1. science
2. mathematics
3. statistics

Presentation outline
1
2
• Why DFB lasers with (LC-RWG) surface gratings?
• Particularities of LC-RWG surface gratings
• Application & target characteristics: ESA requirements
• Epilayer structures adjusted for LC-RWG gratings
• Coupling coefficient & modal selectivity
• Requirements on coupling coefficient
and fabrication limitations
• End-mirror effects
• Device fabrication & fabricated device characteristics
• Towards improved device characteristcs
• narrower spectral linewidth
Narrow linewidth 894 nm DFB lasers
with laterally-coupled ridge-waveguide surface gratings
fabricated using nanoimprint lithography
M. Dumitrescu,
Dumitrescu, J. Telkkä
Telkkälä, J. Karinen, A. Laakso, J. Viheriä
Viheriälä,,
T. Leinonen, J. Lyytikä
Lyytikäinen, L. Toikkanen, M. Pessa
Optoelectronics Research Centre, Tampere University of Technology, Tampere,
Finland
Development of Extremely Narrow-Band
Semiconductor Distributed Feedback Laser
Technology
The long path from idea to product: idea (ne w structure, ne w technology) +
acquiring the facilities + proof of conce pt experime nts + project proposal +
detailed analysis + design + expe rime nts + (many) ite rations
CAS 2010
Tuesday, Oct. 12, O2
CAS 2010
Tuesday, Oct. 12, O2
3
Why DFB lasers with
LCLC-RWG surface gratings?
Differences in coupling coefficient dependence
on grating structure
4
3
4
DFB laser with a buried grating:
Surface gratings:
single growth and processing sweep
=> better yield and lower cost
limited interaction between the carriers & the grating
=> more controllable performances
=> more stable (GaAs) devices
reduced coupling coefficient
DFB laser with a LC-RWG grating:
Akiba et al. IEEE J. Quantum Electron. 19, 1052 (1983)
smaller Γ,
higher Δn
=> high k is
achievable
The conventional buried-grating fabrication
techniques require overgrowth, which leads
to problems with yield, performance and
device cost.
Fabricated using UV-based nano-imprint lithography
(UV-NIL), which enables pattern resolutions beyond the
limitations set by the diffraction and scattering for the
conventional techniques.
basic dimensions: t, W, D,
grating parameters: m, Λ, γ
CAS 2010
longitudinal parameters: L, R1, R2, d1, d2
Tuesday, Oct. 12, O2
The effects of the grating order and
filling factor on SMSR are not
discussed in the literature (since
buried-grating devices have been
used).
Λ1/Λ = filling factor
k
sin( πmγ )
κ = 0 ⋅ (n22 − n12 ) ⋅ Γg ⋅
πm
2 neff
=> with surface gratings high
filling factor (i.e. small trench
width) should be targeted
Both transverse and longitudinal mode
discrimination have to be taken into account
CAS 2010
Tuesday, Oct. 12, O2
6
Epilayer structure
Application & target characteristics
Active region (894 nm)
5
6
The 1955 Cesium Atomic Clock at the National Physical Laboratory, UK.:
Cesium clocks measure frequency with an accuracy of from 2 to 3 parts in 10 to the
14th, with an averaging time of 5 days. i.e. 0.00000000000002 Hz; this corresponds
to a time measurement accuracy of one second in 1,400,000 years. It is the most
accurate realization of a unit that mankind has yet achieved.
For the emission at 894 nm
GaInAs/GaAs QWs are too shallow
and don’t have high enough gain for
effective operation.
In 1967, the 13th General Conference on Weights and Measures first defined the
International System (SI) unit of time, the second, in terms of atomic time as the
duration of 9,192,631,770 cycles of microwave light absorbed or emitted by the
hyperfine transition of cesium-133 atoms in their ground state undisturbed by
external fields.
(a) Material gain simulated for 8 nm GaIn0.20 As/GaAs QW, and (b) the
corresponding peak material gain compared with the variation reported in
the literature (blue curve)
The target: low-cost mass-produced atomic clock no larger than a sugar cube, which could run on an AA battery,
yet accurate to one second in 300 years.
The competition: National Institute of Standards and Technology (NIST), Boulder, USA, under the Defense Advanced
Research Projects Agency’s (DARPA’s) Chip-Scale Atomic Clock (CSAC) program
Cesium atomic clocks (ESA project for developing narrow linewidth DFB lasers)
emission at 894.6 nm (tunable ± 2nm) (Cesium D1 line)
=> grating period Λ ≈ 140, 280, 420 nm
output power ≥ 15 mW (@ Ibias < 100 mA)
single mode CW operation with SMSR > 25 dB
spectral linewidth << 1.5 MHz
171,173 Yb
87,88Sr
40Ca+
27Al+
199,201 Hg
Cooling lasers (1st, 2nd , ...)
Auxiliary lasers
Lattice lasers
Probe lasers
(a) The peak material gain simulated for 5 nm GaInAs/Al0.30GaAs,
GaInAs/GaIn0.485 P, and GaInAs/GaAs QWs, and (b) for 5 nm
GaInAs/GaIn0.485 P QWs at 300, 325 and 350 K
GaInAs/AlGaAs or GaInAsP/GaAsP
QWs would be deep enough, but the
growth temperatures for GaInAs and
AlGaAs are quite different and it is not
recommended to use AlGaAs next to
GaInAs QWs for reliability reasons .
The experimental studies indicate that
the group V elements ratio in
GaInAsP/GaAsP QWs is not easily
controllable.
=> GaInAs/GaInP active regions
(GaInP can be used as etch-stop)
CAS 2010
Tuesday, Oct. 12, O2
CAS 2010
Tuesday, Oct. 12, O2
7
8
Epilayer structure
Epilayer structure
Waveguide and cladding layers
(for high grating coupling coefficient)
7
Threshold current (mA)
Al0.6GaAs cladding
Al0.7GaAs cladding
350
50
325
45
300
40
275
250
35
225
Higher Al-content in the cladding layer
• Wider far-field
• Narrower near-field → cladding thickness
can be reduced (fabrication of the grating is
easier), but coupling coefficient is smaller
due to smaller achievable Γg
200
30
175
150
50
SQW
κ=
100
150
8
200
250
300
25
Waveguide thickness (nm)
k0
sin(πmγ )
⋅ ( n22 − n12 ) ⋅ Γg ⋅
2neff
πm
500
0
50
100
150
200
250
Surface gratings have low coupling
coefficients (??)
300
0,016
400
300
0,014
200
0,012
100
0
0,010
-100
Drastic changes of the epilayer
structure in order to extend the optical
field penetration into the cladding
affect other critical parameters and
small changes do not bring significant
coupling coefficient improvement.
0,008
-200
”Barrier reduction” layers added to
GaAs-Al 0.7GaAs interfaces in order to
improve current-voltage characteristics.
The QW confinement factor is usually
maximized for low threshold and high
bandwidth but this affects the coupling
coefficient.
single GaInAs quantum well (QW)
120 nm GaInP waveguide (WG),etch-stop layer.
100 nm Al0.3Ga0.7 As barrier-reduction layers
1000 nm thick Al0.7Ga0.3 As cladding layers,
200 nm GaAs contact layer on the p-side.
QW confinement factor
Al0.5GaAs cladding
375
FF-FWHM (degrees)
High QW confinement factor
• Small threshold current
55
NF FWHM extension beyond the WG [nm]
Near-field FWHM [nm]
400
0
50
100
150
200
250
Etching depth should be used to ensure
adequate coupling coefficient
300
GaInP waveguide thickness [nm]
InGaAs
InP
WS
L
InGaAlAs
GRINSCH
QDashes
GRINSCH
InGaAlAs
InP
DW
⇒Etching through the active region
increases dramatically the losses
⇒Etching should be stopped above the
active region (is this enough?)
L
CAS 2010
Tuesday, Oct. 12, O2
CAS 2010
Tuesday, Oct. 12, O2
9
10
Coupling coefficient calculation
Coupling coefficient calculation
(effect of approximations)
9
2
∫ ∫ ∆ε ( x, y, z ) ⋅ Ψ ( x, y ) dxdy
k0 − ∞− ∞
⋅
∞ ∞
2neff
2
∫ ∫ Ψ ( x, y) dxdy
− ∞− ∞
∞ ∞
κ=
k0
2neff
[
∞ ∞
∫ ∫n
2
1
−∞ −∞
∞
∫

(x, y)Ψ 2 ( x, y)dxdy 

∞

2
∫ Ψ ( x, y)dxdy 

− ∞− ∞
sin(πmγ )
k0
⋅ ( n22 − n12 ) ⋅ Γg ⋅
2neff
πm
κ=
2
2
Conventional approximation
≈ k0 ⋅ (n2 − n1 ) ⋅ Γg ⋅
The ’standard’ formula for calculating the
coupling coefficient is over-estimating the
LC-RWG grating coupling coefficient
300
] sin(ππmmγ )
∆ε m ( x, y , z ) = n2 ( x, y ) − n1 ( x, y ) ⋅

 ∫ ∫ n22 ( x, y )Ψ 2 ( x, y)dxdy
sin(πmγ )  − ∞− ∞
⋅
⋅
−
∞ ∞
πm
2

∫ ∫ Ψ ( x, y )dxdy


−∞ −∞
κ=
K2139LD980 with W=1.5 µm and D=0.5 µm
Coupling coefficient (1/cm)
∞ ∞
κ=
10
Dielectric perturbation Δεm(x,y,z) for
m-th order rectangular-shaped grating:
According to coupled-wave theory:
sin( πmγ )
πm
=> ove r-estimatimation
k0
sin(πmγ )
2
2
⋅ ( neff ,2 − neff ,1 ) ⋅
2 neff
πm
Crosslight's PICS3D
'Standard' formula
Convolution approach + adjusted formula
Conventional approach + adjusted formula
250
200
κ=
150
κ0
2 neff
(
)
⋅ n22 − n12 ⋅ Γg ⋅
sin( πmγ )
sin( πmγ )
≈ k 0 ⋅ ( n 2 − n1 ) ⋅ Γg ⋅
πm
πm
over-estimatimation since
100
50
0
0
50
100
150
200
250
300
2n eff > ( n2 + n1 )
The adjusted formula can produce underestimation if the optical field is calculated
separately for each grating section
k
sin(πmγ )
κ = 0 ⋅ ( n eff , 2 2 − neff ,1 2 ) ⋅
πm
2neff
Thickness of the un-etched cladding (nm)
Third-order LC-RWG grating coupling coefficient values calculated
with the standard formula (stars), with the adjusted formula and
conventional method (triangles), with the adjusted formula and
convolution method (circles) and obtained with PICS3D (squares)
neff , i =
∫Ψ
2 ( x, y ) ⋅ n 2 (x , y ) ⋅dx ⋅ dy
i
2
∫ Ψ (x, y) ⋅dx
(neff2_conventional-neff1_conventional) < (neff2_convolution - neff1_convolution)
unde r-estimatimation
possible
CAS 2010
(for certain dimensional ranges)
Tuesday, Oct. 12, O2
CAS 2010
Tuesday, Oct. 12, O2
11
Transverse modal selectivity
SMSR calculation
(coupling coefficient discrimination)
11
SMSR depends mostly on the bias current and the mirror
loss margin, Δα:

SMSR(dB) ≈ 10 log10 

where
δ G = (α i + α m (λ0 ) )β spη r
∆α + ∆G
δG
12
 ∆α + ∆G 
SMSR( dB) ≈ 10 log10 
+ 1
 δG


+ 1

1
I / I th − 1
∆α = αm(λ1) - αm(λ0) is the loss margin, ∆G = Γg(λ 0) - Γg(λ 1) is the modal gain
margin and δG is the net modal gain for the main mode, βsp is the spontaneous
emission factor and ηr is the radiative recombination efficiency etc.
Modal positions and corresponding
mirror losses are solved with
Transfer Matrix Method
Threshold gain condition:
Shown for 980 nm grating
+ alignment with
gain spectrum
2
HR
AR
R1 r (λ , α ) = 1
Phase condition:
Both transverse and longitudinal mode
discrimination have to be taken into account
r(λ,α) has to be real and positive
CAS 2010
Tuesday, Oct. 12, O2
higher order transverse modes might
have higher coupling coefficients but ...
CAS 2010
Tuesday, Oct. 12, O2
Transverse modal selectivity
Longitudinal mode selectivity
(confinement factor discrimination)
(κL)
13
Thickness of the
un-etched cladding (nm)
 ∆α + ∆G 
SMSR( dB) ≈ 10 log10 
+ 1
 δG

( Γ1 - Γ2 ) / Γ1
500
450
400
350
300
250
200
150
100
50
0.5
1.0
1.5
mirror loss discrimination
modal gain discrimination
2
Gm =
∫ ∫ Ψm ( x , y ) ⋅ g m ( x , y )
∫∫
( Γ1 - Γ3 ) / Γ1
2.0
2.5 0.5
1.0
1.5
2.0
2
± ∞ Ψm ( x, y ) dx dy
SMSR increases with κL only up to
a certain κL value:
• small κL-value
→ not enough modal selectivity
• high κL-value
→ spatial hole burning
→ multiple longitudinal modes
2.5
Simulated SMSR for a DFB structure with phase-matched facets, R 1 = 95 %, R 2 = 2 %,
L = 500 µm and αi = 5, 10, 15 or 20 cm-1 when (a) κ is either 0 or 20 cm-1 and I is
varied, and (b) I/Ith = 3 and κ is varied
Width of the middle section, W (µm)
Thickness of the
un-etched cladding (nm)
2
(Γ1-Γ2)*(Γ1-Γ3)/Γ1
500
450
400
350
300
250
200
150
100
50
0.5
1.0
1.5
2.0
0
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
2.5
14
dxdy
=> κL≈1-2
gm ( x, y ) act.reg _ under−ridge = g
2
Gm ≈ g
∫ ∫ act reg , under ridge Ψm ( x, y )
∫∫
Width of the middle section, W (µm)
Γ =
1m
dx dy
= gΓm
Ψ 2 ( x, y ) dx dy
±∞ m
g Γ − gΓ
Γ −Γ
m= 1 m
1
gΓ
Γ
1
1
κL≈1-2 => 1-2 longitudinal modes
within the grating stopband
CAS 2010
Tuesday, Oct. 12, O2
CAS 2010
Tuesday, Oct. 12, O2
Coupling coefficient
Modal degeneracy
(targets & limits)
(phase(phase-shift section & mirror reflections)
80
3,205
3,204
70
60
3,203
50
40
3,202
30
20
3,201
10
0
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
Filling factor, γ
grating trench width is an
essential parameter for
achieving high κ
0,9
3,200
1,0
100
Use of end-mirror reflections
yellow
blue
Devices without and with a phase-shift section when
end facets have perfect anti-reflection (AR) coatings
the SMSR deteriorates rapidly for facet reflectivities
of only a few percent due to the phase randomness of
the facet reflections
K2281LD894 with W=1.5 µm, D=0.5 µm and t=0 nm
st
1 transverse mode
90
Coupling coefficient
and
80
st
nd
1 -order grating,
rd
3 -order grating
16
Use of a phase-shift section
The product κL=1-2 is important for
longitudiunal mode selectivity
High κ is critical only when devices
have to be short (high modulation
bandwidth)
When the devices can be long (narrow
linewidth) the technological limits
matter more
Coupling coefficient, κ [1/cm]
Coupling coefficient, κ [1/cm]
90
K2281LD894 with W=1.5 µµm, D=0.5 µµm and t=0 nm
st
1 transverse mode
st
nd
Coupling coefficient
1 -order grating,
2 -order grating
rd
3 -order grating; and
Average effective refractive index
and
Average effective refractive index
15
100
2 -order grating
HR
70
Etch aspect
ratio
> 10 < 10
60
50
3rd-order grating
with γ=½
20
yellow
AR / high reflection (HR) –coated device without a phase-shift
section when the distance between HR coating and the last
grating period is varied
=> all DFB laser designs have an element of
randomness as a result of the uncontrollable
phase shift associated with the facet
reflections.
40
30
red
10
0
0
50
100
150
200
250
300
Trench width, t [nm]
350
400
450
CAS 2010
Tuesday, Oct. 12, O2
AR
CAS 2010
Tuesday, Oct. 12, O2
End-mirror influence
Power – linewidth
(randomness in laser characteristics)
(compromise evaluation)
0.1
0.0
0.0 0.1 0.2 0.3 0.4 0.5
0.0 0.1 0.2 0.3 0.4 0.5
0.0 0.1 0.2 0.3 0.4 0.5
0.5
0.85
0.4
0.95
0.3
1.05
0.2
1.15
0.1
1.25
SMSR (dB)
23.0
25.0
27.0
29.0
31.0
33.0
35.0
37.0
39.0
41.0
(Optical thickness of d1)/λ
1.35
0.0
0.0 0.1 0.2 0.3 0.4 0.5
0.0 0.1 0.2 0.3 0.4 0.5
0.5
Ith (mA)
∆ν (MHz)
0.3
0.2
0.1
0.0
0.0 0.1 0.2 0.3 0.4 0.5
0.7
0.8
0.9
1.0
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2.0
SMSR (dB)
0.0 0.1 0.2 0.3 0.4 0.5
improves
degrades
21.0
23.3
25.5
27.8
0.0 0.1 0.2 0.3 0.4 0.5
30.0
6.3
improves
κκL = 0.0
κκL = 0.5
κκL = 1.0
κκL = 1.5
κκL = 2.0
1
7.1
6.6
9.1
6.9
11.1
0,1
0
13.1
20
15
10
κ, L increase => Δν↓, Pout↓
achieving both Pout>15 mW and Δν < 1.5 MHz
while 0.5 < κL < 2 (in order to have high SMSR)
possible only in limited κ, L ranges
5
10
20
30
Current (mA)
40
0
50
Calculated spectral linewidth (blue lines) and output
power (red lines) variations for 1.2 mm long 894 nm
LC-DFB structure with phase-matched R1=0.95
& R2= 0.02.
15.1
17.1
19.1
20.4
8.4
0.0 0.1 0.2 0.3 0.4 0.5
25
P (mW)
8.1
Ith (mA)
λ = 894 nm, neff = 3.2,
k = 22 cm-1, L = 700 μm,
Ibias = 90mA, αeff = 6 &
αi = 25 cm-1
R1 = 0.95 & R2 = 0.02
7.8
(Optical thickness of d2 )/λ
R 2↑
18.8
7.5
0.0 0.1 0.2 0.3 0.4 0.5
22
24
26
28
30
32
34
36
38
40
42
44
46
16.5
18
10
14.3
7.2
(Optical thickness of d2)/λ
0.4
12.0
0.0 0.1 0.2 0.3 0.4 0.5
R1 = 0.95 & R2 = 0.275
5.4
6.3
7.2
8.1
9.0
9.9
10.8
11.7
12.6
13.5
14.4
P (mW)
0
3
6
9
12
15
18
21
24
27
30
⇒ some important parameters
are contradictory influenced
by mirror phase (e.g. output
power, bandwidth & linewidth)
0.0 0.1 0.2 0.3 0.4 0.5
κ ≈ 22 cm-1, L = 550 µm => κL=1.21
simulated Δν≈0.7 MHz, Pout≈20 mW
κ ≈ 22 cm-1, L = 1000 µm => κL=2.2
simulated Δν≈0.35 MHz, Pout<10 mW
R1 = 0.95 & R2 = 0.95
6
40
-1
30
κ = 10 cm
4
25
-1
κ = 5 cm
3
20
15
2
1
10
κ=
-1
20 cm
-1
10 cm
-1
5 cm
5
0
1000
1500
2000
L ( µm)
CAS 2010
Tuesday, Oct. 12, O2
Nanoimprint lithography (NIL)
35
-1
500
degrades
0.5 < κ L < 2.0
κ = 20 cm
5
2500
3000
3500
0
4000
CAS 2010
Tuesday, Oct. 12, O2
NIL stamp structure
19
Contact technique involving pressure (small features =>
potential for very high pressure) and mechanical deformation =>
requirements:
Power (mW) when I = 50 mA
(Optical thickness of d1)/λ
(Optical thickness of d2)/λ
∆ν (MHz)
P (mW)
6.8
7.3
7.8
8.3
8.8
9.3
9.8
10.3
10.8
11.2
Linewidth (MHz) when I = 50 mA
0.2
Ith (mA)
24.0
26.0
28.0
30.0
32.0
34.0
36.0
38.0
40.0
42.0
44.0
46.0
Increasing kL
0.3
SMSR (dB)
0.97
1.10
1.23
1.36
1.49
1.62
1.75
1.88
2.00
Linewidth (MHz)
∆ν (MHz)
0.4
Power (mW)
(Optical thickness of d1)/λ
17
0.5
Stamp should be soft and flexible (to distribute
pressure and to accomodate particle impurities
and wafer non-flatness)
Stamp should have minimal lateral deformation
during imprinting (mostly for small features)
Applied pressure should be low and uniform,
despite varying imprinting area aspect ratio (avoid
breaking the wafer and avoid local deformation of
small size features – high local pressure)
No temperature cycling and stress (=> UV-NIL =>
stamp should be UV transparent)
Resist should have low viscosity – has to flow
almost freely (surface tension forces)
No bubbles should be present (vacuum process +
low viscosity + surface tension)
Small residual resist layer after imprint (residual
layer has to be removed => smallest feature size is
comparable with residual layer thickness)
Durability of master template ( contamination)
and stamps
CAS 2010
Tuesday, Oct. 12, O2
20
Three-layer NIL masks:
PDMS (polydimethylsiloxane), which is an UV transparent polymer
•
•
Hard PDMS (h-PDMS) is used for copying the nanopatterns from the master (does not deform –good replication of small features where the
local forces are very high - potential high deformation)
Soft PDMS (s-PDMS), which is very poor for replicating small, sub 300 nm, patterns, is used as a cushion to ensure stamp softness.
Three-layer sequence:
•
•
•
•
a thick s-PDMS cushion layer (softness, deformation, distribution of pressure),
a thin layer of glass (which preserves the stamp flexibility but prevents the lateral stretching)
a relatively thin nano-patterned h-PDMS layer.
This three-layer soft stamp (B) may be placed
on a relatively thick glass plate, generating
a hard stamp (A).
accommodates particles
and non-flat wafer
10 µm
CAS 2010
Tuesday, Oct. 12, O2
NIL imprinting
Improving the etching profile
21
Previous NIL masks produces thick residual
layers (100 nm), which limits grating resolution
New mask design with resist escape structures
allow a reduction of residual layers below 10 nm
below 10 nm features are imprintable
22
lateral leakage control
AFM
1µ
µm
K2281LD894, grating order 1; W = 1.5 µm, t = 0 nm
Coupling coefficient
of the 1st mode, κ [1/cm]
Imprint
Coupling c oefficient
of the 1st mode, κ [1/cm]
Template
DeLight New Dilute-Nitride:
W = 1.5 µ m D = 2.0 µm t = 0 nm
35
30
25
20
15
10
5
0
0
=> etching profile accuracy is
essential in achieving high κ
100
200
300
Triangle side length, x [nm]
for an ideal
etching profile
0,5
1,0
1,5
2,0
2,5
CAS 2010
Tuesday, Oct. 12, O2
23
Narrow trench etching experiments
K2281LD894 with W=1.5 µµm, D=0.5 µµm and t=0 nm
st
1 transverse mode
st
nd
Coupling coefficient
1 -order grating,
2 -order grating
rd
and
3 -order grating
90
80
70
Etch aspect
ratio
60
50
30
25
20
15
10
5
0
20
10
0
100
150
200
300
400
D=3.0 µm & ideal etching profile
rd
3 -order grating
with γ=½
30
50
100
Triangle side length, x [nm]
40
0
24
DeLight New Dilute-Nitride:
W = 1.5 µm D = 2.0 µ m t = 0 nm
35
0
> 10 < 10
200
250
300
350
400
Trench width, t [nm]
D=0.5 µm & ideal etching profile
450
Coupling coefficient, κ [1/cm]
Coupling coefficient, κ [1/cm]
Coupling coefficient
of the 1st mode, κ [1/cm]
23
100
3,0
Width of the lateral extension, D [µm]
The coupling coefficient improvement with extended lateral
gratings is derived mainly from improved etching profile
and less from the increased lateral extension itself.
CAS 2010
Tuesday, Oct. 12, O2
Higher κ by using wider lateral extension of
the gratings & narrow trench widths
400
55
50
45
40
35
30
25
20
15
10
5
0
0,0
K2281LD894 with W=1.5 µµm, D=3. 0 µµm and t=0 nm
st
1 transverse mode
140
130
120
110
100
90
80
70
60
50
40
30
20
10
0
Coupling coefficient
and
0
50
100
=> trench widths < 100 nm are desirable
irrespective of grating order
150
st
1 -order grating,
rd
3 -order grating
200
250
nd
2 -order grating
SEM views of gratings with period Λ=180 nm and trench widths of 140 nm (left panel),120 nm (middle
panel) and 100 nm (right panel). Gratings were etched to a depth of 1.2, 1.2 and 1.1 µm in GaAs/AlGaAs
300
350
400
450
High aspect-ratio-patterns with BCB
planarization:
-Better mechanical stability
-Less optical losses due scattering
-Higher bandwidth
Trench width, t [nm]
=> 100 nm trench widths are achievable
CAS 2010
Tuesday, Oct. 12, O2
CAS 2010
Tuesday, Oct. 12, O2
Experimental evaluation
of the coupling coefficient
Fabricated 894 nm LC-DFB laser
characteristics
25
26
-56
8
0,40
2,0
-62
-58
6
Power (mW)
Power (dBm)
Power (dBm)
7
-60
-60
-62
o
10 C
o
20 C
o
30 C
o
40 C
o
50 C
o
60 C
0
70 C
0
80 C
5
4
3
2
-64
891
892
893
894
Wavelength (nm)
-64
891
895
1
892
3rd-order grating with γ=0.5
893
Wavelength (nm)
894
895
0
0
5
10
Coupling coefficient (1/cm)
-54
Power (dBm)
-56
-58
-60
-62
891
892
893
Wavelength (nm)
894
Upper error limit
Fitted κ
Lower error limit
20
25
30
35
40
45
1,0
0,5
0,25
0,20
0,15
0,10
0,05
0,0
50
0,00
cm-1,
m = 3, γ = 0.5, κ = 22
L=550 μm, κL=1.21
(simulated / measured)
emission at 894.6 nm (tunable ± 2nm)
Ith ≈ 10 / 10 mA
output power (@ I bias = 90 < 100 mA) > 25 / ≈ 15 mW
single mode CW operation with SMSR > 40 / 35 dB
spectral linewidth (@ L=1000 µm) < 0.35 / ≈ 1.2 MHz
25
20
15
895
15
0,35
0,30
1,5
Current (mA)
30
Measurement (8.0mA)
Fitting by LAPAREX
0,45
2,5
Voltage (V)
-58
9
Measurement (7.6mA)
Fitting by LAPAREX
7,0
7,5
Current (mA)
8,0
Coupling coefficient extraction by fitting
simulated and experimental sub-threshold spectra
CAS 2010
Tuesday, Oct. 12, O2
CAS 2010
Tuesday, Oct. 12, O2
Further linewidth reduction
Linewidth evaluation @ 894 nm
27
100 MHz
AO MOD
υ
DET
υ +100 MHz
υ
Technion
100 MHz
Spectrum
analyzer
Fiber delay line length L
28
The spectral linewidth, Δν, of a laser due to spontaneous emission noise and carrier noise:
spectral linewidth simulation (@ L=1000 µm) < 0.35 MHz
DFB
Slope (W/A)
-54
Measurement (6.9mA)
Fitting by LAPAREX
Self-homoheterodyne
spectrum obtained with
a 2 km long fiber
(assuming Δν=1 MHz,
the coherent length in
fiber is ≈ 200 m)
Self-homoheterodyne experimental set up
∆ν =
ΓR sp'
4πN p
2
'
sp
qυ g n
(1 + α ) = Γ4Rπηqυ(I −g IV ) (1 + α ) = 4Γπη
(α
(I − I )
2
g
2
th act
eff
g
th sp
eff
i
th
i
th
i
(
+ α m ) 1 + α eff
2
2
) = 4πηqυ(I n− I
g
i
sp
th
)
(α i + α m )2 (1 + α eff 2 )
Decreasing αm is more straightforward than decreasing αi and αeff. However, the output power is
proportional to αm/(αi +αm).
Linewidth and light-current measurements suggest that αeff and αi are higher than evaluated in
simulations => significant linewidth reduction can be obtained by the reduction of αi with
epilayer and grating design and/or by the reduction of αeff with a phase-shifted DFB design
Short-delay self-homodyne experimental setup
Short-delay homodyne spectrum fitting (left panel) linewidth evaluation
at different bias levels (right panel) for a 1000 µm long LC-DFB laser
CAS 2010
Tuesday, Oct. 12, O2
αeff =6 was used in our linewidth simulations
CAS 2010
Tuesday, Oct. 12, O2
Acknowledgements
29
Development of Extremely Narrow-Band Semiconductor
Distributed Feedback Laser Technology (Narrow DFB)
Thank you!
CAS 2010
Tuesday, Oct. 12, O2