The most direct way to generate laser pulses is to add a modulator external to the continuous laser. This method can generate the fastest picosecond-level pulses. Although simple, it wastes light energy and the peak power cannot exceed the continuous optical power. Therefore, a more efficient method to generate laser pulses is intracavity modulation of the laser, which stores energy in the off-time of the pulse train and releases it in the on-time. A comparison of the two methods is as follows:
Four common techniques for generating pulses through laser intracavity modulation are gain switching, Q switching (loss switching), cavity emptying, and mode locking.
- The gain switch generates short pulses by modulating the pump power. For example, semiconductor gain-switched lasers can generate pulses ranging from a few nanoseconds to hundreds of picoseconds through current modulation. Although the pulse energy is low, this method is very flexible, such as providing adjustable repetition frequency and pulse width. Researchers at the University of Tokyo reported a femtosecond-level gain switching semiconductor laser in 2018, which meant a breakthrough in the 40-year technical bottleneck.
- Strong nanosecond pulses are generally generated by Q-switched lasers. The laser is emitted within several round trips within the cavity. The pulse energy ranges from a few millijoules to a few joules, depending on the size of the system.
- Medium energy (generally below 1 μJ) picosecond and femtosecond pulses are mainly generated by mode-locked lasers. There are one or more ultrashort pulses in continuous circulation in the laser resonant cavity. Each intracavity pulse is emitted through the output coupling mirror. Pulse, repetition frequency is generally between 10 MHz and 100 GHz. The figure below shows a fully normal dispersion (ANDi) dissipative soliton femtosecond fiber laser device, most of which can be built using Thorlabs standard components (fiber, lens, mounting base and displacement stage).
- Cavity depletion technology can be used not only for Q-switched lasers to obtain shorter pulses, but also for mode-locked lasers to increase pulse energy at lower repetition frequencies.
Time and frequency domain pulses
The linear shape of the pulse changing over time is generally relatively simple and can be represented by Gaussian and sech² functions. Pulse time (also called pulse width) is most commonly represented by the half-maximum (FWHM) value, that is, the width spanned by the optical power at least half of the peak power; nanosecond-level short pulses are generated by Q-switched lasers, and several nanosecond pulses are generated by mode-locked lasers. Ultrashort pulses (USP) from ten picoseconds to femtoseconds. High-speed electronics can only measure tens of picoseconds at best, and shorter pulses can only be achieved with purely optical technologies, such as autocorrelators, FROG and SPIDER.
If the pulse shape is known, the relationship between pulse energy (Ep), peak power (Pp) and pulse width (𝜏p) is calculated according to the following formula:
where fs is a coefficient related to the pulse shape, which is approximately 0.94 for Gaussian pulses and approximately 0.88 for sech² pulses, but is generally calculated approximately as 1.
The bandwidth of a pulse can be expressed as frequency, wavelength, or angular frequency. If the bandwidth is small, the wavelength and frequency bandwidth are converted using the following formula, where λ and ν are the center wavelength and frequency respectively, and Δλ and Δν are the bandwidth expressed in wavelength and frequency respectively.
Bandwidth limit pulse For a specific pulse shape, the spectrum width of the pulse is minimum when there is no chirp. At this time, we call it bandwidth limit or Fourier transform limit pulse. The product of its pulse time and frequency bandwidth is a constant. This constant is called time. Bandwidth Product (TBP). The time-bandwidth products of the bandwidth-limited Gaussian and sech² pulses are approximately 0.441 and 0.315 respectively; based on this, the actual pulse chirp amount and accumulated group delay dispersion can also be calculated.
Therefore, a narrower pulse width requires a wider Fourier spectrum. For example, the bandwidth of a 10 fs pulse must be at least on the order of 30 THz, while the bandwidth of an attosecond pulse must be larger, and its center frequency must be much higher than any visible light frequency.
|
1 ms (millisecond) = 10−3 s |
1 ps (picosecond) = 10−12 s |
|
1 μs (microsecond) = 10−6 s |
1 fs (femtosecond) = 10−15 s |
|
1 ns (nanosecond) = 10−9 s |
1 as (attosecond) = 10−18 s |
Factors affecting pulse width
While nanosecond or longer pulses barely change in width as they propagate, even over long distances, ultrashort pulses can be affected by a variety of factors:
Chromatic dispersion can cause large pulse broadening, but the opposite dispersion can be used to recompress. The figure below shows the working principle of Thorlabs' femtosecond pulse compressor to compensate for microscope dispersion.
Nonlinearity generally does not directly affect the pulse width, but it widens the conduction bandwidth, making the pulse more susceptible to dispersion during propagation.
Any type of optical fiber (including other gain media with limited bandwidth) may affect the bandwidth or the shape of ultrashort pulses, and the reduction in bandwidth may cause time broadening; there are also cases where the pulse width of a strongly chirped pulse becomes shorter when the spectrum narrows. .
