Lasers and Femtosecond Lasers

Transcription

1 1/26/2004 Lasers and Femtosecond Lasers As you learned in the lecture, a Ti:sapphire laser can operate either as a tunable, continuous-wave (CW) laser, or a pulsed, self-modelocked, laser. A slight alignment change of the cavity allows you to switch between these two modes of operation. In this laboratory, you will learn how to characterize the laser in both modes of operation. Pre-Lab Exercises 1. How many longitudinal modes in a typical HeNe laser? A typical Ti:sapphire laser? 2. What is the Brewster angle for ti:sapphire at 532 nm? 800 nm? (The laser mode is along the optical axis.) 3. What is the distance separation between the 532 nm pump and the 800 nm laser mode on the curved mirrors of the laser if the two modes are perfectly mode matched within the crystal? The radius of curvature of the mirrors is 10 cm. 4. Estimate the repetition rate of the laser. I. Align the laser for CW operation: The laser and beam paths are shown in Fig. 1. High reflector Pump Laser Prisms Ti:sapph crystal Focusing Lens Curved mirrors Output coupler 1

2 CM L CM XTAL Fig. 2(a). Close up of rail that holds the two curved mirrors (CM1, CM2), Ti:sapphire crystal (XTAL), and focusing lens (L). Procedure: 1. Dial the first prism so that the fluorescence from the laser misses it and hits the turning mirror. Do not move the mount, just turn the micrometer 2. Move the high reflector from behind the second prism and place it in the postholder shown below (labeled CW ). Adjust it to retroreflect the fluorescence that is bouncing off of the turning mirror (the turning mirror should be set to send the fluorescence to the high reflector, but may need adjustment. You need to get the fluorescence exactly retracing the path. 3. Get it lasing. You should place a card after the output coupler, then search around with the alignment of the high reflector. You will see a flash on the card once it begins lase. If you are far off, it helps to look at the card with an IR viewer. With the IR viewer, you may be able to see the spot from the high reflector on the card. It should lase once you have the high reflector spot aligned with the other spots. 4. Optimize for maximum power, you will need to move the curved mirrors, primarily the one going that reflects toward the high reflector. Note you should make a small adjustment to and then optimize the power with the end mirrors and see if the power went up or down. Fig. 2(b) Turning mirror for CW cavity shown. Put high reflector in post holder. 2

3 Fig. 2(c). Back end of cavity showing high reflector in position for modelocked operation a second prism. Record the following 1. Maximum output power (average) for 5W pump power 2. Threshold pump power (pump power for which lasing just begins) 3. Slope efficiency slope of output power vs. pump power above threshold. 4. Output wavelength. 5. Polarization can you tell if the beam is polarized? Is the polarization in the direction you expect? Why? 6. Divergence angle II. Operate with prisms and modelock. 1. Dial the prism back in, restore the high reflector to behind the second prism and get it lasing again. 2. Optimize power by adjusting power. 3. Get it modelocking. The approximate procedure is as follows, a more detailed description is in the laser manual (in the lab). a. Diagnostics. Let the beam propagate a few meters so you can observe the spatial mode, or alternatively, set up a CCD camera. Observing the spectrum is also helpful, either use the Ando optical spectrum analyzer in the Keck lab (you don t need much power, so efficient coupling into fiber is not needed), or bounce a portion of the beam off of a diffraction grating. Setting up the fast photodiode and connecting it to the fast oscilloscope can also be useful. b. Prisms. Set the prisms so the beam is going with in 1-2 mm of the apex of the prisms. c. Misalignment. Translate the curved mirror that is farthest from the lens inward towards the ti:sapphire crystal until beam looks elongated vertically as shown in Fig. 3. If the center looks hollow, move the lens and crystal to fill it in. d. Search. Search for the point where it starts to modelock. To initiate modelocking, shove the prism table. Quick is good, but not too hard. Watch your diagnostics, particularly the spectral width and for pulses on the fast photodiode. If you are close, you will see the spectrum broaden and a pulse train on the diode that remains after you prism is still, but both will only last for a fraction of a second. Once you find this, make small adjustments to 3

4 Fig. 3. Laser spot for CW operation (left) and ML operation (right) when aligned for modelocking. increase the spectral width and duration of pulses. At this point you can also start adjusting the prism insertion. Record the following: 1. Output power for CW and ML (average and peak powers) 2. Output spectrum (use Ando optical spectrum analyzer) 3. Polarization 4. Divergence angle 5. Pulsewidth (or bandwidth since you can assume that the phase is flat across the spectrum so that a fourier transform of the spectrum will give the laser pulsewidth) 6. Repetition rate 7. Energy per pulse Advice and suggestions: Aligning a laser is an optimization problem in a large dimensional parameter space (every knob on the laser is a dimension). This is generally a hard problem, if you get lost in a ~20 dimensional space, you may never be seen again. Avoid Panic tweaks : It is important to remember that You can t tweak zero. * Thus if the laser stops lasing, don t panic and do something stupid. Stop. Think. Ask yourself, what did I just do? The best strategy is to try to undo it (of course this is hard to do if you smashed an elbow into a mount and spun it around by o, so clumsiness is generally a bad thing). Put another way, if you are turning knob A and the laser quits, don t turn knob B to try to get it lasing again, once you do that you are lost in a two dimensions rather than just one dimension. * Attributed to John L. Hall 4

5 Hill climbing: You will often need to optimize two parameters at the same time. However it can be the case that the two parameters do not represent the path of steepest ascent to optimum (i.e. highest power). Worse yet it can be the case simply optimizing each parameter individually won t get you to the highest value. What you need to do is make a small change to one, which may actually make the power go down, and then optimize the second and see if the new optimum is higher or lower. If it is higher move the first in the same direction and repeat, if it is lower, move the first in the opposite direction and repeat. If neither direction makes it go up, then you re at the maximum. To visualize this, imagine that Joe Laserjock is trying to find his way to the top of a hill, there is a dense fog, so he can t see where he is going, all he has is a compass and altimeter (back in the pre- GPS dark ages). So he decides on the following algorithm: 1) walk along an east-west line (A in figure) until he finds a maximum altitude and then along a north-south line to a maximum (B). If he has a nice symmetric hill, as shown, he finds the maximum. North A B But Josephine Laserjockette is in a tougher situation, she is trying to find the top of long narrow hill that is aligned along a SW-NE axis. If she just follows the same algorithm as Joe, she gets stuck at point B, which is not the maximum. But by using the algorithm above (shown in green), she can find the top. North A B The moral is: sometime you have to go down to go up. 5

6 Useful data Index of refraction of sapphire Table λ (nm) n e n o Sellmeir n Bλ B λ B λ = λ C1 λ C2 λ C3 n e B B B C C C n o 6