Open Access
Add to BibSonomy
Abstract
Graphene is widely recognized as an outstanding and multi-functional material in various application fields such as electronics, photonics, mechanics, and life sciences. We propose a neurotransmitter sensor with ultra-small volume for detecting the photonic light-matter response. Such detection can be achieved using surface-activated monolayer graphene sheets and CMOS-compatible silicon-photonic circuits. Patterned pieces of CVD-grown graphene are integrated on the top of a silicon micro-ring resonator, which induce the adsorption of catecholamine molecules originated from the π-stacking effect. We used dopamine to demonstrate such detection and examine the sensitivity of graphene-dopamine coupling. To avoid high optical insertion loss and degradation of resonance characteristics caused by a graphene’s extremely high optical absorption coefficient in the near infrared region, a ring resonator with adjusted coupling design is used to compensate for the drawbacks. Owing to the advanced nano-sensing platform and measurement system, an activated graphene-sensing surface of only ∼30 µm2/ch enables π coupling to dopamine with enough sensitivity to detect less than 10-µM solution concentration. The detection mechanism through the surface reaction is also verified by optical simulation and atomic force microscopy measurement, revealing that the flowing dopamine molecules can only occupy the outermost surface of graphene. We expect this sensor to contribute to the development of an innovative label-free and disposable bio-sensing platform with accurate, sensitive, and fast response.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
The function and organization of neurons and neurotransmitters in the human brain have been elusive research topics in the life sciences [1–3]. Although invasive surgery has been done to understand what is going on in our brain, it often causes undesirable physical and psychological suffering. To avoid this, a simple tool involving indirect probing methods, such as through the use of sweat and saliva, is necessary. A neurotransmitter-detection tool will also eventually contribute to diagnosing and monitoring various diseases, not limited to the nervous system, and lead to the discovery of new drugs. The most important monoamine neurotransmitters [4] are dopamine, adrenaline, noradrenaline, histamine, serotonin, and melatonin. The first three monoamines are known as catecholamines. A benzene ring with a side chain of an amino group is their common structure, which works as a transporter across synapses. From the viewpoint of chemistry, aromatic compounds provide delocalized π electrons along the two-dimensional plane surface. This means that a coupling interaction can be observed with them if other substances are formed as acceptors underneath that hold similar π electrons to form a conjugated system. Taking dopamine, a well-known neurotransmitter discharged from synapse vesicles, as an example, abnormalities in its concentration in the human body are associated with schizophrenia, attention deficit hyperactivity disorder, and restless leg syndrome as well as with intractable diseases such as Parkinson’s and HIV infection [5,6]. While these concrete examples serve as a brief introduction to the importance of neurotransmitters detection, a notable fact is that cost-effective and mature analysis methods have not yet been developed. Over the past several decades, various transducers have been used to identify samples by means of electrochemistry [5], surface plasmon resonance (SPR) [7], and fluorescence [8], all of which perform well regarding dopamine detection in the nM to µM detection range. However, the smart and compact systems of an electrochemical sensor require parallel configuration and the monitoring of a small amount of current change. Tracing all individual channels will restrict the integration scalability of future sensor systems. Unlike optical sensors, electrical measurement may be affected by external environmental fluctuations, such as electromagnetic radiation, which will interfere with the signal. The SPR sensor has recently recorded the highest detection sensitivity; as low as sub-pM order (cf. linearity was limited to nM) [9]. This sensitivity can be achieved with a large localization of electromagnetic energy in the layer immediately adjacent to the metal surface. Although remarkable performance has been achieved, it is difficult to achieve scalable-mass production with an optical-fiber-based SPR system along with guaranteeing low cost or disposability owing to expensive Au or Ag coating.
We propose a real-time and label-free dopamine sensor by applying a nano-photonic probing approach. Its primary advantages are scalability enabled by the potential of Si-CMOS integration, simplicity owing to the serial formation of a large number of sensor devices (i.e. operating with only one light source and spectrometer), and reliability provided by mature fabrication processes. For label-free dopamine detection, a monolayer-graphene-integration process onto a silicon-ring photonic platform is used, which has recently been developed for an ultra-compact optical intensity modulator and nonlinear supercontinuum generation light source [10,11]. The hexagonal rings of a graphene sheet are beneficial to observe catecholamine molecule adsorption by π stacking, which can potentially increase the absorption amount and sensitivity with additional treatments [12], but similar devices have not been proposed on a silicon photonic platform. We expect this device, as the first proof of concept for a graphene-integrated sensor on a silicon ring resonator, will play an important role in future advances in biological sciences.
2. Optical circuit design for sensitive detection
Figure 1(a) illustrates the sensing configuration for observing neurotransmitter detection. The fundamental photonic chip is composed of a high-Q silicon ring resonator with partially integrated CVD-grown monolayer graphene sheets on both sides. The probing laser source is coupled into the sensing device by a focused light through a polarization-maintained lensed fiber (NA: 0.33, MFD: 3.3 ± 0.6 µm) and propagates to the reaction region. A ring-resonant peak on the frequency axis can be modulated by the outer environment surrounding the silicon waveguide core on the ring because the refractive index change has a direct response to wavelength shift Δλ, which can be expressed as $\varDelta$λ = λc·Γ·Δneff / neff, where λc is the center wavelength in the resonance peak, Γ is the confinement factor of the optical waveguide, neff and Δneff are the effective refractive index and changing value affected by the outer environment [13,14], respectively. Figures 1(b) and 1(c) correspond to the ring resonator sample in optical-microscope and SEM images, respectively. Sections of graphene (lengths (LGr) of 0, 5, and 10 µm; width of 3 µm) are loaded on both left and right sides of the circumference. The detailed fabrication steps of silicon photonic devices are based on our previous studies [15,16]. Contrary to common methods of forming a graphene layer on the upper planarized surface [17–19], graphene in this case is directly defined on an as-fabricated silicon photonic wafer with 250-nm height steps. This transfer technique is used to gain more overlapping as interactively as possible between the confined optical mode field distribution and stacked analytes on the top surface as well as sidewalls of the silicon waveguide. In µ-Raman spectroscopy characterization by two-dimensional mapping, we found neither significant disorder nor staked multi-layer graphene by observing the D, G, and 2D (G’) bands [20,21]. For an efficient optical coupling with high-NA lensed fibers, we formed 1-mm long (500 µm for inversed taper) selective spot-size converters (SSCs) with a taper edge width of 200 nm on both sides (see green stripes in the optical microscope image in Fig. 1(b)). The SSCs contribute to improving the coupling efficiency as well as suppressing the ripples on the spectral response due to internal light reflection. In a TE-like mode, the coupling loss is as low as 1.4 dB/facet [16]. Next, we prepared a microfluidic device made of polydimethylsiloxane (PDMS) to construct a basic micro-total analysis system (µ-TAS). A fluidic channel with a width of 200 µm and a height of 50 µm was designed for guiding analyte solution. The alignment between the silicon photonic chip and microfluidic device was implemented with as-etched silicon pedestals under an optical microscope. In addition to those alignment marks, a deposited SiO2 film for the SSCs also coarsely supported the alignment process for the rotation angle.
Fig. 1. (a) Sensing configuration for neurotransmitter detection. Sample consists of high-Q silicon ring resonator and partially integrated mono-layer graphene sheets on it. Bottom right is expanded view of graphene-covered reaction field, where adsorption of graphene and dopamine is illustrated. (b) Optical microscope image of fabricated sensing sample. Total of 9 channels and reference waveguides are integrated. Green sidelines are over-cladding layers of SiO2 for efficient taper coupling, and rest is open (air cladding). (c) SEM image of silicon ring resonator. Sections of graphene (length 5 µm; width 3 µm in image) are loaded on both left and right sides of circumference.
Intrinsic graphene always induces huge optical absorption due to the interband transition [22–25], which causes sharp Q-factor degradation in nano- and micro-cavities such as silicon photonic crystals and ring resonators [15,26]. In our previous study, we found that the initial Q factor of nearly 8000 in a silicon ring resonator sharply reduces to 1000 with 20-µm long graphene loading. This structure also absorbs the propagation light larger than 10 dB at the peak intensity. These phenomena can be explained by the imbalance between the internal resonant section (internal Q) and the external power-supplying section (external Q). In other words, the intensity coupled out via through and drop ports can be purposefully tuned by designing the directional coupling coefficient (Fig. 2(a)). The adjustment (called “adjusted coupling”) can be done in two ways, either by changing the gap between the two waveguides or the coupling length with parallel configuration. We chose the latter because controlling the absolute coupling length is highly accurate in defining the geometry. The light transmittance from the drop port is derived from the following Eq. (1) via the transfer function of a ring resonator [14],
Fig. 2. (a) Schematic image of proposed sensing device. (b) Color-contour plots of transmitted peak intensity from drop port. Note that a and r are single-pass amplitude transmission of ring resonator and self-coupling coefficient of DC, respectively. Experimental results are plotted with designed parameters to compare conventional (red diamonds), and adjusted coupling designs (blue circles). White dashed line corresponds to “a = r”. (c) Color-contour plots of Q factor in same manner as (b). (d) Spectrum comparison of adjusted coupling and conventional ring resonators. Peak transmittance was improved by 13 dB (20 times). (e) Summary of adjusted coupling design LGr was set to 0, 5, and 10 µm, respectively.
3. Integrated dopamine sensor
Before flowing the analyte sample solution into the sensor, the PMMA bound to the surface during the graphene transfer process must be removed. This additional treatment step, by annealing at 400 °C in a mixture atmosphere of H2 and Ar for 30 minutes, significantly improves the activation on the graphene surface [28]. Such a high temperature (e.g. conventional temperature is around 250–300 °C) was used to activate over the graphene pattern because the opening area significantly affects the sensing sensitivity and adsorption observation. This effect was clarified through atomic force microscopy (AFM) measurement, as illustrated in Fig. 6, which revealed that the initial thickness of 4–5 nm reduced to about 0.5 nm, and approached an ideal thickness of 0.34 nm as pristine graphene. The dynamic-liquid-flowing measurement with µ-fluidic devices was carried out on a semi-auto fiber alignment stage. First, a PDMS-based µ-fluidic was gently pressed onto the silicon photonic chip along the alignment marks and SiO2 grooves on both sides. To suppress the thermal drift and noise, the temperature of the silicon sensor substrate was adjusted to 35.0 °C by using a thermal-electric controller. A silicon ring resonator is very sensitive to outer-air-temperature variation because bulk crystal silicon has a notably high thermo-optic coefficient σn/σT of over 1.8×10−4 K−1 at room temperature [29]. For this reason, all analytes, syringes and a part of a delivery tube attached with a micro-syringe pump were sunk into a dry thermos stat. A Y-branch liquid combiner was installed for seamless solution switching throughout the entire flow measurement. The liquid-delivery velocity was set to 2.0 µL/min for analyte sensing, and 20.0 µL/min for cleaning process of the inside of a delivery tube as well as µ-fluidics. The measurement configuration feeding into the sample is shown in Fig. 3(a). We estimated sensitivity in terms of wavelength displacement per unit change in the refractive index. This value can be determined by measuring the peak shift of the resonance spectrum, and referred by the relationship between solution concentration and refractive index with well-known substances such as diluted sodium chloride (NaCl) [30,31] or glucose [32] as a reliable reference. Figure 3(b) shows the slope sensitivity, which we found to be 120 nm/RIU for the NaCl solution. Although the graphene pieces were loaded on a part of the ring resonator, this sensitivity is high enough to detect unknown material that has a refractive index unit smaller than 10−5 [13]. We used the slope sensitivity for the computational simulation as well as determining the refractive index change with our target analyte, i.e. dopamine. Consequently, dopamine hydrochloride (DA) solution of 5, 16, 50, 160, 500 µM, and DI were delivered into the fluidics alternately at approximate 10-minute intervals. DA was selected instead of dopamine because it is soluble in water and chemically stable [33] (cf. HCl prevents dopamine from becoming an oxide). Note that the dominant framework and adsorption mechanism based on the π stacking were assumed to be equivalent.
Fig. 3. (a) Sensing measurement setup. µ-fluidics made of PDMS on silicon photonic circuit was mounted onto optical fiber alignment stage with temperature control. (b) Measured relationship of refractive index change and resonant peak shift with NaCl-solution calibration. Slope sensitivity of 120 nm/RIU was observed.
The detailed configuration on the experimental setup for real-time flow measurement is illustrated in Fig. 4. The measurement system is composed of a tunable laser diode (TLD) with a line width of around 100 kHz, and fast-scanning and data-logging combination unit (Agilent TLD: 8163B + 81600B; Agilent Power Meter: N7744A). A scan dataset ranging from wavelengths of 1540 to 1560 nm (cf. designed FSR of the ring is 6–7 nm) with a 0.1-pm resolution can be stored in the unit memory within 2–3 seconds with numerical processing of Lorentzian-function fitting and peak decision, data transfer, and deliberate interval time. Since a stable measurement environment is required, this experiment was conducted under synchronized temperature control for the silicon photonic chip with a built-in thermo-electric controller, delivery pump, and tubes by using a dry thermostatic tank filled with metal beads.
We measured two reaction curves obtained from LGr of 10, and 0 µm (reference) samples. Almost identical peak shifts were observed in both cases. All peak shifts during DA flowing had a linear-like response as the solution concentrations increased (5, 16, 50, 160, 500 µM), as shown in Fig. 5(a). These absolute values approximately agree with our calculation results, for which we used a finite-element-method-based commercial simulation software (COMSOL) to estimate the effective refractive index change by varying analyte solution concentration and confinement factor. The inset image corresponds to the mode distribution of the Ex component (electrical field parallel to the substrate) at the maximum DA concentration of 500 µM. According to [13], a modified expression taken in consideration of calibration schemes by using DI and NaCl solution appeared as $\varDelta$λ = λc·Γ·(neff,DA – neff,DI)·η / neff, where Γ is the confinement factor of the graphene-integrated silicon waveguide, neff,DA and neff,DI are the effective refractive indices from optical mode simulation, and η is the occupied length ratio of the graphene-integrated part to silicon part within the circumference of the ring. Although the refractive index of diluted DA solution is an unknown parameter, the measurement peak shift and calibrated sensitivity with NaCl solution provide the necessary parameters for the computation process.
Fig. 5. (a) Peak shift as function of DA-solution concentration obtained from flowing measurement and simulation with COMSOL. Upper-right inset is three-dimensional image of DA structure, right is simulated mode distribution (Ex) at DA concentration of 500 µM. (b) Reaction curve of graphene-integrated sample at concentrations of 5, 16, and 50 µM. (c) Reaction curve of reference silicon ring resonator.
Fig. 6. AFM measurement results on thickness profiles of pristine graphene (after all fabrication steps), activated pristine graphene at 400 °C, and DA-flowed graphene samples with 5 and 50 µM. Inset is topography image of tested pattern at DA concentration of 50 µM.
Another interest from the reaction curves is the DA-adsorption behavior during DI flushing. Figures 5(b) and 5(c) show the extracted curves at the DA concentrations of 5, 16, and 50 µM. In the reaction curve for a graphene-integrated sample (Fig. 5(b)), despite a smooth peak shift of ∼17 pm observed from the 5-µM injection followed by DI flushing, the shift suddenly fell to 50 pm during DI flushing (note that the DI-flushing velocity was 10 times faster than that for all DA-injection periods) after the 16-µM injection (see the blue shaded region). Furthermore, this tendency of staying around 50–60 pm during DI flushing was observed until the entire measurement was completed. This phenomenon can be explained by the matter interaction between the activated graphene surface and DA molecule that couple with each other based on the π-stacking theory. Since the π-bond is a relatively weak coupling contrary to a general covalent bond, every flow cycle causes a cleaning effect rather than sticking to the graphene surface permanently. In other words, only the outermost surface of graphene can capture a few layers of DA molecules, which corresponds to a saturated peak shift of 50–60 pm. Since the reference sample does not have an apparent adsorption feature, as shown in Fig. 5(c), every DI flushing process peels the DA residue up. Thus, the peak shift always returns to the original state instead of increasing or maintaining a constant value. It should be noted that small perturbations (i.e. less than 10 pm shift on average) were observed even for Fig. 5(c), which is possibly caused by some graphene residue or particles around the silicon waveguide, which can capture the DA. The reference silicon ring resonator is integrated on the same chip, so it is fully covered by a graphene sheet once during the fabrication process. Therefore, a small amount of residue can be detected by carrying out the µ-Raman spectroscopy.
To prove this hypothesis in terms of the adsorption mechanism, we carried out a topography measurement for revealing the thickness profile of the graphene surface by AFM in non-contact mode. Figure 6 and the inset show the thickness distribution along the sliced line and the entire topography image of an evaluated square pattern (10 × 10 µm2) of graphene, respectively. To maintain the AFM measurement accuracy, no waveguide structure was defined underneath the graphene sheet, but all other fabrication procedures were identical. While the thickness of pristine graphene is ∼0.5 nm immediately after surface-activation annealing, a thickness of about 1.5–2.5 nm was observed at both 5 and 50 µM. Both conditions were cleaned once by DI flushing before starting AFM measurement to assume DA adsorption. The ripple morphology is related to the structure of the activated graphene surface. The AFM measurement data indicates that the flowing DA was certainly adsorbed on the activated graphene, but the thickness almost saturated rather than increasing linearly. The slowly saturated tendency around 50–60 pm (cf. no change until 500 µM) during DI flushing in Fig. 5(b) reflects this AFM measurement result. We also confirmed that the DA structure in theory has a similar height over that from AFM measurement. By using the gap distance of 0.5–1.0 nm between the pristine graphene and DA-adsorbed one, the calculated thickness (∼0.8 nm) corresponds to that of a few layers of the DA.
The following bonding atomic radius, distance and angle: H:1.2 Å, C:1.7 Å, O:1.4 Å, N:1.5 Å, O-H:0.96 Å, N-H:1.01 Å, C-C:1.40 Å (cf. assuming delocalized state), C-H:1.09 Å, C-N:1.47 Å, C-O:1.43 Å, ∠H-N-H:107.5°, ∠C-C-C:120°, ∠H-C-H:109.5°, ∠C-O-H:104.5° were taken into account in the molecular height calculation. Regarding the three-dimensional image of a DA shown in Fig. 5(a)’s inset, the dominant factor of height is the side chains monoamine. This reveals that the top benzene ring of DA is almost facing down on the lower graphene sheet; otherwise, the total thickness observed from the topography must be much thicker. In other words, if a benzene ring with reduced side chains has a centrosymmetric bonding structure (e.g. benzene and hexafluorobenzene), the AFM observation results and detection limit should be different from our case: thinner, and higher peak shift and adsorption density. This prediction has been theoretically analyzed through molecular dynamics simulation on the interaction between the solubilization/suspension of pristine graphene flakes and other aromatic solvents [34]. Those authors also clarified that there is a difference in the number of interacted layers to graphene flakes for each type of molecule, as well as the coupling density distributions and orientation.
For future work, we will apply two scenarios for our sensor. One will involve immobilizing new linkers, such as DNA or RNA aptamers, on the graphene surface to enable selectivity or enhance the current bio-chemical reaction field. It will allow our sensor to be versatile and multi-functional, as well as overcome the current limitation due to the reaction field occupation. The other scenario will involve expanding the graphene-detection area or adding additional gimmicks to the silicon photonic chip. Thanks to the pattern flexibility in graphene and silicon photonics design enabled by well-organized mature semiconductor processes, the detection dynamic range can be adjustable by changing the integration length.
4. Conclusions
We proposed an on-chip dopamine sensor using an adjusted-coupling designed graphene-silicon ring cavity that suppresses the significant degradation of the Q factor by graphene’s optical absorption. Through the sensing and topography measurements of DA, our sensor revealed meaningful adsorption characteristics including sensitivity from 5 to 500 µM (cf. minimum ∼10−5 in refractive index unit) by the ring resonator featuring graphene, saturation thickness (one to a few layers of DA) on activated clean graphene, and adsorption mechanism due to the π-stacking effect. Both technologies of silicon photonics and graphene have huge potential for changing the world through the various fields of electronics and photonics. We expect this synergy effect will lead to these sensing devices becoming more sophisticated in the near future.
Funding
Ministry of Education, Culture, Sports, Science and Technology (MEXT LEADERs program); JSPS KAKENHI (19K15054).
Acknowledgments
The authors thank T. Watanabe, S. Tanabe, K. Warabi, Y. Hori, H. Nishi, T. Hiraki, and K. Okazaki for their support in fabricating the sensor and helpful discussions.
References
1. T. C. Sudhof, “The synaptic vesicle cycle,” Annu. Rev. Neurosci. 27(1), 509–547 (2004). [CrossRef]
2. M. Zhou, Y. M. Zhai, and S. J. Dong, “Electrochemical Sensing and Biosensing Platform Based on Chemically Reduced Graphene Oxide,” Anal. Chem. 81(14), 5603–5613 (2009). [CrossRef]
3. R. Fernandez-Chacon, A. Konigstorfer, S. H. Gerber, J. Garcia, M. F. Matos, C. F. Stevens, N. Brose, J. Rizo, C. Rosenmund, and T. C. Sudhof, “Synaptotagmin I functions as a calcium regulator of release probability,” Nature 410(6824), 41–49 (2001). [CrossRef]
4. M. A. Kurian, P. Gissen, M. Smith, S. J. R. Heales, and P. T. Clayton, “The monoamine neurotransmitter disorders: an expanding range of neurological syndromes,” Lancet Neurol. 10(8), 721–733 (2011). [CrossRef]
5. A. Pandikumar, G. T. S. How, T. P. See, F. S. Omar, S. Jayabal, K. Z. Kamali, N. Yusoff, A. Jamil, R. Ramaraj, S. A. John, H. N. Lim, and N. M. Huang, “Graphene and its nanocomposite material based electrochemical sensor platform for dopamine,” RSC Adv. 4(108), 63296–63323 (2014). [CrossRef]
6. J. Ng, A. Papandreou, S. J. Heales, and M. A. Kurian, “Monoamine neurotransmitter disorders clinical advances and future perspectives,” Nat. Rev. Neurol. 11(10), 567–584 (2015). [CrossRef]
7. D. Rithesh Raj, S. Prasanth, T. V. Vineeshkumar, and C. Sudarsanakumar, “Surface plasmon resonance based fiber optic dopamine sensor using green synthesized silver nanoparticles,” Sens. Actuators, B 224, 600–606 (2016). [CrossRef]
8. H. Y. Wang, Q. S. Hui, L. X. Xu, J. G. Jiang, and Y. Sun, “Fluorimetric determination of dopamine in pharmaceutical products and urine using ethylene diamine as the fluorigenic reagent,” Anal. Chim. Acta 497(1-2), 93–99 (2003). [CrossRef]
9. W. J. Hu, Y. Y. Huang, C. Y. Chen, Y. K. Liu, T. A. Guo, and B. O. Guan, “Highly sensitive detection of dopamine using a graphene functionalized plasmonic fiber-optic sensor with aptamer conformational amplification,” Sens. Actuators, B 264, 440–447 (2018). [CrossRef]
10. A. Ishizawa, T. Goto, R. Kou, T. Tsuchizawa, N. Matsuda, K. Hitachi, T. Nishikawa, K. Yamada, T. Sogawa, and H. Gotoh, “Octave-spanning supercontinuum generation at telecommunications wavelengths in a precisely dispersion- and length-controlled silicon-wire waveguide with a double taper structure,” Appl. Phys. Lett. 111(2), 021105 (2017). [CrossRef]
11. R. Kou, Y. Hori, T. Tsuchizawa, K. Warabi, Y. Kobayashi, Y. Harada, H. Hibino, T. Yamamoto, H. Nakajima, and K. Yamada, “Ultra-fine metal gate operated graphene optical intensity modulator,” Appl. Phys. Lett. 109(25), 251101 (2016). [CrossRef]
12. P. O. Patil, G. R. Pandey, A. G. Patil, V. B. Borse, P. K. Deshmukh, D. R. Patil, R. S. Tade, S. N. Nangare, Z. G. Khan, A. M. Patil, M. P. More, M. Veerapandian, and S. B. Bari, “Graphene-based nanocomposites for sensitivity enhancement of surface plasmon resonance sensor for biological and chemical sensing: A review,” Biosens. Bioelectron. 139, 111324 (2019). [CrossRef]
13. M. Iqbal, M. A. Gleeson, B. Spaugh, F. Tybor, W. G. Gunn, M. Hochberg, T. Baehr-Jones, R. C. Bailey, and L. C. Gunn, “Label-Free Biosensor Arrays Based on Silicon Ring Resonators and High-Speed Optical Scanning Instrumentation,” IEEE J. Sel. Top. Quantum Electron. 16(3), 654–661 (2010). [CrossRef]
14. W. Bogaerts, P. De Heyn, T. Van Vaerenbergh, K. De Vos, S. K. Selvaraja, T. Claes, P. Dumon, P. Bienstman, D. Van Thourhout, and R. Baets, “Silicon microring resonators,” Laser Photonics Rev. 6(1), 47–73 (2012). [CrossRef]
15. R. Kou, S. Tanabe, T. Tsuchizawa, T. Yamamoto, H. Hibino, H. Nakajima, and K. Yamada, “Influence of graphene on quality factor variation in a silicon ring resonator,” Appl. Phys. Lett. 104(9), 091122 (2014). [CrossRef]
16. R. Kou, Y. Kobayashi, K. Warabi, H. Nishi, T. Tsuchizawa, T. Yamamoto, H. Nakajima, and K. Yamada, “Efficient- and Broadband-Coupled Selective Spot-Size Converters With Damage-Free Graphene Integration Process,” IEEE Photonics J. 6(3), 1–9 (2014). [CrossRef]
17. N. Youngblood, Y. Anugrah, R. Ma, S. J. Koester, and M. Li, “Multifunctional Graphene Optical Modulator and Photodetector Integrated on Silicon Waveguides,” Nano Lett. 14(5), 2741–2746 (2014). [CrossRef]
18. D. Schall, D. Neumaier, M. Mohsin, B. Chmielak, J. Bolten, C. Porschatis, A. Prinzen, C. Matheisen, W. Kuebart, B. Junginger, W. Templ, A. L. Giesecke, and H. Kurz, “50 GBit/s Photodetectors Based on Wafer-Scale Graphene for Integrated Silicon Photonic Communication Systems,” ACS Photonics 1(9), 781–784 (2014). [CrossRef]
19. X. Gan, R.-J. Shiue, Y. Gao, I. Meric, T. F. Heinz, K. Shepard, J. Hone, S. Assefa, and D. Englund, “Chip-integrated ultrafast graphene photodetector with high responsivity,” Nat. Photonics 7(11), 883–887 (2013). [CrossRef]
20. M. S. Dresselhaus, A. Jorio, M. Hofmann, G. Dresselhaus, and R. Saito, “Perspectives on Carbon Nanotubes and Graphene Raman Spectroscopy,” Nano Lett. 10(3), 751–758 (2010). [CrossRef]
21. A. Das, S. Pisana, B. Chakraborty, S. Piscanec, S. K. Saha, U. V. Waghmare, K. S. Novoselov, H. R. Krishnamurthy, A. K. Geim, A. C. Ferrari, and A. K. Sood, “Monitoring dopants by Raman scattering in an electrochemically top-gated graphene transistor,” Nat. Nanotechnol. 3(4), 210–215 (2008). [CrossRef]
22. F. Bonaccorso, Z. Sun, T. Hasan, and A. C. Ferrari, “Graphene photonics and optoelectronics,” Nat. Photonics 4(9), 611–622 (2010). [CrossRef]
23. M. Liu, X. B. Yin, and X. Zhang, “Double-Layer Graphene Optical Modulator,” Nano Lett. 12(3), 1482–1485 (2012). [CrossRef]
24. M. Liu, X. B. Yin, E. Ulin-Avila, B. S. Geng, T. Zentgraf, L. Ju, F. Wang, and X. Zhang, “A graphene-based broadband optical modulator,” Nature 474(7349), 64–67 (2011). [CrossRef]
25. R. Kou, S. Tanabe, T. Tsuchizawa, K. Warabi, S. Suzuki, H. Hibino, H. Nakajima, and K. Yamada, “Characterization of Optical Absorption and Polarization Dependence of Single-Layer Graphene Integrated on a Silicon Wire Waveguide,” Jpn. J. Appl. Phys. 52(6R), 060203 (2013). [CrossRef]
26. A. Majumdar, J. Kim, J. Vuckovic, and F. Wang, “Electrical Control of Silicon Photonic Crystal Cavity by Graphene,” Nano Lett. 13(2), 515–518 (2013). [CrossRef]
28. Y. C. Lin, C. C. Lu, C. H. Yeh, C. H. Jin, K. Suenaga, and P. W. Chiu, “Graphene Annealing: How Clean Can It Be?” Nano Lett. 12(1), 414–419 (2012). [CrossRef]
29. G. Cocorullo, F. G. Della Corte, and I. Rendina, “Temperature dependence of the thermo-optic coefficient in crystalline silicon between room temperature and 550 K at the wavelength of 1523 nm,” Appl. Phys. Lett. 74(22), 3338–3340 (1999). [CrossRef]
30. H. H. Li, “Refractive index of alkali halides and its wavelength and temperature derivatives,” J. Phys. Chem. Ref. Data 5(2), 329–528 (1976). [CrossRef]
31. W. Lu and W. M. Worek, “Two-wavelength interferometric technique for measuring the refractive index of salt-water solutions,” Appl. Opt. 32(21), 3992–4002 (1993). [CrossRef]
32. W. M. Haynes, CRC handbook of chemistry and physics95th ed. (CRC, 2015-2016).
33. T. E. Peddicord, K. M. Olsen, T. L. ZumBrunnen, D. J. Warner, and L. Webb, “Stability of high-concentration dopamine hydrochloride, norepinephrine bitartrate, epinephrine hydrochloride, and nitroglycerin in 5% dextrose injection,” Am. J. Health-Syst. Pharm. 54(12), 1417–1419 (1997). [CrossRef]
34. A. J. Oyer, J.-M. Y. Carrillo, C. C. Hire, H. C. Schniepp, A. D. Asandei, A. V. Dobrynin, and D. H. Adamson, “Stabilization of Graphene Sheets by a Structured Benzene/Hexafluorobenzene Mixed Solvent,” J. Am. Chem. Soc. 134(11), 5018–5021 (2012). [CrossRef]
