Comparison of Selective Transmitters for Solar Thermal Applications

ROBERT A. TAYLOR,1,2,* YASITHA HEWAKURUPPU,1 DREW DEJARNETTE,3 TODD P. OTANICAR3 1School of Mechanical and Manufacturing Engineering—The University of New South Wales, Gate 14, Barker St., Sydney, Australia, 2052 2School of Photovoltaics and Renewable Energy Engineering—The University of New South Wales, Gate 14, Barker St., Sydney, Australia, 2052 3Department of Mechanical Engineering, The University of Tulsa, 800 S. Tucker Dr., Tulsa, USA, 74104 *Corresponding author: [email protected] Received 10 March 2016; revised 13 April, 2016; accepted XX Month XXXX; posted X Month XXXX; published XX Month XXXX

OCIS codes: (310.6860) Thin films, optical properties, (310.7005) Transparent conductive coatings, (350.6050) Solar Energy.

ABSTRACT 1. INTRODUCTION Solar thermal collectors are radiative heat exchangers. Their efficacy is dictated predominantly by their absorption of short- An ideal selective solar component perfectly absorbs (or wavelength solar radiation and, importantly, by their emission of transmits) short wavelength radiation - e.g. for wavelengths long-wavelength thermal radiation. In conventional collector cut-off – while perfectly reflecting long designs, the receiver is coated with a selectively absorbing wavelength radiation – e.g. for wavelengths cut-off. This cut- surface (Black Chrome, TiNOx, etc.) which serves both of these betweenoff wavelength 0 and depends λ on operating temperature and the solar aims. As the leading commercial absorber, TiNOx consists of concentration ratio. ‘Selectivity’ is made possible> λ by the fact several thin, vapor deposited layers (of metals and ceramics) on a that terrestrial solar energy does not have a true blackbody metal substrate. In this technology the solar absorption to spectrum after passing through the atmosphere (e.g. after H2O thermal emission ratio can exceed 20. If a solar system requires and CO2 absorption). Thus, incident solar radiation has very an analogous transparent component – one which transmits the little infrared energy, enabling an energy balance ‘win’ with a full AM1.5 solar spectrum, but reflects long wavelength thermal cut-off absorber (or filter). Modern solar thermal receivers use emission – the technology is much less developed. Bespoke ‘heat thin, selectively absorbing/emitting coatings on the outer mirrors’ are available from optics suppliers at high cost, but the surface of an opaque absorber. Selective surfaces are most closest mass-produced, commercial technology is low-e glass. advantageous when the operation temperature is high, but not Low-e glasses are designed for visible light transmission and, as too high (e.g. 100-600oC). They are also best applied when the such, they reflect up to 50% of available solar energy. To address hot surface is encased in a vacuum package to minimize this technical gap, this study investigated selected combinations convective heat transfer loss and chemical oxidization/ of thin films which could be deposited to serve as transparent, degradation. When these conditions are met, the solar and the selective solar covers. A comparative numerical analysis of blackbody spectrums have little overlap and, importantly, feasible materials and configurations was investigated using a radiation heat loss is dominant. This is the case in evacuated non-dimensional metric, termed the ‘Efficiency Factor for tube collectors – a technology which represents the dominant Selectivity’ (EFS). This metric is dependent on the operation solar thermal collector on the market with approximately 350 temperature and solar concentration ratio of the system, so our million square meters installed in China alone [1]. This analysis covered the practical range for these parameters. It was technology operates well in the intermediate temperature found that thin films of indium tin oxide (ITO) and ZnS-Ag-ZnS range (with and without solar concentrating optics), and is provided the highest EFS. Of these, ITO represents the more well-suited to provide heat for industrial processes or to drive commercially viable solution for large-scale development. Based organic Rankine cycles. Since there is an immense demand for on these optimized designs, proof-of-concept ITO depositions 100-600oC heat (some 25-30% of the total energy usage in were fabricated and compared to commercial depositions. developed countries [2]), it is expected that this type of solar Overall, this study presents a systematic guide for creating a new thermal energy technology will increasingly displace class of selective, transparent optics for solar thermal collectors. conventional fossil fuels [3,4]. At present, the global installation rate of solar thermal collectors is about 9% per year [1], yielding a doubling in global installed capacity every 7- 8 years – signifying a large and rapidly growing global market.

A. Solar Selectivity cut-offs in the visible spectrum and are typically Fabry-Perot – based depositions which consist of 100s of layers of alternating Although selective absorbing surfaces are well developed high and low refractive index materials [6–8]. As an example, and very successful commercially, little research has gone into the Brightline ‘shortpass filter’ from Semrock has good long the development of analogous selective transparent wavelength cut-off for this application (1326 nm), but comes a components. If developed, these could be useful in similar solar poor shortwave cut-off at 700nm, which actually makes it a thermal applications, but would enable alternative receivers bandpass filter for solar applications [6]. Since these filters are which: a) employ volumetric (or direct) absorption receiver or sold in small quantities (for optics laboratories), their list price b) use simple, low-cost black absorbers with a selective heat is in the range of 0.1-1 Million USD/m2, depending the on mirrors to improve their performance. Thus, this type of supplier/product [6–8]. This is orders of magnitude out of the transparent component could facilitate a new class of low-cost range of solar collection technologies, which need to be << solar receivers that have a similar radiative exchange 1,000 USD/m2 of aperture area [9] [10]. performance with TiNOx. With this as our motivation, the current paper reports on feasible selective coatings for transparent substrates which allow transmission of solar energy, but reflect infrared emission. 1. Selective Absorbers (Commercial) At present, the state-of-the-art commercial selective absorption material is TiNOx, which consists of multiple thin layers on top of a relatively thick metal substrate [1]. A proprietary multi-step method involving advanced vapor deposition techniques is used to produce these coatings. TiNOx is generally used on the inner (metal) tube of an evacuated tube receiver, where it is never exposed to oxygen. It is then sealed inside an outer (glass) tube, which generally stays at temperatures well below 100oC during operation. This outer tube is exposed directly to the ambient meaning that convection (not radiation) heat loss dominates. The outer glass tube can be given either an anti-reflective coating or simply left Figure 1. Measured transmission spectra of commercial uncoated. It should be noted that the metal-to-glass seal of this transparent materials compared with the normalized direct technology is difficult and is the source of many failure modes. normal solar irradiance (DNI) spectrum (from [11])

2. Selective Transmitters (Commercial) B. Incorporating Transparent Selective Components There is one product on the market which is indeed Due to the fact that the transparent components of designed to be selective in transmitting sunlight and reflecting conventional solar collectors operate at low temperature and infrared light – low-e glass. Low-e glass is mass-produced and usually experience more convective than radiative heat loss, is rapidly replacing ordinary clear glass in buildings and there has been little impetus to develop selective transparent automobiles the world over [5]. Unfortunately, low-e glasses components. While selective solar transmitters are under- seek only to deliver maximum visible light and, in fact, minimize developed, they could be gainfully employed in the following overall solar heat gain (which is clearly not the goal of solar two applications. Note that for both selective surfaces are either thermal collectors). Low-e glasses reflect much of the solar not possible and/or would likely be of higher cost. spectrum which could be converted to useful heat. To demonstrate this, Figure 1 gives experimental measurements 1. Direct Solar Absorption using UV/Vis (from a Shimadzu UV-Vis 2600) and IR (Perkin With the advancement of nano-fabrication techniques, Elmer Spectrum Two) spectrophotometer data for a few direct absorption solar collectors are now possible to create at commercially available low-e glasses as compared to clear relatively low cost through ‘nano-engineering’ [12,13]. Direct glass. It can be seen that low-e glasses, if used as solar absorption (or volumetric absorption) systems need selective collectors, would reject 30-50% of the energy available in the solar transmitters since they will generally have a transparent AM 1.5 direct normal irradiance (DNI) spectrum. Figure 1 also component in which radiative heat losses are large. Aside from shows that normal ‘clear glass’ (window glass) also rejects the optical tunability afforded by these systems, one of the main ~20% of the solar spectrum, due to the fact that they contain advantages of going towards these absorbers is cost. It is iron and other impurities which absorb sunlight. Low-e glasses relatively cheap to mix nanoparticles in with common do involve advanced thin film coatings – such as magnetron (transparent) working fluids – gases, water, oils, glycols and sputter vacuum deposition – indicating that this type of even molten salts [3,14–20] at low volume/mass fractions (i.e. fabrication can be commercially viable. At present, 1 Billion < 0.01% by volume) [21,22]. Liquids are desirable for this square meters of low-E glass is produced annually [5]. application since they are much better heat carriers than gasses, but liquids have intrinsically high long wavelength 3. Selective Transmitters (Scientific) emissivity due to molecular stretching and bending. In fact, since the vast majority of solar collectors use a liquid heat It also is possible to obtain custom heat (or hot) mirrors transfer fluids [23], a simple substitution of a TiNOx tube with from scientific optical component suppliers. Unfortunately, a glass tube filled with a direct absorbing fluid (as has been even ‘stock’ items are very expensive and are rarely suitable as discussed in literature [15,24–31]), results in much higher selective solar transmitters. Most of the available products have thermal emission. To demonstrate this point, Figure 2 shows the EFS. This analysis zooms in on the cover/absorber system the long wavelength optical depth of water, propylene glycol and its radiative exchange with the environment. All other (PG), and Therminol 55 (synthetic heat transfer oil). Since these energy transfer (conduction/convection/heat removal) is fluids have long wavelength optical depths exceeding 3 (e.g. neglected since these would be common to all designs (or, in >95% of a blackbody emittance in Figure 2). It should be noted the case of convection, suppressed with vacuum insulation). that while there are other fluids that are used in solar thermal Thus, the EFS only considers radiative properties of the cover collectors, most have similarly high absorption/emission in the and the solar and blackbody radiation to and/or through it. wavelengths shown in Figure 2. Thus, without a solar selective This significantly simplifies the analysis and focusses it on transmitter, direct or volumetric absorption systems are salient radiative exchange parameters. doomed to be much less efficient selective surface absorbers.

Opaque

Figure 3. Generalized cover configuration – top right boxed inset shows tube receivers also have (near-normal) incidence Transparent

In the configuration of Figure 3, the absorber is assumed to be a black body absorber and emitter at all wavelengths (e.g. ). Solar radiation falls on this thin, generalized collector Figure 2. Long wavelength optical depth of pure liquids/heat at normal incidence, which is a reasonable assumption for transfer fluids, as measured with FTIR (using a 1 cm path length cell) α/εconcentrating ~ 1 systems with tracking. It should be noted that the assumption of near-normal incidence is valid for low 2. Black Absorbers concentration flat receivers and for tube receivers in which An alternative scenario is one where incorporating a concentrated radiation is distributed around their curvature selective surface, such as TiNOx, is untenable. In this case, it (see inset of Figure 3). Diffuse radiative emission from the may be desirable to utilize a lower-cost, black surface absorber absorber is either: i) transmitted through the cover (e.g. lost to (e.g. Pyromark paint) and then cover it with a solar selective the environment), ii) reflected back to the absorber (e.g. re- transmitter. This could prove advantageous if the total cost of absorbed), or iii) absorbed by the cover. Meanwhile, the cover incorporating the selective absorber is higher than a selective can also emit black body radiation to the environment in accordance with its temperature. Note that glass can also be transmitter. This is possible on either: a) a pure $/m2 considered a black body at long wavelengths – see the long comparison or b) due to geometric or material compatibility wavelength side of Figure 1. It should also be noted that the considerations. An example for b) is to choose glass-to-glass cover also emits thermal radiation back to the absorber, but (rather than metal-to-glass) vacuum seals. this quantity is not tracked since it is re-absorbed. With these Either direct absorption system or a simplified black assumptions, the EFS for the cover can be defined as [32]: absorber could benefit from a selective transmitter, if a 4 4 transparent component is operating at temperatures between GC τ −τ σT − E σT EFS = r SOL BB ABS BB cov er (1) 100 and 600oC. As these solar thermal receivers become more GC prevalent [13], this technology is in critical need of r development for high radiative exchange efficiency. In this equation, G is the incoming solar irradiance and Cr is the concentration ratio. These terms are constants, and help to 2. METHODS normalize the EFS for different solar resources and collector designs (since they appear together as a product in the To develop selective solar transmitters, this paper denominator). The terms τSOL and τBB are the total conducted a numerical and experimental analysis of readily transmittance of solar and black body radiation, respectively, available materials which can be used to cost-effectively which represent design variables of the selective cover design. formulate these devices. In order to compare among possible Note that blackbody emission (from a 100-600oC absorber) and options, we first defined a key performance metric, the solar radiation (a ~5500oC source) have opposing propagation ‘Efficiency Factor for Selectivity’ ( ). Then, we selected EFS directions and very little in common in terms of their spectrum. candidate materials and computed/compared their EFS values. The temperatures, TABS and Tcover, relate to the absorber and A. Efficiency Factor of Selectivity (EFS) cover, respectively. The absorber temperature is fixed in solar applications by controlling flow conditions. Although Figure 3 In this analysis, ‘selectivity’ was defined most simply shows separation between the cover and the absorber, the EFS through radiative heat exchange. Figure 3 shows a generalized here was calculated for situations where they are coupled collector configuration which serves as our baseline to define thermally – i.e. TABS = Tcover – as is likely the case in direct absorption systems. In practice, the EFS is applicable to any we limited the scope of materials to those which are readily cover temperature equal or less than that of the absorber, but available and suitable for solar applications. Using the EFS as a the condition TABS = Tcover, give a good indication of ranking mechanism enabled us to determine the advantage, if performance in the worst case scenario of emissive losses by any, of these materials over a bare glass substrate. the cover. EBB is the total cover emittance, which represents the long wavelength design variable for transparent, selective 1. Cover materials covers. Under these assumptions, the EFS gives a simple, non- Commercial solar collectors use a variety of transparent dimensional parameter which can be used to compare a wide materials, but glass generally provides the best long term range of options, while still capturing all the important performance and reliability. For the purposes of this research, technical features of a selective transparent optical component. we used low-iron glass was used as a substrate [37] [33]. This It can be seen in Figure 3 that solar radiation is falling normally boro-aluminosilicate glass was chosen because it has high on the front side of the cover. This is a reasonable transmission of the solar spectrum (see Figure 1). This specific approximation for i) non-concentrated solar collectors, ii) glass is also mass produced for displays, so it can be considered concentrated systems with focal distances >> aperture width, a well-known, relatively low-cost, base material. and iii) linear concentrators with tube receivers. Additionally, for solar applications, we can simplify the design space further 2. Coating materials by assuming common parameters for these designs in Table 1. Since the goal was to produce something commercially Table 1. Parameters common to all designs competitive with TiNOx, complex thin films which require Parameter Value several deposition steps are unlikely to be cost-competitive. Substrate (thickness) Low-Iron Glass (2 mm) [33] Thus, thin film-based interferometers which are produced with Surrounding medium Air, Vacuum (n=1, k=0) alternating high and low refractive index layers (e.g. the Operating temperature, 100 to 600oC (100oC increments) Brightline filter mentioned above) were not chosen for this TABS = Tcover work due to their high fabrication cost. Although this category Concentration ratio, Cr 10 and 100 of material can achieve effective optical control through needle optimization and other techniques, it requires very pure As can be seen in Table 1, the operating temperature and materials and 10s to 100s of layers of well-controlled concentration ratio range used here should span a wide range thicknesses to obtain the desired properties [38]. Eliminating of the intermediate collector technologies, and as such covers Fabry-Perot based thin films, there are two remaining viable the vast majority of installed solar thermal collectors. options. These were classified as ‘Category 1’ and ‘Category 2’ films. Category 1 films consist of a dielectric-metal-dielectric B. Calculation Method sandwich, in which the materials and the thicknesses of the Using the EFS equation means that we must calculate (or three layers each represent a degree of design freedom. measure) the ‘effective’ spectral optical properties of the Category 2 materials are single layer dispositions of doped substrate and film together as a single unit. For this work we semi-conductors, namely, transparent conductive oxides. To have prepared our own code in Matlab based on the the reduce infrared thermal emission, the selected semi-conductors theoretical relationships presented by Macleod [34]. When the must have much lower band gaps than silicon (1.1 eV). optical properties and dimensions of the thin films and 3. Category 1 Materials substrate are known, we can determine the directional and spectral transmittance, absorbance, and reflectance (and also The most common dielectric materials used for optics in emittance, using Kirchoff’s law) of the complete cover. The the literature are TiO2, ZnS, and Al2O3 [39,40]. Metals, such as directional quantities are integrated over their whole Ag, Au, Cu, Al are also commonly used, so these were selected as hemisphere. The spectral properties weighted with, and the middle layer due to their potential to transmit solar numerically integrated, over the solar (or temperature radiation and reflect black body radiation [41]. After some dependent black body) energy distributions. This process initial analysis, this work selected 8 different Category 1 thin yields the final scalar variables τSOL, τBB and EBB, for any film sandwiches, as describe in Table 2. The complex refractive desired material, operational, or geometric configuration. indexes of these materials can be found in [35]. To model these radiative properties we need to know the Table 2. Category 1 thin film materials complex refractive index. For metals and dielectric materials, Film No. Materials Film No. Materials this is somewhat straight-forward, since the properties are readily available from optical property handbooks – e.g. [35]. 1 TiO2-Ag-TiO 2 5 ZnS-Ag-ZnS For semi-conductors, the complex refractive index is dependent 2 TiO2-Au-TiO 2 6 ZnS-Au-ZnS on the dopant level which is determines the free carrier 3 TiO2-Cu-TiO 2 7 ZnS-Cu-ZnS concentration. In this case, determination of the complex refractive index requires the classic Drude model and a 4 TiO2-Al-TiO 2 8 ZnS-Al-ZnS conversion from the dielectric function to the refractive index (via Kramers-Kronig dispersion relations [36]). Although the 4. Category 2 Materials refractive index can be altered by temperature and deposition method, these factors were not considered here. Optimization of Category 2 films, similarly, can only be carried out for a small subset of all possibilities. Only low C. Materials Selection energy band-gap materials were considered to match the required IR properties. Thus, oxides of In, Sn, Cd and Zn doped The aim of the current work was to find feasible, relatively with materials like Sb, F, Sn, B, Al and Ga were selected [42–45]. near-term, thin film materials for solar applications. As such, After some initial literature review, this left 8 feasible candidate thin films. Table 3 presents key details about these thin films, size was chosen to be small enough that convergence of an including the dopant range used as well as deposition type and optimum was achieved. The variation of free carrier the associated conditions. In addition, the effective electron concentration (ne) and DC resistivity (ρdc) for each Category 2 mass to free mass ratio (m*/m) and the high frequency material was obtained as a function of the doping concentration dielectric constant used for each thin film is also given in the from the corresponding sources listed in Table 3. The specific table as this information is needed to determine the optical values of ne and ρdc were calculated at by interpolating the properties. values found in literature for the current doping concentration. During the Matlab modelling process the concentration range Table 3. Details of the selected Category 2 thin films found in each reference was divided into 50 equally spaced Film # Type Doping Temp. m*/m ε(∞) units, which was found to achieve a converged result for the range (oC) best EFS value. Thus, EFS values for ~8,000 Category 2 films were calculated. It should be noted that thickness was assumed 1-Fluorine SP 0-180 490 – 0.45 3.7 to have no influence on the optical constants that and ne and doped tin F/Sn 525* ρdc and that these parameters are uniform across the thickness. oxide (at.%) [46] (FTO) 3. RESULTS: EFS RANKINGS AND DISCUSSION 2-Sb doped SP 0-12 400* See 4 tin oxide Sb /Sn [47] [48] [48] This section presents and explains the trends in ranking of (ATO) (at.%) the EFS the selected thin films. First off, it should be noted that 3-Cadmium DCRS 4-30 20-200 0.23 5.82 in many cases a bare glass substrate out performed these films, tin oxide O2(%) [49] [50] [50] depending on the temperature and concentration ratio (CTO) conditions. In general, thin films were found to out-perform 4-Tin doped PLD 0-15 250 0.4 3.57 [5 bare glass especially at low solar concentration ratios and at indium SnO 2 [51] [52] 2] high operating temperatures. For these situations it is very oxide (ITO) (wt.%) important to control losses, and a small penalty in solar 5-Boron SGSC 0-1.0 Anneal at transmission is acceptable. At high concentration ratios and doped zinc B (at.%) 450 low temperatures, however, it was sufficient to have bare glass oxide (BZO) [43] that allows more solar radiation to transmit to the absorber 6-Al doped AP- 0.2-0.7 400 and, in comparison, the amount of thermal energy loss is zinc oxide CVD Al (at.%) [45] negligible. As the temperature increases, having a thin film

(AZO) became increasingly important relative to bare glass. 0.28 3.85 7-Gallium AP- 0.5-10 370 [51, [51, 52] A. Category 1 Material Results doped zinc CVD Ga [44] 52] oxide (GZO) (at.%) Our analysis revealed that that film 5 (ZnS-Ag-ZnS) was the 8-Fluorine SP 0.2-1.0 425* clear winner among Category 1 thin films since it achieved the doped zinc F/Zn [53] highest EFS at all operating conditions compared to the other oxide (FZO) films. However, it was found that bare glass produced higher * Substrate temperature selected to maximize solar transmittance and EFS for high concentration ratios and low operating DC conductivity [Abbreviations: SP – Spray Pyrolysis, DCRS – DC temperatures. Film 1 (TiO2-Ag-TiO2) was found to be the next Reactive sputtering, PLD – Pulsed Laser Deposition, SGSC – Sol-Gel Spin best choice for most conditions. Interestingly, both of these Coating, AP-CVD – Atmospheric Pressure Chemical Vapour deposition films used Ag as the metal. This is because Ag has a low refractive index in the solar spectral region compared to the 5. Analysis Process other metals, resulting in a higher transmission [41]. The thin The variables, τSOL, τBB and EBB, and the resulting EFS, films based on Au and Cu generally had lower EFS values than were determined for the 16 films shown in Tables 2 and 3 the Ag-based coatings, and Al -based thin films generally across the range of operation temperatures and concentration performed the worst. In fact, Al -based Category 1 materials ratios mentioned in Table 1. For Category 1 materials, the metal only had a higher EFS than bare glass at 500-600oC. layer thickness was varied from 1 to 30 nm, in increments of 1 nm. The upper limit of 30 nm was enforced because a thicker Among the dielectric layers, ZnS had the highest EFS and metallic layer will create too much solar absorption. For the was thus identified as the best amongst the three materials lower limit, 1 nm was chosen. The thicknesses of the dielectric considered here. Alumina, Al2O3, was found to be the next best, layers were allowed to vary from 2 to 200nm, in 2 nm although the difference in the EFS between TiO2 and Al2O3 increments. It should be noted thinner films are less likely to based films was typically less than 0.01, so choosing between suffer delamination and may be more stable, but very thin, the materials is more a matter of price than performance. uniform films are exceedingly difficult to fabricate. This range Trends of film layer thickness were similar across all four again covers the dimensions used in previous works that Category 1 materials. When the operating temperature considered these Category 1 materials, which have been increased, the thickness of the metal layer was increased to considered in the literature [41,54,55]. Thus, EFS values for optimize reflection of the long wavelength thermal energy. At ~24,000 Category 1 films were calculated – a calculation which low temperatures a very thin metal layer was advantageous as takes a few days on a standard desktop computer. more solar radiation can be transmitted. In the best films for For Category 2 films, the films thicknesses were varied in each metal, the thickness of the dielectric layer also increased the range of 50 to 1000nm, at 50nm intervals. While this range with increasing temperature to match the thickness of the covered the thicknesses used in previous literature, the interval metal layer. This can be explained by the fact that the better films had improved solar transmittance through index matching. When the concentration ratio (Cr) was increased, the what was seen in Category 1 films. For instance, the optimal thickness of the optimized metal layer was reduced. In this case layer thickness of the ITO films monotonically increased with it was beneficial to transmit more solar radiation. In this case, increasing temperature, but for the other materials it increased the dielectric layer thickness also followed the trend of metal to a point, but then reduced with increasing temperature. The layer thickness to achieve good solar transmittance. Overall, it trends with dopant concentration were even less clear, since was concluded that among the materials studied Ag in a ZnS the optimal dopant concentration of some films increased with sandwich provided the best Category 1 EFS value. Table 4 temperature while others remained constant. These can shows the top three ranked candidates for all conditions. roughly be tracked back to trade-off in τSOL versus TBB versus EBB. As a summary, Table 5 shows the top three ranked Table 4. Category 1 film numbers (corresponding to Table Category 2 candidates for all conditions. 2) and EFS values are given in each cell. Dielectric and metal film thicknesses are given in parenthesis, in nanometers. Table 5. Category 2 film numbers (corresponding to Table 3) and EFS values are given in each cell. Film thicknesses (nm) Rank Cr = 10 and dopant levels (from Table 3) are in parenthesis.

100oC 200oC 300oC 400oC 500oC 600oC Cr = 10 1 #5, #5, #5, #5, #5, ZnS- #5, ZnS- Rank ZnS- ZnS-Ag- ZnS- ZnS- Ag-ZnS, Ag-ZnS, 100oC 200oC 300oC 400oC 500oC 600oC Ag- ZnS, Ag-ZnS, Ag- 0.31 0.13 # 4, # 4, # 4, # 4, # 4, ZnS, 0.71 0.56 ZnS, (28/11) (28/16) Bare ITO, ITO, ITO, ITO, ITO, 1 glass 0.70 0.63 0.52 0.39 0.20 0.79 (10/3) (18/5) 0.47 0.77 (150/ (200/ (250/ (300/ (350/ (4/1) (26/8) 4.06) 4.06) 4.06) 5.94) 5.94) # 6, # 1, # 1, # 1, # 1, 2 #7, #1, #1, #6, #6, ZnS- #6, ZnS- # 4 AZO, FTO, FTO, FTO, FTO, 2 0.76 0.67 0.57 0.44 0.24 -0.06 ZnS- TiO2- TiO2- ZnS- Au-ZnS, Au-ZnS, (50/5.94) (250/ (200/ (300/ (350/ (450/ Cu- Ag- Ag- Au- 0.26 0.10 0.2) 180) 180) 180) 180)

ZnS, TiO2, TiO2, ZnS, (34/12) (34/19) Cr = 100 0.78 0.69 0.56 0.42 Rank (4/1) (2/2) (4/4) (32/8) 100oC 200oC 300oC 400oC 500oC 600oC Rank Cr = 100 # 4, # 4, 100oC 200oC 300oC 400oC 500oC 600oC Bare Bare Bare Bare ITO, ITO, 1 glass glass glass glass 0.71 0.66 0.85 0.84 0.81 0.76 (100/ (150/ 1 Bare Bare Bare #5, #5, ZnS- #5, ZnS- 4.06) 4.06) # 4, # 4, # 4, # 1, # 1, Glass Glass Glass ZnS- Ag-ZnS, Ag-ZnS, # 4, ITO, ITO, ITO, ITO, FTO, FTO, 0.80 0.85 0.84 0.81 Ag- 0.71 0.64 2 0.79 0.77 0.74 0.68 0.63 (50/ (50/ (50/ (50/ (150/ (150/ ZnS, (6/2) (10/3) 4.06) 4.06) 4.06) 5.94) 180) 180) 0.77 (4/1) C. Inter-Material Comparison 2 #5, #5, #5, Bare #1, #1, The best Category 1 film was only slightly better (e.g. <

ZnS- ZnS-Ag- ZnS- Glass TiO2- TiO2- +0.05 EFS) than the best Category 2 film at low temperatures (<220oC) for a concentration ratio of 10. At the higher Ag- ZnS, Ag-ZnS, 0.76 Ag-TiO 2, Ag-TiO 2, concentration ratio (100), Category 2 films start to outperform ZnS, 0.83 0.81 0.69 0.61 Category 1 films by a small amount (e.g. < +0.05 EFS), but only 0.84 (4 /1 (4/1) (2/2) (2/3) at high temperature (> 500oC). Figure 4 compares the optimized EFScover of the two best Category 1 (ZnS-Ag-ZnS) and (4 /1) /4) Category 2 (ITO) thin film materials under different operating conditions. The performance of the bare low iron glass substrate was also added to Figure 4 to show areas where thin B. Category 2 Material Results films have good performance. These results indicate that In Category 2 thin films, film 4 (ITO [56]) consistently Category 1 films are slightly better solar transmitters, but ended up with the highest EFS. The FTO film [53] and the AZO Category 2 films are better black body reflectors. Since film [52] were found to have provide the next best EFS values. Category 1 film depositions are more complex (3 thin layers Other than this, Category 2 materials are more difficult to versus 1 thicker layer) and require expensive equipment, this compare since their optical properties are subjective to modest performance boost is unlikely to be economic. deposition methods and dopants. For the Category 2 films, the trends in their optimal layer thickness were not as uniform as them is small, and ITO films are easier to produce (possible with sol-gel processing and liquid coating methods), ITO was considered to be a more likely commercially candidate and was thus chosen as a material for further experimental analysis. D. Other Considerations Several other factors were considered with respect to the applicability of these films. Due to space limitations these can only be mentioned briefly. 1. Coating Side An important element in these designs is the choice of which side of the substrate (or both) to do the deposition. Although this aspect will not be discussed in detail here, we checked all Category 1 and 2 films for deposition on the front (facing the incident solar radiation), back (facing the absorber), and both sides. It was found that front-sided coating performed best due to the fact that emission from the glass substrate can be minimized (which is not possible with a back-side Figure 4. ZnS-Ag-ZnS, ITO and bare glass EFS - dashed blue lines deposition). Additionally, the front side coating was found to be show the crossing points between Zns-Ag-ZnS and ITO slightly better than a double sided coating, due to improved Breaking the EFS down into τsol, TBB, EBB allows an in- solar transmission. Thus the following analysis will focus on depth consideration as to the underlying mechanism for front-sided disposition. achieving better EFS values. Table 6 summarizes these values for both film types at four operating conditions that represent 2. Temperature the regions where each film dominates. The film with the In this study, the temperature of the transmitter was highest EFS at each condition is marked with an asterisk. assumed to be the same as the absorber, which can be regarded Overall the results in Table 6 show that the difference in as a worst-case performance scenario. However, this is not performance can be tracked to differences in their capability to always true in practice. Reducing the cover temperature transmit solar radiation and suppress emissive loses similarly, increases the performance of the DAC as emission by the film’s with the transmission of black body radiation being relatively substrate reduces. This, in turn, directly affects the EFS. To small. explore this we analyzed situations where Tcover = Z*TABS, where 0 < Z < 1. It was found that if the cover temperature is Table 6. Solar weighted absorbance (τSOL), blackbody weighted reduced drastically (to Z < 0.5), bare glass performs better than transmittance (TBB) and emittance (EBB) of the best films the vast majority of the thin films. For smaller reduction of the

Operating Film cover temperature (e.g. Z = 0.8), the EFS can change slightly (by τSOL TBB EBB Conditions material up to 0.05) and the optimal layer thickness can alter substantially (by up to than 50%). Thus, it is important to use Cr = 10 * ZnS-Ag-ZnS 0.845 0.007 0.530 the correct cover temperatures in any design for best Case 1 Tabs = performance. 100oC ITO 0.803 0.004 0.408 Cr = 10 ZnS-Ag-ZnS 0.655 0.024 0.145 3. Incidence Angle Case 2 Tabs = 400oC ITO* 0.710 0.008 0.156 The incident angle of solar radiation is a secondary parameter that directly affects the solar transmittance, and Cr = 100 * ZnS-Ag-ZnS 0.845 0.100 0.604 thus the EFS. In a real absorber – even a tubular absorber – Case 3 Tabs = sunlight does comes from a range of angles (rather than the 400oC ITO 0.803 0.064 0.448 normal incidence assumed above), depending on the design. = 100 Cr ZnS-Ag-ZnS 0.793 0.138 0.347 However, little angular variation was found from 0-30 degrees Case 4 = Tabs by analyzing the solar weighted transmission at a variety of 600oC ITO* 0.761 0.063 0.237 angles for the selected films. Beyond 35 degrees, a sharp cut-off * Indicates the more efficient film in transmission was found due to total internal reflection. By At Cr = 100, neither of the thin films perform well as 45 degrees, less than 1% of the incident sunlight will transmit selective covers until high temperatures were reached. This is through to the absorber. It should be noted that this is also true fortuitous since higher concentration ratios and higher for bare glass, so the thin films studied herein are only slightly operation temperatures are linked. When Tabs was 400-500oC, less tolerant of large incidence angles. In the real application, ZnS-Ag-ZnS was the best, but the best ZnS-Ag-ZnS film has a this 30 degree incident angle limit is unlikely to pose a serious very thin metal layer (1 to 2 nm). As such it has high solar issue since evacuated tube receivers are cylindrical, with light transmittance, but is much harder to fabricate that the thicker incident at near normal angles around the tube. ITO film. For Case 4, where Cr = 100 and Tabs = 600oC, ITO 4. Surrounding Medium (Refractive Index) outperformed ZnS-Ag-ZnS. This results from the ITO reducing the emissive losses (see the EBB values). ZnS-Ag-ZnS may be the The optical properties of the media surrounding the cover best choice for low temperatures, but at higher temperatures were set to be non-participatory (air or a vacuum) in this ITO has less emissive losses. Since the EFS difference between analysis. However, a liquid or another medium may be present on one or both sides of the selective transmitter. When a liquid spin rate was achieved. The substrate was then spun for 90 is present (e.g. water or oil), the difference in refractive indices seconds at the designated speed. The 1 mL volume was chosen between the glass and back surrounding reduces, resulting in to ensure complete coverage of the substrate was achieved, less solar reflectance at the boundary. This net increase in τSOL although a significant volume of excess was spun off (e.g. is, unfortunately, balanced (and sometimes exceeded) by an wasted) at all speeds. [Note: A uniform 100 µm coating on a 4 increase in the long wavelength emission through the cover. As inch wafer requires ~0.8 mL] The coated wafer was allowed to a characteristic example, if an ITO cover is exposed to a air dry afterwards. For the dip-coated samples, the substrate synthetic heat transfer oil one side (for Cr = 10 and Tabs = was dipped into a bath of the ITO gel and then removed. These 250oC), the EFS reduces from 0.668 to 0.665. In this instance, samples were then placed on a hot plate at 50oC in a fume τSOL is improved from 0.756 to 0.79, but the EBB is also cabinet for approximately 30 minutes to ensure the ITO film increased considerably from 0.219 to 0.32. was fully dried. This method provides a thicker (than spin coating), double sided coating. All samples were placed in both 5. Substrate Thickness UV/Vis (Shimadzu UV-Vis 2600) and IR (Perkin Elmer The thickness of the substrate can affect the performance Spectrum Two) spectrophotometers (all detectors having an of the cover. To determine the influence of this we held all other accuracy of less than 0.01% of the full parameters constant and varied the substrate thickness from 0 transmittance/reflectance scale) to determine the transmission to 20mm. It was found that, mainly due to absorption in the and reflection spectra. substrate, the EFS reduced nearly linearly with thickness. B. Experimental Results 4. FABRICATION AND EXPERIMENTS It was found that with this fabrication method, high spin In this paper, ITO films were prepared ‘in-house’ through speeds produced coatings which only marginally changed the sol-gel chemical methods and deposited using spin and dip base substrate. Slow spin speed produced better results since coating. It should be noted, however, that there are other the thickest layers of ITO had the largest impact on the methods of fabrication (notably spray coating), that are transmission of the bare substrate. Furthermore, the dip-coated expected to be scalable for (low-cost) larger scale production samples were found to have a transmission spectrum very close o after the concept is explored here. The results of these films to the ideal cut-off for a selective surface operating at 300 C were compared to commercially available methods which with a Cr of 10. Figure 5 shows the best results achieved for the usually rely on vapor deposition techniques. spin and dip coated ITO samples from this study as compared with an off-the-shelf commercial ITO sample (from MTI [57]). A. Fabrication method To produce ITO films, indium nitrate hydrate (In(NO)3:xH2O, Sigma Aldrich) and tin acetate (Sn(CH3CO2)2, Sigma Aldrich) were used as precursors, using a 10% basis of tin. Additionally, acetic acid (AA) (CH3COOH, Sigma) and ethylene glycol (EG) (C2H4(OH)2, VWR) were used together as a complexing and esterification agent in the mixture. The molar ratio between the cations and the organic additives was 1: 1.5. Next, stoichiometric amounts of the indium nitrate and tin acetate and the AA and EG were dissolved in de-ionized water and kept on a hot-plate at 50oC to make sure they were homogenous, as well as to evaporate the solvent. After 3 hours of heating, a low viscosity gel was formed. The gel was then heated to 400oC for 3 hours so that it would undergo calcination. This step produced a powder which was then Figure 5. Sol-gel ITO (> 1µm film thickness) transmission cooled. The powder was pulverized and mixed with ethanol spectrum compared to commercial ITO and a bare substrate (EtOH) and de-ionized water containing 3% polyvinyl alcohol (VWR) at a 1:1 ratio. The final solution was used without It can be seen that the dip-coated Eagle XG borosilicate further treatment for coating the substrates. The substrates substrate transmitted at approximately 90% until 2,700 nm, used in this study were 4 inch wafers of optically flat, 1 mm where its transmission dramatically drops. As a transparent thick borosilicate glass (Corning Eagle XG), which had even selective transmitter, this represents a good cut-off wavelength higher transmission than the model assumptions above. Before where the window would ideally begin to reflect IR light of deposition, the substrates were thoroughly cleaned with dish longer wavelengths. The thickness of the dip-coated sample soap, rinsed with distilled water, and the rinsed with ethanol. was estimated to be >1µm, which is larger than the ranking After cleaning they were dried with compressed air prior to results from Section 3. This was likely due to the fact that the film deposition. As noted above, two methods of deposition free carrier concentration was lower in the ‘in-house’ samples were chosen – spin and dip coating. Since the spin or dip rates as compared to the modelled optimal design. While the can significantly affect the layer thickness and uniformity, a transmission spectrum of this sample appears to be ideal, parametric study of various spin speeds ranged from 1,000 to Figure 6 shows that the reflection spectrum for all the samples 10,000 rpm was chosen along with a range of dip speeds from tested across the infrared region. Figure 6 also shows a 0.1 to 10 cm/s was used. Together these two methods covered commercial sample of ITO, which was ordered from Geomatec most of the feasible range of film thicknesses. During spin Co., Ltd – a custom Japanese supplier of optical coating. From coating, a syringe loaded with 1 mL of the solution was quickly the results of our Matlab optimization study, we provided deposited onto the center of the substrate after the specified Geomatec with the designed film parameters and they produced samples which were much closer to the desired thickness and dopant concentration (e.g. with a final product Future experimental work is still required, however, to which has 30 Ohm/sq.). It was observed in Figure 6 that these obtain an ITO sample with the correct cut-off wavelength, samples were indeed, very reflective at long wavelengths. which has anti-reflective properties at short wavelengths, and reflects well at long wavelengths. Table 7 show the resulting properties of the best ‘in-house’ ITO films compared to the two commercial films. In the table it can be seen that at low concentration ratios the ‘in-house’ films had a negative EFS, while at high concentration ratios they performed even better than the commercial samples. This is due to the fact that they in-house films had high solar transmission, but poor black body emission. It should be noted that the substrate transmission for the ‘in-house’ films was better than the modelled substrate, yielding a higher τSOL than the model. It should also be noted that since we are using fixed films (rather than optimizing) τSOL, TBB, and EBB do not change with concentration ratio. Table 7. Comparison of commercial versus ‘in-house’ ITO films for 400oC operation temperature and two concentration ratios

Figure 6. Reflection spectra of various ITO coated glass substrates (the ‘Geomatic’ sample used a thicker substrate) Film EFS, EFS, τSOL TBB EBB material Cr = 10 Cr = 100 1. Anti-reflective coatings 0.49 0.67 It can be seen even in Figure 1 that while the transmission ITO MTI 0.69 0.005 0.17 of substrate is high, there was still a ~8% reflection of the ITO Geomatec 0.8 0.007 0.25 0.50 0.77 incoming solar energy. Compared to a TiNOx absorber (which -0.19 0.76 has <5% solar reflection), this represents a large optical loss, ITO Dip Coat 0.87 0.01 0.91 and necessitates an anti-reflective coating. There are many ITO Dip w/ Silica 0.88 0.01 0.91 -0.18 0.78 options for this, but recent studies have indicated that a thin layer of silica nanoparticles – i.e. approximately one particle diameter thick – can be used to reduce the reflection of glass 5. CONCLUSIONS substrates [58]. For this application, it is proposed that this coating could be deposited on the outside of the ‘selective’ A wide range of selective transmitter materials was studied coating. Thus, aqueous phase nanoparticles (produced using for feasibility as selective solar transmitters. A simple metric, the Stöber method [58]) were spin cast onto the dip-coated the EFS, was proposed to rank the material combinations ITO substrates to investigate whether a simple, low-cost anti- across a wide range of temperatures and for typical solar reflective coating was achievable. A wide variety of spin speeds concentration ratios. A wide range of known dielectric-metal- were tested, but the best result was achieved at 10,000 rpms, as dielectric ‘Category 1’ films and transparent conductive oxide is shown in Figure 7. Figure 7 indicates that this simple method ‘Category 2’ films were analyzed. The highest EFS candidates had a marginal reduction (~2%) in reflectivity of the substrate (among some 32,000) were found for each category. A ZnS-Ag- in the short wavelength range of incoming solar energy. In ZnS film was found to be the best Category 1 film and an ITO comparison with commercial samples of ITO, our dip-coated coating was found to be the best Category 2 film. Since ZnS-Ag- ITO and our dip-coated ITO with a thin silica nanoparticle outer ZnS required tight, nanometer-scale layer thickness tolerances, layer, proved to have lower reflection. ITO was identified as a leading candidate for low-cost production for solar applications. From this finding, selected ‘in-house’ production (laboratory) methods were experimentally evaluated, revealing that it is indeed possible to create ITO coatings using (scalable) sol-gel chemistry and dip coating, which have an appropriate transmission cut-off. The long wavelength reflection of these samples was not in agreement with the model prediction, so there is scope for future work in using low-cost fabrication methods to achieve high reflection of long wavelength black body radiation. This research could prove to have significant socio-economic impacts [59]. Additionally, this paper concludes that commercial ITO (custom and stock) coatings, if chosen well for the application, can achieve high long wavelength reflection and have EFS values in line with the EFS values found in the optimized model of this study. It was also found that an anti- reflective coating (for short wavelengths) could help improve ITO further, and to approach the performance TiNOx. A simple Figure 7. Short wavelength comparison of the ‘in-house’ and thin film of silica nanoparticles (fabricated using high speed commercial ITO formulations with the Eagle XG substrate spin coating) was found to be a feasible solution for this challenge, although the anti-reflectivity enhancement was only R. A. Taylor, "Multifunctional Core-Shell Nanoparticle a couple percent from the ‘in-house’ films tested here. Overall, Suspensions for Efficient Absorption," J. Sol. Energy the present work is a key step towards defining and realizing Eng. 135, 021005 (2013). next-generation, low-cost direct (or simple black) absorbers 17. R. A. Taylor, P. E. Phelan, T. P. Otanicar, C. A. Walker, M. which can be used to provide 100-600 oC heat for industrial Nguyen, S. Trimble, and R. Prasher, "Applicability of applications. nanofluids in high flux solar collectors," J. Renew. Sustain. Energy 3, 1–8 (2011). 6. References 18. R. A. Taylor, T. P. Otanicar, Y. Hewakerrppu, F. Bremond, G. Rosengarten, E. Hawkes, X. Jiang, and S. 1. F. Mauthner and W. Weiss, Solar Heat Worldwide: Coulombe, "Feasibility of Nanofluid-Based Optical Markets and Contribution to the Energy Supply (2015). Filters," Appl. Opt. , 1413–1422 (2013). 2. C. Brunner, Solar Heat Integration in Industrial 52 19. R. A. Taylor, P. E. Phelan, T. Otanicar, R. J. Adrian, and R. Processes: 2014 Highlights (2016). 3. T. P. Otanicar, R. A. Taylor, and C. Telang, S. Prasher, "Vapor generation in a nanoparticle liquid "Photovoltaic/thermal system performance utilizing suspension using a focused, continuous laser beam," thin film and nanoparticle dispersion based optical Appl. Phys. Lett. 95, 161907 (2009). 20. Y. L. Hewakuruppu, L. A. Dombrovsky, C. Chen, V. filters," J. Renew. Sustain. Energy 5, 033124 (2013). 4. A. Mojiri, R. Taylor, E. Thomsen, and G. Rosengarten, Timchenko, X. Jiang, S. Baek, and R. A. Taylor, "Spectral beam splitting for efficient conversion of "Plasmonic “pump – probe” method to study semi- solar energy—A review," Renew. Sustain. Energy Rev. transparent nanofluids," Appl. Opt. 52, 6041–6050 (2013). 28, 654–663 (2013). 21. T. P. Otanicar, P. E. Phelan, R. A. Taylor, and H. Tyagi, 5. Presswire, "Global LOW-E glass industry 2015 market research report," M2 Presswire (October 2015). "Spatially Varying Extinction Coefficient for Direct 6. Semrock, "BrightLine short-pass filter," Absorption Solar Thermal Collector Optimization," J. https://www.semrock.com/FilterDetails.aspx?id=FF01 Sol. Energy Eng. 133, 024501 (2011). -1326/SP-25. 22. S. Mesgari, S. Coulombe, N. Hordy, and R. A. Taylor, 7. Edmund Optics, "High Performance Hot Mirrors," "Thermal stability of carbon nanotube-based http://www.edmundoptics.com/optics/optical- nanofluids for solar thermal collectors," Mater. Res. mirrors/hot-cold-mirrors/high-performance-hot- Innov. 19, 650–653 (2015). mirrors/3150/. 23. N. Boerema, G. Morrison, R. A. Taylor, and G. 8. Thorlabs, "Hot and Cold Mirrors," Rosengarten, "Liquid sodium versus Hitec as a heat http://www.thorlabs.hk/NewGroupPage9.cfm?ObjectG transfer fluid in solar thermal central receiver roup_ID=6108. systems," Sol. Energy 86, 2293–2305 (2012). 9. A. Shirazi, R. A. Taylor, S. D. White, and G. L. Morrison, 24. A. Veeraragavan, A. Lenert, B. Yilbas, S. Al-Dini, and E. "A systematic parametric study and feasibility N. Wang, "Analytical model for the design of volumetric assessment of solar-assisted single-effect, double- solar flow receivers," Int. J. Heat Mass Transf. 55, 556– effect, and triple-effect absorption chillers for heating 564 (2012). 25. T. Yousefi, E. Shojaeizadeh, F. Veysi, and S. Zinadini, "An and cooling applications," Energy Convers. Manag. 114, 258–277 (2016). experimental investigation on the effect of pH variation 10. T. Otanicar, R. A. Taylor, P. E. Phelan, and M. of MWCNT–H2O nanofluid on the efficiency of a flat- Engineering, "Prospects for Solar Cooling – An plate solar collector," Sol. Energy 86, 771–779 (2012). Economic and Environmental Assessment," Sol. Energy 26. T. Yousefi, F. Veysi, E. Shojaeizadeh, and S. Zinadini, "An experimental investigation on the effect of Al2O3–H2O 86, 1287–1299 (2012). 11. J. A. Jeong, H.-K. Kim, and M.-S. Yi, "Effect of Ag nanofluid on the efficiency of flat-plate solar interlayer on the optical and passivation properties of collectors," Renew. Energy 39, 293–298 (2012). 27. R. Saidur, T. C. Meng, Z. Said, M. Hasanuzzaman, and A. Kamyar, "Evaluation of the effect of nanofluid-based Appl. Phys. Lett. 93, 033301 (2008). 12. R.flexible A. Taylor, and transparentS. Coulombe, Al2O3∕Ag∕Al2O3 T. P. Otanicar, P. multilayer," E. Phelan, A. absorbers on direct solar collector," Int. J. Heat Mass Gunawan, W. Lv, G. Rosengarten, R. S. Prasher, H. Tyagi, Transf. 55, 5899–5907 (2012). and Tyagi, "Small Particles, Big Impacts: A Review of 28. V. Khullar, H. Tyagi, N. Hordy, T. P. Otanicar, Y. the Diverse Applications of Nanofluids," J. Appl. Phys. Hewakuruppu, P. Modi, and R. A. Taylor, "Harvesting solar thermal energy through nanofluid-based 113, 011301 (2013). 13. P. Phelan, T. Otanicar, R. Taylor, and H. Tyagi, "Trends volumetric absorption systems," Int. J. Heat Mass and opportunities in direct-absorption solar thermal Transf. 77, 377–384 (2014). 29. Z. Said, M. H. Sajid, M. a. Alim, R. Saidur, and N. a. collectors," J. Therm. Sci. Eng. Appl. 5, 021003 (2013). 14. Y. L. Hewakuruppu, R. A. Taylor, H. Tyagi, V. Khullar, T. Rahim, "Experimental investigation of the Otanicar, S. Coulombe, and N. Hordy, "Limits of thermophysical properties of AL2O3-nanofluid and its selectivity of direct volumetric solar absorption," Sol. effect on a flat plate solar collector," Int. Commun. Heat Mass Transf. , 99–107 (2013). Energy 114, 206–216 (2015). 48 15. T. P. Otanicar, P. E. Phelan, R. S. Prasher, G. 30. E. Natarajan and R. Sathish, "Role of nanofluids in solar Rosengarten, and R. A. Taylor, "Nanofluid-based direct water heater," Int. J. Adv. Manuf. Technol. 1 (Special, 3–7 (2009). absorption solar collector," J. Renew. Sustain. Energy 2, 033102 (2010). 31. R. A. Taylor, P. E. Phelan, T. P. Otanicar, R. Adrian, and 16. W. Lv, P. E. Phelan, R. Swaminathan, T. P. Otanicar, and R. Prasher, "Nanofluid optical property characterization: towards efficient direct absorption "Bandgap and effective mass of epitaxial cadmium solar collectors," Nanoscale Res. Lett. 6, 225 (2011). oxide," Appl. Phys. Lett. 92, 022101 (2008). 32. R. A. Taylor, Y. Hewakuruppu, D. DeJarnette, and T. P. 51. H. Kim and C. M. Gilmore, "Electrical , optical , and Otanicar, "Fabrication and comparison of selective, structural properties of indium – tin – oxide thin films transparent optics for concentrating solar systems," in for organic light-emitting devices," J. Appl. Phys. 86, Proc. of SPIE (2015), Vol. 9559, p. 955905. 6451–6461 (1999). 33. MTI, "Corning Eagle XG Glass Substrates," 52. S. Lauz, N. Kaiser, A. Zoller, R. Gotzelmann, H. Lauth, http://www.mtixtl.com/CorningEAGLEXG- and H. Bernitzki, "Room-temperature deposition of 252507.aspx. indium tin oxide thin fims with plasma ion-assisted 34. H. A. Macleod, Thin-Film Optical Filters, 4th ed. (Taylor evaporation," Thin Solid Films 335, 1–5 (1998). & Francis, 2010). 53. A. Sanchez-Juarez, A. Tiburcio-silver, A. Ortiz, E. P. 35. E. D. Palik, Handbook of Optical Constants of Solids, Five- Zironi, and J. Rickards, "Electrical and optical Volume Set (Academic Press, 1997). properties of fluorine-doped ZnO thin films prepared 36. M. Fox, Optical Properties of Solids (Oxford University by spray pyrolysis," Thin Solid Films 333, 196–202 Press, USA; 2 edition, 2010). (1998). 37. Corning, "Corning EagleXG Material Informations," 54. J. C. C. Fan, "Transparent heat-mirror films of http://www.sydor.com/wp-content/uploads/Corning- EAGLE-XG-Display-Glass.pdf. insulation," Appl. Phys. Lett. 25, 693 (1974). 38. F. Crisostomo, R. A. Taylor, T. Zhang, I. Perez-Wurfl, G. 55. TiO2∕Ag∕TiO2J. C. Fan and F. for J. Bachner,solar energy "Transparent collection heatand radiationmirrors Rosengarten, V. Everett, and E. R. Hawkes, for solar-energy applications.," Appl. Opt. 15, 1012– "Experimental testing of SiNx/SiO2 thin film filters for 1017 (1976). a concentrating solar hybrid PV/T collector," Renew. 56. H. Kim and C. M. Gilmore, "Electrical , optical , and Energy 72, 79–87 (2014). structural properties of indium-tin-oxide thin films for 39. X. Liu, X. Cai, J. Qiao, J. Mao, and N. Jiang, "The design of organic light-emitting devices," J. Appl. Phys. 86, 6451– ZnS/Ag/ZnS transparent conductive multilayer films," 6461 (1999). Thin Solid Films 441, 200–206 (2003). 57. MTI, "ITO Coated Glass," 40. J. A. Jeong, H. K. Kim, and M. S. Yi, "Effect of Ag http://www.mtixtl.com/itocoatedglasssubstrates.aspx. interlayer on the optical and passivation properties of 58. B.-T. Liu and W.-D. Yeh, "Antireflective surface flexible and transparent Al2 O3 Ag Al2 O3 multilayer," fabricated from colloidal silica nanoparticles," Colloids Appl. Phys. Lett. 93, (2008). Surfaces A Physicochem. Eng. Asp. 356, 145–149 41. C. Lee, S. Chen, and C. Jaing, "Optical monitoring of (2010). silver-based transparent heat mirrors," Appl. Opt. 59. R. A. Taylor, P. E. Phelan, T. Otanicar, R. S. Prasher, and (1996). B. E. Phelan, "Socioeconomic impacts of heat transfer 42. B. Stjerna, E. Olsson, and C. G. Granqvist, "Optical and research," Int. Commun. Heat Mass Transf. 39, 1467– electrical properties of radio frequency sputtered tin 1473 (2012). oxide films doped with oxygen vacancies, F, Sb, or Mo," J. Appl. Phys. 76, 3797–3817 (1994). Funding Information. ARC DE160100131; ARENA, USO-1 MUSIC 43. V. Kumar, R. G. Singh, L. P. Purohit, and R. M. Mehra, "Structural, Transport and Optical Properties of Boron- Acknowledgment. R.A.T. gratefully acknowledges support from the doped Zinc Oxide Nanocrystalline," J. Mater. Sci. Australian Research Council’s DECRA program. RAT and T.O. Technol. 27, 481–488 (2011). acknowledge the financial support of the Australian Renewable Energy 44. J. Hu and R. G. Gordon, "Atmospheric pressure chemical Agency. vapor deposition of gallium doped zinc oxide thin films from diethyl zinc, water, and triethyl gallium," J. Appl. Phys. 72, 5381–5392 (1992). 45. J. Hu and R. G. Gordon, "Textured aluminum-doped zinc oxide thin films from atmospheric pressure chemical- vapor deposition," J. Appl. Phys. 71, 880–890 (1992). 46. A. E. Rakhshani, Y. Makdisi, and H. A. Ramazaniyan, "Electronic and optical properties of fluorine-doped tin oxide films," J. Appl. Phys. 83, 1049 (1998). 47. S. Shanthi, C. Subramanian, and P. Ramasamy, "Growth and characterization of antimony doped tin oxide thin films," J. Cryst. Growth 197, 858–864 (1999). 48. E. Shanthi, V. Dutta, A. Banerjee, and K. L. Chopra, "Electrical and optical properties of undoped and antimony-doped tin oxide films," J. Appl. Phys. 51, 6243–6251 (n.d.). 49. N. Miyata, K. Miyake, and S. Nao, "Physical properties of cadmium-tin oxide films deposited by d.c. reactive sputtering," Thin Solid Film 58, 385–389 (1979). 50. P. H. Jefferson, S. a. Hatfield, T. D. Veal, P. D. C. King, C. F. McConville, J. Z iga–P rez, and V. Mu oz–Sanjos ,

ú ñ é ñ é