Nanosized polyphenylene dendrimers based upon pentaphenylbenzene units Frank Morgenroth, Christian Ku� bel and Klaus Mu�llen* Max-Planck-Institut fu� r Polymerforschung, Ackermannweg 10, D-55128 Mainz, Germany A facile divergent synthesis of monodisperse polyphenylene dendrimers having diameters of 21–55 A° is presented. These nanoparticles have been prepared via a [2+4]cycloaddition–deprotection sequence using an excess of tetraphenylcyclopentadienone 1a as monomer and the tetraethynylbiphenyl 2 as core.Due to the dense packing of 22, 62 or 142 benzene rings in generations G1, G2 and G3 , respectively, the conformational freedom of the higher generations G2 and G3 is limited. Molecular mechanics calculations as well as molecular dynamics simulations are included in a discussion of the structure and the shape-persistence of G2.The calculations revealed that selected inner distances of the molecule varied only 5–10% during the molecular dynamics simulations, thus indicating that the overall shape of the molecule essentially did not change throughout the simulation time. Within the past decade, chemists have established new methods Results and Discussion to synthesize complex molecules and also to control their Synthesis macromolecular and supramolecular architecture.These eorts are connected tightly to a transition in molecular scale from We have recently introduced 3,4-bis(4-triisopropylsilylethynyl- the picometre to the nanometre range.1 Supramolecular chem- phenyl)-2,5-diphenylcyclopenta-2,4-dienone (1a) as a building istry and nanostructure engineering assemble organic and block for the synthesis of monodisperse polyphenylene dendri- inorganic molecules using covalent as well as non-covalent mers such as 3a and 4a (Scheme 1).8 A key step of our interactions to obtain complex systems such as liquid crystals approach was the [2+4] cycloaddition of 1a to a core such or nanotubes.2 By focusing on all-hydrocarbon, covalent bond- as 3,3¾,5,5¾-tetraethynylbiphenyl (2).Compound 1a contained based dendritic nanostructures,3–6 Moore and his co-workers a diene function and two dienophile functions and could thus built nanosized phenylacetylene dendrimers.3 Hart et al. have be considered an AB2 building block. Since the two dienophile reported the construction of polyaromatic iptycenes.4 A functions in 1a were blocked by triisopropylsilyl groups, four remarkable feature of these nanostructures is their essentially equivalents of the tetraphenylcyclopentadienone 1a could be invariant shape (shape-persistence).7 Shape-persistence is a reacted selectively in a four-fold [2+4] cycloaddition with the requirement for the design of nanoparticles with tailor-made four dienophile functions of 2.Under the applied reaction topographic features and well-defined functional group dispo- conditions (180 to 200 °C, diphenyl ether–a-methylnaphthalene sitions. In the case of Moore’s phenylacetylene-based macro- 151) the Diels–Alder reaction proceeded with the loss of molecules, shape-persistence was accomplished using rigid carbon monoxide to yield the first generation 3a of the phenylacetylene subunits.The cascades of Hart are highly rigid dendrimer. Compound 3a was used for an eight-fold addition due to the connection of bicyclo[2.2.2]octane moieties with of 1a after an easy deprotection with tetrabutylammonium benzenes. In this paper, we present an alternative route to fluoride (Bu4NF) in THF resulting in the second generation generation G3 , of a new type of nanosized all-hydrocarbon 4a (Scheme 1).As discussed below in detail, characterization dendrimers. Our idea for obtaining shape-persistence was to was accomplished by mass spectrometry and 1H and 13C NMR create an extremely dense packing of benzene rings, thereby spectroscopy. limiting the conformational freedom of the dendrimer.This paper outlines the synthetic aspects of our dendrimer Considering these structural aspects, we present molecular construction and its extension to the fabrication of G3. The mechanics calculations of G2, which provide insight into the procedure employed worked for generation growth as well as three dimensional structure of this molecule. Additionally, in activation both in high yields and with no significant side this report we will discuss molecular dynamics simulations reactions.The yields of white amorphous polyphenylene den- investigating the shape-persistence of G2. drimers, from the Diels–Alder growth reaction, were higher than 80% and the deprotection reaction was nearly quantitative. However, the growth steps had to be optimized to avoid an incomplete reaction of the terminal ethynyl groups since this severely aected the growth scheme.Simply heating a solution of 5 equiv. of 1a and 2 in diphenyl ether–a-methylnaphthalene to 200 °C for 1 h led to a mixture of two-, three- and the desired four-fold Diels–Alder reaction product. As expected, this mixture could not be separated chromatographically. In contrast, a quantitative formation of G1 was obtained by adding 2 in a-methylnaphthalene slowly to a hot solution of 5 equiv.of 1a in diphenyl ether–amethylnaphthalene. By focusing on the growth step from G1 to G2, a longer reaction time was essential to avoid growth imperfections. This was expected since our divergent approach required the cycloaddition to be carried out at an increasing number of ethynyl groups.In addition, the use of pentaphenyl substituted benzene rings as repetitive units resulted in a very J. Mater. Chem., 1997, 7(7), 1207–1211 1207Scheme 1 Chemical structure and procedure for the preparation of the polyphenylene dendrimer generations G1 to G3. The dotted lines mark ‘half-G1’ and ‘half-G2’, respectively (see text). Overlapping phenyl rings in the 2D projection of 5 are plotted in bold.Reagents and conditions: i, 1a, heat; ii, Bu4NF. compact structure. Furthermore, steric crowding in the vicinity methylnaphthalene to a solution of 1a in diphenyl ether–amethylnaphthalene at 180 to 200 °C over a period of 25 h. The of the ethynyl groups had presumably lowered the reaction rate (see Fig. 1). Accordingly, the addition of 3b to 5 equiv. 1a mixture was stirred for an additional 15 h at 180 °C. In the case of G3 , the addition of fresh 1a was required in order to at 180 to 200 °C over 2 h and an additional reaction time of the same time, yielded a mixture of seven- and eight-fold sustain a sucient excess of the cyclopentadienone during reaction, otherwise traces of 10- to 15-fold cycloaddition [2+4] cycloaddition product.However, the addition of 3b over 5 h and an additional reaction time of 4 h selectively led products were observed. The latter could be converted into 5a by a second treatment with an excess of 1a. We obtained 5a to the desired eight-fold Diels–Alder product. By extending this concept, the synthesis of G3 was achieved as a white amorphous powder in 81% yield. A comparison of our divergent approach with the convergent synthesis of an by adding a mixture of 4b and 1a in diphenyl ether–a- 1208 J.Mater. Chem., 1997, 7(7), 1207–1211caused a mass dierence of ca. 717 g mol-1 (1a less CO) relative to the completely reacted product. The perfect agreement between calculated and experimentally determined m/z ratios for G1 to G3 confirmed the monodispersity of the dendrimers (see Experimental).Computer simulations Molecular mechanics calculations12 performed for G1 and G2 (without acetylene units) provided a first insight into the 3D structure of the dendrimers (Fig. 1). In the first step, one half of the molecule (see Scheme 1) was optimised by energy minimisation using 144 and 1024 dierent conformers (obtained by torsion about the marked bonds in Scheme 1) of G1 and G2, respectively.Thereby we obtained one conformer for ‘half-G1’ (see Scheme 1) with an energy 3 kcal mol-1 (1 cal=4.184 J) lower than any other and another 20 conformers diering by less than 10 kcal mol-1. This observation was even more pronounced in the case of ‘half-G2’ (see Scheme 1 and Fig. 1), where wd more than 100 conformers diering by less than 15 kcal mol-1.In both cases we used the conformer with the lowest energy to generate the complete dendrimer. This was finally optimised by varying the Fig. 1 The 3D structure of G2 (without acetylene units) according to torsional angle of the central phenyl–phenyl unit (bond a, see molecular mechanics calculations12 Fig. 2). The latter was generated by connecting two ‘half-G2’ to obtain G2 itself.oligophenylene-dendrimer with 46 benzene rings carried out Therefore, we obtained a 3D structure of G2, which could by Miller and Neenan9 revealed that in our experiments a be regarded as consisting of a twisted central biphenyl unit faster dendrimer growth, in terms of the number of benzene carrying four branches which were nearly orthogonal to the rings, was achieved in fewer steps and without the use of central biphenyl unit.The two pentaphenylbenzene units in catalysts. Our synthetic approach has also enabled products each branch were twisted. They were nearly parallel to each to be purified simply by precipitation from acetone with other, but the protons of the central phenyl rings pointed in ethanol.dierent directions (Fig. 1). The dimensions of the dendrimer The larger dendrimer generations G2 and G3, accessible by G2, in the three main directions, were 36, 27and 14 A° according our process, had 62 and 142 benzene rings, respectively. We to the molecular mechanics calculations. used computer models to estimate the dimensions of these The existence of several local energy minima of similar nanoparticles.As an indication of the impressive size of our energy is a requirement for a flexible molecule, but the dynamic dendrimers, the diameters of G2 and G3 were found to be behaviour is controlled by the activation energy for trans- approximately 38 and 55 A° (Table 1). forming the conformers into one another. A first insight into The success of our dendrimer synthesis was complemented the dynamic behaviour of the dendrimer was gained by molecu- by the full characterization of the new structures.lar dynamics (MD) simulations13 of G2. We analysed the trajectory of the MD simulations by inspecting specific intra- Characterization molecular distances and torsional angles (specified in Fig. 2 and summarized in Table 2) to obtain information about the Mass spectrometry is an excellent tool for determining the motions within the dendrimer.purity and structural composition of dendrimers.10 The triisopropylsilyl- protected as well as the oligoethynyl substituted dendrimers exhibited an unexpectedly good solubility in common solvents such as dichloromethane, acetone or tetrahydrofuran (THF). Therefore, characterization was performed using matrix-assisted laser desorption ionization time of flight mass spectrometry (MALDI-TOF-MS) as well as 1H and 13C NMR spectroscopy.By 1H NMR spectroscopy it was only possible to check the correct intensity ratio of aromatic and aliphatic (or acetylenic) protons.11 Employing MALDI-TOFMS allowed potential growth imperfections during the [2+4] cycloaddition to be detected without any ambiguity, even at the higher generations, since each unreacted ethynyl group Table 1 Summary of the molecular mass and the approximate diameter of the dendrimers.These diameters were estimated from molecular models generated using the CERIUS2 program number formula of mass/ benzene generation formula gmol-1 rings diameter/A° G0, 2 C20H10 250 2 7 G1, 3b C148H90 1868 22 21 Fig. 2 2D projection of G2 (without acetylene units). The distances G2, 4b C404H250 5104 62 38 and torsional angles used to determine intramolecular rotations and G3, 5b C916H570 11577 142 55 the overall shape of the dendrimer G2 are labelled. J. Mater. Chem., 1997, 7(7), 1207–1211 1209Table 2 Intramolecular distances and torsional angles (values obtained of the dendrimer during the whole simulation time.Therefore, by MD simulations,13 compare Fig. 2) used to determine intramolecu- the functional groups were exposed for further reaction and lar rotations and the overall shape of the dendrimer G2 had a well-defined chemical environment due to the shapepersistence of the dendrimer. distance/A° The next generation, G3, had an even higher packing density number (%) torsion angle/° of phenylene units.We expected the dierent branches of this 1 2.8±0.16 a 20–60 dendrimer to be even more correlated (details have yet to be (±5.3) revealed by further MD simulations) resulting in an even more 2 4.7±0.15 b1 115–140 shape-persistent 3D structure with a diameter of ca. 55 A° . (±3.2) 3 6.7±0.25 b2 -25–(-50) (±3.7) Summary and Outlook 4 17.1±1.3 (±7.3) Our innovative approach can be summarized as follows: 5 11.3±1.0 $ A simple synthetic process led quickly to nanosized mol- (±8.8) 6 7.4±0.6 ecules with well-defined size and shape.(±8.0) $ A new type of polyphenylene dendrimer based on penta- 7 8.0±1.0 phenylbenzene units was obtained. (±12.5) $ The synthesized materials were soluble in common organic 8 13.8±1.0 solvents and, therefore, were readily processable.(±7.2) $ It was possible to control the chemical functionalization 9 27.5±2.5 (±9.1) of the nanoparticles. 10 31.0±1.5 With respect to the last point, it should be mentioned that (±4.8) we are working on the functionalization of our dendrimers. A 11 34.2±1.7 straightforward approach would be afinal Diels–Alder reaction (±5.0) with tetraphenylcyclopentadienones carrying functional groups 12 35.8±2.2 such as MNR3+ (1b), MCN (1c), MOH (1d), MCOOH (1e) (±6.1) or M(CH2)nCH3 (1f ).Such groups are relevant for chemical, physical and potentially catalytic properties.14 Our approach uses charged as well as neutral lipophilic or hydrophilic and The observed distribution of distances had several reasons.hydrogen-bonding substituents. In addition, units which facili- There were three (approximately) independent motions: tate the covalent coupling of electronically active components vibration, rotation and bending of the oligophenylenes. In such as redox-systems (in particular viologen), dyes, NLO- order to distinguish between these motions we compared two phores or active metal complexes are attractive.The synthesis distance distributions: a distance sensitive to the rotations of of such species is the subject of our current investigations. interest and one with similar connectivity, but being insensitive to any rotation (for example, distances 2 and 3 or 6 and 7, Fig. 2). The dierence between both distributions was assumed Experimental to be due to independent rotations.Furthermore, the corre- Melting points are uncorrected. 1H NMR spectra were sponding torsional angles revealed that the rotations were recorded at 500 MHz on a Bruker DRX 500 spectrometer. (nearly) frozen or that they were correlated strongly. 13C NMR spectra were recorded on the same instrument at The calculations indicated the following qualitative picture 125 MHz. Chemical shifts are given in parts per million (ppm) for the internal rotations of the dendrimer: using the solvent signal as reference;15 J values are in Hz. Mass $ We observed a rotation of the central phenyl–phenyl unit spectral analyses were carried out on ZAB2-SE-FPD (VG (bond a) with the torsional angle varying by ±20° and Analytical) and Bruker Reflex-TOF.MALDI-TOF-MS spec- an average torsional angle of ca. 40°. The average torsional tra were measured using 1,8,9-trihydroxyanthracene as matrix. angle was not the one found during the energy minimiz- Compounds 1a and 2 were prepared from commercially avail- ation, instead a continuous change of the torsional angle able starting materials by methods to those already in of bond a was observed throughout the equilibration time literature.16–18 (distance 1, torsional angle of bond a).$ A rotation about bond b was observed (torsional angle Generation G1, 3a, 3b of bond b1 and b2 ), but this rotation was accompanied only by a small change in distance 3, thus indicating a To a degassed solution of 1a (3.27 g, 0.00440 mol) in a mixture of diphenyl ether–a-methylnaphthalene (3 ml/3 ml), 2 (55.0 mg, strongly correlated rotation of both branches of the dendrimer.The same results were obtained by comparing 0.220 mmol) in a-methylnaphthalene (2 ml) was added over 30 min at 180–200 °C under a stream of argon. After stirring distances 4 and 5. $ There was an (partially) independent rotation of the for 30 min at this temperature the mixture was allowed to cool.The cold reaction mixture was diluted with methanol pentaphenylbenzene units about the bonds c4 or d4 (comparison of the distances 6 and 7). (40 ml) and stirred overnight at 25°C. A crude red product was filtered. Contaminating 1a was removed either by reprecip- In spite of the aforementioned rotations, the overall shape of the molecule changed only slightly. The distances 8 to 12 itating from acetone with ethanol or by passage of the crude product through a short column of silica gel using light generated a network measuring the diameter of the dendrimer in several directions. All these distances changed by only ±5 petroleum–dichloromethane (351) as eluent.Yield: 0.611 g (0.196 mmol, 89%) white amorphous solid, mp 169–170°C. to 10% throughout the whole simulation time, thereby indicating that the dendrimer was a shape-persistent molecule.We MS m/z (FD): 3118 (M+, 100%), calc. for C220 H250 Si8 3119. Deprotection was achieved by adding tetrabutylammonium suppose that this behaviour was due to the high packing density of phenylene rings, which caused the strongly correlated fluoride (0.820 g, 3.14 mmol) to a stirred solution of 3a (0.611 g, 0.196 mmol) in THF (50 ml).After stirring for 5 h at 25°C, rotations within the molecule. Thus the shape-persistent dendrimer retained some flexibility which enhanced solubility in the THF was evaporated under reduced pressure to give a yellow solid, which was taken up in acetone. The resulting common solvents. In addition, the phenylene groups carrying the functional groups (acetylene units) in G2 were pointing out solution was added dropwise to ethanol (100 ml).The desired 1210 J. Mater. Chem., 1997, 7(7), 1207–1211product precipitated as white solid and was filtered. Foundation of the Chemical Industries (Fonds der Chemischen Industrie) for scholarships. Deprotection was nearly quantitative, mp>300°C, dH (500 MHz, CD2Cl2, 303 K): 7.25–7.21 (br, 12 H), 7.14–6.97 (br, 28 H), 6.88–6.68 (br, 38 H), 6.45–6.38 (br, 4 H), 3.03–3.00 References (br, 8 H).dC (125 MHz, CD2Cl2 , 303 K): 141.54, 141.49, 141.27, 1 J.-M. Lehn, Supramolecular Chemistry, Concepts and Perspectives, 141.11, 141.00, 140.67, 140.04, 139.31, 138.66, 132.11, 131.87, VCH, Weinheim, 1995. 131.84, 131.79, 131.18, 130.88, 130.31, 128.12, 127.64, 127.01, 2 F.Vo� gtle, Supramolekulare Chemie, Eine Einfu� hrung, 126.93, 126.50, 119.80, 119.50, 83.78, 77.43, 77.33. MS m/z B. G. Teubner, Stuttgart, 1992. (MALDI-TOF): 1907 in the presence of K, calc. for 3 Z. Xu and J. S. Moore, Angew. Chem., 1993, 105, 1394; Angew. C148H90K+, 1907. Chem., Int. Ed. Engl., 1993, 32, 1354; Z. Xu and J. S. Moore, Acta Polym., 1994, 45, 83. 4 S. B. Singh and H.Hart, J. Org. Chem., 1990, 55, 3412; K. Shahlai, Generation G2, 4a, 4b H. Hart and A. Bashir-Hashemi, J. Org. Chem., 1991, 56, 6812. Preparation of 4a and 4b was analogous to 3a and 3b except 5 A. Rajca, J. Org. Chem., 1991, 56, 2557; A. Rajca and for the reaction time of the [4+2] cycloaddition. Compound S. Utamapanya, J. Am. Chem. Soc., 1993, 115, 10688. 6 G. R. Newkome, C.N. Moorefield and F. Vo� gtle, Dendritic 3b was added over 5 h to the hot reaction mixture and the Molecules, VCH, Weinheim, 1996; J. M. J. Fre� chet and solution was allowed to stir for an additional 4 h. Yield: 87%, C. J. Hawker, in Comprehensive Polymer Science, 2nd. Suppl., ed. white amorphous solid, 4a, mp>300°C. MS m/z (MALDI- G. Allen, S. L. Aggarwal and S. Russo, Elsevier, Oxford, 1996, TOF): 7645 in the presence of K, calc.for C548H570Si16K+, pp. 70–129. 7645. 7 J. K. Young and J. S. Moore in Modern Acetylene Chemistry, ed. 4b, mp>300 °C, dH (500 MHz, CD2Cl2, 303 K): 7.46–6.45 P. J. Stang and F. Diederich, VCH,Weinheim, 1995, p. 415. 8 F. Morgenroth, E. Reuther and K. Mu�llen, Angew. Chem., 1997, (br, 388 H), 3.03–2.99 (br, 404 H). dC (125 MHz, CD2Cl2, 109, 647. 303 K): 142.22, 141.71, 141.66, 141.29, 141.22, 141.12, 140.41, 9 T. M. Miller and T. X. Neenan, Chem. Mater., 1990, 2, 346; 139.90, 139.48, 139.39, 138.95, 138.56, 131.90, 131.84, 131.43, T. M. Miller, T. X. Neenan, R. Zayas and H. E. Bair, J. Am. Chem. 131.30, 131.15, 130.86, 130.46, 130.22, 128.92, 128.60, 128.08, Soc., 1992, 114, 1018. 127.98, 127.37, 126.89, 126.27, 126.19, 121.31, 119.77, 119.46, 10 K.L. Walker, M. S. Kahr, C. L. Wilkins, Z. Xu and J. S. Moore, 102.69, 83.80, 77.39, 77.33. Am. Soc.Mass Spectrom., 1994, 5, 730. 11 Due to the limited accuracy of the integration, traces of side products due to incomplete [2+4] cycloaddition could not be detected Generation G3, 5a by 1H NMR spectroscopy, but appeared in the MALDI-TOF-MS as distinguished peaks. Compound 1a (1,87 g, 2.51 mmol) was taken up in a mixture 12 The molecular mechanics calculations were carried out using the of diphenylether–a-methylnaphthalene (8 ml/8 ml).The stirred MM2 (85) forcefield with the CERIUS2 program package and solution was degassed and heated to 180–200°C under argon. applying the Conjugate Gradient 200 algorithm. G1 and G2 were Then a degassed solution of 4b (80.0 mg, 0.0157 mmol) and 1a minimized by considering only one half of the molecule.In the case (1.17 g, 1.57 mmol) in a mixture of diphenyl ether–a-methyl- of G1, two torsional angles b1 and b2 (as shown in Fig. 2) were naphthalene (5 ml/5 ml) was added dropwise over 25 h under varied stepwise by 30°, whereas four torsional angles were varied in the case of G2 (as shown in Fig. 2) in 45° steps.In a second step, a stream of argon. After the addition was complete, the reaction two of those ‘half-molecules’ were connected and the torsional mixture was allowed to stir for an additional 15 h at 180 °C. angle of this newly created, central phenyl–phenyl bond was The cold reaction mixture was diluted with methanol (40 ml) optimised.and stirred overnight at 25°C. The crude red product was 13 NVT-molecular dynamics simulations were carried out at 300 K filtered and reprecipitated from acetone with ethanol. Yield: (temperature dumping, 0.1 ps relaxation time) with a time step of 210.9 mg (0.0127 mmol, 81%), white amorphous solid, 0.001 ps. For the evaluation of the molecular shape a simulation time of 35 ps was used after an initial equilibration time of 5 ps.mp>300 °C, dH (500 MHz, CD2Cl2, 303 K): 7.43–6.41 (br, Intramolecular distances were calculated by averaging the dis- 538 H), 1.21–0.99 (br, 672 H). dC (125 MHz, CD2Cl2, 303 K): tances measured for every tenth conformer, torsional angles were 142.21, 141.89, 141.50, 141.24, 141.12, 140.77, 140.40, 140.33, measured every ps. 140.04, 139.54, 139.46, 138.92, 138.65, 138.51, 138.21, 131.81, 14 R. Dagani, Chem. Eng. News, 1996, June 3, p. 30. 131.40, 131.20, 131.05, 130.78, 130.45, 130.22, 128.90, 128.59, 15 H. O. Kalinowski, S. Berger and S. Braun, 13C-NMR- 128.10, 127.96, 127.40, 127.23, 126.85, 126.24, 121.18, 120.88, Spektroskopie, Thieme, Stuttgart 1984, p. 74. 16 S. Takahashi, Y. Kuroyama, K. Sonogashira and N. Hagihara, 119.22, 107.30, 90.95, 90.86, 18.80, 18.45, 11.93, 11.71, 11.48. Synthesis, 1980, 627. MS m/z (MALDI-TOF): 16619 in the presence of K, calc. for 17 M. A. Ogliaruso, M. G. Romanelli and E. I. Becker, Chem. Rev., C1204H1210Si32K+, 16619. 1965, 65(3 ), 261; W. Broser, J. Reusch, H. Kurreck and P. Siegle, Chem. Ber., 1969, 102, 1715. Financial support by the Volkswagenstiftung and the 18 F. L. W. van Roosmalen, Recl. T rav. Chim. Pays-Bas, 1934, 53, 359. Bundesministerium fu�r Bildung und Forschung is gratefully acknowledged. C. K. and F. M. would also like to thank the 2D; Received 2nd January, 1997 J. Mater. Chem., 1997, 7(7), 1207–1211 1211