Explanation of the LID Phenomenon in ...

1 downloads 0 Views 1MB Size Report
Sep 28, 2016 - Explanation of the LID Phenomenon in Monocrystalline. Silicon by the A. Si. -Si i. Defect Model. K. Lauer1, C. Möller1, D. Schulze2, C. Ahrens3, ...
Explanation of the LID Phenomenon in Monocrystalline Silicon by the ASi-Sii Defect Model K. Lauer1, C. Möller1, D. Schulze2, C. Ahrens3, J. Vanhellemont4, N.V. Abrosimov5 1CiS, 2TU

Ilmenau, 3Infineon Technologies AG, 4U Ghent, 5IKZ Berlin

PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

Overview

• •

ASi-Sii defect model Monitoring of ASi-Sii defect kinetics at varying temperature and illumination by: •

Fast in-situ carrier lifetime measurements with high spatial resolution by MWPCD



Low temperature photoluminescence spectroscopy



Low temperature FTIR spectroscopy



Hydrogen passivation



Conclusion

2 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

ASi-Sii defect model

3

Light-induced degradation (LID) of c-Si



First observed in n-in-p solar cells for space application

after electron irradiation and subsequent photon irradiation •

Defect degrades charge carrier lifetime in silicon by carrier injection (illumination or forward bias)



Microscopic structure of responsible defect not finally clarified

R. L. Crabb, Proceedings of the 9th IEEE Photovoltaic Specialists Conference, New York, 329 (1972)

4 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

Degradation in indium doped silicon In_1 In_2 Ga B

350

• Degradation is fully reversible

300 210

eff. lifetime [µs]

effective lifetime [µs]

• LID occurs in indium-doped silicon [1,2]

180

200

• Why does the defect occur in boron and indium and not in gallium doped silicon?

Annealing 200 °C 10 min

180 160 140 0

50

100

0

50 100 150

time [min]

150

120 2

10

3

4

5

10 10 10 time of light exposure [s]

[1] C. Möller and K. Lauer, phys. stat. sol. RRL 7, 461 (2013) [2] M.J. Binns et al., in 42nd IEEE Photovoltaic Specialist Conference (PVSC), (IEEE, 2015), p. 1.

5 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

Formation energies of interstitial acceptor atoms

• Configuration of interstitial acceptor atom with lowest formation energy varies for the type of acceptor • Gallium on tetrahedral position • Boron and indium forming an acceptor silicon interstitial pair (ASi-Sii)

M. Hakala et al., Phys. Rev. B 61, 8155 (2000) C. Melis et al., Appl. Phys. Lett. 85,4902 (2004) P. Alippi et al., Phys. Rev. B 69, 085213 (2004) P. Schirra et al., Phys. Rev. B 70, 245201 (2005)

6 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

ASi-Sii defect model • Total (electronic + elastic) energy diagram deduced from simulation results [1] • Three charge states (+, 0, -) and two configurations (S1, S2) of ASi-Sii possible

• Charge state change induced configuration changes cause the slow LID phenomenon

[1] M. Hakala et al., Phys. Rev. B 61, 8155 (2000)

K. Lauer, C. Möller, D. Schulze, and C. Ahrens, AIP Advances 5, 017101 (2015).

Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

7

Progress in ASi-Sii defect model development p-type SRC

n-type SRC

FRC 5 3

ASi-Sii-

4 1

ASi-Sii+

2

5

ASi-Sii0

ASi-Sii0

ASi-Sii-

2 4

ASi-Sii+

3

1

D

H

T

D

H

T

• Energy-configuration-diagram for LID phenomenon will be more complex

• Deduced from calculated defect configurations of the silicon interstitial R. Jones et al., Mat. Sci. Eng. B 159–160, 112 (2009).

Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

8

Defect composition

• ASi-Sii defect model requires LID defect to be composed of one acceptor atom and one silicon interstitial atom. • This can not be ruled out by experimental data related to:

• oxygen clustering during crystal cooling [1]. • the “boron interstitial” defect [1,2].

[1] K. Lauer et al., Solid State Phenomena 242, 90 (2015). [2] Lasse Vines et al., Physical Review B 78, 085205 (2008).

9 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

Operation of ASi-Sii defect model

10

Fast in-situ carrier lifetime measurements with high spatial resolution by MWPCD • Microwave detected photoconductance decay (MWPCD) setup from Semilab • Equipped with hot plate • Carrier lifetime extracted at fixed excess carrier density Δn = 5x1014 cm-3 [1]

hotplate [1] K. Lauer et al., J. Appl. Phys. 104, 104503 (2008).

Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

11

Typical carrier lifetime change due to LID phenomenon indium doped CZ silicon -2 Tannealing = 120°C Iav = 8.5 kWm Tannealing = 200°C

-1

inverse carrier lifetime  [s ]

• Indium doped CZ crystal -3

6x10

-3

-1

7x10

• 2 inch diameter, orientation • ρ = 3.8 Ωcm • [Oi] = 7.8x1017cm-3

5x10

• [Bs] = 1.9x1014cm-3

-3

Tmeasurement = 50°C 10

0

10

1

10

2

10

3

10

4

5 10 0.0

illumination time t [s]

• [Ins] = 2.9x1015cm-3 3

4

5.0x10 1.0x10 1.5x10

4

annealing time t [s]

• Inverse carrier lifetime measured as function of time using the MWPCD setup • LID generation by illumination

• LID annihilation by annealing Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

• Surface passivation by silicon nitride K. Lauer et al., to be published in Physica Status Solidi (c) (2016), presented at EMRS Spring Meeting 2016, Symposium K (Preprint available at researchgate.net)

12

indium doped CZ silicon -2 Tannealing = 120°C Iav = 8.5 kWm Tannealing = 200°C

-1

inverse carrier lifetime  [s ]

LID generation process

-3

-1

7x10

mainly 3 to 4

6x10

-3

mainly 1 to 2

5x10

-3

Tmeasurement = 50°C 10

0

10

1

10

2

10

3

10

4

5 10 0.0

illumination time t [s]

3

4

5.0x10 1.0x10 1.5x10

4

annealing time t [s]

• ASi-Sii defect model transitions are electron or hole captures (e.g. 4 to 5) and configuration changes (e.g. 3 to 4) • All 10 transitions simultaneously present

• LID generation: equilibrium shifts from state 1 to state 5 during illumination Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

13

indium doped CZ silicon -2 Tannealing = 120°C Iav = 8.5 kWm Tannealing = 200°C

-1

inverse carrier lifetime  [s ]

Transition rate measurement

-3

-1

7x10

mainly 3 to 4

6x10

5x10

exponential fit

-3

mainly 1 to 2

-3

Tmeasurement = 50°C 10

0

10

1

10

2

10

3

10

4

5 10 0.0

illumination time t [s]

3

4

5.0x10 1.0x10 1.5x10

4

annealing time t [s]

• LID generation rate usually obtained from exponential fit • Measured rate represents the solution of a rate equation system including each transition in the ASi-Sii defect model 14 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

Interpretation of measured transition rate • Rate equation system for 5 states (Zi) of the ASi-Sii defect model:

dZ i  kij Z j dt

• kij includes the possible transition rates of the model • E.g.: solution Z5(t) represents slow LID component • Systematic error occurs if a measured transition rate is assigned to only one transition • Same situation as in case of ASi-Fei defects • If association and dissociation reaction of ASi-Fei defects are not separated, the activation energy of the association is considerably overestimated [1,2].

[1] C. Möller et al., J. Appl. Phys. 116, 024503 (2014). [2] K. Lauer et al., Solid State Phenomena, 242, 230 (2015).

15 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

indium doped CZ silicon -2 Tannealing = 120°C Iav = 8.5 kWm Tannealing = 200°C

-1

inverse carrier lifetime  [s ]

LID annihilation process

-3

6x10

-3

5x10

-3

-1

7x10

mainly 5 to 2

Tmeasurement = 50°C 10

0

10

1

10

2

10

3

10

4

5 10 0.0

illumination time t [s]

3

4

5.0x10 1.0x10 1.5x10

4

annealing time t [s]

• Equilibrium of ASi-Sii defect is shifted from state 5 to state 1 during annealing • Dependence of equilibrium on temperature indicates impact of other transitions during annihilation process (e.g. 3 to 4) 16 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

photoluminescence intensity [a. u.]

Low temperature PL spectroscopy 200°C 10min 13.5h illumination 1h illumination 200°C 10min 10h illumination 1h illumination untreated

T = 4.2K

sample: F_2 BTO(BE)

P line

K. Lauer et al., AIP Advances 5, 017101 (2015).

ITO(FE) InNP(BE)

1090

BTA(BE)

1100

1110

R line

1120

1130

1140

1150

wavelength  [nm] • Unidentified photoluminescence peak called P line appears in In implanted samples with a silicon interstitial rich region placed on the In implantation peak • P line follows LID annealing and illumination cycle Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

17

Impact of LID cycle treatment

PL intensity ratio P line/I TO(FE)

8 7 6

C_2 D_2 F_2

5 4 3 2 1 0

ation C 10 min mination mination C 10 min n i m u u llu 200° 200° 1h ill 13.5 h ill 10h i • Photoluminescence peaks can be attributed only to one specific state • P line identified as InSi-Sii0 in C2v configuration (state 2) 18 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

Hydrogen passivation 10

0

Possible mechanism [1]:

10

-1

10

-2

EA = (1.36±0.06) eV

ASi-Sii0(S2) + H0

-1

transition rate R5,1 [s ]

BSi-Sii-defect (transition 5=>1)

0

10

-3

10

-4

10

-5

10

-6

16

H3 -center dissociation EA = (1.37±0.02) eV

-3

p0 = 10 cm

2.0

ASi-Sii-H+ + e-

Lim et al. [48] Bothe et al. [6] Rein et al. [49] Yarykin et al. [47] 2.1

2.2

2.3

2.4

Regeneration

Destabilization

2.5

2.6

2.7

2.8

ASi-Sii+(S1) + H0 + e-

-1

inverse temperature T [1000/K] • DLTS center called H3 [2] could be identified

[1] K. Lauer et al., Solid State Phenomena 242, 90 (2015). [2] N. Yarykin et al., Physical Review B 69, 045201 (2004).

with BSi-Sii–H defect 19 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

Conclusion

• Monitoring defect kinetics at varying temperature and illumination by • Fast in-situ charge carrier lifetime measurements with high spatial resolution by MWPCD • Low temperature photoluminescence spectroscopy

• Low temperature FTIR spectroscopy • ASi-Sii defect model is a promising approach to explain phenomena of LID and P line • LID phenomenon due to charge state change induced configuration changes • Defect kinetics have to be described by solving a rate equation system • Regeneration of LID phenomenon explained by capture of hydrogen by the ASi-Sii defect

K. Lauer, C. Möller, D. Schulze and C. Ahrens, AIP Advances 5, 017101 (2015). K. Lauer, C. Möller, D. Schulze, C. Ahrens and J. Vanhellemont, Solid State Phenomena 242, 90 (2015). K. Lauer, C. Teßmann, D. Schulze and N.V. Abrosimov, to be published Physica Status Solidi (c) (2016).

20

Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

Thank you for your kind attention!

BMWi (VF140015)

21 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

Migration of interstitial boron at room temperature?

Common understanding of Bi, which is the BSi-Sii-defect [1]: • Trapping self-interstitials by Bs produces the interstitial boron species, Bi, which is mobile above ~240 K [2,3] and can be trapped by other defects e.g. BiBs. • In contradiction to ASi-Sii-defect model since LID remains after high temperature anneals • Problem: In these papers [2,3] there is no “direct evidence” that migration of boron is the cause for the disappearance of Si-G28 or DLTS peaks! • The annealing kinetics of these signals are still an open question [4].

[1] [2] [3] [4]

E. Tarnow, Europhys. Lett. 16, 449 (1991). G. Watkins, Physical Review B 12, 5824 (1975). J. Troxell et al., Phys. Rev. B 22, 921 (1980). R. Harris et al., Mat. Sci. For. 10-12, 163 (1986)

22 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

Migration of interstitial boron at room temperature? 10

-1

-1

transition rate R [s ]

EPR, Watkins VOC, Bothe and Schmidt 10

10

-2

E2

= (0.41 ± 0.03) eV

-3

0

+

(BSi-Sii) (C3v)

-

(BSi-Sii) (C3v) + e

2 10

1

1

-4

2.6

2.8

3.0

3.2

3.4

3.6

-1

inverse temperature 1000/T [K ] • Disappearance of Si-G28 coincides with annihilation of fast LID component => ASi-Sii-defect model in agreement with experimental data of boron interstitial defect K. Lauer et al., AIP Advances 5, 017101 (2015).

23 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

Dependency of ASi-Sii-defect density on [Oi]2

• LID defect density experimentally found to be proportional to [Oi]2 • Explanation in frame of ASi-Sii-defect model : – Oi forms precipitates during crystal cooling [1] – For two precipitating oxygen atoms approximately one silicon atom is released [2] – Sii migrates to substitutional acceptor atom and forms ASi-Sii-defect [3,4]

• Consequence: – Oxygen precipitate density in as grown crystals must be proportional to [Oi]2 • Can this be excluded by present available experimental data?

• No!

[1] [2] [3] [4]

J. Vanhellemont et al., Electrochem. Soc. Proc. 98-13, 101 (1998). U. Gösele et al., Appl. Phys. A 28, 79–92 (1982) V. Akhemetov et al. Phys. Status Solidi A 72, 61 (1982) A. Bean et al. J. Phys. C 5, 379 (1972)

24 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

V1 V2 V3 V4 V5 V6

-3

precipitated oxygen [Oi] [cm ]

Oxygen precipitation by annealing steps

10

18

10

17

10

16

6x10

ASTM 1239-94: 1050 °C, 16 h

[Oi] ~ [Oi]

2.2±0.3

black: Lab 3 red: Lab 4 17

8x10

17

10

18

1.2x10

18

-3

initial interstitial oxygen concentration [Oi] [cm ] -2

calibration factor  = 3.14 cm (IOC88) •

Precipitated oxygen concentration shows in relevant region a quadratic dependency on interstitial oxygen concentration 25 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

Dependency of ASi-Sii-defect density on [Oi]2 • Comparison of measurement methods for precipitate density in as-grown CZ crystal

FTIR + Ham >90% of BMDs are invisible by C&E, SIRM, LST Defect etching SIRM IR-LST

G. Kissinger et al., Mat. Sci. Sem. Proc. 9, 236 (2006).

Experimental basis for current understanding of precipitates in as-grown crystals 26

Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

Dependency of ASi-Sii-defect density on [Oi]2

• Relation of invisible precipitate density on [Oi] is not yet investigated experimentally. • Theory of homogenous oxygen nucleation predicts a [Oi]2 dependency [1] => An [Oi]2 dependency of the majority of the oxygen precipitate density in as-grown silicon cannot be excluded and is even expected from theory. => An indirect involvement of oxygen in the LID defect cannot be excluded by present available experimental data on oxygen clustering during crystal cooling! • Interesting: Thermal donor generation rate depends on [Oi]2 as well.

[1] T. Tan et al., in „Oxygen in Silicon“ (1994)

27 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

Measuring iron acceptor reaction rates RA RD

relative interstitial iron content [Fe i]rel

• Dynamic approach applied

complete dissociation

1.0

• Association and dissociation reaction of iron acceptor pairs are simultaneously present

9 sun illumination 1.4 sun illumination

0.8

equilibrium between association and dissociation reaction

0.6

0.4

[Fei]rel(t) = {RD + RAexp(-(RA + RD)t)} / {RD + RA} 0.2

0.0

1000

-1

-4

-1

RA = (8.9±0.2)x10 s

T = 70°C boron sample 0

-4

RD = (7.8±0.2)x10 s 2000

3000

time t [s] Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

• Fitting of [Fei]rel by solution of differential equation system of two first order defect reactions [1,2] • Reaction rates of association as well as dissociation can be obtained from one measurement

4000 [1] C. Möller et al., J. Appl. Phys. 116, 024503 (2014). [2] K. Lauer et al., Solid State Phenomena, 242, 230 (2015).

28

Impact of illumination intensity

• Example for impact of illumination on iron acceptor defect reaction • Equilibrium is shifted to 100% interstitial iron by increasing illumination • Time constant of monoexponentially evaluated association reaction changes, too!

[1] Möller et al., J. Appl. Phys. 116, 024503 (2014).

29 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16

ASi-Sii-defect model • Energy levels within the model • Most reliable: transition from 4 to 5 (degraded state)

b) 0

EC )

energy

+ 0 +

-

)

0 )

2

2

)

0

)

1

)

)

)

3

0

)

4

5

EV

[] M. Hakala, M. J. Puska, and R. M. Nieminen, Physical Review B 61, 8155 (2000). [] R. D. Harris,J. L. Newton, G. D. Watkins, Physical Review B 36, 1094 (1987). [] K. Bothe and J. Schmidt, J. Appl. Phys. 99, 013701 (2006). [] S. Rein and S. W. Glunz, Applied Physics Letters 82, 1054 (2003). [] T. Mchedlidze and J. Weber, Phys. Status Solidi RRL 9, 108 (2015).

30 Kevin Lauer, PV Days 2016, Fraunhofer CSP, Halle, 28.09.16