PHD Thesis (Final Ver.2)

HIGH SPEED VERY THIN FILMS
WITH REVERSE ROLL COATING
SUBTITLE: An experimental investigation of reverse roll coating of
fluids using rigid and deformable rolls at high speeds
Yusuke SHIBATA (BEng, MEng)
Submitted as a PhD thesis for the
School of Engineering, Design and Technology
University of Bradford
2012
Y.SHIBATA
HIGH SPEED VERY THIN FILMS
WITH REVERSE ROLL COATING
Keywords: Reverse roll coating, Rigid, Deformable, Very thin films, Ribbing, Air entrainment,
Cascade
Abstract
The objective of a coating operation is to transfer a defect free liquid film
onto a continuous substrate in order to meet the requirements of the final products.
Mainly two concerns govern the process. The first concern is the economics of the
process and the second concern is the quality of the coated film. The economics of
the process are dictated by the speed of coating and the film thickness. Clearly,
higher speeds mean better productivity hence less cost of operation and thinner
films are desirable because less material is being used. Quality is governed by film
uniformity and integrity, indicating that the film will perform as designed. Film
defects such as streaks or tiny air bubbles are indication that the film properties are
not uniform rendering it unacceptable to customers. One of the most versatile
coating systems to achieve thin films at high speeds is reverse roll coating which has
been used for a long time all over the world. At low speed, typically 1m/s, this
coating operation is inherently stable and with small gaps of order 100 microns can
i
lead to film thickness of order 30-50 microns. Much research, theoretical and
experimental, has been devoted to this coating flow but only at low speeds and for
large gaps (>100 microns). There are no comprehensive data how very thin films, 20
microns and less (particularly lower limits in the region of 5 microns) can be
achieved at high speeds, of 2 or more metres per second. This study is concerned
precisely with this aim, that of investigating the effect of large speeds and small
roller gaps (rollers nearly touching or in elastohydrodynamic contact) to achieve the
very thin films desired by modern applications (electronics, medical and others). In
order to achieve this aim, a rig was designed and built to enable to understand the
effect of various coating conditions and liquid properties on the metered film
thickness and coating instability. To achieve thin films at high speeds, small roll gap
and low viscosity are needed, however flow instabilities will develop under these
conditions. To achieve stable coating window at high speeds high surface tension is
needed. It was found that the roll gap and the viscosity have complicated effect on
the coating window. In the case of low viscosity liquid (7mPa.s), small roll gaps are
needed, whereas in the case of high viscosity liquid (more than 30mPa.s), large gaps
are needed. It was found that Weber number is better describer for ribbing
instability in rigid reverse roll coating unlike in rigid forward roll coating in which
capillary number is the one.
In addition the potential of reverse deformable roll coating (rolls in
elastohydrodynamic contact) was investigated in order to achieve much thinner
films at higher speeds. As a result of the investigation of reverse deformable roll
coating, it was found that there is a possibility to get much thinner stable films at
much higher speeds compared to reverse rigid roll coating.

ii
The liquid transfer from an applicator roller to a PET film was investigated in
this study. It was found that air stagnation at downstream meniscus and air
entrainment at upstream meniscus depend on the liquid properties such as viscosity
and surface tension and coating conditions such as web tension and wrap angle of
web. As a result, wet film instability also depends on liquid properties and coating
conditions. It was found that air stagnation causes streaks on the wet film and air
entrainment caused bubbles on the wet film. To get a stable wet film, it was found
that suitable viscosity and high surface tension were needed.
iii
Table of Contents
Chapter 1 : INTRODUCTION ...................................................................................... 1

1.1 COATING: INDUSTRIAL APPLICATIONS & CHALLENGES .......................................... 1
1.2 COATING FLOWS: DEFINITION, CLASSIFICATIONS & INSTABILITIES....................... 2
1.3 AIM AND OBJECTIVES OF RESEARCH ....................................................................... 8
Chapter 2 : LITERATURE REVIEW ........................................................................ 10
2.1 INTRODUCTION ...................................................................................................... 10
2.2 BRIEF REVIEW OF VARIOUS COATING METHODS ................................................. 10
2.2.1 Free coating .................................................................................................... 10
2.2.2 Metering coating ............................................................................................ 13
2.2.3 Pre-metered coating....................................................................................... 13
2.2.4 Gravure or print coating ................................................................................. 21
2.3 ROLL COATING........................................................................................................ 25
2.4 FILM THICKNESS IN ROLL COATING FLOWS .......................................................... 29
2.4.1 Forward Rigid Roll Coating ............................................................................. 29
2.4.2 Reverse Rigid Roll Coating.............................................................................. 40
2.4.3 Forward Deformable Roll Coating ................................................................. 49
2.4.4 Reverse Deformable Roll Coating .................................................................. 52
2.5 RIBBING INSTABILITIES IN ROLL COATING FLOWS ................................................ 53
2.6 AIR ENTRAINMENT IN ROLL COATING FLOWS ...................................................... 61
2.6.1 The Nature and Physics of Air Entrainment .................................................. 62
2.7 LIQUID TRANSFER FROM ROLL TO WEB ................................................................ 72
2.8 THE NEW ROLL COATING RESEARCH CHALLENGES ............................................... 73
2.9 CONCLUSIONS ........................................................................................................ 74
Chapter 3 : EXPERIMENTAL METHOD ................................................................ 76
iv
3.1 INTRODUCTION ...................................................................................................... 76
3.2 THE COATER & MEASURING TECHNIQUES (at the University of Bradford) ......... 76
3.2.1 Rollers & Coater Configuration ...................................................................... 76
3.2.2 Feed supply ..................................................................................................... 78
3.2.3 Measurements of wet film thickness ............................................................ 84
3.2.4 Observation of the coating instabilities ........................................................ 84
3.3 THE COATER & MEASURING TECHNIQUES (at Toyobo) ........................................ 87
3.3.1 Rollers & Coater Configuration ...................................................................... 87
3.3.2 Measurements of wet film thickness ............................................................ 88
3.3.3 Observation of the coating instabilities ........................................................ 88
3.4 CHARACTERISATION OF THE COATING FLUIDS ..................................................... 89
Chapter 4 : RESULTS & DISCUSSION .................................................................... 93
4.1 INTRODUCTION ...................................................................................................... 93
4.2 PRELIMINARY MEASUREMENT WITH THE TRANSPARENT ROLLER...................... 93
4.2.1 Film Thickness and Instabilities ..................................................................... 94
4.2.2 Meniscus Changes .......................................................................................... 95
4.3 MEASUREMENT WITH A STEEL PRECISION METERING ROLLER ........................... 96
4.3.1 Effect of roller speed ...................................................................................... 96
4.3.2 Effect of roller gap at constant viscosity ....................................................... 99
4.3.3 Effect of supply film thickness ..................................................................... 102
4.3.4 Effect of viscosity at constant roller gap ..................................................... 108
4.3.5 Coating Window dependence on all operating variables ........................... 111
4.3.6 Effect of surface tension............................................................................... 118
4.3.7 Film thickness analysis ................................................................................. 122
4.3.8 Summary ....................................................................................................... 128
4.4 ANALYSIS BY MEASUREMENT WITH TRANSPARENT ROLLER............................. 128
4.4.1 Effect of roller speed .................................................................................... 128
4.4.2 Effect of roll gap ........................................................................................... 129
4.4.4 Effect of viscosity .......................................................................................... 132
4.4.6 Ribbing mechanism at high speeds.............................................................. 137
4.5 MEASUREMENT WITH DEFORMABLE ROLLER .................................................... 141
4.5.1 Effect of roller speed .................................................................................... 142
v
4.5.2 Effect of Viscosity ......................................................................................... 144
4.5.3 Effect of rubber hardness ............................................................................. 146
4.5.4 Effect of roller gap ........................................................................................ 149
4.4.6 Effect of surface tension (negative gap) ...................................................... 159
4.5.7 Stable condition............................................................................................ 162
4.5.8 Effect of roller speed at Positive Gap .......................................................... 164
4.5.9 Effect of hardness (positive gap).................................................................. 171
4.5.10 Effect of roll gap (negative and positive gap) .......................................... 175
4.5.11 Conclusions ............................................................................................... 178
4.6 LIQUID TRANSFER FROM APPLICATOR ROLLER TO PET FILM ............................. 178
4.6.1 Effect of viscosity .......................................................................................... 180
4.6.3 Effect of wrap angle...................................................................................... 192
4.6.4 Effect of web tension ................................................................................... 195
4.6.5 Summary ....................................................................................................... 197
Chapter 5 : THEORETICAL CONSIDERATIONS ............................................... 200
5.1 INTRODUCTION .................................................................................................... 200
5.2 LUBRICATION MODEL FOR RIGID REVERSE ROLL COATING ............................... 200
5.3 LUBRICATION MODEL FOR DEFORMABLE REVERSE ROLL COATING .................. 211
5.4 CONCLUSION ........................................................................................................ 213
Chapter 6 : CONCLUSIONS ..................................................................................... 214
6.1 INTRODUCTION .................................................................................................... 214
6.2 RIGID REVERSE ROLL COATING ............................................................................ 215
6.3 DEFORMABLE REVERSE ROLL COATING .............................................................. 218
6.4 LIQUID TRANSFER FROM ROLL TO WEB .............................................................. 221
6.5 RECOMMENDATIONS FOR FUTURE WORK ......................................................... 223
REFERENCES ............................................................................................................ 225
vi
List of Figures
Figure 1.1: Free coating ........................................................................................................... 3

Figure 1.2: Metering coating .................................................................................................... 4
Figure 1.3: Pre-metered coating .............................................................................................. 5
Figure 1.4: Gravure or print coating ........................................................................................ 5
Figure 1.5: Coating instabilities (a) Ribbing, (b) Air entrainment ............................................ 6
Figure 2.1: Free coating to a vertical moving surface (Spiers et al., 1974) ............................ 12
Figure 2.2: Slot die coating onto a roller (Lee et al., 1992) .................................................... 14
Figure 2.3: Tensioned-web slot die coating onto a substrate (Lin et al., 2007)..................... 16
Figure 2.4: Slide coating system (Hens and Abbenyen, 1997) ............................................... 18
Figure 2.5: Curtain coating system (Blake and Clarke, 1994) ................................................ 21
Figure 2.6: Schematic representation of the gravure geometries (Patel and Benkreira, 1991)
............................................................................................................................................... 22
Figure 2.7: Different modes of gravure roll coating (Patel and Benkreira, 1991) ................. 22
Figure 2.8: Typical roll coating operation (Coyle, 1990a) ...................................................... 26
Figure 2.9: Roll coating arrangements ................................................................................... 27
Figure 2.10: Variety of substrate configurations ................................................................... 28
Figure 2.11: Lubrication model (Greener and Middleman, 1979) ......................................... 31
Figure 2.12: Coating thickness as a function of modified Ca number (Greener and
Middleman, 1979) .................................................................................................................. 32
Figure 2.13: Semi-empirical model (Benkreira et al., 1981b) ................................................ 33
Figure 2.14: Results of semi-empirical model (Benkreira et al., 1981b) ................................ 34
Figure 2.15: Finite element analysis (Coyle et al., 1986) ....................................................... 36
Figure 2.16: Flow rate from the finite element analysis (Coyle et al., 1986)......................... 37
Figure 2.17: Coating regimes ................................................................................................. 38
Figure 2.18: Comparison of inlet/outlet film thickness ratio versus speed ratio results of
meniscus roll coating with classical roll coating predictions (Malone, 1992) ....................... 38
Figure 2.19: Effect of viscoelasticity on the film thicknesses (Han et al., 2009).................... 39
Figure 2.20: Simple model of reverse roll coating system (Greener, 1981) .......................... 41
Figure 2.21: Measured flow rate as a function of speed ratio compared to predictions (Coyle,
1990b) .................................................................................................................................... 42
Figure 2.22: Reverse roll coating arrangement (Benkreira et al., 1981a).............................. 43
Figure 2.23: Metered film thickness as a function of speed ratio and Capillary number
(Coyle, 1990a) ........................................................................................................................ 44
Figure 2.24: Dependence of measured metered film thickness on the arriving film thickness
(Coyle, 1990a) ........................................................................................................................ 45
Figure 2.25: Metered film thickness as function of arriving film thickness, roll radius and roll
radius ratio (Hao and Haber, 1999) ....................................................................................... 46
Figure 2.26: Effect of reservoir height for (a) flow rate, (b) wetting line position, as a
function of speed ratio (Thompson et al., 2001) ................................................................... 47
Figure 2.27: Measured metered film thickness of (a) shear-thinning solutions and (b)
viscoelastic solutions (Coyle et al., 1990d) ............................................................................ 48
Figure 2.28: Effect of viscoelasticity on the film thickness (Han et al., 2009) ....................... 48
vii
Figure 2.29: Simple one-dimensional elastohydrodynamic model of forward deformable roll
coating (Coyle, 1988) ............................................................................................................. 50
Figure 2.30: Deviation of the experimental data from Coyle’s theoretical predictions as a
function of the ratio between the cover thickness and the contact half-width (Cohu and
Magnin, 1997) ........................................................................................................................ 51
Figure 2.31: Flow rate at different elasticity numbers and gaps (positive and negative)
(Carvalho, 2003) ..................................................................................................................... 52
Figure 2.32: Schematic of a sinusoidal perturbation to a coating meniscus (Coyle, 1997) ... 54
Figure 2.33: Computed pressure along the main flow direction in forward roll coating (Coyle,
1997) ...................................................................................................................................... 55
Figure 2.34: Critical capillary number for the onset of ribbing as a function of the
gap/diameter (Coyle, 1990c) ................................................................................................. 57
Figure 2.35: Computed pressure along the main flow direction in reverse roll coating
(Coyle, 1997) .......................................................................................................................... 59
Figure 2.36: Computed pressure gradient at metered film meniscus as a function of speed
ratio and Ca number (Coyle, 1990a) ...................................................................................... 59
Figure 2.37: Three-phase junction in dip coating: moving substrate/coating
liquid/surrounding air ............................................................................................................ 62
Figure 2.38: Break up of the dynamic wetting line (Deryagin and Levi, 1964) ...................... 63
Figure 2.39: Schematic representation of saw-teeth （Blake and Ruschak, 1979）............ 65
Figure 2.40: Inclined dip coating (Cohu and Benkreira, 1998)............................................... 66
Figure 2.41: Curtain coating (Blake, 1994)............................................................................. 67
Figure 2.42: Relationship between gas pressure and entrainment speed (Benkreira et al.,
2008) ...................................................................................................................................... 67
Figure 2.43: Computed steady-state free surface profiles (Coyle, 1990a) ............................ 69
Figure 2.44: Coating window in reveres roll coating (Coyle, 1990a) ..................................... 71
Figure 3.1: Schematic diagram and picture of reverse roll coating rig with a steel round bar
............................................................................................................................................... 79
Figure 3.2: Picture of drive and control box arrangement ................................................... 79
Figure 3.3: Pictures of accurate stainless shims placed between the roller bearing blocks 80
Figure 3.4: Pictures of scraper systems ................................................................................ 80
Figure 3.5: Picture of a micrometer ...................................................................................... 80
Figure 3.6: Relationship between roll/bar gap and actual wet film thickness ...................... 81
Figure 3.7: Schematic diagram of reverse roll coating rig with a die .................................... 83
Figure 3.8: Schematic diagram of die position....................................................................... 83
Figure 3.9: Film thicknesses as a function of the gear pump frequency and the gap between
an applicator roller and a die ................................................................................................. 84
Figure 3.10: Configuration for visualisation of the meniscus region ..................................... 86
Figure 3.11: Photograph of a graticule and a small bead placed in the gap between the
rollers (a) Graticule, (b) Small bead ....................................................................................... 86
Figure 3.12: Illustration of the technique of setting centre position..................................... 87
Figure 3.13: Schematic diagram of reverse roll coating rig ................................................... 88
Figure 3.14: Liquid viscosity as a function of shear rate ........................................................ 92
Figure 4.1: Film thicknesses on the applicator roller and the metering roller as a function of
roller speed ratio.................................................................................................................... 94
viii
Figure 4.2: (1) Images of meniscus situations changes; (2) Images of applicator roller after
metering................................................................................................................................. 95
Figure 4.3: Effect of applicator roller speed. Supply film=50 m, Roll gap=25 m ................. 97
Figure 4.4: Effect of applicator roller speed. Supply film=100 m, Roll gap=50 m ............... 97
Figure 4.5: Effect of applicator roller speed. Supply film=200 m, Roll gap=100 m ............. 98
Figure 4.6: Relationship between applicator roll speed and minimum film thickness. Roll
gap=25 m, Supply film=50 m ............................................................................................... 98
gap=50 m, Supply film=100 m ............................................................................................. 99
gap=100 m, Supply film=200 m ........................................................................................... 99
Figure 4.9: Effect of the roll gap. Applicator Roll Speed= 30m/min .................................... 100
Figure 4.10: Effect of the roll gap. Applicator Roll Speed=60m/min ................................... 101
Figure 4.11: Effect of the roll gap. Applicator Roll Speed=90m/min ................................... 101
Figure 4.12: Effect of the roll gap. Applicator Roll Speed=120m/min ................................. 102
Figure 4.13: Effect of the supply film thickness. Applicator Roll Speed=6m/min................ 104
Figure 4.14: Effect of the supply film thickness Applicator Roll Speed=15m/min ............. 104
Figure 4.15: Effect of the supply film thickness. Applicator Roll Speed=30m/min.............. 105
Figure 4.18: Effect of the supply film thickness. Applicator Roll Speed=120m/min............ 106
Figure 4.19: Dimensionless minimum metered film thickness (Effect of the supply film
thickness ) Liquid: Millmax5, Roll gap=100 m................................................................... 107
Figure 4.20: Dimensionless minimum metered film thickness (Effect of the supply film
thickness ) Liquid: Millmax5, Roll gap=25, 50, 100, 200 m ............................................... 107
Figure 4.21: Effect of the viscosity. Applicator Roll Speed=30m/min ................................. 109
Figure 4.24: Effect of the viscosity. Applicator Roll Speed=120m/min ............................... 110
Figure 4.25: Coating window (Effect of the roll gap). Applicator Roll Speed=30m/min...... 112
Figure 4.28: Coating window (Effect of the roll gap). Applicator Roll Speed=120m/min ... 115
Figure 4.29: Coating window (Effect of the roll gap). Applicator Roll Speed=150m/min ... 116
Figure 4.30: Coating window (Effect of viscosity). ............................................................... 118
Figure 4.31: Effect of the surface tension. Applicator Roll Speed=30m/min ...................... 119
Figure 4.34: Effect of the surface tension. Applicator Roll Speed=120m/min .................... 120
Figure 4.35: Effect of the surface tension. Applicator Roll Speed=150m/min .................... 121
Figure 4.36: Minimum metered film thickness (Effect of the surface tension). .................. 121
Figure 4.37: Coating window (Effect of the surface tension). ............................................. 122
Figure 4.38: Film thickness measured in this study and comparison with data of Coyle et al.
(1990). .................................................................................................................................. 123
Figure 4.39: Correlation between Ca number and dimensionless minimum film thickness124
ix
Figure 4.40: Meniscus shape................................................................................................ 124
Figure 4.41: Critical speed ratio vs. Ca: data from this study (¨) and comparison with Coyle
et al. (1990) data ( ) and Kang and Liu (1991) correlation (-------). ..................................... 125
Figure 4.42: Comparison of film thickness between Coyle’s and this study (Ca=0.1) ......... 126
Figure 4.43: Comparison of film thickness between Coyle’s and this study (Ca=0.6-0.7, 2.3-
2.4) ....................................................................................................................................... 127
Figure 4.44: Minimum Film thickness measured in this study (¨) and comparison with data
of Coyle et al. (1990) ( ). ..................................................................................................... 127
Figure 4.45: Effect of roll speed ratio on meniscus positions. Liquid: Millmax5 ................. 130
Figure 4.46: Effect of applicator roller speed on meniscus positions. ................................. 130
Figure 4.47: Effect of roll gap on meniscus positions. ......................................................... 131
Figure 4.48: Effect of supply film thickness on meniscus positions..................................... 132
Figure 4.49: Effect of roll speed ratio on meniscus positions. Liquid: Millmax15 ............... 133
Figure 4.50: Effect of roll speed ratio on meniscus positions. Liquid: Millmax68 ............... 133
Figure 4.51: Effect of viscosity on meniscus positions......................................................... 134
Figure 4.52: Pictures of meniscus regions with various viscosities at 90m/min ................. 135
Figure 4.53: Pictures of meniscus regions with various viscosities at 120m/min ............... 135
Figure 4.54: Effect of roll speed ratio on meniscus positions Liquid: Water/Glycerine .... 136
Figure 4.55: Effect of surface tension on meniscus positions ............................................. 137
Figure 4.56: Pictures of meniscus regions with various surface tension at 120m/min ....... 137
Figure 4.57: Ca number as a function of roll gap with various liquids (individuals) ............ 139
Figure 4.58: Ca number as a function of roll gap with various liquids (all in one) .............. 140
Figure 4.59: Re number as a function of roll gap with various liquids (all in one) .............. 140
Figure 4.60: We number as a function of roll gap with various liquids (all in one) ............. 141
Figure 4.61: Effect of roller speed. Liquid: Millmax5 ........................................................... 143
Figure 4.62: Effect of roller speed. Liquid: Millmax15 ......................................................... 143
Figure 4.63: Effect of roller speed. Liquid: Millmax68 ......................................................... 144
Figure 4.64: Effect of Viscosity. Applicator roll speed=30m/min ........................................ 145
Figure 4.65: Effect of Viscosity. Applicator roll speed=90m/min ........................................ 145
Figure 4.66: Effect of Viscosity. Applicator roll speed=150m/min ...................................... 146
Figure 4.67: Effect of Rubber hardness. Liquid: Millmax5 (7mPa.s) ................................... 147
Figure 4.68: Effect of Rubber hardness. Liquid: Millmax15 (30mPa.s) ............................... 148
Figure 4.69: Effect of Rubber hardness. Liquid: Millmax68 (180mPa.s) ............................. 149
Figure 4.70: Effect of roller gap. Liquid: Millmax5, Applicator roll speed=30m/min .......... 150
Figure 4.71: Effect of roller gap. Liquid: Millmax5, Applicator roll speed=90m/min .......... 151
Figure 4.72: Effect of roller gap. Liquid: Millmax5, Applicator roll speed=150m/min ........ 151
Figure 4.75: Effect of roller gap. Liquid: Millmax15, Applicator roll speed=150m/min ...... 153
Figure 4.78: Effect of roller gap. Liquid: Millmax68, Applicator roll speed=150m/min ...... 154
Figure 4.79: Effect of supply film thickness. Rubber shore hardness=30A, Liquid: Millmax68
............................................................................................................................................. 156
x
............................................................................................................................................. 157
............................................................................................................................................. 158
............................................................................................................................................. 159
Figure 4.83: Effect of surface tension on the metered film thickness. ................................ 161
Figure 4.84: Effect of surface tension on the coating window. ........................................... 161
Figure 4.85: Film thickness with steel and deformable rollers at positive gaps. Liquid:
Millmax5 .............................................................................................................................. 165
Figure 4.86: Coating window with steel and deformable rollers at positive gap. Roll
gap=50 m, Liquid: Millmax5................................................................................................ 166
gap=100 m, Liquid: Millmax5 ............................................................................................. 166
Millmax15 ............................................................................................................................ 167
gap=100 m, Liquid: Millmax15 ........................................................................................... 168
Millmax68 ............................................................................................................................ 169
gap=100 m, Liquid: Millmax68 ........................................................................................... 170
Figure 4.94: Effect of rubber hardness. Liquid: Millmax5 .................................................... 172
Figure 4.95: Effect of rubber hardness. Liquid: Millmax15.................................................. 173
Figure 4.96: Effect of rubber hardness. Liquid: Millmax68.................................................. 174
Figure 4.97: Effect of roll gap (negative and positive). Applicator roll speed=30m/min..... 176
Figure 4.98: Effect of roll gap (negative and positive). Applicator roll speed=90m/min..... 177
Figure 4.99: Effect of roll gap (negative and positive). Applicator roll speed=150m/min... 177
Figure 4.100: Minimum Film thickness measured in this study .......................................... 179
Figure 4.101: Critical speed ratio vs. Ca: data from this study ............................................ 180
Figure 4.102: Effect of PET film / Applicator roll speed ratio on wet film instabilities.
Viscosity=3mPa.s, Surface tension=26mN/m ...................................................................... 181
Viscosity=44mPa.s, Surface tension=26mN/m .................................................................... 181
Viscosity=105mPa.s, Surface tension=26mN/m .................................................................. 182
Figure 4.105: Pictures of transfer regions at various F/A ratio............................................ 183
Figure 4.106: Pictures of transfer regions with different viscosity liquids. ......................... 183
Figure 4.107: Effect of PET film / Applicator roll speed ratio at various applicator roll speed
on meniscus positions. ......................................................................................................... 185
Figure 4.108: Effect of viscosity on meniscus positions. Applicator roller speed: 30m/min185
xi
Figure 4.111: Effect of PET film / applicator roll speed ratio on liquid transfer ratio.
Viscosity=3mPa.s, Surface tension=26mN/m ...................................................................... 187
Viscosity=44mPa.s, Surface tension=26mN/m .................................................................... 187
Viscosity=105mPa.s, Surface tension=26mN/m .................................................................. 188
Figure 4.114: Effect of PET film / Applicator roll speed ratio on wet film instabilities. ....... 189
Figure 4.115: Pictures of transfer regions with different surface tension liquids. .............. 190
Figure 4.116: Effect of surface tension on meniscus positions. Applicator roller
speed=30m/min ................................................................................................................... 191
speed=60m/min ................................................................................................................... 191
speed=90m/min ................................................................................................................... 192
Figure 4.119: Effect of PET film / applicator roll speed ratio on liquid transfer ratio. ........ 192
Figure 4.120: Pictures of transfer regions at various wrap angle conditions. ..................... 194
Figure 4.121: Effect of wrap angle on meniscus positions. ................................................. 195
Figure 4.122: Effect of wrap angle on liquid transfer ratio.................................................. 195
Figure 4.123: Pictures of transfer regions at various web tension conditions. ................... 196
Figure 4.124: Effect of web tension on meniscus positions. ............................................... 197
Figure 4.125: Effect of web tension on liquid transfer ratio................................................ 197
Figure 4.126: Diagrams of transfer regions with low and high viscosity liquids.................. 199
Figure 5.1: Geometry of roll gap region............................................................................... 201
Figure 5.2: Effect of Applicator roller speed. ....................................................................... 205
Figure 5.3: Effect of the roll gap........................................................................................... 205
Figure 5.4: Effect of the supply film thickness. .................................................................... 206
Figure 5.5: Effect of the viscosity. ........................................................................................ 206
Figure 5.6: Effect of the surface tension. ............................................................................. 207
Figure 5.7: Predicted Pressure Distribution. ........................................................................ 208
Figure 5.8: Velocity distribution in (a) forward and (b) reverse modes .............................. 209
Figure 5.9: Geometry at the roll gap region ........................................................................ 210
Figure 5.10: Critical roll speed ratio from Lubrication model .............................................. 210
xii
List of Tables
Table 3.1: Physical properties data for liquids used (20 C) .................................................... 90
Table 3.2: Effect of temperature on the viscosity of liquids used (20 C, 25 C) ...................... 91
Table 3.3: Physical properties data for liquids used (25 C) .................................................... 91
Table 3.4: Effect of temperature on the viscosity of liquids used (20 C, 25 C) ...................... 92
Table 4.1: Stable conditions in deformable roll coating ...................................................... 163
Table 6.1: Effect of various parameters on the film thickness and coating window in rigid
roll coating ........................................................................................................................... 217
Table 6.2: Effect of viscosity and roll gap on the film thickness and coating window in rigid
roll coating ........................................................................................................................... 218
Table 6.3: Effect of various parameters on the film thickness and coating window in
deformable roll coating........................................................................................................ 218
Table 6.4: Effect of various parameters on wet stability and transfer ratio at transfer region
in reverse kiss coating .......................................................................................................... 221
xiii
NOMENCLATURE
b Thickness of rubber layer
Ca Capillary number Ca= V/
CaA Capillary number (Applicator roller)
CaM Capillary number (Metering roller)
E Young modulus
Es Elasticity number Es= V/ER
g Gravitational acceleration
h Roll gap (at any positions)
H Dimensionless roll gap (at any positions) H=h/h0
h0 Roll gap (at roll centre)
H0 Dimensionless roll gap (at roll centre) H0=h0/2R
ha Film thickness (applicator roller)
hB Film thickness (bottom roller)
HB Dimensionless film thickness (bottom roller)HB=hB/h0
hf Final film thickness
hi0 Initial set roll gap (at roll centre)
hi Initial set roll gap (at any positions)
hm Metered film thickness
HM Dimensionless metered film thickness HM= hm/ h0
HM, min Dimensionless minimum metered film thickness
hS Roll gap (stagnation point)
hT Film thickness (top roller)
hW Film thickness (withdrawal)
h* Characteristic film thickness h *( V / g )1 2
xiv
h¥ Final film thickness
NCa Modified capillary number NCa =Ca (R/H0)1/2
N Modified capillary number N= h ¤R /Ca
p Pressure
P Dimensionless pressure P=p h /R
Q Flow rate
qB Flow rate (bottom roller)
qL Leakage flow rate
qT Flow rate (top roller)
R Roll radius
RB Roll radius (bottom roller)
Rd Roll radius (deformable roller)
Re Equivalent roll radius
Rr Roll radius (rigid roller)
RT Roll radius (top roller)
rc Radius of curvature of meniscus
Re Reynolds number Re= V h/
ReA Reynolds number (applicator roller)
rr Ratio of roll radius (RT/RB)
rS Roll speed ratio (VM/VA or uT/uB)
rS* Critical roll speed ratio
T Dimensionless film thickness T=hm/h0
TW Dimensionless film thickness (withdrawal) TW =hW/h*
U Dimensionless velocity U=u/uB
uB Roll speed (bottom)
uT Roll speed (top)
V Velocity
xv
V* Maximum velocity of wetting
Vae Velocity (air entrainment)
VA Velocity (applicator roller)
VM Velocity (metering roller)
W Load
We Weber number We= V2h/
X Dimensionless x position X = x/ Rh
XD Dynamic wetting line position
Xs Dimensionless stagnation position
yB Position of the wall of the bottom roller
yT Position of the wall of the top roller
Y Dimensionless y position Y=y/h0
Inclination of web to liquid pan
Viscosity
 Density
Surface tension
q Dynamic contact angle
Inclination of wetting line
l Dimension ess f ow rate
lL Dimension ess eakage f ow rate
b
Dimensionless rubber deflection  p
Ehio
xvi
Acknowledgements
I would like to thank everyone for supporting me in my efforts of making this PhD thesis
possible.
I am especially grateful to my supervisors Professor Hadj Benkreira, Dr Raj Patel and Dr
Katsuya Ito who have continually helped and guided me in the right direction. I would also
like to thank Mr David Steele for his technical expertise when most needed and TOYOBO for
the grant that made all this possible for me.
I would also like to thank to my family and friends who encouraged and supported me
throughout the study.
Thank you all very much.
xvii
Chapter 1 : INTRODUCTION
1.1 COATING: INDUSTRIAL APPLICATIONS & CHALLENGES
Coating or the deposition of a thin film on a surface is a technique used for
many years in a variety of applications. Make-up and painting are probably the
original coating techniques which over the years have developed into more
sophisticated applications. Now coating is used not only for decorative or protective
ends but also to impart functionality to a surface and a material. For example food
packaging films whilst transparent are coated with a gas barrier layer to limit the
exposure of food to oxidation and moisture. Audio and video films are coated with a
magnetic layer to record and play sound and images and the porous surface of paper
is filled with a solution of mineral particles to enable clear printing. Applications are
many and are continuously being developed in this modern world, particularly in the
electronic, healthcare and energy capture industries. For instance optical films are
used for liquid crystal display (LCD), plasma display panel (PDP) and mobile
phones; drug release patches are now substituting to drugs which are injected or
taken orally and solar cells are coated with semi-conducting layers to capture the
energy of the sun and transform it into electricity. Coating as an industry is truly a
very big, multi-billion dollars global business, employing many people all over the
world. Inevitably, it is competitive and as a result attracts a lot of research attention.
Clearly, as a coating operation is simply the deposition of a thin film at speed, the
competitive requirements and challenges are thinner films, faster speeds whilst
maintaining uniformity of the surface. This thesis is concerned precisely with these
aspects in one, particularly widely used, coating flow: reverse roll coating.
1
1.2 COATING FLOWS: DEFINITION, CLASSIFICATIONS & INSTABILITIES
Most industrial coating processes involved fluid flows forming thin films
onto moving surfaces called substrates or webs. The liquid films are usually
formulated from a mix of liquid, resin, particle, surfactant and other additives to
exhibit specific physical, chemical and optical properties in order to meet the
requirements of the applications. After deposition on the substrate the liquid films are
dried or cured to form solid layers. The liquids used in the coating industry are
solutions based on water or organic solvents. In order to meet current demand for
ecological products and processes, many attempts have been made to reduce the use
of organically based solutions. Therefore many types of liquids with various
properties (from low viscosity to high viscosity and from low surface tension to high
surface tension) have been used recently.
Although there are many different ways of depositing a film continuously onto
a moving substrate, coating flows can be classified mainly into four basic categories
(Benkreira, 1994). Any one or a combination of any of these may be used. The
choice of selection strongly depends on the application of the finished product and
coating liquid properties. The following are the basic categories into which different
coating flows may be classified:
1) Withdrawal of liquid from a pool by a moving substrate, i.e. free coating
flow (Figure 1.1).
2) Metering an excess amount of liquid in flow geometry to form a film onto
a moving substrate i.e. metering coating flow (Figure 1.2). Reverse roll
coating, the subject of this study, falls into this category. As shown in
Figure 1.2b, it is simply the flow between two rollers, separated by a gap
and rotating in opposite directions at the nip. The film of interest, labelled
2
h2 in Figure 1.2b is the one taken by the applicator roller at the nip exit. In
actual practice, it is transferred to a web driven by (wrapped around) a
third rubber roller in kiss contact with the applicator roller (see Figure
1.2d).
). Clearly, the film thickness h2 will
ill depend critically on the gap. The
lower the gap, the lower h2 will be and this leads to considering the
situation where the gap is negative, i.e. the rollers in deformable contact
with one roller being rigid and one roller covered with a rubber sleeve
sleeve. As
we are concerned in this research with assessing the feasibility of
producing very thin films, we must consider this condition. Interestingly
no prior work has been done in deformable reverse roll coating. The
research we propose is thus novel.
3) Delivering
livering the exact amount of liquid in flow geometry to form a film and
then transferring it onto a moving substrate, i.e. pre-metered

metered coating flow
(Figure 1.3)
1.3).
4) Allowing a moving substrate to wipe a proportion of a coating trapped in
the cells of a printed or gravure roller, i.e. print or gravure coating flow
(Figure 1.4)
1.4).
Figure 1.1: Free coating
(a) a rotating cylinder and (b) an inclined plane
3
(d)
Figure 1.2: Metering

ing coating
(a) blade coating, (b) reverse roll coating (c) forward roll coating (d) reverse roll
coating with a third rubber roller
4
Figure 1.3: Pre-metered
metered coating
(a) kiss coating, (b) slide coating, (c) slot coating, (d) die coating
Figure 1.4: Gravure or print coating
All coating flows are designed to produce a thin, uniform and air free film.
Inevitably as the speed is increased and or the thickness is decreased, the flows
delivering these films become unstable and give way to:
5
a) Surface instabilities in the form of ribbing on the free surface forming the
film (see Figure 1.5(a))

1.5(a)), and
b) Dynamic wetting line failure which leads to air entrainment in the film
(see Figure 1.5(b))

1.5(b)).
These two instabilities are the most serious limitations

ions to coating flows, which
in practical operations should be overcome

me to produce films of the required
requi
specifications. We now explain the mechanism and detailed studies connected to
these instabilities are further cover

covered in Chapter 2.
(a) Ribbing
(b) Air entrainment
Figure 1.5: Coating instabilities (a) Ribbing, (b) Air entrainment
6
a) Ribbing instability
Free surface ribbing is one of the most prominent three-dimensional coating
flow instability. It often arises in thin liquid layers applied by brushing, spreading or
rolling. Ribbing is also quite common in various pre-metered bead-coating devices. It
is triggered by hydrodynamic instabilities when the flow from a coating nip splits to
form a meniscus which leads to the final formed on the substrate. Ribbing is an
important area of study in thin film coating and has attracted many investigations.
According to previous researchers, in roll coating, the ribbing instability is mainly
driven by the pressure gradient generated by viscous flow. This thesis re-examines
this aspect, critically in reverse roll coating.
b) Air Entrainment
Coating processes involve high-speed contact between a liquid and a solid
substrate in the presence of a gas phase (usually the surrounding air) at the
solid/liquid/gas interface or the wetting line. Such processes are speed restricted
because above a certain speed, the wetting line which is initially straight breaks and
allows air bubbles to be entrained in the film. This causes defects in the final product
and any measure to postpone the speed of onset of air entrainment to higher values is
highly desirable in industry. This aspect is also considered in this thesis with regard
to reverse roll coating.
As described at the introductory part, in these days the demands of customers
have become higher and higher, which are for high qualities and high productivity,
i.e. non-defect, uniform and thin coating at high speed. Therefore the understanding
of coating mechanism as a scientific and technological discipline of thin and uniform
7
film at high speed is significantly needed by many coating companies all over the
world.
1.3 AIM AND OBJECTIVES OF RESEARCH
A great deal of research has been invested theoretically and experimentally
with the coating methods explained in the previous pages. In particular reverse roll
coating has been used for producing thin films at high speeds because this coating
system has been thought as one of the most versatile, cheap and inherently stable
operation. Although many researchers have investigated reverse roll coating, there is
no comprehensive data how very thin film coating (less than 10 m) can be achieved
at high speeds (more than 100m/min).
The aim of this research is thus:
To study the fluid mechanics of reverse roll coating with both rigid and
deformable rollers and its feasibility to produce very thin (less than 10 m) stable
films at high speeds (more than 100m/min). As explained earlier, these are the
requirements giving the competitive edge in actual applications. They are also
scientific challenges. As just explained, with increasing speeds, coating flows are
limited by ribbing and air entrainment instabilities.
To achieve this aim, the following objectives have been set for this research
1. Review critically prior work of relevance to reverse roll coating with regard to
film thickness, ribbing instabilities and air entrainment.
8
2. Design and build a prototype reverse roll coating rig that simulates the operation
with (a) two rigid rollers operating in a positive gap and (b) one rigid roller and
one deformable roller operating in a deformable gap that may be negative;
3. Instrument the coating rig and use transparent rollers so that roller speeds and
gap can be accurately controlled, the flow in the nip visualised and the film
thickness and stability of the film formed measured;
4. Carry out flow experiments (i.e. measure film thickness and onset of
instabilities) in a range of conditions of roller speeds, speeds ratios and gaps and
a controlled flow inlet film thickness with fluids of different viscosities and
surface tensions;
5. Draw from the above experiments, stable coating windows that may be used in
practice to achieve desired film thickness at prescribed speeds of operation with
fluids of set properties, the aim being to push towards higher speeds and thinner
films and to confirm limit of operation of reverse roll coating;
6. Underpin theoretically the measured flow properties at these higher speeds with
lower and/or negative gaps by developing flow models able to capture these
extreme features.
9
Chapter 2 : LITERATURE REVIEW
2.1 INTRODUCTION
Reviewing the various coating flows and methods to produce films of various
thicknesses at various speeds is of great interest not only to form an overview of the
subject but also to identify areas which require further research. This task however is
very big indeed and will require reviewing everything written on the subject. This is
not intended here. Rather, the focus of this research is in the production of very thin
films on the order of 10 m at speeds higher than 100m/min. In this chapter, a variety
of coating systems are reviewed briefly especially from the point of view of how
very thin film can be achieved at high speeds. Secondly, roll coating which is the
subject of this research is reviewed comprehensively.
2.2 BRIEF REVIEW OF VARIOUS COATING METHODS
As discussed in the previous chapter, we can classify all coating methods into
four main categories, namely free coating, metering coating, transfer coating and
gravure or print coating. In this section, three types of coating systems except for roll
coating (reviewed in the section 2.3 below) are briefly reviewed from the point of
view on how very thin film can be achieved at high speeds.
2.2.1 Free coating
Dip coating has been studied since an earlier stage as a free coating system.
(see Landau and Levich (1942)). As its name implies, it considers the extent of the
film thickness picked up from a pool of liquid by a surface emerging out of the liquid
at speed V. This surface may be a moving plane or a rotating cylinder, as illustrated
10
in Figure 1.1 (a) (b). Essentially the rotating roller is an inclined plane in curvilinear
co-ordinates and the case of angular withdrawal represents a general free coating
flow. No complete solution of a flow problem of this kind is available.
Approximations of the film thickness, hw, developed far upstream are obtained
depending on the conditions - which can be represented by the capillary (Ca= V/)
and Reynolds (Re =  V hw/) numbers. These dimensionless numbers define the
relative contribution of the viscous, surface tension and inertia forces with , ,
defined as density, viscosity and surface tension respectively. Since in most coating
flows, viscous and surface tension forces always play a part, we note immediately the
fundamental role of the capillary number. Previous approximate analyses (see for
example the work of Landau and Levich, 1942 and Spiers et al., 1974) supported by
experimental data for low Ca and negligible Re show that the film thickness hW
relate with operating condition as:
Tw hw / h *0.944Ca1 / 6 (1  cos ) 1 2 (2.1)
Here dimensionless numbers are used with h *( V / g )1 2 as a
characteristic film thickness (an order of magnitude) obtained by balancing viscous
and gravitational forces. In this equation g is the gravitational constant and , the
angle of withdrawal.
This equation, verified by other workers, implies that it is difficult to obtain a thin
film at high speed even with low viscosity liquids. As an example, we may consider
the following coating liquid properties ( =0.005Pa.s, =0.03N/m, =1g/cm3), and
find that at speed=150m/min and =90 (vertical withdrawal), the film thickness hw
calculated is 921 m. This calculation clearly indicates that free coating flows
produce relatively thick films as expected as there is no metering obstacle to the flow.
11
Figure 2.1: Free coating to a vertical moving surface ((Spiers
Spiers et al., 1974)
One interesting feature of dip coating is very well suited to the study of air
entrainment. In this situation, the substrate is plunged at speed into the fluid and the
critical conditions leading to air entrainment are measured. Clearly such flow
situation is easy to reproduce experimentally and this explains why so much work
has been done on air entrainment using dip coating. Interestingly, when we observ
observe
all coating flows, the dynamic wetting region is always similar, showing a solid
surface entering a liquid phase. The difference between air entrainment in dip
coating and other coating flows is the condition of the liquid near the dynamic
wetting line. If, for example, the liquid is impinging on the surface as in curtain
coating, there will be what is referred to as hydrodynamic assistance which acts
against air entrainment. Therefore studying air entrainment in dip coating is a
simplification that does

oes not take into consideration the dynamics in other coating
flows. Further review of air entrainment is presented in section 2.6.1 below.
12
2.2.2 Metering coating
Metering coating flows can be viewed as a continuation of free coating flows
but with a fixed or moving boundary put in place to reduce the film thickness formed.
The simplest metering coating flow is blade coating where the metering boundary is
fixed. In roll coating, the boundaries (rollers) move. These boundaries can be rigid
(roll coating with a positive gap between rigid rollers) or deformable (flexible blade
coating or deformable roll coating). The flow in all these situations is in a narrow
gap where pressure develops. Downstream of the nip a separation region is formed
leading to the film(s) of interest. Upstream of the nip, there is the feed region.
Metering coating is very popular in industry because it is simple to engineer and the
size of the final film formed is of order about 50% of the minimum gap if we
consider flow between a moving surface and a plate placed parallel to it at a gap.
The actual film thickness formed in roller coating, blade coating, etc. can be difficult
to predict precisely because in these real situations, the flow is pressured and drag
driven in the nip and a free surface form at the exit. Further details are given in
Section 2.3 with reference to roll coating as this is the subject of this research.
2.2.3 Pre-metered coating
In this category of coating flows, a uniform film flow per unit width, q, is
transferred to a moving substrate using the three main types of pre-metered coating
systems: slot die coating, slide coating and curtain coating.
Slot die coating is known as the method for the production of thin films by
using low to high viscosity liquids. Many researchers studied the minimum wet film
thickness at a given web speed, at which the wet film is stable.
13
Lee et al. (1992) studied the m
minimum
inimum wet thickness experimentally in slot
die coating with liquids ranging from 10 to 1000mPa.s. They found that there is a
critical Ca (around 0.2

0.2-1.0
1.0 depending on coating gap from their figure 11)
11 below
which the minimum wet thickness depends on the co

coating
ating speed and the fluid
viscosity. Above the minimum limit, the film thickness is independent of Ca and is a
function only of the coating gap. The dimensionless (with respect to the gap)
minimum wet thickness is around 0.6

0.6-0.7 for Ca > Ca critical. The range
r of the coating
gap in the experiments was 0.2 - 1.0mm, therefore the minimum thickness was about
120 m.
Figure 2.2: Slot die coating onto a roller (Lee et al., 1992)
Carvalho and Kheshgi (2000) discovered a new region in which the minimum
wet film thickness decreases as Ca increases (Ca > around 0.3-3.0

0.3 depending on
coating gap).. Their result indicates that it is possible to get a 12.7 microns film using
a 22mPa.s liquid at the coating gap of 100 microns at 70m/min. This kind of region
is different from the previous research by Lee et al. (1992). The difference between
the study of Lee et al. and that of Carvalho and Kheshgi is the maximum web speed
in their experiments. The maximum web speed in the study of Lee et al. is 66m/min,
whereas that of Carvalho and Kheshgi is 107m/min.
14
Chang et al. (2007) studied the minimum wet thickness using very low
viscosity liquids from 1.2 to 50mPa.s. They found that there are critical Re numbers
and three regions in which minimum stable film can be achieved. In the first region,
the minimum wet thickness increases as Ca increases. In the second region, it
becomes independent of Ca. These trends are the same as those found by Lee et al.
(1992). In the third region above the critical Re (=20), the wet thickness decreases as
Re increases. This trend is the same as that found by Carvalho and Kheshgi (2000).
In their experiments, the minimum wet thickness is about 12 microns using a
1.2mPa.s liquid at the coating gap of 200 microns and speed of 72m/min (Re =200).
Chin et al. (2010) found that when the die angle to the substrate and the
geometries of die lips are set appropriately, wider coating windows can be achieved
i.e. higher speeds and thinner films. For instance in their experiments they achieved
higher coating velocity (10m/min) with optimised die lip condition and inclining
angle compared to an original condition in which the maximum coating velocity was
3m/min. However in their experiments, the maximum substrate speed was limited at
20m/min, therefore the conditions to achieve thin film at higher speeds remain
unknown. Romero and Carvalho (2008) found that the die geometry may be
optimised in order to minimise the film thickness oscillation. According to these
literatures, to optimise the die lip geometry is very crucial to decrease the wet
thickness oscillation as well as the minimum wet thickness. Optimisation usually
involve getting the correct die length at entry and exit as well as the angle of
inclination if the die is not a straight channel. The idea being is that for oscillations
not to occur the flow must remain pinned in the entire geometry of the die and not
change separation point positions.
15
As Chang et al. (2007) indicated (see above),, the minimum wet thickness can
be decreased if the gap is decreased. H

However there is a limit to how small the gap
can be made because of the potential of clash between the roll and the die. Also, as
the gap is reduced, the accuracy of the rotating roll

roll-die
die gap diminishes and the
operating gap will fluctuate around a value causing fluctuation in the film thickness
formed. To overcome this limitation, the roller was removed and tensioned-web slot
die
ie coating was developed in the early 1980s. Lin et al. (2007) studied
experimentally the tensioned

tensioned-web
web slot coating with PVA solutions and found the
minimum wet thickness to be between 0.5 and 1.5 microns, which is much thinner
than that in the convention

conventional
al slot coating. However the range of the speed is from 0.5
to 10.0m/min, which is slow from the practical point of view. Lin et al. (2008)
studied theoretically the tensioned

tensioned-web
web slot coating and found that their experimental
data are in good agreement wi

with their theoretical results.
Figure 2.3: Tensioned

Tensioned-web slot die coating onto a substrate (Lin et al., 2007)
Nam and Calvalho (2010) studied theoretically the effect of die lip design on
the minimum wet thickness in the tensioned

tensioned-web
web slot coating and found the
geometry of downstream die lip is an important factor to reduce the minimum wet
16
thickness. These effects are to be expected but they do not radically change the film
thickness or increase the speed.
As another attempt to reduce the minimum wet thickness at high speed, non-
Newtonian effects have been studied. Ning et al. (1996) examined the effect of
polymer additives (Carboxy-methyl-cellulose and poly-acrylamide powders) on the
slot die coating. They found that there is an optimal concentration for each polymer
additive, at which the maximum coating speed can be increased. However the
minimum thickness was more than 100 microns in their experiments.
Romero et al. (2004) studied the low flow limit in slot coating using low
molecular weight polyethylene glycol and high molecular weight polyethylene oxide
in water. From their experimental results, as extensional viscosity increased, the
minimum wet thickness increased, i.e. there was no effect of adding high molecular
polymer on reducing the minimum wet thickness. However, their theoretical analysis
(Romero et al., 2006) showed that at low Weissenberg number (less flexible polymer
molecules), the onset of the low flow limit occurs at lower flow rate than with the
Newtonian liquids and above a certain value, the viscoelastic stresses destabilises the
process.
Chu et al. (2010) measured the effect of the molecular weight and
concentration of polyvinyl alcohol (PVA) in slot die coating. They found that the
coating window expanded with increasing PVA concentrations and molecular weight
at low concentration regions. However beyond a critical point, the effect of fluid
elasticity became dominant, and the coating window began to shrink. The range of
viscosity was from 39 to 860mPa.s, and the minimum wet thickness was more than
100 microns at the maximum speed (84m/min).
17
As another way of making a coating window wide, adding inorganic particle
in solutions was reported. Chu et al. (2006) for example measured the effect of
adding TiO2 and SiO2 in aqueous polyvinyl alcohol (PVA) solutions in slot coating.
They found that the strong interaction between polymer and particles resulted in
higher surface tension, and this made the coating window wider.. In their experiments,
the range of the liquid viscosity was from 11.5 to 26.7mPa.s and the minim
minimum wet
thickness was more than 100 microns at the maximum speed (54m/min
54m/min). Therefore,
as this review describes, there is no data showing that very thin films,
film less than 10
microns can be made at high speeds,, more than 100m/min using slot coating either in
its standard form or in the tensioned

tensioned-web mode.
We now examine slide coating, a similar technique to slot coating but where
the liquid does not pass through a narrow gap before ending up on the substrate. The
transfer slide to substrate is direct.
Slide coating
oating system has been widely used for multi
multi-layer
layer coating,
coating especially
in the photographic industry where the production of colour films requires the
superposition of many (up to twenty) layers

layers. The study of Hens and Abbenyen
(1997) concluded that in slide coating the minimum wet thickness is usually around
20 m and the speed no more than 100m/min, limited by the onset of air entrainment.
Figure 2.4: Slide coating system (Hens and Abbenyen, 1997)
18
Prior to Hens and Abbenyen (1997), Gutoff and Kendrick (1987) had
measured the low flow limits of coatability on a slide coating and established that the
maximum velocities are identical to the dip coating, increasing only slightly with the
application of vacuum, upstream of the bead. According to their data, the minimum
wet film thickness was not less than 100 m at more than 60m/min with liquids of
viscosity in the range of 21 to 1110mPa.s. The low flow limit is clearly controlled by
air entrainment as to achieve a thin film, the speed has to be increased up to a critical
condition and this condition is when the air begins to be entrained. Gutoff and
Kendrick (1982) had studied earlier air entrainment in dip coating and found that the
air entrainment velocity depended essentially only on viscosity as:
Vae  5.11  0.67 (2.2)
In this equation, Vae is in units of m/s and the viscosity is expressed in mPa.s.
Many more researchers have examined air entrainment in order to understand
the mechanism of this important phenomenon (reviewed in greater details in Section
2.6.1). Three studies are however critical. The first is the work of Deryagin and
Levi (1964) who first observed that at Vae, the dynamic wetting line turns from
straight into an appearance of sawteeth with air entrained at vertices of the “vvvv”
formed (see Figure 2.39). The second study is due to Blake and Ruschak (1979) who
observed the geometry of the “vvvv” at speeds higher than Vae and established,
experimentally, that the component of the speed normal to the straight-line segments
of the “vvvv” was independent of the substrate velocity. They termed this component,
the maximum speed of wetting V* or the maximum speed at which the wetting line
can advance normal to itself. They observed that the substrate could be wetted at
speeds V higher than V* only if the wetting line slanted so that the speed of the solid
19
normal to it did not exceed the maximum speed of wetting. They determined from
this that the wetting line segments angle adopted the minimum possible inclination
such that:
cos = V*/V (V >~ V*) (2.3)
This would explain the break-up of the wetting line into a sawteeth pattern and the
subsequent occurrence of air entrainment at the point where two straight-line
segments intersect. A corollary from this observation (not substantiated
experimentally by Blake and Ruschak (1979)) is that air entrainment could be
postponed to velocities Vae greater than V* provided that the substrate enter the pool
of liquid at an angle so that the wetting line is not perpendicular to the direction of
substrate motion as shown in depicted in Figure 2.40. This brings us to the third
work carried by Cohu and Benkreira (1998) who established this experimentally and
applied it to delay air entrainment in slide coating. In their experiments, the
minimum wet thickness was however more than 100 m, i.e. still very large
compared to our objective.
In conclusion, slide coating is limited by air entrainment much like dip
coating. The main interest in slide coating is its application to multilayer coating and
understanding how the interface between layers behave (instabilities) and how the
thickness of each layers changes (see for example the work of Lin et al., 2005 and
Noakes et al, 2002). According to these works, it is difficult to make very thin film in
the slide coating.
Finally, we now examine curtain coating, another important pre-metered
coating technique which capable of coating films of liquids on substrates at
extremely high speeds, more than 600m/min according to Blake et al. (2004). Blake
and Clarke (1994) explain that the reason for this is “hydrodynamic assistance” of
20
the curtain of liquid impinging from a height on the substrate thus preventing the air
from being entrained

entrained. Thus the onset of air entrainment can be postponed by
increasing the height of curtain, by adjusting the solution flow rate and angle of
impingement (Blake et al., 2004). Because of the nature of liquid curtains, there
t will
be a minimum flow rate below which the curtain is not stable and will tend to curve
at the sides as a result of surface tension. This minimum flow rate

rat is found to be
about 1.0cm2s-1. If we assume a high coating speed of 150m/min, a film thickness of
40 m will be obtained which is below our 10 m objective. Curtain

urtain coating is thus
unsuitable to produce very thin films.
Figure 2.5: Curtain coating system (Blake and Clarke, 1994)
2.2.4 Gravure or print coating
Gravure coating is an established method for the production of thin films
using low viscosity liquids. Gravure cells can have many different patterns and
geometries, but there are mainly three popular patterns: pyramidal, quadrangular and
tri-helical
helical (see Figure 2.6). In gravure roll coating these patterns are set on the rollers
21
which can be operated in the forward or reverse mode using direct or offset transfer
to the substrate using one or more rubber covered offset rolls as shown in Figure 2.7.
Figure 2.6: Schematic representation of the gravure geometries (Patel and Benkreira,
1991)
(a) Direct gravure roll coating (b) Off set gravure roll coating
Figure 2.7: Different modes of gravure roll coating (Patel and Benkreira, 1991)
22
The main interests of researchers of gravure coating are that how efficiently
doctor blades wipe clean the surplus of liquid from above the cells as the roller rotate
and how much liquid is picked out of the cells by a passing substrate in close contact
with the cells. These factors decide the final wet film thickness.
As for the doctor blade, Patel and Benkreira (1991) found experimentally that
the properties of the doctor blade (applied load, elasticity, positioning) determine the
amount of liquid remaining on the lands of the gravure roller and effective doctoring
is not easily achieved unless very low viscosity liquids are used. Hanumanthu (1999)
studied the variation of gravure coating thickness during the early stages of doctor
blade wear. He investigated how variables such as doctor setting angle, doctor blade
tip thickness affect coating thickness theoretically.
With regard to the amount of pick out, Pulkrabek and Munter (1983) first
studied forward gravure roll coating and found that the amount of pick out was
around 59% of the volume of the roll cells.
Benkreira and Cohu (1998b) found that the film thickness was of order 15 to
20% of the gravure cell depth for direct forward gravure coating and that stable films
can only be achieved in a narrow coating window at very low substrate Ca number
(less than 0.02) equivalent to substrate speeds typically less than 20m/min.
According to this result, it can be thought that the reverse gravure coating would be
better than forward gravure coating when it comes to at high speed coating.
Benkreira and Patel (1993) measured the wet film thickness for reverse direct
gravure coating under various operating conditions (coating speed, roller roughness,
substrate wrap angles, and liquid properties) and concluded that the wet film
thickness only depends on the shape of the gravure cell (constant value of about 1/3).
23
Kapur (2003) studied direct reverse gravure roll coating experimentally,
focusing on roll-to-web fluid transfer. His data showed that the pick out of fluid from
the gravure cells was significantly affected by the ratio of the web-to-roll speed, the
fluid properties and the shape and size of the gravure cells. As the speed ratio was
increased, the amount of pick out approached one and a coating instability occurred
due to starvation.
Hewson et al. (2006) investigated reverse gravure roll coating with tri-helical
gravure roller experimentally and theoretically. They measured the wet film
thickness, pick out and meniscus location as a function of roll speed ratio. The results
indicated that as the speed ratio was increased, the amount of pick out also increased,
although the actual film thickness decreased. They also found that the depth of the
gravure cell and Ca number affected the coating instability with shallower cells
causing loss of stability.
All the researchers named above studied the direct transfer from a gravure
roller to a substrate. Kapur et al. (2001) on the other hand investigated the indirect
offset mode roller varying the speed ratio offset roller/substrate and measuring the
critical ratio at which coating instabilities occurred. They found that as the applicator
roller speed was increased, the critical speed ratio decreased. The maximum speed of
the substrate they studied was 1.2m/s.
Now visualising the meniscus region is clearly useful to understand the
coating mechanism of fluid transfer in gravure coating although, the region is very
small therefore difficult to observe. Yin and Kumar (2006) carried out such
observations and established that the pick out is a function of Ca. Similarly, Chuang
et al. (2008) observed the transfer mechanism using scaled-up cells. They found that
the stability of the liquid bridge that connects the moving roller and the gravure cell
24
was the most critical factor to determine the pick out and the most stable condition
that yields the maximum pick out is that when the moving roller just touches the
gravure cell and the coating gap is zero.
Hewson et al. (2010) developed a mathematical model of a tri-helical gravure
coater and assessed the effect of grooves of different cross sectional shape and
aligned at different angles of pitch on the pick out of liquid from the cells. Their
results show that the amount of fluid pick out decreased with increasing groove depth
while the film thickness increased and an increase in groove aspect ratio caused a
reduction in both pick out and film thickness for a given land width and groove
cross-sectional shape.
The review of gravure coating implies that very thin films can be achieved
with this technique. However all the research published deals with pick out from
cells without stating specifically the minimum film thickness achieved and the speed
at which it is achieved. Clearly this is an area of research worthy of further
investigation.
2.3 ROLL COATING
Roll coating is an important and common coating process widely used in the
film coating and printing industries. This process uses two or more rollers (see
Figure 2.8) to create a “nip flow” between rotating rollers that serves to produce thin
films. The nip or gap between the rollers is usually less than 1mm and the films
produced are of order greater than 10 microns at speed typically less than 100m/min.
As the technology has developed over the years, the impetus now is to produce very
thin films of order less 10 microns at speeds in excess of 100m/min. Clearly
depending on the direction of roller rotation and the number of rollers used, roll
25
coating can appear to be very difficult to classify. Benkreira et al. (1981a), however,
offered a very simple classification by considering the primary coatin

coating flow created
by two rollers only. When both rollers move in the same direction, we call this type
forward roll coating (Figure 2.9(a)) and when the rollers move in opposite direction
we refer to this type as reverse roll coating (Figure 2.9(b)). In addition we

w can sub-
classify forward or reverse roll coating further depending on whether rigid or
deformable rollers are used and on the nature of the gap formed. When the two
rollers are rigid and the gap formed is positive, we can refer to this type of roll
coating flow as rigid roll coating flow ((Figure 2.9(c))) When the gap can be negative
as a result of one roller being deformable (as result of it being covered by a rubber
sleeve for example) we refer to this as deformable roll coating ((Figure

igure 2.9(d)). In all
configurations- rigid, deformable, forward or reverse, in practice only one film ends
up as the final film to be coated on the substrate, hence the substrate can be arranged
in a variety of configurations (see Figure 2.10).
Figure 2.8: Typical roll coating operation (Coyle, 1990a)

19
26
(a) Forward roll coating (b) Reverse roll coating
(c) Rigid roll coating (d) Deformable roll coating
Figure 2.9: Roll coating arrangements
27
Figure 2.10: Variety of substrate configurations
The main concern however is to obtain a film which is “absolutely” flat and
“integral”. By absolutely flat, we mean that the film thickness is uniform across the
width of the substrate and not undulating

ndulating in thickness. By integral, we mean that the
film does
oes not contain any air bubbles even if these are ve
very
ry fine. Ribbing instability
(undulating
ndulating variation in film thickness across the width of the substrate) and air
entrainment are the two main limits of a good coating operation. These as we shall
explain later are inherent to all coating flows and occur because of flow instabilities:
surface instabilities in the case of ribbing and dynamic wetting instability in the case
of air entrainment. Clearly from this introduction, three issues have concern
concerned
researchers. First, how does the film thickness vary depending on roller gaps, speeds
and coating fluid properties

properties? Secondly, when do ribbing instabilities occur and what
are the mechanisms behind their onset and how can these be prevented
prevented? Thirdly,
when does
es air entrainment occur, what are the mechanisms behind it and how can it
be postponed to higher speeds

speeds? In this chapter, we present a critical review of the
research
earch work carried out in these three areas in order to situate the purpose of this
new study which is to push the limits further so that very thin films can be coated
faster without ribbing instabilities or air entrainment arising. The review of these
28
three issues is organised under the heading of rigid roll coating and deformable roll
coating and subdivided further under forward and reverse roll coating.
2.4 FILM THICKNESS IN ROLL COATING FLOWS
2.4.1 Forward Rigid Roll Coating
Greener and Middleman (1979) studied the model of forward roll coating for
Newtonian liquids for the simple case of two rolls of equal rotational speeds and
equal diameters by utilizing the usual lubrication approximations. These assume that
the flow in the nip is Newtonian, steady and one dimensional and develops as a result
of the balance between pressure and viscous forces. The governing flow equations
reduce to the following simple equation:
= (2.4)
Using velocity boundary conditions at the wall of the rollers and the so-called
Prandtl-Hopkins separation conditions (Prandtl, 1937, Hopkins, 1957) and balancing
the pressure at exit with surface tension effects, (see Figure 2.11 summarising the
model), Greener and Middleman (1979) were able to predict the net flow rate
through the nip, Q or its dimensionless equivalent l as depending on a modified
capillary number NCa=Ca (R/H0)1/2. These results are presented in Figure 2.12 and
show that at the limit of negligible surface tension effect, the pressure at exit is
approximately zero and =4/3. This result applies only for the case of equal roller
speeds.
*Lubrication approximation equation:
29
=
*Velocity boundary condition:
u=uT at y=yT, u=uB at y=yB
yT = H + , yB = - (H + )
2H = yT - yB = 2H +
* Prandtl-Hopkins separation conditions:
The velocity and the velocity gradient are zero at a separation point (x , y ).
u= =0 at x= x
y = 0 because of symmetry
*Pressure boundary condition:
x ® -∞ P(x ) = P = 0
at x = x P(x ) = - r = (2H – 2H)
30
(a) Definition sketch of the nip region
(b) An approximate model of the film

film-splitting
splitting region
Figure 2.11: Lubrication model ((Greener and Middleman, 1979)
31
Figure 2.12: Coating thickness as a function of modified Ca number (Greener and
Middleman, 1979)
Clearly, the above model does not cater for the situation when the rollers
rotate at different speeds and some flow separation conditions are required. We may
still use the Pradtl-Hopkins

Hopkins conditions which merely state that at the separation point
both the velocity and velocity gradient are zero

zero. However, the pressure at exit cannot
be evaluated as we do not know the individual film thickness on the top and bottom
roller. To overcome this problem, Benkreira et al. (1981

(1981b, 1982a)) developed a semi-
semi
empirical model using a flux distribution measured experimentally. These
experiments (Benkreira et al
al., 1981a) showed that the flux ratio depends only on the
speed ratio and is given by:
.•‚
0.87 € (2.5)
q : Top flux, q : Bottom flux, u : Top roller speed, u : Bottom roller speed
The constant 0.87 is peculiar to the arrangement used by Benkreira et al

al. (1981a) and
accounts for the effect of gravity. Using this approach, Benkreira et al. (1981a)
obtained the following result for the total flow rate Q(=qT+qB) through the nip:
32
ƒ
Q=1.31
31ho (2.6)
This is in good agreement with Greener and Middleman (1979) prediction at equal
speeds.
Figure 2.13 describes the semi

semi-empirical
empirical model of Benkreira et al.
al (1981b)
and Figure 2.14 gives the results which the variation of the dimensionless flow rate l
=Q/Uh ) with both the modified capillary number (N=„h ¤R /Ca.)

Ca.), the speed ratio
( r = u /u ) and the radius ratio ( r = R /R ). As shown in Figure 2.14,

2. the
dimensionless flux is a weak function of the three dimensionless numbers which are
the modified capillary number, the speed ratio and the radius ratio in most practical
forward coating operation range in which N is less than 0.3. Th

These
ese model results are
in good agreement with their experimental data.
Figure 2.13: Semi-empirical

empirical model (Benkreira et al
al., 1981b)
33
Figure 2.14: Results of semi-empirical model (Benkreira et al.,, 1981b)
1981b
(a) …… =10, (b) …… =1.0,

1.0, (c) …… =0.1
34
Variations of the work carried out by Greener and Middleman (1979) and
Benkreira et al. (1981b, 1982a) have been published but none presenting a step
change in the knowledge until Coyle et al. (1986) treatment of the flow analysis
using finite elements. Such a new approach, possible by the advance of computer
power, paved the way to considering both the bulk flow but also the important free
surface that limits the flow and that is indeed the solution to the problem. Figures
2.15, 2.16 explain the main lines of this new model and the importance of the free
surface. The comprehensive numerical modelling work of Coyle et al. (1986) showed
that the simple lubrication model was useful only as an approximation at high
capillary numbers (modified Ca number is more than unity). The finite element
model on the other hand, gave solutions that provided full details of the steady two
dimensional free surface flows including complex recirculation patterns in the film
splitting region. All the work that followed Coyle et al. (1986) model are variant of
his initial model and do not provide drastic advances in our understanding of how the
flow and film thicknesses in forward roll coating develop. After this work, the
challenge was to study surface instabilities and consider the role of the dynamic
wetting line. Whereas as we shall see later, surface instabilities have been well
researched and to certain extent well understood, how to fix the dynamic wetting line
remains an enigma and a challenge. It is important to understand that mathematically
speaking, this line is critical to resolving the flow problem. Indeed one criticism of
Coyle et al. (1986) and all models to date are in their arbitrary treatment of the
dynamic line, often fixed to enable a solution of the entire problem to be obtained.
*No-traction condition:
n・T = 0
35
*The normal stress in the liquid must balance the Ca number
number:
n・T = - nPa
*Kinetic boundary
oundary condition
condition:
n・u = 0
pointing unit normal vector

n: outward-pointing
T: total
otal stress (sum of the pressure and viscous stress)
t: unit
nit tangent vector in the direction of increasing arc length s along the free surface
Pa: ambient
mbient gas pressure
Finite-element
element discretizations for flow in forward roll coating
Figure 2.15: Finite element analysis (Coyle et al., 1986)
36
Figure 2.16: Flow rate from the finite
inite element analysis (Coyle et al., 1986)
All the work so far described was concerned with large gap (>
(>100 microns)
and moderate speeds (less than 100m/min) and assumes the inlet flow to be fully
flooded (see Figure 22.17 (a)). Clearly, one way of controlling further forward roll
coating and reducing the net flow rate through is by manipulating the inlet flow.
Indeed, it is possible that the inlet film thickness be

become smaller than
tha the minimum
gap (meniscus coating as shown in Figure 22.17 (b)). In such a case, the arriving flow
just about bridges the nip causing full recirculation and as low net flow as could be
obtained. This is precisely the flow situation investigated by Malone (1992) who
stated (he did not give supporting evidence) that only small
all speeds can be tolerated
in meniscus coating. The result of film thickness in forward meniscus roll coating
indicates that the film thickness in meniscus roll coating was smaller than that in
classical roll coating (see Figure 2.

2.18).
37
(a) Fully flooded regime (b) Meniscus regime
Figure 2.17: Coating regimes
Figure 2.18: Comparison of inlet/outlet film thickness ratio versus speed ratio results
of meniscus
eniscus roll coating with classical roll coating predictions (Malone, 1992)
h0: Inlet film thickness, h1, h2: Outlet film thickness

thickness,, h0/H0=0.1/0.2,
h0/H0=0.1/0.2 H0: Roll gap
Gaskell et al. (1995, 1998a) have also investigated meniscus roll coating in
which inlet films are starved and flow rates are small. They showed that it was
possible to get ultra thin film, less than dimensionless film thickness 0.1
0.1, but only in
the low Ca number liquid which is equivalent to low speeds operation.
Clearly although meniscus coating is able to develop very thin films, there are
limits on the speed that can be achieved. It is thus interesting to follow the work of
Malone and Gaskell et al. and establish the limiting speed that can be attained with
coating
ting of set viscosities and surface tensions.
38
All the above describe
describes work carried out on fluids that are Newtonian. In
practice, coating fluids by their very nature (mixture of polymers or very fine solids
with solvents) are non

on-Newtonian and theses effects
cts may be important on the net
flow.
Benkreira et al. (1981a) studied shear thinning and inelastic fluids as well as
Newtonian fluids.. They found that shear thinn

thinning
ing and inelastic fluids result in an
increase in the dimensionless flux value. As for the flux distribution, there are no
major differences between Newtonian and non Newtonian fluids.
Q=1.43
1.43ho (2.7)
.
0.99 (2.8)
Han et al. (2009) examined the effect of vicsoelasticity on the film thickness
in forward roll coating experimentally. The results showed that the metered film
thickness decreased in the case of viscoelastic liquids (see Figure 2.19).

2.1
Figure 2.19: Effect of viscoelasticity on the film thicknesses (Han

Han et al., 2009)
N: Newtonian, P1: 200ppm PAAm, P2 500ppm PAAm
(PAAm : Polyacrylamide)
39
In conclusion, forward roll coating, unless operated at very low gaps (not an
attractive proposition in practice because of rollers clash) or in the meniscus coating
mode is unable to produce very thin films. Even, in meniscus coating mode, which
does lead to very thin films, the operation is limited by the coating speed. Clearly
another strategy is required to reduce the net flow (the flow that forms the film
eventually). One simple solution is to meter the flow further, not by reducing the gap
or the arriving inlet flow but simply by changing the rotation of the metering roller
and reversing it rather than operating it in the forward mode. This takes us to
consider the next flow which is reverse roll coating.
2.4.2 Reverse Rigid Roll Coating
Following the same approach as that used for forward roll coating, Ho and
Holland (1978), Benkreira et al. (1982c) and Greener and Middleman (1981) applied
the lubrication approximation to predict the net flow hence the coating thickness in
reverse roll coating. The result is the same if we take the sign of the speed into
consideration and the net flow is again given by:
Q=0.63ho (2.9)
Following this, Greener and Middleman (1981) proposed the system depicted
in Figure 2.20 as a simple model of reverse roll coating. In this the applicator and
metering rollers are arranged horizontally and are half-submerged in a coating liquid.
40
Figure 2.20: Simple model of reverse roll coating system (Greener
Greener, 1981)
In such a system, there

here is a high shear flow in the narrow gap between the
rollers, however there

here is no dynamic wetting line to interact with this flow and no
instabilities are observed. By using lubrication theory with two different boundary
conditions, Greener and Middleman (1981) indicated the thickness can be written as
follows:
T = 0.650 1 (Prandtl-Hopkins condition) (2.10)
T = 0.613 1 (Swift-Stieber (Reynolds) condition) (2.11)
At low speed ratio (Vm/Va <1),

1), they observed strong recirculation in the flow
and their measured flow rate through the gap deviated strongly from predictions of a
model based on lubrication theory (see Figure 2.21).
Coyle (1990bb) have examined this model configuration experimentally and
by finite element solutions of the Navier

Navier-Stokes
Stokes equations in two dimensions. The
result indicated that the film transfer free surface and the recirculation under it did
not significantly influence the flow rate through the gap. Deviations from lubrication
theory occur only under conditions of low speed ratio and large gap when the effect
41
of gravity becomes appreciable. Under such conditions, he found that there is a large
component of the flow alon

along the roller axes as fluid flows off each end of the roller.
Figure 2.21: Measured flow rate as a function of speed ratio compared to predictions
(Coyle, 1990b)
The data of Benkreira et al. (1981a, 1982c) obtained with the bottom applicator roller
feeding the metering roller (see Figure 22.22)) give the dimensionless film thickness
as:
T = 0.630 1 (2.12)
This result is very similar to that obtained by Ho and Holland (1978) who obtained a
slightly different expression:
T = 0.667 1 (2.13)
42
Figure 2.22: Reverse roll coating arrangement (Benkreira et al
al.,, 1981a)
In order to gain a better understanding of the 2D effect, Coyle (1990a)

(1990 have
examined this model configuration experimentally and by finite element solutions of
the Navier-Stokes
Stokes equations in two dimensions. They concluded that the lubrication
model fails to describe the flow adequately at high roll speed ratios and capillary
number. A comparison of their results and those obtained from the lubrication model
is presented in Figure 2.
2.23 which shows that at high speed ratios and capillary
numbers, the metered

ed film flow deviated strongly from predictions of lubrication
theory. Finite
inite element m
model reveals that as the speed ratio is increased, the wetting
line moves through the gap centre

centre, the metered film thickness passes through a
minimum and then increases again with further increase in the speed ratio.
ratio At this
critical speed ratio, the dynamic contact line is exactly located at the centre of the
gap between the two roll

rollers. Beyond this critical speed ratio, cascade happens (later
discussed), which causes th

the metered film to thicken.
43
Figure 2.23: Metered film thickness as a function of speed ratio and Capillary
number (Coyle,, 1990a
1990a)
Kang and Liu (1991a) data provided further confirmation of the existence of
a minimum film thickness first observed by Coyle et al.(1990a). After examining the
effect of viscous and inertial forces, surface tension and gravity on coating flow, they
found that the minimum coating thickness depended only on the capillary number
and could be expressed by:
.
= 0.29Ca (2.14)
In an effort to obtain very thin films, Coyle (1990a) observed that at speed
ratio greater than 1 and with an arriving film on the applicator roller thinner than the
minimum gap clearance (meniscus coating),, dimensionless film thickness less than
0.05 could be achieved (see Figure 2.24).

). His data indicated that stable operation is
possible at speed ratios of 1 to 2 when the Ca number is on the order of 0.005 or less.
The range of viscosity studied was from 100 to 2000mPa.s and the range of roll gap
from 125 to 1000 m. Therefore although this set-up

up is able to achieve very thin films,
44
it can only do so at very low speeds and with relatively more viscous fluids than
those of interest to prese

present day technology (typically water like, less than 20mPa.s).
2
Figure 2.24: Dependence of measured metered film thickness on the arriving film
thickness (Coyle,, 1990a
1990a)
Malone (1992) also investigated meniscus roll coating in the reverse mode.
He stated that meniscus coating, although allowing only limited coating speeds, can
provide far thinner, more uniform films than the classical roll coating process.
However he did not provide data to support this claim.
Hao and Haber (1999) examined reverse roll coating, using finite element
method pointing out that tthe arriving film thickness strongly affects the metered film
thickness and the dynamic contact line position if the speed ratio is beyond the
critical value. This is particularly true for the case of arriving film thickness ratios
smaller than unity (see Figure 2.

2.25 (a)).. Thinner arriving films always cause the
upstream side free surface to be drawn to the minimum gap and the metered film to
become thicker at high speed ratio

ratio. Hao and Haber (1999) also found that the
t
principal roll diameter had a comparatively important effect in reverse roll coating at
45
high speed ratios. For speed ratios beyond the critical point, a far thinner metered
film can be achieved with smaller principal roll radius (see Figure 2.25(b)).
2. However,
within a practical parameter value range (where the minimum film thickness can be
achieved), both the arriving film thickness and roll radius had no significant effects
on the metered film thickness

thickness.
(a) Effect of arriving film thickness H /H (R/H =1000, R / R =1)
R/H (H / H =1.25, R / R =1)

(b) Effect of principal roll radius R/
(c) Effect of radius ratio R / R (H / H =1.25, R/H

H =1000)
Figure 2.25: Metered film thickness as function of arriving film thickness, roll
radius and roll radius ratio (Hao and Haber, 1999)
46
Thompson et al. (2001) investigated reservoir-fed, rigid reverse roll coating
and found that the gravity for the liquid in the reservoir affected the meniscus
position and the metered film thickness (see Figure2.26).
Figure 2.26: Effect of reservoir height for (a) flow rate, (b) wetting line position, as
a function of speed ratio (Thompson et al., 2001)
As for the effect of non

non-Newtonian
Newtonian behaviour, only limited work has been
carried out. Coyle et al. (1990d) found that purely viscous shear thinning liquids
behave in much the same manne

manner as Newtonian
ewtonian liquids at the same Ca,
Ca defined with
an apparent viscosity at a characteristic shear rate of the process, d[-s]=(Va+Vm)/h0.
In the case of viscoeelastic liquids, the film thickness is decreased and is much less
sensitive to variations in speed ra

ratio (see Figure 2.27).
47
(a) shear-thinning
thinning solutions (b) viscoelastic solutions
Figure 2.27: Measured metered film thickness of (a) shear

shear-thinning
thinning solutions and (b)
viscoelastic solutions (Coyle et al., 1990d)
Han et al. (2009) examined the effect of vicsoelasticity on the film thickness
in reverse roll coating, the same as forward roll coat

coating
ing experimentally. The results
showed that the metered film thickness decreased in the case of viscoelastic liquids
(see Figure 2.28).
Figure 2.28: Effect of viscoelasticity on the film thickness (Han et al., 2009)
N: Newtonian, P1: 200ppm PAAm, P2 500ppm PAAm
48
2.4.3 Forward Deformable Roll Coating
As explained above, the main purpose of using deformable roll coating is to
avoid the risk of clashing two hard rolls when the gap is very small or negative. As
low gaps are naturally required to obtain very thin films, deformable roll coating is
normally used for such applications. In all cases, deformable roll coating is operated
in the forward mode and usually with the rollers rotating at equal speeds to avoid
tearing the rubber sleeve through long exposure to shear. As a result all the studies
reported to date are concerned with forward deformable roll coating. One important
such study was carried out by Coyle (1988) who developed a simple one-
dimensional elastrohydrodynamic model of the flow. Essentially the flow equation
is the same as for rigid roll coating but with an unknown gap that deforms. If the
deformation of the roll is linear with pressure, i.e.:
d = P (L/E) / H0 (2.15)
d: total roll deflection, P: pressure, L: thickness of deformable roller, E: Young’s
modulus, H0: minimum roll half-gap
Then as Coyle showed, the film thickness becomes a function of the Young’s
modulus of the rubber cover and the applied load as well as speeds and viscosity.
Specifically, Coyle (1988) obtained the following expression for the film thickness:
T = R ( V) W (small deflection limit) (2.16)
/ / / /
T=R V E W (large deflection limit) (2.17)
: viscosity, V: roll speed, E: Young’s modulus, W: load, R: roll radius
49
Figure 2.29: Simple one
one-dimensional elastohydrodynamic model of forward
deformable roll coating (Coyle, 1988)
(a) Process definition

definition,, (b) Definition of variables for positive gap case, (c)Definition
of variables for negative gap case
Note in these equations, E is assumed to remain constant and not depend upon
operating conditions. Cohu and Magnin (1997) remarked that the characteristics of
the rubber sleeve may play an important part as the elastic modulus may change. In
50
other word, E is not constant and may be referred to as a dynamic elastic modulus,
determined
mined according to the time scale of the process, rather than a static constant
elastic modulus. The data of Cohu and Magnin (1997) with thin and thick sleeves
prove that (see Figure 2.30).
Figure 2.30: Deviation of the experimental data from Coyle

Coyle’ss theoretical predictions
as a function of the ratio between the cover thickness and the contact half
half-width
(Cohu and Magnin,, 1997)
b: cover thickness, d/2:

/2: contact half
half-width
Carvalho (2003) took-up

up this aspect in his comprehensive finite element
model and was able to predict the effect of the viscoelastic properties of the rubber
sleeve and its thickness on the film thickness formed (see Figure 2.31).
2.
51
Figure 2.31: Flow rate at different elasticity numbers and gaps (positive and
negative) (Carvalho,, 2003
2003)
Es (Elasticity number): V/ER

ER (the ratio of viscous to elastic forces)
2.4.4 Reverse Deformable Roll Coating
As just explained, deformable roll coating is normally run in the forward
mode. This explains the scarcity of research studies of this flow in the reverse mode.
For example, Kang et al. (1991

(1991b), Wang (1998) and Tiu et al. (1999) all assumed in
their experimental study of reverse deformable roll coating that all the incoming feed
flow was transferred to the applicator roller with no net flow flowing through the gap
when clearly, it cannot be the case, always. Clearly this method is in principle
capable of delivering very thin films indeed and a full evaluation of this is necessary
and forms one objective of this research.
52
2.5 RIBBING INSTABILITIES IN ROLL COATING FLOWS
Ribbing instabilities are common in coating flows and they result in wavy
patterns on the free surface as shown in Figure 1.5(a). Clearly these imperfections
spoil the uniformity of the film and several studies have been carried out to
understand their origin and find solution to prevent their occurrence. Forward roll
coating is particularly prone to these instabilities and has been the subject of many
theoretical and experimental investigations.
Pearson (1960) was the first to carry out a theoretical investigation of ribbing
by applying the lubrication approximation and linear stability theory to the flow
associated with a wedge spreader. He found that Ca number is an important factor
how ribbing instability changes.
Pitts and Greiller (1961) considered the evolution of the disturbance to be
quasi-steady-state, and they supposed that at some location in the third dimension the
coating meniscus moves slightly upstream because of a disturbance. As illustrated in
Figure 2.32, the pressure under rib is different from the pressure in the unperturbed
region for two reasons: there is a pressure gradient in the flow, and the radius of
curvature of the meniscus has been changed. If the pressure under the perturbation is
the greater of the two, it will generate lateral flow out of the perturbation, causing it
to level. If the pressure under the perturbation is the lesser of the two, it will generate
lateral flow into the perturbation, causing it to grow. This latter situation corresponds
to a flow which is unstable to a ribbing disturbance.
53
Figure 2.32: Schematic of a sinusoidal perturbation to a coating meniscus (Coyle,
1997)
Following on from this, Pitt and Greiller (1961) approximated the force
balance at the perturbed meniscus to produce the intuitively appealing criterion that
the flow becomes unstable if the pressure gradient in the liquid exceeds a certain
value, i.e.
< ( +N ) (2.18)
r:: radius of curvature of the meniscus,
N: Dimensionless wave number of the disturbance
s:: streamwise coordinate in the direction from the gap to the free surface
The right part of this equation is positive therefore surface tension stabilizes the flow
withh respect to ribbing. This equation implies that the steeper the pressure increases
in the flow direction (toward the meniscus), the greater the tendency of the flow to
become prone to ribbing, in other words being more unstable. On the other hand, if
the pressure decreases in the flow direction, the flow is stable to ribbing.
54
The pressure gradient at the downstream region in forward roll coating
system is always positive (Figure 2.33). Therefore we can understand that ribbing
instability occurs when the capillary number increases.
Figure 2.33: Computed pressure along the main flow direction in forward roll
coating (Coyle, 1997)
Left side: Upstream, Right side: Downstream
Using a lubrication approximation analysis of the flow, Pitts and Greiller
(1961) applied this criterion to forward roll coating and obtained the following
condition for the onset of ribbing:
Ca = 14 (h /R) (2.19)
The discrepancy between this correlation and their measured correlation (Ca =
31(h /R)) was attributed to the experimental difficulties in recognizing reliably the
first appearance of uneven flow.
55
Benkreira et al. (1982b) used Pitts and Greiller’s instability criteria in the
unsymmetrical forward roll coating (unequal roller radius and speeds) and using a
lubrication approximation type flow model found the following correlation for the
onset of instabilities:
/
Ca = 8.5 h /R (2.20)
It is important to note the contribution of Savage (1977) to the study of
ribbing instabilities. Indeed in his paper on cavitation in lubrication flows of journal
bearings, and roll coating flows, he observed that Pitts and Greiller (1961) instability
criterion is applicable to all such flows.
Coyle et al. (1990c) studied the onset of ribbing theoretically using a full 2D
flow analysis solved numerically using finite elements. His results together with
other purely experimentally observed data (Mill and South (1967), Greener et al.
(1980)) are shown in Figure 2.34 and suggest that Pitts and Greiller and Benkreira et
al. lubrication approximation approach give equally good predictions of the onset of
ribbing instabilities. However, although Coyle et al. (1990c) visual observations are
in agreement with the theoretical predictions, finer disturbances as viewed with a
light reflections method are not “picked” by these models. The important practical
observation in all cases is that according to the experimental and theoretical data,
large gaps and low Ca number are required to prevent the onset of ribbing which
means that high speeds and thin films will almost always be unstable. This is a
major challenge to the aim of this research and rules out forward roll coating as a
process to develop thin stable films at high speeds.
56
Figure 2.34: Critical capillary number for the onset of ribbing as a function of the
gap/diameter (Coyle, 1990c)
It is useful to note that recent work by Lie et al. (2009) who analysed
analy
numerically ribbing in forward roll coating by use of CFD

CFD-RC
RC software package
support these observations. Lie et al

al. found from these computations that the forward
roll coating becomes stable by increasing the roll gap or decreasing the flow
viscosity, which means increasing Re number. They concluded that Re number and
Ca number both increase with roll velocity but eac

eachh has opposite effects on ribbing
instability, therefore,
herefore, the roll velocity, itself, may not determinate the ribbing
instability.
Newtonian effect which are always important to consider in

As for non-Newtonian
practice, Han et al. (2009) examined the effect of visco

viscoelasticity
elasticity on the coating
window in forward roll coating experimentally and found these effects to be
57
stabilising, i.e. the critical Ca number at which ribbing occurs decreases in the case
of viscoelastic liquids.
Ribbing in reverse roll coating can be explained by pressure gradient at a
meniscus region in the same way as in the forward roll coating. Figure2.35 indicates
the pressure along the main flow direction in reverse roll coating as computed
numerically by Coyle (1997) using a full 2D finite element method analysis. At high
speed ratio the pressure gradient at the downstream region changes into negative. In
reverse roll coating the pressure gradient thus stabilises the flow compared with
forward roll coating. This feature explains why in practice, reverse roll coating is the
configuration used and the one that was taken in this research to investigate the
conditions to achieve very thin films at high speeds. Also, from the profile of the
pressure gradient in the meniscus area as a function of speed ratio and Ca (see Figure
2.36), it appears that at low Ca, the flow first is destabilised as speed is increased,
before finally becoming stabilised again at high enough speed ratio, while at high Ca,
the flow becomes stabilised as speed ratio is increased (Coyle, 1990a). This feature
indicates that there are possibilities to manipulate the speed ratio and the viscosity so
that high speed operation that produces very thin film remains stable.
58
Figure 2.35: Computed pressure along the main flow direction in reverse roll coating
(Coyle, 1997)
Figure 2.36: Computed pressure gradient at metered film meniscus as a function of

speed ratio and Ca number (Coyle, 1990
1990a)
In addition to the above discussion, the following observations are also
important. Inertia
nertia can be the driving force for the ribbing instability, as it also can
59
create a pressure gradient (Yih, 1960). Non-Newtonian shearing thinning effects, as
the study of Coyle et al. (1990d) shows, aggravate ribbing as does viscoelasticity
where ribbing became irregular and time dependent (Han et al., 2009). Also
although Coyle et al. (1990a), Malone (1992), and Hao and Haber (1999)
investigated the effect of the feed film thickness on the metered film thickness
formed, they did not consider this effect on ribbing instability. This is a void in the
research of reverse roll coating that is worthy of consideration in the present
programme as the inlet film thickness is one means of obtaining very thin metered
films.
Carvalho and Scriven (1997) used the same stability criteria described above
to analyse the onset of ribbing in deformable forward roll coating with positive gaps.
This is really not such a different system from the rigid roller case. The results
however indicated that a deformable cover may be useful in order to delay the onset
of ribbing for a desired coating thickness. An explanation, consistent with the
observation made above on pressure at exit, is that the pressure distribution with a
deformable roller is comparatively lessened with lower maximum and minimum
pressures and a moderated pressure gradient. Carvalho and Scriven (1999) extended
his analysis to the negative gap case and observed also that the deformation of the
roll cover delayed the onset of ribbing, i.e., deformable gaps can be operated at
higher speeds without ribbing. Carvalho and Scriven (1999) did not however
provide experimental data to support their theoretical findings.
Chong et al. (2007) carried out a comprehensive experimental study to assess
ribbing instabilities in deformable forward roll coating with negative gaps and
60
reported that ribbing was always present. Large negative gaps, a reduction in roll
speed and the use of a thinner compliant covering were all found to produce ribbing
of shorter wavelength. This finding is particularly important in the context of the
present research and is worth following as in practice shorter wavelength ribs may be
“acceptable” as they do not excessively affect the smoothness of the film and in
certain cases cannot be seen with the naked eye. As with rigid roll coating, Lee et al.
(2010) observed experimentally that viscoelasticity aggravates ribbing.
No research has been reported on ribbing in reverse deformable roll coating.
This is because no research has been done altogether on reverse deformable roll
coating, probably because in practice it is not attractive to use this configuration
which may tear the rubber layer. In the context of the present research, this is an area
worth investigating if it does produce very thin films at very high speeds which
cannot be produced in any other ways. With advance control technique to minimise
the risk of tearing the rubber sleeve, there is no reason why reverse deformable roll
coating could not be operated.
2.6 AIR ENTRAINMENT IN ROLL COATING FLOWS
In addition to ribbing instabilities, air entrainment also limits stable operation
as it creates fine air bubbles to be trapped in the film and these naturally mar the
quality of the final product. As this is a serious limitation, a presentation of the
problem is given here and relevant studies carried out in roll coating presented.
61
2.6.1 The Nature and Physics of Air Entrainment
Air entrainment is at the heart of the physics of all coating flows and is
concerned with the state of the dynamic wetting line as the speed is increased (see
Figure 1.5(b)).
). In all coating flows, this line delimits three phases, air
air-solid-liquid,
and although the substrate is moving this line appears to be stationary and hence
there is a singularity in the flow. So how does this line break and give way to air to
be entrained? One simple coating flow that has been used to understand this problem
is dip coating in which a contact angle can be defined (see Figure 2.37). When
examining roll coating it is easy to see that in the dynamic wetti

wetting
ng region there is a
similar flow situation: a substrate plunging into a pool of liquid.
Figure 2.37: Three

Three-phase
phase junction in dip coating: moving substrate/coating
liquid/surrounding air
Deryagin and Levi (1964) were the first to establish the mechanism of
dynamic failure in dip coating. They observed that as the substrate speed was
increased, the dynamic contact angle increased until it reached a value of 180 and
air was being entrained. At this critical point, just before they could see air bubbles
being entrained, they observe that the dynamic wetting line broke from being straight
into a serrated line much like a series of vvv (see Figure 2.38)) in which the angle in
62
the single “vees”” are essentially the same. Many other researchers followed from
Deryagin and Levi’s work and reported the same observation and also experimental
equations correlating the speed at which these vvv appear and air entrainment occurs
with the physical proper

properties of the liquid being coated.
Figure 2.38: Break up of the dynamic wetting line (Deryagin

Deryagin and Levi,
Levi 1964)
As indicated by Blake et al. ((1997)) in his review of air entrainment in dip coating,
viscosity appears to be the only significant parameters with Gutoff and Kendrick
(1982) correlation being the one most quoted in the literature:
V*  5.11  -0.67 (2.21)
In the practical context of coating operation, it means that to achieve high speed
coating it is required to use coating of low viscosities. This, at least, is compatible
with the aim of this work which is to seek high speed operation but with lower
63
viscosities coatings of the type currently developed for high technological
applications.
Although the fact that viscosity is important in air entrainment would suggest
that it is worthy to investigate non-Newtonian effects, shear thinning in particular,
very little work has been done in this area. Interestingly, Bolton and Middleman
(1980) showed that elasticity in a coating fluid postponed the onset of air entrainment
to higher speeds. This may be because the dynamic line has better tendency to
deform when elasticity is present compared with a purely viscous fluid.
Although, the above discussion is helpful in that it describes the onset of air
entrainment and the importance of the viscosity of the coating liquid, it does not
explain the mechanism of how the vvv is formed and how their appearance may be
postponed to higher speeds. Blake and Ruschak (1979) reported that Deryagin and
Levi (1964) zoomed in on these vvv and measured the angle F of the “v” and
observed that the product VcosF (V is the substrate speed) remained constant (see
Figure 2.39). In terms of vectors terminology, this term is the component of the
speed normal to the straight-line segments of the vvv line. This term, Blake and
Ruschak (1979) defined as the “maximum speed of wetting” V*. It is constant and is
the maximum speed at which the wetting line can advance normal to itself, i.e. at
speed equal or higher than V*, the wetting line breaks so that:
V* = V cos F V V* (2.22)
As the speed is increased, F increases until it reaches 90 at which point air is
engulfed.
64
Figure 2.39: Schematic representation of saw
saw-teeth （Blake and Ruschak, 1979）
1979
Now although the work of Blake and Ruschak (1979) iss very revealing, it was
not exploited until Cohu and Benkreira (1998) observed that the equation (V* = V
cos ) meant that if the substrate was tilted by an angle ,, the speed at which air
entrainment would occur would be increased by a factor of 1/cos.

1/cos Cohu and
Benkreira (1998) gave experimental proof that this was indeed the case (see Figure
2.40).. This “trick” is one of the ways of creating what is referred to as
“hydrodynamic assistance” of wetting after Blake observed that in curtain coating the
speed at which air entrainment was observed to occur was much mor
more greater than in
dip coating. They attributed the reason to hydrodynamic assistance resulting from th
the
pressure that a curtain of liquid creates as it impinges on a substrate several
centimetres below it (see Figure 2.41).
65
Following from these results, Benkreira and Khan (2008) and Benkreira et al.
(2008) went to develop other ways of developing hydrody

hydrodynamic
namic assistance in dip
coating via the control of roughness for a given viscosity (Benkreira and Khan
Khan, 2008)
and the control of gas or air viscosity (Benkreira et al., 2008).. The control of air or
gas viscosity is critical because as Benkreira et al. (2008) measured, a significant
increase in the air entrainment velocity is observed by reducing the viscosity of the
gas or air film near the dynamic wetting line (see Figure 2.42).
Figure 2.40: Inclined dip coating (Cohu and Benkreira, 1998)
66
Figure 2.41: Curtain
ain coating (Blake, 1994)
Figure 2.42: Relationship between gas pressure and entrainment speed (Benkreira et
al., 2008)
The discussion above gave an insight into the nature and mechanism of air
entrainment. In principle, the observations made in dip coating hold also in roll
67
coating because as explained earlier the dynamic wetting region in both is the same
except for hydrodynamic assistance which may or may not be effective as in the case
of curtain coating. Clearly, if the air viscosity can be controlled by reducing
substantially the pressure in the dynamic wetting region, the air entrainment should
be postponed. Also, if a gas of much lower viscosity than air could be used, also the
air entrainment speed should increase. There are however no data supporting this
claim.
Benkreira (2002a) made the same observation with forward roll coating and
could by carefully manipulating the inlet feed flow develop a little hydrodynamic
assistance and air entrainment speed slightly larger than those measured in dip
coating. As the present research is concerned with roll coating, it is important to note
that a large hydrodynamic assistance is difficult to achieve unless low viscosity gases
are used.
One important air entrainment issue in reverse roll coating, hence of
relevance to the present research is what is described as cascade. Coyle (1990a)
explained that cascade instability is a large-amplitude, three-dimensional, time
periodic disturbance of the metered film which occurs as a result of the dynamic
wetting line moving inside the nip as the metering roll speed is increased. Figure
2.43 shows how the wetting line moves from position (a) to (b) then to (c). Between
positions (a) and (b), the metered film begins with a thickness larger than the
68
minimum gap and stable flow is maintained. Between positions (b) and (c), the
dynamic wetting line crosses the minimum gap position which implies that the
metered film thickness now begins with a thickness larger than the minimum gap and
it touches the metering roller downstream

downstream.. The bead, i.e. the flow length in the nip is
now very narrow and uns

unstable,
table, causing point (c) to jump forward beyond the
minimum gap position and backward, trapping in the process a large air pocket.
These oscillations and air trapped cause cyclic transverse perturbations of the flow
and these are described as bars or casc

cascade
ade which leaves regular transverse bars on the
film both on the metering roller and the applicator roller.. The onset of cascade was
measured by Coyle (1990a) and found to correlate with the applicator roll capillary
Ca number and the metering roll to appl

applicator roll speed ratio as:
.
Ca = / . 0.2< Ca < 4 (2.23)
Figure 2.43: Computed steady

steady-state free surface profiles (Coyle,
Coyle, 1990a)
1990a
As the position of the wetting line is central to cascades, Coyle also presented
finite-element
element computations of how the wetting line moves when the Ca number and
the speed ratio are varied. Their predictions are however only qualitative, as they
depend on the assumed value of the contact angle and the slip coefficient near the
wetting line. Benkreira (2002b) have investigated these experimentally and found
them to correlate with applicator and metering roll capillary numbers and applicator
roller Reynolds
lds numbers as follows:
69
. .
X / RH = [1.189Ca ] – [3.902Ca ] rS (2.24)
. .
q = 240.37 Ca Re (2.25)
X : Dynamic wetting line position, q : Dynamic contact angle
Benkreira (2002b) observed that by manipulating the operating conditions, the
dynamic wetting line position and contact angle could be controlled and
hydrodynamic assistance of wetting achieved to prevent both air entrainment and
cascade. Recently, Shiode et al. (2009) carried out flow simulations of reverse roll
coating using VOF (volume of fluid) method. They investigated X and q and
compared to the results of Benkreira (2002b). The calculated values were in good
agreement with his experimental data.
Note that Coyle et al. (1990d) investigated also cascade instability using non-
Newtonian liquids. He found that in the case of highly elastic liquids, there is no
sharp transition from ribbing to cascade, rather the film gradually became mottled.
Coyle et al. (1990a) investigated coating windows which represent ribbing (as
reviewed in the section 2.5.2), stable and cascade (as reviewed in this section) areas
as a function of roll speed ratio (see Figure 2.44). In this figure the maximum Ca is
2.3, from which we can calculate roll speeds of 45m/min (200mPa.s) and 22m/min
(400mPa.s). Therefore no comprehensive data is available at high speeds.
70
Figure 2.44: Coating window in reveres roll coating (Coyle, 1990a)
○: 400mPa.s, :200mPa.s
No study of air entrainment by observing the wetting line has been carried out,
either experimentally or theoretically in deformable roll coating. Ascanio et al.
(2004) reported the phenomenon of misting droplets being ejected from the gap at
exit at high speeds ranging from 500 to 1100m/min. Their interpretation of this
phenomenon is that it is due to both film splitting and air entrainment into the nip.
Clearly, in relation to the present study which is concerned with high speeds, this
phenomenon is worth further investigation.
71
As stated earlier, no work on deformable reverse roll coating, theoretical or
experimental and on any aspects, has been carried out. This makes the present
research interesting and novel.
2.7 LIQUID TRANSFER FROM ROLL TO WEB
Gaskell et al. (1998b) studied transfer coating from rubber roll to web in
reverse roll coating using lubrication model. They found the transferred film
thickness depends on web-roll speed ratio, Ca, web tension and wrap angle.
= ( - Ca 1 S ( /T )), S<1, ¹0 (2.26)
H : Transferred film, H : Inlet film, T: web tension, : Wrap angle
S: Web to roll speed ratio
However they focused on wet film thickness and did not study coating instabilities.
As for deformable roll coating, there is a transfer region between a rubber
roll and a web. There is no literature about transfer coating from a rubber roll to a
web. Instead of that, a few researchers studied transfer regions from a deformable
roll to a steel roll or vice versa. In all studies they assumed all the liquid is
transferred and no liquid pass through the gap. (For more detail, please see the
section 2.4.4 literature review about film thickness in reverse deformable roll
coating.)
Much research has been done about liquid film transfer from a gravure roll
to a web in gravure coating. This is because the transfer is a key factor to determine
72
the final wet film thickness on a web. However there is relatively little work
published on coating instabilities in the transfer region. (For more detail, please see
section 2.2.4 literature review on gravure coating.)
2.8 THE NEW ROLL COATING RESEARCH CHALLENGES
The comprehensive literature review presented above has enabled the
identification of the following voids and challenges to address in future studies of
roll coating:
1. Although there have been many studies of rigid reverse roll coating, the focus
has not been on producing very thin stable films at high speeds. Rather, data
and simulations have been reported to cover the classical low speed
operations of order 1m/s with gap of order 100 microns and greater leading
to films of order 50 microns and greater. Studies at much lower gaps,
approaching zero, are required if very thin films are to be achieved.
2. Except for the very limited data provided by Coyle, the important effect of
inlet feed flow on the applicator roller film thickness has not been
investigated thoroughly. This aspect together with very small gaps is clearly
requirements to obtain very thin films.
3. The new technological advances are using very low viscosities fluids
(typically water like) and this is an area not covered by earlier studies in roll
coating.
On the basis of the above three observations, a study of reverse roll coating with
gaps approaching zero, controlled inlet film and low viscosities fluids (water like)
and at high speed is clearly required to underpin the new technological needs of very
73
thin low viscosity films. This justifies the main aim and objectives of this research
as stated in the introductory chapter.
In addition to the above observation, one striking void in the research
concerns reverse deformable roll coating which has not been studied at all. In the
context of the present research, it may be more practical to operate a negative gap
with a deformable roll coater than a near zero gap with rigid roller (roller clash). This
aspect will complement fully the quest to develop very thin films and this too is
amongst the objectives set for the present research.
Finally as the film formed is to be transferred to a substrate, such study would
be highly valuable to assure that the film formed remain stable when transferred to
the substrate. This aspect also will be the objective of the present research.
2.9 CONCLUSIONS
In this chapter, a literature review on roll coating flows has been carried out
to assess the important issues regarding how the film thicknesses and their stability
depended on the design and operating conditions of the coater. It was found that
forward rigid and deformable roll coating have been the subject of much research,
both experimental and theoretical. The practical lesson from this survey is that
forward roll coating is not an attractive method to produce very thin films that are
stable. Rigid reverse roll coating was found to be more stable and capable of
delivering thin films by virtue of the fact that the metering roller reduces the net flow
through the nip and produce flow (pressures) that is inherently more stable to ribbing.
Previous studies of this flow have however been limited to relatively large gaps
(>100 microns), moderate to high viscosities (>100mPa.s) and moderate speeds
74
(~1m/s). Clearly, research at gap approaching zero with much higher speeds and
lower viscosities is required to meet the new technological demands of very thin
films at high speeds. The literature survey also revealed that deformable reverse roll
coating has not been studied at all when potentially it can by definition deliver the
very thin stable films sought, in spite of the practical problem of potentially shearing
the rubber covered roller.
In conclusion, this literature survey has revealed that the aim and objectives
set in the present research are original and studying them either experimentally or
theoretically will help advance further our understanding of roll coating flows and
thin film coating science and technology.
75
Chapter 3 : EXPERIMENTAL METHOD
3.1 INTRODUCTION
The research described in this thesis is concerned with developing a new
stable regime of operation of reverse roll coating, which can produce very thin films
(less than 10 microns) at high speeds (greater than 100m/min). As explained in the
introductory chapter such films are in high demand for a whole range of applications.
As theoretical work and previous experimental investigations have not been able to
determine such a regime, a new thinking is necessary and this can be helped through
experiments. Hence the requirement to build an experimental rig that is able to
operate over a wide range of conditions and permit not only measurements of film
thickness but also visualisation of the flow to detect the onset of instabilities when
they occur and establish the stable window of operation. As the gap between the
rollers and the feed flow play a crucial role in lowering film thickness, provisions in
the design of the rig must be made to narrow the gap and reduce the feed as much as
possible, including using negative gaps and feed film thickness smaller than the gap.
We describe here the set-up, the control techniques and other aspects important to the
experimental work.
3.2 THE COATER & MEASURING TECHNIQUES (at the University of
Bradford)
3.2.1 Rollers & Coater Configuration
A reverse roll coater is in principle a simple set-up, two rollers separated by a
gap and rotating at opposite direction at the nip. The film thickness formed will
depend strongly on maintaining the gap size and constant roller speeds. The rollers
76
must thus be machined accurately and set in a solid well designed steel frame in
which the gap can be changed and controlled very accurately. The drive must also
provide constant rotation transmitted to the roller shafts. In addition, because
visualisation is important, provisions must be made to facilitate this.
Three types of rollers were used: steel, steel with a rubber sleeve and
transparent Perspex. The steel rollers were 0.20m long and 0.20m in diameter,
chromed and polished to provide an accuracy in the roller radius of <±2.5 m with a
chrome finish of ±0.1 micron. The rubber sleeved roller had a shell of 0.20m in
diameter and a rubber sleeve 15mm in thickness. A nitrile based rubber material (30,
50 and 70 shore A hardness) was selected as cover for the deformable roller. The
eccentricities of these rollers were less accurate at ±25 m. The transparent Perspex
roller although machined to the highest standard had an eccentricity of ±25 m. This
is because during machining of the surface, the exposure of the Perspex to heat
cannot produce a better finish.
For any given combination of rollers, steel-steel, steel-rubber sleeved, steel-
Perspex, the rollers pair were slotted and held in place in a solid stainless steel frame,
the design and dimensions of which are shown in Figure 3.1. The rollers were driven
independently by inverter-controlled AC geared motors via a combination of timing
pulleys and belts. By using a tachometer initial calibration of the inverters provided
accurate control of the roller speeds. The accuracy of the roll speed is ±0.1%. Figure
3.2 shows the drive and control box arrangement. As for setting the gap between the
two steel rigid rollers this was achieved using accurate stainless shims (FEELER
STOCK 667 series ®STARRETT) placed between the roller bearing blocks and
checked using stainless slip gauges, shown in Figure 3.3. The stainless shims used in
this study are various in steps of 5 m, therefore the accuracy of gap setting is
77
±2.5 m. In the case of setting the gap between steel and deformable rollers, due to
the deformable nature of the metering roller it was not possible to check the roll gap
using stainless slip gauges. Therefore the gap was calibrated by observing a light
source placed between the two rollers. Accurate calibration of the roll gap was
achieved by moving the deformable metering roller towards the steel applicator roller
until the light just disappeared when looking from side between the two rollers. At
this point the rollers were just touching and the roll gap was zero. As the deformable
roller has an eccentricity which is less accurate at ±25 m, this calibration was done
at six different points on the deformable roller. The gap at which the light is not seen
at three points and is seen at other three points is considered zero gap.
As in practice reverse roll does not allow for the return of the film on the
metering roller, this film was wiped clean in our system using a scraper attached to a
funnelling device to enable collection for measuring if required. Figure 3.4 shows
the arrangement.
3.2.2 Feed supply
As indicated earlier, besides the gap, the film thickness can clearly be
controlled by the amount of fluid arriving to the nip. In the “flooded” situation
where a surplus is available, this supply has no effect but in the “starved” situation
(see Figure 2.17(b)), the film thickness should in principle be critically dependent on
the amount supplied. As in this research, it is desired to produce various supply film
thickness (25 - 200 m), arranging a controlled feed supply was important. Two
methods were used and the feed supply achieved by both compared before starting
the experimental programme of this study. The first method was simple and
78
consisted of placing a steel round bar actin
actingg to meter the flow of liquid picked up by
the applicator roller from a trough in which it was immersed (see Figure 3.1). In
such arrangement, the gap between the applicator roller and the round bar provided a
means to control the feed suppl

supply to the nip. In order to ensure accurate control along
the length of the roller, the round bar was attached to a plate which pushed in and out
accurately using a micrometer (see Figure 3.5).
Figure 3.1: Schematic diagram and picture of reverse roll coating rig with a steel
round bar
Figure 3.2: Picture of drive and control box arrangement
79
Figure 3.3: Picturess of accurate stainless shims placed between the roller bearing
blocks
(a) Metering roller (b) Applicator roller

Figure 3.4: Pictures of scraper systems
Figure 3.5: Picture of a micrometer
80
The advantages of this method are simple configuration, easy to set and clean
the rig after experiments and a stable film can be achieved.

achieved Before starting
experiments, the relationship between the gap (the applicator roller and the steel bar)
and the supply film thickness into the roller gap region was investigated. In this
investigation the metering roller was not used, in other words the gap between the
applicator roller and the metering roller was set more than 10mm so that the liquid
did not touch the metering roller. Figure 3.6 shows the result of the relationship
between the roll/bar gap and the actual film thickness

thickness.
Figure 3.6: Relationship between roll/bar

oll/bar gap and actual wet film thickness
(Liquid: Millmax5)
The
he disadvantage of using the round steel bar is the difficulty to produce a
flooded regime especially at low speed condition

condition.. This is because the amount of
liquid from the pan is limited

limited. Note that the pick-up film can be calculated using
experimentally found correlation from earlier work in our laboratory (Spiers et al.,
1974).
/ /
h = 0.944
0.944Ca ( (3.1)

81
This equation implies that it is difficult to get thick film with low viscosity liquid at
low speed. The film thickness of the liquid used in this experiment (7mPa.s,
6m/min)) is calculated as 146 m. This value is higher than the data in Figure 3.6
(65 m at 6m/min) probably due to the difference of liquid surface position in the pan.
In this study the position of liquid surface was lower than the centre of the applicator
roller. Nevertheless it is obviously difficult to get films of thickness of 200 m with
low viscosity liquids at low speeds from the theoretical calculation. Therefore, a
second method was investigated using a supply generated by a gear pump via a die
(Figure 3.7). Here the amount of fluid supplied could be controlled at any level
precisely by controlling the speed of the gears in the pump and the gap between the
die and the applicator roller. The effect of gravity in this arrangement was also
assessed by changing the die position. At a die position 17°(see Figure 3.8(a)), it
was established from the experiments that films of thickness up to 125 m could be
achieved. The position was thus changed to a higher position, 44°(see Figure
3.8(b)), to assess if even thicker film could be obtained. This indeed resulted with an
increase in film thickness but the coating bead between the applicator roller and the
die became unstable. This is shown in Figure 3.9 which compares the film produced
at the two die positions and the unstable operation points. There were thus no
advantages to using a die and a pump system and the simpler round bar method was
adopted.
82
Figure 3.7: Schematic diagram of reverse roll coating rig with a die
d
(a) 17° (b) 44°
Figure 3.8: Schematic diagram of die

ie position
83
(a) Die
ie position 17° (b) Die
ie position 44°
Figure 3.9: Film thickness

thicknesses as a function of the gear pump frequency and the gap
between an applicator roller and a die
Solid mark: stable, Blank mark: unstable
3.2.3 Measurement
easurements of wet film thickness
The measurement of film thickness was carried out using the simple but
reliable method of scraping the film of

offf the roller and weighing it (see Figure3.1 and
3.4 (b)).. The accuracy was improved by carrying out the scraping over a reasonably
long period of time, between 30 to 120 seconds, depending on the amount of the
metered liquid. In all cases, the measuremen

measurements were repeated and the result averaged,
giving the accuracy in the film thickness of ±3%.
3.2.4 Observation of the coating instabilities
Detecting the onset of ribbing and other instabilities was a very important
consideration of the research programme. This permitted to establish the window of
stable operation to guide industrial practice. As ribbing instabilities can easily be
84
seen by the naked eye (in practice this is indeed the test), it was sufficient to provide
good illumination. Using the Perspex roller gave another dimension to this study as
we could link it to what was happening to the flow in the nip (particularly the
position of the separation meniscus) (see Figure 3.10).
Visualisation of the meniscus region was achieved by focusing a TK-1270
Color video camera (JVC) coupled with a Canon Macro lens (50mm, 1:3.5) directly
above the Perspex metering roller and mounted on an accurately moveable x-y-z
platform. The images captured by the camera were stored directly into a PC with
software (Compro DTV4). The setup was properly calibrated by first determining the
degree of magnification of the image captured using a photograph of a graticule
placed in the gap between the rollers (see Figure 3.11(a)). The location of the centre
axis of the gap as viewed from the camera was determined by making a small bead of
fluid trapped between the two rollers, and increasing the gap between the rollers so
the bead became elongated. The fluid bead naturally positioned itself centre axis of
the gap, which allowed the centre to be clearly defined (see Figure 3.11 (b)). Typical
image is shown in Figure 3.12 as an illustration of the technique.
85
Figure 3.10: Configuration for visualisation
isualisation of the meniscus region
(a) (b)
Figure 3.11: Photograph

hotograph of a graticule and a small bead placed in the gap between
the rollers (a) Graticule, (b) Small bead
86
Figure 3.12: Illustration of the technique of setting centre position
3.3 THE COATER & MEASURING TECHNIQUES (at Toyobo)
3.3.1 Rollers & Coater Configuration
A reverse roll coater was used for an analysis of wet film transfer from an
applicator roller to a PET film (see Figure 3.13). The applicator roller was 0.40m
long and 0.22m in diameter and the metering roller was 0.40m long and 0.18m in
diameter, chromed and polished to provide an accuracy in the roller radius of
<±2.5 m with a chrome finish of ±0.1 m. PET Film “A4100” 50 m (made by
TOYOBO) was used. The accuracy of PET film speed was measured ±1%. A coating
liquid was supplied from a die to the applicator roller and excess liquid was metered
by the metering roller. The accuracy of roll speed was measured ±0.1%. The roll gap
was adjusted so that 10 m wet film could be achieved on the applicator roller. The
accuracy of tension of PET film was measured ±5% during running the coating line.
87
3.3.2 Measurements of wet film thickness
The measurement of film thickness was carried out using a light interference
wet film gauge (PTT Ltd. Co.,)

Co.,). Inn all cases, the measurements were continued in a
certain time and the result

results were averaged.
3.3.3 Observation of the coating instabilities
Detecting the onset of coating instabilities on a wet film was observed by the
naked eye. Visualisation of the transfer region was achieved by focusing a
microscope VW-6000
6000 (Keyence Ltd. Co.,) through a running transparent PET film.
An upstream and a downstream meniscus positions were measured by comparing
reference pictures taken beforehand and also bubbles at the transfer region were
observed.
Figure 3.13: Schematic diagram of reverse roll coating rig
88
3.4 CHARACTERISATION OF THE COATING FLUIDS
In order to ensure the collection of a well defined set of experimental data, all
the fluids coated were tested for their physical properties, including viscosity, surface
tension and density. Densities were measured by weighting a set volume of the fluid,
collected in an appropriate flask. The surface tension measurements were carried out
using the FTA 188 video tension meter which operates on the principle of creating a
drop from a syringe, recording a series of pictures of the pendant drops and
processing these information to obtain a volume and a weight of the pendant drop
from which the surface tension can be calculated from a force balance (Harkins and
Brown, 1919), i.e.:
W = 2 rsf (3.2)
In the above equation, W is the weight of the drop, the surface tension and f a
correction factor which accounts for the fact that a small amount of the drop always
remains attached to the capillary syringe. Harkins and Brown (1919) compiled these
correction factors from a series of measurements using syringe of various diameters.
As for viscosities, these were measured using a modern rheometer (Physica
MCR 301, Anton Paar) which uses as principle the shear flow created between a
rotating cone of very small angle and a fixed plate. In our measurement, the CP50-1
(50 mm diameter cone and 1 angle cone) system was used. Such a flow arrangement
creates at a constant speed and constant gap a constant shear rate. The flow system
then measures the torque from which the shear stress exerted by the fluid can be
calculated. The experimental approach was thus to collect torque versus angular
velocity data, and calculate shear stress, and shear rate, from which the
apparent viscosity, can then be calculated from the equation:
89
= (3.3)
Viscosity measurements were made before and after coating experiments and
showed no variation.
Densities of the test liquids were calculated from accurate mass and volume
measurement using the simple relationship:
Density = mass / volume (3.4)
The properties of liquid used in this study at 20 C are shown in Table 3.1. The effect
of temperature on the viscosity is shown in Table 3.2. It is noted that all four types of
liquids are Newtonian which is suitable for understanding basics of fluid mechanics
though Non Newtonian liquids are widely used in industries. All coating experiments
were conducted at room temperature between 20 and 25 C.
Table 3.1: Physical properties data for liquids used (20 C)
Surface
Viscosity tension Density
Sample (mPa.s) (mN/m) (kg/m3)
Millmax5 6.6 26.8 820
Millmax15 27 28.2 860
Millmax68 176 28.6 870
Water/Glycerine(50/50) 5.9 65.4 1130
90
Table 3.2: Effect of temperature on the viscosity of liquids used (20 C, 25 C)
Temperature
Sample
20 C 25 C
Millmax5 6.6 5.9
Millmax15 27 21
Millmax68 176 128
Water/Glycerine(50/50) 5.9 5.1
In this study (at Toyobo) Polyester resin, water and isopropyl alcohol were
used for coating liquids. The properties of liquid used in this study at 25 C are shown
in Table 3.3. The effect of temperature on viscosity changes is shown in Table 3.4.
All coating experiments were conducted at room temperature between 20 and 25 C.
Figure 3.14 shows liquid viscosity as a function of shear rate. For measuring
viscosity at low shear rates a cone and plate geometry was used and for high shear
rates a parallel plate geometry was used and both results were combined into one
figure.
Table 3.3: Physical properties data for liquids used (25 C)
Viscosity Surface
Density
Liquid (mPa s) tension
(kg/m3)
at 100(1/s) (mN/m)
1 105 26.2 1058
2 44 25.5 1054
3 2.9 34.5 1025
4 2.8 26.2 968
91
Table 3.4: Effect of temperature on the viscosity of liquids used (20 C, 25 C)
Temperature
Liquid
20 C 25 C
1 110 105
2 46 44
3 3.4 2.9
4 3.3 2.8
Figure 3.14: Liquid viscosity as a function of shear rate
92
Chapter 4 : RESULTS & DISCUSSION
4.1 INTRODUCTION
In this chapter the results of experiments with various rollers as a metering
roller will be showed. A steel roller is always used as an applicator roller. First of all
the experiment with a transparent roller was carried out in order to observe the nip
region from the top position of the Perspex metering roller, and understand the
movement and stability of the meniscus. After that, and in order to investigate more
precisely, the fluid mechanics (measurement of film thickness and onset of instability
at precise gap), a steel roller was used as the metering roller. The stability assessment
of the flow required following this with complementary visualization experiments
with the Perspex roller. In addition, a deformable roller was also used as a metering
roller in an attempt to reduce film thickness further whilst producing stable films at
high speeds. In the last part of the experimental programme, carried out at Toyobo,
liquid transfer from an applicator roller to a PET film was investigated to achieve a
full understanding of the entire operation.
4.2 PRELIMINARY MEASUREMENT WITH THE TRANSPARENT
ROLLER
In these experiments, the low viscosity lubricating oil (Millmax5) was used.
The accuracy of Perspex roller was about ±25 m. The roller gap was set at 50 m
which is the minimum roller gap due to the eccentricity of the Perspex roller. The
supply film thickness was fixed at 100 m, equivalent to a flooded inlet regime. The
speed of the applicator roller was fixed at 60m/min and the speed of the metering
roller (Perspex roller) was gradually increased from at zero to 90m/min.
93
4.2.1 Film Thickness
hickness and Instabilities
Figure 4.1 shows the film thickness

thicknesses measured on the applicator roller and
the metering roller as a function of roller speed ratio. In the case when the metering
roller is not moving, there is much amount of liquid on the applicator roller after
metering. When the speed of the metering roller is increased,

eased, thinner film can be
achieved on the applicator roller. However beyond a critical ratio (Vm/Va = 0.75),
the film thickness increased again. It was observed that the region where a minimum
film is obtained correspond to the stable coating window. This trend is the same as
the result of Coyle’ss study ((Coyle, 1990a) but a difference that there is an area of
ribbing after and before the stable area (see Figure 4.1 and Figure 2.44).
2. This may be
due to the difference in the viscosity of the liquids used between this study and
Coyle’s study in which the viscosity was more than 100mPa.s

100mPa.s.. This point will be
discussed further in the Chapter 4.3 with the more precise experiments using a steel
metering roller.
Figure 4.1: Film thickness

thicknesses on the applicator roller and the metering roller as a
function of roller speed ratio
Liquid: Millmax5, Applicator roller speed: 60m/min,
Supply film thickness: 100 m, Roll gap: 50 m

94
4.2.2 Meniscus Changes
Figure 4.2 shows how the meniscus changes at various speed ratios.
ratio Images
(1) indicate meniscus regions through the transparent roller captured from the top
position. Images (2) indicate the applicator roller after metering.. At the speed ratio
Vm/Va =0.75,, where the minimum and stable film can be made, the meniscus
situation is stable as well. On the other hand at other conditions (Vm/Va =0, 1.00,
1.17),, unstable meniscus situations and instabilities of the wet film on the applicator
roller can be seen. Th

Therefore in order to produce a stable film, a stable meniscus
situation has to be established

established.
Figure 4.2: (1) Images of meniscus situations changes; (2) Images of applicator
roller after metering
95
4.3 MEASUREMENT WITH A STEEL PRECISION METERING
ROLLER
In this section a steel roller was used as a metering roller in order to carry out
more precise tests, the accuracy of the roller diameter being ±2.5 m. The applicator
and metering roller speeds were set in the range 6-150m/min and 0-150m/min
respectively with the roller speed ratio (Vm/Va) set in the range 0 - 1.7. The gap
between the two rollers was set in the range 25 - 200 m and the supply film
thickness was varied from 25 to 200 m. Three liquids (lubricating oils) of viscosity 7,
30 and 180mPa.s were used. The surface tension of these oils was 28mN/m. In order
to assess the effects of surface tension, a 6mPa.s glycerine/water solution was also
used. This solution had a surface tension of 65mN/m.
4.3.1 Effect of roller speed
In this section, the data with Millmax5 of viscosity 7mPa.s are presented.
Figures 4.3(a) - 4.5(a) show the metered film thickness as a function of roller speed
ratio and Figures 4.6 - 4.8 show the minimum metered film thickness. It can be seen
that as the applicator roller speed increases, the minimum metered film thickness also
increases and the speed ratio at which a minimum film can be achieved decreases.
Figures 4.3(b) – 4.5(b) give the coating window which shows that the higher
the applicator roller speed, the narrower the coating window in which stable films
can be achieved.
96
(a) Film
ilm thickness (b) Coating window
Figure 4.3: Effect of applicator roller speed. Supply film=50 m, Roll gap=25
gap m
(a) Film
Figure 4.4: Effect of applicator roller speed. Supply film=100

100 m, Roll gap=50
gap m
97
(a) Film
Figure 4.5: Effect of applicator roller speed. Supply film=200

200 m, Roll gap=100
gap m
Figure 4.6: Relationship between applicator roll speed and minimum film thickness.
Roll gap=25
25 m, Supply film
film=50 m
98
Roll gap=50 m, Supply film
film=100 m
Roll gap=100 m, Supply film
film=200 m
4.3.2 Effect of roller gap at constant viscosity
Figures 4.9 (a) - 4.12 (a) show the metered film thickness as a function of
roller speed ratio and roll gap for Millmax5. The results indicate that thin metered
99
film can be achieved at small roller gap
gaps. Also, there
ere are no major differences in the
roller speed ratio (Vm/Va) at which the minimum metered film is achieved.
Figures 4.9 (b) - 4.12 (b) show the coating window at applicator speeds
varying from 30 to 120 m/min. We observe that at as the applicator roll speed
increases, the coating window narrows until it dissappears totally at 120 min/min.
min/min At
60m/min there is not a big difference at all roller gaps. Att 30m/min, the coating
window widens as the gap is increased. This is probably due to the fact that the Ca
number for these runs is very low. At 60m/min, no widening of the window is
observed as the gap is increased then at 90m/min, the stable coating window
disappears at large gaps. One of the reason of this result may be due to high speeds
and large gaps causing high inertia (Re). Another reason is that the wave length of
ribbing at larger gap is long making ribbing patterns easily seen.
(a) Film thickness (b) Coating window
Figure 4.9: Effect of the roll gap. Applicator Roll Speed= 30m/min
Supply film=200 m,, Liquid: Millmax5
100
Figure 4.10: Effect of the roll gap

gap. Applicator Roll Speed=60m/min
0m/min

0m/min
101

0m/min
4.3.3 Effect of supply film thickness
Figures 4.13(a)
(a) – 4.18(a) show the metered film thickness as a function of
roller speed ratio for a fixed roller gap of 100 m and for a range of supply film
thickness, 25 to 200 m. Figures 4.13(b) – 4.18(b) show the corresponding coating
windows and Figures 4.19, 4.20, the minimum metered film thickness.
thickness
It was expected that the metered film thickness would decrease when the
supply film thickness was reduced, however there was no effect on the minimum
metered film thickness at all supply film thicknesses. As the data indicates, the film
thickness measured on the applicator roller is clearly sensitive to the supply film
thickness when this supply film thickness is smaller than the gap (100 m). At supply
film thickness equal or large

largerr than the gap, the sensitivity disappears. As Coyle
(1990) had carried out similar experiments it is useful to compare the data of his
study with those measured here. There is however one important difference and it is
102
in the values of Ca used which had the value of 0.005 in Coyle’s study. Also in
Coyle’s study, R/h0 was equal to 1000 making the gap h0=300 m as the roller radius
was 0.15m. As for the viscosity and surface tension values used by Coyle, these were
not given (glycerine and water solutions with a range of viscosity 0.1 to 2Pa.s was
quoted and for which we may assume a surface tension of 65mN/m). Using his
lowest viscosity value, the corresponding Va would be equal to 0.2m/min which is
too low in practical conditions. In our study, the minimum Ca used was 0.025
(6m/min, 27mN/m, 7mPa.s); this is because 6m/min is the minimum speed at which
the roller rotates stably. It is thought that the supply film thickness does not have an
effect on the metered film thickness in practical or industrial range. This is probably
because the form of downstream meniscus does not change even if supply film
thickness is changed.
The reason why the results by Coyle showed the decreasing of the metered
film thickness when the supply film thickness was reduced might be that there was a
gravity effect at very low speeds. Gaskell (1998a) reported that the gravity at the
downstream meniscus affected the film thickness in forward meniscus roll coating.
In Coyle’s study, as it was extremely low speed conditions, the gravity effect may
have occurred.
Taking into account the data shown so far, it can be also thought that the
metered film thickness depends only on the roller speed and roller gap in the case of
using the same liquid at practical speeds (in this study: more than 6m/min).
Regarding the coating window, it becomes narrow when supply film
thickness is reduced. This is thought to be due to the air entrainment occurring easily
when the amount of liquid at the roller gap region is small. ["× no" in the figures of
103
coating window means that there is no coatin
coatingg bead at the roller gap region because
the supply film thickness is much thinner than the roller gap. ]
(a) Film thickness

kness (b) Coating window
Figure 4.13: Effect of the supply film thickness. Applicator Roll Speed=6m/min
Speed
Roll gap=100 m,, Liquid: Millmax5

Figure 4.14: Effect of the supply film thickness Applicator Roll Speed=15m/min
Speed
104
Speed

Speed
105
Speed
Speed
106
Figure 4.19: Dimensionless
imensionless m
minimum metered film thickness (Effect
Effect of the supply
film thickness ) Liquid: Millmax5, Roll gap=100 m
Figure 4.20: Dimensionless minimum metered film thickness (Effect

Effect of the supply
film thickness ) Liquid: Millmax5, Roll gap=25, 50, 100, 200 m
107
4.3.4 Effect of viscosity at constant roller gap
Figures 4.21 – 4.24 show the effect of viscosity (7, 30 and 180mPa.s) at a
constant roll gap of 50 m and a supply film thickness of 200 m (flooded inlet). As
previously, the data give both the film thickness formed at exit on the applicator
roller (Figures 4.21(a) – 4.24(a)) and the corresponding coating windows (4.21(b) –
4.24(b)).
With regard to the metered film thickness, we observe that it is shifted higher
as the viscosity increasing. The linear lubrication approximation region is much
narrower with the higher viscosity fluid causing the critical speed ratio where the
film thickness upturns and increases to be comparatively lower. Typically for the
180mPa.s fluid, the critical speed ratio is about 0.4 whereas at 7mPa.s it is
approaching 0.9. These values are for the low applicator speed of 30m/min. As the
applicator speed is increased, the shift in the metering film thickness curves is even
higher, upwards with the lubrication approximation region narrowing further and
critical speed ratios being 0.2 and 0.6 for viscosities 180mPa.s and 7mPa.s
respectively at 120m/min.
As for the coating windows, they become narrower when the viscosity and or
the applicator roll speed are increased. This is equivalent to increases in Ca causing
the same effect. For example, at high applicator speed (90m/min), the coating
window is narrow even with the low viscosity 7mPa.s liquid. It is also found that the
coating window of the low viscosity liquid has a ribbing area beyond the speed ratio
at which a stable film can be achieved unlike the high viscosity liquid where cascade
appears immediately beyond the stable area. At 120m/min, there is no stable coating
window with any types of liquids at the gap of 50 m (see Figure 4.24(b)).
108
Figure 4.21: Effect of the viscosity. Applicator Roll Speed=30m/min

m/min
Supply film=200 m,, Roll gap=50 m

m/min
109

m/min

m/min
110
4.3.5 Coating Window dependence on all operating variables
Figures 4.25 - 29 bring together the coating windows dependence on roll gap,
viscosity, applicator speed and speed ratio in an attempt to identify the stable set of
operating conditions. Although in the case of the very low viscosity liquid
(Millmax5), the coating windows change depending on the roll gap at 30m/min,
fundamentally there are no major differences in the width of the coating windows at
various roller gaps with Millmax15 and Millmax68 (see Figure 4.25). This is thought
because the capillary number in the case of the 30mPa.s and 180mPa.s are higher
than that of 7mPa.s, therefore the effect of the large roll gap on reducing ribbing is
not enough. At 60m/min and with the three liquids, there are no major differences in
the width of the stable coating window at various roll gaps (see Figure 4.26). At
90m/min, as observed earlier, the stable coating window of the low viscosity liquid
disappears at large gaps; however in the case of the higher viscosity liquids, there is a
stable coating window even at large gaps (see Figure 4.27). This may be attributed
Re being lower with the high viscosity liquids. At 120 and 150m/min, the stable
coating window of the low viscosity liquid disappears. However there are stable
coating windows at large gaps with higher viscosity liquids (see Figures 4.28, 29).
The ribbing mechanism at high speed will be discussed later (section 4.4.6 ).
111
(a) 7mPa.s (b) 30mPa.s
(c) 180mPa.s
Figure 4.25: Coating window ((Effect of the roll gap).. Applicator Roll
Speed=30m/min
Supply film=200 m
112
(c) 180mPa.s
Speed=60m/min
Supply film=200 m
113
(c) 180mPa.s
Speed=90m/min
Supply film=200 m
114
(c) 180mPa.s
Speed=120m/min
Supply film=200 m
115
(c) 180mPa.s
Speed=150m/min
Supply film=200 m
Figure 4.30 shows aan additional feature at high Ca (> 4) not covered in Coyle
et al.(1990a) work. One such a feature is that at gaps = 25 m and 50 m we
observed an upper limit in Ca above which no stable flow occurs whatever the speed
ratio is set at and the flow changes directly from ribbing into cascade (see Figure
116
4.30 (b) (c)) or always ribbing (see Figure 4.30(a)). A second feature is that with the
fluid of viscosity 27 mPa.s, the flow, at all gaps 25, 50, 100 and 200 m, goes
ribbing-stable-ribbing-cascade (see Figure 4.30(b)) whereas the fluid of viscosity 176
mPa.s always goes ribbing-stable-cascade (see Figure 4.30(c)). This can be explained
with reference to the movement of the wetting line towards the minimum gap as Ca
and rS are increased. With the high viscosity liquid or Ca, it is closer to the minimum
gap position than with the low viscosity fluid or Ca and as the speed ratio is
increased, it moves past the minimum gap and induces cascade. With the low
viscosity fluid, it remains shifted away from the minimum gap position so becomes
prone to ribbing instead as the speed ratio is increased until with further increase in
speed ratio it goes past the minimum position and triggers cascade instabilities.
These observations are consistent with the CFD simulations of Coyle et al. (1990a)
on the movement of the wetting line towards the minimum gap as Ca and rS are
increased.
117
Figure 4.30: Coating window (Effect of viscosity).
Roll gap=50µm, Supply film=200µm
4.3.6 Effect of surface tension
Figures 4.31 - 4.35 show the effect of surface tension at applicator roller
speed of 30, 60, 90, 120, 150m/min, respectively

respectively. As previously, the data give both
the film thickness formed at exit on the applicator roller ((Figure

Figures 4.31(a) - 4.35(a))
and the corresponding coating windows ((4.31(b) - 4.35(b)).
With regard to the metered film thickness, we observe that the film
thicknesses differ only in the region above the critical speed ratio with the film
118
thickness of the low surface tension liquid becoming very much higher than that of
the high surface tension liquid

liquid. The coating windows on the other hand become
wider with the higher surface tension liquid, as expected.
Figure 4.31: Effect of the surface tension. Applicator Roll Speed=

=30m/min

=60m/min
119

=90m/min

=120m/min
120

=150m/min
Figure 4.36 shows the minimum metered film thickness as a function of roller speed.
The metered film using liquid with =65mN/m is slightly thinner than that using
liquid with =27mN/m.
Figure 4.36: Minimum metered film thickness (Effect of the surface tension).
tension)
Supply film=100
100 m, Roll gap
gap=50 m
121
Figures 4.37(a)
(a) (b) show the coating window with 27mN/m liquid and
65mN/m liquid respectively. In the case of high surface tension liquid,

liquid even at
150m/min stable film

films are able to be achieved. This is due to the surface tension
force making the meniscus between the two rollers stable.
(a) 27mN/m (b) 65mN/m
Figure 4.37: Coating window ((Effect of the surface tension).
(a) Liquid: Millmax5 (lubricating oil) (27mN/m)

(b) Liquid: Water/Glycerine (65mN/m)
4.3.7 Film thickness analysis
In this section all data so far presented will be discussed. The film thickness
data are presented in Figure 4.38 as HM vs. rS at fixed Ca for comparison with the
only other available data of this kind, those due to Coyle et al (1990a). The first
observation is the qualitative agreement regarding the form of the curve, a linear
lubrication approximation region followed by an upturn above a critical speed ratio,
122
rS*. The second observation is the added value of the present work: it pro
provides
complementary data in the interesting region where the film thickness drops
significantly below 0.45 down to as low as 0.05. Observe from Figure 4.38 that the
bulk of Coyle et al. (1990a) data were in the higher range of film thickness, except
for twoo data sets at Ca=0.1 and Ca=0.005.
Figure 4.38: Film thickness measured in this study and comparison with data of
Coyle et al. (1990).
Figure 4.39 shows the correlation between Ca and the dimensionless
minimum film thicknes

thickness H (=ha/h0).. According to this graph, as Ca number increases
H also increases which implies that in order to decrease the film thickness,
thickness Ca must
be decreased. Interestingly, Landau-Levich (1942) studied dip coating and derived
the following equation between the thickness of the film formed and the
corresponding Ca:
h ¥ / rc = 1.34 Ca2/3 (4.1)
123
h ¥ = final film thickness, rc: radius of curvature of meniscus
There is a good agreement with the experimental data in this study and Landau-
Levich model up to about Ca = 0.1 in Figure 4.39. However at Ca>0.1

Ca there is a
large discrepancy. This is due to the assumption of Landau

Landau--Levich model that
assumes that the radiu

radius of meniscus remains constant at any places.
places In reality such as
in roll coating flow and with the gap confinement, the radius of meniscus changes
and this affects the final film thickness (Figure 4.40).
Figure 4.39: Correllation

ation between Ca number and dimensionless minimum film
thickness
(a) Landau-Levich
Levich model (b) Roll coating at high Ca number
Figure 4.40: Meniscus shape
124
From a practical point of view, an other important operation information is the roll
speed ratio, rs* at which the minimum film thickness Hmin occurs. This is given
Figure 4.41 which shows that as Ca increases, rs* decreases. As stated in the
literature review chaper Kang and Liu (1991) found a empirical relationship between
Ca and rs*. Figure 4.41 shows the comparison and suggests a more appropriate
correlation covering the wide range of Ca tested as:
* .
0.46 (4.2)
Figure 4.41: Critical speed ratio vs. Ca: data from this study (¨) and
comparison with Coyle et al. (1990) data (D) and Kang and Liu (1991)
correlation (-------).
125
Figures 4.42 and 4.4
4.43 show the dimensionless film thickness
hickness H as a function
of rs* for Ca»0.1, 0.6 and 2.3. At low Ca values of 0.1 and 0.6-0.7,
0.6 there are
significant difference
differences between the data from this study and that of Coyle. This is
probably due to the submerged configuration or the difference in roll speed in
Coyle’s study where relatively high viscosity liquids were used. At high values of Ca
(=2.3-2.4), there is no major difference in the data from this study and that of Coyle.
This is probably because the speed is not too low. Following from this comparison,
the possible reason why the wet film thickness at low Ca in this study is lower than
that of Coyle is the difference in viscosity or the configuration of the coating rig.
Figure 4.44,, which compare

compares both sets of data indicates
ndicates that there is a significant
difference at low Ca and little difference of at high Ca number. An experimental
correlation on how the minimum film thickness varies with Ca and rS can also be
developed to guide operation. This correlation derives from the data given in Figure
4.44 and can be expressed as:
.
, 0.33 (4.3)
Figure 4.42: Comparison of film thickness between Coyle’ss and this study (Ca=0.1)
126
Figure 4.43: Comparison of film thickness between Coyle
Coyle’ss and this study (Ca=0.6-
(Ca=0.6
0.7, 2.3-2.4)
Figure 4.44: Minimum Film thickness measured in this study (¨) and
comparison with data of Coyle et al. (1990) (D).
127
4.3.8 Summary
On the basis of this data, we can achieve film thicknesses of 7 - 8 m at
speeds higher than 90m/min using a gap of 25 m with liquids of viscosity of 7mPa.s.
Clearly films of 5 m at speeds higher than 90m/min can be achieved by decreasing
the gap below 25 m which is problematic because the rollers will touch. In addition
the smaller the roll gap, the bigger the fluctuation of film thickness caused by roll
accuracy. Hence another coating system will be needed. Forward deformable roll
coating is known to produce thinner films at high speeds so it may also be possible to
use deformable rollers in the reserve mode of operation. As this has never been tried
before, there is a clear interest in assessing this possibility.
4.4 ANALYSIS BY MEASUREMENT WITH TRANSPARENT ROLLER
We now analyse further the data obtained with steel-steel rollers using
observations made with steel-transparent roller. As mentioned earlier the accuracy of
the Perspex roller is not very precise, however it is still useful to observe the flow in
the meniscus region in order to form a learned view of the data on film thickness and
instability. Since the accuracy of the Perspex roller is about ±25 m, setting the roller
gap to 25 m could cause a clash, hence the choice of a 50 m gap is made for these
experiments.
Figure 4.45(a) shows the metered film thickness as a function of the roll
speed ratio at the applicator roller speed of 90m/min derived from Figure 4.4. The
128
positions of the meniscus captured by the camera are plotted in Figure 4.45(b). [The
meniscus position zero indicates position at the centre of the rollers.] It can be seen
that as the speed ratio is increased, the downstream meniscus moves upstream and
the upstream meniscus moves downstream. As a result, the coating bead or flow
region narrows. At the roll speed ratio at which minimum metered thickness is
achieved, the downstream meniscus position is zero. This result is in good agreement
with the data reported by Coyle et al. (1990a).
Figures 4.46(a)(b) show the effect of applicator roller speed on the minimum
metered film thickness and the meniscus positions. The roll speed ratios in these
graphs are identical to that at which the minimum metered film thickness can be
achieved. Therefore the downstream meniscus positions at any applicator roller
speeds are zero. On the other hand the upstream meniscus position shifts toward the
centre of the rollers as the applicator roller speed is increased.
4.4.2 Effect of roll gap
Figures 4.47(a)(b) show the effect of roll gap on the minimum metered film
thickness and the meniscus positions. There are no big differences in the meniscus
positions at the various gaps tested.
129
(a) Metered film thickness (b) Meniscus positions
Figure 4.45: Effect of roll speed ratio on meniscus positions. Liquid: Millmax5
Roll gap=50 m, Supply thickness

thickness=100 m
Applicator roller speed

speed=90m/min
(a) Minimum metered

etered film thickness (b) Meniscus positions
Figure 4.46: Effect of applicator roller speed on meniscus positions.

positions
Liquid: Millmax5, Roll gap

gap=50 m, Supply thickness=100 m
Vm/Va:: The ratio at which minimum thickness can be achieved
130
(a) Minimum metered
Figure 4.47: Effect of roll gap on meniscus positions.
Liquid: Millmax5, Supply thickness

thickness=200 m, Applicator roll speed=90m/min
speed
Vm/Va= The ratio at which minimum thickness can be achieved
As revealed in the previous section, with flooded nip, the supply film
thickness has no effect on the minimum metered film thickness (see Figure 4.48(a)).
As the supply film thickness is reduced, the upstream meniscus position moves
toward the centre of rollers and the downstream meniscus stays at the same positio
position
which is the centre of the rollers. From this data, it can be concluded that only
downstream meniscus position decides the metered film thickness.
131
(a) Minimum metered
Figure 4.48: Effect of supply film thickness on meniscus positions.

positions
Liquid: Millmax5,, Roll gap

gap=100 m, Applicator roll speed=30m/min
0m/min
Vm/Va= 0.9
4.4.4 Effect of viscosity
Here, the effects of viscosity are presented in Figures 4.49-

4.49 4.51 which show
that the movement of the meniscus position with roll speed ratio is more sensitive
with higher viscosity liquids. This result explain

explains why the speed ratio at which the
minimum metered film thickness is achieved shifts to lower value as the viscosity is
increased.. The reason why the meniscus position moves more sensitively with higher
viscosity fluids than lower viscosity fluids is because the shear stress is
comparatively higher
higher.
132
Millmax

thickness=100 m

speed=90m/min
Millmax

thickness=100 m

speed=90m/min
133
Figure 4.51: Effect of viscosity on meniscus positions.
Roll gap=50 m, Supply thickness=100 m, Applicator roller speed=90m/min

speed
Interestingly in the previous section with steel rollers about the observations
of wet film instabilities on the applicator roller after metering, it was found that the
roll gap and the viscosity have a complicated

mplicated effect on the coating window. In other
words, inn the case of low viscosity liquid

liquids (7mPa.s), small roll gaps are better for
getting stable films,, whereas in the case of high viscosity liquids

liquid (more than
30mPa.s), large gaps are required. To understand the flow in the bead, pictures were
taken and compared at various viscosities and roll gaps as shown in Figures 4.52,
4.53 at applicator roller speeds of 90 and 120m/min respectively.

respectively The roll speed
ratios were set at which the minimum film tthickness

hickness could be achieved. These
conditions are identical for most stable flow situations. Inn the case of low viscosity
liquids (7mPa.s), the downstream meniscus become

becomes more stable with smaller roll
gaps, whereas in the case of high viscosity liquid

liquids (>30mPa.s),
0mPa.s), it becomes more
stable with larger gaps

gaps.. This trend is the same as the observations of the wet films on
the applicator roller. This proves that ribbing on the applicator roller was caused by
134
the meniscus formation, not bubble stagnations at the me
meniscus
niscus region. In addition,
addition
we can conclude from this observation that there is a different mechanism of ribbing
at high speeds with low viscosity liquids. With high viscosity liquids,
liquids the cause of
ribbing is the high level of the viscous force and this is well established. With low
viscosity liquids at high speeds

speeds,, inertia rather than viscous forces increase and they
become the destabilising forces. This difference of ribbing mechanism will be
discussed further in section 4.4.6 below.
Figure 4.52: Pictures of meniscus regions with various viscosities at 90m/min
Figure 4.53: Pictures of meniscus regions with various viscosities at 120m/min
135
In this section, observations at high surface tension (with a solution of water
and glycerine)) are presented in Figure 4.54 and compared in Figure 4.55
4.5 with those
of lower surface tension. It is found that there were no major differences in both the
downstream and the upstream (except the two points at very low speed ratios)
meniscus positions. As for stability, whereas for the 27mN/m, the picture of the
meniscus clearly showed ribbing, for the 65mN/m, the images wer
were less clearly
defined (Figure 4.56),

), suggesting smaller wavelength ribbing.
Figure 4.54: Effect of roll speed ratio on meniscus positions Liquid:

Water/Glycerine

thickness=100 m
speed=90m/min
136
Figure 4.55: Effect of surface tension on meniscus positions

thickness=100 m, Applicator roller speed=90m/min
speed
Figure 4.56: Pictures of meniscus regions with various surface tension at 120m/min
4.4.6 Ribbing mechanism at high speeds
As presented earlier, there is a different trend on the effect of gap on ribbing
instabilities. Inn the case of low viscosity liquids there is a stable coating window at
small gaps but with high viscosity liquids

liquids, the stable coating window is at large gaps.
In order to analyse this

his result, Ca for each condition were calculated and plotted as
function of roll gap in Figures 4.57(a) – (d) for all the liquids tested: Millmax5
137
(7mPa.s, 27mPa.s), Water/Glycerine (6mPa.s, 65mN/m), Millmax15 (30mPa.s,
28mN/m) and Millmax68 (180mPa.s, 29mN/m). [In these graphs “stable” means that
there is a certain stable area in a certain roll speed ratio area and “unstable” means
that there are no stable areas at any roll speed ratios].
In the case of high viscosity liquids (more than 30mPa.s) (see Figures 4.57 (c)
(d)), we can see the same trend as in previous research on forward roll coating
studies, i.e. large gaps and small Ca are better for obtaining stable wet films. On the
other hand in the case of low viscosity liquids (see Figures 4.57 (a) (b)) it is found
that there is an opposite trend for roll gap, i.e. small gaps and small Ca are suitable.
If we insert the previous four figures (Figures 4.57 (a) – (d)) into one single
graph, we cannot see any border line (see Figure 4.58). Clearly, ribbing in this
6.6mPa.s case is being controlled in a way different from that normally encountered,
i.e. not by Ca or the capillary forces. We are thus justified in assuming that in this
situation, the other forces at play- inertia forces- are the destabilizing factor. As in all
cases, surface tension stabilizes the flow, the Weber number, We (Inertial forces/
surface tension forces) is the appropriate parameter to consider as an alternative to
Ca. To test this hypothesis, we processed the data under all conditions (i.e. including
low and large viscosities) with respect to Ca in Figure 4.58, Re in Figure 4.59, and
We in Figure 4.60. The better fit with We proves indeed that it is the case that in
reverse roll coating it is We and not Ca that is the control parameter of ribbing
instabilities. The best fit correlation of all the data describing the onset of instability
is given by:
.
2.35 10 (4.4)
138
(a) 7mPa.s, 27mN/m (b) 6mPa.s, 65mN/m
(c) 30mPa.s, 28mN/m (d) 180mPa.s, 29mN/m
Figure 4.57: Ca number as a function of roll gap with various liquids (individuals)
139
Figure 4.58: Ca number as a function of roll gap with various liquids (all in one)
Figure 4.59: Re number as a function of roll gap with various liquids (all in one)
140
Figure 4.60: We number as a function of roll gap with various liquids (all in one)
4.5 MEASUREMENT WITH EFORMABLE ROLLER
As stated in the literature review chapter, reverse deformable roll coating may
have a potential to achieve very thin film at high speeds,, however this coating
method has not been studied so far. Therefore this study is the first attempt at
understanding the fluid mechanics of deformable reverse roll coating.

coating
Following from the data just described with rigid roll coating, the minimum
film thickness
hickness obtained was 7 - 8 m at speeds higher than 90m/min using a gap of
25 m, and liquids of viscosity 7mPa.s. Clearly, films of 5 m at speeds higher than
90m/min can be achieved by decreasing the gap below 25 m but this is problematic
because the rollers will touch. In addition the smaller the roll gap, the bigger the
fluctuation of film thickness caused by roll tolerance. When it comes to coating
window, there are stable coating window at high speeds in some cases, however they
are very narrow. Hence in order to achieve thin films which are in wide stab
stable
141
coating windows at high speeds, the need to develop an alternative coating system,
with deformable reverse roll coating proposed here.
As described in the experimental chapter, nitrile based rubber materials (30,
50 and 70 shore A hardness) were selected as the deformable layer (15mm) covering
the core of the steel roller giving an overall accuracy in the roller diameter of ±25 m.
It should be noted that any damage on rubber rollers could not be observed by naked
eye even after the experimental study.
The data here are given for three liquids of viscosities 7, 30 and 180mPa.s
tested with a deformable rubber roller of 70A shore hardness at constant negative gap
of -100 m and with a supply film thickness of 100 m.
Figures 4.61(a), 4.62(a) and 4.63(a) show that the metered film thickness, hm
measured decreases continuously down to almost zero as the roller speed ratio, rs (=
Vm/Va) increases, unlike in the rigid reverse roll coating data presented earlier where
an up-turn is observed at a critical speed ratio, rs*. This is observed for the three
applicator roll speeds tested, 30, 90 and 150 m/min. With regard to the stable coating
window, there are wider ranges than in rigid roller (compare Figures 4.61(b) 4.62(b)
and 4.63(b) with Figures 4.3(b), 4.4(b) and 4.5(b)). [In the figure of coating window
(b) ‘Zero’ means that there is no film on the applicator roller after metering].
Interestingly, now film thinner than the 7 - 8 m limit observed with rigid reverse roll
coating are obtained at high speed (2 m at 90-150m/min).
The data in Figures 4.61(a), 4.62(a) and 4.63(a) also show that as the
applicator roller speed increases, the metered film thickness also increases, more so
142
comparatively with increasing viscosity. This can be eexplained
xplained in relation to the
higher viscous forces enlarging the coating gap as the speed and or the viscosity is
increased.
Figure 4.61: Effect of roller speed. Liquid: Millmax5
Supply film=100 m,, Roll gap

gap= -100 m, Rubber shore hardness=70A
=70A

=70A
143

=70A
4.5.2 Effect of Viscosity
In this section, the data just presented are transformed into figures at various
viscosities (Figures
Figures 4.64, 4.65 and 4.66
4.66). [The vertical short lines in the figures
indicate the border between the ribbing and stable areas]. We can see that as the
applicator speed (the actual coating speed) is increased, the favourable wider stable
window enables very thin films to be observed, particularly with the low viscosity
liquid. For example, Figure 4.6

4.66 shows stable film thickness 2 m and lower being
achievable at 150m/min with the 7mPa.s. fluid and indeed at viscosity higher (30 and
180mPa.s).
As stated before, the film thickness increases as high viscosity liquids are
used. The advantage of using high viscosity liquids is that the stable border extends
to lower roll speed ratio. Also, there is no cascade instability.
144
Figure 4.64: Effect of Viscosity. Applicator roll speed=30m/min

=70A

=70A
145
0m/min

=70A
4.5.3 Effect of rubber hardness
Here the effect of rubber elasticity ((shore

shore hardness 30A, 50A and 70A)
70A is
presented in Figures 4.6

4.67, 4.68 and 4.69 for the three liquids tested (7,
( 30 and 180
mPa.s). In all cases, as the rubber hardness is soft, the film thickness increases,
increases more
so with increasing applicator speed and viscosity on account that the viscous load
increases and enlarges the coating gap. With regard to the border line between
ribbing and stable area

areas, we observe no major differences
ences with the three types of
rollers. In some cases (for example Figure 4.6

4.67(c)), it may seem that the border with
70A shore hardness is located at low

lower roll speed ratios. However this is probably
due to the difficulty of picking exactly the onset of ribbing with the naked eye when
the film thickness is so low, less than 2 m.
146
(a) 30m/min (b) 90m/min
(c) 150m/min
Figure 4.67: Effect of Rubber hardness

hardness. Liquid: Millmax5 (7mPa.s)
Supply film=100
100 m, Roll gap
gap= -100m
147
(c) 150m/min

mPa.s)
Supply film=100
00 m, Roll gap
gap= -100m
148
(c) 150m/min

mPa.s)
Supply film=100
100 m, Roll gap
gap= -100m
4.5.4 Effect of roller gap
Here gap effects are assessed with negative gaps set from zero to -300 m on
the 30A shore hardness rubber roller using three liquids (7, 30 and 180 mPa.s) and
applicator roller speeds of 30, 90 and 150m/min. Figures 4.70 – 4.78 show the
149
corresponding data for the variations of film thickness with speed ratio and the
coating windows. In all cases and as expected the film thickness decreases as the
negative gap iss reduced from -50 to -300 m. Except for the zero gap, there are no
cascade instabilities.
Figure 4.70: Effect of roller gap. Liquid: Millmax5, Applicator roll speed=30m/min
speed
Supply film=100 m,, Rubber shore hardness=30A
150
Figure 4.71: Effect of roller gap. Liquid: Millmax5, Applicator roll speed=90m/min
speed
Figure 4.72: Effect of roller gap. Liquid: Millmax5, Applicator roll

speed=150m/min
151

speed=30m/min

speed=90m/min
152

speed=150m/min

speed=30m/min
153

speed=90m/min

speed=150m/min
154
In the data so far presented, the supply film thickness in the deformable mode
was fixed at 100 m. Here, we present data to assess the effect of varying the supply
film thickness from 200 to 25 m, where possible (with Millmax68 of viscosity
180mPa.s at 150 m/min, 25 m supply film thickness is insufficient).
Figure 4.79 gives the data for the condition in which the rubber deformation
is thought to be largest i.e. with the softest deformable roller (30A shore hardness)
and the highest viscosity liquid (180mPa.s). Differences are measurable only for
applicator roll of 90m/min at speed ratio less than 0.2, suggesting lower film
thickness with the 25 m supply film. When the shore hardness is increased (50A),
the same observation can be made. However a further increase in shore hardness
(70A), shows no difference in the effect of supply film thickness, although the actual
value of the film thickness in the sensitive region near zero speed ratio has now
decreased from 42-47, 27-32 and 15 when the shore hardness was increased from 30,
50 to 70A , respectively. Noted also, that these measurable differences are only in
the unstable region, i.e. below the critical speed ratio. Therefore we can conclude that
the supply film thickness has practically little effect (Figures 4.79-4.82).
155
(c) 150m/min
Figure 4.79: Effect of supply film thickness. Rubber shore

hore hardness=30A,
hardness Liquid:
Millmax68
Roll gap=-100
00 m(negative)
156
(c) 150m/min

hore hardness=50A,
hardness Liquid:
Millmax68
Roll gap=-100
00 m(negative)
157
(c) 150m/min

hore hardness=70A,
hardness Liquid:
Millmax68
Roll gap=-100
00 m(negative)
158
(c) 150m/min

hore hardness=30A,
hardness Liquid:
Millmax15
Roll gap= -100

00 m(negative)
4.4.6 Effect of surface tension (negative gap)
Figures 4.83 (a)(b)(c) show the effect of surface tension on the metered film
thicknesses with the liquids which have the viscosity of 7mPa.s and different surface
tensions of 27 and 65mN/m at the applicator roller speed of 30, 90, 150m/min
159
respectively. According to these data, there are no major differences in the metered
film thickness and this trend is the same as in the rigid reverse roll coating results.
Figures 4.84 (a) (b) show the effect of surface tension on the coating windows with
the liquids of surface te

tensions of 27 and 65mN/m respectively. In the case of high
surface tension liquid (65mN/m), the border between the ribbing and the stable areas
is located at lower roller speed ratio making the coating window wider and the
maximum stable film thicker. The ttrend

rend is the same as in the rigid reverse roll
coating, which is due to the meniscus getting stable with higher surface tension
liquid.
160
(c) 150m/min
Figure 4.83: Effect of surface tension on the metered film thickness.

thickness
Liquid: Millmax5 (7mPa.s, 27mN/m), Water/Glycerine ((6mPa.s,

mPa.s, 65mN/m)
Supply film=100
100 m, Roll gap
gap= -100 m (negative), Rubber shore
hore hardness=30A
hardness

speed=(a) 30m/min, (b) 90m/min, (c) 150m/min
(a) 27mN/m (b) 65mN/m
Figure 4.84: Effect of surface tension on the coating window.
Supply film=100
100 m, Roll gap
gap= -100 m (negative), Rubber shore
hore hardness=30A
hardness
Liquid: (a) Millmax5 (7mPa.s, 27mN/m), (b) Water/Glycerine ((6mPa.s,

mPa.s, 65mN/m)
161
4.5.7 Stable condition
In this section the conditions in which stable wet films can be achieved in the
deformable reverse roll coating at the negative gap of -100 m with various viscosity
liquids and rubber rollers of different rubber hardness are summarised.
Table 4.1 shows the stable minimum and maximum wet film thicknesses at
various conditions. The minimum thickness means the possible film thickness in this
study and may not mean the actual minimum film thickness because there is a
limitation of the roll speed. For example at 150m/min, the maximum roll speed ratio
equals Vm/Va=1 due to the difficulty in increase the roll speed more than 150m/min
in the coating rig used in this study. The maximum thickness means the film
thickness above which ribbing instability occurs at the lower roll speed ratio.
According to Table 4.1, with higher rubber hardness, lower viscosity and lower
speed (equivalent to low hydrodynamic forces at the nip region), the maximum film
thickness is very small and the range of the stable film thickness is narrow. At the
lowest hydrodynamic force condition (i.e. shore hardness 70A, 7mPa.s liquid and
30m/min), too small to deform the gap and allow flow, no stable film can be
achieved. On the other hand, using stronger hydrodynamic force conditions (i.e.
softer rubber roller, higher viscosity liquid at higher speed), the maximum film
thickness increases and the range of stable film that can be achieved becomes wider.
162
Table 4.1: Stable conditions in deformable roll coating
163
4.5.8 Effect of roller speed at Positive Gap
In the previous sections, data on the deformable reverse roll coating in the
negative gap conditions were presented and discussed. Now, data at positive gap are
presented noting the limited accuracy of the gap in this case causing much
fluctuation in the film thickness measured. Although, in industrial practice, such set-
up will not be tolerated, for the purpose of fundamental analysis, these additional
data will be useful to complement the study and provide data for future mathematical
modelling.
Figures 4.85-4.87, 4.88-4.90 and 4.91-4.93 show the data with the 70 A shore
hardness roller for the three liquids tested, Millmax5 (7mPa.s), Millmax15 (30mPa.s)
and Millmax68 (180mPa.s). The forms of the curves are identical in both the rigid
and deformable modes, as expected as these situations are essentially similar except
for the fact that the viscous loads, when high (high speed and high viscosities), will
enlarge the gap through pressing on the deformable rubber. Differences in the film
thickness values will follow accordingly. Interestingly, the data indicate that in
certain conditions, the loads are so high that the differences in film thickness are
significant suggesting that rubber deformation indeed happen even at positive gap.
As for the coating window, interestingly in all cases it is wider in the deformable roll
coating than that in the rigid roll coating. This can be attributed to the comparatively
reduced pressure resulting from the roller deformation. As stated in the literature
survey, in relation to both forward and reverse roll coating, ribbing occurs as a result
of the large pressures and pressure drops developing at exit of the nip. Any
reduction will relieve the instability and this has been found by Carvalho and Scriven
(1997) in their analysis of forward deformable roll coating.
164
(a) Roll gap: 50 m, 30m/min (b) Roll gap: 100 m, 30m/min
(c) Roll gap: 50 m, 90m/min (d) Roll gap: 100 m, 90m/min
(e) Roll gap: 50 m, 150m/min (f) Roll gap: 100 m, 150m/min

Figure 4.85: Film thickness with steel and deformable rollers at positive gaps
gaps.
Liquid: Millmax5
165
(a) eformable roller (b) Steel roller
Figure 4.86: Coating window with steel and deformable rollers at positive gap
gap.
Roll gap=50 m, Liquid: Millmax5
gap.
166

gaps.
Liquid: Millmax15
167
gap.
Figure 4.90: Coating window with steel and deformable rollers at positive gap.
168

gaps.
Liquid: Millmax68
169
gap.
Figure 4.93: Coating window with steel and deformable rollers at positive gap.
gap
170
4.5.9 Effect of hardness (positive gap)
In the previous section 4.5.8, a deformable roller of shore hardness 70A was
used. In this section, data from two other rubber rollers (shore Hardness 30A and
50A) are presented to assess the effect over the range on both the metered film
thickness and the coating window as shown in Figures 4.94, 4.95 and 4.96. Again
the trend is similar to the data with the rigid rollers but here with increased rubber
elasticity and viscous forces (higher applicator speed and increased viscosity), the
gap becomes wider increasing thus the film thickness and relieving pressures and
pressure drops thus ribbing instability. This effect is clearly seen with the highest
viscosity liquid (Millmax 68) and applicator roll speed of 150m/min. Again, the
importance of the data here is that they provide the actual values which are necessary
to assess mathematical models. They also explain the role of pressure in relation to
the ribbing phenomenon.
171
(a) 30m/min
(b) 90m/min
(c) 150m/min
Figure 4.94: Effect of rubber hardness
hardness. Liquid: Millmax5

gap=50 m
172
(a) 30m/min
(b) 90m/min
(c) 150m/min


gap=50 m
173
(a) 30m/min
(b) 90m/min
(c) 150m/min


gap=50 m
174
4.5.10 Effect of roll gap (negative and positive gap)
In this section the film thicknesses obtained with the rigid system are
compared with the deformable system which had the softest, most elastic roller (30A
shore hardness), over the complete range of viscosities and applicator speeds, in both
positive and negative gaps settings. The purpose is to assess the extent of the film
thickness reduction, important in the context of the aim of this thesis which is to
produce the thinnest film at the highest speed. The data are presented in Figures 4.97,
4.98 and 4.99.
In the case of the rigid system, the film thicknesses are the minimum film
thickness measured [Recall that in this case the film thickness exhibit a minimum at
the critical speed ratio]. As shown in the data, in the positive gap setting, the
deformable system also exhibits a minimum and this thickness is used here. The
deformable system in the negative gap setting mode however does not exhibit a
minimum. Rather, as the data showed, the film thickness continuously decreased. So
for the purpose of comparison, the film thicknesses at the same speed ratio as those
in the rigid system and deformable positive gap system were used.
The observations from this comparison are as follows. In the positive gap
settings, the slope of the lines film thickness vs. gap are steep, indicating that small
changes in the gap setting will cause large changes in the film thickness. However, in
the negative gap setting, the slopes are flatter, indicating less sensitivity of the film
thickness with small variation in the gap setting. Carvalho (2003) made a similar
observation in his predictive values but for forward deformable roll coating (see
Figure 2.31). As the uniformity of films is an important requirement in industrial
application, the suggestion is that reverse deformable roll coating is a good system to
175
achieve this, in addition as discussed earlier to offering a wider stable coating
window.
Figure 4.97: Effect of roll gap (negative and positive)

positive). Applicator roll
speed=30m/min
Vm/Va= 0.9 (7mPa.s), 0.6 (30mPa.s), 0.3 (180mPa.s)
176
speed=90m/min
Vm/Va=0.7
0.7 (7mPa.s), 0.4 (30mPa.s), 0.2 (180mPa.s)

speed=150m/min
Vm/Va=0.7 (7mPa.s), 0.3 (30mPa.s), 0.2 (180mPa.s)

177
4.5.11 Conclusions
The data presented on deformable roll coating in the negative gap setting
demonstrate that it is possible to achieve very thin wet films, less than 10 m at
150m/min. Using low viscosity liquid (7mPa.s), with the softest rubber roller (30A
shore hardness), a stable wet film ranging from 1.2 to 4.6 m at 150m/min can be
obtained. With relatively high viscosity liquids of 30mPa.s, stable wet films ranging
from 1.5 to 15.7 m at 150m/min are achieved. Moreover it was found that the
coating window at high speeds in the deformable reverse roll coating was
significantly wider than that in the rigid reverse roll coating.
One of the biggest problems in industry is that there are fluctuations of wet
film thickness in the machine direction due to the eccentricity of the steel rollers. The
data show that deformable roll coating, in the negative gap setting, exhibits a flatter
variation of the film thickness with gap than in the positive gap setting. This is an
advantage when considering fluctuation in film thickness.
Finally and importantly in relation to research, these data are the first as no
previous study on reverse roll coating has been published. Because they were
collected over a wide range of conditions, they will provide theoretical modellers the
information necessary to assess their predictions.
4.6 LIQUI TRANSFER FROM APPLICATOR ROLLER TO PET FILM
In this section, the transfer of the film formed on the applicator roller to a
substrate is analysed using experimental data collected on an industrial line as
described in the Experimental Method chapter. The merit of these data is that they
178
replicate industrial practice of typical roll coating systems. In most if not all
previous studies of roll coating, the analysis was always limited to studying film
formation on the applicator roller only. These data offer an insight on the transfer
mechanism
anism particularly to check the premise that if the film is stable on the
applicator roller it will remain stable upon transfer to the substrate.
Before conducting the experiments, the correlation between Ca and
dimensionless minimum film thickness on the applicator roller was checked and it
was confirmed that the data was in good agreement with the data taken at Bradford
coating lab (see Figure 4.100). Also, the co

correlation
rrelation between Ca and roll speed ratio
at which minimum film can be achieved was checked and it was found that the data
taken at Toyobo was in good agreement with the data taken at Bradford (see Figure
4.101).
Figure 4.100: Minimum Film thickness measured in this study
(¨ at Bradford, at Toyobo) and comparison with data of Coyle et al. (1990) (D).
179
Figure 4.101: Critical speed ratio vs. Ca: data from this study
(¨ at Bradford, at Toyobo) and comparison with Coyle et al. (1990) data (D) and
Kang and Liu (1991) correlation ((-------).
4.6.1 Effect of viscosity
Figures 4.10
4.102-4.104 show the effect of speed ratio on instabilities at three
viscosity values of 3, 44 and 105mPa.s and a surface tension of 26mN/m. These
conditions were designed to replicate the experiments done on the coating rig using
lubricating oils. The values are thus marginally different as the coating fluids used
for the transfer experiments required non

non-lubricating
lubricating oils. The observations from the
data indicate that with the low viscosity fluid (3mPa.s), a stable flow was obtained
only at low substrate speed of about 30 m/min and a speed ratio in the range 0.8
0.8-1.0.
The speed ratio is defined as PET film/Applicator roll (F/A). As the viscosity was
increased to 44mPa.s, the substrate speed in the stable regime could be increased to
60m/min. The corresponding speed ratio for stable flow was now reduced to the
range 0.7-0.9.
0.9. A narrow stable regime was also located for this viscosity when the
180
substrate speed was set to 90m/min and the speed ratio set to nearly 1. Upon further
increase in viscosity to 105mPa.s, a stable flow was observed at speed ratio of 1 for
substrate speed in the range of 60

60-90m/min.
Figure 4.102: Effect of PET film / Applicator roll speed ratio on wet film
instabilities. Viscosity
Viscosity=3mPa.s, Surface tension=26mN/m
181
Figure 4.104: Effect of PET film / Applicator roll sspeed ratio on wet film
The stability of the flow on the substrate was further investigated using
visualisation experiments. The images are shown in Figures 4.10

4.105 and 4.106. They
show large bubbles and “white” lines of aggregated small bubbles indicative that the
instabilities are not surface based but dynamic wetting related and strongly sensitive
to the speed ratio substrate/applicator speed. As anticipated, the substrate speed,
above a certain speed

peed will fail to wet the fluid and entrain air in the form of streaks.
This is observed in Figure 4.10

4.105 at speed ratio of 0.71. However, when the speed
ratio is further increased, the streaks instead of increasing diminish in severity, until
they almost disappear

isappear at speed ratio of 1, suggesting bubble size get smaller or
stagnation of large bubbles being prevented as the web speed is increased. Further
increases in the speed ratio towards 1.25 now cause cascade suggesting the coating
bead has moved passed the mid-point

point when the applicator and the substrate are in
contact.
182
Figure 4.105: Pictures of transfer regions at various F/A ratio
Viscosity=3mPa.s,
mPa.s, Surface tension
tension=26mN/m,, Applicator roller speed=60m/min
speed=60m/mi
Figure 4.106: Pictures of transfer regions with different viscosity liquids.
Surface tension= 26mN/m

mN/m, Applicator roller speed=60m/min
183
In attempt to understand the film transfer mechanism further, the meniscus
position was measured and the data are shown in Figure 4.107-4.110 for a range of
operating conditions. The data clearly show the importance of both the speed ratio
and the applicator speed in controlling the meniscus positions, particularly upstream.
Whereas downstream, the position remains relatively insensitive with speed ratio and
applicator speed, the upstream position varies considerably from being negative (far
away at the inlet to zero and above, i.e. past the mid-position, as the speed ratio is
increased.
As for the effect of viscosity, Figure 4.106 provides the relevant data
showing that with low viscosity liquids, large bubbles are formed and stagnate
downstream with air entrainment upstream being relatively low. However with high
viscosity liquids, no such large bubbles stagnated downstream and air entrainment
upstream was comparatively greater. This is consistent with air entrainment
observation in dip coating for example where we see that air entrainment occurs at
much lower speed with high viscosity liquids in comparison with low viscosity
liquids.
184
Figure 4.107: Effect of PET film / Applicator roll speed ratio at various applicator
roll speed on meniscus positions.
Viscosity: 3mPa.s,
mPa.s, Surface tension: 26mN/m
Figure 4.108: Effect of viscosity on meniscus positions

positions. Applicator
licator roller
roll speed:
30m/min
185
positions. Applicator roller
roll speed:
60m/min

positions. Applicator roller
roll speed:
90m/min
With regard to transfer ratio of the film from the applicator roller to the substrate, the
relevant data are described in Figure

Figures 4.111-4.113 showing, as expected the
importance of the speed ratio (through mass balance). The transfer ratio is defined as
186
transferred wet film on the web / wet film on the applicator roller. As the speed ratio
increases towards 1, th
thee ratio tends to 1 and below speed ratio of 1, the variation is
broadly linear (subject to experimental errors) but shifted downwards with increasing
applicator roll speed.
ratio
Viscosity=3mPa.s,
3mPa.s, Surface tension
tension=26mN/m
ratio
Viscosity=44mPa.s,
tension=26mN/m
187
ratio
Viscosity=105mPa.s,
tension=26mN/m
Figure 4.1144 shows a coating window with a high surface tension liquid
which has the viscosity of 3mPa.s and the surface tension of 35mN/m. Coating
window with high surface tension liquid is wider than that with low surface tension
liquid (compare with Fi

Figure 4.102). Figure 4.115 shows the images of transfer
regions with the liquids of 3mPa.s viscosity but different surface tension (26,
35mN/m) at 60m/min. There was no large bubble at the downstream meniscus
despite existence of white lines with high surfa

surface
ce tension liquid. ownstream
meniscus with a high surface tension liquid locates slightly at more downstream
positions (see Figures 4.

4.116 -118).
). In the F/A ratio = 0.91 much air entrainment was
observed with 26mN/m liquid, on the contrary in the case of high surface tension
liquid (35mN/m), only few air entrainment was observed. The relatively stable
upstream meniscus and the location of the downstream meniscus are the possible
reason to prevent stagnations of large bubbles at downstream meniscus.
188
instabilities.
Viscosity=3mPa.s,
tension=35mN/m
189
Figure 4.115: Pictures of transfer regions with different surface tension liquids.
liquids
Viscosity=3mPa.s,, Applicator roller speed

speed=60m/min
190
Figure 4.116: Effect of surface tension on meniscus positions.
positions Applicator roller
speed=30m/min

speed=60m/min
191
speed=90m/min
ratio
Viscosity=3mPa.s,
tension=35mN/m
4.6.3 Effect of wrap angle
Figure 4.120
20 shows the images of transfer regions with the different wrap
angle of PET film against the applicator roller with 44mPa.s viscosity liquid. The
192
wrap angle is defined as in Figure 3.13. As the wrap angle is increased, which means
the PET film is pressed towards the applicator roller more, weak streaks were
observed. This is probably due to stagnations of relatively large bubbles at
downstream meniscus. As the wrap angle is decreased, air entrainment occurred
more often and much amount of small bubbles was seen at the downstream meniscus,
however the wet film was stable. From this observation, it can be said that
stagnations of small bubbles do not affect the wet film instabilities as long as the size
of bubbles is small. Figure 4.121 shows the meniscus positions as a function of the
wrap angle. Reasonably as the wrap angle is large, the distance between down and
upstream meniscus become long. In this study as a hold roller which is located below
the applicator roller was moved, the position of downstream meniscus changed but
upstream meniscus was at a same position. Figure 4.122 shows a transfer ratio as a
function of wrap angle. As the wrap angle is large, the transfer ratio is slightly
decreased. The mechanism is not clear at the moment.
193
Figure 4.120: Pictures of transfer regions at various wrap angle conditions
conditions.
Viscosity=44mPa.s,
speed
194
Figure 4.121: Effect of wrap angle on meniscus positions.
Applicator roll speed=

=60m/min, F/A ratio=0.83, viscosity=44mPa.s
44mPa.s
Figure 4.122: Effect of wrap angle on liquid transfer ratio.

44mPa.s
4.6.4 Effect of web tension
Figure 4.1233 shows the images of transfer regions in various web tension
conditions. There was no major difference in wet film instabilities. As the web
tension is increased, although air entrainment from the upstream meniscus decreased,
195
there was no major difference in the situation at downstream meniscus (Small
bubbles
ubbles were observed at downstream meniscus but large bubbles were not
observed.). Figure 4.12

4.124 shows the meniscus positions as a function of web tension.
There is no difference in meniscus positions with different web tensions. Figure
4.125 shows the liquid

id transfer ratio as a function of web tension. As the web tension
was increased, the liquid transfer ratio slightly increased. The mechanism of this
trend is not clear at the moment.
Figure 4.123: Pictures of transfer regions at various web tension conditions.

conditions
Viscosity=44mPa.s,
speed
196
Figure 4.124: Effect of web tension on meniscus positions.

=60m/min, F/A ratio=0.83, viscosity= 44mPa.s
Figure 4.125: Effect of web tension on liquid transfer ratio.

44mPa.s
4.6.5 Summary
In this section effects of liquid properties and coating rig configurations on
liquid transfer from an applicator roller to a PET film were studied. As main causes
197
of wet film instabilities, stagnations of large bubbles at downstream meniscus and air
entrainment from upstream meniscus were observed. To prevent stagnations of large
bubbles at downstream meniscus, high viscosity liquids have advantage. This is due
to a difficulty of high viscosity liquids occurring vortex at meniscus regions (see
Figure 4.126). To prevent air entrainment from upstream meniscus, low viscosity and
high surface tension are useful. The phenomenon that low viscosity liquids have
advantage for air entrainment is identical to the previous studies about air
entrainment in dip coating. In this study three types of liquids which have different
viscosities (3, 44 and 105mPa.s) were used and 44mPa.s liquid had the widest
coating window. In the case of the liquid of 3mPa.s, strong streaks were observed on
the wet film due to stagnations of large bubbles at downstream meniscus and in the
case of the liquid of 105mPa.s liquid, many bubbles were observed on the wet film
due to many bubbles were entrained from upstream meniscus. In the case of 44mPa.s
liquid, although a few bubbles were entrained from the upstream, there was no
stagnation of large bubbles at downstream meniscus. This is the reason why the wet
film was the most stable with 44mPa.s liquid. From the studies about wrap angle and
web tension, these factors also affect the air entrainments and stagnations at
meniscus. Therefore a suitable viscosity for achieving stable wet film depends on the
coating configuration such as web tension and wrap angle. As for surface tension,
high surface tension is useful to prevent air entrainment from upstream meniscus and
more stable wet film was achieved.
198
(a) Low viscosity (b) High viscosity
Figure 4.126: iagrams of transfer regions with low and high viscosity liquids
199
Chapter 5 : THEORETICAL CONSIDERATIONS
5.1 INTRODUCTION
The effort undertaken in this research was to approach the analysis of reverse
roll coating using experimental measurements. These are essential, not to only to
guide confidently successful operation but also provide theoreticians with the means
to check their predictions which in the case of coating flows are very challenging
thus probably not entirely reliable. The reasons are that coating flows have complex
features. First the flow begins at the dynamic wetting line where a discontinuity
exits: the substrate moves whilst the wetting line is stationary. Second, at exit a free
surface forms. Both features are mathematically complex and require much skill to
attempt to develop meaningful analyses. Work by Carvalho (2003) shows the
complex nature of such full analyses. Even with these apparently robust
mathematical treatments, assumptions are made, particularly on the wetting line,
usually assuming a certain contact angle or ignoring altogether the dynamic wetting
line (flooded inlet).
In this chapter, rather than attempting a full analysis (not in the scope of this
research), we present an approximate treatment based on the lubrication
approximations as these enable first estimates of film thickness which we can
compare with the data.
5.2 LUBRICATION MODEL FOR RIGID REVERSE ROLL COATING
The analysis begins with the usual Navier-Stokes equations and various
assumptions on the flow, mainly steady state, incompressible, one dimensional and
Newtonian. Figure 5.1 describes the flow and coordinates system.
200
Figure 5.1: Geometry of roll gap region
Navier-Stokes equation:
r( + u・ u ) = - P+ u + rg (5.2.1)
We consider only one component in the direction of “x”.
r( +u +v +w )=- + + rg (5.2.2)
Mass continuity equation:
+ ・(r u 0 (5.2.3)
Under the incompressible assumption in which density is a constant, the mass
continuity equation simplifies to a volume continuity equation:
・u = 0 (5.2.4)
If we assume the flow is only in one direction (x direction) and ignore the effect of
gravity and the ratio of the roll gap to radius is very small, i.e. h/R<<1, the complex
Navier-Stokes equation simplifies to:
= (5.2.5)
Velocity boundary conditions:
These invoke that the applicator and metering rollers are moving at speed uT and uB:
201
u= -uT at y=yT, u=uB at y=yB (5.2.6)
Here yT and yB are the positions of the walls of the rollers in the coordinates system
of Figure 5.1. They relate to the geometry of the gap as:
y = h + and y =-( h + ) with h=yT-yB= h + (5.2.7)
= being the inverse equivalent radius (5.2.8)
imensionless Variables
As recommended in the treatment of such analysis, we introduce the appropriate
dimensionless variables of speed, speed ratio, pressure, X and Y positions and gap H:
U= , = u /u , P = p €h /R , X = x/€Rh , Y=y/h , H = h/h =1+X2
(5.2.9)
imensionlessVelocity Profile:
Starting with Eq. (5.2.5), integrating twice and using the velocity boundary
conditions Eq. (5.2.6) leads to:
• ƒ„ •‰Š‹
U=‚ ƒ…
(† ‚ - (†‡ †ˆ ) Y + †‡ †ˆ ) – Œ
(† • †ˆ ) + 1 (5.2.10)
imensionless Flow Rate
Non dimensionalising the flow rate qL with respect a nominal drag flow that can be
taken by the roller through the gap (uB-uT) h0 then integrating the velocity Eq.
(5.2.10) with respect to Y between the top YT and bottom YB wall leads to:
•• ’• ƒ„ •
Ž• ‘ˆ ’‘‡ “”
Œ• + ‚ Œ (5.2.11)
•‚ •’Š‹ ƒ…
202
This equation can be rearranged to express the pressure drop as a function of flow
rate:
ƒ„ – ‚Ž•
ƒ…
= Œ‚ (1 - Š‹ ) • • Œ
) (5.2.12)
imensionless Pressure
Starting with the pressure drop Eq. (5.2.12) and integrating it with respect to X from
a position far upstream (-¥) where the pressure may be assumed to be atmospheric
(0) leads to the following pressure profile along the gap position X:
… • š … •… •
P(X)=6(1- Š‹ ) {( ‚ •‰… ‚ + ‚ —˜™’• X + › ) -2Ž• ( › •‰… ‚ ‚
+ œ •‰… ‚ + œ —˜™’• X +
•š
•–
)} (5.2.13)
Separation Boundary Conditions
As stated in the introduction of this chapter, the intention with this simple model is to
obtain an estimate of the film thickness formed. Therefore the above equation needs
to be solved for the flow rate. To achieve this requires the setting of separation
boundary conditions where the film forms. Two sets of conditions may be used and
are known as the Reynolds separation conditions (Reynolds, 1886) and the Prandtl-
Hopkins separation condition (Prandtl, 1937, Hopkins, 1957) used originally in
solving flow in lubrication bearings.
Reynolds Separation conditions:
These simply state that at the end of the flow both the pressure and pressure gradient
vanish, i.e:
P(X ) = 0, (X ) = 0 (5.2.14)
•
Applying the pressure drop conditions to Eq. (5.2.12) leads to:
203
X = €2 ž • 1 (5.2.15)
This separation position can then be substituted in the pressure Eq. (5.2.13). Using
then the pressure separation condition P(X ) = 0 leads to finding the dimensionless
flow ž from the following equation:
€ Ÿ ’ π
6(1 - ) (( ¡Ÿ
+ tan’ €2lž • 1 + ¡
)
€ Ÿ ’ £€ Ÿ ’ £ £π
- 2lž ( ¢Ÿ
+ ¢Ÿ
+ ¤ tan’ €2lž • 1 + ¢
)) = 0
A unique value lž independent of the speed ratio rS is then calculated by trial and
error giving:
lž = 0.613 (5.2.16)
¥
As lž is defined as ’
and qL as uBhB, the dimensionless film thickness is
given as:
HB=0.613(1-rS) (5.2.17)
This results states that the dimensionless film thickness decreases linearly with speed
ratio and has a value of 0.613 at rS=0. This result is surprisingly a very good
engineering estimate of the actual film thickness, particularly at low speed ratio and
in the regime before cascade instabilities set in. As the series of figures, Figure 5.2-
5.6 show, the data for a wide range of conditions (speeds, gap, viscosity and surface
tension) compare well with this simple prediction before the upturn where cascade
occurs provided the lubrication approximations are obeyed. For example Figure 5.4
shows a strong disagreement with the data labelled 25 and 50 m. This is because for
these sets of data, the incoming film thickness is smaller than the actual gap (100 m).
Also, the strong disagreement between this simple prediction and the data in the
cascade regime are to be expected as the assumptions of 1 flow no longer stand. In
204
the cascade regime the bead is narrow, the separation point invades the gap, past the
zero position and strong recirculation prevails.
Figure 5.2: Effect of Applicator roller speed

speed.
Supply film=50
50 m, Roll gap
gap=25 m.
Figure 5.3: Effect of the roll gap.
Liquid: Millmax5, Applicator Roll Speed=120m/min, Supply film=200

film m
205
Figure 5.4: Effect of the supply film thickness.
Liquid: Millmax5, Applicator Roll Speed=120m/min, Roll gap=100

100 m
Figure 5.5: Effect of the viscosity.
Applicator Roll Speed

Speed=120m/min, Supply film=200 m, Roll gap=50
gap m
206
Figure 5.6: Effect of the surface tension.
Applicator Roll Speed

Speed=120m/min, Supply film=100 m, Roll gap=50
gap m
In conclusion, we can state on the basis of this comparison, that the simple
lubrication approximations prediction provide a good estimate of the linear variation
of the film thickness with speed ratio for a very wide range of operating conditions
that comply with the lubrication approximations assumptions, i.e. flooded nip and
not in the cascade regime. And these are conditio

conditions
ns that are operational in practice.
The analysis also enables to obtain an estimate of how the pressure varies in the nip
(see Figure 5.7)) and where the flow ends, giving from Eq. (5.
(5.2.15)
15) and the value of
l just calculated (0.613), Xs = 0.507.
207
Figure 5.7: Predicted Pressure Distribution.
Prandtl-Hopkins Separation conditions:
These simply state that at the end of the flow the velocity, velocity gradient and
pressure vanish, i.e:
U(X ) = 0, (X ) = 0 P Xs 0 (5.2.18)
Applying these zero conditions to the original Eq. (5.

(5.2.10),
10), Eq. (5.2.12)
(5. and (5.2.13)
leads to a system of 3 equations with 3 unknowns Xs, Ys and l which in principle
can be solved. The actual values are best obtained by trial and error for various
values of the speed ratio, rS. The calculations show that no solution exists. After
further consideration, we observe that whereas these separation conditions are
plausible in forward roll coating, they cannot be in reverse roll coat

coating because of the
nature of the velocity profile making the velocity gradient at separation variable and
non-zero
zero (see Figure. 5.
5.8).
208
(a) Forward mode (b) Reverse mode
Figure 5.8: Velocity distribution in (a) forward and (b) reverse modes
We observed that at the critical condition the dynamic wetting line is located
at XD=0 as indicated in Figure 5.9. We can analyse this flow using the lubrication
model which gives the pressure at the stagnation point from (5.2.13) as:
P(0) = 6(1-rs) ( - ) (5.2.19)
We can now approximate P(Xs) using surface tension force, i.e.
P(0) = - (5.2.20)
Now we approximate the radius of curvature rc= 0.5(hs – hf)
hs: roll gap at the stagnation point, hf: final film thickness
If we assume hf<<hs,
P(0) = - (5.2.21)
We equate (5.2.19) and (5.2.21):
6(1-rs) ( - l )=-
We define = 6( - l )
(1-rs) = -
rs = 1 + 2 / ( Ca)
209
This result indicates that the critical roll speed ratio depends on dimensionless flow
rate, roll gap, roll radius and Ca number. Here we use roll radius equal 0.1m which is
the same as in this study’s experiments. We can draw the graph which shows critical
roll
ll speed ratio as a function of Ca number depending on dimensionless flow rate and
roll gap (see Figure5.10)

Figure5.10). Figure 5.10 also shows experimental data in this study.
There is a discrepancy between the theoretical result and the experimental data. This
is probably
robably because as the speed ratio increases the lubrication approximation model
fail hence the errors in predicting the critical speed ratio because the flow is clearly
2D not 1D.
Figure 5.9: Geometry at the roll gap region
Figure 5..10: Critical roll speed ratio from Lubrication model
210
5.3 LUBRICATION MODEL FOR DEFORMABLE REVERSE ROLL
COATING
The approach is similar to the one just described except that here the top
roller surface has a deformable layer which compresses under the action of pressure.
The deformation of this rubber layer of thickness b may be assumed to be linear and
described simply by Hooke’s law as:
b
 x   p( x)  k  p( x)
E (5.3.1)
E is the modulus of elasticity of the rubber layer and p is the pressure exerted by the
fluid. The actual gap h(x) at any position x along the flow must now account for the
initial gap set hi(x) and the deformation d (x) both of which vary also along the x
flow direction:
h( x)   ( x)  hi ( x) (5.3.2)
As in the rigid case, the initial profile hi(x) must be related to the geometry and the
negative gap set at hi (0). The geometrical correlation is simply (see Figure 5.1):
1 1 1 2 x2
hi ( x)  hi 0  (  ) x  hi 0  (5.3.3)
2 R Rr Re
Both Eq.(5.3.2) and Eq.(5.3.3) can now be made non-dimensional, using the same
approach as in the rigid case, i.e. with respect to the initial set gap hio to obtain:
b
H ( X )  ( X )  H i ( X ) with  p (5.3.4)
Ehio
x
Hi (X ) 1  X 2 with X (5.3.5)
Re hio
These combine into:
H    1 X 2 (5.3.6)
Note that the variation of H and with X has been dropped as it is clearly implicit.
211
The flow equation given earlier for the rigid case also applies here as it
describes the pressure drop through a nip flow, i.e.:
€•
= (1 - r ) 1 ) (5.3.7)
The system of equations to solve is now composed of Eq.(5.3.6) and Eq. (5.3.7). We
can again use the Reynolds separation conditions as they proved successful in the
analysis of the rigid case to find L as a function of the variables of the system.
First as with the rigid roller case, the condition that the pressure drop is zero
at the separation point Xs leads to:
X = ‚2l 1 (5.3.8)
Using Eq.(5.3.6), enables to locate the separation position XS as:
X = ÖH Lƒ1 (5.3.9)
These XS values can be applied together with the other separation boundary
conditions which states P(XS)=0. Integrating first Eq. (5.3.7) with the condition
upstream P(-¥) enables the pressure profile as:
€•
P(Xs)-P(-¥) = „ 1 r 1 with H    1  X 2 (5.3.10)
The Reynolds condition (P(Xs)= dP/dXs=0) is now been applied transforming Eq.
(5.3.7) to:
…† ‡
l = ˆ
Eq. (5.3.10) is non-linear and to solve it will require using a numerical technique.
The procedure to follow is in principle as follows:
1) Start at the inlet, assume a far enough position corresponding to -¥
2) Assume ‰Š hence calculate XS from Equation (5.3.10) knowing that both pressure
and pressure drop are zero at XS.
212
3) At this XS, the pressure should be zero and if not another ‰Š must be assumed and
the calculations repeated until convergence is obtained.
In the present research, the calculations were not performed and are part of the
recommendation for future work. The objective here was to demonstrate that in
principle a simple model can be formulated. Improved model can also be devised by
changing the way in which the deflection varies with pressure as it may not be
necessarily linear. Also the modulus of elasticity can change depending on the
applied load.
5.4 CONCLUSION
Despite the complexities of reverse roll coating, the simple lubrication
approximation model balancing pressure and viscous forces together with the
Reynolds separation conditions is able to provide a good estimate of the linear
variation of the film thickness with speed ratio for a very wide range of operating
conditions that comply with the lubrication approximations assumptions, i.e. flooded
nip and not in the cascade regime. These are conditions that are operational in
practice. The analysis also enables to obtain an estimate of how the pressure varies in
the nip and where the flow ends. Although not evaluated here, the same model can
be extended to deformable reverse roll coating.
It is important however to note that reverse roll coating in both rigid and
deformable modes is a complex flow especially at large speed ratios and it requires a
full 2D analysis which is mathematically involved and outside the scope of this
research. The data collected in this research however can serve to test any
sophisticated model developed as the range of operating conditions covered was
comprehensive extending to speed ratio 1 and above and gaps positive and negative.
213
Chapter 6 : CONCLUSIONS
6.1 INTRODUCTION
The aim of this predominantly experimental study was to investigate the
viability of reverse roll coating to produce good quality thin films (<10 micron) at
substrate speeds greater than 1 m/s (60 m/min). At the outset, it was clear that in
order to achieve these low thicknesses, the operation had to be run at low metering
gaps, less than 20 microns (rigid or positive gap reverse roll coating). Although,
there was a risk that roll clash could occur at these conditions, the operation was
nevertheless possible and a wide range of parameters were tested to produce stable
coating windows with the required low thickness and high speed. Also and in order
to validate the experimental method, operations at higher gaps were also conducted
and results compared with previous studies, in particular that of Coyle et al (1990a).
To push the film thickness further down, negative gaps operation was also tested
(deformable or negative gap reverse roll coating). This is the first study ever
attempted on reverse roll coating. As with positive gaps, a wide range of parameters
were tested to arrive at stable coating windows.
In addition to investigating rigid and deformable roll coating in the laboratory,
industrial trials were also conducted to assess the state of the films formed on the
rollers when they are transferred onto substrates, i.e. do they remain in the same
stable state or does the transfer process affect stability?
The conclusions and recommendations from these experiments are as follows.
214
6.2 RIGID REVERSE ROLL COATING
The effects of operating parameters are described in the tables below.
Viscosity plays a critical role particularly with regard to stability and we observe the
following relationships with the gap and speeds:
 In the case of low viscosity liquid (7mPa.s) at low speed (30m/min), the
coating windows at smaller gaps become narrower (see Figure 4.25(a)). In the
case of low viscosity (7mPa.s) at high speed (higher than 90m/min), there is a
stable coating window at small gap, however there is no stable coating
window at large gaps (see Figure 4.27(a)).
 In the case of high viscosity (higher than 30mPa.s) at low speed (30 -
90m/min), there is no major difference at the same speed condition (see
Figures 4.25(b)(c)-4.27(b)(c)). In the case of high viscosity (higher than
30mPa.s) at high speed (higher than 120m/min); although there is no stable
window at small gaps, there is stable coating window at large gaps (see
Figures 4.28(b)(c)-29(b)(c)).
In order to achieve thin film the following conditions are required:
(1) Low speed, not good for productivity
(2) Small roll gap, limited by roll accuracy
(3) Low viscosity, limited as already very low viscosity liquids are being used in
industry.
(4) High surface tension, slightly effective for reducing the film thickness.
(5) Supply film thickness, showing no effect on the minimum metered film
thickness.
215
Low speed (1), low viscosity (3) and high surface tension (4) equal low Ca number.
Therefore we can conclude that low Ca number and small gap are effective for
reducing film thickness.
In order to achieve wide stable coating window the following conditions are
required:
(1) Low speed, not good for productivity
(2) Roll gap: It depends on the liquid viscosity and roll speed. Small roll gap is
needed for low viscosity liquid at high speed. Large roll gap is needed for
high viscosity liquid at high speed.
(3) Viscosity: It depends on the roll gap and roll speed. Low viscosity is needed
for any gaps at low speed. High viscosity is needed for large gap at high
speed.
(4) High surface tension: It is useful to obtain a wider coating window.
However it may cause reduction of the wetting ability of a coating liquid onto
a substrate. Therefore it might be necessary to adjust the surface energy of the
substrate before the coating operation.
(5) Thick supply film thickness: If the supply film thickness is decreased to less
than the roll gap, the coating window becomes narrower.
According to dimensionless analysis, low Weber number and large gap are effective
at stabilising flow (eliminate ribbing) at high speeds. This means that low inertia and
high surface tension forces are effective at reducing ribbing at high speed in reverse
roll coating. However the criterion does not completely divide the stable and unstable
216
groups of data and therefore further analysis is needed to assess what other
parameters are at play in this complex flow.
On the basis of the data in this study, we can achieve 7 - 8 m at speeds higher
than 90m/min using a gap of 25 m, viscosity of 7mPa.s. Clearly films of 5 m at
speeds higher than 90m/min can be achieved by decreasing the gap below 25 m
which is problematic because the rollers will touch. In addition the smaller the roll
gap, the bigger the fluctuation of film thickness caused by roll accuracy. When it
comes to the coating windows, there are stable coating windows at high speeds in
some cases, however they are very narrow. Hence in order to achieve thin films
which are in wide stable coating windows at high speeds, another coating system will
be needed and this lead us to considering deformable reverse roll coating.
Table 6.1: Effect of various parameters on the film thickness and coating
window in rigid roll coating
Film thickness: - means thick, ¯means thin, ® means unaffected
Coating window: - means wide, ¯means narrow, ® means unaffected
217
Table 6.2: Effect of viscosity and roll gap on the film thickness and coating
window in rigid roll coating
6.3 DEFORMABLE REVERSE ROLL COATING
The effects of operating parameters are described in the tables below.
Table 6.3: Effect of various parameters on the film thickness and coating
window in deformable roll coating
By using a deformable roll as a metering roller, it is possible to produce films and
achieve wide coating windows at high speeds.
In order to achieve thin film the following conditions are required.
(1) Low speed: However it is also possible to make very thin film even at high
speeds
218
(2) Roll gap: Negative gap is needed to make very thin films and even if the
negative gap differs, the film thickness does not change dramatically. This
means the sensitivity of the roll gap difference is small for the metered film
thickness.
(3) Low viscosity: However it is also possible to make very thin film even with
high viscosity liquids.
(4) Surface tension: There is no major effect of surface tension on the film
thickness.
(5) Supply film thickness: There is no major effect on the metered film thickness.
(6) Hard rubber roller: However it is also possible to make very thin film even
using soft rubber roller.
On the basis of the data in this study, we can achieve stable wet film ranging from
less than 0.5 to 40 m by optimising liquid viscosity and rubber hardness of the
roller at high speed (150m/min).
In order to achieve wide stable coating window the following conditions are required.
(1) High speed: This is because the critical roll speed ratio between ribbing and
stable states becomes lower at high speeds and there are no cascade
instabilities at high roll speed ratio areas unlike in rigid roll coating. It should
be noted that even at low speeds it is possible to achieve a wide stable coating
window.
(2) Roll gap: There is no major difference at any negative gap condition. It is
possible to obtain a wide stable coating window at any negative gap (in this
study more than minus 50 m).
219
(3) High viscosity: This is because the critical roll speed ratio between ribbing
and stable states becomes lower with high viscosity liquids and there are no
cascade instabilities at high roll speed ratio unlike in rigid roll coating. It
should be noted that even with low viscosity liquids it is possible to achieve a
wide stable coating window.
(4) High surface tension: With high surface tension liquids the coating window
becomes marginally wider because the critical roll speed ratio between
ribbing and stable states becomes slightly lower with high surface tension
liquids and there are no cascade instabilities at high roll speed ratio unlike in
rigid roll coating. It should be noted that even with low surface tension
liquids it is possible to achieve a wide stable coating window.
(5) Supply film thickness: There is no major effect of the supply film thickness
on the coating window.
(6) Rubber hardness: There is no major effect of the rubber hardness on the
coating window.
On the basis of the data in this study, we can achieve wider stable coating windows
compared to rigid roll coating at high speed (150m/min).
In positive deformable roll coating, the following results were achieved.
 The film thickness in the deformable roll coating is bigger than that in the
rigid roll coating with high viscosity liquid even at positive gap settings. This
result indicates that rubber deformation occurs even at positive gap settings
by using deformable roll and high viscosity liquid.
 Also it is found that there is an effect of the rubber hardness on the metered
film thickness. With a softer rubber roller, thicker metered film thicknesses
are achieved.
220
 As for the coating window, interestingly in all cases it is wider in the
deformable roll coating than that in the rigid roll coating. This phenomenon is
probably due to the pressure in the nip region in the deformable roll coating
getting less than that in the rigid roll coating. As a result, the occurrence
ribbing may be prevented.
6.4 LIQUID TRANSFER FROM ROLL TO WEB
The effect of the transfer parameters are shown in the below table. What is
important here are both air entrainment and surface instabilities.
Table 6.4: Effect of various parameters on wet stability and transfer ratio at
transfer region in reverse kiss coating
In order to prevent air stagnation at downstream meniscus, the following conditions
are required.
(1) Low speed: However it is not good for productivity
(2) High F/A ratio: However at high speed ratio, air entrainment occurs and at too
large ratios, the downstream and upstream meniscus get close and the bead
breaks.
(3) Low wrap angle: As the wrap angle becomes too large, air stagnation occurs.
(4) Web tension: No major difference was observed in this study.
221
(5) High viscosity: As the viscosity is reduced, air stagnation occurs.
(6) High surface tension: There is still air stagnation at the downstream meniscus.
However probably due to the air entrainment being prevented at the upstream
meniscus, air stagnation reduces with high surface tension liquids.
In order to prevent air entrainment at upstream meniscus, the following conditions
are required.
(2) Low F/A ratio: However at low speed ratio air stagnation occurs.
(3) High wrap angle: As the wrap angle is increased, air entrainment is prevented.
(4) High web tension: As the web tension is increased, air entrainment is prevented.
(5) Low viscosity: As the viscosity is increased, air entrainment occurs.
(6) High surface tension: As the surface tension is increased, air entrainment is
prevented.
In order to make stable wet films, the following conditions are required.
(2) F/A ratio: There is a suitable ratio for obtaining stable films. When the ratio is
too small, air stagnation occurs and streaks occur on the wet film. However when
the ratio is too large, air entrainment occurs and bubbles exist on the wet film
which could be a cause of coating defects.
(3) Wrap angle: In this study as the wrap angle is increased, air stagnation and
streaks occur. However when the wrap angle is too small, air entrainment and
bubbles occur.
222
(4) High web tension: In this study there is no major effect of web tension on the wet
film instability. However as the web tension is increased, air entrainment is
prevented, so bubbles can be prevented at some conditions.
(5) Viscosity: According to this study, there may be a suitable viscosity with which
stable wet film can be achieved. With low viscosity air stagnation tends to occur
and streaks occur. With high viscosity air entrainment tends to occur and bubbles
exist on the wet film.
(6) High surface tension: As the surface tension is increased, air entrainment is
prevented and as a result streaks are prevented.
6.5 RECOMMENDATIONS FOR FUTURE WORK
This study was able to arrive at a new method of producing very thin films
(less than 0.5µm) at high speed (150m/min), deformable reverse roll coating. It has
also identified an important critical role for viscosity, suggesting that low viscosities
can be conducive to facilitating stable operation at low gaps in rigid reverse roll
coating. There are however several areas worthy of further investigation:
1. A theoretical analysis of deformable roll coating. We have shown in this thesis
how a simple lubrication analysis can be used to formulate a model but have not
actually produce the solution which requires a numerical procedure. Further, as
deformable roll coating at high speed cannot really be described by the
lubrication approximation (full 2D flow), a full flow analysis required. This is a
real theoretical challenge.
2. A theoretical analysis of the transfer flow from roll to web to assess if a stable
flow on the roll always remain stable when transfer to the web.
223
3. An experimental study to measure the effect of non-Newtonian liquid properties
on the wet film thickness and instabilities. This is important as in practice,
coatings are formulated with polymeric fluids and/or solid particles in a binder
and these solutions are a-priori non-Newtonian.
4. An experimental study to measure the effect of dynamic surface tension on the
wet film instabilities in reverse roll coating. This is important as during flow in
very thin films, surface tension additives require time to diffuse from the bulk of
the fluid to the surface.
5. An experimental study to measure more fully the effects on film thickness and
instabilities (ribbing and air entrainment) of the scraping extent of the film
formed on the metering roller, i.e. the effect of doctor blade properties (hardness,
thickness and setting angle).
6. An experimental study to measure the effect of web properties (roughness and
surface energy) on the transfer flow (film thickness, surface stability and air
entrainment).
224
REFERENCES
Ascanio, G., Carreau, P.J., Brito-De La Fuente, E., Tanguy, P.A., 2004. Forward
Deformable Roll Coating at High Speed with Newtonian Fluids. Chemical
Engineering Research and Design 82, 390-397.
Benkreira, H., Edwards, M.F., Wilkinson, W.L., 1981a. A semi-empirical model of

the forward roll coating flow of newtonian fluids. Chemical Engineering Science 36,
423-427.
Benkreira, H., Edwards, M.F., Wilkinson, W.L., 1981b. Roll coating of purely
viscous liquids. Chemical Engineering Science 36, 429-434.
Benkreira, H., Edwards, M.F., Wilkinson, W.L., 1982a. Flux distribution in forward
roll coating: a simple analysis. The Chemical Engineering Journal 25, 211-214.
Benkreira, H., Edwards, M.F., Wilkinson, W.L., 1982b. Ribbing instability in the roll
coating of newtonian fluids. Plastics and rubber processing and applications 2, 137-
144.
Benkreira, H., Edwards, M.F., Wilkinson, W.L., 1982c. Mathematical modelling of

the reverse and metering roll coating flow of Newtonian fluids. Chemical
Engineering Science 37, 277-282.
Benkreira, H., Patel, R., 1993. Direct gravure roll coating. Chemical Engineering
Science 48, 2329-2335.
Benkreira, H., Patel, R., Edwards, M.F., Wilkinson, W.L., 1994. Classification and
analyses of coating flows. Journal of Non-Newtonian Fluid Mechanics 54, 437-447.
Benkreira, H., Cohu, O., 1998a. Angling the wetting line retards air entrainment in
premetered coating flows. AIChE Journal 44, 1207-1209.
Benkreira, H., Cohu, O., 1998b. Direct forward gravure coating on unsupported web.
Chemical Engineering Science 53, 1223-1231.
Benkreira, H., 2002a. Dynamic wetting in metering and pre-metered forward roll
coating. Chemical Engineering Science 57, 3025-3032.
Benkreira, H., 2002b. Experimental study of dynamic wetting in reverse-roll coating.

AIChE Journal 48, 221-226.
Benkreira, H., Khan, M.I., 2008. Air entrainment in dip coating under reduced air
pressures. Chemical Engineering Science 63, 448-459.
225
Benkreira, H., Ikin, J.B., Patel, R., 2008. Advances in the study of dynamic wetting
in coating flows Experiments with air and other gases under reduced pressures,
Presented at the 14th International Coating Science and Technology Symposium,
Marina del Rey, California.
Blake, T.D., Ruschak, K.J., 1979. A maximum speed of wetting. Nature 282, 489-
491.
Blake, T.D., Clarke, A., Ruschak, K.J., 1994. Hydrodynamic assist of dynamic
wetting. AIChE Journal 40, 229-242.
Blake, T.D., Ruschak, K.J., 1997. Wetting: Static and dynamic contact lines, in:
S.F.Kistler and. P.M.Schweizer, C.H., London (Ed.), Liquid Film Coating, p. 63.
Blake, T.D., Dobson, R.A., Ruschak, K.J., 2004. Wetting at high capillary numbers.
Journal of Colloid and Interface Science 279, 198-205.
Bolton, B., Middleman, S., 1980. Air entrainment in a roll coating system. Chemical
Carvalho, M.S., Scriven, L.E., 1997. Deformable roll coating flows: steady state and
linear perturbation analysis. Journal of Fluid Mechanics 339, 143-172.
Carvalho, M.S., Scriven, L.E., 1999. Three-Dimensional Stability Analysis of Free

Surface Flows: Application to Forward Deformable Roll Coating Journal of
Computational Physics 151, 534-562.
Carvalho, M.S., Kheshgi, H.S., 2000. Low-flow limit in slot coating: Theory and
experiments. AIChE Journal 46, 1907-1917.
Carvalho, M.S., 2003. Effect of thickness and viscoelastic properties of roll cover on
deformable roll coating flows. Chemical Engineering Science 58, 4323-4333.
Chang, Y.-R., Chang, H.-M., Lin, C.-F., Liu, T.-J., Wu, P.-Y., 2007. Three minimum
wet thickness regions of slot die coating. Journal of Colloid and Interface Science
308, 222-230.
Chong, Y.H., Gaskell, P.H., Kapur, N., 2007. Coating with deformable rolls: An
experimental investigation of the ribbing instability. Chemical Engineering Science
62, 4138-4145.
Chin, C.-P., Wu, H.-S., Wang, S.S., 2010. Improved Coating Window for Slot
Coating. Industrial & Engineering Chemistry Research 49, 3802-3809.
Chu,W.B., Yang, J.W., Wang, Y.C, Liu, T.J., Tiu, C., Guo, J., 2006. The effect of
226
inorganic particles on slot die coating of poly(vinyl alcohol) solutions. Journal of
Colloid and Interface Science, 297, 215–225
Chu, V., Tsai, M.-Z., Chang, Y.-R., Liu, T.-J., Tiu, C., 2010. Effects of the molecular
weight and concentration of poly(vinyl alcohol) on slot die coating. Journal of
Applied Polymer Science 116, 654-662.
Chuang, H.K., Lee, C.C., Liu, T.J., 2008. An experimental study on the pickout of
scaled-up gravure cells. International Polymer Processing 23, 216-222.
Cohu, O., Magnin, A., 1997. Forward roll coating of Newtonian fluids with
deformable rolls: an experimental investigation. Chemical Engineering Science 52,
1339-1347.
Cohu, O., Benkreira, H., 1998. Air entrainment in angled dip coating. Chemical
Coyle, D.J., 1984. The fluid mechanics of roll coating: steady flows, stability, and
rheology. PhD thesis, University of Minnesota, Minneapolis.
Coyle, D.J., Macosko, C.W., Scriven, L.E., 1986. Film-splitting flows in forward roll
coating. Journal of Fluid Mechanics 171, 183-207.
Coyle, D.J., 1988. Forward roll coating with deformable rolls: A simple one-
dimensional elastohydrodynamic model. Chemical Engineering Science 43, 2673-
2684.
Coyle, D.J., Macosko, C.W., Scriven, L.E., 1990a. The fluid dynamics of reverse roll
coating. AIChE Journal 36, 161-174.
Coyle, D.J., Macosko, C.W., Scriven, L.E., 1990b. A simple-model of reverse roll
coating. Industrial & Engineering Chemistry Research 29, 1416-1419.
Coyle, D.J., Macosko, C.W., Scriven, L.E., 1990c. Stability of symmetric film-
splitting between counter-rotating cylinders. Journal of Fluid Mechanics 216, 437-
458.
Coyle, D.J., Macosko, C.W., Scriven, L.E., 1990d. Reverse roll coating of non-
Newtonian liquids. Journal of Rheology 34, 615-636.
Coyne, J.C., Elrod, H.G., 1970. Conditions for the rupture of a lubricating film.
Part1: Theoretical model. Journal of Lubrication Technology 73, 451-456.
Coyle, D.J., 1997. Knife and Roll Coating, in: S.F.Kistler and. P.M.Schweizer, C.H.,
London (Ed.), Liquid Film Coating, p. 539.
227
Deryagin, B.V., Levi, S.M., 1964. in: Focal Press, L. (Ed.), Film Coating Theory p.
137.
Gaskell, P.H., Savage, M.D., Summers, J.L., Thompson, H.M., 1995. Modelling and
analysis of meniscus roll coating. Journal of Fluid Mechanics 298, 113-137.
Gaskell, P.H., Innes, G.E., Savage, M.D., 1998a. An experimental investigation of

meniscus roll coating. Journal of Fluid Mechanics 355, 17-44.
Gaskell, P.H., Rees, S.R., Savage, M.D., Storey, S.P., 1998b. A mathematical model
of roll-to-web kiss coating. Trans IChemE 76, 29-37.
Greener, J., Middleman, S., 1979. Theoretical and Experimental Studies of the Fluid
Dynamics of a Two-Roll Coater. Industrial & Engineering Chemistry Fundamentals
18, 35-41.
Greener, J., Sullivan, T., Turner, B., Middleman, S., 1980. Ribbing instability of a
two-roll coater: newtonian fluids. Chemical Engineering Communications 5, 73-83.
Greener, J., Middleman, S., 1981. Reverse roll coating of viscous and viscoelastic
liquids. Industrial & Engineering Chemistry Fundamentals 20, 63-66.
Gutoff, E.B., Kendrick, C.E., 1982. Dynamic contact angles. AIChE Journal 28, 459-
466.
Gutoff, E.B., Kendrick, C.E., 1987. Low flow limits of coatability on a slide coater.
AIChE Journal 33, 141-145.
Han, S.K., Shin, D.M., Park, H.Y., Jung, H.W., Hyun, J.C., 2009. Effect of
viscoelasticity on dynamics and stability in roll coatings. The European Physical
Journal - Special Topics 166, 107-110.
Hanumanthu, R., 1999. Variation of gravure coating thickness during early stages of
doctor blade wear. AIChE Journal 45, 2487-2494.
Hao, Y., Haber, S., 1999. Reverse roll coating flow. International Journal for
Numerical Methods in Fluids 30, 635-652.
Harkins, W.D., Brown, F.E., 1919. The determination of surface tension (free surface
energy), and the weight of falling drops: the surface tension of water and benzene by
the capillary height method. Journal of the American Chemical Society 41, 499-524.
228
Hens, J., Abbenyen, W.V., 1997. Slide coating, in: S.F.Kistler and. P.M.Schweizer,
C.H., London (Ed.), Liquid Film Coating, p. 427.
Hewson, R.W., Kapur, N., Gaskell, P.H., 2006. A theoretical and experimental
investigation of tri-helical gravure roll coating. Chemical Engineering Science 61,
5487-5499.
Hewson, R.W., Kapur, N., Gaskell, P.H., 2010. Tri-helical gravure roll coating.
Ho, W.S., Holland, F.A., 1978. Between-rolls metering coating technique -

theoretical and experimental-study. Tappi 61, 53-56.
Hopkins, M. R., 1957. Viscous flow between rotating cylinders and a sheet moving
between them. British Journal of Applied Physics 8, 442-447.
Kapur, N., Gaskell, P.H., Bates, A., 2001. A parametric study of offset gravure
coating. Chemical Engineering Research & Design 79, 41-50.
Kapur, N., 2003. A parametric study of direct gravure coating. Chemical Engineering
Science 58, 2875-2882.
Landau, L., Levich, B., 1942. Dragging of a liquid by a moving plate. Acta.
Physicochim. URSS 17, 42-54.
Lee, J.H., Han, S.K., Lee, J.S., Jung, H.W., Hyun, J.C., 2010. Ribbing instability in
rigid and deformable forward roll coating flows. Korea-Australia Rheology Journal
22, 75-80.
Lee, K.-Y., Liu, L.-D., Ta-Jo, L., 1992. Minimum wet thickness in extrusion slot
Lin, F.-H., Liu, C.-M., Liu, T.-J., Wu, P.-Y., 2007. Experimental study on tensioned-
web slot coating. Polymer Engineering & Science 47, 841-851.
Lin, F.H., Liu, C.M., Liu, T.J., Wu, P.Y., 2008. A macroscopic mathematical model
for tensioned-web slot coating. Polymer Engineering and Science 48, 307-315.
Lin, Y.N., Liu, T.J., Hwang, S.J., 2005. Minimum wet thickness for double-layer
slide-slot coating of poly(vinyl-alcohol) solutions. Polymer Engineering and Science
45, 1590-1599.
Kang, Y.-T., Liu, T.-J., 1991a. Minimum film thickness for reverse roll coating.
Kang, Y.T., Lee, K.Y., Liu, T.J., 1991b. The effect of polymer additives on the
performance of a two-roll coater. Journal of Applied Polymer Science 43, 1187-1195.
229
Lie, K., Chiu, Y., Jang, J., 2009. Ribbing Instability Analysis of Forward Roll
Coating. Journal of Mechanics 25, 167-175.
Malone, B., 1992. An experimental investigation of roll coating phenomena. PhD

thesis, University of Leeds.
Mill, C.C., South, G.R., 1967. Formation of ribs on rotating rollers. Journal of Fluid
Mechanics 28, 523-529.
Nam, J., Carvalho, M.S., 2010. Two-layer tensioned-web-over-slot die coating:

Effect of die lip geometry. Chemical Engineering Science 65, 4065-4079.
Ning, C.Y., Tsai, C.C., Liu, T.J., 1996. The effect of polymer additives on extrusion
slot coating. Chemical Engineering Science 51, 3289-3297.
Noakes, C.J., Gaskell, P.H., Thompson, H.M., Ikin, J.B., 2002. Streak-line defect
minimization in multi-layer slide coating systems. Chemical Engineering Research &
Design 80, 449-463.
Patel, R., Benkreira, H., 1991. Gravure roll coating of newtonian liquids. Chemical
Pearson, J.R.A., 1960. The instability of uniform viscous flow under rollers and
spreaders. Journal of Fluid Mechanics 7, 481-500
Pitts, E., Greiller, J., 1961. The flow of thin liquid films between rollers. Journal of
Fluid Mechanics 11, 33-50
Prandtl, L., 1937. General discussion on lubrication. Proceedings of the Institution of

Mechanical Engineers 182, 102-104.
Pulkrabek, W.W., Munter, J.D., 1983. Knurl roll design for stable rotogravure
Reynolds, O., 1886. On the theory of lubrication and its application to beauchamp
tower's experiments, including an experimental determination of the viscosity of
olive oil. Philos. Trans. R. Soc., London, Sr. A 117: 157-233.
Romero, O.J., Suszynski, W.J., Scriven, L.E., Carvalho, M.S., 2004. Low-flow limit
in slot coating of dilute solutions of high molecular weight polymer. Journal of Non-
Newtonian Fluid Mechanics 118, 137-156.
Romero, O.J., Scriven, L.E., Carvalho, M.S., 2006. Slot coating of mildly
viscoelastic liquids. Journal of Non-Newtonian Fluid Mechanics 138, 63-75.
230
Romero, O.J., Carvalho, M.S., 2008. Response of slot coating flows to periodic
disturbances. Chemical Engineering Science 63, 2161-2173.
Ruschak, K.J., 1985. Coating Flows. Annual Review of Fluid Mechanics 17, 65-89.
Savage, M.D., 1977. Cavitation in lubrication. Part 1. On boundary conditions and

cavity—fluid interfaces. Journal of Fluid Mechanics 80, 743-755.
Shiode, H., Tamai, H., Isaki, T., 2009. Flow simulations of dynamic wetting line at a
reverse-roll coater using the VOF method. Nihon Reoroji Gakkaishi(Journal of the
Society of Rheology, Japan) 37, 59-64.
Spiers, R.P., Subbaraman, C.V., Wilkinson, W.L., 1974. Free coating of a Newtonian
liquid onto a vertical surface. Chemical Engineering Science 29, 389-396.
Tiu, C., Wang, L., Liu, T.-J., 1999. Non-Newtonian effects on pre-metered reverse
roll coating. Journal of Non-Newtonian Fluid Mechanics 87, 247-261.
Thompson, H.M., Kapur, N., Gaskell, P.H., Summers, J.L., Abbott, S.J., 2001. A
theoretical and experimental investigation of reservoir-fed, rigid-roll coating.
Wang, L., 1998. Coating fluid rheology in reverse roll coating operation. PhD thesis,
Monash University.
Yih, C.-S., Kingman, J.F.C., 1960. Instability of a Rotating Liquid Film with a Free
Surface. Proceedings of the Royal Society of London. Series A. Mathematical and
Physical Sciences 258, 63-89.
Yin, X., Kumar, S., 2006. Flow visualization of the liquid-emptying process in
scaled-up gravure grooves and cells. Chemical Engineering Science 61, 1146-1156.
231
Appendix (Lubrication model for rigid reverse roll coating)
This appendix shows calculation procedure how to derive equations of velocity,
pressure and pressure gradient.
Navier-Stokes equation:
r( + u・ u ) = - P+ u + rg (A.1)
We consider only one component in the direction of “x”.
r( +u +v +w )=- + + rg (A.2)
Mass continuity equation:
+ ・(r u 0 (A.3)
Under the incompressible assumption in which density is a constant, the mass
continuity equation simplifies to a volume continuity equation:
・u = 0 (A.4)
If we assume the flow is only one direction (x direction), the following can be
derived.
“u” is constant in the x direction, therefore = 0, =0, =0.
v=w=0 (A.5)
We consider only x-y dimension, i.e. eliminate z direction and the effect of gravity.
= 0, =0, g =0
The complex Navier-Stokes equation can be simplified to the equation (A.7) in case
the ratio of the roll gap to radius is very small, i.e. h/R<<1,.
= (A.7)
232
= (A.8)
This equation can be integrated twice with respect to y.
u= +c y+c (A.9)
*Velocity boundary condition:
u= -uT at y=yT, u=uB at y=yB (A.10)
Using the velocity boundary conditions the following two equations can be derived.
At y = y u = +c y +c (A.11)
At y = y u = +c y +c (A.12)
Subtraction (A.12) from (A.11) gives:
u -u = ( - )+c y y ) (A.13)
c = - (y y ) (A.14)
Substituting (A.14) into (A.12) gives:
c =u - - y + y (y y ) (A.15)
Substituting (A.14) and (A.15) into (A.9) gives:
u= + - (y y )y +u - - y
+ y (y y ) (A.16)
h= y y , = u /u (A.17)
€
u= (y - (y y )y+y y )– •
(y y )u +u (A.18)
(A.18) divided by u gives:
U= (A.19)
233
€
U= (y - (y y )y+y y )– •
(y y )+1 (A.20)
•‚ ƒ•‚ /„ €
U= (y - (y y )y+y y )– •
(y y )+1
•‚ ƒ„•‚ ƒ•‚ /„ /ƒ„•‚
(A.21)
•‚ ƒ•‚ /„ €
U= •‚
(y - (y y )y+y y )– •
(y y )+1 (A.22)
/ƒ„•‚
•‚
P=p ƒh… /R , X = x/ƒRh… , H = h/h… , Y=y/h… (A.23)
† ˆ‰ † Ž•
U=‡ ˆŠ
(‹ ‡ - (‹Œ ‹• ) Y + ‹Œ ‹• ) – •
(‹ ‹• ) + 1 (A.24)
Integrating the dimensionless velocity we can get the dimensionless leakage flow
rate.
“
Q‘ = ’“ UdY (A.25)
“ • “ • “ “
Q‘ = ”
( –
- (Y Y ) +Y Y Y Y )
€ “ “
– —
( Y Y Y + Y Y ) (A.26)
— “ “ “ “ — “ “
Q‘ = ”
( –
- + HY Y )
€ — “ “
– —
( HY +H (A.27)
Q‘ = (2H Y Y Y Y - 3H Y Y + 6HY Y )
”
€ — “ “
– —
( HY +H (A.28)
Q‘ = H (2Y Y Y Y )– H 1 +H
”
= H Y Y – H 1 +H
”
= H– + H 1 (A.29)
”
Dimensionless total flow rate l‘ is defined as:
234
l = (A.30)
Q = = = (1 - r ) l (A.31)
(A.29) and (A.31) give:
(1 - r ) l = H + H 1 r (A.32)
= (1 - r ) ) (A.33)
‚ƒ
= (1 - €) • ) (A.34)
„ ∞
dP = 6(1 - r ) ( „ ∞
2l „ ∞ …
) (A.35)
Let us consider geometry of the roll gap region (see Figure 5.1).
R † – ‡y + x = R† (A.36)
‡y =0 ‡y = ˆ
(A.37)
y† = h‰ + ‡y = h‰ + ˆ
(A.38)
We can derive yŠ in the same way:
yŠ = - ( h‰ + ˆ
) (A.39)
Subtraction (A.39) from (A.38) gives:
h = y† - yŠ = h‰ + (ˆ + ˆ = h‰ + ˆ
(A.40)
H= =1+ˆ =1+X (A.41)
Substituting (A.41) into (A.35) gives:
„ ∞
dP = 6(1 - r ) ( „ ∞
2l „ ∞ … ) (A.42)
‹ ‹
„ = + tan X (A.43)
‹ ‹
„ ∞
= + tan X - tan (-∞) (A.44)
‹ ‹
235
π
„ ∞ ‹
= ‹
+ tan X + Œ
(A.45)
„ ‹ … =Œ ‹
+• ‹
+ • tan X (A.46)
„ ∞ ‹ … =Œ ‹
+• ‹
+ • tan X - •
tan (-∞) (A.47)
π
„ ∞ ‹ … =Œ ‹
+• ‹
+ • tan X + Ž
(A.48)
Substituting (A.45) and (A.48) into (A.42)
π
„ ∞
dP = 6(1 - r ) (( ‹
+ tan X + Œ
)
π
- 2l ( Œ ‹
+• ‹
+ • tan X + Ž
)) (A.49)
π
P(X) – P( ∞) = 6(1 - r ) (( ‹
+ tan X + Œ
)
π
- 2l ( Œ ‹
+• ‹
+ • tan X + Ž
)) (A.50)
Applying pressure boundary condition: P( ∞) = 0
• ’
P(X) = 6(1 - €) (( •‹
+ ••‘ • X + “ )
” ” ”’
- 2‚ƒ ( “ •‹
+ • •‹ + • ••‘ • X + •
)) (A.51)
Above three equations (A.24) (A.34) (A.51) need to be solved the flow rate. To
achieve this requires the setting of separation boundary conditions where the film
forms. Two sets of conditions may be used and are known as the Reynolds separation
conditions and the Prandtl-Hopkins separation condition as shown in Chapter 5.
236

PHD Thesis (Final Ver.2)

Uploaded by

Copyright:

Available Formats

PHD Thesis (Final Ver.2)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PHD Thesis (Final Ver.2)

Uploaded by

Copyright:

Available Formats

What is reverse roll coating and what are its advantages?

What is reverse roll coating and what are its advantages?

What are the challenges in achieving very thin films at high speeds using reverse roll coating?

What are the challenges in achieving very thin films at high speeds using reverse roll coating?

HIGH SPEED VERY THIN FILMS

WITH REVERSE ROLL COATING

SUBTITLE: An experimental investigation of reverse roll coating of

fluids using rigid and deformable rolls at high speeds

Yusuke SHIBATA (BEng, MEng)

Submitted as a PhD thesis for the

School of Engineering, Design and Technology

HIGH SPEED VERY THIN FILMS

WITH REVERSE ROLL COATING

The objective of a coating operation is to transfer a defect free liquid film

not uniform rendering it unacceptable to customers. One of the most versatile

roller gaps (rollers nearly touching or in elastohydrodynamic contact) to achieve the

capillary number is the one.

In addition the potential of reverse deformable roll coating (rolls in

elastohydrodynamic contact) was investigated in order to achieve much thinner

films at higher speeds. As a result of the investigation of reverse deformable roll

much higher speeds compared to reverse rigid roll coating.

entrainment at upstream meniscus depend on the liquid properties such as viscosity

that suitable viscosity and high surface tension were needed.

Chapter 1 : INTRODUCTION ...................................................................................... 1

Figure 1.1: Free coating ........................................................................................................... 3

b Thickness of rubber layer

Ca Capillary number Ca= V/

CaA Capillary number (Applicator roller)

CaM Capillary number (Metering roller)

Es Elasticity number Es= V/ER

h Roll gap (at any positions)

H Dimensionless roll gap (at any positions) H=h/h0

h0 Roll gap (at roll centre)

H0 Dimensionless roll gap (at roll centre) H0=h0/2R

ha Film thickness (applicator roller)

hB Film thickness (bottom roller)

HB Dimensionless film thickness (bottom roller)HB=hB/h0

hf Final film thickness

hi0 Initial set roll gap (at roll centre)

hi Initial set roll gap (at any positions)

hm Metered film thickness

HM Dimensionless metered film thickness HM= hm/ h0

HM, min Dimensionless minimum metered film thickness

hS Roll gap (stagnation point)

hT Film thickness (top roller)

hW Film thickness (withdrawal)

h* Characteristic film thickness h *( V / g )1 2

NCa Modified capillary number NCa =Ca (R/H0)1/2

N Modified capillary number N= h ¤R /Ca

P Dimensionless pressure P=p h /R

qB Flow rate (bottom roller)

qL Leakage flow rate

qT Flow rate (top roller)

RB Roll radius (bottom roller)

Rd Roll radius (deformable roller)

Re Equivalent roll radius

Rr Roll radius (rigid roller)

RT Roll radius (top roller)

rc Radius of curvature of meniscus