basis of efficiently scattering colloidal metal nanoparticles. The second one is the use of plasmon-induced hot electrons ejected by metal nanoparticles into Si for ...
1
PHYSICS, CHEMISTRY AND APPLICATION OF NANOSTRUCTURES, 2015
ON THE APPLICATION OF PLASMONIC METAL NANOPARTICLES IN THIN-FILM SILICON SOLAR CELLS VLADIMIR S. CHIRVONY, RAÚL GARCÍA-CALZADA, JUAN MARTÍNEZPASTOR UMDO, Instituto de Ciencia de los Materiales, Universidad de Valencia, P.O. Box 22085, 46071 Valencia, Spain SVEN RING, BERND STANNOWSKI, RUTGER SCHLATMANN Helmholtz Zentrum Berlin/PVcomB, Schwarzschildstr. 3, 12489 Berlin, Germany RAFAEL ABARGUES, PEDRO RODRIGUEZ-CANTÓ Intenanomat S.L., C/Catedrático José Beltrán 2, 46980 Paterna, Spain Last achievements in two main applications of plasmonic metal nanoparticles for improving the efficiency of thin-film silicon solar cells will be reviewed and our own results presented. One of the applications is concerned with fabrication and investigation of diffuse rear reflectors on the basis of efficiently scattering colloidal metal nanoparticles. The second one is the use of plasmon-induced hot electrons ejected by metal nanoparticles into Si for extending Si solar cell photoresponce toward wavelengths longer than Si bandgap.
1. Introduction Thin-film (TF) Si-based materials, owing to their low manufacturing cost and because of the high natural abundance and low toxicity of Si, are considered now as main materials for the 2nd generation solar cells. However, micrometer scale thickness of the light absorbing layers of these materials, which are usually hydrogenated microcrystalline Si (mc-Si:H) and/or hydrogenated amorphous Si (a-Si:H), makes them semitransparent in the near-infrared (NIR) region. In such a situation, integration of light trapping elements into solar cells is essential for increasing the optical path length in the absorbing Si layer. The idea of using plasmonic metal nanoparticles (MNPs) to increase light trapping in optoelectronic devices, first of all in solar cells, has been actively discussed in the literature in the early 2000s. The results of the first stage of the experimental and theoretical studies of the mechanisms of influence of the plasmonic MNPs on the effectiveness of light trapping in Si-based solar cells were summarized in 2010 in a well-known review [1] where the authors gave very optimistic prognosis on the possibility of increasing light trapping efficiency and light absorption in Si solar cells due to deposition of efficiently scattering MNPs on a Si solar cell front surface. However, later, it has been
2
recognized that the deposition of MNPs on the front surface of Si solar cell is equivalent to deposition of an ARC layer having parasitic absorption at wavelengths near the surface plasmon resonance. Thus, since the beginning of 2010s the application of big-size (100 nm and more) Ag and Au NPs as plasmonic back reflectors (BRs) became the main approach for increasing light trapping in TF Si solar cells. However, only very recently, through the careful common optimization of plasmonic BRs optical and electronic circuits, light trapping efficiency was achieved comparable with stateof-the-art surface-textured substrates [2]. In previous works, incorporation of metallic NPs in TF Si solar cells was typically achieved through physical approaches, such as thermal annealing of evaporated or sputtered metal films. Although this method is straightforward, it is challenging in regulating independently size of NPs and the coverage. Furthermore, the annealing method in principle is not suitable for superstrate TF Si solar cell geometries, which are the only geometries used for production of commercial TF Si solar cells. In such a situation we propose a more flexible approach, which consists in room-temperature deposition of preliminary chemically synthesized, homogeneous in size colloidal metal NPs onto the rear side of TF Si solar cells as a back scattering diffuse reflector. This approach is not yet widely described in the literature. Below we present main achievements of our group in the fabrication and optical characterization of monolayers of Ag NPs (coverage 10-20%) deposited on glass, Si and other substrates as well as the first results of the monolayer application as a back reflector to TF Si solar cells. 2. Experimental Results and Discussion We deposited a sub-monolayer of 200 nm colloidal (in water suspension) Ag NPs on the rear side (i.e. on ZnO surface) of 10x10 cm2 ZnO/a-Si:H/glass solar cell structure. Providing homogeneous distribution of deposited Ag NPs on micrometer-centimeter scales is rather challenging and we achieved it by a special treatment of ZnO surface to overcome its hydrophobicity, by a special choice of organic ligands around Ag NPs, and by applying a special method of deposition. SEM image of a layer of 200 nm Ag NPs on ZnO/glass surface is shown in the inset of Fig.1. The most important optical characteristics of the plasmonic NPs layers, which characterize the ability of the layers to scatter light diffusively without parasitic absorption, are the spectra of diffuse reflectance Rd(λ), diffuse transmittance Td(λ), and true absorption A(λ). Investigation of these spectra requires a use of spectrophotometers provided with integrating spheres with the working diapason from UV until near-IR. The above-mentioned spectra measured for the Ag NPs layers deposited on ZnO/glass are shown in Fig. 1. The
3
parameter Rd+Td characterizes the percentage of the incident light, which will diffusively reflected either directly by NPs or diffusively transmitted with the following reflection by a specular mirror, which is usually mounted behind the metal NPs layer. All this light, which is characterized by Rd+Td, will be cached by waveguide modes of a thin Si layer, directed along it and finally absorbed.
Samples 0526-1,2,3 50
Rd+Td
Rd, Td, A
40
3 2 1
30 Rd
20 10 0
A 400
600
800
1000
Wavelength, nm Figure 1. Spectra of diffuse reflectance Rd, diffuse transmittance Td, and true absorption A measured for a monolayer of colloidal 200 nm AgNPs (coverage about 20%) deposited on ZnO/glass substrate (the spectra were measured for three samples, which demonstrated slightly different coverage factors). The inset shows SEM image of the same Ag NP layer.
It is worth to note that (i) absorption spectrum is quite different from scattering spectra (because they belong to different multipoles), and (ii) there is an unexpected parasitic absorption (10-15%) between 700 and 1200 nm which can be preliminary explained as a result of Ag NPs aggregation and/or Ag oxidation. In spite of this, such 20%-coverage Ag NP layer deposited on the rear side of ZnO/a-Si:H/glass solar cell structure provides sufficiently marked enhancement of efficiency (see Fig. 2).
4
144501-3
Moulin et al.
deposited through a mask, on half of the substrate only, so that devices without NPs can be used as references. The devices without NPs (referred to as i-n and i-p) exhibit an absorptance larger than 60% between 330 nm and -‐2.00 600 nm, reaching a maximum of nearly 100% at 550 nm (Fig. 3(a), squares). Between 550 nm and 700 nm, the ab-‐4.00 sorptance decreases abruptly. For k > 700 nm, values below 10% are measured. For k > 900 nm, the absorptance is negligible. This rapid decrease is explained by the low absorption -‐6.00 coefficient of a-Si:H in the NIR spectrum. The integration of NPs leads to a significant increase in the absorptance, for -‐8.00 both devices, at k > 600 nm (Fig. 3(a), circles). The absorptance reaches almost 100% at 800 nm and 900 nm for the NPs-i-p and NPs-i-n devices, respectively, before declining -‐10.00 at longer wavelengths. The slight differences found for both devices probably result from the use of different front layers, -‐12.00 namely the p- and n-doped layers, and a remaining inhomogeneity in the layer thicknesses of the co-deposited devices. The broad maximum in absorptance between 700 nm and -‐14.00 V 0.6 nm associated with the LSPP0.8 resonance 0 0.1 0.2 0.3 1100 0.4 0.5 0.7 0.9 is presumably 1 caused by (i) the broad distribution of shapes and sizes of the 9,10,28,29 9 (ii) the interaction between the NPs, and (iii) NPs, Figure 2. IV characteristics for a-Si:H TF solar cells in superstrate configuration with (orange) and the presence of the silver back reflector, which leads to interwithout (blue) 200 nm colloidal Ag NPs as diffuse BR integrated between a-Si:H and Ag mirror. ferences that partly mask the LSPP resonance. At k < 580 nm, the EQE of the i-p device is higher than Another possible application of plasmonic MNPs in TF theto a the EQE of the i-n device (cf. Si Fig.solar 3(b)). cells This isisdue lower parasitic absorptance in the front a-SiC:H p-layer than use of plasmon-induced photoemission of electrons from a metal. Indeed, if in the a-Si:H n-layer. In accordance with the absorptance metal NPs are embedded into semiconductor the extremely data, the EQEs ofmaterial, both devices without NPs drop high between
mA/cm2
0.00
electric field around a MNP can result in excitation of high-energy (“hot”) electrons, which overcome the metal-semiconductor Schottky barrier and are transferred into the semiconductor conductive band. Therefore, an excitation of metal NPs within localized surface plasmon absorption band will result in photocurrent in semiconductor even if the metal NP plasmonic band energetically lies below the semiconductor bandgap. Recent literature results in this field are shown in Fig.3. Results of our group in this field will be also presented. Figure 3. External quantum efficiency (EQE) of i-n and i-p photosensitive devices (filled and open symbols, respectively) with (circles) and without (squares) Ag NPs. The inset shows an enlargement of the EQE in the long wavelength region (delimited by the dashed box) [3].
References 1. 2. 3.
FIG. 3. Absorptance as a function of wavelength (a) and external quantum
efficiency at !0.5 V (b) of i-n and i-p photosensitive devices (filled and H. A. Atwater et al., Nature Mater. 9, 205, 2010. open symbols, respectively) with (circles) and without (squares) nanoparH. Tan, et al., Nano Lett. 12, 4070 (2012). ticles. The inset in (b) shows an enlargement of the EQE in the long waveregion (delimited by the dashed box). Please note that the EQE data E. A. Moulin et al., J. Appl. Phys.length 113, 144501 (2013). of the NPs-i-p device measured at !2 V is added in the inset of (b).
J. App
550 nm and 700 nm and reach a n 750 nm. The EQEs are not modified range by incorporating the NPs. T extraction and collection of the gen noticeably altered by the presence of EQE of the NPs-i-n device has alread value, confirming the satisfying extra carriers under moderate bias voltage. in the NIR wavelength region is no collection, a definitive number can be ternal quantum efficiency (IQE) of th generation mechanism of the NPs-i that most of the light in the NIR reg NPs (or by the first nanometers of NPs), an IQE of around 5% and 3% 900 nm, respectively. The above result can be expl based on the LSPP-induced photoge trons in the NPs (or from defect stat the transport of these electrons tow (cf. Fig. 1). However, if the n-doped p-doped layer, we would expect the from the NPs towards the front co prohibited. Interestingly, though, th shows a photocurrent in the NIR wa bias voltage of !0.5 V, the EQE sign bias voltage of !2 V, the EQE reac (Fig. 3(b), inset, open circles). As th of trap-assisted tunnelling of electr contact towards the NPs can be e thickness of the absorber i-layer use study, this result indicates the exis based on the photogeneration and tr device. This underlines the reciproc involved in the photogeneration pro port for both the NPs-i-n and the N basis of the model illustrated in Fig. Fig. 4 that depicts the LSPP-induc “hot” holes in the NPs (process 1)— vicinity (process 2)—and the tra towards the p-doped contact. As the photocurrent of the NPs-i rate in the NIR wavelength region w bias, no definitive number can be giv the associated photogeneration proce The photocurrent might suffer from collection of holes, even at a reverse of the following reasons: (i) the lo a-Si:H,30 (ii) the existence of a barri port through the i-layer, and/or (iii) th the NPs/a-Si:H interface resulting i recombination. To investigate this, a symmetri device has been prepared. With this ap eration process—based on the photo (see Fig. 1) as well as holes (see Fig. 4 port mechanism can be addressed inverting the polarity of the applied bi the same device can be employed for
Downloaded 15 Apr 2013 to 216.47.136.20. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jap.a
