Effect of annealing on carrier transport properties of GaN-incorporated silicon

Abstract

GaN nanocrystals were formed in a silicon matrix by sequential implantation of Ga⁺ and N₂⁺ ions followed by either Furnace Annealing (FA) or Rapid Thermal Annealing (RTA). The formation of Ga-rich clusters and Ga–N bonds was confirmed by X-ray photoelectron and photoluminescence spectroscopy. A Schottky top contact and ohmic bottom contact were fabricated and current–voltage (I–V) characteristics of the implanted samples were studied. Current values were found to be higher in RTA samples and lower in FA samples. It is observed that the height of the Schottky barrier strongly depends on the kind of annealing and is found to be 0.65 ± 0.02 eV for FA and 0.56 ± 0.02 eV for RTA compared to the value of 0.61 ± 0.02 eV for the pristine Si sample. A carrier transport mechanism is discussed based on experimental results for both kinds of annealing.

---

Introduction

GaN nanocrystals have received much interest in recent years due to their potential applications in optoelectronic and nanoelectronic devices.¹ GaN nanostructures are grown on various substrates by several techniques such as molecular beam epitaxy, metalorganic chemical vapour deposition and chemical routes.^2–4 However, the growth of high quality GaN nanostructures is still a difficult task due to their lattice and thermal expansion coefficient mismatch with CMOS compatible materials like Si.⁵ The biggest challenge is to grow the embedded GaN nanocrystals in a Si matrix. Ion implantation is an appropriate method, by which the embedded nanostructures can be formed inside the matrix. Ion implantation technique is compatible with standard Si technology and can be used to modify optical and electrical characteristics of semiconductor materials. Unlike the conventional thin films or nanocrystals on a substrate, this method provides encapsulation and passivation, as the nanocrystals are formed directly in a substrate matrix. Another advantage of this technique is the ability to control the concentration, size and localization of the implanted species, which allows tailoring optical, electrical and magnetic properties. Sequential implantation of individual elements creates metastable supersaturated solid solution of individual elements and their compounds in the near-surface region of the matrix. Thermal annealing then causes precipitation, chemical bonding and the formation of nanocrystals or nanoclusters, the size and distribution of which can be controlled by the implantation and/or annealing conditions. The precipitation and compound formation depend on the solubility of implanted elements in the substrate and chemical affinity between the elements. Ion implantation followed by thermal annealing is used, for instance, to synthesize binary and ternary metal silicide in silicon matrices.^6,7 Metal nanocrystals (such as Au, Cu, and Pt) also synthesized in silicon matrices by using ion implantation and subsequent annealing.^8,9 The synthesis of elemental and compound semiconductor nanocrystals by using ion implantation technique in Si, SiO₂, Al₂O₃ and other matrices have been also demonstrated.^10,11 GaN nanocrystals embedded in CMOS compatible substrates/matrices exhibit great potential for novel photonics and memory devices. The synthesis of GaN nanocrystals has been reported in previous studies under the high dose ion implantation of nitrogen into GaP¹² or GaAs¹³ and sequential ion implantation of Ga and N into dielectric matrices.^14,15 However, unlike other substrates/matrices, GaN nanocrystals formation in silicon matrix and the effect of implantation and annealing on the transport characteristics of GaN-incorporated Si are not widely studied. The implantation of ions induces the formation of localized defects and amorphization of a subsurface layer. As a result of high-temperature annealing, the point defects are transformed to extended defects,¹⁶ and the recrystallized layer (especially at high doses required for ion synthesis) becomes rich of structural imperfections such as precipitates and grains. Moreover, the ion synthesis of GaN in silicon can be accompanied by the formation of SiN_x layer due to the chemical reaction of implanted nitrogen with silicon. All this inevitably influences the electrical properties of implanted layer and, as a consequence, the properties of metal/Si contact, which is a part of device fabricated on the basis of ion-synthesized nanostructures. It is therefore important to investigate the carrier transport in the implanted Si layers and to optimize the implantation/annealing conditions.

In this work, we report the results of such investigation for the GaN-containing layers in Si matrix synthesized by sequential implantation of Ga⁺ and N₂⁺ ions and discuss the influence of different kinds of thermal annealing – Furnace Annealing (FA) and Rapid Thermal Annealing (RTA) on carrier transport characteristics of Schottky contacts to implanted surface.

Experimental details

To synthesize GaN nanostructures, the sequential implantation of Ga⁺ (80 keV) and N₂⁺ (40 keV) ions was carried out at room temperature on the ILU-200 ion implanter. The scanning ion beam was used with current density kept lower than 3 μA cm⁻². The n-Si substrate (4.5 Ω cm, (100) orientation) was used as a target material. The trimethylgallium and N₂ were used as ion source materials. The molecular N₂ ions used for the implantation splits into two N atoms with equal energies at the impact with the target surface and this allow reducing the implantation time budget. The doses of N₂⁺ and Ga⁺ implantation were 2.5 × 10¹⁶ cm⁻² and 5 × 10¹⁶ cm⁻², respectively. The good overlap of Ga and N concentration profiles in the as-implanted Si substrate was predicted by the SRIM calculation.¹⁷ The projected ranges of Ga⁺ and N⁺ at the given energies in Si are about 59 and 56 nm, respectively. Thermal annealing was carried out in either (a) tube furnace for 30 minutes under N₂ atmosphere (FA sample) or (b) RTA under Ar atmosphere for 20 seconds (RTA sample). The JIPELEC JetFirst 100 system was used for RTA treatment. The implantation and annealing sequences are shown in Table 1.

Table 1 Conditions of ion implantation and implantation/annealing sequences

Samples	Ion	Dose, cm^−2	Energy, keV	Implantation and annealing sequence
FA	Ga^+	5 × 10^16	80	Ga^+ → FA 800 °C → N2^+ → FA 800 °C
	N2^+	2.5 × 10^16	40
RTA	Ga^+	5 × 10^16	80	Ga^+ → FA 800 °C → N2^+ → RTA 800 °C
	N2^+	2.5 × 10^16	40

---

Investigation of phase and chemical composition of implanted Si layers was carried out by using the ultrahigh-vacuum Omicron Multiprobe RM system with the pressure lower than 10⁻¹⁰ mbar. Photoelectron emission was excited by the X-ray radiation at the MgK_α (1253.6 eV) characteristic line. The energy spectrum of secondary electrons was scanned with a semispherical energy analyzer with a radius of 125 mm. The diameter of the region of collection of secondary electrons from the sample surface was 4 mm. The energy analyzer operated in the constant-analyzer energy (constant transmission energy – 50 eV) mode, with an absolute resolution better than 0.3 eV. Photoelectrons were detected with a set of detectors consisting of five secondary-electron multipliers. The charge shift of the XPS lines was verified by comparing the energy position of five or six peaks lying in different spectral regions of the position of reference lines. The fine structure of the XPS lines that arises from chemical shifts (binding energies) was analyzed with the mathematical software Spectral Data Processor v. 4.3 applied for spectral data processing. The photoluminescence (PL) emission spectra of the implanted samples were recorded at room temperature using an Acton SP-150 grating monochromator and a Hamamatsu R-928 photomultiplier tube in the spectral range of 350–900 nm. The blue LED was used as an excitation source at a wavelength of 385 nm. The top Schottky electrodes were formed by thermal evaporation of circular Au contact of diameter 400 μm on the surface of implanted layer. The bottom ohmic contacts of large area were formed by thermal evaporation of Al on the back side of silicon substrate followed by 30 min annealing at 300 °C. The current–voltage (I–V) measurements were carried out on the probe station with Keithley 4200 semiconductor characterization system.

Result and discussion

The XPS measurements were performed with MgK_α radiation to excite photoelectrons from Ga 2p and N 1s core levels.¹⁸ Fig. 1 shows typical XPS spectra of implanted sample after FA etched to the depth of 70 nm. The peak at binding energy of 1117.3 eV corresponds to the Ga 2p core level of elemental Ga (Ga⁰) and the peak at 1118.9 eV to Ga bonded with N (Ga–N) (Fig. 1a). The spectrum shown in Fig. 1b corresponds to the N 1s core level with the peaks at 397.4 eV and 397.9 eV, which confirm the presence of GaN as well as the formation of SiN_x phase. The GaN phase can be present in the Ga-rich nanoclusters previously detected by electron microscopy in the similar Ga and N co-implanted Si layers.¹⁹ The implanted layer was found to be recrystallized after annealing and contain large heterogeneities and extended defects at the depths around the projected range of implanted ions. Unfortunately, the used XPS technique cannot give sufficiently precise quantitative information about different bonding states of the implanted nitrogen and catch the difference between the RTA and FA annealed samples. Fig. 2 shows the comparison of PL spectra of implanted Si samples after FA and RTA. The broad emission peak centered at 530 nm is not typical of silicon matrix and can be associated with GaN (the reference samples implanted only with nitrogen do not possess visible PL). Indeed, the transition between the conduction band and the deep level attributed to the vacancy in GaN²⁰ can be the source of such visible luminescence peak. The current–voltage characteristics of GaN implanted Si samples were measured using Keithley 4200SCS. Fig. 3a shows semi-log I–V characteristics of Au/p-Si diode and GaN-incorporated Si substrate after FA and RTA treatment. High recombination current is observed in RTA sample compared to FA sample. The three distinct regions on I–V curves can be seen in Fig. 3b. A linear ohmic dependence I ∝ V is observed in the region of V < 0.4 V (region I). The current increases exponentially with voltage in the region II. In the region III (V > 1.25 V), the current increases with voltage obeying the power law relation I ∝ Vⁿ, where n = 2.85 for FA and n = 1.38 for RTA. The obtained results can be interpreted in the following manner. According to the XPS data (Fig. 1), the implanted Si layers contain Ga and GaN nanoclusters after FA and RTA. In addition, the nonstoichiometric SiN_x layer is evidently synthesized in view of the presence of high concentration of Si–N bonds. This layer is less expressed in the case of RTA compared to FA because of the shorter time period for the reaction Si + N → SiN_x. Thus, the circuit of current flow through the system “Au – implanted layer – silicon substrate” consists in general of series-connected resistances including Schottky barrier at the Au/Si:N interface, SiN_x layer and Si substrate (the resistance of Si substrate can be neglected in the first approximation). When low and medium voltages are applied, a substantial contribution to the impedance makes the Au/Si contact, the current transport through which can be described by the Schottky relation eqn (1):where V is the voltage drop on Schottky barrier, φ_B is the Schottky barrier height, I_S is the reverse saturation current, A is the junction area, A* is the Richardson constant, η is the ideality factor, T is the temperature. At low voltages, eqn (1) gives the proportionality between the current and voltage (Fig. 4a), from which the Schottky barrier height and ideality factor can be calculated. In Fig. 4b, we can see an increased barrier height for the FA sample (φ_B = 0.65 eV), whereas the barrier height is lower for the RTA sample (φ_B = 0.56 eV) compared to the value for pristine Si (φ_B = 0.61 eV). The ideality factor is around 4.5 for Si, FA and RTA samples. The measured reverse saturation currents are 4.7 × 10⁻⁸ A, 1.26 × 10⁻⁶ A and 2.04 × 10⁻⁸ A for FA, RTA and pristine Si samples, respectively.

Fig. 1 XPS spectra of GaN-incorporated FA sample showing (a) Ga 2p and (b) N 1s peaks.

Fig. 2 PL spectra of GaN-incorporated FA and RTA samples.

Fig. 3 (a) Semi-log I–V characteristics of Au/p-Si diode and GaN-incorporated Si substrate after FA and RTA. (b) Experimental and fitted I–V characteristics of GaN-incorporated FA and RTA samples with top Au and bottom Al contacts.

Fig. 4 (a) Experimental and fitting curves of log(I)–V of FA, RTA and pristine Si. (b) Barrier heights for the pristine Si, GaN-incorporated FA and RTA samples.

The ion damage is expected to be repaired by the first annealing and reamorphization of silicon takes place on subsequent implantation. The final recrystallization by FA or RTA should determine the transport characteristics of the sample. In general, thermal annealing is expected to change the defects distribution²¹ and to repair the ion induced damage.¹⁶ Although the concentration of radiation defects is decreased, a number of defects can exist near surface after annealing. The Coulomb interaction between charged defects results in a decrease in Schottky barrier.²² It is naturally to assume that in the RTA sample, due to a short annealing duration, the concentration of defects is higher and the tendency of the barrier height decrease is more expressed compared to the FA sample. The increased value of φ_B in the FA sample compared to pristine Si is apparently due to the presence of a larger number of Si–N bonds, which leads to an increase in the band gap (the band structure of the material is approaching to that of silicon nitride). Currently, it is not clear whether it is possible to consider the implanted layer adjacent to Au contact as a supersaturated solid solution of nitrogen in silicon or as a layer with a number of SiN complexes or nonstoichiometric inclusions. Irrespective of this, the Ga and N co-implanted layer can be considered in a first approximation as a Ga and N-incorporated material of increased effective band gap, which results in a higher value of energetic barrier for electron to transfer from the Fermi level in Au to the conduction band in the underlying material.

As it follows from Fig. 3, at higher voltages, the I–V dependence deviates from the Schottky mechanism for both kinds of annealing and is described by the power low, which is characteristic of the space charge limited conduction (SCLC).²³ The transition from Schottky to SCLC mechanism is correlated with the fact that, with increasing voltage, the voltage drop on the Schottky barrier becomes lower than that on the SiN_x layer, which becomes dominating in the total sample resistance. The SCLC mechanism is typical of many dielectrics²⁴ and differs in the value of power n, which is either equal to 2 (trap-free SCLC) or higher than 2 (trap-mediated SCLC).²³ As discussed before, the SiN_x layer probably is thicker in the FA sample resulting in the higher probability of electron trapping into the trap cites. That is why the trap-mediated mechanism governs the conduction in the FA sample (n ≈ 2.85), whereas the RTA sample reveals the trap-free SCLC (n ≈ 1.38).

Based on the I–V characteristics, we show schematically in Fig. 5 the formation model of GaN-incorporated layers in Si by ion implantation followed by thermal annealing. In step (a), Ga⁺ ions are implanted and the sample is annealed in furnace at 800 °C for 30 minutes. N₂⁺ ions are implanted later and the sample annealed in furnace (FA) for 30 minutes at 800 °C (b) or RTA for 20 seconds at 800 °C (c). The SiN_x region formed after FA is thicker due to the duration of annealing which provides enough time for many implanted N atoms to react with Si. The Ga-rich nanocrystals/nanoclusters form within the SiN_x dielectric medium which gives rise to the trap-mediated SCLC mechanism as shown before. In the RTA sample, the trap-free SCLC is realized due to a thinner SiN_x dielectric layer, whereas the trap-mediated SCLC is characteristic for the FA sample.

Fig. 5 The model of structure of implanted layer for GaN-incorporated FA and RTA samples.

The PL intensity is affected by the defects, which are located within or on the surface of GaN nanoclusters, whereas the transport properties of Schottky barriers are affected by all defects located near the metal-semiconductor interface. Our PL and I–V data indicate that the concentration of defects of the first type is approximately the same for both kinds of annealing, whereas the concentrations of defects located outside the GaN nanoclusters are different.

Conclusions

We have investigated the current transport characteristics of GaN-incorporated silicon layers by realizing Schottky diode. The formation of Ga or GaN nanoclusters is confirmed by XPS and PL studies. We found that the carrier transport depends on annealing kinds (FA and RTA). The I–V characteristics reveal the Schottky-like behavior at low voltages and the SCLC mechanism at high voltages. The barrier height values are extracted as 0.65 eV and 0.56 eV for the FA and RTA samples, respectively. The SCLC transport is favored by the formation of SiN_x layer in series with the Schottky barrier. For the FA sample, the SiN_x is thicker than for the RTA sample resulting in different power values in the I ∝ Vⁿ SCLC dependence (n ≈ 2.85 instead of n ≈ 2).

Acknowledgements

The study is supported by the Ministry of Education and Science of the Russian Federation (Project identifier RFMEFI58414X0008) and the Department of Science and Technology, India (Project No. INT/RUS/RMES/P-04/2014).

References

1. P. Dimitrakis, P. Normand, C. Bonafos, E. Papadomanolaki and E. Iliopoulos, Appl. Phys. Lett., 2015, 102, 053117.
2. T. J. Goodwin, V. J. Leppert, S. H. Risbud, I. M. Kennedy and H. W. H. Lee, Appl. Phys. Lett., 1997, 70, 3122.
3. Y. Xie, Y. Qian, W. Wang, S. Zhang and Y. Zhang, Science, 1996, 272, 1926.
4. T. Ogi, Y. Kaihatsu, F. Iskandar, E. Tanabe and K. Okuyama, Adv. Powder Technol., 2009, 20, 29.
5. M. Kumar, B. Roul, T. N. Bhat, M. K. Rajpalke, P. Misra, L. M. Kukreja, N. Sinha, A. T. Kalghatgi and S. B. Krupanidhi, Mater. Res. Bull., 2010, 45, 1581.
6. A. E. White, K. T. Short, R. C. Dynes, J. P. Garno and J. M. Gibson, Appl. Phys. Lett., 1987, 50, 95.
7. Z. Tan, F. Namavar, S. M. Heald and J. I. Budnick, Appl. Phys. Lett., 1993, 63, 791.
8. P. Kluth, B. Johannessen, L. L. Araujo and M. C. Ridgway, AIP Conf. Proc., 2007, 882, 731.
9. R. Giulian, P. Kluth, L. L. Araujo, D. J. Llewellyn and M. C. Ridgway, Appl. Phys. Lett., 2007, 91, 093115.
10. C. W. White, J. D. Budai, J. G. Zhu, S. P. Withrow, R. A. Zhur, Y. Chen, D. M. Hembree Jr, R. H. Magruder and D. O. Henderson, Mater. Res. Soc. Symp. Proc., 1994, 358, 169.
11. E. Borsella, C. de Julian Fernandez, M. A. Garcıa, G. Mattei, C. Maurizio, P. Mazzoldi, S. Padovani, C. Sada, G. Battaglin, E. Cattaruzza, F. Gonella, A. Quaranta, F. D'Acapito, M. A. Tagliente and L. Tapfer, Nucl. Instrum. Methods Phys. Res., Sect. B, 2002, 191, 447.
12. K. Kuriyama, H. Kondo, N. Hayashi, M. Ogura, M. Hasegawa, N. Kobayashi, Y. Takahashi and S. Watanabe, Appl. Phys. Lett., 2001, 79, 2546.
13. X. W. Lin, M. Behar, R. Maltez, W. Swider, Z. Liliental-Weber and J. Washburn, Appl. Phys. Lett., 1995, 67, 2699.
14. J. A. Wolk, K. M. Yu, E. D. Bourret-Courchesne and E. Johnson, Appl. Phys. Lett., 1997, 70, 2268.
15. E. Borsella, M. A. Garcia, G. Mattei, C. Maurizio, P. Mazzoldi, E. Cattaruzza, F. Gonella, G. Battaglin, A. Quaranta and F. D'Acapito, J. Appl. Phys., 2001, 90, 4467.
16. J. F. Gibbons, Ion implantation in semiconductors – part II: damage production and annealing, Proc. IEEE, 1972, 60, 1062.
17. http://www.srim.org.
18. N. Elkashef, R. S. Srinivasa, S. Major, S. C. Sabharwal and K. P. Muthe, Thin Solid Films, 1998, 333, 9.
19. D. S. Korolev, A. N. Mikhaylov, A. I. Belov, V. K. Vasiliev, D. V. Guseinov, E. V. Okulich, A. A. Shemukhin, S. I. Surodin, D. E. Nikolitchev, A. V. Nezhdanov, A. V. Pirogov, D. A. Pavlov and D. I. Tetelbaum, Semiconductors, 2016, 50, 271.
20. S. Limpijumnong and C. G. Van de Walle, Phys. Rev. B: Condens. Matter Mater. Phys., 2004, 69, 035207.
21. T. N. Morgan, Phys. Rev., 1965, 139, A343.
22. K. M. Kim, B. J. Choi, M. H. Lee, G. H. Kim, S. J. Song, J. Y. Seok, J. H. Yoon, S. Han and C. S. Hwang, Nanotechnology, 2011, 22, 254010.
23. M. A. Lampert and P. Mark, Current Injection in Solids, Academic Press, New York/London, 1970.
24. K. C. Kao and W. Hwang, Electrical Transport in Solids, Pergamon press, Oxford, New York, 1981.

---