1. G. Kartopu*, A.K. Gürlek, A.J. Clayton, S.J.C. Irvine Centre for Solar Energy Research, OpTIC Glyndŵr, St. Asaph, UK B.L. Williams, V. Zardetto, W.M.M. Kessels, M. Creatore Department of Applied Physics, Eindhoven University of Technology, The Netherlands * Now at: Swansea University, College of Engineering, Swansea, UK Improvement of CdZnS/CdTe solar cells via use of a highly-resistive ZnO transparent buffer layer PVSAT-12, Liverpool University, 7 March 2016

2. Motivation  CdTe PV technology more interesting than ever with new record efficiencies: 22.1% for cells & 18.6% for modules  Metalorganic chemical vapour deposition (MOCVD) for thin film PV: Effectively used for III-V epitaxial multi-junction cells Not widely used for polycrystalline thin film PV Polycrystalline MOCVD CdTe best cell efficiency ~15% (CSER)

3. Motivation  Additional strategies needed to improve MOCVD CdTe devices  One could be to use a highly-resistive transparent (HRT) buffer layer btw. the front contact & window layers (e.g. C.S. Ferekides et al., Thin Solid Films, 2005)  Atomic layer deposition (ALD) can provide uniform, pin-hole free conformal oxide/nitride coatings at low temperatures: P. Banerjee et al., Nature Nanotech. (2009)  We used a 50 nm ALD ZnO film as the HRT layer for MOCVD CdTe devices  Effect of substrate treatment by O 2 plasma & CdZnS thickness investigated

4. 1 ITO & 1 HRT/ITO substrate side-by-side H 2 carrier gas, T sub = 200-420  C 60-150nm Cd 0.3 Zn 0.7 S window (E g  2.9 eV) CdTe:As bi-layer (low/high doping) absorber CdCl 2 anneal (420  C, 10min, in H 2 ) & air- anneal (170  C, 90min) for activation 8  0.25 cm 2 Au contacts MOCVD CdTe cell deposition (CSER) Oxford Instruments OpAL reactor, T sub = 100  C, ~0.2 nm/cycle ITO/aluminosilicate (4-8 Ω/  ) substrate, diethylzinc & H 2 O precursors; Ar carrier/purge Resistivity: 1-5 × 10 6 Ohm·cm for 50 nm ZnO film on SiO 2 /Si ZnO ALD growth (Eindhoven University) Experimental ZnO/ITO ITO spacer Flow susceptor

5. Case 1: CdZnS  150nm (ZnO also O 2 plasma-cleaned) JV setup calibrated with Si detector V oc loss by 10-20mV; FF loss by 0.2-1.4% J sc improves by 0.5-0.7 mA/cm 2 (ZnO/ITO also O 2 plasma cleaned)

6. Case 2: CdZnS  150nm (no plasma treatment for ZnO) JV setup calibrated with GaAs detector V oc loss by 20mV; FF improves by 2.3-4.4% J sc improves by 2.2-2.6 mA/cm 2 (no O 2 plasma cleaning for ZnO/ITO)

7. Case 3: CdZnS  60nm (no plasma treatment for ZnO) JV setup calibrated with GaAs detector V oc improves by 200-210mV; FF improves by 2.5-3.8% J sc improves by 1.5-1.8 mA/cm 2 (no O 2 plasma cleaning for ZnO/ITO)

8. Spectral response (EQE): ITO vs. ZnO/ITO In general – for ZnO/ITO spectral response more uniform & stable (compare, e.g. Devices 1 & 4 with the same CdZnS thickness) Above-bandgap absorption by ZnO (i.e. 500nm) response enhances ZnO

9. AM1.5 Light J-V: ITO vs. ZnO/ITO In general – ZnO/ITO gives higher J sc & FF by as much as 2 mA·cm 2 & 4%, respectively Plasma cleaning does not to lead to FF enhancement FF seems to benefit from the rise in shunt (R sh ) & small drop in series (R s ) resistances The V oc is lowered by 20mV (to be discussed next) Device 4 - best cells:

10. On the V oc loss: How is the absorber doped? Less As (by ~0.6 times) incorporated into CdTe absorber for ZnO/ITO substrate Lower dopant concentration correlates well with carrier density (N a ) profiles MOCVD growth kinetics are likely to differ on ITO and ZnO/ITO surfaces (due to roughness, nucleation behaviour, etc.) Reduction in N a would lead to lower V oc (Theory: J. Sites & J. Pun, Thin Solid Films, 2007; Experiment: G. Kartopu et al., PiP, 2015)

11. Modelling results V oc & J sc remain nearly constant while FF expected to improve (by 1.5%) Reducing carrier density to 0.6 times lowers V oc (by ~10mV) & enhances J sc (by 0.1 mA·cm 2 ) in line with experimental observations J sc enhancement with modelling too small compared to experiment  quality of absorber improved (leading to lower bulk recombination)? or  an optical enhancement occurred, such as increased transmission for CdZnS/ZnO/ITO vs. CdZnS/ITO stack (allowing more photons to absorber)? J-V parameters calculated by SCAPS for CdZnS/CdTe solar cells with ITO and HRT/ITO front contacts (ZnO mobility = 60cm 2 /V·s; carr. density = 1E17/cm 3 )

12. Discussion  Alternative explanation to J sc enhancement with the use of a ZnO HRT buffer layer: UPS data show that “CdS/ITO” interface has high barrier height of 0.9eV, but this splits to smaller steps for “CdS/ZnO/ITO” (0.5eV at CdS/ZnO & 0.4eV at ZnO/ITO) (L. Tingliang et al., J. Semicon., 2012) This suggests lower probability for “interfacial recombination” & higher J sc  Effect of ZnO mobility & carrier density: Only FF noticeably affected with the properties of ZnO & controlled the efficiency Carrier density more influential than the mobility

13. Conclusions  Effect of an ALD ZnO film as HRT layer on MOCVD CdTe PV cells studied  For ZnO/ITO (150nm CdZnS) J sc up by 2 mA·cm 2 V oc down by 20mV FF up by 4% & ɳ up by 1.7%  J sc enhancement linked to better collection at long  For ZnO/ITO cells spectral response more uniform & stable  V oc drop linked to higher density of As & carriers in bulk CdTe  SCAPS supports experimental observations qualitatively well

14. Thank you for listening Team effort Funding / support Acknowledgements