1. Introduction

Thermophotovoltaics (TPV) can contribute to high-efficiency solar energy conversion by shaping the solar spectrum to match with a photovoltaic (PV) cell’s useful wavelengths via an intermediate thermal radiation emitter, which can be regarded as a photon-to-photon converter. Technologies of thermal radiation spectral shaping through thermal emitters, which suppress the emission of low-energy photons and increase the emission of photons with higher energies than the bandgap of PV cells, especially at high temperatures (>1000 °C), have been studied [1–10]. In solar-TPVs, the system design is essential to minimize thermal radiation loss. There have been many analytical studies on solar-TPV systems [11–17], whereas the number of experimental studies is limited. Yugami et al. demonstrated a solar-TPV system composed of a dish-type solar concentrator with a concentration factor of 25000, a gallium antimonide (GaSb) TPV cell (E_g = 0.7 eV), a graphite-cavity absorber, and an Al₂O₃/Er₃Al₅O₁₂ eutectic composite spectrally selective emitter [18]. Datas et al. also constructed a solar-TPV system with a concentration factor of 3813 using a Fresnel lens and secondary meniscus lens, arrayed germanium TPV cells (E_g = 0.66 eV), and a tungsten cyrindrical absorber/emitter coated with hafnium oxide as an antireflection coating [19]. Based on these early studies, the overall efficiency of solar-TPV systems has been drastically improved. Lenert et al. first achieved an overall efficiency of greater than 3% with an InGaAsSb (E_g = 0.55 eV) cell, carbon nanotube coating absorber, and one-dimensional photonic crystal emitter under Xe arc light irradiation with a power density of 75 W/cm² at the highest efficiency point [20]. Our group reported a solar-TPV system using a monolithic planar spectrally selective absorber/emitter [21] with an overall system efficiency of 5.1% [22] using a GaSb cell and spectrally selective absorber and emitter using a few-layer metal-dielectric coating [6] under irradiation from a high-power solar simulator with a 102 W/cm² power density. Bierman et al. achieved an overall efficiency of 6.8%, which exceeds the efficiency of direct irradiation of sunlight to a TPV cell, showing the advantage of the solar-TPV concept experimentally [23]. In their experiments, they used an experimental setup similar to that reported in [20], but they employed an optical filter to suppress unconvertible photons from being absorbed by the cell. Very recently, experimental demonstration of thermophotovoltaics have been reported, which does not focus on the solar-TPV spectral conversion process. Suemitsu et al. demonstrated an overall efficiency of 11.2%; however, their emitter was heated by Pt/Ti wires [24]. Bhatt et al. reported an overall efficiency exceeding 8%. Their device was implemented under laser illumination (λ = 808 nm) with an input power density of approximately 200 W/cm² at the highest efficiency point [25,26].

As shown above, it can be concluded that one of the reasons for the recent improvement in the overall conversion efficiency is the use of a planar-shape absorber/emitter. It allows increasing the temperature even with low input power by decreasing the surface area. Also, thermal stability improvement and controlling the thermal-radiation spectrum [5,7,25] has been progressed, which enables high-temperature operation. However, in the reported systems using a planar absorber/emitter, the distance between the emitter and the cell should be controlled precisely within ${\approx} $1 mm to obtain a high view factor, which would be a problem from a practical point of view.

In addition, planar systems have the problem of fully utilizing photons in terms of photon recycling [27]. Effective photon usage in a TPV system is slightly different from that in conventional PV systems. As the light source is close to the cells, non-absorbed photons that are reflected at the cell surface or boundary of the rear electrode surface can be recovered through re-absorption by the light source, i.e., emitter. TPV cells equipped with a back-surface reflector for lower energy photon recycling have been already reported [28–31]. Omair et al. realized a 29% conversion efficiency from blackbody-like thermal radiation to electricity using a single-junction InGaAs cell [32]. Also, Fan et al. measured heat-input-to-power-output efficiency, which is considering absorbed power to the cell, is more than 30% using an air-bridge TPV cell with near-perfect reflection of low-energy photons [33]. To take advantage of this feature in solar-TPVs, the photon transfer to the cell and the reflection of photons, which do not contribute to the PV process, back to the emitter should be efficient. To obtain a mutual photon transfer efficiency, i.e., emitted photons from the emitter to the cell and reflected photons from the cell to the emitter, the areas of the emitter and TPV cell should be large. In addition, to suppress re-emission loss from the absorber surface, it has been reported that the absorber area should be smaller than the emitter area [20,22]. However, if we make a large area difference between the absorber and the emitter in a planar absorber/emitter, thermal radiation loss from the surroundings of the absorber area cannot be neglected [22], and the in-plane emitter temperature distribution would be a problem. Therefore, planar systems would not be the optimized configuration from the points of view of both effective photon usage and practical realization.

Here, we propose a solar-TPV system equipped with a monolithic spectrally selective cuboid absorber/emitter confined in enclosed space consisting of TPV cells. By surrounding the cuboid emitter by TPV cells, it is possible to achieve a view factor of 100% between the emitter and the TPV cells mutually, even if the gap between the emitter and the cells is not as small as <1 mm, with the assumption that the reflectance of the reflector and surface electrode of the TPV cell is 100%. This means that photons emitted from the emitter can be used fully by absorbing photons with higher energy and recycling photons with lower energy than the PV cell’s bandgap. In addition, the area ratio (AR) between the absorber and the emitter is easily controlled to increase the relative emissive power from the emitter without any loss from the absorber surroundings. This type of geometry has been briefly and analytically discussed in a part of the paper reported by Wang et al. [15], but detailed analysis and experimental studies have not been performed yet. Moreover, Datas et al., constructed a system in which a cylindrical emitter was surrounded by cells [19]. In their paper, the advantage of obtaining a large AR is mentioned, but the optical confinement effect is not focused on. In this paper, we perform a detailed analysis of the proposed solar-TPV system that effectively utilizes emitted photons, and then experimentally demonstrate the advantage of the proposed system. We conclude that the advantages of the proposed system over planar systems can be shown experimentally.

2. Method

2.1 Analysis of the enclosed-space confined emitter

The proposed solar-TPV system, which employs a cuboid absorber/emitter confined by TPV cells, is shown in Fig. 1(a). The absorber/emitter is fabricated on the surfaces of a solid cuboid metal, which has a high melting point, such as molybdenum or tungsten. The top of the cuboid surface acts as a solar absorber, and the other sides and bottom of the five surfaces act as emitters. Each TPV cell faces the emitter surface, and a metal reflector with an aperture, whose size is equal to the absorber surface, is placed on top of the TPV cell box; hence, the emitter is confined in an enclosed TPV cell box.

 

Fig. 1. (a) Schematic illustration of the confined cuboid emitter system. The top of the cuboid acts as an absorber, and the other five surfaces act as emitters. Each of the emitter surface is faced to TPV cells, which form a TPV cell box, so that thermal radiation from the emitters can be confined in a box. The red and yellow arrows show convertible photons and recoverable photons by reflecting back at the rear electrode boundary, respectively. (b) Schematic illustration of a conventional planar emitter configuration. (c) Absorptance (emittance) spectrum of the absorber (red line) and emitter (blue line), which are measured in an actual absorber/emitter consisting of molybdenum and hafnium oxide layers. At infrared wavelengths, absorptance (emittance) is assumed to be at approximately 0.15 under operating temperature conditions, as shown by dashed lines. Actual hemispherical reflectance of the reflector and the GaSb TPV cell, applied in our experiment, measured by integrating a spherical setup, is shown with the black line and green line, respectively.

Download Full Size | PDF

With this configuration, a large AR can be easily obtained, as compared to planar absorber/emitter systems depicted in Fig. 1(b). In planar systems, to make the absorber area smaller than the emitter area, the surrounding of the absorber area becomes a dead area; i.e., does not act as an absorber or an emitter. Usually, such an area is covered by high-reflectance metals to reduce thermal radiation loss, although thermal radiation loss from the area cannot be neglected when the AR is increased. In addition, because the side of the planar absorber/emitter is also a dead area, the thermal radiation loss becomes large when the absorber/emitter thickness becomes high. However, the cuboid absorber/emitter has no, or very small, dead area. The AR can be easily designed by varying the height of the cuboid. In addition, the in-plane temperature distribution in the emitter becomes more uniform, compared with that in large-AR planar absorber/emitters, which are designed to be thin to minimize radiation losses from the sides. Furthermore, most photons from the emitter reach the TPV cells, and then higher energy photons than the bandgap of the TPV cell are absorbed by the cell, and lower energy photons are reflected back to the emitter, which contributes to the emitter temperature increase.

To assess this system, the overall conversion efficiency, i.e., from incident solar power to electrical output, the so-called system efficiency ${\eta _{\textrm{system}}}$, is evaluated by the following equation, ${\eta _{\textrm{system}}} = {\eta _{\textrm{ext}}} \times {\eta _{\textrm{trans}}} \times {\eta _{\textrm{PV}}}$. This equation is similar to that in our previous papers [16,22]. The transfer efficiency is additionally defined instead of using the view factor because energy transfer between the emitters and the TPV cells becomes more complex when considering multireflection.

In this equation, the extraction efficiency ${\eta _{\textrm{ext}}}$, which shows the amount of emissive power extracted from the emitters, is calculated by ${\eta _{\textrm{ext}}} = P_{\textrm{eff}\_\textrm{emi}}^{\textrm{emit}}/{P_{\textrm{in}}}$, where $P_{\textrm{eff}\_\textrm{emi}}^{\textrm{emit}}\; $ is the effective emissive power from the emitters and ${P_{\textrm{in}}}$ is the incident solar power. The effective emissive power $P_{\textrm{eff}\_\textrm{emi}}^{\textrm{emit}}$ is obtained by $P_{\textrm{eff}\_\textrm{emi}}^{\textrm{emit}} = P_{\textrm{emi}}^{\textrm{emit}} - P_{\textrm{return}}^{\textrm{emit}}$, where $P_{\textrm{emi}}^{\textrm{emit}}$ is the emissive power without considering multireflection and $P_{\textrm{return}}^{\textrm{emit}}$ is the re-absorbed power that is returned to the emitters through multireflection between the emitters and the TPV cells as shown in Fig. 1(a). The transfer efficiency ${\eta _{\textrm{trans}}}$ shows the amount of emissive power absorbed by the TPV cells from the emitters; in other words, it can be regarded as an effective view factor. It is calculated by ${\eta _{trans}} = P_{\textrm{eff}\_\textrm{abs}}^{\textrm{cell}}/P_{\textrm{eff}\_\textrm{emi}}^{\textrm{emit}}$, where $P_{\textrm{eff}\_\textrm{abs}}^{\textrm{cell}}$ is the effective power absorbed by the TPV cells. The effective absorbed power is defined by $P_{\textrm{eff}\_\textrm{abs}}^{\textrm{cell}} = P_{\textrm{abs}}^{\textrm{cell}} + P_{\textrm{return}}^{\textrm{cell}}$, where $P_{\textrm{abs}}^{\textrm{cell}}$ is the power absorbed by the TPV cells without considering multireflection and $P_{\textrm{return}}^{\textrm{cell}}$ is the absorbed power that is returned to the TPV cells through multireflection as depicted in Fig. 1(a). The PV conversion efficiency ${\eta _{\textrm{PV}}}$ is calculated by ${\eta _{\textrm{PV}}} = {P_{\textrm{out}}}/P_{\textrm{eff}\_\textrm{abs}}^{\textrm{cell}}$, where ${P_{\textrm{out}}}$ is the maximum electrical power extracted from the TPV cells. ${P_{\textrm{out}}}$ is evaluated by calculating a number of convertible photons by the cells using the quantum efficiency obtained based on the literature [34] and $P_{\textrm{eff}\_\textrm{abs}}^{\textrm{cell}}$.

In the evaluation, an absorber, an emitter, and TPV cells are assumed to have optical properties that are measured from the absorber/emitter used experimentally, as shown in Fig. 1(c). For the absorber and the emitter, a few-layer structure consisting of molybdenum and hafnium dioxide (HfO₂) is assumed to be applied, as in our previous reports [6,22]. The details of the absorber and emitter are explained in the following experimental section. For the TPV cell, a GaSb cell purchased from JX-Crystals Inc. was used in our experiments. Normal reflectance is measured, including surface electrodes. However, in the calculation, we assumed that the front electrode is a perfect reflector and photons are only absorbed by the cell. Therefore, the active area of the cell is assumed to have the measured optical properties and to convert photons to electrons. The reflector placed on top of the TPV cell box was also assumed to be a perfect reflector. The size of the cuboid absorber/emitter was assumed to have a 12 mm square top, which is similar to the experimental setup. Each parameter used in the analysis considering the ideal confinement model and the experimental model, which is explained in Sec. 2.2, is summarized in Table 1.

Table 1. Parameters and methods used in the analysis of the ideal confinement model and the experimental model.

View Table

Various ARs were realized by varying the height of the cuboid. The distance between each TPV cell and the faced emitter was fixed at 1 mm in both the planar and proposed configurations. Note that even when the distance was increased to 10 mm in the proposed system, the PV conversion efficiency drop, according to the irradiated power density decrease in the TPV cells, was small (<1%) because ${\eta _{\textrm{trans}}}$ does not change with respect to the distance if the reflector is assumed to have 100% reflectance. Therefore, a small distance is not essential for high PV conversion efficiency in this configuration, leading to reduced practical difficulties.

The extraction efficiency ${\eta _{\textrm{ext}}}$ was calculated for the proposed structure and a planar system with various incident power densities. In these analyses, the area of the absorber was fixed, and the area of the emitter was varied with increasing the height of the cuboid. In the planar system, the absorber and the emitter were assumed to have the same area; i.e., AR = 1. In the calculation, the absorptance and emittance of the absorber and emitter at infrared wavelengths are fixed at 0.15, as shown by the dashed line considering the increase in free-carrier absorption (emission), referring to a previous report [22]. In the planar absorber/emitter case, the thickness of the absorber/emitter is considered to be zero; thus, the thermal radiation loss from the side is neglected, and the effective emissive power $P_{\textrm{eff}\_\textrm{emi}}^{\textrm{emit}}$ is calculated by considering multireflection between the GaSb cells distant by 1 mm.

In each case, the extraction efficiency ${\eta _{\textrm{ext}}}$ is not largely dependent on the incident power density. A slight decrease can be observed in the low incident power density case owing to the shift of the emissive power peak toward long wavelengths with respect to the temperature decrease. It strongly depends on the AR, as mentioned in previous reports [16]. The extraction efficiency ${\eta _{\textrm{ext}}}$ is below 50% with the planar system, whereas it reaches 70% when AR = 5 and 75% when AR = 10, as shown in Fig. 2(a). In the ideal configuration shown in Fig. 1(a), the effective emissive power from the emitters is equal to the effective absorbed power by the TPV cells; thus, the transfer efficiency ${\eta _{\textrm{trans}}}$ becomes 100%. In the planar configuration, the emitted photons are not fully transferred to the TPV cell and are not fully reflected back to the emitter. The transfer efficiency ${\eta _{\textrm{trans}}}$ becomes 79% when the emitter and a TPV cell are of the same size even if faced with a 1 mm distance. The confinement performance of enclosed space is not only important for ${\eta _{\textrm{trans}}}$ but also for ${\eta _{\textrm{PV}}}$ because the emitter temperature can be increased by photon recycling. Nevertheless, under these conditions, the planar configuration shows the highest ${\eta _{\textrm{PV}}}$, as shown in Fig. 2(b). It can be understood that the planar absorber/emitter has the smallest surface area and reaches the highest temperature, as shown in Fig. 2(c). For a GaSb cell with a bandgap of 0.67 eV, the optimum operating temperature should be approximately greater than 2000 K, as shown in Supplement 1 (Fig. S1) because the emissive power spectrum is shifted toward short wavelengths according to Planck’s law and is well matched to the GaSb cell’s useful wavelengths at this temperature. This indicates that even if the photon-recycling effect is insufficient, the planar system is better for obtaining high ${\eta _{\textrm{PV}}}$ if the thermal stabilities of the absorber and emitter are sufficient. The photon-recycling effect can be recognized by comparing the solid and dashed lines in Fig. 2(c). The temperature difference increases with the temperature range. The maximum temperature difference shown in this graph is 24 K at 200 W/cm² with AR = 3. This temperature difference indicates that with photon recycling, it contributes to a 6.5% reduction of the input power, which is attributed to a 1.5% increase in ${\eta _{\textrm{PV}}}$. If the back-surface reflection of the TPV cell can be increased as reported [32,33], the photon-recycling effect becomes more explicit with an increase in ${\eta _{\textrm{PV}}}$.

 

Fig. 2. Calculated (a) extraction efficiency, (b) photovoltaic conversion efficiency and (c) temperature of the emitter in various geometries are shown versus input power density. The dashed lines in (c) show the reached temperature without the photon-recycling effect.

Download Full Size | PDF

By multiplying each efficiency shown above, the system efficiency ${\eta _{\textrm{system}}}$ can be obtained, as shown in Fig. 3. At a low input power density, ${\eta _{\textrm{system}}}$ in the planar system is comparable to that in the proposed system. However, the difference can be clearly seen at high-input power density. In the planar system, even with high ${\eta _{\textrm{PV}}}$, ${\eta _{\textrm{ext}}}$ is significantly lower than in the proposed system because the thermal radiation loss from the absorber is relatively owing to the low AR and ${\eta _{\textrm{trans}}}$ is not sufficiently high even if the emitter and the PV cell are closely faced with a 1 mm distance.

 

Fig. 3. Calculated system efficiency in various geometries are plotted versus input power density.

Download Full Size | PDF

As a result, the system efficiency only reached 10% at 200 W/cm². For the proposed system, the low-AR system (AR = 3) shows the highest efficiency of 14% at this input power density, which makes it possible to build a system with a Fresnel lens concentrator for practical use.

From the analysis, it is shown that the proposed system is superior to the planar system owing to its high extraction efficiency ${\eta _{\textrm{ext}}}$ and high transfer efficiency ${\eta _{\textrm{trans}}}$. In addition, the photon-recycling effect by the enclosed-space configuration contributes to a high PV conversion efficiency ${\eta _{\textrm{PV}}}$.

2.2 Experimental setup and procedure of the power generation test

The performance of the proposed solar-TPV system with the enclosed-space confined emitter was experimentally evaluated. A solid cube of molybdenum with a multilayer coating was used as the monolithic cuboid absorber/emitter, as shown in Fig. 4(a). The black colored top surface corresponds to the absorber, and the five metallic blue colored surfaces correspond to the emitters. These colors indicate that the absorber can absorb most of the visible light and that the absorption (or emission) peak of the emitters is at a longer wavelength than that of the absorber, which is in line with the optical properties shown in Fig. 1(c). If the confinement performance degrades and ${\eta _{\textrm{trans}}}$ decreases from 100%, ${\eta _{\textrm{ext}}}$ becomes more important. Therefore, a cube absorber/emitter with AR = 5 was used in the experiment. For the absorber surface, to have a spectrally selective solar absorption property, a layered structure based on the coherent perfect absorption concept as presented in previous reports [6,22] was fabricated, consisting of a molybdenum nanometric layer sandwiched by an HfO₂ layer. Each layer thickness was 50 nm of bottom HfO₂, 9 nm of molybdenum, and 60 nm of top HfO₂. This multilayer was also applied to the emitters, with thicknesses of 190 nm for bottom HfO₂, 11 nm for molybdenum, and 90 nm for top HfO₂. The absorptance spectra of the absorber and emitter are shown in Fig. 1(c). The cube was surrounded by five GaSb PV cells placed on a copper substrate, as shown in Fig. 4(b). Each cell was 12 mm wide, 16 mm high, and 0.7 mm thick, and placed on a copper substrate, which was electrically insulated by a thin ceramic layer from the PV cell. These copper substrates were placed on a water-cooled heat sink to keep the cells at around 25°C.

 

Fig. 4. (a) Picture of the fabricated cubic absorber/emitter. (b) Schematic illustration of the equipment for the power generation test. (c) Picture of the emitter set at the equipment. Although the GaSb TPV cells are attached in front of every five surfaces of the emitter, one cell is connected with a cable to evaluate the power generation performance. The temperature of the emitter is measured by an R-type thermocouple attached at the bottom of the cube shown in the left picture. The right picture was taken immediately after the incident light was off. The absorber surface becomes red and hot, and almost no emission is leaked from the TPV cell box.

Download Full Size | PDF

The cell facing the side surfaces of the cube was placed 1 mm away from the emitter surface, and that facing the bottom was placed 5 mm away from the emitter. Because of small gaps (${\approx} $1 mm) between adjacent cells, the copper substrate was polished to be a mirror surface so that emitted photons passed through these gaps could be reflected. A polished molybdenum pinhole-plate was employed to obtain homogeneous illumination of the concentrated simulated sunlight onto the absorber surface. A pinhole was designed to be 11 mm in diameter, which is slightly smaller than the absorber area, to avoid direct incidence of light into the PV cells. The pinhole-plate also functioned as a reflector of the photons emitted from the emitters. The reflectance spectrum of the reflector is shown in Fig. 1(c). To avoid the contact between the pinhole-plate and the cube, a 1 mm gap was introduced between them. Three platinum wires with a 0.2 mm diameter were used to suspend the cube. One end of each wire was attached to the absorber edge using spot welding, and the other end was tied to a zirconium oxide screw through a small hole of the pinhole-plate shown in Fig. 4(b). The temperature of the cube was measured by a thermocouple attached to the bottom of the emitter surface.

As explained above, as the established enclosed space is not perfect as assumed in the analysis, the transfer efficiency ${\eta _{\textrm{trans}}}$ is evaluated by the ray-tracing method. By counting the number of rays that exit the enclosed space and are absorbed by the reflector and the walls (except the cells), the efficiency is analyzed to be 82%; i.e., it is still higher than in the planar setup with a 1 mm gap. Because the enclosed space is designed considering the use of the existing GaSb TPV cell size, the efficiency is supposed to be increased by improving the enclosed-space design further.

Concentrated simulated sunlight generated by a high-power solar simulator (Yamashita Denso, YSS-C1000HR) was used as the incident light in the experiment. The light was incident from the front window of a vacuum chamber and reflected down by a silver mirror placed on a heat sink at an angle of 45°. The spectrum of the light from the solar simulator is shown in Fig. S2 in Supplement 1. Note that this spectrum matches well with that of sunlight and is classified as Class-A in the JIS class specification. Therefore, we believe the difference in system efficiency arising from using actual concentrated sunlight would be very small. In the solar simulator, a cylindrical rod light guide is placed in front of the lump between concentrating lens optics. Therefore, by changing the lump position to change the power input to the light guide, the power density of the light irradiated to the absorber can be controlled. The actual input power density was evaluated using a heat flux sensor (Medtherm, model GTW-300-40-484).

A picture of the setup immediately after turning off the input sunlight is shown in Fig. 4(c). The absorber/emitter is heated to a high temperature as can be observed from the red absorber surface from the pinhole. However, almost no emission light can be seen from the side of the copper block, indicating that the light emitted from the emitter is well confined in the enclosed space.

3. Results and discussion

A power generation test was conducted using the proposed solar-TPV system under several input power density irradiations. During the experiments, the system was vacuumed at 1×10⁻³ Pa.

The measured temperatures of the emitter are in good agreement with the analyzed ones, as shown in Fig. 5(a), considering the emission light leakage in the actual enclosed space and conductivity loss from the wires and the thermocouple. In the low-input power density range, the measured temperature is slightly above the calculated line because the emittance at wavelengths corresponding to lower energy than the bandgap might not increase to 0.15 in this temperature range. Thus, we overestimated the emissive power from the emitter. To evaluate each efficiency, ${\eta _{\textrm{ext}}}$, ${\eta _{\textrm{trans}}}$, and ${\eta _{\textrm{PV}}}$, the reflected power and the emissive power from the absorber were calculated using the absorber optical property, as shown in Fig. 1(c), and the measured temperatures, as plotted in Fig. 5(a). The effective emissive power from the emitters and the effective absorbed power by the TPV cells were similarly analyzed using the emitter and the PV cell’s optical properties and the measured temperatures. The agreement between the measured and analyzed temperatures indicates that the model used in the analysis of the system energy flow, such as energy leakage from the fabricated enclosed space and heat conduction loss through the wires, well reproduces the characteristics of the actual setup.

 

Fig. 5. (a) Obtained temperature of the emitter and (b) transfer and extraction efficiencies estimated from the measured temperature are plotted.

Download Full Size | PDF

The maximum output power from the PV cell is obtained by measuring the current–voltage plots using a source measure unit (DC Voltage Current Source/Monitors 6244, ADCMT). It should be noted that the current–voltage measurements are conducted for only one cell facing the side surface of the emitter. Next, the measured electrical power is multiplied by five to obtain the maximum output power, which is used in the calculation of the PV conversion efficiency and system efficiency. A PV conversion efficiency of 13.3% was obtained at an input power density of 81 W/cm² at 1401 K. As shown in Supplement 1, Fig. S3, temperature difference between top and bottom of this cuboid at around 1400 K range is only less than 10°C. Therefore, we consider the total efficiency evaluated with this way is highly expected with a full set of the cells.

The analyzed system efficiency is shown in Fig. 6(a). The measured efficiencies are shown by square dots, which are calculated from the measured output power of one cell. The efficiency becomes 7.0% after multiplying the power measured from one cell by five. Detail of the energy flow analysis using the model is shown in Fig. 6(b), in which 66.4% of the input power is transferred to the emitter. A total of 11.0% of the input power is leaked from the enclosed space, but 4.0% is reabsorbed by the emitter and contributes to an increase in its temperature. Consequently, 7.0% of the input power is converted to electrical power.

 

Fig. 6. (a) Measured system efficiencies are plotted with red dots. The analyzed efficiency using the model of the fabricated setup (η_trans = 0.82, R_{λ>1.8 µm} = reflectance of the actual GaSb cell) is shown by red line. The blue line shows the estimated efficiency assuming R_{λ>1.8 µm} = 0.9, and the green line assuming η_trans = 1, R_{λ>1.8 µm} = 0.9. (b) Energy flow at an input power density of 81 W/cm² where the maximum efficiency is obtained in the experiment.

Download Full Size | PDF

By increasing the input power density to increase the temperature of the emitter, the system efficiency is expected to reach 10% with this setup. For further improvement, one option is to increase the reflectance of the TPV cell at a lower photon energy range than the bandgap. If the reflectance is increased to as high as 90%, the PV conversion efficiency is increased by 2% at an input power density of 200 W/cm² because the emitter temperature is increased by increasing the photon recycling rate. As a result, the system efficiency is expected to become 13% with the same setup. The photon-recycling effect strongly appears if the confinement performance of the enclosed space becomes large. Therefore, a system efficiency of >20% can be obtained by improving the transfer efficiency to be 1 with perfect emission light confinement in the enclosed space.

4. Conclusion

In this study, an enclosed-space confined emitter system is proposed, and its performance is evaluated experimentally. With this system, photons emitted from the emitter effectively reach the TPV cell, ideally, 100%, even if we increase the gap between the emitter and the TPV cell.

A cuboid absorber/emitter is used to realize the confinement geometry. Another reason for adopting the cuboid absorber/emitter is that it contributes to the increase in the AR between the absorber and the emitter because the dead area can be zero, whereas a planar absorber/emitter has a dead area if the absorber area is reduced. Furthermore, the temperature distribution can also be diminished, as compared to the planar absorber/emitter. The extraction efficiency can be greater than 70% if we increase the AR of the cuboid absorber/emitter to greater than 5.

The proposed system is also effective not only for increasing the photon transfer efficiency to the TPV cell but also for photon recycling. The temperature of the emitter can be increased by photon recycling in the system. For example, it is analyzed to increase by 24 K, which is attributed to a 1.5% increase in the PV conversion efficiency ${\eta _{\textrm{PV}}}$ at 200 W/cm² input in the system with AR = 3. However, in our analysis, the planar absorber/emitter shows the highest ${\eta _{\textrm{PV}}}$ because the temperature becomes considerably higher than the cuboid one. The use of high-reflectance cells at wavelengths corresponding to lower energy than the bandgap of the TPV cell can contribute to minimizing the gap.

Consequently, the system efficiency ${\eta _{\textrm{system}}}$ of the planar system is comparable, but the proposed system shows the advantage of the higher input power density range. Almost 10% higher efficiency is expected in the proposed system than that in the planar system at 200 W/cm².

A power generation test was conducted with the proposed system, and a transfer efficiency ${\eta _{\textrm{trans}}}$ of 82% was predicted through an analysis using the ray-tracing method. The measured and analyzed temperatures are consistent. In addition, the extraction efficiency ${\eta _{\textrm{ext}}}$ and ${\eta _{\textrm{PV}}}$ calculated from the measured temperature are also consistent with the analyzed results. Hence, the analyzed model reproduces the experimental system well. At an input power density of 81 W/cm², ${\eta _{\textrm{system}}}$ becomes 7.0%, which is obtained after multiplying the power measured from one cell by five. At 200 W/cm², ${\eta _{\textrm{system}}}$ = 11% is expected with the same setup. Furthermore, it can be increased to ${\eta _{\textrm{system}}}$ = 13% by increasing the reflectance in infrared wavelengths, according to photons with lower energy than the bandgap, owing to the photon-recycling effect. In addition, if ${\eta _{\textrm{trans}}}$ can be increased to 100%, by improving the enclosed-space configuration, ${\eta _{\textrm{system}}}$ is increased to greater than 20%.

According to the results, it is concluded that the enclosed-space confined cuboid absorber/emitter system contributes to effective photon usage in solar-TPV systems. Moreover, it is expected to have higher system efficiency than the recent planar systems, even within 200 W/cm², which is available with practical Fresnel lens optics.