- Email: [email protected]
GaN Schottky barrier photodiode on Si (1 1 1) with low-temperature-grown cap layer
Journal of Alloys and Compounds 481 (2009) L15–L19
Contents lists available at ScienceDirect
Journal of Alloys and Compounds journal homepage: www.elsevier.com/locate/jallcom
Letter
GaN Schottky barrier photodiode on Si (1 1 1) with low-temperature-grown cap layer L.S. Chuah a,∗ , Z. Hassan a , H. Abu Hassan a , N.M. Ahmed b a b
School of Physics, Universiti Sains Malaysia, 11800 Penang, Malaysia School of Microelectronic Engineering, Universiti Malaysia Perlis, 02600 Perlis, Malaysia
a r t i c l e
i n f o
Article history: Received 25 September 2008 Received in revised form 26 February 2009 Accepted 26 February 2009 Available online 17 March 2009 Keywords: AlN GaN Photodiode Schottky barrier height Thermal annealing
a b s t r a c t In this work, GaN films were grown on three-inch silicon substrates by plasma-assisted molecular beam epitaxy (PAMBE) with AlN (about 200 nm) as the buffer layer. Finally, a thin AlN cap layer (50 nm) was grown on the GaN surface. Current–voltage (I–V) measurements before and after heat treatment were carried out. Different annealing temperatures (500–700 ◦ C) were investigated. Under dark condition, the Schottky barrier height (SBH) derived by the I–V method is 0.48 eV for as-deposited Ni/AlN/GaN/AlN Schottky diode. On the other hand, the effective barrier heights of 0.52, 0.55, and 0.57 eV were obtained for Schottky diodes annealed at 500, 600, and 700 ◦ C, respectively. We found that the SBHs of annealed Schottky diodes under dark and illuminated conditions were observed to be higher relative to the asdeposited Schottky diode. When annealed at 700 ◦ C, the resulting Schottky diodes show a dark current of as low as 5.05 × 10−5 A at 10 V bias, which is about two orders of magnitude lower than that of as-deposited Ni/AlN/GaN/AlN Schottky diode (2.37 × 10−3 A at 10 V bias). When the sample was under illumination condition, the change of current was significant for the annealed samples as compared to the as-deposited sample. © 2009 Elsevier B.V. All rights reserved.
1. Introduction Semiconductors of the III–V nitride group and their alloys have received much attention due to their contribution in optoelectronics. These nitrides form a continuous alloy system with direct band gaps ranging from 6.2 eV for AlN to 0.7 eV for InN [1]. Among all the III–V nitrides, gallium nitride (GaN) is the most extensively studied [2,3]. GaN normally crystallizes in the stable wurtzite structure, although it has also been observed to have the metastable zinc blende polytype when grown on a cubic substrate [4]. At room temperature, the band gap of wurtzite GaN is ∼3.4 eV, which corresponds with the blue-ultraviolet region. The common operating region for semiconductor optical devices is from the infrared to green. By extending this range into the blue, semiconductor components would be able to emit and detect the three primary colors of the visible spectrum. Also, GaN possesses large intrinsic dielectric breakdown fields, good thermal conductivity and chemical stability at elevated temperatures [5]. This makes it a desirable material for high speed, high power visible-to-UV optoelectronic devices which can operate in high temperatures and caustic environments.
∗ Corresponding author. Tel.: +60 4 6533673; fax: +60 4 6579150. E-mail addresses: [email protected] (L.S. Chuah), [email protected] (Z. Hassan). 0925-8388/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.jallcom.2009.02.151
Silicon (Si) is one of the most common elements of the earth crust and the substrates are of very low price and are available in very large size due to its mature development and large-scale production. The thermal conductivity is higher than that of sapphire and is close to that of GaN. The crystal perfection of Si is better than that of any other substrate material and it has good thermal stability under GaN epitaxial growth condition. The growth of GaN on Si enables the possibility of integrating GaN optoelectronics devices with Si-based electronics. As GaN devices are usually made from hexagonal GaN epitaxial layers, Si (1 1 1) can provide the hexagonal template for AlN deposition. According to the literature, X-ray diffraction (XRD) patterns showed that full width at half maximum (FWHM) of AlN (0 0 0 2) peak grown on Si (1 1 1) substrates was smaller than that grown on Si (1 0 0) substrates. XRD results also indicate that the preferred orientation of AlN films on Si (1 1 1) substrates is more easily controlled than those on Si (1 0 0) substrates. It can be attributed to the more matched lattice template with hexagonal structures of AlN films provided by (1 1 1) plane of silicon. Vibrational characterization by Fourier transform infrared spectroscopy (FTIR) revealed that the stress in the AlN films deposited on Si (1 1 1) substrates was also smaller than AlN films deposited on Si (1 0 0) substrates. The lattices in AlN (0 0 0 1) and Si (1 1 1) are both hexagonal, and thus Si (1 1 1) can provide matched template for AlN (0 0 0 1) plane. The lattice mismatch between these two planes is 19% (dSi (1 1 1) − dAlN (0 0 0 1) /dSi (1 1 1) , here dSi (1 1 1) refers to the Si
L16
L.S. Chuah et al. / Journal of Alloys and Compounds 481 (2009) L15–L19
Fig. 1. (a) Typical RHEED image at few monolayers of Al before AlN buffer layer growth, (b and c) show the RHEED images with 90 s and 15 min growth of AlN layers, (d) RHEED image of the GaN layer and (e) AlN cap layer.
lattice distance in Si (1 1 1) plane and dAlN (0 0 0 1) refers to the AlN lattice distance in AlN (0 0 0 1) plane). The lattice in Si (1 0 0) is square, which is unmatched with hexagonal lattice in AlN (0 0 0 1) plane. The lattice mismatch between AlN (0 0 0 1) and Si (1 0 0) is 42.7% (dSi (1 0 0) − dAlN (0 0 0 1) /dSi (1 0 0) , here dSi (1 0 0) refers to the Si lattice distance in Si (1 0 0) plane). The larger lattice mismatch between AlN (0 0 0 1) and Si (1 0 0) is a main contribution to the larger strain in the formed films. In the case of GaN layers grown on AlN buffer layer the crystalline quality is much improved because the lattice mismatch between GaN and AlN is only 2.5%. For these reasons, the Si (1 1 1) substrate was used. Aluminium nitride, AlN (wurtzite structure type) has specific physical properties like high thermal conductivity (91–190 W/(m K)), large breakdown electric field, high electrical resistivity (1011 to 1014 cm), high melting point and large energy gap [6]. Its applications as a component of refractory ceramics or buffer layers for GaN epilayers grown on sapphire are widely
known. In the last decade considerable interest arose in the use of thin films of AlN for various applications, from coatings for magneto-optic media, to thin films transducers and gigahertz-band surface acoustic wave devices. Although AlN is a wide band gap semiconductor, its properties are like those of an insulator with a high dielectric constant, good conductivity, and large breakdown electric field. Therefore, AlN may be a good insulator for blocking the leakage current. Previously, it has been shown that one can reduce gate leakage current and interface state density in GaN-based effect transistor by utilizing the AlN layer [7]. One of the most important considerations in fabricating a photodetector is achieving a low dark current condition, which is critical in producing UV photodetectors with a high signal-to-noise ratio. The effects of thermal annealing on GaN Schottky barrier photodiode with a thin AlN cap layer are investigated, mainly due to the high thermal stability of GaN that has prompted us to bring out the
L.S. Chuah et al. / Journal of Alloys and Compounds 481 (2009) L15–L19
L17
Fig. 2. XRD scan of AlN/GaN/AlN/Si. Fig. 3. XRD rocking curve (RC) of (0 0 0 2) plane for AlN/GaN/AlN/Si.
best from thermal treatment to the electrical characteristics of the UV photodetectors. Thermal annealing has been proven to be useful in reducing the dark current in Schottky photodiodes which is the subject that we are addressing in this paper. 2. Experimental details The film growth has been performed in a Veeco Gen II MBE system, using standard effusion sources for evaporation of Al and Ga and an EPI RF plasma source to supply activated nitrogen. The base pressure in the system was below 5 × 10−11 Torr. The nitrogen flux through the plasma source created a nitrogen partial pressure in the MBE chamber of about 2.0 × 10−5 Torr during growth. Growth mode and surface structures have been monitored by a 15 kV reflection high energy electron diffraction (RHEED) system. N-type Si (1 1 1) wafers (resistivity 0.02 cm) were used as the substrates. The substrates were ultrasonically degreased in solvents and etched in buffered HF prior to loading into MBE load lock chamber. In the preparation chamber, the substrates were outgassed for 10 min at 400 ◦ C prior to growth. In the growth chamber, Si substrate was heated at 750 ◦ C, and then a few monolayers of Ga was deposited on the substrate for the purpose of removing the SiO2 by the formation of GaO2 . A RHEED reconstruction with prominent Kikuchi lines is then observed, that turns into clean Si (1 1 1) surfaces at 750 ◦ C. Then a few monolayers of Al are deposited on silicon prior to an AlN buffer layer growth. A way to avoid this amorphous Six Ny layer formation is to start the growth with an AlN buffer
layer, because the bond formation between Al and N atoms prevails over the Si–N one. Then, AlN (200 nm) deposition is started by opening both Al (cell temperature at 1120 ◦ C) and N cell shutters and starting the N plasma simultaneously. The role of this layer is not only to improve the crystalline quality of the GaN layer, but also to electrically insulate the epitaxial film from the conductive substrate. Unintentionally doped GaN (800 nm) layers were grown at cell temperature of 1080 ◦ C. The samples were grown under Ga-stable conditions (Ga/N flux ratio greater than 1) that result in high quality unintentionally doped GaN growth. The GaN is unintentionally doped n-type, due to native defects commonly thought to be nitrogen vacancies [6]. Finally, thin AlN cap layer (50 nm) was grown on the GaN surface. To determine the exact orientation relationship and the content of the sample, high-resolution PANalytical X’Pert Pro MRD XRD system with a Cu-K␣1 radiation source ( = 1.5406 Å) was used. For conventional XRD phase analysis ω/2 scan, the intensity data was collected in two dimensions by performing ω (sample angle)–2 (detector angle) scan at a range of different values. For ω scan of X-ray diffractometer rocking curves (RCs), the incident beam is monochromatized and collimated by a Ge (2 2 0) four-crystal monochromator, and the diffracted beam optics is the rocking curve-triple axis PreFIX module. Prior to metallization, the samples were cleaned in boiling aqua regia (HCl:HNO3 = 3:1) for 10 min as a way to etch away surface contaminants and native oxides, then they were rinsed with distilled water and blown dry by nitrogen gas blower. After surface treatment, aluminium (100 nm) metal stripes were deposited
Fig. 4. I–V characteristics of the fabricated photodiodes annealed at different temperatures: (a) as-deposited, (b) 500 ◦ C, (c) 600 ◦ C and (d) 700 ◦ C.
L18
L.S. Chuah et al. / Journal of Alloys and Compounds 481 (2009) L15–L19
on the sides of the samples as ohmic contacts by thermal evaporation method. The samples with the ohmic contacts were then annealed under flowing nitrogen gas environment in the furnace at 400 ◦ C for 10 min. The ohmic behaviour of the contacts was confirmed by I–V measurement. Subsequent to the ohmic contacts deposition, the nickel (150 nm) was deposited as the Schottky contact metal for all of the fabricated devices. The metal mask which was used for Schottky contacts fabrication consists of an array of dots with diameter of 200 m. Both contacts were made on the top surface of the sample. The fabricated photodiodes were then annealed at temperatures from 500 to 700 ◦ C in a conventional tube furnace in flowing nitrogen environment. For sample (photodiode) annealed at 500 ◦ C, the annealing duration was 15 min, while the 600 ◦ C sample was annealed for 5 and 2 min for the 700 ◦ C sample [8]. Photocurrent and dark current of the fabricated photodiodes were then measured by Keithley high-voltage-source-measure-unit model 237-semiconductor parameter analyzer. The ideality factors (n) and effective Schottky barrier heights (SBHs) are deduced from the I–V measurement.
The I–V characteristics of the Schottky diode, ˚B and n, were determined assuming thermionic emission [10]:
I = I0 exp
Fig. 1(a) exemplifies typical RHEED images at few monolayers of aluminium. Fig. 1(b) and (c) shows the RHEED images with 90 s and 15 min of AlN buffer layer growth. After 5 min of AlN buffer growth, a stable pattern which is different from Si (1 1 1) was formed. Fig. 1(d) shows how the RHEED pattern changes during GaN growth. The AlN cap layer was grown under Al-stable conditions (Al/N flux ratio greater than 1), as shown in Fig. 1(e). Fig. 2 shows the structure of the thin films which has been determined by means of conventional XRD phase analysis ω/2 scan. Only the (0 0 0 2) and (0 0 0 4) peaks of GaN and AlN were present, along with reflections from Si (1 1 1) peak. The double-peak structure shown in the diffraction spectra is due to the K␣1 and K␣2 . In order to examine the quality of the films, ω scan of XRD RC at (0 0 0 2) plane was carried out. Fig. 3 shows the ω scan of the XRD RC of (0 0 0 2) plane for the AlN/GaN on silicon. It can be seen that one intense and sharp peak corresponding to AlN (0 0 0 2) diffraction plane is observed at 18.1◦ . The FWHM value of this peak is 26 arcmin. The structural quality of the thin film is comparable to the reported values in the literature [9]. Fig. 4 shows the I–V characteristics of the Ni/AlN/GaN/AlN Schottky barrier photodiodes annealed at different temperatures. The higher turn-on voltage observed for all measurements under dark currents could be attributed to the highly resistive nature of the AlN cap layer. The dark current was 2.37 × 10−3 A under 10 V applied bias for as-deposited Ni/AlN/GaN/AlN Schottky diode. On the other hand, for Schottky diodes annealed at 500, 600, and 700 ◦ C, the dark currents were 3.25, 4.97, and 5.05 × 10−5 A, respectively, under 10 V applied bias. By referring to Table 1, annealed samples exhibited more significant changes to the dark current characteristics compared to the as-deposited Ni/AlN/GaN/AlN Schottky diode. This finding was further confirmed by measuring the contrast ratio of photocurrent and dark current at 10 V. The contrast ratio for annealed sample (at 700 ◦ C) and as-deposited Schottky diode was found to be 25 and 2, respectively. High temperature (600 and 700 ◦ C) annealing treatment increased the barrier height as well as reduced the dark current as compared to the low-temperature annealing (500 ◦ C).
nkT
I0 = A∗ AT 2 exp
−1
(1)
q˚ B −
(2)
kT
where I0 is the saturation current, n is the ideality factor, k is Boltzmann’s constant, T is the absolute temperature, ˚B is the barrier height, A is area of the Schottky contact and A* is the effective Richardson coefficient. The theoretical value of A* can be calculated using A∗ =
3. Results and discussion
qV
4m∗ qk2 h3
(3)
where h is Planck’s constant and m* = 0.27m0 is the effective electron mass for AlN [11]. The value of A* is determined to be 32.4 A cm−2 K−2 . Using Eq. (2) and the theoretical value of A* , under dark condition, the Schottky barrier height derived by the I–V method is 0.48 eV for as-deposited Ni/AlN/GaN/AlN Schottky diode. On the other hand, the effective barrier heights of 0.52, 0.55, and 0.57 eV were obtained for Schottky diodes annealed at 500, 600, and 700 ◦ C, respectively. In our previous study, a thin AlN cap layer was incorporated in GaN Schottky diode to enhance the effective Schottky barrier height and reduces the dark current. A barrier height of 0.52 eV for normal Ni/GaN Schottky diode was increased to the effective barrier height of 0.63 eV for Ni/GaN Schottky diode with thin AlN cap layer. For GaN Schottky barriers with the thin AlN cap layer, the low dark currents might be attributed to the AlN cap layer, resulting in a higher potential barrier, as compared to conventional sample. It is also possible that the detrimental effect of interface states situated near the metal–semiconductor interface was less pronounced for sample owing to the insertion of the AlN cap layer. It was found that AlN cap layers could effectively suppress the dark current of the GaN Schottky diodes and result in improved device characteristics [12]. In the last few years, various kinds of nitride-based photodetectors have been reported [13,14]. However, an adverse situation often occurs, when large differences in the lattice constant and thermal expansion coefficient of GaN and silicon inevitably lead to high dislocation density in the GaN epitaxial layer. Such a result contributes to a large dark current and smaller photocurrent/dark current ratio for nitride-based Schottky barrier photodetectors. However, we achieved low dark current from the GaN metal–semiconductor–metal photodiode with a thin lowtemperature GaN (50 nm) barrier enhancement layer [15]. With a view to reducing the dark current of Schottky barrier photodetectors, metal–insulator–semiconductor (MIS) structures are being widely investigated. In previous reports, a number of gate dielectrics such as SiO2 have been used in MIS structures [16,17]. However, these insulators were all ex situ deposited and the contamination might occur at the insulator/semiconductor interface. Therefore, we deposited an in situ AlN cap layer by MBE in our study.
Table 1 Summary of the dark and photocurrent (I–V) characteristics of the samples annealed at different temperatures. Temperature As-deposited 500 ◦ C 600 ◦ C 700 ◦ C
Samples (photodiodes) Dark Illumination Dark Illumination Dark Illumination Dark Illumination
Ideality factor, n 1.08 1.10 1.07 1.08 1.03 1.04 1.05 1.04
Barrier height, ˚B (eV) 0.48 0.46 0.52 0.50 0.55 0.53 0.57 0.53
Current at 10 V (A) −3
2.37 × 10 4.29 × 10−3 3.25 × 10−4 1.15 × 10−2 4.97 × 10−5 1.20 × 10−3 5.05 × 10−5 2.00 × 10−3
Current at 0.1 V (A) 1.44 × 10−5 3.61 × 10−5 4.85 × 10−6 5.21 × 10−6 3.13 × 10−7 7.08 × 10−6 3.01 × 10−7 6.77 × 10−6
L.S. Chuah et al. / Journal of Alloys and Compounds 481 (2009) L15–L19
However, not much report on GaN-based Schottky barrier photodiodes capped with an AlN layer can be found in the literature, to our knowledge. Reddy et al. [18] hypothesized that the temperature dependence of Schottky barrier height may be due to the changes of surface morphology of Pt films on the n-GaN and variation of nonstoichiometric defects at the interface. Khanna et al. [19] investigated temperature dependence of W2 B5 -based rectifying contacts to n-GaN and showed that the Schottky barrier height (0.65 eV) increased with annealing temperature up to 200 ◦ C. Wang et al. [20] investigated that the variation of barrier height upon annealing may be attributed to changes of surface morphology and variation of nonstoichiometric defects at the interface vicinity. In the present work, the increase of the Schottky barrier height may be due to the reduction of nonstoichiometric defects in the metallurgical interface. The region involving the defects can be reduced due to the interdiffusion of Ni and Al. Thus, the consumption of the defect region is followed by an increase in the value of the Schottky barrier height extracted from the I–V characteristics for the samples annealed from 500 to 700 ◦ C. As generally known, chemical reaction between the metal and the semiconductor interfaces can play an important role in the electrical properties of metal/semiconductor contact. The change in the barrier height of Ni/AlN/GaN/AlN Schottky contact with annealing temperature may also be ascribed to the interfacial reaction occurring between metals and AlN and their alloys which extend to AlN films. These interfacial layers may have different work functions than the Ni/AlN contact layers, which is responsible for the increase of barrier height. From the value of the ideality factors that we have obtained, the deviation of the diodes’ ideality factor from unity may be due to tunneling effects, image force, and edge leakage. Besides, a unique value of ideality factor may be associated with a given set of diode conditions like temperature, bias, and doping. From the literature, the Schottky barrier heights varied widely from 0.50 to 1.15 eV for Ni, depending on the measuring methods, doping concentration and quality of the GaN [21,22]. This also suggests that the metal work function should not be the only factor affecting the Schottky characteristics of the diodes [22]. 4. Conclusion In summary, GaN-based Schottky photodiodes using a thin AlN cap layer were fabricated. The effect of post annealing in nitrogen ambient on the electrical properties of Ni/AlN/GaN/AlN Schottky photodiodes is studied by I–V measurement. Annealed samples exhibited more significant changes to the dark current characteristics compared to the as-deposited Ni/AlN/GaN/AlN Schottky diode.
L19
The best Schottky contact is obtained when the fabricated photodiodes was annealed at 700 ◦ C. The contrast ratio for the annealed sample (at 700 ◦ C) and as-deposited Schottky diode was found to be 25 and 2, respectively. In summary, the incorporation of the AlN cap layer and thermal annealing treatment has resulted in improved device characteristics by the enhancement of Schottky barrier height, and suppression of dark current of the fabricated Schottky photodiodes. Acknowledgments Financial support from FRGS grant and Fellowship from Universiti Sains Malaysia are gratefully acknowledged. References [1] E. Kawano, Y. Uchibori, T. Shimohara, H. Komaki, R. Katayama, K. Onabe, A. Fukuyama, T. Ikari, Jpn. J. Appl. Phys. 45 (2006) 4601. [2] V. Bondar, I. Kucharsky, B. Simkiv, L. Akselrud, V. Davydov, Yu. Dubov, S. Popovich, Phys. Status Solidi (A) 176 (1999) 329. [3] P.C. Chang, C.L. Yu, S.J. Chang, C.H. Liu, Thin Solid Films 516 (2008) 3324. [4] A. Philippe, C. Bru-Chevallier, H. Gamez-Cuatzin, G. Guillot, E. MartinezGuerrero, G. Feuillet, B. Daudin, P. Aboughe-Nze, Y. Monteil, Phys. Status Solidi (B) 216 (1999) 247. [5] C.R. Miskys, M.K. Kelly, O. Ambacher, M. Stutmann, Phys. Status Solidi (C) 6 (2003) 1627. [6] A. BenMoussa, J.F. Hochedez, R. Dahal, J. Li, J.Y. Lin, H.X. Jiang, A. Soltani, J.-C. De Jaeger, Appl. Phys. Lett. 92 (2008) 022108. [7] D.H. Cho, M. Shimizu, T. Ide, H. Ookita, H. Okumura, Jpn. J. Appl. Phys. 41 (2002) 4481. [8] Y.C. Lee, Z. Hassan, M.J. Abdullah, M.R. Hashim, K. Ibrahim, Microelectron. Eng. 81 (2005) 262. [9] A. Dagar, A. Krost, J. Christen, B. Bastek, F. Bertram, A. Kstschil, T. Hempel, J. Blasing, U. Haboeck, A. Hoffmann, J. Cryst. Growth 297 (2006) 306. [10] E.H. Rhoderick, R.H. William, Metal–Semiconductor Contacts, 2nd ed., Oxford University Press, New York, 1998. [11] J.H. Edgar, S. Strike, I. Akasaki, H. Amano, C. Wetzel, Properties: Processing and Application of Gallium Nitride and Related Semiconductor, INSPEC, London, 1999. [12] L.S. Chuah, Z. Hassan, H. Abu Hassan, F.K. Yam, C.W. Chin, S.M. Thahab, Int. J. Mod. Phys. B, in press. [13] E. Monroy, E. Munoz, F.J. Sanchez, F. Calle, E. Calleja, B. Beaumont, P. Gibart, J.A. Munoz, F. Cusso, Semicond. Sci. Technol. 13 (1998) 1042. [14] Q. Chen, J.W. Yang, A. Osinsky, S. Gangopadhyay, B. Lim, M.Z. Anwar, M. Asif Khan, D. Kuksenkov, H. Temkin, Appl. Phys. Lett. 70 (1997) 2277. [15] L.S. Chuah, Z. Hassan, H. Abu Hassan, C.W. Chin, S.M. Thahab, J. Nonlinear Opt. Phys. Mater. 17 (2008) 59. [16] H.C. Casey Jr., G.G. Fountain, R.G. Alley, B.P. Keller, S.P. DenBaars, Appl. Phys. Lett. 68 (1996) 1850. [17] B. Gaffey, L.J. Guido, X.W. Wang, T.P. Ma, IEEE Trans. Electron Dev. 48 (2001) 458. [18] V.R. Reddy, P.K. Rao, C.K. Ramesh, Mater. Sci. Eng. B 137 (2007) 200. [19] R. Khanna, S.J. Pearton, F. Ren, I. Kravchenko, Appl. Surf. Sci. 252 (2006) 5814. [20] J. Wang, D.G. Zhao, Y.P. Sun, L.H. Duan, Y.T. Wang, S.M. Zhang, H. Yang, S. Zhou, M. Wu, J. Phys. D: Appl. Phys. 36 (2003) 1018. [21] A.C. Schmitz, A.T. Ping, M. Asif Khan, Q. Chen, J.W. Yang, I. Adesida, Semicond. Sci. Technol. 11 (1996) 1464. [22] E.V. Kalinina, N.I. Kuznetsov, A.I. Babanin, V.A. Dmitriev, A.V. Shchukarev, Diam. Relat. Mater. 6 (1997) 1528.
Contents lists available at ScienceDirect
Journal of Alloys and Compounds journal homepage: www.elsevier.com/locate/jallcom
Letter
GaN Schottky barrier photodiode on Si (1 1 1) with low-temperature-grown cap layer L.S. Chuah a,∗ , Z. Hassan a , H. Abu Hassan a , N.M. Ahmed b a b
School of Physics, Universiti Sains Malaysia, 11800 Penang, Malaysia School of Microelectronic Engineering, Universiti Malaysia Perlis, 02600 Perlis, Malaysia
a r t i c l e
i n f o
Article history: Received 25 September 2008 Received in revised form 26 February 2009 Accepted 26 February 2009 Available online 17 March 2009 Keywords: AlN GaN Photodiode Schottky barrier height Thermal annealing
a b s t r a c t In this work, GaN films were grown on three-inch silicon substrates by plasma-assisted molecular beam epitaxy (PAMBE) with AlN (about 200 nm) as the buffer layer. Finally, a thin AlN cap layer (50 nm) was grown on the GaN surface. Current–voltage (I–V) measurements before and after heat treatment were carried out. Different annealing temperatures (500–700 ◦ C) were investigated. Under dark condition, the Schottky barrier height (SBH) derived by the I–V method is 0.48 eV for as-deposited Ni/AlN/GaN/AlN Schottky diode. On the other hand, the effective barrier heights of 0.52, 0.55, and 0.57 eV were obtained for Schottky diodes annealed at 500, 600, and 700 ◦ C, respectively. We found that the SBHs of annealed Schottky diodes under dark and illuminated conditions were observed to be higher relative to the asdeposited Schottky diode. When annealed at 700 ◦ C, the resulting Schottky diodes show a dark current of as low as 5.05 × 10−5 A at 10 V bias, which is about two orders of magnitude lower than that of as-deposited Ni/AlN/GaN/AlN Schottky diode (2.37 × 10−3 A at 10 V bias). When the sample was under illumination condition, the change of current was significant for the annealed samples as compared to the as-deposited sample. © 2009 Elsevier B.V. All rights reserved.
1. Introduction Semiconductors of the III–V nitride group and their alloys have received much attention due to their contribution in optoelectronics. These nitrides form a continuous alloy system with direct band gaps ranging from 6.2 eV for AlN to 0.7 eV for InN [1]. Among all the III–V nitrides, gallium nitride (GaN) is the most extensively studied [2,3]. GaN normally crystallizes in the stable wurtzite structure, although it has also been observed to have the metastable zinc blende polytype when grown on a cubic substrate [4]. At room temperature, the band gap of wurtzite GaN is ∼3.4 eV, which corresponds with the blue-ultraviolet region. The common operating region for semiconductor optical devices is from the infrared to green. By extending this range into the blue, semiconductor components would be able to emit and detect the three primary colors of the visible spectrum. Also, GaN possesses large intrinsic dielectric breakdown fields, good thermal conductivity and chemical stability at elevated temperatures [5]. This makes it a desirable material for high speed, high power visible-to-UV optoelectronic devices which can operate in high temperatures and caustic environments.
∗ Corresponding author. Tel.: +60 4 6533673; fax: +60 4 6579150. E-mail addresses: [email protected] (L.S. Chuah), [email protected] (Z. Hassan). 0925-8388/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.jallcom.2009.02.151
Silicon (Si) is one of the most common elements of the earth crust and the substrates are of very low price and are available in very large size due to its mature development and large-scale production. The thermal conductivity is higher than that of sapphire and is close to that of GaN. The crystal perfection of Si is better than that of any other substrate material and it has good thermal stability under GaN epitaxial growth condition. The growth of GaN on Si enables the possibility of integrating GaN optoelectronics devices with Si-based electronics. As GaN devices are usually made from hexagonal GaN epitaxial layers, Si (1 1 1) can provide the hexagonal template for AlN deposition. According to the literature, X-ray diffraction (XRD) patterns showed that full width at half maximum (FWHM) of AlN (0 0 0 2) peak grown on Si (1 1 1) substrates was smaller than that grown on Si (1 0 0) substrates. XRD results also indicate that the preferred orientation of AlN films on Si (1 1 1) substrates is more easily controlled than those on Si (1 0 0) substrates. It can be attributed to the more matched lattice template with hexagonal structures of AlN films provided by (1 1 1) plane of silicon. Vibrational characterization by Fourier transform infrared spectroscopy (FTIR) revealed that the stress in the AlN films deposited on Si (1 1 1) substrates was also smaller than AlN films deposited on Si (1 0 0) substrates. The lattices in AlN (0 0 0 1) and Si (1 1 1) are both hexagonal, and thus Si (1 1 1) can provide matched template for AlN (0 0 0 1) plane. The lattice mismatch between these two planes is 19% (dSi (1 1 1) − dAlN (0 0 0 1) /dSi (1 1 1) , here dSi (1 1 1) refers to the Si
L16
L.S. Chuah et al. / Journal of Alloys and Compounds 481 (2009) L15–L19
Fig. 1. (a) Typical RHEED image at few monolayers of Al before AlN buffer layer growth, (b and c) show the RHEED images with 90 s and 15 min growth of AlN layers, (d) RHEED image of the GaN layer and (e) AlN cap layer.
lattice distance in Si (1 1 1) plane and dAlN (0 0 0 1) refers to the AlN lattice distance in AlN (0 0 0 1) plane). The lattice in Si (1 0 0) is square, which is unmatched with hexagonal lattice in AlN (0 0 0 1) plane. The lattice mismatch between AlN (0 0 0 1) and Si (1 0 0) is 42.7% (dSi (1 0 0) − dAlN (0 0 0 1) /dSi (1 0 0) , here dSi (1 0 0) refers to the Si lattice distance in Si (1 0 0) plane). The larger lattice mismatch between AlN (0 0 0 1) and Si (1 0 0) is a main contribution to the larger strain in the formed films. In the case of GaN layers grown on AlN buffer layer the crystalline quality is much improved because the lattice mismatch between GaN and AlN is only 2.5%. For these reasons, the Si (1 1 1) substrate was used. Aluminium nitride, AlN (wurtzite structure type) has specific physical properties like high thermal conductivity (91–190 W/(m K)), large breakdown electric field, high electrical resistivity (1011 to 1014 cm), high melting point and large energy gap [6]. Its applications as a component of refractory ceramics or buffer layers for GaN epilayers grown on sapphire are widely
known. In the last decade considerable interest arose in the use of thin films of AlN for various applications, from coatings for magneto-optic media, to thin films transducers and gigahertz-band surface acoustic wave devices. Although AlN is a wide band gap semiconductor, its properties are like those of an insulator with a high dielectric constant, good conductivity, and large breakdown electric field. Therefore, AlN may be a good insulator for blocking the leakage current. Previously, it has been shown that one can reduce gate leakage current and interface state density in GaN-based effect transistor by utilizing the AlN layer [7]. One of the most important considerations in fabricating a photodetector is achieving a low dark current condition, which is critical in producing UV photodetectors with a high signal-to-noise ratio. The effects of thermal annealing on GaN Schottky barrier photodiode with a thin AlN cap layer are investigated, mainly due to the high thermal stability of GaN that has prompted us to bring out the
L.S. Chuah et al. / Journal of Alloys and Compounds 481 (2009) L15–L19
L17
Fig. 2. XRD scan of AlN/GaN/AlN/Si. Fig. 3. XRD rocking curve (RC) of (0 0 0 2) plane for AlN/GaN/AlN/Si.
best from thermal treatment to the electrical characteristics of the UV photodetectors. Thermal annealing has been proven to be useful in reducing the dark current in Schottky photodiodes which is the subject that we are addressing in this paper. 2. Experimental details The film growth has been performed in a Veeco Gen II MBE system, using standard effusion sources for evaporation of Al and Ga and an EPI RF plasma source to supply activated nitrogen. The base pressure in the system was below 5 × 10−11 Torr. The nitrogen flux through the plasma source created a nitrogen partial pressure in the MBE chamber of about 2.0 × 10−5 Torr during growth. Growth mode and surface structures have been monitored by a 15 kV reflection high energy electron diffraction (RHEED) system. N-type Si (1 1 1) wafers (resistivity 0.02 cm) were used as the substrates. The substrates were ultrasonically degreased in solvents and etched in buffered HF prior to loading into MBE load lock chamber. In the preparation chamber, the substrates were outgassed for 10 min at 400 ◦ C prior to growth. In the growth chamber, Si substrate was heated at 750 ◦ C, and then a few monolayers of Ga was deposited on the substrate for the purpose of removing the SiO2 by the formation of GaO2 . A RHEED reconstruction with prominent Kikuchi lines is then observed, that turns into clean Si (1 1 1) surfaces at 750 ◦ C. Then a few monolayers of Al are deposited on silicon prior to an AlN buffer layer growth. A way to avoid this amorphous Six Ny layer formation is to start the growth with an AlN buffer
layer, because the bond formation between Al and N atoms prevails over the Si–N one. Then, AlN (200 nm) deposition is started by opening both Al (cell temperature at 1120 ◦ C) and N cell shutters and starting the N plasma simultaneously. The role of this layer is not only to improve the crystalline quality of the GaN layer, but also to electrically insulate the epitaxial film from the conductive substrate. Unintentionally doped GaN (800 nm) layers were grown at cell temperature of 1080 ◦ C. The samples were grown under Ga-stable conditions (Ga/N flux ratio greater than 1) that result in high quality unintentionally doped GaN growth. The GaN is unintentionally doped n-type, due to native defects commonly thought to be nitrogen vacancies [6]. Finally, thin AlN cap layer (50 nm) was grown on the GaN surface. To determine the exact orientation relationship and the content of the sample, high-resolution PANalytical X’Pert Pro MRD XRD system with a Cu-K␣1 radiation source ( = 1.5406 Å) was used. For conventional XRD phase analysis ω/2 scan, the intensity data was collected in two dimensions by performing ω (sample angle)–2 (detector angle) scan at a range of different values. For ω scan of X-ray diffractometer rocking curves (RCs), the incident beam is monochromatized and collimated by a Ge (2 2 0) four-crystal monochromator, and the diffracted beam optics is the rocking curve-triple axis PreFIX module. Prior to metallization, the samples were cleaned in boiling aqua regia (HCl:HNO3 = 3:1) for 10 min as a way to etch away surface contaminants and native oxides, then they were rinsed with distilled water and blown dry by nitrogen gas blower. After surface treatment, aluminium (100 nm) metal stripes were deposited
Fig. 4. I–V characteristics of the fabricated photodiodes annealed at different temperatures: (a) as-deposited, (b) 500 ◦ C, (c) 600 ◦ C and (d) 700 ◦ C.
L18
L.S. Chuah et al. / Journal of Alloys and Compounds 481 (2009) L15–L19
on the sides of the samples as ohmic contacts by thermal evaporation method. The samples with the ohmic contacts were then annealed under flowing nitrogen gas environment in the furnace at 400 ◦ C for 10 min. The ohmic behaviour of the contacts was confirmed by I–V measurement. Subsequent to the ohmic contacts deposition, the nickel (150 nm) was deposited as the Schottky contact metal for all of the fabricated devices. The metal mask which was used for Schottky contacts fabrication consists of an array of dots with diameter of 200 m. Both contacts were made on the top surface of the sample. The fabricated photodiodes were then annealed at temperatures from 500 to 700 ◦ C in a conventional tube furnace in flowing nitrogen environment. For sample (photodiode) annealed at 500 ◦ C, the annealing duration was 15 min, while the 600 ◦ C sample was annealed for 5 and 2 min for the 700 ◦ C sample [8]. Photocurrent and dark current of the fabricated photodiodes were then measured by Keithley high-voltage-source-measure-unit model 237-semiconductor parameter analyzer. The ideality factors (n) and effective Schottky barrier heights (SBHs) are deduced from the I–V measurement.
The I–V characteristics of the Schottky diode, ˚B and n, were determined assuming thermionic emission [10]:
I = I0 exp
Fig. 1(a) exemplifies typical RHEED images at few monolayers of aluminium. Fig. 1(b) and (c) shows the RHEED images with 90 s and 15 min of AlN buffer layer growth. After 5 min of AlN buffer growth, a stable pattern which is different from Si (1 1 1) was formed. Fig. 1(d) shows how the RHEED pattern changes during GaN growth. The AlN cap layer was grown under Al-stable conditions (Al/N flux ratio greater than 1), as shown in Fig. 1(e). Fig. 2 shows the structure of the thin films which has been determined by means of conventional XRD phase analysis ω/2 scan. Only the (0 0 0 2) and (0 0 0 4) peaks of GaN and AlN were present, along with reflections from Si (1 1 1) peak. The double-peak structure shown in the diffraction spectra is due to the K␣1 and K␣2 . In order to examine the quality of the films, ω scan of XRD RC at (0 0 0 2) plane was carried out. Fig. 3 shows the ω scan of the XRD RC of (0 0 0 2) plane for the AlN/GaN on silicon. It can be seen that one intense and sharp peak corresponding to AlN (0 0 0 2) diffraction plane is observed at 18.1◦ . The FWHM value of this peak is 26 arcmin. The structural quality of the thin film is comparable to the reported values in the literature [9]. Fig. 4 shows the I–V characteristics of the Ni/AlN/GaN/AlN Schottky barrier photodiodes annealed at different temperatures. The higher turn-on voltage observed for all measurements under dark currents could be attributed to the highly resistive nature of the AlN cap layer. The dark current was 2.37 × 10−3 A under 10 V applied bias for as-deposited Ni/AlN/GaN/AlN Schottky diode. On the other hand, for Schottky diodes annealed at 500, 600, and 700 ◦ C, the dark currents were 3.25, 4.97, and 5.05 × 10−5 A, respectively, under 10 V applied bias. By referring to Table 1, annealed samples exhibited more significant changes to the dark current characteristics compared to the as-deposited Ni/AlN/GaN/AlN Schottky diode. This finding was further confirmed by measuring the contrast ratio of photocurrent and dark current at 10 V. The contrast ratio for annealed sample (at 700 ◦ C) and as-deposited Schottky diode was found to be 25 and 2, respectively. High temperature (600 and 700 ◦ C) annealing treatment increased the barrier height as well as reduced the dark current as compared to the low-temperature annealing (500 ◦ C).
nkT
I0 = A∗ AT 2 exp
−1
(1)
q˚ B −
(2)
kT
where I0 is the saturation current, n is the ideality factor, k is Boltzmann’s constant, T is the absolute temperature, ˚B is the barrier height, A is area of the Schottky contact and A* is the effective Richardson coefficient. The theoretical value of A* can be calculated using A∗ =
3. Results and discussion
qV
4m∗ qk2 h3
(3)
where h is Planck’s constant and m* = 0.27m0 is the effective electron mass for AlN [11]. The value of A* is determined to be 32.4 A cm−2 K−2 . Using Eq. (2) and the theoretical value of A* , under dark condition, the Schottky barrier height derived by the I–V method is 0.48 eV for as-deposited Ni/AlN/GaN/AlN Schottky diode. On the other hand, the effective barrier heights of 0.52, 0.55, and 0.57 eV were obtained for Schottky diodes annealed at 500, 600, and 700 ◦ C, respectively. In our previous study, a thin AlN cap layer was incorporated in GaN Schottky diode to enhance the effective Schottky barrier height and reduces the dark current. A barrier height of 0.52 eV for normal Ni/GaN Schottky diode was increased to the effective barrier height of 0.63 eV for Ni/GaN Schottky diode with thin AlN cap layer. For GaN Schottky barriers with the thin AlN cap layer, the low dark currents might be attributed to the AlN cap layer, resulting in a higher potential barrier, as compared to conventional sample. It is also possible that the detrimental effect of interface states situated near the metal–semiconductor interface was less pronounced for sample owing to the insertion of the AlN cap layer. It was found that AlN cap layers could effectively suppress the dark current of the GaN Schottky diodes and result in improved device characteristics [12]. In the last few years, various kinds of nitride-based photodetectors have been reported [13,14]. However, an adverse situation often occurs, when large differences in the lattice constant and thermal expansion coefficient of GaN and silicon inevitably lead to high dislocation density in the GaN epitaxial layer. Such a result contributes to a large dark current and smaller photocurrent/dark current ratio for nitride-based Schottky barrier photodetectors. However, we achieved low dark current from the GaN metal–semiconductor–metal photodiode with a thin lowtemperature GaN (50 nm) barrier enhancement layer [15]. With a view to reducing the dark current of Schottky barrier photodetectors, metal–insulator–semiconductor (MIS) structures are being widely investigated. In previous reports, a number of gate dielectrics such as SiO2 have been used in MIS structures [16,17]. However, these insulators were all ex situ deposited and the contamination might occur at the insulator/semiconductor interface. Therefore, we deposited an in situ AlN cap layer by MBE in our study.
Table 1 Summary of the dark and photocurrent (I–V) characteristics of the samples annealed at different temperatures. Temperature As-deposited 500 ◦ C 600 ◦ C 700 ◦ C
Samples (photodiodes) Dark Illumination Dark Illumination Dark Illumination Dark Illumination
Ideality factor, n 1.08 1.10 1.07 1.08 1.03 1.04 1.05 1.04
Barrier height, ˚B (eV) 0.48 0.46 0.52 0.50 0.55 0.53 0.57 0.53
Current at 10 V (A) −3
2.37 × 10 4.29 × 10−3 3.25 × 10−4 1.15 × 10−2 4.97 × 10−5 1.20 × 10−3 5.05 × 10−5 2.00 × 10−3
Current at 0.1 V (A) 1.44 × 10−5 3.61 × 10−5 4.85 × 10−6 5.21 × 10−6 3.13 × 10−7 7.08 × 10−6 3.01 × 10−7 6.77 × 10−6
L.S. Chuah et al. / Journal of Alloys and Compounds 481 (2009) L15–L19
However, not much report on GaN-based Schottky barrier photodiodes capped with an AlN layer can be found in the literature, to our knowledge. Reddy et al. [18] hypothesized that the temperature dependence of Schottky barrier height may be due to the changes of surface morphology of Pt films on the n-GaN and variation of nonstoichiometric defects at the interface. Khanna et al. [19] investigated temperature dependence of W2 B5 -based rectifying contacts to n-GaN and showed that the Schottky barrier height (0.65 eV) increased with annealing temperature up to 200 ◦ C. Wang et al. [20] investigated that the variation of barrier height upon annealing may be attributed to changes of surface morphology and variation of nonstoichiometric defects at the interface vicinity. In the present work, the increase of the Schottky barrier height may be due to the reduction of nonstoichiometric defects in the metallurgical interface. The region involving the defects can be reduced due to the interdiffusion of Ni and Al. Thus, the consumption of the defect region is followed by an increase in the value of the Schottky barrier height extracted from the I–V characteristics for the samples annealed from 500 to 700 ◦ C. As generally known, chemical reaction between the metal and the semiconductor interfaces can play an important role in the electrical properties of metal/semiconductor contact. The change in the barrier height of Ni/AlN/GaN/AlN Schottky contact with annealing temperature may also be ascribed to the interfacial reaction occurring between metals and AlN and their alloys which extend to AlN films. These interfacial layers may have different work functions than the Ni/AlN contact layers, which is responsible for the increase of barrier height. From the value of the ideality factors that we have obtained, the deviation of the diodes’ ideality factor from unity may be due to tunneling effects, image force, and edge leakage. Besides, a unique value of ideality factor may be associated with a given set of diode conditions like temperature, bias, and doping. From the literature, the Schottky barrier heights varied widely from 0.50 to 1.15 eV for Ni, depending on the measuring methods, doping concentration and quality of the GaN [21,22]. This also suggests that the metal work function should not be the only factor affecting the Schottky characteristics of the diodes [22]. 4. Conclusion In summary, GaN-based Schottky photodiodes using a thin AlN cap layer were fabricated. The effect of post annealing in nitrogen ambient on the electrical properties of Ni/AlN/GaN/AlN Schottky photodiodes is studied by I–V measurement. Annealed samples exhibited more significant changes to the dark current characteristics compared to the as-deposited Ni/AlN/GaN/AlN Schottky diode.
L19
The best Schottky contact is obtained when the fabricated photodiodes was annealed at 700 ◦ C. The contrast ratio for the annealed sample (at 700 ◦ C) and as-deposited Schottky diode was found to be 25 and 2, respectively. In summary, the incorporation of the AlN cap layer and thermal annealing treatment has resulted in improved device characteristics by the enhancement of Schottky barrier height, and suppression of dark current of the fabricated Schottky photodiodes. Acknowledgments Financial support from FRGS grant and Fellowship from Universiti Sains Malaysia are gratefully acknowledged. References [1] E. Kawano, Y. Uchibori, T. Shimohara, H. Komaki, R. Katayama, K. Onabe, A. Fukuyama, T. Ikari, Jpn. J. Appl. Phys. 45 (2006) 4601. [2] V. Bondar, I. Kucharsky, B. Simkiv, L. Akselrud, V. Davydov, Yu. Dubov, S. Popovich, Phys. Status Solidi (A) 176 (1999) 329. [3] P.C. Chang, C.L. Yu, S.J. Chang, C.H. Liu, Thin Solid Films 516 (2008) 3324. [4] A. Philippe, C. Bru-Chevallier, H. Gamez-Cuatzin, G. Guillot, E. MartinezGuerrero, G. Feuillet, B. Daudin, P. Aboughe-Nze, Y. Monteil, Phys. Status Solidi (B) 216 (1999) 247. [5] C.R. Miskys, M.K. Kelly, O. Ambacher, M. Stutmann, Phys. Status Solidi (C) 6 (2003) 1627. [6] A. BenMoussa, J.F. Hochedez, R. Dahal, J. Li, J.Y. Lin, H.X. Jiang, A. Soltani, J.-C. De Jaeger, Appl. Phys. Lett. 92 (2008) 022108. [7] D.H. Cho, M. Shimizu, T. Ide, H. Ookita, H. Okumura, Jpn. J. Appl. Phys. 41 (2002) 4481. [8] Y.C. Lee, Z. Hassan, M.J. Abdullah, M.R. Hashim, K. Ibrahim, Microelectron. Eng. 81 (2005) 262. [9] A. Dagar, A. Krost, J. Christen, B. Bastek, F. Bertram, A. Kstschil, T. Hempel, J. Blasing, U. Haboeck, A. Hoffmann, J. Cryst. Growth 297 (2006) 306. [10] E.H. Rhoderick, R.H. William, Metal–Semiconductor Contacts, 2nd ed., Oxford University Press, New York, 1998. [11] J.H. Edgar, S. Strike, I. Akasaki, H. Amano, C. Wetzel, Properties: Processing and Application of Gallium Nitride and Related Semiconductor, INSPEC, London, 1999. [12] L.S. Chuah, Z. Hassan, H. Abu Hassan, F.K. Yam, C.W. Chin, S.M. Thahab, Int. J. Mod. Phys. B, in press. [13] E. Monroy, E. Munoz, F.J. Sanchez, F. Calle, E. Calleja, B. Beaumont, P. Gibart, J.A. Munoz, F. Cusso, Semicond. Sci. Technol. 13 (1998) 1042. [14] Q. Chen, J.W. Yang, A. Osinsky, S. Gangopadhyay, B. Lim, M.Z. Anwar, M. Asif Khan, D. Kuksenkov, H. Temkin, Appl. Phys. Lett. 70 (1997) 2277. [15] L.S. Chuah, Z. Hassan, H. Abu Hassan, C.W. Chin, S.M. Thahab, J. Nonlinear Opt. Phys. Mater. 17 (2008) 59. [16] H.C. Casey Jr., G.G. Fountain, R.G. Alley, B.P. Keller, S.P. DenBaars, Appl. Phys. Lett. 68 (1996) 1850. [17] B. Gaffey, L.J. Guido, X.W. Wang, T.P. Ma, IEEE Trans. Electron Dev. 48 (2001) 458. [18] V.R. Reddy, P.K. Rao, C.K. Ramesh, Mater. Sci. Eng. B 137 (2007) 200. [19] R. Khanna, S.J. Pearton, F. Ren, I. Kravchenko, Appl. Surf. Sci. 252 (2006) 5814. [20] J. Wang, D.G. Zhao, Y.P. Sun, L.H. Duan, Y.T. Wang, S.M. Zhang, H. Yang, S. Zhou, M. Wu, J. Phys. D: Appl. Phys. 36 (2003) 1018. [21] A.C. Schmitz, A.T. Ping, M. Asif Khan, Q. Chen, J.W. Yang, I. Adesida, Semicond. Sci. Technol. 11 (1996) 1464. [22] E.V. Kalinina, N.I. Kuznetsov, A.I. Babanin, V.A. Dmitriev, A.V. Shchukarev, Diam. Relat. Mater. 6 (1997) 1528.