Fa-Jun Ma

Advanced modeling of the effective minority carrier lifetime of passivated crystalline silicon wafers

A strong injection level dependence of the effective minority carrier lifetime (τeff) is typically measured at low injection levels for undiffused crystalline silicon (c-Si) wafers symmetrically passivated by a highly charged dielectric film. However, this phenomenon is not yet well understood. In this work, we concentrate on two of those possible physical mechanisms to reproduce measured τeff data of c-Si wafers symmetrically passivated by atomic layer deposited Al2O3. The first assumes the existence of a defective region close to the c-Si surface. The second assumes asymmetric electron and hole lifetimes in the bulk. Both explanations result in an adequate reproduction of the injection dependent τeff found for both n- and p-type c-Si wafers. However, modeling also predicts a distinctly different injection dependence of τeff for the two suggested mechanisms if the polarity of the effective surface charge is inverted. We test this prediction by experimentally inverting the polarity of the effective surfac...

Deposition temperature independent excellent passivation of highly boron doped silicon emitters by thermal atomic layer deposited Al2O3

In this work, we demonstrate that by using H2O based thermal atomic layer deposited (ALD) Al2O3 films, excellent passivation (emitter saturation current density of ∼28 fA/cm2) on industrial highly boron p+-doped silicon emitters (sheet resistance of ∼62 Ω/sq) can be achieved. The surface passivation of the Al2O3 film is activated by a fast industrial high-temperature firing step identical to the one used for screen printed contact formation. Deposition temperatures in the range of 100-300 °C and peak firing temperatures of ∼800 °C (set temperature) are investigated, using commercial-grade 5″ Cz silicon wafers (∼5 Ω cm n-type). It is found that the level of surface passivation after activation is excellent for the whole investigated deposition temperature range. These results are explained by advanced computer simulations indicating that the obtained emitter saturation current densities are quite close to their intrinsic limit value where the emitter saturation current is solely ruled by Auger recombinatio...

Progress in Surface Passivation of Heavily Doped n-Type and p-Type Silicon by Plasma-Deposited AlO

We report an outstanding level of surface passivation for both n⁺ and p⁺ silicon by AlO_x/SiN_x dielectric stacks deposited in an inline plasma-enhanced chemical vapor deposition (PECVD) reactor for a wide range of sheet resistances. Extremely low emitter saturation current densities (J_0e) of 12 and 200 fA/cm² are obtained on 165 and 25 Ω/sq n⁺ emitters, respectively, and 8 and 45 fA/cm² on 170 and 30 Ω/sq p⁺ emitters, respectively. Using contactless corona-voltage measurements and device simulations, we demonstrate that the surface passivation mechanism on both n⁺ and p ⁺ silicon is primarily due to a relatively low interface defect density of <;10¹¹ eV^-1cm^-2 in combination with a moderate fixed negative charge density of (1-2) × 10¹² cm^-2. From advanced modeling, the fundamental surface recombination velocity parameter is shown to be in the order of 10⁴ cm/s for PECVD AlO_x/SiN_x passivated heavily doped n⁺ and p⁺ silicon surfaces.

_{\bm x}

Extremely low emitter saturation current density (J_0e) values of 6 and 45 fA/cm², respectively, are reported for 220 and 30 Ω/sq planar p⁺ boron emitters passivated by an AlO_x/SiN_x stack deposited in an industrial plasma-enhanced chemical vapor deposition (PECVD) reactor. The thermal activation of the AlO_x films is performed in a standard industrial fast firing furnace, making the developed passivation stack industrially viable. For textured p⁺ emitters the J_0e values are found to be 1.5 - 2 times higher compared to planar emitters. This excellent surface passivation is attributed to a high negative charge density of -(3-6)×10¹² cm^-2 in combination with a low interface defect density of ℒ10¹¹ eV^-1cm^-2. Assuming a short-circuit current density of 40 mA/cm² and the ideal diode law, the J_0e result for the 80 Ω/sq emitter represents a 1-sun open-circuit voltage limit of 736 mV at 25°C.

/SiN

This paper presents a three-dimensional numerical analysis of homojunction/heterojunction hybrid silicon wafer solar cells, featuring front-side full-area diffused homojunction contacts and rear-side heterojunction point contacts. Their device performance is compared with conventional full-area heterojunction solar cells as well as conventional diffused solar cells featuring locally diffused rear point contacts, for both front-emitter and rear-emitter configurations. A consistent set of simulation input parameters is obtained by calibrating the simulation program with intensity dependent lifetime measurements of the passivated regions and the contact regions of the various types of solar cells. We show that the best efficiency is obtained when a-Si:H is used for rear-side heterojunction point-contact formation. An optimization of the rear contact area fraction is required to balance between the gains in current and voltage and the loss in fill factor with shrinking rear contact area fraction. However, the corresponding optimal range for the rear-contact area fraction is found to be quite large (e.g. 20-60 % for hybrid front-emitter cells). Hybrid rear-emitter cells show a faster drop in the fill factor with decreasing rear contact area fraction compared to front-emitter cells, stemming from a higher series resistance contribution of the rear-side a-Si:H(p+) emitter compared to the rear-side a-Si:H(n+) back surface field layer. Overall, we show that hybrid silicon solar cells in a front-emitter configuration can outperform conventional heterojunction silicon solar cells as well as diffused solar cells with rear-side locally diffused point contacts.

_{\bm x}

Multidimensional numerical simulation of boron diffusion is of great relevance for the improvement of industrial n-type crystalline silicon wafer solar cells. However, surface passivation of boron diffused area is typically studied in one dimension on planar lifetime samples. This approach neglects the effects of the solar cell pyramidal texture on the boron doping process and resulting doping profile. In this work, we present a theoretical study using a two-dimensional surface morphology for pyramidally textured samples. The boron diffusivity and segregation coefficient between oxide and silicon in simulation are determined by reproducing measured one-dimensional boron depth profiles prepared using different boron diffusion recipes on planar samples. The established parameters are subsequently used to simulate the boron diffusion process on textured samples. The simulated junction depth is found to agree quantitatively well with electron beam induced current measurements. Finally, chemical passivation on planar and textured samples is compared in device simulation. Particularly, a two-dimensional approach is adopted for textured samples to evaluate chemical passivation. The intrinsic emitter saturation current density, which is only related to Auger and radiative recombination, is also simulated for both planar and textured samples. The differences between planar and textured samples are discussed.

Dielectric Stacks

The effective minority carrier lifetime of p-type silicon wafers passivated by silicon nitride and of n-type silicon wafers passivated by aluminium oxide often decreases significantly as the excess carrier concentration decreases. Several theories have been postulated to explain this effect. The main ones are asymmetric carrier lifetimes, high recombination within a surface damage region, and edge recombination. As in some cases, the effective lifetime measurements can be fitted quite well by all these effects, it is challenging to determine the main cause for the suppressed performance at low illumination. This is partly due to the fact that no study has yet included a sufficiently large set of wafers and advanced modelling to examine all these theories. The aim of this study is to determine the most likely theory based on a set of undiffused p- and n-type wafers of different sizes, passivated with both silicon nitride and aluminium oxide. Quasi-steady-state photoluminescence measurements were used in order to investigate effective lifetime at very low carrier densities, without artifact effects that commonly limit photoconductance-based measurements. Advanced modelling using Sentaurus was used to investigate the impact of different parameters—such as the fixed charge within the dielectric—on the recombination at the edge and within the surface damage region. These models were then used to simulate the measurement results. It is shown that asymmetrical surface lifetime cannot explain the observed reduction when the dielectric is highly charged (either positively or negatively). It is also shown that although edge recombination influences the effective lifetime at low excess carrier concentration, it alone cannot explain the effective lifetime reduction. It is therefore concluded that the presence of a surface damage region is the more likely explanation for the effective lifetime decrease of the studied wafers.

State-of-the-art surface passivation of boron emitters using inline PECVD AlO x /SiN x stacks for industrial high-efficiency silicon wafer solar cells

We present state-of-the-art results on boron emitter passivation (J_0e <; 25 fA/cm² and S_n0 <; 400 cm/s) with industrially fired positively-charged low-temperature PECVD SiO_x/SiN_x dielectric stacks deposited in an industrial reactor. These films feature a very low fixed charge density (~ + 6×10¹⁰ cm^-2) and excellent interface quality (D_{it, midgap} of ~3×10¹⁰ eV^-1 cm^-2) after an industrial firing step. Based on contactless corona-voltage measurements and device simulation, we explain the mechanism of surface passivation to be dominated by chemical passivation rather than field-effect passivation. With excellent optical and passivation properties, these films are suitable for high-efficiency cost-effective industrial n-type silicon wafer solar cells.

Three-dimensional numerical analysis of hybrid heterojunction silicon wafer solar cells with heterojunction rear point contacts

Efficient optimization of the boron-doped region of silicon solar cells requires reliable process simulation of boron tube diffusion. Established simulation models and parameters are mostly calibrated for complementary metal oxide semiconductor device fabrication, where the doping processes are significantly different from those used in solar cell fabrication. In this paper, we present models and a set of corresponding parameters that are suitable for process simulation of BBr3 tube diffusion for solar cell applications with Sentaurus TCAD. Experimental doping profiles obtained with a wide range of diffusion recipes are compared with simulation results. Additionally, with the process parameter sensitivity analysis, we demonstrate the dominant process parameters that alter the boron distribution profile and its effect on the electrical performance.

Two-dimensional numerical simulation of boron diffusion for pyramidally textured silicon

Effective minority carrier lifetime reduction at low injection levels is observed on 125 mm undiffused lifetime samples whose surfaces are under inversion due to field-effect passivation. With numerical analysis, we show that edge recombination is insufficient to account for this phenomenon on these samples. Between surface damage and asymmetric bulk lifetimes mechanisms that can account for the reduction, surface damage is confirmed to be more plausible. We demonstrate that the measured effective lifetime curves can be well reproduced assuming surface damage, a 700 nm thin layer with much lower bulk lifetimes, with numerical simulation.