TRI-BN-25-14: Electrostatic toroidal bender and its fringe fields
Rick Baartman, TRIUMF · 2025
Read TRI-BN-25-14: Electrostatic toroidal bender and its fringe fields by Rick Baartman, TRIUMF (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-21365.txt.
Introduction
Rick Baartman of TRIUMF writes this 2025 technical note on a device that bends charged particle beams using electric fields. The device is shaped like a section of a doughnut, and stray fields at its edges shape how the beam passes through. For about thirty years his group has used its own software, built within the COSY Infinity system, to chart beams through such bends to fine detail.
A puzzle runs through the note. COSY's built-in tools for the same bends gave answers that disagreed with the group's own code, and no one had settled which was right. Baartman lays out the case to clarify the issue for others who design beam lines. The note is a short, candid record of a long-standing mismatch between two trusted tools.
Catalog record
| CREATOR | Rick Baartman, TRIUMF |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 28 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-21365
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
TRI-BN-25-14: Electrostatic toroidal bender and its fringe fields Rick Baartman, TRIUMF Abstract I describe the COSY code GES that we have used in various forms in the past 30 years to calculate maps to third order through electrostatic bend elements. It has been a mystery that COSY ’s in-built procedures ES , ESP and ECL disagreed with our own code. This note is intended to clarify the issue. 1 Introduction A long time ago (the 90s), I discovered two things concerning fringe fields of optical elements. The first is that the nonlinear effects of ‘fringe fields’ are intrinsic to the elements. That is, they are not reducible by modifying element boundaries. So it was a surprise to me that codes such as GIOS [ 1 ] and COSY [ 2 ] (and many others) allow higher order (beyond linear) calculations with fringe fields omitted. The cost is that non-specialists tasked with designing (for example) a beamline, design without these effects at the start, thinking that they can deal with their complicating effects later, or not at all. Further, that the elements themselves are somehow at fault, and that there is a way of shaping elements to reduce these effects. The second discovery was that the lowest order irreducible effect of the fringe field is independent of the shaping of the fringe field falloff. This is a general result and applies to all beam optical elements [ 3 ] . For solenoids and einzel lenses, this effect is in first order so is correctly handled by all transport codes. But in dipoles, the effect is in second order and for quadrupoles in third order [ 4 , 5 ] . I suggest that it would be valuable if the major transport codes include the fringe field effects by default. These maps are simple and would be easy to add. Then the motion would always obey all conservation laws, even with the details of the fringe fields turned off. This is the way in which TRANSOPTR [ 6 ] operates, but it is only a linear code, so the effects are used only to estimate the rms emittance growth, to be used in optimization calculations. In COSY , such a second order map of a fringe field would use negligible computing time as compared with Runge-Kutta integrating through an Enge function, but would still give most of the effect needed when for example designing the optics of a transfer line matching section. In this paper, the theory applied to electrostatic deflectors is recapped, and used to derive the second order ‘hard-edge’ transfer map. A new COSY routine named GES was written and is described. It agrees with the developed theory (and with GIOS ). COSY ’s built in ESP, ECL elements agree neither with theory, GES nor GIOS . 2 Theory To find the transfer map of a fringe field, one must find a canonical transformation that eliminates the terms in the expanded Hamiltonian that become singular in the limit of hard-edge fringe fields. I have done this for quadrupoles and in the particular case of electrostatic bends [ 7 ] . But the hard-edge limit is not given explicitly in that reference so will be amplified here. But first let us recap the theory for the toroidal bend and how the potential function was derived. 2.1 Toroidal coordinates Inside the toroidal bend, there are curvatures in both the s (longitudinal) and the y (off-plane, usually vertical) directions, and these are constant. But in the fringe field regions at entry to and exit from the bend, the former varies from zero to its body constant value, while the y curvature remains constant. The potential is found in a coordinate system corresponding to these curvatures, as an expansion with respect to the reference trajectory, assumed to be at ground potential. In this system, the potential has no y -dependence, and is given to third order 1 1 1 It can be expanded to any order, and the fourth order term is used in the code shown in the Appendix, but not used in the analysis below. in [ 7 , eq. 3] . V_{\rm T}(x,s)=hx-h(k+h)\frac{x^{2}}{2}+[2h(k^{2}+kh+h^{2})-h^{\prime\prime}]\frac{x^{3}}{6} (1) h=h(s)=1/\rho(s) , and k=1/\rho_{y}:=c/\hat{\rho} , a constant: c=1 for spherical electrodes, c=0 for cylindrical ones. Note that we assume no y -dependence in toroidal coordinates, even in the fringe field region where h is changing. 2.2 Frenet-Serret Coordinates Switching to the Frenet-Serret coordinate system, where the y direction has no curvature, we know that there has to be y -dependence, and that only even powers of y occur. At the same time, the x -dependence for y=0 must remain as in the above equation, i.e. V(x,0,s)=V_{\rm T}(x,s) . Writing the potential in Berz’s [ 2 ] notation, V(x,y,s)=\sum_{i}\sum_{j}a_{ij}(s)\frac{x^{i}y^{j}}{i!j!}, (2) we have (remember: h is a function of s but k , the vertical curvature, is constant): \displaystyle a_{10} \displaystyle= \displaystyle h (3) \displaystyle a_{20} \displaystyle= \displaystyle-h(h+k) (4) \displaystyle a_{30} \displaystyle= \displaystyle 2h(h^{2}+hk+k^{2})-h^{\prime\prime}. (5) Inserting into the Laplace equation with curvature in the xs -plane that varies only in the s -direction, we can find the remaining coefficients. More simply, we can insert into Berz’s recursion relation [ 8 , eq. 3.19] 2 2 2 Note that the earlier paper [ 2 , eq. 23] has a misprint. to find that a_{20}+a_{02}+ha_{10}=0 and a_{12}+a_{30}+ha_{20}+h^{2}a_{10}+a_{10}^{\prime\prime}=0 , giving: \displaystyle a_{02} \displaystyle= \displaystyle hk (6) \displaystyle a_{12} \displaystyle= \displaystyle-h(h+k) (7) In the case of using COSY , however, we need not work these out since given V_{\rm T}(x,s) in the plane, it can back out the y -dependence using Laplace equation. This is done inside the routine POFF . 2.3 Fringe field effect in the hard edge limit The second order in optics (third in potential) term in the series expanded Hamiltonian that becomes singular in the limit of a hard edge fringe field, is h^{\prime\prime}x^{3}/6 . This results in a shift in P_{x}^{\prime} of h^{\prime\prime}x^{2}/2 . Integrating by parts twice for entry into the field where h goes from zero to 1/\hat{\rho} , we find \Delta P_{x}=-hxx^{\prime}=-\frac{xP_{x}}{\hat{\rho}}. (8) But if P_{x} is to change discontinuously, x must also change. We can find this from the canonical transformation [ 7 , eq. 20] , since we know that while x and P_{x} suffer shifts in the hard edge limit, the transformed coordinates X and P_{X} do not. From outside where h=h^{\prime}=0 to inside the field where h=1/\hat{\rho} , h^{\prime}=0 , we recover the change above in P_{x} , but also \Delta x=\frac{x^{2}}{2\hat{\rho}}. (9) These two equations (plus of course the first order identity matrix) are the second order map for entry into the bender field. For exit, the signs are reversed. Note that since the singular coefficient of the x^{3} term, h^{\prime\prime} , depends only upon the electrode curvature in the bend plane, this second order shift applies to all toroidal cases; cylindrical and spherical bends and anything between. 3 GES : new COSY code GES , the general electrostatic bender element was originally written (with the help of Dobrin Kaltchev in 1995) as a straightforward application of the general element GE , in which the curvature h and the median plane electric potential V are given on a grid of