Optical pulse amplifier

Abstract

There is described a laser pulse generator, comprising: one or more seed laser sources for generating a plurality of sequentially-timed seed pulses; input optics arranged to direct the plurality of sequentially-timed seed pulses through gain medium; the gain medium configured to be pumped by pump diodes to increase the energy of the pulses; and synchronisation optics arranged to guide the plurality of pulses from the gain medium along respective delay paths to temporally synchronise the pulses at one or more targets. Spatial recombination may also be included. Corresponding methods are described.

Description

[0001] Optical Pulse Amplifier

[0002] The present invention relates to generation and amplification of optical pulses. In particular, the invention relates to generation of high intensity short pulses such as nanosecond pulses for use in inertial confinement fusion.

[0003] Introduction

[0004] In many applications there is a trade-off between the optical power of pulses, the duration of the pulses and the damage threshold of components in the optical chain generating the pulses. The physics or engineering of the application also dictates the power or energy in the pulses and the duration. For example, inertial confinement fusion requires high energy pulses having a duration of the order of nanoseconds or tens of nanoseconds. Other applications have other requirements on the laser pulses.

[0005] For inertial confinement fusion applications two of the key difficulties in generating pulses with sufficiently high energy and the correct duration are: i) the requirements on pump diodes; and ii) the damage threshold of optical components such as optical gain media.

[0006] For efficient pumping of optical gain media the energy supplied by the pump diodes is required to be supplied within timescales approximately that of the fluorescence lifetime of the gain media. Pumping for durations significantly longer than the fluorescence lifetime increases the losses of the pumped energy in the gain media. Based on current pump diode costs and laser designs, pump systems for inertial confinement fusion could cost £1 Obillion for the diodes alone. As current pump diode costs are proportional to their output optical power, schemes to reduce the power requirement are highly desirable.

[0007] The laser amplifier materials which typically have the longest fluorescence lifetimes, and hence are suitable for the reduction of expenditure on pump diodes, usually also have relatively high saturation fluences. In order for a large proportion of the energy stored within the amplifier to be extracted, the fluence of the amplified laser pulse is usually preferred to exceed the saturation fluence by at least 5 times.

[0008] Damage thresholds of optical components in the nanosecond pulse regime are specified in terms of optical fluence which is a measure of the laser energy per optical area (J / cm2). Above a certain fluence threshold optical components and materials are damaged by laser light. In the case of laser amplifier materials which have long fluorescence lifetimes, and hence high saturation fluences, the aforementioned preference for the amplified laser pulse to exceed the saturation fluence by at least 5 times may create a

[0009] 17215838.MDE.MDE conflict with the damage threshold of the optical components; with current laser designs it may not be possible to achieve this fluence requirement without damaging the optics. This results in limitations and added costs to high energy short pulse generation systems, such as nanosecond pulse amplification systems for inertial confinement fusion.

[0010] US 11 ,296,478 B2 describes techniques for scaling the average power of high energy solid-state lasers to high values of average output power while maintaining high efficiency. However, the techniques describe propagating many pulses through gain medium before decay due to the fluorescence lifetime, but the techniques do not create laser pulses with the characteristics required for the applications considered here, such as high pulse energy with modest (few per second) pulse repetition rates.

[0011] Accordingly, there is a need to provide improved pulse generation and amplification techniques for high energy short period pulses.

[0012] Summary of the invention

[0013] The present invention may be refered to as “Sequential Pulse Amplification and Resynchronisation”. The present invention provides a laser pulse amplifier, such as for generating or amplifying nanosecond laser pulses or beams. The pulses may be particularly useful for laser inertial confinement fusion. The pulse generator or amplifier comprises: one or more seed laser sources for generating a plurality of sequentially-timed seed pulses; input optics arranged to direct the plurality of sequentially-timed seed pulses through gain medium; the gain medium is configured to be pumped to increase the energy of the pulses; and synchronisation optics arranged to guide the plurality of pulses from the gain medium along respective delay paths to temporally synchronise the pulses once amplified. For example, the pulses may be temporally synchronised for arrival at one or more targets. If spatial recombination is required for the application, this can be performed by directing or focussing the pulses / beams to approximately the same location. We use the terms pulse amplifier and pulse generator interchangeably, although it is understood that a pulse generator may be considered to comprise an amplifier along with seed lasers, for example. We refer to pulses and beams interchangeably. The pulse generator or amplifier may be used for generating high energy, short pulses, such as nanosecond pulses. For example, pulses may have a duration between 1 and 1000ns but more preferably have a duration of 10s of nanoseconds. We describe a pulse generation cycle above. The gain medium increases the energy in the pulses without substantially changing the individual pulses temporal or spatial size. The pulse generation cycle may be repeated multiple times per second such as 10 times per second. By generating multiple sequential pulses,

[0014] 17215838.MDE.MDE amplifying them and synchronising them it is possible to generate a pulse that has an optical fluence that is greater than the damage fluence threshold of the gain material for an individual pulse. This is because the optical energy is spread out over time, which has the effect of increasing the damage threshold of the optical materials.

[0015] The input optics and one or more seed laser sources may be arranged to guide each of the sequentially-timed seed pulses along spatially different paths, i.e. creating multiple laser beams. The spatially different paths may be angularly separated. The spatially separated paths may be arranged such that the sequentially-timed seed pulses are incident on substantially the same portion or portions of gain medium. By having the pulses propagate on spatially separate paths they can be readily separated so that after amplification they can be synchronised. In this scenario the beams may also have different wavelengths.

[0016] The synchronisation optics may comprise mirrors arranged to guide the pulses along different respective delay paths.

[0017] Alternatively, instead of propagating the seed pulses along spatially separate paths through the gain medium, they may have different wavelengths, propagate along the same path, and be separated by wavelength filtering. The one or more seed laser sources may comprise at least two seed lasers sources configured to output different wavelength pulses. Alternatively, the one or more seed laser sources may comprise a source, such as a broadband source, configured to output multiple wavelengths. If different wavelengths are used, then the input optics may be arranged to guide the pulses along substantially the same paths through the gain medium.

[0018] The synchronisation optics may comprise one or more filters for spatially separating the different wavelength pulses and guiding them along different respective delay paths.

[0019] The pulse generator may further comprise multi-pass optics arranged to direct each pulse through the gain medium a plurality of times.

[0020] The multi-pass optics may comprise one or more sets of mirrors, each set of mirrors comprising a pair of mirrors arranged to direct each pulse for a return pass through the gain medium. The multi-pass optics may also include other components such as relay-imaging telescopes, spatial filters, etc.

[0021] The input optics may be arranged to guide each of the sequentially-timed seed pulses along spatially different paths. The mirrors of the multi-pass optics may be arranged to guide each of the seed pulses through substantially the same region of gain medium. The multi-pass optics may be arranged around the optical axis of the gain medium to direct the seed pulses multiple times through the gain medium. In one example, the multi-pass

[0022] 17215838.MDE.MDE optics may be arranged in a circle, square, or other shape with an axis through the circle, square or other shape passing through the gain medium.

[0023] The one or more seed laser sources may comprise a plurality of seed laser sources. Each seed laser source may generate one, for example seed pulse per pulse generation cycle. A controller is configured to turn-on and off the seed laser sources to sequentially time the seed pulses. Alternatively, delay paths may be used to sequentially time the seed pulses to the gain media.

[0024] The one or more of the seed lasers sources may be arranged to generate pulses at different wavelengths, or a broadband pulse(s) may be used.

[0025] The pulse generator may further comprise splitter optics arranged to divide each of the seed pulses in to a plurality of sequentially-timed seed pulses.

[0026] The one or more seed lasers may be configured to generate seed pulses and splitter optics may be arranged to divide each of the seed pulses in to a plurality of sequentially-timed seed pulses. This may be achieved using one or more beam splitters, in which case the spectral content of the seed is conserved, or by splitting a broadband seed or seeds spectrally using filters.

[0027] The input optics and one or more seed laser sources may be arranged to direct the seed pulses through the gain medium such that seed pulses are sequentially timed to form a train of pulses. In other words, the seed pulses follow each other through the gain media one after the other, preferably with no overlap. In the case of ‘square’ pulses (which have approximately constant peak power) any overlap should be less than 1 / 2 of the peak power levels of the pulses.

[0028] The pulse generator may further comprise multi-pass optics arranged to direct each pulse through the gain medium a plurality of times. The multi-pass optics may be arranged to have path lengths such that after the plurality of pulses have passed through the gain medium a first time, the pulses pass through the gain medium a second time, or more. Preferably, the pulses pass through a second time after all of the pulses have passed through a first time. Subsequent passes may be similarly timed.

[0029] In order to reduce the cost of the pump diodes, it is desirable for the gain medium to have as long a fluorescence lifetime as possible. We refer to gain media to indicate that there may be multiple slabs of gain material.

[0030] The gain medium should preferably have a long fluorescence lifetime (Tm:YLF is an example with Tfiuor~15ms) which allows the pump pulses to be provided over a longer period of time (for example, compared to other gain media such as Nd:glass (Tfiuor~240us) or Yb: YAG (Tfiuor~1 ms)) and hence the pump diodes can be lower power for a given

[0031] 17215838.MDE.MDE quantity of stored energy. For a given amount of stored energy, the approximate diode power requirement can be calculated as follows: Pdiodes 2 — P diodes 1 * (Tfluor 1 / Tfiuor 2) , where Pdiodes i refers to the diode power requirement for material I, and Tfiuori refers to the fluorescence lifetime of material i.

[0032] In order to extract a large fraction of the energy stored within the amplifier, the fluence of the amplified laser beam should exceed the saturation fluence of the gain medium, preferably by at least 5 times, as described by the Frantz Nodvik equation. The pump fluence should also exceed this but this requirement is less challenging due to the far longer pulse duration. In the case of gain media with high saturation fluences, this can mean that the fluence required for efficient extraction greatly exceeds the optical damage threshold, For example in the case of Tm:YLF, which has a saturation fluence of 20 J / cm2, the fluence requirement for efficient extraction is >100 J / cm2far exceeding the typical damage limit for 10 ns pulses of ~10 J / cm2, implying efficient extraction may not be possible from such gain media. However, the optical damage threshold increases as the pulse duration increases (specifically as the square root of the pulse duration) thus by using multiple sequential beams (or separable pulses), which therefore have a longer total duration, the damage threshold of the amplifier optics can be greatly increased. If a sufficient number of beams, with sufficient total duration are employed, the fluence requirement for efficient extraction can thus be met. The pulse generator may further comprise the pump diodes and a controller. The pump diodes may be turned on to pump the gain medium for a period in the range 0.2 to 50ms for each pulse generation cycle, but modelling suggests approximately the fluorescence lifetime is optimal. The relatively long fluorescence lifetime of the gain medium allows the material to be pumped over a longer period. This decreases the output power requirements of the pump diodes, which also significantly decreases their cost, which are currently proportional to total output diode power.

[0033] The extraction phase during which the pulses are passing through the gain medium may be a period of 5ns to 5ms.

[0034] The pulse generator may be configured such that an output pulse is generated on a ‘single shot’ basis, potentially up to 1 / TfiUOr times per second, although this may be limited to lower extraction frequencies by heat extraction considerations. This design is most advantageous for applications requiring high repetition rates and high extraction efficiency.

[0035] The present invention further provides an inertial confinement fusion facility comprising the pulse generator set out herein. For inertial confinement fusion the target or

[0036] 17215838.MDE.MDE targets may be the fuel. The pulse generator may also be used in other applications such as space debris removal, defence related directed energy systems, secondary particle or light sources, etc.

[0037] The present invention provides a method of generating laser pulses, such as high energy, nanosecond laser pulses. The method comprises: generating a plurality of sequentially-timed seed pulses; pumping gain medium by pump diodes; directing each of the plurality of seed pulses through the gain medium, optionally with multiple passes; and delaying the pulses (or the plurality of pulses) to synchronise them once directed to the target or to form a single, temporally combined pulse.

[0038] The method may further comprise directing each of the sequentially-timed seed pulses along spatially different paths to the gain medium.

[0039] The spatially different paths may be angularly separated and may be arranged such that the sequentially-timed seed pulses are incident on substantially the same portion(s) of the gain medium.

[0040] The step of generating a plurality of seed laser pulses may comprise generating a plurality of seed laser pulses at different wavelengths.

[0041] The method may further comprise filtering the different wavelengths to spatially separate them to direct them along different delay paths to synchronise the pulses.

[0042] The method may further comprise directing each pulse through the gain medium a plurality of times.

[0043] The step of directing the seed pulses through the gain medium may comprise directing the seed pulses such that they are sequentially-timed to form a train of pulses.

[0044] The method may further comprise directing all of the plurality of pulses through the gain medium a first time and subsequently directing the pulses through the gain medium a second time, or more.

[0045] The method may comprise generating an output pulse up to a plurality of times per second.

[0046] The present invention further relates to use of the methods described herein in inertial confinement fusion reactions.

[0047] Brief description of the drawings

[0048] Embodiments of the invention will now be described, with reference to the drawings, of which:

[0049] 17215838.MDE.MDE Figure 1 is a schematic diagram of a laser pulse generator apparatus for generating high intensity short pulses;

[0050] Figure 2 is a schematic diagram of a laser pulse generator apparatus which includes multiple passes of each of the pulses through gain media;

[0051] Figure 3 is a timing diagram showing the timings of pulses from three seed lasers passing through the gain media of the amplifier four times;

[0052] Figure 4 is a schematic power-time graph comparing the pump power and durations of Nd:glass and Tm:YLF;

[0053] Figure 5 is a graph showing the saturation fluence of Tm:YLF as a function of wavelength for TC- and a- crystal orientations;

[0054] Figure 6 is a timing diagram showing the stored energy within the gain medium as a function of time;

[0055] Figure 7 is a flow-chart showing a method of generating high fluence, short, laser pulses;

[0056] Figure 8 is a graph showing the Tm:YLF emission and absorption spectra for the crystal orientations o and TC; and

[0057] Figure 9 is a schematic diagram showing Brewster rotation of gain medium relative to pulse propagation direction.

[0058] Detailed description of embodiments

[0059] Figure 1 is a schematic diagram of a laser pulse generator apparatus 100 for generating high intensity short pulses, such as may be required for inertial confinement fusion. The apparatus 100 comprises a plurality of seed lasers 112, 114, 116 which generate seed laser pulses. Each of the pulses has a pulse length equal to the desired pulse length at the output. The figure shows three seed lasers but other numbers of seed lasers may be used such as two or four or more. Later in the description we will also describe an alternative to having a plurality of seed lasers. The timing of generating the seed pulses may be controlled by controller 105. The seed pulses are directed by input optics 122, 124, 126, to an amplifier 130 comprising gain media. The input optics may comprise mirrors and / or beam or pulse conditioning optics. Not shown in figure 1 , the input optics may comprise a telescope arrangement to expand the beam size to that desired for passing through the gain media, for example, such that it optimally covers the available area of the gain media. The input optics may provide a collimated beam directed to the gain media. The input optics or mirrors preferably guide the seed pulses through the gain medium along different angular paths which, due to the small differences in angle, result in

[0060] 17215838.MDE.MDE the seed pulses encompassing essentially the same spatial regions in the gain medium. However, so that the seed pulses are not affected substantially differently by the gain media, it is desirable that the pulses pass through the same regions of gain media. Hence, the pulses are directed along paths having angular separation but that converge at the amplifier, such as at the centre of the amplifier, to pass through substantially the same region. In the figure the angular separation may be considered to be shown as 10-20°, but this is just the schematic representation and the actual angular separations for realistic designs are expected to be substantially less than this. We note that while the amplifier head geometry illustrated in figures 1 & 2 is a transmissive design, then scheme is equally applicable to reflective amplifier designs.

[0061] The amplifier may comprise slabs of gain medium and may be liquid or gas cooled. In the figure seven slabs of gain medium are shown but other numbers of slabs may be used. The gain slabs may be contained in a manifold having windows, so as to allow the gas or liquid cooling. Not shown in figure 1 the gain slabs are pumped by pump diodes. The pump diodes may be an array of diode stacks, each stack consisting of an array of bars, each bar made up of multiple diodes on one chip. The diode bars provide a brief pulse of pump energy and then switch off until the next pulse is required (at the next pulse generation cycle), which may be around 1 / 10thof a second later.

[0062] At the output of the amplifier are synchronisation optics which comprise delay paths 152, 154, 156. Optionally, the synchronisation optics may comprise mirrors 142, 144, 146, arranged to guide the pulses from the output of the amplifier to respective delay paths. As shown in the figure, after passing through the gain media of amplifier 130, the pulses have a greater intensity and substantially the same temporal length. The delay paths 152, 154, 156 add a delay to each of the paths so that the pulses can be synchronised to arrive at a target 160 (or multiple targets) at the same time. The delay paths may comprise a straight or folded path length. For pulse lengths of the order of nanoseconds and each pulse timed to pass sequentially through the gain media with each following closely behind the previous, the delay paths would need to be of the order of metres for each pulse. For around 10 pulses the maximum delay may be up to hundreds of metres. The delays paths may be straight lines in free space or a vacuum, or may be folded paths using mirrors. The pulse arriving last may not have to travel along a specific delay path. The seed pulses are timed to pass through the gain media sequentially. By temporally spreading out the pulse energy the gain media is less likely to be damaged, as we will discuss in more detail later. Preferably, the pulses are timed such that they do not overlap but also may have little or no

[0063] 17215838.MDE.MDE temporal gap between them. The latter requirement reduces the space the laser beamline occupies.

[0064] The arrangement of multiple sequential seed lasers enables a reduced flux through the gain media reducing the likelihood of damage. We will describe this further in the following sections. We also describe passing the seed pulses repeatedly through the gain media to increase pulse energy and also how efficient pumping and extraction is achieved for the gain media.

[0065] Above a certain fluence threshold (laser energy per unit optical area) optical materials are damaged by laser light. This fluence threshold is typically determined experimentally. It has been found in the nanosecond pulse duration regime that the threshold scales with laser wavelength and laser pulse duration, such that:

[0066] F(2, T2) = F(XbTI) (X2 / Xi) (T2 / TI)1 / 2(1 ) where F( i, Ti) is an experimentally measured optical damage fluence threshold at a certain reference wavelength i and laser beam pulse duration TI, and F(k2, T2) is the scaled damage fluence threshold at a new wavelength2and laser beam pulse duration T2. Thus, for longer wavelengths and / or longer laser pulse durations, the optics’ damage fluence threshold increases.

[0067] Gain Medium Saturation Fluence

[0068] The saturation fluence (Fsat) of a given laser gain material dictates the laser fluence which is required in order to reduce the gain to ~36.7% (1 / e) of its original value. The gain (G = Fout I Fn , where Fin, and Fout are respectively the fluence of the incident and exiting beam from the amplifier) is dictated by two attributes, namely: the laser fluence, and the stored fluence within the gain medium, both of which are normalised by the saturation fluence, as described by the Frantz Nodvik (FN) equation:

[0069] Gout = log ( 1 +exp(Gstored)(exp(Gin)-1 )) (2) where:

[0070] Gstored = Fstored I Fsat

[0071] Gin = Fjn I Fsat

[0072] Gout = Fout / Fsat

[0073] 17215838.MDE.MDE and Fstored, in, and Foutare respectively the fluence stored in the gain medium, the fluence of the incident laser beam, and the fluence of the beam exiting the amplifier. The implications of the FN equation on laser design are as follows.

[0074] On the first pass of a laser pulse through the amplifier, when Finis likely to be very small relative to Fsat, gain will only significantly exceed unity if Fstored is sufficiently larger than Fsat, i.e. Gstored is large. However, on subsequent passes through the gain medium, Fstored will have decreased as energy will have been transferred to the laser pulse, so Gstored will reduce. Thus, for later passes, when Fstored is small, in order to extract a substantial fraction of the initially stored energy (requiring gain above one), Finmust be substantially greater than Fsat- In other words Gin has to become large to compensate for the reduction in Gstored caused by energy extraction.

[0075] For gain media with saturation fluences sufficiently below the optical damage threshold, the latter requirement (Fin> Fsat ) does not present a problem, however other gain media which have desirable properties (examples are discussed in more detail below) have saturation fluences substantially higher than the optical damage fluence threshold. This implies that efficient extraction from gain media with high saturation fluences may not be possible without damaging the optics, which would render the laser inoperable.

[0076] As described with reference to figure 1 the use of multiple seed laser beams or pulses for generating each output beam or beamline has advantages increasing the total combined duration of the laser beams within the gain media, thereby increasing the damage threshold, as per equation 1 . However, to be able to temporally synchronise the pulses or beams, each laser beam or pulse must be in some way distinguishable from the others such that once the beams or pulses have been amplified they can then be separated from one another and synchronised temporally. In the arrangement shown in figure 1 , the pulses are propagated through the gain media with a difference in the angular propagation directions of the pulses. Alternatively, different wavelength pulses or beams could be used and spectrally selective optical components such as wavelength selective mirrors or filters, or dispersive components, such as gratings or prisms, used to separate the different pulses for synchronisation after amplification. In another alternative, a fast optical switch may be able to achieve a similar effect.

[0077] In the arrangement of figure 1 , the amplifier is shared by all the beams of the beamline. The various beams or pulses are ‘launched’ one after another and are timed such that they propagate through the gain-medium sequentially. As we will describe with reference to figure 2, the pulses can then perform multiple passes through the amplifier.

[0078] 17215838.MDE.MDE For example, the number of passes may be determined based on the detailed beamline design, but may for example be two or more passes. As shown in figure 2, for each beam an arrangement of mirrors may be used to propagate the beams or pulses through the gain media multiple times. Lenses, Pockels cells and / or waveplates may also be used (although not shown in figure 2). The beams or pulses are arranged so that within the gain media of the amplifier they propagate through the same transverse region of the media, or as close as possible given angular constraints and limits on alignment precision.

[0079] Figure 1 shows three seed lasers 112, 114, 116, each generating a seed pulse such that three pulses are directed at the gain media. Other numbers of seed lasers may be used such as to generate a higher number of pulses such as up to 10 or 12 pulses or more. Controller 105 may control the timing of the seed pulses such that they are generated sequentially. The controller may also control the length of time each seed laser is turned on, the power vs time profile of the pulses, and how often the pulses are turned on per second to determine the repetition rate which may for example be 10Hz. In an alternative arrangement, a single seed laser may generate a single seed pulse which is then divided into multiple seed pulses. For example, this may be done using beamsplitter optics, or wavelength-discriminating filters. The pulses may then be delayed by respectively increasing amounts to generate the train of sequential pulses. It is generally more flexible to use multiple seed lasers, but using a single seed laser to generate multiple seed pulses may be more cost effective. A combination of the two approaches may also be used such that each seed laser may generate a small number of seed pulses but multiple seed lasers are provided. For example, three seed lasers could generate three pulses, each of which is split into four to generate twelve pulses in total. In this arrangement the controller also controls various timing aspects of the seed pulses, such as duration and repetition rate.

[0080] Referring to figure 2 in more detail, two seed pulses are shown generated by two seed lasers 112, 114. We have shown only two seed lasers in figure 2 for convenience of illustration but other numbers may be used such as three or more. Similarly to figure 1 the seed pulses are directed by input optics 122, 124, such as mirrors, to the gain media of amplifier 130. On exiting the gain media for the first time, instead of being incident on synchronisation optics to direct the pulses to delay paths as in figure 1 , the pulses are incident on multi-pass optics, such as mirrors, to direct the pulses through the gain media at least once more. In figure 2, the multi-pass optics comprise a pair of mirrors. For example, for the first seed pulses, from seed laser 112, the pulses pass through the gain media to mirrors 222a and 222b which reflect the pulses such that they are directed on a return path to the gain media. As illustrated in figure 2, the mirrors 222a and 222b act like a

[0081] 17215838.MDE.MDE reflector returning the pulses on a slightly different path than that which they first passed through the amplifier, although in alternative configurations they may re-trace the same path on subsequent passes. After the second pass through the gain media the pulse from seed laser 112 is incident on mirror 242, which may be considered part of synchronisation optics because it directs the pulse to the delay path 252 and on to spatially combine the pulses at the target 160 in a similar way to figure 1 . Seed pulse from second laser 114 also passes through gain media in a similar way to figure 1 and is then incident on a pair of mirrors 224a and 224b which return the pulses back to the gain media, in a similar way to the first pulse. After passing the second time through the gain media the pulse is then incident on mirror 244 which like synchronisation optics of figure 1 directs the pulse to the delay path and to target. Although figure 2 shows two pulses, each passing through the gain media two times, a greater number of pulses and passes may be used. In one embodiment 12 seed lasers generate 12 seed pulses, and each passes through the gain media four times.

[0082] At the left of figure 2 there are shown eight circles surrounding a ninth shaded circle. This is an illustration of how multiple mirrors may be arranged to direct multiple beams or pulses or for multiple passes, as viewed end on looking past the gain media which is schematically shown by the shaded circle in the centre. For example, as shown by the arrows, on a first pass a pulse might arrive at the mirror just to the right of the top of the circle, be reflected to the mirror just to the left of the bottom of the circle and is then reflected back towards the gain media for a second pass through. The circle of mirrors comprises eight mirrors and only two of them have been used by the first pulse. The other six mirrors may be used to direct second, third and fourth pulses such that they also pass twice through the gain media. For example, as shown in the figure pulses from the second seed laser 114 are incident and reflected by mirrors 224a and 224b. The use of 2 passes through the gain media is illustrative and it is likely more passes would be employed. For more pulses or more passes, additional mirrors may be arranged in a circle or in concentric circles. Such arrangements of mirrors may be used at both sides of the gain media if more than two passes through the gain media are desired. Other arrangements of mirrors or reflectors may be used. The circular arrangement is a convenient and compact arrangement but other arrangements of mirrors may be used, for example square beams may benefit from linear arrangements.

[0083] Figure 3 shows the timings of pulses from three seed lasers passing through the gain media of the amplifier four times. A pulse P1 from first seed laser passes through the gain media first, sequentially followed by pulses P2 and P3 respectively from second and

[0084] 17215838.MDE.MDE third seed lasers. As shown in the figure it is likely preferable if the pulses P1-P3 closely follow each other in time. After this first pass the pulses are returned to pass through the gain media a second time in the opposite direction (as indicated by the arrows). Following this second pass, the pulses pass through the gain media a third time and a fourth time. With each pass the pulses are amplified increasing their power. Note that in this simplified illustration, each pulse’s power (i.e. an individual pulse’s power as a function of time, or ‘shape’) is shown as constant as a function of time for clarity. Furthermore, the effects of gain depletion, which in reality would mean that subsequent pulses would require different initial power profiles is also not accounted for in this illustration. In an alternative implementation, the order of the beams’ passes through the amplifier could be inverted after each pass to more evenly balance the effect of pump depletion between the beams, although this would require a longer path length.

[0085] As the laser pulses propagate sequentially through the amplifier, from the perspective of optical damage, they are effectively a single long-duration laser pulse. For a beamline with a given number of beams (nbeams) each of which has duration Tbeam, the effective total laser pulse duration is nbeamsTbeam. Thus in equation (1) above,

[0086] T2 = nbeams'Cbeam which means that for a given Tbeam, the optical damage threshold in the amplifier will scale as ribeams1 / 2

[0087] By increasing nbeanis above unity, the optical damage threshold within the amplifier will be increased. This means that the laser system can operate at a higher fluence than that which would normally be possible before incurring amplifier optics damage. This will reduce system costs, as the optics area is a key driver of cost.

[0088] A further advantage relates to the range of laser gain media which can be employed whilst achieving efficient extraction. In a conventional (single beam) laser beamline, the requirement for laser fluence to be substantially above the damage threshold would imply that efficient extraction cannot be achieved with some high saturation fluence gain-media without damaging the optics. However, equation 1 tells us that a longer laser pulse duration will increase the optics damage fluence threshold. Therefore, if the beam duration is increased it may become possible to achieve efficient extraction using materials that it has not previously been possible to use.

[0089] 17215838.MDE.MDE System physical requirements determine the laser pulse duration which is required. For example, in laser inertial confinement fusion a typical pulse duration might be 10-25 nanoseconds (ns). While this duration can and does vary, increasing it by an order of magnitude for example would enable an approximately 3.1 times increase in damage threshold fluence. However, such an increased pulse duration is not compatible with almost all laser inertial confinement fusion designs, other than the highest-energy implosion designs which in-turn would require a large and likely very expensive laser system.

[0090] The present invention addresses the pulse duration limitation by using multiple laser beams per beamline. Each beam has a pulse duration matched to the physics of the system in which it is to be used, but the sequence of laser beams or pulses has a total pulse duration which is a factor of nbeams times longer, thereby increasing the optics damage fluence threshold by nbeams, / 2.

[0091] Using this technique, extremely efficient energy extraction from the gain medium may be achieved whilst remaining below the damage threshold, even when using high saturation fluence materials. We describe below an example of using a high saturation fluence material, namely one containing thulium dopant. After sufficient energy has been extracted from the gain media of the amplifier, the beams or pulses exit the amplifier after their final pass.

[0092] Pump Diode Cost and Fluorescence Lifetime

[0093] We have described above that the gain medium is pumped to provide the energy for amplifying the pulses as they pass through it. Such pumping is preferably performed by diodes due to their high electrical-to-optical efficiency, which may approach 60%.

[0094] The total peak power the diodes must supply (Pdiode) is equal to the energy required (Eiaser) divided by the pump duration (tpUmp), that is

[0095] P diode = Elaser tpiimp-

[0096] Therefore, increasing the pump duration tPumPwill decrease the power Pdiode the diodes are required to supply. The cost of such diodes is directly proportional to the diode peak power Pdiode, so any reduction in Pdiode will reduce cost.

[0097] For high-energy, high-efficiency laser systems the ability to reduce the diode power is important because the cost of pump diodes is expected to be the single largest cost. For example, based on current costs in 2024 of approximately GBP 2 / per Watt, a laser inertial confinement fusion plant could spend GBP 10 billion on pump diodes alone.

[0098] There are a wide range of potential applications for such lasers, including but not limited to, laser fusion research facilities or power-plants, such as pulsed laser defence

[0099] 17215838.MDE.MDE systems, laser peening, etc. Any reduction in the required diode pump power, and hence costs, will have important consequences for the commercial viability of these and other applications requiring high-energy, high-efficiency lasers.

[0100] For laser systems which require high energy-efficiency, the pump duration tpumpis determined by the gain-medium’s fluorescence lifetime, tfiUOr- The fluorescence lifetime is a measure of how long pump energy is stored within the gain medium. The amount of stored energy decays exponentially and after a time tfiuorhas passed, only 36.7% (1 / e) of the energy stored will remain. Hence, for a laser to be energetically efficient, the pump duration should be short and preferably less than tfiUOr, whereas to keep the cost of pump diodes low, the pump duration should be long. While exact values will depend on the cost of diodes etc., an optimal pump duration balancing costs against efficiency is preferably when the pump duration is approximately equal to the fluorescence lifetime, that is tPumP~ tfiuor. Looking at it another way, for a given stored energy, the diode costs are approximately inversely proportional to the fluorescence lifetime of the amplifier gain medium.

[0101] Embodiment using Tm:YLF gain material

[0102] We now describe an example embodiment which uses a thulium (Tm) doped laser gain medium, specifically Tm:YLF, where YLF is the ‘host’ material. Tm:YLF has a relatively long (15ms) fluorescence lifetime in comparison to most other gain media. Table 1 below compares properties of three gain media materials.

[0103] Gain medium Fluoresence lifetime Diode cost relative Saturation material (Us) to Tm:YLF fluence (J / cm2)

[0104] Tm:YLF 15,000 1 17 (@1900nm & )

[0105] Yb:YAG @ 140K 950 16 2.8 Nd :G lass 240 62 ~1

[0106] Nd:glass has a much shorter fluorescence lifetime compared to Tm:YLF, by around 60 times. As shown in figure 4, this means that the length of time the diodes are turned on is reduced, and hence, the power required for pumping Nd:glass is much greater (shown by 1 in figure 4) as compared to Tm:YLF (shown by 2 in the figure) which has a longer fluorescence lifetime, even though the same amount of energy is stored in both cases, which is represented by the area of the rectangle in figure 4.

[0107] Tm:YLF has a high saturation fluence in comparison to many other gain media. As discussed previously, the saturation fluence effectively determines the laser fluence within

[0108] 17215838.MDE.MDB the gain medium which is required to extract a large proportion of the energy stored in the amplifier. If the stored energy cannot be extracted, the laser will be energetically inefficient, and hence not suitable for many applications. Preferably, the fluence of the laser light passing through the gain medium should be at least 5-6 times the saturation fluence for efficient extraction. However, if the laser fluence passing through the gain medium is too high, the light will damage the laser’s optics, requiring expensive repairs. Therefore, current lasers use a gain medium with a low saturation fluence (a few J / cm2) in order to maintain the laser fluence below the damage threshold (~3J / cm2for dielectric coated optics).

[0109] Figure 5 shows the saturation fluence of Tm:YLF as a function of wavelength for JI- and o- crystal orientations. The figure shows that although the saturation fluence varies, it is generally around 20J / cm2for the o-orientation over a range of wavelengths from around 1735 to 1900nm. This level of saturation fluence is well above optical damage thresholds even before we exceed the saturation fluence by 5-6 times to achieve efficient energy extraction.

[0110] By using the technique of the present invention we use multiple beams per beamline (nbeams). Each beam has a duration (Tbeam) matched to the physics of the system. The duration between the start of each beam being ‘launched’ (xiaunch) preferably exceeds or is equal to Tbeam in order to ensure the beams propagate through the amplifier sequentially and do not overlap in space. Therefore the total effective beam duration within the amplifier is given by

[0111] ^amplifier = nbeams'Tlaunch

[0112] From equation 1 we see that the optics damage threshold scales as TamPiifier1 / 2. Thus, the damage threshold within the amplifier can be increased by increasing the number of beams, or the time between the start of each beam being launched, or both. This is illustrated in figure 6, which shows the stored energy within the gain medium as a function of time. The timings used in figure 6 are representative of those which may be used for Tm:YLF, with different ‘diode on’ durations required for gain media with differing fluorescence lifetimes. From time 0ms to a time of around 15ms the pump diodes are turned on and they are injecting energy to be stored in the gain medium. The stored energy increases up to around 15ms when the extraction phase commences. Multiple beams or pulses are then sequentially incident on the gain medium which can be seen in the zoomed in view as a series of steps reducing the stored energy. In the figure the number of steps reducing the stored energy is the number of beams or pulses passing through the gain medium. The length of each step is the duration of each beam or pulse. In this illustration,

[0113] 17215838.MDE.MDE only one energy extraction pass is shown for simplicity. The total extraction duration in this scheme is nbeamsTbeamnpasses, where nPasses is the number of times the beams pass through the amplifier (this assumes there is no additional delay between passes). In this scheme the extraction phase is nbeams times longer than in a conventional laser system. This means that a substantially higher fluence can occur within the amplifier without damage occurring. Note that in this simplified illustration, each beam is assumed to deplete the gain medium equally, in reality the depletion caused by previous energy extraction has to be accounted for in order to equalise the beams’ power profiles at the output.

[0114] As can be seen in the figure the extraction phase commences at around 15ms, which is also the fluorescence lifetime of the Tm:YLF material. This timing is used because it is the approximate optimal balance between efficient storage, which requires as short a pump duration as possible to minimise fluorescence losses, with cost, which requires as long pump duration as possible.

[0115] Figure 7 is a flow-chart of the method of the present invention based on the apparatus set out in figures 1 and 2. The method commences at step 710 by pumping the gain medium and at step 720 by generating seed laser pulses. As discussed previously, these pulses may be generated from separate sources and launched at times such that one starts immediately after an earlier one ends. Hence, the pulses should be launched sequentially in time. The seed pulses are directed at the gain medium at step 730, for example, by input optics. The beams or pulses are preferably directed at the gain medium along angularly separate paths so that they can be separated spatially outside the amplifier head. The beams or pulses pass through the gain medium and are amplified, extracting stored energy from the gain medium. Optionally, at step 740 the pulses may be returned to the gain medium for further passes to extract more energy and further increase the pulses’ energies. At step 750 the pulsesare delayed such that each of the pulses are temporally synchronised. The pulses may then be directed to arrive at a target (or targets) at essentially the same time (within the precision obtainable), as shown at step 760. Alternatively, the pulses may be spatially combined to form a high energy pulse. For nanosecond duration beams incoherent beam re-combination is sufficient, as coherence between beams is not required for efficient laser absorption. There may be applications with shorter pulse durations which would benefit from coherent recombination of the beams. The latter is expected to be more technically challenging.

[0116] An advantage of the amplifier being able to support a substantially higher fluence before damage occurs, is the ability to extract a very high proportion of the energy stored within the amplifier gain medium. We estimate that extraction efficiency may be able to

[0117] 17215838.MDE.MDE exceed 95%, with the limiting factors on efficiency being other losses such as reflections at optical components.

[0118] The total effective beam duration in the amplifier Tamper (= nbeams iaunch) may determine the length of the laser beamlines either side of the amplifier. This is particularly a factor after the pulses leave the amplifier for temporal synchronisation, because space required for the synchronisation will be set by TampiitierC, where c is the speed of light. We can minimise beamline space requirements by setting Tiaunch = Tbeam such that one pulse is launched immediately as a preceding pulse ends, although we note that by setting tiaunch > tbeam the damage threshold will increase according to equation 1 .. At the input side, if the pulses are launched from separate seed lasers to meet the sequential timing requirements through the amplifier, then the space requirements on the input may be minimized. If a single seed laser is used to generate multiple pulses by dividing and delaying an initial seed pulse then the space requirements on the input side will be similar to the output / delay side. To make the most efficient use of space, the beamline can be ‘folded’ either side of the amplifier, using mirrors as previously discussed. For further increased efficiency of space the paths can be folded multiple times.

[0119] We previously discussed that as each beam or pulse propagates through the gain medium sequentially, with a slightly different propagation direction through the gain medium, namely with a slightly different propagation angle. Various different beam-layouts can be employed to achieve this. The variations in beam angle can be less than 1° to achieve the desired effects. With such similar beam angles the beams can effectively fill the pumped region of the gain medium, with maximal overlap, so as to achieve efficient energy-extraction from the amplifier.

[0120] As indicated at step 740, having passed through the amplifier once, the beams may be reflected back through the amplifier using a multi-pass architecture for each beam. In a particular example, each beam or pulse may have four passes through the amplifier. For twelve beams or pulses, i.e. nbeams = 12, and assuming Tiaunch = Tbeam = 25ns this creates a system which, from the perspective of the amplifier, sees a 48-pass single beam system with a combined pulse duration of 300ns.

[0121] At the mirrors used for the multi-pass architecture, the energy and duration of the pulses incident on these optics will differ from the effective total duration and energy within the amplifier. This is because the amplifier effectively sees all of the pulses sequentially thereby forming one longer pulse of greater total energy, compared to the mirrors which only see individual pulses because each of the pulses takes a spatially different path. The number of beams can be used as a design parameter to balance the damage fluence at

[0122] 17215838.MDE.MDE the mirrors and amplifier. The energy on each mirror scales as 1 / nbeams, while the damage threshold within the amplifier scales as (nbeamsTiaunch)172. Hence, the system design should take account of the damage thresholds for both the mirrors and amplifier, which may differ depending, for example, on the coatings used.

[0123] Before the beams leave the system to be sent to a target, they are synchronised in time. If the pulses at the seed laser end have been generated by a single seed, this may be viewed as re-synchronisation. As previously discussed in relation to figure 1 , the synchronisation will be done by having mirrors at different path lengths, thereby resynchronising them before sending the beams to target. Due to the use of nanosecond duration laser beams or pulses there is not usually a need to phase-match the various beams / pulses.

[0124] Figure 8 shows the Tm:YLF emission and absorption spectra for the crystal orientations o and TI. Tm:YLF emits over a broad range of wavelengths between around 1650 and 1950. However, as shown in figure 5 the saturation fluence varies significantly as a function of wavelength. For some systems instead of requiring a single wavelength laser a broad bandwidth laser output is required for the physics of the application. For Tm:YLF attempting to use the full bandwidth for, for example, the Tm:YLF sigma crystal orientation from 1700-1925nm (see o line in figure 5) would be extremely challenging due to the large variations in gain as a function of wavelength which would occur due to the variations in saturation fluence with wavelength. To overcome this, the laser system may be seeded at a number of discrete wavelengths which have equal saturation fluence. An example is shown by the intersection of the dashed horizontal line in figure 5 with the a line. In principle a separate, relatively narrow band seed could be used at each intersection point such that each beam has a different wavelength with modest bandwidth. Cumulatively the various beams would have nbeams times the bandwidth of an individual beam. By design each seed would have the same gain because the saturation fluence at each of the wavelengths is equal. The variation in re-absorption with wavelength as shown in figure 8 should also be taken into account. This approach for a broadband source has the additional advantage that, as each beam is relatively narrow-band, efficient frequency up-conversion is possible (which we describe in more detail in the following).

[0125] According to equation 1 , Tm:YLF’s longer emission wavelength, in comparison to the ~1000nm laser light which is more typically used, means that the laser damage threshold will be increased in comparison to the ~1000 nm laser light. However, emission

[0126] 17215838.MDE.MDE around 1900nm may not be suitable for certain applications, such as inertial confinement fusion. Nevertheless, to achieve the desired wavelength, frequency conversion may be used. For example, frequency quadrupling of the 1900nm light to ~475nm can be achieved using two sequences of second harmonic generation (SHG) crystals. Such frequency quadrupling introduces losses but good efficiencies can still be achieved.

[0127] For the example of Tm:YLF, it is proposed to use 793nm diodes to pump the gain medium, because thulium has an absorption peak in this spectral region. This might be expected to have a very low quantum efficiency, however a cross-relaxation process enables one incident pump photon to excite two Tm ions to the upper laser level stimulating emission of two photons, meaning the quantum efficiency is almost doubled, so long as sufficiently high doping levels of Tm are used (> ~2%). Achieving the required doping levels may require co-doping with aluminium.

[0128] For wavelength or frequency conversion it may be desirable to spatially combine the pulses into a single pulse for passing through the SHG crystal, assuming it can withstand the optical fluence. The combination into a single pulse may be achieved by a combination of mirrors, combiner cubes and prisms. Hence, having started with a number of seed pulses which are temporally and spatially separated, after combining, the pulses are temporally and spatially aligned providing an output pulse with greater intensity. Combining the amplified pulses in this way may also have other applications. If the optical fluence is too great for the SHG crystal, each amplified seed pulse may be passed through separate SHG crystals and directed to combine at the target.

[0129] In large aperture amplifiers transverse gain is often the limiting factor in achieving high energies and efficiencies. Transverse gain is undesired gain in the amplifier in the direction transverse to the laser’s main propagation axis. Evaluations using Tm:YLF as the gain medium indicate this is manageable using the following techniques:

[0130] 1 . By exploiting the high saturation fluence of Tm:YLF, lots of energy can be stored in a relatively small aperture system, thus reducing the transverse dimension of the amplifier medium while enabling a high-energy system. This high energy is practical within the damage limits due to the use of the scheme we have described.

[0131] 2. Brewster rotation of the gain medium and amplifier windows can be used to reduce the laser fluence incident on the amplifier optics (in the frame of the optics). We describe this in more detail below, but by having the gain medium at an oblique angle to the beams or pulses, as shown in figure 9, the rotation has the effect of increasing the gain medium’s size in one direction. Normally this would mean that the transverse gain would be

[0132] 17215838.MDE.MDE worse in the direction of the gain medium which has the larger dimension. Tm:YLF is however a bi-refringent crystal, which means that by orienting the crystal correctly, the smaller crystal gain coefficient can be aligned with the longer direction of the crystal, while that in the shorter crystal direction is aligned with the larger gain coefficient. In this way, a high energy design is possible while controlling transverse gain in both directions within the amplifier.

[0133] 3. If the gain remains too high in one direction, the amplifier crystals could be split in the appropriate direction in order to reduce the path length over which the transverse gain can occur. An index-matched cladding can then be introduced both at the split and on the periphery of the gain medium to absorb any transverse fluorescence.

[0134] We mentioned Brewster rotation in the above paragraphs. This may also be used to further increase the damage threshold fluence within the amplifier. Brewster angle is the angle of incidence at an optical surface at which all of the incident light is transmitted when the polarisation of light is parallel to the plane of incidence of the surface (p-polarisation). Brewster-angled gain medium slabs may be used in the amplifier. Windows for the amplifier may also be configured at the Brewster angle. Angling the optics in this way has the following advantages: i) It increases the surface area of the optics, this will: a) Further increase the damage threshold energy because the energy is spread across a greater area; and b) Aid heat extraction ii) It minimises undesired reflections and thus: a) Maintains efficiency; and b) Does not require dielectric coatings which reduce damage thresholds.

[0135] Brewster angled slabs will define a particular polarisation state for the beamline. Optics such as those coated with micro-structured surfaces may offer alternative, or additional, advantageous characteristics. In particular these will not impose a polarisation state that may then result in depolarisation losses.

[0136] We now describe a detailed embodiment using Tm:YLF and having multiple beam passes through the amplifier. The seed wavelength is 1915nm and the pulse duration is 25ns. The beam diameter is 20cm. 12 beams are generated for the beamline and each beam is set to pass through the amplifier four times, resulting in an effective number of 48

[0137] 17215838.MDE.MDE passes through the amplifier. The maximum seed energy is 10mJ. (The seeds may have different energies to balance the beams' energies at the output. This is the maximum energy used.) The Tm:YLF gain medium is pumped with a diode peak power of around 7.9MW, corresponding to a total energy of 118kJ for a 15ms pump pulse duration. For 12 beams, each having a duration of 25ns, sequentially timed means a total pulse duration through the amplifier of 300ns. After amplification, to synchronise the pulses a delay path of up to 90m is used, although this may be folded to reduce size.

[0138] Using reference damage fluences for the amplifier components of 10 J / cm2and 5 J / cm2for the mirrors (both for a 10 ns pulse at 1 micron) and equation 1 , the damage fluence threshold in the amplifier is calculated as 105J / cm2for a 300ns pulse. At mirrors in the system the damage fluence threshold is determined as 15J / cm2for a 25ns pulse. The calculated fluence in the amplifier is 91 J / cm2and at the mirrors is 14J / cm2.

[0139] The resulting output energy per beam is around 4.2kJ at 1915nm and after synchronisation the beamline output energy is around 50kJ at 1915nm, corresponding to a diode input to 1915nm output efficiency of 47%. If the output beamline is upconverted for inertial confinement fusion to 475nm, the beam or pulse energy is 30kJ. Although this is a loss of energy it still represents relatively efficient performance which is also cost effective compared to other pulse generation techniques. Overall the wallplug efficiency at 1915nm is 28% and at 475nm is 18%. In going from an optical-to-optical efficiency of 47% to a wallplug efficiency of 28%, the loss is caused by the diodes’ electrical to optical efficiency, assumed to be 60%.

[0140] As discussed, other gain materials (examples include hosts YAG, YLF, fluorides, garnates, glasses, sesquioxides, vanadates, phosphates, aluminates, and their derivatives and active ions Tm, Yb, Nd and Ho) and wavelengths may be used. For example, Yb:YAG, Yb:YLF or Tm:YAG may be used. Also, instead of a single wavelength, multiple seed wavelengths could be used.

[0141] As discussed the pulse generator of the present invention could be used as part of an inertial confinement fusion power plant or research facility. To generate the required powers for such plants and facilities optical pulses of the order of 1 MJ may be required.

[0142] The person skilled in the art will readily appreciate that various modifications and alterations may be made to the above described apparatus and methods without departing from the scope of the appended claims. For example, different gain medium, such as Yb:YAG, wavelengths or arrangement of beam paths may be used. In particular this

[0143] 17215838.MDE.MDE invention would be appropriate for the efficient extraction of energy from room temperature Yb:YAG, which has a relatively high saturation fluence. The number of pulses and the number of passes through the gain material may be changed. Adjustments to the optical arrangement may also be made without departing from the scope of the present invention, as defined by the appended claims.

[0144] 17215838.MDE.MDE

Claims

CLAIMS:1 . A laser pulse generator, comprising: one or more seed laser sources for generating a plurality of sequentially-timed seed pulses; input optics arranged to direct the plurality of sequentially-timed seed pulses through gain medium; the gain medium configured to be pumped by pump diodes to increase the energy of the pulses; and synchronisation optics arranged to guide the plurality of pulses from the gain medium along respective delay paths to temporally synchronise the pulses at one or more targets.

2. The pulse generator of claim 1 , wherein the input optics and one or more seed laser sources are arranged to guide each of the sequentially-timed seed pulses along spatially different paths.

3. The pulse generator of claim 2, wherein the spatially different paths are angularly separated and arranged such that the sequentially-timed seed pulses are incident on substantially the same portion of gain medium.

4. The pulse generator of claim 2 or claim 3, wherein the synchronisation optics comprise mirrors arranged to guide the pulses along different respective delay paths.

5. The pulse generator of claim 1 , wherein the one or more seed laser sources comprise: at least two seed lasers sources configured to output different wavelength pulses, or a broadband source configured to output pulses having multiple wavelengths.

6. The pulse generator of claim 5, wherein the input optics are arranged to guide the pulses along substantially the same paths through the gain medium.

7. The pulse generator of claim 5 or claim 6, wherein the synchronisation optics comprise one or more filters, gratings or prisms for spatially separating the different wavelengths or different wavelength pulses and guiding them along different respective delay paths.17215838.MDE.MDE8. The pulse generator of any preceding claim, further comprising multi-pass optics arranged to direct each pulse through the gain medium a plurality of times.

9. The pulse generator of claim 8, wherein the multi-pass optics comprise one or more sets of mirrors, arranged to direct each pulse for a return pass through the gain medium.

10. The pulse generator of claim 9, wherein the input optics are arranged to guide each of the sequentially-timed seed pulses along spatially different paths, and the mirrors of the multi-pass optics are arranged such that each beam passes through substantially the same region of the gain medium.11 . The pulse generator of any preceding claim, wherein the one or more seed laser sources comprise a plurality of seed laser sources, each seed laser source generating one seed pulse per pulse generation cycle, and a controller configured to control the seed laser sources to sequentially time the seed pulses.

12. The pulse generator of claim 11 , wherein one or more of the seed lasers sources are arranged to generate pulses at different wavelengths.

13. The pulse generator of claim 11 or claim 12, further comprising splitter optics arranged to divide each of the seed pulses in to a plurality of sequentially-timed seed pulses.

14. The pulse generator of any of claims 1 to 10, wherein the one or more seed lasers are configured to generate seed pulses and splitter optics are arranged to divide each of the seed pulses in to a plurality of sequentially-timed seed pulses.

15. The pulse generator of any preceding claim, wherein the input optics and one or more seed laser sources are arranged to direct the seed pulses through the gain medium such that seed pulses are sequentially timed to form a train of pulses.

16. The pulse generator of claim 15, further comprising multi-pass optics arranged to direct each pulse through the gain medium a plurality of times, wherein the multi-pass optics are arranged to have path lengths such that after the plurality of pulses have passed17215838. MDE.MDEthrough the gain medium a first time, the pulses pass through the gain medium a second time, or more.

17. The pulse generator of any preceding claim, wherein the pulses are nanosecond pulses.

18. The pulse generator of any preceding claim, wherein the gain medium has a fluorescence lifetime of greater than 200ps, greater than 1000ps or greater than 5000|is.

19. The pulse generator of claim 18, wherein the gain medium is Tm:YLF, Yb:YAG, Yb:YLF, or Tm:YAG.

20. The pulse generator of claim 19, further comprising the pump diodes and a controller, wherein the controller is arranged to turn on the pump diodes to pump the gain medium for a period in the range 0.2 to 50ms for each pulse generation cycle.21 . The pulse generator of claim 19 or claim 20, wherein the extraction phase during which the pulses are passing through the gain medium is a period of 5ns to 5ms.

22. The pulse generator of any preceding claim, configured such that an output pulse is generated a plurality of times per second such as up to five times per second, up fifty times per second or more.

23. The pulse generator of any preceding claim, wherein the synchronisation optics are further arranged to spatially combine the pulses into a single output pulse.

24. An inertial confinement fusion facility comprising the pulse generator of any preceding claim.

25. A method of generating laser pulses, the method comprising: generating a plurality of sequentially-timed seed pulses; pumping gain medium by pump diodes; directing each of the plurality of seed pulses through the gain medium; and delaying pulses of the plurality of pulses to temporally synchronise the plurality of pulses at one or more targets.17215838.MDE.MDE26. The method of claim 25, further comprising directing each of the sequentially-timed seed pulses along spatially different paths to the gain medium.

27. The method of claim 26, wherein the spatially different paths are angularly separated and arranged such that the sequentially-timed seed pulses are incident on substantially the same portion of the gain medium.

28. The method of any of claims 25 to 27, further comprising directing each pulse through the gain medium a plurality of times.

29. Use of the method of any of claims 25 to 33 in an inertial confinement fusion reaction.17215838. MDE.MDE

Images