The influence of weak hydrogen bonds on the properties of 3-cyano-N-methylpyridinium chloride and iodide

Article abstract of DOI:10.1023/A:1024230130658

The iodide salt of 3-cyano-N-methylpyridinium cation crystallizes as yellow needles in the monoclinic space group P2₁/n with a = 4.8726(5), b = 12.9380(15), c = 13.4629(15) A, β = 97.463(2)°, and Z = 4 (mp 196°C). The chloride salt forms colorless plates in the monoclinic space group P2₁/m with a = 7.9488(8), b = 6.4023(7), c = 8.0844(8) A, β = 112.987(2)°, and Z = 2 (mp > 220°C). For ring hydrogens interacting with the anion, each salt has C-H ... X distances and angles consistent with weak hydrogen bonds as described by recently published criteria (Brammer et al. Cryst. Growth Design 2001, 1, 277; Steiner Acta Crystallogr., Sect B 1998, 54, 456; Crystallogr. Rev. 1996, 6, 1). The chloride salt has an additional interaction between H5 and the cyano nitrogen on an adjacent cation in the same layer, or a total of four coplanar hydrogen bonds per cation. It forms sheets which stack along the b axis. The iodide has fewer hydrogen bonds from the ring and the cations are tilted relative to each other. There is an unusual hydrogen bond from a methyl hydrogen to a cyano nitrogen in the iodide. Roughly oval (3.8 x 5.7 A) open channels bounded by H5 and iodides run along the a axis. Stabilization of these solids by hydrogen bonding is estimated to be at least 1.9 kcal/mole more for the chloride than for the iodide.

Full text of DOI:10.1023/A:1024230130658

C
Journal of Chemical Crystallography, Vol. 33, Nos. 5/6, June 2003 ( 2003)
The influence of weak hydrogen bonds on the properties
of 3-cyano-N-methylpyridinium chloride and iodide
Lynn Vogel Koplitz,(1) Kevin D. Bay,(1) Neil DiGiovanni,(1) and Joel T. Mague(2)
Received August 4, 2002
The iodide salt of 3-cyano-N-methylpyridinium cation crystallizes as yellow needles in the
˚
monoclinic space group P21/n with a = 4.8726(5), b = 12.9380(15), c = 13.4629(15) A,
= 97.463(2) , and Z = 4 (mp 196 C). The chloride salt forms colorless plates in the
˚
monoclinic space group P21/m with a = 7.9488(8), b = 6.4023(7), c = 8.0844(8) A,
=
112.987(2) , and Z = 2 (mp > 220 C). For ring hydrogens interacting with the anion, each
---
X distances and angles consistent with weak hydrogen bonds as described
salt has C
H
by recently published criteria (Brammer et al. Cryst. Growth Design 2001, 1, 277; Steiner
Acta Crystallogr., Sect B 1998, 54, 456; Crystallogr. Rev. 1996, 6, 1). The chloride salt has
an additional interaction between H5 and the cyano nitrogen on an adjacent cation in the
same layer, or a total of four coplanar hydrogen bonds per cation. It forms sheets which
stack along the b axis. The iodide has fewer hydrogen bonds from the ring and the cations
are tilted relative to each other. There is an unusual hydrogen bond from a methyl hydrogen
˚
to a cyano nitrogen in the iodide. Roughly oval (3.8 5.7 A) open channels bounded by
H5 and iodides run along the a axis. Stabilization of these solids by hydrogen bonding is
estimated to be at least 1.9 kcal/mole more for the chloride than for the iodide.
KEY WORDS: Weak hydrogen bond; 3-cyano-N-methylpyridinium chloride; iodide.
Introduction
idinium iodide (CMPI) melts at 196 C while
CMPCl decomposes by charring at around 220 C,
While investigating 3-cyano-N-methyl-
we suspected that the chloride was more ionically
bound while the iodide, with a larger, softer
anion, might exhibit more covalency. However,
when we were able to grow and analyze crystals
of the chloride for comparison with the iodide, we
determined that hydrogen bonding also plays an
important role in stabilizing both solids as well as
influencing the details of their crystal structures.
Recent analyses of intermolecular interac-
pyridinium chloride (CMPCl) as a possible ana-
lytical reagent for certain nucleophiles, we dis-
covered that the iodide precursor readily formed
long yellow needles from concentrated aqueous
solutions. We used the crystals to help character-
ize this intermediate product by x-ray diffraction.
Since we found that 3-cyano-N-methylpyr-
tions in a wide variety of compounds by Brammer,
Bruton, and Sherwood,1 and by Steiner 2,3 show
persuasively that even H atoms bound to carbon
can form weak hydrogen bonds not only with con-
ventional hydrogen bond acceptors such as O, N,
(1) Chemistry Department, Loyola University-New Orleans,
New Orleans, Louisiana 70118.
(2) Department of Chemistry, Tulane University, New Orleans,
Louisiana.
To whom correspondence should be addressed. E-mail: koplitz
@loyno.edu
391
C
1074-1542/03/0600-0391/0 2003 Plenum Publishing Corporation

392
Koplitz, Bay, DiGiovanni, and Mague
and F but also with halide ions. We give a detailed
analysis of the extent to which the interactions in
CMPI and CMPCl conform to the geometric pref-
erences for hydrogen bonds established in these
reviews. Cation distortions in the solids relative
to a calculated gas-phase ion are also consistent
with the hydrogen bonding interactions described
herein.
liquid remains in the bottom of the flask. Addition
of absolute ethanol to this liquid causes CMPCl to
precipitate as a white solid which can be collected
and dried by suction filtration and desiccation. It
decomposes by charring at about 220 C so that
no melting point was determined. It is typical to
obtain yields of no more than 50%.
Crystal growth
Experimental
CMPI grows long (1–2 cm) yellow needles
from aqueous solutions ( 0.2 g per mL) within
a few hours. Crystals can also be grown from
methanol, ethanol, or a 1:1 mixture of acetoni-
trile and ethanol ( 0.02 g in 0.8 mL). CMPI crys-
tals have been observed to be stable in air for
more than a year. We have only been successful at
growing small CMPCl crystals from a 1:1 mixture
of acetonitrile and ethanol ( 0.02 g in 0.8 mL)
with one drop of 1 M HCl(aq). Left in a shal-
low pool in a tilted Petri dish, this mixture evapo-
ratesovernighttogivecolorlessflatplates. CMPCl
crystals degrade if exposed to air for more than a
few days.
CMPI synthesis (see also ref. 4)
The synthesis of 3-cyano-N-methylpyr-
idinium iodide (CMPI) was carried out by
dissolving 10.5 g of 3-cyanopyridine in 20-mL
benzene and 21.5 g of methyl iodide in 20-mL
benzene. (Handle methyl iodide with extreme
care—it is a vesiccant which causes blisters on
contact with skin.) These solutions were mixed to-
gether in a round-bottom flask, closed with a glass
stopper and stirred for at least 2 h. The resulting
solution is orange yellow. This solution was then
refluxed for 5–8 h. All work was performed in a
fume hood. The resulting bright yellow slush was
stored in a refrigerator overnight then suction
filtered to collect the solid product which was
subsequently dried in a desiccator. Typical yields
were 30–45%. Melting point is 196 C (melt is
blood red) in agreement with Pfleiderer et al.4
X-ray crystallography
Crystals of CMPI and CMPCl, obtained as
described earlier, were cut to size and mounted on
thin glass fibers with silicone grease. Full spheres
of data were collected at 100 K on a Bruker
Smart APEX CCD diffractometer by using nar-
row frames (0.300 scans in ω at = 0, 120,
and 240 ; detector swing angle = 28.00 ) and
converted to integrated intensities using standard
procedures.5 Details specific to the present study
are presented in Table 1. In both instances, the
choice of space group was made with a combina-
tion of observed systematic absences and inten-
sity statistics. An empirical absorption correction
(SADABS5) was performed for CMPI. The loca-
tion of the halide ion in each compound was ob-
tainedfromaPattersonfunctionandtheremainder
of the structure developed by successive cycles of
full-matrix, least-squares refinement followed by
CMPCl synthesis (see also ref. 4)
The chloride is obtained from the iodide
by anion exchange of an aqueous solution us-
ing Dowex 1 8-400 (strongly basic anion, 8%
cross linking, 200–400 mesh) prepared accord-
ing to the manufacturer’s directions. (The ion ex-
change resin tends to turn yellow, probably be-
cause of an irreversible reaction with some of
the hydrolyzed cation. It is only possible to use
it once.) Eluent containing dissolved 3-cyano-N-
methylpyridinium chloride (CMPCl ) is collected
and rotovapped to remove excess water. When
most of the solvent has been removed, an oily

H-bonding in N-methylpyridinium halides
393
Table 1. Crystal Data and Structure Refinement
201734 201735
CCD deposit no.
Compound
CMPI
CMPCl
Empirical formula
Formula weight
Temperature, K
C7H7IN2
246.05
100(1)
C7H7ClN2
154.6
100(1)
˚
Wavelength, A
0.71073
Monoclinic
P2(1)/n
0.71073
Monoclinic
P2(1)/m
Crystal system
Space group
˚
˚
˚
Unit cell dimensions
a = 4.8726(5) A
= 90
a = 7.9488(8) A
= 90
˚
˚
b = 12.9380(15) A
= 97.463(2)
= 90
b = 6.4023(7) A
= 112.987(2)
= 90
˚
c = 13.4629(15) A
c = 8.0844(8) A
3
˚
Volume, A
841.53(16)
4
1.942
3.733
378.75(7)
2
1.356
0.424
160
Z
Density (calculated), Mg/m3
Absorption coefficient, mm
F(000)
1
464
Crystal size
range for data collection
Index ranges
0.271.183 0.083 mm
2.19–28.72
0.285 0.217 0.12 mm
2.74–28.26
6
h
6, 16
k
17,
10
10
h
l
10,
10
8
k
8,
17
7141
l
17
Reflections collected
3495
Independent reflections
2017 (Rint = 0.0161)
999 (Rint = 0.0287)
Completeness (to
)
92.1% (28.72 )
97.3% (28.26 )
Refinement method
Full-matrix least squares on F2
2017/0/92
1.139
R1 = 0.0167, wR2 = 0.0412
R1 = 0.0172, wR2 = 0.0414
0.0171(6)
Full-matrix least squares on F2
999/0/61
1.135
R1 = 0.0288, wR2 = 0.0752
R1 = 0.0301, wR2 = 0.0758
None
Data/restraints/parameters
Goodness of fit on F2
Final R indices [I > 2 (I)]
R indices (all data)
Extinction coefficient
Largest diff. peak and hole, eA
3
˚
0.586 and 0.632
0.335 and 0.223
calculation of a difference map. In both instances,
good positions for all of the hydrogen atoms could
be obtained from difference maps calculated fol-
lowing completion of the refinement of the non-
hydrogen atoms. These locations were used for
the hydrogen atoms, which were allowed to ride
on the attached carbon atoms with fixed isotropic
displacement parameters 20% larger than those
of the latter. All calculations were performed with
the SHELXTL-PLUS6 program package.
the cation geometry. All resultant bond lengths
and angles are given in Tables 2 and 3.
Results and discussion
The crystallographic features of CMPI and
CMPCl are summarized in Table 1 (see CIF files
and Tables 3–9 for greater detail). As can be seen
from this summary, exchange of chloride for io-
dide produces drastic differences in the properties
of the CMP halide crystals even though these sim-
ple anions are very similar in that they have the
same charge, come from the same chemical fam-
ily, and differ by only about 20% in ionic radius.
We might have reasonably expected the two salts
to have essentially the same structure but with a
somewhat smaller unit cell for the chloride. This
proved not to be the case. The unit cell of CMPI
Calculation of gas-phase CMP + properties
MacSpartan Plus software7 was used to cal-
culate properties of the gas-phase cation. Within
this software constellation, we ran the ABINITIO
program, Release 1.1.9, using the Hartree–Fock
approximation and 6-31G* basis set8 to optimize

394
Koplitz, Bay, DiGiovanni, and Mague
Table 2. A Comparison of Selected Experimental Bond Angles and Lengths for the Cation
( )
Cyano group
CMP+
CMPI
CMPCl
CMP+
CMPI
CMPCl
˚
(A)
--- ---
N2 C7 C3
179.8
177.7(2) 179.59(18)
Methyl group
--- ---
H1a C1 H1b
110
111
110
110(1)
109
109
106
118
107
110(7)
109
107
109
113
109
110(2)
108
109
N1-C1
1.486
1.496(3) 1.481(2)
0
--- ---
H1a C1 H1c (or a0)
--- ---
---
C1 H1a
H1b C1 H1c (or a )
1.079
1.077
0.94
1.05
0.93
0.94
1.00
---
Average (SD)
C1 H1b
--- ---
---
C1 H1c
N1 C1 H1a
0
--- ---
N1 C1 H1b
--- ---
N1 C1 H1c (or a )
Average (SD)
---
C1 H1a
1.079
1.078(1) 0.97(7)
0.94
0.96(3)
0
109
109(<1)
110
109(2)
108
108(1)
avg (SD)
Ring hydrogens
--- ---
--- ---
H2 C2 N1
---
H2 C2
H2 C2 C3
122
117
121
120
120
122
118
124
118
116
122
118
120
122
124
1.071
1.074
1.072
1.072
0.96
1.02
1.07
1.02
0.82
--- ---
---
H4 C4
H4 C4 C5
0.86
--- ---
H4 C4 C3
--- ---
---
H5 C5
H5 C5 C6
0.90
--- ---
H5 C5 C4
--- ---
H6 C6 N1
121
116
124
120
116
115
---
H6 C6
0.98
--- ---
H6 C6 C5
123
119
125
Average (SD)
120(2)
120(3)
120(4)
avg (SD)
1.072(1) 1.02(5)
0.89(7)
Note: Significant differences with calculated gas-phase cation noted by boldface (at least 4 for angles including H atoms);
such differences between the two salts are indicated in italic.
consists of four ion pairs having the cation rings
˚
tilted relative to one another and oriented with H5
Table 3. Bond Lengths (A) and Angles ( ) for the Cation (Calculated
Gas-Phase for CMP+)
pointed in toward an open oval channel bounded
˚
by anions about 5.7 A apart on the major axis
CMP+
CMPI
CMPCl
˚
and H5 ring hydrogens about 3.8 A apart on the
minor axis (see Fig. 1). The cations spiral along
the a axis of the crystal alternating with the io-
Bond lengths
---
N(1) C(2)
1.332
1.342
1.486
1.134
1.380
1.390
1.440
1.388
1.370
1.345(3)
1.354(3)
1.496(3)
1.146(3)
1.392(3)
1.402(3)
1.446(3)
1.393(3)
1.392(3)
1.341(2)
1.344(2)
1.481(2)
1.145(2)
1.379(2)
1.390(2)
1.443(2)
1.388(2)
1.379(2)
---
N(1) C(6)
---
N(1) C(1)
---
N(2) C(7)
---
Table 4. Atomic Coordinates (104) and Equivalent Isotropic
C(2) C(3)
3
2
˚
Displacement Parameters (10 A ) for CMPI
---
C(3) C(4)
---
C(3) C(7)
---
x
y
z
Ueq
C(4) C(5)
---
C(5) C(6)
I
54(1)
7016(3)
4616(4)
5834(5)
5901(4)
6927(4)
9174(4)
10309(4)
9183(4)
5651(4)
8454(1)
5991(1)
8818(2)
5602(2)
6830(2)
7185(2)
6693(2)
5840(2)
5494(2)
8091(2)
1505(1)
2413(1)
4382(2)
1401(2)
2794(2)
3746(2)
4311(2)
3883(2)
2938(2)
4119(2)
15(1)
16(1)
27(1)
22(1)
16(1)
16(1)
18(1)
19(1)
18(1)
20(1)
Bond angles
N(1)
N(2)
C(1)
C(2)
C(3)
C(4)
C(5)
C(6)
C(7)
---
---
C(2) N(1) C(6)
120.8
120.4
118.7
120.8
119.0
119.3
121.7
119.1
118.9
121.3
179.8
120.65(18)
119.50(17)
119.85(17)
120.24(18)
120.76(19)
117.92(18)
121.30(19)
117.31(19)
120.17(19)
120.84(19)
177.7(2)
121.68(15)
118.22(14)
120.11(14)
119.42(16)
120.77(15)
118.29(15)
120.94(15)
117.94(16)
119.79(16)
120.39(15)
179.59(18)
--- ---
C(2) N(1) C(1)
--- ---
C(6) N(1) C(1)
--- ---
N(1) C(2) C(3)
--- ---
C(2) C(3) C(4)
--- ---
C(2) C(3) C(7)
--- ---
C(4) C(3) C(7)
--- ---
C(5) C(4) C(3)
--- ---
C(6) C(5) C(4)
--- ---
N(1) C(6) C(5)
--- ---
Note: Ueq is defined as one third of the trace of the orthogonalized
ij
N(2) C(7) C(3)
U
tensor

H-bonding in N-methylpyridinium halides
395
Table 7. Atomic Coordinates (104) and Equivalent Isotropic
3
2
˚
Table 5. Anisotropic Displacement Parameters (10 A ) for CMPI
3
2
˚
Displacement Parameters (10 A ) for CMPCl
U11
U22
U33
U23
U13
U12
x
y
z
Ueq
I
13(1)
15(1)
33(1)
27(1)
14(1)
16(1)
15(1)
14(1)
15(1)
21(1)
15(1)
15(1)
23(1)
22(1)
15(1)
13(1)
20(1)
18(1)
15(1)
19(1)
17(1)
17(1)
26(1)
17(1)
18(1)
19(1)
18(1)
23(1)
23(1)
18(1)
1(1)
1(1)
2(1)
4(1)
3(1)
1(1)
2(1)
5(1)
3(1)
2(1)
1(1)
2(1)
5(1)
1(1)
2(1)
5(1)
1(1)
1(1)
5(1)
1(1)
0(1)
1(1)
4(1)
1(1)
1(1)
1(1)
3(1)
1(1)
0(1)
1(1)
Cl
9045(1)
9316(2)
5521(2)
11327(2)
8639(2)
6773(2)
5589(2)
6347(2)
8220(2)
6081(2)
7500
2500
2500
2500
2500
2500
2500
2500
2500
2500
6794(1)
7748(2)
1525(2)
8729(2)
5946(2)
4999(2)
5890(2)
7755(2)
8661(2)
3062(2)
18(1)
17(1)
32(1)
23(1)
18(1)
18(1)
20(1)
22(1)
20(1)
22(1)
N(1)
N(2)
C(1)
C(2)
C(3)
C(4)
C(5)
C(6)
C(7)
N(1)
N(2)
C(1)
C(2)
C(3)
C(4)
C(5)
C(6)
C(7)
Note: The anisotropic displacement factor exponent takes the form:
2[h2a 2U11
+
+ 2hka b U12].
Note: Ueq is defined as one third of the trace of the orthogo-
nalized U tensor.
2
ij
dide ions. The cyano and methyl groups follow
each other in a head-to-tail fashion and are not dis-
ordered. In contrast, the chloride salt crystallizes
in a layer structure with a unit cell of less than
half the volume observed for CMPI. The result is
that the cations in the chloride salt have crystal-
lographically imposed mirror symmetry and the
anions are coplanar with them (see Fig. 2). Suc-
cessive layers are rotated by 180 with respect to
one another allowing the anion in one plane to lie
roughly between cations in the adjacent planes.
ORTEP drawings of the ion pairs of CMPC1 and
CMPI are given, respectively, in Figures 3 and 4.
By carefully examining the interatomic dis-
tances and angles found in the crystal structures of
CMPI and CMPCl, we show herein that the chlo-
ride possesses four hydrogen bonds per CMP+,
three from ring hydrogens to chloride and one
with a cyano nitrogen, while the iodide compound
has similar interactions between the cation and the
anion but lacks the ring-hydrogen–cyano-nitrogen
---
(H5 N2) bond. Apparently the smaller ionic ra-
dius and higher electronegativity of the chloride
ion allow, or promote, the formation of this addi-
tional coplanar hydrogen bond. These hydrogen
bonds may be responsible for holding the CMPCl
cationsinflatsheetswiththechlorideanionswhile
the iodide arranges itself very differently.
Table 10. summarizes the important bond
lengths and angles for our samples and compares
them to results from Brammer et al.1 and Steiner.2
Steiner’s analysis included separating out chem-
---
to
ically different types of bonds, adjusting C H
distances to 1.083 A (common practice
1,2,10
˚
correct for asphericity by using neutron diffrac-
tion measurements11), recalculating associated
3
2
˚
Table 8. Anisotropic Displacement Parameters (10 A ) for CMPCl
U11
U22
U33
U23
U13
U12
Table 6. Hydrogen Coordinates (104) and Isotropic
Cl
18(1)
18(1)
26(1)
17(1)
22(1)
23(1)
17(1)
24(1)
24(1)
21(1)
18(1)
15(1)
48(1)
26(1)
16(1)
16(1)
20(1)
25(1)
19(1)
26(1)
16(1)
16(1)
19(1)
24(1)
18(1)
14(1)
21(1)
21(1)
16(1)
20(1)
0
0
0
0
0
0
0
0
0
0
6(1)
4(1)
6(1)
5(1)
10(1)
5(1)
4(1)
12(1)
7(1)
8(1)
0
0
0
0
0
0
0
0
0
0
3
2
˚
Displacement Parameters (10 A ) for CMPI
N(1)
N(2)
C(1)
C(2)
C(3)
C(4)
C(5)
C(6)
C(7)
x
y
z
Ueq
H(1A)
H(1B)
H(1C)
H(2)
H(4)
H(5)
5292
4028
7480
4452
9886
12006
9960
4835
6038
5617
7184
6986
5403
4847
1534
1190
971
2378
5001
4248
2653
21
21
21
16
22
40
19
Note: The anisotropic displacement factor exponent takes the form:
2
H(6)
2[h2a 2U11
+
+ 2hka b U12].

396
Koplitz, Bay, DiGiovanni, and Mague
Table 9. Hydrogen Coordinates (104) and Isotropic
Displacement Parameters (10 A ) for CMPCl
for our data are given in Table 10 for compar-
3
2
˚
ison. Steiner found the mean H
Cl distance
for hydrogens bound to C2 or C6 in pyridine
rings to be 2.64 0.02 A (the uncertainty given
x
y
z
Ueq
˚
H(1A)
H(1B)
H(2)
H(4)
H(5)
11793
11623
9360
4427
5684
8859
1280
2500
2500
2500
2500
2500
8427
10052
5453
5328
8435
9966
27
27
19
22
25
23
here is two standard deviations representing the
95% confidence interval). For the same group of
atoms, he found the C
3.66 0.02 A. The three ring hydrogens H2, H4,
Cl distance to average
˚
H(6)
and H6 in CMPCl easily fit Steiner’s hydrogen
bond criteria since all the H
Cl distances are
˚
˚
shorter than 2.64 A by at least 0.24 A and the
C Cl distances are all shorter than 3.66 by at
H
X distances, then averaging interactions
˚
˚
where H
H
X was not more than 3.0 A (3.2 for
---
least 0.19 A.
For iodides, Steiner found the average H
I) and the C H
X angle was not less than
I
˚
140 . Similarly recalculated distances and angles
distance to be 2.99 0.04 A and C
I to be
Fig. 1. Packing diagram emphasizing the open channel outlined by H4
I and H6
I close contacts in
CMPI. Looking down the a axis, asymmetric unit labeled. Iodide ions are stacked along a axis separated
˚
by 4.91 A.

H-bonding in N-methylpyridinium halides
397
---
Fig. 2. Layer of CMPCl showing four different hydrogen bonds, three C
H
Cl and one
˚
C5H5
N2. Looking along the b axis. Distance between layers is 3.35 A.
˚
4.00
interval). Again, our CMPI data for H2 I,
C2 I, and C4 I are easily within their respec-
tive ranges. The H4
0.04 A (uncertainty as 95% confidence
of bonds. They have determined that the mean
value of RHX for weak hydrogen bonds between
---
C H donors and halide ion acceptors is 0.960
˚
I distance of 3.03 A just
for chloride and 0.982 for iodide by considering
---
matches the upper end of the 95% confidence in-
terval so this interaction must be considered to be
just cases where RHX
<
1.05 with C H
X
angles of 110 . These investigators also found
---
a marginal case. H6
I is clearly too long since
that, for both halides, the C H
drogen bonds tends to be close to 180 especially
for shorter H X distances.
Because the H Cl distances for the ring
X angle in hy-
˚
the recalculated distance of 3.08 A exceeds even
the 95% confidence limit of the distribution. Note
that the H I interactions for both H6 and H4
would not have been considered at all by Steiner
hydrogens H2, H4, and H6 in CMPCl all re-
sult in RHX values of about 0.8 (<1 and near
0.960) with angles 165 , we can conclude that
all three of these interactions are described ap-
propriately as hydrogen bonds. We also reach the
same conclusion for the iodide case because all
of the RHX and angular parameters for the ring
hydrogens fit into the distributions determined by
Brammer and coworkers (see distribution maps
---
since the C H
I angles are less than 140 .
To map out the geometric preferences for
halogens as weak hydrogen bond acceptors,
Brammer et al.1 used a normalization procedure
fromLommerseetal.9 wherebyeachclosecontact
ofinterestinthecrystalisdividedbythesumofthe
van der Waals radii for the two atoms in contact,
or RHX = d(H
X)/(rH + rX). Here d(H X)
---
is also recalculated after the C H bond length
has been extended along the experimental vec-
on p. 279 of ref. 1). The H2
ifies as a hydrogen bond with RHX = 0.858 and a
---
I interaction qual-
1
˚
tor to 1.08 A. Unlike Steiner, these authors did
C2 H2
I angle of 168 . H4 I can be clas-
notdistinguishbetweenchemicallydifferenttypes
sified as a hydrogen bond since RHX = 0.959 is

398
Koplitz, Bay, DiGiovanni, and Mague
Table 10. Close Contacts and Angles Compared to Criteria Characteristic of Weak Hydrogen
Bonds from Refs. 1 and 2 (Given in Boldface (SD))
˚
Bond lengths (A)
Experimental
Recalculated
Recalculated
--- ---
C
X
H
X
H
X
C
H
X ( )
RHX
CMPCl
C2H2Cl
C4H4Cl
C6H6Cl
C1H1bCla
3.66(1)
3.388(1)
3.469(1)
3.463(1)
3.745(1)
na
2.64(1)
2.33
2.40
2.39
2.73
na
0.960
0.790
0.814
0.810
0.925
na
2.59
165
172
170
157
2.62
2.50
2.80
C5H5N2
3.359(2)
3.216(2)
4.00(2)
3.776(2)
3.843(2)
3.856(2)
na
2.55
2.85
2.40
2.84
2.99(2)
2.71
3.03
3.08
na
148
142
0.857
1.014
0.982
0.858
0.959
0.978
na
C1H1bN2a
CMPI
C2H2I
C4H4I
C6H6I
2.83
3.07
3.13
168
132
130
C1H1cN2
3.556(3)
2.60
2.58
151
0.921
˚
Note: RHX = d(H
X)/(rH + rX) where rH = 1.20, rCl = 1.75, rI = 1.96, and rN = 1.60 A
from Ref. 9. Criteria for comparison with CH
N not available (na).
aThese interactions not counted in stabilization estimate.
---
less than 1 and close to 0.982 with a C4 H4
I
=
be hydrogen bonds. By Steiner’s more restrictive
criteria, we identified only two of these interac-
tions as hydrogen bonds in CMPI.
angle of 132 . H6 I also qualifies with RHX
0.978 0.982 and a C6 H6
---
I angle of 130 .
So using the broader criteria from Brammer et al.,
we arrive at the same count of three ring H Cl
hydrogen bonds per mole of CMPCl, but we find
As can be seen in Table 10 and Figs. 1 and 2,
there is an additional important close contact in
CMPCl that does not exist in CMPI. Here H5
˚
three ring H
I interactions per mole of CMPI to
on one cation ring is only 2.40 A (when the
˚
---
C5 H5 distance is adjusted to 1.08 A) away from
the cyano nitrogen, N2, on the nearest cation in
---
CMPCl. The corresponding C5 H5
N2 angle
is 148 . We were unable to locate an analysis of
a large set of CH N hydrogen bonding inter-
actions so the best comparison available is with a
large data set of CO H bonds.12,13 The average
˚
distance found for terminal CO H was 2.62 A,
˚
so CMPCl has a contact that is 0.22 A shorter.
Typical angles are around 140 while CMPCl
has an even larger, more nearly linear angle of
about 148 . It is also true that Reddy et al.14 re-
ported the synthesis of a 1:1 molecular complex
between 1,3,5-tricyanobenzene and hexamethyl-
benzene where the tricyanobenzene layers show
linear CH N hydrogen bonds. The distance be-
tween the ring carbon and the cyano nitrogen in
Fig. 3. ORTEP perspective of CMPCl (50% probability).

H-bonding in N-methylpyridinium halides
399
Fig. 4. ORTEP perspective of CMPI (50% probability).
˚
Reddy’s compound is either 3.47 or 3.52 A so that
the H
are not disordered. H1c is always closest to the
˚
˚
---
N distance is no more than 2.6 A. There-
nearest N2 at 2.58 A with a C1 H1c
N2 an-
fore, we conclude that CMPCl has a fourth hydro-
gen bond between cations. Apparently this type
of bond is weak relative to other packing forces
in the iodide so that interactions between ring hy-
drogens and cyano nitrogens do not exist in these
CMPI crystals. Figure 1 shows that all H5 sites
are far away from all cyano groups in the iodide
salt.
However, CMPI shows evidence of a very
unusual weak hydrogen bond between a methyl
hydrogen and the cyano nitrogen of an adjacent
cation. As we noted when describing the overall
crystal structures, the cyano and methyl groups
follow each other in a head-to-tail fashion and
gle of 151 . As this distance is less than the
˚
sum of the van der Waals radii (1.60 A for N
9
˚
˚
and 1.20 A for H = 2.80 A ) and less than the
˚
CO
H distance of 2.62 A discussed in the
N angle is
preceding paragraph, and the CH
greater than 140 , it warrants consideration even
though methyl hydrogens are not generally con-
sidered to be acidic enough to form hydrogen
bonds. A similar interaction is noted by Thalladi
et al.15 between a methyl hydrogen and an oxy-
gen atom in trimethylisocyanurate as part of the
1:1 molecular complex formed between this com-
pound and 1,3,5-trinitrobenzene. As discussed by
Steiner and Desiraju16, CH O/N interactions

400
Koplitz, Bay, DiGiovanni, and Mague
are the archetypes for weak hydrogen bonds in or-
ganic and organometallic systems. These authors
also point out that even methyl group hydrogens
can be activated, or acidified, by attachment to
electron-withdrawing groups or atoms (e.g. nitro-
gen atoms in pyridine or isocyanurate rings). Fur-
ther evidence for the existence of this hydrogen
bond is given in our analysis of cation geometries
(vide infra).
to compare cation geometries in the crystals (see
Figs. 3 and 4) to that calculated for a gas-phase
cation. MacSpartan software was used to perform
a Hartree–Fock calculation for this purpose
(6-31G basis set, geometry optimization).7,8
Perhaps the most interesting outcome of this
analysis is that it convincingly supports the exis-
tence of the H1c
N2 hydrogen bond, between
the methyl group on one cation and the cyano
group on another, in CMPI. Useful data for CMPI,
CMPCl, and gas-phase CMP+ (calculated) are
given in Table 2. Note that the bond from the pyri-
We might wonder whether a similar methyl–
cyano hydrogen bond exists in CMPCl. It is true
that H1b is closer than H1a or H1a0 to the nearest
˚
--- ---
---
N2 by about 0.25 A at a C H N angle of 142 .
dinenitrogentothemethylcarbon, N1 C1, issig-
˚
---
However, the H N distance of 2.84 A slightly
exceeds the van der Waals contact of 2.80 A. We
nificantly longer in CMPI than in either CMPCl or
CMP+ since the respective 99% confidence inter-
vals of this bond length in CMPI and CMPCl just
meet at their limits, 1.496 3(0.003) = 1.487 =
1.481 + 3(0.002). This distance is essentially
the same in both CMP+ and CMPCl. In addi-
tion, the almost perfectly linear trio of atoms
˚
choose not to call this interaction a hydrogen bond
until a better basis for judging CH N bonds is
available.
This same methyl-group hydrogen (H1b)
˚
makes a 2.73 A contact with a chloride ion. Cri-
---
--- ---
teria specific to methyl group C H
Cl inter-
N2 C7 C3 is bent by about 2 at C7 in CMPI.
actions are not available but this distance is just
outside the range of 2.64 0.02 A for Steiner’s
This angle also differs significantly (99% con-
fidence level) since 177.7 + 3(0.2) = 178.3 <
179.05 = 179.59 3(0.18). It is easy to envision
these clearly complementary distortions with one
cation bending slightly at C7 while a neighboring
˚
criterion for (NC)Csp2-H. In addition, the C
distance (3.745 A) is longer than Steiner’s crite-
Cl
˚
˚
--- ---
rion of 3.66 0.02 A. However, the C H Cl
angle of 157 and the RHX value of 0.925 are con-
sistent with broader criteria from Brammer et al.
Along with comparisons of bond distances
---
cation elongates somewhat at N1 C1, thereby al-
lowing N2 on one cation and H1c on another to be
closer together in order to facilitate an attractive
force between them (see Fig. 1).
Table 2 also summarizes the experimentally
determined bond angles and lengths for the methyl
groupandringhydrogensfoundinCMPI, CMPCl,
and gas-phase CMP+ (calculated). Because the
hydrogen atom positions were not refined, no
derived esd’s are available. However, esd’s for
and angles with published criteria, cation dis-
tortions relative to non-hydrogen-bonded CMP+
might offer further evidence for the existence of
these weak hydrogen bonds. The only other pub-
lished example of a crystal structure containing
CMP+ is that of 3-cyano-N-methylpyridinium
tetraphenylborate.17 The structure of this com-
pound is dominated by the large anion and affords
no opportunities for hydrogen bonding between
the cations. However, the methyl and cyano
groups of the cation are found to occupy similar
spaces in the anion framework such that the
whole cation is disordered in the crystal. Because
of the disorder, the estimated standard deviations
(esd’s) associated with bond distances and angles
for the cation are too large to allow useful
comparisons with our data. Ultimately we chose
--- ---
the C C H ring angles can be estimated to be
approximately six to eight times those for the
--- ---
C C C angles, or 1.2–1.6 . Thus one needs an-
gular differences of >3.2 to be significant at the
95% confidence level. This crude estimate sug-
gests that differences of 4 may be meaningful so
we have indicated discrepancies of at least that
size between gas-phase CMP+ and either crystal
with boldface while such differences between the
two crystals are shown in italic.

H-bonding in N-methylpyridinium halides
401
--- ---
One might notice that H1a C1 H1c in
locations determined with x-rays attributable to
asphericity as described by Allen.11 It is perhaps
sufficient to note that CMPCl exhibits greater
variance than CMPI in these bond lengths, as well
as for the bond angles as mentioned previously,
even though esd’s for the refined atoms in the
chloride structure are smaller. Wider variation in
these parameters could be an indication of more
extensive, probably stronger, hydrogen bonding
CMPI is the only methyl group angle showing
meaningful differences with the other two cation
types. Indeed, 118 is by far the largest methyl
group angle for any of the three cation types
with almost all of the angles very close to the
sp3 theoretical value of 109.5 . This observation
fits with the discussion in the previous paragraph
since it is H1c that hydrogen bonds with N2 on
another cation.
---
in CMPCl. All of the methyl C H bonds are
˚
calculated to be about 1.08 A while the salts each
The largest distortion in this data for the ring
hydrogens relative to the calculated gas-phase
result is the H5 C5 vector tilted toward C4
˚
exhibit two bonds at about 0.94 A and one longer
˚ ˚
---
bond (1.05 A in CMPI, 1.00 A in CMPCl). We
---
by 5 in CMPCl; the two crystals also differ
from each other by 8 in the orientation of this
vector. This comparison supports the idea that the
note slightly greater variance in C H lengths for
the CMPI methyl group in agreement with the
notion that it participates in hydrogen bonding
with N2 while the methyl group in the CMPCl
cation probably does not.
H5
N2 interaction is an important, structure-
determining force for CMPCl. It was apparently
this type of interaction that created sheets of
1,3,5-tricyanobenzene molecules in Reddy’s
complex.14 Differences between the two solids of
at least 4 for the angles about C4 and C6 are con-
sistent with assignments to ring H X hydrogen
bonds based on reviews as discussed previously.
However, neither salt has angles appreciably dif-
ferent from those about C2 in the gas-phase ion,
or from each other, even though both structures
clearly exhibit hydrogen bonding from this ring
position. Interestingly, CMPCl has the greatest
variance for ring hydrogen angles (standard
deviation of 4 ) even though the CMPCl structure
generally has lower esd’s for the refined atoms. It
should be noted that all of the interior ring angles,
observed as well as calculated, are 120 2 with
As mentioned previously, the cation in 3-
cyano-N-methylpyridinium tetraphenylborate is
disordered because of its location on a site of crys-
tallographic mm2 symmetry. It is noteworthy that
such disorder is not found in CMPI or CMPCl.
This contrast with the tetraphenylborate is one
more testament to the influence of the hydrogen
bonds on the structures of the halide compounds
because those forces are clearly absent, or at least
unable to develop, in the tetraphenylborate.
While it is beyond our expertise to construct
an adequate overall theoretical model for these
crystals, we can make a crude attempt at assign-
ing energies for the special interactions identified
previously to estimate and compare the amount of
stabilization they confer upon these compounds.
An appropriate estimate for the minimum differ-
ence in stabilization between CMPCl and CMPI
can be obtained by assigning a reasonable en-
ergy to each of the hydrogen bonds discussed
previously, summing these energies for each salt,
then subtracting the answers. To obtain a mini-
mum difference, questionable assignments are in-
--- ---
the exception of C5 C4 C3 that is calculated
to be 119 but observed to be 117 in CMPI
and 118 in CMPCl (see Table 3 for complete
list).
---
With respect to C H bond length distortions
---
for the cation in the solids, all of the ring C H
bonds for the gas-phase ion are calculated to be
˚
about 1.07 A while the observed distances from
cluded for CMPI (C6H6
I) but excluded for
CMPCl (C1H1b Cl, C1H1b N2). Follow-
ing Desiraju and Steiner (Ref. 16, p. 110), the
cation–cation interactions can be assigned. This
˚
x-ray diffraction data are about 1.02 A in CMPI
˚
and 0.89 A for CMPCl (see Table 2). Lower
values in the solids are probably primarily due
to typical bond length shortening for hydrogen
approach gives H5
N2 in CMPCl a value of

402
Koplitz, Bay, DiGiovanni, and Mague
---
2 kcal/mol by analogy with C H
O hydrogen
bonds with acidic donors, and H1c N2 in
interactions are observed in CMPCl, either the op-
posing packing forces are weaker than the hydro-
gen bonds or the other packing forces are synergis-
tic with the hydrogen bonds, e.g., chloride is small
enough to allow closer anion–anion packing than
iodide so the hydrogen bonding partners can get
close enough to find one another. In any case, hy-
drogen bonds have a marked effect on the crystal
structures and properties of these two compounds.
CMPI a value of 1 kcal/mole since H1c is part
of an activated methyl group. We did not find
the energy of stabilization for H X interac-
tions with halide ions to be clearly described in
any reference. So we assign H I a value of
0.5 kcal/mole since we expect this bond to be
stronger than H3CH
but probably weaker than NH2
about 1.3 kcal/mole.19 In addition, we count
Cl bonds as about 1.6 times as strong as
analogous H I bonds on the basis of the work
of Peris et al.19 Combining all of this reason-
ing leads to 3 (H N2) = 3(0.5) +
I) + (H1c
1 = 2.5 kcal/mole in CMPI and 3 (H Cl) +
(H5
FCH3 at 0.2 kcal/mole18
--- --- ---
I Ir at
Acknowledgments
H
The authors thank Dean Frank E. Scully for sug-
gesting the original project using CMPCl. We are also
grateful for grant support from the J. Bennett Johnston
Science Foundation.
N2) = 3(1.6)(0.5) + 2 = 4.4 kcal/mole in
CMPCl, or an additional 1.9 kcal/mole stabiliza-
tion in the chloride salt. This greater stabilization
for the chloride salt is consistent with the observed
lower melting point of CMPI. Indeed, the fact that
CMPCl chars at about 220 C rather than changing
phase implies that the energy required to decom-
pose the cation is less than that needed to dissoci-
ate intermolecular forces in the solid.
References
1. Brammer, L.; Bruton, E.A.; Sherwood, S. Cryst. Growth Design
2001, 1, 277.
2. Steiner, T. Acta Crystallogr. Sect B 1998, 54, 456.
3. Steiner, T. Crystallogr. Rev. 1996, 6, 1.
4. Pfleiderer, G.; Sann, E.; Stock, A. Chem. Ber. 1960, 93, 3083.
5. SMART APEX V 5.0, User’s Manual, Bruker-AXS, Madison,
WI, 2001.
6. SHELXTL-PLUS V. 5.1, Bruker-AXS, Madison, WI, 1997.
7. MacSpartanPlus, Wavefunction, Inc., Irvine, CA, 1996.
(www.wavefun.com )
8. Hehre, W.J.; Yu, J.; Klunzinger, P.E.; Lou, L. A Brief Guide
to Molecular Mechanics and Quantum Chemical Calculations;
Wavefunction: Irvine, CA, 1998.
Summary
Four hydrogen bonds are present per cation
in CMPCl, three between ring hydrogens H2, H4,
H6, and chloride ions, and one between H5 and
a cyano nitrogen on another cation. Similar hy-
drogen bonds exist in CMPI between ring hydro-
gens and iodide ions. There is also a hydrogen
bond between a methyl group hydrogen and the
cyano nitrogen in an adjacent cation in CMPI.
9. Bondi, A.J. J. Phys. Chem. 1964, 68, 441.
10. Lommerse, J.P.M.; Stone, A.J.; Taylor, R.; Allen, F.H. J. Am.
Chem. Soc. 1996, 118, 3108.
11. Allen, F.H. Acta Crystallogr. Secr B. 1986, 42, 515.
12. Braga, D.; Grepioni, F. Acc. Chem. Res. 1997, 30, 81.
13. Braga, D.; Grepioni, F.; Biradha, K.; Pedireddi, V.R.; Desiraju,
G.R. J. Am. Chem. Soc. 1995, 117, 3156.
14. Reddy, D.S.; Goud, B.S.; Panneerselvam, K.; Desiraju, G.R. J.
Chem. Soc., Chem. Comm. 1993, 663.
15. Thalladi, V.R.; Panneerselvam, K.; Carrell, C.J.; Carrell, H.L.;
Desiraju, G.R. J. Chem. Soc., Chem. Commun. 1995, 341.
16. Desiraju, G.R.; Steiner, T. The Weak Hydrogen Bond; Oxford
University Press: New York, 1999.
17. Zhu, D; Kochi, K. Organometallics 1999, 18, 161.
18. Howard, J.A.K.; Hoy, V.J.; O’Hagan, D.; Smith, G.T.; Tetrahe-
dron 1996, 52, 12613.
While H5
N2 appears to be the strongest of
all of these interactions based on the work of
other investigators, this type of bond is missing
in CMPI suggesting that other kinds of oppos-
ing packing forces dominate. Since four coplanar
19. Peris, E,; Lee, J.C.; Rambo, J.R.; Eisenstein, O.; Crabtree, R.H.
J. Am. Chem. Soc. 1995, 117, 3485.