Introduction

Non-covalent interactions are essential for sustaining life and performing ubiquitous biological functions1. Among different non-covalent interactions, π interactions form a significant category, describing the interactions between aromatic molecules/fragments and nearby chemical groups. Distinguished by the interacting moieties, π–π2, XH–π3,4, cation–π5,6, anion–π7,8, and lone-pair–π interactions9,10,11 have garnered substantial research interests in the past few decades. By strategically manipulating these π interactions, scientists are able to craft intricate supramolecular structures12,13,14,15, achieve precise molecular recognition16, and produce catalysts with high efficiency that mimic enzymatic functions17. The identification of different intermolecular forces creates a strong foundation for innovations within the field of supramolecular chemistry.

Intermolecular interactions are often discovered through gas phase studies or by examining known protein structures18,19. The structural features are then corroborated through energy analyses and spectroscopic characterizations to confirm the existence of interactions. Given that many discoveries in this field have been protein-centric, the moieties involved in known π-interactions are frequently observed in biological systems. Examples include aromatic rings, cations, and halogen anions20,21,22. While biological molecular structures have significantly contributed to our understanding of intermolecular interactions, they may not fully reveal the existence of interactions involving less common moieties in nature.

An example of such an overlooked component is the alkyne group. Alkyne groups are not typically found in life-forming molecules, such as amino acids, nucleic acids, and electrolytes. This absence is supported by the “Raman-silent spectral window of cells (1800–2800 cm−1)23, where the characteristic Raman signal of the C≡C stretching vibration is located”. The relative invisibility of alkyne groups in biological systems may have led to their underrepresentation in studies of supramolecular interactions. In the literature, few structural descriptions hint at the possible interactions involving alkyne groups. For instance, Siegel et al.24 described a scenario where “one radical benzene ring sits atop the acetylene of its neighbor”, and Sakamoto et al.25 noticed that “an alkyne from an adjacent monomer overlaps with each anthracene”. However, none of these studies definitively validate the existence and properties of alkyne-involved intermolecular interactions.

Meanwhile, despite their scarcity in cellular components, alkyne-containing compounds play a pivotal role in biological and medical sciences, as the distinctive C≡C stretching vibration of the probe molecules can be tracked to monitor specific biological processes. For example, alkyne-tagged probes are widely studied for visualizing proteins, nucleic acids, glycans, and other small metabolites, with stimulated Raman scattering (SRS) microscopy26,27. Researchers have dedicated to develop protocols and strategies, for enhancing signal resolution and enabling multi-channel imaging28. These efforts lead to a diverse library of probes with C≡C stretching signals covering a broad spectrum ranging from 2017 to 2262 cm-1. Meanwhile, besides their utility in bio-imaging, alkyne-containing molecules are also integral to numerous commercially available medications29. Notable examples include Mifepristone, used for contraception, and Selegiline, which is prescribed for the treatment of depression, Parkinson’s, and Alzheimer’s diseases. Given the extensive applications of alkyne-containing molecules in complex biological systems from bioimaging to biomedicine, it is crucial to investigate the intermolecular interactions involving alkyne groups and understand their impact from both chemical and physical perspectives.

In this work, we compare two structurally analogous compounds, Red-1 and Orange-1, containing the same metal-organic cage {[Cp3Zr3(μ3-O)(μ2-OH)3]2L3}Cl2. By focusing on the interactions between alkyne groups and aromatic rings, we examine their effects on the chemical and physical properties while minimizing the differences introduced by other intermolecular interactions. Experimental evidence and theoretical analysis both suggest that alkyne-π interactions are a genuine intermolecular interaction, pivotal in determining the spectroscopic properties of the compounds. This makes the alkyne-π interaction an intriguing and functional intermolecular interaction.

Result and discussion

Structure description of Red-1 and Orange-1 and identification of an alkyne-π interaction in Red-1

During the synthesis of metal-organic cages (MOCs) with trinuclear zirconium-oxo cluster nodes {Cp3Zr3(μ3-O)(μ2-OH)3} (where Cp = C5H5-), we obtained two structurally similar compounds. It is well-known that zirconocene MOCs can be synthesized by hydrolyzing zirconium bis(cyclopentadienyl) dichloride (ZrCp2Cl2)30,31 (Fig. 1a). Here, we utilized 4,4’-(anthracene-9,10-diylbis(ethyne-2,1-diyl)dibenzene carboxylic acid (H2L) as the ligand, featuring alternating aromatic rings and alkyne groups (Fig. 1b). Upon crystallization, a capsule-like MOC, {[Cp3Zr3(μ3-O)(μ2-OH)3]2L3}2+ ((Cp3Zr3)2L3-MOC) crystallizes with Cl- counterions. Crystallization below 30 °C yields red block-shaped crystals {[Cp3Zr3(μ3-O)(μ2-OH)3]2L3}Cl2, referred to as Red-1, while crystallization above 60 °C result in orange-colored crystals {[Cp3Zr3(μ3-O)(μ2-OH)3]2L3}Cl2, labeled as Orange-1.

Fig. 1: The assembly process, structures, and morphologies of Red-1 and Orange-1 crystals.
figure 1

a Chemical structure of the ZrCp2Cl2. b Chemical structure of the ligand H2L. c Chemical structure of the MOCs. Zr, blue sphere; O, red sphere; C, gray sphere; H, light gray sphere. Counterions (Cl-) and solvent molecules are omitted for clarity. d Top view of the cluster node. e–g the (Cp3Zr3)2L3-MOC dimer structure, lattice packing, and high-resolution electrospray ionization mass spectrum of Red-1. The mass spectrum peak of 1,231.9768 corresponds to [M − 2Cl]2+ (M = {[Cp3Zr3(μ3-O)(μ2-OH)3]2L3}Cl2, Red-1), exptl m/z = 1,231.9768; calcd m/z = 1,231.9949. h–j the MOC dimer structure, lattice packing, and high-resolution electrospray ionization mass spectrum of Orange-1. The mass spectrum peak at 1,231.9740 correspond to [M − 2Cl]2+ (Orange-1), exptl m/z = 1,231.9740; calcd m/z = 1,231.9949.

Full size image

Single-crystal X-ray diffraction (SCXRD) revealed that both Red-1 and Orange-1 crystallize in the P\(\bar{1}\) space group, with comparable cell volumes (Supplementary Table 1). Each structure is composed of (Cp3Zr3)2L3-MOCs (Fig. 1c), consisting of two {Cp3Zr3(μ3-O)(μ2-OH)3} nodes linked by three 4,4’-(anthracene-9,10-diylbis(ethyne-2,1-diyl)dibenzene carboxylate) ligands (L) (Fig. 1b, d). X-ray Photoelectron Spectroscopy (XPS) analysis confirms that the oxidation state of Zr is tetravalent (Supplementary Fig. 1) in both compounds. Within the (Cp3Zr3)2L3-MOC, two of the three L ligands have their aromatic groups almost parallel to each other, while the aromatic ring of the third ligand oriented at an angle to the other ligands. Two (Cp3Zr3)2L3-MOCs further stack to form dimers (Fig. 1e, h), which assemble into the overall structure (Fig. 1f, i). By dissolving Red-1 and Orange-1 in methanol, we confirm that the same (Cp3Zr3)2L3-MOCs are the major species in both solutions by high resolution mass spectroscopy (Fig. 1g, j). These structural and compositional analyses indicate that Red-1 and Orange-1 have the same building units.

Although their building units are the same, Red-1 and Orange-1 intriguingly show different colors. The differences are likely caused by variations in packing and intermolecular interactions. A close examination of the (Cp3Zr3)2L3-MOC dimers reveals that each (Cp3Zr3)2L3-MOC interacts with its neighbor primarily through ligand-ligand interactions (Fig. 2b, g). In each (Cp3Zr3)2L3-MOC, the three L ligands are crystallographically distinct (color-coded in Fig. 2: L1-yellow, L2-rose red, and L3-blue) (Fig. 2a, f). L1 and L2 are parallel to each other, while L3 forms a certain angle with both L1 and L2. Because an initial analysis of the angles and distances between these ligands within the molecular cage does not immediately clarify their interactions (Supplementary Figs. 2, 3 and Supplementary Table 2), we proceed with a detailed geometric analysis32. We gain a deeper understanding of π-π interactions by quantifying the interplanar distance (di), π slipping distance (ds), and π-overlap area (Soverlap). For the definitions of these three parameters, please refer to S2.4 of the Supplementary Information. The analysis process consists of three steps:

Fig. 2: Molecular geometric analysis of intra-cage and inter-cage packing in Orange-1 and Red-1, listing the interplanar distance (di), π slipping distance (ds), and π-overlap area (Soverlap) of aromatic rings.
figure 2

a, b the monomer and their stacked structures in Orange-1. c π-π stacking of L1-L2 in Orange-1 cage monomers. d π-π stacking of L1-L1 in the Orange-1 dimer. e the stacking of L3-L3 between dimers in Orange-1. f–h the monomer and their stacked structures in Red-1. h π-π stacking of L1-L2 in Red-1 cage monomers. i π-π stacking of L1-L1 in the Red-1 dimer. j the stacking of L3-L3 between dimers in Red-1. Note: Four consecutive alkyne-π stackings are observed between L1 ligands in (i).

Full size image

First, the (Cp3Zr3)2L3-MOC monomers of Red-1 and Orange-1 exhibit similar structural characteristics. L1 and L2 appear to interact through offset stacked π-π interactions, while L3 interacts with L1 and L2 via edge-to-face stacking. However, our study focuses on analyzing face-to-face stacking and offset stacking interactions. Therefore, the di and Soverlap between L1 and L2 are 3.39 Å and 7.65 Ų, for Orange-1 (Fig. 2c and Supplementary Fig. 4), and the corresponding values for Red-1 are 3.42 Å and 9.35 Ų (Fig. 2h).

Second, the two (Cp3Zr3)2L3-MOCs monomers of Red-1 and Orange-1 form dimers through L1-L1 interactions. In Orange-1, the two L1 ligands form π-π interactions, with the di and Soverlap being 3.39 Å and 4.68 Ų, respectively (Fig. 2d). In contrast, there is no significant overlap between the aromatic rings of L1 ligands in Red-1 (Fig. 2i), even though they are separated by a similar distance of 3.40 Å as in Orange-1.

Third, the stacking arrangement between the dimers of Red-1 and Orange-1 is different. In Orange-1, the dimers form π-π interactions with the neighboring ones via the aromatic rings on L3 ligands, with the di and Soverlap being 3.30 Å and 5.50 Ų, respectively (Fig. 2e). Conversely, in Red-1, the di and Soverlap between the two L3 ligands are 3.37 Å and 0.38 Ų, respectively (Fig. 2j), indicating much weaker π-π slip-stacking interactions.

The above assessments suggest that Orange-1 exhibits stronger π-π interactions than Red-1. Specifically, the intra-dimer and inter-dimer Soverlap values of Orange-1 are 19.98 (7.65 × 2 + 4.68) and 5.50 Å2, larger than the corresponding Soverlap (18.70 (9.35 × 2) Å2 and 0.38 Å2) of Red-1. It is well-known that molecular packings involving more effective π-π interactions often lead to a more significant red-shift in light absorption and emission32,33. However, Red-1, with overall less π-π interactions, exhibits a more red-shifted color in appearance. Such a puzzle raises the possibility that other interactions may contribute to the color of the crystals.

Validation of the alkyne-π stacking interactions

After further examining the ligand interactions (Fig. 2i), we find an intriguing pattern of intra-dimer interactions in Red-1. Particularly, the two neighboring L1 ligands are nearly parallel, with the dihedral angle smaller than 1.1° and the distance within 3.50 Å (Supplementary Fig. 5). Besides, the alkyne in one L1 ligand stacks with the aromatic rings in the other L1 ligand, and four pairs of such alkyne-aromatic-ring stackings occur consecutively between the two L1 ligands. By sharp contrast, the alkyne-alkyne and alkyne-aromatic-ring distances in Orange-1 all exceed 4.60 Å (Supplementary Fig. 6), beyond the typical range for considering intermolecular interactions34. Although crystal packing factors like symmetry and solvent molecules may influence these molecular arrangements, the stacking patterns of Red-1 and Orange-1 are distinctly different. We thus propose that unique interactions between the alkyne and aromatic rings in Red-1 contribute to its counterintuitive red-shifted color.

To examine the existence of such alkyne-π interactions, we performed comprehensive spectroscopic studies on the C≡C bonds in both Red-1 and Orange-1 using phase-pure polycrystalline powders (XRD patterns in Supplementary Fig. 7). Fourier Transform Infrared (FTIR) spectroscopy (Fig. 3a and Supplementary Fig. 8a) reveals a characteristic C≡C stretching signal at 2195 cm−1 for Orange-1. For Red-1, the corresponding signal is red-shifted to 2190 cm−1, indicating that the C≡C bonds in Red-1 are indeed distinct. Given the intrinsic weakness of the C≡C stretching signal in FTIR spectra35, we also characterized the samples using Raman spectroscopy (Fig. 3b and Supplementary Fig. 8b). Importantly, we observed a unique C≡C stretching signal at 2182 cm−1 for Red-1, clearly red-shifted from the normal C≡C stretching peak at 2190 cm−1 in both samples. This interesting observation suggests that a portion of C≡C bonds are influenced by interactions only present in Red-1, i.e., those stacked with the aromatic rings. The red-shifted signals in FTIR and Raman spectra provide direct evidence to support the existence of alkyne-π interactions36,37.

Fig. 3: Spectroscopic verification of alkyne-π interactions.
figure 3

a Fourier Transform Infrared (FTIR) spectroscopy. The inset shows the vibrational signals of the C≡C bonds. b Raman spectroscopy. The inset shows the vibrational signals of the C≡C bonds. The C≡C stretching signal of the Red-1 crystal is composed of two components, one matching Orange-1 and one not. c 13C Solid-State Nuclear Magnetic Resonance spectroscopy. The inset provides a magnified view of the C≡C bond region (60-100 ppm). The C≡C signals of the Red-1 crystal are composed of two components.

Full size image

We further characterize the chemical environment of the C≡C bonds by 13C Solid-State Nuclear Magnetic Resonance (13C SSNMR). Although the signals in the 13C SSNMR spectra are relatively broad, they are distinguishable for identifying different carbon atoms (Fig. 3c and Supplementary Fig. 8c). Red-1 and Orange-1 both exhibit clear signals corresponding to the carboxylate groups (170.0 to 190.0 ppm) and the aromatic rings (100.0 to 150.0 ppm). In the alkyne region (60.0 to 100.0 ppm), Orange-1 exhibits only one broad peak at 96.0 ppm, corresponding to the alkyne groups that are not involved in intermolecular interactions. Red-1, on the other hand, exhibits two peaks at 97.0 ppm and 76.0 ppm. The upfield signal at 76.0 ppm results from the shielding effect of the aromatic ring current38, confirming the alkyne groups are involved in the alkyne-π stacking interactions. These findings align well with the FTIR and Raman spectroscopic results, lending credence to the assertion that the alkyne-π interaction is a true intermolecular interaction.

Effect of alkyne-π interactions

Having confirmed the presence of alkyne-π stacking interactions, we proceed to investigate their influence from three key aspects, including energy analysis, vibrational spectroscopy, and light absorption/emissions.

Computational analysis on the energetics

To estimate the energy contributions from alkyne-π interactions, we conduct a detailed analysis of the interaction energies within the dimers, mainly focusing on the seven pairs of π-π and alkyne-π interactions between the adjacent L1 ligands as depicted in Fig. 4a (for the computational details, please refer to the Supplementary Information and Supplementary Data 1). Notably, although the relative positions and orientations of two L1 ligands in the Orange-1 and Red-1 dimers are different (Fig. 4b), the π-π interactions are consistently observed to be more robust than their alkyne-π counterparts (Supplementary Figs. 9 and 10). However, a discernible difference emerges when we closely examine the relative strength of these interactions in each dimer. To quantify these variations, we calculated ΔΔEInteraction, defined as the interaction energy between the two L1 ligands in Orange-1 minus that in Red-1 (Fig. 4c). In this definition, a positive ΔΔEInteraction value indicates that the interaction is stronger in Red-1, whereas a negative value indicates a stronger interaction in Orange-1. Based on this metric (Supplementary Table 3), our results show that π-π interactions contributed by anthracene-anthracene and phenyl-phenyl interactions in the Orange-1 dimer are stronger than those in the Red-1 dimer (Fig. 4c). On the contrary, the alkyne-π interactions consisting of alkyne-phenyl and alkyne-anthracene interactions in the Red-1 dimer are stronger than in the Orange-1 dimer (Fig. 4c). Therefore, transitioning from the Orange-1 dimer to the Red-1 dimer, the predominant alterations in the intra-dimer interactions involve a reduction in π-π interactions and a concomitant enhancement of alkyne-π interactions. Hence, our analysis reveals that, although alkyne-π interactions are energetically less significant compared to π-π interactions, they elaborate the main difference in the intra-dimer interactions in the Red-1 dimer relative to the Orange-1 dimer.

Fig. 4: Energy calculations provide critical insights into the energy contribution of π-π and alkyne-π interactions.
figure 4

a Calculations of the π-π and alkyne-π interactions between the adjacent L1 ligands in dimeric units. Seven pairs of interactions are numbered as shown in the inset and their energy values are plotted accordingly (Numbers 1 and 3 represent the π–π interactions between benzene rings; number 2 represents the π–π interactions between anthracene rings; numbers 4 and 7 represent the alkyne–π interactions between alkyne groups and benzene rings; numbers 5 and 6 represent the alkyne–π interactions between alkyne groups and anthracene rings). b Diagram showing the relative positions and orientations of two L1 ligands in Red-1 and Orange-1 dimers. c Variations of π-π and alkyne-π interaction energies from Orange-1 to Red-1 dimer.

Full size image

Spectrum-changing: Part 1 - On vibrational spectroscopy

Alkyne-π interaction clearly has an effect on vibration spectroscopies. In Section 2, we showed that the C≡C stretching signal in Red-1 shifts to lower wavenumbers in both FTIR and Raman spectra. The red-shift, observed within a range of 10 cm−1, suggests a minor reduction in the C≡C bond strength. As we know, the red-shifting vibrational signals have been extensively studied in the context of “back-bonding” from transition metals to CO molecules in carbonyl complexes39. Similar behaviors have been observed in the C≡N vibrational signal when interacting with the Ag nanoparticles in a “side-on” configuration40. Therefore, the red-shifted C≡C vibration signal in Red-1 is likely caused by the donation of π electrons from aromatic rings to the anti-bonding orbitals of the alkynes, which is a direct consequence of the alkyne-π interaction.

Spectrum-changing: Part 2 - on light absorption and emission

Finally, the change in light absorption and emission naturally falls within our interest. The UV-Vis spectra of the crystalline powders of the two compounds show different absorption ranges. For Orange-1, absorption takes place from 200 nm to 580 nm, while Red-1 absorbs from 200 nm to 620 nm. In the visible region, the absorption edge of Red-1 is clearly red-shifted from that of Orange-1 (Fig. 5a), leading to their color differences. We use the Tauc plot to determine the bandgap from UV-Vis data41. The fitting of (αhν)1/n = B(hν - Eg) indicates that n = 2, confirming that both materials are direct bandgap semiconductors. Specifically, Orange-1 exhibits a bandgap of 2.14 eV, whereas Red-1 has a bandgap of 2.09 eV, with a difference of 0.05 eV (Supplementary Fig. 11). This variation is mainly attributed to the alkyne-π interactions in Red-1, which cause a reduction in the optical bandgap and a red shift in the absorption spectrum. Meanwhile, we find that the methanol solution of Red-1 and Orange-1 exhibit high similarity in the UV and luminescent spectra (Fig. 5b) (see discussion in the Supplementary Information and Supplementary Figs. 12–15 and Supplementary Table 4). This suggests that the differences in solid-state stacking structures, in particular the alkyne-π interactions, is likely the primary cause of the absorption red-shift in Red-1 crystals.

Fig. 5: Effects of alkyne-π interaction on absorption and emission.
figure 5

a Fluorescence spectrum of the methanol solution of Red-1/Orange-1 (λex = 410 nm). The inset shows the appearance of the solutions under 365-nm UV exposure. b UV-visible absorption spectra of the polycrystalline powders, indicating the red-shifted absorption edge of Red-1. c Fluorescence emission spectrum of the polycrystalline powders (λex = 410 nm), with a pronounced red-shift in Red-1. The inset displays the photographs (under 365-nm UV exposure). d Fluorescence lifetime analysis, fitted with a double exponential decay model. e Visualization of the highest occupied molecular orbital (HOMO) of Red-1 and Orange-1.

Full size image

We further studied the fluorescent emission properties of the two compounds. Under 410 nm excitation, Orange-1 exhibits maximum emission at 600 nm (with a quantum yield PLQY of 0.56%), while the maximum emission peak of Red-1 is located at 640 nm (PLQY = 8.60%). Therefore, the “red-shift” characteristic is again observed in the fluorescent emission of Red-1, accompanied by a 15-fold increase in PLQY (Fig. 5c). In addition, time-resolved fluorescence spectra are fitted with a double exponential decay model42, yielding an average lifetime of 10.36 ns for Red-1 (Fig. 5d). This fluorescence lifetime is about twice as long as that of Orange-1 (5.84 ns). With the higher PLQY and longer lifetime, the nonradiative rate (knr = (1-PLQY)/τ) of Red-1 is calculated to be 0.088 ns−1, clearly lower than that of Orange-1 (knr = 0.38 ns−1)43,44. As mentioned, increasing π-π overlap typically induces a more pronounced red-shift in the optical properties32,33,45. However, despite the more significant π-π interactions in the Orange-1, its optical spectra, counterintuitively, exhibit less red-shifting when compared to Red-1. Since the primary distinction between Orange-1 and Red-1 is identified as a decrease in π-π interactions and a simultaneous strengthening of alkyne-π interactions in the latter, we hypothesize that the alkyne-π interactions may predominantly govern the red-shift in absorption and emission properties observed in Red-1.

To verify this conjecture, we investigate the electronic structures of Orange-1 and Red-1 in the crystal environment by DFT calculations. First, we find that Red-1 has a narrower thermal bandgap than Orange-1, mainly because the conduction band shifts to a lower energy region (Supplementary Fig. 16 and Supplementary Table 5). In accordance with the bandgap, computations also suggest a smaller HOMO-LUMO gap in Red-1. These results are consistent with the UV-Vis data, indicating that our computational models can reproduce the spectral features observed in the experiment. To further decipher the molecular origin of the red-shift observed in the spectra, we examine the orbital distribution of frontier molecular orbitals over the molecular fragments46. We focus on the regions of L1 ligands, as they are essential for determining the interactions between (Cp3Zr3)2L3-MOCs. Interestingly, the HOMOs of Red-1 and Orange-1 show clear discrepancies. In particular, in Red-1, alkyne groups in the L1 ligand on one monomer appear to form “bonding interactions” with the anthracene rings of the L1 ligand on the other monomer (Fig. 5e). However, such interactions are not observed in Orange-1. This discrepancy between the HOMOs of Red-1 and Orange-1 helps us rationalize the smaller HOMO-LUMO gap in Red-1 and offers molecular insights into the red-shift observed in the spectra.

It is worth noting that the effect of alkyne-π interactions on light absorption/emission can be extended to other phases beyond Red-1. Particularly, by controlling the synthetic temperature at 0 °C, a new phase (Red-2) is successfully obtained (Supplementary Fig. 17a). Meanwhile, by dissolving Red-1 in methanol followed by evaporation, we obtain another new compound with orange color (Orange-2) (Supplementary Fig. 17b, c). The detailed analyses of the intermolecular interactions in Red-2 and Orange-2 are shown in the Supplementary Discussions and Supplementary Figs. 18 and 19. Interestingly, alkyne-π stacking interactions are also observed in Red-2 but not in Orange-2. These observations further corroborate the significant influence of the alkyne-π interactions on light absorption and emission.

Exploring the broader impacts of alkyne-π interaction

Upon discovering the existence of alkyne-π stacking interaction in our newly prepared functional materials, we conducted a structural search in the Cambridge Crystallographic Data Centre (CCDC). Three parameters are defined for this analysis: α is the angle between the C≡C bond axis and the ring normal; θ represents the angle between the ring normal and a vector connecting the C≡C bond center and the ring center; and d is the distance between the C≡C bond center and the ring center (Fig. 6a). As illustrated in the heatmap (Fig. 6b), C≡C bonds predominantly orient parallel to the aromatic ring (with α ~ 90°, perpendicular to the ring normal). After zooming in, two hotspots are identified, one at ~3.5 Å and the other at ~5.0 Å (Fig. 6c). By plotting these hits against θ (Fig. 6d), we observe that the 3.5 Å hotspot remains at θ near 0°. This 3.5 Å hotspot closely aligns with the alkyne-π stacking interaction we described here, representing a total of 436 structures (Supplementary Figs. 20a, 20b, Supplementary Table 6, and Supplementary Data 2), e.g., MEFREX47 and CIMKOB48 (Fig. 6e and Supplementary Figs. 21, 22). On the other hand, the 4.3 Å hotspot in Fig. 6d, including a total of 1704 structures (Supplementary Figs. 20c, 20d, Supplementary Table 7, and Supplementary Data 2), e.g., BUGVIK49 (Fig. 6e and Supplementary Fig. 23), are attributed to the parasitic stacking geometry caused by slip-stacking π-π interactions, which out-compete the alkyne-π interactions. These findings unveil the alkyne-π stacking as a subtle but characteristic interaction mode.

Fig. 6: Exploration of alkyne-π stacking interactions using the Cambridge Crystallographic Data Centre.
figure 6

a Three parameters, α, θ, and d, are defined for data retrieval and analysis. b Heat map (α 0°~90°; θ 0°~90°; d 3~5 Å) with hotspots near α at 90°. c Heat map with α restricted to 80°~90°, showing hotspots located at ~3.5 Å and ~5.0 Å. d Heat map relating d with θ (α = 80°~90°), showing the hotspot at 3.5 Å persists (highlighted by the red dashed line). This demonstrates the alkyne-π stacking as a characteristic interaction pattern. e MEFREX and CIMKOB represent structures from the 3.5 Å hotspot in (d), while BUGVIK represents a structure from the 4.3 Å hotspot.

Full size image

Regarding the “spectrum changing” effect, alkyne-π interactions may affect the resolution of alkyne-tagged probes during bio-imaging. In particular, previous researchers have reported a series of alkyne-tagged imaging probes (i.e., Carbow)28, with the C≡C vibration signals differing by only 5 cm−1. The “high resolution” of alkyne-tagged probes could be diminished by alkyne-π interactions, because we found that alkyne-π interactions could lead to a red-shift of the Raman signal up to ~10 cm−1 (Fig. 3b). Therefore, caution is needed when applying alkyne-tagged probes in complex biological systems, where the probes might undergo substantial alkyne-π interactions with biomolecules.

Furthermore, we notice a recent report on the successful synthesis of polyyne polyrotaxanes with multiple threaded macrocycles50. For unthreaded polyynes, the C≡C stretching signals are observed at 1944 cm−1 (compound C28) (Supplementary Fig. 24a) and 1913 cm−1 (compound C48) (Supplementary Fig. 25a). After threading through the macrocycles, the corresponding signals red-shift to 1940/1940 cm−1 (compound C28·(Ma)2/ C28·(Mb)2) (Supplementary Fig. 24b) and 1909/1907 cm−1 (compound C48·(Ma)3/ C48·(Mb)3) (Supplementary Fig. 25b). Based on the single crystal structure of similar compounds reported by the same paper (Supplementary Fig. 26), we speculate that this red-shift might be caused by alkyne-π stacking. Therefore, the red-shifted C≡C Raman signal could serve as a signature of the successful synthesis of polyyne-threaded macrocycles.

Finally, it is worth noting that although the existence of alkyne-π interaction is evident in the solid state, we haven’t found a solvent that seems to maintain the alkyne-π interaction. Therefore, the presence of this interaction in the solid state might not guarantee its persistence after entering the solution. We are currently still exploring the right solvent that might maintain this intermolecular interaction.

Conclusion

In summary, we systemically analyze two zirconocene metal-organic cage compounds and unveil the existence of alkyne-π interactions. The stacking of alkyne groups directly above aromatic rings is recognized as the unique structural feature, with spectroscopic studies indicating the changing C≡C bond characters because of the alkyne-π interactions. Despite being energetically less potent than π-π interactions, alkyne-π interactions significantly influence the spectroscopic properties of the compounds, as evidenced by red-shifting effects observed across vibrational, absorption, and emission spectra. Together, our findings reveal the role of alkyne-π interactions in modulating the spectroscopic properties with functional relevance. This research expands our understanding of intermolecular forces and highlights the role of alkyne-π interactions in developing functional materials for advanced applications in chemistry and biology. It also underscores the importance of considering subtle yet impactful interactions in the development of supramolecular chemistry.

Methods

Starting materials

All reagents, including bis(cyclopentadienyl)zirconium dichloride and 4,4’-(anthracene-9,10-diylbis(ethyne-2,1-diyl))dibenzoic acid used in this study, were purchased from Shanghai Bide Pharmatech Co., Ltd. and used without further purification. All solvents were obtained from Sinopharm Chemical Reagent Co., Ltd. and used directly without additional purification.

Synthesis of {[Cp3Zr3(μ 3-O)(μ 2-OH)3]2L3}Cl2 (Cp = η 5-C5H5) (Red-1)

Bis(cyclopentadienyl)zirconium dichloride (ZrCp2Cl2, where Cp = η5-C5H5) (32.6 mg, 0.112 mmol) and 4,4-(anthracene-9,10-diylbis(e-thyne-2,1-diyl))dibenzoic acid (H2L) (4.1 mg, 0.00879 mmol) were dissolved in DMA/THF mixed solvent (2.0 mL, v/v = 1/1). Shake to dissolve and add distilled water (190 μL). The mixture was ultrasonicated at 0 °C for 30 minutes followed by volatilization at 30 °C to obtain red crystals (Red-1). Yield: 59% (based on Zr). IR (2000-400 cm−1): 3578 (w), 2190 (w), 1576 (m), 1523 (m), 1406 (s), 1176 (s), 1064 (w), 1016 (m), 809 (s), 778 (s), 761 (s), 614 (s), 591 (m). HRMS (m/z): [M − 2Cl]2+ (M = {[Cp3Zr3(μ3-O)(μ2-OH)3]2L3}Cl2), found m/z = 1231.9768; calculated m/z = 1231.9949. 1H NMR spectrum for Red-1 (500 MHz, CD3OD) δ6.57-6.63 (s, 10H), 6.64-6.69 (m, 4H), 7.42-7.48 (d, 4H), 7.71-7.76 (d, 4H), 7.97-8.01 (m, 4H).

Synthesis of {[Cp3Zr3(μ 3-O)(μ 2-OH)3]2L3}Cl2 (Cp = η 5-C5H5) (Orange-1)

Bis(cyclopentadienyl)zirconium dichloride (ZrCp2Cl2, where Cp = η5-C5H5) (32.6 mg, 0.112 mmol) and 4,4-(anthracene-9,10-diylbis(e-thyne-2,1-diyl))dibenzoic acid (H2L) (4.1 mg, 0.00879 mmol) were dissolved in DMA/THF mixed solvent (2.0 mL, v/v = 1/1). Shake to dissolve and add distilled water (190 μL). The mixture was ultrasonicated at 0 °C for 30 minutes followed by volatilization at 60 °C to obtain orange crystals (Orange-1). Yield: 68% (based on Zr). IR (2000–400 cm−1): 3586 (w), 2195 (w), 1576 (m), 1523 (m), 1404 (s), 1176 (s), 1066 (w), 1016 (m), 809 (s), 778 (s), 761 (s), 614 (s), 591 (m). HRMS (m/z): [M − 2Cl]2+ (M = {[Cp3Zr3(μ3-O)(μ2-OH)3]2L3}Cl2), found m/z = 1231.9740; calculated m/z = 1231.9949. 1H NMR spectrum for Orange-1 (500 MHz, CD3OD) δ6.56–6.63 (s, 10H), 6.63-6.69 (m, 4H), 7.43-7.48 (d, 4H), 7.71-7.76 (d, 4H), 7.96-8.02 (m, 4H).

Synthesis of {[Cp3Zr3(μ 3-O)(μ 2-OH)3]2L3}Cl2 (Cp = η 5-C5H5) (Red-2)

The synthesis of the reaction mixture is identical to that of Red-1. The reaction mixture is subjected to ultrasonic treatment at 0 °C for 30 minutes, followed by evaporation at 0 °C to obtain red block-shaped crystals (Red-2).

Synthesis of {[Cp3Zr3(μ 3-O)(μ 2-OH)3]2L3}Cl2 (Cp = η 5-C5H5) (Orange-2)

Dissolve the dry Red-1 crystals (10.0 mg) in methanol (50.0 mL), noting that the crystals are only slightly soluble. After filtering and sealing the solution, let it stand at room temperature to slowly evaporate over 3 months, yielding orange rhombic crystals (Orange-2).