Abstract

We report on the prototyping and development of a highly reflective dielectric back reflector for application in thin-film solar cells. The back reflector is fabricated by Snow Globe Coating (SGC), an innovative, simple, and cheap process to deposit a uniform layer of TiO2 particles which shows remarkably high reflectance over a broad spectrum (average reflectance of 99% from 500 nm to 1100 nm). We apply the highly reflective back reflector to tandem thin-film silicon solar cells and compare its performance with conventional ZnO:Al/Ag reflector. By using SGC back reflector, an enhancement of 0.5 mA/cm2 in external quantum efficiency of the bottom solar cell and an absolute value of 0.2% enhancement in overall power conversion efficiency are achieved. We also show that the increase in power conversion efficiency is due to the reduction of parasitic absorption at the back contact; that is, the use of the dielectric reflector avoids plasmonic losses at the reference ZnO:Al/Ag back reflector. The Snow Globe Coating process is compatible with other types of solar cells such as crystalline silicon, III–V, and organic photovoltaics. Due to its cost effectiveness, stability, and excellent reflectivity above a wavelength of 400 nm, it has high potential to be applied in industry.

1. Introduction

Thin-film solar cells are very promising candidates for cost effective and high efficient photovoltaics. Several of the existing and developing thin-film solar cell technologies, for example, made of organic materials or thin silicon layers, apply physically and optically thin absorber layers and thus require light trapping concepts [16] as well as highly reflective back reflectors [710] to achieve high efficiencies. In this work, we report on the fabrication, prototyping, and implementation of a highly reflective dielectric back reflector prepared by the Snow Globe Coating (SGC) technique which is a simple and cheap method to form a uniform layer of TiO2 particles.

In order to demonstrate the superiority of SGC back reflector, we test it on tandem thin-film silicon solar cells and compare its performance with conventional ZnO:Al/Ag back reflectors which have proven their excellent performance in thin-film silicon solar cells [1113]. For tandem thin-film silicon solar cells which consist of a hydrogenated amorphous silicon (a-Si:H) top solar cell and a hydrogenated microcrystalline silicon (μc-Si:H) bottom solar cell, light management at the rear side is crucial to achieve high short circuit current density, as the μc-Si:H bottom solar cell absorbs weakly in the wavelength range from around 550 nm to 1100 nm [14, 15]. Improved reflectivity at the back reflector in tandem thin-film silicon solar cells will allow an increase of the charge carrier generation in the optically thin absorber layers and/or a reduction of the overall thickness of the tandem structure. In this way, the devices will benefit from improved power conversion efficiency and/or decreased deposition time and material, which will reduce the overall production costs per installed kWp. Moreover, reducing the thickness of the tandem thin-film silicon solar cells will result in better stability of the cells [16].

Diffuse reflectors such as white paint are widely investigated as back reflector (BR) and prove to be very effective for both crystalline silicon [1719] and thin-film solar cells [17, 2025]. Compared to the alternative BR materials such as Ag, diffuse reflectors based on dielectric materials such as TiO2 have great potential for being applied as BR due to lower cost, better reflectivity, and lower parasitic absorptivity. A technique called Snow Globe Coating first described by Basch et al. in [25] is a simple and cheap way to form a layer of TiO2 diffuse reflector on the rear side of solar cells.

In this paper, we compare the optical and electrical performance of SGC back reflector and conventional ZnO:Al/Ag reflector on a-S:H/μc-Si:H tandem thin-film silicon solar cells and demonstrate the outstanding characteristics of SGC with high reflection, low parasitic absorption, and compatibility with textured cell structure. By simulating the plasmon-induced optical losses on the surface of the conventional ZnO:Al/Ag reflector, we elaborate the physical reasons for better optical performance of SGC. External quantum efficiency (EQE), current-voltage characteristics, and absorptance () spectra of the prototype solar cells are experimentally studied.

2. Experiment

2.1. Fabrication of Tandem Thin-Film Silicon Solar Cells

A three-dimensional (3D) schematic of the tandem thin-film silicon solar cell with SGC reflector is shown in Figure 1. Tandem thin-film silicon solar cells made of stacked a-Si:H top solar cell and a μc-Si:H bottom solar cell are prepared on as-grown textured commercially available SnO2:F superstrates (Asahi U-type from Asahi Glass Company [26]). The p-doped, intrinsic, and n-doped a-Si:H and μc-Si:H layers of the solar cells are deposited using a single-chamber plasma enhanced chemical vapor deposition process (PECVD) at a frequency of 13.56 MHz. More details on our deposition process of the solar cells can be found in [27]. On the rear side of the bottom solar cell, 80 nm ZnO:Al is sputtered, followed by a few nanometers (<5 nm) of SiO2 as a capping layer. The SiO2 layer maintains the stability of ZnO:Al in water during the Snow Globe Coating process [28] and provides sufficient ohmic contact for carrier collection. Then 700 nm of cross-finger Ag contact is thermally evaporated through a shadow mask with 2 mm pitch (as shown in Figure 1) to minimize the Ag covered area as well as maintaining good electrical conductivity. Finally, a TiO2 back reflector is fabricated by using the Snow Globe Coating technique, as described in the following section. In order to achieve the most relevant comparison between two types of back reflectors, after characterization the SGC reflector is removed and a layer of 700 nm thermally evaporated Ag is deposited on the same cells to define the conventional ZnO:Al/Ag back reflector which is used as the reference structure in this experiment.

2.2. Highly Reflective Back Reflectors with the Snow Globe Coating Process

The Snow Globe Coating process is based on commercially available titanium dioxide particles with average diameter of around 1 μm (TiO2-100, L32090 Auftrag no. 4497, Treibach-Althofen, Austria). A suspension of TiO2 particles with concentration of 0.5 wt% in 1000 mL deionized water is prepared in ultrasonic bath for 20 min. The tandem thin-film silicon solar cells are located at the bottom of a beaker and the suspension is slowly poured into the beaker. The mixture is left to stand for two hours until the particles settle; then the water is drawn out leaving TiO2 on the cells for drying at room temperature.

2.3. Characterization Methods

External quantum efficiency (EQE) of the top cell and bottom cell is measured separately. In order to measure one subcell, current density saturation of the other subcell is needed since the cells are connected in series. For top cell measurement from 300 nm to 800 nm, a bias light with RG695 filter is applied to saturate the bottom cell; an IF450 filtered bias light is used to saturate the top cell and measure the bottom cell from 450 nm to 1100 nm. The current-voltage characterizations are performed with a Wacom Class A (AM 1.5) solar simulator with a power density of 100 mW/cm2 (AM 1.5) at a constant temperature of 25°C.

We have also measured the optical properties of various types of back reflectors including SGC back reflectors, textured ZnO:Al/Ag back reflectors prepared on a textured substrate, and super flat ZnO:Al/Ag on glass substrates using a dual beam Perkin Elmer LAMBDA 950 (UV-Vis-NIR) spectrophotometer with integrating sphere. The same optical characterization method is applied on tandem thin-film silicon solar cells with SGC and conventional ZnO:Al/Ag back reflectors.

3. Result and Discussion

3.1. Tandem Thin-Film Silicon Solar Cells Employing Snow Globe Coating Reflector

In Figure 2, the EQE and absorptance of tandem thin-film solar cells with SGC back reflectors and ZnO:Al/Ag back reflector are presented. The absorptance of each structure is obtained by reflectance measurements via (where is measured reflectance since the transmittance is zero). EQE results of the top cells (a-Si:H) with SGC back reflector and the standard back reflector are almost identical at the wavelength range of 300–800 nm. This underlines the comparability of the performance of the tandem cells with two different rear structures. In contrast, for the bottom cell (μc-Si:H) from 450 nm to 1100 nm, EQE is improved significantly in the wavelength range from 650 nm to 1100 nm. The short circuit current density of the bottom solar cell increases from 10.4 mA/cm2 with conventional ZnO:Al/Ag back reflector to 10.9 mA/cm2 with SGC back reflector (shown in Figure 2). This enhancement clearly indicates the better optical performance of the solar cell with the SGC back reflector compared to the cell with ZnO:Al/Ag reflector. As we can see from Figure 2 (right -axis), absorption spectra of the cells with the two types of reflectors are identical for wavelengths shorter than 650 nm. The optical impact of back reflectors comes in the wavelength range > 650 nm, where the photons bounce once or more times within the structure and then escape through the front side of the solar cell. In this wavelength range, the absorptance of the solar cell with the SGC back reflector is slightly lower than the solar cell with the ZnO:Al/Ag back reflector. In combination with the EQE results, this is evidence that there is a larger portion of the absorbed light which does not contribute to the photocurrent generation in the solar cells with ZnO:Al/Ag back reflector. In other words, there is significant optical loss in the conventional back reflector compared to SGC back reflector. With lower overall absorptance but a higher portion coming from the active layers, the cell with SGC back reflector outperforms the one with conventional ZnO:A/Ag back reflector in terms of photocurrent generation.

Figure 3 shows the total internal quantum efficiency (IQE) defined by the fraction of collected carriers per photon absorbed. The results clearly indicate that a larger number of carriers are generated in the tandem thin-film silicon solar cell with SGC back reflector compared to the cell with ZnO:Al/Ag back reflector.

The characterizations of tandem thin-film silicon solar cells with both types of back reflectors are presented in Table 1. The SGC back reflector sample has slightly lower fill factor FF (71% to 72%) and open circuit voltage (1350 mV to 1367 mV) than the one with ZnO:Al/Ag back reflector. This small degradation in FF of the tandem thin-film silicon solar cells with SGC back reflector is caused by the different photocurrent mismatch of the component cells and the application of the cross-finger contact, which has a slightly lower collection efficiency than the conventional ZnO:Al/Ag back reflector, but less than 10% of the Ag consumption. The reason for small degradation is unclear and requires further investigation to clarify. However, the optical advantage of the SGC back reflector compensates the electrical loss and improves the overall efficiency by an absolute value of 0.2%.

The state-of-the-art tandem a-Si:H/μc-Si:H solar cells have achieved initial efficiency of over 13.5% and stabilized at around 12% on the sized cells in laboratory scale [15, 2932] and the module efficiency of over 10% [15, 30, 31]. The tandem solar cells that are used in this experiment are fabricated with standard process reported in [27]. The overall characteristics are less outstanding than cells with record efficiencies. However, the main focus of this study is trying to demonstrate the experimental proof of the concept of better performing diffuse dielectric reflector than the conventional metal reflector on a promising device configuration. On top of that the optical performance of the reference back reflector is comparable to the best known back reflector [33, 34]. The result of higher efficiency of SG coated sample confirms our assumption. Since the efficiency gain is mostly due to the short circuit current enhancement, a higher overall conversion efficiency increase for the SG coated cells can be expected on an electrically optimized device.

In conclusion, the SGC back reflector outperforms the conventional ZnO:Al/Ag back reflector because of its higher reflectivity and lower parasitic absorptivity over a broad spectral range. In the following section, we make a detailed explanation of the reasons for the better optical performance of SGC reflector compared to the ZnO:Al/Ag reflector.

3.2. Absorptance of SGC Back Reflectors and Conventional ZnO:Al/Ag Back Reflectors

We present absorptance of SGC and standard textured ZnO:Al/Ag back reflector, as well as a special type of super flat ZnO:Al/Ag as a function of wavelength in Figure 4. SGC back reflector shows close to 0% absorptance above 400 nm wavelength which indicates much lower parasitic absorption compared to the randomly textured ZnO:Al/Ag back reflector. The reflectance is even comparable to or lower than the super flat ZnO:Al/Ag back reflector for wavelengths beyond 450 nm. The super flat Ag structure was prepared by template stripping of an evaporated Ag layer from a conventional Si wafer with atomic scale flatness [35]. This layer serves as an optimal reference here and has proven to have excellent reflectivity in the literature [9, 35, 36]. For the textured ZnO:Al/Ag back reflector, which is comparable to the ZnO:Al/Ag back reflector applied on tandem thin-film silicon solar cells in the previous section, significant parasitic absorptance is measured for wavelengths longer than 400 nm. The enhanced absorption in the ZnO:Al/Ag back reflector introduces optical losses to the solar cells. This leads to lower EQE for wavelengths longer than around 550 nm where the incident light impinges multiple times at the rear reflector due to the light trapping effect of the randomly textured front side. Moreover, absorptance shows a specific characteristic including a distinct absorptance peak at wavelengths around 425 nm (Figure 4, blue dashed line). This peak is accompanied by an absorption tail that extends significantly into the wavelength range (500 nm < < 1100 nm). The enhanced parasitic absorptance could be due to two types of plasmonic losses which appear at textured Ag surfaces, namely, dissipative surface plasmon polariton (SPP) resonances [3740] at the textured ZnO:Al/Ag [8, 41, 42] interface as well as dissipative localized surface plasmon (LSP) resonances [43, 44] at protrusions of the textured ZnO:Al/Ag interface [4547]. These plasmonic losses are clearly related to the textured ZnO:Al/Ag interface since they are not apparent on the absorptance spectrum of the super flat ZnO:Al/Ag interface.

In order to discuss the plasmon-induced absorption of incident light at textured surfaces of Ag back reflector, the absorptance of thick Ag layers covered by ZnO:Al is studied using theoretical methods. We begin by investigating the possible contribution of SPP resonances to the parasitic absorption of textured ZnO:Al/Ag reflectors. SPP resonances are electromagnetic eigenmodes which propagate along the metal/dielectric interface with a wave-vector () parallel to the interface. The dispersion relation of the SPP eigenmodes at a flat ZnO:Al/Ag interface is calculated according to the following equation [44]:where is the angular frequency of the photon and is the speed of the light in the medium. and denote the dielectric function of the dielectric and metal, respectively. In Figure 5(a), the spectral dependence of (calculated by (1)) of SPP modes at a flat ZnO:Al/Ag interface is shown. As the incident light at normal angle exhibits no component of the wave-vector parallel to the Ag surface (i.e., = 0), no coupling of light to the SPP modes at a flat interface is possible due to momentum conservation. Therefore, in order to allow coupling of incident light and SPP modes at the Ag surface, an additional wave-vector component () must be provided by scattering at the randomly textured Ag surface. As shown in Figure 5(a), at the resonance wavelength (), the density of suitable wave-vector components () per wavelength interval is strongly enhanced. Thus, the excitation of propagating SPP modes is much more probable. Due to the damping of the SPP modes, the enhanced excitation of SPP modes leads to enhanced absorptance of incident light at randomly textured Ag surfaces at . Thus, the absorptance peak of the randomly textured ZnO:Al/Ag surface at the wavelength around 425 nm can be explained by coupling of the light to SPP modes at the randomly textured Ag surfaces.

Localized surface plasmons are also well known to induce optical losses at small Ag nanostructures [40, 43]. In Figure 5(b), the simulated maximum values of the absorption efficiency (max. ) at the dominant LSP resonance of hemispherical Ag nanostructures embedded in ZnO:Al are shown. The three-dimensional electromagnetic simulations are conducted with commercially available software JCMwave which was used extensively in previous studies to investigate at high precision localized plasmonic effects in Ag nanostructures [45]. For the investigated hemispherical Ag nanostructures embedded in ZnO:Al, strong absorptance is found. The absorption efficiency () denotes the absorbed intensity of incident light normalized by the intensity irradiated on the cross section of the nanostructures. In other words, if = 1 the entire light incident on the cross section of nanostructure is absorbed. Considering the values of shown in Figure 5(b) which are well above 1 for small nanostructures with radius < 100 nm, a significant fraction of the light incident on a nanostructured ZnO:Al/Ag interface can be absorbed, even with a fill fraction of nanostructures well below unity. Similar strong absorption as a function of the resonance wavelength has been observed for other geometries of the nanostructures as well as coupled nanostructures [48]. For this reason, considering the broad distribution of shapes, sizes, and arrangements of Ag nanostructures at the randomly textured Ag surfaces, LSP-induced optical losses are expected to be significant on the Ag surfaces. Note that substantial LSP-induced losses are found in small nanostructures of radius below 100 nm, which show strong nonradiative damping, in contrast to larger Ag nanostructures which have proven to show a beneficial strong light scattering effect [18, 4953].

Both LSP and SPP resonances can be excited by incident light on randomly textured Ag surfaces such as the Ag back contacts of thin-film silicon solar cells with the state-of-the-art random texture for light trapping. The agreement of the calculated with the spectral position of the absorptance maxima of the randomly textured Ag surfaces indicates that these absorptance maxima are likely to be caused by SPP-induced optical losses [7, 42]. However, the enhanced absorption at longer wavelengths on the surface of Ag back contacts with the random texture can be explained by both LSP-induced and SPP-induced optical losses. It is concluded that, for incident light of wavelengths longer than 500 nm, both types of plasmonic effects induce parasitic absorptance at the textured Ag back contacts covered with ZnO:Al. Since these plasmonic losses are not apparent in the absorptance of the super flat ZnO:Al/Ag device shown for comparison in Figure 4, the plasmonic losses are related to the textured ZnO:Al/Ag interface. However, textured cell structures are essential to thin-film solar cells for providing light trapping [54]. Therefore, the highly reflective SGC back reflector examined in this study provides an alternative to this dilemma. SGC reflector provides excellent reflectivity comparable to a super flat ZnO:Al/Ag back reflector as well as allowing a textured cell structure to be maintained.

4. Conclusion

Localized and propagating plasmonic resonances excited by incident light on randomly textured ZnO:Al/Ag back contacts of thin-film silicon solar cells can introduce significant amount of parasitic absorption thereby reducing the overall cell performance. Dielectric diffuse reflectors prepared by SGC show close to 100% reflectance over a spectral range of 400–1100 nm which indicates much lower parasitic optical losses compared to the standard textured ZnO:Al/Ag reflector. By replacing ZnO:Al/Ag with SGC reflector on tandem thin-film silicon solar cells as a rear reflector, the overall cell efficiency can be improved from 10.2% to 10.4%. Moreover, due to the relatively simple fabrication process of SGC reflector, it can be easily adapted to other types of solar cells such as mono/multicrystalline silicon solar cells, III–V solar cells, and perovskite solar cells. Given the fact of parasitic absorption and higher cost of Ag, SGC reflector has a very promising future in the PV industry as a replacement of Ag back reflector due to excellent optical properties, cheap price, and massive production of TiO2.

Competing Interests

The authors declare that there is no conflict of interests regarding the publication of this paper.

Acknowledgments

The authors would like to thank S. Gall and M. Weizman from HZB for the SEM image of SGC and T. Merdzhanova for helpful discussion.