lable at ScienceDirect

Energy 181 (2019) 1e10

Contents lists avai

Energy

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

Life Cycle Assessment of tandem LSC-Si devices

Marina M. Lunardi a, David R. Needell b, Haley Bauser b, Megan Phelan b,Harry A. Atwater b, Richard Corkish a, *

a The Australian Centre for Advanced Photovoltaics (ACAP), School of Photovoltaic and Renewable Energy Engineering, University of New South Wales,Sydney, 2052, Australiab California Institute of Technology, Department of Applied Physics and Materials Science, Pasadena, CA, 91125, USA

a r t i c l e i n f o

Article history:Received 4 November 2018Received in revised form19 April 2019Accepted 12 May 2019Available online 16 May 2019

Keywords:Life cycle assessmentLCATandem solar cellsEnvironmental impactsPV technologyLuminescent solar concentratorsLSC

* Corresponding author.E-mail address: r.corkish@unsw.edu.au (R. Corkish

https://doi.org/10.1016/j.energy.2019.05.0850360-5442/© 2019 Elsevier Ltd. All rights reserved.

a b s t r a c t

Given the increasing interest in tandem silicon-based solar cells and the recent advances in luminescentsolar concentrators, the luminescent solar concentrators/silicon tandem structure has been proposed asan option for a four-terminal tandem structure. As part of the evaluation of a new type of solar cell, it isimportant to conduct a Life Cycle Assessment to effectively guide research efforts towards cell designswith minimum environmental impacts. Here, we carry out a process-based Life Cycle Assessment toassess global warming, human toxicity (carcinogenic and non-carcinogenic), freshwater eutrophicationand ecotoxicity and abiotic depletion potential impacts associated with three luminescent solar con-centrators/silicon tandem cell structures, considering a bottom layer as being a passivated emitter rearcontact silicon solar cell. The results are based on experimental parameters and show that the increase inthe performance of the cells and modules using the studied tandem structure can produce lowerenvironmental effects than the passivated emitter rear contact technology (single-junction) for theimpact categories studied. These results encourage the studies on cell and module performance im-provements using such tandem luminescent solar concentrators/silicon structures.

© 2019 Elsevier Ltd. All rights reserved.

1. Introduction

The continuous development of photovoltaic (PV) technologiesfocuses on reducing costs and enhancing efficiencies and stabilityfor solar cells and modules [1]. Silicon (Si) remains the dominantcommercial PV technology (above 90% share) in the industry [2],but alternative materials and structures are being reported. PVsingle-junction solar cell limiting power conversion efficiency(PCE), even with improvements, is approximately 30% (dependingon specific assumptions) [3]. Tandem technology makes use of sub-cells optimized for each part of the spectrum, connected eitherelectrically in series (two-terminal) or kept electrically separate(four-terminal), achieving higher efficiencies when compared withsingle junction solar cells [4,5]. Researchers have explored differentorganic and inorganic materials for tandem solar cells to producehigher PCEs [6].

Recent advances in luminescent solar concentrators (LSC), forexample, present a method to capture diffuse sunlight and

).

maintain a relatively low manufacturing cost [7], offering a way tooptimise the value of tandem solar cells. The theoretical efficiencylimits of an LSC that is optically coupled to a flat-plate Si cell (LSC/Sitandem) were explored for a four-terminal tandem structure. Thisstructure exhibits strong power conversion efficiencies for poten-tial high-performance PV [8].

An LSC structure consists of a polymer waveguide withembedded luminophores (photoactive particles), coupled to a PVcollecting material [9,10]. Incident light is absorbed and re-radiatedat down-shifted wavelengths by the luminophores e termedphotoluminescence (PL) e within the polymeric layer. For addi-tional light trapping, common LSC structures optically encase theluminophoric waveguide within filters that are designed to reflectwavelengths of light near the PL center [6]. For such an LSC/Sitandem module, the design necessitates integration of variousoptical and electrical components, as shown in Fig. 1.

The top filter transmits short-wavelength (300e550 nm) andlong-wavelength (700e1100 nm) light for luminophore and subcellabsorption, respectively. The LSC structure employs a one-dimensional photonic crystal (PC) structure as a top, wavelength-selective filter, using alternating layers of low/high refractive in-dex material to achieve the notch-filter reflectance shape. In this

Fig. 1. Three-dimensional schematics of the tandem LSC/Si structure (fabricateddevice).

M.M. Lunardi et al. / Energy 181 (2019) 1e102

analysis we assess two different configurations of LSC cell, whichare differentiated by the top filter layers (dielectrics and polymers).

The dielectric PC consists of SiO2 (low index, ƞavgz 1.45 visible)and TiO2 (high index, ƞavgz 2.5 visible) and has approximately 100layers each of varying thickness on the order of 50 nm. Such a filtercan be constructed via sputter deposition and, in this study, iscalled top filter “a”.

The polymeric PC uses poly(methyl-methacrylate) (PMA) (lowindex, ƞavgz 1.48 visible) and poly(styrene sulfide) (PSS) (highindex, ƞavgz 1.65 visible) and has approximately 1000 layers eachof varying thickness on the order of 50 nm. Polymeric filters such asthis can be fabricated using extrusion and roll-to-roll fashion. Inthis study the polymeric PC is called top filter “b”.

Together with the development of advanced technologies andprocesses there should also be a concern for the environmentalimpacts. Manufacturing a solar cell involves complex process stepsand the use of diverse materials. Each process or resource has itspeculiarities and can come frommany different sources, involving aspecific series of inputs (material or energy which enters a unitprocess [11]) and outputs (material or energy which leaves a unitprocess [11]). Each one of these processes and the use of everymaterial has potential environmental impacts. Several tools andmethodologies can be used to calculate those effects. Life cycleassessment (LCA) is a method used to evaluate potential environ-mental impacts associated with the production, use and disposalphases of a product during its lifetime. The depth of detail of an LCAstudy varies with each case and depends on the goal, scope defi-nition and the assumptions made [11].

This LCA study can assist in sustainable technology develop-ment by focusing on the life cycle environmental consequences of afresh technology that is still in the early stages of development. Theadvance of PV technologies and structures require comprehensiveLCA studies to guide manufactures and policy makers to the bestenvironmental choice of technology such as identifying opportu-nities to improve the environmental aspects of products at variouspoints in their lifecycle, decision-making in industry, governmentalor non-governmental organisations, selection of relevant indicatorsof environmental performance and marketing [11].

This report explores the potential environmental impacts from afour-terminal tandem LSC/Si solar cell [12], considering a standardcommercially available passivated emitter rear contact (PERC) Sisubcell. To date, the environmental analysis of this structure has notbeen reported to the best of the authors’ knowledge. Besides, asimportant outcome from this study is the publication of usefulinventories for LCA studies, which are continuously changing forinnovative Si technologies. There may be found one or more pro-duction steps that dominate the environmental cost of a product, as

has been found for Si PV modules and for all the Si-based tandemmodules studied to date [13e16]. If so, LCA can identify the step(s)and illuminate the areas where attention should be paid to alleviatethe environmental costs.

LCA is a reliable method to calculate and analyse environmentalimpacts and support claims. It provides designers, regulators andengineers with valuable evidence that help them tomake decisionsabout each life stage of materials and products. The importance ofconducting an LCA study in new technologies, such as LSC/Si tan-dem solar modules, is to identify the environmental hot spots inthis new device in order to improve its environmental performancebefore it can be industrialised. LCA is also important during thedevelopment of new products when environmental footprint isimportant to the future marketing or cost structure of this product.

2. Methods

The goal of this LCA is to assess global warming potential (GWP),human toxicity potential - cancer effects (HTP-CE), human toxicitypotential e non-cancer effects (HTP-nCE), freshwater eutrophica-tion potential (FEuP), freshwater ecotoxicity potential (FEcP) andabiotic depletion potential (ADP), comparing LSC/Si tandem and Sisolar modules. The reference and characterisation factor for eachenvironmental impact category is shown in Table 1.

2.1. Global warming potential

The Intergovernmental Panel on Climate Change (IPCC) is theinternational body for assessing the science related to climatechange and their data are revised as the models used in the cal-culations are developed [17]. As solar energy is considered a greensource of energy, GWP is the most common calculated environ-mental impact for LCAs on PV technologies [16,18,19]. GWP is usedto determine the effectiveness of these technologies in reducinggreenhouse gas emissions during the manufacturing processes, useand end of life phases [19]. The calculation of GWP impacts basedon the IPCCmethod is reported in the IPCC's 2001 Third AssessmentReport [20], which defines GWP as the ratio of the time-integratedradiative forcing from the instantaneous release of 1 kg of a tracesubstance relative to that of 1 kg of a reference gas, which is CO2 inthis case (Equation (1)):

GWPðxÞ ¼

ðTH0

ax � ½xðtÞ�dtðTH0

ar � ½rðtÞ�dt(1)

where TH is the time horizon over which the calculation isconsidered; ax is the radiative efficiency due to a unit increase inatmospheric abundance of the substance (in Wm�2kg�1) and [x(t)]is the time-dependent decay in abundance of the substancefollowing an instantaneous release at time t¼ 0. The denominatorcontains the corresponding quantities for the reference gas (CO2).

2.2. Toxicity potential (human and freshwater)

The use of heavymetals and other toxic and hazardousmaterialsduring these processes should also be assessed, for their impact notonly on human health, but also on fauna and flora. HTP (cancer andnon-cancer effects), FEcP and FEuP are comprehensive indicatorsfor these impacts [21]. USEtox calculates human toxicity andfreshwater ecotoxicity impacts by assessing how the toxicologicaleffects of a chemical emitted into the environment affect thecauseeeffect chain that links emissions to impacts, considering

Table 1Impact categories chosen to be analysed in this LCA.

Impact CategoryCharacterisation factor

Global Warming Potential (IPCC [17]) kg of carbon dioxide equivalent (kg CO2 eq.)Human toxicity e cancer effects (USEtox [17]) Comparative toxic units e humans (CTUh)Human toxicity - non-cancer effects (USEtox [17]) Comparative toxic units - humans (CTUh)Freshwater eutrophication (EUTREND [17]) kg of phosphorus equivalent (kg P eq.)Freshwater ecotoxicity (USEtox [17]) Comparative toxic units - ecotoxicity (CTUe)Abiotic depletion (CML 200217) kg of antimony equivalent (kg Sb eq.)

M.M. Lunardi et al. / Energy 181 (2019) 1e10 3

environmental fate, exposure and effects. Equation (2) representshow these impacts are calculated [22].

CF ¼ EF � XF � FF ¼ EF � iF (2)

where CF is the specific characterisation factor (normally in kge-mitted), FF is the corresponding factors that link the quantityreleased into the environment to the chemical masses (or con-centrations) in a given space, XF is the exposure per person (onlyhuman toxicity), EF are the effects, measured by the quantity takenin via a given exposure route by a population or organism to theadverse effects (or potential risk) of the chemical and iF is theintake fraction matrix, which is the result of the fate and theexposure matrices combination and describes the fraction of theemission that is taken in by the overall exposed population [22].

2.3. Eutrophication potential

For the calculation of the eutrophication potential the concen-trations of total nitrogen and phosphorus are the two basicassessment parameters. The calculation is shown in Equations (3)and (4) [23].

TNI¼X

Wj � TNIj (3)

Wj ¼ r2ij.X

r2ij(4)

where TNI is the sum of indices of all nutrient parameters, TNIj is theTNI of parameter j, Wj is the proportion of j parameter in the TNI,and rij is the relation of chlorophyll a (Chla) to other parameters.The available parameters concerned include total nitrogen (TN),total phosphorus (TP), Chla, dissolved oxygen (DO), chemical oxy-gen demand by K2MnO4 oxidation method (CODMn), biologicaloxygen demand (BOD5), etc [23].

2.4. Abiotic depletion potential

ADP, expressed in kg of antimony equivalents, is the relationshipbetween the annual production and the reserves of a resource [17].Since producing solar cells and modules use natural resources, andthe decreasing availability of these natural resources is an impor-tant concern, we are also calculating this environmental impact.The calculation of the ADP is shown in Equation (5),[24], whichrepresents the function of natural reserves (stocks/deposits in theenvironment) of the abiotic resources combined with their rates ofextraction. The natural reserves of these resources are based onconcentrations of the elements and fossil carbon in the Earth'scrust.

ADPi ¼DRi

.ðRiÞ2

DRref��

Rref�2 (5)

where ADPi is the abiotic depletion potential of resource i (in kg Sbeq/kg of resource i); Ri is the ultimate reserve of resource i (kg); DRiis the extraction rate of resource i (kg/yr) (regeneration is assumedto be zero); Rref is the ultimate reserve of the reference resource (Sbin kg) and DRref is the extraction rate of the reference resource, Rref(in kg/yr).

2.5. Scope and assumptions

The environmental impacts calculation is based on the arearequired for a module to produce 1 kWh of electricity, which is ourfunctional unit (FU). This LCA is a cradle-to-grave, which meansthat the stages within the system boundaries of this study initiatewith the raw materials necessary for the cells' production andfinish at the modules’ end of life (assumed to be disposed tolandfill).

The processes from the raw materials until the cell production,as well as the end-of-life and landfill scenarios, are described in theliterature [13]. This analysis considers the material and energyflows of primary processes, together with extraction of raw mate-rials and, whenever available in the Ecoinvent database and/or theliterature, the intermediate processes. The inclusion or exclusion ofany step can significantly affect the outcome, particularly the end oflife, which might change the results considerably. The main likelypossibilities for end of life of solarmodules are landfill, incineration,reuse and recycling. The comparison of the environmental impactsfrom each of these scenarios have been published by a few authors[25,26].

Besides the impacts of different ends of life included in thesystem boundaries, there are also impacts from the balance ofsystem (BOS) that might affect the overall results of the LCA.However, the inclusion of BOS in LCA studies of PV technologies isnot common, and thus, the inventory is scarce. Hence, the BOS isnot considered in this LCA either. If the efficiency is increased, thenecessary quantities of area-related BOS components, such as landarea, mounting frame and cable, decrease to generate the sameamount of electricity. Consequently, the environmental impactsarising from the BOS generally tends to reduce as well. Hence, wecan argue that exclusion of the BOS from our LCA is likely to lead toan under-rather than over-estimation of the environmental bene-fits arising from cell and module efficiency improvement.

This study is assuming a lifetime of 30 years for all modules(panel manufacturers commonly guarantee that their panels willproduce electricity at 80%e90% of their power output rating at theend of 25e30 years [27]) and that all the modules go to the landfillafter the end-of-life. However, it is important to notice that theimpacts of disposal may be shown to be significant [28e30].Possible end of life scenarios for Si-based PV modules, such as

M.M. Lunardi et al. / Energy 181 (2019) 1e104

landfill, incineration, reuse and recycling were assessed and theresults validate the environmental benefits of the recycling pro-cesses [25]. Similar results can be expected if the impacts from theLSC/Si tandem are produced mainly from the bottom layer of thecell (PERC Si).

The materials considered for the module assembly are ethylvinyl acetate (EVA) encapsulant, aluminium frame, polymer back-sheet, cover glass, copper/tin tabbing and lead-containing solder[31], which are the most common materials for Si-based solarmodules. This LCA assumes that the cell and module production islocated in the United States of America (USA) and uses an electricitymix for that country, although it is well known that it could occur inother countries too [32]. The USA electricity mix is mainly based infossil fuels, such as petroleum, natural gas, and coal that, combined,accounted for about 77.6% of the county's primary energy produc-tion in 2017 [33]. This value is similar to the estimated share of totalfinal energy consumption worldwide in 2016, which is 79.5% [34].China remains the biggest producer and customer for Si cells [35]and its electricity mix is based in approximately 72% in fossil fuels(coal based) [36]. Considering the similarities between USA, themajor producer of solar cells (China) and the average of electricitymix in the world, the environmental impacts from the use of fossilfuels described in this study are acceptably representative. How-ever, the results for this LCA considering the electricity mix fromChina are also presented in the SI (Figure SI-1) and discussed insection 3 Results and Discussion. The inventory is from the Ecoin-vent database [37] and simulation work [8] using GaBi software[38], whichmodels a scaled-upmanufacturing process for all layersof the device (Tables SI-1 and SI-2 in the SI). Ecoinvent lists theimpacts of every element of a product or system and estimates theimpacts of all stages of a product's life cycle [38].

We undertake the LCA for this module by estimating thenecessary energy inputs and processing requirements for each ofthe individual components as well as for the full device integrationand assembly. We assemble two variations for PL-trapping, topfilter designs: dielectric and polymeric Bragg stack filters composedof alternating high/low refrative indexmaterials. Such designs haveshown high tunability and optimized reflectance/transmittancecharacteristics [39,40]. Such structures are grown via sputterdeposition (dielectric) or extrusion processing (polymeric). Poly-meric waveguides are fabricated via doctor blading technology,wherewe deposit poly(laurylmethacrylate) (PLMA) and cure underUV exposure. CdSe/CdS QD lumihophores are grown in a colloidalsynthesis [41]. We grow InGaP micro-cells via MOCVD and mesa-etch in order to isolate micro-cells which are 0.4� 0.4mm in area[42]. Via pick-and-place machining and screen printing technology,we arrange and electrically interconnect the micro-cells into a gridpattern, arranging the cells to lie such that the total waveguide tocell area (termed the geometric gain) is on the order of 100. Wefabricate the bottom AlSb high contrast grating (HCG) metasurfacefilters first by co-sputtering thin films of AlSb from Al and Sb tar-gets. We then pattern the film into cylindrical pillars in a hexagonalarray via nano-imprint lithography followed by a dry or wet etchprocess. The hexagonal array of the pillars generates the desiredoptical properties of a reflectance peak centered aroun the QDemission and transmission in all other wavelength regions [43].

The LSC/Si tandem module is composed of five primary com-ponents: a polymeric/dielectric top filter, a CdSe/CdS QD wave-guide, an InGaP micro-cell PV array, a bottom metasurface HCGfilter, and a Si subcell [44]. For the top cell, specifically the LSC layer,we consider two possible materials for the top notch filter (poly-meric and dielectric). For the Si cell layer, we adopt the currentindustry standard process for passivated emitter and rear cell(PERC) [5]. It is important to highlight that the results from LCAscan differ depending on the system boundaries, the functional unit

(FU) and other assumptions [6]. The tandem structures analysed inthis LCA are shown in Fig. 2 and Table 2, where the different topfilters, explained in the introduction section, are shown.

The calculation of the environmental impacts of these technol-ogies is based on the area required for a solar module to produce1 kWh of electricity (functional unit). This calculation considers themodule's assumed efficiency and lifetime (Table 3), the average USinsolation condition (assumed to be 1800 kWh/m [2]/year [45]) andthe performance ratio (set as 0.75 16 for all cases in this study). Thefinal environmental impacts results are highly sensitive to theseassumptions which were chosen carefully to better represent themajority of LCA studies of PV technologies [46], in order to berelevant and applicable for further studies.

3. Results and Discussion

The environmental impacts from the PV technologies analysedare calculated based on the LCA methodology described previouslyand the collected inventory (Tables SI-1 and SI-2 in the SI).

Fig. 3 presents the results for GWP, HTP-CE, HTP-nCE, FEuP, FEcPand ADP of LSC/Si tandem compared to Si (PERC) solar modules.

The analysis of the results (Fig. 3) shows that the tandemtechnology's efficiency improvement influences the environmentaloutputs positively, due to the lower energy usage to produce thesame amount of solar-derived energy (1 kWh, which is the func-tional unit of this LCA) during the module's lifetime, at relativelylow environmental costs.

3.1. Global warming potential

The GWP effects show thatmost of the impacts come from the Silayer of the tandem solar cells, which is in accord with other LCAstudies of Si-based tandem technologies [15,47]. The energy inputis an important factor in terms of environmental impacts, especiallyfor GWP. The Si that is used in the PV industry has tomeet very highpurity standards and therefore requires energy-intensive refine-ment techniques. The Si treatment processes, especially the monoc-Si (Czochralski method [48]) crystallisation process generatesrelatively little waste, but the main concern is the energy requiredand the amount of argon gas used during this process, which resultsin the most significant impacts on this category [48].

As mentioned above, this analysis used the USA electricity mixin its calculations. Fossil fuel usage accounts for approximately77.6% of the county's primary energy production, which is similar tothe total final energy consumption worldwide (79.5%). BecauseChina remains the biggest producer and customer for Si cells, wealso calculate the impacts of the studied cell structures consideringthe electricity mix from China, which is presented in the SI(Figure SI-1). Due to the similarities of the percentage use of fossilfuels in the electricity mix of the USA and China, the environmentalimpact results for these two countries are similar. The GWP impactis almost the same value, as expected. This category is directlydependent on the type of energy source used during the productionprocesses and, because of that, the results are compatible. The otherimpacts categories follow this trend and are very similar comparingthe two chosen countries. The impacts would be expected to belower in the case of an electricity mix that uses renewable energiesas the primary source of energy such as Iceland, Sweden andDenmark. However, this is not yet the reality formost of the PV cellsand modules producers (China, Japan, USA, etc).

One possibility to reduce the impacts of the Si treatment is touse different approaches such as the use of solar grade Si (SGS). Theenergy consumption to produce SGS is estimated to be 25e30 kWh/kg of product, which is a significant reduction, compared to the EGSproduction via the conventional Siemens (approximately

Table 2LSC layer materials (including lower end and upper end), parameters and processes analysed.

Layer Material Parameter Materials Required (g/cm2) Processing

Lower End Upper End

Dielectric Filter SiO2 Layer Thickness 6.63E-04 6.90E-04 SputteringTiO2 1.06E-03 1.55E-03

Polymeric Filter PMA 3.00E-03 3.00E-02 Roll-to-RollPSS 2.50E-03 2.50E-02

LSC Waveguide CdSe/CdS Concentration 1.05E-06 2.10E-06 Colloidal SynthesisPLMA Thickness 1.39E-03 2.80E-03 Doctor Blade

Micro-cell Array NOA Adhesive Required 1.29E-04 2.58E-04 Spin CoatedInGaP Geometric gain 7.50E-06 3.75E-05 MOCVDSiNx 3.80E-07 1.90E-06 Screen PrintingCu 8.40E-06 4.20E-05Au 6.00E-06 3.00E-05Glass Type 1.23E-01 6.99E-01

High Contrast Grating Al Pattern Thickness 1.35E-02 2.70E-02 Nanoimprint LithographySb 3.35E-02 6.70E-02

Fig. 2. Tandem LSCeSi structures analysed (including a 2D breakdown of the dominant fabrication procedure required for each layer). They differ only in the top filter.

M.M. Lunardi et al. / Energy 181 (2019) 1e10 5

120e160 kWh/kg of product) [49], but at some cost to efficiency.

3.2. Toxicity potential (human and freshwater)

Considering the HTP-CE and HTP-nCE, the most significantimpact results from EGS and mono c-Si processes. As discussed,these processes require substantial amounts of electricity, whencompared with the other methods considered. These impacts resultfrom the assumed use of USA electricity mix, based on fossil fuels(62.7%) [50], as the source of primary energy. The contributions tothis category are mostly from the mining process, plant operation,and fossil fuels transportation in the USA [51].

The Si PERC cells and module fabrication also present significantimpacts for most of the environmental categories analysed. Thesubstances that contribute to this category most significantly arelead, arsenic, and mercury and the concern is their emission to air,

Table 3Assumed efficiency and lifetime of the solar modules.

Parameter Unit PERC (EGS) Tandem LSC-Si (a) Tandem LSC-Si (b)

Efficiency % 21.2 24.5* 24.5*Lifetime years 25 25 25*Average based on computational analyses [8].

mainly generated from the silver paste (conductive metal paste,used to form the front contact grid), electricity supply (USA elec-tricity mix) and glass production process for the modules [52]. Thedisposal of the PV modules also has a significant impact, since weassume that all materials go to landfill. Particularly, lead exhibitshigh cancer and non-cancer toxicity potential for humans andfreshwater [53]. In the future, improved recycling technologies arelikely to mitigate these costs [54].

Lead is a health hazard, since when inhaled (in the case ofemission to air) is stored in the bones, teeth and blood, and maydamage liver, kidneys and brain affecting the health of children,unborn babies and adults [55]. The metabolism of arsenic inhumans can result in the formation of dimethylarsenic acid (DMA),which is carcinogenic to mammals [56]. Mercury emissions fromthe coal smoke is the main source of anthropogenic discharge andmercury pollution in atmosphere [57].

Specifically for the CdSe/CdS layer in the LSC device, there is anenvironmental concern with the use of cadmium QDs. Cadmium'spossible toxicity and use in these QDs may be addressed throughorganic coatings which may potentially protect the QDs and pre-vent the release of cadmium to the environment [58]. These im-pacts are not included in the present study, due to the scarcity ofdata, but it will be important to study the environmental impacts ofthese substances in various use and disposal scenarios in the future.

Fig. 3. Global Warming Potential (GWP), Human Toxicity Potential e Cancer and non-Cancer Potential (HTP-CE and HTP-nCE), Freshwater Eutrophication Potential (FEuP),Freshwater Ecotoxicity Potential (FEcP) and Abiotic Depletion Potential (ADP) results for the three technologies studied (PERC Si, LSC(a)/PERC Si and LSC(b)/PERC Si, where “a”represents the top filter: TiO2/SiO2 Stack Filter and “b” represents the top filter: PMA Stack Filter.

M.M. Lunardi et al. / Energy 181 (2019) 1e106

M.M. Lunardi et al. / Energy 181 (2019) 1e10 7

A case study showed that the impacts from a CdSe quantum dotsolar systems would exhibit lower environmental emission levelsthan other types of PV solar systems for a few categories, but thereare significant negative effects of heavy metal emissions andquantum dots to ecological and public health [59].

Considering assumptions made in this study, it can be seen thatfor HTP (CE and nCE) and FEcP the impacts from the tandemstructures are lower compared with the single-junction Si module,mainly because of the high efficiency achieved by the tandemconfiguration.

3.3. Eutrophication potential

The FEuP impacts of the Si are dominated by emissions to air andfreshwater, including nitrogen oxides, phosphate and nitrate dur-ing the Si treatment and cell manufacturing processes. In themodule production stage, encapsulant, backsheets and aluminiumframe are the main contributors to this impact category becausethey emit phosphate and nitrogen oxides during their life cycles.These excessive nutrients in the water can cause unexpectedgrowth of water-weeds (phytobenthon), free-floating plant or-ganisms (phytoplankton) and other plant forms [60].

Emissions in the form of nitrogen oxides and phosphorus witheutrophication potential also generate some impacts during theingot growing, however not as significative as the cellmanufacturing and module production processes.

3.4. Abiotic depletion potential

Materials like glass and aluminium have the most influence onADP, through the module production phase, while metals such ascopper and silver are the ones that show the most significant im-pacts for the cell manufacturing phase (for all structures studied inthis LCA). Silver-based metallization paste was a significantcontribution to ADP. This impact has been addressed and thequantities of silver per m2 of cell are decreasing considerably [61]over time with improved cell technology.

Importantly, it has been shown that the recycling and reuse ofmetals and other materials from solar cells and panels canconsiderably decrease the ADP impacts from the production of Si-based PV modules [25,29].

3.5. Sensitivity analysis

Compared with the bottom layer (PERC Si) of the tandemstructures analysed, the impacts from the top layer (LSC “a” and“b”) are very low. However, it is important to understand whichprocess steps and materials used in the LSC layer have the mostsignificant environmental effects for this innovative technology.

A sensitivity analysis is essential for the assessment of theimpact from the LSC top cell (shown in Fig. 4), which uncertaintiesare due to specific parameters that are different depending on eachlayer of the LSC cell. For the top filters, the layer thickness param-eter controls the overall amount of material required, and this willchange given need for spectral profile control, which entails un-certainties for TiO2/SiO2 stack filter (“a’) and PMA stack filter (“b”).The impact of these parameters is more significant for the top filter“b”, mainly because, comparedwith the polymeric stack filter, moreexperiments have been done involving the dielectric stack filter andnow the methods to fabricate this layer structure have been care-fully optimized, while for the polymeric stack filter, the processoptimization is much harder from a design perspective. The reasonfor that is because the index of refraction of each layer (consideringthe polymeric stack filter) is limited given that these are polymermaterials. As such, a more substantial error is given for the

thickness of each layer and the numbers of layers altogether of thepolymeric stack filter.

Considering the other LSC layers, for the waveguidematerial theconcentration can be cut in half if there is an excellent high contrastgrating in place, which represents less material use and a reductionof the environmental impacts. For the micro-cell array layer, thelower geometric gain (GG) yields (in this analysis we considerGG¼ 100 for lower end material use and GG¼ 20 for the higherend), the higher device performance but comes at the cost ofgreater amounts of material. For the glass substrate, it is assumed avalue from 0.12500 to 0.02200 Gorilla® Glass substrate. For the HCGthe pitch, radius, and height of the pillars of the grating determinethe amount of material required (in this LCAwe assumed the lowerend¼ 100 mm height, and the higher end¼ 200 mm height).

Fig. 4 shows only the impacts from the LSC layer (including thesensitivity analysis), not considering the impacts from the bottomlayer (PERC Si). It is important to highlight that the values in Fig. 4are much slower than Fig. 3, due to the low impact from the LSC “a”and “b” layers.

From Fig. 4 it can be seen that, due to the higher use of elec-tricity, the impacts are more significant for (a) than (b). Comparedwith the polymeric process, the dielectric process uses more elec-tricity. These results are expected, as it has already been reported inthe literature that tools and equipment used in a roll-to-roll processusualy use less energy per unit area of manufactured product,which included solar cells, for a shorter duration relative to con-ventional manufacturing processes, such as the sputtering process.Benefits of the roll-to-roll technology include the reduced con-sumption of materials and the reduction of equipment size andfloor space (lower-cost), compared with other technologies [62].

The use of InGaP, copper, gold and antimony results in higherHTP-nCE impacts when compared with the other materials,because these metals can cause various health problems to humansrelated to their mining processes and/or use (eg., problems relatedto the nervous system, bones, intestine, cardiac and liver) [63].Some of these metals, such as gold and indium, also impact on theADP because they are not very abundant.

4. Conclusions

An LCA of LSC/Si (PERC) tandem solar modules was conducted toguide researchers, manufacturers and policymakers to the possibleenvironmental impacts related to this new technology. A set of sixenvironmental impacts, considered by the authors to be the mostrelevant for the studied product and materials, were analysed:GWP, HTP-CE and HTP-nCE, FEuP, FEcP and ADP. The resultsdemonstrate that the increase in the efficiency of the cells and,consequently, of the modules caused by the use of LSC/Si (PERC)tandem structure generates lower environmental impacts whencompared with a single junction Si (PERC) technology, consideringthe categories studied. This result is expected since the processesinvolved to create the top layers of the tandem solar cell demand alow quantity of energy andmaterials input, and energy is here, as inother Si-based photovoltaic modules [15,47], the dominant sourceof environmental costs.

In particular, the use of electricity is the major cause of most ofthe environmental impacts assessed. This contribution is mostlyfrom the mining process, plant operation, greenhouse gases emis-sions and transportation. During the Si treatment process (Czo-chralski method [48]) there is an intensive use of energy, whichaffects most of the impacts, particularly global warming and humantoxicity effects, where it can be seen that the Si (PERC) layer pre-sents the most important impacts.

Besides electricity supply, the Si PERC cells and modules fabri-cation also present some significant impacts for all categories

Fig. 4. Global Warming Potential (GWP), Human Toxicity Potential e Cancer and non-Cancer Potential (HTP-CE and HTP-nCE), Freshwater Eutrophication Potential (FEuP),Freshwater Ecotoxicity Potential (FEcP) and Abiotic Depletion Potential (ADP) results for the LCS layer only, not considering the bottom layers included in Fig. 3 (PERC Si). Where “a”represents the top filter: TiO2/SiO2 Stack Filter and “b” represents the top filter: PMA Stack Filter.

M.M. Lunardi et al. / Energy 181 (2019) 1e108

because of the use of metals (including lead and the silver), andglass production process for the modules. Particularly for ADP, theuse of non-renewable resources, such as glass, copper, silver andaluminium have the most influence, through the module produc-tion and cell manufacturing phases (for all structures studied in thisLCA). Our results encourage cell and module performance im-provements using tandem structures and new materials and

processes and are in accordance with other published studies inthis area.

Mostly because of the assumption that all the end-of-life ma-terials go to landfill, the disposal of the PV modules also has somesignificant impacts. Particularly, lead exhibits high toxicity poten-tial for humans and freshwater and the use of QDs may raisetoxicity concerns. However, as discussed in the literature

M.M. Lunardi et al. / Energy 181 (2019) 1e10 9

confirming the environmental benefits of the recycling processeswhen compared with other possible end of life possibilities, such aslandfill, incineration and reuse [25,26], changes in the end of lifescenario can affect these environmental outcomes.

It is important to note that all the results presented in this LCAare based on preliminary studies, since no industry data is availablefor the new LSC technologies analysed. Also, as renewables arepenetrating electricity grids over the world, the impacts fromelectricity use are expected to fall in future [64]. It should further benoted that the IEA PVPS inventory of impacts for Si cells was pub-lished in 2012 and improvements in manufacturing methods since.

In summary, the key finding of this study is that the improve-ments in the modules’ efficiency using LSC/Si (PERC) tandem de-vices can result in lower environmental impacts compared with Si(PERC) technology. This LCA shows that it is possible to increase theefficiency of solar cells and modules without increasing the envi-ronmental impacts produced.

Acknowledgements

MML and RC acknowledge the support of the Australian Gov-ernment through the Australian Renewable Energy Agency(ARENA, Grant SRI001). Responsibility for the views, information oradvice expressed herein is not accepted by the Australian Gov-ernment. Additionally, the first author would like to acknowledgeCoordenaç~ao de Aperfeiçoamento de Pessoal de Nível Superior(CAPES) for her scholarship.

Appendix A. Supplementary data

Supplementary data to this article can be found online athttps://doi.org/10.1016/j.energy.2019.05.085.

References

[1] Green MA. Commercial progress and challenges for photovoltaics. NatureEnergy 2016;1:15015.

[2] Burger B, Kiefer K, Kost C, Nold S, Philipps S, Preu R, Schindler R, Schlegl T,Stryl-Hipp G, Willeke G. Photovoltaics report. Fraunhofer Institute for solarenergy systems. Freiburg 2014;24.

[3] Shockley W, Queisser HJ. Detailed balance limit of efficiency of p-n junctionsolar cells. J Appl Phys 1961;32(3):510e9.

[4] Green MA. Silicon wafer-based tandem cells: the ultimate photovoltaic solu-tion?. In: International society for optics and photonics; 2014. 89810L-89810L-6.

[5] Almansouri I, Ho-Baillie A, Green MA. Ultimate efficiency limit of single-junction perovskite and dual-junction perovskite/silicon two-terminal de-vices. Jpn J Appl Phys 2015;54(8).

[6] Debije MG, Verbunt PP. Thirty years of luminescent solar concentratorresearch: solar energy for the built environment. Advanced Energy Materials2012;2(1):12e35.

[7] Debije MG, Rajkumar VA. Direct versus indirect illumination of a prototypeluminescent solar concentrator. Sol Energy 2015;122:334e40.

[8] Needell DR, Ilic O, Bukowsky CR, Nett Z, Xu L, He J, Bauser H, Lee BG, Geisz JF,Nuzzo RG. Design criteria for micro-optical tandem luminescent solar con-centrators. IEEE Journal of Photovoltaics 2018;99:1e9.

[9] Batchelder J, Zewai A, Cole T. Luminescent solar concentrators. 1: Theory ofoperation and techniques for performance evaluation. Appl Optic1979;18(18):3090e110.

[10] Batchelder J, Zewail A, Cole T. Luminescent solar concentrators. 2: experi-mental and theoretical analysis of their possible efficiencies. Appl Optic1981;20(21):3733e54.

[11] International Organization for Standardization. ISO 14040: environmentalmanagement-life cycle assessment-principles and framework. 1997.

[12] Atwater HA. New avenues for photonic design in photovoltaics & solar fuelgeneration. In: Presented at UNSW School of photovoltaic & renewable en-ergy engineering presentation; 2018. Available at: http://www2.pv.unsw.edu.au/videos/Harry-Atwater-22January2018/seminar.php. [Accessed 2 April2018].

[13] Lunardi MM, Alvarez-Gaitan JP, Chang NL, Corkish R. Life cycle assessment onPERC solar modules. Sol Energy Mater Sol Cell 2018;187:154e9.

[14] Lunardi MM, Moore S, Alvarez-Gaitan JP, Chang NL, Corkish R. Life cycleassessment on advanced silicon solar modules. In: Asia-pacific solar researchconference, melbourne; 2017.

[15] Lunardi MM, Moore S, Alvarez-Gaitan JP, Yan C, Hao X, Corkish R.A comparative life cycle assessment of chalcogenide/Si tandem solar modules.Energy 2018, Vol. 145.

[16] Frischknecht R, Itten R, Sinha P, de Wild-Scholten M, Zhang J, Fthenakis V,Kim HC, Raugei M, Stucki M. Life cycle inventories and life cycle assessment ofphotovoltaic systems. International Energy Agency (IEA) PVPS Task 12 2015:4.Report T12.

[17] JRC E. ILCD handbook Recommendations for Life Cycle Impact Assessment inthe European context-based on existing environmental impact assessmentmodels and factors. first ed. Publication Office of the European Union; 2011.

[18] Louwen A, Sark W, Schropp R, Turkenburg W, Faaij A. Life-cycle greenhousegas emissions and energy payback time of current and prospective siliconheterojunction solar cell designs. In: Progress in photovoltaics: research andapplications; 2014.

[19] Hsu DD, O'Donoughue P, Fthenakis V, Heath GA, Kim HC, Sawyer P, Choi JK,Turney DE. Life cycle greenhouse gas emissions of crystalline silicon photo-voltaic electricity generation. J Ind Ecol 2012;16(s1):S122e35.

[20] Watson RT. Climate change 2001: synthesis report, contribution of workinggroups I, II, and III to the Third assessment report, intergovernmental Panel onclimate change. Cambridge University; 2001.

[21] Monteiro Lunardi M, Wing Yi Ho-Baillie A, Alvarez-Gaitan JP, Moore S,Corkish R. A life cycle assessment of perovskite/silicon tandem solar cells. In:Progress in photovoltaics: research and applications; 2017.

[22] Rosenbaum RK, Bachmann TM, Gold LS, Huijbregts MA, Jolliet O, Juraske R,Koehler A, Larsen HF, MacLeod M, Margni M. USEtoxdthe UNEP-SETACtoxicity model: recommended characterisation factors for human toxicityand freshwater ecotoxicity in life cycle impact assessment. Int J Life CycleAssess 2008;13(7):532.

[23] Cheng X, Li S. An analysis on the evolvement processes of lake eutrophicationand their characteristics of the typical lakes in the middle and lower reachesof Yangtze River. Chin Sci Bull 2006;51(13):1603e13.

[24] van Oers L, Guin�ee J. The abiotic depletion potential: Background, updates,and future. Resources 2016;5(1):16.

[25] Lunardi MM, Alvarez-Gaitan JP, Bilbao J, Corkish R. Comparative life cycleassessment of end-of-life silicon solar photovoltaic modules. Appl Sci2018;8(8):1396.

[26] Fthenakis VM. End-of-life management and recycling of PV modules. EnergyPolicy 2000;28(14):1051e8.

[27] Weaver J. Solar module lifetime predictions are getting better. PV magazine -Photovoltaics Markets and Technology 2018. Available at: https://www.pv-magazine.com/2018/12/10/solar-module-lifetime-predictions-are-getting-better/. Acessed on: 05/01/2019.

[28] Vellini M, Gambini M, Prattella V. Environmental impacts of PV technologythroughout the life cycle: importance of the end-of-life management for Si-panels and CdTe-panels. Energy 2017;138:1099e111.

[29] Latunussa CE, Ardente F, Blengini GA, Mancini L. Life Cycle Assessment of aninnovative recycling process for crystalline silicon photovoltaic panels. SolEnergy Mater Sol Cell 2016;156:101e11.

[30] Latunussa C, Mancini L, Blengini G, Ardente F, Pennington D. Analysis ofmaterial recovery from silicon photovoltaic panels. Luxembourg: PublicationsOffice of the European Union; 2016.

[31] ITRPV, International. Technology roadmap for photovoltaic results 2016. isavailable at:. Eight Edition 2017.. https://www.eia.gov/todayinenergy/detail.php?id=33092. accessed on: 05/01/2019.

[32] Zhao Z-Y, Yang H-J, Zuo J. Evolution of international trade for photovoltaiccells: a spatial structure study. Energy 2017;124:435e46.

[33] U.S. Departament of Energy. The United States uses a mix of energy sources.U.S. Energy Information Administration; 2018.

[34] REN21. Renewables 2018 global status report. Paris: REN21 Secretariat; 2018.[35] Hutchins M. The weekend read: cell manufacturer ranking. PV magazine -

Photovoltaics Markets and Technology. 2018. Available at: https://www.pv-magazine.com/2018/03/30/weekend-read-cell-manufacturer-ranking/.Acessed on: 08/10/2018.

[36] West B. Chinese coal-fired electricity generation expected to flatten as mixshifts to renewables. EIA’s International Energy Outlook 2017. https://www.eia.gov/todayinenergy/detail.php?id=33092. Accessed on: 05/01/2019.

[37] Jungbluth N, Stucki M, Frischknecht R. Photovoltaics. Sachbilanzen vonEnergiesystemen: grundlagen für den €okologischen Vergleich von Ener-giesystemen und den Einbezug von Energiesystemen in €Okobilanzen für dieSchweiz. Ecoinvent report 2009, Vol. 6.

[38] GaBi LCA software. 2015. Available at: http://www.gabi-software.com/australia/index/. Acesses on: 20/01/2016.

[39] Tibuleac S, Magnusson R. Reflection and transmission guided-mode resonancefilters. JOSA A 1997;14(7):1617e26.

[40] Torres CS, Chigrin D, Vivas J, Goldschmidt A, Zankovych S, Ferrand P,Romanov S. Polymer-based photonic crystals. In: Proceedings of 2003 5th

international conference - IEEE, transparent optical networks, vol. 190; 2003.1.

[41] Bronstein ND, Yao Y, Xu L, O'Brien E, Powers AS, Ferry VE, Alivisatos AP,Nuzzo RG. Quantum dot luminescent concentrator cavity exhibiting 30-foldconcentration. ACS Photonics 2015;2(11):1576e83.

[42] Geisz JF, Steiner MA, Garcia I, Kurtz SR, Friedman DJ. Enhanced externalradiative efficiency for 20.8% efficient single-junction GaInP solar cells. ApplPhys Lett 2013;103(4):041118.

[43] Bauser H, Needell DR, Bukowsky CR, Ilic O, Nett Z, Lee BG, Geisz JF,

M.M. Lunardi et al. / Energy 181 (2019) 1e1010

Alivisatos AP, Atwater HA. Metasurfaces as wavelength selective mirrors intandem luminescent solar concentrators. In: 7th edition of the world con-ference on photovoltaic energy conversion. USA: WCPEC) - IEEE Waikoloa, HI;2018.

[44] Bauser HC, Needell DR, Bukowsky CR, Ilic O, Nett Z, Lee B, Geisz JF,Alivisatos AP, Atwater HA. Metasurfaces as wavelength selective mirrors intandem luminescent solar concentrators. In: 7th edition of the world con-ference on photovoltaic energy conversion. USA: WCPEC) - IEEE Waikoloa, HI;2018.

[45] Otkin JA, Anderson MC, Mecikalski JR, Diak GR. Validation of GOES-basedinsolation estimates using data from the US Climate Reference Network.J Hydrometeorol 2005;6(4):460e75.

[46] Gerbinet S, Belboom S, L�eonard A. Life Cycle Analysis (LCA) of photovoltaicpanels: a review. Renew Sustain Energy Rev 2014;38:747e53.

[47] Lunardi MM, Ho-Baillie AWY, Alvarez-Gaitan JP, Moore S, Corkish R. A lifecycle assessment of perovskite/silicon tandem solar cells. Prog PhotovoltaicsRes Appl 2017;25(8):679e95.

[48] Zulehner W, Huber D, Grabmaier J. Crystals growth, properties and applica-tions. Springer Verlag; 1982 (8).

[49] de Wild-Scholten M, Alsema E. Towards cleaner solar PV: environmental andhealth impacts of crystalline silicon photovoltaics. Refocus 2004;5(5):46e9.

[50] International Energy Agency (IEA). World energy outlook 2017 - executivesummaryvol. 13; 2017.

[51] Hedemann J, K€onig U, Cuche A, Egli N. Technical documentation of theecoinvent database. Final report ecoinvent data v2 2007;2.

[52] Chen W, Hong J, Yuan X, Liu J. Environmental impact assessment of mono-crystalline silicon solar photovoltaic cell production: a case study in China.J Clean Prod 2016;112:1025e32.

[53] European Commission. Directive 2008/98/EC of the european parliament andof the council of 19 november 2008 on waste and repealing certain directives(waste framework directive). LexUriServ; 2008.

[54] Weckend S, Wade A, Heath G. End-of-life management solar photovoltaicpanels. IRENA and IEA-PVPS; 2016.

[55] Yun YH, Sim JA, Jung JY, Noh DY, Lee ES, Kim YW, Oh JH, Ro JS, Park SY, Park SJ,Cho KH, Chang YJ, Bae YM, Kim SY, Jung KH, Zo ZI, Lim JY, Lee SN. The asso-ciation of self-leadership, health behaviors, and posttraumatic growth withhealth-related quality of life in patients with cancer. Psycho Oncol2014;23(12):1423e30.

[56] Bissen M, Frimmel FH. Arsenicda review. Part I: occurrence, toxicity, speci-ation, mobility. Acta Hydrochim Hydrobiol 2003;31(1):9e18.

[57] Zahir F, Rizwi SJ, Haq SK, Khan RH. Low dose mercury toxicity and humanhealth. Environ Toxicol Pharmacol 2005;20(2):351e60.

[58] Mahendra S, Zhu H, Colvin VL, Alvarez PJ. Quantum dot weathering results inmicrobial toxicity. Environ Sci Technol 2008;42(24):9424e30.

[59] Sengül H, Theis TL. An environmental impact assessment of quantum dotphotovoltaics (QDPV) from raw material acquisition through use. J Clean Prod2011;19(1):21e31.

[60] Smith VH, Tilman GD, Nekola JC. Eutrophication: impacts of excess nutrientinputs on freshwater, marine, and terrestrial ecosystems. Environ Pollut1999;100(1):179e96.

[61] International technology roadmap for photovoltaic (ITRPV), internationaltechnology roadmap for photovoltaic results 2017. 2018.

[62] Randolph MA. Commercial assessment of roll to roll manufacturing of elec-tronic displays. Massachusetts Institute of Technology. Masters Thesis. Mas-sachusetts Institute of Technology, Dept. of Materials Science andEngineering; 2006.

[63] Smith KS, Huyck HL. An overview of the abundance, relative mobility,bioavailability, and human toxicity of metals. The environmental geochem-istry of mineral deposits 1999;6:29e70.

[64] United States Environmental Protection Agency. Electricity delivery and itsenvironmental impacts. 2018. Available at: https://www.epa.gov/energy/electricity-delivery-and-its-environmental-impacts. Acessed on: 18/02/19.