Third-Order Nonlinear Optical Properties of Sulfur-Rich Compounds

Article abstract of DOI:10.1021/jp990137f

The molecular third-order optical nonlinearity γ_R (second hyperpolarizability or nonlinear refractive index) was measured for a series of sulfur-rich molecules: sulfur (S₈), carbon-sulfur compounds, and metal thiolate clusters. Z-sca

Full text of DOI:10.1021/jp990137f

6930
J. Phys. Chem. A 1999, 103, 6930-6937
Third-Order Nonlinear Optical Properties of Sulfur-Rich Compounds
†
Jonathan G. Breitzer,* Dana D. Dlott, Lawrence K. Iwaki, Sean M. Kirkpatrick, and
Thomas B. Rauchfuss
School of Chemical Sciences, Chemical and Life Sciences Laboratory, UniVersity of Illinois at
UrbanasChampaign, Urbana, Illinois 61801
ReceiVed: January 11, 1999; In Final Form: April 19, 1999
The molecular third-order optical nonlinearity γ
R
(second hyperpolarizability or nonlinear refractive index)
), carbon-sulfur compounds, and metal thiolate
was measured for a series of sulfur-rich molecules: sulfur (S
8
clusters. Z-scan techniques (pulse width 27 ps, wavelength 527 nm) were used to measure these properties in
solution by comparing the solution to the pure solvent. Our approach is an efficient way to evaluate a number
of different compounds and to quickly direct synthetic strategies for improved nonlinear and linear optical
-
1
properties. The optical nonlinearities were evaluated in terms of figures of merit, |W|/I
0
and [T] , where
is the ratio of nonlinear refraction to linear absorption and [T] is the ratio of nonlinear refraction to
had the largest figures of merit: |W|/I
-
1
|W|/I
0
nonlinear absorption. Among the carbon-sulfur compounds, C
6
S O
8 2
0
-
13
2
-1
-1
)
4.3 × 10 m W and [T] > 5. The metal thiolate cluster with the largest second hyperpolarizability
4- -56 4 -3 -31
4 R
was [Zn10
S
(SPh)16
]
(γ
) -7.8 × 10 C m V , -6.3 × 10 esu). This cluster exhibited no measurable
linear or nonlinear absorption, so the figures of merit were effectively infinite. Previous work on the second
hyperpolarizability of sulfur-rich compounds examined species that were hampered by substantial linear
absorption coefficients. The present work shows that high figures of merit can be achieved without significant
linear or nonlinear absorption.
Introduction
In this work, we evaluate the NLO properties of candidate
molecules as solutions, correcting for the contributions of the
solvent. With this method, we can choose any convenient solvent
for a particular sample, so it allows us to survey the optical
properties of a wide range of molecules.
The prospect of applying the molecular level control inherent
in organic and organometallic syntheses has motivated many
chemists to pursue the development of new second- and third-
order nonlinear optical (NLO) materials. Indeed, many molecular
organic and organometallic compounds have been shown to
possess large molecular hyperpolarizabilities. Furthermore, it
has been shown that macroscopic assemblies of these species
Ideally, NLO measurements would probe the nonlinear
refraction and absorption of the candidate molecules at all pulse
widths and at all wavelengths. While this is usually not practical
because of instrumental limitations, we stress the importance
of making the NLO measurements with ultrashort light pulses
(
2)
(3)
exhibit sizable second-order (ø ) or third-order (ø ) optical
nonlinearities.1
(
picoseconds or less) in order to obtain the intrinsic contribution
An ultimate practical goal for designing molecules with large
third-order nonlinearities is to incorporate them into devices used
in all-optical signal processing. Realization of this goal entails
recognizing that the molecule must not only have a large, fast
optical nonlinearity (øR > 10-30 C m V , 10
species must also have a small absorption, be it linear or
nonlinear. A low absorbance is necessary; otherwise, slower
photothermal effects dominate the desirable fast photorefractive
effects and the optical damage threshold may be unacceptably
low. Although not considered here, additional practical consid-
erations influence the selection of materials for NLO applica-
tions, including factors related to processability and chemical
stability.
(3)
to ø and to choose a wavelength region that holds practical
interest. In the present case, we chose to work at 527 nm in
order to evaluate materials for visible wavelengths in which
there exists interest for switchable holographic filters and
2
(3)
-3
-11
esu), these
4
refractive optical limiters. In addition, tabulated data at a nearby
wavelength of 532 nm, at about the same pulse duration, exist
for some of the solvents used here, enabling us to verify our
measurement technique. The use of 27 ps pulses allows us to
measure contributions to ø( arising from electronic response
and molecular reorientation. Other effects such as electrostriction
and thermal lensing are much slower, and we wish to avoid
measuring their effects because they are not relevant to the
properties of useful fast-response devices. The build-up times
for these effects are roughly equal to d/c, the laser-beam diam-
3)
3
Figures of merit have been used to evaluate how well a
molecule satisfies the requirement of a large nonlinear refractive
index and small absorption. One does not wish to synthesize a
molecule and then fabricate the NLO material, for example, by
polymerizing or crystallizing it, only to find that the material’s
absorptive properties outweigh its useful refractive properties.
5
eter divided by the acoustic velocity. For the 40 µm diameter
beam used here and a typical acoustic velocity of 4 km/s, the
build-up times are about 10 ns, which are too slow to affect
our measurements. However, the decay of a thermal lens via
thermal diffusion can be much slower than the build-up; in fact,
it may be as slow as several milliseconds. If the laser repetition
rate is too large, undesirable cumulative effects emerge. Two
methods can be used to eliminate these cumulative effects of
*
Author to whom correspondence should be addresssed.
Present address: Air Force Research Laboratory, AFRL/MLPJ Bldg.
†
6
51, 3005 P ST STE 1, WPAFB, OH 45433-7702.
1
0.1021/jp990137f CCC: $18.00 © 1999 American Chemical Society
Published on Web 08/13/1999

Sulfur-Rich Compounds
J. Phys. Chem. A, Vol. 103, No. 35, 1999 6931
thermal lensing. Either the time between pulses can be made
much longer than the thermal lens decay time or a flowing
sample can be used. In our work, we chose to use a high
repetition rate (640 Hz) to maximize our signal-to-noise ratio,
so we cycled the liquid sample through a flow cell fast enough
(3)
R
ø
γR )
(4)
4
L N
and the imaginary part γI by
(
about 150 mL/min, 25 cm/s linear velocity) to ensure a fresh
(
I
3)
sample volume for each pulse.
ø
γI )
(5)
The present work was premised on two ideas: that sulfur
anions and organosulfur compounds are known to be quite
polarizableslarge, conjugated sulfur-containing molecules (e.g.,
the tetrathiafulvalenes ) especially sosand that highly polariz-
able species tend to exhibit large second hyperpolarizabilities.
The latter idea is supported qualitatively by previous third-order
4
L N
where L is the Lorenz correction factor,13 L ) (n0 + 2)/3, and
2
6
N is the number density.
It is useful to have figures of merit that combine the nonlinear
susceptibility and the linear or nonlinear absorption, referred
to as W/I0 and T , respectively. For almost any practical
material, the linear absorption coefficient, R0, must be small.
More transparent materials will therefore have larger values of
the figure of merit W/I0, defined as
7
8
NLO studies on polydiacetylene, oligothiophenes, and caro-
tenoids. This paradigm has led to recent studies on organ-
-1
14
9
10
otransition metal complexes. We were interested in extending
these studies to “sulfur-rich” molecules, i.e., molecules where
there is a high atomic concentration of S.1 Carbon disulfide
can be regarded as a prototypical sulfur-rich molecule, and it
has long been used as a standard in third-order NLO studies.
This work provides specific examples and general insights into
how sulfur-rich molecules may be tailored for NLO applications
while purposefully minimizing the nonlinear absorption to
achieve a favorable figure of merit.
1
I
2
n
W
I0
)
(6)
R0λ
At the large intensities needed to operate nonlinear optical
devices, the nonlinear absorption must be small as well. This
requirement is expressed as the figure of merit T ,
-
1
Theoretical Section. We assume our measurements are
(3)
dominated by the bulk third-order susceptibility ø . Of course,
ø
as ø should be negligible compared to ø . The real part, ø ,
gives rise to an intensity-dependent phase change, i.e., a
I
2
(2)
(4)
n
and ø vanish in liquids, and higher-order odd terms such
T-1
)
(7)
(5)
(3)
(3)
2λR2
R
(3)
T-1 is larger for materials with smaller nonlinear absorption.
nonlinear refraction. The imaginary part, ø , gives rise to an
I
intensity-dependent amplitude change, i.e., a nonlinear absorp-
tion. These susceptibilities are given (in SI units) in terms of
Experimental Section
I
2
the nonlinear refractive index, n , and the nonlinear absorption
Synthesis. Syntheses employed standard Schlenk techniques,
but workups were conducted in air. Unless noted otherwise,
reagents and solvents were used as purchased. The heterocyclic
compounds, C3S7O and C6S6O2, were prepared according to
coefficient R2,
(
R
3)
I
2
2
0
2
0
ø
) 2n n ꢀ c
(1)
15
16
literature methods starting from (Bu4N)2[Zn(C3S5)2]. The zinc
clusters, (Me4N)2[Zn4(SPh)6] and (Me4N)4[Zn10S4(SPh)16], were
and
1
7
synthesized according to Dance’s methods. n-Butylferrocene
n-BuFc) was purchased from Aldrich and used as received.
(
8
2
0
1
0 n R ꢀ
(
I
3)
2
0
The n-butylferrocene disulfide polymer (BuFcS2)n was made
from n-BuFc according to a literature procedure.
ø
)
(2)
18
2πkc
1
9
Synthesis of R-C3S3(SCH2Ph)2. This species was prepared
by treating a solution of (Bu4N)2[Zn(R-C3S5)2] (1.06 g, 1.12
mmol) in 30 mL of acetone with benzyl chloride (0.78 mL,
6.78 mmol) and stirring it for 16 h at room temperature, during
which time the solution turned from deep purple to orange. To
isolate the compound, the orange solution was concentrated
under reduced pressure until a precipitate began to form; the
resulting mixture was then diluted with an approximately equal
where n0 is the ordinary refractive index, c the velocity of light,
and k the magnitude of the wave vector, k ) 2π/λ, where λ is
1
2
the wavelength.
I
In our experiments, we measure the nonlinear refraction n2
and the nonlinear absorption coefficient R2 for a dilute solution
containing a known mole fraction f of solute. We also measure
the same properties for the pure solvent. By use of eqs 1 and 2,
(3)
volume of 2-propanol. Further concentration yielded R-C S -
these can be converted to third-order susceptibilities ø . We
then assume the solute and solvent properties are additive so
that
3 3
(
SCH2Ph)2 in the form of yellow crystals (0.70 g, 82%).
2
0
Synthesis of â-C3S3(SCH2Ph)2. To a 2.0 M solution of
sodium methoxide in methanol (16.5 mL) was added â-C3S5-
20
(COPh)2 (2.24 g, 5.52 mmol), and the resulting brown mixture
(
3)
(3)
solvent
(3)
solute
ø
) (1 - f)ø
+ fø
(3)
solution
was stirred at ambient temperature. Ammonium acetate (2.8 g)
was added, and the mixture was diluted with water, resulting
in a red solution. A solution of benzyl chloride (0.650 mL, 5.65
mmol) in methanol (14 mL) was added dropwise over 15 min,
during which time a yellow, crystalline solid formed. This was
collected by filtration, washed with water, and dried in vacuo
(0.62 g, 30%). Anal. Calcd for C17H14S5: C, 53.93; H, 3.73.
Equation 3 is assumed to be valid for both the real and imaginary
(3)
(3)
parts of ø . Using eq 3, we can determine ø for the solute.
To compare one molecular species to another, it is convenient
to define molecular nonlinear properties, termed molecular
hyperpolarizabilities γ. The real and imaginary parts, γR and
γI, characterize the nonlinear refraction and nonlinear absorption,
respectively. The real part γR is given by
1
Found: C, 53.53; H, 3.91. H NMR (CD3COCD3, 500 MHz):
δ 4.12 (s, 2H), 4.49 (s, 2H), 7.24 (m, 5H), 7.37 (m, 5H).

6932 J. Phys. Chem. A, Vol. 103, No. 35, 1999
Breitzer et al.
Synthesis of C6S8O2. A slurry of C6S1021 (2.83 g, 7.22 mmol)
in chloroform (100 mL) was treated with a slurry of 9.19 g
28.8 mmol) of mercuric acetate in glacial acetic acid (60 mL).
(
The orange suspension was heated to reflux for 17 h and allowed
to cool to ambient temperature, and a brown solid was removed
by filtration, leaving a yellow solution. This was washed twice
with water (0.1 L portions), then with 0.1 L of 3% aqueous
sodium carbonate, then again with two 0.1 L portions of water.
The organic fraction was dried over magnesium sulfate and
concentrated to half its volume. The golden solution was diluted
with an equal volume of isopropyl alcohol and concentrated
until yellow crystals (0.136 g, 5.2%) formed. IR (KBr): 1722
(
w), 1662 (s, νC-O), 1588 (sh), 996 (w), 872 (w), 734 (w), 420
-
1
(w) cm .
Synthesis of (Me4N)2[Cu4(SPh)6]. This compound was syn-
Figure 1. Apparatus used for Z-scan measurements. Key: BS ) beam
splitter; L ) lens; D ) silicon photodetector.
thesized by adapting Dance’s procedure.22 A 500 mL flask was
charged with 250 mL of oxygen-free ethanol followed by 4.00
g (25.9 mmol) of Me4NBr, 5.60 mL (54.5 mmol) of ben-
zenethiol, and 6.90 mL (49.5 mmol) of Et3N. The mixture was
stirred and heated at reflux for 30 min at which stage most of
the quaternary ammonium salt dissolved. After allowing the
mixture to cool to ca. 50 °C, the reaction solution was treated
with a solution of 2.90 g (12.5 mmol) Cu(NO3)2‚2.5H2O in 50
mL of oxygen-free ethanol over a course of 15 min. The
resulting pale-yellow, turbid mixture was reheated to the boiling
point and diluted with 40 mL of acetonitrile to partially dissolve
the precipitate. The mixture was allowed to cool to room tem-
perature. After the slurry stood undisturbed for 10 h, pale-yellow
flakes were collected by filtration in an oxygen-free atmosphere,
washed with 20 mL of water and 20 mL of methanol, and dried
in vacuo. Yield: 2.45 g (74% based on copper). Anal. Calcd
for C44H54N2Cu4S6: C, 49.97; H, 5.15; N, 2.65. Found: C,
transmitted through the sample is split into two parts. One part,
which generates the “open-aperture” data, is focused onto D2
(ThorLabs DET1-SI) with a lens. As the sample moves in and
out of the focus, nonlinear absorption causes the intensity seen
by D2 to vary. Nonlinear refraction (due to the real part of the
nonlinear susceptibility) has no effect on the signal seen by D2.
The second part, which generates the “closed-aperture” data,
goes through a 1.5 mm aperture in the far field before reaching
D1 (ThorLabs DET1-SI). As the sample moves in and out of
the focus, both nonlinear refraction and nonlinear absorption
effects cause the pulse diameter in the far field to vary. The
aperture converts this diameter variation into an intensity
variation detected by D1.
In our setup, a solution is pumped through an optical cell
(Uvonic 1 mm, type 48) at a velocity that ensured that a fresh
4
(
9.74; H, 5.20; N, 2.51. The bright-yellow pentacopper cluster,
Me4N)2[Cu5(SPh)7],22 can be converted to (Me4N)2[Cu4(SPh)6]
using the following procedure. (Me4N)2[Cu5(SPh)7] (0.264 g,
.214 mmol) was treated with acetonitrile (60 mL), resulting
in a yellow solution and some undissolved solid. Me4NBr (0.360
g, 2.34 mmol) and thiophenol (0.22 mL, 2.1 mmol) were added.
The mixture was then treated with triethylamine (0.5 mL),
whereupon the remaining yellow solid dissolved immediately,
the yellow color of the solution became much paler, and some
white solid precipitated. The mixture was then diluted with
-propanol (60 mL), stirred until the white solid redissolved
20 min), and concentrated until the liquid became slightly
turbid. The mixture was sealed and allowed to stand for 72 h,
at which time pale-yellow crystals formed (0.150 g, 53% based
on copper).
volume is irradiated by each succeeding pulse. That occurs when
the linear flow velocity V is much greater than dF, where d is
the Gaussian beam diameter and F the laser repetition rate. For
0
-1
a diameter d ) 40 µm and a repetition rate F ) 640 s , the
-1
velocity should exceed 2.5 cm s . The actual linear flow
-1
velocity used was V ) 25 cm s . In this manner, we measured
the fast electronic and rotational contributions while eliminating
undesirable cumulative effects such as thermal lensing. The cell
is mounted on a motorized translation stage with a location
sensor. A computer, with an analog-to-digital converter, samples
the output of the three detectors and the location sensor.
The intensity of the laser pulse at the focus in air is computed
knowing the energy, the Gaussian beam radius (r0 ) 20.0 ( 3
µm), and the pulse duration (tp ) 27 ps). Gaussian beam radius
2
(
2
denotes the 1/e intensity point. These parameters were deter-
Z-Scan Technique. Fundamentals of the Z-scan technique
have been discussed elsewhere,12 and so only a brief description
is given here. Figure 1 shows the experimental setup. A diode-
mined with a pulse energy meter, a knife-edge beam profile
measurement, and background-free autocorrelation. For each
sample, we obtained data on a dilute solution and on the pure
solvent. Data were taken at two intensities, 1.3 ( 0.2 and 3.6
3+
pumped mode-locked Nd :YLiF (YLF) oscillator generates a
20 ps, 1.053 µm pulse, which is injected in a YLF regenerative
∼
(
0.2 µJ, corresponding to peak intensities of 16 and 43 GW/
amplifier. The amplifier is cavity-dumped at a rate of 640 Hz.
This continuously pumped YLF amplifier produces a high-
quality beam with a single transverse mode Gaussian (TEM00)
profile. The cavity-dumped pulse of ∼800 µJ energy is
frequency-doubled to λ ) 527 nm. The duration of the 527 nm
pulse is 27 ( 2 ps. The pulse energy could be continuously
varied using a half-wave plate and a polarizing beam splitter.
2
cm . With the lower value, nonlinear effects are reduced in
magnitude, but any effects caused by optical inhomogeneities
in the moving cuvette remain, allowing the data to be corrected
for such effects.
The Z-scan analytical functions have been thoroughly detailed
elsewhere. We used the thin-sample analysis functions derived
12
by Sheik-Bahae et al. The thin-sample approximation is
In the Z-scan apparatus, the pulse is focused into a sample,
which is translated along the axis of pulse propagation (the
Z-axis). Three solid-state detectors are used. A portion of the
incoming pulse is split off into D3 (ThorLabs DET2-SI), which
is used to normalize for intensity fluctuations. The pulse
derived in the limit where the Rayleigh range exceeds the sample
thickness. Here, the Rayleigh range in air is 2.4 mm and the
sample thickness is 1 mm. In ref 12, it is shown that the distance
between the peaks and valleys of the closed-aperture scan should
be about 1.7 times the Rayleigh range. That would mean a peak-

Sulfur-Rich Compounds
J. Phys. Chem. A, Vol. 103, No. 35, 1999 6933
-
11
-1
5
27 nm was R2 ) 1.4 × 10 m W . The dashed curve in the
left-hand graph of Figure 2 is the theoretical fit. The difference
between the data and the fitted curve is attributed to the
somewhat too-simple treatment of the self-focusing, which in
a full treatment should be considered to vary as the cell moves
through the beam waist as in the beam propagation methods
discussed in ref 24. In the closed-aperture data, shown in Figure
3
, a dip for Z < 0 and a peak for Z > 0 are observed. This
behavior is indicative of a nonlinear lens with a positive value
of n . When the data were fit to the closed-aperture transmis-
I
2
sion function (dashed curve in Figure 3) using the above values
I
2
I
2
of R2 and the Rayleigh range, while varying n , the value n )
.3 × 10
-
19
2
-1
8
m W was obtained (see Table 1).
When the measurements of pure DMF were completed, a
solution of (Me4N)4[Zn10S4(SPh)16] (0.809 mM) in DMF was
studied. The solution had no detectable absorbance at 527 nm,
so R0 ) 0. The open-aperture data (Figure 2) are indistinguish-
able from pure DMF, indicating that R2 for the solute alone
was essentially zero. In the closed-aperture configuration in
solution (Figure 3), there is a peak at Z < 0 and a dip at Z >
Figure 2. Open-aperture data for pure dimethylformamide (DMF) and
a DMF solution of (Me
theoretical fits using the thin-sample functions of Sheik-Bahae et al.
ref 12).
4 4 4
N) [Zn10S (SPh)16]. The dashed curves are
(
to-valley distance of about 4 mm in our experimental data. If
the peak-to-valley distance is significantly different, there would
be two likely possibilities. If the sample were too thick, the
Z-scan data would take on a distinctive shape. There would be
0
. These features indicate that the nonlinear refraction of the
solution is greater than and opposite in sign from that of the
I
2
solvent. The value of n could then be varied to fit the solution
2
3
data, as shown in Figure 3.
a characteristic flat region around the focus at Z ) 0, and the
peak-to-valley separation would become greater than predicted
by the thin-sample model, by an amount that increases with
increasing sample thickness (for example, see Figure 2 of ref
To obtain the properties for the solute alone, we use eq 3.
I
2
(3)
We could convert the above R2 and n measurements into ø
and plug them into eq 3. However, since the theoretical curves
do not exactly fit the pure solvent data, we used another
procedure that we believe is more accurate and not dependent
on how the pure solvent data are interpreted. We simply
subtracted the closed-aperture and open-aperture data for the
pure solvent from the solution data. We then fit the difference
2
3). The other possibility is self-focusing. As discussed in detail
24
by Hughes et al., significant self-focusing effects on Z-scan
data have been previously observed in thin samples with pulses
(15 ps duration, 532 nm) similar to those used here. If self-
focusing were to occur in the sample, the effective Rayleigh
range would be decreased. The peak-to-valley distance would
then become smaller than predicted by the thin-sample model.
I
data as above to obtain the solute’s contribution to R and n .
2
2
I
-18
2
-1
We obtained the value n2 ) -1.3 × 10
m
W
for
12
In that case, following a suggestion from Sheikh-Bahae et al.,
(
Me4N)4[Zn10S4(SPh)16]. We then computed molecular hyper-
an approximate correction for self-focusing can be obtained by
allowing the Rayleigh range to vary. In general, our data do
not show the characteristic flat region indicating a thick sample,
but they do show peak-to-valley separations that are somewhat
less than predicted by the thin-sample treatment. That indicates
significant effects from self-focusing, so we allowed the
Rayleigh range to vary in fitting the data. To verify that no
significant errors were introduced by using the thin-sample
approximation, some of the data were fit using the full thick-
sample treatment,25 which produced results that were always
identical to those obtained with the thin-sample approximation.
polarizabilities γR and γI given in Table 2 using eqs 4 and 5
and figures of merit W and T using eqs 6 and 7.
-1
We also measured the NLO properties of several common
solvents, which are given in Table 1. Table 2 presents results
for 12 sulfur-rich molecules (plus CS2). For the n-butylferrocene
persulfide polymer, the mole fraction f in eq 3 was calculated
as the mole fraction of the monomer. Table 3 lists the figures
-1
of merit W/I0 and T for the sulfur-rich molecules examined
in this work.
Discussion
Results
The structures of some of the sulfur-rich molecule are shown
in Figure 4. They fall into five categories: (1) molecular sulfur,
We first discuss in detail how measurements are made and
interpreted for a specific example, the sulfur-rich compound
2
-
(2) derivatives of the isomers of C3S5 , (3) a series of C3S5-
derived dithiocarbonates, (4) ferrocenes, and (5) metal sulfido
clusters. All the solutions were either colorless or yellow. The
yellow solutions had some linear absorption at 527 nm, which
resulted in a depressed figure-of-merit W/I0 (Table 3).
(
Me4N)4[Zn10S4(SPh)16], and then cite results for other com-
pounds in Tables 1-3. We chose N,N-dimethylformamide
DMF) as the solvent for (Me4N)4[Zn10S4(SPh)16]. First, mea-
(
surements were made on pure DMF, which, like the other
solvents discussed in this work, is colorless, i.e., R0 ) 0. The
data for pure DMF are shown in Figures 2 and 3.
Solvents. Both CS2 and toluene have been studied extensively
and are used as standards to test the accuracy of our absolute
5
,12,23
I
2
-18
2
measurements.
W
For CS2, we found n ) 3.0 × 10
m
In Figure 2, there is a dip in the signal near the focus at Z )
where the intensity is at a maximum. This dip indicates
-
1
at 527 nm and 27 ps. Converting into units used by other
0
authors, we obtain a nonlinear refractive index n2 ) 1.16 ×
reduced transmission of 527 nm pulses near the focus, caused
by nonlinear absorption. The value of the nonlinear absorption
coefficient was determined as discussed above, by fitting the
data to the open-aperture transmission function of Sheik-Bahae
-
11
I
2
10
esu (the conversion is n2 ) n0cn /(40π)). Other authors
-
11
working at λ ) 532 nm and 30 ps report n2 ) 1.2 × 10
1
2
-11
26
esu and 1.05 × 10 esu, in good agreement with our results.
1
2
I
2
-19
2
-1
et al., with the Rayleigh range and the nonlinear absorption
For toluene, we found n ) 9.5 × 10
m W at 527 nm
-
12
coefficient R2 as free parameters. The result for pure DMF at
and 27 ps, which converts to n2 ) 3.3 × 10
esu. Other

6934 J. Phys. Chem. A, Vol. 103, No. 35, 1999
Breitzer et al.
TABLE 1: Nonlinear Properties at λ ) 527 nm of Solvents Used in This Work
I
(3)
R
(3)
I
n
ø
γ
R
-57
R
2
-12
ø
γ
I
2
-
18
-33
-33
-57
concn
(M)
N
(10 m2
(10
C
)
(10
C
)
(10
m
)
(10
C
)
(10
C
4
-3
W-1
m V-3
4
-3
solvent
CS
toluene
(1024 m-3
)
n
0
W-1
)
m V-3
m V
m V )
2
16.63
9.39
15.60
19.15
12.91
10015
5650
9390
11530
7774
1.627
3.0 ( 0.1
370
98
-43
-23
80
0.0065
0.0043
-0.0014
-0.00077
0.0031
0.8
2.7
0.2
0.3
84
240
16
21
1100
0.0015
0.0106
0.00052
0.00071
0.044
1.4960
1.4240
1.3440
1.4310
0.93
a
CH
CH
2
Cl
CN
2
-0.44
3
-0.27
DMF
0.83
14
a
Value obtained with lower intensity (1.3 µJ pulses). With higher intensity pulses, a positive component to the nonlinear refraction was also
observed.
TABLE 2: Nonlinear Optical Properties at λ ) 527 nm of Sulfur-Rich Molecules
I
2
(3)
R
-33
(
3)
n
ø
(10
γR
R2
ø
γI
I
-
18
-57
-12
-33
-57
-3
concn
(mM)
N
(10 m2
C
)
(10
C
(10
m
(10
(10
C
R0
-1
-3
4
-3
-1)
C m V-3)
4
molecular species
CS2
solvent
(1024 m-3
)
n0
W-1
)
m V
m V
)
W
m V
)
(m )
neat
16630
21.0
7.01
7.01
10.0
10015
12.6
1.627
1.4960
3.0 ( 0.1
370
0.0065 0.8
84
18
630
320
40
0.0015
0.35
46
23
2
30
0
0
10.9
76.5
2
5.6
4.6
18
201
44
1.3
S8
toluene
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
CH3CN
CH2Cl2
CH3CN
-0.25 ( 0.02
-0.15 ( 0.02
-0.93 ( 0.02
-26
-14
-89
-0.52
-1.04
-6.5
0.2 ( 0.02
7.9 ( 0.2
4.0 ( 0.2
0.5 ( 0.2
0.8 ( 0.2
0.0 ( 0.2
2.1 ( 0.2
18.2 ( 0.5
0.0 ( 0.2
0.0 ( 0.2
0.8 ( 0.2
0.0 ( 0.2
R-C3S3(SCH2Ph)2
â-C3S3(SCH2Ph)2
C3S7O
C6S6O2
C6S8O2
4.22 1.4240
4.22 1.4240
6.02 1.4240 -(<0.1)
0.656 1.4240
1.132 1.4240
6.80 1.4240
6.781 1.4240
3.30 1.3440
-(<10) -(<0.5)
1.09
1.88
11.3
11.26
5.48
4.24
-1.36 ( 0.1
-130
-100
-75
340
-60.8
-27.2
-3.4
15.6
64
-1.05 ( 0.1
-0.79
((<16)
((<4)
BuFc
170
7.6
(BuFcS2)n
3.6 ( 0.2
1460
((<15)
((<16)
57
66
((<2)
((<2)
(Me4N)2[Cu4(SPh)6]
-0.13 ( 0.02
-11
-1.3
ZnS6(TMEDA)
2.55 1.4240 -(<0.1)
3.04 1.3440
0.487 1.4310
-(<10) -(<1)
(Me4N)2[Zn4(SPh)10]
5.04
0.809
0.38 ( 0.02
-1.3 ( 0.1
32
-130
4.1
-78
7.2
0
0
(Me4N)4[Zn10S4(SPh)16] DMF
((<16) ((<10)
TABLE 3: Figures of Merit at 527 nm for Sulfur-Rich
Molecules
15
2
-1
-1
molecular species
|W|/I
0
10- m W
|T|
3.6
>1.2
0.02
0.22
<0.2
1.6
CS
2
large
large
26
S
8
3
S
3
(SCH
(SCH
2
Ph)
Ph)
2
â-C
S
3
2
2
23
C
3
C
6
C
6
O
O
O
<100
460
430
34
2
2
>5
(
BuFcS )
2 n
0.19
0.36
>0.6
BuFc
83
6
(
Me
4
6
4
4
N)
2
[Cu
4
(SPh)
6
]
ZnS
Me
Me
(TMEDA)
<150
large
large
very large
0.5
(
(
N)
N)
2
[Zn
4
(SPh)10
]
4
[Zn10 (SPh)16
S
4
]
>6.2
authors have reported 6 × 10-12 esu at 532 nm and 25 ps, which
is within a factor of 2 of our measurement.24 Some of this
difference may be attributed to experimental errors and some
to impurities in the toluene.
Figure 3. Closed-aperture data for pure dimethylformamide (DMF)
and a solution of (Me N) (SPh)16]. The dashed curves are
theoretical fits using the thin-sample functions of Sheik-Bahae et al.
ref 12).
4
4 4
[Zn10S
(
Some interesting results were obtained with CH2Cl2. The
nonlinear parameters were intensity-dependent. At lower inten-
sity, the nonlinear refraction was relatively small and it had a
negative sign, which is unusual for a colorless solvent. At higher
intensity, a larger positive nonlinearity was observed. The values
for CH2Cl2 cited in Table 1 were obtained with lower intensity
pulses. A negative nonlinear refractive index is usually associ-
ated with the presence of nearby electronic states. For example,
CH2Cl2 has some low-lying electronic states near 450 nm. We
can speculate that the negative nonlinear refraction might be
associated with the tail of this transition or with a transition
involving two photons at 527 nm or possibly due to stabilizers
or impurities in the solvent. Acetonitrile also showed negative
nonlinear refraction, which was very small. Since acetonitrile’s
electronic excited states are deep in the ultraviolet, this small
negative nonlinearity might well be due to impurities.
our determination of nonlinear parameters for the sulfur-
containing molecules because in our method we simply subtract
8,12
the neat solvent data from the dilute solution data. However,
two points are worth mentioning about this method. First, the
results may be solvent-dependent; this effect would be quite
severe to the extent that the degree of aggregation, solvation,
or protonation is affected. Such complications are negligible
for our samples, the solution properties of which are normal.
Second, nonlinear properties of solutes are best obtained when
the solvent contributes less to the nonlinear refraction and
absorption than does the solute. As shown in Table 1, acetonitrile
and dichloromethane are the best solvents in this regard and
CS2 is the worst.
Whereas our approach is the simplest way of screening many
different materials, once an interesting material is found, more
extensive studies should be performed. These might include
concentration dependence, wavelength dependence, and pulse-
width dependence. Concentration dependence could be espe-
It should be emphasized that the determination of nonlinear
parameters for the neat solvents themselves has no effect on

Sulfur-Rich Compounds
J. Phys. Chem. A, Vol. 103, No. 35, 1999 6935
investigated the nonlinear refractive properties of the core
2
-
heterocycle. Two isomers of C3S5 are known. The great
majority of previous studies has focused on the R isomer (see
Scheme 1 and Figure 4), also known as dmit. A second,
thermodynamically more stable â isomer of C3S52 is also
known, wherein the three carbon atoms are adjacent. For this
study we prepared the dibenzyl derivative of both isomers of
-
2
-
C3S5 , i.e., R- and â-C3S3(SCH2Ph)2. These neutral dibenzyl
compounds have high solubility in nonpolar solvents.
Both the R and â isomers of C3S3(SCH2Ph)2 have absorptions
at 527 nm, with linear absorption coefficients of R0 ) 10.9 and
-1
7
6.5 m , respectively. (Note that these linear absorption values
are concentration-dependent and are reported for the concen-
trations used in the z-scan measurements; the extinction coef-
-
1
-1
ficients for the R and â isomers are 15.5 and 109 M cm ,
respectively.) The open-aperture measurements show that both
species exhibit substantial nonlinear absorption. The nonlinear
refraction for the â isomer is greater than that of the R isomer.
It is difficult to rationalize the differing hyperpolarizability
values for the C3S5 ring compounds except to note that the â
isomer is a better Bronsted base; this might suggest that it is
more polarizable. To some extent their linear absorption
coefficients precludes further interest on our part. However, the
R isomer is a useful reference material because the core is easily
functionalized at the terminal thiolato groups (Scheme 1); this
versatility might make it possible to optimize or fine-tune optical
and processing properties. Additionally, as discussed in the next
section, some of its derivatives may exhibit both substantial
second hyperpolarizabilities as well as low linear absorption
coefficients.
Figure 4. Structures of some of the sulfur-rich molecules studied in
this work.
SCHEME 1
Dithiocarbonates. We evaluated a recently synthesized
1
5,21
series
isomer of C3S5. These compounds exhibit extended structures
of dithiocarbonates, which are derived from the R
(bicyclic or tricyclic) that should enhance their polarizabilities
(
and therefore their γR values). Three dithiocarbonates were
examined: the bicyclic species C3S7O, which contains a
polysulfur chain-linked to the dithiocarbonate, and two tricyclic
species C6S6O2 and C6S8O2, which feature two dithiocarbonates
attached to C4S2 and C4S4 cores, respectively (see Figure 4).
The R isomer of C3S5 is a trithiocarbonate (i.e., bearing the
functionality (RS)2CdS); we made the dithiocarbonate by
substituting an oxygen atom for the terminal sulfur. This was
done because the tricyclic trithiocarbonates, C6S8 and C6S10,
are nearly insoluble; C6S8 is only slightly soluble in CH2Cl2,
and C6S10 is soluble (slightly) only in CS2. Additionally, it is
known that carbonyls have a larger HOMO-LUMO gap than
thiocarbonyls; this could be expected to result in a lower linear
absorption in the visible spectrum. Therefore, the dithiocarbon-
ates could exhibit a larger nonlinear refraction due to the
increased conjugation, without a correspondingly larger linear
absorbance.
cially useful because it might provide insights into how the
solute would perform in the solid state. Furthermore, at high
solute concentrations, solute-solute interactions could slow
molecular reorientation. Of course, in the limiting case of a solid,
molecular reorientation is totally suppressed and the intrinsic
electronic and nuclear hyperpolarizabilities dominate.
Sulfur. Given that this project focused on sulfur-rich
molecules, it seemed worthwhile to study sulfur itself. Elemental
sulfur exists in a number of allotropes, the most common being
S8. It is known, however, that in solution, S8 exists in
equilibrium with small amounts of other cyclic species, such
30
2
7
as S7. Toluene solutions of S8 were found to have no
measurable linear or nonlinear absorption at 527 nm. The γR
value of S8 is about 80 times larger than CS2 and opposite in
sign. Thus, on a molecule-to-molecule basis, S8 is superior to
CS2 in its nonlinear properties. Of course, we are ultimately
comparing pure liquid CS2, where one has an extremely large
number density of small molecules that rotate quickly, to a dilute
solution of S8. In future work, it would be interesting to compare
the bulk nonlinear parameters of liquid CS2 to crystalline S8.
Consistent with the expected influence of the oxygen vs the
sulfur on the energy of the associated n to π* transition, C3S7O
-
1
did indeed exhibit a diminished extinction coefficient (2 M
-
1
-1
cm , with an absorption coefficient of 2 m at the concentra-
tion used) compared to the C3S3(SCH2Ph)2 isomers. On the other
hand, the two tricyclic molecules, C6S6O2 and C6S8O2, had
Derivatives of C3S52 . Anions of the formula C3S5 have
attracted much attention in the materials science community
because they are widely employed as precursors to two major
classes of electronic materials, metal dithiolenes and tetrathia-
fulvalenes (Scheme 1).28 Recent work has shown that transition
-
2-
-1
-1
extinction coefficients at 527 nm of 51 and 24.5 M cm ,
which were between those of the R and â isomers of C3S3-
(SCH2Ph)2. (The small linear absorption coefficients, 5.6 and
-1
4.6 m , respectively, are due to the lower concentrations used).
However, C6S6O2 and C6S8O2 have much larger γR values
-
56
-56
4
-3
metal derivatives of C3S5 exhibit high nonlinear refraction as
(-6.08 × 10
and -2.72 × 10
C m V , respectively)
well.1
0,29
Given the intense interest in these materials, we
than the C3S3(SCH2Ph)2 isomers. The nonlinear absorption

6936 J. Phys. Chem. A, Vol. 103, No. 35, 1999
Breitzer et al.
CHART 1
Figure 5. Structure of the [Zn10S
4
(SPh)16]4 anion, taken from ref 17.
-
coefficients are modest for two of these species, C3S7O and
C6S8O2. For C6S6O2, however, γΙ is large and comparable to
those seen for the C3S5 derivatives; the corresponding figure of
merit, T , reflects this deficiency. It is also interesting to
compare the effect of cumulated S-S bonds, cf. S8, vs the effect
of cumulated C2S2CO rings; the monodithiocarbonate displays
only a modest nonlinear refraction.
Starting with ZnS (TMEDA), the linear absorption, which is
6
weak, is attributed to a ligand (sulfido) to metal charge transfer
-
1
band centered in the UV. This species displays only a modest
γR, reminiscent of the results for other polysulfide molecules,
such as C S O and S (see above). On the other hand, the
3
7
8
tetrazinc species (Me N) [Zn (SPh) ] has a sizable γ value.
4
2
4
10
R
Ferrocenes. Ferrocenes have been examined as candidates
for photonic materials because of the discovery of high first
hyperpolarizabilities for ferrocene-substituted olefins by Marder,
Despite a modest linear absorption, the encouraging nonlinear
refraction data were offset somewhat by nonlinear absorption
2-
effects. We now turn to the copper cluster Cu4(SPh)6 , which
31
33
Green, and their co-workers. Ferrocenes are effective donors
Shi and co-workers have reported to exhibit an “extraordinary”
(2)
22
in push-pull ø molecules. In the present context, ferrocenes
are of interest in view of the recent synthesis of ferrocene
persulfide (S-S) polymer (Figure 4). Electrochemical measure-
ments indicate that the subunits in these polymers are strongly
second hyperpolarizability. As implied by Dance et al., this
Cu4 cluster is chemically labile in solution; bright-yellow crystals
2
-
of the pentacopper cluster, Cu5(SPh)7 , often form along with
2
-
(or instead of) Cu4(SPh)6 , which crystallizes in the form of
pale-yellow flakes (see Experimental Section). In view of this
complication, it is likely that both this and Shi et al.’s studies
were conducted on a mixture of the Cu4 and Cu5 clusters. We
confirmed, however, that these samples suffer a substantial linear
absorption.
3
2
33
coupled. This coupling implies a delocalized ground state,
hence, the potential for a large value of γR.
Our results show that the sulfur-linked ferrocenes indeed
exhibit higher nonlinear refraction than ferrocene itself. The
polymers, however, suffer from substantial linear and nonlinear
absorption. In terms of the figures of merit in Table 3, the
absorption effects offset the benefits of the enhanced γR.
Studies on Zn10S4(SPh)164-. We derived two conclusions
based on the experiments described in the previous section: (i)
these metal clusters exhibit higher optical nonlinearities than
mononuclear complexes and (ii) the undesirable absorption
characteristics of the Cu4 cluster were not exhibited by the Zn4
analogue. These results encouraged us to extend the series to
Small Metal Sulfido Ensembles. Recent work has shown
that metal sulfido clusters display substantial nonlinear refraction
34
and nonlinear absorption. It appears that most such systems
have substantial linear absorption properties,35 and we were
4-17,37
include the still larger cluster [Zn10S4(SPh)16],
the structure
interested in examining species without this detracting feature.
A series of three species were selected for this part of the study
of which is depicted in Figure 5. This colorless anion was
studied as its tetramethylammonium salt, which confers solubil-
ity in DMF. Like the Zn4 cluster, the Zn10 compound exhibits
no linear absorption. Furthermore, it displays virtually no
nonlinear absorption. Its second hyperpolarizability is the
greatest of the series of compounds studied in this work. The
favorable juxtaposition of absorption and refraction properties
is reflected in the exceptional figures of merit (Table 3).
(
Chart 1). As a reference compound, we started with the simple
2-
complex ZnS6(TMEDA). In this compound the S6 chain is
bound at its termini to the tetrahedral Zn2+ center, whose other
two coordination sites are occupied by tetramethylethylenedi-
3
6
amine (TMEDA). We examined two thiolato clusters,
Zn4(SPh)102- and Cu4(SPh)62-. Despite their structural complex-
ity, these species are easily synthesized by high-yielding
reactions from simple reagents. Both clusters feature tetrahedral
Conclusion
arrays of d10 metal centers (Cu , Zn ). The zinc species has
both bridging and terminal thiolato ligands, whereas in the
copper clusters, the thiolates are bridging. The copper thiolato
cluster had been previously examined where it was claimed to
exhibit one of the highest values of γR for a transition metal
complex. We were interested in reconfirming this observation
at shorter pulse widths and comparing these results with the
Zn4 species. The Zn4 cluster is very robust and, relative to the
Cu4 species, has improved linear optical properties relative to
the Cu4 species.
+
2+
In this work we have examined the nonlinear absorption and
nonlinear refraction of solutions of a series of sulfur-rich
molecules and anions. Molecular properties γR and γΙ were
extracted by subtracting the contribution from the solvent. This
study represents a quantitative evaluation of the qualitative idea
that species containing large numbers of the highly polarizable
sulfur centers would exhibit large nonlinear refraction in the
absence of significant linear or nonlinear absorption. This
hypothesis is confirmed. The accumulation of a large number

Sulfur-Rich Compounds
J. Phys. Chem. A, Vol. 103, No. 35, 1999 6937
of S atoms does not, by itself, give rise to optimal properties;
rather, the electronic polarizability correlates with the number
(12) Sheik-Bahae, M.; Said, A. A.; Wei, T.-H.; Hagan, D. J.; Van
Stryland, E. W. IEEE J. Quantum Electron. 1990, 26, 760. Values given
here are in SI units. To convert to esu, the following conversion is used:
2-
of sulfido (S ) centers. Our survey identified two structural
families with particularly promising optical properties, the Zn-
S-SPh clusters and the condensed dithiocarbonates. In both
cases the greatest figures of merit were obtained for the largest
molecules. Further synthetic work is planned to develop related
structures with still more extended structures.
Previous work on the second hyperpolarizability of sulfur-
rich compounds examined only species with substantial linear
absorption coefficients. The present work shows that high figures
of merit can be achieved without the undesirable effects of linear
absorption.
4 2
(
3 × 10 )
(
3)
(3)
SI
ø
)
ø
esu
4
πꢀ
0
(
13) Winter, C. S.; Oliver, S. N.; Manning, R. J.; Rush, J. D.; Hill, C.
A. S.; Underhill, A. E. J. Mater. Chem. 1992, 2, 443.
14) W and T are used in the literature. We wanted to define figures of
(
merit whose terminology was consistent with this prior use, yet express
them in such a way as to be intensity-independent and large when favorable.
(
15) Galloway, C. P.; Doxsee, D. D.; Fenske, D.; Rauchfuss, T. B.;
Wilson, S. R.; Yang, X. Inorg. Chem. 1994, 33, 4537.
(16) Yu, L.; Zhu, D. Phosphorus, Sulfur Silicon Relat. Elem. 1996, 116,
25.
2
(
17) Dance, I. G.; Choy, A.; Scudder, M. L. J. Am. Chem. Soc. 1984,
Acknowledgment. This work was supported by grants from
the Department of Energy, DOE DEFG (T.B.R.) through the
Seitz Materials Research Laboratory, the National Science
Foundation, DMR 97-14843 (D.D.D.), and a contract from the
Air Force Office of Scientific Research, F4960-97-1-0056
1
06, 6285.
(
18) Compton, D. L.; Brandt, P. F.; Rauchfuss, T. B.; Rosenbaum, D.
F.; Zukoski, C. F. Chem. Mater. 1995, 7, 2342.
19) Steimecke, G.; Sieler, H. J.; Kirmse, R.; Hoyer, E. Phosphorus
Sulfur 1979, 7, 49.
20) Steimecke, G.; Sieler, H.-J.; Kirmse, R.; Dietzsch, W.; Hoyer, E.
(
(
(D.D.D.). L.K.I. acknowledges support from the AASERT
Phosphorus Sulfur Relat. Elem. 1982, 12, 237.
(21) Yang, X.; Rauchfuss, T. B.; Wilson, S. R. J. Chem. Soc., Chem.
Commun. 1990, 34.
program, sponsored by the U.S. Army Research Office, Contract
DAAG55-98-1-0191. We thank one referee for insightful
comments and criticisms.
(22) Dance, I. G. Aust. J. Chem. 1978, 31, 2195.
(
23) Tian, J.-G.; Zang, W. P.; Zhang, C. Z.; Zhang, G. Appl. Opt. 1995,
2
1, 4331.
(24) Hughes, S.; Burzler, J. M.; Spruce, G.; Wherrett, B. S. J. Opt. Soc.
References and Notes
Am. B 1995, 12, 1888.
(
1) Nonlinear Optics of Organic Molecules and Polymers; Narwa, N.
(25) Handbook of Nonlinear Optics; Sutherland, R. L., Ed.; Marcel
Dekker: New York, 1996.
S., Miyata, S., Eds.; CRC Press: Tokyo, 1996.
(
2) Clymer, B.; Collins, S. A. Opt. Eng. 1985, 24, 74.
(26) Zhao, M.-T.; Singh, B. P.; Prasad, P. N. J. Chem. Phys. 1988, 89,
5535.
(3) Mizrahi, V.; DeLong, K. W.; Stegemann, G. I. Opt. Lett. 1990,
1
4, 1140.
4) Natarajan, L. V.; Sutherland, R. L.; Tondiglia, V. P.; Bunning, T.
(27) Tebbe, F. N.; Wasserman, E.; Peet, W. G.; Vatvars, A.; Mayman,
A. C. J. Am. Chem. Soc. 1982, 104, 4971. Steuel, R. Top. Curr. Chem.
1982, 102, 149.
(
J.; Adams, W. W. J. Nonlinear Opt. Phys. Mater. 1996, 5, 89-98. (b) Tang,
N.; Su, W.; Cooper, T.; Adams, W.; Brandelik, D.; Brant, M.; McLean,
D.; Sutherland, R. SPIE 1996, 2853, 149.
(28) Cassoux, P.; Valade, L.; Kobayashi, H.; Kobayashi, A.; Clark, R.
A.; Underhill, A. E. Coord. Chem. ReV. 1991, 110, 115.
(29) Zuo, J.-L.; Yao, T.-M.; You, X.-Z.; Fun, H.-K.; Yip, B.-C. J. Mater.
Chem. 1996, 6, 1633. Papadopoulos, M. G.; Waite, J.; Winter, C. S.; Oliver,
S. N. Inorg. Chem. 1993, 32, 277 and references therein.
(30) Scheithauer, S.; Mayer, R. Thio- and Dithiocarboxylic Acids and
Their DeriVatiVes in Topics in Sulfur Chemistry; Senning, A., Ed.;
Thieme: Stuttgart, 1979; pp 2-4.
(31) Long, N. J. Angew. Chem., Int. Ed. Engl. 1995, 34, 826.
(32) Compton, D. L.; Brandt, P. F.; Rauchfuss, T. B.; Rosenbaum, D.
F.; Zukoski, C. F. Chem. Mater. 1995, 7, 2342.
(33) Manners, I. Angew. Chem., Int. Ed. Engl. 1996, 35, 1602.
(34) Hou, H.-W.; Xin, X.-Q.; Shi, S. Coord. Chem. ReV. 1996, 153, 25.
(35) Shi, S.; Zhang, X.; Shi, X. F. J. Phys. Chem. 1995, 99, 14911.
(36) Verma, A. K.; Rauchfuss, T. B.; Wilson, S. R. Inorg. Chem. 1995,
34, 3072.
(
5) Nonlinear Optics; Boyd, R. W., Ed.; Academic Press: New York,
992.
6) Solid State Chemistry, An Introduction; Smart, L., Moore, E., Eds.;
Chapman and Hall: London, 1992.
7) For leading references, see the following. Kobayashi, T. Pure Appl.
Chem. 1995, 67, 387-400.
8) Hein, J.; Bergner, H.; Lenzner, M.; Rentsch, S. Chem. Phys. 1994,
79, 543. (b) Thienpont, H.; Rikken, G. L. J. A.; Meijer, E. W.; ten Hoeve,
W.; Wynberg, H. Phys. ReV. Lett. 1990, 65, 2141.
9) Marder, S. R.; Torruellas, W. E.; Blanchard-Desce, M.; Ricci, V.;
1
(
(
(
1
(
Stegeman, G. I.; Gilmour, S.; Br e´ das, J.-L.; Li, J.; Bublitz, G. U.; Boxer,
S. G. Science 1997, 276, 1233.
(
10) Whittall, I. R.; Humphrey, M. G.; Samoc, M.; Swiatkienicz, J.;
Luther-Davies, B. Organometallics 1995, 14, 5493.
11) Draganjac, M. E.; Rauchfuss, T. B. Angew. Chem., Int. Ed. Engl.
985, 24, 742.
(
(37) For an overview of the area, see the following. Dance, I.; Fisher,
K. Prog. Inorg. Chem. 1994, 41, 637.
1