LETTERS PUBLISHED ONLINE: 3 NOVEMBER 2013 | DOI: 10.1038/NMAT3789
Potassium-induced surface modification of Cu(In,Ga)Se2 thin films for high-efficiency solar cells Adrian Chiril˘a1 *† , Patrick Reinhard1† , Fabian Pianezzi1 , Patrick Bloesch1 , Alexander R. Uhl1 , Carolin Fella1 , Lukas Kranz1 , Debora Keller1 , Christina Gretener1 , Harald Hagendorfer1 , Dominik Jaeger2 , Rolf Erni3 , Shiro Nishiwaki1 , Stephan Buecheler1 and Ayodhya N. Tiwari1 Thin-film photovoltaic devices based on chalcopyrite Cu(In,Ga)Se2 (CIGS) absorber layers show excellent light-topower conversion efficiencies exceeding 20% (refs 1,2). This high performance level requires a small amount of alkaline metals incorporated into the CIGS layer, naturally provided by soda lime glass substrates used for processing of champion devices3 . The use of flexible substrates requires distinct incorporation of the alkaline metals, and so far mainly Na was believed to be the most favourable element, whereas other alkaline metals have resulted in significantly inferior device performance4,5 . Here we present a new sequential post-deposition treatment of the CIGS layer with sodium and potassium fluoride that enables fabrication of flexible photovoltaic devices with a remarkable conversion efficiency due to modified interface properties and mitigation of optical losses in the CdS buffer layer. The described treatment leads to a significant depletion of Cu and Ga concentrations in the CIGS near-surface region and enables a significant thickness reduction of the CdS buffer layer without the commonly observed losses in photovoltaic parameters6 . Ion exchange processes, well known in other research areas7–13 , are proposed as underlying mechanisms responsible for the changes in chemical composition of the deposited CIGS layer and interface properties of the heterojunction. Photovoltaic devices, which directly convert abundant light into solar electricity, are important for the implementation of renewable energy supply systems. Today, silicon-based modules are dominating the photovoltaic market14 , but various emerging technologies based on thin-film inorganic semiconductors15 , dyesensitized materials16 and organic materials17 are rapidly progressing. Thin-film photovoltaic devices can be manufactured on flexible substrates, enabling the employment of roll-to-roll deposition techniques that have already revolutionized the food-packaging industry. Besides cost-reduction, flexible photovoltaic devices allow new opportunities in areas such as building integration, portable electronics, and space applications. However, the conversion efficiency of flexible solar cells has always been—until now—significantly lower compared with photovoltaic devices on rigid substrates18 . Amongst thin-film technologies, CIGS has yielded maximum conversion efficiencies of up to 20.3% with rigid soda lime glass (SLG) as a substrate2 . SLG substrates were so far favoured because
they allow high processing temperatures of 600 ◦ C or even more, have a coefficient of thermal expansion similar to CIGS, and provide the right amount of Na necessary for high-efficiency devices directly during film growth19 . In contrast, polyimide film as a flexible substrate can withstand processing temperatures only of well below 500 ◦ C, posing significant constraints on growing highquality CIGS layers, have a large thermal mismatch with CIGS and do not contain alkaline elements, which necessitates an adequate extrinsic supply method19 . Recently, we have shown that several of the coercive aspects of polyimide substrates described above can be overcome as we demonstrated an 18.7%-efficiency flexible solar cell on polyimide, which represented the highest reported value for any type of solar cell grown on a flexible substrate20 . We used a NaF postdeposition treatment (PDT) of the CIGS layer to supply Na after the growth of the CIGS (ref. 21) whereas other strategies focus on Na addition using NaF deposition before or during CIGS growth19 . Improvements in cell efficiency are achieved irrespective of the methods used for addition of Na, but the PDT treatment yielded the highest gains, especially at the low growth temperatures that are required for polyimide films as substrate21 . SLG substrates contain a vast amount of Na and a lower amount of K in oxide form and these alkaline elements are known to diffuse into CIGS during its growth. Although K has been frequently observed in CIGS, its influence on the properties of the final device has mostly been neglected or considered as similar to the effects stemming from Na (refs 22–26). Up to now, it was commonly believed that Na alone is sufficient for the improved electronic quality of CIGS devices. Enamelled steel substrates, which contain a high amount of K and a lower amount of Na, have indicated the potential for further improvements in device performance when the substrate contains a high amount of K (refs 18,26). Here, we demonstrate a distinctly beneficial role for further addition of K in CIGS layers subjected to prior addition of Na with a PDT. This resulted in a flexible solar cell with a new remarkable efficiency of 20.4%. K is added by simultaneous coevaporation of KF and Se directly after CIGS deposition and Na incorporation (see Methods). The secondary-ion mass spectroscopy (SIMS) profile of the sample with additional K treatment shown in Fig. 1a reveals a higher count rate of K compared with Na throughout the
1 Laboratory
for Thin Films and Photovoltaics, Empa—Swiss Federal Laboratories for Materials Science and Technology, Ueberlandstrasse 129, CH-8600 Duebendorf, Switzerland, 2 Nanoscale Materials Science, Empa—Swiss Federal Laboratories for Materials Science and Technology, Ueberlandstrasse 129, CH-8600 Duebendorf, Switzerland, 3 Electron Microscopy Center, Empa—Swiss Federal Laboratories for Materials Science and Technology, Ueberlandstrasse 129, CH-8600 Duebendorf, Switzerland. † These authors contributed equally to this work. *e-mail: [email protected] NATURE MATERIALS | VOL 12 | DECEMBER 2013 | www.nature.com/naturematerials
© 2013 Macmillan Publishers Limited. All rights reserved.
1107
1108
100,000
Mo
CdS
ZnO:AI i-ZnO
a
71Ga 113In
Intensity (counts)
10,000
27AI
39K 63Cu
64Zn
1,000
23Na
100 114Cd
10
0
5
98Mo
10
15 20 25 Sputter time (min)
30
35
b 100,000
71Ga 113In
27AI
10,000
Intensity (counts)
CIGS absorber layer. Using inductively coupled plasma mass spectroscopy (ICPMS) measurements, the amount of K was determined to be 1,700 ppm whereas the amount of Na was estimated to be about 50 ppm with respect to the CIGS layer. Figure 1b shows the SIMS profile for a representative sample that was not subjected to the additional KF-PDT. A significantly higher count rate of Na compared with K is measured, and it reveals a much higher count rate of Na compared with the SIMS profile of the sample with additional K treatment. These results show that the KF-PDT leads to a significant incorporation of K and removal of Na in the CIGS layer, which can be explained by an ion exchange mechanism that favours the presence of K over Na at the given KF deposition conditions. Ion exchange phenomena have been known for a long time7 and are used in a broad range of material systems and applications8,9 . For example, the exchange of Na with K is used for chemical strengthening of glass10 or tuning of its refractive index11 . Heterojunction formation has also been achieved by ion exchange methods, for example in the development of Cux S/CdS thin-film solar cells by topotactic conversion of the surface of a CdS layer12 . Ion exchange reactions between Cd and Cu were also observed at the CdS/CIGS interface13 . The surface composition of a bare CIGS absorber without alkaline treatment, as well as after NaF-PDT or after KF-PDT, was investigated by X-ray photoelectron spectroscopy (XPS). Figure 2a illustrates the three differently prepared samples. To remove possible NaF- or KF-related remnants from the surface, the samples subjected to NaF- or KF-PDT were etched before the XPS measurements (see Methods). No clear difference in the surface morphology due to the different PDT treatments is observed (Supplementary Fig. 1). Spectra of the relevant elements are presented in Fig. 2b–e. After treatment of the CIGS surface with only NaF-PDT the Cu and Ga peaks decrease by a factor of about 1.3 as compared with the sample that did not undergo any PDT, whereas the In and Se peaks remain nearly unchanged. KF-PDT leads to a much stronger reduction of the Ga peak intensity, the disappearance of the Cu peak, and little or no significant effect on Se and In peak intensities, respectively. Sputtering of approximately 20 nm is necessary to obtain a similar composition compared to the case when no KF is added as shown for the Cu 2p3/2 peak in Fig. 2f. Moreover, K is clearly detected in the first few nanometres of the CIGS surface as shown in Fig. 2g for the K 2p3/2 peak, whereas no Na is detected in the case of the CIGS absorber subjected to NaF-PDT (not shown). These measurements suggest that the changes of the CIGS surface chemical composition induced by the KF-PDT are limited to a thickness below 30 nm. To compare the near-surface region of the CIGS before and after buffer layer deposition, XPS measurements were performed on CIGS layers subjected to only NaF-PDT or only KF-PDT and covered with a thin CdS layer. The CdS layer was removed by sputtering as illustrated in Fig. 2h. The obtained spectra from the CdS/CIGS interface region are shown in Fig. 2i–l. In the case of the KF-PDT, a similar depletion of Cu and Ga is observed, revealing that the altered surface properties owing to the KF-PDT remain present after the CdS deposition process. Furthermore, K is again clearly detected at the interface, as shown in Fig. 2m with spectra recorded at different depths ranging from the CdS surface towards CIGS bulk, proving that it is not completely removed during the CdS deposition process. In contrast, the Na concentration at the interface was below the detection limit of XPS. Furthermore, it is found that the XPS peak positions, respectively binding energies, of In and Se shift in a different way on NaF or KF addition, but not sufficiently to support the formation of a new type of chemical bond (Supplementary Fig. 2). We conclude that besides the partial ion exchange of Na with K within the CIGS film, the KF-PDT also induces a modification of the chemical composition profile near the surface, depleting
CIGS
NATURE MATERIALS DOI: 10.1038/NMAT3789
LETTERS
64Zn 23Na
1,000
63Cu
114Cd
100
98Mo
39K
10 0
5
10
15 20 Sputter time (min)
25
30
Figure 1 | Elemental depth profiling. a,b, SIMS profiles showing all relevant elements from the front contact (ZnO) to the back contact (Mo) obtained from representative devices with addition of NaF and KF after the CIGS growth (a), and addition of NaF alone (b). In a, the significantly high count rate of 39 K, which is about 25 times higher than the count rate of 23 Na, indicates that a comparably high amount of K is incorporated within the device. In b, if no KF is added, clearly a higher count rate of 23 Na is observed compared with 39 K.
mostly Cu and Ga, with implantation of a significant amount of K. Different mechanisms and concepts have been discussed in the literature to explain the Cu depletion observed at CIGS surfaces. Among others, electrochemically driven Cu migration into the bulk due to increased Fermi energy at the surface27 and oxygenation-induced Cu redistribution28 are frequently used for explaining this phenomenon. Even if significant oxygenation in high vacuum during the KF-PDT is unlikely, the alkali content of the surface could enhance oxygenation during transfer in air to the buffer layer deposition. For an efficient photovoltaic device, in view of the modified surface chemistry of the CIGS absorber due to the KF-PDT, it is necessary to adapt the interface formation with the subsequent buffer layer. Generally, the best performance of CIGS solar cells is obtained using a CdS buffer layer deposited by chemical bath
NATURE MATERIALS | VOL 12 | DECEMBER 2013 | www.nature.com/naturematerials © 2013 Macmillan Publishers Limited. All rights reserved.
NATURE MATERIALS DOI: 10.1038/NMAT3789 a
No PDT
NaF
LETTERS h
KF
NaF
KF
CdS CIGS
Mo
Mo
KF
KF
448
e
Ga 2p3/2
No PDT NaF
No PDT NaF
931
930
Intensity (a.u.)
Intensity (a.u.) Se 3d 5/2
NaF
KF
1,121
1,119 1,117 Binding energy (eV)
m
K 2p 3/2 (KF)
K 2p 3/2
Sputtering time (approx. depth) 10 min (30 nm)
0 min (0 nm)
1,115
59
57 55 53 51 Binding energy (eV)
K 2p 3/2 (KF)
Se (Auger)
225
K 2p3/2
231 229 227 Binding energy (eV)
Se (Auger)
Intensity (a.u.)
NaF
CdS
0 min (0 nm)
932
446 444 442 Binding energy (eV)
CIGS
10 min (30 nm)
933
233
g
Binding energy (eV)
448
l
Se (Auger)
1,115
Sputtering time (approx. depth)
934
930
KF
Cu 2p3/2 (KF)
935
NaF KF
Ga 2p3/2
KF
Intensity (a.u.)
f
1,119 1,117 Binding energy (eV)
935 934 933 932 931 Binding energy (eV)
k
KF
1,121
NaF KF
446 444 442 Binding energy (eV)
Se 3s
In 3d 5/2
Intensity (a.u.)
NaF
934 933 932 931 930 Binding energy (eV)
Intensity (a.u.)
No PDT
NaF
j
Cu 2p3/2
Intensity (a.u.)
d
Intensity (a.u.)
Intensity (a.u.)
No PDT
935
i
In 3d 5/2
Intensity (a.u.)
c
Cu 2p3/2
Intensity (a.u.)
b
CIGS
300 298 296 294 292
306 302 298 294 290 286 282
Binding energy (eV)
Binding energy (eV)
Figure 2 | Surface chemical analysis. a, Schematic view of the three investigated absorbers. The purple layer on the KF absorber indicates the modified surface composition. b–e, XPS peak of Cu 2p3/2 , In 3d5/2 , Ga 2p3/2 and Se 3s, respectively, obtained from the surface of CIGS absorbers with no alkali evaporation (no PDT), only NaF addition and only KF addition. f, Sputtering of the CIGS absorber subjected to KF-PDT shows the appearance of the Cu 2p3/2 peak within the first approximately 20 nm with similar intensity as in the case of no PDT. g, K is clearly measurable at the surface up to a depth of approximately 20 nm. h, Schematic view of two absorbers measured after sputtering through the CdS layer. i–l, XPS peak of Cu 2p3/2 , In 3d5/2 , Ga 2p3/2 and Se 3s, respectively, at the CdS/CIGS interface with only NaF addition and only KF addition. m, XPS spectra of K at different sputtering depths.
deposition (CBD). However, the relatively low bandgap energy of about 2.4 eV leads to optical losses in the low-wavelength region of the solar spectrum. A way to reduce losses induced by CdS is to reduce its layer thickness. In the present work, we have identified the absorber modified with the KF-PDT to facilitate significant reduction of the CdS layer thickness without the commonly observed decrease in open-circuit voltage (VOC ) and fill factor values of the solar cell6 . Figure 3 presents external quantum efficiency (EQE) measurements of solar cells with absorbers subjected to the KF-PDT and different CBD durations, respectively CdS thicknesses. Up to now, state-of-the-art processes required CdS thicknesses of 40 nm or more22 , which is comparable to 20–22 min deposition time with our recipe (see Methods). However, the KFPDT enables a significant reduction in deposition duration down
to 14 min without any deterioration in device performance with efficiencies exceeding 18%. The gain in current density arising from better spectral response is apparent in Fig. 3 where the contribution to the short circuit current density (JSC ) from absorption below 540 nm wavelength is shown in the inset. This correlates well with the reduction of the CdS layer thickness calculated from ICPMS (inset) or observed in transmission electron microscopy (Supplementary Fig. 3). A minimum amount of evaporated KF is found to be required to allow the reduction of CdS thickness with maintained high efficiency. A pronounced deterioration of the photovoltaic parameters is observed if the duration of the KF-PDT is reduced to below 15 min (Supplementary Fig. 4). The changes in CIGS surface chemical composition induced by the KF-PDT, especially
NATURE MATERIALS | VOL 12 | DECEMBER 2013 | www.nature.com/naturematerials
© 2013 Macmillan Publishers Limited. All rights reserved.
1109
NATURE MATERIALS DOI: 10.1038/NMAT3789
LETTERS 90
70
Current density < 540 nm (mA cm¬2)
60 50 40 30 20 10 0 300
400
7.0
70 60
6.5
50 6.0
40
5.5
30 20
5.0 4.5
500
10 10 12 14 16 18 20 22 CBD time (min)
600
700
800
12 35
CdS thickness (nm)
External quantum efficiency (%)
80
a Voc = 736 mV Jsc = 35.1 mA cm¬2
25
FF = 78.9% η = 20.4%
8
20
6
15 4 10
0
0
900 1,000 1,100 1,200
10
2
5
b
0
100
200
300 400 500 Voltage (mV)
600
700
0
100 90
Figure 3 | Variation of CdS film thickness. EQE measurements from devices grown with various CBD deposition durations of 10 min (blue) up to 22 min (yellow). The inset figure shows the related contribution to JSC from photons with wavelength below 540 nm together with corresponding CdS layer thickness as determined with ICPMS.
the Cu deficiency, alter the interface properties of the resulting p–n junction in a way that enables a shorter CdS deposition time and therefore CdS thickness. This can be explained on the basis of an earlier model29 that describes the role of CBD–CdS, suggesting that during the CBD process Cd tends to occupy Cu vacancy sites, with the CdCu donors causing electronic inversion of the CIGS surface and forming a p–n homojunction in the CIGS layer. This inversion requires a certain density of CdCu that depends on the density of Cu vacancies and conditions of CBD to fill the Cu vacancies with Cd. A recent theoretical study further supports this model30 . With a similar CBD duration, XPS measurements of the CdS/CIGS samples discussed above reveal a higher Cd content in the near-surface region of the CIGS absorber subjected to the KF-PDT (Supplementary Fig. 5). Therefore, it seems that in a shorter duration a higher density of CdCu is achieved by diffusion of Cd into the Cu vacancies owing to the Cu-depleted CIGS surface composition. This can explain the possibility to reduce the CdS deposition time while maintaining a high junction quality, as evidenced by the efficiency above 18% for the cells with a CdS deposition time longer than 14 min shown in Fig. 3. Finally, improvements in the chemical composition of the overall CIGS layer, interface modification with additional KF, and overall stack and process optimization enabled the fabrication of a η = 20.4% efficiency device on polyimide substrate, representing the highest efficiency for the CIGS technology. Figure 4a presents the certified current–voltage and power–voltage measurements and Fig. 4b shows the corresponding EQE measurement. The EQE measurement of our previously reported best device20 is also given to highlight the improved spectral response related to the thinner CdS layer thickness. In summary, we have shown that the roles of Na and K in CIGS have to be differentiated, especially when being supplied after the growth of the CIGS layer. We show an approach of sequential in situ addition of both Na and K in CIGS after the layer growth that leads to a significant reduction in Na content in the CIGS bulk and a strong depletion of Cu and Ga in the CIGS surface region, confined to a depth of less than 30 nm, without any significant change in the microstructure. The KF-PDT-induced CIGS surface modification
External quantum efficiency (%)
Wavelength (nm)
1110
30
Power (mW)
10 min 12 min 14 min 16 min 18 min 20 min 22 min
Current density (mA cm¬2)
100
80 70
η = 18.7%
60 50 40 30 20 10 0 300
400
500
600 700 800 900 1,000 1,100 1,200 Wavelength (nm)
Figure 4 | Characteristics of the best device. a,b, J–V (diamonds), P–V (triangles) measurements (a) and EQE (b) of the new champion device (area = 0.52 cm2 ) obtained on flexible polyimide substrate (solid line). The EQE of the previous best device20 is also shown for comparison (dashed line). The solar cells were independently certified by Fraunhofer ISE, Freiburg, Germany. FF, fill factor.
facilitates Cd diffusion in the Cu-depleted surface of the absorber, and results in an improved CIGS/CdS heterojunction quality, even with a shorter CdS deposition time, respectively CdS layer thickness. Concepts of ion exchange reaction can explain changes of the K and Na concentration in the CIGS layer and the interface formation with the buffer layer. This new treatment led us to increase the highest efficiency of flexible CIGS solar cells on plastic to an independently certified 20.4% efficiency. This is the first time that the efficiency of a flexible solar cell on plastic has surpassed the cell efficiency on rigid glass, and it even matches the performance of the best polycrystalline Si wafer-based device, which is considered as a benchmark for photovoltaic technologies. Furthermore, the shown interaction and effects of different alkaline metals, based on the solid-state ion exchange mechanism, opens up new research directions not only in the field of thin-film photovoltaics but as well in various types of material system.
Methods Device preparation. Devices were fabricated on commercially available polyimide film on the basis of the methods described previously20 . CIGS layers were grown by co-evaporation from elemental effusion cells in a high-vacuum chamber (base pressure ∼10−8 hPa) equipped with an additional effusion cell for KF. The addition of KF was carried out directly after the NaF-PDT (∼1.3 nm min−1 for 20 min) by evaporating KF (∼1 nm min−1 , 15 min for optimal KF treatment) in the presence
NATURE MATERIALS | VOL 12 | DECEMBER 2013 | www.nature.com/naturematerials © 2013 Macmillan Publishers Limited. All rights reserved.
NATURE MATERIALS DOI: 10.1038/NMAT3789 of Se onto the finished CIGS layer at a substrate temperature of about 350 ◦ C. The average [Ga]/([In] + [Ga]), [Cu]/([Ga] + [In]) and film thickness of the CIGS layers are in the range of 0.33–0.38, 0.78–0.82 and 2.5–3.0 µm, respectively, as determined by means of X-ray fluorescence measurements. Devices are finished by CBD of a CdS buffer layer (see details below), radiofrequency sputtering of an i-ZnO/ZnO:Al window layer and electron-beam evaporation of Ni/Al/Ni grids. All cells are covered with a MgF2 anti-reflection coating and single cells are defined by mechanical scribing. CdS deposition. The CdS buffer layer was grown by a CBD method. A Cd acetate (Cd(CH3 COO)2 ) aqueous solution (0.028 M) and ammonium hydroxide aqueous solution (NH4 OH, 14.5 M) are first mixed together with high-purity water (18 M cm) in a volume ratio of 3:7:37 and preheated in a water bath at 70 ◦ C for 2 min. Thiourea (SC(NH2 )2 ) aqueous solution (0.374 M) is then added and the samples are immersed therein to begin the actual deposition. The whole solution is placed into a 70 ◦ C water bath under slow stirring until the desired film thickness is reached. The samples are then washed with high-purity water and dried with an ionized N2 gun, before an annealing step at 180 ◦ C in air for 2 min. SIMS. Depth profiling data were obtained with a time-of-flight SIMS 5 system from ION-TOF. Bi+ ions were used as primary ions and positive ions were detected. Sputtering was performed using O+ 2 sputtering ions with 2 keV ion energy, 400 nA ion current and a 300 × 300 µm2 raster size. An area of 100 × 100 µm2 was analysed + using Bi ions with 25 keV ion energy. ICPMS. For ICPMS analysis, approximately 1 cm2 of CIGS material was detached from the thin-film solar cell at the Mo/CIGS interface, directly transferred into 50 ml trace metal-free polyethylene tubes and fully dissolved in a mixture of 10 ml HNO3 and 1 ml HCl. After filling to 50 ml with 18 M cm deionized water, the sample was not further diluted for analysis. Metal analyses were performed on an Agilent 7500ce ICPMS with external calibration using certified metal standards (1,000 µg ml−1 , Merck CertIPUR). For quality assurance, analysis of reference materials (NIST SRM 1643e) and spiking experiments were performed, with recoveries between 90 and 110%. XPS. All XPS measurements were performed in a Quantum2000 from Physical Electronics using a monochromatized Al Kα source (1,486.6 eV). The measurements were recorded in the fixed analyser transmission mode, under an angle of emission of 45◦ and at an instruments base pressure of below 5×10−9 mbar. Spectra were recorded with a pass energy of Ep = 58.7 eV and an energy step size of 1E = 0.125 eV. Ar+ ions and electrons were used to compensate possible surface charging. Depth profiles have been obtained using an ion energy of 1 keV over a range of 2 mm × 2 mm. The work function at 45◦ has been adjusted in such a way that the binding energy of Au 4f7/2 is found at 83.96 eV with a full-width at half-maximum of 0.8 eV; the binding energies of gold have been measured regularly. Before the measurements, the surface of the CIGS bare absorbers subjected to a PDT are etched with a low-concentration ammonia aqueous solution to remove any unnecessary NaF or KF compound from the surface. EQE. EQE of solar cells was measured with a lock-in amplifier. A chopped white light source (900 W, halogen lamp, 360 Hz) and a dual grating monochromator generated the probing beam. The beam size was adjusted such that the illumination area was smaller than the device area. A certified monocrystalline Si solar cell from Fraunhofer ISE was used as the reference cell. Cell temperature was controlled at 25 ◦ C with Peltier cooling and white light bias was applied. Solar cell performance measurements. J –V characteristics of solar cells were measured under simulated standard-test conditions (25 ◦ C, 1,000 W m−2 , AM 1.5 G illumination) in a sun-simulator with a HMI light source. A Keithley 2400 source meter with four-terminal sensing was used to acquire J –V characteristics. J –V and EQE characteristics of the solar cell with 20.4% efficiency were independently certified by Fraunhofer ISE, Freiburg, Germany.
Received 14 May 2013; accepted 25 September 2013; published online 3 November 2013
References 1. Green, M. A., Emery, K., Hishikawa, Y., Warta, W. & Dunlop, E. D. Solar cell efficiency tables (version 41). Prog. Photovolt. 21, 1–11 (2013). 2. Jackson, P. et al. New world record efficiency for Cu(In,Ga) Se2 thin-film solar cells beyond 20%. Prog. Photovolt. 19, 894–897 (2011). 3. Hedstrom, J. et al. 23rd IEEE Phot. Spec. Conf. (PVSC) 364–371 (IEEE, 1993). 4. Contreras, M. A. et al. 26th IEEE Phot. Spec. Conf. (PVSC) 359–362 (IEEE, 1997). 5. Bodegård, M., Granath, K. & Stolt, L. Growth of Cu(In,Ga) Se2 thin films by coevaporation using alkaline precursors. Thin Solid Films 361, 9–16 (2000).
LETTERS 6. Contreras, M. A. et al. Optimization of CBD CdS process in high-efficiency Cu(In,Ga) Se2 -based solar cells. Thin Solid Films 403–404, 204–211 (2002). 7. Jenny, H. Studies on the mechanism of ionic exchange in colloidal aluminium silicates. J. Phys. Chem. 36, 2217–2258 (1932). 8. Clearfield, A. Role of ion exchange in solid-state chemistry. Chem. Rev. 88, 125–148 (1988). 9. Rivest, J. B. & Jain, P. K. Cation exchange on the nanoscale: An emerging technique for new materials synthesis, device fabrication, and chemical sensing. Chem. Soc. Rev. 42, 89–96 (2013). 10. Nordberg, M. E. et al. Strengthening by ion exchange. J. Am. Ceram. Soc. 47, 215–219 (1964). 11. Zou, J. et al. Two-step K+ –Na+ and Ag+ –Na+ ion-exchanged glass waveguides for C-band applications. Appl. Opt. 41, 7620–7626 (2002). 12. Das, S. R. et al. The preparation of Cu2 S films for solar cells. Thin Solid Films 51, 257–264 (1978). 13. Ramanathan, K. et al. Proc. 2nd World Conf. on Photovoltaic Solar Energy Conversion 477–481 (Joint Research Centre, European Commission, 1998). 14. Wolden, C. A. et al. Photovoltaic manufacturing: Present status, future prospects, and research needs. J. Vac. Sci. Technol. A 29, 030801 (2011). 15. Chopra, K. L., Paulson, P. D. & Dutta, V. Thin-film solar cells: An overview. Prog. Photovolt. 12, 69–92 (2004). 16. Grätzel, M. Photoelectrochemical cells. Nature 414, 338–344 (2001). 17. Bernede, J. C. Organic photovoltaic cells: History, principle and techniques. J. Chil. Chem. Soc. 53, 1549–1564 (2008). 18. Reinhard, P. et al. Review of progress toward 20% efficiency flexible CIGS solar cells and manufacturing issues of solar modules. IEEE J. Photovolt. 3, 572–580 (2013). 19. Kessler, F. & Rudmann, D. Technological aspects of flexible CIGS solar cells and modules. Sol. Energ. 77, 685–695 (2004). 20. Chirilă, A. et al. Highly efficient Cu(In,Ga) Se2 solar cells grown on flexible polymer films. Nature Mater. 10, 857–861 (2011). 21. Rudmann, D. et al. Sodium incorporation strategies for CIGS growth at different temperatures. Thin Solid Films 480, 55–60 (2005). 22. Jackson, P. et al. High quality baseline for high efficiency, Cu(In1−x ,Gax )Se2 solar cells. Prog. Photovolt. 15, 507–519 (2007). 23. Ishizuka, S., Yamada, A., Fons, P. & Niki, S. Flexible Cu(In,Ga) Se2 solar cells fabricated using alkali-silicate glass thin layers as an alkali source material. J. Renew. Sustain. Energ. 1, 013102 (2009). 24. Cojocaru-Mirédin, O. et al. Characterisation of grain boundaries in Cu(In,Ga) Se2 films using atom-probe tomography. IEEE J. Photovolt. 1, 207–212 (2011). 25. Abou-Ras, D. et al. Confined and chemically flexible grain boundaries in polycrystalline compound semiconductors. Adv. Energy. Mater. 2, 992–998 (2012). 26. Wuerz, R. et al. CIGS thin-film solar cells and modules on enamelled steel substrates. Sol. Energ. Mater. Sol. Cells 100, 132–137 (2012). 27. Klein, A. et al. Fermi level-dependent defect formation at Cu(In,Ga) Se2 interfaces. Appl. Surf. Sci. 166, 508–512 (2000). 28. Rau, U. et al. Oxygenation and air-annealing effects on the electronic properties of Cu(In,Ga) Se2 films and devices. J. Appl. Phys. 86, 497–505 (1999). 29. Nakada, T. & Kunioka, A. Direct evidence of Cd diffusion into Cu(In,Ga) Se2 thin films during chemical-bath deposition process of CdS films. Appl. Phys. Lett. 74, 2444–2446 (1999). 30. Kiss, J. et al. Theoretical study on the structure and energetics of Cd insertion and Cu depletion of CuIn5 Se8 . J. Phys Chem. C 117, 10892–10900 (2013).
Acknowledgements This work was supported by the Swiss National Science Foundation and the Swiss Federal Office of Energy. The laboratory for Nanoscale Materials Science and the Laboratory for Electronics/Metrology/Reliability at Empa are acknowledged for SIMS and scanning and transmission electron microscopy measurements, respectively.
Author contributions A.C., P.R., F.P., P.B., S.B., S.N. and A.N.T. designed the research and experiments. A.C., P.R. and P.B. fabricated the solar cells. A.C., P.R., F.P., S.B., A.R.U., C.F., C.G., H.H., D.J., L.K., D.K., R.E. and A.N.T. performed the characterization and analysis. A.C., P.R., F.P., S.B. and A.N.T. wrote the paper. All authors contributed to discussions.
Additional information Supplementary information is available in the online version of the paper. Reprints and permissions information is available online at www.nature.com/reprints. Correspondence and requests for materials should be addressed to A.C.
Competing financial interests The authors declare no competing financial interests.
NATURE MATERIALS | VOL 12 | DECEMBER 2013 | www.nature.com/naturematerials
© 2013 Macmillan Publishers Limited. All rights reserved.
1111
Potassium-induced surface modification of Cu(In,Ga)Se2 thin films for high-efficiency solar cells Adrian Chiril˘a1 *† , Patrick Reinhard1† , Fabian Pianezzi1 , Patrick Bloesch1 , Alexander R. Uhl1 , Carolin Fella1 , Lukas Kranz1 , Debora Keller1 , Christina Gretener1 , Harald Hagendorfer1 , Dominik Jaeger2 , Rolf Erni3 , Shiro Nishiwaki1 , Stephan Buecheler1 and Ayodhya N. Tiwari1 Thin-film photovoltaic devices based on chalcopyrite Cu(In,Ga)Se2 (CIGS) absorber layers show excellent light-topower conversion efficiencies exceeding 20% (refs 1,2). This high performance level requires a small amount of alkaline metals incorporated into the CIGS layer, naturally provided by soda lime glass substrates used for processing of champion devices3 . The use of flexible substrates requires distinct incorporation of the alkaline metals, and so far mainly Na was believed to be the most favourable element, whereas other alkaline metals have resulted in significantly inferior device performance4,5 . Here we present a new sequential post-deposition treatment of the CIGS layer with sodium and potassium fluoride that enables fabrication of flexible photovoltaic devices with a remarkable conversion efficiency due to modified interface properties and mitigation of optical losses in the CdS buffer layer. The described treatment leads to a significant depletion of Cu and Ga concentrations in the CIGS near-surface region and enables a significant thickness reduction of the CdS buffer layer without the commonly observed losses in photovoltaic parameters6 . Ion exchange processes, well known in other research areas7–13 , are proposed as underlying mechanisms responsible for the changes in chemical composition of the deposited CIGS layer and interface properties of the heterojunction. Photovoltaic devices, which directly convert abundant light into solar electricity, are important for the implementation of renewable energy supply systems. Today, silicon-based modules are dominating the photovoltaic market14 , but various emerging technologies based on thin-film inorganic semiconductors15 , dyesensitized materials16 and organic materials17 are rapidly progressing. Thin-film photovoltaic devices can be manufactured on flexible substrates, enabling the employment of roll-to-roll deposition techniques that have already revolutionized the food-packaging industry. Besides cost-reduction, flexible photovoltaic devices allow new opportunities in areas such as building integration, portable electronics, and space applications. However, the conversion efficiency of flexible solar cells has always been—until now—significantly lower compared with photovoltaic devices on rigid substrates18 . Amongst thin-film technologies, CIGS has yielded maximum conversion efficiencies of up to 20.3% with rigid soda lime glass (SLG) as a substrate2 . SLG substrates were so far favoured because
they allow high processing temperatures of 600 ◦ C or even more, have a coefficient of thermal expansion similar to CIGS, and provide the right amount of Na necessary for high-efficiency devices directly during film growth19 . In contrast, polyimide film as a flexible substrate can withstand processing temperatures only of well below 500 ◦ C, posing significant constraints on growing highquality CIGS layers, have a large thermal mismatch with CIGS and do not contain alkaline elements, which necessitates an adequate extrinsic supply method19 . Recently, we have shown that several of the coercive aspects of polyimide substrates described above can be overcome as we demonstrated an 18.7%-efficiency flexible solar cell on polyimide, which represented the highest reported value for any type of solar cell grown on a flexible substrate20 . We used a NaF postdeposition treatment (PDT) of the CIGS layer to supply Na after the growth of the CIGS (ref. 21) whereas other strategies focus on Na addition using NaF deposition before or during CIGS growth19 . Improvements in cell efficiency are achieved irrespective of the methods used for addition of Na, but the PDT treatment yielded the highest gains, especially at the low growth temperatures that are required for polyimide films as substrate21 . SLG substrates contain a vast amount of Na and a lower amount of K in oxide form and these alkaline elements are known to diffuse into CIGS during its growth. Although K has been frequently observed in CIGS, its influence on the properties of the final device has mostly been neglected or considered as similar to the effects stemming from Na (refs 22–26). Up to now, it was commonly believed that Na alone is sufficient for the improved electronic quality of CIGS devices. Enamelled steel substrates, which contain a high amount of K and a lower amount of Na, have indicated the potential for further improvements in device performance when the substrate contains a high amount of K (refs 18,26). Here, we demonstrate a distinctly beneficial role for further addition of K in CIGS layers subjected to prior addition of Na with a PDT. This resulted in a flexible solar cell with a new remarkable efficiency of 20.4%. K is added by simultaneous coevaporation of KF and Se directly after CIGS deposition and Na incorporation (see Methods). The secondary-ion mass spectroscopy (SIMS) profile of the sample with additional K treatment shown in Fig. 1a reveals a higher count rate of K compared with Na throughout the
1 Laboratory
for Thin Films and Photovoltaics, Empa—Swiss Federal Laboratories for Materials Science and Technology, Ueberlandstrasse 129, CH-8600 Duebendorf, Switzerland, 2 Nanoscale Materials Science, Empa—Swiss Federal Laboratories for Materials Science and Technology, Ueberlandstrasse 129, CH-8600 Duebendorf, Switzerland, 3 Electron Microscopy Center, Empa—Swiss Federal Laboratories for Materials Science and Technology, Ueberlandstrasse 129, CH-8600 Duebendorf, Switzerland. † These authors contributed equally to this work. *e-mail: [email protected] NATURE MATERIALS | VOL 12 | DECEMBER 2013 | www.nature.com/naturematerials
© 2013 Macmillan Publishers Limited. All rights reserved.
1107
1108
100,000
Mo
CdS
ZnO:AI i-ZnO
a
71Ga 113In
Intensity (counts)
10,000
27AI
39K 63Cu
64Zn
1,000
23Na
100 114Cd
10
0
5
98Mo
10
15 20 25 Sputter time (min)
30
35
b 100,000
71Ga 113In
27AI
10,000
Intensity (counts)
CIGS absorber layer. Using inductively coupled plasma mass spectroscopy (ICPMS) measurements, the amount of K was determined to be 1,700 ppm whereas the amount of Na was estimated to be about 50 ppm with respect to the CIGS layer. Figure 1b shows the SIMS profile for a representative sample that was not subjected to the additional KF-PDT. A significantly higher count rate of Na compared with K is measured, and it reveals a much higher count rate of Na compared with the SIMS profile of the sample with additional K treatment. These results show that the KF-PDT leads to a significant incorporation of K and removal of Na in the CIGS layer, which can be explained by an ion exchange mechanism that favours the presence of K over Na at the given KF deposition conditions. Ion exchange phenomena have been known for a long time7 and are used in a broad range of material systems and applications8,9 . For example, the exchange of Na with K is used for chemical strengthening of glass10 or tuning of its refractive index11 . Heterojunction formation has also been achieved by ion exchange methods, for example in the development of Cux S/CdS thin-film solar cells by topotactic conversion of the surface of a CdS layer12 . Ion exchange reactions between Cd and Cu were also observed at the CdS/CIGS interface13 . The surface composition of a bare CIGS absorber without alkaline treatment, as well as after NaF-PDT or after KF-PDT, was investigated by X-ray photoelectron spectroscopy (XPS). Figure 2a illustrates the three differently prepared samples. To remove possible NaF- or KF-related remnants from the surface, the samples subjected to NaF- or KF-PDT were etched before the XPS measurements (see Methods). No clear difference in the surface morphology due to the different PDT treatments is observed (Supplementary Fig. 1). Spectra of the relevant elements are presented in Fig. 2b–e. After treatment of the CIGS surface with only NaF-PDT the Cu and Ga peaks decrease by a factor of about 1.3 as compared with the sample that did not undergo any PDT, whereas the In and Se peaks remain nearly unchanged. KF-PDT leads to a much stronger reduction of the Ga peak intensity, the disappearance of the Cu peak, and little or no significant effect on Se and In peak intensities, respectively. Sputtering of approximately 20 nm is necessary to obtain a similar composition compared to the case when no KF is added as shown for the Cu 2p3/2 peak in Fig. 2f. Moreover, K is clearly detected in the first few nanometres of the CIGS surface as shown in Fig. 2g for the K 2p3/2 peak, whereas no Na is detected in the case of the CIGS absorber subjected to NaF-PDT (not shown). These measurements suggest that the changes of the CIGS surface chemical composition induced by the KF-PDT are limited to a thickness below 30 nm. To compare the near-surface region of the CIGS before and after buffer layer deposition, XPS measurements were performed on CIGS layers subjected to only NaF-PDT or only KF-PDT and covered with a thin CdS layer. The CdS layer was removed by sputtering as illustrated in Fig. 2h. The obtained spectra from the CdS/CIGS interface region are shown in Fig. 2i–l. In the case of the KF-PDT, a similar depletion of Cu and Ga is observed, revealing that the altered surface properties owing to the KF-PDT remain present after the CdS deposition process. Furthermore, K is again clearly detected at the interface, as shown in Fig. 2m with spectra recorded at different depths ranging from the CdS surface towards CIGS bulk, proving that it is not completely removed during the CdS deposition process. In contrast, the Na concentration at the interface was below the detection limit of XPS. Furthermore, it is found that the XPS peak positions, respectively binding energies, of In and Se shift in a different way on NaF or KF addition, but not sufficiently to support the formation of a new type of chemical bond (Supplementary Fig. 2). We conclude that besides the partial ion exchange of Na with K within the CIGS film, the KF-PDT also induces a modification of the chemical composition profile near the surface, depleting
CIGS
NATURE MATERIALS DOI: 10.1038/NMAT3789
LETTERS
64Zn 23Na
1,000
63Cu
114Cd
100
98Mo
39K
10 0
5
10
15 20 Sputter time (min)
25
30
Figure 1 | Elemental depth profiling. a,b, SIMS profiles showing all relevant elements from the front contact (ZnO) to the back contact (Mo) obtained from representative devices with addition of NaF and KF after the CIGS growth (a), and addition of NaF alone (b). In a, the significantly high count rate of 39 K, which is about 25 times higher than the count rate of 23 Na, indicates that a comparably high amount of K is incorporated within the device. In b, if no KF is added, clearly a higher count rate of 23 Na is observed compared with 39 K.
mostly Cu and Ga, with implantation of a significant amount of K. Different mechanisms and concepts have been discussed in the literature to explain the Cu depletion observed at CIGS surfaces. Among others, electrochemically driven Cu migration into the bulk due to increased Fermi energy at the surface27 and oxygenation-induced Cu redistribution28 are frequently used for explaining this phenomenon. Even if significant oxygenation in high vacuum during the KF-PDT is unlikely, the alkali content of the surface could enhance oxygenation during transfer in air to the buffer layer deposition. For an efficient photovoltaic device, in view of the modified surface chemistry of the CIGS absorber due to the KF-PDT, it is necessary to adapt the interface formation with the subsequent buffer layer. Generally, the best performance of CIGS solar cells is obtained using a CdS buffer layer deposited by chemical bath
NATURE MATERIALS | VOL 12 | DECEMBER 2013 | www.nature.com/naturematerials © 2013 Macmillan Publishers Limited. All rights reserved.
NATURE MATERIALS DOI: 10.1038/NMAT3789 a
No PDT
NaF
LETTERS h
KF
NaF
KF
CdS CIGS
Mo
Mo
KF
KF
448
e
Ga 2p3/2
No PDT NaF
No PDT NaF
931
930
Intensity (a.u.)
Intensity (a.u.) Se 3d 5/2
NaF
KF
1,121
1,119 1,117 Binding energy (eV)
m
K 2p 3/2 (KF)
K 2p 3/2
Sputtering time (approx. depth) 10 min (30 nm)
0 min (0 nm)
1,115
59
57 55 53 51 Binding energy (eV)
K 2p 3/2 (KF)
Se (Auger)
225
K 2p3/2
231 229 227 Binding energy (eV)
Se (Auger)
Intensity (a.u.)
NaF
CdS
0 min (0 nm)
932
446 444 442 Binding energy (eV)
CIGS
10 min (30 nm)
933
233
g
Binding energy (eV)
448
l
Se (Auger)
1,115
Sputtering time (approx. depth)
934
930
KF
Cu 2p3/2 (KF)
935
NaF KF
Ga 2p3/2
KF
Intensity (a.u.)
f
1,119 1,117 Binding energy (eV)
935 934 933 932 931 Binding energy (eV)
k
KF
1,121
NaF KF
446 444 442 Binding energy (eV)
Se 3s
In 3d 5/2
Intensity (a.u.)
NaF
934 933 932 931 930 Binding energy (eV)
Intensity (a.u.)
No PDT
NaF
j
Cu 2p3/2
Intensity (a.u.)
d
Intensity (a.u.)
Intensity (a.u.)
No PDT
935
i
In 3d 5/2
Intensity (a.u.)
c
Cu 2p3/2
Intensity (a.u.)
b
CIGS
300 298 296 294 292
306 302 298 294 290 286 282
Binding energy (eV)
Binding energy (eV)
Figure 2 | Surface chemical analysis. a, Schematic view of the three investigated absorbers. The purple layer on the KF absorber indicates the modified surface composition. b–e, XPS peak of Cu 2p3/2 , In 3d5/2 , Ga 2p3/2 and Se 3s, respectively, obtained from the surface of CIGS absorbers with no alkali evaporation (no PDT), only NaF addition and only KF addition. f, Sputtering of the CIGS absorber subjected to KF-PDT shows the appearance of the Cu 2p3/2 peak within the first approximately 20 nm with similar intensity as in the case of no PDT. g, K is clearly measurable at the surface up to a depth of approximately 20 nm. h, Schematic view of two absorbers measured after sputtering through the CdS layer. i–l, XPS peak of Cu 2p3/2 , In 3d5/2 , Ga 2p3/2 and Se 3s, respectively, at the CdS/CIGS interface with only NaF addition and only KF addition. m, XPS spectra of K at different sputtering depths.
deposition (CBD). However, the relatively low bandgap energy of about 2.4 eV leads to optical losses in the low-wavelength region of the solar spectrum. A way to reduce losses induced by CdS is to reduce its layer thickness. In the present work, we have identified the absorber modified with the KF-PDT to facilitate significant reduction of the CdS layer thickness without the commonly observed decrease in open-circuit voltage (VOC ) and fill factor values of the solar cell6 . Figure 3 presents external quantum efficiency (EQE) measurements of solar cells with absorbers subjected to the KF-PDT and different CBD durations, respectively CdS thicknesses. Up to now, state-of-the-art processes required CdS thicknesses of 40 nm or more22 , which is comparable to 20–22 min deposition time with our recipe (see Methods). However, the KFPDT enables a significant reduction in deposition duration down
to 14 min without any deterioration in device performance with efficiencies exceeding 18%. The gain in current density arising from better spectral response is apparent in Fig. 3 where the contribution to the short circuit current density (JSC ) from absorption below 540 nm wavelength is shown in the inset. This correlates well with the reduction of the CdS layer thickness calculated from ICPMS (inset) or observed in transmission electron microscopy (Supplementary Fig. 3). A minimum amount of evaporated KF is found to be required to allow the reduction of CdS thickness with maintained high efficiency. A pronounced deterioration of the photovoltaic parameters is observed if the duration of the KF-PDT is reduced to below 15 min (Supplementary Fig. 4). The changes in CIGS surface chemical composition induced by the KF-PDT, especially
NATURE MATERIALS | VOL 12 | DECEMBER 2013 | www.nature.com/naturematerials
© 2013 Macmillan Publishers Limited. All rights reserved.
1109
NATURE MATERIALS DOI: 10.1038/NMAT3789
LETTERS 90
70
Current density < 540 nm (mA cm¬2)
60 50 40 30 20 10 0 300
400
7.0
70 60
6.5
50 6.0
40
5.5
30 20
5.0 4.5
500
10 10 12 14 16 18 20 22 CBD time (min)
600
700
800
12 35
CdS thickness (nm)
External quantum efficiency (%)
80
a Voc = 736 mV Jsc = 35.1 mA cm¬2
25
FF = 78.9% η = 20.4%
8
20
6
15 4 10
0
0
900 1,000 1,100 1,200
10
2
5
b
0
100
200
300 400 500 Voltage (mV)
600
700
0
100 90
Figure 3 | Variation of CdS film thickness. EQE measurements from devices grown with various CBD deposition durations of 10 min (blue) up to 22 min (yellow). The inset figure shows the related contribution to JSC from photons with wavelength below 540 nm together with corresponding CdS layer thickness as determined with ICPMS.
the Cu deficiency, alter the interface properties of the resulting p–n junction in a way that enables a shorter CdS deposition time and therefore CdS thickness. This can be explained on the basis of an earlier model29 that describes the role of CBD–CdS, suggesting that during the CBD process Cd tends to occupy Cu vacancy sites, with the CdCu donors causing electronic inversion of the CIGS surface and forming a p–n homojunction in the CIGS layer. This inversion requires a certain density of CdCu that depends on the density of Cu vacancies and conditions of CBD to fill the Cu vacancies with Cd. A recent theoretical study further supports this model30 . With a similar CBD duration, XPS measurements of the CdS/CIGS samples discussed above reveal a higher Cd content in the near-surface region of the CIGS absorber subjected to the KF-PDT (Supplementary Fig. 5). Therefore, it seems that in a shorter duration a higher density of CdCu is achieved by diffusion of Cd into the Cu vacancies owing to the Cu-depleted CIGS surface composition. This can explain the possibility to reduce the CdS deposition time while maintaining a high junction quality, as evidenced by the efficiency above 18% for the cells with a CdS deposition time longer than 14 min shown in Fig. 3. Finally, improvements in the chemical composition of the overall CIGS layer, interface modification with additional KF, and overall stack and process optimization enabled the fabrication of a η = 20.4% efficiency device on polyimide substrate, representing the highest efficiency for the CIGS technology. Figure 4a presents the certified current–voltage and power–voltage measurements and Fig. 4b shows the corresponding EQE measurement. The EQE measurement of our previously reported best device20 is also given to highlight the improved spectral response related to the thinner CdS layer thickness. In summary, we have shown that the roles of Na and K in CIGS have to be differentiated, especially when being supplied after the growth of the CIGS layer. We show an approach of sequential in situ addition of both Na and K in CIGS after the layer growth that leads to a significant reduction in Na content in the CIGS bulk and a strong depletion of Cu and Ga in the CIGS surface region, confined to a depth of less than 30 nm, without any significant change in the microstructure. The KF-PDT-induced CIGS surface modification
External quantum efficiency (%)
Wavelength (nm)
1110
30
Power (mW)
10 min 12 min 14 min 16 min 18 min 20 min 22 min
Current density (mA cm¬2)
100
80 70
η = 18.7%
60 50 40 30 20 10 0 300
400
500
600 700 800 900 1,000 1,100 1,200 Wavelength (nm)
Figure 4 | Characteristics of the best device. a,b, J–V (diamonds), P–V (triangles) measurements (a) and EQE (b) of the new champion device (area = 0.52 cm2 ) obtained on flexible polyimide substrate (solid line). The EQE of the previous best device20 is also shown for comparison (dashed line). The solar cells were independently certified by Fraunhofer ISE, Freiburg, Germany. FF, fill factor.
facilitates Cd diffusion in the Cu-depleted surface of the absorber, and results in an improved CIGS/CdS heterojunction quality, even with a shorter CdS deposition time, respectively CdS layer thickness. Concepts of ion exchange reaction can explain changes of the K and Na concentration in the CIGS layer and the interface formation with the buffer layer. This new treatment led us to increase the highest efficiency of flexible CIGS solar cells on plastic to an independently certified 20.4% efficiency. This is the first time that the efficiency of a flexible solar cell on plastic has surpassed the cell efficiency on rigid glass, and it even matches the performance of the best polycrystalline Si wafer-based device, which is considered as a benchmark for photovoltaic technologies. Furthermore, the shown interaction and effects of different alkaline metals, based on the solid-state ion exchange mechanism, opens up new research directions not only in the field of thin-film photovoltaics but as well in various types of material system.
Methods Device preparation. Devices were fabricated on commercially available polyimide film on the basis of the methods described previously20 . CIGS layers were grown by co-evaporation from elemental effusion cells in a high-vacuum chamber (base pressure ∼10−8 hPa) equipped with an additional effusion cell for KF. The addition of KF was carried out directly after the NaF-PDT (∼1.3 nm min−1 for 20 min) by evaporating KF (∼1 nm min−1 , 15 min for optimal KF treatment) in the presence
NATURE MATERIALS | VOL 12 | DECEMBER 2013 | www.nature.com/naturematerials © 2013 Macmillan Publishers Limited. All rights reserved.
NATURE MATERIALS DOI: 10.1038/NMAT3789 of Se onto the finished CIGS layer at a substrate temperature of about 350 ◦ C. The average [Ga]/([In] + [Ga]), [Cu]/([Ga] + [In]) and film thickness of the CIGS layers are in the range of 0.33–0.38, 0.78–0.82 and 2.5–3.0 µm, respectively, as determined by means of X-ray fluorescence measurements. Devices are finished by CBD of a CdS buffer layer (see details below), radiofrequency sputtering of an i-ZnO/ZnO:Al window layer and electron-beam evaporation of Ni/Al/Ni grids. All cells are covered with a MgF2 anti-reflection coating and single cells are defined by mechanical scribing. CdS deposition. The CdS buffer layer was grown by a CBD method. A Cd acetate (Cd(CH3 COO)2 ) aqueous solution (0.028 M) and ammonium hydroxide aqueous solution (NH4 OH, 14.5 M) are first mixed together with high-purity water (18 M cm) in a volume ratio of 3:7:37 and preheated in a water bath at 70 ◦ C for 2 min. Thiourea (SC(NH2 )2 ) aqueous solution (0.374 M) is then added and the samples are immersed therein to begin the actual deposition. The whole solution is placed into a 70 ◦ C water bath under slow stirring until the desired film thickness is reached. The samples are then washed with high-purity water and dried with an ionized N2 gun, before an annealing step at 180 ◦ C in air for 2 min. SIMS. Depth profiling data were obtained with a time-of-flight SIMS 5 system from ION-TOF. Bi+ ions were used as primary ions and positive ions were detected. Sputtering was performed using O+ 2 sputtering ions with 2 keV ion energy, 400 nA ion current and a 300 × 300 µm2 raster size. An area of 100 × 100 µm2 was analysed + using Bi ions with 25 keV ion energy. ICPMS. For ICPMS analysis, approximately 1 cm2 of CIGS material was detached from the thin-film solar cell at the Mo/CIGS interface, directly transferred into 50 ml trace metal-free polyethylene tubes and fully dissolved in a mixture of 10 ml HNO3 and 1 ml HCl. After filling to 50 ml with 18 M cm deionized water, the sample was not further diluted for analysis. Metal analyses were performed on an Agilent 7500ce ICPMS with external calibration using certified metal standards (1,000 µg ml−1 , Merck CertIPUR). For quality assurance, analysis of reference materials (NIST SRM 1643e) and spiking experiments were performed, with recoveries between 90 and 110%. XPS. All XPS measurements were performed in a Quantum2000 from Physical Electronics using a monochromatized Al Kα source (1,486.6 eV). The measurements were recorded in the fixed analyser transmission mode, under an angle of emission of 45◦ and at an instruments base pressure of below 5×10−9 mbar. Spectra were recorded with a pass energy of Ep = 58.7 eV and an energy step size of 1E = 0.125 eV. Ar+ ions and electrons were used to compensate possible surface charging. Depth profiles have been obtained using an ion energy of 1 keV over a range of 2 mm × 2 mm. The work function at 45◦ has been adjusted in such a way that the binding energy of Au 4f7/2 is found at 83.96 eV with a full-width at half-maximum of 0.8 eV; the binding energies of gold have been measured regularly. Before the measurements, the surface of the CIGS bare absorbers subjected to a PDT are etched with a low-concentration ammonia aqueous solution to remove any unnecessary NaF or KF compound from the surface. EQE. EQE of solar cells was measured with a lock-in amplifier. A chopped white light source (900 W, halogen lamp, 360 Hz) and a dual grating monochromator generated the probing beam. The beam size was adjusted such that the illumination area was smaller than the device area. A certified monocrystalline Si solar cell from Fraunhofer ISE was used as the reference cell. Cell temperature was controlled at 25 ◦ C with Peltier cooling and white light bias was applied. Solar cell performance measurements. J –V characteristics of solar cells were measured under simulated standard-test conditions (25 ◦ C, 1,000 W m−2 , AM 1.5 G illumination) in a sun-simulator with a HMI light source. A Keithley 2400 source meter with four-terminal sensing was used to acquire J –V characteristics. J –V and EQE characteristics of the solar cell with 20.4% efficiency were independently certified by Fraunhofer ISE, Freiburg, Germany.
Received 14 May 2013; accepted 25 September 2013; published online 3 November 2013
References 1. Green, M. A., Emery, K., Hishikawa, Y., Warta, W. & Dunlop, E. D. Solar cell efficiency tables (version 41). Prog. Photovolt. 21, 1–11 (2013). 2. Jackson, P. et al. New world record efficiency for Cu(In,Ga) Se2 thin-film solar cells beyond 20%. Prog. Photovolt. 19, 894–897 (2011). 3. Hedstrom, J. et al. 23rd IEEE Phot. Spec. Conf. (PVSC) 364–371 (IEEE, 1993). 4. Contreras, M. A. et al. 26th IEEE Phot. Spec. Conf. (PVSC) 359–362 (IEEE, 1997). 5. Bodegård, M., Granath, K. & Stolt, L. Growth of Cu(In,Ga) Se2 thin films by coevaporation using alkaline precursors. Thin Solid Films 361, 9–16 (2000).
LETTERS 6. Contreras, M. A. et al. Optimization of CBD CdS process in high-efficiency Cu(In,Ga) Se2 -based solar cells. Thin Solid Films 403–404, 204–211 (2002). 7. Jenny, H. Studies on the mechanism of ionic exchange in colloidal aluminium silicates. J. Phys. Chem. 36, 2217–2258 (1932). 8. Clearfield, A. Role of ion exchange in solid-state chemistry. Chem. Rev. 88, 125–148 (1988). 9. Rivest, J. B. & Jain, P. K. Cation exchange on the nanoscale: An emerging technique for new materials synthesis, device fabrication, and chemical sensing. Chem. Soc. Rev. 42, 89–96 (2013). 10. Nordberg, M. E. et al. Strengthening by ion exchange. J. Am. Ceram. Soc. 47, 215–219 (1964). 11. Zou, J. et al. Two-step K+ –Na+ and Ag+ –Na+ ion-exchanged glass waveguides for C-band applications. Appl. Opt. 41, 7620–7626 (2002). 12. Das, S. R. et al. The preparation of Cu2 S films for solar cells. Thin Solid Films 51, 257–264 (1978). 13. Ramanathan, K. et al. Proc. 2nd World Conf. on Photovoltaic Solar Energy Conversion 477–481 (Joint Research Centre, European Commission, 1998). 14. Wolden, C. A. et al. Photovoltaic manufacturing: Present status, future prospects, and research needs. J. Vac. Sci. Technol. A 29, 030801 (2011). 15. Chopra, K. L., Paulson, P. D. & Dutta, V. Thin-film solar cells: An overview. Prog. Photovolt. 12, 69–92 (2004). 16. Grätzel, M. Photoelectrochemical cells. Nature 414, 338–344 (2001). 17. Bernede, J. C. Organic photovoltaic cells: History, principle and techniques. J. Chil. Chem. Soc. 53, 1549–1564 (2008). 18. Reinhard, P. et al. Review of progress toward 20% efficiency flexible CIGS solar cells and manufacturing issues of solar modules. IEEE J. Photovolt. 3, 572–580 (2013). 19. Kessler, F. & Rudmann, D. Technological aspects of flexible CIGS solar cells and modules. Sol. Energ. 77, 685–695 (2004). 20. Chirilă, A. et al. Highly efficient Cu(In,Ga) Se2 solar cells grown on flexible polymer films. Nature Mater. 10, 857–861 (2011). 21. Rudmann, D. et al. Sodium incorporation strategies for CIGS growth at different temperatures. Thin Solid Films 480, 55–60 (2005). 22. Jackson, P. et al. High quality baseline for high efficiency, Cu(In1−x ,Gax )Se2 solar cells. Prog. Photovolt. 15, 507–519 (2007). 23. Ishizuka, S., Yamada, A., Fons, P. & Niki, S. Flexible Cu(In,Ga) Se2 solar cells fabricated using alkali-silicate glass thin layers as an alkali source material. J. Renew. Sustain. Energ. 1, 013102 (2009). 24. Cojocaru-Mirédin, O. et al. Characterisation of grain boundaries in Cu(In,Ga) Se2 films using atom-probe tomography. IEEE J. Photovolt. 1, 207–212 (2011). 25. Abou-Ras, D. et al. Confined and chemically flexible grain boundaries in polycrystalline compound semiconductors. Adv. Energy. Mater. 2, 992–998 (2012). 26. Wuerz, R. et al. CIGS thin-film solar cells and modules on enamelled steel substrates. Sol. Energ. Mater. Sol. Cells 100, 132–137 (2012). 27. Klein, A. et al. Fermi level-dependent defect formation at Cu(In,Ga) Se2 interfaces. Appl. Surf. Sci. 166, 508–512 (2000). 28. Rau, U. et al. Oxygenation and air-annealing effects on the electronic properties of Cu(In,Ga) Se2 films and devices. J. Appl. Phys. 86, 497–505 (1999). 29. Nakada, T. & Kunioka, A. Direct evidence of Cd diffusion into Cu(In,Ga) Se2 thin films during chemical-bath deposition process of CdS films. Appl. Phys. Lett. 74, 2444–2446 (1999). 30. Kiss, J. et al. Theoretical study on the structure and energetics of Cd insertion and Cu depletion of CuIn5 Se8 . J. Phys Chem. C 117, 10892–10900 (2013).
Acknowledgements This work was supported by the Swiss National Science Foundation and the Swiss Federal Office of Energy. The laboratory for Nanoscale Materials Science and the Laboratory for Electronics/Metrology/Reliability at Empa are acknowledged for SIMS and scanning and transmission electron microscopy measurements, respectively.
Author contributions A.C., P.R., F.P., P.B., S.B., S.N. and A.N.T. designed the research and experiments. A.C., P.R. and P.B. fabricated the solar cells. A.C., P.R., F.P., S.B., A.R.U., C.F., C.G., H.H., D.J., L.K., D.K., R.E. and A.N.T. performed the characterization and analysis. A.C., P.R., F.P., S.B. and A.N.T. wrote the paper. All authors contributed to discussions.
Additional information Supplementary information is available in the online version of the paper. Reprints and permissions information is available online at www.nature.com/reprints. Correspondence and requests for materials should be addressed to A.C.
Competing financial interests The authors declare no competing financial interests.
NATURE MATERIALS | VOL 12 | DECEMBER 2013 | www.nature.com/naturematerials
© 2013 Macmillan Publishers Limited. All rights reserved.
1111
