We demonstrate an electrically pumped InAs quantum dot (QD) two-section passively mode-locked laser (MLL) on a silicon substrate by low temperature (250 °C) Pd-GaAs wafer bonding technology. The saturable absorber of the QD-MLL is electrically isolated by a 15-μm wide dry-etching gap which resulted in ∼30 kΩ resistance from the gain regions of the MLL. At room temperature, the laser operates in the O-band (1.3 μm) telecommunication wavelength regime with a threshold current of 94 mA and laser bar cavity and absorber lengths of 6 mm and 300 μm, respectively. The optimum mode-locked conditions are observed under injection current and reverse bias voltage of 124 mA and −7 V, which generates pulses at a repetition rate of 7.3 GHz, an optical bandwidth of 0.97 nm, and a nearly transform limited pulse width of 2 ps (sech2 pulse profile). These results enable QD-MLLs to be integrated with silicon photonic integrated circuits, such as optical time division multiplexing and optical clocks.

Silicon photonics is one of the most promising solutions to achieve low cost CMOS-compatible photonic integrated circuits (PICs) and nearly all the components of silicon photonics interconnections have been individually demonstrated.1–5 However, a proper way to monolithically integrate robust and efficient lasers into silicon photonics platform is still under investigation. Specifically, current approaches are inefficient and not robust against temperature variations (requires T < 40–60 °C for operation).6,7 An attractive solution to these problems is the use of InAs quantum dot (QD) lasers due to their capability for high temperature operation.8–10 Recently, InAs QD lasers were monolithically grown on silicon substrates.11–13 However, these novel direct growth technologies require 4°–6° off-cut or pre-patterned silicon substrates and thick buffer layers to eliminate threading dislocations (TDs) between Si and GaAs, both of which limit CMOS compatibility. Alternatively, QD lasers on Si have been demonstrated using wafer bonding technologies where the performance is influenced by the high temperature during the bonding process.14,15 Recently, we have demonstrated a metal wafer bonding platform using Palladium (Pd) to integrate InAs QD lasers on silicon substrates with a very low bonding temperature (<250 °C).16 This metal-mediated bonding process provides an ohmic contact to Silicon and excellent thermal conductivity, resulting in p-side down bonded lasers with enhanced performance over counterparts on native GaAs substrates.16 Here, we experimentally demonstrate a mode-locked QD laser on Si at a repetition rate of 7 GHz by using this Pd wafer bonding platform.

Mode-locked lasers generate ultrashort pulses for applications in optical clock generation, optical time division multiplexing (OTDM), wavelength division multiplexing (WDM), and high-speed electro-optic sampling systems. In addition to using bulky Ti:sapphire lasers or other vibronic lasers to generate pulses of light, semiconductor mode-locked lasers (MLLs) and hybrid quantum well MLLs with pulsewidths from hundreds of femtoseconds to tens of picoseconds have been demonstrated in the near infrared, which provide opportunities for photonic integration.17–19 Compared to quantum well lasers, QD lasers are well suited for mode-locked applications due to their inhomogeneous gain spectrum and ultrafast carrier dynamics.19 Currently, ultrafast, high repetition rate, low-jitter, temperature insensitive quantum dot mode-locked lasers (QD-MLLs) have been realized.20–22 To further integrate QD-MLLs into an on-chip system, we have demonstrated an OTDM system using a QD-MLL which produces a 5 Gb/s clock signal externally coupled into a silicon photonics chip, generating a 40 Gb/s signal.23 In order to integrate the QD-MLLs on chip, we have developed hybrid QD-MLLs on Si using the Pd-GaAs bonding technology. In this report, the growth design, fabrication, and characterization of the bonded QD-MLLs will be discussed.

The structure of the hybrid QD-MLL is shown in Figure 1(a). The active region was grown by molecular beam epitaxy (MBE) on an n-type GaAs (100) substrate, containing 10 stacks of self-assembled InAs QD layers (grown by Innolume GmbH, Dortmund, Germany). The InAs quantum dot density is ∼3 × 1010 cm−2 and the photoluminescence (PL) has a linewidth of ∼28 meV. The ground state modal gain is expected to be approximately ∼21 cm−1 based on results obtained from similar Innolume epitaxy wafers.24 In this laser heterostructure, the p-type and n-type doped Al0.35Ga0.6As claddings are designed to confine the QD emission in the III–V waveguide. A 200-nm Al0.7GaAs etch stop layer is grown between the laser structure and GaAs buffer layer for epitaxy membrane transfer. The low temperature Pd-GaAs wafer bonding technology used to fabricate QD-MLLs has been previously developed to make ridge waveguide Fabry-Perot lasers.16 Additionally, this p-side down low temperature wafer bonding technology shows the potential to enhance heat dissipation properties, which lower the threshold current of bonded lasers compared to unbonded ones. The schematic of the passive QD-MLL is shown in Figure 1(b), where the MLL has a 6-mm long gain section and a 300-μm saturable absorber section separated by a 15-μm wide electric isolation gap. The QD-MLLs are fabricated with waveguide and mesa widths of 5 μm and 25 μm, respectively (Figure 2(a), inset).16 Electric isolation gaps are made together with the laser ridges by dry etching to a depth of 2 μm (Figure 2(b)). The resistance between the two laser sections is characterized by I–V measurements (Figure 2(b), inset). The measured resistance is ∼30 kΩ which allows the sections to be independently biased. After fabricating the QD-MLLs on Si, the silicon substrate is mechanically polished to a thickness of ∼100 μm before cleaving in order to realize a mirror facet that is free of striations (Figure 2(a)). Although the facets could be high reflectivity (HR) coated, the facets were left as-cleaved for this study.

FIG. 1.

(a) Heterostructure schematic and (b) diagram of the InAs QD mode-locked laser diode on silicon.

FIG. 1.

(a) Heterostructure schematic and (b) diagram of the InAs QD mode-locked laser diode on silicon.

Close modal
FIG. 2.

(a) Magnified cross-section SEM image of the laser facet with a 70° tilted angle and (inset) unmagnified SEM image of the same region. (b) Top view SEM image and I–V characteristics (inset) of a 15 μm wide dry etched electrical isolation gap between the gain and saturable absorber sections.

FIG. 2.

(a) Magnified cross-section SEM image of the laser facet with a 70° tilted angle and (inset) unmagnified SEM image of the same region. (b) Top view SEM image and I–V characteristics (inset) of a 15 μm wide dry etched electrical isolation gap between the gain and saturable absorber sections.

Close modal

The DC characteristics of bonded QD-MLLs are tested with current injection (Ig) and a constant reverse bias voltage (Va) applied to the saturable absorber section. The devices are mounted on an aluminum oxide submount using Indium-Tin solder (Indium Corporation ribbonin-10809) on top of a thermoelectric cooler (TEC) to maintain room temperature (RT) operation during measurement. The continuous wave (CW) light–current (L–I) characteristics of bonded QD-MLLs with forward and backward sweeping of the injection current under zero bias (0 V), 4 V, and 7 V reverse bias conditions are shown Figure 3. When the absorber is short circuited (Va = 0 V), the QD-MLL shows a threshold current (Ith) of 94 mA, and a single facet output power of ∼20 mW at an injection current of 240 mA. At a reverse bias of Va = −7 V, the threshold current is increased to 100 mA due to the higher cavity losses generated from the reverse-biased saturable absorber. We note that a bistability effect around threshold current region is not obvious due to the high ratio of the gain/absorber section.21,25

FIG. 3.

(a) Light current (L-I) characteristics of the bonded mode-locked laser under reverse bias voltages of 0 V, −4 V, and −7 V with forward and backward sweep current. Inset: electroluminescence (EL) of the bonded mode-locked laser with a reverse bias voltage of 0 V with an injection current of 150 mA.

FIG. 3.

(a) Light current (L-I) characteristics of the bonded mode-locked laser under reverse bias voltages of 0 V, −4 V, and −7 V with forward and backward sweep current. Inset: electroluminescence (EL) of the bonded mode-locked laser with a reverse bias voltage of 0 V with an injection current of 150 mA.

Close modal

The RF properties are characterized using a high-index single-mode fiber (Nufern UHNA7) directly coupled from the saturable absorber facet (absorber and cavity facets exhibited similar RF and optical spectra). The coupled light is then sent through an O-band semiconductor optical amplifier (SOA) (Thorlabs S9FC1132P) to boost the power intensity and is followed by an optical splitter for simultaneous characterization of the RF spectrum, optical spectrum, pulse train, and pulse width. In order to characterize the mode-locking dynamics of the bonded QD-MLL, optical and RF spectra 2D evolutions are generated from the OSA (optical spectrum analyzer) and RF spectrum analyzer measurements simultaneously. These 2D optical and RF spectra evolutions are characterized over reverse bias voltages from 0 V to −7 V with −1 V steps. Figure 4 shows the evolution of the optical and RF spectra at reverse biases of −7 V and 0 V, respectively, which are selected to indicate the typical differences of bonded MLLs functioning at the two ends of the bias conditions. The RF spectra exhibit mode-locking under high or low reverse bias conditions; however, higher reverse bias voltages result in locking over a wider range of conditions as well as a stronger RF signals due to the faster absorption recovery in the saturable absorber. It is worth noting that the laser operates at a higher RF frequency at higher reverse bias conditions since the elevated bias voltage induces more loss, essentially shortening the cavity length of the laser. In addition, under both reverse bias conditions, the bonded MLL shows a stable narrow RF spectrum around the threshold current, and the linewidth broadens as the optical spectrum transitions from single to dual band with increased injection current. This dual band phenomenon was also observed in unbonded QD mode-locked lasers, and the band splitting may be due to internal optical power, AC stark effect, or the pumping effects from coupled groups of QDs.26,27 The gap between dual spectral bands results in refractive index differences, leading to two competing groups of pulses in the cavity, which destabilizes operation. Optimum mode-locking conditions are achieved under high reverse bias with an injection current set below the bifurcation of the optical spectrum. This set of conditions usually occurs around the threshold current (Va = −7 V and Ig = 124 mA).

FIG. 4.

(a) Optical and (b) RF spectra evolutions with injection current from 100 mA to 240 mA at 0 V bias. (c) Optical and (d) RF spectra evolutions with injection current from 100 mA to 240 mA at −7 V bias. Inset: RF spectra at optimized injection (Ig = 125 mA) and at bifurcated spectrum operation (Ig = 215 mA). The RF signal degrades significantly at higher injection currents where the optical spectrum splits.

FIG. 4.

(a) Optical and (b) RF spectra evolutions with injection current from 100 mA to 240 mA at 0 V bias. (c) Optical and (d) RF spectra evolutions with injection current from 100 mA to 240 mA at −7 V bias. Inset: RF spectra at optimized injection (Ig = 125 mA) and at bifurcated spectrum operation (Ig = 215 mA). The RF signal degrades significantly at higher injection currents where the optical spectrum splits.

Close modal

Under these optimized MLL conditions, we have characterized the temporal and spectral characteristics of the laser. In the Figure 5(a) inset, a pulse train is observed by using a digital sampling oscilloscope with an optical detection module and triggered with an amplified RF signal. The pulse period is approximately 136 ps, in agreement with the 7.3 GHz RF spectrum shown in Figure 5(a), and corresponding to the fundamental cavity repetition rate. The side peaks and broadened pulse width observed in each pulse are due to the non-ideal response of the photodetector. In order to measure the actual pulse width, the SOA amplified pulse train is sent into a Femtochrome autocorrelator (FR-130HS). The inset in Figure 5(b) shows the autocorrelation trace with a Full-Width-Half-Maximum (FWHM) of 2.96 ps, which corresponds to a deconvolved 1.9 ps pulse using a sech2 fitting. The corresponding optical spectrum is measured to have a FWHM of 0.97 nm [∼172 GHz] (Figure 5(b)). The resulting time-bandwidth product is 0.326 which is comparable to the sech2 shape Fourier transfer limited theoretical value of 0.315, i.e., 1.029× the transform limit or within 5% of our experimental value. The asymmetric pulse shape is attributed to gain clipping in the SOA or chirp effects inside the bonded MLL.28,29

FIG. 5.

(a) RF spectrum and pulse train (inset); and (b) optical spectrum and temporal pulse width (inset) of the bonded mode-locked QD laser under the optimized bias condition of Ig = 124 mA and Va = −7 V, respectively.

FIG. 5.

(a) RF spectrum and pulse train (inset); and (b) optical spectrum and temporal pulse width (inset) of the bonded mode-locked QD laser under the optimized bias condition of Ig = 124 mA and Va = −7 V, respectively.

Close modal

In conclusion, we have demonstrated nearly transform limited quantum dot mode-locked lasers on silicon by low temperature Pd-GaAs wafer bonding technology. The optimized mode-locked lasers operate near 1.3 μm wavelength with a 7.3 GHz repetition rate and about a 2 ps pulse width. In addition, mode-locking was investigated over a wide range of operating conditions by adjusting gain injection currents and reverse bias voltages of the saturable absorber. These hybrid InAs QD MLLs are a promising solution to integrating laser sources together with silicon photonic integrated circuits (PICs).

The work was supported by NSF under Grant No. ECCS-1309230. This work was performed in part at the Cornell Nano Scale Facility, a member of the National Nanotechnology Coordinated Infrastructure (NNCI), which is supported by the National Science Foundation (Grant No. ECCS-1542081).

1.
W.
Bogaerts
,
R.
Baets
,
P.
Dumon
,
V.
Wiaux
,
S.
Beckx
,
D.
Taillaert
,
B.
Luyssaert
,
J.
Van Campenhout
,
P.
Bienstman
, and
D.
Van Thourhout
,
J. Lightwave Technol.
23
(
1
),
401
412
(
2005
).
2.
S. J.
McNab
,
N.
Moll
, and
Y. A.
Vlasov
,
Opt. Express
11
(
22
),
2927
2939
(
2003
).
3.
M.
Notomi
,
A.
Shinya
,
S.
Mitsugi
,
E.
Kuramochi
, and
H.
Ryu
,
Opt. Express
12
(
8
),
1551
1561
(
2004
).
4.
S.
Saeedi
and
A.
Emami
,
Opt. Express
23
(
17
),
21875
21883
(
2015
).
5.
Q.
Xu
,
B.
Schmidt
,
S.
Pradhan
, and
M.
Lipson
,
Nature
435
(
7040
),
325
327
(
2005
).
6.
A. W.
Fang
,
H.
Park
,
O.
Cohen
,
R.
Jones
,
M. J.
Paniccia
, and
J. E.
Bowers
,
Opt. Express
14
(
20
),
9203
9210
(
2006
).
7.
D.
Liang
,
M.
Fiorentino
,
S.
Srinivasan
,
J. E.
Bowers
, and
R. G.
Beausoleil
,
IEEE J. Sel. Top. Quantum Electron.
17
(
6
),
1528
1533
(
2011
).
8.
D.
Bimberg
,
M.
Grundmann
, and
N. N.
Ledentsov
,
Quantum Dot Heterostructures
(
John Wiley & Sons
,
1999
).
9.
S.
Mikhrin
,
A.
Kovsh
,
I.
Krestnikov
,
A.
Kozhukhov
,
D.
Livshits
,
N.
Ledentsov
,
Y. M.
Shernyakov
,
I.
Novikov
,
M.
Maximov
, and
V.
Ustinov
,
Semicond. Sci. Technol.
20
(
5
),
340
(
2005
).
10.
M.
Sugawara
,
N.
Hatori
,
M.
Ishida
,
H.
Ebe
,
Y.
Arakawa
,
T.
Akiyama
,
K.
Otsubo
,
T.
Yamamoto
, and
Y.
Nakata
,
J. Phys. D: Appl. Phys.
38
(
13
),
2126
(
2005
).
11.
S.
Chen
,
W.
Li
,
J.
Wu
,
Q.
Jiang
,
M.
Tang
,
S.
Shutts
,
S. N.
Elliott
,
A.
Sobiesierski
,
A. J.
Seeds
, and
I.
Ross
,
Nat. Photonics
10
,
307
311
(
2016
).
12.
Y.
Wan
,
Q.
Li
,
A. Y.
Liu
,
W. W.
Chow
,
A. C.
Gossard
,
J. E.
Bowers
,
E. L.
Hu
, and
K. M.
Lau
,
Appl. Phys. Lett.
108
(
22
),
221101
(
2016
).
13.
A. Y.
Liu
,
J.
Peters
,
X.
Huang
,
D.
Jung
,
J.
Norman
,
M. L.
Lee
,
A. C.
Gossard
, and
J. E.
Bowers
,
Opt. Lett.
42
(
2
),
338
341
(
2017
).
14.
K.
Tanabe
,
D.
Guimard
,
D.
Bordel
,
S.
Iwamoto
, and
Y.
Arakawa
,
Opt. Express
18
(
10
),
10604
10608
(
2010
).
15.
T.
Tatsumi
,
K.
Tanabe
,
K.
Watanabe
,
S.
Iwamoto
, and
Y.
Arakawa
,
J. Appl. Phys.
112
(
3
),
033107
(
2012
).
16.
Z.
Wang
,
R.
Yao
,
S. F.
Preble
,
C.-S.
Lee
,
L. F.
Lester
, and
W.
Guo
,
Appl. Phys. Lett.
107
(
26
),
261107
(
2015
).
17.
P.
Vasil'ev
,
Ultrafast Diode Lasers: Fundamentals and Applications
(
Artech House
,
London
,
1995
).
18.
A. W.
Fang
,
B. R.
Koch
,
K.-G.
Gan
,
H.
Park
,
R.
Jones
,
O.
Cohen
,
M. J.
Paniccia
,
D. J.
Blumenthal
, and
J. E.
Bowers
,
Opt. Express
16
(
2
),
1393
1398
(
2008
).
19.
B. R.
Koch
,
A. W.
Fang
,
O.
Cohen
, and
J. E.
Bowers
,
Opt. Express
15
(
18
),
11225
11233
(
2007
).
20.
X.
Huang
,
A.
Stintz
,
H.
Li
,
L.
Lester
,
J.
Cheng
, and
K.
Malloy
,
Appl. Phys. Lett.
78
(
19
),
2825
2827
(
2001
).
21.
E.
Rafailov
,
M.
Cataluna
,
W.
Sibbett
,
N.
Il'inskaya
,
Y. M.
Zadiranov
,
A.
Zhukov
,
V.
Ustinov
,
D.
Livshits
,
A.
Kovsh
, and
N.
Ledentsov
,
Appl. Phys. Lett.
87
(
8
),
081107
(
2005
).
22.
J.
Mee
,
M.
Crowley
,
N.
Patel
,
D.
Murrell
,
R.
Raghunathan
,
A.
Aboketaf
,
A.
Elshaari
,
S.
Preble
,
P.
Ampadu
, and
L.
Lester
,
Appl. Phys. Lett.
101
(
7
),
071112
(
2012
).
23.
A. A.
Aboketaf
,
A. W.
Elshaari
, and
S. F.
Preble
,
Opt. Express
18
(
13
),
13529
13535
(
2010
).
24.
H.
Shahid
,
D.
Childs
,
B.
Stevens
, and
R.
Hogg
,
Appl. Phys. Lett.
99
(
6
),
061104
(
2011
).
25.
A.
Gubenko
,
D.
Livshits
,
I.
Krestnikov
,
S.
Mikhrin
,
A.
Kozhukhov
,
A.
Kovsh
,
N.
Ledentsov
,
A.
Zhukov
, and
E.
Portnoi
,
Electron. Lett.
41
(
20
),
1124
1125
(
2005
).
26.
M.
Tahvili
,
L.
Du
,
M.
Heck
,
R.
Nötzel
,
M.
Smit
, and
E.
Bente
,
Opt. Express
20
(
7
),
8117
8135
(
2012
).
27.
Y.-C.
Xin
,
D. J.
Kane
, and
L.
Lester
,
Electron. Lett.
44
(
21
),
1255
1257
(
2008
).
28.
Y.
Li
,
L. F.
Lester
,
D.
Chang
,
C.
Langrock
,
M.
Fejer
, and
D. J.
Kane
,
Opt. Express
21
(
7
),
8007
8017
(
2013
).
29.
M.
Margalit
,
M.
Orenstein
,
G.
Eisenstein
, and
V.
Mikhaelshvili
,
Opt. Lett.
19
(
24
),
2125
2127
(
1994
).