We generated an ultrashort but weak laser pulse from plasma, and we observed its interference patterns by using Mechelson interferometer. If we want to know its duration or width, how to extract this information from the interference patterns ?
You will get only a coherence time from the interference pattern rather than pulse duration. Try to use cross-correlation or autocorrelation procedure for measuring the laser pulse width.
@ A.A. Lonin Thank you for rapid response ! Ok, I see. But our laser beam is so weak (few nJ) that it seems have no chance to do cross-correlation or autocorrelation measurement, because it will be totally attenuated in the nonlinear crystall. Do you have other suggestions about how to obtain the pulse width of weak laser ?
well there are several ways to do it. probably the simplest one for week pulses is to use some nonlinear absorption (say two photon) fluorescing die and use pulse shaper to tune spectral phases to get maximal fluorescence (look at MIIPS from Marcus Dantus group ~ 2009-2010). In any case, what is important for any kind of nonlinear measurements is not so much the energy of the pulse, but, rather peak power. There is no way for measuring pulse duration without using some kind of optical nonlinearity. Any other measurement will provide you with coherence time - namely the theoretical limit of pulse duration, but you could get that simply measuring its spectrum with simple spectrometer.
@Yuri Paskover Thank you! So as you said, the coherence time gives the limit of pulse duration. Could you give me more details about it ?
I mean lower limit: basically it goes back to Fourier Transform relations (uncertainty principle) if you wish. Pulse is superposition of all its spectral components. The shortest pulse in time you will get if there is an instance in time where all the spectral components arrive with same the phase (say 0). So, taking Fourier transform of the spectrum of your pulse (probably, you should take square root of the spectrum prior to FT to work with amplitudes) you get temporal signature of your pulse assuming that spectral phase is 0 across your entire bandwidth.
This, be definition gives you the shortest possible pulse. Surprisingly, this action is very much similar to your interferogram, the width of which is also connected to the bandwidth of your pulse. Actual pulse duration has to do both with amplitudes and relative phases of the different spectral components.
In order to make different spectral frequencies to "talk to each other" you need some kind (in fact any kind) of nonlinear response to your pulse.
Coherence time (the width of the visibility envelope in your interferometer divided by speed of light) has to do only with amplitudes, simply because different spectral components do not interfere.
By the way.. few nJ should be plenty for autocorrelation, MIIPS, crosscoreallation.
I am less familiar with plasma pulse generation, what is central wavelength of your pulse? What is the bandwidth? I think I am only familiar with works on generation THz radiation from plasma, in such case, you can probably measure directly electric field, and not just intensity. In such cases, you will get your full pulse profile.
Igor, yeah, I kind of missed that point, few nJ might be more than enough for any kind of characterization. The only question is what is their peak power. I guess no special effort for pulse compression were taken, so duration can go to several ns. In such a case any nonlinearity might be rather challenging.
Perfect answers!
If we could only new approximately the wavelength of your pulse and what do you call ultrashort?
for example dozens femtoseconds pulses of Ti:Sa laser are usually fouirier-bounded and the described technique with simple spectrometer gives reasonable results.
few nJ... what is the repetition rate?
@ Yuri, @Igor, @Krutyanskiy Thanks !
The central wavelength of pulse is 337 nm, and we expect it will have dozens picosecond. The pump laser we used has a repetition rate of 100 Hz, so the lasing pulse at this wavelength from plasma also has same repetition rate.
As your opinion, it seems 1 nJ is enough for characterization. MIIPS method is new to me. Mabe it works !
for MIIPS you will need pulse shaper working with your wavelength range. as far as I know, only reflective ones might get under 400 nm. You will also need some process nonlinear with your light - say two photon absorption at 170nm, followed by fluorescence. I would check photoacids used in lithography, those might be rather close to this absorption band.
Instead, you can probably use small portion of your pump (if sufficiently shorter than your signal pulse) to do the characterization, using say simple Kerr cell, or whatsoever.
I mean crossed polarizers, cell of say CS2 pump and signal at 45 degrees polarization and controlable delay. Just look at say polarization gating FROG by R Trebino or something.
Good luck.
The full answer to your question shall be found there : http://www.swampoptics.com/tutorials_interferometric.htm
Yep, now it is clear that the task is not easy. You have to check the suggested above two-photon techniques for working with deep UV output. I'm afraid that 170 nm will be too short, it will be absorbed by everything. At the same time with 100Hz you will have very low average power.
The suggestion of Yuri to mix with the pump is a good way, if pump is really much shorter.
The other option - is to measure directly by a streak camera, but I never dealt with it http://en.wikipedia.org/wiki/Streak_camera
Few dozens of ps... well you should be able to measure that directly with fast PD.
I always forget that different people think of very different pulse durations when they say 'ultrafast' :) few dozens of ps - 'barely ultrafast'. Good luck. Streak should work too. Post the findings.
- Badges
- Science topic
- Similar topics
- Biological Science
- Microbiology
- Virology
- Virus