<<

Radiation Damping by Thomson Scattering

Nickolai Muchnoi∗ Budker Institute of Nuclear Physics of Siberian Branch Russian Academy of Sciences and Novosibirsk State University

Synchrotron radiation of relativistic in storage rings naturally leads to the damping of betatron oscillations. Damping time and transverse beam emittance can be reduced by wigglers or while the beam parameters are still well defined by the common radiation integrals, based on the properties of radiation. However the quantum excitation of betatron oscillations in principle can be considerably reduced if radiation occurs due to the Thomson scattering in the periodic electromagnetic field. After a brief introduction we compare radiation properties for different cases and suggest the corresponding modification of the radiation integrals.

I. INTRODUCTION

Radiation of relativistic electrons from the point of view of quantum electrodynamics occurs due to their θ scattering on the real or virtual photons. After intro- ε ducing the basic properties of the Compton effect we come to the equivalence of the Thomson scattering and photon ω = hν ε = γmc2 0 0 the radiation of electrons in a weak harmonic . 0 The scattering probability is defined as a product of the Thomson cross section and the density of photons along the electron propagation trajectory. Consequently the θ changes in average energy, energy spread and the trans- ω verse emittance of the electron beam are evaluated as the product of the number of scattering acts and the aver- age change of the corresponding parameter in a single scattering. Within this approach we derive the radiation damping effect from undulator radiation and compare the FIG. 1. Inverse Compton scattering: the thickness of the results with those obtained by the application common arrows qualitatively represents the energies of the particles. radiation integrals [1] for the case of a wiggler/undulator. This comparison shows that the common radiation inte- grals only include the dispersion mechanism for excita- For our case, after Lorentz transformations we have: tion of betatron and synchrotron oscillations. At present dσ r2  u2 uhu i  the radiation damping by wigglers is a well established = e 2+ +8 − 1 sin2ϕ , (1) technique [2, 3] but also there were several suggestions du dϕ κ(1 + u)2 1 + u κ κ to use a laser backscattering for the same purpose [4– 2 6]. The aim of the present study is to find the universal where κ = 4γω0/mc is twice the ratio between the pho- description of the harmonic damping devices for an ar- ton and electron energies in the electron rest frame and bitrary combination of the field strength and the field u ∈ [0, κ] is the dimensionless parameter [8]: period (or wavelength). ω θ ω ε − ε Let us start by consideration of the scattering of u = = ε = = 0 . (2) an ultra-relativistic electron in the oncoming plane ε θω ε0 − ω ε monochromatic electromagnetic wave. In Fig. 1 there is ϕ is the angle between the plane of polarisation and the an illustration for the process of inverse Compton scatter- scattering plane. θω and θε in Fig. 1 are defined as:

arXiv:1606.08602v5 [physics.acc-ph] 20 Nov 2016 ing of a soft photon with energy ω0 on an ultra-relativistic 2 electron with energy ε0 = γmc . The electron transfers 1 rκ urκ θ (u) = − 1; θ (u) = − 1 . (3) a part of its energy ω  ω0 to the photon so that the ω γ u ε γ u energy conservation law looks like ε0 ' ε + ω. We take from [7] the cross section for scattering of lin- The energy of the scattered photon and its maximum early polarised photons on unpolarised electrons at rest, possible value are: summed over the polarisation states of the final photons. u κ ω(u) = ε ; ω = ε . (4) 0 1 + u max 0 1 + κ In the limiting case when κ  1 the approximations ∗ [email protected] ωmax ' ε0κ, du ' dω/ε0 and u/κ ' ω/ωmax allow to 2 rewrite Eq. (1) as: According to Eq. (11) if PL = 1 [W], the scattering −11 probability W . 1.4 · 10 and it does not depend on dσ 2r2  ω  ω   = e 1 + 4 − 1 sin2 ϕ . (5) laser radiation wavelength. The RTS of intense laser dω dϕ ωmax ωmax ωmax pulse had been proposed in [5] for an electron beam cool- ing in a low-energy . In [9] this approach was Integration of (5) over ϕ yields: suggested for cooling of TeV muons. For laser cooling an ! extremely high laser power is required to produce a suffi- dσ 4πr2 ω  ω 2 = e 1 − 2 + 2 . (6) ciently strong field in the scattering area. The power of a dω ωmax ωmax ωmax gaussian laser beam is coupled with the on-axis magnetic induction amplitude B0 [T] as: Integration of (6) in the range from 0 to ωmax gives the 2 πc 2 2 Thomson cross section σT = 8πre /3. Thus the case of PL = σ(z) B0(z) , (12) scattering with κ  1 may be considered as Relativistic µ0 −7 −2 Thomson Scattering (RTS). Further we will need aver- where µ0 = 4π · 10 [N A ]. aged values of some RTS parameters, which can be easily obtained using the definitions (3), (4) and the cross sec- tion expression (1). The lowest order in the parameter κ B. Undulator yields the following results: 2 Now consider a plane undulator with a period λu hωi = 2γ ω0, (7) (¯λu = λu/2π) and on-axis magnetic induction By(z) = 2 28 2 2 7 2 ω = (γ ω ) = hωi , (8) Bu sin(z/λ¯u). The dimensionless undulator parameter is: 5 0 5  2 eBuλ¯u Bu λ¯u D 2E 6 λc 1 2 K = = , (13) (θε cos ϕ) = = θε , (9) mc Bc λ¯c 5 λ0 2 −3  2 where Bcλ¯c = mc/e ' 1.7 · 10 [T m] is the product D 2E 4 λc 1 9 2 of the Schwinger field strength Bc = 4.414 · 10 [T] and (θε sin ϕ) = = θε , (10) 5 λ0 3 the Compton wavelength of the electron. The maximum photon energy of the basic harmonic of undulator radia- where λ0 is the wavelength of incident photon and λc is the Compton wavelength of an electron. tion is: 2γ2hc ωmax = 2 . (14) λu(1 + K /2) II. RELATIVISTIC THOMSON SCATTERING ωmax decreases with increase of the undulator parameter due to effective decrease of the electron velocity along z- A. Laser Radiation axis caused by increase of the curvature of its trajectory in a strong field. Note that ωmax(K  1) in (14) and Let us explore a case when an ultra-relativistic electron ωmax(κ  1) in (4) are the same if λu = λ0/2. So are propagates along the axis of a 100% linearly polarised the RTS photon energy spectrum (6) and the scattered gaussian CW laser beam in the direction, opposite to the photon distribution for a weak undulator radiation (see light propagation. If the electron is on-axis of the laser ref. [10]). The only difference between RTS and weak beam within z ∈ [−a : a], the probability of scattering undulator radiation is the difference between head sea is the product of the Thomson cross section σT and the and standing waves. density of the photon target: If we replace B0(z) by By(z) in (12) and use the result a in Eq. (11), we obtain the average number of RTS events Z PLλ0 dz PL arctan(a/ZR) in a long (L  λu) undulator: W = σT 2 2 = , (11) πhc σ(z) PC π/2 1 σ 2πα L −a T 2 2 W = · BuλuL = K , (15) 2 µ0hc 3 λu where where L is an undulator length and α is the fine structure 1 • PL [W] and λ0 [m] are the laser power and the wave- constant. Factor 2 in Eq. (15) comes from the decrease 2 −1 length of its radiation. PLλ0/hc [m ] is the lon- of equivalent photon density in a standing wave. gitudinal density of laser photons. Equations (7–10) can be used for undulator, bearing in mind the substitutions λ = 2λ and ω = hc/2λ . p 2 0 u 0 u • σ(z) = σ0 1 + (z/ZR) is the transverse RMS laser The average electron energy loss in an undulator is de- beam size with the waist size σ0 at location z = 0. 2 termined as a product of Eq. (15) and hωi from Eq. (7): ZR = 4πσ /λ0 is the Rayleigh length. 0  2 h∆Ei α Bu L • P = c2/2σ ' 0.7124·1011 [W] is the critical laser u = γ , (16) C ~ T E 3 B λ¯ power when the probability of scattering is close to c c 2 100% (assuming a  ZR). where E = γmc andλ ¯c = λc/2π. 3

C. Emittance Excitation A spontaneous emission of the photons leads to the excitation of the betatron oscilla- tions, while this process takes place in the orbit regions If an electron beam with the horizontal emittance x 0 passes through an undulator (or laser beam), located with the non-zero dispersion functions, D and D . If a in the dispersion-free section of the beam orbit, beam wiggler is located in the dispersion-free section of the or- emittance is increased by radiation. For a single case bit, it will induce the dispersion of RTS the average increase of x depends on the beam 2 β-function [11]: λ¯w 0 λ¯w Dx(z) = sin(z/λ¯w),Dx(z) = cos(z/λ¯w). (23) ρ0 ρ0  2 β 2 3 λc h∆1iu = θx = β , (17) The excitation of horizontal emittance is determined by 2 20 λu w −1 radiation integral I5 [m ]: where θ = θ cos ϕ (θ cos ϕ was introduced in Eq. (9)). x ε ε 55α Z H(z)dz If we locate the centre of an undulator at the local mini- 2 5 w w h∆xiw = √ λ¯c γ I5 , where I5 = . (24) 2 |ρ3(z)| mum of β-function, β0, we have β(z) = β0 + z /β0. Con- 48 3 sequently, the average value of β-function inside this un- If the centre of a wiggler (z = 0) is located in a local dulator is hβi = β 1 + (L/β )2/12. One can write the 0 0 minimum of the β-function β(0) = β , H(z) is: expression for the emittance excitation in an undulator 0 as the product of Eq. (15) and Eq. (17): 2 02 0 2  H(z) = β0 D + (sD − D) β0. (25)  2 πα Bu L Integration yields the result: h∆xiu = hβi . (18) 10 Bc λu  2 11α Bw L 3 Note that the vertical emittance excitation is smaller, h∆xiw = √ hβi K , (26) 18 3 Bc λw h∆yi = 2/3 h∆xi according to Eq. (10). In the longi- tudinal dimension, the energy spread σE in the electron where hβi is the same as in Eq. (18). The increase in the beam is increased by RTS. This effect could be evaluated energy spread of the electron beam is determined by the as the product of Eq. (15) and Eq. (8): w −2 radiation integral I3 [m ]: 2  2 ∆σE u 14πα 2 Bu L 2 Z = γ . (19) ∆σE 55α 2 dz E2 15 B λ w = √ λ¯ γ5Iw, where Iw = . (27) c u E2 48 3 c 3 3 |ρ3(z)| After integration we obtain: III. WIGGLER 2 2 ∆σ 55α B  L E w √ 2 w Let us now switch to the consideration of a wiggler, 2 = γ K. (28) E 18 3 Bc λw i. e. an undulator with K  1. The standard approach for this case was taken from [1]. The following expres- Comparing Eq. (16) and Eq. (22) one can see that sions describe a wiggler field By(z) and local radius of an the average electron energy losses in either undulator electron trajectory ρ(z): or wiggler are the same and independent of the analy- sis approach. But the results for emittance excitation 1 sin(z/λ¯w) By(z) = Bw sin(z/λ¯w); = , where are different due to essential differences in the RTS and ρ(z) ρ0 the synchrotron radiation properties. The RTS-based ex- λw E (Bρ) pressions (18, 19) are valid only for K  1, where the λ¯w = , ρ0 = = 2π ecBw Bw standard radiation integrals show zero emittance excita- 3 (20) tion rate (h∆xi ∝ K , Eq. (26)). This is because RTS is not included to the radiation integrals and it should and (Bρ) = P/e = γλ¯cBc is the electron rigidity. Aver- be if we want them to be universal. Since the transverse age electron energy loss in a wiggler is determined by its emittance and the mean squared energy deviation repre- w −1 radiation integral I2 [m ] as follows: sent the squared amplitudes of betatron and synchrotron Z oscillations, the correct way of combining correspond- h∆Eiw 2 3 w w dz ing contributions from RTS and synchrotron radiation is = αλ¯cγ I2 , where I2 = . (21) E 3 ρ2(z) their direct summation. This yields the something like:

Integration along the wiggler of length L  λw yields  2 11α Bw L 3 the same result as was obtained in Eq. (16) for RTS: h∆xi ' √ hβi 0.89 + K , (29) 18 3 Bc λw h∆Ei α B 2 L ∆σ2 55α B 2 L w w E √ w 2 = γ . (22) 2 ' γ (1.66 + K) . (30) E 3 Bc λ¯c E 18 3 Bc λw 4

These expressions can be used to update an undula- of K ' 10, where the impact of the constant factor 0.89 tor/wiggler radiation integrals: in Eq. (33) is negligible. The situation when this im- pact could become noticeable is represented by the dot-    2 w L Bw πK L Bw ted curve in Fig. 2. The parameters of this curve are I2 = = , (31) λw Bρ γ 2 Bρ  2 w L Bw (1.66 + K) I3 = , (32) 100 λw Bρ γ  2 3 minimal DBA w L Bw 8 hβi 0.89 + K I = . (33) 80 Bw = 1 T, Lw = 100m 5 λ Bρ 15 γ3 w Bw = 10 T, Lw = 1 m 60 IV. RADIATION DAMPING 40 In an electron storage ring the equilibrium beam emit- tances are formed by radiation damping while the longi- 20 tudinal momentum losses are replaced by rf-system [1]: horizontal emittance [pm] emittance horizontal 2 σE  2 I3 0 = Cqγ , (34) 0.1 1 10 E 2I2 + I4x + I4y 2   undulator parameter K γ I5x,y 55¯λc x,y = Cq Cq = √ . (35) Jx,y I2 32 3 FIG. 2. Horizontal emittance x vs K, hβi = 10 m. The concept of damping wigglers application is to in- crease the radiation integral I2 and therefore decrease taken as Lw = 1 m and Bw = 10 T in order to keep the equilibrium emittances. Neglecting I4 and assuming w the same value of the damping integral I2 . This dot- Jx = 1 the following expressions describe this equilibrium ted curve shows small decrease in equilibrium emittance for a storage ring with damping wigglers: starting from K = 0.1 up to K ' 1. The reason is that in w 2 2 w this range I is inverse proportional to the wiggler period σE  Cqγ I3 + I3 5 = w , (36) λw, see Eq. (33). The parameter combination [K = 1, E 2 I2 + I2 w Bw = 10 T] means that λw ' 1 mm and so it is beyond 2 I5 + I5 x = Cqγ , (37) the existing technologies. According to Eq. (12) this is w 11 I2 + I2 the case when ∼ 10 W of radiation power is focused to 1 mm transverse size along 1 m length. Though only 1 m where I2,3,5 are radiation integrals of the storage ring it- w of a damping section length provides the same damping self and I2,3,5 are due to the damping wigglers. The lim- w effect as an 100 m section of regular wigglers. iting case when Ii  Ii (i = 2, 3, 5) gives the emittance values in the wiggler-dominated storage ring: σ 2 C 0.5 E = q γ (1.66 + K) , (38) E λw minimal DBA Bw = 1 T; Lw = 100m Cq 16 hβi 3 0.4 x = 0.89 + K . (39) Bw = 10 T; Lw = 1 m λw 15 γ For K = 0 case Eqs. (38, 39) yield the same results as 0.3 w were obtained in [5]. However the value of I5 if (K = 0) is so small that practically it can not be comparable to I5 0.2 of a ring (this issue seems have not been properly taken into account in [5]). Let us consider an imaginary storage energy spread [%] spread energy 0.1 ring with the DBA lattice [1] and the parameters:

3 • Beam energy – 2 GeV (γ = 4 · 10 ). 0.0 • 100 bending dipoles with B = 0.1 T. 0.1 1 10 undulator parameter K • Minimal DBA emittance x = 96 pm.

In Fig. 2 the dashed curve represents the horizontal emit- FIG. 3. Beam energy spread σE /E vs K. tance change due to wiggler damping section of length Lw = 100 m and Bw = 1 T. One can see that the damp- In the longitudinal dimension the equilibrium beam en- ing effect does not depend on K until it reaches the value ergy spread is shown in Fig. 3 for the same parameters as 5 they were in Fig. 2. The energy spread is always growing polarisation on the excitation of betatron oscillations, see with the decrease in the undulator parameter K. It hap- Eqs. (9) and (10). We point out the possibility of a sub- pens due to the valuable impact of the constant factor stantial reduction of the length of the cooling section, 1.66 in Eq. (32). preserving its effectiveness by increasing the field and re- ducing the period. However, the parameter combination like Bw ' 10 T and λw ' 1 mm is presently beyond the V. CONCLUSION existing technology limits for either laser or undulator cases. But if this limit could be overcome, the pending We have examined the impact of relativistic Thomson amendment will have a decisive role. scattering to the radiation integrals of undulators and wigglers. The solution is suggested for the correct evalu- ation of these integrals for arbitrary values of undulator ACKNOWLEDGMENTS parameter. When K = 0, these results are in agreement with the ones obtained earlier [5, 6]. However, we used The author wants to express his gratitude to Eugene a somewhat different approach, provide an opportunity Levichev for inspiration of these studies and for useful for optimisation of the damping parameters and also a discussions. This work was supported by Russian Science proper account of the influence of the direction of wave Foundation (project N 14-50-00080).

[1] A. W. Chao and M. Tigner, Handbook of Accelerator [6] A. N. Lebedev, in 19th Russian Accelerator Conferences Physics and Engineering (World Scientific, 1999). (RuPAC 2004) Dubna, Russia (2004) pp. 9–12. [2] H. Wiedemann, Nuclear Instruments and Methods in [7] V. Berestetskii, E. Lifshitz, and L. Pitaevskii, Quan- Physics Research Section A: Accelerators, Spectrometers, tum Electrodynamics, Course of theoretical physics Detectors and Associated Equipment 266, 24 (1988). (Butterworth-Heinemann, 1982). [3] D. Schoerling, F. Antoniou, A. Bernhard, A. Bragin, [8] E. V. Abakumova, M. N. Achasov, D. E. Berkaev, V. V. M. Karppinen, R. Maccaferri, N. Mezentsev, Y. Pa- Kaminsky, N. Y. Muchnoi, E. A. Perevedentsev, E. E. paphilippou, P. Peiffer, R. Rossmanith, G. Rumolo, Pyata, and Y. M. Shatunov, Phys. Rev. Lett. 110, S. Russenschuck, P. Vobly, and K. Zolotarev, Phys. Rev. 140402 (2013). ST Accel. Beams 15, 042401 (2012). [9] F. DeJongh, Nucl. Instr. Meth. A 472, 585 (2001). [4] V. Telnov, Phys. Rev. Lett. 78, 4757 (1997). [10] A. Hofmann, The Physics of Synchrotron Radiation [5] Z. Huang and R. D. Ruth, Phys. Rev. Lett. 80, 976 (Cambridge University Press, 2004). (1998). [11] H. Bruck, Accelerateurs circulaires de particules: intro- duction a la theorie (Presses Universitaires De France, 1966).