Session Fr-C2

Nano-and Microcrystalline Silicon: Photovoltaics II

Chair: Ruud Schropp, Eindhoven University of Technology

Fr-C2.1 10:30–10:50

Development of PECVD Microcrystalline Silicon Oxide as a Replacement for n-Type and Back TCO Layers in Amorphous Silicon Thin Film Solar Cells

Shin-Wei Liang, Hung-Jung Hsu, Li-Shin Chang, Cheng-Hang Hsu, and Chuang-Chuang Tsai

Department of Photonics, National Chiao Tung University, 1001 University Road, Hsinchu, Taiwan

Hydrogenated microcrystalline silicon oxide (μc-SiOx:H) has attracted much interest for the thin-film solar cell applications due to its superior combination of optical and electrical properties [1–3]. The characteristics of μc-SiOx:H can be tuned over a wide range by varying the deposition conditions in PECVD systems. For a-Si:H thin-film solar cells, a thin absorber is desirable for the device stability, so a good back reflector is important to provide sufficient photocurrent. A sputtered TCO layer and metal contact is used as highly reflective combination to enhance absorption in the absorber. Typically, an ex-situ sputtering step for TCO is needed. We have found that μc-SiOx:H can be a good replacement for back TCO, which can be deposited by PECVD in-situ. Consistent interface quality can be obtained without breaking the vacuum. The use of μc-SiOx:H as part of back reflecting structure can reduce the optical loss due to its wide optical bandgap and simplify the fabrication by the in-situ PECVD process. In this work, we employed μc-SiOx:H as a replacement for n-layer and back TCO layers in a-Si:H single-junction and a-Si:H/a-Si:H tandem solar cells. The influence of μc-SiOx:H(n) on the optical absorption and interface quality will be discussed.

The silicon-based thin films were deposited by a 27.12 MHz PECVD system with NF3 in-situ plasma cleaning and a load-lock system. The oxygen incorporation in μc-SiOx:H films was achieved by adding CO2 to highly H2-diluted SiH4. X-ray photoelectron spectroscopy, Raman spectroscopy, optical transmission, electrical conductivity, AM 1.5G solar simulator and external quantum efficiency were employed for the material and device characterization.

The μc-SiOx:H(n) films with different composition have been prepared. The oxygen content can be controlled from 3.5 to 37.3 at.%, with the corresponding conductivity changed from 31.2 to 3.4x10–9 S/cm. By using μc-SiOx:H(n) as a replacement for a-Si:H(n) and ITO in a-Si:H single-junction cells, the quantum efficiency at the wavelength from 500 nm to 680 nm increased. The short circuit current density (JSC) significantly increased from 14.00 to 15.01 mA/cm2, with the cell efficiency of 9.81%, VOC = 0.91 V and FF = 71.82%. The major improvement was originated from the increased optical absorption. The a-Si:H/a-Si:H tandem cell using μc-SiOx:H(n)/Ag back reflector exhibited an efficiency of 9.86%, with JSC = 7.60 mA/cm2, VOC = 1.78 V and FF = 72.80%.

In conclusion, we have found that μc-SiOx:H can be a replacement for both n-layer and back TCO. The resultant single-junction a-Si:H solar cell has an efficiency of 9.81%, while that of a-Si:H/a-Si:H tandem cell has an efficiency of 9.86%. The all in-situ process can reduce interface contamination and eliminate ion bombardment damage during sputtering. The simplified fabrication process may also be beneficial to industrial production.

[1] K. Haga et al., Jpn. J. Appl. Phys. 25 (1986) L39

[2] V. Smirnov et al., Phys. Status Solidi C 3 (2010) 1053

[3] A. Lambertz et al., J. Appl. Phys. 109, (2011) 113109

Keywords: solar cells, amorphous silicon, microcrystalline silicon oxide, PECVD

Fr-C2.2 10:50–11:10

p- and n-Type Microcrystalline Silicon Oxide (μc-SiOx:H) for Applications in Thin Film Silicon Tandem Solar Cells

V. Smirnov, A. Lambertz, S. Tillmanns, and F. Finger

IEK-5 Photovoltaik, Forschungszentrum Jülich GmbH, 52425 Jülich, Germany

In tandem solar cells, consisting of amorphous (a-Si:H) and microcrystalline (μc-Si:H) silicon sub-cells, silicon oxide (μc-SiOx:H) layers are placed between the sub cells, serving as an intermediate reflector (IR). Such IR layers should fulfill requirements like low parasitic absorption and appropriate refractive index, high carrier concentration to build up the electric field in the diode, serve as a nucleation layer to promote crystalline growth, and low series resistance in the recombination junction. Using μc-SiOx:H alloys we can enhance the optical band gap energy, adjust the refractive index over a considerable range, and dope the material p type as well as n type to make the material suitable for all these various tasks.

μc-SiOx:H layers were deposited by RF PECVD at 185°C from mixtures of silane (SiH4), carbon dioxide (CO2) and hydrogen (H2), with admixture of phosphine (PH3) and trimethylboron (TMB) for n- and p- doping, respectively. The oxygen content was varied by the changes in rCO2 = CO2/SiH4 during the deposition. Films were investigated by dark conductivity measurements, photothermal deflection and Raman spectroscopy to evaluate optical band gap (E04) and crystallinity, respectively. Solar cells were investigated by current-voltage (J-V) measurements under AM 1.5 illumination, quantum efficiency (QE) and reflectance measurements.

To be used as intermediate reflector, low refractive index (n) and a high band gap at a sufficient conductivity (above 10–5 S/cm) are desired. We show that conductivity values of 10–5 S/cm, high E04 around 2.6 eV and low n around 2 can be achieved for both p- and n-type μc-SiOx:H materials. We have found that at a given rCO2, adding of PH3 gas tends to increase crystallinity of μc-SiOx:H layers, while TMB tends to suppress crystalline growth. With respect to solar cell performance, the optimal μc-SiOx:H material was found to have n around 2.4. When implemented in the tandem cell as p-type μc SiOx:H IR, top cell current (JSCtop) is increased by 0.6 mA/cm2 up to 13.7 mA/cm2. An application of n-type μc SiOx:H IR results in the increase of the JSCtop up to of 14.2 mA/cm2. While the JSCtop is higher when n-type μc SiOx:H IR is used, the total cell current (defined as the sum of currents from top and bottom sub-cells) is reduced relative to the cell with p-type μc SiOx:H IR, by 0.5 mA/cm2. Improvement of reflectance spectra and internal quantum efficiencies indicates that the gain in the total cell current can be attributed to the reduced parasitic absorption in the spectral range between 600 nm to 900 nm.

Our results evidence that the properties of μc-SiOx:H can be conveniently tuned over a wide range to fulfill the necessary set of requirements for applications in tandem solar cells. A high efficiency of 13.5% was achieved in the case of tandem solar cell utilizing n-type μc-SiOx:H intermediate reflector (FF = 71.7%, VOC = 1.33V, JSC = 14.1mA/cm2). A remarkable increase in top cell current and overall efficiency due to incorporation of both p- and n-type μc-SiOx:H intermediate reflector demonstrates the suitability and high potential of μc-SiOx:H as a functional layer in tandem devices.

Keywords: thin film silicon, microcrystalline silicon oxide, tandem solar cells

Fr-C2.3 11:10–11:30

Light Trapping in Silicon Thin Films Measured by Raman Spectroscopy

Martin Ledinský (1), Kristína Ganzerová (1), Antonín Fejfar (1), Fanny Meillaud (2), Grégory Bugnon (2), and Christophe Ballif (2)

1. Institute of Physics, Academy of Sciences of the Czech Republic, v.v.i., Cukrovarnicka 10, 162 00, Prague 6, Czech Republic

2. Photovoltaics and Thin Film Electronics Laboratory, Ecole Polytechnique Fédérale de Lausanne (EPFL), Rue A.-L. Breguet 2, CH-2000 Neuchâtel, Switzerland

In order to enhance insufficient absorption in μc-Si:H thin film solar cells in the near infrared (NIR) part of the spectra, rough front contacts (made of transparent conductive oxides, TCOs) are used. Rough substrate scatters the incoming light and scattered photons may enter the silicon layer under large angles. This effect together with multiple internal reflections prolong the photon path in the cell and lead to the enhanced absorption in NIR region. This effect is essential for all indirect semiconductors and is intensively studied. In case of full PV cells, light trapping may be assessed from the short circuit current JSC or external quantum efficiency. However, both measurements are sensitive to the electronic quality of the film and contacts and so the effect of light trapping may be hidden by recombination from bulk/interface defects, imperfect doping, etc.

This situation motivated us to test a novel use of Raman spectroscopy for light trapping assessment. Raman spectroscopy is a nondestructive and contactless tool frequently used for silicon characterization. In case of microcrystalline silicon (μc-Si:H) thin films, Raman spectra are used to calculate the crystalline volume ratio (crystallinity [1]) and additionally stress may be deduced from the shift of the LO-TO phonon band from the standard position at 520 cm–1. Usually, the absolute intensity of the Raman band is not used for characterization purposes. But in case of weakly absorbed Raman excitation laser (from the NIR part of the spectra), the Raman intensity of the μc-Si:H Raman peak depends on the sample roughness. Indeed, light trapping effect leads to an enhancement of the photons path in the μc-Si:H material and the Raman interaction path thus increases [2]. Therefore the Raman intensity may be used to assess light-trapping efficiency in μc-Si:H.

In this paper, a series of μc-Si:H layers deposited on substrates with different roughness is hence studied. A 785 nm laser was used to excite the Raman spectra. The Raman intensity is compared with root mean square (RMS) roughness of the sample measured by atomic force microscopy in the tapping mode. With increased roughness the Raman intensity also increases. The same effect was found also for the frequently used 633 nm laser. By contrast, intensity of Raman signal excited by highly absorbed 514 nm and 442 nm laser show nearly no dependence on substrate roughness.

These results demonstrate the use of Raman spectroscopy as a noncontact method suitable for light-trapping characterization. On the other hand, the Raman intensity (and thus also the light-trapping) depends not only on the RMS roughness, but also on the substrate morphology. Therefore the final goal of this study is to find an appropriate substrate which offers good light-trapping properties, while enabling low crack density μc-Si:H layers.

[1] M. Ledinský, A. Vetushka, J. Stuchlík, T. Mates, A. Fejfar, J. Kočka, J. Non-Cryst. Sol. 354 (2008) 2253–2257

[2] M. Ledinský, M. Hakl, L. Ondič, K. Ganzerová, A. Vetushka, A. Fejfar and J. Kočka, 27th European Photovoltaics Solar Energy Conference and Exhibition (2012) Frankfurt, 3CV.2.24

Keywords: light trapping, Raman spectroscopy, AFM, roughness

Fr-C2.4 11:30–11:50

Improved Light Trapping Effect for Thin-Film Silicon Solar Cells by New White Glass

Hidetoshi Wada (1), Bancha Janthong (1), Porponth Sichanugrist (1), and Makoto Konagai (1,2)

1. Department of Physical Electronics, Tokyo Institute of Technology, 2-12-1, Oookayama, Meguro-ku, Tokyo 152-8552, Japan

2. Photovoltaic Research Center (PVREC), Tokyo Institute of Technology, 2-12-1, Oookayama, Meguro-ku, Tokyo 152-8552, Japan

The optimized W-textured substrate is a considerable approach in order to further improve light-trapping effect and thus increase solar cell performance. Recently, the cost-effective white glass was introduced for applications to solar cells as a substrate. We have successfully deposited the W-textured ZnO:B on the RIE-etched white glass manufactured by Company A (named as Type A) as a front transparent conductive oxide (TCO) film using MOCVD technique and the improved cell performance of 13.3% has been obtained. However, this kind of glass was manufactured by floating method which will have SnO2 existed at its surface. This SnO2 will make RIE glass etching difficult. Furthermore, as reported by AIST of Japan for the n-i-p type microcrystalline (μc-Si) solar cells fabricated on Si wafer, we consider that in order to obtain the same result with high short-circuit-current density for glass type substrate, an optimum feature size of ZnO:B coated on a glass substrate have to be equal to or slightly larger than the cell thickness. Therefore, the further investigation of the ZnO-coated substrate is needed to optimize the optical properties and surface morphology of the ZnO-coated substrates. In this study, we have tried to use another type of white glass manufactured by Company B (named as Type B) and enlarge the feature size or period of the ZnO coated on it by changing the condition of RIE.

Up to now, we have deposited the 1.6 μm thick ZnO:B film on the RIE-etched white glass (type B) to compare with the one deposited on the type A and it was found that the surface morphology of the type B white glass was mountain-chain-like double texture which is the same as the one on type A. However it was found that it is much easier to etch type B glass. This may come from the reason that type B glass is manufactured by up-draw technology process which has no protective film on its surface as found in the case of type A. When longer RIE-etching time was used, the feature size of ZnO became larger in length so that the size of more than 10 μm has been observed. However, its haze ratio at 800 nm was lower down to 46% comparing with 93% which has been observed for the shorter etching time. Further investigation are carried out to find out the reason.

Also, an existence of fewer cracks at the bottom of valley in the textured substrate is observed as the diameter of concave becomes larger. The density of cracks of μc-Si are supposed to be lower if we are able to larger diameter or feature size of the ZnO:B film. Therefore, higher short circuit current density as well as the conversion efficiency of solar cells can be expected. We will also report about the evaluation results of p-i-n μc-Si:H solar cell fabricated on this newly developed white glass substrate.

Keywords: light-trapping, double-texture, RIE, ZnO:B

Fr-C2.5 11:50–12:10

Thermodynamic Behavior of Periodic and Random Light-trapping Structures in Thin Film Silicon Solar Cells

Brian Maynard, Hui Zhao, and Eric Schiff

Dept. of Physics, Syracuse University, Syracuse, NY 13244-1130

Light-trapping is crucial to thin-film silicon solar cells: on one hand, it compensates for the slow rise of absorption coefficient with photon energy, and on the other it takes advantage of the fairly large refractive index n. The well-known enhancement factor for the absorptance is 4n2 ≈ 50 with ideal "thermodynamic" light-trapping, but in practice thin-film solar cells have enhancements of less than half of this magnitude.

In this paper we present general thermodynamic calculations for light-trapping enhancements for both periodic and random light-scattering morphologies, and we compare them both with measurements on thin-film silicon solar cells in both substrate and superstrate configurations. Such calculations are the basis for the original 4n2 result with random morphologies, and for more recent work showing enhancements beyond 4n2 with some periodic structures. The calculations assume equipartition for all accessible waveguide modes of the thin-film solar cell. This approach differs from calculations for particular periodic morphologies using finite-difference time-domain (FDTD) and related calculations, which also predict enhancements beyond 4n2.

In this paper we present extensions of the thermodynamic approach to incorporate parasitic absorption by conducting oxide layers in both random and periodic structures. For random structures we obtain good agreement between the calculations and external quantum efficiency and reflectance measurements in a set of substrate solar cells. The result is of practical importance because it suggests that backreflector structures can support thermodynamic light-trapping without contributing substantially to parasitic loss. We also examine recent measurements showing nearly equal light-trapping with periodic and random morphologies for superstrate solar cells. For the periodic morphologies, quantum efficiency measurements do not exhibit the sharp spectral structures that are typical of FDTD calculations; the thermodynamic calculations are in closer agreement with the measurements.

Keywords: thin-film silicon, solar cells, light trapping

Fr-C2.6 (invited) 12:10–12:40

Nanophotonic Light Trapping in Ultrathin Film Solar Cells

Vivian E. Ferry

University of California, Berkeley, and Lawrence Berkeley National Laboratory, Berkeley CA 94720 USA

The integration of nanostructures with solar cells offers the ability to guide and confine light in nanoscale dimensions. By designing the complex dielectric function and nanoscale geometry, enhanced absorption and decreased reflection can be achieved in ultrathin volumes. Ultrathin semiconductor volumes in photovoltaics are attractive for a number of fundamental and practical reasons, enabling improved efficiencies through increased open-circuit voltages while also reducing the cost and fabrication time of devices.

Plasmonic nanostructures possess high scattering cross sections, improving coupling of incident sunlight into guided and localized resonant modes of a solar cell. These metallic nanostructures can be built directly into the back contact of a thin film silicon solar cell using an inexpensive imprinting process, which results in both large area and precise nanopatterns. Semiconductor nanostructures with highly absorbing Mie resonances can also be used to enhance light absorption and coupling into thin films. By combining back-contact plasmonic nanostructures with front-surface semiconductor nanostructures, broadband light trapping can be achieved by tuning resonances to different regions of the solar spectrum.

Coupling to guided modes can be controlled and understood through the use of different nanostructure patterns, including periodic, quasiperiodic, and pseudorandom arrangements. The ability to precisely locate these nanostructures allows for the construction of devices with broadband and isotropic light trapping in less than 100 nm of amorphous Si. In another design, core-shell nanowire geometries are used to enhance photocurrent by coupling to leaky-mode resonances.

Electromagnetic simulation agrees closely with experimental quantum efficiency measurements on these devices. For a more complete model, a three-dimensional electromagnetic model is combined with a three-dimensional device physics model to fully calculate device efficiency. In nanophotonic solar cells, the spatial modification of light absorption due to nanostructuring can influence carrier collection, and here nanostructures are shown that preferentially couple sunlight to regions of the device where carriers are efficiently collected.