透過您的圖書館登入
IP:3.14.142.115
  • 學位論文

積體化矽基高速發射器應用於光連結系統

Integrated High Speed Si-based Transmitters for Optical Interconnect

指導教授 : 李明昌

摘要


由於矽光子學元件高度相容於先進半導體製程技術,具備大量生產能力、高密度整合特性,透過光纖,以光訊號傳輸,有高頻寬、低損耗優勢,因此近年來吸引許多研究人員投入,過去幾年來的努力,積體化光調變器、光偵測器可實現40 Gbit/s以上高速的數據傳輸,未來非常有機會可應用在終端的光發射器以及光接收器,作電/光或光/電轉換。 透過自由載子的濃度改變量造成的色散效應是目前應用在光調變器中,常見的機制。一般可以分為注入式與空乏式的光調變器,空乏式光調變器操作速度受多數載子生命週期主控,因此操作速度比注入式光調變器要快,但,空乏式光調變器的調變效率不如注入式光調變器來得好,因此元件尺寸較大,為了使空乏式光調變器有更好的調變效率,重參雜區域必須儘量靠近光波導中心附近,然而,卻同時造成更多的自由載子吸收,因為如此,許多研究團隊,提出pipin diode與doping compensation method的方式,改善調變效率,同時抑制額外的自由載子吸收,但,這兩種方式需要很精確的控制參雜濃度分布。有鑑於此,我們提出邊緣場pn接面(fringe field pn junctions)光調變器,可有效提升調變效率與降低自由載子吸收,此外,以元件製程的角度看,fringe field junctions的位置是由自我對準離子佈植(self-aligned ion implantation)所定義,因此pn接面的位置可被準確地控制在光波導兩側。 積體化光發射器,因尺寸小,若要與光纖結合,因光波導與光纖界面的面積尺寸差異一百倍以上,因此波導與光纖界面的耦合損耗至少會超過20 dB,目前,grating coupler與inverted taper是目前常見的改善方法,grating coupler收光面積大,對於光纖的對準誤差容忍度高,但穿透頻譜窄,因此可利用的光波長有範圍的限制,而inverted taper的穿透頻譜寬,有利於結合WDM作多通道的光發射器,但是,受限於截面積大小(~3 um),光纖的對準誤差容忍度大約1 m。因此,我們以SU-8光阻利用玻璃基板壓印的方式作垂直方向taper,微影製程作側向taper,其taper截面積從10 um x 10 um(寬 x 高)漸變地縮小至4 um x 1 um的SiON光波導,再透過inverted taper,將光波從SiON 波導耦合進入矽奈米線,透過實作與量測,從光纖到矽奈米線的耦合損耗為2.8 dB (TE模態),穿透頻譜趨勢平坦,有利於結合WDM作多通道光發射器,光纖的對準誤差容忍度為~3 um,相對降低對於元件封裝上的困難度。

並列摘要


Silicon photonics, with the unique advantages in cost-effective, mass-production and high dense integration resulting from the compatibility with the advanced CMOS technology, have developed rapidly in the recent years and been one of the key building blocks for optical interconnection. Various high performances of optical modulators and detectors have been implemented on silicon substrate for high speed E/O and O/E conversion. The plasma dispersion effect is the often seen approach applied to the optical modulation on silicon substrate over the past one decade. Compared with the injection type modulators, the depletion type modulators provide faster switching rate resulting from the virtue of short majority lifetime. But the modulation efficiency of carrier-depletion modulator is not superior to that of carrier-injection leading to a large device area. To address this problem, the location of heavily doped region of pn junction should be as close to the waveguide center as possible to obtain maximal overlap integral between modulated depletion region and optical mode so as to achieve best modulation efficiency. Nevertheless, the free carrier absorption inevitably increases because of the heavily doped region. Many attempts, to date, have been made for reducing phase shifter loss such as pipin diode and doping compensation method. But, each of them needs to stringently control the doping profile to keep the modulation efficiency while avoid strong free carrier absorption as well. In this study, we propose a new device structure to exploit fringe field pn junctions by deploying heavily doped regions just near the corners of the waveguide to mitigate free carrier absorption. In the meantime, the large fringe field at these doped regions effectively depletes the carriers resulting in extended depletion region across the waveguide center. These doped regions can be precisely controlled by self-aligned ion implantation as well as a post annealing process. Silicon integrated photonics shows great potential for applications in optical interconnect and optical signal processing. Many state-of-the-art active components such as Si modulators and Si/Ge photodetectors have been demonstrated for high-speed data transmission exceeding 40 Gbps. However, a large core-dimension discrepancy between the submicron silicon waveguides (or called silicon photonic wires) and single-mode fibers (SMF) introduces significant coupling loss. To solve this problem, many approaches have been proposed, for example, surface coupling by grating couplers, end-butt coupling by tapered lensed fibers or Si inverted nanotapers. Silicon inverted nanotapers have been studied widely with high coupling efficiency and small polarization- and wavelength-dependence. Nevertheless, through this method, another coupler with low index of refraction is usually employed to enclose the inverted taper as an inter-medium that confines light for better fiber-to-silicon-wire coupling. The core dimension of this extra coupler is usually limited to several m to achieve effective coupling. Though the lensed fiber has better mode matching with this extra coupler, the misalignment tolerance is very stringent, typically less than 1 m. A practical coupler design not only should be low-loss but ought to tolerate fiber misalignment as well. In this study, this 3D SU-8 taper is cascaded with an inverted nanotaper through the SiON waveguide for light coupling from SMF to silicon photonic wires. In the section 4-2, we have a brief discussion about the integration between fringe field junction modulator and wideband fiber coupler in the process point of view.

參考文獻


[1]C. Dragone, "An N*N optical multiplexer using a planar arrangement of two star couplers," IEEE Photonics Technology Letters, vol. 3, pp. 812-815, 1991.
[2]C. Dragone, C. A. Edwards, and R. C. Kistler, "Integrated optics N*N multiplexer on silicon," IEEE Photonics Technology Letters, vol. 3, pp. 896-899, 1991.
[3]X. Fang and R. O. Claus, "Polarization-independent all-fiber wavelength-division multiplexer based on a Sagnac interferometer," Opt Lett, vol. 20, pp. 2146-8, Oct 15 1995.
[4]D. W. Kim, A. Barkai, R. Jones, N. Elek, H. Nguyen, and A. Liu, "Silicon-on-insulator eight-channel optical multiplexer based on a cascade of asymmetric Mach-Zehnder interferometers," Opt Lett, vol. 33, pp. 530-2, Mar 1 2008.
[5]H. Takahashi, K. Oda, and H. Toba, "Impact of crosstalk in an arrayed-waveguide multiplexer on NxN optical interconnection," Journal of Lightwave Technology, vol. 14, pp. 1097-1105, Jun 1996.

延伸閱讀