Transition metal oxides as hole-selective contacts in silicon heterojunctions solar cells

TL;DR: In this paper, a comparative study comprising three transition metal oxides, MoO3, WO3 and V2O5, acting as front p-type contacts for n-type crystalline silicon heterojunction solar cells was performed.

About: This article is published in Solar Energy Materials and Solar Cells.The article was published on 2016-02-01 and is currently open access. It has received 326 citations till now. The article focuses on the topics: Crystalline silicon & Heterojunction.

Summary

1. Introduction

• The last years much effort has been devoted by the photovoltaic community to find crystalline silicon (c-Si) solar cell technologies with competitive manufacturing costs.
• Cost reduction strategies include using ultra-thin wafers or lower-quality substrates, but in any case lower thermal budgets and simplified fabrication processes would be desirable.
• Hybrid c-Si/organic structures where the p-doped layer is replaced by PEDOT:PSS have already been demonstrated [3,4], achieving an outstanding open-circuit voltage (VOC) of 657 mV and a conversion efficiency above 20% [5].
• These TMOs work as hole-selective contacts due to their large work functions (>5 eV) laying close to the Highest Occupied Molecular Orbital (HOMO) level of several ptype organic semiconductors, favoring ohmic contact formation.
• Particularly, heterojunction solar cells based on p-type c-Si (p-Si) and TMOs acting as p-type Back Surface Fields (BSFs) were reported recently [18,19], demonstrating low contact resistivities and efficiencies of 15%.

2. Experimental methods

• Surface composition of the TMOs was determined by X-ray Photoelectron Spectroscopy (XPS).
• Scans were performed using a non-monochromated Al-Kα X-ray excitation source at 1486.6 eV, detecting photoelectrons (Phoibos 150 MCD-9 detector) at a 25 eV pass energy in 0.1 eV steps.
• The structure of the TMO/n-Si heterojunction solar cells fabricated with V2O5, MoO3 and WO3 is depicted in Fig. 1(a), while Fig. 1(b) summarizes the main processing steps.
• The substrates were immediately loaded into a PECVD system (Elettrorava, Italy) to deposit a stack of layers on the rear side.
• Then, the rear side was laserfired to obtain an array of locally-diffused point contacts (0.5% contacted area fraction) [26], resulting in a highly passivated back contact with contact resistivities <1 mΩcm 2 .

3.1 Properties of Transition Metal Oxides

• The adequacy of V2O5, MoO3 and WO3 as large work function carrier-selective materials depends on their specific electronic properties.
• This is supported by an O/M ratio that results in an oxygen-rich WO3.2, contrary to the observed oxygen deficiency in V2Ox and MoOx.
• Accordingly, TMOs exhibit a double functionality depending on the relative work function difference ΦTMO–ΦSi of both materials [33].
• Fig. 3 shows the optical absorbances measured by spectrophotometry, subtracting the contribution of the glass slide.
• Overall, the identification of oxygen-vacancy defects in the TMOs under study (at least for the V2Ox and MoOx cases), as well as their inherently large work functions (>5eV) and excellent optical transparency, strongly indicate that these materials possess the functionality required for carrier-selective contacts in c-Si heterojunctions.

3.2 Transition Metal Oxide/n-Si Solar cells

• Since TMO thickness is a critical design variable, preliminary solar cell devices were fabricated on flat wafers with varying MoOx thicknesses (30, 60 and 90 nm).
• The obtained open-circuit voltages (VOC) were almost invariant at 603 ±7 mV, suggesting that the capability of MoOx to passivate silicon’s surface does not depend on its thickness.
• In contrast, the shortcircuit current densities (JSC) decreased from 30.6 mA/cm 2 (30 nm) to 25.6 (90 nm) due to parasitic absorbance and reflectance losses of the MoOx/ITO layers.
• By assuming a constant VOC of 600 mV and a conservative FF of 70%, a maximum power conversion efficiency (PCE) of 16.9% was obtained.
• A shortfall of silicon heterojunction solar cells is the passivation damage and VOC loss induced by sputtering and e-beam evaporation [37].

3.3 Origin of rectification in TMO/n-Si heterojunctions

• In order to elucidate the carrier transport mechanism in TMO/n-Si heterojunctions, their dark J–V characteristics were measured at varying temperatures and fitted for the double diode solar cell model, extracting J01 and J02 for the diffusion and recombination diodes respectively (Fig. 7(a)).
• The thermally-activated behavior of each J0 is defined by a specific activation energy (Ea) [34]: where J00 is the saturation current pre-factor, kT the thermal voltage and n the ideality factor for each fitted diode (n1 = 1 and n2 = 2).
• From the exponential fit of the Arrhenius plots in Fig. 7(b), the obtained activation energies seem to be related to crystalline silicon´s bandgap (Ea1 = Egap, Ea2 = ½Egap).
• Parting from equilibrium conditions, the Fermi level difference between both semiconductors is distributed between the induced built-in potential Vbi across the heterojunction and the dipole Δ that very likely occurs at the interface [33]: (3) This Vbi is expected to be mostly allocated on the silicon bulk, given the thinness of the TMO layer.
• Similarly, up-bending of the conduction band creates a barrier for electrons, promoting the separation of photogenerated carriers.

4. Conclusion

• Three n-type transition metal oxides (V2Ox, MoOx and WOx) with work function values >5eV were effectively used as hole-selective contacts for n-Si heterojunction solar cells, obtaining a maximum VOC of 606 mV for the V2Ox/n-Si device with a corresponding PCE of 15.7%.
• After MoOx (13.6%), the lowest performance was for WOx (12.5%), possibly due to the absence of oxygen vacancies in its atomic structure as determined from XPS analyses.
• A wide energy band gap resulted in an estimated JSC gain of ~1.2 mA/cm 2 (for 300–600 nm wavelengths) when compared to a reference a-Si:H emitter.
• Given the obtained results, prospective applications of these doping-free heterojunctions could be extended to advanced silicon technologies such as ultra-thin wafers or interdigitated back-contacts, taking advantage of the demonstrated low-temperature and solution processability of transition metal oxides.

Figures

Fig. 5(a) (90 mm width)

Fig. 6 (90 mm width)

Fig. 3 (90 mm width)

Fig. 4 (90 mm width)

Table 1 Peak positions (eV) of fitted XPS spectra and relative content of each oxidation state identified, including O/M ratios and reported work functions (ΦTMO) for as-deposited and air-exposed oxides. The oxidation state variations M n → M n–x identified for each transition metal were V +5 → V +4 , Mo +6 → Mo +5 and W +6 → W +5 . Note also the energy band gap (Egap) sensitivity to redox conditions.

Table 2 Photovoltaic parameters obtained for the fabricated solar cells with transition metal oxides (15 nm) as holeselective contacts, compared to a reference heterojunction device with a-Si:H emitter.

Fig. 5(a) (90 mm width)

Fig. 2 (140 mm width)

Frequently asked questions

Q1. What have the authors contributed in "Transition metal oxides as hole-selective contacts in silicon heterojunctions solar cells" ?

This work reports on a comparative study comprising three transition metal oxides, MoO3, WO3 and V2O5, acting as front p-type emitters for n-type crystalline silicon heterojunction solar cells. 

Q2. What are the future works in "Transition metal oxides as hole-selective contacts in silicon heterojunctions solar cells" ?

In future works, the possibility to include additional passivation interlayers to further increase VOC values should be investigated. 

Q3. What is the important question regarding the extraction mechanism of photogenerated holes?

Although diffusion of injected minority carriers appears to be the predominant transport process, questions remain regarding the specific extraction mechanism of photogenerated holes via gap states in these oxides. 

Q4. What is the effect of the absence of the W +5 oxidation state on the air?

the absence of the W +5 oxidation state on the air-exposed surface could be caused by tungsten reoxidation by air, not excluding oxygen vacancies from the material bulk. 

Q5. Why are TMOs a natural doping alternative?

Since TMOs are more stable than their organic counterparts [15] and possess the same low-temperature and solution-based processability, it is natural to explore their potential as doping alternatives for c-Si solar cells. 

Q6. What is the recent research on organic thin-film photovoltaics?

recent research on organic thin-film photovoltaics has provided a considerable number of carrier-selective materials (i.e. with preferential conductivity for either electrons or holes) which can be deposited by low-temperature or solution processes. 

Q7. How is the recovery of the VOC done?

In standard a-Si:H/c-Si solar cells, the recovery of the VOC is routinely done by post-fabrication thermal annealing (160 ºC for 20 minutes) [37], enhancing the current collection efficiency. 

Q8. What was the effect of Shirley background subtraction and fitting?

After Shirley background subtraction and fitting by Gaussian-Lorentzian curves, a multi-peak deconvolution of the spectra was performed by use of the binding energies referenced in the literature, allowing to quantify the relative content of each oxidation state and the oxygen to metal (O/M) ratios from the integrated peak areas. 

Q9. What could be done to improve the conductivities of the oxide films?

Further study of preparation methods and post-deposition treatments could promote oxide crystallinity in order to enhance film conductivities. 

Q10. How many x-rays were used to measure the deposition rate of TMOs?

The deposition rate was ~0.2 Ȧ /s, as controlled by quartz micro-balance previously calibrated with Scanning Electron Microscope (SEM) measurements of lamella samples. 

Q11. What is the effect of the n-type transition metal oxides on photogenerated holes?

given their n-type nature and their relatively low density of gap states, it can be argued that the transport of photogenerated holes across the oxide bulk does not occur, but instead they recombine in the TMO/n-Si interface with those electrons supplied by the ITO contact [42]. 

Q12. What is the way to reduce production costs?

In this sense, the utilization of risk-free materials deposited at low temperature is a comprehensive alternative to further decrease production costs. 

Q13. What was the main process of the TMO/n-Si heterojunction solar cells?

Further TMO characterization included spectrophotometry measurements (Shimadzu UV3600) on soda-lime glass slides and lateral resistivity measurements between two gold electrodes (length/width = 0.01 mm/mm) deposited upon an insulating SiO2/c-Si surface. 

Q14. Why were the hereby reported solar cells not annealed?

the herebyreported solar cells were not annealed after having observed FF absolute losses of 25–35%, attributed to TMO instabilities but needing further investigation. 

Q15. What is the reason why WOx is underperforming?

The underperformance of WOx could be partially explained by the absence of oxygen vacancies or merely by its lower passivation potential. 

Q16. How can the authors improve the work function of transition metal oxides?

a significant improvement could be achieved by finetuning the transition metal oxide work function (for instance, by avoiding air-exposure) or by eliminating the i-VOC losses of ~20 mV caused by sputtering damage. 

Q17. How did the shortcircuit current densities decrease?

In contrast, the shortcircuit current densities (JSC) decreased from 30.6 mA/cm 2 (30 nm) to 25.6 (90 nm) due to parasitic absorbance and reflectance losses of the MoOx/ITO layers.