Technical Field

The present invention relates to a compact regenerative optical amplifier with wavelength conversion such as intracavity wavelength conversion. Embodiments provide a regenerative optical amplifier with second harmonic generation and more particularly a source for pumping a titanium :sapphire laser.

Background

Lasers with high output power are required for a number of applications such as materials processing, investigation of material properties, laser induced fusion for energy production, particle acceleration, high resolution radiography and military applications. Lasers for these applications are required to provide high energy, high repetition rate pulses. One of the challenges associated with obtaining stable and reliable high energy pulse generation at high pulse rate is the heating of optical elements, especially within amplifier modules. Often to achieve high energy, large aperture laser beams are amplified up to their high energies by multiple passes through an amplifier or passes through multiple amplifier modules. Compact designs tend to use multiple passes through a single amplifier module, whereas other designs can use multiple amplifier modules cascaded together. However, the use of large aperture beams results in systems that are large and cannot be easily reduced in size, especially if pulse energies are increased.

For high resolution x-ray radiography, and other applications, high energy, short pulse titanium sapphire lasers are commonly used. Such lasers require a pump source which is preferably high repetition rate and pumps in the green region of the visible spectrum between 450 and 550 nm. To achieve this wavelength for pumping, frequency doubling of Nd:YAG (532nm), Nd:YLF (523 or 527nm) or Yb:YAG (515nm) lasers are often used. For other applications, lasers at other visible wavelengths are desirable.

Figure 1 shows a prior art system for generating high energy frequency doubled pulses such as for pumping a titanium :sapphire laser. The system 100 comprises a seed oscillator 110 which is used as the seed source. The pulses from the seed oscillator 110 may be expanded in area and then amplified by pre-amplifier 120 or a series of preamplifiers (not shown). The pulses are then transmitted towards the amplifier module 140 via isolation and focussing optics 130. The isolation and focussing optics comprise an optical isolator that prevents back-reflection into the pre-amplifier 120 and seed oscillator 110. The isolator is accompanied by focussing optics to expand the pulses before they pass through the isolator and then to further expand the pulses before being directed by mirrors to the amplifier module 140.

The pulses are directed to the laser amplifier module by mirror ml for a first pass through amplifier module 140. Amplifier module 140 comprises five slabs 142 of gain material. After passing through amplifier module 140 for a first time, the pulses are incident on a pair of mirrors m2 and m3 which redirect the pulses back towards the amplifier module 140 for a second pass. The path of the pulses for the second pass is slightly offset from the path of the pulses for the first pass. After the second pass the pulses are redirected to the amplifier module by mirrors m4 and m5 for a third pass. After the third pass the pulses are redirected at the amplifier module by mirrors m6 and m7 for a fourth pass through the amplifier module. On exiting the amplifier module for the fourth time the pulses are directed by mirror m8 to a second harmonic generation crystal, SHG, 150 for frequency doubling. Finally, the frequency-doubled pulses are output at 160. SHG crystals have a typical efficiency of 70% conversion, and conventionally the remainder of the input light which is not converted is not used further.

WO 2018/215771 A1 , by the applicant of the current application, describes a laser amplifier module similar to that 140 of figure 1. Accordingly, WO 2018/215771 A1 is incorporated by reference herein. In WO 2018/215771 A1 the middle slab of the amplifier slabs 142 is replaced by a polarisation rotator. In examples, the group of amplifier slabs on one side of the polarisation rotator has the same number of amplifier slabs as on the other side of the polarisation rotator. Although figure 1 of the current application may be considered to show a polarisation rotator with two amplifier slabs on each side of the polarisation rotator, other numbers are possible, such as two groups of three or four slabs. Although figure 1 shows four passes through the amplifier module, any number of passes is possible. The polarisation rotator may provide 90° rotation of polarisation to compensate for depolarisation effects on the pulses passing through the amplifier, as described in WO 2018/215771 A1.

It is evident from figure 1 that the system 100 is not compact and to increase energies will likely require larger aperture amplifier slabs and larger aperture beams to prevent optical damage. Limits on the size of available amplifier slabs and SHG crystal may limit the energy of the frequency-doubled output.

Summary of the Invention

The present invention is directed to providing a high intensity source for pumping a titanium :sapphire laser. Such pump sources usually emit in the green wavelength part of the spectrum such as at 515nm. Other wavelengths and uses for the pump source are possible.

The present invention is directed to a different approach to energy scaling than using large aperture beams to spread the pulse energies spatially, such as for building up high pulse energies for pumping of titanium:sapphire lasers. Instead the present invention is directed to temporally spreading the pulse energies. The laser induced damage threshold of optical components scales as the square root of pulse duration. Hence, increasing the pulse duration by a factor of four can double the amount of energy that can safely be included in a pulse. Hence, by temporally spreading the energy of a pulse smaller beams or higher energies can be used without increasing the risk of damage to optical components.

The present invention uses the temporal spreading of pulses in combination with the requirement to convert infra-red pump laser light to visible light.

The present invention provides a laser amplifier formed of a regenerative cavity which includes a wavelength converting element in the cavity (intracavity). The regenerative cavity may be a ring cavity or a linear cavity. Preferably, the wavelength conversion is frequency doubling by second harmonic generation, SHG. By including the SHG in the cavity light that is not wavelength converted can be recycled around the cavity. In prior art arrangements with the SHG outside of the cavity the portion of the pulse that is not converted is wasted or not further used. Hence, by having the SHG (or other wavelength converter) in the cavity energy of the unconverted portion of the pulse is not lost but recycled, thereby improving overall efficiency of SHG pulse generation. The recycled portion of the pulse travels around the cavity and is amplified and wavelength converted to produce a further pulse. This process may repeat until the gain medium starts to become depleted of energy. The output consists of a burst of pulses closely spaced in time.

The invention is directed to an amplifier that can amplify a seed pulse to high energy levels. The cavity of the amplifier does not generate the seed pulse or starting energy itself, as would be the case for a laser, but receives it from a source external to the cavity.

The present invention aims to provide a capability to pump a high-repetition rate, high intensity titanium: sapphire laser. The capability preferably increases the infra-red to green-conversion efficiency and reduces the risk of optical damage by using lower fluence.

The output wavelength converted pulses such as converted by second harmonic generation may be pulses in the green part of the electromagnetic spectrum for pumping a titanium:sapphire laser. For example, the green may be in the range 450nm to 550nm or more preferably 505nm to 525nm, such as at 515nm. This wavelength may have been converted down from the infrared, for example between 1010nm and 1050nm, such as from 1030nm. The energy of each of the pulses in the burst of pulses may be greater than 0.1 J per pulse. The energy of the pulses in the burst may vary, for example, the first pulse may be lower energy. For the highest energy pulse in the burst of pulses the energy of the pulse may be greater than 1 J or 10J, or in the range of 10s of Joules. Preferably, the majority of the pulses in the burst of pulses will each have an energy greater than 1 J, 10J, or more preferably have an energy of 10s of Joules. In such a case the burst of pulses together and preferably the majority of the pulses in the burst may be described as multiJoule. The bursts of pulses may number between 10 and 100 and the time between pulses is in the order of tens of nanoseconds and at least greater than 5ns such as 50ns. The repetition rate for bursts of pulses may be less than 100Hz such as 10Hz. The burst repetition rate is determined by how quickly the gain medium can be recharged with energy by pumping which will be limited by thermal characteristics and available pump technology.

High energy regenerative optical amplifiers of this kind are not hand held devices but tend to have beam sizes of the order of centimetres and may for example have beam sizes that are 10cm square across. Beam size is a somewhat relative term. For systems of the type described herein the beams may have a flat-top intensity or amplitude profile, or a high order super-Gaussian intensity or amplitude profile. In such cases the beam size may be taken to be full width half maximum. Use of such beam profiles is to make use of the full area of the gain slabs to maximise extraction efficiency and provide uniform intensity to avoid damage to optical components. The regenerative optical amplifiers may be at least a metre in length. Furthermore, to avoid the gain slabs cracking due to heating the gain slabs are mounted in a fluid stream and cooled for example, by a high velocity stream of helium gas at cryogenic temperatures.

The present invention provides a regenerative optical amplifier with intracavity wavelength conversion, the regenerative optical amplifier for generating a burst of pulses, and comprising: pulse guiding optics arranged to form a cavity and guide a received pulse along an optical path within the cavity; an optical gain module disposed in the optical path in the cavity and configured to amplify the pulse; a wavelength converter disposed in the optical path in the cavity and configured to convert at least a portion of the amplified pulse from a first wavelength to a second wavelength; and a splitter disposed in the optical path and configured to separate the portion of the amplified pulse having the second wavelength from the portion of the amplified pulse having the first wavelength and to output the portion of the pulse having the second wavelength from the cavity. By the term cavity we mean a closed path or loop in which a pulse or portion thereof can be cycled around or can traverse round-trips. The cavity may be linear or form a ring or loop. The portion of the pulse having the first wavelength is preferably recycled around the cavity to generate a burst of pulses. In this way at least a portion of the pulse is confined within the cavity for a plurality of passes around the cavity. By the term intracavity wavelength conversion we mean that the wavelength conversion takes place in the optical path in the cavity. The splitter may be a wavelength selective splitter of filter, i.e. dual-band dielectric coated or a prism, or a polarisation splitter because the converted and unconverted portions of the pulse are likely to have orthogonal polarisations. One or more gain modules may be disposed in the optical path in the cavity and be configured for amplifying the pulse.

The pulse guiding optics may be arranged to guide or cycle the portion of the pulse having the first wavelength, after having being left unconverted by the wavelength converter, around the cavity a plurality of cycles, for example, if it is a ring cavity, or back and forth, to and fro, if the cavity is linear. This is to generate the burst of pulses at the second wavelength.

The splitter may be arranged to output from the cavity the burst of pulses at the second wavelength. The burst of pulses may form an output pulse train.

The pulse guiding optics may be arranged to direct the amplified pulse to the wavelength converter to convert at least a portion of the amplified pulse to a second wavelength. The pulse guiding optics may be further arranged to direct the remaining portion of the amplified pulse having the first wavelength to the optical gain medium for further amplification and subsequent wavelength conversion.

The number of round trips of the pulse around the cavity is at least the number of pulses in the burst of pulses. In one embodiment the number of round trips is the same as the number of pulses. In another embodiment the number of round trips is one more than the number of pulses. The time-period between pulses in the burst of pulses may be the round trip time for a pulse around the cavity.

The second wavelength may be less than the first wavelength. The wavelength converter may preferably be a second harmonic generation medium, but may instead be a third harmonic generation medium or a parametric amplification medium. The conversion efficiency of the second harmonic generation crystal may be controlled by rotating the crystal axis relative to the polarisation of the pulse input to the crystal. A controller may be provided to receive a measure of the energy in a burst of pulses or across pulses in the burst, and the controller may rotate the wavelength conversion medium to adjust the conversion efficiency (that is the amount of the input pulse at a first wavelength that is converted to a second wavelength). This may even be done actively to tailor the distribution of energy in the pulses across the burst or train.

The second wavelength may be half the wavelength of the first wavelength and the wavelength converter may be a second harmonic generation medium.

The splitter may be a wavelength selective mirror, such as a dichroic. The wavelength selective mirror may be arranged to reflect the portion of the amplified pulse having a second wavelength and transmit the portion of the amplified pulse having a first wavelength. Alternatively, the wavelength selective mirror may be arranged to transmit the portion of the amplified pulse having a second wavelength and reflect the portion of the amplified pulse having a first wavelength.

The pulse guiding optics may comprise: pulse size reduction optics to spatially reduce the extent of the pulse for input to the wavelength converter; and pulse expansion optics to spatially expand the extent of the pulse for input to the optical gain module. One or each of the pulse size reduction optics and the pulse expansion optics may be formed of a telescope arrangement.

The regenerative optical amplifier may further comprise a polarisation rotator arranged to rotate the plane of polarisation on a pulse passing through the polarisation rotator, such as by 90°.

The optical gain module may comprise a plurality of slabs of optical gain medium and the polarisation rotator may be disposed between a first group of one or more slabs of the plurality of slabs of optical gain medium and a second group of one or more slabs of the plurality of slabs of optical gain medium. Alternatively, the polarisation rotator may be arranged outside of the optical gain medium, elsewhere in the optical path in the cavity. In a further alternative, rods of gain material may be used instead of gain slabs. In other alternatives, thin disk gain material or active mirror geometries may be used. Gain materials may be used in transmission or reflection geometries.

The regenerative optical amplifier may comprise a plurality of optical gain modules. In such a case, the polarisation rotator may be provided in the beam path between the gain modules, such as between a first gain module and a second gain module. Preferably, equal numbers of similar gain modules will be provided in the beam path before and after the polarisation rotator. Alternatively, the gain and depolarisation profiles of the gain media before and after the polarisation rotator will be substantially similar. The regenerative optical amplifier may further comprise a polarisation controller which may be a switchable polarisation rotator switchable between two polarisation rotation modes, the polarisation rotator arranged in the cavity and configured to rotate the polarisation of a pulse passing through it by 90° in one mode compared to the polarisation state in the other mode.

The net polarisation rotation provided to a pulse in a complete round trip around the cavity may be zero or 180° between pulses in a burst of pulses.

The switchable polarisation rotator may be configured to change the polarisation state through the rotator by 90° after a first pulse has passed through the switchable polarisation rotator. This change may remove or add polarisation rotation to the optical path around the ring. The switchable nature of the polarisation rotator may be used to introduce the seed pulse to the ring and keep the seed pulse in the ring. The switchable polarisation rotator may be a Pockels cell. The switchable polarisation rotator may be switched a further time to end the cycling of the pulse or pulses around the cavity, and therefore control the number of pulses in the burst.

The regenerative optical amplifier may further comprise a polariser arranged to receive a seed pulse. The seed pulse, which may be the afore-mentioned received pulse, may be received at a first polarisation and transmitted or reflected by the polariser into the cavity. The polariser may be further arranged to receive a portion of the amplified pulse having the first wavelength that has passed through the wavelength converter and maintain said portion around the cavity while the polarisation state of the pulse is maintained.

The regenerative optical amplifier may further comprise a second polariser, wherein: the second polariser has a crossed polarising direction to the first polariser, or the first and second polarisers have the same polarising direction and the polarisation rotator is disposed in the optical path between the first and second polarisers.

The regenerative optical amplifier may be formed as a linear cavity, and the pulse guiding optics may comprise a pair of mirrors defining the cavity there between such that pulses guided between the mirrors travel a linear path. The optical gain module, the wavelength converter, and the filter may be disposed in the cavity.

The regenerative optical amplifier may further comprise a controller for monitoring energies of pulses and adjusting the rotational position of the wavelength converter medium relative to the pulse polarisation to change the conversion efficiency of the wavelength conversion medium.

The spatial cross-section of pulses passing through the gain medium may be square and may be at least 0.5cm by 0.5cm, and is preferably at least 1.0cm by 1.0cm such as around 3cm by 3cm or 10cm by 10cm. The repetition rate between bursts of pulses may be less than 1000Hz, such as around 100Hz.

The finesse of the cavity may be less than around 15.

The present invention further provides a burst pulse laser system, comprising a seed oscillator and any of the regenerative optical amplifiers described herein. The seed oscillator may be arranged to provide a seed pulse to the regenerative optical amplifier. The seed oscillator may comprise one or more preamplifier(s).

The present invention further provides a titanium :sapphire laser comprising, as a pump source, any of the regenerative optical amplifiers described herein or the burst pulse laser system described herein.

The present invention further provides a method of regenerative optical amplification including wavelength conversion, the method comprising: receiving, into an optical cavity having an optical path, a seed pulse of a first wavelength; amplifying, in the optical cavity, the seed pulse using optical gain medium; converting, in the optical cavity, a first portion of the amplified pulse to a second wavelength; separating the first portion at the second wavelength from the remaining portion of the pulse at the first wavelength and outputting the first portion from the optical cavity, and directing the remaining portion of the pulse at the first wavelength around the cavity a plurality of cycles so as to generate a burst of pulses at the second wavelength.

The method may further comprise directing the remaining portion of the pulse at the first wavelength to the optical gain medium for further amplification and subsequently for wavelength conversion to the second wavelength, and separating the first and second wavelengths.

The portion of the pulse at the second wavelength may be used to pump a titanium :sapphire laser or amplifier. The wavelength conversion may be second harmonic generation. The first wavelength may be in the range 1010 to 1050nm, such as 1030 nm, and the second wavelength may be in the range 505 to 525nm, such as at 515nm. The optical cavity is linear or a ring cavity.

Brief Description of the Drawings

Embodiments of the present invention, along with aspects of the prior art, will now be described with reference to the accompanying drawings, of which: figure 1 is a schematic diagram of a system for generating high energy frequency doubled pulses according to the prior art; figure 2 is a schematic diagram of an amplifier module similar to that of figure 1 but having eight passes of the pulses through an amplifier module; figure 3a is a schematic conceptual diagram of a regenerative amplifier with intracavity second harmonic generation; figure 3b is a graph showing example experimental results for a burst of pulses output from a regenerative amplifier of figure 3a; figure 4 is a graph comparing the energy of single pulse amplified by a prior art amplifier with the energies of a burst of pulses amplified by a regenerative amplifier such as that of figure 3a; figure 5 is flow-chart of a method of operating a regenerative optical amplifier with intracavity wavelength conversion; figure 6a is a schematic diagram of a ring regenerative amplifier according to a first arrangement, whereas figures 6b to 6f are variations to the first arrangement; and figure 7 is a schematic diagram of a linear regenerative amplifier according to a second embodiment.

Detailed Description

Figure 1 shows schematically the arrangement of an amplifier system 100 as described in WO 2018/215771 A1 . The arrangement of figure 1 provides four passes of pulses through an amplifier module 140. Figure 2 shows schematically an arrangement with eight passes of the pulses through an amplifier module 240. A seed pulse 220, which may be infra-red, is directed to the amplifier module 240 and optical arrangements 230, such as the mirrors of figure 1 , redirect the pulse back to the amplifier module 240 for subsequent passes. An optical arrangement is required for guiding the pulses back to the amplifier for each pass. Hence, for eight passes there are needed seven optical arrangements. On each pass the energy in the pulse grows and finally leaves the amplifier module as a single, high intensity pulse 248. The spatial cross-sectional size of the pulse is chosen so that the output fluence, that is the output energy per unit area, is close to the saturation fluence of the laser gain material and below the laser damage threshold of the optical elements, such as the gain material, mirrors, lenses, windows etc., that make up the system. The pulses 248 are high energy, high repetition rate (nanosecond duration) in the infra-red. The SHG material 250 converts the infra-red to pulses 252 in the visible spectrum which may then be output 260 and used, for example, to pump a titanium :sapphire amplifier. The resulting output from the titanium :sapphire amplifier is high brightness, sub- picosecond pulses suitable for the production of x-ray and gamma ray pulses at high repetition rate, such as 10Hz to 100Hz, for pulsed radiography.

Figure 3a is a schematic diagram of a regenerative ring cavity according to an embodiment of the present invention. In figure 3a the gain medium such as an amplifier module 340 along with a wavelength converter 350 such as an SHG medium are disposed in a ring cavity. Similarly to figure 2a, seed pulse 320 is directed to laser gain medium of the amplifier module 340. However, in figure 3a the seed pulse is introduced into the ring cavity by a polarisation selective optical switch 315. The amplifier module 340 is disposed in the pulse path after the switch. The wavelength converter or SHG 350 is arranged in the pulse path after the amplifier module. A splitter 360 may also be included. The splitter 360 is arranged to separate, or filter, frequency doubled light in the pulse from non-frequency doubled light. For the path taken by the non-frequency doubled light, in the pulse path after the splitter is an optical arrangement 330. The optical arrangement 330 is configured to return the pulse to pass through the gain medium and SHG again, and may also comprise optical components for spatially adjusting the pulse shape. For example, the optical components may include pin holes, apertures, focusing optics such as telescope arrangements, and mirrors for guiding the path of the pulse.

We will now describe the operation of the regenerative ring cavity of figure 3a. The polarisation-selective optical switch 315 receives the seed pulse 320 and the seed pulse is passed through the switch 315 towards the gain medium of the amplifier module 340. The amplifier module 340 increases the optical intensity of the pulse. The wavelength converter or SHG 350 converts a portion of the pulse received from the amplifier module 340 to a different wavelength. If the wavelength converter is an SHG medium the wavelength will be halved (the frequency will be doubled). For a near infra-red input pulse (that is, the lower end of the infra-red spectrum) this means the converted portion 364 of the pulse output from the SHG will be in the visible spectrum such as green. The splitter 360 separates the converted portion 364 of the pulse from the unconverted portion 362 of the pulse. The splitter 360 may be a wavelength-selective mirror such as a dichroic. The remaining unconverted light 362 is recycled back to the input of the amplifier module 340 via the optical arrangement 330 and polarisation-selective optical switch 315. The amplifier module 340 may include a polarisation rotator such as a 90° polarisation rotator similarly to the amplifier module described above in relation to figure 1. The polarisation rotator may be disposed between two groups of equal numbers of gain slabs so as to compensate for depolarisation caused by the gain slabs. The polarisation rotator may be a quartz plate. The operation of the polarisation-sensitive optical switch 315 may be as follows. On receipt of the input pulse 320 the switch transmits the polarisation of the input pulse. After receipt of the input pulse 320 and before the unconverted portion 362 of the pulse arrives back at the switch the polarisation-sensitive optical switch 315 is switched to change its polarisation status. After switching the polarisation-sensitive optical switch does not transmit but reflects the polarisation of the recycled pulse so as to direct the unconverted portion of the pulse back to the amplifier module 340.

The polarisation-sensitive optical switch may be a Pockels cell in combination with a polariser such as a thin film polariser. The Pockels cell may arranged to provide a change of 90° of polarisation rotation between the switched and unswitched states. The 90° change is provided between the initial input pulse being transmitted by the switch and the recycled pulse being reflected by the switch. Other alternatives may be used instead of a Pockels cell. The polarisation-sensitive optical switch 315 is used to inject the initial seed pulse 320 in to the regenerative amplifier. When the switch is returned to its initial state the recycled pulse will be transmitted through the switch and hence will not be further recycled around the regenerative amplifier.

For each pass through the amplifier module 340 another wavelength converted pulse will be output from the splitter 360. The length of time the polarisation sensitive optical switch is in the switched state will determine the length of time the regeneration process continues and hence, the number of pulses that are output. The output pulses will be closely spaced in time and may be considered to be a burst of pulses.

By using a ring amplifier that recycles the pulses, the amplifier can be more compact than the prior art approaches by using fewer optical elements reducing optical losses and reducing cost. Furthermore, by spreading the energy over time by using a burst of pulses, the risk of optical damage is reduced. This also allows smaller diameter beams to be used which may allow standard optical components to be used thereby saving further costs. In an example, if a burst of ten pulses is to be provided such as through ten round trips around the ring or cavity then in comparison to a single pulse, each of the ten pulses may be 1/10 ^th the energy of the single pulse.

In an embodiment in which the wavelength converter converts infra-red to green a burst of green pulses is obtained. Example experimental results are shown in figure 3b. The time between pulses is determined by the time it takes for a pulse to complete one circuit of the regeneration ring cavity. In the example results of figure 3b, the ring design had a 50 ns time to traverse the ring. The number of pulses is determined by the time delay between injecting the input pulse and re-opening the polarisation-sensitive optical switch, which in figure 3b can be seen to be long enough for five pulses.

Figure 4 provides a schematic graphical comparison of a single pulse that has passed through a prior art amplifier module such as the arrangement of figure 1 with that of a burst of pulses such as has been generated by the regenerative amplifier of figure 3a. The graph shows fluence on the ordinate and time on the abscissa. Fluence is a measure of the energy per unit area in the pulse(s). In the single pulse of the prior art amplifier the fluence is approaching the damage threshold for components in the amplifier, whereas by temporally spreading the energy into a burst of pulses the fluence of each of the burst of pulses is well below the damage threshold. In figures 3b and 4 the power in the pulses of the burst grows initially to a peak and then reduces. The increase is because the power in the pulses is continuing to grow initially as they travel around the ring passing through the amplifier, and then later the energy stored in the gain slabs starts to be depleted so the pulse energy reduces.

Figure 5 is a flow-chart setting out steps of a method similar to the method of operation described in relation to figure 3a. The flow chart begins at step 1010 by receiving a seed pulse, such as pulse 320, at a first wavelength. The pulse is directed to pass through an amplifier at step 1020, such as an amplifier 340, to amplify the pulse. At step 1030 the amplified pulse passes through a wavelength converter such as an SHG medium 350, for example for second harmonic generation. At step 1040 the wavelength converted portion of the pulse, which may be in the green part of the spectrum, is separated from the unconverted portion, which may be in the infrared. At step 1050 the unconverted portion of the pulse is directed around the cavity such as by optical arrangement 330 to direct the pulse back to the amplifier. The portion of the pulse that has been wavelength converted is at a second wavelength and is output at step 1060. The step of separating and outputting may be performed by a splitter or filter. Optionally, the output pulse may be directed to a titaniurmsapphire laser or amplifier for pumping 1070 of the laser or amplifier.

Figures 6a-6f and 7 show more detailed example configurations of regenerative amplifiers with intracavity wavelength conversion.

A first configuration example 600, which is shown in figure 6a, is a ring regenerative amplifier with intracavity frequency conversion. The amplification is provided by an amplifier module 640 and the wavelength conversion is provided by wavelength conversion medium 650. Other components are included to control the flow of pulses around the ring. The amplifier module 640 includes a number of slabs of gain material. In a particular embodiment, the gain material is cryogenically cooled by gas flowing between the slabs. The slabs of gain material may be Yb:YAG. The slabs are preferably arranged normal to a beam or pulse path direction. The gain material may be pumped from both sides simultaneously by beams or pulses. The pump beams or pulses are directed to the gain slabs by wavelength selective mirrors 638 and 642. The pump beams or pulses are of a different wavelength to that of pulses circulating around the ring amplifier. The wavelength selective mirrors 638 and 642 are selected to transmit the wavelength of the circulating pulses and reflect the pump pulse wavelength. The wavelength selective mirrors 638, 642, may be dichroic such as dichroic mirrors. In a preferred embodiment the pump pulses are provided at a wavelength of around 940nm and the amplifier operating wavelength (that is, the wavelength of the circulating pulses) is around 1030nm. In an alternative arrangement, the wavelength selective mirrors 638, 642, may transmit the pump pulse wavelength and reflect the circulating pulse wavelength. The power handling capability of the wavelength selective mirrors in transmission and reflection at the relevant wavelength should be considered so as to avoid damage. Although we have described the amplifier module 640 as having slabs of gain material, the amplifier module may instead be provided with a rod or rods of gain material operating in transmission or a thin disc or discs of gain media working in reflection.

The configuration further comprises an input polariser 681 for receiving a seed pulse 620 to initiate circulation of pulses around the ring. The input polariser 681 is configured to reflect polarisation corresponding to the polarisation of the input pulse 620 and transmit the polarisation orthogonal to that. Following the input polariser in the ring is provided a polarisation rotator 682, which may for example be a quartz rotator. The polarisation rotator 682 rotates the plane of polarisation of pulses passing through it by, for example, 90°. Hence, an input pulse input to the ring at polariser 681 and having vertical polarisation, V, will be rotated by the polarisation rotator 682 to horizontal polarisation, H.

In the pulse path after the polarisation rotator 682 is provided a polarisation controller 665 such as a Pockels cell. Such a device comprises a birefringent material whose birefringence can be adjusted based on a voltage applied. The polarisation controller 665 is preferably configured to be switchable to add in 90° polarisation rotation to the optical path around the ring. In the arrangement of figure 6a the polarisation controller in combination with a second polariser 683 is provided to change the polarisation of the seed pulse and by doing so maintain the pulse cycling around the ring. When the power from the amplifier is nearing exhaustion, or for other reasons, the polarisation controller 665 may be switched to switch the pulses out of the ring.

The second polariser 683 is located in the ring after the output from the polarisation controller 665. The second polariser 683 reflects vertically polarised light and transmits horizontally polarised light. Following the second polariser is a telescope arrangement 631 which is an image-relaying telescope that converts the beam or pulse size back to the input beam size. The telescope 631 does not add magnification to the path but instead compensates for any beam size reduction or increase, or defocussing effects. The telescope 631 may provide focussing (indicated by the double-headed arrow) to compensate for thermal lensing in the polarisation controller. This may be especially true if the polarisation controller is a Pockels cell because the Pockels cell may absorb a small proportion of the incident pulses. The amount of absorption depends on the design of the PC and choice of the active material. For example, DKDP has higher absorption but is available in larger apertures than BBO, which has much lower absorption but is limited in aperture. The smaller beams made possible by the burst pulses technique allows BBO to be used.

Wavelength conversion medium is provided at 650. The wavelength conversion medium 650 may be a second harmonic generation crystal. This may be, for example, LBO or DKDP (lithium triborate or deuterated potassium dihydrogen phosphate) although other materials may be used. In our example of the seed pulse being around 1030 nm, these two example materials provide second harmonic generation lowering the wavelength to around 515 nm.

In the pulse or beam path after the wavelength conversion medium is provided a wavelength selective splitter or filter 660, which may be a wavelength selective mirror such as a dichroic. The wavelength selective splitter 660 is configured to separate the portion of the pulse that has been wavelength converted 664 from that which has not been converted 662. In our example of a converted pulse at 515nm wavelength and a 1030nm unconverted pulse, one of the converted and unconverted pulse wavelengths is reflected and the other is transmitted. Although shown in figure 6a with the converted portion of the pulse reflected and the unconverted pulse transmitted, it is possible instead to use a dichroic with the opposite configuration such that the unconverted portion of the pulse is reflected and the converted portion is transmitted. In general, optical components damage more easily at shorter wavelengths, for example in the green as compared to the infrared, because the photon energy is higher. For green/infrared separation, dichroic splitters or filters that reflect in the green and transmit in the infrared are generally used so that the green beam does not pick up, or picks up reduced, aberrations as a result of passing through the substrate at an angle (dichroic substrates are often wedged). However, overall there is little difference between the two alternatives for transmission and reflection by the dichroic and actual choice may depend on the specific optical arrangement used.

Although the path of the pulses in the regenerative ring amplifier is shown as a rectangle in figure 6a other path shapes are possible. Mirrors (not shown in figure 6a) will be required to direct the pulses around the ring, such as at the corners of the rectangle.

We now describe the polarisation manipulation and pulse handling of the example of figure 6a. A seed pulse 620 is directed at the polariser 681 for input into the ring. The seed pulse is shown as having vertical polarisation V (direction of polarisation is perpendicular to the page). This is shown by the large dot on the line representing the input 620 to first polariser (P1) 681 . Since the polariser 681 is arranged to reflect vertically polarised light, the seed pulse 620 is reflected at the polariser into the optical path of the ring. Polarisation rotator 682 rotates the polarisation of the input pulse by, for example, 90° such that it is horizontally polarised. On this first pass around the ring the polarisation controller 665 does not change the polarisation of the pulse and subsequently the horizontally polarised pulse is transmitted by second polariser 683. The pulse is reimaged by telescope 631 and then arrives at wavelength converter medium 650. Here a proportion of the pulse is wavelength converted. However, because the pulse arriving at the wavelength converter on this first pass has not passed through the amplifier, the intensity of the pulse is low and the amount of converted pulse is very low.

The wavelength conversion medium may be a Type I second harmonic generation medium such as LBO or DKDP. In such a case the polarisation of the wavelength converted pulse is output at 90° with respect to the polarisation of the pulse input to the SHG medium. The polarisation of the unconverted portion of the pulse at the output of the wavelength conversion medium is unchanged from the polarisation at the input to the SHG medium. In other words, the Type I SHG medium does not change the polarisation of the infra-red pulse that passes through unconverted but the polarisation of the converted green output is orthogonal to them. Alternatively, Type 0 or Type II SHG media may be used.

Wavelength selective splitter 660 reflects the converted pulse 664 such as the green pulse and transmits the unconverted pulse 662. The unconverted pulse continues around the loop to amplifier module 640. The wavelength selective mirrors 638 and 642 transmit at the wavelength of the seed pulse and so the seed pulse passes through the first mirror 638 to the amplifier module 640. The amplifier module 640 amplifies the pulse. To provide sufficient and uniform gain may require the gain medium to be pumped from both sides. This is indicated in the figure by the two arrows labelled “Pump”.

The amplified pulse is transmitted by first polariser 681 and then on passing through the polarisation rotator 682 its polarisation is rotated to vertical polarisation. To maintain the pulse in the loop the polarisation controller now rotates the pulse polarisation by 90 degrees such that it is horizontally polarised. The pulse again passes through the telescope 631 and is wavelength converted by medium 650. The converted pulse is again output at splitter 660. The unconverted part of the pulse continues around the loop and on each pass a wavelength converted pulse is output thereby resulting in a burst of output pulses. To stop producing output pulses the polarisation controller (Pockels cell) is switched to remove 90° degrees of polarisation rotation from the ring. The result is that the pulse from the polarisation rotator stays vertically polarised, V, and is rejected from the loop by polariser 683.

A wavelength converted pulse will be output for (almost) every pass around the ring. However, the conversion efficiency of the wavelength converter may increase in proportion to the fluence squared. Hence, for the embodiment of figure 6a the converted pulse on the first pass around the ring will be very low energy because the seed pulse power has not yet passed through the amplifier to increase the pulse power. The time from one output pulse to the next will be the time for the pulse to travel around the ring. If the output is being used such as for a pump for a titanium :sapphire laser it is likely to be preferable if the time between pulses is as short as possible, or tuned over a time period to provide effective pumping while reducing the risk of damage to optical components. In one example, the round trip time and hence the time from one output pulse to the next may be around 25 ns. If the time between pulses of the burst is too long, and by implication the total time of the green pulse train is long, the energy stored in the titanium :sapphire may start to decay. Conversely, if a single pulse is provided the instantaneous gain may be sufficiently high that amplified spontaneous emission may start transverse lasing in the titanium: sapphire amplifier. Hence, a balance needs to be reached for the timing between pulses in the burst and the round-trip time of the pulse being amplified in the titanium :sapphire amplifier to provide optimum pumping of the titanium :sapphire.

As discussed in relation to figures 3b and 4, for the embodiment of figure 6a the energy of pulses in the burst increases as the seed pulse is amplified more on each pass and then the energy in the pulses decreases as the available power in the gain medium is depleted. The cavity of figure 6a (as well as others described below) outputs a significant portion of the pulse energy for each pass in its wavelength converted pulse. This means the finesse of the cavity is low such as less than around 10-15. For example, if 70% of the energy of each pulse is output the finesse will be around 10. This is different to conventional regenerative amplifiers, which have high finesse such as much greater than 1000.

The use of two polarisers 681 , 683 and a polarisation rotator 682 such as a quartz rotator (QR) between them creates a crossed-polarisers configuration. Alternatively, actual crossed polarisers could be used but the addition of the quartz rotator helps with pulse introduction and polarisation control. The crossed polariser configuration is used to prevent (longitudinal) self-lasing around the ring such as is caused when any two optical components are aligned causing a laser-like cavity. The crossed polariser configuration provides significant loss to prevent these unwanted effects from building up. In more detail, the amplifier gain material, such as Yb:YAG is pumped to build up energy to provide gain and will emit at a range of wavelengths and polarisations in all directions, especially before the seed pulse is injected into the system. A small proportion will be emitted in the correct direction to propagate around the ring. Before the seed pulse is injected into the ring (and up until its first pass around the ring) the polarisers 681 , 683, will be effectively crossed such that any polarisation passing around the ring will be reduced in power significantly. The presence of the polarisation rotator will mean that the polarisers may actually be oriented for transmitting the same polarisation of light but in combination with the polarisation rotator the net effect is to substantially extinguish light passing through them both. As a result, light emitted by the gain material that is coupled around the ring will be extinguished so that self-lasing of the gain material around the ring is not supported. When a seed pulse is injected in to the ring the pulse will be amplified by the gain material thereby taking energy from the gain material reducing the emission that can cause selflasing. Hence, on injecting the seed pulse, the possibility of self-lasing is reduced. Furthermore, the pumps to the gain material may be operating on millisecond timescales, whereas the seed pulse will be incident on the gain material of the order of 10s of nanoseconds from injection and repeatedly over these timescales until switched out of the ring. With the seed pulse propagating around the ring, the gain material will be rapidly reducing in stored energy and therefore will be much less likely to be emitting in a way that causes self-lasing of the cavity.

The discussion in the preceding paragraph relates to self-lasing around the ring. Another unwanted effect in the amplifier is amplified spontaneous emission (ASE). The gain slabs will emit in the transverse direction, that is, along the slab in a direction transverse to the pulse direction. The ASE may reflect from the edges of the slab or from the vane plate or mount in which the slab is held resulting in feedback within the slab. This effect is managed by adding an absorbing material to the edge of the slab which has a similar refractive index to the gain material. This prevents reflection at the gain medium-cladding interface and attenuates the unwanted transmitted signal, thus preventing lasing in the transverse direction.

The polarisation controller 665 is shown and described as a transmissive Pockels cell. However, reflective Pockels cells are also available. In such a case the polarisation of the reflected beam may be controlled by application of a voltage to the cell in the same ways as for the transmissive Pockels cell. Accordingly, a transmissive Pockels cell may be used in the arrangement of figure 6a at the corner of the ring close to the illustrated position of the Pockels cell and may replace a mirror (not shown) at that corner.

The presence of polarisation rotator 682 (QR) allows the second polariser 683 to be orientated the same way as the first. This keeps beams rejected from the ring travelling in the same plane as each other, such as horizontally, or to keep them on an optical table rather being directed vertically up in the air. This means that they are easier to manage safely.

We have discussed how the regenerative amplifiers discussed herein may be particularly suited to pumping a titanium :sapphire laser. The use of burst pulses, for pumping, as provided by the arrangements described herein will result in the gain of the titanium :sapphire laser being spread out in time compared to a conventional single large pump pulse. Hence, the instantaneous gain is reduced thereby lowering the risk of ASE or self-lasing occurring in the titamium:sapphire laser.

We now provide a simplified numerical example of the power levels that could be seen around the ring regenerative amplifier of figure 6a. In this example we will consider a seed pulse of spatial dimensions 10mm x 10mm forming a pulse with a substantially square cross-section. Consider the seed pulse having energy of 0.33J injected into the ring at 620. The seed pulse will travel almost a full pass around the ring before reaching the gain module 640. Passing through the various components on the way may reduce the seed pulse power slightly. The gain medium 640 may be double pumped with two pump lasers delivering 25J pulses. This may result in the gain module providing a small signal gain of 3x which amplifies the seed pulse to around 1 J. The gain medium may be slabs of Yb:YAG mounted in an amplifier module having cryogenic gas cooling of the slabs. As described above, on the first pass around the ring there is almost no conversion of the pulse by the wavelength conversion medium because of its low intensity. The pulse travels further around the ring and passes through the gain module for the second time. Loss around the ring may reduce the pulse power to just less than 1 J, for example. On passing through the gain module for the second time the gain module increases the power in the pulse to 3J. The wavelength conversion medium may be arranged to convert 60-70% of the received pulse, such as 66%. If the seed pulse is near infra-red at, for example, 1030 nm, the converted pulse would be in the green at 515nm. With the conversion efficiency mentioned, the green pulse would be 2J and 1 J of infra-red would be recycled around the ring. By recycling 1 J the same amount of power as on the second pass would be recycled, theoretically providing a burst of pulses of equal energy. However, changes in gain of the amplifier will cause the actual output powers to vary. The polarisation controller will receive the pulses after amplification and so will have to handle the highest powers of pulses around the ring. For example, it should be able to receive a burst of 10 pulses, and handle a power of 300W when operated at a 10Hz repetition rate. The wavelength conversion medium should similarly be able to handle 300W. In a real system, the energy per pulse in the burst will vary and therefore the actual power-handling requirement may be lower.

Figures 6b-6f provide alternative arrangements of components for the regenerative ring amplifier. Preferably, the arrangements balance the requirement to prevent, or minimise, self-lasing as caused by emission from pumping the gain slabs. This is achieved by having crossed polarisers or polarisers with a polarisation rotator between them. In the latter case, this may be polarisers arranged with the polarisation direction parallel with a polarisation rotator arranged to provide around 90° polarisation rotation between the polarisers. Another requirement is to be able to inject a seed pulse and controllably maintain it in the ring for as long as desired. This may be achieved using a switchable element such as a polarisation rotator, for example, a Pockels cell. Furthermore, there are many possibilities for the position around the ring at which the seed pulse may be injected. The injection point may be at the position of one of the polarisers. In considering this, to prevent the self-lasing discussed above, it is preferable if the polarisers are located such that stimulated emission from the gain medium does not get reflected or otherwise returned to the gain medium for further amplification. Accordingly, the following figures provide examples of different locations for the components such as locating the polarisation rotator in the amplifier module. Figure 6b is an alternative arrangement of the ring regenerative amplifier of figure 6a. Features identified by the same reference numbers represent the same features as in figure 6a.

The input to the ring regenerative amplifier of figure 6b is polariser 681 which is arranged to transmit horizontally polarised light and reflect vertically polarised light, as for polariser 681 of figure 6a. Following polariser 681 ' is amplifier module 640’. The use of reference number 640' indicates that amplifier module is modified in comparison to amplifier module 640 of figure 6a. Amplifier module 640’ includes a polarisation rotator 682 between the gain slabs. The polarisation rotator may be a quartz rotator and may preferably provide 90° of polarisation rotation. Similarly to amplifier module 640 of figure 6a, amplifier module 640’ of figure 6b is pumped from both sides and wavelength selective mirrors 638, 642, direct the pump beams or pulses to the amplifier module 640’.

In figure 6b the polarisation rotator 682 is indicated as the central slab in the amplifier module. The gain slabs are preferably symmetrically arranged either side of the polarisation rotator. For example, equal numbers of gain slabs may be arranged either side of the polarisation rotator. As described herein, the polarisation rotator is included to prevent lasing resulting from spontaneous emission caused by pumping of the amplifier module 640’ and also as part of the pulse control methods provided by the polarisation controller. However, by including the polarisation rotator in the amplifier an additional advantage is provided. This arrangement may provide compensation against depolarisation effects in the gain slabs, as described in WO 2018/215771 A1 . The polarisation rotator may be particularly effective in compensating against depolarisation effects because the pulses will pass through the amplifier module and polarisation rotator on axis. In the prior approach of figure 1 the pulses are necessarily slightly off-axis to allow the multiple passes through the amplifier module. The present example of having the pulses on-axis means that the depolarisation effect from a pass through a first group of gains slabs will be very effectively compensated for by a pass through a second group of slabs. This is because the amount of depolarisation can be dependent on the lateral position across the pulse or gain slab, and this will be largely the same with the pulse on-axis between a pass through a first group of slabs and a second group of slabs in the amplifier module.

In comparison to figure 6a the main change is that the polarisation rotator such as quartz rotator 682 is included in the amplifier module 640. The crossed polariser effect for stopping or reducing self-lasing is maintained. However, to keep the polarisation rotator between the two polarisers the first polariser (now numbered 681 ’) is moved in the ring path to before the amplifier module 640’. Hence, the input is introduced before the amplifier module 64O’.This also has the advantage that even on the first pass around the ring the pulse is amplified, which is not the case for figure 6a. Hence, in figure 6b a higher energy pulse can be built up before the wavelength conversion commences. Much of the rest of the arrangement of figure 6b is the same as figure 6a. Accordingly, it can be seen that as long as the polarisation rotator (quartz rotator) is provided between the two polarisers, which will always be the case for a ring, the polarisation rotator can be placed in, or outside of, the laser amplifier module.

Operation of figure 6b is otherwise similar to figure 6a.

Figure 6c is a further modified version of the arrangements of figures 6a-6b. In figure 6c the location in the ring of the wavelength converter medium 650 and wavelength selective splitter 660 has been changed in comparison to figures 6a and 6b. This allows pulse size reduction optics 632 and pulse size expansion optics 631 ” to be included. The optics 631 ” and 632 may each take the form of a telescope arrangement. The pulse size expansion optics 631 ” may take the form of a magnifying telescope, Te, and is arranged to increase the spatial size or extent of the pulses passing through. The pulse size reduction optics 632 may take the form of a demagnifying or beam-reducing telescope, Tr, and is arranged to decrease the spatial size or extent of the pulses passing through. The pulse size reduction optics 632 may follow the amplifier module 640’. The telescope arrangement of figure 6a provides refocussing but does not provide magnification or demagnification. The pulse size reduction optics 632 and pulse size expansion optics 631 ” may also provide refocussing.

The wavelength conversion medium 650 follows the pulse size reduction optics 632. The wavelength conversion medium 650 may have a smaller aperture than the amplifier gain slabs. For example, smaller aperture wavelength conversion materials are lower cost. Furthermore, the conversion efficiency may increase with greater input fluence. Hence, the pulse size reduction optics are provided to reduce the size of the pulse and increase the intensity of the pulse from the amplifier and passing to the wavelength conversion medium 650. Correspondingly, after the wavelength conversion medium 650’, later in the path of the ring, pulse expansion optics 631 ’ may be provided to increase the spatial extent of the pulse for input to the gain module 640'. By expanding the size of the pulses the energy in the pulses is more spread out as they reach the gain slabs of the amplifier module. The power density in the pulses as they are amplified through the gain slabs is kept below the damage threshold. As mentioned, following amplification and before reaching the wavelength conversion medium 650 the spatial size of the pulses is reduced. Increased energy density in the pulses as they reach the wavelength conversion medium provides greater wavelength conversion efficiency. Accordingly, the arrangement of figure 6c has advantages in gain and conversion efficiency compared to the arrangements of figures 6a and 6b.

In figure 6c the wavelength conversion medium 650 is located in the ring before the Pockels cell. This is different to figures 6a and 6b where the Pockels cell is located before the wavelength conversion medium. The result is that the orientation of the wavelength conversion medium is different in figure 6c such that the wavelength conversion occurs for the vertical polarisation.

Figure 6c operates by receiving a vertically polarised pulse which is amplified and polarisation rotated by gain module 640’. On the first pass through wavelength conversion medium the pulse is in the wrong polarisation for wavelength conversion and passes straight through without conversion. The pulse continues in horizontal polarisation through the polarisation controller and both polarisers to return to gain module. Here it is amplified further and rotated to vertical polarisation. Part of the pulse is converted by wavelength conversion medium 650’ and output in the horizontal polarisation. The polarisation controller is switched to add 90° of polarisation rotation to maintain the pulse in the ring. The pulse continues to cycle around the ring until the polarisation controller is switched so as not to add the 90° of polarisation rotation. In figure 6c the Pockels cell is located in the ring after the wavelength splitter 660’. This means that the Pockels cell can be a lower power handling unit than in figures 6a and 6b because some of the power has been output at the wavelength splitter 660.

Figure 6d is an alternative arrangement of the ring regenerative amplifier of figure 6a-6c. Features identified by the same reference numbers represent the same features as in figures 6a-6c.

The input to the ring regenerative amplifier of figure 6d is polariser 615, which is arranged to transmit horizontally polarised light and reflect vertically polarised light similarly to the polarisers 681 and 681 ’ of figures 6a-6c. In figure 6d, following polariser 615 is amplifier module 640’. The use of reference number 640’ indicates that amplifier module is largely the same as the amplifier module of figures 6b and 6c, in that it includes a polarisation rotator 682 between the gain slabs.

Following amplifier module 640’ are pulse size reduction optics 632, which may take the form of a telescope arrangement and may be similar to the pulse size reduction optics of figure 6c. After the wavelength conversion medium 650, later in the path of the ring, pulse expansion optics 631 ’ may be provided to increase the spatial extent of the pulse for input to the gain module 640’. The pulse expansion optics 631 may also be of a telescope configuration. Hence, similarly to figure 6c the pulse expansion optics may increase the spatial extent of the pulses so as to improve gain and keep the power in the gain slabs below the damage threshold, and the pulse size reduction optics may increase the fluence in the wavelength converter to increase the conversion efficiency.

In the ring after the pulse size reduction optics 632 is wavelength conversion medium 650 which, as for figures 6a-6c, converts a portion of the seed pulse to a different wavelength such as by SHG. Wavelength selective splitter 660 is provided after wavelength conversion medium 650. Polarisation controller 665 which may be a Pockels cell is provided next in the ring. Pulses in the ring are then directed from Pockels cell to pulse size expansion optics and back to input polariser 615 for further passes around the ring.

Since the arrangement of figure 6d is different to that of figures 6a-6c, by including only one polariser, we now describe the process of ring regenerative amplification with wavelength conversion for the arrangement of figure 6d.

A seed pulse 620” is received in horizontal polarisation at polariser 615. The polariser is arranged to transmit horizontally polarised light such that the seed pulse passes through the polariser. The seed pulse next reaches amplifier module 640’ which includes polarisation rotator 682. The amplifier amplifies the seed pulse and rotates the polarisation to vertical. The pulse size is reduced at optics 632, by a demagnification factor M, before reaching wavelength conversion medium 650. As previously discussed, wavelength conversion medium 650, such as Type I SHG, may be polarisation dependent such that wavelength conversion only occurs for a particular linear polarisation of pulses or beams. In the arrangement of figure 6d the wavelength converter 650 or SHG is oriented such that in this first pass around the ring, in which the seed pulse is incident on the wavelength conversion medium 650 vertically polarised the pulse is not converted. Accordingly, the pulse simply passes through wavelength selective splitter 660 to the polarisation controller 665. On the first pass through the polarisation controller no polarisation rotation is imparted on the pulse and it remains in the vertical polarisation. In a particular arrangement, the polarisation controller may be configured such that no voltage is required to be applied to the controller to provide no rotation. The pulse is then directed to the pulse expansion optics 631” to spatially expand the pulse, such as by a magnification factor M, and is further directed back to polariser 615. The vertical polarisation of the pulse is shown as V at the input of the polariser 615. The pulse is now orthogonal to the polarisation that on the first pass was transmitted by the polariser 615. Hence, the pulse, on this second pass, is reflected by the polariser 615 and directed to the amplifier module 640’ for a second time. The pulse is amplified again and this time around the polarisation of the pulse is rotated to horizontal. The pulse then passes to pulse expansion optics and is incident on the wavelength conversion medium 650. This time around, the pulse polarisation is correctly oriented for wavelength conversion. The converted portion of the pulse is polarised vertically, whereas the unconverted portion of the pulse remains horizontally polarised.

The converted and unconverted portions of the pulse exit the wavelength conversion medium 650 and are incident on the wavelength selective splitter 660 which separates the two portions. The converted portion may be output at 664. To keep the unconverted portion of the pulse cycling around the ring, the polarisation rotation provided by the polarisation rotator 682 needs to be compensated for. That is, the polarisation of the unconverted pulse needs to be rotated back to vertical, V, polarisation to avoid transmission at polariser 615 and rejection from the ring. Hence, the polarisation controller 665 is activated to provide 90° of rotation to the polarisation. As described above, the polarisation controller may be thought of as having a first setting which is used when the seed pulse has just been injected, and a second setting which is used to act on the polarisation of the pulse to cause the pulse to cycle around the ring.

After passing the wavelength conversion medium and the polarisation controller the pulse is directed around the ring for a further pass. The pulse will continue to go around the ring generating a wavelength converted pulse for each pass around the ring. The number of wavelength converted pulses that are output will be almost the same as the number of passes of the pulse around the ring, except that the first pass through the wavelength conversion medium does not produce an output wavelength converted pulse.

After a number of pulses it may be desirable for the pulse to be stopped from circulating around the ring. For example, after a number of passes through the amplifier module the gain slabs may start to become depleted of energy and not provide the desired amount of gain. To stop the pulse from circulating around the ring the polarisation controller 665 is again switched to provide no rotation of the polarisation of the unconverted pulse. After doing so upon circulation around the ring back to the polariser 615 the pulse is in a polarisation state, H, that is transmitted instead of being reflected by the polariser 615 such that the pulse leaves the ring. For example, in the arrangement of figure 6d the pulse will pass straight through the polariser 615 (in a direction down the page).

In the arrangement of figure 6d, similarly for figure 6c, by placing the polarisation controller in the ring after the wavelength selective splitter or filter the optical power- handling requirement of the polarisation controller is reduced in comparison to if it is placed before the wavelength converter, such as in figures 6a and 6b.

The pulse expansion optics 631 ” which expand the spatial extent of the pulse may also include focussing (indicated by the double-headed arrow) to compensate for thermal lensing in the polarisation controller. The focussing is similar to that described for figure 6a.

The polarisation rotator 682 and single polariser 615 prevent the gain medium from self-lasing in a similar manner as described for figure 6a-6c. When the gain medium is pumped but no seed pulse is injected, after a complete pass around the ring the polarisation rotator rotates spontaneous emission light from the gain slabs by 90°. Hence, vertically polarised light passing around the ring will be converted to horizontally polarised and then ejected by transmission through the polariser 615. Any horizontally polarised light will be ejected straight through polariser. The difference between figure 6d and figures 6a- 6c is that in one direction horizontally polarised emission from the gain slabs will pass nearly two cycles around the ring before being ejected. In figures 6a-6c having two polarisers separated around the ring means that ejection will be after a little more than one cycle. That said, for all embodiments emission will pass around the ring for no more than one additional pass through the gain medium. As result, in all embodiments self-lasing will be avoided.

The use of a single polariser instead of two means that the polariser is being asked to perform two functions, namely reject the unwanted polarisation and act as a mirror for the cavity for the correct polarisation. This means it is not possible, or is at least more difficult, to adjust the cross polarisers angular position to maximise pulse rejection because the adjustment may cause beam deviation and so not provide the best alignment. Hence, this may provide that two polarisers may be preferable.

Earlier we provided a simplified numerical example of the power levels that could be seen around the ring regenerative amplifier of figure 6a. We now consider the case for figure 6d. We will consider a seed pulse of spatial dimensions 30mm x 30mm forming a pulse with a substantially square cross-section and having energy of 0.33J injected into the ring at 620”. The gain medium 640 may again be double pumped with two pump lasers delivering 25J pulses. This may result in the gain module providing a small signal gain of 3x which amplifies the pulse to 1 J. After amplification the spatial size of the pulse may be reduced by a factor, such as 3x, to bring the pulse size to around 10mm x 10mm to fit the wavelength conversion material, which may be Type I SHG formed of LBO. As described above, on the first pass around the ring there is no conversion of the pulse by the wavelength conversion medium because it is in the wrong polarisation. Hence, the pulse travels around the ring a cycle to pass through the gain module for the second time. Loss around the ring may reduce the pulse power to less than 1 J, for example. On passing through the gain module for the second time the gain module increases the power in the pulse to 3J. The wavelength conversion medium may be arranged to convert 66% of the received pulse. If the seed pulse is near infra-red at, for example, 1030 nm, the converted pulse would be in the green at 515nm. With the conversion efficiency mentioned the green pulse would be 2J and 1 J of infra-red would be recycled around the ring. By recycling 1 J the same amount of power as on the second pass would be recycled, theoretically providing a burst of pulses of equal energy. However, changes in gain of the amplifier will cause the actual output powers to vary. The polarisation controller will receive the pulses after wavelength conversion and separation so will not have to handle the highest powers of pulses around the ring. However, it should be able to receive the 10mm x10mm pulse and handle a power of 100W for a burst of 10 pulses at a repetition rate of 10Hz. The wavelength conversion medium, which will receive the full power of the converted and unconverted portions, should be able to handle 300W. In a real system, the energy per pulse in the burst will vary and therefore the actual power-handling requirement may be lower. The beam expansion optics 631’ convert the beam size back to 30mm x 30mm for passing through the gain module a further time.

Figure 6e is an alternative arrangement of the ring regenerative amplifier of figure 6b. Features identified by the same reference numbers represent the same features as in figures 6a-6d. The operation of the amplifier of figure 6e is substantially the same as that of figure 6b. The embodiment of figure 6e differs from that of figure 6b in that the gain module 640’ has been replaced by two gain modules 640a and 640b. The polarisation rotator 682 is located between the two gain modules 640a, 640b. Embodiments may include more than two gain modules. Furthermore, the other embodiments described herein may include two or more gain modules. Figure 6e also shows gain modules in which the gain medium of each module comprises a rod of gain medium instead of the plurality of slabs shown in figure 6b and other figures. The use of rods of gain media is also applicable to the other embodiments described herein. The rods of gain media are pumped from the side as shown in figure 6e. In this arrangement, the pump coupling optics, such as the wavelength selective mirrors 638, 642, of figures 6a-6d are not required.

As mentioned, although figure 6e shows two gain modules, more than two gain modules may be used. The use of two or more gain modules may allow higher pulse energies to be achieved. ln figure 6e the polarisation rotator 682 is located between the two gain modules 640a, 640b. If the gain media of the two modules are similar then the combination of the polarization rotator 682 and similar gain media will provide depolarization compensation. That is, depolarisation effects produced by gain slabs in one gain module will be substantially reversed in the other gain module due to the transverse polarisations with which the pulses pass through each module. Depolarisation compensation is best achieved if the gain slabs before the polarisation rotator 682 are substantially similar to those after the polarisation rotator. This can be achieved by having equal numbers of gain modules either side of the polarisation rotator. Alternatively, this can be achieved irrespective of the numbers of gain modules either side of the polarisation rotator as long as the gain material at the first side of the polarisation rotator is substantially similar to that at the second side of the polarisation rotator. In any of the embodiments described herein a gain module having a polarisation rotator arranged between first and second groups of gain slabs may be replaced with multiple gain modules with a polarisation rotator there between.

Figure 6f is an alternative arrangement of the ring regenerative amplifiers of figures 6b and 6e. Features identified by the same reference numbers represent the same features as in figures 6a-6e. The operation of the amplifier of figure 6f is substantially the same as that of figures 6b and 6e. The embodiment of figure 6f differs from that of figure 6e in that the gain modules 640a and 640b have been replaced by four gain modules 640n, 640n+1 , 640n+2 and 640n+3. The gain modules 640n to 640n+3 are illustrated as gain modules operating in reflection and may, for example, be thin disk or active mirror geometry amplifier modules. Similar to figure 6e, the gain modules may be pumped from the side, i.e. transversely. Although many of the embodiments described herein show transmission geometries for amplifier modules, reflection geometries for amplifier modules may alternatively be used. The use of four amplifier modules or other numbers of amplifier modules in a chain may allow higher pulse energies to be achieved. In figure 6f the polarisation rotator 682 is located between the second and third amplifier modules 640n+1 and 640 n+2 such that it is half way along the gain path for maximising depolarisation compensation.

As will be apparent from the arrangements of figures 6a to 6f there are a number of possible configurations for how the components are arranged in the ring. The arrangements discussed in these figures are not considered exhaustive and other arrangements of components in the ring are possible. A further example configuration 900 is shown in figure 7. This configuration is a linear regenerative amplifier with intracavity frequency conversion instead of a ring regenerative amplifier as shown in figures 3a and 6a-6f. The linear regenerative amplifier of figure 7 is similar to the ring regenerative amplifiers of figures 6a-f but the arrangement of components is necessarily different to produce the linear arrangement. Additionally, to create a linear cavity mirrors 980 and 990 are provided at the ends of the cavity. Components that correspond to features in figures 6a-6f are indicated with like reference numbers.

As mentioned, the cavity of the linear regenerative amplifier 900 includes mirrors 980 and 990 at the ends of the cavity to confine the pulses. Within the cavity there is provided a pair of polarisers 681 and 683. Similarly to the arrangements of figures 6a-6f there is a polarisation rotator also included in the cavity. The gain module 940 is similar to that of figures 6a-6d. The gain module 940 differs from that of figures 6b-6d because it does not include the polarisation rotator therein. Instead, similarly to figure 6a, the polarisation rotator 682 is provided outside of the gain module elsewhere in the cavity. The gain module is pumped by pump lasers 638, 642. As for figures 6a-6d the pump pulses are directed to the slabs of the gain module by wavelength selective mirrors 638 and 642. Between the first polariser 681 and the gain module 940 is provided a polarisation controller 665 which may be a Pockels cell. Between the gain module 940 and second polariser 683 is provided a polarisation rotator 682, which may be a quartz rotator. After second polariser is provided wavelength conversion medium 960 and splitter 660. In the arrangement of figures 7 the splitter 660 is provided in the path before the wavelength conversion medium 950. Splitter 660 may be a wavelength selective splitter or filter such as dichroic mirror, or it may be a polarisation-based splitter.

The requirements on mirror 980 and mirror 990 are different. Mirror 980 will receive the seed pulse 920 which will preferably be in the infra-red. Hence, mirror 980 is required to have high reflection at the desired infra-red wavelengths. The mirror 990 at the other end of the cavity will also receive the infra-red of the seed pulse, but will also receive the wavelength converted pulses which may for example be green. The mirror 990 will therefore be required to have high reflectivity in the infra-red and in the green, for example at 1030nm and at 515nm.

We now describe operation of the linear regenerative amplifier with intracavity wavelength conversion shown in figure 7. A seed pulse 920 is injected into the cavity at polariser 681. The seed pulse may be vertically polarised as indicated by the dot and the V in the figure. The polariser 681 is set to reflect vertically polarised light and transmit horizontally polarised light. Hence, the vertically polarised light is reflected by the polariser. In the arrangement of figure 7 the polariser is arranged to reflect the vertically polarised light along the path towards the gain module 940. The pulse having being reflected towards the gain module by the polariser 681 is directed at polarisation controller 665. On this first pass through the polarisation controller, the polarisation controller is set not to provide polarisation rotation. The seed pulse is next incident on the gain slabs in gain module 940 which amplify the seed pulse. Following amplification, the quartz rotator 682 rotates the pulse to horizontally polarised. The pulse reaches polariser 683 which is arranged to transmit the horizontally polarised pulse. The pulse also passes through splitter 660 and is incident on wavelength conversion medium 950. The wavelength conversion medium converts a portion of the pulse to a second wavelength such as by SHG to green. As described in relation to previous embodiments, the wavelength conversion medium may be Type I SHG. In such a case the converted portion of the pulse has orthogonal polarisation to the seed pulse, and the remaining portion of the seed pulse maintains its polarisation. In the present example, the wavelength converted portion of the pulse has a vertical polarisation and the unconverted portion has horizontal polarisation. The two portions are reflected by mirror 990 at the end of the cavity. On travelling back along the cavity the two portions are again incident at the wavelength conversion medium. This will have no effect on the already converted portion of the pulse, but further conversion of the unconverted portion will occur. The double pass through the wavelength conversion medium will increase the amount of the converted portion such that it is possible to think of the double pass as being equivalent to using double the length of wavelength conversion material. Alternatively, the length of wavelength conversion material used may be halved. The portions of the pulse are incident on the splitter 660 and the wavelength converted portion is reflected from the linear cavity and output at 964. The unconverted portion is transmitted by the splitter 660 and continues back through polariser 683, polarisation rotator 682, through gain module 940 to polarisation controller 665. On arriving at polarisation controller 665 the polarisation rotator has been switched to add 90° of polarisation rotation to the cavity. The pulse is rotated from vertical to horizontal polarisation by polarisation controller 665. The polariser 681 is arranged to transmit the horizontal pulse and so the pulse travels on to the mirror 980. The polarisation controller is used to maintain the seed pulse in the linear cavity. The amplifier module amplifies the pulse twice in every return trip of the pulse around the cavity. Similarly to the earlier arrangements, the example of figure 7 uses two polarisers and a polarisation rotator to form crossed polarisers. This arrangement is used to prevent or reduced self-lasing of the gain medium, in the same way as described for preceding arrangements described herein.

When it is desirable to stop the pulses cycling back and forth in the linear cavity, the polarisation controller is switched to remove the 90° of polarisation added. The pulse will be output from the cavity at one of the polarisers. The actual polariser will depend on the timing of switching the polarisation controller. If the polarisation controller is switched as the pulse travels from polarisation controller towards mirror 990 (left to right in the figure) or as it travels towards mirror 980 (right to left in the figure) but before it reaches the polarisation controller the pulse will be output by polariser 681 when the pulse reaches it. On the other hand, if the polarisation controller is switched when the pulse is between the polarisation controller and mirror 980 it will be output from polariser 683. Given that it takes 10s of nanoseconds for the pulse to travel around the cavity, careful control of the timing of switching of the polarisation controller will be needed for outputting of the pulse at the desired polariser. If this cannot be achieved, output from either polariser is possible and safety precaution should be taken.

As discussed in relation to figures 6a- 6f, it is also the case for figure 7 that the order of the various components in the linear cavity can be changed. For example, in figure 7 the polarisation rotator is provided between the gain module 940 and the polariser 683, but as for the arrangements of figures 6b-6d, the polarisation rotator may be included in the gain module 940 between two groups of gain slabs. Other arrangements are also possible such as changing the positions of the polarisers, polarisation controller and/or wavelength conversion medium, but the arrangement of these components as shown in figure 7 may be preferred for the reasons set out in the preceding paragraphs. In a further modification to the arrangement of figure 7 a single polariser could be used as for figure 6d.

Although not shown in figure 7 the linear cavity may be provided with a telescope to provide focussing and/or beam size adjustment. This may be placed in the beam path at a point between the gain module 940 and the wavelength conversion medium 950. For example, the telescope may be arranged such that the spatial size of the pulse may be increased as it travels towards the gain medium and decreased as it travels towards the wavelength conversion medium. The telescope may also include focussing to compensate for defocussing in the polarisation controller.

In this arrangement compared to the ring arrangements of the preceding figures, the pulses pass through the gain medium twice per round trip compared to once per round trip. Hence, for a desired timing between pulses the gain slabs of the arrangement of figure 7 may operate at higher temperature than those of figures 6a to 6f. It was discussed above in relation to figure 6a that a reflective Pockels cell may be used instead of the illustrated transmissive Pockels cell. A reflective Pockels cell may be also used in figure 7 to replace the transmissive Pockels cell 665.

In a modification to the arrangement of figure 7, at both sides of the wavelength conversion medium there may be arranged wavelength selective splitters for separating the wavelength converted portion of the pulse from the unconverted portion. One wavelength selective splitter, arranged towards the mirror 990, may separate the converted portion of the pulse on the first pass through the wavelength conversion medium. The second wavelength selective splitter may be arranged where the splitter 660 is shown in figure 7 and may split out the converted portion of the pulse after a second pass through the wavelength conversion medium. The wavelength selective splitters may be mirrors such as dichroics. The two wavelength selective splitters may direct the converted portions of the pulse in opposite directions. For example, one may direct the converted portion in a downwards direction such as shown by splitter 660 in figure 7, whereas the other wavelength selective splitter may direct the unconverted portion in an upwards direction. However, the two splitters may alternatively be arranged such that the portions are directed in largely the same direction such as up or down and may be combined into the same optical path. When opposite directions are used this may be because one of the portions is going to be used to pump a separate titanium-sapphire amplifier.

The amount of wavelength conversion by the wavelength conversion medium 950, such as an SHG, may be controlled by the setting the angular orientation of the wavelength converter medium relative to the polarisation of the approaching pulse. In theory if the two are aligned perfectly and the power input to the medium is high enough then greater than 90% conversion efficiency could be achieved. The orientation of the wavelength conversion medium could be adjusted to reduce the conversion efficiency leaving more of the pulse for conversion on the return pass through the medium. However, it would be difficult to arrive at two sets of converted pulses of equal powers. For example, if on the first pass 50% of the pulse is converted, then on a second pass only 50% of the remainder will be converted, that is 25%. The power levels in the two pulses could be made more even by the wavelength selective splitter not being 100% efficient, such that a portion of the converted pulse is not output by the first splitter and is returned to the wavelength conversion medium. Here it may be combined with the portion of the pulses converted on the return pass through the wavelength conversion medium. The provision of two pulse bursts of approximately equal power would allow double pumping of a titanium sapphire laser with bursts of pulses. Furthermore, high-energy titanium :sapphire amplifiers are typically pumped by two green beams from either side to ensure uniform heat load and gain. So this arrangement of producing two green bursts of pulses is readily applicable thereto. The second pulse train may be weaker than the first but equal pumping is not always required. Alternatively, the arrangement of figure 7 may be used as shown and the output pulses split by a power divider such as a mirror to produce two sets of output pulses for double pumping.

In embodiments the wavelength conversion may be third harmonic generation, THG, instead of second harmonic generation. To achieve THG would require both an SHG medium and a THG medium. As described above, the output of the SHG medium would be a combination of both the frequency doubled portion of the pulse and the unconverted portion of the pulse. In the examples described above the outputs would be at 515 nm and 1030 nm. Both these outputs are input to the THG medium to provide a third harmonic output at around 343 nm. From the THG medium would also be some of the original pulse and the frequency doubled pulse. The portion that is the original pulse can be recycled around the ring.

Another embodiment may couple two SHGs together in series to provide a four-fold reduction in wavelength, for example for 1030 nm pulses the output would be around 250nm such as at 257 nm.

A further embodiment could use parametric amplification on the input pulses to produce output pulses of different wavelengths. An example would be to take the already frequency doubled 515nm pulse and convert it to two pulses. The two pulses would have frequencies f1 and f2 which would sum to be equal to the frequency of the 515nm pulse. Alternatively, the input pulses could be 1030nm pulses and using parametric amplification these could be converted to two frequencies f1 and f2, signal and idler, in the near infra-red having a degeneracy of around 2pm, for example. The actual frequencies f1 and f2 depend on the non-linear property of the crystal.

In the arrangements described the conversion efficiency of the second harmonic generation crystal may be adjusted by rotating the crystal axis relative to the polarisation of the pulse input to the crystal. We have discussed this in relation to the arrangement of figure 7 but it may be applied to any of the arrangements. If the crystal is mounted for rotation the conversion efficiency of the wavelength conversion process may be controlled by adjusting the rotation angle of the crystal. A controller may be provided to receive a measure of the energy in a burst of pulse or across pulses in the burst, and the controller may rotate the wavelength conversion medium to adjust the conversion efficiency (that is the amount of the input pulse at a first wavelength that is converted to a second wavelength). This may even be done actively to adjust the rise and fall in energy of pulses across the burst or train.

The person skilled in the art will readily appreciate that various modifications and alterations may be made to the above described regenerative amplifiers with intracavity wavelength conversion without departing from the scope of the appended claims. For example, different materials, powers, polarisations, pulse paths, numbers of gain modules, transmission or reflection gain modules and different optical configurations may be used. In one arrangement anamorphic prism pairs may be used to reduce the spatial extent of the pulse such as to increase intensity and efficiency of wavelength conversion. These prisms pairs could be used instead of the above described telescope arrangements. To maintain a pulse as having a square area two anamorphic prism pairs may be required. In arrangements a transmissive Pockels cell may be used instead of a reflective Pockels cell.