ARTICLE pubs.acs.org/JPCA
Extending the Supramolecular Synthon Based Fragment Approach (SBFA) for Transferability of Multipole Charge Density Parameters to Monofluorobenzoic Acids and their Cocrystals with Isonicotinamide: Importance of CH 3 3 3 O, CH 3 3 3 F, and F 3 3 3 F Intermolecular Regions Venkatesha R. Hathwar, Tejender S. Thakur, Ritesh Dubey, Mysore S. Pavan, Tayur N. Guru Row,* and Gautam R. Desiraju* Solid State and Structural Chemistry Unit, Indian Institute of Science, Bangalore 560 012, India
bS Supporting Information ABSTRACT: An extension of the supramolecular synthon-based fragment approach (SBFA) method for transferability of multipole charge density parameters to include weak supramolecular synthons is proposed. In particular, the SBFA method is applied to CH 3 3 3 O, CH 3 3 3 F, and F 3 3 3 F containing synthons. A high resolution charge density study has been performed on 4-fluorobenzoic acid to build a synthon library for CH 3 3 3 F infinite chain interactions. Libraries for CH 3 3 3 O and F 3 3 3 F synthons were taken from earlier work. The SBFA methodology was applied successfully to 2- and 3-fluorobenzoic acids, data sets for which were collected in a routine manner at 100 K, and the modularity of the synthons was demonstrated. Cocrystals of isonicotinamide with all three fluorobenzoic acids were also studied with the SBFA method. The topological analysis of inter- and intramolecular interaction regions was performed using Bader’s AIM approach. This study shows that the SBFA method is generally applicable to generate charge density maps using information from multiple intermolecular regions.
’ INTRODUCTION Charge density distributions obtained from high resolution experimental X-ray diffraction studies provide perhaps the best method of rigorously evaluating intermolecular interactions.1 These interactions are difficult to model with established theoretical approaches if one needs to account for all components of the interaction energy. Recent developments in instrumentation, both in terms of speed and accuracy, have made these X-ray measurements economical and easy to use. The Atoms in Molecules (AIM) theory2 of Bader provides a useful means for the analysis of the modeled electron density obtained from both experiment and theory. The bonding and nonbonding features of the electronic distribution in both the intra- and intermolecular regions can be evaluated. Additionally, multipole parameters obtained from refinements of experimental charge density have been used to compute the electrostatic component of intermolecular interaction energies.3 The topological properties of electron density (Fbcp, r2Fbcp) at the bond critical points (bcp) provide information about the nature and strength of intermolecular interactions. These parameters have been used to quantitatively classify various interactions as van der Waals or hydrogen bonds.4 They have also been used in characterizing intramolecular agostic interactions in organometallic compounds.5 High resolution charge density studies are possible only in compounds where the crystals obtained are of good quality, r 2011 American Chemical Society
Scheme 1. Fluorobenzoic Acids and Their Cocrystals with Isonicotinamide
Special Issue: Richard F. W. Bader Festschrift Received: April 29, 2011 Revised: July 20, 2011 Published: August 03, 2011 12852
dx.doi.org/10.1021/jp2039866 | J. Phys. Chem. A 2011, 115, 12852–12863
The Journal of Physical Chemistry A
ARTICLE
Table 1. Crystallographic and Structure Refinement Details for 16 compound
1
2
3
4
5
6
CCDC No.
823577
823572
823570
823578
823571
823569
molecular formula
C7H5O2F
C7H5O2F
C7H5O2F
C13H11O3N2F
C13H11O3N2F
C13H11O3N2F
formula weight
140.11
140.11
140.11
262.24
262.24
262.24
crystal system
monoclinic
monoclinic
monoclinic
monoclinic
monoclinic
triclinic
space group
P21/c
P21/n
P21/n
C2/c
C2/c
P-1
a (Å)
3.7041(2)
6.746(1)
6.552(1)
22.046(3)
22.913(2)
7.0275(9)
b (Å)
6.2642(4)
3.7192(7)
3.7524(8)
5.2078(6)
5.2046(5)
7.4474(9)
c (Å) R ()
26.660(2) 90
24.285(5) 90
25.021(5) 90
20.995(3) 90
20.013(2) 90
12.601(2) 74.371(5)
β ()
92.960(3)
92.91(3)
92.82(3)
95.888(8)
100.212(7)
87.185(6)
γ ()
90
90
90
90
90
70.004(5)
V (Å3)
617.78(6)
608.6(2)
614.5(2)
2397.7(5)
2348.8(4)
596.1(1)
Z
4
4
4
8
8
2
Fcalc (gcm3)
1.506
1.529
1.515
1.453
1.483
1.455
F(000)
288
288
288
1088
1088
270
μ (mm1) T (K)
0.13 100(2)
0.132 100(2)
0.13 100(2)
0.115 100(2)
0.117 100(2)
0.115 100(2)
λ (Å)
0.71073
0.71075
0.71075
0.71075
0.71075
0.71075
(sin θ/λ)max (Å1)
1.08
0.648
0.644
0.648
0.649
0.649
reflns. collected
81207
5565
5097
12029
11736
6404
unique reflns.
6493
1376
1360
2741
2687
2736
completeness (%)
99.9
99.9
98.9
99.9
99.8
99.8
redundancy
12.51
4.04
3.74
4.39
4.37
2.34
Rint
0.037
0.077
0.128
0.054
0.035
0.019
R1 (F2)
0.044
0.047
0.056
0.045
0.041
0.039
wR 2 (F2)
0.122
0.135
0.183
0.139
0.106
0.107
goodness-of-fit
1.133
1.090
1.140
1.130
1.060
1.030
spherical atom refinement
multipole refinements reflns. used [I > 3σ(I)]
5039
No. of parameters
268
R1 (F2)
0.025
wR2 (F2) goodness-of-fit
0.057 1.48
ΔFmin,max (eÅ3)
0.14, 0.16
diffract well at higher resolutions, and are free from disorder. Because of these limitations, such studies cannot be employed in each and every system. The ability to transfer charge density derived multipole parameters from one system to another provides a good alternative approach.68 Current transferability methods use atom based information obtained either from available experimental charge density libraries or from theory. The modeled charge density multipole parameters are very sensitive to the chemical environment of the atom, and hence, a direct atom-wise transferability can pose problems. In such situations, molecular fragment based approaches like the supramolecular synthon9 based fragment approach (SBFA) method10 provide a more convenient and accurate way of obtaining essential topological features for the intra- and intermolecular space in an organic crystal. We have previously detailed this method with respect to two most commonly observed supramolecular synthons, the OH 3 3 3 O carboxylic acid dimer and the NH 3 3 3 O amide chain.10 The SBFA method uses the supramolecular information that is inherent in intermolecular interaction patterns. Supramolecular
synthons are robust in nature (they are repeated in different structures) and retain their geometrical features over a variety of compounds.9 Consequently, the charge density distribution around the atomic framework of a synthon fragment is also conserved and becomes an intrinsic property of these modular units. Accordingly, the system of interest is divided into chemically reasonable fragments based on the interaction environment and the multipole parameters derived from charge density experiments are transferred directly to these synthon regions. The modeled densities obtained from the SBFA method have shown promising transferability and can be used to obtain “synthetic” charge density distributions. In this paper, we extend our earlier study10 on the transferability of experimental charge density multipole parameters to some new types of hydrogen bonded patterns. More specifically, we consider supramolecular synthons involving weaker interactions like the CH 3 3 3 O and CH 3 3 3 F hydrogen bond. It is by now well established that these weak interactions play an important role in directing supramolecular assembly in chemical and biological systems.11,12 12853
dx.doi.org/10.1021/jp2039866 |J. Phys. Chem. A 2011, 115, 12852–12863
The Journal of Physical Chemistry A Scheme 2. Supramolecular Synthons and the Corresponding in-House SBFA Library Entry
The topological features of weak CH 3 3 3 O bonds have been studied by many researchers using both experimental and theoretical approaches.4,11a11e Notably, the essential characteristic features given by Koch and Popelier permit a classification of these interactions as weak hydrogen bonds as opposed to a van der Waals type interaction.4 It has been well documented that despite the high electronegativity of the F-atom, it does not behave as a strong hydrogen bond acceptor in fluoro-organic compounds. Yet, weak interactions such as CH 3 3 3 FC and F 3 3 3 F are observed often in the crystal structures of these compounds.12 A survey of the Cambridge Structural Database (CSD)13 has shown that these CH 3 3 3 FC interactions lead to the formations of specific synthons.12a Like other halogen 3 3 3 halogen interactions, short F 3 3 3 F contacts are also observed in the crystal structures of fluoro-organic compounds.14 However, the stabilizing nature of these F 3 3 3 F interactions is still being debated within the scientific community and has drawn interest from many researchers.14 The present study reports the synthon based transferability of multipole parameters to model the intermolecular interaction regions in monofluorobenzoic acids and their cocrystals with isonicotinamide. The molecular diagrams for the chosen systems are given in Scheme 1.
ARTICLE
’ EXPERIMENTAL DETAILS Materials. All the compounds used in this study were purchased from Sigma Aldrich and used without further purification. Crystallization. 4-Fluorobenzoic acid (1) and 3-fluorobenzoic acid (2) were crystallized from benzene by slow evaporation at room temperature. Crystals of 2-fluorobenzoic acid (3) were grown by layering using an ether-petroleum ether mixture kept at 5 C. Colorless, needle-shaped crystals were harvested after a few days. The 4-fluorobenzoic acidisonicotinamide cocrystal (4) was obtained by dissolving equimolar quantities of the components in ∼1015 mL of MeOH. To the solution were added a few drops of water and it was kept for slow evaporation at room temperature for two days. Crystals of the 3-fluorobenzoic acidisonicotinamide cocrystal (5) were obtained by dissolving equimolar quantities of the components in CH2Cl2MeOH. The solution was allowed to evaporate for two days. The 2-fluorobenzoic acidisonicotinamide cocrystal (6) was obtained by dissolving equimolar quantities of the components in ∼1015 mL of CH3NO2. The solution was allowed to evaporate for a few days, and colorless, block-shaped crystals were obtained. High Resolution Charge Density Data Collection and Structure Refinement Details. A good quality single crystal of 1, of size 0.25 0.16 0.14 mm, was selected under a polarizing microscope and affixed to a Hampton Research cryoloop using Paratone-N oil for data collection. The crystal was cooled to 100 K with a liquid nitrogen stream using an Oxford cryosystems N2 open flow cryostat. High resolution X-ray data sets up to (sin θ/λ)max = 1.08 Å1 with redundancy (∼13) and completeness (∼100%) were collected on a Bruker Kappa Apex II CCD diffractometer using Mo KR radiation at 100 K (Table 1). Data collection strategies were generated using the COSMO module of the Bruker software suite.15 The crystal-to-detector distance was fixed at 40 mm and the scan width was 0.5 per frame during the data collection. Cell refinement, data integration, and reduction were carried out using the SAINTPLUS program.15 The crystal face indexing was used for a numerical absorption correction.15 Sorting, scaling, and merging of the collected data sets were carried out using the SORTAV program.16 The crystal structure was solved by direct methods and refined in the spherical-atom approximation (based on F2) by using SHELXL9717 with the help of the WinGX package suite.18 Multipole Modeling of the Experimental Data. The charge density modeling and multipolar aspherical atom refinements were performed with XD200619 using the Hansen and Coppens multipole formalism.20 The function minimized was ∑w{|Fo|2 K|Fc|2}2, for all reflections with I > 3σ(I). The core and valence scattering factors of all atoms were derived from Su, Coppens, and Macchi wave functions.21 Initially, the scale factor was refined against the whole resolution range of diffraction data. The scatter plots showing the dependence of Fobs/Fcal with sin θ/λ and the variation of Fobs with Fcal clearly depict the quality of the collected data sets after scaling (see Supporting Information, Figure S1). These plots suggest that the data sets are of very high accuracy to the resolution limits. The positional and anisotropic displacement parameters of the non-hydrogen atoms were refined against the reflections with sin θ/λ > 0.7 Å1. In the next step, the position and displacement parameters of all nonhydrogen atoms were kept fixed to the obtained values and XH bond lengths were constrained to the values determined by neutron diffraction experiments.22 The isotropic displacement parameters for H-atoms were refined using reflections sin θ/λ < 0.7 Å1. The converged 12854
dx.doi.org/10.1021/jp2039866 |J. Phys. Chem. A 2011, 115, 12852–12863
The Journal of Physical Chemistry A
ARTICLE
Figure 1. (a) Residual, (b) deformation, and (c) Laplacian of density maps obtained after multipolar refinement of the experimental and theoretical (CRYSTAL09 B3LYP/6-31G(d,p) periodic calculations) charge density data of 1. Blue (solid lines), red (broken lines), and black (dotted lines) colors represent positive, negative, and zero contours, respectively. Contours are drawn at the intervals of (0.05 e Å3.
Table 2. Topological Analysis of Charge Densities Obtained from Experimental X-ray Data and CRYSTAL09 Calculations for the Intramolecular Region of Compound 1 Fbcp bond
experiment
6-31G(d, p)
r2Fbcp TZPV-DFT
experiment
6-31G(d, p)
ε TZPV-DFT
experiment
6-31G(d, p)
TZPV-DFT
F1C4
1.904
1.795
1.794
12.270
11.237
10.294
0.03
0.01
0.03
O2C7
2.932
2.779
2.728
36.296
32.021
31.121
0.12
0.12
0.11
O1C7
2.448
2.321
2.298
25.230
27.486
24.661
0.15
0.06
0.02
O1H1 C1C2
2.235 2.164
2.300 2.072
2.240 2.014
43.733 18.420
36.473 16.959
29.606 14.336
0.01 0.21
0.01 0.18
0.01 0.18
C1C6
2.164
2.071
2.032
18.407
16.609
14.547
0.19
0.20
0.21
C1C7
1.868
1.846
1.811
14.389
14.019
12.302
0.20
0.18
0.14
C2C3
2.184
2.098
2.071
17.907
17.607
16.244
0.23
0.21
0.23
C2H2
1.985
1.924
1.896
20.276
20.483
18.605
0.05
0.04
0.03
C3C4
2.223
2.157
2.146
20.040
19.732
18.443
0.25
0.23
0.26
C3H3
2.006
1.901
1.878
19.803
19.487
17.913
0.07
0.05
0.05
C4C5 C5C6
2.260 2.184
2.177 2.095
2.141 2.075
21.363 18.571
20.118 17.336
18.182 16.193
0.21 0.22
0.22 0.20
0.24 0.20
C5H5
2.038
1.906
1.866
22.729
20.052
17.833
0.08
0.05
0.04
C6H6
2.042
1.935
1.913
21.632
20.975
19.853
0.06
0.04
0.03
model was used to calculate anisotropic displacement parameters (ADPs) of H-atoms using the SHADE2 analysis.23 Estimated ADPs for H-atoms were kept fixed during the subsequent multipole refinements. This methodology of obtaining estimated ADPs for H-atoms was found to be superior in deriving good experimental charge density models.24 For nonhydrogen atoms, the scale, positional, and anisotropic displacement parameters, Pval, Plm, k, and k0 were allowed to refine in a stepwise manner until convergence was
reached. No chemical symmetry constraints were applied to the structures, whereas separate k and k0 were used to define different atom type based chemical environments. The multipole expansion was truncated at the octupole level (l = 3) for nonhydrogen atoms. For H-atoms, only the monopole, bond-directed dipole (dz) and quadrupole (q3z21) components were allowed during the multipole refinements. The quantitative analysis of the electronic structure distribution was performed with the XDPROP module of the 12855
dx.doi.org/10.1021/jp2039866 |J. Phys. Chem. A 2011, 115, 12852–12863
The Journal of Physical Chemistry A Scheme 3. Supramolecular Synthons (Enclosed Region, IIII) Observed in the Crystal Structure of 3-Fluorobenzoic Acid and the Dissection of the Molecule into Two Supramolecular Synthon Based Fragments (Shown in Red and Blue) for Multipole Transferabilitya
ARTICLE
Table 3. Intermolecular Interaction Metrics for the Structures 16 XH interaction 1
2
3 a
The illustration is with respect of 3-fluorobenzoic acid but is equally appropriate for 1 and 3.
XD software suite. Experimental and all relevant crystallographic details are summarized in Table 1. Routine Data Collection and Structure Refinement Details. Routine data sets for compounds 26 were collected at 100 K (see Table 1) on a Rigaku Mercury375R/M CCD (XtaLAB mini) diffractometer using graphite monochromated Mo KR radiation, equipped with a Rigaku low temperature gas spray cooler. The variable temperature data sets for 1 were collected at 293, 250, 210, and 170 K (Supporting Information, Table S1). In all these cases, data were processed with the Rigaku CrystalClear software.25 Structure solution and refinements were performed with SHELX97 using the WinGX suite. Transferability of Multipole Parameters Using the Supramolecular Synthon Based Fragment Approach (SBFA). The target molecules are divided into chemically reasonable molecular fragments based on their interaction environments (supramolecular synthons). The refined multipole parameters (Pval, Plm, k, and k0 ) obtained from the in-house library of experimental charge density data sets were used for SBFA transferability. The direct transfer of experimental charge density derived multipole features from the SBFA library (Scheme 2) to synthon fragments present in the target molecules was carried out. Scaling and initial refinement of the positional and thermal parameters of the all atoms was carried out using the XD2006 package.19 The H-atoms were fixed to neutron values and the anisotropic thermal parameters of H-atoms were computed using the SHADE2 server.23 Charge neutralization was obtained by fixing the individual atomic monopole to neutral atom values. Thereafter, the atomic monopoles for all atoms were allowed to refine to obtain realistic atomic charge values. All other multipole parameters including k and k0 were kept fixed during the refinements.
’ COMPUTATIONAL DETAILS Single point periodic quantum mechanical calculations at the B3LYP/6-31G(d,p) level26 were carried out using CRYSTAL0927 with geometries obtained from the experimental charge density refinement as input. The shrinking factors (IS1, IS2, and IS3) along with the reciprocal lattice vectors were set to 4 (30 k-points in irreducible Brillouin zone). The bielectronic Coulomb and exchange series values for the truncation parameter were set as ITOL1ITOL4 = 8 and ITOL5 = 17 respectively for the CRYSTAL09 calculations. The level shifter was set to 0.7 Hartree/ cycle. The SCF convergence limit was chosen to ∼106 Hartree.
6
X3 3 3A (Å)
— XH 3 3 3 A ()
0.97 1.08
1.65 2.39
2.610(5) 3.384(5)
171(2) 152(1)
C5H5 3 3 C3H3 3 3
1.08
2.46
3.385(5)
143(3)
1.08
2.51
3.413(7)
141(2)
1.18(4)
1.44(4)
2.622(2)
177(4)
0.94(2)
2.59(2)
3.377(2)
142(2)
1.00(2)
2.66(2)
3.560(2)
150(1)
0.96(2)
2.53(2)
3.405(2)
153(2)
1.21(5) 0.97(2)
1.44(5) 2.58(2)
2.638(2) 3.444(2)
172(4) 149(2)
0.85(3)
2.72(2)
3.486(2)
151(2)
1.00(3)
2.58(3)
3.441(2)
144(2)
1.00(3)
1.60(3)
2.598(2)
176(2)
0.89(2)
2.00(2)
2.885(2)
173(2)
0.93(2)
2.03(2)
2.952(2)
172(2)
0.95(2)
2.41(2)
3.325(2)
163(2)
0.95(2) 0.91(2)
2.57(2) 2.59(2)
3.198(2) 3.192(2)
124(2) 124(2) 145(1)
3 O2 3 F1
O1H1 3 3 3 O2 C6H6 3 3 3 O1 C5H5 3 3 3 O2 C4H4 3 3 3 F1 O1H1 3 3 3 O2 C6H6 3 3 3 O1 3 O2 3 F1
O1H1 3 3 3 N1 N2H2A 3 3 3 O3 N2H2B 3 3 3 O2 C11H11 3 3 3 O2 C3H3 3 3 3 O3 C5H5 3 3 3 O1
5
H3 3 3A (Å)
O1H1 3 3 3 O2 C6H6 3 3 3 O1
C5H5 3 3 C3H3 3 3 4
(Å)
F3 3 3F O1H1 3 3 3 N1 N2H2A 3 3 3 O3
-
2.63(2)
3.835(2)
1.12(3)
1.46(3)
2.576(2)
174(2)
0.90(2)
1.99(2)
2.885(2)
171(2)
N2H2B 3 3 3 O2 C11H11 3 3 3 O2
0.91(2)
2.02(2)
2.924(2)
170(2)
0.95(2)
2.47(2)
3.348(2)
155(2)
C4H4 3 3 3 F1 O1H1 3 3 3 N1 N2H2A 3 3 3 O3
0.97(2)
2.66(2)
3.627(2)
174(2)
1.01(3) 0.89(2)
1.59(3) 2.02(2)
2.592(1) 2.896(2)
175(2) 169(2)
N2H2B 3 3 3 O2 C9H9 3 3 3 O3
0.90(2)
2.10(2)
2.977(2)
168(1)
0.97(2)
2.55(2)
3.402(2)
147(1)
0.97(2)
2.60(3)
3.377(2)
147(1)
0.95(2)
2.62(2)
3.550(2)
167(2)
C5H5 3 3 C3H3 3 3
3 O2 3 F1A
Multipole Modeling Using Theoretical Structure Factors. Theoretical structure factors obtained from the CRYSTAL09 single point calculations were used in the multipole refinements using the XD software package. Molecular geometry and the atomic thermal displacement parameters for all atoms were kept fixed throughout the multipole refinement of the static model. Refinements and analysis of the theoretically obtained charge density data were performed using the same methodology as used for the experimental charge density modeling.
’ RESULTS AND DISCUSSION A high resolution X-ray data was collected for 4-fluorobenzoic acid (1) at 100 K (Table 1). The experimental charge distribution for compound 1 was modeled using the Hansen and Coppens multipolar formalism (see Experimental Section for refinement details).20 The Hirshfeld rigid bond test28 was applied to all covalent bonds involving nonhydrogen atoms. The largest difference in the mean-square displacement amplitudes (DMSDA) values, 4(2) 104 Å2, was observed for the C(1)C(2) bond. The minimum and maximum residual electron density peaks observed in the difference Fourier maps obtained after multipole refinements (with I > 3σ (I)) of 0.185 and 0.175 eÅ3 were 12856
dx.doi.org/10.1021/jp2039866 |J. Phys. Chem. A 2011, 115, 12852–12863
The Journal of Physical Chemistry A
ARTICLE
Table 4. Topological Analysis of Charge Density Obtained from Experimental X-ray data and CRYSTAL09 Calculations for the Intermolecular Regions of Compound 1 interaction
O1H1 3 3 3 O2
C6H6 3 3 3 O1
C5H5 3 3 3 O2
C3H3 3 3 3 F1(a)
C3H3 3 3 3 F1(b)
Rij
Fbcp
r2Fbcp
d1
d2
experimental
1.6507
0.293
3.162
1.1204
0.5303
2.23
2.20
7.60
0.01
B3LYP/6-31G(d,p) B3LYP/TZPV-DFT
1.6510 1.6508
0.330 0.329
2.850 3.394
1.1035 1.0919
0.5475 0.5590
2.54 2.40
2.52 2.38
7.91 8.17
0.01 0.01
method
λ1
λ2
λ3
ε
Experimental
2.3941
0.037
0.777
1.4911
0.9030
0.18
0.16
1.12
0.13
B3LYP/6-31G(d,p)
2.3985
0.052
0.658
1.4540
0.9445
0.25
0.23
1.13
0.06
B3LYP/TZPV-DFT
2.3949
0.050
0.723
1.4476
0.9473
0.23
0.21
1.17
0.07
experimental
2.4846
0.041
0.510
1.5107
0.9739
0.21
0.18
0.90
0.15
B3LYP/6-31G(d,p)
2.4696
0.043
0.601
1.4887
0.9809
0.20
0.18
0.98
0.12
B3LYP/TZPV-DFT
2.4602
0.042
0.689
1.4756
0.9845
0.18
0.17
1.04
0.11
experimental B3LYP/6-31G(d,p)
2.5203 2.5170
0.020 0.031
0.593 0.503
1.5190 1.4805
1.0012 1.0365
0.09 0.14
0.08 0.12
0.76 0.76
0.11 0.17
B3LYP/TZPV-DFT
2.5069
0.028
0.563
1.4742
1.0328
0.12
0.11
0.79
0.14
experimental
2.6389
0.019
0.486
1.5340
1.1049
0.08
0.08
0.64
0.02
B3LYP/6-31G(d,p)
2.6088
0.020
0.457
1.5282
1.0806
0.09
0.06
0.61
0.53
B3LYP/TZPV-DFT
2.6070
0.022
0.481
1.5085
1.0985
0.09
0.07
0.65
0.20
found to be within acceptable limits (Figure 1). The static deformation density and Laplacian maps represent the essential chemical features such as the aspherical nature of the electron density around the F-atom and the lone pairs of the O-atoms. A deformation density map shows poor accumulation of electron density (
