Optical and electrical analysis

η [%]

EWT 591 37.4 75.1 16.6

Conv. with full area BSF 612 35.2 77.1 16.6

Conv. with local BSF 607 34.2 75.6 15.7

Table 4-4: Parameters of the Two-Diode model extracted from the dark, illuminated and Jsc-Voc

characteristics for BC-EWT solar cells and two types of conventional BCSCs.

Cell type J₀₁

[10^-12A/cm²]

J₀₂ [10^-8A/cm²]

R_sh [Ωcm²]

R_s [Ωcm²]

EWT 2.5 8.9 1600 0.72

Conv. with full area BSF 1.3 4.1 1900 0.62

Conv. with local BSF 1.5 4.3 1200 0.84

The previous short discussion of the IV-characteristics gives only a very basic understanding of the BC-EWT cells in comparison to the conventional ones. Further cell characterisation in the next sections will answer the following questions leading to a quantitative explanation:

• How large is the current gain due to the second carrier collection on the rear surface?

• What is the reason for the reduced Voc mainly caused by a significantly higher J01?

• What is limiting the fill factor of the EWT cell?

4.4.4 Optical and electrical analysis

The increase in Jsc for the EWT cell compared to the cell with full area BSF was 2.2 mA/cm² (6.3% rel.), to the one with local BSF 3.2 mA/cm² (9.4% rel.). This is caused by two effects:

1. A negligible shadowing loss of the front grid

2. An additional carrier collection at the p/n-junction at the rear surface

A detailed analysis of the current gain requires the measurements of the reflectivity as well as local LBIC-mappings, which will be presented in the first part. The analysis of the two contributions to the current gain is carried out after these characterisations.

400 500 600 700 800 900 1000 1100 0

20 40 60 80 100

EWT local BSF full BSF

EQE [%], refl [%]

λ [nm]

Figure 4.14: External Quantum Efficiency (EQE) and reflectivity of a BC-EWT cell and two designs of conventional BCSCs.

4.4.4.1 Spectral response and reflectivity measurements

The results of the measurements (EQE and reflectivity) on an illuminated cell area of 2x2 cm² are shown in Figure 4.14. The reduced shadowing losses of the EWT solar cell can be seen in the reflection curves. All three cell types have the same EQE in the short wavelength range up to about 800 nm. In this range the EQE is slightly lower than for other BCSCs processed in this work due to recombination in the higher doped emitter (Rsheet=80 Ω/sqr instead of 100 Ω/sqr). The highest EQE for wavelengths larger than 900 nm is obtained for the EWT solar cell mainly caused by double sided collection of the minority charge carriers.

A spectral analysis of the IQE was carried out to extract the effective bulk diffusion length (see Chapter 1). The effective bulk diffusion length is Leff=240 µm for the BCSC with full area rear contact and Leff=210 µm for the locally contacted one. The spectral analysis in this form can not be used for EWT solar cells, since the deviation of the spectral analysis is not valid for double sided collection. The reason for the difference in Leff for the two conventional BCSCs will be analysed by LBIC scans in the next section.

4.4.4.2 LBIC measurements

A spatially resolved characterisation of the three cell designs was carried out by SR-LBIC [Per02] and is illustrated as Leff-mappings in Figure 4.15. For the EWT solar cell, the analysis for the determination of Leff is not valid in the n-type regions on the rear. The black box in each mapping indicates the area, where the spectral response and reflectivity measurements of the previous section were carried out.

Figure 4.15: Mappings of the effective diffusion length Leff obtained by SR-LBIC. (top) EWT solar cell, (middle) conventional BCSC with full area rear contact ,(bottom) conventional BCSC with local rear contact. The black box indicates the regions where the spectral response and reflectivity measurements of the previous section were carried out.

The spatially resolved LBIC scan of the EWT solar cell clearly shows the increased current collection underneath the n-type finger contacts on the rear which can be seen in the horizontal linear structure. The circular structure originates from local variations in the bulk lifetime caused by a non-optimised thermal cycle during high temperature processing. Cz-Si should be processed with fast temperature ramps (e.g. loading and unloading at high temperatures), whereas mc-Si and FZ-Si is best processed with slow temperature ramps.

The high temperature processes in this work were optimised for mc-Si leading to non-optimum conditions for Cz-Si.

In the Leff-mapping of the conventional cell with full area rear contact, the circular structure is reduced. Only a small region in one corner has a lower Leff.

Two linear structures can be seen in the LBIC scan of the locally contacted conventional cell (see bottom in Figure 4.15). The one in horizontal direction is due to the finger metallisation. The higher current in the vicinity of the buried contact finger is due to the vertical emitter within the contact grooves which leads to an improved current collection.

The vertical structure in the LBIC scan is due to the local rear contacting scheme. In the lower right part of the cell corresponding to a high Leff, this vertical structure is not pronounced. This indicates, that the rear surface recombination velocity is similar for the LPCVD SiNx and for the thin evaporated Al-BSF which is in the range of 5000 cm/s. Only in regions with a high diffusion length, the rear surface recombination velocity influences Leff (see Figure 2.19). In the upper left part of the LBIC-mapping the vertical structure due to the local rear contact scheme is very pronounced. In this region, Leff is low and therefore any differences in rear surface recombination are not responsible for local variations in Leff. Local variations in this part are caused by differences in the bulk diffusion length LB. A possible explanation is an Al-gettering effect leading to an increase in LB on the regions covered by Al since regions coated with LPCVD SiNx can not be gettered. A further indication of this effect is the LBIC-mapping of the conventional cell with full rear contact, since a circular structure is almost not visible for this cell.

4.4.4.3 Current gain due to double sided charge carrier collection

The effect of double sided minority charge carrier collection leads to an increase in Jsc

and was investigated by several authors including [Ker00b], [Kre01], [Kna01]. Parts of the investigations in [Ker00b] are applied to the EWT solar cells of this work. The left side of Figure 4.16 shows the Collection Probability (CP) for (a) a conventional cell with unpassivated rear contact and for (b) a partly double sided collecting structure [Ker00b].

This figure shows that minority charge carriers generated deeply inside the bulk above the area with two emitters have a higher CP therefore leading to an increase in Jsc.

The right part of Figure 4.16 shows the relative current gain due to the double sided collection for different rear surface recombination velocities SB of the single sided cell for a cell thickness of 280 µm [Ker00b]. In these simulations the rear surface is completely covered with a second collecting junction. The relative difference in Jsc can be as high as 8%

if compared to a single sided cell with poor rear surface passivation. If the rear surface passivation of the single sided cell is improved, a current gain up to 5% can be achieved if the bulk diffusion length corresponds to about half of the cell thickness. For the EWT cells in this work, SB for the thin evaporated Al-BSF as well as for the regions passivated by LPCVD SiNx, will be about 5000 cm/s. The diffusion length is varying between 100 and 200 µm (see LBIC scans). For these parameters, the corresponding gain in Jsc is in the range of 5 to 7%.

Figure 4.16: (left) Collection probability (CP) of minority charge carriers for (a) a conventional solar cell with unpassivated rear contact and (b) partly double sided collecting cell for a bulk diffusion length of 100 µm. (right) Simulated current gain of double sided collecting designs compared to conventional designs for a wafer thickness of 280 µm (both from [Ker00b]).

In the investigated design of BC-EWT cells, a second collecting junction is only underneath the regions opened by laser ablation with a width of 400 µm and amount therefore to about 20% of the area. However, the physical width and the “electrically active width” of the n⁺⁺-region is not the same, as these regions are “smeared out” (see Figure 4.16). To further quantify this effect, linescans of the IQE at λ=980 nm are given for the EWT cell in Figure 4.17 for regions with high (black) and low (red) effective diffusion lengths. The electrically active width was determined as the width when the IQE reaches half of the value between high IQE (with the second collecting junction) and low IQE (LPCVD SiNx and Al-BSF, p-type regions). For the regions with a high Leff, the width is about 600 µm, for the region with a low Leff it is 750 µm. As an average value 700 µm was taken for the electrical width of the n-type fingers for the analysis performed in the next section since a large fraction of the EWT solar cell has a lower Leff.

2 3 4 5 6 7 8 9

50 55 60 65 70 75 80 85 90

750 µm 600 µm

IQE at λ=980 nm

distance [mm]

Figure 4.17: Linescans of the IQE determined from an LBIC-mapping at λ=980 nm. The black curve corresponds to the region with high Leff in the upper right part of the Leff-mapping in Figure 4.15, the red curve to the region with low Leff in the lower right part. The red curve intercepts the holes leading to dips in the regions of a high IQE.

4.4.4.4 Discussion

1. Current gain due to reduced shadowing losses

The non-active cell area for the EWT solar cell due to the holes is below 0.2%. The contact fingers of the conventional cells were slightly wider (45 µm) as in standard cell processing (30 µm) since the defect etch of the contact grooves was enhanced (see section 4.4.3). The shadowing losses of the contact finger therefore amount to 3%, from the tabbed busbar also to 3%. Therefore the reduced shadowing losses correspond to a difference in Jsc

of 2.2 mA/cm².

2. Current gain due to second collecting junction on the rear

The model calculations of [Ker00b], introduced in the previous section, showed, that a second collecting junction on the complete rear side leads to an increase in Jsc of 5-7%. The electrically active width of the n⁺⁺-type region on the rear corresponds to 700 µm wheareas the “physical” width was only 400 µm. Therefore the double sided collecting junction on the rear covers 35%. The increase in Jsc will be in the range of 1.8-2.5% corresponding to 0.6-0.85 mA/cm². In order to determine the experimental increase in Jsc due to rear side current collection, the difference between the EWT cell and the locally contacted conventional BCSC has to be taken, since these cells have a very similar distribution of Leff. The difference in Jsc between these two cells amounted to 3.2 mA/cm² leading to an increase of 1 mA/cm² due to current collection in the rear side junction. This is very close to the calculated increase in Jsc and lies within measurement uncertainty.