GaN CoreâShell ... - ACS Publications

Subscriber access provided by UNIV OF CAMBRIDGE

Communication

Nanoscopic insights into InGaN/ GaN core-shell nanorods: structure, composition, and luminescence Marcus Müller, Peter Veit, Florian F. Krause, Tilman Schimpke, Sebastian Metzner, Frank Bertram, Thorsten Mehrtens, Knut Müller-Caspary, Adrian Avramescu, Martin Strassburg, Andreas Rosenauer, and Juergen Christen

Nano Letters is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society.


Nano Lett., Just Accepted Manuscript • Publication Date (Web): 12 Aug 2016

Downloaded from http:// pubs.acs.org on August 13, 2016

Just Accepted

“Just Accepted” manuscripts have been pe online prior to technical editing, formatting Society provides “Just Accepted” as a dissemination of scientific material as soon appear in full in PDF format accompanied b fully peer reviewed, but should not be consi readers and citable by the Digital Object Ide to authors. Therefore, the “Just Accepted” in the journal. After a manuscript is techni Accepted” Web site and published as an A changes to the manuscript text and/or gra



and ethical guidelines that apply to the j or consequences arising from the use of i


(a)

Page 1 of 27

(c)

InGaN AlGaN

n-GaN core n-GaN shell

50 nm GaN cap In-concentration (%)

1 2 3 4 5 6 7 8 9 10 11 12 (b) 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

Nano Letters

15

10

InGaN SQW

n-GaN shell

(d) medium [In] (11.5 +- 1.3) %

5

ACS Paragon Plus Environment

1 µm

0

Position across InGaN QW

(a)

(b) 500Nano nm Letters

500 nm Page 2 of 27

Norm. CL Intensity (a. u.)

1 2 (II) (III) 3 4 5 (I) 6 7 8 (IV) 9 10 11 12 13 (V) 14 15 16 17 CL Intensity (a. u.) 350 400 450 500 550 Wavelength (nm) 18 19 (c) QW 20 458 nm 21 (V) InGaN QW 22 QW 23 443 nm 24 (IV) InGaN QW 25 0 26 (D ,X) 27 357 nm (III) n-GaN shell 28 29 30 AlGaN 31 351 nm (II) AlGaN shell 32 33 GaN YL 34 550 nm 35 ACS Paragon Plus(I)Environment c-GaN core 36 400 450 500 550 600 650 37 350 Wavelength (nm)

600

(a)3 of 27 Page 8

471 nm Nano Letters

top

7

Position (µm)

Position (µm)

1 26 3 45 5 64 7 3 8 92 10 11 1 12 0 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38

bottom

410 nm 350

400

CL Intensity (a. u.) 450

500

550

600

650

Wavelength (nm) Average indium-concentration (at %) 10.5 8

11.0

11.5

12.0

12.5

13.0

(b)

13.5

top

7 6 5 4 3 2

bottom 1

quantum-well thickness In-concentration ACS Paragon Plus Environment

0 5

6

7

8

9

10

11

12

13

Quantum-well thickness (nm)

14

1 2 3 4 5 6 7 (c) 8 9 10 11 12 13 14 15 16

(b)

InGaN QW

Nano Letters

Page 4 of20 27 15

GaN

10 5

5 nm InGaN QW

5 nm

0 20

(d)

15

GaN

10 5

ACS Paragon Plus Environment 0

5 nm

In-concentration (at. %) In-concentration (at. %)

(a)

5 nm

(a)

(b)

Page 5 of 27Nano Letters

InGaN GaN QW shell

integral intensity

CL Intensity (a. u.)

1 2 3 4 5 6 7 8 30 nm 30 nm 9 10 11 (c) InGaN 12 430 nm 13 FWHM: 250 meV 14 15 16 17 18 19 20 21 350 400 450 500 550 600 650 22 Wavelength (nm) 23 24 405 - 424 nm (e) 425 - 439 nm (d) 25 26 27 28 29 30 31 32 33 34 35 30 nm 30 nm 36 37 (f) 38 440 - 455 nm (g) 456 - 474 nm 39 40 41 42 43 44 45 46 47ACS Paragon Plus Environment 48 49 30 nm 30 nm

600

550 1 2 3 4 500 5 6 7 450 8 9 10 11 400 12ACS Paragon Plus Environment 13 14 350 15

CL Wavelength (nm)

500Letters nm Page 6 of 27 Nano

Page 7 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

Nanoscopic insights into InGaN/GaN core-shell nanorods: structure, composition, and luminescence Marcus Müller1, Peter Veit1, Florian F. Krause2, Tilman Schimpke3, Sebastian Metzner1, Frank Bertram1, Thorsten Mehrtens2, Knut Müller-Caspary2, Adrian Avramescu3, Martin Strassburg3, Andreas Rosenauer2 and Jürgen Christen1 1

Institute of Experimental Physics, Otto-von-Guericke-University Magdeburg, Magdeburg,

Germany 2

Institute of Solid State Physics, University of Bremen, Bremen, Germany

3

OSRAM Opto Semiconductors GmbH, Regensburg, Germany

KEYWORDS: scanning transmission electron microscopy, cathodoluminescence spectroscopy, HAADF STEM, composition determination, nanorods, InGaN, Table of Contents Graphic:


1

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 27

ABSTRACT Three-dimensional nitride based core-shell nanorods (NRs) are promising candidates to achieve highly efficient optoelectronic devices. For a detailed understanding of the complex core-shell layer structure of InGaN/GaN NRs, a systematic determination and correlation of the structural, compositional, and optical properties on a nanometer-scale is essential. In particular, the combination of low-temperature cathodoluminescence (CL) spectroscopy directly performed in a scanning transmission electron microscope (STEM) and quantitative high-angle annular dark field imaging enables a comprehensive study of the nanoscopic attributes of the individual shell layers. The investigated InGaN/GaN core-shell NRs, which were grown by metal-organic vapor phase epitaxy using selective area growth exhibit an exceptional low density of extended defects. Using highly spatially resolved CL mapping of single NRs performed in cross-section, we give a direct insight into the optical properties of the individual core-shell layers. Most interesting, we observe a red shift of the InGaN single quantum well from 410 nm to 471 nm along the non-polar side wall. Quantitative STEM analysis of the active region reveals an increasing thickness of the SQW from 6 nm to 13 nm, accompanied by a slight increase of the indium concentration along the nonpolar side wall from 11 % to 13 %. Both effects, the increased quantum well thickness and the higher indium incorporation, are responsible for the observed energetic shift of the InGaN SQW luminescence. Furthermore, compositional mappings of the InGaN quantum well reveal the formation of locally indium rich regions with several nanometers in size, leading to potential fluctuations in the InGaN SQW energy landscape. This is directly evidenced by nanometer-scale resolved CL mappings which show strong localization effects of the excitonic SQW emission.


2

Page 9 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

In recent years, several studies have shown that three-dimensional nitride based nanorod (NR) light emitting diodes provide promising candidates for future highly efficient solid state lighting (1; 2; 3; 4). The controlled fabrication of GaN NRs with high surface-to-volume ratio and small footprint reduces the density of extended defects significantly, offering the advantage of high crystal quality shell layers grown around the initial core (5; 6). Moreover, the epitaxy of the active region on the non-polar side walls decreases the negative impact of polarization fields and the associated quantum-confined Stark effect (QCSE), which is expected to boost the efficiency of the device (7) (8). Most promisingly, the novel three-dimensional design of core-shell NR LEDs leads to a significant enlargement of the effective light emitting area compared to conventional planar heterostructures (9), reducing the local current density and thus the “droop” occurring in the InGaN material system (10). High aspect ratios of the NRs, uniformity over the full wafer, as well as controlled size, position, and alloy composition of the core-shell layer structure are required to gain an advantage over state-of-the-art planar LEDs. Nevertheless, not only the homogeneity of the NR ensemble is important, but also the uniformity of the luminescence and compositional properties along single NRs are crucial for the efficiency of commercial devices. Consequently, new requirements for the characterization methods of individual NRs arise from these novel semiconductor heterostructures. When going from conventional planar LEDs to three dimensional nanostructures, most standard characterization techniques are challenged with the investigation of the nanoscopic optical, structural and chemical NR properties. Performing cathodoluminescence spectroscopy (CL) directly in a scanning transmission electron microscope (STEM) enables nanometer-scale spatially resolved correlation of the optical characteristics with the real crystalline structure. Several actual studies on nitride-based heterostructures have demonstrated the high spatial resolution of STEM-CL (11; 12; 13). In addition, the quantitative


3

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 27

analysis of the high-angle annular dark field (HAADF) contrast when using STEM enables mapping the composition of ternary alloys (14; 15). Combining low temperature CL spectroscopy and quantitative HAADF analysis in a STEM allows an exceptional look inside the nanostructure of InGaN quantum wells and is fundamental for the further improvement of three-dimensional NR LEDs. In this letter, we present a unique approach for the direct one-by-one correlation of the optical, structural, and chemical properties of high-aspect ratio nitride-based core-shell NRs with a spatial resolution of a few nanometers. The InGaN/GaN core-shell NR array was fabricated by metal-organic vapor phase epitaxy (MOVPE) in a commercial, high-volume, vertical flow MOVPE reactor on a 4” c-plane sapphire substrate using an optimized n-doped GaN template. A 30 nm thick SiO2-mask with circular holes for position control of the NRs was obtained by nanoimprint lithography. The hole pattern was arranged in a hexagonal lattice with a pitch of 3 µm and a hole diameter of 800 nm, respectively. Consequently, the MOVPE process under high SiH4 flow rate results in a well-defined threedimensional selective area growth of n-doped GaN NRs out of the mask openings. Previous studies have reported that a high silane flow increases the vertical growth rate resulting in high-aspectratio NRs (16). Thus, the NRs were grown with a silane flux of 450 nmol/min and a V/III ratio of 100 which results in a nominally free-carrier concentration of ~ 1 x 1020 cm-3 as recently published by Mohajerani et al. (17). To avoid the formation of a SiNx passivation layer on top of the core as previously observed by Hartmann et al., the SiH4 concentration was reduced towards the end of the growth leading to a decreased free-carrier concentration of ~ 2 x 1019 cm-3 in the upper part of the nanorod (16; 17). An AlGaN layer and an n-doped GaN shell were deposited epitaxially on top of the high Si-doped GaN core. The AlGaN shell acts as a marker layer to clearly identify the interface of the n-doped core and the GaN shell layer in STEM and CL measurements (18). After


4

Page 11 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

the GaN shell, a single InGaN quantum well (SQW) was grown as the optically active shell layer around the complete structure. Finally, the whole core-shell NR was capped with an undoped GaN layer. The V/III ratio was increased by two orders of magnitude for the shell deposition. Detailed information about the growth conditions can be found in reference (4; 16; 17). Low-temperature CL imaging was performed in a FEI STEM Tecnai F20 to correlate the optical and structural properties of the core-shell heterostructure with a nanoscale spatial resolution. In particular, low temperature experiments are essential to investigate the optical properties of bound excitonic complexes located in the potential minima and thus study the influence of local indium fluctuations on the luminescence characteristics. When raising the temperature towards roomtemperature, excitons are no longer frozen out within potential minima and cross the mobility edge resulting in the free-exciton recombination in the InGaN quantum wells (19). Therefore, the specimen was cooled down to T = 16 K in a modified liquid helium TEM sample holder. The generated CL is collected by a parabolically shaped aluminum mirror located above the sample, coupled into a grating monochromator MonoCL4 (Gatan) and is detected by a liquid N2 cooled SiCCD. For STEM-CL measurements, the sample was prepared in cross-section using a conventional mechanical wedge-polishing technique combined with Ar+ ion milling in a Gatan PIPS2 ion polishing system as reported in our previous work (20). To avoid surface damage of the TEM sample and formation of amorphous layers, the acceleration voltage and the angle of incidence ranged from 3–5 kV and 4–6°, respectively. The lamella was thinned down to a thickness of about 200 nm to obtain an optimum ratio of electron transparency for high resolution and CL intensity. STEM-CL imaging was performed under low electron doses per area, to avoid damage of the InGaN layer due to electron beam irradiation. Further information about the experimental method and preparation process can be found in earlier publications (11; 21).


5

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 27

Chemical mapping of the composition in the InGaN single quantum well structure was achieved by quantitative HAADF STEM investigations performed in a FEI Titan 80/300 G1 microscope with a corrector for spherical aberration. The measurements were executed at 300 kV with a probe semiconvergence of 9 mrad at a probe current of 3.3 pA measured from the single electron signal of the HAADF detector (22). It has been clearly demonstrated that with STEM at these parameters no significant beam damage is caused in InGaN over time periods much longer than the acquisition times used here. Especially artificial In clustering due to the irradiation does not occur (14). The electrons scattered into high-angles (36 – 250 mrad) were recorded by a Fischione Model 3000 HAADF detector. A quantitative evaluation of the HAADF image contrast by comparison of the normalized experimental intensities with frozen lattice calculations was utilized to determine the indium concentration. Unlike the previously described CL investigation, selected NRs have been cut out from the array for the HAADF analysis and thinned down to electron transparency by focused ion beam preparation (FIB). Using a FIB prepared specimen, a homogeneous sample thickness of ~ 120 nm is achieved, which is necessary for the compositional analysis of the layers. Detailed information about the compositional analysis by HAADF STEM can be found elsewhere (14; 15; 23). Figure 1(a) shows a SEM image in bird’s eye view of the as-grown core-shell NR sample. The selective area growth results in a homogeneous array of highly uniform nanorods with a mean density of 1.3  105 rods/mm2 and cover 10 % of the complete planar surface area of the underlying GaN template. Typically, the hexagonal shaped NRs are terminated by {1-100} m-plane side walls and {1-101} semi-polar tips. The average diameter and height of the NRs was measured to be 1.1 µm and 8.8 µm, respectively. Calculating the aspect ratio of the three-dimensional structure


6

Page 13 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

and taking the pitch of 3 µm in consideration, a 3.5 times higher effective area is achieved compared to conventional planar devices. One representative NR from the ensemble is depicted in

Figure 1. The SEM image (a) reveals an ordered and homogeneous NR array on top of the GaN template. The STEM image in HAADF contrast (b) obtained in [11-20]-projection presents one representative NR from the ensemble. (c) HAADF image of the core-shell region (position marked in (b) with yellow dashed square) showing the InGaN quantum well in bright contrast and the AlGaN in darker contrast surrounded by GaN matrix. The indium concentration profile (d) evaluated from the HAADF contrast exhibits a medium [In] of (11.5 ± 1.3)% in the middle part of the NR.


7

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 27

the cross-sectional STEM HAADF image in Fig 1(b). For structural analysis the NRs have been investigated in [11-20] zone axis geometry. The n-doped GaN template can be seen in bright contrast in the bottom part. On top of the template, the SiO2 mask appears in darker contrast. The GaN NR core is grown homogeneously out of the mask openings in the [0001]-direction. STEM and TEM investigations reveal a very high crystal quality and the absence of extended structural defects such as basal plane stacking faults and threading dislocations in the whole NR. However, we observe a facet break in the bottom part of the NR. The shell layers were not grown continuously down to the base of the initial core. According to Hartmann et al., the initial high silane flux during the MOVPE process can result in the formation of a SiNx passivation layer which can affect the shell growth in the bottom part (16). Going towards the tip of the nanorod a slight diameter increase from 1080 nm to 1180 nm is observed. The surface tip facets are tilted 62° with respect to the c-direction and are identified as semi-polar {1-101} facets. Figure 1(c) shows the higher magnified HAADF image of the active core-shell region from the middle part of the NR (position marked with the yellow dashed square in Fig. 1(a)). In the HAADF image, the InGaN SQW appears as a brighter layer and AlGaN as a darker one. Both layers run parallel to the {1-100}-plane. The thickness of the AlGaN layer and the InGaN single quantum well (SQW) amounts to tAlGaN = (5.0 ± 0.2) nm and tSQW = (6.3 ± 0.2) nm, respectively. The HAADF image shows that the AlGaN shell does not form a completely closed layer on the core. STEM micrographs recorded along the non-polar side wall exhibit an increasing thickness of the AlGaN layer from the NR base to the very top of the NR. In the upper part, we observe a completely closed AlGaN layer with a thickness of tAlGaN = (13.7 ± 0.2) nm. A closer look at the InGaN SQW in Figure 1(c) reveals slight intensity variations of the HAADF contrast inside the quantum well. As the HAADF contrast is primarily given by the atomic number Z of the analyzed material, the


8

Page 15 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

contrast variations indicate local fluctuations of the indium concentration (24). Figure 1(d) presents the indium concentration profile evaluated from the HAADF contrast (position marked in Fig. 1(c)) by comparing normalized intensities with frozen lattice calculations. The average indium concentration in the quantum well in the middle part of the NR is [In] = (11.5 ± 1.3)%. Local variations of the indium concentrations will be discussed later. To directly visualize the optical properties of the individual core-shell layers, the CL mapping at low temperature (T = 16 K) from the upper part of the NR is depicted in Fig 2. The CL intensity image in Fig 2(a) clearly shows the strongest emission from the InGaN SQW on the non-polar side walls. In contrast, only weak luminescence is detected from the Si-doped GaN core. The initial core is grown under high SiH4 concentrations to enhance the vertical growth. We assume that high Si-doping can result in an increased density of point defects which can lead to a quenching of the luminescence due to non-radiative recombination processes at these defects. The formation of point defects depends strongly on the Fermi level which is determined by the free-carrier concentration (25). In accordance, a reduction of the luminescence intensity in heavily Si-doped GaN microrods has also been observed by Wang et al. (26). Nevertheless, the optical quality of the overgrown shell regions is not affected by the high Si-doping of the core. In Fig. 2(b), the corresponding color-coded CL peak wavelength image is displayed. Beginning at the NR center, we observe the yellow luminescence (YL) from the GaN core, which is frequently observed in planar GaN layers (27). The local CL spectrum (I) presented in Fig. 2(c) at the position of the core exclusively exhibits the YL no luminescence contribution from the near-band edge emission is detected. The GaN YL is commonly associated with the optical transitions of Ga-vacancies (VGa) and oxygen complexes (VGa-ON) (28). Unlike the Si-doped core, the GaN shell layer exhibits dominantly the donor-bound exciton emission at 357 nm indicating a high optical quality of


9

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 27

the overgrown shell (see local CL spectrum III). In addition to the GaN shell luminescence, we observe an emission of the AlGaN shell layer at  = 351 nm on top of the semi-polar tip of the core (in the local CL spectrum II). Going in the direction of the non-polar side walls, the CL

Figure 2. Cross-sectional panchromatic CL intensity image (a) and simultaneously recorded spectrally resolved CL peak wavelength image (b) from the upper part of the NR at T = 16 K. The color-coded CL peak wavelength image clearly resolves the emission of the individual coreshell layers. The most intense emission is detected from the InGaN SQW grown on the non-polar side walls. Local spectra (c) obtained from the different layers (position marked in (a)) present the optical properties of the distinct layers in detail.


10

Page 17 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

mapping is dominated by the InGaN SQW emission. Here, we observe a clear energetic shift of the InGaN emission along the non-polar side wall from  = 443 nm to  = 471 nm from the medium part to the very top of the side wall. The local CL spectra obtained from the InGaN SQW shown in Fig. 2(c) reveals a quite broad full width at half maximum (FWHM) of 190 meV. Furthermore, we observe locally slight variations of the peak wavelength which can be caused by localization effects of generated carriers. The quantum well grown on the {1-101} semi-polar facets emits at about  = 492 nm. The STEM images from the semi-polar tip reveal a reduced SQW thickness compared to the non-polar SQW. Consequently, despite the reduced SQW thickness, we observe a red shift of the SQW emission, which can be accounted by the impact of the polarization fields arising in semi-polar directions. Those inherent polarization fields lead to a band tilting and, consequently, to a local separation of the electron and hole wave function within the quantum wells. In addition, a higher indium incorporation on semi-polar facets can also lead to the observed red shifted SQW emission (29; 30). For a detailed analysis of the InGaN SQW luminescence along the NR, a CL spectrum line scan was performed from the bottom part to the very top along the non-polar side wall. The color-coded CL spectrum line scan in Fig. 3(a) clearly shows a red shift of the InGaN emission together with an increase of the SQW CL intensity towards the upper part. Starting at the NR base, the GaN NBE emission is exclusively detected. As already seen in the STEM image in Fig 1(b), the shell layers were not grown completely on top of the GaN core at the base of the NR. At the position x = 1 µm the InGaN SQW CL appears in the line scan with an emission wavelength of  = 410 nm and exhibits a weak red shift of 50 meV up to a height of x = 5 µm. In the bottom part of the NR the FWHM is typically in the range of 140 meV. Going further towards the upper part of the NR the red shift becomes stronger leading to an energetic shift of 340 meV and finally ends at a


11

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 27

wavelength of about  = 471 nm at the very top. Moreover, the FWHM of the SQW emission is slightly broadened up to 200 meV in comparison to the lower part of the NR. Similar energetic shifts of the quantum well emission have also been observed by Tchernycheva, et al. and Riley et al. for NRs with a lower aspect ratio, and were attributed to compositional gradients along the non-polar side wall (31) (32). Nevertheless, the quantum well thickness can also strongly affect the luminescence properties of the InGaN SQW. Therefore, scanning transmission electron microscopy investigations of the InGaN single quantum well along the {1-100} facet were performed. The quantum well thickness along the NR is depicted in Fig. 3(b). The STEM analysis exhibits a significant increase of the SQW thickness from the base (tSQW = (6.1 ± 0.2) nm) to the top part (tSQW = (13.1 ± 0.2) nm) of the NR. In detail, the STEM images yield a homogeneous quantum well thickness of about tSQW = (6.1 ± 0.2) nm up to a height of x = 5 µm. Above x = 5 µm a clear increase of the SQW thickness is observed. Simultaneously, the quantitative analysis of the HAADF STEM images indicates a higher indium concentration towards the top. Similar to the quantum well thickness, the indium concentration is constant in the bottom part. At x = 5 µm we observe an increasing mean indium content from 11 % to 13 % to the upper part. Both effects, increased quantum well thickness and the compositional gradient lead to the observed red shift of the InGaN SQW emission along the non-polar side wall. Though, the increase of the SQW thickness yields only a smaller contribution to the overall energetic shift. The changing SQW thickness and composition are attributed to different growth rates along the NR side wall. During the growth process a reduced gas diffusion between the nanostructures perpendicular to the template leads to a compositional gradient in vertical direction. Thus, a very high indium incorporation occurs at the very top of the NR and leads to a higher indium content and thicker quantum well. Similar behavior has been found by Wunderer et al. for InGaN QWs grown on


12

Page 19 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

three-dimensional, selectively fabricated semi-polar light emitters (33). The gas diffusion is strongly related to the growth temperature and growth pressure.

Figure 3. The spectrally resolved STEM-CL line scan (a) along the non-polar side wall exhibits a strong red shift of the InGaN SQW emission from the bottom part to the very top of the NR. Detailed analysis (b) of the InGaN SQW (black points) by high resolution HAADF images indicates an increasing SQW thickness accompanied with an increasing indium concentration (red points) along the facet. Both effects are reasons for the observed energetic shift of the InGaN luminescence.


13

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 27

A closer examination of the InGaN SQW emission also reveals CL spots on a spatial scale of several hundreds of nanometers indicating non-continuous gradient along the side wall. The overall wavelength shift is attributed to the reduced gas diffusion between the nanostructures, whereas the step-like behavior could be caused by a surface roughening of the GaN core due to the high silane flow during the initial MOVPE process (16; 34). Lowering the pressure during the shell growth is supposed to reduce the thickness and composition gradient for future devices to achieve a homogeneous SQW emission along the complete vertical structure. Local CL spectra obtained from distinct positions in the CL linescan in Fig 3(a) exhibit also sharp emission lines from the InGaN SQW arising in regions with nanometer-scale size. Due to the absence of extended defects in the shell region, the sharp emissions lines can be caused by nanometer-scale, short range fluctuations of the indium concentration and/or variations of the quantum well thickness which was also observed in planar InGaN quantum wells (35; 36; 37). Recently, Shahmohammadi et al. have observed the formation of indium rich clusters in InGaN quantum wells grown on nanowires in time-resolved CL measurements (38). In particular, indium fluctuation effects within the SQW lead to localization effects of the generated excitons and thus distinct emission lines (39) . Despite indium clustering, localization effects can also appear in random alloy distributions (40). To get insight into the specific indium distribution within the non-polar quantum well on the nanorod and investigate indium concentration fluctuations, quantitative nanometer-scale resolved HAADF imaging was applied. Fig. 4 shows a set of high resolution STEM images in HAADF contrast from InGaN SQW in the upper part (a) and the lower part (c) of the NR. Only slight thickness fluctuations from the mean values tmean = (6.3 ± 0.2) nm (top) and tmean = (13.1 ± 0.2) nm (bottom) are observed (see Fig. 4(c)). Correspondingly, the indium concentration mappings evaluated from the HAADF contrast are presented in Fig. 4(b) and (d), respectively. The mean


14

Page 21 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

indium content from the upper part to the lower part is decreased. The indium concentration mapping from the upper part (Fig. 4(b)) shows that the indium concentration tends to be higher near the SQW/n-GaN shell interface. Furthermore, it exhibits In-rich agglomerates with sizes of only a few nanometers. Locally indium rich regions with [In] > 18 % can be found. Although these concentrations are averaged over the specimen thickness in electron beam direction, which makes it difficult to quantify these concentration fluctuations or make statements about their shape, the analysis of the HAADF intensity allows a statistical assessment of the alloy homogeneity (41). In this regard, the nm-scale fluctuations of the concentration averaged in beam direction, which are significantly larger than expected for a random alloy distribution, are a clear indicator for indium concentration fluctuations on this scale. Accompanying frozen phonon multislice simulations showed, that In rich clusters do indeed cause HAADF STEM intensity distributions that lead to

Figure 4. The high resolution STEM images in HAADF contrast of the InGaN quantum well from the upper part (a) and the bottom part of the NR (c) reveal the thickness increase along the facet. The color-coded composition maps of the indium concentration in (b) and (d) are evaluated from the corresponding HAADF images. The composition maps clearly show local indium rich regions with a size of a few nanometers.


15

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 27

concentration maps as shown in Fig. 4, while fully random alloys yield very homogeneous maps, as is presented in the supporting information for this publication. In comparison to the upper part, we observe a more homogeneous indium distribution within the SQW in the lower part of the NR, as can be seen in Fig. 4(d). Nevertheless, slight fluctuations of the indium concentration can also be found with local [In] up to 15 %. The effects of the indium concentration fluctuations on the luminescence properties of the nonpolar InGaN quantum well are visualized by highly spatially resolved CL mappings depicted in Fig. 5. A detailed STEM image in bright field contrast from the middle part of the NR is presented in Fig. 5 (a). The InGaN SQW appears as darker contrast in the bright field image and is surrounded by the GaN barriers as can be seen in brighter contrast. The quantum well appears apparently thicker in the bright field image in comparison to the previously presented high resolution HAADF images in Fig. 4 due to a slight tilt of the TEM-specimen in the STEM-CL measurements. Fig. 5(b) shows the corresponding spectrally integrated CL mapping acquired at T = 16 K. Here, the most intense emission originates from the InGaN quantum well with a spot-like CL contrast. The spatially averaged CL spectrum from the depicted region exhibits the InGaN SQW CL at  = 430 nm with a FWHM of 250 meV. Probing different spectral positions of the quantum well luminescence, a set of spectrally resolved CL intensity mappings (CLI) depicted in Fig. 5 (d) – (g) identifies the spatial distribution of the InGaN emission. The monochromatic mapping from  = 405 nm to  = 425 nm on the high energetic side of the InGaN emission presented in Fig. 5 (d) exhibits spottily distributed SQW emission. In comparison, the InGaN emission between  = 425 nm and 439 nm stems from spatially different locations. When going to longer wavelengths different luminescence spots are addressed. As an example the CL mapping from  = 440 nm – 455 nm (Fig. 5(f)) reveals only one distinct emission spot, whereas the mapping from  = 456 nm


16

Page 23 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

– 476 nm shows two spatially separated CL spots. Comprehensive studies of m-plane InGaN/GaN QWs from Schulz et al. reported already on localization effects in random alloy distributions (40).

Figure 5. (a) Bright-field image of the InGaN SQW surrounded by GaN barriers and the integral intensity image (b) at T = 16 K. The TEM-specimen is slightly tilted in the STEM-CL measurements. The corresponding spatially integrated CL spectrum is shown in (c). Monochromatic mappings (d) – (g) reveal a spot-like CL contrast of the InGaN emission due to nanometer-scale potential fluctuations.


17

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 27

In contrast, the observed localization effects in our case are primarily attributed to the formation of indium rich regions as evidenced by HAADF imaging. In particular, the formation of indium rich agglomerates within the quantum well is suggested to lead to band gap fluctuations, where the excitons are localized at potential minima. Quantum well thickness fluctuations can also contribute to the localization effects. Nevertheless, monolayer thickness fluctuations lead only to small energetic shifts and cannot explain the strong emission variations, exclusively. These band gap fluctuations caused by indium clustering effects can form efficient localization centers for generated carriers (42; 43). Thus, the generated carriers diffuse to the indium rich clusters and lead to the observed spot-like CL contrast which is in complete agreement with the STEM HAADF results. The nanoscopic potential fluctuations lead to an inhomogeneous broadening of the overall SQW emission. In summary, we have demonstrated a unique approach for a direct one-by-one correlation of the optical, structural, and compositional properties of InGaN/GaN core-shell NRs. The core-shell nanostructures exhibit an extraordinary low density of extended defects. Highly spatially and spectrally resolved CL mappings at low temperature reveal the most intense luminescence from the InGaN single quantum well on the non-polar side walls. Due to an increased quantum well thickness and indium concentration, an energetic red shift from 410 nm to 471 nm of the quantum well luminescence along the side facets in c-direction is observed. The changing quantum well thickness and indium content is attributed to a gas composition gradient during the MOVPE process along the NR. Indium rich regions with a size of several ten nanometers are formed within the quantum well. These short range indium variations result in strong potential fluctuations of the InGaN quantum well as evidenced by nanometer-scale resolved CL mappings.


18

Page 25 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

Supporting Information Available To demonstrate the capability of HAADF STEM to detect indium fluctuations within the InGaN quantum well, the experimental HAADF intensity maps are compared in detail to frozen phonon multislice simulations of a model system with the same specimen thickness and lateral extent in the supporting information. This material is available free of charge via the Internet at http://pubs.acs.org.

ACKNOWLEDGMENT We gratefully acknowledge the German Research Foundation (DFG) for financial support within the Research Instrumentation Program INST 272/148-1, the Collaborative Research Center SFB 787 "Semiconductor Nanophotonics: Materials, Models, Devices", and the "Materials World Network" Program. Furthermore, the work was supported by the DFG under Contracts RO2057/42 and MU3660/1-1. The sample fabrication was partly supported by the European Community within the FP7 project GECCO (contract number 280694). Many thanks to Silke Petzold for her work regarding the very difficult TEM preparations.

Note: The authors declare no competing financial interest


19

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 27

REFERENCES (1) (2) (3) (4)

(5) (6) (7) (8) (9) (10)

(11) (12) (13)

(14)

(15) (16)

(17) (18)

(19) (20)

Kikuchi, A.; Kawai, M.; Tada, M.; Kishino, K. Jpn. J. Appl. Phys 2004, 43 (12A), 1524 –1526. Waag, A.; Li, S. J. Appl. Phys. 2012, 111, 071101. Dai, X.; Messanvi, A.; Zhang, H.; Durand, C.; Eymery, J.; Bougerol, C.; Julien, F. H.; Tchernycheva, M. Nano Lett. 2015, 15, 6958−6964. Mandl, M.; Wang, X.; Schimpke, T.; Kölper, C.; Binder, M.; Ledig, J.; Waag, A.; Kong, X.; Trampert, A.; Bertram, F.; Christen, J.; Barbagini, F.; Calleja, E.; Strassburg, M. Physica Status Solidi (RRL) - Rapid Research Letters 2013, 7 (10), 800 - 814. Hersee, S. D.; Rishinaramangalam, A. K.; Fairchild, M. N.; Lei, Z.; Petros, V. J. Mater. Res 2011., 26, 2293-2298. Colby, R.; Liang, Z.; Wildeson, I. H.; Ewoldt, D. A.; Sands, T. D.; García, R. E.; Stach E. A. Nano Lett. 2010, 10, 1568–1573. Romanov, A. E.; Baker, T. J.; Nakamura, S.; Speck, J. S. J. Appl. Phys. 2006; 100, 023522. Waltereit, P.; Brandt, O.; Trampert, A.; Grahn, H. T.; Menniger, J.; Ramsteiner, M.; Reiche, M.; Ploog, K. H. Nature 2000, 406, 865 – 868. Kölper, C.; Sabathil, M.; Römer, F.; Mandl, M.; Strassburg, M.; Witzigmann, B. Physica Status Solidi A 2012, 209, 2304–2312. Laubsch, A.; Sabathil, M.; Bergbauer, W.; Strassburg, M.; Lugauer, H.; Peter, M.; Lutgen, S.; Linder, N.; Streubel, K.; Hader, J.: Moloney, J. V.; Pasenow, B.; Koch, S. W. Physica Status Solidi C 2009, 6, 913 - 916 Schmidt, G.; Müller, M.; Veit, P.; Bertram, F.; Christen, J.; Glauser, M.; Carlin, J. F.; Cosendey, G.; Butté, R.; Grandjean, N. Appl. Phys. Lett. 2014, 105, 032101. Zhou, X.; Lu, M.; Lu, Y.; Jones, E. J.; Gwo, S.; Gradecak, S. ACS Nano 2015, 9, 2868−2875. Griffiths, J. T.; Zhang, S.; Rouet-Leduc, B.; Fu, W. Y.; Bao, A.; Zhu D.; Wallis, D. J.; Howkins, A.; Boyd, I.; Stowe, D.; Kappers, M. J.; Humphreys, C. J.; Oliver, R. A. Nano Lett. 2015, 15, 7639– 7643. Rosenauer, A.; Mehrtens, T.; Müller, K.; Gries K.; Schowalter, M.; Venkata Satyam, P.; Bley, S.; Tessarek, C.; Hommel, D.; Sebald, K.; Seyfried, M.; Gutowski, J.; Avramescu, A.; Engl, K.; Lutgen, S. Ultramicroscopy 2011, 111, 1316–1327. Rosenauer, A.; Gries K.; Müller, K.; Pretoriusa, A.; Schowalter, M.; Avramescu, A.; Engl, K.; Lutgen, S. Ultramicroscopy 2009, 109, 1171–1182. Hartmann, J.; Wang, X.; Schuhmann, H.; Dziony, W.; Caccamo, L.; Ledig, J.; Mohajerani, M. S.; Schimpke, T.; Bähr, M.; Lilienkamp, G.; Daum, W.; Seibt, M.; Straßburg, M.; Wehmann, H. H.; Waag, A. Physica Status Solidi A 2015, 12, 2830–2836 Mohajerani, M. S.; Khachadorian, S.;Schimpke, T.; Nenstiel, C.; Hartmann, J; Ledig, J., Avramescu A.; Strassburg M.; Hoffmann, A; Waag, A., Appl. Phys. Lett. 2016, 108, 091112. Schimpke, T.; Mandl, M.; Stoll, I.; Pohl-Klein, B.; Bichler, D.; Zwaschka, F.; Strube-Knyrim, J.; Huckenbeck, B.; Max, B.; Müller, M.; Veit, P.; Bertram, F.; Christen J.; Hartmann, J.; Waag, A.; Lugauer, H. J.: Strassburg, M. Physica Status Solidi A 2016, 1- 8, published online: DOI 10.1002/pssa.201532904. Bell, A.; Srinivasan, S.; Plumlee, C.; Omiya, H.; Ponce, F. A.; Christen, J.; Tanaka, S.; Fujioka, A.; Nakagawa, Y. J. Appl. Phys. 2004, 95, 4670. Müller, M.; Schmidt, G.; Metzner, S.; Veit, P.; Bertram, F.; Leute, R. A. R.; Heinz, D.; Wang, J.; Meisch, T.; Scholz, F.; Christen, J. Physica Status Solidi B 2015, 253, 112 – 117.


20

Page 27 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

(21) (22) (23)

(24) (25) (26) (27) (28) (29) (30) (31) (32) (33)

(34) (35) (36) (37) (38) (39) (40)

(41)

(42) (43)

Urban, A.; Müller, M.; Karbaum, C.; Schmidt, G.; Veit, P.; Malindretos, J.; Bertram, F.; Christen, J.; Rizzi A. Nano Lett. 2015, 15, 5105−5109. Krause, F. F.; Schowalter M.; Grieb, T.; Müller-Caspary, K.; Mehrtens, T.; Rosenauer, A. Ultramicroscopy 2016, 161, 146–160. Schowalter, M.; Stoffers, I.; Krause, F. F.; Mehrtens, T.; Müller, K.; Fandrich, M.; Aschenbrenner, T.; Hommel, D; Rosenauer A. Microscopy and Microanalysis 2014, 20, 1463 – 1470. Hartel, P.; Rose, H.; Dinges, C. Ultramicroscopy 1996, 63, 93–114. Van de Walle, C. G.; Neugebauer, J.; J. Appl. Phys. 2004, 95, 3851. Wang, X.; Hartmann, J.; Mandl, M.; Mohajerani M. S.; Wehmann H. H.;Strassburg, M; and Waag, A. J. Appl. Phys. 2014, 115, 163104. Ponce, F. A.; Bour, D. P.; Götz, W.; Wright, P. J. Appl. Phys. Lett. 1996, 68, 57–59. Reshchikov M. A.; Morkoc, H. J. Appl. Phys. 2005, 97, 061301. Northrup, J. E. Appl. Phys. Lett. 2009, 95. 133107. Wernicke, T.; Schade, L.; Netzel, C.; Rass, J.; Hoffmann, V.; Ploch, S.; Knauer, A.; Weyers, M.; Schwarz, U.; Kneissl, M. Semicond. Sci. Technol. 2012, 27, 024014. Tchernycheva, M.; Lavenus, P.; Zhang, H.; Babichev, A. V.; Jacopin, G.; Shahmohammadi, M.; Julien, F. H.; Ciechonski, R.; Vescovi, G.; Kryliouk, O. Nano Lett. 2015, 14, 2456−2465. Riley, J. R.; Padalkar,S.; Li, Q.; Lu, P.; Koleske, D. D.; Wierer, J. J.; Wang,G. T.; Lauhon, L. J. Nano Lett. 2013, 13, 4317−4325 Wunderer, T.; Feneberg, M.; Lipski, F.; Wang, J.; Leute, R. A. R.; Schwaiger, S.; Thonke, K.; Chuvilin, A.; Kaiser, U.; Metzner, S.; Bertram, F.; Christen, J.; Beirne, G. J.; Jetter, M.; Michler, P.; Schade, L.; Vierheilig, C.; Schwarz, U. T.; Dräger, A. D.; Hangleiter, A.; Scholz, F. Physica Status Solidi B 2010, 248, 549 – 560. Markurt, T.; Lymperakis, L.; Neugebauer, J.; Drechsel, P.; Stauss, P.; Schulz, T.; Remmele, T.; Grillo, V; Rotunno, E.; Albrecht, M., Phys. Rev. Lett. 2013, 110, 036103. Kisielowski, C.; Liliental-Weber, Z.; and Nakamura, S. Jpn. J. Appl. Phys. 1997, 36, 6932 – 6936. Bartel, T.; Jinschek, J. R.; Freitag, B.; Specht, P.; Kisielowski, C. Phys. Stat. Sol. A 2006, 203, 167– 175. Gerthsen, D.; Neubauer, B.; Rosenauer, A.; Stephan, T.; Kalt, H.; Schön, O.; Heuken, M. Appl. Phys. Lett. 2001, 79, 2552. Shahmohammadi, M; Ganière, J. D.; Zhang, H.; Ciechonski, R.; Vescovi, G.; Kryliouk, O.; Tchernycheva, M.; Jacopin, G. Nano Lett. 2015, published online. Chichibu, S.; Sota, T.; Wada, K.; Nakamura, S. J. Vac. Sci. Technol. B 1998, 16, 2204. Schulz, S; Tanner, D. P.; O’Reilly, E. P.; Caro, M. A.; Martin, T. L.; Bagot, P. A. J.; Moody, M. P.; Tang, F.; Griffiths, J. T.; Oehler, F.; Kappers, M. J.; Oliver, R. A.; Humphreys, C. J.; Sutherland, D.; Davies, M. J.; Dawson P. Phys. Rev. B. 2015, 92, 235419 Krause, F. F.; Ahl, J.P.; Tytko, D.; Choi, P.; Egoavil, R.; Schowalter, M.; Mehrtens, T.; MüllerCaspary, K.; Verbeeck, J.; Raabe, D.; Hertkorn, J.; Engl, K.; Rosenauer, A. Ultramicroscopy 2015, 156, 29-36. Schömig, H.; Halm, S.; Forchel, A.; Bacher, G.; Off, J.; Scholz, F. Phys. Rev. Lett. 2004, 92, 106802. Wu, Y. R.; Shivaraman, R.; Wang, K.-C.; Speck, J. S. Appl. Phys. Lett. 2012, 101, 083505.


21

GaN CoreâShell ... - ACS Publications

Recommend Documents

GaN CoreâShell ... - ACS Publications