Presentation on theme: "Femtosecond Pump / Probe Operation and Plans at the LCLS"— Presentation transcript:
1Femtosecond Pump / Probe Operation and Plans at the LCLS Josef Frisch for the LCLS Commissioning team
2Ultra-Fast ScienceSome experiments use multiple images on an already evolving systemAll feet off the groundMost experiments are pump probe:Stimulate the system (fire bullet)WaitMeasure with a probe pulse (flash bulb)Measurement resolution is set by the length of the pump / probe pulses AND the accuracy of the time delay between the pump and probe.H. Edgerton
3Short Bunch Operation at LCLS Low charge (20pC) operating mode for very short pulsesNo direct pulse length measurement available, but believed to be < 5fs FWHMPhase = +1 degPhase =+0.5 degPhase = 0 degΔT=5.0fsPhase = -0.5 degΔT=2.3fsPhase = -1 degGenesis Simulation for over compression: 5fs FWHMΔT=1.1fsΔT=1.9fsΔT=4.2fs we typically operate here
4Narrow or Double X-Ray Pulses from a Slotted Foil PRL 92, (2004).P. Emma, M. Cornacchia, K. Bane, Z. Huang, H. Schlarb (DESY), G. Stupakov, D. Walz0-6 mm0.25 mmpulses not coherenttime (fs)Power (GW)1050-150 fs2 fs
5Low Charge AND Slotted Foil X-ray spectrum with 20pc operation – few spikes suggest ~5 fs pulsesWith 20pc and slotted foil see single spike spectrum suggests very short pulsesNo direct measurement but LCLS may be producing ~1fs X-ray pulses
6Short Pulse LasersCommercial Ti:Sapphire lasers can produce pulses as short as 15fs. (25-50fs more typical).High harmonic generation can produce ~100aS, pulses in the XUV ~100eV.Assume lasers will produce shorter pulses in the futureAttosecond XUV generation, Max-Plank_institut fur Quantemoptik / ATLAS
7Experiment Requirements This talk will concentrate on laser pump / X-ray probe experimentsMost common experiment at LCLSRight now operating with ~ fs X-ray pulses and ~50fs laser pulseIn the future we expect few-fs X-rays and few-fs laser pulsesTiming control at the few fs level will be required.Typical temperature coeficient for either coaxial cables or fiber optics is 2x10-5/C° ->1 meter is 60 femtoseconds / C°Thickness of a sheet of paper = 100fsWhen describing timing drift or jitter, need to be careful to clarify what reference is used for comparison.
8Experiment Requirements Pump LaserX-raybeamX-rays todetectorSystem evolves from pump to probe timeIdeally would scan time differenceGenerally OK to let jitter vary the timing and measure shot-to-shotTiming jitter relative to an external clock isn’t important
9Sources of Timing Jitter LaserGunRF off crestBunch CompressorRFRFLaser pulse is compressed typically 2X in gun, then an additional factor of 100 in the bunch compressorsChanges in laser time are compressed, so gun laser jitter is not very important. Beam time is mostly set by the RF in the compression system. (both amplitude and phase contribute)Synchronizing the gun laser to the experiment laser doesn’t fix the jitter
10Conventional Timing System StabilizedtransmitterStabilizedReceiverFemtosecond LaserLaser AmplifierX-raysUndulatorBeam time pickupExperimentE-beamdump~100MBeam pickup typically responds to electric field of bunch: either RF cavities or electro-optical pickups are usedStabilization system typically feeds back on the length of the cable / fiber.
11Timing Jitter in LCLS ~ 1km Master Source Phase Shift Stabilized link ~20fs stability10ps drift over hours10fs jitter, 50fs stabilityFew fs jitter in a few msHigh power RFPhase DetectLaser~50fs RMS jitter shot to shot~50fs jitterFeedbackRF in compressor sets beam timeAcceleratorPhase cavityFELExperimentExperiment data corrected offline with phase cavity data report fs stability
12Beam arrival time cavity (LCLS) Similar to a cavity BPM but use the monopole modePhase drift from cavity temperature is the most significant problem1us time constant, 10-5 /C° temperature coefficient -> 10ps/C° (!)Raw SignalPhase slope gives cavity temperatureRMS difference between cavities ~12 femtoseconds RMS at 250pC, 25 femtoseconds at 20pC. Drift is ~100 femtoseconds p-p over 1 day.
13RF Phase Detection Limits Oscillators: unlocked timing noise relative to an “ideal” clock increases with timeConventional oscillators: 1fs RMS above 1 KHzSapphire oscillators 1fs RMS above 10HzRF phase measurement (2X thermal noise)1GHz, 1ms, 1mW power -> 20aS (theoretical)SLAC summer students actually measured a noise level corresponding to 30aS in a 1KHz bandwidthIn a 1MHz bandwidth, still expect 1 fs.Phase cavity system noise is about 7fs RMS. (best conditions)Electronics noise is not a stringent limit!Drift: few fs / °C for mixers.Drift: ~30fs / °C for 1 M cable.
14EO Beam Time Measurement (Several versions, simplified concept shown)Short pulse laserFree space or fiber-opticDetectorOutput intensity depends on relative timing of laser plulse and E-beamBunch fieldsF. Loehl et alDESY/FLASHElectric field from bunch6 femtosecond timing noise published(Believe ~3 fs achieved)Electro-optical intensity modulationAllows direct conversion from beam timing to optical signal: significant advantage for some types of timing systems
15Long Distance Timing Transmission Adjust DelayTransmitterUse fibers:Low loss as high transmitter frequencyGood directional couplersLow costmirrorTiming SignalFeedbackCompare forward and reflected signalsEnvelope scheme (DESY, MIT Bates):Transmit short (ps) pulses at ~100MHz rate.Timing of the reflected pulses is used to measure the fiber length.Control fiber length with feedbackPulses detected at the receiver end are used for timingPulses allow direct locking to experiment laserExcellent resolution – based on optical wavelengthDifference between phase and group velocity is important an must be compensatedCarrier scheme (LBNL, used at LCLS)Frequency stabilized laser used in an interferometerInterferometer determines fiber lengthControl fiber length with feedback (feed forward in this case).Both systems work at <20fs over 100M fibers
16Optical to Electronic Conversion Even with perfect fiber stabilization systems, this can be the performance limit.Photo-diodes: Tradeoff between noise and linearityNonlinearity: Charge extraction -> changes bias voltage -> changes capacitance -> changes phase delayHigh frequency diodes have small area, low capacitance.For S-band (3GHz) diode -> 150fs single shot resolutionFor X-band (12 GHz) diode -> 60fs single shotFor high repetition rate systems (oscillators) this isn’t too bad: 68MHz, 100us TC -> 1fs (ideal)For amplifiers, this is a large problem – single shot measurements are very difficult.Can in principal use an optical resonant cavity (etalon) to average signals. For Q = 100 -> ~10fsOther techniques have been developed for fiber based systems: Rely on electro-optical mixing between laser and RF signal.
17Laser StabilizationConventional Ti:Sapphire laser oscillators can be locked to ~50fs to a RF reference.Several limitations:Phase detection from photodiodesAcoustic noise changing the cavity lengthPump laser fluctuations change the effective cavity length through nonlinearitiesLaser chirp pulse amplifier system can add jitterWavelength changes can change the delay through the compressors (if the wavelength response of the amplifier isn’t flat)Pulse shape changes with laser power from changes in amplifier saturationVery active area of research both at labs and in industry.At least at LCLS this is the limit to stability.The pulsed DESY / FLASH system allows direct optical cross correlation between the experiment laser and the timing system!(A. Winter et al).DESY optical master oscillator
18Superconducting vs RT Accelerators The beam timing jitter relative to the accelerator timing reference system is similar for room temperature and superconducting accelerators: 30-50fs RMS.FeedbackCompressorRF StructureRF StructureBeam time pickupGunIn an superconducing accelerator the beam timing can be measured for each pulse at the ~MHz beam rate, much faster than the typical 100us energy storage time in the accelerator cavitiesThis allows the use of a fast timing feedback to reduce the timing jitter.
19Other Limits Ground Motion SASE process Tidal stretching is 30um / kilometer. (100fs/km)In principal predictable, but in practice trickyFast ground motion varies with location.Measured at SLAC as 10s of nanometers over 14 M separation.Needs more studySASE processStatistical fluctuations give a minimum timing jitter of [(1/12)rL]1/2 with r the slippage distance and L the bunch length.If only part of the bunch lasers, X-ray time will not match electron beam time.Location of experimental IP (1 um -> 3fs)Looks difficult to reach 1fs even if the individual technical system problems can be resolved.Tides observed in LEP frequency corresponding to ~2x10-8(L. Araudon et al, CERN SL/94-07)
20Optical / X-ray Cross Correlator X-raysReflected optical beam measured on array sensorLaserGaAs or similarTests at SXR (W. Schlotter et al) have demonstrated <60fs RMS (consistent with 0) single shot X-ray to laser optical timing measurement.Note that electronic timing will still be needed for “crude” 100fs timing
22Cross CorrelationVarious physics is available, but need to find a way to operate over the full wavelength range and with femtosecond resolutionKeVOperate at few uJ pulse energies (1fs operation)Final version should do cross correlation in the experimental chamber1fs is 300nm, very difficult to control long lengths at this level.Need to find appropriate physics to use for thisMay need an XFEL to study this physics!
23THz Timing Experiments THz pump / X-ray probeThe high peak current beams used for XFELs can also serve as sources of very intense THz radiationThis radiation is precisely timed to the electron beam.Unfortunately since the beams are ultra-relativistic the THz can never “catch” the X-raysTHz delayed relative to X-rays. Need to use 2 bunches, one generates THz, second X-rays.THzX-raysFELFor hard X-rays can use crystals to delay to match the THzTiming error limited by mechanical stabilityTHzX-raysFELPlans to test both schemes at SLAC / LCLS.
24Future Timing Systems“Conventional” systems presently have fs rms timing resolutionCan probably extend to ~10-30fs RMSConventional lasers now produce <25fs pulses, with ~100as available from XUV lasers.XFELS at <10fs, with <1fs likely in the near future.For single femtosecond timing will need new approaches like direct X-ray / optical cross correlation.