Goal & Challenge:

We want to use SLM for Laser pulse shaping and controlling of the emitted electron bunch (Distribution, Size...). 

SLM doesn't work with UV so it has to be installed at the beginning parts of the optical table → many non-linear elements (THG, quart rod) afterwards can distort/destroy the desired shape. So we need a feedback control.


  • A Master student, Carl Sax, is working on the SLM test setup.
  • SLM works now with native Hamamatsu-controller and a red laser.
  • Carl starts implementing an in-house controller (python, expects to be EPIC compatible) & algorithm for generating hologram
  • Virtual Cathode: still problem with YAG-screen, will get to it soon ( when Michael has time & Carl has time)

Next Steps:

  • Install a camera for the test setup
  • Use the camera signal as feedback and use (persumably) a NN-algorithm to iteratively improve the image
  • Add some non-linear elements to test its capability
  • If succeeded, install the SLM on the optical table & use the virtual cathode as feedback signal

Some useful Information

Current works on SLM (
Basically they use NN-algorithm & camera feedback to train and iteratively improve the SLM performance.
Paper (to be published Dec. 2020)  and source code 

A KIT logo generated using red laser & SLM.

  • Manuel Kirchen: Prediction of the electron beam quality in a laser-plasma accelerator
  • Jun Zhu: Application of multi-task deep learning at the injector of European XFEL

Flute Laser Confluence Page:

  • The laser is in a cleanroom. Geared to industrial use. 
  • Oscillator: repetition rate of 83 MHz (short pulses, broadband). 800 nm = central wavelength. Gives the timing for synchronization for the machine. 
    • The laser is reflected between two mirrors that give the timing.
    • Pulses stretched in time. Chirped pulse amplification → short pulses with a lot of power: stretch them (200 ps),  amplifying them (1 kHz), compress them again (35 fs, 6 W).
    • Difficult to transport high power density pulses, so the compressor is not used. The stretched beam is sent.


  • Virtual Cathode: for diagnostics (pulse shape...), camera
  • ICT (Integrated Current Transformer): interesting signal.
  • BPM (some problems need to be solved, not in the control system).

Tuning knobs

  • Motorized beam splitter (rather slow, not thought for feedback, PV)
  • Focusing lens onto the cathode (almost in the control system)
  • Pulse picker
  • Laser stabilization system (spatial direction and angle)
  • Compressor (not in the control system, but it is remotely controllable (35 fs 10 pc)
  • Timing not in the control system (ask Nigel, Thiemo)

It is difficult to change the pulse length as of now (non-linear behavior in THG). We lose efficiency in the UV conversion.

Future work

From March 2021 with Matthias (Ph.D. student):

  • Spatial light modulators: (like a liquid crystal display), Liquid Crystal on Silicon (~1000x1000 pixels, electronics for each pixel). You apply voltage to each pixel, so you can change the amplitude/phase of the light → manipulate your laser beam. Installed in the optics table, but first, try it out in a test set-up to learn how to program it, set-up the software. Time estimation: ~1 year.
  • Could be able to generate bunch trains.

From November 2020 with Carl (master student):

  • Set-up of the virtual cathode.
  • Set-up of photodiodes. Very sensitive, the electronics are not fast enough to resolve the signal at short pulses. The amplitude changes depending on how you sample it. Thiemo switches the bias off, but you lose a lot of sensitivity.
  • Also include the manipulation with magnets.

To Do

  • Aims of FLUTE: smallest bunch length, smallest bunch size overall, high bunch charge.
  • Start to end simulations:
    • prefer to have it all in one simulation code (ocelot).
    • includes transfer lines.
  • SRR: limited by the bunch charge and bunch length. It's a destructive technique. Idea: try to provide small beam sizes for the SRR → We can use much shorter laser pulses (fs range).
  • EO setup:
    • Not easy to install before the linac. Some limitations about why it is not suitable for that energy: at lower energy the coulomb fields are not relativistically compressed, so you lose resolution.
    • Ask Gudrun & Erik if it makes sense? A student already worked on this.
    • Electron bunches are long now (3-5 ps). Single-shot measurements. 50-100 fs.
  • There is a bunch creation routine in ASTRA that models laser-cathode interaction. The routine is working and reliable. There's an additional program that you run apart from  ASTRA (increasing complexity). The space charge part of OCELOT should be the same as in ASTRA.
  • Interaction of laser with the cathode: very big enterprise.
  • Meeting with Thiemo and Michael: for laser properties.
  • Nigel and Dima: diagnostics.
    • Possibilities to install new diagnostics in Spectrometer arm?
    • Transverse properties: button BPMs (will be installed)
  • Summary of old measurements (EO Simulations): (20% error at 7 MeV)
  • Beam starting from the cathode has hot spots (surface roughness).

  File Modified
PDF File 2020-10-02_FLUTE_Meeting.pdf Oct 02, 2020 16:01 by Andrea Santamaria Garcia