Crystal engineering of some 2,4,6-triaryloxy-1,3,5-triazines: Octupolar nonlinear materials

Article abstract of DOI:10.1021/ja972830r

The principles of crystal engineering have been used to design a family of structures with potential as octupolar nonlinear optical (NLO) materials. The major aim in such an exercise, a carry-over of molecular symmetry into the crystal, is possible with a retrosynthetic approach. An appropriate choice of precursor trigonal molecules leads from the concept of the dimeric Piedfort unit. The crystal structures and NLO properties of a series of 2,4,6-triaryloxy-1,3,5-triazines, 1-6, are reported. These compounds consistently form quasitrigonal or trigonal networks that are two-dimensionally noncentrosymmetric. Substitutional variations on the phenyl moieties that were expected to maintain or to perturb this trigonal network have been explored. Molecular nonlinearities have been measured by Harmonic Light Scattering (HLS) experiments. Among the compounds studied, 2,4,6-triphenoxy-1,3,5-triazine, 1 adopts a noncentrosymmetric crystal structure with a measurable SHG powder signal. All these crystal structures are stabilized by weak intermolecular interactions such as herringbone, π...π, C-H...O, and C-H...N hydrogen bonding. These octupolar molecules are more isotropic than the classical p-nitroaniline based dipolar NLO molecules, and this is advantageous from the viewpoint of potential electrooptic applications. The principles of crystal engineering have been used to design a family of structures with potential as octupolar nonlinear optical (NLO) materials. The major aim in such an exercise, a carry-over of molecular symmetry into the crystal, is possible with a retrosynthetic approach. An appropriate choice of precursor trigonal molecules leads from the concept of the dimeric Piedfort unit. The crystal structures and NLO properties of a series of 2,4,6-triaryloxy-1,3,5-triazines, 1-6, are reported. These compounds consistently form quasitrigonal or trigonal networks that are two- dimensionally noncentrosymmetric. Substitutional variations on the phenyl moieties that were expected to maintain or to perturb this trigonal network have been explored. Molecular nonlinearities have been measured by Harmonic Light Scattering (HLS) experiments. Among the compounds studied, 2,4,6- triphenoxy-1,3,5-triazine, 1 adopts a noncentrosymmetric crystal structure with a measurable SHG powder signal. All these crystal structures are stabilized by weak intermolecular interactions such as herringbone, π···π, C-H···O, and C-H···N hydrogen bonding. These octupolar molecules are more isotropic than the classical p-nitroaniline based dipolar NLO molecules, and this is advantageous from the viewpoint of potential electrooptic applications.

Full text of DOI:10.1021/ja972830r

J. Am. Chem. Soc. 1998, 120, 2563-2577
2563
Crystal Engineering of Some 2,4,6-Triaryloxy-1,3,5-triazines:
Octupolar Nonlinear Materials
Venkat R. Thalladi,† Sophie Brasselet,‡,§ Hans-Christoph Weiss, Dieter Bla1ser,
Amy K. Katz,| H. L. Carrell,| Roland Boese,*, Joseph Zyss,*,‡,§ Ashwini Nangia,*,† and
Gautam R. Desiraju*,†
Contribution from the School of Chemistry, UniVersity of Hyderabad, Hyderabad 500 046, India,
France, Te´le´com, CNET, Laboratoire de Bagneux, De´partement d’ Electronique Quantique Mole´culaire,
196, aVenue Henri RaVera, 92225 Bagneux, France, Institut fu¨r Anorganische Chemie, FB 8,
UniVersita¨t-GH Essen, UniVersita¨tsstrasse 5-7, D-45177 Essen, Germany, The Institute for Cancer
Research, Fox Chase Cancer Center, 7701 Burholme AVenue, Philadelphia, PennsylVania 19111
ReceiVed August 12, 1997
Abstract: The principles of crystal engineering have been used to design a family of structures with potential
as octupolar nonlinear optical (NLO) materials. The major aim in such an exercise, a carry-over of molecular
symmetry into the crystal, is possible with a retrosynthetic approach. An appropriate choice of precursor
trigonal molecules leads from the concept of the dimeric Piedfort unit. The crystal structures and NLO properties
of a series of 2,4,6-triaryloxy-1,3,5-triazines, 1-6, are reported. These compounds consistently form quasi-
trigonal or trigonal networks that are two-dimensionally noncentrosymmetric. Substitutional variations on the
phenyl moieties that were expected to maintain or to perturb this trigonal network have been explored. Molecular
nonlinearities have been measured by Harmonic Light Scattering (HLS) experiments. Among the compounds
studied, 2,4,6-triphenoxy-1,3,5-triazine, 1 adopts a noncentrosymmetric crystal structure with a measurable
SHG powder signal. All these crystal structures are stabilized by weak intermolecular interactions such as
herringbone, π‚‚‚π, C-H‚‚‚O, and C-H‚‚‚N hydrogen bonding. These octupolar molecules are more isotropic
than the classical p-nitroaniline based dipolar NLO molecules, and this is advantageous from the viewpoint of
potential electrooptic applications.
Introduction
been engineered all the way from optimized molecular units to
state-of-the-art low-threshold optical parametric oscillators.6,7
Crystal engineering aims at the design of crystal structures
of molecular solids with specific attributes, be they topological
features, chemical functions, or physical properties.1,2 Distinct
aims and goals within the subject of crystal engineering may
now be discerned:3 (1) strategic and methodological concerns;
(2) aesthetically motivated supramolecular synthesis based on
symmetry, topology, and network properties; (3) efforts directed
toward functional materials of practical importance4 (nonlinear
optics (NLO), ferromagnetism, microporosity, electronics, solid-
state reactivity). Crystal engineering targeted at solids exhibiting
large optical nonlinearities has attracted much interest.5 Varied
goals, ranging from fundamental to application-oriented, have
converged so that optically nonlinear molecular crystals have
(3) Some very recent papers in the area of crystal engineering include
the following: (a) Russell, V. A.; Evans, C. C.; Li, W.; Ward, M. D. Science
1997, 276, 575. (b) Allen, F. H.; Hoy, V. J.; Howard, J. A. K.; Thalladi, V.
R.; Desiraju, G. R.; Wilson, C. C.; McIntyre, G. J. J. Am. Chem. Soc. 1997,
119, 3477. (c) Karle, I. L.; Ranganathan, D.; Haridas, V. J. Am. Chem.
Soc. 1997, 119, 2777. (d) Endo, K.; Ezuhara, T.; Koyanagi, M.; Masuda,
H.; Aoyama, Y. J. Am. Chem. Soc. 1997, 119, 499. (e) Fe´lix, O.; Hosseini,
M. W.; De Cian, A.; Fischer, J. Angew. Chem., Int. Ed. Engl. 1997, 36,
102. (f) Coe, S.; Kane, J. J.; Nguyen, T. L.; Toledo, L. M.; Wininger, E.;
Fowler, F. W.; Lauher, J. W. J. Am. Chem. Soc. 1997, 119, 86. (g) Hirsch,
K. A.; Wilson, S. R.; Moore, J. S. Chem. Eur. J. 1997, 3, 765. (h) Melendez,
R. E.; Sharma, C. V. K.; Zaworotko, M. Z.; Bauer, C.; Rogers, R. D. Angew.
Chem., Int. Ed. Engl. 1996, 35, 213. (i) Lewis, F. D.; Yang, J.; Stern, C. L.
J. Am. Chem. Soc. 1996, 118, 12029. (j) Yaghi, O. M.; Li, H.; Groy, T. L.
J. Am. Chem. Soc. 1996, 118, 9096. (k) Abrahams, B. F.; Batten, S. R.;
Hamit, H.; Hoskins, B. F.; Robson, R. Angew. Chem., Int. Ed. Engl. 1996,
35, 1690. (l) Eichhorst-Gerner, K.; Stabel, A.; Moessner, G.; Declerq, D.;
Valiyaveettil, S.; Enkelmann, V.; Mu¨llen, K.; Rabe, J. P. Angew. Chem.,
Int. Ed. Engl. 1996, 35, 1492. (m) Kra¨utler, B.; Mu¨ller, T.; Maynollo, J.;
Gruber, K.; Kratky, C.; Ochsenbein, P.; Schwarzenbach, D.; Bu¨rgi, H.-B.
Angew. Chem., Int. Ed. Engl. 1996, 35, 1204. (n) Glidewell, C.; Ferguson,
G. Acta Crystallogr. 1996, C52, 2528.
(4) Selected recent examples of property-directed crystal engineering
include the following: (a) Kahn, O. Curr. Opin. Solid State Mater. Sci.
1996, 1, 547. (b) Brunet, P.; Simard, M.; Wuest, J. D. J. Am. Chem. Soc.
1997, 119, 2737. (c) Sarma, J. A. R. P.; Allen, F. H.; Hoy, V. J.; Howard,
J. A. K.; Thaimattam, R.; Biradha, K.; Desiraju, G. R. Chem. Commun.
1997, 101. (d) Zyss, J.; Nicoud, J. F. Curr. Opin. Solid State Mater. Sci.
1996, 1, 533. (e) Hulliger, J.; Rogin, P.; Quintel, A.; Rechsteiner, P.; Ko¨nig,
O.; Wu¨bbenhorst, M. AdV. Mater. 1997, 9, 662. (f) Pan, F.; Wong, M. S.;
Gramlich, V.; Bosshard, C.; Gu¨nter, P. J. Am. Chem. Soc. 1996, 118, 6315.
(g) Hoss, R.; Ko¨nig, O, Kramer-Hoss, V.; Berger, U.; Rogin, P.; Hulliger,
J. Angew. Chem., Int. Ed. Engl. 1996, 35, 1664. (h) Toda, F.; Hyoda, S.;
Okada, K.; Hirotsu, K. J. Chem. Soc., Chem. Commun. 1995, 1531.
†University of Hyderabad.
‡ France Te´le´com, CNET.
Universita¨t-GH Essen.
| Fox Chase Cancer Center.
§ Present address: Laboratoire de Photonique Quantique et Mole´culaire,
De´partement de Physique, Ecole Normale Supe´rieure de Cachan, 61, Avenue
du Pre´sident Wilson, 94235 Cachan Cedex, France.
(1) (a) Desiraju, G. R. Crystal Engineering: The Design of Organic
Solids; Elsevier: Amsterdam, 1989. (b) Gavezzotti, A. Curr. Opin. Solid
State Mater. Sci. 1996, 1, 501. (c) Desiraju, G. R. Curr. Opin. Solid State
Mater. Sci. 1997, 2, 451. (d) Desiraju, G. R. Chem. Commun. 1997, 1475.
(e) Aakero¨y, C. B. Acta Crystallogr. 1997, B53, 569.
(2) (a) Philp, D.; Stoddart, J. F. Angew. Chem., Int. Ed. Engl. 1996, 35,
1154. (b) Schwiebert, K. E.; Chin, D. N.; MacDonald, J. C.; Whitesides,
G. M. J. Am. Chem. Soc. 1996, 118, 4018. (c) Wright, J. D. Molecular
Crystals, 2nd ed.; Cambridge University Press: New York, 1995. (d)
Structure and Properties of Molecular Crystals; Pierrot, M., Ed.; Elsevier:
Amsterdam, 1990.
S0002-7863(97)02830-8 CCC: $15.00 © 1998 American Chemical Society
Published on Web 03/07/1998

2564 J. Am. Chem. Soc., Vol. 120, No. 11, 1998
Thalladi et al.
These efforts have been generally based on the dipolar paradigm
wherein electron donating and withdrawing substituent groups
interact via conjugated π-electron reservoirs such as a phenyl
ring (e.g., p-nitroaniline, pNA and analogues),8 two phenyl rings
(stilbene and azo-dye derivatives),9 or more elongated carotenoid
push-pull structures.10,11
they are therefore not the best-suited for NLO-type applications.
More symmetrical molecules organized in more isotropic yet
noncentrosymmetric arrays are expected to alleviate such
problems. The traditional design of NLO active materials has
concentrated exclusively on dipolar molecules and has thus
ignored a wealth of possibilities that could arise from two- and
three-dimensional self-assembly. Diversified investigations of
more isotropic molecules with attached octupolar and multipolar
nonlinearities have been proposed.16-18 For example, 2,4,6-
triamino-1,3,5-trinitrobenzene (TATB) has attracted much at-
tention as the trigonal analogue of pNA.16a,b,19 While octupolar
nonlinearity has been experimentally demonstrated in molecular
systems,16c,d its demonstration in supramolecular, that is crystal-
line, systems has remained a challenge.20
In polar molecule-based crystal engineering of NLO materials,
the main aim is to ensure that a noncentrosymmetric assembly
of molecular dipoles is established. This is a prerequisite for
quadratic NLO effects5 such as second harmonic generation
(SHG) and for the cascading of such effects for cubic molecular
dephasing.12 A useful hierarchic classification of molecular
crystal classes with respect to their quadratic nonlinear efficiency
was established in the early 1980s and has served as a guideline
for NLO crystals engineered with polar quasi one-dimensional
molecules with a single dominant molecular hyperpolarizability
coefficient along the charge-transfer axis.13 At the magic angle
of 54.7°, the inclination of the molecule within the unit cell
provides optimal tradeoff between birefringence phase-matching
constraints and enhancement of the effective molecular coef-
ficient, as exemplified by the well-known substance N-(4-
nitrophenyl)-L-prolinol (NPP).7,14 However, the significant
ground-state dipole moment in pNA-like dipolar molecules has
been recognized as being detrimental toward the establishment
of a noncentrosymmetric structure because dipole-dipole
interactions tend to favor antiparallel stacking of neighboring
molecular units. Reducing the ground state dipole moment
while preserving optical nonlinearity had been a long-sought
goal leading eventually to the push-pull structure of 3-methyl-
4-nitropyridine-1-oxide (POM).15
A typical symmetry pattern that leads to crystalline octupolar
nonlinearity is the trigonal network A constituted with trigonal
molecules (Figure 1). This two-dimensional network is non-
centrosymmetric and arises from specific attractive interactions
between structural elements that are schematically represented
by bold and dashed lines.16a The main task in the crystal
engineering of such a structure lies therefore in identifying the
complementary elements and eventually a molecule that contains
these elements in appropriate locations. This is not trivial. The
carry-over of molecular symmetry into crystal symmetry (or
even pseudosymmetry) is not expected from Kitaigorodskii’s
theory of close-packing,21 and a majority of trigonal molecules
routinely adopt close-packed crystal structures of low symmetry.
This is so because there is little reason to expect that the
symmetry of the supramolecular web surrounding any molecule
should conform to the internal symmetry of the molecule itself.22
Thus, the key to successfully extending molecular symmetry
into the crystal lies in identifying a design strategy whereby
the patterns of supramolecular interactions possess the (high)
symmetry of the molecule. Of course, it is also necessary to
ensure that the noncentrosymmetry of the two-dimensional
network shown in A extends to the third dimension in the
crystal, that is to bulk noncentrosymmetry. However, this is
not the primary aim of this paper which concentrates on two-
The high anisotropy in these optimized structures13 (pNA,
NPP, POM) does not provide polarization independence, and
(5) (a) Boyd, R. W. Nonlinear Optics; Academic Press: New York, 1992.
(b) NoVel Optical Materials and Applications; Khoo, I.-C.; Simoni, F.;
Umeton, C. Eds.; John Wiley: New York, 1997. (c) Nonlinear Optical
Properties of Organic Molecules and Crystals; Chemla, D. S., Zyss, J.,
Eds.; Academic Press: Boston, 1987. (d) Molecular Nonlinear Optics:
Materials, Physics and DeVices; Zyss, J., Ed.; Academic Press: Boston,
1994. (e) Zyss, J. J. Phys. D. Appl. Phys. 1993, B198, 26.
(6) (a) Bosshard, C.; Sutter, P.; Preˆtre, P.; Hulliger, J, Flo¨rsheimer, M.;
Kaatz, P.; Gu¨nter, P. Organic Nonlinear Optical Materials; Gordon and
Breach: Basel, 1995. (b) Zyss, J. J. Mol. Electron. 1985, 1, 25.
(7) (a) Josse, D.; Don, S. X.; Zyss, J.; Andreazza, P.; Pe´rigaud, A. Appl.
Phys. Lett. 1992, 61, 121. (b) Don, S. X.; Josse, D.; Zyss, J. J. Opt. Soc.
Am. B. 1993, 10, 1708. (c) Kodja, S.; Josse, D.; Samuel, I. D. W.; Zyss, J.
Appl. Phys. Lett. 1995, 67, 3841.
(8) Oudar, J. L.; Hierle, R. J. Appl. Phys. 1977, 48, 2699.
(9) (a) Wang, Y.; Tam, W.; Stevenson, S. H.; Clement, R. A.; Calabrese,
J. C. Chem. Phys. Lett. 1989, 148, 136.
(16) (a) Ledoux, I.; Zyss, J.; Siegel, J.; Brienne, J.; Lehn, J.-M. Chem.
Phys. Lett. 1990, 172, 440. (b) Zyss, J.; Dhenaut, C.; Chau Van, T.; Ledoux,
I. Chem. Phys. Lett. 1993, 206, 409. (c) Dhenaut, C.; Ledoux, I.; Samuel,
I. D. W.; Zyss, J.; Bourgault, M.; Bozec, H. L. Nature 1995, 374, 339. (d)
Zyss, J.; Ledoux, I. Chem. ReV. 1994, 94, 77.
(17) (a) Zyss, J. Nonlinear Opt. 1991, 1, 3. (b) Zyss, J. J. Chem. Phys.
1993, 98, 6583. (c) Zyss, J. In Nonlinear Optics: Fundamentals, Materials
and DeVices; Miyat, S., Ed.; North-Holland: Amsterdam, 1992; pp 33-
48. (d) Zyss, J.; Ledoux, I.; Nicoud, J.-F. In Molecular Nonlinear Optics:
Materials, Physics and DeVices; Zyss, J., Ed.; Academic Press: Boston,
1994. (e) Bre´das, J.-L.; Meyers, F.; Pierce, B.; Zyss, J. J. Am. Chem. Soc.
1992, 114, 4928. (f) Joffre, M.; Yaron, D.; Silbey, R. J.; Zyss. J. J. Chem.
Phys. 1992, 97, 5607. (g) Brasselet, S.; Zyss, J. J. Nonlinear Opt. Phys.
Mater. 1996, 5, 671.
(18) Brasselet, S.; Zyss, J. J. Opt. Soc. Am. B 1998, in press.
(19) An early crystal structure report on the TATB showed it to belong
to the centrosymmetric space group, P1h (Cady, H. H.; Larson, A. C. Acta
Crystallogr. 1965, 18, 485). Potential energy calculations were performed
on the TATB crystal, and two energetically similar noncentrosymmetric
(P1) crystal structures were proposed (Filippini, G.; Gavezzotti, A. Chem.
Phys. Lett. 1994, 231, 86). A recent electron diffraction study of thin
hexagonal crystals of TATB revealed that they belong to the noncentrosym-
metric space group, P31 (Voigt-Martin, I. G.; Li, G.; Yakimanski, A.; Schulz,
G.; Wolff, J. J. Am. Chem. Soc. 1996, 118, 12830).
(10) Barzoukas, M.; Blanchard-Desce, M.; Josse, D.; Lehn, J.-M.; Zyss,
J. Chem. Phys. 1989, 133, 323.
(11) The long rodlike structures (ref 10) are less prone to crystallization
and are useful candidates for poling in amorphous polymer structures. (a)
Singer, K. D.; Andrews, J. H. In Molecular Nonlinear Optics: Materials,
Physics and DeVices; Zyss, J., Ed.; Academic Press: Boston, 1994. (b)
Pinsard-Levenson, R.; Liang, J.; Toussaere, E.; Bouadma, N.; Carenco, A.;
Zyss, J.; Froyer, G.; Guilbert, M.; Pelous, Y.; Bosc, D. Nonlinear Opt. 1993,
4, 233. (c) Michelotti, F.; Toussaere, E.; Pinsard-Levenson, R.; Liang, J.;
Zyss, J. J. Appl. Phys. 1996, 3, 80.
(12) (a) Stegeman, G. I.; Sheik-Bahae, M.; VanStryland, E.; Assanto,
G. Opt. Lett. 1993, 18, 13. (b) Kim, D. Y.; Toruellas, W. E.; Kang, J.;
Brosshard, C.; Stegeman, G. I.; Vidakovic, P.; Zyss, J.; Moerner, W. E.;
Twieg, R.; Bjorklund, G. Opt. Lett. 1994, 19, 868.
(13) Zyss, J.; Oudar, J. L. Phys. ReV. A 1982, 26, 2028.
(14) (a) Zyss, J.; Nicoud, J. F.; Coquillay, M. J. Chem. Phys. 1984, 81,
4160. (b) Ledoux, I.; Lepers, C.; Badan, J.; Pe´rigaud, A.; Zyss, J. Opt.
Commun. 1990, 80, 149. (c) Andreazza, P.; Josse, D.; Le Faucheux, F.;
Robert, M. C.; Zyss, J. Phys. ReV. B 1992, 45, 7640.
(15) (a) Zyss, J.; Chemla, D. S.; Nicoud, J. F. J. Chem. Phys. 1981, 74,
4800. (b) Zyss, J.; Ledoux, I.; Hierle, R.; Raj, R.; Oudar, J. L. IEEE J.
Quantum Electron. 1985, QE-21, 1286. (c) Andreazza, P.; Le Faucheux,
F.; Robert, M. C.; Josse, D.; Zyss, J. J. Appl. Phys. 1990, 68, 8. (d) Josse,
D.; Hierle, R.; Ledoux, I.; Zyss, J. Appl. Phys. Lett. 1988, 53, 23.
(20) It may be noted that statistical octupolar macroscopic order has
recently been demonstrated in host-guest polymer structures by a new, all
optical poling technique. Fiorini, C.; Charra, F.; Nunzi, J. M.; Samuel, J.
D. W.; Zyss, J. Optics Lett. 1995, 20, 2469.
(21) Kitaigorodskii, A. I. Molecular Crystals and Molecules; Academic
Press: New York, 1973.
(22) This distinction between molecular and crystal symmetry has been
appreciated since the earliest days of crystallography. See, for example:
Astbury, W. T.; Bragg, W. H. Ann. Rep. Prog. Chem. 1923, 20, 233.

Crystal Engineering of Some 2,4,6-Triaryloxy-1,3,5-triazines
J. Am. Chem. Soc., Vol. 120, No. 11, 1998 2565
Figure 1. Retrosynthesis of a trigonal octupolar network. A. Trigonal network. B. Recognition of trigonal species. C. Stacked molecular diads of
triazines. D. 2,4,6-Triaryloxy-1,3,5-triazine.
Figure 2. Piedfort units with 3-fold symmetry.
dimensional lamellar structures.23 For a new family of potential
octupolar NLO compounds, even this more limited goal of
dissecting out critically large lower-dimensional noncentrosym-
metric substructures is useful, and we show here that such
trigonal networks are a common structural feature in the triazine
family even when the overall crystal structures are centrosym-
metric.24
Retrosynthetic analysis, as applied to network A (Figure 1),
is not new to molecular synthesis, but its formal application to
crystal engineering, that is supramolecular solid state synthesis,
has been suggested only recently.25 Considering a target crystal
structure as a network, one may logically dissect it to identify
structural units that are constructed with intermolecular interac-
tions and ultimately arrive at the origin of the desired macro-
scopic architecture, that is the supramolecular synthons. These
synthons are the crucial fusing elements of a crystal structure,
and their identification can lead to molecular precursors that
will yield the desired network structure upon crystallization.
Accordingly, our search for a crystalline octupolar nonlinear
material led us to think in terms of synthons such as B.
Crystalline triaryloxy substituted 1,3,5-triazines form stacked
diads called Piedfort units (PUs).26 PUs possessing C3i, C3, and
D3 symmetry (hereafter C3i-PU, C3-PU, and D3-PU) have
been identified27 (Figure 2). Here we use the PU, which in
itself is a supramolecular species C, as the starting point for
the generation of higher level supramolecular assemblies. It
was anticipated that synthon C of C3 or D3 symmetry PUs
should lead to B en route to the desired trigonal network A.
Such an ordered self-assembly was expected on the basis of
specific electrostatic interactions between phenyl rings.28 Ret-
rosynthetic analysis of the PU C lead to substituted 2,4,6-
triphenoxy-1,3,5-triazines D as starting materials for the gen-
eration of trigonal octupolar networks.29 In summary, the
retrosynthetic analysis A w B w C w D had to take into
account that the 3-fold molecular symmetry in D be faithfully
transmitted at each level of the supramolecular hierarchy.
Implied here is that at each level from D to A, only those
combinations of intermolecular interactions that are consistent
with high symmetry be optimized, while the numerous others
that would be expected to lower the crystal symmetry be
suppressed.
In this work we have investigated the crystal structures of
the symmetrical 2,4,6-triaryloxy-1,3,5-triazines 1-6, and have
analyzed the effects of substitution on the network structures
obtained. Halogen derivatives 7 and 8 are briefly mentioned.
We discuss the reasons for the choice of the different molecular
systems, the relationship among the observed crystal structures,
(23) Suggestions for three-dimensional structural control have been
proposed in refs 17b and 17c.
(24) Panunto, T. W.; Urba´nczyk-Lipkowska, Z.; Johnson, R.; Etter, M.
C. J. Am. Chem. Soc. 1987, 109, 7786. These authors have demonstrated
the merits of dissecting a structure, even of a centrosymmetric crystal, into
constituent units which may be polar. These units have been termed
“acentric”, and their identification may be useful in crystal engineering,
provided they are large enough. The concept of local noncentrosymmetry
is therefore a viable one in the intermediate stages of building up a (three-
dimensionally) noncentrosymmetric crystal.
(27) MacNicol, D. D.; Downing, G. A. In ComprehensiVe Supra-
molecular Chemistry; MacNicol, D. D., Toda, F., Bishop, R., Eds.;
Pergamon: Oxford, 1996; Vol. 6, pp 421-464.
(28) (a) Williams, D. E. J. Chem. Phys. 1965, 45, 3770. (b) Burley, S.
K.; Petsko, G. A. J. Am. Chem. Soc. 1986, 108, 7995. (c) Gavezzotti, A.;
Desiraju, G. R. Acta Crystallogr. 1988, B44, 427.
(29) The choice of triazines is also justified from the viewpoint of
molecular engineering because the molecules have appreciable bond dipole
moments. While the molecular 3-fold symmetry results in a net zero dipole
moment, the magnitude of the individual dipoles ensures high octupole
moments.
(25) Desiraju, G. R. Angew. Chem., Int. Ed. Engl. 1995, 34, 4, 2311.
(26) Jessiman, A. S.; MacNicol, D. D.; Mallinson, P. R.; Vallance, I. J.
Chem. Soc., Chem. Commun. 1990, 1619.

2566 J. Am. Chem. Soc., Vol. 120, No. 11, 1998
Table 1. Crystallographic Data for Triazines 1-6
Thalladi et al.
1
2
3
4
5
6
emp. formula
formula wt.
crystal system
space group
a (Å)
b (Å)
c (Å)
C21H15N3O3
357.36
monoclinic
Ia (9)
6.6010(13)
20.903(4)
12.649(3)
90
97.93(3)
90
4
1728.6(6)
1.373
C21H12N3O3Cl3
460.69
trigonal
P3hc1 (165)
12.957(2)
12.957(2)
14.193(2)
90
90
120
4
2063.6(5)
1.483
C21H12N3O3Br3
594.07
trigonal
P3hc1 (165)
13.1205(7)
13.1205(7)
14.4265(9)
90
90
120
4
2150.8(2)
1.835
C24H21N3O3
399.44
trigonal
P3hc1 (165)
13.023(2)
13.023(2)
14.334(3)
90
90
120
4
2105.3(6)
1.260
C21H9N3O3Cl6
564.01
trigonal
P3hc1 (165)
13.2925(6)
13.2925(6)
15.9132(9)
90
90
120
4
2435.0(2)
1.538
C27H27N3O3
441.52
trigonal
P3hc1 (165)
13.2987(13)
13.2987(13)
15.962(2)
90
90
120
4
2444.8(4)
1.200
R (deg)
â (deg)
γ (deg)
Z
V (Å3)
D
calc (Mg/m3)
R1
wR2
GOF
0.0832
0.2033
1.126
0.0692
0.1546
1.156
0.0448
0.1034
1.070
0.0748
0.1624
0.996
0.0956
0.2738
1.111
0.0507
0.1320
1.009
N-totala
N-independentb
variables
crystal shapec
solvent
Ck* d
3685
2044
244
needle
CHCl3
0.69
2226
902
96
h-rod
CHCl3
0.66
2330
947
96
h-rod
CH2Cl2
0.66
2263
920
96
h-needle
CH3CO2C2H5
0.65
9070
1452
104
h-tablet
CHCl3
0.63
2437
1072
106
h-rod
CH3CO2C2H5
0.64
a N-total is the total number of reflections collected. b N-independent is the number of independent reflections. c h-rod, h-needle, and h-tablet
stand for hexagonal rod, hexagonal needle, and hexagonal tablet. d Ck* is the packing fraction calculated from the program PLATON.34
Figure 3. Laboratory reference frame (X,Y,Z) for the Harmonic Light
Scattering (HLS) depolarization ratio measurements. The fundamental
and our observations on molecular and crystal nonlinearities in
this class of compounds.
incoming beam propagates along Z and is polarized in the (X,Y) plane.
The observation direction is along Y with the polarization analysis along
X or Z.
Experimental Section
NLO Measurements. The lack of a permanent dipole moment in
the molecules studied here makes the classical Electric Field Induced
SHG (EFISH) experiment unsuited to the determination of molecular
nonlinearities, â.32 We therefore performed measurements in solution
by use of the HLS experiment.16c,d,33 This incoherent nonlinear process
results from the effect of room temperature fluctuations in centrosym-
metric media on the optically nonlinear response and allows for the
determination of the required orientational average âXâ of the â
quadratic hyperpolarizability tensor. The â2 term of interest here has
the additional advantage, with respect to EFISH, of being polarization
dependent and thus reflects the tensorial nature of the molecular
hyperpolarizability.
General Methods. Commercially available cyanuric chloride and
the appropriate phenols were used as received without further purifica-
tion. Reagent grade solvents were used for extraction, and distilled
solvents were used for all recrystallizations.
General Procedure for the Synthesis of 2,4,6-Triaryloxy-1,3,5-
triazines.30 Cyanuric chloride (10 mmol) was heated with a small
excess of the appropriate phenol (35 mmol) at 185-210° C for 5 h
under a reflux air condenser. HCl gas was evolved vigorously during
the first few hours. The crude reaction product was extracted with
boiling EtOH leaving a residue of crude triaryloxytriazine. This residue
was then recrystallized from CHCl3 to give the pure crystalline product
in 80-90% yield. All the triazines were characterized by their IR and
NMR spectra.
X-ray Crystallography. Single crystals suitable for X-ray diffrac-
tion were grown from common organic solvents (Table 1). Data for 1
were collected on an Enraf-Nonius FAST area detector diffractometer
and for 2-4 and 6 on a Siemens P4 diffractometer, while those for 5
were collected on a SMART diffractometer using Mo-KR radiation.
The structure solutions and refinements were carried out using the
programs SHELXS-8631a and SHELXL-9331b built-in with the Siemens
SHELXTL (Version 5.03) package. The details of the X-ray data
collection, structure solution, and refinement are given in the Supporting
Information.
In the experimental setup (Figure 3), the fundamental pulsed Nd3+
:
YAG near-IR source at 1.064 µm corresponds to a scattered second
harmonic wavelength in the transparent region of the molecules of
interest. Notably, all the compounds in this study are completely
transparent in the visible region. The scattered intensity, collected at
90° from the incoming beam main axis, can be expressed as I2ω
)
G(N1 â1 + N2 â2 )(Iω)2 for a solution with N2 (respectively N1)
molecules per cm3 of solute (respectively solvent; CHCl3 in the present
case), where the G coefficient includes geometrical factors and
2
2
(31) (a) Sheldrick, G. M. SHELXS-86: Program for the Solution of
Crystal Structures; University of Go¨ttingen: Germany, 1986. (b) Sheldrick,
G. M. SHELXL-93: An Integrated System for Refining and Displaying
Crystal Structures from Diffraction Data; University of Go¨ttingen: Germany,
1993.
(30) Schaefer, F. C.; Thurston, J. T.; Dudley, J. R. J. Am. Chem. Soc.
1951, 73, 2990.
(32) Oudar, J. L.; Chemla, D. S. J. Chem. Phys. 1977, 66, 2664.

Crystal Engineering of Some 2,4,6-Triaryloxy-1,3,5-triazines
J. Am. Chem. Soc., Vol. 120, No. 11, 1998 2567
Table 2. Geometrical Parameters for Various Interactions in the
Structure of 1
interactiona
Db (Å)
dc (Å)
θd (deg)
ortho-C-H‚‚‚O
ortho-C-H‚‚‚O
ortho-C-H‚‚‚O
ortho-C-H‚‚‚N
ortho-C-H‚‚‚N
ortho-C-H‚‚‚N
para-C-H‚‚‚Xe
para-C-H‚‚‚X
para-C-H‚‚‚X
meta-C-H‚‚‚X
meta-C-H‚‚‚X
π‚‚‚πf
3.455
3.544
3.566
3.541
3.449
3.468
3.556
3.924
3.968
3.759
3.805
3.301
2.509
2.603
2.618
2.535
2.401
2.461
2.699
3.289
3.341
2.936
2.994
145.63
145.20
146.14
154.57
163.23
154.70
135.96
118.67
118.24
133.29
132.27
a All the C-H bonds are normalized to the standard neutron length
(1.08 Å) along the C-H vector. b D is the distance between C and the
acceptor (O, N, or ring centroid). c d is the distance between H and the
acceptor (O, N, or ring centroid). d θ is the angle at H in C-H‚‚‚X (X
) O, N, or ring centroid). e Herringbone interactions are expressed as
C-H‚‚‚X, where X is the centroid of the aromatic ring acting as the
C-H acceptor. f For π‚‚‚π interactions, D is the perpendicular stacking
distance.
Figure 4. View down [100] showing a D3-PU in 1. C-H‚‚‚O and
C-H‚‚‚N bonds are shown as dashed lines. Molecules related by a-glide
form D3-PUs.
identified (Figure 4). The two molecules in the D3-PU are
related by an a-glide and are held together by π‚‚‚π stacking
interactions between triazine rings and also by C-H‚‚‚O and
C-H‚‚‚N hydrogen bonds.36 One of the molecule in the D3-
PU donates three C-H‚‚‚O hydrogen bonds to the other which
in turn donates three C-H‚‚‚N hydrogen bonds back to the first.
The molecules of 1 use ortho H-atoms of the phenoxy groups
for such hydrogen bonding. In effect, a D3-PU is stabilized
by π‚‚‚π interactions and six weak (C-H‚‚‚O and C-H‚‚‚N)
hydrogen bonds. The geometrical parameters for these interac-
tions are given in Table 2.
experimental correction terms and where Iω is the fundamental incident
intensity. Further insight on molecular symmetry is available from
polarized configurations of the HLS experiment.17g The overall â2
coefficient has two independent components, namely â2
(respec-
XXX
tively â2 ) measured for an analyzer parallel (respectively perpen-
ZXX
dicular) to the fundamental incident polarization. In this context, the
evaluation of the depolarization ratio17g D ) â2 / â2
helps
ZXX
XXX
determine the dipolar, octupolar, or mixed character of the nonlinear
hyperpolarizability. This terminology arises from irreducible repre-
sentation of the â tensor in a general spherical formalism. It has been
shown18 that the octupolar symmetry corresponds to a minimum of
the depolarization ratio D ) 2/3.
The mean plane of the D3-PU is almost parallel to (100)
and makes an angle of 4.2° with it. A layered structure parallel
to (100) arises from the “T”-geometry of the herringbone
interactions between the phenyl groups (Figure 5, Table 2). All
the para H-atoms on the phenyl groups and two of the meta
H-atoms are engaged in these herringbone interactions and thus
play a very important role in the generation of the trigonal layer
structure.
Results and Discussion
Table 1 gives the details of the crystallographic information
for 1-6. While crystals of triazine 1 are monoclinic, triazines
2-6 adopt trigonal space groups. It is of note that triazines
1-6 contain the trigonal network A in their crystal structures.
Triazines 7 and 8 belong to the hexagonal crystal system and
possess large cavities in their structures. The packing fractions34
for triazines 1-6 (Ck*, Table 1) are in the range 0.63-0.69,
typical of many close-packed organic solids. The description
and analysis of the individual crystal structures is now presented.
Triazine 1 Crystallizes as a Unique Noncentrosymmetric
Structure. Triazine 1 crystallizes in the noncentrosymmetric
space group Ia (number 9).35 D3-symmetric Piedfort units (D3-
PUs) formed by an assembly of two molecules of 1 may be
The D3-PUs are translationally stacked along [100] in an
eclipsed manner and produce a columnar structure. Signifi-
cantly, all the C-H‚‚‚O bonds run in the [100] direction whereas
all the C-H‚‚‚N bonds run in the opposite [1h00] direction. Since
these C-H‚‚‚O and C-H‚‚‚N hydrogen bonded strands are
polar, such translated stacking results in overall noncentrosym-
metry with a macroscopic nonlinear coefficient one-tenth that
of urea.37 Three-dimensional structural control is a major
endeavor in crystal engineering today, and it is realized here
because the herringbone and hydrogen bonding interactions
occur in roughly orthogonal directions (Figure 6). It may be
noted that the D3-PUs in 1 are only approximately D3-
symmetric and hence the network generated is quasi-trigonal
and not perfectly trigonal. The supramolecular synthesis of
octupolar network A (Figure 1) leading to the crystal structure
of SHG active 1 underscores the importance of logic-driven
supramolecular retrosynthesis in the quest for target networks
with specific properties.
(33) (a) Terhune, R. W.; Maker, D.; Savage, C. M. Phys. ReV. Lett. 1965,
14, 681. (b) Maker, D. Phys. ReV. A 1970, 1, 923. (c) Clays, K.; Persoons,
A. Phys. ReV. Lett. 1991, 66, 2980. (d) Kaatz, P.; Sheldon, D. P. ReV. Sci.
Instrum. 1996, 67(4), 1439.
(34) Platon-97: Spek, A. L. Acta Crystallogr. 1990, A46, C34.
(35) The X-ray diffraction on the crystals of 1 yielded an initial triclinic
cell with dimensions a ) 6.601(2), b ) 12.426(6), c ) 12.649(4) Å, R )
118.23(2), â ) 97.93(5) and γ ) 101.22(3)o. The structure was solved and
refined in the space group P1, with two molecules in the asymmetric unit.
During the refinements two of the C-atoms became nonpositive-definite. It
was identified from an analysis of the refined structure that the two
symmetry independent molecules are stacked along [100] and they could
be related by an a-glide perpendicular to [010]. This and the pseudo-
hexagonal packing of the structure led us to search for the higher symmetry
in the crystal. It was found that a higher symmetry space group is indeed
possible, and this is the one reported in this paper.
A comparison of the structure of 1 to that of cis,cis-
cyclohexane-1,3,5-tris(R-picolin-6-yl)carboxamide, 9, is perti-
(36) (a) Desiraju, G. R. Acc. Chem. Res. 1991, 24, 290. (b) Desiraju, G.
R. Acc. Chem. Res. 1996, 29, 441. (c) Steiner, T. Cryst. ReV. 1996, 6, 1.
(37) Details are provided in the section on NLO measurements.

2568 J. Am. Chem. Soc., Vol. 120, No. 11, 1998
Thalladi et al.
Figure 5. View down [100] showing the trigonal layer structure in 1. Molecules at different heights along [100] are respectively unshaded and
shaded. The packing of D3-PUs within the layer is governed by herringbone interactions. The D3-PUs are translation related along [001].
nent.38 The structure of 9, wherein successive molecules are
interconnected by three N-H‚‚‚O hydrogen bonds between
amide functionalities, shows a supramolecular columnar struc-
ture similar to that found in 1. In the noncentrosymmetric
structure of 9, the hydrogen bonds run along a polar direction
(as they do in 1); 9 has a trigonal octupolar network structure
similar to 1, and it has an SHG activity of 0.06 × urea. While
the columnar structure in the triamide is stabilized by three
strong (N-H‚‚‚O) hydrogen bonds, the structure of 1 is
stabilized by six weak (three C-H‚‚‚O and three C-H‚‚‚N)
hydrogen bonds. Thus, we have serendipitously produced a
C-H‚‚‚O/C-H‚‚‚N topological equivalent of an N-H‚‚‚O
based supramolecular structure. This indicates that when weak
hydrogen bonds are employed collectiVely they can work as
effectiVely as their stronger counterparts.
The following points emerge from an analysis of the structure
of 1: (1) Such a compound may be a good starting point for
producing PUs. (2) The aggregation of PUs is governed by a
(38) Fan, E.; Yang, J.; Geib, S. J.; Stoner, T. C.; Hopkins, M. D.;
Hamilton, A. D. J. Chem. Soc., Chem. Commun. 1995, 1251.

Crystal Engineering of Some 2,4,6-Triaryloxy-1,3,5-triazines
J. Am. Chem. Soc., Vol. 120, No. 11, 1998 2569
Table 3. Geometrical Parameters for Various Interactions Found
in the Structures of 2-6
D3-PU
C3i-PU
Db
(Å)
dc
θd
Db
dc
θd
triazine interactiona
(Å) (deg) (Å)
(Å) (deg)
2
3
4
5
6
o-C-H‚‚‚O
o-C-H‚‚‚N
p-C-H‚‚‚Xe
3.655 2.682 149.64 3.344 2.685 118.96
3.738 2.892 135.33 3.640 2.672 148.98
3.720 2.760 147.92
m-C-H‚‚‚Cl 3.847 2.767 178.37
π‚‚‚πf
o-C-H‚‚‚O
o-C-H‚‚‚N
p-C-H‚‚‚X
m-C-H‚‚‚Br 3.910 2.835 173.58
π‚‚‚π
o-C-H‚‚‚O
o-C-H‚‚‚N
p-C-H‚‚‚X
m-C-H‚‚‚CH3 4.160 3.230 178.96
π‚‚‚π
3.557
3.567
3.636 2.644 152.50 3.373 2.668 122.42
3.791 2.972 132.94 3.685 2.702 151.21
3.835 2.846 153.10
3.620
3.618
3.691 2.735 147.39 3.410 2.755 118.88
3.727 2.829 140.64 3.689 2.726 148.64
3.810 2.840 149.50
3.569
3.627
o-C-H‚‚‚O
o-C-H‚‚‚N
p-C-H‚‚‚X
Cl‚‚‚Clg
3.850 2.853 152.53 3.600 2.837 127.66
4.030 3.119 142.63 3.920 2.882 161.23
4.049 3.038 156.03
3.120
4.025
π‚‚‚π
3.948
o-C-H‚‚‚O
o-C-H‚‚‚N
p-C-H‚‚‚X
3.860 2.813 163.28 3.634 2.874 127.51
4.022 3.106 143.00 3.945 2.934 155.91
4.067 3.081 156.03
h
CH3‚‚‚CH3
π‚‚‚π
3.478
4.029
3.970
a All the C-H bonds are normalized to the standard neutron length
(1.08 Å) along the C-H vector. b D is the distance between C and the
acceptor (O, N, Cl, Br, methyl C-atom or centroid). c d is the distance
between H and the acceptor (O, N, Cl, Br, methyl C-atom or centroid).
d θ is the angle at H in C-H‚‚‚X (X ) O, N, Cl, Br, methyl C-atom
or centroid). e Herringbone interactions are expressed as C-H‚‚‚X,
where X is the centroid of the aromatic ring acting as the C-H acceptor.
f For π‚‚‚π interactions D is the stacking distance. g For Cl‚‚‚Cl
interactions D is the distance between Cl-atoms. h For methyl-methyl
close packing, the distance between two C-atoms of the interacting
methyl groups is expressed as D.
4 are discussed together because of their structural similarity.
Triazines 2-4 crystallize in the trigonal space group P3hc1
(number 165) and are isostructural. The molecules lie on 3-fold
axes, and D3-PUs, assembled from two stacked molecules
related by a 32-axis, are found in these structures (Figure 7a)
as in 1. Unlike the D3-PUs in 1, the PUs found in 2-4
maintain perfect D3 symmetry. The two molecules in the D3-
PU are related by π‚‚‚π stacking interactions of the central rings
as well as by C-H‚‚‚O and C-H‚‚‚N hydrogen bonds. The
geometrical parameters for various interactions found in these
structures are given in Table 3. Each molecule in the D3-PU
donates and accepts three C-H‚‚‚O bonds via one of the two
symmetry independent ortho H-atoms. The same ortho H-atom
is also involved in the formation of a longer, bifurcated39
C-H‚‚‚N hydrogen bond with the N-atom of the central triazine
ring. Effectively, each D3-PU is stabilized by stacking
interactions and 12 weak hydrogen bonds, 6 C-H‚‚‚O, and 6
C-H‚‚‚N bonds.
Figure 6. Two views of eclipsed stacking of D3-PUs in 1. (a) View
down [100]. The unshaded D3-PU is above the shaded one. (b) Side
view of the same stacked pair of D3-PUs. C-H‚‚‚O and C-H‚‚‚N
bonds are indicated as dashed lines. The sense of these hydrogen bonds
maintains the polarity. The shading scheme adopted is C - crescent,
H - unshaded, N - dotted, O - hatched. The same shading scheme
is followed in the remaining figures.
combination of π‚‚‚π stacking and C-H‚‚‚O/C-H‚‚‚N interac-
tions. The ortho H-atoms must be available as hydrogen bond
donors. (3) The herringbone interactions are critical for the
trigonal layer structure and involve participation of all the para
and a third of the meta H-atoms. The trigonal network is thus
expected to be the most sensitive to substitution in the para
position. (4) The only site on the phenoxy group that appears
not to disrupt formation of the D3-PUs and the trigonal network
is the meta position. Thus, we turned our attention to triazines
2-6.
The mean planes of the D3-PUs in 2-4 are constrained by
crystallographic symmetry to be parallel to (001). Unlike in 1
where three para and two meta hydrogen atoms are involved
in herringbone intractions, only the para H-atoms are involved
in 2-4, and therefore the octupolar networks found in these
triazines are ideally trigonal (Figure 8). This is reflected in their
macroscopic, crystallographic 3-fold symmetry. Each D3-PU
Triazines 2-4 Produce Perfectly Symmetric D3-PUs in
Their Crystal Structures. The structures of 2,4,6-tris(3-
chlorophenoxy)-1,3,5-triazine 2, 2,4,6-tris(3-bromophenoxy)-
1,3,5-triazine 3, and 2,4,6-tris(3-methylphenoxy)-1,3,5-triazine
(39) Jeffrey, G. A.; Saenger, W. Hydrogen Bonding in Biological
Structures; Springer-Verlag: Berlin, 1991.

Figure 7. View down [001] showing the Piedfort units in 2-4. (a) D3-PUs. C-H‚‚‚O bonds are indicated as dashed lines. (b) C3i-PUs. C-H‚‚‚N bonds are indicated as dashed lines. Bifurcation is shown
in neither case.

Crystal Engineering of Some 2,4,6-Triaryloxy-1,3,5-triazines
J. Am. Chem. Soc., Vol. 120, No. 11, 1998 2571
Figure 8. View down [001] showing the trigonal layer structure in 2. Molecules at different heights along [001] are respectively unshaded and
shaded. The packing of D3-PUs within the layer is governed by herringbone interactions. The D3-PUs are translation related parallel to (001).
Contrast this with Figure 5.
in Figure 8 is connected to six neighbors within the layer through
12 herringbone interactions. Since the D3-PUs in 2-4 are
strictly D3-symmetric, the trigonal network structure is also D3-
symmetric.40
The successive D3-PUs in 2-4 stack along [001] in a
staggered manner because eclipsed stacking, as in 1, will lead
to unfavorable steric interactions with the meta substituents. Two
views of the staggered stacking of the 3h axis related D3-PUs
are shown in Figure 9. Such a staggered stacking of the D3-
PUs alleviates steric interactions between the bulky meta
substituents on the phenoxy groups.41 The isostructural nature
of 2-4 illustrates the similar steric demand of chloro, bromo,
and methyl groups. Incidentally, the staggered stacking of D3-
PUs defines a new PU with C3i symmetry in these structures.
Piedfort units of C3i symmetry (C3i-PUs) result if the lower
molecule of the upper D3-PU and the upper molecule of the
lower D3-PU are selected from a staggered stacked pair of D3-
PUs. Figure 7b shows the C3i-PUs found in the structures of
2-4.
The two 3h-related molecules in the C3i-PU are held by π‚‚‚π
stacking and also by C-H‚‚‚O and C-H‚‚‚N hydrogen bonds.
One of the two symmetry independent ortho H-atoms is
involved in bifurcated C-H‚‚‚O and C-H‚‚‚N hydrogen
bonding between D3-PUs. The other symmetry independent
ortho H-atom on the phenoxy group is bifurcated between
(41) Based on similar arguments Schwiebert et. al. have suggested in a
recent paper (ref 2b) that chloro, bromo, and methyl derivatives of
2-benzimidazolones show offset stacking of tapes as opposed to perfect
stacking of tapes in the unsubstituted hydrogen compound.
(40) A similar situation is observed in the structure of hexaphenyl-
melamine (Lindeman, S. V.; Shklover, V. E.; Struchkov, Yu. T.; Kuznetsov,
S. N.; Pankratov, V. A. Kristallografiya 1982, 27, 65).

Figure 9. Two views of staggered stacking of D3-PUs in 2-4. (a) View down [001]. The unshaded D3-PU is above the shaded one. (b) Side view of the stacked pair of D3-PUs. C-H‚‚‚O and C-H‚‚‚N
bonds are indicated as dashed lines. Contrast these with Figure 6.

Crystal Engineering of Some 2,4,6-Triaryloxy-1,3,5-triazines
J. Am. Chem. Soc., Vol. 120, No. 11, 1998 2573
Figure 10. View down [001] showing the Piedfort units in 5 and 6. (a) D3-PUs. C-H‚‚‚O bonds are indicated as dashed lines. (b) C3i-PUs.
C-H‚‚‚N bonds are indicated as dashed lines. Bifurcation is shown in neither case.
C-H‚‚‚O and C-H‚‚‚N hydrogen bonds between the two
molecules in the C3i-PU. The data in Table 3 show that
C-H‚‚‚N bonds are relatively stronger than C-H‚‚‚O bonds
in C3i-PUs, while the trend is opposite in D3-PUs. Thus, the
O- and N-atoms are responsible for effective hydrogen bonding
in the alternating D3-PUs and C3i-PUs. In contrast we note
that bifurcation is absent in 1 which has better C-H‚‚‚O and
C-H‚‚‚N bonds.
The continuous, staggered stacking of D3-PUs along [001]
in 2-4 produces a supramolecular columnar structure wherein
the C-H‚‚‚O and C-H‚‚‚N hydrogen bonded strands do not
maintain structural polarity as in 1. This is because the
successive D3-PUs are staggered such that inversion related
molecules are stacked. Accordingly, successive trigonal octu-
polar layers are inversion related in 2-4, and the structures are
centrosymmetric.
bulky meta substituents. If attractive C-H‚‚‚Cl/Br interactions
are structure determining, the crystal structures of 2 and 3 should
be different from the methyl derivative 4 because C-H‚‚‚Me
contacts are not attractive.42 That 2-4 are isostructural and
the fact that the intermolecular contacts are long (Table 3)
indicates that layer inversion is a consequence of the similar
space demand of the meta substituent.43 To further examine
the endurance of this structure type toward meta substitution,
the structures of 5 and 6 were determined.
Chloro‚‚‚Chloro and Methyl‚‚‚Methyl Close Packing in
the Structures of 5 and 6. The structures of 2,4,6-tris(3,5-
dichlorophenoxy)-1,3,5-triazine 5 and 2,4,6-tris(3,5-dimethyl-
phenoxy)-1,3,5-triazine 6 are isostructural to 2-4 and belong
(42) Sarma, J. A. R. P.; Desiraju, G. R. Acc. Chem. Res. 1986, 19, 222.
(43) The corresponding fluoro triazine could not be crystallized for single-
crystal X-ray analysis. Based on the powder X-ray spectrum of the material
it was not possible to distinguish between noncentrosymmetric 1 and
centrosymmetric 2. However, the absence of a powder SHG signal suggests
that it adopts a centrosymmetric structure.
Such stacking inversion could be a result either of attractive
C-H‚‚‚Cl/Br interactions or of the space filling demand of the

2574 J. Am. Chem. Soc., Vol. 120, No. 11, 1998
Thalladi et al.
Figure 11. View down [001] showing the trigonal layer structure in 5. Molecules at different heights along [001] are respectively unshaded and
shaded. The packing of D3-PUs within the layer is governed by herringbone interactions. The D3-PUs are translation-related parallel to (001).
Compare this with Figure 8.
to the same trigonal space group P3hc1 (number 165). In these
structures too the triazine molecule sits on a 3-fold axis and
D3-PUs and C3i-PUs may be identified (Figures 10 and 11).
The π‚‚‚π stacking separation and H‚‚‚O/N distances in 5 and
6 are longer when compared with the corresponding distances
in triazines 2-4 (Table 3). A supramolecular columnar structure
is obtained by the staggered stacking of D3-PUs along [001]
(Figure 12).
The successive trigonal layers in 5 and 6 are inversion related
and this reduces the repulsive interactions between the meta
substituents. Interestingly in 5 the inverted layers are related
by short Cl‚‚‚Cl geometries between the two symmetry inde-
pendent Cl-atoms (Cl‚‚‚Cl: 3.12 Å). Usually, short Cl‚‚‚Cl
interactions have a structure influencing effect,42,44 and in such
cases the corresponding methyl analogue has a different
structure.45 When the Cl-atom merely has a space-filling role,
the chloro and methyl groups (volumes: 20 and 24 Å3,
respectively) can be interchanged without much variation in the
structure. That 5 and 6 are isostructural suggests that the chloro
group is not involved in specific, attractive Cl‚‚‚Cl interactions.46
The short Cl‚‚‚Cl separation in 5 is rationalized as a repulsive
contact that is forced to accommodate itself within the extremely
robust trigonal packing. This is corroborated by the almost
linear C-Cl‚‚‚Cl-C geometry (type I) and not the distinctive
inclined geometry (type II) that is characteristic of the polariza-
tion induced and stabilizing Cl‚‚‚Cl contact.47
(45) Desiraju, G. R.; Sarma, J. A. R. P. Proc. Ind. Acad. Sci. (Chem.
Sci.) 1986, 96, 599.
(46) (a) Jones, W.; Theocharis, C. R.; Thomas, J. M.; Desiraju, G. R. J.
Chem. Soc., Chem. Commun. 1983, 1443. (b) Theocharis, C. R.; Desiraju,
G. R.; Jones, W. J. Am. Chem. Soc. 1984, 106, 3606.
(44) Desiraju, G. R.; Parthasarathy, R. J. Am. Chem. Soc. 1989, 111,
8725.

Crystal Engineering of Some 2,4,6-Triaryloxy-1,3,5-triazines
J. Am. Chem. Soc., Vol. 120, No. 11, 1998 2575
Figure 12. Two views of staggered stacking of D3-PUs in 5 and 6. (a) View down [001]. The unshaded D3-PU is above the shaded one. (b) Side
view of the stacked pair of D3-PUs. C-H‚‚‚O and C-H‚‚‚N bonds are indicated as dashed lines. Compare this with Figure 9.
It is now instructive to compare the crystal structure of 1
with those of 2-6. Triazine 1 yields a polar arrangement of
monoclinic symmetry Ia. The molecules, with a chiral con-
formation in the crystal, are related by translation within
noncentrosymmetric layers and are stacked along [100] and
related by a-glide symmetry. Minor changes at the molecular
level give 2-6 which yield noncentrosymmetric layers akin to
1, but the molecules are stacked along [001] by 3h and 32
symmetry in the centrosymmetric space group P3hc1. Though
1 is monoclinic, it has quasi-trigonal octupolar layers comparable
to the perfectly D3-symmetric layers in 2-6 (compare Figures
5, 8, and 11). Herringbone interactions between the D3-PUs
in these layers are found in 1 as well as in 2-6. The stacking
distances vary linearly with size, and this is attributed to the
space demand of the meta substituents: the shortest and the
longest stacking distances are found in 1 and 6, respectively
(Tables 2 and 3). The similarities and differences between the
network in 1 compared to those in 2-6 may be visualized by
the stacked shaded-unshaded D3-PU molecular diads in
Figures 5, 8, and 11. While the herringbone interactions within
and between the linear arrays are different in 1, shaded-to-shaded
and unshaded-to-unshaded within a linear array and shaded-to-
unshaded and unshaded-to-shaded between the linear arrays
(Figure 5) the interactions are all the same in 2-6, always
shaded-to-shaded and unshaded-to-unshaded (Figures 8 and 11).
This differentiates the quasi and perfect D3 symmetry of the
trigonal networks in 1 and 2-6, respectively. Two-dimensional
supramolecular octupolar nonlinearities are well demonstrated
in the structures of 2-6, and all these triazines are excellent
examples of two-dimensionally chiral systems.24 The main
difference between 1 and 2-6 lies in the mode of stacking of
D3-PUs: eclipsed stacking is observed in 1, while staggered
(47) Pedireddi, V. R.; Reddy, D. S.; Goud, B. S.; Craig, D. C.; Rae, A.
D.; Desiraju, G. R. J. Chem. Soc., Perkin Trans. 2 1994, 2353.

2576 J. Am. Chem. Soc., Vol. 120, No. 11, 1998
Thalladi et al.
Figure 13. Hexagonal networks in the crystal structure of 7. Notice that molecular synthons (triazine ring) and supramolecular synthons (Cl3
trimer) alternate to produce the hexagonal sheet structure.
Table 4. Experimental Results at the Microscopic Level for
the two âyyy and âyxx coefficients with âyyy ) -âyxx, y lying
Triazines 1-6 from Harmonic Light Scattering Measurements at
along one of the three charge-transfer axes. After isotropic
1.064 µm (CHCl3)a
averaging, the microscopic âyyy values can be inferred from the
orientational averaging relation17g,18 â2 ) 8/21â2 in the
1
2
3
4
5
6
yyy
octupolar case (Table 4).
18 ( 2 19 ( 2 20 ( 4 15 ( 2 22 ( 5 21 ( 3
â2 (10-30 esu)
x
yyy (10-30 esu)
29.2
30.7
32.4
24.3
35.6
34
An estimation of the nonlinear efficiency of crystalline 1,
using the oriented gas model, was performed noting that it is
the only compound in the series that crystallizes in a non-
centrosymmetric structure. Triazine 1 exhibited a SHG powder
signal measured as one tenth that of the urea powder signal.
Triazines 2-6 that crystallize in centrosymmetric structures did
not show any measurable SHG signal.
â
a The microscopic tensorial coefficients, âyyy are deduced from these
measurements.
stacking is observed in 2-6. This leads to the following
differences between the structure of 1 and those of 2-6: (1)
the absence and presence of C3i-PUs, (2) the absence and
presence of interactions between the meta substituents, (3) the
absence and presence of inversion between layers, and (4) the
presence and absence of bulk noncentrosymmetry.
Change in Structure Type for the Para-Substituted 7. The
similarity between the structure types exemplified by 1 and by
2-6 is the trigonal network stabilized by herringbone interac-
tions between the para H-atoms and the phenyl rings. To better
understand the role of the para H-atoms in establishing the
trigonal networks, the crystal structures of triazines 7 and 8 were
determined. 2,4,6-Tris(4-chlorophenoxy)-1,3,5-triazine, 7, and
2,4,6-tris(4-bromophenoxy)-1,3,5-triazine, 8, are isostructural
and crystallize in the hexagonal space group P63/m (number
176). Crystals of 7 and 8 grow as solvates from chloroform,
dichloromethane, ethyl acetate, benzene, and toluene. However,
they readily lose their solvent of recrystallization and become
opaque when taken out of the mother liquor. Also, they could
not be cooled without destroying them. The structural results
at room temperature are of low accuracy, and the solvent
molecule could not be located. Since 7 and 8 are completely
isostructural, we report the structure only of 748 and that too
Nonlinear Characterization of Triazines 1-6 at the
Molecular and Crystal Level. The molecular nonlinear
coefficients, â2 of compounds 1-6 determined from HLS
x
measurements at 1.064 µm are given in Table 4. The molecular
hyperpolarizability of these compounds is of the same order of
magnitude as that of the classical dipolar pNA molecule
(
â2 ) 10 × 10-30 esu) implying that there is a comparable,
x
moderately efficient charge transfer in these molecular struc-
tures. The depolarization ratio D ) â2 / â2
is about 0.63
ZXX
XXX
for the different triazines, and this confirms their octupolar
symmetry since the calculated D value for pure octupolar
symmetry is 0.67.18 Assuming a planar octupolar structure for
these compounds, the resulting â tensor components reduce to

Crystal Engineering of Some 2,4,6-Triaryloxy-1,3,5-triazines
J. Am. Chem. Soc., Vol. 120, No. 11, 1998 2577
only in the context of demonstrating the structural sensitivity
of the para position of the phenyl ring in this family of
compounds.
area of research. In particular, this work shows that an octupolar
NLO substance may be realized at the supramolecular level.
Octupolar systems offer advantages over dipolar systems with
regard to manipulation of packing characteristics because they
can be expressed as higher-dimensional networks and as such
a greater proportion of the structure is amenable to retroanalysis.
We find here that the generation of a target supramolecular
network may be achieved not only from a molecular species
but also from a convergent, relatively small supramolecular
species. Typical examples are the herringbone interaction-
based-synthons in 1-6. Keeping this template intact, both
structural mimicry and structural diversity may be manipulated
at will in this family of triazines.
The crystal structure of triazine 7 has large hexagonal cavities
with a diameter of 14.25 Å (Figure 13). The para Cl-atom
forms intermolecular contacts with two other symmetry related
atoms through trimeric Cl3 supramolecular synthons. These
synthons are structurally equivalent to the well-known (OH)3
trimer synthon.49 Cl‚‚‚Cl contacts have two preferred geom-
etries, type I (θ1 = θ2) or type II (θ1 = 180°, θ2 = 90°) where
θ1 and θ2 are the C-X‚‚‚X angles.50 While space filling type
I contacts are found in the structure of 5, the contact present in
7 corresponds to the polarization induced type II category.47
The trimer X3 synthon is cyclic, and each halogen is ap-
propriately polarized, behaving both as a donor and an acceptor.
Such cooperativity39 effects could enhance the strength of the
X‚‚‚X interactions.51 It is clear that adoption of a particular
structure type in the triazine family is very sensitive to
substitution at the para position on the phenoxy groups. So
long as the para H-atom is intact, a robust polar trigonal layer
results. A para-halogen substitution results in a hexagonal layer
structure. In summary, the supramolecular retrosynthetic ap-
proach provides excellent structural control in two dimensions
even as we realize that control in the third dimension still
remains to be achieved in a general sense.
At the outset, we have stated that contemporary crystal
engineering can be motivated by strategic/methodological,
aesthetic, or application-oriented concerns. And yet, these
motivations stand in no real contradiction to each another. The
present paper illustrates that rather than being in conflict, these
three fundamental concerns can converge to yield a family of
solids that has much potential for development as functionalized
materials.
Acknowledgment. G.R.D. and A.N. acknowledge financial
support from the Department of Science and Technology,
Government of India (SP/S1/G19/94) and helpful assistance
from Molecular Simulations, Cambridge, England and San
Diego, U.S.A. R.B. thanks the Fonds der Chemischen Industrie
for support. V.R.T. thanks the Department of Atomic Energy,
Government of India for fellowship support and the Heinrich
Hertz-Stiftung des Landes Nordrhein-Westfalen for travel as-
sistance. H.-C.W. acknowledges the Prof. Werdelmann-Stiftung
im Stifterverband fu¨r die Deutsche Wissenschaft for travel
assistance. A.K.K. and H.L.C. acknowledge financial support
from NIH Grant CA-10925.
Conclusions
Current advances in crystal engineering and nonlinear optics
show that the intersection of these fields constitutes a fertile
(48) Crystal data for 7: Hexagonal, P63/m, Z ) 2, a ) 15.364(3) Å, c
) 6.855(2) Å, V ) 1401.3(6) Å3, Dc ) 1.092 Mg m-3, µ(Mo-KR) ) 0.348
mm-1, F(000) ) 468, T ) 293 K, 1.53 e θ e 24.99°, -1 e h e 15, -18
e k e 1, -1 e l e 8, 2352 measured reflections, 898 independent
reflections, 58 variables, full-matrix least-squares refinement on F2, R )
0.1236, wR2 ) 0.3172, GOF ) 1.280, C, N, O, Cl anisotropic, H isotropic.
(49) Ermer, O.; Ro¨bke, C. Angew. Chem., Int. Ed. Engl. 1994, 33, 1755.
(50) Sakurai, T.; Sundaralingam, M.; Jeffrey, G. A. Acta Crystallogr.
1963, 16, 354.
(51) The CSD has been searched for the occurrence of X3 trimer synthons
with Cl‚‚‚Cl and Br‚‚‚Br separations 3.5 and 3.7 Å, respectively. An
unsymmetrical Cl3 synthon was found (PCPBPT.; Miravitlles, C.; Solans,
X.; Germain, G.; Declercq, J. P. Acta Crystallogr. 1979, B35, 2809).
Symmetrical Br3 synthons equivalent to those found in 8 are seen in
BROFRM03 (Myers, R.; Torrie, B. H.; Powell, B. M. J. Chem. Phys. 1983,
79, 1495) and JUHJIF (Tebbe, F. N.; Harlow, R. L.; Chase, D. B.; Thorn,
D. L.; Campbell, G. C., Jr.; Calabrese, J. C.; Herron, N.; Young, R. J., Jr.;
Wasserman, E. Science 1992, 256, 822).
Supporting Information Available: ORTEP diagrams and
tables giving crystal data and structure refinement, atomic
coordinates, isotropic and anisotropic displacement parameters,
and bond lengths and angles for 1-6 (38 pages). See any
current masthead page for ordering and Web access instructions.
JA972830R