Source-linked AI summary
Thermal Energy Grid Storage Using Multijunction Photovoltaics
Caleb Amy, Hamid Reza Seyf, Myles A. Steiner, Daniel J. Friedman, Asegun Henry
TL;DR
Containing liquid silicon in tanks is a central realization challenge for TEGS-MPV. The paper proposes storing heat in pumped liquid and converting emitter light with multijunction photovoltaics; it estimates that low costs and RTE above 50% may be achievable.
Problem
A central realization challenge is containing liquid silicon in tanks, because material choices at these temperatures are severely limited.
Source: A New System Concept · p. 8Method
The design stores heat in pumped liquid and converts emitter light with multijunction PV, reflecting sub-bandgap photons back to the thermal emitter.
Source: A New System Concept · p. 7, A B E · pp. 10–11Results
The authors judge TEGS-MPV potentially capable of CPE < $50/kWh-e, CPP < $0.5W-e, and RTE > 50%.
Source: Conclusions · p. 14Takeaways & Limitations
Some TEGS embodiments may exceed the ~36% RTE lower bound identified for arbitrage while achieving extremely low CPE and CPP.
Source: Conclusions · p. 13
Abstract
from arXiv · showhide
As the cost of renewable energy falls below fossil fuels, the most important challenge to enable widespread sustainable power generation has become making renewables dispatchable. Low cost energy storage can provide this dispatchability, but there is no clear technology that can meet the need. Pumped hydroelectric and compressed air storage have low costs, but they are geographically constrained. Similarly, lithium-ion batteries are becoming ubiquitous, but even their lower bounding asymptote cost is too high to enable cost-competitive dispatchable renewables. Here, we introduce a concept based on thermal energy grid storage (TEGS) using a multijunction photovoltaic heat engine (MPV) with promising initial experimental results that could meet the low cost required to enable cost competitive dispatchable renewables. The approach exploits an important tradeoff between the accession of an extremely low cost per unit energy stored, by storing heat instead of electricity directly, while paying the penalty of a lower round trip efficiency. To understand why this tradeoff is advantageous, we first introduce a framework for evaluating storage technologies that treats round trip efficiency (RTE) as a variable, in addition to cost per unit energy stored (CPE) and cost per unit power (CPP). It is from this perspective that the TEGS-MPV concept offers a compelling economic proposition.
A New System Concept
The concept stores electricity as heat in pumped liquid silicon and converts the heat back to electricity with multijunction photovoltaic cells. Initial tank experiments and the proposed system design address key containment and operating challenges.
A New System Concept
- The proposed cycle stores excess electricity as sensible heat in liquid silicon across 1900–2400°C, then converts emitter light back to electricity with MPV cells.The design uses a tungsten-foil emitter and returns cooled silicon to the cold tank.Source: A New System Concept · pp. 5–6
- The 1900–2400°C regime seeks higher RTE and suits higher-bandgap, multijunction cells with a back-surface reflector, though 2400°C nears practical refractory limits.Higher temperatures support high-bandgap operation and power density, while vapor pressure can deposit emitter material on cells.Source: A New System Concept · pp. 6–7
- A flowing liquid can support steady-state power and efficiency, whereas solid storage can develop conductive resistance and declining output during discharge.The liquid can be pumped to maintain rated power and RTE at peak output.Source: A New System Concept · p. 7
- MPV is favored over turbines because external-heat turbines above 1000°C do not exist, MPV may cost less, and modules can actuate in seconds.The authors argue that faster load following could increase grid value and revenue.Source: A New System Concept · p. 8
- A 10–30 m graphite tank cannot be made as one monolithic piece, so the design requires impermeable sections joined by leak-tight seals.The miniature silicon-containment experiment provides initial evidence for this modular-tank approach.Source: A New System Concept · p. 8
- Above 2,000°C for 60 minutes, a sealed graphite miniature tank contained 553-grade silicon and formed a dense 20-µm SiC interface layer.The experiment showed silicon penetration of approximately 400 µm before the protective layer stopped further penetration.Source: A New System Concept · p. 8
A B E
The analysis models multijunction PV cells and system costs for TEGS-MPV, while addressing pumping, optical, and electrical losses. It estimates that two-junction cells could enable RTE above 50%.
A B E
- Pumping liquid silicon at 2400°C has not been demonstrated, but the authors cite ceramic pumping at 1400°C and graphite strength increasing up to 2600°C.They conclude these precedents leave no obvious issue that should prevent operation at the higher temperature.Source: A B E · p. 9
- Existing power electronics can supply the heater inputs at less than 1% parasitic loss, shifting attention to losses in the MPV power block.The authors characterize the heater loss as small and almost negligible.Source: A B E · p. 9
- Two junction bandgaps are chosen to convert separate parts of the thermal spectrum and mitigate thermalization loss from a single bandgap mismatch.Thermalization loss arises because the emitted spectrum spans photon energies that no one junction bandgap matches.Source: A B E · p. 10
- For III-V cells, the roughly 0.3–0.4 eV voltage penalty is proportionally smaller at higher bandgaps, which suit the high-energy photons from hot emitters.The passage attributes much of the voltage loss to entropy increase and notes additional non-radiative recombination penalties for silicon.Source: A B E · p. 10
- At least 98% reflectivity for below-bandgap photons is needed to return unusable light to the emitter and realize the predicted high performance.The remaining absorption is treated as parasitic loss at the reflector or semiconductor.Source: A B E · p. 11
- More than 50% RTE is possible with dual-junction MPV; modeled practical efficiencies exceed 40% for one-junction cells and 50% for two-junction cells.For a 2100°C emitter, the reported optimal bandgaps are roughly 1.0–1.2 eV for one junction and {1.2, 1.0} to {1.4, 1.2} eV for two.Source: A B E · p. 12
- The nominal cost estimate covers a 100 MW-e system with 10 hours of storage; the lower-cost case assumes scrap steel and near-silicon-cell costs.CPE covers storage and construction costs, while CPP includes the heater, MPV cells, inverter, emitter, insulation, and cooling system.Source: A B E · p. 12, Figure 5A · p. 13
Conclusions
The analysis identifies a tradeoff among round-trip efficiency (RTE), cost per unit energy (CPE), and cost per unit power (CPP), and presents TEGS-MPV as a potentially attractive grid-storage approach. Initial experiments support feasibility, though a prototype is still needed to confirm system reliability.
Conclusions
- The analysis identifies an RTE lower bound of about 36%, below which storage may be unlikely to earn arbitrage revenue.The conclusion frames this estimate in relation to previous work by Denholm [4].Source: Conclusions · p. 13
- The proposed system uses liquid ultra-high-temperature storage and MPV conversion, with potential for load following and RTE above the estimated lower bound.The conclusion presents these as potential benefits of the new embodiment, not as demonstrated full-system performance.Source: Conclusions · p. 13
- A prototype is needed to verify reliable storage-medium pumping, leak-free flow-loop operation, and reliable MPV cell fabrication and testing under the described conditions.The conclusion identifies these as the most notable remaining practical challenges.Source: Conclusions · p. 13
- CPE < $50/kWh-e, CPP < $0.5/W-e, and RTE > 50% appear feasible for TEGS-MPV, though realizing them remains challenging.The conclusion says the cost and performance targets seem feasible, but calls for a prototype to confirm reliable pumping, leak-free operation, and reliable MPV cell fabrication and testing.Source: Conclusions · p. 14
- Experiments support sealing liquid silicon with grafoil gaskets and achieving the assumed 2% parasitic absorption with gold- or silver-backed MPV cells.Reflectivity measurements confirmed that the assumed below-band-gap parasitic absorption is possible.Source: Conclusions · p. 14
Methods:
The value model estimates storage economics under current market conditions using capacity payments and arbitrage as revenue streams. It defines arbitrage value and estimates capacity payments using Net CONE.
Methods:
- The model evaluates storage under current market conditions using capacity payments and arbitrage as the two revenue streams unlikely to be saturated.Arbitrage value is defined as annual device revenue minus the cost of buying energy at off-peak times.Source: Methods · p. 17
- Capacity payments are estimated from Net CONE, the net loss a balancing resource would need compensated to reach zero NPV.The estimate uses the 2018 average Net CONE in the PJM market: $95/kW-yr.Source: Methods · p. 17
Multijunction Photovoltaics (MPV)
This section examines a two-junction MPV design and the thermal-storage considerations for TEGS-MPV, including modeled performance factors and discharge challenges associated with solid sensible-heat storage. It presents liquid storage as the preferred option for high-efficiency operation.
Multijunction Photovoltaics (MPV)
- At the envisioned several hundred kW/m2 flux, cell current density exceeds 10 A/cm2, so front-contact grids must balance reduced series-resistance loss against increased shadowing.Blocked photons can be reflected back to the emitter for reuse; module losses may also be mitigated with more small cells in series for high-voltage, low-current operation.Source: Multijunction Photovoltaics (MPV) · p. 17
- A thin two-junction MPV with 1.2-eV Al0.15Ga0.55In0.3As and 1.0-eV Ga0.7In0.3As, a removed compositionally graded buffer, and a high-reflectivity BSR could achieve >50% RTE.The design grows the top junction first, then the bottom junction, and removes the substrate after bonding the device to a handle.Source: Manufacturing Approach · p. 18, Figure 1A · p. 18
Radiative Heat Transfer in Optical Cavity
- For Eg =1.43 eV, Aemitter/Acell =4 raises above-bandgap incident power 1.6-fold over the equal-area baseline; convective heat loss is <5 kW/m2 of cell area.The analysis also finds that increasing emitter-to-cell area ratio can trap hot gas between fins, reducing convective heat loss.Source: Radiative Heat Transfer in Optical… · p. 19, Figure 2 · p. 20, Figure 3 · p. 22
- Longer tungsten fins lower emitter heat loss by trapping hot, low-velocity gas and thickening the thermal boundary layer; smaller MPV cells also reduce heat loss by increasing cell-emitter distance.At small cell-emitter distances, conduction dominates and natural-convection circulation has little effect.Source: Convective Heat loss · p. 22, Figure 3 · p. 22
Multijunction Photovoltaics (MPV)
- A 100 MW plant with 10 hours of storage would require 10^7 kg of graphite, a roughly 18 m by 18 m tank, and about 1,000 m2 of MPV surface area.At the assumed 100 kW/m2 MPV power density, the graphite would need divisions for cell slots, while the characteristic conduction length would likely exceed 1 ft (~0.3 m).Source: Convective Heat loss · p. 29
- For graphite, the reported temperature decrease exceeds 40°C/sec throughout the first two minutes of discharge; the authors say this transient would cause cracking and mechanical failure within 100 cycles.The 100-cycle estimate is based on the authors’ previous unpublished experiments thermally cycling and cracking graphite with 100°C/min heating rates.Source: Convective Heat loss · pp. 29–30
- A ~3300 K/m surface gradient implies about a 1000°C center-to-surface temperature difference across a 1 ft solid, reducing efficiency relative to predicted 1900–2400°C emitter temperatures.By contrast, the passage estimates only a 15–20°C drop across a 5 mm pipe wall; transient conductive resistance in a solid is described as fundamental to discharge.Source: Convective Heat loss · p. 30
- Because solid storage’s discharge thermal-management problems are likely insurmountable, the authors identify liquid storage as highly preferred and likely necessary for sufficiently high efficiency.The passage links that efficiency to enabling eventual 100% penetration of renewables.Source: Convective Heat loss · p. 30