Ultrafast shock synthesis of nanocarbon from a liquid precursor

Carbon nanoallotropes are important nanomaterials with unusual properties and promising applications. High pressure synthesis has the potential to open new avenues for controlling and designing their physical and chemical characteristics for a broad range of uses but it remains little understood due to persistent conceptual and experimental challenges, in addition to fundamental physics and chemistry questions that are still unresolved after many decades. Here we demonstrate sub-nanosecond nanocarbon synthesis through the application of laser-induced shock-waves to a prototypical organic carbon-rich liquid precursor—liquid carbon monoxide. Overlapping large-scale molecular dynamics simulations capture the atomistic details of the nanoparticles’ formation and evolution in a reactive environment and identify classical evaporation-condensation as the mechanism governing their growth on these time scales.

scales obtains Hugoniots that are equivalent to longer time scale experiments, e.g. using gas gun driven impactors, which are known to generate final states over 100s of ns to µs. Substantial existing work [3][4][5] demonstrates that for materials which equilibrate rapidly prior to the onset of chemistry (e.g. liquids) or undergo very rapid chemistry, ultrafast compression does obtain an equivalent Hugoniot. A comparison of the gas-gun Hugoniot data of Nellis et al. 6 and the measured Hugoniot from this work at pressures below 16 GPa (see Supplementary Figure 2) demonstrates this here as well for liquid CO.
Supplementary Figure 2: Ultrafast shock Hugoniot data compared with gas gun data from Nellis et al. [6]. The line is a third order spline interpolation of the gas gun data (including points outside of the domain of the graph). Error bars are single prediction errors from a linear fit to the ultrafast data. Source data are provided as a Source Data file.
Probe signal loss through the sample at shock pressures higher than ≈16 GPa is likely due to absorption and Rayleigh scattering caused by the precipitation of liquid carbon nanoparticles. Attenuation through Rayleigh scattering is however orders of magnitude smaller than from absorption for particle sizes in the nanometer range 7 , so we neglect it here. To model probe absorption subsequent to shock compression we assume (as shown schematically in Supplementary Figure 3) that carbon nanoparticles form behind the shock front and grow to sufficient size to exhibit liquid carbon bulk dielectric response on a time scale of tens of ps. Thereafter, consistent with simulations, we assume that their volume fraction remains approximately constant at ≈10%, while the volume of compressed material grows with shock propagation, thus increasing the length of the nanoparticle-containing region through which probe light must pass.
Consequently, the probe reflection from the piston (Al/sample) interface will be absorbed according to a function of time, is the absorption constant, and ( ) is the thickness of the absorbing material. (We neglect here the reflection from the shock front, which is typically much smaller than the piston reflection).
We approximate the (time-dependent) thickness of the absorption layer by: The absorption constant is given by: , where rn is the number density of nanoparticles and the absorption cross section is given by 7,8 : Here is the volume of a nanoparticle with dielectric function = = + I ; N is the relative dielectric constant of the surrounding medium, and is the probe wavelength. This expression is valid when the nanoparticle diameter (d) is much smaller than , which should be well satisfied here since the simulations suggest that d =1-10 nm while recovered samples have characteristic particle sizes of 5-30 nm, and λ = 800 nm. Due to the particle volume dependence, the absorbance is constant for a fixed volume fraction of nanoparticles, thereby removing an explicit dependence on the nanoparticle size. To estimate σabs we employ the dielectric function of liquid carbon inferred from femtosecond spectroscopy 9 at conditions similar with our experiment and estimate N = 2.6 using the Clausius-Mossotti relation along with thermochemical calculations that yield the molecular composition (approximately equimolar mixture of CO and CO2 with trace amounts of O2) and density (≈2 g/cm 3 ) of the background fluid. This yields absorption depths α -1 of 380 nm and 514 nm, corresponding to the two possible liquid carbon dielectric functions proposed in Ref. 9; these two values generate the intensity decay grey band in Fig. 2 (main text).
The above analysis suggests that carbon nanoparticle formation lags only ≈50 ps behind the shock arrival in the sample, as indicated by the strong deviation of the reflectance from the Al background. This is likely an upper bound, since smaller particles may exist at earlier times, but not exhibit sufficiently metallic behavior to affect the absorption.
Imaging, diffraction, and EELS data for a single agglomerate are shown in Supplementary  Figure 3c in the main manuscript) is compared with reference spectra from graphite, nanodiamond, and amorphous carbon, and is a close match to the amorphous carbon reference.
For the sake of completeness, we also include below a comparison of the original and marked TEM images shown in Fig. 3a, b (main text).

) and yellow c) boxed regions in a). Diffraction signal d) from agglomerate region in c) and radial integrations e) of d) and
diffraction signal from bare SiO2 membrane. Diffraction pattern from agglomerate is consistent with amorphous carbon. Carbon K edge EELS spectra f) from agglomerate in a) and reference carbon materials ("Amorphous" corresponds to the lacey carbon of TEM grids. Agglomerate spectrum matches well that of amorphous carbon. Scale bars are 50 nm (a -c) and 5 1/nm d). Source data are provided as a Source Data file.

Supplementary Note 3
The resulting Raman spectrum shows bands representative of disordered carbon -both the D and G bands are observed at 1354 and 1566 cm -1 , respectively -see Supplementary Figure 6; no 2D+G band is present at +2500cm -1 . Comparisons of this Raman spectrum with the published spectra of various forms of carbon suggest that the solid residue is predominantly amorphous carbon, with possible nanodiamond content.
Previous studies have shown that the G-band is associated with the presence of sp 2 bonds, while the Dband points to structural disorder and/or sp 3 bonding [10][11][12][13][14][15] . We also note the presence of a low intensity peak at 1190 cm -1 , which was previously observed in the spectrum of nanocrystalline diamond 15 . Besides the carbon Raman bands we also observed a broad Raman background at high frequencies, which is likely due primarily to amorphous nanocarbon surface effects. The origin of the sharp Raman modes at lower frequencies (below 1000 cm -1 ) is not entirely clear at this time. Al2O3 and/or Al4C3 could plausibly form due to the Al ablator used in these experiments. The observed Raman modes are not due to Al2O3 and we could only partially, not conclusively, assign them to Al4C3. It is worth noting that neither Al2O3 nor Al4C3 have Raman peaks close to the 1354 cm -1 (carbon D band) peak.