Open Access
Add to BibSonomyAbstract
This paper presents results on a highly efficient experimental solar thermophotovoltaic (STPV) system using simulated solar energy. An overall power conversion efficiency of 6.2% was recorded under solar simulation. This was matched with a thermodynamic model, and the losses within the system, as well as a path forward to mitigate these losses, have been investigated. The system consists of a planar, tungsten absorbing/emitting structure with an anti-reflection layer coated laser-microtextured absorbing surface and single-layer dielectric coated emitting surface. A GaSb PV cell was used to capture the emitted radiation and convert it into electrical energy. This simple structure is both easy to fabricate and temperature stable, and contains no moving parts or heat exchange fluids.
© 2015 Optical Society of America
1. Introduction
Solar thermophotovoltaic (STPV) systems are energy conversion methods that are capable of overcoming the Shockley-Queisser efficiency limit of 32.1% for silicon photovoltaic (PV) cells [1]. In fact, the upper theoretical limit for the efficiency of STPV systems is 85.4% [2]. This is possible because in STPV systems the broadband solar spectrum is converted into a narrow spectrum tailored for use in a PV cell. This emission spectrum typically consists of photons with an energy just above the bandgap energy (Ebg) of the PV cell, resulting in greatly reduced thermalization and transmission losses in the cell.
Figures 1(a) and 1Figures 1(a) show two common system architectures in this field: planar and cylindrical systems. While the cylindrical architectures allow more control over the size of the emitting surface, they require large PV cell areas and are extremely difficult to achieve. Planar structures are much more simple and allow for a reduced PV cell area, making them ideal for this application. The key to realizing highly efficient STPV systems is precise control of the optical properties of the light absorbing and emitting surfaces. The absorbing surface must efficiently absorb solar energy, while simultaneously minimizing the emission of thermal energy, and the emitting surface must have high emission in a narrow band just above the Ebg of the PV cell used. This spectral control is typically achieved through the use of nanostructures or thin film coatings; this paper will consider a combination of both methods.
Fig. 1 Diagram of a) a flat STPV system and b) a cylindrical STPV system, both with solar absorber and thermal emitter.
The choice of PV cell in an STPV system is primarily determined by the operating temperature. The maximum efficiency for an STPV system occurs when the blackbody peak of the emitting surface is near the Ebg of the PV cell used [2]. This allows the PV cell to absorb a large portion of emitted thermal energy, increasing the efficiency of the system. Narrowing the emission spectrum by controlling the optical properties of the emitting surface will further increase the portion of emitted energy that is usable by the PV cell. The wavelength of peak emission of a blackbody at a certain temperature is given by Wien’s displacement law. Due to these restrictions, PV cells commonly used in STPV systems include germanium (Ge), gallium antimonide (GaSb), and indium gallium arsenic antimonide (InGaAsSb) cells. Silicon (Si) cells would require an operating temperature of about 2600 K for efficient operation, making them a poor choice for most STPV systems. Ge and GaSb cells have high efficiencies at operating temperatures around 1600 K, while InGaAsSb cells operate efficiently at temperatures around 1250 K [3].
2. State of the art
Despite the extremely high theoretical efficiency possible in STPV systems, experimental efficiencies remain low. Early STPV systems focused on cylindrical geometries, with large cavities for an absorbing surface and bulk tungsten or rare earth compounds as emitters [4–6 ]. Sunlight was concentrated onto these systems via Fresnel lens, raising their temperatures as high as 1680 K, and emitted radiation was collected by Ge or GaSb PV cells. These early designs all had <1% efficiency, with PV cell heating, temperature gradients along the cylinders, and a lack of efficient absorbing and emitting surfaces, lowering the efficiency.
In 2014, a paper was published by A. Lenert et al. demonstrating an STPV system with 3.2% efficiency, sparking a renewed interest in the field [7]. This system used a planar geometry to simplify the architecture, reduce the required PV cell area, and remove the problem of a temperature gradient across the emitting surface. InGaAsSb PV cells with a 0.55 eV bandgap were used to allow the system to operate at lower temperatures (the 3.2% efficiency was recorded at 1,285 K), which afforded increased material stability and reduced the level of solar concentration required for operation. A multi-walled carbon nanotube blackbody absorber was used for the absorbing surface, and a Si/SiO2 Bragg stack was used for the emitting surface. Experimental validation of this system was performed using a Xenon-arc light source and concentrating lens system to simulate the solar spectrum. Due to the fact that a blackbody absorber was used, spectral mismatch between solar energy and the Xenon-arc light source used in the experiment would have a negligible effect on the results; however, solar concentrator systems have optical losses up to 50% that would not be present in this simulated setup. The low efficiency of this system is primarily due to the low operating temperature and low efficiency PV cells.
A recent 2015 paper by M. Shimizu et al. showed a ground-breaking experimental efficiency of 8% using a planar geometry and GaSb PV cells [8]. Due to the 0.75 eV Ebg of these cells, high operating temperatures were required for efficient operation, and the system was operated at a temperature of 1640 K. Both the absorbing and emitting surfaces consisted of a stack of a yttria-stabilized zirconia (YSZ) layer followed by a tungsten (W) layer, followed by an additional YSZ layer and a W substrate. This resulted in reduced thermal emission from the absorbing surface; however, the absorption band was narrow and a 20% reflection loss from the absorbing surface was reported. Again, a solar simulator was used to illuminate the setup, and potential solar concentrator losses were not taken into account.
Despite the large advances in STPV systems, further research is needed to improve system efficiencies to make them competitive with traditional PV systems. This article aims, through modeling and experimental studies, to examine the losses in an experimental system in detail to provide a path forward to a more efficient STPV system.
3. System design considerations and simulation
The STPV system presented here consists of a solar concentration system, a vacuum chamber to reduce convective losses and ensure material stability, a tungsten absorbing/emitting structure with a laser-textured blackbody absorbing surface and dielectric-coated emitting surface, and a GaSb PV cell with a water-based cooling system. Tungsten is used as a substrate due to its high thermal stability and good intrinsic reflectance properties for this application [5].
For efficient operation, the absorbing and emitting surfaces of an STPV system must meet the following requirements: a) sunlight must be efficiently absorbed by the absorbing surface, b) the emission of thermal radiation by the absorbing surface must be minimized, c) a large amount of power must be emitted by the emitting surface, and d) the emitting surface must emit primarily in a narrow band just above the Ebg of the PV cell.
To maximize the solar absorption (α) of the absorbing surface, a laser-textured blackbody with an anti-reflective dielectric coating was used. Since blackbodies have high thermal emission as well as high solar absorption, the surface area of the absorbing portion of the top surface must be minimized to lower emission loss. In a planar STPV system, the surface area of the top and bottom surfaces must be equal; however, laser texturing allows only a small portion of the top surface to be textured. This small textured area, combined with the naturally low thermal emittance of tungsten in untextured areas, resulted in a low overall thermal emittance for the top surface of the STPV system while maintaining a very high solar α. Additionally, an infrared (IR) reflecting mirror may be placed over the untextured portions of the top surface to further reduce thermal loss.
The emitting surface consisted of a single dielectric layer on the tungsten substrate of a thickness to minimize reflectance around a wavelength of 1400 nm. This simple structure maintains the high efficiency of more complex structures while being easy to fabricate and having a high thermal stability. Emission is spectrally matched via controlling the thickness of the dielectric layer on the tungsten. GaSb PV cells are used in this design due to their having high efficiency compared to cells with similar bandgaps, and the fact that their Ebg corresponds well with STPV system temperatures.
The STPV device was modeled as a thermodynamic system that can be broken down into a series of components with individual efficiencies that are multiplied together to obtain an overall efficiency [9]. While each loss is affected by the system temperature, losses can still be separated and examined as individual functions of temperature, not directly related to other losses. Incoming solar energy is simulated with an air-mass 1.5 (AM 1.5) solar spectrum with no diffuse energy component since it is not captured by solar concentrators [10].
The temperature of the STPV system rises until thermal equilibrium is reached, with the assumption that temperature is constant at every point on the absorbing/emitting structure. This is accurate for the thin planar systems examined here but may be less accurate for other system designs. Energy leaves this structure via heat conducted through the mechanical supports, convection, thermal energy radiated out of the top and sides, and thermal energy emitted from the emitter. The overall efficiency of the STPV system is given by the efficiency of the PV cell output compared to the input optical power into the system.
4. Experimental
Polished W substrates with dimensions 25×25×1 mm were purchased from the MTI Corporation and cleaned with acetone and then methanol. On one substrate, the top surface, a 0.6 cm2 area, was laser-textured to reduce reflectance, while on another substrate the entire top surface was microtextured. A nanosecond 1064 nm laser (IPG Photonics, GLP-10) operating with a pulse width of 50 ns, focused to a spot with an average power of 24 W and 30 kHz frequency, was scanned across a 6.5 cm2 tungsten substrate at a rate of 50 mm/s by means of a Galvo.
The top and bottom surfaces were then coated by a 160 nm thick Si3N4 protective layer. This layer was deposited by plasma-enhanced chemical vapor deposition (PECVD) at a temperature of 100°C. The diffuse reflectance of the laser-microtextured absorbing surface was measured using lasers with 405 nm, 532 nm, 633 nm, 790 nm, 980 nm, and 1064 nm wavelengths and a Labsphere RTC-060-SF integrating sphere with a Spectraflect barium sulfate reflecting surface. The reflection of the Si3N4 coated flat tungsten substrate was measured by a PerkinElmer Lambda 950 UV/VIS spectrometer.
Four square GaSb PV cells were purchased from JX Crystals, with an active area of 1.48 cm2 each. These cells have an Ebg of 0.67 eV, fill factor (FF) of 0.61, and an external quantum efficiency (EQE) of 0.72 at a wavelength of 1500 nm. The cells were soldered to copper substrates via reflow soldering and connected together in series.
To simulate solar radiation under laboratory conditions, a 300 W continuous wave, 808 nm laser diode stack was purchased from Sino-Laser, Inc. Although this laser emits radiation at a single wavelength, the absorbing surface used in this case is a blackbody across the visible and near IR spectra, making the wavelength of radiation irrelevant. The laser was focused to a spot size of 0.6 cm2 to match the textured area on the absorbing surface, and the relative solar concentration was calculated assuming an incident solar power of and a concentration factor ranging from 1 to 2500. 2500 was chosen as the upper concentration limit due to its being achievable with Fresnel lens systems, and to avoid unrealistic solar tracking requirements [11]. The half cone-angle of the incoming laser radiation was 14.5 degrees. Laser power was measured using a power meter to find the power of a portion of the beam reflected off a glass reflector.
The PV cells were placed inside a vacuum chamber on a water-cooled copper heat sink, with Arctic Silver 5 thermal compound used to reduce thermal resistance between the substrate and heat sink and prevent air pockets from forming. A thin aluminum cover plate was then placed over the heat sink and PV cell, with a hole cut out to allow radiation to reach the cell. The hole cut in the plate was slightly larger than the total PV cell area so as to not affect the view-factor between the emitter and cell. This plate prevented direct laser radiation from reaching the heat sink and PV cell frame, and reflected back some emitted radiation that would not reach the PV cell. The view-factor between the emitter and PV cells was 86%. A glass slide was placed over this window to prevent convective heating of the PV cells by the substrate and to reduce the amount of long wavelength IR radiation that reached the cell. The slide was purchased from Corning, Inc. and made from soda-lime glass. While this glass is transparent in the visible and near IR regime, it is reflective to wavelengths longer than 2500 nm. Next, the W substrate was suspended above the cover plate on two 0.5 mm diameter fused silica rods. Thin fused silica rods were used due to their high melting point, low thermal conductivity, and optical clarity. Lastly, a reflective plate was positioned over the absorbing surface of the substrate with a small hole cut out of it over the textured portion of the absorbing surface to reflect back thermal radiation from the untextured portion of the top surface. This allowed a small absorbing surface area to be used without incurring large efficiency losses from thermal emission from the inactive area. This assembly was kept under a vacuum of 10 mTorr during the experiment. Figure 2(a) shows a cross-sectional drawing of this device and Fig. 2(b) shows a top-down drawing of the setup, while Fig. 3(a) shows the flow of power through this system.
Fig. 3 a) The flow of power through an STPV system, showing each component and the power remaining after each loss. Total efficiency (ηtotal) is given by Isc × Voc × FF divided by the input power, ϕinput. Losses are quantified and described in Table 1. b) measured and simulated reflectance of an emitting surface made of a 160 nm thick Si3N4 coating on a W substrate.
Temperature measurements were taken with a type R thermocouple purchased from Nordic Sensors, Inc. This thermocouple was bonded to the bottom surface with a high-temperature thermally-conductive cement purchased from Sigma-Aldrich. The location of the thermocouple was 1 cm in from the edge of the tungsten substrate. The sample was kept under vacuum, and thermal conduction away from the substrate was minimized via the use of fused silica rods as supporting structures. The temperature gradient was not measured; however, we do not expect a significant gradient across this sample. Temperature measurements were reliably taken at temperatures up to 1200°C. The open circuit voltage (Voc) and short circuit current (Isc) of the PV cells were recorded with a multimeter. The I-V curve of a cell was measured under a Xenon arc lamp source at 1 sun, using a voltage source and multimeter, and is pictured in Fig. 4. The fill factor (FF) of this cell was 0.61, and the other three cells have measured FFs of 0.61, 0.59, and 0.62. The FF of the cells under thermal illumination was not measured. The overall laser light to electrical power conversion efficiency was calculated using the formulae efficiency η = Isc ×Voc FF/Iinc. The radiation from the emitter was incident on GaSb solar cell Isc and Voc were measured using a multimeter, the fill factor was used as measured with the lamp (0.61) and the incident power was based on the value of laser power incident on absorber.
Fig. 4 The measured current vs. voltage curve of the GaSb cell used in this experiment. A FF of 0.61 was measured.
5. Results and conclusions
The α of the microtextured and coated portion of the absorbing surface was measured using lasers with 405 nm, 532 nm, 633 nm, 790 nm, 980 nm, and 1064 nm wavelengths with a Labsphere RTC-060-SF integrating sphere with a Spectraflect barium sulfate reflecting surface. The sample was angled so that direct reflection would be captured by the sphere, which allowed for measurement of total (direct + scattered) reflectivity. It was found that the reflectance of the surface remained invariant across the wavelengths measured. An absorbance of 92% was found before annealing the surface at 1700 K, and an absorbance of 90% was found after annealing. The reflectivity is essentially insensitive to incidence angle as indicated in our previous published research [12].
Figure 5(a) shows an SEM of this surface before annealing, while Fig. 5(b) shows an image of the surface after annealing. Very little change can be seen between these samples, showing the stability of the surface. The emittance of the emitting surface was measured and simulated via the Fresnel method, and can be seen in Fig. 3(b). This surface shows good spectral selectivity with increased emittance at photon energies just higher than the Ebg of the PV cell.
Fig. 5 a) SEM of the absorbing surface texture before heating and b) SEM of the absorbing surface after heating to 1700 K.
The temperature and efficiency of the STPV system were recorded at various laser power levels. These measurements were then compared to simulated temperature and efficiency values. Figure 3(a) shows the flow of power through this system. Good matching between simulated and experimental results was found, showing that the basic premise of the simulation is sound. The measured and simulated temperature and efficiency of the system under these conditions is shown in Figs. 6 a and b. The temperature of the PV cell reached a maximum of 72°C during the experiment. A maximum efficiency of 6.2% was measured at a laser power of 149 W, corresponding to a solar concentration factor of 2483, and simulations show that this efficiency would rise to 7% if the PV cell was kept at room temperature during this measurement [13].
Fig. 6 Comparison of simulated and experimental values for a) system temperature and b) system efficiency with concentration factor.
Table 1 shows the breakdown of how much energy is lost to each factor shown in Fig. 3(a). This demonstrates that the quality of the absorbing and emitting surfaces is paramount to the operation of an efficient STPV system. It also shows that PV cell losses play a large role, and that the use of more advanced GaSb PV cells could greatly improve STPV system efficiency. Simulations show that improving the Voc and FF from the values found on cell testing in this paper to values found in literature (an increase from 413 mV to 586 mV for Voc and from 0.61 to 0.82 for FF) would increase system efficiency to 11.8% [13]. Additionally, the use of a hot mirror with a 1.7 μm cutoff wavelength between the emitting surface and PV cell could both improve system efficiency by returning longer wavelength radiation to the emitting surface and help to keep the PV cell cooler, which in turn would increase the PV cell efficiency. The increase in efficiency due to the reflection of long wavelength radiation back to the emitter could result in system efficiencies up to 20.2%; however this figure assumes a perfect heat mirror which would be very difficult to achieve in practice.
Table 1. List of losses in the simulated STPV system using a tungsten substrate at a concentration ratio of 2500 (losses shown in Fig. 3(a), where FPV is the view-factor between the emitting surface and PV cell, and Einc is the optical energy available to the absorbing surface.
In conclusion, this paper demonstrates a simple way to fabricate an STPV system utilizing nanostructures that operates at high efficiencies. The system is studied under simulated solar conditions, and an experimental efficiency of 6.2% was recorded. This is the highest efficiency recorded for an STPV system using nanostructures, and this can be further improved. For example, our theoretical modeling results match the experimental results well and the losses in the system could be analyzed through this model. A path forward for increased STPV system efficiency is found by analyzing the simulated losses as well.
Acknowledgments
We would like to thank the NASA Langley Professor and National Science Foundation IUCRC programs for their support of this project. This work was performed, in part, at the Center for Nanoscale Materials, a U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences User Facility under Contract No. DE-AC02-06CH11357.
References and links
1. W. Shockley and H. J. Queisser, “Detailed balance limit of efficiency of p-n junction solar cells,” J. Appl. Phys. 32, 510–519 (1961). [CrossRef]
2. D. L. Chubb, “Fundamentals of Thermophotovoltaic Energy Conversion” (Elsevier, 2007).
3. M. A. Green, K. Emery, Y. Hishikawa, W. Warta, and E. D. Dunlop, “Solar cell efficiency tables (Version 45),” Progress in Photovoltaics: Research and Applications 23, 1–9 (2014). [CrossRef]
4. H. Yugami, H. Sai, K. Nakamura, H. Nakagawa, and H. Ohtsubo, “Solar thermophotovoltaic using Al2O3/Er3Al5O12 eutectic composite selective emitter,” IEEE Photovoltaic Specialists Conference28, 1214–1217 (2000).
5. A. Datas and C. Algora, “Development and experimental evaluation of a complete solar thermophotovoltaic system,” Progress in Photovoltaics: Research and Applications 890, 327–334 (2012).
6. A. S. Vlasov, V. P. Khvostikov, O. A. Khvostikova, P. Y. Gazaryan, S. V. Sorokina, and V. M. Andreev, “TPV systems with solar powered tungsten emitters,” AIP Conference Proceedings890, 327–334 (2007).
7. A. Lenert, D. M. Bierman, Y. Nam, W. R. Chan, I. Celanovic, M. Soljacic, and E. N. Wang, “A nanophotonic solar thermophotovoltaic device,” Nature Nanotech. 9, 126–130 (2014). [CrossRef]
8. M. Shimizu, A. Kohiyama, and H. Yugami, “High-efficiency solar thermophotovoltaic system equipped with a monolithic planar selective absorber/emitter,” J. Photon. Energy 5, 053099 (2015). [CrossRef]
9. Y. Nam, Y. X. Yeng, A. Lenert, P. Bermel, I. Celanovic, M. Soljacic, and E. N. Wang, “Solar thermophotovoltaic energy conversion systems with two-dimensional tantalum photonic crystal absorbers and emitters,” Sol. Energ. Mat. Sol. Cells 122, 287–296 (2014). [CrossRef]
10. “Reference solar spectral irradiance: ASTM g-173,” http://rredc.nrel.gov/solar/spectra/am1.5/ (2012). Accessed: 2015-06-10.
11. K. Ryu, J.-G. Rhee, K.-M. Park, and J. Kim, “Concept and design of modular Fresnel lenses for concentration solar PV system,” Solar Energy 80, 1580–1587 (2006). [CrossRef]
12. V. V. Iyengar, B. K. Nayak, and M. C. Gupta, “Ultralow reflectance metal surfaces by ultrafast laser texturing,” Appl. Opt. 49, 5983 (2010). [CrossRef]
13. L. G. Ferguson and L. M. Fraas, “Theoretical study of GaSb PV cell efficiency as a function of temperature,” Sol. Energ. Mat. Sol. Cells 39, 11–18 (1995). [CrossRef]
