- Slides: 32
Radiation Sources: Horses for Courses Graeme Hirst STFC Central Laser Facility Cockcroft Institute Laser Lectures July 2011
Lecture 2 Plan • The big picture • The issues • Comparisons • Wrap-up
The Stakeholders Source providers Decision makers Users
The Options Thermal Plasmas Electron impact Sunlight “Electrons in flight” Lasers
Brainstorm - What Matters to Users ? TECHNICAL NON-TECHNICAL • Spectrum • Access • Pulse length & timing • Flexibility • Transverse beam quality • Reliability • Polarisation • Support & infrastructure • Future-proofing • Cost
Health Warning ! Light sources occupy a parameter space with many dimensions Sources and user requirements are often quite “localised” within this parameter space (the degree of overlap is, perhaps, surprising !) Sources which can deliver unusual performance in more than one dimension simultaneously are very much the exception, not the rule !
Photon Energy Range BENDING MAGNETS: Output Coherent enhancement if bunch length < wavelength THz Photon Energy X-ray
Photon Energy Range UNDULATORS: Output Single ID tuning envelope Fundamental tuned by B-field and/or electron energy Harmonics (odd, on axis) THz Photon Energy X-ray
Photon Energy Range FELS: Output Tuning envelope Fundamental tuned by B-field and/or electron energy Harmonics THz Photon Energy X-ray
Photon Energy Range Dyes 0. 01 0. 1 F 2 X-ray Lanthanide lasers ions Excimers Nd Ti Molecular gas masers CO 2 CO Output LASERS: 1 10 Photon Energy (e. V) 100 (Energies in the FIR and X-ray are indicative only) 1000
Photon Energy Range Extension LASERS: Nonlinear optics (frequency multiplying, sum- and difference-frequency generation, optical parametric processes) can shift NIR laser photons to give complete coverage from <0. 1 e. V to >6 e. V (limited by crystal opacity) with reasonable efficiency Harmonic generation gives complete coverage from 10 e. V to several hundred e. V with 10 -5 - 10 -8 efficiency Plasma conversion can give incoherent broadband output from ~100 e. V to tens of ke. V and Ka emission from ~2 ke. V to many tens of ke. V Laser particle acceleration promises X-rays from IDs and e. g. betatron radiation (they’re not lasers though)
Photon Energy Tuning/Scanning BENDING MAGNETS: Tuning can be by a dichroic mirror/monochromator or a dispersive detector, which are easy but often inefficient, or (in the IR) by a Fourier Transform spectrometer. UNDULATORS AND FELS: Tuning is usually by changing the gap which is relatively easy for undulators but gets increasingly difficult for FELs as the photon energy rises. An order of magnitude range is possible. LASERS: Commercial Ti: S and dye lasers and optical parametric systems are tuned by optic alignment, which can be automated. An order of magnitude range is possible.
Spectral Linewidth Df can never be less than ~1/Dt (pulsewidth) This is the (Fourier) transform limit The exact constant of proportionality depends on pulse shape and on how “width” is defined LASERS: the beams are often “near transform limited” which means they are longitudinally (temporally) coherent. Spectral amplitude and phase can be stabilised. Longitudinal coherence can cause “speckle”. ACCELERATOR BASED SOURCES: the beams are usually not transform limited, the exceptions being cavity FELs, seeded FELs and spectrally filtered beams. Beams which are not transform limited are not temporally smooth.
Temporal Pulse Profiles Simulations from DESY Temporal profile of a SASE FEL pulse generated by an electron bunch with s = 250 fs
Photon Energy Range - Summary ACCELERATOR BASED SOURCES: cover the widest spectral range from <0. 01 e. V to >100 ke. V. In principle tuning can be continuous. In practice the tuning range depends on the source and the optical transport system. In general the spectral linewidth is broader than the transform limit. This may be significant for some experiments (high resolution spectroscopy, coherent control etc). LASERS: cover the spectral range from <0. 1 e. V to ~6 e. V with continuous tuning using nonlinear optics in crystals. Harmonic generation can extend this to several hundred e. V. In general the spectral linewidth can be near to the transform limit.
Pulse Duration Among others, this matters a) To users who need high peak power or intensity b) To users who want to do time-resolved studies c) To users whose sample explodes when irradiated (These seem to represent a large fraction of laser users but only a small fraction of synchrotron users) Important aside: Laser people would call this a 235 fs 100 235 pulse (FWHM). Accelerator people would call it a 100 fs pulse (s). Laser people would call this one a 50 fs pulse on a pedestal
Pulse Duration BMs AND UNDULATORS: The pulse duration essentially reflects the electron bunch length. In storage rings this can be as short as 10 ps and in linacs less than 100 fs. Exotic techniques (laser slicing, crab cavities, slits in chicanes …) can allow further shortening. FELS: FEL gain is sensitive to peak current. Slippage can lengthen pulses, superradiance can shorten them, as can seeding. ~10 fs pulses are possible and exotic schemes promise lower. 30 fs FWHM Simulated output from saturated cavity FEL driven by 250 fs FWHM electron bunch
Pulse Duration LASERS: Systems working from cw to <10 fs FWHM are commercially available, as is the technology for CEP stabilisation. Harmonic techniques can deliver “attosecond” pulses. Short pulses are measured using correlation techniques. Below 10 fs it has been said that measuring the pulses is more difficult than generating them ! “Grenouilles” from Swamp Optics Inc for measuring ultra short laser pulses using FROG Frequency Resolved Optical Gating is one of a number of “amphibian” measurement techniques (TOAD, TADPOLE, SPIDER …)
Timing and Synchronisation Multi-beam experiments usually require the timing of the different pulses to be managed. In 10 fs light travels just 3 mm. In some cases “management” can just be measurement and the experimental data can be time-stamped. LASERS: Laser oscillator pulses can be synchronised to one another and to RF references with ~20 fs rms jitter in experimental laboratories and sub-fs jitter in standards labs. ACCELERATORS: Accelerators are large and many beam transport elements are dispersive. This makes bunch timing control difficult. Stabilisation and feedback control at the ~10 fs level are very active research areas.
Pulse Rates and Time Structures These matter if they a) affect average power b) interrupt relaxation/recovery of sample and/or diagnostic c) are seriously mismatched in multi-beam set-ups EO and AO choppers work for NIR/vis/UV photon energies and mechanical ones work into the X-ray (on MHz beams !) LASERS: Oscillator pulse rates vary from tens of MHz to GHz and are set by optical cavity length. The time structure is inherently uniform. Amplifiers can be pulsed from single-shot to hundreds of k. Hz or cw. But matching to few hundred MHz accelerator rates is rare.
Pulse Rates and Time Structures ACCELERATORS: Maximum pulse rates are set by the machine RF, typically hundreds of MHz to few GHz (30 GHz for CLIC !) Not every RF bucket needs to have electrons in, but gaps reduce current and may affect stability Secondary time structures are imposed by orbit length (storage rings) and macrobunching to limit average power (linacs) On multi-user machines there can be conflict over time-structures (multi-bunch vs single-bunch) FEL pulse rates can be limited by optical power handling
Average Power Average power = Pulse energy × Pulse rate It is important for: Improving signal-to-noise Acquiring data before a sample degrades Acquiring data before beamtime runs out Supplying multiple experiments It may be limited by: Drive capacity and conversion efficiency to light Nonlinear effects in the light source (due to peak power ? ) Heat removal from the light source Damage/distortion to the transport optics Damage/disruption to the sample and/or detectors Controlling beam power without affecting other parameters can be surprisingly difficult.
Average Power LASERS: Commercial “research” lasers tend to be limited to a few tens of watts if tuneable short pulses are required. Over the next few years lasers in the 100 W-1 k. W class are expected to be developed and may be commercialised. HHG conversion at 10 -6 efficiency promises m. W powers. ACCELERATORS: High current machines (storage rings, ERLs etc) should be capable of watts or tens of watts from undulators. A 14 k. W IR FEL has been built and a 100 W blue/100 m. W VUV FEL is being commissioned (Jlab IR and UV FELs). Linacs’ inherent inefficiency reduces average powers by orders of magnitude.
Average Power 4 GLS IR FEL Data from 4 GLS CDR 4 GLS VUV FEL Commercial 1 k. Hz OPA (ph/s) 4 GLS Bending Magnet 10 k. Hz OPAs are becoming commercially available 4 GLS XUV FEL Diamond undulators HU 64 U 30, 10 m U 23 4 GLS undulators U 60, 5 m Max III U 62. 5 4 GLS 3. 5 T wiggler Commercial lasers compare well with high current accelerators.
Transverse Beam Quality Beam focusing is important for intensity f and for microscopy/small samples d Without tricks, focal spot diameter f cannot be less than ~fl/d f This “diffraction limit” requires full transverse coherence LASERS: the beams are often “near diffraction limited”. ACCELERATOR BASED SOURCES: can be effectively “point like”. Long, thin undulators can act as spatial filters, imposing transverse coherence. Positional stability may be an issue. Transport optics can distort under high heat load. Achieving adequate surface figure is hard at short wavelength.
Polarisation LASERS: Oscillators and nonlinear optical converters are generally linearly polarised. Over the photon energy range typical of lasers polarisation can easily be adjusted by passive optics. Harmonic generation is inherently linearly polarised. BENDING MAGNETS: Output is inherently linearly polarised. UNDULATORS AND FELS: Polarisation can be controlled by using an APPLE helical undulator section to generate linear, elliptical or circular polarisation.
Source Combinations These can be considered in (at least) two ways: • A composite light source could be produced using lasers to cover, say, the photon energy range 0. 1 e. V-6 e. V and accelerator-based sources for higher and lower energies • Users should be unaware of the source of their photons (concentration on similarities) Alternatively: • A laser could be used to do something for which it is uniquely suited – e. g. create a warm-dense-matter plasma or a Bose-Einstein condensate • An accelerator-based source could then probe the result (concentration on differences)
Summary - Source Strengths LASERS: • Can deliver high energy pulses • Can deliver sub-femtosecond pulses • Can be very tightly controlled (spectral, spatial and temporal phases and amplitudes, time structures and synchronisation) • Are relatively cheap and quick to produce, so the field develops rapidly and small groups can build custom systems ACCELERATORS: • • Can reach the spectral extremes with high average powers Can operate at high photon energy with variable polarisation Can support multiple users Are large enough to justify the provision of substantial supporting infrastructure
Wrap-Up • The choice between a laser and an accelerator-based light source will depend on the application and on the user’s preferred “way of working” as much as on the sources’ technical performance • Comparisons are complicated by the very significant difference in investment to date between a typical laser source (few £M) and a typical accelerator-based source (few hundred £M) “. . . ” Report from the Review of UK Light Sources Jan 2008
Thank you !
An Example – EXAFS and XANES Extended X-ray Absorption Fine Structure (EXAFS) and X-ray Absorption Near-Edge Structure (XANES) “simply” involve measuring and analysing X-ray absorption around elemental edges Choice of the edge allows a particular element to be studied The details of the absorption fine structure reveal information about the atom’s local environment Absorption features are often weak
EXAFS Requirements Wikipedia says: “Since EXAFS requires a tunable X-ray source, data are always collected at synchrotrons. . . ” most often Edge energy: EXAFS is possible from <1 ke. V to ~35 ke. V but 3 -10 ke. V covers many interesting elements Spectrum: continuous and smooth over 1 ke. V Signal resolution: DI/I of 10 -4 as a design target