Skip to content

Diamondiyne - A 3D Carbon Allotrope With Mixed sp–sp 3 Hybridization

FieldValue
JournalAngewandte Chemie International Edition
Published2026-09-14
AuthorsYizhou Yang, Jie Xu, Yanyan Chen, Yu Xia, Sami Zeliouche
DOI10.1002/anie.4062963

Diamondiyne is the second known carbon allotrope that forms covalent bonds in all three dimensions, after diamond. The authors produced amorphous carbon films containing diamondiyne crystallites through a liquid-liquid interfacial reaction. Electron microscopy supports an asymmetric two-fold interpenetrated network in the crystalline regions. The films reached areas up to 5 cm², and reaction time produced films below 5 nm and above 60 nm. These results support a new 3D carbon phase at the crystallite level, not a bulk diamondiyne material.

Diamondiyne expands the diamond-topology network by linking tetrahedral sp³ carbon centers with linear diyne groups. The illustrated structure contains sp carbon and sp³ carbon in an 8:1 ratio. The diyne links use triple bonds, while the tetrahedral centers retain sp³ character. This combination preserves three-dimensional covalent connectivity but enlarges the repeating unit and introduces pores. The authors describe diamondiyne as a carbo-mer of diamond with an expanded, more porous network.

The ideal noninterpenetrated framework was expected to have face-centered cubic Fd-3m symmetry and a 16.22 Å unit cell edge. Its observed interpenetration and structural disorder complicate that ideal model.

The reaction placed tetra(triethylsilylethynyl)methane and CuF₂ in separate phases. Chloroform lay below water, and pyridine was distributed in both phases. F⁻ from CuF₂ removed triethylsilyl protection groups to release tetraethynylmethane. Cu²⁺ then catalyzed coupling between terminal alkynes. Keeping both reactions at the interface released the unstable monomer gradually and coupled it as it formed.

The authors held the protected monomer concentration at 7 mM. Their kinetic model assigns first-order growth to monomer addition at a crystallite and second-order kinetics to unwanted homocoupling in the bulk. The low concentration favors molecule-by-molecule addition, which matters because carbon-carbon bonds do not self-correct like reversible bonds used in many covalent organic frameworks.

Time-resolved ¹H NMR of the CDCl₃ phase detected terminal alkyne protons a little less than one week after reaction began. Slice-selective ¹H NMR on day 7 mapped the chloroform phase at 1.25 mm slice spacing. Deprotected monomer was concentrated near the interface, while protected monomer increased farther away. This persistent gradient showed that diffusion did not homogenize the phase during film growth and supported the interfacial reaction mechanism.

The film grew normal to the interface and covered areas up to 5 cm². Optical microscopy showed a homogeneous film, with wrinkles attributed to transfer onto silicon. AFM found a smooth surface without observable pinholes or cracks, and SEM supported film continuity. The representative AFM section profile gave a thickness of 16 nm.

Film thickness increased linearly with reaction time. The authors therefore concluded that catalyst or monomer mass transfer through the film did not limit growth. Films below 5 nm were difficult to transfer, whereas films above 60 nm supported transfer of pieces with areas of 10⁴ to 10⁵ square micrometers. The reported reaction yield was on the order of up to 1%, and yield optimization had not been performed.

The authors assume that the solvent-volume-to-interface-area ratio strongly affects yield. Changing that ratio would also change monomer transport and require reoptimization of concentration and solvent composition. QCM indicated porosity qualitatively, but the film mass was too small for gas isotherm analysis. The measured Young’s modulus was 2.7 GPa. The authors report that this is slightly higher than for PMMA and relatively close to the predicted 19 GPa for diamondiyne.

FTIR showed no C-H stretching bands at 2873 to 2956 cm⁻¹, indicating that silyl protection groups were absent. The terminal-alkyne band at 3300 cm⁻¹ was also absent, which argues against residual deprotected intermediates. No significant carbonyl bands appeared at 1700 to 1725 cm⁻¹, unlike evidence from Feldman’s earlier oxidative Glaser-Hay attempts.

Raman spectroscopy produced a weak but reproducible diyne band at 2234 cm⁻¹. It lies close to the 2261 cm⁻¹ band of hexa-2,4-diyne and above the 2192 cm⁻¹ band of the protected monomer. The film band was also above the reported 2094 to 2139 cm⁻¹ range for amorphous and silicon diamondiyne variants. The authors interpret this frequency placement as evidence for a diyne group connected to nonaromatic groups.

Solid-state ¹³C NMR gave signals at 65 to 75 ppm, matching the 72.2 and 64.9 ppm diyne carbons of hexa-2,4-diyne. These measurements confirm sp-hybridized carbon in the film. Signals at 120 ppm indicated sp² byproducts, while minor signals near 220 ppm indicated oxidized defects. XPS found carbon and oxygen on the surface, and EDX found less oxygen in the interior. The authors attribute greater top-surface oxidation to terminal alkyne exposure to water. Only minor traces of F⁻ were detected.

HRTEM and STEM resolved ordered lattices within the amorphous film. A 3.4 Å periodicity was close to the predicted 3.31 Å diamondiyne spacing. An iDPC-STEM image showed a 9.6 Å periodicity with uneven lattice fringes, which does not fit symmetric interpenetration. Electron diffraction showed a weak 222 reflection and stronger 333, 444, and 555 reflections. This intensity pattern was inconsistent with symmetric interpenetration but consistent with two asymmetrically displaced nets separated by about 3 Å.

The [010] diffraction pattern matched the asymmetric two-fold model. Along [133], the projected electrostatic potential map agreed with a simulated iDPC-STEM image. The authors therefore assigned the crystalline regions to tetragonal space group I41/amd, with a = b = 11.47 Å and c = 16.22 Å. The combined observations support this structure, but individual crystallites varied in interpenetration and amorphous content.

Several observed d-spacings exceeded the lowest-angle reflection of CuF₂, so the authors rejected encapsulation of the catalyst as the source of those lattices. TEM at 5, 14, and 21 days showed clear lattice fringes by day 5. That timing supports direct formation of crystalline domains rather than later correction of an initially disordered network.

The model treated a five-membered ring as a defect formed by removing one node from ideal diamond topology. Its fully relaxed geometry defined the zero-energy reference. As more tetrahedral carbon atoms were constrained to the ideal structure, the defect energy increased to, and likely above, 18 kJ/mol per defect. The authors relate this value to an equilibrium constant of roughly 10⁻³.

The calculation covers defect relaxation but omits the larger inherent penalty for creating the defect. The authors did not estimate that penalty because hydrogen-capped cluster models would require different numbers of hydrogen atoms. The calculation therefore supports, rather than measures, their conclusion that framework strain and low monomer concentration limit errors during growth.

Engineering implications and open questions

Section titled “Engineering implications and open questions”

The work establishes a synthetic route to a 3D all-carbon network with nonreversible carbon-carbon bonds. It also shows that kinetic control at an interface can produce crystalline domains without the self-correction used in covalent organic frameworks.

For materials selection, the data do not support treating diamondiyne as a higher-performance form of crystalline diamond. The measured product is an amorphous film containing crystallites, with different interpenetration between regions and a measured modulus well below the predicted value.

Several limits remain. The yield was only on the order of 1%, ultrathin films had poor transfer robustness, and synthesis required days. Oxygen was concentrated at the water-exposed surface, sp² byproducts and minor carbonyls were detected, and the film mass prevented a gas isotherm. The paper reports no electrical transport, thermal transport, long-term stability, or device performance.

The demonstrated engineering result is a process for building three-dimensional covalent carbon networks from irreversible bonds. Diamondiyne remains an experimental allotrope film rather than a device material with established specifications.

View Original Abstract

ABSTRACT The creation of carbon allotropes is synthetically highly challenging, but their practical significance to science and technology cannot be underestimated. Here, we report the synthesis of a new carbon allotrope, diamondiyne, which is the second known one (after diamond) that forms covalent bonds in all three dimensions. Diamondiyne has a mixed sp-sp 3 hybridization, resulting in a carbo‐mer of diamond with an expanded diamond‐topology network. On the nanoscopic scale, two asymmetrically interpenetrated 3D networks form the tetragonal space group I 4 1 / amd with unit cell parameters a = b = 11.47 Å and c = 16.22 Å. Amorphous carbon films containing diamondiyne crystals are formed using a cascade reaction at a liquid-liquid interface. Just as for a covalent organic framework, molecular nodes are connected based on geometrical constraints to form an extended structure, and we show that the method is scalable in both the thickness and the lateral direction of the film. New carbon allotropes have historically found widespread use in materials science, and we look forward to what applications might emerge for diamondiyne.