Picosecond and sub-picosecond pulse generation

0 downloads 0 Views 2MB Size Report
Jan 1, 1992 - Picosecond and sub-picosecond pulse generation in .... have the evident advantage of being compatible with integrated electronic circuits.
Picosecond and sub-picosecond pulse generation in semiconductor lasers J.-M. Lourtioz, L. Chusseau, N. Stelmakh

To cite this version: J.-M. Lourtioz, L. Chusseau, N. Stelmakh. Picosecond and sub-picosecond pulse generation in semiconductor lasers. Journal de Physique III, EDP Sciences, 1992, 2 (9), pp.1673-1690. .

HAL Id: jpa-00248835 https://hal.archives-ouvertes.fr/jpa-00248835 Submitted on 1 Jan 1992

HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destin´ee au d´epˆot et `a la diffusion de documents scientifiques de niveau recherche, publi´es ou non, ´emanant des ´etablissements d’enseignement et de recherche fran¸cais ou ´etrangers, des laboratoires publics ou priv´es.

Phys.

J.

III

France

(1992)

2

1673-1690

SEPTEMBER

1992,

PAGE

1673

Classification

Physics

Abstracts

42.60F

42.55P

42.80W

Picosecond sub-picosecond and semiconductor lasers Lourtioz,

J.-M.

Chusseau

L.

d'Electronique Orsay Cedex,

Institut 91405

(Received

5

and

N.

Fondamentale,

generation

pulse

in

Stelmakh

URA

22

CNRS,

du

Universitd

Paris

XI,

220,

Bit.

France

I99J,

December

accepted

20

February

1992)

g6n6ration

d'impulsions dans les semiconducteurs d'int£rEt lasers h courtes est d'applications. aussi intdrdt EIIe pr6sente fondamental de pour un un relation 6troite dtudes les de la dynamique milieu semiconducteur. du Apr6s une par sa avec introductive, les performances des techniques de commutation de gain, de revue nous comparons commutation de penes de cavitd et de blocage de modes. illustr6e La comparaison est aussi bien r6sultats exp£rimentaux que par des r6sultats de moddlisation. Certains rdsultats de ces par des Un particulier est mis sur [es aspects de modulation de phase dans Ie milieu sont accent nouveaux. semiconducteur. Nous discutons finalement quelques perspectives de d6veIoppements nouveaux. R4s~1m4.

La

pratique

nombre

croissant

short optical pulses in semiconductor lasers is of practical interest for applications. It is also of fundamental interest because of its tight relation with the field of semiconductor laser dynamics. After introductory review, we the an compare performances of gain-switching, Q-switching and mode-locking techniques. The comparison is illustrated by experimental results as well as by laser modelling, of which A special some are new. emphasis is given on the aspects of phase modulation in the Finally, semiconductor laser medium. discuss perspectives for further developments. we some

Generation

Abstract.

an

1.

increasing

Introductory

High-speed picosecond demand exists

for a

need

review.

light

semiconductor time

scales

picosecond of

extemally systems [4, 5].

rate

of

of

number

sources

modulated

are

laser for

sources

essential diodes soliton

transmission

for

a

generating short pulses in the picosecond large variety of applications. To date, the

remains transmission

systems

in

field

the

in and

optical for

of

optical

fibers

time-division

communications

[1-3]

as

well

multiplexed

as

for

sub-

and

strongest where

high

bit

transmission

(Sub)-picosecond laser diodes are also expected to serve as clock generators in other hand, semiconductor all-optical parallel On the high-power pulsed lasers processors. sizes YAG, interesting altemative conventional of larger (e.g. laser to systems are an more dye and time-resolved solid-state laser lasers) for electro-optic sampling [6, 7], spectroscopy, time-domain. addition their semiconductor ranging and metrology in the In to compactness, electronic circuits. lasers have the evident advantage of being compatible with integrated

JOURNAL

1674

PHYSIQUE

DE

III



9

pulse characteristics desired according to the application which is envisaged. are soliton long-distance transmission in fibers require Fourier transform-limited pulses of adjustable length, while studies of nonlinear phenomena in various materials essentially need high peaks powers. For each application, different techniques and optical schemes be explored. As for other lasers, there basically three techniques for must are operating semiconductor lasers in pulsed regime : I) gain-switching (I.e. pulsed pumping), it) Q-switching, iii) mode-locking. These techniques were investigated early on in the three development of semiconductor lasers. Moreover, laser the first emission from an homojunction diode achieved in 1962 with the injection of pulses [8]. Two years structure current was Lasher later, proposed an inhomogeneous pumping of the active medium realize the to conditions for regularly pulsed emission [9]. Q-switching was effectively demonstrated in 1966 by Kumosov al. with a two-section laser [10]. Mode synchronisation in an et geometry injection semiconductor GaAs laser with an extemal cavity was claimed by Morozov et al. in 1968 [1il. All these first experimental investigations performed with homojunction were below operated low 77 The shortest K. pulses were already found temperature structures at a 100 ps, less than with pulse repetition of gigahertz [12]. several to be rates Different

instance,

For

Progress accomplished in the of semiconductor developments 1970,

measurements.

In

semiconductor

laser

Q-switched

of

for

search P.T.

lasing materials reported in laser operated

heterostructure

mode-locking

with

a

Better

same

pulses by

laser

saturable

well

as

1978

al.

et

decade

materials. Harris

[13]. In the

GaAs

new

Ho

E.P.

following

at

clearly year, Glodge

followed

accuracy demonstrated

also

was

active

technological in pulse mode-locking in a the

achieved

reported accurate measurement autocorrelation [14]. After a long period devoted to the improvement of CW laser spectral properties, as to the the

absorber

and

Lee

mode-locking

first

[15]

temperature

room

obviously

:

demonstrated

was

20 ps three

experiments

with

pulses

obtained.

were

after

years

double-

a

by Van

Passive

der

Ziel

et

by proton implantation through the diode facets. For the : was first time, subpicosecond pulses (0.65 ps) were measured in semiconductor lasers. This was performances were achieved (0.58 ps) followed by the work of Yokoyama et al., where similar GaAs with aged laser [17]. Gain-switching experiments in double lasers heterostructure an which experimental carried out in parallel, output pulses of 15 ps [18]. All these were gave improvements in terms of short pulse duration. represented decisive results [16]

al.

From was

of

the

the

certainly realizing

carried

InGaASP

year the

1982,

lasers

research

the

investigation

coherent

in

out

created

absorber

with

emitting

in

was

pursued

effects

optical development spectral regions near

emission

parallel

of

work

chirping

for

the

in

in

new

modulated

directions. and

[19-21]. single-mode

communication of 1.3

~m

and

1.5

pulsed These

The lasers

main with

direction the

investigations

distributed-feedback ~cm.

In

goal

1986,

Takada

were

(DFB) et

al.

picosecond pulse generation at 1.3 ~m by fiber compression of further pulses [22]. The efficiency of this novel technique was for the spectral region 1.5 [23, 24]. Fiber compression confirmed by several authors near ~m Q-switched laser where strong chirping effects also successfully applied to a single-mode was and powerful (3W) pulses [25]. Other generation of very short (~lps) led the to for and interferometers used compression schemes based gratings compensate to were on obtained by Silberberg Transform-limited chirping effects in mode-locked lasers. pulses were active mode-locking, respectively. Kuhl et al. [27] using passive and and et al. [26] transform-limited reported gain-switched DFB laser

A

second

concemed

locking, repetition repetition

important

improvement of (sub)-picsecond laser diodes during the last decade pulse repetition rate. For gain-switching as well as for active moderesulted the of modulation bandwidths. A from lasers with large progress use of 12 GHz 1986 [22]. achieved with gain-switched DFB laser in Pulses rate was a scaling from 16 GHz to 40 GHz were reported in different active works rates on the

increase

in

(SUB)-PICOSECOND

9



mode-locking monolithic

[28-31], in

demonstrated

350

to

up

InGaASP

without

lasers

SEMICONDUCTOR

IN

highest performance being

the

cavity

extended

PULSES

[31].

laser

extended

cavities

1675

LASERS

in 1989 by Tucker et al. with a recently, passive mode-locking was repetition rates [32, 34], leading to ultimate achieved

More

[34].

GHz

of studies directed peak-power optimization. In a few at was exceeding I W were directly obtained from solitary diodes [25, 35attained by complementing the 37]. However, in most high-power performances cases, were laser diode with several amplifier sections and nonlinear compression stages. A peak power of mode-locked elaborate 70 W reported by Delfyett et al. for an system comprised of a was AlGaAs laser diode with external cavity, grating and travelling-wave two compressor an a amplifiers [38]. Peak powers scaling from 100 W to I kW were achieved semiconductor in two experiments by using high-order soliton effects in fiber compression stages : the laser recent gain-switched DFB laser in one case [39] and a synchronously pumped surfacewas source a emitting InGaAs [40]. Extremely short pulse durations of 0.46ps, laser in the other case in references [38, [40] respectively, the 0.2ps and 21fs reported 39] and but were

Finally, an increasing experiments, peak powers

number

characteristics performances. laser for a little in these were Following this introductory review, the rest of the paper is organized in four sections. The techniques for short pulse generation, I-e-, gain-switching, Q-switching and modethree 4, locking, presented in sections 2, 3 respectively. Experimental results and and are calculations from Special is laser modelling used illustrate these techniques. emphasis to are (chirping) which are of prime importance in given on the aspects of phase modulation semiconductor finally discuss lasers. We perspectives of further developments in some semiconductor

5.

section

Gain-switching.

2.

Gain-switching is population

strong turn,

as

with and

lasing takes level rapidly

fast

a

excitation

high gain

thus

places,

as

soon

laser

upper

achieved inversion

decrease

below

is

threshold.

medium.

laser be

can

light

emitted

the

of the

level

reached intense

very The laser

Under before

and

the

emission

the

this laser

condition,

then

ceases

In

starts.

populations

of the and

an

optical pulse is obtained. Semiconductor lasers characterized by high gain values and short are cavity roundtrip times (w10ps). These features give the possibility of short pulse two generation at high repetition rates by gain-switching. The simplest mathematical description of the gain-switching process is provided by the two familiar equations for the carrier density N and the photon density S : rate ~ ~~

=

~

~

~~

~n

" ~~ The the

three

terms

on

Those

on

the

numerical case

step before

of

solutions simulations

side

are

side

rate

the

in

table

a

(2)

shown

figures

I

steady-state

value

respectively

total

the rate.

injection rate, respectively. cavity-loss rate, the Parameters entering

carrier

rate,

I.

and

in

are

the

recombination

stimulated-emission

(Fig. la) corresponds to an ideal step amplitude exceeds threshold current,

reaching

(2j

s.

represent

stimulated

the

equations (I) are

Noj

es j (N

equation (I)

of

and

equation (2)

of

listed

(1)

S

~n

and

rate

(i

rA

+

~p

right-hand

the

right-hand

spontaneous-emission equations (I) and (2) Exact

+

No)

ES) (N

(I

~~~

~ =

recombination

spontaneous

A

different

only

can

and 2 for excitation the

from

laser zero.

be found numerically. of Results particular cases of excitation. The first with an infinitely risetime. short If the

output

exhibits

Actually,

the

relaxation laser

response

oscillations is that

of

1676

JOURNAL

Table

I.

of parameters entering into figures1, 2 and 3 (right).

List

calculations

in

PHYSIQUE

DE

III

equations (left)

rate

1.5

photon

A F

carrier

o

p

spontaneous laser

filter

second-order second

pulse

case

emitted relaxation

is

first

ns

10-6 cm3/s

x

I

jransparency

at

x

I

x

10-~? cm3 ' 8 cm-3

10

0.5

emission

factor

2

10-5

x

factor

enhancement

4

efficiency

quantum

0.4

differential is govemed by two equations. The rectangular pulse excitation. single gain-switched A a of the pulse with nearly coincides the end of the current (Fig. 2) corresponds to a strong The last microwave case this situation being of practical interest for high-bit-rate current, time

whose

(Fig. lb)

2

coefficient factor

linewidth

a

a

coefficient

confinement

~m3

I

I ps

gain density

r

1~

lifetime

lifetime

nonlinear

N

for

laser

=

170 carrier

gain

linear

used

~m)

400

volume

spontaneous

r~

r~

DFB

~m

(L active

values

parameter

9

;

Parameters

V



evolution

corresponds to the back edge

when

oscillation.

injection Figures 2a, 2b and 2c represent the simultaneous time-evolutions of and laser frequency, respectively. Note that figure 2a also represents output power of the density since the gain is assumed to linearly depend on time-evolution carrier the N in equations (I) and (2). Figure 2c will be discussed later. As the laser pulse is seen, emitted reached its maximum. This is the norrnal situation after the laser gain has where only pulse is emitted per period. The maximum-gain to threshold-gain ration is presently close one microwave 1.2. It rarely 1.5 under modulation [41]. exceeds to The laser frequency evolution reported in figure 2c merits further instantaneous comments. It is implicitely that the laser is single-mode. In the assumed semiconductor laser theory, the interrelated dimensionless gain and refractive index variations via parameter are a modulation

optical the gain,

of the

communication.

known

a,

linewidth

the

as

frequency

variations

(Av (t)

c

=

enhancement

(t)/n1),

An

Au

it)

also

are

the

laser

fi

Ag(t)

=

factor

directly frequency $ =

(rA

[19]

An

:

(t)

refractive

to

evolution

can

No)

by

the

variations

:

I/r~)

(3) ,

reference frequency is taken at the laser threshold. equation (3), a linear decrease of the gain is accompanied by a linear laser frequency (downchirp). This situation is illustrated in figures 2b and 2c. quasi-linear downchirp. from lasers exhibit Non-linearities semiconductor a beginning and at the end of the pulse. Consequently, pulse compression can passing the downchirped pulses through a linearly dispersive medium such as the

where

From

(Within I

Since

index

described

be

ES) (N

ii

(1/4 aT) Ag (t).

a

=

related

+

a~=

Figure

3

a

the

Gaussian

pulses,

the

maximum

compression

Pulses occur

be an

rate

of the emitted at

the

by optical fiber is given by realized

[42].

shows

gain-switched

assumption of

decrease

DFB

the

laser

results at

of

1.52

pulse ~m

and

compression a

I km

experiments

dispersion-shifted

recently fiber

with

performed with a dispersion a total

(SUB)-PICOSECOND

N° 9

PULSES

LASERS

SEMICONDUCTOR

IN

1677

400

~°°

E

'~

200

o 5

0.2

loo

0.35

0

~

(

~ _

~

§

0.25

o,1

(

0.7

Ins)

Time

j~

D-1

>

~

$

0

0

"

(psi

Time

O-OS

(

600

400

200

0

0 0

0.35

D-S

0.2

0.7

(ns)I

Time

i oo

2

I

Z

~

0.2Sfl

0.1

?

5 M

'

I

o

~

~

~

~~~~

I.

Fig.

2.

Laser

°.35

j

Fig.

Fig. pulse

o

600

~~o

0

Fig-

I.

under

responses

electrical

strong

single-step

a)

excitation.

0 7

Ins)

Time

2.

b)

excitation,

rectangular

excitation.

laser

Time-evolutions

under

amplitude

modulation

laser

modulation.

line

Calculations

(Tab. I) characteristics, the

parameters

current

frequency

resonance

Under

deduced

maximum). 120

GHz

pulse

tail

width

of

to

2a

represents

from

laser

spectra

from

to

be

over

the

I~=

those

under

a

threshold,

a

I.

and

1.5

~m

The

bias, =3GHz,

f

DEB

dc

comparison. The including the light-

presented

are

for

measurements

the

l~

for

table

l~=150mA

optical

of the

measurement

characteristics and

in

gain.

of

set

reported

80mA,

threshold

the

=

spread

are

frequency c)

emitted

are

equations (1-3)

rate

determined

are

b) and

parameters

small-signal modulation modulation ~f~ 2.I GHz, 120 GHz (Fig. 3a, bottom)

and

microwave

strong

measured

figure

in

power

Laser

frequency

modulation

and

dotted

ps/nm [24].

20

of

The

gain a), output

of the

microwave

strong

respectively.

is

loo

0

0

various

at

200mA),

the

laser

=

the

pulsewidth

before

biases

laser

[43].

spectmm

is

compression

~

~18ps

from

autocorrelation

measurements

(FWHM,

full-width

half-

experiments. The calculated spectrum (Fig. 3a, top) is the low-frequency side, the latter being related to the wide and exhibits on a peak pulse (Fig. 3b) is asymmetrical and has a of the downchirped pulse. The calculated pulse after autocorrelation of the 18ps (FWHM). Figure 3c shows the trace Calculations

agree

with

1678

JOURNAL

DE

PHYSIQUE

al

9



III

b) i

j ~

so

i

~~~~

0

E

~ =

fl

$

3

i.

~

~

So

0

0 loo

~

c) 3

]

0

200

loo

Frequency

100

200

(GHz)

300

(psi

Time

d) '00

2

~

p

E 200

~ c

f

$

~

y o

3

~

1

X ~ ~

~

0

50

100

200

(ps)

Time

Fig.

So

compression

T;me

300

(ps)

gain-switched DFB pulses at 1.5~m laser (experiments and reported in table modulation I. The dc bias, amplitude and are modulation 200 mA frequency are I~ 90 mA, and f 2. I GHz, I~ respectively. The fiber dispersion is -20ps/nm. a) laser spectrum calculations (bottom) and (top) b) measurements calculated pulse before compression ; c) autocorrelation of the compressed pulses trace measurecalculations (top) ; d) calculated pulse after compression. (bottom) and ments 3.

Fiber

calculations).

Laser

of

those

parameters

=

compression. curve).

Again,

The

trace

autocorrelation

=

=

experiments (bottom curve) are is 8ps (FWHM). The

width

reveals

curve

deviation

some

from

reproduced

well

existence

either

a

by

(top wings in the Gaussian hyperbolic secant or a compression give rise to small of

calculations

extended

in the pulse before shaped pulse. Actually, chirp non-linearities in the compressed pulse (Fig. 3d). Owing to the overall fit between side-lobes calculations and experiments, a pulsewidth of 5.5 ps can be estimated after compression, which leads to a timebandwidth product At. Au of ~0.6. The compression (~3) is close to the value rate for a (~ 4). The peak power at the fiber output (350 mW) is more estimated than one order of magnitude higher than the CW power delivered by the laser. Still shorter pulses scaling from 2 to 4ps have been obtained by applying the same technique to different [44]. Ultimate lasers performances of gain-switching can be evaluated from an analytical approach of rate equations. Ignoring emission and electrical spontaneous pumping in equations (I) and (2) during gain-switching, analytical expressions are found for the

[41, taken

total

45]. into

number The

of

pulsewidth

account,

photons,

emitted

it

is

is

then shown

S dt, deduced

that

for

and

the

from

these

very

high

peak photon density, S~~~, respectively

quantities.

If

gain

initial-inversion-ratios

non-linearities the

are

minimum

(SUB)-PICOSECOND

N° 9

pulsewidth

given by

is

PULSES

SEMICONDUCTOR

IN

1679

LASERS

:

f

~~~ r~ is

where

and

A

photon

the

entering

e,

semiconductor

into

laser

lifetime

f

and

f are a priori terminology, 4

~

~

P

~

e/Ar~.

=

material aT~ e/A

(4)

f-1

~

The

linear

parameters is

known

the

as

gain

nonlinear

and

while

r~ is K-factor

cavity

a

parameter.

f~

and

coefficients, Al

=

/

In

aT

the is

e

material [46]. achievable with a given laser easily understood. e/A is much larger than In most cases, pulsewidth is close to than I in Eq. (4)) so that the ultimate r~ (I,e., oscillations. related to the period of laser relaxation The e/A. This smallest lasers is 0.17 ns [47], which gives minimum date for long-wavelength to 4 ps. Still expected for short-wavelength lasers wich exhibit weaker F/A smaller values are gain-non-linearities. Complementing the gain-switching technique with fiber compression, sub-picosecond pulse generation thus appears to be feasible for semiconductor with lasers larger than 5. a different limit is predicted for lasers with very long cavities A since f becomes less than I in equation (4). The minimum identifies in pulsewidth with the photon lifetime, that case with since propagation effects be taken be neglected must cannot care r~. However, this result far in rate equation models. Indeed, the cavity roundtrip time, r~, appears to be the true so physical limit instead of r~. the

modulation

maximum

bandwidth

equation (4) f is much larger value is obviously K-value reported

limits

The

of

be

can

=

Q-switching.

3.

Q-switching

in

series

the

with

implantation

of

multi-section 5

semiconductor

gain ions

laser

is

lasers

medium

through geometries,

(Fig. 4). diode

the

where

by creating

achieved Two

main

facets some

[48-50].

of the

regions of

solutions

are

The

sections

second are

absorption

saturable

used.

first

The

one

unbiased

is or

one

is

realization

the

in the

of

[9,

reverse-biased

Ii-

saturable

Fig.

4.

the

absorber JOURNAL

Basic

DE

scheme

PHYSIQUE III

of

a

time,

relaxation -T

2,

N'9,

semiconductor the

active

absorber

gain

SEPTEMBER

laser

recovery 1992

with

a

time

and

saturable the

medium

absorber.

cavity

r~,

roundtrip

r~

and

r~

are

respectively

time. 59

JOURNAL

1680

mechanisms

The time.

latter

The

involved must

be

diode

in

always

PHYSIQUE

DE

laser

shorter

than

Q-switching the gain

pulsed emission. However, it may be shorter depending on the cavity length and the absorber mode-locking. The laser dynamics in the to switching. Let

focus

us

on

the

second

One

situation.

III

depend

or

material.

The

with

difference

in

situation similar

are

gain-switching

relaxation

order

cavity

the

first

situation

second

time

than

absorber

the

on

relaxation

longer

9



to

obtain

roundtrip may to

is the

a

those

priori of

a

time, lead

gain-

possibility

of

generating optical pulses with a dc laser bias. In this pure Q-switching case, the pulse periodicity is fixed by the recovery medium. difference time of the gain Another is related to than gain-switched pulses due to the pulse shape. Q-switched pulses exhibit shorter tails from the absorber end. However, the major interest of Q-switching at pulse stems recovery of very intense pulses under electrical excitation. saturation achievement pulse Absorber simultaneously effects combined with gain switching and the gain at maximum may be two are times higher than the saturated cavity losses. description of Q-switching is provided by rate mathematical As for gain-switching, a good equation models. Equations (I) and (2) of section 2 are accompanied by a third equation for population density N~. the absorber

N~

dN

N ~

~

~~

(5)

N~ S

A

=

~

~a

carrier right-hand side of this equation represent the spontaneous absorber respectively. N~ is the equilibrium absorption, absorber time and A~ is the relaxation population in the absence of light, r~ is the absorber stimulated absorption term must be included as a loss term in equation (2). cross-section. The reference [49]. been proposed in solutions of rate equations have Approximate The precise description of phase modulation effects in the Q-switching regime is somewhat difficult due to the of the saturable absorber. The simplest approach is to presence assume an value for the (Eq. (4) in Sect. 2). Under this assumption, the chirp a-parameter average amplitude during the pulse is proportional to the gain variations as well as to the electron density variations, a linear relation existing between g and N. Since the number of emitted photons is approximately equal to the number of lost electrons, the total chirp amplitude is expected to be proportional to pulse energy. time-resolved Figure 5 shows the results of spectroscopic experiments perforrned with Qof different switched AlGaAs lasers pulse energies [52]. The lasers are operated in singleradiation in the cavity [53]. The pulse monochromatic mode with the injection of a small CW ion-implantation conditions, I.e., by changing the spatial energy is varied by changing the Electrical pumping is achieved with 2 ns of the saturable absorber in the cavity. extension made with a streak pulses at a repetition rate of 40 MHz. Optical pulse measurements are increase with pulse energy. As seen, the chirp amplitude is found to regularly For the camera. highest energy (13 pJ), the chirp amplitude (15 A) is about four times the cavity mode The

two

terms

relaxation

spacing. Figure energy.

6

the

stimulated

results

shows

The

pulsewidth by streak

on

the

and

obtained

pulsewidth measured

measured

after

in

compression

fiber

by

autocorrelation

compression is

l.3

ps,

experiments 22ps

is

which

agrees

laser of highest compression. The previous measurements

with

the

before

with

semiconductor reported to date for a Q-switched laser with electrical pumping. Beyond the performance, notice the extremely high one may of the compression Such a result value (~17), the latter being directly related to a. rate which non-linearities with semiconductor laser models include gain and predict an agrees [54]. However, evolution forrn for high-power lasers of the I +P/P~ the a ao camera

[25].

This

result

=

is the

best

(SUB)-PICOSECOND

9



Fig. with

a)

300keV

=

33

1400

HLP

implantation E

LASERS

1681

spectroscopic

Time-resolved ~m.

SEMICONDUCTOR

IN

images of single-mode gain-IQ-switched laser pulses at implantation b) HLP 1400 Hitachi E =1.2 pJ laser implantation -E=2.5 pJ c) HLP 1400 Hitachi laser with 11 MeV proton oxygen -E=5.8pJ; AlGaAs d) gain-guided with 20MeV implantation structure oxygen

5.

0.83

PULSES

Hitachi

without

laser

pJ.

0.8

d d

0.6

~'

0.4

2ps

I

38 ps 2

?

o 60

40

20

0

20

60

40

Delay, ps

Fig.

Second-harmonic

6.

AlGaAs

with

laser

compression. 24 ps/nm.

It A

20 becomes

factor

autocorrelation

MeV

of

2ps 0.65

oxygen after is

used

Q-switched

of

traces

implantation. The compression by a to

deduce

the

trace

laser

width

pulses

long pulsewidths

~200m

emitted

(FWHM) fiber in

the

by the

gain-guided

38 ps without extemal dispersion of with a total is

text.

contribution of the saturable absorber should be explicitly into taken for a account more complete analysis. The previous experimental illustrations clearly reveal the possibility of very large phasemodulation in single-mode Q-switched lasers. Due to high levels resulting in strong power laser non-linearities, the ultimate performances of Q-switching may be different from those previously established for gain-switched lasers. In principle, the ultimate limit of chirp

1682

JOURNAL

amplitude is given by the gain-bandwidth. be expected at 0.8 ~m for a pulse can compression. 240 fs pulses after

DE

PHYSIQUE

More

energy

N° 9

III

realistically, chirp amplitudes ~30pJ, which should of

of

about

40

correspond

h to

Mode-locking.

4.

Mode-locking pulses from the gain- and well

known

as

sources.

Q-switching minimum

the

as

is laser

width

the

most

For

technique to produce extremely short optical mode-locking is effectively superior to concerning the amplitude and timing jitter [30, 55] as efficient

semiconductor

techniques of the

obtainable

lasers,

pulses [16, 17, 29, 38].

However,

it is

also

more

in large variety of effects observed mode-locked semiconductor lasers, still of them remain unexplained. some Basically, mode-locking is achieved by introducing a mechanism inside the laser cavity to the longitudinal interact with another thereby locking them in phase. One modes to cause one is to modulate the laser gain at a frequency close to multiple of the frequency means or spacing between the longitudinal modes. This technique is referred to as active mode-locking. However, for semiconductor laser devices, standard the cavity spacing is beyond mode 100 GHz and it is very difficult modulate the laser gain at these frequencies. modeActive to locking requires the extension of the laser cavity. The most solution consists of using current the diode facet being antireflection extemal reflector, inner (AR) coated (Fig. 7a). The an second solution is to realize special devices such as monolithic extended-cavity lasers where the gain region is coupled to a low-loss passive waveguide [31, 56]. In both the cases, minimum pulse repetition rate is fixed by the cavity mode spacing whereas the maximum repetition rate is limited by the laser bandwidth. modulation laser pulses is to use a The other of achieving mode-locked absorber in the saturable means laser cavity. This technique which is referred mode-locking is best described in to as passive the time domain. saturable As for Q-switching (see Sect. 3), the absorber sharpen to serves risetime while gain saturation effects combined with fast bring absorber up the pulse recovery about an abrupt termination difference with Q-switching lies in the of the pulse [57-59]. The supplementary conditions imposed to the absorber relaxation time, r~ and to the cavity roundtrip time, r~. The most favourable situation for passive mode-locking when : occurs Since r~ is of the order of I ns in time of the gain medium. recovery r~ r~ r~, r~ being the semiconductor lasers, the previous condition be only fulfilled if the cavity length exceeds a can few As for active mode-locking, is to use an centimeters. the experimental solution external cavity (Fig. 7b). On the hand, passive mode-locking other also when: may occur extemal cavity is needed in that case, but very fast absorbers no r~ m r~ r~. In principle, be used. Mode-locked pulses were recently reported in two experiments without using must extemal cavity [33, 34] region of saturable the absorption realized by ion an : was implantation in the first case [33] while a three-section laser with reverse-biased section one of in the second [34]. Interest such configurations from the achievement used stems was case of very high pulse repetition The major drawback is the level of emitted weak peak rates. fully pulses, the the gain medium between successive As cannot two recover power. inversion remains population always value. rate at a low mode-locking from the The complexity of the semiconductor lasers mostly results process in of extemal cavity. Even if residual reflectivity of the AR coating is the low, use an very say

complex.

Among

~

the

~

~

10-~,

laser

1980

in

presence

into

existence

the

locked

must

of

be

the

small

treated

as

laser a

[60, 61]. Figure 8 gives of the diode cavity, the

different

clusters,

each

cluster

diode

a

cavity

can

never

be

ignored.

In

fact,

the

mode-

analyzed by Haus schematic representation in frequency-domain. Due to the longitudinal modes of the extended cavity can be gathered being centered around of the diode cavity one resonances.

two-cavity

system.

This

situation

was

first

(SUB)-PICOSECOND

N° 9

PULSES

SEMICONDUCTOR

IN

LASERS

1683

a)

b)

absorber

Fig. 7. Basic schemes of a mode-locked locking, b) passive mode-locking.

semiconductor

h~~~~ """'j

with

laser

I

/

$~~)~

cavity. a)

active

mode-

~~~~

~

""""

-.,

_.-

extended

-.

' '

compound cavity gain

_/

cluster locked

of modes

J~l/~ext Fig.

Mode-locking of a nonideally antireflection laser coated diode in an extemal cavity. The the diode cavity is simulated by a periodic gain profile (compound gain) [60]. resonances time in the extemal cavity while r~ is the roundtrip time in the diode sub-cavity. r~~~ is the roundtrip Modes from neighboring clusters necessarily in phase with one another. not are 8.

influence

Using

of

simple picture,

different of a given easily that the modes cluster can neighboring clusters necessarily in phase with not are another. Mode-locking is then only partially realized and the emitted pulses are far to be one description proposed by Haus was based on a single master transforrn-limited. The theoretical longitudinal mode [60, 61]. The equation which gave the steady-state conditions for each diode cavity simulated using periodic profile for the laser gain (see by resonances were a Fig. 8) and the pulsewidth limits evaluated in this condition. were lock

this

together.

In

contrast,

one

imagines

modes

from

PHYSIQUE

DE

JOURNAL

1684

N° 9

III

mode-locking of semiconductor lasers in an extemal authors [30, 62-67]. Except for one case analyzed in time-domain by several directed active mode-locking and numerical simulations at were were equation models. For the long pulse regime (AT r~) (*), the basic rate identical feedback almost those given in section 2, except that a delayed to (**) in equation of form k. S (t included the photon density the term to account r~~~) was with observations, for the effect of the extemal cavity [64, 65]. In agreement experimental instabilities when the roundtrip time in the external cavity was laser where shown to occur for [64]. regime shorter period of the pulses For the short pulse than the current recurrence modified diode cavity (AT « r~), the rate equations include propagation effects in the to were conditions inner diode facet the external reflector together with the boundary both the and at [30, 66-68]. In gain non-linearities ignored for the sake of simplicity. The contrast, were

Following the cavity was further [63], analyses performed from equations were

of

existence

of

laser successive

work

non-ideal

a

Haus,

AR

coating

for

the

a

residual

facet

found

was

pulses, being the roundtrip [29, 30], pulse mode-locked

measurements

reflectivities

diode

inner

of

as

low

as

the time trains

to

in

in the

emission

separation

between

result

time

the

diode

consisting

of

cavity. In three pulses

10-~.

still models remain unadapted to predict long pulse regime to the short pulse regime, I-e- from partial modeExperimentally, this transition abruptly for a certain locking to full mode-locking. occurs tuning of the extemal cavity and/or for a careful adjustment of the feedback coupling [29]. active mode-locking. In view of results behaviour observed in passive and The is indifferently chirping effects play an important role in sections 2 and 3, it is clear that reported in models into semiconductor Improved then necessarily them lasers. take must account. experiments on Transverse effects importance as revealed by recent some may be also of Sapphire lasers [69, 70]. Experimental and theoretical mode-locked other systems such as Ti investigations carried this direction for better understanding of should be in out a dynamics. semiconductor laser results obtained mode-locked AlGaAs Figure 9 shows by the authors on a passively recent commercial The experimental scheme close that presented in figure 7b. A laser. is to HLP1400 Hitachi diode is used. The saturable absorber is by an lo MeV laser created implantation of oxygen ions through the rear facet of the diode. The other facet is used for coupling to the external cavity. The absence of AR coating in the present experiments places 30 fb. The cavity is 15 cm long and is the facet reflectivity to a standard value of extemal microscope objective for beam collimation and an end-mirror comprised of an AR coated

However,

the

in

transition

spite

at

regular given pulse train

instead

autocorrelation

with

calculated

of

~

pulse trains pulses of

agreement were

earlier

from

of

theoretical

these

fits,

laser

the

~

with

total

reflection.

Under

additionally

norrnal

conditions,

the

laser

diode

is

dc

biased.

A

modulation

synchronize the pulses on an extemal reference clock. discussion. autocorrelation reported in figure 9b well illustrate the previous It is The traces clear that the mode-locked laser output is composed of pulse trains, the pulse spacing being amplifier (~ 8 ps). For most of the experimental the roundtrip time in the semiconductor modulation ratio is less than 60 fb conditions, the between pulses is moderate : the contrast However, for certain tunings of the extemal (dotted curve). abrupt change is observed an increases 90fb Simultaneously, the cavity and the modulation ratio (solid curve). up to emitted towards shorter wavelengths and the spectral width is increased spectrum is shifted 45h (Fig. 9a). The minimum pulsewidth from 25 deduced in these conditions is to of modes corresponds to phase-locking of five clusters ~0.7ps, which approximately source

can

be

used

to

(~) The pulsewidth, AT, is larger than the roundtrip time in the diode (*~) z~~j is the roundtrip time in the external cavity while k determines

cavity,

r~.

the

of

rate

optical

feedback.

(SUB)-PICOSECOND

9



PULSES

SEMICONDUCTOR

IN

1685

LASERS

io

8pS o-g

s

0.6

I

401

fi

6

~

4 0.4

4

0.2

2

0

0

835

837

839

Ml

Wavelength,

15

5

10

9.

with

a

Measured

15

different

(Al

long tunings of cm

A).

=15

a) and

autocorrelation

cavity. Dotted curve extemal cavity (see text).

extemal the

The

output

average

perforrnances

The

I W.

spectrum

solid

those

to

b) of

traces

and

is 4 mW,

power

comparable

are

5

10

15

ps

b)

a) Fig.

0

Delay,

nm

mode-locked AlGaAs passively b) respectively correspond to

a

laser

in

curve

thereby indicating peak-powers previously published by Van der

two

around Ziel

et

[16].

al.

Q-switching, the minimum pulsewidth by the inverse of the gain-bandwidth. pulses (~ 20 fs) since the gain-bandwidth typically for

As

by

obtainable

determined

mode-locking is, priori lead to a

should

This scales

from

20 to 60

rate-equation

nm

in

in

principle, femtosecond

semiconductor

placed the mode-locking [68]. Note that the finally authors case questioned existence of pure « active mode-locking ». A more direct estimation can be made from the spectral width of the laser emission. In pulsed regime, laser diodes emit can of reasonable amplitude (15 modes are obtained of in Fig. 9). In the up to 30 modes case AlGaAs lasers at 0.8 ~m, this approximately corresponds to a spectral width of 90 A and transform-limited pulses of 100 fs can be envisaged. lasers.

limit

5.

On

just

I ps about the

of

also

of

confirm

this

pulse

be

expected

pulses.

Research

fundamental

interest

media.

already

300fs

below

of

from

models

active

developments. can

laser

calculations

recent

some

the

further

femtosecond

semiconductor

short

in

developments

towards is

hand,

other

below

Prospects

Future it

the

From

appear

[40].

assertion

generation,

but

to

in

several

in this for

a

discussions be

Mode-locking optimized AR

not

1, 2

sections Recent

remains

important direction is to go by applications, but of transient phenomena in and 3, it is clear that pulsewidths obtained by optical pumping efficient technique in terms of

One

only knowledge

is

better in

feasible.

directions.

field

results

the

most

motivated

(R~10~~)

coatings

are

needed

eliminate

to

switching in vertical gain or Q of VCSEL Interest (VCSEL) offers emitting lasers surface great promise [71]. (~ l ~m) leading in tum of the gain medium thickness from tile weak optical stems structures pulse-to-pulse timing jitters can be Sub-femtosecond short cavity lengths (~ 10 ~m). to very thickness. emission is localized in an active region of such a small expected when spontaneous parasitic cavity

effects

in

extended

Multiple-quantum-well

VCSEL

cavities.

On

structures

the

other

consisting

hand,

of

wells

of

different

widths

also

appear

JOURNAL

1686

be

to

interesting

an

solution

for

PHYSIQUE

DE

realization

the

of

N° 9

III

semiconductor

lasers

with

wide

gain-

bandwidth.

important

Another in

progress

require

field

the

Fourier

challenge

is

picosecond

of

optical

soliton

semiconductor

of

realization

the

lasers

Long-haul adjustable length at coherent

of

is

soliton

transmission.

pulses

transform-limited

surprising

most

evolution

of

1.55

dictated

by

transmission ~m.

semiconductor

in

Perhaps, laser

the

rapid fibers of the

one

pulses

with

repetition rates scaling from 2.5 to 10 Gbit/s. tens Actually, long pulses are needed to minimize pulse distorsions and dispersive radiation, wave regularly spaced along the optical link [72]. both resulting from the use of lumped amplifiers mode-locking in an extemal selective cavity One technique for realizing such pulses is to use [73], the pulsewidth being adjusted with spectral filtering in the cavity. However, solutions monolithic cavities preferable for optical communication applications. Investibased on are mode-locking gations are presently carried out in this direction. A first solution is to achieve monolithic cavities with integrated Bragg reflector [56]. A second solution is to in very long compression technique to the case of gain-switched pulses of long duration. extend the fiber Single-mode lasers with small phase-amplitude coupling factors be designed in order to must dispersive fiber needed for compression. Strained-layer multiplelimit the length of the quantum-well DFB lasers are good candidates [74]. picosecond pulses of high energy are desired Finally, semiconductor lasers able to deliver 200pJ pulses are typically required for of applications. For instance, for a large number circuits. electronic Up to now, the output energy of optical triggering of fast semiconductor rarely exceeds 30 pJ in the short-pulse regime and there is room for improvement. The lasers intensity which can propagate in the laser medium is determined by the two-photon maximum absorption threshold. A typical value is 109 W/cm2. For a laser stripe of standard transverse (~0.2 x 2.5 ~m~), the peak-power is then 5W dimensions maximum around and the emitted in a lo ps pulse is 50 pJ. Further maximum improvements require the use of energy with larger emitting surfaces. Several solutions lasers be considered broad stripe can : I) iii) lasers with stripe of large thickness (1-2 ~m), iv) vertical lasers, it) laser diode arrays, cavity surface emitting lasers (VCSEL). The two first solutions commercially available, are but mode control is difficult. Solution iii) presents the advantage of output transverse beams ellipticity. However, CW operation is excluded. with weak Solution iv) a priori transverse provides the largest emitting surfaces. A quick estimate indicates that peak powers up to 100 W should be obtained solutions I), it) or iii). Still one with order of magnitude could be gained with solution iv). durations

6.

lasers. some

several

Concluding

This

have

of

picoseconds

of

remarks.

reviewed the paper has The different methods of the been

and

results

also

presented

discussed.

current

used for

be

generate

illustration

However,

low-dimensionality could not of low-dimensionality lasers applications to short pulse performances are not related

understanding to

some

specifically

of short

pulse generation in semiconductor pulses have been analyzed in details, Possibilities of further improvements

short

being new. important aspects treated

in this

related context.

for

instance

to

lasers

of

Actually, the development necessarily motivated by

other developments is not recent generation high-speed modulation. and Moreover, the laser in a simple material instance, it is For to the parameters. manner established that well waveguide lasers do not exhibit better perforrnances than quantum now bulk lasers of the In acomplished in the growth of thin geometry. contrast, same progress semiconductor layers allows the realization of new geometries such as vertical cavity lasers which of great promise for high-power short-pulse generation. are as

(SUB)-PICOSECOND

N° 9

PULSES

1687

LASERS

SEMICONDUCTOR

IN

Acknowledgment. The

acknowledge

authors

valuable

the

and

assistance

expert

of

D.

in

Pascal

the

course

of

experiments.

References

Ill

L.F.,

MOLLENAUER

Demonstration

M.,

NYMAN

B.

soliton

transmission

of

M.J.,

NEUBELT at

2.4

Gbit/s

RAYBON 12 000

over

S.G., G., EVANGELIDES km, Elecn.on Lett. 27 (1991)

178-179.

[2]

Conference

pean l55.

[3]

[5]

on

soliton

using

communication

Optical

Communications

ECOC'

Erbium-doped fibre 91 (Pads 1991)

amplifier, Invited

17th

paper,

Euro-

pp.

150-

Observation of 10 Gb/s optical soliton NISHI S., SARUWATORI M., over 17th Conference Optical gain-switched DFB-LD pulse European on a source, Communications ECOC'91 (Paris 1991) Post-deadline paper, pp. 88-91. time-division multiplication SARUWATARI M., 100 Gbit/s optical signal generation by TAKADA A., distributed feedback (DFB) laser modulated and compressed pulses from gain-switched of 1406-1408. diode, Electron Lett. 24 (1988) time-division multiplexing for very highTUCKER R. S., EISENSTEIN G., KOROTKY S. K., Optical CLEO 88 (Anaheim 1988), Conference Laser Electro-Optics bit-rate and systems, paper on IWATSUKI

K.,

10 000

[4]

Optical

M.,

NAKAzAWA

K.,

SusuKI

using

km

TUKS.

[6]

circuits using LouRTioz J.-M., Electro-optic Sampling of GaAs L., lasers, 3rd European Conference Electron and Optical Beam Testing of on Integrated Circuits (Como 1991), Conf. Digest, pp. 246-253. BuRRus C. A., Electro-optic WIESENFELD J. M., TAYLOR A. J., TUCKER R. S., EISENSTEIN G., sampling using injection lasers, SPIE Proc. 795 (1987) 339-344. Coherent light HALL R. N., FENNER G. E., J. D., SOLTYS T. J., CARLSON R. O., KINGSLEY

L.,

DUVILLARET

CHussEAu

semiconductor

[7]

[8]

emission

from

junctions,

GaAs

Phys.

Rev.

Lett.

9

(1962)

366-368.

[9] LASHER G. J., Analysis of a proposed laser, Solid State 707-716. bistable Electron. 7 (1964) L.A., V.G., [IOjKURNOSOV MAGALYAS V. I., PLESHKOV A.A., RIVLIN TRUKHAN V.D., V. V., Self modulation of emission from an injection semiconductor laser, JETP TsvETKov 303-305. Lett. 4 (1966) SHERONOV A. A., Self synchronization of modes in a GaAs [I Ii MoRoZov V. N., NIKITIN V. V., semiconductor injection laser, JETP Lett. 7 (1968) 256-258. [12] ZAKHAROV Y. P., KOMPANETS I. N., NIKITIN V. V., SEMENOV A. S., Synchronization of gallium 895-896. JETP 26 (1968) arsenide laser light pulses, Sov. Phys. [13] HARRIS E. P., Spiking in current-modulated CW GaAs extemal cavity lasers, J. Appl. Phys. 42

(1971) [14] [15]

[16j

[17]

882-893.

D., LEE T. P., Study of a self-pulsing GaAs in Lithium laser by intensity correlation Iodate, J. Appl. Phys. 42 (1971) 307-309. GLASSER L. A., IPPEN E. P., Ho P. T., HAus H. A., Picosecond pulse generation with a CW GaAlAs laser diode, Appl. Phys. Lett. 33 (1978) 241-242. VAN ZIEL J. P., TSANG W. T., LOGAN R. A., MIKULYAK R. M., AUGUSTYNIAK W. M., DER Sub-picosecond T.P., from passively mode-locked GaAs buried pulses optical guide semiconductor lasers, Appl. Phys. Lett. 39 (1981) 525-527. YOKOYAMA H., ITO H., INABA H., Generation of sub-picosecond coherent optical pulses by GLODGE

passive

mode-locking

of

an

AlGaAs

semiconductor

diode

Appl. Phys.

laser,

40

(1982)

picosecond

pulse

Lent.

105-107.

[18]

G. J.,

CAROLL

generation

with

ASPiN

39

(1981)

J.

E.,

highly

860-861.

PLUMB rf

R. G.,

modulated

The

AlGaAs

effect

double

of

cavity

length

heterostructure

on

lasers,

Appl. Phys.

Lent.

1688

[19]

JOURNAL

(1982) [20]

T, L.,

KOCH

BowERs

E.,

J.

Electron.

Lett.

LINKE

R. A.,

T. L.,

KOCH

A.,

TAKADA

GHz

12

pulse [23]

T.,

SUGIE

Effect Lent.

IEEE

J.

Quantum

QE-18

Electron.

K.,

directly

in

modulated

semiconductor

M.,

semiconductor

on

5.6ps

Transform-limited

Electron.

SARUWATARI

reduction

laser

wavelength

613-615.

gain-switched

pumped

diode

gain

nonlinear

(1981)

technique,

chirping

1038-1040.

of

48

from

rate

IwATsuKi

by laser

12 W

9



lasers,

semiconductor

wavelength

of

(1984)

SARUWATARI

repetition compression

A.,

TAKADA

of

Nature 20

chirping, Appt. Phys. [22]

linewdith

of the

III

259-264.

lasers,

[21]

Theory

C. H.,

HENRY

PHYSIQUE

DE

distributed

Lent.

M.,

22

(1986)

Picosecond

erbium-doped

fiber,

optical pulse

feedback

diode

laser

generation at by employing

1347-1348. laser

iEEE

diode

pulse Technol.

Photon.

amplification Lent.

2

(1990)

up to 122-

l24.

[24j

L.,

CHUSSEAU

J.M.,

XIE

with

laser

[25]

[27]

0.3

fibre

J.-M.,

LOURTIOZ

W

a

1.53

ACCARD ~m

A.,

microwave

HEBERT

J.-P.,

modulated

DFB

1085-1087.

BARTHELEMY Ultrashort pulse generation from a Q-switched A., injection, Appl. Phys. Lent. 59 (I99I) 624-626, SILBERBERG Subpicosecond pulses from a Y., SMITH P. W., mode-locked semiconductor laser, IEEE J. Quantum. Electron. QE-22 (1986) 759-761. SERENYi Bandwidth-limited KUHL J., M., GOBEL E. O., picosecond pulse generation in an actively mode-locked GaAs laser with intracavity chirp compensation, Opt. Lett. 12 (1987)

N.,

STELMAKH

J.-M.,

LouRTIoz

AlGaAs

[26]

L.,

DUVILLARET

picosecond pulses (4 ps) from compression, Electron Lett, 26 (1990)

Bandwidth-limited

with

laser

CW

334-336.

[28]

G.,

EISENSTEIN

of

(1986)

[29j

TUCKER

R.

InGaASP-single

CORZINE

S.,

mode

KOREN

U.,

KOROTKY

K.,

Actively

lasers,

IEEE

S.

cavity

composite

fiber

J.

mode-locking characteristics Electron. Quantum QE-22

142-148.

W.,

S.

Actively

BOWERS

J.

E.,

G.,

PRZYBYLEK

GaInASP

mode-locked

laser

with

KOREN

U.,

Subpicosecond

MILLER

output,

B.

I.,

SOCCHOLICH

Appl. Phys. Lett.

C.

E.,

(1987)

52

348-350.

[30]

P. A., MAR A., CORzINE S. W., Actively mode-locked semiconductor Quantum Electron. QE-25 (1989) 1426-1439. TUCKER R. S., KOREN U., RAYBON G., MILLER B. I., KocH T. L., EISENSTEIN G., Active modelocked in a 1.5 ~m monolithic extended 621-629. cavity laser, Electron. Lent. 25 (1989) SERGEEV A. B., Generation of bandwidth-limited I00GHz VASIL'EV P. P., 2ps pulses with Electron. 1049-1050. repetition rate from multisegmented injection laser, Lett. 25 (1989) CHELNOKOV A. V., ZARRABI J. H., PORTNOI E. L., Passive mode locking of a multistripe single GaAs well laser diode with an intracavity saturable absorber, Appl. Phys. Lent. 59 quantum 1526-1528. (I99I) CHEN Y. K., Wu M. C., TANBUN-EK T., LOGAN R. A., CHIN M. A., Subpicosecond monolithic colliding-pulse mode-locked multiple quantum well lasers, Appl. Phys. Lett. 58 (1991) 1253BowERs

E.,

J.

lasers,

[3 II

[32] [33]

[34]

MORTON

IEEE

J.

l255.

[35]

Y.,

TSUCHIYA

[36]

VASIL'EV

P.

Quantum [37]

PORTNOI

E.

with

a

(1989) [38j

DELFYETT

MIWA

pumped

current

M.,

KoisHi

M.,

Direct

laser,

semiconductor

J.

observation of picosecond light pulses from 6671-6672. Appl. Phys. 53 (1982) technique for ultrafast Q-switching, a new

injection laser : (1988) 2386-2391. Characteristics L., STEL'MAKH N. M., CHELNOKOV A. V., of heterostructure saturable absorber fabricated by deep ion implantation, Sov. Tech. Phys. P.,

Picosecond

Electron.

a

pulse-

iEEE

J.

24

lasers Let'.

15

N.

C.,

432-433. P.

J.,

LEE

C.

H.,

FLOREZ

L.

T.,

STOFFEL

N.

G.,

GMITTER

T.

J.,

ANDREAKIS

sub-picosecond high-power optical pulses mode-locked semiconductor I37I-1373. from a hybrid laser, Opt. Lent. 15 (1990) SusuKi K., YAMADA E., Femtosecond optical pulse generation using a [39] NAKASAWA M., distributed-feedback laser diode, Electron. 2038-2040. Lett. 26 (1990) WATANABE K., MACHIDA S., SAKAI Y., IWAMURA H., [40j XIANG W. H., FRIBERG S. R., Sub-100 femtosecond YAMAMOTO Y., pulses from extemal-cavity surface emitting an ALPHONSE

G. A.,

CONNOLLY

J.

C.,

Generation

of

(SUB)-PICOSECOND

N° 9

multiple

InGaAs/InP

(1991) [41]

K.,

LAU

pulse

Short

(1985) [43]

generation

in

pulses

laser

semiconductor

pulses

Appl. Phys.

semiconductor

lasers,

Lett.

59

Lighnvave

J.

in

Appl. Opt.

fibers,

monomode

24

N.,

G.,

J.-M.,

XIE

(~2.sps) up

rates

B. A.,

LENGYEL

Dynamic modeling

behaviours and

experiments

CHUSSEAU

L.,

LOURTIOZ

semiconductor

of

lasers

1.3~m,

at

under

iEEE

J.

633-640.

Picosecond

repetition

with

W.

modulation:

(1990)

26

STELMAKH

N.,

BOUADMA

J.-M.,

LouRTioz

current

Electron.

M.,

CAVELIER

WAGNER

L.,

CHussEAu

Quantum

[45]

compression,

400-419.

chirped

of

sinusoidal

strong

[44j

soliton-effect

1689

LASERS

4196-4198.

E.,

HEMERY

with

laser

high frequency signal

and

(1989)

LT-7

Propagation

J.,

Buus

well

quantum

SEMICONDUCTOR

IN

2076-2078.

Technol.

[42]

PULSES

wavelength-tunable GHz,

12

to

Evolution

Electron.

Lett.

in

C.,

KAZMIERSKI

semiconductor

(1992)

28

giant pulse

of the

J.-M.,

(~20nm)

laser

224-226.

laser, J. Appl. Phys. 34 (1963)

a

2040-2046.

[46]

High-speed

R. S.,

TUCKER

modulation

of

lasers, J. Ligthwave

semiconductor

Technol.

(1985)

LT-3

1180-1192.

[47]

R.,

NAGARAJAN

(1991) 148]

VAN

developed

TSANG

D.

QE-19

152]

[54]

[55]

COLDREN L., High-speed damping, Appl. Phys. Lett. 58

S.,

R.

with

low

27

PORTNOI

E.

lasers,

P.,

Intensity lasers,

J.,

D.

iEEE

A.,

MAR

of

Lett.

dependence BowERs

J.

STELMAKH

N.

Phys.

linewidth

Technol.

E.,

12th

and

iEEE

Q-

Quant.

J.

Electron.

lasers,

AlGaAs

laser

with

and

its

implication

timing jitter

in

actively

for

212-214.

absolute

Intemational

from

452-453.

semiconductor

factor

1(1989)

well-

pulses

(1986)

Q-switched

of

of

segment

Picosecond 12

enhancement

Lett.

Relative IEEE

M.,

Q-switched

Injection-locking (1991) 160-161.

27

bombarded

lasers,

in

of

output

introduction

587-592.

Lent.

diode

chirp

modal

of the

Photon.

lasers,

(1985)

Tech.

deliberate

proton

a

Pulsating

M.,

W. the

to

QE-21

semiconductor

JULIEN F. H.,

Electron.

semiconductor

L., Sov.

Q-switched

J.-M.,

absorber,

semiconductor DERICKSON

Electron.

due

with

lasers

Quant.

J.

B.,

A.

39

semiconductor

J.-M., Analysis (1991) 1856-1858.

LouRTioz

saturable G.

Lent.

AUGUSTYNIAK

guide lasers (1981) 376-378.

LOURTIOZ Lett.

N.,

STELMAKH

locked

[56j

GEEL

lasers

A.,

R.

LOGAN

heterostructure

Z., WALPOLE J. N., (1983) 145-156. N.,

AGRAWAL

iEEE

ZHULAREV

injection

STELMAKH

fast

J. well

optical

of

pulsations,

Electron.

[53]

E.,

BowERs

quantum

T.,

W.

buried

Pulsations

I.,

Zh.

switched

[51]

TSANG

Appl. Phys.

loss,

M.,

KUZNETSOV

ALFEROV

P.,

J.

confinement

saturable

[50]

multiple

strained

2326-2328. ZIEL

DER

separate

[49]

T.,

FUKUSHIMA

InGaAs/GaAs

Semiconductor

mode-

Conference

Laser

(Davos 1990) Conf. Digest, pp. 138-139. G., HANSEN P. B., KOREN U., MILLER B. I., YOUNG M. G., CHIEN M., BURRUS C. A., ALFERNESS R. C., A monolithic extended-cavity laser with an integrated Bragg reflector for Conference Optical Communications active mode-locking at 8.3 GHz, 17th European on

RAYBON

[59]

40-43. (Paris 1991) Postdeadline paper, pp. G., LETOKHOV V. S., Fluctuation mechanism of ultrashort pulse generation by laser with Electron. 766-782. saturable absorber, iEEE J. Quantum QE-8 (1972) saturable absorber, iEEE J. Quantum Electron. HAus H. A., Theory of mode-locking with a slow 736-746. QE-11 (1975) absorber, J. Appl. Phys. 46 (1975) 3049HAus H. A., Theory of mode-locking with a fast saturable

[60]

HAus

ECOC'91

[57] [58]

KRYUKOV

P.

3058.

H. A.,

Phys. [61]

HAus

H.

51

A.,

Theory

of

mode-locking

active

(1980) 4042-4049. Mode-locking of

of

semiconductor

a

laser

semiconductor

diodes,

laser

Japan

J.

in

an

extemal

Appl. Phys.

cavity, 20

J.

(1981)

Appl. 1007-

l020.

[62]

VAN

DER

ZIEL J. P.,

Appl. Phys.

52

AcIive

(1981)

mode-locking 4435-4446.

of

double

heterostructure

lasers

in

an

extemal

cavity,

J.

JOURNAL

1690

[63]

HAus

A.,

H. well

[64]

B.

locked

[65]

J.,

CHESNOY

J.

KLEIN

DEMOKAN

S.,

M.

Electron.

[67]

60

M.,

SCHELL

A

M.,

SCHELL

G.,

A.

WEBER

facet

[69]

NEGUS

D.

Lens

SALIN

BnTTCHER

G.,

SCHOLL

E.,

K.,

L.,

SPINELLI

self-focusing,

multiple-quantum-

a

actively

of

mode-

Period-doubling and periodcavity, J. Appl. Phys. 67

extended

with

by gain

mode-locked

SCHOLL E.,

H.,

E.

modelocking pulsewidth, IEEE J.

mode-locking Postdeadline

diode

actively

active

A.

active

F., SQUIER J.,

with

dynamics

the

J.-M.,

LouRTIoz

laser

laser

reflectivities,

1991) [70]

diode

modelocking of IEEE J. Quantum

generation by

L.,

mode-locked

a

diode

modulation,

J.

Int.

67-85.

pulse generation by cavity limit for the [68]

of

model

of

B 2

CHussEAu

actively

(1986)

WEBER

C.,

M.

quadrupling for an (1990) 7615-7617. [66]

N° 9

III

Mode-locking of a laser (1985) 1237-1243. Influence of diode facet reflectivity on K., KEMPF P., semiconductor lasers, Appl. Opt. 25 (1986) 3846-3849. Y., Theory Opt. Soc. Am.

SILBERBERG

structure,

GARSIDE

PHYSIQUE

DE

GOLDBLATT

A1203,

InTi

of

Quantum

BIMBERG

D.,

N.,

Fundamental

in

(1991) limits

an

extemal

402-409.

sub-ps

of

width

pulse and

the

1661-1668. fs

Sub-100

State

sub-picosecond

spectral gain

the

(1991)

QE-27 FEUGNET G., Solid

Theory of amplifier

QE-27

lasers

Electron.

Advanced

laser

Electron.

semiconductor

a

D.,

BIMBERG

semiconductor

a

pulse

generation by (Hilton

Conference

Lasers

Kerr Head

PD4.

paper MouRou

G.,

Advanced

Solid

PicHt

State

M.,

Mode-locking

N.,

MCCARTHY

(Hilton

Conference

Lasers

Head

of Ti

A120~ using

199I)

Postdeadline

PDS.

[7ij

paper KARIN J. R.,

MELCER

L.

G.,

NAGARAJAN

R.,

BOWERS

J.

R.,

CORZmE

W.,

S.

MORTON

P.

A.,

picosecond pulses with a gain-switched GaAs surface-emitting laser, Appl. Phys. Lett. 57 (1990) 963-965. [72] MOLLENAUER L. F., EVENGELIDES S. G., HAus H. A., Long-distance soliton propagation using lumped amplifiers and dispersion-shifted fiber, J. Lighnvave Technol. 194-197. 9 (1991) [73] LUNDQVIST S., ANDERSSON T., ENG S. T., Generation of tunable single-mode picosecond pulses from an AlGaAs semiconductor laser with grating feedback, Appl. Phys. Lent. 43 (1983) 715GEELS

R.

S.,

COLDREN

L.

A.,

Generation

of

717.

[74]

KANO

3

F., YosHiKuNi Y., FUKUDA M., YOSHiDA J., InGaAsPflnP strained-layer multiple-quantum-well

(1991)

877-879.

Linewidth DFB

lasers,

enhancement

IEEE

Photon.

factor

of

Technol.

1.3

~m

Lent.