Cyclotron Info

Vacuum & Beam Survival Calculator

A cyclotron beam dies by collision: every residual gas molecule along the spiral is a chance to charge-exchange or scatter out of the machine. This calculator converts a chamber pressure into a mean free path and an estimated surviving beam fraction over your total path length (from the path length calculator), and tells you what pressure keeps losses under 10%.

10⁻⁷10⁻⁶10⁻⁵10⁻⁴10⁻³ 0255075100 beam surviving (%) pressure (torr), log scale 90 % survival 10 % loss at 4.2×10⁻⁵ torr 97.5 % survives
Surviving beam fraction F = exp(−L/λ) over the total path L, versus chamber pressure (log scale). The filled dot is the current operating point; the dashed line is 90 % survival (10 % loss), with the open marker showing the pressure that meets it. Hover for values; exact numbers are in the results table.

Results

Pressure (all units)
Gas number density n
Gas–gas mean free path (kinetic theory)
Beam loss mean free path λ = 1/nσ
Beam surviving the modeled gas losses over L
Pressure for <10% loss over L

The math

From the ideal gas law, the number density of residual gas molecules is

n = p / kBT

A fast beam particle traversing (effectively stationary) gas with a loss cross-section σ has a mean free path λ = 1/nσ, and — in the simple single-collision-loss model used here, where every such collision removes the particle — the fraction surviving a path L is

F = exp(−nσL) = exp(−L/λ)

Requiring F ≥ 0.9 gives the pressure bound p ≤ −ln(0.9)·kBT/(σL). For reference the calculator also shows the ordinary kinetic-theory mean free path of the gas itself, λgas = kBT/(√2 πd2p), using hard-sphere diameters d = 0.274 nm for H₂ and 0.366 nm for air (O'Hanlon App. B.2) — note the √2 applies only to gas–gas collisions, not to the fast beam.

Along the orbit

The cross-section is not constant: for protons in hydrogen it peaks near 10⁻¹⁵ cm² at 7 keV and falls nearly three orders of magnitude by 200 keV, so a beam loses most of its particles in its early turns. Orbit mode replaces the single σ with the sum over turns. With ΔT = ngaps·q·Vp per turn (every crossing at the voltage peak, as in the path-length calculator), the beam completes ⌊TfinalT⌋ whole turns — after k of them it has energy kΔT and the radius of that energy, r(E) = √(2mE)/qB — plus a final partial turn evaluated at its midpoint energy and radius, so the energy used for σ and the energy used for geometry agree at every step. The survival is

F = exp(−n [ Σk σ(kΔT)·2πr(kΔT) + frac·σ(Ē)·2πr(Ē) ])

The second term is the final partial turn: frac is the fractional turn remaining and Ē its midpoint energy. It vanishes when TfinalT is an integer.

Since every term is proportional to n, the optical depth −ln F is linear in pressure (the lost fraction 1 − F is approximately linear only while it is small) and the curve above keeps its shape; what changes is the effective σ, reported as the path-weighted mean. σ(E) is interpolated log-log in the tables below; outside their range it is held at the end values — an unsourced extrapolation, not data, and the result flags it whenever a lookup leaves the tabulated range. For H₂⁺ the tables are tabulated per nucleon, so a 150 keV H₂⁺ ion is looked up at 75 keV/amu.

The cross-sections

Recommended values from the Oak Ridge compilation (Barnett et al. 1990, ORNL-6086 Vol. 1; OSTI full text, retrieved 2026-08-23). Protons lose by electron capture, H⁺ + H₂ → H + H₂⁺ (p. A-28; stated accuracy 10% above 100 eV). H₂⁺ is destroyed by capture and by collisional dissociation together; the total-destruction cross-section (p. G-46; accuracy 20%) is tabulated only to 50 keV/amu. Above that the calculator holds the last value — the measured trend to 50 keV/amu is falling, but whether the held value over- or understates the true loss beyond the table is not established by the source.

Energy (keV/amu)σ, H⁺ + H₂ → H + H₂⁺ (cm²)σ, H₂⁺ + H₂ → total destruction (cm²)
14.26 × 10⁻¹⁶
1.58.12 × 10⁻¹⁶
26.88 × 10⁻¹⁶8.18 × 10⁻¹⁶
49.33 × 10⁻¹⁶8.30 × 10⁻¹⁶
79.57 × 10⁻¹⁶8.15 × 10⁻¹⁶
108.66 × 10⁻¹⁶8.18 × 10⁻¹⁶
158.46 × 10⁻¹⁶
205.79 × 10⁻¹⁶7.75 × 10⁻¹⁶
402.50 × 10⁻¹⁶4.69 × 10⁻¹⁶
503.81 × 10⁻¹⁶
707.78 × 10⁻¹⁷
1002.91 × 10⁻¹⁷
2001.76 × 10⁻¹⁸
4006.09 × 10⁻²⁰
7003.64 × 10⁻²¹
10004.14 × 10⁻²²

Simple mode is exact when handed the right path-weighted σ for the trajectory, and conservative only when handed a true upper bound over it — which "the low-energy end" is not automatically, since the proton cross-section rises from 1 to 7 keV/amu. For the reference 2–150 keV proton spiral the two modes agree closely (its path-weighted σ lands near 10⁻¹⁶ cm²), but that agreement is the trajectory's and does not transfer: a 20 keV, 1 kV proton machine's path-weighted σ is 7.5 × 10⁻¹⁶ cm², and the default 10⁻¹⁶ would permit about seven times too much pressure there. For H₂⁺ the default understates the loss several-fold at any tabletop energy — the destruction cross-section stays within a factor of about two of its peak across the table's range.

Assumptions and limits

  • Simple mode: one constant σ over the whole spiral. Orbit mode: σ(E) per turn from the tables, every gap crossing at peak voltage, orbit radius from the non-relativistic energy formula.
  • Orbit mode's stopping convention: whole turns at their own energy and radius, then one partial turn evaluated at that segment's midpoint energy and radius — an approximation, not a gap-by-gap orbit integration.
  • Every loss collision removes the particle; small-angle scattering that merely dilutes the beam is not counted separately.
  • Uniform pressure and gas composition throughout the chamber — in particular, the pressure inside the dee is taken equal to the chamber pressure. That can be optimistic: dee conductance limits and the source's local gas load usually make the beam-path pressure higher than the gauge reads. The vacuum page quantifies the dee-interior excess.
  • The cross-sections are for hydrogen gas, and the calculator enforces it: orbit mode locks the gas selector to H₂. Air cross-sections differ and no air loss model is offered.
  • The result is survival against the modeled gas-loss processes only — capture for protons, total destruction for H₂⁺. Source efficiency, phase loss, scattering acceptance, centering, and extraction are not modeled.

Worked checks

Simple mode: at 1 × 10⁻⁵ torr of H₂ at 20 °C, n = 3.29 × 10¹⁷ m⁻³. With σ = 10⁻¹⁶ cm² (10⁻²⁰ m²), λ = 304 m; over L = 7.6 m survival is 97.5%. With a low-energy σ = 10⁻¹⁵ cm² and L = 15 m, λ = 30.4 m and survival drops to 61% — the survival arithmetic behind the 10⁻⁵–10⁻⁶ torr operating pressures in the census's record (which pressure a given machine needed also turned on RF holdoff, source load, and how much loss it could accept).

Orbit mode, the site's reference point: protons, B = 0.6 T, 150 keV, 1 kV peak on the dee, two crossings — 75 turns to a 9.33 cm radius, 29.6 m of path. At 1 × 10⁻⁵ torr survival is 89.9%, against 90.7% from simple mode with σ = 10⁻¹⁶ cm². H₂⁺ to 75 keV at the same radius — 37.5 turns, 14.9 m — survives 71.5% at 1 × 10⁻⁵ torr, where simple mode with 10⁻¹⁶ cm² would have said 95.1%. Both values are recomputed against an independent implementation whenever this page's model changes; the last such revision (energy-consistent stopping) is dated in the changelog.

Sources

  • C. F. Barnett et al., Atomic Data for Fusion, Vol. 1: Collisions of H, H₂, He and Li Atoms and Ions with Atoms and Molecules, ORNL-6086/V1, 1990 — H⁺ + H₂ capture p. A-28, H₂⁺ + H₂ total destruction p. G-46. OSTI, retrieved 2026-08-23.
  • J. F. O'Hanlon, A User's Guide to Vacuum Technology, 3rd ed., Wiley, 2003 — kinetic theory; molecular diameters from Appendix B.2 (H₂ 0.274 nm). Library entry.

The narrative behind this calculator — gas load, effective pumping speed, the dee interior, and a worked budget — is the vacuum page; the sourced rules are the design guide’s vacuum domain. What gas scattering does to the beam itself is covered in beam quality.

Educational reference, not an operating procedure: results reflect the stated model and assumptions. Verify anything safety-critical against primary sources, and read the safety fundamentals before applying numbers to real hardware. Last reviewed: · Report a correction.