Nd:YAG laser side pumped by a single diode laser array

At the start of this work, the most powerful diode laser array that was available commercially was the SDL-3220-J, from Spectra Diode Labs, which was capable of a peak power of 25W in a 200 |is pulse producing 5 mJ. These 'quasi-cw' devices could be operated up to a maximum rate of 100 Hz due to heating considerations. This laser was used to side-pump a 1.3 % (atomic) NdiYAG rod of 1.5 mm diameter and 12 mm

Ch. 4 : The Pump Laser

length. Nd:YAG was chosen as it was an established material which was readily available and would allow a good match between the emission wavelength of the diode and the absorption peak at ~ 809 nm. The work that is described in this subsection was performed in collaboration with Callum Norrie, and can also be found in the final section of his thesis [37].

The laser diode array had to be characterised so that its use as a pump laser could be optimised. As mentioned above, the diode could be operated with up to a 200 |is pump pulse, which allows good energy storage due to the 230 jis upper state lifetime. The limit on this pump pulse length was the ability to remove heat from the device. If the temperature of the diode changes, then so does its wavelength, which must be matched to the 1 nm absorption bandwidth of the NdiYAG. The emission wavelength of the diode array was found to vary both across the array length and also with time due to the dynamic heating of the array by the driving current pulse [38]. This gave a time averaged linewidth from the array of approx. 5 nm. To achieve the best possible match between the emission and absorption requires control of the temperature of the diode array, or at least of its heatsink. Temperature control was obtained by inserting the laser diode inside an insulated copper block, and removing heat using two peltier coolers. The current to the peltier coolers was controlled with a Photon Control 290 Peltier Driver, with temperature sensing using a thermistor. The heat was then removed from the hot sides of the peltier coolers by finned heat sinks and passive air cooling.

The emitting dimension perpendicular to the plane of the laser array is only 1 jxm which results in a large divergence angle of the emitted light in this dimension, -40° (FWHM), while in plane of the array the divergence is - 10°. The close coupled scheme adopted here allowed the use of the rod surface as a lensing element to control the divergence in the vertical plane. The diode facet can be aligned to be parallel to a reference axis which will contain the laser rod, by observing the diffraction pattern produced when a He-Ne is directed at the diode facet. The gain region which is produced by this arrangement was observed by detecting the 1 jim fluorescence that is

produced with a Si CCD camera (EEV photon). Use of a small aperture at the camera lens allows the whole length of the rod to be imaged. The fluorescence profile was sampled with an in-house video frame grabber controlled by a BBC Archimedes micro­ computer, and is shown in fig.4.8. The fluorescence, and hence gain, profile consisted of a stripe opposite the diode with the peak close to the diode side of the rod where the pump intensity is highest. The rod was held in a polished brass holder which would reflect unabsorbed pump light back into the rod.

Using a short cavity, - 12 cm, with a Im highly reflecting (HR) mirror and a 10 % piano output coupler (o/c) and with the rod roughly centred, a best free running energy of 0.9 ml in a single transverse mode was obtained for 5 mJ pump. Energy

Ch. 4 : The Pump Laser

fig. 4.8 Fluorescence profile viewed through one end of the rod

measurements were made with a Molectron J50 pyroelectric detector and JIOOO energy meter. Normally the laser mode size is kept as small as possible to obtain a low threshold provided that the pump mode can be confined to be within this size. However, when the pump mode is spatially extended, as is the case here, the mode size is a trade off between low threshold, good overlap between pump and lasing mode, and the requirement that only the TEMqo mode is allowed to oscillate. Single transverse mode

is obtained here by aligning the cavity such that the TEMqo mode uses the peak gain

close to one side of the rod, and the rod itself acts as an aperture to prevent higher order modes. The gaussian output profile and measured spot sizes agreed with those expected from cavity mode calculations and was therefore taken to be diffraction limited, although no direct measurement of the closeness to diffraction limited (i.e. number of times diffraction limit or the 'M^' parameter) was made with this laser. The temporal properties were measured with a BPX65 Si photodiode and seen to give the expected shape of relaxation oscillations tending to a steady state output.

For non-linear optical applications a high peak power is required and so the next step was to Q-switch the laser. This was achieved with the set-up shown in fig. 4.3. Firstly the laser cavity had to be extended in length to allow insertion of the polariser and electro-optic Q-switch. The change in mode size due to extension of the cavity (from 310 |im to 380 |im at the rod) reduced the free-running energy to -0.7 mJ. With the polariser and Q-switch in the cavity the energy was reduced further to -0.5 mJ. Using the Findlay-Clay technique [39] (which looks at the way threshold varies with output coupling) the intracavity losses were estimated to be -9 %. Most of these losses were found to be due to the polariser.

Ch. 4 : The Pump Laser

Q-switching of the laser produced a best output of 0.41 mJ in a 38 ns (FWHM) pulse giving a peak power of over 10 kW, though a more typical output was 0.33 mJ in 40 ns. The drop in output energy compared to free running is attributed to the relatively large losses which reduced the number of times above threshold to which the laser could be pumped which affects the extraction efficiency. The loss due to the elasto-optic effect, as mentioned earlier, was not too serious for this laser as the long pulse build up time meant that the Q-switch was almost fully opened when the pulse had built up. The pulse to pulse stability was good, as can be seen in fig. 4.9 which shows a persistence trace over 100 pulses indicating a stability of approx. ± 5% .

20ms I

fig. 4,9 100 pulse persistence trace of Q-switched pulses

The characterisation of the Q-switched output agrees well with standard Q-switch theory, as found in advanced textbooks, e.g. Siegman [40], giving a slope efficiency of

17 %, as shown in fig. 4.10.

Lack of success at this point with the OPO lead to the requirement of a laser which was capable of producing higher output powers. The logical step to achieve this was to use more diodes for pumping, and also to use the higher powered diode arrays which were available by this time.