Silicon surface passivation via ultra-thin SiO2, TiO2, and Al2O3 layers

• Main Author: Ek, Anton
• Format: Others
• Language: English
• Published: Luleå tekniska universitet, Institutionen för teknikvetenskap och matematik 2019
• Subjects: surface passivation; ALD; silicon; solar cell; SiO2; Al2O3; TiO2; stacks; characterization; optimization; RSM; temperature; Materials Engineering; Materialteknik
• Online Access: http://urn.kb.se/resolve?urn=urn:nbn:se:ltu:diva-75913

Energy traps at the silicon surface originating from discontinuities in the lattice is detrimental to the performance of solar cells. Acting as recombination centers, they offer a location where the charge carriers may easily return to their original energy band after excitation. Surface passivation is an effective method to combat this and can be done either by suppressing traps (lowering trap density) or by forming an electric field, preventing the carriers from reaching the defect states. Silicon oxide, SiO2, and aluminum oxide, Al2O3, are two materials which have previously been shown to provide good passivating qualities. In this thesis, SiO2 and Al2O3 have been used both as single layers and in a stack configuration to passivate the surface of crystalline silicon (c-Si). Using a response surface methodology approach, temperature optimization with respect to deposition and annealing temperature has been conducted for SiO2/Al2O3 stacks deposited with plasma-enhanced atomic layer deposition, PEALD. It was shown that the same deposition temperature (Tdeposition = 140 °C, Tanneal = 395 °C) could be used for both materials and provide good passivation with an effective surface recombination velocity, Seff, of 5.3 cm/s (1Ωcm n-type Si wafers). From FTIR measurements, an increase in hydroxyl groups was seen as the SiO2 deposition temperature increased while the opposite was observed for Al2O3 which also showed fewer carbon related impurities with increasing temperature. Increasing the SiO2 temperature strongly affected the fixed charge density, causing it to decrease and even switch polarity. The fixed charge density could also be controlled by varying the thickness of the intermediate SiO2 layer. At a thickness of 1-2 nm, a minimum in the effective lifetime was observed and was correlated to Si close to flat-band conditions. N-type wafers showed a larger negative fixed charge density than p-type wafers which results in stronger field-effect passivation. For phosphorous doped emitters (200 Ω/sq on 10 Ωcm p-type wafer), it was seen that SiO2/Al2O3 stacks with a SiNx anti-reflection coating performed better than SiO2 or Al2O3 single layers. By depositing SiO2 at 130 °C in SiO2/Al2O3 stacks and annealing at 450 °C, an implied open circuit voltage (iVoc) of 710 mV was measured (AM1.5G) together with an implied fill factor (iFF) of 84.1% and a recombination parameter (J0) of 19.2 fA/cm2. Al2O3 single layer showed an extremely low J0 of 10 fA/cm2 but suffered from a decreased iFF and strong injection dependent lifetimes which originates from an inversion layer. ALD ozone processes were successfully developed for SiO2 and Al2O3.  The deposition rate per cycle for SiO2 was found to be only ~0.175 Ǻ/cycle (PEALD ~1.1 Ǻ/cycle), making it rather unpractical for use outside of research. Single layer SiO2 deposited with ozone showed, similarly to a plasma process, almost no surface passivation. Al2O3 however proved to be highly passivating on its own with a τeff = 3.8 ms, Seff = 1.2 cm/s (1 Ωcm n-type) after depositing at 250 °C. Studies on the effect of annealing showed that an annealing temperature of 450 °C is necessary to completely activate the passivation. The low Seff values were attributed to a very high negative fixed charge density ~1013 cm-2 together with strong chemical passivation.