Question about aligning a cavity

[Malamala]

Hello! I have a laser that I want to mode match to a bow tie cavity. For my setup I only care about power amplification inside the cavity, not about the actual frequency that gets amplified. So the cavity itself is not super stable, but I want to use a servo to control the laser wavelength such that the laser follows the cavity, such that it is always mode matched and hence I get amplification.

The cavity has a linewidth of 25 kHz and the laser has a linewidth of 5 kHz (at 1064 nm and 20 W power). Currently I am trying to align the cavity and I am scanning the laser frequency while adjusting the mirrors. I am able to see (by measuring with a diode the transmission from one of the mirrors) the cavity modes (see attached figure), but they look kinda weird. I am pretty sure these are due to amplification inside the cavity, as if I block one of the paths inside the cavity all the peaks are gone (also I can see the light getting transmitted out of the cavity by naked eye, using an IR viewer card, and I wouldn't be able to, unless I have some power amplification).

However, as you can see in the figure, the height of the peaks varies a lot. Also if I zoom into only one of the peaks, it doesn't have a nice Lorentzian shape. Instead, it has multiple, smaller and spread peaks inside of it (I can take a picture of that, too, if needed). I would think that vibrations might play a role, as I am currently not floating the table I am using, but it is still an optical table still, with good damping, so I wouldn't expect vibrations to play such a big role (but I am not sure).

It is the first time I am aligning a cavity, so I would really appreciate any insight into this (or ways in which I can debug this issue). Please let me know if you need further details about the setup. Thank you!

 

[Twigg]

Do you have a CCD camera lying around? You'll also need a bunch of attenuators (maybe you can re-purpose the ones you used on your photodiode?). Normally, CCD's are silicon-based and not sensitive to 1064, but since you have a whole boatload of power you should be able to find a sweet spot where you can see it without frying anything. If you can get an image of your beam, you'll see which way you need to walk the mirrors to get rid of the higher order modes.

Instead, it has multiple, smaller and spread peaks inside of it

What kind of splitting are we talking here? Also, is there any chance you could draw a frequency scale on the scan waveform you posted? You can do this by plotting the scan waveform side-by-side, and by knowing the frequency range of your scan.

What's the FSR of the cavity? If you don't know it, what's the resonator length (the total path length for going through the whole bow-tie shaped loop)?

The last thing that'd be nice to know is what your beam waist is inside the cavity. This should be determined by the mode-matching lenses.

My guess is that you are just still far from gaussian and exciting many different transverse modes. Back to the optics table with you

I wouldn't worry about vibrations when your laser is 5kHz broad (edit: at least, not under these conditions). Vibrations cause broadening, they don't really create new modes.

On a more humorous note, I'm jealous! I'm walking a high finesse cavity right not and I haven't even seen any transmission yet, much less transverse mode structure! Lucky!

 

[Malamala]

Do you have a CCD camera lying around? You'll also need a bunch of attenuators (maybe you can re-purpose the ones you used on your photodiode?). Normally, CCD's are silicon-based and not sensitive to 1064, but since you have a whole boatload of power you should be able to find a sweet spot where you can see it without frying anything. If you can get an image of your beam, you'll see which way you need to walk the mirrors to get rid of the higher order modes.What kind of splitting are we talking here? Also, is there any chance you could draw a frequency scale on the scan waveform you posted? You can do this by plotting the scan waveform side-by-side, and by knowing the frequency range of your scan.

What's the FSR of the cavity? If you don't know it, what's the resonator length (the total path length for going through the whole bow-tie shaped loop)?

The last thing that'd be nice to know is what your beam waist is inside the cavity. This should be determined by the mode-matching lenses.

My guess is that you are just still far from gaussian and exciting many different transverse modes. Back to the optics table with you

I wouldn't worry about vibrations when your laser is 5kHz broad (edit: at least, not under these conditions). Vibrations cause broadening, they don't really create new modes.

On a more humorous note, I'm jealous! I'm walking a high finesse cavity right not and I haven't even seen any transmission yet, much less transverse mode structure! Lucky!

Thanks a lot for your reply! I attached below a new figure, after adjusting the cavity a bit more. The yellow is the measured transmission and the green is the ramp frequency. The FSR of the cavity is around 300 MHz and the peak to peak voltage of the ramp should correspond to about 400 MHz. The peaks are still changing amplitude quite a lot, as well as the distance between any 2 consecutive peaks changes. Moreover, the width of these peaks appears to be of the order of ~100 MHz, while the linewidth of the cavity should be around 25 kHz. I am using this camera to see the modes and I do see many different modes coming and going while the frequency is scanned.

I am currently not mode matching the cavity, that will be the next step I will do when I go back to lab (I thought that adding a telescope for mode matching would be just a fine tuning, but it seems like it plays a much bigger role than I thought).

Could you please tell me a bit about how should I proceed in order to end up just with a gaussian mode? Adjusting the height of the peaks it's easy, just by looking on the oscilloscope, but I am not sure how I would know if I am going into the right direction (while watching the modes changing on the camera) in terms of obtaining just one mode.

 

[Twigg]

Lookin' good! Nice work

You might want to slow your scanning frequency so you don't see so much ringdown (that exponential decay on each peak). In theory some modes could be hiding behind that ringdown signal.

Since you've gotten the alignment this nice, the CCD won't be needed anymore. The advantage of a CCD is that when you're very far from aligned, you can see the elliptical path the beam traces out (at least, on a Fabry-Perot cavity its elliptical, might be funkier in a bowtie). When you walk the beam with steering mirrors, you see that ellipse collapse on itself into a single point, scanning over a couple different modes. Based on your waveform. If you looked now on a CCD now, you'd probably see a single dot flickering between 2 or 3 modes, so it's not helpful anymore.

Remember the process you used to get this alignment, because you're going to need to do it again if you add a telescope.

As far as getting a gaussian mode, you've pretty much got it. Try taking a beam burn when the piezo is parked at maximum transmission.

 

[Twigg]

oh, if you've never taken a beam burn before, don't put the paper close to any optics! Sometimes it coats them in soot

 

[Malamala]

Lookin' good! Nice work

You might want to slow your scanning frequency so you don't see so much ringdown (that exponential decay on each peak). In theory some modes could be hiding behind that ringdown signal.

Since you've gotten the alignment this nice, the CCD won't be needed anymore. The advantage of a CCD is that when you're very far from aligned, you can see the elliptical path the beam traces out (at least, on a Fabry-Perot cavity its elliptical, might be funkier in a bowtie). When you walk the beam with steering mirrors, you see that ellipse collapse on itself into a single point, scanning over a couple different modes. Based on your waveform. If you looked now on a CCD now, you'd probably see a single dot flickering between 2 or 3 modes, so it's not helpful anymore.

Remember the process you used to get this alignment, because you're going to need to do it again if you add a telescope.

As far as getting a gaussian mode, you've pretty much got it. Try taking a beam burn when the piezo is parked at maximum transmission.

Thank you for your reply. But now I am confused about what to do next, beside adding the telescope. Right now, adjusting the mirrors won't increase the peaks anymore. But the peaks are about 3 orders of magnitude broader than expected and the power transmitted is much lower than expected, too (also around 3 orders of magnitude). I haven't measure the power yet, but just by eye, using an IR viewer, it is a lot less than what I expected. Also the location of the peaks is not fixed. The peaks move left and right continuously (that might be vibrations, tho). I have a video of that, but it seems like I can't upload it here. Would a telescope actually fix all these issues or am I still missing something? Shouldn't the peaks actually have a linewidth close to the cavity linewidth? And shouldn't I see more power transmitted, even without the telescope?

What is a beam burn?

 

[Twigg]

adjusting the mirrors won't increase the peaks anymore

Normally this would suggest that you are as well aligned as can be already.

Also the location of the peaks is not fixed. The peaks move left and right continuously (that might be vibrations, tho).

Vibrations don't cause drifts on a seconds timescale, they're usually in the 100's of hertz (they can broaden, but not drift). The number one culprit for drifts on this timescale is thermal fluctuations. Did you make sure the laser's temperature servo was running? Are you sure the TEC isn't just railed to one side, unable to reach steady state? If not, that's what's causing the drifts, I (almost) guarantee it.

If it's not temperature causing the drift, it's probably optical feedback. I don't know how you would rectify that for a 20W laser. For a low power laser (~100mW) you'd use optical isolators. I don't know if they make those for such high power, but it's worth checking.

On second thought, since you don't care about the laser's exact frequency (per your first post), maybe you don't care if there's feedback or not. If you were trying to maximize feedback, you would block the path between the laser and the bowtie, then turn the diode current down to juuuust below the lasing threshold. Then you'd unblock the path to the cavity, and the laser should lase. Then turn the current down again to juuuuust below lasing threshold, and walk the cavity alignment until the laser starts to lase again. Rinse and repeat. I'm not saying you *should* try to maximize feedback, I'm just not sure I understand what your goals are here, so I'm covering all possibilities.

But the peaks are about 3 orders of magnitude broader than expected and the power transmitted is much lower than expected, too (also around 3 orders of magnitude).

Hmmm, that's not good. Just to clarify, you are saying that the ratio of transmitted power to incident power is low, correct? It's not simply that your incident power is low? Just checking.

If you're getting the 3x broader than expected from the waveform you shared above, don't sweat it. What you saw in the above plot was ringdown, not a scan of the cavity linewidth. The exponential shape is a giveaway. Decrease the frequency at which you apply the frequency scan by a factor of 10x and watch what happens to the peaks' linewidth.

Would a telescope actually fix all these issues or am I still missing something?

That depends on the radius of curvature of the bowtie mirrors. The flatter they are, the less mode-matching matters (and the more alignment matters). Can you tell us?

 

[Twigg]

What is a beam burn?

You shine your laser on a special kind of paper that burns and makes an image of your beam profile. The paper is made with some combination of photographic reagents and/or saltpeter IIRC. For when your beam would turn a CCD sensor into slag.

 

[Malamala]

Normally this would suggest that you are as well aligned as can be already.Vibrations don't cause drifts on a seconds timescale, they're usually in the 100's of hertz (they can broaden, but not drift). The number one culprit for drifts on this timescale is thermal fluctuations. Did you make sure the laser's temperature servo was running? Are you sure the TEC isn't just railed to one side, unable to reach steady state? If not, that's what's causing the drifts, I (almost) guarantee it.

If it's not temperature causing the drift, it's probably optical feedback. I don't know how you would rectify that for a 20W laser. For a low power laser (~100mW) you'd use optical isolators. I don't know if they make those for such high power, but it's worth checking.

On second thought, since you don't care about the laser's exact frequency (per your first post), maybe you don't care if there's feedback or not. If you were trying to maximize feedback, you would block the path between the laser and the bowtie, then turn the diode current down to juuuust below the lasing threshold. Then you'd unblock the path to the cavity, and the laser should lase. Then turn the current down again to juuuuust below lasing threshold, and walk the cavity alignment until the laser starts to lase again. Rinse and repeat. I'm not saying you *should* try to maximize feedback, I'm just not sure I understand what your goals are here, so I'm covering all possibilities.Hmmm, that's not good. Just to clarify, you are saying that the ratio of transmitted power to incident power is low, correct? It's not simply that your incident power is low? Just checking.

If you're getting the 3x broader than expected from the waveform you shared above, don't sweat it. What you saw in the above plot was ringdown, not a scan of the cavity linewidth. The exponential shape is a giveaway. Decrease the frequency at which you apply the frequency scan by a factor of 10x and watch what happens to the peaks' linewidth.That depends on the radius of curvature of the bowtie mirrors. The flatter they are, the less mode-matching matters (and the more alignment matters). Can you tell us?

I am not sure I know much about the TEC or the temperature properties of the laser. Currently I am using it at the factory setup (I was told that they were optimized for proper functionality). But I will look more into that.

I don't think the optical feedback is a problem. The input coupling mirror is at a 3 degrees angle, so the light reflected from the back of the mirror doesn't go back into the laser (I actually need to use a beam block, in order to block that reflected light). Can there be other sources of optical feedback?

In terms of power, the incident power is the desired one (actually right now I am using 0.5W, but the maximum will be 20). My cavity is expected to amplify the power by a factor of 1000, so I would have 500W inside the cavity. And the transmission of the mirror I used for the power measurement is $10^{-4}$, so I expect 0.05W leaving the cavity. It looks smaller than that by a significant amount.

For the curvature, 2 of the mirrors (including the input coupling) are flat. The other 2 are concave with a radius of 25 cm.

 

[Twigg]

I am not sure I know much about the TEC or the temperature properties of the laser. Currently I am using it at the factory setup (I was told that they were optimized for proper functionality). But I will look more into that.

Hmmm, at the very very least I'd expect there to be a "lock" or "servo on/off" switch for the temperature control. Usually I'd also expect there to be a temperature control setpoint, but if it's just optimized for power at 1064 then who knows.

For the curvature, 2 of the mirrors (including the input coupling) are flat. The other 2 are concave with a radius of 25 cm.

Yeah, I'd go ahead and add the mode matching telescope, and see where that takes you. You have to do it sooner or later anyways.