Higher Order Modes in the Project-X Linac V. Yakovlev, Fermilab

The talk is on behalf of the FNAL team: M. Champion, C. Ginsburg, I. Gonin, Ch. Grimm, T. Khabiboulline, A. Lunin, Th. Nicol, Yu. Orlov, A. Saini, D. Sergatskov, N. Solyak, A. Sukhanov, and A. Vostrikov

PROJECT X

3-GeV, 1-mA CW linac provides beam for rare processes program ~3 MW; flexible provision for beam requirements supporting multiple users; <5% of beam is sent to the Main Injector. Project X multi-experiment facility: 3 3 GeV GeV GeV

RF Cavities of the Project-X linac: HWR, MHz SSR1, 325 MHz SSR2, 325 MHz (ANL) (FNAL) (FNAL) HWR and spoke cavities of the Project X front end

Elliptical cavities of the high-energy part of CW linac Single-cell models: 5-cell model, HE650 LE 650 MHz (JLAB) HE 650 MHz (FNAL)

CM designs for Project X CW linac: HWR SSR1 HE650

Motivation for HOM investigation in PX : HOM damper is a vulnerable, expensive and complicated part of SC acceleration structure (problems – multipactoring, damage of feed-through, etc; additional hardware – cables, feed-through, connectors, loads); SNS SC linac experience shows that HOM dampers may limit a cavity performance and reduce operation reliability; SNS linac experience doesn’t show necessity of the HOM couplers. The PX cavities are under development and it is necessary to understand necessity of HOM dampers for these cavities.

Effects of HOMs (collective instabilities, cavity heating) depend on the bunch population, beam current, beam spectrum (chopping!) and cavity HOM spectrum. “Average” HOM power per cavity (incoherent losses): P av = k loss Q b I av Project X beta=0.9 SNS beta=0.81 ILC beta=1 I av, mA114.8e-2 Q b, pC25.6* k loss, V/pC P av,mW/cavity The ILC-type 1.3 GHz cavities contain HOM couplers that reduce the loaded Q-factors for transverse and longitudinal HOMs down to For SNS cavities no measureable HOM signals from beam observed. *162.5 MHz beam sequence frequency.

In ILC linac HOM dampers are necessary. All the 1.3 GHz ILC cavities are equipped by HOM couplers, which work successfully at DESY. Some problem with long-pulse operation: J. Sekutowicz, HOM Workshop, Cornell, Analysis of the HOM resonance excitation and collective effects (BBU and klystron-type) in SNS linac does not show critical influence of the HOMs on the beam dynamics; Our goal is to understand the HOM influence on the beam dynamics in Project X in order to decide whether we need the HPM dampers in high energy part of the linac and in the low energy part as well.

Scope of investigations: Loss factor calculations for non-relativistic beam; Beam spectrum investigation for different Project X operation options; Cavity spectrum investigations (simulations and measurements) for all the Project X sections - HWR, SSR1, SSR2, LE650 and HE650 – impedance and spectrum density versus frequency; Resonant frequency spread for HOMs estimations and if possible measurements for all the cavities of the Project X; Q estimations for different modes; HOM impedance spread (estimations and measurements) for all the cavities of the Project X; HOM impedance dependence on the particle velocity over the linac; Microphonics for different HOMs; Stability of HOMs versus operation mode tuning/detuning;

Cryogenic loss estimations for different option of the Project X operation (taking into account propagating and non-propagating modes); Estimation of the beam emittance (longitudinal and transverse) dilution caused by HOMs; Longitudinal “klystron type” instabilities caused by HOMs; Transverse instability – BBU; Comparison to operating accelerators (SNS)

Specifics of Higher Order Mode effects in the Project-X CW Linac: Non-relativistic beam; Small bunch population (26 pC/bunch) Small current (1 mA); No feedback (linac); CW operation; Complicated beam timing structure (dense frequency spectrum). Possible issues: Resonance excitation of HOMs; Collective effects; Cryo-losses; Emittance dilution (longitudinal and transverse).

Non-relativistic beam in Project X linac: HE electron linac Proton linac (Project X) (ILC, XFEL or NGLS) f max ~ c/σ bunch f max ~ c/a for σ bunch = 50µ, f max < 6 THz for a = 50mm, f max < 6 GHz Loss factor depends strongly on the σ field !

Incoherent loses per CM in HE 650 MHz section of the Project X linac. Longitudinal impedance versus frequency: R/Q =R0exp(-f/f0)

The beam splitter The beam spectrum in Project X is determined by the beam distribution scheme:

RF separator directs Muon pulses (16e7) MHz, 100 nsec at 1 MHz 700 kW; Kaon pulses (16e7) 20.3 MHz 1540 kW Nuclear pulses (16e7) MHz 770 kW Deflecting structure should operating at the frequency f 0 (m±1/4),where f 0 is the bunch sequence frequency (f 0 =162.5 MHz). 1  sec period at 3 GeV Operating scenario at the bunch sequence frequency of MHz: t, mksec

Beam after splitter 10 MHz bunches 20 MHz bunches 1 MHz pulses Muon pulses Nuclear pulses Kaon pulses

Injection to the pulse linac for neutrino experiments: Pulse width – 4.3 msec; Repetition rate – 10 Hz.

The beam current spectrum contains ● harmonics of the bunch sequence frequency of MHz and ● sidebands of the harmonics of MHz separated by 1 MHz. Idealized beam spectrum

DimensionBeta=0.61Beta=0.9 Regular cell End cell Regular cell End cell r, mm R, mm L, mm A, mm B, mm a, mm b, mm α,° Dimensions of the 650 MHz cavities beta=0.61 beta=0.9 (bottom).

Monopole mode spectrum β = 0.61 β = 0.9

Dipole mode spectrum β = 0.61 β = 0.9

Impedances of monopole modes For β = 0.61: all the modes have (r/Q) below 10 Ohms; For β = 0.90: -two modes have (r/Q) =12 Ohm: F=1988 (df=2.2 MHz) and 2159 MHz (df=3.9 MHz), -one mode has (r/Q) = 22 Ohm: F= MHz (df=0.44 MHz), and -one mode has (r/Q) = 130 Ohm: F= MHz (df=2.0 MHz) df is the difference between the HOM frequency and nearest main beam spectrum line.

Impedances of dipole modes β = 0.61 β = 0.9 For β = 0.61 three modes have (r/Q) above 10 4 Ohm/m 2 (F=974, and 1293 MHz); For β = 0.90 four modes have (r/Q) above 10 4 Ohm/m 2 (F=946.6, 950.3, 1376 and 1383 MHz).

(r/Q) for HOM modes depends on the particle velocity β: 650 MHz, β=0.9 cavity a) b) Monopole (a) and dipole (b) impedances of “the most dangerous” modes for beta=0.9 cavity versus accelerated particle velocity.

HOM have frequency spread caused by manufacturing errors.  For ILC cavity r.m.s. spread σ f of the resonance frequencies is 6-9 MHz depending on the pass band, according to DESY measurement statistics: J. Sekutowicz, HOM damping,” ILC Workshop, KEK, November 13-15, However, in a process of “technology improvement” σ f reduced to ~1 MHz;  Cornell: σ f ≈ 10.9·10 -4 ×(f HOM -f 0 ), for PX σ f = 1-2 MHz;  SNS: σ f ≈ (9.6· ·10 -4 )×(f HOM -f 0 ); Δf max =|f HOM,calculated -f HOM,measured | ~ σ f

Effect of the HOMs:  Resonance excitation;  Collective effects.  Resonance excitation, monopole modes. Monopole modes should not increase the beam longitudinal emittance ( = 1.6 keV*nsec): is average energy gain caused by HOM, is a bunch length. For high-Q resonances and thus, is the deference between the HOM frequency and the beam spectrum line frequency ( ). Is a beam spectrum line amplitude.

The worst case: beginning of the high-beta 650 MHz section. = 7.7e-3 nsec (or 1.8 deg). For = 0.5 mA and (R/Q)= 130 Ohm (HOM with the frequency of 1241 MHz) one has >> 70 Hz  When the distance between the beam spectrum lines is 5 MHz, and the frequency spread is 1 MHz, the probability that the cavity has the resonant frequency close enough to the beam spectrum line is <1e-4.  The gain caused by the HOM is <300 keV, that is small compared to the operating mode gain, ~20 MeV, and does not contribute to the cryogenic losses because 1241 MHz mode is TM 011 mode in a cell, and, thus, it’s surface distribution is “orthogonal” to one of the operating mode. Q 0 ~ 5e9.  If the HOM mode is in resonance, it’s Q loaded << 2e7. One should take care on the 2d band monopole HOMs in order to avoid resonance accidental excitation!

 Resonance excitation, dipole modes.  Dipole modes should not increase the beam transverse emittance  Transverse kick caused by the HOM is: (k=2π/λ)  Emittance increase may be estimated the following way: is beta-function near the cavity.  Thus, is proton rest mass in eV.  For f =1376 MHz, (R/Q) 1 =60 kOhm/m 2 (worst case), proton energy of 500 MeV, β f =15 m and x 0 =1 mm one has δf >> 2.5 Hz.  If the HOM is in resonance, Q << 1.4e9..5 Does not look to be a problem. ( ε = 2.5e-7/ βγ m).

Approach:  Cavity HOM simulations taking into account manufacturing misalignments;  Compare the frequency and R/Q spread with the measured data if available;  Calculation and measurements of R/Q and Q of the HOMs;  For propagating HOMs: variation of the reflection from the ports and determining maximal R/Q, which is used in RF loss and emittance dilution estimations. Calculate surface field as a sum of contributions from exited modes; Calculate losses by integrating squared surface field; Calculate total losses by the sum of losses from individual beam harmonics; Calculate energy change caused by HOMs for every bunch and emittance dilution caused by HOM Statistics accumulation and analysis.

a) b) 1/5π 2/5π 3/5π 4/5π π HOMs in an ideal cavity HOMs in a “realistic” cavity, i.e., in presence of misalignments. In both cases the operating mode is tuned to correct frequency and field flatness.

Calculations of the power losses in the HE 650 MHz cavities: Q=1.e7 Q=1.e10

red – maximum R/Q vs beta for all HOMs, blue –R/Q for an ideal cavity, green – “realistic” R/Q for a cavity with mechanical errors. Probability of monopole HOMs RF losses per cryomodule. (HE 650 MHz cavities)

I = 1ma I = 5ma

Longitudinal emittance dilution: Because HOM and beam components frequencies are not multiple, the resonance HOM excitation by one of the beam component will cause that other parts will see its voltage at a “random” phase. Thus, in a case of a high-Q resonance, such a HOM mode may introduce a significant energy variation along the beam train. This variation is proportional to the amplitude of the beam current spectrum line closest to the HOM frequency. HOM Phase, [deg] Bunch number Resonance monopole HOM excitation in HE 650 by various beam components, a) – “muon”, b) – “nuclear” and c) “kaon” There is no this effect if the beam structure is regular – for example, for NGLS

The probability of longitudinal emittance growth for I = 1mA average current Based on the predicted deviations of monopole HOMs frequency and Q ext we generated ~10 5 random linacs in order to estimate the probability of the beam longitudinal emittance growth.

For 5 mA there is no problems with the emittance; For 10 mA thee may be some problems.

HOM damping through the main coupler. Main coupler should provide Q ext ~ 2-3e7 for the operating mode. D=17mm D t =100mm D c =40mm DaDa DaDa Qext

HOM damping through the main coupler (perfect window transmission). The coupler window is optimized to provide a good transmission for the 2d pass band

Monopole Band #2 Ez* on axis Mode # Freq [GHz] Q_extR/Q max [Ω] β max 1/5π E /5π E /5π E /5π E π E R/Q vs β

650 MHz coupler, coupler window and WG-coax transition: 1 st PB 2d PB 3d PB 4 th BP 5 th PB 1 st and 2d BPs have high (r/Q), 3d and 4 th BPs have modest (r/Q) and damping; 5 th BP has modest (r/Q) and good damping.

Map of the electric fields in the chain of cavities connected with bellows

Project X versus SNS (HE parts) Q=1.e8Q=1.e9 Power of losses (per cavity) distribution for high energy cavities for beta=0.9 (PX) and beta=0.81 (SNS). SNS: Lorentz detuning and pulse-to-pulse jitter may reduce the losses; Project X: microphonics may reduce the losses.

Microphonics for HOMs in HE 650 MHZ cavity of the PX linac. df/dP in Hz/mbar (blue) for different cavity modes; mode # is in black. The cavity + He vessel are optimized in order to minimize df/dP for the operating mode #5, but not HOMs.

HOM investigations for SSR1: Bead-pull measurement test setup for an SSR1 cavity prototype. Results of analysis show that HOM losses and corresponding cryogenic heat load are extremely small in the range of QL from 1e6 to 1e9 ‣ mean power loss is less than 1 nW ‣ 99% quantile loss is less than 1 uW

 Even in the case when it happens, it is possible to move the HOM frequency away from the spectrum line simply detuning the cavity by tens of kHz, and then tune the operating mode back to the resonance.  A special test was made with the 1.3 GHz, 9-cell ILC cavity. The cavity was tested at 2 K.  The operating mode was detuned by Δf =90 kHz, and then was tuned back.  The frequencies of HOMs moved after this procedure by δf = Hz because of small residual deformation of the cavity. What to do if the HOM has resonance frequency close to the beam spectrum line*? f, MHzΔf, kHzδf, HzPassband Monopole Dipole Dipole Dipole Dipole Dipole Dipole Dipople Dipople Quadrupole Quadrupole Monopole Monopole Monopole Monopole *Timergali Khabiboulline, this workshop

Collective effects:  Beam break –up (BBU), transverse.  “Klystron-type”, longitudinal. Why collective effects may not be an issue: 1.No feedback as in CEBAF; 2.Different cavity types with different frequencies and different HOM spectrum are used; 3.Frequency spread of HOMs in each cavity type, caused by manufacturing errors; 4.Velocity dependence of the (R/Q); 5.Small beam current.

BBU estimations for 650 MHz part of the Project – X linac: Simple model:  Short bunches;  Current lattice design - N. Solyak, et al;  Two types of the 650 MHz cavities, beta=0.61 and beta=0.9;  Five dipole pass bands are taking into account;  Random transverse misalignment of the cavities;  Beam time structure – S. Nagaitsev (see above)  Model: (P-W) Parameters:  Beam current: 1 mA;  RFQ frequency: 325 MHz;  r.m.s cavity off-set: 0.5 mm; Transverse dynamics.

Q = 10 7 Q = 10 8 Q = 10 9 Low beta section, Resonance case, One HOM only, ( MHz, 24 kOHm/m 2 ) No dependence of (R/Q) on beta. Δε~Δf -2 Transverse emittance dilution vs. HOM frequency spread Δf.

“Realistic” linac. Transverse emittance dilution vs. Δf. Q = 10 7 Q = 10 8 Q = 10 9 No noticeable effect.

Klystron-type longitudinal instability*, longitudinal emittance dilution caused by monopole HOMs J. Tuckmantel, “Do we need HOM dampers on superconducting cavities in proton linacs?”, HOM Workshop, CERN, July 2009 Q = 10 7 Q = 10 8 Q = 10 9

σ f = 10 kHz σ f = 100 kHz σ f = 1 MHz σ f = 10 MHz Cavity voltage distribution for different HOM frequency spread No noticeable effect.

σ f = 10 kHz σ f = 100 kHz σ f = 1 MHz σ f = 10 MHz HOM Voltage

Summary. To damp or not to damp?  BBU in 650 MHz section should not be a problem;  “Klystron-type” longitudinal instability does not look to be a problem as well.  Resonance excitation of the dipole modes does not look to be an issue;  Accidental resonance excitation of the 2d monopole band in beta=0.9 section may lead to longitudinal emittance dilution, but probability is small. No HOM dampers!

Flange for HOM damper “Safe” option: make flanges (~40 mm)for HOM dampers, seal them and install the dampers in future for possible upgrade if necessary (as they discuss at SPL).