Questions about Lock-in detection in optical reflectivity

[alanwake90]

Hi guys~
I have a question about lock-in detection method that commonly used in optical spectroscopy experiments, for example, the optical pump-probe reflectivity exeperiments on some semiconductor surfaces, or kinda like that.

Usually, the sinario is: you use laser pulses, divided as pump and probe, for inducing and detection, not so hard to get that picture, and you need an optical chooper or some other modulation methods, also the frequency of your chopper is connected to a lock-in amplifier, of course, to get the signal buried in noises. I think some of you are definitely pretty familiar with this sort of experiments.

here comes my question, which confused me alot, let me put it this way:
it's said that lock-in amplifier will ideally and only select the signal that equals to your reference signal, other components are filtered and reduced. but how do you know that your reference signal from the optical chopper is comparable to the real signal you want to measure?
specifically, let's say you want to measure some really fast phenemenon such as phonon oscillations in semiconductor, with ultrafast laser pump-probe setup, in this case, how to choose the frequency of your chopper (and also it's the reference frequency for lock-in)? is that the case that if your signal is oscillating at GHz level, you are going to choose the chooper frequency at the same level? that sounds rediculous so that I don't get it.

I don't know if I made myself clear, hopefully someone would enlighten me on this.
Thx in advance.

 

[tech99]

Hi guys~
I have a question about lock-in detection method that commonly used in optical spectroscopy experiments, for example, the optical pump-probe reflectivity exeperiments on some semiconductor surfaces, or kinda like that.

Usually, the sinario is: you use laser pulses, divided as pump and probe, for inducing and detection, not so hard to get that picture, and you need an optical chooper or some other modulation methods, also the frequency of your chopper is connected to a lock-in amplifier, of course, to get the signal buried in noises. I think some of you are definitely pretty familiar with this sort of experiments.

here comes my question, which confused me alot, let me put it this way:
it's said that lock-in amplifier will ideally and only select the signal that equals to your reference signal, other components are filtered and reduced. but how do you know that your reference signal from the optical chopper is comparable to the real signal you want to measure?
specifically, let's say you want to measure some really fast phenemenon such as phonon oscillations in semiconductor, with ultrafast laser pump-probe setup, in this case, how to choose the frequency of your chopper (and also it's the reference frequency for lock-in)? is that the case that if your signal is oscillating at GHz level, you are going to choose the chooper frequency at the same level? that sounds rediculous so that I don't get it.

I don't know if I made myself clear, hopefully someone would enlighten me on this.
Thx in advance.

I think the chopper is applying pulse modulation to the optical carrier in the transmitter. In the receiver, the pulse modulation is first recovered by envelope detection, and is then synchronously detected using the TX pulse stream as a reference. This will give a zero frequency DC output to our own signal and allow noise to be low pass filtered.
If the phenomenon is varying, it will amplitude modulate the pulse train. If it is fast, the LPF will then need to be widened in order to see it, and you may need to raise the chopper frequency to avoid fold-over of the modulation spectrum at zero frequency. The chopper frequency must be higher than the phenomenon being observed (by factor >2?).

 

[alanwake90]

I think the chopper is applying pulse modulation to the optical carrier in the transmitter. In the receiver, the pulse modulation is first recovered by envelope detection, and is then synchronously detected using the TX pulse stream as a reference. This will give a zero frequency DC output to our own signal and allow noise to be low pass filtered.
If the phenomenon is varying, it will amplitude modulate the pulse train. If it is fast, the LPF will then need to be widened in order to see it, and you may need to raise the chopper frequency to avoid fold-over of the modulation spectrum at zero frequency. The chopper frequency must be higher than the phenomenon being observed (by factor >2?).

Thank u so much for ur reply, But I still don't get it, I saw some papers taking about measuring THz oscillation using the optical pump-probe method I mentioned, on coherent phonon dynamics, for example. and they said that the chopper is used to chopping the pump laser at kHz level.I don't think currently we have a chopper that can operate at THz, sounds impossible.
In a word, I just confused that, since the lock-in amplifier only select the signal (closely) equals to the reference, then how to exactly understand the reason why signals could be obtained when they are really fast.
Thx in advance.

 

[tech99]

Thank u so much for ur reply, But I still don't get it, I saw some papers taking about measuring THz oscillation using the optical pump-probe method I mentioned, on coherent phonon dynamics, for example. and they said that the chopper is used to chopping the pump laser at kHz level.I don't think currently we have a chopper that can operate at THz, sounds impossible.
In a word, I just confused that, since the lock-in amplifier only select the signal (closely) equals to the reference, then how to exactly understand the reason why signals could be obtained when they are really fast.
Thx in advance.

Maybe they are using a chopper and just looking at the received signal with a CRO or spectrum analyser.

 

[alanwake90]

Maybe they are using a chopper and just looking at the received signal with a CRO or spectrum analyser.

Thx for ur reply very much.
I think they were using the photodetector(or a combination of photodetectors, the signal to lock-in is A-B, photodetector A - Photodetector B to cut the background at first)to collect the signal.
I believe that nowadays the bandwide of photodetector is at GHz level at best, here I mean the response of sensitivity of the photodetector.
is this the case that the signal is locked-in at a frequency far from the "real" phenemena, and it's required that you need to have a lot of time for averaging? then how the lock-in here can be a "bandpass" filter?

 

[tech99]

Thx for ur reply very much.
I think they were using the photodetector(or a combination of photodetectors, the signal to lock-in is A-B, photodetector A - Photodetector B to cut the background at first)to collect the signal.
I believe that nowadays the bandwide of photodetector is at GHz level at best, here I mean the response of sensitivity of the photodetector.
is this the case that the signal is locked-in at a frequency far from the "real" phenemena, and it's required that you need to have a lot of time for averaging? then how the lock-in here can be a "bandpass" filter?

I am not sure, but I suspect that the set up is being used like a radar. The light beam, maybe chopped at quite a low frequency, is exciting optical resonances in the material under study. When the light pulse ends, we can see the continuing resonance. We cannot see individual cycles at THz, but we can see the envelope by using a photo detector with perhaps a CRO or a spectrum analyser.