Structural and optical characterization of GaN ... - Semantic Scholar
Microelectronics Journal 35 (2004) 891–895 www.elsevier.com/locate/mejo
Structural and optical characterization of GaN grown on porous silicon substrate by MOVPE N. Chaabena,*, T. Boufadena, M. Christophersenb, B. El Jania a
Unite´ de Recherche sur les He´te´ro-Epitaxy et Applications, Faculte´ des sciences de Monastir, Monastir 5000, Tunisia b Center for Future Health, University of Rochester, 601 Elmwood Ave, Rochester, NY 14642, USA Received 2 June 2004; received in revised form 20 July 2004; accepted 26 July 2004
Abstract GaN was grown on porous silicon (PS) substrates by Metalorganic Vapour Phase Epitaxy at temperature of 1050 8C. An additional AlN buffer layer is used between GaN and PS. The crystalline quality and surface morphology of GaN films were studied by X-ray diffraction and scanning electron microscope (SEM), respectively. Preferential growth of hexagonal GaN with h00.1i direction is observed and is clearly improved when the thickness of AlN buffer layer increases. Morphological changes in PS layer appearing after growth have been also discussed. GaN optical qualities were determined by photoluminescence at low and room temperature (RT). q 2004 Elsevier Ltd. All rights reserved. Keywords: GaN; AlN buffer; Porous silicon; SEM; Photoluminescence
1. Introduction The wide band gap semiconductors of the nitrides III–V and their alloys have received enormous attention for their practical use in both optoelectronic and high temperature/ high power electronic devices [1]. In addition to the application fields for crystalline and perfect GaN, polycrystalline GaN may also have utility as electronic and optoelectronic materials. It could grow with high band edge photoluminescence (PL) efficiency on a wide variety of large and low cost substrates promising the possibility to fabricate large area and low cost photonic devices operating in the blue and UV regions [2]. Recently, the demonstration of the fabrication of light emitting devices using polycrystalline InGaN/GaN layers were reported [3]. Growth of strong PL emission polycrystalline GaN on polycrystalline metal substrates has been successfully demonstrated [4,5]. Poly-GaN presents high efficiency of PL without yellow emission [6]. * Corresponding author. E-mail address: [email protected] (N. Chaaben). 0026-2692/$ - see front matter q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.mejo.2004.07.009
Porous silicon (PS) is also a good candidate to easily grow polycrystalline GaN. It is easily obtained on Si substrate and pore formation is intensively studied. Morphology, thickness and porosity are well controlled on n- and p-type oriented (100)/(111) silicon substrates [7]. PS keeps its crystalline character and has a supple structure that can reduce the thermal strain and lead to free-dislocations of GaN grains. We have recently performed the direct growth of GaN on PS in the temperature range of 450–1000 8C [8–10]. X-ray analysis showed cubic inclusions with proportion depending on growth temperature. At high growth temperature, the adherence of GaN to PS was poor and PS surface was partially covered. The PL spectrums were dominated by narrow deep emissions related to stacking faults and structurally perturbed regions [11]. Those emissions seem to depend on the growth temperature. We have also studied the annealing effect on PS in temperature range from 300 to 1000 8C [12]. We observed that the porous structure could remain steady under GaN growth conditions. In this paper, we report on the results of GaN growth on PS at high temperature (1050 8C). We used AlN buffer layer
892
N. Chaaben et al. / Microelectronics Journal 35 (2004) 891–895
to improve the interfacial sharpness and wetting between GaN and PS. AlN can also reduce the lattice mismatch between GaN and Si to 2.5% and consequently leads to high quality polycrystalline GaN with large grain size.
2. Experiments The used PS substrates were obtained by standard anodisation on oriented (100) p-type Si substrate. The growth of GaN was performed by atmospheric pressure MOVPE [12]. The obtained samples were characterized by scanning electron microscope (SEM) and by X-ray analysis. PL measurements were made using an argon laser with doubled frequency (244 nm) as the excitation source with power density ranging from 5 to 50 mW. A Jobin Yvon triax spectrometer and cooled CCD camera completed the set-up.
3. Results In the cross section SEM micrograph (Fig. 1a) of the PS layers, we observe (at the top) a low porosity PS layer (lp-PS) with a flat and continuous surface and a deeper high porosity PS layer (hp-PS). The thickness of lp-PS was around 1 mm whereas one of the hp-PS was a few hundred nanometers. The interface between these two PS layers is rough and not straight. The surface morphology of hp-PS was observed after detachment of the top layer (lp-PS) after AlN growth. Its plan view SEM is presented in Fig. 1b. Highly distributed macropores with diameter up to 250 nm are clearly shown. These macropores are perpendicularly oriented to the surface and do not extend the PS surface. This observed gradual morphology, i.e. lp-PS/hp-PS is well known in PS and is related to the etch conditions [7]. The hp-PS appears lighter than the lp-PS and also than bulk Si. This contrast, probably due to its insulating behaviour that generally increases with porosity, prevents more detailed structural information such as the depth of macropores. The growth of AlN and GaN layers was in situ controlled by laser reflectometry. In this experiment we observed that the AlN growth begins by an incubation period and then the reflectivity increases in some cases and decreases in others. The study of this behaviour permitted us to directly reach to a mean porosity of the cut PS samples. The measured value was around 60% indicating a relatively good homogeneity as confirmed by SEM observations. In GaN growth phase, the reflectivity decreases due to the rough surface showing a three-dimensional growth mode. More details of reflectivity results are shown in Ref. [12]. Fig. 2 shows surface micrograph of GaN obtained on the thickest AlN (growth time was 600 s). The surface is rough and presents grains of crystallites having very varied shapes and micrometrics sizes. Their average size is estimated to be approximately 2 mm with a maximum size
Fig. 1. Cross section SEM micrograph (a) of PS layer supported by p-type (100) silicon substrate revealing a gradual PS structure formed by lp-PS/ hp-PS. Plan view SEM (b) of macroporous structure obtained after detachment of lp-PS during the AlN growth. More details are given in Fig. 5.
up to approximately 3 mm. These crystallites are randomly distributed and irregularly disoriented. However, the surface is completely covered compared to obtained morphologies by high temperature direct growth on PS [8–10]. Similar GaN structure is observed for the other AlN thickness. The X-ray diffraction (XRD) analysis showed a polycrystalline GaN that grows in merely hexagonal structure with a preferential h00.1i direction. But there were no XRD peaks corresponding to cubic GaN observed from all of our grown GaN on PS samples. The diffraction spectra were described elsewhere [12]. The percentages in volume of oriented crystallites in the hhk.li direction were estimated using the model detailed in Ref. [13] and results are given in Table 1. The contribution of h00.1i is dominant compared to the other directions. The percentage of oriented crystallites according to the c-axis
N. Chaaben et al. / Microelectronics Journal 35 (2004) 891–895
Fig. 2. Plan view SEM micrograph of GaN layer grown on PS with AlN buffer layer. The AlN growth time is 600 s.
h00.1i of the wurtzite structure increases as function of the AlN thickness. We compare the intensity of the (10.1) peak which is the strongest reflection for perfectly random polycrystals to the one of (00.2) peak. The intensity ratio of (10.1)–(00.2) peak (Fig. 3) increases from 0.1 to 0.7 as a function of AlN thickness proving that the increase of the AlN buffer thickness improves the crystalline quality and seems to adapt more GaN and PS layers. AlN buffer with appropriate thickness can limit the substrate surface perturbation due to diffusion and surfactant effect of Si under high temperature GaN growth conditions [14]. Cross section observation (Fig. 4) shows that the interface between GaN and the PS is continuous and flat. The lp-PS layer is clearly discernable and has a thickness of about 1 mm as like the original PS layer. However, as shown in this figure, lp-PS layer is detached from the Si substrate. The lp-PS layer is kept intact with flat bottom face and remains bounded to GaN layer. Under the detached lp-PS, the interface has a structure similar to that seen in the original PS/Si (Fig. 1a). Depressions on Si surface were formed to a depth of a few hundred nanometres. This depth corresponds to one of the described hp-PS layer. We conclude that the hp-PS was fractured and was more fragile than the lp-PS. This detachment may be associated to
893
Fig. 3. Intensity ratio of (10.1)–(00.2) peak evolution as a function of AlN growth time.
the stress induced in PS during the post growth-cooling phase. Generally, the PS layers obtained by standard anodisation method often show a gradient of porosity which increases in depth related to PS growth conditions where the top of PS layer has a lower porosity than the bottom. Therefore, we think that the interfacing between PS and Si substrate is more flexible than the one between AlN/ PS and GaN and can undergo most of the thermally induced deformation in the porous layer. This is confirmed when we just grow AlN (400 s) followed by annealing at 1050 8C and cooling down to room temperature (RT). In Fig. 5, we show that the AlN/PS (zone 1) layer is completely cracked revealing a second zone (zone 2) where we observe that the porous structure was taken off and a macroporous structure is clearly shown. We confirm that the thermal instability of hp-PS is responsible for the appearance of crack after AlN growth. These cracks are not observed in the final GaN layer
Table 1 Percentages of oriented crystallites in hhk.li direction calculated for GaN films grown with different AlN growth time: 200 s (c1), 400 s (c2) and 600 s (c3) 2qhkl
(hk.l)
c1 (%)
c2 (%)
c3 (%)
32.4170 34.5979 36.8746 48.1385 57.8251 63.4953 72.9847
(10.0) (00.2) (10.1) (10.2) (11.0) (10.3) (00.4)
10 45 16 8 6 15 –
11 53 16 7 3 10 –
– 80 3 3 – 4 10
Fig. 4. Cross section SEM micrograph of GaN layers grown on PS with AlN buffer layer. The AlN growth time is 400 s. The PS layer is detached from the Si substrate. The AIN layer is very thin and is not discemable in the SEM resolution.
894
N. Chaaben et al. / Microelectronics Journal 35 (2004) 891–895
Fig. 5. Surface SEM micrograph of AlN/PS annealed at 1050 8C. The AlN growth time is 400 s. The lp-PS (zone 1) is cracked and detached revealing a macroporous hp-PS layer (zone 2).
probably due to the insufficient coalescence of GaN crystallites. We show that the instability of hp-PS is responsible for detachment of lp-PS and crack appearance after AlN growth. These two phenomena, strongly dependent, are principally related to the thermal growth conditions. We have previously signalled that porous structure could remain steady under GaN growth conditions. However, thermally activated micro structural changes can appear during all GaN growth steps and can consequently affect the thermal strain level in PS layer. They can consist of PS recrystallisation and residual oxide desorption inducing atom mobility and reorganization of PS structure. These have not been revealed by our SEM observations. The RT PL measurements of GaN, given in Fig. 6a, show near band edge emission at 3.4 eV that becomes more accentuated for the thickest AlN buffer. These optical qualities are consistent with the crystalline quality improvement versus AlN thickness. Then, we studied the temperature PL dependence (5–100 K) for the best crystalline and ambient optical qualities GaN film (Fig. 6b). At low temperature, the donor bound exciton (DBE) emission at 3.470 eV dominates the PL spectrum. Two others not well pronounced peaks at 3.476 and 3.510 eV are present and separated from DBE peak by 6 and 34 meV, respectively, that can be assigned to free exciton (FE) transitions. In Fig. 6c, we plotted the integrate intensity of the DBE emission versus inverse of temperature and fitted the experimental data to the well known expression [15] IðTÞ Z
I0 ; 1 C C exp K KEBaT
where I0 and C are constant parameters, KB is the Boltzman constant and Ea is the activation energy. The fit gives Ea of about 5.8 meV that can be assigned to the dissociation
Fig. 6. PL measurements of GaN. (a) Room temperature PL for two AlN growth duration: 200 s (1) and 600 s (2). (b) Temperature dependence (5–100 K) of PL for GaN grown in the thickest AlN buffer (the insert figure give the temperature dependence of the yellow emission). (c) Integrate intensity of DBE emission versus inverse of temperature.
energy of DBE into FE [16]. The DBE quenches hardly above 50 K. At relatively low energy, two other lines at 3.44 and 3.42 eV are also observed and dominate the PL spectrum above 50 K. The position of the first peak (3.44 eV) coincides with that of the two-electron satellites of the DBE in GaN [16,17]. Although, as shown in Fig. 6b, the DBE peak quench faster than 3.44 eV peak, we consequently rule out such identification. Recent high spatial resolution PL studies demonstrated that the bright
N. Chaaben et al. / Microelectronics Journal 35 (2004) 891–895
emission at 3.45–3.46 eV originates from the inversion domain boundaries and this peak was attributed to shallow trap [18]. Reshchikov et al. [17] show that the 3.45 eV peak dominated PL spectrum for N polar GaN having a high density of inversion domain. They attributed this emission to excitons bound to inversion domain boundaries. Knowing that our GaN samples were grown under nitrogen-rich conditions that can favour N polar GaN formation, we agree the assumption made by Reshchikov that the corresponding transition is related to excitons bound to inversion domain boundaries. For the second peak at 3.42 eV, it was extensively studied and found related to excitons bound to structural and surface defects [19,20]. At about the same energy, 3.41 eV emissions have been shown in low structural quality and related to excitons bound to structural defect [20]. Kim et al. [19] have been reported that crystal structural imperfections and rough surface morphology are responsible for the luminescence band around 3.42 eV in GaN. The band position does not have fixed energy level, but changes from 3.402 to 3.428 eV depending upon the GaN crystal qualities. Reshchikov [17] also attributed the 3.42 eV emission to excitons bound to some surface defects rather than to dislocations. All the mentioned assumptions are consistent with the polycrystalline structure of our GaN samples and allow us to conclude that the 3.42 eV peak is related to excitons bound to structural and surface defects. We also notice the presence of a weaker broad yellow band emission at around 2.2 eV (see the inset figure in Fig. 6b) that does not quench with increasing temperature.
4. Conclusion The adhesion between PS and GaN is improved by using AlN buffer layer. GaN layers showed a preferential growth towards the c-axis and the alignment is enhanced with increasing AlN thickness and RT optical qualities were consistent with the crystalline quality improvement. Low temperature PL revealed excitonic transition energies below the DBE line at 3.44 and 3.42 eV related to structural and surface defects in polycrystalline GaN. The thickness of porous layer seems to be insufficient to absorb the residual thermal strain between GaN and Si. To succeed in the GaN growth, properties of PS (porosity and thickness) and growth conditions of AlN buffer (thickness and growth temperature) need to be optimized.
895
Microstructural changes in PS should be also profoundly studied versus GaN growth conditions.
Acknowledgements The authors would like to thank H. Fo¨ll (University of Kiel) for providing the PS samples.
References [1] S. Strite, H. Morkoc, GaN, AlN and InN: a review, J. Vac. Sci. Technol. B 10 (1992) 1237–1266. [2] K. Iwata, H. Asahi, K. Asami, R. Kuroiwa, S. Gonda, Jpn. J. Appl. Phys. 36 (1997) L661. [3] D.P. Bour, N.M. Nickel, C.G. Van De Walle, M.S. Kneissl, B.S. Krusor, P. Mei, N.M. Johnson, Appl. Phys. Lett. 76 (2000) 2182. [4] K. Yamada, H. Asahi, H. Tampo, Y. Imanishi, K. Ohnishi, K. Asami, IPAP Conf. Ser. 1 (2000) 556. [5] K. Yamada, H. Asahi, H. Tampo, Y. Imanishi, K. Ohnishi, K. Asami, Appl. Phys. Lett. 78 (2001) 2849. [6] H. Asahi, H. Tampo, K. Yamada, K. Ohnishi, Y. Imanishi, K. Asami, Phys. Status Solidi (a) 188 (2) (2001) 601–604. [7] C. Ja¨ger, B. Finkenberger, W. Ja¨ger, M. Christophersen, J. Carstensen, H. Fo¨ll, Mater. Sci. Eng. B 69/70 (2000) 199. [8] A. Missaoui, M. Saadoun, H. Ezzaouia, B. Bessaı¨s, T. Boufaden, A. Rebey, B. El Jani, Phys. Status Solidi (a) 182 (2000) 189. [9] A. Matoussi, T. Boufaden, A. Missaoui, S. Guermazi, B. Bessaı¨s, Y. Mlik, B. El Jani, Microelectron. J. 32 (2001) 995. [10] A. Missaoui, H. Ezzaouia, B. Bessaı¨s, T. Boufaden, A. Matoussi, M. Bouaı¨cha, B. El Jani, Mater. Sci. Eng. B 93 (2002) 102–106. [11] T. Boufaden, A. Matoussi, S. Guermazi, S. Juillaguet, A. Toureille, Y. Mlik, B. El Jani, Phys. Status Solidi (a) 201 (3) (2004) 582–587. [12] T. Boufaden, N. Chaaben, M. Christophersen, B. El Jani, Microelectron. J. 34 (2003) 843–848. [13] P. Joubert, B. Loisel, Y. Chouan, L. Haji, J. Electrochem. Soc. 134 (10) (1987) 2541. [14] H. Lahre`che, P. Venne´gue`s, O. Tottereau, M. Lau¨gt, P. Lorenzini, M. Leroux, B. Beaumont, P. Gibart, J. Cryst. Growth 217 (2000) 13. [15] G. Davies, Phys. Rep. 176 (1989) 83. [16] K. Kornitzer, T. Ebner, K. Thonke, R. Sauer, C. Kirchner, V. Schwegler, M. Kamp, M. Leszczynski, I. Grzegory, S. Porowski, Phys. Rev. B 60 (3) (1999) 1471–1473. [17] M.A. Reshchikov, D. Huang, F. Yun, P. Visconti, L. He, H. Morcoc¸, J. Jasinski, Z. Liliental-Weber, R.J. Molnar, S.S. Park, K.Y. Lee, J. Appl. Phys. 94 (9) (2003) 5623–5632. [18] P.J. Schuck, M.D. Mason, R.D. Grober, O. Ambacher, A.P. Lima, C. Miskys, R. Dimitrov, M. Stutzmann, Appl. Phys. Lett. 79 (2001) 952. [19] S.T. Kim, Y.J. Lee, S.H. Chung, D.C. Moon, Semicond. Sci. Technol. 14 (1999) 156–160. [20] S. Fischer, G. Steude, D.M. Hofmann, F. Kurth, F. Anders, M. Topf, B.K. Meyer, F. Bertram, M. Schmidt, J. Christen, L. Eckey, J. Holst, A. Hoffmann, B. Mensching, B. Rauschenbach, J. Cryst. Growth 189/190 (1998) 556–560.
Structural and optical characterization of GaN grown on porous silicon substrate by MOVPE N. Chaabena,*, T. Boufadena, M. Christophersenb, B. El Jania a
Unite´ de Recherche sur les He´te´ro-Epitaxy et Applications, Faculte´ des sciences de Monastir, Monastir 5000, Tunisia b Center for Future Health, University of Rochester, 601 Elmwood Ave, Rochester, NY 14642, USA Received 2 June 2004; received in revised form 20 July 2004; accepted 26 July 2004
Abstract GaN was grown on porous silicon (PS) substrates by Metalorganic Vapour Phase Epitaxy at temperature of 1050 8C. An additional AlN buffer layer is used between GaN and PS. The crystalline quality and surface morphology of GaN films were studied by X-ray diffraction and scanning electron microscope (SEM), respectively. Preferential growth of hexagonal GaN with h00.1i direction is observed and is clearly improved when the thickness of AlN buffer layer increases. Morphological changes in PS layer appearing after growth have been also discussed. GaN optical qualities were determined by photoluminescence at low and room temperature (RT). q 2004 Elsevier Ltd. All rights reserved. Keywords: GaN; AlN buffer; Porous silicon; SEM; Photoluminescence
1. Introduction The wide band gap semiconductors of the nitrides III–V and their alloys have received enormous attention for their practical use in both optoelectronic and high temperature/ high power electronic devices [1]. In addition to the application fields for crystalline and perfect GaN, polycrystalline GaN may also have utility as electronic and optoelectronic materials. It could grow with high band edge photoluminescence (PL) efficiency on a wide variety of large and low cost substrates promising the possibility to fabricate large area and low cost photonic devices operating in the blue and UV regions [2]. Recently, the demonstration of the fabrication of light emitting devices using polycrystalline InGaN/GaN layers were reported [3]. Growth of strong PL emission polycrystalline GaN on polycrystalline metal substrates has been successfully demonstrated [4,5]. Poly-GaN presents high efficiency of PL without yellow emission [6]. * Corresponding author. E-mail address: [email protected] (N. Chaaben). 0026-2692/$ - see front matter q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.mejo.2004.07.009
Porous silicon (PS) is also a good candidate to easily grow polycrystalline GaN. It is easily obtained on Si substrate and pore formation is intensively studied. Morphology, thickness and porosity are well controlled on n- and p-type oriented (100)/(111) silicon substrates [7]. PS keeps its crystalline character and has a supple structure that can reduce the thermal strain and lead to free-dislocations of GaN grains. We have recently performed the direct growth of GaN on PS in the temperature range of 450–1000 8C [8–10]. X-ray analysis showed cubic inclusions with proportion depending on growth temperature. At high growth temperature, the adherence of GaN to PS was poor and PS surface was partially covered. The PL spectrums were dominated by narrow deep emissions related to stacking faults and structurally perturbed regions [11]. Those emissions seem to depend on the growth temperature. We have also studied the annealing effect on PS in temperature range from 300 to 1000 8C [12]. We observed that the porous structure could remain steady under GaN growth conditions. In this paper, we report on the results of GaN growth on PS at high temperature (1050 8C). We used AlN buffer layer
892
N. Chaaben et al. / Microelectronics Journal 35 (2004) 891–895
to improve the interfacial sharpness and wetting between GaN and PS. AlN can also reduce the lattice mismatch between GaN and Si to 2.5% and consequently leads to high quality polycrystalline GaN with large grain size.
2. Experiments The used PS substrates were obtained by standard anodisation on oriented (100) p-type Si substrate. The growth of GaN was performed by atmospheric pressure MOVPE [12]. The obtained samples were characterized by scanning electron microscope (SEM) and by X-ray analysis. PL measurements were made using an argon laser with doubled frequency (244 nm) as the excitation source with power density ranging from 5 to 50 mW. A Jobin Yvon triax spectrometer and cooled CCD camera completed the set-up.
3. Results In the cross section SEM micrograph (Fig. 1a) of the PS layers, we observe (at the top) a low porosity PS layer (lp-PS) with a flat and continuous surface and a deeper high porosity PS layer (hp-PS). The thickness of lp-PS was around 1 mm whereas one of the hp-PS was a few hundred nanometers. The interface between these two PS layers is rough and not straight. The surface morphology of hp-PS was observed after detachment of the top layer (lp-PS) after AlN growth. Its plan view SEM is presented in Fig. 1b. Highly distributed macropores with diameter up to 250 nm are clearly shown. These macropores are perpendicularly oriented to the surface and do not extend the PS surface. This observed gradual morphology, i.e. lp-PS/hp-PS is well known in PS and is related to the etch conditions [7]. The hp-PS appears lighter than the lp-PS and also than bulk Si. This contrast, probably due to its insulating behaviour that generally increases with porosity, prevents more detailed structural information such as the depth of macropores. The growth of AlN and GaN layers was in situ controlled by laser reflectometry. In this experiment we observed that the AlN growth begins by an incubation period and then the reflectivity increases in some cases and decreases in others. The study of this behaviour permitted us to directly reach to a mean porosity of the cut PS samples. The measured value was around 60% indicating a relatively good homogeneity as confirmed by SEM observations. In GaN growth phase, the reflectivity decreases due to the rough surface showing a three-dimensional growth mode. More details of reflectivity results are shown in Ref. [12]. Fig. 2 shows surface micrograph of GaN obtained on the thickest AlN (growth time was 600 s). The surface is rough and presents grains of crystallites having very varied shapes and micrometrics sizes. Their average size is estimated to be approximately 2 mm with a maximum size
Fig. 1. Cross section SEM micrograph (a) of PS layer supported by p-type (100) silicon substrate revealing a gradual PS structure formed by lp-PS/ hp-PS. Plan view SEM (b) of macroporous structure obtained after detachment of lp-PS during the AlN growth. More details are given in Fig. 5.
up to approximately 3 mm. These crystallites are randomly distributed and irregularly disoriented. However, the surface is completely covered compared to obtained morphologies by high temperature direct growth on PS [8–10]. Similar GaN structure is observed for the other AlN thickness. The X-ray diffraction (XRD) analysis showed a polycrystalline GaN that grows in merely hexagonal structure with a preferential h00.1i direction. But there were no XRD peaks corresponding to cubic GaN observed from all of our grown GaN on PS samples. The diffraction spectra were described elsewhere [12]. The percentages in volume of oriented crystallites in the hhk.li direction were estimated using the model detailed in Ref. [13] and results are given in Table 1. The contribution of h00.1i is dominant compared to the other directions. The percentage of oriented crystallites according to the c-axis
N. Chaaben et al. / Microelectronics Journal 35 (2004) 891–895
Fig. 2. Plan view SEM micrograph of GaN layer grown on PS with AlN buffer layer. The AlN growth time is 600 s.
h00.1i of the wurtzite structure increases as function of the AlN thickness. We compare the intensity of the (10.1) peak which is the strongest reflection for perfectly random polycrystals to the one of (00.2) peak. The intensity ratio of (10.1)–(00.2) peak (Fig. 3) increases from 0.1 to 0.7 as a function of AlN thickness proving that the increase of the AlN buffer thickness improves the crystalline quality and seems to adapt more GaN and PS layers. AlN buffer with appropriate thickness can limit the substrate surface perturbation due to diffusion and surfactant effect of Si under high temperature GaN growth conditions [14]. Cross section observation (Fig. 4) shows that the interface between GaN and the PS is continuous and flat. The lp-PS layer is clearly discernable and has a thickness of about 1 mm as like the original PS layer. However, as shown in this figure, lp-PS layer is detached from the Si substrate. The lp-PS layer is kept intact with flat bottom face and remains bounded to GaN layer. Under the detached lp-PS, the interface has a structure similar to that seen in the original PS/Si (Fig. 1a). Depressions on Si surface were formed to a depth of a few hundred nanometres. This depth corresponds to one of the described hp-PS layer. We conclude that the hp-PS was fractured and was more fragile than the lp-PS. This detachment may be associated to
893
Fig. 3. Intensity ratio of (10.1)–(00.2) peak evolution as a function of AlN growth time.
the stress induced in PS during the post growth-cooling phase. Generally, the PS layers obtained by standard anodisation method often show a gradient of porosity which increases in depth related to PS growth conditions where the top of PS layer has a lower porosity than the bottom. Therefore, we think that the interfacing between PS and Si substrate is more flexible than the one between AlN/ PS and GaN and can undergo most of the thermally induced deformation in the porous layer. This is confirmed when we just grow AlN (400 s) followed by annealing at 1050 8C and cooling down to room temperature (RT). In Fig. 5, we show that the AlN/PS (zone 1) layer is completely cracked revealing a second zone (zone 2) where we observe that the porous structure was taken off and a macroporous structure is clearly shown. We confirm that the thermal instability of hp-PS is responsible for the appearance of crack after AlN growth. These cracks are not observed in the final GaN layer
Table 1 Percentages of oriented crystallites in hhk.li direction calculated for GaN films grown with different AlN growth time: 200 s (c1), 400 s (c2) and 600 s (c3) 2qhkl
(hk.l)
c1 (%)
c2 (%)
c3 (%)
32.4170 34.5979 36.8746 48.1385 57.8251 63.4953 72.9847
(10.0) (00.2) (10.1) (10.2) (11.0) (10.3) (00.4)
10 45 16 8 6 15 –
11 53 16 7 3 10 –
– 80 3 3 – 4 10
Fig. 4. Cross section SEM micrograph of GaN layers grown on PS with AlN buffer layer. The AlN growth time is 400 s. The PS layer is detached from the Si substrate. The AIN layer is very thin and is not discemable in the SEM resolution.
894
N. Chaaben et al. / Microelectronics Journal 35 (2004) 891–895
Fig. 5. Surface SEM micrograph of AlN/PS annealed at 1050 8C. The AlN growth time is 400 s. The lp-PS (zone 1) is cracked and detached revealing a macroporous hp-PS layer (zone 2).
probably due to the insufficient coalescence of GaN crystallites. We show that the instability of hp-PS is responsible for detachment of lp-PS and crack appearance after AlN growth. These two phenomena, strongly dependent, are principally related to the thermal growth conditions. We have previously signalled that porous structure could remain steady under GaN growth conditions. However, thermally activated micro structural changes can appear during all GaN growth steps and can consequently affect the thermal strain level in PS layer. They can consist of PS recrystallisation and residual oxide desorption inducing atom mobility and reorganization of PS structure. These have not been revealed by our SEM observations. The RT PL measurements of GaN, given in Fig. 6a, show near band edge emission at 3.4 eV that becomes more accentuated for the thickest AlN buffer. These optical qualities are consistent with the crystalline quality improvement versus AlN thickness. Then, we studied the temperature PL dependence (5–100 K) for the best crystalline and ambient optical qualities GaN film (Fig. 6b). At low temperature, the donor bound exciton (DBE) emission at 3.470 eV dominates the PL spectrum. Two others not well pronounced peaks at 3.476 and 3.510 eV are present and separated from DBE peak by 6 and 34 meV, respectively, that can be assigned to free exciton (FE) transitions. In Fig. 6c, we plotted the integrate intensity of the DBE emission versus inverse of temperature and fitted the experimental data to the well known expression [15] IðTÞ Z
I0 ; 1 C C exp K KEBaT
where I0 and C are constant parameters, KB is the Boltzman constant and Ea is the activation energy. The fit gives Ea of about 5.8 meV that can be assigned to the dissociation
Fig. 6. PL measurements of GaN. (a) Room temperature PL for two AlN growth duration: 200 s (1) and 600 s (2). (b) Temperature dependence (5–100 K) of PL for GaN grown in the thickest AlN buffer (the insert figure give the temperature dependence of the yellow emission). (c) Integrate intensity of DBE emission versus inverse of temperature.
energy of DBE into FE [16]. The DBE quenches hardly above 50 K. At relatively low energy, two other lines at 3.44 and 3.42 eV are also observed and dominate the PL spectrum above 50 K. The position of the first peak (3.44 eV) coincides with that of the two-electron satellites of the DBE in GaN [16,17]. Although, as shown in Fig. 6b, the DBE peak quench faster than 3.44 eV peak, we consequently rule out such identification. Recent high spatial resolution PL studies demonstrated that the bright
N. Chaaben et al. / Microelectronics Journal 35 (2004) 891–895
emission at 3.45–3.46 eV originates from the inversion domain boundaries and this peak was attributed to shallow trap [18]. Reshchikov et al. [17] show that the 3.45 eV peak dominated PL spectrum for N polar GaN having a high density of inversion domain. They attributed this emission to excitons bound to inversion domain boundaries. Knowing that our GaN samples were grown under nitrogen-rich conditions that can favour N polar GaN formation, we agree the assumption made by Reshchikov that the corresponding transition is related to excitons bound to inversion domain boundaries. For the second peak at 3.42 eV, it was extensively studied and found related to excitons bound to structural and surface defects [19,20]. At about the same energy, 3.41 eV emissions have been shown in low structural quality and related to excitons bound to structural defect [20]. Kim et al. [19] have been reported that crystal structural imperfections and rough surface morphology are responsible for the luminescence band around 3.42 eV in GaN. The band position does not have fixed energy level, but changes from 3.402 to 3.428 eV depending upon the GaN crystal qualities. Reshchikov [17] also attributed the 3.42 eV emission to excitons bound to some surface defects rather than to dislocations. All the mentioned assumptions are consistent with the polycrystalline structure of our GaN samples and allow us to conclude that the 3.42 eV peak is related to excitons bound to structural and surface defects. We also notice the presence of a weaker broad yellow band emission at around 2.2 eV (see the inset figure in Fig. 6b) that does not quench with increasing temperature.
4. Conclusion The adhesion between PS and GaN is improved by using AlN buffer layer. GaN layers showed a preferential growth towards the c-axis and the alignment is enhanced with increasing AlN thickness and RT optical qualities were consistent with the crystalline quality improvement. Low temperature PL revealed excitonic transition energies below the DBE line at 3.44 and 3.42 eV related to structural and surface defects in polycrystalline GaN. The thickness of porous layer seems to be insufficient to absorb the residual thermal strain between GaN and Si. To succeed in the GaN growth, properties of PS (porosity and thickness) and growth conditions of AlN buffer (thickness and growth temperature) need to be optimized.
895
Microstructural changes in PS should be also profoundly studied versus GaN growth conditions.
Acknowledgements The authors would like to thank H. Fo¨ll (University of Kiel) for providing the PS samples.
References [1] S. Strite, H. Morkoc, GaN, AlN and InN: a review, J. Vac. Sci. Technol. B 10 (1992) 1237–1266. [2] K. Iwata, H. Asahi, K. Asami, R. Kuroiwa, S. Gonda, Jpn. J. Appl. Phys. 36 (1997) L661. [3] D.P. Bour, N.M. Nickel, C.G. Van De Walle, M.S. Kneissl, B.S. Krusor, P. Mei, N.M. Johnson, Appl. Phys. Lett. 76 (2000) 2182. [4] K. Yamada, H. Asahi, H. Tampo, Y. Imanishi, K. Ohnishi, K. Asami, IPAP Conf. Ser. 1 (2000) 556. [5] K. Yamada, H. Asahi, H. Tampo, Y. Imanishi, K. Ohnishi, K. Asami, Appl. Phys. Lett. 78 (2001) 2849. [6] H. Asahi, H. Tampo, K. Yamada, K. Ohnishi, Y. Imanishi, K. Asami, Phys. Status Solidi (a) 188 (2) (2001) 601–604. [7] C. Ja¨ger, B. Finkenberger, W. Ja¨ger, M. Christophersen, J. Carstensen, H. Fo¨ll, Mater. Sci. Eng. B 69/70 (2000) 199. [8] A. Missaoui, M. Saadoun, H. Ezzaouia, B. Bessaı¨s, T. Boufaden, A. Rebey, B. El Jani, Phys. Status Solidi (a) 182 (2000) 189. [9] A. Matoussi, T. Boufaden, A. Missaoui, S. Guermazi, B. Bessaı¨s, Y. Mlik, B. El Jani, Microelectron. J. 32 (2001) 995. [10] A. Missaoui, H. Ezzaouia, B. Bessaı¨s, T. Boufaden, A. Matoussi, M. Bouaı¨cha, B. El Jani, Mater. Sci. Eng. B 93 (2002) 102–106. [11] T. Boufaden, A. Matoussi, S. Guermazi, S. Juillaguet, A. Toureille, Y. Mlik, B. El Jani, Phys. Status Solidi (a) 201 (3) (2004) 582–587. [12] T. Boufaden, N. Chaaben, M. Christophersen, B. El Jani, Microelectron. J. 34 (2003) 843–848. [13] P. Joubert, B. Loisel, Y. Chouan, L. Haji, J. Electrochem. Soc. 134 (10) (1987) 2541. [14] H. Lahre`che, P. Venne´gue`s, O. Tottereau, M. Lau¨gt, P. Lorenzini, M. Leroux, B. Beaumont, P. Gibart, J. Cryst. Growth 217 (2000) 13. [15] G. Davies, Phys. Rep. 176 (1989) 83. [16] K. Kornitzer, T. Ebner, K. Thonke, R. Sauer, C. Kirchner, V. Schwegler, M. Kamp, M. Leszczynski, I. Grzegory, S. Porowski, Phys. Rev. B 60 (3) (1999) 1471–1473. [17] M.A. Reshchikov, D. Huang, F. Yun, P. Visconti, L. He, H. Morcoc¸, J. Jasinski, Z. Liliental-Weber, R.J. Molnar, S.S. Park, K.Y. Lee, J. Appl. Phys. 94 (9) (2003) 5623–5632. [18] P.J. Schuck, M.D. Mason, R.D. Grober, O. Ambacher, A.P. Lima, C. Miskys, R. Dimitrov, M. Stutzmann, Appl. Phys. Lett. 79 (2001) 952. [19] S.T. Kim, Y.J. Lee, S.H. Chung, D.C. Moon, Semicond. Sci. Technol. 14 (1999) 156–160. [20] S. Fischer, G. Steude, D.M. Hofmann, F. Kurth, F. Anders, M. Topf, B.K. Meyer, F. Bertram, M. Schmidt, J. Christen, L. Eckey, J. Holst, A. Hoffmann, B. Mensching, B. Rauschenbach, J. Cryst. Growth 189/190 (1998) 556–560.
