WO2002058200A2 - Quantum dot lasers - Google Patents

Abstract

A quantum dot active region is disclosed in which quantum dot layers are formed using a self-assembled growth technique. In one embodiment, growth parameters are selected to control the dot density and dot size distribution to achieve desired optical gain spectrum characteristics. In one embodiment, the distribution in dot size and the sequence of optical transition energy values associated with the quantum confined states of the dots are selected to facilitate forming a continuous optical gain spectrum over an extended wavelength range. In another embodiment, the optical gain is selected to increase the saturated ground state gain for wavelengths of 1260 nanometers and greater. In other embodiments, the quantum dots are used as the active region in laser devices, including tunable lasers and monolithic multi-wavelength laser arrays.

Description

QUANTUM DOT LASERS

STATEMENT AS TO FEDERALLY SPONSORED RESEARCH AND

DEVELOPMENT

[0001] The U.S. Government may have certain rights in this invention

pursuant to research conducted under the following grants: Grant No. F49620-

95-1-0530 awarded by the Air Force Office Of Science and Research, Grant No.

DAALOl-96-02-0001 awarded by the Army Research Lab, Grant No. F4920-99-1-

330 awarded by the Air Force Office of Science and Research, and Grant No.

MDA972-98-1-0002 awarded by the Defense Advanced Research Projects

Agency.

RELATED APPLICATIONS

[0002] This application claims priority under 35 U.S.C.§ 119(e) to the

following United States Patent Application Nos.: 60/238,030 entitled "Broadband

Continuously Tunable-Wavelength Quantum Dot and Quantum Dash

Semiconductor Lasers with Low-Threshold Injection Current" filed October 6,

2000; 60/252,084 entitled "Quantum Dot and Quantum Dash Semiconductor

Lasers For Wavelength Division Multiplexing (WDM) System Applications" filed November 21, 2000; 60/276,186, entitled "Semiconductor Quantum Dot Laser

Active Regions Based On Quantum Dots in a Optimized Strained Quantum

Well," filed March 16, 2001; 60/272,307, entitled "Techniques for Using

Quantum Dot Active Regions In Vertical Cavity Surface Emitting Lasers," filed

March 2, 2001; and Attorney Docket No. 22920-06322, entitled "Quantum Dot

And Quantum Dash Active Region Devices," filed August 31, 2001 (Application

number not received from the United States Patent and Trademark Office at the

time of filing of the instant application). The contents of all of the above

applications are hereby each incorporated by reference in their entirety in the

present patent application.

[0003] This application is also related to Attorney Docket No. 22920-6391

"Quantum Dash Devices," by Stintz et al., filed in the United States Patent and

Trademark Office on September 20, 2001, commonly owned by the assignee of

the present patent application, the contents of which are hereby incorporated by

reference in their entirety.

BACKGROUND OF THE INVENTION

1. Field of the Invention [0004] The present invention relates generally to self-assembled

semiconductor quantum dot lasers. More particularly, the present invention is

directed towards quantum dot lasers with improved optical gain characteristics.

2. Description of Background Art

[0005] Quantum dot lasers are of interest for a variety of applications. Each

quantum dot of a quantum dot laser is a three dimensional quantum-confined

heterostructure which confines electrons and holes in a region having a size,

along each of three dimensions, that is less than a thermal de Broglie wavelength

at room temperature (e.g., less than about 100 nanometers in many applications).

The quantum confinement produces quantum confined energy states within each

dot for electrons and holes. There are also corresponding optical transition

energies for both electrons and holes associated with a ground state transition

energy, first excited state transition energy, second excited state transition

energy, etc. A quantum dot laser typically includes a substantial total number of

quantum dots within a gain producing region.

[0006] Theoretical studies indicate that quantum dot lasers have many

potential performance advantages over conventional quantum well lasers. First,

a quantum dot laser has a lower fill factor (volume of material to be pumped)

and an improved density of states function compared with a quantum well laser. Referring to FIG. 1, the theoretical density of states function becomes sharper as

the carrier dimensionality decreases. FIG. 1A shows a theoretical density of

states function for a bulk material, which has a square root dependence on

energy. FIG. IB shows the theoretical density of states function for a quantum

well (one dimension of quantum confinement) that increases in steps at each

quantum well energy level. FIG. 1C shows the theoretical density of states

function for a quantum wire (two dimensions of quantum confinement). FIG. ID

shows the theoretical density of states function for a quantum dot (three

dimensions of quantum confinement) that has a delta-like density of states

function (e.g., a finite number of states available only at the quantum dot).

Theoretical calculations indicate that the threshold current of a semiconductor

laser may be improved by using quantum dot active regions due to the smaller

volume of material and reduced number of states.

[0007] One technique to fabricate a quantum dot laser exploits the tendency

of strained layer semiconductors to form islands when the strain exceeds a

certain threshold strain. In particular, InAs tends to form islands when a

sufficient thickness of InAs is grown on a layer that is psuedomorphically

strained on a GaAs substrate due to strain associated with the difference in

relaxed lattice constant of the two materials. The InAs islands may be embedded

in another layer, such as a GaAs quantum well, to form quantum dots. [0008] Self-assembled quantum dot lasers having a low threshold current

density are disclosed by Lester, et. al. in the article entitled "Optical

Characteristics of 1.24-μm InAs Quantum Dot Laser Diodes," IEEE Photonics

Technology Letters, Vol. 11, No. 8, August (1999). The quantum dot laser

structure disclosed in this reference includes a single layer of InAs quantum dots

formed by a self-assembled growth process on a GaAs substrate. FIG. 2A shows

measured laser spectra for different cavity lengths for the quantum dot lasers of

Lester, et al. FIG. 2B shows measured current density versus cavity length for

conventional Fabry-Perot lasers having uncoated facets.

[0009] Referring to FIG. 2B, the lowest threshold current density operation

occurs for a cavity length of almost 3.94 millimeters, for which an emission

wavelength of 1.24 microns was achieved. The emission wavelength at 1.24

microns is attributed to a ground state transition energy. When the cavity length

is reduced to less than about one millimeter, the lasing wavelength blue-shifts to

below 1.15 microns, which is attributed to lasing off of the first excited state.

When the cavity length is further reduced to about 500 microns, the emission

wavelength is further blue shifted to about 1.05 microns.

[0010] One drawback of the quantum dot lasers of Lester, et al., is that the

maximum wavelength is shorter than desired, particularly for cavity lengths less

than one millimeter. In a variety of commercial applications, a wavelength of greater than 1.260 nanometers (1.260 microns) is desired. This is because a

wavelength of at least 1260 nanometers is of interest for use in the OC-48 and

OC-192 standard compliant lasers used in short and medium length optical

network links. OC-48 and OC-192 are optical carrier (OC) standards for fiber

optic networks conforming to the SONET standard. OC-48 has a data rate of

2.4888 Gbps whereas OC-192 has a data rate of 9.952 Gbps.

[0011] Another drawback of the quantum dot lasers of Lester, et al. is that the

modal gain of the ground state is lower than desired, making it impractical to

design short-cavity lasers that lase from the ground state transition energy level.

The desired laser length depends upon the application. First, many

commercially available packages are designed for a cavity length no greater than

500 microns. Second, directly modulated lasers with conventional facet coatings

must typically have a cavity length less than 500 microns, and preferably less

than about 300 microns, in order to have a photon lifetime that is sufficiently

short to permit high speed direct current modulation. The threshold gain, gth,

of the Fabry-Perot lasers is

where L_cav is the length of the

cavity, R is the facet reflectivity of both facets, and on is the internal optical loss.

This corresponds to a minimum modal gain of at least 25 cm^-1 for a typical 500

micron long cavity with uncoated facets and preferably greater than 40 cm^-1 for a

typical 300 micron long cavity with uncoated facets. Unfortunately, the ground state saturated gain reported by Lester, et. al., is less than 9 cm^-1. Referring to

FIG. 2A, it can be seen that the quantum dot laser structure disclosed by Lester,

et al. lases at a wavelength of about 1.05 microns for a cavity length of 500

microns or less, associated with an abrupt jump to lasing at higher excited

energy states.

[0012] A general problem with self-assembled quantum dot fabrication

processes is that heretofore the growth processes have not been understood well

enough to adjust quantum dot parameters that determine the optical gain

characteristics. Consequently, the optical gain spectrum of self -assembled

quantum dot lasers may lack sufficient saturated modal gain at a desired

wavelength, have a larger than desired separation between the ground state and

excited state transition energies, or have either too little or too much

inhomogenous broadening.

[0013] What is needed are techniques for forming self-assembled quantum

dot active regions having desirable optical gain characteristics.

SUMMARY OF THE INVENTION

[0014] Techniques for forming optical devices having layers of quantum dots

having desirable optical gain characteristics are disclosed. In one embodiment,

the quantum dots are self-assembled InAs quantum dots formed in InGaAs quantum wells that are grown on a GaAs substrate by molecular beam epitaxy.

A first barrier layer of AlGaAs or GaAs is grown. A first well layer of InGaAs is

grown on the first barrier layer. A sufficient monolayer equivalent thickness of

InAs is grown on the InGaAs to form self-assembled islands. A second well

layer of InGaAs is grown over the InAs islands to embed the quantum dots. A

second AlGaAs or GaAs barrier layer is then grown to complete the quantum

well. The optical gain characteristics of the quantum dot layers are influenced by

the compositional uniformity of surrounding layers, the dot size distribution, the

dot density, and the number of layers of dots that can be placed in an active

region without exceeding a critical thickness for forming dislocations.

[0015] In one embodiment, the density of dots is adjusted by selecting the

composition of the underlying well material that the dots nucleate on and by

selecting the growth temperature of the dots. The growth temperature also

influences the size distribution of the dots such that in one embodiment the

temperature and equivalent monolayer coverage of InAs for the dots is selected

to achieve a desired size distribution of the dots. In one embodiment, the well

material has an Indium alloy composition of between about InO.15Gao.ssAs to

Ino.₂oGao.sAs. In one embodiment the growth temperature of the dots is selected

to be in the range of between about 450 °C to 540 °C . [0016] In one embodiment, the compositional uniformity of the

underlying InGaAs is improved by depositing a floating layer of InAs to pre-

saturate the InGaAs, thereby permitting an extremely thin bottom well layer to

be grown prior to dot formation. The underlying bottom well layer of InGaAs

that the dots nucleate may have a thickness of about two nanometers or less, and

in one embodiment has a thickness of about one nanometer.

[0017] In one embodiment the spatial uniformity of the dots is improved

by performing a desorption process to desorb excess InAs from the top InGaAs

well layer that is grown over the dots. The desorption step may be carried out at

a temperature of between 560 °C to 650 °C. In one embodiment, the desoprtion

process is continued for a sufficient length of time to planarize portions of InAs

dots extending above the top InGaAs well layer, thereby improving the

uniformity of the InAs dots.

[0018] The growth temperature of layers grown subsequent to the

quantum dots may be selected to prevent a blue-shift of the emission wavelength

of the dots to shorter wavelength. In one embodiment, the growth temperature

of cladding layers grown subsequent to the dots is selected to be less than 610 °C.

[0019] In one embodiment, quantum dot lasers are formed having one or

more quantum dot layers. In one embodiment, the growth conditions are

selected to achieve a size distribution of the dots that produces substantial inhomogenous broadening of the optical gain spectrum, which is beneficial for

tunable laser and arrays of lasers having a wide range of operating wavelength.

In another embodiment, the layer structure and growth conditions are selected to

achieve a saturated ground state modal gain greater than about 25 cm^-1.

BRIEF DESCRIPTION OF THE DRAWINGS

[0020] FIGS. 1A, IB, 1C, and ID illustrate the density of states function for

bulk semiconductors, quantum wells, quantum wires, and quantum dots,

respectively.

[0021] FIG. 2A is a prior art plot showing quantum dot light output intensity

versus wavelength for quantum dot lasers of various lengths.

[0022] FIG. 2B is a prior art plot showing threshold current versus laser cavity

length for the lasers of FIG. 2A.

[0023] FIG. 3 is a perspective view illustrating an idealized quantum dot layer

of quantum dots embedded in a quantum well.

[0024] FIGS. 4A-4C each illustrate portions of a growth layer sequence form

forming quantum dot layers.

[0025] FIG. 5 is an atomic force microscopy image of InAs quantum dots. [0026] FIG. 6 shows plots of quantum dot density versus growth temperature

for two different compositions of a nucleation layer.

[0027] FIG. 7 is a plot showing photoluminescence intensity versus

wavelength of quantum dots for several different capping layer growth

temperatures.

[0028] FIG. 8 shows plots of photoluminescence intensity versus wavelength

of quantum dots for several different annealing temperatures and times.

[0029] FIG. 9 is a prior art plot of the surface segregated indium coverage for

different nominal InGaAs growth thicknesses.

[0030] FIG. 10 illustrates a growth sequence for pre-saturatihg a bottom

InGaAs well layer with a floating layer of indium.

[0031] FIG. 11 A is a side view illustrating how size variation in the thickness

of quantum dots may cause a portion of some quantum dots to protrude from

the quantum well into the barrier layer.

[0032] FIG. 11B illustrates a first growth sequence of layers with asymmetric

placement of dots within the well to facilitate embedding quantum dots in wells.

[0033] FIG. 11C illustrates a growth sequence of layer in which the quantum

dots are trimmed in thickness after the well layer is grown.

[0034] FIG. 12 illustrates a growth sequence for an active region having a

plurality of quantum dot layers. [0035] FIG. 13 shows a plot of a relationship for calculating a critical thickness

for a single layer.

[0036] FIG. 14A shows a growth sequence for a quantum dot laser.

[0037] FIG. 14 B shows a conduction band diagram for the laser structure of

FIG. 14A.

[0038] FIG. 15 shows a growth sequence for a quantum dot laser with

preferred temperature ranges for growing each layer.

[0039] FIG. 16 shows plots of measured modal gain versus threshold current

density for quantum dot laser having two different quantum dot densities.

[0040] FIG. 17 shows plots of ground state modal gain versus threshold

current density for quantum dot lasers having a single layer of quantum dots

with three different compositions of the InGaAs quantum well.

[0041] FIG. 18 shows a plot of modal gain versus threshold current density

for a quantum dot lasers having one and three layers of quantum dots.

[0042] FIG. 19 shows a growth sequence for quantum dot laser having a

sequence of quantum dot layers selected to achieve a high ground state modal

gain.

[0043] FIG. 20A illustrates transition energies for a single, ideal quantum dot.

[0044] FIG. 20B illustrates inhomogenous broadening of transition energies

associated with variance in quantum dot size for an ensemble of dots. [0045] FIG. 21 illustrates a method of forming a continuous quantum dot

optical gain spectrum over an extended wavelength range.

[0046] FIG. 22 shows plots of electroluminescence intensity versus

wavelength for a quantum dot laser structure with three different driver currents

selected to achieve a continuous optical gain spectrum over and extended

wavelength range.

[0047] FIG. 23 shows plots of wavelength tuning range for quantum dot and

quantum well lasers versus current.

[0048] FIG. 24 is a block diagram of a generic external cavity laser structure.

[0049] FIGS. 25A and 25B are block diagrams illustrating two external cavity

laser arrangements.

[0050] FIG. 26 is a perspective view of tunable laser.

[0051] FIG. 27A shows single wavelength laser structure with Bragg gratings.

[0052] FIG. 27B shows a comparison of wavelength response versus

temperature for conventional quantum well active regions and quantum dot

active regions compared with the wavelength response of a Bragg grating.

[0053] FIG. 28 shows a multi-wavelength array of lasers having a quantum

dot active region. [0054] FIG. 29 is a block diagram illustrating an embodiment in which the

multi-wavelength array of FIG. 28 is coupled by an optical multiplexer to an

optical fiber.

[0055] FIG. 30 shows plots of optical gain spectrum for an array at two

different operating temperature with corresponding DFB wavelengths

superimposed.

[0056] FIG. 31 is a side view of a quantum dot laser cavity.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0057] The present invention is directed towards techniques for fabricating

self-assembled layers of quantum dots for optical devices, including techniques

for improved control of quantum dot characteristics that determine the optical

gain characteristics, including the density of the dots, size distribution of the

dots, uniformity of the dots, position of dots within wells, and the strain-

thickness product that limits the spacing between neighboring layers of quantum

dots. Additionally, laser structures are disclosed having improved optical gain

characteristics.

[0058] FIG. 3 is a perspective view of an idealized quantum dot layer 330

that includes a multiplicity of individual quantum dots 320 embedded in a

quantum well layer 310 sandwiched between barrier layers 305. The individual quantum dots 320 comprise a low bandgap semiconductor material, the

quantum well layer 310 comprises a semiconductor with an intermediate

bandgap energy, and the barrier layers 305 comprise a high bandgap

semiconductor material. The semiconductor materials preferably comprise III-V

compound semiconductors.

[0059] It will be understood that the quantum dot layer may be included in

the active region of a p-i-n laser diode structure that includes an optical

waveguide structure to provide optical confinement for light generated in the

active region. For example, the laser structure may comprise a bottom optical

cladding layer grown having a first doping polarity; a first undoped

waveguiding core layer; a quantum dot active region, a second undoped

waveguiding core layer, and a top optical cladding layer of a second doping

polarity.

[0060] The macroscopic optical gain provided by a multiplicity quantum

dots 320 will depend upon the optical gain characteristics of the ensemble of

quantum dots. The modal gain of an active region including a single quantum

dot layer 330 or a sequence of quantum dot layers is the result of the cumulative

gain of the ensemble of dots within the active region and the coupling (often

known as the "optical confinement" for transverse optical waveguide modes) of

light to the quantum dots 320 due to other waveguiding layers (not shown in FIG. 3). The modal gain as a function of wavelength is often described as the

"optical gain spectrum."

[0061] An individual quantum dot 320 has an optical gain responsive to

an injected current associated with optical transitions of its quantum-confined

states. A single, ideal quantum dot would have a sharp atomic-like density of

states and would have an emission spectrum having sharp peaks at transition

energies determined by the size of the dot. Homogeneous broadening occurs

due to the effects of injected current, resulting in a broadening of the emission

energy from an individual dot by more than 20 meV.

[0062] Inhomogeneous broadening of the transition energies of an

ensemble of quantum dots occurs due to the influence of dot size. A quantum

dot can be modeled as a quantum box having a height, a width, and a length.

The height corresponds to the as-grown thickness of the dot. The width and

length correspond to the dimensions in the plane of the well. As is well known,

to a first order approximation, the first confined quantum states along each

dimension of an ideal quantum box have energy levels that vary inversely with

the square of the length, i.e. the energy levels, E ni,_n2,n₃, of an ideal quantum box

with infinite barriers having a length Lx, a width Ly, and a height Lz is: E _nι,n2,

_n3=ΔEo(nι²/Lx² + n^/Ly² + n₃ ²/Lz²), where ni, nz, and n₃ each integers equal to or

greater than 1 and ΔEois a material constant. For a quantum box with finite energy barriers, the separation in energy between quantum states tends to vary '

more slowly due to the penetration of the quantum mechanical wave function

into the barrier layers. For an ensemble of dots, the size of each dot varies about

a mean value, broadening the quantum confined energy levels for electrons and

holes in each dot.

[0063] Embodiments of the present invention include growth techniques

that permit improved control over the characteristics of individual dots,

individual quantum dot layers, and sequences of quantum dot layers. As

described below in more detail, these techniques facilitate forming laser

structures with desirable optical gain characteristics for various device

applications.

[0064] FIGS. 4A-4C are diagrams of growth layer sequences illustrating

some of the steps used to form self-assembled quantum dots layers, although it

will be understood that additional layers (e.g., waveguiding layers) may be

required to form a complete laser or optical amplifier structure. A preferred

growth technique is molecular beam epitaxy with the quantum dot layer grown

at a temperature between 450 °C to 540 °C. A conventional optical pyrometer

may be used to determine the temperature. The arsenic flux is preferably chosen

to achieve an arsenic stabilized surface. As shown in FIG. 4A, a bottom barrier

layer 402 is formed on a substrate 401. In one embodiment the substrate is a GaAs substrate and bottom barrier layer 402 is a layer of GaAs or a layer of

AlGaAs.

[0065] A bottom well layer 404, preferably comprised of InGaAs, is deposited

having a thickness, Twl. In one embodiment, the thickness of bottom well layer

404 is between about 0.5 to 5 nanometers. In one embodiment, the surface of the

bottom barrier layer 402 is pre-saturated with an Indium floating layer of about

0.5 to 1.6 monolayers prior to growth of the bottom InGaAs well layer.

[0066] As shown in FIG. 4B, a dot layer, preferably comprised of InAs, is

deposited on the bottom well layer 404, forming InAs islands 406. Since the InAs

has a relaxed lattice constant that is more than about 2% greater than the

underlying semiconductor layers, a Stranski-Krastanow (S-K) growth mode

occurs once a sufficient number of equivalent monolayers of InAs is deposited.

(As described below in more detail, in one embodiment the equivalent

monolayers of deposited InAs include the InAs deposited on the bottom well

layer 404 and the InAs monolayer thickness associated with depositing a pre-

saturation layer of InAs on bottom barrier layer 402 prior to growth of the

bottom well layer 404.) In the S-K growth mode, the driving force for the

formation of islands is the reduction in strain energy afforded by elastic

deformation, i.e., for S-K growth it is more energetically favorable to increase

surface energy by islanding than by relaxing strain by dislocation generation. In a S-K growth mode, the growth becomes three dimensional after a critical

thickness of the larger lattice constant material is grown upon an initial wetting

layer. Consequently, InAs islands 406 grow on the bottom InGaAs layer 404.

Typical ranges of equivalent InAs monolayer coverage for forming islands 404 is

between about 1.8 to 4.0 monolayers depending upon other growth parameters

and the desired dot size and density. A larger InAs equivalent monolayer

coveragage is used for denser and/or larger quantum dots. A thin wetting layer

405 of InAs may remain on the bottom InGaAs well layer 404 for some growth

conditions.

[0067] A top well layer 408, preferably comprised of InGaAs, is then grown to

cap (embed) the InAs islands 404. In one embodiment, the top InGaAs well layer

408 may have a thickness of between four to twelve nanometers. If desired, a

growth interruption may be performed after the dot layer to adjust the growth

temperature for the top well layer and/or to allow for Ostwald ripening of the

dots.

[0068] A top barrier layer 410 of GaAs or AlGaAs is then grown to complete

the quantum well. In one embodiment, after several monolayers of GaAs are

grown in the top barrier layer 410 a growth interruption is performed during

which the temperature is raised to between 580 °C to 650 °C to desorb excess Indium floating on the growth surface. The growth temperature of subsequent

layers may be selected to preserve the optical quality of the quantum dots.

[0069] The bandgap of an InGaAs well is intermediate between that of InAs

islands and surrounding GaAs or AlGaAs barrier layers. In one embodiment,

the IriχGaι-χAs has an alloy fraction of between about 0.1 to 0.3. A high indium

alloy fraction has the advantage of increasing carrier confinement to quantum

dots but also increases the strain of the layer. Each quantum dot consists of an

island of low bandgap material surrounded on all sides by a higher bandgap

material. It will be understood that interdiffusion, phase segregation, and

spinodal decomposition during the embedding process need to be taken into

account because they may affect the shape and composition of the quantum dots.

[0070] The InAs islands that form the core of the quantum dots in grown

structures have statistical characteristics that depend upon the growth

conditions. FIG. 5 is an atomic force microscopy image showing a perspective

view of InAs islands 505 deposited on an InGaAs layer. The islands were grown

by molecular beam epitaxy. The dots have a mean diameter of between about 20

to 30 nanometers in the plane of the InGaAs layer and a height (perpendicular to

the plane of the InGaAs layer) of about 7 nanometers. The dots have a width-to-

length ratio that is typically less than 2:1. It can be seen, however, that the dots

have a size distribution such that the dots can be characterized by a mean size and an associated variance. Over a sufficiently large area, the dots in a layer of

dots can also be characterized by a dot density (e.g., dots per unit area).

[0071] FIG. 6 is a plot of quantum dot density versus growth temperature for

quantum dots grown on two different InGaAs well layer compositions. It can be

seen that the dot density depends strongly upon temperature and also upon the

composition of the bottom well layer. Dot densities of greater than 1 x 10" cm^~2

may be achieved at a growth temperature of about 470 °C. The dot density can

be adjusted by more than a factor of five by selecting a growth temperature

between 470 °C to 540 °C. Experiments indicate that the dot density is at least a

factor of two higher when the dots are grown on an InGaAs layer compared with

a GaAs layer at a comparable temperature. The dot density also increases when

the InGaAs alloy composition is increased from Ino.ιGao. As to Ino.₂Gao.sAs.

Experiments by the inventors indicate that the thickness of the bottom InGaAs

well layer may be extremely thin and still have the same effect as a thick layer in

regards to the nucleation of quantum dots on the bottom InGaAs layer. Thus, to

achieve a reproducible dot density, the bottom well layer need only have a

thickness consistent with it having a reproducible thickness and alloy

composition. The bottom well layer may have a thickness as low as 0.5 nm,

although a thickness of about one nanometer may be easier to reproducibly

grow. [0072] As can be seen in FIG. 6, there is a strong effect of growth temperature

on dot density. Thus, the quantum dot density may be selected over a

considerable range by choosing the composition of the bottom InGaAs layer and

the growth temperature. Additionally, the dot size increases and the dot density

decreases with increased growth temperature. The dots also tend to become

more uniform in size with increasing growth temperature. The variance in dot

size with growth conditions can be characterized with atomic force microscopy.

However, for structures with the dots embedded in a higher bandgap material,

the variance in dot size may be inferred from photoluminescence measurements

of the full width half maximum (FWHM) for different growth conditions.

Generally speaking, a larger variance in quantum dot size tends causes a

corresponding increase in photoluminescence FWHM. Thus, for a particular

application, a series of quantum dots may be grown under different temperature

and the FWHM of photoluminescence used as an indication of the size variation.

[0073] The growth conditions of layers grown subsequent to the quantum

dots influences the optical quality of the quantum dots. In particular, the growth

temperature of the top quantum well layer influences the photoluminescence

quality of the quantum dots. FIG. 7 shows photolumiscence measurements of

InAs quantum dot test structures having upper well InGaAs layers grown at

different temperatures. The change in photoluminescence is attributed to decomposition of the upper (capping) well layer at elevated temperatures.

Consequently, in one embodiment the growth temperature of the upper well

layer is selected to improve the optical quality of the quantum dot layer. In one

embodiment of a laser, the growth temperature of the upper well layer is

selected to be between about 460 °C to 490 °C, since this temperature range

facilitates achieving a combination of both good material quality and a ground

state transition energy corresponding to a comparatively long emission

wavelength.

[0074] The growth of thick layers subsequent to the quantum dot layers may

also influence the optical characteristics of the quantum dots. FIG. 8 shows

photoluminescence spectra for quantum dot test structures annealed at different

temperatures to simulate the growth of waveguide cladding layers in a laser

structure. A considerable blue-shift (shift to shorter wavelengths) occurs for high

annealing temperatures. This indicates that for a long wavelength laser (e.g., one

having a wavelength beyond 1.24 microns) it is desirable to select a

comparatively low growth temperature of cladding layers grown subsequent to

the quantum dot layers (e.g., a growth temperature of AlGaAs cladding layers of

less than about 610 °C).

[0075] It is desirable that the surface of the InGaAs semiconductor comprising

bottom layer 404 have a reproducible InGaAs alloy compositon. This is, in part, due to the strong dependence of dot density on the InGaAs alloy composition of

bottom layer 404, as shown in FIG. 6. However, it is known in the art of field

effect transistors that when an InGaAs is grown on top of a layer that does not

include indium, such as GaAs or AlGaAs, that the InGaAs may have a graded

alloy composition for a finite thickness. This is because indium tends to

segregate on the surface of InGaAs during molecular beam epitaxy growth. The

segregated indium is sometimes known as a "floating layer" of indium. The

amount of indium in the floating layer eventually builds up to a steady state

value for a particular growth temperature and nominal (thick film) InGaAs

growth temperature. For an Ino.₂Gao.sAs grown at around 500 °C, the floating

layer of Indium may correspond to about one monolayer of InAs. FIG. 9 shows a

prior art calculation of the segregated indium (equivalent monolayer coverage)

versus nominal InGaAs thickness. It can be seen that three to four nanometers of

InGaAs may have to be grown to reach steady state conditions if the initial

concentration of Indium on the surface (where Xo, in monolayers (MLs) is zero).

[0076] FIG. 10 shows an embodiment of a growth sequence in which the

growth is interrupted subsequent to bottom barrier layer 402 and a floating layer

of InAs 1010 is deposited having an Indium monolayer thickness approximating

the equilibrium monolayer coverage (e.g., about one monolayer equivalent of

InAs, although an equivalent monolayer coverage between 0.5 to 1.5 may be beneficial). The bottom well layer 404 is then grown. As indicated in FIG. 9, if

the InGaAs layer begins with an indium monolayer coverage close to its

equilibrium value, only a comparatively thin layer of InGaAs need to be grown

to achieve a reproducible alloy composition. Consequently, the inventors have

recognized that a a bottom well layer 404 may be grown having a thickness of

less than two nanometers while also achieving a reproducible alloy composition

of the surface of bottom well layer 404. In one embodiment, bottom well layer

404 has a thickness of one nanometer or less. Note that the floating layer of InAs

1010 contributes to the total InAs monolayer coverage on the surface which

results in the formation of dots when InAs is deposited for forming dots 406.

Consequently, it will be understood that in one embodiment the equivalent

monolayer coverage for forming InAs dots 406 includes the InAs monolayers

deposited after the growth of bottom well layer 404 combined with the InAs

monolayer coverage of presaturation floating layer 1010. As an illustrative

example, if the desired equivalent InAs monolayer coverage for forming

quantum dots is 2.4 monolayers and one monolayer of InAs is deposited as a

floating layer of InAs 1010, then only about 1.4 monolayers of InAs needs to be

deposited subsequent to the growth of bottom well layer 404 to achieve the

desired equivalent monolayer coverage of InAs of 2.4 equivalent monolayers. [0077] Photoluminescence studies by the inventors indicate that it is desirable

that all of the quantum dots not significantly penetrate the upper barrier layer

410. FIG. 11A illustrate a portion of the growth process after the top InGaAs well

layer 408 has been grown over InAs islands 406. As previously discussed, there

is a random size distribution of the InAs islands 406 due to the statistical nature

of the growth process. Consequently, some of the islands 406 may have

protruding portions 1105 that are not embedded within top well layer 408 unless

well layer 408 is made thicker than the largest thickness variation of islands 406.

Studies by the inventors of the present patent application indicate that

protruding portions 1105 may reduce the optical quality of the structure,

especially upon annealing, and may have other deleterious effects. Since

protruding portions 1105 may cause a substantial change in the optical

properties of its corresponding quantum dot 406 it is desirable to reduce the

percentage of dots having a protruding portion 1105.

[0078] FIG. 1 IB is a side view showing a layer sequence for ensuring that the

quantum dots are completely embedded in the quantum well. In the

embodiment shown in FIG. 11B, the thickness, d₂, of the upper well layer 408, is

selected to be greater than the mean dot thickness td. Note that the dots do not

have to be symmetrically placed in the center of the well. For example, in one

embodiment, d₂ is much thicker than the thickness, di, of lower well layer 404. In one embodiment, the thickness of bottom well layer 404 is selected to be as thin

as practical while achieving a reproducible composition (e.g., 1 to 2 nanometers

thickness). As an illustrative example, lower well layer 404 may have a nominal

thickness of 2 nanometers and the upper well layer 408 may have a nominal

thickness of 8 nanometers. For a constant total well thickness, tw, this

configuration permits the widest possible variation in dot height with all of the

dots embedded in the well.

[0079] FIG. 11C illustrates a growth sequence for an embodiment in which

protruding portions 1105 of dots 406 are removed using a desorption step. In

this embodiment after the upper well layer 408 is grown the growth is

interrupted at a sufficiently high temperature and for a sufficient length of time

to desorb InAs from protruding regions 1105 (shown in phantom in FIG. 11C

because they are removed during the desorption process resulting in a

substantially planar surface). In one embodiment, a small number of monolayers

of GaAs may be grown prior to the desorption process. The growth is resumed

after the desorption step and the barrier layer 410 is grown.

[0080] Photoluminescence or other characterization techniques may be used

to select a desorption time and temperature that removes protruding regions

1105. Typical times and temperatures are a desorption temperature of 590 °C for

one minute at a sufficient arsenic flux to maintain an arsenic stable GaAs surface. In one embodiment, the temperature is selected so that a GaAs surface is stable

but an InAs surface desorbs. As previously described, in one embodiment, a

small number of monolayers of GaAs may be deposited prior to the desorption

process to facilitate preserving the quality of planar portions of the upper

InGaAs well layer during the desorption process. As indicated by the hatched

regions, the desorption step ensures that the resulting quantum dots 406 do not

extend into the upper barrier layer 410 in the final structure. Photoluminescence

experiments indicate that the processs illustrated in FIG. 11C has a

photoluminescence FWHM for the quantum dots of about 25 to 35 meV, which is

about a factor of two lower than the FWHM for quantum dots formed using the

process illustrated in FIG. 11B, which is attributed to an improved uniformity in

quantum dot thickness. Consequently, the embodiment shown in FIG. 11B

provides two benefits: it eliminates protruding portions and also improves the

thickness uniformity of the quantum dots. In one embodiment, the mean

thickness of the as-grown islands is selected to be greater than the thickness of

the upper InGaAs well layer. Consequently, in this embodiment all of the dots

are trimmed in height by the desoption step and will have a thickness that is

approximately the same as the upper InGaAs well layer.

[0081] One benefit of the growth techniques of the present invention is that

they permit a high density of quantum dots to be grown in a comparatively thin quantum well, which reduced the strain-thickness product to achieve a particular

dot density and corresponding peak optical gain of the quantum dot layer.

Moreover, this can be simultaneously achieved at comparatively long ground

state emission wavelengths of commercial interest, such as emission wavelengths

of at least 1260 nanometers. For example, with a well thickness of 9 nanometers,

a well Indium alloy fraction of 0.15, an InAs monolayer coverage of 2.1

monolayers grown at a temperature corresponding to a dot density of 1 x 10¹¹ per

square centimeter, the corresponding average strain-thickness product

(calculating an average indium alloy fraction in the well of 20% by a weighted

average of the equivalent InAs thickness associated with the InAs dots and the

InGaAs well layers) is the strain associated with the average composition (1.45%)

multiplied by the well thickness (90 Angstroms), or a strain thickness product of

130.5 A-%.

[0082] FIG. 12 shows a growth layer sequence and layer thicknesses for an

active region 1200 of an optical device. It will be understood that active region

1200 may comprise part or all of a waveguide core within other optical

waveguide cladding layers. The capability to form individual quantum dot

layers having a low strain-thickness product facilitates stacking four or more

quantum dot layers in the active region of a suitable laser waveguide having a core waveguide thickness, Tactive, of between about 200 nanometers to 300

nanometers.

[0083] An important limitation in designing an individual quantum dot layer

or a sequence of quantum dot layers is the strain-thickness product of the active

region. An InGaAs layer grown on a GaAs substrate is a strained layer. If the

strain-thickness product of the strained layer is sufficiently low, a high quality

strained layer may be achieved. However, if the strain-thickness product is

excessively high, deleterious misfit dislocations tend to form if a critical thickness

is exceeded. There are a variety of techniques used in the art to calculate the

critical thickness of a single strained layer.

[0084] FIG. 13 is a plot of the commonly used Matthews-Blakeslee

relationship for calculating a critical thickness for strained layers having a

uniform alloy composition (which may be used to calculate a lower bound for an

embedded quantum dot layer). Curve 1305 is a curve indicating a critical

thickness. Regions above curve 1305 tend to form dislocations. The Matthews-

Blakeslee curve 1305 indicates that a single quantum dot layer having an average

composition of well layers and dots of Ino.₂Gao.sAs has a critical thickness of at

most 12 nanometers. However, provided the quantum dots are not dislocated

themselves, the Matthews-Blakeslee critical thickness is a conservative estimate

for non-equilibrium crystal growth. [0085] Additionally, for a sequence of quantum dot layers of an active region

1200, an average strain-thickness product should be below a threshold average

strain (e.g., 0.5%). The strain thickness product of an individual quantum dot

layer is EwTw, where Ew is the strain of a well layer and Tw is the thickness of the

well. The strain thickness product of an individual barrier layer is EbTb, where Eb

is the strain of the barrier layer and Tb is the thickness of the barrier layer. For a

sequence of n layers of dots, the average strain, Eav, is:

_[0086] R^ (» + Wπ + ^ήEwT* (n + ϊ)Tb + nTw ^

[0087] For GaAs barriers (which are unstrained), this simplifies to:

[0089] Equation 1 can be re-expressed as a relationship between the barrier

thickness, well thickness, modified average strain, strain in the barriers, and

strain in the well:

[0090] τb₌ nTw(Ew-Eav) ₃

¹ (n + ϊ)(Eav- Eb) ^

[0091] For GaAs or AlGaAs layers grown on a GaAs substrate Eb »0 so that

the barrier thickness is:

[0092] Tb= ^nTw(Ew-^Εav E_qΛ. (n + l)Eav [0093] Equation 4 can be used to derive a relationship for a minimum barrier

layer thickness. If the average strain is selected to be less than a maximum

average strain (for example, and average strain less than about 0.51%), Eavmax,then

the following relationship holds:

_r~~ _^ rr-n Tw[n(Ew -Eav max)] ^ _

[0094] Tb> — — -Eq.5.

(n + l)Eα max

[0095] As an illustrative example, if Eavmax is 0.4 and Ew =1.45 for an average In

alloy composition of about Ino.₂Gao.sAs then Tb>2.625Tw(n/(n+l)). If Tw is 9 nm

for a structure, then the minimum barrier thickness for a structure with 6

quantum dot layers is about 20 nanometers.

[0096] The quantum dots may be utilized in a variety of laser structures. For

example, referring to FIG. 31, AlGaAs cladding layers 3120 and 3125 may be

used to provide vertical optical confinement of light to a quantum dot active

region 3105 of an optical cavity having a length L. Optical feedback for the

optical cavity may be provided using any suitable means, such as reflective facets

or a Bragg grating. Lateral optical confinement may be provided using any

suitable process. For example, a ridge waveguide may be used to provide lateral

optical confinement. Individuals lasers may be fabricated as Fabry-Perot lasers,

distributed bragg reflector lasers, distributed feedback lasers, or external cavity

lasers. By selecting growth conditions to avoid deleterious blue-shifting, the size of the quantum dots (height, width, and length) may be selected to provide gain

at a variety of wavelengths, such as the commercially important 1.3 wavelength

range.

[0097] FIG. 14A is an illustration of a growth sequence for a laser having a

single layer of quantum dots and FIG. 14B shows a corresponding conduction

band energy diagram. It will be understood from the conduction band energy

diagram that the quantum well assists the quantum dots to capture and retain

injected carriers due to the lower bandgap energy of the central quantum well

layer compared with surrounding quantum well barrier layers. Referring back to

FIG. 14A, an n-type GaAs buffer layer is grown on a GaAs substrate. An

approximately two micron thick first AlGaAs cladding layer is then grown. This

is followed by an undoped GaAs waveguiding layer about 0.11 microns thick,

which is preferably undoped to reduce absorption losses. A presaturation layer

1415 comprising approximately one monolayer of InAs is deposited during a

growth interruption step. The growth temperature is adjusted to approximately

490 °C during the growth interruption. An approximately 2 nanometer InGaAs

bottom well layer 1420 is then grown. An InAs layer corresponding to a

monolayer coverage of about 1.4 monolayers is deposited to form InAs islands.

Note that since a one monolayer presaturation layer 1415 of InAs was deposited

prior to the bottom well layer 1420 that the total InAs coverage on the surface of the bottom InGaAs well that produces the InAs islands is approximately 2.4

equivalent monolayers. An approximately 7.6 nanometer thick InGaAs top well

layer 1440 is then grown. In one embodiment, several monolayers of GaAs are

then grown, followed by a growth interruption step in which the substrate

temperature is raised to about 610 °C. The growth interruption step preferably

lasts sufficiently long to desorb excess segregated indium from the surface prior

to commencing growth of a second GaAs waveguidng layer 1445 which has

about the same thickeness as the first waveguiding layer. An upper AlGaAs

cladding layer 1450 is then grown, followed by a GaAs cap layer 1455.

[0098] Layers 1410, 1420, 1430, 1440, and 1445 form a waveguide core region

having a higher refractive index than surrounding AlGaAs cladding layers 1405

and 1450. Consequently, a transverse optical mode will be confined by the laser

structure. A fraction of the waveguide mode wβl be confined in the portion of

the structure occupied by the quantum dots.

[0099] The p-type layers, undoped layers, n-type layers form a p-i-n laser

structure. While one substrate polarity is shown, it will be understood that

doping polarity of the layers may be reversed from what is shown. The quantum

well layers 1420 and 1440 provide an additional benefit of providing a means to

improve carrier capture to the quantum dots and also serve to reduce thermionic

emission of carriers out of the dots. This provides an important benefit for quantum dots lasers. In a quantum dot laser the fill factor of quantum dots in an

individual quantum dot layer is low, typically less than 10%, depending upon

the dot density and mean dot size. In some cases, the fill factor is less than 5%.

Since the quantum dots are disposed within the quantum wells, carriers captured

by the well layer of the quantum wells may enter the quantum dots.

Additionally, the barrier layers of the quantum well also serve to reduce

thermionic emission out of the quantum dots.

[00100] FIG. 15 shows a similar layer sequence as FIG. 14. However, as

indicated by FIG. 15, the growth temperature of the active region may be varied

to control the density of quantum dots.

[00101] FIG. 16 shows plots of measured modal gain versus current density for

laser structures grown at growth temperatures producing two different dot

densities. Referring to FIG. 16, experiments data indicates that the saturated

ground state modal gain increases approximately linearly with dot density.

[00102] FIG. 17 shows plots of ground state modal gain for a laser structure

similar to that of FIG. 14 with the alloy composition of the InGaAs well layers

varied. It can be seen that the saturated (maximum) ground state modal gain

increases as the In alloy fraction increases.

[00103] As previously discussed, multiple quantum dot layers can be used in

an active region as long as the separation between quantum dot layers is large enough that the strain within a quantum well remains below a critical thickness

for forming dislocation and the average strain with the waveguide core remains

below a threshold average strain. FIG. 18 shows a plot 1805 of modal gain versus

current density for a laser structure having a single layer of quantum dots and a

corresponding plot 1820 for a laser structure having three quantum dot layers,

with the quantum well of each quantum dot layer separated from its neighboring

quantum dot layers a 10 nanometer thick barrier layer. For this example, the

quantum dot layers are spaced sufficiently close that the optical confinement for

each layer of quantum dots is about the same as for a single layer of quantum

dots, resulting in a cumulative three-fold increase in total modal gain.

[00104] The growth conditions and layer sequence of a quantum dot laser may

be adapted for a particular application. In some applications a comparatively

high ground state saturated modal gain is desired. For example, in some

commercial laser applications, such as OC-48 and OC-192 compliant lasers, an

emission wavelength of at least 1260 nanometers and a cavity length of no more

than 500 microns is required, with a cavity length of 300 microns being desirable

to reduce the photon lifetime. This wavelength may be achieved by designing

the laser to operate off of the ground state (longest wavelength emission) and by

selecting growth conditions that minimize blue-shifting of the quantum dots

during growth (e.g., by appropriately selecting the growth conditions of cap layer and thick cladding to minimize blue-shifting as described above).

Moreover, the dot height, dot diameter, dot composition and the InGaAs well

width and composition influence the ground state transition energy and may be

selected to achieve a ground state optical transition having an emission

wavelength of greater than 1260 nanometers.

[00105] One benefit of the quantum dot layers of the present invention is that

they have an extremely low linewidth enhancement factor. A low linewidth

enhancement factor reduces the deleterious chirp when a laser is directly

modulated by changing its drive current (the current injected into the p-i-n laser

diode to change its light output. The linewidth enhancement factor, , describes

the degree to which variations in the carrier density, N, alter the index of

refraction, n, of an active layer for a particular gain, g, at the lasing wavelength, Λ

. The linewidth enhancement factor can be expressed mathematically as:

[00107] Experiments by the inventors indicate that the linewidth enhancement

factor of the quantum dot lasers is as low as 0.1, which almost twenty times

lower than for comparable quantum well lasers at about the same wavelength.

The low linewidth enhancement factor correspondingly reduces wavelength

chirp for high frequency direct modulation of the lasers. Chirp becomes a progressively more severe problem in conventional lasers used in fiber-optics

systems as the modulation data rate increases. Consequently, in some

conventional fiber optics system, an external modulator (a modulator distinct

from the laser) is used to modulate a constant power output of a laser. By way of

contrast, a low linewidth enhancement factor of less than 0.2 for the quantum dot

lasers of the present invention would permit direct high speed modulation of the

lasers with extremely little chirp. This makes the quantum dot lasers of the

present invention attractive for applications in which the laser has a wavelength

of at least 1260 nanometers and is directly modulated at data rates of greater than

one Gbps, thereby eliminating the need for external modulators.

[00108] A variety of commercial applications, such as OC-48 and OC-192

compliant lasers, require modulation rates in excess of one Gbps. In a directly

modulated laser the relaxation oscillation frequency, fr is given by the

expression:

[00109] fr= , where P is the photon density, η the effective

refractive index of the waveguide, c the speed of light in vacuum, τ the photon

lifetime, and dg/dN the differential gain.

[00110] For direct modulation of a quantum dot laser, a low photon lifetime is

required for a laser that is directly modulated at high data rates, which implies a cavity length of less than 500 microns and preferably no more than 300 microns.

The saturated ground state modal gain should be at least about 25 cπr¹ for a 500

micron long cavity with 10%/90% facet reflectivity of front and rear facets,

respectively; and at least 40 cm^-1 for a 300 micron long cavities with 10%/90%

facet reflectivity of front and rear facets, respectively. It is also desirable that the

lasers be operated at a threshold gain that is selected to be sufficiently below the

saturated ground state modal gain that the differential gain (dg/dN) still remains

high.

[00111] For applications requiring a high saturated ground state modal gain, a

plurality of quantum dot layers are preferably included within a waveguide core

region no greater than about 300 nanometers in thickness, since a waveguide

core thickness of between about 200 to 300 nanometers provides the optimum

optical confinement per quantum dot layer. (Note that a waveguide core having

a thickness between 200 to 300 nanometers corresponds to a distance comparable

to about a half wavelength for 1.3 micron emission for the refractive index of

AlGaAs waveguiding layers, for which a high optical confinement per quantum

dot layer may be achieved). Since a saturated ground state modal gain of greater

than 7 cm^-1 can be obtained for a single quantum dot layer, a total number of

quantum dot layers in the range of four to eight is sufficient for many

applications requiring a high peak optical gain. An exemplary growth sequence of a laser having 4 to 8 quantum dot layers is shown in FIG. 19. The total

waveguide core thickness is preferably in the range of about 200 to 300

nanometers, since this provides the highest optical confinement per quantum dot

layer. The corresponding AlGaAs cladding layers are preferably ALGai_→As

cladding layers with the molar fraction, x, between x=0.6 to x= 0.8 (which can also

be described in percentage terms as an aluminum arsenide percentage of

between 60% to 80%), since this range tends to produce a high optical

confinement consistent with growing high quality optical layers. However, it

will be understood that the AlGaAs cladding layers may be grown with a lower

molar fraction of aluminum but that the optical confinement may decrease,

somewhat, as the molar fraction of aluminum in the AlGaAs is decreased.

[00112] It will be understood that a directly modulated laser structure may use

any known packaging and contacts for supplying a modulated current to the

laser p-i-n diode. For example, the laser may be a ridge waveguide laser

packaged to provided a current to the diode that may be modulated at rates in

excess of one Gbps.

[00113] It will also be understood that there are tradeoffs between growth

conditions that increase dot density and the size uniformity of the dots. A high

number density of dots in each laser increases the number of dots available to

provide gain. However, this is offset, somewhat, by the fact that growth conditions that increase the dot density may also increase the size variance of the

dots. For a laser structure with a high saturated ground state optical gain it is

desirable to select growth conditions consistent with a high dot density per

quantum dot layer with as high a dot size uniformity (minimal inhomogenous

broadening) as possible. Consequently, in one embodiment of a multiple

quantum dot layer laser, the growth sequence of each quantum dot layer is

selected to trim the quantum dots similar to the embodiment shown in FIG. 11C.

[00114] In some laser applications it is desirable to have an optical gain

spectrum that extends over a substantial range of wavelengths and at a

comparatively low pumping level. For this situation, a comparatively low

saturated ground state modal gain may be desirable to facilitate wavelength

tuning. FIG. 20A is an illustrative diagram of optical transition energies for a

single quantum dot. An individual quantum dot has quantum confined energy

states associated with its length, width, and height. There will be a ground state

transition energy, a first excited state transition energy, a second excited state

transition energy, and possibly additional transition energies, depending upon

the dot size, composition, and shape. For InAs quantum dots with a nominal 7

nanometer height and 30 nanometer diameter, the difference in transition energy

between the ground state transition and the first excited state transition is about

80 meV. Referring to FIG. 20B, for an ensemble of quantum dots having a distribution in size about a mean size, there will be a corresponding

inhomogeneous broadening, as indicated by the hatched areas. When the

quantum dots are pumped, homogeneous broadening will also occur. The

degree of inhomogenous broadening can be adjusted be selecting the growth

conditions that increase or decrease the size variation of the dots.

Photoluminescence studies indicate that the growth conditions may be adjusted

to vary the inhomogeneous broadening from a minimum of about 30 meV to at

least 70 meV. Homogenous broadening of greater than 24 meV may be achieved

at pump levels of several kA/cm². The homogenous broadening may be

attributed, for example, to collisions of electrons with phonons and other

electrons.

[00115] The interplay of homogeneous and inhomogeneous broadening acts to

broaden the optical gain spectrum. The interplay of homogenous and

inhomogenous broadening may more than double the width of the gain

response. A lasing-mode photon will receive gain from not only the

energetically resonant dots but also from other non-resonant dots that lie within

the range of homogenous broadening. Consequently, if the difference between

the central ground state transition energy and successive excited states is the

range of between about 30 meV to 80 meV, a continuous optical gain spectrum

may be achieved at a comparatively low current density because the combination C of homogeneous and inhomogeneous broadening produces a continuous gain

spectrum.

[00116] In one embodiment of the present invention, the growth parameters

are selected to achieve a difference in energy value of successive quantum dot

transition energies that is between about 30 to 80 meV. In particular, in one

embodiment, growth conditions may be selected to vary the mean size of the

quantum dots to adjust their transition energies. The growth conditions are

preferably further selected to achieve a sufficient degree of inhomogeneous

broadening. In one embodiment, the inhomogeneous broadening is selected to

be at least half the difference in energy value between the ground state transition

energy value and the first excited energy state value. In another embodiment,

the inhomogeneous broadening is selected to be about 20-30 meV below the

difference in energy levels such that the combination of homogeneous and

inhomogeneous broadening forms a continuous optical gain spectrum at

comparatively low threshold current.

[00117] A semiconductor laser including a quantum dot active region may be

designed to operate over a wide range of wavelengths at a comparatively low

current. FIG. 21 is an illustrative plot 2110 of cavity optical gain (e²s^L where g is

the gain per unit length and L is the length of the gain medium in the cavity) of a

laser quantum dot active region at a selected current. Referring to FIG. 21, a quantum dot active region may be designed to have a ground state quantum

level corresponding to a peak 2105 at a wavelength λi. In the illustrative

diagram the current is selected such that the ground state modal gain at λi is

saturated with a saturated value greater than g in, a minimum gain, to overcome

a resonator loss that includes a mirror loss and an absorption loss. If an extended

tuning range is desired, the drive current is selected to populate the excited

states. For example, the first excited quantum states (e.g., first excited state 2110

and second excited state 2115) associated with quantum confinement along the

length and width of the dots may be designed to provide additional higher

energy states, as indicated by the peaks 2110 and 2115 between Λi and . By way

of contrast, a conventional quantum well active region providing gain over a

comparable wavelength would not have a saturable gain at the first ground state

of the quantum well, necessitating an extremely high current density to achieve

the minimum gain over the entire wavelength range. Rough estimates by the

inventors indicate that a quantum dot laser can achieve a wide tuning range with

approximately ten times less current than a comparable quantum well laser. In

one embodiment the inhomogeneous broadening and the difference in energy of

the ground state transition energy, the first excited state transition energy, and

the second excited state transition energy is selected to permit a continuous

optical gain spectrum of at least 150 nanometers for a maximum threshold current density. The drive current is preferably selected to at least saturate the

ground state modal gain. In one embodiment with multiple layers of quantum

dots, the growth parameters of the dots in different layers may be adjusted such

that each layer of quantum dots has a different sequence of optical transition

energies selected to form a sequence of transition energies for the layers that

facilitates forming a broad, continuous optical gain spectrum. For example, a

first quantum dot layer could have a first ground state transition energy and

associated excited state transition energy levels. The second quantum dot layer

could have dot characteristics (e.g., size, shape, and composition) selected to

have its ground state transition energy and excited state transition energy levels

(e.g., first and second excited state transition energy) offset in value (e.g., by 10-

40 meV) with respect to corresponding transition energy levels of the first layer

of quantum dots.

[00118] FIG. 22 is plot of amplified spontaneous emission intensity versus

wavelength for a laser cavity fabricated from a structure similar to that shown in

FIG. 14B and having a single quantum dot layer. The nominal dot height is

about 7 nanometers grown at a temperature of approximately 490 °C with a dot

density corresponding to about 7.5 x 10¹⁰ cm². The front facet is anti-reflection

coating to suppress Fabry-Perot lasing and the cavity length is about 1.7 mm.

Plot 2205 is for a drive current of 500 mA, plot 2210 is for a drive current of 600 mA, and plot 2215 is for a drive current of 700 mA. The drive current for all

three plots is sufficiently high that the ground state modal gain is saturated,

which accounts for the convergence of the curves between 1.20 to 1.30 microns.

The drive current is sufficiently large that excited quantum dot states are

populated, corresponding to a continuous gain over more than 200 nanometers.

At the higher pump levels of plots 2210 and 2215, the onset of lasing at the

second excited state (1.05 microns) begins to occur.

[00119] FIG. 23 is an illustrative plot of tuning range versus current density for

a nominal 1.3 micron wavelength laser tuned to shorter wavelengths. The

quantum dot plot 2380 is based upon experimental data of the inventors. The

quantum well plot 2390 is based upon various numbers published in the

literature. Referring to plot 2280, quantum dot lasers permit about 70 nm of

tuning per each 1 kA/cm² increase in pump current density. By way of contrast a

conventional quantum well requires about 2.3 kA/ cm² to achieve a 70 nm tuning

range. It can be seen in plot 2390 that the quantum well has an abrupt increase

in threshold current range when the tuning range exceeds 10% of the nominal

wavelength (150 nm for this case). Above a 10% tuning range, the bias current

increases dramatically, typically exceeding 10 kA/ cm². However, for a long

lifetime operation, quantum well lasers used in tunable lasers are commonly

operated with a maximum current density corresponding to about a 70 nm tuning range. As can be seen in plot 2380 a quantum dot active region of the

present invention may be tuned to greater than 150 nm (e.g., greater than 200

nm) with current densities in the 2-3 kA/ cm² range. This permits, for example, a

single quantum dot laser to be used to achieve a tuning range of 180 nanometers

at a current density of less than about 3 kA/ cm² which is impractical with

conventional quantum well lasers.

[00120] A quantum dot laser of the present invention may be used as the active

gain medium of a tunable external cavity laser. FIG. 24 is a top view of a generic

external cavity laser. The threshold condition for lasing is that the summation of

the resonator losses (mirror and internal losses) is balanced by the optical gain of

the gain medium. In a tunable laser, a wavelength selector 2440 is included that

has a reflectivity that is a function of wavelength. Typically a rear facet 2405 of a

laser diode 2402 retains a high reflectivity whereas a front facet 2410 is processed

to have an extremely low reflectivity. The threshold condition for the external

cavity system is: Rιe²te-^αi>^L1R(λ)e⁽-^2αeL2)=l, where RI is the reflectivity of the rear

facet of the laser diode, g is the gain per unit length of the laser diode, αi is the

loss per unit length of the laser diode, LI the cavity length of the laser diode, R(λ)

the reflectivity of the wavelength selector, αe is the average loss per unit length

of the external cavity, and L2 is the effective length of the external cavity. [00121] In an external cavity tunable laser, the wavelength selector may be any

combination of external elements whose transmissivity or reflectivity response as

a function of wavelength may be controlled such as to limit the optical feedback

to the laser to a narrow wavelength band. Tunable external cavity lasers

typically use an external grating arranged to provide wavelength-selective

optical feedback based on wavelength-selective dispersion. Referring to FIGS.

25A and 25B, in an external cavity semiconductor laser typically one facet of a

laser diode is coated with a high reflectivity (HR) material and the other facet is

coated with an antireflection (AR) coating. An external grating is typically

arranged to provide wavelength selective feedback back to the laser diode and to

also serve as an output coupler. The reflectivity of the AR coating and the length

of the laser cavity are preferably selected to suppress Fabry-Perot (FP) modes in

the laser. It will also be understood that any known technique to suppress

Fabry-Perot modes from the cleaved facets over a range of wavelength may also

be used, such as incorporating unguided window regions or tilting the laser

stripe with respect to the facet during device fabrication. The length of the

semiconductor laser and residual reflectivity of the AR coated facet is preferably

selected such that the external grating solely determines the lasing wavelength.

In one embodiment, the length of the semiconductor laser and the residual

reflectivity of the AR coated facet is selected so that the laser is incapable of lasing off the ground state and at least the first excited state of the quantum dots

without feedback from the external grating.

[00122] In an external cavity laser, the minimum gain must exceed the cavity

losses. Thus, referring back to FIG. 24, to achieve a wide tuning range the

saturated laser gain for the ground state should be selected to be greater than the

total resonator losses, which include the absorption losses and mirror losses.

However, the grating selected wavelength only depletes the optical gain within

the homogenous gain broadening range such that the grating selected

wavelength only suppress the Fabry-Perot mode only within approximately 20

nm of the free-run emission wavelength. Consequently, in one embodiment, the

peak gain at high quantum states, g_P, is preferably selected to be sufficiently low

to prevent deleterious lasing from FP modes due to residual facet reflectivity of

the front facet.

[00123] Two common external cavity configuration are the Littman-Metcalf

external cavity and the Littrow external cavity. FIG. 25B is an illustration of a

Littman-Metcalf laser cavity. FIG. 25A is an illustration of a Littrow laser cavity

2508. In the Littrow cavity the angle of incidence of the light received by the

grating 2530 from the laser 2510 is such that the beam is reflected back to the

laser serving the function of one mirror of the laser cavity. The angle of the

grating determines the wavelength. An etalon is sometimes included in a Littrow configuration to decrease the cavity bandwidth. In the Littman-Metcalf

configuration 2500, the grating 2530 diffracts the light towards a tuning mirror

2510 (also known as a retroreflector), which determines the feedback.

Collimating lenses 2505 are typically included in an external cavity laser to

improve the coupling of the laser output to the grating.

[00124] One application of a tunable laser in accord with the present invention

is for testing & monitoring (T&M) applications. A practical tunable laser for

T&M applications must have a maximum operating current selected to achieve a

reasonable laser lifetime (e.g., a current density of greater than 10 kA/cm² is

typically correlated with a degraded lifetime). In commercially available tunable

lasers using quantum well gain medium the quantum well lasers have useful

wavelength range of about 70 nanometers. However, in many applications it is

desirable to have a larger tuning range. Conventionally, three tunable lasers,

each having quantum well gain media optimized for different wavelength

ranges, would be required for T & M over a 200 nanometer tuning range. By

way of contrast, the quantum dot lasers of the present invention have a

continuous tuning range of at least 200 nanometers at practical current densities.

[00125] It will be also be understood that the quantum dot active regions of the

present invention may be used in laser structures having an integrated

wavelength selector element to tune the lasing wavelength. A variety of other semiconductor laser structure with wavelength selector elements are known in

the art. In particular, a variety of tunable distributed Bragg reflector (DBR) and

distributed feedback laser structures (DFB) are known in the lasing art. DFB and

DBR lasers include a grating that define a Bragg wavelength condition given by

where Λ is the grating period, n is the refractive index of the material,

and m is the diffraction order. The Bragg wavelength condition defines a

wavelength having a high effective reflectivity associated with the grating. As

shown in FIG. 26, multi-section DFB and DBR lasers are known in the art having

a plurality of sections 2605, 2610, 2615 in which a current may be adjusted in one

section of the laser to adjust the refractive index of a portion of the grating 2620

of the laser and hence its Bragg wavelength. A suitable tunable DFB or DBR

laser may be modified to include a quantum dot active region 2630 of the present

invention.

[00126] Additionally, the output wavelength of a semiconductor laser having a

quantum dot active region may be temperature tuned. Referring to FIG. 27A, a

distributed feedback laser having a quantum dot active region 2705 may include

any suitable grating structure to establish a Bragg lasing condition from the

periodicity of a grating fabricated on waveguide layers 2720, 2730 having

different refractive indices. In a conventional DFB laser, thermal expansion and

the temperature dependence of the refractive index causes a shift in the Bragg wavelength of about 0.1 nm/°C. In conventional 1.3 micron lasers the gain peak

shifts by about 0.4°C. The result is that there is a limited temperature range (e.g.,

typically about 40 °C) over which the DFB laser retains good modal properties,

i.e., the DFB laser tends to have too high a threshold current or has poor mode

discrimination if the temperature rises such that the gain peak is substantially

longer in wavelength than the Bragg wavelength. FIG. 27B shows the shift in

Bragg wavelength and gain peak versus temperature for conventional 1.3 micron

lasers. Quantum dot active regions have a delta function density of states

associated with the three dimensional quantum confinement of the quantum

dots that results in reduced temperature sensitivity. Experiments by the

inventors indicated that quantum dots have a measured shift in peak gain of

only about 0.17 nm/°C due to the delta function density of states associated with

the three dimensional quantum confinement. The reduced temperature

sensitivity of the gain peak along with the broad gain function permits a

temperature tuned DFB quantum dot laser to achieve an increased tuning range

compared to a conventional bulk or quantum well DFB laser. In one

embodiment of a DFB laser, the Bragg wavelength is selected to operate on the

long wavelength side of the optical gain spectrum at a first operating

temperature to facilitate operating the laser at higher operating temperatures

(which shifts the optical gain spectrum to longer wavelengths). [00127] The quantum dot active region of the present invention may also be

used in monolithic multi-wavelength arrays of lasers. FIG. 28 is a perspective

view of a multiwavelength array 2800 of lasers 2820. Each laser 2820 may be

fabricated as a ridge laser, buried heterostructure laser, or other laser structure

providing lateral optical confinement. Each laser has a longitudinal cavity

length, L. A conventional cleaving or etching process may be used to form a

laser facet 2815. The array 2800 is formed on a common substrate 2805 having a

quantum dot active region 2810. The growth parameters of the quantum dot

active region 2810 are selected to achieve a desired wavelength tuning range

(e.g., 100 to 200 nm) within a desired range of current densities. Each laser is

preferably a DFB or DBR laser having at least one grating section 2830 with the

grating periodicity, Λ, of its grating selected to achieve a desired wavelength of

the laser. One benefit of the monolithic multi-wavelength array 2800 is that the

large tuning range of the quantum dot active region 2810 permits DFB or DBR

lasers 2820 having a large number of different output wavelengths to be

simultaneously fabricated for dense wavelength division multiplexed (DWDM)

applications. Alternately, array 2800 may be used for wide wavelength division

multiplexed (WWDM) applications in which it is desirable to have a smaller

number of wavelengths but with a significant wavelength spacing. For

example, in one embodiment, array 2800 could be used to cover the wavelength range of 1270 nanometers to 1355 nanometers (e.g., a 75-85 nanometer range). In

one embodiment the inhomogeneous broadening and the difference in energy of

the ground state transition energy and the first excited state transition energy is

selected to permit a continuous optical gain spectrum of at least 75 nanometers

for a maximum threshold current density.

[00128] The optical characteristics of the quantum dot active region 2810 of

array 2800 also permits each laser to be directly modulated at high data rates.

Theoretical investigations by the inventors indicate that optimized quantum dot

lasers should have a linewidth enhancement factor that is approximately a factor

of five -to- ten lower than for conventional quantum well lasers along with a

higher differential gain. This makes it practical to directly modulate quantum

dot lasers at high data rates (e.g., 10 - 40 Gbit/s) with acceptable chirp. This is in

contrast to conventional quantum well lasers, which are typically modulated

with external modulator at high data rates to reduce chirping effects.

Additionally, the low threshold current density of the quantum dot lasers of the

present invention facilitate direct laser modulation. Typically, directly

modulated high-frequency lasers require drive currents that are several times the

threshold current. The comparatively low threshold current of each laser of

multiwavelength array 2800 facilitates direct modulation at high data rates.

Consequently, in one embodiment, each laser of the multiwavelength array is directly modulated by varying its drive current. This eliminates the needs for

external modulators, filters, and optical isolators used in conventional DWDM

systems.

[00129] As shown in FIG. 29, if each laser of the array is directly modulated

their outputs may be directly received by an optical combiner (MUX) 2950

module and coupled to an optical fiber. It will be understood that in a direct

current modulation embodiment that any suitable high frequency packaging

technique may be used to facilitate coupling microwave frequency drive currents

to each laser. In one embodiment, each laser is driven by a wire bond. However,

a wire bond has a parasitic inductance such that long lengths of bonding wire

may degrade performance at very high data rates. In another embodiment,

microwave transmission lines are fabricated on array 2900 to facilitate coupling

microwave drive current to each laser. In still another embodiment, array 2900 is

mounted on a submount adapted to provide microwave current to each laser.

[00130] One benefit of a multiwavelength array 2800 of the present invention is

that temperature tuning may be used to fine-tune the wavelength of a large

number of wavelengths. FIG. 30 illustrates a gain spectrum for two different

temperature with Bragg wavelengths of an array of lasers (having a "picket"

fence shape for an optical network grid) superimposed on the same plot.

Referring to FIG. 30, in one embodiment, the nominal DFB wavelengths at a first temperature, Ti, are selected to cover an upper wavelength range of the gain

spectrum. Tuning the temperature of array 2800 to a higher temperature T2 shifts

the gain spectrum at a slightly higher rate than the Bragg wavelengths. Due to

the large initial width of the gain curve and the reduced temperature shift of the

gain response of quantum dot lasers, a monolithic array of DFB lasers spanning a

large wavelength range can be simultaneously tuned over a wide range.

[00131] Another benefit of array 280 is that the low threshold current density

and slow drift of the gain response with temperature renders the lasers less

susceptible to heating and thermal cross-talk. This may permit, for example,

individual lasers to be more closely packed (e.g., a small inter-laser separation)

compared with conventional quantum well lasers. Additionally, the thermal

characteristics of array 2800 may facilitate operating the array junction up, i.e.,

with the substrate mounted to a heat sink. As is well known in the laser art, a

junction down configuration (epitaxial layer mounted to the heat sink) has less

thermal resistance but typically increased the packaging complexity and cost.

Still another benefit of array 2800 is that for some applications, such as WWDM,

the need for the heat sink to be cooled may be eliminated.

[00132] While embodiments of the present invention have been discussed in

detail in regards to quantum dot layers comprising InAs embedded in InGaAs

quantum wells, it will be understood that invention may be practices in related III-V compound semiconductor materials. For example, the InGaAs quantum

wells may be replaced with AlInGaAs wells. Similarly, the barrier layers may

comprise a variety of III-V materials, such as AlGaAs or AlGalnAsP. It will be

understood, for example, that the barrier layers may be comprised of a material

having a lattice constant selected so that the barrier layers between quantum dot

layers serve as strain compensation layers.

[00133] The present invention has been discussed in detail in regards to laser

structures grown on GaAs substrates. GaAs substrates have many advantages

over other III-V semiconductor substrates, such as a comparatively large wafer

size. However, while the present invention has been discussed in regards to

materials formed on a GaAs substrate, it will be understood that embodiments of

the present invention may be practiced on other types of substrates, such as InP

substrates. Additionally, while molecular beam epitaxy has been described as a

preferred fabrication technique, it will be understood that embodiments of the

present invention may be practiced using other epitaxial techniques as well.

[00134] While particular embodiments and applications of the present

invention have been illustrated and described, it is to be understood that the

invention is not limited to the precise construction and components disclosed

herein and that various modifications, changes and variations which will be

apparent to those skilled in the art may be made in the arrangement, operation and details of the method and apparatus of the present invention disclosed

herein without departing from the spirit and scope of the invention as defined in

the appended claims.

Claims

CLAIMSWhat is claimed is:

1. A semiconductor laser, comprising:

a semiconductor optical waveguide;

quantum dot means optically coupled to the optical waveguide having an

ensemble of quantum dots for providing a saturated ground state modal gain of

greater than about 25 cm"¹ at an emission wavelength of at least about 1260 nm

responsive to an injected current; and

quantum well means for providing carrier confinement of injected current

to the quantum dot means.

2. The laser of claim 1, wherein the laser has a cavity length not greater

than 500 microns.

3. A directly modulated laser, comprising:

a semiconductor optical waveguide;

quantum dot means optically coupled to the optical waveguide having an

ensemble of quantum dots for providing a saturated ground state modal gain of

greater than about 25 cm^-1 at an emission wavelength of at least about 1260 nm

responsive to an injected current, the quantum dot means having a linewidth

enhancement factor of less than about 0.2; means for injecting a modulated current into the quantum dot means to vary the

light output of the laser; and

quantum well means for providing carrier confinement of injected current

to the quantum dot means.

4. The laser of claim 3, wherein the laser has a cavity length not greater

than 500 microns.

5. The laser of claim 3, wherein the laser has a cavity length not greater

than 300 microns.

6. A semiconductor laser, comprising:

a semiconductor optical waveguide;

quantum dot means optically coupled to the optical waveguide having an

ensemble of quantum dots with a sequence of at least two inhomogeneously

broadened optical transition energies for providing a continuous optical gain

spectrum over a wavelength range of greater than 75 nanometers responsive to a

threshold drive current; and

quantum well means for providing carrier confinement of injected current

to the quantum dot means.

7. A semiconductor laser, comprising:

a semiconductor optical waveguide; quantum dot means optically coupled to the optical waveguide having an

ensemble of quantum dots with a sequence of at least three inhomogeneously

broadened optical transition energies for providing a continuous optical gain

spectrum over a wavelength range of greater than 150 nanometers responsive to

a threshold drive current; and

quantum well means for providing carrier confinement of injected current

to the quantum dot means.

8. A semiconductor laser, comprising:

a semiconductor optical waveguide including a plurality of spaced-apart

quantum wells within an active region disposed within the waveguide; and

a plurality of quantum dots embedded in the plurality of spaced-apart quantum

wells, the quantum dots shaped and positioned within each quantum well to

provide a saturated ground state modal gain to the waveguide of greater than

about 25 cm^-1 at an emission wavelength of at least about 1260 nm responsive to

an injected current.

9. The laser of claim 8, wherein the laser has a saturated ground state

modal gain of at least 40 cm⁴.

10. The laser of claim 8, wherein there are between four to eight quantum

wells layers and the active region has a thickness of not more than 300

nanometers.

11. The laser of claim 10, wherein the thickness of the active region is in

the range of between 200 nanometers to 300 nanometers.

12. The laser of claim 8, wherein a barrier layer thickness between

neighboring quantum wells is greater than 20 nanometers.

13. A semiconductor laser, comprising:

a semiconductor optical waveguide formed on a GaAs substrate including a

plurality of spaced-apart quantum wells within an active region disposed in the

waveguide; and

a plurality of quantum dots embedded in the plurality of spaced-apart

quantum wells, the quantum dots having a size distribution forming a sequence

of at least three inhomogeneously broadened optical transition energies for

providing a continuous optical gain spectrum over a wavelength range of greater

than 150 nanometers responsive to a threshold drive current.

14. A semiconductor laser, comprising:

a semiconductor optical waveguide including a plurality of spaced-apart

quantum wells within an active region disposed in the waveguide; and

a plurality of quantum dots embedded in the plurality of spaced-apart

quantum wells, the quantum dots having a size distribution forming a sequence

of at least two inhomogeneously broadened optical transition energies for providing a continuous optical gain spectrum over a wavelength range of at least

75 nanometers responsive to a threshold drive current.

15. A semiconductor active region for providing optical gain, comprising:

an InGaAs quantum well semiconductor quantum well having a

substantially planar well layer sandwiched between first and second barrier

layers; and

a plurality of InAs quantum dots embedded in the InGaAs quantum well;

the quantum dots sized to have a ground state emission energy at room

temperature of at least 1260 nanometers.

16. A semiconductor active region for providing optical gain, comprising:

an InGaAs quantum well semiconductor quantum well having a

substantially planar well layer sandwiched between first and second barrier

layers;

a plurality of InAs quantum dots embedded in the InGaAs quantum well, the

quantum dots having a ground state with an associated first optical transition

energy value and a first excited state having an associated second optical

transition energy value, the second optical transition energy value being in the

range of between 30 meV to 80 meV greater than the first optical transition

energy value; and p-type and n-type diode layers positioned to inject current into the quantum

dots;

the quantum dots having an inhomogenous broadening associated with size

variations of the quantum dots sufficient with respect to the energy separation of

the optical transition energies for the quantum dots to have a continuous optical

gain spectrum responsive to a threshold electrical current density.

17. A tunable laser, comprising:

a first optical cavity having a first end and a spaced-apart second end;

quantum dot active region means positioned in the first optical cavity

having a sequence of quantum confined energy states with energy levels selected

to provide continuous optical gain over a wavelength range of greater than 150

nanometers at a preselected current density;

a first reflector reflecting light into the first end of the first optical

cavity; and

an external optical cavity including an optical element reflecting a

selected wavelength of light into the second end of the first optical cavity.

18. A monolithic multiwavelength array of lasers, comprising:

a substrate;

quantum dot active region means disposed on the substrate for

providing an extended optical gain spectrum; quantum well means for confining carriers in the quantum dot active

region means; and

a plurality of lasers formed on the quantum dot active region means,

each of the plurality of lasers having a Bragg grating with an associated a grating

period positioned to provide optical feedback to the laser.

19. A method of growing self-assembled InAs quantum dots embedded in

InGaAs quantum wells, comprising:

growing a first AlxGal-xAs barrier layer with the molar fraction, x, less

than 1.0;

preadsorbing a floating layer of InAs to substantially match a steady-

state surface segregated layer of InAs for growth of bulk InGaAs;

growing a first InGaAs layer on the bottom AlGaAs barrier layer

having a thickness of less than about 2 nanometers;

depositing a sufficient equivalent monolayer coverage of InAs to form

InAs islands;

growing a second InGaAs layer over the InAs islands to embed the

InAs islands in InGaAs; and

growing a second AlxGal-xAs barrier layer with the molar fraction, x,

less than 1.0.

20. The method of claim 19, further comprising: selecting a growth

temperature of the InAs for a desired dot density.

21. The method of claim 19, further comprising: selecting the In alloy

composition of the first InGaAs layer to achieve a desired quantum dot density.

22. The method of claim 19, further comprising: stopping growth after the

second InGaAs layer is grown and desorbing exposed InAs for a sufficient length

of time to planarize tops of the InAs islands.

23. The method of claim 22, wherein at least one monolayer of GaAs is

deposited prior to desorbing exposed InAs.

24. A method for growing quantum dots with a controllable dot density,

the method comprising:

growing a first InGaAs layer having a reproducible surface InGaAs alloy

composition;

selecting a growth temperature for forming InAs islands with in a range of

between about 450 °C to 540 °C;

growing InAs for a selected equivalent monolayer coverage of between

about 1.8 to 4.0 at the growth temperature to form InAs islands; and

growing a second InGaAs layer to embed the islands, forming a plurality of

quantum dots; the plurality of quantum dots having a density determined by the alloy

composition of the first InGaAs layer and the growth temperature.

25. A method of growing self-assembled InAs quantum dots embedded in

InGaAs quantum wells, comprising:

growing a first AlxGal-xAs barrier layer with the molar fraction, x, less

than 1.0;

preadsorbing a floating layer of InAs to substantially match a steady-

state surface segregated layer of InAs for growth of bulk InGaAs;

growing a first InGaAs layer on the bottom AlGaAs barrier layer

having a thickness not greater than about 2 nanometers;

depositing a sufficient equivalent monolayer coverage of InAs to form

InAs islands;

growing a second InGaAs layer over the InAs islands, the second

InGaAs layer having a thickness selected to embed substantially all of the InAs

islands in InGaAs; and

growing a second AlxGal-xAs barrier layer with the molar fraction, x,

less than 1.0.

26. For a quantum dot laser having an ensemble of quantum dots for

providing optical gain at a wavelength of at least 1260 nanometers with a linewidth enhancement factor of less than about 0.2, a method of operating the

laser, the method comprising:

selecting a modulation bit rate in excess of about one Gbps; and

supplying a drive current to directly modulate the laser at the modulation

bit rate.

27. For a quantum dot laser having an ensemble of quantum dots with a

sequence of inhomogeneousy broadened optical transition energies extended

over a wavelength range of at least 150 nanometers, a method of operating the

laser, the method comprising:

providing a drive current for which the optical gain spectrum is continuous

over the the wavelength range; and

supplying wavelength-selective optical feedback to generate a lasing

wavelength within the wavelength range.