High throughput tabletop shock techniques and measurements

Although shock experiments are traditionally performed in large facilities using guns that shoot disks called “flyer plates,” explosives or high-energy lasers,¹ tabletop experiments that provide convenient high-throughput shock testing have been growing in importance. The convenience and small scale of tabletop experimentation allows researchers to expand the range of possible shock measurements to new and complex systems, especially novel materials that are available only in small quantities, and to develop new diagnostic techniques that provide high time and space resolution of shocked materials. Here, we will focus on tabletop techniques using shocks produced by laser-launched flyer plates, although there are complementary tabletop methods that use direct laser drive, usually femtosecond pulses,² or laser-launched microparticles.³ Among these, the flyer plate method stands out for its ability to transfer much more momentum to the target in a planar geometry.

Our laboratory has developed a “shock compression microscope” which is a tabletop microscope equipped with a laser flyer plate launcher and a photon Doppler velocimeter (PDV).^4–6 The laser flyer plates are small, 0.5 mm in diameter and 10–100 μm thick, producing shock durations from 4 to 20 ns,⁷ but despite their small size they can achieve hypersonic velocities (0–6 km/s) with commercial single-box solid-state pulsed lasers.^7–9 Shocks in condensed matter typically propagate at several μm/ns,¹⁰ and our shock rise time in a homogeneous material such as an optical flat can be as short as 0.5 ns, although 1–2 ns is typical.¹¹ With the shock experiments in a microscope, physical processes within the 0.5 mm diameter shock can be observed with 2 μm spatial resolution.

Laser flyer plates are laser-accelerated disks, typically metal foils, whose kinetic energy is a fraction (perhaps 30%)^8,12 of the laser pulse energy. In our tabletop set up with a 2.5 J pulsed laser, the maximum flyer plate kinetic energy is about 1 J. We can achieve pressures equal to or greater than a detonating explosive.^13,14 Geometrically, the run distance is limited by the arrival of radial edge waves. Our optical probe methods (with the exception of video imaging) look at the central 0.1 mm of the 0.5 mm shocked volume, and within this volume, a typical run distance where a constant pressure is maintained is about 250 μm.¹⁵ 

Large facilities use guns with flyer plates that are millimeters or centimeter thick which produce shocks lasting many microseconds.^1,10 A long-duration shock can give a steady-state measurement that can be used for accurate determinations of equation of state. Similar experiments are possible with flyer plates, but require a large laser facility, for instance, a 400 J multiamplifier laser that could launch 250 μm thick 4 mm diameter flyers at several hundred meters per second.¹⁶ Our short-duration shocks can briefly produce steady-state conditions that can be probed with sufficiently fast diagnostics but there is less time for precision measurement. Our shocks can also be used for transient experiments analogous to widely used optical pump-probe methods. A short-duration shock can be used to trigger a process such as a chemical reaction that persists long after the shock has dissipated.¹⁷ 

Here, we describe our shock compression microscope with emphasis on methods for laser launching flyer plates that produce reproducible planar impacts, where the key to achieving excellent results is control over the spatial and temporal profile of the launch laser beam, and on techniques for preparing multitarget sample arrays needed for high-throughput studies. Several applications are described that utilize the unique features of tabletop shock microscopy, including studies of shock propagation and attenuation in nanoporous materials,^18–20 emission spectroscopy of organic dyes^21–24 and quantum dots^25–27 to label individual constituents of inhomogeneous materials, shock deformation of quasispherical quantum dots to measure shear relaxation in solids and fluids,²⁷ methods to measure transient optical absorbance needed for nonfluorescent or highly quenched absorbers, high-speed temperature measurements of shocked plastic-bonded explosives (PBX),^17,28,29 and shock chemistry at a fuel/oxidizer interface.^30,31

Our shock compression microscope with laser flyer launcher and PDV^4–6,32 is depicted in Fig. 1(a), and a schematic for a typical multielement flyer plate launcher and target array is shown in Fig. 1(b). We use an all fiberoptic 8 GHz PDV⁷ based on a design by Weng et al.³³ We use four illumination sources, a light-emitting diode, a Xenon flashlamp, a nanosecond green laser, and a femtosecond Ti:sapphire laser. For detectors we use a homebuilt optical pyrometer, which is a high dynamic range optical emission spectrograph,³⁴ a nanosecond video camera, and a streak camera. Often these diagnostics can be employed simultaneously using dichroic beamsplitters. For instance, the 1.55 μm PDV can be used to monitor flyer and sample velocities while the camera and pyrometer image and record thermal emission in the visible range.^35,36

We have developed a laser flyer plate launcher that produces flat parallel and reproducible hypervelocity (0–6 km/s) impacts with targets.^9,11 We use a high-energy Q-switched multimode Nd:YAG laser^7,8 with a Faraday isolator to protect the laser from backreflections. The flyer plate optic is simply a metal foil cemented to glass substrates [Fig. 1(b)] using a low viscosity optically clear epoxy such as Eccobond 24 or equivalent. A wide variety of foils can be launched but the research described here used 12–100 μm thick Al or Cu foils.⁹ We have found it is not necessary to fabricate many individual sacrificial flyer assemblies as was done in prior works.^16,37 With our simple flyer optic, a 50 × 50 mm² foil on glass can launch hundreds of flyers.⁸ 

Since the invention of the Q-switched laser, there have been many methods developed for laser launching of flyer plates.^16,38 Until recently, however, these methods were not in widespread use for a variety of reasons involving the difficulty of mass-producing single-use flyer launch assemblies³⁷ and concerns about the state of the flyer plate upon target impact.

While the laser is accelerating the flyer plate, it can damage the flyer in several ways, and we have developed methods to minimize such damages. Laser pulses of sufficient energy to propel hypersonic flyer plates generally come from multimode lasers or multimode laser amplifier assemblies. Pulses from multimode lasers have complex spatial and temporal profiles. For example, the profile from our laser is shown in Fig. 2(a). However, it is necessary to apply a uniform force on the flyer plate so that a large amount of laser energy can be transferred to the plate without distorting or destroying it. A nonuniform spatial profile will accelerate different parts of the flyer to different velocities, causing the flyer to tear itself apart during flight.³⁹ The short Q-switched laser pulse duration, typically 10 ns, can launch reverberating shocks in the flyer plate that can cause damage through spallation.⁷ Laser heating can melt, vaporize, erode, or ablate the flyer.

Multimode laser beams are typically characterized by the M² parameter. The focused beam diameter is M² times the diameter of a single-mode laser beam,

where λ is the wavelength and d the focused beam diameter for an input beam diameter D and a lens of focal length f.

The laser used in our laboratory has M² = 40, a pulse duration of 10 ns, and a maximum energy of 2.5 J which, due to optical losses, is reduced to 1.8 J at the flyer surface. For thicker flyer plates, 10 ns pulses can produce damaging reverberating shocks in the flyer,⁷ so we have the capability of stretching the pulse to 20 ns using a low-loss all-reflective pulse stretcher,^7,40 and these 20 ns pulses can reliably launch flyer plates up to 100 μm in thickness.⁷ 

It is important to realize that output beam profiles obtained with a camera, such as Fig. 2(a), represent a time average of the laser pulse spatial profile. The pulse spatial profile changes somewhat from shot to shot, but in addition, it changes throughout the nanosecond pulse. With a Q-switched multimode laser, as the pulse builds up the lower order modes in the center of the beam reach threshold sooner, so the modal pattern of the beam profile changes with time. A beam homogenizing technique that appears to produce a uniform spatial average on a camera that time-averages the image may not generate a uniform force on the flyer plate at all times.

We use a first-order diffractive beam shaper from Silios Technologies (Peynier, France),⁸ which we strongly recommend for this application. The principle of operation is depicted in Fig. 2(b). The laser beam is expanded to fill the shaper, a 75 mm diameter window with many diffractive elements etched onto the surface. The first-order diffraction from any one of these elements will produce a uniform disk of light in the far field when irradiated by a beam whose intensity is uniform across the element. Due to the large area of the laser beam compared to the size of the elements, no matter how the beam shape changes, on every shot each element is nearly uniformly illuminated. Beam homogenization takes place at the level of the individual elements. As the spatial profile varies during the pulse, the intensity on each element correspondingly varies in time, so the intensities of all the disks in the far field also vary with time, but each disk remains spatially homogeneous [Fig. 2(b)]. These uniform disks of light are combined at the focus of the objective lens. The diffractive elements have a range of diffraction angles to produce the desired pattern of uniform overlapping disks. Figure 2(c) shows the resulting beam profile at the focus of the homogenized laser beam. The beam consists of a uniform 0.5 mm diameter disk with soft edges, and, as per the discussion above, its profile hardly varies during the nanosecond pulse or as the laser flashlamps age or the laser alignment drifts. We typically use a 150 mm focal length aspheric objective lens, and with the resulting 0.5 mm beam, the maximum laser fluence at the flyer plate surface is 500 J/cm². The setup also works well with longer focal length objectives, producing similarly uniform but larger diameter focused beams.⁷ 

In designing a laser launch system, it is important to use a laser with an adequate value of M² and to use a correspondingly designed diffractive optic since the angular spread of the diffractive elements should be optimized for a particular value of M². The larger the value of M² the larger the focused beam size so the angular spread must be correspondingly larger. Silios data indicate that the uniformity of the homogenized beam increases with the value of M². However, large M² beams can be difficult to work with since they have high divergence, requiring short focal length objectives, short working distances, and short depths of field, so some compromise is in order. Our M² = 40 laser appears to be a good compromise since it produces good beam uniformity with a convenient working distance of 150 mm from objective to flyer plate.

This beam homogenization method launches flyers that produce spatially uniform impacts with minimal tilt. Two measurements demonstrating this, where the target was a glass flat, are shown in Figs. 2(d) and 2(e). In Fig. 2(d), the flyer impact was imaged with a femtosecond strobe approximately 2 ns after impact.⁹ The strobe, at a high angle of incidence, begins to see the flyer plate when the distance to the window is roughly a wavelength of light.⁹ The image in Fig. 2(d) shows that the flyer, immediately after impact, is a uniform 0.5 mm disk. We have estimated the impact tilt as being less than 1 mrad,⁹ which is comparable to what is achieved with gas guns.⁴¹ Although we ordinarily launch flyers in a vacuum of a few mTorr, Fig. 2(e) shows an image of a 2 km/s flyer launched in air. The hypersonic flyer compresses a column of air which is strongly heated. Just prior to impact with the window the air density and temperature are high enough to create an emissive plasma, and Fig. 2(e) shows this air plasma, which appears as a uniform disk the diameter of the flyer plate.

Compared to typical methods for launching flyers, which use the laser to vaporize part of the flyer plate to produce an expanding plasma drive,⁴² we reduce the damaging effects of the laser by depositing up to 75% of the laser pulse energy at the interface between the substrate, the optical cement, and the flyer plate,⁷ as depicted in Fig. 3(a). The absorbed energy creates a shock in the material just underneath the flyer plate resulting in a “shock-assisted launch”^7,43 greatly reducing the laser energy incident on the flyer plate. It is well known that focused Q-switch laser pulses are absorbed and produce damage primarily at window surfaces⁴⁴ but we need to avoid absorption at the input face which otherwise would attenuate the pulse before it reaches the flyer. Because the beam on the diffractive optic has a large diameter of 75 mm compared to a 0.5 mm focused spot size, the convergence angle is large [Fig. 3(a)], so the energy density can be below the absorption threshold at the input face when we use 6.35 mm thick glass windows.⁷ With this arrangement, we can generate laser fluences up to 500 J/cm² at the glass/cement/flyer interface.⁷ 

Our group modeled the flyer launch and the impact of an Al flyer plate with various target materials using a two-dimensional code called ALE3D, which uses arbitrary Lagrangian–Eulerian (ALE) computational techniques.⁴⁵ To avoid simulating the complex interfacial nonlinear optical absorption process, we simulated inputting heat to the cement layer that bonds the foil to a glass substrate using the measured temporal and spatial profiles of our laser pulses. The flyer plate was launched by the thermal expansion of this adhesive layer.⁴⁵ Our calculations were in good agreement with our measured flyer plate acceleration profiles and with the shock waves observed in glass and other target materials.¹¹ 

Figure 3(b) shows the range of velocities we achieved with Al and Cu metal foils with different foil thicknesses. The shock duration in the target is roughly proportional to the foil thickness with 25 μm foils producing nominal 4 ns shocks, 100 μm foils 20 ns, etc.⁷ Figure 3(c) shows the results of a flyer reproducibility test. We shot 33 flyers in rapid succession, and the mean velocity was 3.29 km/s with a 0.02 km/s variance.⁹ 

New methods for ameliorating the dangerous effects of shock waves are important to protect personnel and equipment from impacts and blasts. One common way to accomplish this is to use a shield having high dynamic shear strength, for instance, a Kevlar vest or a polyurea⁴⁶ coating. The high dynamic shear strength of these materials prevents projectiles from penetrating the shield, and such materials are often evaluated by determining how thick a material is needed to stop a given projectile.

A potential method to improve shock wave protection is to develop multifunctional materials having multiple shock dissipation mechanisms. Those, for instance, might involve shock wave absorption, defined as the attenuation of input shock pressure by converting the directed energy of the shock into heat or stretching the shock pulse to reduce its peak pressure.

We have developed a method to measure and characterize shock waves transmitted through small quantities of novel materials to measure shock wave absorption and dispersion.^19,47,48 Our absorption characterization is analogous to the well-known Lambert’s law used to describe optical absorption, which states that absorbance is proportional to thickness. In applying Lambert's law to electromagnetic absorption, the incident I₀ and transmitted I electromagnetic intensities are measured to determine the absorbance A = −log₁₀(I/I₀). The analogous relation proposed for shock wave absorption would be A = −log₁₀(F/F₀), where F denotes the shock wave incident and transmitted energy fluences (units J/cm²).^19,47,48

Shock wave absorption measurements use the arrangement in Fig. 1(b). A flyer plate drives a shock into targets with various thickness layers of the test material. The shock emerges into a window, typically Pyrex glass, and the physical motion of the sample/glass interface imparted by the shock is measured using PDV. The window is coated with a thin (<100 nm) nonperturbative metal film mirror, allowing PDV to measure the time-dependent velocity U_p(t) at the interface of the window and the test material. If the window has a known shock response, the interfacial velocity profile can be used to determine the fluence F out of the sample into the window, knowing the window density ρ₀ and the window Hugoniot, which have been tabulated. Often the Hugoniot has the linear form U_s = mU_p + c, where U_s is the shock velocity and U_p the material (i.e., interface) velocity. For Pyrex glass, ρ₀ = 2.23 g/cm³, c = 5.03 km/s, and m = −0.18 in the velocity (U_p) range of 0.77–1.5 km/s. The shock energy fluence F transmitted through the test material into the window is given by⁴⁷ 

An example measurement of the velocity profile U_p(t) is shown in Fig. 4(a), where the test samples consisted of 4–110 μm thick layers of microcrystal powder of the metal-oxide framework (MOF) denoted ZIF-8, having a few percent of polymer binder. The impact velocity was 0.6 km/s.⁴⁸ ZIF-8 is a good candidate for a multifunctional shock absorbing material because it exhibits powder compaction, it has a nanoporous structure that is stable when the solvent used to synthesize the MOF is removed, and nanopore collapse endothermically breaks multiple chemical bonds.⁴⁸ Figure 4(a) shows that the transmitted shock loses amplitude and is stretched in time. It develops a structure characteristic of pore compaction.⁴⁹ 

Although this method gives the transmitted shock fluence F, it does not give the incident fluence F₀. The incident fluence from the flyer plate is simply the flyer kinetic energy (from flyer mass and velocity) divided by the flyer area but due to impedance mismatch between flyer and test sample we do not necessarily know the fraction of the flyer's kinetic energy that is transmitted to the sample. Our solution was to approximate the incident fluence F₀ as the fluence transmitted through such a thin layer of sample that attenuation was minimal.⁴⁸ Knowing the incident and transmitted fluences, we can compute the shock wave absorbance A and the shock energy absorption coefficient α for a sample of thickness l,

Figures 4(b)–4(e) show measurements of the shock absorbance A and the absorption coefficient α (the slope of A vs sample thickness l).⁴⁸ These measurements used four kinds of powders, all with a small quantity of polymer binder: Al metal, Al₂O₃, ZIF-8, and graphene. The flyer velocities were varied from 0.6 to 1.9 km/s and the sample thicknesses from a few μm to >100 μm.⁴⁸ 

Figures 4(b)–4(e) show that absorbance vs thickness plots are always linear, so at least in these four cases, just as with optical absorption, shock energy absorption obeys Lambert's law. For Al and Al₂O₃, all the absorbance points were independent of impact velocity and the lines in Figs. 4(b) and 4(c) were fit using all those points. In the cases of ZIF-8 and graphene, the absorption coefficient α decreased with increasing impact velocity, and the lines in Figs. 4(d) and 4(e) were the fits at each velocity. The decrease in absorption coefficient for ZIF-8 and graphene indicates that the target material has a limited capacity to absorb shock energy so that shock absorption can become saturated. For instance, with powder compaction, after the voids collapse the ability to absorb shock energy is decreased. By analogy with standard optical saturation theory, we propose⁴⁸ 

where α₀ is the unsaturated absorption coefficient and F_sat is the saturation fluence, a measure of the material's shock energy absorption capacity. The linear fits to the data in Figs. 4(b)–4(e) are fits to Eq. (4) where the unsaturated absorption coefficient was varied for Al and Al₂O₃ and the absorption coefficient and saturation fluence were varied for ZIF-8 and graphene. In all cases, the fits to Eq. (4) are excellent. This method provides an absorption coefficient along with a measure of a sample's capacity to absorb shock energy.⁴⁸ For example, the saturation fluence for ZIF-8 was 70 kJ/m².⁴⁸ 

For shock protection, a useful figure of merit is shock absorbance per unit mass since shock protection materials are ideally both highly protective and lightweight. ZIF-8, which is highly nanoporous and lightweight, was an outstanding performer by this metric. For ZIF-8, using the saturation fluence needed to fit the data in Fig. 4(d) and the ZIF-8 density, we obtained a specific shock absorption coefficient of 55 m²/kg.⁴⁸ To put this into perspective, we compared ZIF-8 with published results on various thickness layers of Plexiglas, which is used to make “bulletproof” glass. ZIF-8 turns out to absorb seven times more shock energy than an equivalent mass of Plexiglas. However, ZIF-8 does not have high dynamic shear strength, so a possible method to improve shock protection might be to incorporate related multifunctional materials into matrices with high shear strength.

Although Lambert's law with saturation appears to work well for these diverse materials, there is no certainty that this conclusion can be generalized to all materials. However, when applicable, this method can be used to predict shock wave energy absorption given the impact velocity and sample thickness.

Shocked heterogeneous solids produce a complex pressure and density field. Photoemissive probes such as organic dyes or quantum dots can be used to study the individual constituents of a heterogeneous mixture under shock compression since it is possible to bind different probes to different constituents, a technique widely used in biological research.⁵⁰ In order to do this, however, it is helpful to understand some of the mechanisms that affect the photoemission intensity,²¹ lifetime,⁵¹ and wavelength shift^25,52 under shock. Generally speaking, these properties depend on the interactions between the probe and nearby atoms, in other words shock effects on photoemitters are local properties.

In photoemission experiments, we excite the probe species with a laser pulse that is long in duration (250 ns) compared to the shock experiment while we monitor the time-resolved emission with video imaging,²³ which is sensitive to changes in emission intensity, or spectroscopy which can measure density-dependent spectral shifts and widths.^22,25,26 In the experiments described below, the excitation source was a 527 nm Nd:YLF laser with a 250 ns pulse width, and the excitation light was blocked from the detector using a narrow-band notch filter.

One example of selective probing of an inhomogeneous material is shown in Fig. 5, where the sample consisted of 40 μm spherical silica particles.²³ We used sol-gel techniques to grow a thin coating of Rhodamine 6G dye-doped silica²² on these particles [Fig. 5(a)], so the surface of the particle is photoluminescent while the bulk is not. Figure 5(b) shows a fluorescence image of the sample, which consists of these particles along with 5% poly-vinyl alcohol binder. Figure 5(c) shows another fluorescence image taken immediately after a 15 ns duration shock produced by a 1 km/s flyer plate.

Figure 5 shows that shock decreases the overall fluorescence intensity. Shocking a Rhodamine dye decreases the singlet–triplet gap, increasing the intersystem crossing rate and quenching the emission.^21,51,53 The shock also cracks the particles, revealing the interior which lacks a photoemitter and appears even darker than the particle surfaces. With this localized photoemissive probe, it is possible to distinguish between the surface and interior of the shocked particles and visualize shock damage that damages the particles.

Quantum dots are another attractive photoemission probe. Figure 6 describes experiments using photoemissive CdSe quantum dots (QD). With hydrostatic pressure, the QD emission blueshifts since the electronic bandgap, often described in a particle-in-a-box picture, increases.^25,26 In Fig. 6, these QDs with appropriate surfactants were dispersed into three types of media, a fluid (glycerol), an organic polymer (poly-vinyl alcohol), and a hard glass (sol-gel silica).²⁷ These three samples were shocked to 5 GPa with an 8 ns duration shock. In all these media, the shock causes a photoemission intensity loss, which we attribute to enhanced transfer of electrons from the QD to the surrounding media.

The shock also causes the emission spectrum to temporarily shift, although the intensity loss is persistent. In the fluid, Fig. 6(a), we see the expected transient blueshift that lasts for about the duration of the shock.

However, in the two solids, the polymer and the glass, we instead see a transient redshift, as indicated by the red lines in Figs. 6(b) and 6(c). The explanation for these effects involves the shear stress in solids that is not present in fluids. The planar input shock temporarily deforms the nominally spherical QS into elliptical QD, and this deformation lasts longer in hard matrices than in fluids. The deformation splits the emission transition into two components, one polarized along the long axis and the other along the short axis. The long-axis transition is redshifted compared to the short-axis transition, so during the brief period that the QDs are elliptical, the redshifted transition dominates the emission spectrum. Thus, these QDs can be used to monitor local shear relaxation in shocked media.

The photoemissive probe technique only works for probes that are strongly emissive, and as the results in Figs. 5 and 6 show they tend to lose emission intensity when shocked. Many shock-induced chemical and physical processes of interest can quench emission. Emission measurements are usually not quantitative due to lack of knowledge of the extent of quenching process. For this reason, we have recently implemented optical absorption spectroscopy in our shock compression microscope. With optical absorption, we can observe the photoemissive probe as well as any new absorbing species that might be formed in the shock, even if the new species are not photoemissive. We can determine concentrations quantitatively using the Beer–Lambert law, which states that the absorbance is proportional to absorber concentration.

Absorption spectroscopy has rarely been used in shock experiments, and we are aware of only a handful of examples.^54–56 For the probe source, we use a small broadband 2 μs duration Xenon flashlamp (Ocean Insight PX-2). Since the flashlamp emission is weighted toward the blue, we use a color balancing filter to flatten the visible spectrum. The flashlamp reflects off a mirror on top of the sample to measure a double-pass absorption. The signal is the spectrum during shock and the reference is a separate shot measuring the spectrum without shock. The reflector may be the flyer plate, the flyer plate with a shiny Ag coating, or a metal film deposited on top of the sample. In the work described here, we studied liquid solutions. The solutions are in microcuvette arrays described elsewhere,¹³ and the cuvettes are sealed with a thin 9 μm Al foil that also acts as a reflector.

Figure 7 shows photoemission and absorption from a 40 μm thick ethanol solution of 30 mM Rhodamine 6G dye with a 3.3 km/s impact lasting 8 ns. The flashlamp spectrum fluctuates a bit from shot to shot but we have found we can get excellent results by averaging ten or more shots, as was done in Fig. 7. Using PDV measurements along with the tabulated ethanol Hugoniot, we find the input pressure is 14 GPa and the time for the shock to traverse the sample is 7 ns.

The 14 GPa photoemission results in Figs. 7(a) and 7(b) show that the input shock (at t = 0) shuts down about 90% of the emission intensity. Figure 7(b) shows wavelength-resolved transients over four narrow wavelength bands including the emission peak (536–538 nm) and to the red and blue of that peak. The intensity loss to the blue edge (black curve) is larger than the intensity loss to the red, denoting a shock redshift [which can be seen at shorter times in Fig. 7(a)]. The redshift occurs because the higher-density shocked ethanol stabilizes the excited state more than the ground state.

Figures 7(c) and 7(d) show absorption spectra. In contrast to emission, a strong absorption is observed at all times. Figure 7(d) shows the absorption loses some intensity when the shock arrives. The intensity loss is greater on the blue edge than on the red edge, indicating the same redshift that was seen in the emission.

An entirely new absorption transition is observed in the 552–554 nm region [Fig. 7(d)] which is to the red [arrow in Fig. 7(c)] of the parent R6G transition at 520 nm. This new transition peaks at about 60 ns. A corresponding transition is not observed in the photoemission spectra, presumably because the species giving rise to this transition is not strongly photoluminescent.

The most likely origin of the new transition is a new molecular species due to proton transfer from ethanol.⁵⁷ Ethanol, a weak acid, becomes a stronger acid under shock. Other chemical reactions such as oxidation seem less likely since oxidation would damage the chromophore so no absorption transient would be observed.

High-performance energetic materials (EM) are generally in the form of plastic-bonded explosives (PBX) which consist of energetic powders or powder mixtures, a polymer binder, and the inevitable voids.⁵⁸ It is desirable to minimize the void fraction to reduce shock sensitivity caused by ignition of EM from the heat generated in collapsing voids.⁵⁹ The aim of our shock experiments is to determine how a shock ignites hot spots in a PBX microstructure, how they grow into a deflagration, how the deflagration evolves into a detonation if geometry permits, and how this all depends on PBX composition and microstructure.

We have developed methods to mass-produce arrays of hundreds of small cylindrical PBX charges [Figs. 8(a) and 8(b)] with various compositions and microstructures, all with minimal void volumes [Figs. 8(c) and 8(d)].¹⁷ Warning: energetic materials can be hazardous and should only be handled by personnel with adequate safety training. Each PBX charge is about 0.1 mg. To fill an array of 200 charges takes about 20 mg of EM.

Array templates are fabricated by bonding a Kapton adhesive tape, thicknesses from 25–500 μm, to a glass substrate then using a focused laser to cut a matrix of 1.0 mm holes. The wells formed by these holes and the glass substrate can be packed by an extrudable PBX with the consistency of, say, toothpaste or putty which can then be heat cured into a solid structure. There are at least two extrudable explosives described in the literature, XTX-8003 and XTX-8004, which both use 20% poly-dimethyl siloxane (PDMS) binder and 80% of either PETN (pentaerithritol tetranitrate) or RDX (1,3,5-trinitroperhydro-1,3,5-triazine).^60–63 Since the PDMS binder is compatible with a large variety of energetic and inert ingredients, with this method, it is possible to produce PBX with widely varying compositions and microstructures.

To prepare these PBX, we mix fine powders with Sylgard 182, a type of PDMS, and mix them into a putty. Now, and this is very important, the putty is placed in a hydraulic press to minimize voids.⁶⁴ The pressed putty is then packed into the array with a Teflon spatula followed by a sharp blade to clear excess material from the surface. Figure 8(b) shows an array using a PBX based on TATB explosive (1,3,5 triamino 2,4,6 trinitrobenzene), which is yellow. Figures 8(c) and 8(d) show slices of x-ray computed tomography (CT) scans of TATB and PETN PBX where the white regions are the voids. Typically, these charges have <5% void volume.

Figures 8(a) and 8(e)–8(g) depict some different array configurations, where each depiction shows a single element of a multitarget array. In Fig. 8(a), a flyer plate is used to initiate the PBX, which can be observed through the glass substrate (witness plate). A thin Au mirror (<100 nm thick) can be coated onto a portion of the glass for velocity measurements of the shock emerging from the PBX at the PBX/glass interface.⁶⁵ As discussed in Sec. III, knowing the Hugoniot of the glass window, it is possible to determine the energy and temporal profile of this emergent shock.

Because flyer plates have kinetic energies limited by the laser, for insensitive explosives, such as TATB, we can use the flyer to initiate a booster charge that launches a more powerful shock into the test explosive, as depicted in Fig. 8(f). To fabricate such a double-layer array, we start with an array like Fig. 8(b) and then cement a stainless steel plate with a matching hole pattern. The holes in the stainless steel are then packed with the booster charge.^29,66

Figure 8(g) is an arrangement used to study explosive casing fragmentation. The flyer plate initiates the PBX which drives a shock wave into a test casing material which can be observed through the glass window. The goal of such experiments is to develop reactive materials that are ductile enough to fabricate into casings, but which fragment in predetermined patterns so the fragments can release energy on a target.⁶⁷ 

Figure 9 shows a result using the two-layer PBX method. The sample is a 40 μm thick cylinder of TATB in PBX form. TATB combines high explosive power with high shock insensitivity. The booster charge uses 90 μm thick CL-20 (hexanitrohexaazaisowurtzitane), which is one of the most powerful EM known. This measurement uses PDV to monitor the velocity of the shock exiting the TATB. When a 3 km/s flyer plate is used to initiate the TATB directly without the booster charge, the peak interface velocity is 1.6 km/s and the shock produced by the TATB lasts for about 10 ns. With the CL-20 booster, the peak velocity increases to 1.8 km/s while, significantly, the output shock duration increases by a factor of three. In this way, TATB can be made to produce a much larger explosion than with flyer plate shock alone.

While velocimetry can provide an equation of state for an energetic material, temperature, which is essential, is lacking. Temperature probes such as thermocouples inserted into the energetic material are perturbative and have slow response. But the temperature of a hot body (i.e., Planck's blackbody) can be determined remotely by measuring the spectral radiance, which is the emission spectrum vs a calibrated intensity standard.⁶⁸ Energetic materials produce copious thermal emission when they react but they are not necessarily the classic blackbody, so remote determination of EM temperatures uses the graybody approximation, which is Planck's blackbody law with an emissivity ɛ < 1. A significant problem in graybody temperature determination is the presence of interfering light sources, typically atomic line and molecular band emission,^34,69 or absorption of the thermal emission by the explosive itself or its decomposition products.⁷⁰ We have developed an optical spectrograph to measure time-dependent thermal emission from shocked EM [Fig. 1(a)], in which case it functions as an optical pyrometer.⁶⁸ Our optical pyrometer consists of a high-throughput low aberration prism spectrograph with 32 fiberoptics in the image plane that transport the light to 32 photomultiplier tubes and 32 digitizers that measure spectral radiance every 0.8 ns out to 150 μs and beyond.³⁴ The spectrograph is calibrated using a ∼2900 K tungsten halogen source that is spatially homogenized with an integrating sphere. The sphere manufacturer, Labsphere, Inc., then calibrates this spatially uniform spectral radiance against a NIST-traceable standard.

Our pyrometer obtains about 200 000 spectra on a single shot. We use a logarithmic averaging routine in each time decade to decrease the total number of spectra to 70 per shot. So, the output of the pyrometer on a single shot is a 32 × 70 matrix of spectral radiances.³⁴ 

Using 32 detectors in the visible is useful to identify interfering emission sources or regions perturbed by the presence of absorbing species in order to verify that only thermal emission is being detected.³⁴ When interfering sources are identified, temperature measurements can simply select wavelength channels that avoid those interferences.

Wien's high-temperature approximation for graybody emission is^68,71

where S(T,λ) is the spectral radiance of an object at temperature T, C₁ = 3.72 × 10⁻¹⁶ W/m², and C₂ = 1.439 × 10⁻² m K. The EM we use typically produce temperatures in the 2000–6000 K range where Eq. (5), which is the classical limit of Planck's blackbody, is valid.

Equation (5) can be reformulated to yield

Equation (6) shows that if we plot the function Z vs 1/λ, a linear function will be obtained with a slope inversely proportional to T.

Figure 10 shows the 32 × 70 output matrix of spectral radiances from our pyrometer plotted as Z vs 1/λ. The sample was an HMX (cyclotetramethylene-tetranitramine) PBX, and the flyer plate velocity was 4 km/s. Figure 10 shows that the spectral radiances plotted in this form give linear plots at all times, hence the emission from HMX is purely thermal (at least in this wavelength range). At shorter times, the slope is lower, indicating higher temperatures that decrease as time passes.

The corresponding time-dependent temperature profile is shown in Fig. 11. It indicates three regimes as shown in Fig. 11. At shorter times, the thermal emission originates from the so-called hot spots,^59,72 regions where the PBX microstructure concentrate the shock energy to produce localized ignition. The initial hot spot temperatures are above 6000 K.⁶⁴ With such a high initial temperature, the hot spot combustion process is a mixture of deflagration, combustion that propagates subsonically, and a thermal explosion that propagates supersonically.⁷³ For tiny hot spots embedded in a colder medium, thermal conduction and adiabatic expansion result in rapid cooling to below 5000 K within 30 ns (Fig. 11). Subsequently, a second cooling phase occurs where temperature drops to about 3000 K in 200 ns. High-speed video has shown that this second cooling process results from a bulk thermal explosion and adiabatic expansion of product gases produced by HMX decomposition.²⁸ Finally, the temperature stabilizes at ∼2800 K which represents the combustion of the unshocked HMX PBX in the 1 mm diameter charge surrounding the 0.5 mm shocked volume.

Shock hot spot generation and growth in a microstructured PBX is a complex process, and due to the opacity and extreme conditions of shocked PBX, hot spots have rarely been observed in real time. To observe hot spots directly, we have developed a model system that consists of EM embedded in a transparent polymer binder.^35,36,74 We have focused on HMX single and polycrystals with a polyurethane (PU) binder.^35,36,74 As illustrated in Fig. 12(a), the sample geometry has the flyer plate impacting a flat PU surface, launching an initially planar shock into the HMX crystal. We have shown that for HMX single crystals, hot spots form at the HMX–PU interface³⁵ where shock energy piles up due to the impedance mismatch.⁷⁵ For defective crystals, hot spots can also be produced at internal voids, cracks, and polycrystal junctions.³⁵ 

We observe the hot spots with high-speed video with 20 ns integration time and 2 μm spatial resolution and with simultaneous optical pyrometry which determines the spatial-averaged temperature.³⁵ The temperature profile in Fig. 12(b) is similar to the profile for HMX in PBX form in Fig. 11 with two notable differences. In the PBX, the hot spots are 6000 K and the longer-time temperature is 2800 K, while in the single crystal, the hot spots are 4000 K and the longer-time temperature is 3200 K. The hot spots are hotter in PBX due to shock collapse of voids not present in the polymer-bonded single crystal. The longer-time combustion temperature is lower in the PBX than in the single HMX crystal since the PBX is 20% binder.

The model system of a single crystal embedded in a polymer, or other engineered microstructures embedded in a polymer, allow us to produce hot spots and observe their time evolution in space at known temperatures. It presents a unique opportunity to understand kinetics under extreme conditions of high pressures, temperatures, and fast and copious energy release. In other words, we have developed the ability to produce and observe highly energetic chemical reactions that are initially localized in both time and space.

Figure 12(d) shows the time-dependent radiance (wavelength-integrated spectral radiance) profile of the hot spot in Fig. 12(c). We can see the hot spot grow anisotropically, presumably due to the HMX crystal structure, which is known to have anisotropic thermal conductivity.⁷⁶ We can assign an overall rate for the growth of the hot spot, and it was about 550 m/s³⁶ which is subsonic but fast for a combustion reaction. For comparison, the reactive molecular simulations of Joshi and Chaudhuri⁷⁷ where a thermal pulse created ∼10 nm hot spots in 20.5 GPa RDX, gave a deflagration propagation velocity of 205 m/s.

Although EM such as HMX or CL-20 release large quantities of energy (roughly 40 kJ/cm³), an even more energetic reaction is the reaction of metal fuels such as Al with oxidizers such as O or F, which can release —two to three times more energy.^78,79 Metal-based explosives are fundamentally different from organic molecules such as HMX since the fuel and oxidizer are physically separated so that metal oxidization requires transport of the oxidizer to the metal fuel.⁸⁰ Diffusive transport limits reaction rates of metal fuel oxidizer systems. For instance, even when well-mixed nanoparticles were excited by a short-duration high voltage discharge, reactions occurred on microseconds or longer.⁸¹ 

Although many useful explosive formulations include metal powders such as Al,⁵⁸ those additives do not contribute much to the detonation itself, where reactions typically occur in tens of nanoseconds. Instead metal powders are used to generate additional heat subsequent to detonation. The desire to increase the energies of chemical explosive detonations has motivated us to study shocks in systems with model fuel/oxidizer interfaces with the intent of finding methods for fast nondiffusive mixing of metals and oxidizers.

A model system for the fuel/oxidizer interface is the reactive nanolaminate (RNL),⁸² which consists of a stack of alternating thin layers of fuel and oxidizer. The model system we investigated, produced in the Maria laboratory at Pennsylvania State University,⁸² is depicted in Fig. 13(a), where the oxidizer was CuO, the fuel was Zr, and there was a single reactive interface. The RNL was 0.5 mm thick and the chemistry was stoichiometric, so CuO was 160 nm and Zr was 340 nm. The shock-induced reactivity was probed by the same pyrometry-imaging method used for single HMX crystals.^31,83

The striking feature of Fig. 13(b) is that initially, intense light emission, which our graybody measurements show is approximately 3000 K, arises from the edges of the flyer plate disk and not from the center of the disk. The radiance measurement in Fig. 13(c) shows that light emission from the shocked RNL occurs in two bursts. The first, from the flyer plate edges, occurs within about 50 ns, whereas the much longer-lived light from the center of the flyer plate occurs in about 2 μs.

The material deformation in the RNL is quite different at the center and at the edges of the flyer plate. At the center of the flyer plate, the planar slabs of the RNL are subjected to an almost purely compressive load, whereas at the edges of the flyer plate the load has a high component of shear.

The data in Fig. 13(c) indicate that it is possible to produce high rates of energy release from metal fuel/oxidizer systems comparable to organic explosives. This rapid energy release arises from fuel/oxidizer interfaces subjected to intense shear waves and can be attributed to shear mixing processes.⁸³ 

With the goal of developing more powerful EM utilizing metal particles, future focus should be on developing systems with geometries that promote rapid shear mixing to overcome the diffusion-limited barrier to fast chemical reactions.

High-throughput tabletop shock methods are growing in significance due to the lower barrier to entry in the field, at least compared to guns and massive laser facilities, and because of the new opportunities available in this format, for instance, the ability to evaluate small quantities of new materials or to safely and conveniently produce extreme dynamic conditions of high temperatures and pressures. The high time and space resolution of tabletop methods reveal fundamental mechanisms and can generate large data sets needed for machine learning.

In this paper, we described in some detail how to build a reliable high-throughput tabletop laser hypersonic flyer launcher using a single-box pulsed solid-state laser, and we characterized flyer reproducibility and flatness and parallelism of impact. The key to achieving excellent results is to design a laser and beam shaping system that generates a uniform force on the flyer with minimal damage. There are myriad ways of generating and shaping laser launch pulses but we cannot guarantee that any of those methods other than the one we describe can give such good results.

Besides the laser launcher, there are two other components needed for high-throughput tabletop shock measurements, namely, the appropriate optical diagnostics, especially velocimetry, and a convenient method for fabricating single-use multitarget arrays. Our use of an inverted microscope (which could be a commercial unit or one built from individual components), which is common in many biological and materials science experiments, means that the many optical probe methods used in those fields can be adapted to shock studies.

To illustrate some of the new possibilities for tabletop high-throughput experimentation, we focused on a few examples. We described methods to measure shock propagation and shock attenuation and showed that shock attenuation, like electromagnetic wave attenuation, can be described by Lambert's law, that absorbance is proportional to sample thickness. We also showed that rather complex materials such as the metal-organic framework (MOF) ZIF-8 can exhibit multifunctional shock absorbance.

Inhomogeneous materials such as composites, mixtures, laminates, and so forth have both a bulk shock response and a local shock response. We have shown it is possible to probe local responses by binding photoemissive dyes or quantum dots to different constituents or different microstructures. We have also studied the fundamental mechanisms that determine how such probes respond to local changes in density and temperature. Along the way we have developed an instrument that can measure absorption spectra in shocked materials, which can be used for nonemissive probe species or to study probes with low quantum yields.

Energetic materials, especially in the form of plastic-bonded explosives (PBX), have been a mainstay of shocked material studies. These are microstructured materials that, remarkably, release huge energies when subjected to impact. However, these processes are difficult to study due to the harsh environment and the hazard to researchers so that they are poorly understood at a fundamental level. Here, we have shown how to fabricate arrays of tiny cylindrical PBX charges in a wide variety of compositions and microstructures. Here, we have demonstrated an optical method that measures the emission from shocked PBX allowing the temperature (>1500 K) to be determined with high time and space resolution with confidence that the emission is thermal in origin. We have also been able to directly observe hot spots generated when shock energies are localized in microstructures by embedding the microstructure in a transparent matrix.

Interfacial processes, always difficult due to the small fraction of atoms or molecules at an interface, are a robust field of study, but have hardly ever been studied with shocks, although nonlinear coherent interface-selective spectroscopies show promise.⁸⁴ A unique and greatly complicating phenomenon at shocked interfaces involves the massive material deformation that does not happen in typical interfacial studies. One interfacial system of interest is the fuel/oxidizer interface where high-energy reactions occur as a result of mixing the fuel and oxidizer at an atomic or molecular level. Here, we have shown that rapid compression of a planar interface produces only a slow, presumably diffusion-limited reaction, but that a unique feature of shock compression, namely, rapid shear mixing, can produce energetic reactions at the same rate as chemical explosives.

Although the possible applications of tabletop shock methods are limited only by imagination, here we will close by mentioning a few research areas that appear worthwhile.

Water under extreme conditions has been studied by both static high pressure and shock, which reveal a rich phase diagram,^85–87 but there is little information about water's reactivity with solutes. In 1969, Holzapfel⁸⁸ used shock conductivity measurements to show that water's ionic strength increased dramatically, resulting in a massive pH jump.⁸⁹ This happens because ionic forms of water have smaller volumes than the ambient hydrogen-bonded structure. Many shock measurements reach high pressure by letting the shock ring up, which causes water to freeze.⁹⁰ But with a single-stage shock, water remains liquid and can be driven into a liquid superionic state.^91,92 This hot superionic water must be highly reactive and it must have a rich, still unknown chemistry.

There are several biomedical applications that involve shock waves. One is shock wave cell killing where a shock breaks up the cell membrane.⁹³ Targeted cell killing by laser-generated shocks is possible by binding optical absorbers to cells.⁹⁴ Laser shocks are also used to break up kidney stones⁹⁵ and reshape eyeball corneas.⁹⁶ However, the fundamental mechanisms of these processes remain largely unexamined.

As shown in Fig. 2(e), laser-launched flyer plates can be used to conveniently generate dense plasmas that can be probed in situ.

Currently, hypersonic vehicles and missiles are of keen interest for national defense. In the atmosphere, a hypersonic missile interacts with dust and water droplets,⁹⁷ eroding structural and window materials. Figure 3(b) shows that laser-launched flyer plates can function as tiny hypersonic test beds since hypersonic missiles typically operate at velocities below 3 km/s. Orbital spacecraft interactions with dust or debris, orbital reentry vehicles interacting with the atmosphere, or cometary collisions might also be studied, but such collisions typically happen at velocities up to 12 km/s, so we would have to more than double the velocities in Fig. 3(c). But that simply means four times the laser launch energy, adding an amplifier stage to our current 2.5 J laser.

The research described in this study is based on work at the University of Illinois, supported by the U.S. Air Force Office of Scientific Research under Award Nos. FA9550-19-1-0027 and FA9550-19-1-0318 and the U.S. Army Research Office under Award Nos. W911NF-19-2-0037 and W911NF-16-1-0406. Fabing Li acknowledges support from the China Scholarship Council (CSC).

Authors declare no conflicts of interest.

The data that support the findings of this study are available from the corresponding author upon reasonable request.