organic compounds
˚
refined to values of 1.420 (3) and 1.443 (2) A, respectively
(shown as a square in Fig. 2, indicated with an arrow). A
disorder model incorporating the two different components,
with the sum of the occupancies constrained to unity, was
attempted. However, in order for the refinement to converge
successfully, the displacement parameters for the ꢀ-C atom
C18 and its disordered component C18A had to be
constrained as isotropic. The model converged, yielding
occupancies of the two inden-2-yl components of 0.57 (2) and
0.43 (2). The resulting C C and C—C bond lengths about
˚
C18/C18A were 1.29 (2)/1.52 (2) A for component 1 and
˚
1.35 (2)/1.60 (3) A for component 2 (lower and upper triangles
in Fig. 2, respectively). Further refinement cycles in which
additional bond-length constraints were applied to all bonds
about the ꢀ-C atom led to instability in the refinement. From
this analysis it was concluded that the disorder model was
insufficient and so the data presented here is based on the
ordered model.
For the inden-2-yl moiety, the five-membered C18–C20/
C25/C26 ring is planar, with C18—C19—C20—C25 and C20—
C25—C26—C18 torsion angles of 1.36 (16) and ꢁ0.69 (16)ꢂ,
respectively, whereas the five-membered C9–C11/C16/C17
ring of the indan-2-yl moiety adopts an envelope conforma-
Figure 2
A scatterplot of C C versus C—C bond lengths for inden-2-yl fragments
in the CSD. See the Comment for an explanation of the symbols.
potential H-atom positions in a difference Fourier synthesis
about each of atoms C19 and C26, both ꢀ to atom C18. The
potential disorder in this group was also revealed through a
Hirshfeld rigid-bond test (Hirshfeld, 1976), where the differ-
ences in the components of the anisotropic displacement
parameters along the C18—C19 and C18—C26 bonds exceed
6 s.u.
A simple Conquest (Macrae et al., 2008) search of the
Cambridge Structural Database (CSD, Version 5.33; Allen,
2002) for inden-2-yl fragments as shown in (II) (see Scheme),
where R1 is defined as any substituent other than hydrogen,
returned a total of 33 entries. The disorder present in the
inden-2-yl fragment was documented in a number of struc-
tures and the use of a two-part disorder model to separate the
two components was attempted [see, for example, CSD
refcodes APOVUX (Nikitin et al., 2010), CORBOB (Nikitin et
al., 2009), OGEKAN (Nikulin et al., 2008) and TENBAP (Li et
al., 1996)], although in the case of APOVUX it was specifically
noted that the disorder model failed. A scatterplot of C—C
distances versus C C distances is shown in Fig. 2. The
correlation between these two parameters is clear, such that
for structures where there is no disorder present, or where the
disorder model has been successfully implemented, the values
of the C C and C—C bond lengths are clearly different, ca
˚
tion or E form, with atom C9 displaced by 0.478 (2) A from
the mean plane defined by the other four atoms.
The absolute configuration of (I), viz. S, S and S at the chiral
centres C9, C10 and C33, respectively, was determined by
reference to the a priori knowledge of the chirality of the (S)-
(ꢁ)-methylbenzylamine used in the salt-formation step and by
anomalous dispersion methods (Flack, 1983). The determi-
nation of the absolute configuration of (I) by anomalous
dispersion methods was likely to be challenging, given that the
molecular formula and asymmetric unit contain only a single
N and three O atoms. To maximize the likelihood of success, a
full sphere of data was collected using Cu Kꢀ radiation to a
˚
maximum resolution of 0.80 A. A total of 25 532 reflections
were collected, yielding a Flack parameter x and standard
uncertainty u for this structure of 0.00 (15) based on 2343
Friedel pairs. The value of u is slightly beyond the limit of
enantiopure-sufficient distinguishing power (Flack & Bernar-
dinelli, 1999, 2000), and for further confirmation of the abso-
lute configuration a determination using Bayesian statistics on
Bijvoet differences (Hooft et al., 2008), as implemented in the
program PLATON (Spek, 2009), was performed. This gave
probability values p3(ok), p3(twin) and p3(wrong) of 1.000,
0.000 and 0.000, respectively. The calculation was based on
2343 Bijvoet pairs. The absolute structure parameter and
standard uncertainty calculated using this method was
0.11 (4). An improvement in the absolute structure parameter
can be made using a Student t distribution rather than a
Gaussian distribution (Hooft et al., 2010), giving ꢁ0.03 (13)
for a ꢁ parameter of 9.79. The overall p3 probability values
calculated using this method remain unchanged at 1.000, 0.000
and 0.000.
˚
1.35 and 1.50 A, respectively (e.g. OGEKAN), whereas for
structures that demonstrate the disorder phenomenon these
two bond lengths appear to be correlated and ultimately
˚
equilibrate to a value of ca 1.42 A. The outlier point circled in
Fig. 2 (QUGWUK; Basavaiah et al., 2001) is due to the
incorrect assignment of the C-atom type when geometrically
˚
placing the H atoms; the 1.378 and 1.463 A bond lengths
should be assigned as C C and C—C bonds, respectively, and
not vice versa. The complete list of structures contained within
this data set is available in the Supplementary materials.
In keeping with the findings above, the C C and C—C
bond lengths (C18 C19 and C18—C26, respectively) in (I)
The packing arrangement for (I) is best descibed as an
infinite two-dimensional hydrogen-bond network in the (100)
plane of the unit cell. The primary building block of this
o324 Frampton et al. C8H12N+ꢀC26H21O3
Acta Cryst. (2012). C68, o323–o326
ꢁ
ꢃ