Surface & Coatings Technology 347 (2018) 252–256
Contents lists available at ScienceDirect
Surface & Coatings Technology
journal homepage: www.elsevier.com/locate/surfcoat
Control and broadband monitoring of transparent multilayer thin films T
deposited by magnetron sputtering
A. Voronova, S.A. Atarahb,⁎
a Thin Film Centre, University of the West of Scotland, High Street, Paisley PA1 2BE, UK
bDepartment of Physics, University of Ghana, Legon, Ghana
A R T I C L E I N F O A B S T R A C T
Keywords: A multilayer thin film deposition system has been developed that implements broadband process control and
Magnetron sputtering monitoring. Running on a simple algorithm, the system stringently controls the thickness of each of the mul-
Broadband monitoring tilayers thereby avoiding thickness error build up throughout the deposition process. The system which also has
Thin transparent fil process monitoring capability can be implemented on any platform. Antireflection coatings deposited by the new
Film deposition system yielded spectral characteristics of commercial standards.
1. Introduction measuring the transmission spectrum of deposited thin films in situ. The
transmission spectra of the evolving layers measured during deposition
Thin films have many and important applications. Their applica- matched very well with the (calculated) theoretically expected char-
tions span across industry, health and research sectors. Materials coated acteristics at various stages of deposition. The system was used for
with thin films or a cascade of thin films produce unique surface and depositing multilayer filters and anti-reflection coatings using Nb2O5
optical properties not observable in conventional materials. Techniques and SiO2 and initial results have shown good matches with the targeted
for thin film coating include electron-beam evaporation with ion-beam specifications. The spectral characteristics of filters deposited with same
assistance [1–3], ion beam sputtering [4,5] and reactive magnetron optical design specifications were reproducible and also showed no
sputtering [6,7]. Magnetron sputtering is said to provide dense stable ageing effects. Simple modifications allowed the new monitor to be
films and is the commonly preferred deposition method for the most remotely controllable over a LAN network.
demanding applications. It has been noted that without effective pro-
cess control, magnetron sputtering processes can lead to poor run-to- 2. Experiment
run repeatability, poisoned target state [8], process drift and arcing
[8–11]. The advent of nanoscience requiring applications in the nano- 2.1. Material and deposition set up
regime also places low tolerance limits on the precision for thin film
coating. The above underpin the need for robust control and monitoring Anti-reflection (AR) coatings were deposited using Nb2O5 and SiO2.
systems during magnetron sputtering. Four layers were deposited (in alternation) on each side of BK7 glass
Various systems exist for monitoring the thicknesses of thin films substrate by microwave plasma-assisted pulsed-DC sputtering.
which include quartz crystal monitoring (quartz tooling), multi-wave Microwave power of 3 kW was applied and the frequency was
ellipsometry and Real-time Spectroscopic Ellipsometry [12]. An ex- 2.54 GHz. Fig. 1. is a schematic of the multilayer AR coating. The
tensively review by Kildemo et al. shows that besides their sensitivity to TFCalc software [14] was used to design the layering.
noise, most of these systems are unsuitable for multilayer films due to Details of the layers are given in Table 1.
cumulative thickness errors [13]. The pulse D-C power and duty cycle of the system were preset at
A magnetron sputtering system with real time thickness control 5 kW and 70% respectively by the power supply unit. The deposition
capability would be required in order to meet target optical char- was done using a MicroDyn 40000 series magnetron sputtering system
acteristics of thin film deposited components. (DSI Inc.). Si and Niobium metal (Metal, Grey 99.98 wt% “SuperVac®”
In this paper, we demonstrate a broadband monitoring technique Evaporation grade as supplied by Testbourne LTD) were used as the
that has been integrated into a magnetron sputtering deposition system. target material whilst Ar and O2 were used as sputtering and reactive
The technique enables monitoring of layer thickness evolution by gases respectively. Thus SiO2 was the low (L) whilst Nb2O5 was the high
⁎ Corresponding author.
E-mail addresses: a.voronov@uws.ac.uk (A. Voronov), saatarah@ug.edu.gh (S.A. Atarah).
https://doi.org/10.1016/j.surfcoat.2018.05.004
Received 25 October 2017; Received in revised form 4 April 2018; Accepted 3 May 2018
Available online 04 May 2018
0257-8972/ © 2018 Elsevier B.V. All rights reserved.
A. Voronov, S.A. Atarah Surface & Coatings Technology 347 (2018) 252–256
typically maintained at 25 sccm: 14 sccm. This ratio is close to those
recommended in a recent investigations on the effect of Ar/O2 gas ratio
on oxide multilayer film properties [15]. The total pressure in the
chamber was about 5×10−3 Torr during deposition but deposition
was not commenced till the chamber was pumped down to
5×10−6 Torr. In order to avoid out-gassing or spitting, a 10–15min
interval was allowed for pre-heating to slowly bring the target material
up to deposition temperature. Twelve samples were deposited in one
run and five runs were made from the same design with data collected
on one representative sample during each run. For each process, the
samples were arranged on a revolving drum in a chamber. Fig. 2 is a
schematic of the system that was used. The drum was set to rotate at
Fig. 1. Schematic of the anti-reflection coating. The schematic also suggests the 60 rpm because at this speed deposited films showed more repeatable
transmittance (T) and suppressed reflectance (R) of a typical ray of light character over several runs compared to those deposited at the drum
through the AR coating.
speed of 30 rpm. The layer deposition rate under these conditions was
2 Å/s.
Table 1 Optical transmission spectra of deposited films were measured in
Target details showing the thickness for each layer as obtained by the design the range 385–1100 nm using two systems: Peckin Elmer Lamda 19 and
software (TFCalc®). Scalar Technologies spectrometers. The latter spectrometer (Scalar) was
Layer Dielectric Thickness (nm) Layer Dielectric Thickness (nm) used for the real time measurements whilst the former (Lamda 19) was
used after deposition for the purpose of comparison. Both systems
Front side corroborated each other in the spectra obtained. The refractive index
1 H 110.83 5 H 87.63
2 L 119.37 6 L 127.57 and absorption coefficients were obtained from the spectral data using
3 H 99.36 7 H 71.19 the Cauchy method [16] which was integrated into the monitoring
4 L 105.17 8 L 143.30 software. In order to verify repeatability of spectral characteristics of
Back side deposited films, the spectral data was periodically taken of random
1 H 100.20 5 H 79.14 samples from different runs for comparison. Similar test measurements
2 L 115.44 6 L 145.01 were also made after several weeks and the spectra compared with
3 H 92.71 7 H 183.51 those of freshly deposited samples to check for any ageing effects.
4 L 103.85 8 L 78.60
Key H Nb2O5 L SiO2
2.2. Process monitoring and control
Glass substrates (of type BK7), on which films were deposited, were
placed on holes on the rotating drum in the deposition chamber. A fibre
optic cable, carrying white light, was introduced into the deposition
chamber such that it run into the inner part of the drum directing the
light to its outer part through the holes (on the drum). See Fig. 2. A
geometrical optical system configured in the chamber just outside the
drum collected the light from the fibre optic cable and relayed it to a
diode array spectrometer (Scalar Technologies Ltd).
The geometrical optical arrangement was constructed in such a rigid
way that any losses due to misalignment which have been reported [17]
were minimized. By the arrangement, all samples (on the drum) cut the
optical path normally in each revolution of the drum enabling the
transmittance of the thin film samples to be measured during deposi-
tion. Both the spectrometer and the deposition system were connected
to a PC. An in-house built software enabled concurrent process control
and in situ measurement of film spectral data. During deposition the
transmission spectrum of the sample and of the background reference
were captured in each revolution of the drum. The background re-
ference was essentially constant and so no stray reflections could affect
the measured transmission spectrum. With systems that use rotation on
same plane, a witness sample is needed.
The in-house developed software was the hub of the automation of
the deposition system. In addition to implementing tight thickness re-
quirements by the algorithm, the software also enabled remote control
and monitoring over a LAN/Intranet connection. Details of the inter-
relations between the program modules and systems have been ex-
plained earlier [18].
Fig. 2. Experimental setup for real time broadband monitoring and control of
the thicknesses of samples on the revolving drum. Adapted with permission
from earlier work [21]. 3. Theory
3.1. Thickness monitoring
(H) refractive index material. Both gases were fed in through a PC
regulated mass flow controller by which the gas concentration was The point of focus of effort in this work was development of a
maintained. During deposition Ar: O2 gas concentration ratio was system capable of repeatedly depositing stable layers of thin films with
253
A. Voronov, S.A. Atarah Surface & Coatings Technology 347 (2018) 252–256
minimum deviation from the required spectrum. In order to meet the
0.95
target spectrum, layers must be deposited with minimal error in the (a)
thicknesses specified by the design software. 0.90
For thickness control the algorithm proposed by Wilbrandt et al.
[19] was adopted. The algorithm follows the thickness, dk, of the cur- 0.85
rent layer being deposited with the object being to minimize as much as
0.80
possible the difference, Δ, between the theoretically required thickness
DN and the actual deposited thickness dN for the Nth layer. The layers 0.75
are numbered 1, 2, 3…k− 1, k, k+1, …, N.
0.70
Assume that k− 1 layers were deposited within a window of j
spectral points [λ1, λ2, λ3,… λi,… λj]. Then the transmittance TM (λi) 0.65  Lamda9
recorded at each spectral point λi is  Scalar
0.60
TM = TM (λi ), i = 1, 2, 3, …j (1)
0.55
The measured transmittance was then compared with the theoreti- 300 400 500 600 700 800 900
cally calculated transmittance. Computationally, the transmittance was Wavelngthg (nm) 
obtained by considering the thickness dk of the layer being deposited.
Optical wave propagation in the medium can be described by a transfer 2.6 7.0
matrix [20]. (b)
n - value 6.5
cos(κdk) 1 sin(κdk)
Mk =
⎛ κ ⎞ k - value 6.0
⎜ k k ⎟ 2.5
⎝− κ sin(κd ) cos(κd )⎠ (2)
5.5
where k refers to the current layer and κ is the wave number. When
5.0
transmission across a system of N stack of layers is considered, the 2.4
system transfer matrix, Ms, becomes 4.5
Ms = MN ·MN−1·MN−2⋯M1. (3) 4.0
The electric field transmitted across a layer can be expressed as 2.3 3.5
E (z) = tE eiκ0 Rz. (4) 3.0
where E0is the amplitude of the wave at the boundary (z=0)and the 2.2 2.5
transmission amplitude t is given by (Born & Wolf [20]) 400 600 800
Wavelength (nm)
t = 2iκ eiκL ⎡ M11M22 − M12M21L ⎤⎢ ⎥.⎣−M12 + κLκRM12 + i (κRM11 + κLM22) ⎦ (5) Fig. 3. (a) Transmission spectrum for a single layer of Nb2O5 as measured by
two commercial spectrometers at room temperature.
In expression (5), Mmn refers to the components of the system (b) Refractive index and absorption coefficient variation with wavelength of
transfer matrix whilst κR and κL are respectively the wave numbers in Nb2O5.
the medium to the right and left of the boundary. The transmittance
was then obtained as Tcal=|t|2. thickness, dk. Typically, for each layer, dk was averaged from 15 gen-
The extent to which the measured transmittance agreed with Tcal erated values. The system was then returned from the suspended state
was assessed at the current (kth) layer by use of a merit function, F, of after 30–45 s of delay and the monitoring advanced to the next layer for
the form. deposition, if the current layer was not the last one. As discussed below,
j
F (dk) = ∑ [T dk, λ − T (λ )]2; 0 < dk ≈ Dk for each layer in the design file, the program calculated and displayedi=0 cal ( i) M i (6) the theoretical spectrum as a function of percentage completion for that
Application of the merit function was facilitated by computing T layer. The current total thickness of the layer, where desired, was alsocal,
at the same grid as the T . At each spectral point, the merit function used for calculating the spectrum. The software could be used to si-M
depends only on the thickness of the current layer, dk. Thus the function mulate the evolution of the film spectrum as well as closely follow the
was vital in obtaining optimal thickness of the currently deposited process development during deposition. The final spectrum of the last
layer. The minimum solution, d , of the merit function was used layer in the software repeatedly corresponded to the target spectrum asend
during deposition as the stopping criterion for the current layer: optically designed.
dk kend > D − Δ (7)
4. Results
3.2. Algorithm implementation Fig. 3(a) shows the transmission spectrum of a single layer of Nb2O5
on BK7 glass substrate. Included in the figure for comparison is the
The lab configuration for implementing the system was close to that transmission spectrum for the BK7 substrate only. Within the visible
described earlier [21]. Suppose a certain layer was being deposited. For spectrum it can be seen that the maxima for the transmittance for
any thickness, the transmittance was measured and compared with the Nb2O5 intersect that of the substrate indicating that deposition was
calculated value so as to determine the thickness of the layer, dend, for homogeneous and nearly absorption free [22]. This was a pointer to the
which merit function was acceptable (see Eq. (6)). Once the acceptable high quality of the target used for sputtering and is also an indication of
stopping criterion, dend, was obtained, an interrupt signal was triggered optimal process conditions. Fig. 3(b) shows the dependence of the re-
which suspended the deposition system. At that instant the routine for fractive index, n(λ), and the absorption coefficient, k(λ), on wave-
the merit function was repeated several times to generate many values length, i.e. n− k characteristics. The n− k data was extracted from the
of the current layer thickness which was averaged to be the target transmittance data by integrating the Cauchy method in the broadband
254
n ( )
%T
-4
k ( ) x 10
A. Voronov, S.A. Atarah Surface & Coatings Technology 347 (2018) 252–256
monitoring software. The n− k values, as indicated by arrows in
1.0
Fig. 3(b), were plotted for same wavelength. At 550 nm and 800 nm the
refractive index values of 2.34 and 2.24 respectively match those ob-  Deposited (in situ)0.9
 Required (Designed) 
tained by other workers recently, [22,23].
 Deposited (vented)
It is well know that absorption within or below the order of 10−4 0.8
has a minor effect on transmittance. It can be observed from Fig. 3(b)
0.7
that k-values were indeed of this order throughout the wavelength
window of measurement. One can therefore say that with the process 0.6
conditions fixed, absorption would have an insignificant effect on the
results presented. The spectrum for the anti-reflection filter would thus 0.5
be expected to be determined entirely by the precision with which the
0.4
deposited film thicknesses conformed to the designed and also by the
refractive index. Variation in the refractive index would affect trans- 0.3
mittance and hence the thickness calculation. However, the refractive
index would vary if there are changes in the composition of plasma (i.e. 0.2
the ratio Nb2O5:SiO2) [3] or in the deposition conditions. Under fixed
process conditions therefore deposited components would have negli- 400 500 600 700 800 900 1000 1100 1200 1300
gible errors in layer thicknesses and hence the repeatability of the Wavelength (nm)
spectra. In this work, the deposition pressure was fixed throughout the Fig. 5. The transmission spectrum of a deposited filter as measured in air
deposition process and the gas concentration was maintained by mass (Vented) is compared with the spectral data for the theoretical design target
flow sensors controllers. (Design) and that taken in situ (Vacuum).
Fig. 4 shows the transmission spectrum as captured during the de-
position of the first layer of the anti-reflection coating. Calculated Under-coating and over-coating therefore compensate each other,
spectra are shown for the layer as a function of completion stages at minimizing over all thickness errors. Where over-coating or under-
20% intervals. For comparison with measured data, the spectra were coating occurred, an error table indicating the deviation from the the-
also calculated for two values of layer thicknesses (91.30 nm and oretical target layer thicknesses was kept. Five runs were made for the
118.71 nm) at which measurements were made during deposition. In particular deposition presented and data on thickness error for each
Fig. 4, the measured spectral data are shown as symbols whilst lines layer was recorded. Out of the total of 96 layers in one run, a mean
represent calculated spectral data. It can be seen that there is an ex- value of 0.63 ± 0.21 nm was obtained for Δ. Thus on the average the
cellent agreement between the calculated and measured spectra. It thickness could be controlled to within 1 nm.
suggests that the thicknesses deposited closely matched the values Fig. 5 is the spectrum of the anti-reflection coating taken after the
targeted. Similar spectra were observed for all layers throughout the
final layer for one side was deposited but under process temperature
process. It must be stated, though, that in few instances over coating and pressure. The theoretical target spectrum along with the spectrum
and undercoating were observable but these often compensated each of the same sample measured several weeks later is included in the
other. Infinitesimal deviations in the preceding layer thicknesses add up
figure for comparison. A good agreement with the target spectrum can
to later layer thicknesses. Thickness deviations can arise due to de- be observed. It can be seen that age had shown little effect on the de-
position history as follows. The merit function determines only the posited filter. After backside coating the anti-reflection spectrum com-
thickness of the current layer required to match the theoretical spec- pared favourably with a commercial neutral density filter (part #
trum but the measured current spectrum is for the totality of layers NE201B, Thorlabs, Germany) for the visible regime.
already deposited. Thus, the function may under evaluate the current
layer thickness if, due to positive residual errors, the measured thick-
ness is more than the theoretical value. In such a case, a smaller value 5. Discussion
for the stopping criterion was obtainable resulting in under coating of
the current layer. In a similar manner a layer may be over-coated. An important advancement realised by the new system is an en-
hanced process monitoring, i.e. real time availability of thickness and
spectral information of deposited layers as well as strict thickness
0.95
control leading to achievement of desired optical characteristics. A
Initial spectrum (substrate) common problem in coating with multilayers of thin films is that errors0.90
in each of the layers add up towards the end, thereby making it hard to
achieve the final target thickness and the desired spectral character-
0.85 20 % of layer deposited
istics. The developed system though cannot be claimed to completely
solve this problem but can minimize thickness errors to under 1 nm.
0.80
40 % One more important advantage of the new deposition and mon-
0.75 itoring over existing commercial systems is the response to (layer)
80 %
118.71 nm process termination. Fig. 6 compares the spectra of two filters deposited
60 % 
0.70 from the same optical design but using the newly implemented mon-
104.71 nm
itoring system for one deposition (Fig. 6 (a)) and quartz monitoring for
91.30 nm
0.65 the other (Fig. 6 (b)). For clarity, data is shown from 650 nm for both
cases. It can be seen that over all the spectra are more reproducible
0.60 when monitored by the new system. With the quartz monitoring, the
difficulty in reproducing the spectra especially at the edges is because of
400 500 600 700 800 900 failure by quartz crystal monitor to consistently stop deposition at the
Wavelength (nm) defined layer thickness. Quartz tooling factor can vary [24]. Whilst the
new broadband monitoring system uses the stopping criterion, dend, to
Fig. 4. Transmission spectrum as measured in situ and predicted by the algo- promptly interrupt the deposition process, quartz tooling factor changes
rithm. with time. Other broadband monitoring systems control the layer
255
Transmittance
%T
A. Voronov, S.A. Atarah Surface & Coatings Technology 347 (2018) 252–256
Broadband monitoring monitoring system has been developed. Using a compact algorithm, the
100 transfer matrix method and basic computer connectivity, the system
enables precise spectral measurements and real time determination of
 ScalarRun1
90 the thickness of a growing film. Implementation of the system was
a)  Desgn demonstrated by deposition of several anti-reflection coatings. Data
80  ScalarRun2 collected from initial test runs using the new deposition and monitoring
system shows that, on the average, the deviation from target layer
70
thickness was controlled to 0.63 ± 0.21 nm. A comparison of trans-
60 mission spectra of deposited filters as monitored by quartz and by the
new broadband monitor shows how superior the latter system is in
50 adhering to designed layer thicknesses during deposition. The mon-
itoring was implemented across a LAN introducing flexibility to
40
monitor the deposition system remotely.
30
Acknowledgement
20
The authors would like to acknowledge Dr. S. Song (University of
10
700 800 900 1000 1100 1200 1300 the West of Scotland, UK) who was a research assistant at the Thin Film
Centre at the time of the work for technical assistance.
 (nm)
Quartz monitoring References
[1] C. Xu, et al., Comparison of laser-induced damage in Ta2O5 and Nb2O5 single-layer
films and high reflectors, Chin. Opt. Lett. 9 (2011) 013102.
90 b)  Quartz1 90 [2] G. Abromavicius, et al., The microstructure and LIDT of Nb2O5 and Ta2O5 optical
 Quartz2 coatings, Proc. SPIE 6403 (2007) 640315.
 Desgn [3] J. Sancho-Parramon, V. Janicki, H. Zorc, Compositional dependence of absorption
coefficient and band-gap for Nb2O5–SiO2 mixture thin films, Thin Solid Films 516
(2008) 5478.
[4] C.-C. Lee, J.C. Hsu, D.H. Wong, Low loss niobium oxides film deposited by ion beam
60 60
sputter deposition, Opt. Quant. Electron. 32 (2000) 327.
[5] A. Melninkaitis, et al., Characterization of zirconia and niobia silica mixture coat-
ings produced by ion-beam sputtering, Appl. Opt. 50 (2011) C188.
[6] S. Jakobs, et al., Characterization of metal-oxide thin films deposited by plasma-
assisted reactive magnetron sputtering, Chin. Opt. Lett. 8 (2010) 73.
30 30 [7] F. Elsholz, et al., Roughness evolution in thin-film growth of SiO2 and Nb2O5, J.
Appl. Phys. 98 (2005) 103516.
[8] J. Musil, J. Vlcek, P. Baroch, Chapter 3 — magnetron discharges for thin films
plasma processing A2, in: Y. Pauleau (Ed.), Materials Surface Processing by Directed
Energy Techniques, Elsevier, Oxford, 2006, pp. 67–110.
[9] E. Hollands, D.S. Campbell, The mechanism of reactive sputtering, J. Mater. Sci. 3
0 0 (1968) 544.
700 800 900 1000 1100 1200 1300 [10] S. Venkataraj, O. Kappertz, M.W.R. Jayavel, Growth and characterization of zir-
 (nm) conium oxynitride films prepared by reactive direct current magnetron sputtering,
J. Appl. Phys. 92 (2002) 2461.
Fig. 6. The transmission spectrum of a gain flattening filter as deposited using [11] M. Audronis, V. Bellido-Gonzalez, B. Daniel, Control of reactive high power impulse
magnetron sputtering processes, Surf. Coat. Technol. 204 (2010) 2159.
(a) the new broadband monitor and (b) by quartz monitoring. [12] S. Marsillac, S.A. Little, R.W. Collins, A broadband analysis of the optical properties
of silver nanoparticle films by in situ real time spectroscopic ellipsometry, Thin
Solid Films 519 (2011) 2936–2940.
thickness by estimating the time required to deposit a certain layer from [13] M. Kildemo, B. Drivillon, Real-time control of the deposition of optical coatings by
the deposition rate, the refractive index and the layer thickness [25]. In multiwavelength ellipsometry, Surf. Coat. Technol. 100-101 (1998) 480–485.
such systems the stopping criterion (time to deposit) depends on three [14] M. Tilsch, et al., Design and Demonstration of a Thin-Film Based Gain EqualizationFilter for C-Band, EDFAs NFOEC, 1999.
factors each of which can cause an error in the thickness of the de- [15] X.F. Luo, Z.E. Ning, J.H. Wang, J.J. Yang, Y.J. Feng, J.L. Liao, Y.Y. Yang, K.M. Feng,
posited layer. The deposition rate for our system at the set gas con- N. Liu, M. Gong, Effects of Ar/O2 ratio on preparation and properties of multilayer
Cr2O3/α-Al2O3 tritium permeation barrier, Surf. Coat. Technol. 339 (2018) 132.centration was typical of those reported recently [22]. The system was [16] R.S. Adve, T.K. Sarkar, Generation of accurate braodband information from nar-
tested for room temperature processing. The results are promising be- rowband data using the Cauchy method, Microwave Opt. Tecnhnol. Lett. 6 (10)
cause low temperature depositions are known to be good for processing (1993) 569–573.[17] P.A.v. Nijnatten, Optical monitoring tools and strategies for controlling coating
of low scattering loss optical coatings. deposition in large area continuous coating processes, Thin Solid Films 502 (2006)
The broadband monitoring system has subsequently been applied to 147–152.
[18] A. Voronov, F. Placido, I. Bain, In Situ Broadband Optical Monitoring and
deposit output gain flattening filters, high reflection and antireflection Characterization of Thin Films in 51st Annual Technical Conference Proceedings of
components with repeatable characteristics. the Society of Vacuum Coaters, (2008) (Chicago IL, USA).
The system in its current state is however limited to depositing only [19] S. Wilbrandt, N. Kaiser, O. Stenzel, In-situ broadband monitoring of heterogeneous
optical coatings, Thin Solid Films 502 (2006) 153–157.
transparent thin films on transparent substrate as it relies on light [20] M. Born, E. Wolf, Principles of Optics: Electromagnetic Theory of Propagation,
transmitted through the layers for transmittance and thickness calcu- Interference and Diffraction of Light, Pergamon Press, Oxford, (1964).
lation. However, with a little modi cation, re ected light from samples [21] S. Atarah, et al., Integrated method for control and broadband monitoring offi fl multilayer thin films, Optical Interf. Coating, Optical Society of America, Tucson,
can also be collected for process monitoring. Further works are un- Arizona, USA, 2010.
derway to implement reflectance monitoring in the system and extend [22] K. Juškevičius, et al., Fabrication of Nb2O5/SiO2 mixed oxides by reactive magne-
tron co-sputtering, Thin Solid Films 589 (2015) 95–104.
its applicability to non-transparent substrates. [23] Ö.D. Coşkun, S. Demirel, The optical and structural properties of amorphous Nb2O5
thin films prepared by RF magnetron sputtering, Appl. Surf. Sci. 277 (2013) 35–39.
[24] H. Muramatsu, K. Maki, S. Tanabe, Basic characteristics of quartz crystal sensor
6. Conclusions with interdigitated electrodes, Anal. Chem. Res. 7 (2016) 23–30.
[25] L. Qipeng, M. Huang, S. Deng, G. Li, Fabrication of broadband antireflection
coatings using wavelength-indirect broadband optical monitoring, Optik -
A thin film coating instrument that embodies broadband process International Journal for Light and Electron Optics 156 (2018) 325–332.
256
% T % T