Applied Physics Letters 103 (16), 163904 (2013) Characterization of defects in 9.7 % efficient Cu2ZnSnSe4-CdS-ZnO solar cells G. Brammertz1,2*, M. Buffière3,4, S. Oueslati1,2,5,6, H. ElAnzeery1,2,5,7, K. Ben Messaoud1,2,5,6, S. Sahayaraj1,2,4, C. Köble8, M. Meuris1,2, J. Poortmans3,4. 1
imec division IMOMEC - partner in Solliance, Wetenschapspark 1, 3590 Diepenbeek, Belgium. Institute for Material Research (IMO) Hasselt University, Wetenschapspark 1, 3590 Diepenbeek, Belgium. 3 imec, Kapeldreef 75, 3001 Leuven, Belgium. 4 Department of Electrical Engineering, KU Leuven, Kasteelpark Arenberg 10, 3001 Heverlee, Belgium. 5 KACST-Intel Consortium Center of Excellence in Nano-manufacturing Applications (CENA), Riyadh, KSA. 6 Department of Physics, Faculty of Sciences of Tunis, El Maner, Tunisia. 7 Microelectronics System Design Department, Nile University, Cairo, Egypt. 8 Helmholtz-Zentrum Berlin für Materialien und Energie GmbH, Hahn-Meitner-Platz 1, 14109 Berlin, Germany. 2
We have fabricated Cu2ZnSnSe4-CdS-ZnO solar cells with a total area efficiency of 9.7 %. The absorber layer was fabricated by selenization of sputtered Cu10Sn90, Zn and Cu multilayers. A large ideality factor of the order of 3 is observed in both illuminated and dark IV-curves, which seems to point in the direction of complex recombination mechanisms such as recombination through fluctuating potentials in the conduction and valence bands of the solar cell structure. A potential barrier of about 135 meV in the device seems to be responsible for an exponential increase of the series resistance at low temperatures, but at room temperature the effect of this barrier remains relatively small. The free carrier density in the absorber is of the order of 1015 cm-3 and does not vary much as the temperature is decreased. In the past years an increasing amount of research on Cu2ZnSnSe4 (CZTSe), Cu2ZnSnS4 (CZTS) and Cu2ZnSnSxSe4-x (CZTSSe) based solar cells has been conducted for application as thin film photovoltaic devices1-5. They have shown a serious potential to rival CIGS and CdTe technology, with 11.1% efficiency already demonstrated6. Nevertheless, further improvements to the technology are necessary in order to reach the 15 to 20 % total area efficiency range, which is necessary to make CZTS a serious competitor for CIGS and CdTe thin film solar cell technology. In this work we have fabricated 9.7 % total area efficient devices using an industry relevant technology composed of sputtering of metal layers followed by selenization. We have done extensive physical, electrical and optical characterization in order to get information about the defects in our device structure. The device presented in this work was prepared by DC-sputtering of Cu, Zn and Cu10Sn90 layers onto 5x5 cm2 Mo-coated soda-lime glass substrates. The sputtered layers were then annealed in a 10% H2Se in N2 environment for 15 minutes at a temperature of 460°C. For solar cell processing, a standard procedure for CIGS based solar cells was used, consisting of KCN etch, chemical bath deposition of a thin n-type CdS buffer layer and AC-sputtering of 120 nm of intrinsic ZnO followed by 250 nm of highly Al-doped ZnO7,8. A 150 nm thick MgF2 anti reflection coating was also deposited. Finally, a 50 nm Ni - 1 µm Al finger grid pattern was evaporated through a shadow mask for top contact formation. Lateral isolation of the cells was performed by needle scribing. In this way 23 solar cell devices with sizes ranging from 0.25 cm2 to 1 cm2 were fabricated on the 5x5 cm2 1
sample. The final layout of the sample can be seen in the insert of figure 1(c), which shows an image of a finished device. More details of the fabrication procedure can be found in Ref. 9.
Figure 1: Cross-section SEM image of the solar cell with the different layers being highlighted in different colors (a). Current density versus voltage curve of the best 1 cm2 solar cell under AM1.5G illumination (b). J-Jsc for the dark case (dark blue curve) and the AM1.5G illuminated case (light blue curve) of the 1 cm2 solar cell on a logarithmic scale. The one diode model fits to the data are shown as well (dashed lines). The insert shows a picture of the complete 5x5 cm2 sample with a total of 23 cells of different sizes 2
Figure 1(a) shows a cross-section SEM image highlighting the different layers of the device in different colors. The average composition of the absorber layer, as measured from energy dispersive X-ray spectroscopy, was determined as Cu/(Zn+Sn) = 0.7, Zn/Sn = 1 and Se/(Cu+Zn+Sn) = 1.08. The sample is therefore very Cu-poor, stoichiometric with respect to Zn and Sn and quite Se-rich. This composition diverges a bit from other best compositions reported in literature, where stoichiometry of Cu/(Zn+Sn) = 0.8 and Zn/Sn = 1.2 are generally reported to yield best results10-14. The current-voltage curve of the 1 cm2 cell with the highest efficiency measured under AM1.5G illumination is shown in figure 1(b). A total area efficiency of 9.7 % was measured on a 1 cm2 cell with an open circuit voltage Voc of 408 mV, a short circuit current density Jsc of 38.9 mA/cm2 and a fill factor of 61.4 %. Whereas this is the efficiency of the best 1 cm2 device, the homogeneity of the sample was also relatively good, as all devices of the 5x5 cm2 sample showed efficiencies above 8 %. One particularity of our samples becomes apparent in the cross section SEM image, namely the large amount of holes on the backside of the sample. The origin of the holes could either be due to the unstable nature of CZTSe in contact with Mo, as reported in Ref. 15, or it could arise because we use the low melting point Cu10Sn90 metal layer as the first layer in contact with the Mo. This layer could liquefy during the first stages of the selenization and not present good wetting properties on Mo. It is not quite clear currently whether these holes improve the backside contact passivation and thereby the charge collection or not. Nevertheless, it is quite clear that they are not very helpful for good adhesion of the absorber layer on the Mo backside contact, as the absorber layer easily peels off the Mo backside upon mechanical stress. Therefore, careful probe needle placement is necessary in order not to damage the samples during measurements. Figure 1(c) shows the current density versus the voltage for both the dark and AM1.5G illuminated case on a logarithmic scale. The short circuit current density was subtracted for better visibility of the diode parameters on a logarithmic scale. One diode model fits according to the equation:
− = − +
(1)
are shown on the figure as well. The diode parameters for the fit were extracted using the procedure outlined in Ref. 16. The values of the different diode parameters are shown in table I. Table I: One diode model parameters for fit to experimental data in figure 1(c). Dark curve AM1.5G curve Rs (Ω cm2) 1.44 1.05 2 Rsh (Ω cm ) 3130 680 A (/) 3.5 3.0 J0 (A cm2) 1.6 10-6 2.7 10-6 What is quite remarkable is the large ideality factor A of the diode equation. In both the dark and the illuminated case A is larger than 2. A somewhat unusual recombination mechanism seems to be responsible for the current behavior of the device. Possible candidates of this recombination mechanism are tunneling enhanced recombination, recombination via coupled defects, donoracceptor pair recombination or fluctuations of the activation energy of the dominant recombination process17. From low temperature photoluminescence measurements it was shown18,19 that the most likely recombination mechanism responsible for this high ideality factor is recombination through fluctuating potentials in the conduction and valence band, arising either from strong compensation in the material or from lower bandgap phase inclusions in the main kesterite absorber material. This 3
unusual recombination mechanism is likely responsible for the low Voc values observed in kesterite solar cells.
Figure 2: External quantum efficiency of the best 1 cm2 solar cell as a function of illumination wavelength (a). Time resolved photoluminescence spectrum of the solar cell sample. The dashed line is a fit to the longer lifetime component representative of the minority carrier lifetime in the absorber. The insert shows a photoluminescence spectrum of the sample (b). External quantum efficiency (EQE) measurements of the best cell are shown in figure 2. As could already be expected from the large Jsc values, the EQE of the solar cell is quite good. Charge collection in the optical wavelength range approaches 100 %. Towards the infrared wavelength range, for photons absorbed deeper in the absorber, charge collection probability decreases, but is still above 80 % at 1000 nm. At zero bias voltage, there is therefore not too much problems for collecting most of the charge carriers. The bandgap derived from the quantum efficiency curve is about 1 eV, which can be confirmed from the photoluminescence peak position, shown in the inset of figure 2(b). Time and spectrally-resolved photoluminescence measurements shown in figure 2(b) were acquired with a Hamamatsu C12132 near infrared compact fluorescence lifetime measurement system. An area of 3 mm diameter was illuminated on a solar cell with a 15 kHz, 1.2 ns pulsed 532 nm laser with 1.38 mW of average laser power. The minority carrier lifetime was derived using a two exponential fit to the photoluminescence decay curve shown in figure 2(b): IPL(t) = C1e−t/τ1 + C2e−t/τ2 ,
(2)
where IPL(t) represents the PL intensity as a function of time. C1 and C2 are coefficients, τ1 and τ2 are two different decay times, with τ1 being the faster decay time and τ2 being the slower decay time. The slower decay time τ2 is reported to be the low injection minority carrier lifetime whereas the faster decay time τ1 is linked to the charge separation time20. The minority carrier lifetime at zero bias 4
voltage in our finished solar cell structure is of the order of 7 ns, which seems to be sufficient to collect most of the minority carriers excited in the absorber, as was visible from the EQE curves.
Figure 3: Open circuit voltage of the solar cell as a function of temperature. The dashed line is a fit to the linear part of the curve (a). Semi-log plot of the series resistance of the dark IV-curve as a function of temperature (b). Arrhenius plot of the series resistance for activation energy determination (c). In order to gather more information about the electrical behavior of the cell, low temperature current density and admittance measurements were acquired. Figure 3(a) shows the Voc of the cell as a function of the temperature. The behavior is linear down to 180 K and the extrapolation of the linear trend to zero K yields approximately the bandgap value (0.97 eV), therefore the activation energy of the main recombination process is not limited by a reduced bandgap at the CZTSe-CdS interface. The series resistance of the devices, shown in figure 2(b) increases exponentially as the temperature decreases. An Arrhenius plot of the series resistance (Fig 2c) reveals an activation barrier of about 5
135 meV. Similar results have been presented elsewhere21. Nevertheless, the effect on the room temperature series resistance remains relatively small.
Figure 4: Capacitance of the solar cell at zero Volt as a function of frequency for different temperatures (a). Doping profile in the cell as determined from the Mott-Schottky plot derived from the capacitance data at 100 kHz for different temperatures (b).
On the other hand, the doping in the absorber, as measured from a 100 kHz capacitance-voltage measurement by means of a Mott-Schottky plot does not vary much with temperature. Figure 4(a) shows the capacitance voltage curves at zero bias voltage and figure 4(b) shows the doping as a function of position in the absorber for temperatures going down to 213 K. The measurement was made down to only 213 K, as the exponentially increasing resistance at lower temperatures introduces an additional signal in the capacitance response, thereby influencing the extraction of doping and other admittance properties22. It can be stated, despite the doping decreasing a little with temperature, that a freeze out of the doping cannot be observed. Therefore, the increasing series resistance at low temperature must be due to a barrier somewhere in the solar cell structure. The 6
doping in our solar cell is low, of the order of 1015 cm-3, in agreement with the exponential relationship between the Zn/Sn ratio and the doping presented in Ref. 23 for Cu-poor CZTSe devices.
Figure 5: Admittance spectra derived from the capacitance data of figure 4(a) for different temperatures down to 213 K (a). Arrhenius plot of the frequency corresponding to the maximum in the admittance peak (b).
The conductance peaks at zero bias voltage for different temperatures are shown in figure 5(a). A small peak can be observed, which shifts to lower frequencies with decreasing temperature. Temperatures lower than 213 K have been omitted for the same reason as mentioned previously, namely an exponentially increasing series resistance at lower temperatures which distorts the admittance results. An Arrhenius plot of the peak maximum reveals an activation energy of 150 meV. Theoretical calculations in Ref. 24 show that a lot of physical defects in CZTSe have transition level energies close to 150 meV, such as the Zn vacancy, Sn vacancy, Cu on Zn antisite, Cu on Sn antisite and even the Zn on Sn antisite. Attribution of this level to any precise physical defect is therefore difficult. But, on the other hand, this activation energy of 150 meV is also close to the one of the barrier resistance. Therefore, it could be that this admittance peak is just a trace of the barrier resistance in the admittance signal and not corresponding to any physical defect level at all. Other 7
peaks could not be observed in the signal, probably showing that overall deep defect density in the CZTSe is small. In summary, we have fabricated 9.7 % total area efficient solar cells using selenization of sputtered metal layers. We have observed a large ideality factor, of the order of 3, probably due to complex recombination mechanisms such as recombination mediated by fluctuating conduction and valence bands in the absorber. The average stoichiometry of the device was found to be Cu/(Zn+Sn) = 0.7 and Zn/Sn = 1. This stoichiometry corresponds to a low average free carrier density of the order of 1015 cm-3 as measured from Mott-Schottky plots, which does not change much as the temperature decreases. Charge collection at zero bias is good in the device, with a short circuit current density of 39.8 mA/cm2 and an external quantum efficiency at zero bias voltage approaching 100 %. We observe an exponentially increasing series resistance as the temperature decreases, due to a 135 meV barrier somewhere in the structure, but the resistance of this barrier at room temperature remains small. Admittance spectroscopy reveals a peak with an activation energy of about 150 meV, which is possibly a trace of the barrier resistance in the admittance signal. Otherwise no peaks could be observed in the admittance spectrum, probably showing that deep defect density in CZTSe is small. Acknowledgments We would like to acknowledge Tom De Geyter, Greetje Godiers and Guido Huyberechts from Flamac in Gent for sputtering of the metal layers. AGC is acknowledged for providing substrates. Hamamatsu Photonics is acknowledged for providing the time resolved photoluminescence measurement system. References 1
S. Siebentritt and S. Schorr, Prog. Photovolt: Res. Appl. 20, 512 (2012). S. Delbos, EPJ Photovoltaics 3, 35004 (2012). 3 K. Ito, T. Nakazawa, Japanese Journal of Applied Physics 27, 2094 (1988). 4 C. Platzer-Björkman, J. Scragg, H. Flammersberger, T. Kubart, M. Edoff, Solar Energy Materials & Solar Cells 98, 110 (2012). 5 C. Persson, J. Appl. Phys. 107, 53710 (2010). 6 T.K. Todorov, J. Tang, S. Bag, O. Gunawan, T. Gokmen, Y. Zhu and D.B. Mitzi, Adv. Energy Mater. 3, 34 (2013). 7 J. Klaer, I. Luck, A. Boden, R. Klenk, I. Gavilanes Perez, R. Scheer, Thin Solid Films 431, 534 (2003). 8 Ch. Köble, D. Greiner, J. Klaer, R. Klenk, A. Meeder and F. Ruske, Thin Solid Films 518, 1204 (2009). 9 G. Brammertz, Y. Ren, M. Buffière, S. Mertens, J. Hendrickx, H. Marko, A. E. Zaghi, N. Lenaers, C. Köble, J. Vleugels, M. Meuris, J. Poortmans, Thin Solid Films 535, 348 (2013). 10 L. Grenet, S. Bernardi, D. Kohen, C. Lepoittevin, S. Noel, N. Karst, A. Brioude, S. Perraud, H. Mariette, Solar Energy Materials & Solar Cells 101, 11 (2012). 11 A. Redinger and S. Siebentritt, Appl. Phys. Lett. 97, 092111 (2010). 12 I. Repins, C. Beall, N. Vora, C. De Hart, D. Kuciauskas, P. Dippo, B. To, J. Mann, W. C. Hsu, A. Goodrich, R. Noufi, Solar Energy Materials & Solar Cells 101, 154 (2012). 13 B. Shin, Y. Zhu, N. A. Bojarczuk, S. J. Chey, and S. Guhaa, Appl. Phys. Lett. 101, 053903 (2012). 14 A. Fairbrother, E. García-Hemme, V. Izquierdo-Roca, X. Fontané, F. A. Pulgarín-Agudelo, O. Vigil-Galán, A. Pérez-Rodríguez, and E. Saucedo, J. Am. Chem. Soc. 134, 8018 (2012). 15 J. J. Scragg, P. J. Dale, D. Colombara and L. M. Peter, ChemPhysChem 13, 3035 (2012). 2
8
16
S.S. Hegedus and W.N. Shafarman, Prog. Photovolt: Res. Appl. 12, 155 (2004). R. Scheer and H. W. Schock, Chalcogenide Photovoltaics, Wiley-VCH, Weinheim, Germany (2011). 18 D. P. Halliday, R. Claridge, M. C. J. Goodman, B. G. Mendis, K. Durose and J. D. Major, J. Appl. Phys. 113, 223503 (2013). 19 S. Oueslati, G. Brammertz, M. Buffiere, C. Köble, M. Meuris, T. Walid and J. Poortmans, Photoluminescence study and observation of unusual optical transitions in ZnO/CdS/Cu2ZnSnSe4 solar cells, submitted to Solar Energy Materials and Solar Cells (2013). 20 A. Kanevce, D. H. Levi and D. Kuciauskas, Prog. Photovolt: Res. Appl., DOI: 10.1002/pip.2369 (2013). 21 O. Gunawan, T. K. Todorov, and D. B. Mitzi, Appl. Phys. Lett. 97, 233506 (2010). 22 T. P. Weiss, A. Redinger, J. Luckas, M. Mousel and S. Siebentritt, Appl. Phys. Lett. 102, 202105 (2013). 23 G. Brammertz, M. Buffière, Y. Mevel, Y. Ren, A. E. Zaghi, N. Lenaers, Y. Mols, C. Koeble, J. Vleugels, M. Meuris, and J. Poortmans, Appl. Phys. Lett. 102, 013902 (2013). 24 S. Chen, A. Walsh, X.G. Gong, S.H. Wei, Adv. Mater. 25, 1522 (2013). 17
9
