Additional experimental details can be found in [14]

1.2 Solar simulator and spectral response results

The table 1 summarises the results gained from the I-V curve measurements (AM1.5g), as well as the results obtained by integrating the spectral response (SR) multiplied by the solar spectrum density. The initial current density of the cells was in the range of 30-31 mA/cm2. Both groups of cells show an increase in the short circuit current after encapsulation. The group with the AR layer outperforms nevertheless the group prepared without the AR layer in average by 2.65%. The results obtained by measuring the spectral responses are similar and the calculated short circuit current is around 2.73% higher with the AR layer.

The small difference of (2.73-2.65)%=0.11% between the results obtained with the direct Isc measurements (solar simulator) and the results obtained from the SR measurements indicate that spectral mismatch under the solar simulator did not influence the direct Isc measurements significantly. By considering the statistical nature of our measurement error [9,14], we estimate that the global improvement of the short circuit — current is 2.65%±0.25 %, which will be the value taken for the simulation of the yearly energy yield. The higher standard deviation for the results of group AR is due to a single cell which showed a lower improvement of Isc after embedding.

Table 1: Summary of the results obtained on the 22 mini-modules. NG (Normal Glass) indicates the group of the mini-modules without AR layer.

Group

Name

improvement in Isc solar simulator[%]

Improvement in Isc spectral response [%]

Modules NG

2.69

2.95

Standard deviation

0.22

0.32

Modules with AR layer

5.34

5.68

Standard deviation

0.48

0.66

Gain AR Layer

2.65

2.73

Figure 2a shows the spectral responses before and after encapsulation, averaged for the two groups of cells. For all mini-modules, a degradation of the SR below 400 nm is observed, which is linked to glass and EVA absorption. The SR is significantly improved for both groups in the range of 400 to 500 nm. This is mainly linked to a good match between the glass refractive index and the SiN/Si system. The AR layer brings an enhanced contribution in the range of 600-1000 nm. For the curves of the modules with the AR layer (Fig.2a), the apparent improvement of the SR has a minimum around 600­700 nm, which results from the fact that the reflection minimum before encapsulation (i. e., the reflection minimum given by the SiN layer) was already in that range, and could hence not be further improved. Although the effect is difficult to quantify, we estimate that diffuse light reflection on the screen-printed fingers followed by total internal reflection in the glass and subsequent collection by the cells, has a positive effect on the current collection in for both the aR and NG glasses cases, probably around 1%. Fig. 2b displays the results differently by showing the increase G in percent of the SR given by the Ar layer compared to the normal glass, according to

G = Gain [SRar (A)]-Gain [SRvg(^)],

where Gain [SR] = 1-SRafter/SRbefore-

It shows a broad maximum around 900 nm where it reaches around 3.5%. Note that the results are noisier below 400 nm and above 1000 nm because of the reduced system sensitivity at these wavelengths (lower cell response and weaker lamp intensities).

Wavelength [nm] a)

&_

<D

>,

ГО

СҐ.

<

.c

1

c

‘ro

О

400 600 800 1000 1200

Wavelength [nm] b)

5 4 3 2 1 0

Fig. 2. a) Spectral response as a function of the wavelength before and after encapsulation for the two types of glasses. b) Gain in current in percent given by the the AR layer when compared to the case without layer.