Text from the interactive explanation
Tailored laser wakefield acceleration for decaying particles
Chiara Badiali, Rafael Almeida, Bernardo Malaca, Ricardo Fonseca, Thales Silva, J. R. Vieira
Original paperPhysical Review Research 8, 023077 (2026) Badiali, Almeida, Malaca, Fonseca, Silva & Vieira
A laser focus that travels slower than light drags a plasma wake behind it. Shape the plasma density so the wake keeps a muon riding the same bucket from 0.94c to near light speed.
1 Density falls: the wake slows down to catch the muon
2 Density rises: the wake speeds up with the muon
3 The bucket reaches the laser focus: the stage ends
Upper half: plasma electrons (bright sheets = piled-up electrons)
Lower half: field that accelerates (teal) or brakes (amber)
Longitudinal field Ez along the central axis: the bunch sits where Ez is negative (below the axis line), which pushes these negatively charged particles forward
Muon bunch
Laser focus, 0.96c
Plasma along the stage
density n/n0 (log)
muon γ
Eq. (1) taper
laser focus
Density from the paper’s matching condition, Eq. (1): a gentle dip lets the wake slow down to catch the muon, then a rise speeds the wake up as the muon accelerates.
Whole run
Final γ ≈ 5.2 after 3.8 mm · stayed in bucket 1
Taper 3.82 mm
36 of 36 bunch particles kept in bucket 1
The tailored density is designed for one particle: the one at the centre of the bucket, starting at exactly 0.94c. A particle that starts a little slower, or a little off-centre, meets a bucket that moves at slightly the wrong speed for it. As the paper notes, a later bucket reacts more strongly to a given density change, so these small mismatches grow fastest in buckets 3–5, in the steep rise at the end of the stage. Heavy kaons speed up slowly, so a mismatch in speed lasts longer and a few of them still drop out.
The paper’s Eq. (1) also makes one simplification: when it turns time into distance, it uses the focus speed instead of the particle’s own speed. In this page’s model, that alone made the centre particle slip back one bucket at the very end in buckets 4 and 5. The scene therefore solves the same matching condition with the particle’s own speed. The profile comes out about 1% longer and the final energy is unchanged, and the centre particle now stays locked in every bucket.
The bunch drawn here is short and has almost no spread in speed. In one of the paper’s full simulations, a short 80 pC bunch was trapped completely at optimal beam loading.
Scene and track: the paper’s Eq. (1) matching condition, solved with the particle’s own speed (see “Why can particles fall out of the bucket?”), with the wake and the bunch computed live in a simplified one-dimensional plasma model (see “What this page computes”). The two-dimensional look of the channel is illustrative. The run stops where the particle’s bucket has been squeezed up against the laser focus, the limit the paper describes as the beam overtaking the driver.
In this story:
Born slow, dying fast
A focus slower than light
Stretch the wave
The recipe
Full simulations
How many joules?
Outlook, cooling & sources
1
Born slow, dying fast
Muons live 2.2 microseconds in their own rest frame. Charged pions live 26 nanoseconds and charged kaons 12 nanoseconds. Because these particles are heavy, they are often created moving slower than light, so they need to be accelerated quickly, before too many of them decay.
The paper estimates the surviving fraction as S ≃ (γi/γf)ν with ν = m c ⁄ (q Eacc τ). In words: the survivors are the ratio of starting to final energy, raised to a power that grows with the particle’s mass and shrinks with the accelerating field and the lifetime. A stronger field means fewer decays.
For a μ⁻ bunch taken from 200 MeV to 1 GeV, the authors quote a loss of about one particle in a million in a 0.2 GV/cm plasma stage, against 1% in a 20 MV/m radio-frequency linac. For π⁻ the contrast is starker: about 10−3 in plasma versus 94% in the linac.
Kaons are the hardest case: they are 4.7 times heavier than muons and have the shortest lifetime of the three. The paper does not quote a kaon figure for this example, so this page applies the same formula with τ = 12 ns, reading 200 MeV and 1 GeV as kinetic energies. In the linac, only about 1 kaon in 200 would survive. In the plasma stage, about 99.5% would survive.
Radio-frequency linac · 20 MV/m · about 40 m
Plasma stage · 0.2 GV/cm · about 4 cm
rf linac: 99 of 100 muons survive · plasma: 99.9999 of 100
Both lanes are drawn at the same width; the real stages differ in length a thousandfold. Muon and pion decay fractions are the paper’s quoted estimates; kaon fractions are this page’s estimate from the paper’s formula. Sparks cluster near the start, where the particles are slowest and their clocks run fastest.
2
A laser focus slower than light
A plasma wake travels with whatever drives it, and an ordinary laser pulse drives it at nearly the speed of light, too fast for a particle starting at 0.94c. The paper uses a space-time wave packet, or “flying focus”: a pulse whose bright intensity peak, about 20 µm long, travels at a chosen speed vf below c, inside a much longer envelope (about 500 µm in Fig. 1; 275 µm in the simulations) that moves at about c.
The peak therefore slips backward through its envelope. It keeps driving a wake for a time t ≃ L ⁄ (c − vf), covering a distance l ≈ L vf ⁄ (c − vf). Once the peak reaches the tail of the envelope, it no longer drives an efficient wake.
Pulse energy grows with the envelope length, as in the paper’s Eq. (2): ε[J] ≈ 0.1 a02 w02 λ−2 L (w0, λ in µm; L in mm).
Where the focus runs out of envelope
How to read it: the shaded band is the envelope, with its front on the right. Moving up means moving forward in the lab. The focus (red) drifts back through the band, so its path is a steep diagonal. The faster the focus, the steeper the path, and the wider the envelope, the longer the path. The wake (pale streaks, one plasma wavelength ≈ 16 µm apart) follows the focus until it reaches the tail. The top strip shows the pulse at the current moment. Geometry follows l = L vf ⁄ (c − vf); the carrier wave is drawn schematically. Energy assumes a0 = 0.8, w0 = 3.6 µm and λ = 0.8 µm, the paper’s parameters for its laser-energy estimate.
3
Stretch the wave, slow the wave
A slow driver is not enough. As the particle gains speed it eventually overtakes the focus and drops out of the accelerating phase; this is dephasing. The paper’s second idea is to control where the buckets sit by shaping the plasma.
The wake’s wavelength scales as λp ∝ 1 ⁄ √n, and the nw-th bucket sits about nw wavelengths behind the focus. Where the density falls, λp stretches as the focus advances, so the bucket drifts backward: it moves slower than the focus and can catch a slow beam. Where the density rises, λp shrinks and the bucket moves forward, faster than the focus, keeping pace with a particle that has sped up.
The paper’s Eq. (A5) gives the bucket speed, vw = vf − nw vf dλp/dz. Buckets farther back react more strongly to the same density change.
Ten buckets behind a 0.96c focus
Flat density: every bucket moves at 0.960c, the focus speed.
Speeds from Eq. (A5) with |dλp/dz| up to 0.006. The drift in the drawing is exaggerated so it is visible within one crossing.
4
The recipe: dip, then rise
Requiring the bucket to move exactly with the particle gives the paper’s Eq. (1), two coupled equations: the field, proportional to √n, pushes the particle, d(γv)/dz = α0 (q/e)(me/mp)(c/vf) ωp0 √(n/n0); and the bucket’s speed must equal the particle’s, v = vf − (2π nw vf2/ωp0) d/dz √(n0/n). Solving them gives the ideal taper. It begins with a gentle down-ramp that traps the beam. The slope can stay gentle because the driver is only slightly faster than the beam, which mitigates, though does not prevent, electron self-injection. Then comes an up-ramp that prevents dephasing as the beam passes the driver speed. Acceleration ends when the beam reaches the driver. Two overlapping Gaussian gas jets, or linear ramps from shock-front targets, can approximate such profiles.
Press Fig. 3(b) check to solve Eq. (1) for buckets 1–7 with the paper’s Fig. 3(b) parameters and compare with its theory points.
Computed live from the paper’s Eq. (1) with n0 = 4 × 1018 cm−3 (c/ωp0 = 2.55 µm). Plasma density is capped at 20 n0, where the beam has reached the driver. α0 is the field in units of mecωp0/e (α0 = 0.36 ≈ 0.7 GV/cm). Do not read this model for injection speeds far below the focus speed; for those regimes see Fig. 3(a) and Fig. 4.
5
Does it survive a full simulation?
The authors tested the model with OSIRIS, a fully relativistic particle-in-cell code, simulating the laser, the plasma electrons and the witness beam self-consistently. These results are simulations, not experiments.
In the waterfall of Fig. 2(b), time runs upward and the horizontal axis follows a window moving at c. Each dark streak is a sheet of piled-up electrons. The streaks bend : they lean back in the down-ramp and forward in the up-ramp, while the laser focus keeps a straight 0.96c line. The green muon trace stays in its bucket throughout. Theory and simulation agree on the final energy for every mass tested (Fig. 2c), for injection from 0.5c, 0.7c and 0.9c (Fig. 3a), and for buckets 1–7 (Fig. 3b).
The same picture can be built from the run in the instrument above, using the simplified one-dimensional model. Switch the instrument between Tailored and Uniform and this plot follows it.
Waterfall of the instrument’s current run 1D model
Horizontal: position relative to a window moving at c (ticks in mm). Vertical: lab time, increasing upward. Magenta: laser focus at 0.96c. Green: reference muon.
100%
of an 80 pC witness beam trapped at the optimal beam loading, Λ = (nb/n0)(kpσr)2 = 0.13 (mμ/4 beam from 0.9c, 0.97c driver)
all ≥ 0.97c
Every trapped particle reaches at least 0.97c.
76% ≥ 0.98c
About three in four pass 0.98c.
50% > 0.99c
Half exceed 0.99c, fast enough to hand over to a second plasma or radio-frequency stage.
Beam-loading results from the paper’s OSIRIS simulations. The authors compared negligible (Λ = 10−5), optimal (0.13) and over-loaded (1.3) beams.
6
Heavier particles, how many joules?
Extending Eq. (1), the model predicts that muons and pions from 0.2c, kaons from 0.4c and protons from 0.5c, driven by a 0.7c focus, reach about 0.9c within tens of millimetres (Fig. 4, model only).
To estimate the laser needed, the paper combines the pulse energy of Eq. (2) with the energy gained over the slip distance, giving Eq. (3): ε[J] ≈ (a02/20) Δγ (mp/me)(c/vf − 1)(w0/λ)2 ⁄ Eacc[GV/m]. For Δγ = 5 the authors find 1, 1.3, 4.9 and 9.2 J for muons, pions, kaons and protons, within reach of existing high-intensity lasers.
Laser energy estimate, Eq. (3)
≈ 1.0 Jpaper’s value: 1 J
Fixed: a0 = 0.8, w0 = 3.6 µm, λ = 0.8 µm (joules). Small differences from the paper’s numbers come from rounding of the particle masses.
7
Where this could lead
The authors suggest the scheme could serve as a rapid booster stage for decaying particles in muon-collider designs , or as a compact source of high-energy pions, kaons and muons for neutrino physics . Because it runs on few-joule drivers, which have already been demonstrated in spatiotemporal optics experiments, plasma acceleration of decaying particles is within reach. They also note it might improve electron acceleration in plasma wakefields and help with isotope separation. In the linear regime, the model extends to positive particles such as antimuons and antipions.
Once a beam is relativistic, conventional wakefield drivers can take it further. This stage does the hard first step: catching particles that are born slow.
What this page computes
Tapers are solved in your browser from the paper’s matching condition, using its Fig. 2 setup: injection at 0.94c behind a 0.96c focus. The recipe explorer uses Eq. (1) exactly as printed. The instrument keeps the particle’s own speed where Eq. (1) uses the focus speed, which makes the taper about 1% longer and leaves the final energy unchanged. Its bunch is short, with a phase spread of ±0.2 rad and a speed spread of ±0.0001c.
The wake and bunch in the instrument use a one-dimensional local-oscillator plasma model: each plasma slice rings at its own local plasma frequency once the focus passes. The field amplitude α0 = 0.36 was chosen so the muon taper length resembles Fig. 2(a). The model ignores electron self-injection, beam loading, the evolution of the laser and all transverse dynamics. The two-dimensional look of the channel and the laser carrier spacing are illustrative.
Simulation results (OSIRIS particle-in-cell, beam loading, trapping and Figs. 1–4) are quoted from the paper.
Survival numbers in the decay race are the paper’s quoted estimates for muons and pions; the kaon numbers are estimated on this page with the paper’s survival formula. Laser energies use the paper’s Eqs. (2) and (3).
Cool first, then hurry
The paper opens with the muon collider’s central problem: muons “must be collected, cooled, and rapidly accelerated before a significant number of them decay.” Muons are born as a diffuse, “hot” cloud flying at many angles. Ionization cooling (demonstrated by the MICE collaboration, the paper’s reference [5]) squeezes it into a narrow, parallel beam:
Absorber: the muons lose momentum in every direction, sideways included.
RF cavity: momentum is given back only in the forward direction.
Repeat: the sideways spread shrinks cell after cell, and the beam narrows.
The cooled beam is still slow and still decaying, so it must then be accelerated fast. That is where the paper places its plasma stage, as a rapid booster. For muons going from 200 MeV to 1 GeV, it estimates a loss of about 1% in a 20 MV/m conventional accelerator, against about one in a million with a 0.2 GV/cm plasma field.
Ionization cooling, step by step illustrative
A schematic with exaggerated numbers that shows the idea only. It is background to the paper, not one of its results.
Source
Chiara Badiali, Rafael Almeida, Bernardo Malaca, Ricardo Fonseca, Thales Silva and Jorge Vieira, “Tailored laser wakefield acceleration for decaying particles,” Physical Review Research 8 , 023077 (2026). doi.org/10.1103/1pp8-qx35 · full text (PDF)
Earlier version of the same work: preprint arXiv:2506.23668.
Figures 1–4 are reproduced from the published article under the Creative Commons Attribution 4.0 license. They were converted to WebP and resized, with no other changes. The article is published by the American Physical Society under the same license.
Topics in this iPaper
Physics
We introduce a plasma wakefield acceleration scheme capable of boosting initially subrelativistic particles to relativistic velocities within millimeter-scale distances.
Abstract, page 1
laser wakefield acceleration
we propose a plasma wakefield acceleration method working in a copropagating geometry.
Title and Section I, pages 1-2
plasma physics
A subluminal light pulse drives a wake whose velocity is continuously matched to the beam speed through a tailored plasma density
Abstract, page 1
particle in cell simulation
We validate our theoretical predictions with particle-in-cell (PIC) simulations with the code OSIRIS
Section I, page 2
muon acceleration
can accelerate initially subrelativistic decaying particles (e.g., muons and pions) to relativistic energies over remarkably short distances.
Section IV, page 4
spatiotemporal pulse shaping
Spatiotemporal laser shaping, particularly flying-focus pulses with programmable focal velocity, enables precise control over the trajectory of the laser intensity peak.
Section I, page 1