Highly Transparent and Highly Passivating Silicon Nitride for Solar Cells. Yimao Wan The Australian National University (ANU) 23/10/2014

Transcription

1 Highly Transparent and Highly Passivating Silicon Nitride for Solar Cells Yimao Wan The Australian National University (ANU) 23/10/2014

2 2 Outline Motivation Reviews of SiNx properties Process development Recombination studies Planar Texturing Cell simulation and application

3 3 Motivation Success of SiNx on silicon solar cells PECVD SiN x is incorporated into most laboratory and industrial silicon solar cells, fulfilling three functions: i. it comprises the antireflection coating; ii. iii. it provides surface and bulk passivation; and it forms a chemical barrier to protect underlying interfaces from the degrading effects of moisture, humidity and sodium ions.

4 4 Motivation Success of SiNx on silicon solar cells PECVD SiN x is incorporated into most laboratory and industrial silicon solar cells, fulfilling three functions: i. it comprises the antireflection coating; ii. iii. it provides surface and bulk passivation; and it forms a chemical barrier to protect underlying interfaces from the degrading effects of moisture, humidity and sodium ions.

5 5 Motivation Challenge of SiNx on silicon solar cells Planar undiffused FZ p-si Classic trend: SRV decreases as n increases, irrespective of deposition techniques. High absorption associated with Si-rich SiN x

6 Refractive index n at 632 nm Stoichiometric: N/Si = 4/3 6 Review of SiNx properties Optics: refractive index N/Si ratio Claassen et al (LF direct, RBS) Bustarret et al (RF direct, ERD) Lenkeit et al ( W remote, ERD) Verlaan et al (ERD) Hotwire W remote RF direct LF direct Calculated by Eq. (2.1) (Bustarret et al. 1988) N Si = n n

7 Extinctin coefficient k at 360 nm Stoichiometric: N/Si = 4/3 Absorption coefficient at 360 nm (cm -1 ) 7 Review of SiNx properties Optics: extinction coefficient Doshi et al (RF direct) van Erven et al (ETP) Duttagupta et al ( W remote inline) This work ( W/RF dual-mode) N/Si ratio

8 [Si N] (10 22 cm -3 ) Stoichiometric: N/Si = 4/3 8 Review of SiNx properties Structures: [Si N] Mäckel and Lüdemann 2002 (RF direct) Cuevas et al ( W remote) Chen et al ( W/RF dual-mode) van Erven et al (ETP) This work ( W/RF dual-mode) N/Si ratio

9 [Si H] or [N H] (10 21 cm -3 ) Stoichiometric: N/Si = 4/3 9 Review of SiNx properties Structures: [Si H] and [N H] [Si H] Lauinger et al ( W remote) Chen et al ( W/RF dual-mode) Cuevas et al ( W remote) Verlaan et al (Hotwire) van Erven et al (ETP) This work ( W/RF dual-mode) 1 [N H] Lauinger et al ( W remote) van Erven et al (ETP) Chen et al ( W/RF dual-mode) Cuevas et al ( W remote) Verlaan et al (Hotwire) This work ( W/RF dual-mode) N/Si ratio

10 Si H peak wavenumber (cm -1 ) Stoichiometric: N/Si = 4/3 Increase of the back-bonded N atoms 10 Review of SiNx properties Structures: [Si H] peak wavenumber 2220 H Si N H Si N 2 H H Si N 2 Si H Si NSi 2 Mäckel and Lüdemann 2002 (RF direct) Cuevas et al ( W remote) Verlaan et al (Hotwire) Verlaan et al Hotwire W remote RF direct LF direct This work ( W/RF dual-mode) N/Si ratio H Si Si 3

11 Q eff (cm -2 ) Stoichiometric: N/Si = 4/3 11 Review of SiNx properties Electronics: insulator charge density Q eff Hezel et al (LF direct) Bagnoli et al (RF direct) Landheer et al & 1998 (ECR) Garcia et al (ECR) Dauwe 2004 ( W remote) De Wolf et al (LF direct) Lelievre et al (LF direct) Lamers et al ( W remote) This work ( W/RF dual-mode) Temperature NH 3 :SiH 4 ratio Pressure W plasma power N/Si ratio

12 D it (ev -1 cm -2 ) Stoichiometric: N/Si = 4/3 12 Review of SiNx properties Electronics: interface defect density D it Hezel et al (LF direct) Landheer et al (ECR) Garcia et al (ECR) Lamers et al (remote W) This work ( W/RF dual-mode) Temperature NH 3 :SiH 4 ratio Pressure W plasma power N/Si ratio

13 13 Review of SiNx properties Conclusion 1 Irrespective of deposition techniques, (i) the bulk structural and optical properties are universally correlated to the N/Si ratio; and (ii) the electronic properties (Q eff and D it ) appear independent of the N/Si ratio. Promoting an opportunity of decoupling SiN x surface passivation and optical transmission properties; and therefore Circumventing the trade-off between the two properties.

14 14 Process development Methodology 1) Varying deposition parameters in Roth & Rau AK400: NH 3 :SiH 4 ratio Pressure Temperature Microwave plasma RF plasma 2) Monitoring surface passivation eff (Δn) 3) Monitoring optical properties n & k

15 15 n at 632 nm Process development Results NH 3 :SiH 4 ratio eff at n = cm -3 ( s) p-type 0.85 cm NH 3 /SiH 4 gas flow ratio NH 3 /SiH 4 gas flow ratio

16 16 Refractive index n at 632 nm Process development Results Pressure and temperature eff at n = cm -3 ( s) ºC eff on p-0.85 cm 290 ºC eff on p-0.85 cm 290 ºC eff on n-0.47 cm 235 ºC 290 ºC Pressure (mbar) Pressure (mbar) 1.8

17 17 Process development Results S eff,ul vs. n

18 18 Process development Surface passivation: eff (Δn)

19 19 Process development Optical transmission OPAL 2 Random textured Si surface Si thickness: 180 µm Cell under air Cell encapsulated beneath glass/eva Ref. n 632nm S eff,ul (cm/s) J inc t SiNx (nm) J gen J Abs J relf J Inc t SiNx (nm) (ma/cm 2 ) J gen J Abs J relf (ma/cm 2 ) DuttaGupta et al This work at 290 C ΔJ ΔJ

20 S eff,ul at n = cm -3 (cm/s) 20 Process development Thermal stability 10 4 Poor thermal stability As-deposited Post RTA Correlated to [Si N] Further studies required S eff,ul-post RTA / S eff,ul-as deposited [Si-N] (10 22 cm -3 ) RTA 800 C for 5 seconds

21 21 Process development Conclusion 2 A single SiN x layer can provide both high passivation of c-si surface high transmission at short wavelength

22 Recombination studies planar 22 Methodology 1) Extracting S eff,ul (Δn) from measured τ eff (Δn) 2) Probing electronic properties by C-V measurements: D it and Q eff 3) Modeling S eff,ul (Δn)

23 S eff,ul (cm s -1 ) Recombination studies planar 23 Extracting S eff,ul (Δn) 10 3 Lauinger et al cm (1996) Schmidt and Kerr cm (2001) Kerr and Cuevas cm (2002) DuttaGupta et al cm (2011) 10 2 This work 0.85 cm ºC This work 0.85 cm ºC S eff ( n) = w 2 ( 1 τ eff ( n) 1 τ bulk ( n) ) w : Si wafer thickness τ eff ( n) : effective minority carrier lifetime τ bulk ( n) : Si bulk lifetime 10 1 FZ p-type {100} c-si Excess carrier density n (cm -3 ) Similar injection-level dependence Lowest Auger-corrected S eff,ul (1.6 cm/s)

24 Recombination studies planar 24 Probing electronics: D it, Q eff and n/p D it-midgap (cm -2 ev -1 ) Q eff (cm -2 ) Extracted from quasi-static (QS) and high-frequency (HF) capacitancevoltage (C-V) measurements Schmidt et al. (1997)

25 S eff,ul (cm s -1 ) Recombination studies planar 25 Modeling S eff,ul (Δn) 10 5 FZ p-type {100} c-si Defect C B A Excess carrier density n (cm -3 ) Notes for the S eff,ul (Δn) modeling: Adapting the latest Si intrinsic bulk lifetime model Richter et al. (2012) Assuming a single defect at a single energy level (midgap) Assuming n & p saturate and constant for the unmeasured gap regions Defect A or B (or both) is likely to dominate the recombination at the Si SiN x interface. Defect C is excluded.

26 Recombination studies planar 26 Conclusion 3 We obtained a low S eff,ul = 1.6 cm/s on 0.8 Ωcm p-type FZ Si, with D it = 3.0 x ev 1 cm 2 at midgap, and Q eff = 5.6 x cm 2. Defect A or B (or both) is likely to dominate the recombination at the Si SiN x interface, and Defect C is excluded.

27 Recombination studies texturing 27 Methodology 1) Employing the optimum deposition condition and varying the NH 3 :SiH 4 ratio to obtain a wide range of film refractive indices (n at 632nm = ) 2) Monitoring S eff,ul as a function of n at 632nm on three types of Si surfaces: i. Planar {100} ii. Planar {111} iii. Texturing with random upright pyramids 3) Probing electronic properties at textured surfaces by by depositing corona charge

28 Recombination studies texturing 28 Drivers of enhanced recombination 1. Enlarged surface area 2. Orientation of surface planes: f O = {111}/{100} 3. Presence of concave and convex surface features: f V = texture/{111}

29 Recombination studies texturing Surface area-corrected S eff,ul at n = cm -3 (cm s -1 ) 29 Results: surface area-corrected S eff,ul Consistent trend: S eff,ul first decreases and then saturates Two exceptions: 10 2 Lauinger et al NH 3 :SiH 4 gas ratio {100} {111} Rantex low S eff,ul saturates at lower n Low and constant S eff,ul over a range of n ( at 632 nm) surface morphologies 10 1 a-si:h Refractive index n at 632 nm

30 Recombination studies texturing Surface area-corrected 30 Results: surface area-corrected S eff,ul ratio f O is constantly low ( ) less orientation dependence f O : {111}/{100} f V : rantex/{111} f V relates strongly to n high at low n sensitive to morphology close to 1.0 when n 2.3 negligible increase at texture, after area-correction S eff,ul ratio: f O, or f V (b) a-si:h Refractive index n at 632 nm

31 S eff,ul at n = cm 3 (cm/s) Recombination studies texturing 31 Probing electronics by corona-lifetime technique 10 2 S it Planar {100} n-type FZ Si n at 632 nm S 0 S it 10 1 S 0 Q eff Q eff Applied surface charge density Q s (10 12 cm 2 )

32 Surface area-corrected S it (cm/s) Recombination studies texturing Surface area-corrected Q eff (10 12 cm -2 ) 32 Results: S it or Q eff versus S 0 (a) n-type 1.1 cm FZ Si (b) {100} {111} rantex Surface area-corrected S 0 (cm/s) {100} {111} rantex Surface area-corrected S 0 (cm/s)

33 Recombination studies texturing 34 Conclusion 4 A constant and low SRV is achieved on planar and textured Si surfaces over a wide range of n ( ), When passivated by N-rich SiNx, the increase in recombination at textured surfaces is high, and it is i. due to the presence of vertices/edges on pyramids rather than due to the {111} orientation (i.e., high f V and low f O ), and ii. primarily caused by an increase in D it rather than a reduction in Q eff. When passivated by Si-rich or a-si:h, the increase in recombination at textured surfaces is negligible after are-correction (i.e., f V and f O are close to unity).

34 Cell simulation and application Refractive index, n Extinction coefficient, k 35 Simulating impact of SiN x on cells SiN x functions as front surface passivation and ARC layer p-type PERC (n + -diffused front) and n-type IBC (undiffused front) Cell simulation using Quokka 2 [Fell 2012] Optical simulation using OPAL 2 [McIntosh et al. 2012] Random upright pyramids Spectrum AM 1.5g Operating in air (a) , (b) , Wavelength (nm) Wavelength (nm)

35 J Refl (ma/cm 2 ) J Abs (ma/cm 2 ) Cell simulation and application J Gen (ma/cm 2 ) Optimised SiN x thickness (nm) 36 Simulation results (1/3): ARC Refractive index n at 632 nm 2.5 (a) 2.5 (b) 43 (c) n at 632 nm n at 632 nm n at 632 nm

36 FF (%) V OC (mv) Cell simulation and application J SC (ma/cm 2 ) Simulation results (2/3): p-perc cell (a) 39 (b) High recombination Refractive index n at 632 nm (c) Refractive index n at 632 nm 21.5 (d) High absorption (%) Refractive index n at 632 nm Refractive index n at 632 nm 37

37 FF (%) V OC (mv) J SC (ma/cm 2 ) Cell simulation and application Simulation results (3/3): n-ibc cell (a) (b) IBC PERC IBC PERC High recombination (c) Refractive index n at 632 nm Refractive index n at 632 nm (d) IBC PERC High absorption 81.5 (%) IBC PERC Refractive index n at 632 nm Refractive index n at 632 nm 38

38 Current density J (ma/cm 2 ) 39 Cell simulation and application Application on n-type IBC C k i or o ta ts r th r a i 2 C i ta Measured Simulated V OC J SC FF η (mv) (ma/cm 2 ) (%) (%) ± ± 0.5 Under standard testing conditions (25 C, AM1.5G spectrum) Voltage (V)

39 40 Cell simulation and application Conclusion 5 The optimum SiN x for both p-type PERC and n-type IBC has n = 1.9, owing to: high surface recombination when n < 1.9 high film absorption when n > 1.9 Application of the optimum SiNx onto the front textured undiffused surface of an n-type IBC cell enabled a conversion efficiency of 24.4 ± 0.5%.

40 41 Summary No universal correlation exists between surface recombination and bulk structural/optical properties. A highly transparent and highly passivating SiNx is attained, enabling a 24.4%-efficient n-type IBC cell. An increase in recombination of the textured surfaces is i. related to the presence of vertices and/or edges of the pyramids rather than to the presence of {111}-orientated facets, and; ii. primarily attributable to an increase in D it rather than a decrease in Q eff.

41 Thank you! 42