Accelerator Test Facility

The ATF electron gun is a resonant pi-mode 1-1/2 cell cavity operating at 2856 MHz. It uses a metal photocathode that forms part of the wall of the first 1/2 cell. The 4.5-MeV electrons exiting the gun are accelerated in two SLAC sections to 70 MeV for performing users’ experiments. There are three beam lines presently dedicated to the Compton backscattering experiment, the high-gain harmonic generation experiment and the self-amplified spontaneous emission experiment. The nominal electron beam requirements for these experiments are 2-4 mm-mrad normalized rms emittance at 100-180 A peak current with good stability.

The laser used in illuminating the photocathode is a Nd:YAG modelocked laser with a four-pass preamplifier and a two-pass amplifier located in a Class-100 clean room. The high-power 1.06 µm beam is split into two beams, one high-energy pulse (20 mJ) to irradiate a germanium switch for slicing a 50-ps CO₂ laser pulse, and one lower-energy pulse (1 to 2 mJ) for irradiating the photocathode. The photocathode infrared laser beam is frequency quadrupled and then directed to the "gun hutch" where optics are used to shape the laser beam for oblique incidence on the photocathode. Currently, magnesium is used as the photocathode in the ATF gun because it has reasonably high quantum efficiency (~5 x 10^-4) and low susceptibility to poor vacuum and arcing. The magnesium cathode surface is pretreated to achieve the high quantum efficiency.

Before the drive laser was relocated, the ATF gun had produced a whole-beam emittance of 2.5 mm-mrad at 1 nC, with a slice emittance of 1.4 mm-mrad. This performance has not been reproduced since the laser was moved to the new location. Below are the problems that appeared after the laser was moved:

Poor emittance, e.g. 8 mm-mrad at bunch charge less than 1 nC

Possible cause: Poor spatial beam of drive laser. This problem was fixed after the laser engineer and another colleague discovered a small leak from the cooling line and coolant on part of the amplifier rod end face. They also found that the fourth harmonic generating crystal had poor optical quality. However, the emittance remained high after these problems were fixed.

The ATF electron beam diagnostics indicate the existence of two separate electron beams. These two electron beam are separated in time by ~5-20 ps (deduced by changing the injection phase and observing the appearance and disappearance of these two pulses)

Possible cause: A spurious reflection off a 1-3 mm thick piece of optics

Drift in beam centroid over time. The ATF laser engineer has to realign the laser frequently to keep the electron beam exiting the gun at the same location.

Possible cause: The gun hutch and drive laser are on different slabs and thermal drift causes the laser beam to move off its align position.

Recommendations for the Short Term

The prime suspect for the spurious reflection is the gun laser entrance window. The Panel urge ATF personnel to measure the reflected UV from the gun entrance window. This reflection should be less than 8% of the incident beam energy. If it is more than 8%, replace it with an AR V-coated window immediately. Also check for reflections from any piece of optics that is 2-3 mm thick.
Improve the air conditioning system for better temperature and humidity control. If possible, also reduce vibration and air currents in the drive laser room.
Check for drift between drive laser room and gun hutch using a test HeNe laser. We recommend the ATF team set up a far-field diagnostic camera for pointing stabilization along with the existing near field diagnostic (i.e., beam image) camera. Acquire a commercially available beam pointing stabilizer to correct for the drift.
Check with Lightwave Electronics about the photodiode voltage for the phase-lock loop. One of the Panel members recalled it should be between 1 and 2 volts. Currently it is 0.5 volt.
Verify the output wavelength of the Time-bandwidth Nd:YVO₄ oscillator matches with the Nd:YAG line at 1.064 µm.
Move the Lasermetric Pockels cell to upstream of the second harmonic crystal.

Recommendations for the Intermediate Term

Decouple the two functions of the drive laser, i.e. pulse slicing and photocathode irradiation, by splitting the laser beam early, before the amplifiers. Use the multipass amplifier for the pulse slicer beam; use a regenerative amplifier and maybe another double-pass amplifier for the photocathode beam.
Redesign the optical train so that pulse splitting occurs earlier in the chain, preferably right after the oscillator, to minimize the numbers of Pockels cells. This should reduce optical distortion that leads to poor beam quality.
Replace KD*P with LBO as a SHG crystal.
Put Brewster angle on the exit window of the evacuated spatial filter. Care should be taken to avoid astigmatism, since the beam is expanding through this window, having come from a focus in the evacuated tube. An additional entrance Brewster window with normal at 90° to the exit window would compensate the astigmatism.
Explore normal incidence of the UV beam onto the photocathode to replace the existing oblique incidence optics. Study the issues associated with normal incidence such as mirror degradation and plan to address these problems early on.
ATF should work on generating a uniform spatial profile in the regenerative amplifier and Fourier relaying that image onto the SHG crystals. Random phase plates or Gaussian mirrors should be explored as possible spatial profile beam shapers.

Recommendations for the Future

Based on the specifications provided by the ATF personnel, the Panel believes the baseline for the ATF photocathode drive laser oscillator should be a modelocked Ti:sapphire laser followed by a 1-KHz, 2-mJ/pulse Ti:sapphire regenerative amplifier that is pumped by a frequency-doubled pulsed Nd:YAG laser. The latter could be pumped by either cw semiconductor diode lasers or arc lamps. The specified output energy at 800 nm is 2 mJ/pulse with an rms amplitude jitter of 1%. This specification, though aggressive, is within the capability of commercial lasers. The output repetition rate can be divided down from the 1-KHz repetition rate by switching the Pockels cell inside the regenerative amplifier at a reduced frequency. If more energy is needed, a Ti:Sapphire amplifier pumped by a frequency-doubled diode-array-pumped Nd laser may be added. This system is identical for use in a high-rep rate, low energy applications or low-rep rate, high-energy applications. The risk involved in using the additional amplifier stage is an increase in pulse-to-pulse amplitude fluctuations.

Photon Energy

Based on the required photon energy of 3.5 – 4.6 eV, the laser beam at 800 nm will have to be frequency tripled to obtain 266 nm light.

Bandwidth
The design pulse shape for the future drive laser has 100-500 fs rise and fall time and pulse width of 5-10 ps, with the ability to program the laser intensity at the edges and in the middle of the pulse. For example, the laser pulse may have a small dip in the center and variable intensity at the front and back of the pulse.
Strategies: The strategy that the panel adopted was pulse stacking with multiple short (100-200 fs) pulses, either with a Michelson interferometer or preferably by a Fourier spectral mask technique. The latter technique would require a pair of gratings sandwiching a Fourier mask to produce a train of short pulses. This should be done upstream of or inside the regenerative amplifier so that any loss in optical power due to pulse shaping can be recovered by the regenerative amplifier or any downstream amplifiers. The Panel also expected that the short (<200 fs) electron pulses would stretch out and fill in the space between adjacent pulses to provide a temporal square pulse (or one with a dip in the middle). The Panel discussed the following strategies but discounted them for the reasons shown.
- Shaping square pulses with spectral mask: Too much energy under the main spectral feature can cause self-focusing in the amplifiers.
- Partial recompression: sensitivity to input intensity and complication in the nonlinear frequency conversion. Also, this method offers no fine control of the leading and trailing edges.
- Shaping with temporal electro-optics: The output pulse is not transform limited (too much bandwidth).
- Gating with semiconductor gates: Necessity for another 100 fs laser
- Self-phase modulation and dispersion to generate square pulses: Possibility of poor transverse profile and lack of flexibility.
- Saturable absorbers: This method is deemed unreliable and offers no control for the middle part of the pulse.
The Panel has a genuine concerns regarding the diagnostic capability to measure the short ultraviolet pulse with 100-fs resolution. ATF needs to develop a diagnostic technique with <200 fs resolution for the UV pulses. Possible methods include sub-picosecond streak camera, Kerr effect, and two-photon cross-correlation (Check out publications by A. L. Gaeta, Cornell. For example: "Femtosecond Ultraviolet Autocorrelation Measurements based on 2-photon Conductivity in Fused Silica", Optics Letters, 23, 798 (1998)).
Uniform transverse profile (10-20% amplitude variation)

The best approach is to employ a supergaussian mirror in the regenerative amplifier. Fourier relay imaging should be implemented to relay the supergaussian mirror onto the SHG crystal, from the SHG crystal to the THG crystal, and from the THG crystal to the photocathode.

Laser pulse energy (Required 20 µJ to obtain 1 nC with oblique incidence)
ATF personnel estimated that with normal incidence the pulse energy required is 50 µJ for Mg and 200 µJ for Cu. Since there is no plan for reverting to Cu photocathodes, the present design calls for 150 µJ in UV assuming a factor of three loss in the UV beam transport.
Taking a conservative conversion efficiency of 10%, the required near-IR pulse energy is 1.5 mJ.
Contrast (100:1)
The Panel feels that this is not an issue since typical regenerative amplifier leakage in the second pulse is at most 1 to 2%. With third harmonic generation, this translates into less than 0.001% leakage in the UV.
If leakage is a problem, add another Pockels cell.
Timing jitter (0.2 ps rms)
The desired 0.2 ps rms is very difficult to obtain and to maintain with the existing commercial systems. The current state-of-the-art is 0.4 ps rms, 2 ps peak-to-peak.
Time-bandwidth: ANL experience shows that this system exhibits 10 degrees drift in 2 hours.
Lightwave: The best timing jitter with the Model 131 is <1 ps rms.
FESTA said they measured 77 fs jitter with their Ti:Sapphire laser. One should check to see if their measurements are relevant (that is, if the modelocked laser is synchronized to an external RF source).
Pointing stability (<5 µrad)
This should not be a problem if uniform spatial beam generation is implemented with Fourier relay imaging. However, a diagnostic setup with both near-field and far-field images will help assessing the pointing stability.

Amplitude stability (<5% fluctuations)
This stability would be difficult if additional pulsed diode-pumped amplifiers were to be used. It is feasible with the commercially available regenerative amplifier systems.
Repetition rate (>120 Hz)
- The cw-pumped frequency-doubled Nd:YAG as 1 KHz pump for Ti:Sapphire regenerative amplifier can achieve close to this design stability specification for output energy (at 800 nm) around 1.5 mJ. More energy can be obtained with additional amplifier stages operating at a lower repetition rate. The use of diode arrays to pump the additional amplifiers raise questions about their lifetime. Lawrence Livermore claimed to have a system that puts out 100 mJ of green pump at 120 Hz and lasts 1 billion shots. The lifetime of cw diode is typically 10,000 hours.
- The cost of a 150-200 watts system is $150K each.
- The output green (532 nm) pump energy is 15 mJ. With 10% conversion efficiency to 800 nm, the Ti:Sapphire regenerative amplifier output is 1.5 mJ. . (Actually, there is about ~20% conversion to 800 nm in the regen, but with ~50% grating compressor loss, the compressed output at 800 nm is about 1.5 mJ). After harmonic conversion, again assuming 10% efficiency after compression, the output in the UV is 150 µJ. This gives no headroom for spatial and temporal shaping, and thus it is imperative to perform spatial and temporal shaping earlier in the optical chain. Furthermore, ATF personnel should exercise care to reduce the optical loss in the beam transport system.

Last Modified: December 3, 2007
Please forward all questions about this site to: Vitaly Yakimenko