Edmund Optics(cid:15)(cid:3)(cid:44)(cid:81)(cid:70)(cid:17)(cid:3)
`(cid:40)(cid:91)(cid:75)(cid:76)(cid:69)(cid:76)(cid:87)(cid:3)(cid:20)(cid:19)(cid:19)3(cid:3)
`(cid:3)
`
`Rapid Prototyping of Optical Thin Film Filters
`
`K. Starke, T. GroI3, M. Lappschies, D. Ristau,
`Laser Zentrum Hannover e.V., Hannover, Germany
`
`ABSTRACT
`
`In the course of the rapid development of laser technology and modem optics, an ever increasing demand for optical coatings
`with extraordinary specifications can be observed. In practice, the production of such high quality optics with special
`requirements in respect to bandwidth, edge steepness or wavelength accuracy regularly requires an extended optimization of
`the coating process.
`
`In many cases, the resulting high production cost delays the development of new promising concepts in laser and optics
`technology. For the realization of new optical designs, generally two difficulties occur: At first, the physical properties of the
`coating materials change after completion of the coating process due to environmental influences. Furthermore, the accuracy
`of the commonly utilized methods for thin film thickness monitoring is not sufficient for a reliable thickness control.
`
`In this paper, an ion beam sputtering (IBS) coating process is described for the completely automated fabrication of optical
`coatings with extremely stable characteristics. In contrast to conventional arrangements with witness glasses, the presented
`thickness monitoring during the coating process can be directly performed for the optics. The precise transmittance
`measurement over a bandwidth of one octave is achieved by a fiber-coupled multi-channel spectrophotometer. With this
`arrangement also very small layer thickness errors are detected and may be compensated by optimizing the subsequent layers
`in the stack in order to meet the specifications. The combination of the innovative lBS-process with the broad-band
`spectrophotometric thickness monitoring is the key for new laser applications, e. g. low loss edge filters for high power diode
`laser wavelength multiplexing or phase-optimized mirrors for ultrashort pulse laser systems.
`
`Keywords: broad-band optical monitoring, ion beam sputtering, edge filters, chirped mirrors
`
`1. INTRODUCTION
`
`A steadily growing amount of new laser applications demands for optical coatings with extraordinary specifications.
`Especially extreme spectral requirements concerning broad-band reflection, steep edges or wavelength accuracy complicate
`the manufacture and oblige to optimize the coating process. In many cases, the resulting cost expenditure inhibits the
`development in optics technology.
`
`Utilizing conventional coating procedures, two basic difficulties concerning the manufacture of new optical designs can be
`observed: the alteration of optical properties after coating completion and the deficiency in optical thickness control during
`the coating process.
`
`Since the conventional e-beam PVD-coating process operates under high vacuum and high temperature, optical properties
`basically changes if coatings are objected in normal environmental conditions. Water from air moisture can penetrate the
`column-like morphology of the layer stack and cause alterations in the refractive indices of the coating materials. If the optic
`with the coating is warmed up, e.g. due to laser operation, a fraction of water is removed and the spectral characteristics is
`shifted. This behavior can be diminished utilizing advanced ion beam coating processes due to compaction within the layers.
`High performance laser optics produced by the ion beam sputtering (IBS) process are characterized by a high spectral
`stability, low A+S-losses and high laser damage thresholds'. During the last decade, the lBS process was steadily cultivated
`and optimized at the Laser Zentrum Hannover3. Therefore, the lBS process is preferred for the automated manufacture of
`optical coatings with stable characteristics described in this paper.
`
`For layer thickness control, several procedures and devices are commonly used in coating industry: A very simple method is
`to deposit each material for a specific duration corresponding to the desired layer thickness. This procedure is only applicable
`for exactly known coating conditions, e.g. in the large-scale series production after an exhausting optimization cycle. In
`small-scale series production, oscillator crystal and single-wavelength optical monitoring devices are commercially available.
`
`Optical and Infrared Thin Films, Michael L. Fulton, Editor,
`Proceedings of SPIE Vol. 4094(2000) 2000 SPIE . 0277-786X/OO/$1 5.00
`
`83
`
`0001
`
`

`
`For these methods, a maximum accuracy of about 1 -2 of the layer thickness can he noted. In many cases, this uncertainty is
`unacceptable. e.g. for extreme steep edge filters or chirped mirrors. In order to achieve an enhanced accuracy. broad—hand
`optical illonitonng was introducer!4 and steadily optimized by several iesearchers4' obtaining a layer thickness determination
`accuracy down to kw tenth percents'
`Therefore, at the Laser Zentrum Hannover. this broad—hand optical monitorine
`technique was elaborated within the framework of a German Ministry of Education and Research I BMBFresearch project
`concerning the manufacture of optics for high power diode laser waveIength—muftiplexin. The re1ativel low deposition rate
`of the lBS process implies that the development of this monitoring device was. alread\ at an early stage. orientated at the
`latter automation of the whole coating process. I.e. the monitoring device should he able to recogniie the completion of each
`I aver, control the plant to change the coating material and resume coating the subsequent lave r. In the present 1)llr. the basic
`design and performance of the broad—hand optical monitoring device will he described. In order to demonstrate the
`performance. selected coating results will be presented.
`
`The prospects of success for realizing an optical coating depends on the design as well as on the monitor perforiiiaiice. As
`reported in several papers'''. an error compensation effect can he observed by utilizing optical monitoring. especially hw
`quuter—\a\'e optical thickness (QWOT)—designs. For anti—reflective coatings or other non—QW( )T designs. even minor
`thickness errors may destroy the desired performance specifications. In order to estimate the prospects of success for realizing
`difficult optical designs. a coating simulator was developed at the Laser Zentrum Hannover. This siniulator computes
`transmittance curves due to characteristic parameters of the utilized spectrophotometer. which are transferred to the optical
`monitor for simulating increasing layer thicknesses. The results of anti—reflection design simulations describe the influence of
`refractive index variations on the maintenance of spectral specifications.
`
`2. EXPERIMENTAL
`
`2.1. Monitor Hardware
`
`Larrii:i
`
`Filt
`L.iif. Sriirre
`
`1=1:11 ic al Fib p
`
`''L -L et ekt cr
`
`Va': uurn 1'liairjbi
`
`IAcru:i chroniatcir
`Figure 1 Set-up ott/ic broad-hand optical monitor hard'ii'are
`
`84
`
`Proc. SPIE Vol. 4094
`
`0002
`
`

`
`The broad—hand optical monitor is impleniented in a commercial coating plant (Varian Model 3 125) supplied with an ion
`beam source (RIM ID. Pliifter Vakuumtechnik). The basic structure is preseiited in figure I. It consists of a light source.
`optical fibers and a spectrophotometer. As light source, a tungsten lamp 'ith silver coated reflector is utiliied. Before
`coupling the light into the fiber, the spectral range can be limited with long pass filters inhibiting the appearance of the
`second order spectrum within the measured spectrum. Guided through the fiber into the vacuum chamber, the light is out—
`coupled. collimated and propagating to the test substrate position. After transmitting the substrate, the light is coupled into the
`out—fiber transfrring it to the imaging spectrophotometer with a I 024x256 pixels UV—graded silicon CCD—uiay detector
`(Triax 320. ISA Johin Yvon . The 1024 pixels direction is usually illuminated with a spectrum from approx. 500 to 1000 nm
`by utilizing a ISO lines/mm holographic grating. Therefore. nearly one optical octave can he acquired without moving the
`grating. In the other CCD—arrav direction, the spectra were averaged for reducing the quantum noise of the signal curve. The
`approximately 60 ills exposure time of the CCD—chip is controlled by a mechanical shutter.
`
`Applying a suitable trigger circuit, the optical monitor allows to measure the transmittance of the optic—of—interest or an other
`rotating test substrate on the same calotte radius. Hence, usual calibration problems of witness glasses or oscillatorcrystals do
`not occur. The measurement of one transmittance curve consists of three single measurements perfoiiiied during each rotation
`cycle: a dark—current spectrum without illumination, a refirence spectruni through a hole in the calotte and a sample
`spectrum. The transmittance curve is the quotient curve of signal to reh.rence curve, each reducedh the dark-current curve
`first.
`
`2.2.
`
`Data Processing
`
`cjinitiation
`
`load design, actual layer n=1
`
`-
`
`wait for substrate
`
`4.
`transmission measurement
`(dark-current, reference and sample spectrum)
`
`'I,
`determine thickness of the actual layer
`(least square fit + layer matrices)
`4.
`calculate actual coating rate
`(weighted linear fit of thicknesses)
`
`wait
`
`1
`
`4.
`change material
`or end
`
`Figure 2 F/oar diagram of data processinc.
`The data processing part of the optical monitor is obtained utilizing the data acquisition and processing environnienl
`LahVIEW (Version 5. I. National Instruments). Au important advantage of LahVIEW is to provide so-called "virtual
`instruments". These program modules enables the developer to structure and easily change separate program parts. For the
`
`Proc. SPIE Vol. 4094
`
`85
`
`0003
`
`

`
`described application, the program is designed to use the same core code for different spectrophotometers and for the coating
`simulator, only by changing the data acquisition virtual instruments.
`
`The basic design of the data acquisition is displayed in figure 2. The program starts with an initiation of the measurement
`devices and the loading of the optical design. Usually, the monitor controls the coating plant to start with layer number one.
`During deposition, the program triggers the spectrophotometer to perform transmittance measurements as described above.
`After receiving the transmittance curve, the program determines the thickness of the actual layer by least-square-fining of a
`calculated transmittance curve to the measured one. The theoretical transmittance curve, according to the given design
`utilizing the well-known matrices formalism'4, is produced for the former layers plus the optimized thickness of the actual
`layer. In order to compute as much layer stack matrices as possible, one for each wavelength position, these calculations were
`performed in C++ code included as a "Code Interface Node (CIN)" in the LabVIEW program. With this CIN and a common
`personal computer. the thickness optimization for even 900 wavelength positions and 60 layers usually lasts not longer than 1
`second.
`
`After plotting the actual thickness against the measurement time, the actual coating rate is computed for the estimation of the
`remaining coating duration. If this remaining time is shorter than 15 seconds, the program waits until it terminates the actual
`layer and changes the coating material if needed. Otherwise, the cycle restarts with the transmittance measurement.
`
`2.3.
`Coating Simulator
`For verifying the feasibility of a given optical design, a simulator was developed in order to model the coating process with
`respect to the optical monitor measurement uncertainties. Except the transmittance curve acquisition, this program applies the
`same LabVJEW routines as the real monitor. The simulator has to be supplied with monitor parameters: number and locations
`of wavelength positions, signal noise amplitude, curve offset due to CCD-shutter imperfections and spectral resolution.
`Assuming a constant coating rate, the simulator calculates a transmittance curve corresponding to the increasing thickness of
`each layer. On this ideal transmittance curve, several error contributions are superposed: The signal noise is modeled by
`white noise with an absolute peak-to-peak amplitude of 0.5%. Since the mechanical shutter has a limited reproducibility in
`the subsequent measurement of reference and sample spectrum, an normally distributed offset of
`is added to the
`transmittance curve. The influence of the spectral resolution of the imaging spectrophotometer is modeled by applying the
`floating average over 3 CCD-pixels corresponding to a resolution of about 1 .6 nm.
`
`The realization of a difficult optical design depends also crucially on the knowledge of coating material optical constants.
`Since a long-term variation of refractive indices is possible, e.g. due to target degradation, the influence of insufficiently
`known Sellmeier-Coefficients has also to be proved with this simulator: The production of the simulated transmittance curve
`is performed with the "real" optical constants and the determination of the actual thickness with assumed values. Therefore,
`the design sensitivity against optical constants variation can be estimated.
`
`After the termination of each layer, the simulator registers the real layer thickness and compares it to the desired value.
`Additionally, the resulting spectrum of the simulator run reveals, if specifications could have been maintained.
`
`2.4.
`Design Re-optimization
`In the lBS coating practice, it could be observed that the refractive indices especially of Ti02 depends highly on the coating
`operation parameters. On one hand, coating parameters with an influence on the coating material, as partial pressure of
`utilized process gases, ion current and acceleration voltage, has to be kept as constant as possible during the whole coating
`run. On the other hand, non-operation periods, e.g. due to servicing may be followed by degradation effects of the target
`surfaces also influencing the refractive indices.
`
`Additionally, the termination of each layer is connected with an uncertainty, which may be accumulated during the
`subsequent layers. This accumulation mostly appears in an increasing deviation of the curves corresponding to theoretical and
`measured transmittance. According to an increasing deviation, the termination accuracy of following layers may become
`worse.
`
`For accomplishing even very sensitive designs, the broad-band optical monitor supplies an interface to our SPEKTRUM thin
`film design software and allows for re-optimizing during the coating run. After the determination of each layer, the monitor
`
`86
`
`Proc. SPIE Vol. 4094
`
`0004
`
`

`
`stores the resulting transmittance curve acquired from the monitor. If a re-optimization is necessary, the existing curves can
`be transferred to SPEKTRUM in order to fit thicknesses and refractive indices of the former layers to the measured curves.
`Based on this "real" design, the subsequent layers may be re-optimized towards the desired specifications and the altered
`design delivered to the monitor.
`
`3. RESULTS
`
`Influence of Refractive Index Variations on Different AR-Designs
`3.1.
`In the following section, the investigations concerning the influence of refractive index variations on the performance of the
`optical monitor are presented. As refractive index variations, certain values were selected, which were observed for different
`operation conditions due to coating parameter variations and maintenance periods as mentioned in the introduction.
`
`As an example, these simulations were calculated for a double layer and a broad-band anti-reflection coating, respectively. At
`first, the simulator performed a coating run supposing the exact knowledge of optical constants, i.e. the simulated
`transmittance curve was computed with the same refractive indices as utilized for the layer thickness determination. Since
`refractive indices cannot be exactly known in principle, a relatively small variation of 0. 1 % is added to the refractive index
`of TiO and 5i02, respectively, in a further simulation run. Additionally, several simulator runs were performed with a higher
`variation of the Ti02 refractive index corresponding to its sensible behavior.
`
`1.000
`
`0.999
`
`ci)0 0.998
`
`0.997
`
`0.996
`
`C I
`
`0.995
`750
`
`760
`
`780
`770
`wavelength [nm]
`
`790
`
`800
`
`Figure 3 Simulator results for double layer AR (on afused silica substrate)
`
`Proc. SPIE Vol 4094
`
`87
`
`0005
`
`

`
`In figure 3, the results of the simulator runs concerning the double layer anti-reflection coating are presented. For the case of
`no variations and for variations up to 0.8 %, no major difference of all simulated spectra can be observed, i.e. the
`specifications of minimum reflectance at 777 nm were maintained. Even for a very unusual alteration of 2.9 %, only a little
`wavelength shift can be noticed for this given design.
`
`a)
`C-)
`
`E0
`
`750
`
`800
`
`850
`900
`wavelength [nm]
`
`950
`
`1000
`
`Figure 4 Simulator results for broad-band AR 777-9 70 inn (8 layers on afused silica substrate)
`
`The results for the same variations in the case of the broad-band anti-reflection coating are plotted in figure 4. For these
`variations a totally different behavior can be observed: The simulator run with exact optical constant knowledge shows that
`the specifications of 0. 1 % reflectance over the spectral range of 777 to 970 nm could be maintained except a little exceeding
`area. This exceeding reveals that the monitor device has a limited layer termination accuracy inducing crucial errors to this
`sensible design. For a little variation of 0. 1 % in refractive indices, also an acceptable spectrum mainly within the
`specifications can be achieved. With increasing variations of the Ti02 refractive index, an enlarging spectral region departs
`from the specifications. The 2.9 % spectrum is not displayed, because the corresponding transmittance is located below
`99.5 %.
`
`The results of these two simple designs show, how distinctly the prospect of success of an automated coating run is
`dependent on the design. Especially, the appearance of rather thin layers below 0. 1 QWOT may complicate the realization of
`broad-band anti-reflection coatings. Furthermore, the knowledge of the exact optical constants is crucially necessary for a
`reliable automatic coating procedure. Only minor variations in these constants will be tolerated if complicated designs are
`desired.
`
`88
`
`Proc. SPIE Vol. 4094
`
`0006
`
`

`
`3.2.
`Performed Coatings
`In the following section, selected results of performed coating runs will be presented. All described coatings were
`manufactured using Ti02 as high and Si02 as low refractive index material on fused-silica substrates. The transmittance
`measurements were performed with a commercial spectrophotometer (Lambda 19, Perkin-Elmer).
`
`As first example, a short pass edge filter with a band pass area from 665 to 790 nm under an AOl of 0° is presented.During
`the coating procedure, layer thicknesses and refractive indices were regularly re-optimized for the band pass area as described
`in section 2.4. As can be regarded in figure 5, the band pass transmittance area is extremely broad and homogeneous. The
`design transmittance is limited by the reflection of the substrates back surface. Neglecting the back surface, thetransmittance
`varies in the range of 99.7 and 100 % in the band pass region. Comparing the design and the measured transmittance in the
`magnified graph, a maximum deviation of about 'A % can be observed. For maintaining these extreme specifications,it was
`advantageous that the band pass area was entirely located within the detectable spectral range.
`
`1,0
`
`0,8
`
`0,6
`
`0,4
`
`0,2
`
`a)0
`
`E0=
`
`0,0
`600
`
`700
`
`900
`800
`wavelength [nm]
`
`1000
`
`1100
`
`Figure 5 Edge filter HT 665-790nm, HR >850nm
`As a second example, an interference filter with a two-cavity design is presented. The transmittance peak was inserted in
`order to broaden the reflectance area up to a spectral range of 800 to 1100 nm reflecting several high power diode laser
`wavelengths. This coating run was performed automatically, i.e. no optimization of existing and subsequent layers were
`performed.
`
`Proc. SPIE Vol. 4094
`
`89
`
`0007
`
`

`
`In figure 6, the transmittance of the corresponding ideal QWOT-layer stack and the transmittance of the realized coating are
`displayed. Regarding the coarse spectrum from 400 to 2000 nm, an excellent agreement of both curves can be observed.
`Hence, the refractive indices utilized in the design were sufficiently known to obtain the reflectance area and the position of
`the interference peak, both mainly in the monitoring spectral range. Additionally, the plot shows that the supposed dispersion
`is also suitable for controlling spectral characteristics outside the monitor region.
`
`1,0
`
`0,8
`
`0,6
`
`0,4
`
`0,2
`
`ci)0
`
`E(
`
`I)
`
`0,0
`400
`
`600
`
`800
`
`1000 1200 1400 1600 1800 2000
`wavelength [nm]
`
`Figure 6 Inteiferencefilter (2-cavity design), spectral overview
`In figure 7, a detailed view on the transmittance peak is given. As an indicator for the layer thickness accuracy, the design
`transmittance is supplied with error bars corresponding to the design sensitivity against thickness errors. For this, the
`SPEKTRUM-design software is provided with a procedure inducing statistically distributed thickness errors up to a certain
`value to each layer in the stack. For a tolerance estimation, an amount of 50 calculation cycles of the transmittance with
`disturbed layer thicknesses are performed. The error bars represent the maximum deviation values from the ideal curve
`during all 50 cycles.
`
`For the tolerance test of the given interference filter, a maximum thickness error of 0.5 % was adjusted. The plot in figure 7
`reveals that this uncertainty, which is rather good for conventional monitoring devices mentioned in the introduction, would
`sensibly influence the shape and amplitude of the transmittance peak. Since the agreement of the design and the measured
`transmittance is far better than the tolerance estimation, a distinctly lower thickness error can be assumed.
`
`90
`
`Proc. SPIE Vol. 4094
`
`0008
`
`

`
`01.
`
`8'0
`
`CD
`C)
`
`9'0
`
`VVB
`4.,'(
`
`CI)
`
`.-'
`
`0'0
`006
`
`01.6
`
`06
`0C6
`q6ueeAM
`
`[wu]
`
`0176
`
`096
`
`ainXzq
`
`L
`
`vad
`aauvmuisuwj
`'pVjdp
`
`0
`%ç
`SJOLId SS?U)/37t/l
`JOJS'd3UWd/OJ
`
`
`
`SNOIS1T1TMOD
`uozoi
`qi
`sq
`sup qi
`uj
`-sal
`qiq
`'idid
`putq-pioiq
`ioiuom
`ioj
`jo
`uooq
`poiiuisuomop
`uio
`wosaid
`oon'uuojiod
`ID!1d0
`qi
`suiio
`jrnioppt
`ocuiuausuu
`jo
`qiM
`sum
`siuuoinsiow
`'ioiuow
`isoi
`uisuodsp
`pondo
`-jo-Dudo
`r
`musqns
`jso
`ouj
`uuiioi
`uinp
`q2 spoop
`
`JJ
`Jo
`om
`iopuom
`poauojiod
`1SOJOU
`&npuodsauoD
`uo!11UiluJo2
`uo1Tsodop
`j1
`uioz
`'Jojoio
`jo
`put
`oq
`UOD
`iuid
`soiiuoo
`ut qi
`oqi
`iuowow
`OU1j3
`XIP0!UW01fl
`01
`po11!_uo1n1
`so
`put
`paiojr
`uinpijnuru
`ssooid
`oA1DoJp
`u!1ioD
`iOJ
`so!jddns
`onpoooid
`SflOLItA
`pDudo
`U!A1S
`10S131
`S&ffflO3
`suouDijddi
`io
`i
`qi
`piiuoni
`uiiiuso
`uioc,
`Jo
`sizdsoid
`JO
`oqi
`'un.i
`oloidwoD
`SSOODflS
`popiosoid
`jo uonjjui
`
`sioiowtnd
`SUO!11JtA
`Sf11
`pi.m
`'iO11flWTS
`J1owo1oiTdoJ1Dods
`UOSJOdSIp
`i!do
`OA1
`01
`UMOU)
`S1U1SUOD
`U!AO!qM
`•sU!sop
`Oq
`1OJ
`xoIdwoD
`ixis
`qii
`uwo
`
`jo sinsoj
`1
`oq
`suni
`putq-pioiq
`2tq
`pouuojd
`pudo
`poqinsop
`iouuow
`S1M
`poiuosoid
`OJOM
`UMO1jS
`inoqi
`pu2qo
`pno
`uoq
`!qiIoJ
`iqiuo
`szis.ioiwmqc
`puods
`oAiq
`UOA
`°LL
`pouDJdtuoo
`oinpooid uomziumdo
`
`xpui
`jo
`pui
`kiqssod
`uunp
`ioido
`uiuiuuoop
`uisop
`Ai1D1uJoJ
`SUOTLmA
`sosloJDu!
`uoiiisothp
`utzauT1do-oJ
`UOAO
`'qi
`ziuarixo
`wiiods
`UI
`2uTUp1ull1u
`suoutDUicods
`
`qi
`ionuis
`sondmoc,
`Moq
`'poww!1s0
`1ciioixo
`
`iouuow
`UD
`Oq
`
`Od
`
`2IdS
`IOA
`
`t6Ot'
`
`0009
`
`

`
`ACKNOWLEDGEMENTS
`
`This work would not have been performed without the grateful financial support within the framework of the BMBF-project
`"Wellenlängenmultiplex-Verfahren für die Direktanwendung von Hochleistungs-Laserdioden" under contract no. 13N7293.
`Furthermore, the assistance of the VDI-TZ -Physikalische Technologien- concerning the administrational management of the
`project is very much appreciated.
`
`REFERENCES
`
`1. D.T. Wei: "Ion beam interference coating for ultralow optical loss", Appi. Opt. Vol.28, No. 14, pp.2813-2816,1989
`2. P. Amaya and C. Masser: "Ultra-low-loss Optics", Lasers and Optr. June 1992, pp.7-8, 1992
`
`3. R. Henking, D. Ristau, H. Welling: "Optical characterization and damage thresholds of low loss mirrors", Laser Induced
`Damage in Optical Materials: 1994, SPIE 2428, 281-292 (1994)
`
`4. B. Vidal, A. Fornier, and E. Pelletier: "Optical monitoring of nonquarterwave multilayer filters", Appl. Opt. Vol. 17 (7),
`pp. 1038 (1978)
`5. Fred J. Van Milligen, Bertrand G. Bovard, Michael R. Jacobson, James Mueller, Ross H. Potoff, Richard L. Shoemaker,
`and H. Angus Macleod: "Development of an Automated Scanning Monochromator for Monitoring Thin Films", Appi.
`Opt. Vol. 24 (12), pp. 1799 (1985).
`6. Powell, J.C.M Zwinkels, and A.R. Robertson: "Development of optical monitor for control of thin-film deposition",
`Appl. Opt. Vol. 25 pp. 3645 (1986)
`7. Gabriele Emiliani, Angela M. Piegari, and Enrico Masetti: "Fast Scan Spectrometer for Monitoring of Thin Film Optical
`Properties", Proc. SPIE Vol. 1012 pp. 35 (3, 1988).
`
`8. Li Li, Yi-Hsun Yen: "Wideband Monitoring and Measuring System for Optical Coatings", Appi. Opt. Vol. 28 pp. 2889
`(7, 1989).
`
`9. Xue-Qun Hu, Yu-Ming Chen, and Jin-Fa Tang: "Apparatus for Wideband Monitoring of Optical Coatings and its Uses",
`Appl. Opt. Vol. 28 pp. 2886 (7, 1989).
`10. Brian T. Sullivan, J.A. Dobrowolski: "Deposition Error Compensation for Optical Coatings. I. Theoretical Description.",
`Appl. Optics vol. 31 pp. 3821 (1992).
`11. Brian T. Sullivan, J.A. Dobrowolski: "Deposition Error Compensation for Optical Multilayer Coatings. II. Experimental
`Results-Sputtering System", Appi. Opt. vol. 32 pp. 2351(1993).
`12. Harry H. Bauer, Erwin Nuessler: "In-situ optical multichannel spectrometer system", Proc. SPIE Vol. 2253 pp. 423-43 1
`(11, 1994).
`13. Harry H. Bauer, Erwin Nuessler: "Computer-Integrated Manufacturing of Advanced Thin Film Coatings", Proc. SPIE
`Vol. 2253 pp. 432-436 (1994).
`
`14. H. A. Macleod, "Thin-Film Optical Filters", Adam Hilger, London, 1969
`
`92
`
`Proc. SPIE Vol. 4094
`
`0010