Optical properties of self assembled GaN polarity inversion domain boundary

(1)

Optical properties of self assembled GaN polarity inversion domain boundary

M.-C. Liu, Y.-J. Cheng, J.-R. Chang, S.-C. Hsu, and C.-Y. Chang

Citation: Applied Physics Letters 99, 021103 (2011); doi: 10.1063/1.3610449

View online: http://dx.doi.org/10.1063/1.3610449

View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/99/2?ver=pdfcov Published by the AIP Publishing

Articles you may be interested in

Germanium doping of self-assembled GaN nanowires grown by plasma-assisted molecular beam epitaxy J. Appl. Phys. 114, 103505 (2013); 10.1063/1.4820264

Self-confined GaN heterophased quantum wells

Appl. Phys. Lett. 96, 222105 (2010); 10.1063/1.3442497 Inversion domains and pinholes in GaN grown over Si(111) Appl. Phys. Lett. 82, 4471 (2003); 10.1063/1.1584072

Termination mechanism of inversion domains by stacking faults in GaN J. Appl. Phys. 93, 3264 (2003); 10.1063/1.1544427

Evidence for multiple atomic structure for the {1010} inversion domain boundaries in GaN layers Appl. Phys. Lett. 74, 947 (1999); 10.1063/1.123418

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: 140.113.38.11 On: Wed, 30 Apr 2014 09:01:45

(2)

Optical properties of self assembled GaN polarity inversion domain

boundary

M.-C. Liu,1Y.-J. Cheng,1,a)J.-R. Chang,2S.-C. Hsu,3and C.-Y. Chang2

1

Research Center for Applied Sciences, Academia Sinica, Taipei 11529, Taiwan

2

Institute of Electronics, National Chiao Tung University, 1001 Ta Hsueh Rd., Hsinchu 300, Taiwan

3

Department of Chemical and Materials Engineering, Tamkang University, No.151, Yingzhuan Rd., Danshui Dist., Taipei, Taiwan

(Received 12 November 2010; accepted 23 June 2011; published online 13 July 2011)

We report the fabrication of GaN lateral polarity inversion heterostructure with self assembled crystalline inversion domain boundaries (IDBs). The sample was fabricated by two step molecular-beam epitaxy (MBE) with microlithography patterning in between to define IDBs. Despite the use of circular pattern, hexagonal crystalline IDBs were self assembled from the circular pattern during the second MBE growth. Both cathodoluminescent (CL) and photoluminescent (PL) measurements show a significant enhanced emission at IDBs and in particular at hexagonal corners. The ability to fabricate self assembled crystalline IDBs and its enhanced emission property can be useful in optoelectronic applications.VC _{2011 American Institute of Physics. [doi:}_{10.1063/1.3610449}_]

III-Nitride semiconductors, in particular GaN, have attracted great research interests in past few years due to their promising applications for UV to blue optoelectronic devi-ces.1One of the important properties of wurtzite GaN is its strong spontaneous and piezoelectric polarization, which can induce surface charges and create large internal electric field in the film. This electric field can strongly affect the electri-cal and optielectri-cal properties of GaN based devices.2,3 It can reduce the electron-hole wave function overlap in quantum well devices and have adverse effect on its light emitting effi-ciency. The direction of internal field depends on crystal c-axis orientation, which is not symmetric. Conventionally, the surface in the positive and negative c-axis direction is labeled as Ga- (0001) and N-polar (000-1) surface, respectively.

Recently, there are interests in studying the physical property of inversion domain boundary between Ga- and N-polar regions. The inversion domain boundary (IDB) has been found to exist at microscopic scale in Ga-polar GaN thin film grown by molecular-beam epitaxy (MBE).4 The IDB can also be created by intentionally growing lateral po-larity heterostructures, where patterned Ga- and N-polar regions are laterally grown on the same substrate to form IDBs.5–7The ability to fabricate controlled polarity pattern can open up additional device design dimension and applica-tions. The IDB has been studied theoretically,8,9and experi-mentally imaged by high-resolution transmission electron microscope,4,5 and piezoelectric force microscope.10,11 It was theoretically predicted that IDBs in wurtzite GaN would not have electronic states in the band gap, implying that they would not affect photoluminescent (PL) efficiency.8The op-tical property of IDB has been investigated by high-resolu-tion spatially resolved PL measurement.12The measurement was done at 10 K low temperature and significantly brighter emission was reported at IDBs. However, the observed emis-sion from IDB was neither spectrally nor spatially uniform. The nonuniformity could be due to the coexistence of

differ-ent crystalline planes in the intdiffer-entionally patterned IDB, which was suggested possibly having mixed crystalline planes {10-10} and {11-20}. The fabricated IDB was a straight line in tens of lm and not oriented along any specific crystal plane. That could be the reason for leading to mixed crystalline planes in IDBs. The observed PL peaks have some variations and are about 30–40 meV lower than the bulk Ga-polar emission. A theoretical calculation however predicts a zero shift in emission peak at IDB.9The discrep-ancy may come from the mixed crystal planes in IDBs and possible defects associated with them. The ability to fabri-cate crystalline IDB pays the way for better physical prop-erty study. It may also open up new applications due to its enhanced emission property.

Here we report the fabrication of self assembled IDB along {10-10} crystalline plane and the observation of enhanced light emission at IDB by cathodoluminescent (CL) and PL measurement. The crystalline IDBs were fabricated by two step rf-plasma-enhanced MBE, where microlithogra-phy patterning process was used in between two growth steps to define lateral polarity heterostructure boundary. Circular patterns were used in microlithography patterning process. However, hexagonal crystalline IDBs along {10-10} crystal-line planes were self assembled from the original circular pattern after the second MBE re-growth. Both CL and PL measurements show an enhanced light emission at IDBs and a zero shift in emission peak.

It is known that GaN grown on c-plane sapphire by MBE normally has an N-polar surface. It has also been dem-onstrated that the polarity of GaN can be switched to Ga-po-lar surface by pre-growing a high temperature AlN buffer layer on c-plane sapphire surface. The fabrication steps, as shown in Figs. 1(a)–1(d), use these two growth techniques and microlithography patterning to grow patterned lateral polarity heterostructure. First, a high temperature (930 C) thin AlN buffer layer (30 nm) was grown on top of a (0001) sapphire substrate followed by a thin layer of GaN growth (810 C, 50 nm) [Fig. 1(a)]. Ga-polar GaN was grown on AlN buffer layer, which was confirmed by in situ (2 2)

a)_{Author to whom correspondence should be addressed. Electronic mail:} [email protected].

0003-6951/2011/99(2)/021103/3/$30.00 99, 021103-1 VC2011 American Institute of Physics APPLIED PHYSICS LETTERS 99, 021103 (2011)

(3)

reflection high energy electron diffraction (RHEED) pattern.13–15 Photoresist (PR) was then spun on the sample and patterned using microlithography to create 3 lm circular mesa masks [Fig.1(b)]. The sample was etched in an induc-tively coupled plasma (ICP) etcher using Ar/Cl2 gas until the unmasked GaN regions were etched down to expose sap-phire substrate surface. The PR mask was then removed, leaving 3 lm circular Ga-polar GaN mesas on sapphire sub-strate [Fig.1(c)]. The sample was subsequently reintroduced into the MBE system to grow 1 lm (810 C) GaN layer. Since AlN buffer layer was removed by the etching process, the re-grown GaN on the exposed sapphire surface would be N-polar.13 The same second step growth parameter was applied to a separate blank sapphire substrate to verify that it did grow N-polar GaN, which was confirmed byin situ (3 3) RHEED pattern. The GaN grown on the circular Ga-polar GaN mesa, which worked as a seeding layer, will still follow the same Ga polarity direction. As a result, we obtained a patterned lateral polarity heterostructure with pre-defined IDBs.

A scanning electron microscope (SEM) plane view of the sample is shown in Fig.2(a). It is interesting to observe the self assembly of hexagonal shapes after the second MBE growth even though circular pattern was originally used to define polarity inversion boundary. From the growth process design, the polarity inside hexagon is Ga-polar, while the surrounding area is N-polar. The IDBs were self assembled at {10-10} hexagonal crystal planes. There have been reports showing the growth speeds of different crystalline planes in selective area lateral overgrowth on patterned oxide mask with circular openings.16 The observed growth rate of (10-11) plane is much slower than that of (0001) plane. As a result, the selective lateral overgrowth often leads to a hex-agonal boundary in the in-plane direction despite of the orig-inal circular pattern. In our experiment, it is a 2-D thin film growth instead of a 3-D lateral overgrowth. As a result, IDBs are self-assembled at {10-10} hexagonal planes. The SEM was switched to cathodoluminescent detection under the same magnification condition. The CL image is shown in Fig.2(b). The focused e-beam spot size is less than 10 nm, which defines the CL imaging spatial resolution. The catho-doluminescent intensity is significantly stronger at IDBs and maximum at hexagonal corner. The luminescent intensity at

IDB gradually decreases with position moving away from corner along the boundary. A SEM cross section and CL images are shown in Figs.2. It shows that the hexagonal Ga-polar region is about 80 nm higher than the surrounding N-polar surface, which is consistent with the thickness of Ga-polar seeding layer. The cross section profile shows a slow inclined side wall of about 160 nm at IDBs. The cross sec-tion CL does not show visible enhancement as the plane view CL does, neither at the IDBs nor the IDB mesa edge. This might indicate that the enhancement is a shallow sur-face effect on c-plane and requires further investigation. A cross section TEM picture is shown in Fig. 2(e), which shows the IDB propagating from the edge of patterned AlN buffer layer to the top surface.

The plane view CL intensity spectra taken at the hexag-onal corner (A), middle of hexaghexag-onal boundary (B), center of hexagonal Ga-polar region (C), and N-polar region (D) are shown in Fig. 3(a). The intensity magnitudes of these four points in descending order follow the same sequence.

FIG. 1. (Color online) (a)–(d) Processing flow for fabricating patterned po-larity IDB. (a) Grow Ga-polar GaN layer. (b) Photoresist patterning. (c) ICP etching to remove unmasked region followed by PR removal (d). The sec-ond growth to form polarity heterostructure.

FIG. 2. (a) SEM plane view image of the sample surface. (b) CL image at the same magnification showing the enhanced luminecent intensity at IDBs. (c) SEM cross section image. (d) CL image. (e) TEM cross section view.

FIG. 3. (Color online) (a) CL intensity spectra at various sample locations. (b) PL intensity spectra at various sample locations.

021103-2 Liu et al. Appl. Phys. Lett. 99, 021103 (2011)

(4)

The full width at half maximum (FWHM) linewidth is about 14 nm for all spectra. The intensity at Ga-polar region is sig-nificantly higher than that at N-polar region. Ga-polar GaN is known to have better material quality than N-polar GaN, which explains the observed higher luminescent intensity. The intensity at IDB, in particular at the hexagonal corner, is much higher than both Ga- and N-polar region. The peak wavelength of the enhanced luminescence at IDB is the same as those for Ga- and N-polar GaN, consistent with the theoretical prediction.12 This is in contrast to the 30 meV redshift reported from IDB with mixed crystal planes where the redshift could be due to shallow defect traps created among the mixed crystal planes. The PL spectra of the sam-ple were also measured using 325 nm He-Cd laser pumping source. The laser was focused on the sample by a 39 UV objective with a focused spot size of 0.8 um. The PL spectra at similar locations are shown in Fig.3(b). It reconfirms the enhanced light emission at IDBs. The relative intensity mag-nitudes at different locations are similar to the CL case except that the intensity at corner is not as strong. This is likely due to the area average effect of the much larger exci-tation source spot size in PL measurement and implies that the great enhancement is localized to a very small region. A theoretical model based onab initio density functional calcu-lation has shown that the electronic potential tends to trap electrons and holes in the IDB vicinity.9This could provide better e-h recombination efficiency and explain the enhanced emission. However, the inclined surfaces at IDBs, as shown in Fig.2(c), may provide better light extraction and contrib-ute to the observed enhanced emission as well.

To estimate the light extraction effect, we analyze the escape light cones using a simplified ray model as shown in Fig.4, where point source locations are shown in dotted line. The surface inclined angle D is about 26.5 from SEM cross section. The total internal reflection angle h is 24from GaN refractive index of 2.45. Each point source has two rays defining the escape light cone (ELC). For all the points to the left of point A, the ELC is the nominal 2h value. For points between A and B, ELC has its maximum value 2hþD. For points between B and C, it is back to 2h. Going further from point C to D, ELC decreases from 2h to h. For points between D and E, ELC is at minimum value of 2hD. ELC then increases from h at point E to 2h at point F. From the

analy-sis, we see that the upper bound of light extraction enhance-ment is (2hþD)=2h ¼ 55% occurring for points between A and B. This is still smaller than the observed 75% CL emis-sion enhancement at the middle of IDB, compared to Ga-po-lar region. This indicates that even though light extraction makes significant contribution to the observed enhanced emission, the material property at IDBs also contribute to a significant portion of enhancement. The much brighter emis-sion at IDB corners could be due to the sharp turning angle, which may increase light extraction by scattering the wave-guide mode wave-guided along the IDB into free space.

In summary, the fabrication of polarity inversion GaN heterostructure with self assembled {10-10} crystal plane IDBs is demonstrated. The fabrication process involves two step rf-enhanced MBE growths and lithography patterning in between two growths to define IDBs. Despite the use of cir-cular patterns for defining IDBs, hexagonal crystalline IDBs are self assembled from the original circular patterns in the second epitaxial growth. CL and PL measurements show that luminescent property is significantly enhanced at IDBs. The ability to fabricate crystalline IDB and the demonstra-tion of its enhanced luminescent property offers an addi-tional device design dimension and could be of useful in GaN based light emitting devices.

This work was financially supported by Sinica Nano-program and in part by the National Science Council of Republic of China (ROC) Taiwan under contract NSC97-2112-M-001-027-MY3 and Sinica Nano Program.

1

T. Mukai, K. Takekawa, and S. Nakamura,Jpn. J. Appl. Phys.Part 2 37, L839 (1998).

2_{O. Ambacher, J. Smart, J. R. Shealy, N. G. Weimann, K. Chu, M. Murphy,} R. Dimitrov, L. Wittmer, M. Stutzmann, W. Rieger, and J. Hilsenbeck,J. Appl. Phys.85, 3222 (1999).

3

M. Stutzmann, O. Ambacher, M. Eickhoff, U. Karrer, A. Lima Pimenta, R. Neuberger, J. Schalwig, R. Dimitrov, P. Schuck, and R. Grober,Phys. Status Solidi B288, 505 (2001).

4

C. Iwamoto, X. Q. Shen, H. Okumura, H. Matuhata, and Y. Ikuhara,Appl. Phys. Lett.79, 3941 (2001).

5_{F. Liu, R. Collazo, S. Mita, Z. Sitar, G. Duscher, and S. J. Pennycook,} Appl. Phys. Lett.91, 203115 (2007).

6

S. Pezzagna, P. Venne´gue`s, N. Grandjean, A. D. Wieck, and J. Massiesc, Appl. Phys. Lett.87, 062106 (2005).

7_{A. Chowdhury, H. M. Ng, M. Bhardwaj, and N. G. Weimann,}_{Appl. Phys.} Lett.83, 1077 (2003).

8

J. E. Northrup, J. Neugebauer, and L. T. Romano,Phys. Rev. Lett.77, 103 (1996).

9_{V. Fiorentini,}_{Appl. Phys. Lett.}_{82, 1182 (2003).}

10_{B. J. Rodriguez, A. Gruverman, A. I. Kingon, R. J. Nemanich, and O.} Ambacher,Appl. Phys. Lett.80, 4166 (2002).

11

R. Katayama, Y. Kuge, K. Onabe, T. Matsushita, and T. Kondo, Appl. Phys. Lett.89, 231910 (2006).

12_{P. J. Schuck, M. D. Mason, R. D. Grober, O. Ambacher, A. P. Lima, C.} Miskys, R. Dimitrov, and M. Stutzmann,Appl. Phys. Lett.79, 952 (2001). 13

X. Q. Shen, T. Ide, S. H. Cho, M. Shimizu, S. Hara, and H. Okumura, Appl. Phys. Lett.77, 4013 (2000).

14_{D. Huang, P. Visconti, K. M. Jones, M. A. Reshchikov, F. Yun, A. A.} Baski, T. King, and H. Morkoc,Appl. Phys. Lett.78, 4145 (2001). 15

A. R. Smith, R. M. Feenstra, D. W. Greve, M.-S. Shin, M. Skowronski, J. Neugebauer, and J. E. Northrup,Appl. Phys. Lett.72, 2114 (2001). 16_{B. Beaumont, S. Haffouz, and P. Gibart,} _{Appl. Phys. Lett.}

72, 1227 (1998).

FIG. 4. (Color online) Escape light cone analysis around IDB.

021103-3 Liu et al. Appl. Phys. Lett. 99, 021103 (2011)