Low-temperature Surface Passivation of Moderately Doped Crystalline Silicon by Atomic-layer-deposited Hafnium Oxide Films

Transcription

1 Available online at Energy Procedia 15 (2012) International Conference on Materials for Advanced Technologies 2011, Symposium O Low-temperature Surface Passivation of Moderately Doped Crystalline Silicon by Atomic-layer-deposited Hafnium Oxide Films F. Lin a,*, B. Hoex a, Y.H. Koh a, J.J. Lin a and A.G. Aberle a,b a Solar Energy Research Institute of Singapore, National University of Singapore, 7 Engineering Drive 1, Block E3A, Singapore , Singapore b Department of Electrical and Computer Engineering, National University of Singapore, 4 Engineering Drive 3, Block E4, Singapore , Singapore Abstract Hafnium oxide (HfO 2 ) films synthesized by thermal atomic layer deposition (ALD) are investigated for lowtemperature surface passivation of moderately doped crystalline silicon (c-si). At intermediate bulk injection levels, effective surface recombination velocities of 55 cm/s and 24 cm/s are achieved on 2.1 cm p-type and 3.3 cm n- type c-si, respectively, demonstrating a good level of surface passivation. Fourier transform infrared spectroscopy and cross-sectional transmission electron microscopy experiments are conducted to provide insight into the surface passivation mechanism of HfO 2 on c-si. The good passivation quality is shown to be due to both chemical passivation and field-effect passivation. The latter is due to built-in positive charges in the HfO 2 film, which is particularly beneficial for the passivation of n-type c-si Published by Elsevier by Elsevier Ltd. Selection Ltd. Selection and/or peer-review and/or peer-review under responsibility under of responsibility the organizing of committee Solar Energy of International Research Institute Conference of Singapore on Materials (SERIS) for Advanced National Technologies. University Open of access Singapore under CC (NUS). BY-NC-ND license. Keywords: Hafnium oxide, atomic layer deposition, surface passivation 1. Introduction The passivation of electrically active surface defects ( surface passivation ) is of importance for a range of crystalline silicon (c-si) based electronic devices, and in particular for high-efficiency c-si solar * Corresponding author. Tel.: ; fax: address: lin.fen@nus.edu.sg Published by Elsevier Ltd. Selection and/or peer-review under responsibility of the organizing committee of International Conference on Materials for Advanced Technologies. Open access under CC BY-NC-ND license. doi: /j.egypro

2 F. Lin et al. / Energy Procedia 15 (2012) cells [1, 2]. The traditional surface passivation method for these solar cells is thermal oxidation at high temperature ( C). Thermal oxide provides good passivation on moderately doped n- and p-type c-si surfaces, as well as on heavily doped n-type c-si surfaces [1]. Disadvantages of thermal oxides for solar cell applications include a long processing time and a high processing temperature. Due to these process related issues, significant efforts have been spent in the past by the solar cell community on the development of low-temperature surface passivation methods for both moderately doped and heavily doped c-si surfaces. A major result of these efforts has been the development of hydrogenated amorphous silicon nitride (a-sin x :H) films deposited at low temperature ( C) by plasma-enhanced chemical vapour deposition (PECVD) [1, 3, 4]. Such films provide very good passivation on moderately doped n- and p-type c-si surfaces, as well as on heavily doped n-type c-si surfaces [1]. Since they provide far superior antireflective properties than a thermal silicon dioxide film, a large fraction of today s industrial c-si wafer solar cells features a plasma-deposited silicon nitride film on the front surface. Other low-temperature passivation schemes for moderately doped c-si that have been developed in the past include amorphous silicon carbide (a-sic x :H) [5, 6], hydrogenated amorphous silicon (a-si:h) [7] and aluminium oxide (Al 2 O 3 ) [3, 8-11]. Al 2 O 3 is a high-κ dielectric material with negligible absorption in the visible part of the solar spectrum that demonstrates an excellent level of surface passivation on n-type and p-type c-si surfaces, as well as on heavily doped p-type c-si surfaces, driven by a reasonably low interface defect density in combination with a high fixed negative charge density in the film [8-10]. Besides Al 2 O 3, hafnium oxide (HfO 2 ) is also an important high-κ dielectric material with negligible absorption in the visible spectrum [12]. In the semiconductor industry, HfO 2 has been investigated as an alternative to silicon dioxide due to its high dielectric constant, good electrical interface properties, and relatively high thermal stability. Recently, Wang et al. reported a surface recombination velocity of 73 cm/s for n-type c-si passivated by HfO 2 grown by atomic layer deposition (ALD) [13]. Similar results were also obtained on p-type c-si passivated by ALD-grown HfO 2 [14]. In this work, HfO 2 films synthesised by thermal ALD are studied for surface passivation of moderately doped n-type and p-type c-si wafers. Structural properties of the HfO 2 film before and after annealing are studied in detail to gain insight into the surface passivation mechanism of HfO 2 on c-si. 2. Experiments HfO 2 thin films were grown in a thermal ALD reactor (Genus, StrataGem-200) at a substrate temperature of 300 C, using trimethylhafnium (TMHf) and H 2 O as the precursors for hafnium and oxygen. One deposition cycle consisted of one H 2 O pulse, one nitrogen purge, one TMHf pulse, and another nitrogen purge. The total cycle time was about 8 s and the growth rate about 0.6 Å/cycle. To study the surface passivation quality, an identical film was deposited in two separate ALD runs onto each surface of double-side polished c-si substrates, resulting in a symmetrically coated carrier lifetime sample. Moderately doped p-type (280 µm, <100>, 2.1 cm) and n-type (280 µm, <100>, 3.3 cm) float-zoned (Fz) c-si substrates were used in this study. Prior to the insertion into the ALD reactor, the c- Si substrates received a standard RCA (Radio Corporation of America) cleaning process, followed by a HF dip, water rinse and drying step in gaseous nitrogen (N 2 ). This sequence ensures a H-terminated c-si surface [15]. Upon completion of the lifetime measurements, the carrier lifetime sample was annealed at 440 C in a N 2 ambient for 30 minutes and then re-measured. The effective carrier lifetime of each carrier lifetime sample was measured by the photoconductance decay method, using a calibrated lifetime tester (Sinton Consulting, WCT100) in both the quasi-steady-

3 86 F. Lin et al. / Energy Procedia 15 (2012) state and the transient mode [16]. The upper limit S eff.max of the effective surface recombination velocity is obtained by assuming an infinite bulk lifetime in the c-si substrate. It is given by [17]: W Seff. max (1) 2 eff where W is the substrate thickness, and τ eff is the effective carrier lifetime of the sample. Infrared absorption spectra were measured using a Fourier transform infrared (FTIR) spectrometer (Varian, model 3100) in the range of cm -1. Cross-sectional transmission electron microscopy (TEM) images were obtained using a JEOL 2010F system operating at 200 kv. For the TEM measurements, the as-deposited samples were prepared using a focused ion beam, whereas the annealed samples were prepared by means of manual grinding followed by ion milling. 3. Results and discussion 3.1. Electrical properties Figure 1(a) shows the injection level dependent effective carrier lifetime for the p-type c-si substrate (2.1 cm) passivated by a 10-nm and 15-nm HfO 2 thin film, respectively. For the 10-nm HfO 2 film, the effective carrier lifetime τ eff of the sample (the τ eff value is quoted at an injection level of cm -3 unless indicated otherwise) increased from 81 µs to 256 µs due to the post-deposition anneal, corresponding to a S eff.max value of 55 cm/s after annealing. The injection level dependent effective lifetime of the 15-nm HfO 2 sample is also shown in Fig. 1(a). It can be seen that the passivation quality for both HfO 2 thicknesses is quite similar, both before and after annealing. Fig. 1. Measured injection level dependence of the effective lifetimes of (a) the moderately doped p-type samples symmetrically passivated by 10 nm and 15 nm thick HfO 2 films and (b) a moderately doped n-type sample symmetrically passivated by 15 nm thick HfO 2 films. The measurements were performed before (as-deposited) and after the 440 C thermal annealing step (annealed). The corresponding results for n-type c-si substrates (3.3 cm) are shown in Fig. 1(b), for a HfO 2 film thickness of 15 nm. As can be seen, the τ eff at an injection level of cm -3 are higher than in the case of the p-type substrates (Fig. 1(a)), both before and after annealing (267 µs and 599 µs, respectively, corresponding to S eff.max values of 54 cm/s and 24 cm/s).

4 F. Lin et al. / Energy Procedia 15 (2012) In Figs. 1(a) and 1(b), the effective lifetimes decrease for injection levels above cm -3 due to Auger recombination in the bulk of the c-si substrate. The effective lifetimes of the p-type samples demonstrate a very strong injection level dependence in the as-deposited state, with the effective lifetime decreasing by about one order of magnitude when the injection level is decreased from to cm -3. This injection level dependence is much weaker for the annealed samples. For the n-type sample (Fig. 1(b)), the effective lifetime decreases only slightly with reducing injection level. The injection level dependence of the effective lifetime of HfO 2 passivated samples is quite similar to that of a-sin x :H, a-sic x :H, and thermal oxide passivated samples [1-7], indicating that it is most likely related with the positive polarity of the fixed charges in the film [18]. The passivation mechanism will be discussed later in more detail. As seen from Fig. 1(a), there is no significant difference in the effective lifetimes of p-type c-si substrates passivated with HfO 2 thin films of different thickness (10 nm and 15 nm). The improved surface passivation after the post-deposition anneal for both thicknesses could be attributed to a reduction in the electrically active defect density at the c-si/hfo 2 interface. Hydrogen from the HfO 2 bulk could diffuse to the interfacial region and provide the chemical passivation of the dangling bonds, reducing the defect related recombination rate. For as-deposited 15-nm HfO 2 films, the effective lifetime of the n-type sample (267 µs) is more than two times higher than that of the p-type sample (105 µs). The post-deposition N 2 anneal improved the effective lifetime of the n-type sample to 599 µs (S eff.max = 24 cm/s). It is noted that these two samples were fabricated using the identical HfO 2 process and given the same annealing process. It has been reported that HfO 2 films on c-si contain positive fixed charges with a density of up to elementary charges per cm 2 [14, 19, 20]. These positive built-in charges in the HfO 2 film could be attributed to oxygen vacancies, which are known to be the dominant intrinsic defect in both bulk HfO 2 and thin-film HfO 2 [21, 22]. It has been found that, in three-fold coordinated Hf sites, positively charged oxygen vacancies V 2+ and V + are energetically favourable and stable [23]. The positive fixed charges are particularly beneficial for passivating n-type c-si substrates, because they repel the bulk minority charge carriers (holes) from the c-si surface and consequently reduce the surface recombination rate. In the case of p-type substrates, the positive built-in charges in the HfO 2 film produce an inversion layer along the c-si surface. It is noted that the doping concentration of the p-type wafer is 5 times higher than that of the n-type wafer. This is one reason for the higher S eff.max of the p-type wafer, because it is well known that, for a given fixed insulator charge density, field-effect passivation becomes less effective with increasing doping concentration of the wafer [1]. As suggested in the literature [24, 25], the pronounced injection level dependence of the effective lifetime in Fig. 1(a) for HfO 2 passivated p-type c-si samples could mainly be due to the recombination losses in this surface space charge region. Shockley-Read-Hall (SRH) recombination in the bulk or at the surface could also result in a small decrease in effective lifetime at lower injection level [26, 27], as can be observed in Fig. 1(b) for HfO 2 passivated n-type c-si samples. More recently, Steingrube et al. attributed this injection dependence to a damaged region in a thin layer underneath the passivating film, which is mainly due to H-induced defects introduced during sample preparation or deposition process [28] Structural properties As the post-deposition anneal in N 2 results in a significant increase in the c-si surface passivation quality on both n- and p-type c-si, the structural changes in the c-si/hfo 2 system caused by this anneal were studied in more detail to get more insight into the surface passivation mechanism of HfO 2. Crosssectional TEM images of an as-deposited and the annealed 10-nm HfO 2 film on c-si are shown in Figs.

5 88 F. Lin et al. / Energy Procedia 15 (2012) (a) and 2(b). As can be seen, the as-deposited HfO 2 film is amorphous while the film is transformed into a polycrystalline state by the post-deposition anneal in N 2. An interfacial oxide layer can be seen between the c-si substrate and the HfO 2 thin film. This layer was determined to have a thickness of about 0.9 nm and 1.1 nm for the as-deposited and annealed sample, respectively. This interfacial oxide film was most likely formed when the H-terminated c-si surface was exposed to H 2 O during the first few ALD cycles. Fig. 2. Cross-sectional transmission electron microscope images of a p-type c-si substrate passivated with a 10 nm thick HfO 2 film, (a) before and (b) after thermal annealing at 440 C. The interfacial oxide layer is also confirmed by the presence of Si-O bonds in the infrared absorption spectra of as-deposited and annealed 10-nm HfO 2 films, see Fig. 3. The intensity of the Si-O stretching mode is found to be higher after the post-deposition anneal, which agrees well with the slightly thicker interfacial oxide film observed by TEM. The absorption band for the Si-O rocking mode overlaps with the monoclinic HfO 2 absorption band near 512 cm -1 and thus cannot be properly resolved. The absorption bands near 752, 600, 512 and 410 cm -1 (see arrows) are typical for monoclinic HfO 2 [29]. The enhancement of the monoclinic HfO 2 absorption bands due to post-deposition annealing indicates the transformation of the HfO 2 film into the monoclinic phase. This is in good agreement with the TEM observation that the HfO 2 film is polycrystalline after the post-deposition anneal. As suggested by the infrared spectra and the cross-sectional TEM images, the interfacial oxide layer thickness increases during the post-deposition anneal. This thickening of the interfacial oxide layer could result from the oxidation of the underlying Si substrate by oxygen diffusing from the HfO 2 film to the c- Si substrate, because HfO 2 thin films are known to be prone to the phenomenon of oxygen out-diffusion [22, 30]. Consequently, the positively charged oxygen vacancy density in the HfO 2 film would increase during the post-deposition anneal, resulting in a large improvement of the effective lifetime of n-type samples. In addition, it has been reported that the fixed positive charge density in HfO 2 films increases during ultraviolet (UV) radiation, owing to photon-induced electron injection [31]. Hence, the field-effect passivation by HfO 2 may be further improved by exposure to UV light. This could be beneficial for its application in solar cells. It is also noted that the as-deposited amorphous HfO 2 film is converted to a polycrystalline film during the post-deposition N 2 anneal. One may think that the grain boundaries in the polycrystalline HfO 2 may

6 F. Lin et al. / Energy Procedia 15 (2012) become diffusion paths for defects and charged carriers, and therefore compromising the effective carrier lifetime. As indicated in this work, the crystallization of the HfO 2 film does not seem to affect the level of surface passivation. Similarly, in their studies of the electrical properties of HfO 2, Lee et al. [32] and Kim et al. [33] reported that the leakage currents in amorphous and polycrystalline HfO 2 are similar. Fig. 3. Infrared absorption spectra of a p-type c-si wafer passivated by a 10 nm thick HfO 2 film, before and after thermal annealing at 440 C. The spectrum for the annealed sample is shifted upwards by for better comparison. The absorption bands typical for monoclinic HfO 2 (near 752, 600, 512, 410 cm -1 ) are indicated by arrows. The absorption spectra were obtained by dividing the infrared absorption spectrum measured on the c-si/hfo 2 sample by the infrared absorption spectrum measured on the c-si reference sample (edge region of the same wafer, which was not coated by the HfO 2 film). 4. Summary In this paper, HfO 2 thin films fabricated at low temperature with thermal ALD were shown to provide good surface passivation of moderately doped p-type and n-type crystalline silicon wafers. Effective surface recombination velocities of 55 cm/s and 24 cm/s were obtained on 2.1 cm p-type and 3.3 cm n-type c-si wafers, respectively. The good surface passivation is due to both chemical passivation and field-effect passivation. The latter results from built-in positive fixed charges, which are particularly beneficial for the passivation of n-type c-si surfaces. Acknowledgements The Solar Energy Research Institute of Singapore (SERIS) is sponsored by the National University of Singapore (NUS) and Singapore s National Research Foundation (NRF) through the Singapore Economic Development Board (EDB). References [1] Aberle AG. Surface passivation of crystalline silicon solar cells - A review. Progress in Photovoltaics 2000; 8: [2] Blakers AW, Wang A, Milne AM, Zhao J, Green MA. 22.8% efficient silicon solar cell. Appl. Phys. Lett. 1989; 55: [3] Hezel R, Jäger K. Low-temperature surface passivation of silicon for solar sells. J. Electrochem. Soc. 1989; 136: [4] Aberle AG. Overview on SiN surface passivation of crystalline silicon solar cells. Sol. En. Mat. Sol. Cells 2001; 65:

7 90 F. Lin et al. / Energy Procedia 15 (2012) [5] Martin I, Vetter M, Orpella A, Puigdollers J, Cuevas A, Alcubilla R. Surface passivation of p-type crystalline Si by plasma enhanced chemical vapor deposited amorphous SiC x:h films. Appl. Phys. Lett. 2001; 79: [6] Martin I, Vetter M, Orpella A, Voz C, Puigdollers J, Alcubilla R. Surface passivation of n-type crystalline Si by plasmaenhanced-chemical-vapor-deposited amorphous SiC x:h and amorphous SiC xn y:h films. Appl. Phys. Lett. 2002; 81: [7] Sawada T, Terada N, Tsuge S, Baba T, Takahama T, Wakisaka K, Tsuda S, Nakano S. High efficiency a-si/c-si heterojunction solar cell. Proc. 1st World Conf. on Photovoltaic Energy Conversion, Hawaii; 1994, p [8] Hoex B, Heil SBS, Langereis E, van de Sanden MCM, Kessels WMM. Ultralow surface recombination of c-si substrates passivated by plasma-assisted atomic layer deposited Al 2O 3. Appl. Phys. Lett. 2006; 89: [9] Hoex B, Schmidt J, Bock R, Altermatt PP, van de Sanden MCM, Kessels WMM. Excellent passivation of highly doped p- type Si surfaces by the negative-charge-dielectric Al 2O 3. Appl. Phys. Lett. 2007; 91: [10] Benick J, Hoex B, van de Sanden MCM, Kessels WMM, Schultz O, Glunz SW. High efficiency n-type Si solar cells on Al 2O 3-passivated boron emitters. Appl. Phys. Lett. 2008; 92: [11] Agostinelli G, Delabie A, Vitanov P, Alexieva Z, Dekkers HFW, De Wolf S, Beaucarne G. Very low surface recombination velocities on p-type silicon wafers passivated with a dielectric with fixed negative charge. Sol. Energy Mat. Sol. Cells 2006; 90: [12] Wilk GD, Wallace RM, Anthony JM. High-κ gate dielectrics: Current status and materials properties considerations. J. Appl. Phys. 2001; 89: [13] Wang J, Baroughi MF, Shanmugam M, Tarighat RS, Sivoththaman S, Paul S. Passivation of silicon surfaces using atomic layer deposited metal oxides. Mater. Res. Soc. Symp. Proc. 2009; 1153: A [14] Wang J, Baroughi MF, Bills B, Galipeau D, Samadzadeh R, Sivoththaman S. Substrate dependence of surface passivation using atomic layer deposited dielectrics. Proc. 34th IEEE Photovoltaic Specialists Conf., Philadelphia, Pennsylvania; 2009, p [15] Kern W. The evolution of silicon wafer cleaning technology. J. Electrochem. Soc. 1990; 137: [16] Sinton RA, Cuevas A. Contactless determination of current-voltage characteristics and minority-carrier lifetimes in semiconductors from quasi-steady-state photoconductance data. Appl. Phys. Lett. 1996; 69: [17] Sproul AB. Dimensionless solution of the equation describing the effect of surface recombination on carrier decay in semiconductors. J. Appl. Phys. 1994; 76: [18] Hoex B, Schmidt J, Pohl P, van de Sanden MCM, Kessels WMM. Silicon surface passivation by atomic layer deposited Al 2O 3. J. Appl. Phys.2008; 104: [19] Pereira L, Barquinha P, Fortunato E, Martins R. Low temperature processed hafnium oxide: Structural and electrical properties. Mater. Sci. Semicond. Process 2006; 9: [20] Chiou YK, Chang CH, Wu TB. Characteristics of hafnium oxide grown on silicon by atomic-layer deposition using tetrakis (ethylmethylamino) hafnium and water vapour as precursors. J. Mater. Res. 2007; 22: [21] Gavartin JL, Muñoz Ramo D, Shluger AL, Bersuker G, Lee BH. Negative oxygen vacancies in HfO 2 as charge traps in high-k stacks. Appl. Phys. Lett. 2006; 89: [22] Xiong K, Robertson J, Gibson MC, Clark SJ. Defect energy levels in HfO 2 high-dielectric-constant gate oxide. Appl. Phys. Lett. 2005; 87: [23] Foster AS, Lopez Gejo F, Shluger AL, Nieminen RM. Vacancy and interstitial defects in hafnia. Phys. Rev. B 2002; 65: [24] Glunz SW, Biro D, Rein S, Warta W. Field-effect passivation of the SiO 2-Si interface. J. Appl. Phys. 1999; 86: [25] Hoex B, Gielis JJH, van de Sanden MCM, Kessels WMM. On the c-si surface passivation mechanism by the negativecharge-dielectric Al 2O 3. J. Appl. Phys. 2008; 104: [26] Hall RN. Electron-hole recombination in germanium. Phys. Rev. 1952; 87: 387. [27] Shockley W, Read WT. Statistics of the recombinations of holes and electrons. Phys. Rev. 1952; 87: [28] Steingrube S, Altermatt PP, Zielke D, Werner F, Schmidt J, Brendel R. Reduced passivation of silicon surfaces at low injection densities caused by H-induced defects. Proc. 25 th European Photovoltaic Solar Energy Conf., Valencia, Spain; 2010, 2CV.2.9. [29] Neumayer DA, Cartier E. Materials characterization of ZrO 2 SiO 2 and HfO 2 SiO 2 binary oxides deposited by chemical solution deposition. J. Appl. Phys. 2001; 90: [30] Robertson J. High dielectric constant gate oxides for metal oxide Si transistors. Rep. Prog. Phys. 2006; 69: [31] Upadhyaya GS, Shohet JL. Comparison of the vacuum-ultraviolet radiation response of HfO 2/SiO 2/Si dielectric stacks with SiO 2/Si. Appl. Phys. Lett. 2007; 90: [32] Lee BH, Kang L, Nieh R, Qi WJ, Lee JC. Thermal stability and electrical characteristics of ultrathin hafnium oxide gate dielectric reoxidized with rapid thermal annealing. Appl. Phys. Lett. 2000; 76: [33] Kim HS, Marshall A, McIntyre PC, Saraswat KC. Crystallization kinetics and microstructure-dependent leakage current behavior of ultrathin HfO 2 dielectrics: in situ annealing studies. Appl. Phys. Lett. 2004; 84: