Towards 28 %-efficient Si single-junction solar cells with better passivating POLO junctions and photonic crystals

https://doi.org/10.1016/j.solmat.2021.111560Get rights and content

Highlights

  • Reconsideration of the influence of increased photogeneration by photonic crystals on energy conversion efficiency under experimentally feasible surface passivation and wafer thicknesses.

  • Insight that the surface passivation quality of our polysiclicon on oxide (POLO) junctions needs to be improved, deduction of optimization strategy by quantification of c-Si/SiOx interface state density on sophisticated test structures and MarcoPOLO analysis.

  • Experimental reduction of J0 values of nPOLO (pPOLO) junctions from 4 (10) fA/cm2 to 0.5 (3.3) fA/cm2 by advanced hydrogenation, achievement of >70 ms effective lifetime (@ Δn = 1015 cm−3) on symmetric p-type lifetime samples with nPOLO passivation.

  • Simulation-based prediction of efficiency potential of 29.1% for POLO2 IBC cells with PCs on 150 μm thick (and thus still handleable) wafers.

  • Simulation-based prediction of efficiency potential of 27.9% for POLO2 IBC cells with random pyramid texture (i.e. without PCs) on 150 μm thick wafers .→ “practical efficiency limit” for POLO2 IBC cells above 27% even when subtracting perimeter losses

Abstract

We conduct numerical device simulations to study to what extend poly-Si on oxide (POLO)2 IBC solar cells can be optimized. In particular, we evaluate the benefit of the concept of photonic crystals (PCs) for “standard” cell thicknesses compatible with industrial wafer handling. We find that for our current surface passivation quality, implementing PCs and decreasing the wafer thickness down to 15 μm would increase the efficiency by „only“ 1% absolute due to limiting surface recombination losses. We deduce a high c-Si/SiOx interface state density Dit of 2.9 × 1012 eV−1cm−2 by analyzing special two-terminal IV measurements on small pads that contact the intact interfacial oxide between pinholes with our MarcoPOLO model. Consequently, we improve the hydrogenation process of our POLO junctions by an Al2O3/SiNx/Al2O3 rear-side dielectric layer stack. For n-type POLO (p-type POLO) J0 is reduced from 4 (10) fA/cm2 down to 0.5 ± 0.3 (3.3 ± 0.7) fA/cm2. For this improved surface passivation, our numerical device simulations predict an efficiency potential of 29.1% (27.8%) for POLO2 IBC cells with (without) PCs for a standard thickness of 150 μm. This shows that the “practical limit” for Si solar cells with poly-Si on oxide-based passivating contact schemes is above 27%, and, in general, that the efficiency potential of Si single-junction cells is still far from being exhausted. The first implementation of the improved POLO junctions into cell precursors confirms the predicted improvement on the level of suns - implied open-circuit voltage curves.

Introduction

Lighthouses guide the way in a complex field. This is also the case in photovoltaics (PV), where many different absorber materials, their combinations, and, within these technologies, many different approaches and structures compete with each other with respect to efficiency improvement as a lever for further reduction of the Levelized Costs of Electricity. More specifically, an important emerging question is to what extent the current backbone technology of PV – crystalline (c-) Si single-junction cells with a market share >95% – can be optimized further before a transition to another technology – most likely perovskite/silicon tandem cells – is required. It is clear that there is still a difference of ∼6%/∼5.5%/∼5%/∼4.5% between the current median production efficiencies of the c-Si technologies PERC (Passivated Emitter and Rear)/TOPCon1/SHJ (H-rich amorphous Si/c-Si heterojunctions)/IBC (interdigitated back-junction cells with passivating contacts) and the theoretical limit of 29.56% for Lambertian light trapping and only intrinsic recombination [1]. However, this gap can hardly be closed completely due to extrinsic loss mechanisms that occur in every real solar cell. To what extend these extrinsic losses can be mitigated is difficult to quantify by simulation studies due to uncertainties in the input parameters.

Therefore, the guiding lighthouses in PV are often record-efficiencies on laboratory cells, demonstrating the potential of a technology and thus directing subsequent application-related developments to the most promising approaches.

The one sun record efficiencies for solar cells based on a single Si absorber have remained unchanged2 in the last ∼3 years at 26.7% [2,3] for c-Si cells with passivating contacts based on SHJ and at 26.1% for passivating contacts based on polycrystalline Si on oxide (POLO) junctions [4]. These values are very close to previously announced “practical limits” of 27.2% [2] and (26.0% [5]) 26.2% [6] for the respective contact schemes. By contrast, record efficiencies for small area perovskite/Si tandem cells have increased rapidly in the last years to 29.15% [7] or even 29.8% [8]. There is also impressive progress in upscaling [9,10] and stability [10] of perovskite (sub) cells. We nevertheless think that Si single-junction solar cells will stay competitive for a while, particularly if they succeed in further significant efficiency improvement.

In this work, we intend to depict pathways that allow an improvement of c-Si single-junction record efficiencies in order to revise the impression that the potential of Si is already exhausted. We argue that the practical limit, particularly for Si cells with POLO junctions and perhaps even the theoretical “intrinsic limit,” can be shifted upwards. The inspiration for the latter stems to a large extent from the theoretical work of Bhattacharya and John [11] who showed that one way to exceed the Lambertian light trapping and thus the corresponding the Auger & radiative efficiency [1,12] is to implement photonic crystals (PCs) [13,14] on the front side of the solar cell. The PCs, which are in this case just inverted pyramids with dimensions3 of ∼1–3 μm but arranged in a periodic pattern, improve the solar absorption due to the formation of optical modes propagating parallel to the surface with reduced group velocity [11]. Thus, the absorber thickness can be reduced, and intrinsic recombination can be mitigated. Based on this concept, Bhattacharya and John simulated a 15 μm-thin 31.1% efficient Si single-junction cell, including all losses [11].

We conduct a numerical device simulation study to evaluate to what extent this approach can be transferred to our current and future POLO2-IBC cells4 (Fig. 1) and to “standard” cell thicknesses that are compatible with industrial wafer handling. The experimental findings of our 26.1%-efficient record cell serve as a starting point [19]. We find that the current surface passivation quality, including our poly-Si on oxide (POLO) junctions, is not sufficient (i.e., not as good as assumed in Ref. [11]) and would prevent a large efficiency improvement by the PCs. Thus, we focus in the experimental part of this work on a significant improvement of our POLO junctions that is compatible with our cell process. We perform special two-terminal IV measurements on pads that are small enough to contact only the intact oxide area between pinholes [20]. Since the in this case dominating tunneling currents are very sensitive to the band alignment, it is possible to deduce the c-Si/SiOx interface state density Dit from a fit with our MarcoPOLO model [21]. A high Dit value of 2.9 × 1012 eV−1cm−2 is obtained. Consequently, we improve the hydrogenation process of our junctions. Using the reduced prefactors of the recombination current densities as input parameters, a simulation-based roadmap to efficiencies beyond 28% with PCs and beyond 27% without PCs is provided – for POLO2-IBC cells on a standard wafer thickness of 150 μm and even when subtracting perimeter losses of ∼0.3%abs.

In our first attempt to implement the improved POLO junctions into solar cells (still w./o. PCs 5), we measured an improvement of the implied pseudo efficiency of 0.4%abs, which is consistent with the predictions from the simulations.

Section snippets

Device simulations

For the numerical device simulations, we use the Quokka2 implementation [22] of the Conductive Boundary Model [23] and the Free Energy Loss Analysis (FELA) concept [24]. We use our 26.1% efficient POLO2 IBC cell (on 290 μm thick wafers and with random pyramid front side texture) as a starting point, which has already been simulated by Hollemann et al. [19] based on experimentally determined input parameters. These parameters are listed in Table A1. One should remark that the surface

Deduction of optimization strategy for POLO junctions

We characterize the properties of the SiOx/c-Si interface from IV curves at room temperature of a sample that contacts a pinhole-free area fraction of the pPOLO junction. For that, we characterize the IV curves measured on the two-terminal test structures with a contacted junction area of (5 × 5) μm2. For this area fraction of the pPOLO junction, we expect a small number of 2–3 pinholes per pad from the deduced pinhole density with the etch-pit method [43]. One interesting finding is that none

Improved hydrogenation

Fig. 4 (a) shows the improvement of the passivation quality of our POLO junctions by our new rear side dielectric layer stack intended as a more efficient hydrogen donor. The effective lifetimes measured on the symmetric lifetime test structures with nPOLO, pPOLO and intrinsic poly-Si at an injection level of Δn = 1015 cm−3 are shown at different processing stages. The post-deposition anneals are performed cumulatively. The respective benchmark values for our previous passivation by

A simulation-based roadmap to efficiencies beyond 28%

Fig. 5 shows the main IV parameters efficiency η, open circuit voltage Voc, short circuit current density Jsc and fill factor FF as simulated for our current experimental POLO2-IBC cell as a starting point, as well as for different hypothetical solar cells including different improvements.

The potential path to directly implement photonic crystals with the previous surface passivation quality is indicated in blue squares. This approach – together with a reduction of the wafer thickness down to

Discussion

It is clear that these predictions from the simulations call for an experimental verification first. Nevertheless, it is worth discussing the perspectives: A 15 μm thin Si single-junction solar cell with an efficiency close to 30% would be very attractive. In particular, its mechanical flexibility could be utilized in special applications such as Vehicle-Integrated PV (VIPV). Energy yield modeling and measurements would be required to assess aspects like dependency on the angular distribution

Conclusion

In this work, we evaluated how far we can push the efficiencies of our POLO2-IBC Si single-junction solar cells under normal incident light. As an extension of previous completely idealized considerations [19], we attempt here to base our projection as much as possible on experimental input data, in particular for improved POLO junctions. Furthermore, we take into account the concept of photonic crystals [11] with excellent light trapping even for very thin wafers.

Our numerical device

Declaration of competing interest

Herby, I declare that I do not have any competing financial or non-financial interests.

Acknowledgments

We would like to thank H. Fischer, S. Spätlich, R. Winter, A. Raugewitz,G. Glowatzki and R. Zieseniβ for sample processing, M. Wolf, A. Dietrich, R. Reineke-Koch for discussion and support with the measurement systems, and Sajeev John for fruitful discussions. This work is funded by the German Ministry for Economic Affairs and Energy (grant FKZ 003EE1056A), by the federal state of Lower Saxony and by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany's

References (64)

  • D.B.M. Klaassen

    A unified mobility model for device simulation—I. Model equations and concentration dependence

    Solid State Electron.

    (1992)
  • S. Schäfer et al.

    Accurate calculation of the absorptance enhances efficiency limit of crystalline silicon solar cells with lambertian light trapping

    IEEE Journal of Photovoltaics

    (2018)
  • K. Yoshikawa et al.

    Silicon heterojunction solar cell with interdigitated back contacts for a photoconversion efficiency over 26

    Nat. Energy

    (2017)
  • M. Green et al.

    Solar cell efficiency tables (version 57)

    Prog. Photovoltaics Res. Appl.

    (2021)
  • R.M. Swanson

    Approaching the 29% limit efficiency of silicon solar cells

  • D.D. Smith et al.

    Toward the practical limits of silicon solar cells

  • A. Al-Ashouri et al.

    Monolithic perovskite/silicon tandem solar cell with >29% efficiency by enhanced hole extraction

    Science

    (2020)
  • A.S. Subbiah et al.

    High-performance perovskite single-junction and textured perovskite/silicon tandem solar cells via slot-die-coating

    ACS Energy Lett.

    (2020)
  • Z. Liu et al.

    A holistic approach to interface stabilization for efficient perovskite solar modules with over 2,000-hour operational stability

    Nat. Energy

    (2020)
  • S. Bhattacharya et al.

    Beyond 30% conversion efficiency in silicon so-lar cells: a numerical demonstration

    Sci. Rep.

    (2019)
  • S. John

    Strong localization of photons in certain disordered di-electric superlattices

    Phys. Rev. Lett.

    (1987)
  • E. Yablonovitch

    Inhibited spontaneous emission in solid-state physics and electronics

    Phys. Rev. Lett.

    (1987)
  • A. Richter et al.

    Design rules for high-efficiency both-sides-contacted silicon solar cells with balanced charge carrier transport and recombination losses

    Nat. Energy

    (2021)
  • J. Zhao et al.

    24·5% Efficiency silicon PERT cells on MCZ substrates and 24·7% efficiency PERL cells on FZ substrates

    Prog. Photovoltaics Res. Appl.

    (1999)
  • C. Hollemann et al.

    26.1%-efficient POLO-IBC cells: quantification of electrical and optical loss mechanisms

    Prog. Photovoltaics Res. Appl.

    (2019)
  • P. Bayerl et al.

    Contacting a single nanometer-sized pinhole in the interfacial oxide of a poly-silicon on oxide (POLO) solar cell junction

    Prog. Photovoltaics Res. Appl.

    (2021)
  • N. Folchert et al.

    Modeling recombination and contact resistance of poly-Si junctions

    Prog. Photovoltaics Res. Appl.

    (2020)
  • A. Fell

    A free and fast three-dimensional/two-dimensional solar cell simulator featuring conductive boundary and Quasi-Neutrality approximations

  • R. Brendel

    Modeling solar cells with the dopant-diffused layers treated as conductive boundaries

    Prog. Photovoltaics Res. Appl.

    (2012)
  • R. Brendel et al.

    Theory of analyzing free energy losses in solar cells

    Appl. Phys. Lett.

    (2008)
  • F. Haase et al.

    Perimeter recombination in 25%-efficient IBC solar cells with passivating POLO contacts for both polarities

    IEEE Journal of Photovoltaics

    (2018)
  • Cited by (0)

    View full text