Wien-Filter Hamiltonian and Transfer Matrix

The paper derives the Hamiltonian and the six-dimensional transfer matrix for an ideal Wien filter with a vertical magnetic field, starting from first principles. The authors define an ideal Wien filter as one with hard-edged electric and magnetic fields and no misalignments or imperfections. They use the variables x, x′, y, y′, z, and δ = ∆p/p to describe the full six-dimensional phase space, complementing earlier work that derived transfer matrices in four and five dimensions.

The analysis begins with the general Hamiltonian from a referenced formalism, assuming the paraxial approximation with small angles. For a Wien filter, the trajectory is straight (1/ρ = 0), and the electric and magnetic potentials are given by Φ = Ex and As = −B0 x, where B0 is the vertical magnetic field and E is the horizontal electric field. After inserting these potentials and expanding the root to second order in the dynamical variables, the Hamiltonian simplifies considerably when the Wien condition B0 = −E/β0c is fulfilled. The resulting Hamiltonian is:

H(x, x′, y, y′, z, δ; s) = (1/2)x′2 + (1/2)y′2 + δ2/(2γ02) + x2/(2R2) + xδ/(γ0R)

where R = −γ0β0cp0/eE = γ0(Bρ)/B0. The first three terms describe a drift space, the term proportional to x2 describes horizontal focusing, and the term proportional to xδ describes the energy-dispersive properties of the Wien filter.

From this Hamiltonian, the equations of motion are derived. In the horizontal plane, the solutions are:

x(s) = x1 cos(s/R) + x′1 R sin(s/R) − (R/γ0)(1 − cos(s/R))δ

x′(s) = −(x1/R) sin(s/R) + x′1 cos(s/R) − (δ/γ0) sin(s/R)

In the vertical plane, the motion is simply:

y(s) = y1 + y1′s

y′(s) = y1′

For the longitudinal plane, δ(s) = δ1, and integrating the equation for dz/ds gives:

z(s) = (x1/γ0) sin(s/R) + (R x′1/γ0) cos(s/R) + (R/γ02) sin(s/R)δ1

Using these solutions with the Wien filter length L and the abbreviation ϕ = L/R, the authors construct the six-dimensional transfer matrix. The matrix is presented in Equation 14, with the top-left 4×4 part resembling that of a horizontal sector dipole, but with R differing from the bending radius by a factor of γ0. The matrix elements include:

• R11 = cos ϕ, R12 = R sin ϕ, R16 = −(R/γ0)(1 − cos ϕ)

• R21 = −(1/R) sin ϕ, R22 = cos ϕ, R26 = −(1/γ0) sin ϕ

• R33 = 1, R34 = L

• R51 = (1/γ0) sin ϕ, R52 = (R/γ0)(1 − cos ϕ), R56 = (R/γ02) sin ϕ

• R66 = 1

The authors note that the matrix elements agree with those calculated in a previous reference. They then relate the beam optics to the spin-rotation effect, showing that the spin-rotation angle Θ = B0L/(γ0(Bρ)) = L/R = ϕ. This means the transfer matrix is already parametrized by the spin-rotation angle.

For a practical example, the authors consider 200 kV electrons in the CEBAF injector with γ0 ≈ 1.39 and (Bρ) = 1.65×10−3 Tm. A Wien filter of length L = 0.43 m requiring a 90° spin rotation needs R ≈ 0.27 m and B0 = 8.4×10−3 T. The approximate focal length from the R12 matrix element is f ≈ 0.17 m, though this formula is used outside its range of validity since L > R. In June 2025, a measured focal length of fy = 0.78 m for a 40° spin rotation was reported, while the equations estimate f = 0.88 m, giving confidence in the model.

The authors highlight an important consequence: flipping the sign of the polarization by reversing the field polarity also reverses the sign of R. The top-left 4×4 part of the transfer matrix is unaffected, leaving beta functions unchanged, but the dispersive contribution (R16) changes sign. For a 90° spin rotation, the dispersion magnitude is R16 = R/γ0 ≈ 0.2 m, which is likely non-negligible. Additionally, the R51 matrix element couples the horizontal orbit into the longitudinal plane: a change of ∆x1 = 1 mm causes ∆z2 = ∆x1/γ0 ≈ 0.72 mm, which translates to a phase change of about 1.3° for an RF wavelength of 0.2 m. This is substantial and must be considered in high-precision experiments like MOLLER.

The conclusion states that the transfer matrix is already parametrized by the spin-rotation angle, reversing the angle leaves beta functions unaffected but flips the sign of dispersion, and the arrival time in cavities is affected via R51 and R52 unless the beam is well-centered. For a horizontal magnetic field, the matrix can be obtained by sandwiching the vertical-field matrix between two 90° coordinate rotations. The authors emphasize that fringe fields are not considered and refer to numerical methods for such effects.

Improvements to AI Systems Based on This Paper:

1. Analytical Beam-Optics Solver

• Implement the derived 6D transfer matrix (Eq. 14) as a closed-form module for Wien filter simulation, eliminating the need for numerical integration in paraxial regimes.

• Enable fast parameter sweeps (e.g., field strength, length, particle energy) for design optimization in accelerators.

2. Spin–Orbit Coupling Prediction

• Integrate the relation Θ = L/R (spin-rotation angle = focusing parameter) to automatically compute spin precession alongside orbital dynamics, enabling simultaneous spin–beam matching in injector design.

3. Dispersion and Coupling-Aware Control

• Use R16, R51, R52 elements to predict and correct:

• Energy-dependent position shifts (dispersion) for polarized beams.

• Time-of-arrival jitter at RF cavities caused by horizontal orbit offsets (via R51), enabling feed-forward stabilization for high-precision experiments (e.g., MOLLER).

4. Polarity-Reversal Impact Analyzer

• Automatically compute the effect of reversing field polarity (flipping sign of R):

• Beta functions unchanged (top-left 4×4 block invariant).

• Dispersion sign flips—flag potential beam-size or emittance growth in downstream optics.

5. Coordinate-Transformation Adapter

• For horizontal-field Wien filters, apply the paper’s sandwiching method (two 90° rotations around the vertical-field matrix) to generate the correct 6D matrix, extending the tool to arbitrary field orientations.

6. Fringe-Field Warning System

• Detect when the hard-edge assumption (L >> fringe-field extent) is violated, and automatically switch to numerical solvers (e.g., GPTrack or COSY INFINITY) while flagging the model’s validity limits.

7. Experimental Validation Assistant

• Compare measured focal lengths (e.g., fy = 0.78 m for 40° rotation) against the analytic prediction (f ≈ 0.88 m) to provide uncertainty estimates and suggest calibration corrections.

8. Coupled Longitudinal–Transverse Phase-Space Visualizer

• Generate 6D phase-space plots showing how x, x′, y, y′, z, δ evolve through the Wien filter, highlighting the x–δ and x–z couplings for diagnostic or teaching purposes.

9. Automatic Hamiltonian Reduction

• Reuse the paper’s derivation steps (paraxial expansion, Wien condition) to symbolically simplify Hamiltonians for other straight-axis electromagnetic elements, accelerating future analytic modeling.

10. RF Phase-Stability Optimizer

• Use the R51/R52 elements to compute allowable horizontal orbit tolerances for a given RF phase error budget, directly informing alignment specifications in accelerator commissioning.

We derive the Hamiltonian and the transfer matrix of a Wien-filter with vertical magnetic field from first principles.