Plasma processing of microcrystalline silicon films : filling in the gaps

A.C. Bronneberg

Research output: ThesisPhd Thesis 2 (Research NOT TU/e / Graduation TU/e)

348 Downloads (Pure)

Abstract

Hydrogenated microcrystalline silicon (µc-Si:H) is a mixed-phase material consisting of crystalline silicon grains, hydrogenated amorphous silicon (a-Si:H) tissue, and voids. Microcrystalline silicon is extensively used as absorber layer in thin-film tandem solar cells, combining the advantages of a low (indirect) band gap (1.1 eV), which results in an enhanced absorption of red and (near) infrared light, with an improved stability under light exposure (reduced Staebler–Wronski effect). However, due to the indirect nature of the band gap, relatively thick (1–2 µm) µc-Si:H films are necessary to achieve an efficient absorption of red and (near) infrared light, even when light trapping concepts are applied. Therefore, from a cost-perspective point of view, high growth rates (>1 nm/s) are required, preferably in combination with large-area (roll-to-roll) processing. The most common, and so far most successful, deposition technique is the capacitively-coupled plasma (CCP) in parallel plate configuration using radio or very high excitation frequencies (RF or VHF, respectively) and highly hydrogen diluted hydrogen and silane gas mixtures. Approaches to increase the growth rate include an increase of the plasma power, moving from a low-pressure to a high-pressure depletion (LPD to HPD, respectively) regime, and/or by increasing the excitation frequency from 13.56 MHz to 27–300 MHz. The combination of HPD-VHF has resulted in high deposition rates (2-3 nm/s) while maintaining high solar cell efficiencies (7-8%). In this work, the use of an ultra-fast (2-20 nm/s) deposition technique, i.e. the expanding thermal plasma, has been explored for the deposition of µc-Si:H films. Characteristic for ETP-grown µc-Si:H films is the lack of a sufficient amount of a-Si:H tissue, which is necessary to passivate the grain boundaries and fill the intergranular space, resulting in a network of (inter-connected) cracks and voids. As a consequence, the µc-Si:H films are prone to post-deposition oxidation, resulting in low solar cells efficiencies (<2%). The post-deposition oxidation has been monitored by means of Fourier transform infrared (FTIR) spectroscopy over a period of 8 months. This study revealed a two-timescale oxidation: on short timescales (<3 months) the crystalline silicon grain boundaries oxidize, on longer timescales the oxidation involves also the a-Si:H tissue. This indicates that in order to prevent post-deposition oxidation, it is not sufficient to fill the intergranular space, but that the a-Si:H tissue needs to be of sufficient quality, i.e. dense and not susceptible for post-deposition oxidation. One process that could be responsible for the insufficient amount of a-Si:H tissue, is hydrogen-induced etching of a-Si:H tissue. Atomic hydrogen is, under µc-Si:H growth conditions, abundant in the plasma, and is known to preferentially etch a-Si:H over crystalline silicon (c-Si). In addition, the interaction of atomic hydrogen with the (growing) film can result in the formation of an hydrogen-rich sub-surface layer, caused by the insertion of atomic hydrogen into strained Si-Si bonds, which possibly explains the porous quality of the a-Si:H tissue. Monitoring the etch rate of a-Si:H films during Ar/H2 plasma exposure by real time spectroscopic ellipsometry showed that the hydrogen-induced etch rate was at least one order of magnitude lower than typical deposition rates. In addition, FTIR spectroscopy revealed that insertion of atomic H in the sub-surface layer (top ~30 nm) during Ar/H2 plasma exposure did not result in an increased porosity. These results suggest that the interaction of atomic hydrogen with the growing film is not responsible for the insufficient amount of (dense) a-Si:H tissue. The fact that the interaction of atomic hydrogen is not responsible for the poor material properties of ETP-grown µc-Si:H, the question "what mechanism is then responsible?" arises. To address this question the plasma chemistry and the resulting growth mechanism of ETP is compared to CCP, which so far is the only technique with which solar-grade µc-Si:H is obtained. One difference between the two techniques is the absence of an ion bombardment effect in ETP. In CCP the HPD and the use of VHF are employed to suppress a (potentially uncontrolled) ion bombardment effect, hypothesized to be responsible for an amorphization of the crystalline growth and defect incorporation. However, there is always some form of ion bombardment present. The extent to which ions contribute to the growth depends on the ion flux, the ion energy, and the chemical nature of the ion. Under HPD-VHF conditions SinHm+ is identified as the dominant ion in H2/SiH4 plasmas, but no direct ion energy and ion flux measurements under HPD conditions have been reported so far. Therefore, the ion energy and flux in a CCP reactor have been studied. For this purpose, a capacitively-coupled plasma reactor in parallel plate configuration has been designed and built, in close collaboration with the Institute of Photovoltaics at Forschungszentrum Jülich (Germany). This reactor has been especially designed for the implementation of plasma and (in situ) film diagnostics. Under solar-grade µc-Si:H deposition conditions the contribution of ions to the film growth has been studied by means of a capacitive probe. The ion to Si deposition flux ratio was found to be large, ~0.30. However, since the ion energy is rather low,
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • Applied Physics
Supervisors/Advisors
  • van de Sanden, M.C.M. (Richard), Promotor
  • Creatore, M. (Adriana), Copromotor
Award date3 Jul 2012
Place of PublicationEindhoven
Publisher
Print ISBNs978-90-386-3174-5
DOIs
Publication statusPublished - 2012

Fingerprint Dive into the research topics of 'Plasma processing of microcrystalline silicon films : filling in the gaps'. Together they form a unique fingerprint.

Cite this