Acoustic transmission of photovoltaic mini-modulesAbstractWe have measured sound velocities in photovoltaic mini-modules before and after forced ageing.The mini-modules are a series connection of three high efficiency c-Si cells with a cell area of 6×6 mm2. Thesemodules are produced by IXYS Corporation. The modules under test were type KXOB22-04X3-ND from theIXOLAR palette [1]. The physical dimension as given in the datasheet was 22 mm×7 mm×1.8 mm with a massof 500 mg. It is encapsulated and intended predominately for indoor use in small electronic devices.

IntroductionWith the increased use of photovoltaic electricity generation surveillance the of degradation becomes importantin order to prevent a drop out of parts or the whole power plant. Currently the methods of degradation monitor-ing focus on the electrical performance of the cells themselves. IR and electroluminescence imaging are widelyin use [2,3]. Much less attention is paid to changes of the rest of the modules, like encapsulation though in sev-eral cases like the potential induced degradation effect they were found to be responsible for cell degradation [4].Sound velocity measurements are a well established technique for material fatigue and failure recognition [5].

ExperimentalAcoustic transmissionThe set up for the acoustic transmission measurements was based on the sonic viewer from Oyo corporation,model 5210. For the experiments two matched - in frequency and size - piezo trancducers for shear wave gen-eration/reception were used. The resonance frequency was 33 kHz. The circular surface had a diameter of31.3 mm. Originally the instrument was intended for sound velocity measurements of geological samples. Itconsists of a precise time base, a high voltage pulse generator which excites oscillations of the transmitter and apreamplifier for the receiver with an output that could be directly connected to a digital storage oscilloscope,DSO. The measurement procedure consists of two steps. First the receiver was placed directly onto the trans-mitter and the signal captured by the DSO was stored in the PC. Then the sample was inserted and placed in thecentre, The resulting signal was stored again.In our case the sample’s area is smaller than the transducer’s area and the thickness is much smaller than thewavelength of the sound wave. For this reason an excitation which is perpendicular to the direction of propaga-tion was chosen. In order to estimate how suitable the set up could be for our measurements model calculationswith a copper sample of the module’s dimensions are made. The sound velocity of 2325 m/s for shear waves incopper was taken from the literature [6]. For a frequency of 33 kHz the sound wavelength, λ0 in copper is calcu-lated to be 70 mm. The quarter wavelength condition for destructive interference is still about 10 times the sam-ple’s actual thickness of 1.8 mm. Furthermore absorption and internal reflection on boundaries are expected tobe negligible. Therefore the observable attenuation should arise solely from the ratio of the sample’s surfaces tothe transducer’s and receiver’s surfaces. In our case that is 154 mm2/770 mm2=0.2. Due to the increased distancebetween transmitter and receiver the sound wave will be received with delay. For 1.8 mm copper the receivedsignal will be delayed for ~0.8µs compared to the observation with directly coupled transducers.Our experimental observation with one of the mini-modules resulted in an attenuation of the amplitudes of 0.18

Fig. 1: Plots of the recorded sound waves in the time domain (left graph) and the frequency domain(right graph).

and a time delay, ∆t=0.96 µs. Both values are in agreement with our expectation from the model calculation forcopper. The overall shear wave sound velocity for the composite structure of the module was calculated to beabout 1900 m/s. The time delay was directly measured by overlying the two received curves in the time domain.The left graph in Fig. 1 shows the plots enlarged around a zero crossing of the amplitudes. The black line corre-sponds to the measurement with transmitter and receiver face to face and the red line with the inserted module.The right graph show both signals in the frequency domain together with an inset of the experimental arrange-ment.Forced ageing of a mini-moduleIn order to find out how sensitive the acoustic transmission measurement can be to physical changes of the mod-ule a time series with one module exposed to elevated temperatures was made. Beside the acoustic transmissionmeasurement the mass, thickness variations and the electroluminescence was determined in time increments ofdays during the sample’s heat treatment.A temperature controlled drying unit was used to keep the sample at a constant temperature of 87°C under ambi-ent air conditions.Monitoring the weight of the sample is a fast and easy indicator of changes within the module due to heating.The first time when the sample was weighted was one year ago and a mass of m=514.2 mg was found. Immedi-ately before the module was transferred to the heating chamber m=514.45 mg was measured.The sample’s thickness was measured in order to derive the sound velocity from the transfer time of then acous-tic shear wave through the sample. The thickness in the centre of the module was determined with a profilometerfrom Heidenhain with a resolution of 1 µm. A thickness of 1.838 mm was measured prior to the heating treat-ment.The EL intensity of the module was recorded in order to observe whether changes in the cell’s performance takeplace during the heat treatment. A glass fibre bundle with a diameter of ~3mm was placed above the centre ofeach of the three individual cells in the module. The emitted EL intensity was detected by an InGaAs photodiode(DET10C of Thorlabs). The module current was varied from 1 mA up to 35 mA in increments of 2 mA. Withrespect to the short circuit current, Isc of ~15 mA under STC [1] the current variations cover a range from 0.1 –2 × Isc. The signal of the middle cell before the heat treatment was taken as reference for all subsequent record-ings.

ResultsThe results of the time series are summarised in table 1. In Fig. 2 the relative change from the initial values ofthe measured parameters are plotted. The mass as well as the measured thickness in the middle of the moduledecreases continuously. After one week a reduction of more than one per cent was observed. Within the resolu-tion of our experiment the EL intensity remains unchanged indicating that the heat treatment has no effect on thecells performance. Although the acoustic transmission measurements have a low resolution compared to theother measurements a significant reduction of the sound velocity from 1800 m/s to 1200 m/s was recordedalready after one day at elevated temperatures.

Table 1: Summary of measured module parameters

EL intensity [a.u.]Date/Time Mass

[mg]

Thickness

[mm]

Time-Delay

[µs]

Velocity

[m/s] Left Centre Right

18.8/11:00 514.45 1.838 1.01±0.13 1820 1.00±0.04 1.00 0.87±0.02

19.8/11:00 511.2 1.834 1.56±0.13 1176 1.01±0.06 1.01±0.05 0.90±0.04

20.8/9:30 510.6 1.821 1.56±0.12 1167 1.01±0.10 0.97±0.10 0.90±0.05

22.8/9:30 509.5 1.816 1.73±0.12 1050 1.00±0.06 0.98±0.07 0.88±0.06

26.8/9:30 508.2 1.811 1.44±0.07 1258 1.00±0.06 0.99±0.04 0.89±0.05

It seems that the sound velocity responds faster to temperature stress than mass or thickness which are in goodcorrelation with each other.

DiscussionWe have investigated the sound velocity of a 1.8 mm thin photovoltaic module composed of several layers ofunknown materials. A measurement series was conducted where the module was exposed to high temperatures.A reduction of the sound velocity from greater than 1800 m/s to below 1200 m/s was observed after 1 week. Ourresults show that the sound velocity was highly sensitive to changes in the mechanical properties of the module.With the exception of about 200 µm crystalline silicon which do not exhibit a resolvable ageing effect of itsopto-electrical properties thickness and composition of all other layers in our sample was not known. It isunlikely that the sound velocity of crystalline silicon was affected by the thermal treatment. No furtherinterpretation of the results we have made can currently be given. However the experiments were primarilyintended as a proof of our concept.

References[1] Datasheet „KXOB22-04X3-DATA-SHEET-20110808.pdf“ from IXYS. Online i.e. at

http://www.digikey.com/product-highlights/us/en/ixys-ixolar-solarbit/823

Fig. 2 Observed changes in the module's mass and thickness (a), emitted EL intensity(b) and sound velocity for shear waves (c). Dashed lines are guidlines to the eye.

[2] T. Kirchartz, A. Helbi, W. Reetz, M. Reuter, J. H. Werner, U. Rau1, “Reciprocity between electrolumines-cence and quantum efficiency used for the characterization of silicon solar cells”, Prog. Photovolt: Res.Appl., 17, 2009, 394-402, doi: 10.1002/pip.895.

[3] Morgan D. Bazilian, Harry Kamalanathan, D.K. Prasad, „Thermographic analysis of a building integratedphotovoltaic system“, Renewable Energy, 26, 2002, 449-461, doi: 10.1016/S0960-1481(01)00142-2

[4] V. Naumann, D. Lausch, A. Graff, M. Werner, S. Swatek, J. Bauer, A. Hähnel, O. Breitenstein, S. Großer, J.Bagdahn, C. Hagendorf, „The role of stacking faults for the formation of shunts during potential-induceddegradation of crystalline Si solar cells“, Phys. Status Solidi RRL 7, No. 5, 2013, 315–318, doi10.1002/pssr.201307090

[5] “Harris‘ Shock and Vibration Handbook”, A.G. Piersol, T. L. Paez (editors), McGraw-Hill, 6th edition, 2010[6] David R. Lide, CRC Handbook of Chemistry and Physics, 7th Edition, 1990-1991.

AppendixAppended are the sound measurements of the other samples summarised in the table below.

Sample mass [mg]original

mass [mg]aftertreatment

∆m [mg] d [mm] ∆t [µs] Velocity[m/s]

A1 484.3 480.6 -3.7 1.682 1.14±0.19 1475

A2 476.1 472.1 -4.0 1.624 1.55±0.38 1048

A3 513.4 508.5 -4.9 1.798 1.72±0.25 1045

A4 488.6 483.8 -4.8 1.681 2.01±0.24 836

A5 500.0 494.0 -6.0 1.741 1.37±0.08 1271

A6 511.0 510.5 -0.5 1.812 1.17±0.15 1549

A7 513.4 512.3 -1.1 1.826 1.04±0.11 1756

A8 513.2 512.0 -1.2 1.818 0.98±0.09 1855

A9 514.2 514.45 +0.25 1.838 1.01±0.13 1820

The plots in the following Fig.s are derived from the table.

Fig. 3 Correlation between thicknessand weight of the samples.

Fig. 4 Linear correlation betweenshear wave sound velocity and loss ofmass due to heat treatment.