Site-controlled growth of GaN nanorod arrays by magnetron sputter epitaxy

(1)

Site-controlled growth of GaN nanorod arrays by

magnetron sputter epitaxy

Elena Alexandra Serban, Justinas Palisaitis, Per Ola Åke Persson, Lars Hultman, Jens

Birch and Ching-Lien Hsiao

The self-archived postprint version of this journal article is available at Linköping

University Institutional Repository (DiVA):

http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-147648

N.B.: When citing this work, cite the original publication.

Serban, E. A., Palisaitis, J., Persson, P. O. Å., Hultman, L., Birch, J., Hsiao, C., (2018), Site-controlled

growth of GaN nanorod arrays by magnetron sputter epitaxy, Thin Solid Films.

https://doi.org/10.1016/j.tsf.2018.01.050

Original publication available at:

https://doi.org/10.1016/j.tsf.2018.01.050

Copyright: Elsevier

(2)

Contents lists available atScienceDirect

Thin Solid Films

journal homepage:www.elsevier.com/locate/tsf

Site-controlled growth of GaN nanorod arrays by magnetron sputter epitaxy

Elena Alexandra Serban, Justinas Palisaitis, Per Ola Åke Persson, Lars Hultman, Jens Birch,

Ching-Lien Hsiao

⁎

Thin Film Physics Division, Department of Physics, Chemistry and Biology (IFM), Linköping University, SE-58183 Linköping, Sweden

A R T I C L E I N F O

Keywords: Gallium nitride Magnetron sputter epitaxy Selective-area growth Nanorods

Lithography Focused ion beam Nanosphere

A B S T R A C T

Catalyst-free GaN nanorod regular arrays have been realized by reactive magnetron sputter epitaxy. Two na-nolithographic methods, nanosphere lithography (NSL) and focused ion beam lithography (FIBL), were applied to pattern Si substrates with TiNxmasks. The growth temperature was optimized for achieving selectivity and

well-faceted nanorods grown onto the NSL-patterned substrates. With increasing temperature from 875 to 985 °C, we observe different growth behaviors and associate them with selective insensitive, diffusion-domi-nated, and desorption-dominated zones. To further achieve site-specific and diameter control, these growth parameters were transferred onto FIBL-patterned substrates. Further investigation into the FIBL process through tailoring of milling current and time in combination with varying nanorod growth temperature, suggests that minimization of mask and substrate damage is the key to attain uniform, well-defined, single, and straight nanorods. Destruction of the mask results in selective area growth failure, while damage of the substrate surface promotes inclined nanorods grown into the openings, owning to random oriented nucleation.

1. Introduction

Gallium nitride (GaN) is a direct bandgap semiconductor, with technologically relevant properties such as: high thermal stability (when compared to common semiconductors), high electric break-downﬁeld, and chemical inertness [1]. Due to this, it is already a well-established semiconductor used in solid-state lighting devices, [2,3] and in high-temperature as well as high-power operation electronic applications [4–6].

GaN nanorods (NRs) combine the intrinsic properties of the GaN material with distinctive features induced by the reduced dimension and the NR geometry like: increased photon extraction eﬃciency, high crystal quality and strain relaxation [7,8]. The NRs fabrication pro-cesses include: self-assembled (SA), [9] catalyst-induced, [10] and se-lective-area growth (SAG) [11]. SA processes are characterised by a random growth of NRs leading to non-uniform properties and hindering device processing [12,13]. This can be overcome by SAG of NRs, using pre-patterned substrates by lithographic methods such as: photo-lithography, [14] electron-beam lithography, [15] focused ion beam lithography (FIBL), [16] nanoimprint lithography [17] or nanosphere lithography (NSL) [11]. The usage of mask layers, limits the growth to speciﬁc areas, resulting in ordered NR arrays with controllable sizes, shapes, positions, and densities.

The dominating techniques used for the epitaxial growth of GaN

NRs are metal-organic chemical vapor deposition [18–20] and mole-cular-beam epitaxy [21–23]. However, magnetron sputter epitaxy (MSE) of GaN and related alloys, has gained momentum in recent years [24–26]. MSE is a versatile, industrially-mature method, and a series of high-quality structures which include SA [27] and SAG [11] NRs, and thinﬁlms [28] have been reported.

In this paper we demonstrate the growth of GaN NR arrays on Si substrates pre-patterned by two techniques: NSL and FIBL, employing TiNxmask layers. GaN NRs were grown by reactive MSE onto these patterned substrates. A temperature-dependent series was grown onto NSL-patterned substrates in order to attain the optimum growth con-ditions to achieve selectivity and well-defined morphology. NSL is a simple, fast and cheap method that offers the possibility of obtaining size-uniform NRs. However, it does not offer the control on the size, position or density. To circumvent this, we transferred the growth parameters onto FIBL-patterned substrates. FIBL enables accurate con-trol of position and size of the openings that are prepared in the mask layer, enabling the growth of uniform, well-defined NR arrays. For this, the optimum patterning conditions were tested for obtaining well-de-fined openings in the TiNxmask and for minimizing substrate damage. FIBL optimization included tailoring of the milling current (2–50 pA) and milling time (5–50 s). Low milling currents are necessary to avoid damage to the whole mask layer. A minimum ion beam induced sub-strate damage is achieved at low currents of 2 pA and short milling

https://doi.org/10.1016/j.tsf.2018.01.050

Received 18 November 2017; Received in revised form 22 January 2018; Accepted 24 January 2018

⁎_{Corresponding author.}

E-mail address:hcl@ifm.liu.se(C.-L. Hsiao).

(3)

times. Longer milling times induce extensive substrate damage as ob-served by scanning transmission electron microscopy (STEM) and en-ergy-dispersive x-ray spectroscopy (EDX). The resulted rough substrate surface conducts to the growth of multiple, tilted NRs inside one opening. Growth temperature optimization was also performed. At lower growth temperatures (950 °C) nanostructures resulted from the coalescence of multiple, tilted, and irregular NRs are observed. The tilting of the NRs is reduced when increasing the growth temperature to 980 °C resulting in hexagonal, mostly single, straight NRs with uniform sizes and controlled position.

2. Experimental details

The growth of GaN NRs was performed by direct current-MSE in an ultrahigh vacuum chamber with a base pressure of 1.33 × 10−6Pa. A liquid Ga (99.99999% pure) target, placed in a stainless-steel crucible is used as a sputtering target. Details may be found elsewhere [29]. The NRs deposition on the pre-patterned Si(001) substrates was optimized by changing the growth temperature in the interval 875–985 °C. The working pressure was kept constant at 2.67 Pa N2.

Two patterning methods were employed: NSL and FIBL. In both cases a TiNxlayer was employed as a mask layer with thicknesses of 20 nm and 6 nm for the NSL- and FIBL-patterned substrates respec-tively. Details about the NSL process can be found in our previous work [11]. FIBL was performed using a Carl Zeiss Cross-Beam 1540 EsB system. A 30 keV Ga+_{ion beam was used for patterning. The milling} current (2–50 pA) and milling time (5–50 s) were tailored in order to achieve the optimum patterning conditions. The sample surface was tilted 54° from the horizontal and placed at 5 mm working distance.

Sample morphologies were characterized in side- and plan-view with a Zeiss Leo 1550ﬁeld-emission gun scanning electron microscope (SEM), operated at 10 kV. Microstructural and elemental analysis were performed by STEM and energy-dispersive x-ray spectroscopy

(STEM-EDX) using the double-corrected Linköping FEI Titan3₆₀_–300, oper-ated at 300 kV.

3. Results and discussion

3.1. Growth optimization on NSL-patterned substrates

To study the temperature effect on the morphology of the GaN NRs and achieve the optimum growth conditions, a temperature-dependent growth series was prepared.Fig. 1presents the top- and side-view SEM images of GaN NRs grown on NSL-patterned Si substrates at 875, 900, 925, 940, 950, 965, 975, and 985 °C. As it can be seen, the increase in temperature results in improved selectivity of the SAG GaN NRs. At temperatures of 875 °C and lower, selectivity is not achieved, and NRs randomly grow both on the mask and inside the openings. The NRs grown at these temperatures are characterized by accentuated size non-uniformity and a large deviation from verticality is observed in the case of the NRs that nucleate onto the mask layer. Increasing the growth temperature to 900 °C leads to a tendency to grown only inside the openings, however, growth on the mask is still visible. The improved selectivity results also in a smaller NR length variation. The NRs have the tendency to develop a pencil-shaped top and are highly coalesced. At 925 °C selectivity is achieved, and the side-view SEM shows faceted NRs that grow inside the openings and the coalescence is reduced. When the growth temperature is raised further, selectivity is main-tained and the NRs shape becomes better defined with an improved aspect ratio. Starting from 940 °C the NRs develop aflat top-shape and better defined hexagonal cross-section. The sizes of the samples grown at 925, 940, and 950 °C are similar. At 965 °C, the correlated increased diffusion length conducts to a reduced growth along semipolar direc-tion and c-axis growth becomes the dominant growth direcdirec-tion after less than 100 nm height. By further increasing the growth temperature to 975 °C, desorption of the Ga atoms is enhanced, and the NRs have a

Fig. 1. Top- and side-view SEM images of GaN NRs grown on NSL-patterned Si substrates at diﬀerent temperatures. The ﬁgures have the same scale.

E.A. Serban et al. Thin Solid Films xxx (xxxx) xxx–xxx

(4)

similar diameter, but a decreased length. Finally, at 985 °C, the deso-rption process becomes dominant and no NRs are formed, growth oc-curring only in the form of small seeds inside the openings.

Fig. 2 shows the temperature dependence upon the dimensional distribution (diameter–Fig. 2a, and length–Fig. 2b) of the GaN NRs. The data was obtained from the measurements performed on the side-view SEM images of the NRs. The sample grown at the lowest tem-perature, 875 °C, characterized by a self-assembled growth regime, presents the smallest diameters, of around 110 nm, and a length of ~ 310 nm. At 900 °C, the diameter of the NRs increases to an average of 220 nm, as an effect of the mixed self-assembled and selective-area growth regimes. The size of the opening starts influencing the mor-phology of the NRs, resulting in a larger diameter correlated with shorter and more uniform lengths of around 280 nm. By further in-creasing the temperature, the NRs selectively grow, only in the desig-nated opening areas. The SAG combined with the characteristic di ffu-sion-induced growth mechanism results in an improved aspect ratio of the NRs, with decreasing diameters and increasing lengths with in-creasing growth temperature. The average diameters are: 270, 250, 230, and 160 nm, correlated with increasing lengths of 330, 345, 360, and 400 nm, for growth temperatures of, 925, 940, 950, and 965 °C, respectively. By further increasing the growth temperature to 975 °C, desorption of the Ga atoms is enhanced, and the NRs follow the trend of decreasing diameters, to 130 nm, but with decreasing lengths also, to an average value of 330 nm. Due to the high desorption rate at 985 °C, only small nuclei of around 30–40 nm are formed and the openings are not

Fig. 2. GaN NRs (a) diameter and (b) length distribution as a function of growth tem-perature. The dashed lines demarcate the 3-temperature dependent growth regimes. The blue and pink hachures represent the temperature intervals where the growth is domi-nated by diﬀusion and desorption process, respectively. (For interpretation of the refer-ences to color in thisﬁgure legend, the reader is referred to the web version of this article.) Fig. 3. Top view SEM images of GaN NRs deposited at 950 °C on substrates pre-patterned using 50 s milling time and: 50, 20, 10, 5, and 2 pA milling currents.

(5)

ﬁlled due to the limited supply of material. 3.2. Growth optimization on FIBL-patterned substrates 3.2.1. Milling current eﬀect

Theﬁrst step in optimizing the FIBL conditions consisted in ﬁnding the optimum milling current for patterning. The top-view SEM images of samples grown at 950 °C on substrates patterned with 50, 20, 10, 5, and 2 pA for 50 s milling time are presented inFig. 3. It can be observed that a high milling current, of 50 pA results in the complete destruction of the mask layer so that NRs grow in a self-assembled mode directly

onto the Si substrate. The self-assembled process, combined with the rough substrate surface created by the high milling current, results in a high density of non-uniform, tilted NRs. A decrease in the milling current to 20 and 10 pA induces less damage to the mask layer, so that the proximity effect is less pronounced, and portions of the mask re-main unaffected. However, multiple tilted NRs are still formed on the exposed Si areas. At 5 pA, the pattern gets better defined, but the in-duced substrate damage results in the growth of multiple NRs with diameters smaller than 100 nm. When the milling current is reduced to 2 pA, the pattern is well defined. However, no single NRs are formed, but nanostructures with an average size of 350–400 nm, resulting from

Fig. 4. Top view SEM images of GaN NRs deposited at (a) 950 °C and (b) 980 °C on 2 pA pre-patterned substrates using: 30, 15, 10, and 5 s milling times.

Fig. 5. (a) SEM image of the TEM lamella prepared by FIB and showing the side-view of GaN NRs deposited at 980 °C on 30 s milled Si substrates. (b) Overlaid STEM-EDX elemental maps acquired inside the marked area, outlining the interface between Si substrate, GaN NRs, and the TiNxmask layer.

Fig. 6. (a) STEM image of an inclined NR and (b) the lattice-resolved image acquired inside it.

(6)

the coalescence of multiple initial NRs. 3.2.2. Milling time and growth temperature eﬀect

To further optimize the FIBL process we tried toﬁnd the shortest milling time that would induce a minimum substrate damage and still make it possible to obtain well-deﬁned patterns. As suggested in the previous section, 2 pA current was used and the milling time was varied for: 30, 15, 10, and 5 s. The results can be observed in the SEM images of the NRs grown at 950 °C onto these patterned substrates, pre-sented inFig. 4a. For 30 s milling time, the coalesced NRs result in the formation of irregular nanostructures, with sizes of 200–300 nm inside the openings. The decrease in milling time to 10–15 s leads to the coalescence of a smaller number of NRs which grow with a higher degree of verticality. Finally, when using 5 s milling time, a combina-tion of single and coalesced NRs can be observed, with average sizes of 100–200 nm.

The mask layer used in the case of FIBL-patterned samples (6 nm) is thinner than that in the NSL case (20 nm). This allows an increase in growth temperature, since we can grow SA GaN NRs on Si substrates at temperatures up to 1000 °C [27]. The same series of samples, but de-posited at 980 °C is presented inFig. 4b. In this case, when using 30 s milling time, fewer NRs form inside the openings and we can observe the presence of single NRs also. Most of the openings contain 2–3 NRs that coalesce. This tendency is increasingly accentuated for shorter milling times, and mostly single NRs are obtained with a diameter of around 150 nm. Some of the NRs grow tilted and hexagonal faceting appears. Finally, the sample milled for 5 s presents a better morphology, featuring single hexagonal NRs of around 100 nm diameter and with improved verticality.

3.3. Compositional and structural characterization 3.3.1. Multiple nanorods

To further study the microstructure, composition and substrate-NRs interface, STEM-EDX was employed.Fig. 5a shows the SEM image of the TEM lamella prepared from the sample grown at 980 °C onto the 30 s milled substrate, and prepared by FIB, conﬁrming the formation of nanostructures from multiple, tilted, and irregular NRs. The overlaid Ga, Si, and Ti STEM-EDX maps acquired at the interface for the struc-ture marked inFig. 5a are presented inFig. 5b. The presence of a thin TiNxmask layer and two coalesced GaN NRs is conﬁrmed. No GaN growth can be observed on the mask layer, further stressing the ob-served growth selectivity. By analyzing the Si map, a pronounced substrate damage at the opening can be observed. The ion beam-in-duced damage results in a rough substrate surface responsible for the growth of multiple tilted NRs inside the openings.

The lattice-resolved HR-STEM image, acquired inside a tilted NR is presented inFig. 6.Fig. 6b, conﬁrms the deviation from verticality. The NR grows tilted with the c-axis oriented approximately 15° from the vertical position.

3.3.2. Single nanorods

Fig. 7a shows the SEM image of the TEM lamella prepared by FIB from the sample grown at 980 °C onto the 5 s milled substrate, con-ﬁrming the formation of single, straight and uniform NRs. The pattern consisted of 100 nm wide openings, spaced 500 nm distance from each other. These characteristics are maintained, resulting in ~550 nm long NRs with diameters in the range of 115–125 nm. The initial growth of one NR, starting at the interface was analyzed by STEM-EDX (Fig. 7b). The maps conﬁrm the selective formation of a single GaN NR in the TiNxmask opening. The Si map, does not show extensive ion beam-induced substrate damage, but a smooth interface resulting in the for-mation of a straight, single NR.

This investigation of the FIBL process suggests that minimization of mask and substrate damage is a signiﬁcant key for growing uniform and well-deﬁned NR arrays. Ion beam-induced damage of the substrate surface promotes a random oriented nucleation of the NRs and results in the growth of multiple, tilted NRs inside one opening. Softer milling conditions with low current and short time, improves the smoothness of the substrate surface. These conditions, correlated with an optimum growth temperature, conduct to the formation of single, uniform, and vertical NRs which develop exclusively inside the opening areas.

4. Conclusions

We report on the selective-area growth of GaN NRs by reactive MSE. The controlled growth was achieved by employing Si substrates, pre-patterned by NSL and FIBL. A thin TiNxmask layer was used in both cases. For the NSL-patterned samples, the growth temperature (in the range 875–985 °C) effect upon the morphology of the NRs was studied. At lower temperatures, selectivity is not achieved and NRs randomly nucleate on both the mask layer and inside the openings. The increase in temperature results in improved selectivity of the SAG GaN NRs. Starting from 925 °C, the NRs develop strictly inside the openings. At higher growth temperatures, NRs with an improved aspect ratio are obtained, exhibiting reduced diameters and increased lengths. This trend is terminated at 985 °C, when the process is governed by deso-rption and no NRs growth occurs. The growth conditions were trans-ferred on FIBL-patterned substrates, which offers better control on the NRs sizes and positions. The tailoring of the milling current and milling time during the FIBL process, helped to attain the optimum patterning conditions which resulted in the growth of straight, single NRs inside the pre-defined openings. High milling currents and long milling times induce ion beam damage in the substrate. The rough surface resulted in these conditions, leads to the growth of multiple, tilted NRs inside one opening. Finally, by reducing the substrate damage to the minimum with low milling currents and short milling times, the high growth temperature resulted in well-defined, single, uniform, and straight NRs.

Fig. 7. (a) SEM image of the TEM lamella prepared by FIB and showing the side-view of GaN NRs (indicated by arrows) deposited at 980 °C on 5 s milled Si substrates. (b) Overlaid Ga, Si, and Ti STEM-EDX elemental maps outlining the interface between Si substrate, the marked GaN NR, and the TiNxmask layer, respectively.

(7)

Acknowledgements

This work was supported by the Swedish Research Council (VR) under grants Nos.621-2012-4420, 621-2013-5360, and 2016-04412 and the Swedish Governmental Agency for Innovation Systems (VINNOVA) under the VINNMER international qualification program. The authors further acknowledge the Swedish Foundation for Strategic Research (SSF) through the Research Infrastructure Fellow Program no. RIF 14-0074 and the Knut and Alice Wallenberg Foundation for support of the electron microscopy laboratory in Linköping. The Swedish Government Strategic Research Area in Materials Science on Functional Materials at Linköping University (Faculty Grant SFO-Mat-LiU 2009-00971) isfinally acknowledged for financial support.

References

[1] S.N. Mohammad, H. Morkoç, Progress and prospects of group-III nitride semi-conductors, Prog. Quantum Electron. 20 (1996) 361–525.

[2] D. Zhu, D.J. Wallis, C.J. Humphreys, Prospects of III-nitride optoelectronics grown on Si, Rep. Prog. Phys. 76 (2013) 106501.

[3] S. Nakamura, Background story of the invention of eﬃcient InGaN blue-light-emitting diodes (Nobel Lecture), Angew. Chem. Int. Ed. 54 (2015) 7770–7788. [4] A.S. Augustine Fletcher, D. Nirmal, A survey of gallium nitride HEMT for RF and

high power applications, Superlattice. Microst. 109 (2017) 519–537. [5] B.J. Baliga, Gallium nitride devices for power electronic applications, Semicond.

Sci. Technol. 28 (2013) 074011.

[6] S.J. Pearton, F. Ren, GaN electronics, Adv. Mater. 12 (2000) 1571–1580. [7] S. Zhao, H.P.T. Nguyen, Md.G. Kibria, Z. Mi, III-nitride nanowire optoelectronics,

Prog. Quantum Electron 44 (2015) 14–68.

[8] N. Wang, Y. Cai, R.Q. Zhang, Growth of nanowires, Mater. Sci. Eng. R 60 (2008) 1–51.

[9] M. Junaid, Y.-T. Chen, J. Palisaitis, M. Garbrecht, C.-L. Hsiao, P.O.Å. Persson, L. Hultman, J. Birch, Liquid-target reactive magnetron sputter epitaxy of high quality GaN(0001̄) nanorods on Si(111), Mater. Sci. Semicond. Process. 39 (2015) 702–710.

[10] S.A. Fortuna, X. Li, Metal-catalyzed semiconductor nanowires: a review on the control of growth directions, Semicond. Sci. Technol. 25 (2010) 024005. [11] E.A. Serban, J. Palisaitis, C.-C. Yeh, H.-C. Hsu, Y.-L. Tsai, H.-C. Kuo, M. Junaid,

L. Hultman, P.O.Å. Persson, J. Birch, C.-L. Hsiao, Selective-area growth of single-crystal wurtzite GaN nanorods on SiOx/Si(001) substrates by reactive magnetron

sputter epitaxy exhibiting single-mode lasing, Sci. Rep. 7 (2017) 12701. [12] M. Forsberg, A. Serban, I. Poenaru, C.-L. Hsiao, M. Junaid, J. Birch, G. Pozina,

Stacking fault related luminescence in GaN nanorods, Nanotechnology 26 (2015) 355203.

[13] R. Calarco, M. Marso, T. Richter, A.I. Aykanat, R. Meijers, A.v.d. Hart, T. Stoica, H. Lu1th, Size-dependent photoconductivity in MBE-grown GaN−nanowires, Nano Lett. 5 (2005) 981–984.

[14] J.-K. Huang, Y. Liu, T.-P. Chen, H.-W. Huang, F.-I. Lai, P.-T. Lee, H. Lin,

C.-Y. Chang, T.S. Kao, H.-C. Kuo, Enhanced light extraction eﬃciency of GaN-based hybrid nanorods light-emitting diodes, IEEE J. Sel. Top. Quantum Electron. 21 (2015) 6000107.

[15] J.E. Kruse, L. Lymperakis, S. Eftychis, A. Adikimenakis, G. Doundoulakis, K. Tsagaraki, M. Androulidaki, A. Olziersky, P. Dimitrakis, V. Ioannou-Sougleridis, P. Normand, T. Koukoula, Th. Kehagias, Ph. Komninou, G. Konstantinidis, A. Georgakilas, Selective-area growth of GaN nanowires on SiO2-masked Si (111)

substrates by molecular beam epitaxy, J. Appl. Phys. 119 (2016) 224305. [16] J. Martinez, F. Barbagini, A. Bengoechea-Encabo, S. Albert, M.A. Sanchez García,

E. Calleja, Fabrication of GaN nanorods by focused ion beam, Microelectron. Eng. 98 (2012) 250–253.

[17] W. Bergbauer, M. Strassburg, Ch. Kölper, N. Linder, C. Roder, J. Lähnemann, A. Trampert, S. Fündling, S.F. Li, H.-H. Wehmann, A. Waag, N-face GaN nanorods: continuous-ﬂux MOVPE growth and morphological properties, J. Cryst. Growth 315 (2011) 164–167.

[18] M. Watson, Metal organic vapour phase epitaxy of AlN, GaN, InN and their alloys: a key chemical technology for advanced device applications, Coord. Chem. Rev. 257 (2013) 2120–2141.

[19] S.-K. Lee, H.-J. Choi, P. Pauzauskie, P. Yang, N.-K. Cho, H.-D. Park, E.-K. Suh, K.-Y. Lim, H.-J. Lee, Gallium nitride nanowires with a metal initiated metal-organic chemical vapor deposition (MOCVD) approach, Phys. Status. Solidus B 241 (2004) 2775–2778.

[20] B.O. Jung, S.-Y. Bae, Y. Kato, M. Imura, D.-S. Lee, Y. Hondaab, H. Amano, Morphology development of GaN nanowires using a pulsed-mode MOCVD growth technique, Cryst. Eng. Comm. 16 (2014) 2273–2282.

[21] G. Suo, S. Jiang, J. Zhang, J. Li, M. He, Synthetic strategies and applications of GaN nanowires, Adv. Condens. Matter Phys. 2014 (2014) 456163.

[22] K.A. Bertness, A. Roshko, N.A. Sanford, J.M. Barker, A.V. Davydov, Spontaneously grown GaN and AlGaN nanowires, J. Cryst. Growth 287 (2006) 522–527. [23] C. Chèze, L. Geelhaar, O. Brandt, W.M. Weber, H. Riechert, S. Münch,

R. Rothemund, S. Reitzenstein, A. Forchel, T. Kehagias, P. Komninou,

G.P. Dimitrakopulos, T. Karakostas, Direct comparison of free and catalyst-induced GaN nanowires, Nano Res. 3 (2010) 528–536.

[24] Y. Isosaki, Y. Yamashita, T. Yagi, J. Jia, N. Taketoshi, S. Nakamura, Y. Shigesato, Structure and thermophysical properties of GaNﬁlms deposited by reactive sput-tering using a metal Ga target, J. Vac. Sci. Technol. A 35 (2017) 041507. [25] J.W. Shon, J. Ohta, K. Ueno, A. Kobayashi, H. Fujioka, Fabrication of full-color

InGaN-based light-emitting diodes on amorphous substrates by pulsed sputtering, Sci. Rep. 4 (2014) 5325–5329.

[26] P. Jakkala, M.E. Kordesch, Electrical properties of InGaN thinﬁlms grown by RF sputtering at diﬀerent temperatures, varying nitrogen and argon partial pressure ratios, Mater. Res. Express 3 (2016) 106406.

[27] E.A. Serban, J. Palisaitis, M. Junaid, L. Tengdelius, H. Högberg, L. Hultman, P.O.Å. Persson, J. Birch, C.-L. Hsiao, Magnetron sputter epitaxy of high-quality GaN nanorods on functional and cost-eﬀective templates/substrates, Energies 10 (2017) 1322–1334.

[28] M. Junaid, P. Sandström, J. Palisaitis, V. Darakchieva, C.-L. Hsiao, P.O.Å. Persson, L. Hultman, J. Birch, Stress evolution during growth of GaN (0001)/Al2O3(0001)

by reactive dc magnetron sputter epitaxy, J. Phys. D. Appl. Phys. 47 (2014) 145301. [29] M. Junaid, C.-L. Hsiao, J. Palisaitis, J. Jensen, P.O.Å. Persson, L. Hultman, J. Birch, Electronic-grade GaN(0001)/Al2O3(0001) grown by reactive DC-magnetron sputter

epitaxy using a liquid Ga target, Appl. Phys. Lett. 98 (2011) 141915.