Nonlinear effects in Silicon Waveguides - The Institute of Optics ...

1/56
Nonlinear effects in Silicon Waveguides
Govind P. Agrawal
Institute of Optics
University of Rochester
Rochester, NY 14627
c○2007 G. P. Agrawal
◭◭
◮◮
◭
◮
Back
Close

Outline
• Why use silicon for photonics?
• Silicon-on-Insulator (SOI) Waveguides
2/56
• Two-Photon Absorption and Free-Carrier Effects
• Self-Phase Modulation and Soliton Formation
• Higher-Order Solitons and Supercontinuum Generation
• Cross-Phase Modulation and Optical Switching
• Four-Wave Mixing and Parametric Amplification
• Raman Amplification and Silicon Raman Lasers
• Concluding Remarks
◭◭
◮◮
◭
◮
Back
Close

Silicon Photonics
• Silicon dominates microelectronics industry totally.
• Silicon photonics is a new research area trying to capitalize on the
huge investment by the microelectronics industry.
3/56
• It has the potential for providing a monolithically integrated
optoelectronic platform on a silicon chip.
Credit:
Intel and IBM Websites
◭◭
◮◮
◭
◮
Back
Close

Silicon-on-Insulator Technology
4/56
• Core waveguide layer is made of silicon (n 1 = 3.45).
• A silica layer under the core layer is used for lower cladding.
• Silica layer formed by implanting oxygen, followed with annealing.
• A rib or ridge structure used for two-dimensional confinement.
• Air on top confines mode tightly (large index difference).
◭◭
◮◮
◭
◮
Back
Close

SOI Waveguide Geometry
[0 0 1]
5/56
[1 1 0]
• SEM image of an SOI device;
Izhaky et al., JSTQE, (2006).
• Typically W ∼ 1 µm,
H ∼ 1 µm, h ∼ H/2.
• Dispersion normal at telecom
wavelengths (λ 0 ∼ 2.2 µm).
◭◭
◮◮
◭
◮
Back
Close

Waveguide Dispersion
• Mode index ¯n(ω) = n 1 (ω) − δn W (ω).
• Material dispersion results from n 1 (ω) (index of silicon).
6/56
• Waveguide dispersion results from δn W (ω).
• Total dispersion D = D M + D W can be controlled.
◭◭
◮◮
◭
◮
Back
Close

Major Nonlinear Effects
• Self-Phase Modulation (SPM)
• Cross-Phase Modulation (XPM)
7/56
• Four-Wave Mixing (FWM)
• Stimulated Brillouin Scattering (SBS)
• Stimulated Raman Scattering (SRS)
Origin of Nonlinear Effects in Silicon
• Third-order nonlinear susceptibility χ (3) .
• Real part leads to SPM, XPM, and FWM.
• Imaginary part leads to two-photon absorption (TPA).
◭◭
◮◮
◭
◮
Back
Close

Third-order Nonlinear Response
˜P (3)
i (ω i ) = 3ε ∫∫
0
16π 2
χ (3)
i jkl (−ω i;ω j ,−ω k ,ω l )Ẽ j (ω j )Ẽ ∗ k (ω k )Ẽ l (ω l )dω j dω k
• Two distinct contributions from electronic and Raman responses:
χ (3)
i jkl = χe i jkl + χ R i jkl.
• Electronic response for silicon is governed by
χ e i jkl = χ e 1122δ i j δ kl + χ e 1212δ ik δ jl + χ e 1221δ il δ jk + χ e dδ i jkl .
• Raman response for silicon is governed by
χ R i jkl = χ R ˜H R (ω l − ω k )(δ ik δ jl + δ il δ jk − 2δ i jkl ) + a similar term.
• A relatively narrow Lorentzian Raman-gain spectrum:
˜H R (Ω) =
Ω 2 R
Ω 2 R − Ω2 − 2iΓ R Ω
(
ΩR
2π = 15.6 THz, Γ R
π = 105 GHz )
.
8/56
◭◭
◮◮
◭
◮
Back
Close

Third-order Nonlinear Susceptibility
• The tensorial nature of χ (3) makes theory quite complicated.
• It can be simplified considerably when a single optical beam excites
the fundamental TE or TM mode of the Si waveguide.
9/56
• Only the component χ1111 e (−ω;ω,−ω,ω) is relevant in this case.
• Its real and imaginary part the Kerr coefficient n 2 and the TPA
coefficient β T as
ω
c n 2(ω) + i 2 β TPA(ω) =
3ω
4ε 0 c 2 n 2 χ1111(−ω;ω,−ω,ω).
e
0
• A recent review provides more details:
Q. Lin, O. J. Painter, G. P. Agrawal, Opt. Express 15, 16604 (2007).
◭◭
◮◮
◭
◮
Back
Close

Measured Nonlinear Parameters
4
3
0.7
0.6
0.5
10/56
2
0.4
0.3
1
0.2
0.1
0
1200 1400 1600 1800 2000 2200 2400
0.0
1200 1400 1600 1800 2000 2200 2400
Lin et al, Appl. Phys. Lett. 91, 021111 (2007)
• We measured n 2 and β TPA for silicon in the spectral range of 1.2 to
2.4 µm using the z scan technique.
• n 2 larger for silicon by >100 compared with silica fibers.
• As expected, β TPA vanishes for λ > 2.2 µm
◭◭
◮◮
◭
◮
Back
Close

Nonlinear Parameters
• Refractive index depends on intensity as (Kerr effect):
n(ω,I) = ¯n(ω) + n 2 (1 + ir)I(t).
11/56
• Material parameter n 2 = 3 × 10 −18 m 2 /W is larger for silicon by a
factor of 100 compared with silica fibers.
• Dimensionless parameter r = β TPA /(2k 0 n 2 ) is related to two-photon
absorption (TPA) occurring when hν exceeds E g /2.
• TPA parameter: β TPA = 5×10 −12 m/W at wavelengths near 1550 nm.
• Dimensionless parameter r ≈ 0.1 for silicon near 1550 nm.
• TPA is a major limiting factor for SOI waveguides because it creates
free carriers (in addition to nonlinear losses).
◭◭
◮◮
◭
◮
Back
Close

Free-Carrier Generation
• TPA creates free carriers inside a silicon waveguide according to
∂N c
∂t
= β TPA
2hν 0
I 2 (z,t) − N c
τ c
.
12/56
• Carrier lifetime is relatively large for silicon (τ c > 10 ns).
• It limits the device response time if carriers cannot be removed
quickly enough.
• Free carriers also introduce loss and change the refractive index.
• Pulse propagation inside silicon waveguides is governed by
∂A
∂z + iβ 2 ∂ 2 A
2 ∂t = ik 0n 2 2 (1 + ir)|A| 2 A − σ 2 (1 + iµ)N cA − α l
2 A.
◭◭
◮◮
◭
◮
Back
Close

Free-Carrier Absorption (FCA)
• Loss induced by FCA: α f = σN c with σ = 1.45 × 10 −21 m 2 .
13/56
• Free carriers also change the refractive index by ∆n = −(µ/2k 0 )σN c
(free-carrier dispersion).
• This change is opposite to the index change n 2 I resulting from the
Kerr effect.
• Parameter µ is known as the “linewidth enhancement factor” in the
context of semiconductor lasers.
• Its value for silicon is close to 7.5 in the spectral region near 1550 nm.
• Absorption and index changes resulting from free carriers affect the
performance of silicon waveguides.
• Quick removal of carriers helps (e.g., by applying a dc electric field).
◭◭
◮◮
◭
◮
Back
Close

Removal of Free Carriers
14/56
Jones et al, Opt. Exp. 13, 519 (2005)
• A reversed-biased p-n junction is used for this purpose.
• Electric field across the waveguide removes electrons and holes.
• Drift time becomes shorter for larger applied voltages.
• Effective carrier lifetime can be shortened from >20 to

Self-Phase Modulation (SPM)
• Refractive index depends on intensity as n ′ = ¯n + n 2 I(t).
• Propagation constant also becomes intensity-dependent:
15/56
β ′ = β + k 0 n 2 (P/A eff ) ≡ β + γP.
• Nonlinear parameter γ = k 0 n 2 /A eff can be larger by a factor of
10,000 compared with silica fibers.
• Nonlinear Phase shift:
φ NL =
∫ L
0
(β ′ − β)dz =
∫ L
Here, P(z) = P in e −αz and L eff = (1 − e −αL )/α.
• Optical field modifies its own phase (SPM).
0
γP(z)dz = γP in L eff .
• Phase shift varies with time for pulses (chirping).
◭◭
◮◮
◭
◮
Back
Close

Chirping and Spectral Broadening
1
0.8
(a)
2
1
(b)
16/56
Phase, φ NL
0.6
0.4
Chirp, δωT 0
0
−1
0.2
0
−2 −1 0 1 2
−2
−2 −1 0 1 2
Time, T/T 0
Time, T/T 0
• In the case of optical pulses, φ NL (t) = γP(t)L eff .
• Chirp is related to the phase derivative dφ NL /dt.
• Phase and chirp profiles for super-Gaussian pulses are shown
using P(t) = P 0 exp[−(t/T ) 2m ] with m = 1 and m = 3.
• SPM creates new frequencies and leads to spectral broadening.
◭◭
◮◮
◭
◮
Back
Close

Self-Phase Modulation and TPA
• Preceding analysis neglected two-photon absorption.
• Its impact on SPM can be studied by solving:
∂A
∂z = iγ(1 + ir)|A|2 A − α l
2 A.
• This equation ignores dispersive and free-carrier effects.
• Using A = √ Pexp(iφ NL ), we obtain the following analytic solution:
P(L,t) = P(0,t)exp(−α lL)
1 + 2rγP(0,t)L eff
,
φ NL (L,t) = 1 2r ln[1 + 2rγP(0,t)L eff].
• TPA converts linear dependence of φ NL on power to a logarithmic
one.
17/56
◭◭
◮◮
◭
◮
Back
Close

Impact of Two-Photon Absorption
• TPA reduces the maximum
phase shift:
φ 0 = ln(1 + 2rφ max )/(2r).
• In the absence of TPA,
φ 0 = φ max = γP 0 L eff .
nonlinear phase shift φ 0
/ π
3
2.5
2
1.5
1
φ max
= 3π
φ max
= 2π
φ max
= π
• Inset shows the reduction
0.5
using r = 0.1. 0 0.2 0.4 0.6 0.8 1
β TPA
/(k 0
n 2
)
• TPA-induced reduction becomes severe at high powers.
• When φ max = 100, φ 0 is limited to a value of 15.
φ 0
/ π
10
8
6
4
2
0
0 2 4 6 8 10
φ max
/ π
18/56
◭◭
◮◮
◭
◮
Back
Close

Impact of Free-Carrier Generation
Spectral density J/(cm 2 ⋅nm)
Spectral density J/(cm 2 ⋅nm)
φ max
= 1.5π, I 0
= 1.2 GW/cm 2 φ max
= 7.5π, I 0
= 6.0 GW/cm 2
0.015 TPA=0
TPA
only
0.025
0.02
0.01
Full 0.015
0.005
0.01
0.005
0
0
1.549 1.5495 1.55 1.5505 1.551 1.548 1.55
λ (µm)
λ (µm)
1.552
0.015
0.01
0.005
φ max
= 15.5π, I 0
= 12.5 GW/cm 2 φ max
= 15.5π, I 0
= 12.5 GW/cm 2
TPA
only
TPA
+FCA
Full
Phase shift φ / π
0
1.548 1.549 1.55 1.551 1.552
λ (µm)
4
3
2
1
0
−1
−2
−3
TPA
+FCA
TPA
only
Full
−5 −4 −3 −2 −1 0 1 2 3 4 5
t / T 0
P(t) = P 0 e −t2 /T 2
T 0 = 10 ps
L = 2 cm
τ c = 1 ns
α l = 1 dB/cm
Yin and Agrawal,
Opt. Lett. (2007)
Free carriers produce a nonlinear phase shift in the opposite direction.
19/56
◭◭
◮◮
◭
◮
Back
Close

Experimental Results
20/56
Boyraz et al., Opt. Exp. 12, 829 (2004).
• First observation of SPM-induced spectral broadening in 2004.
• 4-ps pulses launched inside a 2-cm-long SOI waveguide.
• The 3-peak output spectrum broadened by a factor of 2 when peak
intensity was 2.2 GW/cm 2 (P 0 ≈ 100 W).
◭◭
◮◮
◭
◮
Back
Close

itted intensity but not in spectral shape and position. The spectral full width at half
(FWHM) is identical to pulses measured without sample. Hence, the applied
the two low-power spectra belong to the regime where the waveguide is responding
inear. However, by increasing the peak power to P = 1.55 W (I ≈ 2.6 GW/cm 2 ), the
ape of the pulses broadens dramatically; now exhibiting several spectral side wings.
ening Experimental continues for higher powers which Results can be seen in the spectrum with P = 6.85
6 GW/cm 2 ).
-70
Transmission (dBm)
-75
-80
-85
-90
-95
-100
-105
6.85 W
1.55 W
8 mW
1.8 mW
100 200 300 400
1480 1500 1520
Wavelength (nm)
Fig. 1. Transmission spectra of a 4 mm long SOI waveguide as a function of coupled peak
power. Linear optical response is observed for low powers (black and red) whereas SPMinduced
spectral
Dulkeith
broadening
et
occurs
al.,
for
Opt.
high powers
Exp.
(green
14,
and
5524
blue). Inset: Time- and
spectrally resolved FROG trace of input pulses at 1500 nm.
(2006).
easured spectral pulse distortions can be explained by self-phase-modulation,
nown to cause an intensity dependent phase shift of the pulse carrier frequency.
own duration, the pulse experiences an intensity- and thus time-dependent
index. New frequency components are generated and the initial pulse spectrum
100
200
300
400
500
600
53.2
49.9
43.6
37.4
31.2
24.9
18.7
12.5
6.23
0
• Larger broadening by 2006.
• 1.8-ps pulses launched
inside a 4-mm-long
waveguide.
• Width 470 nm
height 226 nm.
• Spectral asymmetry is due
to free-carrier effects.
• Inset shows the FROG
trace.
21/56
◭◭
◮◮
◭
◮
Back
Close

e broadened and asymmetric. The significant difference is that SPM-induced
g becomes more efficient for longer wavelengths. The spectral distance between the
length SPM oscillation and the originally injected laser wavelength increases from
when the OPA is tuned from 1400 to 1650 nm. As will be discussed in the next
is enhancement for longer wavelengths is probably due to an increase of group
d nonlinear
Wavelength
refractive index n 2 .
Dependence of SPM
Transmission (dB)
-75
-80
-85
-90
-95
-100
-105
1400 1450 1500 1550 1600 1650
Wavelength (nm)
Fig. 4. SPM-induced spectral broadening of optical pulses for the same coupled
peak power (6.85 W) injected into the SOI waveguide at 1400, 1500 and 1600 nm.
The spectral Dulkeith broadening etincreases al., with Opt. the wavelength. Exp. 14, 5524
(2006).
Experimental Results agree with theoretical predictions.
• SPM-broadened spectra at
three different wavelengths.
• 1.8-ps pulses tunable from
1400 to 1650 nm.
• Larger broadening at longer
wavelengths.
• Consistent with a larger n 2
near 1550 nm
rison with theory and discussion
description
e the experimental data we have used a recently developed theoretical model to
e pulse dynamics in Si photonic wires. It accounts for GVD, parametric and nonnonlinear
effects such as SPM, XPM, TPA, or Raman interaction; free carrier-
22/56
◭◭
◮◮
◭
◮
Back
Close

Formation of Optical Solitons
• SPM-induced phase shift φ NL > 1 is easily realized.
• φ NL = 1 occurs at z = L NL = 1/(γP 0 ).
• Nonlinear length L NL ∼1 cm at moderate peak powers

Formation of Optical Solitons
Normalized P ower
S pectral Density
1
0.8
0.6
0.4
0.2
P ulse S hape
0
500 400 300 200 100 0 100 200 300 400 500
time (fs)
P ulse S pectrum
1
0.8
0.6
0.4
0.2
P ath averaged
T P A
P ath averaged
Input
Input
0
1.51 1.52 1.53 1.54 1.55 1.56 1.57 1.58 1.59
Wavelength (µm)
Yin, Lin, and Agrawal, Opt. Lett. 31, 1295 (2006)
130-fs pulses launched inside a 5-mm-long waveguide (N = 1).
T P A
α
24/56
◭◭
◮◮
◭
◮
Back
Close

Path-Averaged Solitons
• Perfect solitons do not for inside silicon waveguides.
• Linear and nonlinear (TPA) losses broaden the pulse as N = 1 is
not maintained along the whole device.
25/56
• Concept of path-averaged solitons can be used to reduce
pulse broadening.
• Input peak power increased such that N = 1 on average.
• Required peak power is estimated using
¯P 0 = 1 L
∫ L
0
P 0 (z)dz = 1
γL D
.
• Solion formation has been observed in a recent experiment.
◭◭
◮◮
◭
◮
Back
Close

Observation of Optical Solitons
26/56
• In a 2007 experiment, Zhang et al. [Opt. Exp. 15, 7682 (2007)]
launched 110-fs Gaussian pulses inside a 5-mm-long waveguide.
• Gaussian spectrum broadened at 1250 nm because of β 2 > 0.
• It narrowed and acquired a “sech” shape at 1484 nm where β 2 < 0.
◭◭
◮◮
◭
◮
Back
Close

Numerical Simulations
0
0
−5
−5
27/56
−10
−10
Normalized Intensity (dB)
−15
−20
−25
Normalized Spectrum (dB)
−15
−20
−25
−30
−30
−35
−35
−40
−8 −6 −4 −2 0 2 4 6 8
Time τ/T 0
• Numerical results under experimental conditions.
−40
1420 1440 1460 1480 1500 1520 1540
Wavelength (nm)
• Input: blue curves; Output: red curves; low power case (green).
• Experimental data in agreement with numerical predictions.
◭◭
◮◮
◭
◮
Back
Close

Supercontinuum Generation
• Ultrashort pulses are affected by a multitude of nonlinear effects,
such as SPM, XPM, FWM, and SRS, together with dispersion.
• All of these nonlinear processes are capable of generating new
frequencies outside the input pulse spectrum.
• For sufficiently intense pulses, the pulse spectrum can become so
broad that it extends over a frequency range exceeding 100 THz.
• Such extreme spectral broadening is referred to as supercontinuum
generation.
• This phenomenon was first observed in solid and gases more than
35 years ago (late 1960s.)
• Since 2000, microstructure fibers have been used for supercontinuum
generation.
28/56
◭◭
◮◮
◭
◮
Back
Close

SC Generation in a microstructured fiber
29/56
Ranka et al., Opt. Lett. 25, 25 (2000)
• Output spectrum generated in a 75-cm section of microstructured
fiber using 100-fs pules with 0.8 pJ energy.
• Even for such a short fiber, supercontinuum extends from
400 to 1600 nm.
• Supercontinuum is also relatively flat over the entire bandwidth.
◭◭
◮◮
◭
◮
Back
Close

SC Generation through Soliton Fission
• Input pulses correspond to a higher-order soliton of large order;
typically N = (γP 0 T 2
0 /|β 2|) 1/2 exceeds 5.
• Higher-order effects lead to their fission into much narrower
fundamental solitons: T k = T 0 /(2N + 1 − 2k).
30/56
• Each of these solitons is affected by intrapulse Raman scattering.
• Spectrum of each soliton is shifted toward longer and longer wavelengths
with propagation inside the fiber.
• At the same time, each soliton emits nonsolitonic radiation at a
different wavelength on the blue side.
• XPM and FWM generate additional bandwidth and lead to a broad
supercontinuum.
◭◭
◮◮
◭
◮
Back
Close

SC Generation in Silicon Waveguides
• SOI waveguides also support higher-order solitons when
N = (γP 0 T 2
0 /|β 2|) 1/2 exceeds 1.
• Higher-order dispersion leads to their fission into much
shorter fundamental solitons: T k = T 0 /(2N + 1 − 2k).
• Spectrum of each soliton shifts toward longer wavelengths by a
smaller amount because of a narrower Raman-gain spectrum of silicon.
• Similar to the case of optical fibers, each soliton is expected to emit
dispersive waves with frequencies on the blue side.
• Numerical simulations confirm the potential of SOI waveguides for
SC generation.
• Spectral broadening over 350 nm is predicted for femtoseond pulses.
31/56
◭◭
◮◮
◭
◮
Back
Close

Spectral and Temporal Evolution
Normalized Power
Normalized Power (dB)
0.1
0.05
0
−0.2 0 0.2 0.4 0.6 0.8
time (ps)
0
−10
−20
Solitons
Dispersive
wave
−30
Cherenkov
radiation
Solitons
−40
1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9
λ (µm)
Yin, Lin, and Agrawal, Opt. Lett. 32, 391 (2007)
50-fs pulses launched with 25-W peak power into a 1.2-cm waveguide.
32/56
◭◭
◮◮
◭
◮
Back
Close

Impact of TPA on SC Generation
Normalized Power (dB)
0
−10
−20
L=3 mm
TPA=0
−30
L=3 mm
with TPA
−40
1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9
λ (µm)
Yin, Lin, and Agrawal, Opt. Lett. 32, 391 (2007)
• TPA reduces SC bandwidth but is not detrimental.
• Nearly 400-nm-wide supercontinuum created within
a 3-mm-long waveguide.
• Required pulse energies are relatively modest (∼1 pJ).
33/56
◭◭
◮◮
◭
◮
Back
Close

Optical Switching
34/56
• A Mach-Zehnder interferometer can be made using two Si waveguides.
Such a device exhibits SPM-induced optical switching.
• In each arm, optical field accumulates linear and nonlinear
phase shifts.
• Transmission through the bar port of theinterferometer:
T = sin 2 (φ L + φ NL ); φ NL = (γP 0 /4)(L 1 − L 2 ).
• T changes with input power P 0 in a nonlinear fashion.
◭◭
◮◮
◭
◮
Back
Close

Cross-Phase Modulation
• Consider two optical fields propagating simultaneously.
• Nonlinear refractive index seen by one wave depends on the
intensity of the other wave as
35/56
∆n NL = n 2 (|A 1 | 2 + b|A 2 | 2 ).
• Total nonlinear phase shift in a fiber of length L:
φ NL = (2πL/λ)n 2 [I 1 (t) + bI 2 (t)].
• An optical beam modifies not only its own phase but also of other
copropagating beams (XPM).
• XPM induces nonlinear coupling among overlapping optical pulses.
◭◭
◮◮
◭
◮
Back
Close

XPM-Induced Spectral Changes
36/56
Hsieh et al., Opt. Exp. 15, 1135 (2007)
• 200-fs pump and probe pulses (at 1527 and 1590 nm) launched into
a 4.7-mm-long SOI waveguide (w = 445 nm, h = 220 nm).
• Pump and probe pulses travel at different speeds (walk-off effect).
• XPM-induced phase shifts occurs as long as pulses overlap.
• Asymmetric XPM-induced spectral broadening depends on pump
power (blue curve); Probe spectra without pump (red curve).
◭◭
◮◮
◭
◮
Back
Close

XPM-Induced Switching
37/56
• A Mach–Zehnder interferometer is often used for optical switching.
• Output switched to a different port using a control signal that shifts
the phase through XPM.
• If control signal is in the form of a pulse train, a CW signal can be
converted into a pulse train.
• Turn-on time quite fast but the generation of free carriers widens
the switching window (depends on the carrier lifetime).
◭◭
◮◮
◭
◮
Back
Close

Experimental Demonstration
38/56
Boyraz et al., Opt. Exp. 12, 4094 (2004)
• A Mach–Zehnder interferometer used for optical switching.
• Short pump pulses (

XPM-Induced Switching
39/56
Boyraz al., Opt. Exp. 12, 4094 (2004)
• Instantaneous switching on the leading edge, as expected, with high
on–off contrast.
• Long trailing edge results from the free-carrier effects.
• Free carriers provide an additional contribution to the probe phase
by changing the refractive index.
◭◭
◮◮
◭
◮
Back
Close

Four-Wave Mixing (FWM)
40/56
• FWM is a nonlinear process that transfers energy from pumps
to signal and idler waves.
• FWM requires conservation of
⋆ Energy ω 1 + ω 2 = ω 3 + ω 4
⋆ Momentum β 1 + β 2 = β 3 + β 4
• Degenerate FWM: Single pump (ω 1 = ω 2 ).
◭◭
◮◮
◭
◮
Back
Close

Parametric Amplifiers
• FWM can be used to amplify a weak signal.
• Pump power is transferred to signal through FWM.
• The idler (generated as a byproduct) acts as a copy of the signal at
a new wavelength (useful for wavelength conversion).
41/56
• Parametric amplifiers can provide gain at any wavelength using
suitable pumps.
• They are also useful for all-optical signal processing.
• Optical fibers are often used, but the use of SOI waveguides would
result in a much more compact device.
• Impact of two-photon and free-carrier absorption requires further
study.
◭◭
◮◮
◭
◮
Back
Close

FWM Theory for Silicon Waveguides
• Theory should include TPA and free-carrier effects fully.
• Polarization and Raman effects should also be included.
• Full vector theory by Lin et al., Opt. Exp. 14, 4786 (2006).
• Free-carrier absorption limits the gain for a CW pump.
• β 2 < 0 (red); β 2 = 0 (blue); β 2 > 0 (green).
42/56
◭◭
◮◮
◭
◮
Back
Close

FWM with Short Pump Pulses
43/56
Lin et al., Opt. Exp. 14, 4786 (2006)
• FCA is reduced significantly for pump pulses much shorter than
carrier lifetime τ c .
• Figure shows the case of 10-ps pump pulses with τ c = 1 ns.
• Phase-matching condition is satisfied even for signal that is shifted
by 70 nm from the pump wavelength.
◭◭
◮◮
◭
◮
Back
Close

Singe and Dual-Pump Configurations
44/56
Lin et al., Opt. Exp. 14, 4786 (2006)
• Parametric amplifiers with a large bandwidth can be realized by
pumping an SOI waveguide with two pumps.
• This is possible because of a relatively short device length.
• Recent experiments with SOI waveguides are encouraging.
◭◭
◮◮
◭
◮
Back
Close

Experimental Results (Single Pump)
45/56
Foster et al., Nature 441, 760 (2006)
• 3.5-ps pump pulses at 1550 nm; signal tunable over 80 nm.
• Up to 5 dB of gain observed in devices

Photon-Pair Generation
46/56
Lin et al., Opt. Lett. 31, 3140 (2006)
• Spontaneous FWM in fibers creates entangled photon pairs but
suffers from the noise induced by Raman scattering.
• The use of SOI waveguides avoids this problem because Raman
scattering does not occur when TM mode is excited.
◭◭
◮◮
◭
◮
Back
Close

Experimental Results
47/56
Sharping et al., Opt. Exp. 14, 12388 (2006)
• 5-ps pump pulses launched into a 9-mm-long SOI waveguide.
• Total (red) and accidental (blue) coincidence rates measured.
• Their ratio exceeded 10 at low pump powers.
◭◭
◮◮
◭
◮
Back
Close

Stimulated Raman Scattering
• Scattering of a pump beam from vibrating molecules creates a
Stokes beam down-shifted in frequency by a specific amount.
48/56
• Frequency shift is set by a vibrational mode (phonons).
• Raman gain spectrum exhibits a dominant peak at 15.6 THz with
a 105-GHz bandwidth (≈1 nm wide near 1550 nm).
• Peak gain for silicon >1000 larger compared with silica.
Claps et al., Opt. Exp.
13, 2459 (2006)
◭◭
◮◮
◭
◮
Back
Close

Raman Amplifiers
49/56
Jalali et al., IEEE JSTQE 12, 412 (2006)
• CW pumping leads to accumulation of free carriers through TPA.
• Free-carrier absorption introduces losses for pump and signal.
• No signal gain occurs for τ eff > 10 ns.
◭◭
◮◮
◭
◮
Back
Close

Pulsed Raman Amplifiers
50/56
Jalali et al., IEEE JSTQE 12, 412 (2006)
• Pulsed pumping can provide >20-dB gain if spacing among pulses
is much larger than τ eff (R p τ eff ≪ 1).
• Free carriers can then decay before the next pulse arrives.
• Pump pulses (∼30 ps) at 1540 used to amplify a 1673-nm signal.
• 20-dB net gain realized at 37-W peak power of pump pulses.
◭◭
◮◮
◭
◮
Back
Close

Raman Lasers
51/56
Boyraz and Jalali, Opt. Exp. 12, 5269 (2004)
• Pumped with 30-ps pulses at 1540 nm at 25-MHz repetition rate.
• Produced 18 ps pulses at 1675 nm at the same repetition rate.
◭◭
◮◮
◭
◮
Back
Close

CW Raman Amplifiers
52/56
Jones et al, Opt. Exp. 13, 519 (2005)
• CW pumping can be used if free carriers are removed quickly.
• A reversed-biased p-n junction is used for this purpose.
• Electric field across the waveguide removes electrons and holes.
• Drift time of carriers is shorter for larger applied voltages.
◭◭
◮◮
◭
◮
Back
Close

CW Raman Amplifiers
53/56
Jones et al, Opt. Exp. 13, 519 (2005)
• A 4.8-cm-long waveguide CW pumped at 1458 nm (signal at 1684 nm).
• Output pump and signal powers increase with applied voltage.
• Effective carrier lifetime decreases from 16 to 1 ns.
◭◭
◮◮
◭
◮
Back
Close

Theory behind Raman Amplifiers and Lasers
• Pump and signal powers satisfy the set of two coupled equations:
∂P p
∂z
∂P s
∂z
= −(α lp + α fp )P p − β Tpp P 2 p − 2β Tps P s P p − g R P s P p
= −(α ls + α fs )P s − β Tss P 2 s − 2β Tsp P p P s + g R P p P s .
54/56
• Signal loss α fs by free carriers (generated from the pump-induced
TPA) limit the performance severely.
• For net amplification to occur, the carrier lifetime should satisfy
τ 0 < τ th ≡ ¯hω pn s ā 2 pp(g R − 2β Tsp ) 2
.
2α ls σ as n 0s β Tpp ā 2 sp
• A Raman laser cannot function if this condition does not hold.
◭◭
◮◮
◭
◮
Back
Close

Concluding Remarks
• Nonlinear effects in silicon waveguides can be used to make many
active and passive components.
55/56
• SPM is useful for soliton formation and supercontinuum generation.
• SPM and XPM can also be used for optical switching, wavelength
conversion, and all-optical regeneration.
• Four-wave mixing converts silicon waveguides into parametric
amplifiers.
• It can also be used for quantum applications requiring entangled
photon pairs.
• Stimulated Raman scattering converts silicon waveguides into
Raman amplifiers.
◭◭
◮◮
◭
◮
Back
Close

Further Reading
• L. Pavesi and D. J. Lockwood, Silicon Photonics (Springer, 2004).
• G. T. Reed, Silicon Photonics: State of the Art (Wiley, 2007).
56/56
• IEEE J. Sel. Topics Quantum Electron., Special issue on Silicon
Photonics, 12 (2006).
• B. Jalali and S. Fathpour, “Silicon Photonics,” J. Lightwave Technol.
24, 4600 (2006).
• B. Jalali, “Teaching silicon new tricks,” Nature Photonics 1, 193
(2007).
• Q. Lin, O. J. Painter, G. P. Agrawal, “Nonlinear optical phenomena
in silicon waveguides,” Opt. Express 15, 16604 (2007).
◭◭
◮◮
◭
◮
Back
Close