Relative efficiency of SHG vs SFG

  • #1
fevemo
2
1
TL;DR Summary
Is a SHG process (936 nm to 468 nm) similar in efficiency to a SFG process (850 nm+1040 nm to 468 nm) assuming same properties of the two lasers (both fs pulsed, same focal spot, same polarization, etc.?
I am trying to align two ~100-140 fs pulsed lasers, one at 1040 nm and one between 750-950 nm through a working custom 2p microscope. The end goal is to do stimulated Raman (CARS and SRS) which requires the two beams to be coincident and the pulses to be perfectly synchronized in time, however for now I am testing using a crystal that generates second harmonics signal.

To find the temporal alignment I look at the blue channel (~475 nm) signal from the microscope and maximize the SFG signal coming from the crystals when both the laser beams are enabled (1040 nm + 850 nm) by moving a delay line for the 1040 nm beam. This procedure works well and I can consistently find a small (200-300 fs) window where the SFG signal is clearly visible. This signal scales linearly with the power of each laser which reassures me it is in fact SFG.

Now the problem I am facing is that once I try to get some CARS/SRS signal there is practically none, even with samples that should generate very large responses. Going back to SFG I noticed that the efficiency is significantly lower than just setting the tunable beam to 936 nm. To get a similar intensity from SFG I need almost 10x more power on the sample than when using SHG with a single 936 nm beam (68 mW vs 7.4 mW). Obviously, I would expect for a single beam to be more efficient than two superimposed beams of different wavelengths, but 10x more power feels like too much. Is this an indicator that the focal points of the two beams are barely overlapping or is the 10x lower efficiency within reason for a properly aligned setup?

P.S.: I am aware CARS and SRS are usually done with ps pulsed lasers because of the spectral resolution required, I'm not concerned about resolution at the moment and fs-based CARS and SRS have been shown to work and even generate a stronger signal in some cases. Also, to add to the mystery, a previous version of this setup worked and gave me a decent SRS signal.
 
Science news on Phys.org
  • #2
I wonder if this is a dispersion thing because the wavelengths are different. Besides the sources, sample(s), and detectors what other optical elements are in your beam path?
 
  • Like
Likes DaveE
  • #3
Andy Resnick said:
I wonder if this is a dispersion thing because the wavelengths are different. Besides the sources, sample(s), and detectors what other optical elements are in your beam path?
Yes, maybe walkoff losses? SFG can sometimes work better if the beams aren't colinear.
 
  • #4
DaveE said:
Yes, maybe walkoff losses? SFG can sometimes work better if the beams aren't colinear.
Another good guess- it depends on the nonlinear crystal and specific phase matching condition.
 
  • #5
Thanks for the replies!

Andy Resnick said:
I wonder if this is a dispersion thing because the wavelengths are different. Besides the sources, sample(s), and detectors what other optical elements are in your beam path?
This has definitely been one of my suspects.
The current setup has the 1040 beam going through a delay line with ~10 silver mirrors, a 2X beam shrinker (to compensate for the divergence of the extra path length), and then joins the tunable beam using a notch filter. After that they both go through a beam expander (not ideal IMO), ~8 more mirrors, and then the microscope (scan + tube lenses, galvos, dichroic mirror, objective)

Some extra info that might be relevant:
  • The laser has GDD pre-compensation for the tunable beam and the best SFG signal is obtained when using the maximum pre-compensation possible (15000 fs2)
  • Using a very old GRENUILLE (FROG) device I estimate both of the beams are still under 160fs before they enter the microscope.
DaveE said:
Yes, maybe walkoff losses? SFG can sometimes work better if the beams aren't colinear.
I had never heard of this but I suspect I have observed it as slightly misaligning the 1040nm beam yields a better SFG signal under the objective. Now I know not to use this as a proxy to improving the SRS signal as there would be no walk-off in that case.

I realized after my original post that my current setup is probably introducing additional divergence for the 1040nm beam as I had the 2X beam shrinker mounted near the exit of the laser, meaning I was shrinking the beam from ~2mm to ~1mm. Could this be the source of issues? In the original implementation (which gave me SRS signal) the same beam shrinker was mounter after several meters of optical path, shrinking the beam from ~5mm to ~2.5mm.
 
  • Like
Likes DaveE

Similar threads

Replies
6
Views
2K
Replies
25
Views
4K
Replies
44
Views
8K
Replies
1
Views
2K
Replies
1
Views
2K
Replies
21
Views
15K
Back
Top