Evaluation of N-aromatic maleimides as free radical photoinitiators: A photophysical and photopolymerization characterization

Article abstract of DOI:10.1021/jp002811v

Photopolymerizable compositions were prepared using acrylate monomers in combination with various N-aromatic maleimides. N-aromatic maleimides were segregated into two groups: those that could adopt a planar conformation and those that could not adopt a planar conformation. The maleimides were characterized using single-crystal X-ray diffraction spectroscopy, laser flash photolysis spectroscopy, UV-vis absorption spectroscopy, and photodifferential scanning calorimetry. Planar N-aromatic maleimides were found to have a low relative excited-state triplet yield, showing significant shift of the primary maleimide UV absorption band with changes in solvent polarity, and did not initiate free radial polymerization upon direct UV excitation. Twisted N-aromatic maleimides have a higher relative triplet yield, show negligible shift of the primary maleimide UV absorption band, with solvent polarity, and initiate free radical polymerization upon direct excitation. Addition of benzophenone was found to dramatically increase the initiation efficiency of both planar and twisted N-aromatic maleimides to levels approaching that of conventional cleavage photoinitiators.

Full text of DOI:10.1021/jp002811v

J. Phys. Chem. B 2001, 105, 2707-2717
2707
Evaluation of N-Aromatic Maleimides as Free Radical Photoinitiators: A Photophysical and
Photopolymerization Characterization
Chris W. Miller,† E. Sonny Jo1nsson,‡ Charles E. Hoyle,*,† Kalyanaraman Viswanathan,† and
Edward J. Valente§
School of Polymers and High Performance Materials, The UniVersity of Southern Mississippi,
Hattiesburg, Mississippi 39406, Fusion UV-Curing Systems, Gaithersburg, Maryland, and
Department of Chemistry, Mississippi College, Clinton, Mississippi 39058
ReceiVed: August 2, 2000; In Final Form: NoVember 27, 2000
Photopolymerizable compositions were prepared using acrylate monomers in combination with various
N-aromatic maleimides. N-aromatic maleimides were segregated into two groups: those that could adopt a
planar conformation and those that could not adopt a planar conformation. The maleimides were characterized
using single-crystal X-ray diffraction spectroscopy, laser flash photolysis spectroscopy, UV-vis absorption
spectroscopy, and photodifferential scanning calorimetry. Planar N-aromatic maleimides were found to have
a low relative excited-state triplet yield, showing significant shift of the primary maleimide UV absorption
band with changes in solvent polarity, and did not initiate free radial polymerization upon direct UV excitation.
Twisted N-aromatic maleimides have a higher relative triplet yield, show negligible shift of the primary
maleimide UV absorption band, with solvent polarity, and initiate free radical polymerization upon direct
excitation. Addition of benzophenone was found to dramatically increase the initiation efficiency of both
planar and twisted N-aromatic maleimides to levels approaching that of conventional cleavage photoinitiators.
Introduction
N-substituted maleimides have been known in organic and
polymer chemistry for many years. Recently we have shown
that N-aliphatic maleimides and N-alkylmaleimides (Figure 1)
may be used as free radical photoinitiators of acrylate mono-
mers.1 N-aliphatic maleimides initiate free radical polymerization
through a hydrogen atom abstraction process in the presence
of hydrogen atom donors such as alcohols and ethers (Scheme
1). In the presence of hydrogen atom donors such as amines
and vinyl ethers, radicals are produced through an electron-
transfer/proton-transfer reaction sequence.2,3 Both direct abstrac-
tion and electron/proton transfer result in the formation of two
radicals capable of initiating polymerization: a succinimido
radical on the maleimide residue, and a radical on the hydrogen
atom donor. In comparison, conventional hydrogen atom
abstraction type photoinitiators produce only a single initiating
radical as a result of the hydrogen-transfer process. Real-time
Fourier transform infrared spectroscopy and real-time UV-vis
spectroscopy have shown that when N-aliphatic maleimides are
used as photoinitiators in acrylic systems, essentially all of the
maleimide chromophores are consumed in the polymerization
process.2,4 Three possible reactions may account for consump-
tion of the maleimide chromophores: (1) initiation of polymer
chains through hydrogen atom abstraction or electron/proton-
transfer processes, (2) copolymerization with the acrylic mono-
mers, and (3) maleimide [2+2] cycloaddition reactions.5-11 At
the low concentrations of maleimide used to initiate free radical
acrylate polymerization, initiation and copolymerization reac-
tions are the dominant pathways for maleimide consumption.
Figure 1. Structure of N-aliphatic maleimides and N-alkylmaleim-
ides: R ) alkyl or aliphatic group.
SCHEME 1. Proposed Initiation Mechanisms for
N-Alkylmaleimides
While N-aliphatic and N-alkylmaleimides have been shown
to initiate free radical polymerization efficiently, they are
typically difficult to synthesize in high purity and high yield.
N-aromatic maleimides are much easier to synthesize. The
objective of the present work was to synthesize a number of
substituted N-aromatic maleimides and evaluate their utility as
free radical photoinitiators for acrylic-based formulations. To
this end, 16 N-aromatic maleimide (ARMI) and isomaleimide
derivatives with various combinations of substituents were
prepared (Figures 2 and 3) and evaluated. The small-molecule
photochemical and photophysical properties of the N-aromatic
maleimides were characterized using UV-vis absorption spec-
troscopy, laser flash photolysis transient absorption spectros-
copy, and low-temperature emission spectroscopy. The relative
photopolymerization efficiency of a difunctional acrylate when
initiated by various N-aromatic maleimides was compared using
photodifferential scanning calorimetry. Trends from the small-
molecule studies and photopolymerizations were correlated with
torsion angles between the phenyl ring and imide ring in the
* To whom correspondence should be addressed. Phone: (601) 266-
4873. Fax: (601) 266-5504.
† The University of Southern Mississippi.
‡ Fusion UV-Curing Systems.
§ Mississippi College.
10.1021/jp002811v CCC: $20.00 © 2001 American Chemical Society
Published on Web 03/21/2001

2708 J. Phys. Chem. B, Vol. 105, No. 14, 2001
Miller et al.
excess DMSO. The aqueous suspension was allowed to stir
overnight and was filtered. The pale white solid thus obtained
was dried under vacuum. Maleimide yields were typically 50-
97% before purification. Products from both steps were analyzed
via 1H and 13C NMR on a Bruker 200 MHz NMR using
deuterated DMSO as the solvent and TMS as the internal
reference. The maleamic acids were characterized by a broad
peak due to the acid proton near 10-14 ppm, a singlet due to
the amide proton near 8-10 ppm, and two doublets due to the
ene protons near 6-7 ppm, relative to TMS in the proton
spectra. The maleamic acids were also characterized by two
peaks near 165 ppm due to the amide and acid carbonyls in the
carbon spectra. The maleimides were characterized by the
absence of the acid and amide proton peaks, and the single ene
proton peak shifted to about 7.6 ppm in the proton spectrum.
In the carbon spectrum, the maleimide typically shows a single
carbonyl peak near 170 ppm. Peak shifts for each of the
N-aromatic maleimides synthesized follow.
Figure 2. N-aromatic maleimide general structure. Substituents were
chosen from among those listed.
1
N-(2,6-Diisopropylphenyl)maleimide (26DIPPMI): H NMR
(d6-DMSP, TMS; δ, ppm) 1.06 doublet, 2.57 multiplet, 7
multiplet, 7.5 multiplet; 13C NMR (d6-DMSP, TMS; δ, ppm)
23.57 singlet, 28.56 singlet, 123.74 singlet, 126.51 singlet,
129.89 singlet, 134.81 singlet, 147.16 singlet, 170.83 singlet.
N-(2-tert-Butylphenyl)maleimide (2tBPMI): 1H NMR (d6-
DMSP, TMS; δ, ppm) 1.22 singlet, 7.1-7.6 multiplet; 13C NMR
(d6-DMSP, TMS; δ, ppm) 31.20 singlet, 127.1 singlet, 128.1
singlet, 129.5 singlet, 131.9 singlet, 135.5 singlet, 149.0 singlet,
171.2 singlet.
Figure 3. Isomaleimide structures.
molecules calculated from single-crystal X-ray diffraction crystal
structures. Photopolymerization of the difunctional monomer
was also conducted for samples with maleimide, amine, and
benzophenone present.
Experimental Section
1
N-(2-Phenylphenyl)maleimide (2PPMI): H NMR (d6-DMSP,
1. Materials and Synthesis. N-Phenylmaleimide (PMI) and
N-methylmaleimide (MMI) were obtained from Aldrich Chemi-
cal Co. and purified by recrystallization and/or sublimation.
N-(4-Methylphenyl)maleimide (4CH3PMI), N-(4-methoxy-
phenyl)maleimide (4OCH3PMI), and N-(4-cyanophenyl)male-
imide (4CNPMI) were prepared by standard procedures. All
other ARMI derivatives were prepared using the general
procedure given below and were purified by sublimation and/
or recrystallization. 1,6-Hexanediol diacrylate (HDDA) was
obtained from UCB Radcure and Aldrich and used as received.
Poly(ethylene glycol)-400-diacrylate (PEG400DA) and all other
reagents, solvents, and materials were obtained from and used
as received unless otherwise specified.
A general synthesis: Maleic anhydride (typical ∼33 g) was
placed in diethyl ether (100 mL) and allowed to dissolve. The
requisite aromatic amine (typically ∼41 mL) was added
squirtwise to the stirring solution. The reaction was allowed to
proceed at room temperature overnight, and a pale white
precipitate was observed. (Note: In cases where the aniline is
unsubstituted or is substituted with an electron-donating sub-
stituent, it is necessary cool the anhydride solution in an ice
bath during addition of the substituted aniline.) The stirring
solution was then warmed for several hours and filtered, yielding
a pale white solid. The mother liquor was then combined with
the wash liquor and allowed to stir again, and additional product
was filtered off. The solids were then combined and dried in a
vacuum. Typical yields of the maleamic acid were 90-98%.
The maleamic acid was then placed in a round-bottomed flask
with 100 mL of toluene and 25 mL of DMSO, and allowed to
dissolve. Concentrated sulfuric acid (1.8 mL) was then added
to the stirring mixture, which was then heated to ∼130 °C. The
reaction mixture was allowed to reflux for 4 h or until an
azeotrope was no longer observed in the Dean-Stark trap used
to remove water from the reaction mixture. Excess toluene was
removed via vacuum distillation. The mixture was then added
to stirring distilled water to precipitate the imide and remove
TMS; δ, ppm) 3.33 doublet, 7.1 multiplet, 7.4 multiplet; 13C
NMR (d6-DMSP, TMS; δ, ppm) 127.6 singlet, 127.7 singlet,
128.4 singlet, 129.0 singlet, 129.6 singlet, 130.0 singlet, 130.4
singlet, 134.8 singlet, 138.3 singlet, 141.0 singlet.
1
N-(2-Fluorophenyl)maleimide (2FPMI): H NMR (d6-DMSP,
TMS; δ, ppm) 7.27 singlet. 7.0-7.6 multiplet; 13C NMR (d6-
DMSP, TMS; δ, ppm) 116.2 singlet, 116.5 singlet, 129.9 singlet,
130.6 singlet, 130.9 singlet, 131.1 singlet, 135.2 singlet, 169.2
singlet.
N-(2-Chlorophenyl)maleimide (2ClPMI): 1H NMR (d6-
DMSP, TMS; δ, ppm) 7.28 singlet, 7.0-7.7 multiplet; 13C NMR
(d6-DMSP, TMS; δδ, ppm) 128.2 singlet, 129.2 singlet, 129.9
singlet, 131.0 singlet, 131.5 singlet, 132.2 singlet, 135.0 singlet.
N-(2-Bromophenyl)maleimide (2BrPMI): 1H NMR (d6-
DMSP, TMS; δ, ppm) 7.28 singlet, 7.5 multiplet, 7.8 multiplet;
13C NMR (d6-DMSP, TMS; δ, ppm) 122.9 singlet, 128.7 singlet,
131.0 singlet, 131.3 singlet, 131.6 singlet, 133.0 singlet, 135.1
singlet.
N-(2-Iodophenyl)maleimide: 1H NMR (d6-DMSP, TMS; δ,
ppm) 7.0-7.3 multiplet, 7.3-7.6 multiplet, 8.0 doublet; 13C
NMR (d6-DMSP, TMS; δ, ppm) 100.4 singlet, 129.3 singlet,
131 singlet, 134.8 singlet, 135.1 singlet, 139.1 singlet, 169.2
singlet.
N-(2-Bromo-3,5-bis(trifluoromethyl)phenyl)maleimide
(2Br35CF3PMI): 1H NMR (d6-DMSP, TMS; δ, ppm) 7.40
singlet, 8.30 singlet, 8.45 singlet; 13C NMR (d6-DMSP, TMS;
δ, ppm) 119.2 singlet, 124.6-126.9 multiplet, 129.3-132.3
multiplet, 135 doublet. (There is possible contamination of NMR
sample by residual toluene.)
1
N-(4-Trifluoromethylphenyl)maleimide (4CF3PMI): H NMR
(d6-DMSP, TMS; δ, ppm) 7.25 singlet, 7.62 doublet, 7.88
doublet; 13C NMR (d6-DMSP, TMS; δ, ppm) 125.9 singlet,
126.9 doublet, 134.9 singlet, 169.5 singlet.
1
N-(2-Trifluoromethylphenyl)maleimide (2CF3PMI): H NMR
(d6-DMSP, TMS; δ, ppm) 7.30 singlet, 7.6-8.0 multiplet; 13
C

Aromatic Maleimides as Free Radical Photoinitiators
J. Phys. Chem. B, Vol. 105, No. 14, 2001 2709
NMR (d6-DMSP, TMS; δ, ppm) 127.2 singlet, 130.4 singlet,
132.4 singlet, 133.9 singlet, 135.4 singlet, 169.7 singlet.
N-(2-Trifluoromethylphenyl)citraconimide (2CF3PCI): 1H
NMR (d6-DMSP, TMS; δ, ppm) 2.11 singlet, 6.90 triplet, 7.6-
7.3 multiplet; 13C NMR (d6-DMSP, TMS; δ, ppm) 120.3 singlet,
125.7 singlet, 127.1 singlet, 127.2 singlet, 127.9 singlet, 128.3
singlet, 128.5 singlet, 129.1 singlet, 129.9 singlet, 130.3 singlet,
132.3 singlet, 133.9 singlet, 146.4 singlet, 169.4 singlet, 170.4
singlet.
Figure 4. Planar and twisted conformations of ARMIs.
shuttered 450 W medium-pressure mercury lamp. On-sample
light intensities were maximal at about 50 mW/cm2, with lower
intensities obtained by using neutral density filters or by
increasing the distance between the lamp and DSC head.
Specific spectral bands were isolated using band-pass filters,
typically 313 and 365 nm. Samples of 2 µL were added to
specially crimped aluminum DSC pans, leading to a theoretical
sample thickness of about 190 µm, and were purged with
nitrogen for 2.5 min prior to and during irradiation. Exotherm
peak maxima are proportional to the relative maximum rates
of polymerization, and peak integrations are proportional to the
monomer conversion. Some exotherm data were acquired using
a Mettler-Toledo Star System DSC modified by the addition of
a Hammamatsu high-pressure xenon-doped Hg light source and
a quartz sample cell cover. The light intensity was set at 50
mW/cm2 for the experiments with the Mettler system.
1
N-(2-Trifluoromethylphenyl)itaconimide (2CF3PII): H NMR
(d6-DMSP, TMS; δ, ppm) 3.76 singlet, 4.0 doublet, 6.2 singlet,
6.7 triplet, 8.0-8.3 multiplet; 13C NMR (d6-DMSP, TMS; δ,
ppm) 34.1 singlet, 120.9 singlet, 127.1 singlet, 127.6 singlet,
130.4 singlet, 131.5 singlet, 134.0 singlet, 168.6 singlet, 173.2
singlet.
2. UV-Vis Absorption Spectroscopy. UV-vis absorption
spectra were measured using a Cary 500 Scan UV-vis near-
IR spectrophotometer by Varian. Solvents for the UV-vis
spectral analyses were primarily Burdick and Jackson brand,
except for some cases in which Optima acetonitrile was used.
Standard quartz cells were used.
3. Laser Flash Photolysis (LFP) Spectroscopy. LFP was
performed using a Continuum Surelite Nd:YAG laser excitation
source and a UV-vis absorption/emission monitoring system
from Applied Photophysics. Excitation wavelengths of 355 and
266 nm were used with laser energies of approximately 160
and 70 mJ, respectively. Samples were prepared in CH3CN
solution to have an optical density (OD) of 1.0 at the excitation
wavelength in 1 cm quartz cells. The UV-vis absorbance at
specific wavelengths was measured for each sample before the
laser pulse (∼15 ns duration) and as a function of time after
the laser pulse. Exponential fits of the resulting absorbance decay
curves allowed calculation of the lifetimes of the transient
excited-state species. The transient absorbance and lifetimes
were measured as a function of monitoring wavelength, allowing
reconstitution of the time-resolved UV-vis absorption spectrum
of the transient excited-state species. Measurement of the
transient lifetime at the transient absorption peak maximum as
a function of quencher concentration allows calculation of the
Stern-Volmer quenching rate constant (kq) through
Results and Discussion
Initial examination of N-aromatic maleimides in our labora-
tories led to the presumption that they were incapable of
initiating free radical polymerization. Literature reports14,15 that
the UV-vis absorption spectra and fluorescence emission
spectra of N-aromatic naphthalimides closely resembled the
absorption and emission spectra of N-alkylnaphthalimides upon
substitution of a bulky group in the ortho position of the phenyl
ring led to the presumption that forcing the maleimide ring to
be antiplanar with respect to the phenyl ring (Figure 4) might
change the photophysical and photochemical properties of
ARMIs such that they would initiate polymerization similarly
to N-aliphatic maleimides. To test this hypothesis, photophysical
and photochemical studies of a series of ARMIs were performed
and the results correlated with relative photoinitiation efficien-
cies inferred from photo-DSC polymerization exotherms.
1. Single-Crystal X-ray Diffraction. SCXD analysis of
single crystals of all of the ARMIs (except for 4CF3PMI, which
could not be crystallized to give a single crystal suitable for
X-ray analysis) allowed calculation of the torsion angles between
the maleimide ring and the phenyl ring in the crystalline state.
Torsion angles (τ) and experimental UV-vis absorption spectra
for five of these compounds (PMI, 2FPMI, 2ClPMI, 2BrPMI,
and 2IPMI) were recently compared16 to torsion angles and
calculated UV absorption spectra based upon energy-minimized
conformations from semiempirical AM1 and ab initio SCF,
DFT, and MP2 calculations. Conformational energy minima
were compared to conformational energies calculated at τ ) 0°
and τ ) 90°. Good correlation was found between the
experimentally observed torsions and UV absorption bands and
the calculated minimized torsions and UV absorption bands.
Ortho substituents larger than fluorine were found to signifi-
cantly increase the conformational energy, indicating that the τ
) 0° conformation is not possible without significant molecular
distortion. Torsion angle data previously reported for these five
ARMIs are included for comparison with the remaining nine
ARMI torsion analyses. 2FPMI exhibited a disorder in the
position of the fluorine that may occupy either of the equivalent
ortho positions in one of the two conformers in the repeating
unit cell. This disorder is not uncommon for o-fluorophenyl
compounds, and the torsion angles for the two major conformers
τ0/τ ) 1 + kqτo[Q]
(1)
where τo is the lifetime in the absence of quencher, τ is the
lifetime at a given quencher concentration [Q], k ) 1/τ is the
rate for transient decay at a given [Q], k0 ) 1/τ0 is the rate for
transient decay in the absence of quencher, and kq is the rate
constant for quenching (L mol-1 s-1).
4. Single-Crystal X-ray Diffraction (SCXD). SCXD was
performed using a Siemens R3m/V single-crystal X-ray dif-
fractometer, with a Mo KR X-ray source producing radiation
with a wavelength of 7.1073 × 10-11 m. The positions of all
non-hydrogen atoms were determined with direct methods (G.
Sheldrick, 1993)12 and refined by full-matrix least-squares (G.
Sheldrick, 1997).13 Hydrogen atom positions were found in
difference Fourier maps and placed in calculated positions. The
resolution of the determinations in each case was below 0.7 Å,
and hydrogen atoms in C-H bonds were resolved, though they
are thermal and the distances are affected by the usual distortions
associated with the nonspherical character of the hydrogen atom
electron density. Complete structural data including cell con-
stants and bond distortions will be published separately.
5. Photo-Differential Scanning Calorimetry (Photo-DSC).
Photo-DSC was typically performed on a Perkin-Elmer DSC-7
modified with quartz windows in the sample head cover and a

2710 J. Phys. Chem. B, Vol. 105, No. 14, 2001
Miller et al.
TABLE 1: Torsion Angles Calculated from SCXD Crystal
Structure Determinations
TABLE 2: Longest Wavelength UV Absorption Peak
Maxima and Molar Extinction Coefficients for MMI and
ARMIs as a Function of Solvent Polaritya
compd
τ (deg)
compd
τ (deg)
peak max
(nm)
ꢀ at peak max
Ortho-Substituted ARMIs
compd
MMI
solvent
(L mol-1 cm-1
)
26DIPPMI
84.5
85.4
88.6
86.6
79.6
2FPMI
2ClPMI
2BrPMI
2IPMI
54.1, 66.8
66.1
67.8
26DIPPMI (benzene solvate)
CHCl3
CH2Cl2
CH3CN
299
298
298
826
821
732
2tBPMI
2CF3PMI
2PPMI
83.9
2Br35CF3PMI 88.9
Twisted ARMIs
Unsubstituted and Para-Substituted ARMIs
2CF3PMI
26DIPPMI
2ClPMI
CHCl3
CH2Cl2
CH3CN
CHCl3
CH2Cl2
CH3CN
CHCl3
CH2Cl2
CH3CN
CHCl3
CH2Cl2
CH3CN
286
286
287
293
293
293
290
289
289
288
287
287
639
568
488
597
551
425
522
519
491
582
538
484
PMI
4OCH3PMI
49.5
63.7
4CF3PMI
a
Isomaleimide Derivatives
85.5 2CF3PII
2CF3PCI
85.5
a Suitable single crystals could not be obtained.
2BrPMI
Planar ARMIs
318
PMI
CHCl3
429
479
458
620
508
513
542
508
471
244
290
306
CH2Cl2
CH3CN
CHCl3
314
307
300
299
294
313
310
304
340
331
315
2FPMI
CH2Cl2
CH3CN
CHCl3
4CF3PMI
4OCH3PMI
CH2Cl2
CH3CN
CHCl3
CH2Cl2
CH3CN
a CHCl3, CH2Cl2, and CH3CN have dielectric constants of 4.81, 9.08,
and 37.5, respectively.
as well as MMI were characterized using UV-vis spectral
analysis as a function of solvent polarity. Data for 2IPMI and
2Br35CF3PMI were excluded because overlap of the longest
wavelength absorption band (primarily on the imide ring) with
the shorter wavelength absorption near 270 nm (primarily
localized on the phenyl ring) prevented determination of an
accurate peak maximum and extinction coefficient for the longer
wavelength band. Peak maxima and molar extinction coefficients
at the peak maxima of the lowest observed absorption bands of
both planar and twisted ARMIs are given in Table 2 in
acetonitrile (CH3CN), dichloromethane (CH2Cl2), and chloro-
form (CHCl3). We note that 4CF3PMI, for which no SCXD
data were obtained, can be classified as planar on the basis of
its spectral characteristics. Average molar extinction coefficients
as well as average solvent shifts for the two ARMI groups
(twisted and planar) are compared in Table 3. Three other planar
ARMIs with single substituents (CH3, CN, and COOCH3) were
also evaluated and gave UV spectral results consistent with those
reported in Table 2.
Figure 5. Molecular structure of 2CF3PMI (A, B) and PMI (C, D) as
determined by single-crystal X-ray diffraction. (A) and (C) are space-
filling models constructed from the atomic coordinates from the crystal
structure. (B) and (D) are stick-model projections based upon the crystal
structure data looking down the nitrogen-phenyl bond to illustrate the
torsion angles.
are given. Torsion angle data for all the structures characterized
are given in Table 1 and example structures in Figure 5. Full
structural data including cell constants, deformations, and atom
coordinates have been submitted separately for publication.
All of the ARMIs examined exhibited a nonzero torsion angle
in the crystalline state. In the cases where there is little energetic
barrier to molecular planarity, i.e., PMI, 2FPMI, and 4OCH3-
PMI, molecular packing energetics may contribute significantly
to the final conformation adopted in the crystalline state.
Generally, the larger the ortho substituent, the closer the torsion
angle approaches τ ) 90°, as expected. While the torsion angles
in solution may be different from those observed in the
crystalline state, there is little doubt that ARMIs with substit-
uents larger than fluorine or chlorine must have τ . 0°, i.e.,
must be “twisted”, in gas and solution, as well as in the solid
state. As will be discussed, changes in molecular conformations
result in partial isolation of chromophores; i.e., the phenyl ring
and the imide ring, give rise to changes in the photophysical
properties of the molecules such as UV-vis absorption spectra
and laser flash photolysis transient absorption spectra, and alter
their photoinitiation characteristics.
Matsuo17 suggested that, as the N-substituent on maleimides
is changed from planar phenyl, to alkyl, to hydrogen, the
hybridization on the nitrogen nonbonding orbital changes from
pure p type to sp3 type, with a concomitant increase in the
overlap of the oxygen lone pair (n) orbital with the nitrogen
orbital, resulting in a perturbation of the n f π* transition on
the carbonyl by an allowed π f π* transition. Other work on
â,γ-unsaturated ketones18 suggests that overlap between the
oxygen orbital on the ketone and the p orbital on the ene can
lead to perturbation of the n f π* transition by the π f π*
transition on the ene, resulting in an increase in the oscillator
strength of the transition, and an intensification of the absorption
band. Matsuo suggests that, in the case of ARMIs, there may
2. UV-Vis Absorption. The absorption characteristics of
N-substituted ARMIs have been reported.17 To gain additional
insight into the nature of ARMI excited states, all of the ARMIs

Aromatic Maleimides as Free Radical Photoinitiators
J. Phys. Chem. B, Vol. 105, No. 14, 2001 2711
TABLE 3: Comparison of the Spectral Features of the Longest Wavelength UV Absorption Bands of Planar and Twisted
ARMIs with an N-Alkylmalimide (MMI)
av ꢀ at peak max
ꢀ at peak max (min, max)
av ∆ peak max (CHCl3 f CH3CN)
∆ peak max (min, max)
compd
MMI
twisted ARMIs
2FPMI
(L mol-1 cm-1
)
(L mol-1 cm-1
)
(nm)
(nm)
793
545
547
440
732, 826
425, 639
508, 620
244, 542
-1
-1.1
-6
0, 2
planar ARMIs
-13
9, 25
be analogous overlap of the carbonyl n orbital with the highest
occupied π orbital on the phenyl ring when the phenyl ring is
nearly planar with respect to the imide ring. He further observed
that there is an additional band around 260 nm for ARMIs
originating on the phenyl ring.17 These bands were also observed
in spectra measured in this study. Observation of the shift of
these bands with increasing solvent polarity shows very small
changes. These data along with the rather large molar extinction
coefficient (∼103) confirm assignment of the band as a π f
π* transition from the phenyl ring, as predicted in the calcula-
tions presented previously.16
In the present investigation, UV-vis absorption spectra
indicate that the average molar extinction coefficients at the
peak maxima of the transition between 275 and 320 nm increase
in the order planar phenyl < twisted phenyl < alkylmaleimide
(Table 3). The twisted N-aromatic maleimides have an average
molar extinction coefficient of the longest wavelength absorption
band approximately 24% larger than those of the planar
analogues. A reasonable comparison can be made between
2CF3PMI and 4CF3PMI, whose inductive electron-withdrawing
effects on the ring should be similar. The molar extinction
coefficients are 639 L mol-1 cm-1 at 285.5 nm and 542 L mol-1
cm-1 at 312.5 nm for 2CF3PMI and 4CF3PMI, respectively.
Both the increase in molar extinction coefficient and the shift
to higher energy transition (∆Eπfπ* > ∆Enfπ*, in general) agree
with the postulate that the transition increases in π f π*
character when the phenyl ring and imide ring are nearly
antiplanar. For the compounds evaluated in this investigation,
the magnitudes of the extinction coefficients range from a
minimum of about 244 L mol-1 cm-1 for 4OCH3PMI (a planar
ARMI with an electron donor group in the para position) to
826 L mol-1 cm-1 for MMI. These values are not large enough
to suggest a pure π f π* transition, which should have a molar
extinction coefficient minimally between 103 and 104 L mol-1
cm-1. They are larger than typical n f π* molar extinction
coefficients, however, which are typically less than 200 L mol-1
Figure 6. UV-vis absorption spectra of MMI in (A) CH2Cl2, (B) CH3-
CN, and (C) CHCl3, with concentrations of 1.68 × 10-3, 1.42 × 10-3
,
and 1.16 × 10-3 M, respectively.
Figure 7. UV-vis absorption spectra of 26DIPPMI in (A) CH2Cl2,
(B) CHCl3, and (C) CH3CN, with concentrations of 1.40 × 10-3, 1.13
× 10-3, and 1.19 × 10-3 M, respectively.
group, planar ARMIs including PMI and para-substituted
ARMIs, on the other hand, exhibit large shifts to shorter
wavelengths, ranging in magnitude from 9 to 25 nm. In general,
the substituents seemed to affect the magnitude of the shift to
shorter wavelength in the order electron-withdrawing < unsub-
stituted < electron-donating. The largest shift of 25 nm observed
for 4OCH3PMI is comparable to the shift of 31 nm obtained
by Matsuo17 for 4OCH3PMI upon changing from hexane to
CH3CN. Shifts of this magnitude are larger than expected simply
on the basis of n f π* character in the transition, suggesting
significant charge-transfer (CT) character in the ground state
of planar ARMIs. 2FPMI seems to fall between the two groups,
exhibiting extinction coefficients in the same range as the other
ortho-substituted ARMIs while giving an intermediate shift
magnitude of 6 nm upon changing from CHCl3 to CH3CN. The
fluoro group is small, and the resulting torsion angle in 2FPMI
is therefore little different from that of PMI, though the
population average conformation of neither compound is likely
to be completely planar in the gas or solution phase. Indeed,
torsion angle data and semiempirical calculations for conforma-
tion energetics and UV-vis absorption bands of compounds in
cm-1 19 Our results agree with those of Matsuo,17 suggesting
.
that the absorption band with peak maxima near 290 nm is
principally an n f π* transition, increasingly perturbed by an
allowed π f π* transition as the N-substituent is changed from
planar phenyl, to twisted phenyl, to alkyl.
UV-vis absorption spectra, obtained for the ARMIs used in
this investigation in chloroform (CHCl3), dichloromethane (CH2-
Cl2), and acetonitrile (CH3CN), with dielectric constants of 4.81,
9.08, and 37.5, respectively, for MMI, 26DIPPMI, and PMI,
are given in Figures 6-8, respectively. On the basis of the
spectral shift data for all of the ARMIs evaluated, the com-
pounds may be segregated again into two main groups. The
first group includes MMI and all the twisted ortho-substituted
ARMIs, with the exception of 2FPMI. The compounds in this
group show only negligible shifts (see Figures 6 and 7 for
examples) upon changing from CHCl3 to CH3CN. These results
are consistent with those reported by Matsuo, who observed a
negligible 2-3 nm shift for N-ethylmaleimide and N-cyclo-
hexylmaleimide when changing from n-hexane (dielectric
constant 1.89) to CH3CN (dielectric constant 37.5).17 The second

2712 J. Phys. Chem. B, Vol. 105, No. 14, 2001
Miller et al.
TABLE 4: Laser Flash Photolysis Transient Absorption
Data for MMI and Selected ARMIs
lifetime at
max (ns)
compd
MMI
λ
max (nm)
340
OD at λmax
0.11
λ
159
Twisted ARMIs (Group 1)
2IPMI
320, 560
330
0.037, 0.017
0.026
0.017
76.9, 118
2Br35CF3PMI
2CF3PMI
2BrPMI
2tBPMI
2ClPMI
775
990
412
358
323
833
330
320
330
0.012
0.010
330
340
0.0087
0.0047
26DIPPMI
Planar ARMIs (Group 2)
4CF3PMI
2FPMI
PMI
360, 480
320, 520
340, 480
320, 520
0.0084, 0.0067
388, 346
1070, 862
417, 400
a
0.008, 0.0063
0.0072, 0.0067
0.0027, 0.002
Figure 8. UV-vis absorption spectra of PMI in (A) CH3CN, (B) CH2-
Cl2, and (C) CHCl3, with concentrations of 9.99 × 10-4, 8.20 × 10-4
,
4OCH3PMI
and 7.51 × 10-4, respectively.
Solvent
0.0018
CH3CN Solvent
a The transient was too weak to measure an accurate lifetime.
Figure 9. UV-vis absorption spectra of (A) 8.62 × 10-4 M 2CF3PMI,
(B) 1.36 × 10-4 M 2CF3PCI, and (C) 8.03 × 10-4 M 2CF3PII in
CH3CN.
the series PMI, 2FPMI, 2ClPMI, 2BrPMI, and 2IPMI reported
previously16 confirm this assertion. Matsuo’s postulate of a
mixed transition, with the n f π* transition increasingly
perturbed by a π f π* transition in the series planar aromatic
< twisted aromatic < alkyl < unsubstituted maleimide,17
explains our observations.
Figure 10. Transient UV-vis absorption spectra for 2CF3PMI in N2-
bubbled CH3CN with OD ) 1 at the excitation wavelength (266 nm).
Spectra are given at 50 ns, 100 ns, 250 ns, 500 ns, 750 ns, and 1 µs
after the laser pulse. The lifetime at 330 nm was calculated to be 990
ns.
UV-vis absorption spectra were also measured for two
maleimide analogues, 2CF3PCI and 2CF3PII. Comparison of
the absorption spectra of these compounds with that of 2CF3PMI
in Figure 9 shows that substitution of the methyl group on the
double bond (citraconimide) shifts the 280 nm band to shorter
wavelength with respect to that of the analogous maleimide.
The itaconimide is essentially a substituted acrylamide, and
exhibits a longest wavelength peak that is shifted even further
toward the 260 nm band.
identical conditions) of transient divides the ARMIs into the
same two groups as before, with the exception of 26DIPPMI.
The twisted ARMIs generally exhibit a higher relative triplet
yield than the planar compounds. The optical densities of the
transients with peak maxima at ∼340 nm are all significantly
lower than that of MMI, suggesting that the relative quantum
yields of excited-state triplet upon direct excitation of the ARMIs
are significantly smaller than that of MMI, which is in turn
significantly smaller than that of benzophenone (BP). The low
yield of the transient (excited-state triplet, as will be shown later)
of 26DIPPMI may be due to enhanced internal conversion from
the excited-state singlet to the ground state, thereby decreasing
the excited-state singlet lifetime and resulting in a decreased
yield of excited-state triplet. Alternatively, the transient yield
may be low due to intramolecular hydrogen abstraction of a
tertiary benzylic hydrogen on the isopropyl groups by the
carbonyl in the excited-state singlet, thereby reducing the
effective triplet yield. The same reaction may also occur in the
triplet, though the relatively long lifetime of 833 ns measured
for the 26DIPPMI transient may suggest that intramolecular
hydrogen abstraction is not a significant deactivation pathway
in the triplet state. In any case, the relative triplet yields of the
3. Laser Flash Photolysis. It has been previously reported1,20
that the transient UV-vis absorption spectra of N-alkylmale-
imides show a single absorption band with peak maxima near
340 nm that is quenched by oxygen, 1,3-cyclohexadiene, and
â-carotene, the latter producing the distinctive triplet absorption
spectrum of â-carotene, thus providing conclusive evidence for
the triplet state. The transient optical densities, peak maxima,
and measured triplet lifetimes for 2CF3PMI, 4CF3PMI, MMI,
and eight other ARMIs are given in Table 4. The data are listed
in order of decreasing relative transient yield, taken to be the
relative optical densities at the transient peak maxima. Repre-
sentative transient absorption spectra for 2CF3PMI and 4CF3PMI
are given in Figures 10 and 11, respectively.
3.1. Transient Spectral Features. The order of the relative
yield (relative absorbance of the transients obtained under

Aromatic Maleimides as Free Radical Photoinitiators
J. Phys. Chem. B, Vol. 105, No. 14, 2001 2713
energetically close (∆ ≈ 4 kcal/mol) excited triplets states exist
for 2PBP, which may interconvert by rotation of one of the
phenyl rings.24 In the case of planar ARMIs, there is little barrier
to planarity or perpendicularity of the two rings, vide supra from
the AM1 rotational conformer calculations previously reported.16
Extension of these trends to the planar ARMIs might suggest
the existence of two excited-state triplets, i.e., one planar and
one perpendicular, which if close enough in energy (∼2 kcal/
mol in the ground state from AM1) might be equally populated,
giving rise to two triplet-triplet absorption bands in the excited
state. If the interconversion between the two triplet states is
hindered such that the bond rotation does not occur within the
lifetimes of the two states, then the two states may be distinct.
Further support of the dual excited-state hypothesis comes from
reports by Valat et al.,14,15 who reported that N-phenylnaph-
thalimides with bulky ortho substituents have UV-vis absorp-
tion (and fluorescence) very similar to that of N-alkylnaphthal-
imides, while unsubstituted and non-ortho-substituted N-phenyl-
naphthalimides showed red-shifted, with respect to the N-
alkylnaphthalimides, longest wavelength UV-vis absorption
peaks (as well as dual-band fluorescence). After characterizing
a host of N-alkyl- and substituted N-phenylnaphthalimides by
UV-vis absorption and fluorescence emission spectroscopies,
Valat postulated that two related, yet distinct excited-state
singlets are responsible for the two fluorescence emission bands
in the non-ortho-substituted N-phenylnaphthalimides. Extension
of this postulate to the case of ARMIs would suggest that two
energetically and conformationally distinct excited singlet states
might upon intersystem crossing lead to two distinct excited
triplet states. Unfortunately, while fluorescence emission spectra
confirmed Valat’s postulate in the case of N-phenylnaphthal-
imides, no emission was observed in this investigation of
ARMIs. (c) A twisted intramolecular charge-transfer state
(TICT) as reported previously for N-aromatic phthalimides25,26
may be operative for planar N-aromatic maleimides. The lowest
possible transition calculated in the AM1 analyses16 of the
o-halo-substituted ARMIs was a HOMO to LUMO transition,
which was not spatially allowed. The HOMO had π character
on the phenyl ring, while the LUMO had π* character on the
maleimide ring double bond and the carbonyls. If such a
transition did occur, even with a very low oscillator strength,
there would be charge separation because the transition would
essentially involve transfer of an electron from an orbital on
the phenyl ring to an orbital on the maleimide ring. The excited
singlet TICT state might be stabilized through intersystem
crossing to an excited TICT triplet state.
Figure 11. Transient UV-vis absorption spectra for 4CF3PMI in N2-
bubbled CH3CN with OD ) 1 at the excitation wavelength (266 nm).
Spectra are given at 50 ns, 100 ns, 250 ns, 500 ns, 750 ns, and 1 µs
after the laser pulse. The lifetime at 360 nm was calculated to be 388
ns, and the lifetime at 480 nm was calculated to be 346 ns.
ARMIs are all low upon direct excitation, with the planar
ARMIs having extremely low yields.
The relative transient yield for 2IPMI (0.037) was more than
double that of 2CF3PMI (0.017), and the 2IPMI transient had
a significantly shorter triplet lifetime than 2CF3PMI, 77 ns
versus 990 ns, respectively. These two observations suggest that
triplet-state formation and decay may be influenced by an
internal heavy atom effect which increases the rates of inter-
system crossing. The effect is generally only significant in π
f π* states, as the spin-orbit coupling effect operative in n
f π* states is of a much higher magnitude.19 Heavy atom
effects not only increase the triplet yields from excited singlets,
but also decrease the triplet-state lifetime by increasing inter-
system crossing rates to the ground-state singlet.19 The lifetime
of the 2IPMI triplet may also be decreased due to a possible
photocleavage reaction. This is certainly consistent with our
observation of a 560 nm peak in the spectrum of 2IPMI
attributed to the formation of a transient photoproduct, likely
from photocleavage of the phenyl-halogen bond, which is well-
known in the literature.21-23 (Interestingly, the 2IPMI solution
was in fact visibly yellow after a time of continuous pulsing,
suggesting the possible formation of molecular iodine).
A correlation between the ARMI structure and the shapes of
the transient absorption spectra was also observed. The twisted
ARMIs all exhibited a single transient peak (Figure 10) similar
to that observed in MMI, with the exception of 2IPMI, which
showed an additional peak at 560 nm as discussed previously.
The planar ARMIs all exhibited a bimodal transient absorption
spectrum (Figure 11), generally with peak maxima between 320
and 360 nm and between 480 and 520 nm. The origin of the
second transient absorption band has not been determined.
Possible explanations include the following: (a) The two bands
might be due to a single species. Similarity of the lifetimes at
the two peak maxima in the transient spectrum of PMI and
4CF3PMI suggests the possibility of a single excited-state
species in each ease. In the case of 2FPMI, the two bands differ
in calculated lifetime by about 20% (1070 ns versus 862 ns),
which may be within experimental error given the low intensity
of the absorption. However, the similarity in lifetimes may
simply connote two species with similar or coupled molecular
structure. (b) Two distinct populated excited triplet states may
exist for planar ARMIs, as in the case of 2-phenylbenzophenone
(2PBP) described by Scaiano, where he claimed that two
Either of the latter two possible explanations for the origin
of the second transient absorption band provides for the
formation of two distinct excited-state triplet species for planar
ARMIs. Further elucidation of these hypotheses for ARMIs is
exceedingly difficult in the absence of fluorescence and
phosphorescence emission spectra. High-level excited-state
triplet to upper excited-state triplet calculations are in progress
to attempt to reproduce the UV-vis absorption spectra of the
ARMI triplet transients, and will be reported in a future
publication.
3.2. Quenching of ARMI Transients. To prove whether the
transient species formed upon photolysis of ARMIs were
excited-state triplets, quenching studies were performed. Cy-
clohexadiene (CHD), which has a low-lying triplet state and
therefore selectively quenches excited-state triplets, was chosen
to quench the N-phenylmaleimide transients. Solutions were
prepared for PMI and 2CF3PMI, representing planar and twisted
ARMIs, respectively, containing various concentrations of CHD.

2714 J. Phys. Chem. B, Vol. 105, No. 14, 2001
Miller et al.
Figure 13. Photo-DSC exotherms from the photopolymerization of
PEG400DA initiated by approximately 5 mol % N-substituted male-
imides, based upon the mole percent of the compound: (A) 4.97%
2Br35CF3PMI, (B) 5.01% 2IPMI, (C) 5.01% MMI, (D) 4.96% BP,
(E) 4.99% 2CF3PMI, (F) 4.99% 2BrPMI, (G) 5.01% 2ClPMI, (H)
5.00% 26DIPPMI, (I) 5.02% 2tBPMI, (J) 4.99% 2FPMI, (K) 4.97%
4CF3PMI, (L) 4.97% PMI. Samples of 2 µL were polymerized under
nitrogen at 25 °C using the full arc of the medium-pressure mercury
lamp with an on-sample light intensity of 32.2 mW/cm2. Samples were
not normalized for initiator absorbance.
Figure 12. Stern-Volmer analysis of data from quenching BP with
2CF3PMI in CH3CN solution excited at 355 nm and analyzed at 520
nm following nitrogen purge. The average kq was determined to be
4.04 × 109 L mol-1 s-1 on the basis of four replicate data sets with a
standard deviation of 9.8 × 107 L mol-1 s-1. The average R2 for the
four least-squares fits is 0.999.
TABLE 5: Rate Constants for Quenching of Benzophenone
by MMI and Selected ARMIs
quencher
kq (L mol-1 s-1
)
quencher
kq (L mol-1 s-1
)
mol-1 s-1 for MMI. The quenching constants in Table 5 are all
quite large and approach the diffusion-controlled rate of ∼1010
L mol-1 s-1. Sensitization of the maleimide triplets by BP could
not be confirmed by laser flash transient absorption spectra due
to overlap of the BP 310 nm band with the maleimide bands,
as well as competitive absorption by the maleimides, resulting
in production of the maleimide transient by direct excitation.
Quenching of the 2CF3PMI transient (formed upon direct
excitation in the absence of BP) by N-methyl-N,N-diethanola-
mine (MDEA) showed that the maleimide triplet was quenched
by the amine, though calculation of a kq for the reaction was
problematic due to the inherently low transient absorbance of
2CF3PMI.
4. Photopolymerization. 4.1. Direct Exposure of N-Alkyl-
maleimides and N-Aromatic Maleimides. As discussed in the
Introduction, N-alkylmaleimides have been shown to initiate free
radical polymerization by a hydrogen abstraction process in the
presence of a hydrogen atom donor synergist.1,2 The hydrogen
atom donor may be added directly to the formulation as a
synergist, or incorporated into the monomers. In the case of
PEG400DA, which is a diacrylate with an average of 10
ethylene oxide units in the spacer group, the hydrogen atoms
on the carbons R to the ether oxygens are labile, and readily
abstracted by hydrogen-abstraction-type photoinitiators. Ac-
cordingly, PEG400DA was found to photopolymerize readily
in the presence of N-alkylmaleimides and certain N-aromatic
maleimides, with the efficiency of the initiation depending upon
the maleimide structure. As shown by the photo-DSC exotherms
in Figure 13 for 10 ARMIs, selected from those depicted in
Figure 2 as well as for MMI and BP, the highest relative
polymerization efficiencies, i.e., exotherm peak maxima, were
obtained for samples with 2Br35CF3PMI and 2IPMI followed
by the sample with MMI. The exotherms for the six maleimides
of lowest reactivity are either less than or not significantly higher
than the exotherm measured for neat PEG400DA, which self-
initiates perhaps by excitation of peroxide impurities or acrylate
groups directly. Interestingly, 2IPMI and 2Br35CF3PMI ex-
hibited the highest triplet yield of the ARMIs, though signifi-
cantly less than that of MMI (Table 4). The polymerization rates
obtained for 2CF3PMI- and BP-initiated samples were signifi-
cantly lower than those for MMI, 2IPMI, or 2Br35CF3PMI. It
MMI
PMI
2CF3PMI
7.8 × 109
3.0 × 109
4.0 × 109
4CF3PMI
26DIPPMI
2.9 × 109
2.2 × 109
LFP measurements were performed under conditions similar
to those described above. Since the ARMIs had inherently low
transient yields, quantitative quenching to obtain quenching rate
constants was impossible. Qualitatively, low concentrations of
CHD were found to completely quench the transient species
observed in 2CF3PMI and PMI (both peaks), strongly suggesting
that the transient is the excited-state triplet, as in the case of
MMI. In the case of PMI, quenching of both bands suggests
that (a) the bands are due to a single triplet species which is
quenched, (b) there are two separate excited-state triplets which
are both quenched, or (c) that the higher energy band is a triplet
which produces the lower energy band species and that
quenching of the former prevents formation of the latter. A
reduction in the lifetime and intensity of the transient absorption
bands for both PMI and 2CF3PMI in air-saturated solutions
suggests oxygen quenching of excited-state triplets. These
observations, along with inferences from MMI experiments,2
provide strong evidence for the characterization of the transient
absorption species of ARMIs as excited-state triplets.
3.3. Quenching of Benzophenone by Triplet ARMIs. A
targeted application of the research on ARMIs in this investiga-
tion is the photoinitiation of radical polymerization in the
presence of absorbing compounds which may act, in part, as
sensitizers for triplet formation of ARMIs. Elucidation of the
relative efficiencies of triplet sensitization of maleimide triplets
by benzophenone is therefore of great interest. To this end,
solutions were prepared containing BP with increasing concen-
tration of various maleimides to determine the rate constants
for quenching (kq) of the excited-state BP triplet by the
maleimides. The lifetime of the BP triplet was measured as a
function of the concentration of the quencher, and the data were
plotted using the Stern-Volmer relation19 (as in Figure 12 for
CF3PMI). Quenching constants (kq) determined from such
Stern-Volmer plots are given in Table 5. The rate constants
for quenching of BP excited-state triplet by ARMIs range from
2.2 × 109 to 4.0 × 109 L mol-1 s-1 compared to 7.8 × 109 L

Aromatic Maleimides as Free Radical Photoinitiators
J. Phys. Chem. B, Vol. 105, No. 14, 2001 2715
Figure 14. Photo-DSC exotherm peak maxima for the polymerization
of HDDA/PEG400DA formulations initiated by 5 mol % 2CF3PMI,
based upon the moles of the functional groups, as a function of
PEG400DA content. Samples of 2 µL were irradiated at 25 °C under
nitrogen atmosphere by the full arc output of a xenon-doped high-
pressure mercury lamp with an on-sample light intensity of 50 mW/
cm2.
Figure 15. Photo-DSC exotherm peak maxima for HDDA initiated
by 5 mol % MMI and various ARMIs in the presence of 1 mol %
MDEA. Polymerizations were performed using 2 µL samples at 25 °C
under nitrogen atmosphere with the full arc of a medium-pressure
mercury lamp giving on-sample light intensity of 29.5 mW/cm2.
4OCH3PMI did not completely dissolve at the concentration and
temperature used.
should be noted that the absorbances of initiator in this series
were not identical; thus, the initiation efficiencies cannot strictly
be compared quantitatively. Ignoring BP, the ranking of
polymerization efficiency does however appear to generally
follow the order of relative triplet yield with one noted
exception, MMI. The molar extinction coefficient of MMI at
266 nm, a principal band in the medium-pressure mercury
spectrum, is approximately a factor of 3 smaller than those
observed for 2IPMI and 2Br35CF3PMI. However, counterbal-
ancing this effect is the fact that MMI has a significantly higher,
by a factor of 3, relative yield of triplet compared to the highest
ARMIs, 2IPMI and 2Br35CF3PMI, with relative triplet yields
of 0.11 versus 0.037 and 0.026, respectively. Finally, the
increased apparent initiation efficiencies, ∼40% higher than
MMI, of 2IPMI and 2Br35CF3PMI may also be partially due
to photocleavage reactions, inferred in the case of 2IPMI from
the LFP experiments, generating highly reactive phenyl radicals.
The polymerization efficiency of binary mixtures of HDDA
and PEG400DA were found to increase when photoinitiated with
5 mol % 2CF3PMI (Figure 14). The maximum polymerization
rate was found to occur for a system with about 70%
PEG400DA. The mixtures with higher concentrations of
PEG400DA polymerize slower, presumably as a result of the
slower inherent polymerization rate of PEG400DA compared
to HDDA. HDDA, which has no readily abstractable hydrogens,
would not be expected to polymerize efficiently in the presence
of N-substituted maleimides without the presence of a source
of labile hydrogens. The efficiency of initiation of HDDA, which
as noted above has no readily abstractable hydrogens, by
N-alkylmaleimides has been shown2 to be enhanced significantly
by the addition of an amine due to a highly efficient electron/
proton-transfer process. To compare results for the ARMIs with
those of N-alkylmaleimides, photo-DSC exotherms of samples
containing 5 mol % of a series of ARMIs in HDDA with 1 mol
% MDEA were measured (Figure 15). The relative ranking of
the exotherm peak maxima roughly follows the relative yield
of triplet transient measured in the laser flash photolysis
experiments (Table 4), with the exceptions of 2CF3PMI and
26DIPPMI. Interestingly, only MMI, 2IPMI, and 2Br35CF3PMI
initiators gave polymerization exotherms larger than that of
HDDA with amine only (the species responsible for the inherent
initiation in the HDDA is probably reduced by competitive
absorption of the maleimides). The reason for the low efficiency
of twisted ARMIs (with the exception of 2IPMI and 2Br35CF3-
PMI) in the presence of MDEA is unknown, although a
contributing factor may be a ground-state charge-transfer
complex formed between MDEA and the N-aromatic maleim-
ides.
4.2. Exposure of N-Alkylmaleimides and N-Aromatic Male-
imides in the Presence of Benzophenone. The quantum yields
of excited-state triplet formation of N-alkylmaleimides upon
direct excitation have been reported by De Schryver and Put6
to be about 0.23-0.24. A recent publication suggests these
values to be even lower.27 As determined by LFP described in
this paper, triplet yields for ARMIs are significantly lower that
those for N-alkylmaleimides. Photolysis of maleimides in the
presence of BP or isopropylthioxanthone (ITX) has been
reported to enhance [2+2] cycloaddition5-11 and free radical
initiation of acrylic systems with N-alkylmaleimide/amine
combinations present.1,2 Representative photo-DSC exotherms
in Figure 16 for several ARMI systems illustrate that both planar
and twisted ARMIs exhibit enhanced radical polymerization
rates of HDDA in the presence of MDEA when BP is added.
All of the maleimides, including N-alkylmaleimides as well as
twisted and planar ARMIs, exhibited polymerization efficiencies
significantly larger than that for the BP/MDEA combination
alone in HDDA. Two possibilities for enhanced rates exist. First,
as suggested by the laser flash photolysis results in section 3.3,
BP may sensitize formation of the maleimide triplet state.
Second, recent results from our lab indicate that semipinacol
radicals formed from the electron/proton-transfer reduction
process involving the BP triplet state and MDEA readily react
with (more exactly, are quenched by) N-aromatic maleimides,
presumably by an electron/proton-transfer process, to generate
BP and the succinimidyl radical. We suppose that both processes
for generation of two initiating radicals (both processes produce
the same radicals) may take place depending upon the concen-
tration of individual components (maleimide, amine). Elucida-
tion of the mechanistic details is in progress and will be reported
in a subsequent publication.
Finally, polymerization of HDDA in the presence of BP (3
wt %), MDEA (1 wt %), and isomaleimides, 2CF3PCI and
2CF3PII (0.1 wt %), indicate that 2CF3PCI/MDEA/BP and
2CF3PII/MDEA/BP systems also operate as efficient free radical
initiators. Qualitatively, 2CF3PII does not yield rates of initiation

2716 J. Phys. Chem. B, Vol. 105, No. 14, 2001
Miller et al.
charge-transfer process in certain para-substituted ARMIs.
Transient UV-vis absorption spectra for the excited-state triplets
of the planar ARMIs show much lower relative triplet-state
quantum yields and exhibit a dual-band absorption, in contrast
to twisted ARMIs and N-alkylmaleimides. The origin of the
second transient absorption band is unknown, though arguments
were presented suggesting the existence of two distinct excited-
state triplet species for planar ARMIs.
Photopolymerization studies indicated that twisted ARMIs
efficiently initiate free radical acrylate polymerization in the
presence of a hydrogen atom donor synergist upon direct
excitation, while planar ARMIs do not. Relative initiation
efficiencies generally correlate with the relative yields of excited-
state triplet. Addition of benzophenone was found to dramati-
cally increase the initiation efficiencies of both planar and
twisted ARMIs. The increase in rate with addition of benzophe-
none was found to be independent of the maleimide structure
(planar ARMI, twisted ARMI, or N-alkylmaleimide). This is
particularly interesting since N-alkylmaleimide/amine/benzo-
phenone systems were previously shown to exhibit initiation
efficiencies approaching those of conventional cleavage photo-
initiators.1 Hence, with noted exception, any N-substituted
maleimide can be used along with benzophenone or closely
related diaryl ketones to initiate acrylate polymerization with
efficiencies comparable to those exhibited by cleavage photo-
initiators.
Figure 16. Photo-DSC exotherm peak maxima for HDDA formulations
initiated by ARMIs in the presence of an amine and BP. Samples
contain 3 mol % BP, based on the moles of the compound, 1% MDEA
and 0.10% maleimide, except for the BP sample, which has no
maleimide. Samples of 2 µL were polymerized at 25 °C under nitrogen
atmosphere, with an on-sample light intensity of 1.86 mW/cm2 at 365
nm, from the full arc spectrum of a medium-pressure mercury light
source.
Acknowledgment. We gratefully acknowledge the financial
support of Fusion UV-Curing Systems and ChemFirst Fine
Chemicals. We thank Jeffrey Zubkowski, Department of
Chemistry, Jackson State University, for access to the diffraction
facility, and the Office of Naval Research (E.J.V.).
as high as those found for 2CF3PCI and the other ARMIs. This
is not unexpected, however, since the itaconimide resembles a
substituted acrylamide much more than a maleimide, and may
yield radicals that have different reactivity toward initiation of
acrylate polymerization than the maleimide radicals and/or are
formed with different efficiencies in the presence of amines or
sources of abstractable hydrogens.
References and Notes
(1) Hoyle, C. E.; Viswanathan, K.; Clark, S. C.; Miller, C. W.; Nguyen,
C.; Jo¨nsson, S.; Shao, L. Macromolecules 1999, 32, 2793.
(2) Clark, S. C. Ph.D. Dissertation, Department of Polymer Science,
The University of Southern Mississippi, Hattiesburg, MS, 1999.
(3) Sonntag, J. V. Ph.D. Dissertation, Faculty of Chemistry and
Mineralogy, The University of Leipzig, Germany, 1999.
(4) Morel, F.; Decker, C.; Jo¨nsson, S.; Clark, S. C.; Hoyle, C. E.
Polymer 1999, 40, 2447.
(5) White, J. E.; Dehnke, M. K.; Tang, L. H. U.S. Patent 4,851,454,
1989.
(6) Put, J.; DeSchryver, F. C. J. Am. Chem. Soc. 1973, 95, 137.
(7) Meir, K.; Zweifel, J. J. Photochem. 1986, 35, 353.
(8) DeSchryver, F. C.; Feast, W. J.; Smets, G. J. Polym. Sci., Part A-1
1970, 8, 1939.
Conclusions
N-aromatic maleimide derivatives were evaluated as free
radical photoinitiators for acrylate polymerization. The ARMI
compounds investigated in this study may be segregated into
two groups on the basis of molecular conformation (torsion
angle between phenyl and imide rings), UV-vis absorption
spectral features, transient UV-vis absorption spectral features
observed by laser flash photolysis, and relative photoinitiation
efficiency. The first group is composed of ARMIs that are
substituted in the ortho position of the phenyl ring by a group
sufficiently large (larger than fluorine) to force the average
molecular conformation to exist with the imide ring antiplanar
with respect to the phenyl ring. These twisted ARMIs exhibited
UV-vis absorption spectra similar to that of N-alkylmaleimides,
with spectral shifts as a function of solvent polarity indicative
of a primarily π f π* longest wavelength absorption band.
Transient UV-vis absorption spectra of the excited-state triplet
species of the twisted ARMIs very closely resemble the transient
spectra observed for N-alkylmaleimides. The second group of
ARMIs do not have substituents in the ortho position on the
phenyl ring, such that there is little conformational energy barrier
to forming a planar molecular conformation. These planar
ARMIs exhibit longest wavelength UV-vis absorption bands
that occur at lower energies than the corresponding bands in
the twisted ARMIs and N-alkylmaleimides as well as large
spectral shifts upon changes in solvent polarity, indicative of a
primarily n f π* transition or possibly an intramolecular
(9) Boens, N.; DeSchryver, F. C.; Smets, G. J. Polym. Sci., Polym.
Chem. Ed. 1975, 13, 201.
(10) Zweifel, H. Photogr. Sci. Eng. 1983, 27, 114.
(11) Timpe, H.-J. Proceedings of the Academic Day, Radtech Europe
‘97; 1997; p 127.
(12) Sheldrick, G. SHELXS-93, A program for the solution of crystal
structures from diffraction data, University of Go¨ttingen, Go¨ttingen,
Germany.
(13) Sheldrick, G. SHELXL-97, A program for the refinement of crystal
structures from diffraction data, University of Go¨ttingen, Go¨ttingen,
Germany.
(14) Valat, P.; Wintgens, V.; Kossanyi, J.; Biczok, L.; Demeter, A.;
Berces, T. J. Am. Chem. Soc. 1992, 114, 946.
(15) Wintgens, V.; Valat, P.; Kossanyi, J.; Demeter, A.; Biczok, L.;
Berces, T. J. Photochem. Photobiol., A 1996, 93, 109.
(16) Miller, C. W.; Hoyle, C. E.; Valente, E. J.; Magers, D. H.; Jo¨nsson,
E.S. J. Phys. Chem. A 1999, 103, 6406.
(17) Matsuo, T. Bull. Chem. Soc. Jpn. 1965, 38, 557.
(18) Labhart, H.; Wagniere, G. HelV. Chim. Acta 1959, 42, 2219.
(19) Turro, N. J. Modern Molecular Photochemistry; University Science
Books: Mill Valley, CA, 1991.
(20) Viswanathan, K.; Clark, S.; Miller, C.; Hoyle, C. E.; Jonsson, S.;
Shao, L. Polym. Prepr. 1998, 39 (2), 644.

Aromatic Maleimides as Free Radical Photoinitiators
J. Phys. Chem. B, Vol. 105, No. 14, 2001 2717
(21) Horspool, W.; Armesto, D. Organic Photochemistry: A Compre-
hensiVe Treatment; Ellis Horwood Ltd.: New York, 1992.
(22) Grimshaw, J.; de Sliva, A. P. Chem. Soc. ReV. 1981, 10, 181.
(23) Bunce, N. J.; Bergsma, J. P.; Bergsma, M. D.; De Graaf, W.; Kumar,
Y.; Ravanal, L. J. Org. Chem. 1980, 45, 3708.
(24) Scaiano, J. C.; Nicodem, D. E. Can. J. Chem. 1984, 62,
2346.
(25) Creed, D.; Hoyle, C. E.; Jordon, J. W.; Pandy, C. A.; Nagarajan,
R.; Pankasem, S.; Peeler, A. M.; Subramanian, P. Macromol. Symp. 1997,
116, 1.
(26) Barlow, J. H.; Davison, R. S.; Lewis, A.; Russel, D. R. J. Chem.
Soc., Perkin Trans. 2 1979, 1103.
(27) Sonntag, J. V.; Knolle, W. J. Photochem. Photobiol., A 2000, 136,
133.