Coupling efficiency of monolithic, waveguide-integrated Si
photodetectors
The MIT Faculty has made this article openly available. Please share
how this access benefits you. Your story matters.
Citation
Ahn, Donghwan, Ching-Yin Hong, Lionel C. Kimerling, and
Jurgen Michel. “Coupling efficiency of monolithic, waveguideintegrated Si photodetectors.” Applied Physics Letters 94, no. 8
(2009): 081108. © 2009 American Institute of Physics.
As Published
http://dx.doi.org/10.1063/1.3089359
Publisher
American Institute of Physics
Version
Final published version
Accessed
Thu May 26 22:46:44 EDT 2016
Citable Link
http://hdl.handle.net/1721.1/80264
Terms of Use
Article is made available in accordance with the publisher's policy
and may be subject to US copyright law. Please refer to the
publisher's site for terms of use.
Detailed Terms
Coupling efficiency of monolithic, waveguide-integrated Si photodetectors
Donghwan Ahn, Ching-yin Hong, Lionel C. Kimerling, and Jurgen Michel
Citation: Appl. Phys. Lett. 94, 081108 (2009); doi: 10.1063/1.3089359
View online: http://dx.doi.org/10.1063/1.3089359
View Table of Contents: http://apl.aip.org/resource/1/APPLAB/v94/i8
Published by the AIP Publishing LLC.
Additional information on Appl. Phys. Lett.
Journal Homepage: http://apl.aip.org/
Journal Information: http://apl.aip.org/about/about_the_journal
Top downloads: http://apl.aip.org/features/most_downloaded
Information for Authors: http://apl.aip.org/authors
Downloaded 24 Jul 2013 to 18.51.3.76. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions
APPLIED PHYSICS LETTERS 94, 081108 共2009兲
Coupling efficiency of monolithic, waveguide-integrated Si photodetectors
Donghwan Ahn,a兲 Ching-yin Hong, Lionel C. Kimerling, and Jurgen Michelb兲
Department of Materials Science and Engineering, Massachusetts Institute of Technology,
Cambridge, Massachusetts 02139, USA
共Received 16 October 2008; accepted 26 January 2009; published online 24 February 2009兲
A waveguide-integrated photodetector provides a small-footprint, low-capacitance design that
overcomes the bandwidth-efficiency trade-off problem of free space optics. High performance
silicon devices are critical to the emergence of electronic-photonic integrated circuits on the
complementary metal oxide semiconductor platform. We have fabricated vertical p-i-n silicon
photodetectors that are monolithically integrated with compact silicon oxynitride channel
waveguides. We report over 90% coupling efficiency of 830 nm light from the silicon oxynitride
共SiOxNy兲 channel waveguide to the silicon photodetector. We analyze the dependence of coupling
on waveguide index by comparing coupling from low index-contrast waveguides and high
index-contrast waveguides. © 2009 American Institute of Physics. 关DOI: 10.1063/1.3089359兴
Monolithic integration of complementary metal oxide
semiconductor compatible waveguides 共WGs兲 and photodetectors 共PDs兲 is one of the most important technologies in the
emerging Si-based integrated photonics.1 A bottom-WGcoupled structure has been commonly adopted in most III-V
compound semiconductor-based WG-integrated PDs2,3 and
more recent Ge PDs integrated with Si WGs,4,5 as the WG
and PD materials are sequentially grown by a heteroepitaxial
process. On the other hand, a top-WG-coupled structure
may fit better into the intrachip optical interconnection
architectures6 because it retains the general Si chip structure,
where active devices are fabricated on the substrate and the
passive interconnections are built in the upper levels.
In addition to a recent report of top-WG-integrated Ge
PD,7 there have been a few previous reports of Si PDs8–11
integrated with upper dielectric WGs. However, most of
them exhibited slow evanescent wave coupling rates and
amorphous10 or polycrystalline11 silicon detectors, which
also suffered from low quantum efficiency and high bias
voltage requirements. In this paper, we present crystalline
p-i-n silicon PDs integrated with compact SiON channel
WGs, which demonstrated high coupling efficiency and high
responsivity with relatively small device size due to high
evanescent coupling rates.
A top-WG-integrated structure provides great flexibility
in the materials choices and device designs. Therefore, it is
critical to understand the trends in coupling behavior and
device performance dependency on design parameters,
which were not addressed in previous studies8–11 as they reported on specific material and design choices. In our current
work, we report the dependence of coupling efficiency on
WG design and the resulting impact on PD performance. The
results can provide design guidance for a broader range of
evanescent coupling structures with various materials in silicon microphotonics.7
For device fabrication, Si p+ wafers 共0.01– 0.02 ⍀ cm兲
with a 5 ␮m thick epitaxial Si layer 共16– 24 ⍀ cm兲 were
used. After the epitaxial layer was etched down to 0.5 ␮m
thickness, phosphorus was implanted with a dose of 3
⫻ 1015 / cm2 at 100 keV, followed by an activation annealing
at 1000 ° C for 30 s. Silicon vertical p-i-n diodes were patterned and a 2.2 ␮m depth was plasma-etched. This was
done in order to ensure optical isolation of the WGs from the
Si substrate. Then, 3.5 ␮m SiO2 was deposited over the Si
PD mesas and then planarized and polished down by chemical mechanical polishing, leaving a thin oxide above the Si
mesas. Patterning and opening of oxide windows followed in
order to expose the top surface of the silicon PDs. Next, the
single-mode WGs were formed by plasma enhanced chemical vapor deposition of SiOxNy followed by etching to define
the WGs. Both relatively low index-contrast WGs 共1.2
⫻ 1.2 ␮m2 with ncore = 1.52兲 and higher index-contrast,
smaller WGs 关1.2 ␮m共width, W兲 ⫻ 0.4 ␮m共height, H兲 with
ncore = 1.58 and 0.9 ␮m共W兲 ⫻ 0.3 ␮m共H兲 with ncore = 1.67兴
were prepared. The dimensions of the WGs were chosen for
single-mode operation and to stay within the resolution limit
of the available lithography tool. The index contrast is defined by ⌬n = ncore − nclad. A 2.2 ␮m thick SiO2 cladding
layer 共nclad = 1.45兲 was deposited, followed by opening contact holes at least 3 ␮m away from the WG in order to
prevent the metal from interacting with the optical mode.
After depositing and patterning Al contact pads, the wafers
were annealed in N2 / H2 forming gas. For testing, while coupling 830 nm transverse-electric light to the cleaved WG
input facets, we simultaneously measured the photocurrent
Iph at ⫺2V reverse bias and the remaining optical power
from the output facet of the through WGs Pout. The device
structures are shown in Fig. 1.
Figure 2共a兲 shows the transmitted optical power
Pout共WG-PD兲, normalized to the power at the reference WG
Pout共ref-WG兲, as a function of detection length L. As the data
show the dependence of exponential decay on the coupling
length L, we can fit the data with the following equation:
Pout共WG-PD兲共L兲
Pout共ref-WG兲
a兲
Present address: Department of Electrical and Computer Engineering, The
University of Texas at Austin, Austin, Texas.
b兲
Author to whom correspondence should be addressed. Electronic mail:
jmichel@mit.edu.
0003-6951/2009/94共8兲/081108/3/$25.00
=
Pin␩in exp共− ␣couplingL兲␩out10−␣WGl⬘/10
Pin10−␣WGl/10
⬇ ␩2 exp共− ␣couplingL兲,
共1兲
where Pin is the optical power at the point of entering the
94, 081108-1
© 2009 American Institute of Physics
Downloaded 24 Jul 2013 to 18.51.3.76. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions
081108-2
Appl. Phys. Lett. 94, 081108 共2009兲
Ahn et al.
FIG. 1. 共a兲 Schematic structure of a top-WG-coupled Si p-i-n PD device.
共b兲 Cross-section of the WG-coupled PD device. Some notations used in the
equations are shown.
PD. Because the optical mode in the WG on top of the PD is
different from the mode in the bus WG, we introduced mode
coupling efficiencies ␩in and ␩out that describe the coupling
from the bus WG mode to the mode in the WG on top of the
PD and vice versa. In approximation, it was assumed that ␩in
and ␩out are equal and that the term 10−␣WG共l⬘−l兲/10 is negligible, as the propagation loss in the WG ␣WG is small.
Fitting the measurement data to Eq. 共1兲 suggests that
incoming photons from the input bus WG are coupled to a
mode in the WG on the PD with ␩in ⬃ 86.3% mode-matching
efficiency and the power in the WG on the PD decreases at a
rate of 1 / ␣coupling = 52 ␮m due to the evanescent wave coupling to the Si PD. We can also conclude that the total coupling efficiency of photons from the input bus WG to silicon
FIG. 3. Measured and fitted data of 共a兲 normalized optical power output
through the WGs coupled with PDs and 共b兲 PD quantum efficiency vs detector coupling length. The samples are coupled with relatively high indexcontrast WGs 共1.2⫻ 0.4 ␮m2 with ncore = 1.58 and 0.9⫻ 0.3 ␮m2 with
ncore = 1.67兲.
PD is well above 90% because nearly all of the photons that
are coupled to the WG mode on the PD 共⬃Pin ⫻ ␩in兲 will be
coupled to the Si PD via evanescent wave coupling in a
sufficiently long PD 关e.g., L ⬎ 3 ⫻ 共1 / ␣coupling兲兴, and some
portion of the remaining 14% photons 共⬃Pin ⫻ 共1-␩in兲兲 also
can contribute to the photocurrent by being directly coupled
to the Si detector.
The measured quantum efficiency as a function of detector length L is shown in Fig. 2共b兲 and can be described as
follows:
QE共L兲 =
h␯
h␯
⫻ Iph共L兲 =
⫻ Iph共L兲
qPin
qPout共ref-WG兲10−␣WGl/10
⬇ A + C关1 − exp共− ␣detL兲兴.
FIG. 2. Measured and fitted data of 共a兲 normalized optical power output
through the WGs coupled with PDs and 共b兲 PD quantum efficiency vs detector coupling length. The sample is coupled with a relatively low indexcontrast WG 共1.2⫻ 1.2 ␮m2 with ncore = 1.52兲.
共2兲
The saturated quantum efficiency, represented by the term
共A + C兲 in Fig. 2共b兲, was about 46%, slightly higher than the
efficiency of Si PD when measured with surface-normal illumination 共41%兲.
The results of coupling from high index-contrast WGs,
shown in Fig. 3, reveal two significant differences compared
to low index-contrast WGs. First, the evanescent coupling to
the detector occurs much faster than for low index-contrast
WGs. The greater index-contrast between core and cladding
of a WG 共ncore − nclad兲 leads to smaller refractive index difference between the PD and WG core 共nPD − ncore兲. In addition to the smaller index difference between WG and detector, the geometrical factors of these WGs, such as smaller
WG dimensions to maintain the single-mode condition and
reduced height, also affect the impedance of the WG layer
Downloaded 24 Jul 2013 to 18.51.3.76. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions
081108-3
Appl. Phys. Lett. 94, 081108 共2009兲
Ahn et al.
FIG. 4. 共Color online兲 Cross-section view of the mode profile immediately
共at 0.5 ␮m兲 after coupling region to the PD begins. 共a兲 In the coupling
structure with smaller high index-contrast WG with flat rectangular shape
共0.9⫻ 0.3 ␮m2, ⌬n = 0.22兲 共b兲 In the coupling structure with lower indexcontrast WG with larger square dimensions 共1.2⫻ 1.2 ␮m2, ⌬n = 0.07兲.
2
2
through an effective index change 共i.e., Z ⬀ 1 / 冑ncore
− neff
兲
and thus lower the impedance barrier between WG and PD.
As a result, a leaky mode has more evanescent field penetrating into silicon and the evanescent coupling rate ␣coupling is
higher, as demonstrated by much shorter 1 / ␣coupling and
quick rise in the quantum efficiency within short coupling
lengths 共e.g., ⬍40 ␮m兲 as shown in Fig. 3.
The second significant difference between low versus
high index-contrast WGs is the observed direct coupling
from the bus WG to modes inside the Si PD. Figure 4共a兲
shows the simulation results for such a scenario. The simulation clearly shows that the light is coupled to modes in the
Si PD. Simulations for the low index-contrast case 关Fig.
4共b兲兴 show that the light remains in the WG and generally
does not couple directly into the Si PD. The direct coupling
into the Si PD explains the relatively low ␩ in Fig. 3共a兲,
indicating that only a small fraction of the light from the bus
WG remains in the WG on top of the detector.
The lower quantum efficiencies for the low-indexcontrast WGs are mainly due to the fact that during evanescent coupling, photons enter the PD nearly perpendicular to
the WG-detector interface. Due to the long absorption length
of 13 ␮m for 830 nm light in Si, only 41% of the photons
can be absorbed in the top 7 ␮m thick layer 共given the vertical p-i-n geometry of our silicon PD, only the photons absorbed within a certain depth of the PD—top 7 ␮m in our
devices—can contribute to the photocurrent兲. Therefore, the
quantum efficiency of the PD coupled with low index-
contrast WG is close to the efficiency of the PD when measured with surface-normal illumination. In contrast, for the
high index-contrast WG cases, the modes, excited due to
direct coupling to the Si PD, travel in the horizontal direction
and therefore are not limited by absorption length of Si. As
the absorption length limitation is lifted, the quantum efficiency increases significantly.
In summary, we have demonstrated WG-integrated silicon PDs that employ a top-to-bottom evanescent-wave coupling structure, achieving over 90% coupling efficiency. PD
devices integrated with single-mode, low index-contrast
WGs exhibit relatively slow evanescent coupling rates and a
slightly higher quantum efficiency than that of surfaceilluminated device. In the case of single-mode, high indexcontrast WGs, coupling to PDs occurred mainly by mode
matching of the input WG mode and modes in the Si PD,
leading to an enhanced 80% quantum efficiency. The observed trend can generally be extended to a wider variety of
top WG-coupled vertical p-i-n PDs, as was recently demonstrated by an integrated Ge PD coupled with silicon nitride
WGs.7
1
L. C. Kimerling, L. Dal Negro, S. Saini, Y. Yi, D. Ahn, S. Akiyama, D.
Cannon, J. Liu, J. G. Sandland, D. Sparacin, J. Michel, K. Wada, and M.
R. Watts, in Silicon Photonics: Topics in Applied Physics, edited by L.
Pavesi and D. J. Lockwood 共Springer, Berlin, 2004兲, Vol. 94, p. 89.
2
R. J. Deri, W. Doldissen, R. J. Hawkins, R. Bhat, J. B. D. Soole, L. M.
Schiavone, M. Seto, N. Andreadakis, Y. Silberberg, and M. A. Koza,
Appl. Phys. Lett. 58, 2749 共1991兲.
3
Y.-S. Wu, J.-W. Shi, J.-Y. Wu, F.-H. Huang, Y.-J. Chan, Y.-L. Huang, and
R. Xuan, IEEE Photonics Technol. Lett. 17, 878 共2005兲.
4
J. F. Liu, D. Ahn, C. Y. Hong, D. Pan, S. Jongthammanurak, M. Beals, L.
C. Kimerling, J. Michel, A. T. Pomerene, C. Hill, M. Jaso, K. Y. Tu, Y. K.
Chen, S. Patel, M. Rasras, A. White, and D. M. Gill, Proceedings of the
Optics Valley of China International Symposium on Optoelectronics, Wuhan 共IEEE, New York, 2006兲, p. 1.
5
T. Yin, R. Cohen, M. M. Morse, G. Sarid, Y. Chetrit, D. Rubin, and M. J.
Paniccia, Opt. Express 15, 13965 共2007兲.
6
A. V. Mule, E. N. Glytsis, T. K. Gaylord, and J. D. Meindl, IEEE Trans.
Very Large Scale Integr. 共VLSI兲 Syst. 10, 582 共2002兲.
7
D. Ahn, C.-Y. Hong, J. Liu, W. Giziewicz, M. Beals, L. C. Kimerling, and
J. Michel, Opt. Express 15, 3916 共2007兲.
8
S. Wunderlich, J. P. Schmidt, and J. Muller, Appl. Opt. 31, 4186 共1992兲.
9
U. Hilleringmann and K. Goser, IEEE Trans. Electron Devices 42, 841
共1995兲.
10
K. Kapser and P. P. Deimel, J. Appl. Phys. 71, 3614 共1992兲.
11
G. Yuan, R. Pownall, P. Nikkel, C. Thangaraj, T. W. Chen, and K. L. Lear,
IEEE Photonics Technol. Lett. 18, 1657 共2006兲.
Downloaded 24 Jul 2013 to 18.51.3.76. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions