Solar Energy Materials & Solar Cells ] (]]]]) ]]]–]]]

Contents lists available at ScienceDirect

Solar Energy Materials & Solar Cells

0927-02

doi:10.1

n Corr

Lyon (IN

E-m1 Pr

Pleas

journal homepage: www.elsevier.com/locate/solmat

Absorbing photonic crystals for silicon thin-film solar cells: Design, fabricationand experimental investigation

Xianqin Meng a,b,c, Guillaume Gomard a,b,c, Ounsi El Daif a,b,c,1, Emmanuel Drouard a,b,Regis Orobtchouk a,c, Anne Kaminski a,c, Alain Fave a,c, Mustapha Lemiti a,c, Alexei Abramov d,Pere Roca i Cabarrocas d, Christian Seassal a,b,n

a Universite de Lyon, Institut des Nanotechnologies de Lyon (INL), UMR 5270, CNRS-INSA-ECL-UCBL, Franceb Ecole Centrale de Lyon, 36 Avenue Guy de Collongue, 69134 Ecully Cedex, Francec INSA de Lyon, Bat. Blaise Pascal, 7 Avenue Capelle, 69621 Villeurbanne, Franced LPICM, CNRS UMR 7647, Ecole Polytechnique, 91128 Palaiseau, France

a r t i c l e i n f o

Keywords:

Photonic crystals

Thin-film devices and applications

Photovoltaics

Holographic lithography

48/$ - see front matter & 2010 Elsevier B.V. A

016/j.solmat.2010.11.020

esponding author at: Universite de Lyon, Inst

L), UMR 5270, CNRS-INSA-ECL-UCBL, France

ail address: Christian.Seassal@ec-lyon.fr (C. Se

esent address: IMEC vzw, Kapeldreef 75, B-3

e cite this article as: X. Meng, et al.,

a b s t r a c t

In this paper, we present the integration of an absorbing photonic crystal within a thin-film photovoltaic

solar cell. Optical simulations performed on a complete solar cell revealed that patterning the

hydrogenated amorphous silicon active layer as a 2D photonic crystal membrane enabled to increase

its integrated absorption by 28 % between 300 and 720 nm, comparing to a similar but unpatterned stack.

In order to fabricate such promising cells, we developed a high throughput process based on holographic

lithography and reactive ion etching. The influences of the parameters taking part in those processes on

the obtained patterns are discussed. Optical measurements performed on the resulting ‘‘photonized’’

solar cell structures underline the regularity of the 2D pattern and a significant absorption increase above

550 nm, similarly to what is observed on the simulated absorption spectra. Moreover, our patterned cells

are found to be robust with regards to the angle of incidence of the light.

& 2010 Elsevier B.V. All rights reserved.

1. Introduction

Second generation solar cells, based on thin deposited films, have avery strong potential to achieve a low production cost. Silicon basedsecond generation solar cells have been developed using hydroge-nated amorphous silicon (a-Si:H) [1], microcrystalline silicon (mc-Si)[2] and crystalline silicon (c-Si) [3]. a-Si:H is an attractive materialbecause it exhibits a very high absorption coefficient [2]. However,considering both fabrication costs and bulk recombination issues, thethickness of this layer should be limited as much as possible.Increasing its absorption through light trapping techniques is thenessential to obtain reasonable conversion efficiency on such devices.More precisely, to enable efficient current generation, light trappingschemes are necessary to capture the red and near-infrared portion ofthe solar spectrum [4,5]. Considering the maturity of nanophotonics,various kinds of advanced light trapping techniques may be con-sidered [6–8]. Absorption enhancement using surface plasmons hasbeen proposed in recent years, with self-assembled nanoparticles [9]or regular arrays of nanostructures fabricated using top-down

ll rights reserved.

itut des Nanotechnologies de

.

assal).

001 Leuven, Belgium.

Sol. Energy Mater. Sol. Cell

technological processes [10]. In these methods, the absorptionenhancement around the plasmon resonances is balanced by lightabsorption in the metallic nanostructures. Photonic crystals (PCs)have also been considered to create either back reflectors [11] orselective filters for tandem solar cells [12]. We recently proposed topattern the absorbing layer itself as a PC membrane. This approachconstitutes an efficient light trapping technique and can lead to asubstantial increase in the solar light absorption over a wide opticalspectral range [7,13–16]. Out of this concept, and considering a-Si:Hbased structures, we proposed to control light capture and absorptionusing planar PC Bloch modes that stand over the light-line. Thisnanopattern allows ultra-thin a-Si:H absorbing layers (with thick-nesses of around 100 nm), i.e. lower than the diffusion length of theminority carriers. The absorption increase related to light trapping isthen expected to lead to short-circuit current densities exceedingthose of similar unpatterned cells.

In order to generate such PC patterns while meeting industrialrequirements, it is essential to develop a high throughput processcompatible with large surfaces. Relevant techniques then includenano-imprint [17] and holographic lithography. Additionally,holographic lithography is a maskless process, particularly wellsuited to the fabrication of periodic patterns like PCs [18–21], butits parameters have to be carefully adjusted and controlledaccording to the desired stack and pattern.

s (2010), doi:10.1016/j.solmat.2010.11.020

X. Meng et al. / Solar Energy Materials & Solar Cells ] (]]]]) ]]]–]]]2

In this paper, we report on a way of patterning the active layer as a2D PC membrane using holographic lithography and reactive ionetching (RIE). The global design of our ‘‘photonized’’ solar cell, whichhas been optimized optically by simulation, is introduced in Section 2.Its optical properties as well as the parameters of the PC membrane,which maximize the absorption in the active layer, are also discussedin this section, which defines the objectives to attain for furtherexperimental development. This last point is developed in Section 3.After presenting the general fabrication process of our patterned cells,we study the influence of the main holographic lithography para-meters, in order to generate the targeted pattern when associatedwith a RIE step. Finally, in Section 4, the optical properties of thefabricated cells are investigated both experimentally and theoreti-cally, and compared to those of an unpatterned cell.

2. Optical design of ‘‘photonized’’ solar cells

The proposed thin-film solar cells are all made of the same stack.As shown in Fig. 1, it consists, from back to front, of a silver (Ag)layer that acts as a reflector and supports the rear metallic contacts,a zinc oxide (ZnO) layer playing the role of an optical spacer, abarrier against the diffusion of silver and a transparent andconductive oxide (TCO), an a-Si:H active layer (p–i–n junction)and finally a front ITO layer, which behaves simultaneously as aTCO and as an antireflective layer. The absorbing PC membranestructures are patterned in the a-Si:H layers; for technologicalreasons, the front ITO layer is also patterned. In this configuration,the photogenerated carriers created within the fully etched a-Si:Hlayer can reach efficiently the two surrounding TCO layers since thethickness of the active layer is lower than the diffusion length of theminority carriers. After their transfer into these TCO layers, thecarriers are then collected by metallic contacts, which consist, forthe top of the cell, of a metallic grid deposited on the ITO layer.

Here, we will consider a 2D PC membrane structure with asquare symmetry (see Fig. 1) as a basis of our ‘‘photonized’’ solarcells. Such a 2D pattern structure is expected to exhibit a higher

Fig. 1. Schematic view of a ‘‘photonized’’ solar cell with 2D pattern. Photogenerated

carriers originating from a-Si:H active layer are collected through two surrounding

TCO layers, before being transported to metallic contacts (not shown) disposed on

Ag layer (for rear contacts) and on ITO layer, beside patterned areas (for top

contacts).

Please cite this article as: X. Meng, et al., Sol. Energy Mater. Sol. Cell

absorption than its unpatterned equivalent, and its behaviour isindependent of the polarization of the incident light [16].

It is then possible to adjust the PC membrane parameters, i.e. itsperiod L and its D/L ratio (where D is the diameter of the holes), so asto maximize the integrated absorption inside the active layer. Thistheoretical study has been conducted using CAMFR [22], and morespecifically its package GARCLED, based on the rigorous coupledwave analysis (RCWA) method [23]. The AM1.5G solar spectralintensity distribution was taken into account over the 300–720 nmspectral range, whose bounds roughly and respectively correspondto the lower limit of the solar spectrum and the a-Si:H gap. Thethicknesses of the different layers were chosen to ensure efficientcollection of the carriers (ITO, a-Si:H and ZnO layers), long-termstability of the cell (ZnO layer) and high reflectivity on the back side(Ag layer). Then, we considered values of L and D/L achievable usingthe selected technological processes. More precisely, the config-urations of the patterned cell were investigated by scanningsimultaneously L between 0.3 and 0.7 mm and D/L between 0.3and 0.7. The range of the D/L ratio has been imposed by thelimitations faced when using holographic lithography.

The results of these simulations, which are displayed in Fig. 2,indicate a maximal value for the integrated absorption of about 82%over the whole patterned cell for L¼0.38 mm and D/L¼62.5%.

Looking for the useful part of this integrated absorption, namelythe one in the sole active layer, we found that 65.7% of the incominglight could be absorbed in the a-Si:H layer for the best configurationobtained. This value represents a 28% relative increase compared tothe case of a similar but unpatterned cell. Provided an efficientpassivation of the processed sidewalls of the holes is created, asubsequent increase in the conversion efficiency is expected for our‘‘photonized’’ solar cells. It should also be noted that the integratedabsorption of the patterned cell is robust with regards to the angle ofincidence of light (a) since at a¼501, only a 5% decrease in theabsolute value is expected compared to its value under normalincidence. Eventually, it can be appreciated in Fig. 2 that highlyintegrated absorptions can be obtained for a large range of L and D/Lvalues around the best configuration, which is an important pointdue to the inherent technological uncertainties. Thus, in practice,and considering the limitations of our holographic lithographysetup, periods between 520 and 620 nm and D/L ratio between50% and 60% have been considered for the experimental part of this

Fig. 2. Integrated absorption in the whole patterned cell in function of the period L

(x-axis, in mm) and D/L ratio (y-axis) of PC membrane. The best configuration

(L¼0.38 mm and D/L¼62.5%) is reported in the figure, as well as the one that stands

within parameters range reachable by experiment (inside the square in dotted

lines).

s (2010), doi:10.1016/j.solmat.2010.11.020

X. Meng et al. / Solar Energy Materials & Solar Cells ] (]]]]) ]]]–]]] 3

study, as depicted in Fig. 2. Within this range of parameters, theexpected gains vary from 22.4% for L¼0.52 mm and D/L¼60%(integrated absorption over the whole stack¼75.8% and integratedabsorption in the sole a-Si:H layer¼63.3%) to 23% for L¼0.6 mm andD/L¼57.5% (integrated absorption over the whole stack¼78.4% andintegrated absorption in the sole a-Si:H layer¼63.7%), when theabsorption in the active layer is considered and compared to the onein a similar but unpatterned cell.

3. Fabrication of PC patterns using holographic lithographyand reactive ion etching

In this section, we discuss the steps leading to the patterning ofour cells, using holographic lithography and RIE techniques. Thestacks, based on the Ag(50 nm)/ZnO(100 nm)/a-Si:H(130 nm)/ITO(60 nm)/SiO2(100 nm) layers, which have been processed forthis study, are based on the design previously presented eventhough the actual thicknesses of these layers may be slightlydifferent due to technological uncertainties. The layers weredeposited on a glass substrate using PECVD in a multi-plasmadeposition system. Moreover, a 100 nm thick silica layer was addedon the top of the stack and acts as a hard mask during the etching ofthe ITO and a-Si:H layers. Starting from such samples, it is possibleto first pattern the hard mask as a 2D PC with square symmetry viaholographic lithography, and then to transfer this pattern to thetwo underlying layers with RIE, as schematized in Fig. 3.

A more detailed description of these techniques will be developedin the following sections. In particular, we will see how the differentparameters involved in this process can be tuned in order to reach thetargeted parameters of the 2D PC (its period L and its D/L ratio) and totransfer the obtained pattern in the different layers, while keeping theroughness of the etched sidewalls as low as possible.

3.1. Generation of PC patterns by holographic lithography

This technique consists of irradiating a photosensitive resistusing an interference pattern. A simple way to create such a patternis to use the system described in Fig. 4a. The sample, covered with aUV sensitive resist, is placed on a sample holder associated to amirror placed at 901 of the sample. This system is illuminated by alaser beam; this generates fringes that irradiate the resist, with

Fig. 3. Main steps of process enabling to pattern active layer as a 2D PC through holo

(b) UV exposure of the photoresist, (c) Development of the photoresist, (d) RIE etching o

etching of the active layer (a-Si:H layer)+photoresist removal.

Please cite this article as: X. Meng, et al., Sol. Energy Mater. Sol. Cell

inter-fringe distance L directly related to angley through the relationL¼llaser/(2siny). In order to obtain a 2D pattern with squaresymmetry at the sample’s surface, the irradiation needs to berepeated after tilting the sample by 901. The resist is then exposedas schematically shown in Fig. 4b. We implemented this genericprocess using a laser source at 266 nm, with a power around5.4 mW, and a NEB 22 negative tone Chemically amplified resist(CAR). While the selected wavelength enables the patterning ofstructures with lattice parameters around 500 nm, using a negativeresist is well suited to the generation of a pattern of holes in theresist, as in the case of our design. At the same time, high sensitivityand resolution may be achieved using such a CAR resist [24]. Withthis system, the processing sequence associated with holographiclithography includes sample cleaning and preparation, hexamethyl-disilizane (HMDS) adhesion promoter spinning, resist spinning,post-apply bake (PAB), UV exposure, post-exposure bake (PEB),resist development and a final descum etch step. The latter consistsof removing through oxygen plasma the undesired resist coveringthe bottom of the holes, namely the SiO2 hard mask, and which wasnot completely removed during the development step [25].

While these steps are well-established for standard siliconsamples, using UV lithography at 248 nm [26], the process para-meters should be adapted to perform holographic lithography overstacked layers deposited on glass, as is the case for our design.Indeed, heat transfer and standing wave effects [27] are expected tobe different in the case of our structures, where using a periodicinterference pattern may induce specific constraints. Our metho-dology therefore consisted of scanning process temperatures andtimes, starting from parameters suited to classical UV lithography.The NEB resist is spun at 3000 rpm for 30 s, followed by a PAB at atemperature of 110 1C for 120 s. After exposure and PEB (at 92 1C),the development is performed in a MF702 solution (supplied byShipley Co) for about 10 s, followed by a descum etch step in anoxygen plasma. The influence of key parameters including the PAB,exposure, PEB and descum etch times is then investigated anddiscussed in the remainder of this section. Such experiments wereperformed on the stack described at the beginning of Section 3.

The first set of experiments was carried out with double UVexposure performed with an exposure time ranging between 50and 130 s. The PAB parameters were fixed at 110 1C and 120 s,while PEB was performed at 92 1C for 240 s, with a 10 s develop-ment time and a 8 s O2 plasma.

graphic lithography (a–c) and RIE (d–f). (a) Spin coating of a negative photoresist,

f the hard mask (SiO2 layer), (e) RIE etching of the front TCO (ITO layer) and (f) RIE

s (2010), doi:10.1016/j.solmat.2010.11.020

Fig. 4. (a) Out of the spatial filter, beams either impinge directly on sample’s surface or are reflected by mirror, leading to interference fringes in the photoresist. (b) After 901 tilt

of sample, second exposure enables to create a square array of unexposed areas.

Fig. 5. Influence of the exposure time on the diameter of holes.

X. Meng et al. / Solar Energy Materials & Solar Cells ] (]]]]) ]]]–]]]4

As shown in the SEM images from Fig. 5, the holes appearcircular, with a strong impact of the exposure time on theirdiameter. As expected in the case of a negative tone resist, thediameter decreases when the exposure time increases. Thisdependence can be used to control the D/L ratio of the patterns,which stands from 0.35 to 0.65 in the whole exposure time range,and for a lattice parameter between 520 and 620 nm. Therefore, thetargeted D/L ratios (50–60%), as calculated in Section 2, stand wellwithin the achievable range. However, parasitic effects such as thepresence of an inner resist ring for an exposure time of 90 s or asecondary resist pattern for low exposure times clearly appear. Theformer may reduce the D/L ratio and therefore, the inner resistshould be eliminated before transferring the pattern to the under-lying layers. This is the role of the descum etch step, which will bediscussed at the end of this section. The latter, which is due toclassical optical standing wave effects [27], does not impact thefinal shape of the holes.

PEB is an essential step in CAR resists, as it is used to catalyselinking reactions, to modify the solubility of the polymer matrixand to homogenize the photo-acid concentration as generatedduring the exposure step. In this sense, it is expected to have a

Please cite this article as: X. Meng, et al., Sol. Energy Mater. Sol. Cell

strong impact on the pattern shape and on the reduction ofstanding wave effects. In the second set of experiments, PEB timewas tuned in a 120–240 s range by fixing its temperature at 92 1C,with a 120 s PAB at 110 1C, a 50 s exposure, a 10 s development andan 8 s descum etch step. Corresponding SEM images are shown inFig. 6.

Although a 120 s PEB time is generally sufficient in the case ofsilicon samples, it is necessary to perform a longer baking in thecase of our samples. This may be accounted for by the lowerthermal conductivity of our glass wafers. PEB time, which should beover 180 s, also appears as a second way to control the size of theholes for a given lattice parameter.

As mentioned above, the photoresist may remain within thepatterned holes. As this affects the size and shape of the final 2D PCstructure, after the ITO and a-Si:H layers have been etched, thisresist material should be removed by a final descum etch step. Thisconsists of a reactive ion etching step, with a 20 sccm O2 flow, at100 W under a pressure of 100 mTorr. The influence of this 8 sdescum etch step is illustrated in Fig. 7 in the case of a 120 s PAB, a50 s exposure, a 210 s PEB, and a 10 s development time. As a resultof this final step, the inner resist ring is removed, and the pattern

s (2010), doi:10.1016/j.solmat.2010.11.020

Fig. 6. SEM images of patterned photoresist for PEB times from 120 to 240 s (with a 30 s step) at 92 1C.

Fig. 7. 2D pattern without and with 8 s O2 plasma.

Fig. 8. SEM profile images of etched SiO2 hard mask, ITO and a-Si:H layers:

(a) general view of the surface and (b) close-up of the holes.

X. Meng et al. / Solar Energy Materials & Solar Cells ] (]]]]) ]]]–]]] 5

roughness tends to decrease. Furthermore, the D/L ratio is increasedfrom 0.89 to 0.93 after the descum step. This geometrical changeshould be anticipated when PEB and exposure time are selected,with a view to reaching the targeted D and L values.

As a conclusion from this investigation, process parameters canbe adjusted in order to achieve the targeted 2D pattern. While thePAB time may be fixed at 120 s, the exposure and PEB times may bejointly adjusted in order to reach the targeted value of the D/L ratio.An appropriate set of parameters consists of a 60 s exposure with aPEB time of 180 s. Finally, to take full profit of the holographiclithography patterning, an 8 s final descum etch step should beperformed.

3.2. Transfer of the 2D pattern by RIE

The photonic crystal patterned by holographic lithography isfirst transferred to a 100 nm thick SiO2 hard mask, and then into theunderlying ITO and a-Si:H layers, using RIE processes. The depthetched in the different layers is controlled thanks to the etchingrates established previously on similar structures with the samematerials. Transfer to the SiO2 hard mask is achieved using a CHF3

plasma (with a 16 sccm flow at 15 mTorr with a power of 100 W for900 s). After elimination of the resist by the O2 plasma, the patternsare transferred to the ITO layer, using a CH4:H2 plasma, with flowsof 10 and 30 sccm, at a 200 W power, for 130 s. This step isperformed at low pressure (30 mTorr) in order to achieve highsidewall verticality. Finally, the a-Si:H layer was etched with aSF6:Ar plasma (5 sccm for each gas, at 50 W for 180 s); a reducedpressure of 15 mTorr was used for this step. Fig. 8 shows that holes

Please cite this article as: X. Meng, et al., Sol. Energy Mater. Sol. Cell

with reasonable regularity and roughness could be transferred;the achieved period and D/L ratio are, respectively, 585 nm, in thetargeted range, and 0.47, below the range corresponding to theoptimum configuration (50–60%).

4. Optical characterization of ‘‘photonized’’ solar cells

In order to determine the optical properties of the patternedsolar cell presented in the previous section, absorption measure-ments were performed using an integrating sphere. In theseexperiments, the sample is illuminated by an unpolarized light,so as to perform reflectance (R) and transmittance (T) measure-ments, with a �1 mm2 spatial resolution and an 81 angle ofincidence. The absorption A(l)¼1�R(l)�T(l) is then simplyderived. The absorption of the whole patterned stack was measuredbetween 300 and 720 nm with a 5 nm step and an experimentaluncertainty of about 5%. The absorption spectrum of the unpat-terned stack was also measured as a reference. The correspondingspectra are reported in Fig. 9a. Absorption spectra simulated byRCWA are displayed in Fig. 9b for comparison. The same stacks asthe ones used for the fabrication of the cells have been consideredfor this purpose, but the top SiO2 hard mask was not taken intoaccount for the simulations. Due to uncertainty in the determina-tion of the geometrical parameters, in particular the ITO layerthickness but also the inhomogeneity of the D/L ratio, the calculated

s (2010), doi:10.1016/j.solmat.2010.11.020

Fig. 9. Comparison between measured (a) and simulated (b) spectra of patterned and unpatterned cells (angle of incidence fixed at 81).

Fig. 10. Absorption spectra measured with linearly polarized light, for two

orientations of the sample, at 81 angle of incidence, and for 161 angle of incidence.

X. Meng et al. / Solar Energy Materials & Solar Cells ] (]]]]) ]]]–]]]6

and measured absorption spectra exhibit different features. How-ever, strong similarities are noticeable and the same general trendscan be deduced.

All spectra exhibit high absorption for lo550 nm, thoughslightly more important in the case of the patterned cell. Indeed,the high values taken by the extinction coefficient of a-Si:H (ka-Si:H)are sufficient to ensure the absorption of the incoming light in onepass within the active layer. Besides, patterning the cell structuredecreases the mean refractive index in the front of the cell, leadingto an antireflective effect; absorption is then higher in the case ofthe patterned cell. The situation differs above 550 nm; a sharpdecrease in absorption is observed for the unpatterned cell due tothe low values of ka-Si:H in this spectral range, whereas the 2D PCpatterning of the active layer leads to additional absorption peaks,and finally to a higher absorption than in the case of the reference.These absorption peaks are attributed to resonant Bloch modes ofthe absorbing PC membrane. The maximum absorption is reachedat the critical coupling conditions, i.e. when the photon lifetime atresonance is matched with the lifetime corresponding to absorp-tion losses in the non-patterned absorbing medium [13]. Moreover,due to the multimode character of this PC membrane structure andtherefore the number of resonances, the absorption integrated inthe whole range between 550 and 720 nm is significantly increasedin the case of ‘‘photonized’’ solar cells.

In addition, the regularity of the 2D pattern transferred to thedifferent layers was investigated by shifting the sample by 901,while keeping all the other parameters fixed. As can be seen inFig. 10, the spectra corresponding to the two configurations (nonshifted or shifted by 901, with an angle of incidence of 81) andobtained using linearly polarized light nearly superimposethroughout the considered wavelength range. This tends to indi-cate that the PC membrane holes have almost an isotropic shapeand are arranged according to the same period in both directions.

Lastly, the influence of the angle of incidence was tested bydoubling its value, from 81 to 161. When increased, the angle ofincidence produces a blue-shift of certain absorption peaks, forinstance around 690 nm in Fig. 10, but it globally leaves theintegrated absorption of the structure unchanged. This can beaccounted for by the multimode nature of the PC structure, since aweak modification in the resonant mode properties will not bevisible in the wide spectral range we consider.

Please cite this article as: X. Meng, et al., Sol. Energy Mater. Sol. Cell

5. Conclusion

Thin-film photovoltaic cell structures integrating a 2D PCmembrane have been designed, fabricated and characterized. Wespecifically focused on the development of a process based onholographic lithography and RIE, enabling the generation of a 2D PCmembrane on a wide surface with a high throughput. By appro-priate tuning of the exposure and PEB times of the lithographicprocess, the desired configurations derived from optical simula-tions could be reached. In order to limit the roughness of thesidewalls and achieve better control of the geometrical parameters,an additional descum etch step was necessary. Absorption spectrawere determined experimentally by optical measurements usingan integrating sphere. Measurements were performed on thepatterned and unpatterned reference cell structures, and comparedto spectra calculated by the RCWA method. Despite some dis-crepancies due to inaccuracies in the geometrical parameters, themeasured and calculated absorption spectra exhibit very similarbehaviour. The absorption, integrated over the whole 300–720 nm

s (2010), doi:10.1016/j.solmat.2010.11.020

X. Meng et al. / Solar Energy Materials & Solar Cells ] (]]]]) ]]]–]]] 7

range, increases by 28% in the case of a solar cell stack patterned as aPC membrane. This increase is mainly attributed to the positive roleof Bloch mode resonances of the PC membrane above 550 nm.Besides, these ‘‘photonized’’ cells, whose 2D pattern regularity wasconfirmed experimentally, present good tolerance towards theangle of incidence of light. The next steps will consist of adding themetallic contacts by optical lithography and vacuum metal deposi-tion to measure the conversion efficiency of these new solar cells.

Acknowledgements

This work was supported by the ANR SPARCS project. PCpatterning was performed at the NanoLyon Technology Platform.X. Meng acknowledges the China Scholarship Council (CSC).

References

[1] D.L. Staebler, C.R. Wronski, Reversible conductivity changes in discharge-produced amorphous Si, Appl. Phys. Lett. 31 (1977) 292–294.

[2] R.E.I. Schropp, M. Zeman, Amorphous and Microcrystalline Silicon Solar Cells:Modeling, Materials, and Device Technology, Kluwer Academic Publishers,Norwell, MA, 1998.

[3] B. Jagannathan, W.A. Anderson, J. Colman, Amorphous silicon/p-type crystal-line silicon heterojunction solar cells, Sol. Energy Mater. Sol. Cells 46 (1997)289–310.

[4] R.H. Franken, R.L. Stolk, H. Li, C.H.M. van der Werf, J.K. Rath, R.E.I. Schropp,Understanding light trapping by light scattering textured back electrodes inthin film n–i–p-type silicon solar cells, J. Appl. Phys. 102 (1) (2007)014503-1–014503-7.

[5] J. Muller, B. Rech, J. Springer, M. Vanecek, TCO and light trapping in silicon thinfilm solar cells, Sol. Energy 77 (6) (2004) 917–930.

[6] M. Peters, J.C. Goldschmidt, T. Kirchartz, B. Blasi, The photonic lighttrap—improved light trapping in solar cells by angularly selective filters,Sol. Energy Mater. Sol. Cells 93 (2009) 1721–1727.

[7] A. Goetzberger, J.C. Goldschmidt, M. Peters, P. Loper, Light trapping, a newapproach to spectrum splitting, Sol. Energy Mater. Sol. Cells 92 (12) (2008)1570–1578.

[8] O.E. Daif, E. Drouard, G. Gomard, A. Kaminski, A. Fave, M. Lemiti, S. Ahn, S. Kim,P. Roca i Cabarrocas, H. Jeon, C. Seassal, Absorbing one-dimensional planarphotonic crystal for amorphous silicon solar cell, Opt. Express 18 (S3) (2010)A293–A299.

[9] F.J. Beck, A. Polman, K.R. Catchpole, Tunable light trapping for solar cells usinglocalized surface plasmons, J. Appl. Phys. 105 (11) (2009) 114310-1–114310-7.

Please cite this article as: X. Meng, et al., Sol. Energy Mater. Sol. Cell

[10] R.A. Pala, J. White, E. Barnard, J. Liu, M.L. Brongersma, Design of plasmonic thin-film solar cells with broadband absorption enhancements, Adv. Mater. 21(2009) 3504–3509.

[11] L. Zeng, P. Bermel, Y. Yi, B.A. Alamariu, K.A. Broderick, J. Liu, C. Hong, X. Duan,J. Joannopoulos, L.C. Kimerlingv, Demonstration of enhanced absorption in thinfilm Si solar cells with textured photonic crystal back reflector, Appl. Phys. Lett.93 (2008) 221105-1–221105-3.

[12] A. Bielawny, C. Rockstuhl, F. Lederer, R.B. Wehrspohn, Intermediate reflectorsfor enhanced top cell performance in photovoltaic thin-film tandem cells, Opt.Express 17 (10) (2009) 8439–8446.

[13] D. Duche, L. Escoubas, J.-J. Simon, P. Torchio, W. Vervisch, F. Flory, Slow Blochmodes for enhancing the absorption of light in thin films for photovoltaic cells,Appl. Phys. Lett. 92 (2008) 193310-1–193310-3.

[14] Y. Park, E. Drouard, O.El. Daif, X. Letartre, P. Viktorovitch, A. Fave, M. Lemiti,C. Seassal, Absorption enhancement using photonic crystals for silicon thinfilm solar cells, Opt. Expr. 17 (2009) 14312–14321.

[15] O. El Daif, E. Drouard, G. Gomard, A. Fave, A. Kaminski, M. Lemiti, Y. Park,X. Letartre, P. Viktorovitch, S. Ahn, H. Jeon, C. Seassal, Photonic crystalenhanced absorption of amorphous silicon for solar cells, in: Proceedings ofthe 24th European Photovoltaic Solar Energy Conference, 2009, pp. 548–552.

[16] G. Gomard, O. El Daif, E. Drouard, X. Meng, A. Kaminski, A. Fave, M. Lemiti,E. Garcia-Caurel, P. Roca i Cabarrocas, C. Seassal, Design and fabrication ofphotonic crystal thin film photovoltaic cells, Proc. SPIE 7725 (2010)77250M-1–77250M-6.

[17] J.Y. Chen, K.W. Sun, Enhancement of the light conversion efficiency of siliconsolar cells by using nanoimprint anti-reflection layer, Sol. Energy Mater. Sol.Cells 94 (2010) 629–633.

[18] M. Campbell, D.N. Sharp, M.T. Harrison, R.G. Denning, A.J. Turberfield, Fabrica-tion of photonic crystals for the visible spectrum by holographic lithography,Nature 404 (2000) 53–56.

[19] C.-O. Cho, Y.-G. Roh, Y. Park, H. Jeon, B.-S. Lee, H.-W. Kim, Y.-H. Choe, Photoniccrystal slab waveguides fabricated by the combination of holography andphotolithography, Jpn. J. Appl. Phys. 43 (2004) 1384–1387.

[20] D.-H. Kim, C.-O. Cho, Y.-G. Roh, H. Jeon, Y.S. Park, J. Cho, J.S. Im, C. Sone, Y. Park,W.J. Choi, Q.-H. Park, Enhanced light extraction from GaN-based light-emittingdiodes with holographically generated two-dimensional photonic crystalpatterns, Appl. Phys. Lett. 87 (2005) 203508-1–203508-3.

[21] W. Bogaerts, P. Dumon, D. Taillaert, V. Wiaux, S. Beckx, B. Luyssaert, J. VanCampenhout, D. Van Thourhout, R. Baets, SOI nanophotonic waveguidestructures fabricated with deep UV lithography, Photonics Nanostruct. 2(2004) 81–86.

[22] Developed by Ghent University, see /http://camfr.sourceforge.net/S, 2007.[23] M.G. Moharam, T.K. Gaylord, Rigorous coupled-wave analysis of planar-grating

diffraction, J. Opt. Soc. Am. 71 (1981) 811–818.[24] H. Ito, C.G. Willson, Chemical amplification in the design of dry developing

resist materials, Polym. Eng. Sci. 23 (1983) 1012–1018.[25] Chris A. Mack, Field Guide to Optical Lithography, SPIE Press Book, 2006.[26] S. Pauliac, S. Landis, J. Foucher, J. Thiault, O. Faynot, Hybrid lithography process

for nano-scale devices, Microelectron. Eng. 83 (2006) 1761–1766.[27] Andreas Grassmann, et al., Contrast transfer function measurements of deep

ultraviolet steppers, J. Vac. Sci. Technol. 10 (6) (1992) 3008–3011.

s (2010), doi:10.1016/j.solmat.2010.11.020