n-type solar cells and characterization

Solar cell process and results – In order to investigate the newly developed rear passivation and contact scheme at the solar cell level, n-type cells with PERL structure (Fig. 6-18 left) were fabricated on (100), 1 Ωcm, n-type Fz silicon wafers with a thick-ness of 230 µm. These cells (2×2 cm²) feature a front surface with inverted pyramids and evaporated Al/Ti/Pd/Ag front contacts defined by photolithography which are thickened by electroplating. A BBr3 diffusion at 890°C followed by a drive-in oxida-tion at 1050°C result in a homogeneous boron emitter with a sheet resistance of 140 Ω/sq. Note that the cells do not feature an additional diffusion for the formation of selective emitter regions. This front side boron emitter is passivated by a layer stack consisting of a 10 nm Al2O3 film [162] followed by SiNx as anti-reflection coating.

The deposition of the Al₂O₃ was performed by plasma-assisted ALD (on an Oxford Instruments OPAL™ setup) at a temperature of 170°C. The PECVD SiNx was depos-ited at 400°C (SINA XS, Roth&Rau). The rear side of the cell is processed as shown in Fig. 6-18 right: analogous to the lifetime samples, a PECVD silicon carbide multi-layer stack (a-Si1-xCx(i)/a-Si1-xCx(n)/C-rich) is deposited onto the whole rear surface

Fig. 6-18: Left: high-efficiency PERL structure for n-type solar cells with industrial feasible local rear BSF by laser doping from passivating a-Si1-xCx stack (PassDop ap-proach). Right: Schema of the PassDop process sequence for the rear passivation and contacting of high-efficiency n-type silicon solar cells.

Silicon solar cells with a-Si1-xCx rear side schemes 115

followed by the laser process which simultaneously opens the rear dielectric films (local contact formation) and leads to a local n⁺ region beneath the contact points. The point spacing was varied between Lp =600-900 µm. No cleaning or etching step is needed after the laser step. Subsequently, a 2 µm thick aluminium layer is evaporated on top by an e-gun process. Finally, the cells are annealed at 350°C for 15 min to decrease the front contact resistance and to anneal surface defects related to e-gun front and rear metallization.

The one-sun parameters of the fabricated n-type solar cells measured on an aper-ture area of 4 cm² are summarized in Table 6-2. The best cell exhibits an open-circuit voltage (Voc) of 701 mV, a short-circuit current density (Jsc) of 39.8 mA/cm² and a fill factor (FF) of 80.1 % resulting in a solar cell efficiency of 22.4 %. The solar cell batch shows the very high process stability resulting in an average of 698 mV and 79.3 % for the Voc and FF of 76 processed solar cells, respectively. The dependence of Voc and the FF on the applied laser pitch is shown in Fig. 6-19 right. A minor impact of the point spacing is observed accounting for a maximum deviation of only 5 mV in open circuit voltage and 2 % in fill factor. Voc slightly increases whereas the FF decreases with spacing. This is in accordance with a reduced surface recombination due to a lowered metal contact fraction and an increased series resistance.

The very high open-circuit voltage in combination with a good fill factor for this industrially feasible rear side approach proves the very high performance of the ap-plied rear passivation and contacting scheme. A very effective suppression of carrier recombination at the rear side is evidenced by an excellent internal quantum efficiency (IQE) in the wavelength region between 900-1200 nm (Fig. 6-19 left). Furthermore, a comparison between quantum efficiencies measured at different bias light intensities revealed no impact on the rear passivation of the cells. Optically, the C-rich silicon

2internal measurements

3confirmed by Fraunhofer ISE CalLab

Table 6-2: One-sun parameters (aperture area of 4 cm²) of n-type cells featur-ing a high-efficiency front side and the PassDop passivation and contacting scheme on the rear.

Voc(mV) Jsc(mA/cm²) FF (%) η (%) average (76 cells)² 698±3 40.1±0.2 79.3±0.8 22.2±0.2

best cell³ 701 39.8 80.1 22.4

116 Silicon solar cells with a-Si1-xCx rear side schemes

carbide layer acts as an effective rear reflector with effective internal reflection values of (93±1) % including the metallization points.

Approximation of Scont from cell parameters - Of special interest is the perform-ance of the laser induced local n⁺ BSF and its influence on the cell parameters. Again, the most valuable quantity for the characterization of the local high-low junction is the effective surface recombination velocity at the contact point Scont (now referring to metallized surfaces) which directly quantifies its electrical impact. An approximation of Scont can be performed departing from the one diode equation

.

With it, the total saturation current density of the best cell J0,total =J0e+J0b results in 56 fA/cm². Assuming a very well passivated front surface and a value of 30 fA/cm² for the emitter saturation current density J0e as proposed by Benick et al. for an equal front cell structure [56], the base saturation current for our cell amounts to 26 fA/cm². As-suming a bulk diffusion length of L =0.23 cm (τbulk=4.5 ms), an upper limit for Seff at the rear side of the cell can be calculated by

( ) ( )

In this equation, q refers to the elementary charge, Dh is the diffusivity of the holes, ni

is the intrinsic carrier concentration and ND is the doping concentration of the bulk.

The calculated Seff amounts to 6 cm/s, which agrees well with values determined from the lifetime experiments (see Fig. 6-14). Using a value of Spass=3 cm/s for the surface

Fig. 6-19: Left: reflection (R), external and internal quantum efficiency (EQE/IQE) of a high-efficiency n-type cell with PassDop rear side scheme. Right: dependence of open circuit voltage (Voc) and fill factor (FF) on the point spacing of the laser process.

Silicon solar cells with a-Si1-xCx rear side schemes 117

recombination velocity of the passivated area between the laser points measured on lifetime samples with a-Si1-xCx multi-layer stack and considering the measured 40 µm for the visible diameter of the laser points, eq. (6-4) yields a Scont of approximately 2000 cm/s. This is in the same order of magnitude as the value previously determined at the lifetime level. For comparison, Benick et al. report on a S_cont of 55 cm/s for their deep diffused phosphorus LBSF [56]. However, considering the simplicity of the process, the reached quality of the laser-induced BSF is very promising for the indus-trial implementation of a n⁺ LBSF.