Here we demonstrate that, when optimally applied, extrinsic passivation can produce .... It is clear that field effect passivation plays a vital role in enhancing cell ...
Available online at www.sciencedirect.com
ScienceDirect Energy Procedia 77 (2015) 774 – 778
5th International Conference on Silicon Photovoltaics, SiliconPV 2015
Extrinsic passivation of silicon surfaces for solar cells Ruy S Bonillaa*,Christian Reichelb, Martin Hermleb, George Martinsa and Peter R Wilshawa a Department of Materials, University of Oxford, Parks Rd, Oxford, OX1 3PH, United Kingdom b Fraunhofer Institute for Solar Energy Systems ISE, Heidenhofstr. 2, Freiburg, 79110, Germany
Abstract In the present work we study the extent to which extrinsic chemical and field effect passivation can improve the overall electrical passivation quality of silicon dioxide on silicon. Here we demonstrate that, when optimally applied, extrinsic passivation can produce surface recombination velocities below 1.2 cm/s in planar 1 ȳcm n-type Si. This is largely due to the additional field effect passivation component which reduces the recombination velocity below 2.13 cm/s. On textured surfaces field effect passivation has a comparable effect, and surface recombination velocities below 32 cm/s are achieved when field effect passivation is optimised extrinsically. Modelling of interdigitated back contact cells showed an absolute increase in efficiency of 0.1 % when passivation was optimised via extrinsic dielectric charge. These results point to the importance extrinsic passivation will have for future dielectric coatings used in solar cell manufacture. © 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license © 2015 The Authors. Published by Elsevier Ltd. (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer review by the scientific conference committee of SiliconPV 2015 under responsibility of PSE AG. Peer review by the scientific conference committee of SiliconPV 2015 under responsibility of PSE AG
Keywords: Silicon PV, dielectrics, surface passivation, field effect.
1. Introduction Reducing surface recombination is a crucial factor in achieving highly efficient silicon solar cells. Silicon surface passivation is usually performed by a dielectric film. The passivation produced by the as-deposited layer (so called intrinsic passivation) is rarely optimal due to film non-uniformities, edge effects and a lack of reproducibility in deposition equipment [1]. But just as importantly, effective passivation requires both a reduction in the density of surface states (chemical passivation) and the presence of an electric field to repel one of the carrier types away from the surface states (field effect passivation) and these two types of passivation can rarely be optimised in a single deposition process. However, it has been shown that extrinsic passivation, performed to the dielectrics after deposition, is able to achieve the highest passivation quality of any dielectric layer [2]. Forming gas and hydrogen-
1876-6102 © 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer review by the scientific conference committee of SiliconPV 2015 under responsibility of PSE AG doi:10.1016/j.egypro.2015.07.109
775
Ruy S. Bonilla et al. / Energy Procedia 77 (2015) 774 – 778
release anneals are good examples of extrinsic chemical passivation processes whilst extrinsic field effect passivation (FEP) has been applied using a corona discharge method. To date extrinsic FEP has only been used in a research environment due to its lack of stability with respect to time. Only recently has extrinsic field effect passivation been reported stable by using new charging methods [3]. This is an important result because, for the first time, it opens the opportunity of applying extrinsic FEP to practical cells which require stability over periods of decades. In the present work we demonstrate the potential of extrinsic field effect passivation when it is optimally applied to a silicon surface. Corona discharge is used to demonstrate remarkably low SRVs using very simple and cost-effective methods. The potential of optimised FEP via extrinsic charge is evaluated using modelled interdigitated back contact cells (IBC), and the improvements in efficiency which result appear to be worthwhile. 2. Experimental methods A point-to-plane corona discharge apparatus was designed to apply extrinsic charge on to silicon dioxide films on silicon. The corona discharge apparatus deposited positive ions with a uniformity better than 5% across a 3 x 3 cm2,1 ȳcm, n-type 200 ߤm thick Float Zone (FZ) silicon sample, dry oxidized to a thickness of 100 nm. The point-tosample distance to achieve uniformity better than 5% was calculated using a model of corona current empirically determined by Warburg in 1923 [4]. Figure 1.a illustrates the layout of the apparatus and the calculated current distribution. Charge was applied to thermally oxidised silicon specimens in sequential steps and the surface potential was recorded after each step using a Kelvin Probe (KP). Figure 1.b illustrates the surface potential of oxidised n-type Si samples with a 100 nm SiO2 film, when corona charged. When charge is deposited on SiO2, an electric field is generated across the film and conduction through the film is activated. A leakage regime is first seen at low field strengths of ~5 MV/cm, followed by a Fowler-Nordheim tunnelling regime that increases exponentially with field of up to ~10 MV/cm, before runaway conduction takes place and the dielectric breaks down [5]. The low field leakage regime has typical current densities in the order of 10-12-10-9 A/cm2 which can leak the corona ions from the surface of the film [5]. This leakage is observed when the field strength is increased beyond ~5 MV/cm and results in a saturation of surface potentials of around 30 V. Oxide films were charged to half of this value to avoid damage of the dielectric yet other reports have shown that such high potentials are possible using corona [6]. The surface potential for three typical samples is shown in Figure 1.b. Charge uniformity was evaluated by mapping surface potential in the KP instrument as illustrated in Figure 1.c. Here the uniformity of the electric field produced at the surface of the dielectric is better than 5%. This optimised extrinsic FEP method has also been applied to IPA/KOH textured n-type Si to reduce surface recombination. The surface recombination velocity has been evaluated by measuring effective lifetime via transient and quasi steady state photo-conductance using a Sinton WCT120 instrument [7]. Upper limits of SRV were calculated by subtracting the Auger and radiative components of recombination using Richter’s parameterisation [8].
Kelvin Probe surface potential [V]
(c)
(b)
20 15 10 5 0 0
20
40
60
80
100
Corona charging time [s]
Sample 1 Sample 2 Sample 3 120 140
4 mm 2
2.5
3
3.5 12 x 10 e/cm2
Fig. 1. a. Scheme of corona discharge set-up, b. KP surface potential as a function of corona discharge time, c. Surface charge map calculated from KP surface potential of uniformly deposited corona charge.
776
Ruy S. Bonilla et al. / Energy Procedia 77 (2015) 774 – 778
3. Optimal extrinsic field effect passivation 3.1. Extrinsic FEP on planar c-Si A dry thermal oxide grown on FZ 1ȳcm planar n-Si was deposited with corona charge. The bulk lifetime in float zone silicon is mainly limited by Auger and radiative recombination processes. Thus, the effective lifetime is a direct measurement of the surface recombination velocity (SRV). Figure 2 illustrates the change in effective lifetime of a 3 x 3 cm2 sample when deposited with charge to a surface potential of 10.65 V. The effective lifetime was seen to increase from 130 ߤs (moderate SiO2 chemical passivation) to 2.87 ms as a result of this extrinsic FEP. This is equivalent to a SRV as low as 2.13 cm/s. This is lower than the SRV of 2.4 cm/s reported for the so called ‘alneal’ SiO2 on silicon [9], and the 6 cm/s quoted value for high quality Si3N4 passivation films [10]. Our set up, however, has improved the field effect passivation component using extrinsic and very low-cost methods. If extrinsic chemical passivation is also applied by performing a 425 oC 25 minutes forming gas anneal (FGA), blue trace in Figure 2, the recombination activity is further reduced such that the effective lifetime increases to 3.8 ms and the SRV decreases to 1.28 cm/s. This value is as good as that achieved using atomic layer deposited (ALD) aluminium oxide. ALD Al2O3 produces SRV~1.3 cm/s [8] yet the cost of achieving such high quality film has been a major draw-back for the commercial implementation of this technology. 3.2. Extrinsic FEP on textured c-Si Improvements in light absorption in silicon solar cells are achieved by texturing the cell’s front surface. Here, a KOH/IPA solution was used to texture both sides of FZ silicon wafers prior oxidation. The oxide films are thus grown on surfaces rather than . The former has 15% more atoms/cm2 and 1.5 times as many available bonds/cm2. This makes a surface more prone to recombination, and leads to poorer performance. Corona charge was applied to this sample such that the KP surface potential was 16.7 V and effective lifetime increased from 17 ߤs (moderate SiO2 chemical passivation) to 300 ߤs, decreasing the SRV to 32 cm/s, red trace in Figure 2. This is a realistic value for the recombination that might be achieved in a textured solar cell when subjected to FEP. Although the SRV obtained in this case is larger than for a planar surface, the relative decrease due to extrinsic FEP is similar. This indicates that extrinsic FEP has an equivalent effect when applied to textured surfaces.
-3
10
-4
Weff [s]
10
Planar Planar+Corona Planar+FGA Planar+FGA+Corona Textured Textured+Corona
FZ n-type Si 1 : cm Dry SiO 2
-5
10 13 10
10
14
10
-3
15
10
16
' p [cm ] Fig. 2. Improvements in effective lifetime as a product of optimal extrinsic FEP for planar and textured silicon surfaces.
777
Ruy S. Bonilla et al. / Energy Procedia 77 (2015) 774 – 778
4. Effect of extrinsic FEP on simulated IBC solar cells Interdigitated back-contact (IBC) solar cells have shown remarkable performance both at research and industrial scale. PC2D [11] was used here to characterise the effect of FEP on the overall cell performance. A model IBC cell was used to calculate efficiency as a function of front surface dielectric charge. The simulation used a 160 ߤm thick unit cell with 20 x 20 elements. It consisted of a p+ emitter 375 ߤm wide, and an n+ BSF 25 ߤm wide on the rear side of the cell. This geometry reflected to the left and right of the unit cell to adjust for boundary conditions. The base doping concentration was 5x1015 cm-3 donors. Table 1 illustrates all the parameters involved in the modelling process. PC2D uses the dark saturation current J01 instead of SRV to account for surface recombination. This is due to a doped front layer which can be included to account for a front surface field. For this work no front surface field was included in the cell since the use of extrinsic FEP means that none is required in a practical cell. Front surface recombination was modelled by computing J01 as a function of dielectric charge using a semi-ideal PN junction in PC1D [12] with no back surface recombination and an intrinsic front SRV of ~500 cm/s –as experimentally found for the uncharged oxide film on textured Si in Figure 2 (߬ ؆ ʹͲ ߤs). Table 1. IBC solar cell parameters used in PC2D software.
Parameter
Value
Units
Thickness
160
ߤm
Unit cell width
500
ߤm
Analysis area
1
Doping concentration Holes diffusion coefficient
5x10 11.6
Electrons diffusion coefficient
32.3
Circuit series resistance
0.4
Illumination AM1.5G
0.1
Temperature
25
Midgap bulk recombination lifetime
10
Rear recombination J at n+ region o1
Rear recombination J at n region o1
Rear recombination J at p+ region o1
4x10 2x10
2
cm
15
-3
cm
2
cm /s 2
cm /s ohm 2
W/cm C ms -13 -14
1.5x10
-13
2
A/cm
2
A/cm
2
A/cm
Figure 3 illustrates the results of these simulations. It is clear that field effect passivation plays a vital role in enhancing cell efficiency if the intrinsic chemical passivation is poor. An astounding 12% absolute increase in efficiency is produced via extrinsic FEP if an uncharged film with a moderate chemical passivation quality is used as the base passivation layer. However, dielectric passivation layers are almost always intrinsically charged such that FEP is already exploited to some extent. Even in such practical cases, if an increase in FEP is produced extrinsically such that the dielectric charge increases from 1012 e/cm2 (average for SiN films) to 1013 e/cm2, then the cell efficiency would increase by 0.1 % absolute. This is a worthwhile improvement to achieve by extrinsically increasing the FEP component of the film. This charge concentration is achievable as commonly reported to be the case for charged Al2O3 films, yet here the charge is of extrinsic nature. Although the charge concentration reported in Section 3 for maximum lifetime was ~ 3 x 1012 e/cm2, it is possible to achieve higher concentrations of extrinsic charge using corona ions [13] or alkali ionic species [3].
778
Ruy S. Bonilla et al. / Energy Procedia 77 (2015) 774 – 778
-11
10
Dark saturation current density
AM1.5 Efficiency
25
22.36
AM1.5 Efficiency
22.34 -12
20
10
22.32
[%]
-13
[%]
J01 [10 -12 A/cm 2 ]
10
15
22.3 22.28 22.26
-14
22.24
10
10
22.22 12 10 -15
10
10
10
10
11
10
5
12
2
Dielectric charge [e/cm ]
10
10
10
11
13
10
12
10
2
Dielectric charge [e/cm ]
Fig 3. Dark saturation current of a semi-ideal PN junction, and AM1.5 efficiency of an IBC cell simulated in PC2D [11], as a function of charge concentration in the front passivation dielectric. Inset: zoom of efficiency for dielectric charge in the range [1012-1013] e/cm2.
5. Conclusion It has been demonstrated that silicon surface passivation can be greatly improved using extrinsic low-cost passivation techniques. In particular, extrinsic FEP has been shown to provide remarkable passivation in 1 ȍcm n-Si surfaces, with SRV as low as 1.28 cm/s. Textured surfaces are more prone to recombination however extrinsic FEP can reduce surface recombination similarly. Application of extrinsic reduction of surface recombination via FEP can lead to significant improvements in device efficiency as demonstrated from PC2D modelled IBC cells. References [1] B. Veith, T. Ohrdes, F. Werner, R. Brendel, P. Altermatt, N.-P. Harder, et al., Injection dependence of the effective lifetime of n-type Si passivated by Al2O3ௗ: an edge effect?, in: Proc. SiliconPV 2013 - Solomat, n.d.: pp. Hamelin, Germany. [2] R.S. Bonilla, F. Woodcock, P.R. Wilshaw, Very low surface recombination velocity in n-type c-Si using extrinsic field effect passivation, J. Appl. Phys. 116 (2014) 054102. [3] R.S. Bonilla, P.R. Wilshaw, A technique for field effect surface passivation for silicon solar cells, Appl. Phys. Lett. 104 (2014) 232903. doi:10.1063/1.4882161. [4] B.L. Henson, A derivation of Warburg’s law for point to plane coronas, J. Appl. Phys. 52 (1981) 3921. doi:10.1063/1.329241. [5] K. Nakamura, T. Takahashi, T. Hikichi, I. Takata, An observation of breakdown characteristics on thick silicon oxide, in: Proc. Int. Symp. Power Semicond. Devices IC’s ISPSD '95, Inst. Electr. Eng. Japan, 1995: pp. 374–379. doi:10.1109/ISPSD.1995.515066. [6] R.S. Bonilla, C. Reichel, M. Hermle, S. Senkader, P.R. Wilshaw, Controlled field effect surface passivation of crystalline n-type silicon and its application to back-contact silicon solar cells, in: Proc. 2014 40th IEEE Photovoltaics Spec. Conf., IEEE, Denver, CO, 2014. [7] R.A. Sinton, A. Cuevas, M. Stuckings, Quasi-steady-state photoconductance, a new method for solar cell material and device characterization, in: Photovolt. Spec. Conf. 1996., Conf. Rec. Twenty Fifth IEEE, 1996: pp. 457–460. doi:10.1109/PVSC.1996.564042. [8] A. Richter, S.W. Glunz, F. Werner, J. Schmidt, A. Cuevas, Improved quantitative description of Auger recombination in crystalline silicon, Phys. Rev. B. 86 (2012) 165202. doi:10.1103/PhysRevB.86.165202. [9] M.J. Kerr, A. Cuevas, Very low bulk and surface recombination in oxidized silicon wafers, Semicond. Sci. Technol. 17 (2002) 35–38. doi:10.1088/0268-1242/17/1/306. [10] M.J. Kerr, A. Cuevas, Recombination at the interface between silicon and stoichiometric plasma silicon nitride, Semicond. Sci. Technol. 17 (2002) 166–172. doi:10.1088/0268-1242/17/2/314. [11] P.A. Basore, K. Cabanas-Holmen, PC2D: A Circular-Reference Spreadsheet Solar Cell Device Simulator, IEEE J. Photovoltaics. 1 (2011) 72–77. doi:10.1109/JPHOTOV.2011.2166376. [12] U. of N.S. Wales, PC1D, (n.d.). http://www.pv.unsw.edu.au/info-about/our-school/products-services/pc1d (accessed August 20, 2012). [13] R.S. Bonilla, C. Reichel, M. Hermle, S. Senkader, P. Wilshaw, Controlled field effect surface passivation of crystalline n-type silicon and its application to back-contact silicon solar cells, in: 2014 IEEE 40th Photovolt. Spec. Conf., IEEE, 2014: pp. 0571–0576. doi:10.1109/PVSC.2014.6924985.
