How big problem is Cu contamination in crystalline silicon - PowerPoint PPT Presentation

How big problem is Cu contamination in crystalline silicon photovoltaics? Hele Savin Aalto University Department of Electronics and Nanoengineering Electron Physics Group Espoo, Finland

(population ~ 5 million) 5.10.2019 2

5.10.2019 3

5.10.2019 4

Merger of three leading Finnish universities Helsinki University of Technology 1849 University of Art & Design Helsinki 1871 2010 Helsinki School of Economics 1911 12 000 A staff of about 4 000 , of which nearly 400 are professors full-time equivalent degree students

Micronova: Research center in Micro - and Nanotechnology • The centre also hosts many companies • Industrial scale facilities, 2600m 2 of cleanrooms • 360 personnel (19 in maintenance) • 60 PhD students • Over 150 major equipment installed • 150 mm R&D Fab, currently upgraded to 200 mm http://www.micronova.fi/

Micronova: Research center in Micro - and Nanotechnology F1 Flip-chip Bonding F2 Wet Processing F3 Nanolithography F4 Plasma and Sputtering F5 Plating F6 Metrology F7 Furnace F8 Lithography F9 ALD F10 Plasma F11 Wafers F12 Chemistry F13 Analysis Lab http://www.micronova.fi/

Micronova: Research center in Micro - and Nanotechnology F1 Flip-chip Bonding F2 Wet Processing F3 Nanolithography F4 Plasma and Sputtering F5 Plating F6 Metrology F7 Furnace F8 Lithography F9 ALD F10 Plasma F11 Wafers F12 Chemistry F13 Analysis Lab http://www.micronova.fi/

Electron Physics Group Crystalline silicon PV activities • Black silicon • DRIE and MACE Electrical simulations • • IBC/lab and PERC/industry • How to explain >100% EQE • Black Ge • ALD metal oxides Surface passivation, passivated contacts • • Up-conversion materials • Coating of module glasses • Light/Carrier induced degradation • Cu-LID • Mitigation of LeTID 2.2.2016 9

Use of “PV” -innovations in other fields Cast-Mono in MEMS&NEMS 10

Use of “PV” -innovations in other fields Certified EQE in UV > 130% Cast-Mono in MEMS&NEMS 11

Use of “PV” -innovations in other fields Utilize LID in Certified EQE in UV > 130% Cast-Mono in MEMS&NEMS Integrated Circuits 12

Can we take advantage of this effect somewhere else? Reference Cu-contaminated 13

Another example how to utilize LID Initial t @ high injection Fe Cu

Another example how to utilize LID Initial t @ After light high injection soaking Fe Cu

Another example how to utilize LID Initial t @ After light Nice Cu (and Fe) map ! high injection soaking Fe Cu ECS Trans. 11, 319 (2007)

Measurement of low-Cu concentrations • The strength of degradation depends on Cu concentration Benefits: • Other common impurities such as Fe can be separated • Contactless, non-destructive, fast, sensitive Appl. Phys. Lett. 87 , 032109 (2005)

How about mc-Si? t = 0 high ref (no Cu) low Cu med Cu high Cu low

How about mc-Si? t = 2 min high ref (no Cu) low Cu med Cu high Cu low

How about mc-Si? t = 17 min high ref (no Cu) low Cu med Cu high Cu low

How about mc-Si? t = 3 hours high ref (no Cu) low Cu med Cu high Cu low

How about mc-Si? t = 8 hours high ref (no Cu) low Cu med Cu high Cu low

Naturally works with PL too Cu contaminated spot

Using PL to measure Cu in Silicon High lifetime Cu contaminated spot Low lifetime 1 1 ∗ = 𝐷 𝐷𝑣 ∝ 𝑂 𝑢 − τ 𝑒𝑓𝑕𝑠𝑏𝑒𝑓𝑒 τ 𝑗𝑜𝑗𝑢𝑗𝑏𝑚

The role of oxygen (BMD) • Sensitivity can be increased by adding oxygen precipitates • Detection limit ppt -level Appl. Phys. Lett. 87 , 032109 (2005)

Boron and oxygen certainly helps… … but you don’t need them FZ silicon n-type silicon Appl. Phys. Lett. 95 , 152111 (2009) 26

What about elevated temperature …? Definitely makes the kinetics faster Can be used for faster detection Appl. Phys. Lett. 107, 052101 (2015) 27

What about higher intensity …? Similarly makes the kinetics faster …. …. further work ongoing on this J. Electrochem. Soc. 150 (12), 2003 28

Gettering during POCl 3 diffusion? Phosphorus-doped region • Cu fast diffuser • Should be easy to getter • … not always true Cu Cu Cu AIP Advances 8 , 015112 (2018) 29

Gettering during POCl 3 diffusion? Phosphorus-doped region • Cu fast diffuser • Should be easy to getter • … not always true Cu Cu Cu Normal cooling after POCl 3 + light soak 410 Diffusion length [  m] 370 330 290 250 210 170 10 20 30 40 50 Position [mm] AIP Advances 8 , 015112 (2018) 30

Gettering during POCl 3 diffusion? Phosphorus-doped region • Cu fast diffuser • Should be easy to getter • … not always true Cu Cu Cu Slow cooling after POCl 3 Normal cooling after POCl 3 + light soak + light soak 410 Diffusion length [  m] 410 Diffusion length [  m] 370 370 330 330 290 290 250 250 210 210 170 170 10 20 30 40 50 10 20 30 40 50 Position [mm] Position [mm] AIP Advances 8 , 015112 (2018) 31

How stable is it during firing? It is not stable, during high-T firing copper diffuses back to the bulk ! Appl. Phys. Lett. 113 (2018) 32

Impact of firing temperature? If firing T is too slow → no Cu -LID Cu prefers to stay in the emitter / surfaces Appl. Phys. Lett. 113 (2018) 33

Impact of ramp rates during firing? Slow ramp rates reduce Cu-LID Simulations verify that Cu goes to bulk during firing but has time to diffuse back to emitter Appl. Phys. Lett. 113 (2018) 34

Cu in PERC cells? AFTER ILLUMINATION INITIALLY Solar Energy Materials and Solar Cells 186, 373-377 (2018) 35

Cu in PERC cells? AFTER ILLUMINATION INITIALLY Solar Energy Materials and Solar Cells 186, 373-377 (2018) 36

Quantum efficiency Solar Energy Materials and Solar Cells 186, 373-377 (2018) 37

Physics behind Cu-LID (+modeling) • Cu i positively charged, fast diffuser, not recombination active • Cu precipitates positively charged, highly recombination active • Hypothesis: Light changes the charge state of Cu precipitates → electrostatic attraction • All parameters known (diffusivity, solubility, precipitation kinetics, recombination parameters… ) • Modeling shows pretty nice correlation with experiments H. Vahlman, PhD thesis 2018 and publications there 38

Physics behind Cu-LID (+modeling) • Cu i positively charged, fast diffuser, not recombination active • Cu precipitates positively charged, highly recombination active • Hypothesis: Light changes the charge state of Cu precipitates → electrostatic attraction • All parameters known (diffusivity, solubility, precipitation kinetics, recombination parameters… ) • Modeling shows pretty nice correlation with experiments Still cannot see anything here in e.g. TEM / DLTS… H. Vahlman, PhD thesis 2018 and publications there 39

What about dark anneal…? Formation of recombination active precipitates - supersaturation (fast reaction) Precipitate dissolution – diffusion to surfaces (slow reaction) Solar Energy Materials and Solar Cells 134 (2018) 40

What about dark anneal…? Solar Energy Materials and Solar Cells 134 (2018) 41

“Le - TID” after dark anneal No LeTID after long dark anneal Solar Energy Materials and Solar Cells 134 (2018) 42

“Le - TID” after dark anneal in PERC Silicon PV 2019 43

What about hydrogenation…? All the prior results have been carried out without PECVD SiN x :H , so we don’t know … yet 44

Conclusions • LID can be also a positive issue Measurement of Cu contamination • • Enhancing Cu gettering in microelectronics • Cu can cause severe LID in solar cells – be aware of contamination risks • It may be difficult to separate Cu-LID from other LID mechanisms … • Looking forward to further collaborations with UNSW! 45

Thank you! Funding acknowledgements: 2.2.2016 46