Substituent Effects on the [N-I-N](+) Halogen Bond.

We have investigated the influence of electron density on the three-center [N-I-N](+) halogen bond. A series of [bis(pyridine)iodine](+) and [1,2-bis((pyridine-2-ylethynyl)benzene)iodine](+) BF4(-) complexes substituted with electron withdrawing and donating functionalities in the para-position of their pyridine nitrogen were synthesized and studied by spectroscopic and computational methods. The systematic change of electron density of the pyridine nitrogens upon alteration of the para-substituent (NO2, CF3, H, F, Me, OMe, NMe2) was confirmed by (15)N NMR and by computation of the natural atomic population and the π electron population of the nitrogen atoms. Formation of the [N-I-N](+) halogen bond resulted in >100 ppm (15)N NMR coordination shifts. Substituent effects on the (15)N NMR chemical shift are governed by the π population rather than the total electron population at the nitrogens. Isotopic perturbation of equilibrium NMR studies along with computation on the DFT level indicate that all studied systems possess static, symmetric [N-I-N](+) halogen bonds, independent of their electron density. This was further confirmed by single crystal X-ray diffraction data of 4-substituted [bis(pyridine)iodine](+) complexes. An increased electron density of the halogen bond acceptor stabilizes the [N···I···N](+) bond, whereas electron deficiency reduces the stability of the complexes, as demonstrated by UV-kinetics and computation. In contrast, the N-I bond length is virtually unaffected by changes of the electron density. The understanding of electronic effects on the [N-X-N](+) halogen bond is expected to provide a useful handle for the modulation of the reactivity of [bis(pyridine)halogen](+)-type synthetic reagents.

Introduction

A halogen bond (XB) is the close to linear, noncovalent interaction between an electron poor region of a halogen and an electron donor.[1] The chemical nature of the interaction was first rationalized by introduction of the σ-hole concept,[2] followed by description of the impact of charge-transfer, dispersion, polarization, and electrostatic forces.[3−9] Over the past decade, halogen bonds have been increasingly applied in a variety of research disciplines including material sciences, crystal engineering, structural biology, anion recognition and transportation, medicinal chemistry, organic synthesis, and organocatalysis.[6,10−13] Halogen bonding was recently utilized in the development of ionic liquid crystals,[14] in multidentate anion receptors,[15−18] and in the enantioselective recognition of chiral anions.[19] It was demonstrated to be applicable as noncovalent activator in organocatalysis,[20] and as a new tool in drug discovery.[11,21] It found application in the development of luminescent[22,23] and phosphorescent[24] optoelectronics, and of functional materials.[25,26] Halogen bonding has the potential to become a chemical tool as widely applicable as hydrogen bonding.

[Bis(pyridine)halogen]+ complexes and their substituted analogues encompass a reactive halogen(I) species that is stabilized by two electron donors in an [N–X–N]+ halogen bond. Both bromine(I) and iodine(I) containing complexes are useful reagents in organic synthesis, for example for electrophilic halogenation of alkenes, alkynes and aromatics, and for oxidations.[27−41] In three-center-four-electron [N–X–N]+ halogen bonds,[42,43] the halogen is stabilized by simultaneous coordination to two nitrogen electron donors. The three atoms of the [N–X–N]+ system may either form a static, symmetric geometry, [N···X···N]+, or two asymmetric structures in a rapid equilibrium, [N–X···N]+ ⇄ [N···X–N]+,[44,45] the latter being analogous to the tautomeric exchange typical of related [N–H–N]+ complexes.[46] In the symmetric [N···X···N]+ geometry, the halogen forms two equally long and strong N···X bonds, and its motion between the two nitrogens is described by a single-well energy potential. In contrast, in the asymmetric, dynamic system, in each asymmetric isomer the halogen forms one shorter and stronger covalent N–X bond, and one longer and weaker N···X halogen bond. As the halogen jumps between the two nitrogens, its motion is described by a double-well energy potential.[42,45] The centrosymmetric [N–X–N]+ halogen bonds are particularly strong,[42,43,45,47] their strength being similar to that of the halogen bond of the isoelectronic [I–I–I]−.[48] We have previously shown that the [N–I–N]+ halogen bonds of [bis(pyridine)iodine]+ (1a, Figure ) and [1,2-bis((pyridine-2-ylethynyl)benzene)iodine]+ (2a) complexes are exceptionally strong, and that their geometry is static symmetric in solution,[45,49] independent of the nature of the counterion present.[50]

Figure 1

[Bis(4-R-pyridine)iodine]+ BF4– (1a–g), and geometrically restrained [1,2-bis((4-R-pyridine-2-ylethynyl)benzene)iodine]+ BF4– (2a–g) were used as model systems for evaluation of the influence of electron density alteration on the [N–I–N]+ halogen bond. A systematic alteration of the electron density of the [N–I–N]+ halogen bond was achieved by variation of the R-substituents, resulting in complexes with an increasing electron density in the order NO2 < CF3 < H < F < Me < OMe < NMe2. Complex 1 permits free rotation and adjustment of N–I distances for the most favorable interaction, whereas the 1,2-diethynylbenzene backbone of 2 inhibits rotation around the N–I–N axis and imposes some strain in the N–I bonds to reach a geometrically optimal [N–I–N]+ interaction. A mixture of 2a–d and their monodeuterated isotopologs 2a–d- were used in IPE NMR experiments for determining the geometry of their [N–I–N]+ halogen bonds.

The influence of electron density alteration on the strength and geometry of the three-center-four-electron [N–X–N]+ halogen bond has so far not been assessed. A decreased electron donating ability of the nitrogens, as induced by electron withdrawing substituents, can be expected to destabilize the symmetric [N–I–N]+ halogen bond, and increase the reactivity of the complex. If the central iodine(I) is unable to efficiently accept electrons into its empty p-orbital from two electron poor nitrogens simultaneously, it may prefer to compensate its electron depletion by forming a covalent bond to one of the nitrogens, instead of forming halogen bonds to both. Hence, electron depletion might induce an asymmetric [N–I···N]+ geometry. In contrast, an increased electron density, induced by electron donating substituents, is expected to stabilize the symmetric [N···I···N]+ geometry by strengthening the two N···I halogen bonds. This is expected to decrease the reactivity of the complex.

In this study, we report our findings, from systematic solution NMR, X-ray diffraction, reaction kinetics, and computational investigations on the influence of electron density alterations on [N–I–N]+ halogen bond strength and geometry. An understanding of electronic effects is, above all, expected to provide a useful handle for the modulation of the reactivity of [bis(pyridine)halogen]+-type reagents.

Results and Discussion

[Bis(pyridine)halogen]+ complexes and their geometrically restrained [1,2-bis((pyridin-2-ylethynyl)benzene)halogen]+ analogues are the model systems typically used to gain understanding of the [N–X–N]+ halogen bond.[42,43,45,49−51] In this study, we introduce substituents of varying electronic properties in the para-positions relative to the pyridine nitrogens (Figure ), to study the influence of electron density alteration on the strength and geometry of their [N–I–N]+ halogen bond.

Synthesis

The 4-substituted electron rich [bis(pyridine)iodine]+ BF4– complexes 1b (4-Me), 1c (4-OMe), and 1d (4-NMe2) as well as their electron poor analogue 1f (4-CF3), (Figure ) were synthesized from their corresponding Ag+ complexes by addition of molecular I2, following our previously reported protocol[50] toward 1a. Complex 1b (4-Me) rapidly decomposed upon isolation.[52]

The 4-substituted [1,2-bis((pyridine-2-ylethynyl)benzene)iodine]+ BF4– complexes 2a (4-H), 2b (4-Me), 2c (4-OMe), 2d (4-NMe2), 2e (4-F), 2f (4-CF3), and 2g (4-NO2), and their selectively monodeuterated, electron rich analogues 2a–d- were prepared from their corresponding 1,2-bis(pyridine-2-ylethynyl)benzene ligands 3a to 3g, and 3b- to 3d-, via their corresponding Ag+ complexes,[50] as shown in Scheme . The electron poor [N–I–N]+ complexes 2e (4-F) and 2f (4-CF3) were observed by 1H NMR to be significantly less stable in CD2Cl2 solution as compared to their electron rich analogues. Consequently, 2e and 2f could only be detected in a mixture with their decomposition product, the analogous [N–H–N]+ complex. Our attempts to generate the electron poor complex 2g (4-NO2) yielded the corresponding [N–H–N]+ complex only, due to the rapid decomposition of the targeted [N–I–N]+ complex in solution.

Scheme 1

General Synthetic Route for the Synthesis of 4-Substituted [1,2-Bis((pyridine-2-ylethynyl)benzene)iodine]+ BF4– Complexes 2a (4-H), 2b (4-Me), 2c (4-OMe), 2d (4-NMe2), 2e (4-F), 2f (4-CF3), and 2g (4-NO2) and Their Selectively Monodeuterated Analogues 2a-d,g-

The 4-substituted 1,2-bis(pyridine-2-ylethynyl)benzene ligands 3a–c,e,f were synthesized using microwave-assisted Sonogashira coupling (Scheme ),[53] following previously optimized conditions.[45] Compounds 3a (4-H),[45]3b (4-Me),[54] and 3c (4-OMe) were prepared from 1,2-diiodobenzene and two equivalents of terminal alkyne 4a–c. Ligands 3e (4-F) and 3f (4-CF3) were obtained from 1,2-diethynylbenzene and 2-bromo-4-fluoropyridine (7e) or 2-chloro-4-trifluoromethylpy-ridine (7f).[54] The electron poor ligand 3g (4-NO2) was prepared from 1,2-diiodobenzene and 2 equiv of 4g using Sonogashira coupling with Et3N as base, instead of Et2NH, and without microwave heating to avoid side reactions. The most electron rich ligand 3d (4-NMe2) was prepared from 3g (Scheme ) via Béchamp reduction,[55,56] followed by reductive amination of paraformaldehyde with sodium cyanoborohydride.[57]

Scheme 2

Reagents and Conditions: (a) 4g (2.2 equiv), Pd(PPh3)2Cl2, CuI, Et3N, 75 °C, 5 h, then rt, 18 h; (b) Fe, AcOH, 55 °C, 1.5 h; (c) CH2O, NaBH3CN, AcOH, rt, 24 h; (d) AgBF4, CH2Cl2, rt, 20 min; (e) I2, CH2Cl2, rt, 30 min; (f) 4g (0.8 equiv), Pd(PPh3)2Cl2, CuI, Et3N, 75 °C, 3 h; (g) 4g- (2.1 equiv), Pd(PPh3)2Cl2, CuI, Et3N, 75 °C, 20 h.

The monodeuterated ligands 3a- to 3c- were prepared in accordance with our reported procedure,[45] including two consecutive microwave-assisted Sonogashira couplings; coupling of 1,2-diiodobenzene with one equivalent of 4a–c, generating 5a–c, was followed by coupling of the corresponding deuterated terminal alkyne 4a- to 4c- (Scheme ). The 4-NO2-substituted analogue 3g- was prepared similarly from 5g and 4g-, and was then converted to the 4-NMe2-subsituted 3d- via Béchamp reduction and subsequent reductive amination, as described above (Scheme ) for the conversion of 3g to 3d. The terminal alkynes 4a- and 4b- were synthesized following our published procedure[45] via regioselective deuteration at C6 of either 2-chloropyridine (7a) or 2-chloro-4-methylpyridine (7b) using BuLi-LiDMAE, subsequent microwave mediated Sonogashira coupling of TMS-acetylene, and TMS-deprotection with KF in MeOH (Scheme ).

The synthetic routes to the 4-MeO- and 4-NO2-substituted terminal alkynes 4c and 4g (Scheme ) both started from the commercially available 2-bromo-4-nitropyridine (7g), whereas those of their selectively monodeuterated analogues 4c- and 4g- started from 2-bromo-4-nitropyridine-6-d (7g-, Scheme ). For selective monolithiation of 2,6-dibromopyridine at C-6 with n-BuLi, a method developed by Cai and co-workers was followed.[58] Successive electrophilic trapping with MeOD furnished 2-bromo-6-deuteropyridine (8-) in nearly quantitative yield. For introducing the 4-NO2-group, 8- was first converted by mCPBA oxidation[59] into its N-oxide 9-, which was then nitrated with H2SO4/HNO3 to provide 10-. Subsequent reduction of 10- with PBr3 generated the desired starting material 7g- in high yield.[60] Sonogashira coupling of either 7g or 7g- with TMS-acetylene at room temperature yielded the TMS-protected alkyne 6g or 6g-, following the procedure described by Sagitullina et al.[61] In subsequent TMS-deprotection, the 4-MeO- and 4-NO2-substituted terminal alkynes 4c/4c- and 4g/4g- were selectively generated by the choice of reaction conditions. Deprotection with KF in MeOH at ambient temperature furnished 4-NO2-substituted 4g/4g-, whereas basic conditions using K2CO3 in THF/MeOH (1:1) at ambient temperature provided 4-MeO-substituted 4c/4c- as a result of simultaneous TMS-removal and nucleophilic replacement of the 4-NO2-substituent. Further details regarding the syntheses described above are given in the Supporting Information.

Scheme 3

Reagents and Conditions: (a) 1. nBuLi (1 equiv), THF, −78 °C; 2. MeOD (5 equiv), −78 °C, 1 h, then rt, 24 h; (b) mCPBA, CHCl3, rt, 24 h; (c) HNO3/H2SO4 2:1, 110 °C, 3 h; (d) PBr3, CH2Cl2, 60 °C 3 h; (e) TMS-acetylene, Pd(PPh3)2Cl2, CuI, Et3N, 5 °C, 1 h, then rt, 2 h; (f) KF, MeOH, rt, 2 h; (g) K2CO3, THF/MeOH 1:1, rt, 1 h.

Alteration of Electron Density

A common approach to increase XB strength is to enhance the electrophilicity of the halogen bond donor using strongly electron withdrawing substituents.[62] Accordingly, perfluoroaryl and -alkyl halides are frequently applied halogen bond donors in model studies. The alternative approach is to increase the electron density of the halogen bond acceptor by the introduction of electron donating substituents.[63] A stronger halogen bond is often presumed to be reflected by a shorter distance between the halogen bond donor and the acceptor.[48,64,65] Computational studies revealed that the Hammett substituent constants of pyridine-based halogen bond acceptors correlate to the strength of their halogen bond.[66,67] Strategic tuning of halogen bond strength is expected to facilitate the development of halogen transfer reagents and of supramolecular systems,[12,20] for example.

The Hammett substituent constants, and the corresponding induction and resonance constants F and R, are reported for benzene derivatives and are not calibrated for pyridine rings. To assess how well they describe electron donation/withdrawal in pyridine rings, we calculated the natural atomic population (NAP[68]) n(N) as well as the π electron population nπ(N) for the nitrogen atoms of 1a–g and 2a–g. The calculations were performed with the computational protocol described below for the investigation of equilibrium geometries and bond energies. The changes of these populations, Δn(N) and Δnπ(N), relative to 1a and 2a, respectively, are listed in Table , together with the corresponding Hammett and resonance constants. Regression coefficients of 0.969 (1a–g) and 0.973 (2a–g) between σpara and Δn(N), and 0.965 (1a–g) and 0.975 (2a–g) between R and Δnπ(N) are found. Thus, the correlations are indeed distinct, but not perfect. We therefore used Δn(N) and Δnπ(N) directly to quantify the electron withdrawing/donating properties of substituents.

Table 1

Calculated Changes Upon Substitution for Natural Atomic Populations Δn(N) and π Orbital Populations Δnπ(N) for the N atoms of Iodine(I) Complexes 1a–g and 2a–g, versus Hammett and Resonance Substituent Constants

structure	4-R	σpara	103 × Δn(N)	R	103 × Δnπ(N)
1a	H	0	0	0	0
1b	Me	–0.17	3.0	–0.18	12.7
1c	OMe	–0.27	5.9	–0.56	42.6
1d	NMe2	–0.83	18.5	–0.98	73.2
1e	F	0.06	–7.9	–0.39	16.7
1f	CF3	0.54	–17.7	0.16	–15.5
1g	NO2	0.78	–30.7	0.13	–25.9
2a	H	0	0	0	0
2b	Me	–0.17	2.7	–0.18	11.6
2c	OMe	–0.27	5.9	–0.56	39.8
2d	NMe2	–0.83	17.1	–0.98	69.2
2e	F	0.06	–6.3	–0.39	16.9
2f	CF3	0.54	–15.1	0.16	–11.4
2g	NO2	0.78	–26.7	0.13	–19.1

The 15N NMR chemical shift, δ15N, is recognized to be a sensitive tool for the assessment of the coordinative bonds of nitrogen heterocycles.[54,69] It has previously been used for the characterization of halogen bonded complexes.[45,49,50,70,71] Upon formation of pyridine-based [N···I···N]+ complexes, analogous to nonsubstituted 1a and 2a, ∼100 ppm 15N chemical shift changes, δ15Ncoord, are induced.[45] This chemical shift alteration is the consequence of the establishment of a strong halogen bond, in which the pyridine nitrogen acts as Lewis base, and the introduction of a positive charge into the molecular system. The bond strength of nitrogen heterocycles is commonly assessed by the 15N NMR coordination shift, δ15Ncoord. It is defined as the 15N chemical shift difference of a complex and of its corresponding free nitrogen base (ligand), i.e., δ15Ncoord = δ15Ncomplex – δ15Nligand. A stronger coordinative bond is commonly seen to be reflected by a larger |δ15Ncoord|.[45,49] To evaluate the influence of substituents on the electron density of the pyridine nitrogen involved in the formation of [N···I···N]+ complexes, we acquired the 15N NMR chemical shift for 1a–d,f and 2a–f (Figure ), δ15Ncomplex, and for their corresponding free nitrogen bases (4-substituted pyridines and 3a–f), δ15Nligand. The δ15N values were detected for CD2Cl2 solutions using indirect detection by 1H,15N HMBC experiments. A single 15N NMR signal was observed for each complex (Table ). Electron donating substituents in the para-position of the pyridine nitrogen increase the shielding of the nitrogen, whereas electron withdrawing substituents have an opposite effect. The influence of the substituents on the δ15N is comparable for the ligands and complexes applied in this study. We determined the regression coefficients r2 for the correlation of δ15N with both Δn(N) and Δnπ(N), as well as σpara and R, both for the series of complexes 1a–d,f and 2a–f and for the corresponding series of ligands (see SI for the complete results). For the correlation with Δnπ(N), the r2 values are in the interval between 0.984 (ligands for 1a–d,f) and 0.995 (complexes 1a–d,f) (Figure ), whereas for the correlation with Δn(N) considerably lower r2 values, in the interval between 0.824 (complexes 2a–f) and 0.920 (ligands 3a–f), were calculated. Likewise, the regression coefficients with respect to σpara and R were lower than the r2 with respect to Δnπ(N). This finding indicates that the substituent effects on the δ15N are governed by the π population rather than the total electron population at the nitrogen atoms. Thus, the δ15N is primarily controlled by the paramagnetic ring currents of the pyridine rings rather than by diamagnetic shielding or deshielding. Consequently, the δ15N change is a specific measure for the π donating or withdrawing property of substituents. In particular, we find that fluorine, in contrast to common beliefs, dominantly acts as an electron donor, via resonance, which is reflected by the comparable δ15N of the fluorine substituted 2e and 3e to those of the methyl substituted analogues 2b and 3b (Table ). This observation is corroborated by the literature.[72] The δ15Ncoord of the iodine(I) complexes do not vary to a significant extent, which is in excellent agreement with previous observations of δ15Ncoord for analogous silver(I) complexes, when detected in a noncoordinating solvent and in the absence of a strongly coordinating counterion.[50,54] Overall, the observed large δ15Ncoord of 1a–d,f and 2a–f confirms formation of [N–I–N]+ complexes; however, its magnitude does not reflect interaction strength.

Table 2

15N NMR Chemical Shift (ppm) of Iodine(I) Complexes 1a–d,f and 2a–f, and of the Corresponding Nitrogen Basesa

structure	4-R	δ 15Ncomplex	δ 15Nligand	δ 15Ncoord
1a	H	–175.1	–67.0	–108.1
1b	Me	–180.2	–71.6	–108.6
1c	OMe	–195.0	–86.0	–109.0
1d	NMe2	–214.2	–104.8	–109.4
1f	CF3	–164.1	–51.6	–112.5
2a	H	–165.5	–64.5	–101.0
2b	Me	–170.2	–69.2	–101.0
2c	OMe	–183.5	–83.5	–100.0
2d	NMe2	–202.9	–102.2	–100.7
2e	F	–170.9	–69.5	–101.4
2f	CF3	–156.7	–50.7	–106.0

The 15N NMR coordination shifts represent the chemical shift change upon complex formation.

Figure 2

15N NMR chemical shift changes (relative to 1a or 2a, respectively), Δδ15N, of substituted pyridines (●), of the corresponding 1,2-bis((pyridine-2-ylethynyl)benzene) ligands 3a–f (○), and of their [N–I–N]+ complexes, 1a–d,f (▼) and 2a–f (Δ), respectively, strongly correlate to the corresponding changes in the π(N) orbital population. This indicates that the variation in δ15N is dominatingly governed by the paramagnetic ring currents in the pyridine moieties.

Halogen Bond Symmetry in Solution

Understanding the factors governing the geometry of the three-center-four-electron [N–X–N]+ halogen bond is a key for the description of its nature.[42,43,45,51] The solution NMR spectroscopic technique isotopic perturbation of equilibrium (IPE) has previously been proven to be a unique tool for distinguishing a rapidly equilibrating mixture from a static, symmetric structure in solution, and was successfully applied for symmetry determinations of hydrogen and halogen bonds, of carbocations, and of metal chelates.[45,46,73,74] IPE relies on the detection of vibrational energy changes in a molecule upon selective isotope labeling, commonly a hydrogen to deuterium substitution close to the interaction site of interest. When analyzing the mixture of the labeled and the unlabeled isotopologs of a compound, two sets of signals are detected; one set originates from the nondeuterated compound, whereas the other set stems from the deuterated analogue. The small shift difference between the signals of the isotopologs is called the isotope shift, Δobs, where n denotes the number of bonds between the observed nucleus, commonly 13C, and the position of the isotopic substitution. The observed chemical shift difference is the sum of the intrinsic isotope shift, Δ0, and the equilibrium isotope shift, Δeq, according to eq :

The intrinsic isotope effect, Δ0, which is present in all systems, is usually small and attenuates as n increases. The 1Δ0 and 2Δ0 isotope shifts, observed for the reporter atoms one and two bonds from the isotopic substitution site, are negative, often in the order of −0.3 ppm and −0.1 ppm, whereas the sign of Δ0 observed for reporter atoms further away may vary. The equilibrium isotope effect, Δeq, in contrast, is present only for systems involved in a dynamic equilibrium process, and is zero for static systems. Its magnitude depends on the equilibrium constant K of the exchange process:where D denotes the chemical shift difference between the signals of the isomeric forms, i.e., halogenated N+–X and nonhalogenated N, in this particular case. The equilibrium constant, K, is temperature dependent according to the van’t Hoff equation,[75] and so is the equilibrium isotope effect, Δeq. We have shown that the magnitude of the observed isotope shift, Δobs, does not allow straightforward differentiation between a static [N···X···N]+ geometry and a dynamic [N–X···N]+⇄[N···X–N]+ system in rapid equilibrium, but the temperature dependence of Δobs is a reliable measure for distinguishing the two systems from each other.[45] The symmetry of the [N–I–N]+ halogen bond of 2a–d was studied in solution with the IPE technique using isotopolog mixtures and 13C {1H,2H} NMR detection, as described in detail earlier.[45] Similar to previous studies of halogen bond symmetry,[45,49,50] isotopolog mixtures of the free ligands 3a–d were used as references for static systems in the evaluation of the temperature dependence of the isotope effects (Table ). Due to their low stability, in combination with their poor solubility, the symmetry of the electron poor complexes 3e–g could not be studied experimentally.

Table 3

Temperature Coefficients (ppm × K) of the 13C Isotope Shifts of Complexes 2a–d, and the Corresponding References 3a–d for a Static Geometry

structure	4-R	C2 1Δobs	C3 2Δobs	C4 3Δobs	C5 4Δobs	C6 3Δobs	Σ |Δobs|
2aa,c	H	–8.9	–10.8	+0.7	0	–2.0	22.4
3aa,d	H	–8.1	–9.1	–1.5	+3.4	–4.5	26.6
2ba	Me	–11.2	–11.3	+1.3	0	–2.5	26.3
3bb	Me	–8.8	–8.6	n.d.e	+3.0	n.d.e	(20.4)
2cb	OMe	–5.4	–11.5	+2.3	0	–3.1	22.3
3cb	OMe	–5.9	–8.9	+2.3	0	0	17.1
2db	NMe2	–5.8	–10.2	+3.2	0	n.d.e	(19.2)
3db	NMe2	–4.1c	–9.2	+0.9	+1.1	–4.9	16.1

The 13C{1H,2H} experiments were run at 125.71 MHz.

The 13C{1H,2H} experiments were run at 201.20 MHz.

Temperature interval −20 to 25 °C.

Temperature interval −10 to 25 °C.

Due to minor temperature dependence and to limited solubility, this coefficient could not be reliably determined.

The temperature coefficients Δobs for complexes 2a–d (Table ) are small and comparable to those of the corresponding reference molecules 3a–d, in both magnitude and sign. The sum of the absolute value of the observed isotope effects, Σ|Δobs|, is comparable for the free ligands and their corresponding [N–I–N]+ halogen bonded complexes. The overall temperature dependence of the isotope effect of systems involved in rapid equilibrium, such as [N–H–N]+ complexes, were previously reported to be significantly higher than those of their corresponding free ligands.[45] In agreement with previous studies,[45,49,50,76] the 2Δobs’s of C3 were largest for both the free ligands 3a–d and their iodine(I) complexes 2a–d. A static, symmetric geometry of 2a–d, independent of the electron density of their [N···I···N]+ bond, was further indicated by the relative magnitude of the temperature coefficients. Thus, a larger temperature dependence was detected for the carbons closest to the position of 1H-to-2H substitution, 1Δobs and 2Δobs. For dynamic systems, such as the corresponding [N–H–N]+ complexes, the magnitude of the temperature coefficients depends not only on the distance of the reporter 13C’s to the deuterium substitution site (n), but also on their distance to the nitrogen that is directly involved in the fast equilibrium process.[45] Consequently, the temperature coefficients of the isotope effects of carbons further away from the point of isotope substitution, yet close to the nitrogen are large for dynamic, but small for static systems.

For a more detailed analysis of the halogen bonds and their symmetry, we calculated equilibrium geometries and energies of 1a–g and 2a–g. All calculations were performed using density functional theory (DFT), employing the B3LYP exchange and correlation functional.[77−79] The LANL08 basis set[80] in conjunction with LANL2DZ effective core potential[81−83] were used for I, whereas Pople’s 6-311+G(d,p)[84−86] basis set was used for N, and Pople’s 6-311G(d,p)[84,85] basis set for the remaining atoms. Solvent effects were accounted for by the Polarizable Continuum Model (PCM)[87,88] with dichloromethane as solvent. Calculations predict symmetric equilibrium geometries for compounds 1a–g and 2a–g (Table ), with N–I distances that are 18% (∼0.41 Å) longer than that of the corresponding N–I covalent bond (2.077–2.094 Å, Table S31, Supporting Information), and considerably shorter than the sum of the van der Waals radii of the participating atoms (3.53 Å, RXB= 0.65[50,64,89]). The bond lengths for the geometrically restrained (2) and the corresponding nonrestrained analogues (1) differ by less than 0.01 Å, indicating that the strain energy required to distort the 1,2-ethynylbenzene backbone is small as compared to the energetic gain upon forming a symmetric three-center [N–I–N]+ halogen bond.[45] Upon formation of the [N–I–N]+ complexes 2a–f, the N–N distance decreases by 0.07 to 0.2 Å, whereas for 2g, it slightly increases by 0.007 Å. The N–I–N angle is linear (180°) for the [bis(pyridine)iodine]+ complexes 1a–g, and is approximately 175.5° for the [1,2-bis((pyridine-2-ylethynyl)benzene)iodine]+ complexes 2a–g, in excellent agreement with previous reports on the geometry of halogen bonds.[6,10,42,48] The N–I bond length shows only minor dependence on the electronic properties of the substituent, with a less than 0.02 Å difference observed between the shortest and longest bonds. The N–I bond length correlates with the π electron population, Δnπ(N), with r2 of 0.953 (1a–g) and 0.935 (2a–g), (Figure ). Significantly lower correlation was found between r(N–I) and the total electron population Δn(N), as reflected by the r2 0.800 for 1a–g, and 0.700 for 2a–g. This indicates that the slight variation in the N–I bond length of the studied complexes is governed by the π population at the nitrogen atoms. To assess the stability of complexes 1a–g and 2a–g, we calculated their stabilization energies according to the formal reactions shown in Scheme . Since we are interested in the strength of the electronic bond rather than wish to make a comparison to experimental thermochemical data, the electronic energies are considered here and not the Gibbs free energies. Whereas the geometry, i.e., the bond lengths and bond angles, of the complexes is virtually independent of electron density, their stability varies by nearly 100 kJ/mol, depending on the electronic nature of the substituents (Table , Figure ). An increased electron density stabilizes the [N···I···N]+ bond (1b–d and 2b–d), whereas electron deficiency reduces the stability of the complexes (1e–g and 2e–g). The correlation coefficients between the stabilization energy, ΔEstab, and the natural atomic population, Δn(N), are 0.991 for 1a–g and 0.989 for 2a–g, whereas the corresponding r2 values for the π electron population, Δnπ(N), are significantly lower, 0.851 for 1a–g, and 0.828 for 2a–g. Thus, whereas the N–I bond length proved to correlate strongly with Δnπ(N), the stabilization energy correlates nearly perfectly with the total natural atomic population (NAP) at the nitrogen atoms. This can be rationalized from the mechanism of the three-center-four-electron [N···I···N]+ bond: the larger the natural atomic population, Δn(N), the higher the energies of the occupied orbitals at the nitrogen, which facilitates the charge transfer to iodine(I). This is in line with the findings of Ebrahimi et al., who investigated the nature of the [N···X···N]+ bond with a variety of theoretical methods.[90] The computationally predicted variation of the [N···I···N]+ bond strength agrees excellently with the experimentally observed higher stability of electron rich complexes 2b–d as compared to the unsubstituted 2a, and the instability of 2e in solution that prevented their IPE study. It should be noted that the rapid decomposition of 2g prevented its experimental investigation completely, in line with its computationally predicted instability.

Table 4

Computationally Predicted N–I and N–N Distances, N–I–N Angles, and Stabilization Energies for Complexes 1a–g and 2a–g and N–N Distances for the Ligands 3a–ga

structure	4-R	r(NI)b (Å)	r(NN)c (Å)	r(NN)d (Å)	∠(NIN) (deg)	ΔEstab (kJ/mol)
1a	H	2.3036	4.6072		180	0
1b	Me	2.3013	4.6026		180	12.7
1c	OMe	2.2980	4.5960		180	42.6
1d	NMe2	2.2921	4.5841		180	73.2
1e	F	2.3032	4.6063		180	16.7
1f	CF3	2.3055	4.6109		180	–15.5
1g	NO2	2.3065	4.6131		180	–25.9
2a	H	2.3034	4.5934	4.6854	175.7	0
2b	Me	2.3011	4.5875	4.7166	175.4	11.6
2c	OMe	2.2982	4.5832	4.6586	175.7	39.8
2d	NMe2	2.2930	4.5715	4.7700	175.4	69.2
2e	F	2.3027	4.5921	4.6756	175.7	16.9
2f	CF3	2.3041	4.5939	4.6624	175.5	–11.4
2g	NO2	2.3043	4.5952	4.5880	175.7	–19.1

All calculations were done for CH2Cl2 solution with the computational protocol described below.

For all compounds, r(NI)1 = r(NI)2.

[N–I–N]+ complexes 1a–g and 2a–g.

Ligands 3a–g.

Figure 3

Correlation of the N–I bond distance with the change in the π(N) population of iodine(I) complexes of substituted pyridines (1a–g) (●) and of the corresponding 1,2-bis((pyridine-2-ylethynyl)benzene) ligands (2a–g) (○).

Scheme 4

Formal Reactions Used to Define the Stabilization Energies of Complexes 1a–g and 2a–g

Figure 4

Correlation of the stabilization energy (ΔEstab) and the change of the natural atom population at the N atoms of the iodine(I) complexes of substituted pyridines (1a–g) (○) and of the corresponding 1,2-bis((pyridine-2-ylethynyl)benzene) ligands (2a–g) (●).

For all compouical">nds, r(NI)1 = r(NI)2.

We emphasize that the correlations found above must not be misinterpreted as causal relationships. For instance, the close correlation between Δn(N) and ΔEstab does not indicate that the halogen bond strength is governed solely by electrostatic interactions between the partial charges of the N and I atoms. Rather, as has been discussed, e.g., in ref (10), halogen bonding is caused by an interplay between electrostatic, charge-transfer and not the least polarization and dispersion effects. The Δn(N) and Δnπ(N) values are to be considered just as indicators for the properties of the substituents, in lieu of properly calibrated Hammett, induction, and resonance constants.

Geometry in the Solid State

Single crystals were obtained via slow diffusion of hexane into dichloroethane solutions of 1c and 1f under cooling from 24 °C to −20 °C over 2 days. Crystallographic data for 1a and 1d were taken from the literature.[91,92] The X-ray crystallographic analysis confirmed that the N–I–N angle is linear (Figure ) and that the iodine-centered complexes are symmetric, even in the solid state (Table ), independent of their electron density. The small difference between the N–I bond lengths within a complex, lower than <0.4%, is likely due to crystal packing forces, and does not reflect a real asymmetry. Supporting the DFT predictions above, the X-ray crystallographic data corroborate that alteration of the electron density of the pyridine nitrogen causes only very minor, <2%, change in the N–I bond length.

Figure 5

Solid state geometries of complexes 1c (top, CCDC-1452897), and 1f (bottom, CCDC-1452897), obtained by single crystal X-ray crystallography. The BF4– counterion is omitted from the figure for clarity. The crystal of compound 1c was obtained as a solvate, and thus each molecular unit contains one molecule dichloroethane. Both complexes possess coplanar pyridine rings and nearly centrosymmetric geometries.

Table 5

X-ray Crystallographically Determined N–I Bond Distances and N–I–N Bond Angles for 1a,c,d,f

structure	4-R	r(N–X)1 (Å)	r(N–X)2 (Å)	σ (N–X–N) (deg)
1a(91)	H	2.260(3)	2.260(3)	180.0
		2.259(3)	2.259(3)	180.0
		2.255(3)	2.260(3)	177.7(1)
1c	OMe	2.252(3)	2.262(3)	178.0(1)
1d(92),a	NMe2	2.232	2.239	179.4
		2.247	2.252	177.7
1f	CF3	2.251(5)	2.272(5)	176.2(2)
		2.256(5)	2.271(5)	175.2(2)

The counterion of 1d is NO3– instead of BF4–. This does not influence the geometry of [bis(pyridine)iodine]+ complexes.

The couical">nterion of 1d is NO3– instead of BF4–. This does not influence the geometry of [bis(pyridine)iodine]+ complexes.

Halogen Bond Strength in Solution Established by UV–vis Kinetics

For experimental evaluation of the electron density dependence of the halogen bond strength of [N–I–N]+ complexes, we have monitored the rate of disappearance of the UV absorbance of [bis(pyridine)iodine)]+ upon iodonium transfer from complexes 1a,c,d,f to 4-penten-1-ol, using a procedure described by Neverov and Brown.[39,93] The mechanistic details of halonium transfer from bis(pyridine)-type complexes to alkenes are well understood.[39,93−97] We have used pseudo-first-order conditions, with a large excess of olefin (3.9–77 mM), and monitored the reaction proceeding in dry dichloroethane solution at 25 °C at the wavelength of maximum change (1a,f: 230 nm, 1c, 257 nm, and 1d, 300 nm). The reactions of 1a,c,d were monitored using a Cary 100 UV–vis spectrophotometer, whereas an Applied Photophysics SX-17MV stopped-flow reaction analyzer was used for the fast reactions with 1f (4-CF3). The observed rate constants (kobs) were obtained by NLLSQ fitting of the absorbance versus time traces for the disappearance of the iodine(I) complex to the standard exponential model:

The saturation profile obtained when plotting kobs of 1a,c,f against the alkene concentration implies that the mechanism for the iodonium transfer is similar to that previously proposed for bromocyclization with 4-penten-1-ol.[39] Saturation was not reached in the corresponding experiment with the electron rich and slowly reacting 1d, suggesting a different rate limiting step. However, in the lower alkene concentration range the kobs versus [4-penten-1-ol] profile was linear permitting determination of the second-order rate constant (k2), even for this reaction. The logarithm of the second-order rate constants (k2 = 9.21 M–1 s–1 (1a, H), k2 = 1.92 M–1 s–1 (1c, OCH3), k2 = 0.00195 M–1 s–1 (1d, N(Me)2), and k2 = 126.9 M–1 s–1 (1f, CF3)), normalized to 1 mM concentration of added 4-substituted pyridine, as a function of the change of natural atomic population, Δn(N), is shown in Figure .

Figure 6

Second-order rate constants (M–1 s–1) of 1a,c,d,f in iodocyclization reactions with 4-penten-1-ol in the presence of 4-R-pyridine (R = H, OMe, NMe2 or CF3), normalized to 1 mM, obtained in dry dichloroethane, are shown as a function of the change of natural atomic population, Δn(N). A linear correlation is seen for all but the most electron rich complex 1d.

The magnitude of the second-order rate constants of complexes 1a,c,d,f follows the inverse order of substituent electronegativity, i.e., k2(1f, CF3) ≫ k2(1a, H) > k2(1c, OCH3) ≫ k2(1d, NMe2). Whereas the log(k2) of complexes 1a,c,f correlates linearly to the change of natural atomic population, Δn(N), which corresponds to the Hammett σpara, the corresponding log(k2) of 1d is an evident outlier (Figure ), corroborating the hypothesis that the iodonium transfer mechanism for 1d is different from that of its less electron rich analogues.

Due to the difficulties in identifying appropriate wavelengths for reaction monitoring, no kinetics experiments were undertaken for the analogous 4-substituted 1,2-bis((pyridine-2-ylethynyl)benzene)iodine(I) complexes 2a–f.

We studied the kinetics of the dissociation of complexes 1a,c,e,f in the presence of the strong XB acceptor 4-dimethylaminopyridine (DMAP), the free base of 1d. This ligand exchange reaction is rapid due to the significantly higher stability of complex 1d as compared to the less electron rich complexes. Moreover, the reaction is easy to follow as 1d has high absorbance. The dissociation rate constants (kobs) at 25 °C (Table ) were determined from stopped-flow experiments by addition of DMAP (0.15 mM; concentration at which rate of the reaction was independent of [DMAP]) to 1a, 1c, or 1e (7.68 × 10–5 M) in dry dichloroethane at 272 nm. We measured the initial rate kinetics for the formation of the DMAP coordinated complex 1d. In addition, the thermodynamic activation parameters ΔH‡ and ΔS‡, shown in Table , for the reactions were determined from the same experiments performed at different temperatures (Table ) by Eyring plots of the obtained kobs rate constants. The obtained activation parameters indicate that the lower the electron density of an [N–I–N]+ complex, the lower the activation barrier for iodine(I) transfer. This observation indicates, in excellent agreement with the DFT computations, that a decrease in electron density of the [N–I–N]+ halogen bond decreases its strength.

Table 6

Dissociation of [Bis(4-R-pyridine)iodine]+ Complexes 1a,c,f (7.68 × 10–5 M) in the Presence of an Excess DMAP (0.15 mM) in Dry Dichloroethane, Observed kobs Dissociation Rate Constants at 298 K, and the Enthalpy and Entropy of Activationa

structure	4-R	kobsb (s–1)	ΔH‡c (kJ mol–1 )	ΔS‡c (J mol–1 K–1)
1a	H	0.392	70.97 ± 1.07	–14.45 ± 3.49
1c	OCH3	0.333	72.83 ± 0.80	–13.23 ± 2.60
1f	CF3	3.322	39.16 ± 1.87	–105.47 ± 6.15

Experiments run under stopped-flow with 7.68 × 10–5 M iodine(I) complex 1a,c,f, and 0.15 mM DMAP in the reaction cell.

Observed rate constants kobs determined at 298 K.

Activation parameters, ΔH‡ and ΔS‡, determined from Eyring plots.

Conclusions

Electron density changes do not disturb the symmetric geometry of the [N···I···N]+ halogen bond of bis(pyridine)-type systems, neither in solution nor in the solid state. Despite having a strong influence on the strength of the halogen bond, the electron density of the pyridine nitrogens does not significantly affect the N–I bond lengths. The slight change (<0.7%) correlates only weakly to the variation in total electron population, Δn(N), but much more strongly to that in the π population Δn(N). This observation is important to stress, as shortening of the halogen bond donor–acceptor distance is one of the most commonly used characteristics for categorizing halogen bond strength.

The 15N NMR chemical shift of halogen bonded complexes is a good measure for the π electron population of the involved nitrogen. The 15N NMR coordination shift of the studied iodine(I) complexes reflects the formation of the [N–I–N]+ halogen bond; however, its magnitude does not directly reflect the strength of the halogen bond. Analogous lack of correlation between bond energy and chemical shift was recently reported for analogous tetrel bonds.[98]

The stability of the [N–I–N]+ halogen bond correlates to the electron density of the nitrogen halogen bond acceptors. A linear correlation to the natural atomic population, Δn(N) of the para-substituted pyridines was seen for most substituents. However, reaction kinetics indicate that the iodine(I) complex bearing the strongly resonance donating N,N-dimethylamino substituent has a different iodonium transfer mechanism as compared to the other analogous complexes studied. Alkene halogenation using the latter 4-NMe2-substituted complex is slow. On the whole, these observations may be helpful for providing a tool for thorough control of electrophilic halogenation reactions. As [bis(pyridine)iodine]+ complexes are common synthetic reagents, this may be of significance for the development of new, stereoselective halogenating agents in the future.