Crystal structure of di­acet­y­lene unveiled by X-ray and neutron diffraction, Raman spectroscopy and periodic DFT

Combining powder X-ray and neutron diffraction, the first crystallographic characterization of di­acet­y­lene (IUPAC designation 1,3-butadiyne) is presented and supplemented by Raman spectroscopy and periodic-DFT calculations. The structure is described in the space group Pnma, with a = 9.348 (2), b = 5.9890 (6), c = 5.6746 (11) Å and V = 317.69 (6) ų at 5 K, with four molecules in the unit cell resulting in a density of 1.0466 g cm⁻³. The di­acet­y­lene molecules are arranged in a layered structure dominated by C—H⋯π interactions, which leads to anisotropic thermal expansion behaviour. No structural phase transitions were observed across the 5–220 K tem­per­a­ture range at ambient pressure. The structural similarity with acetylene identifies it as a potential cocrystal component of particular importance for Titan's surface chemistry in view of the upcoming NASA Dragonfly mission.

1. Introduction

The crystallographic study of simple hydrocarbon species is of great importance for understanding molecular solids, particularly of those relevant to planetary systems (Maynard-Casely et al., 2018; Maynard-Casely et al., 2020; Vu et al., 2022; Brand et al., 2020). Investigating the phase behaviour, density and structural properties of these molecules is key to elucidating the physical processes and geochemical features observed in various bodies across the Solar System.

Saturn's moon Titan has drawn significant attention in recent years due to its dense atmosphere, whose composition is a consequence of photochemical reactions initiated from its primary components: nitrogen and methane (Coustenis et al., 2007; Hörst, 2017; Nixon, 2024). These reactions result in complex chemistry with over 20 molecular species identified in the upper atmosphere (Nixon, 2024). At lower altitudes, the reduced tem­per­a­tures facilitate the condensation and eventual precipitation of heavier hydrocarbons and nitrile molecules, such as acetylene (C₂H₂), hydrogen cyanide (HCN), benzene (C₆H₆) and 1,3-butadiyne (C₄H₂, hereafter referred to as di­acet­y­lene) onto Titan's surface, where the tem­per­a­ture varies in the range 90–94 K (Barth, 2017).

Surface measurements from the Cassini–Huygens mission support the deposition of these small condensates, with images revealing a complex landscape of lakes and dunes where sediments accumulate over time (Lopes et al., 2019). The fate of these compounds is of particular interest to NASA's New Frontiers mission Dragonfly, scheduled to arrive in the 2030s, to investigate the surface composition of Titan (Maynard-Casely et al., 2018; Barnes et al., 2021; Cable et al., 2021; Yu et al., 2023; Yu et al., 2024).

Among these condensates, di­acet­y­lene is potentially sig­nifi­cant in contributing to Titan's atmospheric processes, where its abundance is estimated to be 15 ppb at 150 km altitude (Jolly et al., 2010). It is thought to be formed from the photodissociation of acetylene (Smith et al., 1998):

Once formed, di­acet­y­lene acts as a precursor for higher polyynes by absorbing UV photons at relatively long wavelengths where the solar flux is significant (Barth, 2017). This absorption leads to the potential formation of radical species that are key to forming polycyclic aro­matic hydrocarbons (PAHs) (Huang et al., 2010), and larger organic species that constitute the haze layers observed in Titan's atmosphere (e.g. Fleury et al., 2024).

At lower stratospheric altitudes, the condensation of di­acet­y­lene into icy grains is expected to occur around 80 km, where the tem­per­a­ture reaches 114 K (Barth, 2017; Yu et al., 2023). Once formed, the ice descends and reaches the surface, where it is expected to remain solid (Yu et al., 2023; Yu et al., 2024). Therefore, understanding di­acet­y­lene's solid phase behaviour is crucial for future in situ exploration of its interaction with other molecular species.

Early studies using polarized IR spectroscopy suggested an orthorhombic structure (space group P22₁2₁) at liquid N₂ tem­per­a­ture (Freund & Halford, 1965; Khanna et al., 1988), although this structure has remained unconfirmed by diffraction analysis. Here, we report the first characterization of the crystal structure of di­acet­y­lene by powder X-ray and neutron diffraction, along with its thermal expansion behaviour determined at ambient pressure and tem­per­a­tures ranging from 5 to 220 K. These results are further supported by comparison of ex­per­i­mental Raman spectroscopy with periodic density functional theory (p-DFT) calculated lattice vibrations from the determined crystal structure.

2. Methods

Diacetylene was synthesized according to the procedure described in the literature via the dehydrochlorination of 1,4-dichloro-2-butyne (ClCH₂CCCH₂Cl) (Zhou et al., 2009). Here, 7.5 ml of ClCH₂CCCH₂Cl (AK Scientific, 99%) were added to 15 ml ethanol (C₂H₅OH; Lab Supply, 99.5%) and warmed to 353 K while stirring. Next, 50% aqueous sodium hydroxide (NaOH; Lab Supply, 97%) was added dropwise. A gas evolved and was bubbled through a 13% aqueous sodium hydroxide solution, then dried over calcium chloride (CaCl₂; Lab Supply, 90%) and condensed in a liquid nitrogen trap (77 K).

2.1. Diffraction measurements and structure solution procedure

Neutron powder diffraction ex­per­i­ments were conducted on the Wombat high-intensity diffractometer at the Australian Centre for Neutron Scattering (Maynard-Casely et al., 2025), utilizing a neutron wavelength of 2.41 Å. Confirmation of the wavelength was determined from a refinement of a NIST LaB₆ standard, along with establishing a 2θ zero error. Diacetylene was condensed into a 6 mm diameter vanadium can affixed to a positioning system within a top-loading cryofurnace (Brochier, 1977) designed for the study of condensing gases, adapted from an original design by Lee et al. (2016). During gas condensation, the cryofurnace was maintained at 95 K and connected to a gas delivery line, with the entire assembly placed within a cryofurnace. The tem­per­a­ture was controlled with the variable-tem­per­a­ture insert (VTI) of the cryofurnace – coupled to the sample environment with helium exchange gas, and a heater in a copper block above the sample can (tem­per­a­ture calibration information available in the supporting information). All data were collected without rotation of the cryostat. After sample condensation, the tem­per­a­ture was reduced to 5 K and a diffraction pattern was acquired for 2 h. Next, the tem­per­a­ture was increased to 195 K, and patterns were collected for 10 min at 5 K increments with the tem­per­a­ture stabilized at each point for 2 min before collection. At 100, 165 and 200 K, diffraction patterns were also acquired for 2 h.

Powder X-ray diffraction (PXRD) patterns were acquired on a Stoe Stadi-P diffractometer, operating in Debye–Scherrer geometry with an Mo Kα₁ X-ray source (λ = 0.7107 Å) and three Mythen detectors in a stationary mode, covering a 2θ range of 0–55°. We used a 6 mm divergence slit (dimension parallel to the capillary sample) and a 3 mm × 0.5 mm collimator (dimensions parallel and perpendicular to the capillary, respectively). Using an Oxford Cryosystems nitrogen cryostream to control the tem­per­a­ture, the sample was condensed at 80 K from the gas phase into a thin-walled 0.5 mm quartz capillary (WJM-Glas) and subsequently flame-sealed to iso­late the sample from the atmosphere during PXRD measurements. The tem­per­a­ture was increased to 220 K at a rate of 2 K min⁻¹, with diffraction measurements recorded simultaneously for 15 min in 2 K intervals. All data were acquired while rotating the capillary.

For structure solution, di­acet­y­lene was constructed as a rigid body using Avogadro software (Hanwell et al., 2012) to minimize the number of degrees of freedom during the solution process. The rigid body was then implemented in the Free Objects for Crystallography program (FOX Version 2022.1) (Favre-Nicolin & Černý, 2002), where the Parallel Tempering algorithm was used to perform the structure solution using the PXRD pattern recorded at 132 K. The optimization resulted in a feasible structure that was then further refined via Rietveld refinement (Rietveld, 1969) with GSAS-II software by implementing di­acet­y­lene as a vector rigid body.

The parameters varied for the refinement against the PXRD pattern were scale factor, background function, including a single broad peak to account for the scattering of the borosilicate capillary, the lattice parameters, the orientation of the rigid body, origin position and isotropic displacement of the atoms (U_iso). For the neutron powder diffraction pattern at 5 K, the same parameters were varied, as well as the crystallite size. Analysis of the two-dimensional plot from the neutron diffraction data collected at 5 K (see Fig. 1 in the supporting information) revealed some preferred orientation within the structure, which was then accounted for in the structure refinement.

2.2. Periodic-DFT calculation and micro-Raman spectroscopy

The resulting structure from the refinement against the laboratory X-ray data was used for geometry optimization and harmonic fre­quen­cy calculations applying density functional theory with periodic boundary (p-DFT). The CRYSTAL17 software suite (Dovesi et al., 2018; Pascale et al., 2004) was used for these calculations at the PBE0 hybrid functional level of theory (Perdew et al., 1996) supplemented with triple-ζ quality 6-311 G(d) basis sets, including D3(BJ) dispersion and dampening terms (Heyd et al., 2005; Grimme, 2006). This level of theory has previously been shown to result in an accurate structure prediction and overall good agreement with ex­per­i­mental data of molecular crystal systems (Banks et al., 2020). The coupled-perturbed Hartree–Fock/Kohn–Sham (CPHF/CPKS) approach was used to calculate the dielectric tensor and Raman intensities (Maschio et al., 2013a; Maschio et al., 2013b). The intensities of the polycrystalline Raman spectrum were calculated under the ex­per­i­mental conditions (Veithen et al., 2005), including the laser wavelength and sample tem­per­a­ture (532 nm and 90 K, respectively). Total energy convergence for the geometry optimization and fre­quen­cy calculations were set to 10⁻⁷ and 10⁻⁹, respectively. The absence of imaginary frequencies confirmed convergence to a potential energy surface (PES) minimum. To account for the known overestimation of the calculated vibrational frequencies with the PBE0 functional (Pierre et al., 2011), a scaling factor of 0.9518 was applied to correct for the harmonic approximation (Tikhonov et al., 2024).

Experimental acquisition of the Raman spectra was done using a high-resolution confocal dispersive micro-Raman spectrometer (Horiba Jobin–Yvon LabRam HR). The spectrometer was equipped with a fre­quen­cy-doubled Nd:YAG laser (532 nm, 100 mW), an 1800 grooves/mm diffraction grating and a 50× Olympus BXFM objective. Spectra were acquired at a resolution of 0.4 cm⁻¹ per detector pixel over the 50–4000 cm⁻¹ fre­quen­cy range, with two spectral accumulations and 20 s acquisition per spectral region to improve the signal-to-noise ratio. The fre­quen­cy of the spectrometer was calibrated against a silicon standard reference peak at 520.7 cm⁻¹.

A commercial Linkam LTS350 optical cryostage was mounted on the XYZ motorized translation stage underneath the objective, allowing for the sample tem­per­a­ture to be electronically controlled (to within ±0.1 K) by a resistive heating thermal block counterbalanced by a flow of liquid nitrogen. Before starting the ex­per­i­ment, the cryostage headspace was purged for 15 min with gaseous nitrogen at ∼330 K to remove residual atmospheric moisture. Then, di­acet­y­lene was deposited directly onto a microscope slide placed inside the cryostage at 200 K, to avoid concomitant crystallization of atmospheric carbon dioxide. The sample was then cooled to 90 K and the tem­per­a­ture increased in 10 K increments until 150 K, with an equilibration time of 2 min at each tem­per­a­ture point.

3. Results

3.1. Crystal structure determination

The neutron and X-ray powder diffraction (NPD and PXRD) patterns acquired at 5 and 132 K, respectively, were analysed using the GSAS-II software (Toby & Dreele, 2013), employing a combination of Gaussian and Lorentzian functions to fit the observed peaks. The autoindexing function of GSAS-II in both monoclinic and orthorhombic lattices was applied. The resulting M20 values and overall agreement between reflection positions and ex­per­i­mental peaks indicated a lattice in either the P222 or the P2 space group. Considering previous results on the di­acet­y­lene structure via polarized IR spectroscopy (Freund & Halford, 1965), which proposed di­acet­y­lene to have orthorhombic symmetry (however, no lattice parameters were proposed), indexing was progressed on that basis within GSAS-II and subsequent Pawley refinements (Pawley, 1981) were performed in the space group P222 (Fig. 2 in the supporting information).

It was noted, however, that small peaks are unaccounted for in the NPD and PXRD patterns. In the NPD pattern, the peaks at 39.5 and 41.5° (marked with an asterisk) are unaccounted for. For the PXRD pattern, a peak at 10.77 (marked with an asterisk) and other minor peaks at higher 2θ are also not accounted for in the Pawley fit. These impurity peaks likely suggest the presence of a known byproduct of the synthesis, identified as 2-chlorobut-1-en-3-yne (C₄H₃Cl) (Jolly et al., 2014) by mass spectrometry with m/z 86 and 88, whose crystal structure is unknown. Differences in the peak positions are attributed to the intrinsic differences between the two techniques and the ex­per­i­mental conditions, e.g. the presence of preferred orientation. Therefore, the unaccounted peak was attributed to the presence of this known contaminant and was not considered further in the structure solution.

Initial analysis of reflection conditions (h00, h = 2n; 0k0, k = 2n) and Le Bail refinements (Le Bail et al., 1988) against the PXRD pattern at 132 K indicated P2₁2₁2₁ as a possible space group (more information is provided in the supporting information). The volume of the unit cell obtained (329.51 ų) and the space-group designation suggested the presence of four di­acet­y­lene molecules in the unit cell. Consequently, we first proceeded with structure solution in the space group P2₁2₁2₁, where di­acet­y­lene was implemented as a rigid body with all its six atoms defined and placed in general positions. The molecule was placed with its centre of mass at the origin, and its position and orientation subsequently refined. The resulting Rietveld refinement against the PXRD pattern acquired at 132 K resulted in an R_wp value of 12.12% and a GoF of 2.42 (see Fig. 4 in the supporting information). Rietveld refinement was then performed against the neutron diffraction pattern collected at 5 K [Fig. 1(a)], resulting in R_wp = 3.49% and GoF = 4.41. The lattice parameters were determined to be a = 9.343 (2), b = 5.9859 (7), c = 5.6752 (13) Å and V = 317.39 (8) ų, with a U_iso refined to 0.018 (2) Ų. The presence of preferred orientation was accounted for in the structure refinement with the [010] direction for a March–Dollase correction refined to 0.815 (7).

Figure 1 (a) Rietveld fit of the 5 K NPD pattern with the refined P212121 structure (Rwp = 3.49% and GoF = 4.41). (b) Rietveld fit of the 5 K NPD pattern with the refined Pnma structure (Rwp = 3.57% and GoF = 4.50). The blue line below the data indicates the difference between the observed (dotted black line) and calculated (red line; offset for clarity) patterns. The green tick marks indicate the position of expected reflections from the derived crystal structure.

The molecular arrangement within the refined space group P2₁2₁2₁ results in molecules adopting a slightly off-plane configuration, with y fractional coordinates close to ; consequently, the possibility of describing the crystal structure in a higher-symmetry space group was investigated. The space group P2₁2₁2₁ (No. 19) is a subgroup of Pnma (No. 62) where the inversion and mirror symmetries have been broken. Breaking the mirror symmetry can be achieved by moving the atoms from y = in Pnma, while retaining the same size unit cell. A structural remodel in Pnma was therefore created and tested against the neutron powder diffraction pattern acquired at 5 K [Fig. 1(b)]. This provided a satisfactory fit, and the final R_wp and GoF obtained were 3.57% and 4.50, respectively. The refined lattice parameters were: a = 9.348 (2), b = 5.9890 (6), c = 5.6746 (11) Å and V = 317.69 (6) ų, with a U_iso value refined to 0.0219 (15) Ų and a March–Dollase correction refined to 0.829 (7).

The molecular arrangement in the space group P2₁2₁2₁ is shown in Fig. 2(a) (left). When the molecules are positioned precisely at y = within the space group Pnma, they are placed on mirror planes [Fig. 2(a), right] without significantly altering the intermolecular interactions observed in the space group P2₁2₁2₁.

Figure 2 (a) Comparison between the refined structure at 5 K in the space group P212121 (left) and the molecular arrangement within the space group Pnma (right). The mirror planes in the Pnma structure are shown in light blue. (b) The molecular orientation after refining the molecular positions in the Pnma structure, resulting in a twofold disordered arrangement.

Once the refinement of the molecular position is included in the model, the molecules shift off the mirror plane. This leads to a twofold positional disorder that improves the fit to the diffraction patterns [Table 1 and Fig. 2(b)]. Additional Rietveld refinements of the neutron diffraction patterns collected at 100, 165 and 200 K were also performed for both space groups under consideration, and the resulting R_wp and GoF values are shown in Table 1.

Table 1
Resulting R_wp and GoF of the Rietveld refinement to the NPD patterns acquired at 5, 100, 165 and 200 K in the space groups Pnma and P2₁2₁2₁

	Pnma	Pnma†	P212121
NPD pattern (K)	Rwp (%), GoF
5	3.57, 4.50	3.48, 4.39	3.49, 4.41
100	3.71, 4.67	3.68, 4.65	3.55, 4.49
165	3.08, 3.89	2.88, 3.64	2.86, 3.61
200	2.89, 3.66	2.86, 3.62	2.88, 3.66

†Values after refining the molecular position within the structure.

The presence of preferred orientation, impurities and the real-space resolution of the neutron diffractometer at λ = 2.41 Å imply that the resulting R_wp and GoF, although lower for the refinement in the space group P2₁2₁2₁, might result from the higher positional degrees of freedom and overfitting of the data. Consequently, we concluded that the di­acet­y­lene crystal structure is better described by the space group Pnma. There remains the possibility that the structure is positionally disordered within the Pnma symmetry, but additional ex­per­i­ments would need to be conducted using, for example, inelastic neutron scattering.

The obtained crystal structure in the ordered space group Pnma shows that the di­acet­y­lene molecules form layers of almost perpendicular molecules (Fig. 3), resulting in an intricate network of intermolecular C—H⋯π interactions.

Figure 3 The molecular arrangement within the NPD refined structure at 5 K, viewed down the b axis, highlighting the main intermolecular C—H⋯C interaction.

3.2. Thermal expansion and stability

The thermodiffractogram of the neutron diffraction data taken from 5 to 195 K is shown in Fig. 4. A single phase was observed throughout the tem­per­a­ture range of this study, which extended to 220 K in the PXRD ex­per­i­ment (see Fig. 5 in the supporting information). Neutron diffraction data were not acquired above 200 K to preserve the ex­per­i­mental setup, due to the risk of polymerization in the liquid phase.

Figure 4 A thermodiffractogram of the neutron data from di­acet­y­lene ranging between 5 and 195 K. It can be seen from this plot that there are no phase transitions within this tem­per­a­ture range.

Using the determined crystal structure for di­acet­y­lene, Rietveld refinements were performed sequentially on the patterns collected by neutron diffraction from 5 to 195 K to observe changes in the unit-cell volumes and the dimensions of each crystallographic axis as a function of tem­per­a­ture. Detailed lattice parameters and unit-cell volumes derived from Rietveld refinements are listed in Table 1 of the supporting information.

The associated plot for the thermal expansion on each crystallographic axis and volume as a function of tem­per­a­ture is shown in Fig. 5. Here, an anisotropic behaviour is observed, reflecting the layered molecular arrangement parallel to the (010) plane, as shown in Fig. 2. In such an arrangement, there are no strong intermolecular interactions in the direction of the b axis, i.e. between the 0k0 planes. This reflects the higher relative expansion with increased tem­per­a­ture compared to the other two axes. The a and c axes contrarily, align with the molecular plane and their intermolecular interactions, resulting in a lower expansion with increased tem­per­a­ture. Fig. 6 in the supporting information shows how these interactions change with tem­per­a­ture, based on the Rietveld refinements against the NPD patterns acquired at 5, 100, 165 and 200 K.

Figure 5 (a) Relative expansion of the a, b and c axes over the tem­per­a­ture range studied normalized to 5 K; a change in the thermal expansion rate is observed around 160 K. (b) Variation of the unit-cell volume with tem­per­a­ture, fitted with the Salje equation of state {V0T = [p0 + p1θsatcoth(θsat/T)]^3} (Salje et al., 1991).

The variations of the unit-cell volume with tem­per­a­ture were fitted to the Salje equation of states (Salje et al., 1991) using the EoSFIT7 code (Angel et al., 2014; Gonzalez-Platas et al., 2016). The saturation tem­per­a­ture (θ_sat) was refined to 104 (5) K, while the zero-tem­per­a­ture volume (V₀) and p₁ were refined to 317.25 (13) ų and 129 (4) × 10⁵ K⁻¹, res­pec­tively.

3.3. Structure corroboration via vibrational spectroscopy and periodic DFT

Structural validation was sought from energy minimization calculations, which were then used to calculate the vibrational frequencies to compare with ex­per­i­mental vibrational spectroscopy results. Full geometry optimization of the obtained structure was done via p-DFT calculations, including dispersion corrections. The absence of imaginary frequencies during the fre­quen­cy calculation indicates that the final theoretical structure converged to a PES minimum, showing good agreement with the ex­per­i­mental structure (Fig. 6). A comparison between the two structures was conducted in the Mercury software (Macrae et al., 2020) from the Cambridge Crystallographic Data Centre, using the Crystal Packing Similarity function. This resulted in an r.m.s. deviation of 0.175 for 12 molecules.

Figure 6 Comparison between the structure refined from the NPD data at 5 K (red; outer unit cell) with the output from geometry optimization (blue; inward unit cell).

A comparison between the unit-cell parameters for the calculated and ex­per­i­mental structures is shown in Table 2. The optimized structure exhibited a deviation of less than 5% in the unit-cell parameters and approximately 10% in the volume compared with ex­per­i­mental values obtained at 5 K.

Table 2
Comparison between the lattice parameters and unit-cell volumes of di­acet­y­lene at 5 K obtained from Rietveld refinements of the neutron diffraction data with the output from the geometry optimization

Structure	a (Å)	b (Å)	c (Å)	V (Å^3)
NPD (5 K)	9.3478 (20)	5.9890 (6)	5.6746 (11)	317.69 (6)
p-DFT (0 K)	9.20585	5.72610	5.45640	287.626713

To further validate the proposed structure, we performed p-DFT calculations to obtain simulated Raman and IR spectra (see the supporting information) to compare with ex­per­i­mental spectra. Comparison with ex­per­i­mental Raman spectra is especially important, as Raman spectroscopy provides information on the molecular arrangement and lattice vibrations within the crystalline structure, where the low-fre­quen­cy region can be used as a fingerprint for the crystal (Loudon, 2001). Here, the ex­per­i­mental Raman spectrum acquired at 90 K is compared with the p-DFT Raman spectrum simulated at the same tem­per­a­ture (Fig. 7). The observed and calculated Raman vibrational frequencies and mode assignments are provided in Table 3.

Table 3
Observed and calculated Raman vibrational frequencies (cm⁻¹) and mode assignments for crystalline di­acet­y­lene

Assignment†	Mode	Exp. (90 K)	p-DFT (90 K)	Lit. (liquid ∼250 K)‡
ν4	CH asymmetric stretch	3281.4	3284.9	–
ν1	CH symmetric stretch	3273.1	3262.1	3293
ν2	CC triple-bond stretch	2173.9, 2178.9	2190.2, 2196.7	2172
ν3	CC single-bond stretch	873.5	883.4	877
ν8§	CCH bending	663.4	705.2	–
ν6	CCH bending	650.9	661.4	647
ν7	CCC bending	484.3	495.6	484
	Lattice vibration	96.0	124.5
	Lattice vibration	88.2 (sh)	118.2
	Lattice vibration	67.7	82.3, 98.2

†Jones (1952) and Owen et al. (1987).
‡Jones (1952).
§Jolly et al. (2010).

Figure 7 Comparison between the theoretical and ex­per­i­mental spectra of crystalline di­acet­y­lene at 90 K. (a) The top spectrum corresponds to that acquired ex­per­i­mentally (red), while the calculated Raman spectrum is displayed at the bottom (blue). The black bars correspond to the calculated vibrational modes. (b) The lower-fre­quen­cy range of the Raman spectra, where lattice vibrational modes are active. The correlation between corresponding bands is shown with black dotted lines. (c)/(d) Mid-fre­quen­cy range of the Raman spectra showing the fundamental vibrational modes.

The calculated spectrum has an overall good agreement with the ex­per­i­mental spectrum, with a nearly one-to-one correspondence between the bands, in both their positions and intensities. All fundamental vibrational modes are accurately predicted, and anharmonic features (not calculated) explain the satellite features such as the band around 2170 cm⁻¹, while the band at ∼1000 cm⁻¹ can be attributed to an overtone of the band at 484 cm⁻¹. The lattice vibrational modes calculated from the optimized crystal structure are also well predicted, with the bands around 98 and 82 cm⁻¹ in the calculated spectrum likely corresponding to the broad band with low intensity at 69.9 cm⁻¹ in the ex­per­i­mental spectrum. The 88.2 cm⁻¹ shoulder to the band at 96 cm⁻¹ also has a corresponding calculated band at 118 cm⁻¹, more evident in Fig. 8.

Figure 8 Expanded view of the lattice vibrational modes in the range 90–150 K. Two distinct bands can be identified at 96.0 and 67.7 cm^−1, with a shoulder at 88.2 cm^−1.

In Table 3, our assigned vibrational modes and frequencies are also compared to those identified in previous ex­per­i­mental spectra of di­acet­y­lene in the liquid phase (Jones, 1952). The band shifts observed can be attributed to the different tem­per­a­tures and states (liquid and solid) in which the ex­per­i­ments were performed.

Fig. 8 shows the lattice vibrational region, where no significant changes occur between 90 and 150 K, with bands retaining their profile with minor shifts in position. Here, the varying intensities can be attributed to molecular motion within the lattice, without change in crystal phase. This is consistent with the single phase observed in the PXRD and neutron diffraction ex­per­i­ments.

4. Discussion

4.1. Comparison with the structure proposed from polarized IR spectroscopy

Using neutron and X-ray diffraction, this study demonstrates that di­acet­y­lene crystallizes in a single orthorhombic Pnma phase across the tem­per­a­ture range from 5 to 220 K. The molecular arrangement and anisotropic behaviour are consistent with previous spectroscopic findings.

The refined structure shows that di­acet­y­lene molecules form layers orthogonal to the b axis, while the molecular planes align along the a and c axes. These results agree with observations by Freund & Halford (1965) from polarized IR spectroscopy, where they identified absorptions of fundamental vibrational modes primarily occurring along two directions parallel to the molecular plane. Notably, no absorption was detected along the third direction, corresponding to the absence of dipole moment displacement in that orientation. This direction was later described as an axis orthogonal to the molecular plane. Furthermore, their results showed that both H atoms present in the molecule are involved in hydrogen interactions with the π-electron cloud of the C≡C bonds, supporting the molecular orientation in a coplanar arrangement shown in Fig. 3.

The planar configuration combined with the unit-cell volume refined here confirms a structure containing four molecules in the primitive unit cell, agreeing with the structure pro­posed by Freund & Halford (1965). In the space group Pnma, the molecules are placed on mirror planes, in contrast with results from Freund & Halford (1965) which suggested their placement in non-symmetric sites in the non-centrosymmetric orthorhombic space group P22₁2₁ (Freund & Halford, 1965). However, according to Kitaigorodskii's rules, it is expected that the di­acet­y­lene crystal structure is centrosymmetric due to the molecular D_∞h point group, supporting our proposed structure in the space group Pnma.

4.2. Comparison with the crystal structure of acetylene

Acetylene is the simplest alkyne and exhibits two polymorphs at ambient pressure: an orthorhombic Acam crystal at tem­per­a­tures below 133 K, transitioning to a cubic Pa3 crystal above 133 K (Koski & Sándor, 1975; McMullan et al., 1992). To compare with the refined crystal structure of di­acet­y­lene, the orthorhombic polymorph of acetylene was chosen due to the comparable space group and conditions under which both structures were analysed. As there are no low-tem­per­a­ture hydrogenous acetylene structures, the d-acetylene structure refined at 4.2 K [C₂D₂, Acam (Koski & Sándor, 1975>); CSD refcode ACETYL04] was compared with the structure of di­acet­y­lene refined at 5 K (Pnma, this work).

To investigate intermolecular interactions and explore structural similarities, Hirshfeld surfaces (HSs) were con­structed using the CrystalExplorer program (Spackman et al., 2021) (Fig. 9). These surfaces were mapped with the normalized contact distance (d_norm) (Fig. 9), which considers the van der Waals (vdW) radii of the atoms and the distances to the nearest atoms, both internal and external to the surface (Spackman & Jayatilaka, 2009). Red regions on the HS indicate those contacts shorter than the sum of the vdW radii. White and blue regions represent contacts near and exceeding the length of the vdW limit, respectively (McKinnon et al., 2007).

Figure 9 Hirshfeld surfaces of (a) di­acet­y­lene (refined at 5 K) and (b) acetylene [Acam; CSD refcode ACETYL04 (Koski & Sándor, 1975)] mapped with dnorm surfaces. (c) Relative contributions to the Hirshfeld surface area for the different close intermolecular contacts in acetylene and di­acet­y­lene.

As shown in Figs. 9(a) and 9(b), both structures exhibit layered molecular arrangements with a slightly off-diagonal alignment. The closest intermolecular contacts highlighted in red arise from C—H⋯π interactions. Acetylene molecules lie parallel to the (001) plane and adopt a T-shaped geometry (Koski & Sándor, 1975), whereas di­acet­y­lene molecules align parallel to the (010) plane. The primary intermolecular interactions in both structures involve C—H⋯π interactions within the layers, with distances of 2.492 and 3.008 Å for di­acet­y­lenes placed on mirror planes in the Pnma structure. In contrast, the acetylene molecules lie in the mirror plane of Acam at positions of 2/m symmetry, resulting in a symmetric C—H⋯π interaction of 2.658 Å.

Fig. 9(c) details the relative contributions from the intermolecular contacts within the layers, extracted from the fingerprint plots shown in Fig. 8 in the supporting information. The C—H⋯π interactions dominate both structures, accounting for 74.1% of the contacts of di­acet­y­lene and 66.9% of the contacts of acetylene. Notable differences arise from the contributions of C⋯C and H⋯H close contacts. Diacetyl­ene's layered arrangement results in an approximate interlayer distance of 3.1 Å and a higher crystal density of 1.0466 g cm⁻³ at 5 K, leading to a more closed-packed structure and consequent greater pro­por­tion of C⋯C interactions (17.1%). In contrast, the orthorhombic structure of acetylene has a lower density of 0.902 g cm⁻³ at 4.2 K, with minimal C⋯C interactions (1.4%). The increased H⋯H interactions in the structure of acetylene (31.6 versus 8.7%) can be attributed to the use of d-acetylene, as the larger vdW radius of deuterium influences the d_norm calculations.

The intermolecular interactions within di­acet­y­lene, in particular the C—H⋯π interactions, directly impact the thermal expansion observed in Fig. 5. Notably, the expansion along the c axis closely follows the variation in the intermolecular C—H⋯C distances, as shown in Fig. 6 in the supporting information. In contrast, the expansion along the a axis may result from a combination of two factors: the change in C—H⋯C distances and variation of the angle between the H atom and the C atoms involved in the triple bonds, resulting in the strong anisotropic behaviour observed. Additional ex­per­i­ments and theoretical simulations are needed to elucidate the molecular dynamics of the crystal and provide a more comprehensive description of the underlying mechanism governing the thermal expansion along the a axis.

4.3. Implications for Titan in situ exploration

Future in situ exploration, such as the Dragonfly rotorcraft (Barnes et al., 2021), will encounter a range of materials on Titan's surface. In preparation, attempts are being made to create a database of their properties, compiling information such as phase change and organic ice density (Yu et al., 2023). Understanding these properties is critical for determining the physical state in which these compounds are likely to be found, depending on the type of terrain they encounter once reaching the surface (i.e. dry surface, lakes and seas).

While the crystal structures of some Titan organics remain unstudied ex­per­i­mentally, theoretical models were developed to predict some of these properties. For example, Yu et al. (2023) proposed that di­acet­y­lene's density (ρ) varies with tem­per­a­ture (T) according to equation (3):

where ρ is measured in kg m⁻³ and T in Kelvin. Using this relationship, the predicted densities of di­acet­y­lene at 5, 100, 165 and 200 K are 1191.13, 1130.93, 1089.75 and 1067.57 kg m⁻³, respectively. In our study, however, the refined crystal structure yielded densities of 1046.6, 1029.2, 998.8 and 978.0 kg m⁻³ at 5, 100, 165 and 200 K, respectively, corresponding to deviations of 12.13, 8.99, 8.35 and 8.39% from the theoretical values.

To understand these discrepancies, we compared the di­acet­y­lene results with the published acetylene data [CSD refcode ACETYL06-10 (Koski & Sándor, 1975; McMullan et al., 1992)], as Yu et al. (2023) similarly modelled the density of acetylene within the temperature range 77–188.15 K [equation (4)]:

The resulting densities yielded an average difference of 10.07%, comparable to the deviations observed for di­acet­y­lene. For acetylene, the deviation likely arises from incorporating density data obtained via diffraction techniques and pycnometry in equation (4). Since pycnometry directly measures the volume of the condensed ice, it may include pore space (Yu et al., 2023). The divergences observed for both hydrocarbons highlight the necessity for further ex­per­i­mental data on fundamental parameters that are essential for understanding physical and chemical processes occurring on planetary bodies such as Titan.

The structural similarities between di­acet­y­lene and acetylene may also imply a potential for di­acet­y­lene to participate in similar physicochemical processes. The layered arrangement and C—H⋯π interactions suggest that di­acet­y­lene could also cocrystallize with other hydrocarbons and nitriles. Previous studies have identified acetylene cocrystals with butane (Cable et al., 2019), ammonia (Boese et al., 2009; Cable et al., 2018), acetonitrile (Kirchner et al., 2010; Cable et al., 2020), benzene (Boese et al., 2003; Czaplinski et al., 2020; Francis et al., 2023), carbon dioxide (Gough & Rowat, 1998), pyridine (Czaplinski et al., 2023) and propionitrile (Czaplinski et al., 2025) under Titan-relevant conditions. Thus, these molecules are promising candidates for cocrystallization studies involving di­acet­y­lene.

Cocrystallization can also affect Titan's surface chemistry by influencing the reactions that form complex organic molecules. For instance, previous studies indicate that the photochemistry of co-condensed pyridine and acetylene ices can form precursors to nitrogen-containing polycyclic aro­matic hydrocarbons. Contrarily, cocrystallization can hinder photodegradation and preserve the coformers (Lopes Cavalcante et al., 2024). Both scenarios can potentially alter the range of materials found on Titan's surface, highlighting the need for further ex­per­i­mental investigations, especially those involving di­acet­y­lene due to its absorption of longer wavelength UV photons (λ ≥ 300 nm) compared to acetylene (Couturier-Tamburelli et al., 2015; Fleury et al., 2019; Fleury et al., 2024), where it could initiate photochemical reactions within the mixed ices and form more complex molecules.

5. Conclusion

This study presents the first crystallographic characterization and thermal expansion behaviour of di­acet­y­lene (1,3-butadiyne). Powder neutron and X-ray diffraction data indicate the formation of an orthorhombic structure (space group Pnma) with four molecules in the unit cell. The molecules are on mirror planes and arranged in coplanar layers within which C—H⋯π are the main intermolecular interactions. Variable-tem­per­a­ture ex­per­i­ments indicate that there are no phase transitions in the 5 to 220 K tem­per­a­ture range. Thermal expansion analysis reveals an anisotropic thermal behaviour, with larger expansion along the b direction, where no meaningful intermolecular interactions occur. Periodic-DFT calculations combined with Raman spectroscopy further support the proposed structure, which also agrees with previous spectroscopic considerations (Freund & Halford, 1965). Finally, comparison with the orthorhombic crystal structure of acetylene indicates structural and intermolecular interaction similarities. This implies that di­acet­y­lene may perform a similar role in forming cocrystals and complex organic molecules relevant to Titan's surface chemistry.

6. Related literature

The following reference is cited in the supporting information: Fortes (2018).

Supporting information

Computing details

1,3-butadiyne (NPD5K)

Crystal data

C4H2	c = 5.6746 (11) Å
Mr = 50.06	V = 317.69 (6) Å^3
Orthorhombic, Pnma	Z = 4
a = 9.348 (2) Å	Dx = 1.047 Mg m^−3
b = 5.9890 (6) Å	T = 5 K

Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Ų)

	x	y	z	Uiso*/Ueq
C1	0.25113	0.75000	0.22002	0.0219*
C2	0.17520	0.75000	0.39054	0.0219*
C3	0.07826	0.75000	0.60824	0.0219*
C4	0.00234	0.75000	0.77875	0.0219*
H5	−0.06503	0.75000	0.93006	0.0219*
H6	0.31851	0.75000	0.06872	0.0219*

1,3-butadiyne (Rietveld+RB_NPD100K)

Crystal data

C4H2	c = 5.7013 (19) Å
Mr = 50.06	V = 323.1 (3) Å^3
Orthorhombic, Pnma	Z = 4
a = 9.363 (4) Å	Dx = 1.029 Mg m^−3
b = 6.052 (3) Å	T = 100 K

Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Ų)

	x	y	z	Uiso*/Ueq
C1	0.25163	0.75000	0.21523	0.0260*
C2	0.17617	0.75000	0.38536	0.0260*
C3	0.07983	0.75000	0.60258	0.0260*
C4	0.00438	0.75000	0.77272	0.0260*
H5	−0.06258	0.75000	0.92369	0.0260*
H6	0.31859	0.75000	0.06426	0.0260*

1,3-butadiyne (Rietveld+RB_NPD165K)

Crystal data

C4H2	c = 5.761 (3) Å
Mr = 50.06	V = 332.9 (4) Å^3
Orthorhombic, Pnma	Z = 4
a = 9.378 (5) Å	Dx = 0.999 Mg m^−3
b = 6.162 (3) Å	T = 165 K

Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Ų)

	x	y	z	Uiso*/Ueq
C1	0.24850	0.75000	0.21680	0.0513*
C2	0.17399	0.75000	0.38615	0.0513*
C3	0.07886	0.75000	0.60236	0.0513*
C4	0.00434	0.75000	0.77172	0.0513*
H5	−0.06177	0.75000	0.92199	0.0513*
H6	0.31462	0.75000	0.06652	0.0513*

1,3-butadiyne (Rietveld+RB_NPD200K)

Crystal data

C4H2	c = 5.7971 (15) Å
Mr = 50.06	V = 339.99 (9) Å^3
Orthorhombic, Pnma	Z = 4
a = 9.399 (2) Å	Dx = 0.978 Mg m^−3
b = 6.2398 (11) Å	T = 200 K

Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Ų)

	x	y	z	Uiso*/Ueq
C1	0.24728	0.75000	0.22406	0.0681*
C2	0.17336	0.75000	0.39284	0.0681*
C3	0.07900	0.75000	0.60833	0.0681*
C4	0.00509	0.75000	0.77712	0.0681*
H5	−0.06050	0.75000	0.92690	0.0681*
H6	0.31287	0.75000	0.07428	0.0681*