Spherical sintered bauxite particles between 200 μm and 700 μm in diameter have been shown to be effective in the direct absorption and storage of concentrated solar energy. These particles are commercially available in large quantities and exhibit as-received solar weighted absorptance (αs) greater than 0.90, which gradually degrades with extended heating in air at 700 °C and above. The degradation mechanism is an oxidation reaction that can be reversed via thermal or chemical reduction, resulting in αs > 0.95 along with enhanced resistance to further degradation for some formulations. Certain metal oxide pigments, added to Al2O3:SiO2, have proven to achieve solar weighted absorptance levels similar to those of the commercially available particles and may be promising alternatives to currently available materials.
The falling particle receiver concept was originally conceived in the 1980 s as a means of increasing the operating temperature of solar central receivers beyond the ∼600 °C limit of molten salt-based concepts . In operation, a thin sheet of opaque, highly absorptive ceramic particles, with a sheet thickness of 10–60 particle diameters and a particle diameter of order 102 μm, is cascaded through a beam of concentrated solar energy. The energy is directly absorbed by the particles, heating them rapidly to a temperature beyond the limits of conventional salt-based receivers [2,3]. Higher temperature operation enables the deployment of more efficient power systems, including those based on supercritical steam Rankine and supercritical CO2 Brayton cycles. The falling particle receiver has additional advantages relative to current technology, including: (1) the ability to operate at a greater solar concentration ratio, which decreases thermal losses and (2) the use of chemically inert particles that do not corrode system components, which is a significant challenge in salt-based systems above 550 °C.
Early work on falling particle receivers identified ceramic proppants, spherical sintered bauxite particles used in oil and gas production operations, as the best candidates for use in a direct absorption falling particle receiver . The most important physical properties were identified to be mechanical durability and cost, discussed by Falcone et al. , and solar weighted absorptance was measured for many commercially available particle formulations by Hellmann and McConnell . Further work aimed at assessing sintering at high temperature revealed that some of the proppant formulations under consideration changed in color from relatively dark black, in the as-received condition, to orange upon heating in air above 900 °C for up to 150 h, with most of the change happening within the first 15 h of heating . This color change corresponded to an undesirable decrease in the solar weighted absorptance. While X-ray diffraction analysis was conducted to examine changes in the structure of the materials upon heating , the rate of degradation below 900 °C was not quantified and no mitigation strategies were developed that would allow the solar weighted absorptance of the particles to be maintained at a sufficiently high level over the course of several years, as would be required in a particle receiver-based power plant.
It is likely that future falling particle receivers will operate between 650 °C and 800 °C, a temperature range that can accommodate advanced power cycles based on steam or carbon dioxide working fluids, but not high enough that solar collection losses become prohibitive. In fact, a prototype falling particle receiver, designed to operate at a particle outlet temperature of 700 °C, is currently under construction at Sandia National Laboratories (SNL) with on-sun testing planned for 2015 . The goal of this test, and the larger Department of Energy (DOE)/SunShot-funded project, is to establish the technical feasibility of the falling particle receiver for operation at 700 °C.
In this paper, we present an evaluation of the relevant physical properties, primarily radiative but also composition and specific heat capacity, of several commercially available particle materials for potential use in a falling particle receiver operating at 700 °C and including up to 15 h of storage capacity. We also show how the degradation of solar weighted absorptance due to oxidation with air, a known engineering challenge and one that must be addressed prior to the commercial development of falling particle receiver technology, can be mitigated through the chemical reduction of the particles with forming gas. Finally, we present radiative property data for several new material formulations that are chemically simpler than currently available proppants and may be more stable.
Theoretical and Experimental Background
Radiative Transport Simulation in the Particle Curtain.
Radiative transport within the falling particle receiver has been simulated by several researchers, all of whom used the discrete ordinates formulation of the radiative transport equation to model the multiphase particle-air flow within the receiver. Khalsa et al. compared simulated receiver performance and measured data from on-sun testing at SNL. Their models show good agreement with experimental data when treating the particles as spectrally gray with an emissivity of 0.93 and a particle scattering factor of 4.2 cm−1 . Additional unpublished studies using their models have shown that changing the scattering factor to zero has a negligible effect on the results, indicating that scattering for these highly absorbing particles is of relatively little importance to radiative transport within a curtain and the resulting particle energy gain, under falling particle receiver conditions. In the models developed by Houf and Grief  and Evans et al. , the relative contribution of scattering and absorption to total extinction within the curtain was expressed using the single scattering albedo; the ratio of the scattering coefficient to the sum of the scattering and absorption coefficients. Houf and Grief used data presented by Griffin and Stahl , which was developed through a combination of experiments to separately evaluate particle scattering and absorption coefficients. They found that highly absorbing, opaque particles similar to those that we discuss in this current study have a measured single scattering albedo of less than 0.10. They also showed that the single scattering albedo, which has a spectral dependence, is approximately equal to the measured spectral hemispherical reflectance of a packed bed of particles. A complementary study by Stahl et al. presented measured packed bed reflectance data for candidate particles over a spectral range of 300–2500 nm and for bed temperature between room temperature and 1000 °C . Their results show that room temperature data sufficiently represent spectral hemispherical reflectance over the spectral range of interest for solar energy absorption.
The radiative property data that we present in this work are limited to: (1) spectral hemispherical absorptance on packed particle beds and the equivalent calculated solar weighted hemispherical absorptance ( and (2) hemispherical emittance ( weighted based thermal emission at 700 °C. While scattering does not play a significant role in radiative transport within a falling particle receiver for media of the type considered in this study, other materials are under investigation, which are semitransparent, including silica and olivine sands . In the case of semitransparent media, scattering can be a significant component of extinction and must be considered [15,16].
Our experimental work includes three elements: (1) characterization of the physical properties and degradation mechanisms of prospective absorber materials, (2) chemical modification of absorbers to enhance solar weighted absorptance, and (3) synthesis of novel absorber formulations.
Specific heat capacity as a function of temperature was measured with a Netzsch STA 409 (differential scanning calorimeter). 50 mg samples were placed in a Pt crucible and heated at a rate of 10 °C/min in flowing air (120 ml/min) from room temperature to 1200 °C. The uncertainty of the measured data is ±5% based on the instrument calibration.
The spectral hemispherical reflectance of packed particle beds at room temperature was measured using a Perkin-Elmer Lambda 950 UV–Vis–NIR spectrophotometer with an integrating sphere. Diffuse spectral reflectance was collected at 10 nm increments between 300 and 2500 nm. The measurements were calibrated with a NIST-traceable reflectance standard from Spectralon having a measurement uncertainty of ±0.005 relative reflectance units over the range measured. The spectral data were then weighted with the AM1.5 spectrum (ASTM G173-03) to calculate solar weighted absorptance. Thermal emittance was calculated from room temperature reflectance data taken with a surface optics ET100 Emissometer, which measures spectral reflectance at six wavelength bands between 1.9 μm and 25 μm with an uncertainty of ±0.03 reflectance units. These data were further processed to calculate the total hemispherical emittance at 700 °C .
The chemical reduction of particle materials as a means of increasing solar weighted absorptance was accomplished by placing unmodified particles in an alumina combustion boat and heating in a tube furnace at a temperature of either 1400 °C or 700 °C under 100 ml/min of (5%H2:95%N2) forming gas for 5 h.
Particle stability in an oxidizing environment at high temperature was assessed using thermogravimetric analysis (TGA). In these tests, a ∼30 mg sample of pulverized CARBOHSP (as-received) was placed in an alumina crucible and heated in a TA Instruments SDT Q600 TGA at 10 °C/min to 1400 °C under air flowing at 100 ml/min. Following oxidation in air, the heating process was repeated with the sample exposed to forming gas (5%H2:95%N2). A second, final oxidation in air was then performed with the same heating schedule.
Novel ceramic absorbers based on a composition similar to that of commercial proppants (primarily Al2O3:SiO2) were synthesized by adding an absorbing pigment to Al2O3 and SiO2 powders. To produce pellets for property characterization, candidate pigments were first mixed with Al2O3 and SiO2 powders in a 250 ml jar mill along with 150 g of YSZ grinding media (2.5 mm spheres) and 50 ml of ethanol to produce a slurry. The powders were milled for 24 h after which the contents were dried over flowing air at room temperature. Once dry, the powders were combined with a polyvinyl butyral (PVB) binder solution (2 wt.% PVB in acetone), with binder solution added until the powder was completely wetted. The slurry was then mixed with a mortar and pestle and allowed to dry. Once dry, approximately 4 g of the powder/binder mixture was pressed in a 2.5 cm diameter pellet die and pressed under a load of 10 tons. The pressed pellets were subsequently sintered in a furnace in air using the following heating profile: ramp at 5 °C/min to 1350 °C, hold for 36 h, ramp at 5 °C/min to 1450 °C, hold for 4 h, cool to room temperature at 5 °C/min.
Results and Discussion
Characterization of Commercially Available Materials.
We evaluated a range of commercially available particle media, all supplied by CARBO Ceramics and used industrially as either proppants for oil and gas production or as a casting media for foundry applications. The mean diameter, bulk density, and composition of the particles as reported by CARBO  are summarized in Table 1. The composition of CARBOHSP was measured by CARBO at our request using inductively coupled plasma mass spectrometry (ICP-MS), and the values that we report in Table 1 differ slightly from those published by CARBO. In addition, the composition of CARBOBLACK is proprietary, but is likely similar to CARBOPROP.
The commercial particles are all similar in composition, varying primarily in the fractions of alumina, silica, and iron oxide content. Generally, proppants containing more alumina are mechanically stronger, and those with a higher iron oxide fraction are darker in color . In addition to the major components listed in Table 1, CARBOHSP was found to contain minor components (less than 0.05%) of K2O, CaO, MgO, Na2O, MnO, and P2O5 by ICP-MS. The cost of these materials, in quantities greater than 18,000 kg, ranges from ∼$0.6/kg to $1.1/kg with particles having less alumina content (e.g., less mechanical strength) generally being the least expensive . All of the materials are available in a range of diameters between 100 μm and 1000 μm.
The three thermophysical properties of interest for falling particle receiver heat transfer simulations are density, specific heat capacity, and thermal conductivity. Of these, only specific heat capacity was measured directly in this study, with density data available from the manufacturer and thermal conductivity generally assumed to be approximately equal to that of sintered alumina or mullite, in the range of 2.0–6.0 W/m K. The heat capacity data for three CARBO materials, as a function of temperature, are shown in Fig. 1.
The average value of heat capacity relevant to prototype and near-term commercial falling particle receivers is for CARBOHSP and for both CARBOPROP and ACCUCAST, calculated by numerically integrating the data in Fig. 1 between 300 °C and 700 °C.
The optical properties needed to simulate radiative transport within a falling particle receiver include solar weighted absorptance and thermal emittance at a temperature of 700 °C. As this current work is primarily concerned with assessing and improving the stability of solar weighted absorptance during long term exposure to oxidizing conditions at high temperature, we first measured the optical properties of the particles in the as-received condition. These data are summarized in Table 2.
Both the solar weighted absorptance and the thermal emittance are important with regard to the net absorption efficiency of concentrated solar energy. However, the current designs for falling particle receiver concepts are all cavity-based and operate between 300 °C and 700 °C. Supporting computational analyses have shown that the receiver efficiency depends primarily on solar weighted absorptance. Nevertheless, we measured thermal emittance to identify any materials that may be solar-selective as this quality would be advantageous if the operating temperature of the receiver was increased.
Degradation of Solar Weighted Absorptance.
Hellmann showed that the color of sintered bauxite proppants changes significantly over 150 h when heated in air above 900 °C and that the change is associated with reactions involving the species FeAlTiO5, Fe2TiO5, and Al2TiO5 . Qualitatively, the color changed from dark brown to light brown and orange, which we illustrate in Fig. 2 with CARBOHSP heated in air to 1000 °C for 192 h. Our evaluation of several particle formulations at 700 °C shows similar behavior with respect to color change, albeit less pronounced, which we quantified as a change in solar weighted absorptance. We have also reported XRD spectra for CARBOHSP in a previous study in which we identified the formation of the same species as that found in Hellmann's work at 700 °C .
To assess the degradation of solar weighted absorptance, all of our candidate materials were heated in air at 700 °C for at least 500 h with measurements taken at intermediate points (Fig. 3).
Figure 3(a) shows that all of the commercial particle formulations degrade, but that both CARBOHSP and ACCUCAST converge to αs = 0.92 and αs = 0.90, respectively, after 500 h. The change in αs is most rapid in the first 192 hr, in most cases, with the spectral absorptance decreasing to a relatively greater extent in the near infrared region (Fig. 3(b)) and not strongly manifested as a visible color change. This is somewhat a different behavior than the result obtained when the particles are heated in air above 900 °C.
Subsequent TGA was performed to further elucidate the mechanisms responsible for the degradation of solar weighted absorptance. The results of these studies are shown in Fig. 4. The first oxidation of the as-received material is shown in Fig. 4(a), and the second oxidation process following chemical reduction in forming gas is shown in Fig. 4(b).
Examination of the sample mass rate change as a function of temperature allows the identification of two distinct reactions, centered near 600 °C and 1100 °C. These reactions are more pronounced in Fig. 4(b) than in Fig. 4(a), owing to the fact that the as-received material is more oxidized than the reduced material. Further testing showed that the 600 °C reaction causes significant color change, indicating that any use of CARBOHSP above 600 °C in air will result in color change and a concomitant decrease in solar weighted absorptance over time.
The results shown in Fig. 4 lead to two important conclusions. The first is that the degradation reactions involve oxidation; the reactions at 600 °C and 1100 °C do not occur in the absence of oxygen. The second is that the particles can be chemically reduced and that, upon reduction, are darker in color (Fig. 2(b)) than the as-received materials. Moreover, the solar weighted absorptance of particles that have been substantially degraded by extended exposure to air at 700 °C can be restored after brief (e.g., several hours) exposure to forming gas at 700 °C and above. This finding opens up the possibility of using chemical reduction as a means to maintain the solar weighted absorptance of particles at a sufficiently high level for an extended period of time.
Regeneration of Optical Properties With Chemical Reduction.
Chemical reduction in forming gas improves the solar weighted absorptance of all of the commercially available proppant materials that we evaluated. Data showing the solar weighted absorptance of CARBOHSP and ACCUCAST following chemical reduction and subsequent heating in air at 700 °C are presented in Fig. 5(a), with spectral data for reduced CARBOHSP shown in Fig. 5(b).
A comparison of Figs. 3(a) and 5(a) shows that reduction at 1400 °C increases the solar weighted absorptance of CARBOHSP by 3% and by 2% for reduction at 700 °C. Similarly, of ACCUCAST reduced at 700 °C is increased by 3% relative to the unmodified material. ACCUCAST was also reduced at 1400 °C in a packed bed arrangement, but sintered strongly and could not be removed from the combustion boat. CARBOHSP is unique among the materials tested in that it alone did not sinter strongly at 1400 °C.
In addition to achieving a higher solar weighted absorptance, CARBOHSP reduced at 1400 °C is more stable with respect to the change in its solar weighted absorptance as a function of heating duration. Figure 5(a) shows no significant variation in for CARBOHSP reduced at 1400 °C after heating in air at 700 °C for 1500 h. Figure 4(b) shows that the spectral absorptance of this material did change, increasing above 1000 nm and decreasing between 500 and 1000 nm; however, these changes occurred in the first 1000 h after which the spectral properties can be considered to be stable within the uncertainty limits of the measurement.
Despite the relative stability demonstrated by reduced CARBOHSP over the 1500 h test period, it is possible that the properties of this material are changing, albeit slowly. When considering whether to use this material in a power plant, it is necessary to consider how might vary over an even longer period of time. Our assumption for the lifetime of a ceramic proppant in an operating particle receiver power plant is based on unpublished studies and ranges out to 10 yr, with attrition (e.g., a gradual reduction in the size of the particle) being the limiting factor. Bearing this in mind and considering that the uncertainty limits for the solar weighted absorptance measurements presented in Fig. 5 are ±0.003 reflectance units, it is possible that the absorptance of CARBOHSP reduced at 1400 °C decreases by as many as 0.006 reflectance units over the course of 1500 h. The equivalent annual degradation rate is 0.013 reflectance units per year.2 Assuming an initial absorptance of 0.956, and a minimum allowable absorptance of 0.900, of the particles could degrade below the minimum level after 4.3 yr. At this point, chemical reduction could be used to restore to the original value of 0.956, thus providing another 4.3 yr of operation at the target performance level. A similar analysis performed for particles reduced at 700 °C indicates that the regeneration period could be as short as 7 months.
Whether to regenerate the particles at 700 °C or 1400 °C depends on many factors including regeneration frequency, quantity of forming gas required per regeneration, and the cost of additional processing equipment. Reduction at 700 °C, while resulting in less stable properties, may be favored relative to higher temperature reduction as the processing can be done in situ while the particle resides in the “hot” thermal storage vessel, avoiding the possibly significant costs of additional equipment needed for higher temperature chemical reduction. We are currently assessing the technoeconomic factors associated with particle regeneration in the context of a commercial solar power plant as a means to identify the lowest cost approach.
Novel Material Formulations.
The ideal particle material has and does not require regeneration over its predicted 10 yr lifetime in the thermal storage system. It must also be inexpensive, which means that it should be primarily composed of relatively abundant materials. We have synthesized and characterized 140 unique particle material formulations that are based on major components of alumina (Al2O3) and silica (SiO2), the relatively low cost primary constituents of commercial proppants and inorganic pigments are known or suspected to be good intrinsic solar absorbers. All of our candidate materials were produced from physically mixed powders and then pressed and sintered to form dense, cylindrical pellets. Compositionally, the weight fraction of Al2O3 was varied between 40% and 85%, SiO2 between 10% and 40%, and the pigment between 5% and 20%. Table 3 contains a summary of the pigments included in our study.
Several of the pigments included in our study were selected based on property data presented by Bogaerts and Lampert  and by Levinson et al. . One pigment, MnCo2O4, was developed by SNL to serve as a high temperature selective absorber for central receiver applications . The pigments K38, K40, and K42 were purchased from Kremer Pigments in powder form. In all cases, data pertaining to the radiative properties of the combined pigment material and the Al2O3:SiO2 base material were unavailable.
The characterization of the radiative properties of the candidate materials was done using the same methods and equipment used for the commercial proppants. Thermal exposure testing was also conducted in the same manner, with the sintered pellets heated in air at 700 °C for up to 500 h. The measured values for and for the pigmented, sintered pellets following 500 h of heating in air at 700 °C are shown in Fig. 6.
A total of 12 materials maintain after 500 h and 24 materials maintain after 500 h. These 24 materials include formulations based on seven unique pigments. Formulations using the other ten pigments shown in Table 3 failed to achieve after 500 h. In comparing the results in Fig. 6 with the data for the proppants, it is important to consider that the relatively flat surface of the sintered pellets will reflect light that would have been more completely absorbed by the relatively rough surface of a packed bed of proppants . Put another way, the absorptance and emittance values presented in Fig. 6 are less than would be measured if the pellets were formed into small spherical particles (like proppants). Despite that difference, the absorptance data shown in Fig. 6 are comparable to the data for both unmodified proppants (Fig. 3) and reduced proppants (Fig. 5) with the exception of CARBOHSP reduced at 1400 °C, indicating that these new materials represent a viable alternative to proppants with regard to their use as direct solar absorbers.
The emittance data presented in Fig. 6(b) show little variation across the range of materials. The ratio of solar weighted absorptance to thermal emittance, , was calculated for all of the new materials as a means of assessing selectivity. The average value for all of the materials shown in Fig. 6 is 1.16 with a standard deviation of 0.05. Only four formulations, based on two pigments, demonstrated > 1.21, thus marking them as having a level of spectral selectivity above that achieved by most of the other materials.
The falling particle receiver is one of the few options currently under consideration for providing solar heat input, with thermal energy storage, to thermal power cycles operating at an energy input temperature of 700 °C and above. The performance of the receiver (i.e., receiver efficiency) depends in part on the radiative properties of the particles, with solar weighted absorptance being more important than thermal emittance. We have shown that the solar weighted absorptance of commercially available “proppants,” currently the leading candidates for direct absorption and storage media in falling particle receivers, degrades due to oxidation in air when held at a temperature above 700 °C; however, the undesirable change in may be reversed through chemical reduction in forming gas at a temperature above 700 °C. Chemically reducing one of the formulations, CARBOHSP, at 1400 °C increases to 0.956 and eliminates degradation of upon further heating in air at 700 °C.
Our efforts to develop new particle formulations yielded 24 compositions, based on seven unique pigments, that maintain after heating in air at 700 °C for 500 h, which is comparable to the performance of the commercially available proppants. These chemically simpler materials may ultimately prove to be more stable than the proppants, possibly avoiding the need for chemical reduction, which would help to reduce costs in a commercial-scale system.
The authors would like to thank Eric Coker, Sandia National Laboratories, for providing the heat capacity data for CARBOHSP.
This work was funded by the U.S. Department of Energy, SunShot Initiative, Project No. #DE-EE0000595-1558.
This is based on the assumption that the particles are at temperature, in storage, for 15 h per day and that 10% of the particles is replaced annual with new particles having αs = 0.956.