Categoría: Planificación

An obvious route to improve the performance of this end-pum ped slab system is to shorten the cavity, in order to optimise the pulse duration. An attem pt along these lines, by reducing the resonator length down to 11cm, did not succeed, as the resulting pulse-build-up times proved to be too fast for the switching times of the acousto-optical Q-switch. The opening time of the switch is given by the travelling time of the acousto-optical wave to leave the mode- waist region. In the cavity configuration used this amounts to approximately 100ns. This is a direct consequence of the Q-switch material used, which has a high dam age threshold but comparatively low ultra-sonic speed of sound (approximately 3km /s opposed to 4km /s (PbM0 4) or 51on/s). If the pulse already builds up during the opening phase of the Q-switch, higher losses and bad mode quality in the form of high order modes are the experimentally observed consequences. This limitation is not too great a disadvantage in the current set-up, as the peak power per pulse is already limited by the damage threshold of the slab coatings.

The previously m entioned safe peak power limit of 40kW per pulse for the slabs can be stretched to -70kW w ithout immediate coating damage. However this grey zone was only an option because one slab had already shown signs of coating damage and total failure of this slab would not have been a big loss. With the present coatings this operational mode is not advisable, especially at higher pum p powers.

By pushing the short cavity system w ith low repetition rates and slightly higher pum p powers (12W) to pulse widths below 10ns, a rising edge in the falling flank of the Q-switched pulse starts to appear. This pulse shape is shown in figure 3.16.

This unexpected behaviour could be an indication of a bottleneck effect, either caused by the lifetime of the lower laser level being comparable to the very short pulse durations of around 10ns [9], or a repopulation of the upper laser level, due to relaxation of population from higher excited states, which were previously up-converted by excited state absorption of laser radiation .

c w and Q-switched performance at 1047nm - Chapter 3

Tekamm iGS/s 9 Acqs

^ « •» ~‘j* j

i C h i i W k ir h

1 9.8ns

I y-..

SOOmV 71)1 / - 7?0mV

Fw. 3.16: An unwanted deviation of the pulse shape appears for very short pulses. The rising edge in the falling flank of the pulse starts to appear for pulse durations around 12 - 15ns (see fig. 3.9b). With shorter pulse durations this after-pulse becomes more pronounced, as the 10ns pulse shape above exposes.

In contrast to this there have been reports of pulse widths approaching 3.5ns at 1047nm in Nd;YLF [10], but w ithout displaying the pulse form. These short pulse results w ere achieved by utilising extrem ely short resonator configurations (2 to 3cm) and low cw pum p powers, of about 500mW. A com parison of such different systems is difficult, as the form (or even appearance) of an after-pulse also depends on the time dependent intra-cavity losses during Q-switching.

However, in the present system the 10ns pulse width region is clearly a lower limit in terms of pulse duration and could not be circumvented by careful alignment of the Q-switch or cavity mirrors.

Higher pum p powers than the 20W so far used lead to a further decrease in mode quality. Pumping at the 30W level represented a system limitation, as the degraded beam quality at this level could not be used to increase the diffraction lim ited green or UV output. For other applications, such as, for example, material processing, the high average output powers in the 8W region are an interesting alternative. The change of the output power with repetition rate for diode pum p powers of 30W is shown in figure 3.17 .

c w and Q-switched performance at 1047nm - Chapter 3

I

15

20

25

Repetition Rate [kHz]

Fie. 3.17: In the case of SOW diode pump power the beam quality decreases to = 1.8 at 8kHz. The average output power and peak power per pulse is displayed as a function of the repetition rate.

Although average powers of up to 8.7W were achieved, the beam quality decreased to an of 1.8 at 8kHz repetition rate, and the efforts needed to control and maintain these moderate mode properties also increased. This high power operation is an interesting alternative if the beam quality of the fundamental or converted radiation is only of secondary interest. For example, the conversion efficiency for doubling does not necessarily drop dramatically for beam param eters approaching = 2, so that the advantage of higher average infrared powers at the 30W pum p level can be transferred into the converted green radiation.

However, the mediocre beam quality does not im prove in the frequency doubled case. This is reflected in the frequency tripling results, w here the mixing process of two beams w ith decreased transversal mode properties notably affects the conversion efficiency into the UV . The details of the single pass doubling and tripling experiments will be discussed in Chapter 6. As the main interest was in near diffraction limited beam qualities, the Q-switched experiments were limited to the 30W diode pum p power level. Higher pum p powers would only have increased the average output power of the system at

c w and Q-switched performance at 1047nm - Chapter 3 the further expense of the beam quality.

So far the linewidth of the laser system has not been described satisfactorily, as the recorded linewidth of 0.2nm in figure 3.3 is just at the resolution limit of the optical spectrum analyser. Even a single frequency Helium-Neon laser with a linew idth better than 0.002nm (IGHz), did not record a much im proved linew idth (~0.15nm), than the definitely multi-longitudinal 1.047pm laser source. To obtain a meaningful value for the 1pm spectrum a Fabry-Perot étalon has been used to analyse the frequency doubled infra-red radiation. Frequency doubling has the advantage of easy control of the Fabry-Perot alignment. The étalon mirrors used have a reflectivity of R=80% in the green, and the separation of the m irrors was 1mm. The short separation was necessary to account for the broad spectrum of the free running laser.

The free spectral range (FSR) of this set-up was 150GHz and, given the finesse of F~14, the resolvable linewidth was 12GHz. Three clusters of longitudinal modes within the free spectral range could be resolved and were imaged with a 100mm lens onto the CCD Camera (figure 3.18).

By switching on a cw background, a generally undesirable mode of operation under normal Q-switching conditions, the fringe pattern could clearly be sharpened. In this self seeded mode of operation, although only improvised, the linewidth could be reduced down to 3-5GHz in the green. The reduction of the hold-off, by detuning the Q-switch away from the Bragg-angle, is a demonstration of the well known line narrowing technique by pre-lasing [11]. No relevant im provement in the spectral brightness could be achieved with this very crude seeding, as the output power drops by at least a factor of two, while the pulsew idth broadens and the pulse to pulse fluctuations strongly increase. However, this improvised line narrowing is of use to characterise and demonstrate the linewidth dependence of OPO's, when pum ped with this Q- switched laser operating in the infra-red (Chapter 7).

The main reason for the m easurem ents was to determ ine the linew idth at 1047nm, Taking into consideration that the fringe pattern was observed in the green, the linewidth of the Q-switched laser is in the region of

~0.1nm (25GHz).

c w and Q-switched performance at 1047nm - Chapter 3 This corresponds to -35 longitudinal modes about threshold. The strength of the clustering, as observed through the Fabry Perot étalon, can be slightly varied, without introducing after-pulses, and depends on the alignment of the Q-switch. The characteristic three clusters can also be observed in the signal output of a NCPM KTP-OPO, as shown in Chapter 7.

a)

b)

Fig, 3.18: Fabry Perot fringes of the frequency doubled, Q-switched output (a)). The separation between fringes within one spectral range is >12GHz. Line narrowing could be achieved by switching on a cw background (b)). The reduced hold-off in this case causes irregular after pulses.

c w and Q-switched performance at 1047nm - Chapter 3