thin film optical coatings

373

Thin Film Opt6. Thin Film Optical CoatingsWithin the scientific conception of the modernworld, thin film optical coatings can be inter-preted as one-dimensional photonic crystals. Ingeneral, they are composed of a sequence of sin-gle layers which consist of different transparentdielectrics with a thickness in the nanometer scaleaccording to the operation wavelength range.The major function of these photonic structuresis to adapt the properties of an optical surfaceto the needs of specific applications. By applica-tion of optical thin film coatings with optimizeddesigns, the spectral characteristics of a sur-face can be modified to practically any requiredtransfer function for a certain wavelength range.For example, the Fresnel reflection of a lens ora laser window can be suppressed for a broadwavelength range by depositing an antireflectivecoating containing only a few single layers. Onthe basis of a layer stack with alternating high-and low-refracting materials, high reflectancevalues up to 99.999% can be achieved for a cer-tain laser wavelength. In addition to these basicfunctions, optical coatings can realize a broadvariety of spectral filter characteristics accord-ing to even extremely sophisticated demands inmodern precision optics and laser technology.Moreover, recent developments in optical thinfilm technology provide the means to combineselected optical properties with other featuresconcerning, for instance, the thermal, mechan-ical or chemical stability of a surface. The latestprogress in ophthalmic coatings even includesthe integration of self-cleaning, photoactive oranti-fogging functions in antireflective coatingson glass.

As a consequence of this enormous flexibilityin adjusting the properties of functional surfaces,optical coatings can be found in nearly everyproduct and development of modern optic today.

6.1 Theory of Optical Coatings ..................... 374

6.2 Production of Optical Coatings ............... 3786.2.1 Thermal Evaporation .................... 3796.2.2 Ion Plating

and Ion-Assisted Deposition.......... 3816.2.3 Sputtering ................................... 3826.2.4 Ion-Beam Sputtering .................... 3846.2.5 Chemical Vapor Deposition (CVD) .... 3846.2.6 Other Methods ............................. 3866.2.7 Process Control

and Layer Thickness Determination 386

6.3 Quality Parametersof Optical Coatings................................ 388

6.4 Summary and Outlook........................... 391

References .................................................. 393

In order to keep pace with the rapid developmentof optical technology, innovations in the design,deposition processes and handling of optical coat-ings are some of the crucial factors. Also, highdemands in respect to precision and reproducibil-ity are imposed on the control of layer thicknessduring the production of the coating systems. Forcertain applications in fs lasers or optical mea-surement systems the individual layer thicknesshas to be controlled within the sub-nanometerscale, which can be only achieved on the ba-sis of advanced in situ monitoring techniquesof the growing layers. These skills have to becomplemented by extended knowledge of charac-terization, because optimization and marketing ofoptical coatings can only be performed on the ba-sis of reliable and standardized characterizationtechniques. The present chapter addresses thesemajor aspects of optical coatings and concentrateson the essential topics of optical coatings in theirtheoretical modeling, production processes, andquality control.

PartA

6

374 Part A Basic Principles and Materials

6.1 Theory of Optical Coatings

Photonic structures were present in nature long beforemankind. During their evolution, butterflies and otherinsects developed a variety of nanostructures on theirwings and bodies for camouflage, deterrence or attrac-tion [6.1]. The beginning of technical engineering ofoptical surfaces can be dated back to the Greek culture,when metal surfaces were polished to mirror quality.Tracing the history of optical coatings through the me-dieval times, the production of Venetian mirrors bycovering glass with an amalgam during the 16th cen-tury can be considered as the first application of opticalthin films. The first observation of an antireflective effectfor transparent thin films may be assigned to Fraun-hofer [6.2] who performed studies in aged glass surfacesand observed a reduction of reflectivity by the tarnishlayer in the year 1817. In the same century, DennisTaylor [6.3] investigated adapted etching techniques toachieve antireflective effects on different glass materialsused in optical devices during these times, and AugustinJean Fresnel (*1788; 1827) published his well-knownequations describing the optical function of a singleboundary. As a milestone in optical thin film technol-ogy, the conception of the FabryPerot theory [6.4] in1899 opened the way towards the theoretical descriptionof multilayer structures. This basic theory describes theinterference of partial waves reflected from two paralleloptical surfaces, which can be considered as the funda-mental element of all coating systems (Fig. 6.1): layersof different dielectric materials are deposited in a de-fined sequence on the surface of the optical component.At least two layer materials with different refractive in-dices have to be selected to adjust the spectral transfer

nH

nL

nT

Fig. 6.1 Basic structure of a thin film system: transpar-ent layers of at least two different materials (nH , nL ) aredeposited on a substrate with refractive index nT

function of the layer stack to the specification definedby the application. Besides transparency in the opera-tion wavelength range, major aspects for the choice ofthe coating materials include the contrast in the indicesof refraction as well as their chemical and mechanicalstability. The sequence of the materials and their corre-sponding thicknesses is often referred to as the design ofthe optical coating system and comprises all informationfor modeling of the spectral characteristics.

For the theoretical description of optical thin films,the application of the FabryPerot theory to a singlelayer (Fig. 6.2) can be considered as the starting point.In this approach a single layer of thickness d1 and re-fractive index n1 is formed by two boundaries betweentwo semi-infinite media with indices of refraction n0 andn2, respectively. In the classical model, the light inter-acting with this structure is described by the function ofa plane wave (point x, time t) with amplitude E0 andwave number k = 2/ incorporating the wavelength and frequency :

E(x, t) = E0 exp[i(kx t)] . (6.1)To calculate the spectral transfer function, this planewave is followed on its way through the single layer,and the contributions of the individual partial waves arecalculated and accumulated to form the total transmit-ted and reflected wave. At the first interface, a part ofthe incident plane wave is directly reflected with an am-plitude of A0 = E0r1 (order zero), where the reflectioncoefficient r1 can be calculated by the Fresnel formulae.The reminder of the wave is transmitted through the first

d

r1t1 t1' r2

t1 t1' r1' r22

t1 t1' r1'2r2

3

t1 t1 r1' r2 t1r1'2r2

2

t1 r2 t1 r1' r22 t1 r1'

2r23

t1 t2 t1 t2 r1' r2 t1 t2 r1'2r2

2

0

1

2

Fig. 6.2 Schematic path of a plane wave incident ontoa single layer with refractive index n1. The reflection andtransmission coefficients for the amplitudes at the interfacesm = 1, 2 are denoted by tm and rm , respectively. A primeadded to the coefficients indicates a transfer of the waveopposite to the direction of incidence

PartA

6.1

Thin Film Optical Coatings 6.1 Theory of Optical Coatings 375

boundary (coefficient t1), partially reflected back by thesecond boundary (coefficient r1) and then reaches via thefirst boundary the incident medium with an amplitudeA1 = E0t1r2t1. On its double path through the layer theconsidered partial wave also undergoes a phase shift 1which is dependent on the thickness d1 and the refractiveindex n1 of the layer:

1 = 2n1d1 cos 1

. (6.2)

In (6.2), the case of an arbitrary angle of incidence 1in the layer is taken into account, which has to be cal-culated by Snellius law. On the basis of the precedingconsiderations, the amplitude A1 of the first partial waveemerging from the layer can be written as:

A1 = t1t1r2 E0 exp(i21) . (6.3)In this notation t1 indicates the transmission coefficientfor waves passing the first boundary in the directionopposite to the incident wave. If this notation is appliedin the same sense to the part of the wave reflected backinto the layer by the first boundary (coefficient r 1), theamplitude A2 of this partial wave after back-reflectionby the second boundary and passing the first boundaryis given by:

A2 = t1t1r2 E0 exp(i21)r 1r2 exp(i21) . (6.4)In addition to the expression for the first-order ampli-tude, the second-order partial wave includes a factorr2r

1ex p(i21). Obviously, this additional factor ap-

pears for every further order of reflection, and therefore,the amplitudes Ak of a partial wave of the order k canbe determined:

Ak = t1t1r2 E0 exp(i21)[r 1r2 exp(i21)]k1 .(6.5)

To calculate the reflectance of the layer, all partial wavesemerging from the layer into the incident medium haveto be added to determine the total amplitude Atot of thereflected contributions. This sum can be expressed inclosed form, because the partial waves follow the ruleof a geometric expansion:

Atot =(

r1 + t1t1r2 exp(i2)

1r 1r2 exp(i2))

E0 . (6.6)

Finally, the reflection coefficient rS of the single layeris determined by the ratio between the total amplitudeof the reflected wave and the amplitude of the incomingwave

rs = AtotA0 =r1 +r2 exp(i2)1+r1r2 exp(i2) . (6.7)

In this expression, the coefficients r 1 and t1 have beenreplaced by using the following relations, which can bederived from the conditions of continuity valid for theamplitude at the boundaries:

t1t1 = (1r 1)(1r1) = (1r1)(1+r1) = 1r21 .

(6.8)

In view of the fact that only field intensities but notamplitudes can be measured, it is worthwhile to convertthe expression for the amplitude reflection coefficientinto the corresponding reflectivity value:

Rs = r21 +2r1r2 cos(21)+r22

1+2r1r2 cos(21)+r21r22. (6.9)

In principle, (6.9) provides all the means for the de-scription of the spectral transfer function for a singlelayer also including the cases of an arbitrary angle ofincidence and an absorbing layer material. For this gen-eral case, the refractive index has to be expressed in itscomplex form with n1 = n1 + ik1, where k1 denotes theextinction coefficient of the layer material, and the Fres-nel formulae as well as Snellius law have to be appliedin their complex form.

In order to get a deeper insight into the fundamen-tal operating principle of coatings the discussion ofa few special configurations is of particular interest. Inoptical thin film technology the layer thickness Di is of-ten expressed in units of quarter-wave optical thickness(QWOT) at the design wavelength Z:

Di = 4nidiZ

. (6.10)

If the layer thickness is an integer multiple of 1 QWOT,the phase shift of a wave with the design wavelengthtraveling perpendicularly through the layer correspondsto multiples of /2. Evaluating (6.9) for a 2 QWOTof any dielectric material on the basis of the Fresnelformulae for normal incidence results in:

Rs =(

r1 +r21+r1r2

)2,

with

r1 = n0 n1n0 +n1 , r2 =

n1 n2n1 +n2 ,

Rs =(

n0 n2n0 +n2

)2. (6.11)

Apparently, the expression for the reflectance is reducedto the Fresnel formulae for an interface between the inci-dent medium and the carrier medium, demonstrating that

PartA

6.1


the single 2 QWOT layer has no influence on the spec-tral transmittance of the structure. For the case of a layerwith a thickness of 1 QWOT at the design wavelength,(6.9) can be written in the form:

Rs =[

n2n0 n21n2n0 +n21

]2. (6.12)

The reflectance of the single layer on the substrate canbe totally suppressed if the condition n21 = n0n2 is ful-filled. For a further illustration of the effects typicalfor a single dielectric layer, corresponding reflectancespectra are depicted for different refractive indices inFig. 6.3. Considering the reflectance values at the de-sign wavelength Z the cases n21 = n2 (zero reflectance),n1 < n2 (reflectance lower than the reflectance of theuncoated substrate), n1 = n2 (reflectance equal to thereflectance of the uncoated substrate), and n1 > n2 (re-flectance higher than the reflectance of the uncoatedsubstrate) can be distinguished. As a function of thewavelength, a cyclic behavior of the spectra with a periodof one octave can be observed, which shows commonpoints of tangency at the reflectance level of the bare sub-strate. These points can be attributed to the wavelengthpositions, where the layer thickness is according to aneven multiple of QWOT resulting in the special con-dition described by (6.9). Thus, on one hand, a single1 QWOT layer of a material matched to the refractiveindex of the substrate can be employed as an antire-flective coating. On the other hand, a layer with a highindex of refraction is suitable for an enhancement ofthe reflectance to a certain limit given by the availabil-ity of high-refracting materials. Actually, single-layercoatings can still be found nowadays on many opti-cal components for laser applications, where only onewavelength has to be controlled.

In principle, the outlined theoretical approach forthe single layer can also be transferred to the calcula-tion of multilayer systems. However, considering theenormous number of partial waves, which increases ex-ponentially with the number of layers in a stack, theresulting equations become extremely complicated anddifficult to handle. For example, the expression for thereflectance of a three-layer system would already fillone printed page of this book. Therefore, the idea ofa formalism assigning a matrix to each layer of the de-sign was a substantial step forward to an understandingof thin film design. This so-called matrix formalism,which was born in the 1940s, can be deduced on thebasis of the boundary conditions of the electric andmagnetic field at the interfaces between the layers in

TiO2: n1= 2.25HfO2: n1= 1.95Al2O3:n1=1.62

MgF2: n1=1.38

Y2O3: n1=1.76R (%)

Wavelength (nm)400 1400

24211815129630

600 800 1000 1200

LZH

Fig. 6.3 Spectra calculated for a 1 QWOT single layer(Z = 1000 nm) of different materials on a the surface ofa Nd:YAG rod (n2 = 1.78, ambient refractive index n0 = 1).The refractive indices are typical values for e-beam depo-sition of the corresponding deposition materials: MgF2:n1 = 1.38, Al2O3: n1 = 1.62, Y2O3: n1 = 1.76, HfO2:n1 = 1.95, TiO2: n1 = 2.25. The line corresponds to thereflectivity of the uncoated substrate material

a system [6.58]. Besides the clear representation ofeach layer by a single matrix Mi , this approach offersa second major advantage which is related to a con-secutive multiplication of these single-layer matrices tocalculate the transfer function of a multilayer system.Thus, considering a stack, which is formed by a num-ber K of single layers with layer 1 located at the firstinterface in respect to the incoming wave, the transfermatrix MS of the entire stack can just be determined bythe following multiplication:

MS = M1 M2 M3 . . . Mi Mi+1 . . . MK . (6.13)The elements of the single-layer matrix Mi can be de-rived from the boundary conditions in the layer structurerelating the electric (Ei1) and magnetic (Hi1) fieldstrength at the front face to the field strength values (Eiand Hi ) at the rear face of the layer:(

Ei1Hi1

)= Mi

(EiHi

)

=(

cos ii

nisin i

ini sin i cos i

)(EiHi

). (6.14)

In this formalism, the matrix elements Mij contain therefractive index ni of the layer and the phase shift i ,which are exclusively parameters of the layer number i.For a calculation of the reflection coefficient rSK ofthe entire structure comprising also the substrate (index

PartA

6.1

Thin Film Optical Coatings 6.1 Theory of Optical Coatings 377

of refraction nT ) and the ambient medium (index ofrefraction n0), again the ratios of the amplitudes have tobe considered:

rSK = n0 M11 + in0nT M12 iM21 nT M22n0 M11 + in0nT M12 + iM21 +nT M22 (6.15)

In order to demonstrate the elegance of the matrix for-malism, the reflectance of a high-reflecting dielectricmirror will be considered in the following. The standarddesign of such a mirror is a periodical 1 QWOT stack(see also Fig. 6.1) of two coating materials with a high nHand a low index of refraction nL. For an efficient descrip-tion of the designs, a notation in capital letters, whichindicate a 1 QWOT layer of a certain layer material, isoften used in optical thin film technology. For example,a mirror stack with 11 layers in the described structure isrepresented by the sequence HLHLHLHLHLH (H: re-fractive index nH L: refractive index nL) or even morecondensed by (HL)5H. For the more general case of Nlayer pairs (design: (HL)N H) the matrix MDM of thedielectric mirror can be expressed by:

MDM =(

0 i/nHinH 0

)(0 i/nL

inL 0

)

(0 i/nH

inH 0

). . .

(0 i/nH

inH 0

)

(2N+1) matrices

(6.16)

Obviously, the single matrices for the 1 QWOT-layerreduce to simple expressions containing solely the re-fractive index of the materials for this special caseof normal incidence. Executing the matrix multipli-cation, applying (6.15) and calculating the reflectanceRDM = rDMrDM results in:

RDM 14n0nT n2NL

n2(N+1)H

. (6.17)

This expression is an approximation for a large numberN of QWOT-layer pairs, which is typical for practicalsystems involving 10 to 40 layer pairs. An interpreta-tion of this estimation (6.17) indicates that the contrast(nL/nH) between the indices of refraction governs thenumber of layer pairs necessary to achieve a defined re-flectance value. A high contrast does not only reducethe number of layer pairs, but also results in a more ex-tended reflection band in the spectrum of the QWOTstack (Fig. 6.4). In Fig. 6.5 the development of the re-flectance spectrum is illustrated for the QWOT stackdefined in Fig. 6.4 for different angles of incidence.

The shift of the spectra towards shorter wavelengthcan be explained on the basis of a reduction of thephase difference between the partial waves in the layerstructure. For arbitrary angles of incidence the spectraof p- and s-polarization have to be considered sepa-rately in the matrix formalism resulting in a broaderreflectance band for the s-polarization compared to thep-polarization. These effects have to be taken into ac-count if a dielectric mirror designed for normal incidenceis operated under arbitrary angles of incidence in anapplication.

Besides the described high-reflecting stacks, antire-flective coatings are often applied in laser technologyto reduce reflection losses of windows, laser rods, orlenses. Compared to the mirror designs, antireflective(AR) coatings are regularly built only of a few lay-ers, but often of more than two materials, especiallyto accomplish the demands of lowest residual reflec-tivity for a broad spectral range. For laser technologywith its dominant single-wavelength applications andchallenging specifications concerning the losses and thepower-handling capability of the coatings, single layersor two-layer AR coatings are mainly employed. On thebasis of two layers, a total suppression of the Fresnelreflection of most substrate materials can be accom-plished for one wavelength. Such coatings, which areoften called V-coatings according to the shape of their

Re(k)

Im(k)

UV

IR

k

Vacuum

Vis

Fig. 6.4 Reflectance spectra calculated for 1 QWOT stacks (Z =1064 nm) of different high-refracting materials in conjunction withSiO2 (nL = 1.46) as the low-index material on a quartz surface(nT = 1.46, refractive index of the ambient n0 = 1). The refractiveindices of the high-index layers are typical values for electron-beamdeposition of the corresponding deposition materials: Al2O3: nH =1.62 (green), CeO2: nH = 1.80 (blue), HfO2: nH = 1.95 (magenta),TiO2: nH = 2.25 (black). The number of layer pairs is kept constantat N = 6 for all depicted spectra

PartA

6.1


s

p

R (%)

Wavelength (nm)800 1600

1009080706050403020100

880 960 1040 1120 1200 1280 1360 1440 1520

45

LZH

Normalincidence

Fig. 6.5 Reflectance spectra calculated for 1 QWOT stack withparameters according to Fig. 6.4 and TiO2 (nH = 2.25) as thehigh-index material. Besides the spectrum for normal incidence,the reflectance is depicted for an angle of incidence of 45. Forarbitrary angles of incidence, s- and p-components have to bedistinguished

spectra, offer the advantage of lowest total thicknessproviding minimum losses and highest laser-induceddamage thresholds [6.9].

In regard to the important role of losses and stabil-ity of coating systems for laser technology, the designsare often also optimized with respect to the distributionof the electrical field strength in the structure. In viewof the fact that the matrices associate the field strengthvalues at different positions in the layer system, the cal-culation of the field strength distribution can be readilyaccomplished using the matrix method. Moreover, animplementation of the formalisms into modern computeralgorithms, which allow for the calculation of all trans-fer parameters including many other important aspectsin laser technology, such as absorbance, phase shift orgroup delay dispersion, is an uncomplicated task. In thecourse of the rapid development of computer technology,extensive software tools are available nowadays, whicheven allow for an inverse synthesis of multilayer designs.As input parameters of these modern optimization toolsthe desired spectral characteristics can be entered, andappropriate design solutions are developed on the basisof sophisticated algorithms. In summary, the challengesof optical thin film technology are no longer concen-trated on the theoretical modeling and the calculationof designs. Present and future problems are coupled tothe reproducible and precise production of coatings withhigh optical quality and environmental stability.

6.2 Production of Optical Coatings

Looking back, the production of optical coatings ex-hibits a long and successful technical history. Besides thealready mentioned early aging and etching experimentsof Fraunhofer and Taylor, in 1852 Grove described thesputtering effect, as it was later called, in a gas dischargeapparatus [6.10]. In 1857, Faraday vaporized gold wiresby high electrical currents and studied the optical prop-erties of the deposited material, although he was mainlyinterested in the finest particles [6.11]. Wright reportedin 1877 on the production of transparent metallic filmsby the electrical discharge in exhausted tubes [6.12]. In1884, Edison filed an application for a patent, whichdescribed thermal evaporation within a vacuum ves-sel as well as arc vaporization to plate one materialwith another and was granted in 1894 (Fig. 6.6) [6.13].In the following decades, a multitude of publicationsdocumented the development of industrial productionmethods for optical coatings, starting with single layersand resulting in increasingly complex multilayer sys-tems [6.14, 15]. Within the scope of the present article,a short overview of the most common production tech-

niques for optical thin film filters from the basics to someof the latest developments will be given. Further infor-mation about optical coating technology can be found ina variety of comprehensive technical books [6.9,1619].

Concerning the industrial production, there are fun-damental demands on the deposition process, althoughtheir priorities depend on the specific applications. Highreproducibility of the optical properties of the thin filmsis needed, as well as of the mechanical characteristics.Large areas have to be coated homogeneously, usinghigh deposition rates with simultaneous realization ofa precise layer thickness determination and termina-tion. Furthermore, a high level of automation, a shortprocess time, and inexpensive, nontoxic deposition ma-terials are economically relevant. With regard to the thinfilm material choice, specific requirements also haveto be fulfilled [6.20]. Dielectrics for use in multilayercoatings have to be transparent in the spectral region de-sired by the specific application. Therefore, besides anappropriate refractive index contrast, a low extinctioncoefficient has to be realized. In addition, an adequate

PartA

6.2

Thin Film Optical Coatings 6.2 Production of Optical Coatings 379

Fig. 6.6 Illustration from Edisons patent Art of PlatingOne Material with Another, granted in 1894 [6.13]. Theconductive coating material is vaporized within a vacuumchamber by a continuous arc between the electrodes (a). Forcoating mirrors, Edison placed glass plates on the inner sideof the apparatus. Furthermore, the patent describes thermalevaporation by resistive heating

mechanical and environmental stability must be ensured.Finally, the materials have to be technically controllablewithin the chosen deposition process, and have to ex-hibit the potential to form amorphous, compact thin filmmicrostructures. In the visible and near-infrared spec-tral region, these requirements are covered best by thematerial class of oxides. Common examples are thelow-refractive-index materials SiO2 or Al2O3 and thehigh-refractive-index metal oxides Ta2O5, Nb2O5, orTiO2. Going down to the ultraviolet and vacuum ultra-violet wavelength ranges, the oxides are complementedand successively replaced by fluorides, e.g., AlF3, MgF2,and LaF3. In contrast, the mid- and far-infrared spectralregion is the domain of materials such as Ge, Si, ZnSe,ZnS, and the radioactive ThF4, which is still in use,mainly for high power CO2 laser optics.

6.2.1 Thermal Evaporation

Within the most common deposition processes for op-tics, thin films are produced under vacuum conditions.In general, these processes are divided into two clas-

sification groups: on the one hand, the physical vapordeposition (PVD), comprising thermal vaporization aswell as sputtering, and, on the other hand, chemicalvapor deposition (CVD).

One of the basic ways to realize PVD is thermalevaporation of the thin film material by direct or indi-rect heating of the deposition material within a vacuumchamber. The evaporated material condenses on thesubstrates to be coated, which are usually located ona rotating spherical calotte above the evaporator. If theuniformity of the layer thickness distribution across thecalotte does not fulfil the requirements on the basis ofthe given geometrical set up, additional masks are used.Under vacuum conditions, the evaporation rate exhibitsa strong dependency on the temperature, and is mainlydetermined by the temperature-dependent equilibriumvapor pressure p of the evaporant. On the basis of theexperimental work of Hertz [6.21] and Knudsen [6.22],a theoretical description of the evaporation rate becameknown as the HertzKnudsen equation:

dNeAe dt

= e p p

2mkBT, (6.18)

with

dNe/dt number of evaporating atoms per time unit,Ae surface area of the evaporation source,e evaporation coefficient (sticking coefficient for

vapor atoms onto the surface),p equilibrium vapor pressure of the evaporant,p hydrostatic pressure of the evaporant within the

vacuum chamber,m atomic mass,kB Boltzmanns constant,T temperature.

The strong temperature dependency of the evaporationrate results from the equilibrium vapor pressure as anexponential function of the temperature

p = p0 eL0

kBT , (6.19)

where p0 is a constant factor and L0 is the latent heat ofevaporation per atom or molecule.

With the focus on technical aspects, a variety ofevaporation sources has been developed. In practice,unwanted contamination effects of the films with con-tact material from the evaporation source, as well asits significant corrosion, can occur due to chemical re-actions at the required high temperatures. Furthermore,high evaporation temperatures between some hundredto several thousand Kelvin have to be reached within theevaporation source.

PartA

6.2


An elementary, but not often used, method is thesublimation of a conducting material by direct currentheating. This technique is not applicable to many ma-terials; potential examples are C, Cr, Fe, Ni, Rh, andTi. An early developed and widely applied way to real-ize thermal evaporation is indirect heating of the coatingmaterial by a resistively heated reservoir. The charac-teristic shape of the reservoir gave this process variantits name: boat evaporation. Evaporation boats are madeof conducting high-temperature resistant materials, e.g.W, Mo, Ta, or C, in addition partially equipped witha ceramic insert (liner). Obviously, the maximum evap-oration temperature is limited by the melting point ofthe boat material. In a slightly modified alternative, theboat is reduced to a filament, wetted by molten material,which is kept in place by surface tension. However, dueto the thermal inertia of the boat and the molten coat-ing material, a nearly real-time rate control is difficultto realize. Furthermore, the melting can exhibit disad-vantageous spatter behavior, leading to coating defects.This effect can be partially suppressed by equipping theboats with perforated caps or chimney constructions.

Another technical possibility to melt conductive ma-terials is the application of induction heating. Whileinductive heaters are frequently used for large moltenmasses, e.g., in the field of crystal growth, this solutionis rarely used for the deposition of optical coatings.

Today, the most common thermal evaporationmethod in the industrial production of optical coat-ings is the direct heating of the coating material withina crucible by an electron beam (Fig. 6.7). The character-istic parameters of the electron beam, which is directedby magnetic fields into the water-cooled crucibles, arean acceleration voltage of 610 kV and currents of0.11.5 A. In order to achieve uniform evaporation, theelectron beam can be swept across the coating materialby deflection coils, optionally combined with a rotat-ing crucible. In addition, multi-crucible electron-beamsources allow the sequential evaporation of differentcoating materials within one process; otherwise the de-position plants are equipped with two or more sources.Compared to boat evaporation, the electron-beam heat-ing technique offers important advantages. Due to itshigh power densities, high-melting-point materials canbe evaporated and, furthermore, the cooled crucibleprevents contamination. With regard to rate control,electron-beam sources benefit from a low inertia in com-bination with the high reliability of todays technicalsolutions. However, it must be pointed out that highlocal temperatures may lead to decomposition, if chem-ical compounds are used as coating materials. In the case

Coatingmaterial

Crucible

Heatingvoltage

Accelerationvoltage

Deflectioncoils Electron beam

Anode Cathode

Fig. 6.7 Electron-beam evaporation source. The electronsare accelerated by a potential difference of several kV andare directed by a magnetic field into the water-cooled cru-cible. Additional deflection coils allow for the beam towrite adapted patterns on the material to achieve uniformevaporation

of oxides, an additional oxygen inlet can overcome thisproblem. In common deposition systems, stoichiometricfilms can be attained applying a typical oxygen partialpressure range of 1 to 3 102 Pa. Thermal evaporationin the presence of reactive gases such as oxygen, nitro-gen, or even fluorine is sometimes referred to as reactiveevaporation.

Regarding the layer properties, thin films pro-duced by thermal evaporation exhibit characteristicmicrostructures depending on the individual process pa-rameters. The typical thermal energies below 0.3 eV ofthe condensing particles result in packing densities lowerthan the density values of the corresponding bulk mater-ials. As a consequence of limited surface mobility of thecondensing particles and shadowing effects, a colum-nar growth containing microstructural voids is mostcommon for evaporated films [6.23, 24]. Derived fromexperimental studies, approved structure zone modelshave been developed that describe characteristic zonesdepending on the ratio between the substrate tempera-ture and the melting point of the coating material [6.25].Usually, the process parameters are optimized to achievemaximized packing densities to minimize voids, whichresults in the need for an additional substrate heating,typically in the range of 300 C. Besides increasing me-chanical stability, an enhanced packing density affectsthe optical layer properties, e.g., leading to higher refrac-tive indices in most cases. A major disadvantage of the

PartA

6.2


presence of microstructural voids in the thin films is theingress of moisture from the environmental atmosphere.On the one hand, the adsorbed water and bonded OHgroups result in optical absorption losses, especially inthe near- and mid-infrared (MIR) wavelength range. Onthe other hand, water adsorption influences the effectiverefractive index of the layer. As a result of the vary-ing water content, the spectral characteristic of a porousmultilayer interference filter shows a strong dependencyon environmental parameters such as temperature andhumidity.

Furthermore, the microstructure of the growing layeris affected by the surface structure of the substrate inall deposition processes for optical filters. As a highersurface roughness of the substrate increases the rough-ness of the layer surface, the substrate should be chosenin view of the application to avoid unacceptably highoptical scattering losses. In addition, cleanroom areasare indicated in optical thin film production, due to thefact that particle contamination results in serious layerdefects.

6.2.2 Ion Platingand Ion-Assisted Deposition

As discussed above, thin films produced by thermalevaporation processes exhibit a distinct microstructureas a consequence of the low kinetic energies of thecondensing molecules and atoms. Besides the thermalenergy deposited into the growing layer by substrateheating, additional energy deposit by ion impact wasquickly identified as a promising alternative. In the be-ginning, reactive evaporation processes were optimizedby the application of ionized oxygen [6.26,27]. In 1967a process concept was patented, in which a negativebias voltage accelerated positive ions onto the sub-strates [6.28]. Applied ion species included inert gasdirectly provided to maintain a discharge, as well asionized evaporated coating material. In the course offurther technological progress, this method was en-hanced and became known as the ion plating process,which found its way into the optical thin film indus-try in different variations. An established concept isthe reactive low-voltage ion plating (RLVIP) system,shown schematically in Fig. 6.8 [6.29]. In addition tothe conventional thermal evaporation configuration, anargon discharge is directed from a hot cathode sourcemounted on a chamber wall into the crucible of theelectron-beam evaporator, which acts as anode. As a con-sequence of the interaction with the argon plasma, theevaporated coating material is ionized and accelerated

together with argon ions towards the substrates by theself-bias voltage of the electrically insulated substrateholder. This negative self-bias voltage usually rangesbetween 5 to 30 V with respect to the plasma po-tential and depends on the plasma parameters as well asthe geometrical conditions. As a consequence of the ad-ditional energy input and the effective activation, thinfilms produced by reactive low-voltage ion plating, ex-hibit a compaction of the microstructure and optimizedstoichiometry, respectively.

Employing ions to enhance the optical and mechan-ical thin film properties as well as the process stability,ion-assisted deposition (IAD), also referred to as ion-beam-assisted deposition (IBAD), represents a widelyused process concept. As depicted in Fig. 6.9, the IADprocess is characterized by a separate ion source, in-tegrated in addition to the evaporation sources intothe process chamber. The ion beam is superimposedupon the flux of condensing particles and leads to thedesired densification of the growing layer. The domi-nant densification mechanism is attributed to momentumtransfer, which was verified by calculations on the basisof collision cascade models in accordance with experi-mental results [6.30]. However, excessive ion energies

Substrate holder drive

Vacuumchamber

Substrates

OxygenEvaporationsource(e-beam)

Argon

+

Coatingmaterial

Insulated substrate holder(neg. self-bias potential)

Plasma beam

Plasmasource

Fig. 6.8 Schematic diagram of a reactive low-voltage ion plating(RLVIP) system. An argon discharge is directed from a plasmasource into the crucible of an electron-beam evaporator. The ionizedcoating material and a part of the argon ions are accelerated towardsthe substrates by the self-bias voltage of the electrically insulatedsubstrate holder

PartA

6.2


may lead to ion-induced stoichiometry defects by thepreferential sputtering effect, which results from differ-ent sputtering efficiencies of the single components ofa compound [6.31]. In the case of the widely used metaloxides, the preferential sputtering of oxygen atoms re-sults in an oxygen deficiency and consequently, in anincrease of the optical absorption losses [6.32]. In prac-tice, ion energies covering the range of several 10 eVto several 100 eV are applied within typical IAD pro-cesses. The usual ion-source operating media are inertor reactive gases, in the latter case, in particular, oxy-gen to assist oxide coatings. Consequently, besides thedensification effect, the stoichiometry of oxides benefitsfrom the reactive ions, resulting in homogeneous layerswith reduced optical losses.

Concerning the IAD process environment, some cen-tral demands on the ion source can be stated. High ioncurrents at low ion energies are needed, and the ioncurrent distribution across the calotte has to be exten-sive and homogeneous. The ion source must exhibita high stability and reliability, even at long processtimes. An independent control of ion current and en-ergy within a wide parameter field is also advantageous.Furthermore, contamination of the growing layers bymaterial originating from the ion source has to beavoided. And finally, options for reactive gases that can

Substrate holder drive Vacuum chamberSubstrates

Ionsource

Evaporationsource (e-beam)

Fig. 6.9 Principle arrangement of an ion-assisted deposi-tion (IAD) process. Besides the electron-beam evaporators,an ion source is implemented in the vacuum chamber. Thesubstrates are exposed to the low-energy ion beam, whichresults in an additional energy transfer into the growinglayer

be used, low levels of maintenance, and low operatingcosts are relevant. Nowadays, a variety of gridded andgrid less ion sources, fulfilling nearly all these require-ments for standard IAD applications, are commerciallyavailable [6.33, 34].

In the following, the positive effects of ion-inducedlayer densification are illustrated briefly. With regard tothe microstructure, the aforementioned structure zonemodels for thermally evaporated films have been ex-tended to integrate the dense structures resulting fromion-assisted deposition processes [6.35]. As the mi-crostructural densification by the ions counteracts thevoid development of the columnar growth, the high op-tical losses attributed to adsorbed water can be preventedby applying IAD concepts. In Fig. 6.10, the optical lossesof a SiO2 single layer deposited with a thermal evapo-ration process are compared to those of a SiO2 singlelayer produced within an ion-assisted deposition pro-cess. The plotted extinction coefficients are calculatedfrom spectrophotometric transmittance and reflectancemeasurements and indicate optical losses in the rangebetween 10% and 20% for the thermally depositedlayer. Using spectrophotometric measurements, no wa-ter could be detected in the IAD coating. Besides thereduction of absorption losses, another positive effectof the dense and water-free microstructure is the drasticincrease in thermal spectral stability of the coatings. Ion-assisted deposition can reduce the relative wavelengthshift /(C) of the spectral characteristic from sev-eral hundred ppm/C for a thermally deposited coatingto a few ppm/C [6.36].

Furthermore, ion-assisted deposition offers the pos-sibility to tune the mechanical intrinsic stress of a coatingfrom tensile to compressive stress, which can improvethe layer adherence and the dimensional accuracy ofcritical optical components [6.3739]. And finally, theimplemented ion source enables options for substrateprecleaning process steps.

6.2.3 Sputtering

Sputtering represents an established and versatile classof energetic vacuum deposition processes, applying con-densing particles with significantly higher energies thanthose resulting from thermal evaporation methods. Itsbasic principle, derived from the early gas discharge ex-periments in the 19th century, is depicted in Fig. 6.11.Within a vacuum chamber at a pressure of about 1 Pa,a glow discharge is maintained by a direct-current (DC)voltage in the kV range between the anode, carryingthe substrates to be coated, and the cathode. The cath-

PartA

6.2


Extinction Coefficient k

Wavelength (m)2.25 4.00

0.05

0.04

0.03

0.02

0.01

0.002.50 2.75 3.00 3.25 3.50 3.75

Fig. 6.10 Comparison of the optical losses of SiO2 singlelayers: thermal evaporation versus ion-assisted deposition(absorption band of adsorbed water). The optical lossesare calculated from measured spectrophotometric transmit-tance and reflectance data

ode consists of the coating material which has to beelectrically conductive (target). In the discharge area,positive working gas ions, e.g., argon ions, are generatedby impact ionization and accelerated towards the target.Ions impinging on a solid surface, besides other interac-tions, eject atoms, molecules, and clusters by collisioncascades. This effect is called sputtering; its efficiencydepends on the target material and microstructure aswell as on the species, the energy, and the angle of in-cidence of the ions. Within the depicted DC dischargeprocess, the particles sputtered from the target condenseon the substrates and form a layer. As a consequenceof the high kinetic energies of the adatoms in a typicalrange of several eV (maximum energy of several tensof eV), sputtered coatings exhibit a higher density incomparison to evaporated thin films.

A way to overcome the DC sputtering limitation toconductive targets was found in the technology of radio-frequency (RF) sputtering. In a common set up, the DCelectrodes shown in Fig. 6.11 are converted to RF cou-pling electrodes. Choosing a geometrical arrangementin which the surface area of the substrate electrode ex-ceeds the area of the target electrode, a DC potential witha negative target electrode results from the self-bias ef-fect. Due to this DC potential, the sputtering occurs onthe target, and not on the substrate electrode [6.40].

Today, magnetron sputtering is an important sputter-ing technology for the industrial production of opticalcoatings. The term magnetron is derived from a mag-netic field, crossing the electric field of a DC discharge,

Darkspace shield

Vacuumchamber

CathodeTarget

SubstratesAnode

Working gas

IonCoatingmaterial

Plasma

Fig. 6.11 Schematic diagram of a direct-current sputteringprocess. A glow discharge is driven by a DC voltage in thekV range. Coating material is sputtered from the target onthe cathode and is deposited on the substrates positioned atthe anode

and confining the plasma within the region in front ofthe target (Fig. 6.12). Resulting from the increased de-gree of ionization, a significantly higher deposition rateprovides a reduction in process time. Furthermore, thelayer quality benefits from the less intensive interactionwith the plasma as well as from the reduced pressure,in the range of 101 Pa. However, the application of theDC magnetron source within a reactive process envi-ronment is linked to some limiting complications. Thereactive gas interacts with the target material and thegenerated compounds form an insulating layer on thetarget surface, an effect called target poisoning. Due tothe differences in sputtering efficiency of metals and, forexample, oxides or nitrides, the deposition rate decreasesstrongly while the target is poisoned. Also a modificationof the discharge conditions is attributed to the formationof an insulating layer, which can lead to arcs or even a ter-mination of the discharge. Besides the poisoning of thetarget cathode, the whole chamber, acting as an anode, iscovered successively by the insulating compound theso-called disappearing anode. To overcome these limita-tions, the magnetron process can be driven in a transitionmode between the metallic and the non-metallic mode.However, as this transition mode exhibits considerableinstability, a complex active process control has to be im-plemented to stabilize the relevant parameters, e.g., onthe basis of lambda probes for reactive oxygen [6.41].

A second technological approach is the sequentialprocessing concept, in which the magnetron sourceswork in the metallic mode. Separated from the mag-netrons, plasma sources provide a subsequent treatmentof deposited sublayers, consisting of a few monolayers,

PartA

6.2


Cathode Target

Substrates

Anode

Workinggas

Coating materialPlasma

Vacuum chamberBias voltage(optional)

N S N

Magneticfield

Fig. 6.12 Basic principle of the magnetron sputtering pro-cess. A magnetic field, crossing the electric field of a DCdischarge, confines the plasma within a region in front ofthe target. An optional bias voltage applied to the sub-strate holder allows for an additional energy input into thegrowing layer

with activated reactive gases. In technical implementa-tion, the magnetron and plasma sources are often locatedin a circular geometry with a fast rotating cylindricalsubstrate holder in its center [6.42, 43].

Important advances have been achieved on the ba-sis of dual-magnetron (also called twin-magnetron)configurations, applying medium-frequency voltagesin a typical range of 20100 kHz [6.44]. In a dual-magnetron arrangement, both target electrodes workingalternately as the cathode and anode, preventing elec-trostatic charging as well as the disappearing anodeeffect, mentioned above. Furthermore, improvementshave been made by operating magnetrons in pulse mode,since the limiting thermal load on the target is decreasedby this technique.

Nowadays, inline magnetron sputter systems arestandard for the large-area deposition in the applicationfields of architectural glass, photovoltaic, and displays,whereas sequential processing concepts in combinationwith dual magnetrons are currently finding their wayinto precision optics.

6.2.4 Ion-Beam Sputtering

Ion-beam sputtering (IBS) represents its own class ofsputtering deposition technology, applicable for thehighest-quality optics. The basic principle of an ion-beam sputtering process is illustrated schematically inFig. 6.13 [6.45]. Within a vacuum chamber, a separatedion source is directed to the target, which is consequentlynot in contact with the ion-generating plasma. The set

up provides a rotating substrate holder for the condens-ing thin films in the geometrically preferred direction ofthe sputtered particles. Under typical process conditions,argon ions with kinetic energies in the rage of 1 kV areemployed for sputtering, while an additional reactive-gas inlet provides an option for depositing compoundlayers from metallic targets. In some cases, a second ionsource is directed to the substrates, and allows for as-sistance of the growing layer (compare IAD) as well asprecleaning of the substrates.

In comparison to the described DC approach,alternating-current (AC) and magnetron sputteringprocesses, ion-beam sputtering offers a variety of ad-vantages. A low working pressure (reactive 102 Pa,nonreactive 103 Pa) in combination with the absenceof interactions between the substrates and the plasmaresult in high-quality thin films with minimum contam-ination and defects. However, the decisive factor for theexcellent optical and mechanical film quality in ion-beam sputtering is the high energy (up to 100 eV) of thelayer-forming particles. IBS coatings are dense, amor-phous, and suitable for ultra-low-loss components, astotal optical losses in the range of 1 ppm are achiev-able [6.46]. Furthermore, ion-beam sputtering is anextremely stable process, which allows for a high de-gree of automation. Also, energy and current density ofthe sputtering ions can be adjusted independently withina wide range.

Nevertheless, ion-beam sputtering is less often usedin the industrial mass production of optical coatings.This niche position results from distinct economic dis-advantages, namely a low deposition rate and technicaldifficulties in coating larger areas homogeneously. Com-pared to thermal evaporation or magnetron sputteringprocesses for precision optics, typical IBS depositionrates are 10 times lower (in the range 0.5 /s instead of0.5 nm/s). Major current market segments for ion beamsputtering include high-end precision optics for specialapplications, mainly in the research and science sector,e.g. complex chirped mirrors for femtosecond lasers aswell as next-generation lithography [6.47, 48].

6.2.5 Chemical Vapor Deposition (CVD)

Similar to the PVD processes described above, chem-ical vapor deposition offers a range of process variants.All CVD processes share the basic principle that thedeposited layer is a product of a chemical reaction ofgaseous reactants (precursors) [6.49]. This reaction isactivated within various process types by different kindsof energy input, covering thermal, plasma, and radiation-

PartA

6.2


Substrate holder

Gasinlet

Ionsource

Targets

Fig. 6.13 Schematic illustration of the ion-beam sputteringprocess. A separate ion source is directed to a target, thesputtered particles condense on the rotating substrates. Au-tomatically changeable targets allow for the production ofmultilayer coatings

induced CVD. Except for the coating material, all furtherreaction products have to be gaseous, as they have to beexhausted from the process chamber. Chemical vapordeposition is widely applied in hard coatings for wear-protection purposes, besides being the most widely useddeposition method in the semiconductor industry. In theproduction of optical coatings, CVD processes still playa minor role in comparison with PVD technologies.

Chemical vapor deposition distinguishes two layer-forming mechanisms. In heterogeneous nucleation, thecoating material is generated on the substrate surface. Inhomogeneous nucleation, the chemical reaction produc-ing the coating material takes place above the substratesurface. Subsequently, the reaction product diffuses tothe surface of the substrate, where the film is formed.Since coatings produced by homogeneous CVD pro-cesses in most cases are less dense and exhibit a lowerlayer quality than coatings by heterogeneous CVD pro-cesses, the latter variant is standard. However, someprocess concepts combine both types by applying a ho-mogeneous gas-phase reaction to produce a reactant forthe final heterogeneous layer-forming reaction. In allcases, complex gas dynamics have to be controlled byadaptation of the gas inlets, and a balanced flow andpressure control.

The classic way to perform chemical vapor deposi-tion is by thermal activation of the reactions by heatingof the substrates. Most of these reactions require hightemperatures, typically exceeding 450 C, which areincompatible with optical coatings on substrates forprecision optics or on plastics. As an example, in thesemiconductor industry the pyrolysis of silane is used todeposit silicon (g: gaseous, s: solid):

SiH4(g) 600700C Si(s)+2H2(g) . (6.20)

Applying additional oxygen, silica coatings can be pro-duced:

SiH4(g)+O2(g) 450C SiO2(s)+2H2(g) . (6.21)

Thermally activated chemical vapor deposition pro-cesses are driven at atmospheric pressure (APCVD)or, more commonly today, in low-pressure reactorsat a typical pressure range of 101 103 Pa (LPCVD).Atmospheric-pressure CVD offers higher depositionrates than LPCVD, which, in contrast, provides higheruniformity and a better covering of nonplanar geomet-rical shapes.

Another way of energy input is used in plasma-enhanced chemical vapor deposition (PECVD), a pro-cess variant rapidly gaining importance in manyapplications [6.50]. The PECVD technique offers the ad-vantage of much lower process temperatures than thoseneeded in thermally activated CVD. As a result of theapplied DC, RF or microwave plasma excitation, thegaseous precursors can react while the substrate temper-ature stays below 300 C. In Fig. 6.14, a basic RF-drivenPECVD reactor is shown.

A special pulsed PECVD concept is based onmicrowave impulses and is therefore called plasmaimpulse CVD (PICVD) [6.51]. The PICVD processoffers a feasible technical solution to apply coatingson complex-shaped surfaces, in particular on the innersurfaces of tubes, reflectors, or even bottles. Processpressures in the range of a few 102 Pa result in shortpumping times. On the other hand, the pumping timeis minimized by the smallest reactors, as, in the case ofinner surfaces coatings, the process chamber is often ge-ometrically defined by the substrate itself. Examples ofthis modularly scalable PICVD concept are cold lightmirrors on halogen-lamp reflectors as well as barriercoatings on the inner walls of polyethylene (PET) bottlesor of ampoules for pharmaceutical packaging. As everymicrowave pulse deposits a sublayer of a reproduciblethickness in a few milliseconds, a precise thickness con-trol can be realized by pulse counting. Consequently,

PartA

6.2


Plasma

Vacuumchamber

Electrode

SubstratesHeater

ExhaustPre-cursors

Pre-cursors

RF

Fig. 6.14 Basic reactor set up for RF-excited plasma-enhanced chemical vapor deposition (PECVD). The plottedsubstrate heater is an option to assist the layer-formingmechanism. Without the RF electrode the CVD process canbe driven as a non-PECVD process by thermal activationonly

the deposition rate, which can exceed 10 nm/s for metaloxides, is determined by the pulse repetition rate andthe conditions of the chemical reaction. For example,SiO2 coatings are derived from a SiCl4 precursor by theoverall reaction:

SiCl4(g)+O2(g)microwave pulses plasma SiO2(s)+2Cl2(g) .

(6.22)

In the process chamber, this reaction is divided intotwo parts. In a first homogeneous gas-phase reac-tion SiO is produced (SiCl4 + (1/2)O2 SiO+2Cl2),which diffuses to the surface and is oxidized in the finalheterogeneous reaction (SiO+ (1/2)O2 SiO2).

In an alternative approach for activation, the CVDreactions can be induced by laser radiation [laser CVD(LCVD)]. This process variant is based either on a py-rolytic, i. e., thermal, interaction or direct photolyticactivation [6.52]. In contrast to the other mentionedCVD techniques, LCVD is normally not applied foroptical coatings. Laser CVD is primarily suitable formicro-technical applications, as structures in the rangeof several 10 microns can be deposited selectively, e.g.,for circuit or mask repair in the field of electronicsand lithography, or for the creation of three-dimensionalobjects such as carbon fibres [6.53, 54].

Generally, chemical vapor deposition plays a minorrole compared with physical vapor deposition in opticalthin film technology. However, with regard to coatingson complex surface geometries, CVD processes are su-perior to PVD technologies, as diffusion is an undirected

effect. Furthermore, the precursor chemistry offers thepossibility to obtain intermediate refractive indices, e.g.for the production of rugate filters, by varying the mix-ture [6.55]. In this connection, it has to be mentionedthat these chemicals are mostly toxic and often difficultto handle.

6.2.6 Other Methods

Besides the introduced common deposition techniquesin the fields of precision, laser, and consumer optics,a variety of processes for special applications is availabletoday. Analogous to the laser CVD mentioned above,a laser PVD process has been developed, which has be-come known as pulsed laser deposition (PLD). In PLD,target material is ablated by the laser pulses within a vac-uum chamber and is subsequently deposited onto thesubstrates positioned in the plasma plume. The PLDprocess is mainly employed in laboratory-scale systemsand offers superior reproduction of the target stoichiom-etry in the produced thin films of organic and inorganiccompounds [6.56]. However, up-scaling of this tech-nique to large coating areas and high deposition rates isproblematic and still too cost-intensive.

A more common deposition method that is used insome application fields of optical coatings is the solgelprocess. Starting from a colloidal solution (sol), in mostcases of silicon or metal alkoxides, a transition leads toa gel phase, which is the base material for the coating.In the case of dip-coating, the layer is formed by dip-ping the substrate into the solution. In the spin-coatingprocess, the layer is produced by spreading the solutionon the spinning substrate to achieve a uniform thicknessdistribution, which depends on material-specific param-eters and the speed of rotation. Furthermore, the solutioncan also be applied by spraying, in a laminar-flow-coating process, using capillary forces, or by printingtechniques. A densification of the initially porous solgel films is often done by baking at temperatures ofup to 1000 C. The domain of the solgel technologyin optics is the production of large-area antireflectioncoatings, especially for high-power lasers [6.57].

6.2.7 Process Controland Layer Thickness Determination

With regard to the nanometer precision demanded in op-tical coating production, process control plays a key rolein modern production environments. Recent trends to re-alize advanced process control strategies are often basedon in situ monitoring techniques of important deposition

PartA

6.2


parameters and coating properties. The central require-ment for the production of optical interference filters isa precise thickness control of the growing layers.

In a straightforward approach, the layers can be ter-minated by time, if the deposition rate of the processis stable enough to fulfil the required thickness accu-racy, e.g., in some sputtering processes. An exceptionalposition is taken by plasma impulse CVD, where thethickness can be determined by counting microwavepulses.

A widespread technical solution to measure theactual deposition rate and the layer thickness in theproduction of thin films is a quartz crystal monitoringsystem [6.58]. Since the resonance frequency of the os-cillating quartz varies with the material deposited on itssurface, the rate can be determined from the change inoscillation frequency on the basis of material-specificconstants (e.g., specific weights). As a consequence ofa position of the quartz crystal monitor differing fromthe substrate holder location, another calibration factor,called the tooling factor, has to be considered for an ac-curate determination of the deposition rate. In particular,these factors depend not only on coating materials andthe process parameters, but often also on the layer thick-ness as a result of the developing microstructure in thegrowing thin films.

As indicated above, the performance of a multi-layer filter depends on optical thicknesses, which arethe products of the physical thicknesses and the re-fractive indices of the included single layers. Thus,monitoring methods that provide in situ analysis ofthe optical thickness of the growing layer are oftensuperior to systems monitoring mass deposition, espe-cially under the conditions of small variations in theoptical properties. The natural way to determine opti-cal thickness is a direct optical access in the form ofa transmittance or reflectance measurement [6.59]. Thetheoretical analysis of a single layer reveals a cyclicbehavior of the transmittance or reflectance at a fixedwavelength with increasing optical thickness. This ef-fect can be employed to determine the actual opticaland, moreover, the actual physical thickness of the de-posited layer on the basis of the known optical constants.Especially, for desired single-layer thickness values ofinteger multiples of QWOT at the monitoring wave-length, the termination points are extremal values of themeasured curves. Since certain wavelengths are distin-guished, the monitoring wavelength has to be chosen inaccordance with the coating design, and often multi-wavelength monitoring systems are applied. In mostcases, the measurements are performed on a centred

test substrate, which remains stationary, in contrast tothe products to be coated.

In the course of the rapid development of com-puter and spectrometer technology, advanced opticalbroadband monitoring systems have been implementedin deposition process environments [6.60]. These tech-niques allow for direct broadband in situ transmittancemeasurements on the moving substrates for each ro-tation of the calotte. For example, a spectral rangeof 3501060 nm can be covered by a fibre-coupledcharge-coupled device (CCD) spectrometer system incombination with a free-beam measurement set up insidethe vacuum chamber (Fig. 6.15). Each in situ spec-trum recorded can be calibrated, because three singlemeasurements are typically performed during each rev-olution of the calotte: a dark measurement on an opaquepart of the calotte, a reference measurement throughan open position in the calotte, and the measurementthrough the substrate to be coated. Based on thesethree measurements, the absolute transmittance spec-trum of the substrate is calculated. Furthermore, theonline monitoring software calculates the current layer

Halogen lamp

Ionsource

Trigger

Optical fibre

Process control

Fig. 6.15 Schematic view of the online monitoring systemfor optical broadband in situ transmittance measurements,realized within an IAD process environment. A halogenlamp mounted above the calotte inside of the process cham-ber is used as the light source, and the light is coupled intothe detector fibre through a window in the chamber bottom

PartA

6.2


thickness on the basis of the known optical parametersof the coating material (dispersion values and extinc-tion coefficients). Provided with a target multilayerdesign, the online system can control the depositionplant, determine the end of the layer currently be-ing deposited, and start the next layer. Because of thelarge databases available, broadband monitoring sys-tems are also useful as a supporting tool for processdevelopment and quality management in the produc-tion of modern optical coatings. In contrast to the testsubstrate arrangement in conventional single- and multi-wavelength optical monitors, broadband measurementson the moving substrates make calibration factors andtest runs redundant, if there are only the smallest devi-ations between the vacuum spectra and the spectra afterventing (the vacuum-to-air shift). Therefore, an appro-priate combination of an adequate deposition processand monitoring concept has to be chosen, e.g., an IAD

process providing an extremely low vacuum-to-air shift,as shown above.

Besides a precise layer thickness determination, inthe development and control of coating processes for op-tics, a variety of optimization criteria has to be taken intoaccount. Large areas have to be coated homogenously inflexible and fast processes. Besides the optical quality,the coatings have to exhibit good adherence, low me-chanical stress, high abrasion resistance, and adequateenvironmental stability. Parameters to be controlled are,for example, substrate temperatures, deposition rates,partial pressures of process gases, or ion current den-sities and energies. Finally, it has to be mentioned thatthe thin film processing concepts presented are not onlyapplied for optical coatings, including precision, laser,and consumer optics, but also in the fields of elec-tronics, semiconductors, displays, medical technology,tribology, and even decorative coatings.

6.3 Quality Parameters of Optical Coatings

For the application of optical coatings in laser technol-ogy, a variety of quality parameters have to be taken intoaccount [Table 6.1, [6.6168]]. The spectral characteris-tics are the fundamental properties of the coating systemand have to be adjusted primarily for the optical func-tioning of the laser system or optical device [6.69]. But,as soon as high laser powers or requirements concerningthe precision are involved, additional quality parame-

Table 6.1 Selected quality parameters of optical coatings and surfaces in conjunction with corresponding ISO standardsand measurement principles

Specification Parameter Unit Standard Measurement principleLaser-induced cw-LIDT W/cm ISO 11254-1 Cw-laser irradiationdamage threshold 1 on 1-LIDT J/cm2 ISO 11254-1 Irradiation with single pulses(LIDT) S on 1-LIDT J/cm2 ISO 11254-2 Repetitive irradiation with pulses

Certification J/cm2 ISO 11254-3 Irradiation sequenceOptical losses Absorptance ppm ISO 11551 Laser calorimetry

Total scattering ppm ISO 13696 Integration of scattered radiationTransfer function Reflectance % ISO 13697 Precise laser ratiometric method

Transmittance % ISO 15368 SpectrophotometrySurface quality Form tolerances N ISO 10110 13 parts containing different

Scratch/digs types of imperfectionsRoughness

Stability Abrasion ISO 9211 Different test methodsEnvironmental ISO 9022 21 parts containing a variety ofStability conditioning methods

ters describing the optical losses and the stability of thecoatings have to be evaluated. For many applicationsof high-power lasers, the absorbance, which transformsa fraction of the impinging radiation energy into heat, in-ducing a temperature increase in the coating, is of majorconcern. As an effect of absorption a temperature profileis built up in the optical component, which may influ-ence the transfer behavior or even lead to the thermal

PartA

6.3

Thin Film Optical Coatings 6.3 Quality Parameters of Optical Coatings 389

destruction of the optical component [6.70]. Besides thetechnical problems directly related to thermal effects,absorption always implies a loss of expensive laser en-ergy, demonstrating the economical dimension of opticallosses in modern laser systems. Absorbance in opticalcoatings is mainly governed by defects, stoichiometricdeficiencies, or contaminants generated in the coatingmaterial during the production steps. In particular, di-electrics often suffer from unbalanced stoichiometrywith a slight excess of the metal component, which iscaused by a decomposition of the coating materials dur-ing thermal evaporation or sputtering. Stoichiometriceffects influence the absorbance behavior in the short-wavelength region whereas adsorbed water, which is thepredominant contaminant in optical coatings, may leadto increased absorption values in the MIR spectral re-gion, particularly at the wavelength 10.6 m [6.71] ofthe CO2 laser or the band of the Ho : YAG and Er : YAGlasers in the range 23 m [6.72]. A standard procedurefor the measurement of absorbance in laser compo-nents, which is based on the laser calorimetric principle,is outlined in ISO 11551 [6.62]. For calorimetric ab-sorbance measurement, a temperature sensor is attachedto the specimen, which is located in a thermally isolatingchamber. According to the standard protocol, the sam-ple is irradiated by a laser beam with known power fora heating time tB after having reached thermal equilib-rium with the environment. As a consequence of the heatflow coupled into the sample by the absorption, an expo-nential increase in temperature can be monitored by thesensor element. Subsequently, the laser is blocked andthe sample temperature decreases in proportion to theheat dissipation into the environment. For the determina-tion of the absorbtance, the recorded heating and coolingcurve are evaluated according to the methods describedin the standard. The laser-calorimetric measurementtechnique had been tested in various round-robin exper-iments [6.73,74] and offers the advantage of an absoluteand sensitive assessment of absorbance [6.75].

Scattering is the second loss channel in opticalcomponents and summarizes all effects that deflect theradiation from its specular direction [6.76, 77]. Besidesthe economic loss of radiation, scatter may induce a re-duction of the imaging quality of optical systems andmay even appear as a safety problem, if a significantfraction of laser power is diverted into the environment,endangering personal operating the laser device. Scat-tering losses of optical coating systems can mainly beattributed to microstructural imperfections and inclu-sions in the coatings as well as to the roughness ofthe surface and the interfaces between the individual

layers [6.76]. Theoretical models on the basis of theseeffects reveal a scaling of the scattering with the wave-length which can be described on the basis of a 1/xfunction with an exponent x of 14. In view of thisfundamental relation, high scattering values dominat-ing the losses of an optical component are expectedespecially for the vacuum ultraviolet (VUV)/UV spec-tral ranges. Following the wavelength scale towardslonger wavelengths, optical scattering decreases andmay be even neglected for the MIR spectral range,where absorption losses are more pronounced. Dur-ing the last three decades, an extended scientific andtechnical background has been built up in the field ofscattering measurements resulting in various standardmeasurement procedures for angle-resolved scattering(ARS) [6.78] and total scattering (TS) [6.63, 79]. Espe-cially, for the determination of the TS value, which isdefined by the total amount of radiation scattered intothe 4 fullspace by an optical component, measure-ment set ups with an Ulbricht integrating sphere [6.80]or a Coblentz collecting sphere [6.81] are described inISO 13696. The fundamental principle of the Ulbrichtsphere is based on the integration of scattered radia-tion by a white highly diffuse reflecting coating on theinner wall of sphere and a subsequent monitoring ofpart of the integrated radiation, whereas the Coblentzsphere performs a direct collection and concentration ofthe scattered radiation onto the detector element. Also,the standard ISO 13696 for the measurement of TSvalues has been tested in various measurement cam-paigns [6.82] and has recently been qualified for theDUV/VUV spectral range [6.83, 84].

In particular, high-power lasers impose highdemands on the power-handling capability of opti-cal coatings, which is expressed in terms of thelaser-induced damage threshold (LIDT). Fundamentalparameters limiting the LIDT values of coatings aregiven by the inherent properties as the melting point,the thermal conductivity or the band-gap energy ofthe employed materials [6.85]. Besides these intrinsicproperties, extrinsic effects related to defects and in-clusions in the layer structure or special high-powermechanisms at the layer interfaces have to be takeninto account [6.86]. Frequently, defect-induced damagemechanisms are observed in optical coatings (Fig. 6.16),which can be described on the basis of inclusions catas-trophically heated under laser radiation [6.87]. In thistheory, the generation of heat in the inclusion is mod-eled by the Mie absorption cross section, which is validfor particles with sizes in the range of the interactingwavelength. The diffusion of heat from the inclusion

PartA

6.3


Fig. 6.16 Scanning electron microscope (SEM) picture ofa damage site initiated by defect mechanisms in a single-layer coating of TiO2 subjected to an energy density of10.9 J/cm2 in a Nd:YAG laser beam with a diameter of ap-proximately 200 m. The height of the picture correspondsto a scale of 20 m

into the surrounding coating material can be expressedby a solution of the heat diffusion equation for this spe-cific geometry with polar symmetry [6.88]. The point ofdamage is reached when the perimeter temperature ofthe inclusion attains the melting point of the surround-

Fig. 6.17 Nomarski micrograph (color-enhanced presenta-tion) of a damage site initiated by absorption-dominatedmechanisms in a highly reflecting mirror of Ta2O5/SiO2,objected to an energy density of 45.9 J/cm2 in a Nd:YAGlaser beam with a diameter of approximately 200 m

ing coating material. During the last two decades, thedefect model had been studied by a several workinggroups, resulting in a deep understanding of the un-derlying mechanisms and the corresponding propertiesof the coating defects [6.85, 89]. Besides inclusion-dominated breakdown other mechanisms based onabsorption and on electronic effects are discussed forthe development of high-power laser coatings. In thepicture of absorption-induced damage, the energy is di-rectly coupled into the layer structure by absorptionin the interaction area with the laser beam and leadsto homogeneous temperature increase until the dam-age temperature of the structure is reached. The damagetemperature is specified by transitions in the crystallinestructure, the crossing of a defined coating stress level(Fig. 6.18), or the melting temperature (Fig. 6.17) of thecoating material [6.90]. For short laser pulses in the psand fs time domain, typical diffusion lengths of ther-mal effects are very small compared with the thicknessof the layer structure. As a consequence, thermal ef-fects can be neglected for the modeling of ultrashortpulse damage, which is mainly governed by electronicprocesses. Present theories start from the assumptionthat catastrophic damage takes place at a critical elec-tron density of 1021 cm3 in the conduction band of the

Fig. 6.18 Nomarski micrograph of a damage site initiatedby absorption-dominated mechanisms in an antireflectivecoating of ZrO2/MgF2 on a quartz substrate, subjected toan energy density of 35 J/cm2 in a Nd:YAG laser beamwith a diameter of approximately 200 m. In this case thecatastrophic stress level was reached prior to the thermaldeterioration of the coating material

PartA

6.3

Thin Film Optical Coatings 6.4 Summary and Outlook 391

coating material [6.91]. At this point laser energy is cou-pled efficiently into the electrons, resulting in ionizationand disruption of the material. For the description of thedevelopment of the electron density in the conductionband before arriving at the critical density, rate equationsinvolving multiphoton excitation and the avalanche ef-fect are employed. The major outcome of these theoriesis a clear correlation of the damage threshold to themaximum internal field strength in the layer structureand to the band-gap energies of the layer materials in-volved. These theoretical predictions are supported byexperiments on single-layer and coating systems testedwith ultrashort-pulse lasers in a broad range of pulsedurations [6.92]. Aside the numerous damage mecha-nisms in the depth of the layer structure, features onthe surface of the component may also diminish thepower-handling capability of the optical component. Forexample, the field strength of the incoming wave can beenhanced by more than a factor of two in the vicinity ofgrooves, cracks or other surface imperfection, leadingto weak points on the component (Fig. 6.19) [6.93]. Anadditional cause of surface-initiated damage is the gen-eration of plasmons, which is correlated to the surfaceroughness of the component [6.94]. In these models,the damage threshold values decrease with the increas-ing surface roughness of the layers. Conclusively, eventhough a variety of damage mechanisms have been iden-tified and understood on the basis of adapted models,the power-handling capability of optical thin films isstill a major and vivid research area which plays a keyrole in the development of high-power lasers and theircommercialization.

In many technical applications, the optical com-ponent additionally has to withstand a variety ofenvironmental influences, including mechanical abra-sion, chemical corrosion or severe climatic conditions.For example, the most severe mechanical and chemicalrequirements are imposed onto coatings for ophthal-mology or consumer optics, which are often cleanedwith even abrasive and aggressive cleaning solvents and

Fig. 6.19 Example of damage initiated by a surface imper-fection in an antireflective coating of ThF4/ZnSe on a ZnSesubstrate, subjected to an energy density of 12.7 J/cm2in a TEACO2 laser beam. The width of the picturecorresponds to a scale of 150 m

fabrics. In the course of the individual development ofthe different optical products a variety of standardizedtesting procedures has been cultivated in the differentmarket sectors for the qualification of optical com-ponents. Some of these procedures originating fromophthalmology, defence and medical applications arecompiled in Table 6.1, which illustrates the broad spec-trum of test procedures available. In particular, in lasertechnology the requirements in respect to environmentalstability are less demanding and in most cases certifica-tion according to the conditioning methods with a lowerdegree of severity described in the ISO standard series9211 are sufficient. Typical conditioning methods coverabrasion resistance, which is tested by a series of strokeswith a special cheese cloth or eraser applied to the op-tical surfaces, and climatic stability, which is assessedon the basis of defined conditioning cycles in a climaticchamber.

6.4 Summary and Outlook

Even after more than 60 years of research and de-velopment, optical coatings are often still conceivedas a technology with a large fraction of magic art,which can be mastered only by experience and alwaysleaves a certain probability of unforeseen effects inproduction. Major challenges include the reliable and

reproducible deposition on the basis of well-defined andstable process concepts as well as sensitive and quali-fied characterization of the coatings. In view of the rapiddevelopment of computer systems, online monitoringstrategies of the deposition processes and the optimiza-tion of stable deposition processes, significant progress

PartA

6.4


on the way towards precise and deterministic productionof even complicated coating system are expected in thenear future. Laser technology has always stimulated thedevelopment of coatings to a considerable quality level(Table 6.2) and will also be one of the major pace-settersfor future innovations in the field.

Demanding future challenges will be imposed onoptical coatings by the commercialization of new laserproducts and innovative applications, which are depen-dent on optical coatings with improved optical quality,higher stability and increased complexity in their func-tioning combining several surface properties. This newtechnology generation will also require improved flex-ibility and economy from optical factories, which canonly be achieved on the basis of reproducible produc-tion processes and adapted characterization techniques.In conclusion, optical thin films will remain an enablingtechnology which will play a key role for many futureapplications and products.

Terms and Definitions(in their order of appearance)

nT refractive index of the substrateE0 amplitude of a plane wavek wave number of a plane wave k = 2/ wavelength of a plane wave frequency of a plane wavei phase shift of a plane wave in layer number i

Table 6.2 Selected quality parameters of optical coating systems for laser applications (Types: HR: high-reflecting mirror,AR: antireflective coating, th: thermal evaporation, IBS: ion-beam sputtering)

Laser,wavelength

Type AbsorptionISO 11551

Total scatteringISO 13696

Laser-induced damagethreshold, ISO 11254

157 nm, F2-excimer HR/th 14% 14%193 nm, ArF-excimer AR/th 0.72.5% 0.20.5% 12 J/cm2 (1on1, 20 ns)

HR/th 0.42.0% 0.22.5% 24 J/cm2 (1on1, 20 ns)248 nm, KrF-excimer AR/th < 0.025% 10 J/cm2 (1on1, 30 ns)

HR/th < 500 ppm < 0.2% > 20 J/cm2 (1on1, 30 ns)HR/IBS < 0.1% > 3 J/cm2 (1on1, 30 ns)

633 nm, HeNe Laser HR/th < 30 ppm < 30 ppm -HR/IBS < 5 ppm < 5 ppm -

1.064 m, Nd:YAG AR/th < 20 ppm < 100 ppm > 60 J/cm2 (12 ns, 0.25 mm)HR/th < 50 ppm < 100 ppm > 100 J/cm2 (12 ns, 0.25 mm)HR/IBS < 1 ppm < 1 ppm > 80 J/cm2 (12 ns, 0.25 mm)

10.6 m, CO2-Laser AR/th < 0.16% - > 20 J/cm2 (100 ns, 1.4 mm)> 2 kJ/cm2 (1.2 ms, 250 m)> 3 kW/mm (100 m)

HR/th < 0.10% - > 25 J/cm2 (100 ns, 1.4 mm)> 2 kJ/cm2 (1.2 ms, 250 m)

di thickness of the layer number ini refractive index of the layer number ii angle of incidence in the layer number iA0 amplitude of the partial wave of order zeroA1 amplitude of the partial wave, order oneA2 amplitude of the partial wave, order twot1, t1 transmission coefficients of the ambientlayer

interfacer1, r

1 reflection coefficients of the ambientlayer in-

terfacer2 reflection coefficient at the layersubstrate in-

terfaceAk amplitude of the partial wave of order krS reflection coefficient of a single layerMi matrix of layer iEi , Hi electric field strength, magnetic field strength

at the rear plane interface of layer iMS composite matrix of a layer stack of K single

layersrSK reflection coefficient of a layer stack of K single

layersn0 index of refraction of the ambient mediumDi film thickness expressed in units of QWOTQWOT quarter-wave optical thickness; the unit for the

thickness of the layerZ design wavelengthnH refractive index of a high-index materialnL refractive index of a low-index materialRS reflectance of a QWOT stack

PartA

6.4

Thin Film Optical Coatings References 393

PVD physical vapor depositionCVD chemical vapor depositiondNe/dt number of evaporating atoms per unit timeAe surface area of the evaporation sourcee evaporation coefficientp equilibrium vapor pressure of the evaporantp hydrostatic pressure of the evaporantm atomic masskB Boltzmanns constantT temperaturep0 constant factor (pressure)L0 latent heat of evaporation per atom or molecule

RLVIP reactive low-voltage ion platingI(B)AD ion-(beam-)assisted depositionIBS ion-beam sputteringAPCVD atmospheric pressure CVDLPCVD low-pressure CVDPECVD plasma-enhanced CVDPICVD plasma impulse CVDLCVD laser(-induced) CVDPLD pulsed laser depositionARS angle-resolved scatteringTS total scatteringLIDT laser-induced damage threshold

References

6.1 P. Vukusic, J. R. Sambles: Photonic structures in bi-ology, Nature 424, 852855 (2003)

6.2 J. v. Fraunhofer: Versuche ber die Ursachendes Anlaufens und Mattwerdens des Glases unddie Mittel, denselben zuvorzukommen (Verlagder Kniglich Bayerischen Akademie der Wis-senschaften, Mnchen 1888) p. 1817

6.3 H. D. Taylor: A Method of Increasing the Brilliancyof the Images Formed by Lenses, British patent No29561 (1904)

6.4 C. Fabry, A. Perot: Thorie et applications dune nov-elle mthode de spectroscopie interfrentielle, Ann.Chim. Phys. 16, 115144 (1899)

6.5 M. Born, E. Wolf: Principles of Optics, 7 edn. (Cam-bridge Univ. Press, Cambridge 1999) p. 986

6.6 A. J. Thelen: Design of Optical Interference Coatings(McGraw-Hill, New York 1989) p. 255

6.7 H. M. Lidell, H. G. Jerrard: Computer-Aided Tech-niques for the Design of Multilayer Filters (AdamHilger, Bristol 1981) p. 194

6.8 P. Baumeister: Computer software for optical coat-ings, Photon. Spectra 22(9), 143148 (1988)

6.9 A. Macleod: Thin-Film Optical Filters, 3 edn. (Inst,Publishing, London 2001) p. 641

6.10 W. R. Grove: On the electrochemical polarity of gases,Philos. Trans. R. Soc., London 142, 87101 (1852)

6.11 M. Faraday: The bakerian lecture: experimental re-lations of gold (and other metals) to light, Philos.Trans. R. Soc., London 147, 145181 (1857)

6.12 A. W. Wright: On the production of transparentmetallic films by the electrical discharge in ex-hausted tubes, Am. J. Science Arts 3rd Ser. 13(73),4955 (1877)

6.13 T. A. Edison: Art of Plating One Material with Another,US Patent 526,147 (1894)

6.14 A. Macleod: The early days of optical coatings, Jour-nal of Optics A: Pure Appl. Opt. 1, 779783 (1999)

6.15 D. M. Mattox: The Foundations of Vacuum Coat-ing Technology (Noyes Publications, Norwich 2003)p. 150

6.16 H. Bach, D. Krause: Thin Films on Glass, ed. byH. Bach, D. Krause (Springer, Berlin, Heidelberg 1997)p. 404

6.17 H. K. Pulker: Coatings On Glass, 2 edn. (Elsevier,Amsterdam 1999) p. 548

6.18 R. R. Willey: Practical Design and Production of Opti-cal Thin Films, 2 edn. (Marcel Dekker, New York 2002)p. 547

6.19 P. W. Baumeister: Optical Coating Technology (SPIEPress, Bellingham 2004) p. 840

6.20 M. Friz, F. Waibel: Coatings Materials, Optical Inter-ference Coatings, Springer Ser. Opt. Sci. 88, 105130(2003)

6.21 H. Hertz: I. ber die Verdunstung der Flssigkeiten,insbesondere des Quecksilbers, im luftleerenRaume, Ann. Phys. 17, 177200 (1882)

6.22 M. Knudsen: Die maximale Verdampfungsge-schwindigkeit des Quecksilbers, Ann. Phys. 47,697708 (1915)

6.23 K. H. Guenther: Microstructure of vapor-depositedoptical coatings 23(21), 38063816 (1984)

6.24 I. Petrov, P. B. Barna, L. Hultman, J. E. Greene: Mi-crostructural evolution during film growth, J. Vac.Sci. Technol. A 21(5), 117128 (2003)

6.25 B. A. Movchan, A. V. Demchishin: Study of the struc-ture and properties of thick vacuum condesates ofnickel, titanium, tungsten, aluminium oxide andzirconium dioxide, Phys. Met. Metallogr. 28, 83(1969)

6.26 M. Auwrter: Verfahren und Einrichtung zur Herstel-lung dnner Schichten aus Metallverbindungen ent-haltenden Ausgangssubstanzen durch Aufdampfenim Vakuum und nach dem Verfahren erhal-tene Aufdampfschicht, Swiss Patent No 322265(1957)

6.27 M. Auwrter: Process for the Manufacture of ThinFilms, US Patent 2,920,002 (1960)

6.28 D. M. Mattox: Apparatus for Coating a CathodicallyBiased Substrate from Plasma of Ionized CoatingMaterial, US Patent 3,329,601 (1967)

PartA

6


6.29 H. K. Pulker, W. Haag, M. Buhler, E. Moll: Optical andmechanical properties of ion plated oxide films, ed.by H. K. Pulker, W. Haag, M. Buhler, E. Moll (Proc.IPAT 85, 5th Int. Conf. on Ion and Plasma AssistedTechniques, Munich 1985) pp. 299306

6.30 J. D. Targove, H. A. MacLeod: Verification of momen-tum transfer as the dominant densifying mechanismin ion-assisted deposition, Appl. Opt. 27, 18, 37793781 (1988)

6.31 S. Berg, I. V. Katardjiev: Preferential sputtering ef-fects in thin film processing, J. Vac. Sci. Technol. A17(4), 19161925 (1999)

6.32 J. B. Malherbe, S. Hofmann, J. M. Sanz: PreferentialSputtering of Oxides: A Comparison of Model Pred

thin film optical coatings

Documents