spectrally selective solar absorbing coatings prepared by dc …169700/fulltext01.pdf ·...
TRANSCRIPT
ACTAUNIVERSITATISUPSALIENSISUPPSALA2007
Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 273
Spectrally Selective SolarAbsorbing Coatings Prepared by dcMagnetron Sputtering
SHUXI ZHAO
ISSN 1651-6214ISBN 978-91-554-6803-3urn:nbn:se:uu:diva-7530
To my family
List of Papers
I S. Zhao, C-G. Ribbing, E. Wäckelgård New method to optimize a solar absorber graded film profile Solar Energy, 78 (2005) 125-130.
II S. Zhao, C-G. Ribbing, E. Wäckelgård Optical constants of sputtered Ni/NiO solar absorber film—depth-profiled characterization Solar Energy Materials & Solar Cells, 84 (2004) 193-203.
III S. Zhao, E. Avendano, K. Gelin, J. Lu, E. Wäckelgård Optimization of an industrial DC magnetron sputtering process for graded composition solar thermal absorbing layer Solar Energy Materials & Solar Cells, 90 (2006) 308-328.
IV S. Zhao, E. Wäckelgård Optimization of solar absorbing three-layer coatings Solar Energy Materials & Solar Cells, 90 (2006) 243-261.
V S. Zhao, E. Wäckelgård The optical properties of sputtered composite of Al–AlN Solar Energy Materials & Solar Cells, 90 (2006) 1861-1874.
VI S. Zhao, E. Wäckelgård Theoretical study of a spectrally selective Ni-NiO absorber Manuscript
Comments on my contribution I am responsible for the major part of the writing in the above five publica-tions and one manuscript. I prepared all samples, and performed all experi-ments and measurements apart from the parts mentioned below: X-ray diffraction was measured by E. Avendano, photos of scanning elec-tron microscopy were taken by K. Gelin, and Image of transmission electron microscopy was performed by J. Lu.
Confidential reports written for the company Sunstrip AB whose content is not included in this thesis:
S. Zhao, E. Wäckelgård Profile analysis of Sunstrip samples Uppsala (2003), 10 p.
K.Gelin, S. Zhao and E. Wäckelgård Theoretical and experimental optimization of an antireflection coating on a nickel-nickel oxide absorbing layer for solar thermal applicationsUppsala (2004), 11 p.
Publication not included in this thesis
S. Zhao, K. Gelin, E. Wäckelgård, C-G. Ribbing Roll Coater for gradient film coating Proceeding of Optik I Sverige 2001, Stockholm, Sweden.
Contents
1. Introduction...............................................................................................11
2. Physical fundamentals ..............................................................................142.1. Solar and thermal radiation ...............................................................142.2. Ideal solar spectral selectivity ...........................................................162.3. Different designs of solar selective surface.......................................18
2.3.1. Intrinsic materials ......................................................................182.3.2. Optical trapping surfaces—textured surfaces ............................182.3.3. Tandem absorbers ......................................................................18
2.4. Solid state optics................................................................................202.4.1. Macroscopic description............................................................202.4.2. Optical properties of dielectric material and Lorentz model .....212.4.3. Optical properties of metal and the Drude model......................23
2.5. Effective medium theory—optical constants for composite .............242.5.1. Effective medium theory ...........................................................242.5.2. Optical constants from the Bruggeman model ..........................242.5.3. Optical constants from the Maxwell-Garnett model..................25
2.6. Thin film optics .................................................................................262.6.1. Single thin film reflection..........................................................262.6.2. Multilayer reflection ..................................................................31
2.7. Sputtering ..........................................................................................322.7.1. Sputtering process and dc planar magnetron sputtering ............322.7.2. Reactive sputtering ....................................................................33
3. Experiment set-up and measurements.......................................................353.1. System for sample preparation ..........................................................35
3.1.1. Roll-coater sputtering deposition system...................................353.1.2. Balzer sputtering unit.................................................................36
3.2. Optical characterization.....................................................................373.3. Material characterization...................................................................38
3.3.1. X-ray diffraction (XRD) ............................................................383.3.2. Scanning electron microscopy (SEM) .......................................393.3.3. Transmission electron spectroscopy (TEM) ..............................393.3.4. X-Ray Photoelectron Spectroscopy (XPS) ................................39
3.4. Other measurements ..........................................................................403.4.1. Resistance measurements ..........................................................403.4.2. Thickness measurements ...........................................................41
4. Methodology .............................................................................................424.1. Method for characterization of graded film profiles .........................424.2. Determination of optical constants of cermets ..................................444.3. Theoretical optimization of spectral selectivity ................................45
5. Results and discussions.............................................................................475.1. Tailoring of the film depth profile.....................................................475.2. Optimization sputtering process for Rulle.........................................505.3. Material characterization...................................................................52
5.3.1. X-ray diffraction (XRD) ............................................................525.3.2. X-ray photoelectron spectroscopy (XPS) ..................................545.3.3. SEM micrograph........................................................................555.3.4. TEM micrograph .......................................................................56
5.4. Optical constant of sputtered Ni-NiO composite ..............................575.5. Theoretical simulation of selective surface for Ni-NiO composite as absorber ....................................................................................................60
5.5.1. Three-layer structure..................................................................605.5.2. Graded absorbing layer..............................................................64
5.6. Experimental results of optimized selective surface for Ni-NiO composite as absorber ..............................................................................67
6. Conclusions...............................................................................................69
Acknowledgements.......................................................................................72
References.....................................................................................................75
Tillverkning av sputtrade solabsorbatorer med Ni-NiO komposit................83
List of symbols and constants
T Absolute temperature Wavelength
h Planck’s constant, 6.626*10-34 JskB Boltzman’s constant, 1.381*10-23 J/Kc Speed of light in vacuum, 2.9979*108 m/sR( ) Reflectance T( ) Transmittance A( ) Absorptance ( ) Emittance
EM Electromagnetic field E Amplitude of electric field
iknN Complex optical constant n Refractive index k Extinction coefficient I( ) Intensity of light I0( ) Intensity of light at x = 0Ib( , T) Intensity of blackbody radiation
s Total normal solar absorptance t Total normal thermal emittance
d Film thickness, distance of propagation Absorption coefficient
Angular frequency of electric field 0 Resonance frequency of Lorentz model P Plasma frequency of material = 1+i 2 Complex dielectric constant 1, 2 real and imaginary part of 0 Permittivity in vacuum, 8.86*10-12 As/Vm
f Volume fraction of metallic component r Complex amplitude reflection of thin film rj Complex amplitude reflection at jth bound-
aryj Phase change at jth layer
m interference order Rf Film resistance Rm Metallic film resistance
Film conductivity
AbbreviationsAM Air mass BL Base layer ML Middle layer AR Anti-reflection layer NOF Normalized oxygen flow NNF Normalized nitrogen flow ISO International Organization for
StandardizationXRD X-ray diffraction XPS X-ray photoelectron spectroscopy SEM Scanning electron microscopy TEM Transmission electron microscopy SAED Selected area electron diffraction NC Normalized conductivity RO The relative oxygen flow
11
1. Introduction
Energy is usually and most simply defined as the equivalent of capacity for doing work. The word itself is derived from the Greek energeia: en, “in”; ergon, “work”. Human beings learned to use energy from natural sources to extend his power. Thus energy has been playing an important role for civili-zation. In the early ages, wood was the main source of energy for developing agriculture civilization. Industrialization and modernization started around two hundred years ago were mainly based on fossil fuels. In the past century, fossil fuels have been widely used on a large scale. Huge demands on such energy give the facility of our life, but cause environmental problems and rises in coal and oil prices as well. The limited availability of fossil fuel, and their environmental impact, has led to a growing awareness of the impor-tance of renewable energy sources. Solar energy is the source for most re-newable energy.
Solar energy can be converted into useful forms through different ways [1, 2]. Mass flow in the atmosphere can be used for wind power which uses the solar energy through moving air resulting from the non-uniform heating of the earth’s surface. The water flow can be used for hydroelectric/hydro me-chanical power which uses solar energy through water circulation in the earth driven by the sun. In the biosphere, power from biomass and biogas uses solar energy through energy crops. These cases are regarded as indirect uses of solar energy.
Direct use of solar radiation mainly consists of photovoltaic (PV) and solar-thermal energy conversion, and belongs to the type of technology in which solar energy materials play a key role. PV technology for the conversion of sunlight into electricity is already a cost-effective method for many applica-tions worldwide [3]. At the same time, the technology of solar-thermal en-ergy conversion becomes mature and cost-competitive. Solar-thermal sys-tems are rapidly spreading into many application areas. In many countries people build flat plate solar collectors for domestic hot water and house warming. In China, solar-thermal energy systems are mainly used for do-mestic hot water based on vacuum tube collectors [4]. Meanwhile, solar-thermal energy conversion systems also find their use in solar process heat-ing, swimming pools, air-conditioning, and solar thermal power plants.
12
Like many others, the application of solar-thermal energy depends on the progress of materials in the solar collector. For solar energy materials, most of them involve thin films or surface treatment in one way or another. It is the same for solar-collector materials. A solar collector is a device which absorbs solar radiation, converts it to thermal energy, and delivers the ther-mal energy to a heat-transfer medium. Energy efficiency, i.e., minimized losses associated with the energy transfer, can be achieved by using suitable materials in the components of the solar collector. Most solar collectors use a spectrally selective absorber surface which has high absorptance at solar spectrum range and low emittance at infrared spectrum range to minimize heat radiation losses. Much research and development was conducted during the later half of the 1970s and the early 1980s on coatings and surface treat-ments yielding such properties [5].
There are several design options and physical mechanisms creating a selec-tively solar-absorbing surface. Among them, metallic composition randomly mixed with dielectric composition in nano-particle scale known as metal-dielectric composite or cermet (ceramic metal) is widely used. Dielectric rich composites are, in general, strongly absorbing over much of the solar spec-trum. In the infrared spectral range they rapidly become highly transparent with increasing wavelength. When such a composite coating is formed on a highly reflecting metal surface the resulting absorber/reflector tandem has a good spectral selectivity.
The composite coatings can be produced by a variety of techniques on a large scale. Chemical conversion has been the most widely used technique. For example, electroplated black chromium [6, 7] and black nickel [8, 9] are popular solar selective absorbers. Nicked-pigmented anodic Al2O3 solar ab-sorber [10, 11] is another popular solar selective absorber. However, the chemical process is usually space and material consuming and produces much chemical waste. During the 1990s the manufacturers of absorbers de-veloped new coating techniques based on vacuum technologies. Vacuum techniques such as plasma sputtering and evaporation offer larger possibili-ties in coating compositions: multilayer thin films or composites [12-15]. Especially when magnetron sputtering technology was developed in the 1970s [16] for selective coatings, solar porformance has much improved, in particular, with lower emittance and less enviromental pollution than the electrochemical methods.
For ceramic metal composite absorber, a harmful aspect is the high refrac-tive index which causes a high front surface reflection, and the solar absorp-tance is hence reduced. The most effective way to counter this effect is to prepare a compositional graded absorber such that the metallic content is zero at the front surface and maximum at the composite-substrate interface
13
[17]. An anti-reflection coating can be added on the top to further improve the absorbing property which decreases front surface reflection by choosing low refractive index material.
The work of this thesis mainly concerns graded absorber layer. It has been carried out in the period 2000-2006 (half-time work) at the division of Solid State Physics, Department of Engineering Science, The Ångström Labora-tory, Uppsala University. The project was performed in cooperation with the Swedish company, Sunstrip AB, which is one of the leading companies pro-ducing spectrally selective solar absorbers. A miniature copy of an industrial roll coater unit in Sunstrip was made for study. The main coating material is nickel-nickel oxide which is used in Sunstrip AB as an absorber layer. For comparison, aluminium-aluminium nitride composite is studied which is widely used in vacuum tube collectors in the Chinese market.
The main goal of this work is to optimize process parameters for the im-provement of the optical performance, i.e. how to obtain a high solar absorp-tance, while keeping low thermal emittance for the absorbing coatings. Du-rability issues are not included though they are important for production. For industrial scale, a roll is adopted for continuous coating. The movement of substrate creates a feature which makes it possible to control film profile properties easily. The design of the sputtering zone for absorbing in Rulle isunique in that asymmetric reactive sputtering occurs along the sputtering zone due to the location of a gas spray. This distribution can be transferred into graded film profile by the movement of the substrate. Thus study and characterization of zone property is of primary importance. A lot of work focuses on how process parameters influence the distribution property of the sputtering zone. The study also concerns the optimized depth profile for graded film including theoretical simulation. Determination of optical con-stants for sputtered nickel-nickel oxide composite is one necessary step for the simulation of real samples.
14
2. Physical fundamentals
This chapter provides a fundamental physical background used for perform-ing this work. It starts with thermal and solar radiation. The concept of a spectrally selective solar absorber is of fundamental importance for the work which is the next section. Then it is followed by an introduction of different absorber coating designs during the past decades and in particular concerns the widely used type - composite-reflector tandem. The section on solid state optics treats the interaction between light and solids using two classical models. Using an effective medium theory predicting optical constants of composite, the reader may understand how a composite absorbs in solar spectrum region. In the section on thin film optics, we show that spectral selectivity can be further improved by interference. Finally, the reactive sputtering technique is briefly introduced which is one widely used tech-nique to produce solar selective films.
2.1. Solar and thermal radiation All matter emits thermal electromagnetic radiation. The intensity and wave-length distribution of the radiation depends on its absolute temperature, T. A perfect absorber that absorbs all incoming radiation is called a blackbody. The wavelength distribution of radiation emitted by a blackbody is given by Planck’s radiation law [18, 19]:
112),( /5
2
Tkhcb BehcTI (1)
Where h is Planck’s constant, kB is Boltzman’s constant, and c is the speed of light in vacuum. The symbol Ib represents the intensity of blackbody ra-diation, i.e. energy per unit area per unit time per unit wavelength interval at wavelength emitted by a blackbody surface.
Electromagnetic radiation occurs over widely different wavelength ranges: from cosmic, gamma, X-rays (wavelengths 10-8 m) to long radio waves (wavelengths 1010 m). Sunlight is electromagnetic radiation in the spectral
15
range of 0.3 to 4 m with its maximum intensity around 0.5 to 0.7 m. This spectrum corresponds to an effective blackbody temperature of about 5800 K [20]. The wavelengths for many solar energy applications are found in the range from the ultra violet at 0.3 to the infrared at 50 m covering the solar spectral range and the spectral range of the thermal radiation emitted from a surface having a temperature of about ambient up to 100oC (2 to 50 m).
Terrestrial solar radiation is a low-intensity, variable energy source reaching a maximum of about 1000 W/m2. The intensity of the terrestrial spectrum depends on the distance traveled through the atmosphere. Outside the atmos-phere the spectrum is denoted air mass zero, AM 0, while the radiation that travels through the atmosphere is AM X where X is given by 1/cos z for zless than 70o. z is the angle of incidence with respect to zenith [19]. AM 1.5 is used in the following sections to calculate solar absorptance which corre-sponds to when the sun is about 42o above the horizon.
Figure 1 shows a normalized solar spectrum and normalized blackbody ra-diation spectra at three different temperatures. The solar spectrum has sev-eral local minima which are caused by absorption in the atmosphere mainly by water vapor, carbon dioxide and ozone [21]. The solar and thermal spec-tra do not overlap to any appreciable extent (for temperatures below 500oC,98% of the thermal infrared radiation occurs at wavelength greater than 2
m). This is the physical basis of solar spectral selectivity.
0
0.2
0.4
0.6
0.8
1
0
0.2
0.4
0.6
0.8
1
1 10
Nor
mal
ized
sola
r ra
diat
ion
Norm
alized blackbody radiation
Wavelength ( m)
NormalizedSolarRadiation
NormalizedBlackbodyRadiation
T = 573 K
473 K
373 K
.3 .5 2 5 20
Ideal selectivereflectancespectrum
35
Figure 1. The normalized solar radiation spectrum, blackbody radiation spectra at three different temperatures, and the ideal selective reflectance spectrum.
16
2.2. Ideal solar spectral selectivity For solar-thermal energy conversion systems it would be desirable to capture all incident light to the absorbing surface and then deliver a high possible fraction of the captured energy to the user through a working fluid. Some amount of thermal energy is unavoidably lost from the heated absorber due to conduction, convection, and radiation. These losses are suppressed with different methods. A glass cover and thermal insulation used by a flat plate collector [22] will suppress the loss due to conduction and convection. For a vacuum tube collector, the vacuum will suppress the loss due to conduction and convection. The radiation loss from a hot surface can, in both cases, be suppressed by a spectrally selective absorber surface, i.e. depending on ma-terials used [23].
When incident light strikes on a surface of a solid material, some part of the incident light is reflected, some part is absorbed, and some part is transmit-ted. Reflection, absorption and transmission are wavelength and incident angle dependent. Solar reflectance measurements are usually performed at near normal angle of incidence using spectrophotometers equipped with an integrating sphere. The reason is due to the fact that most sample surfaces show microscopic roughness causing scattering, which usually is fairly ho-mogeneously distributed over the hemisphere. In this case, the integrating sphere will collect all signals of the scattering samples [24]. The following discussion refers to normal incidence unless differently specified.
At the surface, the reflected fraction of incident light intensity at a specific wavelength, , is called reflectance R( ); the absorbed fraction is called ab-sorptance, A( ); and the transmitted fraction is called transmittance, T( ).Energy conservation gives [25]:
1)()()( TRA (2)
The definition of emittance, ( ), is the ratio between infrared light emitted from a surface and that emitted from a perfect blackbody at the same wave-length and the same temperature. According to Kirschoff’s law, the emit-tance is equal to the absorptance [25]:
)()(1)()( TRA (3)
For an opaque surface, the transmittance is zero, so equation (3) becomes:
)(1)( R (4)
17
An ideal absorber surface should have zero reflectance in the solar spectral range to maximize the capture of solar energy and reflectance equal to one in the infrared wavelength range to minimize radiation losses. Thus, an ideal spectrally selective surface should have an abrupt transition from low to high reflectance at a transition wavelength. This behavior has different names, such as ‘crossover behavior’, ‘step function’, etc. by different authors. We call it ‘reflectance transition behavior’ in this summary. The transition wave-length to choose depends on the working temperature. For most low tem-perature applications (below 100oC), a transition wavelength of 2.5 m is used. Figure 1 shows the ideal reflectance curve in this case.
Solar absorptance and thermal emittance are two basic parameters for char-acterizing the selective nature of various coating surfaces. The total solar absorptance calculated with a normal angle of incidence, s, is a weighted average, weighted by the solar spectral radiation, IS( ), as shown in equation (5) where 1=0.3 m, 2=4.1 m. In the following calculations AM 1.5 solar spectrum, defined by the ISO standard 9845-1 (1992) [13], was used.
2
1
2
1
)(
)())(1(
dI
dIR
S
S
s (5)
The total thermal emittance calculated with a normal angle of radiation, t, is a weighted average, weighted by the blackbody radiation, Ib( , T), for the given temperature, T, as shown in equation (6) where 3 depends on tem-perature and 4 = 100 m [26]. For T = 373 K, 3 is 2.0 m. In the following calculations, temperature is chosen as 373 K, if not otherwise specified.
4
3
4
3
),(
),())(1(
dTI
dTIR
b
b
t (6)
18
2.3. Different designs of solar selective surface A variety of selective surfaces have been proposed and fabricated for solar thermal applications. Three main types of absorbing surfaces can be identi-fied from a structural point of view: intrinsic, optical trapping surface, and tandem. Tandem selective surfaces can further be classified into two catego-ries: (1) absorber-reflector tandems; or (2) heat mirrors. Most of the com-mercially selective surfaces are constructed as absorber-reflector tandems [1]. This section therefore mainly focuses on absorber-reflector tandem sur-faces.
2.3.1. Intrinsic materials An intrinsic solar absorbing material should be wavelength selective. How-ever, there is no material occurring in nature which exhibits sufficiently solar selective properties. The transition from low to high reflectance either occurs at the wrong wavelength or is not steep enough. The best absorber of this type, known today, is ZrB2 [27, 28].
2.3.2. Optical trapping surfaces—textured surfaces Special rough surface textures can be very efficient in trapping light by mul-tiple reflections that are 2 m apart, while the long-wavelength emittance is fairly unaffected by this texture. Well known examples include dendritic tungsten [29], textured copper, nickel and stainless steel surfaces [30, 31]. However, they are not widely used in practice.
2.3.3. Tandem absorbers A tandem absorber consists of at least two layers with different optical prop-erties. These layers together make a coating that is suited as a solar absorber. For absorber-reflector tandems, a coating having high absorptance at solar wavelengths is deposited onto a highly IR-reflecting metal substrate (e.g., aluminium, copper, silver, etc.). High absorption of the exterior coating is achieved with different mechanisms: either intrinsic in nature (e.g. semicon-ductor), or may be a combination of the two (e.g. composite).
A heat mirror on black substrate-absorber uses the tandem concept but in an opposite way: the coating is transparent to the solar spectrum but highly reflective in the infrared wavelength interval, while the substrate is a non-selective absorber. SnO2: F on a substrate of black enamel is one example [32]. In the following, only different absorber-reflector tandems are dis-cussed.
19
2.3.3.1. Semiconductor-reflector tandem In order to get a high solar absorptance, a semiconductor should have an energy band gap from about 0.5 eV (2.5 m) to 1.26 eV (1.0 m) which would absorb solar radiation, but is transparent to IR radiation. However, the useful semiconductors usually have high refractive indices, which tend to yield high front reflection losses. Hence they are coated by anti-reflection layer to reduce such losses. Early work on this subject was concerned with Si based designs [33, 34].
2.3.3.2. Multilayer-metal tandems In the case of multilayer interference stacks, the selective effect is a result of a multiplicity of passes through the dielectric layers of the stack. It is com-paratively easy to optimize optical performance of such a stack by an optical multilayer program. One example is an Al2O3/Mo/Al2O3 triple layer, which was originally developed for the US space project. This type of surface has been produced for high temperature applications [35, 36].
2.3.3.3. Metal-dielectric composite tandem Composites are prevalent both in nature and among engineered materials. They are widely used in various fields. What gives composites their utility is that they often combine the properties of the constituent materials. Compos-ites used for solar-thermal energy conversion consist of very fine metal par-ticles embedded in a dielectric host, also known as granular or cermet mate-rials [37]. Optical properties appropriate for good selective solar absorbers may be achieved. Such composites are strongly absorbing in the solar spectral range. In the infrared spectral range, they rapidly become highly transparent with increasing wavelength. When such a composite is coated on a highly reflecting metal surface as a thin film, the resulting absorber-reflector tandem shows good solar spectrally selective character.
The optical properties of such composites can be intermediate between those of the metal and of the dielectric. Effective medium theory (EMT) can be used to predict optical performance (see 2.4). The metal-dielectric concept offers a high degree of flexibility, and the optimization of the solar selectivity can be made with regard to the choice of constituents, coating thickness, particle concentration (metallic content), and the size, shape, and orientation of the particles [38, 39]. Thus, a metal-dielectric composite becomes the most useful material in solar-thermal applications. Such composite coatings are widely studied [40-48] including nickel based cer-mets [49-54]. The composite of nickel-nickel oxide is the coating material mainly studied in my work, which is manufactured by the Swedish company Sunstrip AB [13, 55]. The following discussions are mainly focused on
20
nickel-nickel oxide composite. The composite of aluminium-aluminium nitride is widely used in China [4, 56], and the Optical constants of alumin-ium-aluminium nitride are also discussed in this summary.
2.4. Solid state optics2.4.1. Macroscopic description We consider a plane monochromatic electromagnetic wave which propa-gates in a non-scattering homogeneous material with a complex optical con-stant N=n+ik, where n is the refractive index and k is the extinction coeffi-cient1. Furthermore, in the solar and infrared wavelength regions, the mag-netic permeability approaches unity, we only need to treat the electric part of the EM wave and assume 1. The electric field of a plane wave propagat-ing in the x-direction can then be written as:
))/(exp()/exp(),( 0 cxnticxkEtxE (7)
where, E(x,t) is the amplitude of electric field at position x and time t after entering the medium, is the angular frequency, and c is the speed of light in vacuum. Equation (7) shows that the amplitude attenuates along the path through the medium. And the light intensity, I, is proportional to the square of amplitude:
)/4exp()/2exp(),( 20
20
2 xkEcxkEtxEI (8)
Attenuation of the light intensity obeys an exponential law, i.e. Beer’s law:
deII 0 (9)
where I0 is the intensity at x = 0, d is the distance of propagation, is the absorption coefficient. By comparing (8) with (9), we obtain the relation between and k:
/4 k (10)
1 There are two different notations for the complex optical constant. Notation N = n - ik is adopted by some authors [57]. We adopt notation “+” i.e.
iknN . The final optical result, such as bulk reflectance, is the same.
21
Thus, the imaginary part of the complex optical constant describes the ab-sorption of light (electromagnetic energy) by a medium through equation (9) and (10) [58]. One significant feature is that the absorption coefficient is proportional to k/ , i.e. absorption is stronger at short wavelengths if the variation of k is small. When k is weakly dependent on wavelength as in many practical cases, it yields an absorptivity, =4 k/ , decreasing as 1/ .If a thin film of such material is coated on a metallic substrate, the corre-sponding reflectance increases gradually with increasing wavelength [37]. The transition of reflectance from low to high can thus be controlled to be in a near infrared region by film thickness. Moreover, transition from low to high reflectance becomes steeper if thin film interference occurs at the proper wavelength region which will be discussed in section 2.6.1.
n and k are called ‘constant’ due to historical reasons. In fact, they are func-tions of wavelength. The variation of optical constants on incident wave-length is dispersion.
The complex dielectric function 21 i is related to the complex opti-cal constant by the equation:
2N (11)
Equation (11) gives the following relations:
221 kn (12)
nk22 (13)
The optical constants are more useful for the purpose of investigating what happens to the direction and intensity of light when it interacts with matter, while the dielectric function is more suited when we describe the micro-scopic effects inside a solid. The dielectric function is often in model discus-sions based on polarization.
2.4.2. Optical properties of dielectric material and Lorentz model From a classical microscopic point of view, the optical properties of materi-als are determined by the coupling of various types of oscillators in matter to the electromagnetic radiation field. In other words, an incident electromag-netic field will cause these oscillators to perform forced oscillations. A clas-sical model for the response of bound charge (or electron) to an EM wave is the Lorentz model. It expresses the complex dielectric constant as a function
22
of material parameters, such as resonance frequency, damping constant, and plasma frequency [59].
For nickel oxide, the Lorentz model can be used to model the ionic oscilla-tion in the infrared region. The resonance frequency corresponds to a wave-length of 24 m. Another resonance frequency corresponds to a wavelength of 0.29 m which is outside the interesting region, so it can be neglected [60]. Figure 2 below illustrates the optical properties (n, k and normal bulk reflectance R) of nickel oxide [61]. The normal bulk reflectance is calculated using equation (14)
22
222
)1()1(
11
knkn
NNR (14)
There are some features shown in figure 2: In the most solar spectrum range (i.e. >0.4 m), and extending to 18 m, n> k, k = 0, the material is transparent. At the short wavelength side of the resonance, high reflectance re-gion, referred to as Reststrahlen band [62], is followed where k>n,correspondingly 1 is negative. The light is “rejected” by the oscilla-tor.High absorptance region around the resonance peak where n>k, and k >> 0.
0
0.2
0.4
0.6
0.8
1
1 10
Opt
ical
con
stan
t - n
, kN
ormal bulk reflectance
Wavelength ( m)
n
kR
R
k
.3 .5 2 5 20 30
23
Figure 2. Optical properties (n, k, and normal bulk reflectance R) of nickel oxide as function of wavelength which shows the optical features of material described by the Lorentz model.
2.4.3. Optical properties of metal and the Drude model The complex dielectric constant of free-electron metals can be obtained by the Drude model as a function of plasma frequency and relaxation time [59]. The plasma frequency of metals typically lies in the visible or ultraviolet spectral region in which the reflectance shows a dramatic drop. Starting at the wavelength corresponding to plasma frequency, towards increasing wavelength, 1 is strongly negative, consequently, k>n, light is “rejected” by the oscillator, i.e. the reflectance is high. So for metal, its k is bigger than nin the solar spectrum range.
Nickel is a metal belonging to the iron group, and it is not a real free-electron metal. The reflectance in the solar range shows a noticeable slope towards short wavelengths, and this is caused by low energy interband tran-sitions among the d-electrons [63]. However, its optical properties show clearly the contribution from free-electron behavior. Figure 3 shows the op-tical properties (n, k and normal bulk reflectance R) of nickel [64]: in the wavelength range up to 30 m, 1 < 0, or k >n>0.
0
20
40
60
80
100
0.4
0.5
0.6
0.7
0.8
0.9
1
1 10
Opt
ical
con
stan
t - n
, kN
ormal bulk reflectance
Wavelength ( m)
n
k
R
.3 .5 2 5 20 30
Figure 3. Optical properties (n, k, and R) of nickel illustrated as function of wave-length.
24
2.5. Effective medium theory—optical constants for composite2.5.1. Effective medium theory The optical constants of an inhomogeneous composite medium, consisting of small metallic particles embedded in a dielectric matrix, can be derived from the optical constants of the homogeneous constituents. The condition for this is that: the size of the inhomogeneties is much less then the wavelength of the incident light. Then, the electric and magnetic fields can be regarded constant over this characteristic length. It is called the quasistatic approxima-tion. As mentioned in section 2.4.1, 1 in the solar and infrared wave-length regions, therefore the optical properties can be treated by an effective dielectric function on the basis of effective medium theory (EMT) [42, 65-71]. The most commonly used effective medium theory is the Maxwell-Garnett [72] and Bruggeman models [73]. They can be used to model the effective quantity which depends on the dielectric functions of correspond-ing constituents, the shape of particles, and their volume fractions.
The basic difference between the Bruggeman approach and Maxwell-Garnett is that Maxwell-Garnett assumes the metal to be dispersed as particles em-bedded in the dielectric (separated grain structure); while the Bruggeman model assumes the composite to consist of randomly intermixed particles of dielectric and metal (aggregate structure). This distinction gives different results: the Bruggeman theory tends to predict broader absorption curves than does the M-G theory. Experimentally, if composite coatings show in-termixed micro-crystals of metal and dielectric, they will exhibit broad ab-sorption and agree better with the Bruggeman theory [74]. Such broader absorption bands would be somewhat more favorable for solar energy appli-cations, as will be shown in section 5.5.
2.5.2. Optical constants from the Bruggeman model The Bruggeman effective medium approximation for spherical particles is presented in equation (15) [73]:
02
)1(2 BR
D
BRD
MBRM
BRM
M ff (15)
where fM is the volume fraction of metallic component, stands for dielectric functions, indexed by M for the metal, D for the dielectric component re-spectively, and BR for dielectric function of metal-dielectric composite ob-tained from Bruggeman model.
25
Figure 4 shows optical constants of nickel-nickel oxide composite calculated at three different nickel volume fractions: 0.18, 0.24, 0.30, the choice of these values will be discussed in section 5.5. Optical data for nickel oxide and metallic nickel are the same as shown in figure 2 and figure 3. Figure 4 illustrates how the composite changes the optical constants from the con-stituents. A prominent feature of optical constants calculated from equation (15) includes:
In the solar spectrum region, n>k>0, 1 positive, light is not rejected but absorbed. It is different from both nickel (light rejected) and nickel oxide (transparent). This is the property needed for solar ab-sorbers.Both n and k have a maximum value in the wavelength region be-tween 0.3 to 17 m. Later discussion will show that a sharp drop in n, k around the transition wavelength will boost a steep reflectance transition.In wavelength region between 0.3 to 17 m, k increases with in-creasing metallic content, and n shifts its maximum to longer wave-length.
-5
0
5
10
15
20
0
5
10
15
20
25
1 10
Ref
ract
ive
inde
x - n
Extinction coefficient - k
Wavelength ( m)
n
k
f: 0.30 0.24
0.18
f: 0.30
0.240.18
Optical constants of Ni-NiOcalculated by Bruggeman model
k
n
n
.3 .5 2 5 20 30
Figure 4. Optical constants of nickel-nickel oxide composite calculated using the Bruggeman model for three different nickel volume fractions, as indicated.
2.5.3. Optical constants from the Maxwell-Garnett model As a comparison, optical constants of nickel-nickel oxide composite with three different metallic contents are also calculated with the Maxwell-Garnett model using equation (16) [72]:
26
DM
DMM
DMG
DMG
f22
(16)
where MG is the dielectric function of metal-dielectric composite in the Maxwell-Garnett model, the rest is the same as in equation (15). Figure 5 shows optical constants of nickel-nickel oxide composite calculated for the same nickel volume fractions as in figure 4. The difference includes: n has no maximum in the wavelength range before infrared resonance; and k peak is narrower than that shown in figure 4.
1 100
5
10
15
20
25
Wavelength ( m)
Ref
ract
ive
inde
x - n
Extinction coefficient - k
Optical constants of Ni-NiOcalculated by Maxwell-Garnett model
n
f: 0.30
0.180.24
k
n
n
kf: 0.300.18
.3 .5 2 5 20 30
Figure 5. Optical constants of nickel-nickel oxide composite calculated by Maxwell-Garnett model at three different nickel volume fractions, as indicated.
2.6. Thin film optics2.6.1. Single thin film reflection For a single homogeneous thin film, two boundaries are considered for the optical behaviour. The numbering of the boundary adheres to the convention of optical multilayer, i.e. counting from the top to the substrate. Thus, com-plex amplitude reflection on the first (air/thin film) and second (thin film/substrate) boundary are presented as r1, r2 in equation (17), (18) assum-ing that semi-infinite of materials exist on both sides. This is the case of normal incidence for simplification.
27
10
101 NN
NNr (17)
21
212 NN
NNr (18)
where N0, N1, N2 are the complex optical constants of air (or vacuum), thin film, and substrate material respectively.
The reduction in amplitude inside an absorbing medium is exponential and described by equation (9), so when the condition d >>1 is not fulfilled, we have a thin film case. Then the transmitted light can be reflected from the substrate and transmitted through or re-reflected from the boundary of air/thin film. If the boundaries are smooth and the thin film is homogeneous on a length scale larger than the wavelength of light, it is the case called phase coherence between light waves which are successively reflected be-tween two boundaries. Consequently front surface reflection may account for interacting light with the same phased wave, or opposite phased wave which will raise or reduce the total reflection. This is the well known situa-tion of thin film interference
The complex amplitude reflection of such homogeneous thin film is ex-pressed in equation (19) [57].
1
1
221
221
1 i
i
errerrr (19)
The phase change in the film is given by equation (20):
1112 dN (20)
where d1 is the film thickness, is the wavelength.
The term 12ie contains phase shift information and is further expressed by equation (21):
))/4sin()/4(cos( 11/42 11 dnidnee dki (21)
In equation (21), the factor 4n1d/ controls the periodicity of sine, cosine functions which determines thin film interference property. Some papers [75, 76] call it “interference order” m as defined in equation (22):
28
m = 4n1d1/ (22)
Thus, using equation (22) and (9), equation (21) can further be expressed by equation (23)
))sin()(cos(/ 02 1 mimIIe i (23)
Thin film absorption has two different mechanisms: intrinsic absorption characterized by extinction coefficient k; and interference induced absorp-tion characterized by refractive index n. We will illustrate optical properties of cermet thin film on metal tandem and how n and k of cermet controls the optical properties through an example.
Figure 6 shows nickel-nickel oxide composite thin film reflectance spec-trum. The optical constants were calculated using Bruggeman model with a nickel volume fraction of 0.18 (see figure 4). The tandem contains such ho-mogeneous thin film with 120 nm thickness on 1 mm aluminium substrate. Figure 6 also illustrates the bulk reflectance R1 with the same optical con-stant as the thin film composite. It is obvious that the difference observed in R is due to the light reflected from the substrate. Different regions show dif-ferent features:
Weak interference or no interference is due to strong absorption in short wavelength range, the reflectance in this wavelength range is mainly determined by front surface; Strong interference region afterwards due to less absorption; Transparent in infrared as increasing wavelength due to zero absorp-tion.
29
0
0.2
0.4
0.6
0.8
1
1 10
Ref
lect
ance
: R,
R1
Wavelength ( m).3 .5 2 5 20 30
Int +Int -
R: thin filmreflectance
R1: bulk reflectance
from front material
TransparentRange
m=1
m=.50
m=2
InterferenceRange
AbsorptionRange
Figure 6. Reflectance for a thin film of 120 nm thickness of nickel-nickel oxide composite with metallic content f = 0.18 on 1 mm thick aluminium substrate. Bulk reflectance R1 having optical constants as thin film composite is also illustrated. Differences between R and R1 are due to interference.
Figure 7 shows k, d (left y-axis), and light attenuation of I/I0 (right y-axis, note the zero of y raises for better observation) of above case. The correla-tion between thin film reflectance, R, and the extinction coefficient of cer-met, k, can be seen by comparison of figure 7 to figure 6 in the same wave-length regions. The following features related to the dispersion in k are ob-served:
The absorption wavelength range (in this example, from 0.3 to 0.5m), where absorption is very strong. A very small fraction of light
reaches the substrate, so there is no (or weak) interference in this wavelength range.The interference wavelength range (in this example, from 0.5 to 3
m) where the absorption decreases with increasing wavelength, and some part of the light is transmitted through the thin film and reaches the substrate and re-reflected from the substrate causing in-terference. The observed interference is effective in this range.The transparent range (in this example, > 3 m) where the film is transparent. The high reflectance is mainly due to the metallic sub-strate. The absorption band around 20-30 m will be much weaker taking into account the metal substrate.
30
0
1
2
3
4
5
-1
-0.5
0
0.5
1
1 10
k,
*dIntensity attenuation
k: f 0.18
d: f 0.18
I/I0
= e-ad
f 0.18
Wavelength ( m)
Absorbedby film
Reached at substrate
.3 .5 2 5 20 30
Figure 7. k, d, and I/I0 using optical constant of Ni-NiO composite with f = 0.18 at the specified film thickness d = 120 nm.
As discussed above, the factor of nd determines the interference position which can be estimated by the parameter of interference order m. For m, two values are important for thin film interference. One is around m = 1, corre-sponding to the first interference minimum. Another is around m = 0.5, which roughly corresponds to the high reflectance position after reflectance transition.
The dispersion effect of refractive index n is illustrated in figure 8 in which curves 4nd and as function of wavelength are displayed to find the posi-tions corresponding to m=1 and m=0.5 for Ni-NiO composite with f = 0.18 and 0.30 at film thickness of 120 nm.
The correlation between thin film reflectance and refractive index of cermet, n, can be seen by comparing figure 8 to figure 6 at the marked positions cor-responding to m=1, and 0.5. It shows that the exact interference extreme is shifted due to the absorption factor in equation (23). However, the wave-length range between m=1 to m=0.5 can be used to get an estimate of the steepness of the reflectance transition. As seen in figure 8 the dispersed re-fractive index has a direct influence on the width of the transition. Hence the steepness of the transition is actually limited by optical properties of the cermet.
31
1
10
1 10Wavelength ( m)
.3 .5 2 5 20 30
m = 4nd/
4ndf 0.18
m=1
m= 0.5
4ndf 0.30m= 0.5
(m
)
4n
d (
m)
Figure 8. 4nd, as function of wavelength using same optical constant of Ni-NiO composite with f = 0.18 and f = 0.30 at the specified film thickness d = 120nm.
2.6.2. Multilayer reflection The reflection of a multilayer consisting of several thin, flat and homogene-ous films with known optical constants can be calculated typically according to the matrix method [77]. However, for two- and three-layer thin films, it is more convenient to use a method named summation by extending equations used in the single layer case.
For double layers, there are three interfaces. Equation (17), (18) is extended to the third interface having amplitude reflection:
32
323 NN
NNr (24)
Similarly, the phase change in layer 2:
2222 dN (25)
Starting from the base layer, i.e. layer 2, the total reflected amplitude be-tween boundary 2 and 3 is (compared with equation (19)):
32
2
2
232
232
2 1 i
i
L errerrr (26)
Then the total reflection of double layer can then be obtained by replacing r2with the term rL2 in equation (19),
1
1
221
221
1 iL
iL
errerrr (27)
The above method can be extended to three layers or more. In principle, there is no limitation for layer number, and it is easy to use the Matlab pro-gram to carry out the calculation.
2.7. SputteringThe composite coatings can be produced by a variety of techniques on a large scale, such as chemical vapour deposition [33, 34], electroplating [6, 8], anodization [10, 11]. The costs of production of these two electrochemical methods of coatings are low, so they are widely used. However the emittance of both coatings is high. Magnetron sputtering technology developed in the 1970s [16] has also been used for selective coatings with improved solar porformance, in particular, lower emittance and less enviromental pollution than the electrochemical methods.
2.7.1. Sputtering process and dc planar magnetron sputtering The sputtering is a complicated process with several operation parameters being correlated. In a simple dc sputtering system the target serves as source material to be deposited. Argon is the most commonly used sputtering gas which serves as the medium where a glow discharge is initiated and sus-tained. Microscopically, argon ions in the discharge strike the target plate and eject neutral target atoms through energetic collisions. These atoms enter and pass through the discharge region and they eventually deposit on the substrate as growing film.
The magnetron was developed in the 1970s by placing magnets behind the target. The magnetic field concentrates and intensifies the plasma in the space close to the target, as a result of trapping of electrons near the target
33
surface. This leads to larger discharge current, increased sputter deposition rates, and reduced electron bombardment of substrates. Magnetron sputter-ing is presently the most widely, commercially practiced sputtering method. The chief reason for its success is the high deposition rates achieved. Figure 9 shows a schematic drawing of the process in dc planar magnetron sputter-ing.
N
S
S
N
N
STarget (cathode)
Magnets Magnets
IonsElectronsSputtered atoms
Substrate
Growing film
"B" Field "E" Field
Figure 9. Schematic drawing of process under dc planar magnetron sputtering.
2.7.2. Reactive sputtering Reactive sputtering is a method to deposit films which have a chemical composition different from that of the target, by adding a gas to react with the target material. The most common reactive gas is oxygen, usually mixed with argon to produce an oxide film.
An important phenomenon in reactive sputtering is the hysterisis behavior [78]. A general hysterisis process is illustrated by the change of system pres-sure with varying flow rate of the reactive gas, say oxygen. Typically, the curve measured by increasing oxygen flow does not coincide with that measured by decreasing oxygen flow. There is a critical point along the in-creasing oxygen flow with two target modes. When the oxygen flow is smaller than the critical oxygen flow, the target is metallic, therefore metallic atoms are sputtered and a compound is formed on the substrate. So this is called the ‘metallic mode’. When the oxygen flow is bigger than the critical, the metallic target surface is ‘poisoned’ by formation of oxide on its surface,
34
therefore it is called the ‘compound mode’. For metal targets, the target volt-age after the critical point is dropped due to higher secondary-electron emis-sion yields in the ‘compound mode’. The target voltage versus oxygen flow is used in our study to identify the critical oxygen flow. Figure 10 shows the target voltage for both increasing and decreasing oxygen flow. For increas-ing oxygen flow, target voltage shows a sudden drop at a critical oxygen flow down to lower voltage value. For decreasing oxygen flow, target volt-age keeps low as oxygen decreases pass the critical oxygen flow, and the voltage rises somewhere at oxygen flow which is lower than that critical value. Critical oxygen flow depends on operation conditions.
0.6
0.7
0.8
0.9
1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Arb
itrar
y ta
rget
vol
tage
(-V
)
Normalized oxygen flow
Critical oxygen flow
Compound mode
Metallicmode
If oxygen flow keeps below critical value
Figure 10. Scheme of target voltage versus oxygen flow rate showing hysteresis behavior. For increasing oxygen flow rate, target voltage shows sudden drop at criti-cal oxygen flow.
35
3. Experiment set-up and measurements
This chapter describes the experimental set-up, instruments and measuring techniques used in this work. Two kinds of dc planar magnetron sputtering units are used: a miniature copy of an industrial roll coater and a conven-tional laboratory sputtering unit. The instruments used include spectropho-tometers, X-ray diffractometer, X-ray photo-electron spectroscopy, scanning and transmission electron microscopy, and apparatus for film resistance and thickness determination.
3.1. System for sample preparation3.1.1. Roll-coater sputtering deposition system The roll-coater sputtering deposition system, called Rulle, used in the study is a miniature copy of an industrial roll-coater unit. It contains the main fea-tures of the industrial one used by the Sunstrip AB [13]. The operation con-ditions can be modified to sputter absorbing, or anti-reflection coatings. The following description is limited to the absorbing nickel-nickel oxide coat-ings.
The Rulle is equipped with an Edward HT10 diffusion pump with a pumping capacity of 3000 liter/s (nitrogen). A dc planar magnetron using a ring target is mounted above the sputtering zone. The sputtering power can be adjusted up to 10 kW. The gas flow rate is controlled by flow meters (The Bronkhorst High-Tech B.V. series) through computer software. The Rulle can be loaded with 200 m of aluminum band with a maximum width of 145 mm mounted on a spool in the vacuum chamber below the sputtering zone. The band passes the deposition zone from the left spool and is wound up on the spool on the right band side. The band speed can be adjusted between 0.4 to 1.2 m/min.
A nickel-chromium alloy (Ni80Cr20) target was used in the deposition. The reason for using an alloy is to avoid the difficulty with sputtering magnetic nickel. Pure argon (99.997%) and oxygen (99.997%) were used for the sput-tering. Figure 11 shows a schematic sketch of Rulle.
36
Magnetron
Direction of bandmovement
Spools
Oxygen nozzle
Figure 11. Schematic sketch of roll-coater sputter deposition system-Rulle.
3.1.2. Balzer sputtering unit The solar absorbing coating prepared in Rulle on the moving substrate has graded feature. However, in order to study and optimize three-layer coatings, it is necessary to prepare uniform layers with a controlled amount of metallic content and film thickness. These samples were prepared in a reactive dc magnetron sputtering unit Balzer UTT400.
The working chamber is mounted with turbo molecular pump TMU 521P. The chamber has a base pressure 10-7 mbar after backing. Argon, nitrogen, or oxygen can be sprayed into the chamber from different nozzles controlled by a mass flow controller (Bronkhorst Hi-Tec). The pressure in the chamber can be adjusted by a manual valve mounted between chamber and pump and is measured by a capacitance diaphragm gauge (model of CMH-01S07). The sample holder can be rotated to even the film thickness. There is a separate chamber designed to transfer a new substrate into the working chamber without disturbing the vacuum of working chamber.
The same kind of nickel-chromium alloy (Ni80Cr20) target was used for de-positing the absorbing layer. A pure aluminium (99.99%) target was used for depositing the anti-reflection layer. Figure 12 shows a photograph of the Balzer UTT400 sputtering unit.
37
Figure 12. A photograph of the Balzer UTT400 sputtering unit.
3.2. Optical characterization For determining the optical performance of the spectrally selective solar absorber, the total reflectance was measured in the solar and infrared wave-length ranges between 0.3 and 22 m, and extrapolated from 22 m to 100
m. These data were used to calculate solar absorptance and thermal emit-tance according to equation (5) and (6). For determining optical constants of sputtered film material, the total reflectance and transmittance was measured in the solar wavelength range from 0.3 to 2.5 m.
The instrument used for measurements in the solar spectral range was a Perkin-Elmer Lamda 900 UV/VIS/NIR double beam spectrophotometer. It is a dispersive type which uses a white light source and a few grating mono-chromators to create light with a specified wavelength. A detector collects the reflected or transmitted part of the light according to its setting and trans-fers it to the connected computer for further processing. The instrument is equipped with an integrating sphere to allow for measurements on the rough rolled aluminium sheets [24]. A Spectralon sample was used for reference.
The instrument used for measurements in the infrared range was a TENSOR 27 from Bruker optic. It is a type of Fourier Transform Infrared Reflectance (FTIR) spectrometer. In FTIR, the data are collected as an interference pat-tern by the connected computer and are processed with a Fast Fourier Trans-form technique to reform the spectrum. The Bruker Tensor 27 is equipped
38
with an integrating sphere coated with gold. The measuring wavelength range is from 2.5 to 22 m (450 cm-1). According to the standard EN 673, the thermal emittance should be calculated covering at least to 50 m [26]. The reflectance is therefore extrapolated to 100 m.
3.3. Material characterization 3.3.1. X-ray diffraction (XRD) X-ray diffraction is a non-destructive analysis tool to determine the crystal-line phase and structural properties. During the measurement, the interaction of X-rays with a periodic crystal lattice forms a diffraction pattern due to the X-ray wavelength being approximately of the same order as the unit cell constant. The diffracted beams from the sample atoms interfere construc-tively for certain angles and appear as intensity peaks. The diffraction of the beam is directly related to the crystal atomic arrangement which is described by Bragg’s law [79]:
sin2 BB dm (28)
where mB is an integer, the wavelength, dB the distance between parallel consecutive atomic planes and the angle between these planes and the in-cident beam. Different elements or compounds are then identified by analy-sis of the dB values that are characteristic for the crystalline phase.
XRD measurements for Ni-NiO thin films were performed by using a Sie-mens D5000 diffractometer operating with a grazing incidence angle of one degree in parallel beam geometry using CuK radiation (this wavelength is 1.540598 Å) and the diffraction pattern was analyzed between 30o to 90o
The effective grain size, i.e. the diameter of assumed spherical particle, was determined from the full width at half maximum (FWHM) of the X-ray dif-fraction peaks using the Scherrer equation [80]:
)cos(9.0)2(
LFWHM (29)
where L is the effective grain size. The FWHM was determined by fitting each peak by a Lorentzian curve. Broadening caused by internal strain was not taken into account.
39
3.3.2. Scanning electron microscopy (SEM) The Scanning Electron Microscope (SEM) uses electrons rather than light to form an image. In a scanning electron microscope, a beam of electrons is produced at the top of the microscope by heating a metallic filament. The electron beam follows a vertical path through the column of the microscope. It makes its way through electromagnetic lenses which focus and direct the beam down towards the sample. Once it hits the sample, other electrons (backscattered or secondary ) are ejected from the sample. Detectors collect the secondary or backscattered electrons, and convert them to a signal that issent to a monitor producing an image. The advantage of using the SEM in-cludes higher magnification, larger depth of focus, greater resolution, and ease of sample observation which makes the SEM one of the most heavily used instruments in many research areas today.
A high resolution LEO 1550 SEM with a field emission gun was used to examine grain size, surface and cross-sectional morphologies of different coatings of nickel-nickel oxide composite on aluminum substrates.
3.3.3. Transmission electron spectroscopy (TEM) A transmission electron microscopy (TEM) consists of an electron source (electron gun), condenser, objective, intermediate and projector lenses. The electron gun provides an electron beam. The parallel beam transmits through a thin sample settled above the objective lens and is focused in the back fo-cus plane.
TEM study was carried out using a JEOL2000 FXII operated at 200 kV and a field emission gun TECNAI F30 ST operated at 300 kV with a Gatan Im-aging Filter. An EDAX energy dispersive X-ray spectroscopy system was used to study the chemical composition of the sample. The cross sectional TEM specimen was prepared by gluing two pieces of the thin film sample face to face, followed by cutting, grinding, polishing, dimpling and finally by ion milling.
3.3.4. X-Ray Photoelectron Spectroscopy (XPS) X-Ray Photoelectron Spectroscopy (XPS), also known as Electron Spectros-copy for Chemical Analysis (ESCA), is an analysis technique used for ob-taining chemical information about the surfaces of solid materials. In XPS analysis, the sample is placed in an ultrahigh vacuum environment and ex-posed to a monochromatic X-ray source. X-ray excitation causes the emis-sion of photoelectrons from the atomic shells of the elements present on the surface. The energy of these electrons is characteristic of the element from
40
which they are emitted. By counting the number of electrons as a function of energy, a spectrum representative of the surface composition is obtained. The area under peaks in the spectrum is a measure of the relative amount of each element present, and the shape and position of the peaks reflect the chemical state for each element. XPS is a surface sensitive technique. Only those photoelectrons generated near the surface can escape and become available for detection.
The XPS study was performed with a SCIENTA ESCA 300 spectrometer using monochromatic AlK radiation (1487 eV). The instrument was pro-vided with a rotating anode and the X-rays were monochromated. The X-rays irradiated the sample at an incidence angle of about 45o. All XPS meas-urements were performed with a pass energy of 150 eV.
3.4. Other measurements 3.4.1. Resistance measurements The electrical dc conductivity of the sputtered films is deduced from film resistance Rf measured by using a square-probe which directly touches on the sample surface with a certain pressure. The probe consists of two parallel rows of four pin type contacts. The four contacts in a row are electrically connected and work as one pole, so the measured distance is kept the same.Figure 13 shows a photograph of such a probe, constructed in our laboratory.
The film conductivity is inversely proportional to the product of film resis-tance and thickness expressed by:
1)(1ff dR (30)
where df is the film thickness. In the probe, the current is not restricted be-tween the poles. Therefore the actual position of the probe on the sample influences the result. For this reason, measurements are carried out in the same relative positions on each sample. Moreover, only relative conductivity is used here. Typically, a set of composite samples with different metallic content, as well as a pure metallic sample, are prepared. The normalized dc conductivity of the film is defined as the ratio between the conductivities of the Ni-NiO composite film and a pure Ni film sputtered at the same power and argon flow. From equation (30), the normalized conductivity NC is ex-pressed in equation (31):
41
)/()(/ ffmmmf dRdRNC (31)
where Rm and dm is the film resistance and thickness of a pure metallic sam-ple respectively.
Figure 13. A photograph of the probe constructed in our laboratory for measurement of film resistance.
3.4.2. Thickness measurements The thickness of the films was measured with the mechanical stylus tech-nique, using a Tencor Alpha-step 200 instrument which incorporates profil-ing technology, high magnification image displays and digital data handling for profiling films. The film thickness on the aluminium substrate was esti-mated from the coating on glass prepared at the same time. In practice, a sharp film edge was obtained by attaching a piece of tape onto the substrate before sputtering.
42
4. Methodology
This chapter outlines methods developed in this work for the optimization of selective solar absorbing coatings prepared by Rulle. The first development is a method to be used to analyze graded film profiles. The second develop-ment is a method to determine optical constants of sputtered composite based on effective medium theory. The third part of our development is a method for the optimization of spectral selectivity using theoretical simula-tion.
4.1. Method for characterization of graded film profilesAbsorbing layers prepared in Rulle have a graded compositional profile which is achieved by moving the substrate through the sputtering zone as described in connection to figure 11 in section 3.1.1. An important feature of a moving substrate is that the process contribution along the zone (defined as X-direction) determines the depth profile composition of the deposited film (defined as Z-direction). For such a prepared sample, conventional and more elaborate ways to analyze the profile structure are based on electron or ion beam depth profiling spectroscopy. Obviously, these are expensive and time consuming methods which cannot be used as routine analysis. A new, simple method for preparing samples, directly revealing coating profile, is devel-oped in this thesis.
The non-uniform depth profile composition with a gradual increase of nickel oxide from the base to the surface in the film is a result of the inhomogene-ous distribution of oxygen in the sputtering zone. It provides a very simple way for investigating the film profile just to study a set of samples at fixed positions covering the zone. The moving band works as a shutter: in order to expose the samples for a controlled deposition, a rectangular opening was cut in the aluminum band, far away from the sputtering zone and then driven across the zone after pre-sputtering, just like a shutter is opened for coating. The set consists of twelve substrates of micro slide glass or rolled-aluminium sheet. They were initially screened by the aluminum band to prevent deposi-tion (shutter closed). Position number is according to the conventional count-ing film sub-layer, i.e. sample marked as No. 12 corresponds to the base of film, while No. 1 represents the top. The sample number of twelve was
43
mainly limited by the optical measurement requirements: the sample aper-ture of a Perkin-Elmer Lamda 900 has a diameter of 25 mm, the zone length is about twelve of such a diameter.
Meanwhile, graded coatings were also prepared at the same time by mount-ing substrates, the rolled aluminum sheet or glass, on the moving band. They were used to investigate the relation between zone distribution and spectral selectivity. Figure 14 is a schematic drawing showing the setup used in this method. The positions of each component can be roughly estimated from the drawing.
Substrate of fixed positions
Band (shutter closed)
Opening (shutter opened)
Substrate for graded sample
Band moving directionXZ
Figure 14. Schematic side view of the sputtering zone. Stationary samples were prepared to study the non-uniform zone property. A graded coating was obtained by transporting the substrates from left to right. The positions and size of the target, positions of argon and oxygen inlets are scaled according to the real positions in the sputtering unit.
44
4.2. Determination of optical constants of cermets The optical constants of sputtered materials are usually different from the optical constants of the corresponding bulk materials. Calculation of reflec-tance spectra consisting of sputtered thin film requires experimentally de-termined optical constants. Many methods for determination of optical con-stants have been reported [81-83]. Combined reflectance and transmittance measurement for determination of optical constants is one among them. The method involves two measurements: reflectance and transmittance at a near normal angle of incidence on a semi-transmitting sample. Traditionally, the optical constants are then deduced by the cross point between iso-reflectance and iso-transmittance on n-k plane for each wavelength, i.e. calculated point by point. The major problem can occur when R- and T-contours are almost parallel at their intersection. This will cause uncertainties.
Instead of calculating point by point at each wavelength, determination of the optical constants for the entire measured wavelength range is developed in our study for sputtered composite materials based on models of dielectric functions. This is especially useful for composite thin film where effective medium theory (typically Bruggeman model) describes the dielectric func-tion of the composite. Furthermore, the Drude model can be used for a me-tallic constituent and the Lorentz model for a dielectric constituent. Reflec-tance and transmittance spectra of modeled thin films on glass substrate can then be calculated and compared to the experimental reflectance and trans-mittance. The dielectric function of each component usually contains more than one variable. These variables, as well as volume fraction and film thickness, are then used as fitting parameters. Thus, the optical constants are deduced by fitting the theoretical reflectance and transmittance to the ex-perimental reflectance and transmittance in the whole experimental wave-length range.
Modern computer technique provides a way to make the fitting procedure easy. The software we used is ‘CODE’ (Coating Designer) [84] where simu-lated spectra can be fitted to experimental data, either choosing Auto Fitting or fit manually. Fitting results depend on models. For nickel-nickel oxide composite, a dielectric model was only built for the NiO constituent while literature data were used for nickel. For the Al-AlN composite, dielectric function models [85] for both aluminium and aluminium nitride were used to get better fitting. Figure 15 shows experimental reflectance and transmit-tance data (dots) of sputtered composite of Al-AlN compared to fitted curves.
45
0
0.1
0.2
0.3
0.4
0.5
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 2
Ref
lect
ance
, T
rans
mitt
ance
Wavelength ( m)
R: circle-experiment line--fitted
T: square-experiment line--fitted
Figure 15. Reflectance and transmittance spectra of sputtered composite of Al-AlN: dots indicate experimental data, and lines the fitting curves.
4.3. Theoretical optimization of spectral selectivity Theoretical optimization of spectrally selective reflectance is a good guide for choice of experimental parameters such as metallic content, film thick-ness and so on. The optimization can be based on either experimentally de-termined optical data obtained, as described in section 4.2 or can be calcu-lated from effective medium theory using handbook data for each constitu-ent. Reflectance calculation was usually carried out by ‘CODE’. Further calculation such as interference order, intensity attenuation as discussed in section 2.6 was carried out using the MatLab program.
Beside optical constants, a model of the layer structure is needed. The layer structure was specified as a two-dimensional stratified layer-stack between the half-spaces of air. Layer-stack consists of several sub-layers forming a thin film on an aluminium substrate with a typical thickness of one millime-ter. The number of sub-layers is not limited in principle.
Two types of layer-stack were mainly studied in this work. One is a three-layer structure [86-89]. Such three-layer coatings show a good spectral selec-tivity. The other type was a concentration graded absorbing profile layer [90-92] constructed by using an effective dielectric function with different metal-lic volume fraction on the depth position. The layer of thin film was then equally divided into a set of sub-layers with total defined thickness. In prac-
46
tice, 30 sub-layers were accurate enough for graded thin films with total thicknesses around 200 nm.
A reference spectrum is required in our optimization procedure. Results show that it is not possible to optimize using an ideal selective curve (step function). Two reflectance spectra have been created as references. One was based on a reflectance spectrum of measured nickel-nickel oxide tandem with less steep transition (marked as Ref.2). Another was based on theoreti-cal simulated reflectance spectrum of Al-AlN tandem with steep transition (marked as Ref.1). These two references were then further improved by curve smoothing and suppression of reflectance. They are shown in figure 16.
0
0.2
0.4
0.6
0.8
1
1 10
Ref
lect
ance
Wavelength ( m)
Ref. 10.9850.017
steep transition
Ref. 20.9700.033
less steeptransition
.3 .5 2 5 15
Ideal reflectance
Figure 16. Two reference reflectance spectra for simulation and fitting purpose.
47
5. Results and discussions
This chapter presents the most important results and discussion. It starts with tailoring of the film depth profile based on conductance distribution and optimization of the sputtering process of Rulle. It follows with material char-acterizations, determination of optical constants of the sputtering composite Ni-NiO. Finally, theoretical optimization of spectral selectivity of Ni-NiO coatings and a comparison with the experimental results are analyzed.
5.1. Tailoring of the film depth profile One early task in this thesis work was to find a simple way to determine and tailor film depth profile for obtaining good spectral selectivity. Realizing that a metal-dielectric film is conductive, and its conductivity is roughly propor-tional to the metallic content of the composite which can be deduced from resistance measurement. Thus early results are more or less related to NC(normalized conductivity) distribution along the sputtering zone for absorb-ing cermet coatings. More detailed analysis is then based on these early re-sults.
Figure 17 shows the NC distributions along the sputtering zone for absorbing coatings, which is displayed by sample positions (see in figure 14). Left y axis presents NC from reactive sputtering composite samples. Right y pre-sents NC from pure metallic sputtering, i.e. no oxygen involved. The curves are normalized to the highest conductivity achieved for pure metallic deposi-tion. The location of the target and oxygen spray nozzle is also shown in the figure. The conductivity of pure metallic film is not uniform over the zone. The highest conductivity is found below the target. This result indicates that the microstructure and the composition of the deposited film is not uniform along the sputtering zone.
NC distribution along the sputtering zone indicates the profile features for corresponding graded sample. Due to the position of the oxygen spray noz-zle, the oxygen content is high at the right side of the zone thus giving a dielectric coating for around 40% of the zone length (samples No. 1 to 5), i.e. the top part of graded film shows to be dielectric. From position 5 to 9,
48
the oxygen decreases and as a result, metallic content increases due to less oxidation. From position 10 to 12, though the oxygen decreases, the amount of sputtered metal decreases still more [93]. As a result, metallic content decreases again due to less sputtered metal. A graded film depth profile can be created by moving a substrate from left to right across the zone with a conductive layer in the base and a dielectric layer on the top.
0
0.1
0.2
0.3
0.4
0
0.2
0.4
0.6
0.8
1
24681012
NC
o
f Ni-N
iO fi
lm
Sample position
NC
of Ni film
Oxygen sprayposition
Target position
pure Ni film
Ni-NiO film
1
Figure 17. Normalized conductivity distributions along sputtering zone presented by sample position for pure metallic samples (right y-axis) and composite samples (left y-axis). The curves are normalized to the highest conductivity achieved for pure metallic deposition sample.
NC distribution can be used to tailor film depth profile by screening part of the sputtering zone. In the following example, the leftmost 100 mm and the rightmost 70 mm of the sputtering zone are taken away by screening these parts with a metal sheet. Figure 18 shows NC distributions obtained at two different oxygen flows where the curve marked ‘A’ has a little less oxygen (thus a little higher conductance) than the curve marked ‘B’. Now if the zone exposed to sputtering for case A is between 110 and 260 mm, the NC distri-bution is similar to that for B with an exposed zone between 100 and 250 mm. Therefore, similar film depth profiles can be obtained by the screening technique using an NC distribution analysis. As a result, the reflectance spec-tra are roughly the same in the two cases as shown in figure 19. So the depth profile specified by NC distribution can be used to optimize spectral selectiv-ity from a macroscopic point of view.
49
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
100 120 140 160 180 200 220 240 260
NC
of N
i-NiO
Sputtering zone for BSputtering zone for A
B
A
Sputtering zone ( mm )
Figure 18. NC distributions for two different oxygen flows. Case A has a little lower oxygen flow (thus a little higher NC) than case B. However, screening technique can select same NC part to be exposed for sputtering.
0
0.2
0.4
0.6
0.8
1
1 10
Ref
lect
ance
Wavelength ( m )
BA
.3 .5 2 5 25
Exposed zone: 100-250 mm
Exposed zone: 110-260 mm
Figure 19. The samples (without AR) were sputtered at different oxygen flows, but the profile could be similar by use of screening technique according to NC distribu-tions. Therefore, the reflectance spectra were similar.
50
5.2. Optimization sputtering process for RulleThe operation parameters for the sputtering process (e.g. sputtering power, flow rate, chamber pressure, location of gas nozzles, etc.) strongly depend on the specific machine. Among these parameters, oxygen flow and position of the oxygen nozzle are most important for the desired profile. This section will briefly summarize how to optimize oxygen flow for Rulle in our labora-tory.
The importance of oxygen flow is due to the following factors: Since certain portions of sputtered nickel particles (or clusters) should keep their metallic phase to form metal-dielectric composite, the reaction is not full oxidation, but partial oxidation; Metallic content must have a favorable distribution along the sput-tering zone as shown in figure 17, so the oxygen flow must have the desired distribution along the sputtering zone. This feature requires control of the concentration gradient of oxygen inside the sputtering zone;The reaction of partial oxidation occurs on the substrate surface with the desired distribution and form of intermixed micro-crystals of metal and dielectric.
The absorbing coating is quite tough by using one magnetron with asymmet-ric location of the oxygen nozzle. It is understandable that fluctuation of distribution cannot be avoided, so the same macroscopic process parameters can not reproduce exactly the same depth profile composition for each run. But the advantage of such a design is also obvious: it is simple, compact and low cost.
Several factors influence the distribution, for example, oxygen spray setting, vacuum condition, sputtering power, and gas flow rate, etc [94]. For a cho-sen sputtering power, argon flow rate and specified mechanical setting (e.g. nozzle positions, etc.), one dimensionless parameter roughly determines the zone conditions. It is called “the relative oxygen flow”, RO, and is defined as the ratio of used oxygen flow rate to the corresponding critical oxygen flow. The optimized RO is about 0.80 for different sets of parameters. Thus the most important process parameter, oxygen flow rate, can be determined by measuring a critical oxygen flow under a specified set of other parame-ters. The way to find a critical oxygen flow was described in section 2.7.2 and is further illustrated in figure 20 below. It shows the RO positions on an operation curve with certain sputtering power and argon flow under one fixed mechanical setting. Figure 21 shows the corresponding reflectance spectra.
51
It should be emphasized that this is only a ‘rough determination’. In fact, the critical oxygen flow is also influenced by the target surface condition. There-fore, sputtering condition cannot be exactly determined by RO. However, RO provides a good start value for optimizing the operational conditions.
0.6
0.7
0.8
0.9
1
0.5 0.6 0.7 0.8 0.9 1 1.1 1.2Normalized oxygen flow
Arb
itrar
y ta
rget
vol
tage
(-V
)
RO 0.63 0.8 0.93
Critical oxygen flow
1.07
Metallic modeCompound mode
Figure 20. Target voltage versus normalized oxygen flow rates for an optimized sputtering condition. Note that the curve is valid for an increased flow rate.
0
0.2
0.4
0.6
0.8
1
1 10
Ref
lect
ance
Wavelength ( m )
RO=0.8
RO=0.63
RO=0.93
RO=1.07
2.5 5.3 20
Figure 21. Reflectance spectra of graded absorbing coating samples (without AR) sputtered with sputtering power density 60 W/cm2, argon flow rate of 150 ml/min with different oxygen flow marked by RO.
52
One set of optimal conditions include: sputtering power density of 60 W/cm2, argon flow rate of 150 ml/min and RO at 0.8. Samples deposited under these optimal conditions were used to characterize material properties and optical constants, see section 5.3 and 5.4 which are a part of the opti-mized results.
5.3. Material characterization The main purpose of the materials characterization is to identify the structure properties of the composite material nickel-nickel oxide for optimized sput-tering conditions and interpret, for example, the macroscopic electrical con-ductance distribution. With X-ray diffraction (XRD) and X-ray photoelec-tron spectroscopy (XPS) it was possible to distinguish metallic nickel and nickel oxide. From SEM and TEM (top and cross sectional images), the composite structure on a nano-scale was revealed. The samples analysed with XRD and XPS were distributed in fixed positions in the zone as de-scribed in section 4.1.
5.3.1. X-ray diffraction (XRD)Figure 22 shows the X-ray diffractograms for a set of twelve samples, la-beled from 1 to 12 corresponding to the positions in the zone shown in figure 14. There is also a diffractogram of a graded coating prepared simultane-ously on a moving substrate. Coatings prepared on the right hand side of the zone, No. 1 to 5, consist of crystalline nickel oxide according to peak identi-fication. Sample No. 6 shows very weak peaks for nickel and nickel oxide indicating that this coating has an amorphous structure. In sample No. 7 to 12 nickel metal is recorded. This verifies the conductivity distribution shown in figure 17. Some traces of chromium oxide (Cr2O3) around the angle of 41 degrees are also detected.
The effective grain size obtained using equation 29 is shown in figure 23. The effective grain size of nickel oxide is larger than for the nickel particles. Both are in nano-scale.
53
0
200
400
600
800
1000
1200
1400
1600
30 40 50 60 70 80 90
Inte
nsity
(arb
itrar
y un
it)
2 (Degree)
(111)NiO (200)
NiO
(111)Ni
(220)NiO
(200)Ni
(311)NiO (222)
NiO
1
5
9
12
Graded
Figure 22. Diffractograms for a set of samples positioned in the zone. The numbers refer to the positions according to figure 14. The samples are deposited at optimal conditions. The diffractogram labeled “Graded” belongs to a sample moving across the zone at the same time as the fixed samples are deposited. The Miller indices of Ni and NiO are marked in the figure.
0
5
10
15
24681012
Gra
in si
ze (n
m)
Position of samples
NiO
Ni
1
Figure 23. Effective grain size of the same set of samples as in figure 22 calculated using the Sherrer formula in equation 29. The sizes of both nickel and nickel oxide are marked in position 6
54
5.3.2. X-ray photoelectron spectroscopy (XPS)Figure 24 shows the spectra of binding energy of nickel 2p3/2 in sample No. 1 to 12 and a pure nickel sample taken as a reference.
The spectra are marked with four binding energy positions of different nickel chemical states according to literature [95, 96]. The first is for metallic nickel (Ni 852.7 eV), the second for NiO (Ni2+, 854.3 eV), the third for Ni(OH)2 (Ni2+, 855.6 eV), and the fourth for Ni2O3 (Ni3+, 857.3 eV). It is difficult to identify different valence state in nickel compounds from meas-ured spectra except the pure metal state. Reported binding energies depend on conditions of measurement, preparation, etc. The metal peak can easily be identified from the reference nickel metal spectrum, and it also coincides with the peak position given from the literature.
The spectra for sample No. 7 to 12 clearly show metallic nickel content. The shape of the peaks between 854 to 858 eV is similar to the metallic reference spectrum. Concerning metallic nickel and nickel oxide the XPS results agree in general with the results from the conductivity and XRD measurements.
As it is known from the XRD measurements that samples No. 1 to 5 contain nickel oxide, it is probable that the peaks at 856.5 eV in No. 1 to 3 and 855 eV in No. 4 to 7 belong to NiO due to electrical charging of the surface, as a result, the peak of NiO shifted to higher energies. One is certain that position No. 1 to 5 are not a metallic phase.
55
0
2 104
4 104
6 104
8 104
1 105
1.2 105
850852854856858860
Inte
nsity
(arb
itrar
y)
Binding energy (eV)
L1
Ni 2P3/2
Ni-Met.
L12
L10
L7
L4
NiO
Ni(OH)2
Ni2O
3
Figure 24. XPS spectra of nickel 2p3/2 for a set of samples positioned in the zone. The numbers refer to the positions according to figure 14. The samples are deposited at optimal conditions. A sample deposited at the same conditions but without oxy-gen is shown as a reference. The peak positions of valence states of nickel and dif-ferent nickel compounds from the literature are marked.
5.3.3. SEM micrograph Figure 25 shows a SEM micrograph of a nickel oxide coating deposited at position No. 2 under the same optimized condition as the XRD and XPS samples. From the top view, the film surface exhibits a grainy structure. The grain size can be estimated to be in the range of 20 to 50 nm.
56
Figure 25. SEM image belonging to a sample positioned as No. 2 according to fig-ure 14 sputtered at optimal conditions.
5.3.4. TEM micrograph The TEM cross sectional image in figure 26 was taken from a graded coat-ing. The image shows clearly that the coating consists of two regions. The top sub-layer (B –layer according to the figure) with 100 nm thickness con-tains elongated crystalline grains which indicate a columnar structure to the surface plane. The average diameter of the columns is about 15 nm. The selected area electron diffraction pattern, SAED, shown by the inset b re-veals that this layer has fcc nickel oxide crystal structure with a cell parame-ter of 4.18 Å. The B-layer is deposited in the right part of the zone and con-sists of nickel oxide as confirmed by the XRD measurements.
The base sub-layer (C-layer) has an amorphous-like nano-crystal microstruc-ture. It can still be seen from the SAED pattern shown by inset c that the amorphous-like layer has fcc NiO crystal structure but with very fine grains. This sub-layer originates from deposition in the left part of the zone and consists of metallic nickel according to XRD measurements. The preparation of the TEM sample is done in air and the nickel has probably oxidized, which explains that no traces of nickel metal are found in the TEM prepared samples.
57
Figure 26. TEM image of a graded index coating obtained by moving substrate across the zone at optimized sputtering conditions. The diffraction patterns inserted in the image are taken with SAED. The pattern labeled b belongs to the sub-layer B and c belongs to sub-layer C. Both diffraction patterns are identified as NiO.
5.4. Optical constant of sputtered Ni-NiO compositeA set of optical constants of nickel-nickel oxide have been deduced corre-sponding to the zone position shown in figure 14. This section discusses the optical constants of three positions representing different properties.
Figure 27 shows the dispersion of the refractive index n and the extinction coefficient k obtained from a thin film corresponding to zone position No. 3. The figure also displays optical constants of bulk nickel oxide according to reference literature [61]. Both n and k rise towards short wavelength 0.3 mwhich is due to interband transition at ultraviolet wavelength of 0.29 m. Optical constant of sputtered film at this position shows the basic feature of oxide in visible range: k closes to zero, n is bigger than k. The refractive index of the sputtered dielectric phase is a little smaller than bulk nickel oxide. The rise of n and k towards 0.3 m is also smaller than that of bulk nickel oxide.
As noted, the dielectric constituent is not pure nickel oxide. The target is made of alloy, and reactive sputtering creates different chemical valence states of nickel and chromium oxides. The concept of nickel-nickel oxide composite in our work means that the optical properties of the sputtered ma-terial is replaced by nickel-nickel oxide composite with specified metallic
58
content since the dominating components are nickel and nickel oxide. This is the reason that we call it Ni-NiO composite
0
0.5
1
1.5
2
2.5
3
3.5
4
0
0.5
1
1.5
2
2.5
3
3.5
4
Wavelength m)
Ref
ract
ive
inde
x - n
Extinction coefficient - k
n - NiO
n - zone position 3at Rulle
k - NiO
k - zone position 3at Rulle
Figure 27. The dispersion of refractive index and extinction coefficient obtained from a sample corresponding to zone position No. 3. It is compared to the optical constants of bulk nickel oxide according to reference literature.
Figure 28 shows the dispersion of refractive index n and extinction coeffi-cient k obtained from a thin film at zone position No. 9 which corresponds to the highest metallic content of sputtered nickel-nickel oxide composite. Both the refractive index and the extinction coefficient increases with wavelength. The figure also displays optical constants of nickel-nickel oxide composite with nickel volume fraction f = 0.30 calculated according to Bruggeman and Maxwell-Garnett models as shown in figure 4, and 5. It shows that the opti-cal constants of the sputtered composite under optimal conditions are better described by the Bruggeman model than by the Maxwell-Garnett model. Thus it can be assumed that the composite consists of randomly intermixed particles of dielectric and metal. The optical property of the sputtered com-posite with the highest metallic content is approximately identical to nickel-nickel oxide with f = 0.30 determined by Bruggeman model.
Similar discussion is put on position No. 5b which presents the property between composite and dielectric. Figure 29 shows the result which is closer to the calculated results by the Bruggeman model with nickel volume frac-tion f = 0.10 than that of the Maxwell-Garnett model (f = 0.10). So the opti-cal property of the sputtered composite at position 5b is approximately iden-tical to nickel-nickel oxide with f = 0.10 determined by Bruggeman model.
59
0
0.5
1
1.5
2
2.5
3
3.5
4
0
0.5
1
1.5
2
2.5
3
3.5
4
0.3 0.4 0.5 0.6 0.7 0.80.9 1 2
Ref
ract
ive
inde
x - n
Extinction coefficient - k
Wavelength m)
n - No. 9 at Rulle
k - No. 9 at Rulle
Bruggeman model, f = 0.30
M-G model, f = 0.30
Bruggeman model, f = 0.30
M-G model, f = 0.30
Figure 28. The dispersion of refractive index n and extinction coefficient k obtained from a sample at zone position No. 9. It is compared to the optical constants of nickel-nickel oxide with f = 0.30 calculated according to the Bruggeman and the Maxwell-Garnett model respectively.
0
0.5
1
1.5
2
2.5
3
3.5
4
0
0.5
1
1.5
2
2.5
3
3.5
4
0.3 0.4 0.5 0.6 0.7 0.80.9 1 2
Ref
ract
ive
inde
x - n
Wavelength m)
n - No. 5b at Rulle
k - No. 5bat Rulle
Bruggeman model, f = 0.10
M-G model, f = 0.10
Bruggeman model, f = 0.10M-G model, f = 0.10
Extinction coefficient - k
Figure 29. The dispersion of refractive index n and extinction coefficient k obtained from a sample at zone position No. 5b. It is compared to the optical constants of nickel-nickel oxide with f = 0.10 calculated according to the Bruggeman model and the Maxwell-Garnett model respectively.
60
The rise of n and k towards short wavelength appears in a calculation based on EMT model. It does not appear in experimentally determined results. One possible reason is that the small particle destroys the structure for interband transition.
5.5. Theoretical simulation of selective surface for Ni-NiO composite as absorber
Theoretical simulation provides a way to understand important factors which influence spectral selectivity. The simulation should cover the wavelength range of the reflectance transition which influences both solar absorptance and thermal emittance. The optical constants for sputtered Ni-NiO composite are only experimentally determined between 0.3 to 2.5 m due to the lack of suitable infrared-transparent substrate. Therefore, the optical constants of Ni-NiO are calculated from Bruggeman formula for simulation which covers wavelength range from 0.3 m to 30 m. The optical constants are described in section 2.5. The simulation method is described in section 4.3. This sec-tion summarizes the main results from three-layer and graded layer coating structures.
5.5.1. Three-layer structure Study shows that a three-layer coating is a simple graded structure having a good optical performance [86-88]. Three-layer structure in this work refers to two absorbing layers i.e. base-layer (BL), middle-layer (ML), and an anti-reflection layer (AR).
The base-layer is the main solar absorbing layer in this structure. The reflec-tance spectrum shown in figure 6 in section 2.6.1 was fitted to reflectance spectrum Ref.1 for a single layer. It shows that steep transition can be ob-tained for a single Ni-NiO thin film on an aluminium substrate. The nickel volume fraction of the composite is of 0.18, the corresponding k is not very big in the solar range (figure 4). The solar absorptance for such a single layer tandem is 0.74 with very low thermal emittance of 0.02. Analysis shows (figure 7, 8) that the interference mechanism has an important contribution for steep reflectance transition.
Low solar absorptance for a single composite layer in figure 6 is due to ahigh front surface reflection loss which is a detrimental feature of a metal-dielectric composite absorber. The most effective way to suppress the front
61
surface reflection loss is by producing the absorber with a compositional gradient. In a three-layer structure, this can be done by adding a middle layer with a low metallic content on top of the base layer.
Simulation shows that the suppression is not satisfied for the case shown in figure 6 (i.e. f = 0.18). Figure 30 shows the reflectance for a base layer plus middle layer where ML is coated by 45 nm thick Ni-NiO composite with 0.04 nickel volume fraction. The main problem is that the maximum peak resulted from the base-layer is so strong that it is stretched out of the front surface reflectance of middle layer (marked by ‘R1: ML, air/f .04’). Thus, a strong interference mechanism plays an opposite role for middle layer sup-pression of the reflectance in solar range. The solar absorptance only in-creases from 0.74 for the single base layer to 0.82 when a middle layer is added.
0
0.2
0.4
0.6
0.8
1
1 10Wavelength ( m)
.3
Ref
lect
ance
R: BLf 0.18, d 120 nm
R: BL+MLf 0.04, d 45 nm
.5 2 5 20 30
R1: ML
air/f 0.04
Figure 30. The reflectance spectra corresponding to BL and BL+ML of nickel-nickel oxide composite on an aluminium substrate. Nickel volume fraction is 0.18 for BL, and 0.04 for ML. Layer thickness is 120 nm for BL, 45 nm for ML. The metallic content and film thickness of the middle layer is determined by the three-layer simulation fitted to Ref.1.
The problem with the above coating arises from the metallic content of cer-met used in the base layer being too low. When the nickel content in the BL increases, the interference intensity is reduced, as a result, interference maxima can be successfully suppressed by adding ML. Figure 31 shows the result for a three-layer structure. The nickel volume fraction of Ni-NiO in the base layer increases to 0.30. The optimal layer thickness reduces to 95 nm. The middle layer has the same volume fraction and film thickness as above. The interference maximum resulting from BL is suppressed below the front
62
surface reflectance. Consequently, the solar absorptance increases higher than that in the previous example: from 0.72 for BL to 0.87 for BL+ML.
0
0.2
0.4
0.6
0.8
1
1 10Wavelength ( m)
.3
Ref
lect
ance
R: BLf 0.30, d 95 nm
.5 2 5 20 30
R: BL+ MLf 0.04, d 45 nm
R1
: ML
air / f 0.04
Figure 31. The reflectance spectra corresponding to BL, BL+ML of Ni-NiO com-posite on aluminium substrate. Nickel volume fraction is 0.30 for BL, and 0.04 for ML. Layer thickness is 95 nm for BL, 45 nm for ML. The metallic contents and film thicknesses are determined by three-layer simulation fitted to Ref.2.
The influence of the metallic content in the BL also remains after adding the AR layer. For a base-layer with f = 0.18, the optimized solar absorptance is 0.922 with thermal emittance of 0.019 for the entire three-layer coatings. However if a base layer has f = 0.3, the optimized solar absorptance of the three-layer coating can reach as high as 0.965 by sacrificing a little thermal emittance: increasing from 0.019 to 0.041. This is shown in figure 32 in which both interference maxima are well suppressed under the level of the reflectance by front surface material AR. The third decimal in these values are not significant in the measured values, however it is given such that even small differences between the different types of layer structures can be evaluated.
63
0
0.2
0.4
0.6
0.8
1
1 10Wavelength ( m)
.3
Ref
lect
ance
.5 2 5 20 30
BL + MLf 0.04, d 45 nm
BL + ML+ARd 95 nm
R1: AR
air / AR
Figure 32. The reflectance spectra corresponding to BL+ML, and BL+ML+AR. Ni-NiO composite has volume fraction of 0.3 for BL, and 0.04 for ML. Layer thickness of AR is 95 nm. It is the simulation result by fitting to Ref.2.
To be clear, metallic content throughout this summary means the volume fraction in the cermet which determines its optical properties by Bruggeman model. It should distinguish the total amount of metallic composition in the absorber from the metallic content in cermet. Though the total amount of metallic composition can be increased by increasing film thickness for speci-fied metallic content of cermet, the optical performance is very different from that by increasing metallic content in cermet. Figure 33 shows two reflectance spectra from a single layer: the first is from f = 0.18, d = 158 nm, the second is from f = 0.30, d = 95 nm. The total amount of metallic compo-sition for these two coatings are same (0.30*95=0.18*158), however, they show different reflectance spectra. Increasing thickness (keeping same f=0.18) mainly shifts the interference position (compare to figure 6). On the other hand, increasing metallic content in cermet reduces interference inten-sity and reduces the steepness of reflectance transition as well.
The simulation reveals an important fact for solar absorbers that absorption should mainly rely on an intrinsic, but less on an interference mechanism. Thus, the parameter of the metallic content (volume fraction) has primary importance. If the chosen metallic content in cermet is not suitable for solar selectivity, it cannot be compensated by film thickness since its influence is through different ways.
64
0
0.2
0.4
0.6
0.8
1
1 10Wavelength ( m)
Ref
lect
ance
.3 .5 2 5 20 30
R: BLf 0.30, d 95 nm
R: BLf 0.18, d 158 nm
Figure 33. Two reflectance spectra of single layer: the first is from f = 0.18, d = 158 nm, the second is from f = 0.30, d = 95 nm. The total amount of metallic composi-tion for these two coatings are same, however, they show different reflectance spec-tra.
Thus, in a three-layer structure, the base-layer is the main solar absorbing layer in this structure. The middle-layer is used to link BL and AR (optical matching). The AR layer with low refractive index is to reduce front surface reflection loss.
5.5.2. Graded absorbing layer One purpose for simulation of graded absorbing layer is to investigate whether a further improvement of solar selectivity can be achieved com-pared to a three-layer structure. The parameter for graded compositional change in the absorbing layer is the metallic volume fraction. The gradual change in composition causes a change in the optical constant through the Bruggeman model. The total thickness of graded absorber is then divided into a set of 30 sub-layers of equal thickness. Each sub-layer has its own value of the volume fraction according to the defined depth profile character-ized by volume fraction.
Three different profiles were constructed for absorber as shown in figure 34 marked by ‘A: Linear’, ‘B: Concave-Convex’, ‘C: Concave’ where the x-axis is the normalized distance counted from the top of absorbing layer (i.e. the top corresponds to x=0), y-axis is the nickel volume fraction. Profiles A, B, and C are expressed by equations (32)-(34) respectively:
xy 6.0 (32)
65
1.0)))05.0(cos(*6.05.0( xexy (33) )1(*51 xey (34)
0
0.2
0.4
0.6
0.8
1
0 0.2 0.4 0.6 0.8 1Normalized distance: x
Nic
kel v
olum
e fr
actio
n :
f film topdirection
film basedirection
A: Linear
B: Concave-Convex
C: Concave
Figure 34. Three graded nickel volume fraction profiles according to equations (32)-(34).
The reflectance of a layer-stack consisting of AR layer on the top of such graded layer with an aluminium substrate (1 mm thick) was calculated and displayed in figure 35. It shows
Concave-Convex profile (165 nm thick) has best spectral selectivity which is similar to three-layer coating structure with f = 0.30 in BL; Linear profile (160 nm thick) also gives a good spectral selectivity. However appearance of interference maximum around 1.2 m indi-cates a similar problem discussed in figure 30; Concave profile (220 nm) has poor spectral selectivity. Too little metallic content in cermet in the layer causes interference maximum around 1.5 m which cannot be successfully suppressed.
Figure 36 compares the reflectance spectrum of graded concave-convex to that of three-layer coating (figure 32). The improvement by using graded is smoother in the wavelength range between 0.5 to 1.8 m where an interfer-ence minimum is observed for three-layer coatings. Otherwise the results are very similar so that three-layer is a good and simple coating structure to re-place continuous graded coatings.
66
0
0.2
0.4
0.6
0.8
1
1 10
Ref
lect
ance
Wavelength ( m).3 .5 2 5 20 30
R: Concave-Convex profile
Linear profile
R: Concave profile
Figure 35. Reflectance spectra of layer-stack consisting of AR on top of three graded absorbing layer with aluminium substrate, appearance of interference maxi-mum for linear and concave profile has the same reason discussed in figure 30.
0
0.2
0.4
0.6
0.8
1
1 10Wavelength ( m)
.3
Ref
lect
ance
BL--f .30, d 95 nmML--f .04, d 45 nmAR--d 95 nm
= 0.965, = 0.041
.5 2 5 20 30
Concave-Convexd 165 nmAR d 95 nm
= 0.968, = 0.046
Figure 36. Reflectance spectrum of three-layer (figure 32) compared to that of con-cave-convex graded absorbing film. It shows that three-layer is a good and simple coating structure to replace a continuous graded coating.
67
5.6. Experimental results of optimized selective surface for Ni-NiO composite as absorber Figure 37 shows a reflectance spectrum of a sample in which the absorber was prepared in Rulle on a moving substrate. As analyzed in section 5.3., it can be identified as two sub-layers: base layer contains nickel-nickel oxide with varied metallic content; middle is a dielectric sub-layer. Further, an AR layer was deposited on this sample with thickness of 80 nm in Balzer sput-tering unit. The reflectance spectrum is very similar to the simulated result with graded concave-convex absorber plus AR described in section 5.5.2.
Samples were also studied layer by layer prepared in the Balzer sputtering unit. Figure 38 shows a reflectance spectrum measured from one complete three-layer coating prepared in the Balzer sputtering unit. It is compared to the simulated result with three-layer structure described in section 5.5.1.
The similarity between the simulation and experimental samples in both figures indicate that the description in 5.5 represents the fundamental feature of Ni-NiO absorber.
0
0.2
0.4
0.6
0.8
1
1 10
Ref
lect
ance
Wavelength ( m).3 .5 2 5 3020
Graded absorbing layerprepared in RulleAR prepared in Balzer unit
Simulated withgraded concave-convexprofile + AR
Figure 37. Reflectance spectrum of optimized sample compared to the simulated result with graded concave-convex absorber plus AR shown in figure 36. The ab-sorber was prepared in Rulle using moving band technique. The AR layer (80 nm thick) was deposited in the Balzer sputtering unit.
68
0
0.2
0.4
0.6
0.8
1
1 10
Experimental spectrum:2 absorbing layer(Ni-NiO)+ AR
Simulated spectrum: two absorbing layer(Ni-NiO) + AR
Ref
lect
ance
Wavelength ( m).3 .5 2 5 3020
Figure 38. Reflectance spectrum of three-layer sample compared to simulated result of three-layer structure shown in figure 32. The sample was prepared in Balzer sput-tering unit
69
6. Conclusions
From the above discussions, we can conclude the following:
The design of the sputtering zone in Rulle is unique. Using asym-metric location of the oxygen nozzle, it is possible to form the de-sired metallic concentration distribution along the sputtering zone under optimized conditions. This distribution can be transferred into a graded film profile by moving the substrate, obtaining good spec-tral selectivity. It is a simple, efficient and low cost design.
The optimized conditions contain a set of process parameters, such as sputtering power, argon and oxygen flow etc. For specified me-chanical settings (such as locations of gas sprays, target and pump position etc.), RO, defined as the ratio of used oxygen flow to the corresponding critical oxygen flow, is a dimensionless parameter to control the zone specification. The optimal value is around 0.80 for the Rulle.
Under optimal conditions, the sputtered composite coatings show in-termixed micro-crystals of metal and dielectric in nano-scale. The optical constants exhibit broad absorption which can be described by the Bruggeman model with modified nickel oxide optical data.
Optimized zone shows properties with two main separated parts: the metallic composite part of varied nickel volume fraction and the di-electric part. The highest nickel volume fraction corresponds to 0.3 of simulated composite. Two parts of the sputtering zone can form a graded absorbing layer with the right ratio of base and middle layer by the moving substrate technique.
Distribution of normalized conductivity, NC, along the absorbing sputtering zone is a simple and good specification of zone property. Profile of graded film prepared by the moving substrate technique can be tailored according to NC distribution. XRD and XPS results show that NC distribution reflects the metallic concentration change.
70
Simulation reveals that absorption should mainly rely on the intrin-sic, but less on the interference mechanism. Thus, the parameter of the metallic content (volume fraction) has primary importance. If the chosen metallic content in cermet is not suitable for solar selectivity, it cannot be satisfactorily compensated for by film thickness.
For Ni-NiO composite, metallic content of 0.3 increases both the ex-tinction coefficient and the refractive index. The increase in refrac-tive index is then counteracting a sharp reflectance transition from the solar to the infrared wavelength range. Therefore the limitation in combining a sharp transition with a high solar absorption is caused by intrinsic optical properties of the cermet itself.
Simulation also shows that a graded profile gives a smoother reflec-tance spectrum between 0.5 and 1.8 m, whereas an interference minimum is observed for three-layer coatings. Otherwise the results are very similar so that three-layer coatings are a good and simple coating structure to replace continuous graded coatings.
Experimental results from both graded and three-layer coatings agree very well with simulation results, indicating that simulation explains the main features of the experiment. High solar absorptance of 0.97 can be achieved with thermal emittance of 0.05. This value is determined by the properties of used materials (Ni, NiO and Al2O3)and the method of production (sputtering conditions).
71
72
Acknowledgements
First, I sincerely thank Prof. Claes-Göran Granqvist for giving me the oppor-tunity to work at the division of Solid State Physics as a research engineer, and for encouraging me to pursue a Ph.D. degree in the group. He has set a high standard in both the research atmosphere and the social life in the divi-sion, which has been very enjoyable. His pioneer research work on solar energy materials is a valuable source of inspiration for my study.
I would like to thank my supervisor Prof. Ewa Wäckelgård who established the project in cooperation with the Swedish company, Sunstrip AB. I would like to thank her for all the support and trust that made my research work appreciated and fruitful. I would like to express my gratitude to my co-supervisor Prof. Carl-Gustaf Ribbing for valuable discussions and interaction during these years. Also, I would like to thank him for giving the course “Solid State Optics” that guided me to my current research field.
I am very grateful to Prof. Arne Roos for being an expert in optics and help-ing me with the English text of my manuscripts; to Prof. Gunnar Niklasson for his theoretical knowledge which helped me in solving many problems. The course “Optics of Inhomogeneous Materials” given by Prof. G. Niklas-son and A. Roos taught me how to treat the optical properties of composite materials.
I would also like to express my thanks to Stefan Gustavsson of the Sunstrip AB for sharing the experience of sputtering absorbers.
I would like to express my appreciation to all current and past members of the division of Solid Stated Physics for the friendly and comfortable atmos-phere. Particular thanks go to Bengt Götesson for all the technical assistance, Andris Azen, Jonas Backholm for the discussions on the sputtering tech-nique, and Kristina Gelin, Tobias Boström, Magdalena Lundh for the discus-sions on the solar thermal system…. I wish them all the best and success in their daily lives and future careers.
I would also like to thank Prof. Kai Siegbahn, a Noble prize winner, and Dr. Rein Maripuu, whom I had the opportunity to work with in the ESCA-LASER laboratory for eight years. There I learned the skill to treat basic
73
problems related to modern instruments, and vacuums in particular. My thanks also go to the International Science Program (ISP) who brought me to Sweden as a visiting scholar already in 1979, when China first opened its doors to the rest of the world.
My special thanks are denoted to my host family Harley Thomas and Eva von Wachenfeldt-Thomas, who gave us the feeling of being at home when my family first settled down in Sweden. The friendship and their support through the years has been invaluable.
Finally, I would like to express my heartfelt appreciation to my wife Gia, my sons and their families. Without their understanding and support, I would not have had a chance to complete my PhD. Study. Many thanks go to my lovely grandson who is an endless source of joy and happiness.
Uppsala, January, 2007
Shuxi Zhao
74
75
References
[1] C.G. Granqvist, Solar Energy Materials-Overview and some Examples,Applied Physics A, 52 (1991) 83-93.
[2] C.G. Granqvist and V. Wittwer, Materials for solar energy conversion,Solar Energy Materials & Solar cells, 54 (1998) 39-48.
[3] H.A. Ossenbrink, Future research and development in photovoltaics in The Future for Renewable Energy II, edited by EUREC Agency, pp25.
[4] Z. Yin, Development of solar thermal system in China, Solar Energy Materials and Solar Cells, 86 (2005) 427-442.
[5] G.A. Niklasson and C.G. Granqvist, Surfaces for Selective Absorption of Solar Energy: An Annotated Bibliography 1955-81, J. Mater. Sci. 18 (1983) 3475-3534.
[6] G.E. McDonald, Spectral reflectance properties of black chrome for use as a solar selective coatings, Solar Energy, 17 (1975) 119-122.
[7] G.B. Smith and A. Ignatiev, Black chromium-molybdenum: a new stable solar absorber, Solar Energy Materials, 4 (1981) 119-133.
[8] M. Koltun, G. Gukhman, A. Gavrilina, Stable selective coating ”black nickel” for solar collector surfaces, Solar Energy Materials and Solar Cells, 33 (1994) 41-44.
[9] E. Wäckelgård, Characterization of black nickel solar absorber coatings electroplated in a nickel chlorine aqueous solution, Solar Energy Materials and Solar Cells, 56 (1998) 35-44.
[10] Å. Andersson, O. Hunderi, C.G.Granqvist, Nickel pigmented anodic aluminium oxide for selective absorption of solar energy, J. Applied Physics, 51 (1980) 754-764.
76
[11] S.N. Kumar, L.K. Malhotra, and K.L. Chopra, Nickel pigmented ano-dized aluminium as solar absorbers, Solar Energy Materials, 7 (1983) 439-452.
[12] J.A. Thornton, J.L. Lamb, Sputter-deposited Pt-Al2O3 selective absorber coatings—High temperature applications, Thin Solid Films, 83 (1981) 377-385.
[13] E. Wäckelgård, G. Hultmark, Industrially sputtered solar absorber sur-face, Solar Energy Materials and Solar Cells, 54 (1998) 165-170.
[14] Q. Zhang, Direct current magnetron sputtered W-AlN cermet solar ab-sorber films, J. Vac. Sci. Technol. A, 15 (1997) 2842-2846.
[15] Q. Zhang, Stainless-steel-AlN cermet selective surfaces deposited by direct current magnetron sputtering technology, Solar Energy Materials and Solar Cells, 52 (1998) 95-106.
[16] Robert K. Waits, Planar Magnetron Sputtering: In Thin Film Process, edited by John L. Vossen, Werner Kern, Academic Press, New York, 1978, pp131-162.
[17] R.A. Buhrman and H.G. Craighead, Composite Film Selective-Absorbers, in Solar Materials Science: edited by Lawrence E. Murr, 1980, Academic Press, New York, pp298.
[18] M. Planck, Theory of the heat radiation, English transition of Theorie der Wärmestrahlung, 2nd edition (1913), 1959, Dover, New York.
[19] Frank J. Blatt, Roots of the Quantum Theory in Modern Physics, 1991, McGraw-Hill, Inc. New York, pp71.
[20] J.C. McVeigh, Solar Radiation in Sun Power, 1977, Pergamon Press, Oxford, pp 11-21.
[21] John A. Duffie and William A. Beckman, Solar radiation, Atmosphere attenuation of solar radiation in Solar Energy of Thermal Process, 1991, John Wiley & Sons, New York, pp7-10, and pp64-68.
[22] John A. Duffier and William A. Beckman, Flat-plate collector, in Solar Energy of Thermal Process, 1991, John Wiley & Sons, New York, pp250-329.
77
[23] O.P. Agnihotri and B.K. Gipta, Solar Selective Surfaces, A Wiley-Interscience publication, 1981, New York, pp88.
[24] A. Roos, Use of an integrating sphere in solar energy research, Solar Energy Materials and Solar Cells, 30 (1993) 77-94.
[25] W.L. Wolfe and G.J. Zissis, The infraded handbook, 1989, Environ-mental Research Institute of Michigan, Washinton DC.
[26] European Standard, EN 673, Thermal insulation of glazing calculation rules for determining the steady “U” value (thermal transmittance) of glaz-ing (1998).
[27] E. Randich and D.D. Allred, Chemically vapor-deposited ZrB2 as a selective solar absorber, Thin Solid Films, 83 (1981) 393-398.
[28] E. Randich and R.B. Pettit, Solar Selective properties and high tem-perature stability of CVD ZrB2, Solar Energy Materials, 5 (1981) 425-435.
[29] J.J. Cuomo, J.F. Ziegler, and J.M. Woodall, A new concept for solar energy thermal conversion, Applied Physics Letters, 26 (1975) 557-559.
[30] G.L. Harding and M.R. Lake, Sputter etched metal solar selective ab-sorbing surfaces for high temperature thermal collectors, Solar Energy Ma-terials, 5 (1981) 445-464.
[31] B. S. Richard and D. C. George, Optical reflectance and transmission of a textured surface, Thin Solid Films, 45 (1977) 19-29.
[32] F. Simonis, M. van der Leij, and C.J. Hoogendoorn, Physics of doped tin dioxide films for spectral-selective surfaces, Solar Energy Materials, 1 (1979), 221-231.
[33] B.O. Seraphin, Chemical vapor deposition of thin semiconductor films for solar energy conversion, Thin Solid Films, 39 (1976) 87-94.
[34] M. Janai, D.D. Allred, D.C. Booth and B.O. Seraphin, Optical proper-ties and structure of amorphous silicon films prepared by CVD, Solar En-ergy Materials, 1 (1979) 11-27.
[35] J.A. Thornton and J.L. Lamb, Thermal stability studies of sputter-deposited multilayer selective absorber coatings, Thin Solid Films, 96 (1982) 175-183.
78
[36] R.N. Schmidt and K.C. Park, High-temperature space-stable selective solar absorber coatings, Applied Optics, 4 (1965) 917-925.
[37] R.A. Buhrman and H.G. Craighead, Composite Film Selective-Absorbers, in Solar Materials Science, edited by Lawrence E. Murr, 1980, Academic Press, New York, pp277-292.
[38] G.A. Niklasson and C.G. Granqvist, Selectively Solar-Absorbing Sur-face Coatings: Optical Properties and Degradation, in Materials Science for Solar Energy Conversion Systems, edited by C.G. Granqvist, 1991, Perga-mon Press, New York, pp 70-105.
[39] E. Wäckelgård, G.A. Niklasson and C.G. Granqvist, Selectively Solar-Absorbing Coatings, in Solar Energy-the State of the Art, edited by Jeffrey Gordon, 2001, James & James Ltd, London, pp123-127.
[40] H.G. Graighead, R. Bartynski, R.A. Buhrman, L. Wojcik, A.J. Sievers, Metal/insulator composite selective absorbers, Solar Energy Materials, 1 (1979) 105-124.
[41] C. Sella, A. Kaba, S. Berthier, J. Lafait, Low cost selective absorber based on a Fe-Al2O3 cermet film, Solar Energy Materials, 16 (1987) 143-154.
[42] G. A. Niklasson and C. G. Granqvist, Optical properties and solar se-lectivity of co-evaporated Co-Al2O3 composite films, J. Appl. Phys. 55 (1984) 3382-3410.
[43] T.K. Vien, C. Sella, J. Lafait and S. Berthier, Pt-Al2O3 selective cermet coatings on superalloy substrates for photothermal conversion up to 600°C,Thin Solid Films, 126 (1985) 17-22.
[44] G.L. Harding, Z. Yin, S. Craig and S.P. Chow, Sputtered solar selective absorbing surfaces based on aluminium and stainless steel composite, Solar Energy Materials, 10 (1984) 187-207.
[45] Z. Yin and G.L. Harding, Optical properties of d.c. reactively sputtered thin films, Thin Solid Films, 120 (1984) 81-108.
[46] V. Teixeira, E. Sousa, M.F. Costa, C. Nunes, L. Rosa, M.J. Carvalho, M. Collares-Pereira, E. Roman, J. Gago , Spectrally selective coatings of Cr-Cr2O3 and Mo-Al2O3 for solar energy applications, Thin Solid Films, 392 (2001) 320-326.
79
[47] J.C.C. fan and S.A. Spura, Selective black absorbers using rf-sputtered Cr2O3/Cr cermet films, Applied Physics Letters, 30 (1977) 511-513.
[48] W. Graf, F. Brucker, M. Köhl, T. Tröscher, V. Wittwer and L. Herlitze, Development of large area sputtered solar absorber coatings, J. Non-Crystalline Solids, 218 (1997) 380-387.
[49] M. Okuyama, K. Furusawa, Y. Hamakawa, Ni cermet selective absorb-ers for solar photothermal conversion, Solar Energy, 22 (1979) 479-482.
[50] H. G. Craighead and R. A. Buhrman, Optical properties of selectively absorbing Ni/Al2O3 composite films, Applied Physics Letters, 31 (1977) 423-425.
[51] M. Sikkens, Properties of spectrally selective Ni-C films produced by reactive sputtering, Part 2. Optical properties, Solar Energy Materials, 6 (1982) 415-427.
[52] M. Mast, K. Gindele, M. Köhl, Ni/MgF2 cermet films as selective solar absorbers, Thin Solid Films, 126 (1985) 37-42.
[53] T. S. Sathiaraj, R. Thangaraj and O. P. Agnihotri, Ni-Al2O3 cermet solar absorbers by rf planar magnetron sputtering for high temperature applica-tions, Solar Energy Materials, 18 (1989) 343-356.
[54] M. Farooq, A.A. Green, M.G. Hutchins, High performance sputtered Ni:SiO2 composite solar absorber surfaces, Solar Energy Materials and So-lar Cells, 54 (1998) 67-73.
[55] M. Adsten, R. Joerger, K. Järrendahl, and E. Wäckelgård, Optical char-acterization of industrially sputtered nickel-nickel oxide solar selective sur-face, Solar Energy, 68 (2000) 325-328.
[56] L. Sauques, S. Fagnent, M-C.S. Catherine, C. Sella, Aluminium-aluminium nitride cermet films: preparation by co-sputtering and micro-structure, Surface & Coatings Technology, 102 (1998) 25-34.
[57] O.S. Heavens, Thin Film Optics, in Optical Properties of Thin Solid Films, 1954, Dover publications, Inc., New York, pp46-50.
[58] C.F. Klingshirn, Maxwell’s Equations and Photons, in Semiconductor Optics, 1997, Springer-Verlag, New York, pp.4-22.
80
[59] Frederick Wooten, Absorption and Dispersion, in Optical Properties of Solids, 1972, Academic Press, New York, pp42-55.
[60] G.A. Sawatzky and J.W. Allen, Magnitude and origin of the band gap in NiO, Physical Review Letters, 53 (1984) 2339-2342.
[61] N. Tsuda, K. Nasu, A. Yanase, K. Siratori, Electronic Conduction in Oxide, Springer-Verlag, New York, pp178.
[62] C.G. Ribbing and E. Wäckelgård, Reststrahlen bands as property indi-cators for materials in dielectric coatings, Thin Solid Films, 206 (1991) 312-317.
[63] H. Ehrenreich, H. R. Philipp, and D. J. Olechna, Optical Properties and Fermi Surface of Nickel, Physical Review, 131 (1963) 2469-2477.
[64] David W. Lynch and W.R. Hunter, Optical constants of metal, in Hand-book of Optical Constants of Solids, edited by Edward D. Palik, Academic Press, Inc., 1985, pp313-323.
[65] G.A. Niklasson, Optical properties of inhomogeneous two-component materials, in Materials Science for Solar Energy Conversion Systems: edited by C.G. Granqvist, 1991, Pergamon Press, New York, pp 7-43.
[66] D. E. Aspnes, Optical properties of thin films, Thin Solid Films, 89 (1982) 249-262.
[67] G.B. Smith, Dielectric constants for mixed media, J. Physics D, 10 (1977) L39-L42.
[68] G.B. Smith, The scope of effective medium theory for fine metal particle solar absorbers, Applied Physics Letters, 35 (1979) 668-670.
[69] C.G. Granqvist and O. Hunderi, Selective absorption of solar energy in ultrafine metal particles: model calculations, J. Applied Physics, 50 (1979) 1058-1065.
[70] C.G. Granqvist, Photothermal conversion of solar energy by gold cer-met coatings: applicability of the Bruggeman effective medium theory, J. Applied Physics, 50 (1979) 2916-2920.
[71] G. A. Niklasson, C. G. Granqvist, and O. Hunderi, Effective medium models for the optical properties of inhomogeneous materials, Applied Op-tics, 20 (1981) 26- 30.
81
[72] J.C. M. Garnett, Philos. Trans. R. Soc. London, 203 (1904) 385.
[73] D.A.G. Bruggeman, Berechnung verschiedener physikalischer konstan-ten von heterogenen substanzen, Ann. Phys. 24 (1935) 636-664.
[74] R.A. Buhrman and H.G. Craighead, Composite Film Selective-Absorbers, in Solar Materials Science, edited by Lawrence E. Murr, 1980, Academic Press, New York, pp290-292.
[75] D.B. Kushev, N.N. Zheleva, Y. Demakopoulou and D. Siapkas, A new method for the determination of the thickness, the optical constants and the relaxation time of weakly absorbing semiconducting thin films, Infrared Physics, 26 (1986) 385-393.
[76] J.C. Manifacier, J. Gasiot and J.P. Fillard, A simple method for the de-termination of the optical constants n, k and the thickness of a weakly ab-sorbing thin film, J. Physics E, 9 (1976) 1002-1004.
[77] M. Born and E. Wolf, Principles of Optics, 6th edition 1980, Pergamon Press, Oxford.
[78] W.D. Westwood, Reactive Sputtering, in Physics of Thin Films: edited by M.H. Francombe and J.L. Vossen, 1989, Academic Press, New York, pp1-22.
[79] A. Guinier, Diffraction of X-rays by Crystals, in X-ray Diffraction in Crystals, 1994, Dover Publications, Inc., New York, pp94-95.
[80] A. Guinier, Diffraction by an imperfect crystals lattice, in X-ray Dif-fraction in Crystals, 1994, Dover Publications, Inc., New York,pp159-166.
[81] P.O. Nilsson, Determination of optical constants from intensity meas-urements at normal incidence, Applied Optics, 7 (1968) 435-442.
[82] T.C. Paulick, Inversion of normal-incidence (R,T) measurements to obtain n+ik for thin film, Applied Optics, 25 (1986) 562-564.
[83] D. Pekker and L. Pekker, A method for determining thickness and opti-cal constants of absorbing thin films, Thin Solid Films, 425 (2003) 203-209.
[84] M. Theiss, Hard and Software for Optical Spectroscopy, 2002, Dr.-Bernhard-Klein-Str. 110, D-52078 Aachen, Germany.
82
[85] D.Y. Smith, E. Shiles, and M. Inokuti, The optical properties of metallic aluminium, in: Handbook of optical constants of solids I, edited by E.D. Palik, 1985, Academic Press, INC. New York, pp389.
[86] Q. Zhang, D.R. Mills, New cermet film structures with much improved selectivity for solar thermal applications, Applied Physics Letters, 60 (1992) 545-547.
[87] Q. Zhang, D.R. Mills, Very low-emittance solar selective surfaces using new film structures, J. Applied Physics, 72 (1992) 3013-3021.
[88] Q. Zhang, D.R. Mills, High solar performance selective surface using bi-sublayer cermet film structure, Solar Energy Materials and Solar Cells, 27(1992) 273-290.
[89] Q. Zhang, Metal-AlN cermet solar selective coatings deposited by direct magnetron sputtering technology, J. Physics D, 31 (1998) 355-362.
[90] I.T. Ritchie, B. Window, Applications of thin graded-index films to solar absorbers, Applied Optics, 16 (1977) 1438-1443.
[91] G.L. Harding, Alternative grading profile for sputtered solar selective surfaces, J. Vacuum Science Technology, 16 (1979) 2111-2113.
[92] B. Window, D. McKenzie, G. Harding, Selective absorber design, Solar Energy Materials, 2 (1980) 395-401.
[93] Rointan F. Bunshah etc., Coating deposition by sputtering, in Deposi-tion Technologies for Film and Coatings, edited by Rointan F etc., 1982, Noyes Publications, pp203.
[94] S. Berg, T. Larsson, C. Nender and H-O. Blom, Predicting thin-film stoichiometry in reactive sputtering, J. Applied Physics, 63 (1988) 887-891.
[95] J. Chastain, Handbook of X-ray Photoelectron Spectroscopy, 1992, Perkin-Elmer, Minnesota, pp84-85.
[96] A.G. Grosvenor, M.C. Biesinger, R.C. Smart and N.S. McIntyre, Newinterpretations of XPS spectra of nickel metal and oxides, Surface Science,600 (2006) 1771-1779.
83
Tillverkning av sputtrade solabsorbatorer med Ni-NiO komposit
Sol är en miljövänlig energikälla som både kan användas för att generera elektricitet och värme. In denna avhandling beskrives tillverkning av sputt-rade solabsorbatorytor. En termisk solfångare konverterar solstrålning till värme. Solabsorbatorn är hjärtat i en solfångare. Denna har till uppgift att konvertera den inkommande strålningen från solen till värme i vatten eller luft. Idealt är att så mycket av solenergi som möjligt ska absorberas samti-digt som den infångade värmen inte ska strålas ut till omgivningen igen. För att realisera detta används en så kallad den spektralt selektiv absorbatoryta. Denna avhandling behandlar några möjligheter att förbättra den optiska verkningsgraden på solabsorbatorer som ett sveskt företag tillverkar i sin anläggning för ytbeläggningar av absorbatorstrips. En minikopia av företa-gets ”roll coater” är för närvarande placerad på Ångströmlaboratoriet, Av-delningen för Fasta Tillståndets Fysik vid Institutionen för teknikvetenskaper Uppsala universtitet. Mini-roll coater- ”Rulle” är i sin väsentliga funktion lik den fullskaliga anläggningen.
Ytbeläggningstekniken som används i ”roll coater” är en så kallad plasma-process: material deponeras genom förstoftning av ett material i en ”target” som utsätts för plasmabombardemang. Targetmaterialet transporteras av plasmat till ytbeläggning på ett aluminiumband. Optimeringen gjordes ge-nom att undersöka och systematiskt variera processparametrarnas inverkan på ytornas optiska selektivitet. Denna bestämdes genom att solabsorptansen och den termiska emittansen beräknades från uppmätt optiska reflektans.
En ny metodik för optimeringen av ytan har utvecklats. Det selektiva ytskik-tet på absorbatorn är ca 150 nm tjockt och innehåller i huvudsak nickel och nickeloxid. Skiktet sputtras av ett plasma som tillföres ca 10 kW, från en nickel(80%)krom(20%)-target reaktivt i en processgas av argon och syre. Skiktet sputtras, så att kompositionen i djupled genom skiktet går från hög nickelhalt närmast substratet till ren nickeloxid vid frontytan.
Den gradvisa förändringen i materialkomposition ger upphov till en gradvis förändring i optiskt brytningsindex på ett sådant sätt att ytan erhåller hög absorptans (låg reflektans) för våglängder i solspektrum och låg emittans
84
(hög reflektans i infraröda våglängdsområdet). Target sitter centrerad i taket på zonen. Argon släpps in i vänstra kanten på target och syre bortom dess högra kant. Substratet rör sig från vänster till höger. I den högra delen depo-neras metall men i den vänstra delen kommer nicklet att reagera med syre så att ytan överst beläggs med nickeloxid. Placering av syredysan i detta läge bortom target är avgörande för att erhålla den önskade materialsammansätt-ningen i djupprofil från nickel till nickeloxid.
Vi har studerat ytskiktets egenskaper som funktion av olika processinställ-ningar för att etablera ett samband mellan den brytningsindexgradient som ger den önskade optiska selektiviteten och skiktets sammansättning i djup-profil av nickel och nickeloxid. I studien har vi utvecklat en metod för snabb och tillförlitlig utvärdering av relevanta fysikaliska egenskaper och deras koppling till processparametrar. Metoden går ut på att mäta elektrisk kon-duktans på prover som är placerade i fixa lägen i sputtringszonen. Proverna läggs ut så att de täcker hela zonen och avspeglar på så sätt i sin material-komposition de lokala processförutsättningarna i zonen och ger en karta över materialsammansättning i djupled hos ett prov som transporteras genom zonen.
Den ovan redovisade optimeringen är empirisk, men förfinad jämfört med den som användes vid den ursprungliga utvecklingen av ytan. Metoden base-ras på mätningar av de makroskopiska storheterna reflektans och elektrisk konduktans, och utifrån dessa har vi kvalitativt uppskattat skiktens innehåll av nickel och nickeloxid. I anslutning till detta och för att öka kunskapen om ytskiktet så har vi också gjort mätningar som på mikroskopisk nivå karaktä-riserat provernas struktur och sammansättning. Kristallstrukturen har under-sökts med röntgendiffraktion på prover utlagda i zonen och belagda enligt tre olika processinställningar. Vi har också studerat skikten med ESCA (Elec-tron Spectroscopy for Chemical Analysis) kemisk analys av elektronstruktur. Transmissionselektronmikroskopi (TEM) har utnyttjats för att fastställa kris-tallkornens storlek och förekomsten av dessa i nickel och nickeloxid.
För att ytterligare optimera absorptansen har vi bestämt de optiska konstan-terna för prover utlagda i zonen. De optiska konstanterna n och k (brytnings-index och extinktion) har bestämts från mätningar av reflektans och tran-smittans i våglängdsområdet 0,3 till 2,5 µm på ytskikt som deponerats på glassubstrat utlagda över zonen. De erhållna optiska konstanterna användes för att modellera ett multilagerskikt av fjorton dellager. Genom att göra en anpassning av tjocklekarna hos de ingående fjorton delskikten kunde vi för-bättra överensstämmelsen. I detta fall så minskade antalet skikt från fjorton till tre skikt som gav en bra anpassning. I detta fall finns ett metalliskt del-skikt, ett skikt som är mindre metalliskt plus AR skikt. För hela tre skikt, har
85
vi uppnått solabsorptansen =0,97 och termiska emittansen =0,05 båda in teoretisk simulering och experimentella prover.
Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 273
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 through theseries Digital Comprehensive Summaries of UppsalaDissertations from the Faculty of Science and Technology.(Prior to January, 2005, the series was published under thetitle “Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology”.)
Distribution: publications.uu.seurn:nbn:se:uu:diva-7530
ACTAUNIVERSITATISUPSALIENSISUPPSALA2007