Short pulse laser train for laser plasma interaction experiments

REVIEW OF SCIENTIFIC INSTRUMENTS 78, 083501 共2007兲

Short pulse laser train for laser plasma interaction experiments J. L. Kline, T. Shimada, R. P. Johnson, D. S. Montgomery, B. M. Hegelich, D. M. Esquibel, K. A. Flippo, R. P. Gonzales, T. R. Hurry, and S. L. Reid Los Alamos National Laboratory, Los Alamos, New Mexico 87545

共Received 28 March 2007; accepted 22 June 2007; published online 13 August 2007兲 A multiframe, high-time resolution pump-probe diagnostic consisting of a consecutive train of ultrashort laser pulses 共⬃ps兲 has been developed for use with a chirped pulse amplification 共CPA兲 system. A system of high quality windows is used to create a series of 1054 nm picosecond-laser pulses which are injected into the CPA system before the pulse stretcher and amplifiers. By adding or removing windows in the pulse train forming optics, the number of pulses can be varied. By varying the distance and thickness of the respective optical elements, the time in between the pulses, i.e., the time in between frames, can be set. In our example application, the CPA pulse train is converted to 527 nm using a KDP crystal and focused into a preformed plasma and the reflected laser light due to stimulated Raman scattering is measured. Each pulse samples different plasma conditions as the plasma evolves in time, producing more data on each laser shot than with a single short pulse probe. This novel technique could potentially be implemented to obtain multiple high-time resolution measurements of the dynamics of physical processes over hundreds of picoseconds or even nanoseconds with picosecond resolution on a single shot. © 2007 American Institute of Physics. 关DOI: 10.1063/1.2760687兴

I. INTRODUCTION

The high cost per shot and low repetition rate of high energy pulsed laser systems promote a strong need to increase the data return rate on experiments. Furthermore, on many experiments shot-to-shot reproducibility is often an issue, both due to fluctuating experimental conditions and to a low number of available shots. Innovative methods to increase data return rates on a single shot help increase data statistics and eliminate shot noise, reinforcing scientific results which are often difficult to achieve on large scale, single shot systems. For experiments at the Trident laser1 investigating laser plasma interactions with short laser pulses, ⬃3.5 ps, a pulse train has been implemented using chirped pulse amplification2 共CPA兲 to produce multiple data points at different times with high temporal resolution on a single laser shot. Experiments using short laser pulses can be simulated with full length particle-in-cell simulations of the laser interaction in an effort to make quantitative comparisons with simulations. Laser plasma interaction experiments using longer pulses, ⬎10 ps, cannot be modeled for the full length of the laser interaction and still include the detailed physical processes that occur on space and time scales consistent with the laser wavelength and frequency. In addition, modifications to the background plasma conditions and particle distribution functions by the interaction beam become important, and occur typically on space and time scales such that picosecond detection is required to resolve such processes. Such experiments with multiple short laser pulses take advantage of the evolving plasma conditions to increase the data return rate for laser plasma interaction experiments over a range of plasma conditions. Another potentially revolutionary application for the short pulse train technique can be found in current high in0034-6748/2007/78共8兲/083501/5/$23.00

tensity short pulse experiments. One problem in relativistic laser-matter interactions is preplasma formation, due to the finite intensity contrast of the laser pulse. At a peak intensity of Io ⬎ 1019 W / cm2, even a small prepulse, typically 10−6 at 1 – 2 ns, is strong enough to destroy the target surface and form a plasma. This preplasma and its exact properties will significantly influence laser interaction parameters such as reflection, absorption, beam steering, energy conversion, etc., as well as influence the production of hot electrons, K␣, high harmonics, and fast ions. One way to diagnose these effects would be to exactly measure the contrast ratio and use it to calculate the preplasma evolution using hydrocodes. Not only is this a rather indirect and time consuming method, it is also difficult to implement with enough dynamic range over the complete time domain of relevance. Moreover, due to shot-to-shot fluctuations in current short pulse laser systems, it is necessary to measure those parameters in a single shot and for every shot. Clearly, it is advantageous to directly measure the preplasma evolution, which has been done in the past by many groups using pump-probe interferometry and shadowgraphy.3 The major drawback of all these measurements is that while they have picosecond time resolution, one only obtains a single point in time per shot. For a complete measurement of the preplasma dynamics, however, one needs a series of data at different times. Previous work did this by changing the probe delay on consecutive shots, and building up a series of data over a number of shots. There are two major disadvantages to this method: 共a兲 it is very time consuming and one loses valuable shot and target resources, since one needs to keep the target and other experimental conditions as constant as possible to ensure comparability, and 共b兲 due to shot-to-shot laser fluctuations and target re-

78, 083501-1

© 2007 American Institute of Physics

Downloaded 13 Aug 2007 to 128.165.101.187. Redistribution subject to AIP license or copyright, see http://rsi.aip.org/rsi/copyright.jsp

083501-2

Kline et al.

producibility, and the aforementioned associated laser plasma instabilities at the target interface, this comparability is suspect at best. The new technique presented in this article solves this problem by enabling measurements of the complete time history of plasma dynamics in a single shot with picosecond resolution. Implementing this as a standard diagnostic for high intensity laser experiments will greatly improve the comparability of results, and thus aid in the deduction of fundamental scaling laws for parameters such as laser-energy to electron-energy conversion efficiency, accelerated particle energy, etc., which right now contradict each other in different publications.4–6 While the optical setup to generate the picosecond pulse train has been implemented at low intensities in this article, it also has the potential to be used for high intensity, subpicosecond, short pulse experiments. The pulse train is generated after the short pulse laser oscillator but prior to the regenerative amplifiers and the stretcher, amplifiers, and compressor in a CPA configuration. This helps to maintain the integrity of the beam wave front needed for the creation of a high quality short pulse beam. However, more research is required before implementing such a setup as described in this article to ensure there will be no damage to optical components in the high power CPA system and the pulse shapes can maintain their integrity. For instance, problems may arise from the spatial overlap of the pulses after the stretcher. Overlapping of the stretched pulses could induce high frequency fluctuation of the intensity in the amplifiers or large spatial intensity modulations on the compressor gratings. The overlapping of the pulse may also have a bearing on the shape of the compressed pulses. This problem may be addressed with sufficiently large temporal separation in the pulse train. If the approach presented in this article can be scaled up in energy, multiple pulse hard x-ray or proton radiography sources could potentially be created that are able to diagnose very dense plasmas. Combining this capability with the technique presented here would, for example, enable the measurement of the complete temporal implosion sequence for an inertial confinement fusion 共ICF兲 capsule using a single short pulse laser for the radiographic source. With a high cost per shot on a facility such as the National Ignition Facility7 共NIF兲 or Omega EP,8 and the cost and space constraints for implementing multiple short pulse systems on such a facility, this pulse train technique represents a new possibility worth exploring. Issues, such as the aforementioned overlapping of the stretched pulses in the compressor 共which could be a few nanoseconds and limit the applicability of this technique兲, will need to be carefully examined. In this article, initial proof-of-principle experiments using a short pulse train are described. The optical layout for generation of the pulse train is presented in Sec. II. Sections III and IV describe the Trident experimental setup and results, respectively, in which the pulse train is demonstrated. A brief discussion and conclusions are presented in Sec. V.

Rev. Sci. Instrum. 78, 083501 共2007兲

FIG. 1. Block diagram of CPA laser with pulse train optics.

tem 共Fig. 1兲. The optical setup for generating the pulse train is rather straight forward. A single, 1054 nm, 350 fs Gaussian pulse, as measured with a second order autocorrelator, from the master oscillator is reflected off of a thin film polarizing beam splitter 共TFPBS兲 to a series of high quality, ␭ / 10, laser windows 共Fig. 2兲. The reflection at each window surface provides a single short pulse. Separation of a pair of pulses generated at each surface of a single window is determined by the round-trip transit time of light in the window, i.e., by the window thickness. The temporal separation of each pair of pulses is varied by changing the distance between each window. Pairs of pulses can simply be added by introducing more windows. For these experiments, two 10 mm thick windows separated by ⬃15 mm are used to produce four 3.5 ps pulses 共after compression兲 separated by ⬃100 ps. A ␭ / 4 wave plate is inserted between the TFPBS and pulse train forming windows to convert the linearly polarized light to circular polarization on the first pass. On the outward pass, the circularly polarized light reflected from each window surface is converted back to linearly polarized light that is rotated by 90° from the initial laser light allowing the pulses in the train to pass through the TFPBS to the rest of the laser system. Since each pulse, except for the first pulse which has 4% of the incident energy, passes back through other surfaces, there is a small decrease in intensity for subsequent pulses due to Fresnel losses. Accounting for the reduction in energy as the incident beam propagates through each window the first time and the loss at each surface as each reflected pulse passes back through each surface, we estimate that the second pulse is ⬃3.7% of the incident, the third pulse is 3.4% of the incident, the fourth pulse is ⬃3.1% of the incident, and so on. However, these are the relative energies in each pulse prior to propagating through the amplifiers. The final energies in each pulse depend on the amplification, in particular, the gain saturation

II. PULSE TRAIN OPTICAL CONFIGURATION

The pulse train is generated after the short pulse master oscillator prior to propagating through the Trident CPA sys-

FIG. 2. Schematic of the pulse train optical setup. The pulse forming windows are 10 mm thick and separated by 15 mm to produce four ⬃3.5 ps pulse separated by ⬃100 ps.

Downloaded 13 Aug 2007 to 128.165.101.187. Redistribution subject to AIP license or copyright, see http://rsi.aip.org/rsi/copyright.jsp

083501-3

Short pulse laser train for LPI experiments

FIG. 3. Schematic diagram of the target area experimental setup.

for each pulse. In an effort to even out the final energies in each pulse, the first window was slightly tilted such that the reflections from the surfaces had a small angle with respect to the optical axis. This detunes the coupling of the first pair of pulses to the regenerative amplifier, which comes before the CPA system, reducing the energy in these pulses. The windows used to generate the pulses do introduce a small amount of group velocity dispersion. Since the different pulses pass through different path lengths in the glass, it is not possible to compensate for the group velocity dispersion on each of the pulses. For picosecond pulses, the amount of group velocity dispersion added is negligible, but this effect can limit the pulse durations as the pulse lengths approach the theoretical limit. III. EXPERIMENTAL SETUP

Experiments to demonstrate the viability of the short pulse train for laser plasma interactions were carried out at the Trident laser facility.1 A single 527 nm, 1 ns square pulse, laser beam with up to 200 J is used to generate a plasma 共Fig. 3兲. This heater beam is focused through a 6 mm hexagonal random phase plate using a f / 6 lens with a 120 cm focal length and 1.2 ns flattop pulse shape. The heater beam is focused ⬃750– 1500 ␮m in front of the gas jet target to minimize ionization defocusing which has been observed to decrease electron temperatures. The resulting intensity of the heater beam on the gas jet target is ⬃3.0 ⫻ 1014 W / cm2. The plasma is generated with helium gas producing electron densities and temperatures in the ranges of 2 – 20⫻ 1019 cm−3 and 180– 350 eV, respectively. After the plasma is formed, the interaction beam with the pulse train is fired. The interaction beam leaves the compressor and passes through a KDP crystal to convert the 1054 nm laser light to 527 nm. The interaction beam then propagates through a beam splitter where half the laser light propagates to a wedge that is used as a wavelength and time fiducial.

Rev. Sci. Instrum. 78, 083501 共2007兲

The other leg propagates though a f / 4.5 lens and is focused into the plasma. Backscattered light due to stimulated Raman scattering in the plasma is reflected back through the focusing lens and off of the beam splitter to the backscatter diagnostics. The fiducial leg reflects off the timing wedge where ⬃4% of the light passes back through the beam splitter into the backscatter diagnostics. The path length for the fiducial wedge and the focal point of the focusing lens are intentionally made different. The difference in the path length creates a shift in the relative time between the timing fiducial and backscattered light signals and is controllable by the position of the timing wedge. If the two paths had the same lengths, stimulated Brillouin scattering, which would occur at the same time as the SRS signals on this scale and has a small wavelength shift compared to the incident light wavelength, could not be distinguished from the fiducial signal if present. A pickoff wedge is placed in the timing mirror path to direct a small amount of the laser light in this leg to a pyroelectric calorimeter. The pyroelectric calorimeter is cross calibrated to provide a measure of the total incident energy for the pulse train, since it does not have the temporal resolution to measure the energy in each pulse. As shown in the next section, streak camera measurements of the pulse train from the timing mirror can be used to determine the energy in each pulse of the train. The backscattered light from laser plasma instabilities and the light from the fiducial wedge follow the same path on the back side of the beam splitter to the diagnostics. The light is split into two paths with another 50/ 50 beam splitter. One leg is focused into a 1 / 4 meter Chromex spectrograph to spectrally resolve the signal. The output from the spectrograph is coupled to a Hamamatsu 4187 streak camera to provide temporal resolution and is recorded with a charge coupled device 共CCD兲 camera. Although the streaked spectrograph provides ⬃50 ps temporal resolution, SRS driven by each pulse in the short pulse laser train is much shorter in duration than the temporal resolution. However, the streaked spectrograph allows the backscattered light from each pulse in the train to be resolved, and resolves the slower time scale dynamics as the plasma evolves. The other leg propagates to a diode that measures the backscattered laser light energy. A 4 mm OG550 filter 共see Schott glass colored long pass filters9兲, is placed in front of the diode to block light at wavelengths less than 550 nm from being recorded by the diode. This allows the measurement of only the energy from the SRS backscattered light. As with the pyroelectric calorimeter, the diode does not have the temporal resolution to discern the backscattered energy for each pulse, but can be determined for each pulse with the use of the streaked measurements. IV. MEASUREMENTS OF SRS WITH THE PULSE TRAIN

The pulse train provides a way to increase the data return rate for picosecond-resolution measurements of SRS under different plasma by conditions taking advantage of the plasma evolution in time. Langmuir waves driven by stimulated Raman scattering have been shown to undergo different nonlinear behaviors based on the dimensionless parameter

Downloaded 13 Aug 2007 to 128.165.101.187. Redistribution subject to AIP license or copyright, see http://rsi.aip.org/rsi/copyright.jsp

083501-4

Kline et al.

FIG. 4. Value of k␭D from a one dimensional simulation of laser produced helium gas jet plasma with an atomic density of 6 ⫻ 1019 cm−3, 0.7 mm thickness, and a laser intensity of 5 ⫻ 1013 W / cm2.

k␭D, where k is the electron plasma wave number and ␭D is the plasma Debye length.10,11 In fact, the transition between regimes has been observed during a single 200 ps interaction beam pulse as the plasma cooled and k␭D changed.10 The problem with such a measurement is that a continuous interaction beam pulse can modify the plasma conditions locally affecting the resonant SRS process at later times. In addition, changing hydrodynamic conditions continuously affect the SRS process during the interaction beam pulse. Short laser pulses are not affected by hydrodynamics since the laser pulse occurs over time scales too fast for hydrodynamic effects. Short laser pulses also have the advantage that particlein-cell simulations can be performed over the full length of the interaction. However, using a single short pulse does not take advantage of the changing plasma conditions. When the heater beam shuts off, both the density and temperature decrease at different rates. Thus, k␭D changes as a function of time. A one dimensional hydrodynamic simulation using 12 HELIOS shows the evolution of k␭D for a laser gas jet experiment 共Fig. 4兲. In this example, the laser is on for the first 1.2 ns and k␭D peaks at the end of the laser pulse. After the heater laser pulse terminates, the plasma cools and k␭D decreases. With an interaction beam consisting of multiple short laser pulses, the characteristics of SRS can be sampled for these different values of k␭D on each laser shot. This may improve comparisons of SRS for different k␭D since multiple measurements are made under plasma conditions originated during a single laser shot, unlike previous experiments which compare results from different laser shots causing a wider variation in laser and plasma conditions. Figure 5 shows a streaked measurement of SRS using the pulse train. The four pulses from the fiducial mirror are marked as the input beam and are at a wavelength of ⬃527 nm. The relative energy in each pulse can be determined from the streak measurement by dividing integrated signal in each pulse by the sum of the integrated signal in all

Rev. Sci. Instrum. 78, 083501 共2007兲

FIG. 5. Spectrally and temporally resolved streak record of the incident interaction beam and the SRS backscattered light for a four pulse train.

four pulses. Multiplying these ratios by the total energy in the incident pulse train measured by the pyroelectric calorimeter gives the energy in each pulse. Figure 6 shows the energy in each pulse for the data in Fig. 5 and the corresponding intensity for a beam with a Strehl ratio ⬃0.5.13 The measured energy in each pulse shows that the first pulse in the train is reduced by tilting the first laser window in the pulse train forming optics. At wavelengths near 615 nm, there are four pulses of SRS backscattered light. As time increases, there is a small shift in each pulse towards shorter wavelengths indicative of either a drop in electron density, temperature, or both. The SRS backscattered spectra also exhibit broadening, a phenomenon still under investigation. This could possibly be a result of mechanisms such as a nonlinear frequency shifts due to electron trapping, or an increase in SRS bandwidth due to strong coupling. While this measurement does prove that a train of pulses can be used to measure SRS in a single

FIG. 6. 共쎲兲 Energy in each pulse 共left axis兲 and the corresponding 共䊐兲 irradiance 共right axis兲 determined by the measured total energy in the incident pulse train and streak camera measurements of the incident beam from the timing mirror.

Downloaded 13 Aug 2007 to 128.165.101.187. Redistribution subject to AIP license or copyright, see http://rsi.aip.org/rsi/copyright.jsp

083501-5

Rev. Sci. Instrum. 78, 083501 共2007兲

Short pulse laser train for LPI experiments

laser shot, it should be noted that proper filtering and the dynamic range of the recorder device can prove to be a challenge.

ACKNOWLEDGMENTS

This work was performed at Los Alamos National Laboratory under the auspices of Los Alamos National Security, LLC, for the Department of Energy under Contract No. DEAC52-06NA25396. 1

V. DISCUSSION

The present work has demonstrated the viability of using a short pulse train to increase the data return rate for laser plasma interaction experiments. A higher data return rate for high energy laser experiments could increase data statistics and build confidence in results that typically have too few data points. The increased data return may also allow parametric studies for laser experiments that can use the pulse train. The setup to generate the short pulse train presented in this article also has the potential to be used for variety of other applications, including those requiring high intensity short pulses 共made possible by the fact that the pulse train forming optics come before the CPA system兲. A high intensity laser pulse train opens many new possibilities for short pulse experiments.

N. K. Moncur, R. P. Johnson, R. G. Watt, and R. B. Gibson, Appl. Opt. 34, 4274 共1995兲. 2 D. Strickland and G. Mourou, Opt. Commun. 56, 219 共1985兲. 3 K. Adumi et al., Phys. Plasmas 11, 3721 共2004兲. 4 J. Fuchs et al., Nat. Phys. 2, 48 共2006兲. 5 L. Robson et al., Nat. Phys. 3, 58 共2007兲. 6 Y. Sentoku, K. Mima, H. Ruhl, Y Toyama, R. Kodama, and T. Cowan, Phys. Plasmas 11, 3083 共2004兲. 7 J. A. Paisner, E. M. Campbell, and W. J. Hogan, Fusion Technol. 26, 755 共1994兲. 8 L. J. Waxer et al., Opt. Photonics News 16, 30 共2005兲. 9 See Schott’s optical glass catalogue for transmission function at SCHOTT North America, Inc., Advanced Optics division, 400 York Ave, Duryea, PA 18642, http://www.us.schott.com/optics_devices/filter/english/us/ index.html 10 J. L. Kline et al., Phys. Rev. Lett. 94, 175003 共2005兲. 11 J. L. Kline et al., Phys. Plasmas 13, 055906 共2006兲. 12 Prism Computational Sciences I, HELIOS, 2002. 13 D. S. Montgomery, R. P. Johnson, J. A. Cobble, J. C. Fernandez, E. L. Lindman, H. A. Rose, and K. G. Estabrook, Laser Part. Beams 17, 349 共1999兲.

Downloaded 13 Aug 2007 to 128.165.101.187. Redistribution subject to AIP license or copyright, see http://rsi.aip.org/rsi/copyright.jsp