Wide bandwidth chaotic signal generation in a monolithically integrated semiconductor laser via optical injection

Optics Communications 355 (2015) 551–557

Contents lists available at ScienceDirect

Optics Communications journal homepage: www.elsevier.com/locate/optcom

Wide bandwidth chaotic signal generation in a monolithically integrated semiconductor laser via optical injection Xue-Mei Yin a, Zhu-Qiang Zhong a, Ling-Juan Zhao b, Dan Lu b, Hai-Ying Qiu a, Guang-Qiong Xia a,c,n, Zheng-Mao Wu a,nn a

School of Physical Science and Technology, Southwest University, Chongqing 400715, China Key Laboratory of Semiconductor Materials Science, Institute of Semiconductors, Chinese Academy of Science, Beijing 100083, China c State Key Lab of Millimeter Waves, Southeast University, Nanjing 210096, China b

art ic l e i nf o

a b s t r a c t

Article history: Received 25 April 2015 Received in revised form 11 June 2015 Accepted 12 July 2015 Available online 18 July 2015

Wide bandwidth chaotic signal generation in a three-section monolithically integrated semiconductor laser (MISL) under external optical injection is investigated experimentally. Through evaluating the effective bandwidth of chaotic signals, the influences of the optical injection on the bandwidth of chaotic signal from the MISL are analyzed. The experimental results indicate that, for the currents of the DFB section (IDFB) and the phase section (IP) are fixed at 70.00 mA and 34.00 mA, respectively, the effective bandwidth of chaos signal generated by the solitary MISL reaches its maximum value of 14.36 GHz when the current of the amplification section (IA) takes 23.22 mA. After an external optical injection is introduced into the MISL, the effective bandwidth of the generated chaotic signal can be beyond 2.5 times of the maximum value. Furthermore, the effects of the injection strength and the frequency detuning on the effective bandwidth of the generated chaotic signal are also discussed. & 2015 Elsevier B.V. All rights reserved.

Keywords: Monolithically integrated semiconductor laser (MISL) Wide bandwidth chaotic signal Optical injection

1. Introduction Through introducing external disturbances such as optical feedback [1,2], optical injection [3,4], optoelectronic feedback [5] or current modulation [6], a chaotic system based on semiconductor lasers (SLs) can be established for realizing chaotic signal output. The chaotic signals generated in such SLs-based chaotic systems draw considerable attention due to its unique features and promising applications in secure communications [7,8], fast random-bit generation [9–11], lidar [12,13], etc. In early related researches, most of the SLs-based chaotic systems are usually constituted by discrete components. As one kind of photonic integrated circuits (PICs), since the 1990s, the monolithically integrated semiconductor lasers (MISLs) have attracted intensive attention due to their unique virtues such as smaller size, lower cost, more stable, and better reproducibility for mass production [14,15]. Through specific design and manufacture, MISLs can output different dynamical states [16–21] and then be applied to chaos synchronization and communication [22–24], high-speed physical random bit generation [25–27] and high-quality microwave generation [28]. For example, Chen et al. fabricated a n Corresponding author at: School of Physical Science and Technology, Southwest University, Chongqing 400715, China. nn Corresponding author. E-mail addresses: [email protected] (G.-Q. Xia), [email protected] (Z.-M. Wu).

http://dx.doi.org/10.1016/j.optcom.2015.07.028 0030-4018/& 2015 Elsevier B.V. All rights reserved.

colliding-pulse mode-locking (CPM) laser on a single chip and achieved subpicosecond transform-limited optical pulses [16]. Sartorius et al. demonstrated that the self-pulsation can be electrically switched on and off by adjusting the current of the phase section [17]. Franck et al. mapped the regions of mode-locking, self-pulsation, and continuous waveform (CW) lasing in the parameter space of gain current, absorber bias and RF frequency [18]. Wu et al. reported a three-section MISL, by which the chaotic signal with significant dimension and complexity can be generated [20]. Argyris et al. designed and fabricated a novel four-section MISL to be used as a compact potential chaotic emitter in optical communication [22]. Harayama et al. demonstrated that the chaotic signal generated by MISL can be used for acquiring fast nondeterministic random-bit [26]. Balakier et al. reported that a MISL can be applied to generate high-purity mm-wave signal [28]. It is well known that the use of optical injection technique can greatly broaden the application areas of SLs. The technique is originally developed to lock the frequency and stabilize the oscillation of an optically injected laser. From a viewpoint of laser dynamics, an optical injection means the introduction of an external degree of freedom to the SL. Therefore, various dynamics of SLs can be observed by optical injection, including stable and unstable injection locking, instabilities and chaos, and four-wave mixing [29]. As for the SLs-based chaos applications, the bandwidth of chaotic signal is a key indicator. In order to overcome the bandwidth restriction by the relaxation oscillation frequency of

X.-M. Yin et al. / Optics Communications 355 (2015) 551–557

the active region in SLs, various optical injection schemes are proposed to enhance the bandwidth of the chaotic signal [30–34]. For example, Uchida et al. reported that high-frequency broadband chaos signal can be obtained by using chaotic optical injection [30]. Wang et al. demonstrated that continuous-wave optical injection into a chaotic laser can enhance the bandwidth of chaotic signal [31]. Xiang et al. designated that the bandwidth of chaotic signal can be further enlarged by introducing dual-path chaotic injection from a single laser [32] or dual chaotic injections from two external lasers [33]. Hong et al. investigated the bandwidth enhancement of chaotic signal in vertical-cavity surface-emitting semiconductor lasers with optical injection [34]. However, to our knowledge, most of relevant investigations on the chaos bandwidth enhancement focus on those SLs-based chaotic systems composed of discrete components, and pay little attention on a MISL chaos system. Similarly, due to the restriction by the relaxation oscillation frequency of the DFB section in a MISL, the bandwidth of the chaotic signal generated by a solitary MISL is usually at a level of 10 GHz [22,27], which sometimes may not meet the requirements in some special applications such as beyond 10 Gb/s high-speed optical chaotic communication and ultrahigh-speed random number generation. Based on this consideration, in this work, we emphatically investigate the impacts of external optical injection on the chaotic bandwidth. The results show that, via optical injection, the effective bandwidth of chaotic signal generated by the MISL can be improved from 14.36 GHz to 36.99 GHz under certain operation parameters.

4

ESA TIL

EDFA1 VA1

Power (mW)

1

1543.5

20

40 60 IDFB (mA)

PM

1543.1 100

FC2 10/90

AL3 OI2

AL1

OI1

MISL

BS B AL2 DFB

10/90

80

Fig. 2 shows the variations of the oscillation wavelength and output power of the solitary MISL with the current IDFB of the DFB section for both IP and IA are fixed at 0 mA. From this diagram, it can be seen that the threshold current Ith is about 26.60 mA. If IDFB is biased below Ith, the output power of the MISL is very small and the recorded optical spectrum is a noise spectrum of spontaneous

50/50 PC

1543.9

3. Results and discussion

FC3 EDFA2

FC1

2

3724B), and the currents of three sections are labeled as IDFB, IP and IA, respectively. Throughout the experiment, the temperature is maintained at 18.54 °C. The output of the MISL passes through AL2, BS, OI2 and AL3, and then is divided into two parts by FC2. One part with relatively low power is sent to an optical spectrum analyzer (OSA, Ando AQ6317C) for detecting the optical spectra of the MISL output. The other part with relatively high power is amplified by EDFA2 firstly and then divided into two parts, which are transferred into two electronic signals by two fast photo-detectors (U2T-XPDV3120R, 70 GHz bandwidth), respectively. One electronic signal is sent to an oscilloscope (OSC, Agilent DSOX91604A with 16 GHz bandwidth) for observing the time series of the MISL output, and the other is sent to an electrical spectrum analyzer (ESA, R&Ss FSW, 67 GHz bandwidth) for analyzing the power spectra of the MISL output signal. It should be pointed out that, in this work, the time series recorded by the OSC are for the output of the MISL after DC components have been blocked.

PD1 VA2 PD2 VA3

1544.3

Fig. 2. Measured variations of output power and lasing wavelength of a solitary MISL with the DFB section current IDFB under IP ¼ 0 mA and IA ¼ 0 mA.

The experimental setup is shown in Fig. 1. In this study, the used three-section MISL is designed and fabricated by ourselves, which is composed of a distributed feedback (DFB) section, a phase (P) section and an amplification (A) section with lengths of 220 μm, 240 μm and 320 μm, respectively. The output of a tunable injection laser (TIL, Ando AQ4321A), whose wavelength ranges from 1480 nm to 1580 nm with a linewidth of 200 kHz, firstly passes through an erbium-doped fiber amplifier (EDFA1, Corning PureGainTM 2500C) and a variable attenuator (VA1), and then is separated into two branches by a fiber coupler (FC1). One branch with low power (10% of the total power) is sent to a power meter (PM) for monitoring the injection power. The other branch with high power (90% of the total power) passes through a polarization controller (PC), two aspheric lenses (AL1, AL2), an optical isolator (OI1) and a beam splitter (BS), and then injects into a MISL. The temperature and current of each section of the MISL are controlled by high-accuracy laser diode controller (ILX-Lightwave, LDC-

OSC

3

00

2. Experimental setup

OSA

1544.7 power wavelength Wavelength (nm)

552

P

IDFB IP

A IA

Fig. 1. Experimental setup. TIL: tunable injection laser; MISL: monolithically integrated semiconductor laser; DFB: distributed feedback section; P: phase section; A: amplification section; IDFB: DFB section current; IP: phase section current; IA: amplification section current; EDFA: erbium-doped fiber amplifier; VA: variable attenuator; FC: fiber coupler; PM: power meter; PC: polarization controller; AL: aspheric lens; OI: optical isolator; BS: beam splitter; PD: photo-detector; ESA: electrical spectrum analyzer; OSC: oscilloscope; OSA: optical spectrum analyzer; solid line: optical path; dashes line: electrical path; B: the point for monitoring the output power of the MISL and the injection optical power.

0.1

-10

-50

0

-50

-80 -20

-0.1 0.1

-90 -10

-50

0

-50

-80 -20 -50 -80 -20

Power (dBm)

-20

Amplitude (V)

Power(dBm)

X.-M. Yin et al. / Optics Communications 355 (2015) 551–557

-0.1 0.1 0 -0.1 0.1

-80 -20

-0.1 0.1

-90 -10

-50

0

-50

MP

C

-90 0 10 20 30 40 50 60

Frequency (GHz)

Time (ns)

Wavelength (nm)

P2

-90 -10 -50

2

P1

-50

0

1

S

-90 -10

-50

-0.1 -80 0 1543.5 1544.0 1544.5 1545.0

553

Fig. 3. Optical spectra (first column), time series (second column), and power spectra (third column) for some typical states in a solitary MISL under IDFB ¼ 70.00 mA and IP ¼35 mA for different IA. (a) S: stable state, IA ¼0 mA, (b) P1: period-one oscillation, IA ¼ 17.78 mA, (c) P2: period-two oscillation, IA ¼ 19.75 mA, (d) MP: multi-period oscillation, IA ¼ 21.65 mA, and (e) C: chaotic state, IA ¼ 22.80 mA. The gray lines in power spectra denote the noise floor.

38

15

S P1 P2

36 IP (mA)

C

34

32

30 20

22

24 IA (mA)

26

Bandwidth (GHz)

MP

12

9

28

Fig. 4. Mapping of dynamical states in the parameter space of IP and IA under IDFB ¼ 70.00 mA. S: stable state; P1: period-one oscillation; P2: period-two oscillation; MP: multi-period oscillation; and C: chaotic state.

emission. Once IDFB reaches the threshold current Ith, the output power of the MISL jumps to a relatively large value suddenly, meanwhile the optical spectrum behaves as a typical single mode shape and thus the oscillation wavelength can be read out accurately. With the increase of IDFB from Ith, both the output power and the lasing wavelength show an upward tendency. In the following discussion, IDFB will be fixed at 70.00 mA. Under this circumstance, the output power and lasing wavelength are about 2.06 mW and 1544.070 nm, respectively. Since the amplification section of the three-section MISL is used to adjust the strength of the feedback light, its bias current will inevitably influence the dynamical states of the MISL. We fix the current of the phase section Ip at 35 mA and adjust the current of the amplifier section IA, and then diverse dynamical states can be observed as shown in Fig. 3. For IA ¼ 0 mA (the first row), the optical spectrum is a typical single mode shape, the output power

6 21.5

22.5

23.5 24.5 IA (mA)

25.5

26.5

Fig. 5. Variations of chaotic bandwidth with IA for IDFB ¼ 70.00 mA and IP ¼ 34.00 mA.

is nearly a constant with small ripples due to the noise, and the power spectrum is similar to the noise floor. As a result, the dynamical state of the solitary MISL can be determined to be a stable state (S). For IA ¼17.78 mA (the second row), some peaks with identical interval emerge in the optical spectrum, the time series shows a periodic oscillation whose fundamental frequency is close to the relaxation oscillation frequency and is about 13.577 GHz obtained from the power spectrum. Under these circumstances, it can be judged that the dynamical state of the MISL is a period-one oscillation (P1). For IA ¼19.75 mA (the third row), much more peaks appear in the optical spectrum, both the sub-harmonic frequency (about 6.747 GHz) and fundamental frequency (about 13.511 GHz) present clearly, which are typical characteristics of

X.-M. Yin et al. / Optics Communications 355 (2015) 551–557

Power (dBm)

-20 -50

1544.1 Wavelength (nm)

1544.8 Amplitude (V)

-80 1543.4 0.06 0 -0.06 0 -30

1 Time (ns)

2

BW=14.36GHz

-65 -100 0 10 20 30 40 50 60 70 Frequency (GHz)

Power (dBm)

Power (dBm)

Amplitude (V)

Power (dBm)

554

-20 -50 -80 1543.4

1544.1 Wavelength (nm)

1544.8

1 Time (ns)

2

0.06 0 -0.06 0 -30

BW=36.99GHz

-65 -100 0 10 20 30 40 50 60 70 Frequency (GHz)

Fig. 6. Optical spectra (first row), time series (second row) and power spectra (third row), of the MISL for IDFB ¼ 70.00 mA, IP ¼34.00 mA and IA ¼ 23.22 mA, where the left and right columns are respectively for the solitary MISL and the MISL under optical injection with injection ratio Kin ¼ 2.88 and frequency detuning Δf¼ 60.00 GHz. In power spectra, the gray, green and red lines denote the noise floor, the measured power spectra, and the power spectra after subtracting the background noise, respectively. The shaded areas denote the spectral spans that have been counted toward the effective bandwidth. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

0.05

-30

-50

0

-65

-80 -20

-0.05

-100 -30

Kin=0.32

-80 Kin=2.59

-50 -80 -20 -50

Kin=4.21

Power (dBm)

Kin=1.30

-50

-20

0.05

Amplitude (V)

Power (dBm)

-20

0 -0.05 0.05

-100 -30 -65

-0.05

-100

0.05

-30

0

-65 1 Time (ns)

2

BW=24.17GHz

-65

0

-80 -0.05 1543.5 1543.9 1544.3 1544.7 0 Wavelength (nm)

BW=17.61GHz

BW=29.39GHz

BW=31.97GHz

-100 0 10 20 30 40 50 60 70 Frequency (GHz)

Fig. 7. Optical spectra, time series, and power spectra under Δf ¼45.00 GHz for Kin ¼ 0.32 (first row), 1.30 (second row), 2.59 (third row) and 4.21 (fourth row), respectively, where the meanings of different color lines and shaded areas in power spectra are the same as those in Fig. 6. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

X.-M. Yin et al. / Optics Communications 355 (2015) 551–557

40

Bandwidth (GHz)

35 30 Δf = -45.00 GHz Δf = 45.00 GHz

25 20 15 10 0

1

2 3 Injection ratio

4

5

Fig. 8. Effective bandwidth as a function of the injection ratio Kin under Δf¼  45.00 GHz and Δf¼ 45.00 GHz.

doubled periodicity, and then the dynamics could be identified as a period-two oscillation (P2). For IA ¼ 21.65 mA (the fourth row), some new frequencies emerge around the fundamental frequency and sub-harmonic frequency, which is the feature of a multi-period oscillation (MP). For IA ¼22.80 mA (the fifth row), the time series fluctuates dramatically, and meanwhile the corresponding power spectrum continuously covers a very broad frequency range. All these features indicate that the solitary MISL operates at a chaotic state (C). In a word, a trace of S–P1–P2–MP–C is presented under IDFB ¼ 70.00 mA and IP ¼35.00 mA through scanning I A. The above results show that the dynamical states of the MISL are seriously influenced by IA, and then it can be predicted that the variation of IP also affects the dynamical states of the MISL since the variation of IA will results in the change of the feedback phase. Fig. 4 depicts a mapping of dynamic states in the parameter space of IP and IA under IDFB ¼70.00 mA. As shown in this diagram, through regulating IP and IA, some typical dynamical states such as stable state (S), period-one oscillation (P1), period-two oscillation (P2), multi-period oscillation (MP) and chaotic state (C) can be

observed. In this work, we mainly concern the chaotic region. By carefully observing this region, one can find that for IP ¼34.00 mA, IA possesses a widest variation range of about 21.90–25.90 mA to realize chaotic output. Usually, both effective bandwidth and standard bandwidth can be adopted to define the bandwidth of chaotic signals. In this work, we adopt the effective bandwidth to characterize the bandwidth of chaotic signal from the MISL. The algorithm for the effective bandwidth is as follows: first, the whole power spectrum recorded by the ESA is sorted in a descending order; next, the sorted power values are summed up until the accumulation power arrives at 80% of the total power; and finally, the effective bandwidth can be obtained through summing those cumulated discrete spectral segments [35]. Here, we fix IDFB and IP at 70.00 mA and 34.00 mA, respectively, and analyze the variation of the bandwidth of chaotic signal with IA, which is shown in Fig. 5. With the increase of IA from 21.90 mA to 25.90 mA, the effective bandwidth increases firstly, after reaches a maximum of 14.36 GHz for IA ¼ 23.22 mA, and then decreases. Our related experimental results show that 14.36 GHz is approximately the maximal effective bandwidth no matter how IP is changed. Therefore, we fix IDFB, IP, and IA at 70.00 mA, 34.00 mA, and 23.22 mA, respectively, and then investigate how the introduction of optical injection affects the effective bandwidth. Fig. 6 displays the optical spectra (first row), time series (second row) and power spectra (third row) of the MISL for IDFB ¼70.00 mA, IP ¼ 34.00 mA and IA ¼23.22 mA, where the left and right columns are for the solitary MISL and the MISL under optical injection with injection ratio Kin ¼2.88 and frequency detuning Δf¼ 60.00 GHz, respectively. Here, Kin is defined as the ratio of the injection optical power to the output power of the solitary MISL, where both the injection optical power and the output power of the solitary MISL are measured at point B in Fig. 1, and the frequency detuning Δf is defined as the difference between the lasing frequency of the injection light and the lasing frequency of the solitary MISL under IDFB ¼70.00 mA, IP ¼34.00 mA and IA ¼0 mA. From this diagram, it can be observed that, with or without the optical injection, the MISL output can be identified to be a chaotic state (C) by carefully analyzing the optical spectrum, the power spectrum and the time series under the two cases. Moreover, by comparing Fig. 6 (a3) with (b3), one can see that the

-20

-20 BW=33.21GHz

BW=41.61GHz

-60

-100 0 10 20 30 40 50 60 70 -20 BW=0.10GHz

-60

-100 0 10 20 30 40 50 60 70 Frequency (GHz)

Power (dBm)

-60

Power (dBm)

555

-100 0 10 20 30 40 50 60 70 -20 BW=0.10GHz

-60

-100 0 10 20 30 40 50 60 70 Frequency (GHz)

Fig. 9. Power spectra under Kin ¼2.88 for (a) Δf¼  40.00 GHz, (b) Δf¼  35.00 GHz, (c) Δf ¼  25.00 GHz, and (d) Δf ¼  10.00 GHz, respectively, where the meanings of different color lines and shaded areas in power spectra are the same as those in Fig. 6. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

X.-M. Yin et al. / Optics Communications 355 (2015) 551–557

Bandwidth (GHz)

556

50

50

50

40

40

40

30

30

30

20

20

20

10

10

10

0 0 0 -80-60-40-20 0 20 40 60 80 -80-60-40-20 0 20 40 60 80 -80-60-40-20 0 20 40 60 80 Frequency detuning (GHz) Fig. 10. Variations of the effective bandwidth with Δf under (a) Kin ¼ 1.92, (b) Kin ¼ 2.88, and (c) Kin ¼ 3.85. The filled circle, hollow square, hollow rhombus, snowflake, and solid triangle represent chaotic state (C), stable state (S), multi-period oscillation (MP), period-two oscillation (P2) and period-one oscillation (P1), respectively.

power spectrum of the MISL with optical injection is relatively more broad and smooth than that without optical injection, and the effective bandwidth is increased from 14.36 GHz to 36.99 GHz due to the introduction of optical injection. In the following, we will analyze the impacts of the injection ratio and the frequency detuning on the effective bandwidth of chaotic signal. Firstly, we fix the frequency detuning at 45.00 GHz and examine the influence of the injection strength. The optical spectra, time series, and power spectra under different Kin are showed in Fig. 7. From top to bottom, the injection ratios are 0.32, 1.30, 2.59, and 4.21, respectively. From this diagram, it can be seen that, under above given parameters, all the output dynamical states of the MISL are the chaotic state (C). For a relatively large value of Kin, the amplitude located at the frequency of injection light in optical spectra is also large, and the region contributing to effective bandwidth expands, which leads to a large value of effective bandwidth. Fig. 8 further gives the effective bandwidth of chaotic signal from the MISL as a function of injection ratio Kin for Δf ¼  45.00 GHz and Δf ¼45.00 GHz. A similar trend can be observed for Δf ¼ 45.00 GHz and Δf¼ 45.00 GHz. With the increase of Kin, the effective bandwidth rapidly increases firstly, and then slowly increases to a relative stable value. It should be pointed out that the above rules are only suitable for the case that the Kin varies within 0–5.00, which is restricted by the experimental conditions. Certainly, for a larger value of Kin, the optical injection may result in a change of dynamical state, and then the variation trend of the bandwidth with the injection ratio will be much more complicated. Next, we fix the injection ratio Kin and adjust the frequency detuning Δf to explore the impact of the frequency detuning on the effective bandwidth. Fig. 9 depicts the power spectra of the MISL under optical injection with Kin ¼2.88 and different Δf. As shown in Fig. 9(a), for Δf ¼  40.00 GHz, the power spectrum is broad and its effective bandwidth reaches 33.21 GHz, which demonstrate that the dynamical state of the MISL is chaotic state (C). For Δf¼  35.00 GHz (as shown in Fig. 9(b)), the power spectrum is broad (corresponding effective bandwidth is about 41.61 GHz) but is close to the background of noise spectrum, and then the MISL operates at stable state (S). For Δf¼  25.00 GHz (as shown in Fig. 9(c)), some peaks emerge in the power spectrum, and the main peak locates at about 7.89 GHz. Within the main peak, we can see one visible peak. This indicates that the MISL operates at period-two oscillation (P2) and the effective bandwidth is about 0.10 GHz. For Δf ¼  10.00 GHz (as shown in Fig. 9(d)), the dynamical state of the MISL is period-one oscillation (P1), whose fundamental frequency is about 17.83 GHz. Under this case, the effective bandwidth is about 0.10 GHz, and 80% of the total power is from the main peak. Finally, Fig. 10 shows the effective bandwidth of output signal

from the MISL as a function of Δf under Kin ¼1.92 (a), 2.88 (b) and 3.85 (c), respectively. Here, different marks denote different dynamical states. From this diagram, it can be seen that, within the region that the values of |Δf | are relative small, the dynamical states are very sensitive to the variation of the frequency detuning and the effective bandwidths change dramatically. However, for | Δf | possesses relative large values, the unbroken expanse of chaotic regions can be obtained. For different values of Kin, the continuous chaotic regions own different boundaries, and the maximums of the effective bandwidths are 36.56 GHz under Kin ¼1.92, 36.99 GHz under Kin ¼2.88, and 36.86 GHz under Kin ¼3.85, respectively.

4. Conclusions In summary, the bandwidth enhancement of chaotic output of a monolithically integrated semiconductor laser (MISL) is investigated experimentally via an external optical injection technique. The used MISL during this experiment consists of three sections, namely distributed feedback (DFB), phase (P) and amplification (A) sections, where the latter two mainly are used to adjust the strength and phase of optical feedback. Under certain bias currents of the three sections, the solitary MISL can output a chaotic signal with effective bandwidth of 14.36 GHz. After introducing external optical injection, a chaos signal with effective bandwidth of  36.00 GHz can be obtained through selecting suitable injection strength and frequency detuning. It should be pointed out that this study mainly concerns some special MISLsbased chaotic applications where chaos signals with dozens of GHz bandwidth is required. From a practical point of view, the use of optical injection inevitably partly weakens the compactness of MISLs, though this shortcoming may be gradually improved with the development of PICs. Additionally, the control of the coupling coefficient between the injection light and MISLs is also a difficult point, which need to be paid special attention in practice.

Acknowledgments This work was supported by the National Natural Science Foundation of China under Grant nos. 61178011, 61275116 and 61475127, the Open Fund of the State Key Lab of Millimeter Waves of China under Grant no. K201418, and the Postgraduate Research and Innovation Project of Chongqing Municipality under Grant no. CYB14054. References [1] J. Mork, B. Tromborg, J. Mark, IEEE J. Quantum Electron. 28 (1992) 93. [2] R. Hegger, M.J. Bünner, H. Kantz, A. Giaquinta, Phys. Rev. Lett. 81 (1998) 558.

X.-M. Yin et al. / Optics Communications 355 (2015) 551–557

[3] V. Annovazzi-Lodi, S. Donati, M. Manna, IEEE J. Quantum Electron. 30 (1994) 1537. [4] T.B. Simpson, J.M. Liu, A. Gavrielides, V. Kovanis, P.M. Alsing, Phys. Rev. A 51 (1995) 4181. [5] F.Y. Lin, J.M. Liu, IEEE J. Quantum Electron. 39 (2003) 562. [6] S.T. Kingni, J.H. Talla Mbe, P. Woafo, Eur. Phys. J. Plus 127 (2012) 46. [7] A. Sanchez-Diaz, C.R. Mirasso, P. Colet, P. Garcia-Fernandez, IEEE J. Quantum Electron. 35 (1999) 292. [8] A. Argyris, D. Syvridis, L. Larger, V. Annovazzi-Lodi, P. Colet, I. Fischer, J. GarciaOjalvo, C.R. Mirasso, L. Pesquera, K.A. Shore, Nature 438 (2005) 343. [9] A. Uchida, K. Amano, M. Inoue, K. Hirano, S. Naito, H. Someya, I. Oowada, T. Kurashige, M. Shiki, S. Yoshimori, K. Yoshimura, P. Davis, Nat. Photonics 2 (2008) 728. [10] K. Hirano, T. Yamazaki, S. Morikatsu, H. Okumura, H. Aida, A. Uchida, S. Yoshimori, K. Yoshimura, T. Harayama, P. Davis, Opt. Express 18 (2010) 5512. [11] X. Tang, Z.M. Wu, J.G. Wu, T. Deng, L. Fan, Z.Q. Zhong, J.J. Chen, G.Q. Xia, Laser Phys. Lett. 12 (2015) 015003. [12] F.Y. Lin, J.M. Liu, IEEE J. Sel. Top. Quantum Electron. 10 (2004) 991. [13] W.T. Wu, Y.H. Liao, F.Y. Lin, Opt. Express 18 (2010) 26155. [14] T.L. Koch, U. Koren, IEEE J. Quantum Electron. 27 (1991) 641. [15] S. Charbonneau, E.S. Koteles, P.J. Poole, J.J. He, G.C. Aers, J. Haysom, M. Buchanan, Y. Feng, A. Delage, F. Yang, M. Davies, R.D. Goldberg, P.G. Piva, I. V. Mitchell, IEEE J. Sel. Top. Quantum Electron. 4 (1998) 772. [16] Y.K. Chen, M.C. Wu, IEEE J. Quantum Electron. 28 (1992) 2176. [17] B. Sartorius, M. Mohrle, S. Reichenbacher, W. Ebert, IEEE Photonics Technol. Lett. 7 (1995) 1261. [18] T. Franck, S.D. Brorson, A. Moller-Larsen, J.M. Nielsen, J. Mork, IEEE Photonics Technol. Lett. 8 (1996) 40. [19] M. Yousefi, Y. Barbarin, S. Beri, E.A.J.M. Bente, M.K. Smit, R. Nötzel, D. Lenstra,

557

Phys. Rev. Lett. 98 (2007) 044101. [20] J.G. Wu, L.J. Zhao, Z.M. Wu, D. Lu, X. Tang, Z.Q. Zhong, G.Q. Xia, Opt. Express 21 (2013) 23358. [21] D. Liu, C.Z. Sun, B. Xiong, Y. Luo, Opt. Express 22 (2014) 5614. [22] A. Argyris, M. Hamacher, K.E. Chlouverakis, A. Bogris, D. Syvridis, Phys. Rev. Lett. 100 (2008) 194101. [23] T. Pérez, M. Radziunas, H.J. Wünsche, C.R. Mirasso, F. Henneberger, IEEE Photonics Technol. Lett. 18 (2006) 2135. [24] A. Argyris, E. Grivas, M. Hamacher, A. Bogris, D. Syvridis, Opt. Express 18 (2010) 5188. [25] A. Argyris, S. Deligiannidis, E. Pikasis, A. Bogris, D. Syvridis, Opt. Express 18 (2010) 18763. [26] T. Harayama, S. Sunada, K. Yoshimura, P. Davis, K. Tsuzuki, A. Uchida, Phys. Rev. A 83 (2011) 031803. [27] R. Takahashi, Y. Akizawa, A. Uchida, T. Harayama, K. Tsuzuki, S. Sunada, K. Arai, K. Yoshimura, P. Davis, Opt. Express 22 (2014) 11727. [28] K. Balakier, M.J. Fice, F. van Dijk, G. Kervella, G. Carpintero, A.J. Seeds, C. C. Renaud, Opt. Express 22 (2014) 29404. [29] J. Ohtsubo, Semiconductor Lasers: Stability, Instability and Chaos, third ed., Springer, Berlin, 2013. [30] A. Uchida, T. Heil, Y. Liu, P. Davis, T. Aida, IEEE J. Quantum Electron. 39 (2003) 1462. [31] A.B. Wang, Y.C. Wang, H.C. He, IEEE Photonics Technol. Lett. 20 (2008) 1633. [32] S.Y. Xiang, W. Pan, N.Q. Li, B. Luo, L.S. Yan, X.H. Zou, L.Y. Zhang, P.H. Mu, IEEE Photonics Technol. Lett. 24 (2012) 1753. [33] S.Y. Xiang, W. Pan, B. Luo, L.S. Yan, X.H. Zou, N.Q. Li, H.N. Zhu, IEEE J. Quantum Electron. 48 (2012) 1069. [34] Y. Hong, P.S. Spencer, K.A. Shore, J. Opt. Soc. Am. B: Opt. Phys. 29 (2012) 415. [35] F.Y. Lin, Y.K. Chao, T.C. Wu, IEEE J. Quantum Electron. 48 (2012) 1010.