Structural investigations of three triazines: Solution-state NMR studies of internal rotation and structural information from solid-state NMR, plus a full structure determination from powder x-ray diffraction in one case

Article abstract of DOI:10.1002/mrc.1185

Three model 2,4,6-tris(amino)-1,3,5-triazines, structurally related to a dyestuff molecule previously studied by NMR, were synthesized in order to enable the effects of rotamer exchange on the NMR spectra to be investigated in more detail. Two of the compounds are novel. Internal rotation of the triazine ring substituents was studied by variable-temperature solution-state ¹H, ¹³C and ¹⁵N NMR spectroscopy. All the expected rotamers were detected for each molecule. Rotamer exchange rates varied from slow to fast over the temperature range -40 to 90°C, as observed for the dyestuff molecule itself. Solid-state ¹³C and ¹⁵N NMR provided information about the structures of the solid molecules. A full crystal structure determination from high-resolution powder x-ray diffraction was achieved for one of the molecules using simulated annealing techniques. Ab initio MO and ¹⁵N NMR chemical shift calculations, based on energy-minimized structures derived from the x-ray structure determination, enabled the effect of intermolecular hydrogen bonding on the ¹⁵N NMR chemical shifts to be studied. The results compared favourably with the experimental solid-state ¹⁵N NMR shifts. Copyright

Full text of DOI:10.1002/mrc.1185

MAGNETIC RESONANCE IN CHEMISTRY
Magn. Reson. Chem. 2003; 41: 324–336
Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/mrc.1185
Structural investigations of three triazines:
solution-state NMR studies of internal rotation
and structural information from solid-state NMR,
plus a full structure determination from powder
x-ray diffraction in one case†
Helen E. Birkett,1 Julian C. Cherryman,2 A. Margaret Chippendale,2 John S. O. Evans,1
Robin K. Harris,1∗ Mark James,2 Ian J. King1 and Graham J. McPherson2
1
Department of Chemistry, University of Durham, South Road, Durham DH1 3LE, UK
Avecia Ltd, P.O. Box 42, Blackley, Manchester M9 8ZS, UK
2
Received 1 November 2002; Revised 7 January 2003; Accepted 12 January 2003
Three model 2,4,6-tris(amino)-1,3,5-triazines, structurally related to a dyestuff molecule previously studied
by NMR, were synthesized in order to enable the effects of rotamer exchange on the NMR spectra to
be investigated in more detail. Two of the compounds are novel. Internal rotation of the triazine ring
1
substituents was studied by variable-temperature solution-state H, 13C and 15N NMR spectroscopy. All
the expected rotamers were detected for each molecule. Rotamer exchange rates varied from slow to fast
over the temperature range −40 to 90 ◦C, as observed for the dyestuff molecule itself. Solid-state 13C
and 15N NMR provided information about the structures of the solid molecules. A full crystal structure
determination from high-resolution powder x-ray diffraction was achieved for one of the molecules using
simulated annealing techniques. Ab initio MO and 15N NMR chemical shift calculations, based on energy-
minimized structures derived from the x-ray structure determination, enabled the effect of intermolecular
hydrogen bonding on the 15N NMR chemical shifts to be studied. The results compared favourably with
the experimental solid-state 15N NMR shifts. Copyright  2003 John Wiley & Sons, Ltd.
KEYWORDS: NMR; 1H NMR; 13C NMR; 15N NMR; triazines; structure solution; conformation; rotamers; magic-angle
spinning; powder XRD
INTRODUCTION
NaO3S
N
3′
NH2
α
The effects of internal rotation have been observed in the
1H NMR spectra of the dyestuff molecule 1 in solution in
DMSO-d6 and CD3OD.1
Other groups have also reported2–5 exchange effects
due to rotational isomerism for triazine systems. For
compound 1, multiple peaks are observed for some atoms
H
N
β
N
N
N
δ
H
N
N
γ
OH
SO3Na
HN
NaO3S
°
under some conditions. For instance, in CD3OD at 60 C
a single peak was seen for H-30 which splits into three
SO3Na
°
on decreasing the temperature to ꢀ10 C. Below this
temperature a fourth peak was observed, followed by general
broadening as the temperature was further decreased.
1
The effects were attributed to exchange processes between
conformations involving internal rotation of the substituents
about the triazine ring, indicated by ˛, ˇ, υ and ꢀ on 1. The
bandshapes for the peaks from H-30 were fitted for three-site
†Dedicated to Professor William F. Reynolds on
the occasion of his 65th birthday.
ŁCorrespondence to: Professor R. K. Harris, Department of
Chemistry, University of Durham, South Road, Durham DH1 3LE,
UK. E-mail: r.k.harris@durham.ac.uk
Contract/grant sponsor: EPSRC; Contract/grant number:
GR/N05635.
°
exchange between ꢀ10 and C30 C, and the relevant rate
constants extracted. Activation parameters were calculated:
for exchange between one pair of H-30 sites H† D 76 š 8 kJ
molꢀ1 and S† D 52 š 27 J molꢀ1 Kꢀ1, and for a second pair,
H† D 76 š 4 kJ molꢀ1, S† D 55 š 15 J molꢀ1 Kꢀ1, whereas
Contract/grant sponsor: EPSRC/JREI; Contract/grant number:
GR/M35222.
Copyright  2003 John Wiley & Sons, Ltd.

Structural investigations of triazines
325
SO3Na
SO3Na
SO3Na
SO3Na
R'
R'
R'
R'
H3'
α
H
N
H
β
H
N
H
N
N
N
N
N
N
N
N
H5'
N
N
N
δ
N
R
H
H
R
R
R
H
H
N
H
N
R
N
R
N
N
H
N
N
H
N
R
N
N
R
γ
H
I
II
III
IV
R'
R'
R'
R'
H
H
NaO3S
H
N
NaO3S
H
N
N
NaO3S
N
N
NaO3S
N
N
N
N
N
N
N
N
R
H
R
R
N
H
H
H
R
N
N
N
N
N
N
N
R
N
N
N
H
R
H
H
R
R
H
V
VI
VII
VIII
Figure 1. Rotational isomers for 1. R is CH2CH2SO3Na and R0 is a substituted naphthalene ring joined to the central aromatic ring
by an azo group.
the rate of direct exchange between the remaining pair of
sites appears to be negligible. Exchange involving a fourth
rotamer affects the spectra at somewhat lower temperatures,
which fact assists in a partial assignment of the observed
peaks to the rotamers.
There are 16 possible rotamers arising from two possible
conformations about each of the three triazine C—N bonds
and the two orientations possible for the aromatic ring,
assuming approximate coplanarity of the two six-membered
rings. However, since two of the triazine ring substituents
are identical, these forms comprise eight pairs of rotamers,
shown in Fig. 1. The rotamers are related by internal
rotations, as shown in Fig. 2.
The different possible exchange processes were examined
for isolated molecules using molecular modelling. The results
β
δ
V
VI
Figure 3. Dihedral angle distribution about bonds involved in
(a) ˛ rotation and (b) ˇ rotation. Each based on data from a
total of 39 solid-state structures.
α
α
γ
β
γ
δ
I
II
β
of a preliminary study were presented with the initial work.1
Rotationalenergy barriers calculated from an in-depth study6
were found to be 13, 75, 91 and 91 kJ molꢀ1 for the ˛, ˇ, ꢀ
and υ rotations, respectively. The internal rotations (υ,ꢀ) for
the triazine side-chains have the largest energy barriers, with
˛ having the lowest. This is surprising but is substantiated
by structures in the Cambridge Crystallographic Database.7
Thus, Fig. 3 demonstrates that the smallest barrier, ˛, relates
to a large range of dihedral angles about this bond in known
solid-state structures. This suggests that there is a small
energy difference between different values for the dihedral
angle and is therefore consistent with the barrier to internal
rotation being low in solution. The dihedral angle of ˇ has a
β
γ
δ
γ
VIII
VII
β
α
α
IV
III
δ
β
Figure 2. Relationship between rotamers I–VIII shown in
Fig. 1.
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 324–336

326
H. E. Birkett et al.
structures were also studied by 13C and 15N NMR and, in
one case, solved by powder x-ray diffraction. The principal
aim of this work was to confirm the nature of the four inter-
nal rotation processes and to examine how the barriers are
affected by the nature of the substituents. In addition, some
ab initio molecular orbital calculations of 15N chemical shifts
were carried out for hydrogen-bonded dimers related to the
determined crystal structure.
Standard procedures8 were used to prepare the com-
pounds. The reaction of aniline and 4-chloro-5-carboxyani-
line with cyanuric chloride gave 2 and 5, respectively.
These were readily converted to 3 and 6 by reaction
with diethanolamine, and 2 converts to 4 by reaction with
ethanolamine. Compounds 3 and 6 are new. The preparation
of 4 from 2 has been reported previously.9
much more limited range of angles, suggesting that there is
a greater energy difference between different conformations
and therefore a larger energy barrier. This trend in the size
of the energy barriers supports the molecular modelling.
If there is rapid rotation, on the NMR time-scale, about
˛, but slow rotation for the other processes, there will be
four rapidly interchanging pairs (I$V, II$VI, III$VII and
IV$VIII), which will now be denoted simply as I, II, III
and IV. If there is also rapid rotation about ˇ, on the NMR
time-scale, then rotamers I and IV are indistinguishable. This
°
would leave three distinguishable forms. At 90 C all four
possible internal rotations are rapid on the NMR time-scale,
so that a single averaged 1H NMR spectrum is observed
from all the conformations. As the temperature is decreased,
rotations ꢀ and υ slow, giving rise to separate signals from
three conformations: II, III and an average of I and IV. As the
temperature is decreased further, rotation ˇ begins to slow,
causing the largest intensity peak to split into two, and thus
demonstrating that four distinguishable conformations, I, II,
III and IV, exist. As the temperature is decreased further, the
lines broaden again, suggesting yet another active process,
˛, is slowing. This would lead to the four rotamers becoming
eight. Further reduction in temperature, to see if there are
indeed eight peaks, is precluded by the freezing point of
the solvent.
RESULTS AND DISCUSSION
Solution-state NMR
Compound 3
All the rotamers which would interconvert by ˛, ˇ, ꢀ and υ
rotations are equivalent for 3, so that there is a single distin-
guishable form, as shown in Fig. 4. Moreover, each of the ˛,
ˇ, ꢀ and υ rotations would, if fast on the NMR time-scale,
result in averaging of some of the 1H and 13C resonances for
this structure. Tables 1 and 2, respectively, list the maximum
number of resonances expected for each type of proton and
carbon atom for different combinations of fast and slow ˛,
ˇ, ꢀ and υ rotations, together with the resonances observed
Variable-temperature 13C NMR would have been an ideal
technique to study the origin of the exchange processes
involved. However, the solubility of 1 was not sufficient to
obtain the required sensitivity for such work.
°
at 90, 25 and ꢀ40 C from a DMF-d7 solution. It is assumed
that there is rapid rotation within the saturated side-chains
throughout this temperature range as reported by Katritzky
et al.2 for a range of n-alkylamino-substituted triazines.
R3
R2
3
4
2
1
1
°
In the H NMR spectrum at 90 C, a single band is
5
observed for each type of proton, consistent with fast ˛,
NH
6
t1
°
ˇ, ꢀ and υ rotations. At 25 C, some splitting is observed
t6 N
N t2
on the peaks from the aliphatic protons. On decreasing the
°
temperature further to ꢀ40 C the hydroxyl group resonance
t5
t3
R1
R1
N
splits into four, whilst the aliphatic peaks remain severely
overlapped. Throughout the temperature range the peaks
from the aromatic protons do not split. This is consistent with
t4
°
fast ˛ rotation. The spectrum at ꢀ40 C therefore suggests
2 R1 = Cl, R2 = H, R3 = H
3 R1 = N(CH2CH2OH)2, R2 = H, R3 = H
4 R1 = NHCH2CH2OH , R2 = H, R3 = H
5 R1 = R2 = Cl, R3 = CO2H
rapid ˛ and slow ˇ, ꢀ and υ rotations.
The 13C spectrum, with its wider chemical shift range,
illustrates the situation more clearly. As shown in Table 2,
6 R1 = N(CH2CH2OH)2, R2 = Cl, R3 = CO2H
5
4
This paper reports the synthesis and study of the
exchange processes in three simpler, related, model com-
pounds, 2-anilino-4,6-diamino-s-triazine derivatives (3, 4, 6),
by 1H and 13C NMR spectroscopy. These were chosen so as to
eliminate any possible complications from the presence of the
large naphthyl group in 1. Moreover, 3 and 4 contain a simple
phenyl group attached to the NH nitrogen, thus simplifying
rotation ˛, whilst 3 and 6 have four identical substituents at
the other two exocyclic nitrogens, thus reducing the number
of possible distinguishable rotamers significantly. Thirdly,
they are sufficiently soluble that 13C (and even 15N) spectra
of reasonable quality could be obtained. Their solid-state
6
3
α
H
1
2
N
β
t1
Nt2
t6N
t5
HOCH2CH2
HOCH2CH2
CH2CH2OH
CH2CH2OH
t3
N
N
N
δ
γ
t4
Figure 4. Compound 3 rotamer.
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 324–336

Structural investigations of triazines
327
Table 1. 1H NMR data for 3
1H chemical shifts and coupling constants
Maximum No. of 1H bands
DMF-d7 solution
°
90 C
˛, ˇ,
ꢀ, υ
rapid
˛, ˇ
rapid; ꢀ,
υ slow
˛ rapid;
ˇ, ꢀ,
υ slow
˛, ˇ,
ꢀ, υ
°
°
25 C
ꢀ40 C
Atom No.
slow
υ (ppm)
J (Hz)
υ (ppm)
υ (ppm)
2, 6
3, 5
4
NH
CH2O
CH2N
OH
1
1
1
1
1
1
1
1
1
1
1
2
2
2
1
1
1
1
4
4
4
2
2
1
1
4
4
4
7.76 (d)
7.24 (t)
6.91 (t)
8.34 (br s)
3.78a (m)
3.75a (m)
4.21 (br s)
8.2
8.2, 7.6
7.6
—
—
7.90
7.28
6.94
7.99
7.36
6.98
9.14
9.70
3.7–3.8b
3.7–3.8b
4.91
3.7 to 3.9b
3.7 to 3.9b
5.40
—
—
5.41
5.42
5.46
a Interchangeable pairs of assignments.
b Multiple poorly resolved peaks.
Table 2. 13C NMR data for 3
Theoretical No. of 13C peaks
13C chemical shifts (ppm)
DMF-d7 solution
˛, ˇ, ꢀ, υ
rapid
˛, ˇ rapid;
ꢀ, υ slow
˛ rapid;
ˇ, ꢀ, υ slow
˛, ˇ, ꢀ, υ
slow
°
°
°
Atom No.
90 C
25 C
ꢀ40 C
Solid
t1
1
1
1
1
1
2
1
2
165.15
166.61
164.83
165.87
164.74
163.1
165.7
t3, t5
165.39
165.30
1
1
1
1
1
1
1
1
2
141.72
120.63
141.88
119.89
142.10
119.41
140.7
119.7a
119.2a
2, 6
3, 5
1
1
1
2
128.93
129.05
121.78
129.38
121.71
129.3
128.3
127.4
121.9a
120.8a
4
1
1
1
2
1
4
1
4
122.17
61.32
CH2O
60.82
60.78
60.54
60.48
60.44
60.19
65.1
62.0
60.9
60.0
58.2
CH2N
1
2
4
4
51.88
51.81
51.72
52.25
52.14
51.51
51.25
55.0
53.1
50.4
49.7
48.7
a Interchangeable assignments.
°
the number of peaks observed in the spectrum at 90 C
into four, as shown in Fig. 5. At this temperature, separate
peaks are also resolved for t3 and t5 (see Fig. 4), suggesting
that the ˇ rotation is now slow.
°
demonstrates that all the rotations are rapid. At 25 C, two
peaks are observed for each type of aliphatic carbon atom,
consistent with the ˛ and ˇ rotations being rapid whereas
ꢀ and υ rotations are slow. As the temperature is lowered
A
15N NMR spectrum of a DMSO-d6 solution was
°
obtained at 25 C and the assigned resonances are listed in
Table 6. A single peak is observed from each type of nitrogen
°
further these peaks broaden, and by ꢀ40 C they have split
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 324–336

328
H. E. Birkett et al.
Compound 4
For this compound there are four distinguishable rotamers,
as shown in Fig. 6. They are identified in accordance with
Fig. 1 for 1. Tables 3 and 4, respectively, list the maximum
number of resonances expected for each type of proton and
carbon atom for different combinations of fast and slow
˛, ˇ, ꢀ and υ rotation, together with the signals actually
90°C
°
observed at 90, 0 and ꢀ40 C.
As for 3, a single resonance is observed from each type
°
of proton at 90 C, consistent with fast rotation about all
the bonds. On lowering the temperature, the resonances
°
broaden, but by 0 C they have sharpened again and are
showing splittings. All of them except the PhNH peaks
are, however, poorly resolved. For this hydrogen atom three
peaks are present, from rotamers II, III and an average of I and
25°C
°
IV, consistent with slow ꢀ and υ rotation. By ꢀ40 C ˇ rotation
has slowed and the largest intensity central peak from I and
IV has split in two, as shown in Fig. 7, in an analogous
way to the effects observed for 1. Although the individual
peaks have not been fully assigned to specific rotamers,
their relative intensities are 1.00 (υ9.46) : 0.88 (υ9.58) : 1.08
(υ9.67) : 1.55 (υ9.83).
-40°C
The 13C variable-temperature study shows clearly that
°
at 90 C all the rotations are fast on the NMR time-scale
°
and that at 0 C ˛ and ˇ rotations are fast whereas ꢀ
61 60 59 58 57 56 55 54 53 52 51
°
and υ rotations are slow. At ꢀ40 C only the ˛ rotation
δ(13C)/ppm
is fast, which is as expected since only a phenyl group
is involved (in contrast to the situation for 1). From
Table 4 it can be seen that, with a few exceptions due to
accidental degeneracy, the expected number of peaks are
observed for each type of carbon atom. Figure 8 shows the
triazine carbon resonances for a range of temperatures, with
the t1 carbon atoms giving one peak from each rotamer
Figure 5. Aliphatic region of the solution-state 13C NMR
spectrum of 3.
atom. Separate peaks are not observed for either t2 and t6 or
the two aliphatic amine nitrogen atoms, demonstrating that
there is fast ˇ rotation at this temperature. This is consistent
with the conclusions from the 13C spectrum obtained from
the DMF-d7 solution at the same temperature.
°
at ꢀ40 C, as expected. Carbons t3 and t5 should give
a total of eight peaks, two from each rotamer. In fact
5
4
6
3
α
H
N
H
1
2
N
β
t1
t6N
Nt2
N
N
t5
H
H
H
CH2CH2OH
t3
N
N
N
N
N
N
δ
γ
t4
HOCH2CH2
CH2CH2OH
HOCH2CH2
H
I
II
H
N
H
N
N
t2
N
t6N
N
HOCH2CH2
CH2CH2OH
HOCH2CH2
H
a2
N
a1
N
N
N
N
N
t4
H
H
H
CH2CH2OH
III
IV
Figure 6. Compound 4 rotamers.
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 324–336

Structural investigations of triazines
329
Table 3. 1H NMR data for 4
1H chemical shift and coupling constants
Maximum No. of 1H bands
DMF-d7 solution
°
90 C
˛, ˇ,
ꢀ, υ
rapid
˛, ˇ
rapid;
ꢀ, υ slow
˛ rapid;
ˇ, ꢀ, υ
slow
˛, ˇ,
ꢀ, υ
slow
°
°
0 C
ꢀ40 C
Atom No.
υ (ppm)
J (Hz)
υ (ppm)
υ (ppm)
2, 6
3, 5
4
1
1
1
3
3
3
4
4
4
8
8
4
7.78 (d)
7.22 (t)
6.91 (t)
8.5
8.5, 7.3
7.3
7.9–8.0a
7.2–7.4a
6.9–7.0a
7.90–8.1a
7.2–7.4a
6.9–7.1a
PhNH
1
3
4
4
8.24 (br s)
—
9.11
9.26
9.43
9.46
9.58
9.67
9.83
CH2O
CH2N
CH2NH
OH
1
1
1
1
4
4
4
4
8
8
8
8
8
8
8
8
3.67 (t)
5.9
5.9, 4.4
—
3.6–3.7a
3.4–3.6a
6.9–7.3a
5.1(br)
3.6–3.8a
3.4–3.6a
7.3–8.2a
3.49 (br m)
6.13 (br s)
4.0 (br s)
—
5.40
5.42
5.43
5.75
5.81
a Multiple, sometimes poorly resolved peaks.
90°C
90°C
25°C
0°C
0°C
-40°C
-40°C
9.5
9.0
8.5
8.0
7.5
7.0
δ(1H)/ppm
167.5 167.0 166.5 166.0 165.5 165.0
δ(13C)/ppm
Figure 7. Aromatic and NH region of the solution-state 1
NMR spectrum of 4.
H
Figure 8. Triazine region of the solution-state 13C NMR
spectrum of 4.
seven are observed, so presumably two are accidentally
degenerate.
°
noisy, that at 90 C appeared to contain a single sharp peak
Nitrogen-15 NMR spectra of a DMSO-d6 solution were
from each type of amine nitrogen atom. The peak(s) from
the triazine ring nitrogen atoms are very weak and it is not
°
obtained at 25 and 90 C. Although the spectra were very
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 324–336

330
H. E. Birkett et al.
Table 4. 13C NMR data for 4
Maximum No. of 13C peaks
13C chemical shifts (ppm)
DMF-d7 solution
˛ rapid;
ˇ, ꢀ,
υ slow
˛, ˇ, ꢀ,
υ rapid
˛, ˇ rapid;
ꢀ, υ slow
˛, ˇ, ꢀ,
υ slow
°
°
°
Atom No.
t1
90 C
0 C
ꢀ40 C
Solid
164.3
1
3
4
4
165.65
165.30
165.10
164.98
165.16
164.97
164.94
164.75
t3, t5
1
4
8
8
167.61
167.04
166.99
164.3
166.91
166.80a (ð2)
166.75
166.71
166.61
166.55a (ð2)
166.49
166.30
1
1
1
3
3
4
4
4
8
141.71
120.71
141.92
141.89
141.87
142.14
142.1
119.4
141.95a (ð2)
141.89
2, 6
119.96
119.91a (ð2)
119.88
119.80
119.71
119.52
3, 5
1
1
1
1
3
3
4
4
4
4
8
8
8
4
8
8
128.86
122.16
61.98
129.13
129.02
128.94
129.37
129.29
129.16
129.13
129.4
127.1
4
121.91
121.76
121.72
121.98
121.91
121.80
121.69
119.2
CH2O
CH2N
61.54
61.50
61.40
61.36
61.34
61.24
62.4
60.5
61.18a (>ð1)
61.09
44.25
43.89
43.81
43.64a (ð2)
44.10
43.90
43.84
43.65
43.53
43.90
43.27
43.17
43.0
a Accidental peak degeneracy.
clear whether there is a single peak from all three atoms or
whether a second peak is too weak to detect. This is consistent
with the conclusions from the 1H and 13C NMR spectra, that
Compound 6
For this compound there are only two distinguishable
rotamers, as shown in Fig. 9, because all four substituents
at two of the exocyclic nitrogens are the same. They are
identified in accordance with Fig. 1 for 1. This sample was
°
at 90 C all the rotations are fast on the NMR time-scale.
°
The spectrum at 25 C shows that multiple peaks are present
from each type of nitrogen, although the signal-to-noise
ratio in the spectrum is too low to ascertain with confidence
the total number of peaks. Again these observations are
consistent with the conclusions from the other NMR results.
The assigned resonances are listed in Table 6.
examined by 1H and 13C NMR in DMF-d7 at 25 C. The
°
compound decomposed when the temperature was raised
°
to 90 C, and a low-temperature NMR study was not carried
out. The observed H and 13C chemical shifts are listed in
1
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 324–336

Structural investigations of triazines
331
CO2H
Cl
Cl
3
2
4
5
HO2C
α
1
N
H
N
H
6
N
N
β
t1
t6N
Nt2
N
t5
HOCH2CH2
HOCH2CH2
CH2CH2OH
CH2CH2OH
HOCH2CH2
HOCH2CH2
CH2CH2OH
CH2CH2OH
t3
N
N
N
N
N
γ
δ
t4
V
I
Figure 9. Compound 6 rotamers.
Table 5. Single resonances are observed from all the phenyl
Compound 3
1
group atoms at 25 C in both the H and 13C spectra, which is
Tables 2 and 6 show the 13C and 15N chemical shift
assignments, respectively. The peaks from the carbon atoms
directly bonded to nitrogen atoms, i.e. C-1, t1, t3, t5 and
CH2N, are broader than the other signals. This is caused by
residual dipolar coupling (RDC).15 In both spectra more than
one peak is present from some atom types. This applies to
C-3/5, C-2/6 or C-4, CH2O, CH2N, PhNHTr and AlkylNTr.
In some cases, for example CH2N, this could be due, at
least in part, to the presence of a locked conformation or
loss of molecular symmetry in the solid state. However, this
cannot explain the presence of three peaks from C-3/5 or
two PhNHTr peaks. Therefore the asymmetric unit would
appear to contain more than one molecule or, alternatively,
the sample may contain more than one polymorph. The
chemical shifts for most of the carbon and nitrogen atoms are
very similar to their solution-state values. However, those
for the aliphatic carbon atoms span a larger range than in
solution, possibly suggesting different (or more precisely
defined) conformations of the aliphatic chains.
°
consistent with the ˛ rotation being rapid at this temperature,
as expected. In the 13C spectrum two peaks are observed from
the CH2O group and the CH2N peak is broad, suggesting
that the ꢀ and υ rotations are relatively slow. From the limited
NMR data available, it is unclear why both the peaks from
the triazine carbon atoms are so broad at this temperature, if
˛ rotation is fast.
Solid-state NMR
High-resolution NMR of dilute-spin nuclei such as 13C
and 15N can provide information (such as the number
of molecules in the asymmetric unit,10 loss of molecu-
lar symmetry,11 exchange processes12 and intermolecular
interactions13,14ꢁ about solid-state structures. Solid samples
of 3, 4 and 6 were studied by high-resolution 13C and 15N
NMR. In the solid-state, the compounds are more likely to
exist in a locked conformation.
Table 5. 1H and 13C NMR data for 6
Compound 4
Tables 4 and 6 give the 13C and 15N chemical shift assign-
ments, respectively. Figure 10 shows the spectra. The low-
intensity, unlabelled peaks are spinning sidebands. As with
3, the carbon atoms adjacent to nitrogen give rise to broad
peaks. Two peaks only are observed from each of C-3/5,
CH2O and TrNAlkyl. This is consistent with the asymmet-
ric unit containing a single locked molecular conformation,
which does not have a plane of symmetry perpendicular
to the plane of the triazine ring. In the 13C spectrum the
peaks from t1, t3 and t5 are accidentally degenerate as are
those from C-2/6 and C-4. All the solid-state 13C NMR shifts
are close to the shift ranges observed in solution. However,
there are differences between the solution- and solid-state
15N shifts for one of the triazine nitrogen atoms and for
PhNHTr, which may arise from intermolecular hydrogen
bonding in the solid state.
Chemical shifts and coupling constants
°
DMF-d7 solution, 25 C
Solid
13C
1H
13C
Atom
No.
υ (ppm)
J (Hz) υ (ppm)
υ (ppm)
t1
t3, t5
1
2
3
4
5
6
—
—
—
—
¾163 (br)
160.4
155.2 (br)
137.48
123.02
132.37
127.59
132.23
124.33
59.91
152.9
135.6
122.2
130.1
127.8
130.1
124.4
59.8
—
—
8.43 (d)
—
2.9
—
—
—
7.61 (d)
7.78 (dd)
¾3.9 (m)
8.5
8.5, 2.9
—
CH2O
60.33
54.7
OH
CH2N
¾4.5 (br s)
¾3.9 (m)
—
—
—
52.63(br)
—
53.5
49.9
—
Compound 6
Tables 5 and 6 show the 13C and 15N chemical shift
assignments, respectively. As with compounds 3 and 4,
the carbon atoms adjacent to nitrogen yield broad peaks. It
appears that all the aryl carbon atoms give rise to single
NH
CO2H
11.44 (br s)
¾12.5 (br s)
—
—
166.99
166.5
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 324–336

332
H. E. Birkett et al.
Table 6. Solution- and solid-state 15N NMR data
15N chemical shift (ppm)
3
4
6
Solution
DMSO-d6
Calculated shiftc
Dimer 209
Solution
Atom
No.
DMSO-d6
e
°
°
°
25 C
Solid
25 C
90 C
Solid
Monomer
Dimer 203
Solid
t2
t4
t6
ꢀ205.1
ꢀ204.7
ꢀ205.1
ꢀ268.4
ꢀ205.1d
ꢀ208.8d
ꢀ208.8d
ꢀ266.7
ꢀ268.0
ꢀ192 to ꢀ209 ꢀ202.5a ꢀ201.7
ꢀ192 to ꢀ209 ꢀ202.5a ꢀ214.2
ꢀ192 to ꢀ209 ꢀ202.5a ꢀ201.7
ꢀ195.7
ꢀ208.6
ꢀ207.4
ꢀ262.0
−211.4
ꢀ206.3
ꢀ212.3
ꢀ262.7
ꢀ195.8
−221.6
ꢀ205.9
ꢀ261.0
ꢀ210.0d
ꢀ207.3d
ꢀ210.0d
ꢀ264.9
ꢀ266.5
ꢀ269.2
ꢀ270.6
ꢀ283.1
ꢀ287.8
PhNHTrb
ꢀ264 to ꢀ269 ꢀ269.7
ꢀ256.2
TrbNAlkyl
ꢀ290.4
ꢀ291.7
ꢀ294.3
ꢀ292 to ꢀ295 ꢀ293.0
ꢀ294.1
ꢀ296.5
ꢀ296.3(a1)
ꢀ299.0(a2)
ꢀ299.5(a1)
ꢀ294.3(a2)
ꢀ291.2(a1)
ꢀ298.7(a2)
a Very weak signal(s), one possibly too weak to detect.
b Tr D Triazine ring.
c See Fig. 6, rotamer IV for nitrogen atom labelling.
d Interchangeable assignments.
e The ranges may be underestimated as some peaks are broad and weak.
intensity are present for the former group and two of very
different intensity from the latter. It was thought that some
of these peaks might arise from a related impurity, but the
solution-state 1H and 13C spectra did not contain additional
peaks to support this. The solution-state spectra and the
solid-state 13C spectrum did, however, show the presence
of a small amount of ethanol. In the solid this may have
co-crystallized with some of 6 and could account for at least
some of the additional lower intensity peaks in the solid-state
15N spectrum.
CH2O
C1
CH2N
13C
250
200
150
100
50
0
ppm
Crystal structure determination and molecular
modelling
t4
Compounds 3, 4 and 6 were not amenable to growing
crystals suitable for the determination of single-crystal
x-ray structures. However, 4 was sufficiently crystalline
to allow the measurement of a powder XRD pattern.
Several methods16 for solving indexed powder patterns
have recently been developed which operate in either
reciprocal or direct space. For this work a direct space
simulated annealing protocol, as coded in the DASH software
suite,17 was employed. In this methodology, hundreds of
thousands of trial molecular structures are tested against
experimental x-ray diffraction data. A simulated annealing
protocol performs stochastic structural changes (translations,
rotations and changes of internal torsion angles) until
a good agreement with experimental data is achieved.
The trial structure so generated is then improved using
Rietveld refinement methods. Despite the considerable loss
of information in a 1D powder pattern as compared with a 3D
single crystal experiment, structures of increasing complexity
can now be determined. This approach was used to study 4.
15N
-100
-150
-200
-250
ppm
-300
-350
-400
Figure 10. Solid-state 13C and 15N NMR spectra of 4.
resonances, suggesting that, possibly, there is only one
molecule in the asymmetric unit. The aliphatic peaks show
structure, but the peaks are too broad and overlap too much
to be able to determine the total number present. The 15N
spectrum is more complex than expected. Two sharp peaks
are observed from the triazine nitrogen atoms. However, if
the asymmetric unit contained only a single molecule, one
peak would be expected from PhNHTr and a maximum of
two peaks from TrNAlkyl. In fact, four peaks of varying
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 324–336

Structural investigations of triazines
333
Compound 4
H121
C12
Powder x-ray diffraction revealed 4 to be highly crystalline.
Indexing of data recorded on a sample loaded in a 0.5 mm
capillary on a Bruker d8 diffractometer gave a triclinic cell
of a D 9.00468ꢂ33ꢁ A, b D 9.8930ꢂ4ꢁ A, c D 8.34487ꢂ28ꢁ A,
H111
C11
˚
˚
˚
H131
C13
C16
°
°
°
˛ D 91.5224(21) , ˇ D 110.5189(21) , ꢀ D 87.0330(24) with
a figure of merit of 96. The corresponding cell volume is
consistent with a density of 1.387 g cmꢀ3 and two molecules
in the unit cell. Rietveld refinement of the trial structure
generated by simulated annealing in the DASH software
suite was carried out using the synchrotron data. Final
agreement factors of wRp D 2.73% and RꢂF2ꢁ D 9.71%
in space group P1 were obtained. Figure 11 shows the
experimental x-ray powder diffraction data, the calculated
pattern after Rietveld refinement and the difference curve.
The overall lattice energy for the structure, calculated
using Cerius2, Dreiding Force Field and Gasteiger charges,
is ꢀ163.6 kJ molꢀ1. When the structure was minimized with
respect to the lattice energy, allowing rotation and translation
of the rigid molecule whilst constraining the unit cell, a
value of ꢀ170.5 kJ molꢀ1 was obtained. This small percentage
change in energy confirms that the crystal structure solution
is energetically sensible and realistic.
H161
C14
C15
H171
N17
C6
H151
N1
H101
H102
N5
C10
C2
H201
C4
H71
N9
N3
C20
H202
N7
C8
H91
O21
H81
H82
O19
H211
H182
C18
H191
H181
The structure obtained corresponds to rotamer IV (Fig. 6)
and is shown, with atom labelling, in Fig. 12. The phenyl
and triazine rings are essentially coplanar and the two
alkylamino substituents are located on the same side of
the triazine ring, although with very different orientations to
the ring as shown in Fig. 13. The C4N7C8C18 and C2N9C10C20
dihedral angles are ꢀ85.2 and ꢀ164.4 , respectively. All
the atoms are inequivalent, since the molecule does not
possess any elements of symmetry. The crystal structure
Figure 12. Structure of 4 by x-ray analysis.
is therefore consistent with the solid-state NMR results. In
the 13C NMR spectrum the difference of ¾2 ppm between
the two CH2O resonances is understandable in terms of
their very different environments. [Comment: the modelling
calculation using the monomer has a 1 ppm difference for
°
°
5
10
15
20
25
30
35
40
2-theta (deg)
Figure 11. Final Rietveld plot for 4. Observed data shown as red points, calculated pattern as solid blue line and difference curve as
lower purple line. Small vertical bars represent allowed peak positions.
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 324–336

334
H. E. Birkett et al.
Figure 13. An aspect of compound 4 solid-state structure.
these two resonances. In the dimers the change is up to
6 ppm due to OH hydrogen bonding changing the local
environment.]
Figure 15. Crystal packing for 4.
The packing arrangement in the crystal structure con-
tains an extensive hydrogen bonding network, involv-
ing N—HÐ Ð ÐN, N—HÐ Ð ÐO, O—HÐ Ð ÐO and O—HÐ Ð ÐN
bonds. Figure 14 displays how sheets of hydrogen bond-
ing molecules build up from two different types of dimeric
interaction. Figure 15 shows how these sheets are linked by
weaker O—HÐ Ð ÐO hydrogen bonding. The extensive hydro-
gen bonding in this compound is best described by graph set
notation and is given in Table 7.
˚
3 A) and then a much weaker interaction (with DÐ Ð ÐA greater
˚
than 3 A). It is not straightforward to separate out these
two interactions by calculation. The main intermolecular
contact in one dimer is an N17 —H171Ð Ð ÐO19 2.09 A hydrogen
˚
bond (distance from H to acceptor atom), and in the other
an N3Ð Ð ÐH211 —O21 2.03 A hydrogen bond. In the following
˚
discussion the two dimers will be referred to as dimer 203
and dimer 209. The third dimension of the lattice is formed
by the in-plane and out-of-plane alkylamino groups. This
is an OHÐ Ð ÐO hydrogen bonded chain and has the shortest
It can be seen that the central molecule participates in
four hydrogen bonds in each dimeric pair. For each dimer
there is a strong, dominant interaction (with DÐ Ð ÐA less than
˚
distance (DÐ Ð ÐA 2.75 A) of all the intermolecular hydrogen
bonds.
Dimerization energies of ꢀ50.1 and ꢀ58.0 kJ molꢀ1
were computed for dimers 203 and 209, respectively. The
dimerization energy includes both the hydrogen bonding
and, smaller, rearrangement energies. Such values are
consistent with typical dimerization energies of ¾60 kJ molꢀ1
for a system with multiple, cooperative hydrogen bonds
(e.g. the formic acid dimer). The two dimerization energies
are similar and so the dimers would be expected to be
approximately equally stable.
2.09Å
2.03Å
The 15N chemical shifts calculated using DGauss are listed
in Table 6 for comparison with the solution- and solid-state
experimental data. The energy-minimized monomer struc-
ture is expected to approximate to a gas-phase conformer IV
structure. The rebased 15N chemical shifts calculated from
it fall within, or very close to, the ranges observed in solu-
2.03Å
2.09Å
°
tion at 25 C, when all four conformations are known to
be present. For a proposed crystal structure to be plausible
when confronted with solid-state NMR data, it should be
possible to account for changes in chemical shifts, taking
an isolated molecule as the point of reference, in terms
Figure 14. Hydrogen-bonding interactions, involving the
triazine rings, for 4. The central atom is shown involved in the
two types of dimeric pair. The distances given are H Ð Ð Ð A.
Table 7. Hydrogen bonding in compound 4 crystal structure
Graph set
notation
Hydrogen
bond
D—HÐ Ð ÐA
DÐ Ð ÐA
DHA
˚
˚
°
(A)
(A)
angle ( )
‘209’ dimer
Sub dimer
R22(18)
R22(8)
R22(14)
R22(10)
C11(12)
S11(5)
NHÐ Ð ÐO
NHÐ Ð ÐN
2.09
2.46
2.93
3.23
165.9
150.5
‘203’ dimer
Sub dimer
OHÐ Ð ÐN
2.03
2.33
2.80
3.09
155.8
147.4
NHÐ Ð ÐO
3D chain
OHÐ Ð ÐO
NHÐ Ð ÐO
1.97
2.41
2.75
2.73
160.1
102.1
Intramolecular
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 324–336

Structural investigations of triazines
335
was dissolved in 200 ml of water to form a white suspension.
Diethanolamine (17.0 g, 162.6 mmol) was added and the pH adjusted
to pH 9 with NaOHꢂaqꢁ (10 ml, 2 M). The solution was stirred for 4 h at
of the intermolecular interactions present in the crystal
structure.
For dimer 203, the t4 calculated shift shows an increased
shielding of 13.0 ppm relative to the monomer. Correspond-
ingly for dimer 209, t2 shows an increased shielding of
15.7 ppm. Both of these triazine nitrogen atoms are involved
in intermolecular hydrogen bonding. Such effects have been
discussed in the literature, e.g. Ref. 18. The solid-state 15N
experimental shifts for t2 and t4 can be rationalized and
assigned in terms of such increased shielding. These calcu-
lated dimer shifts are in bold in Table 6. Consideration of
these changes between the monomer and dimer calculations
has been used to assign the experimental solid-state NMR
shifts. Only relatively small differences are observed between
the calculated shifts for the monomer and the dimers for the
other nitrogen atoms.
°
70 C and the reaction, to give 6, was shown to be complete by HPLC.
°
The product was recrystallized from ethanol. M.p. 169.0–171.0 C
(found: C, 43.37; H, 5.47; N, 16.57. C18H25N6O6Cl requires C, 47.31;
H, 5.52; N, 18.40%); IR, ꢃmax(neat)/cmꢀ1 3212 (NH), 2886 (OH), 2588
(OH), 1710 (CO), 1550 (tr), 1504 (Ph), 1428 (tr), 828 (tr), 750 (CCl);
MS, m/z (ESꢀ) [found: 455.1463 (Mꢀ ꢀ 1). C18H25N6O6Cl requires
455.1446].
NMR spectroscopy
Solution-state NMR
Solution-state 1H, 13C and 15N spectra were recorded at
400.14, 100.63 and 40.57 MHz, respectively, on a Bruker
AMX400 spectrometer with either DMF-d7 or DMSO-d6 as
solvent. A 5 mm dual 1H/13C probehead and a 10 mm
broadband probehead were used for the 1H/13C and 15N
spectra, respectively. TMS was used as an internal reference
compound for the 1H and 13C NMR spectra. 15N spectra were
EXPERIMENTAL
1
referenced indirectly to the signal for CH3NO2 via the H
Compound preparation
NMR signal of internal TMS using a value19 of 10.136767%
MHz for , the frequency ratio of the CH3NO2 signal to that
of the 1H resonance of TMS in CDCl3 at ϕ D 1%. Inverse
Compound 2
A solution of aniline (10 g, 108.4 mmol) in acetone (150 ml) was
added to a solution of cyanuric chloride (20 g, 108 mmol) in acetone
°
(183 ml) at 0 C. Aqueous sodium hydrogencarbonate (9.2 g in
15
°
gated decoupling was used for the N spectra. A 30 pulse
111 ml, 108.4 mmol) was then added to the mixture over 30 min
1
°
°
whilst maintaining a temperature of 0–5 C. The solution was stirred
angle was used for the H NMR spectra and a 90 pulse
angle for the 13C and 15N NMR spectra. Pulse delays of 10,
for 1.5 h. Reaction was shown to be complete by high-performance
liquid chromatography (HPLC). HClꢂaqꢁ (20 ml, 2 M) was added until
a precipitate formed which was filtered and washed with acetone.
2 and 10 s were used for the H, 13C and 15N NMR spectra,
1
respectively.
Compound 3
2-Anilino-4,6-bis[N,N-bis(2-hydroxyethyl)amino]-s-triazine. Comp-
ound 2 (5 g, 92.21 mmol) was added to diethanolamine (50 ml) and
Solid-state NMR
°
heated at 90 C for 4 h. The solution was poured into water (500 ml)
Solid-state 13C and 15N magic-angle spinning (MAS) NMR
spectra were recorded on a Varian Unity Plus 300 spectrom-
eter at 75.43 and 30.40 MHz, respectively, at ambient probe
temperature. A Doty Scientific probe was used, with 7 mm
o.d. zirconia rotors and Kel-F end caps. Cross-polarization
(CP) from protons was employed, together with high-power
proton decoupling during signal acquisition. 13C chemical
shifts were referenced by replacement against the high-
frequency line of solid adamantane at 38.4 ppm relative to
the signal for TMS. 15N chemical shifts were referenced by
replacement against the high-frequency line of solid ammo-
nium nitrate at ꢀ5.1 ppm relative to the signal for CH3NO2.
and HClꢂaqꢁ (30 ml, 2 M) was added until an acidic suspension was
formed, which was then filtered and dried. It was recrystallized
°
from ethyl acetate. M.p. 107.5–109.0 C (found: C, 53.79; H, 6.94;
N, 22.18. C17H26N6O4 requires C, 53.95; H, 6.93; N, 22.21%); IR,
ꢃmax(neat)/cmꢀ1 3264 (OH), 2926 (CH2ꢁ, 2860 (CH2), 1578 (triazine
(tr)), 1424 (tr) and 803 (tr); MS, m/z (ESC) 379 (MC C 1).
Compound 4
2-Anilino-4,6-bis[N-(2-hydroxyethyl)amino]-s-triazine. Compound
2 (5 g, 92.21 mmol) was added to ethanolamine (50 ml) and heated
°
at 90 C for 4 h. The reaction mixture was poured into water (500 ml)
and the product extracted into dichloromethane (200 ml). HClꢂaqꢁ
(50 ml, 2 M) was added until the aqueous layer was acidic. The
product was washed until the aqueous layer was clear and colourless
and the solvent then removed. The product was recrystallized
°
from ethanol. M.p. 128.0–128.5 C (found: C, 53.53; H, 6.31; N,
28.67. C13H18N6O2 requires C, 53.78; H, 6.25; N 28.95%); IR,
ꢃmax(neat)/cmꢀ1 3424 (NH), 3291 (OH), 2971 (CH2ꢁ, 2941 (CH2ꢁ,
2897 (CH2ꢁ, 1550 (tr), 1420 (tr), 800 (tr); MS, m/z (ESC) 291 (MC C 1).
X-ray diffraction, molecular structure generation
and refinement
Initial powder diffraction data of 4 were collected using Cu
K˛1 radiation on a Bruker d8 diffractometer equipped with
a Ge(111) incident beam monochromator and an M. Braun
linear position sensitive detector on a sample loaded in a
0.5 mm capillary. Data were indexed using the programme
visser.20 Data for structure solution and refinement were
recorded on beamline x7a of the National Synchrotron Light
Source at Brookhaven National Laboratory at a wavelength
Compound 5
To an aqueous solution of 3-carboxy-4-chloroaniline (25.6 g,
108.4 mmol in 360 ml water), basified to pH 7 with NaOHꢂaqꢁ
°
(40 ml, 2 M) at 0 C was added a solution of cyanuric chloride (20 g,
108.4 mmol) in acetone (180 ml) and the system was stirred for 2 h.
The reaction was shown to be complete by HPLC. This product was
not isolated.
Compound 6
˚
°
of 0.7452 A from 3 to 40 2Â.
2-(3-Chloro-4-carboxyanilino)-[N,N-bis(2-hydroxyethyl)amino]-s-
triazine. Diethanolamine (12.5 g, 119.2 mmol) was added to a
solution of 5 [17.3 g, 54.2 mmol in 90 ml of acetone, 180 ml of water,
20 ml of 2 M NaOHꢂaqꢁ] at room temperature and pH 7 and the
reaction mixture stirred for 18 h. Replacement of the first triazine
ring chlorine atom to give 2-(3-carboxy-4-chlorophenylamino)-4-
diethanolamino-6-chloro-1,3,5-triazine was shown to be complete
by HPLC. HClꢂaqꢁ (20 ml, 2 M) was added until a precipitate formed,
which was filtered and washed with HClꢂaqꢁ (50 ml, 2 M). The product
Structure solution was performed by simulated annealing
using the DASH suite of software.17 A total of six rotations
and translations of the whole molecule along with eight
internal torsion angles were systematically varied and
compared with the experimental powder pattern. From these
calculations a plausible trial structure was readily obtained.
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 324–336

336
H. E. Birkett et al.
Rietveld refinement was then carried out in the GSAS
software suite21 using synchrotron data. Bond distances were
restrained using values derived from similar molecules in
the Cambridge Crystallographic Database.7 Internal bond
angles of the phenyl ring and triazine ring were restrained
4 (39 atoms) has been fully solved from powder XRD
data and is shown to have conformation IV (see Fig. 6).
The solid-state NMR spectrum was consistent with this
structure. Computations on structures involving hydrogen
bonding predicted 15N chemical shift effects which compare
favourably with those observed by CP/MAS NMR.
°
to values of 120, 113.9 and 126.1 , respectively. A planarity
restraint was also applied to the rings. Restraint weights
were adjusted such that the final model had bond distances
Acknowledgements
˚
°
within 0.02 A of their ideal values, bond angles within 2.5
This work was funded by a CASE studentship (to H.E.B.) from the
EPSRC and Avecia Ltd. EPSRC is thanked for financial support
under research grant GR/N05635 and EPSRC/JREI for provision
of laboratory x-ray diffraction equipment under grant GR/M35222.
We also thank Tom Vogt of Brookhaven National Laboratory for
collecting the synchrotron powder diffraction data.
˚
of ideal values, and ring atoms within 0.02 A of planarity.
The unambiguous H atoms were placed geometrically at
standard distances. Positions of H-191 and H-211 were
determined by molecular modelling. For final cycles of
refinement, hydrogen coordinates were allowed to ‘ride’
on the attached C/N/O atoms, and a total of 92 parameters
were refined: 1 scale factor, 6 cell parameters, 15 background
terms, 5 profile coefficients, 1 zero point, 1 overall isotropic
temperature factor and 63 fractional coordinates. Details of
the crystal structure have been submitted to the Cambridge
Crystallographic Database.
REFERENCES
1. Birkett HE, Harris RK, Hodgkinson P, Carr K, Charlton MH,
Cherryman JC, Chippendale AM, Glover RP. Magn. Reson. Chem.
2000; 38: 504.
2. Katritzky AR, Oniciu DC, Ghiviriga I, Barcock RA. J. Chem. Soc.,
Perkin Trans. 2 1995; 785.
3. Katritzky AR, Ghiviriga I, Steel PJ, Oniciu DC. J. Chem. Soc.,
Perkin Trans. 2 1996; 443.
4. Amm M, Platzer N, Guilhem J, Bouchet JP, Volland JP. Magn.
Reson. Chem 1998; 36: 587.
5. Amm M, Platzer N, Bouchet JP, Volland JP. Magn. Reson. Chem
2001; 39: 77.
6. Birkett HE, Cherryman JC, Chippendale AM, Hazendonk P,
Harris RK. J. Mol. Struct. 2002; 602–603: 59.
7. Allen FH, Kennard O. Chem. Desi. Autom. News 1993; 8: 31.
8. Comins DL, O’Connor S. Adv. Heterocycl. Chem. 1988; 44: 243.
9. Kliot VG, Kutepov DF. Plast. Massy 1973; 13.
10. Balimann GE, Groombridge CJ, Harris RK, Packer KJ, Say BJ,
Tanner SF. Philos. Trans. R. Soc. London, Ser. A 1981; 299: 643.
11. Chippendale AM, Mathias A, Aujla RS, Harris RK, Packer KJ,
Say BJ. J. Chem. Soc., Perkin Trans. 2 1983; 1357.
12. McGeorge G, Harris RK, Chippendale AM, Bullock JF. J. Chem.
Soc., Perkin Trans. 2 1996; 1733.
13. Smith EDL, Hammond RB, Jones MJ, Roberts KJ, Mitchell JBO,
Price SL, Harris RK, Apperley DC, Cherryman JC, Docherty R.
J. Phys. Chem. B 2001; 105: 5818.
14. Christopher EA, Harris RK, Fletton RA. Solid State NMR 1992; 1:
93.
15. Harris RK, Olivieri AC. Prog. Nucl. Magn. Reson. Spectrosc. 1992;
24: 435.
16. Harris KDM, Tremayne M, Kariuki BM. Angew. Chem., Int. Ed.
Engl. 2001; 40: 1626.
17. (a) David WIF, Shankland K. Dash Software. Cambridge
Crystallographic Data Centre: Cambridge; (b) David WIF,
Shankland K, Shankland NJ. Chem. Soc. Chem. Comm. 1998; 931.
18. Lorente P, Shenderovich IG, Golubev NS, Denisov GS, Bun-
tkowsky G, Limbach H-H. Magn. Reson. Chem. 2001; 39: S18.
19. Harris RK, Becker ED, de Menezes SMC, Goodfellow R,
Granger P. Pure Appl. Chem. 2001; 73: 1795.
20. Visser JW. J. Appl. Crystallogr. 1969; 2: 89.
21. Larson AC, Dreele RB. V. Los Alamos Internal Report No. 86-748,
1994.
22. UniChem V5.0. Oxford Molecular: Oxford, 2000.
23. Hansen AE, Bouman TD. J. Chem. Phys. 1985; 82: 5035.
24. Kukolich SG. J. Am. Chem. Soc. 1975; 97: 5704.
25. Jameson CJ, Jameson AK, Oppunsunggu D, Wille S, Burrel PM,
Mason J. J. Chem. Phys. 1981; 74: 81.
Ab initio MO and 15N NMR calculations
All ab initio calculations were performed with DGauss,
version 5.0, within UniChem,22 with the BPW91/DZVP
(DFT) level of theory and the LORG23 method to calculate the
nuclear magnetic shielding tensors. Using the powder XRD
structure of 4 as a starting point, calculations were performed
on energy-minimized structures for a single molecule and
for dimers 203 and 209. Calculation for the complete
209/203 trimer would have been too time consuming. The
calculated, isotropic shielding values (ꢄꢁ were converted to
chemical shifts. The isotropic shielding constant of NH145NO3
has been experimentally determined24,25 as ꢀ116 ppm.
The experimental 15N chemical shifts in this paper were
referenced to CH3NO2, which is 5.1 ppm less shielded
than NH145NO3, giving the former an isotropic shielding of
ꢀ121.1 ppm. The DFT method allows for electron correlation,
althoughfor15NNMRchemicalshiftsit tendsto predictNMR
values that are deshielded in comparison with experimental
results. To allow for this, the calculated values were rebased
by 30 ppm, as in the case of the previously reported13
shifts for the theophylline molecule, giving a relationship
υN/ppm D ꢀ151.1–ꢄ/ppm.
CONCLUSIONS
We have shown that for the series of model triazines 3,
4 and 6 the barriers to internal rotation follow the order
˛ < ˇ < ꢀ ¾ υ (see 1). Variable-temperature solution-state
13C and 1H NMR have amply demonstrated this order, and it
is consistent with earlier work1 on the more complicated
dyestuff 1. Moreover, molecular modelling calculations6
support this order. The low-temperature solution-state NMR
spectra provided information on the populations of the
various conformations. The crystal structure of compound
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 324–336