Estimates of Ta strength at ultrahigh pressures and strain rates using thin-film graded-density impactors

We present results from an experimental technique used to estimate the strength of Ta at extreme pressures (150 GPa) and strain rates (${10}^7{\mathrm{s}}^{-1}$). A graded-density impactor (GDI) was fabricated using sputter deposition to produce an approximately $40-\mu \mathrm{m}$-thick film containing alternating layers of Al and Cu. The thicknesses of the respective layers are adjusted to give an effective density gradient through the film. The GDIs were launched with a 2-stage light gas gun, and shock-ramp-release velocity profiles were measured over timescales of $\sim 10$ ns. Results are presented for the direct impact of the film onto LiF windows, which allows for a dynamic characterization of the GDI, as well as from impact onto thin ($\sim 40\mu \mathrm{m}$) sputtered Ta samples backed by a LiF window. The measurements were coupled with mesoscale numerical simulations to infer the strength of Ta, and the results agree well with other high-pressure platforms, particularly when strain-rate, microstructural, and thermodynamic-path differences are considered.

Figure 1

Characteristic conditions for several Ta experimental data sets available in the literature, adapted from [7]. Quasistatic and split-Hopkinson bar experiments cover low-pressure, moderate-strain-rate conditions [8], and recent Richtmyer-Meshkov (RMI) techniques have extended to higher rates [7]. Diamond anvil cells can be used to access low-rate, high-pressure conditions [9]. Magnetic loading on the Z machine has been used to dynamically compress to high pressures at high rates [10], while laser platforms such as Omega [11, 12] and NIF [3] have been used to access similar pressures at even higher rates.

Figure 2

SEM cross section view of the Cu/Al GDI with individual layers oriented vertically. The image in (a) shows the base Cu layer, graded multilayer, and Al cap. Images in (b) and (c) show Cu-rich and Al-rich portions of the multilayer, respectively. The Al layers appear dark while the copper layers are lighter.

Figure 3

Diffractogram obtained from sputter-deposited Cu/Al multilayer showing evidence for Cu, Al, and trace ($<0.1$ wt. % ${\mathrm{CuAl}}_2$). The known reflections of candidate materials are displayed below the diffractogram.

Figure 4

Diffractogram obtained from sputter-deposited Ta thin film demonstrates a body-centered cubic structure. The reflections associated with several candidate Ta structures are included below the diffractogram. A cross section scanning electron micrograph of Ta thin film on LiF substrate showing the full thickness is provided as the inset. The film was locally coated with Pt to minimize rounding of exposed Ta film surface during focused-ion-beam sectioning. The horizontal lines observed in this image are mostly artifacts introduced as part of the sectioning process.

Figure 5

Plate impact experiment configuration. The GDI impacted a sample of interest (either 5 $\mu \mathrm{m}$ of Al or 43 $\mu \mathrm{m}$ of Ta) and PDV was used to measure both the impact velocity and the velocity of the sample/LiF window interface. A PDV array was used to measure the velocity of 12 distinct spots across $\sim 1$ mm to provide the statistics necessary to obtain the required precision in the sample velocity.

Figure 6

Initial (time = 0) spatial density profile for the direct numerical simulation of a GDI impact showing each distinct layer of Al and Cu. The low-pass filter was applied as a visual aid representing the average density gradient through the GDI.

Figure 7

Simulated pressure histories at the interface between the GDI and either a thin (5 $\mu \mathrm{m}$) Al or thick Ta (43 $\mu \mathrm{m}$) sample. Impact velocities were tailored to give the same initial shock pressure.

Figure 8

The subset of the 12 measured PDV spectrograms in (a) represents the range of signal quality, where the gray shaded regions represent the 95% confidence intervals on the velocity extraction (which visually collapse to a line in regions of high signal quality). After aligning the initial shocks to time = 0, the estimated variance for each signal is used to produce (b), the weighted averaged velocity for each experiment. Shaded regions represent the 95% confidence intervals in the mean velocity, although they are mostly indistinguishable from the linewidth.

Figure 9

Drive calculation for experiments GDI-1 and GDI-2. The differences between the GDI-1 drive characterization simulation and experiment are linearly mapped to the simulated loading velocity in the GDI-2 Ta simulation.

Figure 10

Best-effort simulations for the Ta experiments using the mapped velocity boundary condition shown in Fig. 9. The Ta EOS is assumed known and determines the time shift in the experimental data, giving the excellent agreement in the postshock ramp.

Figure 11

The analysis methodology described in [6, 10] is used to determine the Lagrangian wave speed as a function of particle velocity. The initial loading was constrained to agree with the 93524 Sesame EOS, while the unloading represents the elastic-plastic transition that is dictated by the Ta strength response. The inset contains the simulated equivalent plastic strain rates as a function of time halfway into the Ta sample.

Figure 12

Shear modulus ($G$) and change in shear stress ($\mathrm{\Delta }\tau$) estimated from the wave speed profiles shown in Fig. 11. Data from ramp-release experiments on Z [10] and laser RT experiments on Omega [2, 11] are shown for comparison. The strain rate curves in the lower panel are based on the zero-pressure interpolation between split-Hopkinson bar [8] and RMI [7] data and a $G/G_0$ pressure scaling based on the Sesame 93524 $T=0$ shear modulus.