Discussion 5 - Grating Coupler Theory

Discussion today:
Grating couplers
• Today we will discuss how we can couple light between optical

fibers and on-chip photonic waveguides. In particular we will
focus on grating couplers.
• Today we will cover the theory of grating couplers. Next week
we will put the theory to use and design a silicon photonic
grating coupler.
1
EE232 Discussion 2/16/2017
Coupling into and out of waveguides
• Dilemma: how do we couple light to and from an optical fiber

to an integrated on-chip photonic waveguide?
300nm
10 Optical fiber
on-chip waveguide
2
End-fire coupling
• Pros: Coupling efficiency can be very good
• Cons: Poor tolerance to misalignment. Costly to manufacture.
Optical fiber lens photonic waveguide

(not to scale)
Side view
3
Nanotaper coupler
• Pros: Coupling efficiency can be very good. Lens not needed
• Cons: Poor tolerance to misalignment. Requires advanced
lithography equipment
Taper to ~100nm  spatial extent of mode becomes large
Optical fiber
(not to scale)
Single mode silicon waveguide
Top view
4
Top coupling
b er )
• How can we couple light from l fi ale
a confined waveguide mode to ti ca sc
o
Op ot t
an optical fiber placed above (n
the waveguide?
x 𝑛1
𝑛2 𝑘𝑚𝑧
z
𝑛3
5
Grating coupler
• By periodically notching the waveguide
we can make a diffraction grating b er )
l fi ale
such that light that is diffracted off ti ca sc
o
Op ot t
the rulings will constructively (n
interfere toward a direction
into the optical fiber
x 𝑛1
𝑛2 𝑘𝑚𝑧
𝑎
z
𝑛3
6
Grating coupler
• We choose the diffraction period ()
such that scattering from individual b er )
l fi ale
rulings constructively interferes ti ca sc
o
Op ot t
at the desired angle (n
Scattering from
individual ruling
x 𝑛1
𝑛2 𝑘𝑚𝑧
𝑎
z
𝑛3
7
Grating coupler
• We can write a relationship between
the diffraction angle () and the
periodicity () of the grating
𝜃
x 𝑛1
𝑛2 𝑘𝑚𝑧
𝑎
z
𝑛3
8
• Let’s use ray optics to determine the condition for two rays
reflected off adjacent rulings to constructively interfere at an
angle .
Ray 1
• We choose grating periodicity ()
such that the path length difference
() results in constructive Ray 2
interference between 𝑏
Ray 1 and Ray 2. 𝜃
• This will happen when 𝑎
phase difference between
𝑛1 𝑛2
Ray 1 and Ray 2 are
𝑘𝑚𝑧
integer multiples of 2π.
9
Phase of Ray 2 Phase of Ray 1
𝑘𝑚𝑧 𝑎 − 𝑘1 𝑏=2 𝜋 𝑚 𝑚=0 , ± 1, ± 2 … Ray 1
2𝜋 2𝜋
𝑛𝑒𝑓𝑓 𝑎− 𝑛 𝑏=2 𝜋 𝑚
𝜆 𝜆 1 Ray 2
𝑛𝑒𝑓𝑓 𝑎−𝑛 1 𝑏=𝑚 𝜆 𝑏
𝜃
𝑛𝑒𝑓𝑓 𝑎− n1 𝑎sin ⁡(𝜃)=𝑚 𝜆 𝑎
𝑛1 𝑛2
𝑚 𝜆 −𝑛 𝑒𝑓𝑓 𝑎
( )
sin 𝜃 =
−𝑛 1 𝑎
𝑘𝑚𝑧
𝑚𝜆
𝑛𝑒𝑓𝑓 −
𝑎
sin ( 𝜃 )=
𝑛1
10
Coupling into waveguide
• By reciprocity we can use the same
grating to couple light into a b er )
l fi ale
waveguide. ti ca sc
o
Op ot t
• But, let’s examine this problem (n
slightly differently.
x 𝑛1
𝑛2 𝑘𝑚𝑧
𝑎
z
𝑛3
11
Waveguide coupling
• What if I send light onto a waveguide (without a grating) as
shown below?
• How much power can I couple into the waveguide?
x
𝜃
𝑛1
z
𝑛2 𝑘𝑚𝑧
𝑛3
12
Waveguide coupling
• The answer is precisely zero
x
𝜃
𝑛1
z
𝑛2 𝑘𝑚𝑧
𝑛3
13
Waveguide coupling
• Let’s do the math:
(incident)
(reflected)
(transmitted)
• At the air-waveguide interface (x = 0), the tangential component

of electric field must be continuous:
− 𝑗 𝑘1𝑧 𝑧 − 𝑗 𝑘1 𝑧 𝑧 − 𝑗 𝑘𝑚 𝑧 𝑧
𝐸0 𝑒 +𝐸𝑅𝑒 = 𝐸𝑚 𝑒
• This implies a phase matching condition:
𝑘1 𝑧 =𝑘𝑚𝑧
14
Waveguide coupling
• We can write as:
𝑘1 𝑧 =𝑘0 𝑛1 sin 𝜃 𝑚
• Therefore:
𝑘0 𝑛 1 sin 𝜃𝑚 =𝑘𝑚𝑧
• But, for a guided mode we know that that the effective index must
be greater than and as a result
This requires which is of course impossible!
• We cannot couple light into a guided mode without dealing with
this phase match condition.
15
Waveguide coupling
• But, for a guided mode we know that that the effective index must
be greater than and as a result
This requires which is of course impossible!
• We cannot couple light into a guided mode without dealing with
this phase match condition.
16
Waveguide coupling
• Another way of seeing this is to recall that a guided mode must

decay exponentially away from the core. In other words, is
imaginary in the cladding. (If is real, it is radiating!)
• But incident mode must have real components, so cannot couple to
guided mode.
• Can we create an evanescent mode from incident mode?
17
Prism couplers
• Can use prism placed close to
waveguide to satisfy phase
matching condition
𝑘𝑧 ,𝑝𝑟𝑖𝑠𝑚 =𝑘0 𝑛𝑝 sin ( 𝜃𝑝 )=𝑘1 𝑧
• Evanescent mode in region

couples to guided mode
• Same mechanism used in

beamsplitters
• Used mainly for refractive index
measurement of thin films (eg.
metricon in Nanolab)
18
Grating coupler
• What if we have the following geometry?
x
𝜃
𝑛1
z
𝑛2 𝑘𝑚𝑧
a
𝑛3
19
Grating coupler
• At the boundary of
and we can write
• is a periodic function describing the height variation of our

grating. Thus we can write using a Fourier series
𝑚= ∞ − 𝑗𝑚 ( 2a𝜋 )𝑧
e
− 𝑗 𝑘1 𝑥 h ( 𝑧 )
= ∑ 𝐴𝑚 𝑒
𝑚 =−∞
20
Grating coupler
• At the boundary of
and we can write
• Now our new phase matching requirement is:

𝑘1 𝑧 +2 𝜋 𝑚 /𝑎=𝑘1 𝑧 +2 𝜋 𝑛/𝑎=𝑘𝑚𝑧
for some .
• Clearly, , therefore:
𝑘0 𝑛 1 sin 𝜃+2 𝜋 𝑚/𝑎=𝑘𝑚𝑧
21
Grating coupler
• This is the same phase
matching requirement that we
derived under the ray optics
model.
𝑘1 sin 𝜃+2 𝜋 𝑚/𝑎=𝑘𝑚𝑧
• By properly choosing and we can now phase match!
22
Grating examples
23
http://www.oxford-instruments.com/businesses/nanotechnology/plasma-
technology/campaigns/sem-competition-winner-%281%29
24
Design example
• Design the grating period such that light is diffracted out at a
10° angle for 1st order diffraction
• Assume
10°
𝜆 air
𝑛 eff −
𝑎
sin ( 𝜃 )= silicon 𝑎
𝑛 air
2.5 −
1.55 Si O2
𝑎
sin ( 10 ° )=
1
𝑎 670 nm
26
Comments
• Today we have shown how to calculate the diffraction angle of
a grating coupler but we have not said anything about the
intensity of the light that will diffract at that angle or in other
words we do not yet know the coupling efficiency.
• This requires more rigorous theory or simulation tools.
• Next class: Numerical design of a grating coupler using
Lumerical FDTD
27

Discussion 5 - Grating Coupler Theory

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Discussion 5 - Grating Coupler Theory

Uploaded by

Copyright:

Available Formats

Discussion today:

• Today we will discuss how we can couple light between optical

• Dilemma: how do we couple light to and from an optical fiber

Optical fiber lens photonic waveguide

Taper to ~100nm  spatial extent of mode becomes large

Single mode silicon waveguide

𝑘𝑚𝑧 𝑎 − 𝑘1 𝑏=2 𝜋 𝑚 𝑚=0 , ± 1, ± 2 … Ray 1

• At the air-waveguide interface (x = 0), the tangential component

• Another way of seeing this is to recall that a guided mode must

• Evanescent mode in region

• Same mechanism used in

• is a periodic function describing the height variation of our

• Now our new phase matching requirement is:

• By properly choosing and we can now phase match!

You might also like