Novel Mo–Si3N4 based selective coating for high temperature concentrating solar power applications

Novel Mo–Si 3 N 4 based selective coating for high temperature concentrating solar power applications Eva Céspedes a,1 , Men Wirz b , J.A. Sánchez-García a,2 , L. Alvarez-Fraga a , R. Escobar-Galindo a , C. Prieto a, n a Instituto de Ciencia de Materiales de Madrid, Consejo Superior de Investigaciones Científicas, Cantoblanco, 28049-Madrid, Spain b Department of Mechanical and Process Engineering, ETH Zurich, 8092 Zurich, Switzerland article info Article history: Received 3 July 2013 Received in revised form 15 November 2013 Accepted 9 December 2013 Available online 1 January 2014 Keywords: Selective coating material Solar absorber cermets Parabolic trough concentrator for CSP technology abstract A novel selective coating material based on Mo–Si 3 N 4 has been investigated for the first time. Accurate combination of the individual component layers based on optical simulations and precise control of layers thickness and composition leads to high solar absorptivity (α Sol ¼0.926) and low thermal emissivity (ε 25 1C ¼0.017) on a silver film, resulting in a high solar thermal energy conversion efficiency. The tandem absorber is stable in moderate vacuum at temperatures above 600 1C showing low emissivity (estimated value at 600 1Cisε 600 1C ¼0.109) evidencing its relevance for high temperature solar selective applications. A full optical and thermal analysis of a parabolic trough concentrator (PTC) system confirmed the potential of the novel coating material to be used for CSP technology. Crown Copyright &2013 Published by Elsevier B.V. All rights reserved. 1. Introduction Nowadays, thermal conversion of solar energy is one of the more simple methods of harvesting renewable energy. In this direct heating technology, solar energy efficiency is essentially determined by the solar selective coatings, requiring high solar absorbance and low thermal emittance, with the solar absorber material being of fundamental importance. Since efficiency increases with working temperature in concentrated solar power (CSP) systems, research has been aimed at designing selective solar coatings with optimum optical properties at high temperatures (Z400 1C) [1]. For this purpose, the fabrication of coatings by physical vapor deposition (PVD) techniques has become one of the most popular ones in terms of environmental issues and also because those PVD-deposited coatings usually present higher thermal stability than when prepared by wet chemistry methods. Recently, an exhaustive review of selective coatings materials prepared by PVD techniques has been reported [2]. Among these PVD techniques, sputtering appears to be the most suitable for industrial production of large area coatings with very precise control of layer thickness, which is a fundamental parameter to achieve the required optical absorbance. Among the different methods to optimize the required spectral selectivity, multilayer coatings are of special interest because of the very high solar efficiency that may be obtained at medium and high temperatures. In this sense, the approach which has proven to give high solar performances [3–5], by using a simple fabrication procedure is the structure formed by four layers: (i) an infrared reflective metallic layer (IR-mirror), (ii) a high metal volume fraction (HMVF) cermet, (iii) a low metal volume fraction (LMVF) cermet layer, and (iv) an anti-reflective (AR) layer. In a coating with this structure, due to the gradual variation of refractive index, solar radiation is efficiently absorbed internally, and by phase interference between thedoublecermetandtheARlayers. Regarding solar absorber materials, a large list of components has been reported. Most of them are formed by cermets composed of metals such as Mo, Ni, Fe, Cr, W or Pt and ceramics compiled with oxides, oxynitrides or nitrides. Oxide-based cermets have been widely studied, providing some of the few commercially available high-temperature solar selective coatings: for instance, Angelantoni ENEA (Italy) commercializes Mo–SiO 2 [5,6], and W–Al 2 O 3 , while Siemens (Germany) commercializes Mo–Al 2 O 3 and W–Al 2 O 3 coatings [7–9], which are stable in the 350–500 1C temperature range. Not many examples can be found of nitrides, where AlN has been studied [10,11], to form nitride-based cermet absorbers, which have been commercially used by Turbosun (China), being thermally stable at 350–500 1C in vacuum. Additionally, a large list of cermets has been proposed based on several Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/solmat Solar Energy Materials & Solar Cells 0927-0248/$-see front matter Crown Copyright &2013 Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.solmat.2013.12.005 n Corresponding author. E-mail address: [email protected] (C. Prieto). 1 Present address: Institute for Science and Technology in Medicine, Keele University, Guy Hilton Research Centre, Thornburrow Drive, Hartshill, Stoke-on-Trent ST4-7QB, UK. 2 Present address: TECNALIA, Parque Tecnológico de San Sebastián, Mikeletegi Pasealekua, 2. E-20009 San Sebastián, Gipuzkoa, Spain. Solar Energy Materials & Solar Cells 122 (2014) 217–225 mixed oxynitride compounds, such as aluminum oxynitride [12] or oxynitrides used in tandem with a similar nitride, such as TiAlN/ TiAlON [13], and more recently NbAlN/NbAlON [14,15], and HfMoN/HfON [16] for high temperature applications. For practical high temperature applications, low emissivity is a critical parameter, because thermal radiative losses from the absorbers increase with T 4 .Someofthesematerialshavebeenfoundtobe stable in vacuum or air up to 400–600 1C, showing excellent values of absorptivity in the range of 0.93–0.96. However, in most cases, the thermal emissivity ε th is not tuned or even evaluated at working temperatures. ε th is frequently estimated around 100 1C (typical ε 100 1C ¼ε th (100 1C) are between 0.04 and 0.1) while for some materials, ε 350 1C has been reported (ε 350 1C ¼0.08–0.1 for W–AlN and Mo–AlN onto Cu or Mo–Al 2 O 3 ). Just a few evaluations of ε th at hightemperaturehavebeendone,suchasW–Al 2 O 3 (ε 550 1C ¼0.14– 0.25) or Mo–SiO 2 (ε 580 1C ¼0.12) [2]. For higher temperatures, the solar selective stacks generally degrade, causing a decrease in the solar selectivity. The search of new promising materials showing thermal stability but also tuned optical properties at high temperatures is of great interest in CSP applications. To our best knowledge, silicon nitride has not been investigated to date as a candidate for the ceramic component of the cermet in selective coatings. It has been shown that silicon nitride is a promising material to be used as insulator matrix in metal/ceramic granular systems and multilayers, with potential applications in electronic devices due to its optical transparency, transport properties [17] and chemical inertness at high temperatures [18].In addition, superior mechano-elastic properties have been reported for Si 3 N 4 when prepared by reactive sputtering from pure silicon [19] with respect to common silicon oxynitride. Due to this mechanical behavior, Si 3 N 4 is used as coating for magnetic devices [20] or to form superhard TiN–Si 3 N 4 nanocomposite coatings [21]. Furthermore, from a cermet component point of view, since metal nitridation is less energetically favorable than oxidation, silicon nitride cermets are expected to have an enhanced chemical stability at high temperatures compared to silicon oxynitride. In this paper, we report on a new promising solar selective coating based on Si 3 N 4 /Mo–Si 3 N 4 . The Mo–Si 3 N 4 absorber consists of a double layer of high and low metal volume fractions, and on top a Si 3 N 4 layer acts as a dielectric AR coating. Optical properties of Mo–Si 3 N 4 cermets and simulations of the whole stack have been investigated by spectroscopic ellipsometry and by simulation to achieve the optimum optical selectivity at high temperatures. A precise experimental control of composition and thickness of the individual component layers by means of X-ray Reflectivity (XRR) and Rutherford Backscattering Spectroscopy (RBS) characterizations has led to optimized solar absorptivity and thermal emissivity above 600 1C. Additionally, thermal stability under moderate vacuum conditions (110 2 mbar) has been studied. Finally, a full optical and thermal analysis of the developed coatings is conducted, determining the performance of each stack in a parabolic trough concentrator (PTC) system and comparing the results to existing selective coatings. 2. Experimental The tandem absorbers were deposited at room temperature on Si (100) and stainless steel AISI-321 substrates (previously annealed at 600 1C in air, labeled as SSth), using magnetron sputtering. The stack materials were: silver as metallic IR reflector, molybdenum–silicon nitride composites as cermet layers, and silicon nitride as AR layer. The base pressure was around 1 10 6 mbar. The Ag layer was deposited by direct current (DC) sputtering of Ag with Ar at 910 3 mbar at 15 W (deposition rate 5 nm/min). After Ag deposition, the Ag layer surface was passivated at the chamber by flowing oxygen to achieve a pressure of 510 2 mbar during 5 min. Mo– Si 3 N 4 cermets were prepared by simultaneous co-sputtering of Si and Mo targets using N 2 as reactive gas at 710 3 mbar. Radiofrequency (RF) and DC sputtering were used for Si and Mo, respectively. The metal volume fraction, or filling factors (FF), were modified by tuning the power supplied to each target. Parameters for acermetwith37%filling factor (typically used for the HMVF layer) are 100 W for both Si and Mo targets to get 7.1 nm/min deposition rate and for 20% filling factor (typically used for the LMVF layer) are 100 and 17 W for Si and for Mo, respectively, to get 2.7 nm/min deposition rate. Si 3 N 4 AR layers were prepared by RF sputtering of Si with N 2 at 710 3 mbar and 100 W (deposition rate 1.7 nm/min). By varying the Mo content inside the cermet, as well as the thickness of the cermet and AR layers, multilayer structures were optimized for high temperature photo-thermal performance. It should be noted the importance of the Ag layer passivation before depositing the double cermet layer [22]. The reason for this Nomenclature Abbreviations AR anti-reflective CSP concentrated solar power DNI direct normal irradiance FF filling factor HMVF high metal volume fraction HTF heat transfer fluid LMVF low metal volume fraction MG Maxwell-Gamett PTC parabolic trough concentrator PVD physical vapor deposition RBS Rutherford backscattering spectroscopy XRR X-ray reflectivity Symbols A[W/m 2 μm], ASTM AM1.5D solar spectral irradiance A ap [m 2 ], Total aperture area of the concentrator mirrors I DNI [W/m 2 ], Direct normal irradiance L[W/m 2 μm], Black body emission Planck function P pump [W], Required power to pump the HTF Q gain [W], Net heat gain of the HTF Q loss [W], Total heat loss from the absorber tube R[–], Reflectivity Greek symbols α sol [–], Solar absorptivity ε th [–], Thermal emissivity η el [–], Efficiency of power block, generator, and HTF pump η opt [–], Optical efficiency of the solar field loop η th [–], Thermal efficiency of the solar field loop λ[μm], Wavelength θ Sol [1], Incidence angle at the concentrator aperture E. Céspedes et al. / Solar Energy Materials & Solar Cells 122 (2014) 217–225218 passivation is to prevent the reactivity between the IR-mirror and the absorber cermet, which results in worsening the abruptness of transition from low to high values in the reflectance spectra. The explanation should be found in the roughness enhancement allowed by the Ag–Si 3 N 4 reactivity that locally soften the refractive index change from the cermet to the IR-mirror and, consequently, affects the interference pattern of the multilayer. XRR was measured in a D8-Brucker AXS diffractometer with X-ray energy of the Cu K α emission. Microstructural features, such as layer thickness and average roughness were obtained by fitting the experimental XRR data with the XREAL software [23]. In order to determine each layer composition, RBS experiments were performed with the 5 MV HVEE Tandetron accelerator located at the Centro de Micro-Análisis de Materiales of the Universidad Autónoma de Madrid [24]. The RBS experiments were performed both using 3.7 MeV He þ ions to make use of the resonance 14 N(α, α) 14 N, in order to improve the sensitivity to nitrogen, and 2.0 MeV He þ ions with the sample tilted (θ¼451) to improve the depth resolution of the RBS analysis. The experimental spectra were fitted with the RBX software program [25]. No appreciable oxygen amount was detected along the whole film when performing resonant backscattering spectroscopy experiments. Deposited cermets were optically characterized by ellipsometry to enable optical simulations of the reflectance spectrum by using the CODE (Coating Designer version 3.75) software for optical spectroscopy, developed by Theiss [26].UV–vis–IR reflectance measurements were performed using both a Shimadzu SolidSpec-3700 spectrophotometer in the range of 190–2600 nm, and a Varian 660-IR FTIR spectrometer in the 1.5–25 mm wavelength range. Using the calculated or experimentally obtained spectral reflectance data R(λ), the solar-weighted absorptivity α Sol is determined by α Sol ¼R λ 2 λ 1 ½1RðλÞAðλÞdλ R λ 2 λ 1 AðλÞdλ ð1Þ where A(λ) is the ASTM AM1.5D solar spectral irradiance, which is the reference after passing through the atmosphere 1.5 times. α Sol is calculated with integration limits λ 1 ¼0.25 μmandλ 2 ¼4μm. This rangeiswiderthantheonetypicallyreportedinliterature(i.e.in Ref. [6] the authors present α Sol in a 0.3–1.65 μmrange)toaccountfor the whole solar spectrum considered in the AM1.5D data. The thermal emissivity ε th at a specific temperature Tis determined using ε th ðTÞ¼R 1 0 LðT;λÞ½1RðλÞdλ R 1 0 LðT;λÞdλð2Þ where L(T,λ) is the black body emission Planck function. The spectral reflectance data R(λ) is assumed to continue with constant reflectivity values equal to the first and last measurement point for wavelengths below and above the measurement limits, respectively. Even though the spectral reflectance, and consequently the absorptivity and emissivity, is also directionally dependent (R(λ,θ)), a good approximation is obtained from the near normal spectral reflectance (R(λ)) provided by the measurements. Additionally, direct emissivity measurements of the samples were obtained at 82 1C, using a commercial emissometer from Devices & Service Company (model AE1). 3. Results and discussion 3.1. Simulation of optical properties In order to use simulation for studying optical properties of selective coating multilayers, the refractive index and extinction coefficient of each layer have to be known in the whole wavelength range (n(λ) and κ(λ), respectively). For this purpose, the knowledge of these cermet properties as a function of its filling factor is also needed. Once n(λ) and κ(λ) have been obtained, the commercial CODE provides the needed reflectance spectra to evaluate the solar absorptivity and the thermal emissivity. In this section, the optical properties of a cermet formed by metallic Mo particles embedded in ceramic Si 3 N 4 are studied. A detailed description of the performed simulation procedure is reported because, as seen below, a general description of the dielectric constant is needed to simulate the optical behavior due to the high metal fraction in some layers of the coating. Fig. 1(a) shows the experimental tan(ψ) (amplitude ratio of the parallel and perpendicular components of the reflected light) and cos(Δ) (relative phase change) spectra obtained by spectroscopic phase modulated ellipsometry over a wavelength range of 200– 2000 nm for Mo–Si 3 N 4 cermets with two different Mo filling factors (f), obtained by co-sputtering on a Si(100) substrate. Fig. 1. (a) Experimental ellipsometric measurements of Mo–Si 3 N 4 single cermets layers (for two different Mo-contents) deposited on Si substrates (points) and simulated spectra (dashed lines) obtained by the CODE software, where the Bergman representation was used for simulation. (b) Optical constants (refractive index, n, and extinction coefficient, κ) obtained for cermets and Si 3 N 4 . (c) Spectral density function obtained for two cermets. E. Céspedes et al. / Solar Energy Materials & Solar Cells 122 (2014) 217–225 219 Experimental spectra have been fitted with a double layer model, where the optical constants of the Si substrate layer were taken from literature [27].Si 3 N 4 and Mo contributions were simulated by using Cauchy 0 s formula and the Drude model, respectively, for generating both refractive index (n) and extinction coefficient (κ) spectra as a function of wavelength, being in good agreement with values reported elsewhere [28]. The complex refractive index of cermets materials can be calculated from the combination of the refractive indices of the metal and dielectric components using a physical model (Fig. 1(b)). The optical response of these cermets is usually modeled by using the so-called effective medium theories, where the metal particles have a much smaller size than the wavelength of the incident light. The most widely used physical models for the dielectric function of a composite have been proposed by Maxwell-Garnett [29], Bruggeman [30], and Bergman and Milton [31,32]. The Maxwell-Garnett (MG) formula is applicable to systems of low volume fractions (or filling factors) of the embedded metal spheres particles, which must also be far away from each other. The Bruggeman model is probably the most often used effective medium concept. For very low volume fractions, it is equivalent to the MG theory, but for increasing volume fraction its results are quite different. Below f¼1/3, there is no percolation, but above this threshold the embedded particles are assumed to be partially connected [29]. Finally, the so-called Bergman spectral representation theory is the most general form of effective medium approximation. It states that the effective dielectric function (ε eff ) of a heterogeneous medium made up of two phases (labeled m and cfor metal and ceramic, respectively) with dielectric functions ε m and ε c , and with a filling factor of f m ¼fis given by ε ef f ¼ε c 1þfC m ε m ε c 1  þZ 1 0 gðpÞ tþpdp "# ! with t¼ε c ε c ε m ð3Þ where, for the m-phase, g(p) is the spectral density function, C m is the so-called percolation strength, and p(which runs from 0 to 1) can be identified as the depolarization factor of the particles making up the composite. In general, g(p) is a complicated function of the geometric structure of the composite. This function is determined by the shape and distribution of the interfaces between the different phases present in the composite, being constrained by several sum rules [33,34]. Moreover, some theoretical and empirical models of g(p) have been described in the literature [35,36],tofit experimental data. For the system studied here, using the CODE software [25], the fit of experimental reflectance spectra and ψand Δvalues with the Bergman representation has allowed us to obtain the spectral density functions given in Fig. 1(c). The low volume fraction (f¼0.1) exhibits a maximum around p¼0.3, which indicates no percolation and would allow the use of any effective medium theory, because it does not overpass their corresponding filling factor limitations [28,29]. On the other hand, functions for higher fvalues present a definite maximum shifted to lower pvalues, which is proof of mphase percolation and therefore, the Bergman theory is required. The fit for f¼0.4 assumes that Mo is percolated, obtaining a percolation strength of C m ¼0.14. In addition to the refractive indices of metal and ceramic forming the cermet material, IR-mirror and substrate data are necessary to simulate the reflectance of the whole selective stack, from which solar absorptivity and thermal emissivity can be obtained. Ag films and SSth substrates, not shown here, were also characterized by spectroscopic ellipsometry, to have all inputs for the optimization of the whole stack by means of CODE software simulations. Fig. 2 exhibits the results using the CODE software for Mo– Si 3 N 4 cermets on thermally treated AISI-321 stainless steel substrates with (a) different Mo volume fractions (ranging from 10 to 50%) for a selected thickness of 50 nm, as well as (b) for different layer thicknesses (between 20 and 100 nm) for a selected Mo content of 10%. Reflectance curves of Si 3 N 4 and Mo films, as well as of the annealed SS substrate have been included for comparison in Fig. 2(a). Simulation shows how critical the Mo volume fraction and layer thickness are to achieve favorable spectral reflectivity characteristics of the coating. Due to such a critical behavior, accurate control of the composition and thickness of the deposited films is essential. In order to achieve an optimum spectral selectivity, it is necessary to optimize the layer parameters of a multilayer stack (i.e., layer thickness, metal volume fraction of the cermet layers, and number of layers), to minimize or maximize the reflectance for specific wavelength regions. For this purpose, Downhill-simplex and Generalized Reduced Gradient nonlinear optimization codes were used. Plainly speaking, to have a near ideal selective absorber coating, reflectance of the multilayer should be minimized in the UV–vis–NIR optical spectrum and maximized in the IR range, requiring the optimization of the reflectance in both the solar spectrum and the spectrum affected by thermal emission. Therefore, optical simulations of the SSth/Ag/Mo– Si 3 N 4 (HMVF)/Mo–Si 3 N 4 (LMVF)/Si 3 N 4 stack allow inferring the accurate thickness and volume fraction values for optimized selective coatings for high temperature CSP applications. Optimization of thicknesses and filling factors values have been made following a trial and error procedure. Table 1 summarizes four selected stacks with different metal volume fractions and Fig. 2. Spectral reflectivity (R(λ)) simulations using CODE of Mo–Si 3 N 4 single cermets onto thermal annealed stainless steel (SSth) substrates. (a) R(λ) spectra for cermets with different Mo volume fractions, varying from 10 to 50% (with a constant thickness of 50 nm). (b) R(λ) for different layer thicknesses, ranging from 20 to 100 nm (with a Mo volume fraction of 10%). E. Céspedes et al. / Solar Energy Materials & Solar Cells 122 (2014) 217–225220 layer thicknesses, in which solar absorptivity and thermal emissivity might be considered as acceptable magnitudes. Simulated reflectivity spectra of these four selective coating stacks are shown in Fig. 3. It can be observed how such small changes in thickness and composition of the constituent layers strongly affect the reflectance spectrum and, consequently, the optical absorptivity and thermal emissivity parameters. For the simulated spectra, subtle changes in the layer thicknesses for same f(see Sim-1 vs Sim-2) lead to some differences in ε 25 1C , from 0.02 to 0.03, and significant variation of ε 600 1C , increasing from 0.09 to 0.15. Further modifications of thickness and fvalues imply notably differences in α sol and ε th parameters (see Sim-3 and Sim-4), enhancing α sol for Sim-4 (up to 0.95) with an important worsening of ε th (ε 25 1C ¼0.04 and ε 600 1C ¼0.28). Regarding the optical simulations using CODE, solar absorptivity above 0.92 with low emissivity values for high temperature applications (ε 600 1C between 0.1 and 0.2) are expected for this novel system. 3.2. Structural characteristics of deposited selective coatings Based on the optical simulation results of cermet thickness and filling factors, deposition of the selective multilayer stacks was performed to (i) assess the experimental behavior and (ii) to evaluate and optimize the actual solar efficiency at high temperature. To ensure a precise thickness and composition experimental control of the selective stacks, single film layers were investigated to calibrate the deposition rate for different Mo volume fractions. In this way, calibration may be used to set the appropriate thickness and Mo–Si 3 N 4 relative composition. Mo–Si 3 N 4 cermet layers of similar thickness as predicted for optimal optical properties were measured by XRR for an accurate evaluation of the deposition rate. On the other hand, the metal volume fraction of the prepared cermets was determined by RBS. To facilitate RBS analysis, cermet layers were deposited onto Si(100). Fig. 4 reports on several examples to illustrate the accuracy obtained by the followed procedure to control thickness and filling factor of deposited layers. Fig. 4(a) shows XRR spectra of selected Mo–Si 3 N 4 single layers on Si, with different Mo volume fractions and a thickness of around 50 nm. These single cermet layers on silicon are labeled as “MoSN–Fxx”to be distinguished from the “Sxx”notation of the selective stacks. It can be observed that XRR curves exhibit interference fringes, evidencing smooth layers. Thickness values were obtained by fitting the experimental data Table 1 Summary of selective coating stacks characteristics and optical parameters of the simulated reflectivity spectra included in Fig. 3 (FF is the filling factor). Solar absorptivity (α Sol ) has been calculated from R(λ) in the 0.25–4μm range and thermal emissivities (ε 25 1C and ε 600 1C ) in the 0–1range by using the blackbody emission curve at each temperature. Stack Subst. IR-mirror Ag Thickness (nm) HMVF cermet Mo–Si 3 N 4 LMVF cermet Mo–Si 3 N 4 AR layer Si 3 N 4 Thickness (nm) α Sol ε 25 1C ε 600 1C FF (%) Thickness (nm) FF (%) Thickness (nm) Sim-1 SSth 150 40 40 20 40 45 0.905 0.023 0.085 Sim-2 SSth 150 40 50 20 55 60 0.816 0.029 0.146 Sim-3 SSth 150 35 70 25 60 50 0.937 0.036 0.195 Sim-4 SSth 150 45 80 15 80 50 0.942 0.046 0.283 Fig. 3. Simulated reflectivity spectra of selective coating stacks based on Ag/ Mo– Si 3 N 4 (HMVF)/ Mo–Si 3 N 4 (LMVF)/ Si 3 N 4 with different metal filling factors and layer thicknesses. Stacks characteristics and calculated α Sol and ε th values are summarized in Table 1. The reference direct AM1.5 solar spectrum and the black-body emission at T¼600 1C are also included for comparison. Fig. 4. (a) XRR measurements (dots) and simulations (lines) of Mo–Si 3 N 4 single cermets on Si. (b) RBS data of “S10”coating on stainless steel substrate and two Mo–Si 3 N 4 single cermets on Si (dots) and simulations (contributions of “S10”stack have been plotted separately). Experiments were made with 3.7 MeV He þ incident ions. E. Céspedes et al. / Solar Energy Materials & Solar Cells 122 (2014) 217–225 221 with the XREAL software. The excellent agreement of simulated spectra with the experimental XRR data shown in Fig. 4(a) provides the layer thickness values reported in Table 2 with a typical accuracy of 0.5 nm. Fig. 4(b) shows selected experimental RBS data (and their corresponding fits) of single layer cermet samples, including MoSN–F10 and MoSN–F13, as well as “S10”, a whole selective stack. Accurate Mo filling factor values have been achieved by RBS simulations. Actual composition values have been summarized in Table 2. After this calibration procedure, the accuracy for designed selective stacks may be estimated in 71 nm for each prepared layer thickness and 71% for filling factor values. 3.3. Optical properties of deposited selective coatings Evaluation of the optical characteristics of the selective coatings has been carried out by measuring the reflectance over a wide optical wavelength range (from 200 nm in the UV to 25 μm in the IR), to account for the absorption of the whole solar spectrum and for the thermal emission at working temperatures. Fig. 5 shows the reflectivity spectra measured at near normal incidence for several samples, also reported in Table 3, that are representative of the experimentally performed optimization. This optimization has been guided by the trial and error procedure using the CODE simulation. At first glance, Fig. 5(a) shows that very small changes in the multilayer structure produce apparent differences in the optical response. Typically, to quantify those changes, solar absorptivity (α Sol ) and thermal emissivity (ε th ) magnitudes are very useful to know the fraction of absorbed solar energy and the energy lost through thermal emission. Table 3 gives the obtained α Sol and ε 25 1C values for each prepared stack, as well as for spectral data of selective coatings presented in the literature that approximate receivers developed by ENEA and Schott [5,37].“S10”and “S14”stacks have excellent optical characteristics, it should be noted that “S10”has a higher solar absorptivity and may be considered as an optimized coating for low temperature applications and “S14”presents outstanding values of both solar absorptivity and thermal emissivity, which suggest its use for high temperatures applications. Values of α Sol and ε 25 1C , obtained for the “S14”selective coating, give a high selectivity ratio defined by ξ 25 1C ¼α Sol /ε 25 1C ¼54.1. Moreover, at a temperature of 600 1C, the ratio ξ 600 1C ¼8.5 also remains higher than the values obtained for the ENEA (ξ 600 1C ¼7.7) and Schott (ξ 600 1C ¼5.7) selective coatings, although the latter is designed for lower temperatures. It should be commented that this high selectivity ratio is based on two optical characteristics of the multilayer coating: the abrupt transition between the low and high reflectance regimes and the wavelength tunability of this transition. Additionally, the lower achievable value of emissivity is determined by the IR-mirror layer nature. For example, in the “S14”sample, the absorber cermet system does not appreciably increase the silver emissivity allowing maximum values for selectivity ratio when the low-high transition is placed at the appropriate wavelength. 3.4. Optical and thermal analysis of a solar field In addition to calculating solar-weighted absorptivity and thermal emissivity, a full optical and thermal analysis is performed using the detailed 3D heat-transfer model described elsewhere [38]. This model allows a comprehensive evaluation of the potential of the selective coatings in a solar field of a PTC system because the conflicting effects of solar absorptivity and thermal emissivity are both included. Various performance parameters, such as heat loss and heat gain, as well as optical and thermal efficiencies of the solar field are assessed for a location in Seville, Spain. For the operating conditions, weighted yearly average values Table 2 Summary of the RBS results of Mo–Si 3 N 4 cermet layers deposited on Si substrates and “S10”selective coating stacks on air-annealed stainless steel (SSth) substrate. Sample Subst. Ag IR-mirror Thickness (nm) Mo–Si 3 N 4 (HVMF) Mo–Si 3 N 4 (LVMF) Si 3 N 4 (AR-layer) FF (%) Thickness (nm) RBS (%) FF (%) Thickness (nm) RBS (%) Thickness (nm) RBS (%) MoSN–F10 Si –10 50 Mo: 9.570.5 Si: 36.073.0 N: 54.573.0 MoSN–F13 Si –13 54 Mo: 13.070.5 Si: 35.073.0 N: 5273.0 MoSN–F15 Si –15 49 S10 SS-th 175 37 50 Mo: 37.4 20 60 Mo: 19.8 80 Si: 37 Si: 20.9 Si: 26.9 N: 53 N: 41.7 N: 53.3 Fig. 5. (a) Reflectance spectra of several selective coating stacks reported in Table 3. (b) Reflectance spectra of “S14”coating after annealing at several temperatures. E. Céspedes et al. / Solar Energy Materials & Solar Cells 122 (2014) 217–225222 are used for the direct normal irradiance (DNI) I DNI ¼619 W/m 2 and for the solar incidence angle at the concentrator aperture θ Sol ¼30.11. These values represent a design point that produces thermal efficiencies similar to the expected yearly average performance. More detailed information on the used heat transfer model has been reported elsewhere [39]. A single loop of the solar field consists of 48 concentrator modules in series that heat the heat-transfer fluid (HTF) from 290 1C at the inlet to a maximum temperature at the outlet. Two solar field configurations with different HTFs are analyzed: a state-of-the-art system using a commercial molten salt [40] and an outlet temperature of 550 1C, and a current generation system using synthetic oil [41] with a temperature limit of 390 1C. The thermal efficiency of the solar-field loop is defined as the net useable thermal power(netheatgainQ gain of the HTF minus the power required to pump the HTF) divided by the DNI over the total aperture area of the concentrator mirrors [39], η th ¼Q gain P pump =η el I DNI A ap ð4Þ where P pump is the required pumping power for fully-developed flow [42] (determined using a correlation for the friction factor [43]), and η el ¼32.7% is the product of typical efficiencies of the power block, the electrical generator, and the HTF pump [44,45]. The optical efficiency is defined as the total solar power absorbed at the absorber tube divided by the DNI over the aperture area, η opt ¼Q gain þQ loss I DNI A ap ð5Þ where Q loss is the total heat loss from the absorber tube, including radiative and convective heat transfer to the glass envelope, as well as conduction through the supporting structure [38].Inthepresent analysis, the selective coatings are also compared to existing coatings used in commercial PTC plants. Spectral reflectance data is used to approximate the behavior of a commercial Schott receiver [36] and a high-temperature receiver developed by ENEA [5].Allothersystem parameters of the solar field are identical to those listed in Ref. [39]. More information on these parameters and the heat-transfer model can be found in Ref. [38]. Using the spectral reflectance curves of the selective coatings and the spectral data obtained from literature, the solar-weighted absorptivity and the thermal emissivity are calculated at 25 1C, 400 1C, and 600 1C. In Table 3, the calculated values for α Sol and ε th are shown for all coatings. It is seen that the Schott coating displays the highest solar absorptivity of 0.957. Coating candidate “S14”has an absorptivity that is slightly below that of the ENEA coating, both designed for higher temperatures than the coating by Schott. The high-temperature coatings ENEA and “S14”also display more favorable emissivity values at higher temperatures compared to the Schott receiver, with “S14”having a lower emissivity than ENEA. In Table 4, results are shown for the analysis performed assuming a current generation PTC system, using synthetic oil as the HTF and an outlet temperature of 390 1C. The heat gain and therefore the overall thermal efficiency of the different coatings vary considerably since Q loss and η opt depend strongly on the spectral behavior. For this configuration, the Schott receiver displays the best thermal efficiency with η th ¼63.3%. The lower temperatures reduce the heat losses from the absorber and hence increase the importance of the solar absorptivity for the overall thermal efficiency. The high absorptivity of the Schott receiver thus enables a high thermal performance. This effect also equalizes the performance of the ENEA and “S14”coatings, both reaching virtually identical thermal efficiencies. In Table 5, results are shown for the thermal analysis of the state-of-the-art PTC system, using molten salt and a high HTF outlet temperature of 550 1C. Of all the coatings, “S14”displays the best performance with a thermal efficiency of η th ¼59.1%, which is 0.5% higher than that of the ENEA coating. The lower solar absorptivity of “S14”is compensated by the reduced heat losses due to the overall lower thermal emissivity compared to the ENEA coating. The worst performance is observed with sample S2, as the low optical efficiency leads to a very low heat gain that cannot be compensated even by the lowest thermal emissivity of all coatings. 3.5. High temperature stability of deposited selective coatings The practical use of a selective coating at high temperatures is determined by two important factors. The first one is based on the chemical stability of the compounds and the multilayer structure forming the stack, and the second one is based on the thermal efficiency, as thermal emissivity is strongly dependent on temperature. In order to test the stability of the selective coatings reported here, we have carried out thermal annealing under moderate vacuum (10 2 mbar) to simulate working conditions of coated tubes. The subsequent optical characterization has allowed establishing any possible deterioration of the stacks, since thermal Table 3 Solar absorptivity (α sol ) and thermal emissivity (ε th )at251C, 400 1C, and 600 1C, calculated from experimentally measured reflectivity R(λ) and the blackbody emission curve in the 0.25–4μm and 0–1ranges, respectively, for several coating stacks on air-annealed stainless steel and spectral data from literature. Coating IR-mirror Ag Thickness (nm) Mo–Si 3 N 4 (HMVF) Mo–Si 3 N 4 (LMVF) Si 3 N 4 AR layer Thickness (nm) α Sol ε 25 1C ε 400 1C ε 600 1C FF (%) Thickness (nm) FF (%) Thickness (nm) S2 175 20 55 14 60 55 0.687 0.017 0.022 0.037 S5 175 37 50 18 55 80 0.903 0.024 0.089 0.170 S10 175 37 50 20 60 80 0.941 0.028 0.14 0.250 S13 175 32 50 14 55 75 0.866 0.011 0.039 0.075 S14 175 37 50 20 55 60 0.926 0.017 0.055 0.109 S19 175 37 52 18 57 60 0.905 0.010 0.049 0.108 ENEA 0.939 0.022 0.071 0.122 Schott 0.957 0.013 0.087 0.169 Table 4 Results of the heat transfer analysis for a state-of-the-art solar field, using molten salt with a maximum HTF temperature of 550 1C. Stack Q gain (MW) Q loss (MW) η opt (%) η th (%) S2 1.018 0.024 48.7 47.4 S5 1.280 0.079 63.5 59.6 S10 1.291 0.122 66.0 60.1 S13 1.267 0.038 61.0 59.0 S14 1.342 0.051 65.1 62.4 S19 1.318 0.044 63.7 61.3 ENEA 1.345 0.067 66.0 62.6 Schott 1.361 0.077 67.2 63.3 E. Céspedes et al. / Solar Energy Materials & Solar Cells 122 (2014) 217–225 223 degradation due to unstable microstructure causes a decrease in solar selectivity. Fig. 5(b) shows the evolution of the spectral reflectance measured at room temperature after several thermal annealings. In order to release possible stresses in the films, an initial warm up to 450 1C during 15 h was performed with a slow heating rate (2 1C/min). After this first annealing, subsequent thermal cycles were performed up to different temperatures during 15 h with heating rates of 4 1C/min (no dependence on heating rates has been observed). Reflectance spectra depict a very slight change, similar to the previously observed [46,47], and calculations of the optical characteristics show an excellent stability performance. After all those annealing processes performed under moderate vacuum, a maximum variation of 0.003 is obtained for both solar absorptivity (α Sol ¼0.92670.003) and for thermal emissivity at room temperature (ε 25 1C ¼0.01770.003) of coating “S14”. After annealing the “S14”stacks at 640 1C, the variation of emissivity at room temperature might be considered as negligible, but emissivity obtained for 600 1C becomes ε 600 1C ¼0.154, which represents an important variation with respect to the emissivity evaluated for the as-prepared sample (ε 600 1C ¼0.109). The slight change of reflectance after vacuum annealing may be due to a very small variation of layer thickness forming the absorber stack. In order to investigate this assumption, we have performed RBS on “S14”stacks samples after annealing at different temperatures. Fig. 6 shows the signal corresponding to Mo and Ag components obtained in RBS experiments performed with low energy He þ incident ions to maximize the resolution. No variation is observed up to annealing temperatures higher than 500 1C with respect to the as-deposited samples. Since width and intensity of RBS-signal are related to layer thickness and atomic density, respectively, the slight variation obtained after annealing at 600 1C has to be explained by a slight shrinkage of layers. In this way, the micro-structural stability of cermet and IR-mirror is confirmed, and it is evidenced that reflectivity variations obtained after annealing are due to small modification of the optical interference between layers. Much more experimental work has to be done to explain the origin of the observed shrinkage at 600 1C. Nevertheless, it is clear that annealing usually brings about desorption of the typically occluded gas during sputtering deposition. According to the measured RBS-signals, the observed decrease is about 6%. “S14” and “S19”as-prepared samples are 4% different in cermet thickness; their reflectance spectra reported in Fig. 5(a) show the change due to that difference, which is in good agreement with the observed variation from the as-prepared to the 600 1C annealed “S14”samples spectra reported in Fig. 5(b). This is a very important experimental fact that should be taken into account for further optimization of the selective stacks at high temperature. By initially optimizing the effective layer thickness according to the shrinking factor of the annealed coatings, enhanced thermal efficiency at 600 1C, together with important beneficial effects, can be obtained. As stated before, the small variations observed in the reflectance spectra after the annealing processes do not have any appreciable influence on the optical parameters evaluated at room temperature. Nevertheless, such reflectance variation has to modify the high temperature coating behavior. By using Eq. (2) and in the assumption of a temperature independent R(λ) spectra, the thermal emissivity evolution may be easily calculated at different temperatures, to have a complete picture of this Mo–Si 3 N 4 based selective coating. Fig. 7 shows thermal emissivity values as a function of the working temperature for the “S14”coating. Calculations have been made from room temperature up to the annealing temperature by considering the spectra of Fig. 5(b). It should be stated that this commonly used method to evaluate thermal emissivity at high temperatures has been appropriately validated by measuring the spectral emissivity of similar selective coatings, to determine the thermal emissivity dependence at high temperatures [48]. Additionally, the emissivity value obtained at 82 1C with a commercial emissometer is also given, which does not show variation after the vacuum annealing performed at those temperatures. At high temperatures, a small increase of thermal emissivity may be observed, but nevertheless, it should be commented that results are not modified after some repetitive annealing at the same temperature. 4. Conclusions A new solar absorber cermet, based on Mo–Si 3 N 4 , has been investigated. The procedure carried out to develop a suitable selective solar coating and to evaluate the performance potential in a solar system is presented. Selective coating optimization has Table 5 Results of the heat transfer analysis for a current generation solar field, using synthetic oil with a maximum HTF temperature of 390 1C. Stack Q gain (MW) Q loss (MW) η opt (%) η th (%) S2 0.992 0.050 48.7 46.3 S5 1.151 0.208 63.5 53.8 S10 1.087 0.325 66.0 50.8 S13 1.213 0.091 61.0 56.7 S14 1.265 0.129 65.1 59.1 S19 1.241 0.121 63.7 58.0 ENEA 1.255 0.157 66.0 58.6 Schott 1.232 0.206 67.2 57.4 Fig. 6. RBS data of “S14”coating on stainless steel substrate after annealing at several temperatures. Experiments were made with 1.0 MeV He þ incident ions. Fig. 7. Thermal emissivity dependence of the “S14”coating after several annealing processes at different temperatures (calculations are shown up to the temperature value at which the annealing was performed). E. Céspedes et al. / Solar Energy Materials & Solar Cells 122 (2014) 217–225224 included: (i) the cermet intrinsic optical properties (nand κ) ellipsometric characterization, (ii) simulation of the reflectance spectrum of stacks formed by silver IR-mirror, double cermet absorber and on top a Si 3 N 4 AR layer, (iii) solar absorptivity and thermal emissivity optimization by changing metal filling factor and layer thickness of simulated stacks, (iv) deposition of multilayer stacks within a narrow range of metal filling factor and layer thicknesses, (v) experimental characterization of reflectance to test final results and correct from a possible deviation with respect to the simulation optimization. Additionally, UV–vis and FTIR absorption spectroscopy investigations of the stacks before and after consecutive annealings indicate thermal stability above 600 1C. After this procedure, promising high solar absorptivity (α Sol ¼ 0.92670.003) and low thermal emissivity (ε 25 1C ¼0.017 and estimated value at 600 1Cisε 600 1C ¼0.109) have been obtained, leading to a selectivity ratio at high temperature that is among the highest reported ones. Furthermore, the analysis of the optical and thermal performance of the new cermet through simulations of a high temperature PTC system yielded overall thermal efficiencies that are higher than the results achieved with comparable, existing selective coatings. Acknowledgments This work has been supported by European Commission under HITECO Collaborative project contract no. 256830 from the FP7ENERGY program. References [1] C.E. Kennedy, Review of Midto High-Temperature Solar Selective Absorber Materials, NREL/TP-520-31267, National Renewable Energy Laboratory, Colorado, July 2002. [2] N. Selvakumar, H.C. Barshilia, Review of physical vapor deposited (PVD) spectrally selective coatings for midto high-temperature solar thermal applications, Sol. Energy Mater. Sol. Cells 98 (2012) 1–23. [3] Q.-C. Zhang, D.R. Mills, High solar performance selective surface using bisublayer cermet film structuresSol. Energy Mater. Sol. Cells 27 (1992) 273–290. [4] Y. Yin, R.E. Collins, Optimization and analysis of solar selective surfaces with continuous and multilayer profiles, J. Appl. Phys. 77 (1995) 6485–6491. [5] S. Esposito, A. Antonaia, M.L. Addonizio, S. Aprea, Fabrication and optimization of highly efficient cermet-based spectrally selective coatings for high temperature, Thin Solid Films 517 (2009) 6000–6006. [6] A. Antonaia, S. Esposito, Method for Producing a Cermet-Based Spectrally Selective Coating for the Receiver Tube of a Solar Concentrator, and Material thus Obtained, World Intellectual Property Organization, WO2009/107157 A2, 2009. [7] Q.-C. Zhang, D. Mills, A. Monger, Thin Film Solar Selective Surface Coating, US Patent No. 5523132A, 1996. [8] Q.-C. Zhang, Y. Yin, D.R. Mills, High efficiency Mo–Al 2 O 3 cermet selective surfaces for high-temperature application, Sol. Energy Mater. Sol. Cells 40 (1996) 43–53. [9] T. Kuckelkorn, W. Graf, C. Hildebrandt, A. Georg, Radiation Selective Absorber Pipe, Absorber Pipe with Said Coating, and Method of Making Same, US Patent No. 0121225 A1, 2008. [10] Q.-C. Zhang, Metal–AlN cermet solar selective coatings deposited by direct current magnetron sputtering technology, J. Phys. D:Appl. Phys. 31 (1998) 355–362. [11] Q.-C. Zhang, Recent progress in high-temperature solar selective coatings, Sol. Energy Mater. Sol. Cells 62 (2000) 63–74. [12] S. Yue, S. Yueyan, W. Fengchun, High-temperature optical properties and stability of Al x O y –AlN x –Al solar selective absorbing surface prepared by DC magnetron reactive sputtering, Sol. Energy Mater. Sol. Cells 77 (2003) 393–403. [13] H.C. Barshilia, N. Selvakumar, K.S. Rajam, D.V. Sridhara Rao, K. Muraleedharan, A. Biswas, TiAlN/TiAlON/Si 3 N 4 tandem absorber for high temperature solar selective applications, Appl. Phys. Lett. 89 (2006) 191909-1–191909-3. [14] H.C. Barshilia, Spectrally selective NbAlN/NbAlON/Si 3 N 4 tandem absorber for high-temperature solar applications, Sol. Energy Mater. Sol. Cells 92 (2008) 495–504. [15] H.C. Barshilia, V.K. William-Grips, K.S. Rajam, Solar Selective Coating Having Higher Thermal Stability Useful for Harnessing Solar Energy and a Process for the Preparation Thereof, US Patent No. 7585568B2, 2009. [16] N. Selvakumar, N.T. Manikandanath, A. Biswas, H.C. Barshilia, Design and fabrication of highly thermally stable HfMoN/HfON/Al 2 O 3 tandem absorber for solar thermal power generation applications, Sol. Energy Mater. Sol. Cells 102 (2012) 86–92. [17] M. Vila, E. Román, C. Prieto, Electrical conduction mechanism in silicon nitride and oxy-nitride-sputtered thin films, J. Appl. Phys. 97 (2005) 1137101–113710-6. [18] N.S. Jacobson, Corrosion of silicon-based ceramics in combustion environments, J. Am. Ceram. Soc. 76 (1993) 3–28. [19] M. Vila, D. Cáceres, C. Prieto, Mechanical properties of sputtered silicon nitride thin films, J. Appl. Phys. 94 (2003) 7868–7873. [20] M. Gauvin, E.E. Fullerton, F.E. Talke, Investigation of Fe–Si–Nfilms as magnetic overcoat for high density recording disk drives, J. Appl. Phys. 108 (2010) 063925-1–063925-5. [21] H.C. Barshilia, B. Deepthi, A.S. Arun Prabhu, K.S. Rajam, Superhard nanocomposite coatings of TiN/Si 3 N 4 prepared by reactive direct current unbalanced magnetron sputtering, Surf. Coat. Technol. 201 (2006) 329–337. [22] E. Céspedes, C. Prieto, R. Escobar, J.A. Sánchez García, Covering that Selectively Absorbs Visible and Infrared Radiation, and Method for the Production Thereof, Spanish Patent No. PCT/ES2012/070444, World Intellectual Property Organization, WO2012172148 A1, 2012. [23] N.D. Telling, S.J. Guilfoyle, D.R. Lovett, C.C. Tang, M.D. Crapper, M. Petty, Evidence of roughness distributions and interface smoothing in Co/Cu multilayers deposited under energetic particle bombardment, J. Phys. D: Appl. Phys. 31 (1998) 472–481. [24] A. Climent-Font, F. Pászti, G. García, M.T. Fernández-Jiménez, F. Agulló, First measurements with the Madrid 5 MV tandem accelerator, Nucl. Instrum. Methods Phys. Res. B 219–220 (2004) 400–404. [25] E. Kotai, Computer methods for analysis and simulation of RBS and ERDA spectra, Nucl. Instrum. Methods Phys. Res. B 85 (1994) 588–596. [26] W. Theiss, Hardand Software for Optical Spectroscopy, 〈www.mtheiss.com〉, 2013 (accessed: May 2013). [27] G.E. Jellison Jr., Optical functions of silicon determined by two-channel polarization modulation ellipsometry, Opt. Mater. 1 (1992) 41–47. [28] E.D. Palik, Handbook of Optical Constants of Solids, Academic Press, New York, 1985. [29] J.C. Maxwell-Garnett, Colours in metal glasses and in metallic films, Phil. Trans. R. Soc. Lond. A 203 (1904) 385–420. [30] D.A.G. Bruggeman, Berechnung verschiedener physikalischer Konstanten von heterogenen Substanzen, Ann. Phys., Leipzig 24 (1935) 636–679. [31] D.J. Bergman, The dielectric constant of a composite material –a problem in classical physics, Phys. Rep. 44 (1978) 377–407. [32] G.W. Milton, Bounds on the complex dielectric constant of a composite material, Appl. Phys. Lett. 37 (1980) 300–302. [33] R. Fuchs, S.H. Liu, Sum rule for the polarizability of small particles, Phys. Rev. B 14 (1976) 5521–5522. [34] D. Langbein, Normal electric modes at small polyhedral particles, J. Phys. A: Math. Gen. 10 (1977) 1031–1041. [35] C. Pecharromán, F.J. Gordillo-Vázquez, Expansion of the spectral representation function of a composite material in a basis of Legendre polynomials: experimental determination and analytic approximations, Phys. Rev. B 74 (2006) 035120-1–035120-10. [36] C. Pecharromán, F. Estebán-Betegón, J.F. Bartolome, S. Lopez-Esteban, J.S. Moya, New percolative BaTiO3–Ni composites with a high and frequency-independent dielectric constant (ε r E80000),Adv.Mater.13(2001)1541–1544. [37] N. Benz, W. Graf, Z. Hacker, C. Hildebrandt, M. Möllenhoff, J. Schulte-Fischedick, K. Silmy, Advances in receiver technology for parabolic troughs, in: Proceedings of the 14th International SolarPACES Symposium, Las Vegas, 2008. [38] M. Wirz, M. Roesle, A. Steinfeld, Three-dimensional optical and thermal numerical model of solar tubular receivers in parabolic trough concentrators, J. Sol. Energy Eng. 134 (2012) 041012-1–041012-9. [39] M. Wirz, M. Roesle, A. Steinfeld, Design point for predicting year-round performance of solar parabolic trough concentrator systems, J. Sol. Energy Eng. 136 (2013) 021019-1–021019-9, http://dx.doi.org/10.1115/1.4025709. [40] Coastal Chemical Company, LLC, HITEC s Heat Transfer Salt, 〈http://www. coal2nuclear.com/MSR%20-%20HITEC%20Heat%20Transfer%20Salt.pdf〉,2012 (accessed: October 2012). [41] Solutia Inc., Therminol VP-1 –Vapor Phase/Liquid Phase Heat Transfer Fluid, Technical Bulletin 7239115C, St. Louis, 2008. [42] F.P. Incropera, D.P. DeWitt, T.L. Bergman, A.S. Lavine, Fundamentals of Heat and Mass Transfer, 6th Edition, John Wiley & Sons, New York, NY, 2006. [43] T.K. Serghides, Estimate friction factor accurately, Chem. Eng. 91 (1984) 63–64. [44] A.M. Patnode, Simulation and performance evaluation of parabolic trough solar power plants (Master thesis), University of Wisconsin-Madison, Madison, WI, 2006. [45] G. Manzolini, A. Giostri, C. Saccilotto, P. Silva, E. Macchi, Development of an innovative code for the design of thermodynamic solar power plants part A: code description and test case, Renew. Energy 36 (2011) 1993–2003. [46] H.G. Craighead, R.E. Howard, J.E. Sweeney, R.A Buhrman, Graded-index Pt– Al 2 O 3 composite solar absorbers, Appl. Phys. Lett. 39 (1981) 39–41. [47] Y. Yin, Y. Pan, L.X. Hang, D.R. McKenzie, M.M.M. Bilek, Direct current reactive sputtering Cr–Cr 2 O 3 cermet solar selective surfaces for solar hot water applications, Thin Solid Films 517 (2009) 1601–1606. [48] I. Setién-Fernández, T. Echániz, L. González-Fernández, R.B. Pérez-Sáez, E. Céspedes, J.A. Sánchez-García, L. Álvarez-Fraga, R. Escobar-Galindo, J. M. Albella, C. Prieto, M.J. Tello, First spectral emissivity study of a solar selective coating in the 150–600 1C temperature range, Sol. Energy Mater. Sol. Cells 117 (2013) 390–395. E. Céspedes et al. / Solar Energy Materials & Solar Cells 122 (2014) 217–225 225