PDF 1

Review Article
Taking silicon photonics modulators to a higher

performance level: state-of-the-art and a review
of new technologies
Abdul Rahim ,a,b,* Artur Hermans ,a,b Benjamin Wohlfeil,c Despoina Petousi,c Bart Kuyken,a,b
Dries Van Thourhout,a,b and Roel Baetsa,b,*
a
Ghent University, Photonics Research Group, Department of Information Technology, Ghent, Belgium
b
Ghent University, IMEC and Center for Nano- and Biophotonics, Ghent, Belgium
c
ADVA Optical Networking, Berlin, Germany
Abstract. Optical links are moving to higher and higher transmission speeds while shrinking to shorter and
shorter ranges where optical links are envisaged even at the chip scale. The scaling in data speed and span of
the optical links demands modulators to be concurrently performant and cost-effective. Silicon photonics
(SiPh), a photonic integrated circuit technology that leverages the fabrication sophistication of comple-
mentary metal-oxide-semiconductor technology, is well-positioned to deliver the performance, price, and
manufacturing volume for the high-speed modulators of future optical communication links. SiPh has relied
on the plasma dispersion effect, either in injection, depletion, or accumulation mode, to demonstrate efficient
high-speed modulators. The high-speed plasma dispersion silicon modulators have been commercially
deployed and have demonstrated excellent performance. Recent years have seen a paradigm shift where
the integration of various electro-refractive and electro-absorptive materials has opened up additional
routes toward performant SiPh modulators. These modulators are in the early years of their development.
They promise to extend the performance beyond the limits set by the physical properties of silicon.
The focus of our study is to provide a comprehensive review of contemporary (i.e., plasma dispersion
modulators) and new modulator implementations that involve the integration of novel materials with SiPh.
Keywords: high-speed modulators; silicon photonics; plasma dispersion effect; ferroelectrics; graphene; III–V on Si; organic
(electro-optic) materials.
Received Sep. 24, 2020; revised manuscript received Jan. 19, 2021; accepted for publication Mar. 10, 2021; published online
Apr. 29, 2021.
© The Authors. Published by SPIE and CLP under a Creative Commons Attribution 4.0 Unported License. Distribution or
reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI.
[DOI: 10.1117/1.AP.3.2.024003]
1 Introduction of voltage V π applied to the phase shifter to achieve π phase

shift and the length L of the phase shifter—a smaller V π · L rep-
High-speed transceivers for short-reach (few cm to few tens resents higher modulation efficiency; for amplitude modulation,
of km) and long-haul optical communication links require the ratio of extinction ratio (ER) to an applied voltage for a given
high-performance modulators in their transmitter optical length of the modulator represents the modulation efficiency
subassembly.1–3 The key performance parameters for an efficient and large values are desired]; (b) low insertion loss (IL),10–13 both
modulator are (a) high modulation efficiency4–9 in amplitude or low loss in the active and passive parts of the modulator; (c) tens
phase by a drive signal that is compliant with the complemen- of gigabits per second (Gb/s)14–17 modulation speed to support
tary metal-oxide-semiconductor (CMOS) circuitry5,6 [for phase the future capacity demands of the optical communication
modulators, the modulation efficiency is defined as the product networks; (d) energy consumption to be as low as a few tens of
fJ/bit;18–20 (e) large ER—typically ∼4 dB for ∼2 km transmis-
*Address all correspondence to Abdul Rahim, [email protected]; Roel Baets,
sion links and ∼8 dB for tens of km long transmission links—
[email protected] to ensure high optical signal-to-noise ratio for maintaining a low
Advanced Photonics 024003-1 Mar∕Apr 2021 • Vol. 3(2)

Downloaded From: https://www.spiedigitallibrary.org/journals/Advanced-Photonics on 14 Feb 2023
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Rahim et al.: Taking silicon photonics modulators to a higher performance level: state-of-the-art…
bit error rate;21–25 (f) low chirp26–28 to realize higher-order modu- inter-DCIs are limited to a maximum power consumption of
lation schemes27–29 and to enhance dispersion tolerance of around 20 W in total, and new form factors such as octal small
transmission links;26,30 and (g) compact size9,12,31–33 to enable form factor pluggable or quad small form factor pluggable-
high-integration density.1,2,21,34–36 double density are specified to an even lower maximum power
Generally, the modulator figure of merit (FoM) captures consumption. This means that the modulator and driver power
the performance of a modulator. In the literature, a variety of consumption should be carefully optimized. The most widely
definitions for modulator FoM exist.1,2,21,37,38 In its simplest used architecture for inter-DCIs is the Mach–Zehnder modulator
form, V π · L represents the FoM for a phase modulator. V π · L · (MZM) in in-phase, quadrature configuration, whereby MZMs
α is a phase modulator FoM that takes the phase shifter are nested to modulate the in-phase and quadrature component
loss α into account. A phase modulator FoM given as of the signal.55,56 For demonstrating high-order quadrature am-
f c · ½CL · ðV π · LÞ2 · α∕8−1 , where f c represents the phase plitude modulation (QAM), it is extremely critical for the MZM
shifter cutoff frequency and CL represents the phase shifter to show a linear intensity dependence with voltage (low chirp) to
capacitance, relates the FoM in terms of modulator bandwidth, minimize the error vector magnitude.27 Furthermore, the need
loss, and energy per bit.2,4,21 The ratio ER/IL gives the FoM for for more bandwidth in wavelength division multiplexing sys-
amplitude modulators.20,38 Dynamic optical modulation ampli- tems requires broadband and tunable devices, increasing the
tude is a dynamic FoM of a modulator, as it takes modulator need to optimize components for use in optical bands beyond
parameters and the modulation format into account.37 the C-band (1.53 to 1.565 μm), i.e., L-band (1.565 to 1.625 μm)
High-index-contrast silicon photonics (SiPh) provides dense and/or S-band (1.46 to 1.53 μm).
integration of complex photonic functionalities, such as high- • Due to the shorter distance requirements of up to 2 km
speed modulators, using technology and process toolsets from (FR) or up to 10 km (LR), the intra-DCI applications are mainly
CMOS electronics.39 Other photonic integrated circuit (PIC) using multi-channel 56 Gbaud, four-level pulse amplitude
platforms such as indium phosphide,40–42 which typically relies modulation (PAM-4). As a result, the transmitter requirements
on 100 mm wafers, and lithium niobate (LiNbO3 ),43 which com- in this case are more relaxed in terms of linearity, but a high
monly relies on 150 mm wafers, are fabricated in customized dynamic ER (ER >4.5 dB57) is necessary in order to achieve
fabs. SiPh integrated circuits are typically fabricated on 200 multi-level modulation. In this case, QSFP form factors require
or 300 mm wafers. The large wafer size allows for a large num- a maximum total power consumption of 5.5 W. Mainly because
of the very strict power consumption requirements, up to now,
ber of compact dies per wafer at low-cost scaling to large com-
vertical-cavity surface-emitting laser (VCSEL)-based transceiv-
mercial volumes in existing CMOS commercial fabs and R&D
ers and parallel fiber transmission have been used inside DCs,
pilot lines.39,44–48 SiPh modulators promise to provide the best
especially on short reaches. However, with the massive increase
price-performance ratio and possible routes for monolithic inte-
of traffic, there has been a continuous interest to replace
gration of electronics and photonics.2,21,22,39,49 The commercial VCSELs with next generation ultra-compact devices that will
launch of SiPh-enabled transceivers is a testament to the prom- provide even lower cost-energy consumption figures. With
ise of SiPh as a photonic integration technology to deliver effi- the next generations of intra-DCI transceivers aiming for data
cient high-speed modulators.3,44–48,50–52 rates of 800 Gb∕s and beyond, a potential move of coherent
The push for performant high-speed modulators stems from technology into the domain of intensity modulation and direct
the ever booming growth of internet traffic. This growth has detection may be seen, as scaling of intensity modulated signals
driven the modern data centers (DCs) into the so-called to higher capacity is increasingly challenging.58 However, to
“Zettabyte Era,” where the annual global traffic is expected meet the stringent power consumption target for intra-DCI
to exceed 2.2 zettabytes/year. This value corresponds to a three- transceivers, lightweight digital signal processing, lower power
fold increase in just five years. The biggest contribution to this consumption components, and potentially a reduced coherent
huge traffic originates from data moving inside the DCs over system will be required.
inter-DC connections.53 Nonetheless, a remarkable increase in
data exchanged between different DCs is also reported within For both intra- and inter-DCIs, SiPh appears to be a prom-
the last few years (inter-DC connections). Future massive-scale ising technology platform,59,60 mainly because of the dense
DCs are expected to spread widely across the globe in order to integration of photonics59,60 while minimizing the overall cost-
meet the enormous bandwidth demands. An increase in baud power efficiency budget.18,19,39 Currently, the plasma dispersion
rate imposes itself, yet at an even lower power consumption effect, which is based on the movement (injection,7,37,61–67
per bit. Consequently, crucial optimization of the intra- and accumulation,12,68–70 or depletion9,13,14,16,22,32,33,71–87) of carriers to
inter-data center interconnects (DCIs)—where high-speed mod- induce a refractive index change in a silicon waveguide, is the
ulators are a key building block—is needed. More specifically: most widely deployed phenomenon to implement high-speed
modulators in SiPh.5,7–19,22,32,33,37,61–69,69–71,71–121 Plasma dispersion
• The inter-DCI landscape is dominated by coherent dense
wavelength division multiplexing optics for reaching high modulators in SiPh have successfully demonstrated operations
capacities per single wavelength, currently targeting up to of 90 (Ref. 87) and 100 Gb∕s122,123 for the on–off keying (OOK)
38.4 Tb∕s aggregate capacity in a distance between 40 and modulation scheme. Furthermore, various plasma dispersion
1000s of kilometers with data rates of currently 400 Gb∕s modulators implementations for advanced modulation schemes
to 600 Gb∕s (64 Gbaud) per wavelength. By 2025, data rates such as QPSK,55,56,124 QAM,125,126 PAM-4,29,52,127 and PAM-8115
are expected to double 54 with symbol rates exceeding also exist.
100 Gbaud. These values set strict requirements regarding the The plasma dispersion effect is a weak electro-refractive
transmitter performance, primarily in terms of bandwidth. effect.128 A 1017 cm−3 change in carrier concentration introduces
Regarding power consumption, pluggable modules such as an index change of Δn ∼ 10−4 . Furthermore, the phase modu-
C form factor pluggable (CFP) and CFP2 used for current lation provided by the plasma dispersion effect is always

associated with spurious amplitude modulation.21,26 Unlike single lane 100 Gbaud and beyond data rates in an energy-
Pockels modulators, plasma dispersion modulators do not have efficient manner. Table 1 shows a rich landscape of SiPh
a linear dependence of voltage and refractive index change Δn modulators, where SiPh is incorporating a variety of other
(thus, they show chirp), which makes them a less promising can- materials for implementing performant high-speed modulators.
didate for cases that require pure phase modulation.26,55,56,124,125,129 Modulation of amplitude by the Franz–Keldysh (FK) ef-
Nevertheless, R&D efforts have resulted in the performance of fect,20,38,128,133–140 quantum-confined Stark (QCS) effect,141–152 and
SiPh plasma dispersion modulators that is comparable with electrical gating, 153–160 or the modulation of phase by employing
other photonic integration technologies.27 the plasma dispersion effect,161–166 Pockels effect,167–192,192–200 and
Typically, plasma dispersion modulators rely on interfero- inter-band transitions,201–208 is feasible either directly in
metric structures [such as the Mach–Zehnder interferometer silicon20,31,38,133–139,143 or by integrating a variety of efficient
(MZI)13,19,78,84,108,118 or optical cavities9,12,32,33,100,115] for the trans- electro-optic materials with SiPh platforms.161,172,182,184,202 The in-
lation of phase modulation into amplitude modulation. The tegration of new materials into SiPh has to counter various chal-
carrier depletion-based interferometric plasma dispersion mod- lenges. Invariably, any new material integration requires new
ulators, such as MZMs, are large in size (typically a phase technology and process development that allows for low-loss
shifter length of a few mm).21,22 Injection or accumulation of integration with the SiPh platform. The integration process must
carriers offers higher efficiency plasma dispersion effect phase ensure high material quality. The introduction of defects in the
shifters, enabling compact interferometric modulators.37,65 The integration process downgrades the yield of the PIC manufac-
implementations of plasma dispersion high-speed modulators turing process. Consequently, the integration of new materials
involving optical cavities [such as ring resonators (RRs)] result always adds to manufacturing costs. Materials allowing wafer-
in compact devices (typically a few hundreds of μm2 ).9,12,32,33 scale integration and the ones for which the fab can leverage
However, these RR-based devices suffer from high temperature from the existing learning curve of introduction into other plat-
and fabrication sensitivity.130,131 Literature reports solving these forms reduce the financial impact and technological complex-
problems to a certain extent through smart design strategies ities, and they are thus preferred.
and technological advancements.132 Nonetheless, the limited op- The electrical driving of the modulator comprises an electrode
erational optical bandwidth of the ring modulator (RM) is intrin- configuration and the driving scheme (such as single-ended
sic to them. drive with dual-arm push-pull, differential drive with dual-
It is a daunting task to deliver the future demands for single arm push-pull, differential drive with dual-arm push-pull with
lane 100 Gbaud and beyond data rates in an energy-efficient shared drive, or dual-differential drive with dual-arm push-pull)
manner. Recently, there have been a growing number of efforts to feed the electrodes.25 Generally, the length of the phase shifter
to integrate electro-optic materials with a strong Pockels effect determines the electrode configuration.209 Modulators with a
on SiPh platforms. The integration of such electro-optic materi- short phase shifter length (phase shifter length <λRF ∕10)2,25 use
als promises to provide additional routes to upscale the perfor- lumped electrodes.209,210 Lumped driving is applicable to mod-
mance of SiPh modulators to deliver the future demands for ulators comprising interferometric structures,65,210,211 as well as
Table 1 Prominent approaches for high-speed modulation in SiPh. The modulator implementations in SiPh use a variety of physical
phenomenons, materials, optical architectures, and driver architectures.
The landscape of high-speed modulators in silicon photonics

Reported optical Reported driver
Modulation Operating principle Platform implementation implementation
Phase Plasma dispersion effect by carrier Silicon MZI, michelson, resonators Lumped, traveling
(a) Injection (ring, disk, ph. crystal, wave (TW),
Fabry–Perot), slow-light segmented
(b) Accumulation
structure, Bragg reflectors
(c) Depletion
Pockels effect LiNbO3 on silicon MZI Lumped, TW
Organics on silicon MZI, ring resonator Lumped, TW
BTO on silicon MZI, ring resonator Lumped, TW
PZT on silicon MZI, ring resonator Lumped
Interband transitions 2D materials on silicon MZI, ring resonator Lumped, TW
Carrier accumulations/carrier III-V on silicon MZI, ring resonator Lumped, TW
depletion+Franz-Keldysh effect
Amplitude Franz-Keldysh effect Silicon-germanium Waveguide, MZI Lumped
Quantum confined Stark effect Ge-Si-Ge quantum wells Waveguide, Lumped
Fabry–Perot cavity
Electrical gating 2D materials on silicon Waveguide Lumped
Quantum confined Stark effect III-V on silicon Waveguide Lumped

cavity-based structures9,12,62,70,115 and slow-light structures.112,113 and photonic phase shifter design25,52,122 are a key to extracting the
With lumped electrodes, the applied voltage is approximately best performance for all modulator performance attributes.52,213
identical across the entire length of the phase shifter. Lumped This holds for modulators involving monolithic or hybrid inte-
driving does not necessarily require a matched termination of gration with electronics.25,122,213
the driving electrodes. With unterminated lumped driving, the This paper aims to provide a comprehensive review of the
phase shifter’s capacitive load determines the energy consump- new technologies that provide additional, viable routes to en-
tion per bit (Ebit ) for a modulator. Typically, the bandwidth of hance the performance of SiPh modulators. These new technol-
lumped electrode phase shifter driving is limited by its imped- ogies involve the integration of materials with high electro-optic
ance (intrinsic, parasitic, and the one contributed by the driving coefficients with SiPh platforms. As a result, they promise to
circuitry) and the time taken by the photons to transit through deliver modulator performance that is required to meet the
the phase shifter. The latter effect is more pronounced for the blistering surge in transmission rates for the optical links. This
slow-light modulators.113,114 paper also compares these new technologies against the current
The traveling wave (TW) is the most common driving scheme mainstream SiPh modulator implementations. A comprehensive
for carrier depletion plasma dispersion MZMs.2,4,25,210,212 In this coverage in full detail is very challenging and difficult to be
scheme, a single driver drives the entire electrode of the phase performed in a paper with a limited page number. Therefore, we
shifter.25,212 The TW electrode typically comprises a coplanar will limit the discussion in this paper to some of the most im-
waveguide transmission line. It terminates with a resistive portant and relevant implementations of high-speed modulators
impedance matched to the wave impedance of the electrode. in SiPh. It is important to note that the SiPh modulators have
The RF power applied to the TW electrode is consumed by been used to modulate signals by means of advanced modula-
the RF losses in the TW, capacitive loading by the phase shifter, tion schemes such as QPSK,55,56,124 QAM,125,126 PAM-4,29,52,127
and the matched termination. The matched resistance prevents and PAM-8.115 However, in this paper, the operating speed is
electrical reflections and consequent interference with the elec- quoted only for OOK. Furthermore, the discussion in this paper
trical signal stream. But, it also increases the power consump- is limited to C-band and O-band implementations of SiPh
tion of the TW electrode as a fraction of the total input RF power modulators for telecommunication and data communication
is always consumed by the matched termination. Generally, TW applications; though SiPh modulators are also reported for
driving schemes are six to eight times more energy-hungry than mid-infrared (mid-IR) wavelengths214,215 and for non-telecom/
the lumped driving.2,4 The detrimental factors impacting the datacom applications.215
bandwidth of the modulators with TW driving are: (a) RC time
constant of the phase shifter; (b) walk-off (phase mismatch due 2 State-of-the-Art of Contemporary Silicon
to mismatch between the propagation speed of the electrical
signal in the transmission line and the optical signal in the wave- Photonic High-Speed Modulators
guide) between the electrical and optical signals; and (c) RF- Research to develop SiPh modulators dates back to the era when
losses in the TW electrode. it was still a niche choice for photonic integration.61,64,88–90,216
The segmented topology with distributed driving comprises Unstrained silicon has a fundamentally zero linear electro-optic
a series of short phase shifters, which are individually driven effect (the Pockels effect, used in traditional LiNbO3 modula-
by lumped electrodes.6,22,66 Segmentation of the phase shifter tors) due to its crystallographic structure (i.e., centrosymmetric
with distributed driving avoids the RF attenuation due to a long crystal lattice), albeit strained silicon is investigated for high-
transmission line while enabling the same voltage swing across speed all-silicon modulation.217–219 This intrinsic property of
the whole length of the phase shifter.212 As compared to the Si directed the researchers to investigate other optical phenom-
TW architecture, the segmented electrodes require smaller- enons, such as the plasma dispersion effect to realize high-
size voltage drivers.25 The precise delivery of the electrical performance modulators in SiPh. This section provides a
signal to a phase shifter segment requires time synchronization summary of the plasma dispersion modulators to implement
with the transit time of light in that segment, which requires SiPh high-speed modulators.
timing control circuitry for precise delivery of the electrical
signal to each segment. This additional overhead undermines
2.1 High-Speed Modulators Using the Plasma
the energy efficiency of the segmented drivers and adds to the
Dispersion Effect
driver complexity. The number of phase shifter segments and
their length determine the modulator bandwidth and its power The refractive index change by the plasma dispersion effect
efficiency. Typically, segmented electrodes with distributed in silicon is larger than the change by the Kerr61,220 and FK
drivers require tighter connectivity between the electronic effects.31,63,141 The plasma dispersion effect is broadband going
drivers and the phase shifter segments to preserve the signal all the way from telecom to mid-IR wavelengths221 and is rela-
integrity and prevent parasitic effects. Therefore, this driving tively temperature independent.222–225 Furthermore, its implemen-
scheme demands either monolithic integration of electronics tation is inherently simple, using standard CMOS processing
and photonics or flip-chip bonding photonic and electronic technology.81 All of these factors made the plasma dispersion ef-
chips through an array of microbumps. Electrical driving fect one of the current mainstream mechanisms to implement
plays a crucial role in defining the performance (primarily efficient high-speed modulators in SiPh.2,21,35,49 Years of devel-
the power consumption and bandwidth) of high-speed phase opment, led by both industry and academic players, have re-
modulators.2,122,210 In the early years, most of the high-speed sulted in the demonstration of plasma dispersion effect-based
modulator demonstrations focused on the optimization of modu- phase modulators providing losses of <10 dB∕cm,71,80,116,118,121
lator performance in isolation from the driver design.122 Recent phase modulation efficiencies V π · L < 0.3 V · cm,5,69,104 energy
demonstrations of high-speed modulators have established that consumption as low as tens of fJ/bit,18,99,110,111,115 and data rates
concurrent co-design and co-optimization of electrical driving reaching 100 Gb∕s 14,87,122,123 for OOK modulation with minimal

influence of temperature on the performance of the plasma reported using lateral PIN63,65,67,92 and vertical PIN configura-
dispersion modulators (III–V modulators rely on the QCS effect tions (see Fig. 1).63,89,93,216,226 Owing to the feasibility for a short
and are influenced by temperature). It is important to note that phase shifter (∼ a few hundreds of μm), carrier injection mod-
all of these specs are not provided simultaneously by a single ulators typically use lumped driving schemes.7,65,67 The use of
plasma dispersion modulation implementation.21 pre-emphasized electrical signaling schemes, in which a fraction
Plasma dispersion is an electro-refractive effect. The change of the electrical driver signal has a voltage ≫V π of the modu-
of free carrier concentration by the movement of charge carriers lator, boosts the limited speeds of carrier injection-based
into or out of a waveguide results in phase modulation of modulators67,95–97 with the obvious ramification of increased
the optical signal.21,49,61 Phase modulation is translated into power consumption. Recent years have seen the development
intensity modulation by embedding the phase modulator of passive equalization techniques to boost the high-speed
into MZIs,15–17,77,79,108,109 RRs,18,32,33,62,110,111,119 Bragg reflectors,88 operation of the carrier injection modulators.65,67 The studies
Michelson interferometers,8,121 photonic crystal cavities,98,112–114 have shown that the factors that limit the speed of the carrier
and Fabry–Perot cavities.100 The three prominent schemes to injection-based modulation are (a) recombination time of the
introduce change in the free carrier concentration are (a) the injected electron–hole pairs and (b) the sum of the electrical
injection of minority carriers7,37,61–67 by forward biasing a PIN driver output resistance and the bulk resistance of the p- and
junction; (b) the accumulation of majority carriers12,68,69,69,70 of n-doped regions.7,65,67 The slower of these two effects determines
opposing polarity across an insulating section in a waveguide; the speed limit of the carrier injection modulators.67 Practically,
(c) the depletion of majority carriers9,13,14,16,22,32,33,71,71–87 from a it is possible to improve the bandwidth of carrier injection
PN junction by reversely biasing it. modulators by embedding an RE CE passive equalizer such that
Carrier injection is the most efficient plasma dispersion- RE CE ¼ RF CF and CF ≫ CE , where RE;F and CE;F are the
based modulation scheme.7,8,37,98 Most of the early demonstra- equalizer and the forward resistance and the capacitance,
tions of high-speed modulators relied on carrier injection of respectively.7,65,91 Current state-of-the-art carrier injection mod-
minority carriers by forward biasing a PIN junction built along ulators use this passive equalization to demonstrate >20 Gb∕s
a waveguide.63,89,90,93,216,226 The carrier injection-based modula- operation.7,65 Recently, 70 Gbaud operation of a carrier injection
tors provide a high modulation efficiency due to their large MZM was enabled by passive RC equalization.7 The modulator
diffusion capacitance (typically ∼10 pF) when implemented has a 37-GHz 3-dB bandwidth using a 0.25-mm long-phase
on a waveguide with a small cross section with stronger shifter having ∼28 dB∕cm loss and a V π · L of 2 V · cm.7
confinement.7,37,65–67,91–94 Carrier injection-based modulators are One of the best modulation efficiencies of 0.274 V · cm has
(a)
(b)
Fig. 1 (a) Baseline architecture of carrier injection, carrier depletion, and carrier accumulation
plasma dispersion phase shifters. (b) Various configurations of plasma dispersion phase shifters.

been reported in Ref. 67, where a side-wall corrugated MZM scheme to implement the plasma dispersion-based phase
with a bandwidth of 12.5 GHz is operated at 25 Gb∕s using modulator.8–11,13–19,22,32,33,71,71–87,106–121 Depletion-based modulators
a finite impulse response filter for frequency compensation were the first all-silicon modulator to reach speeds of 40106
necessary for the broadband operation.67 The demonstrated and 50 Gb∕s.16 The current highest reported data rates of
modulator had a 250-μm-long phase shifter that exhibited a 100 Gb∕s using OOK modulation were also demonstrated in
propagation loss of 5.3 dB∕mm. carrier depletion modulators.122,123 Carrier depletion phase
The carrier accumulation plasma dispersion phase shifter re- shifters exhibit a relatively limited capacitance of 0.2 to
quires forward biasing of a metal-oxide-semiconductor (MOS) 0.8 pF∕mm21,91 and, hence, a limited modulation efficiency.
capacitor.2,68,101,102 A typical implementation of an MOS (also In order to improve the modulation efficiency, the capacitance
known as silicon–insulator–silicon capacitor—SISCAP) modu- can be increased by either shrinking the mode size (determined
lator comprises a vertical slot waveguide with an overlapping by the waveguide geometry) or by reducing the width of the
stack of a sub-μm-sized thick p-type poly-silicon layer on top depletion region (higher transition capacitance).21 The latter
of a sub-μm-sized thick n-type c-silicon layer with a few nm requires higher doping concentrations, which leads to higher
(typically <20 nm) thick insulating gate oxide layer in the loss due to free carrier absorption. Over the years, a wide variety
middle (i.e., p-Si/insulator/c-Si vertical slot waveguide)—see of PN junction designs are reported to solve this challenge of
Fig. 1(a).69,101,102 Under accumulation conditions, in which a pos- optimizing modulator speed, efficiency, and loss at the same
itive bias is applied to the p-type layer, the vertically confined time.13,15,17,18,22,71–76,78,107–109,116,117,119 Distinct categories of PN junc-
optical mode overlaps strongly with the highly capacitive gate tion implementation are as follows:
section.101,102 This enables <1 V · cm modulation efficiency for
SISCAP modulators37 with state-of-the-art values as low as 0.2 (i) Vertical P(I)N junction, in which the PN junction is hori-
and 0.16 V · cm.5,68,69 Owing to this higher modulation efficiency, zontally placed along the top and bottom surface of the
typically ≤0.5-mm-long phase shifters with lumped electrical waveguide18,107–111 [see Fig. 1(b)]. These junctions are known
driving are implemented in SISCAP modulators.65,67,69,69,103 to provide higher overlap between the optical mode and the
The higher capacitance, which leads to higher modulation effi- depletion region to improve the modulation efficiency (higher
capacitance) and reduced power consumption.18,107–111
ciency, caps the modulation speed of SISCAP modulators.2,104
(ii) Horizontal or lateral PN junction, in which the PN junction
The modulation efficiency and speed can be traded-off in
is vertically placed side-by-side in the center of the waveguide
SISCAP modulators by adopting the capacitance to a value
(laterally symmetric design)17,116,117,119 [see Fig. 1(b)]. The hori-
that can provide either an efficient modulator (i.e., high capaci-
zontal PN junction is abundantly used in various adapted forms
tance) with low V π · L or high-speed modulator (i.e., low
to improve the α · V π · L of the modulator. A few examples
capacitance) without any degradation of the loss parameter
include (a) lateral PN junction with an offset from the center
for the modulator (the increase or decrease in capacitance is in-
of the active region of the modulator,13,22,71–76 and (b) lateral
dependent of the doping level and is controlled by the thickness
junction formed by placing a p-doped slit in the center of the
of gate oxide and the relative permitivity of the material used as
waveguide78 while leaving the waveguide edges or corners
the gate oxide).2,70,101,102,104 Typically, a vertical capacitor con-
un-doped.15,107 The rationale of (a) and (b) is to exploit the higher
figuration is used in SISCAP modulators due to the ease of
index change and lower loss provided by the dominantly
fabrication.5,68,69,69,70,104 It is straightforward to fabricate by the
p-doped active region.13,15,71–75,78,107
deposition of poly-Si.2,101 Techniques such as epitaxial lateral (iii) Interleaved or interdigitated PN junction, which comprises
growth101,102 and laser-induced epitaxial growth crystallize the an alignment tolerant junction implemented by alternating
poly-silicon layer to overcome the higher scattering loss due the p-type and n-type doping regions along the length of the
to defects in the lattice.2,227 Recently, SISCAP modulators using waveguide81,83,84 [see Fig. 1(b)]. This type of junction is known
a c-Si/insulator/c-Si in a vertical slot configuration70 and in a to provide a good modulation efficiency by trading-off the speed
horizontal slot configuration103 without requiring any additional and power efficiency for such a junction.32,80–84
processing steps have been reported [see Fig. 1(b)]. One of the
first 40 Gb∕s Mach–Zehnder SISCAP modulators relied on One of the first 40 Gb∕s modulators relied on a vertical PIN
a vertical capacitor configuration, and it delivered an ER of junction and delivered a modulation efficiency of 4 V · cm
9 dB.68 The design consisted of a 400-μm-long MOS capacitor and phase shifter loss of 18 dB∕cm.106 Using a vertical P(I)N
with a 6.5-dB∕mm loss delivering a V π · L of 0.2 cm for junction design, an athermal microdisk modulator with a high
the O-band. The same MZM is driven by a CMOS driver to modulation efficiency of 250 pm∕V consuming 0.9 fJ∕bit while
demonstrate its low-power operation for NRZ-OOK and for operating at 25 Gb∕s using NRZ-OOK signaling has been
advanced modulation formats such as PAM-4 using a segmented reported.18 At 25 Gb∕s, the modulator had an IL of ∼1 dB and
driver.5 A recent demonstration reports high modulation effi- an ER of 6 dB. One of the highest operating speeds of 60 Gb∕s
ciency of 0.16 V · cm for 25 Gb∕s operation using a 60-μm- with 3.8 dB ER comprised a symmetric PN junction.17 In this
long phase shifter with a loss of 3.5 dB∕mm.69 The MZM case, the modulator operated in a series of push–pull modes by
provided one of the best FoMs (α · V π · L) of <7 dB · V. connecting back-to-back the phase shifter PN junction em-
A microdisk-based SISCAP modulator demonstrated 15 Gb∕s bedded on both arms of the MZI. The series connection of
operation on a small footprint of ∼30 μm2 consuming only the two phase shifters halves the net capacitance, and the
55 fJ∕bit.70 In order to deliver >50 Gb∕s operation using push–pull scheme limits the chirp of the modulator. The modu-
SISCAP modulators, a parametric study has been performed lator shows 50 GHz bandwidth at 4 V reverse bias and an
validating the trade-off between high-speed operation, modula- average modulation efficiency of 3.2 V · cm for a 4-mm-long
tion efficiency, and loss.104 PN junction.17 In another implementation, a symmetric carrier
Depletion of electrons and holes by reversely biasing a PN depletion MZM with sub-1 V drive signal operated at 20 Gb∕s
junction built in or around it is the most widely adopted and delivered a power consumption as low as 200 fJ∕bit.118

Yet another variant of the symmetric lateral PN junction used shifter reported a high modulation efficiency of 0.62 V · cm
the doping compensating technique120 to demonstrate a 50-Gb∕s with a static IL of 2.8 dB while delivering 40 Gb∕s modulation
MZM comprising a 3-mm-long phase shifter with a modulation speed.81 An RM comprising a ring resonator with a 100-μm ra-
efficiency of 1.85 V · cm and a loss of 3.9 dB. An RM relying dius, and an MZM with a 0.95-mm phase shifter length operated
on a symmetric phase shifter configuration delivered a modula- at 40 Gb∕s speed using an interleaved phase shifter delivering
tion efficiency of 25 pm∕V at 0 V119 to demonstrate 36 fJ∕bit a modulation efficiency and loss of 2.4 V · cm and 2.1 dB∕mm,
energy consumption for 40 Gb∕s operation on a compact foot- respectively.32 Apart from standalone implementations of high-
print of 0.5 mm2 . A recent implementation demonstrates a speed phase shifters through these three baseline junction con-
64-Gb∕s MZM, in which the carrier depletion-based phase figurations and their adaptations, there are reports on junction
shifter is implemented through a 200-μm-long photonic crystal implementations that rely on combining these three schemes
slow light structure.114 The modulator used a meandered RF in a single PN junction design.9,33,79,85,86 An example is the sub-
electrode terminated with a 20 Ω resistor to provide a bandwidth strate-removed “zig-zag” phase shifter for an MZM delivering
of 38 GHz by preventing the velocity mismatch of the electrical 1.6 V · cm modulation efficiency for a 2-mm-long phase shifter
and optical signals. The modulator required a voltage swing of with 4.4 dB loss.14 The modulator provided 55 GHz bandwidth
5.5 V to provide an energy consumption of 21 pJ∕bit and pro- and modulation speed of 90 Gb∕s.14,87 Another example in-
vided 4.8 dB ER. The literature also reports various other dem- cludes a phase shifter for 40 Gb∕s operation of a racetrack
onstrations of asymmetric (where there is an offset of the PN RM by combining horizontal, vertical, and interleaved ap-
junction from the waveguide center) carrier depletion-based proaches. This implementation delivered a modulation effi-
phase shifters.8,16,72,74,76 For example, a Michelson interferometer ciency of 0.76 V · cm at −1 V reverse bias and a loss of
with a partially shifted PN-junction provided one of the highest 3.5 dB∕mm.33 A recent result demonstrated an RM with a rec-
modulation efficiencies of 0.72 V · cm at 1 V reverse bias and ord low modulation efficiency of 0.52 V · cm by employing a
4.7 dB loss for a 500-μm-long PN junction with a lumped driv- phase shifter comprising a combination of horizontal and ver-
ing scheme to operate at 40 Gb∕s through a bandwidth exten- tical junctions by wrapping the n-doped region over the intrinsic
sion scheme involving a resistor parallel to the PN junction.8 and the p-doped region.9 The modulator showed 50 GHz band-
The asymmetric PN diodes are also exploited by RRs72,76 to width and delivered 70 fJ∕bit energy consumption for 64 Gb∕s
demonstrate low-power (sub-100 fJ∕bit) and compact depletion operation.
modulators. The literature also reports fully shifted PN junc- In the discussion presented above, we have provided a sum-
tions, where a self-aligning process allows for the alignment of mary of the latest developments made for the improvement of
the n-doped region to the edge of the waveguide.16,74 In one im- plasma dispersion effect-based modulators. Figure 1(a) repre-
plementation of a fully shifted PN junction, a 50-Gb∕s MZM sents the three baseline plasma dispersion phase shifter architec-
modulator provided a modulation efficiency of 2.8 V · cm tures and their respective biasing configuration. Figure 1(b)
and an IL of 4 dB for a 1-mm-long phase shifter.16 By further gives an overview of a variety of other phase shifter configura-
optimizing the implementation of self-aligned PN junctions,74 a tions for carrier injection, carrier depletion, and carrier accumu-
demonstration provided an ∼1 dB∕mm loss without compro- lation phase shifters.
mising the speed of the modulator. The self-aligning process Table 2 summarizes the typical performance for the injection,
is further exploited to implement a PIPIN phase shifter with accumulation, and depletion-type plasma dispersion modula-
a modulation efficiency of 3.5 V · cm for a 0.95-mm-long phase tors. Table 2 also presents the best-reported results for the modu-
shifter introducing 4.5 dB optical loss when operated at lation efficiencies, losses, data rates, and energy consumption.
40 Gb∕s.79 Depletion-based modulators relying on an inter- There is an interplay of a vast pool of variables, such as junction
leaved phase shifter design have also delivered high-speed op- type and its actuation mechanism, doping concentrations (typ-
eration. A modulator comprising a 3-mm-long interleaved phase ically ranging from ∼1017 to ∼1018 cm−3 ) and their positions,
Table 2 Typical and state-of-the-art performance matrix for the plasma dispersion high-speed phase modulators. The parentheses
contain the best-reported result for a performance attribute. The matrix includes the results reported for O-band and C-band demon-
strations.
Modulation Lengtha of
efficiency phase shifter Data rateb
Principle V π · L (V · cm) Loss (dB/cm) (mm) (Gb/s) Energy/bit (fJ/bit)
Carrier injection c
<0.5 (0.058 ) 8
∼70 (28 ) 7
≥0.1 to <0.3 <40 (70 )7
∼1000 for MZMs and RMs (0.1f,98)
Carrier accumulation d
<0.3 (0.16 )69
50 to 80 (∼35 ) 69
≤0.5 ∼40 (40 )5
>200 for MZMs, <200 (3105) for SLMsg
Carrier depletione ∼2 (0.529) 10 to 30 (2.610) >1 >40 (100122,123) ∼200 for MZMs (32.419), <40 for RMs (0.918)
a
The phase shifter lengths are quoted for the non-resonant implementations.
b
The rates for an OOK modulation scheme. The data rates for the carrier injection scheme are presented for approaches that do not require pre-
emphasis of the modulating signal.
c
Typical performance for carrier-injection plasma dispersion phase modulators is based on Refs. 7, 8, 37, 61–67, and 88–100.
d
Typical performance for the carrier-accumulation plasma dispersion modulators is based on Refs. 5, 12, 68, 69, 69, 70, and 101–105.
e
Typical performance for the carrier-depletion plasma dispersion modulators is based on Refs. 8–11, 13–19, 22, 32, 33, 71–87, and 106–123.
f
Based on 1D-SiPh crystal.
g
For photonic crystal cavity SLM.

the thickness of the slab waveguide (typically ranging from of an applied electric field, the band edge tilts to enhance the
1/3 to 1/4 of the guiding silicon thickness), the thickness of absorption coefficient for GeSi bulk material (FK effect) or
the guiding silicon layer (typically <1 μm), choice of the Ge/SiGe quantum well (QW) structures (QCS effect).31,38,145,146
passive structure for the translation of phase modulation to These effects are intrinsically fast through their sub-ps response
amplitude modulation, and the design of the driving electrode time.34,229 The QCS effect is stronger than the FK effect134,230 due
(lumped, TW, and segmented), which can influence the to the strong excitonic effect and discretized density of states.
performance of plasma dispersion modulators. In many cases, The integration of bulk GeSi or Ge/SiGe QW requires epitaxial
one performance parameter has to be delicately traded-off growth on SiPh platforms. The FK effect is considered more
for another specification. This makes the design of a plasma attractive than the QCS effect in the context of integration
dispersion modulator, which can meet all of the performance feasibility. The latter requires a more complex epitaxial growth
requirements concurrently, far from trivial. Operation of process for band alignment and strain balancing.152,230 A large
plasma dispersion modulators at speeds >70 Gb∕s7,14,122 con- change in the absorption coefficient (typically ranging from
suming tens of fJ/bit with high modulation efficiency enabling 100 to 1000 cm−1 in the C-band146) enables the implementation
micrometer-sized footprints has been accomplished. Recent of a compact and energy-efficient EAM using the FK effect
developments have resulted in a co-designed SiPh modulator- and QCS effect.31,230 Both effects rely on band engineering.
driver offering 100 Gb∕s while preserving a good modulation Therefore, they have limited optical bandwidth (∼30 nm for
efficiency122 and without any customized fabrication process the FK effect and ∼20 nm for the QCS effect).134,149 In one
development. The plasma dispersion modulators deliver suffi- of the first SiPh EAMs, the L-band absorption spectrum is
cient linearity and ER to handle advanced coherent modulation shifted to the C-band by a tensiled 0.8% GeSi alloy.31 On a small
schemes,3,55,56,124–126 such as QAM, and amplitude modulation footprint of only 30 μm, the EAM actuated by a vertical PIN
schemes,9,17,29,52,66,115,123 such as PAM. Furthermore, the plasma diode provided an energy consumption of only 50 fJ∕bit.
dispersion modulators have shown modest chirp-induced This EAM could operate over a 14-nm bandwidth from 1539
penalty for short- and long-haul links.3,26,30 The co-design and to 1553 nm and delivered an efficiency (defined as Δα∕αon ,
co-optimization of the driver–modulator design is the key to where Δα is the absorption difference between ON-state and
taking the performance of plasma dispersion modulators to a OFF-state, and αon is the absorption for the ON-sate, in which
new level.122 no electric field is applied) of 2. Despite the limited bandwidth
of 1 GHz, this promising demonstration triggered other demon-
3 New Materials on Silicon for High-Speed strations of EAMs.20,38,128,133–139,231,232 An evanescently coupled
Modulators 55 μm2 GeSi EAM has been reported on a thick SOI platform
Recent years have seen growing efforts to integrate materials (silicon guiding layer thickness of 3 μm), actuated by a horizon-
with strong electro-optic effects on SiPh platforms.153–192,192–208,228 tal PIN junction.135 This EAM is butt-coupled to the thick SOI
The integration of new materials provides routes for optically waveguide through an appropriately designed taper.135 The
broadband and energy-efficient high-speed modulation with modulator has an efficiency of ∼1. This EAM operated in
little or no current flow.176,181,184,188,202 Furthermore, in some cases, the C-band and provided 28 Gb∕s operation with 60 fJ∕bit en-
these new materials provide an excellent decoupling between ergy consumption.135 Literature also reports an L-band Ge EAM
amplitude and phase modulation, which is difficult to achieve and a GeSi C-band EAM with an electrical 3-dB bandwidth of
with contemporary all-silicon plasma dispersion modulation ∼50 GHz for an optical bandwidth of 30 nm.136 In this case, the
schemes.174,178,184 EAM is butt-coupled to a silicon waveguide through a tapered
Through the wafer-scale integration of these new material waveguide section. The Ge or GeSi alloy, which is grown on
systems with SiPh platforms, one can achieve impressive modu- a silicon recess, has a lateral PIN diode. A highly doped contact
lator performance, while at the same time preserving the bene- on the silicon layer biases the EAM. The modulator consumed
fits provided by the mature SiPh technology.181 These new 12.8 fJ∕bit energy at modulation speeds of 56 Gb∕s and pro-
materials can be broadly categorized into the following: vided an FoM of <1. Using this GeSi EAM, a digital to
analog converter-less and digital signal processing (DSP)-free
• (silicon-)germanium, 100 Gb∕s transmission over a 500-m-long single-mode fiber
• ferroelectrics, using NRZ-OOK signaling has been reported.20,137
• III–V semiconductors, There are various reports on the implementation of the Ge/
• 2D materials, and SiGe QCS effect-based modulators.141–152 In one case, a Ge/SiGe
• organic (electro-optic) materials. QW modulator is butt-coupled to a thin SOI rib waveguide in
Early results of high-speed modulators by integrating electro- such a way that the QWs and the rib waveguide are vertically
optic materials with the SiPh platforms have shown a lot of aligned with each other to have maximum overlap.147 The
promise. Many research teams are attempting to take their active section of the modulator, which comprised 15 pairs of
maturity and performance beyond the performance of plasma Ge∕Si0.15 Ge0.85 QWs, is driven by a vertical PIN diode. The
dispersion modulators. The sections below discuss the key 8-μm2 modulator showed 3.5 GHz bandwidth by using a 1-V
modulator implementations by integrating these materials into voltage swing (5.5 to 6.5 V) to provide ∼3 dB of ER. In another
SiPh platforms. implementation, Ge QWs are epitaxially grown at temperatures
below 450°C on the silicon-rich Si1−x Gex low-loss waveguide
3.1 (Silicon-)Germanium platform.141 The x concentration is selected such that it simulta-
neously ensures low indirect bandgap absorption and a suffi-
The FK effect and the QCS effect provide possible routes to ciently small lattice mismatch with the QWs. The 400-μm2
implement compact, low-power, and high-speed electro-absorp- modulator with 6.3 GHz bandwidth and 0 to 3 V swing provided
tion modulators (EAMs) in SiPh.31,143,144,146,149 Under the influence ∼4 dB ER and 3 dB IL.

(a) (b)
Fig. 2 Representative cross sections of (a) FK-based EAMs and (b) QCS effect-based EAMs in
SiPh. This figure also highlights the integration route for the respective EAMs. The term monolithic
refers to the integration of a material with an SOI substrate using wafer-scale epitaxial growth
resulting in PICs made out of one substrate technology in mass on a wafer level.
The FK effect or QCS effect electro-absorption amplitude Figure 2 represents the typical cross sections of FK EAMs and
modulators allow for the implementation of coherent I∕Q231,232 QCS effect-based EAMs in SiPh. This figure also summarizes
and advanced amplitude modulation schemes such as PAM-4.233 the integration routes of these materials with SiPh. Table 3
These modulation schemes are a promising solution for short- summarizes the typical and the best-reported performance for
and long-reach links because they are spectrally efficient while the (silicon-)germanium modulators. This table covers the perfor-
being tolerant of optical fiber transmission impairments.54,55,231 mance of the FK effect-based EAMs as well as QCS effect-based
The implementation of an EAM-based I∕Q or PAM-4 modula- EAMs. The FoM for both FK and QCS effect-based EAMs is
tors relies on balanced MZIs with an EAM and a fixed phase- typically <2, although QCS effect-based modulators can poten-
shift in the arms of the MZI.231–233 Typically, an unequal splitting tially deliver better FoM due to their stronger excitonic absorption
of the optical carrier is required by the splitter of the MZI to peaks.144,149,152 In most cases, FK-effect-based EAMs operate either
achieve the appropriate eye-opening for PAM-4 or constellation in the L-band or C-band.150,152 It is possible to tune the operating
for the I∕Q signal.233 Literature reports an EAM-based PAM-4 wavelength of the QCS effect modulators by a bias voltage.150,152
modulator by integrating an EAM in each arm of a two-port Therefore, the QCS effect modulators can operate in the O-band
balanced MZI.231–233 In this modulator, a fixed phase shift is pro- as well. In terms of high-speed operation, FK effect-based EAMs
vided by embedding a thermal phase shifter in one of the two have reached line rates of up to 100 Gb∕s for OOK modulation
arms of the MZI. The device modulated a 128-Gb∕s PAM-4 schemes,137 whereas the QCS effect-based modulators have not
signal without requiring any DSP. Furthermore, there are reports yet demonstrated such a high speed of operation.
on the demonstrations of modulating QPSK and 16-QAM sig-
nals using EAM-based coherent modulators.231,232 The energy 3.2 Ferroelectrics
consumption of these modulators takes a hit due to the power-
hungry thermo-optic effect required to provide a fixed phase Ferroelectrics are materials possessing spontaneous polarization
shift in the arms of the MZIs. that can be altered by the application of a sufficiently strong
Table 3 Typical and state-of-the-art performance matrix for amplitude modulation by the FK effect and QCS effect. The best-reported
performance attributes are mentioned in parentheses.
Modulator Energy/bit
Principle FoM ER/IL Loss (dB) length (mm) Data ratea (Gb/s) (fJ/bit)
FK effectb <2 (231) <6 (4.8133) ∼0.050 >40 (100137) <50 (1320)
Quantum-confined Stark effectc <2 (7.9149) <6 (1.3149) <0.25 <10 (7147) <100 (16148)
a
The data rate is for the OOK modulation scheme. The highest reported data rate with QCS effect modulators is 7 Gb/s. The highest modulation
bandwidth has been reported to be 23 GHz.
b
Typical performance for the FK EAMs is based on Refs. 20, 31, 38, 128, and 133–139.
c
Typical performance for the QCS EAMs is based on Refs. 141–152.

electric field.234 As ferroelectric materials are non-centrosym- lane 100 Gb∕s NRZ-OOK transmission. Unlike the other re-
metric, they exhibit a Pockels effect.234 Many ferroelectrics ported implementation,173 the integration of LiNbO3 with the
have large Pockels coefficients compared to non-ferroelectric SiPh PIC was carried out using BCB-bonding. Monolithic in-
inorganic materials. The dominant contribution to the Pockels tegration of LiNbO3 with SiPh PICs has yet to be reported.
coefficients in these ferroelectrics tends to come from lattice
vibrations rather than electron movements.234,235 Because of 3.2.2 BaTiO3 on silicon high-speed modulators
their strong Pockels effect, the integration of ferroelectric ma-
In recent years, there has been a lot of interest in the integration
terials such as LiNbO3 ,167–174 barium titanate (BaTiO3 ),175–182
of BaTiO3 with SiPh.175–182 BaTiO3 is a promising material
and lead zirconate titanate (PZT)183–185 with SiPh provides a
for integration with SiPh because of (a) its high Pockels coef-
promising route for the implementation of efficient, high-speed
ficient [in bulk form, LiNbO3 has a Pockels coefficient of
modulators.2,175,228 A variety of integration techniques have been
∼30 pm∕V,27,175,249 whereas BTO offers one of the highest
explored, such as molecular beam epitaxy (MBE),176,178 pulsed
Pockels coefficients (>1000 pm∕V) in bulk form176,251]; (b) fea-
laser deposition,236,237 chemical solution deposition (CSD),238 RF
sibility of integration with a silicon substrate;252 (c) capability to
magnetron sputtering,239 metal organic chemical vapor deposi-
offer high-speed modulation;253 (d) high chemical and thermal
tion,240 and atomic layer deposition.241 The two key requirements
stability of BTO; and (e) feasibility of low-loss hybrid BTO-Si
for integrating ferroelectric materials with SiPh are to ensure
passive circuitry.178,179 The integration of BTO with SiPh relies
a high electro-optic coefficient and low propagation loss.182
on epitaxial growth enabling integration compatible with 200
The following sections will review the latest developments to
and 300 mm silicon wafers.181 The large lattice mismatch
implement high-speed modulators by integrating ferroelectric
(∼27%) between BTO and Si makes the integration of the
materials with SiPh.
two materials far from trivial.254 A SrTiO3 (STO) buffer layer of
a couple of nanometers thickness reduces this lattice mismatch
3.2.1 LiNbO3 on silicon (nitride) high-speed modulators to <2%.228 Among the various possible routes to grow BTO
LiNbO3 is a widely used material for modulators and provides on SiPh platforms, MBE is known to provide the highest
high-speed devices with a bandwidth of >35 GHz.27,242,243 (>900 pm∕V) Pockels coefficient in BTO thin films.176 In one
LiNbO3 modulators have been used for a long time in fiber- approach, the BTO growth is done directly on an SOI wafer.177
optic networks.242,244 These devices are based on low-index- Afterward, an amorphous silicon layer with a thickness equal to
contrast waveguides in bulk LiNbO3 ,242 in which the index the thickness of the c-Si layer of the SOI wafer is grown. The
contrast between the core and cladding of a LiNbO3 waveguide resulting heterostructure is patterned lithographically to form a
is provided by the diffusion of Ti or Li ions.242 The weak optical horizontal slot waveguide. Stronger guiding of the optical signal
guiding resulting from the low-index-contrast requires the in the low-index BTO layer (nBTO ∼ 2.38) enhances the overlap
electrical contacts to be placed far away from the waveguides, of the optical and electrical field in the BTO layer. This
resulting in typical V π · L of 10 to 20 V · cm.242 Recent demon- approach provided a Pockels coefficient of 213 49 pm∕V.177
strations of high-index-contrast LiNbO3 -on-insulator (LNOI) A ring-based modulator with a modulation efficiency of
modulators have shown energy-efficient operation (a few tens of 1.5 V · cm and a bandwidth of 4.9 GHz has been reported, where
fJ/bit energy consumption) with 100-GHz bandwidth, ∼2 V · cm the limited bandwidth is associated with the parasitic electrical
modulation efficiency, and <0.5 dB loss for a device that is effects.177 A key challenge faced by the integration approach
a few millimeters long (typically 2 to 20 mm long).243,245–247 mentioned in Ref. 177 and other integration approaches176,178
These thin-film LNOI modulators rely on LiNbO3 layers of is the high propagation loss (>40 dB∕cm).176–178,255 The origin
a few hundred nanometers thickness, which are bonded on of this high loss is attributed to the absorption of hydrogen
top of SiO2 .243 Several approaches have been reported in the by the STO seed layer during the fabrication of the slot
last two decades to integrate LiNbO3 with silicon at room waveguide.178 Furthermore, it has been shown that a loss of
temperature245,248,249 and to demonstrate high-speed modulation 6 dB∕cm in a strip-loaded horizontal slot waveguide can be
by bonding LiNbO3 with silicon167–169,250 or silicon nitride.170–172 achieved by low-temperature annealing.178 Apart from the afore-
The Pockels coefficient of integrated LiNbO3 is comparable to mentioned approach, where BTO is grown directly on an SOI
that of bulk LiNbO3 ,249 but most of these integration attempts wafer,177 another possible approach entails the formation of a
have resulted in modulator bandwidths that are smaller than the BTO/Si heterostructure by bonding an SOI wafer with a BTO
all-silicon modulator implementations.173 Recently, a tremen- layer grown on a host SOI wafer.176,179 For such a bonded
dous boost in the speeds (not yet the overall performance) of BTO/Si heterostructure, a Pockels coefficient of 923 pm∕V has
such modulators has been reported.173,174 For example, for a been reported (30 times larger than that of a LiNbO3 struc-
5-mm long TW-MZM that was realized by direct oxide bonding ture).179 In one implementation, a plasmonic MZM with a
of thin-film LiNbO3 on top of a silicon waveguide, a bandwidth 10-μm-long BTO/Si phase shifter with 25 dB loss delivered
of 100 GHz and a loss of 7.6 dB have been reported.173 The 72 Gb∕s and required a V π of 14.5 V (modulation efficiency
modulator had a limited efficiency of 6.7 V · cm due to <100% of 0.0145 V · cm).180 The modulator operated successfully for
vertical coupling of light into the unpatterned LiNbO3 mem- temperatures up to 130°C. Furthermore, a wafer-level bonding
brane on top of the silicon waveguide. In another implementa- of Si/BTO heterostructures on the back-end-of-line (BEOL) di-
tion, the complete coupling of light from a silicon waveguide to electric layer of an advanced SiPh platform has been reported.181
a patterned LiNbO3 rib waveguide through a vertical adiabatic This integration led to the demonstration of an MZM with
coupler resulted in a modulation efficiency of ∼2.5 V · cm.174 a limited bandwidth of 2 GHz (mostly limited by the electrical
With a 2.5-dB loss, the 5-mm-long LiNbO3 phase shifter em- bandwidth of the electrodes).181 The modulator had a modula-
bedded in an MZI and driven by a TW electrode in the push–pull tion efficiency of 0.2 V · cm and loss of <6 dB for a 1-mm-long
configuration provided 70 GHz bandwidth and enabled single- phase shifter.181,182

3.2.3 PZT on silicon nitride high-speed modulators speeds of ∼100 Gb∕s. In terms of modulation efficiency, BTO-
PZT is an orthorhombic lead-containing ceramic ferroelectric based modulators have shown a lot of promise with typical
material.256,257 Though BTO has a large Pockels coefficient, reported values of <0.5 V · cm.
its use in SiPh requires complex and expensive integration
technology.178,181,182,255 As compared to BTO, PZT has a smaller 3.3 Organic Electro-Optic Materials
Pockels coefficient, but it can be deposited by a low-cost and
simple CSD method.185,257 Ultra-thin lanthanide-based inter- Organic electro-optic materials are typically based on π-conju-
mediate seed layers enable the deposition of high-quality thin gated molecules with strong electron donor and acceptor groups
(up to 200 nm) PZT layers on silicon,257 with Pockels coeffi- at the ends of the π-conjugated structure.188,258 The delocalized
cients up to 150 pm∕V.228 PZT thin films have been integrated electrons associated with this π-conjugated structure are con-
with silicon nitride photonic circuits. An RM with a bandwidth fined by a strongly asymmetric potential well, giving rise to
of 33 GHz and an MZM with a bandwidth of 27 GHz have been large Pockels coefficients.188,258 In contrast to the ferroelectrics
demonstrated,184 with a modulation efficiency of 3.2 V · cm and discussed above, the Pockels effect in these organic electro-
a loss of 1 dB∕cm. The PZT layer was deposited on a planarized optic materials is mostly electronic in nature and can thus
silicon nitride PIC (i.e., the PIC consisted of silicon nitride strip provide an ultra-fast response.188,190,228,258 Organic electro-optic
waveguides with oxide bottom and side cladding, while the top materials can be processed at room temperature using in-
cladding was air). The PZT thin film was poled by applying expensive solution-based techniques, whereas the growth of
a voltage of 60 to 80 V for 1 h.185 inorganic Pockels materials typically requires high temperatures
Figure 3 represents the typical cross sections of ferroelectric and/or the use of vacuum equipment.177,182,188,255,258
modulators in SiPh. This figure also summarizes the integration A combination of organic materials with SiPh devices
routes of the respective materials with SiPh. Table 4 presents has resulted in the so-called silicon-organic hybrid (SOH)
a summary of the typical and the best-reported modulation class of modulators.186–188,197 The underlying organic materials
efficiencies, losses, data rates, and energy consumption for in SOH modulators are designed to provide high electro-
the ferroelectric-based modulators in SiPh. For all three types optic activity with experimentally demonstrated values of
(i.e., LiNbO3 , BTO, and PZT), modulation speeds ≥40 Gb∕s n3 r > 2000 pm∕V.190 This results in modulators with a low-
have been demonstrated. LiNbO3 -based modulators are out- loss,194 a low-chirp parameter of α ¼ −0.02,193 and a high
standing in terms of low-loss operation and have delivered modulation efficiency (V π · L ≈ 0.03 V · cm),190 along with a
(a) (b)
(c) (d)
Fig. 3 Representative cross sections of (a) LiNbO3 , (b) BTO, (c) PZT, and (d) organics modulators
in SiPh. This figure also highlights the integration route for each material. The term monolithic
refers to the integration of a material with a SiPh substrate using wafer-scale epitaxial growth
resulting in PICs made out of one substrate technology in mass on a wafer level.

Table 4 Typical and state-of-the-art performance matrix for the ferroelectric and organic high-speed phase modulators in SiPh.a The
parentheses contain the best-reported result for a performance attribute.
Modulation efficiency Loss Length of phase Data rateb

Scheme V π · L (V · cm) (dB/cm) shifter (mm) (Gb/s) Energy/bit (fJ/bit)
LiNbO3 integrated with SiPh c
<3 (2.2 )174
∼1 (0.98 ) 174
>1 >70 (100 )174
>100 (170174)
d 182 182 180
BaTiO3 integrated with SiPh <0.5 (0.2 ) >40 (6 ) >1 >40 (72 ) <100179
PZT integrated with SiPhe 1 (1183) 1 (1183) >2 40 (40184) -
f 190 228
Organics integrated with SiPh <0.05 (0.032 ) <45 (20 ) <1.5 >50 (100196) <100 (0.7192)
a
Plasmonic modulators are not included in this table.
b
The data rates are reported for the OOK modulation scheme.
c
Typical performance for LiNbO3 integrated with SiPh is based on Refs. 167–174 and 228.
d
Typical performance for BaTiO3 integrated with SiPh is based on Refs. 175–182 and 228.
e
Typical performance for PZT integrated with SiPh is based on Refs. 183–185 and 228.
f
Typical performance for the organics integrated with SiPh is based on Refs. 186–192, 192–200, and 228.
high modulation bandwidth (>100 GHz)191,259 and low-energy doped silicon slabs that provide an electrical contact with the
consumption of a few fJ/bit.192,228 With π-voltages below 1 V, rails of the slot waveguide.190 These bandwidths may be im-
SOH modulators may be directly driven by standard CMOS proved to more than 100 GHz by exploiting optimized wave-
circuits without the need for any additional drive amplifiers,260 guide designs that rely on multi-step doping profiles.259 To
thereby opening a path toward greatly reduced power dissipa- that end, doped silicon slabs have also been replaced by
tion in highly integrated transceivers. For the fabrication of BTO to feed the modulating RF signal to the EO material in
SOH devices, the organic electro-optic material is deposited the slot waveguide.200 This so-called capacitively coupled
onto fully processed SiPh chips in a postprocessing step, which SOH MZM delivered a bandwidth of 65 GHz, which was suf-
requires removal of the BEOL layer stack of a standard SiPh ficient to produce a 100-Gb∕s OOK signal.200 The modulator
process flow.228 SOH modulators rely on slot waveguides,188,191 had a modulation efficiency of 0.13 V · cm; the loss from the
which guarantees enhanced interaction between the highly con- phase shifter section is not reported. Yet another variant of the
fined optical mode in the slot waveguide and the organic electro- SOH modulator is the plasmonic-organic hybrid (POH) modu-
optic material filling the slot. To pole the electro-optic material lator,258,264–269 where the silicon slot waveguide is replaced by a
after deposition, the devices are heated, and a DC voltage is ap- metal plasmonic slot waveguide to provide even higher optical
plied across the slot waveguides. This non-volatile poling trans- confinement.258,265 The deposition of an organic electro-optic
forms the random orientation of the electro-optic chromophores material on the metal slot waveguide leads to compact high-
in the slot into an acentric orientation, thereby leading to a non- speed devices with bandwidths of hundreds of gigahertz.266,270
zero Pockels coefficient. After cooling down the device to am- POH modulators are compact with typical lengths of a few
bient temperature, the voltage can be removed, and the poled microns and offer a low-energy consumption of a few tens of
electro-optic material preserves its orientation. SOH modulators fJ/bit along with ultra-low voltage-length products of typically
have shown exemplary performance over the years. Some key ∼0.005 V · cm258—nearly an order of magnitude lower than
results are reported in Refs. 189–191, 193, 196, and 261. The those of classical SOH modulators. The obvious limitation
main skepticism faced by the SOH modulators comes from their of the POH modulators is the high propagation loss in the
long-term photochemical and photothermal stability. Recent re- plasmonic slot-waveguide section with typical values of
search results reported in Refs. 195 and 262 have shown long- 200 dB∕mm.200,258 Despite the high efficiencies, this leads to
term stability of SOH devices following Telecordia standards. substantial π-voltage-loss-products of the order of 10 V · dB.188
These early experiments have shown that in storing the SOH Table 4 provides a summary of typical modulation efficien-
modulators at elevated temperatures of 85°C, the degradation cies, losses, speeds, and energy consumption for organic
of the electro-optic activity converges to a stable level where electro-optic modulators in SiPh. When compared with the
the V π · L-product increases by <15% after 2400 h. other Pockels modulators, the organic electro-optic modulators
The limitations of poled organic materials can be overcome provide the best modulation efficiency with a reported value
by substituting them with crystalline organic materials, which of 0.032 V · cm.190 Organic electro-optic modulators have
are inherently resilient to higher temperatures with good shown up to 100 Gb∕s196 operation and sub-fJ/bit energy
photochemical stability and do not require any poling.199 The consumption.192 Figure 3(d) represents the typical cross section
electro-optic coefficients of the organic crystals are smaller of organics modulators in SiPh.
(typically n3 r < 500 pm∕V)263 than those of the poled organic
polymers.190 The integration of organic crystals requires growth 3.4 III–V Semiconductors
to functionalize the silicon waveguides.198,199 An N-benzyl-2-
methyl-4-nitroaniline crystal has been deposited on an SOI The integration of III–V semiconductors with SiPh platforms is
waveguide to demonstrate an MZM that could operate at a well-established research field, as it provides a viable route for
12.5 Gb∕s with a modulation efficiency of 1.2 V · cm and a loss the integration of laser sources with SiPh PICs.271–275 Direct
of 16 dB in the phase modulator with a 1.5 mm length.199 bonding,272 BCB-aided adhesive bonding,271 III–V/Si flip-
Typically, the analogue bandwidth of current SOH modula- chip bonding,275 and transfer printing274 are some of the most
tors is limited to ∼25 GHz by the finite conductivity of the established bonding mechanisms for III–V/Si integration.

Demonstrations of III–V integration with advanced SiPh plat- of III–V on Si MOS modulators is not limited to MZIs
forms have been reported.276,277 Demonstrations of epitaxial only.164 By implementing a III–V on Si MOS modulator in a
growth of III–V on Si have also been reported despite the large racetrack configuration (implementing III–V on Si MOS mod-
lattice mismatch between Si and III–V (for example, InP has ulators in a ring is challenging due to the long taper length
8% lattice mismatch with Si).278,279 The integration of III–V needed for the III–V on Si hybrid waveguide), the energy
on silicon also provides a route for the implementation of effi- consumption of a 40-GHz (3-dB bandwidth) modulator has
cient high-speed modulators because they provide (a) a large been improved by a factor of 3.7 as compared to the III–V
electron-induced refractive-index change, (b) a high electron on Si MZMs, while delivering a modulation efficiency of
mobility, and (c) a low carrier-plasma absorption.166 0.064 V · cm.164 The 100-μm-long phase shifter section of the
The reported implementations of III–V/Si modulators typi- modulators comprised a thin aluminum oxide gate layer sepa-
cally use a MOS (SISCAP) modulator architecture.161–163 rating the n-type III–V and the p-doped Si layer. Recently,
In these modulators, n-doped III–V material, for example, a proof-of-concept, taper-less integration of a few tens of
InGaAsP, is bonded to a p-doped Si layer with an insulating nanometers thick III–V layer with Si has been demonstrated
layer separating the two.161–163 This results in a SISCAP archi- to reduce the footprint and loss of the device281 without compro-
tecture, where the insulating layer acts as a capacitor. A change mising the modulation efficiency of the III–V on Si MOS
in the refractive index takes place due to the charge accumula- capacitor.
tion due to the biasing of the doped silicon and III–V layer. Apart from phase modulators, EAMs have been reported by
According to Drude’s model, the plasma dispersion induced the bonding of III–V QWs with silicon.140,282 These EAMs use
refractive index change −Δn is inversely proportional to the ef- a PIN architecture, where the application of a bias voltage
fective masses of the electrons and holes.61,89 As compared to the changes the optical absorption of the intrinsic multiple QWs
electron mass of Si, the lighter electron mass of n-doped III–V due to the QCS effect.282 A segmented electrode, comprising
results in a 17 times stronger electron-induced refractive index a low-impedance active modulation section cascaded by a
change.161 Furthermore, an n-doped III–V provides a larger ratio high-impedance passive section, provides high-bandwidth oper-
of −Δn to Δk, where Δk is the change in the absorption coef- ation while ensuring low reflection and good modulation
ficient,161 resulting in a phase modulator with reduced spurious efficiency.140 A >67 GHz bandwidth hybrid III–V on silicon
intensity modulation (“pure-phase” modulation). Similarly, EAM has been demonstrated using this scheme.140 In the
p-doped silicon provides larger −Δn∕Δk than p-doped III–Vs O-band, this modulator has an IL of 4.9 dB, and it shows
for carrier concentrations below 1019 cm−3.161–163 Therefore, a >9 dB of dynamic ER for 50 Gb∕s operation over a 16-km-
hybrid III–V/Si MOS modulator brings the best of III–V and long optical fiber link.
Si together. Consequently, this approach offers modulators with Figure 4 represents the typical cross sections of III–V on Si
a 3 to 5 times better modulation efficiency compared to all- phase modulators and EAMs. This figure also summarizes
silicon MOS modulators and a phase shifter loss that is signifi- the routes to integrate III–V on Si. Table 5 presents a summary
cantly lower than in all-silicon MOS modulators.161,163 of the typical and best-reported values for the modulation effi-
In one report, an InGaAsP∕Al2 O3 ∕Si MOS structure pro- ciencies, losses, data rates, and energy consumption for forward
vided a modulation efficiency of 0.047 V · cm and a phase and reverse biased III–V on Si phase modulators as well as for
shifter loss of ∼19.4 dB∕cm.161 In another report, an InGaAsP∕ III–V on Si EAMs. III–V on Si modulators have shown an
SiO2 ∕Si MOS structure provided163 a modulation efficiency of impressive modulation efficiency and low-loss operation, but
0.09 V · cm with an absorption loss of 26 dB∕cm in the phase so far experimental demonstrations of high-speed operation
shifter. Both demonstrations show an impressive modulation ef- (∼100 Gb∕s) remain elusive.
ficiency. But, the electro-optic cut-off frequency is rather limited
due to the large oxide capacitance in the carrier accumulation 3.5 2D Materials
mode of operation.161,163
One possible approach to break the trade-off between the Atomically thick 2D materials, the most prominent graphene,
high-speed operation and high modulation efficiency banks have received considerable interest in the last decade to
on operating the III–V on Si MOS phase shifter in reverse implement efficient modulators by integration with Si/SiN
bias.165,280 The III–V on Si MOS modulator that is operated in PICs.1,153–160,201–208 The implementation of graphene-based mod-
reverse bias is architecturally similar to the forward biased ulators in SiPh involves the transfer of the graphene layer on
III–V on Si modulator.161,163,165,280 For a given thickness of the a Si or SiN waveguide.1 Mature and wafer-scale chemical vapor
oxide layer, the reverse biased III–V on Si phase shifter has a deposition is the widely adopted scheme to deposit graphene on
reduced depletion capacitance as compared to the accumulation a variety of metallic and dielectric substrates.283–285 The transfer
capacitance.165,280 The combined plasma dispersion effect and of the graphene layer on a Si or SiN waveguide is performed
FK effect in the reverse biased III–V on Si MOS phase shifter after delaminating graphene from the substrate while adding
ensure a modulation efficiency that is comparable to that in a polymer carrier to handle the graphene film.154,159 Recently,
the accumulation mode while ensuring a low-loss operation microtransfer printing technology has shown promising results
due to the limited hole density in the n-doped InGaAsP layer to transfer graphene to the target substrate in an automatic
of the modulator.165,280 A III–V on Si MOS modulator with a fashion.159 A single graphene layer, which is only a fraction
250-μm-long phase shifter and a 9-nm-thick oxide layer has of a nanometer thick, can absorb 2.3% of light286 with a wave-
been reported with a modulation efficiency of 0.11 V · cm.165 length ranging from the visible to THz wavelengths (graphene is
As compared to the carrier accumulation MOS, the reverse a gapless material). Furthermore, graphene has a very high elec-
biased MOS provides a 3.3× lower depletion capacitance, hence tron mobility of >100,000 cm2 · V−1 · s−1 (silicon has a carrier
paving the way for a 3.3× improvement in modulation band- mobility of 1400 cm2 · V−1 · s−1 ).228,287,288 These two properties
width and energy consumption per bit. The implementation enable the implementation of high-speed electro-absorptive153 or

(a) (b)
Fig. 4 Representative cross sections of (a) III–V on Si phase modulator and (b) III–V on Si EAM.
Table 5 Typical and state-of-the-art performance matrix for the III–V on Si high-speed modulators in SiPh. The parentheses contain
the best-reported result for a performance attribute.
Modulation efficiency Loss Length of phase Data ratea Energy/bit

Operation V π · L (V · cm) (dB/cm) shifter (mm) (Gb/s) (fJ/bit)
Forward biased III–V on Si as a phase <0.1 (0.047161) <30 (19.4161) <0.5 >30 (32163,b) ≥100
modulatord
Reverse biased III–V on Si as a phase <0.2 (0.11164) (28280) <0.5 200c,280 ≤100280
modulatord
Operation FoM ER/IL Loss (dB) Modulator length Data ratea Energy/bit
(mm) (Gb/s) (fJ/bit)
III–V on Si as an amplitude modulatord ∼2 (2140) (4.9140) ≤0.1 ∼50 (50)140 -

a
b
The value represents the highest reported 3-dB bandwidth of 40 GHz for a racetrack resonator modulator in III–V on Si.
c
The value represents the theoretically predicted value for the 3-dB bandwidth in GHz.
d
The typical performance of III–V on Si high-speed modulators is based on Refs. 140, 161–166, and 282.
electro-refractive201 modulators through the integration of gra- results in modulation of light propagating in the waveguide.
phene with SiPh ICs. The doped Si waveguide in the single-layer graphene modulator
The two prominent types of graphene modulators in SiPh is a source of loss, and it requires an implantation step.153,156
comprise a SiPh rib waveguide, which is either loaded with a Double-layer graphene modulators remedy these limitations.156
single layer or a double layer of graphene153,156 (see Fig. 5). They are separated by a dielectric layer that resides on top of
The operating principle of these modulators relies on the charg- a completely passive waveguide. (This architecture does not
ing and discharging of a capacitor. A single-layer graphene require any doping of the waveguide slab.)156 The first double-
modulator comprises a graphene loaded Si rib waveguide with layer graphene modulator reported 1 GHz bandwidth and an
a doped slab region.153 A few nm thick dielectric spacer, such as electroabsorption modulation of 0.16 dB∕μm.156 The 40-μm-
SiO2 , Al2 O3 , or hBN, separates the waveguide and the graphene long modulator provided a 6-dB ER and required a 5-V
layer to form a capacitor.1 The first reported graphene driving voltage. One of the first multi-Gb/s graphene-silicon
modulator153 had a 4-dB ER and relied on a single-layer ap- modulators157 relied on a single-layer graphene architecture
proach. The 25-μm2 EAM provided 1 GHz operation, limited that provided a 5.2-dB ER for a 50-μm-long EAM operating
by the RC time constant, and an electroabsorption modulation at ∼2.5 V. At 10 Gb∕s, the modulator provided a modulation
of 0.1 dB∕μm.153 The absorption in graphene is dependent on efficiency of 1.5 dB∕V. The ON state loss is 3.8 dB with a
the bias voltage.1 This voltage tunes the Fermi level of graphene power consumption of about 350 fJ∕bit. Another result reports
to enable or inhibit the interband transitions by excitation of a double-layer graphene RM158 using a critical coupling effect in
electrons through the incident photons.201,289 This change in a 40-μm SiN ring. In this implementation, a 50-nm thick Al2 O3
absorption due to carrier accumulation in the graphene sheet layer separates the two graphene layers. By biasing, the change

(a) (b)
Fig. 5 Representative cross sections of (a) single-layer graphene phase/amplitude modulator and
(b) double-layer graphene amplitude/phase modulator in SiPh.
in the transparency of the graphene capacitor led to the modu- disulfide (MoS2 ) have also been integrated with SiPh. Results
lation of the optical signal. The reduced capacitance resulting show phase change relative to the change in absorption for these
from the thick spacer between the graphene layers led to a materials at telecom wavelengths when integrated with SiN.204
modulator showing 30 GHz bandwidth with a modulation Figure 5 represents typical cross sections of a single layer
efficiency of 1.5 dB∕V but required a −30-V DC bias and graphene modulator (a) and a double layer graphene modulator
7.5V p−p .158 The modulator could transmit 22 Gb∕s while con- (b) in SiPh. Table 6 provides the current state of high-speed
suming 800 fJ∕bit. A positive chirp of a graphene EAM is phase and amplitude modulators in SiPh by integrating 2D ma-
an interesting property.160 This feature is useful in compensating terials. The currently reported performances of 2D modulators,
for the chromatic dispersion in an optical transmission link for in particular graphene modulators, are mostly limited by tech-
the transmission of 10-Gb∕s without any optical signal to noise nological immaturity and not by conceptual limits. Graphene
ratio penalty.160 requires more development to reach the best performances.
Apart from amplitude modulation, it is also feasible to use More specifically, graphene needs to preserve the high carrier
graphene for phase modulation.201–208 Phase modulation is more mobility after the integration in the photonic circuit. This pres-
exciting for the encoding of complex modulation formats such ervation is not the case yet. By achieving a high level of tech-
as QPSK. A single-layer graphene architecture, in which the nological maturity, graphene has the potential to provide devices
graphene layer is separated from the silicon waveguide by a thin that can work over a wide range of wavelengths due to its gap-
dielectric spacer layer, acts as a phase modulator by shifting the less band structure, miniaturized devices due to its large EO
Fermi level of graphene.201,205,207,290,291 A single-layer graphene efficiency, feasibility for BEOL integration with the SiPh fab-
phase modulator of 300 μm length is integrated on a Si MZI rication process, and a possibility to offer a complete platform
and showed a modulation efficiency of 0.28 V · cm. The modu- for communication applications, as graphene possesses proper-
lator delivered an ER of 35 dB, an EO bandwidth of 5 GHz, and ties for efficient modulation and detection.
a loss of 236 dB∕cm.201 With a 2-V p−p drive, the transmission of
10 Gb∕s over a 50-km fiber span is demonstrated. The higher 4 Summary and Outlook
loss in the modulator is associated with the suboptimal transfer Years of academic and industrial R&D have enabled a tremen-
of graphene.201 The quality of graphene, which is typically dous boost in the performance of plasma dispersion phase mod-
determined by the scattering time and is related to the carrier ulators. Operation at speeds of 100 Gb∕s with high modulation
mobility, plays a detrimental role in the performance of gra- efficiency is now feasible. Plasma dispersion modulators can
phene modulators.1 Typically, scattering time of >100 fs is re- operate with a sub-1-V drive signal consuming a few tens of
quired to implement energy-efficient graphene modulators with fJ/bit in a μm-sized footprint. Demonstrations of advanced
a high ER.1 Achieving 10,000 cm2 · V−1 · s−1 graphene carrier modulation schemes such as PAM-4, PAM-8, QPSK, 16-QAM,
mobility, which roughly corresponds to 300 fs scattering time, and 32-QAM have been reported in plasma dispersion modula-
upon the complete modulator integration process would be tors. The modulators have an insignificant chirp-induced
ideal. Accomplishing this would lead to extremely low IL.289 penalty for long- and short-haul links. A closer synergy through
In perspective, the combination of low IL in a dual-layer co-design and co-optimization of electronic–photonic modula-
graphene structure, the high graphene electro-absorption, or tor design is the key to take plasma dispersion modulators
electro-refractive efficiency would lead to a modulator with a beyond 100 Gb∕s operation in an efficient manner. The last five
good FoM.1 It is important to mention that graphene can be used years have seen a new paradigm in achieving higher speed and
for detection,1 and this makes graphene suitable for a telecom- lower loss operations of SiPh modulators by integrating materi-
munication platform on a passive integrated circuit operating in als such as ferroelectrics, organics, SiGe alloys, III–Vs, and 2D
several telecommunications bands. materials. Unlike plasma dispersion modulators, which are com-
Inspired by the promising results shown by graphene, other mercially deployed and have a high manufacturing readiness
2D materials such as tungsten disulfide (WS2 ) and molybdenum level (MRL) of >8, these new modulators are still in the early

Table 6 Typical and state-of-the-art performance matrix for graphene on Si high-speed modulators in SiPh. The parentheses contain
the best-reported result for a performance attribute.
Modulation efficiency Length of phase Data ratea

Operation V π · L (V · cm) Loss (dB/cm) shifter (mm) (Gb/s) Energy/bit (fJ/bit)
2D Materials on Si as <0.5 (0.28 ) 201
>200 (236 ) 201
<0.5 d
∼10 (10 )
201
∼1000 (600)1) for MZM;
phase modulatorb <500 for RM
Operation FoM ER/IL Loss (dB) Modulator Length Data ratea (Gb/s) Energy/bit (fJ/bit)
(mm)
2D Materials on Si as <3 (4.9155) <4 (0.9203) <0.15 >10 (50154) ∼100 (40202)
amplitude modulatorc
a
b
The typical performance for 2D materials-based SiPh phase modulators is based on Refs. 201–208.
c
The typical performance for 2D materials-based SiPh amplitude modulators is based on Refs. 153–160.
d
The high IL is the result of the graphene deterioration upon fabrication. This value can be dramatically reduced with a value >5000 cm2 · V−1 · s−1 for
graphene carrier mobility at the end of the fabrication process.
R&D stage (MRL <7) and are currently not included in any fabrication process. For this reason, the optimized solution will
commercial product. depend on the application.
Electro-absorption modulation by (silicon-)germanium FK
or QCS devices allows for micrometer-sized footprints. FK-EA Acknowledgments
modulation provides 100 Gb∕s operation, but these modulators
have a limited ER as well as limited operating optical band- The authors would like to thank Prof. Christian Koos, Prof.
width. QCS-EA modulation has integration challenges with Marco Romagnoli, Prof. Delphine Marris-Morini, and Ashwyn
SiPh. Moreover, their operating speed is <10 Gb∕s. LiNbO3 , Srinivasan for their valuable comments. This research is parti-
BaTiO3 , PZT, graphene, and organic thin films on Si PICs offer ally funded by the ERC starting grant ELECTRIC.
high Pockels coefficients, paving the way for linear and chirp-
free high-performance SiPh modulators. Early results reported References
in the literature indicate that promising performance is achiev- 1. M. Romagnoli et al., “Graphene-based integrated photonics for
able by integrating thin-film LiNbO3 , BaTiO3 , or PZT materials next-generation datacom and telecom,” Nat. Rev. Mater. 3(10),
on Si. The integration of organic materials on Si has also shown 392–414 (2018).
100 Gb∕s operation and impressive modulation efficiency. 2. J. Witzens, “High-speed silicon photonics modulators,” Proc.
Progress has been reported to address the lifetime issues, but IEEE 106(12), 2158–2182 (2018).
more development is necessary before the commercial launch 3. B. Milivojevic et al., “112 Gb∕s DP-QPSK transmission over
2427-km SSMF using small-size silicon photonic IQ modulator
of high-speed modulators relying on organic electro-optic ma- and low-power CMOS driver,” in Opt. Fiber Commun. Conf. and
terials. Bringing together III–V and Si brings the best out of Exposition and Natl. Fiber Opt. Eng. Conf., OTh1D.1 (2013).
III–V and Si to implement MOS-type phase modulators and 4. S. S. Azadeh et al., “Low V π silicon photonics modulators with
EAMs. The product of V π and phase shifter loss is impressive highly linear epitaxially grown phase shifters,” Opt. Express
for the III–V on Si phase modulators, but high-speed operation 23(18), 23526–23550 (2015).
remains elusive. (Theoretically, they have shown impressive 5. M. Webster et al., “Low-power MOS-capacitor based silicon
results.) Integrating 2D materials, in particular graphene, with photonic modulators and CMOS drivers,” in Opt. Fiber
SiPh has the potential to provide a complete solution for modu- Commun. Conf. and Exhib., W4H.3 (2015).
lation and detection at high speed. Multi-Gb/s operation has 6. A. Narasimha et al., “A 40-Gb∕s QSFP optoelectronic transceiver
in a 0.13-μm CMOS silicon-on-insulator technology,” in Conf.
been reported for the graphene phase and amplitude modulators. Opt. Fiber Commun./Natl. Fiber Opt. Eng. Conf., OMK7 (2008).
A technique to preserve the high carrier mobility after integra- 7. Y. Sobu et al., “70 Gbaud operation of all-silicon Mach–Zehnder
tion in the photonic circuit is essential to unleash the true po- modulator based on forward-biased PIN diodes and passive
tential of graphene modulators in SiPh. equalizer,” in 24th OptoElectron. and Commun. Conf. and Int.
To conclude, modulator schemes that rely on the integration Conf. Photonics Switch. and Comput., MD2-2 (2019).
of efficient electro-optic materials with SiPh are gaining rapid 8. D. Patel et al., “High-speed compact silicon photonic Michelson
technological maturity. They provide an additional route to interferometric modulator,” Opt. Express 22(22), 26788–26802
fulfill the compelling needs of bandwidth, loss, and linearity at (2014).
low-power consumption and cost for future high-speed optical 9. J. Sun et al., “A 128 Gb∕s PAM4 silicon microring modulator
with integrated thermo-optic resonance tuning,” J. Lightwave
links. It is important to note that the modulators based on the Technol. 37(1), 110–115 (2019).
integration of efficient electro-optic materials with SiPh will 10. K. Goi et al., “Silicon Mach–Zehnder modulator using low-loss
not necessarily cannibalize or replace the plasma dispersion phase shifter with bottom PN junction formed by restricted-depth
modulators. It is a way to overcome bandwidth and loss limits doping,” IEICE Electron. Express 10(17), 20130552 (2013).
that stem from the physics of the plasma dispersion effect. 11. X. Tu et al., “Silicon optical modulator with shield coplanar
However they will come at the cost of a more complex waveguide electrodes,” Opt. Express 22(19), 23724–23731 (2014).

12. J. Van Campenhout et al., “Low-voltage, low-loss, multi-Gb/s 35. Q. Cheng et al., “Recent advances in optical technologies for data
silicon micro-ring modulator based on a MOS capacitor,” in centers: a review,” Optica 5(11), 1354–1370 (2018).
Opt. Fiber Commun. Conf., OM2E.4 (2012). 36. M. Asghari and A. V. Krishnamoorthy, “Energy-efficient com-
13. X. Xiao et al., “High-speed, low-loss silicon Mach–Zehnder munication,” Nat. Photonics 5(5), 268–270 (2011).
modulators with doping optimization,” Opt. Express 21(4), 37. F. Boeuf et al., “Benchmarking Si, SiGe, and III-V/Si hybrid SIS
4116–4125 (2013). optical modulators for datacenter applications,” J. Lightwave
14. M. Li et al., “Silicon intensity Mach–Zehnder modulator for Technol. 35(18), 4047–4055 (2017).
single lane 100 Gb∕s applications,” Photonics Res. 6(2), 109–116 38. N.-N. Feng et al., “30 GHz Ge electro-absorption modulator
(2018). integrated with 3 μm silicon-on-insulator waveguide,” Opt.
15. X. Tu et al., “50-Gb∕s silicon optical modulator with traveling- Express 19(8), 7062–7067 (2011).
wave electrodes,” Opt. Express 21(10), 12776–12782 (2013). 39. A. Rahim et al., “Open-access silicon photonics: current status and
16. D. J. Thomson et al., “50-Gb∕s silicon optical modulator,” IEEE emerging initiatives,” Proc. IEEE 106(12), 2313–2330 (2018).
Photonics Technol. Lett. 24(4), 234–236 (2012). 40. A. Meighan et al., “Design of 100 GHz-class Mach–Zehnder
17. D. Patel et al., “Design, analysis, and transmission system per- modulators in a generic indium phosphide platform,” in IEEE
formance of a 41 GHz silicon photonic modulator,” Opt. Express Photonics Conf. (2020).
23(11), 14263–14287 (2015). 41. S. Lange et al., “100 GBd intensity modulation and direct detec-
18. E. Timurdogan et al., “An ultralow power athermal silicon modu- tion with an InP-based monolithic DFB laser Mach–Zehnder
lator,” Nat. Commun. 5(1), 4008 (2014). modulator,” J. Lightwave Technol. 36(1), 97–102 (2018).
19. J. Ding et al., “Electro-optical response analysis of a 40 Gb∕s 42. M. Smit, K. Williams, and J. van der Tol, “Past, present, and
silicon Mach-Zehnder optical modulator,” J. Lightwave Technol. future of InP-based photonic integration,” APL Photonics 4(5),
31(14), 2434–2440 (2013). 050901 (2019).
20. S. A. Srinivasan et al., “50 Gb∕s C-band GeSi waveguide 43. K. Luke et al., “Wafer-scale low-loss lithium niobate photonic
electro-absorption modulator,” in Opt. Fiber Commun. Conf., integrated circuits,” Opt. Express 28(17), 24452–24458 (2020).
Tu3D.7 (2016). 44. P. De Dobbelaere et al., “Advanced silicon photonics technology
21. G. T. Reed et al., “Recent breakthroughs in carrier depletion platform leveraging a semiconductor supply chain,” in IEEE Int.
based silicon optical modulators,” Nanophotonics 3(4-5), 229– Electron Dev. Meeting, pp. 34.1.1–34.1.4 (2017).
245 (2014). 45. D. J. Shin et al., “Integration of silicon photonics into DRAM
22. D. Petousi et al., “Monolithically integrated high-extinction-ratio process,” in Opt. Fiber Commun. Conf./Natl. Fiber Opt. Eng.
MZM with a segmented driver in photonic BiCMOS,” IEEE Conf., OTu2C.4 (2013).
Photonics Technol. Lett. 28(24), 2866–2869 (2016). 46. R. Meade et al., “TeraPHY: a high-density electronic-photonic
23. A. Chen and E. E. Murphy, Broadband Optical Modulators: Chiplet for optical I/O from a multi-chip module,” in Opt.
Science, Technology, and Applications, 1st ed., CRC Press, Fiber Commun. Conf., M4D.7 (2019).
Boca Raton, Florida (2012). 47. F. Boeuf et al., “Recent progress in silicon photonics R&D and
24. S. K. Kim et al., “Theoretical and experimental study of 10 Gb∕s manufacturing on 300 mm wafer platform,” in Opt. Fiber
transmission performance using 1.55-μm LiNbO3 -based trans- Commun. Conf., W3A.1 (2015).
mitters with adjustable extinction ratio and chirp,” J. Lightwave 48. S. Fathololoumi et al., “1.6Tbps silicon photonics integrated
Technol. 17(8), 1320–1325 (1999). circuit for co-packaged optical-IO switch applications,” in Opt.
25. N. Qi et al., “Co-design and demonstration of a 25-Gb∕s silicon- Fiber Commun. Conf., T3H.1 (2020).
photonic Mach–Zehnder modulator with a CMOS-based high- 49. X. Chen et al., “The emergence of silicon photonics as a flexible
swing driver,” IEEE J. Sel. Top. Quantum Electron. 22(6), technology platform,” Proc. IEEE 106(12), 2101–2116 (2018).
131–140 (2016). 50. C. Doerr et al., “Single-chip silicon photonics 100-Gb/s coherent
26. L. Chen, P. Dong, and Y. Chen, “Chirp and dispersion tolerance transceiver,” in Opt. Fiber Commun. Conf., Th5C.1 (2014).
of a single-drive push–pull silicon modulator at 28 Gb∕s,” IEEE 51. J. C. Rosenberg et al., “Low-power 30 Gbps silicon microring
Photonics Technol. Lett. 24(11), 936–938 (2012). modulator,” in CLEO: Laser Appl. Photonic Appl., PDPB9
27. M. Jacques et al., “Modulator material impact on chirp, DSP, and (2011).
performance in coherent digital links: comparison of the lithium 52. C. Xiong et al., “Monolithic 56 Gb∕s silicon photonic pulse-
niobate, indium phosphide, and silicon platforms,” Opt. Express amplitude modulation transmitter,” Optica 3(10), 1060–1065
26(17), 22471–22490 (2018). (2016).
28. K. Goi et al., “20-Gbps BPSK silicon Mach–Zehnder modulator 53. “Cisco annual internet report (2018-2023) white paper—Cisco,”
with excellent chirp-free performance,” in 17th Opto-Electron. https://www.cisco.com/c/en/us/solutions/collateral/executive-
and Commun. Conf., pp. 238–239 (2012). perspectives/annual-internet-report/white-paper-c11-741490.html
29. R. Li et al., “High-speed low-chirp PAM-4 transmission based on (accessed 16 July 2020).
push-pull silicon photonic microring modulators,” Opt. Express 54. P. J. Winzer, D. T. Neilson, and A. R. Chraplyvy, “Fiber-optic
25(12), 13222–13229 (2017). transmission and networking: the previous 20 and the next 20
30. K. Goi et al., “11-Gb∕s 80-km transmission performance of zero- years [Invited],” Opt. Express 26(18), 24190–24239 (2018).
chirp silicon Mach–Zehnder modulator,” Opt. Express 20(26), 55. P. Dong et al., “50-Gb∕s silicon quadrature phase-shift keying
B350–B356 (2012). modulator,” Opt. Express 20(19), 21181–21186 (2012).
31. J. Liu et al., “Waveguide-integrated, ultralow-energy GeSi 56. P. Dong et al., “Experimental demonstration of microring quad-
electro-absorption modulators,” Nat. Photonics 2(7), 433–437 rature phase-shift keying modulators,” Opt. Lett. 37(7), 1178–
(2008). 1180 (2012).
32. D. Marris-Morini et al., “Low loss 40 Gbit∕s silicon modulator 57. “802.3bs-2017—IEEE Standard for Ethernet Amendment 10:
based on interleaved junctions and fabricated on 300 mm SOI media access control parameters, physical layers, and manage-
wafers,” Opt. Express 21(19), 22471–22475 (2013). ment parameters for 200 Gb∕s and 400 Gb∕s operation,” https://
33. J. C. Rosenberg et al., “A 25 Gbps silicon microring modulator standards.ieee.org/standard/802_3bs-2017.html (accessed 16 July
based on an interleaved junction,” Opt. Express 20(24), 26411– 2020).
26423 (2012). 58. X. Zhou, R. Urata, and H. Liu, “Beyond 1 Tb∕s intra-data center
34. D. A. B. Miller, “Device requirements for optical interconnects to interconnect technology: IM-DD OR coherent?” J. Lightwave
silicon chips,” Proc. IEEE 97(7), 1166–1185 (2009). Technol. 38(2), 475–484 (2020).

59. B. Jalali and S. Fathpour, “Silicon photonics,” J. Lightwave 82. H. Xu et al., “44 Gbit∕s silicon Mach–Zehnder modulator based
Technol. 24(12), 4600–4615 (2006). on interleaved PN junctions,” in 9th Int. Conf. Group IV
60. W. Bogaerts et al., “Nanophotonic waveguides in silicon-on- Photonics, pp. 201–203 (2012).
insulator fabricated with CMOS technology,” J. Lightwave 83. Z.-Y. Li et al., “Silicon waveguide modulator based on carrier
Technol. 23(1), 401–412 (2005). depletion in periodically interleaved PN junctions,” Opt. Express
61. R. Soref and B. Bennett, “Electrooptical effects in silicon,” IEEE 17(18), 15947–15958 (2009).
J. Quantum Electron. 23(1), 123–129 (1987). 84. M. Ziebell et al., “Ten Gbit/s ring resonator silicon modulator
62. Q. Xu et al., “Micrometre-scale silicon electro-optic modulator,” based on interdigitated PN junctions,” Opt. Express 19(15),
Nature 435(7040), 325–327 (2005). 14690–14695 (2011).
63. G. T. Reed and C. J. Png, “Silicon optical modulators,” Mater. 85. F. Y. Gardes et al., “40 Gb∕s silicon photonics modulator for TE
Today 8(1), 40–50 (2005). and TM polarisations,” Opt. Express 19(12), 11804–11814
64. C. K. Tang and G. T. Reed, “Highly efficient optical phase (2011).
modulator in SOI waveguides,” Electron. Lett. 31(6), 451–452 86. X. Xiao et al., “44-Gb∕s silicon microring modulators based on
(1995). zigzag PN junctions,” IEEE Photonics Technol. Lett. 24(19),
65. T. Baba et al., “25-Gb∕s broadband silicon modulator with 1712–1714 (2012).
0.31-V · cm V π L based on forward-biased PIN diodes embedded 87. X. Xiao et al., “High speed silicon photonic modulators,” in Opt.
with passive equalizer,” Opt. Express 23(26), 32950–32960 Fiber Commun. Conf., Tu2H.1 (2017).
(2015). 88. A. Irace, G. Breglio, and A. Cutolo, “All-silicon optoelectronic
66. S. Tanaka et al., “Ultralow-power (1.59 mW∕Gbps), 56-Gbps modulator with 1 GHz switching capability,” Electron. Lett.
PAM4 operation of Si photonic transmitter integrating segmented 39(2), 232–233 (2003).
PIN Mach–Zehnder modulator and 28-nm CMOS driver,” 89. R. Soref and J. Larenzo, “All-silicon active and passive guided-
J. Lightwave Technol. 36(5), 1275–1280 (2018). wave components for λ ¼ 13 and 1.6 μm,” IEEE J. Quantum
67. S. Akiyama et al., “High-performance silicon modulator for in- Electron. 22(6), 873–879 (1986).
tegrated transceivers fabricated on 300-mm wafer,” in Eur. Conf. 90. A. Cutolo et al., “Silicon electro-optic modulator based on a three
Opt. Commun., P.2.8 (2014). terminal device integrated in a low-loss single-mode SOI wave-
68. M. Webster et al., “An efficient MOS-capacitor based silicon guide,” J. Lightwave Technol. 15(3), 505–518 (1997).
modulator and CMOS drivers for optical transmitters,” in 11th 91. S. Akiyama et al., “Compact PIN-diode-based silicon modulator
Int. Conf. Group IV Photonics, WB1 (2014). using side-wall-grating waveguide,” IEEE J. Sel. Top. Quantum
69. J. Fujikata et al., “High-performance MOS-capacitor-type Si op- Electron. 19(6), 74–84 (2013).
tical modulator and surface-illumination-type Ge photodetector 92. S. J. Spector et al., “High-speed silicon electro-optical modulator
for optical interconnection,” Jpn. J. Appl. Phys. 55(4S), 04EC01 that can be operated in carrier depletion or carrier injection
(2016). mode,” in Conf. Lasers and Electro-Opt. and Conf. Quantum
70. M. Sodagar et al., “Compact, 15 Gb∕s electro-optic modulator Electron. and Laser Sci., CFH4 (2008).
through carrier accumulation in a hybrid Si∕SiO2 ∕Si microdisk,” 93. A. Sciuto et al., “Design, fabrication, and testing of an integrated
Opt. Express 23(22), 28306–28315 (2015). Si-based light modulator,” J. Lightwave Technol. 21(1), 228–235
71. H. Yi et al., “Demonstration of low power penalty of silicon (2003).
Mach–Zehnder modulator in long-haul transmission,” Opt. 94. S. J. Spector et al., “Operation and optimization of silicon-diode-
Express 20(25), 27562–27568 (2012). based optical modulators,” IEEE J. Sel. Top. Quantum Electron.
72. J.-B. You et al., “12.5 Gbps optical modulation of silicon race- 16(1), 165–172 (2010).
track resonator based on carrier-depletion in asymmetric P–N 95. J. C. Rosenberg et al., “Ultra-low-voltage micro-ring modulator
diode,” Opt. Express 16(22), 18340–18344 (2008). integrated with a CMOS feed-forward equalization driver,” in
73. D. J. Thomson et al., “High speed silicon optical modulator with Opt. Fiber Commun. Conf./Natl. Fiber Opt. Eng. Conf., OWQ4
self aligned fabrication process,” Opt. Express 18(18), 19064– (2011).
19069 (2010). 96. C. E. Png et al., “Optical phase modulators for MHz and GHz
74. D. J. Thomson et al., “High performance Mach–Zehnder-based modulation in silicon-on-insulator (SOI),” J. Lightwave Technol.
silicon optical modulators,” IEEE J. Sel. Top. Quantum Electron. 22(6), 1573–1582 (2004).
19(6), 85–94 (2013). 97. Q. Xu et al., “12.5 Gbit∕s carrier-injection-based silicon micro-
75. N.-N. Feng et al., “High speed carrier-depletion modulators with ring silicon modulators,” Opt. Express 15(2), 430–436 (2007).
1.4 V-cm V π L integrated on 0.25 μm silicon-on-insulator wave- 98. A. Shakoor et al., “Compact 1D-silicon photonic crystal electro-
guides,” Opt. Express 18(8), 7994–7999 (2010). optic modulator operating with ultra-low switching voltage and
76. P. Dong et al., “Low Vpp, ultralow-energy, compact, high-speed energy,” Opt. Express 22(23), 28623–28634 (2014).
silicon electro-optic modulator,” Opt. Express 17(25), 22484– 99. W. M. J. Green et al., “Ultra-compact, low RF power, 10 Gb∕s
22490 (2009). silicon Mach-Zehnder modulator,” Opt. Express 15(25), 17106–
77. D. J. Thomson et al., “High contrast 40 Gbit∕s optical modula- 17113 (2007).
tion in silicon,” Opt. Express 19(12), 11507–11516 (2011). 100. S. Meister et al., “High-speed Fabry–Pérot optical modulator in
78. D. Marris-Morini et al., “Low loss and high speed silicon optical silicon with 3-μm diode,” J. Lightwave Technol. 33(4), 878–881
modulator based on a lateral carrier depletion structure,” Opt. (2015).
Express 16(1), 334–339 (2008). 101. L. Liao et al., “High speed silicon Mach–Zehnder modulator,”
79. M. Ziebell et al., “40 Gbit∕s low-loss silicon optical modulator Opt. Express 13(8), 3129–3135 (2005).
based on a pipin diode,” Opt. Express 20(10), 10591–10596 102. A. Liu et al., “A high-speed silicon optical modulator based on a
(2012). metal–oxide–semiconductor capacitor,” Nature 427(6975), 615–
80. H. Xu et al., “High speed silicon Mach–Zehnder modulator 618 (2004).
based on interleaved PN junctions,” Opt. Express 20(14), 103. K. Debnath et al., “All-silicon carrier accumulation modulator
15093–15099 (2012). based on a lateral metal-oxide-semiconductor capacitor,”
81. H. Yu, W. Bogaerts, and A. De Keersgieter, “Optimization of Photonics Res. 6(5), 373–379 (2018).
ion implantation condition for depletion-type silicon optical 104. A. Abraham et al., “Evaluation of the performances of a silicon
modulators,” IEEE J. Quantum Electron. 46(12), 1763–1768 optical modulator based on a silicon-oxide-silicon capacitor,” in
(2010). 11th Int. Conf. Group IV Photonics, WB2 (2014).

105. E. Li et al., “One-volt silicon photonic crystal nanocavity modu- 129. C. G. Bottenfield, V. A. Thomas, and S. E. Ralph, “Silicon
lator with indium oxide gate,” Opt. Lett. 43(18), 4429–4432 photonic modulator linearity and optimization for microwave
(2018). photonic links,” IEEE J. Sel. Top. Quantum Electron. 25(5),
106. L. Liao et al., “40 Gbit∕s silicon optical modulator for high- 3400110 (2019).
speed applications,” Electron. Lett. 43(22), 1196–1197 (2007). 130. L. Chrostowski et al., “Impact of fabrication non-uniformity on
107. Y. Maegami et al., “High-efficiency strip-loaded waveguide chip-scale silicon photonic integrated circuits,” in Opt. Fiber
based silicon Mach–Zehnder modulator with vertical p-n Commun. Conf., Th2A.37 (2014).
junction phase shifter,” Opt. Express 25(25), 31407–31416 131. W. Bogaerts et al., “Silicon microring resonators,” Laser
(2017). Photonics Rev. 6(1), 47–73 (2012).
108. A. Liu et al., “High-speed optical modulation based on carrier 132. B. Guha, B. B. C. Kyotoku, and M. Lipson, “CMOS-compatible
depletion in a silicon waveguide,” Opt. Express 15(2), 660–668 athermal silicon microring resonators,” Opt. Express 18(4),
(2007). 3487–3493 (2010).
109. F. Y. Gardes et al., “A sub-micron depletion-type photonic modu- 133. D. Feng et al., “High-speed GeSi electroabsorption modulator on
lator in silicon on insulator,” Opt. Express 13(22), 8845–8854 the SOI waveguide platform,” IEEE J. Sel. Top. Quantum
(2005). Electron. 19(6), 64–73 (2013).
110. M. R. Watts et al., “Low-voltage, compact, depletion-mode, 134. S. Gupta et al., “50-GHz Ge waveguide electro-absorption modu-
silicon Mach-Zehnder modulator,” IEEE J. Sel. Top. Quantum lator integrated in a 220 nm SOI photonics platform,” in Opt.
Electron. 16(1), 159–164 (2010). Fiber Commun. Conf., Tu2A.4 (2015).
111. M. R. Watts et al., “Vertical junction silicon microdisk modula- 135. D. Feng et al., “High speed GeSi electro-absorption modulator at
tors and switches,” Opt. Express 19(22), 21989–22003 (2011). 1550 nm wavelength on SOI waveguide,” Opt. Express 20(20),
112. H. C. Nguyen et al., “Compact and fast photonic crystal silicon 22224–22232 (2012).
optical modulators,” Opt. Express 20(20), 22465–22474 (2012). 136. S. A. Srinivasan et al., “56 Gb∕s Germanium waveguide electro-
113. Y. Terada et al., “Full C-band Si photonic crystal waveguide absorption modulator,” J. Lightwave Technol. 34(2), 419–424
modulator,” Opt. Lett. 42(24), 5110–5112 (2017). (2016).
114. Y. Hinakura, H. Arai, and T. Baba, “64 Gbps Si photonic crystal 137. J. Verbist et al., “100 Gb∕s DAC-less and DSP-free transmitters
slow light modulator by electro-optic phase matching,” Opt. using GeSi EAMs for short-reach optical interconnects,” in Opt.
Express 27(10), 14321–14327 (2019). Fiber Commun. Conf., W4D.4 (2018).
115. R. Dubé-Demers, S. LaRochelle, and W. Shi, “Ultrafast pulse- 138. A. E.-J. Lim et al., “Novel evanescent-coupled germanium
amplitude modulation with a femtojoule silicon photonic modu- electro-absorption modulator featuring monolithic integration
lator,” Optica 3(6), 622–627 (2016). with germanium p–i–n photodetector,” Opt. Express 19(6),
116. P. Dong, L. Chen, and Y. K. Chen, “High-speed low-voltage 5040–5046 (2011).
single-drive push-pull silicon Mach–Zehnder modulators,” 139. L. Mastronardi et al., “High-speed Si/GeSi hetero-structure
Opt. Express 20(6), 6163–6169 (2012). electro absorption modulator,” Opt. Express 26(6), 6663–6673
117. M. Streshinsky et al., “Low power 50 Gb∕s silicon traveling (2018).
wave Mach–Zehnder modulator near 1300 nm,” Opt. Express 140. Y. Tang, J. D. Peters, and J. E. Bowers, “Over 67 GHz bandwidth
21(25), 30350–30357 (2013). hybrid silicon electroabsorption modulator with asymmetric seg-
118. T. Baehr-Jones et al., “Ultralow drive voltage silicon traveling- mented electrode for 1.3 μm transmission,” Opt. Express 20(10),
wave modulator,” Opt. Express 20(11), 12014–12020 (2012). 11529–11535 (2012).
119. R. Ding et al., “A compact low-power 320-Gb∕s WDM transmit- 141. P. Chaisakul et al., “Integrated germanium optical interconnects
ter based on silicon microrings,” IEEE Photonics J. 6(3), on silicon substrates,” Nat. Photonics 8(6), 482–488 (2014).
6600608 (2014). 142. P. Chaisakul et al., “Recent progress in GeSi electro-absorption
120. X. Tu et al., “Fabrication of low loss and high speed silicon modulators,” J. Lightwave Technol. 15(1), 014601 (2014).
optical modulator using doping compensation method,” Opt. 143. J. Frigerio et al., “Giant electro-optic effect in Ge/SiGe coupled
Express 19(19), 18029–18035 (2011). quantum wells,” Sci. Rep. 5(1), 15398 (2015).
121. X. Li et al., “Highly efficient silicon Michelson interferometer 144. Y.-H. Kuo et al., “Strong quantum-confined Stark effect in ger-
modulators,” IEEE Photonics Technol. Lett. 25(5), 407–409 manium quantum-well structures on silicon,” Nature 437(7063),
(2013). 1334–1336 (2005).
122. K. Li et al., “Electronic–photonic convergence for silicon pho- 145. D. A. B. Miller et al., “Band-edge electroabsorption in quantum
tonics transmitters beyond 100 Gbps on–off keying,” Optica well structures: the quantum-confined Stark effect,” Phys. Rev.
7(11), 1514–1516 (2020). Lett. 53(22), 2173–2176 (1984).
123. H. Zhang et al., “800 Gbit∕s transmission over 1 km single-mode 146. D. A. B. Miller, “Attojoule optoelectronics for low-energy infor-
fiber using a four-channel silicon photonic transmitter,” mation processing and communications,” J. Lightwave Technol.
Photonics Res. 8(11), 1776–1782 (2020). 35(3), 346–396 (2017).
124. D. Pérez-Galacho et al., “QPSK modulation in the O-band 147. S. Ren et al., “Ge/SiGe quantum well waveguide modulator
using a single dual-drive Mach–Zehnder silicon modulator,” monolithically integrated with SOI waveguides,” IEEE
J. Lightwave Technol. 36(18), 3935–3940 (2018). Photonics Technol. Lett. 24(6), 461–463 (2012).
125. F. Fresi et al., “Silicon photonics integrated 16-QAM modulator 148. P. Chaisakul et al., “23 GHz Ge/SiGe multiple quantum well
exploiting only binary driving electronics,” in Opt. Fiber electro-absorption modulator,” Opt. Express 20(3), 3219–3224
Commun. Conf., Th3J.6 (2016). (2012).
126. S. Zhalehpour et al., “All-silicon IQ modulator for 100 GBaud 149. R. M. Audet et al., “Surface-normal Ge/SiGe asymmetric Fabry–
32QAM transmissions,” in Opt. Fiber Commun. Conf. Perot optical modulators fabricated on silicon substrates,”
Postdeadline Papers, Th4A.5 (2019). J. Lightwave Technol. 31(24), 3995–4003 (2013).
127. L. Deniel et al., “DAC-less PAM-4 generation in the O-band 150. E. H. Edwards et al., “Ge/SiGe asymmetric Fabry–Perot quan-
using a silicon Mach-Zehnder modulator,” Opt. Express 27(7), tum well electroabsorption modulators,” Opt. Express 20(28),
9740–9748 (2019). 29164–29173 (2012).
128. Y. Kim et al., “Strain-induced enhancement of plasma dispersion 151. P. Chaisakul et al., “Recent progress on Ge/SiGe quantum well
effect and free-carrier absorption in SiGe optical modulators,” optical modulators, detectors, and emitters for optical intercon-
Sci. Rep. 4(1), 4683 (2014). nects,” Photonics 6(1), 24 (2019).

152. S. A. Srinivasan et al., “High absorption contrast quantum 177. C. Xiong et al., “Active silicon integrated nanophotonics: ferro-
confined Stark effect in ultra-thin Ge/SiGe quantum well stacks electric BaTiO3 devices,” Nano Lett. 14(3), 1419–1425 (2014).
grown on Si,” IEEE J. Quantum Electron. 56(1), 5200207 (2020). 178. F. Eltes et al., “Low-loss BaTiO3 − Si waveguides for nonlinear
153. M. Liu et al., “A graphene-based broadband optical modulator,” integrated photonics,” ACS Photonics 3(9), 1698–1703 (2016).
Nature 474(7349), 64–67 (2011). 179. S. Abel et al., “Large Pockels effect in micro- and nanostructured
154. M. A. Giambra et al., “High-speed double layer graphene electro- barium titanate integrated on silicon,” Nat. Mater. 18(1), 42–47
absorption modulator on SOI waveguide,” Opt. Express 27(15), (2019).
20145–20155 (2019). 180. A. Messner et al., “Integrated ferroelectric BaTiO3 ∕Si plasmonic
155. M. Mohsin et al., “Graphene based low insertion loss electro- modulator for 100 Gbit∕s and beyond,” in Opt. Fiber Commun.
absorption modulator on SOI waveguide,” Opt. Express 22(12), Conf., M2I.6 (2018).
15292–15297 (2014). 181. F. Eltes et al., “A novel 25 Gbps electro-optic Pockels modulator
156. M. Liu, X. Yin, and X. Zhang, “Double-layer graphene optical integrated on an advanced Si photonic platform,” in IEEE Int.
modulator,” Nano Lett. 12(3), 1482–1485 (2012). Electron Dev. Meeting, pp. 24.5.1–24.5.4 (2017).
157. Y. Hu et al., “Broadband 10 Gb∕s operation of graphene electro- 182. F. Eltes et al., “A BaTiO3 -based electro-optic Pockels modulator
absorption modulator on silicon,” Laser Photonics Rev. 10(2), monolithically integrated on an advanced silicon photonics plat-
307–316 (2016). form,” J. Lightwave Technol. 37(5), 1456–1462 (2019).
158. C. T. Phare et al., “Graphene electro-optic modulator with 183. K. Alexander et al., “Broadband electro-optic modulation using
30 GHz bandwidth,” Nat. Photonics 9(8), 511–514 (2015). low-loss PZT-on-silicon nitride integrated waveguides,” in Conf.
159. L. A. Shiramin et al., “High extinction ratio hybrid graphene- Lasers and Electro-Opt., JTh5C.7 (2017).
silicon photonic crystal switch,” IEEE Photonics Technol. Lett. 184. K. Alexander et al., “Nanophotonic Pockels modulators on a
30(2), 157–160 (2018). silicon nitride platform,” Nat Commun 9(1), 3444 (2018).
160. V. Sorianello et al., “Chirp management in silicon-graphene 185. K. Alexander, “Integrated silicon nitride photonics with highly
electro absorption modulators,” Opt. Express 25(16), 19371– nonlinear thin films and 2D materials: properties and devices,”
19381 (2017). PhD thesis, Ghent University (2018).
161. J.-H. Han et al., “Efficient low-loss InGaAsP/Si hybrid MOS 186. C. Koos et al., “All-optical high-speed signal processing with
optical modulator,” Nat. Photonics 11(8), 486–490 (2017). silicon–organic hybrid slot waveguides,” Nat. Photonics 3(4),
162. J. Witzens, “Modulators make efficiency leap,” Nat. Photonics 216–219 (2009).
11(8), 459–462 (2017). 187. D. Korn et al., “Lasing in silicon–organic hybrid waveguides,”
163. T. Hiraki et al., “Heterogeneously integrated III–V/Si MOS Nat. Commun. 7(1), 10864 (2016).
capacitor Mach–Zehnder modulator,” Nat. Photonics 11(8), 188. C. Koos et al., “Silicon-organic hybrid (SOH) and plasmonic-
482–485 (2017). organic hybrid (POH) integration,” J. Lightwave Technol. 34(2),
164. Q. Li et al., “Si racetrack modulator with III-V/Si hybrid 256–268 (2016).
MOS optical phase shifter,” in 45th Eur. Conf. Opt. Commun. 189. R. Palmer et al., “High-speed, low drive-voltage silicon-
(2019). organic hybrid modulator based on a binary-chromophore
165. Q. Li et al., “Efficient optical modulator by reverse-biased III-V/ electro-optic material,” J. Lightwave Technol. 32(16), 2726–
Si hybrid MOS capacitor based on FK effect and carrier 2734 (2014).
depletion,” in Opt. Fiber Commun. Conf., M4A.2 (2019). 190. C. Kieninger et al., “Ultra-high electro-optic activity demon-
166. T. Komljenovic et al., “Photonic integrated circuits using hetero- strated in a silicon-organic hybrid modulator,” Optica 5(6),
geneous integration on silicon,” Proc. IEEE 106(12), 2246–2257 739–748 (2018).
(2018). 191. L. Alloatti et al., “100 GHz silicon–organic hybrid modulator,”
167. L. Cao et al., “Hybrid amorphous silicon (a-Si:H)-LiNbO3 Light Sci. Appl. 3(5), e173 (2014).
electro-optic modulator,” Opt. Commun. 330, 40–44 (2014). 192. S. Koeber et al., “Femtojoule electro-optic modulation using
168. A. Rao et al., “Heterogeneous microring and Mach–Zehnder a silicon–organic hybrid device,” Light Sci. Appl. 4(2), e255
modulators based on lithium niobate and chalcogenide glasses (2015).
on silicon,” Opt. Express 23(17), 22746–22752 (2015). 193. H. Zwickel et al., “Silicon-organic hybrid (SOH) modulators for
169. L. Chen et al., “Hybrid silicon and lithium niobate electro-optical intensity-modulation/direct-detection links with line rates of up
ring modulator,” Optica 1(2), 112–118 (2014). to 120 Gbit∕s,” Opt. Express 25(20), 23784–23800 (2017).
170. S. Jin et al., “LiNbO3 thin-film modulators using silicon nitride 194. C. Kieninger et al., “Silicon-organic hybrid (SOH) Mach–
surface ridge waveguides,” IEEE Photonics Technol. Lett. 28(7), Zehnder modulators for 100 GBd PAM4 signaling with sub-
736–739 (2016). 1 dB phase-shifter loss,” Opt. Express 28(17), 24693–24707
171. A. Rao et al., “High-performance and linear thin-film lithium (2020).
niobate Mach–Zehnder modulators on silicon up to 50 GHz,” 195. C. Kieninger et al., “Demonstration of long-term thermally stable
Opt. Lett. 41(24), 5700–5703 (2016). silicon-organic hybrid modulators at 85°C,” Opt. Express 26(21),
172. N. Boynton et al., “A heterogeneously integrated silicon pho- 27955–27964 (2018).
tonic/lithium niobate travelling wave electro-optic modulator,” 196. S. Wolf et al., “Silicon-organic hybrid (SOH) Mach–Zehnder
Opt. Express 28(2), 1868–1884 (2020). modulators for 100 Gbit∕s on-off keying,” Sci. Rep. 8(1), 2598
173. P. O. Weigel et al., “Bonded thin film lithium niobate modulator (2018).
on a silicon photonics platform exceeding 100 GHz 3-dB 197. T. Baehr-Jones et al., “Optical modulation and detection in
electrical modulation bandwidth,” Opt. Express 26(18), 23728– slotted Silicon waveguides,” Opt. Express 13(14), 5216–5226
23739 (2018). (2005).
174. M. He et al., “High-performance hybrid silicon and lithium 198. H. Figi et al., “Electro-optic modulation in horizontally slotted
niobate Mach–Zehnder modulators for 100 Gbit s−1 and beyond,” silicon/organic crystal hybrid devices,” J. Opt. Soc. Am. B
Nat. Photonics 13(5), 359–364 (2019). 28(9), 2291–2300 (2011).
175. M. Li and H. X. Tang, “Strong Pockels materials,” Nat. Mater. 199. D. Korn et al., “Electro-optic organic crystal silicon high-speed
18(1), 9–11 (2019). modulator,” IEEE Photonics J. 6(2), 2700109 (2014).
176. S. Abel et al., “A hybrid barium titanate-silicon photonics 200. S. Ummethala et al., “Capacitively coupled silicon-organic
platform for ultraefficient electro-optic tuning,” J. Lightwave hybrid modulator for 200 Gbit/s PAM-4 signaling,” in Conf.
Technol. 34(8), 1688–1693 (2016). Lasers and Electro-Opt., JTh5B.2 (2019).

201. V. Sorianello et al., “Graphene–silicon phase modulators with 225. M. Gehl et al., “Operation of high-speed silicon photonic micro-
gigahertz bandwidth,” Nat. Photonics 12(1), 40–44 (2018). disk modulators at cryogenic temperatures,” Optica 4(3), 374–
202. M. Midrio et al., “Graphene-based optical phase modulation of 382 (2017).
waveguide transverse electric modes,” Photonics Res. 2(3), A34– 226. S. M. Jackson et al., “A novel optical phase modulator design
A40 (2014). suitable for phased arrays,” J. Lightwave Technol. 16(11),
203. H. Dalir et al., “Athermal broadband graphene optical modulator 2016–2019 (1998).
with 35 GHz speed,” ACS Photonics 3(9), 1564–1568 (2016). 227. J. Shin et al., “Epitaxial growth technology for optical intercon-
204. I. Datta et al., “Low-loss composite photonic platform based on nect based on bulk-Si platform,” in 10th Int. Conf. Group IV
2D semiconductor monolayers,” Nat. Photonics 14(4), 256–262 Photonics (2013).
(2020). 228. O. Marshall et al., “Heterogeneous integration on silicon photon-
205. S. Ye et al., “High-speed optical phase modulator based on gra- ics,” Proc. IEEE 106(12), 2258–2269 (2018).
phene-silicon waveguide,” IEEE J. Sel. Top. Quantum Electron. 229. S. Kodama, T. Yoshimatsu, and H. Ito, “500 Gbit∕s optical gate
23(1), 76–80 (2017). monolithically integrating photodiode and electroabsorption
206. C. Xu et al., “Characteristics of electro-refractive modulating modulator,” Electron. Lett. 40(9), 555–556 (2004).
based on graphene-oxide-silicon waveguide,” Opt. Express 230. S. A. Srinivasan, “Advanced Germanium devices for optical
20(20), 22398–22405 (2012). interconnects,” PhD Thesis, Ghent University (2017).
207. M. Mohsin et al., “Experimental verification of electro-refractive 231. A. Melikyan et al., “Differential drive I/Q modulator based on
phase modulation in graphene,” Sci. Rep. 5(1), 10967 (2015). silicon photonic electro-absorption modulators,” J. Lightwave
208. L. Yang et al., “Low-chirp high-extinction-ratio modulator based Technol. 38(11), 2872–2876 (2020).
on graphene–silicon waveguide,” Opt. Lett. 38(14), 2512–2515 232. Y. Tong et al., “Integrated germanium-on-silicon Franz–Keldysh
(2013). vector modulator used with a Kramers–Kronig receiver,” Opt.
209. J. Leuthold et al., “High-speed, low-power optical modulators in Lett. 43(18), 4333–4336 (2018).
silicon,” in 15th Int. Conf. Transp. Opt. Networks, We.D2.1 233. J. Verbist et al., “Real-time and DSP-free 128 Gb∕s PAM-4 link
(2013). using a binary driven silicon photonic transmitter,” J. Lightwave
210. A. Moscoso-Mártir et al., “Co-integration of a temperature Technol. 37(2), 274–280 (2019).
tolerant low impedance resonantly enhanced silicon photonics 234. K. M. Rabe et al., Modern Physics of Ferroelectrics: Essential
modulator,” in IEEE 14th Int. Conf. Group IV Photonics, Background, pp. 1–30, Springer (2007).
pp. 101–102 (2017). 235. M. Veithen, X. Gonze, and P. Ghosez, “Nonlinear optical suscep-
211. X. Li et al., “Single-drive high-speed lumped depletion-type tibilities, Raman efficiencies, and electro-optic tensors from first-
modulators toward 10 fJ∕bit energy consumption,” Photonics principles density functional perturbation theory,” Phys. Rev. B
Res. 5(2), 134–142 (2017). 71(12), 125107 (2005).
212. A. Giuglea et al., “Comparison of segmented and traveling-wave 236. H. Buhay et al., “Pulsed laser deposition and ferroelectric char-
electro-optical transmitters based on silicon photonics Mach– acterization of bismuth titanate films,” Appl. Phys. Lett. 58(14),
Zehnder modulators,” in Photonics Switching and Comput. (2018). 1470–1472 (1991).
213. S. Lin et al., “Electronic-photonic co-optimization of high-speed 237. D. J. R. Appleby et al., “Ferroelectric properties in thin film ba-
silicon photonic transmitters,” J. Lightwave Technol. 35(21), rium titanate grown using pulsed laser deposition,” J. Appl. Phys.
4766–4780 (2017). 116(12), 124105 (2014).
214. D. J. Thomson et al., “Optical detection and modulation at 238. T. Kobayashi et al., “Effect of multi-coating process on the ori-
2 μm -2.5 μm in silicon,” Opt. Express 22(9), 10825–10830 entation and microstructure of lead zirconate titanate (PZT) thin
(2014). films derived by chemical solution deposition,” Thin Solid Films
215. M. Montesinos-Ballester et al., “Optical modulation in Ge-rich 489(1), 74–78 (2005).
SiGe waveguides in the mid-infrared wavelength range up to 239. Z. Bi, Z. Zhang, and P. Fan, “Characterization of PZT ferroelec-
11 μm,” Commun. Mater. 1(1), 6 (2020). tric thin films by RF-magnetron sputtering,” J. Phys. Conf. Ser.
216. J. P. Lorenzo and R. A. Soref, “1.3 μm electro-optic silicon 61, 120–124 (2007).
switch,” Appl. Phys. Lett. 51(1), 6–8 (1987). 240. T. Li et al., “Metalorganic chemical vapor deposition of ferro-
217. B. Chmielak et al., “Pockels effect based fully integrated, electric SrBi2 Ta2 O9 thin films,” Appl. Phys. Lett. 68(5), 616–618
strained silicon electro-optic modulator,” Opt. Express 19(18), (1996).
17212–17219 (2011). 241. F. Zhang et al., “Atomic layer deposition of Pb(Zr,Ti)Ox on
218. R. S. Jacobsen et al., “Strained silicon as a new electro-optic 4H-SiC for metal-ferroelectric-insulator-semiconductor diodes,”
material,” Nature 441(7090), 199–202 (2006). J. Appl. Phys. 109(12), 124109 (2011).
219. P. Damas et al., “Wavelength dependence of Pockels effect in 242. E. L. Wooten et al., “A review of lithium niobate modulators
strained silicon waveguides,” Opt. Express 22(18), 22095–22100 for fiber-optic communications systems,” IEEE J. Sel. Top.
(2014). Quantum Electron. 6(1), 69–82 (2000).
220. Y. Shiraki et al., Silicon-Germanium (SiGe) Nanostructures, 243. M. Zhang et al., “Ultra-high bandwidth integrated lithium
Y. Shiraki and N. Usami, Eds., Woodhead Publishing Series in niobate modulators with record-low V π,” in Opt. Fiber Commun.
Electronic and Optical Materials, Woodhead Publishing (2011). Conf., Th4A.5 (2018).
221. L. Vivien and L. E. Pavesi, Handbook of Silicon Photonics, 1st 244. K. Noguchi, O. Mitomi, and H. Miyazawa, “Millimeter-wave
ed., CRC Press (2013). Ti∶LiNbO3 optical modulators,” J. Lightwave Technol. 16(4),
222. M. Zeiler et al., “Radiation damage in silicon photonic Mach– 615–619 (1998).
Zehnder modulators and photodiodes,” IEEE Trans. Nucl. Sci. 245. G. Poberaj et al., “Lithium niobate on insulator (LNOI) for
64(11), 2794–2801 (2017). micro-photonic devices,” Laser Photonics Rev. 6(4), 488–503
223. “Photonic integration and photonics-electronics convergence on (2012).
silicon platform,” https://www.frontiersin.org/research-topics/2972/ 246. A. J. Mercante et al., “Thin film lithium niobate electro-optic
photonic-integration-and-photonics-electronics-convergence-on- modulator with terahertz operating bandwidth,” Opt. Express
silicon-platform (accessed 1 February 2021). 26(11), 14810–14816 (2018).
224. U. Chakraborty et al., “Cryogenic operation of silicon photonic 247. C. Wang et al., “Integrated lithium niobate electro-optic modu-
modulators based on the DC Kerr effect,” Optica 7(10), 1385– lators operating at CMOS-compatible voltages,” Nature
1390 (2020). 562(7725), 101–104 (2018).

248. H. Takagi and R. Maeda, “Room temperature bonding of silicon 272. D. Liang and J. E. Bowers, “Recent progress in lasers on silicon,”
and lithium niobate,” Appl. Phys. Lett. 89(3), 031914 (2006). Nat. Photonics 4(8), 511–517 (2010).
249. P. Rabiei et al., “Heterogeneous lithium niobate photonics on 273. T. Shimizu et al., “High density hybrid integrated light source
silicon substrates,” Opt. Express 21(21), 25573–25581 (2013). with a laser diode array on a silicon optical waveguide platform
250. C. Wang et al., “Nanophotonic lithium niobate electro-optic for inter-chip optical interconnection,” in 8th IEEE Int. Conf.
modulators,” Opt. Express 26(2), 1547–1555 (2018). Group IV Photonics, pp. 181–183 (2011).
251. P. Bernasconi, M. Zgonik, and P. Günter, “Temperature 274. G. Roelkens et al., “III-V/Si PICs based on micro-transfer-print-
dependence and dispersion of electro-optic and elasto-optic ing,” in Opt. Fiber Commun. Conf., W4E.6 (2019).
effect in perovskite crystals,” J. Appl. Phys. 78(4), 2651–2658 275. A. J. Zilkie et al., “Multi-micron silicon photonics platform for
(1995). highly manufacturable and versatile photonic integrated circuits,”
252. R. A. McKee et al., “Molecular beam epitaxy growth of epitaxial IEEE J. Sel. Top. Quantum Electron. 25(5), 8200713 (2019).
barium silicide, barium oxide, and barium titanate on silicon,” 276. B. Szelag et al., “Hybrid III-V/silicon technology for laser inte-
Appl. Phys. Lett. 59(7), 782–784 (1991). gration on a 200-mm fully CMOS-compatible silicon photonics
253. A. Petraru et al., “Ferroelectric BaTiO3 thin-film optical wave- platform,” IEEE J. Sel. Top. Quantum Electron. 25(5), 8201210
guide modulators,” Appl. Phys. Lett. 81(8), 1375–1377 (2002). (2019).
254. A. A. Demkov and A. B. Posadas, Integration of Functional 277. A. Alduino, “Demonstration of a high speed 4-channel integrated
Oxides with Semiconductors, Springer, New York (2014). silicon photonics WDM link with hybrid silicon lasers,” in IEEE
255. M.-H. Hsu, “Monolithic integration of barium titanate on silicon Hot Chips 22 Symp., pp. 1–29 (2010).
for high-speed and power-efficient optical modulator applica- 278. S. Chen et al., “Electrically pumped continuous-wave III–V quan-
tions,” PhD Thesis, Ghent University (2016). tum dot lasers on silicon,” Nat. Photonics 10(5), 307–311 (2016).
256. J. P. George et al., “Lanthanide-assisted deposition of strongly 279. Z. Wang et al., “Room-temperature InP distributed feedback laser
electro-optic PZT thin films on silicon: toward integrated active array directly grown on silicon,” Nat. Photonics 9(12), 837–842
nanophotonic devices,” ACS Appl. Mater. Interfaces 7(24), (2015).
13350–13359 (2015). 280. Q. Li et al., “Optical phase modulators based on reverse-biased
257. J. P. George, “Integration of ferroelectric thin films on silicon for III-V/Si hybrid metal-oxide-semiconductor capacitors,” IEEE
electro-optic devices,” PhD Thesis, Ghent University (2016). Photonics Technol. Lett. 32(6), 345–348 (2020).
258. W. Heni et al., “Silicon–organic and plasmonic–organic hybrid 281. S. Ohno et al., “Taper-less III-V/Si hybrid MOS optical phase
photonics,” ACS Photonics 4(7), 1576–1590 (2017). shifter using ultrathin InP membrane,” in Opt. Fiber Commun.
259. H. Zwickel et al., “Verified equivalent-circuit model for slot- Conf., M2B.6 (2020).
waveguide modulators,” Opt. Express 28(9), 12951–12976 282. Y. Tang et al., “50 Gb∕s hybrid silicon traveling-wave electro-
(2020). absorption modulator,” Opt. Express 19(7), 5811–5816 (2011).
260. S. Wolf et al., “DAC-less amplifier-less generation and transmis- 283. C. Mattevi, H. Kim, and M. Chhowalla, “A review of chemical
sion of QAM signals using sub-volt silicon-organic hybrid mod- vapour deposition of graphene on copper,” J. Mater. Chem.
ulators,” J. Lightwave Technol. 33(7), 1425–1432 (2015). 21(10), 3324–3334 (2011).
261. S. Wolf et al., “Coherent modulation up to 100 GBd 16QAM 284. Y. S. Kim et al., “Direct growth of patterned graphene on SiO2
using silicon-organic hybrid (SOH) devices,” Opt. Express 26(1), substrates without the use of catalysts or lithography,” Nanoscale
220–232 (2018). 6(17), 10100–10105 (2014).
262. G.-W. Lu et al., “High-temperature-resistant silicon-polymer 285. X. Li et al., “Large-area synthesis of high-quality and uniform
hybrid modulator operating at up to 200 Gbit s−1 for energy- graphene films on copper foils,” Science 324(5932), 1312–1314
efficient datacentres and harsh-environment applications,” Nat. (2009).
Commun. 11(1), 4224 (2020). 286. R. R. Nair et al., “Fine structure constant defines visual transpar-
263. M. Jazbinsek et al., “Organic electro-optic single crystalline films ency of graphene,” Science 320(5881), 1308 (2008).
for integrated optics,” Proc. SPIE 7774, 77740Q (2010). 287. K. I. Bolotin et al., “Ultrahigh electron mobility in suspended
264. A. Melikyan et al., “High-speed plasmonic phase modulators,” graphene,” Solid State Commun. 146(9-10), 351–355 (2008).
Nat. Photonics 8(3), 229–233 (2014). 288. A. S. Mayorov et al., “Micrometer-scale ballistic transport in
265. C. Haffner et al., “Plasmonic organic hybrid modulators-scaling encapsulated graphene at room temperature,” Nano Lett. 11(6),
highest speed photonics to the microscale,” Proc. IEEE 104(12), 2396–2399 (2011).
2362–2379 (2016). 289. V. Sorianello, M. Midrio, and M. Romagnoli, “Design optimiza-
266. C. Hoessbacher et al., “Plasmonic modulator with 170 GHz tion of single and double layer graphene phase modulators in
bandwidth demonstrated at 100 GBd NRZ,” Opt. Express 25(3), SOI,” Opt. Express 23(5), 6478–6490 (2015).
1762–1768 (2017). 290. F. Wang et al., “Gate-variable optical transitions in graphene,”
267. C. Uhl, H. Hettrich, and M. Möller, “A 100 Gbit∕s 2 Vpp power Science 320(5873), 206–209 (2008).
multiplexer in SiGe BiCMOS technology for directly driving 291. L. A. Falkovsky, “Optical properties of graphene,” J. Phys. Conf.
a monolithically integrated plasmonic MZM in a silicon photon- Ser. 129, 012004 (2008).
ics transmitter,” in IEEE Bipolar/BiCMOS Circuits and Technol.
Meeting, pp. 106–109 (2017). Abdul Rahim is currently working with Professor Roel Baets at the
268. W. Heni et al., “High speed plasmonic modulator array enabling Photonics Research Group of Ghent University. He is the coordinator of
dense optical interconnect solutions,” Opt. Express 23(23), ePIXfab, the European Silicon Photonics Alliance. From 2009 to 2014,
29746–29757 (2015). he worked in the research group of Professor Klaus Petermann and
269. A. Melikyan et al., “Plasmonic-organic hybrid (POH) modulators was awarded a doctorate degree by Technische Universitaet Berlin for
for OOK and BPSK signaling at 40 Gbit∕s,” Opt. Express 23(8), his work on investigating innovative silicon photonic integrated filters
9938–9946 (2015). for optical communication systems. Prior to his doctoral work, he worked
270. S. Ummethala et al., “THz-to-optical conversion in wireless com- in various capacities at the R&D departments of numerous public and
munications using an ultra-broadband plasmonic modulator,” private sector organizations.
Nat. Photonics 13(8), 519–524 (2019).
271. G. Duan et al., “Hybrid III–V on silicon lasers for photonic in- Artur Hermans received his BSc degree in engineering (electronics
tegrated circuits on silicon,” IEEE J. Sel. Top. Quantum Electron. and information technology) from Vrije Universiteit Brussel (VUB) in
20(4), 158–170 (2014). 2012, and the European MSc degree in photonics from VUB, Ghent

University (UGent), and the University of St. Andrews in 2014. In 2019, he he is a member of the permanent staff of the Photonics Group. Since
was awarded a PhD by Ghent University for his work on second-order 2008, he has had a position as a full time professor. He is a lecturer for
nonlinear optical materials for silicon nitride photonic integrated circuits. four courses at Ghent University (Microphotonics, Advanced Photonics
Currently, he is working as a postdoctoral researcher in the Photonics Laboratory, Photonic Semiconductor Components, and Technology).
Research Group at Ghent University–IMEC. He coordinates the cleanroom activities of the research group and is
coordinator of the NAMIFAB Center of Expertise. His research focuses
Benjamin Wohlfeil is currently a manager in the Advanced Technology on the design, fabrication, and characterization of integrated photonic
Department at ADVA Optical Networking in Berlin and is involved in the devices. Main topics involve silicon nanophotonic devices and the inte-
photonic integration activities of ADVA. He received his Dipl.-Ing. degree gration of innovative materials (III–V, graphene, ferro-electrics, quantum
in computer science from Technische Universität Berlin in 2008 and the dots, etc.) on these waveguides to expand their functionality. He is work-
Dr.-Ing. degree in electrical engineering also from Technische Universität ing on applications for telecom, diatom, optical interconnect, and sensing.
Berlin in 2015. From 2014 to 2015, he worked as a postdoc in the He has submitted 14 patents, has authored and coauthored more than
Computational Nano-Optics Group at the Konrad-Zuse-Center for applied 220 journal papers, and has presented invited papers at all major confer-
mathematics. He has authored more than a dozen papers and is inventor ences in the domain. He is a member of IEEE Photonics Society, OSA,
of several patents. He is an active contributor to OIF and COBO stand- and SPIE. He has coordinated several European Projects (FP6 PICMOS,
ardization groups regarding coherent optical transceivers and is principal FP7 WADIMOS, and FP7 SMARTFIBER), contributed in many more, and
investigator in several national and European research projects. received an ERC Grant (ULPPIC). He received the prestigious “Laureaat
van de Vlaamse Academie Van België” prize in 2012 and was a Clarivate
highly cited researcher.
Despoina Petousi received her diploma in electrical and computer en-
gineering from Aristotle University of Thessaloniki in 2009 where she
received her master’s degree in nanosciences and nanotechnologies Roel Baets received his MSc degree in electrical engineering from UGent
in 2011. From 2011 until 2019 she was a researcher in the silicon pho- in 1980, his second MSc degree from Stanford in 1981, and his PhD from
tonics group of IHP, Leibniz Institut fuer Innovative Mikroelektronik, UGent in 1984. From 1984 to 1989, he held a postdoctoral position at
Frankfurt (Oder). In 2017, she received her PhD from Technische IMEC. Since 1989, he has been a professor at the Faculty of Engineering
Universitä Berlin under the supervision of Professor Klaus Petermann. and Architecture of UGent where he founded the Photonics Research
Her PhD thesis was focused on the development of integrated silicon Group. He is a full professor at Ghent University (UGent) and is associ-
photonic modulators with application in future coherent systems for opti- ated with IMEC. From 1990 to 1994, he was also a part-time professor at
cal communications. Currently, she is a research and development en- Delft University of Technology and, from 2004 to 2008, at Eindhoven
gineer in ADVA mainly focused on the field of photonic integration and University of Technology. He has mainly worked in the field of integrated
next generation transceivers for optical interconnects. photonics. He has made contributions to research on photonic integrated
circuits, both in III-V semiconductors and in silicon, as well as their
Bart Kuyken received his BSc degree in physics and his BSc and MSc applications in telecom, datacom, sensing, and medicine. Web of
degrees in electrical engineering from Ghent University in 2006 and 2008, Science reports more than 600 publications with an h-index more than 60.
As part of a team of eight professors, he leads the Photonics Research
respectively, and his additional MSc degree from Stanford University in
2009. After obtaining his PhD from Ghent University in 2013, he focused Group. With about 90 researchers, this group is involved in numerous
his postdoctoral work on integrated nonlinear circuits and comb genera- (inter)national research programs and has created four spin-off compa-
tion and was a visiting worker at the Max Planck Institute for Quantum nies. The silicon photonics activities of the group are part of a joint re-
search initiative with IMEC. He has led major research projects in silicon
Optics in 2013 and 2014. Currently, he is an associate professor at Ghent
University, where he is looking at electrically pumped comb sources photonics in Europe. In 2006, he founded ePIXfab, the first Multi-Project-
on chip. Wafer service for silicon photonics globally. Since then, ePIXfab has
evolved to become the European Silicon Photonics Alliance. He is also
a director of the multidisciplinary Center for Nano- and Biophotonics
Dries Van Thourhout received his MSc degree in physical engineering (NB Photonics) at UGent founded in 2010. He was a co-founder of
and his PhD from Ghent University, Ghent, Belgium, in 1995 and 2000,
the European MSc Program in photonics. He is an ERC grantee of the
respectively. From October 2000 to September 2002, he was at Lucent European Research Council and a Methusalem grantee of the Flemish
Technologies, Bell Laboratories, New Jersey, USA, working on the Government. He is a fellow of the IEEE, European Optical Society
design, processing, and characterization of InP/InGaAsP monolithically (EOS), and Optical Society (OSA). He is also a member of the Royal
integrated devices. In October 2002, he joined the Department of
Flemish Academy of Belgium for Sciences and the Arts.
Information Technology (INTEC), Ghent University, Belgium. Currently,


PDF 1

Uploaded by

Copyright:

Available Formats

PDF 1

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PDF 1

Uploaded by

Copyright:

Available Formats

Review Article

Taking silicon photonics modulators to a higher

1 Introduction of voltage V π applied to the phase shifter to achieve π phase

Advanced Photonics 024003-1 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-2 Mar∕Apr 2021 • Vol. 3(2)

The landscape of high-speed modulators in silicon photonics

Advanced Photonics 024003-3 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-4 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-5 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-6 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-7 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-8 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-9 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-10 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-11 Mar∕Apr 2021 • Vol. 3(2)

Modulation efficiency Loss Length of phase Data rateb

Advanced Photonics 024003-12 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-13 Mar∕Apr 2021 • Vol. 3(2)

Modulation efficiency Loss Length of phase Data ratea Energy/bit

III–V on Si as an amplitude modulatord ∼2 (2140) (4.9140) ≤0.1 ∼50 (50)140 -

Advanced Photonics 024003-14 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-15 Mar∕Apr 2021 • Vol. 3(2)

Modulation efficiency Length of phase Data ratea

Advanced Photonics 024003-16 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-17 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-18 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-19 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-20 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-21 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-22 Mar∕Apr 2021 • Vol. 3(2)

Advanced Photonics 024003-23 Mar∕Apr 2021 • Vol. 3(2)

You might also like