quantum dots for Þber laser sources - university of … · quantum dots for Þber laser sources...

Quantum dots for fiber laser sources

Jason M. Auxiera, Axel Schulzgena, Michael M. Morrella, Brian R. Westa, Seppo Honkanena,Sabyasachi Senb, Nicholas F. Borrellib, and Nasser Peyghambariana

aOptical Sciences Center, University of Arizona, Tucson, AZ 87521-0094 USAbGlass Research Group, Corning, Inc., Research and Development Center, Sullivan Park,

Corning, NY 14831 USA

ABSTRACT

In this invited paper, we will discuss the use of quantum dots as nonlinear optical elements in fiber lasersources. Furthermore, a review of the fabrication of the first low-loss (< 0.5 dB/cm) ion-exchanged waveguidesin a quantum-dot-doped glass will be presented. We will discuss the coupling, propagation, absorption, andscattering losses in these waveguides. The near-field mode profile along with the refractive index profile of thesewaveguides will be presented.

This PbS quantum-dot-doped glass was chosen due to its attractive optical gain and bleaching characteristicsat wavelengths throughout the near infrared. This bleaching of the ground-state optical transition has beenutilized for passive modelocking of a variety of lasers in the near infrared.

In addition, we will discuss some of the potential integrated and fiber optics applications of our quantum-dot-doped waveguides.

Keywords: Lasing materials, quantum dots, ion exchange, waveguides, fiber laser source, nonlinear optics, lasermodelocking, optical amplifier, integrated optics

1. INTRODUCTION

Semiconductor quantum dots (QDs) have been studied for the past two decades. These exotic structures exhibitoptical properties that can be tailored for a wide variety of applications, including QD modelockers, QD opticalswitches, and QD lasers. This paper will discuss the progression of these QD devices, especially the productionof QD-doped waveguides.

There are a large number of quantum dot systems that have been reported with a wide variety of fabricationmethods. Quantum dots have been formed using deposition in porous glasses,1 chemical preparation andsubsequent suspension in an organic or polymer matrix,2 sol-gel,3, 4 and grown through various epitaxialtechniques.5–7

One primary characteristic which gives one method an advantage over others is the resulting particle sizedistribution and/or density of defects. The other obvious distinction is cost. For many applications, one mightwant both of these to be as small as possible. For other applications, (e.g., tunable lasers) we might want thesize distribution to be moderate, while keeping the number of defects to a minimum.

1.1. Review of the optical properties of quantum dots

Semiconductor quantum dots (QDs) have unique optical properties due to their 3D quantum-confinement ofelectrons and holes. Principally, the 3D quantum-confinement produces a blue-shift in the optical resonanceswith QD size8; therefore, we can use the same constituents to make devices which operate at different opticalwavelengths.

Additionally, this shift (binding energy) is so large that these resonances are easily resolvable over a widetemperature range. Second, this confinement produces an enhancement of these optical resonances and produces

Further author information: (Send correspondence to J.M. Auxier)J.M. Auxier: E-mail: [email protected]

Invited Paper

Fiber Lasers II: Technology, Systems, and Applications, edited by L. N. Durvasula,Andrew J. W. Brown, Johan Nilsson, Proceedings of SPIE Vol. 5709(SPIE, Bellingham, WA, 2005) · 0277-786X/05/$15 · doi: 10.1117/12.594407

249

very large optical nonlinearities.9 QDs are easily invertible, especially in the strong-confinement regime, soQD-lasers can have ultra-low lasing thresholds. Third, the 3D quantum-confinement keeps electron-hole pairsso well contained that the optical properties of quantum dots are often temperature independent, i.e., quantumdot devices operate well at room temperature. This is in contrast to quantum well (QW) systems, wheretemperature-broadening washes out the optical resonances.

For an overview of the electronic structure of quantum dots, then see Peyghambarian et al.,10 Chapter 9 orfor a more detailed description, then see Haug and Koch.11

1.1.1. Particle in a box

Following Peyghambarian et al.,10 the simplest model of a quantum dot is the particle in a spherical box withan infinite potential barrier (ignore many-body effects). In this problem, we solve the Schrodinger equation forelectrons and holes separately: (

− h2

2mi∇2 + V (r)

)ζi(r) = εiζi(r), (1)

where i = e or h and V (r) = 0 for r < R and V (r) = ∞ for r ≥ R. This boundary condition (ideal quantumconfinement) requires that the wavefunction vanish outside the boundary. Also, note that we are ignoring theCoulomb interaction entirely, thus ignoring all many-body effects. The solution of the Schrodinger equation is

ζi(r) =

√1

4πR3

j�

(αn�

rR

)j�+1 (αn�)

Y m� (θ, φ), (2)

where j� is the �th order spherical Bessel function, Y m� (θ, φ) are the spherical harmonics, and αn� is the nth root

of the �th order spherical Bessel function.12

In Eq. 2, n, �, and m are the particle’s quantum numbers and the denominator comes from normalization.Now, inserting Eq. 2 into Eq. 1, we find the energy eigenvalues:

εi =h2

2mi

(αn�

R

)2

. (3)

These eigenstates are denoted using atomic spectroscopy notation as 1s, 1p, 1d, . . . , where s, p, d, . . . correspondto � = 0, 1, 2, . . . , respectively. Note that our spherical confinement potential allows bound states (e.g., 1p) thatare not allowed under the Coulomb potential.

Now, for optical (interband) transitions we must look at the energy separation between electron and holestates. As is customary, let the zero energy be at the top of the valence band, so using Eq. 3, the energy levelsare

εe = E3Dg +

h2

2me

(αne�e

R

)2

, (4)

and

εh = − h2

2mh

(αnh�h

R

)2

. (5)

For one electron-hole pair, the Schrodinger equation is only slightly altered:(− h2

2me∇e

2 + − h2

2mh∇h

2 + VC

)φ(r) = εφ(r), (6)

where φ(r = R) = 0, VC is the Coulomb potential, and φ(re, rh) = ζe(re)ζh(rh). Now, if we ignore the Coulombinteraction between electrons and holes, then

ε = εe + εh = E3Dg +

h2

2me

(αne�e

R

)2

+h2

2mh

(αnh�h

R

)2

≡ E3Dg +

h2

2m∗(αn�

R

)2

. (7)

250 Proc. of SPIE Vol. 5709

Figure 1. Room-temperature absorption spectra of PbS QD-doped glasses with mean QD radii R.

This equations shows the quantum-size effect of quantum dots, that the absorption is blue-shifted from thebulk bandgap E3D

g , with a shift proportional to 1/R2. Therefore, the smaller the quantum dot, the higher thetransition energy or the shorter the transition wavelength. This effect gives us the ability to tune the quantumdot resonances by choosing the QD size (as shown in Fig. 1). Since these transitions are discrete, the density ofstates for quantum dots is a sum of δ-functions.

This analysis was for a single QD with a single electron-hole pair. In a true QD sample, there is an ensembleof quantum dots with a variation in their radii. This produced a variation in the blue-shifted resonances thatgoes as 1/R2, and thus a variation of the optical resonances. This variation produces inhomogeneous broadeningof the optical resonances, making the apparent linewidth broad (See Fig. 1). Theoretically, this broadening ishandled using a density matrix approach13 to sum over all of the dots, each having its own radius and naturallinewidth.

For a realistic description of the optical transitions and electron-hole wavefunctions of quantum dots, onemust consider the detailed geometry and bandstructure of the QD-system under consideration. Additionally,many-body effects should be taken into account using the Semiconductor-Bloch equations.11

1.2. General optical properties of PbS quantum dot doped glass

Our research group has extensively studied the linear and nonlinear optical properties of lead-salt quantum dotsin silicate glass produced by a thermal treatment method.14 This quantum dot system has a small particle sizedistribution of around ∆R/R ≈ 5%. This produces an absorption spectrum with well-resolved optical resonancesas seen in Fig. 1.

The PbS QDs studied here have radii (2-5 nm) smaller than the bulk exciton Bohr radius (18 nm), whichplaces them within the strong 3D-confinement limit (due to the large barrier potential). The small bulk bandgapenergy (0.4 eV @ 300 K) allows us to tune the ground excited state transition throughout the near infrared,which includes telecommunications wavelengths of 1300 and 1550 nm. Figure 1 shows the room temperatureabsorption spectra of PbS QD-doped glasses with different dot radii. Notice the well-defined optical resonances,which is characteristic of QDs. Such optical resonances are not easily seen at room temperature for quantumwells or bulk semiconductors.

The major advantage of semiconductor-doped glasses over epitaxially grown structures is cost. Thesesemiconductor-doped glasses are far less expensive than epitaxially grown heterostructures. Additionally, they

Proc. of SPIE Vol. 5709 251

Figure 2. Absorption spectra of PbS QD-doped glasses with mean QD radii R (left) along with their measured interbandtransition energies (right). The solid and dotted curves in the figure on the right were calculated using the hyperbolic-bandmodel.

are robust and can be manufactured with a wide range of QD concentrations. Recent improvements in the man-ufacture of quantum dots (QDs) embedded in glassy matrices have resulted in structures with more uniform-sizedistribution; fewer vacancies, substitutional defects, and dangling bonds; higher dot concentration; and reducedphotodarkening.14

The QD radii R quoted in Fig. 1 are calculated using a hyperbolic band (HB) model15:

(hω1s)2 =(

hc

λ1s

)2

= E2g +

2h2Eg

m∗( π

R

)2

, (8)

where we used the room temperature (T = 300 K) bandgap energy of Eg = 0.41 eV and effective mass of m∗ =0.12m0 for PbS.16 This slight modification to Eq. 7 includes the non-parabolic nature of the PbS bandstructure.The predictions given by the hyperbolic-band model is shown Fig. 2. This model was further corrected by Kangand Wise.16

2. QUANTUM DOT OPTICAL GAIN AND LASING

Quantum dots can be easily inverted since very few (can be as low as one) electron-hole pairs need to be createdfor inversion. This property gives quantum dots the possibility to experience stimulated emission at extremelylow excitation energies. Quantum dot lasers can have a very low lasing thresholds (or even thresholdless) andquantum dot detectors can be used for single-photon counting.

So far, quantum dot lasers have been made through epitaxial growth techniques7; however, optical gain17

and amplified spontaneous emission18, 19 have been demonstrated in other quantum dot systems. Here, we willprovide a brief overview of these epitaxially-grown QD-lasers and discuss optical gain results in a couple of otherquantum dot systems.

2.1. Quantum dot lasersEpitaxially grown QD-lasers have been developed for a number of lasing wavelengths. Their development wasboot-strapped by the mature field of growing quantum well (QW) heterostructures using epitaxial growth (layer-by-layer growth). The most common epitaxial growth methods are molecular beam epitaxy (MBE) and metal-organic vapor phase epitaxy (MOVPE). Many of these QD-lasers are current-injected devices and some can bedirectly modulated well into the GHz regime.


QD material λ (nm) Research Group

ZnCdSe/ZnSe 560 Hommel - SPIE 4594 294 (2001)22

InP/GaInP 700 Blood - APL 85 1904 (2004)23

GaInAs/(Al)GaAs 980 Forchel - APL 84 2238 (2004)24

InGaAs/GaAs 1000 Bhattacharya - APL 80 3482 (2002)25

AlGaAs/InGaAs 1310 Deppe - JQE 35 1238 (1999)26

Photonic crystal InGaAs/GaAs 1340 Scherer - Elect Lett 38 967 (2002)27

PbSe/PbEuTe 4300 Bauer - APL 79 1225 (2001)28

Table 1. Some examples of quantum dot lasers. We provide the material system, lasing wavelength, along with thereference.

The primary difference between quantum well and quantum dot growth is the use of the Stranski-Krastanowgrowth mode of strained-layer heteroepitaxy20 to produce self-assembled (or self-organized) quantum dots. Inthis process, we grow a few monolayers of layer-by-layer growth on top of a substrate with a slightly differentlattice constant. This lattice mismatch creates strain in the monolayer, which is termed a wetting layer (sincethe monolayer acts somewhat like a liquid). This strained monolayer undergoes surface-energy minimizationresulting in an array of islands of semiconductor (for example, islands of InGaAs on top of a GaAs substrate).For InGaAs/GaAs quantum dots, the near-pyramidal islands have a typical lateral size of 10-40 nm and a heightof 5-8 nm.21 These near-pyramidal islands are then capped by growing an overlayer (another material, often thesame as the substrate) on top. The capped islands are then self-assembled QDs. This process is often repeatedto produce multiple layers of quantum dots.

Self-assembled growth depends on the surface tension created in the wetting layer. This includes the elasticrelaxation on the facet edges, renormalization of the surface energy of the facets, and interaction betweenneighboring islands via the substrate. This substrate and overlayer interaction is strong enough to repeat thegrowth pattern of the quantum dots for multiple layers (i.e., the next layer of QDs are formed exactly on top ofthe first layer), which makes lasing easier. For an extensive review of this fabrication technique used to producethese self-assembled QD lasers, please see the book Quantum Dot Heterostructures by Bimberg et al.7

Some examples of these QD-lasers22–28 are shown in Table 1 along with their references. Without going intogreat detail of the properties of these lasers, we would like to point out a couple of these QD-lasers with uniquefeatures not found in their QW-counterparts.

From this table, we immediately recognize that the mid-infrared (MIR) laser at 4300 nm is unique,28 sincecurrently, there exists no QW-lasers at this wavelength (at room temperature). This is a direct consequenceof the temperature stability of quantum dot resonances. Since the bandgap is required to be narrow (0.29 eV)in this device, the high-temperature stability must be great (thermal energy is one-tenth the bandgap). Thislaser was grown using molecular-beam epitaxy (MBE) into a high-finesse, vertical-cavity surface-emitting laser(VCSEL) structure. This structure is also unique since it was produced using IV-VI semiconductor materials.The trick here was to use Pb1−xEuxTe in order to produce a good barrier material. Good quantum confinementis achieved using only a few percent of Eu content in the alloy.28

The photonic crystal laser27 (optically pumped) shown in the table is of particular interest since it is a singlequantum dot device. This structure is a realization of the thresholdless lasing that quantum dots are capable of.Additionally, the photonic crystal was engineered to match the narrow homogeneous broadening of the quantumdot to produce a high-Q microcavity. Additionally, the photonic crystal can be re-engineered to produce manydifferent cavity modes and/or coupled cavities. In this case, the two- and four-defect, coupled cavity lasersproduced lasing. These devices demonstrate the promise of photonic crystal devices and QD-lasers.


Figure 3. Measurements of the room-temperature, optical gain dynamics in PbS QD-doped glasses. Left shows thepump and probe pulse spectra, center shows the gain dynamics for the pump wavelengths of 1352 nm (light) and 1317nm (dark), and right shows the splitting of the ground and first excited state transitions. In the center figure, the dottedline at −∆αd = 0.08 corresponds to the linear absorption of the ground state around 1350 nm. Thus, the reduction ofabsorption of 0.11 shown at 1352 nm corresponds to a gain of gd = 0.03. Gain is only seen when the correct state splittingis used (right).

2.2. Optical gain in QD-doped glass

Previously, we reported room-temperature optical gain at the ground exciton transition of PbS quantum-dot-doped glasses.17 Here, we will outline this result and correlate it with recent results of optical gain and amplifiedspontaneous emission in lead-salt quantum dots.

For this demonstration of optical gain, we used orthogonally polarized 130 fs pulses which are independentlytunable in frequency. Pump and probe pulses are obtained from two optical parametric amplifiers (OPAs), whichare synchronously pumped by one regenerative Ti:sapphire amplifier at a repetition rate of 1 kHz. The OPAproducing the pump pulses was tuned to 980 nm, which corresponded to the first excited-state transition of thesequantum dots (see the sample with R = 2.7 nm in Fig. 1). We then independently tuned the OPA producingthe probe pulses across the ground-state transition (near 1350 nm). To reduce noise, we used a dual-beam setupfor the probe beam with reference and signal beam paths. These two probe beams were detected simultaneouslyusing an auto-balanced detector (cancels signals common to both beam paths). This allowed us to detect verysmall (spectrally integrated) transmission changes.

Figure 3 shows the pump and probe pulse spectra (left), the gain dynamics for two probe wavelengths(center), and the splitting of the ground and first excited state transitions (right). The pump pulse was tuned tothe common pump wavelength of 980 nm within the first-excited exciton state of the quantum dots. Taking thesplitting between the first-excited state and the ground state shown in Fig. 3(right), we measured a material gainin the quantum dots of 80 cm−1 at a probe wavelength of 1352 nm.17 Additionally, we measured a relaxationtime of 5 ps for the ground-state transition.

This result of optical gain in lead-salt quantum dots was later confirmed in colloidal QDs.18, 19 Thesedemonstrations of optical gain were in PbS19 (260 cm−1) and PbSe18 (100 cm−1) quantum dots in colloidalsolution. These colloids not only demonstrate optical gain, but also produce amplified spontaneous emission(ASE). One reason for the increased gain and ASE is due to a large increase in the volume fraction of thequantum dots. For the PbS QD-doped glass, we had a volume fraction of ∼ 0.1% as compared to a volumefraction of ∼ 20% for the PbSe solution and a volume fraction of ∼ 30% for the PbS solution. The other reasonfor this increase (vital for ASE) is that these colloidal solutions display a much larger Stokes shifts than seenin the PbS QD-doped glass. This allows the photoluminescence spectrum to have little overlap with the linearabsorption spectrum, thus requiring much less absorption change to see the same material (and modal) gain.


3. PASSIVE Q-SWITCHING AND MODELOCKING USING QUANTUM DOTS

3.1. Review of modelocking

In the past, solid-state saturable absorbers have attracted much interest because of their capability to enable pas-sive, self-starting modelocking in solid-state laser systems and ensure simpler and less expensive solutions to laserdesign than additive pulse or active modelocking techniques. This passive modelocking relies on the nonlinear-optical effect of absorption saturation, or bleaching. For this reason, these devices that exhibit bleaching aresometimes called saturable absorbers.

There are two classic classes of saturable absorbers: fast29 and slow.30 In both cases, the saturable absorberis placed within the laser cavity and serves to modulate the net gain of the laser. This modulation acts as a gatewhich can easily be opened (low loss) by a short optical pulse, but is difficult to open (high loss) by a CW beam.The onset (self-starting) of the modelocking occurs from noise or fluctuations in the laser modes in the cavity.These fluctuation spikes swings the gate open, which builds after many round-trips in the cavity. Technically,this is described by the formation of a short net-gain window, which generates and stabilizes the ultrashort pulse.

Mathematically, the modelocking process is analyzed by the Haus master equation29:

TR∂

∂TA(T, t) =

(−iD

∂2

∂t2+ iδ|A(T, t)|2

)A(T, t) +

(g(T ) − l + Dg

∂2

∂t2− s(T, t)

)A(T, t), (9)

with the round-trip time TR, the complex pulse envelope A(T, t), the group-velocity dispersion (GVD) D, thegain dispersion Dg, the self-phase modulation coefficient δ, the linear round-trip losses l, and the saturableabsorber response s(T, t).

In the case of the fast saturable absorber, the absorber saturation occurs much faster than the saturationof the gain medium and the resulting pulse-width (i.e., the absorber’s response follows the pulse-shape).31, 32

Mathematically, this saturable absorber response (transmission) is described by

s(t) =s0

1 + I(t)/Isat, (10)

where s0 is the unsaturated transmission (< 1) of the absorber, I(t) is the instantaneous pulse intensity, andIsat is the saturation intensity of the absorber.

During the modelocking process, the peak of the pulse sees a large net gain due to a major reduction in theloss in the cavity. This produces an amplification of the peak of the pulse and thus reducing the pulse width.This process continues until the pulse becomes short enough to use the available gain bandwidth (ignoring group-velocity dispersion). Fast saturable absorbers can be analyzed numerically by inserting this absorber responseinto Eq. 9. For an ideal fast saturable absorber, the Haus master equation is linearized obtaining the familiaranalytical solution of a simple hyperbolic secant.29, 31

For a slow saturable absorber, there is a temporal differential that is setup between the loss and the gain.33

This assumes that the pulse is short compared to the relaxation time of the gain medium and the saturableabsorber. Mathematically, this is described by

s(t) = −s0

∫ t

0

dt|A(t)|2

Ws, (11)

where A(t) is the mode amplitude (field) and Ws is the saturation energy of the absorber.31 Under thisapproximation, the gain medium follows a similar equation:

g(t) = −g0

∫ t

0

dt|A(t)|2

Wg, (12)

where go is the small-signal gain and Wg is the saturation energy of the gain. As long as the absorber is morestrongly saturated than the gain, we see a net gain during a short period of time.31 During this net-gainwindow, the pulse is amplified ultimately resulting in a pulse-width which closely matches the width of this


Figure 4. (left) Diagram of the Cr:forsterite laser cavity along with the (right) intensity autocorrelation of resultingpulses using PbS QD-doped glass as a saturable absorber.35 This cavity is a Z-fold, astigmatism compensated cavitywith Brewster-cut Cr:forsterite crystal as a lasing medium. This cavity is shown with prism-pair dispersion compensation.

net-gain window. Again, this absorber response and the Haus master equation can be linearized to producean analytical solution. The result is again a hyperbolic secant pulse-shape, but with new constraints on thepulse-width and amplitude.30, 31

Additionally, one can produce soliton-like modelocking using a slow saturable absorber.34 Here, the group-velocity dispersion is balanced by self-phase modulation within the gain medium. The first part of Eq. 9 isthe nonlinear Schrodinger equation, which has a well-known soliton solution (again, hyperbolic secant in pulse-shape).34 Here, the slow saturable absorber starts the modelocking process by producing a long, differential-gainwindow. Soliton-like modelocking takes over once pulsing starts and the effects of the optical nonlinearity becomessignificant.

For more detailed information about the modelocking processes of saturable absorbers, please look at thein-depth mathematical descriptions give by Haus29–31 and Keller.32, 33

3.2. Quantum dot Q-switchers and modelockers

Quantum dots have very large nonlinearities,9 which includes absorption saturation. Quantum dot saturableabsorbers are inexpensive alternatives to epitaxially-grown saturable absorbers. Our research group did anextensive study of the modelocking process of a Cr:forsterite laser using a PbS QD-doped glass.35, 36

Using a Z-folded, Cr:forsterite laser, we produced 5 ps pulses (see Fig. 4) using this PbS QD-doped saturableabsorber and measured a saturation intensity of Isat = 0.18 MW/cm2. An autocorrelation of the pulses producedusing this laser and saturable absorber is shown in Figure 4 (right).

In order to further understand the modelocking process, we measured the bleaching dynamics of the saturableabsorber. These measurements were preformed using a degenerated pump-probe technique using the auto-balanced photo-receiver discussed earlier. Here, we pumped the PbS QD-doped glass with 120 fs pulses froman amplified Ti:sapphire laser system. The pump fluence was varied from 0.1 up to 40 mJ/cm2, covering theestimated intercavity irradiance seen by the saturable absorber. The pulse wavelength was set near to theoperating wavelength of the Cr:forsterite laser, which was on the low-energy side of the lowest electron-hole pairtransition of the quantum dots (see the sample with R = 2.7 nm in Fig. 1). The results showed a saturationenergy of EA = 62 nJ and a two-component decay with a fast component of 5 ps at high fluence (comparableto the fluence seen inside the Cr:forsterite laser). The temporal dynamics of the Cr:forsterite laser using theQD-doped glass as a saturable absorber was limited to this time scale.

This striking resemblance between the bleaching decay time and the resulting pulse-width from the laserlead us to investigate the modelocking process further. Using these measured characteristics of the saturableabsorber, we performed a numerical simulation of the modelocking process. The working equation for this


QD material Laser and Pulse width Research Group

Filter glass (no QDs) Ti:Sapphire - 3 ps Nakano37

PbS QD in a silicate glass Cr:forsterite - 5 ps Guerreiro35 and Wundke36

PbSe QD in a phosphate glass Ho:YAG - 85 ns Lipovskii38

PbS QD in a silicate glass Nd:KGW - 15 ps Malyarevich39

PbSe/PbSe Core/Shell QD Er:Yb - 15 ps Lifshitz40

InGaAs/GaAs QD:SESAM Yb:KYW - 4 ps Ustinov41

InGaAs/GaAs QD self-org. Same (1.3 µm) - 7 ps, 20 GHz Bimberg42

Table 2. Some examples of quantum dot Q-switchers and modelockers. We provide the material system, laser and pulsewidth, along with the research group.

Figure 5. (a) Ion-exchange process and (b) a phase-contrast micrograph of two PbS QD-doped channel waveguides(separated by 250µm) produced by K+-Na+ ion exchange (380 oC, 234 hr).

numerical simulation was Haus’ master equation discussed earlier (Eq. 9). These simulations explained thepulse-width limitations of this modelocking process.36 Evidently, this saturable absorber does not act purelyas a slow or fast saturable absorber. Additionally, this saturable absorber could not demonstrate soliton-likemodelocking. However, using a PbS QD-doped glass with excited-state resonance at the lasing wavelength, oursimulations showed the capability of soliton-like modelocking.36

Since our demonstration of modelocking using quantum dots as a saturable absorber, there have been anumber of demonstrations of passive Q-switching and modelocking using quantum dot saturable absorbers.Table 2 summarizes these demonstrations.

4. QUANTUM-DOT-DOPED WAVEGUIDES

Waveguides in semiconductor-doped glasses have been produced for integrated photonic devices.43, 44 Here, wereview the demonstration of the first low-loss waveguides in a QD-doped glass with a narrow size distribution.45, 46

Our PbS QD-doped waveguides have guiding losses less than 0.5 dB/cm at 1550 nm using a K+-Na+ ion-exchangeprocess.

Thermal diffusion of ions (within a concentration gradient) is the physical mechanism for the ion-exchangeprocess. Here, a potassium nitrate molten salt supplies potassium replacement ions for sodium ions in the glass.This ion exchange produces an index change by altering the local glass density and mean polarizability.47, 48

The Lorentz-Lorenz formula describes the polarizability.

Ag+-Na+ ion exchange was not used here due to the reduction of silver in the glass. This silver nanoparticleformation was confirmed by optical absorption spectroscopy and scanning electron microscopy. K+-Na+ ionexchange requires more time due to the lower self-diffusion coefficient of K+.

In preparation for the ion-exchange process, the glass sample was surface polished, cleaned, and coated withtitanium. The titanium film serves as the ion-exchange mask after the lithographic processing. The titanium


Figure 6. (a) Near-field mode profile and its (b) horizontal and (c) vertical cross-sections from a PbS QD-doped surfacewaveguide(1550 nm).

was coated with photoresist, which was patterned, developed, and cured. The developed photoresist served as amask for titanium etching. After etching and photoresist removal, the sample was ready for ion exchange (seeFig. 5(a)). The K+-Na+ ion exchange was performed by placing the sample in a bath of pure KNO3 moltensalt within a temperature-controlled furnace. After ion exchange, the titanium was removed and the sample wascut and polished for device characterization (see Fig. 5(b)). Figures 6 and 7 show measurements of the modeand index profiles of one of these channel waveguides, respectively.

The resulting index and mode profiles are asymmetric since they are surface waveguides. The index profileis asymmetric since potassium can diffuse along the surface of the glass, resulting in a distributed and ellipticalindex profile; however, the region of highest index change is well confined, which keeps the waveguide single-mode. The mode profile is asymmetric due to the strong surface interaction of the guided mode, that is, thehigh index differential between the glass and the air above the surface draws the mode up against the surface.

To evaluate the quality of the fabricated channel waveguides we analyzed the losses using the fiber-waveguide-objective method. We compared the loss when the waveguide was fiber-coupled at both facets to the loss whenthe waveguide was fiber-coupled at the input only and the light was collected by a microscope objective at theoutput. We measured a guide loss of < 0.5 dB/cm for several channel waveguides.45

Low-loss is not the only desirable characteristic of these waveguides. The principal concern lies upon theQD-doping of these waveguides. In order to ensure that these waveguides were doped with PbS quantum dots, weperformed luminescence measurements of this QD-doped glass before and after ion exchange. These luminescencemeasurements are shown in Fig. 8.

There are two concerns about preforming this ion-exchange process. First, the ion exchange occurs withina furnace at a moderately high temperature of 380 0C. The quantum dots within the glass were formed usinga thermal treatment around 500 0C. We would like to make sure that this ion-exchange temperature does notcause the quantum dots to grow. Second, the ion-exchange process changes the local chemistry of the glass.We would like to make sure that the quantum dots within the waveguides have not been destroyed during theion-exchange process.


Figure 7. Index profile of a channel (PbS QD-doped) waveguide measured by the RNF technique at 1550 nm. WithK+ exchange, one expects a ridge at the surface; however, the dip here is due to the harsh etching process we used tocompletely remove the titanium mask. The width of the index profile increased with increasing mask opening.

Figure 8. Luminescence from PbS QD-doped waveguides (solid) and in this glass before (dotted) and after (dashed) ionexchange along with the QD absorption (open circles). The waveguide spectrum was taken in transmission by collectedthe coupled light from inside the waveguide while pumping through the waveguide (λpump = 1064 nm).


As Fig. 8 shows, there are no noticeable differences between the three luminescence spectra. A comparison ofthe dotted and dashed spectra in Fig.8 represents a comparison of the quantum dots throughout the volume ofthe glass before and after ion exchange. Since this bulk luminescence from the quantum dots remained unchangedthroughout the ion-exchange process, we concluded that the high temperature did not cause the quantum dotsto grow or change shape. Comparing the solid spectra in Fig.8 with the other two, we see no difference betweenthe quantum dots within the waveguide and those throughout the volume of the glass. This demonstrates thatthe chemical composition was unaltered during the ion-exchange process.

5. CONCLUSIONS

In this paper, we discussed the many laser application of quantum dots. We reviewed lasers, modelockers, Q-switches, ASE sources, and other nonlinear optical devices that were made using quantum dots. As an example ofoptical devices incorporating quantum dots, we described the development of QD-doped waveguides, which servesas the first step towards producing integrated nonlinear optical devices using quantum dots. These waveguideswere shown to have very low loss, to be single-mode, and be semi-homogeneously doped with quantum dots.These QD-doped waveguides were shown to have the same optical properties as the original QD-doped glass.This development may lead to the development of waveguide-based signal amplifiers, lasers, and ASE sourcesthat have quantum dots as a gain medium.

ACKNOWLEDGMENTS

The authors would like to thank Corning, Inc. for collaboration and supplying the PbS QD-doped glass samples.We acknowledge Kim Winick’s research group at the University of Michigan for preforming RNF measurements.This research benefited from the generous financial support from TRIF (State of Arizona Photonics Initiative),the Center for Optoelectronic Devices, Interconnects, and Packaging (COEDIP), and the National Science Foun-dation (NSF).

REFERENCES1. N. Borrelli and J. Luong, “Semiconductor microcrystals in porous glass,” Proc. SPIE 866, pp. 104–109,

1988.2. A. Nozik, F. Williams, M. Nenadovic, T. Rajh, and O. Micic, “Size quantization in small semiconductor

particles,” J. Phys. Chem. 89, pp. 397–399, 1985.3. M. Nogami, K. Nagasaka, and K. Kotani, “Microcrystalline PbS doped silica glasses prepared by the sol-gel

process,” J. Non-Cryst. Solids 126(1), pp. 87–92, 1990.4. M. Nogami, K. Nagasaka, and M. Takata, “CdS microcrystal-doped silica glass prepared by the sol-gel

process,” J. Non-Cryst. Solids 122(1), pp. 101–106, 1990.5. D. L. Huffaker and D. G. Deppe, “Electroluminescence efficiency of 1.3 µm wavelength ingaas/gaas quantum

dots,” App. Phys. Lett. 73, pp. 520–522, 1998.6. A. Raab and G. Springholz, “Controlling the size and density of self-assembled PbSe quantum dots by

adjusting the substrate temperature and layer thickness,” App. Phys. Lett. 81(13), pp. 2457–2459, 2002.7. D. Bimberg, M. Grundmann, and N. N. Ledentsov, Quantum Dot Heterostructures, Wiley, Chichester, 1998.8. A. L. Efros and A. L. Efros, “Interband absorption of light in a semiconductor sphere,” Sov. Phys. Semicond.

16(7), pp. 772–775, 1982.9. E. Hanamura, “Very large optical nonlinearity of semiconductor microcrystallites,” Phys. Rev. B 37(3),

pp. 1273–1279, 1988.10. N. Peyghambarian, S. W. Koch, and A. Mysyrowicz, Introduction to Semiconductor Optics, Prentice Hall,

Englewood Cliffs, NJ, 1993.11. H. Haug and S. W. Koch, Quantum Theory of the Optical and Electrical Properties of Semiconductors,

World Scientific, 3rd ed., Singapore, 1990.12. G. Arfken, Mathematical Methods for Physicists, Academic Press, 3rd ed., Boston, MA, 1985.13. C. Cohen-Tannoudji, B. Diu, and F. Laloe, Quantum Mechanics - Volume I, John Wiley and Sons, New

York, NY, 1977.


14. N. F. Borrelli and D. W. Smith, “Quantum confinement of PbS microcrystals in glass,” J. Non-Cryst. Solids180, p. 25, 1994.

15. Y. Wang, A. Suma, W. Mahler, and R. Kasowski, “PbS in polymers. from molecules to bulk solids,” J.Chem. Phys. 87(12), pp. 7315–7322, 1987.

16. I. Kang and F. W. Wise, “Electronic structure and optical properties of PbS and PbSe quantum dots,” J.Opt. Soc. Am. B 14(7), pp. 1632–1646, 1997.

17. K. Wundke, J. M. Auxier, A. Schulzgen, N. Peyghambarian, and N. F. Borrelli, “Room-temperature gainat 1.3 µm in PbS-doped glasses,” App. Phys. Lett. 75(20), pp. 3060–3062, 2001.

18. R. D. Schaller, M. A. Petruska, and V. I. Klimov, “Tunable near-infrared optical gain and amplified spon-taneous emission using pbse nanocrystals,” J. Phys. Chem. B 107, pp. 13765–13768, 2003.

19. V. Sukhovatkin, S. Musikhin, I. Gorelikov, S. Cauchi, L. Bakueva, E. Kumacheva, and E. Sargent, “Room-temperature amplified spontaneous emission at 1300 nm in solution-processed PbS quantum-dot films,” Opt.Lett. 30(2), pp. 171–173, 2005.

20. D. Leonhard, M. Krishnamurty, C. M. Reaves, S. P. Denbaar, and P. Petroff, “Direct formation of quantum-sized dots from uniform coherent islands of InGaAs on GaAs surfaces,” App. Phys. Lett. 63(23), pp. 3203–3205, 1993.

21. P. Bhattacharya, S. Ghosh, and A. Stiff-Roberts, “Quantum dot opto-electronic devices,” Annu. Rev. Mater.Res. 34(1), pp. 1–40, 2004.

22. M. Klude, T. Passow, G. Alexe, H. Heinke, and D. Hommel, “New laser sources for plastic optical fibers:ZnSe-based quantum well and quantum dot laser diodes with 560 nm emission,” Proc. SPIE 4594, pp. 260–270, 2001.

23. G. M. Lewis, J. Lutti, P. M. Smowton, P. Blood, A. B. Krysa, and S. L. Liew, “Optical properties ofInP/GaInP quantum-dot laser structures,” App. Phys. Lett. 85(11), pp. 1904–1906, 2004.

24. S.-C. Auzanneau, M. Calligaro, M. Krakowski, F. Klopf, S. Deubert, J. P. Reithmaier, and A. Forchel,“High brightness GaInAs/(Al)GaAs quantum-dot tapered lasers at 980 nm with high wavelength stability,”App. Phys. Lett. 84(13), pp. 2238–2240, 2004.

25. P. Bhattacharya and S. Ghosh, “Tunnel injection In0.4Ga0.6As/GaAs quantum dot lasers with 15 GHzmodulation bandwidth at room temperature,” App. Phys. Lett. 80(19), pp. 3482–3484, 2002.

26. D. G. Deppe, D. L. Huffaker, S. Csutak, Z. Zou, G. Park, and O. B. Shchekin, “Spontaneous emissionand threshold characteristics of 1.3-µm InGaAsGaAs quantum-dot GaAs-based lasers,” IEEE J. QuantumElectron. 35(8), pp. 1238–1246, 1999.

27. T. Yoshie, O. Shchekin, H. Chen, D. Deppe, and A. Scherer, “Quantum-dot GaAs-based lasers,” Electron.Lett. 38(17), pp. 967–968, 2002.

28. G. Springholz, T. Schwarzl, W. Heiss, G. Bauer, M. Aigle, H. Pascher, and I. Vavra, “Midinfrared surface-emitting PbSe/PbEuTe quantum-dot lasers,” App. Phys. Lett. 79(9), pp. 1225–1227, 2001.

29. H. A. Haus, “Theory of mode locking with a fast saturable absorber,” J. App. Phys. 46(7), pp. 3049–3058,1975.

30. H. A. Haus, “Theory of mode locking with a slow saturable absorber,” IEEE J. Quantum Electron. 11(9),pp. 736–746, 1975.

31. H. A. Haus, “Mode-locking of lasers,” IEEE J. Spec. Top. Quantum Electron. 6(6), pp. 1173–1185, 2000.32. F. X. Kartner, J. A. der Au, and U. Keller, “Mode-locking with slow and fast saturable absorbers - what’s

the difference?,” IEEE J. Spec. Top. Quantum Electron. 4(4), pp. 159–168, 1998.33. U. Keller, K. J. Weingarten, F. X. Kartner, D. Kopf, B. Braun, I. D. Jung, D. Flunk, C. Honninger,

N. Matuschek, and J. A. der Au, “Semiconductor saturable absorber mirrors (SESAM’s) for femtosecond tonanosecond pulse generation in solid-state lasers,” IEEE J. Spec. Top. Quantum Electron. 2(3), pp. 435–453,1996.

34. F. X. Kartner and U. Keller, “Stabalization of solitonlike pulses with slow saturable absorbers,” Opt. Lett.20(1), pp. 16–18, 1995.

35. P. T. Guerreiro, S. Ten, N. F. Borrelli, J. Butty, G. E. Jabbour, and N. Peyghambarian, “PbS quantum-dotdoped glasses as saturable absorbers for mode locking of a Cr:forsterite laser,” App. Phys. Lett. 71(12),pp. 1595–1597, 1997.


36. K. Wundke, S. Potting, J. M. Auxier, A. Schulzgen, N. Peyghambarian, and N. F. Borrelli, “PbS quantum-dot-doped glasses for ultrashort-pulse generation,” App. Phys. Lett. 76(1), pp. 10–12, 2000.

37. N. Sarukura, Y. Ishida, T. Yanagawa, and H. Nakano, “All solid-state cw passively mode-locked Ti:sapphirelaser using a colored filter glass,” App. Phys. Lett. 57(3), pp. 229–230, 1990.

38. A. M. Malyarevich, V. G. Savitsky, P. V. Prokoshin, K. V. Yumashev, and A. A. Lipovskii, “Passive Q-switchoperation of PbSe-doped glass at 2.1 µm,” Proc. SPIE 4350, pp. 32–35, 2001.

39. A. M. Malyarevich, V. G. Savitsky, P. V. Prokoshin, N. N. Posnov, and K. V. Yumashev, “Glass dopedwith PbS quantum dots as a saturable absorber for 1-µm neodymium lasers,” J. Opt. Soc. Am. B 19(1),pp. 28–32, 2002.

40. M. Sirota, E. Galun, V. Krupkin, A. Glushko, A. Kigel, M. Brumer, A. Sashchiuk, L. Amirav, and E. Lifshitz,“IV-VI semiconductor nanocrystals for passive Q-switch in IR,” Proc. SPIE 5510, pp. 9–16, 2004.

41. E. U. Rafailov, S. J. White, A. A. Lagatsky, A. Miller, W. Sibbett, D. A. Livshits, A. E. Zhukov, and V. M.Ustinov, “Fast quantum-dot saturable absorber for passive mode-locking of solid-state lasers,” PhotonicsTech. Lett. 16(11), pp. 2439–2441, 2004.

42. E. U. Rafailov, S. J. White, A. A. Lagatsky, A. Miller, W. Sibbett, D. A. Livshits, A. E. Zhukov, and V. M.Ustinov, “35 GHz mode-locking of 1.3 µm quantum dot lasers,” App. Phys. Lett. 85(5), pp. 843–845, 2004.

43. T. J. Cullen, C. N. Ironside, C. T. Seaton, and G. I. Stegeman, “Semiconductor-doped glass ion-exchangedwaveguides,” App. Phys. Lett. 49(21), pp. 1403–1405, 1986.

44. P. T. Guerreiro, S. G. Lee, A. S. Rodrigues, Y. Z. Hu, E. M. Wright, S. I. Najafi, J. Mackenzie, andN. Peyghambarian, “Femtosecond pulse propagation near a two-photon transition in a semiconductorquantum-dot waveguide,” Opt. Lett. 21(9), pp. 659–661, 1996.

45. J. M. Auxier, M. M. Morrell, A. Schulzgen, B. R. West, S. Honkanen, S. Sen, N. F. Borrelli, and N. Peygham-barian, “Ion-exchanged waveguides in glass doped with PbS quantum dots,” App. Phys. Lett. 85(25),pp. 6098–6100, 2004.

46. J. M. Auxier, B. R. West, S. Honkanen, A. Schulzgen, N. Peyghambarian, S. Sen, and N. F. Borrelli,“Quantum-dot-doped waveguides produced by ion exchange,” CLEO 2004 Technical Digest TOPS 43,2004.

47. N. F. Borrelli, Microoptics technology: fabrication and applications of lens arrays and devices, Mercel Dekker,New York, NY, 1999.

48. e. S. I. Najafi, Introduction to glass integrated optics, Artech House, Boston, MA, 1992.


quantum dots for Þber laser sources - university of … · quantum dots for Þber laser sources...

Documents