Preparation and deposition of stable monolayers of fullerene derivatives

Article abstract of DOI:10.1039/a705950g

Two new fullerene derivatives have been synthesized, which tend to form Langmuir monolayers. Robust monomolecular layers are formed at the air/water interface, that can be transferred onto hydrophilic substrates by Langmuir-Blodgett or Langmuir-Schaefer techniques. The measured areas per molecule (105 A² for 1a dropped from a deuteriochloroform solution at 20°C) reach those expected for a true monolayer of C₆₀ at the air/water interface.

Full text of DOI:10.1039/a705950g

Preparation and deposition of stable monolayers of fullerene derivatives
Ping Wang,a† Bo Chen,a Robert M. Metzger,*a Tatiana Da Rosb and Maurizio Prato*b
aDepartment of Chemistry, University of Alabama, T uscaloosa, AL 35487–0336, USA
bDipartimento di Scienze Farmaceutiche, Universita` di T rieste, Piazzale Europa 1, I-34127 T rieste, Italy
Two new fullerene derivatives have been synthesized, which tend to form Langmuir monolayers. Robust monomolecular layers are
formed at the air/water interface, that can be transferred onto hydrophilic substrates by Langmuir–Blodgett or Langmuir–Schaefer
˚
techniques. The measured areas per molecule (105 A2 for 1a dropped from a deuteriochloroform solution at 20 °C) reach those
expected for a true monolayer of C at the air/water interface.
60
Since fullerenes were first discovered,1 and an efficient synthesis
for them was developed,2 the preparation of thin films contain-
ing fullerenes has become an important issue in the applications
of these new forms of carbon.3 In this connection, either self-
assembled monolayer (SAM) or Langmuir–Blodgett (LB)
deposition techniques may play a significant role.3 A very nice
example of molecular recognition applied to the formation of
molecular monolayers has also been reported recently.4
Whereas monomolecular layers of fullerenes and fullerene
derivatives have already been obtained by several groups, their
deposition has raised some difficulty, with very few successful
results.3 Both C and C form condensed layers at the
Fig. 1(b) shows, for 1a dissolved in deuteriochloroform, a
˚
molecular area of A =105.4 A2 (extrapolated to zero film
o
pressure, P=0), in fair agreement with the theoretically
˚
expected value of 93 A2 molecule−1 (ref. 29) and the experimen-
˚
tally determined value of 96 A2 molecule−1 (ref. 17) for pure
chloroform as a spreading solvent, forms a stable and true
C
. Obviously, the chemically modified 1a, with deuterio-
60
monolayer on air/water interface, but also has a slightly in-
˚
creased area, when compared to the expected C area of 93 A2
60
molecule−1. Using different spreading solution concentrations
(1.44×10−4–1.0×10−3 ) or different subphase temperatures
(10–23 °C) had no effect on the isotherm; thus, monomolecular
layers are formed much more easily for compound 1a than for
60
70
air/water interface.5–7 However, the layers of C are poorly
60
behaved, and their transfer to solid substrates generally results
in films of poor quality.6,8 In fact, the non-amphiphilic C
molecules tend to aggregate, and form layers several molecules
pure C , even with concentrated (10−3 ) spreading solutions.
area at film collapse (A ) is much smaller than the extrapolated
60
It should be noted that, for all molecules in Table 1, the
60
c
thick at the air/water interface,6 instead of true monolayers,
area at zero pressure (A ): it appears that for the very robust
films in this study, A , measured at very high film pressures,
measures an incipient clustering of molecules, rather than the
0
unless very dilute spreading solutions are used.9 For pure C
monolayers form only if the concentration of C solution is
,
60
c
60
lower than 10−5 ,9 and extreme care should be taken during
the film compression, otherwise multilayers are obtained with
˚
an average typical cross-sectional area of ca. 22 A2 molecule−1,
which means that up to five molecules are crowded atop
each other.6
Since then, two strategies have been used to overcome these
problems and prepare stable C monolayers. One is to chemi-
cally modify a surface with a reagent that allows covalent
60
bonding to C ;3,7,10 another is to chemically modify C
itself,11–13 to produce a self-assembling derivative.3,6,14–27 Here,
we report results obtained by the second strategy.
60
60
Scheme 1 (a, R=CH CH OCH CH OCH CH OCH ; b, R=n–
2
2
2
2
2
2
3
C
H )
12 25
Results and Discussion
Compounds 1a,b were prepared according to Scheme 1.
Condensation of glycine derivatives 2a,b with paraformal-
dehyde generates the reactive 1,3-dipole azomethine ylide,
which readily adds to C across the junction between two six-
membered rings.28 Both products possess C symmetry, due
to fast nitrogen inversion, easily detectable by the reduced
60
2v
number of resonances in the 13C NMR spectra and by the
equivalence of the pyrrolidine methylene protons in the 1H
NMR spectra.
Compound 1a was dissolved in deuteriochloroform or tolu-
ene, to form three solutions of concentrations 1.44×10−4 ,
1.03×10−3  or 9.90×10−4 , respectively. The surface press-
ure–area isotherms of 1a are shown in Fig. 1. Both curves (a)
and (b) of Fig. 1 show that the floating films are very rigid, and
have a high collapse pressure (>70 mN m−1).
Fig. 1 Surface pressure–area isotherms of 1a at 20 °C: (a) [1a]
1.03×10−3  in toluene; (b) [1a] 9.9×10−4  in deuteriochloroform
† Present address: Samsung Industries USA, San Jose, CA.
J. Mater. Chem., 1997, 7(12), 2397–2400
2397

˚
Table 1 Molecular areas (A2 molecule−1) A (at first onset of non-
ons
zero film pressure), A (linearly extrapolated from finite pressure to
0
zero film pressure), A (at film collapse) and collapse pressures P as
c
c
a function of temperature T
˚
molecular areas/A2
molecule−1
compound
T /°C
A
A
A
P /mN m−1
—
ons
0
c
c
1a (in C H )
20
140
68
150
97
134.4
67.8
105
2
2
2
—
7
9
#40
<40
#40
>65
>65
>70
1a (in CDCl )
20
20
3
1b (in CDCl )
3
66
ideal area for a ‘relaxed’ film with ‘normal’ intermolecular
distances and with no clustering in the direction normal to the
film plane.
Fig. 3 Optical absorbance at 341.3 nm vs. number of LB layers of 1a
transferred onto a glass microscope slide
When toluene is used as the spreading solvent, Fig. 1(a)
˚
shows a lower molecular area (A =67.8 A2), together with a
0
˚
was studied by STM, using bias voltages between 100 and
300 mV, the constant current mode, and set point currents
between 2 and 5 nA. A typical well ordered region on the LB
shoulder at 134.4 A2. This shoulder appears only sometimes
with toluene, even at different subphase temperatures, but
never with deuteriochloroform: it may be due to some transient
molecular association.17 The smaller molecular area for 1a
spread from a toluene solution [Fig. 1(a)], compared with 1a
in deuteriochloroform [Fig. 1(b)] suggests that the molecules
of 1a ‘dropped’ from toluene are partially clustered.
Using either toluene or deuteriochloroform, the Langmuir
films at the air/water interface were robust, and stable (no
barrier movement) over a period of at least 48 h at an applied
pressure of 25 mN m−1. The water temperature showed very
little influence on the air/water film properties within the range
of 10–25 °C.
˚
film of compound 1a is shown in Fig. 4(a). A 7.9 A separation
˚
in the horizontal axis, a 5.1 A separation in the vertical axis
and an angle of 85° were observed, which are very similar to
our previous results obtained for pure C .6 This indicates that
compound 1a can form a very well ordered region after transfer
60
to a substrate. The image is not perfectly resolved, and the
aliphatic chains are probably not seen clearly. Fig. 4(b) presents
an approximate illustration of the presumed stacking, using
the fcc lattice of pure C
.
60
In order to study the influence of subphase on LB film
properties, the pure water subphase was replaced by a
2.59×10−2  aqueous KCl solution. Typical surface press-
ure–area isotherms are shown in Fig. 5. For compound 1a, the
area per molecule on a dilute aqueous KCl surface is reduced
by about 10% [Fig. 5(b)], relative to the isotherm on a pure
water surface [Fig. 5(a)], and the pressure–area isotherm also
shows a slight change. This may be due to the interaction
between K+ and the triethylene glycol chain, as if the triether
ends of two 1a molecules associated to form, roughly, one 18-
crown-6, to coordinate to one K+ ion.22
The Langmuir film of 1a was transferred at 25 mN m−1
by the LB technique onto quartz at
a dipping speed
<3.5 mm min−1, or carefully transferred onto highly oriented
pyrolytic graphite (HOPG) by the Langmuir–Schaefer (quasi-
horizontal transfer) technique. The film transfer was Y-type. It
is found that good transfer conditions for the first layer were
very important for the successful deposition of subsequent
layers.
For optical absorption measurements, 100 layers of the Y-
type film of 1a were deposited onto isopropyl alcohol-coated
quartz (Spectrasil). The UV–VIS spectrum (Fig. 2) shows
absorption maxima at 219.1, 267.9 and 341.3 nm, in a good
A
deuteriochloroform solution of compound 1b
(1.32×10−4 ) was carefully spread on a purified water sub-
phase. A typical surface area–pressure isotherm is shown in
Fig. 6. The collapse pressure was >70 mN m−1, but, despite
many attempts, the area per molecule (extrapolated to zero
agreement with those for C derivatives in cyclohexane.28 To
investigate the quality of the LB deposition, a stepped thickness
60
structure was built up on a glass microscope slide pretreated
with isopropyl alcohol. The variation of the absorbance with
the number of film layers is shown in Fig. 3, indicating a good
LB deposition up to about the first 20 layers.
˚
pressure) did not exceed A =66 A2 (Table 1). This can be
understood because in 1b, unlike compound 1a, both ends are
0
hydrophobic, as is C itself, and this does not favor monolayer
formation. However, compound 1b forms a stable multilayer
60
The orientation of the LB film for compound 1a on HOPG
film on air/water interface, which did not show any noticeable
area loss at a constant pressure (25 mN m−1) at least for 48 h.
The isotherm is reversible in the range 0–40 mN m−1.
In conclusion, we report the synthesis of an amphiphilic and
a hydrophobic fullerene derivative, prepared to improve mono-
layer formation at the air/water interface. Compound 1a can
be considered a model fullerene for LB film formation. It is
prepared by a general synthetic route, the cycloaddition of
azomethine ylides to C , that relies on the condensation
between an a-amino acid and an aldehyde. The amino acid 2a
60
used here is of great potential value for monolayer studies. In
principle it can be condensed with a variety of aldehydes
(RCHO) for the preparation of an entire new class of com-
pounds. This class is characterized by the presence of a
triethylene glycol chain, introduced to ensure hydrophilicity
and whatever R group (from the aldehyde) in position 2 of the
pyrrolidine ring. By this approach, either donor or acceptor
units can be attached to C and the resulting molecules should
reasonably give stable monolayers. In fact, the model amphi-
Fig. 2 Absorption spectrum for 1a deposited onto a quartz (Spectrasil)
slide (100 layers per side)
60
2398
J. Mater. Chem., 1997, 7(12), 2397–2400

Fig. 5 Surface pressure–area isotherm of 1a (a) on a pure Millipore
Milli-Q water subphase and (b) on a Millipore Milli-Q water subphase
made 2.59×10−2  in KCl. Deuteriochloroform was used as the
spreading solvent.
Fig. 6 Surface pressure–area isotherm of 1b (concentration
1.32×10−4  in deuteriochloroform)
relative to tetramethylsilane. UV–VIS absorption spectra were
taken on a Jasco V550 UV–VIS spectrophotometer. MALDI
(matrix-assisted laser desorption ionization) mass spectra were
obtained in positive linear mode at 15 kV acceleration voltage
on a ReflexTM time of flight mass spectrometer (Bruker) using
2,5-dihydroxybenzoic acid as matrix. Reactions were moni-
tored by thin-layer chromatography using Merck precoated
Fig. 4 (a) STM image of 1a (20 layers deposited on HOPG by LS
transfer). (b) Approximate computer model (CAChe Editor, on
˚
MacIntosh 8100AV), using the C fcc lattice (a=14.11 A) and touching
van der Waals spheres of diameter 10 A for the fullerene cores. The
60
˚
silica gel 60-F
(0.25 mm thickness) plates. Flash column
˚
alkyl tails of 1a are not shown. Repeat distances of 7.1 A in the
horizontal direction are seen assuming that the face diagonal [110]
254
chromatography was performed employing 230–400 mesh
silica gel (from Baker). Reaction yields were not optimized and
refer to pure, isolated products.
of the fcc lattice projects at an angle of 45° to the plane of projection.
Repeat distances of 5.0 A assume a [022] distance between the
˚
C
was purchased from Bucky USA (>99.5%). All other
horizontal row of spheres seen in front, and the next horizontal row
of spheres vertically below them, yet partially behind them.
60
reagents were used as purchased from Fluka or Aldrich. The
synthesis of N-(3,6,9-trioxadecyl)glycine 2a has been described
elsewhere.30 All solvents were distilled prior to use.
Cyclohexane, employed for UV–VIS measurements was a
commercial spectrophotometric grade solvent.
philic fullerene derivative 1a forms a stable ‘true’ monolayer
at the air/water interface when ‘dropped’ from a deuteriochloro-
˚
form solution (area per molecule at zero pressure of 105 A2,
˚
vs. calculated area of 93 A2 for pure C ). This monolayer can
be transferred onto a glass microscope slide or quartz slide by
60
N-(n-Dodecyl)glycine 2b
the LB method, or onto HOPG by the LS method. The same
fullerene derivative 1a has a very strong interaction with a
KCl–water surface. Instead, the hydrophobic fullerene deriva-
tive 1b, used for comparison, does not form a true monolayer
film at the air/water interface.
To a solution of n-dodecyl aldehyde (0.420 g, 2.28 mmol) and
glycine ethyl ester hydrochloride (0.318 g, 2.28 mmol) in 12 ml
of methanol–acetic acid (9951), was added NaBH CN (0.178 g,
2.84 mmol). The solution was stirred at room temperature for
3
2 h, then aqueous NaHCO and ethyl acetate were added. The
organic phase was separated and purified by column chroma-
3
Experimental
tography [eluent: light petroleum–ethyl acetate (951)] after
which 0.180 g (0.663 mmol, 29%) of N-(n-dodecyl)glycine ethyl
General
ester was obtained. 1H NMR (CDCl ): d 4.18 (q, J=7.1 Hz,
3
1H and 13C NMR spectra were recorded on a Varian Gemini
200 in CDCl solutions. Chemical shifts are given in ppm (d)
2H), 3.39 (s, 2H), 2.58 (t, J=7.1 Hz, 3H), 1.96 (br s, 1H), 1.48
(t, J=6.7 Hz, 2H), 1.26 (s, 20H), 0.87 (t, J=6.7 Hz, 3H).
3
J. Mater. Chem., 1997, 7(12), 2397–2400
2399

13C NMR (CDCl ): d 172.59, 60.70, 51.06, 49.70, 31.94, 30.09,
29.64, 29.56, 29.37, 27.27, 22.71, 14.26, 14.13. IR (KCl): 2920,
pretreated in isopropyl alcohol for 2 h, to make their surface
hydrophobic.
3
UV–VIS spectra of LB multilayers were measured on a
Perkin-Elmer Lambda 4B spectrophotometer. The LB films
on a HOPG substrate were studied by scanning tunnelling
microscopy (STM) in air at room temperature, using a Digital
Instruments Nanoscope II equipped with a type A head, and
a Pt/Ir tip, using bias voltages between 100 and 300 mV, the
constant current mode, and set point currents between 2 and
5 nA.
2848, 2210, 1745, 1565, 1465, 1373, 1309, 1198, 1068, 900, 725,
593, 494 cm−1. MS (EI): m/z 271 (M+), 226, 212, 199, 198,
184, 116, 57. Anal. Calc. for C H NO : C, 70.8; H, 12.25; N,
16 33
2
5.16. Found: C, 71.7; H, 12.0; N, 5.17%.
To N-(n-dodecyl)glycine ethyl ester (0.100 g, 0.37 mmol) in
30 ml of methanol–water (151), was added sodium carbonate
(0.400 g, 3.7 mmol) and the mixture was stirred for 72 h at
room temperature. Then 0.5 ml of 12  HCl was added and
the resulting solution extracted with ethyl acetate. 1H NMR
[CDCl –CD OD (852)]: d 3.31 (br s, 2H), 2.76 (br t, J=
References
3
3
7.3 Hz, 2H), 1.59 (br s, 1H), 1.18 (s, 22 H), 0.81 (t, J=6.4 Hz,
3H). 13C NMR [CDCl –CD OD (852)]: d 168.9, 49.7, 49.3,
1
2
H. W. Kroto, J. R. Heath, S. C. O’Brien, R. F. Curl and
R. E. Smalley, Nature (L ondon), 1985, 318, 162.
W. Kratschmer, L. D. Lamb, K. Fostiropoulos and D. R. Huffman,
¨
Nature (L ondon), 1990, 347, 354.
3
3
47.5, 47.3, 47.2, 31.6, 29.3, 29.2, 29.1, 28.8, 26.2, 25.9, 22.3, 13.7.
IR (KCl): 3189, 2920, 2847, 1566, 1463, 1372, 1307, 1153, 1083,
873, 827, 798, 777, 726, 672, 529, 558, 507, 495, 466, 422,
405 cm−1. MS (EI): m/z 243 (M+), 199, 198, 89, 57.
3
4
C. A. Mirkin and W. B. Caldwell, T etrahedron, 1996, 52, 5113.
F. Arias, L. A. God’nez, S. R. Wilson, A. E. Kaifer and
L. Echegoyen, J. Am. Chem. Soc., 1996, 118, 6086.
Y. S. Obeng and A. J. Bard, J. Am. Chem. Soc., 1991, 113, 6279.
P. Wang, M. Shamsuzoha, X. Wu, W. Lee and R. M. Metzger,
J. Phys. Chem., 1992, 96, 9027.
5
6
Fullerene derivatives 1a and 1b
To C (50 mg, 0.07 mmol) in 30 ml of toluene, were added
0.07 mmol of amino acid (2a or 2b) and 10.4 mg (0.35 mmol)
of paraformaldehyde. The solution was heated to reflux for
1 h, the solvent evaporated and the crude product was purified
by chromatography. The products were recrystallized from
dichloromethane–methanol.
7
8
9
G. Williams, A. G. Moore, M. R. Bryce, Y. M. Lvov and M.
C. Petty, Synth. Met, 1993, 55–57, 2955.
G. Williams, G. Pearson, M. R. Bryce and M. C. Petty, T hin Solid
Films, 1992, 209, 150.
L. O. S. Bulhoes, Y. S. Obeng and A. J. Bard, Chem. Mater., 1993,
5, 110.
60
10 K. Chen, W. B. Caldwell and C. A. Mirkin, J. Am. Chem. Soc.,
1993, 115, 1193.
11 A. Hirsch, T he Chemistry of the Fullerenes, Thieme, Stuttgart, 1994.
12 T he Chemistry of Fullerenes, ed. R. Taylor, World Scientific,
Singapore, 1995.
13 F. Diederich and C. Thilgen, Science, 1996, 271, 317.
14 F. Diederich, U. Jonas, V. Gramlich, A. Herrmann, H. Ringsdorf
and C. Thilgen, Helv. Chim. Acta, 1993, 76, 2445.
15 G. Williams, A. Soi, A. Hirsch, M. R. Bryce and M. C. Petty, T hin
Solid Films, 1993, 230, 71.
16 L. M. Goldenberg, G. Williams, M. R. Bryce, A. P. Monkman,
M. C. Petty, A. Hirsch and A. Soi, J. Chem. Soc., Chem. Commun.,
1993, 1310.
1a: C H NO , 38%; 1H NMR d 4.51 (s, 4H), 4.06 (t, J=
69 19
3
5.6 Hz, 2H), 3.90–3.70 (m, 6H), 3.60–3.54 (m, 2H), 3.37 (s, 3H),
3.36 (t, J=5.6 Hz, 2H). 13C NMR d 155.15, 147.32, 146.26,
146.13, 146.08, 145.73, 145.42, 145.32, 144.59, 143.13, 142.65,
142.28, 142.10, 141.91, 140.17, 136.24, 72.07, 70.90, 70.81, 70.71,
70.55, 68.55, 59.13, 54.28. IR (KCl): 2865, 1427, 1340, 1185,
1113, 767, 704, 597, 575, 553, 526 cm−1. MALDI-MS: m/z 909
(M+), 932 (M+Na)+. UV–VIS (cyclohexane) l /nm: 702,
max
2
430, 323, 304, 255, 212. Anal. Calc. for C H NO : C, 91.08;
H, 2.11; N, 1.54. Found: C, 90.0; H, 2.07; N, 1.51%.
16 33
1b: C H N, 41%; 1H NMR d 4.36 (s, 4H), 3.04 (t, J=
74 29
17 N. C. Maliszewskyi, P. A. Heiney, D. H. Jones, R. M. Strongin,
M. A. Cichy and A. B. Smith III, L angmuir, 1993, 9, 1439.
18 C. J. Hawker, P. M. Saville and J. W. White, J. Org. Chem., 1994,
59, 3503.
19 Y. Li, Y. Xu, Y. Mo, F. Bai, Y. Li, Z. Wu, H. Han and D. Zhu, Solid
State Commun., 1994, 92, 185.
20 M. Maggini, A. Karlsson, L. Pasimeni, G. Scorrano, M. Prato and
L. Valli, T etrahedron L ett., 1994, 35, 2985.
21 M. Maggini, L. Pasimeni, M. Prato, G. Scorrano and L. Valli,
L angmuir, 1994, 10, 4164.
7.4 Hz, 2H), 1.90 (qnt, J=7.4 Hz, 2H), 1.59 (m, 2H), 1.15–1.50
(m, 16H), 0.85 (m, 3H). 13C NMR d 153.98, 152.67, 146.19,
145.15, 144.98, 144.60, 144.35, 144.18, 143.48, 142.02, 141.55,
141.17, 141.00, 140.82, 139.11, 135.20, 69.57, 66.98, 54.14, 31.15,
28.95, 28.91, 28.62, 28.07, 26.91, 22.02, 13.35. IR (KCl): 2924,
2849, 1510, 1461, 1182, 526. MALDI-MS: m/z 931 (M+).
UV–VIS (cyclohexane) l /nm: 702, 430, 323, 305, 255, 212.
max
Anal. Calc. for C H N: C, 95.36; H, 3.14; N, 1.50. Found: C,
95.6; H, 3.24; N, 1.50%.
74 29
22 U. Jonas, F. Cardullo, P. Belik, F. Diederich, A. Gugel, E. Harth,
¨
A. Herrmann, L. Isaacs, K. Mullen, H. Ringsdorf, C. Thilgen,
¨
P. Uhlmann, A. Vasella, C. A. A. Waldraff and M. Walter, Chem.
Eur. J., 1995, 1, 243.
23 D. M. Guldi, Y. Tian, J. H. Fendler, H. Hungerbulher and K.-
¨
D. Asmus, J. Phys. Chem., 1995, 99, 17673.
Monolayer and multilayer methodology
Solutions of compounds 1a and 1b in deuteriochloroform or
toluene were carefully spread onto a purified water subphase
(Millipore Milli-Q, resistivity 16 MV cm) in a vibration-iso-
lated Lauda film balance at room temperature and also at
thermostatically controlled lower water temperatures of 10, 15,
and 20 °C. The spread solutions were left for periods ranging
from 20 min to 12 h, after which they were compressed at a
barrier speed of 26 mm min−1.
24 M. Matsumoto, H. Tachibana, R. Azumi, M. Tanaka,
T. Nakamura, G. Yunome, M. Abe, S. Yamago and E. Nakamura,
L angmuir, 1995, 11, 660.
25 H. M. Patel, J. M. Didymus, K. K. W. Wang, A. Hirsch, A. Skiebe,
I. Lamparth and S. Mann, Chem. Commun., 1996, 611.
26 S. Ravaine, F. Le Peq, C. Mingotaud, P. Delhaes, J. C. Hummelen,
F. Wudl and L. K. Patterson, J. Phys. Chem., 1995, 99, 9551.
27 J. Y. Wang, D. Vaknin, R. A. Uphaus, K. Kiaer and M. Losche,
¨
T hin Solid Films, 1994, 242, 40.
28 M. Maggini, G. Scorrano and M. Prato, J. Am. Chem. Soc., 1993,
115, 9798.
The monolayer film at the air/water interface (also known
as a Langmuir, or Pockels–Langmuir film) was transferred
onto quartz (Spectrasil or Suprasil) slides by the Langmuir–
Blodgett (LB, or vertical transfer) technique, or transferred
onto highly oriented pyrolytic graphite (HOPG, Union
Carbide ZYA grade) by the Langmuir–Schaefer (LS, or quasi-
horizontal transfer) technique. The quartz slides were
29 J. Milliken, D. D. Dominguez, H. H. Nelson and W. R. Barger,
Chem. Mater., 1992, 4, 252.
30 T. Da Ros, M. Prato, F. Novello, M. Maggini and E. Banfi, J. Org.
Chem., 1996, 61, 9070.
Paper 7/05950G; Received 13th August, 1997
2400
J. Mater. Chem., 1997, 7(12), 2397–2400