Improved understanding of boron-oxygen-related carrier lifetime degradation and regeneration in crystalline silicon solar cells

Abstract

This thesis examines carrier lifetime instabilities in oxygen-rich boron-doped p-type Czochralski-grown silicon (Cz-Si) with the focus on the permanent deactivation of the boron-oxygen (BO)-related defect center leading to a regeneration in lifetime. In order to resolve contradictory statements reported previously in the literature concerning the mechanism of regeneration in this thesis, passivated emitter and rear solar cells (PERCs) fabricated on boron-doped p-type Cz-Si wafers are regenerated in darkness by carrier injection via application of a forward-bias voltage V_appl at elevated temperatures. The regeneration kinetics is analyzed under regeneration conditions by measuring the total recombination current of the solar cell at the actual regeneration temperature at varying applied voltages Vappl. In parallel, we measure the electroluminescence signal emitted by the solar cell at different time steps during regeneration to directly determine the injected excess carrier concentration Delta{n} at each applied forward-bias voltage V_appl. The deactivation rate constant R_de of the BO defect is determined from the measured time-dependent cell current. The experimental results show unambiguously for the first time that R_de increases proportionally with Delta{n} during the regeneration process, solving the inconsistencies reported in the literature under actual regeneration conditions. To identify the impact of hydrogen on the BO-related lifetime degradation and regeneration kinetics, different amounts of hydrogen are introduced into the silicon bulk by rapid thermal annealing (RTA) treatment in an infrared conveyor-belt furnace quantified by measurements of the silicon resistivity increase. The silicon resistivity increases under dark-annealing due to hydrogen passivation of boron dopant atoms. The hydrogen source in our experiments are hydrogen-rich silicon nitride (SiN_x:H) layers deposited on the silicon wafer surfaces. Varying the peak-temperature of the RTA step indicates that there exists a temperature-dependent maximum in the hydrogen content introduced into the silicon bulk. The exact position of this maximum depends on the composition of the SiN_x:H layers. The highest total hydrogen content, exceeding 10^15 cm^{-3}, is introduced into the silicon bulk from silicon-rich SiN_x layers with a refractive index of n =2.3 (at a wavelength of lamda = 633 nm) at an RTA peak temperature of 800°C. Adding a 20 nm thick Al_2O_3 interlayer in-between the silicon wafer surfaces and the SiN_x:H layers, reduces the in-diffused hydrogen content up to a factor of four, demonstrating that Al_2O_3 acts as a highly effective hydrogen diffusion barrier. By varying the Al_2O_3 thickness, the hydrogen bulk content is varied over more than one order of magnitude. In order to examine the impact of hydrogen on the degradation kinetics, all samples are illuminated at a light intensity of 0.1 suns near room temperature. No influence of the in-diffused hydrogen content on the degradation rate constant is measured, confirming that hydrogen is not involved in the BO degradation mechanism. The regeneration experiments at a light intensity of 1 suns at elevated temperatures, however, show a clear dependence on the hydrogen content with a linear increase of the regeneration rate constant with increasing bulk hydrogen concentration. An extrapolation of this correlation towards a zero in-diffused hydrogen content shows that the regeneration is still working even without any in-diffused hydrogen. Hence, our experiments clearly reveal for the first time that two distinct regeneration processes are taking place, one involving hydrogen, the other not. These results confirm a previous theoretical model, which had not been experimentally verified so far. In another series of experiments, we examine the long-term stability of the carrier lifetime in boron-doped Cz-Si materials with different boron and oxygen concentrations after regeneration in an industrial belt furnace. After firing and subsequent regeneration in the conveyor-belt furnace, the silicon samples are exposed to longterm illumination at an intensity of 0.1 suns and a sample temperature of about 30°C for more than two years. After regeneration, the lifetime samples re-degrade (30-72% reduced compared to the degradation observed without regeneration step). This re-degradation is attributed to an incomplete regeneration within the belt furnace due to the short regeneration period. All in all, the industrial process consisting of firing with subsequent regeneration in the same belt-furnace unit seems to be very effective for industrially relevant silicon materials. Typical industrial silicon wafers with a resistivity of (1.75+-0.03) Ohmcm and an interstitial oxygen concentration of (6.9+-0.3) x 10^{17} cm^{-3} show lifetimes larger than 2 ms after regeneration and two years of light exposure.