1. Introduction

High-concentration photovoltaic (HCPV) cells have been extensively prepared using group III–V compound semiconductors, such as GaInP/GaInAs/Ge tandem solar cells, to achieve high efficiency [1]. In_xGa_{1 − x}N alloys simultaneously cover an extensive solar spectrum range, particularly those higher than 2 eV and lower than 1 eV, which is suitable for specific optimum multiple band gaps for future tandem cells with ultrahigh power conversion efficiency (PCE) [2–4]. This field has drawn significant attention, and numerous reports have been published [5–7]. However, few reports on the HCPV characteristics of InGaN-based materials have been published. Dahal et al. reported their work on 30-sun concentration levels by a multiple quantum well (MQW)-type cell with back metal reflectors [8]. Although the photocurrent was enhanced 15% by the aluminum back reflectors, the fill factor (FF) significantly decreased by 11% mainly because of serious recombination [8]. Based on our previous reports, InGaN/sapphire-based solar cells with blue band GaN/InGaN MQW absorption layers grown on patterned sapphire substrates were characterized under high concentrations of up to 150-sun AM1.5D testing conditions. When the concentration ratio increased from 1 sun to 150 suns, the peak PCE occurred at the 100-sun concentration where the solar cells maintained an FF as high as 0.70 and exhibited a PCE of 2.23% [9]. In other words, the PCE decreased when the concentration ratios increased to 150 suns. This decreased PCE could be attributed to the thermal effect caused by the poor thermal conductivity of sapphire. On the other hand, the thickness of the absorption layer in a solar cell made by using a conventional semiconductor with a direct band gap is only a few micrometers to yield high internal quantum efficiency because of high absorption coefficients (~10⁵ cm⁻¹) around its energy band gap [10]. However, the thickness limit of high-quality In_xGa_{1 − x}N epitaxial layers grown on GaN/sapphire substrate is only hundreds of nanometers because of the large lattice mismatch between In_xGa_{1 − x}N and GaN, particularly for high indium (In)-containing materials [11]. Therefore, achieving In_xGa_{1 − x}N absorption layers with low band gap (less than 2.0 eV) and large thickness remains a challenge. In this study, we propose a green band vertical InGaN/GaN-based solar cell on silicon substrate fabricated by wafer bonding technique to alleviate the thermal effect on performance degradation. This vertical structure with a metal reflector sandwiched between the GaN-based epitaxial layers and the Si substrate can increase the effective thickness of the absorption layer [12]. Solar responses under concentrated levels of up to 300 suns are discussed in the following section.

2. Device fabrication and experiment methods

InGaN epitaxial layers were deposited on c-face sapphire substrates by a metalorganic vapor-phase epitaxy reactor. The layer structure consisted of a 30 nm-thick low-temperature GaN nucleation layer, followed by a 2 µm-thick undoped GaN with chamber pressure at 500 Torr. Thereafter, a p–i–n heterostructure consisting of a 3 µm-thick Si-doped n⁺-GaN (n ~5 × 10¹⁸ cm⁻³), an undoped InGaN/GaN (2.5/14.5 nm for 12 pairs) MQW structure, and a 200 nm-thick Mg-doped p-GaN (p ~5 × 10¹⁷ cm⁻³) were sequentially deposited. The typical peak wavelength of the electroluminescence spectra taken from the InGaN/GaN/Si solar cells was approximately 525 nm, corresponding to the In composition of approximately 28% in the InGaN well layers [13, 14]. After epitaxial growth, a bilayer metal of Ni/Ag (1/200 nm) was deposited onto the p-GaN top layer to serve as reflector/ohmic contact layer [15]. After the formation of the reflector layer, a barrier layer, which was configured to alleviate the diffusion of Ag from the reflector layer to the bonding layer and consisted of a 200 nm-thick TiW layer and a 50 nm-thick Pt layer, was deposited between the reflector and bonding layers. In the proposed vertical InGaN/GaN/Si solar cells, a 3 µm-thick In layer was used as the bonding layer. The Si substrates were dipped in buffered oxide etch solution for 60 s to remove the native silicon dioxide, then a bilayer Ti/Au (20/1,500 nm) metal was deposited onto the surfaces of Si substrates to serve as ohmic contact and bonding layer. These wafers served as receptors for the wafer bonding process. After the bonding process, the sapphire substrate was removed by laser lift-off technique to expose the n⁺-GaN layer. Thus, the In_xGa_{1 − x}N/GaN-based heteroepitaxial layers were transferred to the Si substrate with the n⁺-GaN top layer. Then, the samples were treated in potassium hydroxide (concentration of 3 M) solution at an elevated temperature of 60 °C to texture the n⁺-GaN layer to enhance light trapping. Ti/Al/Ni/Au (20/30/150/2,000 nm) metal layers were then deposited onto the exposed n⁺-GaN layer to form the n-type ohmic contacts (cathode electrodes) on the wafers. Finally, the Si substrates were thinned to 150 µm and coated with Ti/Au (50/500 nm) metals to serve as the back ohmic contact layer. The entire fabricated device area was 1 × 1 mm². Figure 1(a) shows a typical top view scanning electron microscopy image taken from the proposed vertical InGaN/GaN/Si solar cells. The dashed line in Fig. 1(a) indicates where the cross section corresponds to a schematic layer structure of the vertical solar cell, as shown in Fig. 1(b).

 

Fig. 1 (a) Typical top-view SEM image taken from the proposed vertical InGaN/GaN/Si solar cell. The inset shows the enlarged SEM image taken from the surface. (b) Schematic layer structure of the vertical solar cell.

Download Full Size | PDF

3. Results and discussion

Figure 2(a) illustrates the typical characteristics of current density and power density versus voltage (J–V and P–V, respectively) of the vertical InGaN/GaN/Si green band solar cells (vertical green cells) illuminated by the 1-sun condition, which was calibrated by the calibration cell of the National Renewable Energy Laboratory with global air mass of 1.5 (AM1.5G) terrestrial solar spectrum. The typical V_OC, short-circuit current density (J_SC), and fill factor (FF) of the solar cells were 1.90 V, 2.07 mA/cm², and 0.55, respectively, corresponding to the PCE of 2.16%. Although the conversion efficiency was significantly higher than that reported for cells with green-band InGaN absorption layers [16], the PCE was still considerably lower than the theoretical value by single-junction materials with optical band gap at 2.36 eV (525 nm) [17]. The PCE was mainly limited by material quality because of the large lattice mismatch between InGaN and GaN [18]. Figure 2(b) shows the relative external quantum efficiency (EQE) as a function of incident light wavelength. The main response covered a part of visible light in the blue and green regions with a long-wavelength cutoff of approximately 520 nm. This wavelength cutoff was consistent with the emission peak wavelength as the vertical green cells were driven with a forward current of 350 mA, as illustrated in Fig. 2(b). In addition, a steep decrease in EQE in the ultraviolet (UV) region was attributed to the surface absorption of the n⁺-GaN top layer with a thickness of 3 µm, which would lead to a short-wavelength cutoff at approximately 365 nm. Although the incident photons contributed by the UV light are far less than those of the visible and infrared light in the solar spectrum, the thickness of the n-GaN top layer could be further decreased to collect more photogenerated carriers from UV light and alleviate the heating effect. In principle, the thickness of the n-GaN top layer should be considerably less than 1 µm because the absorption coefficient of GaN is as high as ~1 × 10⁵ cm⁻¹ [10]. In a p–i–n homojunction or a lattice-matched heterojunction solar cell with a single absorption layer, conversion efficiency increases theoretically with the decrease in the band gap of the absorption layer. However, the lattice mismatch between the GaN and In_xGa_{1 − x}N limits the critical thickness of the In_xGa_{1 − x}N layer in a p–i–n GaN/In_xGa_{1 − x}N heterojunction solar cell, particularly when the In contents are increased to convert more sunlight. In this study, the proposed vertical green-band cells with a metal reflector were combined with silicon substrate by using the wafer bonding technique to increase the effective thickness of the InGaN absorption layer. Moreover, the Si substrate with higher thermal conductivity replaces the sapphire substrate, allowing the vertical green cells to operate under high solar concentration. A solar simulator concentrator (Oriel Instruments) was used to study the concentrated solar response of MQW-type solar cells made from InGaN/GaN-based materials and to produce different irradiance intensities of 50 suns to 300 suns with the AM1.5D solar spectrum (ASTM G-173-03). In this study, flash-mode light source with light pulse width of 500 ms was applied to irradiance the bare cells (not in a packaged form).

 

Fig. 2 (a) Typical J-V and P-V characteristics of the vertical GaN/InGaN solar cells with green-band absorption layer illuminated by the one-sun AM1.5G condition. (b) Relative external quantum efficiency (EQE) as a function of incident light wavelength and electroluminescence spectrum as the vertical green cells was driven with a forward current of 350 mA.

Download Full Size | PDF

Figure 3 shows the typical J–V characteristics under different irradiance intensities. Evidently, the short-circuit current density (J_SC) increases linearly with irradiance intensity. In an equivalent circuit of a solar cell modeled by an ideal current source (J_SC) in parallel with a diode, the J–V characteristic is expressed in Eq. (1) as follows:

 

Fig. 3 Typical J-V characteristics under different irradiance intensities up to 300 suns.

Download Full Size | PDF

 

Fig. 4 (a) V_OC and FF as functions of solar concentration. (b) PCE as functions of solar concentration.

Download Full Size | PDF

Because the V_OC was approximately 2.0 V for the proposed vertical green cells, Eq. (3) can be estimated roughly as follows:

In accordance with our previous results measured from the lateral InGaN/GaN-based blue band solar cells (lateral blue cells) grown on sapphire, the lateral blue cells with a peak PCE of 2.23% occurred at 100-sun irradiance intensity, and the PCE rolled off because of the decreased FF as the concentration ratio reached 100 suns [5]. The decrease in PCE at high irradiance intensity could be attributed to the fact that the contribution of incremental V_OC in PCE could not countervail the degradation of FF. The lower FF at high irradiance intensity could be attributed to the relatively significant power loss compared with low irradiance intensity. In other words, power consumption in the series resistance (R_s) increased with the increase in cell current (i.e., under high illumination) [20]. Therefore, we should decrease R_s further to allow for GaN-based solar cells operating under high concentration.

To extract R_s from solar cells, we analyzed the J–V characteristics of solar cells under different solar concentrations. The R_s value of the vertical green cells was determined to be 1.77 Ω cm² on average. This R_s value was greater than that of conventional tandem solar cells (e.g., InGaP/GaAs/Ge tandem cells) but lower than our previous blue-band solar cells [5]. In the proposed vertical green cells, InGaN/GaN-based p–i–n heterojunctions have relatively greater band discontinuity and low-conductivity p-GaN contact layer, which result in high R_s. In addition, the bonding interface behaves like a tunneling junction and the thick Si substrate can contribute additional resistance to the devices. Although tunneling junctions between subcells dominate the R_s in the conventional tandem cells, mature epitaxial growth technique and small lattice mismatches between subcells stacked by heteroepitaxy on the same substrate in sequence allow the tandem cells to operate under high solar concentration [21]. By contrast, InGaN-based solar cells grown on sapphire substrates suffer from high threading dislocation density because of large lattice mismatches between III-nitride semiconductors and sapphire. Therefore, the dense structure defects (e.g., threading dislocation) in InGaN-based solar cells significantly degraded the device performance. In a fabricated solar diode, extracting the ideality factor from the J–V characteristics can be indirectly applied to evaluate the effect of material quality on device performance. The ideality factors of the vertical green cells based on Eq. (2) were determined at approximately 1.6. This ideality factor was markedly lower than that reported for GaN-based concentrator solar cells grown on sapphire substrates [8]. In a typical GaN/InGaN heterojunction, numerous factors may result in ideality factors that are significantly greater than 2, including tunneling through barriers, interface defects, and charge-related defects with a pinned Fermi level [1, 8]. In GaAs-based tandem cells, the ideality factor is greater than unity and typically greater than 3. This condition can be attributed to the epitaxial interface profile and mismatched current in each subcell [22, 23]. Although the research on GaN-based tunneling diodes applied to InGaN/GaN-based tandem devices is in its infancy, the vertical InGaN/GaN/Si solar cells fabricated by using the wafer bonding technique to produce subcells with different absorption spectra together is a potential alternative.

4. Conclusions

We demonstrated that vertical InGaN/GaN/Si solar cells operated under high solar concentration of up to 300 suns. The green band InGaN/GaN-based solar cells showed typical characteristics of HCPV cells. The V_OC increased from 1.90 V to 2.15 V, and the FF increased from 0.55 to 0.58 when the concentrated level was increased from 1 sun to 300 suns. At the 300-sun irradiance intensity, the PCE was enhanced by approximately 33% compared with the 1-sun irradiance intensity.