PLAN DE ACCIÓN

With the technology of the growth of semiconductor structures advancing all the time, the power available from laser diode arrays was also increasing. We obtained two SDL- 3230-T diode lasers from Spectra Diode Labs which were each specified to give 12 mJ output in a 200 |is pulse (60 W peak power). To obtain this power requires a power supply capable of current pulses of 80 Amps. The power supply was the same as the one used to drive the single device of the previous section, that is a SDL 922, where the

Ch. 4 : The Pump Laser

diodes were connected in series and driven by a single power supply. The wavelength

0.4 9 0.3

Î

i

diode laser energy / mJ

fig. 4.10 Q-switched laser energy as a function of laser diode pump energy

specification that comes with the diodes is only ± 2 nm, and it was found that to get a

good spectral match to the NdiYAG absorption the diodes had to be run at close to 40 °C. This has the advantage that there is no chance of condensation on the diode facet which can be a problem if the diodes have to be cooled substantially, but has the major disadvantage that operation at an elevated temperature is likely to reduce the lifetime. Therefore, the diode heatsinks which were similar to as described in the previous section and were designed for cooling of the diodes in fact had to be used to heat the diodes.

The output from the diode lasers was again close coupled into the rod, this time with the two diodes pumping opposite sides of the rod, as shown in fig. 4.11. Three different rods in all were investigated with this configuration, as it was found that the quality of rods could vary and affect the output. These consisted of the 1.3 % doped 1.5 mm ^ 12 mm length rod as used in the previous section, a 1 .1 % doped 2 mm <j> 16 mm length

(both of these plano-plano with AR coatings), and a 1.3 % doped 2 mm <}> 12 mm length Brewster angled rod. Use of Brewster angled surfaces should give low loss and also expand the mode inside the rod to give a better spatial match to the linear gain stripe. With the two diodes opposite each other, the gain profile is again a stripe with peaks at either side where the diode light enters, as shown in fig. 4.12. The reason for the slight asymmetry is that due to their mounts, one diode facet can be brought closer to the rod than the other.

Ch. 4 : The Pump Laser

fig. 4.11 Photo of initial pumping arrangement for double side-pumped laser.

fig. 4.12 Fluorescence for double pumped laser, when diodes are opposite each other horizontally

The Brewster rod was used in the same cavity as previously described, i.e. a im HR and 10 % plane o/c separated by - 11 cm. Due to mode expansion on entering at Brewster's angle, the mode dimensions at the rod were approx. 330 x 580 |im. This larger mode size and the fact that the gain profile doesn't peak in the centre are attributed to the relatively poor long pulse efficiency of 2.5 mJ output, with a slope efficiency of 14 %

and a high threshold of 7.5 mJ. Unlike pumping with a single diode laser where the lasing mode could be aligned to the peak of the gain at one side of the rode, with the double pumped case there is a peak at either side of the rod and the gain at the centre, which is used, is slightly lower. A larger overlap between gain mode and lasing mode can be achieved by expanding the laser mode at the expense of threshold, with the

Ch. 4 : The Pump Laser

restriction that too large a mode will experience clipping losses due to the rod.

When Q-switching this laser, the build up time of around 400 ns meant that the Q- switch was not fully open when the pulse was emitted resulting in poor efficiency. The best output was 2.1 mJ in a 50 ns pulse where the output coupler was replaced with a HR and the loss of the Q-switch was used to polarisation output couple.

As the quality of rods was found to vary, this pump geometry was also used with the 1.5 mm piano rod that was used in the single diode pump experiments. With the same short cavity and 10 % output coupling as used for the free-running work above, 4 mJ was obtained with only the rod in the cavity. Lengthening of the cavity and inclusion of the polariser and Pockels cell reduced this to 2.9 mJ. However, again the Q-switch performance was disappointing with only 1 mJ obtained with a 10 % output coupler. Even though the cancellation of the residual Q-switch retardation as described in ^ section 4.4 was being applied, the build up time and associated loss is attributed to the poor performance. Not only does the loss reduce the efficiency of the output coupling, but it also reduces the number of times threshold to which the laser can be pumped, and so reduces the extraction efficiency.

In order to try and increase the efficiency due to overlap between the pump mode and laser mode, a 1 .1 % 2 mm (j) 16 mm length piano rod was used in a concave-convex

cavity. Chesler and Maydan [41] showed that cavities consisting of one concave and one convex mirror could be used to obtain large mode sizes while still maintaining stable cavities which were relatively insensitive to misalignment. They also found experimentally, for the case of flashlamp pumping, the optimum ratio of lasing mode size to rod radius as 2. Using a cavity consisting of 75 cm concave and 100 cm convex mirrors separated by 20 cm the mode size at the rod was ~ 500 jim. A 1/4 plate was used to provide polarisation output coupling as both these mirrors were HR. With the polariser and Q-switch inserted, this configuration gave 3 mJ long pulse, however it was necessary to lengthen the physical length of the cavity to 30 cm to ensure TEMqo operation. Again, despite the improved long pulse output energy, the Q-switching performance was disappointing with only 0.75 mJ.

To try and understand why the efficiency was low the gain and losses of the system were measured. By inserting a Brewster plate into the cavity and rotating until oscillation is only just obtainable this gives the maximum loss which the system can withstand. This is then equal to the gain that the rod provides, which was = 0.215 with the round trip gain as G = exp(2g/) = 1.54. Using a cw NdiYAG the passive losses of the other cavity components, i.e. X/4 plate, polariser and Q-switch, were measured giving a round trip loss of - 8 %. It must then be that the residual retardation loss of the

Q-switch even with a 600 V dc cancellation bias applied was killing the system. Other work in the department had shown that an end-pumped NdiYLF laser exhibited higher

Ch. 4 : The Pump Laser

gain and Q-switched efficiently [42]. It seemed to be that the low gain available from the double side-pumped system was unable to cope with the residual loss of the Q- switch.

To try and improve the side pumped system by obtaining a gain profile with a central maximum an ’angled pump' scheme was attempted in contrast to the previous linear arrangement. It was hoped to angle the diode lasers at 45 ° to the horizontal and in this way the rod could be mounted in a polished brass mount which would reflect back some of the diode light which was not absorbed (58 % was absorbed in a single pass). Unfortunately, the facet of the diode (particularly for one diode) sits back slightly from its mount and it was not possible to angle the diodes at 45 In fact, as is shown in fig. 4.13, the best that could be obtained was approximately half this. Although a gain profile with a central peak was obtained, as the diodes could not be brought as close to the rod due to mount restrictions, the system showed no improvement over the linear pumping scheme.

fig. 4.13 Angled pump geometry. Inset shows how the mounts restricted the angles that could be employed.

The success of the end-pumped NdiYLF system just mentioned encouraged us to build an end-pumped system with NdiYAG.

4.7 End-pumped Nd lasers