Atomic layer deposition of metal oxide thin films for Si heterojunction solar cells
- Author
- B. Macco
- Year
- 2016
- Abstract & Cover
- Source of Information
- https://research.tue.nl/en/publications/atomic-layer-deposition-of-metal-oxide-thin-films-for-si-heteroju
- University
- Eindhoven University of Technology (Eindhoven, Netherlands)
- External Link
- Read Thesis
Atomic Layer Deposition of Metal Oxide Thin Films for Si Heterojunction Solar Cells Within the field of crystalline silicon photovoltaics, the deposition technique of atomic layer deposition (ALD) is mostly known for the preparation of high-quality Al2O3-based surface passivation layers for homojunction solar cells. High-throughput ALD tools specially designed for this application are commercially available, enabling substantial gains in solar cell efficiency. Currently silicon heterojunction (SHJ) or passivating contact solar cells are receiving a lot of attention, mainly because such cells can yield even higher efficiencies and can potentially be made by straight-forward, lowtemperature processing. Whereas for the more conventional homojunction concepts thin films are only used for surface passivation and anti-reflection, heterojunction cells are truly thin-film powered devices: Thin films deposited on the silicon wafer provide all of the functionalities, such as surface passivation, selective extraction of carriers and lateral electrical conduction in the form of transparent conductive oxides (TCOs). In this dissertation, the use of ALD to prepare these functional thin films for these upcoming classes of solar cells has been explored. As described in chapter 3, ALD has been employed for the preparation of the TCO films for SHJ cells, which is conventionally done by sputter deposition. Using atomic layer deposited Al-doped zinc oxide (ZnO:Al), two key merits of ALD over conventional sputtering have been demonstrated. Firstly, whereas it is well-known that the harsh plasma conditions during sputter deposition lead to substrate damage, the absence of a plasma during thermal ALD ensures a high compatibility with the underlying sensitive thin films. Secondly, the use of so-called ALD supercycles has been shown to give an unprecedented level of control over the doping level and doping profile of the TCO. The ability to turn this ”knob” is of great interest: The doping level determines the fundamental trade-off between transparency and conductivity of the TCO. In addition, it has been experimentally shown that the interfacial doping level, which determines the work function of the TCO, is of strong influence on the junction properties. During the course of the project, it became clear that high-mobility hydrogen-doped indium oxide (In2O3:H), a TCO originally developed by sputtering, has enormous potential for heterojunction cells. In such high-mobility TCOs, electrons are able to move much more freely than in conventional Sn-doped In2O3 and doped ZnO. Therefore, the doping level required to achieve the desired level of conductivity is strongly reduced, 207 which mitigates optical losses in the infrared due to free carrier effects as found in the conventional TCOs. As such, high-mobility TCOs have proven to be crucial to obtain high short circuit density Jsc values exceeding 40 mA/cm2 for this class of solar cells. Within this thesis work, an ALD approach to prepare high-mobility hydrogen-doped indium oxide (In2O3:H) has been developed, as described in chapter 4. The process consists of the preparation of mostly amorphous In2O3:H, followed by crystallization at 150-200 oC, which results in polycrystalline In2O3:H with grains a few hundred nm in lateral size. Besides the low-temperature processing and the aforementioned merits of ALD over sputtering, the salient features of the developed ALD process are the record-high carrier mobility and low resistivity of 138 cm2/Vs and 0.27 mΩcm, respectively. In order to gain fundamental understanding for the reasons why this hydrogen-doped TCO works so well, the optoelectronic and structural properties of ALD In2O3:H and the change thereof upon crystallization have been scrutinized. As described in chapter 6, it has been found crucial to prepare the In2O3:H film at 100 oC, since at this temperature a very low density of crystallites is embedded in the otherwise amorphous film. Upon post-annealing grain growth proceeds isotropically from these crystallites without nucleation of additional crystallites. The absence of additional nucleation leads to a large final grain size and is attributed to the high hydrogen content in the films deposited at low temperatures. In addition, the doping mechanism and electron scattering in ALD In2O3:H have been studied, as reported on in chapter 5. Atomic hydrogen has been found to be an excellent dopant in this post-crystallized material: Although only 4 at.% of the embedded H acts as an active dopant, the inactive H has been found to not contribute to electron scattering, which is a strong advantage over the traditional Sn dopant. Ionized impurity from the H dopants and phonon scattering were found to be the only scattering mechanisms of relevance in the crystallized film, with a contribution of 66 % and 34 %, respectively. Since these scattering processes are unavoidable, the carrier mobility in the ALD In2O3:H is at its fundamental limit for its carrier density. The developed ALD In2O3:H has been implemented as the front TCO in a silicon heterojunction solar cell. A convincingly high short-circuit current density Jsc of 40.2 mA/cm2 was obtained, exceeding that of the ITO reference by 1.7 mA/cm2. In addition, evaporated molydenum oxide (MoOx) emerged during the project as a promising, highly transparent hole-selective contact. Initial reports showed that the performance of MoOx was highly thickness-dependent, which made ALD a very interesting candidate to prepare these MoOx films. Therefore, an ALD process for MoOx has been developed within this project, described in chapter 7. The ALD process allows for the preparation of MoOx over a wide temperature range of 50 - 350 oC. At deposition temperatures below 250oC amorphous material is obtained, whereas at higher temperatures a transition to polycrystalline growth has been observed which is accompanied 208 by a strong increase in the growth per cycle. Apart from a reasonable amount of embedded atomic H of 3-11 at. %, the films are of high purity and are slightly substoichiometric with an O/Mo ratio of 2.9. In chapter 8, the developed ALD MoOx has been implemented in a solar cell precursor together with a-Si:H and ALD In2O3:H. A very high level of passivation and optical incoupling have been obtained, as witnessed by excellent implied open-circuit voltage and simulated short-circuit current values of 717 mV and 41.7 mA/cm2, respectively. However, the carrier-selectivity of the ALD material was found to be not yet on par with its evaporated counterpart. This has been found to be due to a lower induced band bending by the ALD MoOx in comparison to the evaporated material, pointing to a lower work function for the ALD material. This lower work function is thought to originate from oxygen vacancies in the substoichiometric ALD MoOx. Therefore, future research focuses on the reduction of the oxygen vacancies in this ALD material. The experience gained within this project on the use of ALD for the preparation of passivation, carrier-selective and transparent conductive oxide layers for silicon solar cells has been culminated in a book chapter, which forms chapter 2 of this thesis. Finally, much of the work presented within this dissertation will be used and developed further within the recently started COMPASS project on passivating contacts for silicon solar cells.
