Sub-250nm light emission and optical gain in AlGaN materialsEmanuele Francesco Pecora, Wei Zhang, A. Yu. Nikiforov, Jian Yin, Roberto Paiella et al. Citation: J. Appl. Phys. 113, 013106 (2013); doi: 10.1063/1.4772615 View online: http://dx.doi.org/10.1063/1.4772615 View Table of Contents: http://jap.aip.org/resource/1/JAPIAU/v113/i1 Published by the American Institute of Physics. Related ArticlesEffects of pumping on propagation velocities of confined exciton polaritons in GaAs/AlxGa1−xAs doubleheterostructure thin films under resonant and non-resonant probe conditions J. Appl. Phys. 113, 013514 (2013) Infrared to vacuum-ultraviolet ellipsometry and optical Hall-effect study of free-charge carrier parameters in Mg-doped InN J. Appl. Phys. 113, 013502 (2013) Nanomechanical and optical properties of highly a-axis oriented AlN films Appl. Phys. Lett. 101, 254102 (2012) 2.8μm emission from type-I quantum wells grown on InAsxP1−x/InP metamorphic graded buffers Appl. Phys. Lett. 101, 251107 (2012) GaN-based platforms with Au-Ag alloyed metal layer for surface enhanced Raman scattering J. Appl. Phys. 112, 114327 (2012) Additional information on J. Appl. Phys.Journal Homepage: http://jap.aip.org/ Journal Information: http://jap.aip.org/about/about_the_journal Top downloads: http://jap.aip.org/features/most_downloaded Information for Authors: http://jap.aip.org/authors

Downloaded 04 Jan 2013 to 168.122.67.21. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

Sub-250 nm light emission and optical gain in AlGaN materials

Emanuele Francesco Pecora, Wei Zhang, A. Yu. Nikiforov, Jian Yin, Roberto Paiella,Luca Dal Negro,a) and Theodore D. MoustakasDepartment of Electrical and Computer Engineering and Photonics Center, Boston University,8 Saint Mary’s Street, Boston, Massachusetts 02215, USA

(Received 29 August 2012; accepted 29 November 2012; published online 4 January 2013)

We investigate the deep-UV optical emission and gain properties of AlxGa1�xN/AlyGa1�yN multiple

quantum wells structures. These structures were grown by plasma-assisted molecular-beam epitaxy

on 6H-SiC substrates, under a growth mode which promotes various degrees of band-structure

potential fluctuations in the form of cluster-like features within the wells. The degree of

inhomogeneities in these samples was determined by cathodoluminescence mapping. We measured

the TE-polarized amplified spontaneous emission in the sample with cluster-like features and

quantified the optical absorption/gain coefficients and gain spectra by the variable stripe length

technique under ultrafast optical pumping. A maximum net modal gain of about 120 cm�1

is measured at 4.9 eV. On the other hand, we found that samples with homogeneous quantum

wells lead to absorption. Numerical simulations are performed to support our experimental findings.VC 2013 American Institute of Physics. [http://dx.doi.org/10.1063/1.4772615]

I. INTRODUCTION

Group III-nitrides (GaN, InN, AlN, and their alloys)

constitute a large class of materials with very appealing opti-

cal properties for a number of engineering device applica-

tions.1–5 In particular, they exhibit a largely tunable and

direct energy band-gaps (from 0.7 eV for InN, to 3.4 eV for

GaN, to 6.2 eV for AlN at room temperature), allowing the

fabrication of highly efficient light emitting devices (LEDs).

Green and blue LEDs based on nitrides have become a

mature technological platform. Currently, the AlGaN mate-

rial is heavily investigated for its potential applications to

deep-UV LEDs and lasers. However, solid-state laser devi-

ces operating below 250 nm are still missing and would

enable a number of applications such as non-line-of-sight

free-space optical communications, biochemical agent

detection, disinfection, and medical diagnostics. Recently,

AlGaN-based light emitting devices have been fabricated

with an internal quantum efficiency of 50% at 250 nm,6,7 a

power efficiency of 40% at 240 nm,8 and, more recently,

external quantum efficiency above 10%.9 Moreover, some

prototypes of deep-UV laser as well as the demonstration of

stimulated emission have been reported in the literature.10–13

In 2004, Takano et al. reported the first evidence of lasing at

241.5 nm under pulsed optical pumping with a threshold

pumping power approximately 1200 kW/cm2 at room tem-

perature.10 In 2011, Wunderer et al. reported an optically

pumped laser fabricated on AlN substrates and working at

267 nm with a threshold power density as low as 126 kW/

cm2.11 The charge separation issues in polar III-nitride quan-

tum wells (QWs) results in reduced optical gain, and recent

approaches by using large overlap QWs or semi/non-polar

QWs had been pursued for suppressing the charge separation

in the active regions.14–17 The employment of multiple

quantum well structures is one of the most effective

approaches for obtaining lasing. However, homogeneous

quantum wells require high carrier density to invert their

population before any stimulated emission process sets in. In

2012, the authors proposed a novel approach of introducing

strong band-structure potential fluctuations, and demon-

strated optical gain under femto-second pulsed excitation

with a maximum net modal gain of 118 cm�1 at 254 nm with

a strongly reduced transparency threshold of 5 lJ/cm2.12

Moreover, we reported on strongly TE-polarized amplified

spontaneous emission (ASE),18 which has been ascribed to

the strain in the active layer.19 Devices are usually optically

or electrically7 pumped, but after an appropriate design of

the sample structure and using high quality materials, Oto

et al. demonstrated8 100 mW deep-UV emission from multi-

ple quantum wells pumped by an electron beam.

In this paper, we report on the optical amplification

properties of high Al content AlGaN quantum wells embed-

ded between AlN cladding layers grown by plasma-assisted

molecular-beam epitaxy (MBE) under a growth mode which

promotes various degrees of band structure compositional

inhomogeneities. Specifically, we compare the stimulated

emission behavior of samples with cluster-like features with

samples whose wells are homogeneous, excited under identi-

cal conditions. The edge emission has been studied as a func-

tion of the pump power, and the optical gain has been

quantified through the variable stripe length (VSL) method.

II. EXPERIMENTAL

AlGaN samples were grown on the Si-face of (0001)

6H-SiC substrates by RF plasma-assisted MBE.20 Growth of

AlGaN alloys on SiC substrates provides a number of chal-

lenges, in particular, the accidental nitridation of the SiC

substrate prior to epitaxial growth and formation of stacking

mismatch boundaries at the step edges due to the polytype

difference between 2H-AlGaN and 6H- or 4H-SiC. The SiC

a)Author to whom correspondence should be addressed. Electronic mail:

dalnegro@bu.edu.

0021-8979/2013/113(1)/013106/7/$30.00 VC 2013 American Institute of Physics113, 013106-1

JOURNAL OF APPLIED PHYSICS 113, 013106 (2013)

Downloaded 04 Jan 2013 to 168.122.67.21. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

substrates were degreased in organic solvents followed by

dipping them into a heated H2SO4:H2O2 (3:1) mix and then

into a buffered HF to remove surface contaminants and

oxides. Prior to growth, the substrates were exposed at high

temperature to a gallium flux followed by desorption

to remove residual oxygen through formation of volatile

gallium oxides. Following this step, the SiC surfaces exhibit

a clear �3� �3 R30� reconstruction in the reflection high-

energy electron diffraction (RHEED) pattern.20 First, a

500 nm thick AlN film was grown at substrate temperature of

800 �C, measured with a thermocouple positioned behind

the substrate. Then the substrate temperature was lowered to

790 �C and 10 AlGaN MQWs consisting of 1.5 nm wells and

40 nm barriers were deposited. Two different structures were

grown and investigated. A schematic of the sample structure

is shown in Fig. 1. Structure A (top) contained Al0.6Ga0.4N

wells and Al0.9Ga0.1N barriers while structure B (bottom)

contained Al0.7Ga0.3N wells and AlN barriers. The AlGaN

quantum wells in both samples were grown under excess gal-

lium, leading to a complete coverage of the surface of the

growing film with liquid gallium. As described elsewhere,

this growth mode is consistent with liquid phase epitaxy

rather than physical vapor phase epitaxy and leads to lateral

compositional inhomogeneities in the films due to statistical

variations of the thickness of the liquid gallium in the sur-

face.21 During growth of structure A, indium has also been

employed as a surfactant in order to improve the homogene-

ity of the film.6,22,23 Both structures were capped with a

100 nm thick AlN layer for waveguiding. The authors have

previously determined from temperature-dependent photolu-

minescence measurements20 that structure B leads to high

internal quantum efficiency (IQE) (68%).

Deep-UV, 150 fs laser pulses at 220 nm were used to

optically pump the multi quantum well structures. We used

the fourth harmonic generated by a proper crystal (Spectra

Physics GWU-24FL) pumped by a mode-locked ultra-fast

high-power Ti:sapphire laser (Spectra Physics MaiTai,

80 MHz) operating at 880 nm as the excitation source. The

high pulse frequency allows the system to be excited in a

quasi-steady-state condition and increases the pumping effi-

ciency (carriers in the active layer are continuously excited

before the conduction band is totally relaxed after every sin-

gle pulse). The pump laser is focused on the sample surface

through a cylindrical lens forming a stripe whose length can

be monitored and adjusted through a blade mounted in a

motorized computer-controlled stage. The beam profile

along the stripe has been measured through the knife-edge

technique, resulting in a height of the stripe of 5 lm and a

maximum stripe length of 250 lm, which provides a homo-

geneous illumination of the sample. Beyond this area, the

pumping intensity cannot be considered uniform anymore

and VSL gain data should not be trusted.12,24 The maximum

fluence on the sample is 60 lJ/cm2. Samples have been char-

acterized in the VSL set-up, exciting the top surface and col-

lecting the ASE from the cleaved edge of the sample through

an UV-transmitting objective, an UV-transmitting movable

analyzer, a computer-controlled f/4 monochromator (Corner-

stone 260) with UV-efficient gratings, and a lock-in amplifier

(Oriel Merlin) coupled to a UV-optimized photomultiplier

tube (PMT, Oriel Instruments 77348). The output power has

been measured considering the PMT responsivity, the trans-

mittivity of all the optical components, and the light coupling

efficiency into the monochromator. The values have been

also verified with a high sensitive power meter detector

(Newport 918D-UV-OD3). A schematic of the measurement

set-up is shown in Fig. 6.

The VSL technique with optical pumping was used in

this work. Alternatively, other approaches based on VSL tech-

nique with current injection,25,26 Hakki-Paoli method,27 and

segmented contact technique28,29 had also been used for char-

acterizing optical gain in electrically injected quantum well

lasers. The VSL technique is a very reliable and widely used

method for the measurement of the optical gain coefficient in

bulk materials, and it allows measuring unambiguously the

FIG. 1. CL spectra and corresponding monochro-

matic CL maps. A schematic of the samples is

also shown for structure A (top) and B (bottom).

013106-2 Pecora et al. J. Appl. Phys. 113, 013106 (2013)

Downloaded 04 Jan 2013 to 168.122.67.21. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

net modal gain coefficient and the entire gain spectrum.30–33

No special sample preparation is needed; however, some ex-

perimental conditions need to be carefully verified to avoid

any artifacts.24 Data are analyzed within the 1D amplifier

model. By integrating the expression for the amplification of

light from a point source, a distance L from the edge, over the

excitation length, the net modal gain can be related to the

ASE intensity by

IASE ¼JspðXÞ

GðeGL � 1Þ; (1)

being Jsp(X) the spontaneous emission intensity emitted

within the solid angle X and G the net modal gain of the

material. The usual VSL scan consists in collecting the IASE

signal as a function of the excitation length. From the fit of

the experimental data, the value of G can be deduced for a

specific wavelength. In addition, the net modal gain spec-

trum can be measured by comparing the ASE intensity spec-

tra collected at two different excitation lengths L1¼L and

L2¼ 2L, with L2¼ 2L1, using the following equation:

G ¼ 1

Lln

I2L

IL� 1

� �� �: (2)

The luminescence properties of these structures were

also investigated using cathodoluminescence (CL) spectros-

copy and mapping in a JEOL JSM-6100 SEM equipped

with a Gatan Mono CL2 system. The UV-optimized

1200 grooves/mm grating was blazed at 250 nm. The time

integrated photomultiplier tube current was used as the mea-

surement of the CL intensity. Top surface CL spectra from

areas of approximately 620 lm by 400 lm and spatially

resolved monochromatic CL maps were collected using a

specially designed adjustable parabolic CL mirror.

III. RESULTS AND DISCUSSION

A. Cathodoluminescence maps

The investigated structures exhibit intense CL at room

temperature. Typical spontaneous surface emission CL spec-

tra obtained with low beam current irradiation are shown for

the two investigated structures in Fig. 1. For structure A, the

CL emission spectrum shown in Fig. 1(a) is symmetric with

the peak at 251 nm. The corresponding 251 nm monochro-

matic CL map shown in Fig. 1(b) indicates a mostly homo-

geneous emission.

The CL emission spectrum from structure B shown in

Fig. 1(c) is also symmetric, but the peak is at 242 nm. The

corresponding 242 nm monochromatic CL map in Fig. 1(d)

reveals bright and dark patches at the micron and submicron

scales suggesting fine-scale compositional fluctuations in the

QWs. The fact that the emission spectrum of structure B is

blue-shifted with respect to that of structure A can be quali-

tatively accounted for by the difference in the composition

of the wells and barriers of these two structures.

B. Amplified spontaneous emission

We have characterized the same samples under optical

pumping for detailed gain quantification. First, we have

investigated the edge emission as a function of the pump

fluence. The measured spectrum intensity is divided by the

excitation fluence in order better emphasize the nonlinear

behavior of the emission (i.e., super/sub-linear emission),

and plotted as a function of the wavelength in the Fig. 2.

Figure 2(a) refers to the structure A. We report three repre-

sentative different pump fluences. The peak wavelength is

252 nm, and a second peak is clearly detected on the left side

of the spectrum, overlapping with the main peak. The origin

of the second peak at 246 nm is not known at this time. One

possibility is that this emission is related to band structure

potential fluctuations due to thickness variation of the wells

since in MBE the accuracy of controlling the well thickness

is one monolayer. The emission of the homogeneous sample

is sub-linear with the fluence, demonstrating absorption in

this structure. On the other hand, Fig. 2(b) refers to the struc-

ture B. The measured trend in this sample is opposite to the

previous one. As the pump fluence increases (in the range

between 2 and 60 lJ/cm2), the signal intensity strongly

increases with a clear superlinear trend. More information is

obtained by the analysis of those spectra. In fact, we notice a

shift of the peak position toward the high energy side of the

spectrum with increasing the pump fluence.

Moreover, we fitted all the spectra as single Gaussian

functions to determine the peak position and the full width

half maximum (FWHM) of the emission. Results are shown

in Fig. 3. Black dots refer to the left side axis, which reports

the energy corresponding to the peak position. Red squares

FIG. 2. Edge emission as a function of the wavelength for the sample with

homogeneous wells (a), and with strong band-structure compositional fluctu-

ations (b). Data are scaled by the excitation fluence.

013106-3 Pecora et al. J. Appl. Phys. 113, 013106 (2013)

Downloaded 04 Jan 2013 to 168.122.67.21. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

are relative to the right axis, which represents the FWHM of

the spectra. Data are plotted as a function of the pumping flu-

ence for the sample with strong compositional fluctuations.

As the pump fluence increases, the edge emission narrows

and blue-shifts. In particular, the FWHM decreases from

12.5 to 8.5 nm while the peak position shifts up from 5.00 to

5.09 eV.

The superlinear emission along with the blue-shift and

the spectral narrowing of the sample with cluster-like fea-

tures are strongly supporting the onset of stimulated emis-

sion in this sample. On the other hand, we have conducted a

similar analysis on the structure A, finding no changes in the

peak position and the spectral width. This is consistent with

the observed sub-linear trend of the measured edge emission

with the pump fluence. The FWHM at different fluences are

plotted in Fig. 4 as a function of the inverse of the peak out-

put power. The continuous line represents the best linear fit

of the experimental data, corresponding to an intercept of

4.5 6 0.2 nm and a slope of (8.9 6 0.4)� 10�4 nm/W�1.

Such a linear trend has been previously reported for a

number of gain materials,34,35 and indicates stimulated emis-

sion in agreement with the Schawlow–Townes relation.36,37

We have also investigated the polarization properties of

the edge emission of both samples in order to better under-

stand the origin of the observed luminescence and gain. In

the polar plot of Fig. 5, we report the peak intensity meas-

ured for both samples as a function of the analyzer angle.

Red square dots correspond to the structure A, while blue

circles are relative to the structure B. 0� corresponds to the

TM polarization; 90� to the TE one. Intensities have been

recorded at the highest pump fluence (60 lJ/cm2) and they

are reported in a linear scale. First, we observe that the sam-

ple with compositional fluctuations is about a factor of 5

brighter than the other. More importantly, the emission from

the homogeneous wells is totally unpolarized, while the com-

positional fluctuations introduce a different band order by

which the ASE results strongly TE polarized. The Al content

in this sample is sufficiently high (70%), and a turnover from

the TE to the TM polarization is expected for a content of

60%–80%, depending on the thickness of the well and on the

strain in the active layer.19,38–40 Since the wells in our struc-

ture are very thin, it is reasonable that the TE polarization is

still predominant. Recent works also have indicated the use

of very thin GaN layer embedded within high Al-content

AlGaN (or AlN) QWs or superlattices also result in domi-

nant TE-polarized light emission41,42

C. Quantification of optical gain

We use the well-known VSL technique to measure the

gain coefficient from the evolution of the peak-emission in-

tensity as a function of the optically pumped sample length.

We report in Fig. 6 the data measured at the peak wavelength

together with their best-fit with Eq. (1). Data are normalized

to be comparable and we plot and process data over a maxi-

mum length of 250 nm. We have ascertained that the laser

beam profile cannot be considered homogeneous anymore

for longer values. As a consequence, for an excitation length

larger than 250 nm, the 1D amplifier model will not be valid.

FIG. 3. Peak position (black dots, left side axis) and FHWM (red circles,

right side axis) of the edge emission spectra from the sample with composi-

tional fluctuations as a function of the pump fluence.

FIG. 4. FWHM measured at different fluences as a function of the inverse of

the ASE intensity. Continuous line is a linear fit of the data.

FIG. 5. Measured peak intensity for the sample with (black dots) and with-

out (red circles) compositional fluctuations as a function of the analyzer

angle.

013106-4 Pecora et al. J. Appl. Phys. 113, 013106 (2013)

Downloaded 04 Jan 2013 to 168.122.67.21. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

Green triangles, red dots, and black squares refer to the struc-

ture A. VSL traces have been recorded at different pump flu-

ences, and even at the highest value, the concavity of traces

indicates a sub-linear trend. We always measure absorption

values in this sample, with an absolute value decreasing with

increasing the pump fluence. On the other hand, we have

conducted the same experiment in the structure B under the

same excitation conditions. We report the recorded trace for

the highest fluence, representative of all the measurements

above threshold. We recognize a change in the concavity of

the VSL trace, which implies positive gain values. Moreover,

after the fitting procedure, we extrapolate a gain in this struc-

ture of 110 cm�1. The absorption/gain spectrum can be

measured for the entire range of wavelengths. We have

detected the edge emission for two stripe lengths (75 and

150 lm) at the highest pump fluence, and we report the gain

spectra for the two samples in Fig. 7. Red squares correspond

to the structure A. The spectrum appears to be featureless

and always negative in values, as expected for an absorbing

1D waveguide, and it represents a measure of the net modal

absorption coefficient of the material. The introduction of

compositional fluctuations in the samples dramatically modi-

fies the gain spectrum. Blue circles are mostly in the positive

side of the graph, indicating that the sample is driven well in

the amplification regime. On the longer wavelength side of

the spectrum, the measured data turn into negative values.

The wavelength of zero gain is at 266 nm (4.66 eV), which

represents an experimental estimation of the effective band

gap of the material. We notice that this effective gap value

should not be confused with the estimated gap of a homoge-

neous material for the nominal Al content. The influence of

Al content on the gap of inhomogeneous materials is still

under investigation.

D. Numerical simulations

The gain properties of the quantum wells under study

were numerically investigated by first computing the

quantum-well sub-band structure with a commercial simula-

tion tool (NextNano3) based on a 6-band k � p model. The

calculated valence sub-bands are plotted in the inset of

Fig. 8, where they are labeled based on the character of their

k¼ 0 Bloch functions. It should be noted that the top-most

sub-band has heavy-hole (HH) character, despite the rela-

tively high Al content of the well material which is expected

to produce a cross-over of the crystal-field split-off hole

(CH) band edge above the HH maximum. The sub-band

FIG. 7. Absorption/gain spectra measured at the highest pump fluence for

the sample with (black dots) and without (red circles) compositional

fluctuations.

FIG. 6. (Top) schematic picture of the VSL measurement set-up. (Bottom)

VSL traces collected at the peak wavelength at different pump fluences.

Data are normalized and continuous lines are the best-fit with Eq. (1). Green

triangles, red dots, and black squares refer to the sample with homogeneous

wells. Blue circles were collected from the sample with compositional

fluctuations.

FIG. 8. Calculated gain spectra of a multiple-quantum-well structure

consisting of 1.5-nm-thick Al0.7Ga0.3N wells and 40-nm-thick Al0.9Ga0.1N

barriers, for different values of the injected carrier density (ranging from

3� 1019 cm�3 to 6� 1019 cm�3 in steps of 0.5� 1019 cm�3, in order of

increasing peak gain). Inset: calculated valence-subband structure of the

same sample.

013106-5 Pecora et al. J. Appl. Phys. 113, 013106 (2013)

Downloaded 04 Jan 2013 to 168.122.67.21. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

ordering observed in Fig. 8 is therefore strongly influenced

by quantum confinement effects (i.e., the HH sub-bands are

displaced from their respective band edge by a smaller

amount compared to the CH sub-bands, due to their larger

effective mass). An important consequence of this valence-

sub-band ordering is that the optical matrix elements

between the top-most valence sub-band and the conduction

sub-bands are predominantly TE polarized.

The calculated sub-band dispersion curves were then

used to compute the material gain spectrum for different

values of the carrier density in the wells, using the model

described in Ref. 43. The results are shown in Fig. 8,

where the different traces correspond to carrier densities

ranging from 3� 1019 cm�3 to 6� 1019 cm�3 in steps of

0.5� 1019 cm�3. Based on these results, we obtain a trans-

parency carrier density of 3.9� 1019 cm�3, which is sub-

stantially larger than what can be produced with the

maximum pump fluence available in our experimental setup.

This prediction is therefore consistent with the absence of

measurable gain in the structure A. In contrast, in the struc-

ture B, only a fraction of the active layer volume (i.e., the

regions with lowest Al content and therefore lowest bandgap

energy) needs to be inverted to produce gain, and the trans-

parency pump-fluence threshold is reduced proportionally.

Furthermore, in-plane quantum confinement effects may

also contribute to further reduce this threshold. Incidentally,

it should be noted that the calculated gain spectra in

Fig. 8 are peaked at photon energies in the range of 5.37 to

5.40 eV, and the corresponding wavelengths of about

230 nm are substantially shorter that the experimental peak

emission wavelength of Fig. 2. This discrepancy may be

attributed to variations in the layer thicknesses and composi-

tions of the experimental samples relative to their target val-

ues. In any case, additional gain calculations based on the

same model show that, even as these parameters are varied

over a wide range, a similarly large transparency carrier

density on the order of a few 1019 cm�3 is obtained. The

main conclusions of this discussion are therefore unaffected

by the detailed structure of the fabricated quantum wells.

IV. CONCLUSIONS

We have investigated the sub-250 nm optical emission

and gain properties of AlGaN/AlN multiple quantum wells

structures under ultrafast optical pumping. Samples have

been grown with different growth approaches, leading to ho-

mogeneous quantum wells or to band-structure composi-

tional fluctuations in the form of cluster-like features within

the wells. We used the variable stripe length technique to

determine the optical absorption/gain coefficients. We meas-

ured only optical losses in the sample with uniform quantum

wells. These results are supported within the 6-band k � p

model of the AlGaN band structure. On the other hand, we

report blue-shift and narrowing of the emission, VSL traces,

gain spectra, polarization studies, and the validity of the

Schalow–Townes relation to demonstrate a maximum net

modal gain of 120 cm�1 at 250 nm in the sample with strong

compositional fluctuations. The type of AlGaN alloys and

QWs investigated in this paper were employed in the growth

of UV LEDs with IQE in excess of 50%,7,44 therefore, they

represent a viable pathway to the fabrication of portable,

solid state, deep-UV laser devices.

ACKNOWLEDGMENTS

The work was supported by the Defense Advanced

Research Projects Agency CMUVT Program (PM: Dr. John

Albrecht) under subcontract from Photon Systems, Inc. (U.S.

Army Cooperative Agreement No. W911NF-11-1-0034).

1S. Nakamura and S. F. Chichibu, Introduction to Nitride SemiconductorBlue Lasers and Light Emitting Diodes (Taylor & Francis, New York,

2000).2F. A. Ponce and D. P. Bour, Nature 386, 351 (1997).3S. Pearton, GaN and Related Materials (Overseas Publisher Association,

Amsterdam, 2000).4J. I. Pankove, and T. D. Moustakas, Gallium Nitride (GaN) I and II in

Semiconductors and Semimetals, vol. 50 and vol. 57 (Academic Press,

1998, 1999).5Y. Taniyasu, M. Kasu, and T. Makimoto, Nature 441, 325 (2006).6A. Bhattacharyya, T. D. Moustakas, L. Zhou, D. J. Smith, and W. Hug,

Appl. Phys. Lett. 94, 181907 (2009).7T. D. Moustakas, Y. Liao, C. Kao, C. Thomidis, A. Bhattacharyya, D.

Bhattarai, A. Moldaver, Proc. SPIE 8278, 8278L-1 (2012).8T. Oto, R. G. Banal, K. Kataoka, M. Funato, and Y. Kawakami, Nature

Photon. 4, 767 (2010).9M. Shatalov, W. Sun, A. Lunev, X. Hu, A. Dobrinsky, Y. Bilenko, J.

Yank, M. Shur, R. Gaska, C. Moe, G. Garrett, and M. Wraback, Appl.

Phys. Express 5, 082101 (2012).10T. Takano, Y. Narita, A. Horiuchi, and H. Kawanishi, Appl. Phys. Lett.

84, 3567 (2004).11T. Wunderer, C. L. Chua, Z. Yang, J. E. Northrup, N. M. Johnson, G. A.

Garrett, H. Shen, and M. Wraback, Appl. Phys. Express 4, 092101

(2011).12E. F. Pecora, W. Zhang, A. Yu. Nikiforov, L. Zhou, D. J. Smith, J. Yin,

R. Paiella, L. Dal Negro, and T. D. Moustakas, Appl. Phys. Lett. 100,

061111 (2012).13J. Mickevicius, J. Jurkevicius, K. Kazlauskas, A. Zakauskas, G. Tamulai-

tis, M. S. Shur, M. Shatalov, J. Yang, and R. Gaska, Appl. Phys. Lett. 100,

081902 (2012).14R. M. Farrell et al., Semicond. Sci. Technol. 27, 024001 (2012).15P. S. Hsu et al., Appl. Phys. Lett. 100, 021104 (2012).16H. Zhao et al., Opt. Express 19, A991–A1007 (2011).17J. Zhang et al., J. Appl. Phys. 110, 113110 (2011).18E. F. Pecora, W. Zhang, J. Yin, R. Paiella, L. Dal Negro, and T. D. Mous-

takas, Appl. Phys. Express 5, 032103 (2012).19J. E. Northrup, C. L. Chua, Z. Yang, T. Wunderer, M. Kneissl, N. M. John-

son, and T. Kolbe, Appl. Phys. Lett. 100, 021101 (2012).20W. Zhang, A. Yu. Nikiforov, C. Thomidis, J. Woodward, H. Sun, C.-K.

Kao, D. Bhattarai, A. Moldawer, L. Zhou, D. J. Smith, and T. D. Mousta-

kas, J. Vac. Sci. Technol. B 30, 02B119 (2012).21T. D. Moustakas and A. Bhattacharyya, Phys. Status Solidi C 9, 580

(2012).22Y. Liao, Ph.D. dissertation, Boston University, 2011.23E. Monroy et al., J. Appl. Phys. 93, 1550 (2003).24L. Dal Negro, P. Bettotti, M. Cazzanelli, D. Pacifici, and L. Pavesi, Opt.

Commun. 229, 337 (2004).25A. Oster et al., IEEE J. Sel. Top. Quantum Electron. 5, 631–636 (1999).26M. Korbl et al., J. Appl. Phys. 92, 2942 (2002).27A. Thranhardt et al., Appl. Phys. Lett. 86, 201117 (2005).28G. Tsvid et al., IEEE J. Quantum Electron. 44, 732–739 (2008).29J. W. Ferguson et al., IEEE J. Quantum Electron. 47, 870–877 (2011).30K. L. Shaklee, R. E. Nahory, and R. F. Leheny, J. Lumin. 7, 284

(1973).31L. Dal Negro, M. Cazzanelli, L. Pavesi, S. Ossicini, D. Pacifici, G. Franz�o,

F. Priolo, and F. Iacona, Appl. Phys. Lett. 82, 4636 (2003).32G. Frankowsky, F. Steuber, V. Harle, F. Scholz, and A. Hangleiter, Appl.

Phys. Lett. 68, 3746 (1996).33M. R€owe, M. Vehse, P. Michler, J. Gutowski, S. Heppel, and A. Hangleiter,

Phys. Status Solidi C 0, 1860 (2003).

013106-6 Pecora et al. J. Appl. Phys. 113, 013106 (2013)

Downloaded 04 Jan 2013 to 168.122.67.21. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

34M. W. Fleming and A. Mooradian, Appl. Phys. Lett. 38, 511 (1981).35S. G. Cloutier, P. A. Kossyrev, and J. Xu, Nature Mater. 4, 887 (2005).36A. L. Schawlow and C. H. Townes, Phys. Rev. 112, 1940 (1958).37C. H. Henry, IEEE J. Quantum Electron. 18, 259 (1982).38J. Zhang, H. Zhao, and N. Tansu, Appl. Phys. Lett. 97, 111105 (2010).39A. A. Yamaguchi, Appl. Phys. Lett. 96, 151911 (2010).

40S. Park, J. Korean Phys. Soc. 59, 357 (2011).41J. Zhang et al., Appl. Phys. Lett. 98, 171111 (2011).42Y. Taniyasu et al., Appl. Phys. Lett. 99, 251112 (2011).43S. L. Chuang, IEEE J. Quantum Electron. 32, 1791 (1996).44Y. Liao, C. Thomidis, C.-K. Kao, and T. D. Moustakas, Appl.

Phys. Lett. 98, 081110 (2011).

013106-7 Pecora et al. J. Appl. Phys. 113, 013106 (2013)

Downloaded 04 Jan 2013 to 168.122.67.21. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions