ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2013

Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 1095

VO2-based Thermochromic andNanothermochromic Materialsfor Energy-Efficient Windows

Computational and Experimental Studies

SHUYI LI

ISSN 1651-6214ISBN 978-91-554-8801-7urn:nbn:se:uu:diva-210016

Dissertation presented at Uppsala University to be publicly examined in Häggsalen, ÅngströmLaboratory, Lägerhyddsvägen 1, Uppsala University, Polacksbacken, Uppsala, Friday,13 December 2013 at 13:15 for the degree of Doctor of Philosophy (Faculty of Theology).The examination will be conducted in English. Faculty examiner: Professor Ivan Parkin(University College London).

AbstractLi, S. 2013. VO2-based Thermochromic and Nanothermochromic Materials for Energy-Efficient Windows. Computational and Experimental Studies. Digital ComprehensiveSummaries of Uppsala Dissertations from the Faculty of Science and Technology 1095.143 pp. Uppsala: Acta Universitatis Upsaliensis. ISBN 978-91-554-8801-7.

VO2-based lms are thermochromic and exhibit high or low infrared transmittance when thetemperature is below or above a critical temperature. The thermochromic switching is passiveand reversible, and therefore VO2 based lms are promising for energy-efcient window appli-cations. However the practicaluse of VO2 for energy-efcient windows has long been hamperedby low luminous transmittance and low solar energy transmittance modulation. The main goalof this dissertation work is to address these issues.

The rst half of the work proposes the concept of nanothermochromics for simultaneousimprovement of luminous transmittance and modulation of solar energy throughput. nanoth-ermochromics considers VO2 nanoparticle composite layers, whose optical properties weremodeled by effective medium theories. Calculations on VO2 spheroids have shown that VO2

nanoparticles, especially nanospheres, can offer dramatically improved luminous transmittanceand solar transmittance modulation that are not possible for lms. Calculations done on coreshellnanoparticles showed comparable improvements and offer an opportunity to reduce the materialcosts. It was also found that the composite of In2O3:Sn (ITO) and VO2 can yield moderately highluminous transmittance, solar transmittance modulation and low-emittance properties.

In the second half of the dissertation work, Mg-doped VO2 lms were sputter deposited.Their band gaps and Mg-content were investigated by means of optical absorption measurementand Rutherford backscattering spectrometry, respectively. The band gaps of VO2 were foundto increase by ∼3.9±0.5 eV per unit of atom ratio Mg/(Mg+V) for 0<Mg/(Mg+V)<0.21.Computations based on effective medium theory were done to estimate the performance of Mg-doped VO2 lms and nanoparticle composite layers. The results suggest that moderately dopedVO2 lms with 0<Mg/(Mg+V)<0.06 perform better than un-doped lms and that the perfor-mance can be further enhanced with one layer of antireection coating. The best results wereachieved by un-doped VO2 nanospheres, closely followed by the VO2 nanospheres with lowMg-content.

Furthermore, the an experimental study on sputter deposited VO2 nanorods has identied thegeometry of the oxygen gas inlet, the type of substrate, the substrate temperature and the layerthickness as important factors that inuence the growth morphology.

Taken as a whole, nanothermochromics offered by VO2 nanoparticles was shown to be thebest solution for VO2 based thermochromic energy-efcient window coatings.

Keywords: VO2, thermochromism, nanothermochromics, energy efficient windows, opticalmodeling, effective medium theory, thin film characterization, doping

Shuyi Li, Department of Engineering Sciences, Box 534, Uppsala University, SE-75121Uppsala, Sweden.

© Shuyi Li 2013

ISSN 1651-6214ISBN 978-91-554-8801-7urn:nbn:se:uu:diva-210016 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-210016)

To my fam ily and friends

List of Papers

This thesis is based on the following papers, which are referred to in the textby their Roman numerals.

I S.-Y. Li, G. A. Niklasson and C. G. Granqvist, Nanothermochromics:Calculations for VO2 nanoparticles in dielectric hosts show much im-proved luminous transmittance and solar energy transmittance modula-tion, J. Appl. Phys. 108, 063525 (2010)

II S.-Y. Li, G. A. Niklasson and C. G. Granqvist, Nanothermochromicswith VO2-based core-shell structures: Calculated luminous and solaroptical properties, J. Appl. Phys. 109, 113515 (2011)

III S.-Y. Li, G. A. Niklasson and C. G. Granqvist, A thermochromic low-emittance coating: Calculations for nanocomposites of In2O3:Sn andVO2, Appl. Phys. Lett. 99, 131907 (2011)

IV Shu-Yi Li, Kyoko Namura, Motofumi Suzuki, Gunnar A. Niklasson,and Claes G. Granqvist, Thermochromic VO2 nanorods made by sput-ter deposition: Growth conditions and optical modeling, J. Appl. Phys.114(3), 033516 (2013)

V Shu-Yi Li, Nuru R. Mlyuka, Daniel Primetzhofer, Anders Hallén,Göran Possnert, Gunnar A. Niklasson, and Claes G. Granqvist,Bandgap widening in thermochromic Mg-doped VO2 thin films:Quantitative data based on optical absorption, Appl. Phys. Lett. 103,161907 (2013)

VI Shu-Yi Li, Gunnar A. Niklasson, and Claes G. Granqvist,Thermochromic undoped and Mg-doped VO2 thin films andnanoparticles: Optical properties and performance limits for energyefficient windows, in manuscript

Reprints were made with permission from the publishers.

Papers not included in the thesis

VII S.-Y. Li, G. A. Niklasson and C. G. Granqvist, Plasmon-induced near-infrared electrochromism based on transparent conducting nanoparti-cles: Approximate performance limits, Appl. Phys. Lett. 101, 071903(2012)

VIII I. Bayrak Pehlivan, E. L. Runnerstrom, S.-Y. Li, G. A. Niklasson,D. J. Milliron and C. G. Granqvist, A polymer electrolyte with highluminous transmittance and low solar throughput: Polyethyleneimine-lithium bis(trifluoromethylsulfonyl) imide with In2O3:Sn nanocrystals,Appl. Phys. Lett. 100, 241902 (2012)

IX Shuanglin Hu, S.-Y. Li, R. Ahuja, C. G. Granqvist, K. Hermansson,G. A. Niklasson and R. H. Scheicher, Optical properties of Mg-dopedVO2: Absorption measurements and hybrid functional calculations,Appl. Phys. Lett. 101, 201902 (2012)

X K. Laaksonen, S.-Y. Li, S. R. Puisto, T. Ala-Nissila, C. G. Granqvist, R.M. Nieminen and G. A. Niklasson, Functional oxide nanoparticles indielectric media: Conditions for low optical scattering, and comparisonbetween effective medium and four-flux theories, Submitted

XI I. Lykissa, S. Li, M. Ramzan, R. Ahuja, C.-G. Granqvist, and G. A.Niklasson, Electronic density of states of amorphous vanadium pentox-ide films: Electrochemical measurements and density functional theorycalculations, in manuscript

Conference papers

XII C. G. Granqvist, I. Bayrak Pehlivan, Y.-X. Ji, S.-Y. Li and G. A. Niklas-son, Electrochromics and Thermochromics for Energy Efficient Fenes-tration: New Functionalities Based on Nanoparticles of In2O3:Sn andVO2, Thin Solid Films, DOI: 10.1016/j.tsf.2013.10.033.

XIII C. G. Granqvist, I. Bayrak Pehlivan, Y.-X. Ji, S.-Y. Li, E. Pehlivan, R.Marsal, and G. A. Niklasson, Electrochromics and Thermochromics forEnergy Efficient Fenestration: New Applications Based on TransparentConducting Nanoparticles, MRS Proceedings, accepted.

XIV Shu-Yi Li, Ilknur Bayrak Pehlivan, Gunnar A. Niklasson and Claes-Göran Granqvist, Progress in Electrochromics and Thermochromics:Two New Applications Involving ITO Nanoparticles, Society of Vac-uum Coaters 55th Annual Technical Conference Proceedings, 41-46,(2012).

XV S.-Y. Li, G. A. Niklasson and C. G. Granqvist, Thermochromic fenes-tration with VO2-based materials: Three challenges and how they canbe met, Thin Solid Films, 520, 3823-3828 (2012).

XVI Shu-Yi Li, Gunnar A. Niklasson and Claes-Göran Granqvist,Thermochromics and nanothermochromics: New options for energyefficient fenestration, Society of Vacuum Coaters 54th AnnualTechnical Conference Proceedings, 29-34 (2011)

XVII C.-G. Granqvist, S. Green, S. Li, N. Mlyuka, and G. A. Niklasson,Progress in chromogenic materials and devices : New data onthermochromic vanadium-oxide-based materials and on electrochromicnickel-tungsten-oxide based foils. Society of Vacuum Coaters 53rdAnnual Technical Conference Proceedings, 75-79, 2010.

XVIII S.-Y. Li, G. A. Niklasson and C. G. Granqvist, Thermochromism ofVO2 Nanoparticles: Calculated Optical Properties and Applications toEnergy Efficient Windows, MRS Proceedings, mrsf10-1315-mm07-07,(2011)

XIX Shuyi Li, Motofumi Suzuki, Kaoru Nakajima, Kenji Kimura, TakaoFukuoka, Yasushige Mori, An approach to self-cleaning SERS sensorsby arraying Au nanorods on TiO2 layer, Proc. SPIE 6647, Nanocoat-ings, 66470J (September 19, 2007)

XX Motofumi Suzuki, Wataru Maekita, Yoshinori Wada, Shuyi Li,Kaoru Nakajima, Kenji Kimura, Takao Fukuoka, Yasushige Mori,Plasmonic nanocoatings tailored for surface-enhanced Raman imagingin near-infrared region, Proc. SPIE 6647, Nanocoatings, 664707(September 10, 2007)

Book chapter

XXI C. G. Granqvist, S. V. Green, S.-Y. Li, N. R. Mlyuka, G. A. Niklas-son and E. Avendaño, Chromogenics for Sustainable Energy: Some Ad-vances in Thermochromics and Electrochromics, ch. 14 in Advances inScience and Technology 8, Nova press, (2011) pp 449-460.

My contribution to the appended papers.

I All the modeling and calculation, most of the analysis and part of thewriting.

II All the modeling and calculation, most of the analysis and part of thewriting.

III All the modeling and calculation, most of the analysis and part of thewriting.

IV All the modeling and calculation, most of the experiment analysis andwriting.

V All the modeling and calculation, most of the experiment analysis andwriting.

VI All the modeling and calculation, most of the experiment analysis andwriting.

Disclaimer: Half of this Ph.D thesis is based on my licentiate thesis titled:VO2-based Nanothermochromics for Energy Efficient Windows: Calculatedresults, which was intended as a half-time report for my Ph.D project.

List of symbols and abbreviations

Greek letters

α Absorption coefficient

αpol Polarizability

Γ Damping coefficient in Lorentz model

∆Tsol Modulation of solar energy throughput, i.e. solar trans-mittance modulation.

δ Phase change during the passage of light in the film

ε Complex dielectric function

εA, εB Dielectric functions of component A and B in Brugge-man theory

εBR Effective dielectric function obtained from Bruggemantheory

εe f f Effective dielectric function

εMG Effective dielectric function obtained from Maxwell-Garnett theory

ε0 Dielectric permittivity in empty space

ε1 Real part of the complex dielectric function

ε2 Imaginary part of the complex dielectric function

εc Dielectric function of the particle core

ε f Dielectric function obtained from films

εm Dielectric function of the hosting matrix

εp Dielectric function of the particle inclusion

εs Dielectric function of the particle shell

ε∞ Dielectric background

εDf , εD

p Drude type terms of the dielectric function of films andnanoparticles respectively

εmax Largest of the dielectric functions in a composite mate-rial

εsub Dielectric function of the substrate

εtherm(λ ), εtherm(Tt) Wavelength and temperature-dependent thermal emit-tance, respectively.

θ Angle of incidence

θscatter the angle between the incident and diffracted beam inXRD.

λ Wavelength

λmax Wavelength where maximum radiated energy occurs

ν Frequency of light

ξ Slope value of linear trend extracted from band gap asa function of Mg/(Mg+V) atom ration z as in Eq. (8.1)

ρ mass density

ρR Resistivity

σSB A constant mentioned in the Stefan-Boltzmann law inEq. (2.2)

τ Temperature

τcr Critical temperature for metal-to-insulator transition inVO2

τ f Mean scattering time for the conduction electrons inthin films (or bulk)

τp, τ f Mean scattering time for the conduction electronsnanoparticles, and thin films (or bulk) respectively.

τsc;n Scattering time, 1/τsc;n is proportional to the width ofeach transitions represented by Lotentz oscillator in Eq.(4.18) and (4.19)

τsc Scattering time

ϕlum Luminous efficiency of the human eye

ϕsol Solar irradiance spectrum when the sun is standing 37

above the horrizon (corresponds to air mass 1.5 or AM1.5)

Ω Volume ratio between the inner and outer spheroids ofa core-shell structure

ω Angular frequency of the electromagnetic wave, i.e.light

ω∗p Screened plasma frequency

ωn Characteristic band-to-band transition energies in Eq.(4.18) and (4.19)

ωp Plasma frequency

ω0 Frequency of the restoring force in Lorentz model

ωP f , ωPp Plasma frequency for electrons in films and nanoparti-cles, respectively

ωres Localized surface plasma resonance frequency forsmall particles

Roman letters

a Major axis of spheroidal particles

An Spectral weight of band-to-band transitions in Eq.(4.18) and (4.19)

A(λ ) Wavelength-dependent absorptance

Ai atom Ai

c Minor axis of spheroidal particles

c0 Speed of light

CWien A constant mentioned in Wien’s displacement law inEq (2.3)

D Dielectric displacement

d thickness

Dp Particle diameter

dt , db thicknesses of top and bottom layers in the structuralmodels describing nanorod samples mentioned in Sec-tion 9.2

dhkl distances between crystall planes (hkl)

E Electromagnetic wave in matter

E0 Incident electromagnetic wave

Eg(0) band gap for zero Mg content as in Eq. (8.1)

Eg(z) band gap as a function of Mg/(Mg+V) atom ratio z asin Eq. (8.1)

EF Fermi level

Eg1, Eg2, Eg3 Band gaps for the semiconducting state of VO2

Eg Band gap from O 2p to the Fermi level for the metallicstate of VO2

eOS,PS A parameter that is a function of principal axes of pro-late or oblate spheroids

Etherm Total integrated hemispherical thermal emittance

f Filling factor, i.e., the volume fraction occupied by theparticles.

ft , fb filling factor of VO2 nanoparticles in top and bottomlayers in the structural models describing nanorod sam-ples mentioned in Section 9.2

Fhkl Structure factor

h Plank’s constant

h Plank’s constant divided by 2π

I(λ ,Tt) Wavelength-denpendent distribution of the radiated en-ergy per unit time per unit volume per unit solid angleper unit wavelength from a black body

Ihkl Intensity of Bragg reflection

Iint Radiated power per unit area of the black body

k Imaginary part of the complex refractive index

kB Boltzmann’s constant

ℓ Electron mean free path in thin films (or bulk)

ℓ f Electron mean free path in the film

L Depolarization factor

Le f f ,i, i=1, 2, 3 The effective depolarization factors of particle aggre-gates.

Li, i =1, 2, 3 Depolarization factors along the principal axes of ellip-soids

Llength The length dimension of sampled area along the flowof current in 4-wire method.

Lwidth The width dimension of sampled area perpendicular tothe current in 4-wire method.

M Interface matrix

m Aspect ratio of spheroidal particles

m∗e The effective mass of electron.

mi mass of atom Ai

N Complex refractive index

n Real part of the complex refractive index

nc Complex refractive index of the nanoparticle core

ne Charge density

ni the number fraction of atom Ai

nm Complex refractive index of the dielectric matrix

ns Complex refractive index of the nanoparticle shell

nx An integer that represents the order of diffraction

Narea number of atoms per area as determined by Rutherfordbackscattering spectrometery

p Dipole moment

r Displacement in Eq. (4.7)

R Intensity reflectance

R(λ ) Wavelength-dependent reflectance

RBaSO4 Reflectance of BaSO4 reference sample.

rcoh Amplitude reflectance for a coherent layer

Rdi f f Diffuse reflectance

Rlum Luminous reflectance

Roi Reflectance from outer half space i to inner halfspace o

Rres Resistance

Rs,p Intensity reflectance of s- and p-polarized light

rs,p Amplitude reflectance of s- and p-polarized light

Rsol Solar reflectance

Rspec Specular reflectance

Rtot Total reflectance

S Layer matrix

S1,tot , S1,di f f The total and diffuse signal recorded using transmit-tance setting in Lambda 900 instrument, respectively.

S2,tot , S2,di f f The total and diffuse signal recorded using reflectancesetting in Lambda 900 instrument, respectively.

T Intensity transmittance

t The thickness of the particle shell

T (λ ) Wavelength-dependent transmittance

Tt Temperature of the radiating black body

tcoh Amplitude transmittance for a coherent layer

Tdi f f Diffuse transmittance

Tlum Luminous transmittance

Toi Transmittance from outer half space i to inner halfspaceo

Ts,p Intensity transmittance of s- and p-polarized light

ts,p Amplitude transmittance of s- and p-polarized light

Tsol Solar transmittance

Tspec Specular transmittance

Ttot Total transmittance

tt Time

vF f Fermi velocity of electrons in thin films (or bulk)

x Displacement in Eq. (4.6)

x The diameter of the particle core

z Mg/(Mg+V) atom ratio

Abbreviations

AR Antireflection

at.% Atomic percent

EC Electrochromic

GIXRD Grazing incidence X-ray diffraction

low-E Low thermal emittance

NIR Near infrared

OS Oblate spheroids

PS Prolate spheroids

RBS Rutherford backscattering spectrometry

SEM Scanning electrom microscopy

UV Ultra-violet

vol.% Volume percent

XRD X-ray diffraction

Contents

Part I: Basics1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232 Energy Efficient Windows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2.1 The Solar Spectrum and Window Performance . . . . . . . . . . . . . 252.1.1 Black Body Radiation and the Solar Spectrum . . . . . . . . . 252.1.2 Basic Light-matter Interaction . . . . . . . . . . . . . . . . . . . . . 272.1.3 Evaluation of Window Performance . . . . . . . . . . . . . . . . . 27

2.2 Low-E and Solar Control Windows . . . . . . . . . . . . . . . . . . . . . 282.3 Chromogenic Smart Switchable Windows . . . . . . . . . . . . . . . . 29

3 VO2-based Thermochromism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313.1 Band structure of VO2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313.2 VO2 based Thermochromic Windows . . . . . . . . . . . . . . . . . . . 32

3.2.1 Spectral and Wavelength-Integrated Optical Properties . . . 323.2.2 Doping and Multilayer Films . . . . . . . . . . . . . . . . . . . . . . 343.2.3 Nanothermochromics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

4 Optics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394.1 Optical Response in Materials . . . . . . . . . . . . . . . . . . . . . . . . . 39

4.1.1 The Complex Optical Constants . . . . . . . . . . . . . . . . . . . . 394.1.2 The Lorentz Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 404.1.3 The Drude Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424.1.4 The Oscillator Model for Real Materials . . . . . . . . . . . . . 424.1.5 Size-dependent Dielectric Functions . . . . . . . . . . . . . . . . 43

4.2 Effective Medium Models for Nanocomposites . . . . . . . . . . . . 454.2.1 The Maxwell-Garnett Theory . . . . . . . . . . . . . . . . . . . . . . 464.2.2 The Bruggeman Theory . . . . . . . . . . . . . . . . . . . . . . . . . . 504.2.3 Tuning the Optical Property of Metallic Nanoparticles . . . 51

4.3 Thin Film Optics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 534.3.1 One Interface: Fresnel Equations . . . . . . . . . . . . . . . . . . . 534.3.2 Two Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 554.3.3 More Interfaces: Multilayer Films . . . . . . . . . . . . . . . . . . 57

5 Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 595.1 Sample Fabrication: DC Magnetron Sputtering . . . . . . . . . . . . 59

5.1.1 The Sputtering of Pure and Mg-doped VO2 Films . . . . . . 605.1.2 The Sputter Deposition of VO2 Nanorods . . . . . . . . . . . . . 60

5.2 Rutherford Backscattering Spectrometery . . . . . . . . . . . . . . . . 625.3 X-ray Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

5.4 Scanning Electrom Microscopy . . . . . . . . . . . . . . . . . . . . . . . . 665.5 Electrical Characterization: Sheet Resistance . . . . . . . . . . . . . . 675.6 Optical Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 695.7 The Determination of Optical Constants . . . . . . . . . . . . . . . . . . 71

5.7.1 The Glass Substrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 715.7.2 Thin Films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

5.8 Determination of Band Gaps . . . . . . . . . . . . . . . . . . . . . . . . . . 746 Input data for Computation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

6.1 VO2 Spheroidal Nanoparticle Inclusions . . . . . . . . . . . . . . . . . 776.2 VO2 Spherical Core-shell Nanoparticle Inclusions . . . . . . . . . . 786.3 Composites of VO2 and ITO . . . . . . . . . . . . . . . . . . . . . . . . . . 79

Part II: Important Results7 Nanothermochromics with Undoped VO2: Modeling Results . . . . . 83

7.1 VO2 Spheroidal Nanoparticle Inclusions . . . . . . . . . . . . . . . . . 837.1.1 VO2 Films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 837.1.2 VO2 Nanoparticles vs. Film . . . . . . . . . . . . . . . . . . . . . . . 84

7.2 VO2 Core-shell Nanoparticle Inclusions . . . . . . . . . . . . . . . . . . 897.3 Composites of VO2 and ITO . . . . . . . . . . . . . . . . . . . . . . . . . . 927.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

8 Mg-doped VO2: Band gaps and Performance Limits . . . . . . . . . . . . 978.1 Band Gaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 978.2 Modeling Results: Mg-doped VO2 Thin Films . . . . . . . . . . . . . 100

8.2.1 Films with Fixed Thickness 50 nm . . . . . . . . . . . . . . . . . . 1008.2.2 Performance Limits for One-layer Films . . . . . . . . . . . . . 1018.2.3 Films with One-layer Antireflection Coatings . . . . . . . . . . 103

8.3 Modeling Results for Mg-doped VO2 Nanospheres . . . . . . . . . 1058.3.1 5-µm-Thick Layer Containing 1 vol.% Nanospheres . . . . 1058.3.2 Performance Limits for Nanospheres . . . . . . . . . . . . . . . . 106

8.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1079 Sputter-deposited VO2 Nanorods . . . . . . . . . . . . . . . . . . . . . . . . . . 109

9.1 Growth Conditions and Morphology . . . . . . . . . . . . . . . . . . . . 1099.1.1 Influence of O2 gas flow . . . . . . . . . . . . . . . . . . . . . . . . . . 1139.1.2 Influence of Substrates . . . . . . . . . . . . . . . . . . . . . . . . . . . 1139.1.3 Influence of Substrate Temperature . . . . . . . . . . . . . . . . . 1139.1.4 Influence of Thickness . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

9.2 Optical Data and Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . 1149.2.1 Experimental Optical Data . . . . . . . . . . . . . . . . . . . . . . . . 1149.2.2 Optical Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

9.3 Possible Growth Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . 11510 Concluding Remarks and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . 11911 Swedish Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

Part I:

Basics

1. Introduction

The industrial revolution has to a great extent raised the material aspects oflife in developed countries as well as in prosperous cities in developing coun-tries: people have abundant food, enjoy the comfort and convenience that aking would have envied 200 years ago [1], have the means to travel half wayaround the globe within a day and access a lot of information with a few clicks.The world is getting smaller and is looking increasingly finite, thanks to theimmense processing power of industry for turning energy and resource intoend products; perhaps too hastily and recklessly. Many have come to realizethat old sources of energy and resources are getting less available, while, theproblem of environmental pollution is deepening.

Most energy-related issues can be included in the following three genres: i)generation; ii) storage; iii) distribution and iv) efficiency, -"to manage a littlewell". i), ii) and iii) points to the new technologies to unlock the potential ofrenewable energies and produce novel batteries and smart power system net-works. iv) however, has existed throughout human history. The ancient Chi-nese people had an efficient way of cultivating rice and fish together so thatno feeding or fertilizing was needed. The quote "to manage a little well" isfrom Isabella Beeton’s masterpiece [2], wherein she documented the systemof keeping up a Victorian well-to-do lifestyle while making sure that every bitof material was put to good use and nothing was lost or wasted. I would like totake this quote as an interpretation of "energy-efficiency" as it does not meandiminishing one’s necessities or even comforts, but cutting back on avoidablewaste and loss by solving a series of design problems. This thesis deals withenergy-efficiency of a fundamental building component, i.e., windows.

Buildings (i.e., the dwellings of modern humans) are among the major en-ergy consumers and have a large environmental impact. Around 30-40 % ofworld primary energy consumption is due to heating, cooling, ventilation,lighting and appliances in buildings [3]. In the industrialized world, these fig-ures tend to be larger. Buildings account for 39% of energy consumption in theUSA [4] (year 2008) and 40-45% in Europe [3] (year 2007). The good newsis the potential for energy saving in buildings is also large, if smart energy-efficient technologies and designs are utilized [4].

Windows are a necessity for buildings for providing lighting and visual con-tact with the outside world, and they are becoming a larger and larger pres-ence in modern architecture. However, in terms of energy efficiency, they area weak link due to the fact that windows very often let in or let out too much

23

energy, and adds to energy consumption especially in summer and winter.Therefore, windows, as they have a big impact on energy consumption/saving,need to be made energy-efficient.

There are many ways to control the energy throughput and daylightingproperties of windows. Low-emittance (low-E) windows and solar controlwindows are the typical energy efficient windows for passive control ofindoor heat loss and solar energy throughput, respectively. These typesof window constantly transmit a part of the light spectrum and block therest. They have very low thermal emittance and are capable of achievinggood heat insulation while maintaining high transparency in the visible.They are already widely in use in many developed countries. Photochromic,thermochromic, electrochromic, and gasochromic materials belong to a classof materials known as chromogenic materials[5]. Chromogenic materials are"smart" and switchable, and therefore offer better options for energ-efficientfenestration. By combining several chromogenic materials, one can worktoward "super fenestration" [6, 7].

The focus of this thesis is on VO2-based thermochromic materials. VO2 hasa switchable infrared optical property when the temperature is changed acrossa critical temperature τcr. It has a great potential for energy efficient fenes-tration [8–12], especially in hot climates [12]. The issues that have hinderedthe practical applications of VO2 in windows are the low luminous transmit-tance and modest solar transmittance modulation. In works covered by thisthesis, a proposal for nanothermochromics is made based on computation re-sults which showed that VO2-based nanoparticles can have very much im-proved luminous properties and solar modulation (Chapter 7). The thesis alsocontains experimental studies on Mg-doped VO2 films that have revealed awell-defined trend between the visible optical band gap of the films and theMg content (Section 8.1). The followup computational studies (Sections 8.2,8.3) show that the moderately Mg-doped films outperform the undoped VO2films, and better performances can be achieved by one layer of antireflectioncoating, but the best performance is still offered by VO2 nanoparticles. As aside project, the growth of sputter deposited VO2 nanorods was investigatedand the findings are presented in Chapter 9.

24

2. Energy Efficient Windows

2.1 The Solar Spectrum and Window Performance2.1.1 Black Body Radiation and the Solar SpectrumA black body by definition is something that absorbs all the incident electro-manetic radiation, i.e., A(λ ) = 1. Black bodies with a temperature of aboveabsolute zero degrees radiate thermal energy in the form of electromagneticwaves. The wavelength-dependent distribution of the radiated energy per unittime per unit volume per unit solid angle per unit wavelength I(λ ,Tt) is de-scribed by Plank’s law:

I(λ ,Tt) =8πhc0

λ 5

1

exp(hc0

λkBTt)−1

(2.1)

where c0, h, kB, λ , Tt refer to the speed of light, Plank’s constant, Boltzmann’sconstant, wavelength and temperature of the radiating black body, respec-tively. A Figure of Plank’s law for different temperatures is shown in Fig 2.1(a).

Integrating Eq. (2.1) over wavelength and solid angle over a half sphere, wehave the Stefan-Boltzmann law, which gives radiated power per unit area ofthe black body Iint :

Iint = σSBT 4t (2.2)

where σt = 5.6705× 10−8 Wm−2K−4. Another important relation, Wien’sdisplacement law, states the relation between the wavelength where the max-imum radiated energy occurs λmax and the temperature of the radiating blackbody Tt :

λmaxTt =CWien (2.3)

with CWien = 2.8978×10−3m K.The Stefan-Boltzmann law and Wien’s displacement law allow us to qual-

itatively estimate that the higher the temperature of the black body, the moreenergy it radiates, and at the shorter wavelength this energy peaks.

25

0

0.5

1

1.5

0.2 0.5 1 2 5 10 20 50 1000

0.5

1

Irra

dia

nce (G

W/m

3)

Effic

iency

Wavelength (µm)

(b)

Solar irradiance

Relative luminous efficiency of the eye

Inte

nsity (A

rb. U

.)

(a)

-50°C0°C

50°C

100°C5505°C (The Sun)

Figure 2.1: Black body radiation at different temperatures (a) and luminous efficiencyof the human eye [13] together with solar irradiance spectra that travels through theatmosphere [14] (b) plotted in the same wavelength scale. Note that in (a) the spectralintensities have comparable scales apart from the one with "5505 C", which is drawnfor convenience.

The sun is a blackbody. The very brightness of the sun is nothing but theblack body radiation. The surface temperature 5778 K (5505 C) of the suncorresponds to the black body radiation at 5505 C, as shown in Fig. 2.1 (a). Itgets partially consumed by absorptions on its path etc. and reaches the earth asa not so perfect bell-shaped spectrum, illustrated in Fig. 2.1(b) and availableat our windows on a sunny day. It is important to note that solar energy is onlylimited to 0.25 < λ < 3 µm, which does not overlap with thermal radiationfrom objects at < 100C. Therefore it is possible to devise materials withdifferent properties regarding solar radiation and thermal radiation from theindoor environment.

Solar energy is divided into 3 intervals: λ < 0.4 µm is the ultra-violet (UV)which contributes to vitamin D formation or causes skin damage. λ > 0.7 µmis the near infrared (NIR) which contributes to solar heat, and the interval inbetween, 0.4 < λ < 0.7 µm is the visible range where the sensitivity of thehuman eyes spans with a peak at 0.555 µm, as shown in Fig. 2.1(b).

26

27

2.1.2 Basic Light-matter InteractionWhen placed in an incident light beam, a material can respond in three ways,it can transmit, reflect or absorb the electromagnetic energy. These processesmust obey the energy conservation law at each wavelength:

T (λ )+R(λ )+A(λ ) = 1 (2.4)

where, λ is the wavelength and T (λ ), R(λ ), A(λ ) refer to the wavelength-dependent transmittance, reflectance, and absorptance, respectively. Part ofthe absorbed energy will heat up the material and give rise to an emittance.For energy conservation Kirchhoff’s law gives:

A(λ ) = εtherm(λ ) (2.5)

where εtherm(λ ) is the wavelength-dependent thermal emittance, which is de-fined as the fraction of the blackbody radiation released at a certain temper-ature and wavelength. The temperature-dependent total hemispherical emit-tance Etherm per unit area of sample surface is then given by:

Etherm(Tt) =! ∞

0A(λ )I(λ ,Tt)dλ/Iint (2.6)

2.1.3 Evaluation of Window PerformanceThe primary functions of architectural windows are visual contact throughthe window, day-lighting and some degree of heat control. These functionsare achieved by letting through or blocking parts of the solar and infraredspectra. The wavelength-integrated properties at an angle of incidence θ , lu-minous transmittance Tlum and reflectance Rlum, solar transmittance Tsol andreflectance Rsol , are some of the important factors used to characterize a win-dow. They are defined as:

Tlum,sol(θ ) =!

dλ T (λ ,θ )ϕlum,sol(λ )/!

dλ ϕlum,sol(λ ) (2.7)

Rlum,sol(θ ) =!

dλ R(λ ,θ )ϕlum,sol(λ )/!

dλ ϕlum,sol(λ ) (2.8)

where ϕlum is the luminous efficiency of the human eye, and ϕsol is the solarirradiance spectrum when the sun is standing 37! above the horrizon (corre-sponds to air mass 1.5 or AM 1.5). The spectra of ϕlum and ϕsol are illustratedin Fig. 2.1.These wavelength-integrated properties allow us to condense the informa-

tion from complex spectra into easily comparable numbers and hence are

frequently used in academia and industry for performance evaluations, es-pecially, for windows with near-neutral colors. For colored windows, moremeasures are needed to take into account the visual comfort of users.

For practical windows, it is often important to limit a certain wavelengthrange in the near-infrared (NIR) or infrared region (IR) while keeping gooddaylighting and see-through properties in order to achieve energy efficiency.This can be easily achieved due to the fact that the human eye is not sensitiveoutside the interval 0.4 < λ < 0.7 µm and it is blind in λ > 0.78 µm, asshown in Fig. 2.1. By blocking part or all of the spectrum within 0.7 < λ < 50µm while keeping high transparency in 0.4 < λ < 0.7 µm (i.e. high Tlum),the indoor heat loss or solar heat gain can be largely reduced on a windowwith good see-through properties. Such spectral selectivity is introduced towindows by coatings, usually on the surfaces facing the confined air (or gas,or vacuum) spaces of multiply-glazed windows. In Sections 2.2 and 2.3, Iintroduce some of the typical window coatings for energy efficiency.

2.2 Low-E and Solar Control WindowsLow-emittance (Low-E) coatings and solar control coatings and are the twomost widely used window coatings for controlling solar heat gain. An ideal-ized definition gives:

low-E

T (λ ) = 1 for 0.4 < λ < 3 µm

R(λ ) = 1 for 3 < λ < 50 µm

solar control

T (λ ) = 1 for 0.4 < λ < 0.7 µm

R(λ ) = 1 for 0.7 < λ < 50 µm

Low-E coatings are common in colder high latitude regions where peoplelike to keep warm inside by letting in the solar heat, while stopping the indoorheat from escaping through the window to the outside. Low-E coatings aretypically installed on the surface of the inner pane facing the confined spaceof a multiple glazed window. Solar control coatings, on the other hand, aredesigned for warmer regions, where maximum blocking of solar heat is desir-able. Solar control coatings are typically installed on the surface of the outerpane facing the confined space of a multiply glazed window.

These coatings are usually based on transparent conducting oxides(TCO) such as tin-doped indium oxide (known as ITO or indium tinoxide and denoted as In2O3:Sn), fluorine-doped tin oxide (FTO, SnO2:F),aluminium-doped zink oxide (AZO, ZnO2:Al) or multilayer sandwichstructures comprising dielectric layers and thin layers of noble metals [15].

28

Solar control coatings and low-E coatings highly reflect the thermal infraredand have a low total integrated hemispherical thermal emittance Etherm. Anuncoated glass is strongly absorbing in the thermal infrared λ > 3 µm and hasa Etherm as large as 0.87. When coated with solar control or low-E coatings,Etherm of the coated glass can be significantly reduced to only a fraction ofEtherm of the uncoated glass. Etherm as low as ∼ 0.2 with 75%< Tlum < 80% isachievable by using coatings based on TCO [15, 16]. An even lower Etherm ≪0.2 can be achieved by using thin layers of Ag. [11]

2.3 Chromogenic Smart Switchable WindowsSmart switchable windows are another class of energy efficient windows. In-stead of offering a static optical control, as solar control and low-E windowsdo, smart switchable windows have variable optical properties. These can bebased on polymer liquid crystal devices, suspended particle devices and chro-mogenic materials. Here I only talk about chromogenic materials.

Chromogenic materials exhibit a color change triggered by external stimuli.They are called electrochromic (e.g. WO3) if the external stimulus is electri-cal current or voltage, thermochromic (e.g. VO2) if the stimulus is tempera-ture, photochromic (e.g. some sunglasses) if the stimulus is energetic light,gasochromic (e.g. some Mg alloys) if the stimulus is adsorbed gases.

Figure 2.2: Illustration of basic design of an electrochromic (EC) device, taken from[15].

Electrochromic (EC) devices comprise 5 layers, as illustrated in Fig. 2.2,and switch color when ions are moved by the external electric field in or out

29

of the electrochromic film. The electric power is only needed for the switch-ing, and the after-switch color of the device can be kept for days at least with-out any power supply. Among all the chromogenic materials, electrochromicmaterials are the most studied and some prototypes are already used in com-mercial products such as windows for buildings, motor cycle helmets, space-craft etc. [17]. Especially, electrochromic windows, whose optical propertiesare switchable when a voltage/current is applied, can give the highest energysaving, which can be as high as 4.5% on the annual energy use in the USAwhen they are installed in commercial and residential buildings [18].

In this thesis I discuss thermochromic materials, which can be as simpleas a single layer structure with switchable optical properties depending ontemperature. Thermochromic materials by themselves cannot save as muchenergy as electrochromic materials, but their sheer simplicity makes them easyto combine with other energy-efficient functional windows to achieve "superfenestration" [6, 7].

Many oxides such as NbO2, FeSiO2, VO2, Ti2O3, V2O3 etc. exhibit ther-mochromism [11]. Among them, VO2-based materials are the most interest-ing. VO2 changes from infrared transmitting to infrared reflecting when it isheated across a critical temperature τcr ∼ 68C. In the next chapter I explainin detail the switching mechanism, optical properties and challenges of ther-mochromic VO2 for energy efficient applications.

30

3. VO2-based Thermochromism

It is well-known that VO2 undergoes a reversible metal-to-insulator structuraltransition at a critical temperature τcr ∼ 68 C in the bulk. At τ > τcr VO2is tetragonal metallic and infrared reflecting; at τ < τcr, it is monoclinic in-sulating and infrared transparent. This phenomenon, namely thermochromismin VO2 was first discovered by Morin [19]. To take advantage of interest-ing functions due to temperature-dependent switching, VO2 has been used invarious applications for sensing, and in optical communication such as in IR-bolometers, photonic crystals etc., and its application as a passive switchablewindow glazing for energy efficient applications has also been discussed fordecades [9, 15].

In this chapter I look into the band structure of VO2, and discuss the advan-tages and problems of conventional VO2-based films for energy efficient win-dow applications. Finally I address the great opportunities that VO2 nanopar-ticles can offer.

3.1 Band structure of VO2VO2 undergoes a first-order metal-to-insulator transition when the temper-ature is altered across τcr. The high temperature metallic state (τ > τcr) istetragonal. This transition was recently shown to be of Mott-Hubbard type[20, 21]. The structural transition of VO2 is of great interest in the field of cor-related electron systems. However, rigorous discussions on this subject lie outof the scope of this work, and therefore we will not attempt to go any furtherthan briefly introduce the band structure of VO2.

Fig. 3.1 shows the schematic band structure based on work by Goodenough[22]. For the metallic state of VO2 corresponding to τ > τcr (3.1 (a)), thecrystal structure is tetragonal with space group P42/mnm. Under the influenceof the crystal field, the V 3d t2g band splits into d|| and π∗ sub-bands. Thehybridization of the V π∗ band and the O 2pπ band causes the π∗ band to bemoved upwards, whereas the weak hybridization between the V d|| band andthe O 2p band leaves d|| at the lowest level of all V 3d bands. The metallicbehavior is caused by the partially filled d|| and π∗ band. The gap between O2p and the Fermi level EF is the band gap Eg, corresponding to an optical gapwith higher energy.

31

(a) (b)

Figure 3.1: Schematic illustration of the band structure of VO2 based on the seminalwork by Goodenough [22]. (a) corresponds to tetragonal metallic state (τ > τcr); (b)corresponds to the monoclinic semiconducting state (τ < τcr).

For the semiconducting state of VO2 (τ < τcr), V atoms dimerize and twistin a zig-zag way along the c-axis, causing the otherwise tetragonal lattice todistort into a monoclinic lattice with space group P21/c. The d|| band splitsinto two sub-bands, the upper d|| band and the lower d|| band. The lower d||band is completely filled leaving the π∗ band completely empty. Eg1 denotesthe energy gap between lower d|| and π∗, Eg2 denotes the gap between O 2pand π∗, Eg3 denotes the gap between lower d|| and upper d||.

Various studies based on photoemission spectroscopy (PES) [23–27], X-ray absorption spectroscopy (XAS) [25, 27], ultra-violet reflection [23], elec-tron energy loss spectroscopy (EELS) [28], optical measurements [29–31],as well as density functional theory (DFT)-based theoretical studies such aslocal density approximation (LDA) [32] and generalized gradient approxima-tions (GGA) [33] have been done to determine the energy gaps in the VO2band structure. In general, for the semiconducting state, the optical transitionat ∼ 0.6 eV is attributed to Eg1; transitions above 1eV can be attributed to Eg2and Eg3.

3.2 VO2 based Thermochromic Windows3.2.1 Spectral and Wavelength-Integrated Optical PropertiesFig. 3.2(a) and (b) show typical spectral reflectance R(λ ) and transmittanceT (λ ) expected from a 50 nm thick VO2 film backed by a glass substrate.Strong optical absorption is present in both states in the UV and in a part ofthe visible, due to the interband transition across the larger gaps, as illustratedin Fig. 3.1. At semiconducting state, T (λ ) ∼ 40% where the relative lumi-

32

nous efficiency of the eye has a peak. This increases with the wavelength andreaches T (λ ) ∼ 75%-80% at 2500 nm. For the metallic state, T (λ ) is almostthe same as that of the semiconducting state in the range where the eye ismost sensitive, whereas, after a broad maximum around 700-800 nm, T (λ )decreases progressively at longer wavelengths and reaches T (λ ) ∼ 20% at2500 nm. Meanwhile, R(λ ) for the metallic state increases with wavelengthfor λ > 1000 nm and reaches R(λ )∼ 40% at 2500 nm. Comparing the opticaldata for the semiconducting state and the metallic state, it can be seen thatthermochromic switching is clearly present. The most efficient switching inboth R(λ ) and T (λ ) occurs at λ ∼ 2500 nm.

0

0.5

1

1.5

0 500 1000 1500 2000 2500 0

0.5

1

Irra

dian

ce (

GW

/m3 )

Effi

cien

cy

Wavelength (nm)

(c)

Solar irradiance

Relative luminous efficiency of the eye

0

20

40

60

80

Tra

nsm

ittan

ce (

%)

Semiconducting

Metallic(b)

0

20

40

60

80

Ref

lect

ance

(%

)

Semiconducting

Metallic

(a)

Figure 3.2: Typical spectral reflectance (a) and transmittance (b) of a 50 nm thick VO2film on a glass substrate in semiconducting state (τ < τcr) and metallic state (τ > τcr)as can be found in work by Mlyuka [34], plotted together with the solar irradiancespectrum and the spectral luminous sensitivity of the eye (c). Note that the same dataas in (c) was given in Fig. 2.1 (b).

33

The definitions of wavelength-integrated optical properties introduced inEq.(2.7) gives Tlum(τ < τcr) ≈ Tlum(τ > τcr) ≈ 37.4%, and Tsol(τ < τcr) ≈42% and Tsol(τ > τcr)≈ 35.5%. Here we introduce solar transmittance modu-lation ∆Tsol , which is frequently used to evaluate the thermochromic switchingperformance of VO2 films. It is defined by:

∆Tsol = Tsol(τ < τcr)−Tsol(τ > τcr) (3.1)

Eq. (3.1) gives ∆Tsol ≈ 6.5% for the optical data in Fig.3.2.Typically for conventional VO2 single layer films, one has Tlum ∼ 50% and

∆Tsol ∼ 5%. The relatively small solar transmittance modulation ∆Tsol stemsfrom the fact that thermochromic switching is most effective at long wave-lengths where the solar irradiance intensity is low. If one requires ∆Tsol ∼ 10%,it would limit Tlum to less than 40%, which is too low for practical application.Apart from low ∆Tsol and Tlum, the critical temperature is too high, τcr ∼ 68C.These challenges need to be effectively addressed in order to achieve practi-cal window applications with VO2-based films. Sections 3.2.2 and 3.2.3 aredevoted to these issues.

3.2.2 Doping and Multilayer FilmsThe critical temperature can be influenced by doping. Studies have shown thatdoping with W, Mo, Ta, Nb and Eu can decrease τcr, while doping with Ge,Al and Ga will increase τcr [22]. In particular, a few percent of W dopingcan decrease τcr to around room temperature [35–38]. However, Tlum is stilllow in this case. Studies on other dopants have been done as well. F-dopingwas shown to decrease τcr [39, 40], and significantly increase Tlum[41] withsomewhat less pronounced switching [41].

A recent study done in our group by Mlyuka et al. showed that Mg-dopingcan jointly enhance Tlum and reduce τcr [36]. A band gap widening effect ofMg-doping was also observed [6, 42, 43], in which Mg-doping was foundto increase the larger optical gap while leaving the small gap Eg1 almost un-changed. A theoretical study based on density functional theories [44] haveconfirmed that Mg-doping causes a lowering of optical absorption in the vis-ible range. This effect is responsible for the increased Tlum in Mg-doped VO2films.

Straight forward optical designs such as multilayer antireflection coatingshave been devised to increase Tlum and ∆Tsol , such as TiO2/VO2 [45],TiO2/VO2/TiO2 [46], SiOx/VO2/SiOx [47], and TiO2/VO2/TiO2/VO2/TiO2[34, 48]. The best of them exhibit ∆Tsol ∼ 12%, Tlum ∼ 45% [34, 48], or∆Tsol ∼ 15.1%, Tlum(τ < τcr) ∼ 49.5%, Tlum(τ > τcr) ∼ 44.8% at normalincidence [45]. A survey of the performances of VO2 films with ARtreatments can be found in the literature [49].

34

To summarize the above, a smart choice of dopants can decrease τcr to roomtemperature, and therefore τcr is not a pressing issue any longer. However, thejoint improvement of Tlum and Tsol is only marginal, despite all the scientificeffort described above. For this reason VO2 is not yet ready for practical en-ergy efficient window applications, and we need something new to tackle thisproblem.

3.2.3 NanothermochromicsVO2 nanoparticles of various shapes have been successfully fabricated by var-ious physical and chemical methods and thermochromic switching has beendemonstrated as mentioned in Refs [50, 51] and references therein. There hasbeen a lot of interest in extracting novel properties that are non-existant in thefilm or the bulk from nanoparticles [52, 53]. Hence, it is only natural to expectnew optical properties from materials based on VO2 nanoparticles.

In 2010 a new idea, namely nanothermochromics was proposed in work[50] covered by this thesis. Nanothermochromics considers dilute suspensionof VO2 nanoparticles embedded in a dielectric medium. This structure hastwo advantages that are well-suited to treating the deficiencies of conventionalVO2 films:• Dilute suspensions of VO2 nanoparticles with little scattering can improve

Tlum in the nanocomposite layer.• By taking advantage of a tunable localized plasmon resonance in metallic

state VO2 nanoparticles, the optical absorption at τ > τcr can be shifted toa shorter wavelength in the near infra-red range where the solar intensityis high, thereby improving the modulation of solar transmittance throuput∆Tsol

The computational study covered in this thesis shows that nanother-mochromics works very well indeed, and the luminous transmittance andsolar transmittance modulation can be significantly improved. As is shownlater in Chapter 7.1, nanocomposite layers with VO2 solid nanospheres yielda spectral transmittance as shown in Fig. 3.3, with the above-mentionedadvantageous features specific for VO2 nanoparticle based layers: Tlumis decidedly higher, and a strong absorption band in the metallic state islocated around 1.2 µm, giving rise to a much larger solar transmittancemodulation ∆Tsol compared to data of films in Fig. 3.2. Moreover, results inChapter 7.2 show that VO2 hollow spheres will improve ∆Tsol even more.Another advantage of VO2 nanoparticles is that by embedding them in otherfunctional layers, one can achieve multifunctional coatings. In Chapter 7.3, acomputation considers a mixture of VO2 and ITO and yields a thermochromiclow-emittance coating with moderately high Tlum, large ∆Tsol and low Etherm.

The above mentioned promises can become reality only if the fabricationof VO2 nanoparticles is possible. It is well-known that VO2 with intermediatevalence V4+ is very difficult to fabricate, the nanoparticle form even more so.

35

0

0.5

1

1.5

0 500 1000 1500 2000 25000

0.5

1

Irra

dia

nce (G

W/m

3)

Effic

iency

Wavelength (nm)

(b)

Solar irradiance

Relative luminousefficiency of the eye

0

20

40

60

80

Tra

nsm

itta

nce (%

) Semiconducting

Metallic

(a)

Figure 3.3: Typical spectral transmittance (a) of a dielectric host containing VO2nanoparticles with the same mass thickness as the sample thin film presented in Fig.3.2 backed by a glass substrate. It is plotted together with the solar irradiance spectraand the spectruml luminous sensitivity of the eye (b).

Despite the difficulties, many experimental works have demonstrated success-ful fabrication of VO2 particles.

First of all, nanoparticles of VO2 embedded in a silica matrix have been fab-ricated using hydrogen-aided reduction of a glass containing V2O5 [54, 55],and ion implantation of V and O ions in a silica host [56, 57] and sol-gel de-position [58]. VO2 particles with near-symmetric shapes have been made bywet chemistry [59–66], molten salt synthesis [67], thermolysis [68], pyroly-sis [69], confined-space conbustion[70], and pulsed laser deposition[71–73].Doped particles of VO2 [61, 64, 68, 74, 75] and VO2:W deposited on SiO2nanoparticles[76], VO2 nanorods [77] W-doped VO2 nanowires [78] and VO2nanosheets [79, 80] have been prepared by wet chemistry.

Core-shell particles have also been extensively demonstrated by experi-ments too. Hollow nanospheres and microspheres of VO2 have been made bychemical sythesis[81], and hollow rodlike VO2 by gas-phase synthesis[82].Transparent polymers with inclusions of VO2 coated SiO2 nanoparticles and

36

similar composites have been made[83]. SiO2 coated with VO2:W have alsobeen produced too [84]. VO2 shells on SiO2 have been prepared by pulsedlaser deposition [85] and VO2 shells on carbon microspheres by means ofchemical methods [86, 87]. VO2 inverse opal and nanoscrolls have also beendemonstrated.

Recently, the interest in synthesizing VO2 nanoparticle compositesis booming due to the advantageous properties for energy-efficientwindow coatings. VO2-ITO composite films [88]. W-doped VO2 andpolysiloxane composites [89], VO2 and ZrV2O7 composites [90], VO2nanoparticle-assembled nanocomposite films [91] were demonstrated.Furthermore, Incorporation of VO2 [92], SiO2/VO2 core/shell structures[93], VO2-Sb:SnO2 composites [94], Mg-doped VO2 nanoparticles [42] intoflexible foils have been realized too.

To summarize, VO2-based Nanothermochromics for energy-efficent win-dows is not only theoretically well-established but also becoming increasinglyeasier to realize. Windows equipped with high-performance VO2 based ther-mochromic components are very likely to finally find their way to our homesin the near future.

37

4. Optics

4.1 Optical Response in Materials4.1.1 The Complex Optical ConstantsA material’s interaction with light can be roughly divided into two kinds, re-fraction and extinction. The refraction process, on the one hand, involves achange of the wavelength, speed and direction of light passing through a ma-terial while preserving the frequency (i.e. energy) of the incident light. Theextinction process consisting of scattering and absorption, which on the otherhand, removes energy from light and causes attenuation of the incident lightamplitude along the path.

Complex optical constants are suitable for describing both refraction andextinction. There are two sets of most commonly used complex optical con-stants: the complex refractive index N and the complex dielectric constant ε ;both are wavelength-dependent. They are inter-related by the simple expres-sion:

ε = N2 (4.1)

with

N = n+ ik (4.2)

and

ε = ε1 + iε2 (4.3)

Hence another relation:ε1 = n2 − k2

ε2 = 2nk(4.4)

with the inverse relation:

n =

[

12

(

(ε21 + ε2

2 )1/2 + ε1

)

]1/2

k =[

12

(

(ε21 + ε2

2 )1/2 − ε1

)

]1/2 (4.5)

39

The complex refractive index comes from the solution of the wave equationbased on the Maxwell equation. This solution describes propagating electro-magnetic wave in matter, by

E = E0 exp[

iω(

Nxc0

− tt

)]

(4.6)

where E0 and E are the amplitudes of the incident electromagnetic wave andthe electromagnetic wave in matter, respectively. ω denotes the frequency ofthe electromagnetic wave, tt denotes time, and x is the position vector. Thepropagating electromagnetic wave is influenced by the matter via the complexrefractive index: n changes the phase velocity, and k attenuates the amplitudealong the light path. The intensity EE∗ contains an exponential attenuationfactor exp(−2kωx/c) = exp(−4πkx/λ ). Factor 4πk/λ is called the absorp-tion coefficient, commonly denoted α .

The complex dielectric constant is more meaningful from a microscopicview point. Contributions due to optical response from different microscopicelements are additive. Such contributions are: dipolar arising from the perma-nent dipoles; ionic due to displacement of charged ions with respect to otherions; electronic due to displacement of electrons with respect to ion cores andthe inter-band transition of valence electrons. These microscopic responsesare nothing but the oscillation of charges under the force generated by the ex-ternal electromagnetic field, i.e. light, and can be represented as oscillators.In 4.1.2 and 4.1.3, I introduce two basic oscillator models, the Lorentz modeland the Drude model.

4.1.2 The Lorentz Model: Oscillator Model for Bound ChargesThe Lorentz Oscillator is a model developed for the description of boundcharges. Here, for convenience, we consider the case that these charges areelectrons. The model is analogous to that of harmonic oscillators: the externalelectromagnetic force -eE(ω) is the force responsible for acceleration of theelectrons; the assumption also includes a restoring force m∗

eω20 r that the elec-

trons are bounded with, and a "friction" m∗eΓ

drdtt

with a damping effect on the

oscillator. The equation representing the balance of the forces is given by

−eE(ω) = m∗e

d2rdt2

t+m∗

eΓdrdtt

+m∗eω2

0 r (4.7)

where r and m∗e are the displacement and the effective mass of the elec-

trons, respectively. By putting E(ω) = E00e−iωtt and r = r00e−iωtt , the so-lution gives:

40

r00 =−eE00/m∗

e

(ω20 −ω2)− iωΓ

(4.8)

The displacement of charged particles gives rise to an induced dipole moment

p(ω) =−er =e2Em∗

e

1(ω2

0 −ω2)− iωΓ(4.9)

Summing up the dipoles and assuming a charge density ne, we have the macro-scopic polarization

P(ω) = nep =nee2E

m∗e

1(ω2

0 −ω2)− iωΓ(4.10)

By using the expression for dielectric displacement

D = ε0εE = ε0E+P (4.11)

with D denoting the dielectric displacement and ε0 denoting the dielectricpermittivity in empty space, we obtain the complex dielectric constant

ε(ω) = 1+ω2

p

(ω20 −ω2)− iωΓ

(4.12)

where the plasma frequency ωp is defined as:

ωp =

(

nee2

ε0m∗e

)1/2

(4.13)

Eq. (4.12) is frequently referred to as the Lorentz model dielectric constant.The real and imaginary parts of ε are written

ε1(ω) = 1+ω2

p(ω20 −ω2)

(ω20 −ω2)2 +ω2Γ2

ε2(ω) =ω2

pωΓ(ω2

0 −ω2)2 +ω2Γ2

(4.14)

The optical properties of ionic crystals and inter-band transitions of valenceelectrons can be described very well with the Lotentz model.

41

4.1.3 The Drude Model: Oscillator Model for Free ElectronsThe Drude model assumes a free electron gas oscillating in the electromag-netic field. The designation "free" means that the restoring force representingthe "bound" effect in the Lorentz model is absent, i.e. ω0 ≈ 0. By introducinga scattering time τsc = 1/Γ, Eq. (4.12) in this case becomes

ε(ω) = 1−ω2

p

ω2 + iω/τsc(4.15)

and separating the real and imaginary part gives

ε1(ω) = 1−ω2

pτ2sc

1+ω2τ2sc

ε2(ω) =ω2

pτsc

ω(1+ω2τ2sc)

(4.16)

Eq. (4.15) and Eq. (4.16) are usually refered to as the Drude model opticalconstants. In the relatively high frequency range ω ∼ ωp, if damping is neg-ligible i.e. h/τsc ≪ hω , hωp, then ε(ω) is progressively real, and expressedas:

ε(ω) = 1−ω2

p

ω2 (4.17)

The Drude model is extensively used to describe optical properties of ma-terials with free electron-like properties, such as alkali metals, noble metals,IIIA metals, some transition metal nitrides, TCOs etc.

4.1.4 The Oscillator Model for Real MaterialsThe dielectric constant of real insulating materials usually contains severalLorentz oscillators; that of metallic materials contains one additional Drudeterm. All of these effects are additive so we have for insulators

ε(ω) = ε∞ +∑n

An

(ω2n −ω2)− iω/τsc,n

(4.18)

and for metals:

ε(ω) = ε∞ +∑n

An

(ω2n −ω2)− iω/τsc;n

−ω2

p

ω2 + iω/τsc(4.19)

42

where ωn and An correspond to characteristic band-to-band transition energiesand their spectral weight, respectively. 1/τsc;n is proportional to the width ofeach transition and ε∞ represents the dielectric background as a result of thecontribution from core electrons (bound charges). Typically, 1 ≤ ε∞ ≤ 10.

Calculations based on the electron density of Ag fromEq.(4.13) leads to ωp = 9.2 eV as the energy where ε1 is ex-pected to approach 0. However, the positive contribution frombound charges from below 6 eV lifts up the ε1(ω) v.s. ω curveso that the ε1(ω) crosses 0 at 3.9 eV instead of at 9 eV. Thisnew plasma frequency ω∗

p is called the screened plasma fre-quency and can be obtained by

ω∗p = ωp/

√ε∞ (4.20)

At this frequency the free electron approximation of ε(ω), atrelatively high frequency provided with negligible dampingh/τsc ≪ hω , hωp, expressed as:

ε(ω) = ε∞

(

1−ω∗2

p

ω2

)

(4.21)

goes to 0.

Silver has ε∞ ∼ 5, hωp ∼ 9.2 eV, h/τsc ∼0.02 eV. hω∗p is then

obtained to be ∼ 4.1 eV, not that far from the observed 3.9 eV.

Screened plasma frequency: the famous example of silver

♥

4.1.5 Size-dependent Dielectric FunctionsHere we consider particle sizes in the nano realm, i.e., much smaller thanwavelength. The same quantum confinement effect as observed in semicon-ductor nanocrystals quantum dots is usually negligible in metal nanoparticles[95]; instead, the confinement effect on the electron mean free path value ℓplays a significant role in the optical response. The mean free path ℓ is definedas the path length that the electron travels between two successive scatteringevents, defined as:

ℓ= vFτ f (4.22)

43

with vF f being the Fermi velocity and τ f being the mean scattering time inthin films (or bulk).

The extra scattering events against the particle boundary give increasingcontributions to the plasma damping (reduces the mean scattering time τ f ) asthe size of the particles gets smaller, and may even become dominating whenthe size gets smaller than ℓ. This confinement effect of particle boundary canbe quite profound in metals like Cu, Ag, Au with room temperature valuesℓ ∼ 42, 52 and 42 nm, respectively. The mean scattering time τp in metallicnanoparticles can be written as a function of the particle diameter Dp:

1τp

=1τ f

+2vF

Dp(4.23)

If the dielectric function for thin films or the bulk ε f is given, the dielectricfunction corrected for particle size εp can be written as:

εp = ε f − εDf + εD

p (4.24)

where the Drude (D) type terms are given by

εDf = 1−

ω2P f

ω(ω + i/τ f )(4.25)

and

εDp = 1−

ω2Pp

ω(ω + i/τp)(4.26)

Here ωPp, f denotes the plasma frequency in particles or films. The aboveprocedure can be found in literature [96].

It is known that the dielectric function of particles is not necessarily thesame as that of bulk, therefore a discussion based on Eq. (4.23) is necessaryin order to determine an appropriate εp for particles. For such discussions, itis convenient to rewrite Eq. (4.23) as

τp =τ f

1+2vF f τ f

Dp

=τ f

1+2ℓ f

Dp

(4.27)

with ℓ f being the electron mean free path in the film. It is given by ℓ f = vF f τ f .

For noble metals in which electrons have a high Fermi velocity and large

mean free path, the term2vF f τ f

Dp(i.e.

2ℓ f

Dp) can be large and the procedure of

44

Eq. (4.24) is necessary. In contrast, for materials with electron mean free paths

fullfilling ℓ f ≪ Dp, the term2vF f τ f

Dpis negligible, so that we have τp = τ f and

thus εp = ε f .

4.2 Effective Medium Models for Nanoparticles andComposite CoatingsA composite coating is one class of inhomogeneous material. For the opticalproperty of such a material, the quasistatic approximation is valid if the sizeof inhomogeneity (e.g. particles, clusters etc) satisfies the condition [97]:

πεmax1/2Dp/λ ≪ 1 (4.28)

where Dp is the diameter of the particle inclusions in the composite and εmaxdenotes the largest of the dielectric functions in the material.

This means that the particle is so much smaller than the wavelength, so thatthe electromagnetic field in the particle is approximately constant. If the in-homogeneity also produces no scattering, the composite material containingsuch inhomogeneities can be considered to have the same optical propertiesas an homogenous material, i.e., the probing light of sufficiently large wave-length cannot tell the difference between the two. The optical property of theinhomogeneous material can be replaced by those of an homogenous materialcharacterized by an effective dielectric function εe f f . Such an inhomogeneousmaterial can be considered to be an "effective medium". The effective mediumtheory is developed to offer a view on the homogenized optical response ofthis kind of composite material, based on mean field theory.

The inhomogeneities in such materials can have different microstructuresand the approximation scheme for the optical properties should be chosen ac-cordingly. The general approach is as follows: for microstructures comprisedof isolated particles, the Maxwell-Garnett theory [98, 99] is appropriate; formicrostructures containing topologically equivalent units, the Bruggeman the-ory [100] should be used instead. The derivation of the Maxwell-Garnett the-ory and the Bruggeman theory for optical properties can be found in the litera-ture [101]. Both theories can be generalized to approximate for other physicalproperties such as electronic conductivity, heat conductivity etc. Here we limitthe discussion to their application for optical properties.

It is well-known that nanocomposite materials can have novel optical prop-erties that are non-existing in film or bulk. The effective medium theory per-tains to a small particle limit and offers an approximate guideline for spectraltuning of optical properties of nanoparticle composites. This is discussed atthe end of this chapter.

45

4.2.1 The Maxwell-Garnett TheoryThe Maxwell-Garnett theory [98, 99] is widely used for estimating opticalproperties of composite layers containing isolated particles. It works bestwhen the particle filling factor f (i.e. volume fraction) is small, such as in thecase of a dilute suspension of nanoparticles in a matrix [102, 103].

Here we describe the procedures for applying the Maxwell-Garnett theoryused in this work. Given that the particle sizes are much smaller than thewavelength as described by Eq. (4.28), the particles are far enough from eachother not to invoke particle-particle interactions. The Maxwell-Garnet theorycan be used to obtain the effective dielectric function εMG by

εMG = εm

1+23

f αpol

1− 13

f αpol

(4.29)

where εm is the dielectric function of the matrix and f is the "filling factor"i.e., the volume fraction occupied by the particles. The parameter αpol is thepolarizability of the particles, which depends on the shape and arrangementof the particles.

4.2.1.1 Solid SpheroidsFor aligned solid spheroidal particles, as discussed in Chapter 7.1, the polar-izability and can be written

αpol =εp − εm

εm +L(εp − εm)(4.30)

where εp is the dielectric function of the particles and the parameter L is theappropriate depolarization factor, which takes particle shape into account. Forrandomly oriented ellipsoidal particles, as shown in Fig. 4.1 (d), αpol can beobtained by averaging αpol over the three principal axes of the ellipsoid ac-cording to

αpol =13

3

∑i=1

εp − εm

εm +Li(εp − εm)(4.31)

where Li denotes the depolarization factors along their three principal axes.The depolarization factors Li fullfil ∑Li = 1. Spheres, as shown in Fig. 4.1

(a), have L1 = L2 = L3 = 1/3. The aspect ratio m is defined in Fig. 4.1 aswell. Spheres have m = 1. Oriented nanoparticles with spheroidal shapes withm 6= 1, as shown in Fig. 4.1 (b) and (c), can be taken into account by modifyingthe depolarization factor L as follows:

46

For prolate spheroids (PS) with L1 < L2 = L3 and m > 1, the depolarizationfactors are related to the major (a) and minor axes (c) of the spheriods by [104]

L1 =1− e2

PS

2e3PS

(

ln1+ ePS

1− ePS−2ePS

)

(4.32)

and L2 = L3 = (1−L1)/2, where

ePS = [1− (c/a)2]1/2 (4.33)

For oblate spheroids (OS) with L1 = L2 < L3 and m < 1, the depolarizationfactors are related to the major (a) and minor axes (c) of the spheriods by [104]

L3 = (e−3OS + e−1

OS)(eOS − arctaneOS) (4.34)

and L1 = L2 = (1−L3)/2 where

eOS = [(a/c)2−1]1/2 (4.35)

For the situations shown in Fig. 4.1 (b) and (c), in which light is incidentnormal toward the layers with oriented nanoparticles with the electric fieldorthogonal to the high-symmetry axes of the nanoparticles, the depolarizationfactors are L3 and L1, respectively.

4.2.1.2 Nanoparticle AggregatesIn real nanoparticle samples, it is not certain that all the nanoparticles are per-fectly dispersed and homogenously distributed. Particles can stick, to formchains and clusters, and thus dipole-dipole interaction between the particlescan no longer be disregarded. Fortunately, the optical properties of such par-ticle aggregates can be estimated. Previous studies [103] have shown that onecan construct effective depolarization factors Li,e f f from examining the res-onance characteristics of the aggregates [105] to represent the dipole-dipoleinteraction effect. Le f f ,i for several geometrical configurations extracted bythe literature [103] are shown in Table 4.1

Unlike the spheroids, the effective depolarization factor Le f f ,is of particleaggregates does not add up to unity, but rather satisfies the relation ∑Le f f ,i ∼1. This approach was used in paper VIII to calculated the optical properties ofchains and fcc clusters of ITO particles dispersed in polymer electrolyte.

4.2.1.3 Core-shell structuresFor randomly-oriented core-shell nanoparticles the polarizability αpol is givenby

47

(a)

(b)

(c)

(d)

Figure 4.1: Schematic drawings for composites of solid nanoparticles with a dielectricfunction εp embedded in a medium with dielectric function εm. The nanoparticles aretaken to be spherical (a), oriented prolate (b), oriented oblate (c), and random prolateand oblate (d) with major axis a and minor axis c respectively. An aspect ratio m isdefined for prolate and oblate spheres.

48

Table 4.1: Equivalent depolarization factors Le f f ,i for shown geometrical configura-tions of nanoparticles determined by literature.[103]

Geometrical Configuration Le f f ,1 Le f f ,2 Le f f ,3

Single sphere 1/3 1/3 1/3

Double sphere 0.250 0.375 0.375

Single strand chain 0.133 0.435 0.435

Double strand chain 0.139 0.342 0.518

fcc lattice 0.0865 0.0865 0.827

αpol =13

3

∑i=1

(εs − εm) [εs +(εc − εs)(Lci −ΩLs

i )]+Ωεs(εc − εs)

[εs +(εc − εs)(Lci −ΩLs

i )] [εm −Lsi (εm − εs)]+ΩLs

i εs(εc − εs)(4.36)

where εc, εs and εm are the dielectric functions of the core (i.e. inner spheroid)with depolarization factors Lc

i , the shell (i.e. coating around the spheroid) withdepolarization factors Ls

i and the surrounding medium. Ω is the volume ratiobetween the inner and outer spheroids. By putting Ω = 1 and Lc

i = Lsi = Li in

Eq. (4.36), εs would cancel out between the numerator and the denominatorand we would have the expression for randomly-oriented solid spheroids Eq.(4.31) with εc equivalent to εp in Eq. (4.31).

In this thesis, we used spherical core shell particles with Lci = Ls

i = 1/3 asshown in Fig. 4.2. Eq. (4.36) can then be simplified to:

αpol = 3(εs − εm)(εc +2εs)+Ω(2εs + εm)(εc − εs)

(εs+2εm)(εc +2εs)+Ω(2εs −2εm)(εc − εs)(4.37)

with

Ω =

(

xx+2t

)3

(4.38)

where x is the diameter of the core and t is the thickness of the shell.Requiring the volume fraction of the material composing the shell to be

0.01, as discussed in Chapter 7.2, the filling fraction of the particles f is

f = 0.01/(1−Ω) (4.39)

49

Figure 4.2: Structural model representing core-shell nanoshells in a surroundingmedium under irradiation of photons with energy hω .

4.2.2 The Bruggeman TheoryFor structures with non-isolated particles and topologically equivalent compo-nents, the Bruggeman theory [100, 101] is the appropriate approach used forobtaining the effective dielectric function. The Bruggeman theory may holdwell when the filling factor is varied from 0 to 1. It can also account for theoptical effects when f is varied across the percolation threshold [11]. Due tothese merits, the Bruggeman theory is a very widely used effective mediumtheory for optical properties.

According to the Bruggeman theory, for a two-component composite layerwith spherical structural units the effective dielectric function εBR can be writ-ten

fεA − εBR

εA +2εBR +(1− f )εB − εBR

εB+2εBR = 0 (4.40)

where εA and εB are the dielectric functions of the two components A and B,and f is the filling factor of A.

This equation is the one used to calculate for results in Chapter 7.3. The so-lution of Eq. (4.40) was obtained by using the approach developed by Berthierand Lafait [106]. An extension to n-component composites can be found inworks by Jansson and Arwin [107].

50

The Bruggeman theory can be extended to randomly oriented spheriodswith dielectric function εp and triplet of depolarization factors Lis densely dis-tributed in a medium composed of spherical particles with dielectric functionεm. In this case, εBR is given by:

εBR = εm

1+23

f αBRpol

1− 13

f αBRpol

(4.41)

and parameter αBRpol is expressed as:

αBRpol =

13

3

∑i=1

εp − εBR

εBR +Li(εp − εBR)(4.42)

where Lis can be obtained from Eq.s (4.32), (4.33) for prolate spheroids andEq.s (4.34), (4.35) for oblate spheroids.

4.2.3 Tuning the Optical Property of Metallic NanoparticlesThe well-known fact that nanoparticle composites can have different proper-ties from those of the bulk lies at the heart of recent technological advancesin nanophotonics. The optical response of metallic nanoparticles is influencedby the dielectric function of the particles and the surrounding medium as wellas by the size and shape. Here is a brief description of how it works.

Under irradiation of an alternating electro-magnetic field, the free electronsin a metallic nanoparticle oscillate following the external field and give rise topolarization oscillating along the field characterized by the dipole moment p.The dipole moment p is a measure of the optical response of a single nanopar-ticle; when p has a peak, the nanoparticle has a plasmon resonance. In thequasi-static picture inherent in the effective medium theories, described earlierin this chapter, the nanoparticles have a dipole moment p that is proportionalto the polarizability αpol . So for a single nanoparticle or dilutely distributednanoparticles ( f → 0) the resonance condition is αpol(ω) = αpol,max. In otherwords, when the denominator in the expression of αpol becomes a minimum,a plasmon resonance will occur which results in a strong absorption of light.

For solid spheroids, one can see that in Eq. (4.30) and (4.31) αpol is relatedto εm, εp and Li. Thus, by changing the materials used for the embedding ma-trix and the particles as well as the shape of the particles, the spectral tuningof plasmon absorption can be achieved. The resonance condition for spheresis εp = −2εm at dilute limit ( f → 0). Aligned spheroids have one resonancefrequency, whereas randomly oriented spheroids have two resonance frequen-cies.

51

The same occurs for core-shell particles in Eq. (4.36). By altering the mate-rials for the core εc, shell εs and the medium εm, the shape of the particles viaLi as well as the volume ratio between the core and the outer sphere Ω, one canspectrally tune the plasmon absorption. Moreover, the additional parameter Ωoffers a much broader range of spectral tuning compared to solid spheres.An example of powerful spectral tuning offered by core-shell nanoparticlesis shown in Figure 3.11 of Ref [11], in which silica-Ag-core-shell nanopar-ticles with different silver shell thickness enabled spectral tuning of plasmonabsorption across the solar spectrum.

A more detailed treatise on plasmonics can be found in Ref. [95].

The localized dipole surface plasmon resonance of a sin-gle metallic particle or metallic nanoparticles at dilute limit( f → 0) happens when the polarizability defined by Eq. (4.30)reaches maximum, which requires approximately:

εp1(ω) =−2εm (4.43)

for spheres with L1 = L2 = L3 = 1/3, provided εp2 ≪ 1. This isknown as the Frölich condition. [52]. Some Drude parametersof VO2 are available from literature [108] roughly ε∞ ∼4.5,hωp ∼ 3.5 eV, and h/τsc ∼0.7 eV. Applying εm =2.25, theassumption of the damped Drude model in Eq. (4.16) com-bined with Eq. (4.43) gives a rough estimation of the localizedsurface plasma resonance frequency hωres =0.93 eV (around1333 nm), where significant light absorption occurs. This phe-nomenon is the basis for the idea of Nanothermochromics.

Surface plasmon resonance of metallic state VO2 nanoparticles

♣

52

This box gives a rough evaluation of plasma frequency andelectron density of experimentally prepared ITO nanoparticlesfrom optical absorbance measurements -an important but un-published procedure in work featured in Paper VIII [109].According to the Frölich condition Eq.(4.43), and Drude di-electric function expressed with screened plasma frequencyEq.(4.21), the resonance frequency of dilutely distributed ITOparticles ωres should roughly satisfy the condition

ε∞(1−ω∗2

p

ω2res

) =−2εm (4.44)

thus

ω∗2p = ω2

res

√

1+2εm

ε∞(4.45)

One can make a very rough estimation of ωp of ITO fromthe optical absorbance measurement performed on a systemconsisting of ITO nanoparticles dilutely embedded in a glassor polymer-like dielectric medium, as studied in Paper VIII[109]. The experimental optical absorbance data in Fig. 1 ofthe paper suggest that the localized plasmon absorption of thenanoparticle composite is located at λ ∼ 1850 nm, equivalentto hωres ∼ 0.67 eV. From literature [16], it can be found thatITO has ε∞ ∼ 4, m∗

e ∼0.35, the embedding medium is takento have εm ∼ 2.25, then from Eq. (4.45), the screened plasmafrequency of ITO is roughly estimated to be hω∗

p ∼ 0.99 eV,and from Eq. (4.20), the plasma frequency (non-screened) ishωp ∼ 1.98 eV. From Eq. (4.13), the electron density can beestimated as ne ∼9.2 × 1020 cm−3. After further fine tuningcompared with experimental data, the electron density was de-termined to be ne ∼9.8 × 1020 cm−3.

Guessing the electron density of ITO nanoparticles

♠

4.3 Thin Film Optics4.3.1 One Interface: Fresnel EquationsAs illustrated in Fig. 4.3, consider an interface between medium 1 and 2 withcorresponding refractive indices N1 and N2, respectively; the angles of inci-dence and transmittance are θ1 and θ2, respectively. The boundary condition

53

Figure 4.3: The transmission and reflection of light of a single interface. Subscripts i,r and t denote the incident, reflected, and transmitted rays. θ1 and θ2 denote the angleof incidence and transmittance, respectively

at the interfaces states that the normal component of the B-field and the tan-gental component of the E-field are continuous at the interface. By applyingthis boundary condition to Maxwell equations, we get Fresnel equations:

tp =2N1 cosθ1

N1 cosθ2 +N2 cosθ1

ts =2N1 cosθ1

N1 cosθ1 +N2 cosθ2

rp =N2 cosθ1 −N1 cosθ2

N1 cosθ2 +N2 cosθ1

rs =N1 cosθ1 −N2 cosθ2

N1 cosθ1 +N2 cosθ2

(4.46)

where t and r denote the interface amplitude transmittance and reflectance.Subscripts s and p denote the polarization of light. The angles θ1 and θ2 arerelated by Snell’s law:

N1 sinθ1 = N2 sinθ2 (4.47)

The Fresnel equations in Eq. (4.46) can be written alternatively without N1and N2 as

54

tp =2sinθ2 cosθ1

sin(θ1 +θ2)cos(θ1 −θ2)

ts =2sinθ2 cosθ1

sin(θ1 +θ2)

rp =tan(θ1 −θ2)

tan(θ1 +θ2)

rs =−sin(θ1 −θ2)

sin(θ1 +θ2)

(4.48)

The intensity transmittance and reflectance are given by

Ts,p =N2 cosθ2

N1 cosθ1ts,pt∗s,p

Rs,p = rs,pr∗s,p

(4.49)

4.3.2 Two InterfacesIn this section, both s and p-polarization give the same expressions so the sub-scripts "s" and "p" are omitted. Light interacting with a coherent layer ex-hibits interference. The total amplitude transmittance and reflectance shouldbe obtained by summing up amplitude contributions from multiply reflectedlight taking into account the phase. Consider a single coherent layer withthickness d, composed of medium 2 characterized by refractive index N2 be-tween two media 1 and 3 characterized by refractive index N1 and N3 as shownin Fig. 4.4. Amplitude reflectance is denoted by r1 for interface 1/2, r′1 forinterface 2/1, r2 for interface 2/3; amplitude transmittance is denoted by t1for interface 1/2, t′1 for interface 2/1, t2 for interface 2/3. In addition a phase

change δ =−2πλ

N2d cosθ2 occurs during the passage of light in the film.

Summing up the contribution from each transmittance and reflectance andtaking into the account the phase change δ , we have for total amplitude re-flectance

rcoh = r1 + t1t′1r2 exp(−2iδ )+ t1t′1r′1r22 exp(−4iδ )+ ...

= r1 +t1t′1r2 exp(−2iδ )

1− r′1r2 exp(−2iδ )(4.50)

55

Figure 4.4: The transmission and reflection of light on a single coherent layer. Phasechange is not noted in this figure.

with r′1 =−r1, and t1t′1 = 1− r21, the final form of Eq. (4.50) is written:

rcoh =r1 + r2 exp(−2iδ )

1+ r1r2 exp(−2iδ )(4.51)

Similarly, the total amplitude transmittance is

tcoh = t1t2 exp(−iδ )+ t1t2r′1r2 exp(−3iδ )+ t1t2r′21 r22 exp(−5iδ )...

=t1t2 exp(−iδ )

1+ r1r2 exp(−2iδ )(4.52)

The intensity transmittance and reflectance in this coherent two-interface caseare

T =N3 cosθ3

N1 cosθ1tcoht∗coh

R = rcohr∗coh

(4.53)

In the incoherent layer, the phase coherence is lost when light is pass-ing through the material. The total intensity transmittance and reflectance areobtained by adding up intensity contributions of multiply reflected light with-out considering their phase. With the same configurations as in Fig. 4.4, con-sider an incoherent layer with thickness d, composed of a material denoted

56

2 characterized by N2 placed in between medium 1 and 3. The bulk inten-sity reflectance is denoted R1 for interface 1/2, R′

1 for interface 2/1, R2 forinterface 2/3; the bulk intensity transmittance is denoted T1 for interface 1/2,T ′

1 for interface 2/1, T2 for interface 2/3. We have T1 = T ′1 . Replacing ampli-

tude transmittance and reflectance with bulk intensity transmittance and re-flectance, phase term −iδ with absorption −αd in Eq. (4.50) and Eq. (4.52),we can obtain the total intensity reflectance and transmittance R and T:

R = R1 +T 2

1 R2 exp(−2αd)1−R′

1R2 exp(−2αd)

T =T1T2 exp(−αd)

1−R′1R2 exp(−2αd)

(4.54)

4.3.3 More Interfaces: Multilayer FilmsFor multilayer films, the calculation requires matrix multiplications based onthe transfer matrix method that can be found in standard text books on optics[110–112]. The procedure that we use in this study is a transfer matrix methodbased on work by Pfrommer et al. [113]. Two kinds of matrices correspondingto possible transformation of the radiation field are considered:

Interface matrix denoted by M, corresponding to transmission of the radia-tion field or intensity across an interface. Here the boundary conditionsat the interface are considered.

Layer matrix denoted by S, corresponding to the transmission of the radia-tion field through a layer. Here the phase change and the attenuation ofthe radiation field or the intensity is considered, depending on the layerthickness.

The total transmittance Toi and reflectance Roi from outer halfspace i toinner halfspace o can be solved from relation:

[

1

Roi

]

= Mo,1S1M1,2S2......Sq−1Mq−1,qSqMq,i

[

Toi

0

]

(4.55)

The detailed procedure is described in literature [113, 114]

57

5. Experimental Methods

An expert is a personwho has found out by his own painful experience

all the mistakes that one can makein a very narrow field.

Niels Bohr

5.1 Sample Fabrication: DC Magnetron SputteringIn the DC sputtering process, the targets are connected to the cathode and thesubstrate and the chamber walls are grounded. Argon (Ar) gas is introducedand ionized by the electric field into plasma containing energetic Ar ions thatbombard the target surface, which leads to the deposition of target atoms onthe substrate. Besides Ar, reactive gas species (such as oxygen and nitrogenetc.) can be introduced to promote chemical reactions to form compounds asthe end product of the deposition; this process is called DC reactive sputtering.

In the magnetron sputtering process, magnets are strategically installed be-hind the target to generate magnetic fields that can effectively confine beamelectrons into helix motions following an orbit trail near the target surface, andenhance the path length of the electrons so that the they can ionize more Arbefore escaping. Thus, the application of magnetron sputtering can help sus-tain the plasma with much lower pressure, and greatly improve the depositionrate.

The reactive DC magnetron sputtering process used in works presented inthis thesis took place in a vacuum system based on Balzers UTT 400 unit, asshown in Fig.5.1, and the inside geometry of the chamber is shown in Fig. 5.2.For thin film fabrications, the setup of a downward-directing tube oxygen inletas shown in Fig. 5.2 (a) was used. Other setup of oxygen inlets are available, asshown in Fig. 5.2 (b) and (c), for the fabrication of samples with more porousmorphologies, as is be discussed in Section 5.1.2. The substrates (1-mm-thickslide glass from the maker "Thermo Scientific") were put on a rotating holderlocated 13 cm from the 5-cm-diameter V (99.5%) at an oblique angle, with50 between the substrate surface normal and the average direction of thesputtered target atom flux. For doping purposes, Mg target (99.9%) can beinstalled at a slot equivalent to the V target, as shown in Fig. 5.2 (a).

59

5.1.1 The Sputtering of Pure and Mg-doped VO2 FilmsThe deposition chamber was pumped down to 6.3 × 10−7 mbar before heatingthe substrates. The substrates were subsequently heated to ∼450±10 C asestimated from thermocouple readings. 80 ml/min of Ar and ∼5 ml/min ofO2 (both 99.997%) were then introduced via mass-flow-controlled inlets, asin Fig. 5.2 (a). A total pressure of ∼9.2 mTorr was maintained during filmfabrication. Sputtering takes place at a power of 172 W and 0-57 W appliedon the V target and Mg target, respectively. Films were grown at 0.05-0.06nm/s to a thicknesses of 40-100 nm and on to glass and carbon substrates.Some films contained ∼ 0.45 at.% Si from system contamination, which waslater fixed with a new design of substrate holder.

5.1.2 The Sputter Deposition of VO2 NanorodsSputter deposition of VO2 nanorods was done to explores the influence ofthe temperature, thickness, substrate and geometry of reactive gas inlet on thegrowth morphology of VO2. Sputter deposition takes place from the V targetonly and follows the same procedure as stated in 5.1.1, with a variety of setupsof gas inlets, substrates and substrate temperatures. The O2 gas inlets were of 3different arrangements, as shown in Fig. 5.2: (a) a simple downward-directingtube (b) a perforated toroid gas inlet with holes obliquely facing the substrate(c) the same perforated toroid gas inlet flipped upside down compared to (b)with holes obliquely directing the gas away from the substrate. The substratesused in this study were of 3 kinds: (i) a 1-mm-thick glass slide from the maker"Thermo Scientific", which is the same as is used in VO2 film production asdescribed in Sect. 5.1.1; (ii) such glass coated with a 4.2-nm-thick gold filmmade by room temperature sputter depositon; (iii) glass coated with ∼ 50-nm-diameter silica nanopillars made by glancing angle depositions topped withgold caps produced in Kyoto, as reported in literature [115]. The substratetemperature during the deposition was set at 450±10 or 550±10 C.

Figure 5.1: The Magnetron sputter deposition chamber based on Balzers UTT 400unit.

60

Figure 5.2: Target-substrate geometery inside the sputter chamber and the differentsetups for the O2 gas inlet. (a) a simple downward-directing tube (b) a perforatedtoroid gas inlet directing gas toward the substrate (c) the toroid gas inlet directing thegas away from the substrate.

61

5.2 Rutherford Backscattering SpectrometeryRutherford backscattering spectrometry (RBS) is a well-established ion beamtechnique for analyzing elemental composition, and it is especially powerfulfor detecting heavier atoms. The collision events between incident ions, usu-ally proton or 4He+ with MeV energy, and the target sample involves twomain processes: the elastic scattering of incident ion by the target nuclei andthe inelastic scattering of the incident ion by the electrons in the target. Theformer yields backscattered ions with energies depending on the mass of thetarget atoms, which is manifested as an abrupt onset of an edge or peak in thespectrum. The latter process gives rise to the energy loss of the impinging ion,which is proportional to the "stopping power" and the number of atoms perunit area of the target, and is manifested as the detailed shape of the spectrumadjacent to the low energy side of the edge or peak. In addition, the intensityof the backscattered ions is proportional to the number of target atoms in thedetected area and their scattering cross section (the scattering cross sectionis proportional to the squared nuclear charge of the target atom). Thus, theinformation extracted from RBS reveals the stoichiometry and the number ofatoms per unit area of the sample, and further more, the asymmetry of thepeaks in the spectrum gives information about the "roughness" or "inhomo-geneity" which is detected as a fluctuation in the number of atoms per unit areain the detected area. In high-resolution RBS, the use of low-energy ions canalso give detailed depth profile down to atomic monolayers [116]. For moreinformation on RBS mechanisms and techniques one should consult standardtextbooks [117–120].

The samples studied in this work were measured in the Uppsala Univer-sity tandem laboratory using 2 MeV 4He+ ions as incidence and detected ata backscattered angle of 170. Information such as stoichiometry, number ofatoms per unit area and "roughness" were extracted by fitting the experimentalspectra with simulated spectra using the SIMNRA program [121]. Examplesof RBS results are shown in Fig. 5.3. The asymmetrical shape of the V peakin Fig. 5.3 (a) is due to flutuation of V atom area density within the mea-sured area as expected from nanorod samples. Fig. 5.3 (b) is an example fitfor determining the value of Mg/(Mg+V) in the Mg-doped thin film samples.

If the elemental number fraction ni of atom Ai with mass mi, total atoms perarea Narea is known and the sample’s geometrical thickness d is provided withanother method, the mass density of the sample can be calculated:

ρ = ∑mini ×Narea/d (5.1)

in which ni satisfies

∑ni = 1 (5.2)

62

0

500

1000

1500

2000

2500

50 100 150 200 250 300 350 400

400 600 800 1000 1200 1400 1600

Counts

Channel

Energy (keV)

Sample 7 Expcalc

V

Si

O

(a)

0

200

400

600

800

1000

1200

50 100 150 200 250 300 350

400 600 800 1000 1200 1400

Counts

Channel

Energy (keV)

C

O

V

Mg/(Mg+V)=0.055 Expcalc

0

20

40

210 240

1000

Mg

(b)

Figure 5.3: Experimental and simulated RBS data for a VO2nanorod sample on glass(sample No. 7 in paper IV)(a) and Mg-doped VO2 thin film with shown Mg content(b).

63

5.3 X-ray DiffractionX-ray diffraction [122–124] was used for studying the crystallographic struc-ture of the samples. X-rays used in this technique typically have a wavelengthof about 0.1 nm, comparable to the distance between crystal lattice planes.The diffraction arises from the interference effects which take place amongthe X-rays elastically scattered by the atoms in the crystal. The condition thatallows the observation of the diffracted beam is roughly divided in two: i) thecondition of constructive interference of scattered X-ray described by Bragg’slaw:

nxλ = 2dhkl sinθscatter (5.3)

where nx is an integer that represents the order of diffraction, and dhkl is theinter-planar distance of the planes with the Miller index (hkl). 2θscatter is theangle between the incident and diffracted beam.

ii) The condition that secures an observable intensity is governed by thescattering power of the atom species and their arrangement throughout thecrytal lattice, which are collectively included in the expression of the structurefactor Fhkl , and related to the intensity of the peak as:

Ihkl ∝ |Fhkl |2 (5.4)

Therefore, combining the above conditions i and ii, the Bragg reflection fora given wavelength of X-ray can be observed at 2θscatter off the direction ofthe incident beam, provided that the intensity is not too weak to observe. Byscanning the 2θscatter, one can obtain a diagram of peak intensity as a func-tion of 2θscatter that allows comparison with such crystallographic databasesas the Inorganic Crystal Structure Database (ICSD) [125] or the Crystallog-raphy Open Database (COD) [126], etc., for the determination of the crystalstructure of the sample.

The grazing incidence X-ray diffraction(GIXRD) data were recorded on thethin film and nanorod samples in the angle range of 10 < 2θscatter < 80 us-ing a Siemens D5000 Th-2Th instrument and Cu Kα as the incidence beam.Rietveld refinement [127] using software PowderCell 2.3 program [128] wasperformed on the nanorod samples for the determination of prefered orienta-tion using the March-Dollase model [129]. An example of XRD spectra andpeak assignment is illustrated in Fig.5.4.

64

10 20 30 40 50 60 70 80

Inte

nsity

(ar

b. u

nits

)

Diffraction angle 2θ (°)

(100

)

(−31

2)(−

122)

(−31

1)(1

21)

(011

)

(−10

2)

(−20

2)(2

11)(

200)

(−22

2)(2

20)(

211)

(022

)

(−40

2)

(020

)(00

2)(−

212)

(210

)

(031

)(01

3)

(−12

1)(1

20)

(−23

1)

(021

)(01

2)Sample 1

Sample 2

Sample 7

Sample 9

(202

)

(−30

2)

(−41

1)

(−22

1)

Figure 5.4: Glancing angle incidence X-ray diffraction taken at room temperature forVO2 thin film and nanorod samples discussed in Paper IV. Samples 1 and 2 refer tothe thin film samples with smaller and larger grains, respectively; samples 7 and 9refer to nanorod samples with smaller and larger mass thickness, respectively.

65

5.4 Scanning Electrom MicroscopyScanning electron microscopy (SEM) is a common method for observingsample morphology. When electrons impinge on the sample, the signal ofbackscattered electrons and the secondary electrons can be recorded. Repeat-ing this process while scanning a sample area, the topological informationin this area can be revealed by an intensity map of the recorded signal. Thesecondary electron signal is more informative about the topological contrast,whereas the backscatter electron signal emphasizes the compositional con-trast, so one can choose which signal to detect accordingly. Such is the prin-ciple of the most basic SEM measurements.

In the study on nanorods, SEM was done using a LEO 1550 FEG Geminiinstrument with an acceleration voltage of 5-15 kV and the secondary electronsignals were collected. Surface images, cross section images, and a "bird-eye-view" image recorded with the sample surface normal tilted 70 off the beamdirection were recorded and discussed in 9.1. Examples of SEM images areshown in Fig. 5.5 for Mg-doped VO2 films. These films have developed sur-face nanostructures.

(a) (b)

(c) (d)

Figure 5.5: SEM image of Mg-doped VO2 films with shown Mg/(Mg+V) atom ratio.(a) and (c) are the top views; (b) and (d) are the "bird-eye-view" imaged at 70 off thesurface normal of the samples.

66

5.5 Electrical Characterization: Sheet ResistanceResistance measurement is done using the 4-wire method. The schematicsetup is illustrated in Fig. 5.7. Samples were cut into bar shapes with fourcontacts painted on the thin film surface with silver glue. The contacts nearthe edges were connected to the power current supply with current readings I,and the inner contacts to the voltammeter with voltage readings V . The sam-pled area is the area between the two voltage contacts with resistance givenby Rres =V/I, from which the resistivity ρR can be obtained as:

ρR =RresdLwidth

Llength(5.5)

where Llength is the length dimension of the sampled area along the flow ofcurrent in the 4-wire method; Lwidth is the width of the sampled area and d isthe thickness of the thin film sample.

In reality, the power, current and the voltmeter are contained by a KEITH-LEY 2400 source meter in 4-wire sensing mode, so that the resistance of themeasured area can be directly read and the resistivity can then be calculatedusing Eq. (5.5). The sample part was fixed on a heater controlled by a temper-ature controller connected to a themometer. To correctly measure the sampletemperature, the thermometer was fixed on top of a piece of glass substrateidentical to that of the sample, located at an equivalent position as the 4-wiresetup. The measured temperature range was 25-127 C, with heating ramprate 3 C/min and cooling by turning off the heater.

The resultant resistivity curve as a function of temperature is shown in Fig.5.6 for VO2 thin films with different Mg-doping. The results shown in this fig-ure are not consistent with trends of lowered τcr proportional to the Mg con-tents as established in literature [36, 42]. The reasons could be i) the numberof samples measured was too small to establish a trend or i) the films have de-veloped nanostructures (Fig. 5.5 ) that influenced the metal-to-insulator tran-sition.

67

Figure 5.6: Resistivity as a function of temperature for shown VO2 samples withshown Mg/(Mg+V) atom ratios.

10−1

100

101

102

103

20 40 60 80 100 120 140

Res

istiv

ity (

Ω·

cm)

Temperature (°C)

Mg/(Mg+V)=z, z=00.0060.0240.0550.088

Figure 5.7: Schematic drawing the the 4-wire measurement setup.

68

5.6 Optical MeasurementsThe measurement of UV/VIS transmittance and reflectance can be done usingthe absolute instrument and a Perkin Elmer Lambda 900 instrument. Bothinstruments use a white light source which passes through a monochromator.

Sample

Detector position

for reflectance

Detector position

for transmittanceParabolic mirror

Light source

Figure 5.8: Schematic illustration of the absolute instrument.

The absolute instrument [130] was used for measuring the specular com-ponent of transmittance and reflectance. The illustration is shown in Fig. 5.8.The advantage is that the reference data is taken on the primary beam itselfand therefore free of the need for a reference plate. The drawback is that theintensity of the light source changes with time and frequent referencing is nec-essary. In this instrument, one first takes a reference by putting the detector atthe "transmittance position" and keep sample out of the light path. Then trans-mittance and reflectance can be measured by putting the sample in the shownposition in Fig. 5.8 and fixing the detectors at the "transmittance position" and"reflectance position" respectively. Angle-resolved reflectance measurementscan be done with this instrument as well.

1. Transmittance

port

3. Exit port for

specular

reflectance

2. Reflectance

port

4. Reference

entry port

5. Reference

exit port

Figure 5.9: Schematic illustration of the integration sphere.

The Perkin Elmer Lambda 900 instrument is equipped with an integrationsphere, as shown in Fig. 5.9, and it is capable of measuring total, specular,diffuse transmittance and reflectance. Ports 1, 2, 3, 4, 5 stand for transmittanceport, reflectance port, exit port for specular reflectance, reference entry portand reference exit port, respectively. Referencing is done by recording thesignal with incident light directly projected on the BaSO4 reference sample

69

fixed on port 2. All the measurements done in this study have port 4 open andport 5 closed so changes the instrument only involve ports 1, 2, 3.1) Transmittance setting:

Measurement of signal S1,tot : BaSO4 reference sample is placed on port 2,and the sample is placed on port 1. Port 3 is closed.

Measurement of diffuse signal S1,di f f : a black cone is placed on port 2, andthe sample is placed on port 1. Port 3 is closed.

Then the specular transmittance Tspec, diffuse transmittance Tdi f f and totaltransmittance Ttot can be obtained via:

Tspec = S1,tot −S1,di f f

Tdi f f = S1,di f f ×RBaSO4

Ttot = Tspec +Tdi f f

(5.6)

Where RBaSO4 is the reflectance of the BaSO4 reference recorded by the abso-lute instrument.2) Reflectance setting:

Measurement of signal S2,tot : sample is placed on port 2 with port 1 openand port 3 closed.

Measurement of diffuse signal S2,di f f : sample is placed on port 2 with port1 open and port 3 open.

Here the specular transmittance Rspec, diffuse reflectance Rdi f f and totalreflectance Ttot can be obtained via:

Rspec = (S2,tot −S2,di f f )×0.96

Rdi f f = S2,di f f ×RBaSO4

Rtot = Rspec +Rdi f f

(5.7)

Where 0.96 is a factor to correct for the geometry of the integration sphere.Optical properties of the thin film samples in this study were measured with

the absolute instrument and Perkin Elmer Lambda 900 instruments. Some ofthe samples were rough and exhibited a small but non-negligible value ofRdi f f ∼ 5± 2% in the short wavelength end of the spectra. To suppress theinfluence of the surface roughness, reflectance was measured both from thesubstrate side and sample side to be used later in a fitting process for thedetermination of optical constants.

All the the nanorod samples were measured with the Perkin Elmer Lambda900 instrument.

70

5.7 The Determination of Optical Constants5.7.1 The Glass SubstrateIn this study, the refractive index of the glass substrate was determined fromnear-normal incidence reflectance and transmittance of the 1-mm-thickglass substrate ("Thermo Scientific", identical to those used in experiments)recorded using the absolute instrument. Recall Eq.(4.54) in 4.3.2,when entrance and exit media have the same refractive index (both areair), N1 = N3 = 1, we have R1 = R′

1 = R2 ≡ R, T1 = T2 ≡ 1 − R withR = |(1−n− ik)/(1+n+ ik)|2, and Eq.(4.54) can be rewritten: [131]

R = R+(1− R)2Rexp(−2αd)

1− R2 exp(−2αd)

T =(1− R)2 exp(−αd)1− R2 exp(−2αd)

(5.8)

Optical constants n and k can then be obtained in the following steps: [132]

R =(T 2 +2)− (R−1)2

2(2−R)−

√

[

(T 2 +2)− (R−1)2

2(2−R)

]2

− R2−R

k =− λ4πd

ln(

R− RRT

)

n =

(

1+ R1− R

)

1+

√

1−(

1− R1+ R

)2

(1+ k2)

(5.9)

The determined refractive index of glass is shown in Figure 5.10. Thismethod was also documented in William E. Vargas’ PhD thesis [133].

5.7.2 Thin FilmsThe determination of optical constants of the thin film samples was done by si-multaneously fitting the simulated transmittance and reflectance to the exper-imental ones. The procedure was achieved by using SCOUT software [134]based on a downhill simplex algorithm. SCOUT allows the building of multi-layer stack thin film models, typically, vacuum/thin film/substrate/vacuum fortransmittance and reflectance collected from the sample surface side. For the

71

1.0

1.2

1.4

1.6

1.8

2.0

500 1000 1500 2000 2500

Opt

ical

con

stan

ts

Wavelength (nm)

(a) glass n

0×100

2×10−6

4×10−6

6×10−6

8×10−6

10×10−6

12×10−6

500 1000 1500 2000 2500

Wavelength (nm)

(b) glass k

Figure 5.10: The optical constants n (a) and k (b) of the 1-mm-thick glass substratedetermined using methods described in 5.7.1.

0.0

0.2

0.4

0.6

0.8

1.0

500 1000 1500 2000 2500

R a

nd T

sem

icon

d.

Wavelength (nm)

(a) Reflectance from surfaceMg/(Mg+V)=0

0.0060.0240.0550.088

expexpexpexpexp

calccalccalccalccalc

500 1000 1500 2000 2500

Wavelength (nm)

(b) Reflectance from back

500 1000 1500 2000 2500

Wavelength (nm)

(c) Transmittance

0.0

0.2

0.4

0.6

0.8

1.0

500 1000 1500 2000 2500

R a

nd T

met

allic

Wavelength (nm)

(d) Reflectance from surface

500 1000 1500 2000 2500

Wavelength (nm)

(e) Reflectance from back

500 1000 1500 2000 2500

Wavelength (nm)

(f) Transmittance

Figure 5.11: Experimentally measured and simulated spectral transmittance T andreflectance recorded from the surface-coated side Rs and substrate side Rb of the VO2thin films with different Mg content as indicated in the figures. The upper panel isfor the semiconducting state (τ < τcr) and the lower panel is for the metallic state(τ > τcr).

72

0.0

1.0

2.0

3.0

4.0

500 1000 1500 2000 2500

Opt

ical

con

stan

ts n

and

k

Wavelength (nm)

(a) semicond.Mg/(Mg+V)=00.0060.0240.0550.088

nnnnn

kkkkk

500 1000 1500 2000 2500

Wavelength (nm)

(b) metallic

Figure 5.12: The optical constants n and k determined with "surface+back" methodfor VO2 films with different Mg content for the semiconducting state (τ < τcr) (a) andmetallic state (τ > τcr) (b).

incoherent substrate layer, dielectric functions of the glass substrate shown inFig. 5.10 were used. For the coherent thin film layer, models containing Bren-del oscillators [134] (usually two in the 1-2 eV range, one in 3-4 eV rangeand one in further UV) were constructed for semiducting state; and modelscontaining Brendel oscillators (usually one in 3-4 eV range, one in the near-infrared) combined with one drude term (for samples with high conductiv-ity) were constructed for metallic state. The fitting parameters for the modelsare resonance frequencies, oscillator strength, damping and distribution width.The film thickness can also be determined by fitting.

As the films have surface roughness and small amounts of scattering inthe short wavelength range, spectral transmittance T and reflectance recordedfrom the sample side Rs and the substrate side Rb were used in the fitting. Twomultilayer stacks were constructed with stack 1 configured as vacuum/thinfilm/substrate/vacuum, and stack 2 as vacuum/substrate/thin film/vacuum. Theaim was to find one set of optical constants and thickness for the film to makea good fit of simulated spectra of stack 1 to Rs and T as well as stack 2 to Rband T simultaneously. Another condition to ensure was that the film thicknessdetermined on the semiconducting state must be identical to that of the metal-lic state. To fulfil this, the thickness was determined for both states separatelyfirst, and then the average of the two was fed back to the fitting procedure as afixed value, followed by a refinement of the oscillator parameters. In this way,the thickness that fulfils agreement between measurements from the samplefront side and substrate side, as well as between the semiconducting state andthe metallic state, was determined with corresponding dielectric functions. Anexample of the fitting results yielded with this method is shown in Fig. 5.11and the resultant n and k are shown in Fig. 5.12. This method is referred to asthe "surface+back" method in the results part and in paper V. Some sampleshave developed nanostructures and could not yield a good fit with this method.

73

In the past, an old sets of n and k were determined by configuring one layerstack only: using the combination of Rs and T with layer stack vacuum/thinfilm/substrate/vacuum later referred to as the "surface" method, or with Rband T using the layer stack vacuum/thin film/substrate/vacuum later referredto as the "back" method. In the case of "suface" or the "back" method, due tothe fact that the only one layer stack was fitted, the constraint of matching thethickness of two layer stack configurations is relieved and therefore allowedthose samples which did not yield a good fit with "surface+back" method toyield a good fit. As a result, the "back" method delivered good fits for 29samples, whereas the "surface+back" methods only delivered good fits for 24samples.

5.8 Determination of Band GapsThe optical absorption coeffient is related to the imaginary part of the com-plex refractive index k, as α = 4πk/λ , and the optical band gaps Eg can bedetermined from:

(αhν)m = A(hν −Eg) (5.10)

where h is Planck’s constant, hν is the photon energy and A is a constant.The exponent m depends on the nature of the optical transition and is taken tobe 1/2. 1/3. 2. 2/3 for indirect-allowed, indirect-forbidden, direct-allowed, anddirect-forbidden optical transitions [135]. A plot of (αhν)m as a function ofhν reveals band gap value as the intercept between axis αhν = 0 and the linearfit near the onset of the absorption. In all the analyses in papers V and XIV,the indirect-allowed transition was assumed and therefore m = 1/2 was takenin the analysis. Two band gaps can be extracted, ∼0.5 eV for the transitionfrom lower d|| to π∗ band and ∼ 1.6 eV for that from O 2p to π∗ band [22,44]. The larger one Eg ∼ 1.6 eV is relevant to the luminous properties ofthe films and was extracted from a large number of samples and discussed atlength in paper V and 8.1. Fig. 5.13 shows example of band gap extractiondescribed above for 2 Mg-doped samples with the assumption of differentnature of transitions. The assumption of indirect allowed transition yielded1.71 eV and 2.04 eV for Mg/(Mg+V)=0 and 0.088, respectively, whereas thecorresponding figures for assuming direct allowed transition were 2.72 eVand 3 eV for Mg/(Mg+V)=0 and 0.088, respectively. It can be seen that theassumption of direct allowed transition yields ∼ 1 eV larger value of Eg thanthe assumption of indirect allowed transition. Nevertheless, both Eg valuesagree well with corresponding literature: refs [30, 31, 136] for Eg ∼ 1.6−1.7eV assuming indirect allowed transition, and refs [137, 138] for Eg ∼ 2.5−2.7eV assuming direct allowed transition.

74

In this work, we used the former assumption of indirect-allowed transitionbecause it falls in line with the theoretical calculated band structure [44].

0

200

400

600

800

1000

1200

1400

0.5 1 1.5 2 2.5 3 3.5 4

(αhν)

1/2 (

eV×c

m−

1 )1/2

Photon Energy hν (eV)

Mg/(Mg+V)=0, Eg = 1.71 eVMg/(Mg+V)=0.088, Eg = 2.04 eV

(a)

0×100

1×1012

2×1012

3×1012

0.5 1 1.5 2 2.5 3 3.5 4

(αhν)

2 (eV

×cm

−1 )2

Photon Energy hν (eV)

Mg/(Mg+V)=0, Eg = 2.72 eV

Mg/(Mg+V)=0.088, Eg = 3.0 eV

(b)

Figure 5.13: (αhν)m as a function of hν for VO2 thin films with shown amount ofMg, with band gap energy Eg extracted by linear fit. (a) m = 1/2 assuming indirectallowed transition; (b) m = 2 assuming direct allowed transition.

Apart from the above described method, an alternative method is availablewhich allows the determination of the band gap without considering the natureof the transition [139]. It is not in the scope of this thesis, so I do not discussit here.

75

6. Input data for Computation

6.1 VO2 Spheroidal Nanoparticle InclusionsIn Sections 7.1 and 8.3 (reported in Papers I and VI), I consider a dilute sus-pension of solid VO2 nanoparticles with spherical or ellipsoidal morphologyembedded in a dielectric matrix and backed by a dielectric substrate, as inthe structure shown in Fig. 4.1. The effective dielectric function εMG of thisstructure is determined by using the Maxwell-Garnett theory as described inChapter 4.2.1.1.

I consider the dielectric matrix to have optical properties similar to that ofglass or polymer and the dielectric substrate to be glass-like. So the dielectricfunction for the dielectric matrix εm and that of the substrate εsub are set to beεm = 2.25 and εsub = 2.25.

The optical constants of VO2 have been determined several times [29, 31,47, 48, 137, 138, 140, 142, 143] as well as in section 5.7.2. A selection ofthese literature values [47, 48, 137, 140, 142] are plotted together with thosedetermined in in section 5.7.2 in Fig. 6.1. As one can see, they agree witheach other fairly well apart from "Tazawa et al. 1998" [140], for n in thesemiconducting state.

For the calculations for results in 7.1, I used data determined by Mlyuka etal. in our group [48], based on a 50 nm thick film and shown in Fig. 5.7.2,whereas results in 8.3 were derived from n and k determined in section 5.7.2 (in Fig. 6.1 marked as "Mlyuka 2009" for Paper I and "Present work" for PaperIV and VI). VO2 in the metallic state is shown to have an exceptionally smallscattering time with h/τ f ∼ 0.7eV [108, 144] and a very small Fermi velocityvF ∼ 2.4× 105 m/s [145]. This results in a very small mean free path, aboutthe order of the lattice constant. VO2 may not even be a conventional Fermiliquid [145]. An alternative work [146] suggests that the mean free path maybe one order of magnitude larger, but not large enough to invoke particle sizedependence. Following the discussion about particle size in Chapter 4.1.5, it

is clear that the term2vF f τ f

Dpis negligible, and hence it is appropriate to put

εp = ε f .In addition, as input parameter for the Maxwell-Garnett theory, the filling

factor of the VO2 particles was fixed at 0.01. The thickness of the layer d wasset to be 5 µm (Section 7.1, 8.3) or to vary 0 < d < 20 µm (Section 8.3.2).which are equivalent to a 50 nm thick VO2 film or films with 0 < d < 200 nm,respectively, in terms of the amount of VO2. The cases for spheres, prolate

77

(a)

(c)

(b)

(d)

0

1

2

3

4

5

6

500 1000 1500 2000 2500

sem

icond. n

wavelength (nm)

Present workKana Kana et al. 2011

Mlyuka et al. 2009Kakiuchida et al. 2008

Tazawa et al. 1998Verleur et al. 1968

0

1

2

3

4

5

6

500 1000 1500 2000 2500

meta

llic n

wavelength (nm)

Present workKana Kana et al. 2011

Mlyuka et al. 2009Kakiuchida et al. 2008

Tazawa et al. 1998Verleur et al. 1968

0

1

2

3

4

5

6

500 1000 1500 2000 2500

sem

icond. k

wavelength (nm)

Present workKana Kana et al. 2011

Mlyuka et al. 2009Kakiuchida et al. 2008

Tazawa et al. 1998Verleur et al. 1968

0

1

2

3

4

5

6

500 1000 1500 2000 2500

meta

llic k

wavelength (nm)

Present workKana Kana et al. 2011

Mlyuka et al. 2009Kakiuchida et al. 2008

Tazawa et al. 1998Verleur et al. 1968

Figure 6.1: Optical contants n and k for an undoped VO2 film determined in this work(marked "Present work") plotted with literature data obtained by Verleur et al. [137],Tazawa et al. [140], Kakiuchida et al. [47], Mlyuka et al. [48], and Kana Kana et al.[142].

spheroids and oblate spheroids are characterized using the aspect ratio m asalso defined in Fig. 4.1.

Optical transmittance and reflectance for normal incidence onto such acomposite layer backed by a substrate was calculated using the matrix for-malism in standard thin film calculations based on Fresnel’s equations. Herewe used the method that is introduced in Ref. [113] as described in Chapter4.3.3. The optical transmittance and reflectance of VO2 thin film with differentthickness, varied from 0.01 µm to 0.1 µm was also calculated for comparison.The wavelength-integrated optical properties, luminous transmittance and so-lar transmittance were calculated using Eq. (2.7).

6.2 VO2 Spherical Core-shell Nanoparticle InclusionsIn Chapter 7.2 and Paper II, calculations were done on spherical core-shellparticles. The model considers a dilute suspension of VO2 nanoparticle shellsenclosing dielectric cores embedded in a glass or polymer-like dielectricmedium and backed by a dielectric substrate. The structural model is shownin Fig. 4.2. The Maxwell-Garnett theory for core-shell nanoparticles,

78

described in Chapter 4.2.1.3, was used to determine the effective dielectricfunction εMG.

For the dielectric medium and the dielectric substrate I used the same inputdielectric function value, i.e., εm = εsub = 2.25. The refractive index of thedielectric core nc was varied to represent different materials: 1 vacuum orair, 1.5 glass or polymer, 2 and 2.5 some common transparent high refractiveindex materials such as ZnO and TiO2.

For the VO2 particle shells, the same arguments as in Chapter 6.1 apply, sothe dielectric function determined from a VO2 film marked as "Mlyuka et al.2009" shown in Fig. 6.1 was used as input dielectric function. For the samereason, the absolute value of x and t were not treated in the calculation in Eq.(4.36).

As input parameters for the Maxwell-Garnett theory, the composite layerwas set to be 5 µm thick and the relative volume amount of VO2 in the layeris kept 0.01 constant. The ratio between the diameter (x) of the spherical coreand the thickness (t) of the shell x/t was varied from 0 to 20, and accordinglythe filling fraction of the particles f goes from 0.01 to 0.04 as explained in Eq.(4.39).

Optical transmittance and reflectance for normal incidence onto the abovementioned composite layer backed with a substrate were calculated using thesame method as described in Chapter 6.1. The wavelength-integrated opticalproperties, luminous transmittance and solar transmittance were calculatedusing Eq. 2.7. The solar transmittance modulation ∆Tsol was calculated usingEq. 3.1.

6.3 Composites of VO2 and ITOIn Chapter 7.3 and Paper III, calculations were done on a composite layer ofVO2 and ITO. The model considers a random mixture of nanopaticles of VO2and ITO. The VO2 and ITO units were taken to be spherical and topologicallyequivalent. The Bruggeman theory, described in Chapter 4.2.2, was used todetermine the effective dielectric function εBR of such a composite layer.

For the input dielectric function of VO2, the same argument as in Chapter6.1 stands; thus I used the dielectric function marked as "Mlyuka et al. 2009"shown in Fig. 6.1.

The dielectric function of ITO was determined for a 200 nm thick ITOfilm by Hamberg et al. [16], in which they derived the optical constants forITO with different values of free electron carrier concentration ne. We usedthe results from ITO with ne = 6.2× 1020 cm−3, as shown in Fig 6.2. Thefree electron mean free path of a high-quality ITO is very short, ∼ 5 nm, andis governed by ionized impurity scattering in the infrared. Considering theeffect of particle size, discussed in Chapter 4.1.5, the situation is analogousto the one in VO2 and it is clear that this effect can be ignored. Hence the

79

0

0.5

1

1.5

2

2.5

500 1000 1500 2000 2500

Opt

ical

con

stan

ts, n

and

k

wavelength (nm)

ITO nk

Figure 6.2: Input spectral optical constants n and k for a 200 nm thick ITO film derivedfrom ε1 and ε2 determined by Hamberg et al. [16] using Eq. (4.5).

optical constants in Fig. 6.2 were used as the input dielectric function for ITOparticles.

In these calculations, the ITO mass thickness was considered to be con-stant at 200 nm, while the VO2 mass thickness varied so that the overall VO2volume fraction in the composite layer f ranged from 0.1 to 0.5.

Optical transmittance and reflectance for normal incidence onto the abovementioned composite layer backed with a substrate were calculated using thesame method as described in Chapter 6.1. The wavelength-integrated opticalproperties, luminous transmittance, solar transmittance and the solar transmit-tance modulation ∆Tsol were obtained using the same method as described inChapter 6.2.

80

Part II:

Important Results

"In theory, there is no difference betw een theory and practice;

In practice, there is."

a Yogi Berra quote spotted in book"Antifragile: Things That Gain From Disorder"

by Nassim Taleb

7. Nanothermochromics withUndoped VO2: Modeling Results

7.1 Dielectric Matrix with VO2 SpheroidalNanoparticle InclusionsIn this section, calculated results from a layer with a dilute suspension of VO2solid spheroidal nanoparticles in a dielectric matrix are shown. A comparisonwith the results for VO2 films is made to show that VO2 nanoparticle lay-ers have very much improved luminous transmittance and solar transmittancemodulation compared with VO2 films.

7.1.1 VO2 Films

0

20

40

60

80

100

500 1000 1500 2000 2500

Tra

nsm

ittan

ce (

%)

Wavelength (nm)

0.01 µm

0.02 µm

0.03 µm

0.05 µm

0.1 µm

Semiconducting

0

20

40

60

80

100

500 1000 1500 2000 2500

Tra

nsm

ittan

ce (

%)

Wavelength (nm)

0.01 µm

0.02 µm

0.03 µm

0.05 µm

0.1 µm

Metallic

(a) (b)

Figure 7.1: Computed spectral transmittance for VO2 films in semiconducting (τ <τcr) (a) and metallic (τ > τcr) (b) states.

Computed spectral transmittance of VO2 films of different thicknesses areshown in Fig. 7.1. For both semiconducting and metallic states, T (λ ) de-creases as the film thickness is increased. The low transmittance dominatingthe UV and a part of the visible range corresponds to the inter-band transitionfrom the O 2p to the d band. For the semiconducting state (Fig. 7.1 (a)), T (λ )increases at long wavelengths; in contrast, the transmittance for the metallicstate peaks at ∼700 nm and goes down at long wavelengths, which marks the

83

characteristic near-infrared thermochromic switching for typical VO2 films.The most efficient switching is located at ∼2500 nm.

Fig. 7.2 shows computed integrated properties (Eq. (2.7)), specifically lu-minous transmittance Tlum, solar transmittance Tsol and solar transmittancemodulation ∆Tsol derived from T (λ ,τ) in Fig. 7.1. An increase of the filmthickness can increase the solar transmittance modulation ∆Tsol , as defined inEq. (3.1), but also causes Tlum to decrease to an undesirably low level. As aresult, requiring a noticable solar transmittance modulation ∆Tsol limits Tlumto less than ∼ 40%. For a 50 nm thick VO2 film, as one can pinpoint in Fig.7.2, we have Tlum(τ < τcr)≈ Tlum(τ > τcr) = 37.4% and ∆Tsol = 6.5%.

0

20

40

60

80

100

0.05 0.1 0.15 0.2 0.25 0.3

Tlu

m (

%)

Thickness (µm)

semiconducting metallic

0

20

40

60

80

100

0.05 0.1 0.15 0.2 0.25 0.3

Tso

l (%

)

Thickness (µm)

semiconducting metallic

(a) (b)

0

5

10

15

20

0.05 0.1 0.15 0.2 0.25 0.3

∆ T

sol (

%)

Thickness (µm)

(c)

Figure 7.2: Computed luminous (a) and solar (b) transmittance and (c) solar transmit-tance modulation corresponding to data in Fig. 7.1 as a function of thickness of VO2films.

7.1.2 VO2 Nanoparticles vs. FilmSpectral transmittance, reflectance and integrated optical properties for a layerwith a dilute suspension of VO2 solid spheroidal nanoparticles in a dielectricmatrix characterized with εm = 2.25 backed by a dielectric substrate with the

84

0

20

40

60

80

100

500 1000 1500 2000 2500

Tra

nsm

ittan

ce (

%)

Wavelength (nm)

VO2 semiconducting

oblate m=0.1m=0.2m=0.5

sphere m=1prolate m=2

m=5m=10

(a)

0

20

40

60

80

100

500 1000 1500 2000 2500

Tra

nsm

ittan

ce (

%)

Wavelength (nm)

VO2 metallic

oblate m=0.1m=0.2m=0.5

sphere m=1prolate m=2

m=5m=10

(b)

0

20

40

60

80

100

500 1000 1500 2000 2500

Ref

lect

ance

(%

)

Wavelength

VO2 semiconductingoblate m=0.1

m=0.2m=0.5

sphere m=1prolate m=2

m=5m=10

0

20

40

60

80

100

500 1000 1500 2000 2500

Ref

lect

ance

(%

)

Wavelength

VO2 metallicoblate m=0.1

m=0.2m=0.5

sphere m=1prolate m=2

m=5m=10

(c) (d)

Figure 7.3: Computed spectral transmittance and reflectance for spheroidal VO2 par-ticles (as in Fig 4.1(a), (b) and (c)), with the shown aspect ratio m and a filling factorof 0.01, dispersed in a dielectric medium. The spheroids have their symmetry axesperpendicular to the substrate. (a) and (c) refer to VO2 in semiconducting state (τ <τcr), (b) and (d) refer to VO2 in metallic state (τ > τcr).

85

30

40

50

60

70

80

90

100

0.1 0.2 0.5 1 2 5 10

Lum

inou

s tr

ansm

ittan

ce (

%)

Aspect Ratio m

VO2 semiconductingmetallic

30

40

50

60

70

80

90

100

0.1 0.2 0.5 1 2 5 10

Sol

ar tr

ansm

ittan

ce (

%)

Aspect Ratio m

VO2 semiconductingmetallic

(a) (b)

0

5

10

15

20

25

30

0.1 0.2 0.5 1 2 5 10

Sol

ar m

odul

atio

n (%

)

Aspect Ratio m

(c)

Figure 7.4: Integrated luminous (a), solar (b) transmittance and (c) solar modulationcorresponding to data in Fig. 7.3 for spheroidal VO2 particles (as in Fig 4.1(a), (b) and(c)), with the shown aspect ratio m and a filling factor of 0.01, dispersed in a dielectricmedium. The spheroids have their symmetry axes perpendicular to the substrate.

86

same dielectric constant were computed. The layer thickness was taken to be5 µm and the content of VO2 particles was 1 vol.%; this is equivalent to a50 nm thick VO2 film (i.e., both contain the same amount of VO2). Differentspheroidal shapes, i.e., prolate spheres and oblate spheres, characterized withan aspect ratio m as defined in Fig. 4.1, are taken into consideration.

The spectral transmittance and reflectance for spheroids oriented with sym-metry axes perpendicular to the substrate as depicted in Fig. 4.1 (a) (b) and(c) are shown in Fig. 7.3. The aspect ratio m is varied from 0.1 to 10. R(λ ) ofboth semiconducting states and metallic states are low and mostly flat. T (λ )increases monotonically as m increases. The low transmittance dominatingthe UV and a part of the visible range is the result of the absorption due tointer-band transitions for O 2p to d band. The most salient feature in Fig.7.3 is the pronounced absorption for the metallic state in the near-infrared,which is centered at ∼ 1.2 µm for spheres, due to a heavily damped plasmonresonance in the metallic state of VO2 [147]. This feature marks an efficientthermochromic switching, because the solar irradiance intensity is high in thiswavelength range. The location of the absorption band goes to shorter wave-length for prolate spheroids and a longer wavelength for oblate spheroids.Moreover, T (λ ) in the visible range is much higher than that of the VO2 film.

Fig.7.4 shows the integrated properties Tlum, Tsol and ∆Tsol as obtained fromthe spectra in Fig. 7.3. The particle shape dependence is consistent with spec-tra in Fig. 7.3. Tlum increases monotonically as m increases. Particles withlarger m offer much higher Tlum than that of a corresponding 50-nm-thick film( Tlum = 37.4%), while ∆Tsol decreases somewhat but is still 2 ∼ 3 times aslarge as that of a corresponding film (∆Tsol = 6.5%). For the intermediate caseof spheres (m = 1), we have Tlum of 72% and 62% for the semiconducting andmetallic states, respectively, and ∆Tsol is 16.7%.

Randomly oriented spheroids are of interest as well. Spectral transmittancefor VO2 particles of different shapes is shown in Fig.7.5, and the integratedproperties Tlum and ∆Tsol are shown in Fig. 7.6. The same as in the case fororiented spheroids, both Tlum and ∆Tsol are high, with a maximum of Tlumoccuring for spheres (m = 1) while Tsol is very much the same for all shapes(across all values of m).

The above computational results show that VO2 particle based materials canoffer greatly improved luminous transmittance and solar transmittance mod-ulation compared to films. This makes VO2-based thermochromic windowsmuch more suitable for practical use in energy efficient applications.

87

0

20

40

60

80

100

500 1000 1500 2000 2500

Tra

nsm

ittan

ce (

%)

Wavelength (nm)

VO2 semiconducting

oblate m=0.1m=0.2m=0.5

sphere m=1prolate m=2

m=5m=10

0

20

40

60

80

100

500 1000 1500 2000 2500

Tra

nsm

ittan

ce (

%)

Wavelength (nm)

VO2 metallicoblate m=0.1m=0.2m=0.5

sphere m=1prolate m=2

m=5m=10

(a) (b)

Figure 7.5: Spectral transmittance for spheroidal VO2 particles (as in Fig 4.1(d)), withthe shown aspect ratio m and a filling factor of 0.01, dispersed in a dielectric medium.The spheroids are randomly oriented. Parts (a) and (b) refer to VO2 in semiconducting(τ < τcr) and metallic states (τ > τcr), respectively.

30

40

50

60

70

80

90

100

0.1 0.2 0.5 1 2 5 10

Lum

inou

s tr

ansm

ittan

ce (

%)

Aspect Ratio m

VO2 semiconductingmetallic

30

40

50

60

70

80

90

100

0.1 0.2 0.5 1 2 5 10

Sol

ar tr

ansm

ittan

ce (

%)

Aspect Ratio m

VO2 semiconductingmetallic

(a) (b)

0

5

10

15

20

25

30

0.1 0.2 0.5 1 2 5 10

Sol

ar m

odul

atio

n (%

)

Aspect Ratio m

(c)

Figure 7.6: Integrated luminous (a), solar (b) transmittance and (c) solar modulationcorresponding to data in Fig. 7.5 for spheroidal VO2 particles (as in Fig. 4.1(d)), withthe shown aspect ratio m and a filling factor of 0.01, dispersed in a dielectric medium.The spheroids are randomly oriented.

88

7.2 Dielectric Matrix with VO2 Core-shellNanoparticle InclusionsSpectral transmittance and reflectance were calculated for a composite layerwith VO2 shells enclosing a dielectric core characterized with nc in a glassor plymer-like dielectric matrix with refractive index nm = 1.5 backed by aglass-like dielectric substrate with refractive index nsub = 1.5 as shown in thestructural model in Fig. 4.2. As in Chapter 7.1, the amount of VO2 was takento be 1 vol.%, and the layer thickness was 5 µm, equivalent to a 50 nm thickfilm.

0

20

40

60

80

100

500 1000 1500 2000 2500

Tra

nsm

ittan

ce (

%)

Wavelength (nm)

nc=1

semiconductingx/t=0

5101520

0

20

40

60

80

100

500 1000 1500 2000 2500

Tra

nsm

ittan

ce (

%)

Wavelength (nm)

nc=1

metallicx/t=0

5101520

(a) (b)

0

20

40

60

80

100

500 1000 1500 2000 2500

Tra

nsm

ittan

ce (

%)

Wavelength (nm)

nc=2

semiconductingx/t=0

5101520

0

20

40

60

80

100

500 1000 1500 2000 2500

Tra

nsm

ittan

ce (

%)

Wavelength (nm)

nc=2

metallicx/t=0

5101520

(c) (d)

Figure 7.7: Spectral transmittance calculated for spherical shells (thickness t) of VO2enclosing cores (diameter x) with the shown values of the dielectric constant nc andembedded in a dielectric medium represented by nm = 1.5, as depicted in Fig. 4.2.Data are shown for τ < τcr (semiconducting VO2) (a) and (c) and for τ > τcr (metallicVO2) (b) and (d). The material was 5 µm thick and contained 1 vol. % of VO2.

In Figure 7.7, spectral transmittance is presented for nc = 1 and 2 for dif-ferent ratios between the diameter x of the dielectric core and the thickness tof VO2 shell, specifically for x/t being 0, 5, 10, 15, 20. At the semiconducting

89

state (τ < τcr), the spectral shape is quite similar to the case of solid spheresin Chapter 7.2. T (λ ) drops with increasing x/t, with the case of nc = 1 beingmost sensitive to x/t. The metallic state (τ > τcr) presents a striking feature: avalley that moves to the long wavelength region with increasing x/t, indicat-ing an absorption as a result of a heavily damped plasmon resonance due tothe conduction electrons in the metallic state of VO2 [147]. The minimum ofthis valley, equivalent to the peak of the plasmon absorption, is shifted mono-tonically from ∼1.2 µm for x/t = 0 to ∼2 µm for x/t = 20. The valley isshallower when nc takes a larger value.

The integrated properties Tlum and Tsol for the semiconducting state and themetallic state and ∆Tsol are computed based on T (λ ) in Fig. 7.7 for cores withthe refractive index being 1, 1.5, 2, 2.5, and x/t is varied according to 0, 1, 2,...20. Data in Fig. 7.8 show that Tlum and Tsol for both semiconducting state(τ < τcr) and metallic state (τ > τcr) decrease monotonically with increas-ing x/t. However, as nc increases, this effect tends to become weaker, withdata lines for nc = 2.5 being almost flat. The solar transmittance modulation∆Tsol increases with decreasing nc, with nc = 1 (hollow spheres) having themaximum ∆Tsol . The largest ∆Tsol is 20.9%, which occurs for nc = 1 hollowspheres at x/t ≈ 10 and is significantly higher than ∆Tsol ∼ 16.7% of VO2solid nanospheres. However, Tlum for τ < τcr at this point is compromised, andconcomitantly decreased to ∼ 59%, i.e., a lower value than 73.5% for VO2solid nanospheres.

The simultaneous gain in solar modulation and luminous transmittance re-quires more discussion since high values of both solar transmittance and lumi-nous transmittance are desirable for most practical applications. It can be in-ferred from the above calculated results that the ideal case may lie at x/t = 10for hollow spheres, so an additional computation for a number of VO2 vol-ume fractions with values smaller than 0.01 were done to extract this informa-tion. The results show that requiring Tlum to be ∼ 73.5% gives ∆Tsol ≈ 14.5%for hollow spheres with x/t = 10 with VO2 volume fraction 0.005, i.e., 50%of the value used for the calculations in Fig. 7.7 and Fig. 7.8. With 60% ofthat amount (VO2 volume fraction 0.006), one can achieve Tlum ≈ 70.5% and∆Tsol ≈ 16.2%. These results, especially the latter, indicate a possibility of sig-nificantly decreasing the amount of VO2 to achieve properties only marginallyinferior to that of solid spheres, by using moderately thin-walled VO2 hollownanospheres.

90

30

40

50

60

70

80

90

100

0 5 10 15 20

Lum

inou

s tr

ansm

ittan

ce (

%)

x/t

semiconducting nc=11.5

22.5

30

40

50

60

70

80

90

100

0 5 10 15 20

Lum

inou

s tr

ansm

ittan

ce (

%)

x/t

metallic nc=11.5

22.5

(a) (b)

30

40

50

60

70

80

90

100

0 5 10 15 20

Sol

ar tr

ansm

ittan

ce (

%)

x/t

semiconducting nc=11.5

22.5

30

40

50

60

70

80

90

100

0 5 10 15 20

Sol

ar tr

ansm

ittan

ce (

%)

x/t

metallic nc=11.5

22.5

(c) (d)

16

18

20

22

24

0 5 10 15 20

Sol

ar m

odul

atio

n (%

)

x/t

nc=11.5

22.5

(e)

Figure 7.8: Luminous transmittance (a) and (b), solar transmittance (c) and (d), andsolar modulation (d) corresponding to data in Fig 7.7 for spherical shells (thickness t)of VO2 surrounding cores (diameter x) with the shown values of the dielectric constantnc and embedded in a dielectric medium represented by nm = 1.5, as depicted in Fig.4.2. Data are shown for τ < τcr (semiconducting VO2) (a) and (c) and for τ > τcr(metallic VO2) (b) and (d). The material was 5 µm thick and contained 1 vol. % ofVO2.

91

7.3 Composites of VO2 and ITOThe effective dielectric functions were derived from Bruggeman theory, asdescribed in Section 6.3. Spectral transmittance T (λ ) and reflectance R(λ )and integrated properties for both semiconducting state (τ < τcr) and metallicstate (τ > τcr) were computed for VO2-ITO composite layers backed by adielectric substrate with εsub = 2.25.

Fig. 7.9 shows T (λ ) and R(λ ) for VO2-ITO composite layers. The ITOmass thickness was kept constant at 200 nm while the VO2 volume fractionranged from 0.1 to 0.5 for both τ < τcr and τ > τcr. It can be seen that T (λ ) hasa higher value at 500 nm< λ <1500 nm, and the thermochromic switching isalso significant in this range. The most effective switching is centered around1200 nm, which is consistent with previous computations on VO2 particlesembedded in a dielectric matrix done in Chapter 7.1 and 7.2. R(λ ) increasesmonotonically when the wavelength gets longer than 1000 nm and it reaches60% at λ = 2500 nm. The spectral shape of T (λ ) and R(λ ) for low filling fac-tors of VO2 preserve the characteristic features of a 200 nm thick ITO layer ina previous study [16], which yielded a low thermal emittance of Etherm ≈ 0.2.The addition of VO2 will no doubt increase Etherm, but for a small VO2 con-tent Etherm should remain low and the composite film with low VO2 contentcan serve as a low-emittance coatings.

Fig. 7.10 shows integrated properties, specifically Tlum, Tsol and ∆Tsol as afunction of VO2 volume fraction. Tlum and Tsol are higher at τ < τcr than atτ > τc, which shows the presence of thermochromic switching. Tlum and Tsoldecrease progressively as the VO2 amount is increased. The fact that both Tlumand Tsol follow the same trend indicates that the thermochromic switching isnot localized in the near infrared but spreads out to the visible region as well.Fig. 7.10 shows ∆Tsol as a function of the VO2 content. ∆Tsol = 10% occursfor f ≈ 0.196, which simultaneously yields Tlum ≈ 59.7% for τ < τcr. ∆Tsolhas a broad maximum of ∼ 12.8% around f ∼ 0.35, where the VO2 units startto percolate in the Bruggeman model. These improvements on Tlum and ∆Tsolshow distinctive advantages over data for multilayer films reported recentlyfor VO2 with a top layer of Pt [148] or ZnO:Al [149], the best of which,yielded modestly Tlum ≈ 46%, Tsol ≈ 4% and Etherm ≈ 0.3 [149] .

92

0

20

40

60

80

100

500 1000 1500 2000 2500

Tra

nsm

ittan

ce (

%)

Wavelength (nm)

f=0.1 semicond.metallic

f=0.2 semicond.metallic

f=0.3 semicond.metallic

f=0.4 semicond.metallic

f=0.5 semicond.metallic

(a)

0

20

40

60

80

100

500 1000 1500 2000 2500

Ref

lect

ance

(%

)

Wavelength (nm)

f=0.1 semicond.metallic

f=0.2 semicond.metallic

f=0.3 semicond.metallic

f=0.4 semicond.metallic

f=0.5 semicond.metallic

(b)

Figure 7.9: Spectral transmittance (a) and reflectance (b) computed for ITO-VO2 com-posites as discussed in the main text. The VO2 volume fraction is denoted by f . Re-sults are shown for τ < τcr (semiconducting VO2) and for τ > τcr (metallic VO2).

93

0

20

40

60

80

100

0 0.1 0.2 0.3 0.4 0.5

Tlu

m (

%)

Filling factor f

semicond.metallic

(a)

0

20

40

60

80

100

0 0.1 0.2 0.3 0.4 0.5

Tso

l (%

)

Filling factor f

semicond.metallic

(b)

0

5

10

15

20

0 0.1 0.2 0.3 0.4 0.5

∆ T

sol (

%)

Filling factor f

(c)

Figure 7.10: Wavelength-integrated luminous (a) and solar (b) transmittance cor-responding to data in Fig. 7.9, and thermochromic modulation of the solar energythroughput (c), computed as a function of the VO2 fraction f for ITO-VO2 compos-ites. Results are shown for τ < τcr (semiconducting VO2) and for τ > τcr (metallicVO2). Circles indicate the data and connecting lines were drawn for convenience.

94

7.4 SummaryI have calculated the optical properties for layers containing suspensions ofsolid spheroidal VO2 particles, VO2 shells enclosing dielectric cores, andcomposites of VO2 and ITO. The results have shown significantly enhancedTlum and ∆Tsol for VO2 nanospheres embedded in dielectric layers, and re-sults on VO2 hollow spherical nanoshells have shown an even more improved∆Tsol and a potential to lower the material costs. Moreover, results on compos-ites of VO2 and ITO have shown the possibility of combining low Etherm, high∆Tsol and a moderately high Tlum in one composite layer. VO2 based nanother-mochromics have certainly opened up new vistas for energy efficient windowapplications.

95

8. Mg-doped VO2: Band gaps andPerformance Limits

This chapter covers the experimental work done on Mg-doped VO2 films, aswell as optical properties of films and nanoparticles revealed by modeling.Section 8.1 covers the study reported in Paper V, which investigated the effectof Mg-doping on the band gaps of VO2. Sections 8.2 and 8.3 correspond tothe studies documented in Paper VI, which explores the performance limits ofMg-doped VO2 films and nanoparticles derived from optical modeling resultsyielded from optical constants determined from experimentally prepared Mg-doped VO2 films.

8.1 Band GapsMg-doping offers lowered τcr, higher luminous transmittance [36] via a low-ered optical absorption in the visible and widened band gap [44]. A recentrealization of nanothermochromics with Mg-doped VO2 also suggests an ad-ditional benefit of bluish color [42]. However, we are not aware of any priorstudy that systematically and quantitatively addresses the effect of Mg-dopingon the band gap widening, and felt the obligation to do so, hence the result ofthis work.

We have grown more than 45 samples of Mg-doped VO2 thin films to athickness of 40 < d < 87 nm. More than 38 of them were evaluated with RBSfor elemental composition. The atom ratio Mg/(Mg+V) ≡ z was found to be0 < z < 0.21. 11 samples showed a presence of Si content of 0.45± 0.07 at.% in the Mg-doped VO2 layer, which was a result of the silicone adhesivefrom the high vacuum tape that was used in the deposition at the time. Thiscontamination was later eliminated by equipping the substrate holder withmetal clamps for sample-fixation. The rest of the samples have shown nothingbut Mg-doping.

The optical measurements were done as described in 5.6, the optical con-stants were determined as discussed in 5.7. The Mg-doped samples were eval-uated with the "surface+back" and "back" method. The (Mg, Si)-codopedsamples were evaluated with surface method. The band gap values were col-lected from 38 samples in total following the methodology as described in5.8. Band gaps were also extracted in the same way (see Section 5.8) fromliterature data on optical constants [47, 48, 137, 140, 142] for comparison.

97

The results were plotted as a function of atom ratio Mg/(Mg+V). The resultsof samples doped with only Mg are shown in Figs. 8.1 (a) and (b) for dataobtained from the "surface+back" method and "back" method, respectively.For the Mg-doped VO2 samples with Si contamination, the results are shownin 8.2. All of the data show a clear rising trend with increasing Mg content,and a linear dependency can be extracted as:

Eg(z) = Eg(0)+ξ z (8.1)

with parameters Eg(0) and ξ shown in Table. 8.1.

Table 8.1: Data for VO2 films with shown dopants and with optical constants evalu-ated with "surface", "back" and "surface+back" methods. Eg(0) and z are the param-eters in Eq.(8.1) acquired from linear fitting as described in the text.

Doping Evaluation method Eg(0) (eV) ξ (eV)

Mg "surface+back" 1.65 ± 0.04 3.85 ± 0.44

Mg "back" 1.80 ± 0.04 4.03 ± 0.39

Mg+Si "surface" 1.68 ± 0.04 7.05 ± 0.85

1.4

1.6

1.8

2

2.2

2.4

2.6

2.8

0 0.05 0.1 0.15 0.2 0.25 0.3

Band g

ap (eV

)

Mg/(Mg+V)

Present workKana Kana et al. 2011

Mlyuka et al. 2009Kakiuchida et al. 2008

Tazawa et al. 1998Verleur et al. 1968

1.4

1.6

1.8

2

2.2

2.4

2.6

2.8

0 0.05 0.1 0.15 0.2 0.25 0.3

Band g

ap (eV

)

Mg/(Mg+V)

Present workKana Kana et al. 2011

Mlyuka et al. 2009Kakiuchida et al. 2008

Tazawa et al. 1998Verleur et al. 1968

(a) (b)

Figure 8.1: Band gap as a function of atom ratio Mg/(Mg+V) for Mg-doped VO2films. Data were evaluated from optical constants obtained from Rs Rb and T us-ing the "surfce+back" and "back" method. Literature data were obtained from opticalconstants determined by Verleur et al.[137], Tazawa et al.[140], Kakiuchida et al.[47],Mlyuka et al.[48], and Kana kana et al.[142]

Regarding the data from Mg-doped VO2 films, 3 data points near z ∼ 0.05and 2 data points near z ∼ 0.11 lie below the linear trend line in Fig. 8.1

98

1.4

1.6

1.8

2

2.2

2.4

2.6

2.8

0 0.05 0.1 0.15 0.2 0.25 0.3

Band g

ap (eV

)

Mg/(Mg+V)

Present workKana Kana et al. 2011

Mlyuka et al. 2009Kakiuchida et al. 2008

Tazawa et al. 1998Verleur et al. 1968

Figure 8.2: Band gap as a function of atom ratio Mg/(Mg+V) for (Mg, Si)-codopedVO2 films with 0.45 ± 0.07 at.% of Si. Data were evaluated from optical constants ob-tained from Rs and T using the "surface" method. Literature data were obtained fromoptical constants determined by Verleur et al.[137], Tazawa et al.[140], Kakiuchida etal.[47], Mlyuka et al.[48], and Kana kana et al.[142]

(a) but on the line in Fig. 8.1 (b). These results were from the samples thatwere metallic as deposited and went through post-anealing, which might havecaused sub-band-gap absorption. Despite the scattering of the data, the slopeof the trend ξ obtained from both the "surface+back" and "back" method wereof a very similar value within each other’s statistical error, which adds to thereliability of our data.

The data from Mg-doped VO2 films containing Si evaluated using the sur-face method in Fig. 8.2 show a stronger trend of band gap increase with Mgcontent z: the magnitude of ξ is close to being doubled, showing a clear ef-fect of Si as a "band gap enhancer". However, the data were only available forsmall amount of doping z < 0.1, and also due to the un-intended nature of thepresence of Si, it is wise to be cautious and more systematic studies should beconducted in the future to confirm the effect of Si-doping on the band gap ofVO2.

In summary, this part of the study has shown the quantitative relation ofMg-doping on the band gap of VO2 with thorough investigation conductedon a large number of samples. The band gap of VO2 doped with only Mgwas shown to increase by 3.9 ± 0.5 eV per unit atom ratio of Mg/(Mg+V)for 0<Mg/(Mg+V)<0.21. In addition, we have also found strong indicationsthat the presence of a small amount of Si may enhance the band gap evenmore. These data shed light on the enhanced luminous transmittance (usuallylinked to a widened band gap) in Mg-doped VO2 films and is a significantcontribution to the knowledge and design parameters of energy-efficient ther-mochromic windows.

99

8.2 Modeling Results: Mg-doped VO2 Thin Films8.2.1 Films with Fixed Thickness 50 nmOptical constants of experimentally prepared films were used as input datafor the computation of spectral transmittance and reflectance for 50 nm thickfilms. The integrated properties Tlum and ∆Tsol were calculated using Eq. (2.7)and (3.1). Fig.s 8.3 (a) and (b) show Tlum as a function of Mg/(Mg+V) forthe semiconducting state and metallic states, respectively. Both Tlums increasewith Mg content, consistent with the findings [44] that Mg-doping decreasesthe luminous absorptance. ∆Tsol as a function of Mg/(Mg+V) shown in Fig.8.3(c) has a decreasing trend with Mg-doping.

0

0.2

0.4

0.6

0.8

1

0 0.05 0.1 0.15 0.2 0.25 0.3

Tlu

m s

emic

ond.

Mg/(Mg+V)

0

0.2

0.4

0.6

0.8

1

0 0.05 0.1 0.15 0.2 0.25 0.3

Tlu

m m

etal

lic

Mg/(Mg+V)

(a) (b)

0

0.05

0.1

0.15

0.2

0.25

0 0.05 0.1 0.15 0.2 0.25 0.3

∆ T

sol

Mg/(Mg+V)

(c)

Figure 8.3: Luminous transmittance Tlum and solar energy transmittance modulation∆Tsol as a function of Mg/(Mg+V) atom ratios for VO2 films of 50 nm. Tlum of semi-conducting (τ < τcr) and metallic (τ > τcr) state are shown in (a) and (b), respectively.Data for solar modulation ∆Tsol are shown in (c).

These results indicates that the advantage of Mg-doping lies in the loweringof luminous absorptance. One has to search for a judicious choice of thickness

100

to allow the advantage of increased Tlum to outweigh the disadvantage of de-creased ∆Tsol , and hence the study presented in the following section.

8.2.2 Performance Limits for One-layer FilmsSpectral transmittance and reflectance were calculated for films with variedthicknesses in the range 0 < d < 200 nm. Integrated properties Tlum and ∆Tsolare computed for each Mg/(Mg+V) atom ratio and also for each thickness, andtheir relation is presented in Fig. 8.4. In this figure Tlum is plotted versus ∆Tsol ,so that data with favorable total performances can be easily identified in theupper-right area of the figure. This plotting scheme gives a big advantage overFig. 8.3, which shows fragmented performance only. In a sense, Fig. 8.4 canbe viewed as a qualitative chart for the performance limit of Mg-doped VO2films. The evolution of data points from small to large d values starts from thetop-left corner and ends at the lower-right corner of the figure. The data aredivided according to Mg/(Mg+V)≡ z atom ratio into 4 groups: z = 0; 0 < z <0.06; 0.06 < z < 0.11; 0.11 < z < 0.21.

Samples with 0.11 < z < 0.21 were "over-Mg-doped" and have bad perfor-mances, so they are not discussed. The rest of the samples all have the sametrend with thickness: after an initial drop in relation to d, the data reaches aplateau region where the increase of thickness brings about increased ∆Tsolwith almost unchanged Tlum. The key to good performance is to sustain a highTlum value before reaching this plateau, which is observed from most of theMg-doped films; consequently, most of the data points from moderately Mg-doped films lie to the top-right to corresponding data from pure VO2, suggest-ing a better performance than films of pure VO2. For a target of ∆Tsol ∼ 10%,one can achieve Tlum ∼ 45 % for the semiconducting state and a slightly lowervalue for the metallic state with Mg-doped films, whereas, undoped films onlyoffer Tlum ∼ 30% at the most.

101

0

0.2

0.4

0.6

0.8

1

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14

Tlu

m s

emic

ond.

∆ Tsol

Mg/(V+Mg) = z, z = 00 < z < 0.06

0.06 < z < 0.110.11 < z < 0.21

(a)

0

0.2

0.4

0.6

0.8

1

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14

Tlu

m m

etal

lic

∆ Tsol

Mg/(V+Mg) = z, z = 00 < z < 0.06

0.06 < z < 0.110.11 < z < 0.21

(b)

Figure 8.4: Tlum as a function of ∆Tsol for VO2 films with shown Mg/(Mg+V) ≡ zatom ratio. The film thickness was varied 0 < d < 200 nm. Data were shown forsemiconducting states (τ < τcr) (a) and metallic states (τ > τcr) (b).

102

8.2.3 Films with One-layer Antireflection CoatingsIn this section, an attempt to improve Tlum and ∆Tsol is made with computa-tion assuming the application of one layer of antireflection (AR) coating withnAR = 1.5 (typical of SiO2 and many polymers) with thickness 0 < dAR <300 nm on the top of the VO2 films with Mg/(Mg+V) values and thicknessd values listed in Table 8.2, and results visualized in Fig. 8.5. Fig. 8.5 showsa trajectory with increasing dAR starting from the points marked with trian-gles with high-performance points marked with circles. It is clear that ARtreatment can improve Tlum by 8% in absolute value while increasing ∆Tsolby 2-3%. Well chosen parameters (bold numbers shown in Table 8.2), canoffer ∆Tsol = 9.2%, with Tlum(τ < τcr) =59.3% and Tlum(τ > τcr) =57%. Al-ternatively, one can and achieve ∆Tsol =12.8% and maintain Tlum(τ < τcr) =47.5% and Tlum(τ > τcr) = 44%; or ∆Tsol =15.4% with Tlum ∼ 30%. Thesehighlighted performances are similar to the best performance of AR treatmentof TiO2/VO2 reported by Chen et al. [45], who reported ∆Tsol ∼ 15% withTlum(τ < τcr) = 49.5% and Tlum(τ > τcr) = 45%.

0.0

0.2

0.4

0.6

0.8

1.0

0.04 0.06 0.08 0.1 0.12 0.14 0.16

Tlu

m

∆ Tsol

(a) semicond.

Mg/(Mg+V)=0, d=70 nmMg/(Mg+V)=0.006, d=90 nmMg/(Mg+V)=0.024, d=80 nm

Mg/(Mg+V)=0.055, d=110 nmMg/(Mg+V)=0.055, d=147.5 nm

0.0

0.2

0.4

0.6

0.8

1.0

0.04 0.06 0.08 0.1 0.12 0.14 0.16

Tlu

m

∆ Tsol

(b) metallic.

Mg/(Mg+V)=0, d=70 nmMg/(Mg+V)=0.006, d=90 nmMg/(Mg+V)=0.024, d=80 nm

Mg/(Mg+V)=0.055, d=110 nmMg/(Mg+V)=0.055, d=147.5 nm

(a) (b)

Figure 8.5: Luminous transmittance Tlum as a function of solar energy modulation∆Tsol for VO2 films with shown Mg/(Mg+V) atom ratios and film thicknesses d. Thecurves form trajectories that evolve from dAR = 0 marked with triangles and end atdAR =300 nm. Data are shown for the semiconducting state (τ < τcr) (a) and metallicstate (τ > τcr) (b). Data points marked with circles are the high-performance pointsof each trajectory and are listed in Table 8.2.

103

Table 8.2: Luminous transmittance Tlum for the semiconducting state (τ < τcr) andmetallic state (τ > τcr) and solar energy modulation ΔTsol for Mg-doped VO2 filmswith shown Mg contents and thickness d. Data are reported for before (numbers inbrakets) and after the application of an antireflection coating. Parameters in bold arethe high-performance points yielded by well-chosen parameters discussed in the text.

Mg/(Mg+V) d dAR ΔTsol Tlum(τ < τcr) Tlum(τ > τcr)

(nm) (nm) (%) (%) %

0 70 265 11.6 (8.4) 45.3 (35.1) 41.5 (32.8)0.006 90 260 9.2 (7.3) 59.3 (50.7) 57 (49.7)0.024 80 265 12.4 (9.3) 44.6 (36.6) 39 (31.6)0.055 110 265 12.8 (10.4) 47.5 (42.5) 44 (39.4)0.055 147.5 95 14.4 (12.3) 37 (30.2) 33.2 (29.7)0.055 147.5 155 15.4 (12.3) 32.1 (30.2) 29.8 (29.7)

104

8.3 Modeling Results for Mg-doped VO2 Nanospheres8.3.1 5-µm-Thick Layer Containing 1 vol.% NanospheresThe optical properties of Mg-doped VO2 nanospheres were calculated us-ing effective medium theory Maxwell-Garnett theory, as explained in section4.2.1, which was also used for calculating results in section 7.1. The calcula-tion assumes 1 vol.% Mg-doped VO2 nanospheres distributed in a 5-µm-thickglass or polymer-like dielectric matrix with n=1.5. The spectral transmittanceand reflectance were calculated and the integrated properties Tlum and ∆Tsolwere calculated. In the section, VO2 nanospheres are of equivalent amounts tothat of films in Section 8.2.1 so that a comparison between nanoparticles andfilms can be made in parallel.

0.5

0.6

0.7

0.8

0.9

1

0 0.05 0.1 0.15 0.2 0.25 0.3

Tlu

m s

emic

ond.

Mg/(Mg+V)

0.5

0.6

0.7

0.8

0.9

1

0 0.05 0.1 0.15 0.2 0.25 0.3

Tlu

m m

etal

lic

Mg/(Mg+V)

(a) (b)

0

0.05

0.1

0.15

0.2

0.25

0 0.05 0.1 0.15 0.2 0.25 0.3

∆ T

sol

Mg/(Mg+V)

(c)

Figure 8.6: Luminous transmittance Tlum and solar transmittance modulation ∆Tsol asa function of Mg/(Mg+V) atom ratios for 5-µm-thick nanocomposite layers contain-ing 1 vol.% of VO2. Tlum for semiconducting (τ < τcr) and metallic (τ > τcr) statesare shown in (a) and (b), respectively. Data for solar modulation ∆Tsol are shown in(c).

105

The same as in Section 8.2.1, it can be seen in Fig.s 8.6 that Tlum increaseswhereas ∆Tsol decreases with Mg content. By the same token as in 8.2.1, it isnecessary to investigate optical properties with different thicknesses, as in thenext section.

8.3.2 Performance Limits for NanospheresLuminous transmittance Tlum for semiconducting (τ < τcr) and metallic state(τ > τcr) and solar transmittance modulation ∆Tsol for nanocomposites con-taining 1 vol.% Mg-doped VO2 were computed for layer thicknesses 0 < d <20 µm. The results of Tlum as a function of ∆Tsol in the same fashion as Fig.8.4 are shown in Fig.8.7, with data divided into 4 groups: z = 0, 0 < z < 0.06,0.06 < z < 0.11 and 0.11 < z < 0.21. Fig.8.7 shows general trends of Tlum asa function of ∆Tsol , to some degree it can be viewed as a qualitative chart forperformance limit of Mg-doped VO2 nanoparticles. Specific data to comparepure and Mg-doped VO2 are listed in Table 8.3.

It can be seen in Fig. 8.7 that in the form of nanospheres, pure VO2 isthe superior option. ∆Tsol =10% can be achieved with high luminous trans-mittance Tlum(τ < τcr) ∼ 82% and Tlum(τ > τcr) ∼ 77.5%, the equivalentTlum for VO2 films less than 35%. Furthermore, a large solar modulation∆Tsol ∼20% (unattainable for films) can be achieved with Tlum(τ < τcr) ∼66% and Tlum(τ > τcr)∼ 57%. The reason for the advantage of undoped VO2nanoparticles can be explained by i) both undoped and the Mg-doped VO2have similar luminous absorption; ii) the positive effect on Tlum seen in Mg-doped VO2 films is absent in Mg-doped VO2 nanoparticles; iii) the plasmonresonance in the NIR in the metallic states is stronger for the undoped VO2nanoparticles. The net effect of a large ∆Tsol in undoped VO2 outweighs theincrease of Tlum offered by the Mg-doped VO2 in nanoparticle composites.

Table 8.3: Luminous transmittance Tlum for semiconducting state (τ < τcr) and metal-lic state (τ > τcr) and solar energy modulation ∆Tsol for nanocomposite layers con-taining 1 vol.% Mg-doped VO2 with shown Mg contents and thickness d.

Mg/(Mg+V) d ∆Tsol Tlum(τ < τcr) Tlum(τ > τcr)

(µm) (%) (%) (%)

0 1 4.6 87.8 86

0 5 16.8 72.4 65.2

0 10 23.7 57 46.5

0.055 1 3.0 88.5 87.4

0.055 5 11.9 75 70.4

0.055 10 18.4 61.2 54

106

0

0.2

0.4

0.6

0.8

1

0 0.05 0.1 0.15 0.2 0.25 0.3

Tlu

m s

emic

ond.

∆ Tsol

Mg/(V+Mg) = z, z = 00 < z < 0.06

0.06 < z < 0.110.11 < z < 0.21

(a)

0

0.2

0.4

0.6

0.8

1

0 0.05 0.1 0.15 0.2 0.25 0.3

Tlu

m m

etal

lic

∆ Tsol

Mg/(V+Mg) = z, z = 00 < z < 0.06

0.06 < z < 0.110.11 < z < 0.21

(b)

Figure 8.7: Tlum as a function of ∆Tsol for nanocopmposite layers containing 1 vol.%VO2 nanospeheres with shown Mg/(Mg+V) atom ratios. Data are shown for semicon-ducting states (τ < τcr) (a) and metallic states (τ > τcr) (b).

8.4 SummaryThis chapter reported our complete study on Mg-doped VO2 for ther-mochromic window applications. Section 8.1 presented a linear relationshipbetween the band gap and Mg-doping, which was found to increase by 3.9 ±0.5 eV per unit atom ratio of Mg/(Mg+V) for 0<Mg/(Mg+V)<0.21. Si maybe a potentially good additive to enhance this effect.

107

Section 8.2 and 8.3 dived further into the optical properties of Mg-dopedVO2 films and nanoparticle composite layers, by means of optical modeling. Itwas found that Mg-doped VO2 films with Mg/(Mg+V) < 0.06 have much im-proved luminous transmittance and solar transmittance modulation. The per-formance can be further enhanced by applying one layer of antireflection coat-ing on top of Mg-doped VO2 films. On the other hand, the modeling resultsfor nanocomposite films showed that the un-doped VO2 is preferable. To in-clude all the speculation on performance limits (joint performance of Tlum and∆Tsol), Fig. 8.8 was constructed as a qualitative/approximate chart showingTlum as a function of ∆Tsol to be expected for undoped and Mg-doped VO2 ofvarious kinds, as shown. It should be noted that well-designed multilayer films[34, 48] can perform better than films with one-layer antireflection treatment,and that one-layer antireflection may also deliver exceptionally good resultsin some cases [45]. Nonetheless, simply by using a nanoparticle composite,one can effortlessly leap over a huge gap and achieve the nanothermochromicperformance that is among the best of all.

(a)

Figure 8.8: Schematic illustration of approximate performance limits of VO2 basedmaterials for thermochromic energy efficient window applications. Data are shownfor Tlum in semiconducting state (τ < τcr) as a function of solar energy transmittancemodulation ∆Tsol for undoped and Mg-doped VO2 based materials: thin films, thinfilms with one-layer AR treatments and nanocomposite layers.

108

9. Sputter-deposited VO2 Nanorods

This chapter covers the studies done in paper IV. In this study we grew VO2nanorods using DC magnetron sputtering and explored the growth parame-ters extensively and found out four factors that promoted rod-like growth ofVO2: substrate seeding, substrate temperature, O2 gas flow, and thickness. Theoptical data showed enhanced absorptance and can be partially explained byoptical modeling using the effective medium theory. A discussion on possiblegrowth mechanisms was attempted.

9.1 Growth Conditions and MorphologyThe fabrication of the VO2 nanorods was described in Section 5.1.2. The im-portant parameters are i) the O2 flow varied as illustrated in Fig. 5.2 (a), (b) and(c); (ii) substrates, three kinds: glass, Au/glass and Au/SiO2 nanopillars/glass:iii) The substrate temperature which was kept at either 450 C or 550 C andiv) thickness, varied 34 ≤ d ≤ 102 nm.

The samples and corresponding growth conditions and their resultant mor-phology are listed in Table 9.1, with links to scanning electron micrographsFigs. 9.1, 9.2 and 9.3.

109

(c) #3 (d) #4

(e) #5 (f) #6

(a) #1 (b) #2

Figure 9.1: Scanning electron micrographs observed from the surface normal (topviews) of the VO2 samples 1-6 grown under the conditions described in Table 9.1.Note that the sale bar is different for samples 1-4 and samples 5-6.

110

(g) #9 (h) #9 (i) #9

(a) #7 (b) #7 (c) #7

(j) #10 (k) #10 (l) #10

(d) #8 (e) #8 (f) #8

Figure 9.2: Scanning electron micrographs for VO2 samples 7-10 grown under theconditions described in Table 9.1. (a), (d), (g), (j) show top view; (b)(e)(h)(k) weretaken 70 off the surface normal of the samples; (c),(f), (i), (l) show cross sectionpictures.

VO2, V6O13

Au

SiO2

nanopillars

(SiO2)

(a) (b) (c)

Figure 9.3: (a) and (b) show scanning electron micrographs viewed from the crosssection for VO2 sample 11, grown under conditions described in Table 9.1 from azoomed-out view and a zoomed-in view, respectively. (c) is a schematic illustration ofsample 11.

111

Tabl

e9.

1:Fa

bric

atio

npa

ram

eter

san

dm

orph

olog

yof

the

VO2

sam

ples

.Sam

ples

inbo

ldsa

mpl

enu

mbe

rsw

ere

chos

enfo

rde

taile

dan

alys

isan

dop

tical

mod

elin

g.

IDN

o.se

tup

inte

mpe

ratu

rem

ass

thic

knes

sm

orph

olog

y

Fig.

5.2

(C

)su

bstr

ate

(nm

)

1a

450

glas

s62

.5co

mpa

ctfil

mFi

g.9.

1(a

)

2b

450

glas

s62

.5ro

ugh

film

with

grai

nsFi

g.9.

1(b

)

3b

550

glas

s34

smal

lelo

ngat

edgr

ains

Fig.

9.1

(c)

4b

550

Au/

glas

s34

shor

trod

sFi

g.9.

1(d)

5b

550

glas

s68

larg

eel

onga

ted

grai

nsFi

g.9.

1(e)

6b

550

Au/

glas

s68

mod

erat

ero

dsFi

g.9.

1(f)

7c

550

glas

s68

shor

trod

sFi

g.9.

1Fi

g.9.

2(a)

-(c)

8c

550

Au/

glas

s68

occa

sion

allo

ngro

dsFi

g.9.

2(d)

-(f)

9c

550

glas

s10

2de

nse

long

rods

Fig.

9.2(

g)-(

i)

10c

550

Au/

glas

s10

2lo

ngro

dsan

dlo

ngw

ires

Fig.

9.2(

j)-(

l)

11a

450

Au/

SiO

2na

nopi

llars

/gla

ss34

long

wir

esFi

g.9.

3

112

9.1.1 Influence of O2 gas flowSEM micrographs of samples 1 and 2 are Fig.s 9.1 (a) and (b). Both are de-posited at the same temperature and same thickness on the same substrate, butsample 2, grown with setup Fig. 5.2 (b), has much larger grains compared tosample 1, grown with Fig. 5.2 (a).

SEM micrographs of samples 5 is shown in Fig. 9.1 (e) and sample 7 in Fig.9.2(a)-(c). With otherwise the same growth conditions, sample 7, grown withsetup in Fig. 5.2 (c), has much more distinct rod-like growth.

To sum up, the effectiveness of setups into promoting large grains and rod-like growth can be ranked:

Fig.s 5.2 (c) > Fig.s 5.2 (b) > Fig.s 5.2 (a).

9.1.2 Influence of SubstratesThe effect of substrate can be observed by comparing sample pairs, namely,samples 5 and 6 (Figs . 9.1 (e) and (f)); samples 7 and 8 (Fig.s 9.2(a)-(c)and (d)-(f)); samples 9 and 10 (Fig.s 9.2(g)-(i) and (j)-(l)). Within each pair,the conditions are the same, apart from the substrate being either glass orAu/glass. Samples 6, 8 and 10 grown on Au/glass has show much more dis-tinct rod-like features than their counterpart samples 5, 7, 9 grown on glasssubstrates.

In addition, it is interesting to compare samples 11 with the rest of the sam-ples. The detailed template Au/SiO2 nanopillars/glass substrate have triggeredthe most salient nanowire growth, with length in µm order at substrate temper-ature 450 C, with quite small thickness compared to other rod-like samples,(with only half the thickness of samples 5-8 and one third that of samples 9and 10).

To rank the effectivity of substrates on promoting rod-like growth:

Au/SiO2 nanopillars/glass > Au/glass > glass.

9.1.3 Influence of Substrate TemperatureSamples 2 and 5 are shown in SEM micrographs Fig.s 9.1 (b) and (e). Both areof similar thickness with the same growth parameters apart from the substratetemperature. Sample 5, grown at 550 C, has much larger and more elongatedgrains than sample 2, grown at 450 C.

One can conclude that to grow nanorods, the 550 C is better than 450 C.

9.1.4 Influence of ThicknessTo clarify the influence of thickness, it is convenient to divide samples into 4groups: A) samples 3 (Fig.s 9.1 (c)) and 5 (Fig.s 9.1 (e)); B) samples 7 (Fig.s9.2 (a)-(c)) and 9 (Fig.s 9.2 (g)-(i)); C) samples 4 (Fig. 9.1 (d)) and 6 (Fig. 9.1

113

(f)); D) samples 8 (Fig.s 9.2 (c)-(f)) and 10 (Fig.s 9.2 (j)-(l)). Within each ofthese groups, the only differing parameter is thickness, and it can be seen thatthe thicker the sample the more rod-like growth it exhibits.

9.2 Optical Data and Modeling9.2.1 Experimental Optical DataThe total and diffuse spectral transmittance and reflectance for samples 1, 2,7 and 9 were recorded with a Lambda 900 instrument (Section 5.6) in 300< λ < 2500 nm, and were subjected to detailed optical analysis. The experi-mentally recorded total and diffuse spectral transmittance and reflectance areshown in Fig.s 9.4 and 9.5, respectively; and absorptance in Fig. 9.6 (a). Itcan be seen that rod-like samples 7 and 9 scatter light significantly more. Inaddition, the absorptance data for samples 7 and 9 are increased in the range300 < λ < 1500 nm for the semiconducting state, and shifted upwards in theentire measured spectral range for the metallic state, whereas the absorptanceof samples 1 and 2 is lower and almost identical. Optical modeling with theeffecive medium theory was applied to offer a partial explanation for this en-hanced absorption, which is not seen in sample 1 and 2 but present in samples7 and 9.

9.2.2 Optical ModelingOptical constants n and k were extracted from thin film sample 1 and wereused as input data for analysis using the effective medium theory. For thesesamples we used the Bruggeman theory for randomly oriented spheroids (Eq.(4.41) and (4.42)) to model the optical properties of the rod-like films, as de-scribed in Section 4.2.2. For the model parameters we relied on close examina-tion of the SEM pictures, which yielded roughly the parameters and suggesteda model containing a dilute top layer with VO2 particles with aspect ratio mt ,filling facotr ft and thickness dt and a dense bottom layer with correspondingparameters mb, fb, db satisfying:

d = ft dt + fbdb (9.1)

where d is the mass thickness as tabulated in Table 9.2. More detailed numberswere estimated by fine tuning the parameters to match spectral transmittanceand reflectance. The parameters are shown in Table 9.2.

The spectral absorptance yielded by the optical modeling is shown in Fig.9.6 (b). The modeled data (Fig. 9.6 (b)) show the same trend as the experi-mental data (Fig. 9.6 (a)) in the metallic state: samples 1 and 2 have relativelylower and almost identical absorptance. In contrast, samples 7 and 9 have an

114

upward shift in the modeled spectral range, especially that in the range 1000< λ < 2500 nm agrees very well with the experimental data. In the semicon-ducing state, the modeling data represent very well samples 1 and 2, but showless absorptance for nanorod samples 7 and 9 compared with the experimen-tal data. This effect maybe caused by the presence of defects such as excessof oxygen that is very likely to form at high deposition temperature. Multi-ple light-scattering in the nanorod samples is another possible cause for thisenhanced absorption.

(a)

(b)

0

0.2

0.4

0.6

0.8

1

500 1000 1500 2000 2500

Tra

nsm

itta

nce

Wavelength (nm)

Expsemicond. sample 1

semicond. sample 2

semicond. sample 7

semicond. sample 9

metallic sample 1

metallic sample 2

metallic sample 7

metallic sample 9

0

0.2

0.4

0.6

0.8

1

500 1000 1500 2000 2500

Reflecta

nce

Wavelength (nm)

Expsemicond. sample 1

semicond. sample 2

semicond. sample 7

semicond. sample 9

metallic sample 1

metallic sample 2

metallic sample 7

metallic sample 9

Figure 9.4: Experimental spectral total transmittance (a) and reflectance (b) for VO2samples tabulated in Table 9.1. Data are shown for semiconducting states and metallicstates.

9.3 Possible Growth MechanismThe growth mechanism is unclear, but may be similar to high-temperatureglancing-angle deposition (HT-GLAD) [150–154], which induces nanowiregrowth in metals when the substrate temperature is higher than around onethird of the metal’s melting point and typically at a high deposition angle of ∼80 off the surface normal. In this study, it was found that significant rod-like

115

0

0.05

0.1

0.15

0.2

0.25

0.3

500 1000 1500 2000 2500

Ref

lect

ance

/Tra

nsm

ittan

ce

Wavelength (nm)

Td sample 2Td sample 7Td sample 9Rd sample 2Rd sample 7Rd sample 9

Figure 9.5: Diffuse transmittance Td and reflectance Rd recorded for VO2 samplestabultaed in Table 9.1.

(a)

(b)

0

0.2

0.4

0.6

0.8

1

500 1000 1500 2000 2500

Absorp

tance

Wavelength (nm)

Expsemicond. sample 1semicond. sample 2semicond. sample 7semicond. sample 9

metallic sample 1metallic sample 2metallic sample 7metallic sample 9

0

0.2

0.4

0.6

0.8

1

500 1000 1500 2000 2500

Absorp

tance

Wavelength (nm)

Calcsemicond. model 1semicond. model 2semicond. model 7semicond. model 9

metallic model 1metallic model 2metallic model 7metallic model 9

Figure 9.6: Spectral absorptance for VO2 samples in semiconducting and metallicstates. (a) shows experimental data for samples tabulated in Table 9.1; (b) shows theresults from theoretical modeling using model parameters given in Table 9.2

116

Table 9.2: Optical models for samples No. 1, 2, 7, 9.

sample/model dt ft mt db fb mb

1 - - - 62.5 1 -

2 - - - 70 0.893 1

7 125 0.05 3 125 0.494 3

9 500 0.01 10 200 0.485 10

growth can be attributed to the the oxygen gas geometry which is related toparticle flux of vanadium and oxygen. The rod-like growth induced by goldseeds maybe related to well-known gold-catalyzed vapor-liquid-solid growthmode [155] that is well-established for growing semiconductor nanowires.

117

10. Concluding Remarks and Outlook

In Chapter 7 It was shown by computation that nanothermochromics, i.e., ther-mochromic material consisting of VO2 nanoparticle composites, can providea promising solution to long sought-after VO2-based thermochromic materi-als for energy-efficient fenestration. By using nanothermochromics, luminoustransmittance Tlum and the modulation of solar energy throughput betweensemiconducting state (τ < τcr) and the metallic state (τ > τcr) ∆Tsol can be in-creased to several times that of conventional thermochromic materials basedon thin films. Core-shell nanoparticles were shown by calculation to increase∆Tsol and produce a good performance with less material. The optical prop-erties of ITO and VO2 composites were revealed by computation to possessboth low-emittance and good thermochromism.

Chapter 8 reports works done on Mg-doped VO2. Experimental work wascarried out on Mg-doped VO2 films. Optical analysis based on absorptionmeasurements showed that a quantitative relation between the larger band gapof VO2 and Mg content can be established: the band gap of VO2 increases by3.9±0.5 eV per unit atom ratio of Mg/(Mg+V) for 0<Mg/(Mg+V)<0.21. Awider band gap has the advantage of moving the absorption out of the visibleregion thereby increasing the visible transmittance. Optical modeling, usingthe optical constants extracted from undoped and Mg-doped VO2 films, wasdone to estimate the optical properties and performance limits of Mg-dopedfilms and nanoparticles. It was found that the Mg-doped films, especially thosemoderately doped with 0<Mg/(Mg+V)<0.06, are superior to those un-dopedand the performance of Mg-doped films can be further boosted with one layerof antireflection coatings with a refractive index of n = 1.5. Calculation resultson nanoparticle composites point to the un-doped VO2 nanoparticles being amore favorable option to the Mg-doped ones. The best illustration to describethe studies on the thermochromic performance of VO2 films and nanoparti-cles is presented in Fig. 8.8, with VO2 based nanothermochromics as the bestchoice.

Another experimental study was done on the growth of VO2 nanorods, re-ported in Chapter 9. The type of substrates, temperature of substrates duringdeposition, the thickness and the geometry of the oxygen gas inlet were iden-tified as important factors that infuence the morphology. Optical modeling ofsuch particles was done and have shown good aggreement between the ex-perimental and simulated optical absorption. Further studies on this subject

119

are on-going in collaboration with Kyoko Namura and Prof. Suzuki from theKyoto University group.

In the future, more synthesis and characterization of an integration of ma-terials combining VO2 nanocomposites and other functional materials suchas ITO, electrochromic layers, solar cells or materials similar to "solar en-ergy scavenging layers" [156] etc. should be done to achieve maximum en-ergy efficiency. Future energy-efficient buildings could be extensively utiliz-ing a clever combination of chromogenic smart windows accompanied bylow-emittance or solar control coatings [7], photocatalytic paints for indoorair-cleaning [157], solar absorbers [158, 159], coatings with radiative coolingproperties for cooling [160] and water condensation [161] as well as photo-voltaics.

As to the "outlook" further into the future, I would like to divert from theso far rather "dry" narrative on VO2 and "dream" a little. Nanomaterials areof great relevance to energy efficiency [11, 157, 162] owing to their smallsize, large surface area, great chemical reactivity and mechanical strength[163]. They will become essential constituents in future devices for light-ing [164–166], detection/sensing [167–171], solar cells [167, 172–174], smartwindows [175, 176], solar absorbers [158], catalysts [177, 178], photocata-lysts [179, 180], batteries [181, 182], drug delivery [183, 184], cancer ther-apy [185, 186], and mechanical/structual applications [187–189], etc. For trueenergy-efficiency, future mankind will need to borrow much from nature’smighty ability to "make order out of randomness" to ensure that design, sythe-sis, process, end-product and recycling are done with ease and generate littleor no waste. Future material scientists will unleash the power of nature’s liv-ing factories, i.e., microorganisms such as bacteria and fungi, for the synthesisof materials, especially nanomaterials. Despite the infancy of the field, thereis already a sizable achievement in synthesizing nanoparticles of metal [190–194], metal oxides [195], sulfides [191, 196, 197], SiOx [197], magnetic iron-based materials [198], ferroelectric materials [199], etc. In addition, materialsof algal [200] and viral [201] origin can serve as batteries themselves. Let usnot forget the ancient biogenic material -silk, which has newly discovered po-tential as a biodegradable building block for flexible electronic displays, tissuerepairs, implantable optical systems and drug delivery, etc.[202]. On a macro-scopic scale, future mankind will take more biomimetic approaches [203] andadapt the concept of "cradle-to-cradle design" [204] to make sure that the flowof energy and materials resembles that of cycles in a naturally self-sustainingecosystem, and that nothing goes to waste. This is how I envision the future.

120

11. Swedish Summary

121

I Kapitel 7 förklaras, med hjälp av beräkningar, att termokroma materialtillverkade av VO2 nanopartiklar är mycket lovande VO2-baseradekompositmaterial för energieffektiva fönster. Dessa material, s.k.nanotermokroma material (nanothermochromics), kan öka transmittansen avdet synliga ljuset och i hög grad öka möjligheten att reglera transmittansen avnära infraröda våglängder (genom att variera temperaturen så att materialetgår mellan halvledartillståndet (τ < τcr) och det metalliska tillståndet(τ > τcr), jämfört med traditionella termokroma material tillverkade avtunna filmer. Beräkningar visade även att nanopartiklar av VO2, med skalav ett annat material, gör att reglering av solljuset kan åstadkommas medmindre mängd använt material. I detta kapitel beskrivs också hur de optiskaegenskaperna hos kompositer av ITO och VO2 beräknades, för att visamöjligheten att utveckla både lågemitterande och termokroma skikt.Kapitel 8 beskriver det experimentella och teoretiska arbetet med

Mg-dopad VO2. Optiska analyser av resultat från absorptionsmätningarpå dessa filmer visade på en tydlig relation mellan ökat bandgap hosVO2 och mängden tillsatt Mg – bandgapet hos VO2 ökar med 3.9 ±0.5 eV per enhet av kvoten Mg/(Mg+V), för dopningskoncentrationer iområdet 0 < Mg/(Mg+V) < 0.2. Fördelen med ett ökat bandgap är attabsorptionen förskjuts till kortare våglängder och ger ökad transmittans i detsynliga våglängdsområdet. Optiska konstanter för odopade och Mg-dopadeVO2 filmer användes i optiska simuleringar och gav teoretiska optiskaegenskaper för dessa Mg-dopade filmer och kompositer innehållandenanopartiklar av VO2. Resultaten visade att Mg-dopade filmer, specielltvid låga dopningskoncentrationer 0 < Mg/(Mg+V) < 0.06, hade mycketbättre egenskaper än odopade filmer. Resultaten visade även att de optiskaegenskaperna för Mg-dopade filmer kan förbättras ytterligare genom attbelägga dem med ett antireflekterande skikt med brytningsindex n =1.5. Däremot visade beräkningarna att kompositer med odopade VO2nanopartiklar hade bättre optiska egenskaper än då nanopartiklarna varMg-dopade, vilket är illustrerat i Fig. 8.8. Experimentella studier av tillväxtenhos nanostavar av VO2 utfördes också, vilket presenteras i kapitel 9. Studienvisade att substrat, mängd av deponerat material, substrattemperatur viddeponering och placering av ventiler för syrgas var faktorer som påverkadestrukturen på dessa nanostavar. Fortsatta studier inom detta område är ett

fortgående samarbete med Kyoko Namura och Prof. Suzuki vid Kyotouniversitet.

Inför framtiden är fortsatt syntetisering och karaktärisering av materialav stor vikt för kommande förbättringar av energieffektivisering. Specielltintressanta förbättringar kan åstadkommas om VO2 nanokompositerkombineras med andra funktionella material, såsom ITO, elektrokroma tunnafilmer, solceller och andra material där solenergi kan utnyttjas [155]. Detskulle t.ex. vara mycket lovande om framtida energieffektiva byggnaderhade kombinationer av elektrokroma ”smarta fönster” och lågemitterandeeller solkontroll skikt [7], fotokatalytisk inomhusfärg för luftrening[156], solfångare [157,158], beläggningar med kylfunktion [159] och förkondensation av vatten [160], samt solceller.

Vidare vill jag gärna tillåta mig att ”drömma mig bort lite” och fundera påenergi- och miljöfrågor i ett större sammanhang och på möjliga framtidascenarier. Nanomaterial är - tack vare deras småskalighet, stora andelyta, utmärkta kemiska reaktionsförmåga och mekaniska tålighet [162] -mycket användbara inom många applikationer för energieffektivisering[11,156,161], t.ex. i belysning [163-165], sensorer [166-170], solceller[166,171-173], elektrokroma ”smarta” fönster [174,175], solfångare[157], katalys [176,177], fotokatalys [178-179], batterier [180-181],målinriktad läkemedelsleverans [182,183], cancerbehandling [184,185] ochmekaniska/strukturella applikationer [186-188], o.s.v. I framtiden är dettroligt att faktisk energieffektivisering kommer att innebära att vi måsteefterapa naturens förmåga att ”skapa struktur från oordning”, och på så sättförsäkra oss om att utformning, syntes, process, slutprodukt och återvinningblir så optimerade som möjligt och ger en minimal mängd restprodukter.Materialforskare borde i framtiden frigöra kraften i naturens livsväv, d.v.s.mikroorganismer som bakterier och svampar, för syntetisering av material,speciellt nanomaterial. Trots att arbetet inom detta område bara har börjat,har stora framsteg redan gjorts inom syntetisering av metallnanopartiklar[189-193], metalloxider [194], sulfider [190,195,196], SiOx [196],magnetiska järnbaserade material [197], ferroelektriska material [198], osv.Dessutom har alger [199] och virus [200] visats kunna fungera som batterier.Låt oss även påminna oss om det ypperliga biogena materialet silke, somnyligen visats vara mycket användbart som komposterbara byggstenar förböjbara elektroniska skärmar, cellreparationer, implanterbara optiska systemoch målinriktad läkemedelsleverans, o.s.v. [201]. På en makroskopisk skalaborde människor i framtiden ha ett mer biomimetiskt förhållningssätt [202]och utnyttja ”cradle-to-cradle” utformingar [203] och på så sätt försäkrasig om att energi och livscykler ger ett självfungerande ekosystem utanrestprodukter, på samma sätt som naturens egen livscykel. Det är så jag villoch hoppas att framtiden ska bli.

122

Acknowledgements

First and foremost I would like to thank my supervisor, Professor Claes-Göran Granqvist, the head of the Division of Solid State Physics, for of-fering me the opportunity to join this prestigious group. Thank you for beinga visionary leader, opening my eyes with your innovative ideas, your guid-ance, readiness to help and for being so kind as to allow me to pursue my ownideas which can sometimes be a bit crazy. Your trust has empowered me andhas given me a lot of confidence. I am very grateful for the "status of a panda"that you allowed me to have during the last month of writing.

I would also like to thank Professor Gunnar Niklasson, my co-supervisor,for your encyclopedic knowledge, your devotion to quality, your patience andfor giving detailed answers for all my academic questions, big or small, andfor encouraging me to dig deeper into the theoretical side of my research.I greatly appreciate your purity and peace of mind, and your rigor when itcomes to scientific pursuit as well as to your passion for teaching.

During these years, I am very fortunate to have collaborated with brilliantscientists: Kyoko Namura and Professor Motofumi Suzuki from the De-partment of Micro Engineering are gratefully acknowledged for collaborationwork done on sputter-deposited VO2 nanorods. Many thanks to ProfessorsDaniel Primetshofer and Göran Possnert from the Department of Physicsand Astronomy, Uppsala University, Professor Anders Hallén from KTH-ICT, Royal Institute of Technology, Stockholm and Per Petersson from theSchool of Electrical Engineering, Fusion Plasma Physics, KTH Royal Insti-tute of Technology, Stockholm for years of kind help with RBS measurementsand analysis of my Mg-doped VO2 samples. Nuru R. Mlyuka from the De-partment of Physics, University of Dar es Salaam, Tanzania is thanked for of-fering helpful tips about the deposition of VO2, as well as providing samplesand data for work done on band gaps of Mg-doped VO2. Shuanglin Hu andProfessor Kersti Hermansson from the Department of Chemistry, UppsalaUniversity, Professors Rajeev Ahuja and Ralph Scheicher from Departmentof Physics and Astronomy, Uppsala University are thanked for the DFT hybridfunctional calculations and brilliant input in the collaboration work on Mg-doped VO2. Ilknur Pehlivan from the Division of Solid State Physics in theDepartment of Engineering Sciences at Uppsala University, Professor DeliaMilliron and Evan Runnerstrom in the Lawrence Berkley National Laboro-tary are thanked for collaboration work done on ITO-polymer nanocompositematerials. K. Laaksonen, T. Ala-Nissila and R. M. Nieminen from the De-

123

partment of Applied Physics and COMP CoE, Aalto University School ofScience, Finland and S. R. Puisto from MatOx Oy, Erottajankatu Helsinki,Finland are thanked for collaboration work done on a comparison betweeneffective medium theory and four-flux theory.

I would like to express my gratitude for all the expertise Shuxi Zhao haveoffered on the sputter unit, optical measurements and optical analysis. Yourold construction of a toroidal gas inlet played an important role in my Paper IV,and I am looking forward to seeing what the "showerhead" gas inlet is capableof. Your kind personality and care for others warms my heart, and I alwaysenjoy your company. Thanks to Andreas Mattsson for your selfless help onmany practical matters regarding equipments and so many other things (rice-transport etc). You keep absolutely 5-star (5 out of 5) standard in everythingyou do apart from your holidays, -since you have chosen not to rest but to"stand by” at home as our "helpdesk hotline", most of the time.

Thanks to Professor Arne Roos, for being a great teacher, sharing your ex-cellent knowledge on optical setups, measurements and analysis. It is alwaysdelightful to talk to you, and your detailed explanation and demonstrationhave enabled me to improve the accuracy of my measurements and helpedme to establish a solid foundation of knowledge in my head. Many thanks toBobbie Roos for your kind and timely help with the English language andlovely British accent. Thanks to Professor Emeritus Carl-Gustaf Ribbing,Professor Mats Boman, Professor Emeritus Sten-Erik Lindquist and Profes-sor Lars Österlund for readily sharing your knowledge, opinions, interestingstories and especially thanks to Professor C-G. R. for delicious apple-relatedproducts and Professor Matts B. for chocolates, -they are very nurturing tothe brain. I am grateful to Professor Laszlo Kish for warm encouragementand fast-paced conversation full of wisdom and gems of physics. Thanks toBengt Göteson for constructing a generally good infrastructure and warm at-mosphere in the whole division, taking care of equipment, coffee, fruit andcapturing lovely moments (including my portrait photo on the back cover ofthis thesis) with your beautiful photographic skills and sharing your nice pho-tos. Many thanks to Maria Skoglund, Ingrid Ringård, Sebastian Alonso,Inger Ekberg, Maria Melin, Anja Hohmann and Ylva Johansson for keep-ing many of the practical matters moving smoothly and in good order. Thanksto Enrique Carrasco and Jonatan Bagge for help with computers and soft-ware installations.

I am grateful to be in the same group with so many fun, helpful and inspir-ing colleagues. Pia Lansåker, Malin Johansson, José Montero Amenedo,Yuxia Ji, Ilknur Pehlivan and Sara Green, you are such great "multifunc-tional" peers, giving me so much from serious consultancy to great fun laugh-ing loud together! Pia L. is gratefully acknowledged for kindly providing ex-cellent Swedish summaries and help with RBS measurements. Annica Nils-son, thank you so much for being helpful and an excellent role model, not tomention your other qualities as the one who survived cycling from Uppsala to

124

Istanbul (yes, that one in Turkey!) and a talented silver smith. Anna-MalouPetersson, my cool competitive athletic mountain/ice-climbing adrenaline-junkie roommate/friend, your many adventure stories have enriched my lifeand all the best to your new conquests. Delphine Lebrun is appreciated forstrong ideals, organizing skills, quality fika cakes and topics, boundless artis-tic imagination and the warm bonding that you have created in the department.Thanks to Miguel Arvizu for adding depth and sophistication to lunch tablediscussions. Professor Ewa Wäckelgård is appeciated for being cool, grace-ful, clear-minded and an admirable female role model in academia. RuitaoWen and Yajun Wei are thanked for being the information hub and updat-ing me about many events. Carlos Triana, your hard working spirit remindsme how diligent and passionate one can be toward physics. Magnus Wik-berg is appreciated for funny and diverse conversation, sharing informationand facts about the Netherlands and care for environmental issues. Christo-pher Maghanga, Maxwell Mageto, Stella Kioko, thank you for sharing re-ally interesting stories, jokes and giving me useful tips about how to avoidleopards. Bozhidar Stefanov, Mikael Andersson, Anil Kumar Puri, Ma-lin J., Matthias Hudl, Delphine L., Esat Pehlivan, Fang Mao, RolandMathieu, Zareh Topalian, Iryna Valyukh, Wei Xia, Ruitao W., Umut Cin-demir, David Lingfors, Joakim Munkhammar, Joakim Widén, Christo-pher Lundgren, Tarja Volotinen and Magnus Åberg are acknowledgedfor offering generous or even life-saving help. Zareh Topalian and SerkanAkansel are appreciated for contagious smily faces. Bozhidar S, thank youfor sharing your insights about XRD, molecular cooking and so many inter-esting and diverse topics in chemistry. Anil K. P., Mikael A., Roland M.,Umut C. and Serkan A. thank you for all the teasing, funny stories and wittyconverstations; the last month of writing was unexpectedly fun. Also appre-ciated was the rather early "magnetism 7.30 am. fika" with a historical touchfrom Professor Per Nordblad, and a hint of Russia from Professor SergeyIvanov. Mattias Strömberg is thanked for his relaxing appearance and shar-ing really strong coffee. Rebecca Bejhed and Sofia Kontos are appreciatedfor adding glitter and color to the department. Professors Peter Svedlindhand Klas Gunnarsson are thanked for being good-humored. I am grateful formeeting so many other nice people who have visited this department, for shortor long periods of time. Special thanks to José’s parents, José Montero andM. Carmen Amenedo for the beautiful 100+ years old artisan nail, which Ihave marveled, treasured from the first day, and will feel very regretful to partwith. (Maybe I will keep it as a "timepiece" for decoration instead of using itto nail the thesis)

I would like to express my gratitude to all the people I have met in Kyoto,Japan, you are such a diverse, sophisticated and enriching crowd. Thanks toKyoko Namura, your diligence, creative ways and delightful chats inspiredme and revived many of the beautiful memories of my old days spent in Kyoto.Professor Motofumi Suzuki, I am grateful for your teaching (which some-

125

times resembles a fierce scolding but neverthless is delightful in hindsight)about many practical experimental skills, and I will always remember yourrule number one, to "think and use common sense instead of blindly executingmanuals" and many many other principles that are nonexistent in textbooks.Professor Kimura Kenji is thanked for being very kind, patient, overwhelm-ingly brilliant and offering instant and mathematically non-trivial answers tomy questions regarding theoretical matters. I am grateful for "Kimura labseminar" and its organizers Profs. Kimura sensei, Suzuki sensei and KaoruNakajima sensei, for showing me the rigorous beauty of physics and thatsome knowledge and skills can only be acquired by digging deep the hardway. Hamachi Kenji is gratefully acknowledged for being helpful, a greatrole model and a continuous source of inspiration. My excellent high-schoolclassmates who happened to also be my university mates in Japan, I didn’tknow loneliness because of the supports and good time spent with you guys!Thanks to Xiaoxing Zhou, Xue Bai and Ayumi Takeda for being one of akind as well as Shuo Zhang, Qingwei Sun, Mengnan An, Chen Wang, YokoMinamide and so many others for lasting friendship. A lot of thanks to theBetter Home Association and the JASSO Honors Scholarship for providingthe financial support during my undergraduate and graduate studies in KyotoUniversity.

Many thanks to my Chinese, Swedish, Canadian, Italian, Dutch and Germanfriends and other friends for nice times spent eating, playing, chatting, and forthe peer-pressure of good cooking. Special thanks to Aron Opheij and hiswarm family, the lovely fairytale-like Laneryd family, Xia Yang, ShuanglinHu, Xin Hu and Jing Ni, Chongyang Sun and Jon Äslund, Fang Mao andTao Cui, Wei Xia and Miao Wu, Lois Tang and Luce and Stefano Rubino,Ralph Scheicher, Zhao Qian, Zhigang Wu, Jiefang Zhu, Shi Cheng, Yan-ling Cai and Di Wu, Yuan Tian, Jia Liu, Bing Cai, Bing Sun, Taha Ahmed,He Xin, Jing Cen and many many others.

Thanks to my family; without your support none of this would have beenpossible for me.

Thanks to the Swedish Research Council for financially supporting myPh.D project and Anna Maria Lundin’s scholarship for financing my confer-ence travels and summer school.

126

Bibliography

[1] J. V. C. Nye, “Standards of living and modern economic growth,” in The Con-cise Encyclopedia of Economics (D. R. Henderson, ed.), Library of Economicsand Liberty [online], 2008.

[2] I. Beeton, Mrs. Beeton’s household management : a guide to cookery in allbranches : daily duties, menu making, mistress and servant, home doctor, host-ess and guest, sick nursing, marketing, the nursery, trussing and carving, homelawyer. London: Ward, Lock, 1907.

[3] UNEP, Buildings and Climate Change: Status, Challenges, and Opportunities.Paris: United Nations Environmental Programme, 2007.

[4] B. Richter, D. Goldston, G. Crabtree, L. Glicksman, D. Goldstein, D. Greene,D. Kammen, M. Levine, M. Lubell, M. Savitz, D. Sperling, F. Schlachter,J. Scofield, and J. Dawson, “How america can look within to achieve energysecurity and reduce global warming,” Rev. Mod. Phys., vol. 80, pp. S1–S109,Dec 2008.

[5] C. Lampert and C. Granqvist, eds., Large-area Chromogenics: Materials andDevices for Transmittance Control, vol. IS4. Bellingham, USA: The Interna-tional Society for Optical Engineering, 1990.

[6] S.-Y. Li, G. A. Niklasson, and C. G. Granqvist, “Thermochromism of vo2nanoparticles: Calculated optical properties and applications to energy efficientwindows,” MRS Proceedings, vol. 1315, pp. mrsf10–1315–mm07–07, 0 2011.

[7] C. Granqvist, S. Green, G. Niklasson, N. Mlyuka, S. von Kræmer, andP. Georén, “Advances in chromogenic materials and devices,” Thin Solid Films,vol. 518, no. 11, pp. 3046 – 3053, 2010. Transparent Oxides for Electronics andOptics (TOEO-6).

[8] C. B. Greenberg, “Undoped and doped VO2 films grown from VO(OC3H7)3,”Thin Solid Films, vol. 110, no. 1, pp. 73 – 82, 1983.

[9] f S. Babulanam, T. Eriksson, G. Niklasson, and C. Granqvist, “ThermochromicVO2 films for energy-efficient windows,” Solar Energy Materials, vol. 16,no. 5, pp. 347–363, 1987.

[10] I. P. Parkin and T. D. Manning, “Intelligent thermochromic windows,” Journalof Chemical Education, vol. 83, no. 3, p. 393, 2006.

127

[11] G. B. Smith and C. G. Granqvist, Green Nanotechnology: Solutions for Sus-tainability and Energy in the Built Environment. CRC Press, Boca Raton, FL,2010.

[12] M. Saeli, C. Piccirillo, I. P. Parkin, R. Binions, and I. Ridley, “Energy mod-elling studies of thermochromic glazing,” Energy and Buildings, vol. 42,no. 10, pp. 1666 – 1673, 2010.

[13] G. Wyszecki and W. Stiles, Color Science: Concepts and Methods, Quantita-tive Data and Formulae. New York, NY, USA: Wiley, 2nd ed., 2000.

[14] Annal Book of ASTM Standards, vol. 14.04, ch. ASTM G173-03: Standard Ta-bles of Reference, Solar Spectral Irradiances: Direct Normal and Hemispher-ical on a 37 Tilted Surface. Philadelphia, PA, USA: American Society forTesting and Materials, 2008.

[15] C. G. Granqvist, “Transparent conductors as solar energy materials: Apanoramic review,” Solar Energy Materials and Solar Cells, vol. 91, no. 17,pp. 1529–1598, 2007.

[16] I. Hamberg and C. G. Granqvist, “Evaporated Sn-doped In2O3 films: Basicoptical properties and applications to energy-efficient windows,” Journal ofApplied Physics, vol. 60, no. 11, pp. R123–R160, 1986.

[17] C. G. Granqvist, Handbook of Inorganic Electrochromic Materials. ElsevierScience, 1995.

[18] D. T. Gillaspie, R. C. Tenent, and A. C. Dillon, “Metal-oxide films for elec-trochromic applications: present technology and future directions,” J. Mater.Chem., vol. 20, pp. 9585–9592, 2010.

[19] F. J. Morin, “Oxides which show a metal-to-insulator transition at the neeltemperature,” Physical Review Letters, vol. 3, pp. 34–36, Jul 1959.

[20] M. M. Qazilbash, M. Brehm, B.-G. Chae, P.-C. Ho, G. O. Andreev, B.-J. Kim,S. J. Yun, A. V. Balatsky, M. B. Maple, F. Keilmann, H.-T. Kim, and D. N.Basov, “Mott transition in VO2 revealed by infrared spectroscopy and nano-imaging,” Science, vol. 318, no. 5857, pp. 1750–1753, 2007.

[21] M. M. Qazilbash, M. Brehm, G. O. Andreev, A. Frenzel, P.-C. Ho, B.-G. Chae,B.-J. Kim, S. J. Yun, H.-T. Kim, A. V. Balatsky, O. G. Shpyrko, M. B. Maple,F. Keilmann, and D. N. Basov, “Infrared spectroscopy and nano-imaging of theinsulator-to-metal transition in vanadium dioxide,” Physical Review B, vol. 79,p. 075107, Feb 2009.

[22] J. B. Goodenough, “The two components of the crystallographic transition inVO2,” Journal of Solid State Chemistry, vol. 3, no. 4, pp. 490 – 500, 1971.

[23] S. Shin, S. Suga, M. Taniguchi, M. Fujisawa, H. Kanzaki, A. Fujimori, H. Dai-mon, Y. Ueda, K. Kosuge, and S. Kachi, “Vacuum-ultraviolet reflectance andphotoemission study of the metal-insulator phase transitions in VO2, V6O13,and V2O3,” Physical Review B, vol. 41, pp. 4993–5009, Mar 1990.

128

[24] T. Christmann, B. Felde, W. Niessner, D. Schalch, and A. Scharmann, “Ther-mochromic VO2 thin films studied by photoelectron spectroscopy,” Thin SolidFilms, vol. 287, no. 1-2, pp. 134 – 138, 1996.

[25] R. Eguchi, M. Taguchi, M. Matsunami, K. Horiba, K. Yamamoto, Y. Ishida,A. Chainani, Y. Takata, M. Yabashi, D. Miwa, Y. Nishino, K. Tamasaku,T. Ishikawa, Y. Senba, H. Ohashi, Y. Muraoka, Z. Hiroi, and S. Shin, “Photoe-mission evidence for a Mott-Hubbard metal-insulator transition in vo2,” Phys-ical Review B, vol. 78, p. 075115, Aug 2008.

[26] R. Eguchi, M. Taguchi, M. Matsunami, K. Horiba, K. Yamamoto, A. Chainani,Y. Takata, M. Yabashi, D. Miwa, Y. Nishino, K. Tamasaku, T. Ishikawa,Y. Senba, H. Ohashi, I. Inoue, Y. Muraoka, Z. Hiroi, and S. Shin, “Electronicstructure of configuration vanadium oxides studied by soft X-ray and hard x-ray photoemission spectroscopy,” Journal of Electron Spectroscopy and Re-lated Phenomena, vol. 156-158, pp. 421 – 425, 2007.

[27] E. Z. Kurmaev, V. M. Cherkashenko, Y. M. Yarmoshenko, S. Bartkowski, A. V.Postnikov, M. Neumann, L. C. Duda, J. H. Guo, J. Nordgren, V. A. Perelyaev,and W. Reichelt, “Electronic structure of vo2 studied by x-ray photoelectronand X-ray emission spectroscopies,” J. Phys.:Cond. Matter, vol. 10, p. 4081,1998.

[28] H. Abe, M. Terauchi, M. Tanaka, S. Shin, and Y. Ueda, “Electron energy-lossspectroscopy study of the metal-insulator transition in VO2,” Japanese Journalof Applied Physics, vol. 36, pp. pp165–169, 1997.

[29] M. M. Qazilbash, A. A. Schafgans, K. S. Burch, S. J. Yun, B. G. Chae, B. J.Kim, H. T. Kim, and D. N. Basov, “Electrodynamics of the vanadium oxidesVO2 and V2O3,” Physical Review B, vol. 77, p. 115121, Mar 2008.

[30] L. Zhen-Fei, W. Zhi-Ming, X. Xiang-Dong, W. Tao, and J. Ya-Dong, “Studyof nanocrystalline VO2 thin films prepared by magnetron sputtering and post-oxidation,” Chinese Physics B, vol. 19, no. 10, p. 106103, 2010.

[31] W. W. Li, J. J. Zhu, X. F. Xu, K. Jiang, Z. G. Hu, M. Zhu, and J. H.Chu, “Ultraviolet-infrared dielectric functions and electronic band structures ofmonoclinic VO2 nanocrystalline film: Temperature-dependent spectral trans-mittance,” Journal of Applied Physics, vol. 110, no. 1, p. 013504, 2011.

[32] R. Mossanek and M. Abbate, “Evolution of the d band across the metal–insulator transition in VO2,” Solid State Communications, vol. 135, no. 3,pp. 189 – 192, 2005.

[33] G.-H. Liu, X.-Y. Deng, and R. Wen, “Electronic and optical properties of mon-oclinic and rutile vanadium dioxide,” Journal of Materials Science, vol. 45,pp. 3270–3275, 2010.

[34] N. Mlyuka, G. Niklasson, and C. Granqvist, “Thermochromic multilayer filmsof VO2 and TiO2 with enhanced transmittance,” Solar Energy Materials andSolar Cells, vol. 93, no. 9, pp. 1685 – 1687, 2009.

129

[35] M. Sobhan, R. Kivaisi, B. Stjerna, and C. Granqvist, “Thermochromism ofsputter deposited WxV1−xO2 films,” Solar Energy Materials and Solar Cells,vol. 44, no. 4, pp. 451 – 455, 1996.

[36] N. R. Mlyuka, G. A. Niklasson, and C. G. Granqvist, “Mg doping ofthermochromic VO2 films enhances the optical transmittance and decreasesthe metal-insulator transition temperature,” Applied Physics Letters, vol. 95,no. 17, p. 171909, 2009.

[37] T. D. Manning and I. P. Parkin, “Atmospheric pressure chemical vapour depo-sition of tungsten doped vanadium(IV) oxide from VOCl3, water and WCl6,”J. Mater. Chem., vol. 14, pp. 2554–2559, 2004.

[38] T. D. Manning, I. P. Parkin, M. E. Pemble, D. Sheel, and D. Vernardou, “In-telligent window coatings: Atmospheric pressure chemical vapor depositionof tungsten-doped vanadium dioxide,” Chemistry of Materials, vol. 16, no. 4,pp. 744–749, 2004.

[39] B. Chamberland, “The hydrothermal synthesis of V2O4−xFx derivatives,” Ma-terials Research Bulletin, vol. 6, no. 6, pp. 425 – 432, 1971.

[40] M. L. F. Bayard, T. G. Reynolds, M. Vlasse, H. L. McKinzie, R. J. Arnott, andA. Wold, “Preparation and properties of the oxyfluoride systems V2O5−xFx andVO2−xFx,” Journal of Solid State Chemistry, vol. 3, no. 4, pp. 484 – 489, 1971.

[41] K. A. Khan and C. G. Granqvist, “Thermochromic sputter-deposited vanadiumoxyfluoride coatings with low luminous absorptance,” Applied Physics Letters,vol. 55, no. 1, pp. 4–6, 1989.

[42] J. Zhou, Y. Gao, X. Liu, Z. Chen, L. Dai, C. Cao, H. Luo, M. Kanahira,C. Sun, and L. Yan, “Mg-doped VO2 nanoparticles: hydrothermal synthesis,enhanced visible transmittance and decreased metal-insulator transition tem-perature,” Phys. Chem. Chem. Phys., vol. 15, pp. 7505–7511, 2013.

[43] S.-Y. Li, N. R. Mlyuka, D. Primetzhofer, A. Hallén, G. Possnert, G. A. Niklas-son, and C. G. Granqvist, “Bandgap widening in thermochromic Mg-dopedVO2 thin films: Quantitative data based on optical absorption,” Applied PhysicsLetters, vol. 103, no. 16, pp. 161907, 2013.

[44] S. Hu, S.-Y. Li, R. Ahuja, C. G. Granqvist, K. Hermansson, G. A. Niklas-son, and R. H. Scheicher, “Optical properties of Mg-doped VO2: Absorptionmeasurements and hybrid functional calculations,” Applied Physics Letters,vol. 101, no. 20, p. 201902, 2012.

[45] Z. Chen, Y. Gao, L. Kang, J. Du, Z. Zhang, H. Luo, H. Miao, and G. Tan, “VO2-based double-layered films for smart windows: Optical design, all-solutionpreparation and improved properties,” Solar Energy Materials and Solar Cells,vol. 95, no. 9, pp. 2677 – 2684, 2011.

[46] P. Jin, G. Xu, M. Tazawa, and K. Yoshimura, “Design, formation and char-acterization of a novel multifunctional window with VO2 and TiO2,” AppliedPhysics A: Materials Science & Processing, vol. 77, pp. 455–459, 2003.

130

[47] H. Kakiuchida, P. Jin, and M. Tazawa, “Control of thermochromic spectrum invanadium dioxide by amorphous silicon suboxide layer,” Solar Energy Mate-rials and Solar Cells, vol. 92, no. 10, pp. 1279 – 1284, 2008.

[48] N. R. Mlyuka, G. A. Niklasson, and C. G. Granqvist, “Thermochromic VO2-based multilayer films with enhanced luminous transmittance and solar modu-lation,” Physica Status Solidi (a), vol. 206, no. 9, pp. 2155–2160, 2009.

[49] A. Taylor, I. Parkin, N. Noor, C. Tummeltshammer, M. S. Brown, and I. Pa-pakonstantinou, “A bioinspired solution for spectrally selective thermochromicVO2 coated intelligent glazing,” Opt. Express, vol. 21, pp. A750–A764, Sep2013.

[50] S.-Y. Li, G. A. Niklasson, and C. G. Granqvist, “Nanothermochromics: Calcu-lations for VO2 nanoparticles in dielectric hosts show much improved luminoustransmittance and solar energy transmittance modulation,” Journal of AppliedPhysics, vol. 108, no. 6, p. 063525, 2010.

[51] S.-Y. Li, G. A. Niklasson, and C. G. Granqvist, “Nanothermochromics withVO2-based core-shell structures: Calculated luminous and solar optical prop-erties,” Journal of Applied Physics, vol. 109, no. 11, p. 113515, 2011.

[52] S. A. Maier, Plasmonics: Fundamentals and Applications. Springer, 2007.

[53] S. V. Gaponenko, Introduction to Nanophotonics, ch. 5. Semiconductornanocrystals (quantum dots), p. 110. Cambridge University Press, 2010.

[54] A. Sidorov, O. Vinogradova, I. Obyknovennaya, and T. Khrushchova, “Syn-thesis and optical properties of vanadium dioxide nanoparticles in nanoporousglasses,” Technical Physics Letters, vol. 33, no. 7, pp. 581–582, 2007.

[55] O. Vinogradova, I. Obyknovennaya, A. Sidorov, V. Klimov, E. Shadrin,S. Khanin, and T. Khrushcheva, “Synthesis and the properties of vanadiumdioxide nanocrystals in porous silicate glasses,” Physics of the Solid State,vol. 50, no. 4, pp. 768–774, 2008.

[56] R. Lopez, T. E. Haynes, L. A. Boatner, L. C. Feldman, and J. R. F. Haglund,“Temperature-controlled surface plasmon resonance in VO2 nanorods,” OpticsLetters, vol. 27, pp. 1327–1329, Aug 2002.

[57] R. Lopez, L. A. Boatner, T. E. Haynes, L. C. Feldman, and R. F. Haglund, “Syn-thesis and characterization of size-controlled vanadium dioxide nanocrystalsin a fused silica matrix,” Journal of Applied Physics, vol. 92, no. 7, pp. 4031–4036, 2002.

[58] S. Mukherjee and A. K. Pal, “Size-dependent magnetic properties of VO2nanocrystals dispersed in a silica matrix,” Journal of Physics: Condensed Mat-ter, vol. 20, no. 25, p. 255202, 2008.

131

[59] Y. Sun, S. Jiang, W. Bi, R. Long, X. Tan, C. Wu, S. Wei, and Y. Xie, “Newaspects of size-dependent metal-insulator transition in synthetic single-domainmonoclinic vanadium dioxide nanocrystals,” Nanoscale, vol. 3, pp. 4394–4401,2011.

[60] S. Ji, F. Zhang, and P. Jin, “Preparation of high performance pure single phaseVO2 nanopowder by hydrothermally reducing the V2O5 gel,” Solar EnergyMaterials and Solar Cells, vol. 95, no. 12, pp. 3520 – 3526, 2011.

[61] C. ling Xu, L. Ma, X. Liu, W. yuan Qiu, and Z. xing Su, “A novel reduction–hydrolysis method of preparing VO2 nanopowders,” Materials Research Bul-letin, vol. 39, no. 7–8, pp. 881 – 886, 2004.

[62] X. Liu, C. Huang, S. Yi, G. Xie, H. Li, and Y. Luo, “A new solvothermalmethod of preparing VO2 nanosheets and petaloid clusters,” Solid State Com-munications, vol. 144, no. 5–6, pp. 259 – 263, 2007.

[63] D. Munoz-Rojas and E. Baudrin, “Synthesis and electroactivity of hydrated andmonoclinic rutile-type nanosized VO2,” Solid State Ionics, vol. 178, no. 21-22,pp. 1268 – 1273, 2007.

[64] J. Shi, S. Zhou, B. You, and L. Wu, “Preparation and thermochromic propertyof tungsten-doped vanadium dioxide particles,” Solar Energy Materials andSolar Cells, vol. 91, no. 19, pp. 1856 – 1862, 2007.

[65] J. M. Booth and P. S. Casey, “Production of VO2 M1 and M2 nanoparticles andcomposites and the influence of the substrate on the structural phase transi-tion,” ACS Applied Materials & Interfaces, vol. 1, no. 9, pp. 1899–1905, 2009.PMID: 20355812.

[66] J. Zhang, Eerdemutu, C. Yang, J. Feng, X. Di, Z. Liu, and G. Xu, “Size- andshape-controlled synthesis of monodisperse vanadium dioxide nanocrystals,”Journal of Nanoscience and Nanotechnology, vol. 10, no. 3, pp. 2092–2098,2010.

[67] F. Wang, Y. Liu, and C. yan Liu, “Molten salt synthesis and localized surfaceplasmon resonance study of vanadium dioxide nanopowders,” Journal of SolidState Chemistry, vol. 182, no. 12, pp. 3249 – 3253, 2009.

[68] Z. Peng, W. Jiang, and H. Liu, “Synthesis and electrical properties of tungsten-doped vanadium dioxide nanopowders by thermolysis,” The Journal of Physi-cal Chemistry C, vol. 111, no. 3, pp. 1119–1122, 2007.

[69] C. Zheng, X. Zhang, J. Zhang, and K. Liao, “Preparation and characteriza-tion of VO2 nanopowders,” Journal of Solid State Chemistry, vol. 156, no. 2,pp. 274 – 280, 2001.

[70] C. Wu, J. Dai, X. Zhang, J. Yang, F. Qi, C. Gao, and Y. Xie, “Directconfined-space combustion forming monoclinic vanadium dioxides,” Ange-wandte Chemie International Edition, vol. 49, no. 1, pp. 134–137, 2010.

132

[71] R. Lopez, L. A. Boatner, T. E. Haynes, R. F. Haglund, and L. C. Feldman,“Switchable reflectivity on silicon from a composite VO2-SiO2 protectinglayer,” Applied Physics Letters, vol. 85, no. 8, pp. 1410–1412, 2004.

[72] R. Lopez, R. F. Haglund, L. C. Feldman, L. A. Boatner, and T. E. Haynes,“Optical nonlinearities in VO2 nanoparticles and thin films,” Applied PhysicsLetters, vol. 85, no. 22, pp. 5191–5193, 2004.

[73] S. A. Pauli, R. Herger, P. R. Willmott, E. U. Donev, J. Y. Suh, and R. F.Haglund, “X-ray diffraction studies of the growth of vanadium dioxidenanoparticles,” Journal of Applied Physics, vol. 102, no. 7, pp. –, 2007.

[74] Y. Gao, C. Cao, L. Dai, H. Luo, M. Kanehira, Y. Ding, and Z. L. Wang, “Phaseand shape controlled VO2 nanostructures by antimony doping,” Energy Envi-ron. Sci., vol. 5, pp. 8708–8715, 2012.

[75] Z. Peng, Y. Wang, Y. Du, D. Lu, and D. Sun, “Phase transition and IR proper-ties of tungsten-doped vanadium dioxide nanopowders,” Journal of Alloys andCompounds, vol. 480, no. 2, pp. 537 – 540, 2009.

[76] Y. Chen, H. Xu, and H. X. Feng, “Preparation of w doped VO2 nanopowdersby coprecipitation,” Materials Research Innovations, vol. 12, no. 2, pp. 78–80,2008.

[77] S. Ji, Y. Zhao, F. Zhang, and P. Jin, “Direct formation of single crystal VO2(R) nanorods by one-step hydrothermal treatment,” Journal of Crystal Growth,vol. 312, no. 2, pp. 282 – 286, 2010.

[78] L. Whittaker, T.-L. Wu, A. Stabile, G. Sambandamurthy, and S. Banerjee,“Single-nanowire raman microprobe studies of doping-, temperature-, andvoltage-induced metal–insulator transitions of WxV1−xO2 nanowires,” ACSNano, vol. 5, no. 11, pp. 8861–8867, 2011.

[79] J. Liu, Q. Li, T. Wang, D. Yu, and Y. Li, “Metastable vanadium dioxidenanobelts: Hydrothermal synthesis, electrical transport, and magnetic proper-ties,” Angewandte Chemie International Edition, vol. 43, no. 38, pp. 5048–5052, 2004.

[80] M. Wei, Z.-M. Qi, M. Ichihara, M. Hirabayashi, I. Honma, and H. Zhou, “Syn-thesis of single-crystal vanadium dioxide nanosheets by the hydrothermal pro-cess,” Journal of Crystal Growth, vol. 296, no. 1, pp. 1 – 5, 2006.

[81] X.-P. Yan, J. Chen, J.-M. Liu, P.-P. Zhou, X. Liu, and Z.-X. Su, “Fabrication ofhollow VO2 microspheres by a facile template-free process,” Materials Letters,vol. 64, no. 3, pp. 278 – 280, 2010.

[82] O. Vinogradova, A. Sidorov, V. Klimov, E. Shadrin, A. Nashchekin, S. Khanin,and V. Lyubimov, “Specific features of the formation of vanadium oxidemicro- and nanocrystals during gas-phase synthesis,” Physics of the Solid State,vol. 50, no. 7, pp. 1227–1233, 2008.

133

[83] H. Suzuki, K. Yamaguchi, and H. Miyazaki, “Fabrication of thermochromiccomposite using monodispersed vo2 coated SiO2 nanoparticles prepared bymodified chemical solution deposition,” Composites Science and Technology,vol. 67, no. 16, pp. 3487 – 3490, 2007.

[84] H. MIYAZAKI, N. KUSUMOTO, S. SASAKI, N. SAKAMOTO,N. WAKIYA, and H. SUZUKI, “Thermochromic tungsten doped VO2-SiO2 nano-particle synthesized by chemical solution deposition technique,”Journal of the Ceramic Society of Japan, vol. 117, no. 1369, pp. 970–972,2009.

[85] I. Karakurt, J. Boneberg, P. Leiderer, R. Lopez, A. Halabica, and R. F. Haglund,“Transmission increase upon switching of VO2 thin films on microstructuredsurfaces,” Applied Physics Letters, vol. 91, no. 9, pp. 091907, 2007.

[86] F. Wang, Y. Liu, and C. yan Liu, “Hydrothermal synthesis of carbon/vanadiumdioxide core–shell microspheres with good cycling performance in both or-ganic and aqueous electrolytes,” Electrochimica Acta, vol. 55, no. 8, pp. 2662– 2666, 2010.

[87] Y. Zhang, M. Fan, W. Wu, L. Hu, J. Zhang, Y. Mao, C. Huang, and X. Liu, “Anovel route to fabricate belt-like VO2(M)@C core-shell structured compositeand its phase transition properties,” Materials Letters, vol. 71, pp. 127 – 130,2012.

[88] A. Llordes, A. T. Hammack, R. Buonsanti, R. Tangirala, S. Aloni, B. A. Helms,and D. J. Milliron, “Polyoxometalates and colloidal nanocrystals as build-ing blocks for metal oxide nanocomposite films,” J. Mater. Chem., vol. 21,pp. 11631–11638, 2011.

[89] Y. Lu, S. Zhou, G. Gu, and L. Wu, “Preparation of transparent, hard ther-mochromic polysiloxane/tungsten-doped vanadium dioxide nanocompositecoatings at ambient temperature,” Thin Solid Films, vol. 534, pp. 231 – 237,2013.

[90] J. Du, Y. Gao, H. Luo, Z. Zhang, L. Kang, and Z. Chen, “Formation and metal-to-insulator transition properties of VO2–ZrV2O7 composite films by polymer-assisted deposition,” Solar Energy Materials and Solar Cells, vol. 95, no. 7,pp. 1604 – 1609, 2011.

[91] S. Ding, Z. Liu, D. Li, W. Zhao, Y. Wang, D. Wan, and F. Huang, “Tun-able assembly of vanadium dioxide nanoparticles to create porous film forenergy-saving applications,” ACS Applied Materials & Interfaces, vol. 5, no. 5,pp. 1630–1635, 2013.

[92] H. Kim, Y. Kim, K. S. Kim, H. Y. Jeong, A.-R. Jang, S. H. Han, D. H. Yoon,K. S. Suh, H. S. Shin, T. Kim, and W. S. Yang, “Flexible thermochromic win-dow based on hybridized VO2/graphene,” ACS Nano, vol. 7, no. 7, pp. 5769–5776, 2013.

134

[93] Y. Gao, S. Wang, H. Luo, L. Dai, C. Cao, Y. Liu, Z. Chen, and M. Kane-hira, “Enhanced chemical stability of VO2 nanoparticles by the formation ofSiO2/VO2 core/shell structures and the application to transparent and flexibleVO2-based composite foils with excellent thermochromic properties for solarheat control,” Energy Environ. Sci., vol. 5, pp. 6104–6110, 2012.

[94] Y. Gao, S. Wang, L. Kang, Z. Chen, J. Du, X. Liu, H. Luo, and M. Kanehira,“VO2-Sb:SnO2 composite thermochromic smart glass foil,” Energy Environ.Sci., vol. 5, pp. 8234–8237, 2012.

[95] S. V. Gaponenko, Introduction to Nanophotonics. Cambridge University Press,2010.

[96] H. Hövel, S. Fritz, A. Hilger, U. Kreibig, and M. Vollmer, “Width of clusterplasmon resonances: Bulk dielectric functions and chemical interface damp-ing,” Phys. Rev. B, vol. 48, pp. 18178–18188, Dec 1993.

[97] C. G. Granqvist, ed., Materials Science for Solar Energy Conversion Systems,ch. 2. Pergamon Press, Oxford, UK, 1991.

[98] J. C. M. Garnett, “Colours in metal glasses and in metallic films,” PhilosophicalTransactions of the Royal Society of London. Series A, vol. 203, no. 359-371,pp. 385–420, 1904.

[99] J. C. M. Garnett, “Colours in metal glasses, in metallic films, and in metal-lic solutions. ii,” Philosophical Transactions of the Royal Society of London.Series A, vol. 205, no. 387-401, pp. 237–288, 1906.

[100] D. A. G. Bruggeman, “Berechnung verschiedener physikalischer Konstantenvon heterogenen Substanzen. I. Dielektrizitätskonstanten und Leitfähigkeitender Mischkörper aus isotropen Substanzen,” Annalen der Physik, vol. 416,no. 7, pp. 636–664, 1935.

[101] G. A. Niklasson, C. G. Granqvist, and O. Hunderi, “Effective medium modelsfor the optical properties of inhomogeneous materials,” Applied Optics, vol. 20,pp. 26–30, Jan 1981.

[102] C. G. Granqvist and O. Hunderi, “Optical properties of Ag-SiO2 cermet films:A comparison of effective-medium theories,” Physical Review B, vol. 18,pp. 2897–2906, Sep 1978.

[103] C. G. Granqvist and O. Hunderi, “Optical properties of ultrafine gold particles,”Physical Review B, vol. 16, pp. 3513–3534, Oct 1977.

[104] L. D. Landau and E. M. Lifshitz, Electrodynamics of Continuous Media. Perg-amon, Oxford, UK, 2nd ed., 1984.

[105] P. Clippe, R. Evrard, and A. A. Lucas, “Aggregation effect on the infraredabsorption spectrum of small ionic crystals,” Phys. Rev. B, vol. 14, pp. 1715–1721, Aug 1976.

135

[106] S. Berthier and J. Lafait, “Effective medium theory: Mathematical determina-tion of the physical solution for the dielectric constant,” Optics Communica-tions, vol. 33, no. 3, pp. 303 – 306, 1980.

[107] R. Jansson and H. Arwin, “Selection of the physically correct solution in the n-media Bruggeman effective medium approximation,” Optics Communications,vol. 106, no. 4-6, pp. 133 – 138, 1994.

[108] A. Gentle, A. I. Maaroof, and G. B. Smith, “Nanograin VO2 in the metal phase:a plasmonic system with falling dc resistivity as temperature rises,” Nanotech-nology, vol. 18, no. 2, p. 025202, 2007.

[109] I. B. Pehlivan, E. L. Runnerstrom, S.-Y. Li, G. A. Niklasson, D. J. Milliron, andC. G. Granqvist, “A polymer electrolyte with high luminous transmittance andlow solar throughput: Polyethyleneimine-lithium bis(trifluoromethylsulfonyl)imide with In2O3:Sn nanocrystals,” Applied Physics Letters, vol. 100, no. 24,p. 241902, 2012.

[110] M. Born and W. Wolf, Principles of Optics. Cambridge, UK: Cambridge Uni-versity Press, 7th ed., 1999.

[111] Z. Knittl, Optics of Thin Films (Pure & Applied Optics). John Wiley & SonsLtd, 1976.

[112] R. M. A. Azzam and N. M. Bashara, Ellipsometery and Polarized Light. North-Holland Physics Publishing, a division of Elsevier Science Publishers B.V.,1977.

[113] P. Pfrommer, K. Lomas, C. Seale, and C. Kupke, “The radiation transferthrough coated and tinted glazing,” Solar Energy, vol. 54, no. 5, pp. 287 –299, 1995.

[114] B. Harbecke, “Coherent and incoherent reflection and transmission of multi-layer structures,” Applied Physics B: Lasers and Optics, vol. 39, pp. 165–170,1986.

[115] K. Namura, M. Suzuki, K. Nakajima, and K. Kimura, “Heat-generating prop-erty of a local plasmon resonator under illumination,” Opt. Lett., vol. 36,pp. 3533–3535, Sep 2011.

[116] K. Kimura, S. Joumori, Y. Oota, K. Nakajima, and M. Suzuki, “High-resolutionRBS: a powerful tool for atomic level characterization,” Nuclear Instrumentsand Methods in Physics Research Section B: Beam Interactions with Materialsand Atoms, vol. 219–220, pp. 351 – 357, 2004. Proceedings of the SixteenthInternational Conference on Ion Beam Analysis

[117] M. Nastasi, J. W. Mayer, and J. K. Hirvonen, Ion-Solid Interactions: Funda-mentals and Applications. Cambridge Solid State Science Series, CambridgeUniversity Press, 1996.

136

[118] P. Sigmund, Stopping of Heavy Ions: A Theoretical Approach, vol. 204 ofSpringer Tracts in Modern Physics. Springer, 2004.

[119] P. Sigmund, Particle Penetration and Radiation Effects: General Aspects andStopping of Swift Point Charges, vol. 151 of Springer Series in Solid-StateSciences. Springer, 2006.

[120] R. Bird and J. Williams, eds., Ion Beams for Materials Analysis. AcademicPress, 1990.

[121] M. Mayer, “SIMNRA, a simulation program for the analysis of NRA, RBS andERDA,” AIP Conference Proceedings, vol. 475, no. 1, pp. 541–544, 1999.

[122] D. L. Bish and J. E. Post, Modern Powder Diffraction, Reviews in Mineralogy.Mineralogical Society of America, 1989.

[123] B. D. Cullity, Elements Of X-Ray Diffraction. Addison-Wesley PublishingCompany, Inc., 1956.

[124] R. Jenkins and R. Snyder, Introduction to X-Ray Powder Diffractometry,vol. 158 of Chemical Analysis: A Series of Monographs on Analytical Chem-istry and Its Applications. Wiley-Interscience, Jun 1996.

[125] Inorganic crystal structure database. http://icsd.fiz-karlsruhe.de/

[126] Crystallography open database. http://www.crystallography.net/

[127] H. M. Rietveld, “A profile refinement method for nuclear and magnetic struc-tures,” Journal of Applied Crystallography, vol. 2, pp. 65–71, Jun 1969.

[128] G. Nolze and W. Kraus, PowderCell 2.3 Program. Federal Institute for Mate-rials Research and Testing, Berlin, Germany, 2000.

[129] W. A. Dollase, “Correction of intensities for preferred orientation in powderdiffractometry: application of the March model,” Journal of Applied Crystal-lography, vol. 19, pp. 267–272, Aug 1986.

[130] P. Nostell, A. Roos, and D. Ronnow, “Single-beam integrating sphere spec-trophotometer for reflectance and transmittance measurements versus angle ofincidence in the solar wavelength range on diffuse and specular samples,” Re-view of Scientific Instruments, vol. 70, no. 5, pp. 2481–2494, 1999.

[131] S. Cunsolo, P. Dore, and C. P. Varsamis, “Refractive index of crystals fromtransmission and reflection measurements: MgO in the far-infrared region,”Appl. Opt., vol. 31, pp. 4554–4558, Aug 1992.

[132] R. C. McPhedran, L. C. Botten, D. R. McKenzie, and R. P. Netterfield, “Un-ambiguous determination of optical constants of absorbing films by reflectanceand transmittance measurements,” Appl. Opt., vol. 23, pp. 1197–1205, Apr1984.

137

[133] Light Scattering and Absorption in Pigmented Coatings. Uppsala University,1997.

[134] M. Theiss, Hard and Software for Optical Spectroscopy. Dr. Bernhard-Klein-Str. 110, 52078 Aachen, Germany, available from http://www.mtheiss.com.

[135] F. Wooten, Optical Properties of Solids. Academic, New York, 1972.

[136] Z. Luo, Z. Wu, X. Xu, T. Wang, and Y. Jiang, “Growth, electrical, and opticalproperties of nanocrystalline VO2 (011) thin films prepared by thermal oxida-tion of magnetron sputtered vanadium films,” Journal of Vacuum Science &Technology A: Vacuum, Surfaces, and Films, vol. 28, no. 4, pp. 595–599, 2010.

[137] H. W. Verleur, A. S. Barker, and C. N. Berglund, “Optical properties of VO2between 0.25 and 5 eV,” Phys. Rev., vol. 172, pp. 788–798, Aug 1968.

[138] M. M. Qazilbash, K. S. Burch, D. Whisler, D. Shrekenhamer, B. G. Chae, H. T.Kim, and D. N. Basov, “Correlated metallic state of vanadium dioxide,” Phys-ical Review B, vol. 74, p. 205118, Nov 2006.

[139] S. K. Panda, S. Chakrabarti, B. Satpati, P. V. Satyam, and S. Chaudhuri, “Opti-cal and microstructural characterization of CdS–ZnO nanocomposite thin filmsprepared by sol–gel technique,” Journal of Physics D: Applied Physics, vol. 37,no. 4, p. 628, 2004.

[140] M. Tazawa, P. Jin, and S. Tanemura, “Optical constants of V1−xWxO2 films,”Applied Optics, vol. 37, pp. 1858–1861, Apr 1998.

[141] H. Kakiuchida, P. Jin, and M. Tazawa, “Control of thermochromic spectrum invanadium dioxide by amorphous silicon suboxide layer,” Solar Energy Mate-rials and Solar Cells, vol. 92, no. 10, pp. 1279 – 1284, 2008.

[142] J. Kana Kana, J. Ndjaka, G. Vignaud, A. Gibaud, and M. Maaza, “Thermallytunable optical constants of vanadium dioxide thin films measured by spectro-scopic ellipsometry,” Optics Communications, vol. 284, no. 3, pp. 807 – 812,2011.

[143] S.-Y. Li, K. Namura, M. Suzuki, G. A. Niklasson, and C. G. Granqvist, “Ther-mochromic VO2 nanorods made by sputter deposition: Growth conditions andoptical modeling,” Journal of Applied Physics, vol. 114, no. 3, p. 033516, 2013.

[144] K. Okazaki, S. Sugai, Y. Muraoka, and Z. Hiroi, “Role of electron-electronand electron-phonon interaction effects in the optical conductivity of VO2,”Physical Review B, vol. 73, p. 165116, Apr 2006.

[145] P. B. Allen, R. M. Wentzcovitch, W. W. Schulz, and P. C. Canfield, “Resistivityof the high-temperature metallic phase of VO2,” Physical Review B, vol. 48,pp. 4359–4363, Aug 1993.

[146] H. S. Choi, J. S. Ahn, J. H. Jung, T. W. Noh, and D. H. Kim, “Mid-infraredproperties of a VO2 film near the metal-insulator transition,” Phys. Rev. B,vol. 54, pp. 4621–4628, Aug 1996.

138

[147] H. Bai, M. B. Cortie, A. I. Maaroof, A. Dowd, C. Kealley, and G. B. Smith,“The preparation of a plasmonically resonant VO2 thermochromic pigment,”Nanotechnology, vol. 20, no. 8, p. 085607, 2009.

[148] L. Kang, Y. Gao, Z. Chen, J. Du, Z. Zhang, and H. Luo, “Pt/VO2 double-layered films combining thermochromic properties with low emissivity,” SolarEnergy Materials and Solar Cells, vol. 94, no. 12, pp. 2078 – 2084, 2010.

[149] L. Kang, Y. Gao, H. Luo, J. Wang, B. Zhu, Z. Zhang, J. Du, M. Kanehira,and Y. Zhang, “Thermochromic properties and low emissivity of ZnO:Al/VO2double-layered films with a lowered phase transition temperature,” Solar En-ergy Materials and Solar Cells, vol. 95, no. 12, pp. 3189 – 3194, 2011.

[150] K. Hamachi, M. Suzuki, K. Nakajima, and K. Kimura, “Effect of the surfacenanomorphology on the growth of al whiskers formed by glancing angle depo-sition on a high temperature substrate,” MRS Proceedings, vol. 1059, 2007.

[151] M. Suzuki, K. Hamachi, K. Nagai, R. Kita, K. Nakajima, and K. Kimura,“Growth of nanowhiskers of Al, Ti, Cr, Mn, Fe, Co, Ni, Zn, Cu, Ag and Auby high-temperature glancing angle deposition,” ECS Transactions, vol. 33,no. 9, pp. 41–48, 2010.

[152] M. Suzuki, R. Kita, H. Hara, K. Hamachi, K. Nagai, K. Nakajima, andK. Kimura, “Growth of metal nanowhiskers on patterned substrate by hightemperature glancing angle deposition,” Journal of The Electrochemical Soci-ety, vol. 157, no. 2, pp. K34–K38, 2010.

[153] M. Suzuki, K. Hamachi, H. Hara, K. Nakajima, K. Kimura, C.-W. Hsu,and L.-J. Chou, “Vapor-liquid-solid growth of Ge nanowhiskers enhanced byhigh-temperature glancing angle deposition,” Applied Physics Letters, vol. 99,no. 22, p. 223107, 2011.

[154] M. Suzuki, K. Nagai, S. Kinoshita, K. Nakajima, K. Kimura, T. Okano, andK. Sasakawa, “Vapor phase growth of al whiskers induced by glancing an-gle deposition at high temperature,” Applied Physics Letters, vol. 89, no. 13,p. 133103, 2006.

[155] R. S. Wagner and W. C. Ellis, “Vapor-liquid-solid mechanism of single crystalgrowth,” Applied Physics Letters, vol. 4, no. 5, pp. 89–90, 1964.

[156] R. R. Lunt and V. Bulovic, “Transparent, near-infrared organic photovoltaicsolar cells for window and energy-scavenging applications,” Applied PhysicsLetters, vol. 98, no. 11, p. 113305, 2011.

[157] C. Granqvist, A. Azens, P. Heszler, L. Kish, and L. Österlund, “Nanomate-rials for benign indoor environments: Electrochromics for “smart windows”,sensors for air quality, and photo-catalysts for air cleaning,” Solar Energy Ma-terials and Solar Cells, vol. 91, no. 4, pp. 355 – 365, 2007.

[158] G. A. Niklasson and C. G. Granqvist, “Optical properties and solar selectiv-ity of coevaporated Co-Al2O3 composite films,” Journal of Applied Physics,vol. 55, no. 9, pp. 3382–3410, 1984.

139

[159] S. Zhao and E. Wäckelgård, “Optimization of solar absorbing three-layer coat-ings,” Solar Energy Materials and Solar Cells, vol. 90, no. 3, pp. 243 – 261,2006.

[160] C. G. Granqvist and A. Hjortsberg, “Radiative cooling to low temperatures:General considerations and application to selectively emitting SiO films,” Jour-nal of Applied Physics, vol. 52, no. 6, pp. 4205–4220, 1981.

[161] T. Nilsson, W. Vargas, G. Niklasson, and C. Granqvist, “Condensation of waterby radiative cooling,” Renewable Energy, vol. 5, no. 1–4, pp. 310 – 317, 1994.Climate change Energy and the environment.

[162] C. G. Granqvist, “Oxide electrochromics: An introduction to devices and ma-terials,” Solar Energy Materials and Solar Cells, vol. 99, pp. 1 – 13, 2012. 9thInternational Meeting on Electrochromism.

[163] T. Zhu and J. Li, “Ultra-strength materials,” Progress in Materials Science,vol. 55, no. 7, pp. 710 – 757, 2010.

[164] Q. Sun, Y. A. Wang, L. S. Li, D. Wang, T. Zhu, J. Xu, C. Yang, and Y. Li,“Bright, multicoloured light-emitting diodes based on quantum dots,” NaturePhotonics, vol. 1, 2007.

[165] Z. M. Wang, ed., Quantum Dot Devices. Lecture Notes in Nanoscale Scienceand Technology, Springer New York, 2012.

[166] S. Pillai, K. R. Catchpole, T. Trupke, G. Zhang, J. Zhao, and M. A. Green, “En-hanced emission from Si-based light-emitting diodes using surface plasmons,”Applied Physics Letters, vol. 88, no. 16, pp. 161102, 2006.

[167] D. V. Talapin, J.-S. Lee, M. V. Kovalenko, and E. V. Shevchenko, “Prospects ofcolloidal nanocrystals for electronic and optoelectronic applications,” Chemi-cal Reviews, vol. 110, no. 1, pp. 389–458, 2010.

[168] S. Tong, B. Ren, Z. Zheng, H. Shen, and G. Bao, “Tiny grains give huge gains:Nanocrystal-based signal amplification for biomolecule detection,” ACS Nano,vol. 7, no. 6, pp. 5142–5150, 2013.

[169] K. Kneipp, M. Moskovits, and H. Kneipp, eds., Surface-Enhanced RamanScattering: Physics and Applications. Springer Berlin, 2006.

[170] E. Hutter and J. Fendler, “Exploitation of localized surface plasmon reso-nance,” Advanced Materials, vol. 16, no. 19, pp. 1685–1706, 2004.

[171] X.-J. Huang and Y.-K. Choi, “Chemical sensors based on nanostructured ma-terials,” Sensors and Actuators B: Chemical, vol. 122, no. 2, pp. 659 – 671,2007.

[172] S. Pillai, K. R. Catchpole, T. Trupke, and M. A. Green, “Surface plasmonenhanced silicon solar cells,” Journal of Applied Physics, vol. 101, no. 9,pp. 093105, 2007.

140

[173] H. a. Atwater and A. Polman, “Plasmonics for improved photovoltaic devices,”Nature Materials, vol. 9, p. 205, 2010.

[174] A. Hagfeldt, G. Boschloo, L. Sun, L. Kloo, and H. Pettersson, “Dye-sensitizedsolar cells,” Chemical Reviews, vol. 110, no. 11, pp. 6595–6663, 2010.

[175] A. Llordés, G. Garcia, J. Gazquez, and D. J. Milliron, “Tunable near-infraredand visible-light transmittance in nanocrystal-in-glass composites,” Nature,vol. 500, pp. 323–326, 2013.

[176] llknur Bayrak Pehlivan, R. Marsal, E. Pehlivan, E. L. Runnerstrom, D. J.Milliron, C. G. Granqvist, and G. A. Niklasson, “Electrochromic deviceswith polymer electrolytes functionalized by SiO2 and In2O3:Sn nanopar-ticles: Rapid coloring/bleaching dynamics and strong near-infrared ab-sorption,” Solar Energy Materials and Solar Cells, in press, 2013. DOI:10.1016/j.solmat.2013.06.010

[177] M. Haruta, “Gold as a novel catalyst in the 21st century: Preparation, workingmechanism and applications,” Gold Bulletin, vol. 37, no. 1-2, pp. 27–36, 2004.

[178] A. Z. Moshfegh, “Nanoparticle catalysts,” Journal of Physics D: AppliedPhysics, vol. 42, no. 23, p. 233001, 2009.

[179] X. Chen and S. S. Mao, “Titanium dioxide nanomaterials: Synthesis, prop-erties, modifications, and applications,” Chemical Reviews, vol. 107, no. 7,pp. 2891–2959, 2007.

[180] X. Chen, L. Liu, P. Y. Yu, and S. S. Mao, “Increasing solar absorption for pho-tocatalysis with black hydrogenated titanium dioxide nanocrystals,” Science,vol. 331, no. 6018, pp. 746–750, 2011.

[181] Z. Wang, L. Zhou, and X. W. (David) Lou, “Metal oxide hollow nanostructuresfor lithium-ion batteries,” Advanced Materials, vol. 24, no. 14, pp. 1903–1911,2012.

[182] P. Bruce, B. Scrosati, and J.-M. Tarascon, “Nanomaterials for rechargeablelithium batteries,” Angewandte Chemie International Edition, vol. 47, no. 16,pp. 2930–2946, 2008.

[183] S. D. Brown, P. Nativo, J.-A. Smith, D. Stirling, P. R. Edwards, B. Venugopal,D. J. Flint, J. A. Plumb, D. Graham, and N. J. Wheate, “Gold nanoparticles forthe improved anticancer drug delivery of the active component of oxaliplatin,”Journal of the American Chemical Society, vol. 132, no. 13, pp. 4678–4684,2010.

[184] P. Ghosh, G. Han, M. De, C. K. Kim, and V. M. Rotello, “Gold nanoparticlesin delivery applications,” Advanced Drug Delivery Reviews, vol. 60, no. 11,pp. 1307 – 1315, 2008. Inorganic Nanoparticles in Drug Delivery.

141

[185] X. Huang, I. H. El-Sayed, W. Qian, and M. A. El-Sayed, “Cancer cell imagingand photothermal therapy in the near-infrared region by using gold nanorods,”Journal of the American Chemical Society, vol. 128, no. 6, pp. 2115–2120,2006.

[186] C. Loo, A. Lowery, N. Halas, J. West, and R. Drezek, “Immunotargetednanoshells for integrated cancer imaging and therapy,” Nano Letters, vol. 5,no. 4, pp. 709–711, 2005.

[187] Y. Wang, M. Chen, F. Zhou, and E. Ma, “High tensile ductility in a nanostruc-tured metal,” Nature, vol. 419, pp. 912–915, 2002.

[188] M. Dao, L. Lu, R. Asaro, J. D. Hosson, and E. Ma, “Toward a quantitativeunderstanding of mechanical behavior of nanocrystalline metals,” Acta Mate-rialia, vol. 55, no. 12, pp. 4041 – 4065, 2007.

[189] E. M. Bringa, A. Caro, Y. Wang, M. Victoria, J. M. McNaney, B. A. Remington,R. F. Smith, B. R. Torralva, and H. Van Swygenhoven, “Ultrahigh strength innanocrystalline materials under shock loading,” Science, vol. 309, no. 5742,pp. 1838–1841, 2005.

[190] T. Klaus, R. Joerger, E. Olsson, and C.-G. Granqvist, “Silver-based crystallinenanoparticles, microbially fabricated,” Proceedings of the National Academyof Sciences, vol. 96, no. 24, pp. 13611–13614, 1999.

[191] T. Klaus-Joerger, R. Joerger, E. Olsson, and C.-G. Granqvist, “Bacteria asworkers in the living factory: metal-accumulating bacteria and their potentialfor materials science,” Trends in Biotechnology, vol. 19, no. 1, pp. 15 – 20,2001.

[192] D. Mandal, M. Bolander, D. Mukhopadhyay, G. Sarkar, and P. Mukherjee,“The use of microorganisms for the formation of metal nanoparticles and theirapplication,” Applied Microbiology and Biotechnology, vol. 69, no. 5, pp. 485–492, 2006.

[193] S. Shankar, A. Rai, A. Ahmad, and M. Sastry, “Rapid synthesis of Au, Ag, andbimetallic Au core–Ag shell nanoparticles using neem (azadirachta indica) leafbroth,” Journal of Colloid and Interface Science, vol. 275, no. 2, pp. 496 – 502,2004.

[194] S. K. Das and E. Marsili, Bioinspired Metal Nanoparticle: Synthesis, Proper-ties and Application, Nanomaterials, ch. 11. InTech, 2011.

[195] G. J. Dick, B. G. Clement, S. M. Webb, F. J. Fodrie, J. R. Bargar, and B. M.Tebo, “Enzymatic microbial Mn(II) oxidation and Mn biooxide production inthe guaymas basin deep-sea hydrothermal plume,” Geochimica et Cosmochim-ica Acta, vol. 73, no. 21, pp. 6517 – 6530, 2009.

[196] X. Li, H. Xu, Z.-S. Chen, and G. Chen, “Biosynthesis of nanoparticles by mi-croorganisms and their applications,” Journal of Nanomaterials, vol. 2011.

142

[197] V. Bansal, D. Rautaray, A. Bharde, K. Ahire, A. Sanyal, A. Ahmad, and M. Sas-try, “Fungus-mediated biosynthesis of silica and titania particles,” J. Mater.Chem., vol. 15, pp. 2583–2589, 2005.

[198] Abhilash, K. Revati, and B. Pandey, “Microbial synthesis of iron-based nano-materials, a review,” Bulletin of Materials Science, vol. 34, no. 2, pp. 191–198,2011.

[199] A. K. Jha and K. Prasad, “Ferroelectric BaTiO3 nanoparticles: Biosynthesisand characterization,” Colloids and Surfaces B: Biointerfaces, vol. 75, no. 1,pp. 330 – 334, 2010.

[200] G. Nyström, A. Razaq, M. Strømme, L. Nyholm, and A. Mihranyan, “Ultra-fast all-polymer paper-based batteries,” Nano Letters, vol. 9, no. 10, pp. 3635–3639, 2009.

[201] Y. J. Lee, H. Yi, W.-J. Kim, K. Kang, D. S. Yun, M. S. Strano, G. Ceder,and A. M. Belcher, “Fabricating genetically engineered high-power lithium-ionbatteries using multiple virus genes,” Science, vol. 324, no. 5930, pp. 1051–1055, 2009.

[202] F. G. Omenetto and D. L. Kaplan, “New opportunities for an ancient material,”Science, vol. 329, no. 5991, pp. 528–531, 2010.

[203] M. Pawlyn, Biomimicry in Architecture. RIBA Publishing, reprint ed., 2011.

[204] M. Braungart and W. McDonough, Cradle to Cradle: Remaking the Way WeMake Things. North Point Press, 1st ed., 2002.

143

Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 1095

Editor: The Dean of the Faculty of Science and Technology

A doctoral dissertation from the Faculty of Science andTechnology, Uppsala University, is usually a summary of anumber of papers. A few copies of the complete dissertationare kept at major Swedish research libraries, while thesummary alone is distributed internationally throughthe series Digital Comprehensive Summaries of UppsalaDissertations from the Faculty of Science and Technology.

Distribution: publications.uu.seurn:nbn:se:uu:diva-210016

ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2013