Vasim Khan & Roger Jones MAIN LINAC DDS DESIGN Vasim Khan & Roger Jones 06.11.09

Outlook CLIC scheme Two Beam Acceleration Optimised parameters What is wakefield Main Linacs Design constraints Present structure DDS design Comparison Forthcoming V. Khan HEP group, The University of Manchester 06.11.09 1/41

CLIC scheme e- e+ collider C. M. Energy : 3 TeV Normal conducting technology Frequency : 12 GHz Acc. Gradient : 100 MV/m Luminosity : ~ 1034 cm-2 s-1 Novel technique : Two beam acceleration Overall site length 48 km........(compact ?) V. Khan HEP group, The University of Manchester 06.11.09 2/41

CLIC parameters Parameters Designed value unit C.M Energy 3 TeV Frequency 11.9942 GHz Acc. Gradient 100 MV/m No. of cells per structure 24 luminosity 5.9 x 1034 cm-2 s-1 Luminosity in 1% of energy 2 x 1034 No. of particles per bunch 3.72 x 109 No. of bunches per pulse 312 Bunch length 44 μm Transverse emittance (x,y) @ IP 660, 20 nm rad Beam size (x,y) @ IP 40,0.9 nm Crossing angle @ IP 20 mrad Beam Power 14 MW Total site length 48.4 km Total site AC power 392 Overall wall plug-beam efficiency 7.1 % V. Khan HEP group, The University of Manchester 06.11.09 3/41

CLIC complete layout V. Khan HEP group, The University of Manchester 06.11.09 4/41

Two Beam Acceleration 140,000 main linac structures. It is difficult to supply power using conventional RF source i.e. Klystron......it will require 10,000’s of such klystrons. A low energy high current beam (drive beam) running parallel to the main beam. Drive beam interacts with the impedance of the Power Extraction and Transfer Structures. Drive beam is thus decelerated. The decelerated energy is used to accelerate main beam. V. Khan HEP group, The University of Manchester 06.11.09 5/41

Why ? 1) Linacs ? 2) Normal conducting ? 3) X-band frequency of 12 GHz ? Synchrotron radiation [1] Energy loss per revolution [1] [1] S.Y. Lee, Accelerator Physics. Colliders Particle Beam energy Circumference ∆E/rev. circular TeV km LHC p-p 7.0 27 6.0 keV LEP e- e+ 0.1 26.7 2.8 MeV CLIC* 1.5 106 * Assume if CLIC was a circular collider V. Khan HEP group, The University of Manchester 06.11.09 6/41

Is there any energy loss in linear acceleration ? In Linac we consider the power radiated to the power supplied by an external source [1] [1] S.Y. Lee, Accelerator Physics. Colliders Particle Beam energy Accelerator length P/(dE/dt) Linear TeV km % CLIC e- e+ 1.5 21 10-12 CLIC* 55 20 1.0 * Assume if CLIC was proposed to accelerate with accelerating gradient of ~2.75 GeV/m V. Khan HEP group, The University of Manchester 06.11.09 7/41

Frequency scaling of rf parameters Normal conducting ? Need high gradient for a feasible site length High gradient cavities will have high surface fields Super conducting (SC) cavities can be operated up to ~ 40 MV/m High gradient in SC cavities will quench the superconductivity. Possible option is Normal Conducting (NC) cavities. Frequency scaling of rf parameters RF parameters Requirement Normal conducting Super conducting RF surface resistance (Rs) Low Power dissipated (Pdis) Quality factor (Q) High Shunt impedance per unit length (R’) V. Khan HEP group, The University of Manchester 06.11.09 8/41

X-band (8-12 GHz)? Initial proposal was 150 MV/m @ 30 GHz Operation with these parameters will suffer major breakdown issues The optimisation procedure has resulted in 100 MV/m gradient with 12 & 14 GHz frequency option. Lower frequency was chosen to utilise the expertise and technical know-how of the discontinued NLC/GLC project which was also proposed at 12 GHz. V. Khan HEP group, The University of Manchester 06.11.09 9/41

What is wakefield ? V. Khan HEP group, The University of Manchester 06.11.09 10/41

What is wakefield ? Fields excited by the ultra relativistic particles Short range wake :tail of the bunch experiences field excited by the head of the bunch Long range wake : trailing bunches experiences fields excited by the leading bunches Transverse wake : luminosity dilution Longitudinal : energy spread V. Khan HEP group, The University of Manchester 06.11.09 10/41

Main Linac V. Khan HEP group, The University of Manchester 06.11.09 11/41

Glossary 1 Monopole mode Dipole mode V. Khan HEP group, The University of Manchester 06.11.09 12/41

Glossary 2 Synchronous mode : Most dominating mode in an accelerating cell, its phase vel. is in synchronous with speed of light Bandwidth : Difference between the synchronous frequencies of the end cells (lowest dipole ) Large BW : 3.3 GHz Small BW : 1 GHz Moderate BW : 2.3 GHz Heavy Damping : Q ~10 Moderate Damping : Q ~500-1000 V. Khan HEP group, The University of Manchester 06.11.09 13/41

Beam dynamics constraints [1],[2] RF breakdown constraint [1],[2] 1) 2) Pulsed surface heating 3) Cost factor Beam dynamics constraints [1],[2] For a given structure, no. of particles per bunch N is decided by the <a>/λ and Δa/<a> Maximum allowed wake on the first trailing bunch Rest of the bunches should see a wake less than this wake(i.e. No recoherence). Ref: [1]: A. Grudiev and W. Wuensch, Design of an x-band accelerating structure for the CLIC main linacs, LINAC08 . [2]: H. Braun, et al. , Updated CLIC Parameters, CLIC-Note 764, 2008. V. Khan HEP group, The University of Manchester 06.11.09 14/41

Accelerating cells : Several designs V. Khan HEP group, The University of Manchester 06.11.09 15/42

Wakefield suppression in CLIC main linacs The present main accelerating structure (WDS) for the CLIC relies on linear tapering of cell parameters and heavy damping with a Q of ~10. The wake-field suppression in this case entails locating the dielectric damping materials in relatively close proximity to the location of the accelerating cells. To minimise the breakdown probability and reduce the pulse surface heating, we are looking into an alternative scheme for the main accelerating structures: Detuning the first dipole band by forcing the cell parameters to have Gaussian spread in the frequencies Considering the moderate damping Q~500 V. Khan HEP group, The University of Manchester 06.11.09 16/41

CLIC_G: Present baseline waveguide damped design Structure CLIC_G Frequency (GHz) 12 Avg. Iris radius/wavelength <a>/λ 0.11 Input / Output iris radii (mm) 3.15, 2.35 Input / Output iris thickness (mm) 1.67, 1.0 Group velocity (% c) 1.66, 0.83 No. of cells per cavity 24 Bunch separation (rf cycles) 6 No. of bunches in a train 312 Ref: A. Grudiev, W. Wuensch, Design of an x-band accelerating structure for the CLIC main linacs, LINAC08 V. Khan HEP group, The University of Manchester 06.11.09 17/41

Damped and detuned design Detuning: A smooth variation in the iris radii spreads the dipole frequencies. This spread does not allow wake to add coherently Error function distribution to the iris radii varion results in a rapid decay of wakefield. Due to limited number of cells in a structure (trunated Gaussian) wakefield recoheres. Damping: The recoherence of the wakefield is suppressed by means of a damping waveguide like structure (manifold). Interleaving neighbouring structure frequencies help enhance the wake suppression V. Khan HEP group, The University of Manchester 06.11.09 18/41

NLC/GLC DDS design Acceleration cells Beam tube Manifold HOM coupler High power rf coupler HOM coupler Beam tube Acceleration cells Manifold V. Khan HEP group, The University of Manchester 06.11.09 19/41

Advantages Disadvantages More sensitive to machine tolerances Moderate damping scheme: Breakdown probability is less Pulse temperature rise is comparatively less Manifolds can be used for beam position monitoring and cell alignments Disadvantages More sensitive to machine tolerances Need bigger bandwidth for adequate detunig and hence more input power to achieve desired accelerating gradient Interleaving neighbouring structure frequencies is necessary, hence structures are not identical which is not cost-efficient V. Khan HEP group, The University of Manchester 06.11.09 20/41

Key parameters for designing an accelerating structure @ 12 GHz & 100 MV/m Iris radii of the end cells Iris thickness <a>/λ Group velocity No. of cells per structure Bunch spacing Bunch charge No. of bunches in a train => pulse length V. Khan HEP group, The University of Manchester 06.11.09 21/41

Uncoupled (designed) distribution of Kdn/df for a four fold interleaved structure Error function distribution V. Khan HEP group, The University of Manchester 06.11.09 22/41

Eight fold interleaved structure Finite no of modes leads to a recoherance at ~ 85 ns. But for a damping Q of ~1000 the amplitude wake is still below 1V/pc/mm/m Why not 3.3 GHz structure? 3.3 GHz structure does satisfies beam dynamics constraints but does not satisfies RF breakdown constraints. V. Khan HEP group, The University of Manchester 06.11.09 23/41

Zero crossing scheme Parameters closely tied to that of CLIC_G with two major changes Gaussian distribution of cell parameters Q= 500 V. Khan HEP group, The University of Manchester 06.11.09 24/41

CLIC_ZC structure # Parameters ZC1 ZC2 Unit 1 <a>/λ 0.102 0.1 - 2 IP/OP iris thickness 1.6 / 0.7 1.6/0.7 mm 3 IP / OP iris radii 2.99 / 2.13 2.87/2.13 4 IP / OP group velocity 1.49 / 0.83 1.45/.83 5 First / Last cell Q0 6366 / 6643 6408/6668 6 First / Last cell Shunt impedance 107 / 138 108/138 MΏ /m 7 Filling time 56.8 58.6 ns 8 IP Power (peak) 48 47 MW 9 RF-to-beam efficiency 27.09 26.11 % 10 Bunch population 3.0 x109 2.9 x 109 11 Esur (max) 285 231 MV/m 13 ∆T max 20 25.2 K 14 14.07 14.36 MW(ns)^1/3/mm V. Khan HEP group, The University of Manchester 06.11.09 25/41

Why not zero crossing scheme ? Though RF breakdown constraints are satisfied it will be very challenging to achieve zero crossing scheme due to tight tolerances. It may not be feasible to build a structure based on zero crossing scheme. Possible option is a moderate bandwidth. V. Khan HEP group, The University of Manchester 06.11.09 26/41

A 2.3 GHz Damped-detuned structure Cell parameters Cell # 1 Cell # 24 Iris radius (mm) 4.0 2.15 Iris thickness (mm) 0.7 Ellipticity 1.0 2.0 Q 4771 6355 R’/Q (kΩ/m) 11.64 20.09 vg/c (%) 2.13 0.9 ∆f = 3.6 σ = 2.3 GHz ∆f/fc =13.75 % <a>/λ=0.126 V. Khan HEP group, The University of Manchester 06.11.09 27/41

Typical DDS cell Accelerating mode (monopole mode) Manifold Coupling slot Accelerating mode (monopole mode) Dipole mode Manifold mode V. Khan HEP group, The University of Manchester 06.11.09 28/41

Spectral function -----(IFT) Wake function 24 cells No interleaving 24 cells No interleaving ∆fmin = 65 MHz ∆tmax =15.38 ns ∆s = 4.61 m 48cells 2-fold interleaving ∆fmin = 32.5 MHz ∆tmax =30.76 ns ∆s = 9.22 m 48cells 2-fold interleaving V. Khan HEP group, The University of Manchester 06.11.09 29/41

Spectral function -----(IFT) Wake function 96 cells 4-fold interleaving ∆fmin = 16.25 MHz ∆tmax = 61.52 ns ∆s = 18.46 m 96 cells 4-fold interleaving 192 cells 8-fold interleaving 192 cells 8-fold interleaving ∆fmin = 8.12 MHz ∆tmax =123 ns ∆s = 36.92 m V. Khan HEP group, The University of Manchester 06.11.09 30/41

CLIC_G vs CLIC_DDS For CLIC_G structure <a>/λ=0.11, considering the beam dynamics constraint bunch population is 3.72 x 10^9 particles per bunch and the heavy damping can allow an inter bunch spacing as compact as ~0.5 ns. This leads to about 1 A beam current and rf –to-beam efficiency of ~28%. For CLIC_DDS structure (2.3 GHz) <a>/λ=0.126, and has an advantage of populating bunches up to 4.5x10^9 particles but a moderate Q~500 will require an inter bunch spacing of 8 cycles (~ 0.67 ns). Though the bunch spacing is increased in CLIC_DDS, the beam current is compensated by increasing the bunch population and hence the rf-to-beam efficiency of the structure is not affected alarmingly. V. Khan HEP group, The University of Manchester 06.11.09 31/41

CLIC_G vs CLIC_DDS Parameters CLIC_G (Optimised) [1,2] CLIC_DDS (Single structure) CLIC_DDS* (8-fold interleaved) Bunch space (rf cycles/ns) 6/0.5 8/0.67 Limit on wake (V/pC/mm/m) 7.1 5.6 5.3 Number of bunches 312 Bunch population (109) 3.72 4.7 5.0 Pulse length (ns) 240.8 273 272.2 Fill time (ns) 62.9 42 40.8 Pin (MW) 63.8 72 75.8 Esur max. (MV/m) 245 232 224 Pulse temperature rise (K) 53 65 RF-beam-eff. 27.7 26.6 26.7 Figure of merit (a.u.) 9.1 8.41 8.29 [1] A. Grudiev, CLIC-ACE, JAN 08 [2] H. Braun, CLIC Note 764, 2008 * Averaged values of structure #1 & #8 V. Khan HEP group, The University of Manchester 06.11.09 32/41

? CLIC_G vs CLIC_DDS CLIC_G CLIC_DDS Undamped Damped 1.56 1.55 @ 4.0 mm iris radius ? Undamped Damped 1.56 1.55 V. Khan HEP group, The University of Manchester 06.11.09 33/41

Circular Square Elliptical Convex Concave V. Khan HEP group, The University of Manchester 06.11.09 34/41

Iris radius = 4. 0 mm Iris thickness = 4.0 mm Undamped cell with various geometries Circular Rectangular Elliptical* ε=10 ε=5 ε=3.33 ε=2.0 ε=1 Facc(GHz) 12.24 12.09 11.98 12.0 11.99 Eacc(V/m) 0.43 0.42 Hsur max /Eacc (mA/m) 3.64 4.86 4.71 4.54 4.29 3.75 3 Esur max /Eacc (mA/m) 2.27 2.33 2.28 V. Khan HEP group, The University of Manchester 06.11.09 35/41

Circular Square V. Khan HEP group, The University of Manchester 06.11.09 36/41

Surface magnetic field in different cell geometries Circular Rectangular ε=10 ε=5 ε=3 ε=2 ε=1 Hsur max /Eacc (mA/V) V. Khan HEP group, The University of Manchester 06.11.09 37/42

Present status 4.35 mA/V V. Khan HEP group, The University of Manchester 06.11.09 38/41

Next ? Surface magnetic field is reduced, need to reduce surface electric field. Efficiency calculations. Dipole simulations, necessary changes in the geometry for critical coupling. Wakefield calculations Mechanical design with power couplers. Hopefully will build one out of eight fold-interleaved structure in collaboration with CERN-KEK-SLAC .................................Thesis writing.... V. Khan HEP group, The University of Manchester 06.11.09 39/41

Acknowledgements We would like to thank our collaborators for their involvement in discussions and many useful suggestions from CERN : W. Wuensch, A. Grudiev, D. Schulte and R. Zennaro KEK : T. Higo SLAC : J. Wang and Z. Li V. Khan HEP group, The University of Manchester 06.11.09 40/41

Thank you ........ Many of life’s failures are people who did not realise how close they were to success when they gave up. .................Thomas Edison V. Khan HEP group, The University of Manchester 06.11.09 41/41