There are five factors that define the performance of SESAM or SBR devices.
i) Ultrafast carrier dynamics are essential for any saturable absorber because successful operation requires that an absorber should be saturated on a short timescale and recover to its original state of high loss (relax) on an equally short timescale. By this token, the recovery time of the absorberA, after a pulsed induced saturationis of
critical importance to the production of ultrashort pulses. Ideally, after complete saturation by a pulse, the absorber should be able to completely relax long before the arrival of the next pulse. Therefore the recovery time of the absorber must be less than the round trip period of the cavity. A similar condition applies to the gain if differential amplification instabilities are to be avoided. Semiconductor lasers can have very short cavity lengths. For instance, a 3mm long laser has a cavity frequency of around 15 GHz and this corresponds to less than 60 ps for its round-trip period. ii) Another very important factor is the saturation fluence FsatA, which is defined as the
pulse fluence required, such that the remaining fractional unsaturated loss of the absorber after such a pulse is equal to 1-e-1. This dynamic process can also be understood in the context of pulse energy. The saturation fluence is equal to the saturation energy EsatA per unit area of the laser mode cross section. To initiate and
gain should be less than that required to saturate the gain as outlined in the expressions below. satA satG h A h A E E a g N N 2.22
In the above expression, is the optical frequency of the laser, h is Planck’s constant while a,g/N are the differential loss and gain i.e. the loss and gain per carrier. In satisfying 2.27 the existence of the gain window shown in figure 2.5 is allowed. Fortunately, in semiconductor material this favourable situation is ensured since g
n n
[24]. In this project, the specific design of the external cavity for the
QD SESAM mode-locked QD lasers enables the fluence to be varied somewhat by varying the cavity length and making small adjustments to the spot size on the SESAM.
Figure 2.6. SESAM reflectivity as a function of incident pulse fluence
Concentrating specifically on mirror based saturable absorbers; there are three main parameters that describe how such absorbers perform (figure 2.6). The modulation depth R of the SESAM is the maximum change in reflectivity between the relaxed absorber state and that exhibited when a pulse with large fluence fully bleaches the absorber. The pulse will have an energy density which is larger than the saturation fluence of the absorber FsatA. By using SESAMs with large modulation
depths, strong modulation of the laser field can be achieved however, excessive modulation depths can cause problems such as an increase in Q-switching instabilities
[25]. These occur because the modulation of the laser field becomes strong enough or the recovery of the gain slow enough such that successive pulses experience differing levels of attenuation. The consequence of this is that the mode-locked pulse train becomes subject to further modulation by the comparatively slowly varying Q- switched envelope [22].
The non-saturable loss, ΔRNS, is the loss of the device which remains after
bleaching by a highly intense pulse. The magnitude of this loss factor is strongly related to the manner in which the device has been designed. Various factors including scattering losses, residual absorption and the reflectivity of the bottom Bragg mirror are taken into account inΔRNS. The quantity R0RNSis the reflectivity of
the device when the pulse fluence incident on the device is close to zero. By a similar token, RNSis the device reflectivity when a very highly fluent pulse fully saturates the
absorber.
Finally, we consider the spectral bandwidth of the SESAM. The spectral bandwidth determines the upper limit on the number of modes which the SESAM can lock together and consequently the minimum pulse duration obtainable with that particular SESAM. This parameter is usually limited by the reflection bandwidth of the Bragg mirror located beneath the absorber layer. Several methods such as selective oxidation have been utilised in order to increase the refractive index contrast between the layers of the mirror stack and thus increasing the bandwidth of the device. In Chapter 5 I describe the use of a QD SESAM that features a very wide spectral bandwidth.
There are several types of commonly used SESAM structures. The first types of SESAM were high finesse antiresonant Fabry-Perot structures (HFA-FPSA). The basic material structure of such a device is shown below in figure 2.7a [26].
Figure 2.7. (a) HFA-FPSA material structure and (b) LFA-FPSA material structure.
In the case of 2.7a, a bottom Bragg reflector consisting of very thin alternating layers of GaAs and AlGaAs with thickness λ/4 is grown on top of a GaAs substrate. The absorber layer typically consists of a quantum well, multiple wells or layers of quantum dots grown on top of this. Finally a top mirror is provided by a highly reflecting SiO2/TiO2Bragg structure. This creates a Fabry-Perot (FP) type of structure
in the absorber region of the device. The thickness of this region is chosen such that it is antiresonant at the wavelength of operation, meaning that minima in the electric field intensity are located at the edges of the Fabry-Perot structure. When this is satisfied, the device exhibits broadband reflection and minimal group velocity dispersion [22, 26]. The bandwidth of the SESAM is limited in most cases by either the reflection bandwidth of the Bragg mirrors or by the free spectral range of the Fabry-Perot cavity. The inclusion of spacer layers between the absorbing layer(s) and the Bragg mirrors can alter the saturation fluence of the device by changing the distribution of the electric field maxima in relation to the absorbing layers of the device.
An alternative SESAM configuration involves the removal of the upper highly reflecting mirror. In this configuration, shown in figure 2.7b, Fresnel reflections provide the feedback at the top of the Fabry-Perot cavity. This low finesse structure requires a thinner absorber and a bottom mirror exhibiting higher reflectivity to minimise losses [22]. In the figure, the absorber areas are placed at positions of
maxima in the F.P. field. The saturation fluence can be altered by changing the position of the absorbing section relative to the maximum of the standing wave inside the Fabry-Perot cavity. Because different wavelengths will have different nodal positions in the standing-wave profile, this can be used to reduce the wavelength dependence of the absorber edge, thereby giving improved broadband performance.
SESAM structures can also be designed to operate in resonance. This is achieved by careful choice of both cavity length and the layer of material terminating the cavity. In this configuration, the standing wave inside the Fabry-Pérot cavity has an anti-node of intensity at the Fabry-Pérot boundary. These types of devices exhibit different GVD behaviour and typically lower saturation fluences due to the enhanced Fabry-Pérot field [27].The quantum-dot SESAM discussed in Chapter 5 was designed to operate close to resonance.