Related:
Diffraction grating
Intro to ultrashort pulsed lasers
A good video on Femtosecond lasers:
Ultrashort pulsed lasers are generated by mode-locked lasers.
They provide high power in a really short time. A good illustration is the holes made by continuous wave laser vs nano vs pico vs femto. In femto lasers, one can but without burn (burn really fast that fire cannot take hold). This is why femtosecond laser are cool and scary (image it falling on the eyes - you will lose photoreceptors without even knowing).
Mode-locked laser:
- Basically the phases of multiple (infinite) frequencies of light are locked together. Or more concretely only when the phases of different frequencies are locked together is when the pulse is generated (Fourier idea applies here).
- Because phase locking repeats, you can get pulse trains.
The Fourier uncertainty idea comes in here nicely:
- In order to get really short pulses, you need a lot of frequencies (Fourier transform).
- If you have only a few frequencies to play with, there is a limitation of the size of the pulse (has to be long).
Dispersion problems
Because the lasers have a lot of frequencies/wavelengths, we have extra problems that arises:
- Spectral (spatial) dispersion due to optics - the pulse is separated into colors in space. This is the standard spatial rainbow. Optics like prisms, diffraction gratings and AODs cause spatial rainbows.1
- Temporal dispersion due to optics - the pulse width begins to decay. The red comes first and blue later, causing the pulse to look longer. This is called a temporal rainbow. Optics like lens causes temporal rainbows.1
Temporal dispersion compensation
A way to compensate for temporal dispersion is to ensure that the red travels after the blue. This way, after the light passes through the optics, the red will speed up, blue will slow down and they will form a narrow pulse.
One can use prisms and diffraction gratings to do this. This is the old approach.
The current approach is to use chirped mirrors which introduces negative dispersion in a compact and precise way. This is used to compress pulses and used to mode-lock lasers (and used inside current femtosecond lasers).
This sort of method has limitations on power. Since the femtosecond laser is energy intense, going beyond mW of power makes it hard as it begins to burn this. The method people use is to make a femtosecond laser, disperse the pulse and amplify it, and then compress the pulse before using it, which can let it go to Petajoules of power and burn things. Interesting and crazy stuff - a noble prize for chirped pulse amplification for the folks who came up with this in 2018.
Measurement of the femtosecond pulse
Use of Fourier transform to obtain the time domain signal of the pulse. It needs two things:
- Amplitude of frequencies in the pulse - can be measured using a spectrometer.
- Phases of the frequencies in the pulse - the tricky bit.
The phase essentially indicates the arrival time of the frequencies. If it is flat, all the frequencies arrive at the same time giving a strong, sharp pulse. Dispersion modifies the phase, making it curved (quadratic), inducing decays of the pulse.
Autocorrelator: splits the beam, delays one beam so that the pulse on beam one can be correlated with beam two. The correlation happens through non-linear optics which respond maximally when both beams are correlated. Autocorrelation provides the convolution between the two pulses. A detector senses the changes in the correlation with respect to the delay.
FROG - similar method. Instead of measuring the convolution using a detector, a spectrophotometer is used instead.
SPIDER - create two spectrally shifted replicas of the pulse to get phase information.
d-scan - dispersion scan. Measures the laser pulse through non-linear optics using a spectrophotometer.