Text from the interactive explanation

Micro-scale fusion in dense relativistic nanowire array plasmas

Curtis A, Calvi C, Tinsley J, Hollinger R, Kaymak V, Pukhov A, Wang S, Rockwood A, Wang Y, Shlyaptsev VN, Rocca JJ.

One laser pulse. Two pieces of the same fusion-fuel plastic. One of them is a forest of plastic wires a few hundred nanometres thick — and it produces about 500 times more fusion neutrons .

IllustrationNeutron sparks are drawn at the measured ≈500 : 1 ratio (Fig. 4): about 490 per shot from the forest, about one per shot from the flat target. Shapes, colours (deuteron energy, blue → red) and timing are illustrative, informed by the paper’s 3D simulation; the femtosecond events are slowed down about twenty trillion times. Drag the scene to look around.

The world’s largest fusion lasers fire pulses of millions of joules and can shoot a few times per hour at most. This team asked a different question: could a compact laser — firing pulses of about one joule, and able to fire again and again — make a useful burst of fusion if the target, rather than the laser, does the hard work?

The fusion lottery

Fusion happens when two atomic nuclei collide hard enough to merge. The fuel in this experiment is deuterium : hydrogen whose nucleus carries an extra neutron. Both targets are made of CD₂ — ordinary polyethylene plastic built with deuterium in place of hydrogen.

Deuterium nuclei — deuterons — are positively charged, so they push each other away. Only a fast deuteron gets close enough to another to fuse. When it does, about half the time the result is a helium-3 nucleus and a neutron that flies off carrying 2.45 MeV. (An MeV , a mega-electron-volt, is the energy unit for single particles: one million electron-volts.)

So each fast deuteron is a lottery ticket. Fire one into the plastic: it slows down as it shoves past electrons, and somewhere along its path it may — rarely — fuse.

Predict first

A 3 MeV deuteron versus a 0.5 MeV deuteron: how much likelier is the faster one to make a fusion neutron before it stops?

Computed on this page from the standard D–D fusion cross-section (Bosch & Hale, Nucl. Fusion 32, 611, 1992) and an approximate stopping power for cold CD₂ plastic. The authors made the same kind of per-deuteron calculation with ENDF cross-sections and SRIM stopping powers (Methods).

So a target that launches many fast deuterons wins this lottery far more often. The ion spectrometer recorded deuterons up to about 0.5 MeV from the flat target and up to about 3 MeV from the nanowire forest, under the same laser pulses. What does the forest do that a flat surface cannot?

Inside 140 femtoseconds

A femtosecond is a millionth of a billionth of a second; the laser pulse lasts about 60 of them. It is violet light (400 nm), so intense that it shakes electrons to nearly the speed of light — physicists call this relativistic intensity , here about 8×1019 watts per square centimetre. Scrub through a slice of each target to see where the light goes.

Simulated Simulated snapshot at −7 fs. Deuteron energy colour scale in MeV. Fig. 1b–d from Curtis et al., Nat. Commun. 9, 1077 (2018), CC BY 4.0 — cropped.

Why the forest wins twice

Faster deuterons. The simulation finds deuterons up to about 3 MeV just 60 fs after the peak of the pulse — far beyond what the team measured from flat targets under the same conditions.

More of them. Because the heat reaches deep into the wires, a much larger volume of material is heated, so many more deuterons are accelerated.

Both matter for the lottery of Chapter 1. Together they produced the measured ≈500-fold increase in fusion neutrons.

If all the light is absorbed and spread evenly through a layer of depth L, the average energy per particle is about

Eav ≈ (a₀²/2)·mc² · (nc/nav) · Z/(Z+1) · cτ/L

a₀ measures how hard the light field shakes an electron (it grows with the square root of intensity); mc² is the electron’s rest energy; nc/nav compares the “mirror” density for this colour of light with the actual electron density, i.e. how many particles share the energy; Z/(Z+1) splits it between ions and electrons; cτ/L compares the length of the light pulse with the depth it is spread over. With the paper’s numbers (a₀≈3, λ = 400 nm, nav = 6.4×1022 cm−3, Z = 2.67, τ = 60 fs, L ≈ 5 µm) this gives ≈0.6 MeV per particle — and this assumes the light is absorbed throughout the depth, which is exactly what the forest allows.

What the instruments saw

The story above comes from physics and simulation. Here is the measurement. For every CD₂ nanowire target the team also shot flat targets of the same CD₂ material, and nanowire arrays of ordinary (non-deuterated) plastic, with the same laser pulses: 60 fs long, 400 nm, up to 1.65 J, focused to a spot about 2–2.6 µm across.

Three kinds of instruments watched each shot. Tap one on the map to jump to what it recorded.

1 · Ion spectrometer: how fast were the deuterons? Measured

A Thomson parabola spectrometer sits 75 cm from the target along its axis. Magnetic and electric fields bend each ion by an amount that depends on its speed, mass and charge, so deuterons, protons and carbon ions draw separate traces on a detector.

Shots on non-deuterated (CH₂) arrays confirmed that the deuteron trace is not carbon ions with the same charge-to-mass ratio. Plates placed around the target showed the fast deuterons stream mostly along the target’s axis and practically vanish beyond 67° (Supplementary Fig. 3).

Combines the shape of the measured nanowire spectrum (Fig. 2c, digitized, 0.1–3.2 MeV) with the fusion chances from Chapter 1. It isolates the speed effect only: the forest also heats a much larger volume and so launches many more deuterons. The paper does not split its ≈500-fold increase into separate factors.

2 · Neutron stopwatch: were they really fusion neutrons? Measured

Four plastic scintillator detectors, shielded by lead against x-rays, sat at different distances. Each shot starts with an x-ray flash from the hot plasma, which reaches every detector at the speed of light; the neutrons follow. Distance ÷ travel time gives their speed, and speed gives their energy.

Neutrons from D–D fusion must carry 2.45 MeV — a fingerprint. Find the neutron energy whose line best matches the measured arrival times.

Points re-plotted from Fig. 3 by digitizing the published figure (positions approximate); several shots per detector. Times include a fixed instrument delay, so only the slope (speed) matters.

White dot: a neutron at your trial speed. Teal glow: arrival time from the best fit to the data.

3 · The comparison: ≈500 times more neutrons Measured

The scintillators were calibrated against a known neutron source (accuracy ≈25%, similar to the shot-to-shot scatter), and two bubble dosimeters outside the chamber gave neutron numbers that agreed.

Flat-target trace magnified ×10, as in Fig. 4.

Schematic redraw after Fig. 4: the neutron dips are scaled by the measured 492 : 1 ratio; the x-ray flash is a timing marker.

4 · More light, many more neutrons — and why

Squares: measured shots; line: simulation using deuteron spectra from the 3D simulation plus a nuclear-kinetics calculation (the laser spot was assumed 5 µm wide). Both re-plotted from Fig. 5 by digitizing (approximate). The spot size was kept fixed, so more energy means higher intensity. Doubling the energy from 0.8 to 1.6 J multiplies the simulated yield by about 6 — a “superlinear” rise.

Re-plotted from Fig. 6 (digitized; arbitrary units, compare shapes). Dashed: 8×1019 W/cm2. The simulations show that the measured rise in neutron yield is caused by this shift of deuterons to higher energies.

A record for neutrons — not for energy

“Fusion” often brings power plants to mind. This experiment aims at something different: making as many fusion neutrons as possible from each joule of laser light. Neutron production and net energy gain are separate questions — and the energy balance here is nowhere near break-even.

Zoom in on the energy budget of the record shot to find the fusion energy.

Each D–D reaction that makes a neutron releases 3.27 MeV; a roughly equally likely second branch (D + D → tritium + proton) releases 4.03 MeV. So 3.6×106 neutrons mean about 4 µJ of fusion energy in total. The violet pulse itself was made by frequency-doubling an infrared pulse of about 6.5 J, with ≈40% conversion (Methods).

So why do these neutrons matter?

Because of how they arrive. The paper points to a source of fusion neutrons that is:

The neutrons come in an ultrashort burst — useful for time-resolved studies of materials and for ultrafast neutron radiography (seeing inside objects).

They come from a tiny region around the laser focus — a “point source”, which helps keep neutron images sharp.

D–D neutrons all start near 2.45 MeV (“quasi-monoenergetic”), helpful for spectroscopy and for developing neutron and neutrino detectors.

Driven by joule-class lasers that can fire repetitively, unlike the giant facilities that shoot a few times per hour or less.

From the paper’s introduction. The National Ignition Facility figure uses deuterium–tritium fuel, which fuses far more readily than the deuterium–deuterium fuel used here, and 1.9 MJ pulses.

Turning up the light

The simulations look beyond the intensities the experiment could reach. Slide the intensity up and watch where the fusion reactions move.

Putting a tritium-containing layer under the deuterated wires could produce significantly more neutrons: the D–T fusion odds are larger and peak at lower deuteron energies.

Higher intensities should drive a directed stream of fast deuterons into a separate “catcher” target — a “pitcher–catcher” neutron source.

With still bigger pulses, the best neutron output may need a balance between higher intensity and a larger irradiated volume. Modelling well beyond ≈2 J will need extra physics, such as heating of the substrate.

The share bar is qualitative, following the trends the paper reports from its simulations, not computed values.

The target did the hard work

The opening question was whether a compact, joule-class laser could make a useful burst of fusion. Shaping the fuel into a forest of nanowires — instead of building a bigger laser — let the light reach deep into near-solid-density plastic, heat it throughout, and launch deuterons fast enough to win the fusion lottery far more often. The result was about 500 times more fusion neutrons than from flat plastic, and a record number per joule for lasers of this size: a step toward bright, ultrafast neutron flashes from lasers that fit in a university lab.

Deuterons up to ≈3 MeV from nanowires vs ≈0.5 MeV from flat CD₂.

Neutrons of (2.48 ± 0.14) MeV: D–D fusion.

≈500× more neutrons than flat targets (ratio 492).

Up to 3.6×106 neutrons per shot, 2.2×106 per joule.

Yield rises faster than laser energy.

Light penetrates and is absorbed deep in the array; wires explode; gaps fill in <100 fs.

Deuterons up to ≈3 MeV, in a much larger heated volume.

The yield rise with energy comes from deuterons shifting to higher energies.

Fusion moves into the substrate at higher intensity.

Higher intensity for still more D–D fusion.

Thicker CD₂ substrate; tritium-containing layer.

Pitcher–catcher neutron sources.

Ultrafast neutron imaging, spectroscopy and detector development.

Source

Alden Curtis, Chase Calvi, James Tinsley, Reed Hollinger, Vural Kaymak, Alexander Pukhov, Shoujun Wang, Alex Rockwood, Yong Wang, Vyacheslav N. Shlyaptsev, Jorge J. Rocca. “Micro-scale fusion in dense relativistic nanowire array plasmas.” Nature Communications 9, 1077 (2018). doi.org/10.1038/s41467-018-03445-z

This page explains the published version of record, open access under a Creative Commons Attribution 4.0 International License.

Figures 1–6 and Supplementary Figs. 1 and 4 are reproduced from the article and its Supplementary Information under CC BY 4.0; they were cropped, resized and converted to WebP.

Charts re-plot data digitized from Figs. 2c, 3, 5 and 6; positions are approximate.

The 3D hero scene, the femtosecond slice, the neutron race, the oscilloscope sweep and the fusion-location view are our illustrations informed by the paper’s descriptions and simulations; they are not to scale.

Fusion chances use the D–D cross-section parameterization of H.-S. Bosch and G. M. Hale, Nucl. Fusion 32, 611 (1992), with an approximate stopping power for CD₂; the energy ledger uses standard D–D reaction energies (3.27 and 4.03 MeV).