This work reports on the fabrication and characterization of a graphene based variable optical attenuator integrated on a photonic Si3N4 waveguide and operating at 855 nm wavelength. The variable optical attenuator utilizes the gate voltage dependent optical absorption of a graphene layer, located in the evanescent field of the waveguide. A maximum attenuation of 17 dB is obtained at −3 V gate voltages for a device length of 700 µm. The measured voltage dependent absorption was found to be in good agreement with theoretical simulations, taking into account inter- and intra-band optical conductivity of graphene. An outlook is given on possible margins for increasing the operation speed and reducing the insertion loss of the device, using an optimized layout and improved fabrication processes.
© 2017 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Variable optical attenuators (VOA) are key components in integrated photonics and optical communication networks finding application in simulating optical links during testing and link power budget verification [1,2], minimizing output signal intensity from laser diodes in order not to overdrive/saturate the receivers to the nonlinear regime , or equalizing the power outputs from optical amplifiers in wavelength division multiplexing (WDM) as optical amplifiers amplify the incoming WDM signal based on individual wavelengths [3–5]. By far the most common type of on-chip VOA is based on silicon pin-diodes biased in the forward direction. These devices have been commercially available for more than a decade and offer excellent key performance parameters, including an attenuation of >25 dB with a total insertion loss of only 2 dB and 1 µs electro-optical response time. However, the operation of these VOAs is limited to near infrared wavelengths around 1310 nm and 1550 nm due to silicon bandgap. For the optical spectrum around 850 nm wavelength, which is also widely used e.g. for short distance optical communication, there are so far no on-chip VOAs available. Instead only discrete and bulky VOAs exist, which are typically based on the mechanical manipulation of an optical system in order to control the optical output power . These devices are not an option for chip-integrated photonic systems and thus another approach is needed to cover this functionality on-chip for the wavelength range around 850 nm.
Graphene has been considered a promising material for on-chip photonic applications because of its unique electronic [7–11] and broadband optical [12–15] properties and because of the possibility of monolithically integrating the graphene based photonic devices on wafer-scale silicon platforms . So far, several waveguide integrated opto-electronic graphene based devices have been realized operating at 1310 and 1550 nm wavelengths including electro-optical phase and amplitude modulators [17,18], on-chip photodetectors [19–21] and thermal tuning elements for silicon waveguides . A graphene based VOA for wavelengths around 800 nm has so far only been demonstrated as a discrete free space component with 5% attenuation at 785 nm . In this work, an on-chip integrated VOA is demonstrated operating at 855 nm wavelength using a gated graphene layer located in the evanescent field of a Si3N4 waveguide. The absorption of the graphene layer is controlled by applying a gate voltage, which tunes the Fermi-level from zero to values below −0.8 eV, so that the inter-band absorption for 855 nm light gets blocked due to Pauli-blocking and the graphene becomes transparent. The presented VOA achieves an optical attenuation up to 17 dB for a device length of 700 µm, which is comparable to silicon photonics VOAs at 1550 nm . These results are in excellent agreement to theoretical simulations taking into account the gate voltage dependent absorption of the graphene layer.
2. Device Fabrication
In order to realize the graphene based on-chip VOAs, we fabricated Si3N4 photonic waveguides on an oxidized silicon wafer and placed graphene on top of the Si3N4 waveguide within the evanescent field of the waveguide. A schematic of the fabricated graphene VOA is shown in Fig. 1.
The first step of the device fabrication process was the thermal growth of 2 µm SiO2 on an Si wafer followed by LPCVD deposition of 360 nm Si3N4. Si3N4 waveguides were then defined using optical lithography (i-line stepper) with subsequent dry etching. The width of the waveguide was 800 nm in the region of the VOA. Grating couplers, optimized for 855 nm wavelength TE polarized light, were used to couple light in and out of the waveguides. These grating couplers were defined by e-beam lithography and dry etching. At this stage of fabrication, the optical propagation loss of the Si3N4 waveguides was measured using the cutback method to be 2 dB/cm, which is a typical value for Si3N4 waveguides at these wavelengths . The insertion loss of the grating coupler is ~8 - 9 dB per coupler. Afterwards, hydrogen silsesquioxane (HSQ) was spin-coated as cladding on the sample and thermally cured [26,27]. The spin-coated HSQ was used to smoothen the step edges of the waveguide and thus avoids cracking of the graphene layer. Afterwards, 5 nm of Al2O3 was deposited by plasma assisted atomic layer deposition. A cross-sectional SEM image of the chip with HSQ cladding is shown in Fig. 2(a), demonstrating the smoothening of the step-edges and a final thickness of the HSQ + Al2O3 on the waveguide of approximately 25 nm.
In the next step, using a standard PMMA wet-transfer method similar to the method described in , CVD grown single layer graphene (Graphenea SE) was transferred onto the sample. Afterwards, the graphene layer was patterned on the waveguides to lengths from 200 to 700 μm in steps of 100 μm using optical contact photolithography and O2 plasma. Subsequently, Ni contacts to the graphene were fabricated by sputter deposition using contact photolithography and lift-off technique. For achieving high doping levels in the graphene layer, 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide (EMIM:TFSI) was spin coated on the sample to act as an ionic liquid gate, enabling high doping levels up to 1014 cm−2. An SEM image of the final device is shown in Fig. 2(b).
3. Results and Discussion
All electrical, optical and electro-optic measurements were performed in ambient air at room temperature using a needle probe station with optical fibers. Figure 3 shows the gate voltage dependent resistance of a VOA for a graphene length of 400 μm. For this device, a resistance modulation of ~4.4 was obtained, which is limited by the contact resistance . The graphene is slightly n-doped with the charge neutrality point (CNP) at Vgate = −560 mV. The charge carrier mobility was measured by Hall measurement on a reference Hall bar on the same chip. Values of 3000 cm2/Vs and 260 cm2/Vs were measured at a carrier concentration of 1x1012 cm−2 and 5x1013 cm−2, respectively, which are typical values for commercial CVD graphene. The resistance modulation in the Hall bar was 12. The gate voltage dependent optical transmission of the 400 µm long VOA is also shown in Fig. 3. The overall change of the optical transmission, for this VOA with 400 μm graphene length, was 9.8 dB when Vgate was varied from 0 to −3 V. At voltages around the CNP, the graphene is absorbing and the transmission through the device is minimal. Increasing the gate voltage to higher negative values causes the Fermi level in graphene to shift away from the CNP resulting in a strong increase in the transmission between −1.5 and −3 V as the Fermi level reaches half of the photon energy (0.73 eV at 855 nm) where Pauli blocking is suppressing the absorption.
In total, we measured the electro-optic properties of 14 VOAs on a single chip. For these VOAs, the graphene length varied from 200 to 700 µm and we found a linear relation between the VOA length and the Vgate dependent optical transmission change, which reached a maximum value of 17 dB for the longest graphene length of 700 µm when Vgate was varied from 0V to −3 V (see Fig. 4(a)). The slope of the linear fit in Fig. 4(a) gives a length dependent transmission change of 0.024 dB/µm. This value is comparable to silicon based VOAs at 1550 nm where a typical value of 20 dB is achieved for a device length of 1 mm (i.e. 0.02 dB/µm)  and enables the realization of compact on-chip VOAs for 855nm wavelength.
In addition to the attenuation amplitude, the insertion loss is also an important parameter for VOAs. The residual fiber-to-fiber insertion loss of the VOAs in the transparent state, i.e. the optical transmission at Vgate = −3 V, is plotted in Fig. 4(b) (black symbols) and ranges from 17 to 23 dB. These values include the loss of the waveguide, two grating couplers and the residual absorption of the graphene layer. As discussed in ‘Device Fabrication’ section, approximately 17-18 dB of this insertion loss originates from the two grating couplers, which is by far the main contribution to the total insertion loss. For a fully packaged device, there are much more efficient couplers available  so that this contribution can be significantly minimized and it will not be dependent on the type of VOA used. The contribution from the straight Si3N4 waveguide is negligible as the measured loss is only 2 dB/cm. Hence, in order to provide a meaningful parameter of the insertion loss for the VOA presented here, the residual insertion loss of the graphene layer needs to be estimated. As all 14 VOAs are identical (grating couplers, length and design of the Si3N4 waveguide), with only the length of the graphene on the waveguide being varied, the slope in Fig. 4(b) at Vgate = −3 V directly gives the length specific insertion loss of the graphene of 0.005 dB/µm. Hence, the ratio of the attenuation amplitude to the residual insertion loss at −3V is found to be ~5, which is comparable to graphene based amplitude modulators operating at 1550 nm , but significantly lower compared to Si based VOAs operating at 1550 nm, where a ratio of 150 is achieved  if losses of the packaging are also not considered. For completeness, the fiber-to-fiber transmission in the absorbing state, i.e. at zero gate voltage, is also plotted in Fig. 4(b) (red symbols). The y-axis intercept of both linear fits in Fig. 4(b), which corresponds to the total losses of the grating couplers and the Si3N4 waveguide, is extracted to be 17.5 dB demonstrating very good consistency of the measurements.
After the DC measurements, the AC response of the device was characterized in order to extract the corresponding rise and fall times. For these measurements, the gate electrode was connected to a square wave AC signal with a frequency of 1 Hz and 3 V amplitude. Figure 5 shows the time-dependent optical transmission for a VOA with 300 µm length. The extracted rise and fall time (10%-90%) are 100 ms and 60 ms, respectively, as depicted in Fig. 5. This relatively low operation speed is because an ionic liquid has been utilized for gating. In addition, the device structure has not been optimized for minimizing the parasitic capacitances and resistances. The ionic liquid is everywhere on the chip and hence not only the active device area is gated, i.e. the graphene on top of the waveguide, but also the access graphene and the contact pads, dramatically increasing the total capacitance. Also, the contacts are further away from the waveguide than necessary and the contact resistance is quite large, both leading to an unnecessarily high access resistance. Assuming a 700 µm long VOA, with a 1 µm wide gated graphene area, a capacitance of 5 µF/cm2 as it was measured in our device, a total device capacitance of 35 pF is calculated. Further, assuming a charge carrier mobility in the graphene of 230 cm2/Vs at a doping level of 5x1013 cm−2, which corresponds to the operation point of the VOA, a distance of the contact metal from the waveguide center of 1 µm, and a reasonable contact resistance of 1 kΩµm , a series resistance of 2.3 Ω is calculated. This leads to an intrinsic bandwidth of 2 GHz, which is fully sufficient for VOA operation. Even higher operation frequencies might be possible if the device layout is further optimized and state of the art graphene (µ > 1000cm2/Vs ) and metallic contacts (contact resistance below 100 Ωµm ) are used. The major challenges for increasing the speed in graphene based VOAs will be the integration of a dielectric into the device providing a sufficiently high dielectric constant and breakdown voltage, which is required to achieve the high carrier concentration and high capacitance. The measured capacitance of 5 µF/cm2 corresponds to an equivalent oxide thickness of 0.7 nm, which is possible with standard high-k dielectrics and used in state-of-the-art CMOS technology. However, the required carrier concentration level of up to 1x1014 cm−2 is not achievable with conventional dielectrics such as HfO2 or Al2O3, because of the dielectric breakdown. SrTiO3 could be a suitable candidate for a dielectric offering a relative dielectric constant well above 100 and a breakdown voltage of up to 3 MV/cm .
In order to gain information on the limiting factor for the insertion loss and the impact of the graphene quality on the VOA performance, FDTD simulations were performed using MATLAB. For such simulations, a material stack consisting of SiO2-Si3N4-HSQ-Al2O3-graphene-ionic liquid-air, which mimics the experimentally fabricated device, is considered as depicted in Fig. 6. Figure 6 also mentions the used refractive indices for each individual material except for graphene, whose refractive index is dependent on the chemical potential and is calculated as follows: The total optical conductivity of graphene (σ) is considered as a sum of inter (σinter) and intra-band (σintra) components which are defined using the Kubo formalism as ,36],
The solution of Eq. (6) gives complex eigenvalues with real and imaginary parts representing neff and absorption, respectively. Figure 7(a) depicts the simulated values of absorption as a function of graphene distance from top of waveguide. The distance between graphene and top of waveguide is filled with varying HSQ thicknesses and 5 nm Al2O3 on top of HSQ. For a distance of 35 nm (30 nm HSQ + 5 nm Al2O3) between graphene and top of waveguide, an absorption matching the experimental value of 0.028 dB/µm is derived from the simulations. This distance is close to the value measured in the SEM picture in Fig. 2(a), and the difference can be explained by uncertainty in measuring the thickness of this layer in an SEM, by sample-to-sample variations of the HSQ spin-coating process and uncertainty in measuring the absorption of the graphene VOA. Using a distance of 35 nm, the simulated graphene absorption as a function of chemical potential (μc) is plotted in Fig. 7(b) for different values of scattering rates (dashed lines). The μc dependent absorption in graphene is fully symmetric around zero, but for clarity only negative values are plotted as the experimental measurements were also only performed for negative gate voltages. For chemical potentials close to zero, the simulated graphene absorption does not depend on different scattering rates but in the transparent regime (|μc| > 0.8 eV), there is a strong dependency of the absorption on the scattering rates with low residual insertion loss obtained for a low scattering rate (i.e. high mobility). The scattering rate (Γ) is inversely proportional to the charge carrier mobility (μ) in graphene via μ = (eυF2)/(Γμc), υF and e denote Fermi velocity and electron charge, respectively. For comparison, the experimentally obtained graphene absorption (solid line) is also plotted in Fig. 7(b) and is consistent with simulated values for Γ = 5x1013 s−1, corresponding to μ = 223 cm2/Vs at μc = 0.726 eV. The Hall measurements on the reference device gave an experimental mobility value of 260 cm2/Vs at carrier concentration of 4.8x1013 cm−2 (corresponding to µC = 0.726 eV) demonstrating excellent consistency between simulations and experiments.
In order to estimate the effect of the scattering rate on the residual absorption of the graphene at higher chemical potentials, which translates into the insertion loss of the VOA, a figure-of-merit (FOM) has been defined as the ratio of graphene absorption at 0 eV to the absorption at −1 eV, corresponding to the ratio of attenuation and insertion loss for a VOA. We observe a nearly linear relation between the FOM and the scattering rate, as depicted in inset of Fig. 7(b), with lower scattering rate providing a higher FOM. For the VOAs, with a scattering rate of 5x1013 sec−1 as in the devices presented here, a FOM of 11 is derived. The measured Hall mobility in our device is well below the phonon limited mobility value, which is predicted to be 2500 cm2/Vs at a carrier concentration of 5x1013 cm−2 . Recently, a mobility value of 1870 cm2/Vs at a high carrier concentrations of 1x1014 cm−2 has been reported for HNO3 doped CVD grown single layer graphene , corresponding to a scattering rate of 3x1012 s−1. Therefore considerably larger values of FOM are reachable if graphene with higher mobility at the required carrier concentration is used and it is also conceivable to approach the values for Si-based VOAs where a FOM of 150 is observed . Such a competitive FOM opens up applications at 1310 and 1550 nm wavelength, which are currently covered by Si-based VOAs, however with the unique advantage of graphene based VOAs of being compatible with dielectric waveguides such as Si3N4 and covering a broadband operation spectrum from 855 to 1550 nm wavelength. In addition, the high mobility not only reduces the insertion loss significantly, but can also increase the operation speed by reducing the graphene sheet resistance.
In conclusion, we reported on the fabrication and characterization of a graphene based VOA on a Si3N4 waveguide for 855 nm wavelength operation, which is the first on-chip VOA available at this wavelength. The length specific attenuation of the measured devices was 0.024 dB/µm, leading to a maximum attenuation of 17 dB for the 700 µm long VOA. Theoretical simulations confirm these values and suggest that the residual insertion loss can be significantly reduced by using graphene with higher carrier mobility. In the measured devices, the operation speed was limited by huge parasitic capacitances and the used ionic liquid gating scheme. An intrinsic RC limited response time of 2 GHz is estimated, with further margins for increasing the operation speed by optimizing the layout and the fabrication process. The VOA presented here is designed for single mode optical communication systems, however, the application can be extended to multimode operation by properly adjusting the dimensions of the waveguide and the coupling from the fiber to the waveguide.
European Commission project Graphene Flagship (contract no. 696656).
References and links
1. S. Zhang, M. Arabaci, F. Yaman, I. B. Djordjevic, L. Xu, T. Wang, Y. Inada, T. Ogata, and Y. Aoki, “Experimental study of non-binary LDPC coding for long-haul coherent optical QPSK transmissions,” Opt. Express 19(20), 19042–19049 (2011). [PubMed]
2. B. Li, Z. Feng, M. Tang, Z. Xu, S. Fu, Q. Wu, L. Deng, W. Tong, S. Liu, and P. P. Shum, “Experimental demonstration of large capacity WSDM optical access network with multicore fibers and advanced modulation formats,” Opt. Express 23(9), 10997–11006 (2015). [PubMed]
3. G. Galliano, P. Motta, and F. Montalti, “Features in spectral attenuation measurement and performances of plug-in optical fixed attenuators,” in Proceedings of IEEE/LEOS Workshop on Fibre and Optical Passive Components (IEEE, 2002), pp. 189–194.
4. Q. Fang, J. Song, G. Zhang, M. Yu, Y. Liu, G. Q. Lo, and D. L. Kwong, “Monolithic integration of a multiplexer/demultiplexer with a thermo-optic VOA array on an SOI platform,” IEEE Photonics Technol. Lett. 21(5), 319–321 (2009).
5. K. Sato, T. Aoki, Y. Watanabe, Y. Oguri, N. Shibata, T. Nishiwaki, M. Oike, and T. Ohta, “Development of a variable optical attenuator,” Furukawa Electric Review, 15–20 (2001).
6. The Fiber Optic Association, http://www.thefoa.org/tech/ref/appln/attenuators.html.
7. K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. V. Dubonos, I. V. Grigorieva, and A. A. Firsov, “Electric field effect in atomically thin carbon films,” Science 306(5696), 666–669 (2004). [PubMed]
8. E. H. Hwang, S. Adam, and S. D. Sarma, “Carrier transport in two-dimensional graphene layers,” Phys. Rev. Lett. 98(18), 186806 (2007). [PubMed]
9. A. H. C. Neto, F. Guinea, N. M. R. Peres, K. S. Novoselov, and A. K. Geim, “The electronic properties of graphene,” Rev. Mod. Phys. 81(1), 109 (2009).
10. T. Ando, “The electronic properties of graphene and carbon nanotubes,” NPG Asia Mater. 1(1), 17–21 (2009).
11. M. F. Craciun, S. Russo, M. Yamamoto, and S. Tarucha, “Tuneable electronic properties in graphene,” Nano Today 6(1), 42–60 (2011).
12. F. Wang, Y. Zhang, C. Tian, C. Girit, A. Zettl, M. Crommie, and Y. R. Shen, “Gate-variable optical transitions in graphene,” Science 320(5873), 206–209 (2008). [PubMed]
13. L. A. Falkovsky, “Optical properties of graphene,” J. Phys. Conf. Ser. IOP Publishing 129(1), 012004 (2008).
14. S. J. Koester and M. Li, “Waveguide-coupled graphene optoelectronics,” IEEE J. Sel. Top. Quantum Electron. 20(1), 84–94 (2014).
15. R. R. Nair, P. Blake, A. N. Grigorenko, K. S. Novoselov, T. J. Booth, T. Stauber, N. M. Peres, and A. K. Geim, “Fine structure constant defines visual transparency of graphene,” Science 320(5881), 1308 (2008). [PubMed]
16. D. Schall, C. Porschatis, M. Otto, and D. Neumaier, “Graphene photodetectors with a bandwidth >76 GHz fabricated in a 6″ wafer process line,” J. Phys. D Appl. Phys. 50(12), 124004 (2017).
17. M. Liu, X. Yin, and X. Zhang, “Double-layer graphene optical modulator,” Nano Lett. 12(3), 1482–1485 (2012). [PubMed]
18. M. Mohsin, D. Neumaier, D. Schall, M. Otto, C. Matheisen, A. L. Giesecke, A. A. Sagade, and H. Kurz, “Experimental verification of electro-refractive phase modulation in graphene,” Sci. Rep. 5, 10967 (2015). [PubMed]
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).
20. A. Pospischil, M. Humer, M. M. Furchi, D. Bachmann, R. Guider, T. Fromherz, and T. Mueller, “CMOS-compatible graphene photodetector covering all optical communication bands,” Nat. Photonics 7(11), 892–896 (2013).
21. 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).
22. D. Schall, M. Mohsin, A. A. Sagade, M. Otto, B. Chmielak, S. Suckow, A. L. Giesecke, D. Neumaier, and H. Kurz, “Infrared transparent graphene heater for silicon photonic integrated circuits,” Opt. Express 24(8), 7871–7878 (2016). [PubMed]
23. J. L. Benítez and D. Mendoza, “Modulation of the optical transmittance in multilayer graphene by an electrical signal,” Appl. Phys. Lett. 103(8), 083116 (2013).
24. H. Nishi, T. Tsuchizawa, T. Watanabe, H. Shinojima, K. Yamada, and S. Itabashi, “Compact and polarization-independent variable optical attenuator based on a silicon wire waveguide with a carrier injection structure,” Jpn. J. Appl. Phys. 49(4S), 04DG20 (2010).
25. A. Z. Subramanian, P. Neutens, A. Dhakal, R. Jansen, T. Claes, X. Rottenberg, F. Peyskens, S. Selvaraja, P. Helin, B. Du Bois, K. Leyssens, S. Severi, P. Deshpande, R. Baets, and P. Van Dorpe, “Low-loss singlemode PECVD silicon nitride photonic wire waveguides for 532–900 nm wavelength window fabricated within a CMOS pilot line,” IEEE Photonics J. 5(6), 2202809 (2013).
26. S. Choi, M. J. Word, V. Kumar, and I. Adesida, “Comparative study of thermally cured and electron-beam-exposed hydrogen silsesquioxane resists,” J. Vac. Sci. Technol. B 26(5), 1654–1659 (2008).
27. C. C. Yang and W. C. Chen, “The structures and properties of hydrogen silsesquioxane (HSQ) films produced by thermal curing,” J. Mater. Chem. 12(4), 1138–1141 (2002).
28. X. Li, W. Cai, J. An, S. Kim, J. Nah, D. Yang, R. Piner, A. Velamakanni, I. Jung, E. Tutuc, S. K. Banerjee, L. Colombo, and R. S. Ruoff, “Large-area synthesis of high-quality and uniform graphene films on copper foils,” Science 324(5932), 1312–1314 (2009). [PubMed]
29. C. A. Chavarin, A. A. Sagade, D. Neumaier, G. Bacher, and W. Mertin, “On the origin of contact resistances in graphene devices fabricated by optical lithography,” Appl. Phys., A Mater. Sci. Process. 122(2), 58 (2016).
30. S. Romero-García, F. Merget, F. Zhong, H. Finkelstein, and J. Witzens, “Visible wavelength silicon nitride focusing grating coupler with AlCu/TiN reflector,” Opt. Lett. 38(14), 2521–2523 (2013). [PubMed]
31. M. Mohsin, D. Schall, M. Otto, A. Noculak, D. Neumaier, and H. Kurz, “Graphene based low insertion loss electro-absorption modulator on SOI waveguide,” Opt. Express 22(12), 15292–15297 (2014). [PubMed]
32. K. Yamada, T. Tsuchizawa, T. Watanabe, H. Fukuda, H. Shinojima, and S. Itabashi, “Applications of low-loss silicon photonic wire waveguides with carrier injection structures,” in 4th IEEE Int. Conf. on Group IV Photonics (IEEE, 2007), pp. 1–3.
33. M. Kim, A. Shah, C. Li, P. Mustonen, J. Susoma, F. Manoocheri, J. Riikonen, and H. Lipsanen, “Direct transfer of wafer-scale graphene films,” 2D Materials 4(3), 1–10 (2017).
34. R. Ifuku, K. Nagashio, T. Nishimura, and A. Toriumi, “The density of states of graphene underneath a metal electrode and its correlation with the contact resistivity,” Appl. Phys. Lett. 103(3), 033514 (2013).
35. C. Neusel and G. A. Schneider, “Size-dependence of the dielectric breakdown strength from nano-to millimeter scale,” J. Mech. Phys. Solids 63, 201–213 (2014).
36. A. B. Fallahkhair, K. S. Li, and T. E. Murphy, “Vector finite difference modesolver for anisotropic dielectric waveguides,” J. Lightwave Technol. 26(11), 1423–1431 (2008).
37. L. Wang, I. Meric, P. Y. Huang, Q. Gao, Y. Gao, H. Tran, T. Taniguchi, K. Watanabe, L. M. Campos, D. A. Muller, J. Guo, P. Kim, J. Hone, K. L. Shepard, and C. R. Dean, “One-dimensional electrical contact to a two-dimensional material,” Science 342(6158), 614–617 (2013). [PubMed]