Article

Burning plasma achieved in inertial fusion

https://doi.org/10.1038/s41586-021-04281-w Obtaining a burning plasma is a critical step towards self-sustaining fusion energy1.
Received: 24 May 2021 A burning plasma is one in which the fusion reactions themselves are the primary
source of heating in the plasma, which is necessary to sustain and propagate the burn,
Accepted: 25 November 2021
enabling high energy gain. After decades of fusion research, here we achieve a
Published online: 26 January 2022 burning-plasma state in the laboratory. These experiments were conducted at the US
Open access National Ignition Facility, a laser facility delivering up to 1.9 megajoules of energy in
Check for updates pulses with peak powers up to 500 terawatts. We use the lasers to generate X-rays in a
radiation cavity to indirectly drive a fuel-containing capsule via the X-ray ablation
pressure, which results in the implosion process compressing and heating the fuel via
mechanical work. The burning-plasma state was created using a strategy to increase
the spatial scale of the capsule2,3 through two different implosion concepts4–7. These
experiments show fusion self-heating in excess of the mechanical work injected into
the implosions, satisfying several burning-plasma metrics3,8. Additionally, we
describe a subset of experiments that appear to have crossed the static self-heating
boundary, where fusion heating surpasses the energy losses from radiation and
conduction. These results provide an opportunity to study α-particle-dominated
plasmas and burning-plasma physics in the laboratory.

Fusion research fundamentally aims to create a system that produces is to reach Qα ≈ 2 (Q ≈ 10), whereas the record from the JET tokamak12
more energy than was required to create it, a necessary condition is Qα ≈ 0.13 (Q ≈ 0.67).)
for energy applications; in practice, the fusion reaction must be A burning plasma is distinct from other scientific milestones in
self-sustaining, with self-heating overtaking loss mechanisms, termed inertial fusion. In 2014, the first milestone of ‘fuel gain’13 (Gfuel > 1) was
‘ignited’9. Such conditions are reached in astrophysical objects includ- achieved14, in which the fusion yield exceeds the energy delivered to
ing the cores of stars, novae and type 1a supernovae, and in thermonu- the fuel; this corresponds to approximately 12–14 kJ of yield at the
clear weapons. Ignition in the laboratory requires heating the fuel to National Ignition Facility (NIF). At 20–22 kJ, the yield was approximately
incredibly high temperatures, where it becomes a ‘plasma’ and fusion doubled by self-heating feedback, termed ‘α-heating’15. The next scien-
reactions readily occur, while also controlling energy losses. Several tific milestone is a burning plasma, as described previously; this is the
approaches have been developed to heat and confine plasma over scientific milestone achieved in this work. No net energy gain, G, relative
the past several decades, with most pursuing deuterium–tritium (DT) to the laser energy is expected for a burning plasma. Because of energy
fuel, which most easily achieves ignition. The dominant approaches to losses incurred in achieving the required compressed state, ICF implo-
plasma confinement are ‘inertial’, an impulsive burn while the fuel is sions must achieve ignition before a net energy gain is possible. A net
confined by its own inertia, and ‘magnetic’, in which specialized con- energy gain would require fusion yields greater than the laser energy,
figurations of magnetic fields provide confinement to the charged 1.9 MJ. Although short of ignition or energy gain, a burning plasma
particles in the plasma. In order for a DT fusion (D + T → α (3.5 MeV) + (Qα > 1) is a new physics regime for laboratory fusion1,11,16. Studying
n (14 MeV)) plasma to become thermally unstable and ignite, it must burning plasmas will elucidate other new physics in this regime, such as
first obtain a ‘burning’ state. In this regime, self-heating from α-particle self-heating-driven instabilities or kinetic effects in the plasma, which
deposition exceeds the external heating input into the DT8; this ratio probably depend on the confinement approach.
is denoted Qα, where the self-heating is taken relative to the heating In a tokamak, the predominant approach to magnetic confinement,
power to the plasma—for inertial fusion this is the PdV compressional once the plasma discharge is generated by resistive heating, exter-
work on the fuel and not the total laser energy (P, pressure, dV, volume nal power sources, such as radio-frequency antenna, provide addi-
change). Qα > 1 is a burning plasma. tional plasma heating as the plasma is brought to fusion conditions.
A burning-plasma state signifies a transformational change to the In indirect-drive ICF, the way energy is delivered to the fusion fuel is
energy and power balance in the DT plasma, opening up the potential different and much less direct. At NIF17, 192 lasers deliver up to 1.9 MJ
for rapidly increasing performance. In the impulsive case of inertial of frequency-tripled light into a high atomic number (Z) ‘hohlraum’
confinement fusion (ICF)10, Qα can be stated either as a power during (Fig. 1) that serves the purpose of an X-ray converter generating a nearly
burn, or as an energy integrated over the burn duration, whereas for Planckian X-ray bath, an approach known as ‘indirect drive’18. The inci-
the near-steady-state operation of magnetic fusion energy (MFE), Qα is dent beam-by-beam laser pointing and power in time are designed6
a statement of power. As α-particles carry 1/5 of the total fusion energy to generate a specific radiation temperature (Trad) history (Fig. 1, bot-
per D + T reaction, Qα = Q/5, where Q is the total fusion energy compared tom left) inside the hohlraum, with sufficient uniformity in a way that
to the heating energy supplied. (Or in the MFE case, stated in terms of is matched to specifics of the target geometry and the desired final
total fusion power over heating power; for example, the goal of ITER11 plasma state. The exposed surface of a capsule at the centre of the

A list of authors and affiliations appears at the end of the paper.

542 | Nature | Vol 601 | 27 January 2022

 Inner cone Implosion process
Capsule schematic 23.5° Bath of hohlraum
30° 23.5°
HDC (80 μm) 30° Outer cone X-ray radiation
44.5° 44.5°
DT ice
1,00
DT (55–65 μm) 50°44.5° 44.5° 50°
vapour 50°
Cryogenic cooling ring DT ice
0–1,0

DT vapour
6.4 mm Target
50 μ

Fuel (DT)
m

Fuel fill tube

11.2 mm
Ablating shell
Radiation drive
Laser power Hohlraum
Hohlraum radiation
temperature 3.64 mm
300 Compress Time
400
Power (TW)

Laser entrance and heat

Trad (eV)

200 hole fuel

200
100
Fusion burn
0 0
0 5 10
Time (ns) Laser quads

Fig. 1 | Schematic of the indirect-drive inertial confinement approach to laser power pulse-shape (blue) and associated hohlraum radiation
fusion. Centre, A typical indirect-drive target configuration with key temperature (green). Right, At the centre of the hohlraum, the capsule is
engineering elements labelled. Laser beams (blue) enter the hohlraum through bathed in X-rays, which ablate the outer surface of the capsule. The pressure
laser entrance holes at various angles. Top left, A schematic pie diagram generated drives the capsule inward upon itself (an implosion) which
showing the radial distribution and dimensions of materials in diamond compresses and heats the fusion fuel during the implosion process.
(high-density carbon, HDC) ablator implosions. Bottom left, The temporal

hohlraum absorbs approximately 10–15% of the X-rays, causing the advances19–21 have generated experiments with approximately 50 kJ
outer edge of the capsule (the ablator) to ionize, generate high pres- fusion yields that were close to the burning-plasma threshold3. These
sures of the order of hundreds of Mbar (1 Mbar = 1011 Pa), and expand experiments used capsules with similar inner radii, between 0.91 and
away from the capsule—a process termed ablation. A shell of cryogenic 0.95 mm. Within the maximum laser energy NIF can deliver, these pre-
DT fuel is layered against the inside surface of the ablator, which is in vious designs were limited in the energy coupled to the capsule, and
partial-pressure equilibrium with DT vapour in the centre of the capsule thus in the fuel kinetic energy, by the ability to control the symmetry
(Fig. 1, top left). The inwardly directed acceleration caused by the abla- of the radiation environment within the hohlraum, primarily because
tion drives the capsule and DT fuel inwards upon itself (an implosion, an ablated plasma bubble expands from where the outer beams hit the
shown schematically at the right of Fig. 1) with enormous acceleration wall (Fig. 1), intercepting the inner beams and thereby suppressing
(about 1014 m s−2) obtaining velocities of approximately 350–400 km s−1 drive at the hohlraum waist22,23. Two tactics have been used to enable
in a matter of nanoseconds. Most of the X-ray energy (about 92–95%) symmetry control with more efficient hohlraums driving larger cap-
absorbed by the capsule is consumed by the ablation process, but as a sules: adjusting cross-beam energy transfer between the outer to inner
result the DT fuel obtains considerable (about 10–20 kJ) kinetic energy beams4,24,25 by changing the laser wavelength separation (Δλ); and
inside a very small volume. incorporating a pocket in the hohlraum wall at the outer beam loca-
Shortly after the DT fuel acquires peak kinetic energy, the pressure tion to delay the bubble propagation5. These tactics have been used
(P) inside the implosion rises markedly, to levels of many hundreds to design higher-efficiency hohlraums that control symmetry; we use
of Gbar (1 Gbar = 1014 Pa), as kinetic energy is converted into internal these hohlraums to drive capsules that are about 10% larger than prior
energy in the DT (a process termed stagnation). An ICF implosion experiments to realize the strategy for achieving a burning plasma laid
is a pressure amplifier, sacrificing absorbed energy to achieve high out in a previous work2. These experimental campaigns are known as
energy density and central pressures that are factors of thousands ‘Hybrid E’ and ‘I-Raum’; the Hybrid-E campaign uses Δλ exclusively,
higher than the pressure at the ablation front. The high central pres- whereas I-Raum uses a combination of the pocket and Δλ. Key data
sure is necessary because only a small fraction of the energy at NIF and analysis supporting this burning-plasma analysis are given in
can ever be coupled into the DT fuel, and heating a large mass of DT complementary papers: Ross et al.7 (experiments) and Kritcher et al.6.
fuel is energetically costly, as reflected in the heat capacity of DT, Four experiments with these new designs have been conducted that
cDT = 115 kJ mg−1 keV−1 (9.9 × 103 J kg−1 K−1). In these experiments the total have generated record performance at NIF, with triple the fusion yield
fuel mass is approximately 200 μg and the hot-spot mass is approxi- compared to past experiments4,19,21, to a maximum of approximately
mately 20–30 μg. As a high ion temperature (Ti) is also needed for 170 kJ reported here. The experiments are referred to by an experiment
fusion, while the fuel stagnates at the centre of the implosion, the DT number denoting the date of the experiment (for example, in the format
forms a hot spot from the fuel’s inner surface and PdV work is done on NYYMMDD, where YY = year, MM = month and DD = day). N201101 and
the hot spot, generating very high ion and electron temperatures in N210207 were experiments using the Hybrid-E platform, and N201122 and
near thermal equilibrium (Ti ≈ Te ≈ 4–5 keV, 1 keV = 1.16 × 107 K, where Ti N210220 were experiments using the I-Raum platform. The experiments
and Te are the ion and electron temperatures). If the conditions of high in November (N201101 and N201122) achieved much higher performance
temperature and pressure are achieved, the hot spot initiates copious relative to past work owing to their increased scale and favourable implo-
DT fusion reactions and self-heating further increases Ti. sion design parameters, yet each suffered from low-mode degradations;
ICF experiments have already demonstrated considerable fusion these low-mode asymmetries were mitigated on the subsequent experi-
performance enhancement from self-heating14,15, and more recent ments (N210207 and N210220), resulting in higher performance6,7.

Nature | Vol 601 | 27 January 2022 | 543

Article
a High Foot b
12 CH LGF 12
HDC N210220
BigFoot
10 Hybrid B 10
I-Raum
HyE-1100 N210207
8 Hybrid-E 8
N201122

Gfuel

Gfuel
6 6
N201101

4 4

2 2

0 0
0 2 4 6

17 1

18 7

19 8

20 0

20 1

21 2

21 7
20
0

0
06

08

01

11

11

11

02

02
1.95 × 107pT 1.6 W

17
N

N
c 35 High Foot d 3.0
CH LGF
HDC N210207
30 N210220 2.5
BigFoot
Hybrid B
25 I-Raum
2.0
HyE-1100
20 Hybrid-E

ED/KEfuel
N201122 N201101
ED (kJ)

1.5
15
1.0
10

5 0.5

0 0
0 5 10 15 20
01

27

28

10

01

22

07

20
06

08

01

11

11

11

02

02
KEfuel (kJ)
17

17

18

19

20

20

21

21
N

N
Fig. 2 | Simple metrics for assessing a burning plasma. a, Total fuel gain distribution c, Total α-heating energy versus fuel kinetic energy, Eα/KEfuel > 1
versus Lawson-like parameter; Gfuel > 5 corresponds to the burning-plasma corresponds to Qα > 1. d, Probability distributions in Eα/KEfuel criteria for
regime. b, Probability distributions for Gfuel for high-performing experiments. high-performing experiments. Error bars in a, c are 1 standard deviation (s.d.)
In these plots the width of the shaded region is proportional to the probability and are shown only for the I-Raum and Hybrid-E points. Historical data are from
distribution and the solid lines mark the 16th, 50th and 84th percentiles of the refs. 4,14,15,19–21,29–31.

On each NIF experiment a comprehensive suite of optical, X-ray As can be seen in Fig. 2a, most experimental series show a linear
and nuclear diagnostics measure key aspects of the implosion per- trend between Gfuel and PT 1.6 i τ and have Gfuel ≤ 5; however, the Hybrid-E
formance. Key data are shown in Extended Data Table 1: the total and I-Raum implosion series show a transition to a super-linear trend
fusion yield in kJ, ion temperature (Ti, measured from DD reactions)7, between Gfuel and PT 1.6 i τ (as expected when self-heating exceeds the
hot-spot volume and burn width in ps. For a full description of the hot-spot internal energy) and have Gfuel > 5. The non-burning-plasma
experimental data and changes between the experiments, see ref. 7. regime is denoted by the grey shaded region (Gfuel < 5). In this and
Analytic models using these data are used to infer characteristics of the following figures, historical data from NIF are shown from
the implosion process and hot spot including the pressure, hot-spot refs. 4,14,15,19–21,29–31, labelled by the names of those predecessor cam-
internal energy, implosion velocity and peak kinetic energy in the paigns.
fuel during implosion, PdV work done on the hot spot, and areal Figure 2b shows the probable distribution of the Gfuel values plot-
density of the hot spot in g cm−2. These quantities are required to ted in Fig. 2a, with the probability distribution in the inferred data
evaluate the burning-plasma criteria. Most of these inferences are quantities included to evaluate the uncertainty (Methods). For com-
described in a previous work26 and Methods; the implosion velocity parison, we include a set of previous high-performing NIF experiments
(vimp) can be inferred from the time of maximum neutron output (‘bang from refs. 4,19,21. The abscissa of Fig. 2b are NIF experiment numbers;
time’) and an implosion dynamics ‘rocket model’ that is calibrated although several experiments in years prior to November 2020 came
to near-neighbour surrogate experiments in which the implosion very close to the threshold of Gfuel = 5, only the experiments reported
trajectory is tracked radiographically27,28. here have so far clearly surpassed it (see Extended Data Table 1 for
Although it would be desirable to have a direct measure that indicates values, where the quoted likelihood is the fraction of the distribution
a burning plasma, such a measurement is not yet known to exist, so above the threshold).
inferences from data must be used instead. Gfuel has a direct connec- Alternatively, comparing the total energy produced in α-particles,
tion to ignited fusion requirements and suggests a simple metric for Eα = Y/5, to the peak kinetic energy of the DT fuel, KEfuel (Fig. 2c), is
assessing a burning plasma from Gfuel = Y/EPdV,tot, where Y is the fusion another simple metric. Similar to Fig. 2b, Fig. 2d shows the probable
yield and EPdV,tot is the total PdV work on the fuel (see Methods for how range of Eα/KEfuel, with normally distributed uncertainties in the input
this quantity is evaluated). Figure 2a (also Extended Data Table 1) shows data versus experiment number for the eight highest performing DT
a plot of Gfuel data from many DT implosions at NIF versus the product experiments at NIF, where again only these four experiments clearly
PT1.6τ, where τ is a confinement time; this is a Lawson-like criterion exceed Eα/KEfuel > 1 (see Extended Data Table 1 for values). Because
applicable for Gfuel (Methods). indirect-drive implosions have a small fraction of ablator mass

544 | Nature | Vol 601 | 27 January 2022

 a Criteria (385 km s–1) b
HDC I-Raum
fDPD > Pb + Pe
BigFoot HyE-1100
High Foot Hybrid B Hybrid-E
CH LGF
6
N210207 2.0
N210220

5
1.5

vcond /vimp
Ti (keV) 4
N201122 1.0
N201101

3
0.5

2
0.1 0.2 0.3 0.4 0

17 1

18 7

19 8

20 0

20 1

21 2

21 7
20
URhs (g cm–2)

0
06

08

01

11

11

11

02

02
17
N

N
c High Foot d 2.5
N210207
CH LGF
HDC
15 BigFoot N210220 2.0
Hybrid B
I-Raum
HyE-1100 N201122
1.5
0.5ED/EPdV
Hybrid-E
0.5ED (kJ)

10 N201101

1.0

5
0.5

0 0
0 5 10 15
01

27

28

10

01

21 2

21 7
20
2

0
06

08

01

11

11

11

02

02
EPdV (kJ)
17

17

18

19

20

20
N

N
Fig. 3 | ICF-specific burning-plasma metrics. a, Criteria on temperature and plasma. c, Criteria on α-heating and PdV work from a previous work8, including
hot-spot ρR established by Hurricane et al. 3. Previous experiments are shown estimates from data inferences (solid symbols) and from 2D simulations (open
as points, and the present four experiments are shown as full probability symbols). d, Probability distribution for experiments exceeding the Betti
distributions (red, N201101; blue, N201122; purple, N210207; grey, N210220), criteria. For these experiments distributions are shown for data-inferred E PdV,hs
with contours enclosing 80% of the distribution. A single contour of (blue) and using 2D simulations (orange). Error bars in a, c are 1 s.d. and are
equation (1) for vimp = 385 km s−1 is given by the solid black line. b, Probability shown only for the I-Raum and Hybrid-E points. Historical data are from
distribution for experiments exceeding the Hurricane criterion, >1 is a burning refs. 4,14,15,19–21,29–31.

remaining at peak velocity it is important to point out that KEfuel is not thermonuclear reactivity ⟨σν⟩ is a function of the hot-spot conditions,
the total implosion kinetic energy at peak velocity; instead, because the specifically the temperature; we use the ⟨σν⟩ evaluation of Bosch and
stagnating shell is compressible and extended radially, only a fraction Hale32. Figure 3a shows the experiments in hot-spot temperature and
of the kinetic energy of the ablator can be converted to internal energy areal density parameter space. Previous experiments are shown as
before peak burn and disassembly. points, and the present four experiments are shown as full probability
Although Gfuel and Eα/KEfuel are suggestive metrics for an ICF burning distributions (red, N201101; blue, N201122; purple, N210207; grey,
plasma, two more rigorous and more stringent metrics already exist N210220), with contours enclosing 80% of the distribution. In Fig. 3a
in the literature3,8. The burning-plasma statement that ‘α-deposition is a single contour of equation (1) for vimp = 385 km s−1, representative
the dominant source of plasma heating’ is complicated by the temporal of these experiments, is shown. When evaluating the criteria for the
nature of an implosion, where the PdV work on the hot spot that does actual inferred velocity of each experiment, with uncertainty, is used.
the heating comes before the time of peak fusion rate, a consideration These are the first experiments to exceed the Hurricane criterion, as
that is not analogous to MFE. Prior works by Hurricane et al. gave a clearly shown by the probability distributions in Fig. 3b. The likelihood
condition on velocity (vcond)2,3 relative to the plasma conditions, which of these four experiments exceeding the criteria is 89% (N201101), 79%
we slightly modify (Methods) to: (N201122), and 100% for both N210207 and N210220.
Equation (1) should be roughly equivalent to the burning-plasma cri-
⟨σv ⟩ teria found by Betti et al. (3.5× yield amplification and 0.5Eα/EPdV,hs > 1)8,
vcond(ρRhs, Ti) = 5.3 × 1025ρRhs > vimp (1)
Ti but for completeness we use both. The first criterion by Betti et al.,
Yamp ≥ 3.5, is satisfied by our inferred yield amplifications given in
in units of keV, g, cm and s. Here, ρRhs is the hot-spot areal density and Extended Data Table 1, inferred with the prescription in ref. 26 and from
⟨σν⟩ is the fusion reactivity. two-dimensional (2D) simulations6 using HYDRA33. Two quantities are
To evaluate the Hurricane metric, the temperature and areal density required to evaluate the second Betti et al. burning-plasma metric.
of the hot spot, and the implosion velocity, are needed (Methods). The The α-deposited energy (Eα) is straightforward as it is simply 20% of

Nature | Vol 601 | 27 January 2022 | 545

Article
High Foot HDC Hybrid B HyE-1100
CH LGF BigFoot I-Raum Hybrid-E

400 N2
10 6 Ignition
20 N210207
N210220 7

N180128

Hot-spot pressure (Gbar)

N201101
300 0.8
× 4
N201122
0.6× 101
0.5 0.6 0.7

Yamp
0.4×
200 N210220
N210207
Burning plasma N201122
0.2×
0.1× N201101

100 N180128

0 100
0 5 10 15 20 0 0.2 0.4 0.6 0.8 1.0
Hot-spot energy (kJ) ITFXnD ≈ F3

Fig. 4 | Parameter space relevant for proximity to ignition. Left, hot-spot highest performing ICF experiments so far and the closest to ignition. The inset
pressure and energy. The product P2 Ehs is representative of proximity to shows these experiments in detail with both inferred (solid) and simulated
ignition; contours of this metric relative to N210207 are shown by the dashed (open) Yamp. Error bars are 1 s.d. and are shown only for the I-Raum and Hybrid-E
grey curves. Right, yield amplification (Yamp) versus ITFXnα. These are the points, plus shot N180128. Historical data are from refs. 4,14,15,19–21,29–31.

the measured total fusion yield (given in Extended Data Table 1), which capsule absorbed energy (giving capsule gain of about 0.7–0.8) and
is approximately 20 kJ for the first two experiments, about 33 kJ for are an order of magnitude greater than the input energy transferred to
N210207, and about 31 kJ for N210220. The second input for these cri- the fusion fuel. Moreover, the total fusion power (5mPα where Pα is the
teria is the PdV work done upon the hot spot, which must be inferred; power per unit mass) generated in the two highest performing experi-
however, such inferences are prone to large uncertainties in the pres- ments are at petawatt levels (for example, approximately 1.6 ± 0.2 PW
ence of considerable α-heating and bremsstrahlung X-ray losses. for N210207).
We perform this inference in two ways (Methods), first using an analytic In the burning-plasma regime, self-heating can overtake loss mecha-
hydrodynamic piston model34 of an implosion, and second by extract- nisms, which include bremsstrahlung losses, thermal conductivity
ing PdV work from the 2D radiation–hydrodynamics simulations that and negative PdV work upon expansion. Simple expressions for the
best match the experimental observables described in ref 6. These two power-balance terms are given in the Methods and values for the
estimates are used to estimate a range in hot-spot PdV work, and are four experiments are given in Extended Data Table 1. Here, we use a
both plotted in Fig. 3c compared to previous experiments at NIF, with bremsstrahlung enhancement factor fb ≈ 1.15 that is inferred from the
a 1–1 line to denote the burning-plasma regime (above the line). Prob- data35. The first two experiments have self-heating comparable to the
ability distributions for the metric quantity itself are shown in Fig. 3d. radiation losses. An important new regime is when self-heating power
From Betti’s criteria, with the experimental (simulated) EPdV,hs, we assess (Pα) is greater than both the radiation (Pb) and conduction losses (Pe)—
that 74% (0%) and 97% (2%) probability for experiments N201101 and that is, fαPα > Pb + Pe, where fα is the fraction of α-particles stopping
N201122, respectively, are in the burning-plasma regime. The difference in the hot spot36. A contour for this regime is shown in Fig. 3a by the
in probability reflects the fact that the simulated PdV work is higher black dashed line. Experiment N210220 is close to entering this regime,
and thus is more pessimistic for satisfying the criteria, with the simu- and we infer that experiment N210207 has entered this regime with
lated values for these experiments being below the burning-plasma 82% likelihood. The level of α-heating in this work is still short of that
threshold. With improved performance, experiments N210207 and required for ignition.
N210220 are assessed to be in the burning-plasma regime with 100% To achieve ignition—defined as a yield amplification (Yamp ≈ 20–30)
confidence by both experimental and simulated methodologies. These consistent with about 1 MJ fusion yield37, and then high gain—further
correspond to an an inferred Qα ≈ 1.4–1.6 for experiment N210207, and progress is needed. Figure 4 shows these experiments in the larger
Qα ≈ 1.3–2.0 for experiment N210220. context of ignition, in the parameter space of hot-spot pressure and
Several metrics for assessing whether these implosions created a energy (Fig. 4, left) and in yield amplification versus a Lawson-like
burning-plasma state have been discussed and presented in Extended parameter called the ‘ignition threshold factor’ experimentally inferred
Data Table 1 and Figs. 2, 3. In each case, burning-plasma likelihoods are (ITFX)8,26,38 for conditions without α-heating (nα) (Fig. 4, right). Figure 4,
calculated by propagating uncertainties in each quantity through the right, plots this quantity as ITFX0.34
nα , which is approximately equivalent
metric (Methods), shown in the figures with likelihood values discussed to χnα as defined previously8. Proximity to ignition can be gauged
and summarized in Extended Data Table 1. Quantitatively, we see that qualitatively in terms of the product P2Ehs (equivalent to (ρRhsTi)3), or
the first two (N201101 and N201122) are probably in the burning-plasma in terms of ITFXnα or χnα ≈ 1, representing ignition. Figure 4, left, shows
regime by all metrics, except Betti’s, when evaluated with the simu- contours of P2Ehs relative to N210207, showing that this metric has been
lated EPdV,hs; the most recent experiments (N210207 and N210220) are improved by a factor of several from previous results. From Fig. 4 we
overwhelmingly likely to have passed this threshold. Qualitatively, our clearly see that these four experiments are the closest to ignition, but
confidence in this conclusion is further increased by the use of multiple a further increase in ITFXnα from approximately 0.6 → 1 is required.
independent metrics. As this Article was being finalized, a new experiment in this series on
The achievement of a burning-plasma state is key progress towards 8 August 2021 produced approximately 1.35 MJ of fusion yield and
the larger goal of ‘ignition’ and overall energy gain in inertial fusion. capsule gain of approximately 5, breaking all previous records. This
The fusion yields reported here (approximately 0.17 MJ) are lower than was announced by our institution in a press release39; this experiment
the input laser energy (approximately 1.9 MJ), but are nearly equal to the will be described in a future publication.

546 | Nature | Vol 601 | 27 January 2022

 As discussed in the complementary papers6,7, these experiments 16. Coppi, B. In Academician A. D. Sakharov. Scientific Works. To His Centenary (eds B. L.
Altshuler, et al.) (Fizmatlit, 2021).
have clear and specific degradation mechanisms, which can be miti- 17. Moses, E. I. et al. The National Ignition Facility: transition to a user facility. In 8th Intl Conf.
gated for further improvement in performance. More generally, the Inertial Fusion Sciences and Applications (IFSA 2013) Vol. 688 012073 (2016).
ICF programme at NIF is pursuing several approaches that can enable 18. Lindl, J. Development of the indirect-drive approach to inertial confinement fusion and
the target physics basis for ignition and gain. Phys. Plasmas 2, 3933–4024 (1995).
additional progress: reducing degradation mechanisms including 19. Le Pape, S. et al. Fusion energy output greater than the kinetic energy of an imploding
low-mode asymmetry40–42 and radiative losses from mix35, further shell at the National Ignition Facility. Phys. Rev. Lett. 120, 245003 (2018).
increasing energy coupled to the capsule4,43, and improving compres- 20. Casey, D. T. et al. The high velocity, high adiabat, “Bigfoot” campaign and tests of
indirect-drive implosion scaling. Phys. Plasmas 25, 056308 (2018).
sion of the fuel44. 21. Baker, K. L. et al. Hotspot parameter scaling with velocity and yield for high-adiabat
In conclusion, we have generated in the laboratory a burning- layered implosions at the National Ignition Facility. Phys. Rev. E 102, 023210
plasma state in which the plasma is predominantly self-heated. (2020).
22. Callahan, D. A. et al. Exploring the limits of case-to-capsule ratio, pulse length, and picket
This was accomplished using inertial fusion implosions at the US energy for symmetric hohlraum drive on the National Ignition Facility Laser. Phys. Plasmas
NIF; previous experiments here were just below the threshold for 25, 056305 (2018).
a burning plasma. We increased the capsule scale relative to previ- 23. Hopkins, L. B. et al. Toward a burning plasma state using diamond ablator inertially
confined fusion (ICF) implosions on the National Ignition Facility (NIF). Plasma Phys.
ous work, increased the coupling efficiency from laser energy to Control. Fusion 61, 014023 (2018).
the capsule, and controlled implosion symmetry using new tactics. 24. Michel, P. et al. Symmetry tuning via controlled crossed-beam energy transfer on the
Four experiments have been conducted that have passed the thresh- National Ignition Facility. Phys. Plasmas 17, 056305 (2010).
25. Glenzer, S. H. et al. Symmetric inertial confinement fusion implosions at ultra-high laser
old for a burning plasma by several metrics, with especially high con- energies. Science 327, 1228–1231 (2010).
fidence on the most recent two experiments. Additionally, the highest 26. Patel, P. K. et al. Hotspot conditions achieved in inertial confinement fusion experiments
performing experiment (N210207) is in a more stringent regime on the National Ignition Facility. Phys. Plasmas 27, 050901 (2020).
27. Landen, O. L. et al. Capsule implosion optimization during the indirect-drive National
where the self-heating surpasses energy losses from radiation and Ignition Campaign. Phys. Plasmas 18, 051002 (2011).
conduction. Although these results are short of total energy gain from 28. Meezan, N. B. et al. X-ray driven implosions at ignition relevant velocities on the National
the system owing to the inherent inefficiencies of ICF, these experi- Ignition Facility. Phys. Plasmas 20, 056311 (2013).
29. Baker, K. L. et al. High-performance indirect-drive cryogenic implosions at high adiabat
ments represent a substantial step towards this goal with record on the National Ignition Facility. Phys. Rev. Lett. 121, 135001 (2018).
values of parameters that assess our proximity to ignition at NIF. 30. Döppner, T. et al. Achieving 280 Gbar hot spot pressure in DT-layered CH capsule
implosions at the national ignition facility. Phys. Plasmas 27, 042701 (2020).
Several promising avenues for further increases in performance are
31. Hohenberger, M. et al. Integrated performance of large HDC-capsule implosions on the
identified and will be pursued by the US inertial fusion programme, National Ignition Facility. Phys. Plasmas 27, 112704 (2020).
in addition to novel physics in the burning-plasma regime such as 32. Bosch, H.-S. & Hale, G. Improved formulas for fusion cross-sections and thermal
reactivities. Nucl. Fusion 32, 611–631 (1992).
α-particle-driven processes.
33. Marinak, M. M. et al. Three-dimensional HYDRA simulations of National Ignition Facility
targets. Phys. Plasmas 8, 2275–2280 (2001).
34. Hurricane, O. A. et al. An analytic asymmetric-piston model for the impact of mode-1 shell
Online content asymmetry on ICF implosions. Phys. Plasmas 27, 062704 (2020).
35. Pak, A. et al. Impact of localized radiative loss on inertial confinement fusion implosions.
Any methods, additional references, Nature Research reporting sum- Phys. Rev. Lett. 124, 145001 (2020).
maries, source data, extended data, supplementary information, 36. Zylstra, A. B. & Hurricane, O. A. On alpha-particle transport in inertial fusion. Phys.
Plasmas 26, 062701 (2019).
acknowledgements, peer review information; details of author contri-
37. Christopherson, A. R., Betti, R. & Lindl, J. D. Thermonuclear ignition and the onset of
butions and competing interests; and statements of data and code avail- propagating burn in inertial fusion implosions. Phys. Rev. E 99, 021201 (2019).
ability are available at https://doi.org/10.1038/s41586-021-04281-w. 38. Spears, B. K. et al. Performance metrics for inertial confinement fusion implosions:
Aspects of the technical framework for measuring progress in the National Ignition
Campaign. Phys. Plasmas 19, 056316 (2012).
1. National Academies of Sciences, Engineering, and Medicine. Final Report of the 39. LLNL. National Ignition Facility experiment puts researchers at threshold of fusion
Committee on a Strategic Plan for U.S. Burning Plasma Research (National Academies ignition. Lawrence Livermore National Laboratory – News https://www.llnl.gov/news/
Press, 2019). national-ignition-facility-experiment-puts-researchers-threshold-fusion-ignition
2. Hurricane, O. A. et al. Beyond alpha-heating: driving inertially confined fusion implosions (18 August 2021).
toward a burning-plasma state on the National Ignition Facility. Plasma Phys. Control. 40. Rinderknecht, H. G. et al. Azimuthal drive asymmetry in inertial confinement
Fusion 61, 014033 (2019). fusion implosions on the National Ignition Facility. Phys. Rev. Lett. 124, 145002
3. Hurricane, O. A. et al. Approaching a burning plasma on the NIF. Phys. Plasmas 26, (2020).
052704 (2019). 41. Tommasini, R. et al. Time-resolved fuel density profiles of the stagnation phase of
4. Zylstra, A. B. et al. Record energetics for an inertial fusion implosion at NIF. Phys. Rev. Lett. indirect-drive inertial confinement implosions. Phys. Rev. Lett. 125, 155003 (2020).
126, 025001 (2021). 42. Casey, D. T. et al. Evidence of three-dimensional asymmetries seeded by high-density
5. Robey, H. F., Berzak Hopkins, L., Milovich, J. L. & Meezan, N. B. The I-Raum: a new shaped carbon-ablator nonuniformity in experiments at the National Ignition Facility.
hohlraum for improved inner beam propagation in indirectly-driven ICF implosions on the Phys. Rev. Lett. 126, 025002 (2021).
National Ignition Facility. Phys. Plasmas 25, 012711 (2018). 43. Amendt, P. et al. Ultra-high (>30%) coupling efficiency designs for demonstrating
6. Kritcher, A. et al. Design of inertial fusion implosions reaching the burning plasma regime. central hot-spot ignition on the National Ignition Facility using a Frustraum. Phys.
Nat. Phys. (in the press). Plasmas 26, 082707 (2019).
7. Ross, J. S. et al. Experiments conducted in the burning plasma regime with inertial fusion 44. Landen, O. L. et al. Yield and compression trends and reproducibility at NIF. High
implosions. Preprint at https://arxiv.org/abs/2111.04640 (2021). Energy Density Phys. 36, 100755 (2020).
8. Betti, R. et al. Alpha heating and burning plasmas in inertial confinement fusion. Phys.
Rev. Lett. 114, 255003 (2015). Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
9. Lawson, J. D. Some criteria for a power producing thermonuclear reactor. Proc. Phys. Soc. published maps and institutional affiliations.
B 70, 6 (1957).
10. Nuckolls, J., Wood, L., Thiessen, A. & Zimmerman, G. Laser compression of matter Open Access This article is licensed under a Creative Commons Attribution
to super-high densities: thermonuclear (CTR) applications. Nature 239, 139–142 4.0 International License, which permits use, sharing, adaptation, distribution
(1972). and reproduction in any medium or format, as long as you give appropriate
11. Green, B. & ITER International Team and Participant Teams ITER: burning plasma physics credit to the original author(s) and the source, provide a link to the Creative Commons license,
experiment. Plasma Phys. Control. Fusion 45, 687–706 (2003). and indicate if changes were made. The images or other third party material in this article are
12. Keilhacker, M. et al. High fusion performance from deuterium–tritium plasmas in JET. included in the article’s Creative Commons license, unless indicated otherwise in a credit line
Nucl. Fusion 39, 209–234 (1999). to the material. If material is not included in the article’s Creative Commons license and your
13. Atzeni, S. & Meyer-ter-Vehn, J. The Physics of Inertial Fusion (Oxford Univ. Press, 2004). intended use is not permitted by statutory regulation or exceeds the permitted use, you will
14. Hurricane, O. et al. Fuel gain exceeding unity in an inertially confined fusion implosion. need to obtain permission directly from the copyright holder. To view a copy of this license,
Nature 506, 343–348 (2014). visit http://creativecommons.org/licenses/by/4.0/.
15. Hurricane, O. A. et al. Inertially confined fusion plasmas dominated by alpha-particle
self-heating. Nat. Phys. 12, 800–806 (2016). © The Author(s) 2022, corrected publication 2022

Nature | Vol 601 | 27 January 2022 | 547

Article
A. B. Zylstra1,10 ✉, O. A. Hurricane1,10 ✉, D. A. Callahan1, A. L. Kritcher1, J. E. Ralph1, K. Newman1, J.-M. G. Di Nicola1, A. Nikroo1, R. Nora1, M. V. Patel1, L. J. Pelz1, J. L. Peterson1,
H. F. Robey2, J. S. Ross1, C. V. Young1, K. L. Baker1, D. T. Casey1, T. Döppner1, L. Divol1, Y. Ping1, B. B. Pollock1, M. Ratledge5, N. G. Rice5, H. Rinderknecht4, M. Rosen1, M. S. Rubery8,
M. Hohenberger1, S. Le Pape3, A. Pak1, P. K. Patel1, R. Tommasini1, S. J. Ali1, P. A. Amendt1, J. D. Salmonson1, J. Sater1, S. Schiaffino1, D. J. Schlossberg1, M. B. Schneider1,
L. J. Atherton1, B. Bachmann1, D. Bailey1, L. R. Benedetti1, L. Berzak Hopkins1, R. Betti4, C. R. Schroeder1, H. A. Scott1, S. M. Sepke1, K. Sequoia5, M. W. Sherlock1, S. Shin1,
S. D. Bhandarkar1, J. Biener1, R. M. Bionta1, N. W. Birge2, E. J. Bond1, D. K. Bradley1, T. Braun1, V. A. Smalyuk1, B. K. Spears1, P. T. Springer1, M. Stadermann1, S. Stoupin1, D. J. Strozzi1,
T. M. Briggs1, M. W. Bruhn1, P. M. Celliers1, B. Chang1, T. Chapman1, H. Chen1, C. Choate1, L. J. Suter1, C. A. Thomas4, R. P. J. Town1, E. R. Tubman1, C. Trosseille1, P. L. Volegov2,
A. R. Christopherson1, D. S. Clark1, J. W. Crippen5, E. L. Dewald1, T. R. Dittrich1, M. J. Edwards1, C. R. Weber1, K. Widmann1, C. Wild9, C. H. Wilde2, B. M. Van Wonterghem1, D. T. Woods1,
W. A. Farmer1, J. E. Field1, D. Fittinghoff1, J. Frenje6, J. Gaffney1, M. Gatu Johnson6, B. N. Woodworth1, M. Yamaguchi5, S. T. Yang1 & G. B. Zimmerman1
S. H. Glenzer7, G. P. Grim1, S. Haan1, K. D. Hahn1, G. N. Hall1, B. A. Hammel1, J. Harte1,
E. Hartouni1, J. E. Heebner1, V. J. Hernandez1, H. Herrmann2, M. C. Herrmann1, D. E. Hinkel1, 1
Lawrence Livermore National Laboratory, Livermore, CA, USA. 2Los Alamos National
D. D. Ho1, J. P. Holder1, W. W. Hsing1, H. Huang5, K. D. Humbird1, N. Izumi1, L. C. Jarrott1, Laboratory, Los Alamos, NM, USA. 3Laboratoire pour l’utilisation des Lasers Intenses chez
J. Jeet1, O. Jones1, G. D. Kerbel1, S. M. Kerr1, S. F. Khan1, J. Kilkenny5, Y. Kim2, H. Geppert École Polytechnique, Palaiseau, France. 4Laboratory for Laser Energetics, University of
Kleinrath2, V. Geppert Kleinrath2, C. Kong5, J. M. Koning1, J. J. Kroll1, M. K. G. Kruse1, Rochester, Rochester, NY, USA. 5General Atomics, San Diego, CA, USA. 6Massachusetts
B. Kustowski1, O. L. Landen1, S. Langer1, D. Larson1, N. C. Lemos1, J. D. Lindl1, T. Ma1, Institute of Technology, Cambridge, MA, USA. 7SLAC National Accelerator Laboratory, Menlo
M. J. MacDonald1, B. J. MacGowan1, A. J. Mackinnon1, S. A. MacLaren1, A. G. MacPhee1, Park, CA, USA. 8Atomic Weapons Establishment, Aldermaston, UK. 9Diamond Materials,
M. M. Marinak1, D. A. Mariscal1, E. V. Marley1, L. Masse1, K. Meaney2, N. B. Meezan1, Freiburg, Germany. 10These authors contributed equally: A. B. Zylstra, O. A. Hurricane.
P. A. Michel1, M. Millot1, J. L. Milovich1, J. D. Moody1, A. S. Moore1, J. W. Morton8, T. Murphy2, ✉e-mail: [email protected]; [email protected]

548 | Nature | Vol 601 | 27 January 2022

Methods
 
Terminology Req = P01 + ∑ δPℓ × Pℓ(cos θ), (4)
Definitions of commonly used mathematical symbols are summarized  ℓ 
in Extended Data Table 2.
 
R po = M01 + ∑ δMm × cos[m(ϕ − ϕi )] , (5)
Reproducibility  m 
Experiments subsequent to those described in this Article have dem-
onstrated the reproducibility of a burning-plasma state at NIF, with ReqRpo
two additional experiments that have performance comparable to Rhs = , (6)
M0
the highest-performing experiments in this Article. These newer
experiments, N210307 and N210605, were conducted in the Hybrid where P0 and M0 are the average measured sizes from each view, δPℓ
E platform. N210307 repeated N210207, albeit using a capsule from a (δMm) is the relative modal amplitude, often referred to as Pℓ/P0 with
different fabrication batch and produced a yield of approximately 145 kJ the nomenclature above chosen for clarity. Pℓ are the Legendre poly-
with an experimentally inferred Qα = 1.34 ± 0.07 from the Hurricane nomials, and ϕm are the orientation of the azimuthal modes. Rhs is the
criterion. Experiment N210605 reduced the thickness of the ice layer hot-spot radius as a function of both θ and ϕ, which is integrated to
relative to N210207 and resulted in a lower yield (135 kJ) but high ion obtain the volume. Here, the dominant parameters are P0, M0 and δP2,
temperature, with Qα = 1.40 ± 0.10, again from the Hurricane criterion. which are given in ref. 7.
These additional experiments confirm that the burning-plasma state Implosion velocity (vimp) is inferred using a rocket model of the implo-
is reproducible at NIF, and full details and analysis on them will be pre- sion46 constrained by both supporting experiments, especially in-flight
sented in future publications. radiography, and the measured time of peak nuclear production on
each experiment. The inferred yield amplification given in Extended
Inferred hot-spot conditions Data Table 1 is a function of the measured yield, shell compression and
Hot-spot conditions must be inferred from measured quantities using fuel mass (mfuel); both the velocity and Yamp inferences use the prescrip-
a model. The simplest hot-spot model is to assume an isobaric volume tion given in a previous work26. The fuel kinetic energy then follows
1 2
of uniform conditions, as used in a previous work14 between equations 2 from 2 mfuelv imp . Our techniques for inferring the PdV work done on
and 3, in which case the hot-spot number density is given by the fuel are discussed in the following section.
A comparison of inferred values using 0D and 1D models are shown in
Y Extended Data Table 3. Inferred pressures are highly consistent between
n = 1.2 × 106 , (2)
σv Vhsτ these calculations, whereas hot-spot energies and areal densities are
higher in the 1D model owing to substantial mass near the 1 keV tem-
where Y is the fusion yield in J, ⟨σν⟩ is the fusion reactivity, which perature cut-off.
depends on the ion temperature (Ti), Vhs is the hot-spot volume in cm3,
and τ is the burn duration in s, for equimolar DT mixtures. The remain- Inferring Gfuel
ing hot-spot quantities follow from the inferred density, including the The total fusion yield produced by a mass (m) of DT, over a characteristic
pressure (P = (1 + Z)nkBTi, with kB Boltzmann’s constant), hot-spot confinement time, τ, is Y ≈ 5mPατ—with Pα = 8.2 × 1024ρ⟨σν⟩ in GJ g−1 s−1 the
energy (Ehs = 1.5PVhs), and areal density (ρRhs = (2.5n /Na) 3 3Vhs /4π ). specific DT fusion power for a given mass density, ρ, of DT with reaction
A more detailed inference is to use a one-dimensional (1D) profile in rate ⟨σν⟩—and the internal energy in that DT is Ehs = cDTmTi. Therefore,
radius for temperature and density, maintaining the isobaric assump- one can write (O.A.H. et al., manuscript in preparation)
tion. A conduction-limited profile follows the expression45:
Y
1
Y E hs
Gfuel = ≈ Efuel q Y
, (7)
  r  2 1+ β EP dV ,tot 1+ − 10
T (r ) = Tmin + (T0 − Tmin) 1 −    , (3) E hs E hs
  R0  
 
with
where Tmin is the temperature at the boundary, T0 is the central tem-
perature and R0 is the hot-spot boundary. β is the thermal conductivity Y σv
≈ 4.6 × 1026 P 2 τ , (8)
power law, 2.5 from classical Spitzer conductivity. Following a previous Ehs T
work26 we use a lower value, β = 2/3, which accounts for additional phys-
ics, dynamical processes and reproduces radiation–hydrodynamics where P is in Gbar, Ti in keV, and τ in s. In equation (7), the total energy
simulations. The density profile is then determined by the isobaric delivered by PdV work, EPdV,tot, is determined from the hot spot and
assumption through P ∝ nT being constant. Tmin is taken as 1 keV leaving compressed, but cold, DT fuel energy at stagnation, Ehs and Efuel, respec-
T0, R0 and P as free parameters in the model; the data are compared to tively, at peak compression. The last term in the denominator repre-
synthetic data calculated from this 1D profile with the model param- sents a correction for additional energy retained by self-heating of the
eters adjusted to minimize residuals. As in the zero-dimensional (0D) fuel from α-particle deposition but not then lost as bremsstrahlung.
model, the hot-spot energy simply follows from pressure and volume, So, EPdV,tot ≈ Ehs + Efuel − qY/10, where q is a ‘quality’ factor, 0 ≤ q ≤ 1, meas-
and the areal density is the mass density integrated over the inferred uring the ability of the implosion to retain self-heating energy (O.A.H.
radial profile. et al., manuscript in preparation). Here we use q ≈ 0.7, inferred from
In either dimensionality the model’s radius is matched to the experi- simulations, and the factor of 10 results from one-fifth of the fusion
mental measurements, which take a contour of emission level, by cal- energy released as α-particles and half of those produced up until the
culating synthetic emission images to calculate an equivalent contour time of peak fusion burn. Albeit generally arrived at in a different fash-
radius. The measurements include 2D and three-dimensional (3D) ion than above, the product P(⟨σν⟩/T2)τ is Lawson’s9 parameter for
asymmetries, so an equivalent spherical volume, and radius, are cal- ignition. Figure 2a uses the useful reaction-rate approximation
culated using the modal decompositions, where the emission contour σv ≈ 4.2 × 10−20T 3.6 i (in units of cm3 s−1 for ion temperature range
measured from the equator (Req) and pole (Rpo) are 3.5 < Ti < 6.5 keV) to simplify the abscissa.
Article
An expression for the fuel gain is given in equation (7). The yield is For analysis of previously published campaigns we use the simple
measured and the hot-spot energy is inferred as described in the pre- relation EPdV,hs ≈ (0.5–0.7)KEfuel(1 − f 2), this is easy to evaluate with the
vious section. Precisely determining the cold-fuel energy from data available data and the factor 0.5–0.7 accounts for a wide range of 1D to
is not straightforward. For the purposes of this analysis we actually 2D/3D behaviour observed on past experiments. For comparison, the
require the total PdV work done on all the DT. This is at a minimum proportionality constant inferred from the first methodology (equa-
the fuel kinetic energy and internal energy at peak velocity, which are tion (10)) is between 0.60 and 0.73 for our four experiments.
both inferred. This neglects any work done by the inflowing remaining We also use radiation–hydrodynamics simulations to estimate
ablator material on the fuel, which can occur in these implosions. In the PdV work done on these implosions. The first simulation-based
this case the hot-spot energy is more than half the previous estimate; methodology is to use 2D simulations with degradation mechanisms
in this scenario we assume equipartition between the hot spot and cold that match the observed performance, and interrogate the work done
fuel to evaluate equation (7). upon the mass elements that form the hot spot to infer EPdV,hs. The simu-
lation methodology is described in ref. 6, and the values of EPdV,hs for
Inferred PdV work this method are given in Extended Data Table 4. The same fusion per-
The primary uncertainty in the Betti metric8 is in the inference of PdV formance can be generated with varying application of degradation
work on the hot spot. Here we use three methodologies: two inferences mechanisms that either degrade EPdV,hs or do not; an estimate of the 2D
using an analytic model, and a direct extraction of PdV work from simu- simulation uncertainty of ±0.5 kJ is estimated by studying multiple
lations that match the experimental observables. simulations.
We use the hydrodynamic piston model of an implosion described A similar energy-balance analysis can be done with 1D simulations, in
previously34. This analytic model abstracts the implosion process using which the work done upon the hot spot is well defined with a Lagrangian
opposed pistons to represent the imploding shell. In spherical geom- mesh. The 1D simulations are tuned to match the measured yields,
etry, the stagnation pressure from this mechanical work on the hot but are expected to underestimate EPdV,hs since they cannot properly
spot is given by (equation 24 in ref. 34): incorporate residual kinetic energy. This estimate is given in Extended
Data Table 4 as an upper bound.
2
ρδRavev imp 2 We have thus develop four methodologies for estimating EPdV,hs.
Ppiston = (1 − f ), (9)
Rhs In the main analysis we use a combination of the empirical piston model
estimate as the more pessimistic data-based inference, and use the
where ρδRave is the average shell areal density, calculated from the 2D simulated EPdV,hs as the most robust computational description of
measured neutron ‘down-scattered ratio’ (DSR) using the relation the experiments.
ρδRave ≈ 19.3DSR, vimp is the implosion velocity and Rhs is the average
hot-spot radius (which can be obtained from the volume, Vhs, given in Modified Hurricane metric
Extended Data Table 1). The factor f 2 represents the effect of mode-1 At peak burn, the time rate of change of hot-spot volume, dV/dt, is nearly
asymmetry and is a measure of the residual kinetic energy (kinetic zero, and therefore so is the heating rate, so time integration is needed.
energy that is never converted into internal energy) in the implosion. Mathematically, a statement of a burning plasma appropriate for ICF is
From the piston pressure we obtain the hot-spot internal energy
t pf t min V
P
(Ehs) from ∫0 Pαdt > − ∫
0 m
dV , (13)
3
Ehs = P V . (10)
2 piston hs where tpf is the time of peak fusion rate, and tminV is the time of minimum
hot-spot volume.
In the absence of α-heating (which adds energy to the hot-spot) and The integrals in equation (13) are easily approximated2 without know-
radiative X-ray losses, or when α-heating exactly balances X-ray losses, ing the details of the actual implosion using the mathematical method
then Ehs = EPdV,hs. For low yield amplification implosions (Yamp < 1.5), X-ray of steepest descent; assuming that the thermodynamic quantities of
losses dominate over α-heating energy gains, so Ehs < EPdV,hs. For higher interest, such as T, P, ρ, and so on, are impulsive, being highly peaked
yield amplification implosions (Yamp > 2), α-heating energy gains start around the time of stagnation. Ultimately, the solution to equation (13),
to dominate over X-ray losses, so Ehs > EPdV,hs. The estimated values for in terms of only burn-average hot-spot areal density, ρRhs, Ti and vimp is
these four experiments are given in Extended Data Table 4 as the piston equation (1) after a correction to the original derivation.
methodology. A recent note from our colleagues at Los Alamos47 discovered an
We can also estimate the stagnated fuel mass in a similar fashion, arithmetic error in the derivation of the criteria as published in ref. 3.
using The error is in going from equation 8 to equation 9 in ref. 3, in which the
conversion to peak temperature (T0) to burn-averaged temperature
2
mshell = 4πRhs ρδRave , (11) (Ths) should be, for n ≈ 4,
n −1
which allows us to then estimate the total mass that stagnates from σv n+ 1 2 σv
mshell + mhs, with mhs from the hot-spot inferences described earlier. ≈ 
T0  n  T Ths
T0
We then estimate the PdV work from (14)
σv
mshell + mhs ≈ 1.40 .
EP dV, hs = 0.73KE fuel , (12) T Ths
mfuel
Additionally, we now believe that the inclusion of the fraction of
where mfuel is the initial fuel mass. The factor of 0.73 is derived from 1D α-particles stopping in the hot spot (fα) in ref. 3 was inappropriate. When
simulations in which the imploding mass stagnates efficiently, and we considering the temperature evolution of a defined mass—for example,
drop the residual kinetic energy factor f 2 because the inferred shell the self-heating criterion in equation (17)—this is necessary because fα
mass does not include non-stagnated material. This estimate leads is fundamentally the fraction of α-particle energy deposited into that
to smaller estimates of EPdV,hs than the first empirical estimate, and are mass. On the other hand, the burning-plasma criteria is one on the
given in Extended Data Table 4 as the stagnated mass estimate. energy of the hot spot,
 becomes an energy loss term. The bremsstrahlung loss can be enhanced
Ehs = cDTmhsThs, (15) beyond the emission of clean DT by the presence of high-Z contamina-
tion of the DT (that is, mix), by a fraction fb. In equation (17), fα is the
and α-particles that escape the hot spot still contribute to its energy fraction of α-particles stopped in the hot spot, evaluated using fits with
via generation of additional hot-spot mass, as seen by examining the modern stopping-power theory36.
time derivative of the above:
Uncertainty analysis
dEhs  dT dmhs  We perform uncertainty analysis for all hot-spot quantities by propa-
= cDTmhs hs + T
dt  dt dt hs  gating the normally distributed uncertainties in measured quantities
(16) through the 0D and 1D models described earlier. The model input
( )
= mhs fα Q α + mhs 1 − fα Q α
parameters are those that fully describe the system, and are constrained
= mhsQ α . by the measured yield, ion temperature, burn widths (from both X-rays
and γ-rays), and volume from the 17% contour of neutron emissivity.
Therefore, the inclusion of fα in a burning-plasma criterion is inap- Distributions of model parameters are generated using Markov chain
propriate. We note that not including an fα factor is consistent with Monte Carlo (MCMC), calculated with the tensorflow49 probability
other criteria, for example, ref. 8. With these two modifications to the package. The log-likelihood function for MCMC is defined by the meas-
criterion published in ref. 3 we use a new criterion (equation (1)). This urements and calculated with the log-likelihood function
modified criterion is slightly more restrictive for the burning-plasma
2
threshold in the regime relevant to these experiments. 1  mi − yi 
− ∑   , (18)
2 i  δyi 
Model uncertainties for Hurricane’s metric
The Hurricane metric3 depends on more quantities than the Betti met- which is summed over all observables (i) where mi is the model value,
ric, although these quantities are more straightforward to infer than yi is the measured value and δyi is the uncertainty in the measurement.
EPdV,hs. The metric reduces to equation (1) where ρRhs and vimp are inferred This methodology produces full distributions of the model param-
as described previously, and Ti is measured. ⟨σν⟩ contains some system- eters including any correlations, from the model parameter distribu-
atic uncertainty from the evaluation used. Data uncertainties are well tions we generate full distributions of all hot-spot parameters, some
defined for Ti and in the inference of ρRhs and vimp, and are propagated of which exhibit correlation, such as in the temperature and areal
as described in the next section; the inferred ρRhs can also vary between density required to evaluate the Hurricane metric, which are partially
models, which will be discussed. anti-correlated (evident in Fig. 3a). Other inferences, such as the implo-
Equation (1) depends on the fusion reactivity; in this work we use the sion velocity or kinetic energy, are treated with normally distributed
evaluation from Bosch and Hale32. Recent publications have presented uncertainties that are uncorrelated with the hot-spot inferences.
alternative evaluations48 which differ by about 2%. We note that the
inferred ρR ∝ 1/ σv from equation (2), so the condition in equation (1) Power-balance relations
depends on the reactivity as 1/ σv . fα is also weakly increasing with In evaluating the power-balance relations relevant to equation (17) we
ρR, leading to the condition being slightly less than square-root depend- use the following expressions for the individual terms:
ent on ⟨σν⟩, so this criterion has <1% uncertainty from the choice of
⟨σν⟩ evaluation. Pα = 8.2 × 1024 ρ σv , (19)
The Hurricane criterion is sensitive to the inferred hot-spot ρR, which
can vary between models depending on the spatial dependence of ρ. As Pb = 3.1 × 107 ρ T , (20)
shown in Extended Data Table 3, the 0D and 1D hot-spot models agree
quite well. We also check these values using a 3D reconstruction of the
T 3.5
hot-spot density and temperature profiles (a yet unpublished method Pe = 5.9 × 103 . (21)
of L. Divol, but briefly described in ref. 35): for N201101 this gives a ρR 2
value of ρRhs ≈ 0.36–0.38 g cm−2 to the 1-keV contour for N201101 and
ρRhs ≈ 0.35–0.36 g cm−2 for N201122. These values are consistent with In these expressions the specific powers are given in units of GJ g−1 s−1
the simple models described earlier. and thus are multiplied by the inferred hot-spot mass to obtain power.
ρ is the hot-spot mass in g cm−3, ⟨σν⟩ is the fusion reactivity evaluated
Self-heating regime as a function of temperature in cm3 s−1, T is the temperature in keV, and
The hot-spot per unit mass power balance is: ρR is the hot-spot areal density in g cm−2. The self-heating power Pα is
multiplied by the fraction of α-particle energy deposited in the hot
dT P dV spot (fα) using the evaluation published in ref. 36; for all four experi-
cDT =f P −f P −P − , (17)
dt α α b b e m d t ments, fα ≈ 0.77–0.80.

which describes the temporal evolution of the temperature (T) in terms

of the balance of self heating (Pα) versus bremsstrahlung (Pb) and elec- Data availability
tron conduction (Pe) losses plus PdV work. Here electron conduction Raw data were generated at the National Ignition Facility. Derived data
losses are calculated relative to a hot-spot boundary that is defined rela- supporting the findings of this study are available from the correspond-
tive to a fraction of the peak burn rate or a specified ion temperature. ing authors upon request.
Thermal conduction cools the hot spot while increasing the mass of
the hot spot. Because the fusion burn rate is more strongly dependent 45. Cerjan, C., Springer, P. T. & Sepke, S. M. Integrated diagnostic analysis of inertial
confinement fusion capsule performance. Phys. Plasmas 20, 056319 (2013).
on the temperature of the spot than its mass in the temperature range 46. Hurricane, O. A. et al. On the importance of minimizing “coast-time” in X-ray driven
achieved by compression alone, α-heating must provide sufficient inertially confined fusion implosions. Phys. Plasmas 24, 092706 (2017).
heating for the hot-spot temperature to increase in the presence of this 47. Albright, B. et al. Comment on the burning plasma condition of Hurricane et. al [Phys.
Plasmas 26, 052704, 2019] and implications for the experimental achievement of a
conduction into an increasing mass. Hot-spot volume change, dV/dt, burning plasma state on the NIF. Report No. LA-UR-21-25149 (Los Alamos National
is negative on implosion, increasing T. During expansion the PdV term Laboratory, 2021).
Article
48. de Souza, R. S., Boston, S. R., Coc, A. & Iliadis, C. Thermonuclear fusion rates for tritium + spectrometer diagnostic; S.H.G. ICF physics; G.P.G. nuclear diagnostics; S.H. capsule physics,
deuterium using Bayesian methods. Phys. Rev. C 99, 014619 (2019). iPOM analysis; K. D. Hahn neutron diagnostics; G.N.H. experiments; B.A.H. capsule physics;
49. Abadi, M. et al. Tensorflow: a system for large-scale machine learning. In Proc. 12th J.H. computational physics; E.H. nuclear time-of-flight diagnostics; J.E.H. MOR and PAM
USENIX Symp. Operating Systems Design and Implementation (OSDI’16) (eds Keeton, K. & stability, SSD improvements, and FC control; V.J.H. MOR and PAM stability, SSD improvements,
Roscoe, T.) 265–283 (USENIX Association, 2016). and FC control; H. Herrman gamma diagnostics; M.C.H. programme management; D.E.H.
hohlraum physics, CBET studies in Hybrid-C; D.D.H. capsule physics; J.P.H. X-ray diagnostics;
W.W.H. management; H. Huang capsule fabrication; K. D. Humbird ensemble simulations; N.I.
Acknowledgements We thank B. Coppi (MIT), S. C. Cowley (PPPL), D. Whyte (MIT), J. Hammer X-ray diagnostics; L.C.J. X-ray diagnostics; J.J. neutron diagnostics; O.J. hohlraum physics;
(LLNL), M. Farrell (GA) and J. Kline (LANL) for discussions. The contributions of NIF operations G.D.K. HYDRA code development; S.M.K. neutron diagnostics; S.F.K. X-ray diagnostics and
and target fabrication teams to the success of these experiments are acknowledged. This work analysis; J.K. diagnostic management; Y.K. gamma diagnostics; H.G.K. gamma diagnostics;
was performed under the auspices of the US Department of Energy by Lawrence Livermore V.G.K. neutron diagnostics; C.K. capsules; J.M.K. HYDRA code development; J.J.K. targets;
National Laboratory under contract DE-AC52-07NA27344. This document was prepared as an M.K.G.K. ICF physics; B.K. ensemble simulations; O.L.L. velocity analysis; S.L. laser plasma
account of work sponsored by an agency of the US government. Neither the US government instability (PF3D) code development; D.L. NIF facility management; N.C.L. optical diagnostics;
nor Lawrence Livermore National Security, LLC, nor any of their employees makes any J.D.L. ICF physics; T. Ma ICF physics; M.J.M. X-ray diagnostics; B.J.M. mode-1 analysis,
warranty, expressed or implied, or assumes any legal liability or responsibility for the accuracy, backscatter; A.J.M. diagnostic management; S.A.M. integrated design physics; A.G.M. X-ray
completeness, or usefulness of any information, apparatus, product, or process disclosed, or diagnostics; M.M.M. HYDRA code development lead; D.A.M. X-ray diagnostics; E.V.M. X-ray
represents that its use would not infringe privately owned rights. Reference herein to any diagnostics; L.M. capsule physics; K.M. gamma diagnostics; N.B.M. advanced hohlraum lead;
specific commercial product, process, or service by trade name, trademark, manufacturer, or P.A.M. LPI physics; M.M. optical diagnostics; J.L.M. hohlraum physics; J.D.M. hohlraum physics;
otherwise does not necessarily constitute or imply its endorsement, recommendation, or A.S.M. neutron diagnostics; J.W.M. hohlraum physics; T. Murphy neutron and gamma
favouring by the US government or Lawrence Livermore National Security, LLC. The views and diagnostics; K.N. project engineering; J.G.D.N. MOR and PAM stability, SSD improvements and
opinions of authors expressed herein do not necessarily state or reflect those of the US FC control; A.N. target fab engineering, capsule and fab planning; R.N. ensembles simulations;
government or Lawrence Livermore National Security, LLC, and shall not be used for M.V.P. HYDRA code development; L.J.P. MOR and PAM stability, SSD improvement and FC
advertising or product endorsement purposes. Document release number: LLNL-JRNL- control; J.L.P. ensembles simulations; Y.P. hohlraum physics; B.B.P. hohlraum physics;
819741-DRAFT. M. Ratledge capsule fabrication; N.G.R. capsule fabrication; H.R. RTNAD mode-1 analysis;
M. Rosen hohlraum physics; M.S.R. X-ray diagnostics; J.D.S. hohlraum physics; J.S. mode-1
Author contributions A.B.Z. hot-spot analysis lead, Hybrid-E experimental lead, wrote sections analysis; S. Schiaffino capsules; D. J. Schlossberg neutron diagnostics; M.B.S. hohlraum
of the paper; O.A.H. capsule scale/burning-plasma strategy, theory, 0D hot-spot models, and diagnostics; C.R.S. HYDRA code development; H.A.S. NLTE opacities (Cretin) code
wrote sections of paper; D.A.C. empirical hohlraum P2 model and hohlraum strategy; A.L.K. development; S.M.S. HYDRA code development; K.S. mode-1 metrology; M.W.S. kinetic
Hybrid-E design lead, integrated hohlraum group lead, wrote sections of paper; J.E.R. N201101 physics; S. Shin sagometer data and particle analysis; V.A.S. capsule physics; B.K.S. ensemble
and N210207 experimentalist and shot-responsible individual (shot RI); H.F.R. original I-Raum simulations; P.T.S. dynamic model, ignition theory; M.S. capsules; S. Stoupin X-ray diagnostics;
design lead; J.S.R. I-Raum experimental lead and N201122 shot RI; C.V.Y. present I-Raum design D. J. Strozzi hohlraum/LPI physics; L.J.S. hohlraum physics; C.A.T. Bigfoot design physics;
lead, wrote sections of paper; K.L.B. Hybrid shot RI; D.T.C. Hybrid shot RI; T.D. Hybrid shot RI; L.D. R.P.J.T. programme management; E.R.T. optical diagnostics; P.L.V. neutron imaging diagnostics;
3D hot-spot analysis; M.H. Hybrid shot RI; S.L.P. Hybrid shot RI; A.P. Hybrid and I-Raum shot RI, C.R.W. capsule/instability physics; K.W. X-ray diagnostics; C.W. capsule fabrication; C.H.W.
physics of capsule engineering defects; P.K.P. 1D hot-spot analysis, Yamp and GLC inference; R.T. neutron diagnostics; B.M.V.W. NIF operations lead; D.T.W. hohlraum physics; B.N.W. project
Hybrid shot RI; S.J.A. capsule microstructure physics; P.A.A. hohlraum physics; L.J.A. engineering; M.Y. capsule fabrication; S.T.Y. MOR and PAM stability, SSD improvements and FC
engineering and targets; B.B. penumbral X-ray diagnostic; D.B. computational physics; L.R.B., C.T. control; G.B.Z. computational physics lead.
X-ray framing camera; L.B.H. HDC design and campaign lead; R.B. ICF physics/ignition theory;
S.D.B. cryo-layering; J.B. capsule fabrication; R.M.B. RTNAD nuclear diagnostic; N.W.B. neutron Competing interests The authors declare no competing interests.
diagnostics; E.J.B. project engineering; D.K.B. diagnostics; T.B. capsule fab and metrology;
T.M.B. cryo-layering; M.W.B. project engineering; P.M.C. DT EOS measurements; B.C. HYDRA Additional information
code development; T.C. LPI physics; H.C. GLEH X-ray diagnostic; C.C. target fab planning; Correspondence and requests for materials should be addressed to A. B. Zylstra or
A.R.C. ignition theory; D.S.C. capsule/instability physics; J.W.C. capsule fabrication; E.L.D. O. A. Hurricane.
experiments; T.R.D. capsule physics; M.J.E. programme management; W.A.F. hohlraum Peer review information Nature thanks Shinsuke Fujioka, Erik Lefebvre and Nigel Woolsey for
physics; J.E.F. 2DConA image analysis; D.F. nuclear diagnostics; J.F. magnetic recoil their contribution to the peer review of this work.
spectrometer nuclear diagnostic; J.G. ensemble simulations; M.G.J. magnetic recoil Reprints and permissions information is available at http://www.nature.com/reprints.
Extended Data Table 1 | Scalar metrics for these experiments

Data, inferred metrics and burning-plasma criteria for these four experiments. Percentages indicate probabilities. The criteria >1 corresponds to a burning plasma, except for Yamp and Gfuel, with
the threshold for those criteria given in brackets. Errors are formal ±1σ standard deviations.
Article
Extended Data Table 2 | Symbols

Definitions for symbols used in this paper.

Extended Data Table 3 | Hot-spot models

Comparison of inferred quantities from 0D and 1D hot-spot models.

Article
Extended Data Table 4 | PdV work methodologies

Inferred hot-spot PdV work (in kJ) by different methodologies, ordered from smallest to larg-
est estimates.