A study of shape optimization on the metallic nanoparticles for thinfilm solar cells
 Shiwei Zhou^{1}Email author,
 Xiaodong Huang^{1},
 Qing Li^{2} and
 Yi Min Xie^{1}Email author
DOI: 10.1186/1556276X8447
© Zhou et al.; licensee Springer. 2013
Received: 1 August 2013
Accepted: 21 October 2013
Published: 29 October 2013
Abstract
The shape of metallic nanoparticles used to enhance the performance of thinfilm solar cells is described by Gielis' superformula and optimized by an evolutionary algorithm. As a result, we have found a lenslike nanoparticle capable of improving the short circuit current density to 19.93 mA/cm^{2}. Compared with a twoscale nanospherical configuration recently reported to synthesize the merits of large and small spheres into a single structure, the optimized nanoparticle enables the solar cell to achieve a further 7.75% improvement in the current density and is much more fabrication friendly due to its simple shape and tolerance to geometrical distortions.
Background
For some years, the high costs of silicon materials and fabrication make photovoltaics less competitive with electricity generation from fossil fuels even though it has potential to meet the soaring energy demands nowadays. Recently, many advanced light trapping techniques such as dye sensitization[1], plasmonic nanostructures[2–8], and nanodent plasmonic substrates[9–11] allow sunlight to be well absorbed within a very thin active absorber layer (just a few hundred nanometers). Therefore, the consumption of the absorber material was considerably reduced[2, 3, 7, 12, 13]. Remarkable outcomes have been achieved by depositing metallic nanoparticles into the dielectric layer between the Si layer and metallic back surface, by which the light path is optically prolonged as the sunlight can be scattered into the active layer at larger angles and induces extraordinarily strong local field intensity in the vicinity of metallic nanoparticles[2–5, 8]. Such a phenomenon caused by the interaction of light with the nanostructures is termed as plasmon. Not all probable factors associated with thinfilm solar cells such as the optical properties of constituting materials and environmental stimuli[6, 14] have been thoroughly considered in the optimization, and this paper is confined to the influence of size and shape of nanoparticles only. In addition to various randomly shaped structures[3, 4], the effects of some primitive geometries such as cylinder, cone, sphere, and hemisphere on light trapping have been investigated extensively[3, 4, 8, 15, 16]. It was reported that the cylindrical and hemispherical particles have better performance than spherical particles[4]. The investigation reveals that the Ag nanorod deposited on a SiO_{2} substrate has the strongest ability to enhance the resonance energy transfer rate when its cross section is a circle[17], which in turn proved the importance of the shape on the surface plasmon. Furthermore, a complicated twolevel hierarchical nanostructure consisting of evenly distributed small spheres on the surface of a large sphere was found to have the virtues of both large and small spheres and therefore benefit the current enhancement considerably[5, 18].
Nevertheless, complex shapes have not been explored for their plasmonic properties so far mainly due to the difficulties in 3D modeling and optimization. As a class of unified mathematical expression, Gielis' superformula demonstrates its simplicity and generality of formulating a wide variety of 3D geometries ranging from common shapes like sphere, cube, octahedron, and cylinder to highly complex structures via changing a small number of parameters[19]. A significant advantage of using Gielis' superformula for shape optimization is to facilitate the integration with evolutionary algorithm[20], enabling to search for such parameters by which the optimal structure can be constructed. In view of its recent success in investigating the geometry for a soft porous system (i.e., an adaptive structure undergoing large deformation[21] and plasmonic nanowires[22] in 2D), it is prospective to explore this superformula in 3D by seeking for the optimal nanoparticle for thinfilm solar cells.
Methods
Since the nanoparticles are periodically deposited and the incident light is normal to the front surface in this study, it is sufficient to restrain the modeling region to a representative volume element (RVE) with periodic boundary conditions applied to its bilateral surfaces parallel with the z axis. The open boundaries on the input and output sides are truncated by perfectly matched layers, whose distances to the top and bottom of the solar cell are 200 and 100 nm, respectively. In the RVE, the solar source is placed on a plane whose distance to the SnO_{2}:F layer is 100 nm. The electromagnetic field within the RVE is governed by Maxwell's equations and solved by the finite difference time domain (FDTD) algorithm[24]. In the present study, we utilize the FDTD program from Lumerical[25] as it has been used by many researchers in this area.
where e denotes the charge on an electron, h the Plank constant, λ the wavelength, and c the speed of light in vacuum. The quantum efficiency QE(λ) = P_{abs}(λ)/P_{in}(λ) is the ratio of the power of the absorbed light P_{abs}(λ) to that of the incident light P_{in}(λ) within the active film. I_{AM1.5} stands for the relevant part of the solar spectral irradiance. Once Maxwell's system is solved, the J_{sc}, electric intensity, absorption, as well as other relevant factors can be computed to any predefined level of accuracy.
Based on the above description, the optimization becomes a problem when searching for a stationary point in the hyperspace defined by seven design variables (s, m, n_{1}, n_{2}, q, n_{4}, n_{5}), so an extreme value of the cost function is achieved. Now that the number of design variables is relatively small, we adopt the evolutional algorithm[20] for the optimization. Such a nongradient method is rather simple but fairly efficient and is especially suitable for optical optimization in which the gradient of the cost function with respect to the design variables is often too sensitive to be controlled and stabilized numerically in terms of our previous studies[22, 23, 27–29]. The evolutionary algorithm starts from a set of parent vectors randomly selected in the given parametric space. For each parent vector, the fitness, namely short circuit current density J_{sc}, is calculated. Then, an offspring is introduced by a mutation rule defined as the summation of the weighted difference between a pair of parent vectors and the third one. The offspring is required to mend by a crossover process for improving the diversity. In this step, the weighting factor and crossover probability are set to be 0.5 and 1, respectively, for the following examples. The mutation and crossover processes are repeated until the best performance is obtained. Importantly, predefined constraints like the bounds of design variables are incorporated into the optimization algorithm to avoid missing feasible solutions and generating unmanufacturable structures.
Results and discussion
Intermediate nanoparticles in the optimization
J _{sc}  m  n _{1}  n _{2}  q  m _{1}  m _{2}  s  Nanoparticle 

17.9778  142  6,917.3889  188.7937  6  246.5756  337.3163  71.1604 

18.6231  158  19,964.3647  711.4337  10  15,278.7838  1,812.3877  81.6840 

19.1814  72  1,774.9  74.6116  12  1,315.1222  473.4534  49.0618 

19.4887  10  1,113.2549  196.8687  116  1,268.8916  3.2545  43.1573 

The Gielis parameters for this optimal lenslike structure are m = 2, n_{1} = 767.6272, n_{2} = 1,379.6088, q = 152, n_{4} = 18,071.6197, and n_{5} = 255.3518, which result in 4.86% nanoparticle coverage for the thinfilm solar cell when the periodicity of nanoparticles is 646.58 nm and the weighting factor is s = 45.08. As the coverage is much smaller than the widely accepted optimal coverage (10%)[5] with randomly shaped nanoparticles, the fabrication cost is likely to be significantly reduced because less silver material is consumed. On the other hand, the volume ratio of this structure to the benchmarking twoscale sphere is around 14.76%, indicating that the increase of J_{sc} is attributed to the shape rather than the volume because larger nanoparticles generally induce stronger surface plasmon and therefore a larger J_{sc}. Additional evidence supporting this claim is that J_{sc} is not directly related to the areas enclosed by the contours as demonstrated in Figure 3b,c.
A glimpse of the nearfield optical contour helps to explain the enhancement of short circuit currency density. Figure 3d,e illustrates the nearfield electric intensity (E^{2}) on vertical and horizontal symmetry planes, respectively, at the time when E^{2} reaches the maximum on the horizontal symmetry plane (the distribution of the E field is recorded by a time monitor in the simulation). It is seen that the nearfield intensity is evidently strengthened in the bottom part (Figure 3d) and the bilateral vertices (Figure 3e) around the nanoparticle. The solar cell can make use of such a strong nearfield enhancement to increase the optical absorption in the active layer, thereby amplifying J_{sc}. Interestingly, similar shapes to the crosssectional geometry of these lenslike structures have been reported elsewhere by Macías et al. recently when they attempted to search for the 2D shapes with maximal scattering geometry[22]. Such a correlation signifies the crucial role played by this structure in inducing strong surface plasmon.
The logarithmic scale of the absorbed power per unit volume ∫ ωϵ″E^{2}dV/2 (ω is the angular frequency and ϵ'' the imaginary part of permittivity) in different layers of the solar cell on the vertical symmetry plane at the frequency where the maximal absorption is attained (λ = 596 nm) is plotted in Figure 3f. It is noted that the enhanced energy absorption in the active layer is ascribed to the strong local field intensity, surface plasmon polariton (SPP) mode, scattering effect of nanoparticles, and FabryPérot (FP) resonance effect. The SPP mode is bounded to tens of nanometers away from the nanostructure surface and decays exponentially; thus, its effects on the Si film, only 20 nm above the nanostructure, are rather evident. Moreover, the extremely strong intensity of local electricity leads to photonic mode predominant in the active layer and contributes to the energy absorption locally. This mechanism explains better energy absorption at the bottom of the active layer which is just above the nanostructure. The gradient direction of energy absorption (as arrowed in Figure 3f) shows that the energy flux interacts with the back Ag surface at large angles, illustrating that the resonant electromagnetic wave impinges upon the back metal and is reflected into the active layer obliquely. Therefore, the reflection from the back Ag surface, together with the scattering and diffraction caused by the nanostructural geometry, results in strong absorption at the bottom corners of the active layer. The superiority of the lenslike nanoparticle to other structures such as cube, sphere, and cylinder in current enhancement can be attributed to the coexistence of sharp tips and smoothing surfaces  the former helps to induce local surface plasmon and thus the strong field nearby while the latter enables the sunlight to be effectively scattered at large angles. These two factors are significant to solar cells and have been analogically implemented by an array of small spheres distributed on a large sphere[5].
Conclusions
This work systemically explored the relationship between a variety of complex shapes parameterized by the 3D Gielis superformula and J_{sc}. We have found that lenslike nanoparticles can generate significantly higher J_{sc} than the twoscale nanospheres that were devised previously. It is important to note that the optimized nanoparticle is more fabrication friendly due to its simpler shape and insensitivity to geometric variations. Numerical simulations have demonstrated that this nanoparticle can yield strong local field enhancement, large scattering angle, FabryPérot resonance, and high surface plasmon polariton, which are suggested as the effective ways to enhance the absorption in the active layer of thinfilm solar cells[2].
Declarations
Acknowledgements
This work was supported by an Australian Research Council Discovery Early Career Researcher Award (project number DE120102906) and an Australian Research Council Discovery Project grant (DP110104698). Dr Shiwei Zhou is the recipient of an ARC Discovery Early Career Researcher Award (project number DE120102906).
Authors’ Affiliations
References
 O'Regan B, Gratzel M: A lowcost, highefficiency solar cell based on dyesensitized colloidal TiO2 films. Nature 1991, 353: 737–740. 10.1038/353737a0View ArticleGoogle Scholar
 Atwater HA, Polman A: Plasmonics for improved photovoltaic devices. Nat Mater 2010, 9: 205–213. 10.1038/nmat2629View ArticleGoogle Scholar
 Beck FJ, Polman A, Catchpole KR: Tunable light trapping for solar cells using localized surface plasmons. J Appl Phys 2009, 105: 114310. 10.1063/1.3140609View ArticleGoogle Scholar
 Catchpole KR, Polman A: Design principles for particle plasmon enhanced solar cells. Appl Phys Lett 2008, 93: 191113. 10.1063/1.3021072View ArticleGoogle Scholar
 Chen X, Jia BH, Saha JK, Cai BY, Stokes N, Qiao Q, Wang YQ, Shi ZR, Gu M: Broadband enhancement in thinfilm amorphous silicon solar cell enabled by nucleated silver nanoparticles. Nano Lett 2012, 12: 2187–2192. 10.1021/nl203463zView ArticleGoogle Scholar
 Ferry VE, Munday JN, Atwater HA: Design consideration for plasmonic photovoltaics. Adv Mater 2010, 22: 4794–4808. 10.1002/adma.201000488View ArticleGoogle Scholar
 Pala RA, White J, Barnard E, Liu J, Brongersma ML: Design of plasmonic thinfilm solar cells with broadband absorption enhancements. Adv Mater 2009, 21: 3504–3509. 10.1002/adma.200900331View ArticleGoogle Scholar
 Pillai S, Catchpole KR, Trupke T, Green MA: Surface plasmon enhanced silicon solar cells. J Appl Phys 2007, 101: 093105. 10.1063/1.2734885View ArticleGoogle Scholar
 Zhu J, Hsu CM, Yu ZF, Fan SH, Cui Y: Nanodome solar cells with efficient light management and selfcleaning. Nano Lett 2010, 10: 1979–1984. 10.1021/nl9034237View ArticleGoogle Scholar
 Battaglia C, Hsu CM, Soderstrom K, Escarre J, Haug FJ, Charriere M, Boccard M, Despeisse M, Alexander DTL, Cantoni M, Cui Y, Ballif C: Light trapping in solar cells: can periodic beat random? ACS NANO 2012, 6: 2790–2797. 10.1021/nn300287jView ArticleGoogle Scholar
 Huang HT, Lu LF, Wang J, Yang J, Leung SF, Wang YQ, Chen D, Chen XY, Shen GZ, Li DD, Fan ZY: Performance enhancement of thinfilm amorphous silicon solar cells with low cost nanodent plasmonic substrates. Ener Environ Sci 2013, 6: 2965–2971. 10.1039/c3ee41139gView ArticleGoogle Scholar
 Gramotnev DK, Bozhevolnyi SI: Plasmonics beyond the diffraction limit. Nat Photonics 2010, 4: 83–91. 10.1038/nphoton.2009.282View ArticleGoogle Scholar
 Mühlschlegel P, Eisler HJ, Martin OJF, Hecht B, Pohl DW: Resonant optical antennas. Science 2005, 308: 1607–1609. 10.1126/science.1111886View ArticleGoogle Scholar
 Ren R, Guo YX, Zhu RH: Design of a plasmonic back reflector for silicon nanowire decorated solar cells. Opt Lett 2012, 37: 4245–4247. 10.1364/OL.37.004245View ArticleGoogle Scholar
 Ouyang Z, Pillai S, Beck F, Kunz O, Varlamov S, Catchpole KR, Campbell P, Green MA: Effective light trapping in polycrystalline silicon thinfilm solar cells by means of rear localized surface plasmons. Appl Phys Lett 2010, 96: 261109. 10.1063/1.3460288View ArticleGoogle Scholar
 Yan W, Stokes N, Jia BH, Gu M: Enhanced light trapping in the silicon substrate with plasmonic Ag nanocones. Opt Lett 2013, 38: 395–397. 10.1364/OL.38.000395View ArticleGoogle Scholar
 Yu YC, Liu JM, Jin CJ, Wang XH: Plasmonmediated resonance energy transfer by metallic nanorods. Nanoscale Res Lett 2013, 8: 209. 10.1186/1556276X8209View ArticleGoogle Scholar
 Chen X, Jia BH, Saha JK, Stokes NS, Qiao Q, Wang QQ, Shi ZR, Gu M: Strong broadband scattering of anisotropic plasmonic nanoparticles synthesized by controllable growth: effects of lumpy morphology. Opt Mat Expr 2013, 3: 27–34. 10.1364/OME.3.000027View ArticleGoogle Scholar
 Gielis J: A generic geometric transformation that unifies a wide range of natural and abstract shapes. Am J Bot 2003, 90: 333–338. 10.3732/ajb.90.3.333View ArticleGoogle Scholar
 Storn R, Price K: Differential evolution – a simple and efficient heuristic for global optimization over continuous spaces. J Glob Optim 1997, 11: 341–359. 10.1023/A:1008202821328View ArticleGoogle Scholar
 Overvelde JTB, Shan S, Bertoldi K: Compaction through buckling in 2D periodic, soft and porous structures: effect of pore shape. Adv Mater 2012, 24: 2337–2242. 10.1002/adma.201104395View ArticleGoogle Scholar
 Macías D, Adam PM, RuízCortés V, RodríguezOliveros R, SánchezGil JA: Heuristic optimization for the design of plasmonic nanowires with specific resonant and scattering properties. Opt Express 2012, 20: 13146–13163. 10.1364/OE.20.013146View ArticleGoogle Scholar
 Zhou SW, Huang X, Li Q, Xie YM: Optimizing twolevel hierarchical particles for thinfilm solar cells. Opt Express 2013, 21: A285A294. 10.1364/OE.21.00A285View ArticleGoogle Scholar
 Yee K: Numerical solution of initial boundary value problems involving Maxwell's equations in isotropic media. IEEE Trans Antennas Propag 1966, 14: 302–307.View ArticleGoogle Scholar
 http://www.lumerical.com
 New grayscale technique opens a third dimension for nanoscale lithography. [http://www.nist.gov/cnst/3d_ebeam_photolithography.cfm]
 Zhou SW, Li W, Sun GY, Li Q: A levelset procedure for the design of electromagnetic metamaterials. Opt Express 2010, 18: 6693–6702. 10.1364/OE.18.006693View ArticleGoogle Scholar
 GarcíaEtxarri A, Apell P, Käll M, Aizpurua J: A combination of concave/convex surfaces for fieldenhancement optimization: the indented nanocone. Opt Express 2012, 20: 25201–25212. 10.1364/OE.20.025201View ArticleGoogle Scholar
 Donelli M: Design of broadband metal nanosphere antenna arrays with a hybrid evolutionary algorithm. Opt Lett 2013, 38: 401–403. 10.1364/OL.38.000401View ArticleGoogle Scholar
Copyright
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.