1. Introduction

The impact of ultra-short laser sources on many scientific and technological applications cannot be undermined. Recent progress in development of ultrafast solid-state lasers has significantly broadened their practical application range, forming the basis for many new inventions and discoveries in different areas of fundamental science. In this context, the development of semiconductor-based saturable absorber structures, namely semiconductor saturable absorber mirror (SESAM) [1] or saturable Bragg reflector (SBR) [2], has significantly improved performance of ultrafast laser systems permitting grater operational tolerances and practicality [3, 4]. Saturable absorber devices are characterized by their centre operation wavelengths, carrier dynamics, saturation fluence, and modulation depth [5]. These properties can be easily adapted to the specific requirements of laser systems through modification of resonant Fabry-Pérot microcavity, for example, an anti-resonant cavity configuration, a resonant microcavity design with a lower-index dielectric cap layer, etc [6, 7]. Semiconductor quantum-well- (QW) and quantum-dot- (QD) based saturable absorber structures have brought a range of complimentary characteristics for efficient and reliable mode-locking of lasers emitting in the near infrared spectral region [4, 7, 8].

Further development of ultrafast laser systems is focused on increasing functionality and flexibility of the sources. The ease of switching between continuous wave (CW) and mode-locked (ML) regimes, and control over the output pulse duration are beneficial for many practical applications, in particular for optical trapping and manipulation [9], multiphoton microscopy [10] and nanosurgery [11].

One way to implement ‘on-demand’ laser operating regimes is via active manipulation of saturable absorber macroparameters. The electrically-enhanced saturable absorber structures have been successfully utilized in the mode-locking of multisection QW- and QD-laser diodes [12, 13]. The pulse shortening was achieved by increasing the reverse bias direct current (DC) voltage on the absorber section [14].

The DC field-enhanced p-i-n junction SESAM structures were also employed for controllable operation of solid-state fibre and bulk lasers [15–18] allowing nearly threefold mode-locked pulsewidth shortening to be demonstrated [17, 18]. Recently the quantum-confined Stark effect (QCSE) has been conclusively identified as the primary mechanism enabling electrical control over the modulation depth in a p-i-n QW-SESAM structure [19], driving the ML output pulse shortening. However the QCSE in QD-based structures is expected to be reduced due to the inhomogeneous QDs size distribution, which masks any field dependent broadening associated with the separation of an electron-hole bound pair [20, 21]. To increase QCSE in the vertically coupled QDs [22] and strain-engineered bilayer QD structures [23] were used.

Carrier dynamics, electro-absorption and electro-refraction are important factors in determining the performance of field-enhanced SESAM structures. These are especially relevant for QD-based SESAMs as devices with great potential for the pulse shaping performance in mode-locked solid-state bulk and semiconductor laser systems. Until now field-enhanced QD-based saturable absorbers have only been investigated in the context of planar waveguide configurations [20–24]. No direct measurements of field-dependent nonlinear optical properties in p-i-n QD-SESAM structures have been presented to date.

In this work, we report the electric-field-enhanced absorption recovery dynamics and nonlinear reflectivity in a p-i-n junction QD-SESAM measured for a DC reverse bias ranging from 0 V to 20 V. The potential of the field-enhanced QD-SESAM devices for controllable ML operation of semiconductor disk lasers is discussed.

2. Experimental methods

The p-i-n junction QD-SESAM structure under investigation was grown using a conventional solid source molecular beam epitaxy (MBE). The structure comprised of a 500-nm Si-doped GaAs buffer layer deposited on an n-type GaAs substrate, an n-doped distributed Bragg reflector (DBR) and an 8 λ-long GaAs saturable absorber microcavity, with a design wavelength of 1060 nm, capped by a p-doped GaAs layer. The highly reflective DBR consisted of 33.5 pairs of Si-doped λ/4 GaAs/Al_0.9Ga_0.1As layers with a stopband ranging from 1020 nm to 1120 nm. The DBR doping concentration was about N_D = 2×10¹⁸ cm⁻³. The undoped QD saturable absorber region was formed in 8 groups consisting of 10 layers of InGaAs QDs, grown in the Stranski-Krastanov regime, with 10-nm thick GaAs spacer layers. The QD groups were positioned at the antinode of the electric field standing wave using GaAs layers. The total thickness of the intrinsic section of the device was ~1365.2 nm. The potential drop across the p-i-n junction was approximately −1.9 V which resulted in the build-in electric field on the QDs of ~13.9 kV/cm. The p-GaAs cap layer had a doping concentration of approximately N_A = 1×10²⁰ cm⁻³. The full area of the n-side was metalized with GeAu/Ni/Au alloy. On the p-side, stripe contacts with a separation distance of 300 μm were formed with a ZnAu/Au alloy. The reflectivity spectrum of the p-i-n SESAM structure exhibits a resonance dip centred at ~1.06 µm as a result of the absorption in InGaAs QDs as it can be seen from the photoluminescence spectrum which corresponds to a Fabry-Pérot resonance. Both spectra are shown in Fig. 1(a) .

 

Fig. 1 (a) Small-signal reflectivity and photoluminescence spectra of the unbiased p-i-n junction QD-SESAM structure. Representative curves of the test laser performance at 1020 nm: (b) output spectrum of the light source; (c) intensity autocorrelation trace.

Download Full Size | PDF

The DC electric-field-enhanced absorption recovery dynamics of the p-i-n junction SESAM device was investigated in the standard degenerate pump-probe configuration with orthogonally linearly polarized pump (TM) and probe (TE) fields. A broadly tunable light source based on a passively mode-locked Er-doped fibre laser and a highly nonlinear dispersion-shifted fibre [25] provided optical pulses of about 44 fs duration at full-width at half maximum (FWHM) centred at a wavelength of 1.06 µm with 35-nm spectral width at a repetition rate of 30 MHz. Although the output spectrum showed typical spectral fringes of a supercontinuum, the optical pulses were only about 20% over the Fourier-limited time-bandwidth product determined under assumption of a Gaussian intensity profile. An example of the laser source performance at 1.02 µm is depicted in Fig. 1(b), 1(c). Both the pump and probe beams were at normal incidence to the sample surface. These were focused to 1/e² spot diameter of ~4 µm on the structure. The nonlinear reflectivity of the p-i-n structure was characterized in the same experimental setup and wavelength region, with the time delay fixed. The pump and the probe pulses were overlapped in time. The pump beam was then attenuated while the probe signal was monitored. Balanced detection and dual beam lock-in technique were used to increase sensitivity and reduce cross-talk with the light from the pump beam. The absolute reflectivity was calibrated with a golden mirror. All measurements were conducted at room temperature.

3. Results and discussion

Fast absorption recovery of saturable absorber is a determinative factor in the ML pulse formation mechanisms. It is especially relevant to ML of semiconductor lasers where significant dynamic gain saturation takes place. At the material level, fast carrier dynamics is classically achieved by increasing the non-radiative recombination rate through the introduction of defects by various means. Standard methods for the controlled incorporation of defects and trap states are ion implantation [26] and low temperature (LT) MBE [27]. In this respect, QD-based material structures demonstrate inherently fast recovery with evidence of fast and slow time components [28].

The absorption recovery dynamics of unbiased p-i-n junction QD-SESAM structure at a pump fluence of ~140 µJ cm⁻² is shown in Fig. 2(a) . The kinetics of absorption bleaching exhibits two distinct time constants with a fast component (τ_F) in the vicinity of 0.95 ps followed by a slower one (τ_S) of 146 ps. The fast component was attributed to the intraband transitions dominated by the hole activation [29] and fast Auger-assisted relaxation [30] processes. The slow component was associated with spontaneous carrier recombination and carrier escape [31].

 

Fig. 2 (a) Absorption recovery dynamics of p-i-n QD- and QW- SESAM structures measured with voltage 0 V applied on the p-i-n junction (pump-probe pulse duration and pump fluence were 44 fs and 140 µJ cm⁻² respectively); (b) pump-fluence-dependent absorption recovery dynamics of unbiased QD-SESAM device.

Download Full Size | PDF

Further the observed absorption recovery dynamics in the QD-based device was compared with one obtained in a resonant LT InGaAs/GaAs p-i-n junction QW-SESAM structure [19] under identical experimental conditions. In Fig. 2(a), a fast temporal decay of ~0.21 ps accompanied by a slow transient of 200 ps was measured. This was attributed to intraband thermalization and carrier trapping respectively [5].

The characteristic recovery time behaviour of as grown QD-based structures is attractive from a ML viewpoint. The slow recovery time component reduces the saturation fluence of the absorber and facilitates a self-starting operation whereas the fast component plays a primary role in the evolution and shaping of the laser pulses.

The dynamic response of the QD structure at different pump fluencies in the range of 26.3 - 1030 µJ cm⁻² is shown in Fig. 2(b). At low fluencies the dynamic behaviour of absorption recovery remains unchanged with characteristic bi-exponential decay. At higher fluencies, the structure has more complex dynamic response, exhibiting second rise in the reflectivity with a time delay of 12 ps. This fluence-dependent behaviour was previously attributed to the carrier generation in the GaAs wetting layers due to two-photon absorption (TPA) [32]. Since TPA is a bias independent process [33], strong suppression of the second peak observed in our experiments at high reverse bias levels (above 10 V) may suggest that some other processes contribute to the observed peak.

The dynamic behaviour of the p-i-n junction QD-SESAM at applied reverse bias voltages ranging from 0 V to 20 V is presented in Fig. 3 and is summarised in Table 1 . It can be seen the absorption recovery responses follow a double exponential decay and both the fast and the slow transients are bias-dependent.

 

Fig. 3 Absorption recovery dynamics of the p-i-n junction QD-SESAM: (a) bias-dependent for applied reverse bias voltages ranging from 0 V to 20 V and a constant pump fluence of 100 µJ cm⁻², (b) dependence of slow recovery time component τ_S on applied reverse bias: solid line is an exponential fit to the experimental data, inset is the exponential fit at high applied fields.

Download Full Size | PDF

Table 1. Summary of the p-i-n Junction QD-SESAM Absorption Recovery Responses at Reverse Bias Voltage Ranging from 0 V to 20V

View Table

The slow recovery component as a function of applied voltage is shown in Fig. 3(b). The measured recovery time decreases by two orders of magnitude from 74.3 ps to 0.52 ps as the reverse bias voltage increases. The observed bias-dependent bleaching dynamics of slow time component can be classified into two groups depending on the reverse bias level. In the low voltage range (up to 10 V) there is a monotonic decrease in the τ_S time component due to the reduced emission time across the junction and shorter transit times across the undoped region. This can be described with an exponential function written as τ_S = τ_S0 exp(-V/V₀), where τ_S0 corresponds to 74.3 ps measured in the unbiased structure, V₀ is a build-in potential drop across the p-i-n junction of −1.9 V. This dependence is indicative of thermionic emission. At higher applied fields (beyond 10 V), the exponential fit is no longer applicable; suggesting significant contribution from additional carrier escape mechanisms, see Fig. 3(b) inset. Previously, this was attributed to the onset of tunnelling processes [20]. The bias dependence of the fast recovery time component is summarised in Table 1. The recovery time of, as short as, 210 fs was observed at a reverse bias of 10 V. It has been suggested earlier that heating of the carriers by the field may result in the contribution of carrier-carrier and carrier-phonon scattering to the bias-dependent fast transient behaviour [34]. Quantitative estimation of the fast recovery component above a reverse bias level of 10 V was found to be difficult due to possible contribution of the coherent artefact [35].

Other important nonlinear optical parameters of a saturable absorber structure are its saturation fluence (F_sat), modulation depth (ΔR) and nonsaturable losses (R_ns). Lower saturation fluence is desirable for suppression of Q-switching instabilities, low-threshold mode-locked operation and stable mode-locking at high pulse repetition rate when intracavity pulse energy is substantially reduced. In case of SESAM device, the value of saturation fluence depends not only on semiconductor material itself but also on the SESAM structure design and can be typically varied in the range of several tens to hundreds of µJ cm⁻². The nonsaturable losses presented in the absorber, when fully saturated, do not directly affect the ultra-short pulse formation and generation, but can significantly degrade the efficiency of the mode-locked laser. Moreover, high level of nonsaturable losses in absorber structure operating in high power laser systems can lead to thermal excursions and subsequent breakdown effects [5].

The results of nonlinear reflectivity measurements performed in the p-i-n junction QD- and QW-based SESAM structures are shown in Fig. 4 . The dependency of the reflectivity on the incident energy fluence was analysed using equations derived in [36]. The saturation fluence of the p-i-n QD device was estimated to be 9.6 µJ cm⁻². The value of saturation fluence of the p-i-n QW structure was found to be three times higher, of ~26.8 µJ cm⁻², for the same resonant microcavity design. This is due to the three-dimensional confinement in QDs, where the density of states is ideally compressed into a delta function, which in turn results in strongly reduced saturation fluence compared to QWs. These results confirm suitability of the QD-based saturable absorber devises for passive mode-locking of solid-state bulk and semiconductor disk lasers, and in particular for high repetition rate systems. It should be noted however that high level of nonsaturable losses of ~17% demonstrated in the QD structure will be detrimental for most application. Further investigation into the source of these losses is required.

 

Fig. 4 Nonlinear reflectivity of the p-i-n junction (a) QD- and (b) QW-based SESAM structures at 0 V reverse bias. Solid lines are fits to the data.

Download Full Size | PDF

The bias dependent saturation fluence and modulation depth are depicted in Fig. 5 . The nonlinear reflectivity for applied reverse bias of 0 V and 15 V is presented in Fig. 4(a) and 5(a). The observed rollovers are due to the TPA in the SESAM structure. It can also be seen that modulation depth of the p-i-n junction QD-SESAM structure is bias independent, at a level that can be resolved in our experimental setup. This is in stark contrast to the previously reported results obtained in the p-i-n QW-based SESAM device [19]. The fivefold reduction in modulation depth for the applied field in the range of 0 - 2 V was demonstrated due to the strong QCSE. In QD-based devices, however, the QCSE is expected to be reduced due to the smaller potential drop across the dot and difficulty of perturbing more strongly confined states [21]. The values of modulation depth were estimated to be in the range of 1.4 - 1.5% for the reverse bias range of 0 - 10 V, see Fig. 5(b).

 

Fig. 5 Nonlinear reflectivity of the p-i-n junction QD-SESAM structure: (a) representative fluence dependent reflectivity of the structure at 15 V reverse bias, solid line is the best fit; (b) dependence of the modulation depth at applied reverse bias voltage ranging from 0 V to 10 V.

Download Full Size | PDF

4. Conclusions

It has been shown that QD-based SESAM structures can serve as a fast, low-saturation fluence and low-loss passive modulator for efficient femtosecond pulse generation from solid-state laser systems, emitting in the near infrared spectral range. Moreover, such devices offer more accessible manipulation of their macroparameters for optimised and controllable mode-locking operation.

To date p-i-n junction SESAMs were only used for ML solid-state bulk and fiber lasers. We are currently exploring the true potential of these devices in semiconductor disk laser configuration. Such an all semiconductor based compact system will hugely benefit from electrically controllable pulse duration. However, reduction of nonsaturable losses will still remain as one of the main challenges. Further investigation of the electroabsorption in the QD-based SESAM structures is needed to unambiguously determine physical mechanisms responsible for the observed carrier dynamics.