Wet-chemistry based selective coatings for concentrating solar power

Eran Maimona, Abraham Kribusa, Yuri Flitsanova, Oleg Shkolnikb, Daniel Feuermannb*, Camille Zwickerc, Liraz Larushc, Daniel Mandlerc, Shlomo Magdassic

aSchool of Mechanical Engineering, Tel-Aviv University, Tel Aviv; bAlexandre Yersin Dept. of Solar Energy & Environmental Physics, Blaustein Institutes for Desert Research, Ben Gurion

University, Sede Boqer Campus; cInstitute of Chemistry, Hebrew University, Jerusalem, Israel.

ABSTRACT

Spectrally selective coatings are common in low and medium temperature solar applications from solar water heating collectors to parabolic trough absorber tubes. They are also an essential element for high efficiency in higher temperature Concentrating Solar Power (CSP) systems. Selective coatings for CSP are usually prepared using advanced expensive methods such as sputtering and vapor deposition. In this work, coatings were prepared using low-cost wet-chemistry methods. Solutions based on Alumina and Silica sol gel were prepared and then dispersed with black spinel pigments. The black dispersions were applied by spray/roll coating methods on stainless steel plates. The spectral emissivity of sample coatings was measured in the temperature range between 200 and 500°C, while the spectral absorptivity was measured at room temperature and 500°C. Emissivity at wavelengths of 0.4–1.7 µm was evaluated indirectly using multiple measurements of directional reflectivity. Emissivity at wavelengths 2–14 µm was measured directly using a broadband IR camera that acquires the radiation emitted from the sample, and a range of spectral filters. Emissivity measurement results for a range of coated samples will be presented, and the impact of coating thickness, pigment loading, and surface preparation will be discussed.

Keywords: Solar Energy, CSP, selective coatings, measurements, emissivity, absorptivity, radiation heat transfer

1. INTRODUCTION

The emissivity of materials and coatings at high temperatures is a very important property in various applications, and especially so in solar energy. However, it is a very difficult property to measure or predict due to the intricate nature of the physical phenomena involved. The available data for emissivity at high temperatures specifically is very limited and often shows great variance between measurements performed under different conditions, preparation, surface structure, etc. [1].

One of the most important applications in which the characterization and manipulation of spectral emissivity and absorptivity is vital, is that of radiative heat transfer management in solar receivers. This is accomplished through the use of selective coatings which absorb the solar spectrum very well, but do not emit much in the infra-red (IR) part of the spectrum [2], [3] (Figure 1). Usually these coatings are prepared by sophisticated CVD/PVD deposition or sputtering methods under high temperatures and high vacuum [4], [5], [6].

In our work, we attempt to develop alternative low-cost CERMET selective coatings that are stable at high temperatures. Coatings are prepared using wet chemistry methods which can be applied to the substrate by simple coating techniques. The CERMET contains black spinel particles embedded in a ceramic matrix (Figure 2). Methods for the preparation of these materials are developed, and feedback is provided by measuring spectral emissivity/absorptivity at high temperatures indirectly in the visible to near infra-red (VIS-NIR) spectrum and directly in the mid-IR to far-IR (MIR-FIR) regions using measurement systems that were designed to be robust enough for selective coatings characterization while maintaining low cost and easy setup using off-the-shelf components.

* Corresponding author: daniel@bgu.ac.il, +972 8 6596927

Figure 1. Nominally ideal spectrally selective coating with absorptance / emittance values of unity up to 2.5 μm and zero from 2.5 μm (dotted line). Superimposed are the standard solar spectrum (Air mass 1.5) and blackbody radiation spectra for 450°C (solid lines)[3].

Figure 2. Schematic: CERMET based Selective coating.

2. COATING PREPERATION

A wet chemistry approach for the preparation of selective coating for high temperatures was taken. A brief description of coating preparation is presented here.

The coatings were applied on stainless steel 316 substrates. First, the substrate was coated with an IR reflective layer. Gold and Silver were tested, and silver was disqualified due to instability at high temperatures (>500°C) when the silver starts to diffuse into the matrix. Samples with and without the IR reflective layer were prepared. It is worth noting that stainless steel (SS) is also reflecting in the IR, though not as well as gold. On top of the substrate, with or without a gold layer, the selective coating was applied.

A typical coating formulation is presented in Table 1. The chemicals are listed in order of their addition. The three matrices in which the solar absorbing pigments were being tested are Silica (Polysiloxane), Alumina sol-gel and Alumina-Silica sol-gel. Each has different optical, mechanical, morphological, and chemical properties. The pigment embedded in those matrices is a commercial black spinel (CuMnO) material. There are several suppliers of this pigment and its exact formulae are commercially classified, thus, pigments from different suppliers are slightly different. The solar coating was applied using hand roller, spraying or spin coating with different wet film thicknesses of 1-40µm on smooth or sandblasted (30µm average grit size) surface of stainless steel (SS) (25x25 mm2, 1mm plate thickness (Figure 3).

After the application of the coating, it was cured by heating to high temperatures for one hour. The coating thicknesses were varied from 1 to 30 μm.

Table 1. Solar selective paint components

Component %wt/wt

Solvent Dispersant Wetting agent Silicate binder Black- pigment Filler Glass additive

20-30 1-2 0.5-1 50 5-20 5-10 1-10

Figure 3. Application methods and curing of the samples. Curing was performed by gradually increasing the temperature up to 600°C, keeping the sample (far right) at maximum temperature for about an hour.

3. INDIRECT ABSORPTIVITY MEASUREMENTS

Samples were irradiated by a Xenon arc discharge lamp (300 W nominal power). A parabolic mirror 200 mm in diameter with focus in the lamp was used to direct nearly collimated light onto the sample. The method of measuring the spectral absorptivity was performed by measuring light reflected from the sample and comparing it to a white reference sample (Spectralon [7]). Measurements were performed with two fiber-optic spectrometers [8], one in the visible (380 to 900nm) and the second in the near infra-red part of the solar spectrum (900 to 1700 nm), and from different directions within the hemisphere (Figure 4). Rotating the sample holder and the optical fiber around a vertical axis, the incident angle of light was varied and the reflected radiation was measured at different polar angles. The effect of azimuthal directions was tested by rotating the sample holder around a horizontal axis (the normal to the sample). The directional hemispherical spectral reflectance was obtained by integrating over all reflectance measurements from different directions with the proper weighting of the respective solid angles. The sample holder can be heated by a custom-made heating element to 700°C.

4. DIRECT MEASUREMENTS OF EMISSIVITY

The surface spectral emissivity at normal incidence was obtained from the ratio of two direct measurements: the radiation flux thermally emitted from the surface at a known temperature, to the flux emitted from a blackbody at the same temperature:

Ti

TiT

b,

,,

(1)

The direct measurement includes two steps: measurement of the blackbody reference emission as a function of temperature, and then measurement of the sample surface emission as a function of temperature. As the emissivity is a function of wavelength, the measurements and calculation of emissivity should be done separately for each relevant wavelength or spectral interval in the spectral range of interest. In our setup the spectral break-down was obtained by using band-pass filters to provide several spectral windows.

Sample

Figure 4. Reflectance measurement system at Ben Gurion University.

An experimental setup was developed for the direct measurement of emissivity using a furnace (E-Instruments TCS-1200 E-Instruments [9]) and an IR camera (Electrophysics PV-320L2E [10]) with measurement capability in the spectral range of 2-14 µm. A chopper and lock-in amplifier system (Thorlabs MC2000+C1F2 [11]) was integrated into the optical path of the system. The signal obtained from the target emission as seen through the chopper, was thus AC modulated, and did not contain the constant part of the signal, which was due to external radiation not passing through the chopper (

Figure 5).

The reference measurement of blackbody radiation used the same setup, except that the sample was removed, and a calibrated blackbody insert was placed in the furnace cavity (

Figure 6). The blackbody temperature was measured by a separate N-type thermocouple.

Each set of measurements included the temperature of the heat flux meter (or of the blackbody in a reference experiment), the heat flux through the sample, and the image of the target (sample or blackbody). The region of interest within the image and the camera internal settings (Gain, Level) were entered manually and saved in the data file together with the measurements.

(a)

(b)

Figure 5. (a) Schematic and (b) picture of the actual setup.

Figure 6. Blackbody inserted instead of the sample inside the furnace aperture.

5. RESULTS

Several contributions may affect the spectral emissivity of a coating: the emission from the substrate (depending on its material, surface roughness, and surface condition); the emission from the matrix (depending on its material, morphology and thickness); the emission from the pigments (depending on the material, particle size distribution, and loading per unit volume, and interactions with matrix material). The objective of the set of experiments reported here was to isolate and quantify the impact of substrate material, coating thickness, surface treatment with sand blasting, and the presence of pigments. Detailed experiments were performed at 500°C and less comprehensive ones at lower temperatures. The following sequence of separate measurements was performed in order to isolate emission contributions:

Bare SS substrate without coating Gold-coated SS substrate (very low emissivity surface as base for matrix characterization) Pure matrix (no pigment) over gold-coated SS substrate, several matrix materials and thicknesses Matrix with pigments over gold-coated SS substrate, several matrix materials and thicknesses

5.1. Substrate emissivity in IR

Figure 7 shows the spectral emissivity of bare SS and of gold-coated SS substrates as measured by our system at 490ºC. The spectral emissivity of gold is below 0.05 across the spectrum, and the total emissivity in the 2–14 µm range is less than 0.02. The bare SS on the other hand has emissivity of around 0.2 over the measured spectrum. These values are consistent with emissivity values in the literature for gold and SS ([12], [13]). The difference is important for thin coatings that may be partly transparent to radiation emitted from the substrate. It is evident that a reflective layer like gold or silver can be effective in reducing IR emission for thin coatings.

Figure 7. Spectral emissivity of SS and gold-coated SS substrates. Two gold coated and two uncoated SS samples where

tested; the substrates were not treated in any way (i.e., no sandblasting).

5.2. Comparison of matrix materials in the IR

Figure 8 compares the emissivity of the three candidate matrix materials, measured for thin layers (between 2 and 6.5 μm thick) of pure matrix materials without pigments on bare SS and gold-coated SS substrates at 490ºC. In the spectral range of 2–4 µm, the results show low emissivity of mostly <0.05 for the gold-coated samples, and higher emissivity of 0.2–0.3 for the SS samples. This indicates that all matrix materials are fairly transparent in this spectral range at thickness up to 6 µm, and the emitted radiation originates mostly from the substrate.

Figure 8. Spectral emissivity of pure matrix materials on SS and gold-coated SS substrates. The substrates were not treated

(i.e. no sandblasting).

In the spectral range of 7–14 µm, a more complicated picture evolves due to a more significant signal from the matrix. Some of the samples show higher emissivity in the range 0.4–0.6, even with the gold base, indicating a significant emission from the matrix itself rather than from the substrate. The highest values are measured for alumina-silica over SS, with somewhat lower values for the same matrix over gold. The silica coating shows high emissivity over gold and lower over SS, an anomaly that cannot be explained based only on the very small difference in thickness. This anomaly may be due to some irregular properties of the specific sample, and before drawing conclusions the repeatability of this effect must be checked. The lowest overall values of emissivity are for alumina over gold. However, the alumina shows higher emissivity then silica when it is combined with pigment, as discussed below.

5.3. Effect of matrix thickness in the IR

Figure 9 shows the effect of matrix thickness for samples of pure matrix without pigment and no SB treatment on total emissivity (2-14µm) at 490ºC.

Figure 9. Matrix thickness effect on SS or gold substrate (no SB and no Pigment)

A slight increase in emissivity with thickness can be observed in most cases as might be expected if the emissivity of the matrix is higher than that of the substrate. The best design is then to have the coating as thin as possible, subject to the effect of absorptivity in the shorter wavelength. Unfortunately, the range of thicknesses applicable for each material is insufficient to detect a saturation effect where the substrate emission is fully absorbed in the coating and further increase in the coating thickness has no effect.

5.4. Effect of the pigment on IR

Adding pigments in the form of spinel particles with sizes of up to 200nm in diameter increases in general the total

emissivity and seems to remove any clear trend in the dependence on coating thickness, as shown in Figure 1010. The irregular behavior may be a result of large variations among the samples in the density and particle size distribution of the pigments (including the presence of agglomerates). The increase in emissivity indicates that the pigment has a significant contribution to emission in the IR, even though it is intended for absorption at shorter wavelengths and was expected to be transparent in the IR. The dependence of emissivity on the matrix materials and on the type of substrate also becomes weaker, further supporting the same conclusion on the significant role of the pigment.

Figure 10. Total emissivity vs. thickness for samples on SS and gold substrates

Figure 11 shows the effect of the pigment on the spectral emissivity for pairs of samples (with and without pigment) with same matrix material and similar thickness on a SS substrate (a) and on a gold substrate (b). In each pair, the addition of the pigment leads to a significant increase in the spectral emissivity at all wavelengths. This again contradicts the expectation that the small pigment particles should have a weak interaction at least with the longer wavelengths. An anomaly in these results can be seen for sample Silica over SS in Figure 11a where the emissivity at long wavelengths is lower than that of bare SS. This cannot be explained by the matrix behavior, and we presume that it is either a measurement error or some damage to the sample. Ignoring this problematic sample, we can make the following general statements:

Alumina without pigment has relatively low emissivity over the entire spectrum, over both SS and gold substrates Silica and alumina-silica without pigment have low emissivity for mid-wavelength <7 µm, but higher emissivity in

the long wave range, over both SS and gold substrates (this is consistent with the properties of pure silica) Silica and alumina-silica with pigment have similar emissivities on gold substrate, but on SS substrate alumina-silica

with pigment shows a much higher emissivity then silica with pigment The pigment has a stronger effect on alumina compared to the other matrix materials, and the emissivity of alumina

with pigment on SS is much higher than silica with pigment.

Silica

Alumina

Al-Si

Silica over Gold

Al-Si over Gold

Figure 11. Effect of the pigment on the spectral emissivity of similar samples (a) on SS substrate, (b) on gold substrate

Table 2 shows the total emissivities of selected samples at 490ºC of silica and Al-Si, the best matrix material candidates. In all compared cases, the silica samples show lower emissivity at similar thickness, or about the same emissivity at much higher thickness. Therefore, it seems that a silica matrix might be the best option.

Table 2 : Total emissivity at 500°C for selected samples of silica and Al-Si, no sandblasting

Substrate Pigment addition

Silica thickness ε

Al-Si thickness ε

Gold no 6.5 0.26 3 0.21

yes 11 0.6 9 0.67

SS

no 9.6 0.34 4.5 0.38

yes 19 0.5 20 0.62

yes 14.1 0.5 9.5 0.56

(a)

(b)

5.5. Effect of sandblasting in IR

Sandblasting helps with adhesion of the coating to the substrate but has an effect on the emissivity. Figure 12 shows the spectral emissivity of two pairs of similar coatings, comparing sandblasted surface vs. natural untreated surface. In both cases the surface treatment resulted in significantly higher emissivity (indicated by the arrows). This was expected as the rougher the surface the closer the microscopic valleys resemble a cavity and the greater the emissivity of same surface becomes.

Figure 12: Effect of sandblasting on the spectral emissivity of similar samples.

Coating thickness does not seem to affect sandblasted samples in a consistent way as can be seen in Figure 13. This might be due to the relatively large pores on the surface (due to sandblasting with coarse grain sand), which leads to high variability in coating thickness, and a significant part of the surface where the local thickness is higher than expected.

Figure 13. Thickness effect on Sand blasted samples

0 10 20 30 40 500

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Thickness [m]

Thickness effect on sandblasted samples

mat: Si SB:1 P:0

mat: AS SB:1 P:0

mat: GX SB:1 P:0

mat: GS SB:1 P:0

Silica

Al-Si

Silica over Gold

Al-Si over Gold

The SB treatment as applied here contributes to higher emissivity in the IR and thus should be avoided. However, this treatment may be modified with different grain size (leading to different characteristic scale of the surface roughness), and the optical impact of this change can be an interesting future investigation.

5.6. Indirect absorption measurements in the VIS-NIR

Measurements in the solar spectrum show very good absorptance for all samples (with pigments). We show here the reflectance of a typical sample for two representative wavelengths. The incidence angle θi of radiation was kept constant for all experiments at θi=30° since in a concentrating system, incidence angles necessarily cover a large range, thus 30° is considered a representative angle.

Figure 14. Directional reflectance for 20 µm alumina silica sample irradiated at θi=30°, for two representative wavelengths

and at two temperatures: (left) ambient temperature (right) 500 °C.

We have found the samples not to be sensitive to azimuthal orientation. However, the samples showed partially specular character when the incident angle and the direction of measurement were equal, this is clearly shown in Figure 14. The same distribution was observed at elevated temperatures. Samples that were sandblasted showed lower specular effects, as expected as the created roughness of the surface was of the order of the thickness of the coating. However, since the ‘specular’ region is of small weight, the effect on solar absorptance was observable, but very small. Solar reflectance was obtained by summing over all directional reflectances while weighting them with their respective solid angle and the standard solar spectrum for air mass 1.5 [14]. Typical results are shown in Table 3. Absorptances were very high for almost all samples with values slightly higher at higher temperature. The effect of coating thickness is very small and saturation occurs already after 1 or 2 μm. Repeated measurements (after heating) showed excellent repeatability (within less than the estimated errors).

6. DISCUSSION AND RECOMMENDATIONS

6.1. Summary and conclusions

This work is an attempt to develop low-cost selective coatings using wet chemistry methods that are stable at high temperatures and which can be applied to the substrate by simple coating techniques. Experimental systems were developed that permit the measurement absorptivities of coatings on substrates from room temperature to 500°C for the solar spectrum and the measurement of emissivities of same coatings at temperatures from 200 to 500°C. The study tries to isolate the effects of several coating parameters: the separate contributions of the substrate, matrix, and pigment; the effect of coating thickness; and the effect of sand blasting surface treatment. Many different samples were measured with different combinations of these parameters. The main conclusions emerging from the results are presented below. Solar spectrum absorption In the VIS-NIR regime, the samples are very good absorbers with above 95% absorptivity. There is a significant specular component for non-sandblasted samples which might be reduced with a very fine sand-blasting that would not affect IR behavior (i.e. smaller than 2µm grain sizes).

0

10

20

30

40

50

60

70

-20 -10 0 10 20 30 40 50 60

refl

ecta

nce

(%

)

polar angle θs(°)

550 nm

1600 nm

0

10

20

30

40

50

60

70

-20 -10 0 10 20 30 40 50 60

refl

ecta

nce

(%

)

polar angle θs(°)

550 nm

1600 nm

Effect of the substrate The contribution of the SS substrate to emissivity in the IR is significant, and the reflective sub-coating (gold in the measured samples) reduces this contribution. Table 3. Typical results for the three matrices on SS at room and elevated temperatures for different thicknesses. Errors for the absorptance measurements are ±0.01, while errors on the emissivities are estimated at ±0.06

matrix coating

thickness

solar absorptance at

ambient temperature

solar absorptance at

500 °C

emissivity at 500°C

Silica

5.1 0.953 0.969 0.37 14.7 0.959 0.975 0.5 19 0.962 0.973 0.5

Alumina-silica

1 0.953 0.969 0.69 1 0.948 0.96 0.68 2 0.952 0.968 0.64 4 0.97 0.971 NA 6 0.946 0.971 0.66 6 0.959 0.966 NA 8 0.919 0.922 0.51

9.5 0.967 0.973 0.56

Alumina

1 0.954 0.959 0.48 4.4 0.962 0.967 0.47 5.2 0.964 0.971 0.63 5.8 0.966 0.972 0.53 6.3 0.966 0.973 0.51 9.2 0.966 0.974 0.64 20 0.959 0.969 0.62

Coating thickness Solar absorption for coatings thicker than about 3µm saturates and does not continue to rise. The emission of the pure matrix without pigment shows a correlation with thickness as expected: thinner matrix has lower emission. The presence of pigment removes any clear trend, probably due to non-repeatable pigment properties (see below) that has a stronger effect than the layer thickness. The important property of uniformity of the coating thickness is under investigation. Non-uniformities can have a major impact on the emission. Sand blasting Surface treatment with coarse-grain sand blasting increases the emission at all wavelengths. Matrix materials Comparing the pure matrix materials without pigment, the alumina matrix offers the lowest emissivity. In the presence of the pigment, the silica coating has the lowest total emissivity and seems preferable to the other two materials. Pigment The pigment causes a significant increase of emission across the spectrum and shows irregular behavior, indicating inconsistent application in different samples (pigment loading, particles shape and size distribution, composition).

Overall performance The total emissivity at 500°C for most samples with pigments was in the range 0.4–0.6. This is higher than the desirable range of around 0.1. The spectral emission shows typically higher values in the long wavelength range >7 µm, and lower but still significant values in the mid wavelength range 2–7 µm. An understanding of the factors that influence the spectral variations is needed, including the longer wavelength range (30% of blackbody radiation emitted at 500°C is above 7 µm).

6.2. Ongoing work

The physical characteristics (loading, particle size and shape distribution) of the pigments for each sample, and their impact on the emissivity is being studied in detail with the goal of achieving lower emissivities. Fine-grain sand blasting of around 1-2µm should be tried in order to assist in coating adhesion to the substrate without hindering IR low emissivity. Further, the repeatability of the fabrication process, control of thickness and pigment distribution is being be investigated; and accelerated testing for establishing the long-term performance of the coatings needs to be performed.

ACKNOWLEDGEMENT

This research was funded in part by the Israel Ministry of Science and Technology, Grant No. 3-8146.

REFERENCES

[1] William B. Stine and M. Geyer, Power From The Sun. 2001. [2] C. M. Lampert, “Coatings for enhanced photothermal energy collection I. Selective absorbers,” Solar Energy

Materials, vol. 1, no. 5–6, pp. 319–341, Jun. 1979. [3] C. E. Kennedy, “Review of Mid- to High- Temperature Solar Selective Absorber Materials Review of Mid- to

High- Temperature Solar Selective Absorber Materials,” NREL/TP-520-31267, July, 2002. [4] H. C. Barshilia, P. Kumar, K. S. Rajam, and a. Biswas, “Structure and optical properties of Ag–Al2O3

nanocermet solar selective coatings prepared using unbalanced magnetron sputtering,” Solar Energy Materials and Solar Cells, vol. 95, no. 7, pp. 1707–1715, Jul. 2011.

[5] P. Oelhafen and a. Schüler, “Nanostructured materials for solar energy conversion,” Solar Energy, vol. 79, no. 2, pp. 110–121, Aug. 2005.

[6] A. Shumway and A. Jaworske, “Solar Selective Coatings for High Temperature Applications,” AIP Conf. Proc. 654, pp. 65-70; doi:http://dx.doi.org/10.1063/1.1541278, 2003.

[7] “Spectralon.” [Online]. Available: http://www.labsphere.com/products/reflectance-materials-and-coatings/default.aspx. [Accessed: 24-Jul-2013].

[8] “Spectrometers | Control Development.” [Online]. Available: http://www.controldevelopment.com/products/spectrometers/index.html. [Accessed: 24-Jul-2013].

[9] “TCS 1200 Multifunction High-Temperature Calibrator | E Instruments.” [Online]. Available: http://www.e-inst.com/temperature-calibrators/products-TCS-1200. [Accessed: 25-Jul-2013].

[10] “PV320L Thermal Imaging Infrared Camera.” [Online]. Available: http://www.electrophysics.com/Browse/Brw_ProductLineCategory.asp?CategoryId=145&Area=IS. [Accessed: 25-Jul-2013].

[11] “MC-2000 Optical Chopper System and Chopper Wheels - Thorlabs.” [Online]. Available: http://www.thorlabs.com/NewGroupPage9.cfm?ObjectGroup_ID=287&pn=MC2000. [Accessed: 30-Jul-2013].

[12] P. D. Jones and E. Nisipeanu, “Spectral-directional emittance of thermally oxidized 316 stainless steel,” International Journal of Thermophysics, vol. 17, no. 4, pp. 967–978, Jul. 1996.

[13] L. N. Aksyutov, “Normal spectral emissivity of gold, platinum, and tungsten,” Journal of Engineering Physics, vol. 27, no. 2, pp. 913–917, Aug. 1974.

[14] in: “ASTM G173-03(2012) Standard Tables Reference Solar Spectral Irradiances: Direct Normal Hemispherical on 37° tilted surface.” [Online]. Available: http://webstore.ansi.org/RecordDetail.aspx?sku=ASTM+G173-03(2012). [Accessed: 25-Jul-2013].