Anda di halaman 1dari 8

International Journal of Engineering Research

Volume No.5, Issue No.7, pp : 610-617

ISSN:2319-6890)(online),2347-5013(print)
1 July 2016

Smart Model for Durable and Efficient Antireflection Coating Process


Khaled N. Faris
Electronics Research Institute, Dokki, Giza, Egypt
khalederi@yahoo.com
Abstract: Anti-reflection coatings are necessary to avoid
optical losses and increase the efficiency of the solar cells.
Stability for Anti-reflection (AR) coating process for solar cells
and optical devices is very sensitive to many variables. These
variables may be due to manufacturing process of depositing
AR coating or due to the environmental effects during normal
operation. In this paper a proposed fuzzy model is introduced
to assure quality of the coating process (CP) that guarantees
the stability, repeatability and durability of the CP.
Experimental data are collected for the coating process at
different coating parameters to build the proposed fuzzy model.
The validity of the model is verified using different input
parameters to the model, the output of the proposed model
agrees with the actual output from the Coating machine.
Current coating machines include sensing devices for all
coating parameters included in the model. Hence introducing
the proposed fuzzy model in the control system of the coating
machines would enable the prediction of the performance of
the coating process before actual running of the coating
process. The proposed model helps for the prediction of the
AR coating performance on optical devices under different
parameters variation; hence the same model can be applied to
the solar panels fabrications.
Keywords Antireflection Coating process (AR) ophthalmic lenses solar cells - fuzzy modelling (FM) Inverse Fuzzy Model (IFM).
I. I ntro d uct io n
Anti-reflective coatings are necessary to avoid optical losses and
increase the current of the solar cell, Solar cell fabrication
involves a number of major process steps, which include
formation of an N/P junction, Contact metallization and
antireflection coatings [1]. PV industry has used materials such
as silicon dioxide (Sio2) and Titanium Oxide (Tio2) for
antireflection coatings. Recently, Silicon nitride (Si3N4) is used.
Chemical vapour deposition technology is used for such AR
coating process for PV modules. It is vital that the coating
process be designed carefully to optimize optical and electronic
properties of the solar cell. Also the stability of antireflection
coating process is the core for obtaining reliable and efficient PV
panels.
The polished Si reflects about 37% of incident radiation, by
applying three layers of thin films, the reflection losses are
reduced to about 5.9% [2]. Figure 1 shows the solar spectrum
and reflectance of Si substrate without AR coating, also with
Silicon Nitride coating, and with three layers graded index
anti-reflection coating. It is obvious that with three layers
coating, the reflection percentage is the minimum. Without AR
coating the curve of the reflection percentage approaches 37%,
the ideal Si3N4 AR coating is considered the optimal as the
IJER@2016

reflection increases in the infrared region to protect the solar cell


from the drawbacks of the heat effect.

Figure 1: solar spectrum and reflectance of Si substrate with


different coatings layers [2]
Optimization for thickness of MgO and ZnO films are
introduced to reduce the front surface reflectivity of the GaAs
solar cell by depositing AR-coating [3]. The efficiency of the
cell approaches 33%. The effect of double layer of (Sio 2/Tio2)
antireflection coating on Silicon solar cells using RF Sputtering
technique is introduced in [4]. These results indicate that Double
layer Sio2/Tio2 coatings could be valuable in achieving highly
absorbent surfaces in optical devices as well as in the production
of high efficiency Silicon solar cells.
In [5] antireflection coating on poly-methyl-methacrylate
(PMMA) and polycarbonate substrate was introduced. A
reduction of about 3.5% to 4.0% in reflection in the visible light
range was achieved. The adhesion of the coatings to both
substrates was good. Some of optical devices are made of
polycarbonate and other polymers, such as CR39 optical lenses.
A proposed fuzzy model for the antireflection coating process at
different coating parameters is carried out on optical lenses.
Since plastic optics are now used for fairly precise single and
multiple-element lens systems, antireflection coating on plastics
are assuming an increasingly important function. Each plasticto-air interface reflects a nominal 4% of incident light per
surface.
Several coating options are available for reducing reflection
losses and increasing light transmission. Selection is based on
wavelength, required efficiency (ratio of transmitted light to
incident light), durability and cost, and reproducibility. So,
antireflection coatings for optical lenses are applied to reduce
glare and the resulting eye fatigue caused by uncoated lenses.

doi : 10.17950/ijer/v5s7/714

Page 610

International Journal of Engineering Research


Volume No.5, Issue No.7, pp : 610-617
The blue and green colour of A. R. coated lenses has turned out
to be an attractive feature for many users. Therefore the
measurement and control of the AR coating process and its
colour has become increasingly important. AR coatings have
several benefits to users; include the increasing of
transmissibility, reduction of the surface reflections, reduction of
ghost images, and the decreasing of the glare [6].
Many literatures handled the application of intelligent modelling
systems in different fabrication processes. In [7] the adaptive
Neuro-Fuzzy Inference System (ANFIS) was implemented to
predict the hardness of Titanium Aluminium Nitrite (TiAlN)
coating. In [8] the fuzzy ruled-based model was proposed for
predicting the roughness of Titanium Aluminium Nitride
(TiAlN) coating for final treatment of cuttings of the tungsten
carbide machine tools used in CNC machines. TiAlN coating
were produced using the magnetron sputtering process. Tungsten
carbide was used as the substrate and titanium alloy is the
coating material. the above mentioned models used in predicting
of hardness and roughness of the cutting machine tools leads to
enhance the finished surface fabrication of the glass moulds [9].
In the present research, a smart modelling system for AR coating
process is proposed to assure stability, quality, and durability of
those coatings for both optical devices and solar cells.
This paper comprises four sections, the first is the introduction,
the second section handles an explanation of the AR coating
process and system, the third introduces the proposed fuzzy
modelling system for that process, and the final section gives the
conclusions and recommended work.
II. Antireflection Coating Process
Anti-reflection coating is a type of optical coating applied to the
optical elements to reduce reflections. However coating would
be useful in improving other properties such as surface hardness,
smoothness, and surface cleanliness. Hence different materials
are used in coating process to achieve the required performance
expected from such optical elements. For example if the optical
device is an optical lens, the main target is to reduce reflection in
the optical spectrum and reflect all UV and IR spectrum,
however for PV modules, it is required to prevent reflection
from the specific part of the spectrum specified by the type of
PN junction used and reflect the harmful part of the spectrum
such as IR and UV. However for both the above mentioned
examples hardness, smoothness, and surface cleanliness are
considered as basic requirements.
Many coatings consist of transparent thin film structures with
alternating layers of different refractive index figure 2. Layer
thicknesses are chosen to produce destructive interference in the
beams reflected from the interfaces, and constructive
interference in the corresponding transmitted beams. This makes
the structure's performance change with wavelength and incident
angle, so that colour effects often appear at oblique angles. A
wavelength range must be specified when designing or ordering
such coatings, but good performance can often be achieved for a
relatively wide range of frequencies: usually a choice of infrared
(IR), visible, or ultraviolet (UV) is offered.

ISSN:2319-6890)(online),2347-5013(print)
1 July 2016

Figure (2): Multilayers coating on an optical lens (with courtesy


to Nikon coating)
Optical lens coatings generally fall into four categories. The
lowest layer closest to the lens surface is an impact-resistance
coat. Above this is a hard coat, which protects the lens from
scratches. The outermost layer is a water-repellent coat (super
hydrophobic) that prevents smudges from forming and allows
easy wiping off of oil films caused by fingerprints. It also
minimizes scratches created by repeated wiping, dust and nearly
invisible sand grains. Between the hard coat and water-repellent
layer are anti-reflective (AR) coats, crucial for the lens to
achieve clear vision as shown in figure 2.
Antireflection coating process is carried out in a large vacuum
chamber capable of achieving vacuum of at least 10 -6 torr., at
the bottom of the chamber is the source of the film material to be
vaporized. The substrates are mounted on a rotating dome.
Figure (3) shows the antireflection coating system, which
includes a Vacuum chamber, that contains, a rotating dome for
holding the substrates, Electron Beam Gun (E. B. G) for
chemical material evaporation process, shutter to control the
amount of evaporated material, Infrared heaters and
thermocouple, Plasma Ion Source for etching process,
refrigerator pipes (Cryo-pump) for frosting process during
chemical vapour deposition, and a peizo-electric crystal for film
thickness measurement.

Figure 3: Schematic diagram of antireflection coating system


IJER@2016

doi : 10.17950/ijer/v5s7/714

Page 611

International Journal of Engineering Research


Volume No.5, Issue No.7, pp : 610-617
The evaporation source is the Electron Beam Gun (EBG) which
is capable of vaporizing materials such as Titanium Oxide (Tio2)
and Zirconium Oxide (Zro2). Using large cooled crucibles
eliminates the chance of reaction between the heated coating
materials and the metal of the crucible.
A high Electron Beam Gun (EBG) of (1A 13KV) is aimed at the
film materials contained in a large, water cooled, copper
crucibles. Intense local heating melts and vaporizes some of the
coating material in the centre of the crucible without causing
excessive heating of the crucible itself. Expected variations of
temperature setting and vacuum setting affect the coating
properties; hence the deviation from the setting value of the
vacuum (Vacuum) and the temperature (Temperature) are of
vital importance.
An optical device that has been coated at an elevated
temperature requires very slow cooling to room temperature, to
avoid thermal stresses that may result from fast cooling, and
hence would improve quality and durability of the AR coating.
Temperature induced stress and poor interlayer adhesion are
the most common thickness-related limitations in optical thin
film production, so using the Plasma Assisted Ion Gun (PAIG)
minimizes overall coating thickness [10] and hence improves
adhesion quality. Plasma ion assisted deposition is a coating
technique, often applied at low temperatures, this technique
leads to minimizing wavelength shift, and achieves the highest
adhesion levels. This performance level is particularly critical in
many semiconductor, microelectronics, and telecommunications
applications.
The function of PAIG is to prepare the surface of the substrate to
improve the coating quality. The period of applying of PAIG
affect the adhesion force between the substrate and the coating
materials, increasing the PAIG period above the setting value
would slightly improve adhesion quality, however decreasing
this period would affect the quality of adhesion seriously.
The EBG is turned on, until the chemical of the source crucible
is at the proper temperature to evaporate. The shutter above the
source is opened to expose the chamber with the rotating dome
to vaporized material; where the distributions of vaporized
materials should be equal. When a particular layer is deposited
to the correct thickness, the shutter is closed and the source is
turned off. This process is repeated for the other sources. Optical
monitoring is the most common method of observing the
deposition process. A peizo-electric crystal is used for this
purpose. The stability and efficiency of the crystal play a vital
role in quality of coating process.
Refrigerating the outer surface of the chamber helps the vacuum
pump to reach its rated value and keeps this vacuum level almost
constant during the coating process. This factor affects the
coating process and introduced as an input to the proposed
model (Cryo-Pump)
A great care and control must be taken at every level in the
coating process. Several tests and measurements are carried out
to assure the dependability and accuracy of antireflection coating
process. So the next section handles a smart modelling to control
the coating process under different coating parameters to assure
high accuracy and durability for AR coating process. In the next
section a proposed fuzzy modelling system is designed taking
into consideration the different above mentioned parameters and
variables that effect on the coating process as inputs. And the
IJER@2016

ISSN:2319-6890)(online),2347-5013(print)
1 July 2016
outputs of the fuzzy model should indicate the quality (accuracy)
and the durability (dependability) of the AR coating.
III. Fuzzy Modelling System for AR Coating Process
Modelling, in general sense, refers to the establishment of a
description of a physical system in mathematical terms, which
characterizes the input-output behaviour of the underlying
system. Most physical systems, particularly those complex ones,
such as, coating process, are extremely difficult to model by an
accurate and precise mathematical formula or equation due to
the complexity of the system structure, nonlinearity, uncertainty,
and randomness.
Fuzzy modelling is needed in the following cases: Incomplete or
vague knowledge about systems, adequate processing of
imprecise information, and in grey modelling and identification.
In our problem according to imprecision in coating parameters,
such as chemicals stability in different coating cycles,
nonlinearities, and system uncertainties, the fuzzy modelling for
coating process is mandatory. So, in this work, a fuzzy
modelling system is introduced as an alternative to mathematical
modelling for many physical systems that are too complicated
such as AR coating process.

Figure3. Proposed smart model inputs and outputs for AR


coating process on ophthalmic lenses
The proposed smart model describes the process or the system
by establishing relations between the relevant system variables
[e.g., inputs and outputs] in the form of [if then] rules. Also the
proposed smart model is based on the fuzzy set theory which
mapping the input-output relations using linguistic membership
functions. The fuzzy modelling or the intelligent CAD system
helps in coating process control, also models the effect of each
parameter on the coating behaviour, the fuzzy modelling inputoutput is shown in figure 3.
A proposed smart coating machine would make use of the
proposed work, where a smart module can be inserted in the
control system of the coating machine which would accept the
introduction of the required characteristics of the output product
and process it to set the optimal values of the machine controls,
that is, this module contains the inverse fuzzy model of the
coating process, Figure 4.

doi : 10.17950/ijer/v5s7/714

Page 612

International Journal of Engineering Research


Volume No.5, Issue No.7, pp : 610-617

ISSN:2319-6890)(online),2347-5013(print)
1 July 2016
Low

Med

High

Degree of membership

0.8

0.6

0.4

0.2

0
40

45

50

55
Temperature

60

65

70

(b)
HC

MC

LC

Degree of membership

0.8

0.6

0.4

0.2

Figure4. Proposed IFM model inputs and outputs for AR coating


process on ophthalmic lenses

0
-130

-120

-110

-100

-90

-80

-70

-60

PolyCold

(c)
Bad

Med

Good

Degree of membership

0.8

0.6

0.4

0.2

0
90

91

92

93

94

95
Crystal

96

97

98

99

100

(d)
Clean

Moderate

Dirty

A training algorithm is used for the proposed fuzzy model,


which is based on the error between the output from the
proposed model and the AR coating process output, the tuning
process continues until the two outputs are matched with an
acceptable level of accuracy.

0.8

Degree of membership

Figure 5: Error based Tuned fuzzy model for A. R. Coating


Process

0.6

0.4

0.2

0
5

10

15

Moderate

SP

Unstable

MP

LP

0.8

Degree of membership

0.8

Degree of membership

25

(e)

3.1 Membership Functions of Input Variables


Stable

20

Chamber

0.6

0.4

0.6

0.4

0.2

0.2
0

0
1

8
Vacuum

10

12

1.5

2.5

3.5

PAIG

14
-5

x 10

(f)

(a)

Figure 6 (a-f): The Membership Functions of the Input Variables


IJER@2016

doi : 10.17950/ijer/v5s7/714

Page 613

International Journal of Engineering Research


Volume No.5, Issue No.7, pp : 610-617

ISSN:2319-6890)(online),2347-5013(print)
1 July 2016
R3: If (Vacuum is Moderate) and (Temperature is Low) and
(PolyCold is HC) and (Crystal is Good) and (Chamber is Clean)
and (PAIG is SP) then (CQuality is ZD)(Reflection is
Med)(CDuarability is MD) (1).

3.2 Membership Functions of Output Variables


NBD

NMD

NSD

ZD

PSD

PMD

PBD

Degree of membership

0.8

R4: If (Vacuum is Stable) and (Temperature is Med) and


(PolyCold is LC) and (Crystal is Good) and (Chamber is Clean)
and (PAIG is SP) then (CQuality is PBD)(Reflection is
Med)(CDuarability is LD) (1)

0.6

0.4

0.2

0
-40

-30

-20

-10

0
CQuality

10

20

30

40

(a)
Low

Med

R5: If (Vacuum is Unstable) and (Temperature is Low) and


(PolyCold is MC) and (Crystal is Bad) and (Chamber is Clean)
and (PAIG is SP) then (CQuality is PBD)(Reflection is
Med)(CDuarability is LD) (1)
IV. Experimental results and the proposed model recorded
outputs

High

Degree of membership

0.8

0.6

0.4

0.2

0
0.4

0.6

0.8

1.2
Reflection

1.4

1.6

1.8

(b)
LD

MD

HD

Figure 8: Coated and Uncoated Spectacle Lens (With courtesy to


Zee Eyes Optical Company)

Degree of membership

0.8

0.6

0.4

0.2

0
0.5

0.55

0.6

0.65

0.7

0.75
CDuarability

0.8

0.85

0.9

0.95

(c)
Figure 7 (a-c): The Membership Functions of the Output
Variables
3.3 Some Rules of the proposed Fuzzy Model:
R1: If (Vacuum is Unstable) and (Temperature is High) and
(PolyCold is LC) and (Crystal is Good) and (Chamber is Clean)
and (PAIG is SP) then (CQuality is PBD)(Reflection is
High)(CDuarability is LD) (1).
R2: If (Vacuum is Stable) and (Temperature is Low) and
(PolyCold is HC) and (Crystal is Good) and (Chamber is Clean)
and (PAIG is LP) then (CQuality is ZD)(Reflection is
Med)(CDuarability is HD) (1).

IJER@2016

Figure 9: Visible Light Spectrum.


The coated lens and the uncoated one is shown in figure (8), the
desired colour quality is obtained according to the visible
spectrum as shown in figure (9). In solar cells AR coating
process, the operating range of the spectrum is (500nm-650nm)
to obtain the maximum energy from the sun light as shown in
figure (1).
So, in the coating process for optical lenses, the preferable the
coating colour is the green (550nm).So, some of optical lenses
are coated using VAC TEC ORION 40 FT COATER at different
operating parameters, SPECTROMETER/DATA SYSTEM ,
JASCO Corp., V-570, Rev. 1.00 is used for measuring
reflectivity percentage. The reflectivity percentage curves are

doi : 10.17950/ijer/v5s7/714

Page 614

International Journal of Engineering Research


Volume No.5, Issue No.7, pp : 610-617

ISSN:2319-6890)(online),2347-5013(print)
1 July 2016

recorded as in Figures (11a-d). Some of the durability tests are


carried out as in [11-13], and the successful and Fail percentages
for different are recorded and the Coating durability
(CDurability) membership functions are constructed.

7
6
Reflection %

4
X: 550
Y: 4.093
Reflection %

5
4
3
X: 550
Y: 1.28

1
0

400

450

500

550
600
wave length in nm

650

700

750

(d)
1

400

450

500

550
600
wave length in nm

650

700

650

700

750

(a)
12
10

Figure 10: Spectrophotometer results for Reflection % at


different coating parameters
Figure 10a, describes the %R curve for uncoated ophthalmic
lens, the remaining curves are recorded at different coating
parameters, as in table (1).
V. Results Of The Fuzzy Model And Actual AR Coating
Process At Different Parameters
Table1: Actual Data from AR Coating system at different
coating Variables

Reflection %

8
6

4
X: 550
Y: 1.175

400

450

500

550
600
wave length in nm

750

Table2: Look up Table for Some of Fuzzy Rules

(b)
10

Reflection %

4
X: 550
Y: 1.192

400

450

500

550
600
wave length in nm

(c)
IJER@2016

650

700

750

Table3: Comparison between actual process and Fuzzy Model at


different coating parameters
doi : 10.17950/ijer/v5s7/714

Page 615

International Journal of Engineering Research


Volume No.5, Issue No.7, pp : 610-617

ISSN:2319-6890)(online),2347-5013(print)
1 July 2016

(c)

(a)

(d)

(b)
(e)
Figure 11 (a-e): Fuzzy Inference system Outputs at different
coating parameters (Rule (1)-Rule (5)).
IJER@2016

doi : 10.17950/ijer/v5s7/714

Page 616

International Journal of Engineering Research


Volume No.5, Issue No.7, pp : 610-617
The outputs of some Fuzzy Rules are shown in figures 11a-11e,
it is clear that the outputs of the AR process in Table (1) and the
output from the proposed fuzzy model in table (2) are matched
with an acceptable level of accuracy as shown in Table (3).
. =

1
=1

100%

(1)

The model performance is measured by the accuracy


determination using equation (1) as in [7], where n = number of
modeled data, Va = actual Value, Vm = modeled Value. The
model accuracy is about 97.7% , which is an acceptable level.
Table 4 shows the abbreviations of the used fuzzy membership
functions.
Table 4: Membership Functions Linguistic Variables
LC: Low Cooling
MC: Medium Cooling
HC: High Cooling
SP: Short Period
MP: Medium Period
LP: Long Period
ZD: Zero Drift

PBD: Positive Big Drift


NBD: Negative Big Drift
NMD: Negative Med. Drift
PMD: Positive Med. Drift
HD: High Durable
MD: Medium Durable
LD: Low Durable

VI. CONCLUSION
Fuzzy model has been built to enhance the efficiency of the
optical devices in its operating optical spectrum. The validity of
the model has been verified using different input parameters to
the model, the output of the proposed model agrees with the
actual output from the Coating machine. Introducing the
proposed fuzzy model in the control system of the coating
machines would enable the prediction of the performance of the
coating process before actual running of the deposition process.
The proposed model helps in prediction of the AR coating
performance on optical devices and solar modules under
different parameters variation to increase the efficiency of the
Photovoltaic Modules (PV). In this paper a super hydrophobic
coating for PV Modules has been recommended to minimize
scratches created by dust and sand grains, also it helps for rapid
slippage of the dead birds to enable the PV modules to behave as
a self-cleaning system.

ISSN:2319-6890)(online),2347-5013(print)
1 July 2016
REFERENCES
i.
Shubham Duttagupta, Fajun Ma, Bram Hoex,
Thomas Mueller, and Armin G. Aberle: Optimized Antireflection
Coatings using Silicon Nitride Textured Silicon Surface based on
Measurements and multidimensional modelling, Energy Procedia 15
(2012) 78-83.
ii.
Sameer Chhajed, Martin F. Schubert, Jong Kyu
Kim and E. Fred Schubert: Nanostructured multilayer graded-index
antireflection coating for Si solar cells with broadband and
omnidirectional characteristics, Applied physics letters 93, 251108,
2008, 2008 American Institute of Physics..
iii.
Ashwini Joshi Thosar, Manoj Thosar, and R. K.
Khanna: Optimization of Anti-reflection Coating for Improving the
Performance of GaAs Solar Cell , Indian Journal of Science and
Technology, Vol 7(5), 637641, May 2014.
iv.
Khuram Ali, Sohail A. Khan, and M. Z. Mat Jafri:
Effect of Double Layer (SiO2/TiO2) Anti-reflective Coating on Silicon
Solar Cells, Int. J. Electrochemical Science. 9 (2014) 7865 7874.
v.
Marieke Burghoorn , Dorrit Roosen-Melsen , Joris
de Riet , Sami Sabik , Zeger Vroon , Iryna Yakimets and Pascal
Buskens: Single Layer Broadband Anti-Reflective Coatings for Plastic
Substrates Produced by Full Wafer and Roll-to-Roll Step-and-Flash
Nano-Imprint Lithography,
Materials 2013, 6, 3710-3726;
doi:10.3390/ma6093710.
vi.
Abdul Syukor Mohamad Jaya and et al:
Application of ANFIS in Predicting of TiAlN Coatings Hardness
Australian Journal of Basic and Applied Sciences, 5(9): 1647-1657,
2011, ISSN 1991-8178.
vii.
Abdul
syukor
mohamad
Jaya,
and
et
al:Application of Fuzzy Rule-Based Model to Predict TiAIN Coatings
Roughness Applied Mechanics and Materials Vols. 110-116 (2012)
PP. 1072-1079.
viii.
Ahmed A. D. Sarhan, M. Sayuti, and M. Hamdi: A
Fuzzy Logic Based Model to Predict Surface Roughness of A Machined
Surface in Glass Milling Operation Using CBN Grinding Tool, World
Academy of Science, Engineering and Technology, Vol. 6, 2012-10-26,
pp. 564-570.
ix.
Optical
Coatings
by
MELLES
GRIOT,
www.mellesgriot.com.
x.
AS Carlson: A Comparison Study of anti-reflection
Coatings, The South African Optometrist September 2003 62 (3) 9298.
xi.
G. Jorgensen and et al: Durability Testing of
Antireflection Coatings for Solar Applications, SPIEs 44th Annual
Meeting and Exhibition, Denver, Colorado, July 18-23, 1999.
xii.
M. Walls and A. G. Spencer: Hard Coating and
the Durability of Anti-Reflection Coatings, Optical World, April 2000.

ACKNOWLEDGEMENT
The author is appreciative to Zee Eyes Optical Company,
especially Mr. Fahd A. Khalil, Dr. Shehab A. Khalil, and AR
Coatings Department, especially eng. Ebtihal, Mahmoud Abd
EL-Rahman for their support in data collection and AR coating
process execution at different operating conditions. Also, the
Author is thankful to Prof. Dr. Mohsen T. EL-Hagry for valuable
notes and discussions. Finally, The Author is Grateful to
Electronics Research Institute, especially, the Power Electronics
and Energy Conversion Department for their support.

IJER@2016

doi : 10.17950/ijer/v5s7/714

Page 617