Silver nanowire networks: Physical properties and potentialintegration in solar cells

D.P. Langley a,b,n, G. Giusti a, M. Lagrange a, R. Collins a, C. Jimnez a, Y. Brchet c, D. Bellet a

a Laboratoire des Matriaux et du Gnie Physique, CNRSGrenoble INP, 3 Parvis Louis Nel, 38016 Grenoble, Franceb Laboratoire de Physique des Solides, Interfaces et Nanostructures, Dpartement de Physique, Universit de Lige, Alle du 6 Aot 17, 4000 Lige, Belgiumc Laboratoire de Science et Ingnierie des Matriaux et des Procds, CNRSGrenoble INP, 1130 rue de la piscine, 38042 Saint-Martin d'Hres, France

a r t i c l e i n f o

Available online 8 October 2013

Keywords:SilverNanowiresTransparent conductive electrodesPhotovoltaicsPercolation

a b s t r a c t

With the growing interest in flexible electronics and the increased utilization of Indium Tin Oxideelectrodes for display and photovoltaic applications the need for new materials is emerging.

In this work we present the electro-optical properties of Ag nanowire networks as an alternativetransparent conductive material. A comparison of different film deposition techniques is made andindicates that the properties of the network are independent of the fabrication method. Analysis of theelectrical behavior as a function of nanowire density is made and compared with theoretical results aswell as Monte Carlo simulations.

Thermal annealing is shown to reduce the sheet resistance from 1000 /sq to 8 /sq; this reduction isachieved by local sintering of the nanowire junctions.

Experimental optimization of Ag nanowire electrodes was undertaken and a peak in the electro-optical properties is observed at approximately 100 mg/m. Finally a discussion of the potentialintegration of Ag nanowire networks into solar cells is undertaken; we observe that these electrodesshow promise as an emerging transparent conductive material, especially for flexible applications.

& 2013 Elsevier B.V. All rights reserved.

1. Introduction

Thin films which exhibit at the same time high electricalconductivity and optical transparency are crucial for many modernelectronic devices such as e-papers, organic light-emitting diodes(OLEDs), liquid-crystal displays (LCDs), and solar cells. Manyarticles in the literature currently highlight the growing need toidentify and develop new methods and materials for fabrication oftransparent conductive materials. There are many reasons for thisneed and they have been discussed in depth by Kumar and Zhou[1] and Ellmer [2].

In the case of solar cells, the transparent electrode usuallyworks as the anode for extracting separated charge carriers fromthe absorbing area. While transparent conductive oxides (TCOs)are usually well adapted for solar cells, they suffer from significantlimitations such as costly fabrication process, scarcity (especiallyconcerning Indium based TCOs like Indium Tin Oxide (ITO)) andbrittleness. Hence a variety of other materials have been inten-sively investigated recently.

Kumar and Zhou highlight three main emerging materials thatmay provide a useful replacement for transparent conductiveoxides (TCOs): graphene, carbon nanotubes and metallic nanowirenetworks. Ag NWs already exhibit very good physical properties,but still some issues inhibit the large-scale application of Ag NWelectrode as for instance the need of a heating step or the lowadhesion of the network onto the substrate. Clearly a betterunderstanding of fundamental properties of Ag NW networks isneeded as well through investigations of the effects of severalparameters such as Ag NW morphology, or the influences of theexperimental conditions and post-deposition treatments (thermalannealing, mechanical pressure, embedding, etc.). While theelectro-optical properties are of prime importance, other proper-ties are also crucial: electro-mechanical properties (often investi-gated under bending fatigue), stability (either thermal orchemical), and diffuse component of the transmitted light (hazefactor). This emerging material has been studied only recently anddeserves thorough investigations to better its physical propertiesand facilitate its integration into devices. Alongside all the physicalproperties, cost will be very important and Ag NW networks canexhibit advantages due to the small required quantity of silver andlow-cost deposition techniques.

This article focuses on the physical properties of metallic nano-wire networks, specifically silver, and whether this material willprovide the necessary balance to meet the needs of photovoltaic

Contents lists available at ScienceDirect

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

Solar Energy Materials & Solar Cells

0927-0248/$ - see front matter & 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.solmat.2013.09.015

n Corresponding author at: CNRSGrenoble INP, Laboratoire des Matriaux et duGnie Physique, 3 Parvis Louis Nel, 38016 Grenoble, France. Tel.: 33 456529337.

E-mail address: daniel.dpl3@gmail.com (D.P. Langley).

Solar Energy Materials & Solar Cells 125 (2014) 318324

www.sciencedirect.com/science/journal/09270248www.elsevier.com/locate/solmathttp://dx.doi.org/10.1016/j.solmat.2013.09.015http://dx.doi.org/10.1016/j.solmat.2013.09.015http://dx.doi.org/10.1016/j.solmat.2013.09.015http://crossmark.crossref.org/dialog/?doi=10.1016/j.solmat.2013.09.015&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.solmat.2013.09.015&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.solmat.2013.09.015&domain=pdfmailto:daniel.dpl3@gmail.comhttp://dx.doi.org/10.1016/j.solmat.2013.09.015

applications. In order to obtain a metallic nanowire based transpar-ent electrode which can be efficiently incorporated into a solar cellone can play with several parameters such as the chemical composi-tion and morphology of the metallic nanowires (NWs), the densityof the NW network and the use of post-processing such asmechanical pressing [3,4], chemical treatments [5] and thermalannealing [3,6]. We report here the influence of the density andannealing on the physical properties of Ag NW networks pertinentto their potential integration into a solar cell.

2. Experimental section

Ag NWs dispersed in isopropyl alcohol were acquired fromSeashell Technology [7]. The average dimensions of the nanowireswere 0.105 mm for the diameter and 37.5 mm for the lengthresulting in an aspect ratio of about 360. Films of differentdensities were fabricated from suspensions of different concentra-tions. Random Ag NW networks were generated by differenttechniques, including spin-coating, drop casting, rod coating andspray injection (as described in Ref. [8]) on low alkaline earthboro-aluminosilicate glass (Corning C1737-S111).

Field-emission scanning electron microscopy (FESEM) imagingwas recorded with a FEI Quanta 250 to investigate the networkmorphology.

In-situ thermal annealing under atmosphere combined with realtime electrical resistance acquisition using the two-point probemethod and with a constant voltage of 1 V was performed to gaininsight into the thermal behavior of the Ag NW networks. Silver paintstrips acted as electrical contacts and the total sample size was12.512.5 mm2. Thus, the resistance values reported in this studyrepresent an average over the surface previously mentioned.

The optical transmittance was recorded by using a Lambda 950Perkin-Elmer spectrophotometer. No substrate subtractions wereperformed on any of the film reported in this study. In addition,optical transmission values are quoted at 550 nm. The hazefactor, quantifying the amount of light scattering, was calculatedfrom the total (T(total)) and direct (T(direct)) transmittances as (T(total)T(direct))/T(total).

Four point probe measurements were performed using a Keithley2400 sourcemeter with a Lucas Labs Pro4-440N probe station.

3. Results and discussions

3.1. Electro-optical properties

Of the properties that are desirable for transparent conductivematerials (TCMs) the obvious selection criteria are the transmission of

light and the electrical resistance. Considering this, it is appropriate tomake an initial comparison of Ag NW networks electro-opticalproperties with those of TCOs. All electro-optical properties that arediscussed are measured on Ag NWnetworks after a thermal annealingstep which was found to significantly decrease the resistance withouthaving an impact on their transmittance. The changes of resistance aredue to several factors that will not be discussed in depth here but inmajority they are caused by local sintering at the junctions betweenthe nanowires (as shown in Fig. 1). The local sintering occurs as aresult of atomic migration to reduce the surface energy at points ofhigh curvature. Fig. 1 exhibits scanning electron microscopy observa-tion of two different junctions between two nanowires before andafter annealing for 2 h at 200 1C in air. Although this is not the samejunction before and after annealing these images represent typicalmorphologies. Local sintering at the junction is present after annealing,which leads to a decrease of sheet electrical resistance from1000sq to 8sq.

The reduction of resistance via thermal annealing allows theproduction of highly conductive layers. Modifying the annealingprofile enables networks with RSo20/sq to be created within2 min at 250 1C or if there is a temperature restriction the samecan be achieved with longer annealing at lower temperatures.At 200 1C the network resistance continues to drop for 2 h thoughthe majority of the change occurs in the first 5 min.

The resistance and transmittance of Ag NW networks aredependent on the wire length and diameter as well as the densityof the network [9]. The nanowire diameter plays a key role in thescattering properties of the network as highlighted by Prestonet al. [10]. It is common to plot the transmittance as a function ofsheet resistance [11]. Fig. 2 shows that the general behavior oftransmittance is somewhat independent of deposition method.

A comparison of Ag NW networks, silver flakes, carbon nano-tubes and graphene was also made by De et al. [11]. Their resultsconcern NW networks created by vacuum filtration of a colloidalsolution of Ag NWonto a membrane to form the network, which isthen transferred to a PET substrate. The experimental results of thecurrent contribution are presented in Fig. 2 for Ag NW networksgenerated by the various techniques mentioned in the legend ofthe figure. Comparison of these results with those of De et al. [11]suggests that there is only a slight dependence of electro-opticalproperties of Ag NW networks on the deposition method used.The majority of the behavior is dominated by the geometry of thewires. In strong agreement with the work of De et al. [11], the datafits well in both regimes: the percolation regime for sparsenetworks and the bulk regime for dense networks. As discussedbelow an optimal density has to be considered to reach a tradeoffin order to get high optical transmittance T and low sheetresistance RS. Generally speaking a good quality transparentelectrode for solar applications corresponds roughly to TE90%

Fig. 1. SEM images of Ag NW junctions before (left) and after (right) annealing; the sheet resistance of this network reduced from 1000 sq to 8 sq. Scale bars indicate100 nm.

D.P. Langley et al. / Solar Energy Materials & Solar Cells 125 (2014) 318324 319

and RSE10sq. When considering the substrate contribution, asin Fig. 2, such requirements would lead to TE80%. As depicted inFig. 2, Ag NWs already meet the requirements for being anefficient transparent electrode for solar cells applications.

Fig. 3 shows a comparison of the transmission spectra for bareglass, Ag NW network and fluorine doped tin oxide (FTO), andillustrates the important difference in the transmission spectrumof Ag NWwhen compared to standard TCOs. There are very similartransmission values from 250 to 1250 nm at which point asignificantly higher rate of transmission is observed for the AgNW network in the region 12502500 nm. This stems from theplasmon absorption of TCOs: above a certain wavelength P (whichdepends upon the carrier density) the electromagnetic wave isdamped thanks to the collective excitation of the carrier gas [2].Increasing the carrier density (by doping) in most TCOs results in adecrease of P leading to a trade off in transparency to gainconductivity. This again illustrates the need to determine anoptimal carrier concentration which balances conductivity andtransmission. This provides an immediate advantage for Ag NWnetworks integration for solar cells employing low band gap activeregions or tandem cell architecture. As the infra-red component ofthe solar spectrum can be transmitted into the cell it can be usedfor energy production. The origin of the absorbance of the TCO is aresult of the material being a continuous thin film. Unlike TCOs

which absorb light for 4P the NW networks provide a largeamount of empty space between wires that allows light to passbetween the wires without being absorbed.

The slightly higher average transmittance of the Ag NW film inthe 2501250 nm range indicates that silver nanowires provide apromising alternative to some TCOs in terms of electro-opticalproperties. Furthermore, recent studies have shown that increas-ing the haze factor (ratio of diffuse light to total transmitted light)of a given transparent electrode can help to improve the efficiencyof a solar cell [12]. Light scattering increases the path length oflight through the absorber layer of a solar cell and consequentlyincreases the current generated in the solar cell [13]. The hazefactor of an Ag NW network is dependent on several key factors:length and diameter of the wires used, NW density and fabricationmethod. It is possible to achieve a significant increase in the hazefactor with only a small tradeoff in optical transmission. As shownin Fig. 4 it is possible to achieve a 3 fold increase in the averagehaze factor and only takes 23% from the optical transmission.This tradeoff is also balanced to a certain extent in that theelectrical resistance of electrode B is 9.5 /sq as compared to17.3 /sq for electrode A.

As stated, the haze factor of a Ag NW network is determined byNW density amongst other properties of the network. Intuitively,it would seem that decreasing the NW density per unit area, n,should lead to higher optical transparency. It is by no means adirect relationship, as complex scattering and graduated refractiveindices convolute the relationship of density to transmission [10].In a simple model though, it is possible to calculate the approx-imate transmission and this follows to some extent the intuitiveidea, decreasing density results in higher optical transparency.However, the electrical resistance of the NW networks is alsodensity dependent. Again the trade off emerges with a three waybalance that needs to be struck to produce the optimal network.The relationship of the network resistance to NW density can bedefined as a percolation problem. Several works in the past havedevoted effort to understanding the problem of 2D conductivestick percolation. For instance Pike and Seager [14] and Li andZhang [15] produced some excellent theoretical results thatdefined the problem in its infancy and have led to further under-standing of the problem. Application of percolation theory so farhas been mainly restricted to theoretical works and it is a goodopportunity to observe whether the theory supports the experi-mental data for Ag NW networks. Coleman's group has alreadyapplied percolation theory to transmission data [16]. We focusbelow more on the electrical properties.

Reducing the number of NWs per unit area creates a sparsernetwork that may fail to form enough conducting paths across thenetwork. For low density, i.e. in the percolation regime, the

Fig. 2. (a) Optical transmittance at 550 nm (including the substrate contribution)versus the sheet resistance after annealing, for different deposition techniques:spin coating, rod coating, drop casting and spray injection The green line representsfits to the bulk regime, while the orange line represents fits to the percolativeregime (see Ref. [4] for more details on the equations used). (For interpretation ofthe references to color in this figure legend, The reader is referred to the webversion of this article.)

Fig. 3. Total transmission spectra of a glass substrate (blue), a Ag NWs network(black) and fluorine doped tin oxide (FTO) (red). The associated sheet electricalresistances for Ag NW network and FTO are 9.5 /sq and 7.4 /sq, respectively. (Forinterpretation of the references to color in this figure legend, The reader is referredto the web version of this article.)

Fig. 4. Transmittance spectra (upper curves, dotted line represents the bare glasssubstrate, dashed and solid lines represent Ag NW networks of different densities:(A) 57 mg/m2 and (B) 117 mg/m2) and the related haze spectra of the same samples(lower curves).

D.P. Langley et al. / Solar Energy Materials & Solar Cells 125 (2014) 318324320

properties of the network differ from the bulk material values.In this percolating regime the electrical conductivity sdc dependson the NW density n and follows a power scaling law:

sdcpnnCt 1where nc is the percolation threshold and t is the universalconductivity exponent (equal to 1.29) [17,18]. Eq. (1) is valid whenn is larger than, but close to nc. The percolation threshold of asystem can be defined by the density of the sticks at which there isa 50% chance that a network with that density will have apercolating cluster that spans through the considered system.For large systems with nanowires of length l, the critical numberdensity nc as defined by Li and Zhang is defined as follows [15]:

nc 5:6372670:00002

l2: 2

Eq. (2) clearly states that longer nanowires are associated with alower percolation NW density. However Eq. (2) is valid only in theconsideration of large scale systems, while for the purpose ofphotovoltaics and many other applications this requirement is notalways met.

The density of nanowires required to make a percolatingnetwork has a minimum value as defined by Eq. (2). However thisvalue is accurate only for large systems where the ratio of thesystem size to nanowires length is greater than 30 [15]. In thenetwork shown in Fig. 5a and b it is clear that there are someregions of the network that will not contribute to the conductionpathway or in a solar cell to the collection of photo-generatedelectrons. This will result in reduced collection efficiency fornanowire based electrodes.

In order to determine the density required to produce anetwork which has a higher percentage of the network contribut-ing to the percolating cluster a characteristic length can bedefined. Defining this length Lc as the minimum distance overwhich the probability of percolation at the given density nc isequal to 50% we can then simulate the required densities forvarious values of Lc. Fig. 5c) shows a result of Monte Carlosimulations performed using the fast Monte Carlo method ofNewman and Ziff [19] and displays the effect of decreasing Lcwhich corresponds to an increase in nc. This result is ratherintuitive since a high collection efficiency should be associatedwith percolation clusters occurring on a shorter scale which isassociated with a higher Ag NW network density.

For exploring the influence of the network density one shouldinvestigate the dependence of both the electrical resistance andoptical transparency (usually considered at a wavelength of550 nm). The electrical resistance of Ag NW networks versusdensity is reported in Fig. 6. The vertical dotted line indicatesthe value of nc as defined in Eq. (2). The electrical resistancecorresponds to the minimum electrical resistance measured in-situ during thermal annealing with a ramp of 15 K/min. Asdiscussed above, thermal annealing causes a decrease of theresistance, by reducing the junction resistance between adjacentnanowires. However longer annealing or annealing at highertemperature can cause morphological change of Ag NWs (leadingeventually to sphereodization) which then prevents the networkfrom percolating. Therefore a minimum of sheet resistance isobserved during a thermal annealing ramp.

The blue line in Fig. 6 corresponds to a fit using both Eqs.(1) and (2) which consider percolation over an infinite size system.A good agreement is observed, showing that the percolationregime is valid for the whole network density range investigated.As already mentioned, any experimental values are certainly notnecessarily associated with an infinite size system but with finitesystem size over which the percolation is observed. Experimen-tally we observe the percolation and electrical behavior in

effectively infinite systems; it is therefore expected that theelectronic behavior will match that of the theoretical infinitesystem. In fact as shown in Fig. 6 a finite value of Lc50 mmconstitutes a better agreement with experimental data. This doesnot suggest that the critical density is in some way restricted inthese experiments, but indicates that the network density is in factsufficiently dense so as to provide percolation to a large proportionof the network.

The optical transmittance can also be investigated versus theNW density and the LambertBeer law is usually employed as afirst approximation [11]. The data of the present study is in goodagreement with this approach and is used to simulate the proper-ties of the Ag NW networks when calculating the figure of merit ofthe obtained transparent electrode.

A commonly used figure-of-merit for transparent conductingelectrodes is defined by Haacke [20]:

H T10

Rsh3

where T is the optical transmittance and Rsh is the electrical sheetresistance. The experimental values of are reported in Fig. 7.An optimal density is observed close to an NW mass density of120 mg/m2 which is associated with a sheet resistance of 9.5sqand a total transmittance of 82.9% without removing the lossesdue to the substrate. Subtraction of the substrate leads to a totaltransmittance of about 90%. The dotted line of Fig. 7 correspondsto the calculated figure-of-merit values using Eqs. (1)(3) as wellas the LambertBeer law. A reasonable agreement is obtainedbetween the calculated and experimental values, especially con-sidering the simplicity of the model used. This clearly indicatesthat the optimal Ag NW density considering the tradeoff betweenhigh optical transparency and low electrical resistance is close toabout 100 mg/m2. This would correspond to the same amount ofsilver (for a given area) as a thin silver film of 10 nm thick ascompared to 200300 nm for usual TCOs. The observed optimalAg NW network density observed in Fig. 7 is in rather goodagreement with the results obtained by De et al. [16], who foundoptimal performances for a density of 47 mg/m2 but with adifferent Ag NW morphological parameter and with anotherdeposition technique. Finally the obtained value (16103)is comparable to those of some TCOs except for the best ITO thinfilms [2].

3.2. Potential for incorporation into solar cells

Many groups have already started to incorporate silver nano-wires as front electrodes for solar applications [2124]. As men-tioned, large haze factors result in light scattering into the deviceand increase the effective absorption cross section. The ability ofAg NW networks to produce high haze networks while maintain-ing sufficient electro-optical properties shows the real potential tointegrate them into solar cells. Gaynor et al. [23] demonstrated anincrease of approximately 10% in conversion efficiency of bulkorganic hetero-junction solar cells, by using a Ag NW compositeelectrode as compared to an ITO standard (3.4% using ITO onplastic and 3.8% for Ag nanowire composite). Taking advantageof the metal nanowire network transparency in the infra-redregion of the spectrum Chen et al. [22] created polymer solar cellsthat were semi-transparent in the visible region and had E4%.These cells have an average transmission of about 61% in the 450650 nm wavelength range [22]. This type of solar cell could forinstance be used for energy producing window tinting. Anotheradvantage of Ag NW networks as front electrodes for photovoltaicapplications is that they allow production of flexible solar cells [21]which is difficult to achieve with TCOs due to their brittleness.Within the field of photovoltaics there are many different device

D.P. Langley et al. / Solar Energy Materials & Solar Cells 125 (2014) 318324 321

architectures which have restrictions associated with them; forsome such as Dye Sensitized Solar Cells and CdTe the transparentelectrode must be deposited first with high temperature pro-cesses. CIGS solar cells on the other hand generally require that theTCM is added last and has to be deposited at low temperature toprevent diffusion of the active layers into one another. Dependingon which application the Ag NW electrodes are to be employed theprocessing steps required will differ. Kim et al. [25] demonstratedthe incorporation of ZnO/Ag NW/ZnO multilayer electrode intoCIGSSe solar cells and also demonstrated that ZnO encapsulationimproves the thermal stability of Ag NWs. Improved thermalstability by encapsulation has also been shown by several othergroups [8,26,27], and there are several low temperature routes tohighly conductive Ag NW networks [35]; hence there are manyoptions to support the application of Ag NWs in a broad variety ofsolar cell architectures.

In considering the incorporation of Ag NW networks into solarcells we must also consider the fact that the network presents adiscontinuous film. This means that as a function of positionrelative to the network the probability of a photo-generated

electron being collected will vary. This is especially importantwhen considering the mobility of electrons in the active region ofthe cell. For cells where electrons and excitons can support a longdiffusion length to the collecting electrode, Ag NW networksprovide an interesting and viable solution as a front electrode.When the diffusion length is below 1 mm such as in organic solarcells, Ag NW networks alone are insufficient. However as shownby Kim et al. the incorporation of Ag NW network embedded in aconductive matrix can aid in the collection efficiency [25]. In thiscase the Ag NW network is used to provide the majority of theelectrical performance of the electrode and the matrix materialallows continuous conductivity. In this manner, materials whichhave a low mobility such as PEDOT:PSS may be enhanced.

Let us also note that roughness can be an issue: if somenanowires are not well aligned along the substrate some short-circuits could then occur. The use of high mechanical pressure hasbeen shown to overcome such problems [4]. As a final remark it isworth mentioning that embedding Ag NWs within a transparentoxide such as ZnO or TiO2 could be interesting; for instance the AgNWs stability (either chemical or thermal) could be improved and

Fig. 5. (a) Simulated image of the 10001000 m2 network, (b) same network with wires that are not part of the percolating cluster removed, and (c) graph showing theprobability of percolation versus the Ag NW network density for various characteristic lengths of percolation (Lc). Decreasing the length of percolation results in an increasein the density required to reach the percolation threshold.

D.P. Langley et al. / Solar Energy Materials & Solar Cells 125 (2014) 318324322

properties such as work functions could be tuned. This might helpas well with the integration of Ag NWs into a solar cell since bandalignment plays a crucial role for the solar cell efficiency.

4. Conclusions

Random Ag networks demonstrate electro-optical propertiesclose to that of TCO materials. They depend both on the morphol-ogy of the metallic nanowires (NWs) and the density of the NWnetwork. Post-processing treatments also play an important rolein the resulting electro-optical properties. Several of these treat-ments are investigated in the literature suggesting that the gapwith TCOs could be bridged in the near future. Furthermore, as

reported in the present study of such random networks, simula-tion is a valuable tool to gain more insight into the electro-opticalproperties. From Monte-Carlo simulations, it was found thatdecreasing the characteristic length of percolation resulted inincreasing the density of nanowires required to reach the percola-tion threshold. From this work, it is clear that Ag NW networkscurrently provide sufficient electro-optical properties to be incor-porated into solar cells. Although not yet matching the best ITOfilms in terms of optimal sheet resistance and transmittance, AgNWs provide sufficient properties and are expected to continue toimprove. They are amenable to low deposition temperatures,solution processing, flexibility and variable haze of the resultingelectrode. These aspects suggest that this material could becomeimportant in emerging applications, particularly in the field of

Fig. 6. (a) Experimental resistance values as a function of network density. The blue line in the graph represents a curve fitted to the data using Eqs. (1) and (2), associatedwith an infinite system. The other curves are associated with two different Lc values (Lc100 mm and 50 mm). (b) SEM images of sparse and (c) dense networks. The scale barsrepresent 10 mm. (For interpretation of the references to color in this figure legend, The reader is referred to the web version of this article.)

D.P. Langley et al. / Solar Energy Materials & Solar Cells 125 (2014) 318324 323

photovoltaics in which Indium-free transparent electrodes couldemerge in a near future.

Acknowledgments

We thank E. Bellet-Amalric and M. Anikin for their help in thinfilms characterization, and D. Nguyen for fruitful discussions. Thiswork has been supported by Grenoble INP through the SEI grantand Erasmus Mundus through the IDS FunMat Program.

References

[1] A. Kumar, C. Zhou, The race to replace tin-doped indium oxide: which materialwill win? ACS Nano 4 (1) (2010) 1114.

[2] K. Ellmer, Past achievements and future challenges in the development ofoptically transparent electrodes, Nature Photonics 6 (12) (2012) 809817.

[3] L. Hu, H.S. Kim, J.-Y. Lee, P. Peumans, Y. Cui, Scalable coating and properties oftransparent, flexible, silver nanowire electrodes, ACS Nano 4 (5) (2010)29552963.

[4] T. Tokuno, M. Nogi, J. Jiu, T. Sugahara, K. Suganuma, Transparent electrodesfabricated via the self-assembly of silver nanowires using a bubble template,Langmuir 28 (25) (2012) 92989302.

[5] C.H. Liu, X. Yu, Silver nanowire-based transparent, flexible, and conductivethin film, Nanoscale Research Letters 6 (75) (2011) 18.

[6] J.-Y. Lee, S.T. Connor, Y. Cui, P. Peumans, Solution-processed metal nanowiremesh transparent electrodes, Nano Letters 8 (2) (2008) 689692.

[7] http:/www.seashelltech.com/index.shtml (accessed 06.07.13).

[8] G. Giusti, D. Langley, C. Jimnez, E. Puyoo, A. Muthukumar, E. Appert,V. Consonni, D. Bellet, Fabrication of transparent conductive thin film electro-des based on Ag nanowire on transparent substrates using the spray methodfor photovoltaic applications, MRS Online Proceedings Library 1439 (2012)5156.

[9] S. Sorel, P.E. Lyons, S. De, J.C. Dickerson, J.N. Coleman, The dependence of theoptoelectrical properties of silver nanowire networks on nanowire length anddiameter, Nanotechnology 23 (18) (2012) 185201.

[10] C. Preston, Y. Xu, X. Han, J.N. Munday, L. Hu, Optical haze of transparent andconductive silver nanowire films, Nano Research (2013).

[11] S. De, P.J. King, P.E. Lyons, U. Khan, J.N. Coleman, Size effects and the problemwith percolation in nanostructured transparent conductors, ACS Nano 4 (12)(2010) 70647072.

[12] T. Chih-Hung, H. Sui-Ying, H. Tsung-Wei, T. Yu-Tang, C. Yan-Fang, Y.H. Jhang,L. Hsieh, W. Chung-Chih, C. Yen-Shan, C. Chieh-Wei, L. Chung-Chun, Influencesof textures in fluorine-doped tin oxide on characteristics of dye-sensitizedsolar cells, Organic Electronics 12 (12) (2011) 20032011.

[13] D.S. Hecht, L. Hu, G. Irvin, Emerging transparent electrodes based on thin filmsof carbon nanotubes, graphene, and metallic nanostructures, AdvancedMaterials 23 (13) (2011) 14821513.

[14] G.E. Pike, C.H. Seager, Percolation and conductivity: a computer studyI,Physical Review B 10 (4) (1974) 14211434.

[15] J. Li, S.-L. Zhang, Finite-size scaling in stick percolation, Physical Review E 80(4) (2009) 040104-1040104-4.

[16] S. De, T.M. Higgins, P.E. Lyons, E.M. Doherty, P.N. Nirmalraj, W.J. Blau,J.J. Boland, J.N. Coleman, Silver nanowire networks as flexible, transparent,conducting films: extremely high DC to optical conductivity ratios, ACS Nano 3(7) (2009) 17671774.

[17] M. eelj, I. Stankovi, From percolating to dense random stick networks:conductivity model investigation, Physical Review B 86 (13) (2012) 134202-1134202-6.

[18] D. Stauffer, A. Aharony, Introduction to Percolation Theory, 2nd revised ed.,Taylor and Francis, London, 2003.

[19] M.E.J. Newman, R.M. Ziff, Fast Monte Carlo algorithm for site or bondpercolation, Physical Review E 64 (1) (2001) 016706-1016706-16.

[20] G. Haacke, New figure of merit for transparent conductors, Journal of AppliedPhysics 47 (9) (1976) 40864089.

[21] L. Yang, T. Zhang, H. Zhou, S.C. Price, B.J. Wiley, W. You, Solution-processedflexible polymer solar cells with silver nanowire electrodes, ACS AppliedMaterials Interfaces 3 (10) (2011) 40754084.

[22] C.-C. Chen, L. Dou, R. Zhu, C.-H. Chung, T.-B. Song, Y.B. Zheng, S. Hawks, G. Li,P.S. Weiss, Y. Yang, Visibly transparent polymer solar cells produced bysolution processing, ACS Nano 6 (8) (2012) 71857190.

[23] W. Gaynor, J.-Y. Lee, P. Peumans, Fully solution-processed inverted polymersolar cells with laminated nanowire electrodes, ACS Nano 4 (1) (2010) 3034.

[24] F.S.F. Morgenstern, D. Kabra, S. Massip, T.J.K. Brenner, P.E. Lyons, J.N. Coleman,R.H. Friend, Ag-nanowire films coated with ZnO nanoparticles as a transparentelectrode for solar cells, Applied Physics Letters 99 (18) (2011) 183307-1183307-3.

[25] A. Kim, Y. Won, K. Woo, C.-H. Kim, J. Moon, Highly transparent low resistanceZnO/Ag nanowire/ZnO composite electrode for thin film solar cells, ACS Nano7 (2) (2013) 10811091.

[26] A. Mayoral, L.F. Allard, D. Ferrer, R. Esparza, M. Jose-Yacaman, On the behaviorof Ag nanowires under high temperature: in situ characterization byaberration-corrected STEM, Journal of Materials Chemistry 21 (3) (2010)893898.

[27] P. Ramasamy, D.-M. Seo, S.-H. Kim, J. Kim, Effects of TiO2 shells on optical andthermal properties of silver nanowires, Journal of Materials Chemistry 22 (23)(2012) 1165111657.

Fig. 7. Experimental (symbols) and calculated (dotted line) values of the figure ofmerit defined by Eq. (3) versus the Ag NW network density.

D.P. Langley et al. / Solar Energy Materials & Solar Cells 125 (2014) 318324324

http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref1http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref1http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref2http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref2http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref3http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref3http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref3http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref4http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref4http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref4http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref5http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref5http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref6http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref6http:/www.seashelltech.com/index.shtmlhttp://refhub.elsevier.com/S0927-0248(13)00473-X/sbref7http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref7http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref7http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref7http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref7http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref8http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref8http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref8http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref9http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref9http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref10http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref10http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref10http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref11http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref11http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref11http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref11http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref12http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref12http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref12http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref13http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref13http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref14http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref14http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref15http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref15http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref15http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref15http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref16http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref16http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref16http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref17http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref17http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref18http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref18http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref19http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref19http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref20http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref20http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref20http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref21http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref21http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref21http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref22http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref22http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref23http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref23http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref23http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref23http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref24http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref24http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref24http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref25http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref25http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref25http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref25http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref26http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref26http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref26http://refhub.elsevier.com/S0927-0248(13)00473-X/sbref26

Silver nanowire networks: Physical properties and potential integration in solar cellsIntroductionExperimental sectionResults and discussionsElectro-optical propertiesPotential for incorporation into solar cells

ConclusionsAcknowledgmentsReferences