Advances in graphene-based optoelectronics, plasmonics and photonics

Since the early days of graphene physics, the idea has emerged of graphene-based electronics as a new, very promising direction of high technologies. Geim and Novoselov [1] have predicted that at the time when Si-based technology is approaching its fundamental limits, graphene would be an exceptional candidate material to take over from Si. Soon after, Avouris et al [2] investigated the structure and function of graphene nanoribbon transistors and also discussed graphene nanoribbon field-effect transistors. Subsequently, the first observation of current saturation in zero-bandgap, top-gated graphene field-effect transistors was reported by Shephard et al [3], and Rogers [4] discussed the synthesis of ultrathin films of reduced graphene oxide with large area and their possible utilization in flexible electronics and other applications. Ryzhii et al investigated the tunneling current-voltage characteristics of graphene and graphene nanoribbon field-effect transistors [5, 6], the device model for graphene bilayer field-effect transistors [7], high-frequency properties of graphene nanoribbon field-effect transistors [8] and an analytical device model for graphene bilayer field-effect transistors, using a weak nonlocality approximation [9]. In reference [10] Duan et al demonstrated the fabrication of high-speed graphene transistors with a self-aligned nanowire gate, a channel length as low as 140 nm, and the highest scaled on-current and transconductance yet reported. In a short communication [11] Avouris et al presented the fabrication of a field-effect transistor on a 2-inch graphene wafer with a cutoff frequency in the radio frequency range, as high as 100 GHz. A comprehensive review of graphene transistors has been performed by Schwierz and was published in reference [12].

Following the above-presented research works on graphene-based electronics, experimental investigations of graphene-based optoelectronic, plasmonic and photonic devices, including graphene-based solar cells, were also rapidly developed. The purpose of the present work is to review the main achievements of this investigation.

In the subsequent section 2 we review the research on graphene-based optoelectronics. The presentation on graphene-based plasmonics is the content of section 3. Section 4 is devoted to the review of research into graphene-based photonic materials and devices. The new progress in the fabrication of graphene-based Schottky junction solar cells is presented in section 5. Section 6 is the conclusion.

Encouraged by the exceptional optical properties of graphene, in reference [13] Avouris et al have explored the use of zero-bandgap, large-area graphene field-effect transistors (FETs) as ultrafast photodetectors. On light absorption, the generated electron-hole pairs in graphene would normally recombine on a time scale of tens of picoseconds, depending on the quality and carrier concentration of the graphene. If an external field is applied, the pairs can be separated and a photocurrent is generated. The same happens in the presence of an internal field formed near the metal electrode-graphene interface. The authors have demonstrated that this internal field can be used to produce an ultrafast photocurrent response in graphene. Owing to the high carrier transport velocity existing even under a moderate E-field, no direct bias voltage between source and drain is needed to ensure ultrafast and efficient (6–16% internal quantum efficiency within the photodetection region) photocurrent generation.

Photocurrent generation experiments were performed at both low and high light intensity modulation frequencies. At or close to the short-circuit condition, the magnitude of the photocurrent strongly depends on the location of the optical illumination and also on the gate bias. To generate a photocurrent in an external circuit, the photogenerated carriers must exit from the photogeneration region before they recombine, resulting in reasonably good internal efficiency (6–16%) within the high E-field photodetection region. Thus the authors have demonstrated ultrahigh-bandwidth photodetectors using single- and few-layer graphene. In these novel photodetectors, the interaction of photons and graphene, the properties of photogenerated carriers, and the transport of photocarriers are fundamentally different from those in conventional group IV and III–V semiconductors. These unique properties of graphene enable very high bandwidth (potentially >500 GHz) light detection, very wide wavelength detection range, zero dark current operation and good internal quantum efficiency.

One year later Xia, Avouris et al [14] reported again the use, for the first time, of a graphene-based photodetector in a 10 Gbit s⁻¹ optical data link. In this interdigitated metal-graphene-metal (MGM) photodetector, an asymmetric metallic scheme was adopted to break the mirror symmetry of the internal E-field profile in conventional graphene FET channels [13], allowing for more efficient photodetection. This was a simple vertical-incidence MGM photodetector with external responsivity of 6.1 mA W⁻¹ at an operating wavelength of 1.55 μm, and represented a 15-fold improvement compared to that reported by the authors in their previous work [13].

The new MGM photodetectors were fabricated on highly resistive silicon wafer with a thick layer of thermal oxide and with geometry similar to that of traditional metal-semiconductor-metal (MSM) detectors. Flakes of single-, bi- and tri-layer graphene were identified and confirmed by Raman spectroscopy, and interdigitated electrodes were then fabricated. One set of fingers was made of Pd/Au and the other-of Ti/Au. The detector was connected with contact pads.

In the graphene FET photodetectors fabricated by the authors in the previous work [13], the internal (built-in) electrical fields responsible for the separation of the photogenerated carriers exist only in narrow regions (∼0.2 μm) adjacent to the electrode/graphene interfaces, where charge transfer between metal and graphene leads to band bending. The absence of a strong electric field in the bulk graphene sheet, where most electron-hole pairs are generated, leads to carrier recombination without contribution to the external photocurrent. In the present work, multiple interdigitated metal fingers are used, leading to the creation of a greatly enlarged, high E-field, light-detection region. However, if both electrodes consist of the same metal, the build-in electric field profile in the channel between two neighbouring fingers is symmetric, and the total photocurrent vanishes. In this experiment the authors demonstrated that an asymmetric metalization scheme can be used to break the mirror symmetry of the built-in potential profile within the channel, allowing for the individual contributions to be summed to give the overall photocurrent.

A broad-band, high-speed, waveguide-integrated electroabsorption modulator based on monolayer graphene has been demonstrated by Wang, Zhang et al [15] for the first time. In this device the modulation is performed by actively tuning the Fermi level of a monolayer graphene sheet. This modulator has following advantages: (1) strong light-graphene interaction, (2) broad-band operation, (3) high-speed operation, (4) compatibility with complementary metal-oxide semiconductor (CMOS) processing.

To fabricate this device, a 50 nm thick Si layer was used to connect the 250 nm thick Si bus waveguide and one of the electrodes. Both silicon layer and waveguide were shallowly doped with boron to reduce the sheet resistance. A spacer of 7 nm thick Al₂O₃ was then uniformly deposited on the surface of the waveguide by atom layer deposition. A graphene sheet grown by chemical vapor deposition (CVD) was then mechanically transferred onto the Si waveguide. To reduce the access resistance of the device, the counter electrode was extended towards the bus waveguide by depositing a platinum (10 nm) film on top of graphene layer. The minimum distance between platinum electrode and waveguide remained undisturbed by the platinum contact. To further improve the electroabsorption modulation efficiency, the silicon waveguide was designed to have the electric field maximized at its top and bottom surfaces, so that interband transitions in the graphene were maximized. As graphene only interacts with the tangential (in-plane) electric field of electromagnetic waves, the graphene modulator is polarization-sensitive.

To measure the dynamic response of the graphene modulator, radio frequency signals generated by a network analyser were added on a static drive voltage ${{\rm{V}}}_{D}$ and applied to the modulator. A 1.53 μm laser was used to test the modulator and the out-coupled light was sent to a high-speed photodetector. The V_D -dependent radio frequency response of the graphene modulator was measured, and gigahertz operation of the device at various driver voltages was performed.

In brief, the authors have demonstrated a graphene-based optical modulator that has broad optical bandwidth (1.35–1.6 μm), small device footprint (25 μm²) and high operation speed (1.2 GHz at 3 dB) under ambient conditions, all of which are essential for optical interconnection. The modulation efficiency of a single-layer graphene sheet is already comparable to, if not better than, traditional materials such as Si, GeSi and InGaAs, which are orders of magnitude larger in active volume. The flexibility of graphene sheets could be also exploited for the fabrication of radically different photonic devices.

Having in mind the integration of the priorities of a graphene photodetector with efficient complementary metal-oxide semiconductor (CMOS) technology, Wang, Xu et al [16] have demonstrated an ultrawide-band CMOS-compatible photodetector based on graphene. The device fabrication consisted of three steps: etching and passivation of the silicon waveguide, deposition and structuring of graphene, and metallization.

In a device of proper length L, the optical mode is almost completely absorbed as the light propagates along the silicon waveguide. The local potential gradient at the interface between the central Ti/Au electrode (signal electrode S) and the graphene layer drive a photocurrent towards the ground (GND) lead. A potential gradient was originated from different dopings in the metal covered and uncovered parts of graphene and additionally could be enhanced by utilizing the waveguide itself as a back-gate electrode to modulate the potential in the graphene channel. A GND-S-GND configuration was used, which allows a doubling of the total photocurrent. Owing to the lack of an electronic bandgap in graphene, the photogenerated carriers pass through the potential barrier at the GND electrodes almost unimpeded, leading to high-bandwidth photodetection even without S-GND bias.

The fraction η of light absorbed in the graphene sheet was calculated. The results showed that efficient light absorption (η > 50%) can be achieved with short device lengths, which enable high-speed operation and dense operation capability. The photoresponsivity S – defined as the ratio of the measured photocurrent to the input power source – can attain the value S ≈ 0.05 A W⁻¹ in the best device prepared from trilayer graphene, which is an order of magnitude larger than that achieved with normal-incidence graphene photodetectors.

Finally the authors summarized the opportunities that graphene offers as a new material for optical interconnects :

• Ultrawide-band operation,
• High-speed operation,
• Low energy consumption,
• Small device footprint
• Compatibilities with CMOS and other technologies
• Simplicity and low cost.

The device had the following structure: Monolayer graphene samples were prepared by standard mechanical exfoliation and transferred to the waveguide. The suspended membrane waveguide was necessary to avoid mid-infrared light by the buried oxide (BOX) and to take full advantage of the transparent wavelength region of silicon, which covers the 1.2–8.0 μm range. Two gold electrodes were fabricated above the graphene and silicon waveguide with a gap of ∼1.5 μm.

The photoresponses were measured using three different types of light sources: visible white light, a commercial tunable laser operating at a wavelength of 1.55 μm for telecommunications, and a mid-infrared fibre laser at 2.75 μm. In the near-infrared region the photodetector was characterized by the narrow linewidth tunable laser. A fibre polarization controller was employed to change the polarization. The transverse electric mode light was coupled into the waveguide via an apodized focusing subwavelength grating. The bias-dependent photoresponse was measured. Distinct from the bipolar white-light photocurrent, the photoresponse was only observable for forward bias (silicon was biased positive with respect to graphene). For the mid-infrared characterization, a single-end forward-pumped ${E}_{r}^{3+}-{P}_{r}^{3+}$co-doped zirconium, barium, lanthanium, aluminium and sodium fluoride fibre laser was used to excite the photodetector. Remarkably, the photocurrent-to dark-current ratio under a −1.5 V bias was larger than 30, which is 15 times larger than that in the near-infrared case.

In brief, the authors have designed and experimentally fabricated a graphene/silicon heterostructure waveguide photodetector, and have observed that the in-plane coupled waveguide can enhance significantly the graphene-light interaction. The heterostructure efficiently suppressed the dark current and enhanced the mid-infrared absorbance. These photodetectors exhibited extremely large ON/OFF current ratio from the visible light to the mid-infrared range. The high responsivity, low dark current and spatial selectivity herald a myriad of applications.

Beside the integration of graphene priorities with efficient CMOS technology, there exists another way to improve the graphene photodetector by integrating graphene onto a silicon optical waveguide on silicon-on-insulator (SOI) material. Following this method Mueller et al [17] demonstrated a graphene/silicon-heterostructure waveguide photodetector on SOI that operated from the visible to mid-infrared spectral range, benefited from a naturally formed graphene/silicon heterostructure and showed a low dark current because of the existence of the junction potential barrier.

In order to overcome the low photoresponsivity of graphene due to its weak optical absorption, Englund et al [18] have demonstrated a waveguide-integrated graphene photodetector that simultaneously exhibits high responsivity, high speed and broad spectral bandwidth. These authors showed that by integrating a graphene photodetector onto a SOI bus waveguide, it is possible to greatly enhance graphene absorption and corresponding photodetection efficiency without sacrificing the high speed and broad spectral bandwidth.

The fabricated device has following structure. A silicon waveguide is backfilled with SiO₂ and then planarized to provide a smooth surface for the deposition of graphene. A thin SiO₂ layer deposited on the planarized chip electrically isolates the graphene layer from the underlying silicon structures. The optical waveguide mode couples to the graphene layer through the evanescent field, leading to optical absorption and the generation of photocarriers. Two metal electrodes located on opposite sides of the waveguide collect the photocurrent. One of these electrodes is positioned ∼100 nm from the edge of the waveguide to create a lateral metal-doped junction that overlaps with the waveguide mode. The junction is close enough to the waveguide to efficiently separate the photoexcited electron-hole pairs at zero bias, but the metal contact-waveguide separation of 100 nm is still far enough to ensure that the optical absorption is dominated by graphene.

Spatially resolved photocurrent measurements were used to confirm the integrity of the metal-doped graphene junction. By deconvolving the photocurrent with the spot size of the excitation laser and numerically integrating it along a line, a relative potential profile across the graphene channel was obtained. The results showed that the graphene has potential gradients around the boundaries of the gold electrodes, yielding the corresponding internal electric field. The graphene beneath the two metal contacts had the same p-type doping level, which was lower than the intrinsic doping of graphene channel. Therefore, band bending with opposing gradient occurred at the two electrode junctions. Unlike the case in conventional semiconductors, both electrons and holes in graphene have very high mobility, and a moderate internal electric field allowed ultrafast and efficient photocarrier separation.

In brief, the authors have demonstrated a high-performance waveguide-integrated graphene photodetector. The extended interaction length between the graphene and the silicon waveguide optical mode resulted in a notable photodetection responsivity of 0.108 A W⁻¹, which approached that of commercial non-avalanche photodetectors. However, the presented device can work with an ultrafast dynamic response at zero-bias operation, allowing low on-chip power consumption.

Although graphene is a good photoconductive material for optical detection due to its broad absorption spectrum and ultrashort response time, it remains a challenge to achieve high responsivity in graphene detectors because of the weak optical absorption and short photocarrier lifetime of graphene. Capasso et al [19] have designed and fabricated an antenna-assisted graphene detector, where optical antennae are used as both light-harvesting components and electrodes to simultaneously enhance light absorption and carrier collection efficiency.

The electrical field intensity enhancement distribution at the antenna resonant wavelength is calculated by finite difference time domain (FDTD) simulations. The optoelectronic characterization of the graphene detectors was performed and the photovoltage maps of the antenna-assisted graphene detector as well as of the reference detector with the same graphene sheet size and contact pads but without antenna were recorded. The wavelength-dependent responsivity of the antenna-assisted graphene detector is measured. As a result of the resonant nature of plasmonic antennae, the responsivity (photovoltage divided by the total incident power on the sample) exhibits a strong wavelength dependence. The detector responsivity is also dependent on the bias of the detector, because the source-drain bias influences the electrical field within the graphene channel between adjacent antenna electrodes. Moreover, the antenna-assisted graphene detector shows a linear photoresponse as the incident light power increases up to 16 mW, indicating that the absorption is not saturated despite the strong field enhancement in the antenna gaps. The time response of the detectors was also measured. It is worth noting that the use of metallic optical antennae to simultaneously enhance the optical absorption and photocarrier collection efficiency in graphene detectors have achieved the successful fabrication of room-temperature mid-IR antenna-assisted graphene detectors with more than 200 times enhancement of responsivity compared to reference devices without antennae.

Although graphene is a highly promising semiconducting material for high-speed, broad-band and multicolor detection, for utilization in fabricating photodetectors it has a drawback: it lacks a bandgap. Therefore there arises the necessity to create the p-and n-regions in graphene and the p-n junctions. Ren, Bao et al [20] have reported a technique for preparing a large-area photodetector on the basis of the controlable fabrication of graphene p-n junctions. The authors have incorporated a new efficient n-type dopant to the chemical vapor deposition (CVD)-grown graphene to enable large area, flexible and transparent IR photodetectors. They demonstrated that charge transfer doping of CVD-grown graphene can be achieved in selective regions to prepare a large number of p-n junctions. The formation of the p-n junction is found to be crucial in determining the polarity and amplitude of the photoresponse in the devices to be fabricated. Furthermore, because no gate voltage is needed to tune the charge carrier density, the charge transfer doped p-n junctions can thus be fabricated onto any substrate, leading to a fully transparent and flexible photodetector. The presence of graphene p-n junctions fabricated by spatially selective n-doping was confirmed by electrical measurements.

The applied efficient and patternable chemical doping technique allowed the authors to prepare large area thin-film photodetectors by forming controlled p-n junctions. Two types of devices were fabricated: a long-channel device (50 μm in length, ∼1 cm in width) and a short-channel device (3 μm in length, 160 μm in width). Note that two geometries featured a significant difference in the ratio between the p-n junction region and the homogeneously doped region, which crucially affects the photoresponse of the devices.

Because the chemical doping-generated p-n junction does not require either the gate or dielectric layers, the device fabrication can easily be accomplished to prepare an ultrathin all-transparent flexible photodetector [21]. The transparent photodetector was fabricated on a flexible polyethylene terephthalate (PET) substrate with indium tin oxide (ITO) as electrodes. The device showed a transmittance greater than 90% over the wavelength range of 400–2000 nm.

In brief, the authors have developed a technique to fabricate large-area, flexible and transparent graphene photodetectors. This was enabled via controlled fabrication of a p-n junction on CVD-grown graphene. Contrarily to most other graphene-based IR photodetectors, the device reported by the authors was fabricated through a selected-area chemical doping process. Together with the broad-band adsorption, the chemically doped CVD-grown graphene photodetector can be fabricated on a large scale. However, the exact mechanisms of the photoresponses in the fabricated device deserve future investigation.

As a semiconducting material with a particular two-dimensional structure, graphene is ideally suited for the integration with planar photonic devices, and the performance of the devices significantly benefits from the elongated optical interaction length in the coplanar configuration [16–18]. With this remark Li et al [22] have fully utilized graphene's extraordinary and tunable optoelectronic properties to demonstrate the first optoelectronic device that acts as both a modulator and a photodetector, where the functionality of the device can be controlled with an integrated electrostatic gate also prepared from graphene separated by a dielectrical layer and integrated on a planarized silicon photonic waveguide. The configuration of the device is that of a simple field-effect transistor (FET): the bottom layer (the channel) acts as an optical absorber and can collect photogenerated carriers, while the top layer acts as a transparent gate electrode, which can tune the electrical and optical properties of the bottom graphene layer. The graphene is grown by chemical vapor deposition (CVD) on copper foil and transferred onto the photonic waveguide substrate. The dielectric layer between the gate and the channel is a thick (100 nm) aluminum oxide (Al₂O₃) one deposited by atomic layer deposition (ALD). The source and drain contacts are made of titanium/gold and palladium/gold which have different work functions and dope graphene n-type and p-type, respectively. The differential metal-graphene contacts induce a lateral p-i-n junction if the middle of the graphene channel is tuned to the charge neutral point (CNP). This allows the device to generate a net photocurrent without the application of a bias voltage and with a higher efficiency than the device with a single-side configuration.

The FET configuration allows the authors to characterize the electrical properties of the graphene channel. The results show that the charge neutral point is reached when a gate voltage of V_g = +33 V is applied, indicating that the graphene channel is heavily p-doped with a hole concentration of p = 1.4 × 10¹³ cm⁻¹ and corresponding Fermi level of E_F = −0.45 eV. This level of doping is relatively high for graphene grown by CVD method and can be attributed to the trapped positive charge at the dielectric interface. Fitting the resistance versus V_g results in an extracted carrier mobility in the graphene of 1150 cm² V⁻¹ s⁻¹, which is relatively low and attributed to disorder introduced by Al₂O₃ deposition and charge trapping in the dielectric.

The transmission spectrum of the Mach-Zehnder interferometer before the graphene layers were integrated on the waveguide was recorded. The interference fringes show an extinction ratio (ER) higher than 40 dB (ER = T_max/T_min, T_max and T_min being the transmission at peaks and valleys, respectively), confirming that there is negligible excess optical loss (less than 0.1 dB) in the interferometer arm. During the fabrication of the device, the ER of the interferometer was measured after every step so that the optical loss caused by each layer can be accounted for. When the device was completed, the ER decreased to 1.6 when zero gate voltage was applied, corresponding to an added loss of 18 dB in the device arm. When voltage was applied to the top graphene gate, the extinction ratio of the interefence fringes was modulated. The authors observed that ER increased (decreased) when positive (negative) gate voltage was applied, indicating reduced (augmented) absorption in the graphene. The authors measured the ER at every step of the applied gate voltage and calculated the linear absorption coefficient in the bottom graphene layer. Knowing graphene's absorption coefficient α, the internal quantum efficiency η of the photodetector can be determined.

Thus the authors have demonstrated a novel multifunctional optoelectronic device based on graphene and integrated on a photonic waveguide that can be operated as both an optical modulator and a photodetector and can be tuned with a gate voltage. The optical absorption and the photocurrent are simultaneously modulated by the gate voltage. While the photocurrent should be proportional to the absorbed optical power and thus approximately proportional to the absorption coefficient, it is also sensitive to the field distribution in the graphene channel which is modulated by the gate. The device can be operated in an unprecedented mode of simultaneous optical modulation and photodetection.

The simplest configuration in various recently proposed photodetection schemes and architectures is the metal-graphene-metal (MGM) photodetector (PD), in which graphene is contacted with metal electrodes as the source and drain [23–26]. These PDs can be combined with metal nanostructures enabling local surface plasmons and increased absorption, realizing the enhancement in responsivity. However, Ferrari et al [27] have remarked that the precise mechanism of the photodetection is still debated, and these authors presented the study of wavelength and polarization-dependent metal-graphene-metal photodetectors. On the basis of this study the authors were able to quantify and control the relative contributions of both photothermoelectric and photoelectric effects, both adding to the overall photoresponse.

MGM-PDs play an important role because they are easy to fabricate, not relying on nanoscale lithography. They operate over a broad wavelength range as the light–matter interaction is mostly determined by graphene itself. Furthermore, ultrahigh operating speed can be achieved as no bandwidth limiting materials are employed. Each MGP-PD consists of a graphene channel contacted by two electrodes of the same metal or two different metals. The difference in work function between the metal pads and graphene leads to charge transfer with a consequent shift of the graphene Fermi level in the region below the metal pads. The Fermi level gradually moves back to that of the uncontacted graphene when crossing from the metal covered region to the metal-free channel. This results in a potential gradient extending ∼100-200 nm from the end of the metal pad to the metal-free graphene channel. This inhomogeneous doping profile creates a junction along the channel. In principle this can be a p-n, n-n or p-p junction between the graphene underneath and within the channel, as the channel Fermi level can be controlled by a back gate.

Currently, two effects are thought to contribute to the photoresponse in graphene-based PDs, both requiring spatially inhomogeneous doping profiles: photothermoelectric and photoelectric. The photothermoelectric effect results from local heating of, e.g., the p-n junction due to the incident light power. The photoelectric effect is as important as the photothermoelectric effect. The potential gradient within the junction separates the photoinduced e-h pairs and leads to a current flow as in a conventional photodiode. The authors investigated the wavelength and polarization dependent responsivity of MGM-PDs. The measured light polarization dependent responsivity, combined with the spatial origin of the photoresponse obtained from photovoltage maps, allowed the authors to determine the photoresponse mechanisms and quantitatively attribute it to photothermoelectric and photoelectric effect.

To further investigate the influence of thermoelectric and photoelectric effects on the overall photovoltage, the authors performed polarization-dependent measurements. Photovoltage maps were acquired at different polarization angles of the incident light. The plots of photovoltage showed two contributions: one polarization dependent, and another polarization independent. The polarization-dependent contribution was assigned to the photoelectric effect due to the polarization-dependent interband optical excitations. Thus the authors have demonstrated the influence of the orientation of the lateral p-n junction in graphene-based photodetectors with respect to the polarization of incident linearly polarized light. The angular dependence was in good agreement with theory and showed that both photothermoelectric and photoelectric effects contribute to the photoresponse in MGM-PDs, with photoelectric effects becoming more pronounced at longer wavelengths.

Having in mind the variety of exceptional electronic and photonic properties of graphene and taking advantage of the mature platform of fiber optics, in reference [28] Tong et al have demonstrated a graphene-clad microfiber (GCM) all-optical modulator at ∼1.5 μm (the C-band of optical communication) with a response time of ∼2.2 ps limited only by the intrinsic graphene response time. The modulation comes from the enhanced light-graphene interaction due to the optical field confined to the wave guiding microfiber and can reach a modulation depth of 38%. The prepared GCM all-optical modulator has the following structure: A thin graphene layer is wrapped around a single-mode microfiber, which is a section with the ends tapered down from a standard telecom optical fiber. The principle of the GCM modulator is as follows: A weak infrared signal wave coupled into the GCM experiences significant attenuation due to absorption in graphene as it propagates along. When a switch light is introduced, it excites carriers in the graphene and through Pauli blocking of interband transition it shifts the absorption threshold of graphene to a higher frequency, resulting in a much lower attenuation of the signal wave. The switch light leads to modulation of the signal output from the fiber, and its response time is limited by the relaxation of the excited carriers.

The GCM structure enables significant enhancement of light-graphene interaction via tightly confined evanescent field guided along the surface of the microfiber. To see how graphene cladding affects the light transmission through a microfiber the authors launched a continuous-wave (CW) broadband light through a GCM. The light power was kept low enough so that the absorption of graphene did not change. The transmission spectrum of GCM was compared with that of the bare microfiber. In the spectral range of 600–1600 nm the bare microfiber has nearly constant transmittance, while GCM has an absorption increasing with the increase of wavelength, which can be explained by the evanescent field for longer wavelength at the graphene interface. The observed absorption of the GCM was an order of magnitude higher than that of a bilayer graphene, because of the large effective interaction length.

At higher light intensities, the band filling (Pauli blocking) effect of the excited carriers can drastically change the absorption spectrum of graphene. At a peak power density below ∼0.2 GW cm⁻², absorption of graphene is in the linear range, leading to a nearly constant transmittance of 15.5%. When the density exceeds 1 GW cm⁻² the transmittance increases rapidly due to the saturable absorption, which saturates as the density approaches ∼2.5 GW cm⁻² to yield a transmittance of ∼24%. The strong pump effect on the absorption of GCM can be readily employed for all-optical modulation. The authors showed that nanosecond pump pulses can be used to switch out signal pulses from a GCM. The signal transmittance depends on the pump intensity.

In reference [29] Liu et al have extended the results presented in their previous work [15], designed and experimentally demonstrated a double-layer graphene optical modulator. This device has a structure similar to the forward/reverse-biased silicon modulator [30] in which the doped silicon is replaced by intrinsic/predoped graphene, removing the insertion loss due to the doped silicon waveguide. Both electrons and holes are injected into the graphene layer to form a p-n like junction, and the optical loss from silicon can be reduced to minimum. This device has an advantage owing to the unique linear band dispersion of graphene with a symmetrical density of states near the Dirac point. Because the interband transition coefficient in graphene is only determined by |E_F| but not the sign of E_F, both graphene layers can become transparent simultaneously at high drive voltage and the device is thus at 'on' state. Such design avoids the participation of electrons/holes in silicon and therefore its operation speed will only be determined by the carrier mobility in graphene. In addition, using two graphene layers for the active medium can further increase the optical absorption and modulation depth, leading to advantages such as a smaller footprint and lower power consumption.

Silicon-on-insulator (SOI) wafers were used in the fabrication process. A wide silicon waveguide with both ends connected to a pair of grating couplers was fabricated using deep reactive-ion etching (DRIE). Atomic layer deposition (ALD) technique was then employed to conformally coat a thick Al₂O₃ isolation layer to prevent potential carrier injection from the bottom graphene layer into the silicon. The chip-sized graphene sheet prepared on Cu film by CVD method was first protected by a poly (methyl metacrylate) (PMMA) film which was baked at 110 °C for 10 min. After removing Cu film by FeCl₃ solution, the graphene sheet was then rinsed and transferred on to the waveguide for overnight baking. E-beam lithography was then used to prepare the active region, and oxygen plasma was applied to remove undesired graphene on one side of the waveguide, leaving the other side for metalization.

Direct deposition of high dielectric constant material through ALD growth on pristine graphene is challenging owing to the hydrophobic nature of graphene basal plane. Therefore the authors deposited aluminum onto the bottom graphene layer, which was immediately oxidized into Al₂O₃ upon exposure to the air. Finally the top graphene layers were mechanically transferred onto the dies forming the desired capacition structure. Subsequently similar patterning and etching procedures were performed to define the active tuning areas of graphene and top metal electrode.

The static optical transmission of the device was measured at the wavelength 1537 nm under different drive voltage. To measure the dynamic response of the modulator, an electric signal generated by a network analyser was superimposed onto a static drive voltage for small signal measurement. To optimize the modulation depth of the device, different waveguide widths were numerically analysed.

For a long time it has been known [31] that a layer of graphene can absorb only 2.3% of the power of the incident light due to its short interaction length. This weak optical absorption is detrimental to active optoelectronic devices. In order to overcome this difficulty Mueller et al [32] have employed a graphene microcavity photodetector (GMPD) with a large increase of the optical field inside a resonant cavity, giving rise to increased absorption. The field enhancement occurs only at the designed wavelength, whereas the radiations with off-resonant wavelengths are rejected by the cavity making these devices promising for wavelength division multiplexing (WDM) systems.

In the fabricated device there are two distributed Bragg mirrors consisting of quarter-wavelength thick layers of alternating materials with varying refractive indices and forming a high-finesse planar cavity. Bragg mirrors are ideal choices for microcavity optoelectronic devices because unlike metal mirrors their reflectivity can be very well controlled and can reach values near unity. The Bragg mirrors are prepared of large bandgap materials that are non-absorbing at the detection wavelength. The absorbing graphene layer is sandwiched between these mirrors. A buffer layer ensures that the maximum of the field amplitude occurs right at the position where the graphene sheet is placed. The response of the conventional device is approximately independent of wavelength, but more than an order of magnitude weaker than that of the microcavity enhanced device.

It is worth noting that the concept of enhancing the light–matter interaction in graphene by use of an optical microcavity is not limited to photodetectors alone. It can be applied to a variety of other devices such as electroabsorption modulators, variable optical attenuators, and possibly future light emitters.

Having noted that graphene plasmons provide a suitable alternative to noble-metal plasmons, because they exhibit much tighter confinement and relatively long propagation distances with the advantage of being highly tunable via electrostatic gating, in reference [33] Koppens et al have proposed to use graphene plasmons as a platform for strongly enhanced light–matter interactions. On the basis of the theoretical study of the interaction between a quantum emitter and single surface plasmons (SPs) in graphene, these authors showed that extreme mode confinement yields ultrafast and efficient decay of the emitter into single SPs of a proximate doped graphene sheet. By analyzing the confinement in two-dimensional homogeneous graphene, the authors have found an increased degree of field enhancement and interaction strength. The authors indicated that graphene opens up a novel route to quantum plasmonics and quantum devices that have so far been difficult to achieve in conventional plasmonics.

In brief, the authors have described powerful and versatile building blocks for advanced graphene plasmonic circuits. These ideas take advantage of the unique combination of extreme field confinement, device tunability and patterning, and low losses that emerge from the remarkable structure of graphene and current experimental capabilities for fabrication. These advances are expected to both remove a number of obstacles facing traditional metal plasmonic and facilitate new possibilities for manipulating light–matter interactions at the nanoscale down to the single-SP level. The simultaneous large bandwidths and field enhancement, for example, should enable novel low power, ultrafast classical or quantum optical devices.

The direct application of graphene in optoelectronics devices is challenging due to the small thickness of graphene sheets and their resultant weak interaction with light. In reference [34] Capasso et al demonstrated the combination of metal and graphene in a hybrid plasmonic structure for enhancing graphene-light interaction and thus in situ controlled the optical response. The optical conductivity of graphene includes the contributions from both interband and intraband transitions. When the Fermi level is increased above half of the photon energy, the interband transitions are blocked, and the dominant intraband ones are highly sensitive to the charge carrier concentration in the graphene sheet; therefore the graphene optical conductivity and permitivity show a strong dependence on the gate voltage making graphene a promising electrically tunable plasmonic material.

The authors exploited graphene tunable optical properties in the intraband-transition-dominated region to achieve electrical tuning of the optic antennae while suppressing the interband absorption in graphene. Although the optical response of graphene is widely tunable, the resonances of plasmonic structures combined with graphene typically exhibit very limited tuning ranges due to the fact that the graphene layer is atomically thin and thus only interacts with a small portion of the plasmonic mode. To improve the graphene-light interaction, the authors incorporated graphene in the nanogap of the end-to-end antennae, where the electrical field is greatly enhanced. Using such a structure with a 20 nm gap size, the authors have developed an antenna design strategy to enhance the interaction of plasmonic mode with underlying graphene along the antenna length and demonstrated antenna structure with a resonance wavelength tuning range of 1100 nm – an increase of almost six times compared with that of a single antenna.

On the basis of the performed design, the authors fabricated the tunable plasmonic device with the following schematic structure: A graphene monolayer grown by atmospheric pressure chemical vapor deposition (CVD) was transferred onto a 30 nm thermal oxide layer of a highly p-doped silicon substrate. A square area of optic antennae and metal contacts was patterned onto the graphene sheet by electron beam lithography (EBL), electron beam evaporation, and lift-off. For probing and bonding purposes, Ti/Au pads are evaporated onto the oxide layer, overlapping with the Pd/Au contacts. Then the gate contact Ti/Au is evaporated onto the backside of the silicon substrate.

The reflectance of the device was measured using a Fourier transform infrared (FTIR) spectrometer with a mid-infrared (MIR) microscope. The time response of the device was characterized by measuring frequency-dependent optical modulation at a fixed wavelength. To explore the factors determining the modulation speed, the authors developed a small-signal, high-frequency circuit model of the device.

Thus the authors have designed and fabricated a new type of plasmonic structure comprising closely coupled optical antennae such that field localization occurs along a significant portion of the antenna length rather than only at the ends. The authors showed that this type of structure interacts particularly strongly with monolayer graphene and that its plasmonic modes are significantly affected by the graphene optical properties which can be dynamically controlled by electrostatic doping. The antenna resonance wavelength can be tuned as much as 1100 nm. This type of metal-graphene structure can be used for tunable sensors, reconfigurable metasurfaces, optical modulators and switches.

In reference [35] Basov et al have implemented a nanospectroscopic infrared local probe via a scattering scanning near-field optical microsope (s-SNOM) under intense near-infrared (NIR) laser excitation to investigate exfoliated graphene single-layers on SiO₂ at technologically significant mid-infrared (MIR) frequencies, where the local optical conductivity becomes experimentally accessible. The authors explored the ultrafast response of Dirac fermions in graphene and showed that the plasmonic effects in graphene can be modified on ultrafast time scales with an efficiency rivaling that of electrostatic gating. The authors analyzed the temporal evolution of the near-field plasmonic response by measuring the spectrally integrated scattering amplitude and briefly outlined the key features revealed by the temporal profile of the pump-probe data.

The authors have reported near-field pump-probe spectroscopy based on s-SNOM combining exceptional spatial, spectral and temporal resolution. The ultrafast s-SNOM was capable of probing a broad spectral region from visible to far-infrared energies and revealed ultrafast optical modulation of the infrared plasmonic response of graphene. The pulse energies needed to modify the infrared plasmonic response are two orders of magnitude smaller than that what is typically necessary for comparable ultrafast switching times in metal-based plasmonic structures at NIR frequencies.

The tunable optical properties of single layer graphene (SLG) due to the Pauli blocking of interband transitions in this semiconducting material was exploited by Boltasseva et al [36] in a graphene-nanoantenna hybrid device where a Fano resonance plasmonic nanostructure was fabricated on the top of a graphene sheet. The use of Fano resonant elements enhances the interaction of incident light with the graphene sheet and enables efficient electrical modulation of the plasmonic resonance.

In their experimental work the authors fabricated a graphene field-effect transistor (FET) by transfering a chemical vapor deposition (CVD) grown single layer graphene (SLG) onto a highly p-doped Si/SiO₂ substrate. Thereafter the authors fabricated the Fano resonant dolmen structures on top of the SLG. This enabled the authors to exploit the large sensitivity of resonance to the local environment and also to achieve electrical control. The optical properties of graphene depend strongly on the carrier density in the graphene sheet. When the graphene sheet is doped, some of interband transitions are blocked and the absorption of graphene exhibits step-like behavior around the interband threshold.

To verify the hypothesis that Fano resonant structures interact strongly with SLG, the authors measured the reflectance from the antenna at four different locations with and without an underlying SLG, and observed a strong impact of the graphene on the measured spectra. The measured data showed a saturation effect, wherein the spectra do not significantly change at large carrier concentrations. This clearly indicated that the graphene carrier concentration around the gold antennas shows a much smaller degree of variation than the changes expected from freestanding graphene. Another direction for improving the tunability of the plasmonic resonance is using several layers of graphene, which have higher optical conductivity, therefore leading to a stronger impact on plasmonic resonance. The achieved results significantly improved on those in a previous work of the authors.

With the purpose to fabricate far-infrared graphene plasmonic crystals for plasmonic band engineering, Ham et al [37] have employed a hexagonal array of apertures in a graphene sheet. This periodic structure perturbation of a continuous graphene medium alters delocalized plamonic dynamics, leading to the formation of plasmonic band structure in a manner akin to photonic crystals. This was demonstrated by resonantly coupling a far-infrared light into particular plasmon modes belonging to a unique set of plasmonic bands, where the light selects these specific modes because the spatial symmetry of the radiation field matched that of the plasmons within these modes.

There may be a variety of methods to introduce the structural periodicity in a continuous graphene medium. The hexagonal lattice of apertures is a proof-of-concept realization of the medium periodicity. To demonstrate the plasmonic band formation in the graphene plasmonic crystal, the authors performed Fourier transform infrared (FTIR) spectroscopy by normally irradiating an unpolarized far-infrared plane wave along the z-axis onto the device lying in the x-y plane.

The symmetry-based selection rule was experimentally proved. The hexagonal lattice possesses the C_6v symmetry point group and thus each T-point mode hosted by the lattice exhibits definite symmetry transformation properties under any symmetry operation belonging to the C_6v group. However, only a few energy bands have the symmetry transformation properties matched those of normally incident plane waves and therefore can interact with the lattice.

Having focused on the intrinsic properties of the graphene-plasmonic nanostructures and overcome the practical limitations in fabrication and device architectures, in reference [38] Iyer, Borondies et al demonstrated a simple two-step method to fabricate large-area freestanding graphene-gold (LFG-Au) nanostructures as well as investigating the plasmonic activity and localized metal-graphene interactions at the nanoscale of the devices. The surface-enhanced Raman scattering (SERS) of the as-prepared LFG-Au structure showed a nine-fold and six-fold enhancement at the 2D (2690 cm⁻¹) and G (1582 cm⁻¹) Raman band, respectively, due to the localized surface plasmon confinement in nanocracks formed in the freestanding Au film. LFG-Au plasmonic nanostructures were fabricated by coupling graphene with the underlying self-assembled array of Au-nanoparticles formed by thermal disintegration of the Au film. The electronic configurations in graphene due to the localized graphene surface-plasmon-metal interactions were reported.

The plasmonic nanostructures were realized by thermally assisted fragmentation of homogeneous metal thin films into nanoparticles (NPs). The near-field confinement in such NPs is known to depend on their size, morphology, and interparticulate separation. Graphene has been widely used as a sensing material to study the plasmonic activity in these structures via surface enhanced Raman scattering (SERS). The as-prepared LFG-Au samples are annealed at various temperatures in Ar atmosphere to form self-assembled Au NPs, which couple with LFG to form LFG-Au plasmonic nanostructures.

The chemical and electronic inhomogeneity across LFG, due to graphene-Au wrapping and the localized graphene-Au interfacial interaction, was further probed by synchrotron-based nano-spectro-microscope technique. The optical density (OD) data were obtained by converting the transmission data considering the l/l₀ ratio, where l is the transmitted photon flux through the sample and l₀ is the incident flux measured at a clear region (free of sample). The spatially resolved near-edge x-ray absorption fine structure (NEXAFS) K-edge spectra of the LFG were extracted from the OD mapping. The samples showed a π* transition at 285 eV and a broad σ* resonance at 291.5 eV. The extracted NEXAFS spectra provided a detailed spatial map of specific unoccupied electronic states such as the π* and the σ* above the Fermi level along with the pre-edge. The positions, relative intensities, shapes and linewidths of these resonances can be used to understand the local chemical and electronic structure of the material under study. The thickness of LFG was monitored by considering the difference in the pre-and post-edge of the extracted NEXAFS spectra from the OD mapping; here the edge-step OD of LFG ∼0.007 was determined. It was the smallest OD experimentally measured for a single graphene layer so far.

Thus in the as-prepared (at room temperature) LFG-Au samples, SERS enhancement is mainly due to the near-field confinement from the nanocracks between the metal islands in the Au film. The enhanced intensity of the D, G and 2D Raman bands validated the SERS enhancement in graphene due to the gold surface plasmon resonance. Further, the red-shift of the 2D band coupled with the emergence of a prominent π* peak in the LFG-Au films indicated strain-induced corrugations in the sample due to gold deposition. The enhanced interaction between Au NPs and graphene led to p-type doping in LFG, which caused an electronic and chemical inhomogeneity in the suspended LFG.

In conclusion, two distinct enhancement phenomena were observed in freestanding graphene–Au film: enhancement through the metal nanogaps via graphene and through strong interactions between thermally formed Au NPs and LFG, leading to a unique graphene surface plasmon resonance.

With the purpose to study the plasmonic enhancement phenomena at a graphene single layer, in reference [39] Kim, Planken et al have performed an experiment to observe the broad-band THz emission from a single layer of graphene excited by femtosecond near-infrared laser pulses. The authors have clarified how the excitation of the surface plasmon resonance (SPR) enhances the THz emission. The experimental results showed that for graphene deposited on a glass substrate, the amplitude of the emitted THz electric field strongly varied and even reversed the sign when the pump-beam polarization direction changed.

The graphene layers were directly synthesized by a chemical vapor deposition (CVD) system and transferred onto a glass slide as well as a thin Au film on a glass substrate. The experiments were performed using a standard THz time-domain spectroscopy setup based on electro-optic sampling. Typical time traces of the detected THz electric field emitted from a single layer graphene on glass were measured in a transmission setup. In general, the THz emission from a single layer of graphene is fairly weak. To enhance the THz emission, both types of propagating and localized SPR excitations at graphene/metal interfaces can be used. In the first case of propagating SPR, the reflected pump power reached a minimum at the incident angle of ∼45°. This incident angle was called the SPR angle, which was sensitive to the surface conditions of the Au layer: the SPR angle shifted from 44.60° (without graphene) to 44.84° (with graphene). In the second case, the local pump intensity increased by the excitation of SPR on semicontinuous percolating Au film. This strong field increase played a major role in the enhancement of the THz emission from a single layer of graphene. It was shown that for graphene deposited on thin Au film, the emitted THz power was significantly enhanced by two orders of magnitude when both propagating and localized SPR were excited.

A typical well-known luminescent nanophotonic device is the quantum dot (QD). With the purpose to find organic materials with superior photovoltaic characteristics Gupta et al [40] have prepared a conjugated polymer blended with graphene quantum dot (GQD) exhibiting a significant enhancement of organic photovoltaic (OPV) characteristics compared to the corresponding conjugated polymer graphene sheet blends. For solar cell applications the authors have functionalized GQDs with aniline (ANI) to form ANI-GQDs. For organic light emitting diode (OLED) applications the authors used fluorescent poly (2-methoxy-5-(2-ethylhexyloxy)-1, 4 phenylenevinylene) (MEH-PPV) polymer mixed with nonfluorescent methylene blue (MB) dye to form the devices denoted MB-GQDs.

The UV–vis absorption spectra of GQDs, ANI-GQDs and MB-GQDs were measured. The photoluminescent spectra of the films of poly (3-hexylthiophene-2, 5 diyl) (P3HT) blended with ANI-GQDs as well as of the films of MEH-PPV blended with MB-GQDs were also recorded. The hybrid solar cells based on P3HT/GQDs and the OLED devices based on MEH-PPV, MEH-PPV/GQDs and MEH-PPV/MB-GQDs were fabricated. The authors have shown that the GQDs dispersed in conjugated polymers show enhanced OPV and OLED characteristic compared to graphene sheets due to improved morphological and optical characteristics.

The bottom-up fabrication of photoluminescent GQDs with uniform morphology was performed by Liu, Muellen et al [41]. Regarding hexa-peri-hexabenzocoronene (HBC) as a nanoscale fragment of graphene [42], the authors have fabricated multicolor photoluminescent disk-like GQDs with the uniform size of ∼60 nm diameter and 2–3 nm thickness by using unsubstituted HBC as a carbon source. The powder of this starting material was pyrolyzed at 600, 900 and 1200 °C, and the final products will be denoted GQD-600, GQD-900 and GQD-1200. The surface functionalization of these devices was realized by using oligometric polyethylene glycol (PEG) diamin and enabled them to exhibit very good dispersibility in water.

The morphology of GQDs was characterized by atom force microscopy (AFM). It was found that GQD-600 consisted mainly of disordered particles, while GQD-900 contained both particles and disk-shaped nanosheets. For GQD-1200 homogeneous nanodisks of ∼60 nm diameter and ∼2.3 nm thickness were observed. The thickness of these nanodisks is three to four times higher than that of reduced graphene oxide, suggesting that they contain more than one layer of graphene. Dynamic light scattering (DLS) and transmission electron microscopy (TEM) studies of GQD-1200 further confirmed the disk-like morphology of this device.

Bearing in mind that the optical properties of GQDs hold the key for their future applications in optoelectronic devices and biological sensors, the authors recorded the UV–vis absorption and photoluminescence (PL) emission spectra of GQD-1200. The GQD-1200 suspension showed a broad UV–vis absorption with a weak shoulder at 280 nm, similar to chemically reduced graphene. Photoluminescence emission spectra indicated that GQDs can emit strong blue radiations under excitation of 365 nm. When the excitation wavelength changed from 320 to 480 nm, the PL peak correspondingly shifted from 430 to 560 nm. The bright and colorful PL may be attributed to the chemical nature of the graphene edges, although the exact mechanisms responsible for the PL from GQDs, especially blue to ultraviolet emission, remain to be explained.

With the purpose to improve electronic and photonic properties of GQDs, Qu et al [43] have fabricated N-doped GQDs with O-rich functional group. By using N-containing tetrabutylamonium perchlorate (TBAP) in acetrontrile as the electrolyte to introduce N atoms into the resultant GQDs in situ, the authors have modified the electrochemical approach reported in their previous work [44] for preparing N-free GQDs.

The solution of prepared N-GQDs exhibited a long-term homogeneous phase without any noticeable precipitation. Transmission electron microscopy (TEM) images showed fairly uniform N-GQDs with diameters of ∼2–5 nm, much smaller than those of the N-free counterparts synthesized hydrothermally (∼10 nm) but very consistent with those of N-free GQDs prepared electrochemically. The corresponding atomic force microscopy (AFM) image revealed a typical topographic height of 1–2.5 nm, suggesting that most of N-GQDs consist of ca 1–5 graphene layers. High-resolution TEM observations confirmed a 0.34 nm interlayer spacing for the few-layer N-GQDs.

X-ray photoelectron spectroscopy (XPS) measurements were performed to determine the composition of the prepared N-GQDs. It was observed that the O/C atomic ratio for the N-GQDs is ca 27%, similar to N-free GQDs and higher than that of the graphene film (ca 15%). This confirmed the successful incorporation of N-atom into the GQDs by electrochemical cycling in the N-containing electrolyte. In addition to the C-N bond, the high-resolution C_1s spectrum of the N-GQDs further confirmed the presence of the O-rich groups such as C-O, C=O and O-C=0, which is consistent with the corresponding Fourier transform infrared (FTIR) spectra.

The UV–vis absorption spectrum of the resultant N-GQDs showed an absorption band at ∼270 nm, which is blue-shifted by ∼50 nm with respect to that of N-free GQDs of similar size. Under the irradiation by a 365 nm lamp the N-GQDs emitted intense blue luminescence, which is different from the green luminescence of the N-free counterparts. It was shown that the O-rich groups as well as the relatively strong electron affinity of N-atom in the N-GQDs contributed to the PL blue shift.

Raman spectra of the original graphene film, N-free GQDs and N-GQDs were measured and compared. The peaks centered at 1365 and 1596 cm⁻¹ are attributed to the D and G bands, respectively, of carbon materials. It was observed that both the N-GQDs and their N-free counterparts have an I_D/I_G ratio of ∼0.7, much lower than that of the original graphene film (∼1.05), indicating that relatively high quality GQDs were prepared by the electrochemical method.

Apart from the specific luminescence properties of N-GQDs, they possess also the electrocatalytic activity for the oxygen reduction reaction (ORR). The authors used a large-area, electrically conductive graphene assembly to support N-GQDs as ORR catalysts. The graphene-supported N-GQDs (N-GQDs/G) were prepared by hydrothermal treatment of a suspension of well-dispersed graphene oxides with N-GQDs. Unlike the Pt/C electrode, the N-GQDs/G electrode exhibited a stable ORR in the methanol-containing electrolyte.

Thus the authors have developed a simple yet effective electrochemical strategy for fabricating N-GQDs with O-rich functional groups, which showed specific optoelectronic features distinctive from those of their N-free counterparts. N-GQD is a metal-free catalyst for the ORR, its specific luminescence properties indicate the potential for use in bioimaging, light-emitting diodes etc.

Since the main characteristics of GQDs depend on their size, for tailoring these characteristics to certain purposes Lee, Rhee et al [45] have demonstrated an efficient approach to prepare size-controlled GQDs via amidative cutting of tattered graphite. In this approach GQDs are synthesized from readily accessible micrometer-sized graphene via two consecutive steps. First, graphite was mildly oxidized with nitric acid (tattering), resulting in graphite flakes of few hundreds of nanometers in size, so-called 'tattered' graphite. Subsequently, tattered graphite was subject to primary amines with long aliphatic chains such as oleylamine (OAm) in an organic medium, followed by in situ hydrazine (N₂H₄) treatment to reduce excess oxygenic carbons. In this step, the size of GQDs can be readily controlled by varying the concentration of OAm, as shown by transmission electron microscopy (TEM). The high resolution TEM images indicated that the GQDs were highly crystalline with a lattice spacing of 0.21 nm (100). From Raman spectroscopy the authors detected the G and D bands with intensity ratio (G/D) around the unity. The x-ray photoelectron spectroscopy (XPS) measurements revealed the C=C, C–O, C=O, O=CO and C–N bondings. The photoluminescence spectra were measured to investigate the energy levels in GQDs. The Kelvin probe analyses showed that the Fermi level of all GQDs were around 4.74 eV – almost constant regardless of their sizes. To explore 'viable' electronic transitions between energy levels, the authors plotted the absorption spectra versus photon energy.

Finally the authors demonstrated organic light emitting diodes (OLEDs) employing 4, 4'-bis (carbazol-9-yl) biphenyl (CBP) as the host and a series of GQDs as dopants. The authors also noted that the prepared GQDs have several advantages such as proper energy-band structures and good organic solubility. The external quantum efficiency (EQE) of the best device was ∼0.1%.

Thus the authors have demonstrated the synthesis of a range of GQDs with certain size distribution via amidative cutting of tattered graphite. The power of this approach is that the size of GQDs could be varied from 2 to over 10 nm by simply regulating the amine concentration. The energy gaps in such GQDs were narrowed down by increasing the size, having shown colorful photoluminescence from blue to brown. The authors have also revealed that the defects play important roles in developing low-energy emission and reducing exciton lifetime through a series of optical analyses. In the practical aspect, the prepared GQDs have several advantages such as high solubility in common organic solvents and almost no undesirable agglomeration between themselves. To ultilize such advantages, the OLEDs employing GQDs as the dopant were demonstrated throughout studies of their energy levels, successfully having rendered white light with the EQE of ∼0.1%.

Pursuant to the strategy of enhancing the optical properties of graphene oxide (GO) by using the functionalization method, Saha et al [46] have functionalized GO sheets with aminoazobenzene (AAB) ligand in such a manner that the diazonium cation was bound to the active carbon centers of the phenolic moieties located at the edges, and amino groups were attached to the active carbon centers of the epoxy moieties on the basal plane of the GO nanosheets, having resulted in the formation of a layered type structure. The synthesized layered AAB-GO material exhibited strong and stable green luminescence emission via surface passivation and the excited-state intermolecular proton transfer (ESIPT) process. Density functional theory (DFT) was used to investigate the stability of the modified structure along with its interlayer separation. The estimated highest occupied molecular orbital-lowest unoccupied molecular orbital (HOMO-LUMO) gaps were compared to the experimental data.

The x-ray diffraction (XRD) pattern of the synthesized AAB-GO composite was shown. For the AAB-GO composite, the main peak appeared at the 2θ value of 9.5°, with the unusual peak at 12.3° (interlayer separation ∼0.72 nm) corresponding to GO. The major peak at 2θ value of 9.5 indicated the intercalated structure with an interlayer separation of ∼0.93 nm. Another peak appeared at 2θ value of 23.2° corresponded to the multilayer graphene with an interlayer separation of ∼0.34 nm.

The Raman spectra of graphene and GO exhibited two main bands. In the Raman of GO, the D band appeared at 1356 cm⁻¹ and the G band appeared at 1602 cm⁻¹. However, for the AAB-GO composite, the G band red-shifted to 1588 cm⁻¹, while the D band shifted upward to 1363 cm⁻¹.

The functionalization of GO with aminoazobenzene was confirmed by Fourier transform infrared (FTIR) spectroscopy. The unusual peaks at 3432, 1702, 1628, 1400 and 1067 cm⁻¹ corresponded to hydrogen-bonded O-H stretching, carbonyl C=O stretching, C=C stretching of the epoxides (C-O-C). In the AAB-GO composite, the broad peak in the region centered at 3430 cm⁻¹ was due to the presence of both the –OH and –NH group. The benzenoid C=C vibrations were observed at 1599 and 1505 cm⁻¹. The peak at 1460 cm⁻¹ showed the presence of the N=N group. The peaks between 1277 and 1219 cm⁻¹ were due to C–N stretching vibrations. In addition, the band centered at 1055 cm⁻¹ represented the C–O group.

To investigate the binding energies of different functional groups in AAB-GO composite, the authors have performed x-ray photoelectron spectroscopy (XPS) measurement. For AAB-GO composite, the characteristic C_1s, N_1s and O_1s core-level photoemission peaks at ca. 285, 400 and 432 eV, respectively, were observed. For GO, the low-range XPS spectra showed the presence of only C_1s and O_1s core-level photoemission peaks.

The AAB-GO composite was further subjected to optical measurements to understand the mode of functionalization in the composite. The UV–vis absorption spectra of pristine GO and AAB-GO composite were measured. GO dispersions showed a maximum absorption at 230 nm and a shoulder between ∼290–300 nm, which were assigned to the π–π* transition of aromatic C=C bonds and n–π* transitions of C=O bonds, respectively. The UV–vis absorption spectra of the AAB-GO composite comprised two peaks at 260 and 433 nm, which corresponded to the π–π* and n–π* transitions of the composite, respectively. This observation of shifting the π–π* transition and the appearance of a new peak compared to the GO indicated the successful functionalization of GO.

Photoluminescence (PL) spectra of GO and AAB-GO composite were measured. The PL spectra of GO solution in the visible range displayed a broad, weak emission band with maximum at ∼565 nm, while in the case of AAB-GO composite the PL maximum is blue-shifted to 563 nm, and the fluorescence intensity increased remarkably by 12 times compared to that of GO. The photoluminescence excitation (PLE) spectra of AAO-GO composite were also investigated. With the increase of the excitation wavelength from 400 to 460 nm, a very small red-shift of the PL peak was observed, but there took place a dramatic change of the PL intensity: when the excitation wavelength increased from 400 to 430 nm, the PL intensity increased, but from 430 to 460 nm the PL intensity decreased.

With the purpose to study for controlling the transfer of energy and charge between graphene and semiconductor quantum dot (QD), Rogez et al [47] investigated the fluorescence and blinking of individual laser-excited CdSeTe/ZnS core/shell QDs on single-layer graphene prepared by chemical vapor deposition (CVD). Fluorescence spectra of QDs in solution were recorded using time-resolved fluorescence spectrometer. The sample was excited with a coaxial xenon arc lamp at 500 ± 2 nm. The spectrum was recorded from 650 to 900 nm with a bandwidth of ±2 nm. On graphene and on glass, the fluorescence spectra of QDs were obtained using an inverted optical microscope in an epifluorescence configuration and laser excitation at 632.8 nm. The emitted light was coupled by an optical fiber to a diffraction grating spectrometer, which was equipped with a liquid nitrogen cooled charge coupled device (CCD). Emission of as-deposited graphene films was similarly recorded. Fluorescence lifetimes of individual QDs were measured using a wide-field, time-resolved fluorescence microscope. The set-up was based on an inverted optical microscope in the epifluorescence configuration. The sample was excited using a pulsed supercontinuum laser source spectrally filtered with a 525 ± 23 nm band-pass filter. Time-gated detection was performed thanks to a high trigger rate gated intensifier optically relayed to a CCD camera. The resulting image series acquired for decreasing decay times was then fitted with an exponential model to obtain the fluorescence lifetime image (FLIM). Fluorescence imaging and blinking measurements were conducted using an inverted optical microscope in an epifluorescence configuration (similar to the emission measurements of QDs on graphene and on glass). The emitted light was focused on a liquid-cooled CCD camera. Blinking statistics of individual QDs were retrieved from the analysis of a series of CCD images.

The fluorescence decay of QDs on glass as well as on graphene was investigated by measuring the fluorescence lifetimes. The collected intensity from QDs on graphene was found to be ten times lower than that from QDs on glass. For QDs on glass the experimental data were fitted with a biexponential decay with approximately equal amplitudes and average lifetimes of ${\tau }_{1}=1.2\pm 0.6ns,$ ${\tau }_{2}=15\pm 7ns,$ while for QDs on graphene ${\tau }_{1}\ll 1ns,$ ${\tau }_{2}=1.7\pm 0.8ns.$ The average values were calculated from those obtained in repeated measurements for over 100(56) individual and optically isolated QDs on glass (graphene). Concerning the blinking, for the first time the authors have observed the characteristic fluorescence intermittency of QDs on graphene: the intensity oscillated between bright and dark states. Moreover, QD fluorescence intermittency on glass was characterized by comparatively short time periods in the bright state, whereas for QDs on graphene the bright state periods were significantly longer.

As a special original method for fabricating graphene quantum dots (GQDs) Suh, Kim et al [48] proposed to use a thermal plasma jet. According to this method, a carbon atomic beam was generated by continuously injecting a very small amount of ethanol as a carbon source into Ar plasma; the beam was then flowed through a carbon tube attached to the anode and then collided with the graphite plate that was placed on the path of the beam, perpendicular to the attached carbon tube. In the subsequent work [49] Suh and Kim have improved this thermal plasma jet method to fabricate size-controllable GQDs with a low cost. A carbon atomic beam was generated by injecting a large amount of ethylene gas continuously into Ar plasma, flowed through a carbon tube attached to the anode and then dispersed into a chamber. Carbon soot together with GQDs were prepared by a gas phase reaction. Almost all of carbon soots were dispersed in ethanol by sonication, while GQDs were dispersed in ethanol by stirring with a stirring rod. The average size of GQDs, with a relatively narrow size distribution, was controlled by varying the length of the attached carbon tube. It was about 10, 14 and 19 nm when the length was 5, 10 and 20 cm. The absolute quantum yields of these GQDs were 13.5%, 12.2% and 9.6%.

High-resolution transmission electron microscopy (HRTEM) images of GQDs showed their high crystalinity with lattice parameter 0.32 nm and (002) lattices fringes of graphene. The corresponding fast Fourier transform (FFT) pattern of hexagonal symmetry without satellite spots showed that GQDs were single-layered. This conclusion was supported by atomic force microscopy (AFM) analysis. The thickness of GQDs was less than 1nm – in good agreement with the reported value of single-layer graphene. The weight percent of GQDs in the carbon soot was about 10%.

UV–vis absorption, photoluminescence (PL) and photoluminescence excitation (PLE) spectra of prepared GQDs were measured. The PL peaks were observed near 375, 393, 406, 430, 460, 490 and 506 nm. The PLE spectra were measured by detecting at 393, 406, 432, 460 and 506 nm. For the PLE spectrum with the detection wavelength of 460 nm, for example, there were three strong absorption peaks at 426.4, 402.7 and 306.1 nm and three weak ones at 380.7, 343.2 and 361.6 nm etc. From the analysis of the above spectra the authors concluded that the electronic structure of GQDs consists of seven levels. The Raman spectra of prepared GQDs were also measured. It was observed that the G band near 1596 cm⁻¹ was stronger than the D band near 1353 cm⁻¹. The D band was known due to the presence of structural disorder in the graphene sheets. A higher G/D intensity ratio could indicate better crystalinity of GQDs. Compared with the results of other works the intensity ratio of GQDs prepared in the present work was relatively high, reaching the value 1.6.

To investigate photonic materials emitting electromagnetic radiations with broad spectral wavelengths covering deep-ultraviolet (DUV), visible (Vis) and near-infrared (NIR) regions, Lan et al [50] prepared nitrogen-doped graphene quantum dots (N-GQDs) with the layered structure possessing broad-band emission range 300–1000 nm. The broad-band emission was attributed to the layered structure of N-GQDs containing a large conjugated system and providing extensive delocalized π electrons. In addition, a broad-band photodetector with responsivity as high as 325 V W⁻¹ was demonstrated by coating N-GQDs onto interdigitated gold electrodes. To achieve the desired purpose, the authors proposed a facile 'one-pot' microwave-assisted hydrothermal (MAH) technique using glucose and aqueous ammonia as the source.

The analysis of transmission electron microscopy (TEM) images of N-GQDs showed that their sizes increased with increasing heating time. The crystalline structures of N-GQDs with the sizes of 1.7, 1.9, 3.0, 4.0 and 5.8 and 9 min, were shown. Atomic force microscopy (AFM) characterizations were performed to investigate the morphology and the height of N-GQDs. In the high-resolution transmission electron microscopy (HRTEM) images of N-GQDs prepared at 5 min heating, the lattice fringes of N-GQDs were clearly observed. The fast Fourier transform (FFT) indicated the hexagonal crystalline structure of N-GQDs. The N-GQDs also revealed the layered structure with an interlayer spacing of ∼0.38 nm, slightly larger than that of bulk graphite (0.355 nm) due to the presence of the functional groups enlarging the basal plane spacing of N-GQDs.

X-ray photoelectron spectroscopy (XPS) was performed to investigate the chemical bonding of N-GQDs. In the N_1s XPS spectra, three types of N-related bonding were identified: pyridine N (∼399.2 eV), pyrrolic N (∼400.2 eV) and graphitic N (∼401.6 eV). The C_1s XPS spectrum can be decomposed into five peaks centered at around 284.5 eV (C=C), 285.8 eV (C-C, C-H, C=N), 286.6 eV (C-OH), 287.2 eV (C-O-C, C-N) and 288.6 eV (C=O), revealing different types of bonding to C. The atomic N/C ratio of N-GQDs was determined to be 8.3/100, indicating that C is the dominant element. This N/C atomic ratio is much higher than that of various N-doped graphene-based materials (0.3–5.6%). The prepared N-GQDs were further characterized by x-ray diffraction (XRD), Fourier transform infrared (FTIR) spectroscopy and Raman scattering.

The UV–vis-NIR absorption spectra of N-GQDs were measured. There were three UV absorption peaks located at 214.5, 268.0 and 310.5 nm related to the C=C, C=N and C=O electron transitions from π (or n) to π*. The visible absorption spectrum between 400 and 700 nm of N-GQDs related to the partial conjugated π electrons in their layered structure were recorded. Importantly, a broad NIR absorption band peaking at ∼812 nm became noticeable as the size of N-GQDs reached 3.0 nm. It could be attributed to the larger conjugated system containing extensive delocalized π electrons in the layered N-GQDs.

The photoluminescence (PL) quantum yields of N-GQDs with various diameters were determined to be between 6.8 and 11.3%. The photoluminescence excitation (PLE) spectra of N-GQDs excited by light with wavelengths of 197, 475 and 808 nm exhibited broadband PL spectra that peaked at 302, 542 and 915 nm.

Time-resolved PL decay measurements were performed on the N-GQDs. A triple-exponential equation fitted the experimental data well. The PL lifetime contained a fast component with lifetime τ₁ changing from 0.01 to 13.8 ns and two slow components with lifetimes τ₂ changing from 2.18 to 4.13 ns and τ₃ > 10 ns.

The obvious advantages of graphene-based Schottky junction solar cells compared with solar cells using an indium tin oxide (ITO) electrode have encouraged the fabrication of graphene-based Schottky junction solar cells on different traditional semiconductor substrates such as Si, CdS, CdSe [51–54] with power conversion efficiency (PCE) ranging from 0.1 up to 2.86%. Recently Tongay, Hebard et al [55, 56] performed a significant improvement of single layer graphene/n-Si Schottky junction solar cells by chemical charge transfer doping of graphene with bis (trifluoromethanesulfonyl)-amide [((CF₃SO₂)₂NH)] (TFSA) and achieved a PCE of 8.6% on this device.

Graphene sheets were grown on copper (Cu) foils by chemical vapor deposition (CVD). Poly (methyl methacrylate) (PMMA) was spin-cast on graphene/Cu, then Cu layer was removed from the PMMA/graphene/Cu foils, yielding PMMA/graphene. Prior to transferring graphene to a new substrate, Au/Cr windows were deposited onto Si(111) wafers with a thick SiO₂ surface layer. After the Au/Cr deposition, exposed parts (3 × 3 mm² area) of SiO₂ were removed to expose the underlying Si. Graphene sheets were transferred onto Si, and PMMA backing layer was dissolved away in an acetone bath. Doping of the graphene with TFSA was accomplished by spin-casting TFSA. Ohmic contacts to Si wafers were made by gallium indium eutectic paint (99.99% metal basis), and J-V and C-V measurements were taken between graphene (metal electrode) and ohmic contact on Si (semiconductor).

To measure the external quantum efficiency (EQE), the devices were illuminated by monochromatic light and the photocurrent was recorded by a lock-in amplifier together with a current amplifier. A Xe-arc lamp was used as the white light source and the monochromator was adopted to generate monochromatic light.

The work function difference between graphene and n-Si resulted in electron transfer from Si to graphene yielding a Schottky junction with its associated depletion layer in Si and the built-in potential ${{\rm{V}}}_{bi}$ across it. Photons absorbed in Si generated electron-hole pairs, and the charges were collected at graphene and semiconductor contacts, thereby generating power from the device. Under illumination the short-circuit current (J_sc) was 14.2 mA cm⁻² with open-circuit voltage (V_oc) and power conversion efficiency (PCE) corresponding to 0.42 V and 1.9%, respectively. The J-V characteristics after doping graphene sheets with TFSA were recorded. Due to the holes doped from TFSA into graphene (p-doping) the resistance of the graphene sheet reduced while its work function increased without changing its optical properties. For the graphene/n-Si solar cells upon doping with TFSA, the J_sc, V_oc and fill factor (FF) all increased from 14.2 to 25.3 mA cm⁻², 0.43 to 0.54 V and 0.32 to 0.63, respectively. These increases in J_sc and V_oc boosted the PCE from 1.9% to 8.6%, which was the highest PCE reported for graphene-based solar cells to date.

The observed enhancement in PCE was attributed to the increase of the Schottky barrier height (SBH) and hence the built-in potential V_bi, and the reduction of resistive losses associated with the increase of the electrical conductivity of doped graphene sheets.

With the intention to develop a new class of graphene-based solar cells combining the advantages of graphene quantum dot (GQD) and oxide semiconductor, Basak et al [57] demonstrated a ZnO/GQD solid state solar cells, in which ZnO material was prepared in the form of nanowires (NWs). Vertical arrays of ZnO NWs grown on Al-doped ZnO (AZO) thin films were infiltrated and covered with the synthesized GQDs. This was done by repeated spin-casting of an ethalonic suspension of GQDs on the NWs arrays until the space between them was filled up completely and a thin layer of GQDs was formed on the top of NWs. Then the authors deposited a 60-70nm thin layer of N-N'-diphenyl-N-N'-bis (3-methylphenyl)-1, 1'-biphenyl)-4, 4' diamine (TPD), which acted as a hole-transporting layer, by spin-casting its solution in chloroform. After that, the device was annealed in an inert atmosphere at 110 °C for 30 min. Then the top Au electrode was sputtered on the TPD.

A high-resolution transmission electron microscope (HRTEM) attached to the energy dispersive x-ray analysis (EDAX) facility, a field-emission scanning electron microscopy (FESEM) equipment and an atomic force microscope (AFM) were used for the microstructural characterizations. The x-ray powder diffraction (XRD) pattern was investigated. The UV–vis transmission spectra were recorded in the range 300–800 nm, Fourier transform infrared (FTIR) spectra were collected and Raman scattering experiments were conducted. The photoluminescence excitation (PLE) spectra were measured by a fluorescence spectrophotometer and also by employing a He-Cd laser with a 325 nm excitation source and a high-resolution spectrometer together with a photomultiplier tube. The current-voltage (J-V) characteristics of the photovoltaic cell and incident photon-to-current conversion efficiency (IPCE) were measured. The internal quantum efficiency (IQE) was estimated by dividing IPCE by the fraction of absorbed light in the active film.

The results obtained by the authors showed that the environmentally friendly GQDs can be potentially harnessed as a replacement of toxic semiconductor QD sensitizers along with an advantage of hole transport.

With the purpose to combine the advantages of both crystalline silicon (c-Si) and graphene quantum dots (GQDs) Jie, Sun et al [58] developed a new type of solar cells based on the c-Si//GQDs heterojunction. Thanks to the unique band structure of GQDs, photogenerated electron-hole pairs could be effectively separated at the junction interface. The GQDs also served as an electron blocking layer to prevent the carrier recombination. With the size-tunable band gap, c-Si/GQDs heterojunction solar cells exhibited significant performance enhancement as compared to the device counterparts without GQDs or with graphene oxide sheets.

GQDs were synthesized by a hydrothermal method for cutting graphene sheets into blue-luminescent GQDs [59]. To construct the c-Si/GQDs solar cells, the colloidal GQDs were dropped onto the silicon wafer surface and baked in air. Thickness of the GQDs film was controlled by adjusting the amount of colloidal GQDs dropped onto the substrate. Afterward, semitransparent gold top electrodes were deposited on the GQDs films using a shadow mask via e-beam evaporation. Indium-gallium (In-Ga) alloy was pasted onto the rear side of the Si substrate to form ohmic contact. The size of GQDs was controlled by varying the ultrasonic time during synthesis. Transmission electron microscopy (TEM) images of GQDs showed relatively narrow size distributions in the range of 2–6, 5–10 and 8–16 nm, respectively. From the cross-sectional view scanning electron microscopy (SEM) image of the device it was seen that the GQDs layer was smooth with relatively uniform thickness. More importantly, the GQDs layer was compact without any visible cracks, avoiding the formation of short circuit channels between the top electrode and c-Si wafer. The electrical conductivity of the GQDs layer was further measured by depositing two Au electrodes on it.

To verify the essential roles of GQDs in the heterojunction solar cells, control experiments were conducted by removing the GQDs layer from the device, or replacing the GQDs layer by a GO layer. It was shown that the characteristics of the device with GQDs layer were significantly enhanced in comparison with its counterparts without GQDs or with GO sheets. From the J-V characteristics of the device it was seen that the open-circuit voltage V_oc increased and the short-circuit current J_sc decreased with the decrease of GQDs size. The device exhibited excellent stability. It can work even after storage for half a year, the J_sc decreased from 28.71 to 22.35 mA cm⁻², while the V_oc and fill factor (FF) were nearly unchanged. By tuning the GQDs size and layer thickness, the authors have achieved the optimum power conversion efficiency (PEC) of 6.63%. The underlying physical mechanisms of the enhancement of the device performance was also studied in detail.

Understanding efficient methods to improve the performance of solar cells, in reference [60] Xie, Zhang et al demonstrated the successful construction of high-efficiency graphene-planar Si solar cells by Si surface passivation as well as interface band engineering. A thin layer of organic film was inserted into the graphene-Si interface as an electron-blocking layer, preventing charge recombination in the graphene anode. Meanwhile, methylation on Si was conducted to suppress the surface recombination and tune the band alignment rear the Si surface as well.

N-type Si substrate with a thick SiO₂ insulating layer was protected by an adhesive tape with an opening window. Then the SiO₂ insulating layer within the window was removed, and by immersing the SiO₂-Si substrate into an aqueous HF solution the authors obtained hydrogen-terminated Si denoted as H-Si. By using PCl₅ the H-Si bond was transformed to the Cl-Si bond, and Cl-terminated Si was then transformed to the CH₃-terminated Si.

Large-area monolayer graphene (MLG) films were prepared via a chemical vapor deposition (CVD) method, and a thick Cu foil was used as the catalytic substrate during the synthesis. After growth, polymethyl-methacrylate (PMMA) in clorobenzene was spin-coated on MLG films, and then the underlying Cu foils were removed. The resulted MLG films were cleaned and then characterized by Raman spectroscopy. The sheet resistances and transmittances of graphene films were measured by a digital four-probe tester and a spectrometer equipped with an integrating sphere, respectively.

To construct the graphene-planar Si Shottky solar cells, a Ti/Au electrode, which served as the electrical contact for graphene, was first deposited on the SiO₂-Si substrate rear the exposed Si window by using a shadow mask via electron-beam evaporation. Then the PMMA-supported MLG films were directly transferred onto the top of the substrate. Graphene then would be in contact with the exposed Si, forming the Schottky junction. The residual PMMA on the MLG films was removed by acetone. By repeating this process, few-layer graphene (FLG) films consisting of one to six MLG layers could be fabricated. An indium-gallium (In-Ga) alloy was then pasted on the rear side of the Si substrate as the Ohmic contact for Si. For preparing the devices with an electron-blocking layer, the organic layer poly(3-hexylthiophene-2, 5 diyl) (P3HT) was deposited on Si within the window area by spin-coating, the substrate was baked in a nitrogen atmosphere, and the P3HT layer outside the window was then removed by an ethanol-dipped cotton swab. Finally, the graphene-P3HT-planar Si were obtained by transferring a FLG film onto the top of the P3HT layer.

The presented method along with the careful control of the graphene doping level and layer numbers gave rise to the power conversion efficiency (PCE) as high as 10.56%.

Interested in the utilization of Si hole array (SiHA) in the fabrication of solar cells, in reference [61] Jie, Zhang et al demonstrated the construction of high-efficiency micro-hole graphene/SiHA Schottky junction solar cells with enhanced device performance and stability. The micro-hole SiHA was fabricated by combining conventional photolithography and reactive ion etching (RIE) techniques. At first, UV photolithography was conducted on Si wafer by using positive photoresist in a mask aligner. After UV exposure and photoresist development, holes with a diameter of 6μm and a period of 8μm were generated on the photoresist film. Afterward, the Si substrate was loaded into a RIE system and the Si etching was performed. The hole depth was adjusted by controlling the etching duration, and the etching duration of 3, 5, 8 and 10 min corresponded to the hole depth of 3.8, 6.4, 10.2 and 12.8 μm, respectively. Significantly, the as-prepared SiHA showed a smooth hole surface in contrast to the rough sidewalls of the SiHA fabricated by the electrochemical etching method. This feature was important for reducing the surface charge recombination and consequently improving the device efficiency. The construction of graphene/SiHA solar cells on n-type Si substrate with a thick $Si{O}_{2}$ insulating layer was performed by the same method as that reported in the preceding reference [60]. The layer number of graphene films was optimized to be four layers. The films were doped by directly spin-coasting AuCl₃ solution (in nitromethane) on the top surface. The Au ions could be reduced to Au atoms by accepting electrons from graphene films. During the reduction reaction the hole concentration in graphene films increased (p-doping), leading to reduced sheet resistance as well as enhanced work function of graphene. The four-layer graphene films after doping were characterized by Raman spectroscopy, and their transmittances were measured by a spectrometer equipped with an integrating sphere. The photovoltaic characteristics of the solar cells were evaluated by using a source meter in an ambient environment.

Efficient light-harvesting is essential to a high-performance photovoltaic device. In order to assess the impact of light absorption capability of the SiHA on device performance, the micro-hole SiHA with various hole depths of 3.8, 6.4, 10.2 and 12.8 μm were investigated. For simplifying the analysis the authors tuned only the hole depth, while keeping the hole diameter and density unchanged, so that the effective Schottky junction areas would be the same for all the devices with different hole depth. Therefore the change of device performance was solely attributed to the variation of SiHA high absorption. The authors demonstrated that the device performance enhanced with increasing hole depth: The J_sc increased dramatically from 20.19 mA cm⁻² for the device with a 3.8 μm thick SiHA to 31.56 mA cm⁻² for that with 12.8 μm thick SiHA. The external quantum efficiency (EQE) value also increased with increasing hole depth, from 61%, 67%, 73% to 80% at 680 nm for 3.8, 6.4, 10.2 and 12.8 μm thick SiHA, respectively. The promotion of J_oc and EQE was ascribed to the enhanced optical absorption of the SiHA depth hole. As the result, the power conversion efficiency (PCE) of the device achieved the optimum value of 10.40% when the hole depth was 12.8 μm.

Graphene not only played the role of active material in Schottky junction solar cells. It can be utilized also in other parts of different solar cells. Pursuing this tendency Dai et al [62] demonstrated tandem solar cells consisting of two or more subcells connected by charge recombination interconnecting layers. In a tandem solar cells of two subcells with stacking complementary absorption profiles, the open-circuit voltage V_oc equals the sum of those of subcells while keeping the short-circuit current J_sc the same as the lower one, leading to an increased overall power conversion efficiency (PCE). Functioning as both an internal anode and a cathode to facilitate the efficient electron-hole recombination for maximizing the ${{\rm{V}}}_{oc}$ and fill factor (FF), the interconnecting layer plays an important role in regulating the device performance. Generally speaking, an ideal interconnecting layer needs to possess an energy level matching with those of donor and acceptor (macro) molecules in the active layer, sufficient conductivity, high transparency, uniform coverage and good chemical stability.

In a previous work [63] the authors showed that simple charge neutralization of the-COOH group in graphene oxide (GO) with Cs₂CO₃ could tune the electronic structure of GO, and the resultant caesium-neutralized GO(GO-Cs) can act as an efficient electron extraction layer in PSCs. By replacing the periphery-COOH groups of GO with-COOCs groups via the charge neutralization, the work function of a GO-Cs modified Al substrate can be reduced to 4.0 eV, matching well with the lowest unoccupied molecular orbital (LUMO) level of [6, 6]-phenyl-C61-butyric acid methyl ester (PCBM) for an efficient electron-extraction. Moreover, the GO-Cs can be well dissolved into ethanol, making the multilayer solution-processing feasible.

In the present work the authors developed a GO-based carbon interconnecting layer consisting of GO-Cs/GO bilayer modified with ultrathin Al and MoO₃. The relatively weak light-absorption characteristic of GO and GO-Cs together with their good solution-processability for ultrathin film formation facilitated the light transmission through the interconnecting layer to the rear cell. By careful design of the energy level alignment within the GO/GO-Cs interconnecting layer, efficient charge carrier collection from the subcells and charge recombination within the interconnection bilayer was achieved. As a result, the tandem cells fabricated with the GO-Cs/GO-based interconnecting layer exhibited a significantly increased V_oc reaching ∼100% of the sum of the V_oc subcell, suggesting a successful serial connection of subcells.

Flexible graphene electrode is another efficient application of graphene in organic photovoltaics. In reference [64] Gradečak, Palacios, Kong et al demonstrated anode-and cathode-based polymer solar cells (PSCs) with record-high power conversion efficiencies (PCEs) of 6.1 and 7.1%, respectively. The high efficiencies were achieved via thermal treatment of MnO₃ electron blocking layer and direct deposition of ZnO electron transporting layer on graphene. The authors used photoactive media composed of a blend of low bandgap semiconducting polymer donor thieno[3, 4-b]thiophene/benzodithisphene (PTB7) and acceptor [6, 6]-phenyl C₇₁-batyric acid methylester (PC₇₁BM) prepared using mixed solvents of chlorobenzen: 1, 8–diiodoctane (CB:DIO). The hole injection layer poly (3, 4-ethylene dioxy thiophenen): poly-(styrenesulfonate) (PEDOT:PSS) was deposited on the transparent graphene electrode. To ensure uniform coverage over the graphene surface, the authors used modified PEDOT:PSS with isopropyl alcohol (IPA) at 3:1 (v/v) ratio. Prior to the active layer deposition, graphene/PEDOT:PSS must be covered by an additional electron blocking layer MoO₃.

For fabricating graphene-based PSCs, the graphene electrode was prepared by stacking three monolayers of graphene film. With incorporation of appropriate PEDOT:PSS and thermally treated MnO₃, the authors observed record-high efficiency from graphene (PCE = 6.1%) approaching that of the ITO reference device (PCE = 6.7%). Both graphene anode-based and inverted cathode-based PSC configurations were investigated.

The effect of solvent treatment on MnO₃ was characterized by scanning transmission electron microscopy (STEM). Significant difference was observed between the as-deposited MnO₃ film and that after CB:DIO treatment. Ultraviolet photoelectron spectroscopy (UPS) was performed to investigate the electronic structure of MnO₃ film after the thermal and solvent treatment. The surface morphology of MnO₃ was also characterized using atomic force microscopy (AFM) after the thermal and solvent treatment.

As the final step the authors have explored the potential of the device structures to realize flexible graphene-based PSCs and prepared both anode-and cathode-based device architectures on polyethylene naphthalate (PEN) substrates. The resulting graphene PSCs on PEN substrates showed excellent device performance for both anode (PCE = 6.1%) and cathode (PCE = 7.1%) configurations. They were robust under mechanical deformations, which is highly desirable for low-cost productions such as roll-to-roll processing and applications that require flexibility.

In this review we have presented the results of the development of graphene-based optoelectronics, plasmonics and photonics, with the emphasis on the recent advances. The successful fabrication of graphene-based photodetectors, modulators, plasmonic nanostructures enhancing or tuning the graphene-light interaction, and graphene quantum dots was reported. Photoluminescence and fluorescence of graphene nanostructures were investigated. The fabrication of graphene-based Schottky junctions solar cells, graphene flexible electrodes in polymer solar cells and interconnecting graphene in tandem solar cells were also presented.

Above presented results of the research on graphene-based optoelectronics, plasmonics and photonics showed significant perspectives of the utilization of graphene in high technologies.

The authors would like to express their deep gratitude to the Vietnam Academy of Science and Technology for the support.