Energy and Buildings 36 (2004) 891–898

Dielectric/Ag/dielectric coated energy-efficientglass windows for warm climates

S.M.A. Durrania,∗, E.E. Khawajaa, A.M. Al-Shukrib, M.F. Al-Kuhaili b

a Center for Applied Physical Sciences, Research Institute, King Fahd University of Petroleum and Minerals,Box 1831, Dhahran 31261, Saudi Arabia

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

Accepted 15 February 2004

Abstract

Energy-efficient glass windows for warm climates were designed and fabricated using a three-layer system of dielectric/metal/dielectric(D/M/D) on glass. Silver was used as a metal layer. The design parameters for optimum performance of D/M/D on glass-systems fordielectrics, having refractive indices in the range 1.6–2.4, were obtained by numerical calculations. Based on these parameters, D/M/Dfilms on glass substrates were deposited using dielectrics such as TiO2, WO3, and ZnS. Upon testing these coated glass windows, it wasconcluded that the window with any of the three dielectrics performed well and the efficiencies of the windows with different dielectricswere nearly the same.© 2004 Elsevier B.V. All rights reserved.

Keywords:Dielectrics; Energy-efficient glass windows; Coated windows

1. Introduction

Over the past 40 years the introduction and develop-ment of an important new area of technology in the ar-chitectural field has occurred, namely, the use of thin filmcoatings to enhance the thermal performance of glass win-dows. Introductory installations of coated glass began in the1960s with the development of large-scale coating facilities,and these were expanded considerably over the last fourdecades as the merits of coatings began to be appreciated[1]. Most of the applications have been in the commercialarea. Now, in view of the growing energy crises, there iswidespread interest world over for higher performance prod-ucts, especially those that could contribute to energy savings[2–4].

The energy content of the solar spectrum is roughly splitbetween the visible and near-infrared regions. A glass win-dow to be energy-efficient in warm climates should havespectrally selective coatings such that it transmits nearly allthe energy in the visible and reflects all the energy in theinfrared. Thus an ideal energy-efficient window for a warmclimate would have transmittance (T) and reflectance (R) to

∗ Corresponding author.E-mail address:smayub@kfupm.edu.sa (S.M.A. Durrani).

be given by:T = 1 andR = 0 for wavelength 400–700 nm(i.e. visible region), and,T = 0 andR = 1 for wavelength>700 nm (infrared, IR, region).

Three-layer systems of dielectric/metal/dielectric (D/M/D)on glass substrates have been used for spectrally selec-tive coatings for various purposes including the energyefficiency (e.g. Ref.[5–11]). By varying the material andthickness of the three layers, the optical properties of theD/M/D films can be tailored to suit different applications.

The main objective of the present work was to develop alaboratory-scale version of thin film coated energy-efficientglass windows for applications in a warm climate. Thethree-layer system of D/M/D on glass was used. Numericalmethod was used for optimizing such a window to givehigh near-infrared reflection and low visual loss. Results arepresented for windows with Ag film between TiO2, WO3,and ZnS layers.

2. Optimization of the optical performances of coatedwindows

The D/M/D films on a glass substrate were used as aspectrally selective filter[5–11] that reflects infrared radia-tion (due to the properties of the metal layer) and transmits

0378-7788/$ – see front matter © 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.enbuild.2004.02.003

892 S.M.A. Durrani et al. / Energy and Buildings 36 (2004) 891–898

most of the visible spectrum. The highly reflective metalfilm, that otherwise transmits very little energy in the visible,was sandwiched between the two dielectric layers that act asanti-reflective coatings so as to enhance the energy transmit-ted in the visible region. Thus, the visible transmittance waschosen to be the parameter to be optimized for a given metalthickness, which in turn controls the IR reflectance. The op-timum thickness of each dielectric layer was estimated. Forthe design of the D/M/D films, a computer program waswritten and implemented in order to perform calculations ofthe optical properties of multilayer films. For this purposethe characteristic matrix formulation, given by Heavens[12]was used.

A systematic search was carried out in order to deter-mine the effect of each layer on the optical properties ofD/M/D type films. The film arrangement was taken to be:

0.548

0.550

0.554

0.552

52 56 58 60 6254

f = 1.0

f = 0.9 f = 1.1 f = 0.9 f = 1.1

f = 1.0

0.65

0.66

0.68

0.67

42 50 54 5846

n = 2.0

n = 1.6 n = 1.8

f = 0.9 f = 1.1

f = 1.0 n = 2.2 f = 1.0

f = 1.1

0.80

0.81

0.82

0.83

0.84

0.85

30 35 40 45

n = 2.4 f = 1.0

f = 1.1

0.87

0.88

0.89

0.90

28 30 32 34 36 38 40

Thickness, d1 (nm)

f = 0.9

f = 0.9

(b)

(d)

(e)

(a)

(c)

35 45 50400.73

0.74

0.75

0.77

0.78

0.76

Tra

nsm

itta

nce

Fig. 1. Computer simulations for optimization of D/M/D layers on glass-system. The two D-layers in the system are of the same dielectric. Effectsof thickness (d1and d3) of dielectric layers on transmittance at wavelength of 550 nm for different values of f (=d3/d1) are shown for dielectrics withrefractive indices of (a) 1.6, (b) 1.8, (c) 2.0, (d) 2.2, and (e) 2.4.

air (n0,k0)/D(n1, k1)/M(n2,k2)/D(n3,k3)/glass (n4,k4), wheren’s andk’s are the refractive and absorption indices, respec-tively, of each of the layers and the media surrounding thethree layers. The absorption of the solar radiation in air, di-electrics and glass was considered to be insignificant, thatis k0 = k1 = k3 = k4 = 0. The two dielectric layers weretaken to be of same material, that isn1 = n3. The calcu-lations (for D/M/D films on glass) were performed for nor-mal incidence transmittance (T) and reflectance (R) usingthe equations given in Ref.[12]. It may be noted that theequation forT [12], gives the transmittance into the glasssubstrate. The transmittance measured in the air across theback-face of the substrate is about 4% less than that in theglass. The results of such calculations are discussed below.The T across the back-face of the substrate is discussed inthe following.

S.M.A. Durrani et al. / Energy and Buildings 36 (2004) 891–898 893

2.1. Effects of (a) thickness and (b) refractive index ofdielectric layer on T

The design wavelength used here is 550 nm. The opticalconstants of Ag at 550 nm aren2 = 0.05 andk2 = 3.55[13]. The thickness of the metal layerd2 = 20 nm was used.It is well known that Ag films, thinner than 15 nm, tendto be inhomogeneous and granular. More realistic valuesfor the silver layer thickness are considered to be those of16–24 nm[11]. Such films are semi-transparent with signifi-cant reflectance in the infrared. InFig. 1a, the transmittanceis plotted as a function of thicknessd1 of the top dielectriclayer for different values of the ratio,f = d3/d1 (whered3 is the thickness of the bottom dielectric layer).Fig. 1ais obtained for a dielectric with refractive indexn1 (or n3)= 1.6. Similar curves for other dielectrics withn1 = 1.8,2.0, 2.2, and 2.4 are shown inFig. 1b-e, respectively. It isapparent fromFig. 1, that for maximum transmittance at thedesign wavelengthf = 1. This means that the dielectriclayers in the D/M/D system should have the same thick-

0.55

0.60

0.65

0.70

0.80

0.85

0.90

0.75

Tm

ax

30

35

40

50

55

60

45

Thi

ckne

ss, d

1 (n

m)

1.6 1.7 1.8 1.9 2.0 2.1 2.3 2.42.2

Refractive Index (n1 or n2)

Dielectric Layers

(a)

(b)

Fig. 2. Results ofFig. 1 are summarized here. Dependences of (a) maximum value of the transmittance and (b) thickness of the dielectric layer(s) onthe refractive index of a dielectric are shown.

ness. The results ofFig. 1a-eare summarized inFig. 2. Us-ing the results ofFig. 1a-e, the maximum in transmittancewas plotted as a function of the refractive index of the di-electric in Fig. 2a. It is clear from Fig. 2a that for hightransmission in the visible region, the dielectric to be usedneeds to have high refractive index. Using the results ofFig. 1a-e, the thicknessd1 of the dielectric layer that gavemaximum transmittance was plotted as a function of the re-fractive index of the dielectric inFig. 2b. Once the dielectricis selected,Fig. 2b may be used to identify the thicknessof the two dielectric layers for the optimum performanceof the D/M/D system. Further calculations show that in aD/M/D system, when the thickness of both dielectric lay-ers are changed by 60%, the change in IR reflectance is amere 2.5%. It can, therefore, be concluded that the thick-ness of the dielectric layers can be altered in order to opti-mize transmission in the visible without any undue changesin the IR reflectance. Based on these findings, the spectralresponses of various D/M/D were calculated, as discussedbelow.

894 S.M.A. Durrani et al. / Energy and Buildings 36 (2004) 891–898

Table 1Optical constants (refractive indexn and absorption indexk) of Ag filmsat different wavelengths (λ) from Ref. [13]

λ (nm) n k

2000 0.25 14.501900 0.22 13.751800 0.20 13.151700 0.18 12.501600 0.16 11.801500 0.14 11.031400 0.12 10.101300 0.10 9.501200 0.09 8.701100 0.07 8.001000 0.06 7.18900 0.06 6.40800 0.05 5.60700 0.05 4.80600 0.05 4.00500 0.05 3.07400 0.05 2.25

2.2. Optical constants of thin films

The optical constants (n and k) for Ag-films given inTable 1, were taken from Ref.[13]. Dielectric films, such asTiO2, WO3, and ZnS, are non-absorbing in the wavelengthregion from 400 nm to 2000 nm; the region of interest in thepresent work. Therefore, the absorption indexk = 0, and weare left with refractive index alone. A method given in Ref.[14] for determining both the refractive index (as a functionof wavelength) and the thickness of a transparent film on atransparent substrate (glass) from measurement of transmit-tance at normal incidence was used in the present work. Av-erage values of the refractive index at different wavelengthsfor the films of different thickness are given inTable 2for

Table 2Refractive index (n) at different wavelengths of TiO2 films (λ) depositedon unheated substrates

λ (nm) n

2000 1.791900 1.791800 1.801700 1.801600 1.801500 1.811400 1.811300 1.821200 1.821100 1.831000 1.83900 1.84800 1.85700 1.87600 1.89500 1.93400 2.05

Table 3Refractive index (n) at different wavelengths (λ) of WO3 films

λ (nm) n

2000 1.941900 1.941800 1.951700 1.951600 1.961500 1.961400 1.971300 1.971200 1.981100 1.991000 1.99900 2.00800 2.00700 2.01600 2.02500 2.04400 2.05

TiO2 deposited on unheated substrates (1.9),Table 3 forWO3 (2.03),Table 4for TiO2 deposited on heated substrates(2.23), andTable 5for ZnS (2.36). The values of the refrac-tive indices of the dielectrics at a wavelength of 550 nm aregiven in parentheses.

2.3. Effects of thickness of Ag layer on spectral responseof R and T

Spectral dependence of reflectance and transmit-tance of D/M/D systems such as (a) TiO2/Ag/TiO2, (b)WO3/Ag/WO3, (c) ZnS/Ag/ZnS, and based on the data ofTables 1–5(optical constants of individual material), werecalculated for different thickness of the Ag layer. Thick-nesses of the dielectric layers used were obtained fromFig. 2. The results are shown inFig. 3. It is clear from

Table 4Refractive index (n) at different wavelengths of TiO2 films (λ) depositedon heated substrates (at 300◦C)

λ (nm) n

2000 2.101900 2.101800 2.101700 2.111600 2.111500 2.121400 2.121300 2.131200 2.131100 2.141000 2.14900 2.15800 2.16700 2.18600 2.21500 2.27400 2.40

S.M.A. Durrani et al. / Energy and Buildings 36 (2004) 891–898 895

0.2

0.4

0.6

0.8

1.0

400 800 1200 1600 2000

T

T

T

T

R R

RR

a

a

a

a

e

e

e

a

a

a

a

e

e

e

e

e

0.0

0.2

0.4

0.6

0.8

1.0

Tra

nsm

itta

nce

(T)

and

Ref

lect

ance

(R

)

400 800 1200 1600

Wavelength (nm)

TiO2/Ag/TiO2

(Unheated Substrate)

TiO2/Ag/TiO2

(Heated Substrate)

WO3/Ag/WO3

ZnS/Ag/ZnS

(a)

(c) (d)

(b)

Fig. 3. Computer simulation of spectral transmittance and reflectance of D/M/D on glass for different thickness of M-layer and for different dielectricmaterials. Curves marked ‘a’ to ‘e’ correspond to Ag-layer thickness of 16, 18, 20, 22 and 24 nm, respectively. For these simulations, data on opticalconstants given inTables 1–5were used.

Fig. 3 that as the Ag-layer thickness increases, an averagetransmitttance in the visible region (TVIS) decreases whilean average infrared reflectance (RIR) increases. An idealheat mirror is such that it hasTVIS = 1.0 andRIR = 1.0(seeSection 1). This ideal behavior cannot be achieved inpractice because, when the Ag film thickness is changed,one of the twoTVIS andRIR, increases while the other de-creases. Therefore, a compromise is needed. For optimumperformance a figure of merit, Z, may be defined as

Z =[(∫

TVIS dλ∫dλ

)VIS

] [(∫RIR dλ∫

dλ

)IR

]

whereλ is the wavelength. For an ideal case, sinceTVIS =RIR = 1, the above equation givesZ = 1. However, for

the actual case, bothTVIS and RIR are less than 1.0 (seeFig. 3). Therefore,Z is expected to be less than 1.0. Us-ing the results ofFig. 3 in the above equation, the valuesof Z were calculated as a function of the thickness of theAg-layer for the different D/M/D systems. The results areshown inFig. 4. For the ZnS/Ag/ZnS and TiO2/Ag/TiO2 (onheated substrate) systems,Z is maximum when the Ag-layerhas a thickness in the vicinity of 18 nm. On the other hand,for the WO3/Ag/WO3 and the TiO2/Ag/TiO2 (on unheatedsubstrate) systems, the maximum value ofZ could be ob-tained for an Ag-layer of thickness that is smaller than 18 nm(Fig. 4). It is well known that Ag films thinner than 15 nmtend to be inhomogeneous and granular. More realistic val-ues for the silver layer thickness are considered to be thoseof 16–24 nm[11]. Thus, Ag film thickness in the vicinity of

896 S.M.A. Durrani et al. / Energy and Buildings 36 (2004) 891–898

Table 5Refractive index (n) at different wavelengths (λ) of ZnS films

Wavelengthλ (nm) Refractive indexn

2000 2.231900 2.231800 2.241700 2.241600 2.251500 2.251400 2.261300 2.261200 2.271100 2.281000 2.29900 2.30800 2.32700 2.33600 2.35500 2.39400 2.56

18 nm seems to be a reasonable compromise. For this rangeof thickness, all the D/M/D systems discussed here, havevalues ofZ that are greater than 0.6.

Design parameters for optimum performance of D/M/Don glass-systems for various dielectrics are listed inTable 6.These parameters were used to fabricate the D/M/D filmson glass substrates.

WO3/Ag/WO3

TiO2/Ag/TiO2

Heated Substrate

TiO2/Ag/TiO2

Unheated Substrate

16 17 18 19 20 21 22 23 24

Thickness of Ag Layer (nm)

0.50

0.55

0.60

0.65

0.70

Z-F

acto

r

ZnS/Ag/ZnS

Fig. 4. Z-factor derived from the results ofFig. 3 as a function of thethickness of Ag-layer for the different dielectrics.

Table 6Design parameters for optimum performance of D/M/D on glass-systemfor various dielectrics

Dielectric Thickness (nm)

Dielectric layer,d1 = d3 Metal layer,d2

[TiO2]a 46 18WO3 43 18[TiO2]b 38 18ZnS 34 18

a Deposited on unheated substrate.b Deposited on heated substrate.

3. Fabrication and results of D/M/D films on glass

Thin films were prepared by physical vapor deposition(PVD) in a vacuum of 10−6 mbar, using a Leybold modelL560 box coater pumped by a turbomolecular pump ontoglass substrates. A tungsten boat was used for evaporationof Ag, while molybdenum boats were used for WO3 andZnS. The TiO2 films were prepared by electron beam evap-oration. Purities of the materials used for evaporation werebetter than 99.9%. In order to produce films with uniformthickness, the substrates were rotated while deposition tookplace. For monitoring the evaporation process of these films,a Leybold Inficon XTC quartz crystal monitor and evapo-ration controller were employed. Calibration of the quartzthickness monitor for individual material was achieved bymeasuring the actual thickness of single films by opticalmethods[14,15]. Normal incidenceT andR from films wasmeasured using JASCO V570 spectrophotometer.

Various three-layered systems of dielectric/metal/dielectricon glass were fabricated using Ag as a metal layer whilefor a set of two dielectric layers either ZnS or WO3 or TiO2was used. Designed parameters used were those given inTable 5. The three-layered system was fabricated in one gowithout breaking the vacuum, which required two differentevaporation sources.

Measured spectral reflectance and transmittance ofD/M/D films on glass substrates are shown inFig. 5 forZnS/Ag/ZnS/glass, WO3/Ag/WO3/glass, and TiO2/Ag/TiO2/unheated-glass. For an ideal energy-efficient glass win-dow the transmittance in the visible region should beclose to one. It may be noted that for an uncoatedglass window the transmittance is about 0.92. The sys-tems such as ZnS/Ag/ZnS/glass, WO3/Ag/WO3/glass, andTiO2/Ag/TiO2/unheated-glass, are acceptable as these haveaverage transmittance of about 0.7 in the visible region(Fig. 5). These three systems perform equally well as faras their transmittance in the visible region is concerned(Fig. 5). On comparing the measured reflectance andtransmittance (Fig. 5) with the corresponding reflectanceand transmittance obtained through computer simulations(Fig. 3) we find that for the different dielectrics the twoare in good agreement (within 10%) for the entire spec-tral region that was covered. However, for the system of

S.M.A. Durrani et al. / Energy and Buildings 36 (2004) 891–898 897

TiO2/Ag/TiO2

WO3/Ag/WO3

(unheated substrate)

0.2

0.4

0.6

0.8

1.0

0.2

0.4

0.6

0.8

R

R

R

T

T

T

ZnS/Ag/ZnS

0.0

0.2

0.4

0.6

0.8

1.0Ref

lect

ance

and

Tra

nsm

itta

nce

400 800 1200 1600 2000

Wavelength (nm)

(a)

(b)

(c)

Fig. 5. Measured reflectance and transmittance of D/M/D on glass-systemsfor the different dielectrics.

TiO2/Ag/TiO2/heated-glass the average transmittance in thevisible region was smaller than 0.4, and thus it may nothave been of real use. Inter diffusion of the two materialsat the interface may have been the cause for such a lowtransmittance.

Field tests of the coated glass were performed using aset-up that is described below. A cube (0.1 m× 0.1m) wasconstructed such that its walls were made from the coatedglass plates, while wooden plates were used for the floorand the roof. Another cube of the same dimensions was con-structed; however, in this case the coated glass plates werereplaced with uncoated glass plates. Each wall of the twocubes was exposed to a 100 W incandescent lamp, and thetemperatures inside both the cubes were measured as a func-tion of time, using two thermometers. It may be mentionedthat the spectrum of radiation emitted by a lamp somewhatresembles the solar radiation spectrum. In fact, there arespecial lamps that are commercially available and have ra-diation spectrum that matches the solar spectrum. However,

0 4 8 12 16 20 24 28 3

Time (min)

220

24

28

32

36

40

44

48

Tem

pera

ture

(o C

)

(a) uncoated glass

(b) coated glass

Fig. 6. Measured temperatures inside the two cubes as a function of time:(a) cube using uncoated glass and (b) cube using D/M/D coated glass.

these lamps were not available locally. Therefore, in thepresent work ordinary filament lamps were used. The re-sults are shown for a WO3/Ag/WO3/glass inFig. 6. Insidetemperature of the cube with coated windows was lower byabout 8◦C when compared with the inside temperature ofthe cube with uncoated windows (Fig. 6). The energy effi-ciency of the glass window is clearly reflected by the plots ofFig. 6. Similar results were obtained for ZnS/Ag/ZnS/glassand TiO2/Ag/TiO2/unheated-glass-systems and these werewithin 5% of those that are shown inFig. 6.

4. Conclusions

Energy-efficient glass windows for warm climates weredesigned and fabricated using a three-layer system of di-electric/metal/dielectric (D/M/D) on glass. Silver was usedas a metal layer. Optimization of the optical performancesof the three layers on glass through computer simulationswas carried out. Based on some of the parameters ob-tained through the simulations, successful fabrication of theenergy-efficient windows was achieved. Overall optical per-formances of the D/M/D systems with different dielectricswere in close agreement with those predicted by the simula-tions. On testing it was found that the D/M/D systems were

898 S.M.A. Durrani et al. / Energy and Buildings 36 (2004) 891–898

highly energy-efficient. Temperature difference of about8◦C was observed inside the two cubes, one using coatedand the other uncoated glass windows.

Acknowledgements

This work is part of an internal project # CAPS 1202,supported by the Research Institute of King Fahd Universityof Petroleum and Minerals.

References

[1] P.H. Berning, Appl. Opt. 22 (1983) 4127.[2] C.M. Lampert, Sol. Energy Mater. 11 (1984) 1.

[3] C.G. Granqvist, Thin Solid Films 193/194 (1990) 730.[4] R.B. Goldner, J. Vac. Sci. Technol. A 13 (1995) 1088.[5] C. John, C. Fan, F.J. Bachner, G.H. Foley, P.M. Zavracky, Appl.

Phys. Lett. 25 (1974) 693.[6] J.A. Pracchia, J.M. Simon, Appl. Opt. 20 (1981) 251.[7] C.G. Granqvist, Appl. Opt. 20 (1981) 2606.[8] H. Köstlin, G. Frank, Thin Solid Films 89 (1982) 287.[9] C.C. Lee, S.H. Chen, C.C. Jaing, Appl. Opt. 35 (1996) 5698.

[10] X. Zhang, S. Yu, M. Ma, Sol. Energy Mater. Sol. Cells 44 (1996)279.

[11] G. Leftheriotis, P. Yianoulis, D. Patrikios, Thin Solid Films 306(1997) 92.

[12] O. S. Heavens, Optical Properties of Thin Solid Films, Dover Pub-lications, New York, 1991.

[13] P.B. Johnson, R.W. Christy, Phys. Rev. B 6 (1972) 4370.[14] E.E. Khawaja, J. Phys. D: Appl. Phys. 9 (1976) 1939.[15] E.E. Khawaja, S.M.A. Durrani, A.M. Al-Shukri, Thin Solid Films

358 (2000) 166.