Phys. Status Solidi C 6, No. S2, S527–S530 (2009) / DOI 10.1002/pssc.200880801
自动化监测GaN nano-pendeo-epitaxy
on Si(111) substrates
Chaowang Liu*, Wang Nang Wang, Somyod Denchitcharoen, Alan Gott, Philip A. Shields,
and Duncan W. E. Allsopp
Department of Electronic and Electrical Engineering, University of Bath, BA2 7AY, United Kingdom
Received 10 September 2008, accepted 25 September 2008
Published online 15 January 2009
PACS 68.55.ag, 81.05.Ea, 81.07.-b, 81.15.Kk, 81.16.Rf
*Corresponding author: e-mail*************.uk, Phone: 0044 1225 383326, Fax: 0044 1225 383325
© 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1 Introduction GaN growth on silicon substrates has made rapid progress in the last decade [1, 2]. Monolayer aluminium deposition on the substrate has been shown to effectively protect the silicon surface from nitridation and to improve the film quality [3]. AlN is routinely used as the buffer layer to avoid melt-back of GaN at high growth temperatures [4, 5]. By inserting an AlGaN layer between the AlN buffer and GaN layer [4, 6], or superlattices such as AlN/GaN [6], AlGaN/GaN [7], or in situ SiNx masks [8-10], or low temperature AlN interlayers [9], the disloca-tion density is reduced and crack-free GaN films are ob-tained.
Conventional epitaxial lateral overgrowth (ELO) [11], selective area growth (SAG) [12], and pendeo epitaxy (PE) [13] on GaN-on-Si(111) substrates have been reported. The structural quality of the GaN epitaxy is further im-proved by using these techniques, but cracking remains a major issue. Growth on patterned bare silicon substrates with the pattern size on the micron-scale has also been studied by many groups [14-18] in order to improve the structural quality and reduce cracking in the GaN layer. Breaking the continuity of the GaN film by patterning the Si substrates in these studies is
proved to be useful; how-ever, it is not possible to obtain flat and smooth film across the whole wafer surface, which is required for applications such as thick GaN growth by hydride vapour phase epitaxy on silicon substrates.
Growth on patterned substrates on the nanoscale has made good progress in recent years both theoretically [19, 20] and experimentally, including GaN growth on silicon substrates [21-25] and on silicon carbide [26] by metalor-ganic vapour phase epitaxy (MOVPE) [21-22, 24-26] and molecular beam epitaxy (MBE) [23]. Dislocations are re-duced and cracks are suppressed by using this nano-heteroepitaxy technique.
This paper reports nano-pendeo-epitaxy (NPE) of GaN on Si(111) substrates by MOVPE. This novel technique is shown to achieve GaN films with superior structural and optical quality, and to overcome the cracking issue widely encountered in this material system. In addition to ena-bling the advantages of low cost silicon substrates to be exploited, the technique also increases the dielectric con-trast with the substrates by generating air voids at the inter-face between GaN and the silicon. This potentially allows the incorporation of a buried photonic crystal structure into optoelectronic devices. Three advances have underpinned the success of the NPE-GaN technology. Firstly, high qual-ity ultrathin GaN templates have been successfully grown by using a high temperature, pulsed
grown AlN buffer lay-er on (111) Si substrates. Secondly, uniform GaN/AlN/Si
Nano-pendeo-epitaxial GaN films have been grown by metal-organic vapour phase epitaxy on nanocolumns fabricated on thin GaN-on-Si(111) templates. Optical and structural per-formances of the films are improved, crack densities are re-duced, and self-separation of GaN films from the substrates at the nanocolumns has been realized. The success of the nano-pendeo-epitaxy of GaN films is attributed to the following three techniques developed in this study: (1) the growth of high quality ultrathin GaN/AlN templates on on Si(111) sub-strates, (2) the fabrication of uniform nanocolumns across the 2-inch wafer surface, and (3) the in situ conversion of the exposed silicon surfaces of the fabricated nanocolumns to silicon nitride.
S528 Chaowang Liu et al.: GaN nano-pendeo-epitaxy on Si(111) substrates
nanocolumns across a 2-inch wafer surface have been suc-
cessfully fabricated. Thirdly, the in situ conversion of the
exposed silicon surfaces of the fabricated nanocolumns to
silicon nitride has been developed.
We also demonstrate the possibility of separating the
GaN structure from the substrates at the nanostructured in-
terface region. Preliminary results show that GaN platelets
were easily separated from the substrate by breaking the
nanocolumns.
2 Experiments Ultra thin GaN templates were grown
on Si(111) substrates using an Aixtron 200/4HT RF-S
MOVPE reactor. The silicon substrates were thermally de-
sorbed for 10 min under H2 at 1140 ºC and a reactor pres-
sure of 100 mbar. Then a thin Al layer was deposited for a
few seconds to prevent the surface from nitridaion before
switching in ammonia for a 40 nm AlN buffer layer growth.
The growth temperature and reactor pressure were then
ramped to 1020 °C and 200 mbar to grow a nominal 30 nm
GaN layer. A pulsed growth technique established previ-ously for GaN growth [27] has been developed further to grow the AlN buffer layer. Figure 1(a) shows a SEM im-age of the surface of a template.
GaN/AlN/Si nanocolumns were fabricated from the templates using a self-assembled nickel nanodot mask [28]. First, a SiO2 layer with thickness around 100-200nm was deposited on the surface, then an about 5 nm thick nickel layer was deposited on top of the SiO2. Nickel nanodots were formed by annealing the films under N2 at around 830 ºC for 30 seconds. A SEM image of the nanodots is shown in Fig. 1(b). RIE and ICP etching were then applied to fabricate GaN/AlN/Si nanocolumns with the Ni/SiO2 ac-ting as a etching mask. Two SEM images in Figs. 1(c) and (d) show the nanocol
umns before and after removing the mask.
After cleaning the samples, the GaN re-growth pro-ceeds from in situ thermal nitridation [29] of the exposed silicon surface to form silicon nitride which is delineated by the red lines in Fig. 2(a). The growth process is moni-tored by a spectral reflectometer [30], which provides rich information for the growth from the inital nucleation as shown in Fig. 2(b) and coalescence to the final flat and smooth growth as shown in Fig. 2(c). The growth tem-perature and reactor pressure for the NP E-GaN growth were 1100 ºC and 100 mbar. The silicon nitride layer acts as growth selective mask. It also acts as a barrier for sili-con outdiffusion and prevents direct contact between gal-lium or GaN and silicon which might otherwise cause the melt-back of the GaN at high temperature [5]. It was found that the presence of this layer is crucial for forming a high quality GaN film with a flat and smooth surface, a neat and sharp bottom interface, and well defined air voids. On some samples MQWs were grown over the coalesced epi-layer so that the pit density in the NPE-GaN and in refer-ence samples can be compared. The samples were charac-terised by SEM, high resolution X-ray diffraction (HR-XRD), room temperature photoluminescence (P L), and atomic force microscopy (AFM).
3 Results and discussion The template surface shown in Fig. 1(a) features a dense coverage of tra
pezoi-dal GaN islands, which however is still smooth enough for fabricating the nanocolumns. This good coverage is impor-tant for assuring that the silicon nanocolumns are covered by GaN/AlN or at least AlN. If by chance, there are a few bare nanocolumns the nitridation process converts the ex-posed Si surface to silicon nitride, preventing direct contact between silicon and overgrowing GaN. The fabricated nanocolumns shown in Figs. 1(c) and 1(d) have diameters in the range of 40 to 120 nm, and a density of 6×109/cm2. The fill factor of the nanocolumns is around 0.5 and the depth into the silicon substrates is ~120 nm.
Figure 2(a) shows a schematic drawing of the growth seed regions formed by the GaN/AlN bi-layer topping the nanocolumns, and silicon nitride masks covering the bare silicon areas. Figures 2(b) and 2(c) are schematic draw-ings showing how the growing nanopillars coalesce, and how air voids between the nanocolumns are thus formed. Fig. 2(d) shows a cross-section SEM image of the interface region of a 1.2 µm thick NPE-GaN layer.
(a) (b)
(c)
(d)
Figure 1 SEM images of (a) an as-grown thin nitride-on-Si template surface, (b) nickel nano-dots formed on the template, (c) and (d) nanocolumns fabricated from the template, before and after removal of SiO2
and /or nickel.
Figure 2(a)-(c)schematic drawings showing the growth process described in the text, and (d) a cross-section SEM image of a NPE-GaN film.
(d)
Phys. Status Solidi C 6, No. S2 (2009)
S529
Figure 3 shows the room temperature PL for a 1.2 μm thick NPE-GaN and a 1.2 μm thick reference sample. The PL intensity on the NPE-GaN layer is at least three times stronger than that of a reference sample. The yellow lumi-nescence band is also much weaker. The reference sample was grown on planar silicon substrate using the same AlN buffer layer, but without any strain control and dislocation reduction layers. Fig. 4 shows the HR-XRD rocking curve measured from the 002 reflecti
on on the 1.2 µm NPE-GaN and the reference sample. The full width at half maximum of the rocking curves are 750 arcsec on NP E-GaN com-pared with 1100 arcsec obtained on the reference sample.
AFM and SEM have been used to study the surface of InGaN/GaN MQWs samples grown on NP E-GaN and on the reference GaN templates. The V-pit density of 6×109/cm 2 on NP E-GaN template, measured on the SEM image shown in Fig. 5(b), is about half of that of reference sample. The results of all these characterization methods clearly and consistently demonstrate that the structural quality of NPE-GaN is markedly superior.
It has been predicted theoretically [19, 20] and proved experimentally [21] that heteroepitaxial materials grown on nano-pillars will not generate new dislocations in the coalesced overgrown materials by lattice mismatch. Not all
dislocations existing in the seed propagate through to the surface of the overgrown layer, instead some are bent and annihilated in the NP E-GaN process [25, 26, 31]. These generally result in the reduction of dislocations. However, greater reductions in the dislocation density will require further study in order to understand the impact of factors such as dislocations in the seed layer and pattern fill fac-tors on dislocation reduction mechanisms. In addition, it is not clear if any new dislocations and how many are gener-ated at the coalescence boundary in the NPE process.
As well as reducing the density of dislocations, the nanocolumns also inhibit cracking of the epitaxial film as shown in Fig. 6. Two possible strain reduction mechanisms contribute to the reduced cracking. Firstly, the strain in the NP E-GaN layer from the nano-heteroepitaxy is relaxed through a three dimensional deformation of the nanocol-umns before coalescence [19-26]. Secondly, probably a more important route for tensile strain built up during the
coalescence and cool-down phases to release is through nanocolumn deformation, with some even breaking to ac-commodate the strain energy after cool-down. In the most extreme situation, a sufficient number of the nanocolumns
P L i n t e n s i t y (a .u .)
Wavelength/nm
Figure 3 Room temperature PL of an 1.2 µm thick NPE-GaN and an 1.2 µm thick reference sample.
Figure 5 (a) The cross-section and (b) the plan view SEM images of a MQWs grown on top of a 1.2 µm NPE-GaN.
(a)
(b)
Figure 6 Crack density comparison in an 1.2 µm reference (a) and an 1.2 µm NPE-GaN sample (b).
(a)
(b)
N o r m a l i z e d X R D i n t e n s i t y (a .u )
Omega rocking angle/degree
Figure 4 002 X-ray rocking curve of an 1.2 µm thick NPE-GaN and an 1.2 µm thick reference sample.
Figure 7 SEM images showing (a) the breaking of nano-columns and (b) a small GaN pallet separated from the substrate.
(a)
(b)
S530 Chaowang Liu et al.: GaN nano-pendeo-epitaxy on Si(111) substrates
can break to cause separation of the epitaxy from the sub-strate.
Figure 7(a) shows a SEM image of an area with broken silicon nanocolumns, evidence of nanocolumn compliance and breakage during cooling down. Figure 7(b) shows a GaN platelet that h
吸塑片材as delaminated from the Si substrate. This is a positive though preliminary result showing that NPE-GaN film can be separated from the substrate. This creates the prospect of separating thick GaN layers as a whole wafer from the substrate by breaking the nanocol-umns without breaking the epitaxy or substrate into pieces.
4 Summary A novel growth technique, i.e. nano-pendeo-epitaxy by MOVP E, for GaN film growth on Si(111) substrates has been demonstrated. The threading dislocation density and tensile strain in the NPE-GaN films are reduced, the optical performance is improved. Well de-fined air voids are formed at the interface between the GaN and silicon by this growth technique. They are shown to make a critical contribution to the formation of a compliant substrate on which otherwise highly strained epitaxial lay-ers can be grown with significantly reduced densities of dislocation. Three advances have underpinned the success of the NPE-GaN technology. Firstly, high quality ultrathin GaN templates have been successfully grown by using a high temperature, pulsed grown AlN buffer layer on Si(111) substrates. Secondly, uniform nanocolumns across the 2-inch wafer surface have been successfully fabricated. Thirdly, in situ conversion of the exposed silicon surfaces of the fabricated nanocolumns to silicon nitride has been developed. A positive although preliminary conclusion is that separation of GaN films from Si(111) substrate is also possible. Study on the further reduction of the threading dislocation density is ongoing.
References
[1] A. Krost and A. Dadgar, Mater. Sci. Eng. B 93, 77 (2002).
[2] A. Dadgar, A. Strittmatter, J. Blasing, M. Poschenrieder, O.
Contreras, P. Veit, T. Riemann, F. Bertram, A. Reiher, A.
Krtschil, A. Diez, T. Hempel, T. Finger, A. Kasic, M. Schu-
bert, D. Bimberg, F. A. Ponce, J. Christen, and A. Krost, Phys. Status Solidi C 0(6), 1583 (2003).
[3] S.A. Nikishin, N.N. Farleev, V.G. Antipov, S. Francoeur, L.
Grave de Perlta, G.A. Seryogin, H. Timkin, T.I. Prokofyeva, M. Holtz, and S.N.G. Chu, Appl. Phys, Lett. 75, 2073 (1999).
[4] H. Ishikawa, G.Y. Zhao, N. Nakada, T Egawa, T. Soga, T.
Jimbo, and M. Umeno, Phys. Status Solidi A 176, 599 (1999).
[5] H. Ishikawa, K. Yamamoto, T. Egawa, T. Soga, T. Jimbo,
and M. Umeno, J. Cryst. Growth 189/190, 178 (1998).
[6] A. Ubukata, K. Ikenaga, N. Akutsu, A. Yamaguchi, K. Ma-
tsumoto, T. Yamazaki, and T. Egawa, J. Cryst. Growth 298, 198 (2007).
[7] A. Dadger, J. Christen, T. Riemann, S. Richter, J. Blasing,
A. Diet, A.Krost, A. Alam, and M. Heukeu, Appl. Phys.
Lett. 78, 2211( 2001).
[8] P.R. Hageman, S. Haffouz, V. Kirilyuk, A. Grzegorczyk,
and P.K. Larsen, Phys. Status Solidi A 188(2), 523 (2001). [9] A. Dadgar, M. Poschenrieder, J. Bläsing, K. Fehse, A. Diez,
and A. Krost, Appl. Phys. Lett. 80, 3670 (2002).
[10] O. Contreras, F.A. Ponce, J. Christen, A. Dadger, and A.
Krost, Appl. Phys. Lett. 81, 4712 (2002).
[11] E. Feltin, B. Beaumont, P. Vennegues, M. Vaille, and P.
Gibart, T. Riemann, J. Christen, L. Dobos, and B. Pecz, J.
Appl. Phys. 93, 182 (2003).
[12] A. Dadger, A. Alam, T. Riemann, J. Blasing, A.Diez, M.
Poschenrieder, M. Strassburg, M. Heuken, J Christen, and A Krost, Phys. Status Solidi A 188,155 (2001).
[13] T. Gehrke, K. J. Linthicum, E. Preble, P. Rajagopal, C. Ron-
ninG, C. Zorman, M. Mehregany, and R.F. Davis, J. Elec-
tron. Mater. 29(3), 306 (2000).
[14] B.S. Zhang, H. Liang, Y. Wang, Z.H. Feng, K.W. Ng, and
K.M. Lau, J. Cryst. Growth 298, 725 (2007).
[15] D. Wang, Y. Dikme, S. Jia, K.J. Chen, K.M. Lau, P. van
Gemmern,Y.C. Lin, H. Kalisch, R.H. Jansen, and M. Heu-
ken, J. Cryst. Growth 272, 489 (2004).
[16] S. Tanaka,Y. Kawaguchi, N. Sawaki, M. Hibino, and K. Hi-
ramatsu, Appl. Phys. Lett. 76, 2701 (2000).
[17] A. Strittmatter, S. Rodt, L. Reißmann, D. Bimberg, H.
Schröder, E. Obermeier T. Riemann, J. Christen, and A.
Krost, Appl. Phys. Lett. 78, 727 (2001).
[18] Y. Honda, Y. Kuroiwa, M. Yamaguchi, and N. Sawaki,
Appl. Phys. Lett. 80, 222 (2002).
[19] S. Luryi and E. Suhir, Appl. Phys. Lett. 49, 140(1986).
[20] D. Zubia and S. D. Hersee, J. Appl. Phys. 85, 6492 (1999).
扭王字块模具[21] 11 D. Zubia, S. H. Zaidi, S. R. J. Brueck, and S. D. Hersee, Appl. Phys. Lett. 76, 858 (2000).
[22] K.Y. Zang, Y.D. Wang, L.S. Wang, S. Tripathy, S.J. Chua,
and C.V. Thompson, Thin Solid Films 515, 4505 (2007). [23] J. Liang, S.-K. Hong, N. Kouklin, R. Beresford, and J. M.
Xu, Appl. Phys. Lett. 83, 1572 (2003).
[24] 15 L.S. Wang, S. Tripathy, B.Z. Wang, and S.J. Chua, Appl.
Surf. Sci. 253, 214 (2006).
[25] S.D. Hersee, D. Zubia, X.Y Sun, R. Bommena, M. Fairchild,
srteS. Zhang, D. Burckel, A. Frauenglass, and S.R.J. Brueck, IEEE J. Quantum. Electron. 38, 1017 (2002).
[26] X. Y. Sun, R. Bommena, D. Burckel, A. Frauenglass, M.N.
Fairchild, S.R.J. Brueck, G.A. Garrett, M. Wraback, and S.D. Hersee, J. Appl .Phys. 95, 1450 (2004).
[27] C. Liu, P.A. Shields, S. Denchitcharoen, S. Stepanov A.
Gott, and W.N. Wang, J. Cryst. Growth 300, 104 (2007). [28] D. Aurongzeb, S. Patibandla, M. Holtz, and H. Temkin,
Appl. Phys. Lett. 86, 103107 (2005).
[29] M. M. Moslehi and K. C. Saraswat, IEEE J. Solid-state Cir-
cuits SC-20,26(1985).
[30] C. Liu, S. Stepanov, P.A. Shields, A. Gott, W.N. Wang, E.
Steimetz, and J.-T. Zettler, Appl. Phys. Lett. 88, 101103 (2006).
[31] L. Meshi, D. Cherns, I. Griffiths, S. Khongphetsak, A. Gott,
C. Liu, S. Denchitcharoen, P. Shields, W. Wang, R. Cam-
pion, S. N ovikov, and T. Foxon, Phys. Status Solidi C 5, 1645 (2007).
豫公网安备41160202000603
站长QQ:729038198
关于我们
投诉建议