Probing the reactivity of photoinitiators for free radical polymerization: Time-resolved infrared spectroscopic study of benzoyl radicals

Article abstract of DOI:10.1021/ja026099m

A series of substituted benzoyl radicals has been generated by laser flash photolysis of α-hydroxy ketones, α-amino ketones, and acyl and bis(acyl)phosphine oxides, all of which are used commercially as photoinitiators in free radical polymerizations. The benzoyl radicals have been studied by fast time-resolved infrared spectroscopy. The absolute rate constants for their reaction with n-butylacrylate, thiophenol, bromotrichloromethane and oxygen were measured in acetonitrile solution. The rate constants of benzoyl radical addition to n-butylacrylate range from 1.3 x 10⁵ to 5.5 x 10⁵ M^-1 s^-1 and are about 2 orders of magnitude lower than for the n-butylacrylate addition to the counterradicals that are produced by α-cleavage of the investigated ketones. Density functional theoretical calculations have been performed in order to rationalize the observed reactivities of the initiating radicals. Calculations of the phosphorus-centered radicals generated by photolysis of an acyl and bis(acyl)phosphine oxide suggest that P atom Mulliken spin populations are an indicator of the relative reactivities of the phosphorus-centered radicals. The α-cleavage of (2,4,6-trimethylbenzoyl)phosphine oxide was studied by picosecond pump-probe and nanosecond step-scan time-resolved infrared spectroscopy. The results support a mechanism in which the α-cleavage occurs from the triplet excited state that has a lifetime less than or equal to the singlet excited state.

Full text of DOI:10.1021/ja026099m

Published on Web 11/21/2002
Probing the Reactivity of Photoinitiators for Free Radical
Polymerization: Time-Resolved Infrared Spectroscopic Study
of Benzoyl Radicals
Christopher S. Colley,† David C. Grills,† Nicholas A. Besley,† Steffen Jockusch,‡
Pavel Matousek,§ Anthony W. Parker,§ Michael Towrie,§ Nicholas J. Turro,‡
Peter M. W. Gill,† and Michael W. George*,†
Contribution from the School of Chemistry, UniVersity of Nottingham, UniVersity Park,
Nottingham, NG7 2RD, United Kingdom, Department of Chemistry, Columbia UniVersity,
New York, New York 10027, and Central Laser Facility, CLRC Rutherford Appleton Laboratory,
Chilton, Didcot, Oxfordshire, OX11 0QX, United Kingdom
Received March 5, 2002
Abstract: A series of substituted benzoyl radicals has been generated by laser flash photolysis of R-hydroxy
ketones, R-amino ketones, and acyl and bis(acyl)phosphine oxides, all of which are used commercially as
photoinitiators in free radical polymerizations. The benzoyl radicals have been studied by fast time-resolved
infrared spectroscopy. The absolute rate constants for their reaction with n-butylacrylate, thiophenol,
bromotrichloromethane and oxygen were measured in acetonitrile solution. The rate constants of benzoyl
radical addition to n-butylacrylate range from 1.3 × 105 to 5.5 × 105 M-1 s-1 and are about 2 orders of
magnitude lower than for the n-butylacrylate addition to the counterradicals that are produced by R-cleavage
of the investigated ketones. Density functional theoretical calculations have been performed in order to
rationalize the observed reactivities of the initiating radicals. Calculations of the phosphorus-centered radicals
generated by photolysis of an acyl and bis(acyl)phosphine oxide suggest that P atom Mulliken spin
populations are an indicator of the relative reactivities of the phosphorus-centered radicals. The R-cleavage
of (2,4,6-trimethylbenzoyl)phosphine oxide was studied by picosecond pump-probe and nanosecond step-
scan time-resolved infrared spectroscopy. The results support a mechanism in which the R-cleavage occurs
from the triplet excited state that has a lifetime less than or equal to the singlet excited state.
Introduction
addition of the radicals (1b-6b, 1c-6c) to alkenes (initiation
step) is one of the most important reactions in free radical
Aryl ketones, such as R-hydroxy ketones (1a, 2a), R-amino
ketones (3a, 4a), and acyl and bis(acyl)phosphine oxides (5a
and 6a, respectively), have been widely used as photoinitiators
in free radical polymerizations.1-4 Upon irradiation these ketones
undergo classical type I cleavage to produce radical pairs with
a high rate constant and good efficiency (eqs 1-6).5-12 The
polymerization. Consequently, the factors controlling the rate
constants have been the subject of much experimental and
theoretical work. Laser flash photolysis with transient UV/visible
detection was employed to determine the addition rate constants
to alkenes for the radicals 1c-6c.13-16 Unfortunately, the
benzoyl radicals 1b-6b have low extinction coefficients at
wavelengths higher than 300 nm and are therefore difficult to
detect directly by their UV/visible absorption bands; e.g., the
unsubstituted benzoyl radical 1b exhibits a weak transient
absorption at 368 nm (ꢀ ≈ 150 M-1 cm-1).17,18 In addition,
overlapping optical absorptions of the initiator molecules 1a-
6a, the radicals 1c-6c, and benzoyl radicals 1b-6b cause the
* Corresponding author: e-mail Mike.George@nottingham.ac.uk.
† University of Nottingham.
‡ Columbia University.
§ CLRC Rutherford Appleton Laboratory.
(1) Crivello, J. V.; Dietliker, K. Photoinitiators for Free Radical Cationic and
Anionic Photopolymerization; John Wiley & Sons: Chichester, U.K., 1998;
Vol. 3.
(2) Dietliker, K. Radiation Curing in Polymer Science and Technology; Elsevier
Applied Science: New York, 1993; Vol. 2.
(3) Rutsch, W.; Dietliker, K.; Leppard, D.; Kohler, M.; Misev, L.; Kolczak,
U.; Rist, G. Prog. Org. Coat. 1996, 27, 227-39.
(4) Herz, C. P.; Eichler, J. Farbe Lack 1979, 85, 933.
(5) Eichler, J.; Herz, C. P.; Naito, I.; Schnabel, W. J. Photochem. 1980, 12,
225-34.
(11) Jockusch, S.; Koptyug, I. V.; McGarry, P. F.; Sluggett, G. W.; Turro, N.
J.; Watkins, D. M. J. Am. Chem. Soc. 1997, 119, 11495-501.
(12) Kolczak, U.; Rist, G.; Dietliker, K.; Wirz, J. J. Am. Chem. Soc. 1996, 118,
6477-89.
(13) Martschke, R.; Farley, R. D.; Fischer, H. HelV. Chim. Acta 1997, 80, 1363-
(6) Jockusch, S.; Landis, M. S.; Freiermuth, B.; Turro, N. J. Macromolecules
2001, 34, 1619-26.
74.
(14) Jockusch, S.; Turro, N. J. J. Am. Chem. Soc. 1999, 121, 3921-25.
(15) Sluggett, G. W.; McGarry, P. F.; Koptyug, I. V.; Turro, N. J. J. Am. Chem.
Soc. 1996, 118, 7367-72.
(7) Eichler, J.; Herz, C. P.; Schnabel, W. Angew. Makromol. Chem. 1980, 91,
39-54.
(8) Leopold, D.; Fischer, H. J. Chem. Soc., Perkin Trans. 2 1992, 513-17.
(9) Rist, G.; Borer, A.; Dietliker, K.; Desobry, V.; Fouassier, J.-P.; Ruhlmann,
D. Macromolecules 1992, 25, 4182-93.
(16) Jockusch, S.; Turro, N. J. J. Am. Chem. Soc. 1998, 120, 11773-77.
(17) Huggenberger, C.; Lipscher, J.; Fischer, H. J. Phys. Chem. 1980, 84, 3467-
74.
(10) Ruhlmann, D.; Fouassier, J.-P.; Schnabel, W. Eur. Polym. J. 1992, 28, 287-
92.
(18) Fischer, H.; Baer, R.; Hany, R.; Verhoolen, I.; Walbiner, M. J. Chem. Soc.,
Perkin Trans. 2 1990, 787-98.
9
14952
J. AM. CHEM. SOC. 2002, 124, 14952-14958
10.1021/ja026099m CCC: $22.00 © 2002 American Chemical Society

Time-Resolved IR Study of Benzoyl Radicals
A R T I C L E S
concentrations ranging from 15 to 50 mM for the TRIR experiments.
Sample purity was confirmed by gas chromatography/mass spectrometry
(GC-MS) (>95%). Acetonitrile (Aldrich, spectrophotometric grade) was
used as received.
kinetic analysis of the transient to be problematic. It has been
demonstrated that time-resolved infrared (TRIR) spectroscopy
is a powerful technique for the direct detection of benzoyl and
aliphatic acyl radicals.19-22 This is due to the fact that these
radicals generally have strong νCO absorptions, in the range
1780-1880 cm-1, with high extinction coefficients; e.g., for
the unsubstituted benzoyl radical with νCO ) 1828 cm-1 in
Diode laser-based TRIR experiments, at Nottingham23 and Colum-
bia,6 employed pulses from a Nd:YAG laser (Spectra Physics GCR-
150-30 or GCR-12) as an excitation source. The IR probe consisted of
a diode laser (Mu¨tek MDS 1150/2) fitted with a monochromator (Mu¨tek
MDS 1400S) and sample chamber with a fast mercury cadmium
telluride (MCT) detector (Kolmar Technologies KMPV11-1-J1). The
signal from the detector was amplified by a low-noise preamplifier
(Stanford Research Systems SR 560) and captured with a digital storage
oscilloscope (Tektronix TDS 360). Argon-saturated acetonitrile solutions
were flowed continuously through a 1 mm path length gastight IR cell
fitted with calcium fluoride windows.
n-hexane, ꢀ ≈ 1300 M-1 cm-1 22
.
Time-resolved step-scan Fourier transform infrared (s2-FTIR) experi-
ments were conducted with a combination of a Nicolet Magna 860
interferometer and a Nd:YAG laser (Spectra Physics GCR-12).24
Synchronization of the laser with data collection was achieved by use
of a pulse generator (Stanford DG535). The interferometer is equipped
with both an internal 100 kHz 16-bit digitizer and an external 100 MHz
12-bit digitizer (Gage 8012A). In these experiments a 1 mm photo-
voltaic MCT detector was used with a 20 MHz preamplifier. This
detector has AC and DC outputs and both outputs are digitized
simultaneously to ensure proper phase matching. The AC signal was
amplified by an external preamplifier (Stanford SR 560) to use the
full dynamic range of the digitizer. Single-sided interferograms were
obtained from one laser pulse at each mirror position. We have used
an external optical bench (Nicolet-TOM) for locating the sample cell
and MCT detector. This allows easy manipulation of the UV/visible
laser beam through the cell. The s2-FTIR spectra in this paper were
recorded at 8 cm-1 resolution with four or eight scans of the
interferometer.
The ps-TRIR facility at the Rutherford Appleton Laboratory has been
described in detail elsewhere.25,26 Briefly, part of the output from a 1
kHz, 800 nm, 150 fs, 2 mJ Ti:sapphire oscillator/regenerative amplifier
was used to pump a white light continuum seeded BBO OPA. The
signal and idler produced by this OPA were difference frequency mixed
in a type I AgGaS2 crystal to generate tunable midinfrared pulses (ca.
150 cm-1 fwhm, 1 µJ). Second harmonic generation of the residual
800 nm light provided 400 nm pulses, which were used to excite the
sample. Changes in infrared absorption were recorded by normalizing
the outputs from a pair of 64-element MCT linear array detectors on a
shot-by-shot basis.
DFT calculations were performed with Q-Chem software.27 The
EDF1 functional28 was used with the 6-31+G* basis set to calculate
the geometrical structure and harmonic vibrational frequencies and
intensities of radicals 1b-6b. EDF1 is an empirical exchange-
correlation density function that was constructed with a set of accurate
experimental data for calibration.28 The justification for using this
functional with the 6-31+G* basis set will not be discussed in this
In this paper we report a systematic TRIR study of the
reactivity of benzoyl radicals 1b-6b toward the acrylic
monomer n-butylacrylate and toward oxygen addition. We have
also measured their reactivity toward hydrogen and bromine
atom donors. The benzoyl radicals were generated by laser
photolysis of photoinitators 1a-6a. The photophysics and
photochemistry of 5a have been studied in detail by both
nanosecond and picosecond TRIR. In addition, density func-
tional theory (DFT) has been used in order to help elucidate
some of the experimental observations.
Experimental Section
(23) George, M. W.; Poliakoff, M.; Turner, J. J. Analyst 1994, 119, 551-60.
(24) Sun, X. Z.; Nikiforov, S. M.; Yang, J.; Colley, C. S.; George, M. W. Appl.
Spectrosc. 2002, 56, 31-39.
The photoinitiators, 1-hydroxycyclohexyl phenyl ketone (1a), 2-hy-
droxy-1-[4-(2-hydroxyethoxy)phenyl]-2-methyl-1-propanone (2a), 2-meth-
yl-1-[4-(methylthio)phenyl]-2-(4-morpholinyl)-1-propanone (3a), 2-benzyl-
2-(dimethylamino)-1-(4-morpholinophenyl)-1-butanone (4a), (2,4,6-
trimethylbenzoyl)diphenylphosphine oxide (5a), and bis(2,6-di-
methoxybenzoyl)(2,4,4-trimethylpentyl)phosphine oxide (6a), (all from
Ciba Specialty Chemicals) were used as received and made up to
(25) Towrie, M.; Bailey, P. D.; Barton, R.; Matousek, P.; Parker, A. W.; George,
M. W.; Grills, D. C. Central Laser Facility, Rutherford Appleton Laboratory
Annual Report 2000/2001, pp 165-67; see http://www.clf.rl.ac.uk/Report/
2000-2001/pdf/75.pdf.
(26) Towrie, M.; Grills, D. C.; Dyer, J.; Weinstein, J. A.; Matousek, P.; Barton,
R.; Bailey, P. D.; Subramaniam, N.; Kwok, W. M.; Ma, C.; Phillips, D.;
Parker, A. W.; George, M. W. Appl. Spectrosc. 2002 (in press).
(27) Kong, J.; White, C. A.; Krylov, A. I.; Sherrill, D.; Adamson, R. D.; Furlani,
T. R.; Lee, M. S.; Lee, A. M.; Gwaltney, S. R.; Adams, T. R.; Ochsenfeld,
C.; Gilbert, A. T. B.; Kedziora, G. S.; Rassolov, V. A.; Maurice, D. R.;
Nair, N.; Shao, Y.; Besley, N. A.; Maslen, P. E.; Dombroski, J. P.; Daschel,
H.; Zhang, W.; Korambath, P. P.; Baker, J.; Byrd, E. F. C.; Van Voorhis,
T.; Oumi, M.; Hirata, S.; Hsu, C.-P.; Ishikawa, N.; Florian, J.; Warshel,
A.; Johnson, B. G.; Gill, P. M. W.; Head-Gordon, M.; Pople, J. A. J.
Comput. Chem. 2000, 21, 1532-48.
(19) Vasenkov, S.; Frei, H. J. Phys. Chem. A 2000, 104, 4327-32.
(20) Sluggett, G. W.; Turro, C.; George, M. W.; Koptyug, I. V.; Turro, N. J. J.
Am. Chem. Soc. 1995, 117, 5148-53.
(21) Brown, C. E.; Neville, A. G.; Rayner, D. M.; Ingold, K. U.; Lusztyk, J.
Aust. J. Chem. 1995, 48, 363-79.
(22) Neville, A. G.; Brown, C. E.; Rayner, D. M.; Lusztyk, J.; Ingold, K. U. J.
Am. Chem. Soc. 1991, 113, 1869-70.
(28) Adamson, R. D.; Gill, P. M. W.; Pople, J. A. Chem. Phys. Lett. 1998, 284,
6-11.
9
J. AM. CHEM. SOC. VOL. 124, NO. 50, 2002 14953

A R T I C L E S
Colley et al.
Figure 2. FTIR spectrum (a) and time-resolved s2-FTIR spectra obtained
(b) 1.1 µs and (c) 4.9 µs after laser excitation (355 nm, 8 ns) of argon-
saturated acetonitrile solutions of 5a.
benefits from spectral multiplexing and fast data acquisition,
can be used to obtain these extra data.30
As a representative example of the photoinitiators, 5a was
studied in detail by both the diode laser-based and s2-FTIR
techniques. Laser flash photolysis (λex ) 355 nm) of deoxy-
genated acetonitrile solutions of 5a affords a readily detectable
transient IR absorption (Figure 1e) centered at 1800 cm-1, which
is in good agreement with the previously published transient
IR spectrum in n-hexane solution (νmax ) 1805 cm-1), and
which can be readily assigned to the 2,4,6-trimethylbenzoyl
radical (5b). The transient absorption is formed within the
response time of the TRIR detection system (<50 ns) and decays
on the microsecond time scale with second-order kinetics (τ1/2
≈ 0.5 µs).
Figure 1. Diode laser-based TRIR spectra recorded 500 ns following laser
excitation (355 nm, 8ns) of argon-saturated acetonitrile solutions of (a) 1a,
(b) 2a, (c) 3a, (d) 4a, (e) 5a, and (f) 6a.
paper since this is the subject of another publication.29 Mulliken spin
populations and frontier orbital energies were calculated by use of the
EDF1 functional with the 6-31G* basis set. The reason for using 6-31G*
rather than 6-31+G* is that in order to calculate the spin population
associated with a particular atom it is necessary to partition the electron
density between the different atoms in a molecule. The + signifies
that diffuse functions are included in the basis set. By definition, such
functions tend to spread over a number of atoms, so that assigning
electron density to particular atoms is problematic.
s2-FTIR has been employed to study the photochemistry of
5a in more detail. Figure 2a shows the ground-state FTIR of
5a in the νCO (1625-1900 cm-1) and νPO (1100-1300 cm-1
)
spectral regions in acetonitrile solution. The s2-FTIR spectrum
obtained 1.1 µs after the laser pulse (λex ) 355 nm) shows
depletion of the parent bands and generation of new absorptions
at 1800 and 1160 cm-1 (Figure 2b). The absorption at 1800
cm-1 is in good agreement with the transient IR spectrum
recorded with diode laser-based TRIR (Figure 1e) and was
assigned to the benzoyl radical 5b. The band at 1160 cm-1 is
tentatively assigned to the diphenylphosphinoyl radical 5c. The
s2-FTIR spectrum obtained 4.9 µs after the laser pulse shows a
decay of the two bands assigned to the radicals 5b and 5c and
the growth of a new absorption centered at 1698 cm-1 (Figure
2c). This new IR absorption band is assigned to mesitil, which
is formed as a secondary photoproduct by dimerization of two
2,4,6-trimethylbenzoyl radicals 5b. This assignment has been
confirmed by steady-state FTIR studies. The time evolution of
the transient IR absorption is shown in Figure 3, where the
decays of the radicals 5b and 5c and the growth of the
dimerization product, mesitil, can be seen.
Results and Discussion
Figure 1 shows the diode laser-based TRIR spectra of a range
of photoinitiators, 1a-6a, in deoxygenated acetonitrile, obtained
500 ns following 355 nm excitation. In all cases, a transient
absorption is observed between 1790 and 1820 cm-1, which
can be assigned to the corresponding benzoyl radicals (eqs 1-6).
Diode laser-based TRIR is an excellent technique for the
characterization of benzoyl radicals since the key information
can be obtained over a small spectral window. However, many
problems can only be solved by obtaining data over a much
wider IR region. Recent advances in s2-FTIR, a technique that
(29) Besley, N. A.; Colley, C. S.; Yang, J.; George, M. W.; Gill, P. M. W.
Manuscript in preparation.
(30) See Appl. Spectrosc., April 1997, Vol. 65, special issue on s2-FTIR.
9
14954 J. AM. CHEM. SOC. VOL. 124, NO. 50, 2002

Time-Resolved IR Study of Benzoyl Radicals
A R T I C L E S
Figure 4. TRIR spectra recorded (a) 10 ps and (b) 500 ps after laser
excitation (400 nm, 150 fs) of an argon-saturated acetonitrile solution of
5a.
Figure 3. Three-dimensional stack plot of time-resolved s2-FTIR spectra
obtained in (a) the νCO region and (b) the νPO region, between 1 and 5 µs
after laser excitation (355 nm, 8 ns) of an argon-saturated acetonitrile
solution of 5a.
TRIR and steady-state product studies of 5a in n-heptane have
suggested the formation of a rearrangement product (7) with a
νCO band at 1738 cm-1 20
Since this band was seen to form
.
within the detector response time (∼50 ns) and formation of 7
was persistent in the presence of the radical quencher bromo-
trichloromethane (BrCCl3), it was concluded that 7 is formed
via in-cage recombination of radicals 5b and 5c. This conclusion
was further supported by 31P chemically induced dynamic
nuclear polarization (CIDNP) studies.12 However, in the present
s2-FTIR study, there is no evidence for the formation of 7 upon
photolysis of 5a in deoxygenated acetonitrile. This rearrange-
ment product will be discussed in more detail later.
Figure 5. Transient IR absorption kinetics observed following picosecond
excitation (400 nm, 1 ps) of 5a in argon saturated acetonitrile solution,
monitored at (a) 1800 cm-1 and (b) 1670 cm-1
.
range. We have therefore used ps-TRIR to elucidate the
formation of 5b. Figure 4 shows TRIR spectra recorded 10 and
500 ps after laser excitation (λex ) 400 nm) of 5a in acetonitrile
solution. In the spectral range 1600-1850 cm-1, two absorption
changes were observed, centered at 1670 and 1800 cm-1. The
kinetics at these wavenumbers are shown in Figure 5. The band
at 1670 cm-1 corresponds to bleaching of the parent, which
occurs within the response time of the instrumental setup used
in these experiments (ca. 1 ps) and remains constant within the
observation time window of 2 ns. Following the laser pulse, a
growth of absorbance of the benzoyl radical 5b at 1800 cm-1
occurs with a time constant of 130 (( 10) ps, which then
It is well established11,12,31,32 that 5a undergoes rapid R-cleav-
age from a triplet excited state to afford the benzoyl-
phosphinoyl radical pair 5b and 5c, respectively (eq 5). By use
of picosecond pump-probe laser flash photolysis and detection
by UV absorption, the formation kinetics of the phosphinoyl
radical 5c have previously been directly observed (τ5c ) 123
ps).11 The benzoyl radical 5b could not be observed with this
technique, because of the weak absorptivity in the UV spectral
(31) Sumiyoshi, T.; Schnabel, W.; Henne, A.; Lechtken, P. Polymer 1985, 26,
141-46.
(32) Baxter, J. E.; Davidson, R. S.; Hageman, H. J.; McLauchlan, K. A.; Stevens,
D. G. J. Chem. Soc., Chem. Commun. 1987, 73-75.
9
J. AM. CHEM. SOC. VOL. 124, NO. 50, 2002 14955

A R T I C L E S
Colley et al.
Scheme 1. Energy Level Diagram for R-Cleavage of 5aa
a See ref 11 for details.
remains constant. The kinetics of the benzoyl radical formation
(5b, τ5b ) 130 ps) are in good agreement with the kinetics of
phosphinoyl radical formation (5c, τ5c ) 123 ps). Moreover,
the observations by ps-TRIR are entirely consistent with the
mechanism for R-cleavage of 5a proposed earlier,11 whereby
the rate-limiting step is intersystem crossing from the singlet
excited state to the triplet state, and thus, the measured rate
constants of growth of 5b and 5c represent the intersystem
crossing rate constants (see Scheme 1).
It is interesting to note that this ps-TRIR experiment gives
no evidence for the formation of 7 at ca. 1738 cm-1. This
observation is inconsistent with previous suggestions that 7 is
formed via in-cage recombination of radicals 5b and 5c, on a
subnanosecond time scale. The formation of 7 has been
investigated further by diode laser-based ns-TRIR spectroscopy,
and preliminary results suggest that the band at 1738 cm-1 is
formed via a multiphoton process and is therefore only observed
under conditions of high photon flux nanosecond pulses.33
The reactivities of the benzoyl radicals 1b-6b were deter-
mined by analysis of the decay kinetics of the IR absorbance
of 1b-6b with the diode laser-based TRIR spectrometer
following photolysis of 1a-6a. For example, in argon-saturated
acetonitrile solutions, laser flash photolysis (355 nm) of 5a
generated 5b, which was monitored at 1805 cm-1. This band
decayed on the microsecond time scale with clean second-order
kinetics (τ1/2 ≈ 0.5 µs), probably via radical-radical dimeriza-
tion. In the presence of radical quenchers, such as n-butylacry-
late, the lifetime of the benzoyl radicals decreases, due to the
addition of 5b to the olefin (eq 7).
Figure 6. Pseudo-first-order decay rate constants (kobs) of the benzoyl
radicals 1b-6b versus n-butylacrylate concentration, measured following
laser flash photolysis (355 nm, 8 ns) of argon-saturated acetonitrile solutions
of 1a-6a in the presence of different n-butylacrylate concentrations. The
rate constants for 2b and 3b were determined at Columbia University, and
those for 1b and 4b-6b, at Nottingham.34
At relatively high n-butylacrylate concentrations, the benzoyl
radical decay follows pseudo-first-order kinetics. The pseudo-
first-order treatment of the decay of the IR absorbance at
different n-butylacrylate concentrations, according to eq 11,
yields the second-order rate constant kacrylate, where kobs repre-
sents the observed pseudo-first-order rate constant at various
concentrations of n-butylacrylate and k0 is the rate constant for
the decay of the radical in the absence of added quencher (see
Figure 6).
kobs ) k0 + kacrylate[acrylate]
(11)
Similar experiments were performed for 1b-4b and 6b. The
n-butylacrylate addition rate constants are summarized in Table
1. The kacrylate values that have previously14,16 been determined
for 1c-6c by UV/visible flash photolysis are also included in
the table for comparison. For each radical pair, the rate constant
for the reaction of the benzoyl radical (1b-6b) with n-
butylacrylate is significantly (ca. 1-2 orders of magnitude)
lower than for the counterradical fragment (1c-6c).
An important common side reaction, which reduces the
initiation efficiency of polymerization, is addition of oxygen to
the radicals 1b-6b (eq 8). The rate constants koxygen for 5b and
(33) Following irradiation (355 nm) of 5a in deoxygenated acetonitrile, a transient
absorption is observed at ca. 1734 cm-1 with diode laser-based TRIR,
suggesting the formation of 7. However, an experiment has been performed
in which the intensity (∆absorbance) of the 1734 cm-1 band has been
monitored as a function of laser pulse fluence. The relationship between
∆absorbance and pulse energy has been found to be nonlinear. A plot of
ln (∆absorbance) vs ln (laser energy) has been fitted to a straight line with
a gradient of ca. 2-3, strongly suggesting that formation of 7 is a
multiphoton process. Therefore, this result provides a possible explanation
for the absence of a band at ca. 1734 cm-1 in the s2-FTIR and ps-TRIR
experiments, where the UV laser energy is lower than in the diode laser-
based TRIR experiment. Observation of 7 is possible with diode laser-
based TRIR since there is sufficient UV photon fluence to pump the
multiphoton process.
(34) The TRIR kinetic experiments at Columbia and Nottingham were carried
out under slightly different conditions of laser power and photoinitiator
concentration. This accounts for the differences in k0 obtained from the
fits to these data. However, the slopes of the lines (kacrylate) will be unaffected
and are therefore directly comparable. The kinetics of n-butylacrylate
addition to 4b were measured in both laboratories, giving identical rate
constants.
9
14956 J. AM. CHEM. SOC. VOL. 124, NO. 50, 2002

Time-Resolved IR Study of Benzoyl Radicals
A R T I C L E S
Table 1. νCO Frequencies and Bimolecular Rate Constantsa for
Reactions of Benzoyl Radicals (1b-6b) and Their Counterradicals
(1c-6c)
Table 2. Experimental and Calculated νCO Frequencies,
Calculated Carbonyl C and P Atom Mulliken Spin Population
(MSP) Values, and ∆Eacrylate for Radicals 1b-6b, 5c, and 6c
radical
ν
CO (cm-1
)
kacrylate (M-1 s-1
)
kPhSH (M-1 s-1
)
kBrCCl3 (M-1 s-1
)
kO2 (M-1 s-1
)
ν
CO expt νCO calc MSP carbonyl
calcd
calcd
calcd
radical (cm-1
)
(cm-1
)
C or P (au)
∆EHOMO (au) ∆ELUMO (au) ∆Etotal (au)
1b
1c
2b
2c
3b
3c
4b
4c
5b
5c
6b
6c
1818
1805
1805
1789
1800
1803
2.7 × 105
1.3 × 107 b
3.5 × 105
1.3 × 107 b
5.5 × 105
6.1 × 106 b
3.6 × 105
6.1 × 106 b
1.8 × 105
2.8 × 107 c
1.3 × 105
1.5 × 107 c
1.8 × 107
1.8 × 108
5.4 × 109 b
6.6 × 109 b
4.3 × 109 b
1b
2b
3b
4b
5b
6b
5c
6c
1818
1813
1807
1806
1795
1800
1812
0.64
0.65
0.65
0.65
0.64
0.65
0.48
0.42
5.8
5.3
5.4
4.8
4.6
4.1
4.5
3.1
5.8
6.5
6.4
7.7
7.1
10.0
2.5
2.0
11.6
11.8
11.8
12.5
11.7
14.1
7.0
1805
1805
1789
1800
1803
5.4 × 107
3.9 × 106
4.2 × 106 c
7.1 × 106
5.8 × 108
4.3 × 109 b
3.0 × 109
5.1
6.9 × 108 c 4.2 × 109 c
1.1 × 108
3.3 × 109
2.4 × 108 c 2.6 × 109 c
5c). It has been suggested20 that this implies that the transition
states for additions of, e.g., 5b and 5c to unsaturated compounds,
such as n-butylacrylate, may be situated slightly further along
the reaction coordinate than the transition states for atom
abstraction reactions, and thus steric and electronic effects are
more important.
a Reactions with n-butylacrylate, thiophenol, bromotrichloromethane, and
oxygen in acetonitrile solution at 298 K were performed. Rate constant
values are (10%. b Reference 14. c Reference 16.
6b were determined by monitoring the radical decay as a
function of oxygen concentration (0.85-3.4 mM). The variable
oxygen concentrations were achieved by saturation of the
acetonitrile solutions with air at different pressures. Determi-
nation of koxygen for 6b was more complex. The absorption
assigned to the benzoyl radical decayed with double-exponential
kinetics. The slower exponential decay was independent of
oxygen concentration, whereas the faster exponential decay was
found to be dependent on the oxygen concentration. These
double-exponential decay kinetics can be explained in terms of
a spectral overlap between the benzoyl radical and the benzoyl
peroxy radical, which is the addition product of oxygen to the
benzoyl radical (eq 8).
Other common reactions, which reduce the initiation ef-
ficiency of polymerization, are atom abstractions from materials
that are required to be present in a photocuring formulation.
To distinguish the reactivity between the different benzoyl
radicals toward atom abstraction, we chose model compounds
with high atom donor tendencies, i.e., thiophenol (PhSH) and
bromotrichloromethane (BrCCl3). The kPhSH values determined
Density functional theory (DFT) has been used in order to
give further insight into these experimental results. Table 2
shows that there is very good agreement between the experi-
mental and DFT calculated IR frequencies. However, it should
be noted that this agreement (average difference between
experiment and theory is (3.8 cm-1) may be somewhat
fortuitous since the IR spectra are calculated for an isolated
single molecule, whereas the experimental spectra are measured
in a polar solvent. Experimentally the νCO frequencies decrease
in the order 1b > 2b ≈ 3b > 6b > 5b > 4b but the calculated
νCO frequencies decrease in the order 1b > 6b > 2b > 3b >
5b > 4b. Thus, it appears that the νCO frequency calculated for
6b is slightly too high relative to the other radicals. A possible
reason for this is that in 6b two different local minima can be
attained: one in which the two methyl groups are eclipsed in
the plane of the aromatic ring and another in which they are
staggered in the plane of the ring. The calculations show that
the barrier for reorientation of the methyl groups is relatively
small. The νCO frequencies were found to be sensitive to the
conformation (eclipsed conformation, νCO ) 1812 cm-1; stag-
gered conformation, νCO ) 1795 cm-1).
for 1b, 4b, 5b, and 6b and the kBrCCl values determined for 1b,
3
4b, and 6b are shown in Table 1. Measurement of the rate
constant for addition of 5b to BrCCl3 has been attempted but
the kinetics were more complicated.35 A previously measured
This agreement between the experimental and theoretical νCO
frequencies provides some confidence that the EDF1 density
functional adequately describes the electronic structure of the
radicals. A number of factors will govern the reactivity of the
radicals 1b-6b, including steric, entropic, and electronic effects.
DFT calculations can provide information about the electronic
structure of the radicals that cannot be obtained experimentally.
Consequently, it is of interest to see if DFT can rationalize and
potentially predict the reactivities of the radicals. While factors
such as steric effects are difficult to quantify, they can often be
regarded as constant when comparing the reactivities of a
number of sterically similar radicals. The extent to which the
orbitals of two reacting molecules can affect the rate of a
reaction can be estimated by perturbation theory. This interaction
can be considered to raise or lower the barrier to reaction. The
change in the barrier height is given by36
kPhSH value for 5c and kBrCCl values for 5c and 6c are also
included in the table. For each benzoyl radical, kPhSH is
3
approximately 1 order of magnitude less than kBrCCl . This
3
observation is consistent with previous work,15,20 which has
identified a qualitative relationship between the rate constant
and the strength of the bond that is broken in the atom
donor: BrCCl3 (56 kcal mol-1) > PhSH (75 kcal mol-1). In
contrast to the reactivity toward n-butylacrylate, the atom
abstraction rate constants for each radical fragment within a
radical pair appear to be strikingly similar to each other (e.g.,
kPhSH ) 3.9 × 106 M-1 s-1 for 5b and 4.2 × 106 M-1 s-1 for
(35) Even at low BrCCl3 concentrations, the kinetic trace at 1800 cm-1 decayed
to a residual absorption. With increasing BrCCl3 concentration, the amount
of residual absorption increased further. In an attempt to understand this
kinetic behavior, the s2-FTIR spectrum of 5a was measured in deoxygenated
acetonitrile, in the presence of 5 mM BrCCl3. It was found that the product
of the atom abstraction, trimethylbenzoyl bromide, has a νCO band at 1800
cm-1, which overlaps with 5b. However, the mechanism of formation of
trimethylbenzoyl bromide is unclear since its yield is highly dependent
upon the concentration of BrCCl3. Furthermore, a new photoproduct with
a strong absorption at ca. 1734 cm-1 was observed only when BrCCl3 was
present. This is consistent with the formation of the rearrangement product
7 during these trapping experiments. The precise mechanism is still unclear
and further experiments are in progress to fully understand this process.
2( c c â )2
∑
ra sb ab
occ unocc
occunocc
ab
∆E )
-
(12)
∑∑ ∑∑
Er - Es
r
s
s
r
9
J. AM. CHEM. SOC. VOL. 124, NO. 50, 2002 14957

A R T I C L E S
Colley et al.
where cra is the coefficient of atomic orbital a in molecular
orbital r, where r refers to the molecular orbitals on one molecule
and s refers to the molecular orbitals on the other, âab is the
resonance integral, and Er is the energy of molecular orbital r.
The greater ∆E is, the more the reaction barrier is lowered and
consequently the rate increased. In a frontier orbital picture of
the reaction between a radical and molecule, only the interaction
between the SOMO (singly occupied molecular orbital) of the
radical and both the HOMO (highest occupied molecular orbital)
and LUMO (lowest unoccupied molecular orbital) of the
reacting molecule are considered. If the resonance integral is
assumed to be constant for all the radicals, then a measure of
the relative changes in ∆EHOMO can be estimated by
Time-resolved electron spin resonance spectroscopy
(TR-ESR) has been previously used to study phosphinoyl
radicals, including 5c and 6c.15,16 The high reactivity of
phosphinoyl radicals has been attributed to a high degree of
s-character and spin localization on the P atom. These charac-
teristics are reflected by large 31P hyperfine coupling constants
(hfcc) in the TR-ESR spectra, which provides an experimental
indicator of the extent of spin localization. The lower 31P hfcc
of 6c (31P hfcc ) 286 G) compared with 5c (31P hfcc ) 369 G)
indicates that the P atom of 6c possesses a lower degree of spin
localization, which is considered to result in the lower reactivity
toward unsaturated compounds and atom donors. The calculated
P atom MSPs for radicals 5c and 6c are 0.48 and 0.42,
respectively, which agree with the experimental observations
where there is a significant difference in reactivity between these
radicals. Furthermore, the computed ∆Etotal values indicate that
this does result in a greater rate. It should be noted that the
computed ∆Etotal for the phosphinoyl radicals should not be
compared to those for the benzoyl radicals since eq 13 is only
valid for comparisons between similar radicals.
MSP2
|EHOMO - ESOMO
∆EHOMO
)
(13)
|
and similarly for ∆ELUMO where MSP is the Mulliken spin
population of the carbonyl carbon. If ∆EHOMO is significantly
greater than ∆ELUMO, then the reaction involving the SOMO
and the HOMO will be favored, and similarly, if ∆ELUMO is
greater the SOMO will interact with the LUMO.
Conclusions
A series of substituted benzoyl radicals (1b-6b) have been
generated by laser flash photolysis of R-hydroxy ketones (1a,
2a), R-amino ketones (3a, 4a), and acyl and bis(acyl)phosphine
oxides (5a and 6a, respectively). The benzoyl radicals were
characterized by TRIR in acetonitrile solution and their absolute
rate constants for addition to n-butylacrylate, thiophenol,
bromotrichloromethane, and oxygen were determined. The rate
constants of benzoyl radical addition to n-butylacrylate range
from 1.3 × 105 to 5.5 × 105 M-1 s-1. DFT calculations have
been used to rationalize the relative rates of these reactions.
These calculations show that from 1b to 6b there is an increasing
tendency of the radical to react via the LUMO of n-butylacrylate.
The rate constants for n-butylacrylate addition to radicals 1b-
6b are about 2 orders of magnitude lower than for the
counterradicals (1c-6c), suggesting that the benzoyl radicals
are less efficient in initiation of free radical polymerization.
Analysis of the spin population at the phosphorus atom supports
previous experimental observations of TR-ESR spectra that
greater localization of the spin on the phosphorus atom results
in faster rates. The experiments outlined in this paper have
demonstrated the power of TRIR to probe photopolymerization
initiation. Further work is in progress to fully elucidate the
propagation steps following the initiation process by use of
TRIR.
The electron spin density is a complicated function of position
and it is conceptually helpful to partition it into atomic
components. Unfortunately, because there is no rigorous quan-
tum mechanical definition of an atom within a molecule, a
unique partition does not exist and it is necessary to adopt one
of the many alternative schemes that have been suggested over
the years. One of the most popular of these, and the one that
we have chosen, is the Mulliken spin population analysis.
Although it is not a sophisticated method, the assumptions of
Mulliken analysis are consistent with those of the semiquanti-
tative frontier molecular orbital theory that we use. Table 2
shows the calculated MSP on the carbonyl C atom, ∆EHOMO
,
and ∆ELUMO of 1b-6b for the reaction with n-butylacrylate.
There appears to be little difference in MSP values between
the radicals 1b-6b. This is in agreement with the experimental
results where a similar reactivity of 1b-6b was found. The
observed small difference in reactivity of 1b-6b probably does
not arise from a large variation in the spin density at the carbonyl
C atom. However, ∆EHOMO and ∆ELUMO do provide some
insight into the changing nature of the reaction. For 1b the
interactions of the SOMO with the HOMO and LUMO are
found to be equivalent. However, for 2b-6b the calculations
suggest that there is an increasing tendency for reaction via the
LUMO. It is probably too much to expect this type of relatively
simple analysis to provide accurate predictions of the relative
rates of these reactions. However, the calculations generally
agree with experiment. The major anomaly is 6b, which is
predicted to have the fastest rate, and for this radical steric
effects may be important. To model these effects requires more
detailed calculations examining the transition states of the
reactions. It should be noted that the Mulliken spin populations
were found to be largely insensitive to changes in molecular
conformation.
Acknowledgment. The authors at Columbia thank the
National Science Foundation and Ciba Specialty Chemicals, Inc.,
for their generous support of this research and Dr. M. Kunz
(Ciba Specialty Chemicals) for samples 1a-6a. The authors at
Nottingham thank the EPSRC and Nicolet Instruments for
financial support. We are especially grateful to Dr. C. Hayes
for many helpful discussions.
Supporting Information Available: Equilibrium geometries
of radicals 1b-6b, 5c, and 6c (PDF). This material is available
free of charge via the Internet at http://pubs.acs.org.
(36) Fleming, I. Frontier Orbitals and Organic Chemical Reactions; John Wiley
& Sons: Chichester, U.K., 1978.
JA026099M
9
14958 J. AM. CHEM. SOC. VOL. 124, NO. 50, 2002