Fluorescence quenching and size selective heterodimerization of a porphyrin adsorbed to gold and embedded in rigid membrane gaps

Article abstract of DOI:10.1021/ja991738n

Octaanionic meso-tetra(3,5-dicarboxylatophenyl) porphyrin 1 was adsorbed to gold electrodes at pH 12 and stayed there after repeated washing with 10^{- 2} M KOH. The fluorescence on sputtered gold surfaces amounted to 10% of the intensi

Full text of DOI:10.1021/ja991738n

J. Am. Chem. Soc. 1999, 121, 9539-9545
9539
Fluorescence Quenching and Size Selective Heterodimerization of a
Porphyrin Adsorbed to Gold and Embedded in Rigid Membrane Gaps
Werner Fudickar,† Jo1rg Zimmermann,‡,§ Laurent Ruhlmann,† Johannes Schneider,†
Beate Ro1der,‡ Ulrich Siggel,§ and Ju1rgen-Hinrich Fuhrhop*,†
Contribution from the Institut fu¨r Organische Chemie, der Freien UniVersita¨t Berlin, Takustrasse 3,
D-14195 Berlin, Germany, 2 Institut fu¨r Physik, der Humboldt-UniVersita¨t zu Berlin, InValidenstrasse 110,
10115 Berlin, Germany, and Max-Volmer Institut fu¨r physikalische Chemie, der Technischen UniVersita¨t
Berlin, Strasse des 17. Juni 135, D-10623 Berlin, Germany
ReceiVed May 25, 1999
Abstract: Octaanionic meso-tetra(3,5-dicarboxylatophenyl) porphyrin 1 was adsorbed to gold electrodes at
pH 12 and stayed there after repeated washing with 10-2 M KOH. The fluorescence on sputtered gold surfaces
amounted to 10% of the intensity observed on an organic subphase. Addition of 10-6 M aqueous solutions of
the manganese(III) complexes of an isomer mixture of tetracationic â-tetraethyl-â′-tetrakis(1-methyl-4-
pyridinium)- and meso-4-(1-methyl-4-pyridinium)phenyl porphyrins 2 and 4 at pH 12 quenched the fluorescence
quantitatively. Visible spectroscopy proved that the amount of porphyrin 1 on the gold surface had not changed.
The octaanionic porphyrin 1 was then embedded in a membrane by self-assembly of a bolaamphiphile containing
two secondary amide groups. Two hydrogen bond chains rigidify such a monolayer. The emission of porphyrin
1 remained after the self-assembly process. 1 was now localized on the bottom of a rigid membrane gap. Its
fluorescence was again quantitatively quenched by the tetracationic manganese(III) porphyrinate 2, which fit
in with the membrane gap. A larger manganese(III) porphyrin with a phenyl spacer between the porphyrin
and methyl pyridinium rings could not enter, and no quenching was observed. The same experiment with a
more fluid membrane made of octadecanethiol showed no such discriminating effect. The entrapment of 1,2-
trans-cyclohexanediol within the “immobile” water volume of the membrane gap is also reported. Water-
soluble compounds have thus been separated within a 2 nm3 water volume from bulk water. So far, the membrane
pores with a porphyrin bottom resemble natural enzyme clefts.
Introduction
canethiol on a gold electrode. Substitution of the steroid by the
long-chain thiol in the second self-assembly process was
prevented by the steroid shielding of the gold sulfur bond. The
resulting ODT electrode coating with 7 Å wide hydrophobic
gaps was then applied in cyclic voltammetry of ferricyanide
ions in bulk water. Strong currents flew through the pores with
a steroid bottom. The ion transport through the membrane gaps
was, however, totally blocked after plunging the electrode for
a few hours in a 0.5 M solution of 1,2-trans-cyclohexanediol.
Interaction with bulk water for several hours did not remove
this blockade; the diol did not diffuse away. This effect was
explained with an immobilization of the water molecules by
local cocrystallization within the hydrophobic Ångstro¨m gap.2
Computer models5,6 for the water structures on hydrophobic
surfaces came to the conclusion of “immobile” layers being 3-4
molecules thick. The issue of how fast such water molecules
exchange with the bulk was, however, not addressed. The effects
of solutes fitting the water arrangement in clusters have not even
been thought of in any of the surface-water models.
Biological organisms select hydrophobic clefts on enzyme
surfaces or within membrane proteins as sites for chemical
reactions. Light- and redox-active sites often contain metal-
loporphyrins. We report here a simple model of such a
hydrophobic reaction center with a porphyrin at the bottom of
a rigid membrane gap. Both the porphyrin and the membrane
are attached to sputtered gold surfaces. The experimental basis
for the development of the fixed membrane system is described
in four earlier publications1-4 and also relates to experience with
pores in vesicle membranes. It can be summarized as follows:
(i)Amphiphiles containing two secondary amide links at the
ends of the hydrophobic core form rigid monolayers on smooth
surfaces because two hydrogen bond chains prevent conforma-
tional changes. Membrane-soluble amines, for example, do not
permeate such a membrane.1
(ii) Ångstro¨m gaps in monolayers have been prepared by
successive deposition of a steroid hydrosulfide and octade-
* To whom correspondence should be addressed.
† Freien Universita¨t Berlin.
(iii) Meso-tetraphenylporphyrin-sulfonates or -carboxylates,
for example, 1 form stable heterodimers with substituted
â-tetraethyl-â′-tetrapyridinium porphyrins 2 in water. The
binding constant is in the order of 107 M-1. If one of the
porphyrins carries a copper(II) or manganese(III) central ion
and the other is a fluorescent free base, then the fluorescence is
‡ Humboldt-Universita¨t zu Berlin.
§ Technischen Universita¨t Berlin.
(1) Bo¨hme, P.; Hicke, H.-G.; Bo¨ttcher, C.; Fuhrhop, J.-H. J. Am. Chem.
Soc. 1995, 117, 5824.
(2) Fuhrhop, J.-H.; Bedurke, T.; Gnade, M.; Schneider, J.; Doblhofer,
K. Langmuir 1997, 13, 455.
(3) Endisch, C.; Fuhrhop, J.-H.; Buschmann, J.; Luger, P.; Siggel, U. J.
Am. Chem. Soc. 1996, 118, 6671.
(4) Ruhlmann, L.; Nakamura, A.; Vos, J.; Fuhrhop, J.-H. Inorg. Chem.
1998, 37, 6052.
(5) Vossen, M.; Forstmann, F. J. Chem. Phys. 1995, 101, 2379.
(6) Gompper, G.; Hauser, M.; Kornyshev, A. A. J. Chem. Phys. 1995,
101, 3378.
10.1021/ja991738n CCC: $18.00 © 1999 American Chemical Society
Published on Web 10/04/1999

9540 J. Am. Chem. Soc., Vol. 121, No. 41, 1999
Fudickar et al.
completely quenched by a slight excess of the metalloporphy-
rin.3,4 This phenomenon provides an easy method to determine
the accessibility of porphyrin A on the bottom of a membrane
gap by porphyrin B.
Scheme 1
(iv) Several fluorescent probes have been applied to charac-
terize the closing and reopening of pores in vesicle mem-
branes.7-10 Edge amphiphiles usually formed domains enclosing
a water-filled ion pore. Fitting hydrophobic stoppers then
interrupted the ion flow. In our case (see (ii) and (iii)), however,
the water within the hydrophobic gap is immobilized for
ferricyanide and porphyrin transport by a molecule which
“freezes” the water in the pore.
We report here on the adsorption of the anionic porphyrin 1
to the surface of solid gold and its remaining fluorescence.
Heterodimerization with 2 or 4 containing a manganese(III)
central ion again leads to fluorescence quenching of 1 on naked
gold. After its embedding in a rigid monolayer made of the
long-chain thiol 3 (Scheme 1), only the small porphyrin finds
its way to 1. The larger porphyrin 4 is rejected by the pore. A
fluid ODT monolayer is also applied for a comparison. It does
not prevent the heterodimerization of 1 and 4.
Experimental Section
Syntheses. The isomer mixture of two â-Tetraethyl-â-tetrakis(1-
methyl-4-pyridinium) porphyrinatomanganese(III) pentachlorides 2
containing the isomers I and III (III is given in Scheme 1 as mayor
component) and the diamidethiol 3 were prepared as reported in
references 11-13. Porphyrin 1 was prepared from 5-formyl isophthalic
ester. Its synthesis is described first.
5-Hydroxymethylisophthalic Acid Diethylester.11 To a solution
of 5 g (16.7 mmol) of 1,3,5-benzenetricarboxylic acid triethylester in
250 mL of absolute THF was added 4 mL of LiBH4 (2 M solution in
THF) dropwise under a nitrogen atmosphere. The solution was refluxed
for 30 min to complete the reaction. After the solution was cooled, 20
mL of water was added at 0 °C. Sulfuric acid (5%) was added dropwise
until a clear solution was obtained. The organic solvent was removed
by evaporation, and the aqueous residue was extracted twice with 200
mL of ether. The separated organic phase was evaporated and the
residue redissolved in 50 mL of hot ethanol. Unreacted benzenetricar-
boxylic acid triethylester crystallized spontaneously after cooling and
was removed by filtration. The ethanol solution containing the desired
monoacid and small amounts of di- and trialcohol was divided in the
three corresponding fractions by chromatography on a silica gel column
(hexane/ethyl acetate 1:1). 5-Hydroxymethylisophthalic acid diethylester
1
(2.55 g) was obtained as a white solid (10.2 mmol, 60%). H NMR
(250 MHz, CDCl3) δ 8.55 (1H, 2, 2-H, phenyl), 8.25 (2H, s, 4-H, 6-H,
phenyl), 4.80 (2H, s, CH2-benzyl), 4.45 (4H, q, CH2-ester), 1.95 (1H,
s, OH), 1.4 (6H, t, CH3-ester). Anal. calcd for C13H16O5 (252.0): C,
61.96; H, 6.34. Found: C, 61.49; H, 6.28.
5-Formyl-diethylisophthalate. Two grams (8 mmol) of 5-hy-
droxymethyl-diethylisophthalate was dissolved in 20 mL of glacial
acetic acid. Ceric ammonium nitrate (9.8 g, 18 mmol) dissolved in 20
mL water was added dropwise to the solution, which was then heated
to 70 °C. The color changed to deep red, then to pale yellow after 30
min. Heating was continued for another 30 min. The reaction mixture
was cooled and diluted with two volumes of water and extracted three
(7) Fuhrhop, J.-H.; Liman, U.; Koesling, V. J. Am. Chem. Soc. 1988,
110, 6840.
(8) Henke, C.; Steinem, C.; Janshoff, A.; Steffan, G.; Luftmann, H.;
Sieber, M.; Galla, H.-J. Anal. Chem. 1996, 68, 3158.
(9) Fyles, T. M.; van Straaten-Nijenhus, W. F. ComprehensiVe Supramo-
lecular Chemistry; Pergamon: Oxford, U.K., 1996; vol 10, p 53.
(10) Odashima, K. AdV. Supramol. Chem. 1997, 41, 211.
(11) Leon, J. W.; Jeffrey, W. L.; Kawa, M.; Fre´chet, M. J. J. Am. Chem.
Soc. 1996, 118, 8847.
(12) Cymerman-Craig, J.; Loder, J. W. J. Chem. Soc. 1956, 100.
(13) Bu¨cher, H.; Drexhage, K. H.; Fleck, M.; Kuhn, H.; Mo¨bius, D.;
Scha¨fer, F. P.; Sondermann, J.; Sperling, W.; Tillmann, P.; Wiegand, J.
Mol. Cryst. 1967, 2, 199.
times with ether. The combined organic layers were washed with 1 N
sodium bicarbonate. The organic layer was dried, and the solvent was
removed by distillation to give 1.8 g of white solid (yield 90%). Mp
68 °C. 1H NMR (250 MHz, CDCl3) δ 10.2 (1H, s, CHO), 8.85 (1H, s,
Ph), 8.65 (2H, s, Ph), 4.45 (4H, q, CH2), 1.45 (6H, t, CH3). MS (EI)
m/z ) 250 (33, [M]+), 205 (100, [M-EtO]+), 177 (13, [M-EtOOC]+).
meso-5,10,15,20-Tetrakis-(3,5-diethoxycarbonylphenyl)porphy-
rin. A 1 L three necked flask fit with reflux condenser and nitrogen
inlet port was filled with 600 mL of dichloromethane. 5-Formyl-

Fluorescence Quenching and Heterodimerization
J. Am. Chem. Soc., Vol. 121, No. 41, 1999 9541
diethylisophthalate (1.6 g, 6.4 mmol) and pyrrole (470 µL, 0.42 g, 6.4
mmol) were added, and the solution was stirred at room temperature
under a slow stream of nitrogen. After 15 min, BF3 etherate (0.34 mL
in 10 mL of dichloromethane) was added and the reaction vessel was
kept in the dark. When 1.18 g of p-chloranil was added to the reaction
mixture after 1 h, the color of the mixture turned to deep violet. After
standing for another 30 min, the solution was concentrated to 20 mL
by rotary evaporation and 10 g of silica was added. The slurry was
dried to afford a dark powder that was poured onto the top of a
chromatography column of silica. The column was washed with a
mixture of dichloromethane/ethyl acetate (3:1). After the first yellow
fraction was disgarded, the major red fraction was again chromato-
graphed with the same solvent mixture (10:1). The violet fraction from
hexane/ethyl acetate was crystallized. Yield 0.5 g (0.42 mmol, 26%).
1H NMR (250 MHz, CDCl3) δ 9.25 (4H, s, Ph), 9.05 (8H, s, Ph), 8.60
(8H, s, pyrrole), 4.45 (16H, q, CH2), 1.25 (24H, t, CH3), -2.80 (2H, s,
NH). MS (EI) m/z ) 1190 (35%, [M]+) Anal. calcd for C68H62N4O16
(1190,74): C, 68.53; H, 5.24; N, 4.70. Found: C, 68.10; H, 5.44; N,
4.44.
meso-5,10,15,20-Tetrakis-(3,5-dicarboxylatophenyl)porphyrin, 1.
A mixture of 0.3 g (4.2 × 10-4 M) of (meso-Tetrakis-(3,5-diethoxy-
carbonylphenyl)porphyrin), 2.0 g of potassium hydroxide, 10 mL of
water, and 100 mL of methanol was refluxed for 6 h. The reaction
mixture was acidified to pH 1 with HCl, and the precipitate was filtered
off. 1 (0.4 g) as a dark green solid was obtained: MS (FAB-) m/z )
966 (1%, [M-]). UV-Vis (H2O, pH ) 12), λmax nm (ꢀ M-1 cm-1) )
414 (446200), 516 (18600), 553 (1200), 580 (800).
meso-5,10,15,20-Tetrakis-4-(2-pyridyl)phenylporphin. p-2-Py-
ridylbenzyldehyde12 (2.3 g, 12.5 mmol) was dissolved in 60 mL of
propionic acid, and 0.83 g (12.5 mmol) pyrrole was added. After being
refluxed for 1 h, the solvent was evaporated and the residue washed
with DMF: 0.65 g (0.71 mmol; 23%) purple crystals were obtained.
1H NMR (250 MHz, CDCl3) δ 8.95 (8Η, s, Por), 8.8 (4H, d, Py), 8.4
(16H, m, Phe), 8.1 (4H, d, Py), 7.9 (4H, dd, Py), 7.4 (4H, dd, Py),
-2.65 (2H, s, Por).
meso-5,10,15,20-Tetrakis(2-pyridyl-5-phenyl)porphyrinatoman-
ganese(III) acetate. Tetrakis(2-pyridyl-4-phenyl)porphin and 88 mg
manganese(II)acetate (230 mg, 0.25 mmol) (36 mmol) was dissolved
in 50 mL of pyridine and refluxed for 4 h. The solvent was then
removed by evaporation, and the crude product was chromatographed
with chloroform/methanol (9:1) on a column of neutral alumina. Yield
150 mg (0.14 mmol, 58%) dark solid. MS (EI) m/z ) 975 (100%,
[M-acetate]+).
an average molecular weight of 8500-11000. The spin coating was
not dried. Porphyrin 1 was first transformed into the corresponding
acid chloride by treatment with SOCl2 at 60 °C for 50 min in order to
provide a solvent-soluble porphyrin, which would decompose to the
acid on the wet surface and bind as ammonium salt. Water could not
be used for self-assembly because PAA would have been solubilized.
The moist PAA-coated glass plate was then immersed into the
dichloromethane solution of the ocataacid chlorite ov 1 for 20 min.
The plate was then washed carefully with dichloromethane, and the
noncoated side was wiped off with a tissue.
Fluorescence Measurements. Steady-state fluorescence of porphyrin
monolayers on gold was determined with a cooled CCD matrix with a
mounted spectrometer (Oriel L. O. T. Instaspec IV). The sample was
oriented on a 5-axis positioning system (Fostec DC 300). Excitation
occurred by an Ar+ laser (100 mW, 30 µm spot size on sample) at
514.5 nm under an incident angle of 45°. Emitted light was collected
perpendicular to the gold electrode surface. Further details are described
in ref 13.
Quenching Experiments. The prepared electrodes were first placed
in a quartz cuvette (20 mL) filled with 10 mL of aqueous potassium
hydroxide (pH 12). A 10-4 M aqueous solution (100 µL) of quencher
2 was then added. For the quenching experiments with porphyrin 4,
larger quantities, namely, 30, 50, and 100 µL, of an aqueous 10-4
M
solution of 4 was added. No fluorescence quenching was observed
(Figure 3). In a control experiment, 2 × 100 µL of porphyrin 2 was
added and quantitative fluorescence quenching occurred. In the case
of the octadecanethiol membrane, 100 µL of the quencher 4 (10-4 M)
was added in the same manner.
The Monte Carlo Simulation. The calculations were performed
using Mathematica 3.0 (Wolfram Research, Inc.).
Cyclic Voltammetry. Voltammograms (CVs) were recorded ex situ
in an aqueous solution of 1 M KCl and 1 × 10-3 M K3[Fe(CN)6]. A
three electrode potentiostat (Versastat, EG & G, Princeton, NJ) was
used. The circular gold electrode with the membrane sample had a
surface area of 0.5 cm2, platinum wire was used as counter electrode,
and an SCE electrode was used as reference. Several sweeps were taken
with each sample. Selected closed sweeps are shown in Figure 5b. The
membrane-coated electrodes were treated with 0,1 M cis- and trans-
cyclohexanediol solutions for 24 h and then splashed with water. The
electrodes covered with a monolayer of porphyrin 1 (see SAM
procedure) showed no change for the CV curve of ferricyanide as
compared to the naked electrode.
Results
meso-5,10,15,20-Tetrakis(1-methyl-2-pyridinium-yl-5-phenyl)por-
phyrinato-manganese(III) pentakis(trifluoromethanosulfonate) 4.
Tetrakis(2-pyridyl-4-phenyl)porphyrinatomanganese(III) acetate (100
mg, 0.1 mmol) was dissolved in 10 mL of DMF, and 0.13 g of methyl
trifluoromethanesulfonate was added at 0 °C via syringe. The mixture
was stirred at room temperature for 1 h, and the solution was cooled
overnight to -18 °C, filtered, and washed with cold water. Yield 140
mg of black solid (80%). UV-Vis (H2O, pH ) 2), λ max nm (ꢀ M-1
cm-1) 380 (37000), 401 (36000), 468 (92000), 513 (4900), 564 (8100),
597 (6100); MS (FAB+) m/z ) 990 (0.36%; [M-3CH3]+).
For a porphyrin lying flat on a gold surface, one may expect
total fluorescence quenching because efficient energy transfer
from the excited dye to the metal surface has been reported for
other dyes on metal surfaces.13 Enhanced fluorescence has,
however, also been observed for dyes adsorbed on rough metal
surfaces,14-17 where the enhancement of fluorescence is due to
the increased local fields near the rough surface. The quenching
by energy-transfer processes is more than overcompensated at
frequencies near the plasma resonance frequency and at an
appropriate distance between dye and surface. Only experiments
could tell whether we could apply a fluorescence test for the
characterization of membrane gaps.
UV/Vis Spectroscopy on Gold Surfaces. UV/Vis absorption spectra
on gold subphases were measured with a Lambda 16 spectrometer
connected with a light conductor.
SAM Procedures. Glass plates (2.5 × 1.5 cm) with depositions of
first a 20 Å layer of chromium and then 200 nm of polycrystalline
gold were used for UV/vis and fluorescence measurements. The gold
electrodes were at first cleaned with piranha solution for 10 s and then
rinsed with water. Afterward they were exposed to a 1 mM solution of
porphyrin 1 at pH 12. After 4 days the electrodes were rinsed with
aqueous sodium hydroxide solution (pH 12) and finally with neutral
water. The dried gold electrodes covered with 1 as described above
were immersed in a 10-3 M solution of the mercaptodiamide 3 in
dichloromethane or ethanol for 24 h. They were washed twice with
ethanol. The same procedure was also followed with octadecythiol
(ODT) instead of the mercaptoamide 3.
We used the meta-tetraphenyl-octacarboxylate-porphyrin18,19
1 for the formation of a porphyrin monolayer adsorbed from
water at pH 12 on gold surfaces. The monolayer luminescence
was investigated using an Ar+ (514 nm) laser for excitation.20
(14) Naujok, R. R.; Duevel, R. V.; Corn, R. M. Langmuir 1993, 9, 1771.
(15) Pope, J. M.; Tan, Z.; Kimbrell, S.; Buttry, D. A. J. Am. Chem. Soc.
1992, 114, 10085.
(16) Wokaun, A.; Lutz, H.-P.; King, A. P.; Wild, U. P.; Ernst, R. R. J.
Chem. Phys. 1985, 79, 509.
(17) Alivisatos, A. P.; Waldeck, D. H.; Jarris, C. B. J. Chem. Phys. 1985,
82, 541.
(18) Niedballa, U.; Gries, H.; Conrad, J.; Hofmann, S.; Weinmann, H.-
J.; Speck, U. German Patent DE 3809671 A1, 1989.
(19) Patel, B. R.; Suslick, K. S. J. Am. Chem. Soc. 1998, 120, 11802.
Adsorption of Porphyrin 1 on PAA Coated Glass Slides. Thin
films from polyallylamin hydrochloride (PAA) were produced by spin
coating of a 2% aqueous solution of PAA (Aldrich Chem. comp.) with

9542 J. Am. Chem. Soc., Vol. 121, No. 41, 1999
Fudickar et al.
Figure 1. (a) Fluorescence of porphyrin 1 on sputtered gold and its
quenching after addition of 10-5 M manganese(III) porphyrinate 2.
Insert: time course of fluorescence quenching. (b) Model of the
heterodimerization on the gold surface.
The typical porphyrin fluorescence bands were observed at 650
and 725 nm. The error in repetitive experiments under identical
conditions was in the range of (10%. The fluorescence was
hardly measurable for mica-smooth21 gold but much stronger
on sputtered22 gold with a rougher surface. The following
experiments were therefore only performed on sputtered gold.
The fluorescence of a self-assembled monolayer of the same
porphyrin 1 on an organic subphase, namely, poly(allylamine)
on glass, was more intense by a factor of about 10. Although
the gold surface has thus a distinct quenching effect, the
measured emission (Figure 1) was more than sufficient for the
quenching experiments with manganese porphyrin 2. In addition
to fluorescence experiments, we also measured the absorption
of porphyrin 1 on gold with a light fiber spectrometer. Porphyrin
1 showed a broad Soret band at 425 nm with a shoulder at 470
nm. A first washing procedure with sodium hydroxide removed
the weakly adsorbed porphyrin. After that two dippings of the
electrode for one minute into aqueous sodium hydroxide at pH
12 did not change the fluorescence or absorption intensities
anymore. Although alkyl carboxylates usually do not bind to
gold, four carboxyl groups in the same molecular plane are
obviously not removed from the gold surface. The fact that
washing at pH 12 did not change the absorption intensity at
425 nm is taken as strong evidence for a porphyrin monolayer.
Multilayers are removed by the first treatment because 1 is well
soluble in sodium or potassium hydroxide solutions above pH
11.
Figure 2. (a) Fluorescence of porphyrin 1 embedded in a membrane
made of 3 and quenching after addition of manganese porphyrinate 2.
(b) Model of the heterodimerization within the rigid membrane gap.
(c) Calculated Monte Carlo model of the porphyrin 1 distribution (9)
in the monolayer made of 3 (O).
the self-assembled monolayer of 1 was plunged into a 10-5
M
solution of 2 for 10 seconds, the fluorescence of 1 was
completely quenched (Figure 1). Visible spectroscopy of the
same sample on gold still showed the 425 nm Soret band with
80% of its original intensity together with a more intense 488
nm band of the manganese porphyrin. Since a 20% loss of Soret
band intensity was also observed upon heterodimerization in
aqueous solution, we believe that practically no porphyrin was
removed during this operation. The quenching experiment thus
worked as well on a gold surface as in an aqueous environment
(Figure 1a,b).
The porphyrin 1 molecules on gold covered about 50% of
the surface. This approximate number was deduced from Monte
Carlo simulation, where we assumed that only one porpyrin
layer would adsorb and that each porphyrin with a diameter of
2.2 nm would stick to the original statistic orientation and not
reorient to form a two-dimensional crystal (Figure 2c). Thus in
every time-step of the simulation one square molecule was
randomly placed on the surface and could successfully be
adsorbed only when it had no overlap with already adsorbed
molecules at earlier times. The justification of this assumption
comes from the observation that, after 4 days of self-assembly
and one washing procedure, the observed emission intensities
The manganese(III) complex of â-tetraethyl-â-tetrapyridinium
porphyrin 2 was chosen as a quencher because its 488 nm Soret
band in the adsorbed state separated well from the 425 nm
absorption maximum of 1. 1 and 2 could therefore be easily
differentiated in UV/vis spectra. When the gold electrode with
(20) Korth, O.; Hanke, Th.; Ru¨ckmann, I.; Ro¨der, B. Exp. Tech. Phys.
1996, 41, 25.
(21) Wagner, P.; Hegner, M.; Gu¨ntherod, H. J.; Semenza, G. Langmuir
1995, 11, 3867.
(22) Whitesides, G. M.; Bain, C. D. Angew. Chem. 1989, 101, 533.

Fluorescence Quenching and Heterodimerization
J. Am. Chem. Soc., Vol. 121, No. 41, 1999 9543
were essentially always the same and then corresponded to
approximately 10% of the fluorescence of 1 on poly(allylamine)-
coated glass. The same reproducibility was observed in UV/vis
spectra. A 10-3 M ethanolic solution of diamide 3 was then
applied for 24 h for embedding the porphyrin in a rigid lipid
monolayer. Less than 10% of the porphyrin was lost by
substitution, as determined by both fluorescence and absorption
measurements. The successful embedding again points first to
the presence of a strongly adsorbed porphyrin monolayer only.
Bulk layers should not survive an ethanol treatment for several
hours. The gold electrode covered by the perforated membrane
was then again plunged into a 10-6 M solution of manganese
porphyrin 2. Porphyrin 2 found its way to the porphyrins at the
bottom of the membrane gap much more slowly than to the
same porphyrin on naked gold. Fluorescence decreased to half
of its original value within 6 min, and quenching was quantita-
tive only after a period of about 30 min (Figure 1c,d). Porphyrin
2 thus finds porphyrin 1, which is fixated on the bottom of the
well and 2 nm away from the membrane surface amides. No
charge interaction between the water dissolved and gold-bound
porphyrins can occur as long as 2 does not penetrate deeply
into the pore.
The quenching experiments on gold surfaces were then
repeated with the tetracationic manganese(III) porphyrinate 4
with phenyl spacers between the porphyrin and pyridinium rings.
The porphyrin diameter was now 35 Å instead of 24 Å. 4 still
quenched the fluorescence of porphyrin 1 lying on naked gold
surfaces, because the ortho-pyridinium nitrogens are close to
the meta carboxylate groups of 1 in heterodimers. It had no
quenching effect whatsoever on the same porphyrin on the
bottom of a rigidly built well made of bolaamphiphile 3 (Figure
3).
In contrast to the gaps in rigid monolayers made of diamide
3, a gap in a monolayer made of octadecanethiol (ODT) should
have only very limited discriminating power, since the edges
of the membrane pores are not defined. The octadecyl chains
should bend toward the porphyrin and thereby open much wider
gaps (Figure 4b). Addition of the large manganese(III) porphyrin
showed indeed total fluorescence quenching after a few minutes
(Figure 4).
Finally we tested whether the rigid, 20 Å wide membrane
gaps around the porphyrins reacted in the same way with the
ferricyanide/cyclohexanediol system as the much narrower
steroidal gaps2 described in the Introduction. It was found that
the current in cyclic voltammograms was practically identical
for the naked gold electrode (Figure 5b, curve 1) and the same
electrode covered with a porphyrin monolayer (not shown). At
first sight a current loss of about 30% was observed, when the
electrodes had been exposed to millimolar porphyrin 1 solutions
for 4 days. After these electrodes were rinsed with pH 12 water
(NaOH), the current loss, however, became negligibly small.
Such an electrode, which still showed strong porphyrin fluo-
rescence, was then immersed in a 10-3 M dichloromethane
solution of the mercaptodiamide 3 for 24 h, washed, and then
again used for cyclic voltammetry of ferricyanide. The current
decreased by about 30%, and the peaks shifted apart (Figure
5b, curve 2). The gold electrode covered with the porphyrin 1
embedded in a monolayer of 3 was then plunged for 24 h into
a 0.1 M solution of 1,2-trans-cyclohexanediol and washed
thoroughly with water, and the cyclic voltammogram measured.
No current leaked through (Figure 5b, curve 3). This situation
did not change for several hours in the presence of bulk water
containing only ferrocyanide and KCl, but no cyclohexanediol.
We then added a 10-5 M solution of manganese porphyrin 2.
Figure 3. (a,b) Heterodimerization of porphyrin 1 with manganese
porphyrinate 4 on a naked gold surface leads to fluorescence quenching
after addition of a 10-5 M aqueous solution of 4 (pH 12). (c,d) The
large manganese(III) porphyrinate 4 cannot reach porphyrin 1 on the
bottom of the rigid membrane made of 3. No fluorescence quenching
is observed.
No fluorescence quenching whatsoever was observed (Figure
5a). The same experiment using the cis-diastereomer of 1,2-
cyclohexanediols showed no blocking effect whatsoever. It was
also not observed when ruthenium dichloride was used instead
of ferricyanide. Treatment with HCl (pH 2) or ethanol (g20%)
opened the cyclohexanediol-blocked pores within several min-
utes.

9544 J. Am. Chem. Soc., Vol. 121, No. 41, 1999
Fudickar et al.
Figure 4. (a) Fluorescence of porphyrin 1 on gold embedded in an
octadecylthiol (ODT) membrane and fluorescence quenching by a 10-5
M solution of manganese porphyrin 4. Insert: time course of
fluorescence quenching. (b) Model of the heterodimerization in the fluid
ODT matrix.
Discussion
The results described in this paper are related to early
experiments of Sagiv, who has shown that self-assembled
octadecyl trichlorosilane monolayers on glass conserve gaps for
cyanine dyes with long alkyl chain substituents, when the latter
are reversibly extracted with chloroform.23,24 Five cyanine dyes
were applied that differed in length by 4.8 Å and were diluted
by octadecyl silanes in a molar ratio of ∼1:25. In our case the
difference in width of porphyrins 1 and 4 is about 11 Å and the
molar ratio 1/ODT is also about 25 as estimated from the
molecular cross sections and the assumption that ∼50% of the
surface is covered by the porphyrin (Figure 2c). In so far both
experimental setups are comparable. Sagiv then found that both
the small dye that had produced the membrane gaps and the
larger dye would adsorb into the holes within 30-60 min. The
longer chromophore would, however, need more time. In
simultaneous competitive adsorption experiments, normalized
molar ratios of up to 2.5 favoring the “fitting” small cyanine
dye were determined. The situation was complicated by the
shape of the molecules, which would allow also perpendicular
orientations within the membrane gap and adsorption of more
than one molecule within a fluid membrane gap. Our results
with the fluid ODT membrane are in agreement with Sagiv’s
results, although the method of fluorescence quenching did not
allow the measurement of competitive adsorption.
Strong size discrimination, however, is not possible with
narrow cyanine dyes and, more important, fluid membranes.
Porphyrins with large surface areas bind more tightly to the
solid subphase, large differences in diameter in combination
with rigid membrane gaps allow specific size discrimination,
and fluorescence-quenching experiments allow a direct mea-
surement of reaction rates within membrane pores. Amide
hydrogen bond chains are obviously the easiest means to
transform soft membranes into concrete blocks. One such chain
is not good enough, as has been demonstrated earlier,1 but two
of them worked perfectly well. An enlargement of the counterion
Figure 5. (a) Fluorescence spectra before and after the addition of
manganese(III) porphyrinate 2 after treatment with 1,2-trans-cyclo-
hexanediol. (b) Cyclic voltammetry of ferricyanide ions (10-3 M; 1 M
KCl) using (1) the naked gold electrode or the membran-coated
electrode depicted in Figure 2c, before (2) and after (3) addition of
trans-1,2-cyclohexanediol. The CV curve of the electrode covered only
with porphyrin 1 after washing is practically identical to the one of the
naked electrode (not shown). (c) Model of the “cyclohexanediol-
immobilized“ filling the membrane gap.
diameter by 10 Å then leads to a complete inactivation of the
octaanionic porphyrin trap on the bottom of the membrane gap.
The experimental results show that several problems involved
with size-selective membrane gaps were solved in a surprisingly
simple manner:
(i) Sputtered gold provides enough surface distortions to allow
quantitative fluorescence-quenching experiments with adsorbed
porphyrins. The probable reason, namely, fluorescence enhance-
ment by local fields, has been mentioned already in the
beginning of the Results section.
(ii) Porphyrin 1 is bound tightly to the gold surface, although
the latter is not totally smooth. The solvents used for self-
assembly, namely, 10-2 M aqueous sodium or potassium
hydroxide solutions, do not remove it from there, and successive
treatments with an excess of manganese porphyrins 1 and 4 in
the same solvents did not diminish the concentration of 1 on
gold either. Porphyrin 1 obviously forms a scattered monolayer
on the gold surface and is held there strongly, perhaps by
hydrogen bonds of the carboxylates to a tightly adsorbed water
layer (see below).
(23) Polymeropoulos, E. E.; Sagiv, J. J. Chem. Phys. 1978, 69, 1836.
(24) Sagiv, J. J. Isr. Chem. 1979, 18, 346.

Fluorescence Quenching and Heterodimerization
J. Am. Chem. Soc., Vol. 121, No. 41, 1999 9545
(iii) Porphyrin 1 monolayers on sputtered gold surfaces also
survive a day long treatment with ethanol or dichloromethane
solution as was evident after the self-assembly of diamide 3.
Less than 20% loss of fluorescence and absorption intensity
were observed after 3 days.
(iv) The surface roughness indicated by fluorescence en-
hancement and AFM measurements inhibited neither total
fluorescence quenching on naked gold nor size recognition by
the rigid membrane gap. Fast quenching was observed on naked
gold, but quantitative quenching with the fitting porphyrin 2
diffusing into the membrane gap was slower by a factor of about
100. The sputtered gold surface thus behaved like a perfectly
smooth subphase. The total amount of the anionic porphyrin 1
was accessible to manganese (III) porphyrin 2.
extremely slow exchange of fluid water with ice-like water has
been reported. Not only does the effect of 1,2-trans-cyclohex-
anediol2 thus reach much farther than the supposed three
immobilized surface molecules in undisturbed surface layers
of water but also, more importantly, it is an unexpected and
unexplained long-term effect. At least six water molecules on
each wall do not equilibrate with bulk water in the presence of
the rigid, cocrystallizing solute for days. Thin layers of surface
water must thus become an integral part of monolayers in the
presence of solutes, which produce a linkage between hydro-
phobic walls and the entrapped water volume. This effect may
also be the basis of the stickiness of porphyrin 1 on gold in
aqueous phases as described under (iii) above. The four carboxyl
groups within one plane may stabilize the surface layer of water
and bind to it. After addition of organic solvents (ethanol,
dichloromethane), the cyclohexanediol is removed from the
membrane gaps. The surface-bound water should then also
equilibrate with the bulk water. The porphyrin dimer on the
bottom of the gap, however, does not dissolve quickly.
Outlook. The present paper shows how one can localize three
water-soluble molecules, namely, two porphyrin and a few
cyclohexanediol molecules, in a water volume of approximately
10 nm3. In this respect the rigid membrane gap resembles surface
clefts of enzymes. The most important task of enzyme clefts,
namely, to catalyze reactions between adsorbed substrates and
to release products, is, however, out of reach with alkyl chains
as only wall material. The cooperation of functional groups
pointing from the walls into the cleft is needed. Functionalization
of the membrane gaps described in this paper should indeed be
possible. We attempt to introduce trans-configured carbon-
carbon double bonds into the hydrophobic chains of bolaam-
phiphile 3 and then to functionalize them with water-soluble,
membrane-impermeable oxidants or from ready-made gaps.
Apart from this futuristic aspect, however, we have already a
means to introduce substrates for porphyrin-catalyzed reactions
into the water-filled hydrophobic holes, namely, reactive phe-
nols25 instead of cyclohexanediol. Furthermore the porous
membranes are currently transferred from flat electrode material
to the curved surfaces of colloidal particles. Routine fluorescence
measurements and solid-state NMR will then allow the study
of exchange reactions between bulk and entrapped water
volumes directly, and systems with a more dilute density of
membrane pores can be characterized.
We would like to stress that the details of the membrane
preparations described are all linked together and were opti-
mized. Several other porphyrins in addition to the tetraphe-
nylporphyrin meta-octacarboxylate 1 have been tried, for
example, tetra-meta isothiocyanato and carboxylato, meso-
tetrathiophene, and various meso-tetraphenol porphyrins. None
of them gave reproducible quenching results. Several thiols
containing two secondary amide groups and different second
headgroups in addition to SH have been used in attempts to
produce functional and thicker rigid membranes around 1. None
of these bolaamphiphiles was satisfactory, mainly because
solubility was too low and precipitation instead of self-assembly
occurred. Other subphases, particularly silicon wafers and glass,
also proved less satisfactory than gold, because the second
component in self-assembly was usually too aggressive. A
tendency to large losses of porphyrin 1 was observed. An
exception was an amino-derivatized glass surface, which will
be explored further.
The transference of the “immobile water concept” from the
7 Å steroid-based gap2 to the 24 Å porphyrin analogue also
deserves a comment. The 7 Å gap allows only for rows of 3
water molecules, if one assumes an average van der Waals radius
of 2.2 Å. For the porphyrin gap a row of 11 water molecules
would be needed to reach from one wall to the other. Model
calculations favor a number of up to three or four immobilized
water molecules. Ferricyanate transport is not inhibited at all
this immobilization. Only after treatment with trans-cyclohex-
anediol the transport is completely and, for many days,
irreversibly blocked. The cis-diastereomer with one axial
hydroxyl group has practically no blocking effect. This finding
of stereoselectivity plus the rapid destruction of the “ice-cube”
by ethanol make it very tempting to consider the formation of
a cocrystal of the edge-amphiphile cyclohexanediol with two
equatorial OH groups and hexagonal solid water. We could,
however, find no support for this model in the literature. To
the best of our knowledge, neither an extraordinary ordering
effect of cyclohexane or benzene diols for water nor an
Acknowledgment. This work was supported by the Euro-
pean Commission within the TMR network “Artificial Photo-
synthesis”, the Deutsche Forschungsgemeinschaft (SFB 312
“Vectorial Membrane Processes” and SFB 348 “Mesoscopic
Systems”), and the FNK of the Free University Berlin.
JA991738N
(25) Li, G.; Fuhrhop, J.-H. manuscript in preparation.