Reflectance and preparation of front-surface mirrors for useat various angles of incidence from the ultraviolet to the far
infrared
Georg Hass
U.S. Army Electronics Research and Development Command, Night Vision and Electro-Optics Laboratory, FortBelvoir, Virginia 22060
Received September 2,1981
The increasing use of reflecting optics in ultraviolet (UV), visible, and infrared (IR) devices has stimulated a greatamount of research on coatings for front-surface mirrors. Methods for measuring the reflectance of front-surfacemirrors at various wavelengths and angles of incidence are discussed, and techniques for preparing reflecting filmswith maximum reflectance and durability are described. Data on the UV, visible, and IR reflectance of the mostfrequently used mirror coatings, Al, Ag, Au, and Rh, are presented. The application of single-layer and multilayerdielectric overcoatings for increasing their durability and normal-incidence reflectance are described, and the effectof these coatings on reflectance at higher angles of incidence is treated. It is shown that Al and Ag overcoated withrather thin layers of silicon oxides or A120 3 (t = 1000-2000 A) have, in the IR from 8 to 12 pm, practically the samehigh reflectance as the unprotected metal at close to normal incidence but greatly decreased reflectance at angleslarger than 400. Since only the parallel component is responsible for the IR-reflectance decrease, such film combi-nations are suitable for producing highly efficient reflection polarizers for the IR, such as for CO2 10.6-pm laser ra-diation. The effect of water absorption in dielectric overcoatings on the mirror reflectance at 3 pm is discussed.
INTRODUCTION
The increasing use of reflecting optics in ultraviolet (UV),visible, and infrared (IR) devices has stimulated a greatamount of research on coatings for front-surface mirrors.Space, laser, and IR applications are the main reason for theirincreased use today, and important progress toward the de-velopment of reflecting coatings with increased reflectance,improved mechanical and chemical durability, and betteradhesion to various substrates has been made in recent dec-ades. The main factors responsible for the improvementsmade in the preparation of mirror coatings are (1) the devel-opment of vacuum systems that make it possible to depositfilms at low pressures or in the presence of oxygen and othergases of well-controlled pressures for reactive evaporation; (2)new deposition techniques, such as electron-gun evaporation,which permits the deposition of many new optical coatings,such as high-melting-point metals and oxides; (3) improveddevices for monitoring the film thicknesses during film de-position; (4) the use of purer and well-outgassed startingmaterials.
The preparation and reflectance of coatings for vacuum-UVspectroscopy and space astronomy has been reported in recentsummary articles.1-3 In this paper, data on the UV, visible,and IR reflectances of the most frequently used mirror coat-ings, Al, Ag, Au, and Rh, are presented, and their reflectancebehavior at various angles of incidence is discussed. Generalrules for predicting the effect of angle of incidence on the re-flectance and polarization of the reflected beam will be givenfor metal coatings.
For many mirror applications utilizing highly reflectingmetal films, hard, transparent, and adherent single-layer andmultilayer overcoatings have to be applied to improve thechemical and mechanical durability of the mirror coatings.
The most frequently used protective layers are films of siliconoxides, SiO, SiOx, Si0 2 t 7 and A12 03 .5 For most purposes,1000-2000-A-thick layers of these materials are sufficient togive adequate protection against abrasion. It will be shown,however, that such silicon-oxide- and A12 03 -protected mirrorshave, in the IR from 8 to 12 Am, practically the same high re-flectance as unprotected ones at close to normal incidence butgreatly decreased reflectance at angles greater than 40°.9,10Therefore mirrors of this type are unsuitable for devices usinga 450 angle of incidence, or scanning optics, in this region.The 8-12-gm region is emphasized because it is one of theatmospheric windows. The drastic IR reflectance decreaseat higher angles of incidence is caused by the fact that theoptical constants n and k of the above oxide films have valuesof less than 1 in this region. Since only the parallel compo-nent is responsible for the reflectance decrease, such filmcombinations are suitable for producing highly efficient re-flection-type polarizers for the IR, such as for 10.6-pm CO2laser radiation.'1 Protected mirror coatings that have highreflectance in the 8-12-gm region for normal to high anglesof incidence are described.12 Since many dielectric over-coatings used on front-surface mirrors absorb water whenexposed, to air, the effect of this water absorption on the re-flectance of overcoated mirrors is discussed.'3"14
REFLECTANCE MEASUREMENTS ANDPREPARATION OF MIRROR COATINGS
Reflectance MeasurementsThe reflectance of front-surface mirrors should be as high aspossible in the wavelength region and at the angles of inci-dence where they are used. To predict the efficiency of de-vices and equipment using reflecting optics, accurate reflec-
tance measurements of the various mirrors have to bemade.
When a beam of radiant energy is incident upon an opaquemirror coating, part of it is reflected and part is absorbed.Reflectance is defined as the ratio of the reflected radiantenergy 1r to the incidence radiant energy Io. It is denotedusually by the letter R (InfIo). The reflected beam can consistof two parts, the specularly reflected component, which is anextension of the incident beam as defined by the geometricallaws of reflection, and the diffusely reflected component,which is the part of the radiant energy scattered from thesurface because of roughness. Since we are dealing in thispaper only with rather smooth reflecting coatings depositedon well-polished substrates, all discussions are limited to thetreatment of the specular reflectance of front-surface mir-rors.
In addition to the term reflectance defined above, anotherterm frequently employed is reflectivity. We suggest that thisterm be used to define the ideal reflectance of a material, suchas the reflectivity of Ag, which at X = 5000 A is 97.9%. Theterm reflection represents a process and should not be usedin connection with numbers. Methods for measuring R canbe divided into two groups: (1) Methods that measure thereflectance without a comparison standard. Devices of thistype are called absolute reflectometers. (2) Techniques thatdetermine R of a sample by comparison with a standardmirror of known reflectance.
There are various techniques for determining R without theuse of a reflectance standard. A combination goniometer-reflectometer is the most straightforward device for deter-mining R of mirror samples at close to normal and at a varietyof angles of incidence. It uses a detector that moves in a circlearound -the sample to be measured while the sample rotatesso that the reflected beam is directed toward the detector.This single-beam instrument requires the measurement of theradiant energy of the direct beam and that of the reflected one.The measurement of R at angles other than normal incidenceis of interest if the mirrors are used at various angles and alsofor the determination of the optical constants from the re-flectance data obtained at many different angles of incidenceof 10 to 800.15,16 The goniometer-reflectometer can easilybe adapted for use in a vacuum system so that R measure-ments can be made in situ. Madden and Canfield17 used suchequipment for measuring R of freshly deposited metal filmsbefore and after exposure to air and used it to determine thevacuum-UV optical properties of various metals. Byemploying different light sources, monochromators, and de-tectors, the use of this instrument can be extended to Rmeasurement from the UV to the IR. Measurements beforeand after air inlet furnish important information on the effectof oxidation and water absorption on the reflectance of metalmirrors with and without overcoatings.
A precision Strong-type reflectometer18 suitable for mea-suring the absolute R values of mirrors with high accuracy atclose to normal incidence has been described by Bennett andKoehler19 and Bennett and Bennett.2 0 With this instrument,they measured R of highly reflecting samples from 0.3 to 32gm with absolute accuracy of 0.1% and an average deviationof 0.04%. Recently, further improvements of this instrumentwere reported that improved its precision to +0.01% in re-flectance measurements.2 ' This technique can be used toprovide data for reflectance standards at normal incidence.
In a review article,20 Bennett and Bennett describe the designof many other devices suitable for measuring the R of mirrorcoatings.
Multiple-pass reflectometers have become one of the mostimportant tools for measuring the absolute reflectance ofhighly reflecting mirrors with high accuracy. Gates et al.22
used a varying number of multiple reflections from identicalparallel specular-reflecting mirrors to determine the absoluteIR reflectances of various metals between 20 and 60° with anestimated accuracy of 0.2%. Harris and Fowler,23 by usinga similar technique, measured the reflectance of Au from 8.5to 84 Am. Herriott and Schulte 2 4 used a high number ofmultiple reflections to measure at normal incidence the ex-tremely high reflectance of all-dielectric mirrors with highprecision. Perry25 measured with the Herriott-Schulte re-flectometer the reflectance of a great number of alternatelayers of ZnS/ThF 2 to be 99.8%. The measurements weremade by using 100 passes. More recently, Armon and Bau-meister26 described a versatile high-precision multiple-passreflectometer. In this instrument, the mirrors can be eitherplano or curved with a few diopters of power, and the angleof incidence can vary from 5 to 70°. Multilayer stacks con-sisting of 33 layers of Ta2 O5/SiO 2 showed a reflectance of99.96% ± 0.02% at X = 520 nm. Their publication also givesmany references to other articles describing various types ofmultiple-pass reflectometers.
Since reflectance values determined with absolute reflec-tometers are usually made wavelength by wavelength, the timerequired to obtain reflectance curves over an extended-wavelength region can be quite long. Therefore most re-flectance-versus-wavelength measurements of samples aremade by comparison with calibrated standards in recordingdouble-beam spectrophotometers that are commerciallyavailable, such as those produced by Perkin-Elmer andBeckman. Since the reflectance data obtained with theseinstruments are relative to those of a calibrated standard, itis necessary to have reliable standards on hand. Mirrorcoatings of Rh, freshly deposited Al, and Au, measured witha high-precision reflectometer, are suitable standards. Withthis method, highest accuracy is obtained by leaving themirror in one beam untouched while using the second beamto record the R curves of the standard and the samples to bemeasured. In this way, the accuracy of relative R measure-ments can be better than 0.5%, and an average deviation of lessthan 0.1% can be obtained by the use of an extended R scale.However, there is a difficulty associated with these devices.The reflectance is usually determined for a particular angleof incidence that is close to normal. Only recently, additionalaccessories have been developed to measure ft also at greaterangles of incidence.
Preparation TechniquesBy far the most widely used technique for depositing reflectingcoatings is evaporation in high vacuum. With no othermethod can films of highest reflectance and of any desiredthickness be prepared with so complete a measure of control.By the use of suitable shutters and rotating sector wheels, andby movement of the mirror substrates during the evaporation,films of extremely uniform thicknesses over large areas maybe obtained. Therefore this article is confined to discussingreflecting coatings prepared by high-vacuum evaporation
Georg Hass
Vol. 72, No. 1/January 1982/J. Opt. Soc. Am. 29
performed at various deposition rates and pressures ontosubstrates of various temperatures.
Al is the most frequently used metal for depositing re-flecting coatings for front-surface mirrors because it has highreflectance from the vacuum UV to the far IR, adheres wellto glass and other substrates, and does not tarnish in normalair. It is also easy to evaporate from helical tungsten coils orby electron bombardment. Obviously, Al coatings are espe-cially important for astronomical mirrors and reflectiongratings for which high reflectance in the UV is required. Inorder to produce highly reflecting films of materials, such asAl, which readily absorb oxygen and other gases, extreme caremust be taken to ensure that the evaporated films are notcontaminated by residual gases present during the deposition.It is well known that Al films of highest reflectance can beproduced in conventional evaporators only by the use of ex-tremely high deposition rates and the use of the purest gradeof Al (99.999%) as the starting material.27
-29 Highly reflecting
Al films can also be produced by slower evaporation in ultra-high-vacuum systems.3 0' 31 Both of these methods eliminatethe absorption of impurities during the deposition. Hut-cheson et a!.32 have made a comparison of the reflectances ofAl films evaporated in conventional and in ultrahigh-vacuumsystems. In both of their vacuum systems, the evaporationsources consisted of six multistrand helical tungsten coils thatwere heated simultaneously. The coils were outgassed andfreed from impurities before being charged with 99.999%-pureAl. This was followed by a second heating to melt the Al andto remove its impurities. With this arrangement, Al can bedeposited at rates up to 1000 A/sec at a distance of about 50cm. The effect of deposition rate on the reflectance at X = 400nm and X = 200 nm of Al films deposited at 3 X 10-6 and 5 X10-9 Torr is shown in Figs. 1 and 2. For Al films depositedat rates of 300 A/sec and greater, there is no measurable dif-ference in reflectance of the two types of films. However, iflower deposition rates are used, the reflectance of Al filmsdeposited at 5 X 10-9 Torr is much less dependent on thedeposition rate than that of filns condensed at 3 X 10-6 Torr.At X = 400 nm the reflectance difference is still quite small andis only noticeable at rates lower than 20-30 A/sec. At X = 200nm the reflectance difference of the two types of films becomespronounced, and it can be as high as 30% if deposition ratesof about 3 A/sec are used. At such low rates even films con-densed at 5 X 10-9 Torr show a considerable decrease in re-flectance. Since for most metals high deposition rates resultin denser films with smoother surfaces,33 rapid evaporationis recommended for almost all mirror coatings.
The fact that evaporated Al does not tarnish in normal at-mosphere is the result of the good protective qualities of itsnatural oxide film, which grows in air to an ultimate thicknessof about 40 .34 This thin oxide filmn results in a negligibleloss of the Al reflectance for wavelengths longer than 1.0 pmand a loss of only 0.3% at 500 nm. However, this is not truefor the vacuum-UV region where the oxide film causes adrastic decrease in reflectance. 35
A detailed description of a 2-m evaporator suitable forcoating large mirrors with Al and Al-plus-overcoatings hasbeen published by Bradford et al.
3 6 Sixteen simultaneouslyheated helical coils of tungsten charged with molten Al wereused as evaporation sources. This makes it possible to depositAl at high rates and thus to obtain films of optimum reflec-tance.
DEPOSITION RATE IN A/SEC -
Fig. 1. Effect of deposition rate on the reflectance at 400 nm of Alfilms deposited at 3 X 10-6 and 5 X 10-9 Torr. All films aged 24 h inair.
DEPOSITION RATE IN A/SEC
Fig. 2. Effect of deposition rate on the reflectance at 200 am of Alfilms deposited at 3 X 0-6 and 5 x 10-9Torr. All films aged for 24h in air.
Optical emission spectrographic analysis was used to studythe purity of the evaporated Al films. The purity was foundto be equal to that of the starting material and showed no in-dication of tungsten (W). Rook and Plumb3 7 studied the Wcontent of Al films evaporated from W heaters by using ra-dioactive 187W prepared by neutron irradiation. They alsofound a negligible amount of W in opaque Al coatings.However, W contamination was found when uncleaned,slightly oxidized W wires were used. This contamination isproduced by vaporization of W03 before the Al is molten andwould be expected to produce a film of contamination of about0.1 A at the Al-substrate interface.
Love and Bower38 reported that they deposited 150-,um-thick Al coatings of ultrahigh purity by evaporation withan electron gun. In their paper they discuss how sourcescausing film contamination were isolated and eliminated.
In contrast to coating materials, such as Al, that are effectivegetters, the reflectance of platinum metals such as rhodium(Rh) and platinum (Pt) is rather insensitive to the evaporationconditions. Rh and Pt films deposited at rates ranging from10 to 100 A/sec at pressures varying from 1 X 10-5 to 1 X 10-6Torr showed the same reflectance. These films were preparedby evaporation with an electron gun. It was, however, foundthat the reflectances of Rh (Ref. 39) and Pt (Ref. 40) aregreatly affected by the substrate temperature during the de-
Georg Hass
30 J. Opt. Soc. Am./Vol. 72, No. 1/January 1982
100 -
90300'C
W 80- - - -
40
- - - - - -
30-0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2
WAVELENGTH IN pm.
Fig. 3. Reflectance of opaque Rh films deposited on glass at 40 and3000 C in the wavelength region from 0.2 to 2.2 ,m.
position. Figure 3 shows the difference in reflectance of Rhfilms deposited at 40 and 3000 C. At X = 546 nm, room-tem-perature films have a reflectance of 73.8%, whereas Rh coat-ings deposited at 3000C show a reflectance of 78.1%. Evap-orated Pt films show a similar dependence of their reflectanceon the substrate temperature. Their reflectance increasesfrom 66.0 to 73.4% at X = 550 nm when the substrate tem-perature is raised from 40 to 3000 C. Both film materials, Rhand Pt, exhibit, when evaporated onto substrates of 300-350°C, no indication of surface roughness, and their reflec-tances are higher than those of room-temperature films fromthe vacuum UV to the far IR. This is not true for films of Al,Ag, and Au, which become rather rough and diffusely re-flecting when deposited at such elevated temperatures. Rhfilms are extremely durable and show even during years ofexposure to air no noticeable change in reflectance. They are,therefore, excellently suited for use as reflectance stan-dards.
Ag and Au that do not wet W should be evaporated from Wor tantalum boats. Films of these materials should also bedeposited at high rates and not thicker than needed to be justopaquely reflecting. Otherwise their surface roughness willbe increased. Their poor adhesion to glass and fused SiO2 canbe overcome by evaporating a binding layer of Nichrome orchromium onto the substrates before the Ag or Au mirrorcoatings are deposited. The smoothness of all evaporatedmetallic mirror films is also affected by the vapor angle ofincidence during evaporation. Holland41 showed that diffusereflecting surfaces form more easily as the vapor incidenceangle is increased. The preparation and properties of di-electric overcoatings will be discussed in connection with themetallic reflecting films on which they are used.
REFLECTANCE OF METALLIC FRONT-SURFACE MIRRORS WITHOUTOVERCOATINGS
Figure 4 shows the UV, visible, and IR normal-incidence re-flectance of freshly deposited opaque metal films of Ag, Al,Au, Cu, Rh, and Pt as reported by Hass and Ritter.4 0 Theonly film material that has high reflectance in all these regionsis Al. Its high reflectance of about 90% extends even into thevacuum UV down to about 1000 A if its surface is completely
Georg Hass
free of oxide.2' 3 The reflectance of the other metal films dropsrapidly in the visible or UV. Al is, therefore, the only filmmaterial that should be used for mirrors and gratings thatrequire high reflectance in the UV. Evaporated Ag films havethe highest reflectance of any mirror coating from the short-wavelength region of the visible to the IR. Therefore a greatamount of research has been performed and is still continuingon techniques for producing adherent and well-protectedfront-surface Ag mirrors. In the IR at X = 10 Am, Al, Au, Cu,and Ag reflect almost equally well, having reflectances of 98.7to 99.5%. At this wavelength even Rh and Pt films reflect97.6% and 96.2%, respectively.
The optical properties of metals are usually characterizedby two parameters, the index of refraction n and the extinctioncoefficient k. The knowledge of these two parameters, whichare called the optical constants or the complex index, N = n- ik, is required to calculate the reflectance of a metal surfaceat various angles of incidence, the effect of surface films on the
WAVELENGTH (pm)
Fig. 4. Reflectance of freshly deposited mirror coatings of Al, Ag,Au, Cu, Rh, and Pt from 0.22 to 10 Am.
100
90 A8.1
60 10 20 350 40 50 60 70 80 90
ANGLE OF INCIDENCE (D,-9,e0
Fig. 5. Calculated reflectance of evaporated Ag as a function of in-cidence angle for 1 = 500 nm and 0 = 10 70m.
10 20 30 40 50 60 70ANGLE OF INCIDENCE (Dqee0)
Fig. 6. Calculated reflectance of evaporated Al as a function of in-cidence angle for X = 546 nm and X = 10 ,um.
Vol. 72, No. 1/January 1982/J. Opt. Soc. Am. 31
Table 1. Values of (n 2 + k 2) and Angle of Incidenceat Which R, Reaches a Minimum for Ag and Al in the
Visible and IR
Reflec- Ag (X = 500 Al (X = 546 Ag (X = 10 Al (X = 10tance nm) nm) Am) 'um)
(n2 + k2
) 8.2 36 4900 5200Angle of Rp 700 800 89.180 89.200
(min)
reflectance, and the phase change on reflection. A rathercomplete list of the optical constants of the metals discussedin this paper with references to their origin has been publishedin a summary article by Hass.42 The article also gives theequations for calculating the reflectances of metals as afunction of angle of incidence, the phase change at a dielec-tric-metal boundary, and the reflectance of metals coated withnonabsorbing surface films of various indices of refraction.The calculated dependence of the reflectances of Ag and Alon the angle of incidence is shown in Figs. 5 and 6. The cal-culations for each metal were made for two wavelengths, onein the visible and one in the IR, by using the equations and listof optical constants published in the above-mentioned arti-cle.42 The reflected intensities follow a certain pattern in allcases: R8, the perpendicular component, increases steadilyfrom the normal-incidence value up to 100% at grazing inci-dence, whereas Rp, the parallel component, first decreases toa minimum and then rises rather rapidly to 100% at 900 angleof incidence. The essential features of the reflectance curvesare given by three quantities:
(1) Normal-incidence reflectance, which is the same forboth components,
R = R = Rp = (n - 1)2 + k2
(n ± 1)2 + k 2
(2)(3)
The angle at which Rp reaches a minimum,The reflectance value of Rp at the minimum.
For most metallic reflecting coatings used from the visible tothe IR it is possible to make an assessment of their reflectancebehavior as a function of incidence angle rather simply with-out resorting to extensive calculations. The ratio k/n and thevalue of (n2 + k2) are the dominating factors for predictingthe reflectance behavior as a function of incidence angle. Itcan be used here to explain the difference between the re-flectance curves of Ag and Al shown in Figs. 5 and 6. Theangle of incidence at which Rp reaches a minimum movescloser to 900 as the value of (n 2 + k 2) increases, as is shownin Table 1.
The ratio k/n has a pronounced effect on the decrease in themagnitude of Rp with angle of incidence. As k/n gets larger,there is a corresponding increase in the minimum value of Rp,which occurs at or close to the principal angle of incidence. Atthis angle the relative phase difference between the s and pcomponents is 900. Ag at X = 500 nm has an extremely highk/n value of 57.4. Therefore, at this wavelength, Rp dropsonly 1.6% between normal incidence and the principal angleof incidence. For Al in the visible with k/n = 7.2, Rp dropsconsiderably lower. The great difference in the minimum
reflectance of Rp for Al and Ag at 10 /im can be explained bythe great difference in their k/n values at this wavelength. Alwith a k/n value of 2.6 has an Rp (minimum) of 47%, whereasAg with k/n = 6.5 drops ony to 73%. In the far IR, where forall metals n and k become large and k/n approaches unity, Rpdrops to a minimum value of 17% close to grazing inci-dence.
REFLECTANCE OF FRONT-SURFACE MIRRORSWITH OVERCOATINGS
General RemarksAll the metal mirrors may be overcoated with protectivelayers, which in some cases also serve to increase or decreasethe reflectance of the underlying metal in certain wavelengthregions. Such overcoatings inevitably alter the intrinsic re-flectance and polarization properties of various metals dif-ferently, because the underlying metals have different opticalconstants. This is demonstrated in Fig. 7, which shows theeffect of single-layer overcoatings on the normal-incidencereflectance of Ag, Al, and Rh at X = 546 nm. The nonab-sorbing surface films have n values from 1 to 3 and were as-sumed to be effectively one-quarter (Rmin) and one-halfwavelength (Rmax) thick. The optical constants of the threemetals reported in Ref. 42 were used for the calculations. Atypical protective coating, such as A1203 with an n value ofabout 1.6, deposited to a thickness of lX/4 or an odd multipleof X/4, will reduce the reflectance of Ag from 98.3 to 96.5%, ofAl from 91.5 to 80%, and of Rh from 78.2 to 55%. This showsclearly that the intrinsic high reflectance of Ag is least affectedby a nonabsorbing surface layer. At the X/2 positions, thereflectance of the three metals is slightly increased by a non-absorbing protective coating, and this is more pronounced forRh than it is for Ag and Al. The reflectance increase causedby X/2-thick surface films also increases with increasing nvalues of the overcoatings.
In Fig. 8, the calculated reflectance of Ag as function ofincidence angle is presented for the perpendicular (R,), theparallel (Rp), and the average (Ray) reflectances at X = 550nm. The calculations were performed for bare Ag and Agovercoated with effectively 1X/4- and 2X/4-thick single-layerfilms of n = 1.6 and a typical high-low-index reflectance-enhancing film pair. It is evident that an evaporated Agmirror without or with a nonabsorbing overcoating exhibitsonly a small change in reflectance at X = 550 nm for angles of
90 A-g -80 I . 80
70
100
90
80
70
60
50
40
INDEX OF REFRACTION
Fig. 7. Calculated reflectance of Ag, Al, and Rh coated with effectiveX/4 or odd multiple of X/4 (Rmin) and X/2 or even multiple of X/4(Rmax) thicknesses of nonabsorbing surface films with n values rangingfrom 1.0 to 3.0 at X = 546 nm.
Georg Hass
I
I
IJ
I
u
32 J. Opt. Soc. Am./Vol. 72, No. 1/January 1982
X- 550 n.
RAV Rp
90Ag +1)I /4. ,1.6
Ag+2>X/4, n.1.6 +IX/4, n2 2.3
800 20 40 60 80 0 20 40 60 80
ANGLE OF INCIDENCE, DEGREES
Fig. 8. Calculated reflectance of Ag and Ag plus three typical non-absorbing surface films as a function of incidence angle at A = 550nm.
incidence ranging from 0 to 900. For all angles of incidence,the polarization Rp/Rs remains close to unity. For a Ag mirrorcoated with a reflectance-enhancing double layer of nj = 1.6(A1203) and n2 = 2.3 (CeO2) (Ref. 43), the normal-incidencereflectance increases from 98.3 to 99.3% and remains virtuallyunchanged up to angles of incidence of 50°.
Figures 9 and 10 contain similarly calculated angular-re-flectance behavior for Al and Rh, respectively. Both metalsshow the same general trend as Ag without and with surfacefilms, but in each case the average reflectances are consider-ably lower than they are for Ag, and there is a much greaterdifference between the Rs and Rp components with increasingangle of incidence. This means that the reflected beams maybe highly polarized at high-incidence angles. The figures alsoshow that the normal-incidence reflectances of Al and Rhmirrors coated with reflectance-enhancing film pairs of nj =1.6 and n 2 = 2.3 are increased from 91.5 to 95.0% and from 78.2to 89.0%, respectively. A more complete discussion of re-flectance-increasing film stacks is presented in a later part ofthis article.
The preceding series of figures clearly shows that evapo-rated Ag films have the highest reflectance in the visible andthat they also introduce less polarization into an optical sys-tem than Al and Rh mirrors when used at angles of incidencegreater than 400.
Evaporated Al Mirrors with OvercoatingsEvaporated Al is undoubtedly the most frequently usedcoating for front-surface mirrors. However, for many mirrorapplications the natural oxide film on Al is too thin to furnishsufficient protection, especially if the mirrors require frequentcleaning. Therefore various methods for overcoating Al withhard and transparent protective layers have been devel-oped.
Thin films produced by evaporation of silicon monoxide(SiO), coatings of SiO 2 or A1203 deposited by electron-gunevaporation, 8 and anodically produced A1203 surface films44
are the most commonly used protective layers for evaporatedAl mirrors.
SiO is usually evaporated from directly heated tantalumor molybdenum boats and containers filled with about 3-mm-thick pieces of SiO. The containers and their chargeshave to be well outgassed before being used to deposit opticalcoatings. The composition and optical properties of siliconoxide layers and the reflectance characteristics of Al coatedwith such protective layers, however, depend greatly on theconditions under which SiO is evaporated. Films deposited
07
. I
0
11
1.ZQ
I
X = 546 nm100
I
ALI
I:
wI
100
90 R,
80 Al 2 X/4, n = 1.6 Al I X/4, n1.6+ 1 X/4, n,= 2.3
70 l lI
0 20 40 60 80 0 20 40 60 80
ANGLE OF INCIDENCE, DEGREES
Fig. 9. Calculated reflectance of Al and Al plus three typical non-absorbing surface films as a function of incidence angle at A = 546nm.
100
0I
S0
80
60
X= 546 nm
I I I I I I F
Rh + I X/4, n = 1.6 X
-/
' I I I I
~RSC R'AV I
- Rh+ 1X/4, n1 =1.6 R U+ 1 X/4, n2 =2.3
l I l I I I l I20 40 60 80
ANGLE OF INCIDENCE, DEGREES
Fig. 10. Calculated reflectance of Rh and Rh plus three typicalnonabsorbing surface films as a function of incidence angle at A = 546nm.
Georg Hass
. I . I I I I
Al, I X/4, n.16 1X
R 1 1 . -R~
I I . I I I
' I ' I ' I I
R i-- f RAV //
RP
Rh\
l I l I l I l I
100 ' II' I ' I I I
10- Rs /180 Rs
60 Rh +2 X/4, n -1.6 /
40 I l I I I l l0 20 40 60 80
at high deposition rates of at least 20 A/sec and low pressuresof less than 1 X 10-5 Torr consist of true SiO. Such films showrather strong absorptance in the UV and the short-wavelengthregion of the visible. They also have a very high index of re-fraction, which reaches about 2.0 in the visible and 1.9 in thenear IR.45 They are therefore not suitable as protective layerson Al if high reflectances in the visible and UV are required.They are, however, excellent antireflection coatings for siliconand germanium in the near IR.46 To produce films withnegligible absorptance in the visible and near IR, low depo-sition rates at rather high pressures of oxygen must be used(rates of 4 A/sec at p = 8-10 X 105 Torr of oxygen). Filmsproduced under such a strongly oxidizing evaporation con-dition, which is called reactive evaporation, consist frequentlyof Si2O3 and have n values of about 1.55 in the visible. Somefilms may have more or less oxygen than Si2O3, depending onthe oxygen pressure and deposition rate. Therefore filmsproduced by reactive evaporation are usually called SiOXcoatings. Figure 11 shows the effect of SiO evaporationconditions on the visible and UV reflectance of silicon-oxide-protected Al mirrors. All three protective coatings areeffectively one-half wavelength thick at X = 550 nm to produce
Vol. 72, No. 1/January 1982/J. Opt. Soc. Am. 33
IOU
I / , P- a 8X 10 5torr OF OXYGEN"6 0- - EOItNRT A/SECLI'
140 I x 165 torrZ / DEPOSITION RATE * 4 A/SEC
X2 P0 --.- P - 8 X 106 torr_ - ,' , ,'DEPOSITION RATE 2 2O A/sEc
01200 300 400 500
WAVELENGTH (nm)600 7- 00
Fig. 11. Effect of SiO evaporation conditions on the visible and UVreflectance of silicon-oxide-protected Al mirrors. Protective coatingseffectively X/2 thick at X = 550 nm.
Fig. 12. Effect of ultraviolet irradiation with a435-W quartz mercuryburner on the reflectance of Al protected with unoxidized and stronglyoxidized silicon oxide coatings. Protective coatings effectively X/2thick at 550-600 nm. Irradiations performed at a distance of 20cm.
the highest reflectance in the visible. The bottom curve showsthe reflectance of Al coated with SiO at low pressure and ahigh deposition rate of 20 A/sec. This true SiO film produceshigh absorptance for all wavelengths shorter than 500 nm andeven some absorptance at longer wavelengths. A mirrorcoated in this way has a yellow appearance. The middlecurve, which still exhibits high absorptance below 400 nm, wasobtained by evaporating SiO onto Al at a much lower depo-sition rate and at pressures of 1 X 10-5 Torr and without theintroduction of oxygen. The top curve represents the re-flectance of Al coated with a strongly oxidized silicon oxidefilm (SiO,) prepared by the use of a low deposition rate of 3A/sec and an oxygen pressure of 8 X 10-5 Torr. This SiOQprotective layer gives the Al mirror high reflectance down to300 nm. Al mirrors that are to be used in the visible and nearUV should therefore be coated with reactively evaporatedSiOX. These oxidized SiO, films are less dense than those oftrue SiO because they are deposited at lower rates and ratherhigh pressures. However, they make excellent protectivecoatings since they adhere strongly to Al and harden rapidlyby further surface oxidation when exposed to air. Below 300nm SiO, films still show rather high absorptance, which in-creases with decreasing wavelength. This had limited theusefulness of SiO, -coated Al mirrors until Bradford andHass5' 7 discovered that UV irradiation completely eliminatesthe UV absorption in these SiO, films and thus makes itpossible to produce well-protected Al mirrors with 91% re-flectance down to 200 nm.
Figure 12 shows the effect of UV irradiation on the visibleand UV reflectance of evaporated Al protected with true SiO
and strongly oxidized SiO. films. The UV irradiations wereperformed with a 435-W Hanovia quartz mercury burner ata distance of about 20 cm. Both films are again effectivelyone-half wavelength thick at X = 550 nm. It can be seen that5 h of UV treatment completely eliminates the initial high-UVabsorptance of the SiO, film deposited slowly at a high pres-sure of oxygen, whereas the same UV irradiation has littleeffect on the coating produced at much lower pressure andwithout oxygen. Additional experiments with much thickerreactively deposited SiO. (t > 1 am) films showed that theirUV absorptance could also be completely removed by UV ir-radiation.
Two completely different effects were found to be respon-sible for the optical changes in reactively deposited SiO,caused by UV treatments:
(1) Ultraviolet irradiation rearranges the oxygen atomsand molecules gathered during the evaporation process andforms well-defined silicon oxide molecules, which are highlytransparent in the UV.
(2) Exposure to air or oxygen increases the oxygen contentin the deposited films, which removes their UV absorptanceand decreases their index of refraction.
An alternative method of increasing the oxidation state ofreactively evaporated SiO films and eliminating their UVabsorptance without UV treatment has been described byHeitmann. 4 7 He performed his evaporations in ionizedoxygen. The ionized gas is produced in a gas-discharge tubeof high current density that is located inside the vacuumsystem. The ionized gas emerges from a nozzle in the wall ofthe discharge tube directly into the vacuum area where theevaporations are performed. Films produced in this way arefree of absorption down to 190 nm and have refractive indicesidentical with those of the fused-silica substrate. A completearrangement of Heitmann's vacuum system suitable for per-forming reactive evaporations in ionized gases is described inRef. 47.
SiO2 and A1203 films properly deposited by electron-beamevaporation are nonabsorbing in the visible and UV8 andshow, when deposited onto evaporated Al, only the expectedinterference maxima and minima in the reflectance curves.
For Al front-surface mirrors that are to be used in the IR,it is important to study the effect of absorptances in the pro-tective layers on their JR reflectance. All silicon oxides, SiO,SiOM, and SiO2, have strong absorption bands in the important8-12-gm atmospheric window region. In practice, siliconoxide layers 1000 to 2000 A thick have been found to be thickenough to give adequate protection against abrasion and hu-midity. The IR reflectances of such protected Al mirrors havebeen measured many times at close to normal incidence,42A45A48and it has been found that the normal-incidence reflectanceof Al coated with such thin protective coatings is nearly thesame as that of uncoated Al in the 8-12-gm region. To ex-plain this effect, one has to remember that light reflected fromhighly reflecting metal surfaces produces a standing wavepattern with a node, or an area of zero vibration, at the metalsurface. Under these conditions, absorbing films that aresmall compared with the wavelength produce little absorp-tance. If thicker protective layers of silicon oxides are applied,high reflectance decreases can be observed. An Al mirrorcoated with a 1-4-m-thick coating of SiO shows at normal
U S 81 1 , as I 'S ! i , l | 1 1
J - - - - - - - -
Georg H~ass
34 J. Opt. Soc. Am./Vol. 72, No. 1/January 1982
q.,
C.
I--A-J
IL.Ad
9WAVELENGTH (Im)
10
Fig. 13. Calculated reflectance of Al coated with 1500 A of SiO2 at600 angle of incidence in the wavelength region from 8 to 10 gm.
incidence a reflectance of almost zero at 10 pm. The wave-lengths at which thick silicon oxide films produce the greatestreflectance decreases depend on the composition or ratio ofSi to 0. For SiO the greatest reflectance decreases occurbetween 10.0 and 10.2 gim, for Si2O3 at 9.7-11.5 gim, and forSiO2 at 9.4-12.5 gim. Al mirrors coated with thick siliconoxide layers are therefore unsuitable as IR reflectance coatingsfor the 8-12-gm region. However, their IR-absorptionproperties have been used to produce surface coatings havinglow solar absorptivity and controllable high thermal emissivityfor controlling the temperature of many satellites.4 9
-'
All the JR-reflectance data of Al coated with thin protectivelayers of silicon oxides that were reported above showed thatthe normal-incidence reflectances of such protected mirrorshave nearly the same high reflectance as bare Al in the 8-12-gm region, provided that the thickness of the oxide layersis kept thin enough (t = 1000-2000 A). However, this situa-tion changes drastically if such mirrors are used at angles ofincidence greater than 400. At high angles of incidence theirreflectances are far below those of their normal-incidencevalues and that of bare Al at the same angle of incidence. 9
This is demonstrated in Fig. 13, which shows the calculatedreflectance of Al coated with 1500 A of SiO2 at a 600 angle ofincidence. The optical constants of SiO2 published byBoeckner 5 2 and of Al listed in Ref. 42 were used for the cal-culations. The figure shows that the Al-SiO2 film combina-tion has at a 600 angle of incidence its lowest reflectance atabout 8.1 gim. At this angle Ray is about 51%, and the R8 andRp values are 99.0 and 2.5%, respectively. That means thatat this wavelength and angle of incidence, the reflected beamis almost completely polarized and that the low Ry is causedsolely by the reflectance drop of the R, component. At a 450
angle of incidence, Rav, R8, and Rp are 60, 98.5, and 25%, re-spectively. The region of high and moderate reflectance de-crease reaches from 8.05 to about 9.6 gm for mirrors used at600 angle of incidence. Al + SiO, and Al + SiO show a similarreflectance behavior. Their reflectance minima at 45 and 600occur at longer wavelengths and are not so low as those causedby an SiO2 surface layer. This angle-of-incidence effect wasfound to be most severe when the optical constants of the thinprotective layers have values of n less than unity and k valuesbetween 0.1 and 0.8 (N = n - ik). Such optical constants ofdielectric materials are always found at the wavelengthsslightly shorter than the reststrahlen high-reflectance bands.If n of the surface layers is larger than unity, there is no greatreduction in reflectance at a 450 angle of incidence no matterwhat the k value is. This is, of course, only true if the pro-tective layers are thin and the wavelength region of interestis 8-12 Am. The effect of reflectance reduction caused by thinsurface layers with n and k values less than unity is not pe-culiar to Al but is essentially the same for any highly reflectingmetal and depends almost entirely on the optical constantsof the protective layers. Table 2 shows the visible and IRreflectance and polarization characteristics of uncoated Agand of Ag overcoated with 300 A of A1203 and with 1500 A ofSiO2 at 0 and 450 angles of incidence.5 3 The thin A1203 layeris used to ensure good adherance between the Ag and SiO2film. In the visible at A = 550 nm, where the surface films arenonabsorbing, the uncoated and overcoated Ag films havepractically the same reflectance at 0 and 450 angles of inci-dence, and the Rp/R8 values are close to unity. In the IR atA = 8.1 gim, the uncoated and overcoated mirrors have thesame reflectance at normal incidence. At a 450 angle of in-cidence, however, the reflectance of the protected Ag filmdrops to Rp = 30.6% and Rv = 64.7% and decreases the Rp/R8value to 0.31, whereas the bare Ag film shows practically nochanges in reflectance between 0 and 450 angles of incidence.Table 2 shows that the drop of Ray is again caused entirely bythe decrease of the Rp component and that R8 remains highfor all angles of incidence.
In our search for a protective film that would not reduce thereflectance of Al in the 8-12-gm region at higher angles ofincidence, electron-beam-evaporated A1203 appeared to bepromising. It forms hard and adherent layers on Al (Ref. 8)and is also one of the few protective coatings that adhere wellto Ag.53' 54 The optical constants of a thin A12 03 film in the8-12-gm region published by Harris 55 were reported to be n= 1.4-1.7 and k = 0.0-0.7. Protective layers with such optical
Table 2. Calculated Visible versus Infrared Reflectance and Polarization Characteristics for Uncoated Ag andAg + 300-A A12 03 + 1500-A SiO2 a
R R, reflectance; R8, perpendicular component; Rp, parallel component; Ray, average reflectance.
Georg Hass
Vol. 72, No. 1/January 1982/J. Opt. Soc. Am. 35
1 40° INCIDENCE | '230OA I
70
7 8 9 10 11 12 13 14 15
WAVELENGTH 1pm A
Fig. 14. Calculated reflectance of Al and measured reflectance ofAl + A12 03 films of three different thickness at 40 and 600 incidenceangles from 7 to 15 ,im.
constants should not cause any considerable reflectance de-crease at higher angles of incidence in this region. However,there are other experimental measurements that cast doubton the validity of the above-reported optical constants ofA1203 in the wavelength region from 8 to 12 ,m. In one of theexperiments, the reflectance of single crystalline A1203 (sap-phire) plates was measured from 7 to 15 ,um. The resultsshowed that at 9.5 Am the measured reflectance is almost zero.This can occur only if the value of n is close to 1 and k is small.Therefore, at X = 10 ,m, n should decrease to values of lessthan 1 since sapphire has the well-known broad reststrahlenreflectance at 14-15 gm. Evaporated A12 03 is amorphous andmuch less dense and has n values much lower than those ofsapphire. The region over which Al coated with evaporatedA1203 has low reflectance at high angles of incidence thereforemay be quite large. To decide whether this prediction wastrue, we coated Al with A1203 layers of three different thick-nesses, 1000, 1750, and 2300 A, by electron-gun evaporationin high vacuum. The reflectances of the samples were mea-sured with unpolarized radiation in the. 7-15-,m region atclose to 0, 40, and 600 angles of incidence. The results areshown in Fig. 14, which also includes the reflectance valuesof unprotected Al in the same wavelength region and at thesame angles of incidence.' 0 The reflectance curves show thatthe reflectance of uncoated Al remains high throughout theentire region from 7 to 15 ,m and at all three angles of inci-dence, whereas A12 03 -protected Al mirrors exhibit consider-able reflectance decreases over an extended wavelength region.This reflectance decrease gets progressively worse with in-creasing film thickness and angle of incidence. The greatestreflectance decrease of all A12 03 -protected Al mirrors occursin the 10.6-10.7-Mm region.
For the previously reported mirror coatings protected withsilicon oxides, the minimum reflectance at higher angles ofincidence occurred at shorter wavelengths since their rest-strahlen reflectances are located at shorter wavelengths.Another important difference is that the low-reflecting regionof A1203 -protected Al is much broader than that of Al coatedwith silicon oxide layers. For example, Al + 1500 A of SiO2
used at a 600 angle of incidence has an average reflectance ofless than 90% from about 8 to 8.8 Am, whereas a coating of Al+ 1750 A of A1203 reflects at the same angle of incidence lessthan 90% from 9.8 to 13.3,um. It should be mentioned againthat in the regions where A120 3- and silicon-oxide-protectedAl mirrors show low reflectance at high angles of incidence,the reflected IR radiation is highly polarized since the re-
flectance decreases are caused entirely by the decrease of theR, component, whereas R, increases with increasing angle ofincidence from about 99.0 to 100% at grazing incidence. Thiseffect makes it possible to prepare efficient reflection-typepolarizers for various regions in the far IR.
Reflection-Type Polarizers for the 10.6-Mm CO2 LaserRadiation-Emission Line Using A1203-Coated Al MirrorsAs was previously mentioned, the measured average reflec-tance of Al coated with A1203 has at high angles of incidencea minimum at about 10.6 Aum. The reflectance of the Rpcomponent must be considerably lower since R, can be as-sumed to be about 99% reflecting. It can be calculated byusing the relation
Rp = 2Rav-Rs
In Fig. 15, Rav and Rp at 10.6 ,m for Al coated with variousthicknesses of A1203 at 40 and 600 angles of incidence arelisted, and schematic diagrams of four- and three-mirror po-larizers are shown.'1 The lowest value of Rp and thereforethe highest degree of polarization is obtained with the thickestA1203 film used and at the highest angle of incidence, 600, forwhich R was measured. The fact that Rp becomes low whileR, stays high makes it possible to design efficient reflection-type polarizers for 10.6-Atm CO2 laser radiation. The opti-mum angle for highest polarization depends on the opticalconstants n and k of the A12 03 surface layer. In order to de-termine this optimum angle the optical constants of A1203were derived from a simultaneous fit of the measured data ofreflectance at normal incidence of a sapphire plate and of re-flectance as a function of angle of incidence of Al + A1203.This analysis yielded values of n 0.65-0.50 and k 0.3-0.5for the wavelength region from 10.6 to 10.8 Am. The optimumangle of incidence for which Rp is a minimum for these opticalconstants is about 75°. However, as is shown below, usefulpolarizers can be made that use angles of incidence less than750
The four-mirror polarizer shown in Fig. 15 uses four Al +2300 A of A1203 mirror coatings, all at a 600 angle of incidence.This results in a polarizer with a ratio R4 /Rp of about 500 atX = 10.6 Am. This shows that an efficient polarizer for10.6-,m radiation can be made that uses four identical mirrorsin a simple arrangement for which the outgoing beam is in the
Refle-tan-- at 10.6O o of Al Coated with VaiousThiknessesof AI,0 at Angle of Incidenc 40° ad 60°
Ooide R,Thikness 40 60- 40 60
1000 A 95.2 84.2 86.0 69.4
1750 7 83.9 67.0 68.8 35.0
2300 A 79.0 60.0 590 21.0
M0
M3
SOhematic diagamof a 4-mirror polaizen. All angles ofiidenceae 60°.
Ma\ T /
Schematic diagram of a 3-mirorg polariaer. Angles ofincidenceae gien by 2a - ) = 90°.
Fig. 15. Reflection-type polarizers for 10.6-,m CO2 laser radiationusing four and three A12 0 3 -coated Al mirrors.
Georg Hass
36 J. Opt. Soc. Am./Vol. 72, No. 1/January 1982
same direction as the incident one. In addition, since R, =99%, the throughput for the s component is very high (96%).Accurate alignment of the mirrors can easily be achieved bythe use of a laser of visible radiation.
Another arrangement, which requires only three Al mirrorscoated with A1203 films, is shown in Fig. 15. Here the beamdirection can also be maintained without deviation, but theoutput beam will be inverted with respect to the input one.For most applications this may not be important. For thisconfiguration, in contrast to the four-mirror design, the anglesare not equal but must obey the relationship 2a - / = 900 tomaintain an undeviated beam direction. If a is chosen to be600, 3 must be 300. However, for this choice of angles, thecalculated ratio of R0IRp for the same A12 03-coated mirrorswould be only 30, compared with the value of 500 for thefour-mirror design. If, on the other hand, a is chosen to be700, for which /3 would be 500, the ratio of R0/Rp would benearly 400, or almost as high as that of the four-mirror po-larizer. The throughput for the s component for the three-mirror design would be about 97%.
The results for the above mirror designs were based onmeasurements of Al protected with A1203. As was notedpreviously, the phenomenon of reduced reflectance at higherangles of incidence depends on the properties of the protectivelayers and not of the underlying metal. It is most pronouncedwhen the optical constants of the protecting layer n and k areboth much less than 1. This is typical for ionic solids at ashort-wavelength side of their reststrahlen region. Here ndecreases usually to values far below 1 as the reststrahlen re-flectance region is approached, while k remains small. Thereare a large number of dielectric materials available that havereststrahlen bands over a very wide wavelength region fromabout 9 Am to beyond 50 Am. This offers the possibility ofchoosing almost any wavelength in this region for which areflection-type polarizer can be made to operate efficiently.The width of any particular polarizing region will, of course,depend on the optical properties and the thickness of the di-electric overcoating. Some will be rather narrow (e.g., SiO2),and others will be much broader (e.g., A12 03 and MgO).
Therefore it can be concluded that rather simple and in-expensive reflection-type polarizers can be built for the IRthat have R3 /Rp ratios in excess of 500, s-componentthroughputs greater than 96%, and large apertures. Opera-tion in many bands in the wavelength region from 9 gim tobeyond 50 gAm can easily be achieved by proper selection ofthe dielectric overcoating material by using a frame in whichmirrors with various surface layers can be easily replaced andcorrectly mounted.
Protected Al Mirrors with High Reflectance in the 8-12-,m Region from Normal to High Angles of IncidenceIn the preceding paragraphs it has been pointed out that Almirrors protected with rather thin (1000-2000-A) silicon oxideand A12 03 films have greatly reduced reflectance in the 8-12-gm region when used at angles of incidence greater than40°. The analysis of this effect also showed that the reductionin reflectance at high angles of incidence occurs in the wave-length region where n is less than 1 and k is small.
A survey of possible alternative dielectric materials thatmight lead to durable mirrors with high reflectance in the8-12-gm region at higher angles of incidence led to a study ofyttria (Y203 ) and hafnium dioxide (HfO2).12 Heitmann5 6 has
100,
4' 40' .d 60'<9 g 60"
854060
75
7 8 9 10 11 12 13 14WAVELENGTH (pl)
Fig. 16. Measured reflectance of Al + Y20 3 and Al + HfO2 at 40 and600 incidence angle from 7 to 14 ,4m. Protective coatings -X/2 thickat X = 550 nm.
shown that reactively evaporated yttria and scandia films havelow absorptance in the 8-12-,m region with excellent me-chanical stability and chemical durability. Therefore Almirrors overcoated with evaporated Y203 and HfO2 wereprepared for IR-reflectance measurements. The evaporationof HfO 2 was made from a copper hearth with an electron gun,and a tungsten boat was used for the Y20 3 evaporation. TheHfO2 and Y203 layers were each about one-half wavelengththick at 550 nm.
The resulting HfO2- and Y202-protected Al mirror coatingsare hard and adherent. The IR reflectance at near-normalincidence is essentialy the same as that of uncoated Al in the8-12-,m region. Measured reflectance curves at 40 and 600angles of incidence are shown in Fig. 16, along with values foruncoated Al in the wavelength region from 7 to 14 gim. Almirrors coated with Y203 have reflectances nearly as high asuncoated Al throughout the entire 7-14-gm region at both 40and 600. The reflectance of Al mirrors coated with HfO2 isalso high to a wavelength of about 12 gm but begins to fall offtoward longer wavelengths. In addition, the two protectivelayers described here, Y203 and HfO2, are nonabsorbing downto the UV when properly deposited. 5 6' 57 Y203 and HfO2 arejust two of several materials that can be used to form durableprotective coatings for Al mirrors that have high reflectancein the 8-12-gm region for all angles of incidence from 0° togreater than 60°.
Rh and Al Mirrors with Reflectance-Enhancing SurfaceLayersAs was previously mentioned, evaporated Rh films are ex-cellently suited for use as front-surface mirrors because theyare extremely hard and chemically durable. However, theiradherence to glass and fused-silica substrates is often poor.Even films deposited in oil-free ion-titanium-pumped vacuumsystems on substrates of 3000 C may frequently be lifted withScotch tape. This poor adherence of Rh coatings can be easilyovercome by evaporating 20-25-A-thick Nichrome films ontothe glass or silicon substrates before the Rh mirror coating isdeposited. This thin adherence-increasing Nichrome innerlayer was found to have no effect on the mechanical andchemical durability and on the reflectance of evaporated Rhmirrors. Its thickness can be monitored by transmittancemeasurements during its deposition by using monochromaticlight of X = 546 nm. Nichrome should be deposited until thetransmittance of the substrate decreases from 92% to about80%.
For some applications, the visible reflectance of plain Rhmirrors is too low. By applying pairs of dielectric coatings
Georg Hass
Vol. 72, No. 1/January 1982/J. Opt. Soc. Am. 37
with alternately low and high indices of refraction, their re-flectance can be enhanced over a rather broad region. Thelow-index film (nL) adjacent to the metal must be effectivelyX/4 and all other films truly X/4 thick to obtain maximumreflectance at normal incidence.
Equations for calculating the reflectance enhancement ofmetals coated with quarter-wave pairs of nL + nH dielectricfilms, including the absolute phase change at the boundaryof the metal and first dielectric layers, are published in Ref.42.
Figure 17 shows the visible reflectance of evaporated Rhdeposited at 3000 C with and without a reflectance-enhancingfilm pair of SiO2 and TiO2 . The reflectance curve of anevaporated Al mirror is shown on the same graph for com-parison. The SiO2 film was produced by electron-beamevaporation, and the TiO2 layer was prepared by depositingmetallic Ti and heat oxidizing it to TiO2 at 4200C. TiO 2 filmsproduced in this way consist of rutile and have the highestindex of refraction (n _ 2.7) and best durability. 58 To obtainthe mirror coating with the reflectance shown in Fig. 17, SiO2is first evaporated onto freshly deposited Rh until the re-flectance decreases to a minimum at about 530 nm. Then Tiis deposited until the transmittance of an uncoated monitoringglass decreases from 92 to 6% at the same wavelength. Thisresults, after the heat treatment, in a TiO2 layer one-quarterwavelength thick at about 530 nm. The maximum reflectanceof Rh-SiO 2-TiO 2 mirror coatings is 93.1% and is even higherthan that of plain Al over most of the visible region.3 9
Rh mirrors with and without reflectance-enhancing surfacefilms of SiO 2 + TiO 2 were boiled for 1 h in 5% salt water andplaced for 10 h in 10% NaOH and 10% HCl acid. None of themirrors showed any damage or change in reflectance afterthese treatments. The heat resistance of an SiO2 -TiO2 -overcoated Rh mirror was found to be better than that of bareRh. During a 16-h heat treatment in air at 4000C, the re-flectance of plain Rh decreased at X = 550 nm from 78.2 to68%, whereas the SiO2 -TiO 2-overcoated Rh mirrors showedno reflectance loss during the same heat treatment and duringexposure to even higher temperatures (450-500°C). Theformation of Rh2O3 on the surface of bare Rh was found to beresponsible for its reflectance decrease. This oxidation of Rhoccurs only at high temperatures.
There are many (nL + nH) X/4 film pairs available to in-crease the reflectance of metals from the UV to the IR. Filmsof SiO2 + HfO 2 are suitable for increasing the reflectance ofmetals in the UV. 5 7 MgF2 + CeO2 , A1203 + TiO2, and reac-
100 I9 -> : - -. --- h + Si( 2+ T;O2
ILIii
4000 5000 6000 0 7000WAVELENGTH (A )
Fig. 17. Visible reflectance of Rh with and without a reflectance-enhancing film pair of sio 2 and TiO2 . Reflectance of Al is includedfor comparison.
C.)
C.)
-J
iLLC.)
a.
ni I I I I I I I I I450 500 550 600
WAVELENGTH (nm)650 700
Fig. 18. Visible reflectance of Al with and without a reflectance-enhancing film pair of MgF2 and CeO2.
tively deposited SiOx + TiO2 are the most frequently usedreflectance-enhancing film combinations for the visible.Figure 18 shows the visible reflectance of Al with and withouta reflectance-enhancing film pair of MgF2 + CeO2.42 Theovercoated Al film has at normal incidence a maximum re-flectance of 97% at X = 550 nm. If two films pairs of MgF 2 andCeO2 are applied, the maximum reflectance increases to 99.0%.A rather complete discussion of the preparation and proper-ties of dielectric films suitable for use as protective and re-flectance-enhancing coatings on metals has been publishedby Ritter.13
Multilayer quarter-wave stacks consisting of alternatingfilms of equal optical thickness but different refractive indicesplay an important role in the preparation of laser mirrors.Most laser mirrors are deposited onto well-polished dielectricsubstrates and should have high reflectance with extremelylow absorption. The film materials used for laser mirrorsshould have a large difference between the values of their re-fractive indices to obtain high reflectance with the leastnumber of layers. Because of the demanding requirementsfor low absorption, scattering losses, and stability, not manyfilm materials have been found to be practical for preparinglaser mirror coatings, and these film materials have to be se-lected and tested anew for different lasers. A complete dis-cussion of substrates, film materials, and combinations forlaser mirrors as well as their stability and susceptibility todamage is outside the scope of this article. A good survey ofthis field has been published by Ritter. 59 His article refer-ences numerous publications dealing with the above topics.
WATER ABSORPTION IN EVAPORATEDDIELECTRIC FILMS
It is well known that many evaporated dielectric film mate-rials, such as SiO 2, SiO,, and MgF 2, absorb water when ex-posed to air. This is especially true for coatings deposited atlow rates and on unheated substrates. This absorbed waterdecreases the reflectance of a protected metal mirror signifi-cantly in the region close to 3 ,m where water has the highestextinction coefficient. An excellent way to study whether amaterial absorbs water and whether the water is absorbed atthe surface or penetrates through the surface film is to depositfilms X/4 and X/2 thick at X = 3 gm on Al and measure thereflectance of the combined coatings at X = 3,m. If there is
Georg Hass
38 J. Opt. Soc. Am./Vol. 72, No. 1/January 1982
wC.z
-JU-w
50[
0 1000THICKNESS (A)
Fig. 19. Calculated reflectance at 3 ,um of Al coated with SiO. filmsX/4 and X/2 thick at 3 jm having water absorbed on the SiOX surfaceonly.
100-
80
23 4WAVELENGTH (plm)
Fig. 20. Measured reflectance of Al coated with SiOX films X/4 andX/2 thick at 3 ,um in the wavelength region from 2 to 4 ,um.
no water present in the film, only a shallow interferenceminimum of Al + X/4-thick SiOX will be observed. Figure 19shows the calculated reflectance of Al coated with A/4- andA/2-thick films of SiOX with water of different thicknessesabsorbed on the surface only. The optical constants of waterpublished by Hale and Querry,6 0 n 1.3 and k 0.3 at A =2.95 ,um, were used for the calculations. The curves show thatthe reflectance at A = 3 ,um of Al coated with A/4-thick SiOXis greatly decreased by water absorbed at the surface, whereasAl + A/2-thick SiOX shows no change in reflectance even forthick absorbed water layers. The situation changes com-pletely if the absorbed water penetrates into or completelythrough the SiOX protective coatings. This is shown in Fig.20, which exhibits the measured reflectance of Al coated withA/4- and A/2-thick coatings of SiOX in the wavelength regionfrom 2 to 4 ,um. The measured curves show clearly that thereflectance of Al protected with A/4 and A/2 SiOX films isgreatly decreased at about 3 ,um. This means that SiQ, filmsabsorb water and that the water penetrates into the SiOXcoating. Increasing the substrate temperature during thedeposition decreases the amount of absorbed water, whichresults in a smaller reflectanlce decrease. To determinewhether the absorbed water penetrates uniformly all the waythrough the SiOX films to the underlying Al layer, calculations
X/2
I H201
I Si 0x I
Al
w
z< 50
-JU-w
0 0.05EXTINCTION COEFFICIENT
0.1
Fig. 21. Calculated reflectance at 3 ,um as a function of the extinctioncoefficient k with n 1.50 for Al coated with surface films X/4 andX/2 thick at 3 Am.
§ | l X- - - -- -
100
u
Georg Hass
. - of reflectance as a function of the extinction coefficient k forX/4- and A/2-thick films were made. The curves are shownin Fig. 21. The measured minimum reflectance value of Alcoated with X/4-thick SiO, was found to be 71%, whereas thatof Al + X/2 of SiO, was 77.5% (see Fig. 20). Both reflectancevalues agree with the calculated ones obtained with uniformcoatings of k = 0.04 and n = 1.50. It can therefore be con-cluded that water penetrates through the entire rather thickSiOx coatings and forms homogeneous SiO,-H 2O films.
MgF2 is probably the most common film material for usein single-layer and as a component in multilayer antireflectioncoatings. It is also employed as a protective coating for Al inthe vacuum UV.34 Its water absorption as a function ofsubstrate temperature has been studied by various meth-ods.61- 63 Measurements of the refractive index of MgF2 invacuum directly after its deposition and again after exposureto air permits the determination of its packing density andwater absorption. It was found that MgF2 films depositedat room temperature have a packing density of 0.82 and a re-fractive index n of only 1.32. Water absorption in air raisesn to 1.38 since holes of n = 1 are filled with water of n = 1.33.Higher substrate temperatures increase the refractive index,and bulk values of n = 1.39-1.40 are obtained at substratetemperatures of 270-340'C. High-temperature films of thistype are hard and do not change when exposed to air. For thisreason durable antireflection coatings of MgF2 can be ob-tained only by evaporation onto heated substrates.
Investigations have shown that the high water content ofSiO, and MgF2 films deposited at room temperature afterexposure to air consists of two parts: one part, representingabout 70%, is reversible, whereas the remaining part of about30% is irreversible and chemically bound in the dielectricfilms.59. 64 That means that about 70% of the absorbed waterin the films is released when the coated samples are placedback into a high-vacuum system. However, the released 70%is immediately reabsorbed when the films are again exposedto air.
ZnS and true SiO deposited at high rates and low pressuresare two of the few film materials that exhibit bulk values ofthe refractive index and density when evaporated onto sub-strates at room temperature. Their optical properties mea-sured in vacuum and after exposure to air are practicallyidentical.
Vol. 72, No. 1/January 1982/J. Opt. Soc. Am. 39
A more complete discussion of the packing density and re-fractive indices of many fluoride, sulfide, and oxide films hasbeen published by Ritter.13' 65
REFERENCES
1. G. Hass and R. Tousey, J. Opt. Soc. Am. 49, 593 (1959).2. R. P. Madden, in Physics of Thin Films, Vol. 1, G. Hass, ed.
(Academic, New York, 1963), pp. 123-186.3. G. Hass and W. R. Hunter, in Physics of Thin Films, Vol. 10, G.
Hass and M. H. Francombe, eds. (Academic, New York, 1978),pp. 72-166.
4. G. Hass and N. W. Scott, J. Opt. Soc. Am. 39, 179 (1949).5. A. P. Bradford and G. Hass, J. Opt. Soc. Am. 53, 1096 (1963).6. E. Ritter, dissertation (University of Innsbruck, Innsbruck,
Austria, 1958).7. A. P. Bradford, G. Hass, M. McFarland, and E. Ritter, Appl. Opt.
4, 971 (1965).8. J. T. Cox, G. Hass, and J. B. Ramsey, J. Phys. (Paris) 25, 250
(1964).9. J. T. Cox, G. Hass, and W. R. Hunter, Appl. Opt. 14, 1247
(1975).10. J. T. Cox and G. Hass, Appl. Opt. 17, 333 (1978).11. J. T. Cox and G. Hass, Appl. Opt. 17, 1657 (1978).12. J. T. Cox and G. Hass, Appl. Opt. 17, 2125 (1978).13. E. Ritter, in Physics of Thin Films, Vol. 8, G. Hass and M. H.
Francombe, eds. (Academic, New York, 1975), pp. 1-49.14. A. P. Bradford, G. Hass, and M. McFarland, Appl. Opt. 11,2242
(1972).15. R. Tousey, J. Opt. Soc. Am. 29, 235 (1939).16. I. Simon, J. Opt. Soc. Am. 41, 336 (1951).17. R. P. Madden and L. R. Canfield, J. Opt. Soc. Am. 51, 838
(1961).18. J. Strong, Procedures of Experimental Physics (Prentice-Hall,
Englewood Cliffs, N.J., 1938), p. 376.19. H. E. Bennett and W. E. Koehler, J. Opt. Soc. Am. 50, 1
(1960).20. H. E. Bennett and J. M. Bennett, in Physics of Thin Films, Vol.
4, G. Hass and R. E. Thun, eds. (Academic, New York, 1967), pp.1-96.
21. H. E. Bennett, Naval Weapons Center Rep. TP-6015 (NavalWeapons Center, China Lake, Calif., 1978), p. 87.
22. D. M. Gates, C. C. Shaw, and D. Beaumont, J. Opt. Soc. Am. 48,88 (1958).
23. L. Harris and P. Fowler, J. Opt. Soc. Am. 51, 164 (1961).24. D. R. Herriott and H. J. Schulte, Appl. Opt. 4, 883 (1965).25. D. L. Perry, Appl. Opt. 4, 987 (1965).26. 0. Armon and P. Baumeister, Appl. Opt. 17, 2913 (1978).27. G. Hass, J. Opt. Soc. Am. 45, 945 (1955).28. G. Hass, W. R. Hunter, and R. Tousey, J. Opt. Soc. Am. 46,1009
(1956).29. G. Hass and J. Waylonis, J. Opt. Soc. Am. 51, 719 (1961).
30. H. E. Bennett, M. Silver, and E. J. Ashley, J. Opt. Soc. Am. 53,1089 (1963).
31. B. Feuerbacher and W. Steinmann, Opt. Commun. 1, 81(1968).
32. E. T. Hutcheson, G. Hass, and J. K. Coulter, Opt. Commun. 3,213(1971).
33. W. Walkenforst, Z. Tech. Phys. 22, 14 (1941).34. P. H. Berning, G. Hass, and R. P. Madden, J. Opt. Soc. Am. 50,
586 (1960).35. R. P. Madden, L. R. Canfield, and G. Hass, J. Opt. Soc. Am. 53,
620 (1963).36. A. P. Bradford, G. Hass, J. F. Osantowski, and A. R. Toft, Appl.
Opt. 8, 1183 (1969).37. H. L. Rook and R. C. Plumb, Appl. Phys. 1, 11 (1962).38. R. B. Love and W. K. Bower, J. Vac. Sci. Technol. 11, 1124
(1974).39. J. K. Coulter, G. Hass, and J. B. Ramsey, J. Opt. Soc. Am. 63,1149
(1973).40. G. Hass and E. Ritter, J. Vac. Sci. Technol. 4, 71 (1967).41. L. Holland, J. Opt. Soc. Am. 43, 376 (1953).42. G. Hass, in Applied Optics and Optical Engineering, R. Kings-
lake, ed. (Academic, New York, 1965), Vol. III, pp. 309-330.43. G. Hass, J. B. Ramsey, and R. Thun, J. Opt. Soc. Am. 48, 324
(1958).44. G. Hass, J. Opt. Soc. Am. 39, 632 (1949).45. G. Hass and C. D. Salzberg, J. Opt. Soc. Am. 44, 181 (1954).46. J. J. Cox and G. Hass, J. Opt. Soc. Am. 48, 677 (1958).47. W. Heitmann, Appl. Opt. 10, 2414 (1971).48. H. E. Bennett, J. M. Bennett, and E. J. Ashley, Appl. Opt. 2,156
(1963).49. G. Hass, L. R. Drummeter, and M. Schach, J. Opt. Soc. Am. 49,
918 (1959).50. A. P. Bradford, G. Hass, J. B. Heaney, and J. J. Triolo, Appl. Opt.
8, 275 (1969).51. A. P. Bradford, G. Hass, J. B. Heaney, and J. J. Triolo, Appl. Opt.
9, 339 (1970).52. C. Boeckner, J. Opt. Soc. Am. 19, 7 (1929).53. G. Hass, J. B. Heaney, J. F. Osantowski, and J. J. Triolo, Appl.
Opt. 14, 2639 (1975).54. G. Hass, J. B. Heaney, and J. J. Triolo, Opt. Commun. 8, 183
(1973).,55. L. Harris, J. Opt. Soc. Am. 45, 27 (1955).56. W. Heitmann, Appl. Opt. 12, 394 (1973).57. P. Baumeister and 0. Arnon, Appl. Opt. 16, 439 (1977).58. G. Hass, Vacuum 2, 331 (1952).59. E. Ritter, in Laser Handbook, F. T. Arecchi and E. 0. Schulz-
DuBois, eds. (North-Holland, Amsterdam, 1972), pp. 899-921.60. G. M. Hale and M. R. Querry, Appl. Opt. 12, 555 (1973).61. E. Ritter and R. Hoffmann, J. Vac. Sci. Technol. 6, 733 (1969).62. D. Hackman, Opt. Acta 17, 659 (1970).63. W. Heitmann and G. Koppelmann, Z. Angew. Phys. 23, 221
(1967).64. H. Koch, Phys. Status Solidi 12, 533 (1969).65. E. Ritter, Appl. Opt. 15, 2318 (1976).
