
Determining the Self-Similar Stage of the Rayleigh-Taylor Instability via LLNL’s NIF
Discovery Science Experiments
A. Shimony1,2, C.M. Huntington3, K.A. Flippo4, Y. Elbaz5, S.A. MacLaren3, D. Shvarts1,6and G. Malamud1,2
1Nuclear Research Center Negev, Beer Sheva 84190, Israel
2University of Michigan, Ann Arbor, MI 48109, USA
3Lawrence Livermore National Laboratory, Livermore, CA 94550, USA
4Los Alamos National Laboratory, Los Alamos, NM 87545, USA
5Israel Atomic Energy Commission, Tel Aviv 61070, Israel
6Ben Gurion University of the Negev, Beer Sheva 84105, Israel
We report a novel measurement of the late time self-similar growth constant αBof the Rayleigh-
Taylor instability (RTI) from controlled initial perturbations. To this end, we have developed a
laser-driven experiment, fielded on the National Ignition Facility (NIF), to explore deeply non-linear,
multimode hydrodynamic growth of a planar interface. The measured value is αB= 0.038 ±0.008,
which is consistent with previously reported 2D simulations and closer to previously reported 3D
simulations. This resolved the known discrepancy between experiments and simulations of RTI.
The Rayleigh-Taylor instability (RTI) evolves when a
light fluid pushes a heavy fluid [1,2] and ubiquitous in
physical systems of substantial different length scales like
Inertial Confinement Fusion (ICF) capsules [3] and super-
novae [4]. In the case of a narrowband multimode initial
perturbation, the RTI reaches the regime of self-similar
evolution after G≈3 [5], where Gis the number of bub-
ble merger generations. At this stage, the magnitudes of
the spike and bubble (hSand hB, respectively) are given
by hS/B =αS/BAgt2, where gis a constant acceleration
at the unstable interface, A=ρ2−ρ1
ρ1+ρ2is the Atwood num-
ber ( ρ1and ρ2are the densities of the lighter fluid and
the density of the heavier fluid, respectively), tis the
evolution time, and αis a constant that describes the
growth, which may be different for the spike and bubble.
The value of αBgoverns the mixing process and has been
a topic of research for decades.
Despite active research, significant discrepancies re-
main between theoretical, computational, and experi-
mental calculations of αB. Recent theoretical model re-
sults suggest a value of ∼0.05 for 3D immiscible flu-
ids [5]. This result is consistent with LEM experiments
[6] and bubble-merger models [7,8], but a factor of ∼2
higher than the results of full numerical 3D simulations
[9,10]. This last result is claimed to be attributed to the
entrainment of the two fluids in the bubble caused by a
3D small scale turbulent flow, which reduces the effec-
tive Atwood number by ∼2 for miscible fluids, as shown
in full numerical simulations [11].
To resolve the discrepancy, introduced above, between
experiments and simulations, a precise comparison be-
tween them should be made. The initial perturbations
in most of the past simulations were of a narrowband, so
that long wavelengths were absent in the initial growth
and the resulted evolution of the mixing zone was of bub-
ble merger dynamics. However, in past experiments, the
control over the initial perturbations was limited and ini-
tial long wavelengths could affect growth and increase the
measured value of αB, as demonstrated by simulations
[12]. In this work, we present a novel experiment, with
a controlled narrowband initial perturbation [13], which
was performed on the National Ignition Facility (NIF).
The experimental results were compared to a model and
to a hydrodynamic simulation.
Several aspects of the present experimental configura-
tion were adapted from similar hydrodynamics studies on
the National Ignition Facility (NIF), including the laser
drive, structure of the ablator [14,15], low-density foam
configuration [15,16], and x-ray source [15,17]. Essen-
tial elements are described below; further details can be
found in the relevant publications.
The unstable interface under investigation consisted of
a plastic disk at a density of 1.43 g/cm3, mated to a foam
cylinder. As shown in Fig. 1, the portion of the plastic
facing the hohlraum and served as the “ablator” was plas-
tic doped with a fraction of iodine (C50H47I3, “CHI”),
which served to stop moderate-energy x-rays (Au M-
band >2 keV) from reaching the interface. Following
the CHI ablator, a layer of polyamide-imide plastic (PAI,
C22H14N2O3) formed the bulk of the dense interface, ex-
cept for a 300 µm wide strip of the CHI that spanned
half of the target width at the interface with the foam.
While the PAI was largely transparent to the diagnostic
x-rays, the purpose of this tracer strip was to absorb the
x-rays and provide contrast in the image, highlighting
the region location of the dense layer as it mixed into the
low-density foam.
The same x-ray opaque tracer-layer concept that was
used in the plastic was employed in the foam. Most of the
foam cylinder was machined from carbon resorcinol foam
(CRF) at an initial density of approximately 0.08 g/cm3.
In a groove machined half-way through the cylinder, a
300 µm wide strip of carbon foam with a nickel dopant
(C88Ni12) [18], also nominally 0.08 g/cm3, was inserted.
The nickel K-edge at 8.3 keV strongly absorbed the 9
keV diagnostic x-ray signal, providing a contrasting dark
layer where the doped foam was mated to the PAI plastic.
arXiv:2210.06631v1 [physics.flu-dyn] 12 Oct 2022