CHARACTERIZATION OF CPV CELLS ON A HIGH INTENSITY SOLAR SIMULATOR: A DETAILED UNCERTAINTY ANALYSIS

Transcription

1 CHARACTERIZATION OF CPV CELLS ON A HIGH INTENSITY SOLAR SIMULATOR: A DETAILED UNCERTAINTY ANALYSIS Mauro Pravettoni 1,2, Monica Cadruvi 3, Diego Pavanello 1, Thomas Cooper 3, and Gabi Friesen 1 1 University of Applied Sciences and Arts of Southern Switzerland, Institute of Applied Sustainability to the Built Environment (SUPSI-ISAAC), Campus Trevano, Canobbio, Switzerland 2 Imperial College London, Blackett Laboratory, London, SW7 2BW, United Kingdom 3 Swiss Federal Institute of Technology, Dept. of Mechanical & Process Engineering, Zurich, Switzerland ABSTRACT The authors modified a large area pulsed solar simulator to a high intensity pulsed solar simulator, moving the test measurement plane to various distances close to the xenon lamp. Measurement results reported in this work have confirmed the dependence of the total irradiance on the cell-to-lamp distance, up to the maximum measured intensity at 3000X. Classification of the solar simulator based on the international standard IEC is reported. Though the standard procedure for current-voltage measurements of non-concentrating cells prescribes a calibrated reference cell to detect the total irradiance independently, a self-reference method is usually preferred at high intensities by the CPV community. In this work the authors apply a detailed uncertainty analysis to both the procedures. The result highlights the pros and contra of the two methods, giving a useful tool in the pre-normative work for the preparation of norms and standard procedures for terrestrial CPV cells characterization. INTRODUCTION Concentrating photovoltaics (CPV) represents nowadays an emerging market at its early stage for terrestrial applications and is expected to increase: several companies worldwide are focusing on it and the market prediction for 2009 and 2010 were 30 MW and 100 MW of installed power, respectively [1]. Two main causes of the growing interest in CPV are: (a) the shortage in silicon feedstock, with the need of decreasing material consumption per silicon solar cell and watt-peak; (b) the high efficiency values of III-V multi-junction cells, which exceeded 42% efficiency under concentrating light [2], but too expensive to be used in non-cpv applications. In 2010 the Swiss PV Module Test Centre at SUPSI- ISAAC, a Swiss centre for standard photovoltaic testing, started a new research activity in CPV, within InPhoCUS [3], a project co-financed by the Swiss Commission for Technology and Innovation (CTI). One of the roles of SUPSI-ISAAC in the project is to perform indoor characterisation of the candidate high efficiency CPV cells. To this purpose, the authors have modified a large area pulsed solar simulators (Pasan IIIa, see Figure 1), moving the test measurement plane at distances close to the xenon lamp, as previously reported in the literature [4]. Figure 1 A scheme of the modified Pasan IIIa measurement setup. The classification of the modified solar simulator based on the available international standard IEC (Ref. [5]) is reported in the first part of this work. An agreement in the standard procedure for indoor current-voltage (IV) characterization of CPV cells is still missing. Though the standard procedure for non-concentrating cells IEC (Ref. [6]) prescribes a calibrated reference cell to detect the total irradiance independently, a self-reference method is usually preferred at high intensities by the CPV community, where the CPV test device itself is assumed to be linear and the total irradiance is given by the ratio between the short-circuit current at the given irradiance and the value at 1000 Wm 2 ( one sun, 1X). The practical advantages and disadvantages of the two different procedures and the need of a procedure to test the CPV device linearity in a wide range of intensities and have been recently proposed [7]. In this work the authors apply a detailed uncertainty analysis according to Ref. [8] to both the procedures. The result gives a useful tool in the pre-normative work for the preparation of norms and standard procedures for terrestrial CPV cells characterization. CLASSIFICATION OF THE HIGH INTENSITY PULSED SOLAR SIMULATOR The total irradiance on the modified Pasan IIIa as a function of the target-to-lamp distance was measured with a c- Si photodiode (11.3 mm 2 cell size), protected by a neutral density filter (optical density, = 3), which was cali /11/$ IEEE

2 brated at 1X. The target-to-lamp distance was recorded by a laser distance meter. Figure 2 shows the measured intensities (in suns) as a function of the target-to-lamp distance, assuming ±2.5 cm position uncertainty due to the non-pointlike lamp source. Measurement data are compared with the theoretical dependence (boundary condition: 1.118X measured at 8.15 m), confirming the linearity of the filtered photodiode used as reference. The deviation from linearity at 3000X is affected by the precision in the positioning system. The modified solar simulator was classified based on the requirements in the International Standard IEC [5,9] as follows. Measurements were performed at =0, =0, = and = with respect of an XY Cartesian coordinate system on the target plane, centred at the point of maximum intensity (positioning uncertainty: ±2 mm). A monitor cell (a filtered c-si reference cell) was fixed outside the measurement plane to normalize all flashes. Figure 4 illustrates the maximum diameter where class A (±2%), B (±5%) or C (±10%) requirements for spatial nonuniformity are met as a function of the intensity. The chart shows that the modified Pasan IIIa simulator has class A spatial non-uniformity above 1100X on a 8 cm diameter target area. Figure 2 Total irradiance (in suns) as a function of the target-to-lamp distance (boundary condition: 1.118X, measured at 8.15 m). The ±. cm uncertainty bars in distance measurements are shown. Figure 3 Spectral match to AM1.5d. The class A requirement within nm is met according to IEC [10]. The spectral mismatch in the additional bands and is highlighted in yellow. Spectral Match Figure 3 shows the spectral match between Pasan IIIa spectrum and AM1.5d standard spectrum [10]. IEC requirement indicates six wavelength bands from 400 to 1100 nm: within these intervals, Pasan IIIa spectrum meets the class A requirement. For CPV purposes, it is beneficial to report also the spectral match in two further wavelength bands: nm (important for the top junction of III-V group multi-junction CPV cells) and nm (important for Ge junction in III-V multijunction CPV cells). The two additional bands are highlighted in yellow in Figure 3. In the extended wavelength band from 300 and 1700 nm, Pasan IIIa spectrum is poorer than from 400 to 1100 nm. This result suggests that a large current imbalance between component junctions may be observed in IV measurements of most III-V group multi-junction CPV cells. Spatial Non-uniformity Spatial non-uniformity of irradiance was measured with a filtered c-si photodiode at various level of intensities. Figure 4 Spatial non-uniformity. The target diameter where class A, B or C requirements are met is shown as a function of the intensity. Temporal Instability Figure 5 shows the first 10 ms profile of Pasan IIIa pulse. Long term instability (LTI) of irradiance meets the class A /11/$ IEEE

3 (±2%) requirement over the 2 ms plateau around the maximum, the class B one (±5%) over 3.5 ms and the class C one (±10%) over an interval of more than 5 ms. The data acquisition system allows simultaneous measurements of current, voltage and irradiance: as a consequence, short term instability (STI) of the irradiance is less than ±0.1% in the 2 ms plateau where class A LTI requirements are met. UNCERTAINTY ANALYSIS The IEC Standard-like Procedure Table 1 lists all sources of uncertainty in current ( ), voltage ( ) and irradiance ( ) measurements on the modified Pasan IIIa described above with the IEC standard-like procedure (see Ref. [6]). All uncertainties are type B (i.e. from the manufacturer data sheet, previous measurement data, reasonable assumptions, general knowledge,...); Gaussian (G) or rectangular (R) uncertainty distributions are indicated. Figure 5 Temporal Instability: time intervals where class A (> 2 ms), B (3.5 ms) or C (> 5 ms) requirements are met. Source of uncertainty Type Distr. [%] [%] [%] 1X 10X 100X 1000X Electrical: Data acquisition system B G ±0.5% TI amplifier (reference cell current) B R ±0.1% Combined: Temperature: Temperature sensor B G ±0.4 C Temperature range B R ±1 C Combined: Optical (target diameter: 10 cm): Spatial non-uniformity (1-10X) ±0.5% Spatial non-uniformity (100X) B R ±1% Spatial non-uniformity (1000X) ±2% Co-planarity (1X) Co-planarity (10X) B G 1.5 mm Co-planarity (100X) Co-planarity (1000X) Orientation of ref. & test cell (1X) ±1 deg Orientation of ref. & test cell (10X) ±3 deg B G Orientation of ref. & test cell (100X) ±6 deg Orientation of ref. & test cell (1000X) ±22 deg Alignment of ref. & test cell B R ±3 deg Combined (1X): Combined (10X): Combined (100X: Combined (1000X): Reference cell: Spectral mismatch (c-si CPV cells) B G 2.8% Drift B R 0.1% Calibration (primary ref. cell., 1X) B G 0.5% Calibration (secondary ref. cell. >1X) B G 1.7% Combined (1X): Combined (>1X): Total expanded, = (1X): Total expanded, = (10X): Total expanded, = (100X): Total expanded, = (1000X): Table 1 Uncertainty calculation: IEC standard-like procedure (see Ref. [6]) with filtered reference cells. A- type (i.e. statistical analysis) or B-type (i.e. from manufacturer datasheet, reasonable assumptions, general knowledge,...) uncertainties, Gaussian (G) or rectangular (R) distribution with the coverage factor k are indicated. 95% confidence ( = ) is given on total expanded uncertainties /11/$ IEEE

4 Source of uncertainty Type Distr. [%] [%] X [%] Electrical: Data acquisition system B G ±0.5% Temperature: Temperature sensor B G ±0.4 C Temperature range B R ±1 C Combined: Optical: Spatial non-uniformity B R ±0.5% Reference cell (*): Calibration at 1X B G 2.94% Total expanded, = ( >1X): Table 2 Uncertainty calculation: self-reference method for IV measurements at >1X. (*) Once its linearity is proven, the reference cell is the test cell itself, calibrated at 1X with the uncertainty calculated as in Table 1. For a measurement procedure to test the cell linearity see Ref. [7]. The value a indicates the expanded uncertainty case by case in the given unit; is the coverage factor, which is related to the distribution type and confidence level, and = is the standard uncertainty (in percent, 68% confidence). The sources of uncertainties are grouped as fol- lows: - Electrical uncertainties arise from the data acquisition system (±0.5% from the manufacturer data sheet, with 99% confidence, i.e. = 2.576) and from the transimpedance (TI) amplifier used for reference cell current measurements, for which a ±0.1% rectangular (i.e., = 3) uncertainty was assumed. - Temperature uncertainties were calculated as in the recent work by H. Müllejans et al. [11]: temperature sensor uncertainty (±0.4 C, G distribution, 99% confidence) is given by manufacturer datasheet; the range is (25±1) C and a ±0.5 C non-uniformity was assumed (both R). Unlike Ref. [11], electrical and temperature uncertainties contribute here only once to, since the current is not corrected to the given irradiance and therefore does not affect as well. - Within the optical uncertainties, a ±0.5% R nonuniformity affects the current measurement, giving = 0.289% at any irradiance: this is the non-uniformity of irradiance within the test cell area. The effect to the measurement of irradiance is affected instead by the nonuniformity along the distance between the test device and the reference cell and therefore depends by the intensity level, as discussed in the section above. To calculate this effect, consider =10cm: the values shown in Table 1 were then calculated from Figure 4. The ±1.5 mm (G, 99% confidence) uncertainty in coplanarity also depends on the intensity level, up to 0.459% standard uncertainty both on the current and on the irradiance measurements at 1000X (target-to-lamp distance of about 25 cm, see Figure 2). The G 99% confidence uncertainty on the orientation of the test and reference cell gives rise to a possible different angle of incidence of light onto the cells and also affects both the current and the irradiance measurement. The maximum angle of incidence as a function of the intensity is given by ( ) = tan (1) ( ) where ( ) is the target-to-lamp distance at the intensity and =10cm as above. The maximum misalignment between the reference and the test devices is ±3 deg (R distribution). Both optical uncertainties and the uncertainties related to the reference cell calibration value affect the voltage measurements logarithmically. The uncertainty in voltage measurements is thus given by = log[1+ ] (2) where the diode coefficient = 5.3% was assumed as in the cited paper [11]. - The uncertainties arising from the reference cell are due to the uncertainties in the spectral mismatch, in the observed historical drift and in the uncertainty in the calibration value itself. The uncertainty on the spectral mismatch has been calculated at SUPSI-ISAAC for the worst case in c-si non-concentrating modules and it is here reasonably assumed that the same uncertainty arises from the spectral mismatch of CPV c-si cells. A deeper analysis is ongoing for the spectral mismatch of III-V multi-junction CPV cells, but is beyond the scopes of this paper. Though reference cells are periodically calibrated, a ±0.1% R drift was reasonably assumed. The expanded ( =2) calibration uncertainty to the primary reference cells is ±0.5%; for secondary reference cells (>1X), is 1.7% /11/$ IEEE

5 The Self-reference Method Once the linearity of the test cell is proven, the test cell itself may be used as a reference to detect the incoming irradiance ( self-reference method) from the following ration = ( ) ( X), (3) where ( ) and (1X) are the short-circuit current values of the test device at the unknown irradiance (in suns) and at 1X, respectively. For further details on how to test the linearity of a CPV cell, see the cited Ref. [7]. Table 2 lists the sources of uncertainties on IV measurements with the self-reference method. In this case, no reference cell is needed: all sources of uncertainties in Table 1 related to it may be neglected and only the data acquisition system, temperature (sensor and range) and spatial non-uniformity uncertainties are present. In addition, the uncertainty in (1X) is added ( = ±2.94% = 2 G uncertainty from Table 1). As a result, = ±3.07% was calculated at all intensities >1X, which is consistent with the uncertainty on measurement of irradiance at 1X with the standard method above. Fill Factor, Maximum Power and Cell Efficiency Figure 6 Expanded ( = ) uncertainty for the maximum power (MPP, see eq. (5)) and for cell efficiency (EFF, see eq. (7)): comparison between the modified standard procedure (at 1X, 10X, 100X and 1000X) and the self-reference method (in yellow, at >1X). (a) The uncertainty contribution to the fill factor ( ) is ±0.70% (R), depending on the electrical connections only. This is a reasonable assumption from the periodical measurements performed at SUPSI-ISAAC to prove the overall stability of the measurement system. Since the maximum power is =, (4) (b) then the standard uncertainty on the maximum power is The cell efficiency is = + +. (5) =, (6) where is cell area and is the total incoming. The standard uncertainty in cell efficiency is therefore = + + +, (7) where in the last term it is assumed that the uncertainty in cell area measurement is small enough to be neglected. (c) Figure 7 Uncertainty source contributions to: (a) current; (b) voltage; (c) irradiance measurements. IEC standard-like (at 1X and 1000X) and selfreference method. DISCUSSION Figure 6 shows the comparison between the total expanded ( =2) uncertainties in the maximum power and in cell efficiency in the two methods described above /11/$ IEEE

6 The chart highlights the increase in both uncertainties with increasing irradiance, up to ±6.30% expanded uncertainty of cell efficiency at 1000X measurements with the standard IEC like method (white background). A significant decrease in the uncertainties is given for measurements with the self-reference method at >1X(yellow background), where the total expanded uncertainty of cell efficiency is reduced to ±4.42%. Figure 7, where the standard ( =1) uncertainties on current (a), voltage (b) and irradiance (c) measurements are analysed, shows the relative impact of each source of uncertainty. The charts highlights that the largest contribution to the uncertainties (especially on current and voltage measurements) arises from the optical uncertainty at 1000X with the IEC standard-like method and from the reference cell uncertainty in both the methods analysed. The latter can be reduced with a better spectral mismatch between the reference cell and the test device. CONCLUSIONS The high-intensity pulsed solar simulator in use at SUPSI- ISAAC (a modification of a Pasan IIIa pulsed solar simulator) was described and classified as class AAA according to the available standard (adapted to CPV purposes). It was highlighted that a better spectral match may be necessary to avoid large current imbalances in IV measurements of multi-junction CPV cells such as In- GaP/InGaAs/Ge, for which even the class A spectrum of Pasan IIIa could be poor. Two methods for the IV measurements at high intensities were considered then and the related uncertainty analysis was proposed. The first method requires the use of reference cells to detect the irradiance at high intensity, in line with the standard procedures for non-concentrators. The second method is based on the linear response of the test device itself and does not require a reference cell in any measurement at >1X(while at 1X the cell is calibrated according to the standard procedure). The result of the uncertainty calculation confirmed that a higher precision can be obtained with the second method, as widely assumed in the CPV community. ACKNOWLEDGEMENTS The authors would like to acknowledge the financial support of the Swiss Confederation, Federal Office for Professional Education and Technology, through the funding of the Innovation Promotion Agency CTI no PFIW- IW (InPhoCUS). They would also like to thank for their kind cooperation Mauro Bernasocchi (SUPSI-ISAAC) and Didier Dominé (SUPSI-ISAAC, now at Oerlikon Solar). Implementation of Photovoltaics. Retrieved May 12, 2011, from [2] M. A. Green, K. Emery, Y. Hishikawa and W. Warta, Solar cell efficiency tables (version 37), Prog. Photovolt: Res. Appl. 19(1), 2011, pp [3] M. Pravettoni, M. Barbato, T. Cooper, A. Pedretti, G. Ambrosetti and A. Steinfeld, InPhoCUS (Inflated Photovoltaic Ultra-light Mirror Concentrators): First Results Of The Project And Future Perspectives, CPV-7, 2010, to be published. [4] M. Pravettoni, R. Galleano, T. Aitasalo, R. P. Kenny, E. D. Dunlop and K. W. J. Barnham, From an Existing Large Area Pulsed Solar Simulator to a High Intensity Pulsed Solar Simulator: Characterization, Standard Classification and First Results at ESTI, Thirty-fifth IEEE PVSC, 2010, pp [5] IEC , Photovoltaic devices - Part 9: Solar simulator performance requirements, Ed. 2.0 (2007). [6] IEC , Photovoltaic devices - Part 1: Measurement of photovoltaic current-voltage characteristics, Ed. 2.0 (2006). [7] M. Pravettoni, R. Galleano, R. Fucci, R. P. Kenny, A. Romano, M. Pellegrino, T. Aitasalo, G. Flaminio, C. Privato, W. Zaaiman and E. D. Dunlop, Characterization of high-efficiency c-si CPV cells, Prog. Photovolt: Res. Appl. (2011). Retrieved May 12, 2011 from Wiley Online Library (wileyonlinelibrary.com). DOI: /pip.1101 [8] ISO-IEC Guide 98-3, Uncertainty of measurements Part 3: Guide to the expression of uncertainty in measurement (GUM: 1995), Ed. 1.0 (2008). [9] M. Pravettoni, R. Galleano, E. D. Dunlop and R. P. Kenny, Characterization of a pulsed solar simulator for concentrator photovoltaic cell calibration, Meas. Sci. Technol. 21, 2010, (8 pp). [10] IEC , Photovoltaic devices - Part 3: Measurement principles for terrestrial photovoltaic (PV) solar devices with reference spectral irradiance data, Ed. 2.0 (2008). [11] H. Müllejans, W. Zaaiman and R. Galleano, Analysis and mitigation of measurement uncertainties in the traceability chain for the calibration of photovoltaic devices, Meas. Sci. Technol. 20, (12 pp) (2009). REFERENCES [1] A. Jäger-Waldau. (2010 September). PV Status Report Research, Solar Cell Production and Market /11/$ IEEE