Zero-field splitting of the electronic ground and lowest excited triplet states of 2,2-dinaphthylcarbene in n-hexane at 1.7 K

Article abstract of DOI:10.1021/jp984581e

The triplet-triplet fluorescence spectra of a new carbene, 2,2-dinaphthylcarbene in nhexane and n-heptane, were studied at cryogenic temperatures. Synthesis of the carbene precursor, 2,2-dinaphthyldiazomethane, is described. Spectral holes were burned wit

Full text of DOI:10.1021/jp984581e

J. Phys. Chem. A 1999, 103, 3155-3162
3155
Zero-Field Splitting of the Electronic Ground and Lowest Excited Triplet States of
2,2-Dinaphthylcarbene in n-Hexane at 1.7 K
B. Kozankiewicz*,† and M. Aloshyna
Institute of Physics, Polish Academy of Sciences, Al. Lotniko´w 32/46, 02 668 Warsaw, Poland
A. D. Gudmundsdottir and M. S. Platz*
Department of Chemistry, The Ohio State UniVersity, Columbus, Ohio 43210
M. Orrit* and Ph. Tamarat
Centre de Physique Mole`culaire Optique et Hertzienne, CNRS et UniVersite Bordeaux I,
351 Cours de la Liberation, 33405 Talence Cedex, France
ReceiVed: December 2, 1998; In Final Form: February 12, 1999
The triplet-triplet fluorescence spectra of a new carbene, 2,2-dinaphthylcarbene in n-hexane and n-heptane,
were studied at cryogenic temperatures. Synthesis of the carbene precursor, 2,2-dinaphthyldiazomethane, is
described. Spectral holes were burned within the inhomogeneous 0,0 fluorescence excitation line assigned to
the pseudo-E/trans conformer. The complicated pattern of burned holes was interpreted with a model taking
into account the zero-field splitting (ZFS) of the ground T0 and excited T1 triplet states and the selectivity of
the intersystem crossing channel. The unusual observation of antiholes was attributed to the extremely long
spin-lattice relaxation time in the T0 state. The analysis provided direct information about the ZFS parameters
of the T0 and T1 states: E0 ) 0.022 ( 0.001 cm-1, D0 ) 0.477 ( 0.001 cm-1, E1 ) 0.006 ( 0.0002 cm-1,
and D1 ) 0.043 ( 0.0005 cm-1.
I. Introduction
The first successful hole-burning experiment on the 0,0 line
of a T0 f T1 fluorescence excitation spectrum was recently
performed on 2-naphthylphenylcarbene (2NPC) dispersed in
n-hexane and n-heptane at 1.7 K.5,6 A complicated pattern of
holes was explained when taking into account the zero-field
splitting (ZFS) of the T0 and T1 states and some assumptions
about the energy dissipation channels. The ZFS of triplet states
originates from the spin-spin interaction between two unpaired
electrons and contains important information about the spatial
distribution of these electrons. The analysis of hole-burning
spectra provided the ZFS parameters of both the T0 and T1 states.
This information is not generally available for the excited, short-
lived T1 state of a carbene.
In the present work we have undertaken spectroscopic studies
of a new aromatic carbene, 2,2-dinaphthylcarbene. This com-
pound can exist in two different conformations (rotameric
forms), which we will call pseudo-E/trans and pseudo-Z/cis
conformers.
Carbenes play a very important role in organic chemistry as
short-lived intermediates produced in many chemical reactions.1
They can be created from the appropriate diazo, diazirine, and/
or ketene precursors by photolysis with visible or UV light.
Experimental studies of reactive intermediates present a chal-
lenge because their concentration in either the gas or liquid phase
at room temperature is extremely small. This problem can be
avoided, however, by using matrix isolation techniques. In a
cryogenic matrix, friction greatly restricts molecular motion and
effectively eliminates intramolecular and intermolecular reac-
tions. Thus, under these conditions it is possible to produce a
relatively large and stable concentration of carbenes and study
them for many days, in practice as long as the matrix is kept
frozen.
Carbenes have two nonbonding electrons and, depending on
their molecular structure, may have either a triplet or singlet
electronic ground state. Information about the geometry and
electronic structure of triplet carbenes can be obtained by EPR,
electronic absorption, and emission spectroscopy of carbenes
immobilized in matrix hosts.2-4 In a low-temperature Shpolskii
matrix the T1 f T0 fluorescence and T0 f T1 fluorescence
excitation spectra4 are often composed of sharp, zero-phonon
quasi-lines accompanied by broad phonon wings. These quasi-
lines are inhomogeneously broadened because of the slightly
different environments of the ensemble of carbenes produced
in a polycrystalline matrix providing a convenient situation for
hole-burning studies using highly selective irradiation with a
narrow-band ring dye laser.
II. Experimental Section
A. Synthesis of 2,2-Dinaphthyldiazomethane. Synthesis of
2,2-dinaphthyldiazomethane (in Columbus) and characterization
of intermediate steps are described below.
* Corresponding author.
† E-mail: kozank@ifpan.edu.pl. Fax: (+48-22) 8430926.
10.1021/jp984581e CCC: $18.00 © 1999 American Chemical Society
Published on Web 04/08/1999

3156 J. Phys. Chem. A, Vol. 103, No. 17, 1999
Kozankiewicz et al.
Preparation of Di-2-naphthalenylmethanone. Di-2-naphtha-
lenylmethanone was prepared by the method of Olah et al.7
Piperidine (10 mL, 0.10 mol) was added dropwise to a solution
of ethyl chloroformate (5 mL, 0.05 mol) in diethyl ether (25
mL) at -78 °C under argon. The resulting solution was stirred
at ambient temperature for an hour and refluxed for another
hour. The reaction was quenched by addition of water. The
organic layer was washed with water, dried over sodium sulfate,
the solvent removed under vacuum, and the residue distilled
under vacuum to yield N-carboethoxypiperdine (5.8 g, 0.037
mol, 71% yield). IR (NaCl plates, neat) νmax: 2981, 2936, 2855,
water layer was extracted with pentane, and the combined
organic layers washed with brine, dried over MgSO4, and
concentrated under vacuum. The resulting red solid was
recrystallized two times from pentane; mp (uncorrected) 123-
1
124 °C. IR (KBr) νmax: 2021, 1625, 1595 cm-1. H NMR
(CDCl3, 200 MHz): δ 8.0-7.5 (m, 14H) ppm. 13C NMR
(CDCl3, 50 MHz): δ 133.9, 131.8, 128.9, 128.4, 127.8, 127.3,
127.0, 126.7, 125.7, 123.7, 123.4 ppm.
B. Experimental Setup. The 2,2-dinaphthylcarbene (22-
DNC) sample was obtained “in situ” by the photolysis of the
2,2-dinaphthyldiazomethane precursor dispersed in a Shpolskii
matrix of n-hexane or n-heptane at 5 K. The 366 nm line,
isolated by appropriate filters from the spectrum of a mercury
lamp, was used during this procedure. Before being inserted
into the liquid helium cryostat, all samples were degassed by
the “freeze-pump-thaw” technique. Samples of 22-DNC were
annealed overnight, when the cryostat was warmed slowly to
approximately 100 K, and subsequently cooled back the next
day to the temperature of boiling liquid helium.
1
1698, 1432 cm-1. H NMR (CDCl3, 200 MHz): δ 4.05 (q, 7
Hz, 2H, 3.40 (m, 4H), 1.5 (m, 6H), 1.20 (t, 7 Hz, 3H) ppm.
MS (m/e relative intensity): 157 (M+, 10), 156 (100), 128 (40),
84 (84). High-resolution mass calculated for C8H15NO2: 157.1103.
Found 157.1075.
A solution of N-carboethoxypiperidine (1.0 g, 0.073 mmol)
in diethyl ether (15 mL) was added to a dispersion of 2-lithium
naphthalene in diethyl ether at 0 °C under argon. This mixture
was stirred for half an hour at room temperature and then
refluxed for an hour. The reaction was quenched by addition
of 10% aqueous HCl (20 mL), and ethyl acetate (50 mL) was
added. The organic layer was washed with saturated sodium
bicarbonate solution and brine solution and dried over MgSO4
and the solvent removed under vacuum to give an off-white
solid (2 g). The resulting solid was recrystallized from an ethyl
acetate-hexane mixture to yield colorless plates of di-2-
naphthalenylmethanone (1.1 g, 4 mmol, 54% yield) and a mother
liquor that contained a mixture of di-2-naphthalenylmethanone
and some unreacted 2-bromonaphthalene (0.88 g).
The fluorescence and fluorescence excitation spectra (in
Warsaw) were measured using two different experimental
techniques. In the photon-counting configuration the samples
were excited either with the 366 nm line isolated from the
spectrum of a HBO200 mercury lamp or with the light emitted
by a Coherent 700 dye laser pumped by a mode-locked Coherent
Antares 76 Nd:YAG laser. In the latter case the laser was
scanned within the lasing frequency range of rhodamine 6G
dye, between 570 and 600 nm. Emission spectra were observed
at the right angle using a 0.25 m Jarrel-Ash monochromator
and an EMI 9659 photomultiplier, cooled to -20 °C and
operating in the photon-counting mode. A LightScan PC card
was used as the pulse-counting electronics and as the master
clock of the experiment. In the photon-sampling configuration
the excitation source was a Lambda Physik FL 1001 dye laser
pumped by an LPX 100 excimer laser. Coumarin 153 was the
lasing dye for the spectral range 530-600 nm and rhodamine
B for the range 595-640 nm. Fluorescence light was detected
with the aid of an EMI 9659 photomultiplier and a Stanford
Research SR250 boxcar averager connected to a previously
mentioned PC card.
The 2-lithium naphthalene was prepared by adding butyl-
lithium to an ether solution of 2-bromonaphthalene at ambient
temperature. The solution was stirred for 15 min and then
refluxed for half an hour; mp 157-8 °C (lit.8 125 °C). IR
1
(CHCl3)νmax: 1654, 1627, 1288 cm-1. H NMR (CDCl3, 200
MHz): δ 8.33 (s, 2H), 8.0 (s, 4H), 7.9 (m, 4H), 7.60 (d of quint,
2 and 7 Hz, 4H) ppm. 13C NMR (CDCl3, 75 MHz): δ 196.7
(CdO), 135.2 135,1 132.3, 131.8 129.4, 128.3, 128.2, 127.8,
126.8, 125.9 ppm. MS (m/e, relative intensity): 282 (M+, 81),
254 (30), 155 (100), 127 (89). Mass calculated for C21H14O:
282.1045. Found: 282.1048.
Fluorescence decays were measured using the “time-cor-
related” single-photon-counting technique. The samples were
excited in this case by the previously mentioned Coherent laser
system, which provided light pulses with about a 20 ps pulse
width and a 3.8 MHz repetition rate. Start and stop signals were
detected by an avalanche photodiode and a Hamamatsu R28090-
07 microchannel plate photomultiplier, respectively. A Tennelec
TC 454 quad constant fraction discriminator, a TC 864 time-
to-amplitude converter, and a Nucleus PCA-II multichannel
analyzer card were used. The estimated time resolution of this
setup was about 50 ps.
Preparation of (Di-2-naphthalenylmethane)hydrazide-4-me-
thylbenzenesulfonic Acid. Di-2-naphthalenylmethanone (1.11 g,
3.9 mol) and tosylhydrazone (0.74 g, 3.9 mol) were dissolved
in toluene and refluxed under argon overnight. The solvent was
removed under vacuum, and the resulting solid was recrystal-
lized from ethanol to yield colorless needles of (2,2-naphtha-
lenylmethane)hydrazide-4-methylbenzenesulfonic acid (1.41 g,
3.1 mole, 80% yield); mp 188-190 °C. IR (KBr) νmax: 3289,
1387, 1166 cm-1. 1H NMR (CDCl3, 200 MHz): δ 8.1-7.8 (m,
8H), 7.7-7.6 (m, 5H), 7.5-7.3 (m, 5H), 7.22 (dd, 8.5 and 1.6
Hz, 1H), 2.45 (s, 3H, -CH3) ppm. 13C NMR (CDCl3, 75
MHz): δ 154.4, 144.2, 135.6, 134.1, 134.0, 133.7, 133.3, 132.7,
130.0, 129.7, 129.0, 128.4, 128.4, 128.1, 128.1, 128.0, 127.7,
127.6, 127.3, 127.1, 126.4, 125.1, 123.8 ppm. MS (m/e, relative
intensity): 450 (M+, 81), 254 (30), 155 (100), 127 (89). Mass
calculated for C28H22N2O2S: 450.14036. Found: 450.13824.
The optical hole-burning setup (of the Bordeaux Laboratory)
used a single-mode dye laser Coherent CR 699-21 with 1-3
MHz frequency resolution and 30 GHz scan width. The lasing
dye was rhodamine 6G. The laser light was stabilized with the
aid of an electrooptic modulator Conoptics Lass-II. Holes were
burned with light intensity between 0.1 and 0.01 mW/cm2 over
a burning time between 100 and 500 s. The monitoring intensity
was about 10 times weaker than that when burning holes. The
fluorescence light emitted from the sample was collected by an
achromatic lens and focused on the slit of a small monochro-
mator (with spectral resolution of a few nanometers). A
monochromator was used to select the main vibronic line
separated by 1353 cm-1 from the 0,0 origin of the fluorescence.
Preparation of Di-2-naphthyldiazomethane. Di-2-naphthyl-
diazomethane was prepared by following a method by Jonczyk
et al.9 (Di-2-naphthalenylmethane)hydrazide-4-methylbenzene-
sulfonic acid was dissolved in dioxane (40 mL) and sodium
hydroxide (2 mL, 50% in water) added to the solution. The
reaction mixture was stirred and heated at 85 °C for an hour.
The resulting solution was cooled to room temperature. The

2,2-Dinaphthylcarbene
J. Phys. Chem. A, Vol. 103, No. 17, 1999 3157
Figure 1. Fluorescence spectra of 22-DNC in n-hexane (solid line)
and in n-heptane (dotted line) at 5 K. The wavelength of excitation
light in both cases was 366 nm. The spectra presented are typical for
the annealed samples.
Figure 2. Fluorescence and fluorescence excitation spectra of 22-DNC
in n-hexane at 5 K after the sample was annealed. The wavelengths of
excitation (in the case of the fluorescence spectrum) and wavelengths
of observation (in the case of the fluorescence excitation spectrum)
are indicated above the spectra. The numbers in parentheses indicate
the vibrational frequencies (given in cm-1) of the 0,0 origins of both
main conformers (given in nanometers).
The excitation light, which was scattered off the sample, was
further removed with a RG630 Schott glass filter. An RCA
31034-A02 photomultiplier, cooled to -20 °C and operating
in the photon-counting mode was used to detect the light.
III. Results
A. Fluorescence and Fluorescence Excitation Spectra of
22-DNC. The fluorescence spectra of 22-DNC in Shpolskii
matrices of n-hexane and n-heptane at 5 K, when excited by
the 366 nm line to a higher excited triplet state, are shown in
Figure 1. The spectra can be attributed to the two main
conformers of 22-DNC. By analogy to 2-NPC,10 we can
presume that the pseudo-E/trans conformer absorbs at higher
energy and contributes to the spectrum with the 0,0 origin at
about 590 nm whereas the pseudo-Z/cis contributes to the
spectrum with the 0,0 band around 635 nm. It has been
experimentally observed that the former conformer dominates
in the n-hexane matrix and that the latter is the only one present
in frozen n-heptane.
The spectra of both conformers are composed of narrow
vibronic lines that can be resolved by using appropriate laser
excitation. With the selective laser excitation of the main
vibronic component of the T1 state of 22-DNC in n-hexane,
1343 cm-1 (λexc ) 546.9 nm) for the high energy and 1312
cm-1 (λexc ) 585.2 nm) for the low-energy conformer, the
fluorescence spectra can be separated, as shown in Figure 2.
The respective 0,0 fluorescence origins are located at 590.4 and
633.7 nm. The spectral resolution of the fluorescence spectra
presented is limited by the monochromator (half-bandwidth
above 30 cm-1). The fluorescence excitation spectra, observed
at the main vibronic component of fluorescence at 1353 cm-1
(λobs ) 643 nm) and measured with a laser spectral resolution
of 0.2 cm-1, are much more narrow. They are composed of
sharp quasi-lines, and the 0,0 origins of the fluorescence and
fluorescence excitation spectra coincide. Both of these observa-
tions indicate that the spectrum of 22-DNC in Shpolskii matrix
is composed of zero-phonon lines. Thus, the system seems to
be a good candidate for hole-burning experiments.
Figure 3. Fluorescence and fluorescence excitation spectra of 22-DNC
in n-heptane at 5 K after the sample was annealed. The wavelength of
excitation (in the case of the fluorescence spectrum) and the wavelengths
of observation (in the case of the fluorescence excitation spectrum)
are indicated above the spectra. The numbers in parentheses indicate
the vibrational frequencies (given in cm-1) of the 0,0 origins (given in
nanometers).
The fluorescence and fluorescence excitation spectra of 22-
DNC in n-heptane are shown in Figure 3. The origins of
fluorescence and fluorescence excitation lines coincide at 637.5
and 628.6 nm. These lines can be identified with the low-energy
conformer, presumably pseudo-Z/cis. In the n-heptane matrix
we were not able to observe any line that can be assigned to
the high-energy conformer. The pseudo-E/trans conformer
should have its origin at around 590 nm.

3158 J. Phys. Chem. A, Vol. 103, No. 17, 1999
Kozankiewicz et al.
Figure 4. The 0,0 fluorescence excitation bands of 22-DNC in
n-hexane at 4.2 K measured just after photolysis (dotted line) and the
next day, after annealing at 100 K (solid line).
The spectra presented in Figures 2 and 3 were collected for
the annealed samples. However, it is a characteristic feature of
the high-resolution spectroscopy of carbenes that the spectrum
depends on the thermal treatment of the sample and the spectrum
changes after annealing. This phenomenon is illustrated in Figure
4. The 0,0 fluorescence excitation origin lines of the high-energy
conformer of 22-DNC in n-hexane at 5 K, observed at 16 866
and 16 890 cm-1 before annealing, rearrange to higher energy
after the overnight annealing, and the main line is located at
16 933 cm-1. After the annealing procedure the fluorescence
spectrum shifts to higher energy (by about 60 cm-1) as well.
The observed spectral changes reflect a rearrangement of the
22-DNC geometry from one resembling that of its precursor,
22-dinaphthyldiazomethane, to the geometry of the relaxed 22-
DNC.
B. Fluorescence Decays. Fluorescence decays for both
conformers of 22-DNC in n-hexane matrix at 5 K are shown in
Figure 5. All of the decays were nonexponential as observed
previously for many other carbenes.4,10,11 This phenomenon has
been interpreted as a result of different intersystem crossing rates
from the three triplet sublevels of the T1 state that were
manifested by a triexponential fluorescence decay. Since the
transition moments are spin-independent, the radiative rate
constants for the three spin sublevels should be equal and the
excitation pulse should equally populate the three spin sublevels
of the T1 state. Thus, the experimental decays are expected to
be a sum of three exponential decays with the same preexpo-
nential factors. The above consideration does not take into
account the Boltzman population of the spin sublevels of the
T0 state. By use of the energy spacing between these spin
sublevels (T0x and T0y are 0.455 and 0.499 cm-1 above the
lowest energy T0z sublevel; this spacing will be found in the
following part of this work), the relative Boltzman populations
of the spin sublevels at 5 K are 1:0.876:0.865. Because of the
lack of symmetry of 2,2-DNC, the electronic transition from
each spin sublevel of the T0 state to each of the T1 state is
possible.
Figure 5. Fluorescence decays of 22-DNC in n-hexane at 5 K: (A)
λexc) 590.4 nm, λobs ) 643 nm corresponds to the pseudo-E/trans, in
which the best fit is given by the dependence (where t is in nanoseconds)
0.0405 exp(-t/5.9) + 0.036 exp(-t/28.1) + 0.0355 exp(-t/28.1); (B)
λexc ) 585 nm, λobs ) 633.7 nm, corresponds to the pseudo-Z/cis
conformer, in which the best fit is given by the dependence 0.049 exp-
(-t/7.0) + 0.0436 exp(-t/15.6) + 0.043 exp(-t/7.0). Plots of residuals
of a least-squares fit to a three-exponential decay dependence are added
below the decays.
in Figure 5. The crucial conditions to be met to obtain a
successful fitting (convolution with the excitation pulse) were
high-quality decay curves (to fulfill this requirement, we
collected the decay curves up to 104 counts in maximum) and
a precise estimation of the baseline. Under these conditions,
the iterative, nonlinear least-squares fitting program found that
the best approximation of the decays was a three-exponential
decay with the preexponential factors proportional to the
Boltzman population and with the component decay times of
5.9, 28.1, and 28.1 ns for the pseudo-E/trans conformer and
7.0, 7.0, and 15.6 ns for the pseudo-Z/cis conformer. In the
case of fitting by three-exponential decays with the same
preexponential factors, the fitting quality was acceptable but
not as good as in the above case (as judged by slightly higher
value of the ø2 and less equilibrated residual plot), but the decay
components are very similar: 5.8, 27.8, and 27.8 ns for the
pseudo-E/trans and 7.0, 7.0, and 15.6 ns for the pseudo-Z/cis
Therefore, we considered two possible situations and fitted
the decays with a three-exponential function with the same
preexponential factors and with a function where preexponential
factors were proportional to the Boltzman population. In each
case there were four fitting parameters. The best fits are shown

2,2-Dinaphthylcarbene
J. Phys. Chem. A, Vol. 103, No. 17, 1999 3159
Figure 6. Typical holes burned within the inhomogeneous 0,0
fluorescence excitation line (around 16 900 cm-1) of 22-DNC in
n-hexane at 1.7 K. The spectral positions of holes are given with respect
to the frequency of the burning light. The scanning range was limited
to 30 GHz, and therefore, the low- and high-energy sides of the hole
spectrum required separate detection.
Figure 7. Detailed structure of holes in the central part of hole-burning
spectrum. Holes are assigned by the same letters as in the spectrum of
holes in the model presented in Figure 9. The fit by seven Lorentzians
is given by the solid line. The Lorentzian parameters are collected in
Table 1.
conformer. The precise value of each decay component changed
slightly with varying pulse and decay baselines, but one decay
component was always different from the other two, which were
approximately the same. The result indicates that in the case of
the pseudo-E/trans conformer the spin-orbit interaction ef-
ficiently couples a singlet manifold with one spin sublevel of
the T1 state and that this sublevel is depopulated more rapidly
than the other two.
The above results suggest that the fluorescence decay curves
in the case of 22-DNC should be well approximated by a
biexponential dependence. Therefore, we also fitted the decays
with a biexponential function and found that fits were even
better, the ø2 values were smaller, than in the case when we
approximated decays by the sum of three exponentials. The best
fits to a biexponential dependence were given by 0.042 exp(-
t/6.1) + 0.069 exp(-t/27.6) for the pseudo-E/trans and 0.103
exp(-t/7.7) + 0.030 exp(-t/17.2) for the pseudo-Z/cis con-
former.
C. Hole-Burning Experiments. Hole-burning experiments
were performed on the 0,0 fluorescence excitation line, which
has its maximum at 16 933 cm-1 with a full width at half of
maximum (fwhm) of about 6 cm-1. It was attributed to the high
energy pseudo-E/trans conformer of 22-DNC. This conformer
dominated the spectrum in n-hexane and was absent in n-
heptane. In the present experimental configuration we were not
able to extend the hole-burning studies to the low-energy
conformer of 22-DNC because our single-mode dye laser system
was limited to the lasing range of rhodamine 6G (570-620 nm
range). Hole-burning studies with the low-energy conformer will
be the subject of forthcoming investigations.
A typical pattern of holes, burned within the inhomogeneous
16 933 cm-1 line of the 0,0 fluorescence excitation, is shown
in Figure 6. The central hole, burned at the frequency of the
applied laser light, had a fwhm of about 0.3 GHz and a depth
of 25% (after prolonged burning we were able to create hole of
a depth reaching 50%). Several substructures appeared around
this line in a frequency range of -2 to +2 GHz, and this part
of the hole-burning spectrum is presented in Figure 7. Two
satellite holes of similar width were located on both sides of
the central hole, 0.36 ( 0.01 GHz away. Two other pairs of
satellite holes, separated by 0.35 ( 0.01 GHz within the pair,
were located 1.11 ( 0.01 GHz from the center of the spectrum.
When we scanned a broader frequency range around the central
hole (Figure 6), we found a pair of deep holes at -13.65 (
0.05 and -14.95 ( 0.05 GHz and a pair of antiholes at 13.6 (
0.1 and 14.9 ( 0.1 GHz. The area of antiholes on the high-
energy side was much smaller than that of the deep holes on
the low-energy side.
IV. Discussion
The observed complicated pattern of holes can be explained
when taking into account the zero-field splitting (ZFS) of the
T0 and T1 states.
The ZFS parameter D is proportional to 1/r3 , where r is
the average separation between the two unpaired electrons and
depends therefore on the distribution of these electrons. In the
electronic ground state, T0, of carbenes, both unpaired electrons
are localized on the central carbene atom, one electron occupy-
ing a σ orbital and the other a π orbital.2,3 This implies that the
ZFS parameter D0 has a large value. In the excited T1 state,
one of the unpaired electrons is substantially delocalized onto
the naphthalene ring (occupying a π* orbital), whereas the
second one remains on the carbenic center.4,12 Consequently,
the average separation of the two unpaired electrons is much
larger in the T1 than in the T0 state and the D1 parameter is
small, much smaller than the D0 value. It is therefore reasonable
to predict that the holes and antiholes located far from the central
hole (13-15 GHz away) should reflect the ZFS of the ground
T0 state, whereas the holes located in the neighborhood of the
central hole (within the (2 GHz range) should reflect the ZFS
of the excited T1 state.
Let us first consider the simplified situation where we can
neglect a splitting of the T1 state. The energy level scheme for
this situation is shown in Figure 8. We used the axis convention
of Brandon et al.,13 where the z axis passed through the central
carbene atom being parallel to the line joining the centers of
the two adjacent carbon atoms and the x axis was perpendicular
to the plane defined by these three carbons. The order and
relative separations of the ZFS sublevels of the ground state,
T0z, T0x, and T0y, were assumed to be the same as that established
from electron spin echo experiments for diphenylcarbene.14
Among the molecules that contribute to the inhomogeneous 0,0
fluorescence excitation line, the hole-burning laser light can
interact only with three groups of molecules, those that have
the same T0z f T1, T0x f T1, and T0y f T1 transition energy

3160 J. Phys. Chem. A, Vol. 103, No. 17, 1999
Kozankiewicz et al.
experimentally demonstrated for diphenylcarbene14 and 2-naph-
thylphenylcarbene.5 If the molecules excited to the T1 state have
to relax to their T0z sublevel, then the only group of molecules
that can efficiently absorb many photons and be burned are those
that have the T0z f T1 transition in resonance with the laser
(hole burning) light. Other groups of molecules (2 and 3) that
also satisfy the energy resonance requirement are protected from
burning as long as they are trapped in the T0z state. Spin-lattice
relaxation, which restores the Boltzman population of the spin
sublevels, is usually very long at 1.7 K, and therefore, the
number of photons that can be absorbed by molecules belonging
to groups 2 and 3 is relatively small. We can even predict that
instead of holes, we should observe temporal (on the time scale
of spin-lattice relaxation time) antiholes, corresponding to the
T
0z f T1 transition energies of molecules from groups 2 and 3.
The observation of persistent antiholes that do not disappear
for a long time (after 15 min we were not able to monitor any
drop of intensity of antiholes presented in Figure 6) may indicate
the existence of carbenes that are characterized by extremely
long spin-lattice relaxation times. Probably this long relaxation
time can be related to a large energy separation between the
spin sublevels of the T0 state, but this problem is not clear and
needs further investigation. As a reminder, in the case of
2-naphthylphenylcarbene,5 where the ZFS of the T0 state is
similar, we were not able to detect any antihole.
Figure 8. Energy level schemes and the main electronic transitions
considered in the text. The thick lines indicate the burning transitions
resonant with laser light. The schemes in this figure neglect the ZFS
of the excited T1 state. Spectral positions of the possible absorption
transitions (and holes) of the three groups of molecules, which can be
in resonance with a laser frequency, are shown on the right side of the
figure.
(in Figure 8 the transitions resonant with the laser light are
indicated by thick lines). If the burning efficiency is the same
for these three groups of molecules, the central hole should be
surrounded by three holes on the high- and three on the low-
energy side. The experimental hole spectrum differs from these
predictions. The observed pair of holes located on the low-
energy side of the central hole can be created by molecules from
group 1 only, whereas the antiholes, located on the high-energy
side of the central hole, are derived from molecules from groups
2 and 3.
The spectral positions of deep holes (and antiholes) provide
direct information about the ZFS of the T0 state. The energy
separations between the spin sublevels are
∆E(T0z,T0x) ) D0 - E0 ) 13.65 ( 0.05 GHz
and
∆E(T0x,T0y) ) 2E0 ) 1.3 ( 0.05 GHz
22-DNC seems to be photochemically stable in low-temper-
atures matrices; we were able to study this compound for several
days (burning new holes after the overnight annealing) with no
detectable drop of the total emission intensity. The hole-burning
process should therefore be nonphotochemical and proceed via
nonradiative relaxation channels. The relaxation energy, when
dissipated to the matrix surrounding the molecule, can change
the molecule-matrix local configuration, leading to a new
energy minimum. Consequently, the zero-phonon line of the
carbene can be spectrally shifted, leaving a spectral hole at the
previous position. The following discussion provides arguments
that this is the mechanism of burning holes in the 0,0
fluorescence excitation line of 22-DNC. There are two possible
channels of nonradiative relaxation, the T1 f To internal
conversion and the T1 f Sn f S1 f T0 intersystem crossing.
The former process does not differentiate among the three
groups of molecules under consideration. The latter process is
known to be highly selective for different ZFS triplet compo-
nents, and if dominant, it may explain the observed pattern of
burned holes. Let us first consider the consequence of high
selectivity of the final step, S1f T0, in the intersystem crossing
pathway. In the T0 and S1 states of carbenes (a good prototype
is methylene, CH2) both unpaired electrons are localized on the
central carbene atom, occupying (px)1(spy)1 and (px)2 orbitals,
respectively. One center spin-orbit coupling on carbon (the
spin-orbit operator l transforms like a 90° rotation) can mix
the T0 and S1 states only through the matrix elements whose
orbital part has the form px|lz|py .15 This implies that the S1 f
These lead to
E0 ) 0.65 ( 0.03 GHz ) 0.022 ( 0.001 cm-1
D0 ) 14.3 ( 0.03 GHz ) 0.477 ( 0.001 cm-1
Discussion of the origin of satellite holes located in the
neighborhood of the central hole is possible if we consider the
ZFS of the excited T1 state, which was neglected in the previous
discussion. Distribution of the electronic density is different in
the T0 and T1 state, and therefore, the principal spin axes x, y,
and z in the T0 state can rotate to new x′, y′, and z′ axes in the
T1 state. Consequently, electronic transitions from each spin
sublevel of the T0 to each spin sublevel T1 are possible.
In the following discussion we will limit our considerations
to molecules that can be burned (and thus the origin of the spin
sublevel in absorption is the T0z state). Three possible transitions,
T0z f T1z′, T0z f T1x′, and T0z f T1y′, define three groups of
molecules (1a, 1b, and 1c) that can efficiently interact with the
laser hole-burning light. The corresponding energy level schemes
and electronic transitions are shown in Figure 9. The model
predicts that the group of molecules 1a should contribute to
the central hole and to two holes on the high-energy side, group
1b to the central hole and to one hole on the high- and one on
the low-energy side, and group 1c to the central hole and to
two holes on the low-energy side of the central hole. The
predicted pattern of holes is shown on the right side of Figure
9, and it can be compared with the spectrum of holes observed
in the neighborhood of the central hole, shown in Figure 7. It
T
0z intersystem crossing channel dominates S1 f T0x and S1 f
T0y. The selective population of the T0z sublevel was already

2,2-Dinaphthylcarbene
J. Phys. Chem. A, Vol. 103, No. 17, 1999 3161
To provide an explanation, we propose a simple kinetic
model. We assume that the number of burned molecules from
groups 1a, 1b, or 1c is proportional to
Azi′kisc
i′
where i′ is the spin sublevels z′, x′, or y′, Azi′ is the rate constant
for the T0z f T1i′ absorption, and kisci′ is the intersystem crossing
rate constants for the T1i′ f Sn depopulation. We further assume
that the hole-reading process is described by the expression
Azi′kr/kisc
i′
where kr is the radiative rate constant, equal for all spin sublevels
of T1. We also neglected radiative and internal conversion
contributions to the total depopulation rate constant of each spin
sublevel of T1. In other words we replace this rate constant by
the dominating intersystem crossing rate. The real hole burning
experiment has a contribution from both burning and reading
processes, and thus, the areas of satellite holes can be ap-
proximated by the following equations:
Figure 9. Energy level schemes and the electronic transitions that
contribute to the spectrum of satellite holes surrounding the central
hole. The thick lines indicate the burning transitions resonant with laser
light. Spectral positions of absorption transitions (and holes) of the three
groups of molecules, which can be in resonance with a laser frequency,
are shown on the right side of the figure.
Azz′kiscz′Azy′kr/kisc
Azz′kiscz′Azx′kr/kisc
S1 ) 0.041
S2 ) 0.032
S3 ) 0.031
(1)
(2)
(3)
(4)
(5)
(6)
y′
TABLE 1: Parameters of Seven Lorentzian Functions
Adopted to Fit the Central Part of the Hole-Burning
Spectrum Presented in Figure 9a
x′
Azx′kiscx′Azy′kr/kisc
ν0 [GHz]
fwhm [GHz]
S
y′
a
a
b
1.47
1.11
0.36
0.29
0.29
0.28
0.27
0.28
0.29
0.29
0.041
0.032
0.031
0.07
0.043
0.033
0.068
(Azz′)2kr + (Azx′)2kr + (Azy′)2kr S4 ) 0.069
a, b, c
0.00
Azy′kiscy′ Azx′kr/kisc
Azx′kiscx′Azz′kr/kisc
S5 ) 0.043
S6 ) 0.033
S7 ) 0.068
c
b
c
-0.36
-1.11
-1.47
x′
z′
a The sum of Lorentzians was given by the solid line in Figure 9.
Each of the Lorentzian functions has been characterized by the formula
(2/π)Sw/(4(ν - ν0)2 + w2), where fwhm is a full width at half of
maximum, S is the area, and ν0 is the frequency of the Lorentzian
maximum.
Azy′kiscy′Azz′kr/kisc
(7)
z′
From (eq 1)/(eq 2) and (eq 6)/(eq 7), we can obtain Azy′
≈
1.42Azx′ and kiscy′ ≈ 1.27kiscx′. From (eq 1)/(eq 3) and (eq 2)/(eq
6), we can obtain Azz′ ≈ 1.35Azx′ and kisc ≈ 0.98kisc
.
x′
is straightforward to identify the predicted holes and assign them
according to the model. To go deeper into the origin of the
burned holes, we fitted the experimental spectrum with seven
Lorentzian curves as shown in Figure 7. The Lorentzian
parameters (frequency at the maximum, ν0; full width at half
of maximum, fwhm; area, S) are collected in Table 1.
z′
If one accepts the above rates, the predicted area of the central
hole (with respect to the area of other holes) should be much
larger than that observed. There are two possible explanations
for the discrepancy. The experimental spectrum of holes
presented in Figure 7 may have been obtained at the fluence
saturation condition. To solve this problem, we would have to
study the shape of the holes as a function of burning fluence
and extrapolate the results to zero fluence. Unfortunately,
because of the poor signal-to-noise ratio at low fluence, this
extrapolation would be very difficult, and therefore, this
experiment was not performed. Alternatively, the fitting pro-
cedure may possibly have overestimated the area of the side
holes compared with the central hole.
The frequencies at the maxima of the holes are directly related
to the ZFS parameters D1 and E1 of the excited T1 state:
∆E(T1z′,T1x′) ) D1 - E1 ) 1.11 ( 0.01 GHz
∆E(T1x′,T1y′) ) 2E1 ) 0.36 ( 0.01 GHz
This gives
The same pattern of holes as those presented on the right
side of Figure 9 can be obtained for a negative value of D1,
when the ordering of spin sublevels energy is E(T1y′) < E(T1x′)
< E(T1z′). In this case, the same set of equations as eqs 1-7
E1 ) 0.18 ( 0.005 GHz ) 0.006 ( 0.0002 cm-1
D1 ) 1.29 ( 0.015 GHz ) 0.043 ( 0.0005 cm-1
leads to Azy′ ≈ 1.62Azx′, Azz′ ≈ 1.55Azx′, kisc ≈ 0.79kiscx′, and
y′
kisc ≈ 1.02kiscx′. Considering the spectrum of holes alone, it is
z′
The depth (and area) of burned holes should be related to
the burning efficiency of the contributing groups of molecules.
A close inspection of Figure 7 and Table 1 shows that the holes
assigned to molecules of group 1c are deeper and have larger
areas than those assigned to groups 1a and 1b, and holes of
these two last groups seemed to have similar areas.
impossible to decide which energy ordering is correct. The
argument to apply one or another order of spin sublevels can
be obtained from fluorescence decays. In the case of the pseudo-
E/trans conformer of 22-DNC we found that fluorescence decays
are composed of one short (5.9 ns) and two long (28.1 ns) time
components. The complex decay was interpreted to be a sum

3162 J. Phys. Chem. A, Vol. 103, No. 17, 1999
Kozankiewicz et al.
of three independent decays from three different spin sublevels.
Thus, one spin sublevel is characterized by fast rate constants
The pattern of observed holes was explained assuming that the
main depopulation channel of the excited T1 state was inter-
system crossing. Both intersystem crossing steps, T1f Sn and
S1f T0, contribute to the shape of the hole-burning spectrum.
The latter step selectively populates the T0z sublevel, and the
only group of molecules that can be burned absorb laser
radiation in this state. Another implication of the S1 f T0z
polarization is the possibility of the observation of (temporal)
antiholes. This unusual observation is probably related to an
extremely long spin-relaxation time in the T0 state. The former
intersystem-crossing step contributes to the relative areas of
satellite holes surrounding the central hole. Another factor
contributing to the areas of these holes originates from the fact
that absorption from T0z to different spin sublevels of T1 is
dependent on the projection of direction z onto new directions
of the principal spin axes z′, x′, and y′. Analysis of frequency
positions of satellite holes provided the ZFS parameters of the
T0 and T1 states: E0 ) 0.022 ( 0.001 cm-1, D0 ) 0.477 (
0.001 cm-1, E1 ) 0.006 ( 0.0002 cm-1, and D1 )0.043 (
of intersystem crossing kisc ) 1/(5.9 ns) ≈ 1.7 × 107 s-1
,
)
whereas the two other spin levels have much slower kisc
1/(28.1 ns) ≈ 3.6 × 107 s-1 rate constants (using the assumption
that intersystem crossing is the main and dominating channel
of the T1 state depopulation). The situation that one intersystem
crossing rate (kiscy′) is bigger than two others (which are nearly
equal) is predicted by a model where the lowest energy sublevel
is the T1z′ sublevel and thus a positive D1 (the competitive model
where T1z′ is the highest energy spin sublevel predicts that kisc
y′
is smaller than the other two). As a reminder, the fluorescence
decay was best fitted by a three-exponential dependence where
the biggest preexponential factor was connected with the shortest
decay component (taking into account the Boltzman population
of the spin sublevels of T0 at 5 K). This observation agrees
with the prediction of the “hole-burning” model, which gives
the highest value for Azy′, for the absorption from the lowest
energy (and thus the highest population) spin sublevel T0z to
the fastest decaying spin sublevel T1y′. The agreement, however,
is only qualitative because from the analysis of the hole-burning
0.0005 cm-1
.
22-DNC is the second system where hole-burning spectra
have been observed and successfully analyzed, but in this case,
in a more elaborate fashion than the previous study of 2-NPC.
We think that the hole-burning method has demonstrated useful
possibilities in the case of triplet-triplet transitions and should
be extended to other systems with the triplet ground states.
spectrum we obtained kisc ≈ 1.27kiscx′, whereas from fluores-
y′
cence decays kisc ≈ 4.6 × kisc
.
x′
y′
An implication of the simple kinetic model presented above
is the possibility of obtaining information about the spatial
orientation of the principal spin axes in the T1 state with respect
to the z direction, defined in the ground state. The absorption
transition moments are spin-independent, and thus, the partition
References and Notes
of the absorption rate constant A into the components Azz′
)
(1) (a) Kirmse, W. Carbene Chemistry; Academic Press: New York,
1964. (b) Carbenes; Jones, M., Jr., Moss, R. A., Eds; Wiley: New York,
1973 and 1975; Vols. I and II. (c) Wentrup, C. ReactiVe Molecules; Wiley:
New York, 1984.
(2) Trozzolo, A. M.; Wasserman, E. In Carbenes; Moss, R. A., Jones,
M., Jr., Eds.; Wiley: New York, 1975; Vol. II, p 185.
(3) Closs, G. L. In Carbenes; Moss, R. A., Jones, M., Jr., Eds.; Wiley:
New York, 1975; Vol. II, p 159.
(4) Migirdicyan, E.; Kozankiewicz, B.; Platz, M. S. In AdVances in
Carbene Chemistry; Brinker, U., Ed.; JAI Press Inc.: Stamford, CT, 1998;
Vol. 2, p 97.
(5) Kozankiewicz, B.; Bernard, J.; Migirdicyan, E.; Orrit, M.; Platz,
M. S. Chem. Phys. Lett. 1995, 245, 549.
0.34A, Azx′ ) 0.25A, and Azy′ ) 0.41A (describing absorption
from the T0z to the T1z′, T1x′, and T1y′) should depend on the
projection of direction z onto the directions of the new principal
spin axes z′, x′, and y′, respectively.
According to the model, each of the deep holes at -13.65
and -14.95 GHz (as well as antiholes on the high-energy side)
should be surrounded by narrow satellite holes in a fashion
similar to that of the central hole. However, this is not the
experimental observation. To explain this result, we have to
consider that the frequency position of these holes is mainly
determined by the D0 parameter, which can differ slightly for
molecules contributing to the hole. It results in a smoothing of
the substructures in the profile of holes.
(6) Kozankiewicz, B.; Bernard, J.; Migirdicyan, E.; Orrit, M.; Platz,
M. S. Mol. Cryst. Liq. Cryst. 1996, 283, 191; 1996, 291, 143.
(7) Prakash, G. K. S.; York, C.; Liao, Q.; Kotian, K.; Olah, G. A.
Heterocycles 1995, 40, 79.
Finally, let us mention that we observed a magnetic field
effect on the shape of the hole-burning spectrum. The result
was a smoothing of the holes under increasing magnetic flux
up to 320 G, which is a consequence of a lack of macroscopic
orientation of carbenes in a Shpolskii matrix. The analysis of
the magnetic field effect will be the subject of forthcoming
investigations.
(8) Beckmann, E.; Liesche, O.; Correns, E. Chem. Ber. 1923, 56, 351.
(9) Jonczyk, A.; Wlostowksa, J. Synth. Commun. 1978, 8, 569.
(10) Despres, A.; Lejeune, V.; Migirdicyan, E.; Admasu, A.; Platz, M.
S.; Berthier, G.; Pariser, O.; Flament, J. P.; Baraldi, I.; Momicchioli, F. J.
Phys. Chem. 1993, 97, 13358.
(11) Kozankiewicz, B.; Despres, A.; Lejeune, V.; Migirdicyan, E.; Olson,
D.; Michalak, J.; Platz, M. S. J. Phys. Chem. 1994, 98, 10419.
(12) Despres, A.; Lejeune, V.; Migirdicyan, E.; Platz, M. S. J. Phys.
Chem. 1992, 96, 2486.
(13) Brandon, R. W.; Closs, G. L.; Hutchison, C. A., Jr. J. Chem. Phys.
1962, 37, 1878.
V. Conclusion
(14) Doetschman, D. C.; Botter, B. J.; Schmidt, J.; van der Waals, J. H.
Chem. Phys. Lett. 1976, 38, 18.
(15) McGlynn, S. P.; Azumi, T.; Kinoshita, M. Molecular Spectroscopy
of the Triplet State; Prentice-Hall: Englewood Cliffs, NJ, 1969.
This work presents and discusses the successful observation
of hole burning on the inhomogeneous 0,0 line of the fluores-
cence excitation spectrum of 22-DNC in n-hexane at 1.7 K.