Experimental Study of the Timing Jitter of a Passively ... - IEEE Xplore

IEEE JOURNAL OF QUANTUM ELECTRONICS, VOL. 51, NO. 4, APRIL 2015

1300107

Experimental Study of the Timing Jitter of a Passively Mode-Locked External-Cavity Semiconductor Laser Subject to Repetition Rate Transitions and Optical Feedback Simon Rauch, Lukas Drzewietzki, Andreas Klehr, Member, IEEE, Joachim Sacher, Member, IEEE, Wolfgang Elsäßer, Senior Member, IEEE, and Stefan Breuer, Member, IEEE

Abstract— We experimentally investigate the timing jitter (TJ) of a passively mode-locked external-cavity diode laser. Variation of the gain current and the absorber reverse bias voltage allows transitions from fundamental mode-locking up to seventh harmonic mode-locking. Hereby, a reduction of the TJ as a function of the harmonic mode-locking order is found. Furthermore, the application of optical feedback results in an additional reduction of TJ for almost the whole investigated operation range. In particular, the reduction increases with harmonic mode-locking order. The highest observed reduction of TJ amounts to a factor of 10 as compared with the freerunning case which corresponds to a repetition-rate linewidth reduction by a factor of 100. Index Terms— Laser mode-locking, timing jitter, optical feedback.

I. I NTRODUCTION

M

ODE-LOCKED lasers allow the generation of femtosecond pulses and cover sophisticated applications including ultrafast optical sampling and frequency comb generation [1]. These applications require high repetition-rates (RRs) along with a low timing jitter (TJ), which corresponds to a narrow RR linewidth [2]. Fiber Manuscript received December 22, 2014; revised February 2, 2015; accepted February 5, 2015. Date of publication February 13, 2015; date of current version February 27, 2015. This work was supported in part by the Bundesministerium für Bildung und Forschung through the Federal Ministry of Education and Research under Grant 13N9819, in part by the State of Hessen within LOEWE Project entitled Kurzpulsdiodenlaser für den Einsatz in THz-Systemen, HA Project under Grant 401/13-40. The work of S. Breuer was supported by the Adolf-Messer-Foundation. S. Rauch, L. Drzewietzki, and S. Breuer are with the Institute of Applied Physics, Technische Universität Darmstadt, Darmstadt 64289, Germany (e-mail: [email protected]; lukas.drzewietzki@physik. tu-darmstadt.de; [email protected]). A. Klehr is with the Ferdinand-Braun-Institut, Berlin 12489, Germany (e-mail: [email protected]). J. Sacher is with Sacher Lasertechnik, Marburg 35037, Germany (e-mail: [email protected]). W. Elsäßer is with the Institute of Applied Physics, Technische Universität Darmstadt, Darmstadt 64289, Germany, and also with the Center of Smart Interfaces, Technische Universität Darmstadt, Darmstadt 64289, Germany (e-mail: [email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/JQE.2015.2402431

lasers can exhibit ultra-low TJ in the attosecond regime [3]. Yet, they are still comparatively costly and bulky and have RRs well below 1 GHz. In contrast, passively mode-locked (PML) semiconductor lasers (SCLs) offer the benefit of compactness and simplicity and allow the generation of sub-picosecond optical pulses at repetition rates in the multi-GHz range. However, the TJ of SCLs is in general higher as compared to fiber lasers. One common method to improve the stability of PML SCLs is applying resonant optical feedback (OFB). It has been studied that this kind of OFB reduces variances in the pulse amplitude [4] and width [5] and the pulse timing jitter [6] compared to the non-resonant case. Experimentally, in [7] a reduction of the RR linewidth for both the exited and the ground state of a PML quantumdot (QD) SCL was demonstrated for different OFB strenghts. In [8], TJ stabilization by short-delay OFB was reported for a quantum-well (QW) based monolithic SCL with an RR of 50 GHz yielding an RR linewidth reduction from 5.4 MHz to 56 kHz. Using a fiber delay line with a long-delay length of 18 m reduced the RR linewidth of a monolithic QD based PML SCL at an RR of 5.25 GHz from 46.2 kHz to 1.1 kHz [9]. Moreover, applying dual OFB to a monolithic QW PML SCL resulted in a decrease of the RR linewidth down to 192 Hz [10]. Applying long-delay OFB to a QD based PML SCL, Breuer et. al. reported a TJ reduction by a factor of 3.5 [11]. It was demonstrated that PML external-cavity diode lasers (ECDLs) offer lower values of TJ as compared to monolithic PML SCLs [12], [13] and in [14], Ippen et. al. demonstrated that OFB even allows the reduction of the TJ of a hybridly mode-locked external-cavity diode laser (ECDL) with a fundamental RR of 9 GHz by a factor of 3.4. External-cavity mode-locking (ML) at RRs of around 300 MHz has been reported in [13] and [15]–[17] and below 100 MHz and sub-threshold ML in [18] and [19]. For such long-cavity ECDLs higher harmonics of the fundamental RR can be achieved by variation of the absorber reverse bias voltage (RBV) and the gain current (GC) [20].

0018-9197 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

1300107


Fig. 1. Schematic of the experimental setup. The two-section diode laser chip is incorporated in an external cavity configuration and the output coupler is a fiber Bragg grating. The optical feedback (OFB) is implemented by the auxiliary fiber cavity (AFC) composed of fibers and a mirror. The delay length amounts to 18 m. Precise mirror displacement allows for fine-tuning of the delay. The OFB ratio can be varied by adjusting the fiber coupling at the end of the AFC.

For a hybridly mode-locked ECDL it was reported that the corner frequency of the timing-phase-noise power-spectraldensity (TPN PSD) decreases with increasing cavity length while keeping the radio-frequency (RF) modulation frequency and thus the RR fixed [21]. This implies a lower RF linewidth and thus also a lower TJ for harmonically PML ECDLs as compared to a fundamentally PML SCLs operating at the same RR [22]. Based on these principal advantages, in this paper we experimentally study the TJ in connection with the harmonic operation of a PML ECDL as a function of the RBV and the GC. The observed RR transitions are explained by Haus’ theory of mode-locking with a slow saturable absorber [23]. Furthermore, the effect of resonant OFB on TJ by a long auxiliary fiber cavity (AFC) is investigated as a function of the RBV, the GC and the OFB ratio. The influence of OFB on the TPN PSD is compared with predictions of a simple model developed for fundamentally PML SCLs. II. E XPERIMENTAL S ETUP A schematic of the experimental setup employed to study the TJ of the harmonically PML SCL embedded in an external cavity is shown in Fig. 1. The external cavity is formed by a fiber Bragg grating. Its peak reflection amounts to 10 %, is centered at 835 nm and the width of the spectral-reflection is 1 nm. The length of the external cavity corresponds to a fundamental RR of f fund = 300 MHz and the optical length of the long-delay AFC amounts to 18 m. The semiconductor laser chip under study is a two-section ridge waveguide laser on an AlN sub-mount. The active layer consists of two InGaAsP quantum wells that are separated by a GaAsP spacer layer. The surrounding waveguide is made up of 1.7 μm-thick Al0.45 Ga0.55 As layers that are coated by Al0.7 Ga0.3 As. Furthermore, the waveguide and coating layers

are p- and n-doped, respectively. This layer configuration results in a center wavelength of the gain spectrum of 840 nm. The total chip length is 1200 μm and the absorber length amounts to 80 μm. The rear facet is high-reflection coated whereas the output front facet is anti-reflection coated with a residual reflection of Rout ≈ 10−4 . Thus, the employed diode chip does not exhibit a distinct lasing threshold without an external reflector. The recovery time of the device’s absorber section is assumed to be similar to values reported for a resembling QW structure [24]. There, the absorber recovery time is in the range of several ten picoseconds and exhibits an exponential dependence on the RBV. Throughout all measurements the device is temperature stabilized at 20 °C. The TJ of a PML laser is commonly specified either by the root-mean-square (RMS) TJ within a selectable frequency bandwidth or by the pulse-to-pulse timing jitter (TJPTP ) [2]. Both can be evaluated from the TPN PSD which can be obtained from the power spectrum of an optical pulse-train [25]. Hereby, the RMS TJ is obtained by integration of the TPN PSD and the TJPTP can be calculated from the width of the TPN PSD or the width of the RR [2]. The TJPTP represents the RMS value of the temporal deviation of two consecutive pulses as compared to the mean pulse-repetition period and is given in by: 3 ))0.5 . σptp = ( f /(2π f rep

(1)

Hereby, σptp is the TJPTP , f is the RR linewidth and f rep is the RR. This equation not only describes the timing deviations on a time scale of ∝ 1/ f rep , but also represents long-term timing fluctuations. That means, on the time scale of some tens of thousands times the round trip time the√RMS timing √ deviation σrms can be expressed by σrms = σptp · n = σLT · n with n being the number of pulses, due to the well-known random-walk nature of the TJ. This long-term timing jitter (TJLT ), represented by σLT , is consequently evaluated from the RR linewidth by Eq. 1. In particular, this long-term timing jitter (TJLT ) is also valid for the case of a PML SCL subject to OFB [6] and thus represents an ideal measure of TJ for the investigations presented here. To evaluate the TJLT it is sufficient to estimate f rep and f . Therefore, a Lorentzian fit to the RR signal is performed to obtain an accurate value for f . This approach is justified since no signs for significant amplitude noise could be found in the vicinity of 0 Hz in the experiments. III. R ESULTS AND D ISCUSSION Initially, the free-running PML ECDL is investigated to demonstrate the transitions of the harmonic mode-locking order (HMLO) and to provide a reference for subsequent OFB experiments. Thus, Fig. 2 shows the TJLT as a function of the GC and the RBV without OFB. Furthermore, information on the present RR is indicated by the corresponding HMLO N according to f rep = N · ffund with ffund = 300 MHz. The black lines mark the transitions between different regimes of HMLO. The grey area in the lower right corner indicates that the laser is operated beneath its threshold and no ML can be achieved. For the investigated areas of ML operation the pulse width ranges from 6 to 8 ps.

RAUCH et al.: EXPERIMENTAL STUDY OF THE TJ OF A PML EXTERNAL-CAVITY SCL

Fig. 2. The long-term timing jitter as a function of the gain current and the absorber reverse bias voltage without optical feedback. N indicates the harmonic mode-locking order.

Fig. 3. Calculated unsaturated gain g0 as a function of the harmonic modelocking order according to [20, eq. 9] with TRT / Tg = 0.2. A rise in g0 and in the boundaries of gi represent an increasing gain current and absorber reverse bias voltage, respectively. The arrows indicate transitions of N if the gain current is varied. The inset exemplarily shows the optical output power for a forward and reverse sweep of the GC and the corresponding HMLOs.

It can be observed from Fig. 3 that the HMLO increases consistently with increasing GC as well as with decreasing RBV. At the lowest applied RBV of 0.25 V and highest current values the highest HMLO of 7 can be observed. Furthermore it can be observed that the transitions between different HMLOs are accompanied by discontinuous transitions of the TJLT as well. In general, the TJLT decreases with increasing HMLO. Within a HMLO region the TJLT is nearly constant. A staircase-like pattern marking transitions of the HMLO and the TJLT can be observed. This structure arises due to the artificial discretization for the GC of 3.5 mA steps and for the RBV where 0.25 V steps are used. This discretization is introduced to account for the observation that a transition from one HMLO to another is not abrupt. The transition area is rather accompanied with a competition and thus fluctuations between two RRs. The RR transitions between different HMLOs N observed in the TJ analysis for the free-running configuration confirm to a large extent the predictions of Haus’ theory of ML with a slow saturable absorber [23]. There, the appearance

1300107

of higher HMLOs N with respect to the fundamental RR is explained by the unsaturated gain g0 . It is shown in [20] that g0 can represent the GC and depends on N. Furthermore it can be concluded from [23] that for a fixed N g0 is bound by limitations of its initial value gi , which depends on the small signal loss of the absorber section q0 , which in turn can represent the RBV. Thus, there is only a certain range of the GC and RBV available for a fixed N. This means that the number of pulses has to change when these constraints are transcended. This is illustrated in Fig. 3. There, the valid calculated unsaturated gain g0 is displayed for a given HMLO N. In this plot g0 is evaluated for three different ranges of gi , where the blue curve indicates the highest and the black one the lowest RBV. The choice of the corresponding numerical values is derived from values given in [23] to represent three successive integer multiples of q0 . The arrows in Fig. 3 indicate transitions of N if the GC is varied. Most importantly, it is evident from Fig. 3 that the laser is expected to switch from fundamental ML to harmonic modelocking (HML) if the GC is increased. This can be confirmed for all absorber reverse voltages in Fig. 2. Moreover, Fig. 2 reveals the trend that the onset of the transitions to higher harmonics N shifts towards lower GCs with decreasing RBV. From Fig. 3 it can be concluded, that this effect occurs due to a reduction of q0 with decreasing RBV. This also explains why the highest HMLOs can be found at high GCs and low RBVs as evident in Fig. 2. Additionally, in Fig. 3 it can be recognized that the range of g0 for a fixed N broadens if the RBV is increased. This trend is also observed for the N = 2 branch in Fig. 2. The shown hysteresis in Fig. 3, indicated by the same values of g0 for different HMLOs, is studied by the means of an exemplary light-current curve at a RBV of 2.8 V, which is shown in the inset of Fig. 3. There, the average optical output power is recorded for a forward and reverse sweep of the GC. The difference in power between the two scans is below 0.1 mW und hence only marginal. Both curves show rising slopes within the region of one HMLO present and exhibit small steps at the transition to another HMLO. Thus only a weak multistability, as reported in [19], is observed here. Passive ML starts shortly after the laser threshold, which is at 120 mA for a RBV of 2.8 V. As the output facet of the laser is anti-reflection coated its threshold is depending on the feedback by the fiber Bragg grating and on the biasing conditions. The lowest investigated RR amounts to 300 MHz. Indeed, in [26] a RR of 191 MHz was reported for a QD PML ECDL and in [27] passive ML at a RR of 126 MHz is demonstrated for a laser chip similar to ours. Having demonstrated the trends of the HMLOs and the correlation with the TJLT , here, the effect of the OFB ratio on the TJ or the TPN PSD is investigated. The aim is to identify the general optimum condition of OFB for the highest reduction of TJ. Hereby, the GC is kept at a fixed value of 154.5 mA and the RBV is kept at a fixed value of 1.75 V. Here, the TPN PSD is evaluated at an offset frequency of 120 kHz which represents an easily accessible measure of TJ [6]. The fine-delay is always optimized towards highest TPN PSD reduction. The optical delay-length amounts to 18 m. Fig. 4 shows the measured TPN PSD at 120 kHz as a function

1300107


Fig. 4. Experimentally obtained timing-phase-noise power-spectral-density at a frequency offset of 120 kHz and corresponding power law fit as a function of the optical feedback ratio for a fixed gain current of 154.5 mA and a fixed absorber reverse bias voltage of 1.75 V. The green and blue straight lines indicate the two power functions of the fit function.

Fig. 5. The long-term timing jitter as a function of the gain current and the absorber reverse bias voltage subject to optical feedback. N denotes the harmonic mode-locking order.

of OFB ratio, which is defined as the ratio of feedback power to output power. To be able to evaluate the feedback ratio, the output power has been mapped for the whole operation regime before the TJ measurement. It can be observed that with increasing OFB ratio the TPN PSD is continuously reduced. No saturation behavior with increasing OFB ratio is evident, which suggests to maximize the OFB ratio in the experiment. However, for decreasing OFB ratio a threshold is indicated by the first shown data point and amounts to 1.85·10−4 . For an OFB ratio below this threshold the laser exhibits amplitude instabilities so that no meaningful data could be recorded. Indeed, in [11] it was mentioned that a reduction of the TJ arises only if a certain value of OFB is exceeded. For an ECDL setup in [14], it was reported that for an AFC length of about 17 m an OFB ratio of 0.01 was required to allow for reduction of TJ. A possible explanation for this deviation of the OFB ratio might be that the reflectivity of the facet could be different, which is important in terms of the optical power which actually enters the waveguide. Furthermore, a power law fit function of the form L ϕ (κ) = 10 log (s1 · κ a + s2 · κ b ), which represents the TPN PSD L ϕ as a function of the OFB ratio κ, is fitted to the data. The second summand dominates for high values of κ and the power amounts to b = −0.93. This result is in agreement with results reported in [6] for a monolithic PML SCL. Thus, the behavior of a harmonically PML ECDL subject to OFB does not differ from the behavior of a monolithic PML SCL, showing the generality of the effect of OFB. It is also found in experiments, that this threshold increases with decreasing OFB delay length and in addition that the TPN PSD reduction is lower. Thus, the longest possible length of the OFB is desirable. These results confirm the trend that longer AFCs and higher OFB ratio have a higher potential to reduce the TJ as has been observed for monolithic PML SCL in [11] and [28]. Having identified the need for the highest possible ratio and length of the OFB, in the following OFB will be applied for the whole operation range of the laser given by Fig. 2. Hereby, the

OFB ratio and the fine-delay are adjusted to obtain the highest TJ reduction. Under the influence of OFB the pulse width still amounts to 6-8 ps and experiences only slight changes of 1.5 ps compared with the free-running case. This indicates that the AFC length equals the external-cavity length very precisely [5]. Thus, Fig. 5 shows the optimized TJLT as a function of the GC and the RBV under OFB. The staircase-like structure similar to Fig. 2 can be recognized here as well. The HMLO regions of 1 and 2 are hardly changed by the OFB. But interestingly the OFB can increase the HMLO for many driving conditions for example from 3 to 4 around an RBV of 3.00 V and a GC of 180 mA. The OFB can also decrease the HMLO for example from 2 to 1 at an RBV of 3.75 V and a GC of 144 mA. This could be explained by a change of the equivalent gain induced by OFB which is expected to affect the calculated results shown in Fig. 3 and thus also the HMLOs. Besides these changes of HMLOs when comparing Fig. 2 and Fig. 5, it is obvious that TJLT is reduced almost in the whole operation range by application of OFB. In both figures the TJLT decreases with increasing GC and decreasing RBV. In order to quantify this qualitative observation ensemble-averages of the TJLT are evaluated for each ensemble of HMLO. Thus, Fig. 6 shows the average TJLT as a function of HMLO for the free-running laser and the laser subject to OFB as well as the corresponding power-law fits. It can be seen that for the free-running laser the average TJLT steadily reduces with increasing HMLO. Hereby a powerlaw fit of the form σLT = a · N b yields a value of bFR = −1.78. It is evident that for the laser subject to OFB the TJLT also steadily reduces with increasing HMLO. Hereby a power-law fit yields a significantly higher value of bFB = −2.34. A comparison of both powers yields a difference of 0.5 which demonstrates that the mean TJ reduction by OFB increases with increasing HMLO. Besides this statement, no explanation can be given regarding the absolute values of both powers, which is explained in the following. First, TJ depends on pulse-width, intra-cavity power and round-trip


Fig. 6. Averaged long-term timing jitter σLT with and without applied optical feedback and the corresponding power-law fits as a function of harmonic mode-locking order.

Fig. 7. Experimentally obtained and simulated timing-phase-noise powerspectral-density for the passively mode-locked external-cavity diode laser operating at a harmonic mode-locking order of 3 with and without applied OFB. The OFB ratio amounts to the highest applied value of 8 · 10−4 and the delay length amounts to 18 m.

gain [29] which change during a HMLO transition. Second, the noise characteristics of HML are not trivial and strongly depend on pulse-to-pulse correlations or the coupling of all the pulses within the cavity [30]. Third, it is not known if an increase of the HMLO at a fixed physical AFC length represents an increase of the equivalent length of the OFB. For a higher length of the OFB in general leads to an increased TJ reduction. Having described the global trends of the TJLT and the HMLOs for the free-running laser and subject to OFB now the highest found TPN PSD and TJLT reduction is reported. Thus, Fig. 7 shows the TPN PSD of the free-running laser and the laser subject to OFB at a GC of 154.5 mA and an RBV of 1.75 V. The OFB ratio amounts to the highest applied value of 8 · 10−4 . At this driving condition the laser operates on the third harmonic RR amounting to frep = 900 MHz and the measured pulse-length amounts to a value of 6 ps. Starting at a frequency of 30 kHz the TPN PSD for the freerunning laser exhibits a proportionality of 1/ f 2 as indicated by the slope of −20 dB and results from the well-known random-walk TJ characteristics. This trend continues up to

1300107

30 MHz towards the noise floor of the electrical instruments at −112.2 dBc/Hz. A broad peak is evident at a frequency of 50 MHz which is attributed to the relaxation oscillation of the ECDL. Under free-running operation the TJLT amounts to 2 ps. When OFB is applied, the TPN PSD also exhibits the 1/ f 2 proportionality for low frequencies. Additionally the typical noise resonance peaks at multiples of the AFC frequency arise [6]. We also observe supermode-noise peaks in the TPN PSD at multiples of the RR of 300 MHz if the HMLO is higher than 1 (not displayed). Nevertheless the noise peaks at multiples of the OFB frequency of 8 MHz, shown in Fig. 7, are not supermode noise. They are different noise resonances stemmming from the external cavity and their contribution to the integrated TJ is negligible. Furthermore, the TPN PSD magnitude is by 20 dB lower as compared to the free-running case. This indicates a significant reduction of the TJLT by a factor of 10, which is equivalent to a reduction of the RR linewidth by a factor of 100 leading to a TJLT of 0.2 ps. Moreover, simulations basing on the model reported in [6] have been performed to investigate if the reported TJ reduction mechanism also applies for HML operation of ECDLs subject to OFB and the results are also shown in Fig. 7. Here, the RR and the TJPTP are used to calculate the TPN PSD of the free running laser. To simulate the TPN PSD for the laser subject to OFB also the pulse-width, the OFB ratio and the delay length of the AFC are used. Based on the given parameters the simulated TPN PSD for the free-running laser reproduces the experimental results with good agreement. The deviations above 3 MHz are attributed to the fact that the model does not incorporate amplitude noise phenomena. With applied OFB the 1/ f 2 dependence is reproduced well. The values predicted by the model are only lower by about 1 dB at 0.03–1 MHz. The typical noise resonance peaks stemming from the AFC are reproduced with good agreement with respect to frequency and shape. The decrease of the peak amplitudes with increasing frequency is also reproduced but the amplitudes of the noise peaks in the simulation are lower as compared to the amplitudes in the experiment. This indicates an amplification of the timing noise stored within the AFC in the experiment. The frequency of the AFC amounts to f aux = 8.33 MHz and deviates from the frequency of the experimentally obtained noise resonance peak amounting to 7.98 MHz. Also the corresponding frequency obtained in simulations deviates towards lower frequencies and amounts to 7.8 MHz. Thus, due to the correspondence of these results in experiment and simulation the deviation from f aux is expected to be intrinsic. Finally, this overall correspondence shows that the TJ behavior of a harmonically PML ECDL subject to OFB can be described within the same framework as for monolithic PML SCLs. This leads to the conclusion that all the pulses within the ECDL are highly correlated or locked. This locking of subsequent pulses can be explained by the periodically occurring net-gain window of the gain segment within the ECDL at the RR. IV. C ONCLUSION We experimentally studied the long-term timing jitter (TJLT ) and the harmonic mode-locking order (HMLO) transitions of

1300107


a harmonically passively mode-locked (PML) external-cavity diode laser (ECDL). The HMLO and thus the repetitionrate (RR) could be tuned up to the 7t h harmonic of the fundamental RR by variation of the gain current (GC) and the absorber reverse bias voltage (RBV). The corresponding findings are in good agreement with theoretical predictions by Haus. Hereby, it was found that the TJLT decreases with increasing HMLO and thus with increasing number of pulses within the ECDL. Furthermore, we confirmed that the TJLT can be reduced by applying long-delay optical feedback (OFB). Hereby it was found that the TJLT reduction also increases with increasing HMLO. In such a way, the timing-phase-noise power-spectral-density (TPN PSD) could be decreased at maximum by 20 dB up to an offset frequency of 1MHz which is equivalent to a RR linewidth reduction by a factor of 100. The agreement between the experimentally obtained and the simulated TPN PSDs led to the conclusion that the observed timing jitter (TJ) reduction originates from the same mechanism identified for monolithic fundamentally mode-locked semiconductor laser (SCL). R EFERENCES [1] U. Keller, “Recent developments in compact ultrafast lasers,” Nature, vol. 424, pp. 831–838, Aug. 2003. [2] F. Kefelian, S. O’Donoghue, M. T. Todaro, J. G. McInerney, and G. Huyet, “RF linewidth in monolithic passively mode-locked semiconductor laser,” IEEE Photon. Technol. Lett., vol. 20, no. 16, pp. 1405–1407, Aug. 15, 2008. [3] Y. Song, C. Kim, K. Jung, H. Kim, and J. Kim, “Timing jitter optimization of mode-locked Yb-fiber lasers toward the attosecond regime,” Opt. Exp., vol. 19, no. 15, pp. 14518–14525, 2011. [4] E. A. Avrutin and B. M. Russell, “Dynamics and spectra of monolithic mode-locked laser diodes under external optical feedback,” IEEE J. Quantum Electron., vol. 45, no. 11, pp. 1456–1464, Nov. 2009. [5] H. Simos, C. Simos, C. Mesaritakis, and D. Syvridis, “Two-section quantum-dot mode-locked lasers under optical feedback: Pulse broadening and harmonic operation,” IEEE J. Quantum Electron., vol. 48, no. 7, pp. 872–877, Jul. 2012. [6] L. Drzewietzki, S. Breuer, and W. Elsäßer, “Timing jitter reduction of passively mode-locked semiconductor lasers by self- and externalinjection: Numerical description and experiments,” Opt. Exp., vol. 21, no. 13, pp. 16142–16161, 2013. [7] C. Mesaritakis et al., “Effect of optical feedback to the ground and excited state emission of a passively mode locked quantum dot laser,” Appl. Phys. Lett., vol. 97, no. 6, pp. 061114-1–061114-3, Aug. 2010. [8] O. Solgaard and K. Y. Lau, “Optical feedback stabilization of the intensity oscillations in ultrahigh-frequency passively modelocked monolithic quantum-well lasers,” IEEE Photon. Technol. Lett., vol. 5, no. 11, pp. 1264–1267, Nov. 1993. [9] C.-Y. Lin, F. Grillot, Y. Li, R. Raghunathan, and L. F. Lester, “Microwave characterization and stabilization of timing jitter in a quantum-dot passively mode-locked laser via external optical feedback,” IEEE J. Sel. Topics Quantum Electron., vol. 17, no. 5, pp. 1311–1317, Sep./Oct. 2011. [10] M. Haji et al., “High frequency optoelectronic oscillators based on the optical feedback of semiconductor mode-locked laser diodes,” Opt. Exp., vol. 20, no. 3, pp. 3268–3274, 2012. [11] S. Breuer et al., “Investigations of repetition rate stability of a modelocked quantum dot semiconductor laser in an auxiliary optical fiber cavity,” IEEE J. Quantum Electron., vol. 46, no. 2, pp. 150–157, Feb. 2010. [12] D. J. Derickson, P. A. Morton, J. E. Bowers, and R. L. Thornton, “Comparison of timing jitter in external and monolithic cavity modelocked semiconductor lasers,” Appl. Phys. Lett., vol. 59, no. 26, pp. 3372–3374, 1991. [13] M. A. Cataluna, Y. Ding, D. I. Nikitichev, K. A. Fedorova, and E. U. Rafailov, “High-power versatile picosecond pulse generation from mode-locked quantum-dot laser diodes,” IEEE J. Sel. Topics Quantum Electron., vol. 17, no. 5, pp. 1302–1310, Sep./Oct. 2011.

[14] L. A. Jiang, K. S. Abedin, M. E. Grein, and E. P. Ippen, “Timing jitter reduction in modelocked semiconductor lasers with photon seeding,” Appl. Phys. Lett., vol. 80, no. 10, pp. 1707–1709, 2002. [15] M. G. Thompson, A. R. Rae, M. Xia, R. V. Penty, and I. H. White, “InGaAs quantum-dot mode-locked laser diodes,” IEEE J. Sel. Topics Quantum Electron., vol. 15, no. 3, pp. 661–672, May/Jun. 2009. [16] M. Xia, M. G. Thompson, R. V. Penty, and I. H. White, “External-cavity mode-locked quantum-dot laser diodes for low repetition rate, subpicosecond pulse generation,” IEEE J. Sel. Topics Quantum Electron., vol. 17, no. 5, pp. 1264–1271, Sep./Oct. 2011. [17] J. C. Balzer, T. Schlauch, A. Klehr, G. Erbert, G. Tränkle, and M. R. Hofmann, “High peak power pulses from dispersion optimised modelocked semiconductor laser,” Electron. Lett., vol. 49, no. 13, pp. 838–839, Jun. 2013. [18] M. Marconi, J. Javaloyes, S. Balle, and M. Giudici, “How lasing localized structures evolve out of passive mode locking,” Phys. Rev. Lett., vol. 112, p. 223901, Jun. 2014. [19] M. Marconi, J. Javaloyes, S. Balle, and M. Giudici, “Passive modelocking and tilted waves in broad-area vertical-cavity surface-emitting lasers,” IEEE J. Sel. Topics Quantum Electron., vol. 21, no. 1, pp. 85–93, Jan./Feb. 2015. [20] S. Sanders, A. Yariv, J. Paslaski, J. E. Ungar, and H. A. Zarem, “Passive mode locking of a two-section multiple quantum well laser at harmonics of the cavity round-trip frequency,” Appl. Phys. Lett., vol. 58, no. 7, pp. 681–683, 1991. [21] T. Yilmaz, C. M. Depriest, A. Braun, J. H. Abeles, and P. J. Delfyett, “Noise in fundamental and harmonic modelocked semiconductor lasers: Experiments and simulations,” IEEE J. Quantum Electron., vol. 39, no. 7, pp. 838–849, Jul. 2003. [22] F. Quinlan, S. Ozharar, S. Gee, and P. J. Delfyett, “Harmonically modelocked semiconductor-based lasers as high repetition rate ultralow noise pulse train and optical frequency comb sources,” J. Opt. A, Pure Appl. Opt., vol. 11, no. 10, p. 103001, 2009. [23] H. Haus, “Theory of mode locking with a slow saturable absorber,” IEEE J. Quantum Electron., vol. 11, no. 9, pp. 736–746, Sep. 1975. [24] J. R. Karin et al., “Ultrafast dynamics in field-enhanced saturable absorbers,” Appl. Phys. Lett., vol. 64, no. 6, pp. 676–678, 1994. [25] D. von der Linde, “Characterization of the noise in continuously operating mode-locked lasers,” Appl. Phys. B, vol. 39, no. 4, pp. 201–217, 1986. [26] Y. Ding, M. A. Cataluna, D. Nikitichev, I. Krestnikov, D. Livshits, and E. Rafailov, “Broad repetition-rate tunable quantum-dot external-cavity passively mode-locked laser with extremely narrow radio frequency linewidth,” Appl. Phys. Exp., vol. 4, no. 6, p. 062703, 2011. [27] J. C. Balzer, B. Döpke, A. Klehr, G. Erbert, G. Tränkle, and M. R. Hofmann, “Femtosecond semiconductor laser system with arbitrary intracavity phase and amplitude manipulation,” Proc. SPIE, vol. 9002, pp. 90020D-1–90020D-6, Feb. 2014. [28] C. Otto, “Dynamics of quantum dot lasers subject to optical feedback and external optical injection,” Ph.D. dissertation, Technische Univ. Berlin, Inst. Theoretische Phys., Berlin, Germany, 2013. [29] R. Paschotta, “Noise of mode-locked lasers (Part II): Timing jitter and other fluctuations,” Appl. Phys. B, vol. 79, no. 2, pp. 163–173, 2004. [30] F. Rana et al., “Characterization of the noise and correlations in harmonically mode-locked lasers,” J. Opt. Soc. Amer. B, vol. 19, no. 11, pp. 2609–2621, 2002.

Simon Rauch received the master’s degree in physics from the Technische Universität Darmstadt, Darmstadt, Germany, in 2014. He pursued a master’s thesis with Sacher Lasertechnik GmbH, where he developed and optimized a highly integrated pulsed semiconductor laser source. He focused on improving the timing jitter of externalcavity pulsed semiconductor lasers with the Technische Universität Darmstadt.


Lukas Drzewietzki received the Diploma degree in physics from the Technische Universität Darmstadt, Darmstadt, Germany, in 2008, where he is currently pursuing the Ph.D. degree in physics.

Andreas Klehr was born in Berlin, Germany, in 1952. He received the Diploma degree in physics from Humboldt-University Berlin, in 1976, and the Ph.D. degree with a focus on polarization switching and bistability of 1.3μm InGaAsP/InP RW laser diodes, in 1995. From 1979 to 1991, he was with the Central Institute of Optics and Spectroscopy, Academy of Sciences of the GDR, where he was involved in semiconductor laser development. From 1991 to 1996, he was with the Max-Born-Institute, Berlin. Since 1996, he has been with the FerdinandBraun-Institut, Leibniz-Institut für Höchstfrequenztechnik, Berlin, where he was involved in high power diode lasers and laser dynamics.

Joachim Sacher (M’94) was born in Koblenz, Germany, in 1962. He received the Diploma and Dr.rer.nat. degrees in physics from Philipps Universität, Marburg, Germany, in 1988 and 1992, respectively. From 1986 to 1992, he worked on nonlinear dynamics and chaos, and picosecond mode-locking of diode lasers. In 1992, he founded Sacher Lasertechnik, where he is involved in research on tunable diode lasers and picosecond mode-locked diodes for spectroscopy applications. He is a member of the German Physical Society, the International Society for Optics and Photonics, and the Optical Society.

1300107

Wolfgang Elsäßer (M’94–SM’97) was born in Pforzheim, Germany, in 1954. He received the Diploma degree in physics from Karlsruhe Technical University, Karlsruhe, Germany, in 1980, the Ph.D. degree in physics from the University of Stuttgart, Stuttgart, Germany, in 1984, and the Habilitation degree in experimental physics from the Philipps University of Marburg, Marburg, Germany, in 1991. He was with the Max Planck Institute for Solid State Research, Stuttgart, from 1981 to 1985. From 1985 to 1995, he was with the Philipps University of Marburg. Since 1995, he has been a Professor with the Institute of Applied Physics, Technische Universität Darmstadt, Darmstadt, Germany, where he is currently the Head of the Semiconductor Optics Group. He is a member of the German Physical Society and the European Physical Society. He was a recipient of the Otto Hahn Medal in 1985, the Werner von Siemens Medal in 1985, the Rudolf Kaiser Prize in 1991, the IEE J. J. Thomson Premium in 1995, the Hassia Cooperation Award in 2004, and the Darmstadt Technology Transfer Foundation Award in 2011.

Stefan Breuer (M’09) received the Diploma degree in physics from the Technical University of Clausthal, Clausthal-Zellerfeld, Germany, in 2005, and the Ph.D. degree in semiconductor laser optics from the Technische Universität Darmstadt, Darmstadt, Germany, in 2010. He is currently a Post-Doctoral Researcher with the Technische Universität Darmstadt.