Chapter 4: Stimulated emission and LASERS

Stimulated Emission
Stimulated emission and photon amplification
Stimulated emission rate and Einstein coefficients
Light amplification
Rate equations for amplifiers
Laser Oscillation Condition
Optical threshold gain
Output characteristics
Semiconductor laser diode
Principles
Gain spectrum under forward bias
Homojunction laser diode
Heterojunction laser diode
Laser diode layout and lateral mode confinement
Laser diode characteristics
Steady state semiconductor rate equations
Optimal out-coupling
Distributed bragg reflection for single mode lasers
Gain spectrum in QW lasers
Low dimensional structures & threshold current reduction
Vertical cavity surface emitting lasers
Late News: Ge electrically pumped laser

TFH

SS 2012

92

Stimulated emission and photon amplification

I

Photon absorption excites electron

from E1 E2 .

Allowed process, if energy (in

semiconductor: and momentum) is
conserved (~Ph = hPh = E2 E1 ).

Two possibilities for de-excitation:

i) Spontaneous emission: Eventually, also the electrons in an isolated atom will return to
their ground state spontaneously, without external action (purely QM process), both
radiative (photon emission hPh = E2 E1 ) and non-radiative transitions [phonon
emission (lattice vibration) etc.] can occur.
ii) Stimulated emission: The interaction of the atoms excited electron with an EM
radiation in resonance with the de-excitation energy E2 E1 triggers the de-excitation.
A photon coherent with the incoming EM field (in phase and with same propagation
direction relative to incoming photons) is emitted [inverse process to (stimulated)
absorption]. Not considered until 1916 (Einstein), but required for an optically
pumped system to reach stationarity, since absorption rate depends on intensity, but
spontaneous emission rate not.

TFH

Stimulated emission is the base process for light amplification.

Shown later: Light amplification requires more electrons in the excited state than in
the ground state (population inversion).
SS 2012

93

Stimulated emission rate and Einstein coefficients

I

N [cm3 ] two level system in thermal equilibrium with

radiation in black body (BB).

2
1

N1 in groundstate @ E1 , N2 in excited state @ E2 ,

N = N1 + N2 .

2
1

2
1

Stimulated excitation (absorption) rate R12 proportional to

N1 , to the number of photons with h21 = E2 E1 per
volume {i.e. to the energy density (h21 ), [] =Jsm3 }:

2
1

2
1

R12 = B12 N1 (h21 ), [B12 ] =m3 J1 s2

spont
stim
De-excitation rate R21 = R21
+ R21
I
I

stim
R21

analog to absorption:
= B21 N2 (h21 )
spont
R21
= A21 N2 , independent of number of photons, only
prop. to N2 .

R21 = A21 N2 + B21 N2 (h21 )

Above equations define Einstein coefficients B12 , B21 , A21

as proportionality factors for the rates.
I

2
1

2
1
2
1

2
1

2
1
2
1

2
1

2
1

2
1

2
1
2
1

Thermal equilibrium:
a) R12 = R21 .
b) Boltzmann statistics: N2 /N1 = e(E2 E1 )/kB T = eh21 /kB T .
c) Photon energy density in BB given by Plancks law: eq (h) =

TFH

SS 2012

8h 3
c3

eh/kB T

94

Stimulated emission rate and Einstein coefficients

I

Three equations for the three Einstein coefficients.

Divide a) by N1 and solve for N1 /N2 :

B12 (h21 )
N2
=
N1
A21 + B21 (h21 )
Using Boltzmann statistics b) results in:
B12 (h21 )
A21 /B21
e21 /kB T =
(h21 ) =
A21 + B21 (h21 )
(B12 /B21 )e21 /kB T 1
Comparision with Plancks law c): Equal for all T only, if

B12 = B21 B and

I

Interpretation:
I

I
I

TFH

3
8h21
A21
(= h Photon DOS)
=
B21
c3

Stimulated emission and absorption are inverse processes and occur with equal
probability per available initial state (2 for stim. emission, 1 for absorption)
population inversion (N2 > N1 ) can never be achieved by intense optical pumping a
two level sytem, at most transparancy (N1 = N2 ) is achieved.
Spontaneous emission prop. and stimulated emission prop. are related, ratio
increases with 3 , harder to construct x-ray lasers.
stim
R21
spon
R21

B
(h)
A

(h)
1
h Photon DOS

Ratio increases with:

i) # photons available optical resonator
ii) decreasing density of photon states optical resonator
SS 2012

95

Light amplification
= I I (x) = I0 ex exponential decay of intensity I .

Absorption law:

For expressing via electron transition rates, we convert intensity I into photon
flux , photon density nPh and : I = h = nPh ch = c(h).

I dI
dx
I
I

TFH

dI
dx

dI
dx

= h d
= ch
dx

dnph
dx

= h

dnph
dt

= h(R12 + R21 ) = (N1 N2 )hB(h).

hB
= (N1 N2 ) c
I

negative for population inversion (N2 > N1 ) exponentially growing intensity,

i.e. gain.
hB
gain coefficient g = (N2 N1 ) nref
c0

SS 2012

96

Rate equations for amplifiers

I

Which pump-rate is required to achieve

gain (i.e. population inversion)?

ideal pumping scheme involves 4 levels

Spontaneous transition rates lumped

together and are expressed as inverse
spon
1
lifetimes (i.e. R21
= 21,spon
)

1
1
1
1
1
example: spontaneous lifetime of level 2: 21 = 21
+ 20
= 21,sp
+ 21,nr
+ + 20

Rate equations [Wi = B(h12 )]:

dN3
N3
=R

dt
32
dN2
N3
N2
=

N2 Wi + N1 Wi
dt
32
2
dN1
N1
N2
=

+ N2 Wi N1 Wi
dt
21
1
dN0
N1
N2
= R +
+
dt
1
20

TFH

I 32  3x N3 = R32
I stationary solution

N2
dN2
=R
N2 Wi + N1 Wi
dt
2

d
=0:
dt

R2 (1 1 /21 )
N 0

1 + Wi [2 + 1 (1 2 /21 )]
1 + Wi s
N 0 : pop. inv. in the absence of amplifier radiation.
s : saturation time.

I N N2 N1 =

Non-linear gain , gain saturation.

typically: nonr. 2 1 decay rate negligible (21 = 21,sp ) &

& 20  21,sp  1 , thus: N 0 R21,sp , s 21,sp .

Attention: for strong pumping, R not independent of N because

N0 + N1 + N2 + N3 = Na (atom density) and N3 N1 0 N0 Na N2 Na N
SS 2012

97

Laser oscillation condition

Optical threshold gain

gain coefficient (stim. em.)

B21 nref h21
g = (N2 N1 )
c0
loss:

abs : absorption: impurities in laser medium, non-inverted allowed transitions with similar
transition energy, free carriers (semiconductors !!)
sc : scattering out of resonator mode: defects and inhomogeneities.
Ri : Out-coupling losses, power reflection coefficient of
. cavity mirrors R1 , R2 / 1
!

stationarity conditions: Pf = Pi
Pf = Pi R1 R2 e2Lgth e2L(abs +sc )

z }| { 1
gth = abs + sc + 2L
ln( R11R2 ) threshold gain.
c0
threshold pop. inv.
B21 nref h0
0
Increase R (i.e N ): g remains clamped at gth
after stationary is re-established N = Nth
Nth = gth

TFH

SS 2012

N/N 0 = g/g0

1.0

laser turn-on
time

0.8
0.6
0.4

steady state
loss=gth

ph. flux
density
0.0
102 101 100 101
Wi s = /s
0.2

102

98

Laser oscillation condition

Output characteristics

Gain clamping: N = Nth

N 0

1 + Wi s


Wi (h12 )
1 N 0
1

Wi =
s Nth


0

s N 1 , N 0 > N th
th
N
=

0,
N 0 N th

Nth =

Simplified description of a laser oscillator.

(N2 N1 ) and coherent output power P0
vs. pump rate R under continuous wave
steady state operation.

steady-state laser-internal
photon-flux density

TFH

Spontaneous emission neglected.

In addition: stationarity of phase

Fabry-Perot resonator modes.

SS 2012

99

PrinciplesoftheLaserDiode

pn junctioninaDegenerate
Semiconductor

Eg

Appliedbiasdiminishesthebuildin
potentialbarrier
Depletionregionisnolongerdepleted
Therearenowmoreelectronsinthe
conductionbandthaninthevalance
bandnearEv Populationinversion

Junction

n+

Ev
EF p

Ho les in V B
Electro ns

Electro ns in C B

n+

p+

eVo

Ec
EF n
Ec

EF n

In v ersio n
reg io n

Ec

Eg

eV
EF p

ChangeintheFermilevelisthework
donebytheappliedvoltage,eV

Ifthejunctionisforwardbiased
suchthatEfn Efp =eV >Eg

TFH

p+
Ec

Whenvoltageisapplied

FermilevelinthepregionisbelowEv
FemilevelinthenregionisaboveEc
Withnoappliedvoltage,Efn=Efp yields
averynarrowdepletionregion
Thereisapotentialenergybarrier,eVo
thatpreventsnsideelectronsfrom
diffusingtothepsideandviceversa

(a)

Ev

(b)

V
The energy band diagram of a degenerately dopedp-n with no bias. (b) Band
diagram with a sufficiently large forward bias to cause population inversion and
hence stimulated emission.
1999 S.O. Kasap, Optoelectronics (Prentice Hall)

SS 2012

100

PrinciplesoftheLaserDiode

Populationinversionregionisa
layeralongthejunctioncalledthe
inversionlayer(activeregion)
Anincomingphotonwithenergy
EcEv cannotexciteanelectronin
Ev toEc astherearehardlyany
presentinthevalanceband
withintheactiveregion
Hencethereismorestimulated
emissionthanabsorption
Theopticalgainpresentinthe
activeregionduetolackof
probabilityofvalanceelectron
absorption

Energy
Optical gain

EF n EF p

CB
EF n
Ec

Electrons
in CB

eV

Eg

Holes in VB
= Empty states

Ev
EF p

At T > 0

VB

At T = 0

Optical absorption
Density of states
(a)

(b)

(a) The density of states and energy distribution of electrons and holes in
the conduction and valence bands respectively at T 0 in the SCL
under forward bias such that E Fn E Fp > E g. Holes in the VB are empty
states. (b) Gain vs. photon energy.
1999 S.O. Kasap,Optoelectronics (Prentice Hall)

TFH

SS 2012

101

Gain spectrum under forward bias

Evolution of the absorption and
gain curves as a function if the
position of the quasi-Fermi level.
The gray (dark) curves
correspond t oa small (large)
displacement from equilibrium.
In this case, the medium absorbs
all photons having energies in
excess of the bandgap. Once the
energy separation between the
two quasi-Fermi levels exceeds
the bandgap, all photons with
energies between Eg and
EF ,c EF ,v are amplified.

From E. Rosencher, B. Vinter,

Optoelectronics, Cambridge University
Press (2002)
TFH

SS 2012

102

Gain spectrum under forward bias

From A. Yariv, Quantum Electronics, 3rd edition, John Wiley & Sons, (1989)

TFH

SS 2012

103

ElectroopticalPerformanceofIII/VDiodes
Current

injectionpumping:Opticalpumpingisachievedbyforward
diodecurrentandthepumpingenergyisanexternalbattery
Forlaserwealsoneedanopticalresonatorcavity.Thisis
achievedthroughtheuseofaslabwaveguidewithahigh
n
1
indexcontrastattheemissionend R = ( nGaAs +1 )2 35%
GaAs
Wavelengthoftheradiationthatcanbuildupinthecavity
dependsonthelength(L)inhalfmultiples

2n

Electrode

GaAs
L

n+

GaAs
Electrode
Active region
(stimulated emission region)

Optical P ower

Laser

A schematic illustration of a GaAs homojunction laser

diode. The cleaved surfaces act as reflecting mirrors.

LED

1999 S.O. Kasap, Optoelectronics (Prentice Hall)

Stimulated
emission

Optical P ower

Laser

Spontaneous
emission

I
It h

Typical output optical power vs. diode current ( I) characteristics and the corresponding
output spectrum of a laser diode.
1999 S.O. Kasap, Optoelectronics (Prentice Hall)

TFH

L
p+

=L

Optical Power

Optical P ower

Cleaved surface mirror

2criticalcurrentidentifiers
Transparency current:Currentabove
whichnonetphotonabsorption
occurs

Threshold current:currentabove
whichopticalgainovercomesall
photonlossesinthecavity
typically: Jth = 500 Amm2 for

homojunction LD !!
SS 2012

104

Heterojunction LaserDiodes

Mainissuewithhomojunction diodesis
thatthelaserthresholdcurrentdensityis
toohighforpracticaluses.

TFH

Ex.Jth =500A/mm2 forGaAs at300K

Heterostructured diodesreducethese
currentdensitiesbyordersofmagnitude
Thisisachievedthroughacombinationof
carrierconfinement(mismatched
materials),andphotonconfinement
(geometricshapeofthewaveguide)
Doubleheterojunction (DH)deviceswith
npp layersallowfordesigned
confinementoftheactiveregion
LowerrefractiveindexoftheAlGaAs
enhancesthemodeconfinementin
comparisontoahomoorsimple
heterojunction device
Significantlyreducesthresholdcurrent
density

n
AlGaAs

(a)

GaAs

AlGaAs

(a) A double
heterostructure diode has
two junctions which are
between two different
bandgap semiconductors
(GaAs and AlGaAs).

(~0.1 m)
Electrons in CB

Ec

Ec

Ec

2 eV
1.4 eV

2 eV

(b)

Ev

Ev
Holes in VB

Refractive
index
(c)

Photon
density

Active
region

n ~ 5%

(d)

(b) Simplified energy

band diagram under a
large forward bias.
Lasing recombination
takes place in the pGaAs layer, the
active layer
(c) Higher bandgap
materials have a
lower refractive
index

(d) AlGaAs layers

provide lateral optical
confinement.

1999 S.O. Kasap, Optoelectronics (Prentice Hall)

SS 2012

105

Figure 15.10 A typical double heterostructure GaAs/GaAlAs laser. Electrons and holes are injected into the active GaAs layer from the
n and p GaAlAs. Photons with frequencies near =Eg/h are amplified by stimulating electronhole recombination.
Cc2007Photonics,6theditionYariv andYeh (OxfordUniversityPress

Photonics,6theditionYariv andYeh

TFH

SS 2012

106

Figure 15.12 The magnitude of the energy gap in Ga1x Alx As as a function of the molar fraction x. For x > 0.37 the bandgap is indirect.
(After Reference [11].)

Photonics,6theditionYariv
andYeh

TFH

11

SS 2012

Cc2007Photonics,6theditionYariv andYeh (OxfordUniversityPress

107

Figure 15.15 IIIVcompounds:latticeconstantsversusenergybandgaps andcorrespondingwavelengths.Thesolidlinescorrespondtodirect

gapmaterialsandthedashedlinestoindirectgapmaterials.Thebinarycompoundsubstratesthatcanbeusedforlatticematchedgrowthare
indicatedontheright.(AfterReference[11].)

Photonics,6theditionYariv
andYeh

TFH

12

SS 2012

Cc2007Photonics,6theditionYariv andYeh (OxfordUniversityPress

108

Heterojunctions Laser
Heterojunction diode: different materials for n & p
Different materials: significantly different index n
Also different lattice constants
Important point: want the lattice matched at layer boundary
Use mixed alloy: eg GaAs and AlAs
AlxGa1-1As
x = mole fraction of Aluminum
1-x = mole fraction of Gallium

Heterojunctions Laser
Single Heterojunctions: one sided confinement
p-GaAlAs: p-GaAs: n-GaAs
Better confinement means lower threshold current for lasing
Thus operates in pulsed mode at room temperature
Double Heterojunction lasers: confines both top & bottom
p-GaAlAs: GaAs: n-GaAlAs: n-GaAs

Double Heterojunctions Laser

Has both Band and Index steps on both top & bottom
Doubly confines light: creates a waveguide as cavity
Requires much less threshold current
Thus CW operation now possible at room temperature

Comparison of Homo/Hetero/D-Heterojunctions Lasers

As add index steps get smaller light spreading
Single hetrojunction threshold current ~5x < homojunction
Double hetrojunction threshold ~50-100x <homojunction
Less current, less heating, more output before thermal limitations

Heterojunctions with Waveguides

Buried heterojunction:
Surrounded both vertical & horizontal by lower material
1-2 microns wide: high efficiency, low threshold
Channeled Substrate
Etch channel in substrate: isolate active area
Low loss
Buried Crescent
Fill grove to get crescent shaped active strip

Heterojunctions with Waveguides

Ridge Waveguide
Etch away a mesa around active region
confines current flow to 2-3 micron strip
Double-channel planar buried heterostructure
Isolate active with mesa, then fill with lower index
used with very high power InGaAsP lasers

EdgeEmittingLasers

VerysimilartoELEDdevicespresentedinchapter
3
AdditionalcontactinglayerofpGaAs nexttothe
p=AlGaAs providesbettercontactingandavoids
Schottky junctionwhichwouldlimitthecurrent
inthedevice.
p andn AlGaAs layersprovidecarrierand
opticalconfinementintheverticaldirection
LaseremissionintheactivepGaAs(oradifferent
AlGaAs constitution)regionisbetween870900
nmdependingondoping.

Cleaved reflecting surface

W

Stripe electrode

TFH

AlGaAs andGaAs havenegligiblelattice

mismatchyieldingveryfewdefectsin
thecrystalthatwouldleadtoexcessive
thresholdcurrents
Also,thestripe electrodeacrossthetop
confinestheelectricfieldandthusthe
opticallyactiveregionproviding
additionalgeometricalconfinement
Suchlasers arecalledgainguided,b/c
thecurrentdensitygeneratedisguided
bytheelectricfieldbetweenthestripe
electrodeandthebottomelectrode

Oxide insulator
p -GaAs (Contacting layer)
p -Al x Ga 1- x As (Confining layer)
p -GaAs (Active layer)
n -Al x Ga 1- x As (Confining layer)
n -GaAs (Substrate)

2
1
Current
paths
Substrate

Elliptical
laser
beam

Substrate
Electrode

Cleaved reflecting surface

Active region where
(Emission region)

J > J t h.

Schematic illustration of the the structure of a double heterojunction stripe

contact laser diode
1999 S.O. Kasap,

SS 2012

Optoelectronics

(Prentice Hall)

115

BuriedHeterostructure LDs

Althoughthestripeelectrodegeometryprovides
somegeometricconfinement,itismore
advantageoustorestrictlateralgeometryphysically
throughtheuseofconfininglayersalongtheside
ofthediode
Creationofaopticalwaveguideinbothverticaland
horizontaldirectionsaidsinreducingopticalcavity
modesandpromotesconfinement
Significantlyreducescurrentdensityrequiredfor
stimulatedemission

FigurefromChapter15Photonics,6theditionYariv andYeh 2007OxfordUniversityPress

TFH

SS 2012

116

ElementaryLaserDiode(LD)Characteristics

Longitudinalmode:lengthdetermined
Lateralmode:widthdetermined
Emissioniseithermultimodeorsingle
modedependingontheopticalresonating
structureandthepumpingcurrentlevel

Relative optical power

Po = 5 mW

Dielectric mirror

P o = 3 mW

Fabry-Perot cavity

P o = 1 mW
(nm)

Length, L
Height, H

TFH

Width W

778

Diffraction
limited laser
beam

780

782

The laser cavity definitions and the output laser beam

characteristics.

Output spectra of lasing emission from an index guided LD.

At sufficiently high diode currents corresponding to high
optical power, the operation becomes single mode. (Note:
Relative power scale applies to each spectrum individually and
not between spectra)

1999 S.O. Kasap, Optoelectronics (Prentice Hall)

1999 S.O. Kasap, Optoelectronics (Prentice Hall)

SS 2012

117

LaserPerformanceasaFunctionofTemperature

Outputspectrumandmodepropertiesare
temperaturedependant
SinglemodeLDsexhibitamodehopatcertain
temperaturescorrespondingtoachangein
peakemissionwavelength
Nominalwavelengthofthelaserincreases
slowlybetweenhopsduetochangein
refractiveindex,n,withtemperature
Slopeefficiencydetermineslaserefficiency
andisnotthesameasLEDconversion
efficiencystatedinchapter3.
slope =

TFH

P o (mW)
10

0 C
25 C

50 C

6
4
2

I (mA)

0
0

20

40

60

80

Output optical power vs. diode current as three different temperatures. The
threshold current shifts to higher temperatures.

Po
I I th

SS 2012

1999 S.O. Kasap, Optoelectronics (Prentice Hall)

118

Steady state semiconductor rate equations

I
edLW

Current (I) pump rate: R =

ph
, where ph is
Photon loss rate: dtph = ph
average time for a photon to be lost from
lasing cavity mode due to transmission at end facets,
scattering and absorption in the semiconductor
Use rate equ. results (p. 97 f):

dN

assume a) 2 = 21 = 21,sp ; b) 1 0 N1 0

N0
1+Wi s

R21,sp
1+Wi 21,sp

; in particular N th N2th = R th 21,sp


2
Stimulated emission rate: dN
= N2 Wi
dt st

th
th
dNph
Nph
dN2th 
1
th N th =
, where
at threshold: dt  = dt = N2th With = N2th CNph
2
C ph
ph

N =
I

Sketch of BHJ laser diode layer structure

N2

st

C = Bh21 / and B is the Einstein coefficient.

I

Jth = Ith /LW = ed R th =ed/ C ph 21,sp

gain clamping: N2 = N2th

Nph =
I

( RRth 1) =

Jth ph J
( Jth
ed

1) =

= R th 21,sp

ph
(J
ed

Jth )

Half of the cavity photons move towards out-couple mirror, a fraction (1 Rm ) Tm

escapes during the transversal time in this direction

Pout =
TFH

1
C 21,sp

R21,sp
1+CNph 21,sp

(0.5Nph )(Cavity Volume) (Photon energy)

(1
time for photon to transverse cavity length

Rmirror ) =

SS 2012

hc02 ph W (1Rmirror )
2enref

(J Jth )
119

Optimal out-coupling I
I

Tm = 1 Rm transmission coefficient

Pout Nph Tm Tm ( RRth 1) = Tm (R C ph 21,sp 1)

Photon lifetime Ph includes out-coupling losses, i.e Ph = Ph (Tm )

each photon escapes with probability T within round-trip time 2nL/c0 .



dNph 
dNph 
dNph
=

+
=

dt
dt
dt 

loss

gain

1/ph (Tm )


dNph 
dt 

loss,out


dN 
+ dtph 

loss,else


dN 
+ dtph 

gain

}|
{
1
Tm c0
= [ (
+
) +N2 C ]Nph
2nL
loss

maximize Pout with respect to Tm :

dPout
d
2nLloss

[Tm (R C 21,sp
1)] 0
dTm
dTm
Tm c0 loss + 2nL
v
u 2Ln
p
2Ln
2Ln
Tmopt =
+u
Floss,int + g0 Floss,int
C R21,sp
t
| {z } c0 loss
c0 loss
c0
N 0 N20

TFH

Floss,int internal loss fraction (probability) per round trip.

g0 open-loop photon multiplication fractor per round trip.

SS 2012

120

Optimal out-coupling II

From A. Yariv, Quantum Electronics, 3rd edition, John Wiley & Sons, (1989)

TFH

SS 2012

121

DistributedBraggReflection
forSingleModeLasers

Ensuresinglemoderadiationinthelasercavityistousefrequencyselectivemirrorsatthe
cleavedsurfaces
DistributedBraggreflectorisamirrorthathasbeendesignedareflectiveBragggrating
Reflectedwaveoccursonlywhenthewavelengthcorrespondstotwicethecorrugation
periodicity,.
Thediffractionorderofthereflectorisinteger,q=0,1,2,
B
n
N=refractiveindexofthemirror
BraggwavelengthofthemirroroutputisB

Corrugated
dielectric structure

= 2

q(B /2n) =

A
B
Active layer

Distributed Bragg
reflector

(a)

(b)

(a) Distributed Bragg reflection (DBR) laser principle. (b) Partially reflected waves
at the corrugations can only constitute a reflected wave when the wavelength
satisfies the Bragg condition. Reflected waves A and B interfere constructive when
q(B/2n) = .
1999 S.O. Kasap, Optoelectronics (Prentice Hall)

TFH

SS 2012

122

DistributedFeedbackLaser(DFB)

InaDFBlaser,thecorrugatinggratingiscalledtheguidinglayerandrestontopofthe
activelayer.
ThepitchofthecorrugationprovidesopticalgainattheBraggwavelength,B.
Travelingwavesareexcitedbytheactivelayerandcoupletotheguidinglayerasthey
reflectbackandforthacrossthegratingtogenerateallowedDFBmodesthatarenot
exactlymatchedtotheBraggwavelength,butareplacedsymmetricallyjustofftheideal
modeoftheguidinglayeratm.

m = B

Corrugated grating

B 2
2nL

(m + 1)
Ideal lasing emission

Optical power

Guiding layer
Active layer

0.1 nm

(a)

(b)

(nm)
(c)

(a) Distributed feedback (DFB) laser structure. (b) Ideal lasing emission output. (c)
Typical output spectrum from a DFB laser.
1999 S.O. Kasap, Optoelectronics (Prentice Hall)

TFH

SS 2012

123

CleavedCoupledCavityLaser

DevicehastwodifferentopticalcavitiesoflengthLandD.
Eachlasercavityispumpedbyadifferentcurrent
Onlymodesresonantinbothcavitiesareallowedtoresonatethroughtheentiredevice,
allowingtheengineertotuneoutcertainmodesfromoneorbothindependentlaser
diodes
Whypumpboththecavities?Ans.Allowedmodesinanunpumped cavitywillundergo
recombinationifthedeviceisnotdriven.

Cavity Modes
In L
Active
layer

D
(a)

In D

In both
L and D

(b)

Cleaved-coupled-cavity (C3) laser

1999 S.O. Kasap, Optoelectronics (Prentice Hall)

TFH

SS 2012

124

QuantumWell(QW)Devices

Devicewithanultrathin(50nm)narrowbandgap activeregionbetweentwowiderbandgap
semiconductors
AssumethatinQWdevicesthatthelatticematchsothatallthesemiconductorshavethesame
latticeconstantasothatcrystallinedefectsareminimized
Badgap changesattheinterfacearethereforeonlyduetodiscontinuitiesbetweenEc andEv of
thedifferingmaterialsyieldingdiscreteallowablequantumstatesthatcanbesolvedasparticle
inaboxtypeproblems.

E
Dy
E

AlGaAs

n=1

Eg2

QW
Ev

2 2me
[E V ( x ) ] = 0
+
x 2
h

Eg1

y
x

Energyinaquantumwell

Bulk

E3
E2
E1

Ec n = 2

d
D z AlGaAs

QW

Ec

Ev

(a)

Bulk

g(E)
Density of states

GaAs

(b)

(c)

A quantum well (QW) device. (a) Schematic illustration of a quantum well (QW) structure in which a
thin layer of GaAs is sandwiched between two wider bandgap semiconductors (AlGaAs). (b) The
conduction electrons in the GaAs layer are confined (by Ec) in the x-direction to a small length d so
that their energy is quantized. (c) The density of states of a two-dimensional QW. The density of states
is constant at each quantized energy level.

E = Ec +

h 2 n 2y
h 2 n 2z
h2n2
+
+
8 m e* d 2 8 m e* D y2 8 m e* D z2

n , n y , n z = 1, 2 ,3,...

Note:potentialenergybarrierofthe
conductionbandisdefinedbyw.r.t.Ec

1999 S.O. Kasap, Optoelectronics (Prentice Hall)

TFH

SS 2012

125

EnergySpectruminaQuantumWell(SQW)
Ec
E
E1
h = E1 E1
E1
Ev

In single quantum well (SQW) lasers electrons are

injected by the forward current into the thin GaAs
layer which serves as the active layer. Population
inversion between E1 and E1 is reached even with a
small forward current which results in stimulated
emissions.
1999 S.O. Kasap, Optoelectronics (Prentice Hall)

TFH

SS 2012

126

Example:AGaAs QW

GaAs QW
Effectiveelectronmassisme*=0.07me
WhatarethefirsttwoelectronenergylevelsforaQW
ofthickness10nm?
WhatistheholeenergybelowEv iftheeffective
electronmassofthehole,mh*=.5me?
Whatistheemissionwavelengthw.r.t.bulkGaAs
whichasanenergybandgap of1.42eV?

n =

h2n2
= 0 .0537 eV
8 m e* d 2

n' =

h 2n 2
= 0 .0075 eV
8 m h* d 2

g =

hc 1240 eV nm
=
= 874 nm
Eg
1 .42 eV

QW

TFH

Figure 16.1 Thelayeredstructureandthebandedges ofa

GaAlAs/GaAs/GaAlAs quantumwell.
Cc2007Photonics,6theditionYariv andYeh (OxfordUniversityPress

1240 eV nm
=
=
= 839 nm
(1 .42 + 0 .0527 + 0 .0075 )eV
E g + n + n'
hc

Differenceinemissionwavelengthbetweenabulk
GaAs LDandaQWLDis35nm

SS 2012

127

DensityofStatesfromourParticleina
BoxSolution(QWDevice)

FigurefromChapter16Photonics,6theditionYariv andYeh 2007OxfordUniversityPress

TFH

SS 2012

128

Gain spectrum in QW lasers

Gain in a QW laser:
a) The Fermi inversion
fc (n, ~) fv (n, ~) at two carrier
densities n2 > n1 for QW thickness
Lz = 200
A.
b) The gain vs. ~ at n1 and n2 .
c) The same as in a) for narrower QW
(Lz = 100
A).
d) The same as in b) for narrower QW
(Lz = 100
A).
e) The same as in a) for a bulk
semiconductor.
f) The same as in b) for a bulk
semiconductor.
The energy Ef in a) and c) corresponds to
the photon energy ~ for which fc fv = 0
which is the transparency condition

TFH

SS 2012

129

DOSDifferentQuantumWellsTypes
for 0,1,2 dimensional quantum
BasedonGeometry
confined structures

ImagefromChapter16,ofFundamentalsofPhotonics,2nd ed.BySaleh andTeich cc2007WileyInterscience

TFH

SS 2012

130

Low dimensional structures & threshold current reduction

from Z. Alferov, Nobel lecture Dec. 2000

TFH

SS 2012

131

VerticalCavitySurfaceEmittingLasers(VCSELs)

Alternatinglayersoflowandhighindexaboveandbelow
theQWregioncreatesadistributedBraggreflectorof
dielectricmirrors
Themirrorsareneededtomatchtheopticalgainlostby
theshortcavitylength.Thuswiththemirrorsthelight
passesthroughthecavitysome2030timestoobtaina
desiredreflectanceof99%
Thehighreflectanceincreasesthegeometriccomponent
ofthegainrequiredforlaseremission

Contact
/4n 2
/4n 1

Dielectric mirror

Active layer
Dielectric mirror

Substrate
Contact

Surface emission

A simplified schematic illustration of a vertical cavity

surface emitting laser (VCSEL).
1999 S.O. Kasap, Optoelectronics (Prentice Hall)

Figure 16.14 Thefielddistributionofthelasermodeinsideavertical

cavitylaserwithL=/nwiththreequantumwells.Notethe
evanescentdecayofthefieldenvelopeinsidetheBraggmirrorsand
theconstantamplitudestandingwavebetweenthemirrors.

Constructiveinterferenceofpartiallyreflectedwaves
Ofwavelength,,attheinterface

n1d1 + n2 d 2 =

Cc2007Photonics,6theditionYariv andYeh (OxfordUniversityPress

TFH

SS 2012

132

VCSELAttributes

TFH

VCSELactivelayersaregenerallyverythin
0.1umandcomprisedofMQWforimprovedthresholdcurrent
Thedeviceiscomprisedofepitaxially depositedlayeronasuitablesubstratewhichistransparentinthe
emissionwavelength
Ex.980nmVCSELdevices
InGaAs istheactivelayer
GaAs isthesubstrate
AlGaAs withdifferentcompositionscomprisethedielectricmirrorstack
ThetopstackisthenetchedafterallthelayershavebeendepositedtocreatetheinvertedTshape
presentedinthepreviousslide
Inpractice,currentflowingthroughthedielectricmirrorsgivesrisetoanundesiredvoltagedropthat
makesthedeviceVERYsensitivetofailurefromelectrostaticdischarge.Infact,thisisthemostcommon
failuremodeduringVCSELoperationandinstallation.
Theverticalcavityandthustheemittedbeamisgenerallycircularincrosssection
Theheightoftheverticalcavityisseveralmicrons.Thusthelongitudinalmodeseparationissufficiently
largetoallowonlyonemodeofoperation.Howeverlateralmodesmaybepresentincertaincavity
geometries
InpracticeVCSELShaveseverallateralmodesbutthespectralwidthisonlynmwhichissubstantially
lessthanthelongitudinalmodesofaDFBorELD.
Also,VCSELShaveanaveragebeamdivergenceofabout812o dependingontheirfabricationand
materialsused
DualwavelengthVCSELemissionisobtainedbyoperatingathighcurrents.

SS 2012

133

Late News: Ge electrically pumped laser

An electrically pumped germanium laser
Rodolfo E. Camacho-Aguilera,1 Yan Cai,1 Neil Patel,1 Jonathan T. Bessette,1
Marco Romagnoli,1,2 Lionel C. Kimerling,1 and Jurgen Michel1,*
1

Massachusetts Institute of Technology, 77 Massachusetts Ave., Cambridge, MA 02139, USA

2
PhotonIC Corporation, 5800 Uplander Way, Los Angeles, CA 90230, USA
*jmichel@mit.edu

Abstract: Electrically pumped lasing from Germanium-on-Silicon pnn

heterojunction diode structures is demonstrated. Room temperature
multimode laser with 1mW output power is measured. Phosphorous doping
in Germanium at a concentration over 4x1019cm3 is achieved. A
Germanium gain spectrum of nearly 200nm is observed.
2012 Optical Society of America
OCIS codes: (140.2020) Diode lasers; (140.3380) Laser materials; (140.5960) Semiconductor
lasers; (160.3130) Integrated optics materials.

References and links

1.
2.

3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.

T.-H.Cheng et al., Appl. Phys.

Lett. 96, 211108 (2010).

15.

D. J. Lockwood and L. Pavesi, Silicon Photonics (Springer-Verlag, 2004).

M. E. Groenert, C. W. Leitz, A. J. Pitera, V. Yang, H. Lee, R. Ram, and E. A. Fitzgerald, Monolithic integration
of room-temperature cw GaAs/AlGaAs lasers on Si substrates via relaxed graded GeSi buffer layers, J. Appl.
Phys. 93(1), 362367 (2003).
H. Park, A. Fang, S. Kodama, and J. Bowers, Hybrid silicon evanescent laser fabricated with a silicon
waveguide and III-V offset quantum wells, Opt. Express 13(23), 94609464 (2005).
J. Liu, X. Sun, D. Pan, X. Wang, L. C. Kimerling, T. L. Koch, and J. Michel, Tensile-strained, n-type Ge as a
gain medium for monolithic laser integration on Si, Opt. Express 15(18), 1127211277 (2007).
J. Liu, X. Sun, Y. Bai, K. E. Lee, E. A. Fitzgerald, L. C. Kimerling, and J. Michel, Efficient above-band-gap
light emission in germanium, Chin. Opt. Lett. 7(4), 271273 (2009).
J. Liu, X. Sun, R. Camacho-Aguilera, L. C. Kimerling, and J. Michel, Ge-on-Si laser operating at room
temperature, Opt. Lett. 35(5), 679681 (2010).
G. Shambat, S.-L. Cheng, J. Lu, Y. Nishi, and J. Vuckovic, Direct band Ge photoluminescence near 1.6 m
coupled to Ge-on-Si microdisk resonators, Appl. Phys. Lett. 97(24), 241102 (2010).
S.-L. Cheng, J. Lu, G. Shambat, H.-Y. Yu, K. Saraswat, J. Vuckovic, and Y. Nishi, Room temperature 1.6 m
electroluminescence from Ge light emitting diode on Si substrate, Opt. Express 17(12), 1001910024 (2009).
M. O. E. Kasper, T Aguirov, J. Werner, M. Kittler, J. Schulze, Room temperature direct band gap emission
from Ge p-i-n heterojunction photodiodes, in Proceedings of Group IV Photonics 2010 (2010).
X. Sun, J. Liu, L. C. Kimerling, and J. Michel, Room-temperature direct bandgap electroluminesence from Geon-Si light-emitting diodes, Opt. Lett. 34(8), 11981200 (2009).
J. Liu, X. Sun, L. C. Kimerling, and J. Michel, Direct-gap optical gain of Ge on Si at room temperature, Opt.
Lett. 34(11), 17381740 (2009).
R. E. Camacho-Aguilera, Y. Cai, J. T. Bessette, D. Kita, L. C. Kimerling, and J. Michel, High active carrier
concentration in n-type, thin film Ge using delta-doping, submitted for publication (2012).
G. Scappucci, G. Capellini, W. M. Klesse, and M. Y. Simmons, Phosphorus atomic layer doping of germanium
by the stacking of multiple layers, Nanotechnology 22(37), 375203 (2011).
R. E. Camacho-Aguilera, Y. Cai, J. T. Bessette, L. C. Kimerling, and J. Michel, Electroluminescence of highly
doped Ge pnn diodes for Si integrated lasers, Proc. 8th IEEE Intern. Conf. GFP, Vol. 190,
10.1109/GROUP1104.2011.6053759 (2011).
S. Xiaochen, L. Jifeng, L. C. Kimerling, and J. Michel, Toward a Germanium Laser for integrated silicon
photonics, IEEE J. Sel. Top. Quantum Electron. 16(1), 124131 (2010).

1. Introduction
It has been long acknowledged that a monolithically integrated laser for silicon (Si) based
photonic circuits would be an enabling technology that could accelerate the implementation of
silicon photonics significantly [1]. Early attempts to integrate III-V semiconductor lasers on a
silicon platform had only limited success [2, 3]. More recently, germanium (Ge) has been
suggested as a gain medium for lasing on Si [4]. Using a combination of tensile strain and ntype doping, efficient direct bandgap emission of Ge can be achieved [5]. Optically pumped

TFH

- $15.00 USD
SS 2012#164840
(C) 2012 OSA

Received 19 Mar 2012; revised 24 Apr 2012; accepted 27 Apr 2012; published 2 May 2012
7 May 2012 / Vol. 20, No. 10 / OPTICS EXPRESS 11316

134

lasing in Ge was demonstrated using a Ge waveguide with polished facets [6]. Furthermore,
attempts in electrically injection have demonstrated pin and pnn Ge diodes emitting between
1590 and 1700nm [710]. Here we present an electrically pumped pnn Ge diode laser that can
be monolithically integrated into a CMOS process. These first laser devices produce more
than 1 mW of output power and exhibit a Ge gain spectrum of over 200nm.
2. Experiments and results
Initial estimates of gain in n-type Ge based on experimental results showed that an n-type
doping level of 1x1019cm3 would yield a gain of about 50 cm1 [11]. Such a gain can lead to
lasing when pumped optically because optical losses are mainly limited to facet losses and
free carrier losses in Ge. For electrical pumping, additional losses due to the electrical
contacts, free carrier losses in doped poly Si and losses due to the interaction with the contact
metal, have to be overcome. Modeling of mode propagation in Ge waveguides with electrical
contacts shows that these additional losses are >100 cm1.To overcome by these losses, the Ge
gain must be increased by increasing the n-type doping to a level of 3-5x1019 cm3 [2].
Recently, we achieved n-type doping levels of > 4x1019cm3 by using a delta-doping
technique during epitaxial growth of Ge [12]. By correlation of photoluminescence (PL)
intensity, n-type doping level, and measured material gain, we have determined that an n-type
doping level of 4x1019cm3 corresponds to a material gain of >400cm1, enough to overcome
the losses in an electrically pumped laser device.
Ge waveguides of 1m width were fabricated by selective growth of n-type Ge-on-Si in
silicon oxide trenches using Ultra-High Vacuum Chemical Vapor Deposition (UHV-CVD)
[3]. A delta-doped Ge layer was grown on top of the n-type Ge to serve as a phosphorous
diffusion source [12, 13]. The delta-doping technique inserts monolayers of P in the Ge film
at low temperatures by alternating the phosphine and germane gas flow in the CVD reactor.
After thermal annealing to drive the phosphorous into the n-type Ge layer, the delta-doped Ge
layer was removed during planarization using chemical mechanical polishing (CMP), to reach
a uniform doping concentration in the gain medium. The remaining thickness of the Ge
waveguide after CMP varied between 100 and 300nm depending on wafer and location on the
wafer. Due to severe dishing of the waveguides after CMP the supported optical modes in the
waveguides could not be determined exactly. Up to six cavity modes can be supported in the
largest waveguides. An 180nm thick amorphous-Si film was then deposited via a PlasmaEnhanced CVD process and subsequently phosphorus-implanted to a doping level of
1020cm3. After a dopant activation anneal at 750C, a metal stack, consisting of Ti and Al
was deposited for top and bottom contacts. The oxide trench provides excellent current
confinement. In order to assure even carrier injection into the n-type Ge, the top contact metal
was deposited on top of the waveguide. After dicing, the waveguides were cleaved to expose
the Ge waveguide facets. A thin oxide layer was deposited on the facets to protect against
contamination and catastrophic optical mirror damage which was observed in devices that did
not have oxide protection.

#164840 - $15.00 USD

(C) 2012 OSA

TFH

Received 19 Mar 2012; revised 24 Apr 2012; accepted 27 Apr 2012; published 2 May 2012
7 May 2012 / Vol. 20, No. 10 / OPTICS EXPRESS 11317

SS 2012

Fig. 1. Schematic of the measurement set-up.

The waveguide emission was measured using a Horiba Micro PL system equipped with a
cooled InGaAs detector with lock-in detection. The emission power measurement was
calibrated using light from a commercial 1550nm laser that was coupled into a single mode
optical fiber with the fiber end at the sample location. In the calibration we verified that the
detection was linear with input power. The electrical pumping was supplied by a pulse
generator with current pulse widths in the range of 20 s to 100 ms. The duty cycle was
varied between 2 and 50%, typically 4% to reduce electrical current heating effects. The laser
was contacted with metal probes and the current was measured using an inductive sensor
placed directly in the biasing circuit. The experimental set-up is shown in Fig. 1.

Fig. 2. Ge laser emission spectrum before (a) and after (b) threshold. The cavity length of the
waveguide is 333m and the waveguide height about 100nm. Current injection employed pulse
widths of 50s at 800Hz and 15C. The detector spectral resolution was 1.2nm.

Figure 2 shows the spectrum of an electrically pumped Ge laser below and above
threshold. The broad, direct band gap related electroluminescence spectrum, observed for
highly doped n-type Ge LEDs, has been reported earlier [14]. The spectra in Fig. 2 employed
short integration times to assure wide spectrum analyses. Measurement time for these large
laser devices is ultimately limited by metal contact breakdown due to the high current flow.
Figure 2(a) shows no spectral features above the noise floor. When the injection current
density is increased above threshold, sharp laser lines appear, as shown in Fig. 2(b). The
observed linewidth of the individual lines is below 1.2nm, the spectral resolution of the
measurement set-up. All measurements were performed with the samples mounted on a
thermo-electric cooler at 15C. Local device temperatures, however, are likely higher due to
the high current injection but could not be reliably determined.

#164840 - $15.00 USD

(C) 2012 OSA

Received 19 Mar 2012; revised 24 Apr 2012; accepted 27 Apr 2012; published 2 May 2012
7 May 2012 / Vol. 20, No. 10 / OPTICS EXPRESS 11318

135

Fig. 3. L-I curve for a 270m long waveguide device. 40s electrical pulses were used at
1000Hz. Measurement temperature was 15C.

Figure 3 shows the L-I spectrum for a typical electrically pumped Ge waveguide laser.
The lasing threshold at about 280kA/cm2 is clearly visible. This measurement was taken with
the set-up in Fig. 1 using a wide instrumental spectral resolution of 10nm, at a wavelength of
1650nm, monitoring a single laser line. The number of datapoints is limited by metal contact
breakdown at high current level. The optical emission power of about 1 mW corresponds to
Fig. 3. Occasionally we observed up to 7 mW. The spectrum in Fig. 2 shows two lines. The
estimate of the cavity free spectral range is 1nm, and the line spacing in Fig. 2, 3nm, is a
possible multiple of the FSR.
These lasers show a dependence of emission wavelength on threshold current density that
is consistent with the expected modal loss variation and that confirms the theoretical
conclusions that the gain spectrum of Ge for the given doping level and strain reaches over
more than 100nm spectral width [15]. For high doping levels of 4x1019cm3, and tensile strain
of ~0.2%, we observed lasing in the range from 1520nm to 1700nm. Figure 4 shows selected
laser lines between 1576nm and 1656nm for different Fabry-Perot cavities of the same gain
material.

devices, the Ge waveguide height is directly related to the modal loss. Since Ge has the
highest refractive index in our device structure, thinner Ge layers expel more of the resonant
mode into the highly-doped poly-Si cladding and into the lossy metal contacts. The
wavelength corresponding to the Ge gain peak (the threshold injection level) and the cavity
loss is the expected emission wavelength of the device. As the modal confinement decreases
with decreasing Ge layer thickness, modal loss and correspondingly threshold current
increases and the emission wavelength blue shifts. In Fig. 5, we show spectral threshold
conditions for two different modal cavity losses using a parabolic band model as described in
[8].
A Ge waveguide of 300nm thickness has a modal loss of about 90 cm1 due to losses in
the doped poly Si and the metal electrode (solid line). A Ge waveguide of 100nm thickness,
however, has a modal loss of about 1000cm1 due to the closer proximity of the mode to the
electrode (dashed line). To overcome the high losses of the thin Ge waveguide, a relatively
high carrier injection level is needed. Lasing is therefore expected at around 1520nm, close to
what we find in Fig. 2. For lower loss waveguides we expect lasing to occur at longer
wavelengths as shown in Fig. 4.

Fig. 5. Simulation of gain clamping condition for two different Ge waveguide thicknesses
(100nm: solid line; 300nm: dashed line). The axes plot the corresponding modal loss and gain
spectrum for the two different injection levels that are needed to overcome the respective
modal losses and to achieve lasing.

3. Conclusions
We have observed lasing from electrically pumped n-type Ge Fabry-Perot cavities. The
threshold current densities decrease with increased modal confinement. The emission
linewidth is less than the 1.2nm resolution of our measurement. Laser emission wavelengths
were observed between 1520nm and 1700nm with a variation consistent with the gain
clamping condition for each device. Measured output powers greater than 1 mW at room
temperature were measured. Improvements in the Ge growth, electrical contacts, and in modal
loss reduction will decrease the lasing threshold to values comparable with Fabry-Perot diode
lasers. The high power and observed gain spectrum of nearly 200nm indicate that the Ge laser
could be used for WDM applications. Since the laser can be monolithically integrated into any
CMOS process flow, novel device applications and systems can be developed.
Acknowledgments

Fig. 4. Spectra of Ge lasers with different Ge waveguide heights. The measured laser line
wavelengths are (a) 1576nm, (b) 1622nm, and (c) 1656nm.

TFH

The CMP-induced variation in cavity height provides self-consistent evidence of the wide
gain spectrum and the gain clamping condition by lasing. Under lasing action optical gain
(and population inversion) is clamped at exactly the value of the resonant cavity losses. In our

This work was supported by the Fully Laser Integrated Photonics (FLIP) program under APIC
Corporation, supervised by Dr. Raj Dutt, and sponsored by the Naval Air Warfare Center Aircraft Division (NAWC-AD) under OTA N00421-03-9-002. R.E.C.-A. was supported by a
NSF Graduate Research Fellowship award number 1122374.

#164840 - $15.00 USD

(C) 2012 OSA

#164840 - $15.00 USD

(C) 2012 OSA

Received 19 Mar 2012; revised 24 Apr 2012; accepted 27 Apr 2012; published 2 May 2012
7 May 2012 / Vol. 20, No. 10 / OPTICS EXPRESS 11319

SS 2012

Received 19 Mar 2012; revised 24 Apr 2012; accepted 27 Apr 2012; published 2 May 2012
7 May 2012 / Vol. 20, No. 10 / OPTICS EXPRESS 11320

136