Synthesis, light harvesting, and energy transfer properties of a Zinc porphyrin linked poly(phenyleneethynylene)

Article abstract of DOI:10.1021/ma048489u

The synthesis of a zinc porphyrin linked conjugated homopolymer is presented. The synthetic strategy employed a directional polymerization of 1-iodo-2,5-dioctyl-4'-ethynyltolane in the presence of [10,20-bis(3,5-bis-tert- butylphenyl]-5,15-diethynylporphinato]zinc(II) (12) or [10,20-bis(3,5-bis-tert- butylphenyl]-5,15-dibromoporphinato]zinc(II) (10). In the case of 12 the reaction was found to give a polymer product with a constant ratio of zinc porphyrin to incorporated monomer units regardless of the molecular weight. In the case of 10, however, the ratio of zinc porphyrin to incorporated monomer units decreased with increasing molecular weight of the polymer product. The polymer products were separated into fractions using preparative SEC. Studies of the energy transfer properties from the polymer to the zinc porphyrin revealed that there was little energy transfer in solution but quantitative energy transfer in the solid state. The polymer products proved difficult to handle in terms of device construction where a thin polymer film is sandwiched between two electrodes. We found this to be a consequence of the palladium catalysis employed during the polymerization. We finally discuss the issues concerning the problems with remnants from the Pd catalyst in the polymer samples and their removal.

Full text of DOI:10.1021/ma048489u

1180
Macromolecules 2005, 38, 1180-1189
Synthesis, Light Harvesting, and Energy Transfer Properties of a Zinc
Porphyrin Linked Poly(phenyleneethynylene)
Kim T. Nielsen, Holger Spanggaard, and Frederik C. Krebs*
The Danish Polymer Centre, RISØ National Laboratory, P.O. Box 49, DK-4000 Roskilde, Denmark
Received July 23, 2004; Revised Manuscript Received November 15, 2004
ABSTRACT: The synthesis of a zinc porphyrin linked conjugated homopolymer is presented. The synthetic
strategy employed a directional polymerization of 1-iodo-2,5-dioctyl-4′-ethynyltolane in the presence of
[10,20-bis(3,5-bis-tert-butylphenyl]-5,15-diethynylporphinato]zinc(II) (12) or [10,20-bis(3,5-bis-tert-bu-
tylphenyl]-5,15-dibromoporphinato]zinc(II) (10). In the case of 12 the reaction was found to give a polymer
product with a constant ratio of zinc porphyrin to incorporated monomer units regardless of the molecular
weight. In the case of 10, however, the ratio of zinc porphyrin to incorporated monomer units decreased
with increasing molecular weight of the polymer product. The polymer products were separated into
fractions using preparative SEC. Studies of the energy transfer properties from the polymer to the zinc
porphyrin revealed that there was little energy transfer in solution but quantitative energy transfer in
the solid state. The polymer products proved difficult to handle in terms of device construction where a
thin polymer film is sandwiched between two electrodes. We found this to be a consequence of the
palladium catalysis employed during the polymerization. We finally discuss the issues concerning the
problems with remnants from the Pd catalyst in the polymer samples and their removal.
Introduction
dioctyltolanyl)ethylene] (Nn), on the NPN structure with
a broad distribution and on fractions with a narrow
distribution prepared by preparative size exclusion
chromatography (SEC). Finally, we discuss the problems
concerning the use of Pd as a catalyst for the polymer-
ization due to the catalyst remnants contained in the
polymer product and provide a solution to this problem
in the context of this work.
Since the discovery in 1974 of doped polyacetylene
showing electrical conductivity,1 there has been a large
interest in conjugated polymers and their application
for e.g. light-emitting diodes,2 polymer electronic cir-
cuits,3 and polymer solar cells.4
Recently, the synthesis of a three-domain (JPJ) and
a five-domain (PF4) structure were reported.5,6 The
three-domain structure consisted of two homopolymer
blocks (J domains) with a porphyrin dye molecule (P
domain) in the middle. In this three-domain structure,
the J domains were poly(2′,5′-dioctyl-4,4′′-terphen-
ylenecyanovinylene), and the distribution of the indi-
vidual incorporated monomer units was a broad distri-
bution. The P domain was a 5,10,15,20-tetraphenylpor-
phyrin derivative, and there was only one porphyrin
molecule in each JPJ unit. In contrast to JPJ the PF4
consisted of four oligofluorene (F domains) linked to a
porphyrin dye molecule (P domain). A very interesting
property observed for these three-domain and five-
domain structures was that both the J and the F domain
served as a light-harvesting antenna for the P domain.
We decided to investigate and evaluate the use of poly-
(phenyleneethynylene)s as the backbone of the light-
harvesting polymer chains for the porphyrin dyes and
have two alkyl groups on every second phenyl ring to
obtain solubility7 and reduce the distortion of the
backbone.8
In this paper we present a synthesis of a conjugated
homopolymer of the poly(phenyleneethynylene) type.
We discuss two different synthetic strategies toward
linking the conjugated homopolymer blocks via a zinc
porphyrin to give the three-domain structure NPN, as
shown in Figure 1.
We present studies of the optical properties of the
compounds with a focus on the band gap, the quantum
efficiency of the fluorescence, and the lifetimes. These
studies were performed on the homopolymer, poly[(2,5-
Experimental Section
1H and 13C NMR spectra were recorded on a 250 MHz
Bruker NMR spectrometer at 300 K unless otherwise stated.
All spectra were recorded in CDCl3 with TMS as an internal
reference. Compound 59 and 1,4-dioctylbenzene10 were pre-
pared according to the procedures described in the literature.
Compounds 9-12 were prepared according to the procedures
reported by Plater et al.11 with the exception that compound
11 was purified by preparative SEC before deprotection of the
TMS groups. Detailed descriptions of synthetic procedures and
analytical data for compounds 1, 2, 3, 4, 6, 7, and 8 can be
found in the Supporting Information. Electroactive device
preparation is described in detail in the Supporting Informa-
tion.
Liquid Chromatography Methods. The apparatus for the
SEC consists of a preparative system with a 3 mL injection
loop. The detector is a UV-vis diode array with a spectral
resolution of 1 nm. A 16-port switch valve allows for separation
into fractions. Two different column systems were used. They
consisted of DVB linked polystyrene gel columns comprising
a guard column and two gel columns. One system had a 100
Å and a 1000 Å column in succession, and the other had a
500 Å and a 10000 Å column in succession. The eluent was
chloroform, and the flow rate was 20 mL/min with a pressure
of 75 bar. The molecular weight was calculated on the basis
of polystyrene standards having molecular weights of 515 000,
127 000, 68 000, 36 300, 19 000, 10 900, 3250, and 1360. In
all cases the polydispersity of the standards was smaller than
1.007.
Photophysical Methods. The UV-vis solution spectra
were recorded in dry CHCl3 degassed with argon. Emission
spectra and lifetimes were recorded in degassed CHCl3. The
absorption maximum was kept below 0.07. All emission
spectra are corrected. Quantum yields where measured rela-
* Corresponding author: e-mail frederik.krebs@risoe.dk.
10.1021/ma048489u CCC: $30.25 © 2005 American Chemical Society
Published on Web 01/25/2005

Macromolecules, Vol. 38, No. 4, 2005
Zn Porphyrin Linked Poly(phenyleneethynylene) 1181
Figure 1. Three-domain structure of the NPN dye linked homopolymer.
tive to DPA.12 The UV-vis solid-state spectra were recorded
on spin-coated films of the polymers.
dissolved in 50/30 mL dry triethylamine/THF. The solution
was flushed with argon for 30 min before the catalyst Pd-
(PPh3)4 (50 mg, ≈2 mol %) and the cocatalyst CuI (40 mg, ≈10
mol %) were added under an argon atmosphere. After the
addition of the catalyst the mixture was refluxed overnight
and was poured into 300 mL of methanol. The following day a
green precipitate formed that was filtered on a glass filter-
funnel. The filter cake was washed with both cold methanol
and cold diethyl ether and was dried in the air to give 1.021 g
Poly[(2,5-dioctyltolanyl)ethynylene] (Nn). A 250 mL
round-bottom flask was dried using a heat gun and flushed
with argon. Compound 8 (1.36 mmol, 0.750 g) was placed in
the flask and dissolved in 40 mL of piperidine. The solution
was flushed with argon for 30 min before the catalyst Pd-
(PPh3)4 (30 mg, ≈2 mol %) and the cocatalyst CuI (25 mg, ≈10
mol %) were added under an argon atmosphere. After the
addition of the catalyst the mixture was stirred under an argon
atmosphere overnight at room temperature. A yellow precipi-
tate quickly formed. The following day the reaction mixture
had solidified. Dry THF (50 mL) was added, and the solution
was heated to reflux temperature. The precipitate dissolved,
and the solution was stirred for another 5 h at reflux. The
solution was then poured into 300 mL of methanol, and a
yellow precipitate formed. Filtration on a glass filter gave a
yellow solid that was washed with both cold methanol and cold
diethyl ether and dried on the filter. The wet product was kept
for further use. Attempts to dry the product in a vacuum oven
at RT gave a darkened product that was no longer soluble in
common solvents. This behavior has been observed earlier.15
The total yield based on a completely dried sample was 0.35 g
(95%) (the wet product weighed 0.6 g containing 40% w/w
solvent and was conveniently stored in this form).1H NMR δ:
0.76-0.95 (m, 6H), 1.13-1.50 (m, 20 H), 1.50-1.83 (m, 4H),
2.70-2.90 (m, 4H), 7.31-7.43 (broad s), 7.43-7.60 (broad s)
(7.31-7.60, 6H).
Porphyrin Linked Polymer with a Constant Porphy-
rin-to-Monomer Ratio (NPN A). A 100 mL round-bottom
flask was dried using a heat gun and flushed with argon.
Compound 8 (7.24 × 10-4 mol, 0.400 g) and compound 12 (6.26
× 10-5 mol, 0.050 g) were placed in the flask and dissolved in
50/30 mL dry triethylamine/THF. The solution was flushed
with argon for 30 min before the catalyst Pd(PPh3)4 (20 mg,
≈2 mol %) and the cocatalyst CuI (15 mg, ≈10 mol %) were
added under an argon atmosphere. After the addition of the
catalyst the mixture was refluxed overnight and was next day
poured into 300 mL of methanol. The following day a green
precipitate formed that was filtered on a glass filter-funnel.
The filter cake was washed with both cold methanol and cold
diethyl ether and was dried in the air to give 0.271 g of
polymer. 1H NMR δ: 0.76-2.18 (m), 2.20 (m) (0.76-2.20, 490
H), 2.81 (broad s, 53 H), 6.80-8.18 (m, 90 H), 8.71-9.06 (m, 4
H), 9.48-9.94 (m, 4 H).
1
of polymer. H NMR δ: 0.73-1.03 (m, 190 H), 1.03-1.91 (m,
740 H), 2.15 (m, 8 H), 2.32 (m, 9 H), 2.56 (m, 13 H), 2.82 (broad
s, 76 H), 6.31-6.61 (m, 14 H), 6.79-8.16 (m, 286 H), 9.01 (m,
4 H), 9.79 (m, 4 H).
Diethylammonium Diethyldithiocarbamate (DDC).
DDC was prepared by the method described in the literature
for sodium methyltiocarbamate.16 Carbon disulfide (0.15 mol,
11.42 g) was added dropwise to a solution of diethylamine (0.4
mol, 29.25 g) in diethyl ether (350 mL), and after a while small
colorless crystals were formed. The material was collected on
a filter and washed with copious amounts of diethyl ether and
dried in a vacuum oven. The yield was 19.29 (58%); mp 81-
1
3
3
82 °C. H NMR δ: 1.24 (t, 6H, J ) 7.3 Hz), 1.37 (t, 6H, J )
3
3
7 Hz), 3.15 (q, 4H, J ) 7.3 Hz), 4.10 (q, 4H, J ) 7 Hz), 8.27
(broad s, 2H). 13C NMR δ: 12.09, 13.05, 41.70, 48.54. Anal.
Calcd for C9H22S2: C, 48.60; H, 9.97; N, 12.60; S, 28.83.
Found: C, 48.73; H, 10.17; N, 12.54; S, 28.37.
General Procedure for the Removal of Pd from Poly-
mer Samples Using DDC. The polymer was dissolved in the
minimum amount of chloroform. DDC (5 equiv relative to the
Pd employed in the catalyst) was added, and the solution was
stirred for 5 min. The solution was poured into methanol (10
time the volume of chloroform) and the polymer precipitated.
The precipitate was filtered and washed with methanol and
diethyl ether.
Results and Discussion
Monomer Synthesis. The synthesis of the monomer
unit is shown in Scheme 1. The synthesis involves the
readily available compounds 1,4-dibromo-2,5-dioctyl-
benzene (1) and 1-(4-iodophenyl)-2-trimethylsilylacety-
lene (TMSA). The first step (Scheme 1A) is a bromine-
iodine exchange of compound 1 using n-BuLi in THF
at -78 °C. The bromoiodo compound (2) was cross-
coupled with TMSA by Pd/Cu catalysis in the following
step to form the 1-(4-bromo-2,5-dioctylphenyl)-2-tri-
methylsilylacetylene (3). In this step it was found
important that the coupling reaction took place at low
temperature to increase the selectivity between iodo vs
bromo substitution. Even at 273 K we obtained ∼2% of
Porphyrin Linked Polymer with One Porphyrin Moi-
ety in Each Polymer Chain (NPN B). A 100 mL round-
bottom flask was dried using a heat gun and flushed with
argon. Compound 8 (2.19 × 10-3 mol, 1.212 g) and compound
10 (2.12 × 10-4 mol, 0.198 g) were placed in the flask and

1182 Nielsen et al.
Macromolecules, Vol. 38, No. 4, 2005
Scheme 1. Synthetic Procedure Leading to the
Monomer Unit
1-Iodo-2,5-dioctyl-4′-trimethylsilylethynyltolane (7)a
Figure 2. SEC traces of the crude product when synthesizing
11 using the Sonogashira reaction. The crude product was
separated into fractions using preparative SEC, which gave a
reasonably pure fraction of the desired product (fraction 2).
ethynyltolane (B3) as a byproduct during the synthesis
of compound 6. Furthermore, we found ∼1% of com-
pound 4. We believe this to be a result of incomplete
reaction. We also detected ∼1% of 1,4-dioctyl-2,5-bis-
((4′-trimethylsilylethynylphenyl)ethynyl)benzene (B4).
The presence of B4 is a result of the difficulty in
separating B1 from compound 3. The RF values in
n-heptane of compounds 4, 6, B3, and B4 are 0.58, 0.38,
0.25, and 0.08, so it was possible by flash chromatog-
raphy to separate 6 from the byproducts. The removal
of byproducts in this step was found to be essential in
order to control the later polymerization process. As a
consequence, compound 6 was obtained in low yield. It
should be noted that, besides the above-mentioned
byproducts, we have detected some minor byproducts,
but the amount of these was so small that we could not
determine the structures.
Porphyrin Synthesis. The porphyrin subunit should
have a structure that allowed for attachment of the
polymer blocks in the 5,15-positions. We chose the
synthetic scheme presented by Plater et al.11 due to the
possibility of large scale synthesis. The reported proce-
dure allowed for the synthesis of the porphyrin (9) (see
Scheme 4) on a 100 g scale of sufficient purity for the
synthesis of the zinc dibromoporphyrin (10).
a (a) (1) n-BuLi, THF, (2) I2; (b) TMSA, Pd(PPh3)2Cl2, CuI,
Et3N; (c) K2CO3, MeOH, CH2Cl2. Compounds B1-B4 were side
products isolated from the reaction mixtures in step b.
1,4-di(trimethylsilylethynyl)-2,5-dioctylbenzene (B1) and
2-4% of 1,4-ditrimethylsilylbutadiyne (B2) as byprod-
ucts which were detected by 1H and 13C NMR, GC-MS,
and MALDI TOF. The TMS group was removed by K2-
CO3, and the 4-bromo-2,5-dioctylphenylethynylene (4)
was obtained.17 The monomer unit 1-iodo-2,5-dioctyl-
4′-trimethylsilylethynyltolane (7) was then obtained by
an additional two-step reaction. The first step being a
Sonogashira cross-coupling reaction between compound
4 and 1-(4-iodophenyl)-2-trimethylsilylacetylene (5)
(Scheme 1B) to obtain 1-bromo-2,5-dioctyl-4′-trimeth-
ylsilylethynyltolane (6). The final step was a standard
bromine-iodine exchange.
It was possible to obtain 10 in high purity, which we
ascribe to the fortuitous low solubility of 10 making the
alleviation of byproducts relatively easy. Plater et al.
also reported the synthesis of bisethynylporphyrin (12)
through Sonogashira coupling of TMSA onto 10 followed
by removal of the TMS groups. The synthesis of 12,
however, proved very difficult in our hands, and while
a product was obtained the analytical data showed that
it consisted of oligomers and other impurities. SEC
revealed many different components as shown in Figure
2. We decided to purify the crude product by preparative
SEC and isolated the purified fraction of 11. The
1
similarity between the H NMR spectra for the crude
The Sonogashira reaction for this type of molecules
has been described by others13 and generally works
quite well. Most authors, however, never comment on
the fact that the Sonogashira alkynyl-aryl cross-
coupling cannot be performed without significant side
reactions. Known byproducts are 1,4-diarylbutadiynes,
1,4-diarylbuta-2-en-1-ynes, and the trimerization prod-
uct 1,3,5-triarylbenzene.18-20 Of those byproducts we
have only detected butadiynes, and this was in connec-
fraction and the purified product was striking, and it
was virtually impossible to establish the extent of
impurities from the 1H NMR spectra alone, so a
technique like SEC is absolutely needed.
We propose that the large molecular weight impuri-
ties are oligomers of 12 to account for the similarities
1
in the H NMR spectra and for the fact that the UV
spectra of the fractions were very similar and contained
the signatures of a zinc porphyrin. The purified fraction
containing mainly 11 could be deprotected followed by
recrystallization from chloroform, and the product was
filtered using a 0.45 µm filter to give pure 12. Similar
tion with the synthesis of compound 3. With 1H and 13
C
NMR and MALDI TOF we instead detected ∼3%
1-bromo-2,5-dioctyl-4′-(4′′-trimethylsilylethynylphenyl)-

Macromolecules, Vol. 38, No. 4, 2005
Zn Porphyrin Linked Poly(phenyleneethynylene) 1183
Scheme 2. Synthetic Procedure Leading to the Homopolymer Poly[(2,5-dioctyltolanyl)Ethynylene] (Nn)a
a (a) K2CO3, MeOH, CH2Cl2; (b) Pd(PPh3)4, CuI, piperidine.
Table 1. Data from the SEC and UV-Vis Analysis for All
The Pd was removed by using DDC (see the section
Problems with Pd).
Three Polymersa
Porphyrin Linked Homopolymer Synthesis. We
attempted two different strategies to synthesize a
molecule with the general structure NPN, as shown in
the lower part of Figure 1. While the two different
procedures shown in Scheme 4 both accord with the
general structure depicted above, they differ in the way
the polymer segments are coupled to the porphyrin
nucleus. The only difference is the distance from the
porphyrin to the first phenyl ring bearing octyl substit-
uents. The proximity of the octyl substituents in the case
of NPN A reduces the degree of conjugation between
the polymer segment and the porphyrin system due to
steric effects.
Mw [g/
absorbance N/P cross-linkage
compound
Nn
NPN A crude 19 299 2.12
mol]
PDI
ratio
ratio (UV-vis Mw)b
92 300 2.90
2.5
2.5
3
2.5
2.5
2.5
2.5
2
8.5
8.5
fraction 1
fraction 2
fraction 3
fraction 4
fraction 5
fraction 6
fraction 7
66 020 1.18
41 255 1.16
22 175 1.15
13 058 1.18
8 373 1.11
4 727 1.09
2 839 1.03
10.0
8.5
8.5
8.5
8.5
6.9
NPN B crude 36 949 4.09
3
6.5
5
10.0
22.2
17.1
15.4
12.0
8.5
3.2c (n.a.)
9.1 (10200)
7.5 (8000)
4.3 (7300)
2.4 (5850)
2.1 (4350)
1.1 (3700)
-d (2900)
fraction 1
fraction 2
fraction 3
fraction 4
fraction 5
fraction 6
fraction 7
92 275 1.07
60 246 1.21
31 612 1.28
14 225 1.19
9 239 1.24
4 074 1.39
1 534 1.17
4.5
3.5
2.5
2
Therefore, there would be a difference between the
electronic properties of the NPN A polymer and the
NPN B polymer. (This was confirmed by the differences
in the UV-vis spectra for the two products.)
6.9
5.1
1.5
a The absorbance ratio is the ratio of the absorbance from the
Nn polymer (at 381 nm) and the porphyrin (at 451 nm). The N/P
ratio listed in this table is the ratio between the N and P units
based on their individual extinction coefficients. (see the section
Method B: Polymerization in the Presence of Dibromoporhyrin
for the results). b Molecular weights of 425 and 750 g mol-1 were
used respectively for the monomer and the porphyrin constituents
in the polymer. c Estimated from the increase in molecular weight
from 11 546 to 36 949 g mol-1 on standing. d A value for cross-
linking of 0.5 is obtained since the extinction coefficient for N is
overestimated at low molecular weights. A value <1 has no
physical meaning in terms of cross-linkage; it has therefore not
been shown.
Method A: Polymerization in the Presence of
Diethynylporphyrin. Our first choice of porphyrin
derivative was the diethynylporphyrin 12 since we
desired a reactivity of the porphyrin derivative that was
comparable or larger than the individual monomers to
ensure attachment of the porphyrin molecules to the
growing polymer chains during the polymerization
reaction.
The polymerization reaction of the monomer 8 in the
presence of the deprotected diethynylporphyrin 12
proceeded smoothly, giving a polymer product. The
evidence for zinc porphyrin content in the polymer was
found by the UV-vis spectrum of the product during
SEC analysis. We employed fractionation of the crude
polymer product using preparative SEC and obtained
seven fractions. The UV-vis spectra of the fractions are
shown in Figure 3, and while they show a gradual red
shift of the absorption maximum for the polymer
component with increasing molecular weight (as ex-
pected) the porphyrin content was roughly constant. The
ratio of the N and P absorption and the molecular
weight (Mw) determined by SEC in the seven fractions
are listed in Table 1. The ratio between the porphyrin
and polymer showed little variation and did not depend
on the molecular weight as expected, if only one por-
phyrin molecule had been incorporated into each NPN
molecule. These results could indicate that not only the
NPN A polymer was produced but also a large amount
of the buta-1,3-diyn-1,4-diyl byproduct derived from
oxidative dimerization (see Scheme 5).
problems have been described earlier by Tomizaki et
al.,21 where a seemingly pure product in terms of
porphyrin content (NMR and UV-vis data) turned out
to be a mixture of molecules with different size (i.e.,
different molecular weights) when SEC was used. This
could be oligomers or higher porphyrin components
(gela¨nder molecules).
Homopolymer Synthesis. Scheme 2 shows the
synthetic procedure of the conjugated homopolymer.
Employing Sonogashira cross-coupling conditions fol-
lowing standard methods, the iodine-terminated ho-
mopolymer poly[(2,5-dioctyltolanyl)ethynylene] (Nn) was
obtained. We have used Pd(PPh3)4 as catalyst instead
of Pd(PPh3)2Cl2 to minimize the formation of the
butadiyne byproducts, where two terminal acetylenes
are coupled when reducing Pd(II) to Pd(0). Pd(0) is most
probably the active species in the catalytic cycle for the
Sonogashira cross-coupling reaction.22 The molecular
weight (Mw) of the polymer Nn was determined using
SEC, and the results are listed in Table 1. Measure-
ments on electroative devices prepared using Nn showed
that it contained residual palladium, which were con-
firmed by inductively coupled plasma sector field mass
spectrometry (ICP-SFMS) measurements. The content
of Pd in the Nn polymer was determined to 49.3 mg/kg.
To account for this finding, we assume that a side
reaction involving the ethynyl groups of either the
growing polymer chain or the porphyrin takes place. In
the case of the polymerization starting from the dieth-
ynylporphyrin the growing polymer chain will always
expose an ethynyl group and thus be susceptible to
oxidative coupling of the ethynyl group to give a
butadiyne, which in principle gives a homopolymer

1184 Nielsen et al.
Macromolecules, Vol. 38, No. 4, 2005
Scheme 3. Two Different Synthetic Producers Leading to the Three-Domain Structure NPNa
a (a) Pd(PPh3)4, CuI, Et3N, THF.
segment with a defect in the growth direction such that
both polymer termini present iodine atoms. Such an
event, however, does not represent a termination of
chain elongation as a monomer or another dieth-
ynylporphyrin can react at the iodinated chain end
hence continue chain growth. The result is a polymer
product with an even distribution of porphyrin mol-
ecules with the amount of porphyrin determined largely
by the ratio of polymer to monomer in the initial
reaction mixture. In conclusion, the synthetic scheme

Macromolecules, Vol. 38, No. 4, 2005
Zn Porphyrin Linked Poly(phenyleneethynylene) 1185
Scheme 4. Synthetic Procedure of Compounds 9-12
Scheme 5. Illustration of Some of the Possible
Byproducts from Cross-Coupling of the Monomer
Unit 8 with the Diethynylporphyrin 12
for the incorporation of a single porphyrin molecule
between two polymer blocks based on a diethynylpor-
phyrin does not work because of the dimerization side
reaction that always takes place during the Sonogashira
reaction.
From a synthetic point of view, a small molecule
product is generally not affected by a small amount of
dimerization (of the order of 1-10%) as means of
purification are available (i.e., distillation, crystalliza-
tion). When employing the Sonogashira reaction for
polymerization, however, the defects introduced by the
side reaction are accumulated in the polymer product
and make the applicability of the Sonogashira reaction
for polymerization questionable, if a product with a high
degree of regioregularity or directionality is required.
Method B: Polymerization in the Presence of
Dibromoporphyrin. In this method we used dibro-
moporphyrin, 10, as the source of porphyrin during the
polymerization reaction. Seen in light of the finding
above, we would expect only one porphyrin molecule to
be incorporated in each polymer chain assuming that
10 does not lead to homocoupled byproducts. We would
also expect a considerable amount of homopolymer since
any dimerization of a pair of ethynyl groups will lead
to homopolymers with halogen termini in both ends of
the molecule that does not allow for the incorporation
of the dibromoporphyrin. From the SEC trace of the
crude product it was clear that the synthetic scheme
worked in terms of a decreasing porphyrin content with
increasing molecular weight. The SEC trace also con-
firmed the presence of porphyrin in molecules of high
molecular weight but revealed nothing about the ho-
mopolymer content.
Figure 3. UV-vis spectra of the seven fractions of the NPN
A polymer formed during the polymerization of the monomer
8 in the presence of the diethynyl porphyrin 12. The spectra
are normalized according to λmax of the polymer. It can be seen
how the absorption of the porphyrin (λ ) 451 nm) does not
vary significantly in intensity as a function of the length of
the Nn polymers.
The crude polymer product had a molecular weight
(Mw) of 11 546 g/mol and a polydispersity of 3.456.
Fractionation of the crude product into seven fractions

1186 Nielsen et al.
Macromolecules, Vol. 38, No. 4, 2005
mismatch in both high and low molecular weights. In
the low molecular weight range the mismatch is a result
of the estimate that the extinction coefficient of both
Nn and P is the same in the polymer and for the isolated
molecules. In the high molecular weight range the
mismatch is due to cross-linking since more than one
porphyrin molecule would be present in the cross-linked
supermolecules. This means that the molecular weight
(Mw) determined by SEC should be given by Mw ) x((N/
P)Mmonomer + Mporphyrin), which is in good accordance
with our observations (see Table 1). In the equation x
is the number of cross-link, N/P is the ratio between
the Nn monomer and the porphyrin obtained by the
UV-vis measurements, and Mmonomer and Mporphyrin are
the molecular weights monomer and porphyrin, respec-
tively. Further, SEC has a tendency to overestimate the
molecular weight of linear rigid molecules when using
polystyrene standards. SEC probes the hydrodynamic
volume of the polymer, and since polystyrene can be
assumed globular in chloroform solution and the poly-
mer systems studied remain linear, this will contribute
to the overestimation as the molecular weight increases.
This effect has not been taken into account in Table 1.
The average of the cross-linking, x, in Table 1 has a
value of 3.8 for the seven fractions, which is in good
accordance with the value of 3.2 obtained for the
molecular weight increase of the homopolymer on
standing (not corrected for the weight of the individual
fractions).
Problems with Pd. The problems concerning the
removal of Pd catalyst from samples is a large forgotten
area in the literature, which only very few authors25-27
have reported. Earlier work25 with palladium-catalyzed
polymerization using the Heck reaction for the polym-
erization have unfortunately shown that the removal
of palladium catalyst remnants can be problematic. It
is probably difficult to envisage complete removal of
catalyst remnants in the case of polymeric products.
With this paper and two recent ones25 we hope to start
a discussion on this topic. It is very important to be able
to remove all of the Pd catalyst, when e.g. a photovoltaic
or a light-emitting device of conjugated polymers is
produced, because remnants from the Pd catalyst in the
form of nanoparticles can lead to a short circuit of the
device.
From our experience with the palladium-catalyzed
Heck reaction, we would expect some remnants of
palladium in both the Nn and the NPN (both A and B)
polymer. The remnants of Pd in the Nn were confirmed
by ICP-SFMS and determined to be 49.3 mg/kg. We had
hoped that the remnants of palladium did not influence
the electrical and optical properties of the polymers too
much. Unfortunately, when we made a standard elec-
troactive device of the polymers produced by sandwich-
ing the polymer between two electrodes (see Supporting
Information) the resistance was as low as 30 Ω.
In connection to removing Pd from our polymers, we
have attempted to use diethylammonium diethyldithio-
carbamate (DDC) with some success as reported by
Jones et al.26 for solid-state synthesis. The reason for
choosing DDC is that it coordinates very strongly to Pd
(log K (Pd[DDC]2) ) 64.9)28 and more strongly than the
alkyne groups in the polymers. Furthermore, Briscoe
and Humphries28 have showed that the coordination of
DDC to Pd is very fast and that the coordination process
is complete after 5 min at room temperature. There were
several indications that DDC coordinated to Pd as
Figure 4. UV-vis spectra of the seven fractions of the NPN
B polymer. The spectra are normalized according to λmax for
the polymer. It can be seen how the absorption for the
porphyrin (λ ) 451 nm and 520-700 nm) varies in intensity
as a function of the length of the Nn polymers.
using preparative SEC as above revealed as expected a
variation of the porphyrin content in the individual
fractions. When SEC analysis was performed on the
individual fractions the following day, it was surprising
that the molecular weight of the fraction had increased,
as if cross-linking of the products took place. The
fractions were then left to stand and evaporate in the
hood in air for 3 days to allow for completion of the cross-
linking process. The resulting solids were boiled in
chloroform that dissolved only a part of the polymer
product. This finding is consistent with earlier reports
on the insolubility of poly(phenyleneethynylene)s once
having been subjected to complete drying.15,18a The
soluble fraction was then refractionated using prepara-
tive SEC, and this gave stable molecular weight frac-
tions. It should be noted that the molecular weight of
the crude polymer was increased to 36 949 g/mol with
a polydispersity of 4.09 during the 3 days. The fractions
show the desired variation of the porphyrin content
between the individual fractions (see Figure 4). The
repurified fractions were used for the emission spec-
troscopy studies. The UV-vis solution spectra of the
seven fractions were recorded, and from these spectra
it can be seen that the absorption maximum, λmax, as a
function of the polymer chain length was red-shifted 9
nm from fraction 7 through fraction 1. From Figure 4
it is observed that the red shift of λmax already stops at
fraction 5, which corresponds to approximately a decam-
er of the Nn in NPN B. These results are in agreement
with the results of Jones et al.,23 who reported studies
on oligomers at a precise length of 1,4-phenylene ethy-
nylenes. Bre´das et al.24 have shown that the band gap
as a function of n-1 (for large n > 5) has a linear
relationship. The average ratio of N and P in the NPN
polymer is obtained on the basis of the measured λmax
at 390 nm (absorption mainly from Nn) and 451 nm
(absorption mainly from P). The ratio for the Nn and P
absorption in the fraction are listed in Table 1 together
with the molecular weight (Mw) determined by SEC.
From the extinction coefficient for Nn and P (we used
the ꢀ431 in 12 for the absorption at 455 nm) an estimate
for the ratio between N and P can be obtained. The two
extinction coefficients were determined to be ꢀ390
)
70 000 M-1 cm-1 and ꢀ431 ) 240 000 M-1 cm-1. This
indicates a N/P ratio of 22.2:1; 17.1:1; 15.4:1; 12:1; 8.5:
1; 6.9:1, and 5:1 for fractions 1 to 7, respectively (see
Figure 4).
A comparison of the molecular size obtained by UV-
vis with the results from the SEC analysis shows a

Macromolecules, Vol. 38, No. 4, 2005
Zn Porphyrin Linked Poly(phenyleneethynylene) 1187
Scheme 6. Illustration of a Probable Trimerization
Cross-Linkage Process between Two Nn Polymers and
a NPN B Polymera
Figure 5. Emission spectra of fractions 1-7 of NPN B in
degassed chloroform. The excitation wavelength was set to the
maximum absorption of the polymer (around 388 nm). Peak
absorption was below 0.07 (concentrations below 10-5 M). The
inset shows the porphyrin emission of fractions 2, 4, and 6 for
comparison.
the optical properties have changed dramatically, when
the polymer was treated with DDC for more than 5 min.
It should be emphasized that there is little change in
the UV-vis spectrum of the polymer, which have been
treated with DDC for 5 min compared with the crude
polymer. It was not possible to observe the dithiocar-
1
bamoyl groups in the H NMR spectra recorded on the
DDC treated Nn polymer product. By comparing the
fluorescence spectra of the Nn polymer in solid state
before and after a long DDC treatment time, it can be
seen that there were significant changes (see Supporting
Information). The changes cannot however explain why
the electroactive devices after long DDC treatment time
did not show electroluminescence. The quantum yield
of fluorescence is approximately constant; there is
however a small blue shift of the emission maximum
upon DDC treatment and a shoulder at 415 nm that
becomes suppressed.
a The trimerization process is observed under normal So-
nogashira cross-coupling conditions.
An elementary analysis of the Nn polymer after a 2 h
DDC treatment time shows that the polymer contains
C, 84.8%; H, 9.1%; N, 0.5%; and S, <0.1% compared with
C, 88.4%; H, 9.5%; N, <0.3%; and S, <0.1% for the
untreated Nn polymer. The elementary analysis indi-
cates that it is not the DDC group itself that is added
to the polymer, since no S could be detected in the
polymer. However, we detected a little increase in the
contents of N corresponding to one N atom for every six
incorporated monomer units. From the elementary
analysis we got an indication that it is an N-containing
compound that is attached to the polymer by treatment
with DDC. While we cannot provide a solid explanation
for the reaction between DDC and the polymer, our
photophysical and electroactive device measurements
show that DDC treatment changes the properties of the
polymer. The NPN B polymer has also been treated with
DDC with success; electroluminescent devices of the Nn
itself and of NPN B polymer were prepared, and the
nature of the electroluminescence has been character-
ized.29 To decide whether the apparent cross-linkage of
the NPN B polymer products could be a result of the
content of Pd in the polymers, we performed a search
in the literature for a useful model study and found that
it has been known for more that 50 years that transition
metals catalyze the trimerization of alkynes.30 In 1997
and 2001, it was further shown by Gevorgyan et al.31,32
that substituted benzenes can be constructed from
enynes and activated alkynes in a [4 + 2] and from
expected and thereby removed Pd from the polymers.
The indications were that the solution shifted color from
yellow to dark brown very quickly (in a few seconds)
upon addition of DDC. Furthermore, the precipitated
polymer product after the treatment with DDC had a
lighter yellow color than the crude polymer product; the
resistance through a standard electroactive device
geometry (produced by sandwiching the polymer be-
tween two electrodes) increased from 30-50 Ω to 0.3-2
MΩ after the treatment with DDC. Our results indicate
that DDC very effectively removed or passivated Pd
from polymer samples. Unfortunately, there were also
some indications that DDC reacted with the polymers.
We observed that treatment with DDC, for periods of
time longer than 5 min (we attempted 2 and 24 h), led
to a dramatic change in some of the properties of the
polymers. The polymer became difficult to precipitate
from in methanol, and no electroluminescence could be
observed when a potential was applied. On the other
hand, we had success in removing Pd from the polymer
samples without the above-mentioned changes in prop-
erties for the polymers, when we used approximately 5
times excess of DDC, compared to the amount of Pd
used in the catalyst, and a reaction time between 3 and
5 min. From the UV-vis spectra of the Nn in the solid
state (see Supporting Information) it can be seen that

1188 Nielsen et al.
Macromolecules, Vol. 38, No. 4, 2005
Figure 6. Typical photon balance of the NPN B assembly in dilute degassed chloroform solution. The absorption is arbitrarily
set to 1000 photons, and the fluorescence emission was measured. Energy transfer as well as nonradiative decay is calculated
from the fluorescence quantum yields of the pure N and P domains. The relative porphyrin emission increases as the N chain
length decreases.
terminal alkynes in a [2 + 2 + 2] fashion in the presence
of a Pd(0) catalyst. Both reactions proceed under normal
Sonogashira conditions without an inert atmosphere. By
comparing these model studies of the trimerization of
alkynes with the known byproducts18-20 from a Sono-
gashira cross-coupling reaction, it is most likely that the
apparent cross-linkage of our polymer products are
formed by a trimerization process. A probable trimer-
ization process could be that two homopolymers Nn with
terminal alkynes (that are not coupled to the porphy-
rins) are forming an enyne in the presence of the Pd(0)
catalysts. This enyne can then react with one of the
triple bonds in both Nn and NPN B molecules to form
the trimerization cross-linkage (see Scheme 6). This can
Figure 7. Normalized fluorescence emission from spin-coated
be used to explain how the molecular weight of a
polymer fraction can increase when standing and ulti-
mately form an insoluble product with a high degree of
cross-linking. To avoid this trimerization cross-linkage
of the polymer, it is very important to remove all the
catalyst during precipitation of the polymers.
films of the Nn polymer (Mw ) 92.300 g mol-1) and the NPN B
assembly. The polymer emission is almost completely quenched
by the porphyrin.
The fluorescence quantum yields for the pure N domain
and pure P domain were determined to be 87% and
10.5%, respectively, using DPA as reference.
Fluorescence Measurements. We have performed
fluorescence measurements on the Nn and the NPN B
polymers. In Figure 5 normalized fluorescence spectra
of the seven fractions of NPN B are shown. In all cases
the excitation wavelength was set to the maximum
absorption of the polymer (around 388 nm). The emis-
sion below 600 nm is from the N domain, whereas the
porphyrin unit emits around 660 nm. The polymer
emission from the seven fractions shows little variation.
However, fractions 6 and 7, with the shortest N do-
mains, differ somewhat from the remaining fractions:
both spectra are more intense at 440 nm. In addition,
the spectrum of fraction 7 is blue-shifted 8 nm compared
to the remaining spectra, probably due to the relatively
short chain segments (oligomers). Another notable
feature is that the polymer emission extends more
toward long wavelengths as the N domains gets larger
as expected. The most interesting difference is that the
porphyrin emission seems to gradually increase as the
N domain gets shorter (see inset in Figure 5). We
interpret this as a consequence of the fact that the whole
polymer chain is probably not perfectly conjugated.
Hence, the chain will consist of a number of segments.
Thus, with increasing polymer length the more likely
it becomes that an exciton gets trapped on a segment
that is not in the porphyrin vicinity, and the energy
transfer efficiency decreases. Since the N domain fluo-
resces in the spectral area where the P domain absorbs,
a Fo¨rster type of energy transfer may be involved. Using
the same methodology as described earlier, photon
balances could be derived for the individual fractions.5,33
Assuming that the ratio between the fluorescence rate
and the nonradiative decay rates are similar in the NPN
B assembly as in the model compounds, a photon
balance can be calculated from the measured fluores-
cence emission. This is shown in Figure 6. The overall
quantum yield for the NPN B assembly was in general
around 60%, which mainly was caused by the N domain
emission. The fluorescence from the P domain was
limited compared with the N domain. The relative
emission of the P domain was found to increase 5-fold
going from fraction 1 to 7. This means that in solution
(or in a diluted matrix) the polymer length may be
crucial in multidomain light-harvesting systems.
Solid-state studies of NPN B shows that whereas
energy transfer is limited in solution, it becomes the
main deactivation route of the N domain in the solid
phase (see Figure 7); thus, N emission is nearly com-
pletely quenched by the porphyrin that emit strongly.
Thus, in the solid state the chain length is much less
important than in solution with regards to efficient
energy transfer. The more efficient energy transfer in
the solid state seems to be a general rule for this kind
of polymer-dye systems.5,33 The N and P emission are
red-shifted by approximately 65 nm compared to the
solution spectra. Also, the relative intensity of the N
domain and P domain peaks are changed.
The lifetime decay of the pure P domain could be
fitted with a single-exponential function, with a decay
of 2.4 ns in degassed chloroform. The lifetime decay of
the N domain needed a linear combination of two

Macromolecules, Vol. 38, No. 4, 2005
Zn Porphyrin Linked Poly(phenyleneethynylene) 1189
(3) Katz, H. E.; Lovinger, A. J.; Johnson, J.; Kloc, C.; Siegrist,
T.; Li, W.; Lin, Y.-Y.; Dodabalapur, A. Nature (London) 2000,
404, 478-481.
(4) Granstro¨m, M.; Petritsch, K.; Arias, A. C.; Lux, A.; Andersson,
M. R.; Fiend, R. H. Nature (London) 1998, 395, 257-260.
(5) Krebs, F. C.; Hagemann, O.; Spanggaard, H. J. Org. Chem.
2003, 68, 2463-2466.
(6) Li, B.; Li, J.; Fu, Y.; Bo, Z. J. Am. Chem. Soc.. 2004, 126,
3430-3431.
(7) Wong, M. S.; Zhong, H. L.; Shek, M. F.; Chow, K. H.; Tao,
Y.; D’Iorio, M. J. Mater. Chem. 2000, 10, 1805-1810.
(8) Gill, R. E.; van Hutten, P. F.; Meetsma, A.; Hadziioannou,
G. Chem. Mater. 1996, 8, 1341-1346.
(9) Yao, Y.; Tour, J. M. J. Org. Chem. 1999, 64, 1968-1971.
(10) Pelter, A.; Jenkins, I.; Jones, D. E. Tetrahedron 1997, 53,
10357-10400.
exponential functions to obtain a reasonable fit in
degassed chloroform: 259 ps (86%) and 1.6 ps (14%).
In the spin-coated film the measured lifetime of the N
domain is significantly larger: 1.4 ns (53%) and 2.7 ns
(47%). This could be caused by the hampering of some
of the deactivation routes that are effective in solution.
A side effect is that the exciton has more time to be
transferred to a porphyrin unit, and this could be one
of the reasons why more efficient energy transfer is
observed. Interestingly, the fluorescence lifetime of the
porphyrin unit was significantly shorter in the solid
state than in solution. The lifetime decay was complex,
and a good fit could not be obtained. This is not
unexpected since a number of interactions and different
local environments are present in the solid NPN B
matrix. The average lifetime was approximately 300 ps.
(11) Plater, M. J.; Aiken, S.; Bourhill, G. Tetrahedron 2002, 58,
2405-2413.
(12) Ware, R. W.; Rothman, W. Chem. Phys. Lett. 1976, 39, 449-
453.
(13) Ziener, U.; Godt, A. J. Org. Chem. 1997, 62, 6137-6142. (b)
Hortholary, C.; Coudret, C. J. Org. Chem. 2003, 68, 2167-
2174. (c) Sabourin, E. T.; Onopchenko, A. J. Org. Chem. 1983,
48, 5135-5137. (d) Kukula, H.; Veit, S.; Godt, A. Eur. J. Org.
Chem. 1999, 277-286. (e) Li, J.; Pang, Y. Synth. Met. 2003,
10383-10389. (f) Jones, L.; Schumm, J. S.; Tour, J. M. J.
Org. Chem. 1997, 62, 1388-1410. (g) Che, C.-M.; Yu, W.-Y.;
Chan, P.-M.; Cheng, W.-C.; Peng, S.-M.; Lau, K.-C.; Li, W.-
K. J. Am. Chem. Soc. 2000, 122, 11380-11392.
(14) Brandsma, L.; Vasilevsky, S. F.; Verkruijsse, H. D. Applica-
tion of Transition Metal Catalysts in Organic Synthesis;
Springer-Verlag: Berlin, 1998.
Conclusions
We have presented the synthesis of a new zinc
porphyrin linked conjugated polymer. We experienced
problems in controlling the nature of polymer product
and found that it was a result of many factors: oxidative
dimerization and cross-linking of the ethynylene groups
and palladium remnants from the catalyst. We thor-
oughly studied products and the reaction by two differ-
ent synthetic approaches and demonstrated that a
polymer product with one porphyrin molecule inserted
in each polymer chain could be obtained by choosing the
right synthetic approach. We separated the polymer
material into fractions using preparative SEC and
studied both the cross-linking of the individual fractions
and purified them further to finally perform thorough
photophysical studies. We showed the light-harvesting
properties of the three-domain structure as a function
of the chain length of the Nn polymer. This new zinc
porphyrin linked homopolymer shows a very effective
energy transfer from the polymer to the porphyrin in
solid state. Finally, we have identified some problems
concerning remnants of Pd from the catalyst in the
polymer product, and devise a procedure for the removal
of the Pd remnants such that standard electroactive
devices could be constructed.
(15) Solomin, V. A.; Heitz, W. Macromol. Chem. Phys. 1994, 195,
303-314.
(16) Klo¨pping, H. L.; van Der Kerk, G. J. M. Recueil 1951, 70,
917-939.
(17) Eaborn, C.; Thompson, A. R.; Walton, D. R. M. J. Chem. Soc.
1967, 1364-1366.
(18) (a) Heitz, W. Pure Appl. Chem. 1995, 67, 1951-1964. (b)
Kukula, H.; Veit, S.; Godt, A. Eur. J. Org. Chem. 1999, 277-
286. (c) Li.; Y.; Pang, Y. Synth. Met. 2003, 10383-10389. (d)
Gevorgyan, V.; Radhakrishnan, U.; Takeda, A.; Rubina, M.;
Rubin, M.; Yamamoto, Y. J. Org. Chem. 2001, 66, 2835-2841.
(19) Heitz, W. Polym. Prepr. 1991, 32, 327-328.
(20) Trumbo, D. L.; Marvel, C. S. J. Polym. Sci., Part A: Polym.
Chem. 1987, 25, 1027-1034.
(21) Tomizaki, K. Y.; Yu, L. H.; Wei, L. Y.; Bocian, D. F.; Lindsey,
J. S. J. Org. Chem. 2003, 68, 8199-8207.
(22) Sonogashira, K. In Metal-Catalyzed Cross-Coupling Reac-
tions; Diederich, F., Stang, P. J., Eds.; Wiley-VCH Verlag:
Weinheim, 1998; Chapter 5.
(23) Jones II, L. R.; Schumm, J. S.; Tour, J. M. J. Org. Chem.
1997, 62, 1388-1410.
(24) Bre´das, J. L.; Silbey, R.; Boudreaux, D. S.; Chance, R. R. J.
Am. Chem. Soc. 1983, 105, 6555-6559.
(25) (a) Krebs, F. C.; Nyberg, R. B.; Jørgensen, M. Chem. Mater.
2004, 16, 1313-1318. (b) Nielsen, K. T.; Bechgaard, K.;
Krebs, F. C. Macromolecules, in press.
(26) Jones, L.; Schumm, J. S.; Tour, J. M. J. Org. Chem. 1997,
62, 1388-1410.
(27) Van Broekhoven, J.; Adrianus, M. European Patent Applica-
tion 0283092 A1, 1988.
(28) Briscoe, G. B.; Humphries, S. Talanta 1969, 16, 1403-1419.
(29) Nielsen, K. T.; Spanggaard, H.; Krebs, F. C. Displays, in
press.
(30) Reppe, W.; Schlicthing, O.; Klager, K.; Toepel, T. Justus
Liebigs Ann. Chem. 1948, 560, 1-7.
(31) Gevorgyan, V.; Takeda, A.; Yamamoto, Y. J. Am. Chem. Soc.
1997, 119, 11313-11314.
(32) Gevorgyan, V.; Radhakrishnan, U.; Takeda, A.; Rubina, M.;
Rubin, M.; Yamamoto, Y. J. Org. Chem. 2001, 66, 2835-2841.
(33) Krebs, F. C.; Spanggaard, H.; Rozlosnik, N.; Larsen, N. B.;
Jørgensen, M. Langmuir 2003, 19, 7873-7880.
Acknowledgment. The Danish Technical Research
Council (STVF) supported this work. We express sincere
gratitude to Mette P. Jeppsen, Ole Hagemann, and Jan
Alstrup for technical assistance.
Supporting Information Available: General analytical
techniques, synthetic procedures, characterization, UV-vis,
and fluorescence spectra of the polymer and the DDC-treated
polymer, and electroactive device preparation. This material
is available free of charge via the Internet at http://
pubs.acs.org.
References and Notes
(1) Chiang, C. K.; Fincher, C. R.; Park, Y. W.; Heeger, A. J.;
Shirakawa, H.; Louis, E. J.; Gau, S. C.; Macdiarmid, A. G.
Phys. Rev. Lett. 1977, 39, 1098-1101.
(2) Burroughes, J. H.; Bradley, D. D. C.; Brown, A. R.; Marks,
R. N.; Mackay, K.; Friend, R. H.; Burn, P. L.; Holmes, A. B.
Nature (London) 1990, 347, 539-541.
MA048489U