Self-Assembly of Zinc Porphyrins around the Periphery of Hydrogen-Bonded Aggregates That Bear Imidazole Groups

Article abstract of DOI:10.1021/jo9615458

This paper describes the preparation and characterization of four aggregates that are based on the rosette of derivatives of isocyanuric acid (CA) and melamine (M). These aggregates comprise a trismelamine, hub(M^Im)₃, that presents imidazole groups around its periphery; these imidazoles organize zinc tetraphenyl porphyrin (ZnTPP) by coordination of the imidazole to the zinc center. Aggregate 1 forms a single rosette upon mixing 1 equiv of hub(M^Im)₃ and 3 equiv of CA. Adding 3 equiv of ZnTPP yields 2. Aggregate 3 forms as a stacked bisrosette upon mixing 2 equiv of hub(M^Im)₃ and 3 equiv of bisCA. Adding 6 equiv of ZnTPP yields 4. The stoichiometries of aggregates 1-4 were obtained by titrating the trismelamines with CA and by titrating the aggregates with ZnTPP. The stoichiometry is defined as the ratio at which additional CA remains insoluble or additional ZnTPP appears as free ZnTPP in the ¹H NMR spectrum. Electrospray ionization mass spectrometry (ESI-MS) is compatible with the measured stoichiometries. The structures of these aggregates were determined using variable-temperature ¹H NMR spectroscopy; analogous structures were inferred for 5 and 6, the tert-butyl analogues of 1 and 2. The shapes of the traces from gel permeation chromatography (GPC) suggest that imidazole groups destabilize the aggregates when they are not involved in coordination to zinc; that is, the stability seems to be 6 ≈ 4 > 3 and 5 ≈ 2 > 1. A direct comparison of the relative stability of 1, 2, and 5 confirms the results of the GPC analysis: mixing 1 (hub(M^Im)₃·3CA) with the trismelamine component of 5 (hub(M)₃) leads to the formation of a 3:2 mixture of 5:1. Adding ZnTPP to this solution leads to a 3:2 mixture of 5:2 with free trismelamines remaining in solution: 1 is not observed. The results of UV/vis spectroscopy are consistent with the other spectroscopic and Chromatographic results and indicate that 3 equiv of ZnTPP are organized around the periphery of 2 and at least 4 equiv around the periphery of 4.

Full text of DOI:10.1021/jo9615458

8994
J . Org. Chem. 1997, 62, 8994-9000
Self-Assem bly of Zin c P or p h yr in s a r ou n d th e P er ip h er y of
Hyd r ogen -Bon d ed Aggr ega tes Th a t Bea r Im id a zole Gr ou p s
Eric E. Simanek, Lyle Isaacs, Xinhua Li, Clay C. C. Wang, and George M. Whitesides*
Department of Chemistry and Chemical Biology, Harvard University, 12 Oxford Street,
Cambridge, Massachusetts 02138
Received August 8, 1996X
This paper describes the preparation and characterization of four aggregates that are based on the
rosette of derivatives of isocyanuric acid (CA) and melamine (M). These aggregates comprise a
trismelamine, hub(MIm)3, that presents imidazole groups around its periphery; these imidazoles
organize zinc tetraphenyl porphyrin (ZnTPP) by coordination of the imidazole to the zinc center.
Aggregate 1 forms a single rosette upon mixing 1 equiv of hub(MIm)3 and 3 equiv of CA. Adding 3
equiv of ZnTPP yields 2. Aggregate 3 forms as a stacked bisrosette upon mixing 2 equiv of hub-
(MIm)3 and 3 equiv of bisCA. Adding 6 equiv of ZnTPP yields 4. The stoichiometries of aggregates
1-4 were obtained by titrating the trismelamines with CA and by titrating the aggregates with
ZnTPP. The stoichiometry is defined as the ratio at which additional CA remains insoluble or
1
additional ZnTPP appears as free ZnTPP in the H NMR spectrum. Electrospray ionization mass
spectrometry (ESI-MS) is compatible with the measured stoichiometries. The structures of these
aggregates were determined using variable-temperature 1H NMR spectroscopy; analogous structures
were inferred for 5 and 6, the tert-butyl analogues of 1 and 2. The shapes of the traces from gel
permeation chromatography (GPC) suggest that imidazole groups destabilize the aggregates when
they are not involved in coordination to zinc; that is, the stability seems to be 6 ≈ 4 > 3 and 5 ≈
2 > 1. A direct comparison of the relative stability of 1, 2, and 5 confirms the results of the GPC
analysis: mixing 1 (hub(MIm)3‚3CA) with the trismelamine component of 5 (hub(M)3) leads to the
formation of a 3:2 mixture of 5:1. Adding ZnTPP to this solution leads to a 3:2 mixture of 5:2 with
free trismelamines remaining in solution: 1 is not observed. The results of UV/vis spectroscopy
are consistent with the other spectroscopic and chromatographic results and indicate that 3 equiv
of ZnTPP are organized around the periphery of 2 and at least 4 equiv around the periphery of 4.
Ta ble 1
In tr od u ction
This paper describes research that is part of a program
whose objectives include the construction of large, soluble
aggregates from simple precursors using molecular self-
assembly. This program has focused on a family of
aggregates based on the hydrogen-bonded lattice formed
from isocyanuric acid and melamine (CA‚M).1 Our choice
of CA‚M reflects the necessity for (i) large numbers of
hydrogen bonds to provide a large enthalpy of formation
and balance the large, unfavorable entropy of formation,
(ii) molecules with easily functionalized sites, and (iii) a
structure for the aggregate that is highly symmetric to
simplify characterization. Aggregates 5 and 6 (Table 1)
are representative of early efforts. Here, we report the
incorporation of zinc tetraphenyl porphyrin‚imidazole
interactions into derivatives of 5 and 6.
There are four advantages to using zinc‚imidazole
interactions in addition to hydrogen bonds to generate
aggregates of high molecular weight. First, zinc
porphyrin‚imidazole interactions are strong (approxi-
mately 8 kcal mol-1).4 Second, zinc‚imidazole interac-
tions do not interfere with the hydrogen-bond interac-
X Abstract published in Advance ACS Abstracts, December 1, 1997.
(1) Whitesides, G. M.; Simanek, E. E.; Mathias, J . P.; Seto, C. T.;
Chin, D. N.; Mammen, M.; Gordon, D. M. Acc. Chem. Res. 1995, 27,
37.
(2) Seto, C. T.; Mathias, J . P.; Whitesides, G. M. J . Am. Chem. Soc.
1993, 115, 1321.
(3) Chin, D. N.; Gordon, D. M.; Whitesides, G. M. J . Am. Chem. Soc.
1994, 116, 12033. Chin, D. N.; Simanek, E. E.; Li, X.; Wazeer, M. I.
M.; Whitesides, G. M. J . Org. Chem. 1997, 62, 1891.
(4) Kadish, K. M.; Shiue, L. R.; Rhodes, R. K.; Bottomley, L. A. Inorg.
Chem. 1981, 20, 1274. Mashiko, T.; Dolphin, D. in Comprehensive
Coordination Chemistry; Wilkinson, G., Gillard, R. D., McCleverty, J .
A., Eds.; Pergamon Press: New York, 1987; Vol. 2, pp 813-898.
tions that provide the major structural elements forming
the central rosette. Third, zinc‚imidazole interactions
can be incorporated into aggregates based on CA‚M on
S0022-3263(96)01545-9 CCC: $14.00 © 1997 American Chemical Society

Self-Assembly of Zinc Porphyrins
J . Org. Chem., Vol. 62, No. 26, 1997 8995
Ta ble 2. Su m m a r y of th e Ch a r a cter iza tion of 1-4a
a
“Aggregate” is defined in Table 1. “Ion” refers to the stoichiometry of the line observed in ESI-MS. “Others” lists other major lines
seen in the mass spectrogram. The “imide region” of the NMR spectra refers to the region between 16.5 and 13.5; the scale of 3 is offset
by 0.5 ppm. The traces obtained by “GPC” are shown. The dashed traces (corresponding to 5 and 6) are included so that comparisons of
5 with 1 and 6 with 3 can be made efficiently. These dashed traces were collected separately and do not represent the result of coinjection
of aggregates. The hatched area corresponds to free ZnTPP. The shaded area is a molecular weight standard (p-xylene). An arrow is used
to identify the sharp leading edge of 3.
the periphery of the central rosette so that established
strategies for characterization can be used. Fourth,
working with zinc‚imidazole interactions allows us to
take advantage of a wealth of literature on self-assembly
based on the coordination chemistry of porphyrins.4-6
The goals of this work were to determine (i) the effect
of the imidazole group on the stability of hydrogen-
bonded aggregates in the absence of metal-containing
groups and (ii) the ability of the imidazole groups to form
stable zinc‚imidazole interactions. We use zinc tetraphe-
nylporphyrin (ZnTPP) as a coordinating group for four
reasons: it associates strongly with imidazole groups, its
coordination to imidazole groups can be followed spec-
trophotometrically, it is commercially available, and it
is well explored in the literature, notably by Sanders5
and Groves.6 The results presented in this paper are
summarized in Table 2.
Or ga n iza tion of Th is P a p er . The results section is
organized into three parts. The first describes syntheses
of the molecules and aggregates; the second discusses the
association of n-butylimidazole (bu-Im) with ZnTPP; the
third summarizes the characterization of aggregates 1-4,
organized by method of characterization so that com-
parisons of 1 and 3 with 2 and 4 can be made efficiently.
by adding one drop of methanol to a suspension of the
components in CH2Cl2, evaporating the resulting homo-
geneous solution to dryness, and redissolving the residue
in CD2Cl2. To prepare 2 and 4, the appropriate amount
(3 or 6 equiv) of ZnTPP was added to the solution of
aggregates 1 and 3.
Bu -Im ‚Zn TP P . The interaction of metal porphyrins
with imidazole groups is well documented.4 In our
system, visual inspection, UV/vis spectroscopy, and 1H
NMR spectroscopy suggest that ZnTPP‚bu-Im forms
rapidly on mixing ZnTPP and bu-Im.
Visu a l In sp ection a n d UV/vis Sp ectr oscop y. Upon
addition of bu-Im to a solution of ZnTPP, the color of the
solution changes from violet to blue. No color change
occurs when derivatives of CA or M lacking an imidazole
group are added; this observation suggests that the
interaction between imidazole and ZnTPP is specific. UV/
vis spectroscopy confirms these visual observations.
Figure 1 shows the UV/vis spectral titration for a fixed
concentration of ZnTPP (10 µM in CH2Cl2) while chang-
ing the concentration of bu-Im (0-60 µM). The Soret
band of ZnTPP shifts from 418 to 428 nm. Nonlinear
least-squares analysis of these data (monitoring at 418
nm) yields the binding constant for bu-Im with ZnTPP;
Ka ) 1.8 × 105 M-1 s-1. In addition to the changes in
the Soret band, the Q-band of ZnTPP at 547 nm weakens
and new bands appear at 565 and 603 nm upon addition
of bu-Im. These changes in the UV/vis spectrum do not
fit to a simple 1:1 binding event. Changes in the position
and intensity of the Soret band will, therefore, be most
useful in monitoring the aggregation events on the
periphery of the CA‚M rosette.
Resu lts a n d Discu ssion
Syn th esis. Scheme 1 summarizes the syntheses of the
components of 1-4. We prepared aggregate 1 by mixing
the components in CD2Cl2 followed by sonication and
brief heating at reflux. Synthesis of 3 was accomplished
1
1H NMR Sp ectr oscop y. The H NMR spectrum at
(5) Anderson, S.; Anderson, H. L.; Sanders, J . K. M. Acc. Chem. Res.
1993, 26, 469 and references therein.
(6) Lahiri, J .; Fate, G. D.; Ungashe, S. B.; Groves, J . T. J . Am. Chem.
Soc. 1996, 118, 2347 and references therein.
240 K suggests that ZnTPP forms a 1:1 aggregate with
bu-Im. On mixing solutions of ZnTPP and bu-Im, all of

8996 J . Org. Chem., Vol. 62, No. 26, 1997
Sch em e 1. Syn th esis of Hu b(MIm
Simanek et al.
a
)
3
a
Key: (a) phthaloyl dichloride, triethylamine; (b) (t-BuOCO)2O, THF; (c) tert-butylbenzyl bromide, diisopropylethylamine (DIPEA); (d)
7 and SOCl2, then (e) DIPEA and 14; (f) trifluoroacetic acid/CH2Cl2; (g) 1,3,5-benzenetricarbonyl chloride, DIPEA, THF; (h) H2NNH2 in
MeOH; (i) cyanuric chloride, DIPEA, THF; (j) NH3(g); (k) aminopropylimidazole, DIPEA, dioxane.
shown by titrations; structure, from 1H NMR spectros-
copy; molecular weights, as established using ESI-MS;
and stabilities, inferred from GPC. The incorporation of
a ZnTPP‚imidazole interaction allows color changes to
be monitored either by UV/vis spectroscopy or by eye.
Stoich iom etr y. To establish the ratio of hub(MIm
)
3
to CA (or bisCA), we slowly added the CA component into
a solution of CD2Cl2 containing hub(MIm)3. We monitored
this titration by 1H NMR spectroscopy and by visual
inspection. Both techniques indicate the stoichiometries
shown in Table 2. As the CA component is added, the
1H NMR spectrum of a solution of hub(MIm)3 in CD2Cl2
goes from broad and featurelesssexcept for solvent and
a broad t-Bu resonancesto sharp lines. The intensities
of these lines increased until 3 equiv of CA had been
added. Additional CA did not affect the spectrum: the
CA was insoluble and was visible as suspended powder.
1H NMR Sp ectr oscop y. Strong spectral correlations
1
involving three different aspects of the H NMR spectra
F igu r e 1. UV/vis spectral titration of ZnTPP (10 µM in CH2-
Cl2) with bu-Im (0-60 µM). The arrows indicate the increase
or decrease in absorbance of the indicated band as bu-Im is
added. The inset shows a plot of the concentration of bu-Im
versus absorbance at 418 nm used to determine the binding
constant.
of 1-4 lead us to the conclusion that 1 and 2 are struc-
turally analogous to 5 and that 3 and 4 are analogous
to 6.
Lin e Wid t h . The spectrum of hub(MIm
)
in the
3
absence of CA is featureless: a rolling base line showing
a line for solvent and a broad resonance corresponding
to the tert-butyl groups. Sharp linessobtained upon
addition of CAsestablish that discrete, nonrandom spe-
cies (1and 3) exist in solution. Similar behavior had been
observed previously for aggregates 5 and 6. Addition of
ZnTPP to solutions of 1 and 3 generates 2 and 4. The
1H NMR spectra of these aggregates are sharp and
indicate that well-defined aggregates exist in solution.
Hyd r ogen -bon d ed P r oton s. Sharp resonances be-
tween 16 and 13 ppm are due to the hydrogen-bonded
imide protons of CA. In DMSO-d6 these protons appear
at 9.5 ppm. The number of lines corresponding to
the aliphatic resonances of bu-Im shift upfield. The
pyrrole protons of ZnTPP shift from 9.0 to 8.9 ppm. At
room temperature, the aromatic resonances of the imi-
dazole broaden into the base line; they appear dramati-
cally shifted upfield at low temperature. No lines
corresponding to free bu-Im appear until the ratio of bu-
Im:ZnTPP becomes greater than 1; we infer that only a
1:1 complex, ZnTPP‚bu-Im, is formed. This observation
suggests that the off-rate for the ZnTPP‚bu-Im interac-
tion is slow on the NMR time scale.
Ch a r a cter iza tion of 1-4. Proof of structure for 1-4
comes from four primary sources: stoichiometry, as

Self-Assembly of Zinc Porphyrins
J . Org. Chem., Vol. 62, No. 26, 1997 8997
hydrogen-bonded protons establishes the symmetry of the
aggregate: for example, two lines are observed for the
C3-symmetric isomer of 5; six lines are observed for its
a range of ionization conditions: only under high voltages
(V > 90) and temperatures (T > 60 °C) was a small line
corresponding to free hub(MIm)3 observed.10 A single line
does not prove unambiguously that 1 and 3 are the only
stable species in solution, since ESI-MS measures the m/z
ratio for aggregates in the gas phase. These results do,
however, agree with those from 1H NMR spectroscopy
and reinforce the inference that a single species (1 and
3) is present.
The mass spectra of 2 and 4 contain more lines than
those of 1 and 3. Intense lines are observed at masses
corresponding to 2‚2Cl- and 4‚3Cl-, but additional lines
complicate the spectrogram. Some of these remaining
lines correspond to the loss of ZnTPP (Table 2). The loss
of ZnTPP is not surprising; even though the imidazole‚Zn
chelate interaction is stronger than a single hydrogen
bond it is probably the easiest bond to break in the entire
assembly since the hydrogen bonds of the rosette act in
concert. Observation of these aggregates, in fact, would
probably not be possible except for the fact that we are
using Ph4PCl as the charge carrier.10 Other linessmost
less than 10% intensitysare difficult to assign and may
result from aggregation of matrix (Ph4Cl) with itself or
molecules of aggregate.
1
C1-symmetric isomer. The imide regions of the H NMR
spectra for 1-4 are shown in Table 2. Aggregates 1 and
2, just like 5, show only two resonances in the imide
region of the 1H NMR spectrum. This spectrum is
consistent, in each case, with the presence of a single
aggregate of C3-symmetry. Aggregate 3, however, shows
12 resonances in the imide region of its 1H NMR
spectrum; two of high intensity and 10 (two more are not
readily identified as a result of coincidental overlap) of
lower intensity. This observation can be rationalized by
the presence of two aggregates, one of D3 symmetry, the
other C1 symmetric.3 The spectra collected previously for
6 showed similar patterns. In contrast, the imide region
of the 1H NMR spectrum of 4 shows only two resonances,
suggesting the presence of a single aggregate of D3
symmetry.
Dia st er eot op ic P r ot on s. Many methylene groups
shouldsand dosappear as sets of diastereotopic protons.
Each diastereotopic proton is easily identified as a
doublet, and several are of markedly different chemical
shift as a result of the relative proximity of the aromatic
rings of the trismelamine. In particular, the methylene
group of the tert-butyl benzyl substituent is diagnostically
valuable. In aggregates 1-4, these protons resonate at
around 4 and 5.5 ppm; the ∼ 1.5 ppm separation
indicates a well-defined structure and can be explained
by the presence of a nearby aromatic ring.
Gel P er m ea tion Ch r om a togr a p h y.11 GPC is a size
exclusion technique. The pores of the stationary phase
readily allow small molecules to enter while exluding
larger molecules. The direct result is that smaller
molecules take longer to pass through the column than
larger molecules. The stability of hydrogen-bonded ag-
gregates based on the CA‚M motif is reflected in the
extent of tailing of the peaks.11 Stable aggregates elute
from the column as sharp peaks. Tailing results from
the dissociation of the aggregates into soluble, slower
moving components. That is, while we expect that 1
should elute similarly to 5 based on the similarity of their
structures, the broad peak (lacking a well-defined leading
edge) observed in the trace (Table 2) suggests that 1 is
unstable under the conditions of GPC. The traces
suggest that 5 and 2 are of comparable stability and that
both are more stable than 1. Similarly, 6 and 4 are more
stable than 3. The traces for 2 and 4 show two peaks:
one peak coelutes with ZnTPP. It is unclear whether the
Assign m en t. The line widths of the 1H NMR spectra,
the intensity, multiplicity and location of the resonances
for the imide protons, and the presence and location of
diagnostic sets of diastereotopic protons for aggregates
1-4 and 5-6 are strikingly similar. The main differ-
ences are due to the presence of ZnTPP in aggregates 3
and 4, which leads to slight broadening of resonances and
a pronounced upfield shift in many resonances on the
periphery of the aggregate. The strong structural simi-
larity of the components of 1-4 and 5-6, when coupled
with these strong NMR correlations among these ag-
gregates, fully justifies the following structural assign-
ments. The number and intensity of resonances in the
1
imide region of the H NMR spectrum for 1 and 2 are
(9) The shifts and broadening observed in the 1H NMR spectra make
the upfield region of the spectra difficult to assign. We can easily
discern, however, that C3 symmetry of the aggregate is maintained
and that the structures of these imidazole-containing aggregates
consistent with a single aggregate existing as a pair of
enantiomers with C3 symmetry.8 Two isomers are ob-
served for 3, whereas a single isomer is observed for 4.
Using 5 and 6 as model compounds, the structural
assignment of these aggregates is straightforward. Ag-
gregate 3 exists as two isomeric forms: the major isomer
has D3 symmetry; the minor is C1 symmetric. Aggregate
4, however, exists as a single species in solution; we
assign D3 symmetry to this aggregate.
are very similar to those of
5 and 6. Assignment of all the
resonancessespecially those corresponding to the chelated imidazole
groupsare, therefore, tentative. The imide region of the spectrum,
however, is diagnostic of the symmetry and number of aggregates
present in solution.
(10) (a) Cheng, X. H.; Gao, Q. Y.; Smith, R. D.; Simanek, E. E.;
Mammen, M.; Whitesides, G. M. J . Org. Chem. 1996, 61, 2204-2206.
(b) Cheng, X. H.; Gao, Q. Y.; Smith, R. D.; Simanek, E. E.; Mammen,
M.; Whitesides, G. M. Rapid Commun. Mass Spec. 1995, 9, 312-316.
(c) Cheng, X. H.; Chen, R. D.; Bruce, J . E.; Schwartz, B. L.; Anderson,
G. A. Hofstadler, S. A.; Gale, D. C.; Smith, R. D.; Gao, J . M., Sigal, G.
B.; Mammen, M.; Whitesides, G. M. J . Am. Chem. Soc. 1995, 117,
8859-8860. The observation of zinc porphyrin‚imidazole assemblies
in the ESI-MS spectrogram was possible in this case due to the special
conditions employed. Such aggregates are expected to be difficult to
observe using ionization from protic solvents since the protonation of
the most basic sitesthe imidazolesleads to destruction of the ag-
gregate. The ESI-MS experiments described here use Ph4PCl as a
charge carrier; we are able to observe [M + Cl-] in the negative-ion
mode. The use of CH2Cl2 as solvent may also limit the extent to which
protonation occurs. The significant number of extra lines remains a
challenge to interpret. Further investigation of other alternatives to
ESI-MS. The spectrograms for 1 and 3 are very
simple: only one line (1‚Cl- or 3‚2Cl-) is observed over
(7) Melamines and derivatives of CA containing covalently tethered
porphyrins have been reported by a number of groups. See: Ichihara,
K.; Naruta, Y. Chem. Lett. 1995, 631. Drain, C. M.; Russell, K. C.; Lehn,
J .-M. Chem. Commun. 1996, 337. Drain, C. M.; Fischer, R.; Nolen, E.;
Lehn, J .-M. J . Chem. Soc. Chem. Commun. 1993, 243 and references
therein.
(8) All three arms twist in the same direction with respect to both
the core benzene-1,3,5-tricarbonyl and the melamine groups. A second
isomer with C1 symmetry has been observed in related structures and
is generated conceptually by rotation of the melamine about the
indicated spoke. The loss of C3 symmetry leads to a 3-fold increase in
the number of lines in the imide region of the 1H NMR spectrum. These
isomers have been investigated in: Simanek, E. E.; Wazeer, M. I. M.;
Mathias, J . P.; Whitesides, G. M. J . Org. Chem. 1994, 59, 4904.
H
+ as the ion source, including Ph4PCl, may lead to spectra with fewer
ambiguous lines.
(11) For a more complete description of our use of GPC see ref 1 or:
Mathias, J . P.; Seto, C. T.; Simanek, E. E.; Whitesides, G. M. J . Am.
Chem. Soc. 1994, 116, 1725.

8998 J . Org. Chem., Vol. 62, No. 26, 1997
Simanek et al.
faster eluting peak corresponds to intact 2 or a species
containing less than three ZnTPP. The position of the
peak in the GPC trace of 2 when compared with the
position of 5 suggests that the species that elutes is larger
than 5. That is, the species that elutes contains at least
one porphyrin. A hint of color in the eluent is consistent
with this belief.
Ultr a violet a n d Visible (UV/vis) Sp ectr oscop y.
The changes in the UV/vis spectrum that occur on
addition of bu-Im to a solution of ZnTPP in CH2Cl2 are
diagnostic for formation of a Zn‚imidazole interaction. We
have, accordingly, used this spectroscopic marker to
follow the association of ZnTPP with self-assembled
aggregates 1 and 3. The experimental design was
crucial; the extremely high extinction coefficient of the
Soret band for ZnTPP gives an upper limit for ZnTPP
concentration of 100 µM. Furthermore, a lower concen-
tration for aggregates 1 and 3 was set at 10 µM, since
aggregates based on the CA‚M lattice are stable at this
concentration.2 On the basis of these considerations, we
performed J ob plots to determine the stoichiometry of the
interaction of 1 and 3 with ZnTPP.
The results of J ob titrations of 1 and 3 (constant total
concentration [1] + [3] ) 100 µM in CH2Cl2) are shown
in Figure 2a. As the concentrations of 1 and ZnTPP are
changed, the relative intensities of the Soret bands at
418 and 427 nm are affected; these changes allow
quantitation of the number of ZnTPP molecules that bind
to 1 at this concentration. The inset (Figure 2a) shows
a J ob plot of the mole fraction of ZnTPP (defined as
[ZnTPP]/([ZnTPP] + [1])) versus absorbance at 427 nm.
This plot shows a maximum when the mole fraction of
ZnTPP is between a ratio of 1:ZnTPP of 1:2 and 1:3. This
1
result, when combined with the results of the H NMR
experiments described above, is strong evidence for the
formation of the 1:3 aggregate 2.
Similar UV/vis J ob experiments were performed using
the bisrosette 3 and ZnTPP. Figure 2b shows the UV/
vis spectra recorded for different mole fractions of 3 and
ZnTPP at a constant total concentration of 100 mM.
Once again, the intensities of the Soret bands of free
ZnTPP and chelated ZnTPP change as a function of mole
fraction. The inset (Figure 2b) shows a plot of the mole
fraction of ZnTPP versus absorbance at 427 nm. This
plot shows a maximum at a mole fraction of ZnTPP of
0.80. This corresponds to a 1:4 molar ratio of 3:ZnTPP,
suggesting that aggregate 4 does not contain a full six
ZnTPP molecules organized around its periphery. This
F igu r e 2. (a) UV/vis spectral titration at
a fixed total
concentration (100 µM) of ZnTPP and 1 in CH2Cl2. The inset
shows a J ob plot of the mole fraction of ZnTPP versus
absorbance at 427 nm. (b) UV/vis spectral titration at a fixed
total concentration (100 µM) of ZnTPP and 3 in CH2Cl2. The
inset shows a J ob plot of the mole fraction of ZnTPP versus
absorbance at 427 nm.
indistinguishable from a trace of ZnTPP. These trends
in stability are born out in competition experiments
described below.
We envision that the destabilizing effect of the imida-
zole group results from their ability to form hydrogen
bonds with both the NH groups of melamine and isocya-
nuric acid groups. For the aggregate to form, the
hydrogen bonds formed with imidazole groups must be
broken at an enthalpic cost. As a result, the net enthalpy
of formation of the hydrogen bonds between isocyanuric
acid and melamine is decreased and the overall enthalpy
of formation of the aggregate is also decreased.
1
result is not inconsistent with the results of H NMR
experiments described above for the following reasons:
(1) the error in J ob plots can be substantial, especially
when dealing with higher order complexes, and (2) this
experiment was performed at a total concentration of 100
1
µM, far below the 5 mM concentration employed for H
NMR. It is plausible that the fifth and sixth equivalent
of ZnTPP require higher concentrations for complex
formation as a result of unfavorable steric interactions.
Im id a zole Gr ou p s Not In volved in Ch ela tion
Decr ea se th e Sta bility of Hyd r ogen -Bon d ed Ag-
gr ega tes. GPC reveals that 5 is more stable than 1 and
6 is more stable than 3. Adding ZnTPPsto form 2 and
4sappears to increase the stability of the aggregates.
That is, these data suggest that the imidazole groups
destabilize hydrogen-bonded aggregates when they are
not coordinated to a metal atom. The traces for 2 and 4
have sharp leading edges and are shaped like the traces
for 5 and 6. A trace of trismelamine and ZnTPP is
Mixin g h u b(M)3 a n d h u b(MIm )3 w ith a n Am ou n t
of CA Less th a n th e 1:3 Stoich iom etr y Requ ir ed for
th e F or m a tion of Agr ega tes Sh ow s th a t 5 Is Mor e
Sta ble th a n 1. Adding hub(M)3 to a solution of 1 leads
to the rapid formation of 5 and free hub(MIm)3 in a ratio
that is greater than 10:1. This result is consistent with
the results obtained from GPC: 5 is more stable than 1.
Unfortunately, the spectrum of this mixture is compli-
cated by the presence of other hydrogen-bonding species.
These species exist in smallsbut significantsamounts
and prevent further analysis of the relative stability of

Self-Assembly of Zinc Porphyrins
J . Org. Chem., Vol. 62, No. 26, 1997 8999
apparatus. Unless otherwise noted, all aggregation experi-
ments were performed at a concentration of 5 mM. GPC was
performed with a Waters 600E HPLC system with a Waters
Ultrastyragel 10 column employing a flow rate of 1 mL min-1
with a 10 µL injection volume, with a Waters 484 detector at
254 nm. UV/vis spectra were acquired on a Hitachi U-3210
instrument as solutions in CH2Cl2; for porphyrin-containing
aggregates, a 1-mm path length cell was employed.
5 and 1. These species do serve as a warning: hydrogen
bonds (on melamines) left unsatisfied can promote the
formation of undesired aggregates.
Addition of ZnTPP to an equimolar solution of 5 and
hub(MIm
) leads to the formation of 2. At 3 equiv of
3
ZnTPPsin terms of hub(MIm)3sthe resulting solution is
a 3:2 mixture of 5:2. Additional resonances (existing in
<20%) correspond to other structurally undefined, hy-
drogen-bonded aggregates.
2-(1,3-D ih y d r o -1,3-d io x o -2H -is o in d o l-2-y l)b e n zo ic
Acid (7). The title compound was prepared by the reported
method:2 1H NMR (400 MHz, DMSO-d6) δ 13.2 (bs, 1H), 8.11
(d, J ) 5.1 Hz, 1H), 7.98 (m, 2H), 7.92 (m, 2H), 7.76 (t, J ) 5.1
The similarity in the stabilities of 2 and 5 is intuitively
reasonable. The change from a tert-butyl group to an
imidazole group on the periphery of a hydrogen-bonded
aggregate should only affect the stability of the aggregate
by competing for hydrogen bonds. Upon formation of a
zinc‚imidazole interaction, the competition for hydrogen
bonds does not occur and the resulting aggregatesbarring
unfavorable steric interactions between the chelating
group and the original scaffoldsshould be comparable in
stability to the tert-butyl analogue.
Hz, 1H), 7.62 (t, J ) 4.3 Hz, 1H), 7.54 (d, J ) 4.3 Hz, 1H); 13
C
NMR (100 MHz, DMSO-d6) δ 167.12, 166.09, 134.80, 132.97,
131.75, 131.48, 131.02, 130.66, 130.13, 129.25, 123.50; HRMS-
FAB (M+Na+) calcd for C15H8NO4Na 290.0249, found 290.0239.
[3-[[2-(1,3-Dih yd r o-1,3-d ioxo-2H -isoin d ol-2-yl)b e n -
zo y l][[4-(1,1-d im e t h y le t h y l)p h e n y l]m e t h y l]a m in o ]-
p h en yl]ca r ba m ic Acid 1,1-Dim eth yleth yl ester (8). The
title compound was prepared by the reported method:2 1H
NMR (400 MHz, DMSO-d6) δ 9.39 (s, 1H), 8.02 (m, 2H), 7.95
(m, 2H), 7.59 (s, 1H), 7.48 (m, 2H), 7.25-7.05 (m, 5H), 7.14
(m, 2H), 6.93 (bs, 2H), 4.90 (bs, 2H), 1.45 (s, 9H), 1.21 (s, 9H);
13C NMR (100 MHz, DMSO-d6) δ 166.97, 166.41, 152.50,
149.04, 142.98, 140.21, 134.79, 133.94, 133.06, 131.56, 130.13,
130.74, 130.09, 129.48, 128.57, 127.80, 126.74, 124.84, 123.47,
121.91, 116.36, 52.24, 31.06, 28.05, 25.06; HRMS-FAB (M +
Na+) calcd for C37H37N3O5Na 626.2631, found 626.2635.
3-[[2-(1,3-Dih yd r o-1-,3-d ioxo-2H -isoin d ol-2-yl)b e n -
zo y l][[4-(1,1-d im e t h y le t h y l)p h e n y l]m e t h y l]a m in o ]-
a n ilin e (9). A solution of 8 (4.5 g, 7.5 mmol) in CH2Cl2 (30
mL) was cooled in an ice bath, and trifluoroacetic acid (1 mL)
was added. The ice bath was removed, and the reaction was
stirred at rt overnight. TLC (19:1 CH2Cl2:MeOH) revealed a
quantitative conversion to 4. The organic phase was washed
with saturated aqueous NaHCO3 (2 × 30 mL) and H2O (30
mL) and dried over Na2SO4. Concentration afforded 4 (3.8 g,
100%) as an off-white solid: 1H NMR (400 MHz, DMSO-d6) δ
8.03 (m, 2H), 7.95 (m, 2H), 7.50 (bs, 2H), 7.24-7.07 (m, 7H),
6.88 (d, J ) 5 Hz, 1H), 6.36 (bs, 2H), 4.95 (s, 2H), 1.22 (s, 9H);
13C NMR (100 MHz, DMSO-d6) δ 167.17, 149.19, 143.78,
134.91, 133.88, 132.83, 131.60, 131.16, 130.32, 129.65, 129.39,
128.87, 128.16, 127.79, 126.84, 126.75, 125.28, 124.93, 123.56,
122.68, 117.93, 115.50, 117.06, 52.33, 34.07, 31.08; HRMS-FAB
(M + Na+) calcd for C32H29N3O3Na 526.2107, obsd 526.2064.
N,N′,N′′-Tr is[3-[[2-(1,3-Dih yd r o-1-,3-d ioxo-2H-isoin d ol-
2-yl)ben zoyl][[4-(1,1-d im eth yleth yl)p h en yl]m eth yl]a m i-
n o]p h en yl]-1,3,5-ben zen etr ica r boxya m id e (10). A solu-
tion of 4 (1.4 g, 2.8 mmol) in CH2Cl2 (10 mL) was added
dropwise to an ice-cold solution of 1,3,5-benzenetricarbonyl
chloride (245 mg, 0.9 mmol) and diisopropylethylamine (DI-
PEA) (180 mg, 0.25 mL, 14 mmol) in CH2Cl2 (200 mL). The
ice bath was removed, and the reaction mixture was stirred
overnight. Following evaporation of the reaction mixture,
EtOAc (100 mL) was added to the residue, and the organic
layer was washed with H2O (2 × 50 mL), dried over MgSO4,
and evaporated onto SiO2. Chromatography (9:1 CH2Cl2:
MeOH) afforded 5 as a white solid (1.3 g, 85%): 1H NMR (400
MHz, DMSO-d6) δ 10.68 (s, 1H), 8.70 (s, 1H), 8.03 (m, 3H),
7,92 (m, 2H), 7.6 (d, 1H, 4.8 Hz), 7.49 (bs, 2H), 7.24 (m, 4H),
7.14 (bs, 4H), 4.99 (bs, 2H), 1.22 (s, 9H); 13C NMR (100 MHz,
DMSO-d6) δ 167.04, 166.54, 164.43, 149.15, 143.00, 139.56,
135.27, 134.84, 133.94, 131.59, 130.92, 130.27, 130.17, 129.86,
129.60, 128.79, 127.85, 126.77, 133.03, 124.94, 123.82, 123.53,
118.64, 117.85, 52.34, 34.07, 31.08; HRMS-FAB (M + Na+)
calcd for C105H87N9O12Na 1686, found 1686; MS ion profile
calcd (M + Na+) m/z (relative intensity) 1689 (82), 1690 (100),
1691 (62), 1692 (27), 1693 (9), found 1689 (88),1690 (100), 1691
(64), 1692 (30), 1693 (11).
In flu en ce of Zn TP P on Isom er s of 3 a n d 4. The
1H NMR spectrum of 3, and in particular the presence
of two plus 12 resonances in the imide region, establishes
that 3 exists as a mixture of two isomeric aggregates of
C3 and C1-symmetry. Similar behavior for 5 had been
observed previously. Computational results3 have been
used to rationalize the observed distribution of isomers;
these results are based on the assumption that eclipsing
interactions between rosettes is unfavorable. When 6
equiv of ZnTPP is added to 3 to generate 4, the imide
1
region of the H NMR is simplified; the spectrum now
contains only two resonances for the hydrogen-bonded
imide protons. This observation suggests that as the
sterically bulky ZnTPP coordinates on the periphery of
these aggregates, the preference for the isomer that
allows the fewest unfavorable steric interactions is
increased. The resulting aggregate consists of only a
single isomer, probably of D3 symmetry.
Con clu sion s
This work establishes that imidazole groups are com-
patible with hydrogen-bonded aggregates based on the
CA‚M lattice. Aggregates containing imidazole groups
are less stable than their tert-butyl analogues. We
hypothesize that this difference in stability reflects the
capacity of imidazole groups to compete for hydrogen
bonds. Addition of ZnTPP yields aggregates of the
expected 1:1 stoichiometry of imidazole:ZnTPP. These
aggregates are more stable than those presenting free
imidazole groups and are of comparable stability to the
tert-butyl analogues.
The incorporation of zinc‚imidazole interactions in
aggregates based on CA‚M is our first attempt at
hierarchical self-assemblysthe noncovalent synthesis of
many particles into a discrete aggregate using multiple
types of interactions. The important conclusion is that
the stability of the aggregates that we have reportedsand
presumably of the family of aggregatessis not affected
by the presence of imidazole groups when ZnTPP coor-
dinates to the imidazole groups.
N,N′,N′′-Tr is[3-[(2-a m in oben zoyl)[[4-(1,1-d im eth yleth -
yl)p h en yl]m et h yl]a m in o]p h en yl]-1,3,5-b en zen et r ica r -
boxya m id e (11). A solution of 5 (1.2 g, 0.7 mmol) in MeOH
(50 mL) was cooled in an ice bath. Hydrazine (61 mg, 2.2
mmol) was added. The ice bath was removed, and the reaction
mixture was heated at reflux for 6 h. Following evaporation
of the reaction mixture, the residue was resuspended in EtOAc
Exp er im en ta l Section
Gen er a l Meth od s. Starting materials were obtained from
commercial sources (Aldrich and Fluka) and were used without
further purification. Compound 14 was available from previ-
ous studies. NMR spectra were collected on a Bruker AM-
400 equipped with a Wescan-1000 temperature monitoring

9000 J . Org. Chem., Vol. 62, No. 26, 1997
Simanek et al.
(100 mL), washed with H2O (3 × 50 mL), dried over Na2SO4,
and evaporated onto SiO2. Chromatography (19:1 CH2Cl2:
MeOH) afforded 11 as a white solid (0.9 g, 90%): 1H NMR
(400 MHz, DMSO-d6) δ 10.54 (s, 1H), 8.61 (s, 1H), 7.74 (s, 1H),
7.55 (d, 1H, J ) 4.6 Hz), 7.33 (d, 2H, J ) 5.1 Hz), 7.26 (d, 2H,
J ) 5.1 Hz), 7.14 (t, 1H, J ) 4.8 Hz), 6.94 (t, 1H, J ) 4.5 Hz),
6.84 (d, 1H, J ) 4.8 Hz), 6.74 (d, 1H, J ) 4.5 Hz), 6.65 (d, 1H,
J ) 5.1 Hz), 6.27 (t, 1H, J ) 4.6 Hz), 5.50 (bs, 2H), 5.05 (s,
2H), 1.24 (s, 9H); 13C NMR (100 MHz, DMSO-d6) δ 170.24,
167.03, 164.33, 149.28, 147.60, 143.79, 139.37, 135.16, 134.60,
134.29, 130.21, 129.78, 128.89, 128.81, 127.14, 125.08, 124.97,
124.96, 122.77, 122.72, 118.69, 118.09, 118.01, 115.78, 114.83,
52.46, 34.11, 31.09; HRMS-FAB (M + Na+) calcd for C81H82N9O6-
Na 1290, found 1290; MS ion profile calcd (M + Na+) m/z
(relative intensity) 1296 (19), 1297 (23), 1298 (100), 1299 (89),
1300 (43), 1301 (16); found 1296 (18), 1297 (23), 1298 (100),
1299 (88), 1300 (43), 1301 (16).
concentrated. The reaction mixture was partitioned between
aqueous 0.1 N NaOH and EtOAc. The aqueous layer was
extracted with EtOAc (6 × 30 mL), the combined organic layer
was dried over Na2SO4, and the crude material was absorbed
onto silica gel. Chromatography (90:10:1 CH2Cl2:MeOH:NH4-
OH) gave 13 (375 mg, 65%) as an off-white solid: 1H NMR
(400 MHz, DMSO-d6) δ 10.56 (s, 1H), 8.59 (s, 2H), 7.75 (s, 1H),
7.58 (m, 2H), 7.32 (d, J ) 5.1 Hz, 2H), 7.26 (m, 3H), 7.14 (m,
2H), 7.05 (bs, 1H), 6.94 (t, J ) 4.5 Hz, 1H), 6.82 (m, 1H), 6.70-
6.40 (m, 3H), 5.08 (bs, 2H), 3.96 (bs, 2H), 3.17 (bs, 2H), 1.90
(bs, 2H), 1.22 (s, 9H); 13C NMR (100 MHz, DMSO-d6) δ 143.84,
140.15, 139.78, 135.69, 134.70, 128.86, 127.60, 125.68, 119.72,
118.86, 118.54, 52.85, 43.71, 34.07, 30.99, 30.65; HRMS-FAB
(M + Na+) calcd for C108H115N30O6Na 1950, found 1950; MS
ion profile calcd (M + Na+) m/z (relative intensity) 1949 (75),
1950 (100), 1951 (66), 1952 (30); found 1949 (85), 1950 (100),
1951 (72), 1952 (42).
Tr ich lor id e 12. Trianiline 11 (400 mg, 0.3 mmol) was
dissolved in THF (3 mL) and then added dropwise to an ice-
cold solution of cyanuric chloride (345 mg, 1.9 mmol) and
DIPEA (1.5 g, 2 mL, 120 mmol) in THF (50 mL). The reaction
was allowed warm to room temperature over 1 h, at which
time NH3(g) was bubbled through the solution for 15 min and
then stirred overnight. The reaction mixture was evaporated
to dryness and partitioned between aqueous 0.1 N NaOH (50
mL) and EtOAc (30 mL). The aqueous layer was washed with
ethyl acetate (2×), and the organic layers were combined and
dried over Na2SO4. The crude product obtained after filtration
and evaporation was used in the next reaction without further
purification.
N ,N ′,N ′′-Tr is[3-[[2-[[4-a m in o-6-(3-im id a zolyl-N -p r o-
p yla m in o)-1,3,5-t r ia zin e-2-yl]a m in o]b en zoyl][[4-(1,1-d i-
m et h ylet h yl)p h en yl]m et h yl]a m in o]p h en yl]-1,3,5-b en -
zen etr ica r boxya m id e, Hu b(MIm )3, (13). The crude trichlo-
ride 12 was dissolved in dioxane (10 mL), and aminopropy-
limidazole (1.0 g, 1.0 mL, 8 mmol) was added. After heating
at reflux overnight, the reaction mixture was cooled to rt and
Ack n ow led gm en t. E.E.S. was an Eli Lilly Predoc-
toral Fellow (1995). L.I. thanks the NIH for a postdoc-
toral fellowship and Dr. J oydeep Lahiri for many helpful
discussions. The spectrometer facility is supported by
NIH Grant No. 1-S10-RR04870-01 and NSF Grant No.
CHE-88-14019. This work was supported by the NSF
(Grant No. CHE-91-22331). Mass spectrometry was
performed by Dr. Andrew Tyler. The Harvard Mass
Spectrometry Facility is supported by NIH RR 08458.
Su p p or tin g In for m a tion Ava ila ble: 1H NMR spectra
and ESI-MS spectrograms for 1-4 (8 pages). This material is
contained in libraries on microfiche, immediately follows this
article in the microfilm version of the journal, and can be
ordered from the ACS; see any current masthead page for
ordering information.
J O9615458