progress towards testing non-linear integrable optics in a ... · 05/11/2019 lucy martin 1 progress...
TRANSCRIPT
-
05/11/2019 Lucy Martin 1
Progress towards testing non-linear integrable optics in a Paul trap
Lucy Martin
4th ICFA mini workshop on space charge – 05/11/19
-
05/11/2019 Lucy Martin 2
Introduction
• What is a Paul trap?
• How can it be used to study accelerator physics?
• How does it work?
• What is Nonlinear Integrable Optics (NIO)
• Why do we want to test NIO on a Paul trap
• Designing the experiment on IBEX
• Can it be done?
-
05/11/2019 Lucy Martin 3
-
05/11/2019 Lucy Martin 4
Motion in a Paul trap
• Hamiltonian of a Paul trap :
where
• Hamiltonian of a conventional accelerator:
where
-
05/11/2019 Lucy Martin 5
IBEX and SPOD: Linear Paul Traps
IBEX at the Rutherford Appleton Lab, UKS-POD at Hiroshima University, JapanPTSX at Princeton Plasma Physics lab, US
• What are the advantages of a LPT?
• What are the limitations?• No longitudinal effects = coasting beam
-
05/11/2019 Lucy Martin 6
Past experiments on LPTs
• Coherent motion
• Excitation of dipole and quadrupole resonances
• Study of how resonances interact at half integer tunes
• Studying resonance crossing for novel FFA accelerators
• Beam Halo and beam size growth
• Cm factor studies8
-
05/11/2019 Lucy Martin 7
Linear Paul Trap
4. Ions confined longitudinally with end caps
1. Argon gas introduced to vessel at ~10-7 mbar
2. Electron gun ionises Ar in trapping region
3. Ions confined transversely via 4 cylindrical rods
-
05/11/2019 Lucy Martin 8
Experimental procedure
• A simple FODO cell is created using a sine wave• 1 MHz
• System scans across a large range of tunes
• Ions are created at the optimal tune
• Tune is then changed to the desired operating point by changing voltage
• At this tune ions are stored for a given time (eg. 100ms)
• Ions are then extracted onto an Micro Channel Plate (MCP) or a Faraday cup
-
05/11/2019 Lucy Martin 9
Diagnostics and CalibrationMCP image
Faraday cup signal
Cell tune
17 mm
Tune scan takes ~ 2 hours
200 ms
[Taken on S-POD, Hiroshima]
-
05/11/2019 Lucy Martin 10
Future accelerators (Nonlinear Integrable Optics)
• In linear accelerators the motion is integrable – it is known to be bounded
• This is exactly what we want, the beam won’t be lost!
• Susceptible to resonances
• Realistically an accelerator can never be totally linear (errors + space charge)
• Nonlinearities -> no longer integrable due to coupling between x and y
• Require integrable system where small perturbations are allowed
• At Fermilab they found such a system2
• Unfortunately it requires a complicated potential
-
05/11/2019 Lucy Martin 11
Non-linear integrable optics
• Requires round beams and np phase advance• This is called a “T-insert”
• Quasi-integrable version involves only octupoles
• Octupole must vary in strength proportional to 1/b3
Image from [3]
To test in a Paul trap we require 2 things:1. To be able to create a good enough T-insert with the Paul trap2. Create correct octupole potential independent of linear focusing
-
05/11/2019 Lucy Martin 12
Why bother testing NIO in a Paul trap?• No dispersion
• No chromatic effects
• Can create a continuous octupole, or simulate a stepping field
• Octupole strength is easily variable
• T-insert parameters are easily variable
• Space charge effects easily included
• Can sit on resonance to study stability and excite resonances at arbitrary frequencies
• Already testing at IOTA and UMER – each facility has different advantages
-
05/11/2019 Lucy Martin 13
1. T-insert
Constraints:• np phase advance in x and y• bx = by in centre of drift• ax = ay = 0 in centre of drift
Solution (first attempt): • Simplify situation by only varying voltage pulse length, not amplitude• 6 constraints -> 6 voltage pulses• Basin hopping algorithm with low amplitude, 25V• At this point did not have a clear idea of the maximum allowable beta function
• Max bx and by• Phase advance close to 0.3 over drift• Bandwidth of amplifiers – pulses can’t be too short
or too large• Try to avoid too much asymmetry – more
susceptible to resonance at high intensity
-
05/11/2019 Lucy Martin 14
2. Quadrupole + Octupole trap• Building heavily on work from Hiroshima4
Vo
ltag
e (V
)
-
05/11/2019 Lucy Martin 15
Quality of octupole
Blue = perfect rod and plate placementRed = example with 20mm error on rod and
plate locationRmeas = 3mm (from simulation)
• Plot shows sum of the harmonic coefficients of higher order fields (from simulation) in an attempt to compare to IOTA and UMER magnets.
• For octupole only desired phase advance tolerance in IOTA7 is 2p * 10-2.
• Magnet tolerance is stated in [8] to be an integrated strength of harmonics relative to octupole field is less than 1% (100 units).
UMER: From [6]IOTA: From [8]
-
05/11/2019 Lucy Martin 16
Dynamic Ap
= 2.6 mm
Dynamic Ap
= 2.1 mm
• Dynamic aperture and
tune spread over 1000
periods.
• The T-insert is applied
using a single matrix
transformation.
• The colour bar shows
the log of the change in
tune over the simulation
for particles within the
dynamic aperture.
• Grey regions are
particles that survived
but are outside of the
dynamic aperture10.
Perfect
octupole
Realistic
Octupole
Log(Du
)
Log(Du
)
Log(Du
)
Log(Du
)
-
05/11/2019 Lucy Martin 17
Tested T-insert on IBEX in Jan 2019…
No IONs
• Trying to run before we can walk, this was before the installation of the MCP• Concluded that we probably were trapping ions but not enough to see• Received custom built amplifiers from Oxford electronics
• Improved band width – however only on central rods.• Installed the MCP, giving us:
• Improved calibration of trap • Ability to see down to 103 ions
-
05/11/2019 Lucy Martin 18
Simple simulationImages from MCP
(rotated through 45o as trap is 45 degrees from accelerator axis)
Cell tune = 0.1 Cell tune = 0.2
Cell tune = 0.3 Cell tune = 0.4
Cell tune = 0.1 Cell tune = 0.2
Cell tune = 0.3 Cell tune = 0.4
Varied the emittance in simulation, fitting to the rms beam size and matching the result to the images on the MCP
-
05/11/2019 Lucy Martin 19
1. T-insert, again• This time we varied pulse strength and length• Created a shorter T-insert that was easier to correct • Looked at 3 different T-inserts scaling voltage and length• To minimise beta function operated at lower tune (0.5 + 0.135)
• UMER also looking to operate at this phase advance6
• For lattice with larger beta function still saw no trapping• Decided to match into the T-insert region (automated)• Read waveforms from scope -> apply calibration factor and calculate optics -> call MAD-
X to calculate matching -> create waveforms to be applied -> send waveforms to AWGs
Length2 = Length1 * 1
𝑘
Voltage2 = Voltage1 * k
-
05/11/2019 Lucy Martin 20
• For the waveforms with larger voltages ran into problems with the amplifiers, especially those on the endcaps, with the lower band width:
• Decided to not apply the alternating voltage to the end caps and to match back to original waveform before extracting:
Central rods
End caps
-
05/11/2019 Lucy Martin 21B
eta
funct
ion (
m)
0 3.02.01.0
Time (ms)
Designed T-insert, MAD-X:
Best beta function reconstruction:
For the shortest (highest voltage/ lowest beta function) lattice studied the optics are reconstructed by taking each voltage point from an oscilloscope trace and
transforming it into a transfer matrix:
-
05/11/2019 Lucy Martin 22
• For lower beta functions more ions were stored• Matching reduced ion loss
-
05/11/2019 Lucy Martin 23
Testing the T-insert• How can we prove that we have the correct phase advance?• How can we prove that we have the correct tune?
Cartoon showing the perturbation applied to the rods (perturbation voltage enlarged)
• Method using a dipole kick5
• Potential to use high speed camera for time resolved imaging
Quadrupole perturbation: See tune
control of ~10-2
-
05/11/2019 Lucy Martin 24
From [8]
• The upper estimate that we have on emittance is 2.16*10-9.
• The largest beta function in the region where the octupole
will be on is 229.89.
• Using the emittance estimate this gives 0. 70467mm as an
rms beam size, v. small due to small beta function.
• Can then use 2 * this as reference radius = 1.4093 mm
Octupole quality
-
05/11/2019 Lucy Martin 25
Simulation of new setup
• Want to vary octupole strength in simulation - can also do this within reasonable bounds experimentally.
• Further simulations of the new T-insert required• Full lattice• Space charge
-
05/11/2019 Lucy Martin 26
Conclusions and further work• IBEX is a useful tool for accelerator physics studies
• Commission of IBEX has ended and we’re now capable of a range of interesting physics
• To test quasi-NIO a number of often competing constraints should be met
• However, it should be possible.
• Further simulations of the new T-insert required• Full lattice• Space charge
• Octupole upgrade to the trap needed – must be well designed
• Experimental testing of quasi integrable NIO
• Potential to vary a wide range of experimental parameters.
-
05/11/2019 Lucy Martin 27
Thank you!
-
05/11/2019 Lucy Martin 28
References1. Martin, L. K. et al., “Can a Paul Ion Trap Be Used to Investigate Nonlinear asi-Integrable Optics?”, 2019, J. Phys.: Conf. Ser. (In Press)
2. Danilov V and Nagaitsev S, “Nonlinear accelerator lattices with one and two analytic invariants”, PRAB, 2010
3. Antipov S et al, JINST 12 P04008, 2017
4. FUKUSHIMA K and OKAMOTO H, “Design Study of a Multipole Ion Trap for Beam Physics Applications”, Volume 10, 1401081, 2015
5. Martin, L. K. et al., “A new method to measure the beta function in a Paul trap”, 2018, J. Phys.: Conf. Ser. 1067 062016
6. Ruisard, K. et al, “Single-invariant nonlinear optics for a small electron recirculator”, 2019, Phys. Rev. Accel. Beams, 10.1103/PhysRevAccelBeams.22.041601
7. S. Antipov, S. Nagaitsev, and A. Valishev, “Single-particle dynamics in a nonlinear accelerator lattice: attaining a large tune spread with octupoles in IOTA”, 2017, J. Instrum. 12, P04008
8. Martin, L. K. et al., “A study of coherent and incoherent resonances in high intensity beams using a linear Paul trap” New J. Phys. 21 053023
-
05/11/2019 Lucy Martin 29
Matching code:
First estimate of matching waveforms applied to the rods
Waveforms from oscilloscopes saved
Code analyses save waveforms calculating the
optics from each
Twiss parameters at the end of the square wave
section and start of the T-insert are found
Start and end parameters for the match are written
into a MADX file
MADX run from python code to calculate correct
matching
Voltage pulse heights extracted from MADX
code
New waveforms to be applied to the rods are created with the new
matching section
New waveforms are applied to the rods via the Labview (requires human
input)
Waveforms from oscilloscope are saved
-
05/11/2019 Lucy Martin 30
-
05/11/2019 Lucy Martin 31