When a person thinks of a laser beam, the general perception is of a continuous beam of light that stretches for a very long distance. This type of beam occurs when a laser is in 'continuous wave' mode. However, lasers can also operate in pulsed modes, by switching on and off rapidly. At first, this may seem to be an odd thing to want to do - after all, surely if the laser is switching on and off then the power that the laser produces will be less? This entry tries to explain why and how this switching is done.
Why Pulse a Laser?
In general there are two reasons why one would wish to pulse a laser beam. One is that, although there may be less power on average in the beam, the instantaneous power in one of the laser flashes may be thousands of times greater than the power in a continuous wave beam. The other is that a pulsed laser beam can give information on very rapidly-occurring processes, rather like a photo-finish in a race. It's a little like a very fast flashlamp.
High Peak Powers
Using a pulsed laser it is possible to create beams that have very low average powers, but very high peak powers. What is meant by these terms? Well, the average power is just what it sounds like. The average power in a laser beam is the power delivered, as an average, over a long period of time, say several seconds1. However, in pulsed mode the laser may spend a long time idle or switched off. One can think of this as an energy storage period2. When the laser switches on, all of the energy that is stored is delivered in a short period of time, which may be a very small fraction of the storage time. This means that the power in the pulse is many times than that of the average, since the average must account for the time that the laser is switched off as well3.
It can be shown that the larger the pulse spacing is compared to the pulse length, the greater the peak power.
High peak power lasers are used in particular for laser machining and to reach the regime of nonlinear optics.
Short Pulse Systems
In order to image or make a picture of a rapid event, one requires a device that responds even more rapidly. A photo-finish camera at the end of a race would be useless if the shutter remained open for a second, since in the 100 metres race the competitors can move ten metres in that time. So it is in science. In order to analyse a rapidly occurring event, a fast camera is required. Pulsed lasers, to date, offer the scientist the shortest manmade events possible4 and so are an invaluable tool for probing certain things. For example, electronic processes in molecules, including the formation and breaking of chemical bonds, can now be monitored using pulsed lasers in a technique called femtochemistry.
The exact duration of these pulses depends on the technique used to create them, and the properties of the laser material and cavity.
How Are Laser Pulses Created?
There are a number of ways to create laser pulses. Listed here are some of the more common techniques.
Gain switching is a process where the gain of the laser is switched low, then high, then low again and so on. Using this system it is possible to generate power fluctuations called 'spikes' with relatively short durations, say around the nanosecond mark. The laser generates spikes just after the gain is switched high5. A simple method of switching the gain is to switch the pump system on and off6; for example, when using a diode laser as a pump source, by modulating the supply current to the laser it is possible to switch it on and off rapidly - thus the gain is switched high and low. This causes the laser which is being pumped to spike with a frequency equal to the frequency of modulation of the diode laser.
Q-switching involves changing the characteristics of the laser cavity itself to cause spiking. Q-switching is also known as Q-spoiling or cavity spoiling7.
The Q-switching process involves the use of some extra optical element in the cavity which changes the loss of the cavity from high to low. One of the conditions for laser action to begin is that the loss in the cavity should be less than the round trip gain of the cavity. By inserting a lossy component into the cavity this condition can be prevented from being satisfied. In this situation the laser medium begins to store the pump energy, much like a capacitor stores electrical energy. Once this reaches some threshold value, the cavity loss is switched to low and this energy comes out of the cavity in one short burst, which has roughly the duration of the length of the cavity divided by the speed of light.
The switching process can be initiated in one of two ways - active or passive. In the active case the element inserted into the cavity has a switch on it which can be manually 'pressed'8; this initiates the pulse. In the passive case the cavity switches automatically, without any external process required. One way of doing this is to use some saturable element which, when the energy stored in the laser medium becomes high enough, temporarily bleaches - that is, becomes transparent. This allows the pulse to form.
Cavity dumping is performed in the opposite way to Q-switching. In the case of cavity dumping the loss of the cavity is switched from low to high.
A feature of all laser systems is that there must be some loss in the cavity, or else one cannot extract light from the system. This is usually in the form of a partially-reflecting mirror at one end, which allows a fraction of the light out. In the case of cavity dumping, this loss is minimised. The energy builds up in the cavity, this time not in the gain medium, but in the light field itself. Once this reaches a certain level, the loss is switched to high, and the light all exits the cavity in one round-trip. This is usually performed in an active fashion.
This is perhaps the most complicated method of creating laser pulses, but is also the most useful, as this seems to be the only method of generating pulses with durations of less than one picosecond9.
The mathematical details of modelocking are somewhat complicated; here only the broad generalities will be discussed. In a laser system, the exact frequency that the light can oscillate at is dependent on the length of the cavity, much like the frequency of sound produced by a wind instrument depends on the length of the column of air vibrating in the instrument10. This allows us to define cavity 'modes'. The first mode of the cavity is that frequency whose wavelength corresponds exactly to twice11 the length of the cavity. Higher frequency waves are called 'harmonics', in a similar fashion again to music. The condition that must be applied to all frequencies is that their corresponding wavelength must equal an even number of cavity lengths, or conversely that the cavity length must equal a whole number of half-wavelengths.
When applied to a laser cavity, the cavity length can reach millions or billions of wavelengths. The difference in frequency between these cavity modes can be relatively small compared to the frequency of the light used, and in many cases the finite bandwidth of the laser gain material can span several cavity modes. Certain materials can have very large bandwidths, and these may contain a large number of cavity modes.
The exact technique can vary, but one way of doing this is by applying some modulation to a wave. By doing this it is possible to create secondary frequencies at the original frequency plus and minus the modulation frequency. This can be applied to a laser beam and if the modulation is applied at a frequency equal to that of the difference between cavity modes, energy can be transferred between modes. The modes can become locked to each other, each one being influenced by the others. When this happens the laser begins to pulse in a regular fashion, with a repetition time equal to that of the round-trip time of the cavity. This is called modelocking and, in particular, when the modulation is induced by some user-controlled process, it is called active modelocking.
Passive modelocking is also possible and there are two types of process. Some lasers can be modelocked in a similar way to passive Q-switching, where there is some element in the laser which can be saturated. This method has the effect of creating a highish-power spike in the cavity which undergoes preferential gain, allowing it to build up and suppress all other power in the cavity. A second approach is called self-modelocking where the laser gain material can cause the system to modelock itself, albeit sometimes with a little help from the user, who can create the noise spike in the system by banging the table, for example12. Other procedures can be used to modelock the laser by introducing perturbations, and sometimes the laser will modelock itself, say if the airflow in the cavity is disturbed. Self-modelocking systems are often based around the optical Kerr effect.
One thing to note about modelocking is that there is a relationship between the pulse length and the frequency bandwidth of the pulse. In other words, the shorter the pulse, the more modes are required. Since the intermode spacing is fixed in frequency, this corresponds to a greater frequency bandwidth.
Looking at the different methods of pulse generation, it can be seen that gain switching gives fairly long pulses, Q-switching and cavity dumping shorter pulses and modelocking shorter yet. Different pulse durations can also be obtained for different types of laser. Looking specifically at Q-switching and cavity dumping, one of the limits on the pulse duration is the length of the cavity. Longer cavities give longer pulses; this can be got around by making the cavity shorter. Diode lasers have short cavities, as do some solid-state lasers, with cavities of less than one millimetre in length.
Modelocked lasers tend to have longer cavities (although it is technically possible to modelock a diode laser) but this is no impediment as the pulse duration is no longer dependent on the length of the cavity (see next section). Modelocked lasers typically give pulses with a duration of around 100 femtoseconds.
|Laser Type||Typ. Pulse Duration||Typical Repetition Rate||Approx. Duty Cycle13||Misc.|
|40 MHz||10 - 100||These are
e.g. neodymium:yttrium orthovanadate
|femtosecond||100 MHz||10000 - 1000000||These are very expensive|
This comparison of pulse techniques shows that modelocking gives the shortest pulses; the current record for a pulse direct from a laser is in the region of 4 femtoseconds. Most laser physicists involved in research into pulsed lasers feel that there is a fundamental limit as to how short a pulse can be, which is that a pulse may not have a duration shorter than one cycle of the electric field; thus, in the visible part of the spectrum pulses may be able to reach 2 femtoseconds before this limit is reached. However, the period of the electric field depends on the frequency of the radiation used, and if radiation in the ultraviolet or X-ray regions can be created, shorter pulses may be possible. To date, the shortest pulse ever created is around about 650 attoseconds (atto = 10-18; one attosecond is one millionth of one millionth of one millionth of a second. This was achieved by the use of high harmonic generation14 of a short optical pulse.