Potential for photogenerated current for silicon based photovoltaic modules in the Atacama Desert

Solar Energy 144 (2017) 580–593

Contents lists available at ScienceDirect

Solar Energy journal homepage: www.elsevier.com/locate/solener

Potential for photogenerated current for silicon based photovoltaic modules in the Atacama Desert Pablo Ferrada a,b,⇑, Aitor Marzo a,b, Enrique Cabrera c, Haifeng Chu c, Valeria del Campo c,d, Jorge Rabanal a,c, Daniel Diaz-Almeida a, Andreas Schneider c, Radovan Kopecek c a

Centro de Desarrollo Energético Antofagasta (CDEA), Av. Angamos 601, Antofagasta, Chile Solar Energy Research Center (SERC Chile), Av. Tupper 2007, 4th Floor, Santiago, Chile International Solar Energy Research Center, Rudolf Diesel-Str. 15, Konstanz, Germany d Departamento de Física, Universidad Técnica Federico Santa María, Av. España 1680, Valparaíso, Chile b c

a r t i c l e

i n f o

Article history: Received 25 August 2016 Received in revised form 10 January 2017 Accepted 23 January 2017

Keywords: c-Si solar cells PV glass Encapsulants UV cutoff Quantum efficiency Solar spectral irradiance

a b s t r a c t In order to evaluate module materials, the maximum theoretical value of the photo-generated current density, was calculated. The calculation was performed for four different solar cells, a standard p-type, passivated emitter and rear contact (PERC), bifacial cell and interdigitated back contact (IBC) considering a solar spectrum of Atacama Desert, the transmittance of several glass-encapsulant-glass structures and quantum efficiency. Regarding the solar spectrum in Atacama, an average air mass (AM) at noon for this location averaged 1.17 and the photovoltaic (PV) modules tilt angle was 20°. When studying the impact of using glass and encapsulants combined with a solar cell under the same solar spectrum, ethylene vinyl acetate (EVA) with low ultraviolet (UV) cutoff led to the higher current density values, up to 2% higher with the IBC solar cell compared to the other solar cells. The highest current gain, when studying the impact of the two spectra in the 300–1200 nm wavelength range, was 7.4% for the IBC solar cell, obtained with a standard 3.2 mm glass, a thermoplastic material (TM) as encapsulant. Considering the UV part of the spectrum, the current gain was maximized with a glass with an anti reflection coating (ARC) combined with the TM encapsulant for the IBC solar cell (25%). A quantification of losses due to reflection in the glass and absorption in the encapsulant revealed that the glass with ARC and the TM encapsulant different than EVA led to the lowest reflection and absorption losses. In this case, the reflection and absorption came down to 4.8% and 0.9%, respectively, contrasting with the 7% and 2.8% loss produced with the standard glass and EVA encapsulants. Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction The photovoltaic (PV) market is widely dominated by crystalline silicon (c-Si) based PV modules with a 93% of total production in 2015. With this regard, 69% of the c-Si technology corresponds to multicrystalline silicon (mc-Si). According to the Fraunhofer Institute for Solar Energy Systems (ISE, 2016), the worldwide cumulative PV capacity reached 242 GWp in 2015 where the Learning Curve shows a module price reduction in the last 35 years of about 19% when doubling the cumulative module production. It is pointed out that one of the important drivers for cost reductions are the technological improvements together with

⇑ Corresponding author at: Centro de Desarrollo Energético Antofagasta (CDEA), Av. Angamos 601, Antofagasta, Chile. E-mail address: [email protected] (P. Ferrada). http://dx.doi.org/10.1016/j.solener.2017.01.053 0038-092X/Ó 2017 Elsevier Ltd. All rights reserved.

the economies of scale. In this way, it is expectable to achieve price reductions expressed in the learning curve. The examination of the solar panel final price divided in different production areas in early 2016 shows that the module represents the highest part with a 41% of the cost, followed by the solar cell and wafer with a 23% each and the poly silicon with a 12% (ITRPV). Consequently, to reduce prices, material costs must be reduced and performance increased. Thus, if module materials and their performance are addressed, potential for price reduction can be obtained. While using less material can lead to a lower cost, reducing optical and resistance losses can enhance the performance of PV devices. Among the materials for PV modules such as, PV glass, encapsulant, backsheet, back foils, solder, conductive adhesive, aluminum frames are most of the main components. Reducing the used volume such as the thickness, substituting expensive materials and reducing wastes are ways to diminish production costs (ITRPV). Regarding the performance improvement,

P. Ferrada et al. / Solar Energy 144 (2017) 580–593

increasing light transmission of cover glass by implementing anti reflection coatings (ARC) is one approach. On these lines, the encapsulant and backsheet are other key materials to be considered. Encapsulants must exhibit, as PV glass, a high transmittance, long lifetime to resist environmental and operation conditions. According to the international technology roadmap for photovoltaic (ITRPV), ethylene vinyl acetate (EVA), polyolefin (POE), polydimethyl silicone (PDMS) and polyvinyl butyral (PVB) are the expected encapsulation materials to dominate from 2015 to 2026. Usually, manufacturers provide two warranties. After 10 years in operation, the final power output (P) is 90% of the initial maximum power (Pmpp). And, after 20 years, P is 80% of Pmpp. The reasons for performance degradation are associated to different effects. First, the power drop can be explained by a decrease in the fill factor due to series resistance increase. A second reason is given by the short circuit current Isc due to optical degradation. Optical degradation can be a consequence of a diminution in the transmittance of glass and encapsulant, and of light induced degradation in boron doped silicon (see Section 1.2) (Skoczek et al., 2009). In that investigation, it was reported that after 20 years the maximum power decreased at rate of 0.8%/year for all tested modules, 1%/ year for the modules connected to battery charger and 0.6%/year for the modules at open circuit condition. Another important finding was related to the module configuration. The glass-glass structure showed larger average degradation than glass-polymer. However there was a large deviation as some glass-glass modules performed very well while others did not. This result was explained by the different permeabilities of polymer backsheet materials, glass-glass design allowing interaction of EVA and aluminum, and high solar cell temperature. Nevertheless, as long the energy provided by PV modules satisfies the user, the end of life is not reached. According to (Skoczek et al., 2009), lifetime of PV modules is not, as assumed, limited to 20 years. Regarding solar cell glass-encapsulant structures, it is important to consider the behavior of the incident light since optical losses occur due to the interaction between light and the structure of PV device., For instance, when the encapsulant is silicone rather than EVA (McIntosh et al., 2009), a 0.9–1.6% increase in the module photo generated current density is obtained. This gain is primarily produced due to the transmission of short wavelength light, 300 < k < 420 nm. The gain is even higher when low-absorbing glass and cells with a high internal quantum efficiency (IQE) at these wavelengths are used. Considerable improvements could be obtained by using encapsulants with higher refraction index n (nEVA = 1.5, for conventional modules with glass-EVA encapsulant) (McIntosh et al., 2006). Nevertheless, optical losses are not trivial to assess due to the following reasons. First, the behavior of refractive index and extinction coefficient of silicon, the antireflection coatings (ARCs) and encapsulants varies with wavelength. Second, solar cells are usually textured in such a way that light reflects multiple times from the front surface, among others (Baker-Finch and McIntosh, 2010). The use of softwares offers the possibility to compute optical losses and to determine the optimal thickness of an antireflection coating with or without encapsulation (McIntosh and Baker-Finch, 2012). Photovoltaic modules are designed based on rated data under standard testing conditions (STC) meaning that measurements are performed with 1000 W/m2 solar irradiation, a temperature of 25 °C and reference AM1.5 spectrum (where AM stands for air mass). The STC conditions are an approximation to the noontime close to spring and autumn equinoxes in the United States. However, the real sunlight and environmental conditions where PV systems are installed can strongly differ to those defined by STC. As a consequence, PV modules may not produce the expected maximum power output. Spectral effect must be taken into consideration (Simon and Meyer, 2011). The fact that the solar spectrum

581

varies from site to site throughout each day and year due to changes in the air mass, turbidity, precipitable water, clouds and albedo can cause large variations in the efficiency of PV modules. The highest increase in relative efficiency (0.5–1.5%) for solar cells with a good ‘blue response’ and for silicones encapsulant of a higher refractive index rather than EVA have been studied in McIntosh et al. (2010) (with the spectrum treated as normally incident light). Additionally, significant contributions have been found for short wavelengths in the morning or late in the afternoon (6– 10%) depending on the power distribution of the spectra. The application of luminescent down shifting layers (LDS) in combination with CdS/CdTe solar cells as a function of the solar spectrum irradiance and power distribution has been considered in Alonso-Álvarez et al., 2012. Subsequently, a great motivation is the design of PV modules with adapted module materials to perform best under local conditions. One place, which has experienced a rapid PV grow and shows further potential for PV implementation and technology development, is the Atacama Desert in Chile. The cumulative installed PV capacity in this country already reached 1.1 GW in March 2016 with 2 GW of approved PV projects to be added next (CIFES Report). Many of these large solar projects are conceived in the Atacama Desert. The Chilean desert exhibits environmental conditions in which the solar spectrum can differ from what is usually found in the north hemisphere (Cordero et al., 2016). In fact, this location represents one of the places with highest surface irradiation on Earth. The high mean altitude, a large number of days with clear skies and low absorption ozone and water vapor columns are characteristics determining the solar resource. Global horizontal irradiation (GHI) can surpass 8 kW h/m2 per day resulting in more than 2500 kW h/m2 per year (Escobar et al., 2012). Atacama Desert is located along the Pacific coast in South America between latitudes 20°S and 30°S with a length close to 1000 km and a surface of approximately 105,000 km2. It is characterized as hyper arid with annual precipitations lower than 50 mm (Larraín and Escobar, 2012). Mean temperature values are 10–20 °C in winter and 20–30 °C in summer, with an air temperature below 38 °C (McKay et al., 2003). Regarding chemical composition, nitrates accounting for 28% of the soil and water-soluble salts such as perchlorates and iodides, which rarely exist anywhere else, are found (Navarro-González et al., 2003). Solar cells are encapsulated into modules to ensure long-term environmental stability. Improved module performance can be achieved by ensuring that the light initially reflected by the solar cell is confined by its encapsulant and covering glass. In this sense, the regular array of inverted pyramids geometry as a surface texture of the solar cell provides enhanced antireflection and favorable light trapping characteristics. Such an advantage may have driven its use in record efficiency solar cells (Baker-Finch and McIntosh, 2011). The key metric to quantify optical performance of a PV device is the short-circuit current density Jsc which depends on front surface transmittance, light trapping and the spatial profiles of photogeneration G(f), where f is the shortest optical path, and collection efficiency gc. In this way, the relatively high Jsc attributable to the capacity of the texture to improve front surface transmission, and determining that for certain designs, losses due to recombination depend on the front surface morphology, is determined (Baker-Finch and McIntosh, 2012). 1.1. Goal and approach The main idea of this work is to evaluate the module materials in terms of photo current densities considering the reference and Atacama solar spectra. This analysis is performed for different encapsulated crystalline silicon (c-Si) solar cells. The technologies consisted of (1) a standard p-type mono c-Si, (2) a passivated

582

P. Ferrada et al. / Solar Energy 144 (2017) 580–593

emitter and rear contact (PERC), (3) a bifacial (BiSoN) n-type mono c-Si and (4) an interdigitated back contact (IBC) solar cell. The photo current generation of laminated solar cells was calculated based on the following measured and generated data: (A) The global solar spectrum of the sunlight in the Atacama Desert and the ASTM G173-03 reference AM1.5 spectrum, (B) the transmittance of glass-encapsulant-glass structures, (C) quantum efficiency (QE) of solar cells. For (A), the simple model of the atmospheric radiative transfer of sunshine (SMARTS) v. 2.9.2 code was used to calculate a global tilted spectral irradiance (GTIk) in the Atacama Desert. For (B), a set of PV glass types of thicknesses 1.5 mm and 3.2 mm was used in combination with three different encapsulants, namely, ethylene vinyl acetate (EVA), a thermoplastic material (TM) and a low UV light cut-off EVA (EVA-U). Thus, the transmittance and reflectance of glass-encapsulant-glass were measured with a spectrophotometer. For (C), the solar cells of type (1) and (2) were fabricated and measured indoor at the International Solar Energy Research Center (ISC Konstanz). Data of solar cells of type (3) and (4) were simulated from PVlighthouse (https://www.pvlighthouse.com.au). Integrating the product of the spectral irradiance, the transmittance, the external quantum efficiency (EQE) and considering physical constants, it is possible to compute the maximum theoretical current density to be generated by each PV device. Likewise, losses due to reflection on the PV glass and absorption in the encapsulant can be computed by this integration. The details of the methodology are described in Section 3. In the following, a description of the solar cell concepts relevant for this work is described.

1.2. Solar cell concepts used in this work The standard p-type and the bifacial n-type mono c-Si solar cells are shown in Figs. 1 and 3 as they were fabricated and measured for this work. The IBC and PERC solar cell structures, for which data were generated from simulation, are shown in Figs. 2 and 4. The p-type substrates have been the main base material for more than 30 years. However, the contribution of n-type based material is expected to be above 30% by 2026 (ITRPV). One of the key features of n-type material is that it is more tolerant to metallic impurities than p-type Si (Macdonald and Geerligs, 2004). Additionally, the absence of boron (B) and oxygen (O) in n-type Si prevents the formation of B-O complexes. Consequently, light induced degradation (LID) can be avoided due to the higher minority carrier lifetime associated to the reduced B-O concentration in n-type Si compared to p-type material (Hallam et al., 2015). Solar cell concepts such as IBC can take advantage of n-type Si yielding highest efficiencies. Concerning the solar cell structure, a simple classification based on the electrical contacts can be made. Two-side contacted and one-side contacted solar cells. While standard p-type Si, bifacial and PERC solar cells belong to the first group, IBC represents the second category. This classification is important due to the approaches for interconnection of the cells in the module, which does not depend on the substrate base doping. Usually, soldering of ribbons is used for two sided contacted solar cells. Meanwhile, improvements for the interconnection have been reported in Geipel and Eitner (2013) in which gluing of electrical conductive adhesives (ECA) are used, multiwire technologies (Ballif et al., 2015; Lorenz et al., 2016) and the new industrial solar cell encapsulation (NICE) module concept (no encapsulant) Dupuis et al., 2012. A review on the materials and device physics for high efficiency silicon solar cells is found in Xiao and Xu (2014).

1.2.1. Standard p-type solar cell A typical industrial solar cell based on p-type silicon (Si) wafers is shown in Fig. 1. For this work, this solar cell and its performance are considered as the reference. With the same technology but some modifications, it is possible to improve front and rear sides. One possibility is the implementation of selective emitters and locally diffused aluminum back surface field (Al-BSF) to fabricate a PERC solar cell. In Lee et al. (2015) the fabrication steps for c-Si solar cells is given. 1.2.2. Passivated emitter and rear contact (PERC) solar cell Compared to the standard p-type solar cell, PERC concept incorporates a rear side passivation (Urrejola et al., 2010) with Al2O3 for example (Dingemans and Kessels, 2012). Producing local openings by laser beam at the rear side, a local aluminum alloying is achieved. Thus, the BSF and contacts are created as shown in Fig. 2. In the end, recombination and optical properties are improved. 1.2.3. Bifacial solar cell A candidate solar cell structure for industrial implementation, starting from the p-type PERC, can be the n-type PERT (passivated emitter rear totally diffused) (Zhao et al., 2003; Richter et al., 2011). The homogenous boron diffusion to create a p-type emitter and phosphorus to form the p-type BSF enables an open rear side grid to achieve a bifacial solar cell as shown in Fig. 3. 1.2.4. Interdigitated back contact (IBC) solar cell This solar cell exhibits all metal contacts on the rear side as Fig. 4 illustrates. The front side can be fully optimized for high photogenerated current without facing any challenge due to metallization at the sunny side. Nevertheless, the rear side is characterized by complex processing to achieve doped regions and metallization. As p-type and n-type doped regions are close to each other, metallization becomes critical due to the possibility of shunting. Three different solar cell processes to achieve back contacted solar cells are described in Woehl et al. (2013). At industrial level, SunPower

Fig. 1. Standard p-type Si solar cell. This solar cell receives a homogeneous phosphorus (P) diffusion for p-n junction (n++ means a doping in the order of 1020 P atoms/cm
3), a plasma enhanced chemical vapor deposition (PECVD) to create a silicon nitride (SiNx) passivation as well as antireflection coating of around 70 nm, a screen-printing metallization of silver (Ag) and aluminum (Al) containing pastes and a firing step. An aluminum-silicon (Al-Si) alloy and a p+ layer (the Al-BSF) are formed during firing covering the full back contact area. Note that Ag busbars are not included in the figure.

Fig. 2. Passivated emitter and rear contact (PERC) solar cell. This solar cell receives a homogeneous P doping and PECVD SiNx layer at the front side and local contact openings (LCO) via laser beam at the rear side. Since the interaction between Al and Si occurs locally, delimited sharp-dark lines in the Al layer are created. As a result, the Al-Si alloy and the local back surface field (L-BSF) are formed in the LCOs at the back contact area.

583

P. Ferrada et al. / Solar Energy 144 (2017) 580–593

Fig. 3. Bifacial n-type Si solar cell. This solar cell requires a homogeneous P diffusion to create the BSF and a homogeneous B diffusion to create a p-type emitter at the front side. The respective precursors for the diffusions are POCl3 and BBr3. During B diffusion a 15 nm boron silicate glass is formed and used as passivation layer. While the n-type BSF is passivated with SiNx by means of PECVD, the p-type emitter receives a thermal oxidation followed by PECVD deposition of SiNx resulting in a BSG/SiNx stack system. Screen-printing metallization with an Al and Ag containing paste for the front p-type emitter and Ag paste for the rear n-type BSF.

Fig. 4. Interdigitated back contact (IBC) solar cell. The IBC solar cell receives a homogeneous B diffusion for the front side to form a front surface field (FSF). The rear side requires patterned diffusions to form multiple p-n junctions. While the passivation for the front side is performed by PECVD to form a SiO/SiNx stack system, the rear side of the solar cell exhibits also a stuck of SiO2/SiNx. The rear side of the IBC cell is composed of 5 surface regions: undiffused, P-diffused and passivated, P diffused and metallized, B diffused and passivated, and B diffused and metallized. Detailed process steps are found in Franklin et al. (2016).

Corporation produces modules with 23.6% efficient IBC solar cells (Smith et al., 2013). From an industrial point of view, the implementation of the conductive backsheet technology can be a suitable solution (Späth et al., 2008). Modules laminated using conductive backsheets rely on the conductive paths at the internal side of the backsheet. In order to fill the holes in the encapsulant layer, solder paste or conductive adhesives are used to form the contact to the electrical circuit of the backsheet. Developments have been reported in Meßmer et al. (2016) in which the contact between copper backsheet and back contact of the solar cell was formed by low temperature solder paste (LTSP). Cell to module loss (CTM) of 2% relative, expressed in FF, was found.

2. Experimental 2.1. Sample preparation Photovoltaic glass of 3.2 mm and 1.5 mm thickness and size 5  5 cm2 were used. Float-glass as standard (Std), float glass with antireflection coating (ARC) and casted glass (Text) were used for this study. Table 1 indicates the types of glass and thickness of the different samples. The tested encapsulants tested were a 450 lm thick EVA (E), a 450 lm thick thermoplastic material (TM) and 400 lm thick EVA with a low light UV cutoff (U). We combined the PV glasses and

Fig. 5. Sketch of the glass-encapsulant-glass structure used to obtain the transmittance and reflectance as a function of the wavelength. This measurement was used to evaluate the module materials in terms of photo current density generation. (a) Shows a structure composed of a 1.5 mm thick front glass, encapsulant and a 1.5 mm rear glass. (b) Depicts the structure composed of a 3.2 mm thick front glass, encapsulant and a 1.5 mm rear glass (G1 type). The rear glass was always 1.5 mm thick without any featured surface.

encapsulants to achieve glass-encapsulant-glass structures. While the front glass was either 3.2 mm or 1.5 mm thick, the glass of the rear side was always 1.5 mm thick. The rear side glass was used as support plate for preparing the structure in a laminator. The thicknesses provided are before lamination (see Fig. 5). In order to process samples similar to commercial PV modules, we inserted the encapsulant between two PV glass layers and laminated in an industrial module laminator. After evacuation the samples were laminated at temperatures of 145 ± 5 °C to polymerize the encapsulant and bond the glass-encapsulant material together. While the structures were heated, a membrane slightly pressed the different layers to the hot plate in order to increase the temperature of the glass sheet with the encapsulant and produce a weather proofed laminate. The applied temperature for TM was 5 °C higher than for EVA and EVA-U while time and pressure were the same. The total time process was 16 min. Table 1 shows the glass-encapsulant combinations for transmittance and reflectance measurements. 2.2. Measurement methods 2.2.1. Optical measurements We measured the wavelength-dependent transmittance (Tk) as well as the reflectance (Rk) of glass, encapsulants and glassencapsulant structures with a Perkin Elmer 950 Spectrophotometer. The device used is able to determine the diffuse as well as direct light. The size of the illuminated spot ranges from 0.5 cm by 3 cm down to 0.1 cm by 0.1 cm. The wavelength sweeps from 180 nm to 3200 nm. This setup includes a 150 mm integrated sphere with an opening of 20 mm. The measurement angle with respect to the surface normal was h = 0° for Tk and h = 8° for Rk. The measurements were performed in steps of 5 nm. 2.2.2. IV and quantum efficiency measurement We measured the current-voltage (IV) characteristic curves for the standard and bifacial solar cells using a flasher device produced by h.a.l.m. elektronik GmbH. The IV parameters such as open circuit voltage (Voc), short circuit current density (Jsc) and efficiency (g) were extracted from the IV curves.

Table 1 PV glass types: standard, antireflection coating (ARC) and textured; combined with encapsulants: EVA (E), thermoplastic material (TM) and EVA with low UV cutoff (U). Glass name

Glass thickness (mm)

Glass surface

Encapsulant

G1 G2 G3 G4

1.5 3.2 3.2 3.2

Standard Standard ARC Textured

E E E E

TM TM TM TM

U U U

584

P. Ferrada et al. / Solar Energy 144 (2017) 580–593

The quantum efficiency (QE) and reflection of metallized solar cells (standard and bifacial) were determined with a spectral response (SR) IQE-SCAN device from pv-tools GmbH. The relative standard deviation for the quantum efficiency measurement is 0.08% for 280 < k < 410 nm, 0.08% 420 < k < 1000 nm, 0.15% for 1010 < k < 1150 nm and 0.96% for 1160 < k < 1180 nm. The QE represents the fraction of charge carriers collected per photon. If all incident photons are counted, the external quantum efficiency (EQE) is defined. The global EQE, which includes the non-shaded and shaded regions of the solar cell, was considered and used to integrate with the solar spectrum. Thus, the integrated current density (J) could be compared with the Jsc from the IV measurement, as described in Section 3. The SR device scanned the whole area of the solar cell with monochromatic light for 300 < k < 1200 nm in steps of 10 nm. IV and QE data for the PERC and IBC solar cells were simulated from PVlighthouse (PV Lighthouse; Fell et al., 2015). 3. Methodolody 3.1. The photo current In order to obtain the maximum current density (JM), the number of photons nph that can impact on an area A within a time interval Dt is used, Let q be the elementary carrier charge. Eq. (1) shows a starting equation for JM,0, assuming that every photon can create an electron-hole pair:

J M;0 ¼

qnph charge ¼ Dt  A Dt  A

ð1Þ

The value of nph can be determined by dividing the optical energy Ek(k) of the radiation at a given wavelength k with the energy of a single photon Eph(k) = hc/k, and integrating over the spectral range of interest, where h is the Planck’s constant and c is the speed of light. Thus, nph depends on the spectral irradiance Uin(k) given in W/m
2 nm
1, according to Eq. (2):

Z nph ¼

kg k0

Ek ðkÞ  dk ¼ Eph ðkÞ

Z

kg

A  Uin ðkÞ  Dt  dk

k0

hc k

:

ð2Þ

On one side, the photons with less energy than the bandgap (or photons of wavelength larger than kg) cannot be absorbed. This fact is accounted as transmission losses. The value of kg is incorporated as the upper limit for the integral. On the other side, a lower integral limit k0 is associated to the surplus, when the photon energy exceeds the bandgap and is expressed as thermalization loss. Combining (1) and (2), Eq. (3) is obtained:

J m;0 ¼

q hc

Z

kg

k0

Uin ðkÞ  k  dk

ð3Þ

Any PV device consists of several layers such as glass and encapsulant, the solar cell and backside materials. While glass and encapsulant must exhibit a high transmittance, the solar cell must absorb as many photons as possible to create electron-hole pairs. Thus, the fraction of charge carriers collected per photon accounts for the probability of generation. The EQE includes both optical (reflection and transmission) as well as recombination losses. Therefore, Eq. (3) can be rewritten in Eq. (4), including EQE (Fische, 2003), as:

J m;1 ¼

q hc

Z

kg

k0

Uin ðkÞ  EQEðkÞ  k  dk:

If only photons reflected back from the transmitted through the cell are excluded, {1
Rcell(k)
Tcell(k)} must be considered. EQE/(1
Rcell
Tcell) refers to the Internal

ð4Þ solar cell and those the following factor Thus, the quantity Quantum Efficiency

(IQE). The IQE allows studying the recombination effects on the solar cell performance but it is also used for quantifying losses due to reflection on the glass (RG) and absorption in the encapsulant (AENC) in terms of current densities (Fische, 2003; BakerFinch and McIntosh, 2010). Wavelength dependence transmittance T(k) and reflectance R(k) are measured quantities from which the absorptance A(k) can be obtained as A(k) = 1
T(k)
R(k). The expression for Jm, which considers the external quantum efficiency as well as the measured k-dependent transmittance of glassencapsulant structures T(k), is shown in Eq. (5):

Jm ¼

q hc

Z

kg

k0

Uin ðkÞ  TðkÞ  EQEðkÞ  k  dk:

ð5Þ

The losses due to reflection on the glass and due to absorption in the encapsulant can be computed according to Eqs. (6) and (7), respectively (Peters et al., 2014; Baker-Finch and McIntosh, 2010).

J loss;R ¼

q hc

J loss;A ¼

q hc

Z

kg

k0

Z

kg

k0

Uin ðkÞ  RG ðkÞ  IQEðkÞ  k  dk:

ð6Þ

Uin ðkÞ  AEnc ðkÞ  IQEðkÞ  k  dk:

ð7Þ

The calculation of integrals with respect to k (Eqs. (3)–(7)) was performed for the 300 < k < 1200 nm and following spectral ranges: 300 < k < 400 nm for ultraviolet (UV), 400 < k < 800 nm for the visible (VIS) and 800 < k < 1200 nm for infrared (IR) part of the spectrum. 3.2. Global solar spectrum in the Atacama Desert The photogenerated current density calculation was carried out by using the Global Tilted Spectral Irradiance (GTIk) for a specific location in Atacama (the Solar Platform of the Atacama Desert or PSDA). The PSDA is located at coordinates 24°050 1400 S. 69°540 4700 W at 963 m.a.s.l, and belongs to the University of Antofagasta, Chile. Photo current density was calculated also with the reference spectrum under which solar cells and modules are usually rated. The GTIk at the PSDA was obtained through SMARTS v. 2.9.2 (Gueymard, 2001, 1995). This spectrum was compared with the ASTM G173-03 Reference Spectra also derived from SMARTS v. 2.9.2. (Gueymard, 2001, 2002, 2004; ASTM, 2012). The current standard G173-03 was issued by the North American PV industry and the American Society for Testing and Materials (ASTM) and the Research and Development Laboratories of the US government. The standard spectral distribution of solar irradiance at ground level is calculated according to the North American atmospheric and geographic conditions. An air mass (AM) equal to 1.5 was considered, a receiving surface inclined at 37° and mean values of atmospheric conditions from US. The authors want to highlight that the solar spectrum calculation for PSDA was performed without information about atmospheric parameters, which are not available yet for Atacama. According to this, atmospheric parameters were not modified from the reference spectrum but only the astronomical inputs. Note that the AM is the main parameter, which affects strongest solar spectrum along the year (day and hour). The AM is derived from equations related to time and location. 3.3. Fraction of photons The fraction of photons f(k) with wavelength below k can be defined as the ratio of the integrated power density (in W/m2) up to k to the total power density (in W/m2) integrated between 300 nm and 1200 nm as Eq. (8) shows (compare with reference

P. Ferrada et al. / Solar Energy 144 (2017) 580–593

(McIntosh et al., 2010). Thus, f measures the photons with a wavelength less than k as a fraction of all photons with k < 1200 nm.

Rk

Uin ðk0 Þ  dk0

300 f ðkÞ ¼ R 1200 300

Uin ðk0 Þ  dk

;

ð8Þ

Eqs. (9) and (10) provide the calculated averages and StdDev values, where X is the variable T or R. A measurement for G4 is not available.

T av g ¼

where the quantity Uin(k) is the spectral irradiance in W/(m2 nm). 4. Results 4.1. The solar spectrum An AM mean value was calculated at noon along the year for the PSDA. The obtained result was 1.17. A tilted surface of 20° was also considered, which corresponds to the tilt angle of installation for PV modules at this place. Atmospheric parameters were set equal to the values of ASTM G173-003 to estimate the solar spectrum. Fig. 6 shows GTI. Under the point of view of energy production with PV technologies, the curves of the two spectra suggest differences between both, mainly in UV and visible spectral ranges. However, there are not notable discrepancies for near infrared. The UV-B (290 < k < 315 nm) at the PSDA is 2.04 times higher than G173-03 GTIk, 1.27 times higher for UV-A (315 < k < 400 nm), 1.09 times higher for VIS (400 < k < 780 nm) and 1.04 higher for IR (780 < k < 800 nm). 4.2. Transmittance and reflectance of glass and encapsulant samples 4.2.1. Photovoltaic glass types A comparison for 400 < k < 1100 nm between G1, G2 and G3 resulted in an average transmittance and standard deviation (StdDev) of TG1 = 91.0% ± 0.2%, TG2 = 91.0% ± 0.3% and TG3 = 94.4% ± 0.8%, respectively. Likewise, same calculation for the reflectance resulted in RG1 = 8.5% ± 0.4%, RG2 = 8.3% ± 0.4% and RG3 = 5.1% ± 0.8%, respectively. The ARC led to an increase of 4% in transmittance and a reduction of 3% in reflection. The calculation of the average and standard deviation values for the transmittance and reflectance was performed taking the data from the spectrophotometer measurements directly. Considering k1 = 400 nm and kN = 1100 nm and N as the number of data points between these wavelengths,

585

N 1X Tðki Þ N i¼1

StdDev ; T ¼

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 XN ðT i
T av g Þ2 i¼1 N

ð9Þ

ð10Þ

4.2.2. Encapsulants The transmittance of E, TM and U encapsulants showed averaged values of TE = 88.9% ± 0.6%, TTM = 87.3% ± 0.6% and TU = 88.6% ± 0.4%, respectively, for the 400 < k < 1100 nm spectral range. Finally, the reflectances in the same spectral range were RE = 9.9% ± 1.0%, RTM = 12.1% ± 1.5% and RU = 10.1% ± 0.9%. Each encapsulant exhibits a different wavelength from which light is transmitted. These cutoffs are kE = 360 nm, kTM = 305 nm and kU = 330 nm. 4.2.3. Glass-encapsulant-glass structures The calculation of Jm was performed using the transmittance of glass-encapsulant-glass structures. The obtained Jm values are an approximation since light will travel through a glass-encapsulant layer before reaching the solar cell. This approximation is valid within the framework of this work due to the high transmittance of the used materials. The transmittance of glass-encapsulant structures is predictable to be between the transmittance of glass and the transmittance of glass-encapsulant-glass structures. According to the sample configuration, taking the worst case into account, which is a 7.3% difference in transmittance between those of the G3-TM-G1 structure and G3, the maximum absolute error is 3.6%. This error is close to the measured standard deviation, ET = 1.8%, calculated according to Eq. (11):

ET ¼

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi StdDev ðT G3
TM
G1 Þ2 þ StdDev ðT G3 Þ2 þ StdDev ðT TM Þ2 :

ð11Þ

Table 2 summarizes the mean and standard deviation values of the transmittance and reflectance of glass-encapsulant-glass structures for the 400 < k < 1100 nm spectral range. While the combination G3-E-G1 achieved the highest transmittance and lowest reflection due to the ARC, the combination G2-TM-G1 led to the lowest transmittance. In order to differentiate how materials respond to short wavelengths, the spectral transmittance and reflectance curves are required (see Fig. 7a and b). In the case of the reflectance, a similar calculation can be given. The worst case is found when R of G3-TM-G1 is compared to that of E resulting in 6.4%. Thus, the maximum absolute error for R is 3.2% contrasting with the instrumental error ET. Using Eq. (11) for R, ET is equal to 1.9%. 4.3. IV and quantum efficiency 4.3.1. IV measurements The IV-values for each solar cell are shown in Table 3, where Std stands for standard p-type, PERC for passivated emitter and rear contact, Bif-f for bifacial front side, and IBC for the interdigitated back contact solar cell. The PERC, Bifacial and IBC cells are compared with the Std solar cell, defined as reference.

Fig. 6. Global tilted solar spectral irradiance, GTIk for Atacama and reference in the 0 < k < 4000 nm wavelength range. The inset shows the 300 < k < 1200 nm spectral range, which is relevant for c-Si technologies. The differences between both solar spectra are more pronounced in the UV, followed by VIS and IR part. However, bands of water vapor absorption occur for both around 950 nm, 1150 nm and 1400 nm. The maximum intensity is found at 480 nm.

4.3.2. External quantum efficiency The average EQE over the whole wavelength range of interest (300 < k < 1200 nm) resulted in 72.3% for the standard, 80.1% for the PERC, 78.6% for the front side of the bifacial, 66.9% for the rear side of the bifacial and 86.6% for the IBC solar cell (Fig. 8). A

586

P. Ferrada et al. / Solar Energy 144 (2017) 580–593

Table 2 Mean and standard deviation values of the transmittance and reflectance of glass-encapsulant-glass structures. Structure

Mean T (%)

Std. dev. T (%)

Mean R (%)

Std. dev. R (%)

G1-E-G1 G1-TM-G1 G1-U-G1

88.0 86.8 88.2

1.1 0.5 0.9

8.7 8.4 8.4

0.3 0.5 0.4

G2-E-G1 G2-TM-G1 G2-U-G1

87.7 84.5 87.7

1.3 0.8 0.9

8.4 8.5 8.6

0.4 0.5 0.4

G3-E-G1 G3-TM-G1

90.7 87.1

1.8 1.5

5.6 5.7

0.7 0.9

G4-E-G1 G4-TM-G1 G4-E-G1

87.2 84.9 87.3

1.5 0.8 1.1

8.5 8.5 8.4

0.3 0.4 0.4

Fig. 7. Transmittance T(k) and reflectance R(k) as a function of the wavelength k. (a) For G1 and G2 structures. (b): For G3 and G4 structures. Highest average transmittance for the 400 < k <
1100 nm spectral range is 90.7% ± 1.9 obtained with G3-E-G1. The insets show the UV to VIS wavelength range.

Table 3 Open circuit voltage (Voc), short circuit current density (Jsc), fill factor (FF) and efficiency (g) were extracted from current-voltage (IV) measurements of standard p-type (std) cell, passivated emitter and rear contact (PERC) cell, bifacial front side (Bif-f) cell, bifacial rear side (Bif-r) and interdigitated back contact (IBC) cell. The table also shows for each solar cell concept the integrated Jm,1 values using Eq. (4) to calculate how Jsc deviates from Jm,1 (Section 4.4.2). The relative differences DJy, calculated for each solar cell concept according to Eqs. (12)–(15), are summarized to compare the performance of encapsulants where the value above is with AM1.5 and the value below is with AM1.17 (Section 5.2). Sample

Voc (mV)

Jsc (mA/cm2)

FF (%)

g (%)

Jm,1 (mA/cm2)

(Jm,1
Jsc)/Jsc (%)

DJ1,TM (%)

DJ1,U (%)

DJ2,TM (%)

DJ2,U (%)

DJ3,TM (%)

DJ4,TM (%)

DJ4,U (%)

Std

630.4

36.9

77.1

18.0

36.6

0.8


0.2 0

1.6 1.7


0.7
0.5

0.6 0.8


2.9
2.7


1.5
1.3

0.9 1.0

PERC

654.7

39.7

79.4

20.7

39.4

0.7


0.2 0.1

1.6 1.8


0.6
0.4

0.7 0.9


2.8
2.6


1.4
1.2

0.9 1.1

Bif-f

653.1

39.2

78.3

20.0

39.9

0.9


0.1 0.2

1.7 1.8


0.6
0.3

0.7 0.9


2.8
2.5


1.4
1.1

1.0 1.1

Bif-r

649.7

34.6

78.2

17.6

34.3

0.8


0.6
0.4

1.4 1.5


1.1
0.9

0.4 0.5


3.2
3.1


1.9
1.7

0.7 0.8

IBC

703.0

42.0

83.1

24.4

42.9

0.3

0.2 0.6

1.9 2.1


0.3 0.1

0.9 1.2


2.5
2.1


1.1
0.8

1.1 1.3

calculation per spectral range led to the following EQE values. In the UV, EQE averaged 57.9% for the standard, 66.3% for the PERC, 67.3% for the front side of the bifacial, 37.8% for the rear side of the bifacial and 87.1% for the IBC solar cell. Likewise, in the VIS, average EQE values were 89.8% for the standard, 92.1% for the

PERC, 91.6% for the front side of the bifacial, 82.4% for the rear side of the bifacial and 96.7% for the IBC solar cell. Finally, in the IR, EQE resulted in 59.2% for the standard, 72.3% for the PERC, 69% for the front side of the bifacial, 59.5% for the rear side of the bifacial and 76.9% for the IBC solar cell.

P. Ferrada et al. / Solar Energy 144 (2017) 580–593

587

role determining the power gain due to the bifaciality. A further work regarding bifacial PV in the Atacama Desert could involve the actual optical properties of the natural or conditioned ground. The optical losses which can occur at the top layer (glass), top layer (encapsulant), Ag contacts, solar cell top interface, solar cell bulk, solar cell rear contact, external interface, solar cell surrounds, bottom layer (encapsulant), transmitted light, are contained in Jm. The calculated current Jm may be lower compared to that of a module since the light trapping effect is not considered. Our work considers a single solar cell, which is encapsulated. However, a PV module in which many solar cells are interconnected (e.g. 60 solar cells), incident rays can reflect from encapsulant/cell or from encapsulant/backsheet interfaces or from the solar cell metallization. In these cases, the reflection can have a diffuse component leading to total internal reflection (TIR) at the air/glass interface. As a result, there is a probability that light can be re directed to the solar cells.

Fig. 8. External quantum efficiency for the standard (Std), PERC, bifacial front side, bifacial rear side and IBC solar cells. The IBC exhibits the highest EQE values for the 300 < k < 1200 nm spectral range. On the contrary, the standard solar cell shows the lowest EQE from the front side measurements and its sensitivity in the UV as well as in the IR is reduced compared to the EQE of the other solar cell concepts.

4.4. Calculated photo current densities 4.4.1. The maximum photo current density (Jm,0) Using Eq. (3) and integrating between 300 nm and 1200 nm, the theoretical maximum achievable photo current density is 46.0 mA/ cm2 with AM1.5 and 49.5 mA/cm2 with AM1.17. 4.4.2. Verification of condition Jsc = Jm,1 with AM1.5 The Jsc values obtained from the IV measurement can be linked to the integrated photocurrent generated density Jm,1 of solar cells if Jm,1 is calculated under AM1.5. Consequently, the integrated values of Jm,1 were computed using Eq. (4) for the 300 < k < 1200 nm wavelength range without considering the glass-encapsulant structures. In that way an error with respect to the Jsc values can be calculated and used as indicator to validate quantum efficiency measurements. The comparison between Jm,1 and Jsc is also possible since the total EQE was measured as mentioned above. Table 3 shows the integrated Jm values and the deviation defined as (Jm,1
Jsc)/Jsc. This value keeps below 1% validating the condition Jsc = Jm,1. 4.4.3. Photo current for laminated solar cells (Jm) The final calculation was performed taking the transmittance of the glass-encapsulant-glass structures into account. The integral of Eq. (5) was calculated for the 300 < k < 1200 nm wavelength range with the solar ASTM AM1.5 and the AM1.17 of the Atacama location. While Fig. 9a summarizes the Jm values obtained under AM1.5, Fig. 9b shows Jm values under AM1.17. The potential of photo-generated current in the Atacama Desert is directly compared to the generated currents calculated by the ASTM AM1.5 spectrum. Fig. 10 shows the absolute difference between Jm values obtained under AM1.17 and AM1.5 to determine where a gain possible. These results are discussed in Section 5. The determination of Jm values for the rear side of the bifacial solar cell was carried out with the same AM1.5 and AM1.17 spectra to illuminate the front side. In this way, the performance of glass and encapsulants was compared as the rear side quantum efficiency of the bifacial solar cell was used. However, several parameters can be varied for an installation of bifacial modules. The altitude from the ground and inclination angle is some of them. Moreover, the reflective capacity of the ground can play a major

5. Discussion 5.1. Glass type First observations regarding the glass indicate that G1, G2, G3 and G4 transmitt light for k > 300 nm. Therefore, based on the UV light cutoff, glass type is expected to have a less impact than encapsulants on Jm values. It is in the UV regime where the transmittance, reflectance and absorption of glass abruptably change. In terms of numbers, TG1 and TG2 were 80%, however with a standard deviation of 17%, whereas TG3 = 81.6% ± 16%. The reflectance RG1, RG2 and RG3 were 9%, 8.6% and 8.3%, respectively, with standard deviations below 1.2%. Accordingly, the absorptance for G1, G2 and G3 become 11.2%, 11.3% and 10.1% with standard deviations from 17% to 19%. The thickness of the glass is discussed in the following.

5.1.1. Thickness Glass types G1 and G2 differentiate each other only by thickness. However, it was found that TG1  TG2 (Section 4.2.1), while the transmittance of the combination G1-encapsulant-G1 is higher than those of G2-encapsulant-G1. The fact that TG1 is slightly higher than TG2 can be expected since the transmittance of glasses (TG) is high exceeding 91% and thus, the reflectance (RG) and absorptance (AG) of glass remain low. Particularly, the reflectance is a property of the surface and therefore is less dependent on the thickness. Conversely, the value of AG is related with the capacity of absorption of the material and depends on the length of the path the light travels through it. Performing average values of TG, RG and AG in the 300 < k < 1200 nm spectral range and comparing G1 and G2, following relative differences were encountered: (RG1
RG2)/RG1 = 2.35% and (AG1
AG2)/AG1 = 10.5%. These values indicate that changing the thickness of the glass from 1.5 mm to 3.2 mm has greater impact in the absorptance than in the reflectance. When considering the encapsultant between the two glass layers, larger Jm values were obtained with G1-encapsulant-G1 instead of with G2-encapsulant-G1 (each comparison considers the same encapsulant E, TM or U). Within this context, higher Jm were determined when using G1 instead of G2 as front glass, holding for all solar cell types. A quantification of percentage differences according to (Jm,G1
Jm,G2)/Jm,G1 led to 0.4% for E, 0.8% for TM and 1.3% for U encapsulants. These differences were equal for each solar cell concept, both under AM1.5 and AM1.17. It follows that the thickness of the glass had only a small impact on the photo generated current.

588

P. Ferrada et al. / Solar Energy 144 (2017) 580–593

Fig. 9. Jm values for the 300 < k < 1200 nm spectral range, (a) under AM1.5 spectrum and (b) under AM1.17 spectrum. The recurrent structure leading to the maximum Jm value for each solar cell was G3-E-G1, under both spectra.

determined. These results imply that ARC produced a gain in Jm of at least 3% compared to its reference (G2). The AM1.17 produced same Jm gain. 5.2. Encapsulant To compare for each solar cell, the Jm produced with one encapsulant with respect to the Jm obtained with a reference encapsulant (encapsulant E), the same front glass and same solar spectrum were considered. A percentage difference was defined in the form of DJ1,y = (Jm,G1-y-G1
Jm-E-G1)/Jm,G1-E-G1, where y can be TM or U encapsulant. For instance, if G1 is considered and TM is compared with E, DJ1,TM is given by (Jm,G1-TM-G1
Jm,G1-E-G1)/Jm,G1-E-G1. Eqs. (12)–(15) are used in this section. Table 3 contains the results of this section.

Fig. 10. Absolute difference between Jm values for the 300 < k < 1200 nm spectral range obtained under AM1.17 spectrum and AM1.5. The IBC reaches the highest gain (2.68 mA/cm2 with G1-TM-G1, and 2.66 mA/cm2 with G3-TM-G1) followed by the PERC, front side of the bifacial solar cell and the standard solar cell. The rear side of the bifacial solar cell leads to the lowest gain (maximum of 2.15 mA/cm2 with G1 and U encapsulant).

5.1.2. Surface Glass type G3 exhibits an ARC coating and G4 is a textured glass. Results obtained with G3 and G4 can be compared with those obtained with G2 as it has the same thickness and a standard surface. Due to the ARC, Jm with G3 as front glass reached the highest values compared to all other cases. Indeed, calculating a percentage difference as DJm(Gx) = (Jm,G3
Jx)/Jm,G3 where x can be G1, G2 and G4 as front glass, following results were obtained (each comparison considering the same encapsulant) under AM1.5. First, DJm(G1) = 3% for E and DJm(G1) = 0.4% for TM was achieved (note that there is no G3-U-G1, see Table 2). Second, with respect to G2, which can be considered a reference for G3 since it has the same thickness, DJm(G2) = 3.4% for E and DJm(G2) = 1.2% for TM was calculated. Third, DJm(G3) = 4% for E and DJm(G3) = 2.6% was

DJ1;y ¼ ðJm;G1
y
G1
Jm;G1
E
G1 Þ=Jm;G1
E
G1 :

ð12Þ

DJ2;y ¼ ðJm;G2
y
G1
Jm;G2
E
G1 Þ=Jm;G2
E
G1:

ð13Þ

DJ3;y ¼ ðJm;G3
y
G1
Jm;G3
E
G1 Þ=Jm;G3
E
G1:

ð14Þ

DJ4;y ¼ ðJm;G4
y
G1
Jm;G4
E
G1 Þ=Jm;G4
E
G1:

ð15Þ

5.2.1. Standard solar cell The TM encapsulant led to the worst Jm compared to that obtained with E. Thus, the Jm loss under AM1.5 was 2.9% when using G3 as front glass, 1.5% with G4, 0.7% with G2 and 0.2% loss with G1. Note that the transmittance of the TM encapsulant for 400 < k < 1100 nm was the lowest producing the highest loss when combined with G3 (glass with ARC). On the contrary, the higher T of U for 400 < k < 1100 nm correlates with the Jm gain, which was 1.6% when using G1 (T values in Section 4.2.1), 0.9% when using G4 and 0.6% when using G2 as front glass under AM1.5. Note that there is no U encapsulant combined with G3 as front glass (Table 1). 5.2.2. Passivated emitter and rear contact solar cell As for the standard solar cell under AM1.5, TM and U encapsulants led to the similar behavior for the PERC solar cell. Using G1, the TM and U encapsulants exactly led to the same differences

P. Ferrada et al. / Solar Energy 144 (2017) 580–593

obtained with the standard solar cell. However, for G2, G3 and G4 it is observed that the loss with TM was reduced by 0.1% and the gain with U is increased by 0.1%. 5.2.3. Bifacial and IBC solar cells Comparing the bifacial to the standard solar cell under AM1.5, a reduction of the Jm loss with TM of 0.1% and gain with U of 0.1% was obtained for all front glass types. The IBC solar cell under AM1.5 together with the TM encapsulant produced a Jm gain within G1 of 0.2% (instead of the 0.2% loss with the standard solar cell) and reduced the loss within G2, G3 and G4 by 0.3%, 2.5% and 1.1%. The use of U encapsulant led to an increased Jm gain with respect to the values obtained with E of 1.9% for G1, 0.9% for G2, and 1.1% for G4 all as front glass (contrasting with the lower gain values of 1.6%, 0.6% and 0.9% for the standard solar cell, respectively for same glass types). 5.3. The solar spectra and quantum efficiency The effect of solar spectra and quantum efficiency of the solar cells can explain the behavior of Jm values. First, the ASTM AM1.5 and the obtained AM1.17 for the Atacama location can exhibit differences. In fact, when calculating the fraction of photons f(k) with wavelength below k with Eq. (8), both spectra differentiate more from each other as k increases. The value of f for a given k within 300–1200 nm means computing the ratio of the power density integrated between 300 nm and k to the total power density of the 300–1200 nm spectral range. Fig. 11 shows values of f at 350 nm are 1.7% and 1.8% for AM1.5 and AM1.17 respectively. At 400 nm, 800 nm and 1100 nm, f values are 5.7%, 64% and 88.3% under AM1.5, and 6.2%, 70.3% and 96.2% under AM1.17. In McIntosh et al. (2010) the influence of solar spectra on optical performance of encapsulants is reported. With regard to the quantum efficiency, the average values for the 300 < k < 400 nm, 400 < k < 800 nm and 800 < k < 1200 wavelength ranges are highly dependent on the solar cell concept. The IBC solar cell is the most sensitive to UV with an EQE of 87% followed by that of the bifacial cell with an EQE of 67%, the PERC with an EQE of 66% and the standard solar cell with an EQE of 58%. These differences explain the possibility or impossibility to take advantage of using an encapsulant with a low UV cutoff such as TM or U. The overall EQE in the 300 < k < 1200 nm range resulted in aver-

Fig. 11. Fraction of photons f(k) of less than 1200 nm with wavelength below k for AM1.5and AM1.17. The inset shows f(k) for 310 < k < 400 nm. As k increases starting from 300 nm, solar spectra differentiate more from each other.

589

age values of 87% for IBC, 80% for the PERC, 79% for the bifacial and 72% for the standard solar cell. The same order is found for the VIS and IR spectral ranges. As a result, the gain values discussed in Section 5.2 are in correspondence with the differences in the solar spectra and solar cell technology. The Jm loss is reduced and Jm gain is increased as more sensitiveness is incorporated in the UV range together with a better overall efficiency. The change of the solar spectrum from AM1.5 to AM1.17 means that more UV photons will be available. This light can be conveniently used when more UV light is transmitted by the glass-encapsulant structure together with a higher EQE. This effect explains the behavior in the loss/gain analysis of Section 5.2. These statements can be supported by the results shown in Fig. 10. The IBC solar cell enables the highest absolute gain in Jm with at least 2.7 mA/cm2 when G1-TM-G1, G1-U-G1 and G3-TMG1 are used. It means that 1.5 mm glass with standard surface combined with TM or U encapsulant, and the 3.2 mm glass with ARC combined with TM encapsulant were optimal for this solar cell concept. The PERC and the front side of the bifacial solar cell exhibit similar sensitivity in the UV and VIS spectral ranges (only slightly better for PERC). The main difference in the performance between them is in the IR, where PERC is 3.3% more efficient than the front side of the bifacial solar cell. Thus, PERC results in Jm gain values of al least 2.46 mA/cm2 up to 2.50 mA/cm2, while the bifacial (front) reaches gain values up to 2.46, both with the recurrent G1-TM-G1, G1-U-G1 and G3-TM-G1 combinations. 5.4. The encapsulants in the UV Considering Eq. (5) with integral limits of 300 nm and 400 nm, Jm values obtained for glass types G1–G4 combined with the TM encapsulant were between 2 and 2.1 times higher than those reached with the E encapsulant. Likewise, Jm values with glass types 1–4 combined with the U encapsulant were 1.7 times higher with respect to those with E encapsulant. These values are valid for the standard, PERC and bifacial solar cells. For the IBC solar cell, glass types 1–4 combined with the TM encapsulant produced a 2.1–2.2 times higher Jm compared to those with E encapsulant. Similarly, the U encapsulant led to a 1.8 times higher Jm compared to those with the E encapsulant. These differences resulted to be the same under AM1.5 and AM1.17. Table 4 shows the absolute Jm values for detailed comparison. A distribution of the spectral contribution per spectral range to the Jm values is shown in Fig. 12 for the standard solar cell and in Fig. 13 for the IBC solar cell, both under AM1.5 (Figs. 12a and 13a) and AM1.17 (Figs. 12b and 13b). The contribution to Jm is defined as the ratio of the photogenerated current density obtained with Eq. (5) (Jm) integrated within the UV, VIS and IR limits to Jm,0 obtained with Eq. (3). Since the performance of the PERC and bifacial solar cell is in between of that of the standard and IBC solar cells, only these two cases are discussed (Std and IBC). The immediate result, which holds for both solar cell concepts under both solar spectra, is that TM encapsulant produced the highest contribution to Jm from UV followed by the U and the E encapsulants. Indeed, the TM material led to the best performance in that range with G1 for the IBC solar cell, with a contribution to Jm up to 3.1% (AM1.5) and 3.6% (AM1.17). Likewise, TM led up to 2.4% (AM1.5) and 2.8% (AM1.17) UV contribution for the standard solar cell. The difference between solar cell quantum efficiencies, shown in Fig. 8, explains the solar cell behavior. EQE of IBC solar cells is notably higher in the UV range, compared to the other solar cells. A comparison of encapsulants’ performance under AM1.5 with G1 as front glass shows that the TM contribution from UV to Jm was 1.6% and 0.6% higher than EVA-U and EVA-E encapsulants, respectively, for the IBC solar cell. These differences are less pronounced with the standard solar cell, as TM contribution from

590

P. Ferrada et al. / Solar Energy 144 (2017) 580–593

Table 4 Integrated Jm values in the UV wavelength range (300 < k < 400 nm) for all glass-encapsulant combinations and solar cell concepts using AM1.5 and AM1.17. Jm,UV (mA/cm2) AM1.5

G1-E-G1 G1-TM-G1 G1-U-G1 G2-E-G1 G2-TM-G1 G2-U-G1 G3-E-G1 G3-TM-G1 G4-E-G1 G4-TM-G1 G4-U-G1

AM1.17

Std

PERC

Bif-f

Bif-r

IBC

Std

PERC

Bif-f

Bif-r

IBC

0.38 0.75 0.63 0.36 0.74 0.61 0.36 0.71 0.36 0.70 0.60

0.43 0.86 0.72 0.41 0.85 0.70 0.40 0.81 0.41 0.80 0.69

0.43 0.88 0.73 0.41 0.86 0.71 0.41 0.82 0.41 0.82 0.70

0.28 0.53 0.46 0.27 0.52 0.45 0.27 0.50 0.27 0.50 0.44

0.52 1.12 0.92 0.49 1.10 0.89 0.49 1.04 0.49 1.04 0.87

0.46 0.94 0.78 0.44 0.93 0.75 0.43 0.88 0.43 0.88 0.74

0.52 1.08 0.88 0.49 1.06 0.86 0.49 1.01 0.49 1.00 0.84

0.52 1.10 0.90 0.50 1.08 0.88 0.50 1.03 0.50 1.02 0.86

0.34 0.66 0.56 0.33 0.65 0.54 0.33 0.62 0.33 0.62 0.53

0.63 1.39 1.13 0.60 1.37 1.09 0.59 1.31 0.60 1.30 1.07

Fig. 12. Contribution per spectral range to Jm of the standard solar cell, (a) under AM1.5 and (b) under AM1.17. In the UV, the TM encapsulant produced more than 2% contribution to Jm. The G1-E-G1, G2-E-G1, G3-E-G1 and G4-E-G1 combinations resulted in the lowest contributions to Jm (1.2%, 1.1%, 1.1% and 1.1%, respectively). The infrared part of the spectrum produce a 34% and 35% contribution to Jm independent regardless the glass-encapsulant-combination.

Fig. 13. Contribution per spectral range to Jm of the IBC solar cell, (a) under AM1.5 and (b) under AM1.17. In the UV, the TM encapsulant led to the highest contribution to Jm compared to EVA encapsulants (up to 3.1%). The infrared part of the spectrum produce a contribution to Jm of 38% with AM1.5 and 37% with AM1.17 regardless the glassencapsulant-combination. However, due to the better quantum efficiency of the IBC with respect to the QE of the standard solar cell, the contribution from IR is higher with IBC.

UV was 1.2% and 0.4% larger compared to the E and U encapsulant materials. If AM1.17 spectrum is used, then these percentages become 2% (TM compared to E) and 0.8% (TM compared to U) for IBC and 1.4% (TM compared to E) and 0.5% (TM compared to U) for the standard solar cell. This result is explained by the differences in the quantum efficiencies (in Fig. 8) and spectra (see fraction of photons in Fig. 11). After performing the same analysis for G2, G3 and G4, a similar behavior was observed regarding the contribution to Jm from UV.

For instance, for G2 and the IBC solar cell, TM produced 1.7% and 0.6% more contribution from UV than E and U, respectively, under AM1.5 (2% and 0.8% under AM1.17). 5.5. Losses due to reflection and absorption The losses in terms of photo current density due to the reflection (Jloss,R) on the PV glass and absorption (Jloss,A) in the encapsulant were calculated according to Eqs. (6) and (7) respectively.

P. Ferrada et al. / Solar Energy 144 (2017) 580–593

The IQE used was obtained by scaling the EQE to the measured Jsc and considering the solar cell reflection (Rcell) as IQE = EQE/ (1
Rcell). 5.5.1. Under reference AM1.5 spectrum Fig. 14 shows that glass types G1 and G2 led to losses due to reflection up to 3.2 mA/cm2 with the standard solar cell, and up to 3.7 mA/cm2 with PERC, bifacial (front) and IBC solar cells. Glass type G3, which is featured with an ARC, led to the lowest Jloss,R = 1.9 mA/cm2 with the standard solar cell and 2.2 mA/cm2 with the other solar cells (maximum Jloss,R difference of 0.2 mA/cm2). In terms of percentages related to the maximum photo current density (Jm,0) obtained under AM1.5, G1 and G2 led to losses due to reflection up to 7% with the standard solar cell and up to 8% with the more advanced solar cell concepts. Glass type G3 showed the lowest loss up to 4.1% with the standard and up to 4.8% with the other concepts due to the ARC. It can be established that compared to a standard glass surface, the ARC could reduce Jloss,R by 2.8% for the standard solar cell and by 3.2% for the other solar cell concepts. With regard to the absorption in the glass expressed by Jloss,A, less than 0.2 mA/cm2 under AM1.5 and less than 0.3 mA/cm2 under AM1.17 spectrum for all solar cell concepts were calculated. These values correspond to 0.4% and 0.5% of Jm,0 under these solar spectra, respectively. Encapsulant E led to the highest loss due to absorption resulting in Jloss,A = 1 mA/cm2 with the standard solar cell and 1.3 mA/cm2 with PERC, bifacial (front) and IBC solar cells (no significant different between the these cell concepts). On the contrary, the TM led to the lowest loss with a maximum of 0.3 mA/cm2 (a slight difference of 0.1 mA/cm2 between Jloss,A values, was observed when comparing solar cell concepts). In terms of percentages related to Jm,0 obtained under AM1.5, TM led to the lowest losses with a maximum of 0.6% with the standard solar cell and 0.8%, 0.9% and 0.7% with the PERC, bifacial and IBC solar cells. The E encapsulant produced the highest losses up to 2.1% with the standard solar cell and equally for the other solar cell concepts up to 2.8%. 5.5.2. Under AM1.17 spectrum The Jloss,R values under AM1.17 were up to 0.3 mA/cm2 above those under AM1.5 spectrum (Fig. 14a), keeping similar behavior as Fig 14b shows. In terms of percentages with regard to Jm,0 (obtained under AM1.17), G1 and G2 led 7% loss with the standard solar cell and 8% loss with the more advanced solar cell concepts, as it was under AM1.5. Glass G3 produced the lowest reflection loss, with 4.2% loss calculated with the standard solar cell and up

591

to 4.8% loss with the more advanced concepts. Regarding Jloss,A under AM1.17 spectrum, the TM encapsulant also led to the lowest absorption loss (0.1 mA/cm2 higher than the absorption loss obtained under AM1.5). The U and E encapsulant resulted in a loss increment of 0.1 mA/cm2 and 0.2 mA/cm2, respectively, compared to the losses obtained under AM1.5 with the standard solar cell. While for the more advanced solar cells the loss increment was of 0.2 mA/cm2 and 0.3 mA/cm2 for U and E encapsulant, respectively. 5.6. UV degradation The impact of the UV part of the solar spectrum can have detrimental long-term effects on the optical response of encapsulants. In fact, photons with energy in the UV wavelength range produce degradation of the encapsulant material. Moreover, the combined influence of temperature, humidity and irradiation levels specifically in the UV range can lead to corrosion, discoloration, and later, to delamination of the PV module. In other words, encapsulant materials need to be resistant to temperature and UV photons. To guarantee long-term stability a variety of formulations for encapsulants (e.g. EVA, POE, silicone) exist on the market. The formulation contains UV absorbers to varying degrees on one hand preventing a discoloration of the material but on the other hand preventing high transmittance in the short wavelength range. A recent interlaboratory study (Miller et al., 2015) reported that encapsulation degradation often results from interaction incompatibilities between formulation additives. It was found that some encapsulant formulations led to a loss in transmittance of up to 3.4% whereas other formulations show a 1% gain if exposed in a chamber with an UVA-340 fluorescent lamp (or a 5.1% transmittance loss in a Ci5000 system with a Xe lamp). After an in-depth evaluation of encapsulant formulations EVA-A to EVA-E and thermoplastic polyurethane (TPU), it came out that the change in optical performance of each EVA formulation (discoloration) depends primarily on additives such as Lupersol 101 (curing agent), Naugard P (antioxidant preventing discoloration), Tinuvin 770 (absorber), or Tinuvin 123 (absorber), instead of the forming resin. The optical degradation after weathering of glass-encapsulantglass samples in Miller et al. (2015) was associated to changes in the concentration of ingredients of the base material. A loss of species due to degradation or migration and the addition of by-products was encountered. To give examples, the reduced transmittance of TPU was due to the formation of chromophore by-product species. For EVA formulation C, the shift in the

Fig. 14. Equivalent losses in terms of photo current density, (a) under AM1.5 and (b) under AM1.17. While the columns with R_G1, R_G2 and R_G3 correspond to the loss due to the reflection in the glass types G1, G2 and G3, the columns with A_E, A_TM and A_U stand for the loss due to the absorption in the E, TM and U encapsulant.

592

P. Ferrada et al. / Solar Energy 144 (2017) 580–593

cutoff-wavelength to lower wavelengths was explained by a loss of the UV absorber due to a degradation and/or diffusion. Likewise, the shift in the UV cutoff wavelength to lower wavelengths for EVA formulation E, most likely is a result from the loss of formulation additives. The discoloration (in EVA formulation A) was also attributed to the formation of optically absorbing chromophore species. The increase in transmittance for EVA formulation C and E was explained by the decomposition or loss of volatile species. This interaction was observed between Lupersol 101 (curing agent), Naugard P (antioxidant) and Tinuvin 770 (absorber) Holley and Agro, 1998. The research conducted in this study came to the conclusion that UV absorbers are the ultimate source of discoloration for many of the EVA formulations in the market. The yellowness does not increase when an absorber is not present (EVA formulation E). In Zimmermann (2008), a model was developed to study the impact of the UV on encapsulants. An equation to describe the transmittance Tenc(k,E) of the encapsulant as a function of wavelength k and UV dose E was conceived for silicone. This discussion underlines the importance of performing standardized (or even modified new defined) aging tests for PV modules, laminated with a variety of encapsulant formulations. The ultraviolet aging test in IEC 61215 (IEC61215, 2005) consists of a UV preconditioning irradiance test to identify materials that are susceptible to UV photons. It requires illuminattion with a total UV irradiation of 15 kW h/m2 in the (UVA + UVB) regions (280 < k < 400 nm), with at least 5 kW h/m2, i.e. 33% in the UVB region (280 < k < 320 nm), while maintaining the module at 60 °C ± 5 °C. The degradation of Isc due to UV exposure was
0.79% and
0.77% for EVA and polyolefin, respectively (
1.05% and 1.99% relative variation in Pmpp) (López-Escalante et al., 2016).

6. Conclusions The potential with regard to photo current generation (Jm) under an average air mass at noon (AM1.17) of a specific location in Atacama Desert and the ASTM G173-03 AM1.5 reference spectrum was calculated for laminated crystalline silicon solar cell concepts. The photo current was calculated in order to determine the performance of PV glass of 1.5 mm and 3.2 mm thickness with standard, antireflection coating and textured surfaces combined with EVA (E), a thermoplastic material (TM) and an EVA with a low UV cutoff (U). On one hand, the glass types with same flat surface but different thickness showed no variations of final Jm values. On the other hand, glass with ARC led to higher Jm values compared to those with textured and flat surface glass. Encapsulants E, TM and U exhibited a UV light cutoff wavelength k at 360 nm, 305 nm and 330 nm, respectively. The TM encapsulant, averaged a slightly lower transmittance than E (
0.3% difference) but 1.3% higher than U, and higher reflectance (12.1%) for the 400 < k < 1100 nm spectral range. After comparison of Jm obtained with one encapsulant with respect to Jm obtained with another encapsulant for each solar cell, with same glass and same solar spectrum, the main findings can be summarized as follows:  Encapsulant U produced higher Jm (with respect to those Jm values obtained with E and TM) since U combines high light transmission (84.6% for 400 < k < 1100 nm) and a UV cutoff in between of those E and TM (305 nm and 360 nm, respectively).  The Jm loss when comparing encapsulants for the same glass (and same AM) is reduced, as well as the Jm gain is increased as more sensitiveness is incorporated in the UV together with a better overall quantum efficiency.

 When considering only the UV spectral range, i.e. Jm integrated between 300 nm and 400 nm, the TM and U encapsulants produced up to a 2.2 and 1.8 times higher Jm compared to that with E encapsulant (under AM1.17). After comparison Jm obtained with AM1.17 with respect to Jm obtained with AM1.5 for each solar cell with same glass and encapsulant main conclusions are the following:  The impact of the two solar spectra on Jm was 7.4% gain for a 3.2 mm standard glass combined with the TM encapsulant in the IBC solar cell.  The more ability of transmitting more UV photons and using them is clearly seen when observing Jm values integrated between 300 nm and 400 nm under AM1.5, and stronger pronounced for AM1.17.  If only UV is considered, Jm gain reached 25% with ARC glass combined with TM encapsulant for IBC solar cell.

Acknowledgments The authors acknowledge Christopth Peter and Jayaprasad Arumughan for the measurement of the quantum efficiency on the bifacial and standard cells, respectively. The authors acknowledge the support provided in the following frameworks:  FONDECYT/CONICYT Project Numer 3160190.  CONICYT/FONDAP/15110019 ‘‘Solar Energy Research Center” SERC-Chile.  BMBF Solar Colaboration between Chile and Deutschland (Solar ChilD) project number 01DN14005. References Alonso-Álvarez D., Ross, D., McIntosh, K.R., Richards, B.S., 2012. Performance of luminescence down shifting for CdTe solar cells as a function of the incident solar spectrum. In: 38th IEEE PVSC, Austin, Texas, pp. 2504–2508, doi:http:// dx.doi.org/10.1109/PVSC.2012.6318103. ASTM International, 2012. ASTM G173-03 (2012). Standard Tables for Reference Solar Spectral Irradiances: Direct Normal and Hemispherical. Baker-Finch, S.C., McIntosh, K.R., 2010. A freeware program for precise optical analysis of the front surface of a solar cell. In: 35th IEEE PVSC, Honolulu, HI, pp. 2184–2187, doi:http://dx.doi.org/10.1109/PVSC.2010.5616132. Baker-Finch, S.C., McIntosh, K.R., 2011. Reflection of Normally Incident Light from Silicon Solar Cells with Pyramidal Texture. Prog. Photovolt: Res. Appl. 19 (406– 416), 1050. http://dx.doi.org/10.1002/pip. Baker-Finch, S.C., McIntosh, K.R., 2012. One-dimensional photogeneration profiles in silicon solar cells with pyramidal texture. Prog. Photovolt: Res. Appl. 20, 51– 61. http://dx.doi.org/10.1002/pip. 1109. Ballif, C., Descoeudres, A., Despeisse, M., De Wolf, S., 2015. Advances in crystalline silicon heterojunction research and opportunities for low manufacturing costs. In: 42th IEEE PVSC, New Orleans LA, pp. 1–5, doi:http://dx.doi.org/10.1109/ PVSC.2015.7356223. CIFES Report. Renewable Energy in the Chilean Electricity Market. May 2016. Cordero, R., Damiani, A., Seckmeyer, G., Jorquera, J., Caballero, M., Rowe, P., Ferrer, J., Mubarak, R., Carrasco, J., Rondanelli, R., Matus, M., Laroze, D., 2016. The solar spectrum in the Atacama Desert. Sci. Rep. 6 (22457). http://dx.doi.org/10.1038/ srep22457. Dingemans, G., Kessels, W.M.M., 2012. Status and prospects of Al2O3-based surface passivation schemes for silicon solar cells. Vac. Sci. Technol. A 30 (040802). doi: http://dx.doi.org/10.1116/1.4728205. Dupuis, J., Sain-Sernin, E., Nickiporuk, O., Lefilastre. P, Bussery, D., Einhaus, R., 2012. NICE modules technology from the concept to mass production: a 10-year review. In: 38th IEEE PVSC, Austin TX, pp. 3183–3186, doi:http://dx.doi.org/10. 1109/PVSC.2012.6318254. Escobar, R., Cortés, C., Pino, A., Bueno Pereira, E., Ramos Martins, F., Cardemil, J.M., 2012. Solar energy resource assessment in Chile: satellite estimation and ground station measurements. Renew. Energy 2014 (71), 324–332. http://dx. doi.org/10.1016/j.renene.2014.05.013. Fell, A., McIntosh, K.R., Altermatt, P.P., Janssen, G.J.M., Stangl, R., Ho-Baillie, A., Steinkemper, H., Greulich, J., Mueller, M., Min, B., Fong, K.C., Hermle, M., Romijn, I.G., Abbott, M.D., 2015. Input parameters for the simulation of silicon solar cells in 2014. IEEE J. Photovolt. 5 (4), 1250–1263. http://dx.doi.org/10.1109/ JPHOTOV.2015.2430016.

P. Ferrada et al. / Solar Energy 144 (2017) 580–593 Fischer, B., 2003. Loss Analysis of Crystalline Silicon Solar Cells using Photoconductance and Quantum Efficiency Measurements, Ph.D. Thesis. University of Konstanz. Franklin, E., Fong, K., McIntosh, K., Fell, A., Blakers, A., Kho, T., Walter, D., Wang, Da., Zin, N., Stocks, M., Wang, E.C., Grant, N., Wan, Y., Yang, Y., Zhang, X., Feng, Z., Verlinden, P.J., 2016. Design, fabrication and characterisation of a 24.4% efficient interdigitated back contact solar cell. Prog. Photovolt: Res. Appl. 24 (4), 411– 427. doi:http://dx.doi.org/10.1002/pip.2556. Geipel, T., Eitner, U., 2013. Kinetics of electrically conductive adhesives for solar cell interconnection. IEEE J. Photovolt. 3 (4), 1208–1214. http://dx.doi.org/10.1109/ JPHOTOV.2013.2270356. Gueymard, C.A., 1995. SMARTS. A Simple Model of the Atmospheric Radiative Transfer of Sunshine: Algorithms and Performance Assessment. SMARTS. Gueymard, C.A., 2001. Parameterized transmittance model for direct beam and circumsolar spectral irradiance. Sol. Energy 71, 325–346. http://dx.doi.org/ 10.1016/S0038-092X(01)00054-8. Gueymard, C.A., 2004. The sun’s total and spectral irradiance for solar energy applications and solar radiation models. Sol. Energy 76, 423–453. http://dx.doi. org/10.1016/j.solener.2003.08.039. Gueymard, C.A., Meyers, D., Emery, K., 2002. Proposed reference irradiance spectra for solar energy systems testing. Sol. Energy 73, 443–467. http://dx.doi.org/ 10.1016/S0038-092X(03)00005-7. Hallam, B.J., Abbott, M.D., Nampalli, N., Hamer, P.G., Wenham, S.R., 2015. Implications of accelerated B-O complex formation for mitigating LID in Czochralski Silicon. In: 42th IEEE PVSC, New Orleans, LA, pp. 1–5, doi:http:// dx.doi.org/10.1109/PVSC.2015.7355689. Holley, W.W., Agro, S.C., 1998. Advanced EVA-Based Encapsulants: Final report, NREL, Golden, CO, Rep. NREL/SR-520-25296. IEC61215, 2005. Crystalline Silicon Terrestrial Photovoltaic Modules – Design Qualification and Type Approval, Ed. Fraunhofer Institute for Solar Energy Systems, ISE, 2016. Photovoltaics Report June 2016. International Technology Roadmap for Photovoltaic (ITRPV), 7th ed., March 2016. International Technology Roadmap for Photovoltaic (ITRPV), 6th ed., 2016. Larraín, T., Escobar, R., 2012. Net energy analysis for concentrated solar power plants in northern Chile. Renew. Energy 41, 123–133. http://dx.doi.org/10.1016/ j.renene.2011.10.015. Lee, Y., Park, C., Balaji, N., Lee, Y.-J., Ai Dao, V., 2015. High-efficiency silicon solar cells: a review. Isr. J. Chem. 55, 1050–1063. http://dx.doi.org/10.1002/ ijch.201400210. López-Escalante, M.C., Caballero, L.J., Martín, F., Gabás, M., Cuevas, A., RamosBarrado, J.R., 2016. Sol. Energy Mater. Sol. Cells 144, 691–699. http://dx.doi.org/ 10.1016/j.solmat.2015.10.009. Lorenz, A., Gredy, C., Beyer, S., Yao, Y., Papet, P., Ufheil, J., Senne, A., Reinecke, H., Clement, F., 2016. Flexographic printing – towards an advanced front side metallization approach with high throughput and low silver consumption. Sol. Energy Mater. Sol. Cells 157, 550–557. http://dx.doi.org/10.1016/ j.solmat.2016.07.025. Macdonald, D., Geerligs, L.J., 2004. Recombination activity of iron and other transition metal point defects in n- and p-type crystalline silicon. Appl. Phys. Lett. 85 (18), 4061–4063. http://dx.doi.org/10.1063/1.1812833. McIntosh, K.R., Baker-Finch, S.C., 2012. OPAL 2: rapid optical simulation of silicon solar cells. In: 38th IEEE PVSC, Austin, doi:http://dx.doi.org/10.1109/PVSC.2012. 6317616. McIntosh, K.R., Swanson, R.M., Cotterm, J.E., 2006. A simple ray tracer to compute the optical concentration of photovoltaic modules. Prog. Photovolt: Res. Appl. 14, 167–177. http://dx.doi.org/10.1002/pip.647. McIntosh, K.R., Costell, J.N., Cumpston, J.S., Norris, A.W., Powell, N.E., Ketola, B.M., 2009. An optical comparison of silicon and EVA encapsulants for conventional silicon PV modules: a ray-tracing study. In: 35th IEEE PVSC 2010, Honolulu, HI, pp. 544–549, doi:http://dx.doi.org/10.1109/PVSC.2009.5411624.

593

McIntosh, K.R., Costell, J.N., Norris, A.W., Powell, N.E., Ketola, B.M., 2010. An optical comparison of silicon and EVA en capsulants under various spectra. In: 35th IEEE PVSC, Honolulu, HI, pp. 269–274, doi:http://dx.doi.org/10.1109/PVSC.2010. 5615830. McKay, C.P., Friedmann, E.I., Gómez-Silva, B., Cáceres-Villanueva, L., Andersen, D.T., Landheim, R., 2003. Temperature and moisture conditions for life in the extreme arid region of the Atacama Desert: four years of observations including the El Niño of 1997–1998. Astrobiology 3, 393–406. http://dx.doi.org/10.1089/ 153110703769016460. Meßmer, T., Demiralp, F., Halm, A., 2016. Low cost semi automated assembly unit for small size back contact modules and low cost interconnection approach. In: 6th Workshop on Metallization and Interconnection for Crystalline Silicon Solar Cells. Miller, D.C., Annigoni, E., Ballion, A., Bokria, J.G., Bruckman, L.S., Burns, D.M., Chen, X., Elliott, L., Feng, J., French, R.H., Fowler, S., Gu, X., Hacke, P.L., Honeker, C.C., Kempe, M.D., Khonkar, H., Köhl, M., Perret-Aebi, L.E., Phillips, N.H., Scott, K.P., Sculati-Meillaud, F., Shioda, T., Suga, S., Watanabe, S., Wohlgemuth, J.H., 2015. Degradation in PV encapsulation transmittance: an interlaboratory study towards a climate-specific test. In: 42nd IEEE PV SC, New Orleans, Louisiana June 14–19, doi:http://dx.doi.org/10.1109/PVSC.2015.7355607. Navarro-González, R., Rainey, F.A., Molina, P., Bagaley, D.R., Hollen, B.J., De la Rosa, J., Small, A.M., Quinn, R.C., Grunthaner, F.J., Cáceres, L., Gomez-Silva, B., McKay, C. P., 2003. Mars-like soils in the Atacama Desert, Chile and the dry limit of microbial life. Science 302 (5647), 1018–1021. Peters, I.M., Khoo, Y.S., Walsh, T.M., 2014. Detailed current loss analysis for a PV module made with textured multicrystalline silicon wafer solar cells. IEEE J. Photovolt. 4 (2). doi:http://dx.doi.org/10.1109/JPHOTOV.2013.2295736. PV Lighthouse. Retrieved October 23, 2016 . Richter, A., Benick, J., Kalio, A., Seiffe, J., Hörteis, M., Hermle, M., Glunz, S.W., 2011. Towards industrial n-type PERT silicon solar cells: rear passivation and metallization scheme. Energy Proc. 8, 479–486. http://dx.doi.org/10.1016/ j.egypro.2011.06.169. Simon, M., Meyer, E.L., 2011. The effects of spectral evaluation of c-Si modules. Prog. Photovolt: Res. Appl. 19, 1–10. http://dx.doi.org/10.1002/pip.973. Skoczek, A., Sample, T., Dunlop, E.D., 2009. The results of performance measurements of field-aged crystalline silicon photovoltaic modules. Prog. Photovolt: Res. Appl. 17, 227–240. http://dx.doi.org/10.1002/pip.874. Smith, D.D., Cousins, P.J., Masad, A., Waldhauer, A., Westerberg, S., Johnson, M., Tu, X., Dennis, T., Harley, G., Solomon, G., Rim, S., Shepherd, M., Harrington, S., Defensor, M., Leygo, A., Tomada, P., Wu, J., Pass, T., Ann, L., Smith, L., Bergstrom, N., Nicdao, C., Tipones, P., Vicente, D., 2013. Generation III high efficiency lower cost technology: Transition to full scale manufacturing. In: 38th IEEE PVSC, Austin TX, pp. 1594–1597, doi:http://dx.doi.org/10.1109/PVSC.2012.6317899. Späth, M., De Jong, P.C., Bennett, I.J., Visser, T.P., Bakker, J., 2008. A novel module assembly line using back contact solar cells. In: 33rd IEEE PVSC, San Diego, USA, May 11–15, doi:http://dx.doi.org/10.1109/PVSC.2008.4922528. Urrejola, E., Peter, K., Plagwitz, H., Schubert, G., 2010. Al-Si alloy formation in narrow p-type Si contact areas for rear passivated solar cells. J. Appl. Phys. 107 (124516). http://dx.doi.org/10.1063/1.3437070. Woehl, R., Ruediger, M., Biro, D., Wilde, J., 2013. All-screen-printed back-contact backjunction silicon solar cells with aluminum-alloyed emitter and demonstration of interconnection of point-shaped metalized contacts. Prog. Photovolt. Res. Appl. 3 (2), 226–237. http://dx.doi.org/10.1002/pip.2418. Xiao, S., Xu, S., 2014. High-efficiency silicon solar cells—materials and devices physics. Crit. Rev. Solid State Mater. Sci. 39 (4), 277–317. http://dx.doi.org/ 10.1080/10408436.2013.834245. Zhao, J., Schmidt, J., Wang, A., Zhang, G., Richards, B.S., Green, M.A., 2003. Performance instability in n-type PERT silicon solar cells. In: 31st IEEE PVSC, Osaka, Japan, vol. 1, pp. 923–926. Zimmermann, C.G., 2008. Time dependent degradation of photovoltaic modules by ultraviolet light. Appl. Phys. Lett. 92 (241110), 1–3. http://dx.doi.org/10.1063/ 1.2947589.