Preorganization: A Powerful Tool in Intermolecular Halogen Bonding in Solution

Abstract

Preorganization is a powerful tool in supramolecular chemistry which has been utilized successfully in intra‐ and intermolecular halogen bonding. In previous work, we had developed a bidentate bis(iodobenzimidazolium)‐based halogen bond donor which featured a central trifluoromethyl substituent. This compound showed a markedly increased catalytic activity compared to unsubstituted bis(iodoimidazolium)‐based Lewis acids, which could be explained either by electronic effects (the electron withdrawal by the fluorinated substituent) or by preorganization (the hindered rotation of the halogen bonding moieties). Herein, we systematically investigate the origin of this increased Lewis acidity via a comparison of the two types of compounds and their respective derivatives with or without the central trifluoromethyl group. Calorimetric measurements of halide complexations indicated that preorganization is the main reason for the higher halogen bonding strength. The performance of the catalysts in a series of benchmark reactions corroborates this finding.

The origin of the strong Lewis acidity of a rotationally locked halogen bond donor is elucidated via a comparison of structurally related compounds with or without a central trifluoromethyl group. Calorimetric measurements of halide complexations and the performances in several organocatalytic reactions indicate that preorganization – and not the electron‐withdrawing effect of this substituent – is the main reason for the higher halogen bonding strength.

1. Introduction

The fields of noncovalent organocatalysis and anion recognition are nowadays shaped by various noncovalent interactions, of which hydrogen bonding was the dominating one for a long time.1 Lately, however, other interactions2 such as anion‐π,3 chalcogen bonding,4 pnictogen bonding5 and halogen bonding6 are receiving steadily increasing interest. The latter in particular is by now relatively widely used,7 as it offers several advantages over hydrogen bonding: the possibility to tune the binding strength via a simple exchange of the Lewis acidic center, the sometimes high solubility in apolar solvents, and especially the high directionality.8 As a result, numerous applications of halogen bonding are by now known in solid‐state chemistry and crystal engineering.9

Halogen bonding (XB) is the interaction of a Lewis acidic halogen substituent, called halogen bond donor (XB donor), and a Lewis basic center. Halogen substituents on sufficiently electronegative backbones feature an anisotropic electron distribution with a region of positive electrostatic potential (σ‐hole)8a, 10 at the elongation of the R−X bond. In addition, particularly for strong XBs, n→σ* orbital overlap will also contribute to the overall interaction energy (Figure 1). For weak XBs, on the other hand, dispersion effects will play an increasing role.7a, 11

Figure 1.

Halogen bonding as n→σ* interaction involving the anti‐bonding orbital of the R−X bond and a lone pair of the Lewis base (LB). R=backbone, X=halogen.

In the last 15 years, applications of XB have also been developed for solution‐phase processes, particularly for anion binding and recognition.8b, 9b, 12 Pioneering work in this field was reported by Resnati, Metrangolo et al. in 2005 in the form of the first heteroditopic receptor involving XB (1 a, Figure 2) which however bound in a monodentate fashion to the anions.13 The first truly multidentate system was published by Taylor and co‐workers in 2010, which was again based on polyfluorinated XB donor motifs(1 b, Figure 2).14 Further halogen bond receptors were developed in the groups of Beer,15 Ghosh16 and Berryman,17 who incorporated imidazolium moieties into macrocyclic systems as well as in bi‐ and tripodal preorganized18 host molecules, yielding enhanced binding strength to halide anions.19

Figure 2.

Heteroditopic XB‐based tripodal receptor 1 a as reported by the group of Resnati and Metrangolo as well as tripodal receptor 1 b by the group of Taylor.

During the last decade, several other host motifs were established as anion receptors, which utilized triazolium, carbazole, catenane and rotaxane moieties, or a combination of them.20 Particularly interlocked rotaxane systems, which bear triazolium and imidazolium moieties, were widely used in that context.21 In parallel, different approaches were followed by the groups of Molina, Schubert and Berryman. Molina et al. showed in 2014 that open chain halogen bond donors can be preorganized via π‐π‐stacking (2, Figure 3), resulting in strong binding to HP₂O₇ ²⁻ in acetone.22 Schubert and co‐workers demonstrated in 2015 that it is possible to rigidify XB donors via internal hydrogen bonds to triazolium substituents (4, Figure 3), thereby increasing the binding strength to halides by about a factor of 30.23 Recently, the group of Berryman introduced preorganization via internal hydrogen bonding to iodine substituents (5, Figure 3).24 To this end, they exploited the anisotropic charge distribution of halogen substituents and coordinated the belt of higher electron density with a hydrogen bond donating amine group, which lead to an order of magnitude stronger binding of chloride, bromide and iodide.

Figure 3.

Schematic representation of Molina's π‐π‐locked triazolium receptor 2, our rotationally locked halogen bond donor 3, Schubert's hydrogen bonded triazolium receptor 4 and Berryman's intramolecular hydrogen‐bonded halogen‐bond‐based pyridinium receptor 5.

With respect to halogen bond based organocatalysis, fewer examples are known compared to anion recognition, although the number is steadily increasing.6a, 6c, 7b First studies were published in the early 2000 s reporting quinoline reductions and halide abstraction reactions by Bolm et al. and our group.25 Today, several examples for carbon‐halide bond,26 carbonyl,27 imine28 and double bond29 activations are known, but the concept of preorganization is, in this context, far less established than it is for anion recognition. To the best of our knowledge, the first such example was published in 2015 by our group:30 rotationally locked, cationic bisimidazolium‐based halogen bond donors 3 (Figure 3) showed superior activity over “non‐locked” analogues (Figure 4, 9 ^Im‐OTf). More precisely, when the catalyst loading was reduced to 0.5 mol‐%, similar yields were obtained as with 10 mol‐% of the non‐preorganized imidazolium systems.30 This trend was also found in a Michael addition reaction31 as well as in an intramolecular Nazarov cyclisation reaction.32 Compared to the “unlocked” XB donor 9 ^BIm‐BArF4 (Figure 4), “locked” catalyst 9 ^{BIm‐CF3‐BArF4} (Figure 4) accelerated the conversion rate in the Michael addition reaction by up to 50 times and in the Nazarov cyclisation by up to 15 times. In these comparisons, it is assumed that the CF₃ group exerts only steric effects (preventing rotation of the heterarene sidearms) and not also electronic ones. Since trifluoromethyl is obviously a strongly electron‐withdrawing group, this hypothesis needs to be verified. In addition, no data on the effect of preorganization had been obtained for other reactions than the ones previously mentioned. Thus, the aim of this study was twofold: on the one hand, we applied preorganized and non‐preorganized XB donors to several further reactions. On the other hand, we systematically investigated the effect of the CF₃ group, inter alia by extending our portfolio with a CF₃‐marked analogue of a freely rotating imidazolium‐based catalyst (see Scheme 1). In this context, next to a comparison of catalytic performance, we were also interested in the binding strengths of preorganized vs. non‐preorganized systems to simple halide anions as reference substrates. For that purpose, ITC titration experiments were performed in different solvents.

Figure 4.

Overview of all halogen bond donors 9 used in this study. Top: 1,3‐bis(iodoimidazolium)‐based compounds. Bottom: 1,3‐bis(iodobenzimidazolium)based compounds. X=OTf or BAr^F ₄. The nomenclature provides information on the counter anion, the presence or absence of the CF₃ group and on whether an imidazolium (Im) or benzimidazolium moiety (BIm) was used.

Scheme 1.

i) 2.5 eq. imidazole, 10 eq. K₃PO₄, DMF (0.18 m), 72 h, 135 °C, yield: 40 %; ii) 2.4 eq. lithium diisopropylamide, 2.4 eq. iodine (0.65 m THF solution), THF (0.04 m), 24 h, −78 °C→RT, yield: 25 %; iii) 4.0 eq. octyl triflate, dry DCM (0.05 m), 72 h, RT, yield: 51 %; iv) 2.4 eq TMABAr^F ₄, dry DCM (0.1 m), 72 h, RT, yield: 77 %.

2. Results and Discussion

2.1. Synthesis of a CF₃‐Marked XB Donor

To complete the library of our bis(benz)imidazolium based halogen bond donors, the CF₃‐marked analogue 9 ^{Im‐CF3‐OTf} of 1,3‐bisimidazolium 9 ^Im‐OTf was synthesized. Its synthesis route followed a sequence (Scheme 1) already published for its 1,3‐bisbenzimidazolium analogue (3, Figure 3) and started from commercially available 1,3‐difluoro‐2‐trifluoromethylbenzene (6).30 The first step involves a nucleophilic aromatic substitution (S_NAr) which afforded the neutral 1,3‐bisimidazole compound 7 in 40 % yield. After deprotonation of compound 7 with lithium diisopropylamide and subsequent addition of elemental iodine, the twofold iodinated compound 8 was obtained. The evaluation of ¹H NMR, ¹⁹F NMR, ¹³C NMR and GC/MS data showed no signs of a rotational barrier for the imidazolium moieties around the C‐N bond, which indicates that the CF₃ group has no preorganizing effect to this framework structure. A comparison of the ¹H NMR data of intermediate 8 to that of the non‐alkylated precursor of 9 ^Im‐X (Figure 4), which lacks the CF₃ group but is otherwise identical, displayed no crucial difference in the chemical shifts of the imidazolium protons. This is a first indication that the CF₃ group induces no substantial electron withdrawal from the sidearm moieties.

Finally, compound 8 was alkylated with octyl triflate, which afforded the cationic halogen bond donor 9 ^{Im‐CF3‐OTf}. Previous publications clearly found that for the activation of neutral compounds, counter anions more weakly coordinating than triflate are necessary.27a, 32 Thus, an exchange to the corresponding BAr^F ₄ salt was performed via simple metathesis with TMABAr^F ₄, affording 9 ^{Im‐CF3‐BArF4}.27a, 28b, 31, 33

2.2. ITC Titration Experiments

With XB donor 9 ^Im‐CF3 now complementing the three already published variants 9 ^Im as well as 9 ^BIm and 9 ^BIm‐CF3 (Figure 4), a comparison of their halogen bonding strength should reveal the effect of the central CF₃ group: the latter should exert a similar electronic effect in 9 ^Im‐CF3 and 9 ^BIm‐CF3, while it only leads to preorganization (hindered rotation) in the latter. Thus, if both 9 ^Im‐CF3 and 9 ^BIm‐CF3 outperform their non‐CF₃‐bearing analogues, there is likely a strong electronic effect of the CF₃ moiety, and if only 9 ^BIm‐CF3 outperforms 9 ^BIm (but 9 ^Im and 9 ^Im‐CF3 perform similarly), the difference is mostly due to preorganization.

To compare the binding strength of the triflate salts of all halogen bond donors 9 shown in Figure 4, isothermal calorimetric titrations (ITC) were performed with tetra‐n‐octylammonium or tetra‐n‐butylammonium halides in either acetonitrile, chloroform or methylene chloride. In THF, either precipitation of the [9 ^OTf ⋅ X] complexes or deiodination of the bis(benz)imidazolium moieties was observed. MTBE was ruled out as solvent due to low solubility of host and guest (e. g. 9 ^{Im‐CF3‐OTf}: <0.50 mm; (nOct)₄Cl: <5.0 mm). The titration results are summarised below in Table 1 and Figure 5.

Table 1.

Binding constant K, energy terms ΔG°, ΔH° and −TΔS° as well as stoichiometries n collected from ITC experiments.

	Entry	9 ‐OTf	X−	K [m −1]	ΔG° [kJ ⋅ mol−1]	ΔΔG° (X−−Cl−) [kJ ⋅ mol−1]	ΔH° [kJ ⋅ mol−1]	‐TΔS° [kJ ⋅ mol−1]	[%] of ‐TΔS° in ΔG°	n
CH3CN, 30 °C	1		Cl	1.23 ⋅ 106	−35.4	0.0	−16.8	−18.5	52 %	1.08
	2*	9 BIm‐CF3	Cl	1.82 ⋅ 106	−36.3	−0.9	−19.7	−16.6	46 %	0.88
	3		Br	2.06 ⋅ 106	−36.6	−1.3	−20.2	−16.4	45 %	0.91
	4		I	1.74 ⋅ 106	−36.2	−0.8	−18.6	−12.9	41 %	0.92
	5		Cl	1.49 ⋅ 105	−30.0	0.0	−15.2	−14.8	49 %	1.05
	6	9 BIm	Br	1.11 ⋅ 106	−35.2	−5.1	−17.0	−18.1	52 %	0.94
	7		I	1.61 ⋅ 106	−36.0	−5.9	−20.0	−16.0	44 %	0.94
	8		Cl	2.79 ⋅ 104	−25.8	0.0	−13.8	−12.1	47 %	1.27
	9	9 Im‐CF3	Br	2.18 ⋅ 104	−25.2	+0.6	−17.2	−7.9	32 %	1.01
	10		I	9.72 ⋅ 104	−29.0	−3.1	−15.9	−13.1	45 %	0.80
	11		Cl	5.71 ⋅ 105	−33.4	0.0	−13.9	−19.5	58 %	1.10
	12	9 Im	Br	6.25 ⋅ 105	−33.7	−0.3	−17.1	−16.6	49 %	0.94
	13		I	3.17 ⋅ 105	−31.9	+1.5	−17.3	−14.6	46 %	0.96
CHCl3, 30 °C	14		Cl	1.41 ⋅ 106	−35.7	0.0	−10.7	−25.0	70 %	1.03
	15	9 BIm‐CF3	Br	2.02 ⋅ 106	−36.6	−0.8	−10.4	−26.1	71 %	0.99
	16		I	3.24 ⋅ 106	−37.9	−2.1	−11.3	−26.5	70 %	0.99
	17		Cl	2.69 ⋅ 105	−31.5	0.0	−7.9	−23.6	75 %	0.91
	18	9 BIm	Br	1.41 ⋅ 106	−35.8	+4.3	−6.1	−29.7	83 %	0.89
	19		I	1.01 ⋅ 106	−34.8	+3.3	−9.9	−24.9	71 %	0.97
	20	9 Im‐CF3	insoluble
	21		Cl	2.58 ⋅ 105	−32.3	0.0	−8.4	−23.9	74 %	1.05
	22	9 Im	Br	5.53 ⋅ 105	−33.4	−1.1	−12.4	−20.9	63 %	1.00
	23		I	1.26 ⋅ 106	−35.5	−3.2	−13.5	−22.0	62 %	0.99
CH2Cl2, 20 °C	24		Cl	2.22 ⋅ 106	−35.6	0.0	−16.3	−19.3	54 %	0.94
	25	9 BIm‐CF3	Br	7.42 ⋅ 106	−38.5	−2.9	−16.4	−22.1	57 %	1.00
	26		I	1.03 ⋅ 107	−39.4	−3.8	−16.7	−22.7	58 %	1.00
	27		Cl	2.29 ⋅ 105	−30.2	0.0	−7.9	−22.2	74 %	1.05
	28	9 BIm	Br	1.15 ⋅ 106	−34.1	−3.9	−14.9	−19.1	56 %	0.95
	29		I	2.91 ⋅ 106	−36.4	−6.2	−16.7	−19.6	54 %	0.99
	30		Cl	1.57 ⋅ 106	−34.8	0.0	−8.4	−26.4	76 %	0.92
	31	9 Im‐CF3	Br	1.28 ⋅ 106	−34.3	+0.5	−11.1	−23.2	68 %	1.04
	32		I	1.02 ⋅ 106	−33.7	+1.1	−12.7	−21.0	62 %	1.07
	33		Cl	1.78 ⋅ 106	−35.1	0.0	−12.2	−22.8	65 %	0.98
	34	9 Im	Br	2.00 ⋅ 106	−35.4	−0.3	−16.4	−19.0	54 %	1.00
	35		I	1.47 ⋅ 106	−34.6	+0.4	−18.5	−16.1	46 %	0.99

*(nBu)₄Cl was dried prior to use and its guest solution prepared using Schlenk techniques.

Figure 5.

Bar graph representation of the binding energies (ΔG) of the 9 ⋅ X adducts.

As tetraalkylammonium salts are commonly hygroscopic, the unknown water content of the ammonium halides used is a possible source of error. To probe for the influence of residual water on the measured binding affinities, an ITC titration was done using a guest solution prepared under inert conditions with pre‐dried (nBu)₄Cl (entry 2). Compared to ITC titrations utilizing non‐dried halide salts, a marginal 0.9 kJ ⋅ mol^‐1 stronger binding was noted, hence all further titrations were performed at ambient conditions.

The bidentate nature of the halide binding with all XB donors 9 ^OTf is apparent from the stoichiometries observed in the titrations: only 3 of 34 titrations are outside a range of 0.90≤n≤1.10.

Of all the compounds titrated, XB donor 9 ^BIm‐CF3 shows the highest affinity towards halides, with the overall binding free energy (ΔG°) ranging from −35.4 up to −39.4 kJ ⋅ mol^‐1. The donors 9 ^BIm (−30.0 up to −36.4 kJ ⋅ mol^‐1) and 9 ^Im (−31.9 up to −35.5 kJ ⋅ mol^‐1) do have comparable affinities, whereas 9 ^Im‐CF3 shows a somewhat reduced binding energy (−25.2 up to −34.8 kJ ⋅ mol^‐1).

A direct comparison of the binding energies of 9 ^BIm‐OTf and 9 ^{BIm‐CF3‐OTf} confirms in almost all cases the hypothesis that preorganization leads to superior binding strength. The highest constant K measured was the binding of 9 ^{BIm‐CF3‐OTf} to iodide in methylene chloride with 1.03 ⋅ 10⁷  m ⁻¹. Under the same conditions, the affinity of 9 ^BIm‐OTf towards iodide decreased by almost one order of magnitude (2.91 ⋅ 10⁶  m ⁻¹). A higher electron deficiency induced by trifluoromethylation cannot explain the superior binding strength found: 9 ^{Im‐CF3‐OTf} shows lower binding free energies to halides than 9 ^Im‐OTf in all tested solvents. Therefore, the electronic contribution of electron‐withdrawing CF₃ substituents to the overall halogen bonding strength is negligible. In fact, a decrease of the binding free energies is observed. Interestingly, the solubility of the bisimidazolium compound 9 ^Im‐OTf is reduced when a CF₃ groups is present in the structure, which could affect the halide binding capabilities of 9 ^{Im‐CF3‐OTf}. However, no precipitation was observed in the ITC experiments mentioned above.

In view of prior results,34 a preference for harder anions and therefore an increased affinity in the order I⁻<Br⁻<Cl⁻ could be expected, yet a minor trend in the opposite direction was observed. While for all XB donors with the exception of 9 ^{Im‐CF3‐OTf}, the binding free energies ΔG° increase when going from chloride to bromide, this trend continues with iodide only for benzimidazolium donors (see ΔΔG°(X⁻−Cl⁻)). However, as ΔΔG° values can only be considered significant when exceeding 3 kJ ⋅ mol⁻¹,35 no clear binding trend towards any halides can be derived from these data.

As already found in earlier titrations,34 the entropic term −TΔS° heavily contributes to the overall binding energy, exceeding 50 % for most of the energy terms (ΔG°) found (exceptions: entries 1–4, 7–10, 12, 13, and 35). The average distribution of the entropic term over all measurements in a given solvent increase from acetonitrile (47 %) over methylene chloride (60 %) to chloroform (71 %), correlating to the relative polarities of the solvents in their inverse order.36 With a few exceptions, however, the overall binding strength does not appear to be affected by the choice of solvent: XB donor 9 ^Im‐CF3 shows halide affinities increased by ΔΔG°_{CH2Cl2‐MeCN}=+9.0 (Cl⁻), +9.2 (Br⁻) and +4.8 (I⁻) kJ ⋅ mol⁻¹, respectively. The binding towards iodides is also notably stronger for 9 ^Bim‐CF3 (ΔΔG°_{CH2Cl2‐MeCN}: +7.9; ΔΔG°_CHCl3‐MeCN: +6.3 kJ ⋅ mol^‐1) and for 9 ^Im (ΔΔG°_{CH2Cl2‐MeCN}: +2.7; ΔΔG°_CHCl3‐MeCN: +3.6 kJ ⋅ mol⁻¹). All other ΔΔG° values do not exceed ±2.0 kJ ⋅ mol⁻¹, which by and large points to an enthalpy‐entropy compensation effect.

2.3. Halogen Bond Organocatalysis – Preface

Overall, the titration studies confirmed that syn‐locked 9 ^{BIm‐CF3‐OTf}, compared to non‐locked 9 ^BIm‐OTf, shows a stronger binding to halide anions. Furthermore, they refuted the possibility that a mere electronic effect of the CF₃‐group is responsible for that observation (via the comparison of the binding constants of 9 ^{Im‐CF3‐OTf} and 9 ^Im‐OTf, Table 1). To see whether the same trends are seen for the catalytic activity of XB donors 9 (Figure 4), we then moved our focus to catalysis studies using several benchmark reactions described below.

2.3.1. Halide Abstraction Reactions in Halogen Bond Catalysis

First, we focused on the activation of benzhydryl bromide (10, Scheme 2) with stoichiometric amounts of XB donors 9 ^OTf (Figure 4), since this reaction a) has virtually no background reactivity, b) can be easily followed via ¹H NMR spectroscopy, c) is unaffected by the presence of Brønsted acids and d) was already activated by several halogen bond donors with success.25d In these examples, reaction times of 2–4 days in presence of 9 ^Im‐OTf and a triazolium catalyst were necessary to yield full conversion to compound 11.25b, 25c

Scheme 2.

Solvolysis reaction of benzhydryl bromide (10) in deuterated acetonitrile in the presence of selected halogen bond donors. For satisfactory and reproducible results, it is necessary to have exactly 2 eq. water present. The reaction time is 9.5 h.

First, we reproduced our results with catalyst 9 ^Im‐OTf: as reported,25b after 96 h reaction time a conversion >90 % was obtained. Next, preorganized bisbenzimidazolium catalyst 9 ^{BIm‐CF3‐OTf} was applied in the reaction, which resulted in nearly full conversion to product 11 within 9.5 h reaction time (Table 2, Entry 5 and Figure 6). After the same period, only 56 % of 11 were obtained in presence of 9 ^Im‐OTf (Table 2, Entry 3). With the unsubstituted bisbenzimidazolium catalyst 9 ^BIm‐OTf, 72 % conversion to 11 was observed, while with CF₃‐marked bisimidazolium compound 9 ^{Im‐CF3‐OTf}, a conversion of 56 % was determined, similar to the one with 9 ^Im‐OTf (Table 2, Entries 4/2).

Table 2.

¹H NMR conversion of benzhydryl bromide (10) to amide 11 activated by XB donors 9 and the determined relative rate accelerations.

Entry	XB donor[a]	Conversion to 11 [%][b,c]	k rel [d]
1	/	<5	1
2	9 Im‐CF3‐OTf	56	150
3	9 Im‐OTf	56	225
4	9 BIm‐OTf	72	540
5	9 BIm‐CF3‐OTf	90	1000

[a] 1 eq. of the respective XB donor was used. [b] The reaction time was 9.5 h. [c] Determined via integration of selected signals of the starting material vs. the product in the ¹H NMR spectrum. [d] Referenced values to the background reactivity after approximately 0.6 h reaction time.

Figure 6.

Conversion vs. time profile for the activation of benzhydryl bromide (10) in presence of different halogen bond activators (monitored via ¹H NMR spectroscopy).

This data represents further evidence that the CF₃ group has only a steric effect on the activating properties of the XB donors, as 9 ^{Im‐CF3‐OTf} should otherwise be superior to 9 ^Im‐OTf. Finally, relative rate constants k _rel were determined for all catalysts (Table 2), which revealed an acceleration of 1000 with 9 ^{BIm‐CF3‐OTf} compared to the background reaction and an acceleration by more than 6 times compared to the weakest activator 9 ^{Im‐CF3‐OTf}. Encouraged by these first promising results, we next studied the benzhydrylation of trimethoxy benzene (12) with benzhydryl bromide (10), shown in Scheme 3.

Scheme 3.

Reaction between benzhydryl bromide (10) and trimethoxy benzene (12) in dry, deuterated chloroform.

Although this reaction is analogous to the solvolysis of benzhydryl bromide (10) in acetonitrile, it was chosen for several reasons: a) due to its similarity, it should follow comparable trends for the activity of our XB donors, which could be considered as a confirmation of the previous results, b) because a different solvent is used, it is possible to probe for any influence of solvents onto the overall activity of our catalysts and c) as monocationic, pyridinium‐based halogen bond donors were already successfully employed in this reaction – resulting in up to 90 % conversion within 12 h reaction time – (benz)imidazolium‐based structures should be activating as well.26a In line with our earlier study,26a Cs₂CO₃ was added to the reaction to rule out hidden Brønsted acid catalysis.

Since twofold, cationic bis(benz)imidazolium catalysts should be stronger activators than pyridinium based structures, we set our observation time to a period of 8 h. In Figure 7, a conversion vs. time profile is shown. Again, 9 ^{BIm‐CF3‐OTf} is the strongest activator, followed by its freely rotatable analogue 9 ^BIm‐OTf and the imidazolium compound 9 ^Im‐OTf (see also Table 3).

Figure 7.

Conversion vs. time profile for the acylation of trimethoxy benzene (12) with benzhydryl bromide (10) in presence of different halogen bond activators (followed by ¹H NMR spectroscopy). Activator 9 ^{Im‐CF3‐OTf} could not be tested in the reaction due to solubility issues.

Table 3.

¹H NMR conversions and relative rate constants for the benzhydrylation of trimethoxy benzene (12) with benzhydryl bromide (10) in presence of selected halogen bond activators.

Entry	XB donor[a]	Conversion to 13 [%][b,c]	k rel [d]
1	/	<5	1
2	9 Im‐OTf	62	210
3	9 BIm‐OTf	68	250
4	9 BIm‐CF3‐OTf	79	290

[a] 1 eq. of the respective catalyst was used. [b] The reaction time was 8 h. [c] Determined via integration of selected signals of the starting material vs. the product in the ¹H NMR spectrum. [d] Referenced values to the background reactivity after approximately 0.6 h reaction time.

Catalyst 9 ^{Im‐CF3‐OTf} could not be tested in the reaction due to solubility issues. With all catalysts, conversions towards compound 13 were determined to be between 62–79 % (Table 3). With catalyst 9 ^{BIm‐CF3‐OTf}, a relative rate acceleration of 290 was determined compared to background reactivity, whereas with the weakest activator 9 ^Im‐OTf, a 210‐fold acceleration was observed (Table 3). Compared to previous results of the solvolysis of benzhydryl bromide (10), only relatively subtle differences in the relative accelerations were found, with the best XB donor 9 ^{BIm‐CF3‐OTf} being only 1.4 times more active than the weakest one (9 ^Im‐OTf). The order of activity of the XB donors was not affected by the solvent, however.

Both reactions discussed so far have the disadvantage that stoichiometric amounts of the respective catalysts are necessary in order to obtain adequate yields. Therefore, we next focused on catalytic applications of these Lewis acids. As first test reaction for this purpose, the activation of 1‐chloroisochroman (14, Scheme 4) was chosen, as several features qualify it as an ideal benchmark reaction: a) it had already been activated by halogen bonding, b) there is no background reactivity at −78 °C, c) several reference compounds (e. g. elemental iodine, HOTf, cationic hydrogen bond donors)25e, 30 are inactive in this reaction and d) very low catalyst loadings are necessary.25e, 30 With merely 0.5 mol‐% of 9 ^{BIm‐CF3‐OTf}, a strong activation had already been observed, and thus we focused in the following on this reaction set‐up.30

Scheme 4.

Reaction of 1‐chloroisochroman (14) with ketene silyl acetal 15 in presence of different halogen bond donors. No background reaction at −78 °C in THF is observed.

After a period of 6 h, 55 % of 16 was isolated with 9 ^{BIm‐CF3‐OTf} as catalyst, while the yield markedly decreased to 25 % when the non‐preorganised catalyst 9 ^BIm‐OTf was employed (Table 4, Entries 5 and 4). This again demonstrates the superiority of the preorganized halogen bond donor. Besides that, with imidazolium catalysts 9 ^Im‐OTf and 9 ^{Im‐CF3‐OTf} only 17 % yield of compound 16 were isolated (Table 4, Entries 2 and 3). The comparable performance of these two catalysts shows again that the CF₃ group does not have a pronounced electron withdrawing influence on the catalysts.

Table 4.

Yield of compound 16 in the reaction between 1‐chloroisochroman (14) and ketene silyl acetal 15 in presence of different halogen bond donors.

Entry	XB donor[a]	Yield of 16 [%][b,c]
1	/	<5
2	9 Im‐CF3‐OTf	17
3	9 Im‐OTf	17
4	9 BIm‐OTf	25
5	9 BIm‐CF3‐OTf	55

[a] 0.5 mol‐% of the respective catalyst was used. [b] The reaction time was 6 h. [c] Isolated yields.

2.3.2. Neutral Compound Activation in Halogen Bond Catalysis

Up to this point, only examples for halide abstraction reactions were discussed, because it is known from many publications in the field of anion recognition,12b, 12c, 34, 37 that XB donors strongly interact with halide anions.

However, many applications in organocatalysis target the activation of neutral compounds, and thus we further investigated the catalytic activity of XB donors 9 (Figure 4) in reactions involving carbonyl or nitro compounds. Previous studies in halogen‐32 and chalcogen bond33b organocatalysis highlighted the relevance of non‐coordinating counter anions to achieve satisfying yields and the necessity to work with dry solvents to suppress a competitive interaction with water. To this end, the BAr^F ₄ counter anion (Figure 8) was used multiple times with great success in our group, but also applications of the BF₄, PF₆ or NTf₂ anion are known.7b

Figure 8.

Structure of the BAr^F ₄ ⁻ counter anion.

As first test reaction, we focused on the Diels Alder reaction of cyclopentadiene (17) and methyl vinyl ketone (18, Scheme 5).26b, 27a Moreover, Diels Alder reactions are powerful tools in organic synthesis, as large ring systems with several stereogenic centers can be formed in a single step.38 To achieve high conversions under mild conditions, Lewis acids are commonly added. In 2014, this reaction was already successfully preformed in presence of 20 mol‐% of XB donor 9 ^Im‐BArF4 as catalyst.27a First orientating experiments were repeated with a catalyst load of 20 mol‐%, but after the first measurement point (approximately after 20 min), each catalyst (9 ^BArF4, Figure 4) had yielded full conversion to product 19. Possible explanations for the lower activity of catalyst 9 ^Im‐BArF4 in previous publications could be an inhibition of the catalyst by water, or the presence of 9 ^Im‐OTf residues as artifact of an incomplete anion exchange, which equally inhibits the iodine centers. A reduction of the catalyst loading to 2.5 mol‐% did not show any improvement: after 20 min reaction time, nearly full conversion for each catalyst again was observed. Finally, the catalyst load was reduced to 0.5 mol‐%, which finally lowered the catalysts activity enough so that only in presence of the strongest catalyst 9 ^{BIm‐CF3‐BArF4}, nearly full conversion to compound 19 was observed within one hour (Table 5, Entry 5 and Figure 9). With the simple imidazolium catalysts 9 ^Im‐BArF4 and 9 ^{Im‐CF3‐BArF4}, approximately 22–25 % conversion to compound 19 was found, whereas with bisbenzimidazolium variant 9 ^BIm‐BArF4, a conversion of 92 % was observed in the same period (Table 5, Entries 2, 3 and 4).

Scheme 5.

Diels‐Alder reaxtion of cyclopentadiene (17) and methyl vinyl ketone (18) in presence of selected halogen bond donors. Cyclopentadiene was freshly prepared by cracking of dicyclopentadiene at 250 °C.

Table 5.

Overview of the ¹H NMR conversion and the relative rate constants for the Diels Alder reaction of cyclopentadiene (17) with methyl vinyl ketone (18) in presence of selected halogen bond activators.

Entry	XB donor[a]	Conversion to 19 [%][b,c,d]	k rel [e]
1	/	10	1
2	9 Im‐CF3‐BArF4	22 (4)	2
3	9 Im‐BArF4	25 (4)	2
4	9 BIm‐BArF4	92 (11)	6
5	9 BIm‐CF3‐BArF4	94 (21)	10

[a] 0.5 mol‐% halogen bond catalysts were used. [b] The reaction time was 1 h. [c] Determined via integration of selected signals of the starting material vs. the product in the ¹H NMR spectrum. [d] Conversion of the reaction after 5 min for entries 4 and 5 and conversion of the reaction after 20 min for entries 2 and 3. [e] Referenced values to the background reactivity after approximately 20–30 min reaction time.

Figure 9.

Conversion vs. time profile for the Diels‐Alder reaction of cyclopentadiene (17) and methyl vinyl ketone (18) in presence of different halogen bond activators. The conversion was determined by integration of selected signals of the ¹H NMR spectrum after defined periods.

As the relative rate constants (Table 5) show, the use of 9 ^{BIm‐CF3‐BArF4} allowed a 10 times faster reaction compared to background reactivity, while with imidazolium catalysts 9 ^Im‐BArF4 and 9 ^{Im‐CF3‐BArF4}, only a reaction acceleration by a factor of two was observed (Table 5). Again, the preorganized catalyst 9 ^{BIm‐CF3‐BArF4} was the strongest activator and the CF₃‐marked imidazolium catalyst 9 ^{Im‐CF3‐BArF4} was equally active as its non‐subststituted analogue 9 ^Im‐BArF4, which corroborates the findings of the halide abstraction benchmark reactions discussed above.

Overall, though, there are some drawbacks to this reaction (comparably strong background reactivity, sensitivity to water and acids, high toxicity of reactants), and hence we focused on a more robust carbonyl activation reaction. In 2017, Breugst et al.27c and our group31 independently published the activation α,β‐unsaturated ketones in a Michael addition reaction by halogen bonding (Scheme 6). Earlier case studies of Breugst et al. showed that molecular iodine can accelerate the reaction through halogen bonding and that a participation and activation by Brønsted acids can be ruled out.39

Scheme 6.

Schematic reaction of indole (20) with trans‐β‐crotonophenone (21) in presence of different halogen bond donors.

Most of our catalysts 9 (except for 9 ^{Im‐CF3‐BArF4}) were already tested in this reaction, and a conversion of only 70 % was previously found when 9 ^{BIm‐CF3‐BArF4} was used as catalyst.31 This result was not reproduced this time but instead, almost full conversion to compound 22 was observed (Figure 10). Again, the most plausible explanation is a too high water content in the earlier experiments, which resulted in partial toxification of catalysts 9. This phenomenon was confirmed for catalyst 9 ^Im‐BArF4, which showed initially no activity, but after removal of water (via addition of molecular sieves), the reaction was catalyzed smoothly. Reference experiments with molecular sieves as only additives showed no conversion to product 22, which excludes the possibility of it being a catalyst.

Figure 10.

Conversion vs. time profile for the Michael addition reaction of indole (20) and trans‐β‐crotonophenone (21) in presence of different halogen bond activators. The conversion was determined by integration of selected signals of the ¹H NMR spectrum after defined periods.

Somewhat surprisingly, the CF₃‐marked imidazolium compound 9 ^{Im‐CF3‐BArF4} was more active than its non‐substituted analogue 9 ^Im‐BArF4 and virtually as active as bisbenzimidazolium catalyst 9 ^BIm‐BArF4 (Figure 10). The reason for this unexpectedly strong performance are currently unclear. As Table 6 shows, the final conversions achieved with 9 ^{Im‐CF3‐BArF4} and 9 ^BIm‐BArF4 are almost identical (55 % vs. 54 %, Entries 2 and 4). With catalyst 9 ^{BIm‐CF3‐BArF4}, 90 % conversion to compound 22 was obtained, whereas in presence of 9 ^Im‐BArF4 only 35 % conversion was achieved, which is comparable to previous results. Relative rate accelerations displayed a 1200‐fold higher activation for 9 ^{BIm‐CF3‐BArF4} with respect to the background reactivity and a 3.5–8 fold faster reaction in comparison to the remaining catalysts. The imidazolium catalyst accelerated the reaction by about 300 times and with the non‐preorganized benzimidazolium catalyst 9 ^BIm‐BArF4 a 240 times faster reaction was observed.

Table 6.

Overview of the ¹H NMR conversion and the relative rate constants for the Michael addition reaction of indole (20) with trans‐β‐crotonophenone (21) in presence of selected halogen bond activators.

Entry	XB donor[a]	Yield of 22 [%][b,c]	k rel [d]
1	/	<5	1
2	9 Im‐CF3‐BArF4	55	325
3	9 Im‐BArF4	35	150
4	9 BIm‐BArF4	54	240
5	9 BIm‐CF3‐BArF4	90	1200

[a] 20 mol‐% halogen bond catalysts were used. [b] The reaction time was 14 h. [c] Determined via integration of selected signals of the starting material vs. the product in the ¹H NMR spectrum. [d] Values referenced to the background reactivity after approximately 1 h reaction time.

Finally, we were interested in the activation of nitro groups in a Michael type addition reaction (Scheme 7) which was recently published by our group in the context of chalcogen bonding organocatalysis.33b To the best of our knowledge, this constituted the first example of a nitro group activation by halogen bonding, albeit elemental iodine is reported to accelerate this type of reaction by an unknown mode of action.39, 40 Conveniently, nearly no background reactivity is observed in the conversion of indole 23 to product 25, while in presence of Brønsted acids, several Lewis acids or selected thiourea‐based hydrogen bond donors, also only minor amounts of 25 are formed.

Scheme 7.

Reaction of 5‐methoxyindole (23) with trans‐β‐nitrostyrene (24) in presence of different halogen bond donors.

The strongest halogen bond donor 9 ^{BIm‐CF3‐BArF4} converted only 39 % of indole 23 to compound 25 (Table 7, Entry 5 and Figure 11). In comparison, a triazolium‐based organotellurium catalyst (10 mol‐%) resulted in 81 % conversion to compound 25 after the same period.33b One possible explanation for this observation is a better fit of the tellurium moieties to the nitro group than is possible for iodine, which is in accordance with findings that an iodinated triazolium catalyst was almost inactive. Next, the remaining halogen bond catalysts were tested in the reaction. With catalyst 9 ^BIm‐BArF4 only 16 % conversion to compound 25 was noted and with 9 ^Im‐BArF4 and 9 ^{Im‐CF3‐BArF4} nearly no reaction was observed (4–7 %, Table 7, Entries 2/3). These observations are fully in line with the trends established earlier.

Table 7.

Overview of the ¹H NMR conversion and the relative rate constants for the nitro Michael addition reaction of 5‐methoxyindole (23) with trans‐β‐nitrostyrene (24) in presence of selected halogen bond activators.

Entry	XB donor[a]	Yields to 25 [%][b,c]	k rel [d]
1	/	<5	1
2	9 Im‐CF3‐BArF	7	20
3	9 Im‐BArF	4	12
4	9 BIm‐BArF	16	55
5	9 BIm‐CF3‐BArF	39	100

[a] 20 mol‐% halogen bond catalysts were used. [b] The reaction time was 48 h. [c] Determined via integration of selected signals of the TES standard vs. the product in the ¹H NMR spectrum. [d] Referenced values to the background reactivity after approximately 6 h reaction time.

Figure 11.

Conversion vs. time profile for the nitro Michael addition reaction of 5‐methoxyindole (23) and trans‐β‐nitrostyrene (24) in presence of different halogen bond activators. The conversion was determined by integration of selected signals of the ¹H NMR spectrum after defined periods.

Both catalysts only led to a 12–20 times faster reaction with respect to the background reaction. For catalyst 9 ^BIm‐BArF4 a 55‐fold rate acceleration was observed and for the strongest catalyst 9 ^{BIm‐CF3‐BArF4} _, this figure reached two orders of magnitude (Table 7, Entry 5).

Regardless of the low catalytic activity of all tested XB catalysts in this reaction, both the superiority of the preorganized system and the lack of any electron withdrawing effect of the CF₃ moiety towards the XB‐donating moieties was demonstrated again.

3. Conclusions

In this study, four structurally very related halogen bond donors were investigated via ITC titrations and in catalysis benchmark reactions. In all cases, it could be clearly demonstrated that a preorganized bis(benzimidazolium)‐based XB donor features the strongest Lewis acidity and catalytic activity. Its performance was clearly superior to the one of its unsubstituted analogues lacking a central CF₃ group (which is pivotal for preorganization via hindered rotation). As the corresponding imidazolium‐based pair of CF₃‐substituted and ‐unsubstitued XB donors were similar active in nearly every reaction, any electronic effect of this group can be ruled out. The comparable performance of these two catalysts also indicates that the CF₃ group does not exert any negative steric influence on substrate binding. Future work will deal with the application of our strongest catalyst in further catalytic reactions and with the continued optimization of other preorganized halogen bonding motifs.

Experimental Section

ITC experiments: Extra dry acetonitrile was purchased from Acros Organics (<10 ppm H₂O). Chloroform was dried over 3 Å molecular sieves and filtered over activated, basic aluminium oxide 90 (63–200 μm) to remove adventitious traces of hydrogen chloride. Methylene chloride was distilled, stored over 3 Å molecular sieves and finally dried on an alox column using a MBRAUN MB SPS‐800 solvent purification system. Synthesized compounds used in measurements were dried under high vacuum prior to use, weighed out and then dissolved in the respective amount of solvent to prepare either 0.5 mm or 1.0 mm solutions, which were titrated against 5.0 mm or 10.0 mm solutions of tetralkylammonium halides, respectively. Commercially available tetraalkylammonium halides were used without further purification.

ITC measurements were performed on a MicroCal VP‐ITC system from GE Healthcare using a reference power of 34.7 μcal/s, a filter period of 2 s, a stirrer speed of 329 rpm, an injection volume of 8.0 μL for guest solutions (ammonium salt) and a time spacing of 160 s between injections. All titrations were performed at a jacket temperature of either 30 °C (303.25 K) when using acetonitrile or chloroform as solvent or 20 °C (293.25 K) when using methylene chloride as solvent. Evaluation of the data sets was performed using Origin 7 with manual integral correction and, if necessary, a subtraction of straight lines.

Part A: General information for catalysis experiments: For all reactions freshly prepared stock solutions of the respective compounds were used. All solvents were previously dried with 4 Å molecular sieves. Unless specified differently, deuterated solvents were used. ¹H NMR spectra were recorded on a Bruker AVIII 300 at 298.5 K. Evaluation of the ¹H NMR data was performed with MestReNova 9.0.1‐13254. The exact acquisition time was extracted from the MestReNova‐fid. file.

Part B: Catalysis Experiments

Activation of benzhydryl bromide (10): The reaction was performed in acetonitrile. In each case, 200 μl of the respective stock solution of the halogen bond activator (9.99 mmol, 0.05 m, 1 eq.), benzhydryl bromide (10) (2.47 mg, 9.99 μmol, 0.05 m, 1 eq.) and wet acetonitrile (0.36 mg, 19,99 μmol, 0.1 m, 2 eq.) were added to an NMR tube, sealed and vigorously shaken for one minute. Finally, periodic ¹H NMR spectra were recorded with a total duration of 10 h. The conversion was determined by the ratio of starting material 10 and the product 11. Hereto, the relative integral of a significant singlet of benzhydryl bromide (6.45 ppm) was set to 1. The conversion is equal to 100 %, divided through the sum of starting material 10 and product 11 (significant doublet at 6.30 ppm) and multiplied with the integral of the product signal.

Activation of benzhydryl bromide (10) in a Friedel‐Crafts alkylation reaction with 1,3,5‐trimethoxy benzene (12): The reaction was performed in chloroform. For the reaction, Cs₂CO₃ (3.26 mg, 9.99 μmol, 1 eq.) and in each case 200 μl of the respective stock solution of the halogen bond activator (9.99 μmol, 0.05 m, 1 eq.), benzhydryl bromide (10; 2.47 mg, 9.99 μmol, 0.05 m, 1 eq.) and 1,3,5‐trimethoxy benzene (12; 1.68 mg, 9.99 μmol, 0.05 m, 2 eq.) were added to an NMR tube, sealed, shaken and sonicated for 2 minutes. Finally, periodic ¹H NMR spectra were recorded with a total duration of 8 h. The conversion was determined by the ratio of starting material 12 and the product 13. Hereto the relative integral of methoxy groups of 1,3,5‐trimethoxy benzene (12) (s, 3.77 ppm) was set as 1. The conversion is equal 100 %, divided through the sum of starting material 12 and product 13 (signals for the two methoxy groups at 3.80 ppm (s) and 3.58 ppm (s)) and multiplied with the integral of the product signals.

Catalytic activation of 1‐chloroisochroman (14): The reaction was performed in non‐deuterated THF. To an oven dried Schlenk finger, equipped with a stirring bar and rubber septum under argon atmosphere were added the respective halogen bond catalyst (0.5 μmol, 0.005 eq.) and 900 μl dry THF. The mixture was cooled to −78 °C and subsequently, 100 μl of a freshly prepared 1 m stock solution of 1‐chloroisochroman (14) in THF were added. The mixture was stirred for 20 min and ketene silyl acetal 15 (28.26 mg, 150 μmol, 32.9 μl, 1.5 eq.) was added and the mixture was stirred at −78 °C for 6 h. Next, the reaction mixture was quenched through addition of 200 μl of a 0.5 m methanolate solution in methanol (100 μmol, 1 eq.), filtered through a plug of silica with diethyl ether and was finally purified via column chromatography (pentane : diethyl ether 9 : 1). Product compound 16 was obtained as colourless oily compound. The purity of the isolated compound was proven by TLC and ¹H NMR measurements.

Catalytic Diels‐Alder reaction of cyclopentadiene (17) and methyl vinyl ketone (18): The reaction was performed in methylene chloride. Cyclopentadiene (17) was freshly prepared from dicyclopentadiene through thermal cracking at 250 °C. In each case, 200 μl of the respective stock solution of the halogen bond catalyst (0.03 μmol, 0.015 m, 0.005 eq.), cyclopentadiene (17; 10.24 mg, 155 μmol, 0.78 m, 10 eq.) and methyl vinyl ketone (18; 1.09 mg, 15.5 mmol, 0.078 m, 1 eq.) were added to an NMR tube, sealed and shaken for 1 min. Finally, periodic ¹H NMR spectra were recorded with a total duration of 1 h. The conversion was determined by the ratio of starting material 18 and the product 19. Hereto the relative integral of methyl group of methyl vinyl ketone (18; s, 2.25 ppm) was set as 1. The conversion is equal 100 %, divided through the sum of starting material 18 and product 19 (signal of the methyl group of the endo‐ and exo‐ product at 2.09 ppm (s) and at 2.18 ppm (s) with a ratio of approximately 10 : 1) and multiplied by the sum of integrals of the product signals.

Catalytic Michael addition reaction of indole (20) and trans ‐β‐crotonophenone (21): The reaction was performed in methylene chloride. In each case, 200 μl of the respective stock solution of the halogen bond catalyst (12 μmol, 0.06 m, 0.2 eq.), indole (20; 7.03 mg, 60 μmol, 0.3 m, 1 eq.) and trans‐β‐crotonophenone (21; 8.77 mg, 60 μmol, 0.3 m, 1 eq.) were added to an NMR tube, sealed and shaken for 1 min. Finally, periodic ¹H NMR spectra were recorded with a total duration of 14 h. The conversion was determined by the ratio of starting material 21 and the product 22. Hereto the relative integral of the methyl group of trans‐β‐crotonophenone (21, dd, 1.99 ppm) was set to 1. The conversion is equal 100 %, divided through the sum of starting material 21 and product 22 (signal of the methyl group at 1.46 ppm (d) and multiplied by the integral of the product signal. Alternatively, the signal of the CH bond (s, 3.78 ppm) in the product 22 was compared with the hydrogen in β‐position (m, 6.55 ppm) of the indole (20).

Catalytic nitro‐Michael addition reaction of 5‐methoxyindole (23) and trans‐β‐nitrostyrene (24): The reaction was performed in methylene chloride. In each case, 200 μl of the respective stock solution of the halogen bond catalyst (3.74 μmol, 0.04 m, 0.2 eq., 88.0 μl), 5‐methoxyindole (23; 8.25 mg, 56.1 μmol, 0.28 m, 3 eq.) and trans‐β‐nitrostyrene (24; 2.79 mg, 18.7 μmol, 0.09 m, 1 eq.) were added to an NMR tube, sealed and shaken for 1 minute. Finally, periodic ¹H NMR spectra were recorded with a total duration of 48 h. The conversion was determined by the ratio of the product 25 against a tetraethyl silyl (TES) standard. Hereto the relative integral of C₂ group of TES (0.38 eq., q, 0.55 ppm) was set as 1. Integration of the newly formed CH₂‐ and CH‐bond (m, 5.04 ppm) of product 25 and multiplication with 100 % directly gave the conversion.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

M. H. H. Voelkel, P. Wonner, S. M. Huber, ChemistryOpen 2020, 9, 214.