Since the discover of crown ethers, the development of hosts for recognizing various guest species has mainly focused on macrocyclic structures (Cram, 1988; Lehn, 1988; Pedersen, 1988; Gong et al., 2010; Chun et al., 2013; Jurícek et al., 2014; Liu et al., 2019). Methylene–bridged macrocyclic arenes, for example calixarenes (Baldini et al., 2007; Guo and Liu, 2012), pillararenes (Ogoshi et al., 2008; Xue et al., 2012; Wang et al., 2016; Yang et al., 2016), coronarenes (Wang, 2018), helic[6]arene (Zhang et al., 2016), biphenarenes (Chen et al., 2015; Dai et al., 2017; Li et al., 2019; Wang et al., 2019b), and etc. (Guo et al., 2018; Luo et al., 2018; Ma et al., 2018) have been widely used in host-guest chemistry, self-assembly materials, and biomedical field (Song and Yang, 2015; Alsbaiee et al., 2016; Li et al., 2017; Jie et al., 2018; Chen et al., 2019; Yang et al., 2019). Naphthalene-based macrocyclic arenes, termed as calixnaphthalenes, have also been produced (Poh et al., 1989; Andreetti et al., 1993; Shorthill et al., 2004; AlHujran et al., 2012; Avetta et al., 2012). However, calixnaphthalenes have not become highly popular receptors because they do not have unique molecular recognition properties. Considering that pillararenes with pillar-shape topologic structures have shown nice host-guest properties, we wondered whether we can create acyclic pillarnaphthalenes (Scheme 1), which would have deep, pillar-shape, and π-rich cavities, and maybe better binding abilities than calixnaphthalenes. As detailed below, we did not get such macrocycles, but succeed in making acyclic pillarnaphthalene oligomers.

Acyclic hosts that contain partially enclosed cavities capable of binding guests provided alternatives with unique synthetic and functional advantages (Goodman et al., 2007; Seebach and Gardiner, 2008; Pan et al., 2017; Wang et al., 2019a). For example, foldamers may provide cavities that are adaptive in recognizing different guest molecules (Zhang et al., 2012; Yashima et al., 2016). Molecular tweezers have made the way from a supramolecular host to a drug candidate, due to their ability to inhibit peptide and protein aggregation through the complexation toward amino acids (Sinha et al., 2011; Schrader et al., 2016).

Isaacs and his co-workers created acyclic cucurbit[n]uril-type receptors, which can function as solubilizing agents for insoluble drugs. Interestingly, the solubility of paclitaxel was increased 2,750 times through the formation of soluble container–drug complex (Ma et al., 2012). These highly soluble acyclic cucurbiturils could also solubilize individual single-walled carbon nanotubes (SWNTs) in water even at a concentration 100–1,000 times lower than typically required for surfactants (Shen et al., 2012). The groups of Schrader and Yoshizawa synthesized beautiful water-soluble clip and tweezer-shaped hosts based on norbornene and anthracene building blocks (Bier et al., 2013; Jono et al., 2017).

Herein, we wish to report the synthesis of a new type of receptors, acyclic pillar[n]naphthalene (n = 2–4, Dimer, Trimer, and Tetramer) oligomers, which are made up of 2,3-diethoxynaphthalene units linked by methylene bridges at the 1- and 4-positions. Tetramer, bearing a pseudo-cavity, has good host-guest properties toward a series of model organic cationic guests.

All the reagents involved in this research were commercially available and used without further purification unless otherwise noted. ¹H NMR, ¹³C NMR, 2D NOESY, and COSY spectra (see Supplementary Material) were recorded with a Bruker AVANCE III 500 MHz instrument. Chemical shifts were referred to TMS. Highresolution mass spectra (HRMS) were determined on a Bruker Daltonics, Inc. APEXIII 7.0 TESLA FTMS instrument. The single crystal X-ray data were measured by direct methods using SHELXS-971 and refined by fullmatrix least-squares procedures on F2 with SHELXL-97.2. All non-hydrogen atoms were obtained from the difference Fourier map and subjected to anisotropic refinement by full-matrix least squares on F2. Hydrogen atoms were obtained geometrically and treated as riding on the parent atoms or were constrained in the locations during refinements. Test parameters and detailed experimental data are shown in the Supplementary Material.

2,3-Diethoxy naphthalene was selected as the building block to condense with paraformaldehyde. Due to the electron-donating ethoxy groups, great regioselectivity can be rationalized, and the reactive sites should be 1- and 4-positions in Friedel–Crafts reaction. It was expected to produce pillar-shape macrocycles, pillar[n]naphthalenes. However, no cyclic oligomers have been obtained after many attempts; a possible reason is that big naphthalene units make the final cyclization quite difficult due to the steric hindrance. Fortunately, we got acyclic pillar[n]naphthalenes (n = 2–4).

Using BF₃·(Et)₂O as the catalyst, the condensation reaction of 2,3-diethoxy naphthalene and paraformaldehyde in CHCl₃ at room temperature (Scheme 2) produced oligomers Dimer, Trimer, and Tetramer with yields of 21, 9, and 15%, respectively. Other Lewis acid catalysts, for example TfOH, FeCl₃, and AlCl₃, could also work, but the reaction yields were lower than that for BF₃·(Et)₂O. The synthesis was considerably easy since it just involved a one–step reaction of commercial starting materials and the isolation was also convenient by column chromatography on silica gel.

Dimer, Trimer, and Tetramer were well characterized by ¹H NMR, ¹³C NMR, NOESY, and COSY spectra (Figure 1 and Supplementary Figures 1–11), and high-resolution mass spectrometry (HRMS). They have rather complex patterns of aromatic and ethoxy peaks in ¹H NMR spectra (Figure 1) because they are not highly symmetrical macrocycles, but acyclic oligomers with low symmetry.

Single crystals of Dimer, Trimer, and Tetramer suitable for X-ray analysis were obtained by diffusion of hexane into a solution of the compounds in dichloromethane at room temperature (Figure 2). As expected, these three acyclic oligomers had the same connecting style, i.e., 2,3-diethoxy naphthalene units were connected by methylene at 1,4-positions. As shown in Figures 2A,B, the acyclic Dimer and Trimer, possessing two and three naphthalene moieties, have ill–defined cavities. Particularly, the Tetramer exhibits a pseudocycle–shaped structure, with all the methylene bridges being orientated outwardly. There exist intramolecular sextuple C–H···π interactions, with H···ring center distances of 2.75–3.23 Å (Supplementary Figure 12), between the middle two ethoxy groups and naphthalenes, resulting in the formation of a pseudo cycle rather than a zigzag structure. More interestingly, the single crystal structures of Tetramer molecules exist in a pair of enantiomers in the solid state (Figure 2D).

The host-guest properties of the acyclic receptors were then tested. Since they possess π-rich cavities, several cationic guests (1⁺-15⁺) and electron–deficient neutral guests (16–17) (Scheme 3) were chosen as model guest molecules to investigate their host-guest chemistry. In most cases, CDCl₃ was used as solvent during the ¹H NMR experiments of host-guest mixture and following NMR titrations; for guests 7²⁺, 9²⁺, and 10²⁺, CD₂Cl₂ was used because of their poor solubility in CDCl₃.

Figure 3 shows the ¹H NMR spectra recorded for quarternary ammonium guest 1⁺ in the absence and presence of Tetramer. As can be readily seen, upon addition of Tetramer, protons H_a, H_b and H_c of 1⁺ display substantial upfield shifts (Δδ = −0.39, −0.29, and −0.21 ppm) due to complexation–induced shielding effects, indicating that 1⁺ was located inside the acyclic host's pseudo-cavity to form a host-guest inclusion complex, and the main binding site is the N⁺(Me)₃ moiety. In contrast, protons H_h−i undergo indistinct NMR changes, suggesting they are located outside the cavity of Tetramer. [24] In the NOESY spectrum of 1⁺ and Tetramer, NOE correlations were observed between methyl protons H_a of the guest and the aromatic protons H₅, H₇ and H₈ of Tetramer, also suggesting the host-guest encapsulation (Supplementary Figure 13). The formation of 1⁺•Tetramer complex was further supported by ESI mass spectrometry analysis of an equimolar mixture of 1•BArF and Tetramer, where an intense peak for the 1:1 complex (m/z 1072.66, calcd. for 1⁺•Tetramer = 1072.67) was observed (Supplementary Figure 14). The encapsulation could also be rationalized by energy-minimized molecular modeling (Figure 3D): the oligomers wrapped around the guest to enhance the host-guest contacts driven by cation···π/ C–H···π interactions.

The addition of Dimer and Trimer could also induce the upfield shifts of guest 1⁺, but the Δδ values are smaller than those for Tetramer (Supplementary Figure 15). These results indicated relatively weak binding interactions occurred for Dimer and Trimer in comparison with Tetramer. These observations were consistent with the association constants (K_a) obtained from ¹H NMR titration experiments. As shown in Table 1, the K_a value of 1⁺ with Tetramer [(4.4±0.6) × 10² M⁻¹] is 18 times larger than that for Trimer, and the affinity for Dimer was too small to be accurately calculated (< 5 M⁻¹).

Since Tetramer showed interesting structure and good recognition behavior, we then examined its binding capacity toward other cationic guests (Table 1 and Supplementary Figures 16–29), revealing that Tetramer can form host–guest complexes with them but the binding affinities are totally different. For the trimethyl ammonium guests 1⁺-6⁺, 3⁺ [K_a = (1.2 ± 0.2) × 10³ M⁻¹] and 4⁺ [K_a = (2.1±0.4) × 10³ M⁻¹] bearing naphthyl moieties give stronger affinities, which should be due to host-guest fitted π ··· π interactions and large contacts. The substitution of naphthyl for smaller phenyl or bigger pyrenyl in 3⁺ and 4⁺, affording 1⁺ or 5⁺, considerably decreases the association constants by one order of magnitude.

Binding affinities of Tetramer toward primary ammonium guests 11⁺–13⁺ were stronger than those of the corresponding quaternary ammonium guests 1⁺–3⁺. For example, the K_a value of Tetramer and octylammonium 11⁺ [(2.5±0.4) × 10⁴ M⁻¹] is about 56-fold higher than that for trimethyloctylammonium 1⁺ [(4.4±0.6) × 10² M⁻¹]. Similarly, the selectivity factors of 12⁺/2⁺ and 13⁺/2⁺ are 15 and 12, respectively. The reason for such high selectivity would be that big and spherical N⁺(Me)₃ group is too larger compared with Tetramer's size, and small NH3+ is a suitable one. It should be noted that the binding affinities of Tetramer and organic ammonium salts, with K_a values in the magnitude of 10²-10⁴ M⁻¹, are comparable to those for macrocyclic arenes such as pillararenes and biphenarenes.

Due to its π-electron rich cavity, the complexation of Tetramer and two π-deficient neutral guests, 16 and 17, were also investigated. From Supplementary Figures 28, 29, no obvious NMR changes were detected, indicating no stable complexes can be formed.

In summary, acyclic pillarnaphthalenes with 2,3-diethoxynaphthalene units bridged by methylenes at 1,4-positions were synthesized through a one-pot reaction of 2,3-diethoxy naphthalene monomer and paraformaldehyde by using Lewis acid as the catalyst. Acyclic pillar[4]naphthalene Tetramer is able to interact organic ammonium guests cations by wrapping around them, giving association constants in the magnitude of 10²-10⁴ M⁻¹. We expect that Tetramer bearing pseudo-cycle cavity, could have significant potential for the applications in host-guest chemistry and self-assembly.

All datasets generated for this study are included in the article/Supplementary Material.

CL, BW, and YJ conceived this project and designed the experiments. YJ and MD contributed to most of the experimental work. CL, MD, and BW co-wrote the paper. All authors discussed and commented on the paper and analyzed the data.