaMarca, C. In Supercritical
Fluid Technology: ReViews in Modern Theory and Applications; Bruno,
T. J., Ely, J. F., Eds.; CRC Press: Boca Raton, FL, 1991; pp 511-524.
(5) Savage, P. E.; Gopalan, S.; Mizan, T. I.; Martino, C. J.; Brock, E. E.
AICHE J. 1995, 41, 1723-1778.
10.1021/ja982289e CCC: $15.00 © 1998 American Chemical Society
Published on Web 11/10/1998

11840 J. Am. Chem. Soc., Vol. 120, No. 46, 1998
Fletcher et al.
have a significant effect on reactions that are diffusion-controlled
or reactions for which cage effects are important, and there are
several conflicting reports pertaining to enhanced cage effects
near the critical point in SCF solvents.26-29
Scheme 1
Scheme 2
Our approach to this issue is to examine reactions that are
already well-understood in conventional solvents to probe for
divergent behavior in an SCF solvent. To assess the nature
and extent of viscosity effects in an SCF solvent, the free radical
chlorination of several alkanes was examined in (a) conventional
organic solvents, (b) the gas phase, and (c) supercritical carbon
dioxide (SC-CO2).30 There is considerable interest in SC-CO2
as an “environmentally benign” reaction solvent,31 and SC-CO2
has been shown to be an excellent solvent for radical
reactions.32-38 The critical properties of CO2 are moderate (TC
) 31 °C, PC ) 74 bar), and SC-CO2 is essentially nonpolar.
(Over the range 74-300 bar, the dielectric constant of SC-CO2
(12) Randolph, T. W.; Carlier, C. J. Phys. Chem. 1992, 96, 5146-5151.
Ganapathy, S.; Carlier, C.; Randolph, T. W.; O’Brien, J. A. Ind. Eng. Chem.
Res. 1996, 35, 19-27.
(13) Sun, Y.-P.; Bunker, C. E.; Hamilton, N. B. Chem. Phys. Lett. 1993,
210, 111-117. Sun, Y.-P.; Bunker, C. E. Ber. Bunsen-Ges. Phys. Chem.
1995, 99, 976-984.
(14) Sun, Y.-P.; Fox, M. A.; Johnston, K. P. J. Am. Chem. Soc. 1992,
114, 1187-1194. Sun, Y.-P.; Bennett, G.; Johnston, K. P.; Fox, M. A. J.
Phys. Chem. 1992, 96, 10001-10007. Sun, Y.-P.; Fox, M. A. J. Am. Chem.
Soc. 1993, 115, 747-750. Rhodes, T. A.; Fox, M. A. J. Phys. Chem. 1996,
100, 17931-17939.
(15) Anderton, R. M.; Kauffman, J. F. J. Phys. Chem. 1995, 99, 13759-
13762.
(16) Urdahl, R. S.; Rector, K. D.; Myers, D. J.; Davis, P. H.; Fayer, M.
D. J. Chem. Phys. 1996, 105, 8973-8976.
(17) Tucker, S. C.; Maddox, M. W. J. Phys. Chem. B 1998, 102, 2437-
2453.
(18) Takahashi, K.; Abe, K.; Sawamura, S.; Jonah, C. D. Chem. Phys.
Lett. 1998, 282, 361-368.
(19) Gehrke, C.; Schroeder, J.; Schwarzer, D.; Troe, J.; Voss, F. J. Chem.
Phys. 1990, 92, 4805-4816.
(20) Combes, J. R.; Johnston, K. P.; O’Shea, K. E.; Fox, M. A. In
Supercritical Fluid Technology: Theoretical and Applied Approaches in
Analytical Chemistry; Bright, F. V., McNally, M. E. P., Eds.; American
Chemical Society: Washington, DC, 1992; pp 31-47.
(21) Ikushima, Y.; Saito, N.; Arai, M. J. Phys. Chem. 1992, 96, 2293-
2297.
(22) Renslo, A. R.; Weinstein, R. D.; Tester, J. W.; Danheiser, R. L. J.
Org. Chem. 1997, 62, 4530-4533.
(23) Randolph, T. W.; O’Brien, J. A.; Ganapathy, S. J. Phys. Chem.
1994, 98, 4173.
(24) Takahashi, K.; Jonah, C. D. Chem. Phys. Lett. 1997, 264, 297-
302.
(25) Kajimoto, O.; Sekiguchi, K.; Nayuki, T.; Kobayashi, T. Ber. Bunsen-
Ges. Phys. Chem. 1997, 101, 600-605.
(26) O’Shea, K. E.; Combes, J. R.; Fox, M. A.; Johnston, K. P.
Photochem. Photobiol. 1991, 54, 571-576.
changes only slightly from 1.3 to 1.5, while the viscosity changes
from 0.02 to 0.12 cP.)
The free radical chlorination of alkanes represents a classic
procedure for the functionalization of alkanes (RH + Cl2 f
RCl + HCl). Many of the details of this reaction have been
well-understood for more than half a century. The mechanism
of this reaction is a free radical chain process, the propagation
steps of which are depicted in Scheme 1: The chlorine atom
abstracts hydrogen from the alkane, yielding an alkyl radical
and HCl. The alkyl radical subsequently reacts with molecular
chlorine, yielding the product alkyl chloride and regenerating
chlorine atom. The chlorine atom is a highly reactive species
and exhibits low selectivity in hydrogen abstractions: In
solution, 3° C-H (4.2) > 2° C-H (3.6) > 1° C-H (1.0), on a
per hydrogen basis (25 °C);39 absolute rate constants for
hydrogen abstraction are just slightly below the diffusion-
controlled limit.40
Results and Discussion
A. The Chlorine Atom Cage Effect as a Probe of Viscosity
Effects in SC-CO2. The chlorine atom cage effect, first
discovered by Skell and Baxter in 1983,41 has been the subject
of numerous investigations.42-44 Put briefly, for the chlorine
atom abstraction step in the free radical chlorination of an alkane
(RH2), the geminate RHCl/Cl‚ caged pair is partitioned among
three pathways (Scheme 2): diffusion apart (kdiff), abstraction
(27) Roberts, C. B.; Chateauneuf, J. E.; Brennecke, J. F. J. Am. Chem.
Soc. 1992, 114, 8455-8463. Roberts, C. B.; Zhang, J.; Chateauneuf, J. E.;
Brennecke, J. F. J. Am. Chem. Soc. 1993, 115, 9576-9582. Roberts, C.
B.; Zhang, J.; Brennecke, J. F.; Chateauneuf, J. E. J. Phys. Chem. 1993,
97, 5618. Roberts, C. B.; Zhang, J.; Chateauneuf, J. E.; Brennecke, J. F. J.
Am. Chem. Soc. 1995, 117, 6553-6560.
of hydrogen from RH2 comprising the cage walls (kRH ) and a
second in-cage abstraction of hydrogen from the alkyl chloride
2
(kRHCl). While the kdiff and k steps result in the formation of
RH2
monochloride (RHCl), the kRHCl step results in the formation
(28) Andrew, D.; Des Islet, B. T.; Margaritis, A.; Weedon, A. C. J. Am.
Chem. Soc. 1995, 117, 6132-6133.
(36) Tanko, J. M.; Blackert, J. F. Science 1994, 263, 203-205. Tanko,
J. M.; Blackert, J. F.; Sadeghipour, M. In Benign by Design: AlternatiVe
Synthetic Design for Pollution PreVention; Anastas, P. T., Farris, C. A.,
Eds.; American Chemical Society: Washington, DC, 1994; pp 98-113.
(37) Srinivas, P.; Mukhopadhyay, M. Ind. Eng. Chem. Res. 1994, 33,
3118.
(38) Hadida, S.; Super, M. S.; Beckman, E. J.; Curran, D. P. J. Am.
Chem. Soc. 1997, 119, 7406-7407.
(39) Russell, G. A. J. Am. Chem. Soc. 1958, 80, 4997-5001.
(40) Bunce, N. J.; Ingold, K. U.; Landers, J. P.; Lusztyk, J.; Scaiano, J.
C. J. Am. Chem. Soc. 1985, 107, 5464.
(29) Bunker, C. E.; Rollins, H. W.; Gord, J. R.; Sun, Y.-P. J. Org. Chem.
1997, 62, 7324-7329.
(30) Tanko, J. M.; Suleman, N. K.; Fletcher, B. J. Am. Chem. Soc. 1996,
118, 11958-11959.
(31) (a) Amato, I. Science 1993, 259, 1538. (b) Wedin, R. E. Today’s
Chemist at Work 1993, 2, 16. (c) Illman, D. Chem. Eng. News 1993, 71
(May 29), 5.
(32) Sigman, M. E.; Leffler, J. E. J. Org. Chem. 1987, 52, 1165-1167.
Sigman, M. E.; Barbas, J. T.; Leffler, J. E. J. Org. Chem. 1987, 52, 1754-
1757.
(33) Suppes, G. J.; Occhiogrosso, R. N.; McHugh, M. A. Ind. Eng. Chem.
Res. 1989, 28, 1152-1156.
(34) Wu, B. C.; Klein, M. T.; Sandler, S. I. Energy Fuels 1991, 5, 453-
458.
(35) DeSimone, J. M.; Guan, Z.; Elsbernd, C. S. Science 1992, 257, 945-
947. Guan, Z.; Combes, J. R.; Menceloglu, Y. Z.; DeSimone, J. M.
Macromolecules 1993, 26, 2663-2669.
(41) Skell, P. S.; Baxter, H. N., III. J. Am. Chem. Soc. 1985, 107, 2823-
2824.
(42) Raner, K. D.; Lusztyk, J.; Ingold, K. U. J. Am. Chem. Soc. 1988,
110, 3519-3524.
(43) Tanko, J. M.; Anderson, F. E., III. J. Am. Chem. Soc. 1988, 110,
3525-3530.
(44) Tanner, D. D.; Oumar-Mahamat, H.; Meintzer, C. P.; Tsai, E. C.;
Lu, T. T.; Yang, D. J. Am. Chem. Soc. 1991, 113, 5397-5402.

Chlorination of Alkanes in Supercritical CO2
J. Am. Chem. Soc., Vol. 120, No. 46, 1998 11841
of polychlorides via RCl‚ + Cl2 f RCl2 + Cl‚. Employing
the steady-state approximation (i.e., d[(RHCl/Cl‚)cage]/dt ) 0),
the ratio of mono- to polychlorides can be expressed in terms
of the rate constants assigned in Scheme 2 (eq 1).
kdiff + k [RH2]cage walls
RH2
M
P
)
(1)
kRHCl
Tanner has shown that in conventional solvents, the ratio of
mono- to polychlorinated products (M/P) depends on solvent
viscosity.44 On this basis, we reasoned that the chlorine atom
cage effect would be a highly sensitive probe for studying the
effect of SCF viscosity and the possible role of solvent clusters
on cage lifetimes and reactivity.30 Toward this end, the ratio
of mono- to polychlorides produced in the free radical chlorina-
tions of cyclohexane, neopentane, and 2,3-dimethylbutane were
examined in SC-CO2 (40 °C at various pressures), with parallel
experiments in conventional solvents. Alkane concentrations
were e0.03 M for all these experiments, and under these
conditions, kRH [RH2]cage walls , kdiff (i.e., ensuring that diffusion
2
of RHCl and Cl‚ apart, rather than reaction of Cl‚ with RH2
comprising the cage walls, was the major source of monochlo-
ride). The results are summarized in the Supporting Information
(Tables S1, S2, and S4).
Cage effects are typically quantified in terms of the Noyes
model. Koenig and Fischer45 derived a general expression based
upon the Noyes model (eq 2), where F ) (1/cage efficiency)
Figure 1. Viscosity dependence of the ratio of mono- to polychlorides
(M/P) produced in the free radical chlorination of cyclohexane
(conventional and SC-CO2 solvent) as a function of inverse viscosity
at 40 °C.
R0 - 2b R0 AT + RAE 1
ATAE 1
2
F )
+
+
(2)
[
]
( )
( )
2b
2b
R
η
R
η
- 1, R0 is the separation between the two reactive components
of the caged pair, R is the probability (per collision) that the
components will react, and η is the viscosity of the solvent. AE
and AT are constants which incorporate terms pertaining to the
mass, radius, and translational energy of the components of the
caged pair, and b is the diffusional radius.45
Equation 2 predicts that F will vary linearly with 1/η, except
perhaps at low viscosities, where upward curvature may be
observed if 1/η2 becomes important. Cage efficiency represents
the fraction of reaction that occurs within the cage and applied
to Scheme 2, is equal to kRHCl/(kdiff + kRHCl + kRH [RH2]).
2
Because the mono- to polychloride ratio (M/P) ) kdiff/kRHCl (at
the low concentrations of alkane employed, kRH [RH2] , kdiff
,
2
kRHCl), it is trivial to show that M/P is also equal to F. Thus,
on the basis of the Noyes model, it is anticipated that M/P will
vary linearly with inverse viscosity (1/η).
In Figures 1-3, M/P ratios observed in the chlorination of
2,3-dimethylbutane, neopentane, and cyclohexane are plotted
as a function of 1/η for the experiments conducted in SC-CO2
and in conventional solvents. Within experimental error, these
plots are linear over a range of viscosities spanning nearly 2
orders of magnitude (from conventional solvents to SC-CO2).
It is also worth noting that although they represent a relative
small portion of the data, the best straight line through the
solution phase results successfully predicts the SCF phase
results.
Figure 2. Viscosity dependence of the ratio of mono- to polychlorides
(M/P) produced in the free radical chlorination of neopentane (con-
ventional and SC-CO2 solvent) as a function of inverse viscosity at 40
°C.
pressures examined is well within what is anticipated on the
basis of extrapolations from conventional solvents.
B. Reaction Selectivity in SC-CO2 Compared to Con-
ventional Solvents and the Gas Phase. The experiments with
2,3-dimethylbutane (23DMB) provide insight into the extent that
Cl‚ selectivity varies as a function of pressure in SC-CO2. In
the gas phase at 40 °C, chlorine atom selectivity (per hydrogen)
for the 3° vs 1° hydrogens of 23DMB (S(3°/1°)) is 3.97.46 In
the condensed phase (neat 23DMB), S(3°/1°) ) 3.08 (40 °C).
In SC-CO2 at 40 °C, the observed selectivity (S(3°/1°)obsd) varies
with pressure.
On the basis of these observations, we find no indication of
an enhanced cage effect near the critical point in SC-CO2
solvent, which could be attributed to solvent/solute clustering.
The magnitude of the cage effect observed in SC-CO2 at all
(45) Koenig, T.; Fischer, H. In Free Radicals; Kochi, J. K., Ed.; Wiley:
New York, 1973; Vol. 1, pp 157-189.
(46) S(3°/1°) ) (yield of 3° RCl/yield of 1° RCl) × 6.

11842 J. Am. Chem. Soc., Vol. 120, No. 46, 1998
Fletcher et al.
Figure 4. Observed selectivity in the chlorination of 2,3-dimethylbu-
tane (conventional and SC-CO2 solvent) at 40 °C as a function of the
ratio of mono- to polychlorides (M/P) produced.
RCl/Cl‚ caged pairs which escape (F3 and F1, respectively) can
Figure 3. Viscosity dependence of the ratio of mono- to polychlorides
(M/P) produced in the free radical chlorination of 2,3-dimethylbutane
(conventional and SC-CO2 solvent) as a function of inverse viscosity
at 40 °C.
be expressed by eqs 3 and 4 (where kesc ) kRH [RH2]cw + kdiff).
2
kesc
F3 )
F1 )
(3)
(4)
3
kesc + kH
kesc
Scheme 3
1
kesc + kH
Because the obserVed yields of 3° vs 1° RCl ) (F3/F1)(k3/
k1) (where k3/k1 represents the true selectivity on a per molecule
basis), an expression can be derived which relates M/P to k3/k1
(eq 5).
(k1/k3)F1 + F3
M
P
)
(5)
(k1/k3)(1 - F1) + (1 - F3)
In SC-CO2, however, S(3°/1°)obsd derived from the relative
yields of 3° vs 1° RCl are not a true measure of chlorine atom
selectivity because, at higher pressures, product yields are
distorted by the cage effect. This point is illustrated in Scheme
3. For the chlorination of 23DMB, there are two types of caged
pairs produced, 3° RCl/Cl‚ and 1° RCl/Cl‚ (designated as C-3
and C-1, respectively), and in-cage hydrogen abstraction by Cl‚
likely occurs at different rates for these two caged pairs (i.e.,
Assuming that k3/k1 in Freon 113 and neat 23DMB is the
same, F1 and F3 can be derived as a function of M/P for
reactions in conventional solvents (Supporting Information,
Figure S-1). Finally, assuming that the same relationship
between F1 or F3 and M/P holds true in SC-CO2 (i.e., assuming
that the relationship between M/P and the fraction of cage escape
is the same regardless of whether cage escape occurs via reaction
of Cl‚ with the cage walls [kRH ] or via diffusion of Cl‚ out of
2
3
kH * kH1).
the cage, [kdiff]), the data obtained in SC-CO2 can be corrected
for the cage effect (Figure 5).
This distortion of the observed selectivity by the cage effect
is readily demonstrated by examining the free radical chlorina-
tion of 23DMB in a conventional solvent (1,1,2-trichlorotrif-
luoroethane, Freon 113) at various concentrations of 23DMB.
In accordance with earlier published results and eq 1, M/P was
found to vary with [23DMB], and in addition, the observed
selectivity as measured by the relative yields of the two
monochlorides also varies. In Figure 4, S(3°/1°)obsd is plotted
against the M/P ratio for reactions conducted in both Freon 113
and SC-CO2. At high values of M/P (i.e., when the cage effect
becomes relatively unimportant and the observed selectivity is
expected to reflect the true selectivity), selectivity is slightly
higher in SC-CO2 compared to a conventional solvent.
As Figure 5 reveals, in SC-CO2 S(3°/1°) varies as a function
of viscosity between the gas- and condensed-phase values. This
variation in S(3°/1°) can be explained as follows: The rate
constants for 1°, 2°, or 3° hydrogen abstractions by Cl‚ from
alkanes are nearly diffusion-controlled in conventional solvents.
Consequently, the intrinsic selectivity of Cl‚ is diminished in
conventional solvents because of the onset of diffusion-control.
In the gas phase, selectivity is slightly higher because the barrier
imposed by diffusion is eliminated. The viscosity of a super-
critical fluid (a) lies between those of conventional fluid solvent
and the gas phase and (b) varies with pressure. Because of the
low viscosity of supercritical fluids, bimolecular rate constants
greater than the 1010 M-1 s-1 diffusion-controlled limit can be
realized in an SCF, and as a consequence, enhanced selectivity
is achieved.
To extract the “true” S(3°/1°), the data obtained in conven-
tional solvents were used to “correct” for the cage effect in SC-
CO2. In accordance with Scheme 3, the fraction of 3° and 1°

Chlorination of Alkanes in Supercritical CO2
J. Am. Chem. Soc., Vol. 120, No. 46, 1998 11843
Figure 6. Product ratio (1,3- to 1,1-dichloro-2,2-dimethylpropane)
formed in the chlorination of neopentane in SC-CO2 solvent as a
function of viscosity (40 °C).
Figure 5. Chlorination of 2,3-dimethylibutane in SC-SO2: selectivity
(corrected) as a function of viscosity.
Table 1. Free Radical Chlorination of Propane in SC-CO2
a qualitative description of viscosity effects on molecular
rotation. At 40 °C, the gas-phase chlorination of neopentane
produces 1,3- and 1,1-dichloro-2,2-dimethylpropane in a 2.42:1
ratio. Because there is no cage effect in the gas phase, this
ratio represents the selectivity associated with the abstraction
of hydrogen from neopentyl chloride by Cl‚ via a diffusive
encounter.
pressure (psi)
viscosity (cP)
S(2°/1°)a
1160
1310
1753
0.029
0.042
0.052
2.94 ( 0.29
2.86 ( 0.29
2.70 ( 0.27
a S(2°/1°) ) (yield of CH3CHClCH3/yield of CH3CH2CH2Cl) × 3.
Scheme 4
In SC-CO2, the ratio of the dichlorides varies with pressure.
In Figure 6, the dichloride ratio is plotted as a function of
viscosity. At high viscosities, the 1,3-isomer dominates because
of the cage effect. At low viscosities, the dichloride ratio
decreases and approaches the gas-phase value, suggesting (at
low viscosities) dichloride formation occurs predominantly from
diffusive encounters and not the cage effect.
The free radical chlorination of propane in the gas phase and
SC-CO2 was also examined. In the gas phase at 40 °C, the
chlorine atom selectivity for the 2° vs 1° hydrogens of propane
(S(2°/1°) was found to be 3.09. In SC-CO2, selectivity was
slightly lower (Table 1). In these experiments, no polychlorides
were detected, which means that the relative yields of 2-chloro-
vs 1-chloropropane provide an accurate assessment of S(2°/1°).47
While a slight variation of selectivity with pressure was noted,
the differences observed were similar in magnitude to experi-
mental error. (An extended study of the effect of pressure on
M/P and S(2°/1°) was not pursued because, at high pressures,
recovery of the volatile reaction products proved to be erratic
and unreliable.)
C. Dichloride Pattern in the Chlorination of Neopentane
in SC-CO2. Abstraction of the second hydrogen within the
RHCl/Cl‚ caged pair occurs at the site(s) in close proximity to
the initially abstracted hydrogen atom, and this second hydrogen
abstraction occurs on a time scale competitive with molecular
rotation.42,43 This phenomenon is especially apparent in the
chlorination of neopentane:43 For the geminate neopentyl
chloride/Cl‚ caged pair, the methyl hydrogens are in closest
proximity to the initially formed chlorine atom (Scheme 4) and
are abstracted more readily. Consequently, the ratio of 1,3-
dichloro-2,2-dimethylpropane and 1,1-dichloro-2,2-dimethyl-
propane produced from the geminate caged pair is higher (on
the order of (4-5):1) compared to the 1,3/1,1 ratio produced
from a diffusive caged pair (ca. 2.2:1, based upon the chlorina-
tion of neopentyl choride),43 where all possible trajectories are
possible.
Conclusions
In SC-CO2, the magnitude of the chlorine atom cage effect
is as anticipated on the basis of extrapolations from conventional
solvents. No evidence was found for an enhanced cage effect
in SC-CO2, which might be attributable to solvent/solute
clustering. Chlorine atom selectivities in SC-CO2 vary slightly
with pressure (viscosity) and are intermediate between gas- and
solution-phase values. Compared to that of conventional liquid
solvents, the higher selectivity observed in SC-CO2 is attribut-
able to the low viscosity of the SCF media, which allows
absolute rate constants to exceed the 1010 M-1 s-1 limit imposed
by the viscosity of conventional solvents. Finally, these results
demonstrate that the “tunable” solvent properties of a super-
critical fluid provide a means of controlling reactivity and
selectivity. Thus, considering its properties as a reaction solvent
and the fact that it offers a nontoxic and less environmentally
threatening alternative to conventional solvents, SC-CO2 emerges
an excellent solvent for radical reactions.
Experimental Section
General Considerations. Gas chromatographic analyses were
performed on a Hewlett-Packard 5890A instrument equipped with FID
detection and an HP 3393A reporting integrator. Analyses were
conducted using either an Alltech SE-30 or SE-54 capillary column
(30 m × 0.25 mm i.d. × 0.25 µm). Products were identified by
comparison of retention time with that of an authentic sample and
quantitated vs a measured internal standard (chlorobenzene) and
appropriate GLC correction factors. All gas chromatographic analyses
were performed in triplicate.
An analysis of the ratio of 1,3- to 1,1- dichlorides formed
from the chlorination of neopentane in SC-CO2 affords at least
Materials. CFCl3, CF2ClCCl2F, decane (Aldrich), and CFCl2CFCl2
(PCR Inc.) were used as received. Cyclohexane (Aldrich or Matheson)
(47) S(2°/1°) ) (yield of 2° RCl/yield of 1° RCl) × 3.

11844 J. Am. Chem. Soc., Vol. 120, No. 46, 1998
Fletcher et al.
and 2,3-dimethylbutane were washed with dilute sulfuric acid, neutral-
ized with saturated sodium bicarbonate, dried over magnesium sulfate,
and distilled. Neopentane (Wiley Organics) was purified by treatment
with Br2 and distilled. SFC grade carbon dioxide was obtained from
Scott Specialty Gases.
vacuum, and placed in the reactor. A second ampule containing the
appropriate amount of Cl2 was added to the reactor (approximately 16
mL in volume). The reactor was then sealed, covered with aluminum
foil, and brought to 40 °C. Following several argon purges, the reactor
was pressurized with CO2 and its contents were allowed to equilibrate
at 40 °C and the desired pressure for several minutes. The aluminum
foil was removed, and the reactor was illuminated as described
previously. Following illumination, the contents of the reactor were
bubbled slowly into hexanes cooled to 0 °C. The internal standard
was added, and direct analyses by GLC were performed in triplicate.
Reactions involving neopentane were conducted in the same manner,
with the neopentane being measured manometrically and condensed
into the degassed ampule. Due to its volatility, propane could not be
sealed in an ampule and was added directly to the reactor. The reactor
was purged several times with propane and sealed at ambient pressure.
The propane:Cl2 ratio for these reactions was approximately 14:1.
Following illumination, the products were collected in hexane cooled
to dry ice/acetone temperatures (-78 °C).
Gas-Phase Chlorination. For 2,3-dimethylbutane and cyclohexane,
the appropriate volume of alkane was added to a large pressure tube
(ca. 180 mL in volume). Typically, a 7:1 ratio of alkane to Cl2 was
used. The pressure tube was subsequently degassed by performing
three or four freeze-pump-thaw cycles. Neopentane or propane was
measured manometrically and condensed into the pressure tube via a
vacuum line. Cl2 was measured with a calibrated gas pipet and purified
immediately before use by condensation from a trap at -78 °C directly
into the degassed pressure tube. The pressure tube was sealed and
wrapped in aluminum foil, and its contents were allowed to equilibrate
to 40 °C in the dark in a thermostatically maintained water bath. After
the foil was removed, the pressure tube was irradiated for 20 min by
a 150 W tungsten lamp. Afterward, the pressure tube was cooled in
liquid nitrogen, the products were collected in either hexanes or decane,
and the internal standard was added. The reaction mixture was analyzed
directly by GLC in triplicate.
Solution-Phase Chlorination. All solution-phase reactions were
conducted in approximately 16 mL of solvent. The solvent was placed
in a ca. 35 mL Pyrex pressure tube, the alkane was added as described
above, and the mixture was degassed three or four times by the freeze-
pump-thaw method. Chlorine was added as described above, the
pressure tube was wrapped in aluminum foil, and its contents were
allowed to equilibrate to 40 °C. During illumination, the gas phase
occupied by the dead space volume of the pressure tube was shielded
from light. After illumination for 20 min, the pressure tube was cooled
to room temperature, the internal standard was added, and the reaction
mixture was analyzed by GLC in triplicate.
Note! Because of the hazards associated with high-pressure work,
we were especially careful to ensure that the pressures utilized in this
study did not exceed the specifications of our system (ca. 15 000 psi).
The Hastelloy reactor (the details of which are provided in ref 36) has
proven especially suitable for reactions involving corrosive materials
such as Cl2 and Br2; after several years of use, there are no visible
signs of corrosion.
Acknowledgment. Financial support from the National
Science Foundation (Grant CHE-9524986) is acknowledged and
appreciated. This paper is dedicated to Prof. Glen A. Russell
(1925-1998).
Supporting Information Available: Tables of experimental
results pertaining to the chlorination of cyclohexane, neopentane,
and 2,3-dimethylbutane and a narrative detailing the procedure
for obtaining corrected values of S(3°/1°) in SC-CO2 solvent
(8 pages, print/PDF). See any current masthead page for ordering
information and Web access instructions.
Supercritical CO2 Chlorination. The details regarding the ap-
paratus for the chlorination of cyclohexane in supercritical carbon
dioxide have been previously reported.36 Briefly, the reactor is
constructed of Hastelloy C-276 and is equipped with a sapphire window
(for irradiation) and a magnetic stir bar. For liquid substrates
(cyclohexane and 2,3-dimethylbutane), the typical procedure follows:
The appropriate volume of alkane was placed in a 1 mL ampule. The
ampule was degassed by the freeze-pump-thaw method, sealed under
JA982289E