Enhanced omni-directional performance of copper zinc tin sulfide thin ...

6 downloads 0 Views 942KB Size Report
Mar 19, 2014 - Thin film solar cells, such as those based on copper zinc tin sulfide ... an important component of a renewable energy supply. Though not yet as ...
Enhanced omni-directional performance of copper zinc tin sulfide thin film solar cell by gradient index coating Zhongyang Ge, Pravakar Rajbhandari, Junjie Hu, Amin Emrani, Tara P. Dhakal, Charles Westgate, and David Klotzkin Citation: Applied Physics Letters 104, 101104 (2014); doi: 10.1063/1.4868104 View online: http://dx.doi.org/10.1063/1.4868104 View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/104/10?ver=pdfcov Published by the AIP Publishing

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 128.226.37.5 On: Wed, 19 Mar 2014 00:51:12

APPLIED PHYSICS LETTERS 104, 101104 (2014)

Enhanced omni-directional performance of copper zinc tin sulfide thin film solar cell by gradient index coating Zhongyang Ge,1 Pravakar Rajbhandari,1,2 Junjie Hu,1 Amin Emrani,1,2 Tara P. Dhakal,1,2 Charles Westgate,1,2 and David Klotzkin1

1 Electrical and Computer Engineering Department, Binghamton University, Binghamton, New York 13902-6000, USA 2 Center for Autonomous Solar Power (CASP Center), Binghamton University, Binghamton, New York 13902-6000, USA

(Received 15 December 2013; accepted 26 February 2014; published online 11 March 2014) Many types of thin-film solar cells have a top, transparent conducting oxide (TCO) coating (such as aluminum-doped zinc oxide (AZO)) through which light is transmitted and current collected. In this paper, we demonstrate an effective antireflective coating for TCO surfaces using a gradient index coating formed from co-sputtered AZO and silicon dioxide (SiO2) targets that reduces reflection loss from the TCO. When applied to an active solar device, the power conversion efficiency of the solar cell increased by >10% when measured at normal incidence and >20% at C 2014 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4868104] angles 10 or greater. V

Thin film solar cells, such as those based on copper zinc tin sulfide (CZTS) absorber layers, have the potential to be an important component of a renewable energy supply. Though not yet as efficient as silicon technology, their fabrication process is much less expensive and they are expected to be readily scalable to very large areas and on flexible substrates. However, for them to be competitive with silicon technology, it is important that they be as efficient as possible. The top surface of a thin film solar cell is usually a layer of transparent conducting oxide (TCO), and one significant source of loss is Fresnel reflection from this layer. Generally, solar cell efficiency can be improved by employing an antireflection layer coating (often a quarter-wavelength thick of a suitable intermediate index material) on the surface to transmit more light into the active region. This works well on silicon but can be challenging for thin film solar cells where the top conducting layers are not usually optically flat. Additionally, typical quarter wave coatings are only effective at completely reducing reflection at one specific angle and wavelength. Typical solar cells collect light ranging from 350 to 1200 nm wavelength over a variety of incident angles throughout the day. Thus, realization of broadband and omnidirectional antireflection technique is desired. Some approaches have been reported to address this problem. Jheng et al. reported that by controlling the morphology of ZnO nanostructures on top of CZTS solar cell the energy conversion efficiency increased from 4.8% to 5.3%.1 Chen and Sun reported that by fabricating nanostructured Si molds with e-beam lithography and chemical wet etching, the light harvesting efficiency of solar cell increased from 10.4% to 13.5%.2 Forberich et al. reported a 3.5% increase in the external quantum efficiency (EQE) peak value by moth-eye anti-reflection coating.3 Improved optical absorption in amorphous silicon nanowire and nanocone arrays is also reported.4 These methods rely on relatively complicated nano-patterning techniques, which are not yet conventional commercial processes. 0003-6951/2014/104(10)/101104/4/$30.00

Gradient index (grin) coatings achieve low reflectivity over a wide spectral and angular range through gradual index changes. In the late 19th century, Lord Rayleigh first mathematically proposed that a gradual index transition between two abrupt mediums has perfect anti-reflective (AR) property.5 After the first gradient index AR film was experimentally realized by Jacobsson in the 1960s,6 this optical coating method has been developed with the advancement of fabrication techniques of dielectric films. A gradient index layer can be normally fabricated by depositing two materials of different refractive indices simultaneously. Recent work on transparent conductors is focused on aluminum-doped zinc oxide (AZO) for thin film solar cells application.7,8 The transparent conductive oxide AZO has become a non-toxic, earth-abundant alternative to commonly used transparent conductors, such as indium tin oxide (ITO),9,10 with resistivity on the order of 104 X cm11 and good transparency. In our previous work,12 we have achieved a 4% reflectivity from 400 nm to 1000 nm at a normal incidence on a bare silicon wafer by a three-layer gradient film coating designed by numerical optimization. Here, we demonstrate the effectiveness of gradient index coatings on CZTS thin film solar cells with aluminum doped zinc oxide (AZO) TCO layers. A co-sputtered coating from AZO to SiO2 effectively increased the cell efficiency over wide incident angles and wavelengths. A gradient index layer, linearly graded from AZO to SiO2, was deposited by simultaneous sputtering of AZO and SiO2 targets. Aluminum zinc oxide (2%/98% Al2O3/ZnO by weight) was chosen as the higher refractive index material as it closely matched the index of the AZO top layer. At the low index end, SiO2 was chosen for its good thermal stability and high transparency. By gradually changing the power on each sputtering target, the relative deposition rates of AZO and SiO2 were linearly changed from 0.25 nm/min to 0 and from 0 to 0.25 nm/min synchronously such that the total material deposition rate at each instant was 0.25 nm/min.

104, 101104-1

C 2014 AIP Publishing LLC V

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 128.226.37.5 On: Wed, 19 Mar 2014 00:51:12

101104-2

Ge et al.

Appl. Phys. Lett. 104, 101104 (2014)

FIG. 1. (a) Refractive index profile of Air/Grin-layer/AZO/glass stack. (b) SEM image of the gradient index layer coated glass. The layers with height H1 and H2 are grin-layer and AZO, respectively.

Other authors have shown that the refractive index of similar mixed oxide materials, either co-sputtered,13 or deposited in very thin layers, is (within a few percent) linearly proportional to the volume fraction of the deposited materials.14 Hence, the assumed refractive index is taken as proportional to the composition fraction of each material for modeling and analysis. To determine the optimal thickness of the graded index layer and the optimal initial and final index, a fit function was defined as the sum of the total reflectivity over a set of different angles, ranging from normal incidence to very oblique incidence. Reflectivity of each arbitrary structure was calculated using a transfer matrix method15 with thin (1 nm) slices of uniform composition. The calculation was done with the film fixed with a linear index gradient between start and end indices, and the thickness and indices were picked to minimize total reflectivity over this sum of angles. The cosputtered materials were AZO and SiO2, constraining the range of indices to be between 1.5 and 1.9. In this case, where the lower index is not that close to 1, the optimal thickness was about 100 nm (similar to the thickness of a conventional quarter wavelength coating in the visible range with these materials) and the optimal final and initial indices were 1.5 and 1.9, respectively. To evaluate directly the improved transmittance, a gradient index coating (160 nm thick) was deposited initially on an AZO-coated glass substrate with a 290 nm thick AZO layer and the transmittance was measured. Figure 1(a) shows the index profile for this experiment, showing the quartz glass (n ¼ 1.5), followed by the AZO layer (n ¼ 1.9), and the gradient index layer from AZO to SiO2, with the index ranging

FIG. 2. Gradient index layer on CZTS solar cell.

from 1.9 to 1.5. Figure 1(b) shows the SEM cross-sectional image of the coated AZO layer on glass. The optimized gradient index layer was deposited on the AZO film surface of a CZTS solar cell. Figure 2 shows the cross section of the solar cell coated with the gradient index layer. Because the underlying CZTS surface is very rough, the AZO top layer and the gradient index coating layer are also quite rough. Details of the solar cell fabrication and performance are described elsewhere.16 The transmittance of the original and coated AZO on glass piece is shown in Figure 3(a). After coating, transmission at normal incidence increased through the visible wavelength range from 400 to 700 nm (except for a band from 475 to 510 nm). A maximum of 11.8% transmittance increase was achieved at 408 nm. For wavelengths longer than 560 nm, the increase is about 7%. To validate the linear index model of the materials, the Lumerical program FDTD was used for a finite difference time domain calculation of the deposited gradient index film (Fig. 1(a)) to compare with experimental results. The boundary conditions were set to perfectly matched layers (PMLs) in the x direction, along which light was incident, and metal boundary conditions for the other directions. A plane wave source was placed in the air above the gradient index layer, and power monitors were placed in quartz glass to measure the net power emitted through AZO into glass. Figure 3(b) shows the simulated transmittance result which agrees well with the experimental result. Both experiment and simulation

FIG. 3. (a) Transmittance measurement of AZO on glass and with the grin layer coating. (b) FDTD simulated transmittance of coating and without coating.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 128.226.37.5 On: Wed, 19 Mar 2014 00:51:12

101104-3

Ge et al.

Appl. Phys. Lett. 104, 101104 (2014)

FIG. 4. Reflectance of coated and uncoated AZO solar cell.

showed the AZO film transmittance improved across a broad band by adding the gradient index coating to a layer on surface. In this calculation, a single nominal value of index was used for each material, and the variation in index with wavelength was neglected. The specular reflectance of coated and uncoated rough AZO solar cell was also directly measured by a reflectometer and is shown in Figure 4. The measured specular reflectance was around 0.5% in the visible region and increased from 0.75% to 1.6% for an uncoated solar cell. This number does not include diffuse reflectance, or absorption, but it does show the improvement due to the gradient coating. To verify the gradient index layer’s improvement in transmittance on an AZO solar cell, the external quantum efficiency of the coated AZO solar cell was measured with a small focused light spot, using a quantum efficiency measurement system from PV Measurements Inc. equipped with a xenon light source and a monochromator with a chopper. As shown in Figure 5, a maximum increase of 11.9% at around 445 nm and an increase of 7.4% at about 550 nm were achieved. The observed absorption edge at 349 nm

FIG. 5. External quantum efficiency of coated and original AZO solar cell at 0 incidence.

FIG. 6. I-V curve of original and grin layer coated solar cell at normal incidence (0 ), and at 40 , 60 , and 70 angles, measured under the solar light spectrum of AM 1.5 flood conditions.

corresponds to a band gap of 3.55 eV, which corresponded to that of AZO.17 Figure 6 shows the I-V curves measured using a Photo Emission Tech solar simulator under air mass (AM) 1.5 conditions at 0 , 40 , 60 , and 70 . Both the short circuit current (Isc) and open circuit voltage (Voc) increased at all angles after coating. Figure 7 shows power conversion efficiencies measured at different angles. The efficiency improvement was 10.7% at normal incidence and above or very close to 20% for other angles. A maximum increase of 24.2% was achieved at 70 . This result illustrates that an AZO to SiO2 grin layer works well for improving solar cell efficiency for various incident angles. The improvement in efficiency is primarily due to the increase in short-circuit current (photo-generated current) corresponding to the improved light harvesting caused by the gradient index layer coating. The improvement in Voc is also observed, which is because of the improved light current according to the following equation:

FIG. 7. Efficiency increase vs. angle of incidence.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 128.226.37.5 On: Wed, 19 Mar 2014 00:51:12

101104-4

Ge et al.

  gkT Isc ln Voc ¼ þ1 ; Io q

Appl. Phys. Lett. 104, 101104 (2014)

(1)

where g, k, q, T, Isc, and Io are diode ideality factor, Boltzman’s constant, electronic charge, temperature, shortcircuit current, and reverse saturation current, respectively. A slight increase in Voc could have also been caused by the reduced reverse saturation current as expected from the improved fill factors. An interesting question is how this grin coating compares to a conventional antireflective coating. Single layer coating of appropriate index and thickness can be designed to reduce reflection to zero at a particular wavelength or index, but gradient index coatings are generally less sensitive to thickness or angle variations, or surface roughness. Other researchers have suggested that combinations of gradient index coatings and interference may be superior to either one separately.18 Overall, the co-sputtered AZO to SiO2 grin layer coating improved the transparency of aluminum-doped zinc oxide film at most wavelengths across the visible range from 400 to 700 nm. Applying the grin layer coating to a CZTS solar cell with AZO as TCO layer improved the solar cell’s quantum efficiency significantly, by a maximum of 11.9% at around 445 nm. It’s power conversion efficiency improved by about 10% at normal incidence and 20% or greater at all other angles. The solar cells short circuit current, open circuit voltage, fill factors, and therefore efficiency were increased for all angles in the experiment. The method used in this paper is repeatable and is a simple technique with minimal microfabrication steps compared to patterned coatings. The constituent elements in the grin layer SiO2 and AZO, are conventional, non-toxic, and earth-abundant. The technique in this work is convenient and effective for efficiency improvement for solar cells capped with transparent conducting oxides. We note that the modest increase in solar cell efficiency is limited by the intrinsic efficiency of

the material; much greater absolute efficiency increases would be expected for those solar cells with higher intrinsic efficiency. This grin layer coating works well over a wide range of wavelengths and a large incident angle range. This technology is also applicable to increasing the efficiency of displays, solid state phosphors, or any other technology in which it would be desirable to emit or collect light across a broad range of wavelengths and angles. This work was supported by the Office of Naval Research, under Grant No. N00014-11-1-0658, and by the Binghamton Advanced Diagnostic Laboratory, under Grant No. ADLG403.

1

B. T. Jheng, P. T. Liu, M. C. Wang, and M. C. Wu, Appl. Phys. Lett. 103, 052904 (2013). 2 J. Y. Chen and K. W. Sun, Sol. Energy Mater. Sol. Cells 94, 629 (2010). 3 K. Forberich, G. Dennler, M. C. Scharder, K. Hingerl, T. Fromherz, and C. J. Brabec, Thin Solid Films 516, 7167 (2008). 4 J. Zhu, Z. Yu, G. F. Burkhard, C. M. Hsu, S. T. Connor, Y. Xu, Q. Wang, M. McGehee, S. Fan, and Y. Cui, Nano. Lett. 9, 279 (2009). 5 L. Rayleigh, Proc. London Math. Soc. S1-11, 51 (1879). 6 R. Jacobsson and J. O. Martensson, Appl. Opt. 5, 29 (1966). 7 T. Dhakal, A. Nandur, R. Christian, P. Vasekar, S. Desu, C. Westgate, D. I. Koukis, D. J. Arenas, and D. B. Tanner, Sol. Energy 86, 1306 (2012). 8 G. B. Murdoch, S. Hinds, and E. H. Sargent, Appl. Phys. Lett. 94, 213301 (2009). 9 H. Kobayashi, T. Ishida, K. Nakamura, Y. Nakato, and H. Tsubomura, J. Appl. Phys. 72, 5288 (1992). 10 H. N. Cui and S. Q. Xi, Thin Solid Films 288, 325 (1996). 11 M. Hiramatsu, K. Imaeda, N. Horio, and M. Nawata, J. Vac. Sci. Technol. A 16, 669 (1998). 12 J. Hu, Z. Ge, A. Piquette, K. Mishra, and D. Klotzkin, Microwave Opt. Technol. Lett. 56, 256 (2014). 13 A. N. Noemaun, F. W. Mont, J. Cho, E. Fred Schubert, G. B. Kim, and C. Sone, J. Vac. Sci. Technol. A 29, 051302 (2011). 14 S. Zaitsu, T. Jitsuno, M. Nakatsuka, T. Yamanaka, and S. Motokoshi, Appl. Phys. Lett. 80, 2442 (2002). 15 C. Mitsas and D. I. Siapkas, Appl. Opt. 34, 1678 (1995). 16 A. Emrani, P. Vasekar, and C. Westgate, Sol. Energy 98, 335 (2013). 17 P. Banerjee, W. J. Lee, and K. R. Bae, J. Appl. Phys. 108, 043504 (2010). 18 M. Schubert, D. Poxson, F. Mont, J. Kim, and E. Schubert, Appl. Phys. Express 3, 082502 (2010).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 128.226.37.5 On: Wed, 19 Mar 2014 00:51:12