Characterization of the laser induced doping concentration

For the determination of the effectiveness of the laser doping process and for the characterization of the induced doping profile, small areas (2x2 cm²) of the a-Si1-xCx

coated c-Si substrates were fully laser processed. The doping polarity of the substrate was chosen in opposition to the doping of the film, allowing therefore for the determi-nation of the sheet resistance of the laser doped region at the substrate surface by means of a four point probe measurement. Subsequently, the same samples were used

Silicon solar cells with a-Si1-xCx rear side schemes 105

for the determination of the induced doping profile by means of the spreading resis-tance profiling (SRP) technique [184]. The laser used for all PassDop processes is a Rofin StarCut Disc 100 ICQ disc laser at 1030 nm and a pulse width of 1 µs. The laser power is adjusted by a variable optical attenuator to ensure the stability of the beam parameters. The beam is directed to the substrate with a galvanometer scanner system which allows high processing speeds (> 15 m/s) and thus point spacing up to 1 mm at a repetition rate of 15 kHz. For the fully processed samples, the spatial pulse overlap was kept small to ensure similar processing conditions as for point contacts.

Laser doping from doped Si-rich a-Si1-xCx films – In a first approach Si-rich, 30-40 nm thick a-Si_1-xC_x single layers were investigated. The resulting sheet resistance as a function of the laser power applied is shown in Fig. 6-7. The measurements evidence a threshold at around 2.5-3.0 W for the occurrence of laser doping. At the threshold, high sheet resistance and standard deviations point to the onset of dopant incorporation into the substrate. At lower laser power, no melting takes place revealed by the meas-ured base resistance of 40 Ω/sq for 250 µm thick 1 Ωcm material. Higher laser power result in a virtually constant sheet resistance pointing to the melting of the a-Si_1-xC_x film and the silicon volume underneath. The entire doping content of the film is dis-solved in this volume and incorporated in the material after recrystallization. Depend-ing on the initial dopDepend-ing concentration in the film, the resultDepend-ing sheet resistance does therefore vary only little with laser power (> threshold) in accordance with

) ,

Fig. 6-7: Sheet resistance vs. laser power of fully laser processed samples. Analysis of laser doping from a single phosphorous (left) and boron (right) doped a-Si1-xCx

film into crystalline silicon from opposite polarity. Altered from [185].

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

q referring to the elementary charge, µe/h to the minority carrier mobility, W to the junction depth and N to the impurity density as a function of the depth z. Minor varia-tions in Rsheet can be related to differing doping profiles and hence varying mobilities.

The laser induced phosphorous and boron profiles are displayed in Fig. 6-8 left and right, respectively. Note that the step-like course of the profiles is due to artifacts of the SRP technique. The measurements reveal a direct correlation between the profile depth (volume of molten silicon) and the laser power. The finding of a quasi linear relation (Fig. 6-9) is in good agreement with theoretical predictions. The latter is valid for laser fluencies around the melting threshold and was experimentally observed by other authors [186, 187]. Conversely, the surface density of the doping species de-creases with increasing laser power. Consistently with the results from the four point

Fig. 6-8: Doping profiles for various laser powers from a single phosphorous (left) and boron (right) doped a-Si1-xCx film measured by SRP. Altered from [185].

Fig. 6-9: Surface doping density and profile depth as a function of laser power for phospho-rous (left) and boron (right) doped a-Si1-xCx films. Altered from [185].

Silicon solar cells with a-Si1-xCx rear side schemes 107

probe measurements, the integrated SRP profiles yield a quasi constant amount of doping atoms for laser power exceeding the threshold. A comparison with the amount of doping atoms available in the a-Si1-xCx film prior to laser processing measured by SIMS shows a discrepancy to the value obtained by integration of the doping profile (Fig. 6-10). Besides inaccuracies in the respective measurements, the different meas-urement principles of the SIMS and SRP techniques themselves may account for this difference: SIMS detects any impurity atom whereas SRP is only sensitive to electri-cally active dopants. A further issue is the possible ablation of dopant containing mate-rial during the laser process (see chapter 6.2.4).

Laser doping under variation of the source layer – As expected, the thickness of the source layer is a crucial parameter for laser doping since it determines the absolute amount of impurity atoms available during the process. This is illustrated in Fig. 6-11 left. Si-rich, phosphorous doped (PH3=100 sccm) a-Si1-xCx films of about 15 nm thick-ness result in sheet resistance of 35 Ω/sq. Fig. 6-11 right reveals the impact of different doping concentrations in the film as well as the influence of the film composition. The latter might be due to a matrix related ability for the incorporation of dopants during deposition of the film, although SIMS measurements of C-rich layers suggest that this effect cannot account for the significant discrepancy between Si- and C-rich films in terms of laser doping efficiency. A different aspect refers to the prevailing bonds in the matrix and, therefore, the melting point of the material. Since C-rich films exhibit a large amount of energetically strong Si-C bonds (see Table 5-1), the melting point is shifted to higher temperatures as compared to their Si-rich counterparts. During the laser process, the C-rich film presumably does not melt (giving rise to a comparably

Fig. 6-10: Comparison between integrated doping density of SRP-profiles and doping content in the a-Si1-xCx film measured by SIMS. Left: a-Si1-xCx (n) with PH3=50 sccm.

Right: a-Si1-xCx (p) with B2H6=60 sccm.

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

slow diffusion from the solid film into the molten silicon) or does not melt entirely.

Both possibilities result in a decreased doping efficiency.

Introducing an intrinsic, Si-rich a-Si1-xCx film between the substrate and the doping source apparently does not alter the resulting sheet resistance (Fig. 6-12 left). The same is true for stack systems exhibiting an additional C-rich silicon carbide layer on top of the doping source (Fig. 6-12 left). Consistently with the preceding observations, the doping of the C-rich film does not further enhance the doping efficiency. The latter multi-layer stack is of major interest for the solar cell rear side application as it allows for a highly efficient surface passivation (intrinsic a-Si1-xCx) and an effective optical

Fig. 6-11: Sheet resistance as a function of applied laser power. Left: dependence on thick-ness of Si-rich, phosphorous doped a-Si1-xCx film. Right: dependence on carbon and phosphorous content in the film.

Fig. 6-12: Sheet resistance as a function of applied laser power. Left: dependence on thick-ness of Si-rich, intrinsic intermediate layer. Right: evaluation of performance of Si-rich/C-rich stacks.

Silicon solar cells with a-Si1-xCx rear side schemes 109

confinement due to the low refractive index of the C-rich film (n ≈1.8).