Project funded by the European Commission under grant agreement n°604391
GRAPHENE Graphene-Based Revolutions in ICT and Beyond
Combination of CP and CSA
WP5 Optoelectronics
Deliverable 5.3 “Report on photodetectors, lasers, OLEDs, modulators and metamaterials”
Main Author(s):
Frank Koppens, ICFO
Andrea Ferrari, UCAM
Due date of deliverable: M24
Actual submission date: M24
GRAPHENE D5.3 30 September 2015 2 / 17
LIST OF CONTRIBUTORS
Partner Acronym Partner Name Name of the contact
69 ICFO Fundacio Institut De Ciencies Fotoniques
Frank Koppens
28 UCAM The Chancellor, Masters and Scholars of The University of Cambridge
Andrea Ferrari, Daniel Popa
2 CNR Consiglio Nationale delle Ricerche Alessandro Tredicucci 70 TU-Wien Technische Universitaet Wien Thomas Mueller 6 UNIMAN The University of Manchester Alexander Grigorenko 130 IMEC Interuniversitair Micro-
electronicacentrum IMEC VZW Dries Van Thourhout
128 QMUL Queen Mary University of London Yang Hao 104 EPFL Ecole Polytechnique Federale de
Lausanne Juan Mosig
GRAPHENE D5.3 30 September 2015 3 / 17
TABLE OF CONTENTS Deliverable Summary ............................................................................................................................... 4 1. Photodetectors ..................................................................................................................................... 4
1.1 Photodetectors for visible and infrared frequencies ....................................................................... 4 1.2 Photodetectors for mid-infrared and THz frequencies ................................................................... 6 1.3 Detectors based on the thermoelectric effect ................................................................................ 7 1.4 New concepts ................................................................................................................................ 8 1.5 New materials ................................................................................................................................ 8
2. Chip-integrated Modulators .................................................................................................................. 9 3. Metamaterials ..................................................................................................................................... 11
3.1 Graphene based metamaterials for optical communications ....................................................... 12 3.2 Metamaterials as perfect absorbers ............................................................................................. 12 3.3 Plasmonic bio-sensors ................................................................................................................. 12 3.4 THz passive devices .................................................................................................................... 13
4. OLEDS ............................................................................................................................................... 13 5. Lasers ................................................................................................................................................ 14
5.1 Wavelength tuning range ............................................................................................................. 14 5.2 Output power ............................................................................................................................... 14 5.3 Repetition rate .............................................................................................................................. 15
References ............................................................................................................................................. 16
GRAPHENE D5.3 30 September 2015 4 / 17
Deliverable Summary
This report covers the integration of graphene/2d- material-based interconnected in
photodetectors, lasers, OLEDs, modulators and metamaterials, focussing on recent advances
since deliverable D5.2
1. Photodetectors
1.1 Photodetectors for visible and infrared frequencies
Photodetection of visible (VIS) and infrared (IR) radiation is of utter importance for a variety of different
applications, most notably, optical communications and (video) imaging. The VIS and IR wavelength
ranges are thus well covered by highly mature photodetection technologies and materials, such as
silicon (VIS; imaging) and compound semiconductors (IR; optical communications). Nevertheless, with
changing demands, the necessity for alternative photodetection materials is becoming more eminent.
E.g., the field of optical communications is shifting towards waveguide-integrated devices (e.g. “silicon
photonics”), where the integration of compound semiconductors is not considered a viable option.
Instead, germanium is typically used whose photodetection performance is comparably poor. For
imaging applications, highly sensitive, flexible and lightweight materials with spectral response
extending into the near-IR would be desirable.
As we will show in detail below, significant progress has been made in the performance of chip-
integrated (waveguide integrated) photodetectors, where competing performance has been shown in
terms of speed and responsivity. This technology is now close to ready for further integration with
graphene-based modulators (discussed below) towards the implementation of a transceiver bank.
The table with KPIs is listed below.
State-of-the-Art 2D materials
Speed 40GHz (waveguide-integrated Ge) 150 GHz (InGaAs) 80 GHz (Ge)
40 GHz (waveguide-integrated graphene) 260 GHz (graphene)
Wavelength range C-band (waveguide-integrated Ge) O-U bands (waveguide-integrated graphene)
Responsivity 0.5-1 A/W 0.36 A/W (waveguide-integrated graphene) 0.11 A/W (TMD)
Responsivity photoconductive
108 A/W (semiconductor) 107 A/W (graphene, slow (Hz)) 3×104 A/W (graphene, fast (kHz)) 106 A/W (TMD)
Flexibility 0.12 A/W (Si and III-V semiconductor membranes)
45.5 A/W (MoS2/graphene)
GRAPHENE D5.3 30 September 2015 5 / 17
1.1.1 Chip-integrated ultrafast photodetectors for optical communications
In order to optimize the ultrafast chip-integrated
photodetectors (PDs) that members of WP5 have
demonstrated previously, we have optimized the
performance of waveguide-integrated photodetectors [1]
and performed a number of ultrafast measurements, both
on graphene and van der Waals heterostructures [2].
In graphene, we have measured the photovoltage
generation time in pn-junctions and found it to be faster
than 50 fs. As a proof-of-principle application of this
ultrafast PD, we used graphene to directly measure, electrically, the pulse duration of a sub-50 fs laser
pulse. We have also examined the spectral response of our devices and found a constant spectral
responsivity between 500 and 1500 nm, consistent with the photo-thermoelectric (PTE) photovoltage
generation effect [2].
As the cooling dynamics of hot carriers determines the efficiency of PTE-based PDs, we are currently
in the progress of studying hBN-encapsulated 2d materials by femtosecond pump-probe experiments.
In a first set of experiments, we measured electron relaxation dynamics of a hBN/graphene/hBN stack
and compared it to graphene on Si or SiO2. Our results indicate a fast electron-optical phonon
relaxation time~200 fs, followed by a slow electron-phonon cooling time~2.8 ps. The cooling in the
hBN/graphene/hBN stack is slower than that observed on the SiO2 substrate, which confirms our
assumption that high-quality graphene will give stronger photoresponse due to reduction of disorder-
induced cooling.
On the device side, members of WP5 have fabricated and characterized a new generation of
waveguide-integrated PDs. These devices exploit the slower cooling dynamics of the electron gas in a
hBN/graphene/hBN stacks and thus show a high responsivity of up to 0.36 A/W – a more than three-
fold improvement over devices without hBN. An electrical bandwidth of more than 40 GHz was
achieved and the device was employed for the direct electrical measurement of ultrafast laser pulses
[1]. The responsivity can be further increased by a factor 2 to 3 by further optimization of the junction
(e.g. by local graphene gates). These results have shown that the most critical bottleneck
(responsivity) for waveguide-integrated detectors can be overcome. More focus will be geared towards
integration with modulators and further optimization of all relevant parameters in one device (including
contact resistance, parasitic capacitances etc.).
Beyond graphene, we studied the ultrafast carrier dynamics in graphene/WSe2/graphene vertical
stacks [3] and MoS2/WSe2 heterojunctions [4]. Although the mobility of transition metal dichalcogenides
(TMDs) is smaller than that of graphene, these devices benefit from the short carrier transit times in
ultrathin van der Waals heterojunctions. In the graphene/WSe2/graphene device, we demonstrated an
intrinsic photoresponse time as short as 5.5 ps, which is tuneable through bias and by varying the TMD
layer thickness. Our study has provided direct insight into the physical processes governing the
detection speed and quantum efficiency of these van der Waals (vdW) heterostructures, namely out-of-
GRAPHENE D5.3 30 September 2015 6 / 17
plane carrier drift and recombination. In the MoS2/WSe2 heterojunctions, we obtained carrier transfer
times lower than 0.2 ps. TMD-based devices thus combine the high photodetection efficiency of
semiconductors with a picosecond photoresponse comparable to that of graphene, thereby optimizing
both speed and efficiency in a single PD. Our observation and understanding of ultrafast and efficient
photodetection has underlined the potential of hybrid TMD-based heterostructures as a platform for
future optoelectronic devices.
1.1.2 (Flexible) photodetectors with high responsivity
As first demonstrated by members of WP5, photogain can be accomplished by integration of 2d
crystals with other materials to produce a hybrid phototransistor (e.g. by sensitizing the surface of a
graphene transistor with quantum dots). Although ultrahigh photogain and responsivities of up to 107
A/W can be achieved in these devices, they initially suffered from very slow photoresponse (≈Hz), but
in an optimized design, we now achieved sensitive photodetection with a bandwidth of up to 2 kHz,
sufficiently large for video imaging [4]. We fabricated field-effect transistors with CVD graphene on
SiO2/p-Si. After synthesis of CdS nanocrystal, their surface ligands are exchanged to
methylbenzenethiol and the NC are transferred in a mixture of chloroform and dichlorobenzene for
spin-coating onto the graphene device. Responsivities up to 3.4 104 A/W were reached. The
measured detector properties are related to the surface chemistry and trap states of the NCs. The use
of short ligands allows the photoexcited electrons to efficiently transfer to graphene. The recombination
is assisted by shallow and deep states, with different timescales. Thus, we have in our system both
short timescales, good for fast detection, and long timescale, for high responsivity
The flexible VIS photodetectors, reported already in D5.2, have further been optimized and
characterized [5]. These devices exhibit external (internal) responsivities as high as 45.5 A/W (455
A/W) with a photoconductive gain of 4 105, vastly outperforming state-of-the-art bulk-semiconductor
flexible photodetectors [6].
1.2 Photodetectors for mid-infrared and THz frequencies
Graphene is a very promising candidate for novel Terahertz PDs. The target here is the development
of a room-temperature technology that would allow sensitive THz cameras, capable of operating at fast
acquisition rates and with high dynamic range. We identified, as final performance goals for these
devices, noise equivalent powers (NEP) in the order of pW/Hz1/2 (in order to be used also in passive
imaging systems), with response speed of hundred MHz and above (for fast automatic in-line operation
in many applications, for instance in industrial process control). Even more important is the objective of
implementing a suitable approach to achieve narrow-band electrically tuneable detection, which would
allow spectrally resolved images even in passive configuration, something that no THz technology at
present can provide.
We reviewed graphene and related-materials based PDs in Nature Nanotechnology, describing the
state-of-the-art [7]. This was summarized in the previous deliverable D5.2. Here we will briefly describe
the major progresses and advances in the past 6 months.
GRAPHENE D5.3 30 September 2015 7 / 17
1.2.1 Monolayer and bilayer graphene plasma wave FETs
We reported room-temperature detection across the far-infrared in monolayer and bilayer graphene
employing an antenna-coupled configuration for the excitation of overdamped non-resonant plasma-
waves in the channel of a graphene FET [8, 9]. In these devices, the measured dependence of the
photovoltage Δu on the gate bias VG was in qualitative agreement with the prediction of a diffusive
theoretical model, proving that the detectors operate in the so-called broadband overdamped regime.
Nevertheless, discrepancies were observed that were attributed to a possible contribution of
thermoelectric origin. The two mechanisms, however, are hard to disentangle since the functional
dependence on VG is expected to be the same. Recent results have now clarified this issue [10]. The
responsivity of FETs made from bilayer graphene produced by thermal desorption of SiC was studied
as function of the impinging radiation in the range 260-360 GHz. By considering the behaviour at the
Dirac point, dominated by the thermoelectric contribution, the temperature heating was evaluated to be
~1 K; the thermoelectric response can then be calculated for every VG and added to the plasma non-
linearity. Experiments show that the two effects tend to cancel each other, thereby strongly supporting
device
1.2.2 Graphene-based broadband THz bolometers
A different detection concept makes use of the bolometric effect in which a radiation-induced
temperature variation is read through the resistance change of a graphene channel [11]. Here,
experiments were carried out at room temperature, still in antenna-coupled devices, demonstrating a
different behaviour for monolayer and bilayer graphene, and verifying the bolometric response through
measurements of the noise spectral function [12]. The estimated NEP values in these detectors in the
1.5-3 THz region are quite promising (in the 10 pW/Hz1/2 range), although the response time is very
slow (s). The latter feature seems to indicate that the graphene channel acts actually mostly as an
electrical read-out, while the absorption/heating involves directly the substrate.
1.3 Detectors based on the thermoelectric effect
Ref. 13 reported the first graphene device explicitly aimed at exploiting the thermoelectric effect as the
main detection mechanism [13]. In this direction the research is presently focusing on increasing the
THz absorption in the graphene layer to enhance responsivity. As the detecting element has to be
rather large to maximize temperature variations, the idea is to keep avoiding the antenna coupling,
since this works best for device sizes much smaller than the wavelength. A possible approach is to use
plasmon enhancement effects in a structure where resonant plasmon modes can be excited. For this
purpose a first demonstration was carried out using a detection element made of graphene
microribbons [14]. Although at the moment a significant performance improvement is still to be proven,
resonant enhancement of the THz absorption around 4-6 THz is evident from both the spectral and the
polarization dependence of the signal, making this a promising route to achieve higher responsivities.
GRAPHENE D5.3 30 September 2015 8 / 17
1.4 New concepts
New THz detection concepts have been proposed. The recent experimental studies on the propagation
of 2d plasmons in graphene, launched by a near-field coupling tip, has prompted a theoretical analysis
of the possibility of extracting an electrical signal from the excitation of such plasmon modes15. While
this could be achieved by using built-in electric fields to separate electron and holes, it is argued that in
a low-damping BN-graphene plasmonic waveguide an electric signal will anyway result from the
intrinsic hydrodynamic non-linearity.
Another proposal is to use quantum point contacts (QPC) [16]. In the tunnelling regime, the current
through the negatively biased barrier of the QPC device is very low but can be enlarged by orders of
magnitude by the radiation field that locally (thanks to electrode tips acting as antenna arms) "heats"
the electron gas. Preliminary measurements at 150 GHz in a GaAs device have shown responsivities
up to 106 V/W, which could even reach larger values in atomically thin graphene sheet geometries
suppressing one of the two mechanisms (Fig. 1).
1.5 New materials
Ideally the responsivity of non-resonant plasma wave detectors is proportional to the non-linearity of
the FET transfer characteristics and is at its maximum near the channel closing point. A FET based on
a 2d material with gap should then possibly yield better transconductance and hence better
responsivities when employed as detector. We have used thin flakes of black-phosphorus (BP) to
realize devices operating at room temperature and integrated on-chip with planar antennas providing
remarkable coupling efficiencies in the strongly sub-wavelength device channel [17]. The achieved
selective detection (~5-8 V/W responsivity, see Fig. 2) and sensitivity performances (signal-to-noise
ratio of 500), have been exploited to demonstrate the first application of a phosphorus-based active
THz device, for pharmaceutical and quality control imaging of macroscopic samples, in real-time and in
a realistic setting. Furthermore, by exploiting the inherent electrical and thermal in-plane anisotropy of
BP, plasmonic, thermoelectric and bolometric detection can all selectively be achieved in a controlled
fashion depending on the device geometrical design.
Fig. 1 Source/drain photovoltage signal of a bilayer graphene FET detector as a function of gate
voltage for two different radiation frequencies. In blue is the plasma wave contribution as computed
GRAPHENE D5.3 30 September 2015 9 / 17
from the transfer characteristics. In red is the computed thermoelectric contribution. In black the
experimental curve (which should result from the sum of the two).
Fig. 2 Left: Responsivity of BP detector as a function of gate voltage (red). In black is the voltage in
absence of THz radiation. Right: SEM picture of the BP FET detector.
2. Chip-integrated Modulators Light transmitted through a graphene layer can be modulated when gating with the appropriate voltage.
This is true over a very broad wavelength range, from the UV up to the MIR. However, when a light
beam is normally incident on a single sheet of graphene, only~2.3% of its power is absorbed, by far not
sufficient for realizing a device that can be used for practical applications in the datacom or telecom
domain. The efficiency can be enhanced by embedding the graphene layer in a Fabry-Perot cavity.
However, this introduces a strong wavelength dependence in the response of the device, negating the
intrinsic broad band response of graphene. An alternative is to integrate the graphene layer on top of a
waveguide. In this way the interaction length can be increased at will. As an example, when integrated
on a 220 nm x 600 nm Si waveguide, ~0.1 dB of the light (@ 1550nm) is absorbed per micrometre for
light propagating in the fundamental quasi TM-mode. In addition, when applying a voltage between the
graphene layer and the Si waveguide the Fermi level and hence the absorption in the graphene layer
can be controlled. This was demonstrated first in Ref.18 where a graphene based intensity modulator
with a bandwidth of 1 GHz and extinction ratio of ~0.08 dB/mm, operating over the wavelength range
1.35-1.60 mm was shown (Fig. 3a,b). In a follow up paper, the authors showed a variant of this device
whereby also the bottom electrode is formed by a graphene layer (see schematic view in Fig. b) [19].
This roughly doubles the extinction ratio, without increasing the capacitance of the device. An
additional advantage of this approach is that the waveguide no longer serves as the bottom electrode
and in principle can be realised from any dielectric material, e.g. silicon nitride. A similar device with
lower loss was demonstrated in Ref [8].
Since these first demonstrations several other groups demonstrated graphene based waveguide
electro-absorption modulators [20, 21, 22, 23, 24, 25, 26]. Recently also graphene based phase
modulation was demonstrated [27]. Also several theoretical studies were published [28, 29, 30]. Initial
devices were limited in operation speed. However, they were relying on non-intentionally doped or
lightly doped Si waveguides, increasing the resistance of the bottom electrode, hence limiting the RC-
0
2
4
6
-1 0 10
2x10-5
VG (V)
Res
pons
ivity
(V/W
)
Δu (µV
)
20
0
x"
y"
1.8 µm 1.0 µm G S
D
2 µm
LG
GRAPHENE D5.3 30 September 2015 10 / 17
bandwidth of the device. Recently an improved version of the device was demonstrated, relying on the
IMEC standard iSIPP25G platform, whereby the waveguide and the silicon contact area were
intentionally doped. This considerably improved the performance of the device, allowing demonstration
of open eye diagrams for data-rates up to 10 GBit/s (Fig. 4).
Fig. 3 a) Schematic view of and b) static transmission through a single layer graphene modulator
integrated on silicon waveguide as function of applied voltage (from [18]). c) On-off transmission of low
loss double-layer graphene modulator [31].
Fig. 4 SEM-picture of graphene modulator (left) and eye-diagrams at 8 and 10GB/s modulation (right)
[27].
The biggest trade-off in designing these modulators lies in the selection of the gate-oxide thickness.
Decreasing the drive voltage, as required for combining the modulators with advanced CMOS drivers,
requires reducing the gate oxide thickness. However, this increases the capacitance of the device and
hence decreases its bandwidth. One approach to overcome this trade-off is to enhance the optical
confinement in the graphene layer(s). This can be achieved by embedding a short modulator section
in a planar resonator, e.g. a ring resonator [22, 23] or a photonic crystal cavity [32], see also Fig. e. Up
to 30GHz modulation was recently shown for such a structure [23]. However, this again introduces
strong wavelength dependence in the response and the loss in the demonstrated structures was
unacceptably high. Alternatives that do not introduce this wavelength dependence are to embed the
graphene layer in the centre of a waveguide (Fig. 5c [28]) or to combine it with a metal contact, which
supports a plasmon-like mode (Fig. 5d [29]). Such configurations have however not yet been shown
experimentally.
GRAPHENE D5.3 30 September 2015 11 / 17
Fig. 5 Cross-section of a number of possible implementations for hybrid silicon-graphene devices (left)
and schematic view of graphene modulator on Silicon Photonic Crystal [32].
Table 1. Most relevant KPIs for current state-of-the-art silicon integrated modulators. (Ref. key:
(a)=[33], (b)=[34], (c)=[35], (d)=[18], (e)=[26]).
Graphene has been used extensively as the saturable absorber in modelocked lasers, allowing
generation of ultrashort pulse [36]. An all fibre-based modulator was demonstrated for that purpose
[37]. Recently, it was demonstrated that also the planar integrated device could be used to control the
saturable absorption in graphene [38]. This paves the way to fully integrated graphene mode locked
devices with controllable pulse shape.
The table below gives values for the most important KPI’s for currently demonstrated graphene
modulators and compares them with other types of modulators integrated on the Si platform. Graphene
modulators excel in footprint, optical bandwidth and operating temperature range. Power consumption
and operating bandwidth need be further improved.
3. Metamaterials Graphene optical metamaterials are composite materials produced from graphene, metals and
dielectrics, which are engineered into a composite structure designed to fit a particular purpose and
realize electromagnetic properties not easily available in natural materials. The main applications of
graphene metamaterials can be found in efficient light modulators, active optical elements, highly
birefringent materials and/or hyperbolic optical materials required for superlens operation, safeguarding
plasmonic properties of existing metals and semiconductors, realizing extremely high electromagnetic
fields and/or field confinement, perfect absorbers among others. One of the most interesting
GRAPHENE D5.3 30 September 2015 12 / 17
applications is modulation of guided light with the help of hybrid graphene/metal metasurfaces, which
could provide a breakthrough in optical communications as a viable alternative to silicon photonics. In
addition, cheaper and better SPR sensors can be produced using graphene protected copper and
silver plasmonics. Since progress in optical metamaterials has been achieved in various areas we will
concentrate our attention only on three promising topics.
3.1 Graphene based metamaterials for optical communications
Progress has been made in development of graphene metamaterials for on-chip hybrid graphene
plasmonic and silicon modulators at telecom wavelengths. Collective resonances have been used to
modulate light reflection from a graphene-metal metasurface by graphene gating [39] WP5 realized a
hybrid structure, which realizes coupled plasmonic resonances of unprecedented quality (Q~300) at
optical communication wavelength 1.5 µm (the standard quality of localized plasmon resonance in
analogous structures is about Q~10). This allowed us to achieve 20% of light modulation at telecom
wavelengths using graphene gating with one graphene layer in reflection – the largest number
achieved so far.
In addition, WP5 realized corrugated metasurfaces to achieve modulation~10% in 12 µm plasmonic
waveguides at telecom wavelengths at low gating voltages with an active device area of just 10 µm2,
characteristics which are already comparable with those of Si-based waveguide modulators while
retaining a benefit of further device miniaturization [40]. We systematically studied propagating
plasmons of various geometries and choose ones which promise the largest modulation for graphene-
based gating [40]. We achieved light guiding at the conditions of diffractive coupled resonances and
are going to use this technique for creation of graphene-based light modulators.
3.2 Metamaterials as perfect absorbers
As first demonstrated by members of WP5, perfect absorbers can be produced from graphene-metal
metasurfaces under various geometries. The perfect absorber based on topological darkness would
allow one to achieve extremely high phase sensitivity, which was demonstrated by graphene
hydrogenation [41]. We studied properties of “dark light” in details and we are preparing a review on
phase sensitivity in various devices and materials [41]. We showed that the phase sensitivity for dark
reflection can be better than 1 fg/mm2, while metamaterials based on graphene Cu multilayers can
provide even better mass sensitivity of 0.1 fg/mm2. In optimized designs, we combined plasmonic
nanoarrays with graphene to achieve zero reflection [42]. We also studied graphene based Fabry-
Perot structures with unusual light modulation due to graphene gating at the condition of zero
reflection. Various types of reflectors have been used, including Bragg mirrors, copper films, etc.
3.3 Plasmonic bio-sensors
We have shown that graphene plasmonics can be used to improve on a well-known molecule-
detection method: infrared absorption spectroscopy, which has important limitations when applied to
GRAPHENE D5.3 30 September 2015 13 / 17
Fig.6 Schematic of photoexcited electron kinetics in graphene, with possible relaxation mechanisms for the non-equilibrium electron population [50].
molecules at the nano-scale [43]. The wavelength of the infrared photon directed at a molecule
is~6µm, while the target measures only a few nm, making it very challenging to detect the vibration of
such a small molecule.
Graphene is capable of focusing light on a precise spot. On its surface the plasmons couple to the
vibration of a nanometric molecule that is attached to it. It is then possible to detect nanometric
compounds in proximity to the surface. In addition to identifying the presence of nanometric molecules,
this sensor is frequency selective and can therefore also reveal the nature of the bonds. The sensitivity
was compared to Au. Due to the significantly larger overlap with the optical mode, a much larger
sensitivity was obtained for the graphene plasmonic bio-sensor.
3.4 THz passive devices
A THz graphene isolator was developed, with non-reciprocity and gate-tuneability as unique aspects
compared to existing technologies [44]. Recently, we also designed several reflectarrays, simulated
with numerical tools developed for graphene devices [45, 46]. The simulations show that the selected
device can reach optimal performances according to our theoretical performance upper-bound [47].
We also demonstrated a graphene based polarization selective device operating at low THz
frequencies [48].
4. OLEDS Organic light-emitting diodes (OLED) are large area light sources that are applied as display in e.g.
mobile phones and televisions. Currently, a thin layer of brittle ITO is used as electrode material.
Graphene is promising as an alternative, especially in flexible applications. However, application in
OLEDs the sheet resistance needs to be less than 100 Ω/o and band alignment should be
accomplished to achieve a high injection efficiency. Using a 5 nm thick molybdenum trioxide (MoO3),
layer as a dopant, the desired work function increase
as well as sheet resistance decrease have been
obtained. The graphene monolayer-based OLED is
somewhat electrically inferior due to the limited
conductivity of the monolayer. However, this is
compensated by a higher current efficiency of 55 cd/A
at 1000 cd/m2 leading to overall superior power
efficiency compared to the state-of-the-art ITO
reference device. Also, the optical transmission of the
device is over 90%, which is an improvement over ITO
that can be attributed to the absence of guiding modes
that cause light loss in the case of ITO.
OLEDs should be protected from exposure to air or
water. For flexible OLEDs this is an issue that currently
asks for a complicated hybrid organic-inorganic
GRAPHENE D5.3 30 September 2015 14 / 17
Fig.7 Few-cycle (28 fs) pulse generated from a graphene all-fiber laser with >50 mW output power [64].
material barrier layer. Graphene is also interesting as alternative for this barrier layer due to its low gas
permeability flexible OLEDs. From determination of the water vapour transmission rate it is evident
that graphene currently still lacks underperforms. The permeation pattern shows close similarity with
the defect pattern on CVD graphene, indicating that not intrinsic properties of graphene are limiting the
performance, but growth and processing related defects. It is therefore expected that improvements in
these areas will lead to improvement of large area graphene barrier properties.
5. Lasers Ultrafast lasers are relevant for many scientific, commercial, and industrial applications. The current
technology is based on Semiconductor Saturable Absorber Mirrors (SESAMs). SESAMs are multi-
quantum well heterostructures. Their fabrication involves expensive epitaxial growth and packaging.
Post-growth ion implantation is required to reduce the response time. Each design typically has a small
bandwidth (~100 nm) and can cover only a narrow wavelength range (~800-1600 nm). The linear
dispersion of the Dirac electrons in graphene offers an ideal solution: for any excitation there is always
an e–h pair in resonance (Fig. 6). The ultrafast carrier dynamics combined with large absorption and
Pauli blocking, make graphene an ideal ultrabroadband, fast Saturable Absorber (SA) [49, 50].
5.1 Wavelength tuning range
To date, Graphene-SAs (GSAs) have been demonstrated for pulse generation at ~800 nm [51], ~970
nm [52], 1 µm [53], 1.2 µm [54], 1.5 µm [55], 2 µm [56],
2.4 µm [57] and 3 µm [58]. A widely tuneable fibre laser
mode-locked with a GSA has been reported [59, 60].
The laser produces ps pulses in a tuning range 1525–
1559 nm, demonstrating its “full-band” operation
performance. A single GSA operating in a range that
covers from 1 to 2 µm was demonstrated [56, 61]. Mode-
locking at 1, 1.5 and 2 µm was achieved by integrating a
graphene composite into ytterbium- [61], erbium- [61],
and thulium-doped [56] fibre laser cavities respectively.
A synchronism of dual-wavelength pulse-trains through
the inclusion of a common graphene-SA in a coupled
cavity configuration was demonstrated [61]. This allowed getting sum frequency generation. A strong
coherent emission was detected at ~800 nm through degenerate four wave mixing, extending the
wavelength operation range.
5.2 Output power
The average output power has increased from a few mW in fibre lasers [55] to over 10 W in
semiconductor lasers [62]. The highest output power to date, ~10W, was achieved in a VECSEL by
GRAPHENE D5.3 30 September 2015 15 / 17
Fig.8 Mid-IR ~ 8 GHz pulses from a graphene mode-locked laser [66].
using a graphene integrated mirror [62]. Other advances towards increasing the output power have
been achieved in all-fibre configurations able to generate over 50 mW output power [63]. The majority
of graphene mode-locked fibre lasers employ polymer composites, because of their simplicity of device
fabrication. However, the average power generated by these lasers is limited to few-mW, mainly due to
optical damage induce into the polymer matrix. By using a polymer-free graphene-SA, Ref. [63]
demonstrated for the first time power scaling up to ~50 mW, one order of magnitude higher than typical
graphene-based fibre lasers. This was only limited by the experimental setup, e.g., the limited power of
the pump laser, with a higher pump power, higher output powers being possible. A few-optical cycle
laser generating ~29 fs pulses with >50 mW output power was demonstrated (Fig. 7) [64]. The duration
of an optical pulse can be measured by counting the optical oscillations of the electric field within the
pulse envelope, with the single oscillation cycle setting the minimum possible duration. By using an all-
fibre setup mode-locked by a graphene-SA, the generation of only 6-cycles of light is possible [64].
This is the shortest pulse duration generated to date, outperforming most conventional technologies in
terms of low-cost, simple to fabricate and broadband. The laser is also able to generate >50 mW
output power, outperforming the current graphene-based fibre lasers.
5.3 Repetition rate
The highest repetition rates to date are ~ 2.5 GHz at
~970 nm in a semiconductor laser [52], ~ 10 GHz at 1.5
µm in a fibre cavity [65] and ~ 8 GHz at 2 µm in a
monolithic waveguide laser [66], by using a single layer
CVD graphene [52] and graphene film on an output
coupler [65, 66], respectively (Fig. 8).
A quantum theory of the nonlinear (third-order)
electrodynamic response of graphene is being
developed, with results on the third harmonic
generation and preliminary results on the optical Kerr effect (saturable absorption) [67]. An experiment
on the gate-voltage tuneable third harmonic generation at near-IR frequencies is planned.
GRAPHENE D5.3 30 September 2015 16 / 17
References
1 Shiue et al., arXiv:1507.00426v3.
2 Tielrooij et al., Nature Nanotechnol. 10, 437 (2015)
3 Massicotte et al., arXiv:1507.06251.
4 Höller et al., manuscript in preparation.
5 De Fazio et al., manuscript in preparation.
6 Yang et al., Appl. Phys. Lett. 96, 121107 (2010).
7 Koppens F.H.L. et al., Nature Nanotechnol. 9, 780 (2014).
8 Vicarelli L. et al., Nature Mat. 11, 865-871, (2012).
9 Spirito D. et al. Appl. Phys. Lett. 104, 061111 (2014).
10 Bianco F. et al., Appl. Phys. Lett., in press (2015).
11 Mittendorff M. et al., Appl. Phys. Lett. 103, 021113 (2013).
12 Mahjoub A. M., et al., Appl. Phys. Lett. 107, 083506 (2015).
13 Cai X. et al., Nature Nanotechnol. 9, 814–819 (2014).
14 Cai X. et al., Nano Lett. 15, 4295 (2015).
15 Torre I. et al., Phys. Rev. B 91, 081402(R) (2015).
16 Levin A.D. et al., Appl. Phys. Lett. 107, 072112 (2015).
17 Viti L. et al. Adv. Materials, in press (2015).
18 Liu M. et al., Nature, 474(7349), 64–67 (2011).
19 Liu M. et al., Nano Lett.,10–13, Feb. (2012).
20 Youngblood N. et al., Nano Lett. 14(5), 2741–2746 (2014).
21 Gruhler N. et al., Opt. Express 21(25) 31678 (2013).
22 Qiu C. et al., Nano Lett., 6811-6815 (2014).
23 Phare C.T. et al. Nat. Photonics, July, 1–5 (2015).
24 Li H. et al., Opt. Express 20(18), 20330 (2012).
25 Mohsin M. et al. Opt. Express 22(12), 15292 (2014).
26 Hu Y. et al., Electron Devices Meeting (IEDM), United States, (2014); Koester S.J. and Li M., Appl.
Phys. Lett. 100(17), 171107 (2012).
27 Mohsin M. e.a., Scientific Reports, 5:10967 (2015)
28 Lu Z. and Zhao W., JOSA B 29(6), 1490–1496 (2012).
29 Ye C. et al, IEEE J. Sel. Top. Quantum Electron 20(4), 3400310 (2014).
30 Sorianello, V. et al, OPTICS EXPRESS 23 (5), 6478-6490 (2015)
31 Mohsin M. et al., Opt. Express 22(12), 15292 (2014).
32 Gao Y. et al., Nano Lett 15(3), 2001-2005 (2015).
33 Streshinsky M. et al., Opt. Express 21, 30350- 30357 (2013).
34 Timurdogan E. et al., Nat. Commun. 5 (2014).
35 Feng D., et al., Opt. Express 20(20), 22224-22232 (2012).
GRAPHENE D5.3 30 September 2015 17 / 17
36 Popa D. et al. Appl. Phys. Lett. 97(20) 203106 (2010).
37 Lee, E.J. et al. Nature Commun. 6 (2015).
38 Alexander et al. “Electrically Controllable Saturable Absorption in Hybrid Graphene-Silicon
Waveguides” 2, 4–5.
39 Thackray B. D. et al., Nano Lett. 15, 3519 (2015).
40 Ansell D. et al., Nature commun., under review.
41 Kravets V. G. et al., Nature Mat. 12, 304 (2013); Kabashin A. V. et al., Nature Materials, in
preparation,
42 Ferrari A. C. et al., Nanoscale, 7, 4598 (2015).
43 Rodrigo D. et al., Science 349, 165-168 (2015).
44 Tamagnone M. et al., “Near optimal graphene terahertz non-reciprocal isolator”, under review.
45 Hasani H. et al. "Tri-band, Polarization-Independent Reflectarray at Terahertz Frequencies: Design,
Fabrication and Measurement", under review.
46 Gómez-Díaz J.S. et al. Nature Commun. 6, 6334 (2015).
47 Tamagnone M. et al. Nature Photon. 8, 556-563 (2014).
48 Li Y. et al. Optics Exp 23(6) 7227-7236 (2015).
49 Sun Z. et al., ACS Nano 4, 803 (2010).
50 Bonaccorso F. et al., Nature Photon. 51 Baek I. H. et al., Appl. Phys. Exp. 5, 032701 (2012). 52 Zaugg C. A. et al., Opt. Exp. 21, 31548 (2013). 53 Mary R. et al., Opt. Express 21, 7943 (2013). 54 Cho W. B. et al., Opt. Lett. 36, 4089 (2011). 55 Popa D. et al., Appl. Phys. Lett. 97, 203106 (2010). 56 Zhang M. et al., Opt. Exp. 20, 25077 (2012). 57 Cizmeciyan M. N. et al., Opt. Lett. 38, 341 (2013). 58 Wei C. et al., Opt. Lett. 38, 3233 (2013). 59 Sun Z. et al., Nano Res. 3, 653 (2010). 60 Popa D. et al., Apl. Phys. Lett. 98, 073106 (2011) 61 Zhang M. et al., CLEO, STu1I.2 (2014). 62 Husaini S. et al., Appl. Phys. Lett. 104, 161107 (2014). 63 Jiang Z. et al., CLEO, JTu4A.67 (2014). 64 Purdie D. et al., Appl. Phys. Lett. 106, 253101 (2015). 65 Martinez A. et al., Appl. Phys. Lett. 101, 041118 (2012). 66 Yingying R. et al., IEEE J. Sel. Top. Quantum Electron. 21, 1 (2015). 67 Mikhailov S. A., Phys. Rev. B 90, 241301(R) (2014); Phys. Rev. B 91, 039904(E) (2015).