Temperature evolution of carrier dynamics in GaNPAs alloys

Temperature evolution of carrier dynamics in GaNxPyAs1−y−xalloysM. Baranowski, R. Kudrawiec, A. V. Luce, M. Latkowska, K. M. Yu, Y. J. Kuang, J. Misiewicz, C. W. Tu, and W.Walukiewicz Citation: Journal of Applied Physics 117, 175702 (2015); doi: 10.1063/1.4919751 View online: http://dx.doi.org/10.1063/1.4919751 View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/117/17?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Growth and characterization of dilute nitride GaNxP1−x nanowires and GaNxP1−x/GaNyP1−y core/shellnanowires on Si (111) by gas source molecular beam epitaxy Appl. Phys. Lett. 105, 072107 (2014); 10.1063/1.4893745 High carrier concentration induced effects on the bowing parameter and the temperature dependence of theband gap of GaxIn1−xN J. Appl. Phys. 110, 103506 (2011); 10.1063/1.3660692 Dynamics of carrier recombination and localization in AlGaN quantum wells studied by time-resolvedtransmission spectroscopy Appl. Phys. Lett. 95, 091910 (2009); 10.1063/1.3222972 Similarities between Ga 0.48 In 0.52 N y P 1−y and Ga 0.92 In 0.08 N y As 1−y grown on GaAs (001) substrates J. Vac. Sci. Technol. B 22, 1495 (2004); 10.1116/1.1752915 Effect of nitrogen on the optical and transport properties of Ga 0.48 In 0.52 N y P 1−y grown on GaAs(001)substrates Appl. Phys. Lett. 83, 5446 (2003); 10.1063/1.1637148

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:

76.167.136.4 On: Sat, 09 May 2015 16:57:13

Temperature evolution of carrier dynamics in GaNxPyAs12y2xalloys

M. Baranowski,1 R. Kudrawiec,1,2,a) A. V. Luce,2,3 M. Latkowska,1 K. M. Yu,2,4

Y. J. Kuang (邝彦瑾),5 J. Misiewicz,1 C. W. Tu,6 and W. Walukiewicz21Laboratory for Optical Spectroscopy of Nanostructures, Department of Experimental Physics,Wroclaw University of Technology, Wybrzeze, Wyspianskiego 27, 50-370 Wroclaw, Poland2Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, USA3Department of Materials Science and Engineering, University of California, Berkeley, California 94720, USA4Department of Physics and Materials Science, City University of Hong Kong, Kowloon, Hong Kong5Department of Physics, University of California, San Diego, La Jolla, California 92093, USA6Department of Electrical and Computer Engineering, University of California, San Diego, La Jolla,California 92093, USA

(Received 5 February 2015; accepted 23 April 2015; published online 4 May 2015)

The temperature dependence of carrier dynamics in GaNxAs1�yPy alloys has been investigated by

time resolved photoluminescence. This investigation has shown that the decay time constant does

not change significantly up to 100K, and then starts to decrease rapidly above this temperature.

Additionally, the decay times at the high-energy side of the spectrum decrease faster than those at

the low-energy side. The effects have been explained by the interplay between carrier capture by

radiative and nonradiative recombination centers. Detailed simulations show that the effect of car-

rier localization in the investigated materials is better described by double-scale potential fluctua-

tions that are related to (i) distribution of localized states energy and (ii) bandgap fluctuations. In

addition, it was observed that the increase in nitrogen concentration leads to a shorter decay time at

room temperature, which is attributed to a larger concentration of non-radiative recombination cen-

ters. Furthermore, a post-growth annealing step leads to a longer decay time at room temperature,

which is attributed to a reduction in non-radiative recombination centers. At low temperatures, the

role of non-radiative centers is suppressed, and therefore the decay time does not differ signifi-

cantly for samples with either different nitrogen concentrations or in both the as-grown and

annealed samples.VC 2015 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4919751]

I. INTRODUCTION

The GaNPAs alloy material system with a small amount

of nitrogen (several atomic percent) is a promising candidate

for application in laser diodes integrated with silicon tech-

nology,1,2 and the next generation of solar cells,3,4 including

intermediate band solar cells.5 This alloy belongs to a group

of materials, the so-called Highly Mismatched Alloys

(HMAs), which are a class of semiconductors in which one

or more of the component elements are replaced with ele-

ments of different size and/or electronegativity.6–10 These

HMAs are interesting for fundamental studies since their

electronic band structure cannot be described within the vir-

tual crystal approximation and their optical properties differ

significantly from properties of regular III–V semiconductor

alloys such as GaAs, GaP, or GaPAs.

Recently, it has been shown that the electronic band

structure of GaNxPyAs1�y�x (x< 0.015 and 0.38� y� 44) is

well-explained by the Band Anticrossing (BAC) model,10

which is widely used to describe the band structure of other

HMAs. This model, applied to GaNxPyAs1�y�x alloys, con-

siders an interaction of the extended conduction band states

of the GaPAs host material and localized states of nitrogen

atoms. For y� 0.4, the nitrogen level is located very close to

the conduction band minimum of GaPyAs1�y host. Above

this P content, the band gap of GaPyAs1�y starts to be indi-

rect and the nitrogen level lies below the conduction band

minimum at the C point of Brillouin zone. Although a direct

gap is expected for GaNPAs alloys, due to the interaction of

nitrogen resonant states with the extended band states of

GaPAs host, the small energy difference between the con-

duction band minima at the C and X points of the Brillouin

zone can have important consequences for the carrier dy-

namics and therefore the application of this material system.

Time resolved photoluminescence (TRPL) spectroscopy is

an ideal tool to study carrier dynamics in semiconductor

alloys, since it measures the rise and decay time of PL signal

that strongly depend on the character of optical transitions

and the type of band gap (direct or indirect). In addition, the

dynamics of PL signal is sensitive to alloy inhomogeneity

and the presence of non-radiative centers.

Thus far, there have been no extensive studies of car-

rier dynamics in GaNxPyAs1�y�x alloys especially close to

the crossover from the direct to indirect band gap, i.e.,

0.38� y� 44. Recently, GaNxPyAs1�y�x layers lattice

matched to Si (x¼ 0.064 and y¼ 0.132) and GaNPAs quan-

tum wells were studied by TRPL and the results supported

by hopping model simulations.11 The observed a strong

non-exponential decay of PL signal at 10 K was explained

by an interplay between a fast capture of carriers by non-

radiative centers and slow radiative recombination via

localized states. However, the temperature evolution of thea)E-mail addresses: [email protected]

0021-8979/2015/117(17)/175702/8/$30.00 VC 2015 AIP Publishing LLC117, 175702-1

JOURNAL OF APPLIED PHYSICS 117, 175702 (2015)

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:

76.167.136.4 On: Sat, 09 May 2015 16:57:13

carrier dynamics and the characteristic times for extended

states was not studied in this report.11 Laser diode and solar

cell applications of GaNPAs alloys require detail knowl-

edge of the characteristic lifetimes for both localized and

delocalized excitons/carriers, as well as understanding of

the mechanism of carrier dynamics in this material system

over a broad temperature range. Furthermore, it is expected

that the tuning of both the nitrogen and phosphorus concen-

tration can influence the characteristic times and the mecha-

nism of carrier dynamics. This paper presents detailed

TRPL studies of GaNxPyAs1�y�x alloys (with y� 0.4) over

a broad temperature range, as well as a model of exciton

hopping in this material system.

The remainder of the paper is organized as follows:

Sec. II provides experimental details; Sec. III. presents the

main results, discussion, and a short description of the hop-

ping exciton model12 modified to explain the temperature

evolution of the TRPL spectra; Sec. IV summarizes the over-

all conclusions.

II. EXPERIMENTAL DETAILS

GaNxPyAs1�x�y samples with nitrogen content, x

between 0.012 and 0.015 and phosphorus content ranging

from 0.38 to 0.45 were grown in a Varian Gen-II Molecular

Beam Epitaxy (MBE) system modified to handle gas sources.

Solid elemental Ga was used to generate Ga atomic beam

through an effusion cell. Thermally cracked PH3 and AsH3 at

1000 �C and RF N plasma excited at 13.56MHz were used

for P, As, and N sources, respectively. After a 0.3lm-thick

GaP buffer layer was grown at 580 �C on the GaP substrate, a

1.5lm-thick linearly graded GaPyAs1�y and 0.5lm-thick

GaPAs with uniform As content were grown at 520 �C.Finally, the 0.5lm-thick GaN0.012As0.608P0.38 dilute-nitride

layer was grown at 520 �C. The samples were annealed at

900 �C for 60 s in 95% N2 and 5% H2 forming gas ambient in

order to improve optical properties of GaNPAs layers. The

composition of GaNPAs layers was measured by channeling

Rutherford backscattering spectroscopy (c-RBS) together

with nuclear reaction analysis (NRA). The fraction of substi-

tutional nitrogen atoms in the films was obtained by compar-

ing the random and channeling yields of the RBS and the

NRA measurements. The channeling c-RBS shows that the

samples grown are of good crystalline quality with a mini-

mum channeling yield of �5%. Channeling RBS and NRA

revealed that the substitutional N fraction is �80% for all

samples. More details on sample growth and structural char-

acterization can be found in Refs. 5 and 13.

For the TRPL experiment, the samples were placed in a

closed-cycle helium refrigerator, allowing for measure-

ments in the temperature range 11 to 300K. Samples were

optically excited by the second harmonic from a 415 nm

mode-locked Ti:Sapphire laser with pulse duration of 150

fs. The laser beam was focused on the sample to a spot di-

ameter of �0.2 mm. The average power of the pulsed laser

beam used in the experiment was 3 mW. The PL signal was

dispersed by a 0.3 m-focal length monochromator and

detected by a Hamamatsu streak camera equipped with an

S-20 photocathode.

III. RESULTS AND DISCUSSION

Figure 1 shows the time dependence of photolumines-

cence of the as-grown GaN0.012P0.38As0.608 layer measured at

11K. The analysis of the time resolved spectra will focus on

this particular sample, though very similar results were

observed in the other remaining samples. The observed PL

decay times are much longer than those reported in Ref. 11

for the samples with lower P content. This clearly indicates

that for P content close to the direct to indirect band gap

crossover (at approximately 45–50% phosphorus) the con-

duction band states in GaNPAs are a complex mixture of dif-

ferent band minima10,14 and the localized states of N. The

scale of the time dependence of PL from our sample is

between hundreds of nanoseconds (typically found for PL dy-

namics in Ga(N)P alloys with the indirect gap GaP host ma-

trix14,15) to single nanoseconds or below (measured in dilute

nitrides with a direct band gap host matrix, e.g., GaInNAs16

or GaNPAs with high As content11). The time dependence of

the PL spectrum presented in Fig. 1 exhibits a strong depend-

ence of PL decay time on energy. Such a decay time disper-

sion is a typical feature of dilute nitride materials.11,16–19

While the low-temperature decay time dispersion is well-

described in the literature, the investigated sample has an

uncommon temperature dependence in that the decay time

constant that does not change significantly up to 100K and

then rapidly decreases at higher temperatures. This behavior

has not been found in any other dilute nitride, where typically

the PL decay time decreases with increasing temperature in

the whole temperature range.17,19 The characteristic features

of the PL dynamics in the investigated samples are presented

in Figs. 2 and 3. Figures 2(a)–2(c) show the decay curves

taken at the maximum of the time-integrated PL peak. As

shown in Fig. 2(a), at low temperature, the PL decay is a sin-

gle exponential represented by a straight line on the log scale.

At higher temperatures, the PL decay becomes non-

exponential (Figs. 2(b) and 2(c)). To provide a simple quanti-

tative measure of this non-exponential behavior, we describe

the time dependence of PL intensity by a decay time s10which is the time at which the intensity of PL signal

decreases 10 times from the maximum. The dependence of

s10 versus temperature is presented in Fig. 2(d). The time s10

FIG. 1. Streak image of time dependence of PL spectra taken at 11K.

175702-2 Baranowski et al. J. Appl. Phys. 117, 175702 (2015)

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:

76.167.136.4 On: Sat, 09 May 2015 16:57:13

does not change significantly for temperatures below 100K,

and then decreases rapidly by three orders of magnitude

between 100 and 300K. Additionally, the spectral dispersion

of the decay time also changes with increasing temperature,

as shown in Fig. 3. The decay times on the high-energy side

of the spectrum decrease faster than those on the low energy

side. For example, s10 at�1.6 eV decreases by about 20 times

for the temperature increasing from 11K to 110K, whereas

the s10 is reduced by only about a factor of 2 for the low

energy emission at 1.5 eV over the same temperature range.

The PL dynamics of the GaNPAs sample shown in Figs.

1–3 exhibit three distinct features; (i) non-exponential decay

curves at higher temperatures, (ii) rapid decrease of the PL

decay time above 100K, and (iii) asymmetric change of PL

decay time dispersion with increasing temperatures. These

features strongly suggest that the PL dynamics in the investi-

gated sample are driven by a temperature-dependent process

of exciton hopping through a population of localized states,

thermal activation to the mobility edge, and capture by either

radiative or nonradiative recombination centers (incorpora-

tion of nitrogen facilitates formation of several Ga interstitial

paramagnetic defects which act as deep defects states re-

sponsible for nonradiative recombination20,21). The observed

temperature dependence of the PL dynamics can be qualita-

tively explained with the model of hopping excitons pro-

posed by Rubel,22 which is schematically shown in Fig. 4(a).

The model assumes two types of exciton trapping states:

shallow localized states and deep non-radiative recombina-

tion centers. The transfer between these two types of states

can occur only via the mobility edge. At low temperatures,

excitons captured by the localized states do not have enough

thermal energy to be transferred to the mobility edge and

they are “protected” by being isolated from non-radiative

recombination centers. At higher temperatures, excitons acti-

vated to the mobility edge move freely in the sample and can

be captured by the center of nonradiative recombination or

again by shallow localizing states. Consequently, the temper-

ature dependence of the decay time depends on the activa-

tion of the excitons to the mobility edge and the probability

of the exciton capture by nonradiative centers. It is important

to note that at high temperatures, the process of carrier

trapping and de-trapping by localized states and exciton

activation to the mobility edge can occur many times

before the exciton finally recombines either radiatively or

FIG. 2. (a)–(c) Examples of decay curve taken at the maximum of PL emis-

sion at different temperatures, (d) temperature dependence of s10 (time

where intensity decreases 10 times) at the maximum of PL spectrum.

FIG. 3. Measured time integrated PL spectra at 11 and 110K (black lines)

together with decay times (s10Þ for different energies (open squares). The

red and green areas are results of photoluminescence simulation with the

Gaussian broadening of the bandgap for 11 and 110K, respectively. Dashed

lines are results of decay time dispersion simulations.

FIG. 4. (a) Schema of exciton dynamics—excitons can recombine radia-

tively from the mobility edge or being localized on localizing center, exci-

tons can be activated to the mobility edge from shallow localizing states.

From the mobility, edge excitons can be again captured by a localized state

or by a center of nonradiative recombination. (b) Schema energy distribution

of localizing states within a material with local bandgap fluctuation. The

activation energy (Ea1, Ea2) can be different for the states with the same

energy of emitted photon.

175702-3 Baranowski et al. J. Appl. Phys. 117, 175702 (2015)

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:

76.167.136.4 On: Sat, 09 May 2015 16:57:13

non-radiatively. The probability of exciton activation to the

mobility edge and thus also the nonradiative recombination

rate increases with increasing temperature. The potential

fluctuations and the exciton separation from nonradiative

centers have been previously invoked in explaining the

increase of photoluminescence efficiency from InGaN/GaN

quantum wells with low indium content.23 In the GaNPAs

alloys studied in this work, we observe that the excitons cap-

tured by localized states are efficiently separated from non-

radiative recombination and the decay constant s10 at PL

maximum is almost constant at temperatures lower than

100K. At higher temperatures, excitons are activated to the

mobility edge and can migrate over the whole sample,

increasing the probability of non-radiative recombination

and reducing the PL decay time at PL maximum. In the

same time, the temperature dependence of integrated PL in-

tensity is different than the decay time constant at the PL

maximum, what is shown in Figure 5. We can observe sys-

tematic decrease of PL intensity with the increasing tempera-

ture. However, this behavior still can be understand within

the presented scenario of carrier dynamics. Observed

decrease of PL intensity with the increase of temperature is

related to the fast quenching of high energy part of the spec-

trum (blue points). This part comes from the weekly local-

ized excitons that can be activated to mobility edge even at

relatively low temperatures. As a result of this, the intensity

integrated over whole spectrum also decreases. In the same

time, the low energy part of the spectrum is not so strongly

quenched what indicates that excitons localized on the

deeper localizing states are protected from nonradiative

recombination. The described scenario also explains the

temperature-induced change of the decay time dispersion.

The high (low) energy side of the PL spectrum originates

from recombination of excitons localized on shallower

(deeper) traps. The lifetime at the high (low) energy side of

the PL spectrum decreases faster (slower) with increasing

temperature because of a smaller (larger) barrier for activa-

tion of excitons to the mobility edge.

To provide a more quantitative description of the pro-

posed mechanism of carrier recombination in GaNPAs, we

perform simulations of the PL spectra using the hopping

excitons model.12,22 Since this model has been already

extensively described before,11,12,19,22,24 here we only dis-

cuss the parameters and some modifications from the origi-

nal approach.

Initially, the energy of the localized states was chosen to

create an exponential density of states (DOS): DOS eð Þ¼ Nr=L

3

e0expð� e

e0Þ, where L is the size of the rectangle volume

and Nr is the number of shallow localized states in this vol-

ume, e0 is the localization parameter describing the localized

states distribution, and e is a localization energy. To imitate

the pulse excitation condition in TRPL experiment at the be-

ginning of simulation, n0 free excitons are created (changing

this ratio between n0 and number of localizing states we are

able to imitate changes in the excitation power density in

real experiment.). Those excitons can be trapped by a local-

ized state or center of nonradiative recombination or recom-

bine radiatively with a rate given by the inverse of its

lifetime vr ¼ s�10 . When excitons are localized, they can be

activated to the mobility edge with the rate given by

va ¼ v0 expð�e=kbTÞ, or hop from localized state i to jwith a rate described by the Miller-Abrahams formula

vij ¼ v0ð� 2rija � ej�eiþjej�eij

2kbTÞ, where v0 is the attempt to escape

frequency. These excitons activated to the mobility edge can

be again captured by the shallow localized states or can be

captured by non-radiative recombination centers. The proba-

bility of being captured by non-radiative states is propor-

tional to the ratio of Nnr=ðNr þ NnrÞ. The values of the

parameters are listed in Table I.

Figure 6 shows the simulated temperature dependence of

s10 for different values of the average energy of shallow

localized states (dashed lines) along with the experimental

data (squares). As shown in Fig. 6, the value of the localiza-

tion parameter determines the temperature at which the decay

time starts to decrease. Moreover, this energy also determines

the decay time reduction rate. The best agreement is obtained

for e0 ¼ 20 meV; however, with this value of e0, it is difficultto reproduce the total intensity of the time integrated PL spec-

trum. According to the model of hopping excitons, the PL

full width at half maximum (FWHM) is related to the value

of e0 through the relation FWHM � 3e0.25 In our case, the

investigated structure has a PL FWHM of about �90meV,

which gives e0 ¼ 30meV. For such a high value of the aver-

age localization energy, we are not able to reconstruct the

temperature dependence of s10, even when using a higher

concentration of nonradiative centers (not shown here).

To resolve the discrepancy in the value of the localization

parameter obtained from the temperature dependence of

decay time and from the spectrum broadening, we have

applied a double scale potential fluctuation model, which is

schematically shown in Fig. 4(b). We assume that the large

broadening of the PL spectrum is the result of both the distri-

bution of localized state energies and local bandgap fluctua-

tions caused by fluctuations of alloy composition.

FIG. 5. Dependence of integrated PL intensity vs. temperature for the whole

spectrum (black points), low and high energy part of the spectrum (red and

blue points, respectively), the bars in the inset show the areas taken for the

low and high energy part integration.

TABLE I. Value of essential parameters used in simulations.

s0 (ns) �0s0 Nnr/(Nr þ Nnr) Nr(a/L)3 n0/(Nr þ Nnr)

20 2�104 0.167 0.026 0.1

175702-4 Baranowski et al. J. Appl. Phys. 117, 175702 (2015)

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:

76.167.136.4 On: Sat, 09 May 2015 16:57:13

Consequently, the overall emission band is a superposition of

excitons recombining in different areas of the sample having

slightly different bandgaps. This approach has been previ-

ously applied to explain the unusual temperature dependence

of the PL spectrum in InGaN layers26 and in GaNAsP quan-

tum wells.27 The solid lines in Figures 3, 6, and 7 show results

of our simulation performed with a Gaussian distribution of

the bandgap energy with a broadening parameter r¼ 25meV.

We show that by adding the local bandgap fluctuations, we

are able to reproduce the temperature dependence of PL decay

time s10 (solid lines in Fig. 6), decay curves (Fig. 7), and

decay time dispersion (Fig. 3). Also, as is shown in Fig. 3, the

PL spectrum itself is well reproduced. The value of s0 has

been determined to be 20 ns. This relatively long carrier life-

time is important for the potential application of GaNPAs as

solar cell absorber material, especially compared to other

III–V semiconductors, which typically have much shorter car-

rier lifetimes. It should be noted that the much shorter carrier

lifetime at room temperature can be improved by optimizing

the growth conditions and/or the post-growth annealing condi-

tions. The local bandgap fluctuation can also explain the non-

exponential behavior of PL decay at higher temperatures.

Similar behavior was previously observed in InGaN

QWs28–31 and explained by alloy composition fluctuations in

QW.32

The emission energy from deeper exciton trapping states

located in the larger gap region can overlap with shallower

exciton trapping states located in a smaller gap region.

Superposition of those two emission channels with distinctly

different lifetimes will result in a nonexponetial PL

decay. An additional mechanism contributing to the non-

exponential decay at higher temperatures comes from a com-

plex interplay between the fast capture of carriers by nonra-

diative centers and these lower radiative recombination via

localized states described in Ref. 11.

As mentioned previously, a similar photoluminescence

dynamics was also observed for the remaining GaNPAs sam-

ples. Fig. 8 shows a direct comparison of decay curves taken

at the maximum of PL emission at 300K for GaNPAs sam-

ples with different nitrogen concentration of 1.2 and 1.5%.

In addition, the decay curve for the annealed sample with

1.2% N is shown in this figure. Comparison of the samples

with different nitrogen concentration shows that the increase

in nitrogen concentration from 1.2 to 1.5% leads to a

decrease of s10 from 0.13 ns to 0.06 ns. This effect is attrib-

uted to an increase in the concentration of N-related non-

radiative recombination centers. On the other hand, the post-

growth annealing increases s10 by 20% for the sample with

1.2% N. This effect is attributed to a reduction of non-

FIG. 6. Temperature dependence of s10 (open squares) together with simu-

lation performed for different values of localized state average energy, solid

lines with bandgap fluctuation and dashed lines without bandgap fluctuation.

FIG. 7. Comparison of measured decay curves at PL maximum with simula-

tions.

FIG. 8. Decay curves taken at the maximum of PL at 300K for all measured

samples. The inset shows temperature dependence of s10 for all investigatedsamples.

175702-5 Baranowski et al. J. Appl. Phys. 117, 175702 (2015)

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:

76.167.136.4 On: Sat, 09 May 2015 16:57:13

radiative recombination centers upon annealing and is typi-

cal for dilute nitrides.33–35 It is worth noting that previously

reported annealing of Ga(In)NAs causes significant changes

of the decay times.35,36 The less pronounced improvement in

the decay time in GaNPAs alloys could be related to non-

optimized annealing conditions. The role of non-radiative

centers is suppressed at low temperatures and therefore the

decay times do not differ significantly for samples with dif-

ferent nitrogen concentrations as well as for the as-grown

and annealed samples, as shown in the inset in Fig. 8.

At the end, we want to make some extensive comment

about differences of nonradiative recombination mechanism

in GaNAsP and Ga(In)NAs alloy. The mechanism of thermal

quenching of excitonic emission from hopping to non-

radiative centers via the mobility edge is expected to be pres-

ent in other dilute nitrides as well. In fact, its significance

depends on the relation between the exciton binding energy

and the activation energy of excitons to the mobility edge.

For Ga(In)NAs alloys, it was observed that the main reason

for thermal quenching of photoluminescence is the dissocia-

tion of excitons captured by localized states.19,24,37 Such

conclusion is consistent with experimental observations pre-

sented in the literature for Ga(In)NAs alloys and QWs.38,39

Thus, our simulation using the hopping excitons model with

the exciton dissociation mechanism as the main reason for

non-radiative recombination reproduces both the photolumi-

nescence and time resolved photoluminescence spectra and

their characteristic features in some materials19,24,37 where

the activation energy of excitons to the mobility edge is

larger than the exciton binding energy. For materials with

the exciton binding energy larger than the average localiza-

tion energy, i.e., samples studied in this work, it is impossi-

ble to simulate the dynamics of photoluminescence without

the assumption that the mechanism of exciton transfer to

non-radiative centers via the mobility edge exists. This sug-

gests that exciton dissociation is not the main reason of non-

radiative recombination GaNPAs samples studied in this

paper. In this case, the activation energy of excitons to the

mobility edge is smaller than the exciton binding energy and

hence the thermal quenching of photoluminescence in the

investigated GaNPAs alloys is due to the transfer of excitons

to non-radiative centers via the mobility edge.

A simple estimation of exciton binding energy in

GaNPAs gives the value of �30meV, which is 6–7 times

larger than the exciton binding energy in Ga(In)NAs

layers,38,39 and 50% larger than the activation energy used

in our simulations (e0 ¼ 20 meV). The electron effective

mass (at the C point) in pure GaP is two times higher than in

GaAs and also the hole masses are higher.40 Additionally,

the incorporation of nitrogen makes the conduction band of

GaNAsP alloy very flat, see Ref. 10 which leads to a further

increase of the electron effective mass. Due to these reasons,

the exciton binding energy in this material system is esti-

mated to be �30meV. This difference between exciton

binding energy inGaNAs and GaNPAs alloys explains why

the dominant nonradiative recombination mechanism in

GaNPAs is different than in the Ga(In)NAs alloys studied in

Refs. 19, 24, and 37–39.

In order to illustrate the effect of exciton binding energy

on the carrier dynamics of photoluminescence, simulations

of PL decay time were performed for a hypothetic alloy with

the same average localization energy (e0 ¼ 20 meV) and a

different exciton binding energy (the rate of exciton dissocia-

tion was set to be 20 times larger than the radiative recombi-

nation rate (typical for dilute nitrides). For a detailed

description of this nonradiative recombination mechanism,

see Ref. 24. Results of these simulations for four different

exciton binding energies are shown in Fig. 9. When the exci-

ton binding energy is large (30meV and 40meV), the tem-

perature evolution of s10 is very similar to the case without

an exciton dissociation mechanism (thick solid line in Fig.

9). On the other hand, when the exciton binding energy is

comparable to the average localization potential (e0 ¼ 20

meV), the impact of exciton dissociation cannot be

neglected, illustrated by comparison of the thick dashed line

with the thick solid line in Fig. 9. In general, the two mecha-

nisms of thermal quenching of photoluminescence are both

present simultaneously, but often one of them is dominant.

FIG. 9. Simulation of temperature dependence of s10 for e0 ¼ 20 meV and

different values of exciton binding energy.

FIG. 10. Simulation of temperature dependence of s10 with different mecha-

nisms of nonradiative recombination for the exciton binding energy lower

than average localization energy (a), and for higher exciton binding

energy (b).

175702-6 Baranowski et al. J. Appl. Phys. 117, 175702 (2015)

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:

76.167.136.4 On: Sat, 09 May 2015 16:57:13

Fig. 10 shows simulations of the temperature dependence of

PL decay time performed assuming different mechanisms of

nonradiative recombination: (i) an exciton dissociation, (ii) a

recombination via nonradiative centers, and both (i) and (ii).

The average localization potential is assumed to be e0 ¼ 20

meV. As shown in Fig. 10(a), when the exciton binding

energy is significantly lower than the localization energy, the

nonradiative recombination is dominated by exciton dissoci-

ation at low temperatures (the blue and red curves overlap in

the range 0–70K), and the influence of nonradiative states

becomes important at higher temperatures. On the other

hand, when the exciton binding energy is higher than the

localization potential, then the impact of nonradiative centers

on the reduction of decay time is rather weak as shown in

Figs. 10(b) and 9.

IV. SUMMARY

In conclusion, the temperature dependence of the carrier

dynamics in GaNxPyAs1�x�y (y� 40%, x� 1.2–1.5%) alloys

has been investigated by time resolved photoluminescence.

The PL decay time was found to exhibit a distinct

temperature-dependent spectral dispersion and a strong

reduction in the decay time constant above 100K. These

changes in PL decay time have been explained by the inter-

play between carrier capture by radiative and non-radiative

recombination centers. The role of non-radiative centers is

suppressed at low temperatures and therefore the decay time

does not differ significantly for samples with different nitro-

gen concentrations and is not affected by sample annealing.

At higher temperature when the non-radiative recombination

starts to play a dominant role, a reduced concentration of

non-radiative centers explains the longer PL decay times in

the lower N-content or thermally annealed samples. Thanks

to the simulation, we estimate the radiative lifetime of exci-

tons in investigated alloys to be about 20 ns.

ACKNOWLEDGMENTS

This work was performed within the grant of the

National Science Centre, Poland (No. 2013/10/M/ST3/

00638). Sample growth by GSMBE at UCSD was supported

by National Science Foundation Grant Nos. DMR-0907652

and DMR-1106369. RBS, NRA, absorption and PR

characterization performed at LBNL was supported by the

Director, Office of Science, Office of Basic Energy Sciences,

Materials Sciences and Engineering Division, of the U.S.

Department of Energy under Contract No. DE-AC02-

05CH11231. A.V.L. acknowledges the support of an NSF

Graduate Research Fellowship. The experiment is partially

performed within the NLTK infrastructure, Project No.

POIG. 02.02.00-003/08-00. M.B. acknowledges the support

from the National Science Centre, Poland (DEC–2012/07/N/

ST3/03173).

1S. Liebich, M. Zimprich, A. Beyer, C. Lange, D. J. Franzbach, S.

Chatterjee, N. Hossain, S. J. Sweeney, K. Volz, B. Kunert, and W. Stolz,

Appl. Phys. Lett. 99, 071109 (2011).

2N. Koukourakis, C. Buckers, D. A. Funke, N. C. Gerhardt, S. Liebich, S.

Chatterjee, C. Lange, M. Zimprich, K. Volz, W. Stolz, B. Kunert, S. W.

Koch, and M. R. Hofmann, Appl. Phys. Lett. 100, 092107 (2012).3J. F. Geisz and D. J. Friedman, Semicond. Sci. Technol. 17, 769 (2002).4K. M. Yu, W. Walukiewicz, J. W. A. Iii, D. Bour, R. Farshchi, O. D.

Dubon, S. X. Li, I. D. Sharp, and E. E. Haller, Appl. Phys. Lett. 88,092110 (2006).

5Y. J. Kuang, K. M. Yu, R. Kudrawiec, A. V. Luce, M. Ting, W.

Walukiewicz, and C. W. Tu, Appl. Phys. Lett. 102, 112105 (2013).6W. Shan, W. Walukiewicz, J. W. Ager, E. E. Haller, J. F. Geisz, D. J.

Friedman, J. M. Olson, and S. R. Kurtz, Phys. Rev. Lett. 82, 1221 (1999).7W. Walukiewicz, W. Shan, K. M. Yu, J. W. Ager, E. E. Haller, I.

Miotkowski, M. J. Seong, H. Alawadhi, and A. K. Ramdas, Phys. Rev.

Lett. 85, 1552 (2000).8J. Wu, W. Walukiewicz, K. M. Yu, J. W. Ager, E. E. Haller, Y. G. Hong,

H. P. Xin, and C. W. Tu, Phys. Rev. B 65, 241303 (2002).9J. Wu, W. Shan, and W. Walukiewicz, Semicond. Sci. Technol. 17, 860(2002).

10R. Kudrawiec, A. V. Luce, M. Gladysiewicz, M. Ting, Y. J. Kuang, C. W.

Tu, O. D. Dubon, K. M. Yu, and W. Walukiewicz, Phys. Rev. Appl. 1,034007 (2014).

11K. Jandieri, B. Kunert, S. Liebich, M. Zimprich, K. Volz, W. Stolz, F.

Gebhard, S. D. Baranovskii, N. Koukourakis, N. C. Gerhardt, and M. R.

Hofmann, Phys. Rev. B 87, 035303 (2013).12S. D. Baranovskii, R. Eichmann, and P. Thomas, Phys. Rev. B 58, 13081(1998).

13Y.-J. Kuang, S.-W. Chen, H. Li, S. K. Sinha, and C. W. Tu, J. Vac. Sci.

Technol. B 30, 02B121 (2012).14I. A. Buyanova, G. Pozina, J. P. Bergman, W. M. Chen, H. P. Xin, and C.

W. Tu, Appl. Phys. Lett. 81, 52 (2002).15T. Niebling, O. Rubel, W. Heimbrodt, W. Stolz, S. D. Baranovskii, P. J.

Klar, and J. F. Geisz, J. Phys. Condens. Matter 20, 015217 (2008).16A. Kaschner, T. Luttgert, H. Born, A. Hoffmann, A. Y. Egorov, and H.

Riechert, Appl. Phys. Lett. 78, 1391 (2001).17R. A. Mair, J. Y. Lin, H. X. Jiang, E. D. Jones, A. A. Allerman, and S. R.

Kurtz, Appl. Phys. Lett. 76, 188 (2000).18I. A. Buyanova, G. Pozina, P. N. Hai, W. M. Chen, H. P. Xin, and C. W.

Tu, Phys. Rev. B 63, 033303 (2000).19M. Baranowski, R. Kudrawiec, M. Latkowska, M. Syperek, J. Misiewicz,

and J. A. Gupta, Appl. Phys. Lett. 100, 202105 (2012).20D. Dagnelund, J. Stehr, A. Y. Egorov, W. M. Chen, and I. A. Buyanova,

Appl. Phys. Lett. 102, 021910 (2013).21N. Q. Thinh, I. P. Vorona, I. A. Buyanova, W. M. Chen, S.

Limpijumnong, S. B. Zhang, Y. G. Hong, H. P. Xin, C. W. Tu, A. Utsumi,

Y. Furukawa, S. Moon, A. Wakahara, and H. Yonezu, Phys. Rev. B 71,125209 (2005).

22O. Rubel, S. D. Baranovskii, K. Hantke, B. Kunert, W. W. Ruhle, P.

Thomas, K. Volz, and W. Stolz, Phys. Rev. B 73, 233201 (2006).23S. Nakamura, Science 281, 956 (1998).24M. Baranowski, M. Latkowska, R. Kudrawiec, and J. Misiewicz, J. Phys.

Condens. Matter 23, 205804 (2011).25H. Gruning, K. Kohary, S. D. Baranovskii, O. Rubel, P. j. Klar, A.

Ramakrishnan, G. Ebbinghaus, P. Thomas, W. Heimbrodt, W. Stolz, and

W. w. Ruhle, Phys. Status Solidi C 1, 109 (2004).26K. Kazlauskas, G. Tamulaitis, A. Zukauskas, M. A. Khan, J. W. Yang, J.

Zhang, G. Simin, M. S. Shur, and R. Gaska, Appl. Phys. Lett. 83, 3722 (2003).27C. Karcher, K. Jandieri, B. Kunert, R. Fritz, M. Zimprich, K. Volz, W.

Stolz, F. Gebhard, S. D. Baranovskii, and W. Heimbrodt, Phys. Rev. B 82,245309 (2010).

28C.-K. Sun, S. Keller, G. Wang, M. S. Minsky, J. E. Bowers, and S. P.

DenBaars, Appl. Phys. Lett. 69, 1936 (1996).29P. Lefebvre, A. Morel, M. Gallart, T. Taliercio, J. Allegre, B. Gil, H.

Mathieu, B. Damilano, N. Grandjean, and J. Massies, Appl. Phys. Lett. 78,1252 (2001).

30S. F. Chichibu, M. Sugiyama, T. Onuma, T. Kitamura, H. Nakanishi, T.

Kuroda, A. Tackeuchi, T. Sota, Y. Ishida, and H. Okumura, Appl. Phys.

Lett. 79, 4319 (2001).31Y.-H. Cho, S. K. Lee, H. S. Kwack, J. Y. Kim, K. S. Lim, H. M. Kim, T.

W. Kang, S. N. Lee, M. S. Seon, O. H. Nam, and Y. J. Park, Appl. Phys.

Lett. 83, 2578 (2003).32M. Gladysiewicz and R. Kudrawiec, Phys. Status Solidi A 209, 752

(2012).33T. Ahlgren, E. Vainonen-Ahlgren, J. Likonen, W. Li, and M. Pessa, Appl.

Phys. Lett. 80, 2314 (2002).

175702-7 Baranowski et al. J. Appl. Phys. 117, 175702 (2015)

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:

76.167.136.4 On: Sat, 09 May 2015 16:57:13

34S. G. Spruytte, C. W. Coldren, J. S. Harris, W. Wampler, P. Krispin, K.

Ploog, and M. C. Larson, J. Appl. Phys. 89, 4401 (2001).35M. Baranowski, R. Kudrawiec, M. Latkowska, M. Syperek, J. Misiewicz,

T. Sarmiento, and J. S. Harris, J. Phys. Condens. Matter 25, 065801

(2013).36R. Kudrawiec, M. Syperek, M. Latkowska, J. Misiewicz, V.-M.

Korpijarvi, P. Laukkanen, J. Pakarinen, M. Dumitrescu, M. Guina, and M.

Pessa, J. Appl. Phys. 111, 063514 (2012).

37R. Kudrawiec, M. Latkowska, M. Baranowski, J. Misiewicz, L. Li, and J.

Harmand, Phys. Rev. B 88, 125201 (2013).38M. Latkowska, R. Kudrawiec, G. Sek, J. Misiewicz, J. Ibanez, M. Henini,

and M. Hopkinson, Appl. Phys. Lett. 98, 131903 (2011).39M. Latkowska, R. Kudrawiec, J. Misiewicz, Y. G. Gobato, M. Henini, and

M. Hopkinson, J. Phys. Appl. Phys. 46, 402001 (2013).40I. Vurgaftman, J. R. Meyer, and L. R. Ram-Mohan, J. Appl. Phys. 89,5815 (2001).

175702-8 Baranowski et al. J. Appl. Phys. 117, 175702 (2015)

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:

76.167.136.4 On: Sat, 09 May 2015 16:57:13