1. Radiotracer diffusion of copper and potassium in Cu(In,Ga)Se2 (CIGS) thin-film solar cells
N. A. Stolwijk, M. Wegner, F. Hergemöller, S. Divinski, G. Wilde
Institut für Materialphysik, Universität Münster
H. Wolf, M. Deicher
Experimentalphysik, Universität des Saarlandes, Saarbrücken
R. Würz
Zentrum für Sonnenenergie- und Wasserstoff-Forschung Baden-Württemberg, Stuttgart (ZSW)
0.5 min.
It is a pleasure to present our proposal titled …
This a joint proposal of the University of Münster with the Saarbrücken group and the Center for Solar Energy ZSW in Stuttgart.
INTC-P-476
CERN-INTC

2. 2015: New World Record for CIGS Solar Cells: 22.6% certified
 cert. = % with ARC
Voc = mV
FF = %
jsc = mA/cm²
Current (mA)
Power (mW)
0.75 min:
CIGS solar cells have a high efficiency of energy conversion:
The new world record is 22.6%. It was achieved by our cooperation partner ZSW in Stuttgart.
The improvement is due to K which is used in addition to Na as the standard „doping“ element.
Voltage (mv)
Improvement due to K used in addition to Na as the standard „doping“ element

3. Solar Cell Structure ZnO:Al (1.0 µm) i:ZnO (0.05 µm) CdS (0.05 µm)
Polycrystalline CIGS layers:
d ≈ 2 µm
Mo (0.5 µm)
ZnO:Al (1.0 µm)
CdS (0.05 µm)
Glas substrate (3 mm)
CIGS (2.0 µm)
i:ZnO (0.05 µm)
0.5 min:
The solar cell is made up of a glas substrate, a metallic contact layer, the active CIGS zone, a CdS buffer layer, and a cap layer consisting of a tranparent conducting oxide.
The CIGS layer is 2 µm thick with grains of about 2 µm in size.

4. Cu(In,Ga)Se2 Crystal Structure
IV II-VI I-III-VI2
0.5 min:
CIGS crystallizes in the chalcopyrith structure which has a diamond-like base lattice.
Compared to the zinc blende structure the group-II element is replaced in equal amounts by group-I and III elements, that is Cu and In.
In a next complexity step, about 1/3 of In is replaced by Ga.

5. Important for optimizing CIGS processing and solar-cell efficiency
Diffusion Exps. in CIGS: Radiotracer Technique
Important for optimizing CIGS processing and solar-cell efficiency
CIGS Mo
Glas
etching
0.75 min:
Diffusion experiments in CIGS are important for optimizing the solar-cell efficiency
In the radiotracer technique we start with samples having a free CIGS surface.
We deposit a tracer, and perform a high-T annealing treatment.
Depth profiling involves sectioning and counting the activity of each section.
DPG Berlin,

6. Diffusion Profiles of 65Zn & 110mAg in CIGSMo
Glas
CIGS
source
65Zn
110mAg
0.75 min:
Here you see 2 examples of diffusion profiles in CIGS.
Both the Zn and Ag profile result from a diffusion treatment at 375 °C.
The Ag profile is deeper than the Zn profile despite a shorter diffusion time, 5 min. against 15 min.
Thus Ag has a higher diffusivity. This is also seen in the next plot.
Zn = front-layer element: ZnS buffer / ZnO transparent oxide
Ag = alloying element: substitute for Cu in (Cu,Ag)(In,Ga)Se2

7. Master Plot: Impurity Diffusion in CIGS
ZSW Stuttgart
Münster group
accommodated by Cu sublattice
0.5 min:
This masterplot collects all diffusivity data obtained by the Münster group.
For comparison, data for Na from ZSW are shown.
All these elements preferably reside on the Cu sublattice.
Cd = front-layer element: standard CdS buffer layer
Fe = substrate element: steel foil as flexible substrate
Na = beneficial impurity: improves solar-cell efficiency

8. Cu diffusion in CIGS
Self-diffusivity on Cu sublattice important for interpretation of impurity diffusion
Cu 1, Cu 2, Cu 3:
various authors,
various (electrical) methods
64Cu:
tracer experiment
bulk single-crystal CIS
Lubomirsky et al., JAP 83 (1998) 4678
0.75 min:
However, there are no reliable data for Cu diffusion in CIGS.
The available data mostly are obtained by indirect methods and vary over many orders of magnitude.
The only tracer experiment was done in a CIS single crystal at one single temperature .
But: Self-diffusivity on the Cu sublattice is important for the interpretation of impurity diffusion
No reliable data exist for Cu diffusion in CIGS !

9. Planned Experiment 1: 67Cu/64Cu Diffusion in CIGS
Self-diffusivity on Cu sublattice data needed to identify diffusion mechanisms
67Cu: t1/2 = 2.6 d ϒ = MeV 49% abundance
UCx target 3.5×108 ions/µC < 30 min/sample
ZrO ×
Half-life time allows for double-tracked experiments:
On-site with ISOLDE/ODC
Off-site at Münster Laboratory (after implantation at ISOLDE)
64Cu: t1/2 = 12.7 h ϒ = MeV 0.5% abundance
UCx target 2.0×108 ions/µC 30 min/sample
0.75 min:
Therefore, we plan to do experiments with Cu isotopes provided by ISOLDE.
The most favorable isotope is Cu-67 with a half-life of 2.6 days.
We prefer to do on-site experiments here at ISOLDE using the Online Diffusion Chamber.
Beyond that, it is also possible to send implanted samples to Münster.
As an alternative, we could take Cu-64, but ist specifications make it less suitable.
Only on-site experiments possible
64Cu less suitable than 67Cu
Beam request: 3 shifts over 2 years

10. Planned Experiment 2: 43K Diffusion in CIGS
Potassium data needed to optimize solar-cell efficiency
43K: t1/2 = 22.2 h ϒ = MeV 87% abundance
UCx target 2.4×107 ions/µC 30 min/sample
On-site experiments preferable
Feasibility demonstrated by exps. on feldspar (July 2016)
0.5 min:
A second experiment concerns the diffusion of K-43 in CIGS.
These data are needed to further optimize the energy-conversion efficiency of CIGS solar cells.
K-43 has a half-life of 22 hours.
So the experiments should be done On-site with the Online Diffuson Chamber.
Beam request: 3 shifts over 2 years

11. 43K Diffusion in Single-Crystal Alkali Feldspar ⊥ (001) (using ISOLDE/ODC facilities, July 2016)
F. Hergemöller, M. Wegner et al., Phys. Chem. Minerals, submitted
Diffusion profiles
Diffusion coefficients
0.5 min:
In fact, we have direct experience with K-43 at ISOLDE.
This concerns experiments on single-crystal alkali feldspar that were done 4 months ago.
Here you see the diffusion profiles measured with the ODC equipment.
We found that the diffusivity of potassium is much smaller than that of sodium.
VF = Volkesfeld (Eifel, Germany), single crystal ⊥ (001) BM = Benson Mines (USA), randomly oriented grains
Conclusion:
DK << DNa

12. On-line Diffusion Chamber (ODC) … to be used in Off-line modus
Concept & Construction: Saarbrücken Group
0.5 min:
Maybe I should emphasize that we have used the On-line Diffusion Chamber in Off-line modus.
To this aim, the equipment was reinstalled in the Solid State Physics Lab.
Figure 1. A representation of the ODC's set-up and the location of its peripherals.
User Guide: Jaime E. Avilés Acosta

13. Resumé Overall request: 6 shifts over 2 years
Our offer: support for ODC maintainance by Münster groups
With thanks to:
Saarbrücken group: Manfred Deicher, Herbert Wolf
ISOLDE team: Karl Johnston, Juliana Schell
0.5 min:
In summary, we request 6 shifts over 2 years.
In turn, we will give support to keep the ODC equipment in a good state of operation.
Finally, I would like to thank the Saarbrücken group and the ISOLDE team for their help over the last year.
Thank you for your attention.

14. Interstitial-Substitutional Diffusion
(Prominent in semiconductors such as Ge, Si, GaAs, etc.) Ai V As
Dissociative mechanism: Ai + V  As
Ai = minority species, fast interstitial, responsible for impurity transport
As = majority species, virtually immobile, determines solubility ( )
V = vacancy, mediates interstitial-substitutional exchange

15. Interstitial-Substitutional DiffusionAi V As
Dissociative mechanism: Ai + V  As Ai As I
Kick-out mechanism: Ai  As + I

16. CIGS Layer & Surface Structure
Side view SEM image of the CIGSe/Mo/soda-lime glass layer structure of a diffusion sample. The CIGS layer thickness is 2.34 mm and the grain size is approximately 2 µm
AFM measurement of the surface topology of the CIGSe layer (RRMS ~ 140 nm).

17. Solar Cell Record Efficiencies
c-Si = 25.6 %1)
mc-Si = 20.8 %1)
CIGS = 22.6 %2)
CdTe = 22.1 %3)
1) M. A. Green et al., Progress in Photovolt. Res. Appl. 23 (19015) 805
2) P. Jackson et al., Phys. Status Solidi RRL, 10 (2016) 583 3)

19. Contamination of 67Cu with 67Ga (10-20%)
Master plot: diffusion in CIGS
67Cu (β-, ϒ) 67Zn: 2.6 d
67Ga (EC, ϒ)67Zn: 3.3 d
No discrimination by ϒ or t1/2
Expected: DCu >> DGa
Discrimination by diffusion depth !

20. Master Plot: Impurity Diffusion in CIGS
accommodated by Cu sublattice
ZSW Stuttgart
Münster group
0.5 min: This masterplot collects all diffusivity data obtained by the Münster group. For comparison, data for Na from ZSW are shown. All these elements preferably reside on the Cu sublattice.
Cd = front-layer element: standard CdS buffer layer
Fe = substrate element: steel foil as flexible substrate
Na = beneficial impurity: improves solar-cell efficiency