Chapter 1

Introduction: Historical Background

The field of high-intensity laser interaction with matter, although barely

two decades old, is already bursting with enough exotic phenomena to keep
researchers busy for years to come. Since the invention of 'chirped pulse
amplification' in 1985, progress in short pulse laser technology has been un-
relenting, see Fig. 1.1. Pulse durations have come down from a picosecond
to less than 5 femtoseconds (10-l5 s); whereas focused intensities have sky-
rocketed six orders of magnitude. At present, several laboratories around
the world are now promising intensities in excess of lo2' WcmP2.
In view of the impending escalation suggested by this diagram, a shift
into the unchartered territory of highly relativistic laser particle physics, it
is perhaps appropriate to take a brief look at the various disciplines from
which this new field has grown. These contributory fields are numerous
and diverse: they include laser physics, atomic physics, plasma physics,
and lately even astrophysics and elementary particle physics. Many of the
theoretical models described in later chapters can be traced back to one
of these more classical areas. This does not mean that researchers in this
business have been able to get their theories or ideas 'off the shelf' - on the
contrary: the extreme conditions under which light and matter are forced
to coexist during such interactions have posed a continual challenge to both
theoreticians and experimentalists alike.
The key to understanding the underlying physics in these interactions
is to realize that ordinary matter - whether solid, liquid or gas - will be
rapidly ionized when subjected to high intensity irradiation. The electrons
released are then immediately caught in the laser field, and oscillate with a
characteristic energy (the right hand column in Fig. 1.1) which then dictates
the subsequent interaction physics.
2 Short Pulse Laser Interactions with Matter

Intensity (w/cm2 ) Electron

energy
Theoretical limit
.............................
1oZ3

I
Particle physics
y- ray sources 1 GeV
Particle acceleration
Fusion schemes
1020 -
1 MeV

Hard X-ray flash lamps

Hot dense matter

Field ionisation of hydrogen -- -- -----.

Multiphoton physics
Laser medicine

1960 1970 1980 1990 2000 2010

Year

Fig. 1.1 Progress in peak intensity since the invention of the laser in 1960.

1.1 Multiphoton Physics

One of the most familiar examples of light inflicting a change in material

properties is the photoelectric effect, predicted by Einstein a century ago
(Einstein, 1905) and validated experimentally by Millikan a decade later
(Millikan, 1916). This process, the ejection of an electron from an atom
by a single photon, occurs when the photon energy fw matches the height
of the atomic potential barrier, I,, which the electron experiences in the
vicinity of the ion, i.e. tw = I,. Even for the outer shells of most atoms, this
energy still runs into many electron-volts, equivalent to photon wavelengths
well into the ultraviolet (UV) range. For inner shells (I, .v keV), one needs
hard x-rays to induce photo-ionisation.
With standard lasers (operating wavelengths 0.25 pm - 13.4 pm ), one
cannot observe the photoelectric effect on normal material because fw <<
 Introduction 3

I,. As lasers grew more powerful in the 1960s and 70s, however, (Fig. 1.1)
it became possible to consider multiphoton ionisation, expressed by the
condition

Thus, instead of one very energetic photon, an electron absorbs n pho-

tons of moderate energy (for example, laser photons with fw z eV) and
is subsequently ejected. Some of the pioneering work on this was done a t
CEA-Saclay, France by the group of Mainfray and Manus (1991), includ-
ing the discovery of above-threshold ionisation by Agostini et al. (1979).
Another ground-breaking event during this period, prompted by the devel-
opment of chirped pulse amplification, was the prediction of high-harmonic
generation via recombination of multiphoton-ionized electrons by Shore and
Knight (1987). This phenomenon was confirmed experimentally soon after-
wards (McPherson et al., 1987; Ferray et al., 1988) by a intensive series of
experimental investigations in a number of labs worldwide. Today it is pos-
sible to generate hundreds of harmonics down to nanometre wavelengths, a
feat which has heralded a new era of highly compact, coherent XUV light
sources.

1.2 Single-Electron Interaction with Intense Electromag-

netic Fields

This subject actually goes much further back than most newcomers to the
femtosecond business would probably guess; predating the invention of the
laser a t the beginning of the 1960s by some margin. One of the first analy-
ses of the behavior of free electrons in the presence of intense radiation was
made by Volkov (1935), who introduced the concept of a dressed state to
describe the enhanced inertia experienced by an electron when oscillating
in an electromagnetic field. Later, motivated by the first experiments with
synchrotrons, Schwinger made a detailed analysis of the radiated power
emitted by accelerated electrons (Schwinger, 1949), pointing out that radi-
ation is preferentially emitted in the direction of motion at high energies.
These early ideas were refined and explored with more urgency when
the invention of the laser brought the prospect of an experimental means
t o study relativistic photon-electron physics in the laboratory (Brown and
Kibble, 1964; Vachaspati, 1962; EberIy and Sleeper, 1968; Sarachik and
Schappert, 1970). These authors defined the figure of merit for 'laser-
4 Short Pulse Laser Interactions with Matter

electron' interaction by a dimensionless parameter q (or q 2 ) , given by:

where e, m and c are the electronic charge, electron mass and speed of
light respectively; EL is the laser electric field strength and w the light
frequency. Needless to say, these theoretical works all lamented the impos-
sibility of achieving truly relativistic conditions (q > 1) with the optical
lasers available a t the time, but speculated that they might be might one
day be reached with 'future' technology. Forty years later, this wishful
thinking has become reality, and in Chapter 3 we will see how these vision-
ary models have inspired new lines of experimental investigation.
Independently of the laser's arrival, astrophysicists were beginning to
suggest mechanisms for cosmic ray generation in the vicinity of pulsars via
the interaction of intense electromagnetic (EM) radiation with free elec-
trons (Ostriker and Gunn, 1969; Gunn and Ostriker, 1971). The numbers
involved here are of course vastly different from laser-plasma interactions:
pulsar radiation has a frequency between 0.3 and 30 Hz, and magnetic field
strengths near the star surface in the region of 1012 Gauss; causing gar-
gantuan oscillation amplitudes of particles in the surrounding plasma. The
interaction physics is very similar however, and is just one of many instances
of scalable laboratory astrophyszcs, where one can emulate the conditions in
astrophysical objects using high-power lasers.

1.3 Nonlinear Wave Propagation

The fact that plasmas can support large-amplitude, nonlinear waves has
been known for almost as long as plasma physics established itself as a main-
stream branch of science. Early seminal works by Akhiezer and Polovin
(1956) and Dawson (1959) set the scene for numerous studies on the behav-
ior of both large-amplitude Langmuir (electrostatic) waves and the prop-
agation of high-intensity electromagnetic radiation in plasmas (Davidson,
1972). One attraction of this then rather obscure field was the tantaliz-
ing possibility of producing long-lived solitons in plasmas (Zakharov, 1972;
Decoster, 1978; Shukla et al., 1986).
The publication of a method for laser-acceleration of electrons in un-
derdense plasmas by Tajima and Dawson (1979) sparked off a fresh wave
of interest in wave propagation. This enthusiasm also encompasses mem-
bers of the accelerator community, who are actively on the look-out for
 Introduction 5

viable alternatives to conventional linear accelerators and synchrotrons as

these devices approach their physical and economic limits. At the time
of writing, the world record for laser-plasma acceleration of electrons lies
a t around 350 MeV, not breathtaking compared to the 50 GeV per elec-
tron/positron beam a t SLAC, until one realizes the former was achieved
with a few millimetres of plasma rather than several kilometres of super-
cooled accelerator structure. Although there is some way to go before laser-
plasma acceleration can compete with existing facilities, scaling up t o the
Petawatt powers now available will soon make the tabletop GeV electron
accelerator a reality.

1.4 Metal Optics

The story of laser interactions with solids also has a nineteenth century
prologue. The simple observation that polished metals behave as almost
perfect reflectors, whereas other materials either absorb or transmit light,
could not be satisfactorily explained until Drude set out his 'Electron The-
ory' (Drude, 1900). Although solid state physics has advanced beyond
recognition since then, the so-called Drude model of electron conduction
still retains its appeal and usefulness in describing the main features of
metal optics (see, for example: Ashcroft and Mermin, 1976, Chapter 1).
Drude's original idea -just three years after J. J. Thomson7sdiscovery
of the electron - was t o suppose that the atoms in a metal somehow share
a limited number of 'valence' electrons, forming a conduction band. These
can wander as far as they like from their parent atoms, carrying current and
heat through the material in the process. For an element with mass density
p and atomic weight A, the free electron density is given by ne = NAZ*p/A,
where NA is Avogadro's constant and Z* is the number of valence electrons
per atom.
The conductivity of a metal will depend on the rate a t which these
free electrons are slowed due to collisions with the ions. Mathematically
this can be expressed by the relation: a, = nee2r/me,where r is the
collision or relaxation time. Even today, precise theoretical treatments to
determine the relaxation time in solids remain a challenge, since the latter
depends on details of the crystal structure, electronic configuration and so
on. For this reason, nearly all the available data on metallic conductivity
has been obtained through experimental measurements using Ohm's law:
j = a,E. This technique - applying a DC voltage and finding the
6 Short Pulse Laser Interactions with Matter

resulting current -can be recast in a form suitable for optical diagnostics

via the AC conductivity:

When combined with Maxwell's equations, this leads to a complex di-

electric constant:
"
w;
E = l -
w(w + iv) '
where

is the plasma frequency of the valence electrons, and u = r-' is their

relaxation rate or collision frequency. This expression basically predicts
that a metal will only become transparent for radiation with wavelengths
X < A, = 2m/w,, which turns out to be in the UV (200-400 nm) range.
Standard laser light (0.5-10 pm) will be reflected or absorbed, depending
on the collision time.
Now, anyone with some basic plasma physics can see that the Drude
model is almost identical to the standard theory of collisional laser absorp-
tion in plasmas, except that in the latter case, T can be calculated with
some accuracy. Suffice t o say that metal optics provides an excellent start-
ing point for studying reflectivity (and other transport properties) of short
pulse laser-produced plasmas (Godwin, 1972). In Chapter 5 we will see how
this leads t o an almost seamless transition between the solid and plasma
states of matter.

1.5 Long Pulse Laser-Plasma Interactions

A common misconception about femtosecond laser-plasma research is that

it has little or nothing t o do with the long pulse laser-plasma interaction
issues relevant to inertial confinement fusion (ICF). It is certainly true that
one cannot do ICF with a table-top laser (at least not yet, anyway!). The
scientific (and often political) demarkation which prevails today masks a
broad thematic overlap, and there are plenty of researchers who happily
jump back and forth between the two fields on a daily basis. Whatever
one's personal view of ICF as a future energy source, and its unavoidable
relationship with nuclear weapons programs, it has to be said that the
 Introduction 7

advances in the short pulse field would have been nowhere near as rapid
without the considerable prior scientific and technical knowledge of laser
science and interaction physics generated by the ICF programs over the
last 30 years. Luckily, training in ICF physics is not necessary to work
on short pulse interactions, but it is nonetheless useful to know where to
find the original literature, on, say, hydrodynamics, parametric instabilities,
energetic particle generation, and so on. A brief outline of ICF would
therefore appear to be in order, even if to mainly draw contrasts between the
new femtosecond phenomena and this long pulse regime in later chapters.
Laser fusion became official in 1972 (having previously been under mil-
itary classification) with the publication of a classic but over-optimistic
paper in Nature by Nuckolls et al., (1972). In this work, the authors de-
scribe how a small micrometer-sized pellet filled with deuterium and tri-
tium fuel can be compressed to enormous densities by irradiating it with
laser beams focused symmetrically onto its surface (Brueckner and Jorna,
1974). By converting the laser energy into thermal plasma energy, the pel-
let shell material ablates radially outwards, thus pushing the fuel inwards
via a rocket-like reaction. By virtue of the spherical symmetry, the fuel
implodes, eventually reaching densities p of several hundred g ~ m -and ~
temperatures T of 10 keV (lo7 degrees Kelvin), thus meeting the require-
ments for thermonuclear confinement encapsulated by the product:

This condition - known as the Lawson criterion - basically states that

the fuel must burn up and release its energy before the capsule blows apart,
leading to a requirement for the areal fuel density pR, where R is the final
capsule radius.
The standard hot-spot scenario (Lindl, 1995) currently being pursued
by the major laboratories in the USA, France and Japan, requires a central
pR = 0.3, which, after allowing for all the inefficiencies and target design
considerations, translates into a laser energy requirement of around 1 MJ
to achieve gain. We will return to these constraints later in section 7.5,
where some speculative new ideas to reduce this driver requirement with
the help of an additional short pulse laser are described.
These more recent schemes aside, the whole process of target ablation,
implosion and ignition is essentially determined by hydrodynamics: any
plasma physics involved is almost invariably destructive, putting additional
constraints on the laser and target design. These coronal processes (so-
called by analogy with stellar objects) are illustrated in Fig. 1.2: as we
 Short Pulse Laser Interactions with Matter

- - Raman: em- em + 1
, '

Inverse bremsstrahlung

Fig. 1.2 Laser-plasma interactions in the corona of an imploding micro-balloon

shall see in later chapters, nearly all of them turn up again in femtosec-
ond interactions. Parametric instabilities such as Raman and Brillouin
scattering (Kruer, 1988) are generally bad for the implosion because they
generate fast electrons. Because of their long range, these electrons preheat
the target core, making the compression less efficient. Resonance absorp-
tion (which we will meet in Sec. 5.5.1) is also undesirable for this reason: in
fact, up to 50% of the laser energy can be wasted on superheated electrons
in this way. The classic signature of such 'anomalous absorption' processes
is a bi-Maxwellian electron distribution, see Fig. 5.17. In the case of reso-
nance absorption, the hot electron component has a temperature TH given
by (Forslund et al., 1975a; Estabrook et al., 1975):
 Introduction 9

where I is the laser intensity in units of 1016 and X the laser

wavelength in microns. The dependence on the product I X 2 (originating
in the quiver motion of the electrons in the laser field) means that long
wavelength lasers (e.g.: C 0 2 ) tend to produce hot electrons much more
readily (than, say KrF) for a given intensity. This was a major factor in
the strategic switch to shorter wavelengths (A < 0.5pm) in ICF programs
a t the beginning of the 1980s. A major advantage of indirectly (X-ray)
driven schemes is that hot electron generation is almost completely avoided
(Lindl, 1995).
We have already seen that the rules of the game describing ultrafast,
ultra-high-intensity (UHI) interactions have t o be rewritten. For example,
the short timescales make it necessary to discard steady-state or adiabatic
models in favour of fully dynamic non-equilibrium formulations. On the
other hand, some of the computational tools developed over 30 years ago to
investigate hot electron generation in nanosecond laser-plasma context are
used today to model femtosecond interactions at intensities of lo2' w ~ m - ~ ,
not least in the fast ignitor context (Sec. 7.5). In fact many of the codes
used for UHI interactions can be traced back t o ICF applications: atomic
physics packages, fluid codes, particle-in-cell or Vlasov codes. These have
usually been adapted to meet contemporary demands: for instance by us-
ing time-dependent ionization models to cope with the breakdown of local
thermal equilibrium (LTE); by allowing for relativistic electron dynamics;
or by introducing wave propagation into hydrodynamic models to handle
the laser pressure and absorption more accurately. Details of these models
are deferred to Chapter 6, but note that there are many areas - including
many optical and diagnostic techniques - where the connection with ICF
can turn out to be very fruitful.

1.6 Femtosecond Lasers

Hopefully, it will be clear by now that the intention of this book is not
t o supply assembly instructions for a Terawatt laser system, but rather to
examine the physics which can be explored with the help of such a device.
The material contained in the chapters which follow is thus very much
aimed a t those researchers either physically or mentally gathered around
the target chamber (see Fig. 5.15), whose knowledge of the laser operation
consists primarily of the four magic numbers: wavelength, energy, pulse
duration and focal spot size. That said, the only way t o appreciate where
10 Short Pulse Laser Interactions with Matter

these numbers are conjured up from is to lift the lid off the box and take a
peek a t the optical wizardry inside.

Fig. 1.3 Front end of the Jena Ti:Sapphire femtosecond laser.

If we do this, we will find something resembling a huge BRIO-BuilderTM

set with an awful lot of glass in it, see Fig. 1.3. Make no mistake, though:
these devices represent the state-of-the-art in optical engineering.'. We
can make more sense of this optical jungle with the help of the schematic
diagram in Fig. 1.4, where we notice that the 'laser' actually comprises sev-
eral autonomous units, namely: an oscillator, producing low-energy short
pulses at regular intervals; a stretcher, converting the femtosecond pulse
into a 50-200 ps chirped pulse; one or more amplifier stages to increase the
pulse energy by a factor of lo7-10'; and finally a compressor, performing
the exact optical inverse of the stretcher t o deliver an amplified pulse of the
same duration as the oscillator.
The key technique in this business is chirped-pulse amplification (CPA),
which since its first application t o lasers by Gerard Mourou and co-workers
in 1985 (Strickland and Mourou, 1985; Maine et al., 1988), is now exploited
by nearly all existing high-power, femtosecond systems. At first sight, the
procedure seems unwarranted: why bother with all the stretching and com-
pressing when one could amplify the short pulse from the oscillator straight
away? The answer is simply that most of the optical components will

'A tip for theoreticians being shown around a laser lab for the first time: resist the
temptation to fiddle with the pieces on the table - it can take several days to put right
again.
Introduction
12 Short Pulse Laser Interactions with Matter

be overheated and damaged if the fluence exceeds a level of around 0.16

1/2, where T,, is the pulse duration in picoseconds. For maximum
Jcm- 2 T~~
efficiency, however, the fluence should be near the saturation level of the
amplifying medium, which is 1 J c ~ for - ~Ti:sapphire. Therefore, the pulse
length throughout the amplifier chain has to be a t least 40 ps (typical val-
ues are 100-200 ps) so that the components (e.g. mirror coatings) have
time t o cool by thermal conduction.
The stretcher-compressor combination separates the pulse generation
and amplification stages, permitting standard techniques t o be used for the
amplifier chain, and furthermore leaving room for more advanced phase-
compensation devices a t either end. Before we consider these refinements,
let us return to the 'front end', where the pulse is generated. Femtosecond
laser sources are mode-locked: the output pulse is made up of a superpo-

-
sition of many electromagnetic waves (or laser modes), and is transform
limited, so that rP l / A v , where Av is the bandwidth. Clearly a large
bandwidth is essential t o generate a short pulse. Consider, for example, a
10 fs Gaussian pulse, for which A v r = 0.44. This gives Av = 4.4 x 1013 Hz,
which for a central wavelength of 800 nm, translates to:

X2
Ax = Av- = 94 nm.
C

This is an absolute requirement of the lasing medium in the oscilla-

tor, as well as the subsequent amplification stages. Most materials have
bandwidths far below this, but titanium-doped sapphire crystals have a re-
markable gain curve ranging from 700 nm to 1100 nm, with AX = 230 nm;
the largest bandwidth of any known material. This property, together with
its high saturation fluence of nearly 1 JcmP2 and high damage threshold,
make Ti:sapphire the ideal choice for most femtosecond laser systems in
operation today.
How do we create such a pulse in the first place? In particular, a means
of converting a continuous-wave (cw) pump laser beam into a train of short
pulses is needed. This is done by exploiting another fortuitous property of
the Ti:sapphire crystal: namely, that it also acts as a nonlinear focusing
lens. This so-called nonlinear Kerr nonlinearity (Shen, 1984) occurs when
the intensity inside the crystal exceeds 10" WcmP2, so that by blocking
the unfocused cw mode with a slit, one ends up with a single, short pulse
bouncing to-and-fro in the oscillator. As it stands, this arrangement is
unstable because dispersion will cause the pulse to spread out longitudinally
(red wavelengths are faster than blue ones). To compensate, a set of prisms
 Introduction 13

is inserted between the crystal and back mirror, which imposes an equal but
opposite dispersion on the pulse (blue faster than red). The combination
of cavity plus dispersion-correction constitutes a highly stable and reliable
femtosecond laser, which these days can even be purchased 'off-the-shelf'.
The true art of femtosecond laser physics comes when we try to amplify
the pulse, as we shall see shortly.
The fs pulse from the oscillator typically contains just a few nJ of energy;
yet we wish to increase its power to the TW level and beyond. As already
mentioned, t o avoid damaging optical elements, the pulse must first be
stretched by a factor of lo3-lo4 or so. This is usually done with a pair
of diffraction gratings, which impose a positive chirp on the pulse, so that
longer wavelengths emerge before shorter wavelengths. Early CPA systems
(Maine et al., 1988) actually used enormous km-length optical fibres to
broaden the bandwidth via self-phase modulation. However, the stretch-
factors obtained by this method are limited because there is no practical
means of correcting the nonlinear, high-order phase distortion introduced
by the fibre, ultimately restricting the final pulse length to values of just
under 1 ps.
The stretched pulse can now be amplified - a process which is usu-
ally split into two or more stages, depending on the final beam energy
required. The Jena system comprises three different modules: a regenera-
tive preamplifier and two multipass power amplifiers. Most of the gain of
the system (- lo7) is obtained by the regenerative amplifier, which works
in a very similar fashion t o the laser resonator itself. The difference is that
it seeded by the chirped pulse, which subsequently makes up t o 20 round
trips through a low gain medium, before being switched out by a combined
Pockels cell/polarizer combination. Since the regenerative amplifier even-
tually becomes saturated a t a few mJ, additional cavity-free techniques are
then used to amplify up t o and beyond the 1 J level. The multipass con-
figuration in Fig. 1.4 is typical of modern tabletop-TW systems, producing
gains of lo2-lo3.
The amplified, long chirped pulse is then recompressed using a grating
pair (or quadruplet in the Jena system shown), ideally reducing the pulse
length back down to a value slightly above the one originally emitted by
the oscillator. Some pulse lengthening is inevitable due t o a combination of
nonlinear dispersion effects and gain-narrowing. The latter arises because
the amplifier medium preferentially enhances central wavelengths over pe-
ripheral ones, leading t o a reduction in bandwidth and hence lengthening
of pulse duration.
14 Short Pulse Laser Interactions with Matter

Due to the high-quality beam profile which can be produced by

Ti:Sapphire amplifier systems, the pulses can be focused to an almost
transform-limited spot containing most of the laser energy, see Fig. 1.5.
This is the form of the laser pulse we want for interaction physics: the
maximum possible photon density for a given wall-plug energy.

Fig. 1.5 Sharp end of the Jena Ti:sapphire femtosecond laser system: a focal spot of
3 pm diameter containing more than 50% of the pulse energy. The peak intensity reached
here is 4 x 1019 ~ c m - ~ .

Naturally there are variations of this generic short pulse system depend-
ing on the intended application. What I have just described is a multi-TW
system typical for a university laboratory. For a high-end user facility, of-
fering powers above 100 TW, one still has to go to one of the dedicated laser
laboratories listed in Table 1.1. On the other hand, there is also a growing
number of smaller, high-repetition-rate (kHz) systems for high throughput
applications such as x-ray sources. More recently, state-of-the-art few-cycle
(< 10 fs) lasers have opened up a new field of attosecond physics, a theme
which crops up again in Chapter 7.
As seen in the table, there are currently three or four Petawatt Nd:glass
lasers coming on line over the next two years (VULCAN has already passed
the P W mark). These will deliver focused intensities of about lo2' WcmP2
- perhaps a bit higher after some tuning and by increasing the energy -

but only up to a point. The damage threshold on the compression gratings

or focusing optics is around 0.3 JcmP2, so a Petawatt facility with 400 J
pulse energy needs gratings with least 40 cm in diameter. To be on the safe
side, the VULCAN laser a t RAL uses 1 m-diameter gratings, right a t the
 Table 1.1 Multi-Terawatt laser systems and laboratories worldwide.
Name Laboratory Country Type X Energy Pulse length Power Focal spot Intensity
(nm) (J) (fs) (TW) ( P I ( w ~ m - ~ )

Petawatta LLNL USA Nd:glass 1053 700 500 1300 - > loz0
VULCAN~ RAL UK Nd:glass 1053 423 410 1030 10 1.06 x 1021
P W laserC ILE Japan Nd:glass 1054 420 470 1000 30 loz0
PHELIX~ GSI Germany Nd:glass 1064 500 500 1000 - -
LULI lOOTW LULI France Ti:Sa 800 30 300 100 -
A P R 100 T W APR Japan Ti:Sa 800 2 20 100 11 2 x l0l9
HERCULES FOCUS USA Ti:Sa 800 1.2 27 45 (1) (8 x 10")
ALFA 2 FOCUS USA Ti:Sa 800 4.5 30 150 (1) (loz2)
Salle Jaune LOA France Ti:Sa 800 0.8 25 35 l0l9
Lund T W LLC Sweden Ti:Sa 800 1.0 30 30 10 > l0l9
MBI Ti:Sa MBI Germany Ti:Sa 800 0.7 35 20 > l0l9
Jena T W IoQ Germany Ti:Sa 800 1.0 80 12 3 5 x l0l9
ASTRA RAL UK Ti:Sa 800 0.5 40 12 l0l9
USP LLNL USA Ti:Sa 800 1 (10) 100 (30) 10 (100) 5 x l0l9
UHI 10 CEA France Ti:Sa 800 0.7 65 10 5 x 1019
a 1996-1999
Petawatt performance achieved on October 5, 2004.
Projected upgrade of PWM - PetaWatt Module.
Commissioned for end 2005.
16 Short Pulse Laser Interactions with Matter

limit of present manufacturing capability.

So just as with the old (non-CPA) pulsed technology, the only way to
increase the power without destroying the laser components is to reduce
the pulse length further: a point where the smaller Ti:sapphire systems
have a distinct advantage thanks t o their broader bandwidth. Fortunately,
an extension of CPA - optical parametric chirped pulse amplification, or
OPCPA - has been proposed by Ross et al., (1997) designed to achieve
exactly that. This technique takes a conventional laser pump beam a t any
wavelength (Nd:glass, Ti:sapphire or KrF) and amplifies a large bandwidth
of the chirped pulse, resulting in 10-20x shorter pulses after recompression
than with the usual CPA scheme. Thus it is possible that the PW-class
lasers may be upgradable to the 10 P W level via OPCPA without too much
additional cost or reconstruction.