transparent heat mirrors based on tungsten oxide–silver multilayer structures

7/16/2019 Transparent heat mirrors based on tungsten oxide–silver multilayer structures

http://slidepdf.com/reader/full/transparent-heat-mirrors-based-on-tungsten-oxidesilver-multilayer-structures 1/7

Transparent heat mirrors based on tungsten oxide–silvermultilayer structures

M.F. Al-Kuhaili *, A.H. Al-Aswad, S.M.A. Durrani, I.A. Bakhtiari

Physics Department, King Fahd University of Petroleum and Minerals, Dhahran 31261, Saudi Arabia

Received 15 October 2008; received in revised form 9 April 2009; accepted 18 May 2009Available online 13 June 2009

Communicated by: Associate Editor Darren Bagnall

Abstract

Transparent heat mirrors based on tungsten oxide/silver three-layer structures were fabricated using thermal evaporation. The opticaland morphological properties of the single layers were first investigated to serve as a basis for the fabrication of the heat mirrors. Onlysilver films with a thickness higher than 18 nm were found to be continuous. Subsequently, WO 3/Ag/WO3 multilayers were deposited,where the WO3 layers thickness was fixed at 35 nm, and the thickness of the silver layer was varied from 18 to 39 nm. The optical prop-erties of the multilayers were measured over the visible and near infrared ranges. These multilayers exhibited the desired heat mirrorbehavior, namely the transmittance was largely confined to the visible range and the reflectance was diminished in that range. The max-imum visible transmittance was 88.3% at 554 nm. Increasing the thickness of the silver films resulted in a decrease of the visible trans-mittance, with a corresponding increase in the infrared reflectance. Optimization of these two opposing trends was evaluated using afigure of merit, from which the best performance was obtained for multilayers with a silver layer of thickness of 24 nm.Ó 2009 Elsevier Ltd. All rights reserved.

Keywords: Heat mirror; Energy efficient coatings; Silver; Tungsten oxide

1. Introduction

A transparent heat mirror (THM) has high visible trans-mittance (400 < k < 700 nm) and high infrared (IR) reflec-tance (700 < k < 3000 nm), where k is the wavelength of light. These mirrors can be used in flat-plate collectorsfor solar heating and cooling, and on windows for the ther-

mal insulation of buildings (Fan, 1981). The basic materialemployed in the fabrication of a THM is a metallic thinfilm. As a result of their high free-electron density, metalshave high reflectance throughout the infrared and visibleranges. In particular, free-electron-like metals, such as sil-ver, have been selected for THM applications (Granqvist,1981; Lee et al., 1996; Karlsson et al., 1981; Valkonenet al., 1984; Lampert, 1981). These metals have their

plasma wavelength in the near ultraviolet because of thed–d transitions (Wang et al., 2006). The plasma wavelengthis close to the wavelength at which the reflectance is mini-mum. In order to suppress their visible reflectance, the met-als are used as thin films. The reflection coefficients of thetwo interfaces of a metal film are in opposite phases, andtherefore interfere destructively, resulting in minimum

reflectance (Lee et al., 1996). Then, the only losses aredue to absorption and spatial traversal of the film. For agiven thickness, these losses can be minimized by selectinga metal with low absorption and low refractive index in thevisible. To further enhance the visible transmittance of themetallic film, multilayer dielectric-metal coatings are used.A dielectric layer produces an antireflection effect whendeposited on the side of light incidence on the metal, andthus serves to increase the transmittance (Kostlin andFrank, 1982). The metallic film can also be embeddedbetween two anti-reflection dielectric layers (Fan, 1981;

0038-092X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.

doi:10.1016/j.solener.2009.05.006

* Corresponding author. Tel.: +966 3 860 3747; fax: +966 3 860 2293.E-mail address: [email protected] (M.F. Al-Kuhaili).

www.elsevier.com/locate/solener

Available online at www.sciencedirect.com

Solar Energy 83 (2009) 1571–1577

mailto:[email protected]

mailto:[email protected]



Granqvist, 1981). Such a three-layer structure allows broadband reflectance and flexibility in band pass selection(Berning, 1983).

Metallic thin films have a strong tendency to retain anisland structure until a substantial film thickness has beendeposited (Kostlin and Frank, 1982). Island formation

has a more dramatic effect on the IR reflectance than onthe visible reflectance (Karlsson et al., 1981). Moreover,island formation results in agglomeration and roughnessin the metallic layer, which reduce the visible transmittance(Lampert, 1981). In addition, single metallic films do notoffer the stability and durability required for practical heatmirrors (Lampert, 1981). This is due to environmentaleffects such as corrosion and oxidation (Karlsson et al.,1981; Valkonen et al., 1984; Lampert, 1981). The thicknessof the metallic layer has to be critically optimized. High IRreflectance necessitates the use of thick films. However, asthe thickness increases, the plasma edge shifts to lowerwavelengths, and the loss due to absorption increases. On

the other hand, very thin metallic films suffer from islandformation. Silver has superior properties for THM applica-tions. It has a plasma wavelength of 320 nm (Wang et al.,2006). Moreover, among metals, silver has the lowest visi-ble absorption (Karlsson et al., 1981). Its optical constantsin the visible (refractive index (n) = 0.06 and extinctioncoefficient (k ) = 3.59 at k = 550 nm (Johnson and Christy,1972)) perfectly match the requirements of a THM.

In a THM, the dielectric layers provide several advanta-ges. First, they act as anti-reflection coatings to enhance thereflectance of the metallic layer (Kostlin and Frank, 1982).Second, the dielectric layers improve the tunability of the

effective transition wavelength and range (Lampert,1981). Third, the top dielectric layer protects the metalliclayer from environmental effects such as abrasion and cor-rosion (Lampert, 1981). Finally, the bottom dielectric layerprovides a nucleation modification layer, which enhancesthe growth of a continuous thin film (Lampert, 1981; Kost-lin and Frank, 1982). The main criterion for the selection of the dielectric, for a given metal, is that the refractive indexof the dielectric should match substantially the extinctioncoefficient of the metal in the desired spectral range (Yol-das and O’Keefe, 1984). Higher refractive indices of thedielectric allow higher IR reflectance, while lower indicesare adequate for broader band transparencies (Pracchiaand Simon, 1981). The thickness of the dielectric is notthe conventional k/4 because the refractive index of thedielectric in the visible is larger than that of the metal(Fan et al., 1974).

In this work, tungsten oxide (WO3) was used as thedielectric. WO3 is the most extensively investigated mate-rial for electrochromic applications, including smart win-dows (Niklasson and Granqvist, 2007). In addition, WO3

thin films have found numerous device applications, suchas photovoltaic devices, photocatalysis, gas and biosensors,and optical data storage (Deb, 2008). Tungsten oxide hasthe basic requirements for incorporation into a THM,

namely its wide band gap of 3.25 eV renders it transparent

in the visible range. Moreover, its refractive index is suit-ably suited for this application.

We have carried out a detailed investigation of the effectof the metal thickness on the performance of WO3-basedTHM. Four sets of samples were investigated: two-layerWO3/Ag with a WO3 thickness of 35 or 70 nm; and

three-layer WO3/Ag/WO3 with a WO3 thickness of 35 or70 nm. In each set, five different thicknesses of the silverlayer were used. Only the three-layer structures with aWO3 thickness of 35 nm resulted in an acceptable heat mir-ror behavior. The remaining three types of samples showedinferior performance, and will not be discussed here.

2. Experiment

Multilayer WO3/Ag/WO3 coatings were deposited usingthermal evaporation. The starting materials were powdersof tungsten oxide (WO3) (Alfa Aesar, 99.8% purity) and sil-ver (Alfa Aesar, 99.999% purity). The films were prepared

in a Leybold L560 box coater pumped by a turbomolecularpump. The materials were slowly out-gassed before evapo-ration. The system was pumped to a base pressure of 5 Â 10À4 Pa. The coatings were sequentially evaporatedfrom molybdenum boats without breaking the vacuum. Itis known that the introduction of oxygen during evapora-tion leads to better stoichiometry of the oxide layer. How-ever, oxygen was not admitted during evaporation in ordernot to oxidize the silver layer. Moreover, in order to reducethe diffusion among the various layers, the substrates werenot heated. The substrates were rotating during the deposi-tion, and the source-to-substrate distance was 40 cm. The

evaporation rate (1.0 nm/s for silver, and 0.5 nm/s forWO3) and thickness of the films were controlled by aquartz crystal thickness monitor.

For different purposes of film characterization, singlefilms and multilayer coatings were simultaneously depos-ited on fused silica substrates (for optical and atomic forcemicroscopy (AFM) measurements), and tantalum sub-strates (for X-ray photoelectron spectroscopy, XPS). Nor-mal-incidence transmittance (T ) and reflectance (R), overthe wavelength range 200–2000 nm, were measured usinga Jasco V-570 double beam spectrophotometer. XPS wasperformed using a VG Scientific MKII spectrometer witha monochromatic Al Ka (1486.6 eV) X-ray source. Priorto the XPS analysis, the samples were transferred in airto the XPS analysis chamber. The C 1s peak of hydrocar-bon contamination, at a binding energy of 284.5 eV, wasused as an energy reference. During the XPS analysis, thesamples were maintained at ambient temperature at a pres-sure of 5 Â 10À7 Pa. Chemical depth profiles of the filmswere obtained by a sequence of etching followed by XPSmeasurement. Etching was performed using a 4-keV Ar+

ion beam. During depth profiling, the chamber pressurewas 10À5 Pa, and the ion current, at the sample surface,was about 1.2 lA. The surface morphology of the filmswas examined by tapping mode AFM (Veeco Innova diS-

PM). The sample surface was probed with a silicon tip of

1572 M.F. Al-Kuhaili et al. / Solar Energy 83 (2009) 1571–1577



10 nm radius oscillating at its resonant frequency of 300 kHz. The scan area was 2 Â 2 lm2, and the scan ratewas 1 Hz. The thickness of the films was measured usinga stylus surface profilometer (Ambios XP-2). In the multi-layer coatings, the thickness of the WO3 layers was 35 nm,and the thickness of silver was 11, 18, 24, 32, and 39 nm.

Finally, X-ray diffraction was performed and revealed thatall samples were amorphous.

3. Results and discussion

3.1. Single films

The surface morphology of the silver films is depicted inthe AFM micrographs shown in Fig. 1. The different filmgrowth modes can be inferred from the evolution of surfacemorphology with film thickness. When the thickness of thesilver film was 11 nm (Fig. 1a), the film was at the nucle-ation stage and the grain morphology was island-like.

However, as the silver film thickness increased, the nucleiand grains grew vertically in a continuous and uniformthree dimensional structure, as is shown for an 18-nm thicksilver film in Fig. 1b. The root-mean-square (RMS) rough-ness of this film was 3.32 nm. Thus, it can be concludedthat silver films thicker than 18 nm were continuous. Theseresults are in agreement with previous AFM investigationsof silver thin films, where the minimum thickness for a con-tinuous thin film was 12 nm (Fu et al., 1997). Conse-

quently, in the fabrication of multilayer coatings, onlysilver films thicker than 18 nm were considered. The nor-mal-incidence reflectance and transmittance spectra of thesilver films are shown in Fig. 2. The reflectance increasedas the thickness of the silver layer increased, whereas thetransmittance showed the opposite trend.

The morphology of the surface of a tungsten oxide film isshown in the AFMmicrograph of Fig.1c.Thefilmwasdensewith a smooth surface. The RMS roughness of this film was1.66 nm. The normal-incidence reflectance and transmit-tance spectra of a typical tungsten oxide thin film are shownin Fig. 3a. The film was transparent down to a wavelength of 300 nm. The optical constants (refractive index and extinc-tion coefficient) were derived from the reflectance and trans-mittance spectra using the spectrophotometric method(Denton et al., 1972). The opticalconstants of tungsten oxideare shown in Fig. 3b. The refractive index changes from2.084 at k = 400 nm to 1.965 at k = 700 nm. Most of trans-parent heat mirrors are optimized for a wavelength of

550 nm, where the refractive index of tungsten oxide is2.014. The extinction coefficient reaches a minimum valueof 0.012 at k = 550 nm. This indicates minimum absorptionin the visible range, which is a desirable property for thedielectric layers in a transparent heat mirror. However, theabsorption increased in the infrared region. This increasein absorption was attributed to the formation of an absorp-tion band by the defects originating from oxygen vacanciesin sub-stoichiometric films (Deb and Chopoorian, 1966).

Fig. 1. Three dimensional AFM images with a scan area of 2 Â 2 lm2: (a) a silver film with a thickness of 11 nm, (b) a silver film with a thickness of 18 nm,

(c) a WO3 film with a thickness of 35 nm.

M.F. Al-Kuhaili et al. / Solar Energy 83 (2009) 1571–1577 1573



Tungsten oxide thin films evaporated without an oxygenatmosphere were shown to be sub-stoichiometric (Leftherio-tis et al., 2001).

3.2. Transparent heat mirrors

The optical spectra of the WO3/Ag/WO3 transparent heatmirrors are shown in Fig.4 for different thicknesses of the sil-ver layer. The thickness of the WO3 layer was fixed at 35 nm.Detailed analysis, based on complex optical admittance, hasshown that the thickness of the dielectric layer must begreater than k/8nD, where nD is the dielectric’s refractiveindex (Berning, 1983). In our case, with nD = 2.014 atk = 550 nm, the minimum thickness of the dielectric shouldbe 34 nm.

The thickness of the silver layer plays a crucial role inthe operation of the heat mirror. A heat mirror with a verythin silver layer (Fig. 4a) did not exhibit the required selec-tivity. Selectivity was improved as the thickness of the sil-

ver layer increased. For thicker films (P24 nm), thefollowing observations can be made:

(i) The transmittance was largely confined to the visiblerange.

(ii) The band width of the transmittance narrowed as thethickness of the silver layer increased.

(iii) The maximum transmittance decreased as the thick-ness of the silver layer increased.

(iv) The infrared reflectance progressively increased asthickness of the silver layer increased.

(v) The minimum reflectance was obtained for a wave-

length around 500 nm. This wavelength shifted to lowervalues as the thickness of the silver layer increased.

These observations are consistent with the results onheat mirrors fabricated using other dielectrics, such asZnS (Leftheriotis et al., 1997) and TiO2 (Yoldas andO’Keefe, 1984). The maximum visible transmittance was88.3% at k = 554 nm, and was obtained for the heat mirrorwith a silver layer thickness of 24 nm. This value is amongthe highest reported values for a heat mirror.

The major factor affecting the stability and durability of a heat mirror is the diffusion of silver atoms across theinterfaces (Chiba et al., 1984). When a dielectric layer isdeposited on top of the silver layer, the silver surface isdamaged to some extent (Chiba et al., 1992). This inducesinstability in the silver layer and subsequent dissociation of silver atoms from the metallic layer (Chiba et al., 1992).The diffusion process reduces the effective thickness of the silver layer (Wang et al., 2007). The selectivity of heatmirrors depends on the sharp interfaces of the componentlayers, which is degraded by inter-diffusion (Fan et al.,1974).

The XPS technique was used to estimate the atomic con-centrations of the elements and their distributions. Thistechnique has several limitations. First, it is a surface tech-

nique that will probe only the top monolayers of the film,

Fig. 2. Visible transmittance and infrared reflectance of silver thin films.

Fig. 3. Optical spectra of a tungsten oxide thin film with d = 152 nm: (a)reflectance (R) and transmittance (T ) spectra, (b) refractive index and

extinction coefficient dispersion curves.




and thus cannot reveal accurate composition of the bulk of the film. Second, the accuracy of determining atomic ratiosusing this technique is 10% (Briggs and Seah, 1983). Third,ion bombardment during depth profiling can cause reduc-tion to lower stoichiometric and non-stoichiometric oxides(Leftheriotis et al., 2001). Thus, the XPS technique givesonly a qualitative analysis of the concentration of the ele-

ments and their distributions. Detailed XPS spectra inthe W 4f, O 1s, and Ag 3d core level regions were obtained.Consequently, the atomic concentrations of the elementswere calculated from these peaks, taking into account theatomic sensitivity factor of each peak (Briggs and Seah,1983). The oxygen/tungsten ratio for the top layer (priorto ion etching) was found to be 2.4, which is lower thanthe stoichiometric value of 3. This low value shows thatthe tungsten oxide layers were sub-stoichiometric, in accor-dance with the increase of the infrared absorption. Theoxygen O 1s spectrum was deconvoluted into two peaks.The firs peak was at a binding energy of 530.5 eV, and isattributed to oxygen due to the W–O bond in slightly

sub-stoichiometric tungsten oxide films (Leftheriotiset al., 2001). This peak accounted for 78.6% of the totalO 1s spectrum. Another smaller peak accounted for theremaining 21.4% of the O 1s spectrum, and occurred at abinding energy of 531.6 eV. This peak is attributed to oxy-gen adsorbed from the moisture in the surrounding atmo-sphere (Leftheriotis et al., 2001). Thus, this component is

an indication of the porosity of the tungsten oxide layers.XPS depth profiling was used to investigate the diffusion

of elements in our heat mirrors. Fig. 5 shows the depth pro-files of a heat mirror, with a silver thickness of 24 nm. Thefigure clearly shows that the interface of each layer was notsharp, and thus inter-diffusion of the tungsten oxide andsilver layers took place. Diffusion of silver into the tungstenoxide layers could take place because of the porosity of these layers. Diffusion of tungsten and oxygen into the sil-ver layer could take place because of the large free path of evaporated species compared to the small thickness of thesilver layer (Briggs and Seah, 1983). Therefore, the effectivethickness of the silver layer was significantly reduced, and

Fig. 4. Reflectance (R) and transmittance (T ) spectra of the heat mirrors (WO3/Ag/WO3/glass). The thickness of WO3 was kept at 35 nm. The thickness of the silver film is indicated in the figures. For comparison, the dashed line shows the 90% value of either quantity.




this would deteriorate its optical performance. One couldovercome this problem by depositing thicker silver layersand more compact oxide layers. However, this will comeat the expense of a sharp decrease in transmittance. Oneother solution is the introduction of ultrathin diffusion-bar-rier layers. Such layers require stringent control of thedeposition process and present a great technological chal-lenge (Wang et al., 2007).

The performance of a heat mirror is evaluated in termsof the integrated visible transmittance (T vis) and infraredreflectance (RIR), defined as:

T vis ¼

R /ðkÞT ðkÞd kR /ðkÞd k

fk : 400 – 700 nmg; ð1Þ

RIR ¼

R RðkÞd kR d k

fk : 700 – 2000 nmg ð2Þ

where T (k) and R(k) are the experimentally-measuredquantities (Fig. 4), and /(k) is the standard luminous effi-ciency for photopic vision. It is desirable to maximize both

functions. However, in practice it is not possible to simulta-

neously maximize both functions. Therefore, the design of the heat mirror must be optimized to yield the highest pos-sible combination of T vis and RIR. The values for these twofunctions for the WO3/Ag/WO3 heat mirrors are shown inFig. 6 as functions of the thickness of the silver layer. Theintegrated infrared reflectance did not show any saturation,

in accordance with observation (iv) above. However, theintegrated visible transmittance showed an optimum valuefor a silver layer thickness around 24 nm, in accordancewith observation (iii) above. Thus, this quantity is the deci-sive factor in the optimization of the heat mirror.

4. Conclusion

Transparent heat mirrors based on WO3/Ag/WO3 multi-layers were fabricated. The thickness of the tungsten oxidelayers was fixed at 35 nm. The thickness of the silver wasvaried over the range of 18–39 nm, for which the silver lay-ers were continuous. It was found that the visible transmit-tance decreased as the thickness of the silver layer increased.The maximum transmittance of 88.3% at 554 nm wasamong the highest reported for metal-dielectric heat mir-rors. The infrared reflectance increased as the thickness of the silver layer increased, as expected for metallic films.The interfaces of the three-layers were not sharp. Therefore,the effective thickness of the layers decreased, which affectstheir optical performance. The performance of the heat mir-rors was evaluated through a figure of merit that takes intoaccount the integrated intensity of the optical functions.This is a more accurate measure compared to consideringonly the maxima of the optical functions. The best perfor-

mance was obtained for the multilayers with a silver filmof thickness of 24 nm. This study not only established thesuitability of tungsten oxide for use as the dielectric in trans-parent heat mirrors but also showed that this material isamong the best dielectrics for this application. Two majorlimitations for the operation of a heat mirror can be envis-aged. The first limitation is a fundamental property and isrelated to the optical properties (transparency and absorp-tion) of the layers. The second limitation is the lack of sharpness of the layers, and this is a process-related matter.

Acknowledgment

This work was supported by the Physics Department of King Fahd University of Petroleum and Minerals.

References

Berning, P.H., 1983. Principles of design of architectural coatings. Appl.Opt. 22, 4127–4141.

Briggs, D., Seah, M.P., 1983. Practical Surface Analysis by Auger and X-ray Photoelectron Spectroscopy. Wiley, New York.

Chiba, K., Nakatani, K., 1984. Photoenhanced migration of silver atomsin transparent heat mirror coatings. Thin Solid Films 112, 359–367.

Chiba, K., Suzuki, K., 1992. Effect of heterogeneous metal atoms on thestability of a silver layer of a heat mirror coating. Sol. Energy Mater.

Sol. Cells 25, 113–123.

Fig. 6. Dependence of integrated visible transmittance (T vis) and inte-grated infrared reflectance (RIR) on the thickness of the silver film.

Fig. 5. XPS depth profile spectra showing the variation of the atomicconcentration of the various elements as a function of etching time for athree-layer structure (WO3/Ag/WO3/glass) with thicknesses of 35/24/35 nm.




Deb, S.K., Chopoorian, J.A., 1966. Optical properties and color-centerformation in thin films of molybdenum trioxide. J. Appl. Phys. 37,4818–4825.

Deb, S.K., 2008. Opportunities and challenges in science and technologyof WO3 for electrochromic and related applications. Sol. EnergyMater. Sol. Cells 92, 245–258.

Denton, R.E., Campbell, R.D., Tomlin, S.G., 1972. The determination of the optical constants of thin films from measurements of reflectanceand transmittance at normal incidence. J. Phys. D 5, 852–863.

Fan, J.C., Bachner, F.J., Foley, G.H., Zavracky, P.M., 1974. Transparentheat mirror films of TiO2/Ag/TiO2 for solar energy collection andradiation insulation. Appl. Phys. Lett. 25, 693–695.

Fan, J.C., 1981. Sputtered films for wavelength-selective applications.Thin Solid Films 80, 125–136.

Fu, J.K., Atanassov, G., Dai, Y.S., Tan, F.H., Mo, Z.Q., 1997. Singlefilms and heat mirrors produced by plasma ion assisted deposition. J.Non-Cryst. Solids 218, 403–410.

Granqvist, C.G., 1981. Radiative heating and cooling with spectrallyselective surfaces. Appl. Opt. 20, 2606–2615.

Johnson, P.B., Christy, R.W., 1972. Optical constants of the noble metals.Phys. Rev. B 6, 4370–4379.

Karlsson, B., Valkonen, E., Karlsson, T., Ribbing, C.G., 1981. Materials forsolar-transmitting heat reflecting coatings. Thin Solid Films 86, 91–98.

Kostlin, H., Frank, G., 1982. Optimization of transparent heat mirrors based ona thin silver film between antireflection films. Thin Solid Films 89, 287–293.

Lampert, C.M., 1981. Heat mirror coatings for energy conservingwindows. Sol. Energy Mater. 6, 1–41.

Lee, C.C., Chen, S.H., Jaing, C.C., 1996. Optical monitoring of silver-based heat mirrors. Appl. Opt. 35, 5698–5703.

Leftheriotis, G., Yianoulis, P., Patrikios, D., 1997. Deposition and opticalproperties of optimized ZnS/Ag/ZnS thin films for energy savingapplications. Thin Solid Films 306, 92–99.

Leftheriotis, G., Papaefthimiou, S., Yianoulis, P., Siokou, A., 2001. Effectof the tungsten oxidation states in the thermal coloration andbleaching of amorphous WO3 films. Thin Solid Films 384, 298–306.

Niklasson, G.A., Granqvist, C.G., 2007. Electrochromics for smartwindows: thin films of tungsten oxide and nickel oxide, and devicesbased on these. J. Mater. Chem. 17, 127–156.

Pracchia, J.A., Simon, J.M., 1981. Transparent heat mirrors: influence of the materials on the optical characteristics. Appl. Opt. 20, 251–258.

Valkonen, E., Karlsson, B., Ribbing, C.G., 1984. Solar optical propertiesof thin films of Cu, Ag, Au, Cr, Fe, Co, Ni and Al. Sol. Energy 32,211–222.

Wang, Z., Cai, X., Chen, Q., Li, L., 2006. Optical properties of metal-dielectric multilayers in the near UV region. Vacuum 80, 438–443.

Wang, Z., Cai, X., Chen, X.Q., Chu, P.K., 2007. Effects of Ti transitionlayer on stability of silver/titanium dioxide multilayered structure.Thin Solid Films 515, 3146–3150.

Yoldas, B.E., O’Keefe, T., 1984. Deposition of optically transparent IRreflective coatings on glass. Appl. Opt. 23, 3638–3643.


transparent heat mirrors based on tungsten oxide–silver multilayer structures

Documents