- Email: [email protected]
Comprehensive analysis and modeling of cell to module (CTM) conversion loss during c-Si Solar Photovoltaic (SPV) module manufacturing
Available online at www.sciencedirect.com
ScienceDirect Solar Energy 130 (2016) 184–192 www.elsevier.com/locate/solener
Comprehensive analysis and modeling of cell to module (CTM) conversion loss during c-Si Solar Photovoltaic (SPV) module manufacturing J.N. Roy School of Energy Science & Engineering & Advance Technology Development Centre, Indian Institute of Technology, Kharagpur 721302, India Received 1 October 2015; received in revised form 12 January 2016; accepted 11 February 2016
Communicated by: Associate Editor Mario A. Medina
Abstract Cell to module (CTM) conversion loss, during Solar Photovoltaic (SPV) module manufacturing, in terms of wattage losses, at critical process steps Tabbing and Stringing (T&S) and Lamination have been analyzed and a comprehensive electrical and optical model presented. The relation between efficiency of the starting cells and CTM loss has been established. The optimization criteria of the T&S process, in terms of ribbon dimensions and the cell parameters, has also been described. CTM conversion loss/gain for lamination process has been modeled using refractive index and thicknesses of various thin film layers on cell with and without lamination. A guideline for selecting these parameters for obtaining optimized efficiency for laminated cells has been presented. The effect of added electrical resistance due to junction box and change of optical property due to anti reflection coating (ARC) on cover glass have also been presented in brief for completeness. Indoor as well as outdoor test data have been used for modules with ARC on cover glass. During outdoor test, measurements have been carried out with varying intensity and angle of incident of the light. T&S and lamination models have been validated by experiments conducted on single cell coupons. The power loss due to junction box and power gain due to ARC on cover glass has been done on full 60 cell modules. The models described here have been successfully used by the author for minimizing CTM conversion loss for two types of cells with known cell process parameters. Ó 2016 Elsevier Ltd. All rights reserved.
Keywords: Photovoltaic; Conversion loss; Optical modeling; Lamination
1. Introduction Crystalline Silicon (c-Si) Solar Photovoltaic (SPV) module manufacturing takes solar cells through a number of process steps. The additional electrical and optical effects introduced during the manufacturing of SPV modules results power loss (or sometime gain) as compared to that of solar cells used to make the module. The difference between input power and the output power is known as cell
E-mail address: [email protected] http://dx.doi.org/10.1016/j.solener.2016.02.020 0038-092X/Ó 2016 Elsevier Ltd. All rights reserved.
to module (CTM) conversion loss which is to be minimized. It is therefore important to understand and model the causes of this loss so that process, materials and design can be accordingly optimized during manufacturing. Improvement of the efficiency of solar cell is to get more power output from same area is a subject of current research. c-Si solar cell technology has made significant progress in last decade. The efficiency improvement is mainly achieved by tackling electrical resistance, External Quantum Efficiency (EQE) Park et al., 2013; Macdonald et al., 2004 and Internal Quantum Efficiency (IQE) Yang
J.N. Roy / Solar Energy 130 (2016) 184–192
et al., 2008; Dimassi et al., 2011. The cell to module (CTM) conversion typically results in loss, which is determined by module manufacturing technology (material and process), and more importantly the type and efficiency of the input cells. It has been seen and described later in this paper that the CTM loss is generally more for high efficiency cells. Although not common, it is possible to achieve a small conversion gain for module made with low efficiency cells. The power output changes due to additional electrical and optical effects introduced by various stages of module manufacturing process, such as Tabbing–Stringing–Bussing (Jung et al., 2014), Lamination (Mickiewitcz et al., 2012) and Junction Box (European committee for electrotechnical standardization, 2012). Reduction of CTM loss occurring due to module manufacturing process has been an important subject. Several innovative techniques have been tried. Enhancement of light capture by reflection (Su et al., 2011; Chung et al., 2012), ARC on Glass (Bunea et al., 2010; Wohlgemuth et al., 2015), conductive adhesive for tabbing and stringing (Hsieh et al., 2010; Zemen et al., 2013), lamination with silicone gel (Poulek et al., 2012) are some of the notable innovations. Several authors reported models of CTM loss attributed to different materials and processes; such as encapsulation (Grunow and Krauter, 2006), resistance introduced due to spot soldering (caballero et al., 2006), mismatch between cells (Louis and Bucciarelli, 1979; Kaushika and Rai, 2007; Webber and Riley, 2013), overall reflection (Scheydecker et al., 1994; Koomen et al., 1996), textured cover glass (Campbell, 1990; Cardona et al., 2008), etc. The effect of temperature on optical loss has also been reported (Lu and Yao, 2007; Krauter and Hanitsh, 1996). More basic models (El-Basit Wafaa et al., 2013) relates illumination and temperature effects to the parameters such as series, shunt resistance and power output. The effect of cell type and efficiency are not addressed by these models. It has been seen by the author that the overall CTM conversion loss has strong relation with the input cell types. In this work, eight different types of cells have been used to develop a comprehensive model to address electrical as well as optical power loss/gain. The cells used are from different manufacturers with varying efficiency, mono and multi crystalline and 2 and 3 bus bars. A guideline for optimum material and process parameters depending on cell types has been presented. The effect of the ARC on cover glass has also been added in this analysis and modeling. It may be noted that the basic process flow and steps used in a SPV module manufacturing have not been disturbed to make this model practical and readily usable. As per the knowledge of the author such comprehensive model and analysis are not available in the literature.
185
collected by making single cell coupons. AAA class solar cell tester from SPIRE Corporation (Spi-Cell TesterTM) www.spiresolar.com has been used to measure powers at various stages of manufacturing; bare cells, tabbing–stinging, bussing and lamination. The power changes due to junction box and ARC on cover glass have been determined by making full size modules of 60 cells (Roy et al., 2010). Experimental data have been collected from single cell coupons made using eight different types of 156 mm 156 mm size cells. The cell description, without disclosing the manufacturer’s name, has been given in Table 1. The cell types chosen are combination of comparatively low (Type-4 and Type-5), moderate (Type-2 and Type-3, Type-6) and high (Type-1, Type-7 and Type-8) efficiency. Fig. 2 shows the configuration of the coupon before and after tabbing–stringing (T&S) and bussing. The same cells are then tabbed–stringed–bussed and measured again. The cells and cell coupons have been tested on a Cell-tester pre and post tabbing–stringing–bussing. Standard tabbing ribbon of 2.4 mm 0.13 mm for 2BB cells and 1.7 mm 0.13 mm for 3BB cells was used. For clarity, the tabbing ribbon for the back bus bar in Fig. 2 is shown a bit wider. The results reported here is the average of five cell coupons of each cell type. LTB of 150 mm and LCB of 20 mm (see Fig. 2b) have been used for making cell coupons. LBB is the property of the cells. The configuration of the 3BB cell coupons is similar. The lamination of the cell coupons was done, keeping the same configuration of Fig. 1, using a Jinchen laminator (www.jinchensolar.com). Lamination temperature of 145 ° C and total cycle time of 18 min have been used. It is not
2. Process flow and experimental details A certified module manufacturing line (Roy et al., 2010) has been used to collect the experimental data. The layup sequence is depicted in Fig. 1. Experimental data has been
Fig. 1. Typical lay-up sequence of a Solar Photovoltaic (SPV) module manufacturing.
186
J.N. Roy / Solar Energy 130 (2016) 184–192
Table 1 Description of the cells used for cell coupons. S. no.
Cell type
1 2 3 4 5 6 7 8
Type-1 Type-2 Type-3 Type-4 Type-5 Type-6 Type-7 Type-8
(Vendor (Vendor (Vendor (Vendor (Vendor (Vendor (Vendor (Vendor
1) 2) 3) 4) 5) 6) 7) 8)
Mono/multi
No. of bus bars
Pmax (WP)
Mono Multi Multi Multi Multi Multi Mono Mono
2BB 2BB 2BB 2BB 3BB 3BB 3BB 3BB
4.028 3.885 3.921 3.638 3.819 3.881 4.257 4.101
Front Contact
LBB
LTB
LCB
Back Contact
(a)
(b)
Fig. 2. (a) Configuration of 2BB bare cell and (b) configuration after Tabbing–Stringing (T&S) and bussing.
practical to determine the effect of junction box effect using cell coupon. The overall junction box related power loss of a module is distributed amongst 60 cells which are connected in series for making a 60 cells module. This effect is therefore determined by measuring power of full 60 cells pre and post junction box fixing. Testing procedure for a module (Roy et al., 2010), using a Spire Sun Simulator (www.spiresolar.com), has been followed. The same methods have been adopted to determine the effect of ARC on cover glass. ARC glass from five different vendors has been used to determine the power gain. Measurements are also done outdoor. Some of the results are also presented in this paper. 3. Experimental results, modeling and discussion – tabbing– stringing (T&S), bussing Short circuit current (ISC) and maximum power (Pmax) of pre and post T&S have been shown in Table 2. In all
cases there is power reduction after T&S. There is a slight decrease in ISC of all cells, which can be explained from shadowing loss. Major loss of CTM conversion primarily occurs due to T&S. This loss can be quantified through the one-diode model (Fig. 3) of a solar cell. Iph is the photo-generated current, Id is current through diode, RS the series resistance and RSh the shunt resistance. The diode has a saturation current I0 and an ideality factor n. The cell voltage (V) and current (I) satisfy Eq. (1) Green, 2014. I ¼ I ph I 0 ½expfðV þ IRS Þ=nV T g 1 ðV þ IRS Þ=RSh
ð1Þ
It is to be noted that the total series resistance applicable in the case of cell coupon are due to inherent bare cell resistance (RSO) and added resistance (Rts) due to T&S and bussing process. In case RSO is the dominating factor, which is applicable for lower efficiency cells, the conversion loss at T&S is lower. It is other-way-around for high efficiency cells. Therefore, for a given T&S process, higher
J.N. Roy / Solar Energy 130 (2016) 184–192
187
Table 2 Electrical data of cell coupons pre and post T&S. S. no.
Cell type
1 2 3 4 5 6 7 8
Type-1 Type-2 Type-3 Type-4 Type-5 Type-6 Type-7 Type-8
(Vendor (Vendor (Vendor (Vendor (Vendor (Vendor (Vendor (Vendor
Pre-T&S
1) 2) 3) 4) 5) 6) 7) 8)
Post-T&S
ISC (A)
Pmax (WP)
ISC (A)
Pmax (WP)
8.622 8.204 8.122 8.035 7.987 8.214 8.690 8.690
4.028 3.885 3.921 3.638 3.819 3.881 4.257 4.101
8.582 8.194 8.111 8.060 7.990 8.207 8.672 8.678
3.785 3.673 3.687 3.451 3.617 3.671 3.961 3.941
efficiency cells have higher CTM conversion loss after T&S. This clear trend can be seen from the results presented in Table 2. It is therefore important to have detail understanding, including the effect of cell properties on the overall CTM conversion loss. In order to quantify the effect of RS on Pmax, Eq. (1) can be used and an expression can be derived relating to copper ribbon parameters for determining conversion loss related to T&S. Calculation of series resistance added due to T&S is not very straight-forward as it is not a simple lumped resistance. A typical ribbon normally makes contact all along the length of the bus bar and hence behaves like a distributed resistance. A lumped equivalent of this distributed resistance can be calculated and used in Eq. (1) to quantify the power loss. Current is collected uniformly along the length of the bus bar ribbon. Current flows through fingers to the nearest bus bar. Once the bus bar collects the current from fingers, the full cell current is collected at the ribbon end. In this configuration, the effective lumped series resistance seen by the cell is given by Eq. (2) (Jung et al., 2014). ð2Þ
where qCu is the resistivity of the copper ribbon; LCu, WCu and tCu are the length, width and thickness of the ribbon respectively. For the gap between cells, direct lumped resistances can be used. For a 2BB cell, the cell-to-cell gap series resistance can be written as: 1=2 qCu LCTC =ðW Cu tCu Þ and for a 3BB cell: 1=3 qCu LCTC =ðW Cu tCu Þ
6.04 5.46 5.97 5.13 5.28 5.40 6.95 6.34
Here, LCTC is the length of the cell-to-cell gap. The resistance of the silver bus bar has been neglected. This actually appears in parallel with the copper ribbon and very thin. The contact resistance has also been neglected. Busing steps connect tabbed strings in series and introduce additional series resistance mainly contributed by bussing leads. Fig. 4 shows current flow in the series connection of two strings. The total series resistance can be split into separate parts arising due to LSS, LBB and LCB. The contribution due to LSS is:
Fig. 3. One-diode model of solar cell.
Rts ¼ 1=3 qCu ðLCu =ðW Cu tCu Þ
CTM loss (%)
qBCu LSS =ðW BCu tBCu Þ due to LCB is: 1=2 qCu LCB =ðW Cu tCu Þ and due to LBB is: 1=2 qBCu LBB =ðW BCu tBCu Þ qBCu is the resistivity, WBCu and tBCu are the width and thickness of the bussing copper. The entire module’s electrical connections can be broken up into one of the aforementioned configurations and the total resistance calculated. This bussing resistance would get equally divided over all the module’s cells as an additional series resistance component. Combining Eqs. (1) and (2) and bussing resistance components, the loss in power after tabbing–stringing and bussing can be quantified. The model was validated on single-cell coupons both pre and post tabbing–stringing as described below: For the tabbed cell configuration shown in Fig. 4, total tabbing–stringing–bussing resistance is depicted by Eq. (3). Rtsb ¼ 1=3 qCu ½LTB =ðW Cu tCu Þ þ qCu ½LCB =ðW Cu tCu Þ þ 1=2 qCu ½LBB =ðW BCu tBCu Þ
ð3Þ
qCu is the resistivity of the ribbon; WCu and tCu are the width and thickness of tabbing ribbon and WBCu and tBCu are the width and thickness of bussing ribbon. LTB is the length of soldered ribbon which is assumed to be the same on the top and back side of the cell. LCB is the gap between the cell edges and bussing ribbon, LBB is the distance between extreme bus bars of the cell. When a bare-cell is measured, its VOC, ISC, Vm and Im can be used to extract the one-diode equivalent circuit parameters (Fig. 3) – Iph, Id, RS and n. Here RSh is assumed to be large and hence
188
J.N. Roy / Solar Energy 130 (2016) 184–192
I/2
LCB
I/2
LBB
I
L
Fig. 4. Current flow in busing copper.
neglected. Once the cell one-diode parameters are known, a theoretical estimate can be made for the tabbing–stringing series resistance using Eqs. (3) or (4). This can then be used to calculate the power loss using Eq. (1). The validation of the model is done from the measured data. I–V data of bare cells (Table 2) is used to extract the one-diode parameters for these cells which are then used to calculate the theoretical change in I–V after tabbing–string ing–bussing. Table 3 shows the extracted parameters. Eq. (3) can be used to determine ribbon series resistance, which turns out to be about 3.4 mX for 3BB cells. Using extracted parameters presented in Table 3 and the calculated addition of series resistance due to tabbing– stringing, the power loss in tabbing–stringing–bussing can be calculated for these single-cell coupons. Comparison between experimental results and calculated values is shown in Fig. 5. There seems to be a good level of agreement between the model and actual calculated values. Similar calculations can be done for a full module and the power loss modeled. However, it is not feasible to do a measurement of 60-series connected stringed cells after T&S and hence it could be difficult to validate the model for a full sized module. The model would however be true irrespective of the number of cells.
4. Experimental results, modeling and discussion – lamination Lamination is another major process step where changes in optical properties of the module lead to changes in power. However, normally there is a power gain due to lamination. The effect due to power loss in T&S–bussing and power gain in lamination typically results net power loss but not as high as due to T&S–bussing alone. Laminated modules have the cells covered by a layer on EVA and glass on the top and a layer of EVA and back sheet on the bottom (Fig. 1). In a solar cell, silicon (refractive index n2) is covered by a thin layer of Anti-Reflection Coating (ARC) of refractive index n1 and thickness d1. When the cell is laminated, two newly added interfaces – Air– Glass–EVA and Glass–ARC change the optical properties of the stack-up, thus changing effective intensity reaching silicon. The EVA–glass composite will be referred to as glass (refractive index n3) in the subsequent discussions.
7 6 5
Table 3 Extracted parameters of one diode equivalent circuit of Fig. 1.
4
S. no.
Cell type
IOP (A)
Id (A)
n
RS (mX)
1 2 3 4 5 6 7 8
Type-1 Type-2 Type-3 Type-4 Type-5 Type-6 Type-7 Type-8
8.622 8.204 8.122 8.035 7.987 8.214 8.690 8.690
6.94E009 5.65E009 2.42E009 9.93E009 4.63E009 4.03E009 2.90E010 8.19E008
1.130 1.138 1.106 1.149 1.019 1.114 1.007 1.307
2.886 3.161 2.925 4.275 3.049 2.921 2.944 1.584
Expt. Theory
3 2 1 0 Type 1
Type 2
Type 3
Type 4
Type 5
Type 6
Type 7
Type 8
(X Axis: Type of Cells, Y Axis: Power Loss)
Fig. 5. Tabbing–Stringing–Bussing power loss in single cell coupons.
J.N. Roy / Solar Energy 130 (2016) 184–192
ISR ¼ qk=hc IQE
ð4Þ
q is the charge of electron, k is the wavelength of light, h is the Planck’s constant and c is the speed of light. The spectrum of light entering silicon is given by Eq. (5):
0.35 0.30
Reflectivity
The effect of the stack-up on incoming radiation can be modeled analytically as a function of wavelength. Current generation, which is determined by Quantum Efficiency (QE), is also a function of wavelength. External Quantum Efficiency (EQE) has been determined first by modeling optical transmission of the stack-up. This gives the spectrum of light entering silicon cell and can be mapped to the generation of current in the cell govern by Internal Quantum Efficiency (IQE). Comparison between an Air– ARC–Silicon stack-up (bare cell) and Glass–Air–ARC–Sili con stack-up (laminated cell) would then give a quantitative estimate of lamination gain/loss. Using standard analytical expression (Green, 2014) the total reflection can be derived for a tri-layer stack-up. An Air–Glass–ARC–Silicon (laminated cell) stack-up, can be modeled as an Air–Glass stack-up in series with a Glass– ARC–Silicon tri-layer. For Air–Glass, Fresnel equations can be used. Typical values for nAir, nGlass and nSi are 1.00, 1.50 and 3.80 respectively. nARC and dARC can be varied typically from 1.80 to 2.20 and from 70 nm to 110 nm respectively. Fig. 6 shows the reflectivity from an Air– ARC–Si stack-up (bare-cell) and an Air–Glass–ARC–Si stack-up (laminated cell). Reflections in case of the laminated cell are lower than the bare cell. This stack-up configuration is thus expected to give a lamination gain. The aim is to have minimum reflection loss of the laminated cell. The value of n and d can be adjusted to vary overall reflection loss. It may be noted that this set of value may not give the minimum reflection loss for the bare cell. The result presented in Fig. 6 is from one of the cell type used and data is available to the author. The lamination gain is the difference between pre and post lamination reflection loss. Bare cell ARC is independently control to get maximum efficiency of the cell. However, the best result are obtained in case the whole stack-up, optical layers of cell and module, are designed together. This has successfully attempted for development of cell and module manufacturing technologies by author. This has been done by determining reflectivity vs wavelength for various refractive index for bare as well as laminated cells. The set of values which give minimum overall reflectivity for the laminated cell are to be selected. This optimum value will ultimately lead to maximum current generation of the laminated cell as described subsequently. The reflectivity determines the External Quantum Efficiency (EQE). Once the External Quantum Efficiency (EQE) is known, Internal Quantum Efficiency (IQE) can be determined by the cell response to portion of the incoming light entering silicon. IQE would be the ratio of number of carriers generated to number of photons entering Silicon. A parameter called Internal Spectral Response (ISR) is defined according to Eq. (4).
189
0.25 0.20 0.15 0.10 0.05 0.00
300
400
500
600
700
800
900
1000
1100
1200
Wavelength (nm) Fig. 6. Reflectivity from Air–ARC–Si (Blue) and Air–Glass–ARC–Si (Green): n1 = 2.13, d1 = 87 nm. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
I Si ðkÞ ¼ I AM1:5G ðkÞ 1 RStack-up ðkÞ
ð5Þ
This spectrum entering Silicon would then generate a current governed by ISR of the cell. JOP, density of optically generated photo-current is given by Eq. (6): Z J OP ¼ I AM1:5 G ðkÞ 1 RStack- up ðkÞ ISRðkÞ ð6Þ The optically generated photo-current of the cell, IOP can be obtained using IOP = Acell JOP, where Acell is the active cell area. Thus, IOP of bare-cells and laminatedcells can be calculated accordingly and the lamination gain can be quantified. Eq. (6) can be used to calculate the current generation for any given stack-up and maximize the JOP of cells after lamination. Fig. 7 shows the generation of current as a function of wavelength for a bare-cell and a laminated-cell. The incident spectrum is AM1.5 G and the ARC has a refractive index of 2.13 and a thickness of 87 nm. It can be seen that the bare-cell has better response than laminated-cell in the wavelength range at around 600–700 nm. However, outside this band the laminated-cell has a better response. The overall lamination gain using this ARC is about 1.9% ˚ ). A more general case is shown in Fig. 8, (7.638–7.785 A which shows IOP variation of a standard 600 bare-cell for ARC refractive index variation from 1.80 to 2.13. This can be repeated for various ARC thicknesses. This graph represents that till about nARC of 2.04, there is a lamination loss. Beyond that value, lamination gain can be obtained. The set of values produce the maximum IOP has to chosen. Fig. 8 shows the optimum possible result. The maximum IOP is obtained at nARC of 2.15 and dARC of 110 nm. Although, it is possible to increase the IOP further by increasing nARC beyond 2.13, the bare cell efficiency will then drops significantly. This is not acceptable for cell manufacturer and therefore not tried. The single-cell coupons used for tabbing–stringing–bus sing experiments have been laminated using standard
190
J.N. Roy / Solar Energy 130 (2016) 184–192 0.8
lamination materials and tested. The results are shown in Table 4. The trend seems to agree with the model. Proper validation can be done once the nARC and dARC for all the cell types used are known. The model has been successfully used by the author, to minimize CTM conversion loss, for two of these cell types (Type-4 & Type-6) with known parameters.
Current Generation (J0 /nm)
0.7 0.6 0.5 0.4
5. Other CTM conversion loss/gain
0.3 0.2 0.1 0.0 300
400
500
600
700
800
900
1000
1100
1200
Wavelength (nm) Fig. 7. Current generation as a function of wavelength: bare cell (Blue) and laminated cell (Red). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) 7.9 7.85 7.8 7.75 7.7 Laminated
7.65
Bare Cell
7.6 7.55 7.5 7.45 7.4 1.75
1.8
1.85
1.9
1.95
2
2.05
2.1
2.15
2.2
(X Axis: nARC, Y Axis: IOP)
Fig. 8. IOP variation with nARC for optimized dARC.
Additional series resistance contributed by the leads and contacts of the junction box introduces additional loss. This comes in series with the cell string having many cells, e.g. 60 or 72 cells for a 60 or 72 cells module. It is found that this resistance introduces 0.1–0.3% CTM conversion loss at module level power; depending on the number of cells in the module, type of junction box used, etc. The CTM conversion loss can be reduced further by using ARC on the cover glass. Experimental results are described in brief to give complete picture on the CTM conversion loss. A detail analysis/modeling is being done and will be reported separately. Five different vendors have been used and the measurements done indoor (www. spiresolar.com) as well as outdoor. The results presented here are average of 5 samples (modules) each and outdoor data taken for 7 sunny days during the month of May in Hyderabad, India. It has been found that CTM conversion gain of 1–2.5% at the module level can be achieved depending on the vendor and test mode. Table 5 gives the indoor test results at STC. All the results presented for ARC on cover glass are percentage difference between the ISC of the module with respective ARC on cover glass and a reference module without any ARC on cover glass under identical outdoor measurement condition.
Table 4 Measured CTM conversion loss/gain. S. no.
Cell type
Power (WP) Pre-T&S
Post T&S
Post lamination
T&S
Lamination
Total
1 2 3 4 5 6 7 8
Type-1 Type-2 Type-3 Type-4 Type-5 Type-6 Type-7 Type-8
4.028 3.885 3.921 3.638 3.819 3.881 4.257 4.101
3.785 3.673 3.687 3.451 3.617 3.671 3.961 3.941
3.879 3.774 3.799 3.558 3.722 3.783 3.870 4.036
6.04 5.46 5.97 5.13 5.28 5.40 6.95 6.34
2.50 2.75 3.05 3.10 2.90 3.05 2.3 2.4
3.54 2.71 2.92 2.03 2.38 2.35 4.65 3.94
Conversion loss (+)/gain () (%)
Table 5 STC power gain with ARC cover glass. S. no.
Type
Average power (WP)
Min. power (WP)
Max. power (WP)
Gain as compared to reference (%)
1 2 3 4 5 6
Reference (No ARC) Vendor-1 Vendor-2 Vendor-3 Vendor-4 Vendor-5
234.13 237.98 237.85 237.58 236.93 236.27
232.58 236.92 235.93 236.29 235.35 234.7
235.52 238.94 239.65 238.7 237.98 237.59
– 1.64 1.59 1.47 1.2 0.91
J.N. Roy / Solar Energy 130 (2016) 184–192
191
Table 6 Gain with outdoor testing of ARC cover glass. Glass type
Average ISC gain as compared to reference (%) 10 AM
10:30 AM
11 AM
11:30 AM
12 PM
12:30 PM
1 PM
1:30 PM
2 PM
3 PM
4 PM
5 PM
Vendor-1 Vendor-2 Vendor-3 Vendor-4 Vendor-5
1.92 2.32 2.33 0.72 1.45
1.69 2.48 2.45 0.69 1.37
1.78 2.19 2.20 0.96 1.3
2.01 2.71 2.2 0.77 1.29
2.16 2.59 2.39 0.61 0.92
1.85 2 2.31 0.8 1.29
1.54 1.38 1.92 0.18 0.61
2.13 2.99 2.65 1.03 1.92
1.31 2.09 2.21 0.02 0.71
1.78 2.68 2.75 0.68 0.5
1.27 2.18 2.98 0.62 0.86
1.16 2.22 2.19 0.50 0.75
Table 7 Gain with outdoor testing of ARC cover glass during early morning. Glass type
Average ISC gain as compared to reference (%) 8 AM
8:30 AM
9 AM
Vendor-1 Vendor-2 Vendor-3 Vendor-4 Vendor-5
2.53 1.09 1.76 0.36 1.15
3.59 2.60 3.54 2.02 2.54
3.44 2.75 3.62 2.09 2.97
Table 8 Average gain with outdoor testing during 10 AM–2 PM. Glass type
Average ISC gain as compared to reference (%)
Vendor-1 Vendor-2 Vendor-3 Vendor-4 Vendor-5
1.82 2.42 2.3 0.64 1.21
The summary of the outdoor testing results is shown in Table 6. A comparative value for early morning, which indicates large gain, is shown in Table 7. Average gain during peak period (10 AM–2 PM) during the day is shown in Table 8. It is clear that overall outdoor power output gain of more than 2% can be achieved by using ARC on cover glass. The output, represented by ISC, is found to vary throughout the day (Table 6) and noticeably higher during early morning (Table 7). This is primarily due to inclined angle for the incident rays and low light behavior of ARC. 6. Conclusion In this paper a detail and comprehensive analysis has been presented for understanding the cell to module (CTM) conversion loss during manufacturing of Solar Photovoltaic (SPV) module. Conversion loss due to Tabbing–Stringing (T&S)–bussing has been modeled and validated by experimental measurements from coupon cells. Eight different cell types have been used for making the coupon modules. It has been shown that the lamination process can give conversion gain under certain conditions. Both T&S–bussing loss and lamination gain/loss depend on the starting cell properties. Higher efficiency cells have higher net conversion loss unless the process, particularly lamination, is optimized for each cell type. This model has been used successfully by the author to minimize the
overall conversion loss for two of the cell types with known cell manufacturing details. There is good agreement between experimental and theoretically calculated values for tabbing–stringing–bussing losses. Optimization of tabbing ribbon dimensions has successfully done by the author for module manufacturing process. The Lamination Optics Model is shown general agreement with the trend of experimental data and used to maximize lamination gain for two of the cell types with known parameters. The effect of ARC on cover glass for further reduction of CTM conversion loss has been presented in brief. Modules with five different ARC glass vendors have been studied. The measurements are done at STC as well outdoor and the results are compared with reference glass having standard glass with no ARC. It has been found that there is a significant difference between STC and outdoor measurements. Output is found to be more during outdoor measurement with significantly more output in the morning hours. Acknowledgement The experimental data used here was from the coupons and modules (Roy et al., 2010) manufactured at Solar Semiconductor Pvt. Ltd, Hyderabad, India. References Bunea, G., Xavier, G., Rose, D., Nelson, L., Peurach, J., 2010. Performance and reliability of modules with antireflective coated glass. In: Proceedings of 25th EUPVSEC, pp. 4103–4106. Caballero, L.J., Sanchez-Friera, P., Lalaguna, P.B., Alonso, J., Vazquez, M.A., 2006. Series resistance modeling of industrial screen printed monocrystalline silicon solar cells and modules including the effect of spot soldering. In: Proceedings of Conference of World Conference on Photovoltaic Energy Conversion, pp. 1388–1391. Campbell, P., 1990. Light trapping in texture solar cells. Sol. Energy Mater. Sol. Cells 21, 165–172. Cardona, M.S., Coronado, A.V., Alvarez, J.L., 2008. An analysis of geometrical shapes for PV modules glass encapsulation. Sol. Energy Mater. Sol. Cells 92, 323–331. Chung, I., Back, U.I., Moon, I.S., Kwon, O., Bae, K., Shin, S., Cho, E.C., Lee, W.J., 2012. Optimization of the output power by effect of backsheet reflectance and spacing between cell strings. In: Proceedings of IEEE Photovoltaic Specialist Conference (PVSC), pp. 2388–2390. Dimassi, W., Derbali, L., Bouaicha, M., Bessais, B., Ezzaouia, H., 2011. Two-dimensional LBIC and internal-quantum efficiency investigation of grooved grain boundaries in multi crystalline silicon solar cells. Sol. Energy 85, 350–355. El-Basit Wafaa, A.B.D., El-Maksood, A.B.D., Soliman, F., El-Moniem Saad, A.b.d., 2013. Mathematical model for photovoltaic cells. Leonardo J. Sci. 23, 13–28.
192
J.N. Roy / Solar Energy 130 (2016) 184–192
European Committee for Electrotechnical Standardization, 2012. EN 50548: Junction Boxes for photovoltaic modules, Brussels. Green, M.A., 2014. Solar Cells: Operating Principles, Technology and System Applications. Prentice-Hall Inc.. Grunow, P., Krauter, S., 2006. Modelling of encapsulation factors for photovoltaic modules. In: Proceedings of World Conference on Photovoltaic Energy Conversion, pp. 2152–2155. Hsieh, H.H., Lee, W.K., Chang, T.C., Hsi, C.S., 2010. Photovoltaic modules of crystalline solar cells using new type of assembly structure. In: Proceedings of International Microsystems Packaging Assembly and Circuits Technology Conference (IMPACT), pp. 1–4. Jung, T., Song, H., Ahn, H., Kang, G., 2014. A mathematical model for cell to module conversion considering mismatching solar cells and the resistance of the interconnection ribbon. Sol. Energy 103, 253–262. Kaushika, N.D., Rai, A.K., 2007. An investigation of mismatch losses in solar photovoltaic cell networks. Energy 32, 755–759. Koomen, E.A.S., Alsema, E.A., Turkenbug, W.C., 1996. A simple model for PV module reflection losses under field condition. Sol. Energy 57, 421–432. Krauter, S., Hanitsh, R., 1996. Actual optical and thermal performance of PV modules. Sol. Energy Mater. Sol. Cells 41–42, 557–574. Louis, L., Bucciarelli, J., 1979. Power loss in photovoltaic array due to mismatch in cell characteristics. Sol. Energy 23, 277–288. Lu, Z.H., Yao, Q., 2007. Energy analysis of silicon solar cell modules based on an optical model for arbitrary layers. Sol. Energy 81, 636– 647. Macdonald, D.H., Cuevas, A., Kerr, M.J., Samundsett, C., Ruby, D., Winderbaun, S., Leo, A., 2004. Texturing industrial multi crystalline solar cells. Sol. Energy 76, 277–283. Mickiewitcz, R.A., Cahill, E., Wu, P.I., 2012. Non-destructive determination of degree of cross linking of EVA solar module encapsulation using DMA shear measurements. In: Proceedings of Photovoltaic Specialist Conference (PVSC), pp. 710–713.
Park, K.M., Lee, M.B., Shin, J.W., Choi, S.Y., 2013. Investigation of surface features using reactive ion etching method for enhanced performance of multi-crystalline silicon solar cells. Sol. Energy 91, 37– 47. Poulek, V., Strebkor, D.S., Persie, I.S., Libra, M., 2012. Towards 50 years lifetime of PV panels laminated with silicone gel technology. Sol. Energy 86, 3103–3108. Roy, J.N., Gariki, G.R., Nagalakhmi, V., 2010. Reference module selection criteria for accurate testing of photovoltaic (PV) panels. Sol. Energy 84, 32–36. Scheydecker, A., Goetzberger, A., Wittwer, V., 1994. Reduction of reflection losses in PV modules by structured surfaces. Sol. Energy 53, 171–176. Su, W.S., Chen, Y.C., Liao, W.H., Huang, C.H., Liu, D.C., Huang, M.Y., Wu, Z.C., Ho, S.J., 2011. Optimization of the output power by effect of backsheet reflectance and spacing between cell strings. In: Proceedings of IEEE Photovoltaic Specialist Conference (PVSC), pp. 3218–3220. Webber, J., Riley, E., 2013. Mismatch loss reduction in photovoltaic array as a result of sorting photovoltaic modules by max-power parameters. Renewable Energy, 1–9. Wohlgemuth, J., Cunnigham, D., Shaner, J., Nguyen, A., Ransome, S., Artigeo, A., 2015. Crystalline silicon photovoltaic modules with antireflective coated glass. In: Proceedings of IEEE Photovoltaic Conference (PVSC), pp. 1015–1018. www.jinchensolar.com. www.spiresolar.com. Yang, W.J., Ma, Z.Q., Tang, X., Feng, C.B., Zhao, W.G., Shi, P.P., 2008. Internal quantum efficiency for solar cells. Sol. Energy 82, 106–110. Zemen, Y., Schulz, S.C., Trommler, H., Buschhorn, S.T., Bauhofer, W., Schulte, K., 2013. Comparison of new conductive adhesive based on silver and carbon nanotubes for solar cells interconnection. Sol. Energy Mater. Sol. Cells 109, 155–159.
ScienceDirect Solar Energy 130 (2016) 184–192 www.elsevier.com/locate/solener
Comprehensive analysis and modeling of cell to module (CTM) conversion loss during c-Si Solar Photovoltaic (SPV) module manufacturing J.N. Roy School of Energy Science & Engineering & Advance Technology Development Centre, Indian Institute of Technology, Kharagpur 721302, India Received 1 October 2015; received in revised form 12 January 2016; accepted 11 February 2016
Communicated by: Associate Editor Mario A. Medina
Abstract Cell to module (CTM) conversion loss, during Solar Photovoltaic (SPV) module manufacturing, in terms of wattage losses, at critical process steps Tabbing and Stringing (T&S) and Lamination have been analyzed and a comprehensive electrical and optical model presented. The relation between efficiency of the starting cells and CTM loss has been established. The optimization criteria of the T&S process, in terms of ribbon dimensions and the cell parameters, has also been described. CTM conversion loss/gain for lamination process has been modeled using refractive index and thicknesses of various thin film layers on cell with and without lamination. A guideline for selecting these parameters for obtaining optimized efficiency for laminated cells has been presented. The effect of added electrical resistance due to junction box and change of optical property due to anti reflection coating (ARC) on cover glass have also been presented in brief for completeness. Indoor as well as outdoor test data have been used for modules with ARC on cover glass. During outdoor test, measurements have been carried out with varying intensity and angle of incident of the light. T&S and lamination models have been validated by experiments conducted on single cell coupons. The power loss due to junction box and power gain due to ARC on cover glass has been done on full 60 cell modules. The models described here have been successfully used by the author for minimizing CTM conversion loss for two types of cells with known cell process parameters. Ó 2016 Elsevier Ltd. All rights reserved.
Keywords: Photovoltaic; Conversion loss; Optical modeling; Lamination
1. Introduction Crystalline Silicon (c-Si) Solar Photovoltaic (SPV) module manufacturing takes solar cells through a number of process steps. The additional electrical and optical effects introduced during the manufacturing of SPV modules results power loss (or sometime gain) as compared to that of solar cells used to make the module. The difference between input power and the output power is known as cell
E-mail address: [email protected] http://dx.doi.org/10.1016/j.solener.2016.02.020 0038-092X/Ó 2016 Elsevier Ltd. All rights reserved.
to module (CTM) conversion loss which is to be minimized. It is therefore important to understand and model the causes of this loss so that process, materials and design can be accordingly optimized during manufacturing. Improvement of the efficiency of solar cell is to get more power output from same area is a subject of current research. c-Si solar cell technology has made significant progress in last decade. The efficiency improvement is mainly achieved by tackling electrical resistance, External Quantum Efficiency (EQE) Park et al., 2013; Macdonald et al., 2004 and Internal Quantum Efficiency (IQE) Yang
J.N. Roy / Solar Energy 130 (2016) 184–192
et al., 2008; Dimassi et al., 2011. The cell to module (CTM) conversion typically results in loss, which is determined by module manufacturing technology (material and process), and more importantly the type and efficiency of the input cells. It has been seen and described later in this paper that the CTM loss is generally more for high efficiency cells. Although not common, it is possible to achieve a small conversion gain for module made with low efficiency cells. The power output changes due to additional electrical and optical effects introduced by various stages of module manufacturing process, such as Tabbing–Stringing–Bussing (Jung et al., 2014), Lamination (Mickiewitcz et al., 2012) and Junction Box (European committee for electrotechnical standardization, 2012). Reduction of CTM loss occurring due to module manufacturing process has been an important subject. Several innovative techniques have been tried. Enhancement of light capture by reflection (Su et al., 2011; Chung et al., 2012), ARC on Glass (Bunea et al., 2010; Wohlgemuth et al., 2015), conductive adhesive for tabbing and stringing (Hsieh et al., 2010; Zemen et al., 2013), lamination with silicone gel (Poulek et al., 2012) are some of the notable innovations. Several authors reported models of CTM loss attributed to different materials and processes; such as encapsulation (Grunow and Krauter, 2006), resistance introduced due to spot soldering (caballero et al., 2006), mismatch between cells (Louis and Bucciarelli, 1979; Kaushika and Rai, 2007; Webber and Riley, 2013), overall reflection (Scheydecker et al., 1994; Koomen et al., 1996), textured cover glass (Campbell, 1990; Cardona et al., 2008), etc. The effect of temperature on optical loss has also been reported (Lu and Yao, 2007; Krauter and Hanitsh, 1996). More basic models (El-Basit Wafaa et al., 2013) relates illumination and temperature effects to the parameters such as series, shunt resistance and power output. The effect of cell type and efficiency are not addressed by these models. It has been seen by the author that the overall CTM conversion loss has strong relation with the input cell types. In this work, eight different types of cells have been used to develop a comprehensive model to address electrical as well as optical power loss/gain. The cells used are from different manufacturers with varying efficiency, mono and multi crystalline and 2 and 3 bus bars. A guideline for optimum material and process parameters depending on cell types has been presented. The effect of the ARC on cover glass has also been added in this analysis and modeling. It may be noted that the basic process flow and steps used in a SPV module manufacturing have not been disturbed to make this model practical and readily usable. As per the knowledge of the author such comprehensive model and analysis are not available in the literature.
185
collected by making single cell coupons. AAA class solar cell tester from SPIRE Corporation (Spi-Cell TesterTM) www.spiresolar.com has been used to measure powers at various stages of manufacturing; bare cells, tabbing–stinging, bussing and lamination. The power changes due to junction box and ARC on cover glass have been determined by making full size modules of 60 cells (Roy et al., 2010). Experimental data have been collected from single cell coupons made using eight different types of 156 mm 156 mm size cells. The cell description, without disclosing the manufacturer’s name, has been given in Table 1. The cell types chosen are combination of comparatively low (Type-4 and Type-5), moderate (Type-2 and Type-3, Type-6) and high (Type-1, Type-7 and Type-8) efficiency. Fig. 2 shows the configuration of the coupon before and after tabbing–stringing (T&S) and bussing. The same cells are then tabbed–stringed–bussed and measured again. The cells and cell coupons have been tested on a Cell-tester pre and post tabbing–stringing–bussing. Standard tabbing ribbon of 2.4 mm 0.13 mm for 2BB cells and 1.7 mm 0.13 mm for 3BB cells was used. For clarity, the tabbing ribbon for the back bus bar in Fig. 2 is shown a bit wider. The results reported here is the average of five cell coupons of each cell type. LTB of 150 mm and LCB of 20 mm (see Fig. 2b) have been used for making cell coupons. LBB is the property of the cells. The configuration of the 3BB cell coupons is similar. The lamination of the cell coupons was done, keeping the same configuration of Fig. 1, using a Jinchen laminator (www.jinchensolar.com). Lamination temperature of 145 ° C and total cycle time of 18 min have been used. It is not
2. Process flow and experimental details A certified module manufacturing line (Roy et al., 2010) has been used to collect the experimental data. The layup sequence is depicted in Fig. 1. Experimental data has been
Fig. 1. Typical lay-up sequence of a Solar Photovoltaic (SPV) module manufacturing.
186
J.N. Roy / Solar Energy 130 (2016) 184–192
Table 1 Description of the cells used for cell coupons. S. no.
Cell type
1 2 3 4 5 6 7 8
Type-1 Type-2 Type-3 Type-4 Type-5 Type-6 Type-7 Type-8
(Vendor (Vendor (Vendor (Vendor (Vendor (Vendor (Vendor (Vendor
1) 2) 3) 4) 5) 6) 7) 8)
Mono/multi
No. of bus bars
Pmax (WP)
Mono Multi Multi Multi Multi Multi Mono Mono
2BB 2BB 2BB 2BB 3BB 3BB 3BB 3BB
4.028 3.885 3.921 3.638 3.819 3.881 4.257 4.101
Front Contact
LBB
LTB
LCB
Back Contact
(a)
(b)
Fig. 2. (a) Configuration of 2BB bare cell and (b) configuration after Tabbing–Stringing (T&S) and bussing.
practical to determine the effect of junction box effect using cell coupon. The overall junction box related power loss of a module is distributed amongst 60 cells which are connected in series for making a 60 cells module. This effect is therefore determined by measuring power of full 60 cells pre and post junction box fixing. Testing procedure for a module (Roy et al., 2010), using a Spire Sun Simulator (www.spiresolar.com), has been followed. The same methods have been adopted to determine the effect of ARC on cover glass. ARC glass from five different vendors has been used to determine the power gain. Measurements are also done outdoor. Some of the results are also presented in this paper. 3. Experimental results, modeling and discussion – tabbing– stringing (T&S), bussing Short circuit current (ISC) and maximum power (Pmax) of pre and post T&S have been shown in Table 2. In all
cases there is power reduction after T&S. There is a slight decrease in ISC of all cells, which can be explained from shadowing loss. Major loss of CTM conversion primarily occurs due to T&S. This loss can be quantified through the one-diode model (Fig. 3) of a solar cell. Iph is the photo-generated current, Id is current through diode, RS the series resistance and RSh the shunt resistance. The diode has a saturation current I0 and an ideality factor n. The cell voltage (V) and current (I) satisfy Eq. (1) Green, 2014. I ¼ I ph I 0 ½expfðV þ IRS Þ=nV T g 1 ðV þ IRS Þ=RSh
ð1Þ
It is to be noted that the total series resistance applicable in the case of cell coupon are due to inherent bare cell resistance (RSO) and added resistance (Rts) due to T&S and bussing process. In case RSO is the dominating factor, which is applicable for lower efficiency cells, the conversion loss at T&S is lower. It is other-way-around for high efficiency cells. Therefore, for a given T&S process, higher
J.N. Roy / Solar Energy 130 (2016) 184–192
187
Table 2 Electrical data of cell coupons pre and post T&S. S. no.
Cell type
1 2 3 4 5 6 7 8
Type-1 Type-2 Type-3 Type-4 Type-5 Type-6 Type-7 Type-8
(Vendor (Vendor (Vendor (Vendor (Vendor (Vendor (Vendor (Vendor
Pre-T&S
1) 2) 3) 4) 5) 6) 7) 8)
Post-T&S
ISC (A)
Pmax (WP)
ISC (A)
Pmax (WP)
8.622 8.204 8.122 8.035 7.987 8.214 8.690 8.690
4.028 3.885 3.921 3.638 3.819 3.881 4.257 4.101
8.582 8.194 8.111 8.060 7.990 8.207 8.672 8.678
3.785 3.673 3.687 3.451 3.617 3.671 3.961 3.941
efficiency cells have higher CTM conversion loss after T&S. This clear trend can be seen from the results presented in Table 2. It is therefore important to have detail understanding, including the effect of cell properties on the overall CTM conversion loss. In order to quantify the effect of RS on Pmax, Eq. (1) can be used and an expression can be derived relating to copper ribbon parameters for determining conversion loss related to T&S. Calculation of series resistance added due to T&S is not very straight-forward as it is not a simple lumped resistance. A typical ribbon normally makes contact all along the length of the bus bar and hence behaves like a distributed resistance. A lumped equivalent of this distributed resistance can be calculated and used in Eq. (1) to quantify the power loss. Current is collected uniformly along the length of the bus bar ribbon. Current flows through fingers to the nearest bus bar. Once the bus bar collects the current from fingers, the full cell current is collected at the ribbon end. In this configuration, the effective lumped series resistance seen by the cell is given by Eq. (2) (Jung et al., 2014). ð2Þ
where qCu is the resistivity of the copper ribbon; LCu, WCu and tCu are the length, width and thickness of the ribbon respectively. For the gap between cells, direct lumped resistances can be used. For a 2BB cell, the cell-to-cell gap series resistance can be written as: 1=2 qCu LCTC =ðW Cu tCu Þ and for a 3BB cell: 1=3 qCu LCTC =ðW Cu tCu Þ
6.04 5.46 5.97 5.13 5.28 5.40 6.95 6.34
Here, LCTC is the length of the cell-to-cell gap. The resistance of the silver bus bar has been neglected. This actually appears in parallel with the copper ribbon and very thin. The contact resistance has also been neglected. Busing steps connect tabbed strings in series and introduce additional series resistance mainly contributed by bussing leads. Fig. 4 shows current flow in the series connection of two strings. The total series resistance can be split into separate parts arising due to LSS, LBB and LCB. The contribution due to LSS is:
Fig. 3. One-diode model of solar cell.
Rts ¼ 1=3 qCu ðLCu =ðW Cu tCu Þ
CTM loss (%)
qBCu LSS =ðW BCu tBCu Þ due to LCB is: 1=2 qCu LCB =ðW Cu tCu Þ and due to LBB is: 1=2 qBCu LBB =ðW BCu tBCu Þ qBCu is the resistivity, WBCu and tBCu are the width and thickness of the bussing copper. The entire module’s electrical connections can be broken up into one of the aforementioned configurations and the total resistance calculated. This bussing resistance would get equally divided over all the module’s cells as an additional series resistance component. Combining Eqs. (1) and (2) and bussing resistance components, the loss in power after tabbing–stringing and bussing can be quantified. The model was validated on single-cell coupons both pre and post tabbing–stringing as described below: For the tabbed cell configuration shown in Fig. 4, total tabbing–stringing–bussing resistance is depicted by Eq. (3). Rtsb ¼ 1=3 qCu ½LTB =ðW Cu tCu Þ þ qCu ½LCB =ðW Cu tCu Þ þ 1=2 qCu ½LBB =ðW BCu tBCu Þ
ð3Þ
qCu is the resistivity of the ribbon; WCu and tCu are the width and thickness of tabbing ribbon and WBCu and tBCu are the width and thickness of bussing ribbon. LTB is the length of soldered ribbon which is assumed to be the same on the top and back side of the cell. LCB is the gap between the cell edges and bussing ribbon, LBB is the distance between extreme bus bars of the cell. When a bare-cell is measured, its VOC, ISC, Vm and Im can be used to extract the one-diode equivalent circuit parameters (Fig. 3) – Iph, Id, RS and n. Here RSh is assumed to be large and hence
188
J.N. Roy / Solar Energy 130 (2016) 184–192
I/2
LCB
I/2
LBB
I
L
Fig. 4. Current flow in busing copper.
neglected. Once the cell one-diode parameters are known, a theoretical estimate can be made for the tabbing–stringing series resistance using Eqs. (3) or (4). This can then be used to calculate the power loss using Eq. (1). The validation of the model is done from the measured data. I–V data of bare cells (Table 2) is used to extract the one-diode parameters for these cells which are then used to calculate the theoretical change in I–V after tabbing–string ing–bussing. Table 3 shows the extracted parameters. Eq. (3) can be used to determine ribbon series resistance, which turns out to be about 3.4 mX for 3BB cells. Using extracted parameters presented in Table 3 and the calculated addition of series resistance due to tabbing– stringing, the power loss in tabbing–stringing–bussing can be calculated for these single-cell coupons. Comparison between experimental results and calculated values is shown in Fig. 5. There seems to be a good level of agreement between the model and actual calculated values. Similar calculations can be done for a full module and the power loss modeled. However, it is not feasible to do a measurement of 60-series connected stringed cells after T&S and hence it could be difficult to validate the model for a full sized module. The model would however be true irrespective of the number of cells.
4. Experimental results, modeling and discussion – lamination Lamination is another major process step where changes in optical properties of the module lead to changes in power. However, normally there is a power gain due to lamination. The effect due to power loss in T&S–bussing and power gain in lamination typically results net power loss but not as high as due to T&S–bussing alone. Laminated modules have the cells covered by a layer on EVA and glass on the top and a layer of EVA and back sheet on the bottom (Fig. 1). In a solar cell, silicon (refractive index n2) is covered by a thin layer of Anti-Reflection Coating (ARC) of refractive index n1 and thickness d1. When the cell is laminated, two newly added interfaces – Air– Glass–EVA and Glass–ARC change the optical properties of the stack-up, thus changing effective intensity reaching silicon. The EVA–glass composite will be referred to as glass (refractive index n3) in the subsequent discussions.
7 6 5
Table 3 Extracted parameters of one diode equivalent circuit of Fig. 1.
4
S. no.
Cell type
IOP (A)
Id (A)
n
RS (mX)
1 2 3 4 5 6 7 8
Type-1 Type-2 Type-3 Type-4 Type-5 Type-6 Type-7 Type-8
8.622 8.204 8.122 8.035 7.987 8.214 8.690 8.690
6.94E009 5.65E009 2.42E009 9.93E009 4.63E009 4.03E009 2.90E010 8.19E008
1.130 1.138 1.106 1.149 1.019 1.114 1.007 1.307
2.886 3.161 2.925 4.275 3.049 2.921 2.944 1.584
Expt. Theory
3 2 1 0 Type 1
Type 2
Type 3
Type 4
Type 5
Type 6
Type 7
Type 8
(X Axis: Type of Cells, Y Axis: Power Loss)
Fig. 5. Tabbing–Stringing–Bussing power loss in single cell coupons.
J.N. Roy / Solar Energy 130 (2016) 184–192
ISR ¼ qk=hc IQE
ð4Þ
q is the charge of electron, k is the wavelength of light, h is the Planck’s constant and c is the speed of light. The spectrum of light entering silicon is given by Eq. (5):
0.35 0.30
Reflectivity
The effect of the stack-up on incoming radiation can be modeled analytically as a function of wavelength. Current generation, which is determined by Quantum Efficiency (QE), is also a function of wavelength. External Quantum Efficiency (EQE) has been determined first by modeling optical transmission of the stack-up. This gives the spectrum of light entering silicon cell and can be mapped to the generation of current in the cell govern by Internal Quantum Efficiency (IQE). Comparison between an Air– ARC–Silicon stack-up (bare cell) and Glass–Air–ARC–Sili con stack-up (laminated cell) would then give a quantitative estimate of lamination gain/loss. Using standard analytical expression (Green, 2014) the total reflection can be derived for a tri-layer stack-up. An Air–Glass–ARC–Silicon (laminated cell) stack-up, can be modeled as an Air–Glass stack-up in series with a Glass– ARC–Silicon tri-layer. For Air–Glass, Fresnel equations can be used. Typical values for nAir, nGlass and nSi are 1.00, 1.50 and 3.80 respectively. nARC and dARC can be varied typically from 1.80 to 2.20 and from 70 nm to 110 nm respectively. Fig. 6 shows the reflectivity from an Air– ARC–Si stack-up (bare-cell) and an Air–Glass–ARC–Si stack-up (laminated cell). Reflections in case of the laminated cell are lower than the bare cell. This stack-up configuration is thus expected to give a lamination gain. The aim is to have minimum reflection loss of the laminated cell. The value of n and d can be adjusted to vary overall reflection loss. It may be noted that this set of value may not give the minimum reflection loss for the bare cell. The result presented in Fig. 6 is from one of the cell type used and data is available to the author. The lamination gain is the difference between pre and post lamination reflection loss. Bare cell ARC is independently control to get maximum efficiency of the cell. However, the best result are obtained in case the whole stack-up, optical layers of cell and module, are designed together. This has successfully attempted for development of cell and module manufacturing technologies by author. This has been done by determining reflectivity vs wavelength for various refractive index for bare as well as laminated cells. The set of values which give minimum overall reflectivity for the laminated cell are to be selected. This optimum value will ultimately lead to maximum current generation of the laminated cell as described subsequently. The reflectivity determines the External Quantum Efficiency (EQE). Once the External Quantum Efficiency (EQE) is known, Internal Quantum Efficiency (IQE) can be determined by the cell response to portion of the incoming light entering silicon. IQE would be the ratio of number of carriers generated to number of photons entering Silicon. A parameter called Internal Spectral Response (ISR) is defined according to Eq. (4).
189
0.25 0.20 0.15 0.10 0.05 0.00
300
400
500
600
700
800
900
1000
1100
1200
Wavelength (nm) Fig. 6. Reflectivity from Air–ARC–Si (Blue) and Air–Glass–ARC–Si (Green): n1 = 2.13, d1 = 87 nm. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
I Si ðkÞ ¼ I AM1:5G ðkÞ 1 RStack-up ðkÞ
ð5Þ
This spectrum entering Silicon would then generate a current governed by ISR of the cell. JOP, density of optically generated photo-current is given by Eq. (6): Z J OP ¼ I AM1:5 G ðkÞ 1 RStack- up ðkÞ ISRðkÞ ð6Þ The optically generated photo-current of the cell, IOP can be obtained using IOP = Acell JOP, where Acell is the active cell area. Thus, IOP of bare-cells and laminatedcells can be calculated accordingly and the lamination gain can be quantified. Eq. (6) can be used to calculate the current generation for any given stack-up and maximize the JOP of cells after lamination. Fig. 7 shows the generation of current as a function of wavelength for a bare-cell and a laminated-cell. The incident spectrum is AM1.5 G and the ARC has a refractive index of 2.13 and a thickness of 87 nm. It can be seen that the bare-cell has better response than laminated-cell in the wavelength range at around 600–700 nm. However, outside this band the laminated-cell has a better response. The overall lamination gain using this ARC is about 1.9% ˚ ). A more general case is shown in Fig. 8, (7.638–7.785 A which shows IOP variation of a standard 600 bare-cell for ARC refractive index variation from 1.80 to 2.13. This can be repeated for various ARC thicknesses. This graph represents that till about nARC of 2.04, there is a lamination loss. Beyond that value, lamination gain can be obtained. The set of values produce the maximum IOP has to chosen. Fig. 8 shows the optimum possible result. The maximum IOP is obtained at nARC of 2.15 and dARC of 110 nm. Although, it is possible to increase the IOP further by increasing nARC beyond 2.13, the bare cell efficiency will then drops significantly. This is not acceptable for cell manufacturer and therefore not tried. The single-cell coupons used for tabbing–stringing–bus sing experiments have been laminated using standard
190
J.N. Roy / Solar Energy 130 (2016) 184–192 0.8
lamination materials and tested. The results are shown in Table 4. The trend seems to agree with the model. Proper validation can be done once the nARC and dARC for all the cell types used are known. The model has been successfully used by the author, to minimize CTM conversion loss, for two of these cell types (Type-4 & Type-6) with known parameters.
Current Generation (J0 /nm)
0.7 0.6 0.5 0.4
5. Other CTM conversion loss/gain
0.3 0.2 0.1 0.0 300
400
500
600
700
800
900
1000
1100
1200
Wavelength (nm) Fig. 7. Current generation as a function of wavelength: bare cell (Blue) and laminated cell (Red). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) 7.9 7.85 7.8 7.75 7.7 Laminated
7.65
Bare Cell
7.6 7.55 7.5 7.45 7.4 1.75
1.8
1.85
1.9
1.95
2
2.05
2.1
2.15
2.2
(X Axis: nARC, Y Axis: IOP)
Fig. 8. IOP variation with nARC for optimized dARC.
Additional series resistance contributed by the leads and contacts of the junction box introduces additional loss. This comes in series with the cell string having many cells, e.g. 60 or 72 cells for a 60 or 72 cells module. It is found that this resistance introduces 0.1–0.3% CTM conversion loss at module level power; depending on the number of cells in the module, type of junction box used, etc. The CTM conversion loss can be reduced further by using ARC on the cover glass. Experimental results are described in brief to give complete picture on the CTM conversion loss. A detail analysis/modeling is being done and will be reported separately. Five different vendors have been used and the measurements done indoor (www. spiresolar.com) as well as outdoor. The results presented here are average of 5 samples (modules) each and outdoor data taken for 7 sunny days during the month of May in Hyderabad, India. It has been found that CTM conversion gain of 1–2.5% at the module level can be achieved depending on the vendor and test mode. Table 5 gives the indoor test results at STC. All the results presented for ARC on cover glass are percentage difference between the ISC of the module with respective ARC on cover glass and a reference module without any ARC on cover glass under identical outdoor measurement condition.
Table 4 Measured CTM conversion loss/gain. S. no.
Cell type
Power (WP) Pre-T&S
Post T&S
Post lamination
T&S
Lamination
Total
1 2 3 4 5 6 7 8
Type-1 Type-2 Type-3 Type-4 Type-5 Type-6 Type-7 Type-8
4.028 3.885 3.921 3.638 3.819 3.881 4.257 4.101
3.785 3.673 3.687 3.451 3.617 3.671 3.961 3.941
3.879 3.774 3.799 3.558 3.722 3.783 3.870 4.036
6.04 5.46 5.97 5.13 5.28 5.40 6.95 6.34
2.50 2.75 3.05 3.10 2.90 3.05 2.3 2.4
3.54 2.71 2.92 2.03 2.38 2.35 4.65 3.94
Conversion loss (+)/gain () (%)
Table 5 STC power gain with ARC cover glass. S. no.
Type
Average power (WP)
Min. power (WP)
Max. power (WP)
Gain as compared to reference (%)
1 2 3 4 5 6
Reference (No ARC) Vendor-1 Vendor-2 Vendor-3 Vendor-4 Vendor-5
234.13 237.98 237.85 237.58 236.93 236.27
232.58 236.92 235.93 236.29 235.35 234.7
235.52 238.94 239.65 238.7 237.98 237.59
– 1.64 1.59 1.47 1.2 0.91
J.N. Roy / Solar Energy 130 (2016) 184–192
191
Table 6 Gain with outdoor testing of ARC cover glass. Glass type
Average ISC gain as compared to reference (%) 10 AM
10:30 AM
11 AM
11:30 AM
12 PM
12:30 PM
1 PM
1:30 PM
2 PM
3 PM
4 PM
5 PM
Vendor-1 Vendor-2 Vendor-3 Vendor-4 Vendor-5
1.92 2.32 2.33 0.72 1.45
1.69 2.48 2.45 0.69 1.37
1.78 2.19 2.20 0.96 1.3
2.01 2.71 2.2 0.77 1.29
2.16 2.59 2.39 0.61 0.92
1.85 2 2.31 0.8 1.29
1.54 1.38 1.92 0.18 0.61
2.13 2.99 2.65 1.03 1.92
1.31 2.09 2.21 0.02 0.71
1.78 2.68 2.75 0.68 0.5
1.27 2.18 2.98 0.62 0.86
1.16 2.22 2.19 0.50 0.75
Table 7 Gain with outdoor testing of ARC cover glass during early morning. Glass type
Average ISC gain as compared to reference (%) 8 AM
8:30 AM
9 AM
Vendor-1 Vendor-2 Vendor-3 Vendor-4 Vendor-5
2.53 1.09 1.76 0.36 1.15
3.59 2.60 3.54 2.02 2.54
3.44 2.75 3.62 2.09 2.97
Table 8 Average gain with outdoor testing during 10 AM–2 PM. Glass type
Average ISC gain as compared to reference (%)
Vendor-1 Vendor-2 Vendor-3 Vendor-4 Vendor-5
1.82 2.42 2.3 0.64 1.21
The summary of the outdoor testing results is shown in Table 6. A comparative value for early morning, which indicates large gain, is shown in Table 7. Average gain during peak period (10 AM–2 PM) during the day is shown in Table 8. It is clear that overall outdoor power output gain of more than 2% can be achieved by using ARC on cover glass. The output, represented by ISC, is found to vary throughout the day (Table 6) and noticeably higher during early morning (Table 7). This is primarily due to inclined angle for the incident rays and low light behavior of ARC. 6. Conclusion In this paper a detail and comprehensive analysis has been presented for understanding the cell to module (CTM) conversion loss during manufacturing of Solar Photovoltaic (SPV) module. Conversion loss due to Tabbing–Stringing (T&S)–bussing has been modeled and validated by experimental measurements from coupon cells. Eight different cell types have been used for making the coupon modules. It has been shown that the lamination process can give conversion gain under certain conditions. Both T&S–bussing loss and lamination gain/loss depend on the starting cell properties. Higher efficiency cells have higher net conversion loss unless the process, particularly lamination, is optimized for each cell type. This model has been used successfully by the author to minimize the
overall conversion loss for two of the cell types with known cell manufacturing details. There is good agreement between experimental and theoretically calculated values for tabbing–stringing–bussing losses. Optimization of tabbing ribbon dimensions has successfully done by the author for module manufacturing process. The Lamination Optics Model is shown general agreement with the trend of experimental data and used to maximize lamination gain for two of the cell types with known parameters. The effect of ARC on cover glass for further reduction of CTM conversion loss has been presented in brief. Modules with five different ARC glass vendors have been studied. The measurements are done at STC as well outdoor and the results are compared with reference glass having standard glass with no ARC. It has been found that there is a significant difference between STC and outdoor measurements. Output is found to be more during outdoor measurement with significantly more output in the morning hours. Acknowledgement The experimental data used here was from the coupons and modules (Roy et al., 2010) manufactured at Solar Semiconductor Pvt. Ltd, Hyderabad, India. References Bunea, G., Xavier, G., Rose, D., Nelson, L., Peurach, J., 2010. Performance and reliability of modules with antireflective coated glass. In: Proceedings of 25th EUPVSEC, pp. 4103–4106. Caballero, L.J., Sanchez-Friera, P., Lalaguna, P.B., Alonso, J., Vazquez, M.A., 2006. Series resistance modeling of industrial screen printed monocrystalline silicon solar cells and modules including the effect of spot soldering. In: Proceedings of Conference of World Conference on Photovoltaic Energy Conversion, pp. 1388–1391. Campbell, P., 1990. Light trapping in texture solar cells. Sol. Energy Mater. Sol. Cells 21, 165–172. Cardona, M.S., Coronado, A.V., Alvarez, J.L., 2008. An analysis of geometrical shapes for PV modules glass encapsulation. Sol. Energy Mater. Sol. Cells 92, 323–331. Chung, I., Back, U.I., Moon, I.S., Kwon, O., Bae, K., Shin, S., Cho, E.C., Lee, W.J., 2012. Optimization of the output power by effect of backsheet reflectance and spacing between cell strings. In: Proceedings of IEEE Photovoltaic Specialist Conference (PVSC), pp. 2388–2390. Dimassi, W., Derbali, L., Bouaicha, M., Bessais, B., Ezzaouia, H., 2011. Two-dimensional LBIC and internal-quantum efficiency investigation of grooved grain boundaries in multi crystalline silicon solar cells. Sol. Energy 85, 350–355. El-Basit Wafaa, A.B.D., El-Maksood, A.B.D., Soliman, F., El-Moniem Saad, A.b.d., 2013. Mathematical model for photovoltaic cells. Leonardo J. Sci. 23, 13–28.
192
J.N. Roy / Solar Energy 130 (2016) 184–192
European Committee for Electrotechnical Standardization, 2012. EN 50548: Junction Boxes for photovoltaic modules, Brussels. Green, M.A., 2014. Solar Cells: Operating Principles, Technology and System Applications. Prentice-Hall Inc.. Grunow, P., Krauter, S., 2006. Modelling of encapsulation factors for photovoltaic modules. In: Proceedings of World Conference on Photovoltaic Energy Conversion, pp. 2152–2155. Hsieh, H.H., Lee, W.K., Chang, T.C., Hsi, C.S., 2010. Photovoltaic modules of crystalline solar cells using new type of assembly structure. In: Proceedings of International Microsystems Packaging Assembly and Circuits Technology Conference (IMPACT), pp. 1–4. Jung, T., Song, H., Ahn, H., Kang, G., 2014. A mathematical model for cell to module conversion considering mismatching solar cells and the resistance of the interconnection ribbon. Sol. Energy 103, 253–262. Kaushika, N.D., Rai, A.K., 2007. An investigation of mismatch losses in solar photovoltaic cell networks. Energy 32, 755–759. Koomen, E.A.S., Alsema, E.A., Turkenbug, W.C., 1996. A simple model for PV module reflection losses under field condition. Sol. Energy 57, 421–432. Krauter, S., Hanitsh, R., 1996. Actual optical and thermal performance of PV modules. Sol. Energy Mater. Sol. Cells 41–42, 557–574. Louis, L., Bucciarelli, J., 1979. Power loss in photovoltaic array due to mismatch in cell characteristics. Sol. Energy 23, 277–288. Lu, Z.H., Yao, Q., 2007. Energy analysis of silicon solar cell modules based on an optical model for arbitrary layers. Sol. Energy 81, 636– 647. Macdonald, D.H., Cuevas, A., Kerr, M.J., Samundsett, C., Ruby, D., Winderbaun, S., Leo, A., 2004. Texturing industrial multi crystalline solar cells. Sol. Energy 76, 277–283. Mickiewitcz, R.A., Cahill, E., Wu, P.I., 2012. Non-destructive determination of degree of cross linking of EVA solar module encapsulation using DMA shear measurements. In: Proceedings of Photovoltaic Specialist Conference (PVSC), pp. 710–713.
Park, K.M., Lee, M.B., Shin, J.W., Choi, S.Y., 2013. Investigation of surface features using reactive ion etching method for enhanced performance of multi-crystalline silicon solar cells. Sol. Energy 91, 37– 47. Poulek, V., Strebkor, D.S., Persie, I.S., Libra, M., 2012. Towards 50 years lifetime of PV panels laminated with silicone gel technology. Sol. Energy 86, 3103–3108. Roy, J.N., Gariki, G.R., Nagalakhmi, V., 2010. Reference module selection criteria for accurate testing of photovoltaic (PV) panels. Sol. Energy 84, 32–36. Scheydecker, A., Goetzberger, A., Wittwer, V., 1994. Reduction of reflection losses in PV modules by structured surfaces. Sol. Energy 53, 171–176. Su, W.S., Chen, Y.C., Liao, W.H., Huang, C.H., Liu, D.C., Huang, M.Y., Wu, Z.C., Ho, S.J., 2011. Optimization of the output power by effect of backsheet reflectance and spacing between cell strings. In: Proceedings of IEEE Photovoltaic Specialist Conference (PVSC), pp. 3218–3220. Webber, J., Riley, E., 2013. Mismatch loss reduction in photovoltaic array as a result of sorting photovoltaic modules by max-power parameters. Renewable Energy, 1–9. Wohlgemuth, J., Cunnigham, D., Shaner, J., Nguyen, A., Ransome, S., Artigeo, A., 2015. Crystalline silicon photovoltaic modules with antireflective coated glass. In: Proceedings of IEEE Photovoltaic Conference (PVSC), pp. 1015–1018. www.jinchensolar.com. www.spiresolar.com. Yang, W.J., Ma, Z.Q., Tang, X., Feng, C.B., Zhao, W.G., Shi, P.P., 2008. Internal quantum efficiency for solar cells. Sol. Energy 82, 106–110. Zemen, Y., Schulz, S.C., Trommler, H., Buschhorn, S.T., Bauhofer, W., Schulte, K., 2013. Comparison of new conductive adhesive based on silver and carbon nanotubes for solar cells interconnection. Sol. Energy Mater. Sol. Cells 109, 155–159.