Diamond-graphene composite nanostructures.

The search for new nanostructural topologies composed of elemental carbon is driven by technological opportunities as well as the need to understand the structure and evolution of carbon materials formed by planetary shock impact events and in laboratory syntheses. We describe two new families of diamond-graphene (diaphite) phases constructed from layered and bonded sp3- and sp2 nanostructural units, and provide a framework for classifying the members of this new class of materials. The nanocomposite structures are identified within both natural impact diamonds and laboratory-shocked samples, and possess diffraction features that have previously been assigned to lonsdaleite and post-graphite phases. The diaphite nanocomposites represent a new class of high-performance carbon materials that are predicted to combine the superhard qualities of diamond with high fracture toughness and ductility enabled by the graphitic units and the atomically defined interfaces between the sp3- and sp2-bonded nanodomains.

I nterest in new carbon allotropes is driven by the desire to identify novel materials for advanced technologies as well as to understand the role and fate of carbon phases in planetary processes. 1 The stable form of carbon at ambient conditions is layered hexagonal graphite with sp 2 bonding. 2 The nanomaterial graphene, which has remarkable electronic and mechanical properties, consists of single or very few stacked layers of the graphite structure. 3 Following highpressure (HP) and high-temperature (HT) treatment, graphite and graphitic carbon are transformed to sp 3 -bonded diamond that forms a superhard material with technological importance. This material has tetrahedrally bonded carbon atoms that are covalently linked to form six-membered rings in a "chair" conformation, and the atoms are stacked into layers according to a cubic (c) close-packing arrangement or the 3C stacking polytype. 2 A metastable sp 3 -bonded carbon allotrope identified within natural impact diamonds and laboratory samples created under static HP and HT or shock conditions is "lonsdaleite", a dense crystalline form displaying diffraction features consistent with hexagonal symmetry. 4−12 Observations of such hexagonal diffraction features provide an important mineralogical marker for the P and T conditions recorded in diamonds from bolide impact sites. 5,8,13 Lonsdaleite was interpreted as a 2H polytype with carbon layers stacked in a hexagonal fashion (h), although recent analyses suggest that natural and experimentally produced lonsdaleite samples are more accurately described as nanotwinning and stacking-disordered sequences of c and h units. 14−18 Density functional theory (DFT) calculations as well as experimental measurements suggest that the hardness and other mechanical properties of lonsdaleite may be superior to that of cubic diamond, 19 thus motivating the search for identifying the sp 3 -bonded hexagonally stacked polytype among natural materials, or preparing it in the laboratory. 20,21 Computational studies suggest a wide range of novel carbon polymorphs that might be produced in the laboratory or discovered in natural samples. 22 −24 In particular, several metastable carbon phases have been identified as being associated with the transition between sp 2 -and sp 3 -bonded structures. 25−30 It was first shown by analysis of high-resolution transmission electron microscopy (HRTEM) images that hexagonal features appearing in the diffraction patterns of impact diamonds could be interpreted as cubic (3C) diamond containing a high density of stacking faults associated with nanoscale twinning. 14 Analysis of the X-ray diffraction profiles of natural and laboratory-shocked diamond samples using the MCDIFFaX modeling technique later correlated the appearance of lonsdaleite features with different proportions of h vs c layer stacking within diamond polytype structures. 16,17,31 The average degree of hexagonality (Φ h ) and stacking disorder among the c and h layers can be represented on a stackogram plot. 16,17,31 These analyses were based on the assumption that the structures contained only sp 3 -bonded carbon atoms.
Introducing hybrid materials with sp 2 -and sp 3 -bonding give rise to additional features into the images and diffraction patterns. From HRTEM studies of an extraterrestrially shocked meteorite, Garvie et al. 27 described crystallites composed of few-layered graphene domains embedded within and exhibiting a topotactic relationship to slabs of diamond. Mixed sp 2 -and sp 3 -bonding is also observed in amorphous carbon forms with important technological applications. 32 For example, amorphous diamond-like carbons exhibiting a high degree of tetrahedral bonding contained within an sp 2 -bonded graphitic matrix can achieve hardness values approaching that of diamond, 33 while "hard carbon" materials with a similar nanocomposite structure are being developed for their reversible Na-ion intercalation properties as battery anodes. 34 A two-dimensional (2D) diamond-graphene crystalline nanostructure termed "diaphite" was proposed to form following photoexcitation of graphite. 35,36 However, the existence of extended diaphite nanostructures within the sp 3bonded matrix of diamond has not been discussed.
Here we show evidence for two families of sp 3 -/sp 2 -bonded carbon nanostructures that form the basis for a new class of crystalline nanocomposites that we call diaphite allotropes. We find that these structures occur within natural as well as laboratory-shocked samples and that features of their diffraction signatures match those previously reported for lonsdaleite as well as the postgraphite phases including Mcarbon, 7,37−40 thus motivating a reappraisal of previous reports of these materials. Despite its high compressive and tensile strength, diamond exhibits a low fracture resistance as a result of crack propagation along cleavage planes. However, inclusion of diaphite nanocomposite structures would allow the material to absorb or deflect incipient crack formation, thus increasing its fracture resistance. 41 HRTEM images of impact diamonds and laboratoryshocked graphite reveal the intrinsic structural complexity of these crystalline carbon materials. Nanostructures that exhibit lattice fringe spacings of both ∼3.4 Å (closely matching the {001} reflection of graphite) and ∼2.1 Å (consistent with both {111} diamond and {100} graphite) have been observed in both natural and synthetic samples (Figures 1, 2, and S1− S3). However, instead of suggesting the presence of independent graphite, lonsdaleite, or diamond within the  Figure S2) from a Popigai diamond indicate mixed sp 2 (graphitic) and sp 3 (diamond) bonding.  S1, and S2). Their lateral extent normal to the layer stacking direction ranges up to a few nanometers, and they terminate within the sp 3bonded lattice. We refer to this as a type 1 diaphite nanostructure.
A TEM image of diaphite is a 2D projection of superimposed nanodomains. As a result, the graphitic and diamond regions are observed most clearly in thin (<20 nm) sections ( Figure 1a). The DFT structure model of few-layered graphene sandwiched between {111} diamond slabs ( Figure  1b) not only reproduces the image contrasts consistent with type 1 diaphite but also reveals the interface corresponding to the Pandey (2 × 1) reconstructed surface. 42 In thicker (<50 nm) samples, the graphene and {111} diamond fringes are less visible, but their contributions can be detected in fast Fourier transforms (FFT) of the HRTEM data (Figures 1c,d and S1). Our electron energy-loss (EELS) spectra show the presence of both sp 2 -and sp 3 -bonded carbon coexisting within the type 1 diaphite materials (Figure 1e). The peak near 285 eV represents electronic transitions from 1s core states to unoccupied π* (2p z ) states of sp 2 -bonded carbon. This peak is most intense in those regions where graphitic domains are clearly visible in the HRTEM images ( Figure S2).
A second diaphite nanostructure, which we call type 2 diaphite, consists of hexagonally arranged graphitic carbon layers inserted within and bonded at high angles to the sp 3bonded diamond surfaces (Figures 2 and S3). HRTEM images of this diaphite are characterized by ⟨121⟩ diamond domains and subnanometer-sized regions containing ∼2.1 Å fringes arranged in a hexagonal pattern (Figures 2a,b and S3). Both the HRTEM images and the FFT indicate a topotaxial relationship between the two domain types (Figure 2b). A similar image from the Canyon Diablo meteorite had previously been interpreted as two-and four-layer-thick {113} diamond twins. 14 However, our DFT calculations described below and further comparison with the simulated HRTEM images indicate that this nanostructure corresponds to a nanocomposite consisting of sp 2 -and sp 3 -bonded carbon regions (Figure 2c−e).
We performed DFT calculations on atomic models of type 1 and 2 diaphite structures (see Supporting Information for details) to reveal the structural relationships between the sp 2and sp 3 -bonded nanodomains and the stability of our diaphite structures relative to other carbon allotropes ( Figure 3). In both structure types, the models were constructed to include varying amounts of diamond (d) vs graphene (g) content (Figure 3a,b). The relative proportion of interface regions between the two structural units was varied by modifying the size of the unit cell, with increasingly large cells producing a Nano Letters pubs.acs.org/NanoLett Letter lower density of interfaces. The two types of diaphite nanostructures give rise to the range of image and diffraction features observed experimentally in natural and laboratoryshocked samples (Figures 1, 2, and S1−S3). One striking aspect of the model structures is the significantly smaller interlayer distances observed for the few-layered graphene domains within type 2 diaphites (∼3 Å). This spacing indicates that the graphene layers are compressed within the nanocomposite structure (Figure 3b), and their bonding, stacking, and electronic properties may be altered from those of bulk graphite or graphene. Although a similar interlayer contraction appears to be predicted for type Nano Letters pubs.acs.org/NanoLett Letter 1 diaphite (Figure 3a), this arises since the dispersion correction used in our DFT calculations overestimates the attractive forces between sp 2 layers. However, that phenomenon does not affect the type 2 structures. The {100} spacings of the graphene layers (2.14 Å) in type 2 diaphites are also expanded relative to those of free-standing graphene (2.12 Å). This lattice spacing closely matches the {100} reflection (2.18 Å) that has been assigned to lonsdaleite. 4−13 The relative energies of type 1 and 2 diaphite nanostructures were plotted on an energy (E)−volume (V) chart ( Figure 3c) and compared with 2H graphite, cubic (3C) and hexagonal (2H) diamond, a range of (... c n h m ...) stacking disordered diamond polytypes, and >200 other metastable carbon structures reported in the SACADA database. 23 The diagram also plots estimated pressures as the E(V) slopes that might be attained in static or dynamic compression experiments, including those encountered during planetary bolide impact events. As the relative proportions of sp 2 -and sp 3bonded domains within the diaphite model structures are varied, the volume evolves systematically between the locus of points for 3C to 2H diamond polytypes and 2H to 3R graphite. The energy can be varied both by changing the relative contribution from each domain and by increasing the size of the unit cell, which has the effect of lowering the density of interface regions resulting in a lower overall energy. Extrapolating calculated results for type 1 and 2 structures with fixed d or g components toward infinitely large cells (i.e., d = 1, g = 1, 2, 3, ... → ∞), allows the prediction of the E−V limits in which the different structures may be produced under equilibrium conditions (Figures S5 and S8). In the limit of d = 1 and g = ∞, the type 1 structures attain the same energy and volume as 2H graphite. However, our type 2 structures always have a lower volume than 2H graphite due to the contracted interlayer distance between graphitic planes, and the locus of points for d = n, g = ∞ structures does not extrapolate to bulk 2H graphite. Our models are constrained by size and threedimensional periodicity, so different structural behavior might be observed in real materials.
The range of type 1 diaphite structures depicted as orange squares in Figure 3c spans a locus of points that could be accessed by compression of graphite to between approximately 10 and 40 GPa, neglecting finite temperature effects and activation energy barriers. Large cells containing low densities of interfaces can achieve energy values that are competitive with the transition pressures from 2H graphite to diamond polytypes with a high degree of hexagonality ( Figure  3c). We demonstrate this by calculating the type 2 (g = 1, d = 41) structure ( Figures S5−S8), which has a lower energy (in eV atom −1 ) than 2H diamond and falls below the transition pressure from 2H graphite to 2H diamond (expanded region of Figure 3c). This plot reveals that the formation of diaphite nanocomposite structures could provide low-energy solutions existing between fully sp 3 -or sp 2 -bonded cubic-hexagonal diamond and graphite phases, while maintaining fully saturated C−C linkages. These diaphite nanostructures might also be encountered during recovery to ambient pressure of sp 3 -bonded cubic-hexagonal diamond polytypes or by heating metastable sp 3 carbon phases at ambient pressure, as they return toward the thermodynamically stable graphitic phase. These structures might also form during diamond surface graphitization according to the ab initio theoretical prediction of De Vita et al. 43 The question arises as to whether the diaphite structures can be revealed from X-ray diffraction and Raman data. Figure  4a shows the diffraction patterns of several Popigai diamonds previously fitted using h/c diamond stacking disorder models. 17 This approach enabled us to reproduce the diffraction pattern including the diffuse diffraction features at ∼ 20 and 35°2θ MoΚα. From these fits, hexagonality indices (i.e., percentages of hexagonal diamond stacking) were obtained (Figure 4a). The only diffraction features that were previously not included in our fits were the sharp, but fairly weak peaks at ∼12°2θ ΜoΚα, which were observed for the two samples with highest hexagonal index. These peaks are commonly attributed to the 00l peaks of graphite. To test if this peak and perhaps the other diffuse features could arise from diaphite structures, we extended our DIFFaX model to contain type 1 diaphite structures, which were found to be most abundant in our samples. This new DIFFaX model includes c/h diamond stacking, the hexagonal/rhombohedral stacking of graphite, and the mixing of diamond and graphite sequences within the structure (Figures S9−S11). Figure 4b shows calculated diffraction patterns obtained by mixing different stacking-disordered graphite and cubic diamond sequences. The top diffraction pattern is that of stackingdisordered graphite showing that the feature at ∼12°2θ ΜoΚα remains sharp despite the disorder in the sp 2 -bonded layered material. Upon mixing into cubic diamond sequences in a random fashion, this feature is predicted to broaden and shifts toward higher angles as increasing amounts of diaphite sequences are created within the diamond lattice. The sharp peak at ∼12°2θ ΜoΚα that is most clearly observed in the experimental data sets is thus thought to originate mainly from extended (potentially stacking disordered) graphitic regions included within the sample or from separate graphite inclusions. Diffraction intensity to the low-angle side of the main peak (∼20°2θ ΜoΚα) occurs for the cubic diamond containing even relatively small amounts of graphitic diaphite sequences. The appearance of this shoulder is typically interpreted as c/h stacking-disordered diamond or "lonsdaleite" (Figure 4c). However, the diffraction intensity appearing on the high angle side of the main peak (∼ 20°2θ ΜoΚα) and that occurring at ∼ 34°2θ ΜoΚα cannot be reproduced by mixing graphitic sequences with diamond, and so these features are likely to be characteristic of the sp 3 -bonded, c/h stacking-disordered, structures. We conclude that although some contribution from type 1 diaphite nanostructures may be present in those samples that we and others have investigated to date, its quantification using X-ray diffraction is challenging. The type 1 diaphite structures observed in the HRTEM images of Figure 1 are only a few nm in width, which explains why they might be difficult to capture in X-ray diffraction studies of the bulk materials.
The implementation of type 2 diaphite into a DIFFaX model is difficult since the stacked layers consist of regions of sp 3 -and sp 2 -bonded carbon. In Figure S12 the calculated diffraction pattern of a type 2 diaphite unit cell with g = 7 and d = 7 is shown. A characteristic feature of a type 2 diaphite is the quite short interlayer distance within the graphitic domains, which results in a Bragg peak at ∼13.4°2θ ΜοΚα (corresponding to ∼3 Å). The absence of such a feature in our XRD data can be attributed to the overall low concentration of type 2 structures. Furthermore, due to the small domain sizes, any type 2 diaphite diffraction features would be expected to be very broad.

Nano Letters
pubs.acs.org/NanoLett Letter The Raman spectrum for the sample with the highest hexagonality value reveals a broad and weak feature in the region of the G band, which could imply the occurrence of diaphite domains within the material, although it could also be associated with a bulk disordered graphitic phase (Figure 4d). In summary, analysis of X-ray diffraction and Raman spectral profiles is most sensitive to determining the overall c/h diamond stacking rather than revealing the existence of diaphite sp 2 −sp 3 bonding within the samples.
The presence of diaphite nanostructures can help us to understand the large number of sp 2 −sp 3 -bonded carbon phases that are reported to form during the graphite to diamond transition, initiated by static or shock compression or during metastable syntheses from precursor compounds during chemical vapor deposition. Static compression of graphite at room T suggests a sluggish transformation to a metastable phase at ∼19 GPa: This material is optically transparent, exhibits superhard properties rivalling diamond, and returns to a mainly sp 2 -bonded material upon recovery. 7,37−40 Proposed structure models for this material include "M-carbon", which according to Wang et al. 39 contains only sp 3 -bonded carbon atoms. However, EELS data presented by Mao et al. 40 suggest that cold compressed graphite transforms at ∼17 GPa to a superhard phase with mixed sp 2 -and sp 3 -bonded C atoms. Furthermore, a diffraction line that appears between the (100) and (101) peaks of graphite (∼ 2.06 Å), that was used to identify Mcarbon, 40 is consistent with the diamond-type contribution from diaphite nanocomposite structures reported here ( Figure  4c). Formation of diaphite nanostructures during compression of graphite could provides an efficient pathway to initiate transformation to sp 3 -bonded phases such as diamond, lonsdaleite, and M-carbon.
Shock studies carried out along the principal Hugoniot of graphite to 100 GPa show an inflection near 20 GPa leading toward the pressure−density relations of sp 3 -bonded structures, although only very small quantities of crystalline diamond are recovered at ambient conditions. 44 Laser shock experiments combined with in situ X-ray diffraction indicate formation of cubic diamond above 50 GPa and of hexagonal diamond above 170 GPa. 20 The evidence for the latter has been questioned as the doublet observed in the XRD patterns could be interpreted as arising from diamond structures experiencing different residual strain regimes. 17 An alternative explanation for data presented in ref 20 could be formation of diaphite nanostructures during dynamic compression, for which diffraction data would also present a doublet peak (Figure 4c). A further study of graphite-diamond transformation using gas-gun shock experiments combined with synchrotron X-ray diffraction concluded that only elastically strained hexagonal diamond was produced above 50 GPa. 21 The projection of these X-ray diffraction data most likely corresponds to that of type 2 diaphite shown in Figure 2. The data from these previous studies are consistent with formation of materials containing diaphite nanostructures, both within the compression phase or the rarefaction wave associated with shock studies, or during their recovery to ambient conditions. Increased hardness and improved mechanical properties have been suggested for lonsdaleite diamond structures containing a high proportion of hexagonal stacking, 45 consistent with results for cubic-hexagonal cBN/wBN polytype assemblies. 19 Although our DFT calculations for bulk, shear, and Young's moduli of different type 1 and 2  In conclusion, we provide evidence for the existence of novel type 1 and type 2 diamond-graphene nanocomposites within diamond materials recovered from natural impact materials and laboratory-shocked graphite. These nanostructures are characterized by the intimate association of sp 3 -(diamond) and sp 2 -bonded (graphene) domains, and the building units have a lateral width of a few nm. Their diffraction signatures exhibit features that are consistent with those previously reported for lonsdaleite as well as the postgraphite "M-carbon" phase. The calculated E(V) relationships reveal that these nanostructures provide low-energy solutions to structural transformation between fully sp 2 -and sp 3 -bonded carbon allotropes that can be sampled under static and dynamic compression, including during the rarefaction wave associated with shock compression and during recovery to ambient conditions. The nanocomposite nature of the diaphite structures is expected to lead to mechanical behavior that preserves the superhard and incompressible properties of the sp 3 -bonded units, while leading to fracture toughening due to the angular flexibility and tensile resistance of the graphitic domains bonded to them.