The search for materials with non-linear optical (NLO) properties has been the subject of intense research, due to their application in a wide range of technologies, such as optical computing and optical communication [1– 3]. Although several inorganic substances are known to exhibit NLO properties, macroscopic assemblies of organic molecules show great promise as new materials for building NLO devices, due to their fast response times [3]. Attempts at designing organic molecules with nonlinear optical activity have traditionally used the push– pull model, where an electron donor group is connected to an electron-acceptor group through a spacer. In order to facilitate charge transfer between the donor and the acceptor groups, the spacer is usually a conjugate, planar or quasi-planar system. Although molecules with reasonably large calculated hyperpolarizabilities (β and γ ) can be designed following this procedure, at the crystalline level, the property can be cancelled due to anti-parallel centrosymmetrical stacking of the molecules induced by their planarity. A well known example of the failure of the push–pull design model is p-nitro-aniline, which was expected to have a large (at least for that time’s standards) non-linear optical activity. Unfortunately, at the crystalline level the NLO activity vanishes. Some of the inconveniences of the 1-D π conjugated systems can be more or less circumvented by other arrangements such as 2-D shaped molecules or 3-D tetrahedral DA compounds [4, 5]. Recently, Barbosa and Nascimento [6] proposed a new approach, where a non-planar unit, with high electron affinity, is used as the acceptor, to which several donors can be linked. This new strategy presents two great advantages over the traditional push–pull model: first, the non-planarity of the central acceptor unit reduces the probability of forming a centro-symmetric crystal; second, by properly choosing the type and number of donor groups, one can design molecules in which the octupolar components of the β and γ tensors can be maximized, according to the proposal of Zyss et al. [7]. The advantages of octupolar crystalline structures over the dipolar ones have been extensively discussed by Zyss et al. [8]. Using this new approach several derivatives of C36 exhibiting large values of β have been designed [6]. However, since the chemistry of C36