Non-Richtmyer–Meshkov instability ejecta production based on shallow bubble collapse

Ejecta have been the subject of broad interest for more than 60 years. Ejecta phenomena are observed across many multi-disciplinary applications, including asteroid impacts on planets, surface shielding on spacecrafts and satellites, engineering applications for powder spraying, and laser-induced material ablation. Experimental work on shock-induced ejecta was carried out as early as the 1960s in the United States¹ and the United Kingdom.² Theoretical work on impulse-driven instabilities by Richtmyer³ and Meshkov,⁴ known as Richtmyer–Meshkov instability (RMI), has been used to underpin the physical model of recent studies,^5–8 and work has been done to examine the source terms for various shock-release surface finishes.⁹ Experiments featuring small grooves machined into the free surface have indicated that RMI causes a jet of material to be emitted from the bottom of a groove.¹⁰ Between grooves, a void (or “bubble”) propagates back into the material, where it feeds material into the jet. Depending on the shock conditions and the geometry of the groove, the jet may break up into individual particles (ejecta) if the material strength is low enough. For stronger materials, the flow may be arrested with little breakup.¹¹ 

Typically, ejecta diagnostics can include a broad spectrum of measurements of momenta, particle velocities and mass–velocity distributions, sizes, and chemistry. Recently efforts have begun to study ejecta temperatures,^12,13 although pyrometry is challenging because of the difficulty in measuring the emissivity. Recent ejecta studies have focused on ejecta production (source terms), transport in gases, and chemical reactions with a surrounding gas.^12,14

Many studies have been conducted focusing on single-shock ejection, and a smaller number have investigated multiple-shock ejection. These efforts have found that the physical picture of material ejection is relatively well characterized by RMI theory when there is a single-shock of sufficient amplitude to leave the material in a molten state: a persistent-melt-after-release (PMAR)¹⁵ condition. However, when two shocks occur and the first leaves the material in a PMAR state, the physical picture becomes significantly more complicated. Depending on the details of the experiment, the first shock can leave the material in a cavitated (high-porosity) state due to colliding release waves. For example, the Taylor wave release from an explosive drive collides with the release waves formed from the shock reflecting at the free surface, and this generates tension that cavitates the now liquified target. In the case of a gun experiment, the release from the back of a thin impactor can collide with the release waves formed from the shock reflecting from the free surface and generate tension, leading to cavitation. In the present study, we find that after the second shock arrival, ejection can range from just that predicted by the RMI model to a significant fraction of the entire cavitated layer mass. If there is no release between shocks, or if a tensile state is only reached deep in the material, or if the first shock and subsequent release do not yield a significant porosity fraction, then RMI theory may describe the behavior well. If there is a release resulting in significant cavitation sufficiently near the material surface, a weak second shock will produce very little ejecta. However, if the second shock is strong, there can be substantial entropy conversion from the second shock that results in a large, high-temperature material ejection event. We find that this mechanism, which we call shallow bubble collapse (SBC), is phenomenologically distinct from RMI-produced ejecta. We determine that if bubbles form at or near the surface due to a tensile state after an initial shock, and if the bubbles have sufficient time to nucleate and grow, then the second shock of sufficient strength can cause substantial mass ejection accompanied by significantly elevated temperatures.

To study SBC physics, we developed a three-layer flyer impactor experimental geometry that enables us to control the key parameters of SBC theory. The first impactor layer has moderate shock impedance to put the target in a PMAR condition. The second layer has low shock impedance so that a release follows the first shock. The third layer has high shock impedance to generate a second shock. The thicknesses of the target and first layer dictate the depth at which the cavitation occurs. The thickness of a second layer of low impedance material, polymethylmethacrylate (PMMA) in this work, controls the time between the first and second shocks. The third layer material then sets the second shock strength. Figure 1 shows an example flyer stack-up for a Sn–PMMA–Ta flyer impacting a Sn target. We use this configuration both experimentally and computationally to study SBC physics.

We use the ARES hydrodynamic code¹⁶ to design and analyze SBC experiments. In our 2D calculations, we use micrometer-scale resolution on centimeter-scale computational domains with periodic boundary conditions to eliminate edge effects. The simulations preserve 1D symmetry, and the cavitation (spallation) manifests as a stack of many thin and perfectly flat material sheets separated by empty space. This is unphysical. To break symmetry and avoid simple liquid spall planes developing, we must provide inhomogeneity in the release profile. To do this, we seed failure regions within the target with the same equation of state and strength, but with a lower spall threshold. This modification yields similar results for different damage models, such as Johnson spall¹⁷ or minimum pressure, so long as they capture the inhomogeneity of the voids in the liquid. To obtain robust results, we need enough seeds to avoid failure in the intervening bulk region between the seeds. This approach is designed to emulate the inhomogeneity that would exist in a liquid with local density variations.

Figure 2 shows a simulation of a Sn–PMMA–Ta flyer impact onto a Sn target where SBC is active. In the first frame (t = 0.60 μs), the first shock, with stress ∼38 GPa, is traveling with no disruption through the seed and bulk regions. After the releases interact (0.80 μs), cavitation bubbles start to develop beneath the surface. These grow and coalesce over time. As the second shock enters the cavitated region (1.40 μs), the bubbles collapse, resulting in substantial entropy production from the shock. At 2.3 μs, the second shock has reached the free surface. From this snapshot, one could imagine RMI theory producing small jets from the perturbations at the free surface. These are indeed present in the snapshot at 2.7 μs, but the field is soon dominated by a large ejection of material from the entire region that was initially cavitated. Within two microseconds from second shock breakout (SSBO), millimeter long tendrils are created. It is difficult to define a “free surface,” as the SBC ejection results in a disruption of much of the target. While imperfect, these calculations have been sufficient to design and analyze experimental configurations for SBC. For very high strength second shocks, we encounter a potential limitation: our equations of state are insufficient to capture vaporization, which may be significant based on the simulated temperatures.

Preliminary simulations showed that with proper design of the reshock and the dwell time between shocks, the reshock could impact and collapse the bubbles near the free surface. In this work, the shallow configuration is designed to produce bubble collapse ∼100 μm from the static free surface. The resulting cavitation compression produced relatively large quantities of ejecta in the simulations. Similar to shock compression of a foam, shocking and collapsing the bubbles produced high predicted temperatures. A second configuration used a thicker impactor layer and thinner release layer to produce voids in the liquid farther from the free surface, near the center of the target, ∼1.4 mm from the surface. This method is referred to as “deep bubble collapse” and resulted in no significant calculated SBC ejection despite achieving very high temperatures from the pore collapse deep within the material. We also performed simulations with periodic surface features to enable RMI ejecta production. We found the calculated ejected mass to be relatively independent of surface finish for the SBC case because the SBC mechanism was the dominant source of ejecta production.

These simulations identified four criteria for SBC to occur. First, the material must achieve at least a partial PMAR condition. Second, there must be a release into tension that results in shallow cavitation bubble formation. Third, there must be sufficient cavitation to result in a substantial increase in the temperature upon subsequent recompaction. Fourth, the second shock must be “strong enough.” The shock strength is system dependent because it depends on the size of the cavitation field and the porosity fraction. Based on work studying shock interactions with porous media,¹⁸ the maximum entropy extraction from the second shock likely occurs around 50% porosity. Since the second shock is dissipating energy into collapsing the cavitation bubbles, our models suggest that a shock that is too weak or too short in duration will result in partial recompaction and eventual release overtake, which can reduce or eliminate SBC ejecta.

We carried out two experiments using a 40 mm diameter, single-stage powder gun to experimentally test SBC ejection. These experiments used impactors designed to produce “shallow” and “deep” cavitation states at an impact velocity of 2.4 km s⁻¹, similar to what was modeled above. This impact velocity ensures a PMAR state for the Sn target after both initial shock and reshock. We measured free surface velocity, mass ejection, and the magnitude and spatial inhomogeneity of surface heating. We chose Sn as the target material because of the extensive published research on RMI ejection in that material. The target was a uniform Sn disc¹⁹ of 2.76 mm thick and 40 mm in diameter. The surface of the target was diamond turned to minimize production of RMI ejecta. We used 38 mm diameter, multi-layered impactors, tailored to impart different time-varying loading profiles to the target. Each impactor consisted of a first layer of Sn of the same composition as the target, backed by a low shock-impedance second layer of PMMA, backed by Ta as a high impedance third layer.

The experiment parameters are listed in Table I, where V is the impactor velocity as measured by optical beam interrupt; d_tgt is the thickness of the Sn target; and d_Sn, d_PMMA, and d_Ta are the thicknesses of the Sn, PMMA, and Ta layers of the impactor, respectively. Shot 1, the shallow bubble collapse experiment, used a thin Sn layer in the impactor so that cavitation bubble collapse occurred at the free surface of the Sn target. Shot 2, the deep bubble collapse experiment, had a thick Sn layer so that cavitation occurred ∼1.4 mm from the free surface. The shock strengths in both experiments were nominally identical. In addition, the thicknesses were selected to match the delay between the first and second shocks. The only difference is the depth of cavitation.

We employed a suite of diagnostic measurements to analyze the time history of the velocity, mass, and temperature of the ejecta material produced in both shallow and deep bubble collapse experiments. A diagram of all diagnostics except the framing camera is shown in Fig. 3.

Three photonic Doppler velocimetry^20,21 (PDV) measurements of the free surface were fielded in each experiment. They were arranged on a 10 mm diameter circle about the center of the target. Laser light of 1550 nm wavelength was sent from a collimated probe toward the target, normal to the free surface. Backscattered, Doppler-frequency-shifted light was collected in the same probe and transmitted to the interferometer. PDV signal breakout gave the first shock arrival time at the Sn free surface as well as the free surface velocity used to determine the stress of the first shock. The ability of PDV to resolve both the target free surface and ejected material velocities (before the second shock produces a more intense ejecta cloud) gave us the SSBO time at the free surface, ∼1.5 μs. After SSBO, the ejecta field possesses much greater optical density, making determination of the presence or velocity of a nominal free surface difficult or impossible. Velocity analysis of the resulting PDV spectrograms (Fig. 4) shows a significant return from particles with a large range of velocities. The ejected mass quantity cannot be determined from PDV data, but qualitative information is obtained on ejecta particle velocities.

To measure the ejecta mass, we used both Asay foil²² momentum sensors and lithium niobate piezoelectric pin pressure sensors.^23,24 Two tantalum discs with 2.5 mm diameters and thicknesses of 0.5 and 1 mm, and one stainless steel disc with 2.5 mm diameter and 0.38 mm thickness, were used as Asay foils, allowing us to estimate ejected mass. Their recoil velocities were measured with PDV. We observed nominally single-velocity spectra with the PDVs, indicating that the foil integrities were maintained. The variation in foil thicknesses allowed us to cover a range of sensor sensitivities and to assess sensor durability in the expected high ejecta mass conditions, which had not been reported previously. The foils were placed 5 mm radially from the target center, equally spaced as shown in Fig. 3, at a stand-off distance of 20 mm from the initial surface. The time-dependent momentum of material impacting the foil can be estimated using momentum conservation, the density of the foil material, and the foil velocity history recorded by PDV. With the further simplifying assumptions that ejecta were all emitted instantaneously and adhered to the foil upon impact (inelastic scattering), we estimated the total momentum deposited in the foil and subsequently the ejecta mass as a function of velocity. The assumptions required for analyzing Asay foils and their consequences are discussed by Steele et al.²⁵ Note that our foils are much more massive than those generally used to detect RMI driven ejecta. This reduces their sensitivity to first-shock RMI ejecta but makes them capable of measuring the intense second shock SBC ejecta without apparent failure. Relying upon the same assumptions of instantaneous sourcing and ballistic transport, the piezo pins agreed with the foil results at lower (<10 mg cm⁻²) areal masses using the well-established pin mass estimation technique.²⁶ Given that ejecta scattering violates the ballistic transport assumption, we did not focus on absolute quantities of mass, but instead we tried to analyze and compare the relative differences in approximate mass–velocity distribution and total mass production between shots using the analysis approach described by Tregillis et al.²⁷ This also requires that the mass be less than the foil areal mass; our 1 mm thick tantalum foils have an areal mass of 1.67 g cm⁻².

We fielded radiance diagnostics on each experiment to attempt to estimate the temperatures of the Sn target free surface, RMI ejecta, and SBC ejecta. We measured radiance in two complementary ways: with a time-resolved, multi-spectral band pyrometer and with a visible light framing camera. Single-point optical pyrometry using an optical fiber provided a continuous record of the total radiance integrated from a large region of the free surface. A high-speed framing camera imaged the radiance across the target at nine times during the experiment. The pyrometer has high time resolution and moderate spectral resolution, which complement the camera's high spatial resolution, moderate speed, and low spectral resolution. We also fielded a gated spectrometer, providing high spectral resolution, spatially integrated over the same collection area as the pyrometer and temporally integrated after second shock. The spectrometer measurements, not shown in this report, confirmed that line emission was not prevalent, allowing us to analyze the surface radiance assuming approximately blackbody character.

The nature of radiance behavior in free surface targets is only partially understood. In prior experiments, it was found that radiance inhomogeneity (apparent “hot spots”) can occur when a target that has been polished (but not diamond turned) is shocked and then released,^28,29 and particularly when this slightly rougher-surface target melts at the free surface.³⁰ In a single-shock experiment, diamond turning smooths the free surface, which then produces spatially homogeneous radiance under shock, and is less likely to produce apparent surface hot spots.

Radiance inhomogeneity in multiply shocked, free-surface experiments has not been extensively studied. Instead, pyrometric methods to test material properties have focused on maintaining surface homogeneity through the use of anvil windows, while methods to test free-surface-release temperatures have evolved tailored geometries to simplify and isolate phenomena such as ejecta.^14,14 These studies have found that RMI ejecta are formed and transport in vacuum at a steady temperature modestly higher than the expected bulk thermodynamic temperature of the sample upon release from the initial shocked state.

The optical pyrometer, described in detail elsewhere,³¹ measured radiance across six visible and near-infrared bands. The entire system of probe, fibers, optics, and detectors was calibrated for absolute spectral radiance against a blackbody standard. The measured field-of-view (FOV) was determined by the collection fiber, which had a numerical aperture of ∼0.22. The diameter of the measurement FOV began at 26 mm at first-shock breakout at the free surface and narrowed as the target surface moved toward the probe. The condition of optical over-fill by post-release material was maintained throughout the experiment.

We imaged the target surface using a nine-frame, high-speed optical framing camera recording 100 ns duration frames with interframe times of 500 ns. No backlighting was used; we recorded only the radiance image of the target surface. The target was imaged through an optical port in the experiment chamber onto the camera sensor with a Nikon 180 mm focal length lens. The camera spatial resolution at the target surface was ∼100 μm, and the camera has an S20 photocathode with spectral response from approximately 400 to 750 nm.

Spectrograms of free surface velocity measurements are shown in Fig. 4. Figure 4(b) shows the velocity–time spectrogram for the deep bubble collapse configuration, experiment 2. The arrival of the first shock and its release into the PMMA (t = 0) produces a surface moving at a steady velocity of ∼2.4 km s⁻¹ continuing for ∼1.5 μs. Initial shock stress in the Sn was 37.8 GPa. Sn is believed to be fully liquid after a shock of ∼38 GPa followed by stress release.³⁰ Minimal backscatter from ejecta is seen by PDV during this time, as is apparent from the lack of structure in or near the free surface velocity. This apparently suppressed RMI ejecta formation follows from the very minimal (∼15 nm peak-to-valley) periodic surface structure residual to the diamond turning process. During this 1.5 μs period, wave interactions occur within the target. Release waves from the target free surface and the tin-PMMA interface in the impactor collide in the target volume (∼1.4 mm from the target surface) creating a region of tensile strain and a zone of damaged material. The shock wave that reflects from the Ta-PMMA interface in the impactor propagates through the PMMA and into the cavitated region. This reshock from the tantalum overtakes the sample surface at ∼1.5 μs. There is evidence of a minimum visible velocity of ∼3.1 km s⁻¹, possibly the free surface, visible briefly after second shock overtake of the free surface. (Conclusively identifying an effective free surface behind a dense ejecta cloud, after multiple ejecta-producing shocks and releases, is a difficult problem that is beyond the scope of this work.) After the second shock, the velocity spectrogram indicates a wide velocity band of ejecta from roughly the free surface velocity up to ∼5.0 km s⁻¹. The middle of the ejecta velocity band appears at ∼4.0 km s⁻¹.

Figure 4(a) shows the velocity time profile for experiment 1, the shallow bubble collapse configuration. The arrival of the first shock (t = 0) creates a surface moving at a steady velocity of ∼2.4 km s⁻¹. Calculated firstshock stress in the Sn was 37.4 GPa. Minimal evidence of ejecta is seen at this time. Over the next ∼1.5 μs, the release waves from the target free surface and the Sn–PMMA bullet interface collide in the target volume just below the surface (modeling suggests a depth of ∼100 μm) and begin to create a zone of damaged material. As with the deep configuration, this zone is eventually compressed by the second shock, which arrives at the free surface at ∼1.5 μs. After this, the spectrograms from the two experiments differ. The possible hint of a free surface seen in the deep collapse experiment is absent here in the SBC experiment, and the ejecta velocity band is centered at a higher median value, at least ∼4.5 km s⁻¹, with velocity components nearing 6.0 km s⁻¹.

Results of the ejecta mass sensor measurements expressed as mass vs velocity for experiments 1 and 2 are shown in Figs. 5(a) and 5(b), respectively. In the deep bubble collapse configuration (experiment 2), the first measurable quantities of ejecta possess velocities just greater than 3.6 km s⁻¹. The sensors then indicate the arrival of much higher density material with a velocity of ∼3.4 km s⁻¹ with an abrupt rise in signal. (This may be the leading edge of the free surface.) The accumulated ejecta mass with velocities >3.4 km s⁻¹ is of the order 50 mg cm⁻². In the shallow bubble collapse experiment, the mass accumulation sensors show the earliest measurable quantities of ejecta at velocities of ∼3.9 km s⁻¹. The foils show continuing mass accumulation at velocities from 3.9 to 3.4 km s⁻¹. Reported mass at lower velocities is likely unreliable due to wave effects in the foil created by substantial quantities of mass striking at high velocities. This may be demonstrated in Fig. 5(a) as the total mass appears to oscillate. In contrast to experiment 2, the accumulated mass with a velocity of ≥3.4 km s⁻¹ is nearly an order of magnitude greater, ∼300–400 mg cm⁻².

While techniques to analyze ejecta velocities and mass from free surfaces have been researched and reported for some time, techniques for analyzing non-uniform temperature distribution across evolving dynamic free surfaces are less advanced. The breakup of the free surface after release, including jetting and interacting sprays of ejecta, creates both non-uniform temperature and non-uniform effective spectral emissivity that evolve over time. To analyze these data, we frame the analysis of emissivity in terms of areal (apparent) emissivity, multiplying the inherent spectral emissivity of the emitting material surface by its fractional geometric fill of the FOV. In the simplest case, when the temperature and surface geometry are nearly uniform, the geometric fill of the FOV is close to unity, and the apparent emissivity becomes that of the free surface. We used two analysis techniques, depending upon whether this condition was true.

The first technique, which we refer to as uniform areal emissivity analysis, is generally applicable in experiments with a spatially homogeneous temperature across the FOV. The true temperature is bounded by a maximum value (using the lowest reasonable emissivity) and a minimum value (using the highest reasonable emissivity). The estimated temperature is reported as the mean of these two values.³² The high and low values of spectral emissivity are taken from dynamic emissivity measurements of Sn free surfaces shocked to the liquid state and released to atmosphere (100 kPa).³³ These emissivity limits are then expanded by 30% to allow for the additional uncertainty in the emissivity of a free surface. This technique has proven a defensible approach for windowed and diamond turned samples in single-shock experiments, and it compares well with cases where measuring dynamic spectral reflectance is possible utilizing an integrating sphere.³⁴ 

The second technique³⁵ allows free variance of a single “graybody” emissivity value: ɛ = k, {1 > k > 0}, where k is a constant independent of wavelength, allowed to vary at each time step in the data, and chosen to be the value that minimizes integrated discrepancy at all wavelengths between calculated Planck radiance and measured radiance at each wavelength. This method performs well in cases where the FOV is composed of small, hot regions (hot spots) superimposed upon a dramatically colder background. In this case, the FOV fraction factor, which is proportional to the fraction of the FOV filled with hot spots, can become very small, greatly limiting the areal emissivity. To determine which pyrometry analysis to use, we looked for hot spots in the camera images (Fig. 7).

In experiment 1, the camera image taken of shallow bubble collapse immediately after SSBO [Fig 7(a)] shows a nearly uniform, high radiance level across the free surface. In this case, we applied the uniform areal emissivity analysis. Temperature values extracted by this method, with an uncertainty of <5%, are shown in blue as the upper curve of Fig. 6. Before SSBO, the temperature is lower than 1000 K, the measurement limit of the pyrometer. After the second shock arrival, the temperature jumps to nearly 2800 K.

Framing camera images of the deep bubble collapse in experiment 2 show a much lower overall temperature, punctuated by small hot spots comprising <2% of the FOV, Fig. 7(b). The bottom of Fig. 6 shows the uniform emissivity analysis results from this experiment. The lower thick orange line shows T ∼1100 K calculated from the largest wavelength channel (1570 nm), while the upper thick orange line shows T ∼1800K calculated for the smallest wavelength (513 nm). The temperatures calculated from the other channels fall in the shaded orange area between. Fitting the experiment 2 data to one free graybody emissivity and one temperature (the second method) produced the best fit with T ∼2300 ± 200 K and apparent emissivity of ∼0.01. We postulate that hot spots with inherent emissivity of very roughly 0.5 filled ∼2% of the FOV, creating an average areal emissivity of ∼0.01. The emission of tiny hot spots at >2000 K added to the emission from the relatively cold bulk produced combined estimated temperatures within the shaded orange region of Fig. 6. The temperatures are dominated at short wavelengths by emission from the hot spots and at long wavelengths by the cold bulk. Thus, the bulk free surface temperature may be best constrained as being no greater than the high-emissivity temperature calculation (∼1100 K) at the largest wavelength, where hotspot emission was the least significant.

The substantially greater predicted mass ejected from the SBC mechanism, vs the RMI mechanism, is clearly demonstrated by the momentum measurements. The foil data indicate both a large increase in mass traveling at higher velocities and a 4- to 10-fold relative increase in total ejected mass faster than the nominal free surface velocity in the shallow bubble collapse experiment in comparison to the deep bubble collapse experiment. A precise quantitative comparison of the mass–velocity distribution in each experiment is impeded by two possible effects. The first is ejecta-foil collisions violating the ballistic transport assumption. The second is the potential of multiple stress waves propagating through the foil material to increase its sound speed, steepening the wave profiles and eventually coalescing them into a shock. This effect distorts the motion history of the foil surface. Subsequent experiments will address the latter issue to improve ejected mass estimation. The velocity spectrograms (Fig. 4) show visible ejecta produced at higher velocities in the shallow bubble collapse experiment as compared to the deep bubble collapse experiment. The appearance of an apparent visible free surface in the deep case also suggests a smaller ejected mass density in experiment 2.

SBC ejects mass with dramatically higher temperature than either the bulk thermodynamic release state of Sn or RMI ejecta from Sn. The SBC experiment demonstrated a relatively uniform temperature of ∼2700 ± 200 K, with variation across the apparent surface likely less than 200 K. The temperatures seen in the shallow bubble collapse experiment are significantly greater than the expected bulk release temperature of Sn, which is well below 1000 K. They are also greater than the first-shock RMI ejecta temperature (also below the ∼1000 K measurement limit of our pyrometer) and the second-shock RMI ejecta in the deep collapse experiment (perhaps 1100 K).

Bubble collapse far from the free surface created a much lower temperature across the majority of the apparent surface. We estimate the majority of the surface is at T ≤ 1100 K, but we see “hot spots” at ∼2100−2500 K, making up ≤2% of the total surface area. The source of these hotspots is unclear.

In this work, we have described a powerful mass ejection mechanism manifested by materials experiencing multiple shock-and-release to fully liquid state events. The nature of the SBC ejection process differs substantially from the RMI ejection process, with SBC producing much greater quantities of total ejected mass at significantly higher temperatures. Future experiments on Sn and other materials will investigate the precise areal mass quantity (mg cm⁻²) and mass-vs-velocity distribution of SBC ejecta, as well as the mechanism responsible for the decreasing radiance (apparent cooling) of the SBC material after sourcing. The relative prevalence of the SBC mechanism at intermediate depths—greater than the order of 100 μm for SBC but less than order 1 mm for deep collapse, warrants further study. In any application possibly featuring a material undergoing multiple shocks and releasing to a liquid state, the SBC mechanism should be considered. If the timing of the shocks is correct, this effect likely dominates the production and properties of ejection.

The authors would like to thank a number of crucial contributors to this work. Ruben Valencia and Michael Grover provided assistance in target fabrication and finishing as well as powder gun operations. Paul Steele, Steve Compton, and Patrick Younk provided assistance in the design, implementation, and analysis of the Asay foils and foil support system. The NNSS dynamic radiometry fielding team including Carl Carlson, Clifford Cochran, Robert Corrow, and David Esquibel supported pyrometry development and fielding. Finally, we thank Lynn Veeser for thoroughly reviewing this manuscript and providing many helpful comments. Lawrence Livermore National Laboratory is operated under the auspices of the U.S. Department of Energy under Contract No. DE-AC52-07NA27344. Nevada National Security Site is operated by Mission Support and Test Services, LLC for the National Nuclear Security Administration of U.S. Department of Energy under Contract No. DE-NA0003624. Los Alamos National Laboratory is operated by Triad National Security, LLC, for the National Nuclear Security Administration of U.S. Department of Energy under Contract No. 89233218CNA000001. This manuscript was approved for release under Document No. LLNL-JRNL-841345.

The authors have no conflicts to disclose.

G. R. Maskaly: Conceptualization (equal); Funding acquisition (equal); Investigation (equal); Writing – review & editing (equal). G. D. Stevens: Conceptualization (equal); Formal analysis (equal); Methodology (equal); Writing – review & editing (equal). B. M. La Lone: Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – review & editing (equal). W. D. Turley: Funding acquisition (equal); Project administration (equal); Writing – original draft (equal); Writing – review & editing (equal). M. D. Staska: Methodology (equal); Resources (equal). F. M. Najjar: Formal analysis (equal); Investigation (equal). T. M. Hartsfield: Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – original draft (equal); Writing – review & editing (equal).

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