How the Toughest Inorganic Fullerene Cages Absorb Shockwave Pressures in a Protective Nanocomposite: Experimental Evidence from Two In Situ Investigations

: Nanocomposites fabricated using the toughest caged inorganic fullerene WS 2 (IF-WS 2 ) nanoparticles could o ﬀ er ultimate protection via absorbing shockwaves; however, if the IF-WS 2 nanomaterials really work, how they behave and what they experience within the nanocomposites at the right moment of impact have never been investigated e ﬀ ectively, due to the limitations of existing investigation techniques that are unable to elucidate the true characteristics of high-speed impacts in composites. We ﬁ rst fabricated Al matrix model nanocomposites and then unlocked the exact roles of IF-WS 2 in it at the exact moment of impact, at a time resolution that has never been attempted before, using two in situ techniques. We ﬁ nd that the presence of IF-WS 2 reduced the impact velocity by over 100 m/s and in pressure by at least 2 GPa against those Al and hexagonal WS 2 platelet composites at an impact speed of 1000 m/s. The IF-WS 2 composites achieved an intriguing inelastic impact and outperformed other reference composites, all originating from the “ balloon e ﬀ ect ” by absorbing the shockwave pressures. This study not only provides fundamental understanding for the dynamic performance of composites but also bene ﬁ ts the development of protective nanocomposite engineering. were employed to examine the morphology and structure characteristic of both as-prepared and postshock composites, particularly to reveal the structure changes by the shock process. Full methods and any associated references are available in the Supporting Information.

D riven by the practical needs, the development of advanced shock-absorbing composites based on nanomaterials has been proposed for over a decade. 1 Nanocomposites, either being stronger and/or more absorbing than their single-phased counterparts, are highly desirable for next generation protective armory applications. 1 In particular, as the toughest cage structures, individual inorganic fullerene WS 2 (IF-WS 2 ) and IF-MoS 2 nanoparticles are superb shock absorbers and can survive shockwave pressures up to 25 GPa, 2,3 whereas nanotubes can withstand 21 GPa shockwaves 4 and exhibit tensile strengths above 16 GPa. 5 Incorporated into a matrix, such composites have been predicted to be a few times better than the best impact resistant materials (e.g., SiC, B 4 C) used in protective armor, 6 which will boost their applications well beyond the proven excellent electronic and tribological properties. 5,7−13 Hence, the structures and extraordinary shock-absorbing properties of these IFs have been associated with an emerging class of energy-absorbing protective nanocompo-sitesnamed "NanoArmour", 6 but this hypothesis highlighted by Hogg in Science 1 has yet to be validated experimentally.
Precisely evaluating the shock/ballistic characteristics and gaining fundamental understanding toward the performance of complex composite systems are extremely challenging because of the difficulty to carry out real-time investigations at highspeed impact. The recent significantly advanced high-speed digital video cameras in combination with computer simulations are probably the best techniques to date; 14,15 however, the resolution at the macroscopic level restricts it from revealing the real-time shock behavior of nanocomposites at the super short moment with details, which also makes the simulations hard to be validated at the nanoscale.
IF-WS 2 nanoparticles have been incorporated with a wide range of polymer matrices previously, 16,17 and they have effectively improved thermal, mechanical, and tribological properties, even the toughness of the composites; 17 however, there were hardly any reports focusing on other matrices reinforced by these IF nanoparticles. Here, we report two very innovative and specialized in situ impact techniques in geophysics: the time-resolved luminescence 18,19 and the velocity interferometer system for any reflector (VISAR) techniques, 20,21 combined with postshock analyses, to investigate the impact performance of such IF-WS 2 containing Al model nanocomposites and to scrutinize their precise roles in impacts, by contrasting with the conventional bulk 2H-WS 2 (the hexagonal WS 2 platelets) composites.

RESULTS AND DISCUSSION
The as-produced IF-WS 2 particles were around 30−80 nm in size (mostly 40 nm), exhibiting a typically multilayered closecaged structural feature (Figure 1a). 22 We used a hot-press to create the IF-WS 2 -reinforced Al model nanocomposites for this research, whereas pure Al and 2H-WS 2 -reinforced Al nanocomposites were also produced as referenced counterparts, following the same process. After being polished and slightly etched, the particle distributions in the Al matrix were clearly shown under scanning electron miscroscopy (SEM). The IF-WS 2 (white phases) were fairly uniformly distributed along the grain boundaries among the larger Al grains (dark phases) (Figure 1b), given the process limitation of powder metallurgy and such a high loading in the matrix. In general, the matrix Al grains in the IF-WS 2 composites were slightly smaller than thoe of the 2H-WS 2 composites and pure Al (Table S1 in  Supporting Information, SI). This showed directly the grain size reduction effect of the fillers in composites and was also an indirect indication of the good distribution of the fillers within the matrix, as a poor filler distribution would lead to dramatically irregular grain sizes of the matrix.
The VISAR experiments measure the particle velocities during impact, 20,23−26 and the detailed setup is described in Figure S1 (SI). A summary for each sample description and experimental parametres are also listed in Tables S1 and S2 in the SI.
We performed three series shots of VISAR measurements at impact velocities of 1.43 ± 0.03, 1.735 ± 0.025, and 1.062 ± 0.027 km/s, as shown in Figure 2a−c, respectively, to evaluate the different impact speed responses of the composites. In general, the window/sample interface particle velocity profiles displayed a sharp increase corresponding to the shockwave arrival, plateau for 0.7−0.8 μs, then followed by a decrease for 0.5−0.6 μs, and a second plateau at nearly half velocity of the first plateau except for samples B3 and C3. The measurements at ∼1 km/s were successful only for a short period of ∼1.2 μs. Using the window shock data (fused quartz), 27 Figure 2c. The initial shock pressure for the sample was slightly higher than the calculated value due to the impedance mismatch between the sample and the window. When the projectile traveled at a speed of 1.4 km/s, there were barely any differences among the samples (Figure 2a). Similarly, the performance of samples being struck at the initial velocity of 1.735 km/s largely resembled those of 1.4 km/s, with minor differences (Figure 2b). The tested samples displayed no distinct differences between impact speeds of 1.43 and 1.735 km/s, and all the resulting velocities were quite stable, except sample D appeared to have lower velocities after shock ( Figure  2a,b).
However, when the projectile traveled at a speed of around 1 km/s, the samples behaved differently. Samples A, B, and C were still quite similar, but samples D and E (containing IF-WS 2 nanoparticles) exhibited vastly different features ( Figure  2c). The velocity signals started to appear at around 2.8 μs, with starting values of about 0.8 km/s for samples D and E, but around 0.9 km/s for samples A, B, and C, that is, a 0.1 km/s (100 m/s) difference. Significantly, an immediate increase in velocities for samples A, B, and C was observed at the time of 3 μs, which peaked at 1 km/s for sample B and at 0.95 km/s for sample A, compared to only 0.85 km/s for samples D and E. Following these sudden surges, the velocities underwent an unstable period (3.1−3.4 μs) and were seen to drop slowly, in contrast to the very steep slopes observed for samples being shocked at 1.43 and 1.735 km/s, although both also demonstrated a significant decay of about 60% in the second state of shot, which is indicative of absorption of shock energy. Because of these significantly different features that were exhibited at the low impact speed region, our second in situ investigations were focused on the 1 km/s velocity.
The time-resolved luminescence experiments recorded the streak image of ruby luminescence when the shockwaves struck the test samples and the ruby window ( Figure S2 in SI), and the results are shown in Figure 3 at impact velocities of ∼1.0 km/s.

Article
It is imperative to note that the points at which the shifted R1 and R2 lines appeared (due to the initial impact-induced compression shockwaves), until the original ambient R signals almost completely disappeared, corresponded to the entire period when the shockwaves had just finished traveling through the ruby window and reached the free surface. Then the shockwaves were reflected back into the ruby window, as a release wave, which was represented by the moment the shifted R lines gradually lost their intensity. The original R lines regained the intensity until the release wave reached the sample/ruby interface. The negatively shifted R line signals observed at around 1 μs (Figure 3a,c) were attributed to the interaction of rarefaction from ruby edges. 28 Based on the pressure-shift characteristics of the ruby crystal, 19,29 we can then determine the shock pressures on the ruby, corresponding to each exact stage of the impact compression.
In this work, the pressure was calculated based on the stronger R1 signal. For sample B, the 3.29 nm shift of R1 (Figure 3b) corresponded to a shock pressure of ∼12.4 GPa. This was below the Hugoniot elastic limit (HEL) of ruby, which was 14 GPa, suggesting the ruby window was shock compressed elastically. 19 In addition, the shifted R1 line is very strong, with intensity almost 66% of intensity for the ambient R1 line. Indeed, the strong and sharp shifted R1 line, as well as the negatively shifted R lines, represented the typical elastic region signals.
Similarly, a strong and sharp R1 line shift had also been observed for sample A ( Figure S3 in SI) and samples C (Al, 30 wt % 2H-WS 2 , Figure 3c,d), as well as the negatively shifted lines, suggesting the typical elastic signals. From the luminescence spectra, we obtained a 3.33 nm shift of the R1 signal for both samples A and C, which corresponded to the shock pressure of 12.5 GPa in ruby, whereas the intensity is 68% compared to that of ambient R1 for sample A and 78% for sample C.
Sample D (Al, 20 wt % IF-WS 2 ), however, exhibited different features if we compare Figure 3e with Figure 3a: the shifted R lines were very blurred and short, and no negatively shifted lines were observed with only slight yet blurry vibrations of the lines. In Figure 3e, the start of shock compression and the final shockwave release took place within a period of 500 ns (between 0.5 and 1 μs in the time scale), whereas in the case of sample B, it lasted for almost 1000 ns (0.5−1.5 μs). The shorter interval implies a faster release of shocked state in the case of sample D, suggesting that the ruby window and the sample have gone through different structural changes from those cases for samples A, B, and C.
The extracted wavelength profiles of sample D (Figure 3f) showed a 3.45 nm shift for the R1 signal, which corresponded to the shock pressure of 12.9 GPa in ruby. However, the compression-shifted R1 signal for sample D (Figure 3f) was much weaker than that of sample B (Figure 3b), consistent with the more blurred signals in the streak image. Significantly, the intensity of the shifted R1 was much weaker, with only 32%   2 μs (a,c,e,f). The two bright parallel vertical lines were the ambient ruby R lines, appearing in the upper and lower part of the image, where the left one was R2 at 692.9 nm (gray) and the right one was R1 at 694.3 nm (highlighted in red). When the ruby crystal was shock compressed, shifted R lines appeared, as shown in the upper part of the image. The ambient R lines then appeared again in their original positions, and negatively shifted R lines were also observed subsequently, followed by another slightly shifted signal and then regained their original intensities and positions, and finally came back to normal ambient R signals. The ruby luminescence spectrum at a chosen instance was obtained by taking the horizontal cross section of the streak image at a time frame of ∼0 .1 μs (b,d,f,h). The ruby luminescence wavelength recorded in the horizontal axis increased from left to right, with a full width range of 15 nm for the chosen instance.

ACS Nano
Article intensity compared to that at ambient pressure (Figure 3d), and the very weak, broad, and short characteristics of such shifted R lines were indicative of an inelastic region signal. 19 This is very surprising, as the shock pressure was below the HEL of ruby.
Sample E (Al, 30 wt % IF-WS 2 ) shock compressed at an impact velocity of 0.97 km/s also exhibited a streak image and luminescence spectra (Figure 3g,h) similar to that of sample D, with weak and blur shifted signals and without negative shift. The shift of R1 signal was 3.21 nm, corresponding to a shock pressure of 12.1 GPa in ruby. The intensity of shift R1 is only 46% of that of ambient R1, close to 32% in sample D, indicating an inelastic region signal.
From the luminescence results, we can distinguish the two different types of behavior for our samples (Figure 3 and Table  1): samples A, B, and C exhibiting a stronger and sharper shifted R1 line accompanied by a negative shift, signifying the typical elastic region signal; samples D and E (containing IF-WS 2 ) having similarly very weak, broad, and short shifted R1 signal, evidencing the inelastic region characteristics. 19 These results have confirmed that for samples D and E, the ruby crystal windows responded inelastically in the elastic region below the HEL of ruby. Further, this suggests that the shockwave propagation characteristics in samples D or E were somehow changed, from a normal plane shockwave to nonplanar. It is interesting to find out that the target composite samples D and E have shown the inelastic responses at low shock pressures, as much as >2 GPa lower than the elastic limit.
In ceramics, fracture and dislocation are normally attributed to inelastic deformation, though the mechanism remains to be fully clarified. 30 In this case, by considering the complementary experiments for other composite samples, we believe that the inelastic response of the ruby window should be attributed to the intrinsic features of the IF-WS 2 nanocomposites. This is quite convincing because such phenomena were only observed for samples containing IF-WS 2 , despite the different concentrations and densities in the composites, not for pure Al and 2H-WS 2 containing composites. It is likely that the previously proposed shock-absorbing property of the IFs 2−4 in the nanocomposites have changed the original feature of the shockwaves, either through their hollow close-caged structure or working as a balloon/spring when they travel through the samples, resulting in the first set of compression-shifted R signals.
The results from both the VISAR and time-resolved luminescence agreed very well with each other; both confirmed similar performance of samples D and E (IF-WS 2 containing composites) and are different from other samples (either pure Al or 2H-WS 2 containing composites). Compared with the VISAR shots, in the time-resolved luminescence experiment, the pressure corresponding to an impact velocity of 1.0 km/s increases due to the high impedance of ruby. Actually, ruby shots in the luminescence experiment correspond almost to Figure 2a at 1.4 km/s (P = ∼13 GPa). VISAR looks at the interface velocity, but the ruby luminescence looks at the pressure it experiences. However, the behavior of samples D and E containing IF-WS 2 nanoparticles differs from others containing 2H-WS 2 , as shown by both VISAR shots and ruby time-resolved luminescence experiments. When the samples were struck by shockwaves with impact velocity of around 1.00 km/s, we have found that this type of IF-WS 2 containing composites had lower particle velocities under shock impacts

ACS Nano
Article and exhibited an unexpected inelastic signal at the elastic region of ruby. The immediate question is why and how the IF-WS 2 caused the above differences in the composites. To answer this question, we have managed to recover sample D after the VISAR test and conducted postshock analyses in an effort to understand what exactly happened to the composites during the shock, to unlock the secret roles of the IFs in the composites, and to reveal the reinforcing mechanism.
Compared with the X-ray diffraction (XRD) results of the original composite samples (for details, see Figure S4 in SI), the IF-WS 2 peaks of the postshock sample were quite similar to the preshock samples ( Figure 4A), and only the (002) peak was broadened and slightly shifted to higher angle, with higher relative intensity. 31,32 These composite samples also exhibited similar Raman features, and the Raman peaks at 349 and 419 cm −1 ( Figure 4B) were assigned to the 1E 2g and A 1g modes of IF-WS 2 , 32 respectively. The slight broadening of the 1 E 2g peak and sharpening of A 1g were observed in the postshock sample, which could be related with the increased nanometer-sized 2H-WS 2 content in the specimens due to shock damage/ breakage. 3,33,34 Figure 5a shows the transmission electron microscopy (TEM) image of an isolated IF-WS 2 nanoparticle in the Al matrix (ca. 70 nm in size), prior to impact tests. The morphologies of the IF-WS 2 in the as-prepared composites exhibited no difference from the loose IF-WS 2 nanoparticles (Figure 1a), though smaller particles (<40 nm) started to exhibit smooth olive-shaped structure and larger particles (>40 nm) adopted a rigid polyhedral shape. This shape difference could affect their performance under shockwaves, as stated in our previous work. 3 After being recovered from a high-impact velocity of 1.74 km/s, damages of different extents have been observed, and several different particles are shown in Figure 5b (1.74 km/s), where the particles were still attached to a large piece of Al matrix whose surface appeared to be cracked. There were also unusal solid dark phases that did not exist in the pristine IF-WS 2 , and it is most likely that they were the evidence left behind from the shockwave impact. Indeed, considering the planar shockwave propagation during impact, we knew that the shockwaves traveled along one direction within the composites. Therefore, based on the trajectory subjected to a similar impact direction, it was shown that the dark phases were postshock insertions (Figure 5b). Smaller intact particles observed on the top of these two broken larger IFs showed that smaller IF particles were indeed capable of sustaining higher shock pressures than larger particles even surrounded by a matrix in the composites. It is in good agreement with our previous studies based on loose particles and confirmed the implication that the smaller the IF size the better the shock-absorbing property. 2 This fact that smaller particles correspond to less extent of damage has also suggested that the shock-absorbing properties of the composites could have been improved if all the IF particles were of uniform and smaller size. Figure 5c (1.74 km/s) showed clearly a third body part insertion from the left into the hollow core of the IF-WS 2 particle (verified by rotation of the sample), and the dark insertion was confirmed to be loose Al grain by using an energy-dispersive X-ray spectroscopy line scan across the particles (bottom left inset in Figure 5c). Severe damages to the top particles were visible, as indicated by the arrows, though the bottom part survived the impact and appeared intact, with perfect WS 2 lattice fringe and smooth corners, akin to the pristine IFs. Nearly intact IF-WS 2 particles subjected to lower shock velocity (1.04 km/s) were observed as the dominant feature in the specimen, and the olive-shaped structure was well-preserved in postshock composites, as shown in Figure 5d.
The postshock characterizations, particularly the highresolution TEM images, have revealed the shockwave-induced breakages/damages of the larger IF-WS 2 particles. As discussed earlier, during the shockwave propagation through the nanocomposites, the breakage or collapse of IFs must also have absorbed the impact pressure in a way similar to other dense and conventional solid materials. In this regard, the stronger and the tougher the materials, the better the performance for the materials to stop penetration during impact. The IF-WS 2 and the 2H-WS 2 resulted in similar densities, filler distribution, hardness (indeed IF-WS 2 resulted in lower hardness than the 2H-WS 2 in composites), and other static mechanical properties for the composites (Table S1 in SI). Therefore, the dynamic performance differences (i.e., the changes in the pressure profiles and the inelastic behavior of the IF-WS 2 nanocomposites against composites containing the same fractions of 2H-WS 2 ) pointed to other new directions, and they must be the consequences of the structural differences between IF-WS 2 and 2H-WS 2 reinforcements. The IF-WS 2 Several different particles were shown in (b); the particles were still attached to a large piece of the Al matrix whose surface appeared to be cracked, due to the Ar ion bombardmant during the TEM specimen thinning process. This was a good indication of the interface connection between the IF-WS 2 and the Al matrix grains. In (c), severe damages were visible at the top part of the particle: missing layers from the inner core (the left arrow), partially broken and mismatched shells or layers (the middle arrow), and peeled off layers (the top right arrow). However, the bottom part survived the impact and appeared intact, demonstrating features similar to those of the pristine IFs. Nearly intact IF-WS 2 particles were observed (d) as the dominant feature in the specimen subjected to lower shock velocity (1.04 km/s).

ACS Nano
Article nanoparticles in the composites reduced the particle velocity of the composites and led to the unexpected inelastic response of ruby at least 2 GPa below its HEL pressure of 14 GPa. The response of the ruby is a direct reflection of the performance during the impact; therefore, the inelastic phenomena we reported here have provided convincing evidence for the damping effect of the IF-WS 2 , at the exact moment of the impact on the composites.
To understand how this damping effect worked under dynamic shockwaves, we need to consider our recovery results again. First, it seems that the perfectly intact hollow structure of the IFs in the composite could act as molecular-sized voids to absorb the shock pressures, which could lead to the inelastic effect ( Figure 5b). Second, the multilayered cages could also act as a "balloon" via a spring-like response to absorb the shock pressure. 2 It is understandable that a spring-like effect will generate an accompanying bouncing effect, which should disturb the propagation of plane shockwaves in the IF composites (Figure 2c) but not in other (2H-WS 2 ) composites. Third, the partially damaged (verified by XRD and Raman in Figure 4) and the third party (Al) insertion into the hollow IF-WS 2 cage, as observed in Figure 5c, could absorb energy and lead to more unstable velocities ( Figure 2e) and quicker velocity drops (Figure 3e,f) compared to that of normal 2H-WS 2 composites. During the shockwave propagation within the sample thickness direction, the observed layer peeling off, cage collapse, and Al grain insertion into the cage would have taken place gradually within the entire process, including the bouncing wave propagations, hence leading to unstable particle velocities. Finally, whereas these damages/breakages gradually happened, the damping effect should be taking place concurrently, thus resulting in faster velocity drop than that with either pure Al or 2H-WS 2 composites, which only suffered from conventional deformations and cracks after being subjected to the same shocking process. We believe this indeed was the case for the IF-WS 2 composites, not for 2H-WS 2 or even other nanoparticulate-reinforced nanocomposites.

CONCLUSIONS
To conclude, the present results provide the experimental evidence to show that the IF-WS 2 nanoparticles effectively dampened the peak shock pressure, reduced the measured velocity by about 40% against the calculated velocity for the second state, changed the shockwave propagation at all impact speeds, and led to the unexpected inelastic response in the composites at low impact speed. Based on this real-time information, assisted with postshock analyses of the composites, we have attributed the interesting performance to the intrinsic shock-absorbing properties of IF-WS 2 nanoparticles that originate from their fullerene-like hollow cages. The 100 m/s particle velocity reduction out of 1000 m/s and the over 2 GPa pressure reduction to achieve an inelastic impact can be viewed as directly converted to equivalent reinforcing effects of this class of shock-absorbing composites, beyond the model Al matrix. The methodology used to gain these fundamental understandings about the intriguing roles of IF-WS 2 in composites during the exact moment of impact will significantly boost nanocomposite research and benefit future advanced nanocomposite design and engineering. More effort is required to create high-performance IF-WS 2 -reinforced composites to turn the "NanoArmour" science fiction into an engineering reality.

EXPERIMENTAL SECTION
IF-WS 2 -and 2H-WS 2 -reinforced Al composites were prepared, respectively, by effective ultrasonic mixing and hot-press sintering, as was the pure Al sample. Time-resolved particle velocities at the interface between the sample and fused quartz window were measured by a velocity interferometer, VISAR (VALYN MULTI-BEAM VISAR system), using a single stage propellant gun with a 30 mm inner-bore diameter. Time-resolved luminescence spectroscopy has been applied to record and reveal the pressure shift of the ruby window during the shock impact generated by a propellant gun. SEM, Raman, XRD, and high-resolution TEM were employed to examine the morphology and structure characteristic of both as-prepared and postshock composites, particularly to reveal the structure changes by the shock process. Full methods and any associated references are available in the Supporting Information.