OSA's Digital Library

Optics Letters

Optics Letters

| RAPID, SHORT PUBLICATIONS ON THE LATEST IN OPTICAL DISCOVERIES

  • Editor: Alan E. Willner
  • Vol. 37, Iss. 21 — Nov. 1, 2012
  • pp: 4356–4358
« Show journal navigation

Compact grating couplers on silicon-on-insulator with reduced backreflection

Yanlu Li, Diedrik Vermeulen, Yannick De Koninck, Gunay Yurtsever, Günther Roelkens, and Roel Baets  »View Author Affiliations


Optics Letters, Vol. 37, Issue 21, pp. 4356-4358 (2012)
http://dx.doi.org/10.1364/OL.37.004356


View Full Text Article

Acrobat PDF (406 KB)





Browse Journals / Lookup Meetings

Browse by Journal and Year


   


Lookup Conference Papers

Close Browse Journals / Lookup Meetings

Article Tools

Share
Citations

Abstract

The backreflection in commonly used grating couplers on silicon-on-insulator (SOI) is not negligible for many applications. This reflection is dramatically reduced in our improved compact grating coupler design, which directs the reflection away from the input waveguide. Realized devices on SOI show that the reflection can be reduced down to 50dB without an apparent transmission penalty.

© 2012 Optical Society of America

Surface grating coupler (GC) structures are extensively used in photonic integrated circuits (PICs) to couple light into and out of PICs. This coupling method does not require polished waveguide facets on the edge of a photonic chip, and can thus be used in wafer-scale testing of PICs. Improvements on the coupling efficiency and bandwidth of GC structures (to free space or to a single-mode fiber) have been thoroughly studied in recent publications [1

1. D. Taillaert, F. Van Laere, M. Ayre, W. Bogaerts, D. Van Thourhout, P. Bienstman, and R. Baets, Jpn. J. Appl. Phys. 45, 6071 (2006).

3

3. A. Mekis, S. Gloeckner, G. Masini, A. Narasimha, T. Pinguet, S. Sahni, and P. De Dobbelaere, IEEE J. Sel. Topics. Quantum Electron. 17, 597 (2011). [CrossRef]

]. Apart from that, the on-chip backreflection of GC structures is also often discussed, because it can deteriorate the behavior of some integrated components, e.g., an integrated laser. Two types of reflections may be introduced in a GC: the second-order Bragg reflection of the grating and the Fresnel reflection due to the mode mismatch between the grating region and the input waveguide. The zenith of the fiber ϕ (the angle between the normal of the PIC surface and the direction of the fiber) is often set as 10° or more to make sure that the second-order reflection is suppressed in the wavelength range of interest. However, the Fresnel reflection can still be reflected back to the waveguide. A suppression in the backreflection was reported by improving the mode matching between the grating region and input waveguide [4

4. C. Alonso-Ramos, A. Ortega-Monũx, I. Molina-Fernández, P. Cheben, L. Zavargo-Peche, and R. Halir, Opt. Express 18, 15189 (2010). [CrossRef]

,5

5. N. Na, H. Frish, I.-W. Hsieh, O. Harel, R. George, A. Barkai, and H. Rong, Opt. Lett. 36, 2101 (2011). [CrossRef]

]. It was also reported that an apodized GC can be used to dramatically reduce the backreflection [6

6. M. Antelius, K. B. Gylfason, and H. Sohlström, Opt. Express 19, 3592 (2011). [CrossRef]

,7

7. R. Halir, P. Cheben, S. Janz, D.-X. Xu, Í. Molina-Fernández, and J. G. Wangüemert-Pérez, Opt. Lett. 34, 1408(2009). [CrossRef]

]. In this Letter, we experimentally demonstrate another method that can be used to strongly suppress the backreflection [8

8. Y. Li, D. Vermeulen, Y. De Koninck, G. Yurtsever, G. Roelkens, and R. Baets, in Proceedings of Optical Fiber Communication Conference (OFC) (Optical Society of America, 2012), p. JTh2A.8.

]. The devices are designed for the silicon-on-insulator (SOI) platform, but the concept can be extended to other photonic platforms.

The design is similar to the compact focusing grating couplers (FGCs) (without taper) reported by Van Laere et al. [9

9. F. Van Laere, T. Claes, J. Schrauwen, S. Scheerlinck, W. Bogaerts, D. Taillaert, L. O’Faolain, D. Van Thourhout, and R. Baets, IEEE Photon. Technol. Lett. 19, 1919 (2007). [CrossRef]

]. The trench shapes in these FGCs can be expressed in a polar coordinate system
r(q,α)=qλ0ngncsinϕcosα,
(1)
where r(q,α) represents the trench radius of the grating, α is the corresponding azimuth, qN is the index of each line, λ0 is the central wavelength of light in vacuum, ϕ is the zenith of the coupling direction in free space, and ng and nc are the effective indices of the grating region and the top cladding material, respectively. In this formula, it is assumed that the effective index value of the slab region ns is the same as ng, which is, however, not exact. Considering this effective index difference, a more general equation is used in our design:
r(q,α)=(qq0)·λ0ngncsinϕcosα+p(q0,α),
(2)
where
p(q0,α)=κq0λ0nsncsinϕcosα,
(3)
q0N is the index of the first line in the grating, and κ is a factor to determine the distance L between the first line and the entrance waveguide. In this Letter, κ is chosen as (nsncsinϕ)/(ngncsinϕ) so that r(q0,0) is the same as that from the conventional design. A similar correction was also reported in [10

10. R. Waldha¨usl, B. Schnabel, P. Dannberg, E.-B. Kley, A. Bräuer, and W. Karthe, Appl. Opt. 36, 9383 (1997). [CrossRef]

].

These curves form an array of ellipses with their first focus points f1,q at the same position. According to Eq. (2), if the end of the entrance waveguide is located on f1,q, light coming from this waveguide can be coupled out of the chip to the predesigned direction in free space. In previously used FGCs [9

9. F. Van Laere, T. Claes, J. Schrauwen, S. Scheerlinck, W. Bogaerts, D. Taillaert, L. O’Faolain, D. Van Thourhout, and R. Baets, IEEE Photon. Technol. Lett. 19, 1919 (2007). [CrossRef]

] (standard FGCs), the entrance waveguide points to the grating region with α=0° [seen in Fig. 1(a)]. Actually, light can be coupled to almost the same predesigned direction if the entrance waveguide is rotated around f1,q and another part of the grating is used (the actual coupling direction might be slightly different due to the second-order reflection). In this case, the backreflection that is refocused on a location near the second focus points of the ellipses f2,q will not be coupled back in the entrance waveguide [seen in Fig. 1(b)]. As a result, the reflection back into the entrance waveguide is suppressed. The reduced reflection in this design has been demonstrated by means of simulation [11

11. D. Vermeulen, Y. De Koninck, Y. Li, E. Lambert, W. Bogaerts, R. Baets, and G. Roelkens, in 8th International Conference in Group IV Photonics (IEEE, 2011), p. 74 (P1.6).

,12

12. D. Vermeulen, Y. De Koninck, Y. Li, W. Bogaerts, R. Baets, and G. Roelkens, Opt. Express 20, 22278 (2012). [CrossRef]

], and the results show that the backreflection of FGCs with α=45° or 90° are suppressed compared to standard FGCs. Besides, to avoid the influence to the adjacent devices, the backreflection can be collected to another waveguide and then sent to a high-loss region. This design slightly breaks the symmetry of the output mode, but its influence on the grating-to-fiber coupling is not much when ϕ=10°. The new design (titled FGC) also has a very compact size compared to the commonly used one-dimensional (1D) GC.

Fig. 1. Schematic of tilted FGCs (exaggerated) on SOI with different azimuths: (a) α=0 and (b) α0. The black region is etched 70 nm into a 220 nm silicon layer, while the rest is unetched. The light diffracting from the waveguide and the Fresnel reflections is indicated with the solid and dashed blue and red lines, respectively.

Tilted FGCs with different α have been fabricated using deep UV (193 nm) lithography through the ePIXfab multiproject wafer service [13], and they are designed to couple 1550 nm light with ϕ=10°. The width of the entrance waveguide aperture (0.9 μm) was chosen to be neither too small to introduce extra reflection nor too large to collect the redirected Fresnel reflection from the grating. The distance L determines the size of the output mode in the GC and thus influences the maximal grating-to-single-mode-fiber coupling efficiency η. In order to know the optimal value of L, a number of FGCs (α=0°) with different first grating indices q0 were fabricated and measured. The η values of different FGCs and the corresponding central wavelengths λ0 are plotted in Fig. 2. It is found that the best q0 is 26, which means L=16μm. All the following designs use this q0 value. The variation of the λ0 is within 5 nm, and it might be caused by the alignment variations of fibers.

Fig. 2. Maximal coupling efficiency between a single-mode fiber and FGCs with different q0 values and corresponding central wavelengths. The width of the waveguide aperture is 0.9μm.

Designs with two identical tilted FGCs connected by a shallowly etched waveguide were measured to retrieve η for different α. These transmission spectra are plotted in Fig. 3. The η and λ0 values are plotted in Fig. 4. The solid horizontal line in Fig. 4 stands for the η of a standard FGC (4.6dB). It can be found that the λ0 values are lower than 1550 nm. Two main reasons are responsible for this shift: (1) ng and ϕ are wrongly estimated in the design, and the real central wavelength will thus become λ=λ0(ngncsinϕcosα)/(ngncsinϕcosα), where ng and ϕ stand for the real values of ng and ϕ, respectively. It is seen that λ is a function of α, which is also shown in the measurement results. (2) A dip caused by the second-order reflection exists in the transmission spectra of the tilted FGCs with α close to 90°, and this reflection dip causes both a suppression of η and a shift of λ0. Note that the central wavelength of the second-order reflection has a blueshift as α increases. For α150° or 60°, the λ0 of the second-order reflection stays far away from the transmission band and thus has less influence. The second-order reflection also causes a bandwidth increase in the tilted FGCs with α close to 90°.

Fig. 3. Transmission spectra of the tilted FGCs with different azimuths.
Fig. 4. Maximal transmission efficiencies of the tilted FGCs and the corresponding wavelengths. The horizontal line stands for the η of a standard FGC.

It is hard to estimate the reflections using the fringes in the transmission spectrum, because the reflections on the output fiber end facets are also mixed in the spectrum. In order to measure the reflection values of our tilted FGCs, we use another set of samples. In those samples, all the input FGCs use the same design with α=0°, while the output tilted FGCs have a scan of the azimuth from 0° to 180°. The frequency average reflections are measured by a simplified optical frequency domain reflectometry setup [14

14. F. Morichetti, A. Canciamilla, C. Ferrari, M. Torregiani, A. Melloni, and M. Martinelli, Phys. Rev. Lett. 104, 033902 (2010). [CrossRef]

]. First we align the input fiber with the help of an output fiber. After alignment, the output fiber is moved away so that the influence caused by the reflection of the output fiber facet is removed. With the help of a circulator, the reflected light signal including reflections from the input fiber facet and reflections from the reflective components in the waveguide is retrieved. A wavelength sweep from 1490 to 1570 nm with a 20 pm resolution was done to obtain the reflection spectrum. The postprocessing on the reflection spectrum is done in a computer, and a spatial distribution of the reflection is calculated from the autocorrelation R(z) of the reflection spectrum. Three steps are done before calculating R(z): (1) interpolating the reflection spectra from the wavelength domain to the frequency domain, (2) multiplying a Gaussian window to the spectrum in the frequency domain, so as to suppress numerical leakages [15

15. F. J. Harris, Proc. IEEE 66, 51 (1978). [CrossRef]

], and (3) zero padding, in order to get a denser frequency mesh in the autocorrelation function, which is good for estimating the 3 dB bandwidth of peaks. After these steps, a power spectral density P(f) for an array of frequencies f was obtained. R(z) was then derived by calculating the inverse Fourier transform of P(f).

The distances between the input and output tilted FGCs are around 500 μm, but not exactly the same. The reflection autocorrelations R(z) for different azimuths with z near 500 μm are plotted in Fig. 5. Two maximal values can be found for each sample, which means that two different reflections are correlated with the reflection at the output tilted FGC. They are the reflections from the input fiber facet and from the input GC, respectively, and the former one corresponds to a larger z. The reflectance values of the tilted FGCs are obtained by calculating the integral over the 3 dB bandwidth around λ0. The frequency average reflections for different α are plotted in Fig. 6. Assuming that a 4% reflection occurs at the input fiber facet, the frequency average reflection for an FGC with α=0° is around 20dB. GCs with α between 45° and 135° have the frequency average backreflection suppressed to less than 40dB. The best reflection suppression is for α=135°, which is around 55dB. The optimal α is not 90°, and that is because the central wavelength value of the second-order reflection is near the transmission peak in this case. This is also shown in the simulation results [11

11. D. Vermeulen, Y. De Koninck, Y. Li, E. Lambert, W. Bogaerts, R. Baets, and G. Roelkens, in 8th International Conference in Group IV Photonics (IEEE, 2011), p. 74 (P1.6).

,12

12. D. Vermeulen, Y. De Koninck, Y. Li, W. Bogaerts, R. Baets, and G. Roelkens, Opt. Express 20, 22278 (2012). [CrossRef]

].

Fig. 5. Peaks in the autocorrelation R(z) caused by the reflection from the output tilted GCs with different azimuths. The autocorrelations are plotted in a logarithmic scale (20×log(R(z))).
Fig. 6. Maximal frequency average reflections of the tilted FGCs for different azimuths.

For tilted FGCs with the azimuth α in the range between 45° and 60°, the frequency average reflection is highly suppressed (up to 50dB) with no apparent transmission penalty. In this design, the Fresnel reflection is directed away from the entrance waveguide rather than being suppressed. This design can work together with the other GC structures that have reduced Fresnel reflection (e.g., apodized gratings), so as to further suppress the on-chip back reflection of the GC strucutures.

The authors acknowledge the Ghent University-Methusalem project “Smart Photonic Chips” for financial support. The authors thank Honghui Shen and Linghua Wang for useful discussions.

References

1.

D. Taillaert, F. Van Laere, M. Ayre, W. Bogaerts, D. Van Thourhout, P. Bienstman, and R. Baets, Jpn. J. Appl. Phys. 45, 6071 (2006).

2.

D. Vermeulen, S. Selvaraja, P. Verheyen, G. Lepage, W. Bogaerts, P. Absil, D. V. Thourhout, and G. Roelkens, Opt. Express 18, 18278 (2010). [CrossRef]

3.

A. Mekis, S. Gloeckner, G. Masini, A. Narasimha, T. Pinguet, S. Sahni, and P. De Dobbelaere, IEEE J. Sel. Topics. Quantum Electron. 17, 597 (2011). [CrossRef]

4.

C. Alonso-Ramos, A. Ortega-Monũx, I. Molina-Fernández, P. Cheben, L. Zavargo-Peche, and R. Halir, Opt. Express 18, 15189 (2010). [CrossRef]

5.

N. Na, H. Frish, I.-W. Hsieh, O. Harel, R. George, A. Barkai, and H. Rong, Opt. Lett. 36, 2101 (2011). [CrossRef]

6.

M. Antelius, K. B. Gylfason, and H. Sohlström, Opt. Express 19, 3592 (2011). [CrossRef]

7.

R. Halir, P. Cheben, S. Janz, D.-X. Xu, Í. Molina-Fernández, and J. G. Wangüemert-Pérez, Opt. Lett. 34, 1408(2009). [CrossRef]

8.

Y. Li, D. Vermeulen, Y. De Koninck, G. Yurtsever, G. Roelkens, and R. Baets, in Proceedings of Optical Fiber Communication Conference (OFC) (Optical Society of America, 2012), p. JTh2A.8.

9.

F. Van Laere, T. Claes, J. Schrauwen, S. Scheerlinck, W. Bogaerts, D. Taillaert, L. O’Faolain, D. Van Thourhout, and R. Baets, IEEE Photon. Technol. Lett. 19, 1919 (2007). [CrossRef]

10.

R. Waldha¨usl, B. Schnabel, P. Dannberg, E.-B. Kley, A. Bräuer, and W. Karthe, Appl. Opt. 36, 9383 (1997). [CrossRef]

11.

D. Vermeulen, Y. De Koninck, Y. Li, E. Lambert, W. Bogaerts, R. Baets, and G. Roelkens, in 8th International Conference in Group IV Photonics (IEEE, 2011), p. 74 (P1.6).

12.

D. Vermeulen, Y. De Koninck, Y. Li, W. Bogaerts, R. Baets, and G. Roelkens, Opt. Express 20, 22278 (2012). [CrossRef]

13.

http://www.epixfab.eu.

14.

F. Morichetti, A. Canciamilla, C. Ferrari, M. Torregiani, A. Melloni, and M. Martinelli, Phys. Rev. Lett. 104, 033902 (2010). [CrossRef]

15.

F. J. Harris, Proc. IEEE 66, 51 (1978). [CrossRef]

OCIS Codes
(050.0050) Diffraction and gratings : Diffraction and gratings
(130.0130) Integrated optics : Integrated optics

ToC Category:
Diffraction and Gratings

History
Original Manuscript: August 6, 2012
Revised Manuscript: September 13, 2012
Manuscript Accepted: September 13, 2012
Published: October 16, 2012

Citation
Yanlu Li, Diedrik Vermeulen, Yannick De Koninck, Gunay Yurtsever, Günther Roelkens, and Roel Baets, "Compact grating couplers on silicon-on-insulator with reduced backreflection," Opt. Lett. 37, 4356-4358 (2012)
http://www.opticsinfobase.org/ol/abstract.cfm?URI=ol-37-21-4356


Sort:  Author  |  Year  |  Journal  |  Reset  

References

  1. D. Taillaert, F. Van Laere, M. Ayre, W. Bogaerts, D. Van Thourhout, P. Bienstman, and R. Baets, Jpn. J. Appl. Phys. 45, 6071 (2006).
  2. D. Vermeulen, S. Selvaraja, P. Verheyen, G. Lepage, W. Bogaerts, P. Absil, D. V. Thourhout, and G. Roelkens, Opt. Express 18, 18278 (2010). [CrossRef]
  3. A. Mekis, S. Gloeckner, G. Masini, A. Narasimha, T. Pinguet, S. Sahni, and P. De Dobbelaere, IEEE J. Sel. Topics. Quantum Electron. 17, 597 (2011). [CrossRef]
  4. C. Alonso-Ramos, A. Ortega-Monũx, I. Molina-Fernández, P. Cheben, L. Zavargo-Peche, and R. Halir, Opt. Express 18, 15189 (2010). [CrossRef]
  5. N. Na, H. Frish, I.-W. Hsieh, O. Harel, R. George, A. Barkai, and H. Rong, Opt. Lett. 36, 2101 (2011). [CrossRef]
  6. M. Antelius, K. B. Gylfason, and H. Sohlström, Opt. Express 19, 3592 (2011). [CrossRef]
  7. R. Halir, P. Cheben, S. Janz, D.-X. Xu, Í. Molina-Fernández, and J. G. Wangüemert-Pérez, Opt. Lett. 34, 1408(2009). [CrossRef]
  8. Y. Li, D. Vermeulen, Y. De Koninck, G. Yurtsever, G. Roelkens, and R. Baets, in Proceedings of Optical Fiber Communication Conference (OFC) (Optical Society of America, 2012), p. JTh2A.8.
  9. F. Van Laere, T. Claes, J. Schrauwen, S. Scheerlinck, W. Bogaerts, D. Taillaert, L. O’Faolain, D. Van Thourhout, and R. Baets, IEEE Photon. Technol. Lett. 19, 1919 (2007). [CrossRef]
  10. R. Waldha¨usl, B. Schnabel, P. Dannberg, E.-B. Kley, A. Bräuer, and W. Karthe, Appl. Opt. 36, 9383 (1997). [CrossRef]
  11. D. Vermeulen, Y. De Koninck, Y. Li, E. Lambert, W. Bogaerts, R. Baets, and G. Roelkens, in 8th International Conference in Group IV Photonics (IEEE, 2011), p. 74 (P1.6).
  12. D. Vermeulen, Y. De Koninck, Y. Li, W. Bogaerts, R. Baets, and G. Roelkens, Opt. Express 20, 22278 (2012). [CrossRef]
  13. http://www.epixfab.eu .
  14. F. Morichetti, A. Canciamilla, C. Ferrari, M. Torregiani, A. Melloni, and M. Martinelli, Phys. Rev. Lett. 104, 033902 (2010). [CrossRef]
  15. F. J. Harris, Proc. IEEE 66, 51 (1978). [CrossRef]

Cited By

Alert me when this paper is cited

OSA is able to provide readers links to articles that cite this paper by participating in CrossRef's Cited-By Linking service. CrossRef includes content from more than 3000 publishers and societies. In addition to listing OSA journal articles that cite this paper, citing articles from other participating publishers will also be listed.


Next Article »

OSA is a member of CrossRef.

CrossCheck Deposited