Two-Dimensional Absorbers for Solar Windows: A Simulation

1 Introduction

Solar windows and other building integrated photovoltaic technologies can play an important role in addressing the issue of future energy production in urban areas. Since the early 2000s, the demand for low- or net zero-energy building is on the rise [1]. Solar windows are of interest because many modern buildings have glass facades or large windows that provide a new option for decentralized energy production. The standard silicon-based photovoltaic materials used today are bulky and not transparent or colour neutral, making them poor candidates for solar windows. A considerable amount of literature has been published on solar windows or transparent solar cells using a range of different technologies including, but not limited to, organic polymer-based solar cells [2], [3], perovskites [4], [5], dye-sensitized solar cells [6], [7], luminescent solar concentrators [8], [9], and thin-film inorganic devices [10], [11]. These studies show that there is a considerable interest in solar technology for transparent applications.

Transition metal dichalcogenides (TMDCs), known for their unique semiconducting properties at the monolayer and few-layer regime [12], are good candidates for solar windows. When coupled with transparent contacting materials, these ultrathin absorbers have the potential to create highly transparent and colour-neutral solar cells with good current generation. Previous research has established that TMDCs, such as MoS₂, MoSe₂, WS₂, and WSe₂, have high absorption coefficients [13], strong photoluminescence [14], [15], high charge carrier mobilities [16], and band gaps ranging from 1.1 to 1.9 eV [17], in addition to naturally passivated surfaces that are nearly free of dangling bonds [18]. These attributes make them suitable candidates for photovoltaics. The main challenge faced by many experiments is the ability to create high-quality layers on large scales. Therefore, only a few small-scale TMDC solar cells have been produced so far [19], [20] However, there are large efforts being made to create new methods of TMDC production and to increase material quality at large scales [21], [22].

Previous work by our group has shown that it is possible to sandwich an ultrathin amorphous germanium absorber inside a Fabry–Perot resonant cavity [23], [24], [25]. The idea of the cavity, as depicted in Figure 1a, is to have light pass through the absorber multiple times by reflecting light between two interfaces, thus significantly increasing absorption. More recent works show that via suitable design of the back reflector these solar cells can be made transparent (Fig. 1b) [26] or spectrally selective [27] or could even be switched between opaque and transparent state using a switchable back reflector (Fig. 1c) [26], [28], making the technology a good candidate for solar windows. We propose that two-dimensional semiconductors, such as TMDCs, can be implemented in similar solar cell configurations.

Figure 1:

Schematic of three different solar cell configurations. (a) Resonant cavity–enhanced solar cell, (b) transparent solar cell, and (c) switchable solar cell.

Transparent contacting and electrode materials are of particular interest in this study. They are the backbone of creating transparent solar cells. Typically, TMDC-based solar cells are created by forming a Schottky junction [13], [29], [30], [31], [32] with a metal contact, where the barrier height depends on the work function of the metal. Akama et al. [32] proposed a semitransparent solar cell design based on a Schottky junction with WS₂ or WSe₂ as an absorber and suggest that this design, using prepatterned electrodes, could be scalable in the future [32]. It has also been reported that an ITO–MoS₂–Au configuration can produce a photovoltaic effect [29], [33]. A Schottky-junction solar cell, beyond the flake size, with a current density of 5.37 mA/cm² was created by Shanmugam et al. [29] with 220 nm of transferred chemical vapor deposition grown MoS₂. Grounded on the previous findings, we propose that an Au–TMDC–ITO solar cell stack with thin-enough gold and TMDC layers would be a good candidate for laboratory scale studies of TMDCs for solar windows.

Another contacting method of interest is using two transparent conducting oxides (TCOs). Using two TCOs would eliminate the need for a typical p–n junction. Lithium fluoride (LiF) [34] and titanium dioxide (TiO₂) [35], [36] have been reported as electron selective contacts, whereas molybdenum oxide (MoO₃) [37] is reported as a hole-selective contact. These materials have been proven to work together as carrier-selective contacts in silicon-based solar cells [38], [39]. It is understood that the extreme work function of both LiF and TiO₂ (low work function) and MoO₃ (high work function) contributes to the carrier selectivity of the materials. Furthermore, these carrier-selective materials are transparent in the visible light range, making them suitable contact material options for transparent solar cells.

This work presents a set of simulations investigating the transmission, quantum efficiency, current density, and colour appearance of solar cell configurations, with different contact materials and TMDCs, while also looking at the effect of monolayer versus few-layer TMDCs.

2 Simulation

Different solar cell stacks, similar to the one presented in Figure 1b, are considered here. First, two contacting methods from literature are compared: metal/TCO and TCO/TCO. Then, the TCO/TMDC/TCO stack is further explored and made colour neutral by changing the thickness of the contacts. Finally, the TMDC materials, MoS₂, MoSe₂, WS₂, and WSe₂, are compared in the same solar cell stack in both monolayer and few-layer forms, and quantum efficiency and current density were calculated.

2.1 Transmission and Colour Appearance

We started with MoS₂ as the TMDC material because it has no very strong absorption peaks, making it a good candidate for colour neutrality. The transmission spectra of two solar cell stacks Au–MoS₂–ITO and LiF–TiO₂–MoS₂–MoO₃ are shown in Figure 2a. In addition, the simulated transmission of monolayer MoS₂ is also shown. The first stack consists of a 10-nm gold contact and a 100-nm ITO contact. The second stack has 2-nm LiF, 50-nm TiO₂, and 50-nm MoO₃. Both contacting schemes have broadband transmission across the visible light range. The Au–MoS₂–ITO stack is the least transparent, with transparency ranging from 35 % to 50 % with a broad peak centred around 500 nm. This peak makes the colour appearance slightly green (Fig. 1b). When we switch to a configuration with two TCOs, the transparency increases. The LiF–TiO₂–MoS₂–MoO₃ stack has a transparency of up to 80 %, but because the transparency has a well-defined peak, it is not colour-neutral and has a turquoise colour appearance.

Figure 2:

Transmission of monolayer MoS₂, Au–MoS₂–ITO solar cell stack and LiF–TiO₂–MoS₂–MoO₃ solar cell stack. (a) Transmission spectra. (b) Colour appearance of light transmitted through the solar cell stack.

The contact configuration with the highest transparency, LiF–TiO₂–MoS₂–MoO₃, was chosen for the colour rendering. To address the colour rendering, the thickness of the TiO₂ and MoO₃ was varied. By taking advantage of the different optical densities of the materials, the transmission spectra can be tuned by thickness variations. Figure 3 shows the evolution of the transmission spectra and colour appearance. When the contact layers (TiO₂ and MoO₃) are very thin, in this case 10 nm each, the colour appearance is yellow. When the contact thicknesses are around 25 or 30 nm on each side of the stack, the colour rendering becomes more neutral. This can be observed in the transmission spectra because the curves are nearly horizontal lines in the visible light range. As the contact thicknesses increase, resonant peaks start appearing in the transmission spectra, making the transmission reach 90 %, but making the colour appearance less colour neutral.

Figure 3:

Variation of TiO₂ and MoO₃ layer thickness in the LiF–TiO₂–MoS₂–MoO₃ solar cell stack. (a) Transmission spectra. (b) Colour appearance of light transmitted through the solar cell stack.

To evaluate the transmission and colour appearance of different TMDC materials, the LiF–TiO₂–TMDC–MoO₃ contacting configuration with 30 nm of TiO₂ and MoO₃ was chosen for comparison. This configuration was chosen because the contact thicknesses are the most colour neutral. Figure 4 shows the transmission spectrum of this stack with four different TMDCs (MoS₂, MoSe₂, WS₂, WSe₂), in few-layer (5 nm) and monolayer forms. The few-layer stacks show transmission between 30 % and 40 %, and as expected, the stacks with the monolayer material have higher transmission of 45 % to 55 %. The different TMDC materials have minimal effect on the colour appearance, as can clearly be seen from the colour appearance boxes (Fig. 4b).

Figure 4:

Variation of TMDC material in LiF–TiO₂–TMDC–MoO₃ solar cell stack with 30-nm TiO₂ and MoO₃ contact layers, with different-thickness TMDCs. (a) Transmission spectra for monolayer (solid line) and few-layer (dotted line) material. (b) Colour appearance of light transmitted through the solar cell stack.

2.2 Quantum Efficiency and Current Density

While colour appearance and transmission spectra are important aspects for solar windows, the electrical aspects are also significant. The calculated quantum efficiency and current density of the same LiF–TiO₂–TMDC–MoO₃ stack, with 30-nm contacts, are shown in Figure 5 and Table 1, respectively. For the configurations with monolayer TMDCs, the quantum efficiency is below 10 %, and the current density is between 0.63 and 1.39 mA/cm². For the few-layer configurations, the maximum quantum efficiency ranges from 20 % to 40 %, and the corresponding current densities fluctuates are between 2.99 and 8.33 mA/cm², depending on the TMDC material. Note that only photons with wavelengths between 350 and 800 nm were counted towards the current density because of missing dielectric function data, so that the actual current densities, especially for WSe₂ and MoSe₂ in the few-layer configuration, could be higher.

Figure 5:

Quantum efficiency of LiF–TiO₂–TMDC–MoO₃ solar cell stack with 30-nm TiO₂ and MoO₃ contact layers with different TMDC materials in monolayer (solid line) and few-layer (dotted line) forms.

Table 1:

Calculated current density of LiF–TiO₂–TMDC–MoO₃ stack with different TMDCs.

	MoS2	MoSe2	WS2	WSe2
	(mA/cm2)
Monolayer	1.11	1.39	0.63	0.75
Few-layer	5.63	8.33	2.99	4.98

3 Discussion

From the results presented above, TMDC materials make promising candidates for transparent solar cells. Their high absorption coefficients, coupled with the right contacting material, provide a good base for future transparent photovoltaic applications. The simulations show the upper bound of what is possible assuming ideal transport and charge extraction efficiency. It is possible that further layer optimization or additional charge carrier collecting infrastructure may be necessary to make working solar cells. The results demonstrate that MoSe₂ has the potential to provide the most current density with calculated values up to 8.33 mA/cm² using 5 nm of absorber material. MoS₂ and WSe₂ also have impressive current densities of 5.63 and 4.98 mA/cm², respectively, when in the few-layer form. The solar cell stacks with few-layer MoSe₂ have significantly higher quantum efficiency in the 700- to 800-nm range, compared to MoS₂ and WS₂, which leads to higher current density.

Figures 2 and 3 show that a simple solar cell stack with monolayer MoS₂ can be transparent and colour neutral and that the choice of contacting materials and thickness can influence both parameters. Due to thin film interference, the optical thicknesses of the contact layers influence the colour appearance. This effect can be used to tune the layer stack to a more neutral colour rendering, as is shown in Figure 3 using different thicknesses of TiO₂ and MoO₃.

In all cases, the solar cell stacks with the few-layer TMDCs have larger quantum efficiency and current densities compared to the solar cell stacks with monolayer material. This is despite the fact that beyond monolayer thickness TMDCs no longer have a direct band gap, which makes light absorption less efficient.

One of the major tradeoffs is transparency versus current generation. The more transparent the solar cell, the less current it generates. This is the result of most of the visible light being transmitted through the solar cell stack and not being absorbed for current generation. Therefore, it is very difficult to produce a highly transparent solar cell that has good current generation. Considering solar protection glass that is available on the market, a very high transparency might not be necessary, as long as the colour appearance is neutral. A lot of solar protection glass solutions show transmission in the visible light range of not more than 50 % to 70 %. Another even more exciting alternative to this problem is the switchable solar cell proposed recently by our group [26], which is sketched in Figure 1c. A first realization of this concept has shown that in such devices a large current generation increase is possible in the opaque state due to active light trapping [28].

4 Methods

Fully coherent 1D optical simulations of different solar cell stacks were performed using the commercially available software package SCOUT/CODE (W. Theiss Hard- and Software, Aachen, Germany). The simulations present the optical feasibility of the proposed solar cells stacks; no electrical parameters of the materials such as charge carrier mobility or exciton/charge carrier separation and transport are considered. To complete the simulations, the complex refractive index data and dielectric functions of the materials were taken from several sources: the TMDC materials, MoS₂, MoSe₂, WS₂, and WSe₂, were simulated using the optical dielectric function data, ε₁ and ε₂, published by Li et al. [40] using different data for monolayer and few-layer materials. The optical data for ITO, TiO₂, Ag, and LiF were provided by the database in the CODE modelling software. ITO, Lif, and TiO₂ are based off models from Jellison et al. [41] and Jellison [42], and the Ag is based off of models from Palik [43]. The refractive index data for the AZO and MoO₃ were modelled with CODE using the transmission and reflection spectra of the material, which were deposited and measured in-house. It is important to note that the optical data for the TMDC materials are extrapolated from 350 to 413 nm because the optical data from Wheeler et al. [5] were calculated only from 1.5 to 3 eV (or about 414–826 nm). Using the software, the transmission, quantum efficiency, current density, and colour appearance were simulated. The simulations assume a direct light source with mixed polarization perpendicular to the solar cell stack. The quantum efficiency was calculated by taking the absorption inside the TMDC layer and assuming that all charge carriers are extracted. The current density was calculated by using the quantum efficiency and the AM1.5 g spectrum. The colour appearance is rendered using the transmission spectra of the solar cell stacks and simulates the light shining through the solar cell at an observation angle of 2 degrees.

5 Conclusion

This article shows that TMDCs have the potential to be integrated into colour-neutral solar windows, by simulating the transmission, colour appearance, quantum efficiency, and current density of different solar cell configurations. When contacting layers are optimized for colour appearance, colour-neutral solar cells with monolayer and few-layer TMDCs materials are achievable with transparencies between 35 % and 55 %. The simulations presented here show that it is possible to get a quantum efficiency of 35 % and a current density of 8.33 mA/cm² with 5 nm of MoSe₂ because of its light absorption in the near infrared. While there are still challenges to overcome in terms of producing TMDC-based solar cells, our simulations act as a guideline for further research.