A.

Concept and components

The mode-locked diode laser system which is specifically designed for micro-integration is configured as an extended-cavity diode laser (ML-ECDL) setup, depicted in Fig. 1. The ML-ECDL consists of an 1 mm long AlGaAs double quantum well ridge-waveguide laser diode, collimation optics, and a plane dielectric mirror. The two-section diode chip features a 100 μm long saturable absorber and a 900 μm long gain section. The front facet of the diode located at the absorber section is coated to 10 % reflection while the resonator-side facet is anti-reflection (AR) coated. Collimation of the diode’s output on both sides is realized by aspheric micro-lenses. Additional intra-cavity beam correction is realized with a cylindrical lens. The optical resonator formed by the front facet of the diode laser and the nearly zero group velocity dispersion (GVD) dielectric mirror is approx. 40 mm long. For the measurements a dc current is applied to the gain section. Fundamental passive mode-locking is realized by reverse biasing the saturable absorber section of the diode.

Fig. 1

Schematic of the mode-locked diode laser system. The optical resonator is formed by the front facet of the diode and the dielectric mirror.

B.

Mode-locking performance

For the results presented here the gain section dc injection current (DCI) was set to 260.00 mA and the saturable absorber bias (SAB) to about -1.16 V. The mount temperature (TMount) was stabilized to the room temperature, i.e. T_Mount is 20.00 °C.

To analyze the optical and RF spectral characteristics the light is coupled into single-mode optical fibers after passing through an optical isolator (Qioptic (Gsänger) DLI-I).

We characterize the RF performance of the mode-locked pulse train using a combination of a fast photodiode (NewFocus 1434), a broadband RF amplifier (Centellax TA0L30VA) and an electrical spectrum analyzer (R&S FSV-30). Fig. 2 shows a typical RF spectrum of our ML-ECDL in its fundamental mode-locking regime. The pulse repetition rate is approx. 4.2 GHz corresponding to the resonator length of the ML-ECDL. A signal-to-noise-ratio (SNR) of more than 60 dB (resolution bandwidth (RBW) = 3 MHz) can be observed for the fundamental RF peak, and still more than 50 dB for its seventh harmonic, indicating high quality mode-locking.

Fig. 2.

(a) Normalized (norm.) RF spectrum of the mode-locked pulses up to 30 GHz. (b) Zoom into the RF spectrum near the pulse repetition frequency f_rep.

A RF linewidth of the pulse repetition rate at full-width-at-half-maximum (FWHM) of about 80 Hz (RBW = 50 Hz) is reached (limited by the analyzer’s resolution bandwidth and low frequency jitter), as shown in Fig. 2(b). Measurement of the frequency noise power spectral density (PSD), as depicted in Fig. 3, allows for a more detailed analysis of the mode-locking stability. Frequency noise PSD spectra are determined for a wide bandwidth, from very low noise frequencies (1 Hz) all the way up to 3 GHz. This is accomplished by combining phase noise data derived from a time series measurement with the IQ (in-phase/quadrature) tool of the RF analyzer [5], which provides access to noise frequencies up to 12.8 MHz, and from the RF phase noise data measured with the phase noise measurement tool (PN tool; option FSV-K40 for R&S FSV-30), which provides access to noise frequencies above 12 MHz. At low noise frequencies (below 1 kHz) increased noise can be observed, which may be attributed to residual thermal and mechanical instabilities of the setup. The white noise floor of the fundamental beat corresponds to about 56 Hz²/Hz (IQ data, 500 ms integration time) while the corresponding RMS integrated timing jitter amounts to ~ 2 ps (PN tool, bandwidth 20 kHz to 1 MHz).

Fig. 3

RF frequency noise PSD versus noise frequency for pulse repetition rate. Data is derived from frequency noise PSD measurements based on IQ de-modulation of the beat note signal [5] (bandwidth 1 Hz to 80 kHz and 100 Hz to 12.8 MHz) and from phase noise measurements carried out with the ESA PN tool (Option FSV-K40 for R&S FSV-30) (bandwidth 10 kHz to 1 MHz). At noise frequencies above 1 MHz the measurement is limited by the analyzer.

The optical characteristics are analyzed using an optical spectrum analyzer (Yokogawa AQ6373). Fig. 4 shows the optical spectrum corresponding to the RF characteristics described above. The mode-locked pulses feature a peak wavelength of approx. 788 nm, an optical bandwidth exceeding 2.5 nm at -3 dB and 7 nm at -20 dB.

Fig. 4.

Optical spectrum

Pulse shape and width of the mode-locked pulses are evaluated by measuring their intensity autocorrelation function (IACF). The IACF corresponding to the data shown above is depicted in Fig. 5. Assuming a Lorentzian pulse shape of the ICAF a FWHM pulse width of 5.5 ps is reached.

Fig. 5.

Intensity autocorrelation function

From the optical and temporal characteristics a time-bandwidth-product of approx. 7 can be calculated, which indicates strongly chirped mode-locked pulses well above the Fourier-limit. Moreover the mode-locked pulses feature an average optical power of about 90 mW, measured with a calibrated photodetector. This corresponds to a peak power of approx. 2.5 W. Hence optimal pulse compression would allow for pulses as short as about 170 fs with a peak power of more than 80 W.