Threshold current variations with temperature

In the following sections, the results refer to the Ioffe laser. The device was mounted p-side up on a copper heatsink, and a Peltier cooler was used to control the operating temperature. The gain section of the laser was electrically pumped by continuous-wave bias, in order to study the performance of the device under real-life working conditions. The absorber section was reverse biased so as to achieve mode-locked operation.

Typical cw light-current characteristics for various heatsink temperatures are illustrated in Fig. 6.1, when the bias applied to the absorber section was 0 V. Under these conditions, a significant thermal rollover behaviour appeared only when the temperature was increased to 80oC.

Fig. 6.1 - Typical cw light-current characteristics of the two-section QD laser as a function of temperature, for a bias of 0V in the absorber section.

For each temperature, the reverse bias was varied between 0V and 10V, and the corresponding threshold current was measured, as depicted in Fig. 6.2. It was also observed that the modulation of losses by the reverse bias in the saturable absorber was more important for higher temperatures – and indeed, pronounced hysteresis was observed when temperatures exceeded 60oC.

In Fig. 6.2, only two threshold current values are featured for a temperature of 80oC. The reason for this is that at this temperature, the laser emission shifted to ES as the reverse bias was increased beyond 1V. The switch to ES at higher temperatures can be explained by the higher level of losses in the laser and also by the thermal excitation of carriers from the GS to the ES levels with increasing temperature [1].

Fig. 6.2 - Dependence of the threshold current on reverse bias and temperature.

Taking into account the phenomenological relationship Ith =Ith0exp

(

)

possible to calculate the characteristic temperature To, which is a measure of the

stability of threshold current Ith with temperature T (and where Ith0 is a constant) [2].

For example, an infinite characteristic temperature will imply that the threshold current is invariant with temperature.

In order to define a realistic To for these two-section lasers, it was taken into

account that the absorber section is never forward biased while in mode-locked

operation, and that current is pumped solely into the gain section. Therefore, To was

calculated using the threshold current values measured while the absorber was reverse

biased in the range of 0V – 10V. For temperatures equal and above 60oC, the lower

threshold current of the hysteresis loop was considered. A characteristic temperature To

of 41K was obtained for the range of operating temperature 20oC - 80oC. To was

relatively constant across the whole range of reverse bias values, confirming that it is essentially dependent on the properties of the gain section, as already demonstrated for

two-section quantum-well lasers, in a previous work [3]. This value of To is thus valid

across the whole mode-locked operating range.

At this point, several important comments are due here, as To is a finite and

actually quite small value. An infinite To had been predicted for QD lasers [4], where

the three-dimensional confinement in an infinite potential would lead to a δ-like density

of states, and therefore no thermal spreading of carriers would occur, even with increasing temperature, provided that higher sub-bands would not become populated – thus meaning that the carriers would always be confined to a single state. However, the

confinement potential is not infinite in a real QD. With increasing temperature, carriers are thermally excited to higher sub-bands [1] and/or to the wetting layers [5], which leads to significant gain saturation.

In line with these investigations, it has been demonstrated that techniques tackling gain saturation can greatly improve To. One of such techniques is p-doping [6,7], used extensively in recent years to decrease temperature sensitivity of the threshold current, resulting in values of To up to 650K [8]. Through p-doping, a surplus of holes is made available within the material, in order to reduce the effect of thermionic hole emission [9]. Indeed, because the hole levels are energetically more closely spaced than electron levels, the excitation of holes into higher subbands may lead to a significant reduction in the population inversion with increasing temperature.

The characteristic temperature To can also be improved by engineering the shape of the QD and increasing the energy separation between the GS and ES levels in the dot [10]. Additionally, gain saturation can also be minimised by increasing the number of QD layers in the laser material, which can also be combined with p-doping techniques [7]. It is important to refer that the Ioffe laser was not designed for temperature resilience in particular, and therefore did not include any of the aforementioned engineering optimisations.

Finally, it is important to add that Auger recombination could ultimately prevent a totally temperature-insensitive behaviour of QD lasers, in particular around room temperature [11]. It has been demonstrated recently that Auger recombination accounts for a very significant proportion of the overall recombination current in QD lasers emitting at 1.3µm or more16 [11,12]. The rate of Auger recombination increases with

temperature, and being a nonradiative process, it also has the effect of increasing the threshold current (although the radiative recombination current does remain constant with temperature in QD lasers).

Auger recombination may also be the main responsible for an increase in the homogeneous broadening with temperature [13], which could also have an impact on the spectral characteristics.

16 _{Auger processes become more important for lower bandgaps, as the carrier scattering becomes more} probable.