AUTOMOTIVE POWDER COATINGS: AN INVESTIGATION OF …

AUTOMOTIVE POWDER COATINGS: AN INVESTIGATION OF PARAMETERS AFFECTING FINISH

APPEARANCE AND DURABILITY

By

LINDITA PRENDI

A Dissertation Submitted to the Faculty of Graduate Studies

through Environmental Engineering in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy at the

University of Windsor

Windsor, Ontario, Canada

2010

© 2010 Lindita Prendi

iii

AUTHOR’S DECLARATION OF ORIGINALITY I hereby certify that I am the sole author of this thesis and that no part of this thesis has been published or submitted for publication. I certify that, to the best of my knowledge, my thesis does not infringe upon anyone’s copyright nor violate any proprietary rights and that any ideas, techniques, quotations, or any other material from the work of other people included in my thesis, published or otherwise, are fully acknowledged in accordance with the standard referencing practices. Furthermore, to the extent that I have included copyrighted material that surpasses the bounds of fair dealing within the meaning of the Canada Copyright Act, I certify that I have obtained a written permission from the copyright owner(s) to include such material(s) in my thesis and have included copies of such copyright clearances to my appendix. I declare that this is a true copy of my thesis, including any final revisions, as approved by my thesis committee and the Graduate Studies office, and that this thesis has not been submitted for a higher degree to any other University or Institution.

iv

ABSTRACT

This project investigated the finish quality of automotive powder coatings in terms of appearance, adhesion and chip resistance. Two powder basecoats (red and black), of three particle sizes, and a colorkey primer (red) were studied in relation to the process temperature and time, particle size, datapaq value (DPV), heating rate, complex viscosity, degree of conversion and film thickness. The appearance was quantified using the contrast values of the structure spectrum elements (Wa, Wb, Wc, Wd, and We). While gravel and scratch tests were used to quantify adhesion and chip resistance properties. The research was conducted in four distinct phases.

It was found that long-waves (Wc, Wd and We) remained unaffected by the factors considered. Short-waves (Wa, Wb) increased with increasing process temperature, time, DPV, viscosity and degree of conversion. The most dramatic increase was observed at high process time and temperatures. High heating rates resulted in smoother film and when combined with high process temperatures (193°C) resulted in ideal spectrum sha pe and good long-wave coverage. As expected from published literature, the finer particle size gave smoother appearance. Smaller contrast values of the long-waves (Wc, Wd, and Wd) were obtained for all cure conditions. This was true for both red and black basecoats. The results for short-waves were not as consistent.

The modified Orchard’s model indicated that time, viscosity and film build can be used to illustrate the progress of levelling. Using the contrast values instead of the amplitudes in the Orchard’s model, still yielded plausible results. However, the model cannot be used to predict individual behavior of each appearance element (Wa to We). The regression model seemed to be the best predictive model for short-waves. However, the behavior of long-waves could not be predicted with the models considered in this study.

This work suggests that powder basecoats and colorkey primers can yield appearance qualities comparable to waterborne counterpart. The datapaq value (DPV) was found as the most important factor controlling the appearance of powder basecoats. In order to achieve better appearance, DPVs should be controlled within the range provided by the paint manufacturer.

v

To my children and my husband for their support and love

vi

ACKNOWLEDGEMENT

I would like to express my gratitude to my supervisor Dr. Paul Henshaw for his guidance, encouragement, patience and support throughout the length of this project.

The support of the ARDC team has been crucial for the successful completion of this work. I would like to thank Tony Mancina, Chris Tighe, Jennifer Di Domenico, Marie Mills, and Jason Bastien for contributing their technical knowledge and assisting with the experimental procedures.

Special thanks to Albert Tse, Chan Skiba and Rob Schwark from the PPG laboratories that assisted with the DSC and viscosity measurements.

Lindita Prendi August, 2010

vii

TABLE OF CONTENTS

AUTHOR’S DECLARATION OF ORIGINALITY ............... .................................. III

ABSTRACT .......................................... ............................................................... IV

ACKNOWLEDGEMENT ................................... .................................................. VI

LIST OF FIGURES .............................................................................................. IX

LIST OF TABLES .................................... .......................................................... XII

CHAPTER 1 - INTRODUCTION ........................................................................... 1

1.1 AUTOMOTIVE PAINT PROCESSES AND THE ENVIRONMENT .......................................................... 1

1.2 VEHICLE PAINTING PHASE AND MATERIALS USED ....................................................................... 2

1.3 OBJECTIVES AND SCOPE ............................................................................................................... 3

CHAPTER 2 – LITERATURE REVIEW ..................... .......................................... 7

2.1. DEVELOPMENT OF POWDER COATINGS ......................................................................................... 7

2.2. CHARACTERIZATION OF APPEARANCE BY SURFACE STRUCTURES AND WAVE-SCAN DOI ........ 9

2.3. CURE KINETICS OF THERMOSETTING POWDER COATINGS (DSC) ............................................ 11

2.4. FACTORS AFFECTING FINISH QUALITY OF POWDER COATINGS ................................................ 13

CHAPTER 3 – MATERIALS AND METHODS ................. .................................. 20

3.1. TESTING FACILITY ........................................................................................................................ 20

3.2. GENERAL APPLICATION PARAMETERS ........................................................................................ 20

3.3. PAINT TYPES................................................................................................................................ 21

3.4. MEASUREMENT PROCEDURE AND INSTRUMENTS....................................................................... 22

3.4.1. Particle Size Analyzer (PSA) .......................................................................................... 22

3.4.2. Viscosity ............................................................................................................................. 23

3.4.3. Cure Kinetics (DSC) ......................................................................................................... 23

3.4.4. Wave-scan DOI ................................................................................................................ 24

3.4.5. Film Build ........................................................................................................................... 24

3.4.6. Scribe and Gravel Tests .................................................................................................. 24

3.4.7. Datapaq ............................................................................................................................. 25

3.5. EXPERIMENTAL PROCEDURE ...................................................................................................... 25

3.5.1. Phase 1 - Cure Window Corner Points Experiments .................................................. 25

3.5.2. Phase 2 Stage A - Isothermal Experiments ................................................................. 28

3.5.3. Phase 2 Stage B - DSC Points Experiments................................................................ 31

3.5.4. Phase 3 - Variable Heating Rate Experiments ............................................................ 34

CHAPTER 4 - PHASE 1: CURE WINDOW CORNER POINTS .... ..................... 37

4.1. PARTICLE SIZE ANALYSIS ............................................................................................................ 37

4.2. EXPERIMENTAL RESULTS AND DISCUSSION ............................................................................... 37

4.2.1. Constant Particle Size ..................................................................................................... 39

4.2.2. Constant Formulation....................................................................................................... 61

4.2.3. Abrasion resistance and Adhesion ................................................................................ 67

viii

4.3. CONCLUSION ............................................................................................................................... 68

CHAPTER 5 - PHASE 2: ISOTHERMAL AND DSC POINTS .... ....................... 70

5.1. STAGE A – ISOTHERMAL EXPERIMENTS ..................................................................................... 70

5.1.1. DPV and Appearance ...................................................................................................... 75

5.1.2. Viscosity and Appearance ............................................................................................... 87

5.1.3. Degree of Conversion and Appearance ........................................................................ 95

5.1.4. Before Ramp Experiments and First Few Minutes Isotherms ................................. 104

5.2. STAGE B – DSC SIGNIFICANT POINTS EXPERIMENTS ................................................................... 109

5.3 CONCLUSION .................................................................................................................................... 112

CHAPTER 6 - PHASE 3: VARYING HEATING RATE EXPERIMEN TS .......... 113

6.1. COMPARISON BETWEEN 5 AND 10°C/ MIN ....................................................................................... 113

6.2. COMPARISON OF 5°C/ MIN AND 10°C/ MIN WITH A VERY FAST RAMP ............................................ 123

6.3. CONCLUSION ................................................................................................................................... 126

CHAPTER 7 – PHASE 4: MODEL GENERATION ............. ............................. 127

7.1 LEVELLING MODEL ........................................................................................................................... 127

7.2 REGRESSION MODEL ....................................................................................................................... 130

7.3 CONCLUSION .................................................................................................................................... 137

CHAPTER 8 – CONCLUSIONS AND RECOMMENDATIONS ....... ................. 139

8.1 CONCLUSIONS .................................................................................................................................. 139

8.2 SIGNIFICANCE OF THIS RESEARCH .................................................................................................. 141

8.3 RECOMMENDATIONS ........................................................................................................................ 141

REFERENCES IN ALPHABETICAL ORDER .................. ................................ 144

APPENDICES .................................................................................................. 149

VITA AUCTORIS ..................................... ......................................................... 150

ix

LIST OF FIGURES FIGURE 1.1 - STAGES IN AUTOMOTIVE PAINT APPLICATION PROCESS ............................ 2

FIGURE 2.1 - FOCUSING ON SURFACE AND WAVENESS PATTERN. .................................... 9

FIGURE 2.2 - IMAGE FORMING QUALITY OF A SURFACE. ..................................................... 10

FIGURE 2.3 - WAVE-SCAN DOI PRINCIPLE OF OPERATION (BYK, 2008) ........................... 11

FIGURE 2.4 - DYNAMIC DSC THERMOGRAM.............................................................................. 12

FIGURE 3.1 - RPA-1 POWDER APPLICATOR (ITWGEMA, 2008) ............................................ 20

FIGURE 3.2 - CURE WINDOW OF POWDER COATINGS .......................................................... 22

FIGURE 3.3 - PANEL RACK SET UP FOR RUNS 1-15 ................................................................ 26

FIGURE 3.4 - HOLDING RACK WITH SPRAYED PANELS ......................................................... 28

FIGURE 3.5 - DATAPAQ THERMOCOUPLES ............................................................................... 30

FIGURE 3.6 - MAP FOR ISOTHERMAL RUNS .............................................................................. 30

FIGURE 3.7 – MAP FOR DSC RUNS AND SPRAY PATTERN ................................................... 32

FIGURE 4.1 – WE BEFORE SPRAY FOR RCP30 ......................................................................... 38

FIGURE 4.2 – WE BEFORE AND AFTER SPRAY FOR RCP30 .................................................. 38

FIGURE 4.3 - FILM BUILDS FOR 30 MICROMETER PAINTS .................................................... 41

FIGURE 4.4 - WA VS. FB AND CURE CONDITIONS FOR BBC30 ............................................ 41

FIGURE 4.5 - WA VS. FB AND CURE CONDITIONS FOR RBC30 ............................................ 42

FIGURE 4.6 - APPEARANCE VS. CURE CONDITION FOR RCP30 .......................................... 47

FIGURE 4.7 - APPEARANCE VS. CURE CONDITION FOR RBC30 .......................................... 48

FIGURE 4.8 - APPEARANCE VS. CURE CONDITION FOR BBC30 .......................................... 49

FIGURE 4.9 - WC VS. CURE CONDITIONS FOR ALL THREE PAINTS ................................... 51

FIGURE 4.10 - STRUCTURE SPECTRUM AT LT1 ....................................................................... 51

FIGURE 4.11 - STRUCTURE SPECTRUM AT LT2 ....................................................................... 52

FIGURE 4.12 - STRUCTURE SPECTRUM AT CNOMINAL ................................................................ 52

FIGURE 4.13 - STRUCTURE SPECTRUM AT HT1 ....................................................................... 52

FIGURE 4.14 - STRUCTURE SPECTRUM AT HT2 ....................................................................... 53

FIGURE 4.15 - APPEARANCE VS. CURE CONDITION FOR RBC25 ........................................ 58

FIGURE 4.16 - APPEARANCE VS. CURE CONDITION FOR BBC25 ........................................ 59

FIGURE 4.17 - WC VS. CURE CONDITIONS FOR RBC25 AND BBC25 .................................. 60

FIGURE 4.18 - DEVELOPMENT OF STRUCTURE SPECTRUM LT1 (RBC25 AND BBC25) 60

FIGURE 4.19 - APPEARANCE VS. CURE CONDITION FOR BBC20 ........................................ 64

FIGURE 4.20 - WD VS. CURE CONDITIONS FOR BBC20, BBC25 AND BBC30 .................... 65

FIGURE 4.21 - STRUCTURE SPECTRUM FOR BBC ALL SIZES AT LT1 ................................ 65

FIGURE 4.22 - WC VS. CURE CONDITIONS FOR RBC25 AND RBC30 .................................. 66

FIGURE 4.23 - STRUCTURE SPECTRUM FOR RBC ALL SIZES AT LT1 ................................ 67

FIGURE 5.1 - DATAPAQ PROFILE FOR RBC25, 163°C AT 95 MINUTES ............................... 72

FIGURE 5.2 - DPV VS TIME AT 163°C ISOTHERM ...................................................................... 77



FIGURE 5.5 - DPV VS TIME AT FOR RBC25 ................................................................................. 78

FIGURE 5.6 - DPV VS TIME AT FOR BBC20 ................................................................................. 79

FIGURE 5.7 - DPV VS TIME AT FOR BBC25 ................................................................................. 79

FIGURE 5.8 - STRUCTURE SPECTRUM ELEMENTS VS. DPV FOR RBC25 ......................... 81

FIGURE 5.9 - STRUCTURE SPECTRUM VS DPV BETWEEN 100 AND 400 (RBC25) .......... 82

x

FIGURE 5.10 - STRUCTURE SPECTRUM ELEMENTS VS. DPV FOR BBC25 ....................... 83

FIGURE 5.11 - STRUCTURE SPECTRUM ELEMENTS VS. DPV FOR BBC20 ....................... 84

FIGURE 5.12 – DULLNESS (DU) VS. DPV FOR ALL THREE PAINTS ...................................... 84

FIGURE 5.13 - WA VS. DPV FOR ALL THREE PAINTS............................................................... 85

FIGURE 5.14 - WB VS. DPV FOR ALL THREE PAINTS............................................................... 85

FIGURE 5.15 - WC VS. DPV FOR ALL THREE PAINTS .............................................................. 86

FIGURE 5.16 - WD VS. DPV FOR ALL THREE PAINTS .............................................................. 86

FIGURE 5.17 - WE VS. DPV FOR ALL THREE PAINTS............................................................... 87

FIGURE 5.18 – COMPLEX VISCOSITY VS. TIME AT THREE ISOTHERMAL TEMPERATURES ........................................................................................................................................................ 88

FIGURE 5.19 - APPEARANCE VS. VISCOSITY FOR RBC25 ..................................................... 91

FIGURE 5.20 - APPEARANCE VS. VISCOSITY FOR BBC25 ..................................................... 91

FIGURE 5.21 – DULLNESS (DU) VS. VISCOSITY FOR RBC25 AND BBC25 .......................... 92

FIGURE 5.22 - WA VS. VISCOSITY FOR RBC25 AND BBC25 ................................................... 92

FIGURE 5.23 - WB VS. VISCOSITY FOR RBC25 AND BBC25 ................................................... 93

FIGURE 5.24 - WC VS. VISCOSITY FOR RBC25 AND BBC25 .................................................. 93

FIGURE 5.25 - WD VS. VISCOSITY FOR RBC25 AND BBC25 .................................................. 94

FIGURE 5.26 - WE VS. VISCOSITY FOR RBC25 AND BBC25 ................................................... 94

FIGURE 5.27 - DSC TYPICAL THERMOGRAM FOR BBC25 ...................................................... 96

FIGURE 5.28 - DSC THERMOGRAM WITH OVERLAYS AT THREE HEATING RATES FOR BBC25 ........................................................................................................................................... 97

FIGURE 5.29 - ARRHENIUS PLOT AT DIFFERENT CONVERSIONS OF BBC25 .................. 99

FIGURE 5.30 - RATE CONSTANTS AT 99% CONVERSION OF BBC25 ................................ 100

FIGURE 5.31 - DEGREE OF CONVERSION (Α) AT THREE ISOTHERMS FOR BBC20 ...... 101

FIGURE 5.32 - STRUCTURE SPECTRUM VS. CONVERSION FOR RBC25 ......................... 102

FIGURE 5.33 - STRUCTURE SPECTRUM VS. CONVERSION FOR BBC25 ......................... 103

FIGURE 5.34 - STRUCTURE SPECTRUM VS. CONVERSION FOR BBC20 ......................... 103

FIGURE 5.35 - STRUCTURE SPECTRUM VS. REAL TIME AT 163ºC .................................... 105



FIGURE 5.38 - STRUCTURE SPECTRUM AT 163ºC ................................................................. 107



FIGURE 5.41 - STRUCTURE SPECTRUM FOR BBC25 AT DSC POINTS ............................. 110

FIGURE 5.42 - STRUCTURE SPECTRUM FOR BBC20 AT DSC POINTS ............................. 111

FIGURE 6.1 - FILM BUILD TREND OVER TIME .......................................................................... 113

FIGURE 6.2 - STRUCTURE SPECTRUM FOR BBC25 AT LT1_R5 AND LT1_10 ................. 115



FIGURE 6.5 - STRUCTURE SPECTRUM FOR BBC25 AT MT1_R5 AND MT1_10 ............... 117

FIGURE 6.6 - STRUCTURE SPECTRUM FOR BBC25 AT MT2_R5 AND MT2_10 (NOMINAL CONDITIONS) ............................................................................................................................ 117

FIGURE 6.7 - STRUCTURE SPECTRUM FOR BBC25 AT MT3_R5 AND MT3_10 ............... 118

FIGURE 6.8 - STRUCTURE SPECTRUM FOR BBC25 AT HT1_R5 AND HT1_10 ................ 118

FIGURE 6.9 - STRUCTURE SPECTRUM FOR BBC25 AT HT2_R5 AND HT2_10 ................ 119

FIGURE 6.10 - STRUCTURE SPECTRUM FOR BBC25 AT HT3_R5 AND HT3_10 .............. 119

xi

FIGURE 6.11 - STRUCTURE SPECTRUM FOR RBC25 AT LT1_R5 AND LT1_10 ............... 120



FIGURE 6.14 - STRUCTURE SPECTRUM FOR RBC25 AT MT1_R5 AND MT1_10 ............ 121



FIGURE 6.17 - STRUCTURE SPECTRUM FOR RBC25 AT HT1_R5 AND HT1_10 ............. 122



FIGURE 6.20 - STRUCTURE SPECTRUM FOR BBC25 AT THREE RAMPS (163°C@14 MIN) ...................................................................................................................................................... 124

FIGURE 6.21 - STRUCTURE SPECTRUM FOR RBC25 AT THREE RAMPS (163°C@14 MIN) ...................................................................................................................................................... 124

FIGURE 6.22 - STRUCTURE SPECTRUM FOR BBC25 AT THREE RAMPS (193°C@35 MIN) ...................................................................................................................................................... 125

FIGURE 6.23 - STRUCTURE SPECTRUM FOR RBC25 AT THREE RAMPS (193°C@35 MIN) ...................................................................................................................................................... 125

FIGURE 7.1 - LAMBDA* FOR APPEARANCE ELEMENTS OF RBC25 ................................... 129

FIGURE 7.2 - LAMBDA* FOR APPEARANCE ELEMENTS OF BBC25 ................................... 129

FIGURE 7.3 - LOG(LAMBDA*) VS. LOG(FB TERM) FOR BBC25 ............................................ 130

xii

LIST OF TABLES TABLE 1.1 – PAINT PROCESSES IN A TYPICAL AUTOMOTIVE APPLICATION .................... 3

TABLE 1.2 – SUMMARY MATRIX OF RESEARCH PHASES ........................................................ 6

TABLE 2.1 - WAVE-SCAN ELEMENTS AND THEIR WAVELENGTH RANGE ......................... 11

TABLE 3.1 - POWDER ZONE SETTINGS AND CONSTANT PARAMETERS .......................... 21

TABLE 3.2 - TYPES OF POWDER COATINGS.............................................................................. 21

TABLE 3.3 - SUMMARY OF PARTICLE SIZE MEASUREMENTS BEFORE SPRAY .............. 22

TABLE 3.4 – CURE CONDITIONS FOR ALL PAINT TYPES ....................................................... 26

TABLE 3.5 - ISOTHERMAL CONDITIONS ...................................................................................... 29

TABLE 3.6 – MATRIX FOR DSC POINTS RUNS ........................................................................... 33

TABLE 3.7 - SUMMARY OF OVEN SETTINGS (FULL TABLE IN APPENDIX 5) ..................... 35

TABLE 4.1 - SUMMARY OF APPEARANCE ELEMENTS BEFORE SPRAY (N = 60) ............. 39

TABLE 4.2 - SUMMARY OF FILM BUILD OVERALL AVERAGES .............................................. 40

TABLE 4.3 – WAVE-SCAN VALUES FOR LT1 CONDITION (N = 36) ........................................ 43

TABLE 4.4 – WAVE-SCAN VALUES FOR LT2 CONDITION (N = 36) ........................................ 43

TABLE 4.5 – WAVE-SCAN VALUES FOR CNOMINAL CONDITION (N = 36) .......................... 43

TABLE 4.6 – WAVE-SCAN VALUES FOR HT1 CONDITION (N = 36) ....................................... 44

TABLE 4.7 – WAVE-SCAN VALUES FOR HT2 CONDITION (N = 36) ....................................... 44

TABLE 4.8 – SUMMARY OF APPEARANCE PARAMETER CHANGES WITH TIME AND TEMPERATURE (30µM) .............................................................................................................. 46

TABLE 4.9 - SUMMARY OF FILM BUILD OVERALL AVERAGES .............................................. 53

TABLE 4.10 - STRUCTURE SPECTRUM FOR LT1 CONDITION ............................................... 55

TABLE 4.11 - STRUCTURE SPECTRUM FOR LT2 CONDITION ............................................... 55

TABLE 4.12 - STRUCTURE SPECTRUM FOR CNOMINAL ............................................................... 56

TABLE 4.13 - STRUCTURE SPECTRUM FOR HT1 CONDITION .............................................. 56

TABLE 4.14 - STRUCTURE SPECTRUM FOR HT2 CONDITION .............................................. 56

TABLE 4.15 - SUMMARY OF APPEARANCE PARAMETERS (25µM) ...................................... 57

TABLE 4.16 - FILM BUILD OVERALL AVERAGES FOR BBC20 ................................................ 61

TABLE 4.17 - SUMMARY OF APPEARANCE PARAMETERS BBC20 .......................................... 61

TABLE 4.18 - STRUCTURE SPECTRUM FOR BBC20 ................................................................. 63

TABLE 4.19 - SUMMARY OF SCRIBE AND GRAVEL TESTS FOR RCP30 ............................. 68

TABLE 5.1 - FB SUMMARY FOR ISOTHERMAL RUNS AT 163°C ............................................ 71

TABLE 5.2 - EXAMPLE OF RAW DATA USED TO CALCULATE AVERAGE FILM THICKNESS AND COV ...................................................................................................................................... 71

TABLE 5.3 - SUMMARY OF STRUCTURE SPECTRUM FOR 163°C ISOTHERM .................. 73

TABLE 5.3 - CONTINUES ................................................................................................................... 74

TABLE 5.4 – EXAMPLE OF RAW DATA USED FOR AVERAGE AND COV CALCULATION 75

TABLE 5.5 - CURE SCHEDULE FOR DPV CALCULATIONS ...................................................... 76

TABLE 5.6 - SUMMARY OF APPEARANCE CHANGES WITH TIME AND TEMPERATURE 80

TABLE 5.7 - ISOTHERMAL VISCOSITY MEASUREMENTS ....................................................... 89

TABLE 5.8 - SUMMARY OF RHEOLOGICAL PARAMETERS ..................................................... 90

TABLE 5.9 - SUMMARY OF TG, TM AND RELAXATION PEAK TEMPERATURE AT EACH HEATING RATE ........................................................................................................................... 95

TABLE 5.10 - SUMMARY OF KINETIC ANALYSIS ........................................................................ 98

xiii

TABLE 5.11 – COMPARISON OF APPEARANCE WITH WATERBORNE BASECOATS ..... 108

TABLE 6.1 - APPEARANCE DATA FOR BBC25 .......................................................................... 114

TABLE 6.2 - APPEARANCE DATA FOR RBC25 .......................................................................... 114

TABLE 7.1 – SUMMARY OF COMMON PREDICTORS ............................................................. 132

TABLE 7.2 – SUMMARY OF REGRESSION ANALYSES USING INDIVIDUAL INDEPENDENT VARIABLES ................................................................................................................................ 134

TABLE 7.3 – SUMMARY OF SIGNIFICANT PREDICTOR VARIABLES .................................. 135

TABLE 7.4 - FINAL PREDICTOR MODELS FOR RBC25 AND BBC25 AT 163°C .................. 135



TABLE 7.7 – CORRELATION MATRIX FOR BBC25 AT 193°C ................................................. 137

1

CHAPTER 1 - INTRODUCTION

1.1 Automotive Paint Processes and the Environment

The automotive manufacturing industry is one of the largest industrial sectors in North America, producing millions of vehicles annually for North American and International export markets (Prendi et al., 2006; Wallace, 2002). As new vehicles are being developed to be more fuel-efficient and less emitting during their operation, the environmental burden of the manufacturing phase becomes proportionately more significant. The car manufacturing process is very complex and involves several stages. The paint process in particular is not only complex, but capitally intensive at the same time. A substantial amount of energy and material use is involved in applying the paint to a vehicle (Hazel, 1997; Kim et al., 2001) and the process generates considerable emissions to air, water and land (Papasavva et al., 2001b). Three unit processes represent the majority of environmental impacts: pretreatment, electrocoat (e-coat) and topcoat. Environmental impacts associated with automotive paint processes include releases to air (e.g., volatile organic emissions), wastewater (e.g., pretreatment chemicals, paint overspray in solution), and land (e.g., waste paint solids from overspray collected from downdraft booths as sludge). Some volatile organic compounds (VOCs) may also be hazardous air pollutants (HAPs). VOCs are known to contribute to ground-level ozone formation, whereas HAPs have direct health effects. Further, materials used are highly varied and processed in large volumes (i.e., immersion tanks are large enough to submerge an entire vehicle body). Finally, substantial energy is required by the automotive paint process (e.g., maintaining immersion tank process temperatures, heating curing ovens, powering robotic spray equipment). Detailed Life Cycle Inventory (LCI) data of material and energy use as well as emissions to air, water and land was recently published by Prendi et al. (2008) and Anastassopoulos et al. (2009).

Stricter environmental legislation, increasing costs of petroleum-based solvents, and competitive pressure to decrease the cost of application, are driving the automotive industry toward using innovative coatings and application technologies, while at the same time maintaining the durability and appearance of their products (Prendi et al., 2006). One of the technologies which would assist in achieving these goals is the use of powder coatings. A distinctive characteristic of powder coatings is minimal or zero VOC emissions. This is important in both regulatory and social aspects. In addition, the reuse of overspray can result in high overall material utilization. The reported increase in use of powder coatings is attributed to their durability, low cost and low VOC emissions (Challener, 2006; Crea, 1996). Powder coatings provide corrosion protection, improved durability, and better transfer efficiency (if overspray is reused) than liquid counterparts. In addition, powder coatings achieve a glossy finish similar to waterborne or solventborne paints. Thermosetting powder coatings are used in a large number of applications (Lee et al., 1999).

2

1.2 Vehicle Painting Phase and Materials Used

The automobile coating process provides corrosion protection to a vehicle’s body structure and a durable and decorative coating to the outer body panel surfaces. Automotive coating is a multi-step sequential process, typically involving the following unit processes: cleaning, pretreatment, e-coat, full body anti-chip primer, and topcoat (i.e., basecoat and clearcoat). This is illustrated in Figure 1.1. Other than pre-treatment, these processes employ thermosetting organic resins (binders) in which the resin polymer molecules chemically bond to other molecules then heat is applied during curing.

Note: Dehydration is needed only for waterborne basecoats

FIGURE 1.1 - STAGES IN AUTOMOTIVE PAINT APPLICATION PROCESS

The welded body panels, known as the body-in-white (BIW), go through a waterborne pre-treatment process that includes a number of dip and rinse stages. One aspect of pre-treatment is the phosphating stage, wherein the metallic surface is converted to a thin layer of metal phosphate, for further application of coatings. Then the application of e-coat is performed by cathodic electrodeposition of an organic resin to seal the metal and protect against corrosion. This anti-corrosion layer is applied by dipping the BIW in the coating material followed by curing in an oven. Before e-coating was used in vehicle coating, the main purpose of primers was to protect against corrosion. Now that e-coat has become a standard stage in coating, the purpose of the “primer” is to provide a barrier to penetration of stone chips, and provide a surface for the adherence of the topcoat. In the basecoat phase, the BIW is sprayed with either waterborne or solventborne paint to cover previous layers with a uniform colour. If the basecoat is solventborne then a short flash time at ambient temperature is needed for evaporation of volatile solvents. Waterborne basecoats however, need a longer flash time at elevated temperature to dehydrate the basecoat from 25-50% solids up to 75-90% solids. Then the vehicle body is sent to clearcoat application and final curing. The clearcoat can be either solventborne or powder and it imparts a glossy appearance to the car, while adding protection to the other layers for longer durability of the finish. The purpose and characteristics of each layer are summarized in Table 1.1.

The types of coating selected for painting a vehicle greatly depend on the desired appearance of the final finish, as well as its resistance to weathering, corrosion and chipping. The typical layering that is applied to a vehicle falls into one of the four combinations listed below. Since pre-treatment and e-coat material and

3

application technology (dipping) are the same for the majority of assembly plants, they are not included in these combinations.

TABLE 1.1 – PAINT PROCESSES IN A TYPICAL AUTOMOTIVE APPLICATION

Order of Application

Name

Purpose

Application Method

Dry Film Thickness

(µm) 1 pre-treatment • remove pressing oil

• pacify surface dip ~0.1 [1]

2 electrocoat (e-coat)

• seal metal from environment

dip 20-30 [2]

3 primer (anti-chip)

• UV protection of e-coat • stone chip resistance

spray 30-35 [2]

4 basecoat • imparts uniform colour spray 15-18 [2] 5 clearcoat • creates gloss

• UV protection for basecoat

• chemical resistance

spray 38-40 [2]

[1] Yasuhara, 2005 [2] Chiou, 1999

Combination 1: solventborne primer/solventborne basecoat/solventborne clearcoat

Combination 2: solventborne primer/waterborne basecoat/solventborne clearcoat

Combination 3: powder primer/waterborne basecoat/solventborne clearcoat

Combination 4: powder primer/waterborne basecoat/powder clearcoat.

Other combinations under investigation are the use of powder basecoats and elimination of the primer application stage. The latter has been developed by PPG and was tested at the Dodge Dakota plant in Campo Largo, Brazil (Davis, 2000).

1.3 Objectives and Scope

Finish quality is one of the most important aspects of automotive coatings. Anecdotal evidence suggests that customers perceive vehicle quality based on the finish appearance. As a result, it is important to achieve a finish that has high gloss and a smooth surface (mirror–like). While this is true for the outer part of the vehicle, it is not necessary for inner surfaces of a car (for example, areas under the hood). Recently, an important innovation has been the use of colorkey primers for areas of the car that do not require a high gloss appearance. Colorkey primers, if properly formulated and applied, could potentially eliminate the basecoat stage in these areas. This would result in significant reductions in

4

material use and process effort. Also the use of powder coatings in basecoat applications, instead of solventborne and waterborne coatings, could potentially reduce the amount of energy used for dehydration of waterborne basecoats, and at the same time reduce air emissions and wastewater generation through powdercoat reclamation and recycling. As indicated in the previous section, the use of powder basecoats could potentially eliminate the powder primer stage.

There are many factors that affect the finish quality, appearance and mechanical properties of automotive powder coatings, such as: paint formulation, application techniques, cure time and temperature, cure kinetics, rheological properties, and film thickness just to name a few.

Waterborne and solventborne topcoats have been used for decades in the automotive industry. The knowledge gained through experience combined with the extensive research findings, in relation to these coatings, has resulted in a vast pool of knowledge explaining their performance qualities. However, there is a gap in research when the appearance qualities of powder basecoats and colorkey primers are concerned. The studies reported have focused more on grouped appearance parameters such as long-wave, short-wave, glossiness and orange peel (OP). However, a detailed investigation of specific wavelength ranges has not been reported for powder coatings. This is important because knowing the surface quality after each paint process step would help to determine which layer is influencing a particular appearance parameter and this knowledge can be used to optimize the process by reducing troubleshooting. This becomes more significant if a specific layer is completely removed (i.e. apply clearcoat on top of colorkey primer and eliminate basecoat application). In addition, studies reported in the literature mostly have focused in the rheological properties of powder coatings as an important factor affecting leveling and consequently appearance. However, for day- to-day in-plant applications viscosity is difficult to measure, so there is a need to use more practical predictors to optimize appearance.

Objective

The intent of the proposed research is to investigate the finish quality of two powder basecoats (red and black), of three different particle sizes, and a colorkey primer (red) in relation to the process temperature and time as well as paint particle size. Datapaq value (DPV), a process parameter related to process time and temperature, was investigated on the effects it has on surface appearance. In addition, the effect of complex viscosity and degree of conversion on paint levelling and consequently on the appearance of the final coating were investigated as well.

The panels were coated at the Automotive Coatings Research Facility (ACRF). This facility has the necessary equipment to duplicate any paint application process on a full-sized vehicle. Initially a literature review was conducted in order to get a better understanding of the coating process, the parameters involved and the effect they have on surface structures. Focus was directed to factors

5

such as process temperature, time, and percent conversion, rheological properties and film thickness.

Scope

The scope of this research was divided in the following phases:

Phase 1 experiments were conducted with the red colorkey primer (RCP) and powder basecoats - red (RBC) and black (BBC) of different particle sizes (20, 25 and 30µm). Different curing scenarios were investigated using times and temperatures corresponding to the corners of the cure window. This stage looked at the effect of particle size and paint formulation on surface appearance as well as gave an initial indication of the effect of time and temperature.

Phase 2 experiments were conducted in two stages as follows:

a. Stage A – Isothermal experiments at three temperatures (163, 171, and 193 °C) with RBC25, BBC25, and BBC20 at varying cure ti mes. Comparisons were made between RBC25 and BBC25 in order to eliminate the effect of particle size. Also between BBC25 and BBC20 to eliminate the effect of formulation.

b. Stage B - Fixed time and temperature corresponding to significant points on the cure exotherm indicated by Differential Scanning Calorimetry (DSC).

The intent of Phase 2 was to determine significant parameters that influenced appearance and so would form the basis for model generation. Factors considered were time, temperature, particle size, Datapaq value (DPV - a unitless property that includes both time and temperature), viscosity and degree of conversion (α). Film build was maintained within a tight range for all the stages in order to minimize the effect of film build on appearance.

Phase 3 considered the effect of ramp rate (time to heat to final process temperature) on surface appearance. Experiments were conducted at fixed time and temperature, corresponding to specific points on the cure window, at different heating rates. The ramp was not considered in the model at this point, but the data collected will lay the ground work for future research.

Phase 4 of this research focused on generation of a model that would predict the finish quality of powder paints with different characteristics such as color, particle size, and cure conditions. The relationship between appearance and Datapaq value (DPV), viscosity and percent conversion were considered as part of the model. Finally, an investigation onto paint levelling was conducted in order to reveal the extent to which particles coalesce and flow on the painted surface. Correlations between levelling and appearance were explained, where data agreed, by known physical mechanisms such as surface tension, degasification and mechanics.

Table 1.2 gives in a matrix format the coatings and factors considered in each phase of this research. Approximately 800 e-coated panels were powder coated

6

for this research. For each scenario, final film build, surface undulations and adhesion properties were determined.

TABLE 1.2 – SUMMARY MATRIX OF RESEARCH PHASES

Paint Temperature Time DPV Viscosity FB Conversion Ramp Phase 1 – Cure Window Corner Points

RCP30 X X X

RBC25 X X X RBC30 X X X BBC20 X X X BBC25 X X X BBC30 X X X

Phase 2 - Stage A: Isothermal RBC25 X X X X X X BBC25 X X X X X X BBC20 X X X X X X

Phase 2 - Stage B: DSC Exotherm Points BBC25 X X X BBC20 X X X

Phase 3 – Ramp Speed Effect BBC25 X X X RBC25 X X X

Phase 4 – Model Generation RBC25 X X X X X BBC25 X X X X X BBC20 X X X X

7

CHAPTER 2 – LITERATURE REVIEW

2.1. Development of powder coatings

Powder coating technology and production methods have improved significantly over the years. An example is the ability to fabricate semi- or low gloss powder paints in addition to only glossy powder paints. Also, it is now possible to produce powder coatings with smaller particle sizes, which allows for a thinner film build and better appearance of the final film (Jones, 2001). The Powder Coating Institute reports that the use of powder coatings by the automotive industry increased by 10% during 1998 (Arjona, 2000). The global market for powder coatings is expected to increase at about 7% a year from 2008 (BCC Research, 2009). This is considered a higher growth than the general coatings market (SpecialChem, 2010).

The increase in use of powder coatings is because of their durability, low cost and low VOC emissions (Crea, 1996). Automotive liquid coating spray operations typically have a high first pass transfer efficiency (FPTE - the ratio of paint adhering to paint used) of 70-80%, but the overspray is disposed as sludge (Ansdell, 1999). Powder clearcoat costs more per kg than liquid clearcoat, and the FPTE is lower (60%) (Papassava, 2001b; Jones, 2001). However, in modern automotive applications, a single type of powder coating is applied in a dry spray booth, wherein the overspray is separated from the booth air by cyclones to be reused. This can result in an overall material utilization of 97%. In addition, the reduction of the need to treat and dispose of paint overspray results in further cost savings (Jones, 2001).

Powder coatings require less energy to cure than solvent and water–borne counterparts, because little make-up air is required in the oven due to the lack of volatiles emitted by powder coatings. So there are energy savings even though powder coatings require higher temperatures to cure. To make the most out of powder coatings it is desirable that coalescence, degassing and levelling as well as cross-linking happen at the lowest possible temperature and in the shortest time (Barletta et al., 2007).

Powder coatings can be applied in the form of dry powders, powder slurries and UV-cure coatings. Application is achieved either by dipping the part in a fluidized bed or by electrostatic deposition. The electrostatic spray is superior since it allows for thinner film applications, which may assist with the “orange peel” problem. However, fine particles pose the problem of clogging the equipment and reducing flowability. The use of flow additives in powder coating compositions significantly increases the flowability of paint and as a result the film formation is more uniform (Conesa et al., 2004).

Powder slurries are powder coatings dispersed in water and as a result fall in between the categories of waterborne and powder coatings. However, the chemical formulation is closer to that of powder coatings. The application as a powder slurry enables plants to use existing liquid spraying equipment. This

8

facilitates the use of low or zero VOC coatings while reducing equipment change-over costs. One problem with this type of coating is that the rheological properties are difficult to control. DaimlerChrysler used powder slurry clearcoat in the Mercedes Benz models marketed in Europe and seems to have overcome this problem (Jones, 2001).

Thermosetting powder coatings account for 80% of the powder paint market (Jones, 2001). Initially, the resins used in powder coatings were epoxy, and the cross-linkers used were dicyandiamide. In the 1970’s other compositions were introduced, such as epoxy/polyester, polyester/ triglycidyl isocyanurate (TGIC), and polyester/blocked isocyanate (resin/crosslinker). The last two types had better durability in weathering. Acrylic copolymers with vinyl ester monomers are a new formulation of powder coatings and have demonstrated the properties required for automotive clearcoats. Powder coating technology and production methods have improved significantly over the years, an example is the ability to fabricate semi or low gloss powder paints in addition to only gloss powder paints. Also, it is now possible to produce powder coatings with smaller particle sizes, which allows for a thinner film build and better appearance of the final film (Jones, 2001). Finally, advancements in powder paint application equipment have increased the FPTE, reducing overspray and material loss.

Even with the improvement in powder formulations and application technologies, powders coatings have mostly found application as anti-chip primers and their use as topcoats has not yet become a standard process. One of the reasons is that it is not clear if the substitution of liquid paint with powder is sufficiently superior to justify the modifications in the manufacturing plants necessary to accommodate the new powder technology. Thick powder films and poor flow during cure lead to “orange peel”. However, Scania has used powder as a basecoat since 1992 at one of its plants in Sweden where robots were used to apply powder basecoat to truck cabs. This application provided for the reduction of VOC emissions and creation of a safer working environment. Using robots to powder coat the trucks also allowed for flexibility of operations (Josefsson, 1997).

Another issue with powder paints is that high overall material utilization is only achieved due to recycling of the overspray. Changing colours necessitates purging the applicator and cleaning the spray booth and equipment so that the overspray of the next colour will not be contaminated prior to recycle. The purge and clean material represents a waste unless it can be stored for the next time that the initial colour is used. Cleaning the booth of residual paint is an operation that is not required in colour change of liquid paints because there is no recycle to contaminate. Kia et al. (1997) investigated powder coat reclamation. The overspray powder is reclaimed usually by using a dry booth system at a high operation cost. A wet booth system is applicable for small parts or irregular product lines where powder reuse is less important. One problem with wet booth systems is the poor wettability of powder, so there is need for a wetting agent in the water-wash in order to collect the powder overspray. They found polyoxyethylene-polyoxypropylene glycol is a suitable agent.

9

For the topcoat stages, solvent borne and waterborne basecoats, and solvent borne clearcoats are more common (Crea, 1996). In contrast, suppliers of smaller, non-glossy parts to the automotive industry such as brake calipers, door handles and interior trim, have been applying powder coatings to their products for a long time.

2.2. Characterization of Appearance by Surface Stru ctures and Wave-scan DOI

The appearance quality of powder coatings, and other decorative coatings in general, is characterized by waviness, which measures the levelling of paints, and by the reflective characteristic known as distinctness of image (DOI). These elements can be seen by looking at the cured film in two ways: 1. focusing the eye on the surface and 2. focusing the eye on the reflected image (BYK, 2008). By focusing on the surface, information is gained about structure size and form. Structures are seen as wavy pattern of light and dark regions. A surface with waviness (orange peel) looks similar to the texture of an orange (bottom left of Figure 2.1). The size range of the surface structures that cause the orange peel effect is between 0.1 and 100 µm (Biris et al., 2001). Orange peel (OP) is quantified using a rating between 0 and 100 (unitless). High values of the orange peel rating indicate that the paint film has a smooth finish (Haldankar et al., 2008).

FIGURE 2.1 - FOCUSING ON SURFACE AND WAVENESS PATTERN. REPRINTED WITH PERMISSION FROM SHERRY BROWN, BYK-GA RDNER

(COLUMBIA, USA). By focusing on the reflected image, the image forming quality of the surface is determined. Distinctness of Image (DOI) indicates the sharpness of the reflected image of an object. For example, if the contours of an object reflected on a smooth surface are sharp and with high contrast, then the structure is perceived

10

as distinct (Figure 2.2a) otherwise the image is perceived as blurry (Figure 2.2b). The same scale used for OP is used for DOI as well, higher values indicating a better appearance. Most automotive applications require high gloss values (above 80), in order to achieve a high DOI (Adams, 1989).

Distinct Image (2.2a) Blurry Image (2.2b)

FIGURE 2.2 - IMAGE FORMING QUALITY OF A SURFACE. REPRINTED WITH PERMISSION FROM SHERRY BROWN, BYK-G ARDNER

(COLUMBIA, USA). The wave-scan instrument was developed by BYK-Gardner (Geretsried, Germany) and simulates the visual impression obtained from optical inspection of surface structures. Structures are analyzed based on their size. A laser point source is directed at a 60 degree angle onto the painted surface and the reflected light is detected at 60 degrees opposite (Figure 2.3). The wave-scan measurement may be made at different locations on the panel and the scan length is about 10 cm with data points recorded every 0.08 mm. Mathematical filtering (Fourier transforms) is used to separate the data collected from the optical profile of the surface into different wavelengths ranging from less than 0.1 to 30 mm (Figure 2.3 and Table 2.1). Data generated by the wave-scan DOI are normalized values of each element. Values are standardized by assigning a value of 1000 to the average and calculating the contrast value. The contrast value is the average of the absolute value of the difference of each individual spectrum element value with 1000 (Equation 2.1).

Contrast = average |� − ��| (Equation 2.1)

The lower the contrast values of the wave-scan elements, the smoother the appearance (Osterhold & Armbruster, 2009). Analyzing the values of each component of the structure spectrum makes it possible to determine factors that affect the finish quality.

11

FIGURE 2.3 - WAVE-SCAN DOI PRINCIPLE OF OPERATION ( BYK, 2008) REPRINTED WITH PERMISSION FROM SHERRY BROWN, BYK-GA RDNER

(COLUMBIA, USA).

TABLE 2.1 - WAVE-SCAN ELEMENTS AND THEIR WAVELENGTH RANGE

Wave-scan Element Wavelength Range [mm] Dullness (du) <0.1

Wa 0.1 – 0.3 Short-waves Wb 0.3 – 1

Wc 1 – 3 Long-waves Wd 3 – 10

We 10 – 30

2.3. Cure Kinetics of Thermosetting Powder Coatings (DSC)

Differential Scanning Calorimetry (DSC) has been widely used to characterize powder coatings. DSC measures the rate of heat generated during a chemical reaction. A calorimeter is used for measurements of the heat flow as a function of temperature (Gherlone et al., 1998). The software generates a DSC thermogram (a plot of heat flow as a function of temperature) from which several important transitions during curing of powder coatings can be deduced such as the glass transition, curing and decompositions (TA instruments, 2008). From a typical thermogram the following characteristic points can be obtained: 1. the glass transition temperature (Tg – temperature where the powder coating changes from a glassy state to a soft or rubbery state), 2. Melting peak temperature Tm, onset flow temperature Tof, onset cure temperature (Toc), reaction peak temperature

12

(Tr) and the heat of reaction (∆Hr) which is found as the area under the reaction (curing) peak. A typical thermogram is shown in Figure 2.4.

This technique makes it possible to follow the curing process of powder coatings. The cure kinetic parameters can be obtained by running the DSC in isothermal or dynamic conditions. The reaction rate constant (k) determined by DSC experiments, when plotted against temperature in an Arrhenius type plot, yields the values for the activation energy and the pre-exponential factor. The cure kinetics of thermosetting powder coatings have been described by either empirical or mechanistic models The autocatalytic models are known to model the cure kinetics of epoxy based systems and the nth order models are used to model the kinetics of polyurethane systems (Lee et al., 1999). The assumption underlying the DSC measurements is that the extent of reaction or degree of conversion (α) is proportional to the change in heat generated by the chemical reaction (Lee et al., 1999; Mafi et al., 2005). So by applying the appropriate kinetic mode it is possible to calculate the degree of conversion. The degree of curing can be checked by using the residual exothermal approach. A sample is subjected to DSC analysis at isothermal conditions in order to get the total heat of reaction. Then after cooling the cured sample is run again through the DSC analysis. The residual heat of reaction is proportional to the degree of undercuring (Gherlone et al., 1998).

FIGURE 2.4 - DYNAMIC DSC THERMOGRAM

-2.0

-1.5

-1.0

-0.5

-0.0

0.5

He

at F

low

(W

/g)

-100 0 100 200 300 400 Temperature (°C)

DSCSample: Red BC R25Size: 5.1100 mgMethod: TG EVALUATION

File: H:\4 Insight\Chan\DSC KINETICS\20°C\dsc-Lindi R25-060509.010Operator: cawRun Date: 5-Jun-09 11:27

Exo up T A I nstruments

203.19°C254.15°C

138.0°C

284.2°C

Heating Rate: 20.10 °C/minOnset Slope: -38.05 mW/°COnset Offset: 0.00 °CCell Constant: 1.954Onset Temperature: 146.4 °C

13

There are several approaches used by the software accompanying the DSC device to automatically calculate the reaction order, activation energy, rate constant and pre-exponential factor. Some of the most commonly used are Borchardt and Daniels (B/D), and the ASTM E-698 Thermal Stability and Isothermal Kinetics (TA Instruments, 2009). The B/D kinetics assumes a reaction of the nth order and Arrhenius behavior. Predictive plots generated by this method are isothermal plots (plots that give information about the degree of conversion with time at isothermal conditions) and iso-conversion plots (plots that provide time/temperature conditions for specific conversion levels). The ASTM approach requires three experiments at different heating rates. It also assumes Arrhenius behavior but the reaction kinetic is assumed first order. Another assumption of this method is that the extent of reaction at the peak exotherm is independent of heating rate and it is constant. The isothermal approach is mainly used for autocatalyzed powder coatings since the B/D and the ASTM methods fail to characterize the cure kinetic of autocatalytic systems. However, the isothermal method can be used for nth order systems as well (TA Instruments, 2009).

The kinetic parameters are very useful in predicting the cure behavior of powder coatings. Specifically, the degree of cure (α) has been widely used to predict the properties of the final cured powder coating such as appearance and durability.

2.4. Factors Affecting Finish Quality of Powder Coa tings

Several research studies have investigated factors that affect the finish quality of powder paints. Factors such as powder application techniques, particle size and particle size distribution (PSD), rheological properties, film thickness, relative humidity (RH), formulation, orientation of surfaces, substrate roughness, and cure conditions are the ones that are found more often in the literature. Biris et al. (2001) also reported that the OP and gloss of powder coatings was improved by controlling electrostatic spray application (i.e. corona voltage). Since the electrostatic spray application technique was used for this research, literature review on other application techniques is not relevant and will not be discussed any further. The particle size and particle size distribution have been reported to impact the surface appearance (Biris et al., 2001; Sims et al., 1999). The film smoothness increased as the mean particle size of the powder coatings decreased (Nix and Dodge, 1973; Yanagida et al., 1998). Kenny et al. (1996) investigated the appearance of acrylic clear powders by analyzing the stages that powdercoat undergoes, until a final film was formed. Panels were treated with electrodeposited primer and solventborne basecoat before the powder clearcoat was applied. The powder was applied using an electrostatic spray system. The powder’s glass transition temperature (Tg) was equal to 60°C and melting temperature (Tm) was 129°C. Panels were baked at 150°C for 25 minu tes. Film thickness was 80µm and the particle size of powders ranged from averages of 10

14

- 40µm. Film appearance was determined by measuring image clarity with a Nippon Paint-Suga Test Instrument (NSIC). Equipment responding to “long-waves” (100s-1000s µm) and “short-waves” (10s-100s µm) was used to determine the image clarity. It should be noted that the range for “short-waves” and “long-waves” reported in the literature differs between authors. Surface roughness was assessed using a needle stylus type device that measured Ra. It was found that as average particle size (D50) increased, image clarity decreased (worse appearance). This corresponded to an increase in the value of Ra. Powder coatings with an average particle size of 10 µm showed 15% better image clarity than 30 µm powders. Also, scanning electron microscopy (SEM) measurements showed a smoother surface with less irregularities for 10 µm (D50) powders than for 30 µm (D50) ones. The appearance of the 10 µm powder film was found to be better than the appearance of solventborne acrylic clearcoatings used by the automotive industry (thickness of 40 µm). Smaller particles showed better coalescence as well and fewer defects such as pinholing or bubbling. Coalescence occurred faster on smaller particle size powders since they have lower heat capacity and require less time for melting. As a result, degassing occurs earlier in the curing process and so bubbling is reduced. If degassing occurs during later stages of cure, reflow tends to cover the defect, but the closer to gelation point the more difficult it is to cover the defect. Also, it was found that a narrower PSD gave better appearance. This was in contrary to the findings by Nix and Dodge (1973), who reported that a wide distribution of particle size gave a better surface appearance than a narrow PSD. In this dissertation a narrow PSD is considered a value of less than 2 for the average of d50/d10 and d90/d50. It was found that cooling did not affect the final finish appearance. An interesting finding is that image clarity improves for the first 5 minutes of heating. Image clarity decreases as short waves develop at about 6 minutes of heating which is the time of cure onset. Then, for the duration of cure the image clarity does not improve. Some problems encountered with fine powder particles were: irregular feed rates due to the low ability to flow, low fluidization, sintering during storage (Kenny et al., 1996) and clogging the equipment (Conesa et al., 2004). Viscosity is an important characteristic of powder coatings and it has been correlated to properties of the final cured film. Rheological behavior is one of the factors affecting the final properties of final finish. Osterhold and Niggemann (1998) studied pigmented and non-pigmented epoxy-based powered coatings and their relationship to rheological properties. The powders were prepared in the lab. From the viscosity-temperature relationship measurements it was found that an increase in heating rate corresponded to a decrease of the minimum viscosity level and the minimum viscosity shifted to higher temperatures. The authors also investigated the effect of viscosity on surface appearance. It was found that increased viscosity corresponded into an increase of long-wave values of the final finish. The authors used powder coatings containing epoxy resins with dicyandiamide as the hardener. The levelling agent was polyacrylate absorbed on silicon dioxide and titanium dioxide treated with aluminum and silicon compounds was the pigment. It is reported in the literature that as the cure

15

temperature increases, the viscosity of powder coatings decreases. As the temperature continues to increase, viscosity reaches a minimum value and then increases with further increases in temperature. The latter is attributed to crosslinking. The thickness required in industrial applications ranges from 60 to 100 µm and the curing temperatures range from 150 to 200°C. Rheological behavior of powder coatings during cure, significantly affects properties of the final film. The viscosity-temperature relationship was highly dependent on the heating rate selected. It was found that high heating rates lead to low minimum viscosities and shift the viscosity minimum to higher temperatures. The glass transition temperature as measured with DSC was reported to range between 52 and 56°C. To study the relationship between surface appearance and viscosity, coated panels were cured at 180°C at the same heating rates as rheological measurements. The film build was 90 µm and cure time 5 minutes. BYK-Gardner wave-scan was used to measure the surface appearance. It was found that long-waviness increases with minimum viscosity and that pigment amount impedes the flow process. Higher heating rates resulted in lower long-wave values and better appearance. Short-wave values tended to increase as the long-wave values decreased with higher heating rates (Osterhold & Niggemann, 1998).

Hannon et al. (1976) also reported that rheological properties can be used to characterize the film appearance. In their investigation of powder coatings, the authors suggest that complex viscosity can be used to explain surface appearance and mechanical properties of coatings. The authors report that at the beginning of cure, complex viscosity is a measure of viscosity that, combined with surface tension, determines finish appearance; while later in the cure, complex viscosity is a measure of modulus that reflects the degree of crosslinking and so indicates mechanical properties. Smooth films are achieved at low viscosity values; however the strength of the film is compromised. Others also report that low viscosity at lower temperatures allows cross linking to occur at a slower pace for acrylic powder coatings, and as a result allows for better levelling of the final film. However, the low viscosity is not sufficient for a complete coalescence to occur in an efficient time frame (Andrei et al., 2000).

Lee et al. (1999) studied the surface structure build up of two thermosetting powder coatings (epoxy and polyurethane based) and its relationship with viscosity. Matting (low gloss) and wrinkling were characteristic of the paints investigated. According to the authors, powder coatings that produce a controlled wrinkled surface and low gloss are not fully developed and this may be a reason why liquid paints dominate the market. Glossiness of the final finish is affected by smoothness or roughness of the film structure. Differential scanning calorimetry (DSC) was used to study the progress of cure at isothermal conditions (130, 140, 150 and 160°C). Complex visco sity measurements were conducted with a torsional parallel-plate rheometer at the same isothermal conditions as DSC experiments and the frequency was 10Hz. Rheological parameters reported are the minimum complex viscosity (η0*) and the gelation time tgel which is the time when the storage modulus G’ is equal to the loss modulus G” (Lee et al., 1999). The viscosity storage modulus G’ measures the

16

elastic behavior of material and the loss modulus G’’ measures the dissipation of the material (amount of deformation energy converted to heat) (Osterhold and Niggemann, 1998). The surface structure was investigated with a laser scanning microscope and reflective optical microscope (Lee et al., 1999). The thickness of coatings was 100 to 150 µm and powder was sprayed on glass panels. Sprayed panels were cured isothermally at 130, 140, 150, 160 and 200 °C and for different time intervals. Rq, the root mean square deviation was used to characterize the surface roughness. The authors found that with increasing temperature, a decrease in minimum viscosity was observed. For epoxy-based coatings it was found that the drop in viscosity was substantial as the cure progressed. Complex viscosity and conversion were correlated for the two paint systems investigated. Also, it was found that gelation occurred at 45% conversion for both paint systems investigated. Surface roughness is one of the factors that affects surface appearance of coatings. The following was observed in relation to surface topography: during the melting stage the surface was rough, after melting and flowing, the surface was very smooth and with further curing the surface was rough again. Authors report that for the polyurethane paints post-curing fine structures are developed before coarse structures. The minimum viscosity and minimum surface roughness occur at the same conversion. Also, it was noted that the fine structures developed even after gelation. For the polyurethane powders, build up of fine structures was observed at 32% conversion and was developed after the gelation point (Lee et al., 1999). The authors also report that cooling has no effect on surface appearance, once the surface structures have been created.

Nix and Dodge (1973) investigated the rheology of thermoplastic acrylic powder coatings, but also looked at some thermosetting polyester and epoxy coatings. Their assumption was that flow time was a function of viscosity, average radius of curvature and surface tension. Powder coatings are not continuous materials when unbaked. The particle dimensions are comparable to the final film thickness and as a result these clusters of particles are larger than the polymer molecules. The major assumption that the authors made was that surface tension was the main factor affecting flow. The data from surface profile measurements indicated that viscosity was not the only factor that controlled powder flow. It was found that values of low orange peel (better surface appearance) did not correspond with the lowest melt viscosity. The authors suggested that particle agglomeration was one of the reasons for orange peel appearance. Pigment dispersion within resin was thought to affect viscosity- the higher dispersion the lower viscosity (Nix and Dodge, 1973). The authors reported that appearance of orange peel was also very much affected by the original profile of the dry sprayed powder. If, after spray and before bake, powder was deposited in clusters or agglomerates, then these areas became peaks of the orange peel structure. One factor causing cluster formation might be electrostatic attraction or repulsion of particles and this agglomeration might occur before spraying, during travel or upon deposition of powder particles. Using a positive potential for deposition showed a decrease in clusters. Also, the least clustering was achieved at zero potential by frictional charging of particles. It was

17

also found that trapped air will increase melt viscosity and so increase surface undulations (orange peel). Also, smaller particles have better flow. Higher heating rates have better flow as well. It was observed that 5 micron particles clung to 40-50 micron particles causing clustering (Nix and Dodge, 1973).

Cure conditions (time and temperature) are considered to greatly influence surface appearance. It has been reported that curing temperatures range between 150 to 200°C for powder coatings with film thicknesses between 60 to 100 µm (Osterhold & Niggemann, 1998). Hannon et al. (1976) reported loss in gloss values for epoxy-based powder coatings at cure temperatures of 120, 168 and 190°C with cure times ranging from 1 min to 20 minutes. The decrease in gloss at high cure temperatures and prolonged time could be attributed to wrinkling of the film by over-curing (Hannon et al., 1976). Barletta et al (2007) reported similar findings: that relatively smooth film with high gloss were obtained at low baking temperature and time, for epoxy-based powder paints with a mean particle size of 20 µm and film thicknesses ranging from 60 to 80 µm. The curing temperatures ranged from 100 to 200°C and for each temperature the cure times were 5, 10, and 20 minutes. They found that high glossy coatings are established by curing at 175°C for 10 minutes. Even though smoo th films were achieved at relatively low baking time and temperature, mechanical properties were not fully developed. While at high temperature and cure time, adhesion and film strength were much better. Baking time and temperature are very important factors affecting paint finish quality and durability, together with paint formulation (Barletta et al., 2007).

The finish quality of powder coatings is affected by film thickness and surface orientation during bake. Andrei et al., (2000) reported that paint levelling is very sensitive to film build values below 300 µm. The thinner the film the more difficult for the paint to level (Biris et al., 2001; Andrei et al., 2000). Increasing clearcoat film thickness up to an optimum value is found to decrease Wc and Wd values which are responsible for the “orange peel” defect (BYK, 2008). The increased film thickness improves flow and levelling of melted powder paint. It was found that vertical surfaces have poorer flow and levelling than horizontal ones. This corresponded to higher long-wave contrast values for vertical surfaces as compared to horizontal ones (BKY Gardner, 2008). Osterhold (1996) similarly reports that bake position affects the long-waves, with horizontally baked substrates yielding smother appearance. The short-wave values were not found to depend on bake position (Osterhold, 1996).

Formulation is the other important factor that affects appearance. Uhlmann & Grundke (2001) studied the appearance of final film as a definitive factor for determining the quality of powder coatings. Stages of film formation such as coalescence of particles, wetting of substrate as the powder coating melts and flow to the surface are important and determine if the powder coating is of good quality. It was found that the addition of commercially used additives in powder paint formulations affected the wetting tension of epoxy resins, but did not have a significant effect on the viscosity. The powders used in this study however, were

18

only binder mixtures and not real powder coatings, which are very complex systems (Uhlmann & Grundke, 2001). The use of flow additives in powder coating compositions significantly increased the flowability of paint and as a result the film formation was more uniform (Kenny et al., 1996). Pigmented powder coatings are reported to have lower appearance qualities as opposed to non-pigmented ones (Osterhold and Niggemann, 1998).

Basu et al. (2005) studied wrinkling in low gloss epoxy and polyester powder coatings. The authors indicate that methylene disalicylic acid (MDSA) crosslinker and amine-blocked Lewis acid catalysts are needed for wrinkle formation. DSC (for reactions) and mechanical profilometry (for surface roughness) were used to monitor the reactions that affected wrinkle size. It was found that the amplitude and wavelength of the wrinkles were lower when the concentration of the catalyst was high. The authors describe wrinkle formation as a step process. As the powder coating is heated it develops a mechanical skin in the top layer, since this layer cures first. The layers beneath are still viscous. This skin is attached to the substrate at the edges by cured coating – edges solidify first. As the coating is heated the resin molecules start bonding and the skin shrinks by following its stress-free state – a state that the skin would have if it was allowed to flow without obstructions. However, the skin is attached at the edges to the substrate and as a result an in-plane tensile stress is developed. At the same time the layers underneath the skin have unreacted material of low molecular weight. Swelling occurs as the skin absorbs material from underneath layers. Swelling generates compressive in-plane stress since the skin cannot swell freely. “This reverses the tensile stress at first present in the skin and it must grow compressive enough to buckle the skin and thereby produce wrinkles”. Coating powders were produced in the lab containing different formulations with varying concentrations of MDSA and catalyst. Particle sizes used were smaller than 120 µm. Electrostatic spraying was done using a cup gun on cold-rolled steel panels. Panels were cured for 10 min at 190°C and the filmbuild ranged from 75 to 110 µm. The gel time is the time at a certain temperature where the material transforms from a dry solid to a liquid state to a specified gel-like condition (Bate, 1990). Gloss was measured at 60 degree angle using the ASTM D 523 method (Basu, 2005). Samples that had epoxy resin, MDSA and catalyst produced the most wrinkles. While those with resin and catalyst only and resin and MDSA only were smooth and glossy. Using the formulation that contained both catalyst and crosslinker as a base formulation, the authors tested the size of wrinkles varying the amount of catalyst while maintaining crosslinker amount constant and vice versa. It was found that low catalyst amount gave coarser wrinkles. The concentration of MDSA did not affect the wrinkle wavelength. However increasing the amount of crosslinker increased the amplitude of wrinkles at first and then it decreased after going through an optimum. Wrinkles depend on curing rate, oven conditions, filmbuild and substrate conditions. For all coatings investigated it was observed that they cured faster at the surface compared to the layer near the substrate. It was found that skin formation was present in all samples that produced wrinkles. Wrinkling was not present in powder coatings with unblocked catalysts (Basu et al., 2005). Other authors describe wrinkling as

19

a phenomenon that occurs due to over-curing (Huang et al., 1997; Luciani et al., 2001).

There is substantial amount of research related to thermosetting powder coatings. However, research is lacking when it comes to correlating specific surface structures such as dullness (du), and Wa to We with different cure factors. Also, to date there is no research conducted to relate appearance to the Datapaq value (DPV), also known as Index of Cure Value, which is used in industry for process development and optimization. The intent of this research is to add to the pool of research findings related to thermosetting powder coatings by considering factors that are not fully investigated. Specifically, possible correlations between wave-scan spectrum elements and DPV as well as temperature and time will be investigated.

CHAPTER 3 – MATERIALS AND METHODS

3.1. Testing Facility

The success of a research project is greatly dependant on the environment in which experiments are performed. The access to research facilities that closely resemble the manufacturing process adds a degree of practicality to the results.

The work described in this dissertation was conducted at the University of Windsor / Chrysler Canada Automotive Research(ARDC). This facility contains the Automotive Coatings Research Facility (ACRF), a full-size, multiresearch in automotive coating process development. The important aspect of conducting the experiments at ARDC is the simulation of plant conditions, and control of the environmental factors such as booth temperature and humidity.painting stages are the same as in automotive assembly plants. The oven, conveyer, and spray booths are large enough to accommodate the painting of an entire car, a feature that is not possible in lab experiments.

3.2. General Application The powder was applied using electrostatic spray deposition. fluidized in a hopper. Compressed air was used to transport the fluidized powder from the hopper to a illustrated in Figure 3.1and sprayed to grounded panels.

FIGURE 3.1 -

MATERIALS AND METHODS


d in this dissertation was conducted at the University of Windsor / Chrysler Canada Automotive Research and Development Centre

This facility contains the Automotive Coatings Research Facility size, multi-stage coating facility designed to support major

research in automotive coating process development. The important aspect of conducting the experiments at ARDC is the simulation of plant conditions, and control of the environmental factors such as booth temperature and humidity.painting stages are the same as in automotive assembly plants. The oven, conveyer, and spray booths are large enough to accommodate the painting of an entire car, a feature that is not possible in lab experiments.

Application Parameters

powder was applied using electrostatic spray deposition. The powder was hopper. Compressed air was used to transport the fluidized powder

RPA-1 Powder Applicator (ITWGema, Model #illustrated in Figure 3.1, mounted on an ABB robot arm. The powder was charged and sprayed to grounded panels.

RPA-1 POWDER APPLICATOR (ITWGEMA, 2008)

20


d in this dissertation was conducted at the University of and Development Centre

This facility contains the Automotive Coatings Research Facility designed to support major

research in automotive coating process development. The important aspect of conducting the experiments at ARDC is the simulation of plant conditions, and control of the environmental factors such as booth temperature and humidity. All painting stages are the same as in automotive assembly plants. The oven, conveyer, and spray booths are large enough to accommodate the painting of an

The powder was hopper. Compressed air was used to transport the fluidized powder

Model #: A11200) ABB robot arm. The powder was charged

(ITWGEMA, 2008)

21

Table 3.1 shows program information that was maintained constant throughout the project. Booth temperature and humidity were monitored before each run to ensure consistency. Several trial and error runs were performed in order to establish process parameters that would yield the desired film build.

TABLE 3.1 - POWDER ZONE SETTINGS AND CONSTANT PARAM ETERS

Item Value Powder Booth temperature 21.1 ± 2°C

Powder Booth humidity 55 ± 5% RH Tip Speed (TS) of robot 375 mm / sec

Shaping air (SA) 20 slpm Downdraft 50-60 fpm

Dilution air (DA) 45slpm Current set point 20 µA

Powder Flowrate (FR) 180 g /min 3.3. Paint Types

Five thermoset polyester basecoat powders and one thermoset polyester primer with BisphenolA/epichlorohydrin epoxy cross-linker [hybrid] were used in this research. Powder paint coatings used are summarized in Table 3.2. These paints were prepared by PPG to have the nominal particle size indicated.

TABLE 3.2 - TYPES OF POWDER COATINGS

Paint Type

Formulation

Nominal Particle Size, d50

[µm]

Abbreviation

PPG-High Gloss Black PZB90100 PCV-8555

Polymer, aliphatic hydrocarbons, amine,

resin, pigment

30 BBC30 25 BBC25 20 BBC20

PPG - High Gloss Red PCV-8555

Polymer, aliphatic hydrocarbons, amine,

resin, pigment

30 RBC30

25 RBC25

PPG – Colorkey Primer (anti-chip) PCV60109

Metal oxide, Urethane resin, polymer, filler, ester

and acid

26

RCP26

Cure time and temperature are important in the paint curing process. Paint manufacturers specify the time/ metal temperature combinations in the form of a cure window. The cure window for the above paints is shown in Figure 3.2. All coated areas should be cured within the time and temperature window in order to yield the proper appearance and mechanical properties of the final finish.

22

FIGURE 3.2 - CURE WINDOW OF POWDER COATINGS

3.4. Measurement Procedure and Instruments

Several instruments were used to measure the variables of interest. They are outlined in the following sections. 3.4.1. Particle Size Analyzer (PSA)

For each powder coating used in this investigation particle size of the virgin material was analyzed with a Beckman Coulter Laser Diffraction Particle Size Analyzer (Model LS 13 320). This instrument has ±1% reproducibility. To see if there was a change in particle size due to electrostatic application, powder coating was collected from an uncured panel after spray and analyzed. The mean particle size for each virgin material together with standard deviation and coefficient of variation values are summarized in Table 3.3. Particle size analysis for after spray powder samples are reported in Chapter 4.

TABLE 3.3 - SUMMARY OF PARTICLE SIZE MEASUREMENTS B EFORE SPRAY

Paint Type / (Abbrev.) D50 [µm] S.D. [µm] C.V. Black Basecoat (BBC20) 20.28 8.63 41.3% Black Basecoat (BBC25) 23.92 11.35 45.4% Black Basecoat (BBC30) 28.12 13.11 45.1% Red Colokey Primer (RCP25) 26.66 12.00 43.6% Red Basecoat (RBC25) 23.42 11.17 45.4% Red Basecoat (RBC30) 29.76 13.71 44.8%

320

330

340

350

360

370

380

160

165

170

175

180

185

190

195

0 20 40 60 80 100

Tem

pe

ratu

re,

T [

F]

Tem

pe

ratu

re,

T [

°C]

Time, t [min]

90 @ 163

9@193 C

14@163 C

35@193 C

Nominal Cure

23

3.4.2. Viscosity

Viscosity was measured for Red Colorkey Primer (RCP25), Red Basecoat 25 (RBC25) and Black Basecoat 25 (RBC25). Measurements were conducted by the PPG Physical Chemistry Lab, Coatings Innovation Center, PA, USA. The complex viscosity (η) was measured as a function of time at 163°C using a Paar Physica UDS200 stress-controlled rheometer with a 25 mm cone plate tool and a 0.5 mm gap. The sample was loaded onto the rheometer bottom plate at 80°C. The oscilliation mode was used at a frequency equal to 1Hz and a strain of 5%. Two measurement intervals were used as follows:

• Interval 1 (total 35 pts collected, 0.2 min per point @ 5% strain and 1 Hz frequency, temperature ramped from 80°C to 163°C in a log scale)

• Interval 2 (total 500 pts collected, 0.2 min per point @ 5% strain and 1 Hz frequency, constant temperature at 163°C)

A new specimen was used for the other two temperatures: 171 and 193°C. The procedure was repeated for each powder coating.

Complex viscosity is η* = G* / ω, where G* = G’ + iG” is the complex modulus and ω is the frequency. The following parameters could be determined from the chemorheology data:

i. Induction time: time required to reach minimum viscosity, ii. Gelation time: time at which G’ = G”, iii. Vitrification time: time when viscosity levels off after the dramatic

increase during cure and iv. Minimum viscosity value.

The UDS 200 is calibrated on a periodic basis for temperature and heat flow with a viscosity standard oil. The stated accuracy is ±1% of the maximum value. From experience, the repeatability of rheological data varied widely with the type of test and materials. At low shear rates [<1 /s] of flow curve measurements, the % difference between runs could be as high as 50%. The error decreased to about 1% at higher shear rates of 100 /s or higher (A. Tse, email communication, September 2009). 3.4.3. Cure Kinetics (DSC)

Cure kinetics of five basecoats and one primer powder coating were analyzed using Differential Scanning Calorimetry (DSC). The samples were sealed in an aluminum hermetic pan and then scanned in a TAI DSC 2920 (Barcode# C10671) in nitrogen from 25°C to 300°C at the heati ng rates of 5, 10 and 20°C/min. A sample size of approximately 5.5 mg was used. The DSC was calibrated on a periodic basis for temperature and heat flow with indium and High Performance Liquid Chromatography (HPLC) grade water standards and the nominal nitrogen purge rate was 50 ml/min. The internal standard operating procedures of calibration followed closely that of ASTM E967 Standard Practice for Temperature Calibration of DSC & DTA and E968 Practice for Heat Flow Calibration of DSC. The melting endotherm and cure exotherm were analyzed by

24

using a linear baseline. Measurements were conducted by the PPG Physical Chemistry Lab, Coatings Innovation Center, PA, USA.

The specifications of the DSC 2920 cell from TA Instruments were: Temperature repeatability +/-0.1°C Calorimetric sensitivity 0.2 µW [rms] Constant calorimetric sensitivity +/-2.5% from -100 to 500°C Baseline noise 0.1 µW [rms] From experience, the repeatability of temperature measurements was usually +/-1°C. The variations in the heat flow were larger a s most endotherms / exotherms were broad. Peak areas or heat flow could vary by the method of peak integration. The error was at least 5% but could be as high as 10 to 20% in some cases (A. Tse, email communication, September 2009). 3.4.4. Wave-scan DOI

The wave-scan DOI instrument (BYK Gardner) was used for appearance measurements by rolling the device over the panel surface three times in the same direction and averaging the values. Panels were scanned before and after spray. It is known that surface structure of the substrate has an influence on the final appearance quality and that is the reason panels were scanned before spray. Repeatability of this device is 4% and reproducibility is 8% (BYK, 2008).

3.4.5. Film Build

A model 355 Elcometer was used to measure the film build, before and after spray. A plastic template (25.4cm x 25.4 cm) with 9 equally spaced holes was used to ensure that the film build was measured at the same representative locations on each panel. Calibration foils provided by the manufacturer were used to calibrate the device before measurements were conducted. The film build in each hole of the template was measured using the conductive probe of the instrument. After 9 measurements, the device calculates and displays the average film build value. The accuracy of the measurements is ± 1% (Elcometer, 2005). 3.4.6. Scribe and Gravel Tests

The adhesion test was performed using the scribe test method as per ASTM procedure D 3359-02: Standard Test Methods for Measuring Adhesion by Tape Test. An Etching Pen/Scriber tool and 3M Scotch tape #898 were used to conduct the test. The gravel test was performed using the Gravel Conformance Rating (Appendix 1). After the panel was bombarded with gravel of different sizes and speed, the number of chips (defects from impact) were counted and their size measured.

25

Based on the number of chips and their size the panels were rated as pass or fail. 3.4.7. Datapaq

The Datapaq (Datapaq Inc., Wilmington, USA) was used as an oven tracker. Several trial and error runs were conducted to establish the time needed for the panel to reach the desired temperature as indicated by the cure window. The Datapaq records the temperature every 5 seconds at thermocouples attached to the unit. Thermocouples measure the metal temperature of the panel and the ambient temperature of an electrical batch oven, Despatch LFD Series Model #34G6. Three points of time/temperature combinations that were on the same cure curve were provided by the paint manufacturer. These values were entered in the Datapaq software which calculated the Datapaq value (DPV). A discussion about DPV together with the equations used to calculate it can be found in Chapter 6. 3.5. Experimental Procedure

A summary of overall experiments for each phase is summarized in a matrix format and included in Appendix 2.

3.5.1. Phase 1 - Cure Window Corner Points Experiments

All six powder coatings were used for these experiments. The cure conditions are summarized in Table 3.4 and correspond to the corners of the cure window provided by the powder coating manufacturer. In Table 3.4 the first letter of the cure condition coding indicates a low or high temperature (L, H) and the number after the “T” indicates shorter or longer curing times (T1,T2) at constant temperature. Three runs were performed for each cure window condition in order to ensure statistical validity of the data. In total, 15 runs per paint type were performed. Panels used for the appearance experiments (4 panels per run) were 25.4 x 25.4 cm (10x10 in) e-coated steel panels. One 10.2 cm x 30.5 cm (4x12 in) e-coated steel panel was sprayed and cured at the same time as the four appearance panels. The upper half of the 10.2 x 30.5 cm panel was used for the scribe test (ASTM D 3359-02), after which the panel underwent the gravel test, which was performed on the whole panel. However, only the lower half was considered for gravel readings (i.e. the panel was not cut in half). Another 10.2 x 30.5 cm e-coated panel was sprayed but not cured and this was used to collect powder for particle size analysis. Panels were mounted on the panel rack in the vertical orientation as shown in Figure 3.3. The panel rack was wrapped in aluminum foil to avoid contamination from paint and allow easy clean up.

TABLE 3. 4

Cure Condition

Temperature (⁰C)

Time (min)

Since 5 panels (four 25.4 the batch electric oven at one temperature of all panels was consistent, 5 probes were attached at the back of 5 e-coated panels and the 6one of the panels. The ambient air probe was used to monitor the temperature in the oven air through an extended tube (Figure 3.5panel since there is no other 16 trials to establish the specified of Datapaq trial 14. The heating rate was controlled and the amount of time needed to reach target temperature was determiexample, the upper cure window temperature needed 7min for the panel to reach 193°C. So panels were put in the oven and timed for 7min, then the panel was left to cure for the specified amount(HT1 = 9’@193°C, HT2=35’@193°C). conditions and for each replicate. Socorners + nominal). For each of these conditions there were 3 replicates (15 runs in total). This was repeated for each powder type (5 basecoats and 1 colorkey primer). Within each run there were 4 panels sprayed (4 replicates within one run). Therefore, there panels.

FIGURE

4 – CURE CONDITIONS FOR ALL PAINT TYPES

LT1 LT2 CNominal HT1

163 (325 F) 171 (340 F) 193 (380 F)

14 90 20 9

Since 5 panels (four 25.4 x 25.4 cm and one 10.2 x 30.5 cm) would be cured in oven at one time (Figure 3.4), in order to check if the metal

temperature of all panels was consistent, 5 probes were attached at the back of 5 and the 6th (oven ambient air) probe was attached at the front of

one of the panels. The ambient air probe was used to monitor the temperature in gh an extended tube (Figure 3.5). It has to be attached to a

panel since there is no other away to support it in the oven. There were in total the specified oven conditions. Appendix 3 shows an output

trial 14. The heating rate was controlled and the amount of time needed to reach target temperature was determined by Datapaq

the upper cure window temperature Datapaq trials indicated that it for the panel to reach 193°C. So panels were put in the oven and

timed for 7min, then the panel was left to cure for the specified amount(HT1 = 9’@193°C, HT2=35’@193°C). The same was done for the other cure conditions and for each replicate. So, in summary there were 5 cure conditions (4 corners + nominal). For each of these conditions there were 3 replicates (15 runs

is was repeated for each powder type (5 basecoats and 1 colorkey primer). Within each run there were 4 panels sprayed (4 replicates within one

were 360 appearance panels and 180 scratch/gravel

FIGURE 3.3 - PANEL RACK SET UP FOR RUNS 1- 15

26

ALL PAINT TYPES

HT2

193 (380 F)

35

cm) would be cured in , in order to check if the metal

temperature of all panels was consistent, 5 probes were attached at the back of 5 (oven ambient air) probe was attached at the front of

one of the panels. The ambient air probe was used to monitor the temperature in ). It has to be attached to a

away to support it in the oven. There were in total conditions. Appendix 3 shows an output

trial 14. The heating rate was controlled and the amount of time Datapaq trial runs. For trials indicated that it

for the panel to reach 193°C. So panels were put in the oven and timed for 7min, then the panel was left to cure for the specified amount of time

ame was done for the other cure in summary there were 5 cure conditions (4

corners + nominal). For each of these conditions there were 3 replicates (15 runs is was repeated for each powder type (5 basecoats and 1 colorkey

primer). Within each run there were 4 panels sprayed (4 replicates within one 60 appearance panels and 180 scratch/gravel test

15

27

For each run the procedure outlined below was followed:

1. Run Datapaq trials to determine ramp time. 2. Label panels indicating run number and condition type 3. Measure film build of e-coated panel before spray using the Elcometer

and record the value (average and standard deviation SD of 9 measurements).

4. Wave-scan the panel before spray. Each panel was measured three times. The wave-scan was rolled across the surface of the panel in the orientation of the paint application.

5. Check powder booth environmental conditions. 6. Fluidize powder paint. 7. Mount panels on the panel rack. 8. Spray powder coating in two passes. 9. Remove four 25.4 x 25.4 cm panels and one 10.2 x 30.5 cm from the

panel rack and place them on the holding rack. Figure 3.4 shows the holding rack with two panels.

10. Scrape an amount of powder from the other 10.2 x 30.5 cm panel. The powder collected this way was then analyzed with the PSA to see if there was any difference in particle size before and after spray. One sample of the virgin material was also analyzed.

11. Place holding rack with the panels in the batch electric oven and cure for the target time and temperature. The oven must start at 32°C (90°F). A stop watch was used to measure the time.

12. Remove panels from the oven and allow to cool down. 13. Measure the film build using the template and Elcometer on all baked

panels. 14. Perform scribe test on the upper half of the 10.2 x 30.5 cm panel and

send it to the lab for the gravel test. 15. Wave-scan the four 25.4 x 25.4 cm panels using wave-scan DOI.

Three measurements per panel were performed by thrice rolling the device across the surface of the panel in the direction that the paint was applied.

16. Change paint type or process temperature and repeat the procedure.

28

FIGURE 3.4 - HOLDING RACK WITH SPRAYED PANELS

3.5.2. Phase 2 Stage A - Isothermal Experiments

This phase required a large number of panels, so in order to minimize cost and energy consumption it was decided to use only three powder coatings and no repetition of runs. Powder coatings used were Black Basecoat 25 µm (BBC25), Black Basecoat 20 µm (BBC20), and Red Basecoat 25 µm (RBC25). However, to ensure statistical validity three 25.4 x 25.4 cm e-coated panels per run were sprayed. Three temperatures were considered for the isothermal runs based on the upper (193°C = 380°F), nominal (171°C = 340°F) and lower (163°C = 325°F) cure window temperatures. A summary of these experiments can be found in Appendix 2. Table 3.5 shows the time/temperature combinations and number of panels per each run. Times highlighted in green (Table 3.5) indicate how long the panel was left in the batch oven without including the ramp time. The oven was at 32°C at the beginning of each experiment. The rest of the times represent the time after the ramp time indicated in each column was completed. For example, at the lower cure window temperature (left column), if the time listed is 2 minutes, that means that the panel was left in the oven for a ramp time of 7.5 ± 2.5 min to reach 163°C, and once this time was up then the pan el was left in the oven for an additional 2 minutes at 163°C to cure. The time tha t it took to reach the desired temperature and the time the panel was cured at the desired temperature were found from the Datapaq measurements. Similar to Phase 1, Datapaq trials were performed to determine the ramp time needed to reach the target temperature. Only 3 probes, one for each panel in the oven, were attached to the panels plus there was one ambient air probe.

29

TABLE 3.5 - ISOTHERMAL CONDITIONS

LC 163°C NC 171°C UC 193°C Number of

Panels Time (min) Time (min) Time (min) 2 2 2 3

4 4 4 3

6 6 6 3

8 8 8 3

With Ramp 7.5±2.5min for

RBC25 8.5±2min for

BBC25 8.5±2.5min for

BBC20

With Ramp 8.5±2.5min for

RBC25 8.5±2min for

BBC25 8.5±1.5min for

BBC20

With Ramp 7±1min for

RBC25 7±2min for

BBC25 7±1min for

BBC20

Number of

Panels

0 0 0 3

2 2 2 3

4 4 4 3

6 6 6 3

8 8 8 3

10 10 10 3

12 12 15 3

14 14 20 3

20 16 25 3

25 18 30 3

30 20 35 3

35 25 40 3

40 30 3

45 35 3

50 40 3

55 45 3

60 50 3

65 55 3

70 60 3

75 65 3

80 70 3

85 3

90 3

95 3

Legend: LC = Lower Cure , NC= Nominal Cure, UC = Upper Cure Green Highlight = panels cured without ramp.

30

In contrast to Phase 1, for experiments of this phase (Phase 2) Datapaq probes were attached at the back of the coated panel and placed in the oven during the actual cure (Figure 3.5). In this way it was possible to collect time/temperature data and from the actual temperature profile measurements to calculate the Datapaq value (DPV) for each cure. Oven ambient air probe

FIGURE 3.5 - DATAPAQ THERMOCOUPLES Panels were wave-scanned and film build measured before and after spray. Twelve panels were sprayed at one time. Three panels for each cure condition were taken from different locations on the panel rack, as shown on the map in Figure 3.6, to ensure randomization in case film builds were not uniform (panels that have the same symbol were removed and baked under the same conditions). The panel rack was wrapped in aluminum foil to avoid contamination from paint and easy clean up.

FIGURE 3.6 - MAP FOR ISOTHERMAL RUNS

Probe attached at the back of the painted panel

O X ∆ + O X

∆ + O X ∆ + Note: Panels with the same symbol baked at the same time

Example: O = 2min @ 163°C

31


1. Run Datapaq trials to determine ramp time. 2. Label panels indicating run number and condition type 3. Measure film build of e-coated panel before spray and record the value

(average and standard deviation). 4. Wave-scan the panel before spray. Each panel was measured three

times. The wave-scan was rolled across the surface of the panel in the orientation of the paint application.

5. Check powder booth environmental conditions. 6. Fluidize powder paint. 7. Mount panels on the panel rack. 8. Spray powder coating in two passes. 9. Remove three 25.4 x 25.4 cm sprayed panels (using the map in Figure

3.6) from the panel rack and place them on the holding rack (Figure 3.4). 10. Set up Datapaq logger. 11. Add an unpainted 10.2 x 30.5 cm panel on the rack and attach the oven

ambient air thermocouple and metal temperature thermocouple to it. Connect thermocouple lines with the Datapaq logger.

12. Place holding rack with the panels and thermocouples in the batch oven and cure for the target time and temperature. The oven must start at 32°C (90°F). A stop watch was used to measure the t ime.

13. Remove panels from the oven and allow to cool down. 14. Remove thermocouples and detach the Datapaq memory. 15. Download the memory and set it up for the next run. 16. Measure the film build using the Elcometer on all baked panels. 17. Wave-scan the three 25.4 x 25.4 cm panels using wave-scan DOI. Three

measurements per panel were performed rolling the device across the surface of the panel in the direction that the paint was applied.

18. Change paint type and repeat the procedure.

3.5.3. Phase 2 Stage B - DSC Points Experiments

The basis for the experiments of this phase were the temperatures of specific points on the exotherm generated by DSC. DSC exotherms for BBC20 and BBC25 are shown in Appendix 4. Two paints of the same formulation (BBC20 and BBC25) but different particle size were used. The reason for that was to avoid the influence of formulation in reaction kinetics. Datapaq trial runs were conducted to determine how long it would take to get the metal temperature to the specific process temperature. As was the case for isothermal runs, Datapaq probes were used during real time cure of sprayed panels. Table 3.6 summarizes the time and temperature for each run and each paint type, respectively. Two panels were prepared for each temperature. For some cure conditions there were replicates conducted, as indicated in Table 3.6, so in those cases there were 4 panels used. Panels were wave-scanned and film build measured before and after spray. Twelve panels were sprayed at one time. Two panels for each cure condition were taken from different locations on the panel rack to ensure

32

randomization as shown on the map in Figure 3.7. So panels that had the same symbol were removed and baked under the same conditions. Panels were mounted on the panel rack in the vertical orientation as shown (Figure 3.7). The panel rack was wrapped in aluminum foil to avoid contamination from paint and easy clean up. Initially six temperatures were selected from the DSC exotherm for BBC25: before melting 40°C (104°F), melting peak 50°C (122 °F), between melting peak and onset of cure 100°C (212°F), onset of cure 145° C (293°F), maximum peak 187°C (369°F), and end of cure 287°C (550°F). Datap aq sensors were placed on the panel before entering the oven and were left on the panel even after panels were removed from the oven. This way it was possible to track temperature changes as they actually happened. It was found that at 40°C (104°F) and 50°C (122°F) the powder was uncured. At a temperature of 100°C ( 212°F), powder was cured t o the extent that was possible to wavescan the panel. The oven could not reach 550°F, the last temperature selected on the DSC chart. As a result, the course of this experiment was changed slightly. One panel was allowed to reach 50°C and then left in the oven at that temperature for 15 and 35 min to see if time would affect powder behavior. Since powder was uncured at 50°C a nd cured at 100°C it was decided to run experiments at a temperature in between (75°C=167°F). One panel was put in the oven and allowed to reach 75°C (167°F) and then removed from the oven. The panel was cured and the paint was not tacky.

FIGURE 3.7 – MAP FOR DSC RUNS AND SPRAY PATTERN

(Figure drawn by Kris June, co-op student that assi sted with the project) Based on the results with the BBC25, the DSC experiments for BBC20 were designed as follows. No runs were conducted at 40°C (104°F), 50°C (122°F), and 287°C (550°F). The last temperature was not run because the oven could not reach that specific temperature, while for the first two temperatures it was

33

concluded from the 25 µm experiments that the powder was not able to melt. As a result, experiments were conducted for after melting point 75°C (167°F), onset of cure 125°C (257°F), and maximum peak 215°C (419° F). The procedure was similar to BBC25 experiments. Film build was measured for all panels before and after spray. Also all panels were wave-scanned before and after spray in order to get appearance data.

TABLE 3.6 – MATRIX FOR DSC POINTS RUNS

Black 25 µm (BBC25) Black 20 µm (BBC20)

Target Temperature

°C (F)

Time at target temperature

[min]

Target Temperature

°C (F)

Time at target temperature

[min] 40 (104) 1 75 (167) 1

50 (122) 1 75 (167) 15

50 (122) 15 75 (167) 35

50 (122) 35 *125 (257) 1

75 (167) 1 *215 (419) 1

*100 (212) 1 * Two replicates *145 (293) 1

*187 (369) 1


1. Run Datapaq trials to determine ramp time. 2. Label panels indicating run number and condition type 3. Measure film build of e-coated panel before spray and record the value

(average and standard deviation). 4. Wave-scan the panel before spray. Each panel was measured three

times. 5. Check powder booth environmental conditions. 6. Fluidize powder paint. 7. Mount panels on the panel rack. 8. Spray powder coating in two passes. 9. Remove two 25.4 x 25.4 cm sprayed panels (using the map in Figure 3.7)

from the panel rack and place them on the holding rack (Figure 3.4). 10. Set up Datapaq logger. 11. Add an unpainted 10.2 x 30.5 cm panel on the holding rack and attach

the oven ambient air thermocouple and metal temperature thermocouple to it. Connect thermocouple lines with the Datapaq logger.

12. Place holding rack with the panels and thermocouples in the batch oven and cure for the target time and temperature. The oven must start at 32°C (90°F). A stop watch was used to measure the t ime.

13. Remove panels from the oven and allow cooling down. Datapaq logger was not detached during cooling.

34

14. Remove probes and detach the Datapaq memory. 15. Download the memory and set it up for the next run. 16. Measure the film build using Elcometer on all baked panels. 17. Wave-scan the three 25.4 x 25.4 cm panels using wave-scan DOI. Three

measurements per panel were performed rolling the device across the surface of the panel in the direction that the paint was applied.

18. Change paint type and repeat the procedure for all powder paints.

3.5.4. Phase 3 - Variable Heating Rate Experiments

The aim of this set of experiments was to investigate the effect of the heating rate (ramp) on appearance quality. Desired heating rates were achieved using a programmable oven. Similar to previous experiments, Datapaq trials were used to determine the specific time it took to reach the desired process temperature at the specific ramp. Two powder coatings were used – RBC25, BBC25 – to see how different colors and paint formulations react to cure rates. Spray settings were maintained the same as those used in previous stages of this project to ensure comparison with previous data (Table 3.3). Panels were sprayed with a single pass on the panel rack (where the panels were mounted to be painted). The single pass consisted of a single vertical sweep, with the applicator being triggered slightly before the panels to ensure steady state painting conditions. Film build was kept constant at 51 ±5 µm (2 ±0.2 mil). A total of thirty-six panels were sprayed for each powder coating used in these experiments. Table 3.7 illustrates the various final temperatures at which each group of panels (for each ramp rate) was left to cure. Complete cure and oven information can be found in Appendix 4. The method used to determine the theoretical time that panels stayed in the oven and to clarify the terms in Appendix 4 is described in the next paragraph using an example. The theoretical ramp time was calculated using the initial and final temperatures and the ramp rate. For example: [163°C (final temp.) - 32°C (initial temp)] / [5°C /min (ramp rate)] = 26.2 min

(ramp time theoretical.) The ramp time is the time to get the oven from ambient temperature to within ± 5°C of the desired final temperature. Even though t he oven ramp is intended to bring the oven to the desired temperature that does not mean that the panel reached the desired temperature. As a result there is need for the set time, which is the time required to allow the panel to achieve the desired final temperature (within ±2°C) once the oven ramp has completed. Aft er the panel reached the target temperature then it was allowed to bake for the specific time (Table 3.7) which was selected based on the cure window points. The soak time: is the combination of the set time and the time that the panel is cured at its final temperature. So, the total time the panels were in the oven was the "ramp time" +

35

"soak time". For example, for the first cure condition (ramp rate 5°C/min, T=163°C @ 14min) the total time in the oven was:

25.4 min + 22 min = 47.4 min where, soak time = 14 min (time at final temp.) + 8 min (set time) = 22 min

TABLE 3.7 - SUMMARY OF OVEN SETTINGS (FULL TABLE IN APPENDIX 5)

Initial Temperature (º C)

Final Temperature (º C)

Ramp Rate

(º C/min)

Time at desired

temperature (min)

32 163 5 14 32 163 5 25 32 163 5 90 32 171 5 11 32 171 5 20 32 171 5 56 32 193 5 9 32 193 5 15 32 193 5 35 32 163 10 14 32 163 10 25 32 163 10 90 32 171 10 11 32 171 10 20 32 171 10 56 32 193 10 9 32 193 10 15 32 193 10 35

Note: These settings were repeated for two powder coatings. For each run the following procedure was followed:

1. Run Datapaq trials to determine ramp time. 2. Label panels indicating run number and condition type 3. Measure film built of e-coated panel before spray and record the value

(average and standard deviation). 4. Check powder booth environmental conditions. 5. Fluidize powder paint. 6. Mount panels vertically on the panel rack. 7. Spray powder coating in one pass. 8. Remove two 25.4 x 25.4 cm sprayed panels from the panel rack and place

them on the holding rack.

36

9. A pre-programmed cure profile was selected, based on the cure settings required by each group of panels. The profile’s first four minutes are dedicated to achieving a steady-state temperature of 90º F.

10. Set up Datapaq logger. 11. Add an unpainted 10.2 x 30.5 cm panel on the holding rack and attach the

oven ambient air thermocouple and metal temperature thermocouple to it. 12. With 30 seconds left in the oven’s 32°C (90 °F) steady-state, the holding

rack was placed in the oven. Immediately, the Datapaq was connected to the thermal probe reader, recording the thermal probe temperatures in five second intervals.

13. Cure for the target ramp rate, time and temperature. A stop watch was used to measure the time.

14. Remove panels from the oven and allow to cool down. 15. Remove thermocouples and detach the Datapaq memory. 16. Download the memory and set it up for the next run. 17. Measure the film built using an Elcometer on all baked panels. 18. Wave-scan the two 25.4 x 25.4 cm panels. Three measurements per

panel were performed. In each measurement the wave-scan DOI was rolled across the surface of the panel in the direction that the paint was applied.

19. This process was repeated as necessary for each pair of panels, varying the final temperature and/or ramp until the desired data had been acquired.

37

CHAPTER 4 - PHASE 1: CURE WINDOW CORNER POINTS

4.1. Particle Size Analysis

Particle size was analyzed before and after spray as mentioned in sections 3.4.1 for all 6 paints (Appendix 6). It was found that for all basecoats, regardless of particle size, the mean particle size did not change significantly after electrostatic spraying. However, for the Red Colorkey Primer (RCP) the mean particle size of the virgin material was 26 µm and for the material collected from panels after spray it was 30 µm. For this reason panels sprayed with RCP were compared with BBC 30 µm and RBC 30 µm. The increase in particle size after spray was not observed in RBC and BBC coatings regardless of the particle size of virgin material. All paints were found to have a narrow particle size distribution (PSD).

4.2. Experimental Results and Discussion It has been reported that the particle size and formulation have a significant impact on the appearance of powder coatings. In order to eliminate the effect of particle size and investigate the effect of formulation, comparisons were made between paints with the same particle size. First, three paints (BBC, RBC, RCP) with a 30 µm particle size were compared. Second, two paints (RBC, BBC) with a 25 µm particle size were compared. Then, the effect of particle size was considered for the same formulation. So, three particle sizes of black basecoat (BBC20, 25, 30) were compared with each other and then two particle sizes of red basecoat (RBC25) were compared with each other.

E-coated panels used for the experiments were chosen based on their smooth surface, in order to minimize the effect of the substrate roughness on appearance. Appearance parameters were measured before and after spray. It was observed that for all panels used, du, Wa, Wb, Wc and Wd values before spray were very consistent (Table 4.1). This means that the initial surface topography was constant from trial to trial and should not have influenced the final film appearance. However, We values (Figure 4.1) seem to have higher scattering, with COVs around 20. It is interesting to see from Figure 4.2 that the contrast values for We before and after spray are essentially the same. The same was true for We for all other coatings. This could mean that cure conditions do not have much effect on contrast values for We. This is in agreement with findings from Giroux (2006), who reported that basecoat was responsible for half of the contrast values of short-waves and the other half was from clearcoat, while for long-waves only clearcoat was responsible for the change in contrast values. It was concluded at this point that the substrate roughness did not have a significant effect on the appearance of the final film and so it was not investigated any further. This is in agreement with findings from Biris et al. (2001), that once the film thickness is above 40 µm, the effect of the substrate smoothness is not important. In this study, film thicknesses were much higher than 40 µm, which

38

supports the assumption that substrate topography does not affect the final finish appearance.

FIGURE 4.1 – We BEFORE SPRAY FOR RCP30

FIGURE 4.2 – We BEFORE AND AFTER SPRAY FOR RCP30

0

5

10

15

20

25

30

35

0 50 100 150 200

We

Data points

RCP30

0

5

10

15

20

25

30

35

0 50 100 150 200

We

Data points

RCP30

We_before spray We_after spray

39

TABLE 4.1 - SUMMARY OF APPEARANCE ELEMENTS BEFORE S PRAY (N = 60)

Du Wa Wb Wc Wd We RCP30

average 56.5 31.0 20.5 14.9 17.8 17.4 SD 0.4 1.3 1.3 1.1 1.5 3.3 COV 0.008 0.043 0.062 0.071 0.087 0.191

BBC20 average 57.3 28.7 18.4 12.3 15.4 16.3 SD 0.5 3.9 2.7 1.8 2.2 3.8 COV 0.008 0.137 0.149 0.148 0.144 0.232



RBC25 average 57.4 28.7 18.0 12.1 14.3 14.2 SD 0.7 3.1 3.0 2.5 3.1 3.8 COV 0.013 0.108 0.165 0.203 0.214 0.271

RBC30 average 57.4 29.1 18.4 12.3 15.8 16.4 SD 0.5 2.0 2.0 1.4 1.6 3.3 COV 0.008 0.068 0.107 0.110 0.101 0.201

Note: averages presented on this table are for all panels used in 15 runs (4 panels/run). Appendix 7 has all the FB data.

4.2.1. Constant Particle Size RBC30, BBC30, RCP30 Film build readings before spray were very consistent and will not be considered any further (Appendix 7). Since there were four panels per run, the average FB of powder coating for each run was calculated from measurements on the 4 panels and data are summarized in Appendix 8 together with the coefficient of variation (COV). The average values for the three runs with the same cure conditions were averaged again and data are summarized in Table 4.2. To ensure that film build did not have an effect on the appearance parameters studied, the film build was maintained within 51-71 µm (2 to 2.8 mils) range (Appendix 7 & 8). Figure 4.3 shows that film build was maintained within the range for all 30 µm powder paints used except for one run of the BBC30, for which it was 78 µm. To find out if this value affected appearance and if a relationship existed between film build and appearance parameters, surface structure elements (Wa – We) were plotted against the cure conditions, with the film build as a secondary y-axis, for all three paints. Figure 4.4 shows the graph of Wa for the BBC30 paint. It can be seen

40

from Figure 4.4 that there is only a slight difference in Wa contrast values (16.7 and 17.8) even though the film build changed from 71 (first point) to 52.3 µm (third point). This finding is more obvious from Figure 4.5, which shows that the film build is very consistent between runs, but the Wa contrast value ranges from 12.2 to 43.1. The same trend was observed for all other spectrum elements and for all three paint types. Comparing Wa trends in Figure 4.4 and 4.5 it can be seen that values vary with treatment and not the film build.

TABLE 4.2 - SUMMARY OF FILM BUILD OVERALL AVERAGES

Run

Cure Time [min]

Process Temperature

[°C] FB

[µm] COV RCP30

10, 11, 12 14 163 61.0 0.033 13, 14, 15 90 163 58.6 0.017

1,2,3 20 171 59.6 0.026 7, 8, 9 9 193 60.5 0.040 4,5,6, 35 193 58.2 0.018

RBC30 10, 11, 12 14 163 55.8 0.023

1,2,3 90 163 57.0 0.029 13, 14, 15 20 171 55.2 0.023

7, 8, 9 9 193 54.4 0.020 4,5,6, 35 193 54.8 0.023

BBC30 4,5,6, 14 163 71.2 0.051 1,2,3 90 163 78.1 0.044

10, 11, 12 20 171 52.4 0.076 7, 8, 9 9 193 54.2 0.072

13, 14, 15 35 193 51.0 0.074

41

FIGURE 4.3 - FILM BUILDS FOR 30 MICROMETER PAINTS

FIGURE 4.4 - WA VS. FB AND CURE CONDITIONS FOR BBC3 0

(Bars of the same color are cured at the same tempe rature, Red = 163°C, Green = 171°C and Blue = 190°C)

42

FIGURE 4.5 - WA VS. FB AND CURE CONDITIONS FOR RBC3 0

(Bars of the same color are cured at the same tempe rature, Red = 163°C, Green = 171°C and Blue = 190°C)

As explained before, for each cure window point there were three runs (replicates) with 4 panels for each run. Each panel was scanned using wave-scan DOI at three locations (three readings) for a total of 12 readings per run. There were 5 cure conditions and three replicates for each condition totaling into 15 runs for each paint type. The averages of 12 readings for each individual run are included in Appendix 9. Similar to film build, an overall average was calculated for each appearance element. Tables 4.3 to 4.7 show overall averages and coefficients of variation (COVs) for all panels baked under the same conditions (4 panels x 3 replicates x 3 readings/panel = 36 readings). It can be seen that We has the greatest variation. This could be related to the fact that for the substrate before spray We had the greatest variation as well. Application of basecoat did not change We contrast values and as a result the same variation persisted even after spray. A table with the raw data for RCP30 at 171°C and 20 minutes, illustrating the calculation of the averages and COVs, is included in Appendix 10.

43

TABLE 4.3 – WAVE-SCAN VALUES FOR LT1 CONDITION (N = 36)

T = 163°C, t = 14 min

du Wa Wb Wc Wd We

RCP30

Average 8.5 15.1 31.2 37.4 35.4 15.2 COV 0.15 0.10 0.05 0.08 0.05 0.21

BBC30 Average 7.0 16.7 30.0 34.0 35.6 14.2

COV 0.25 0.11 0.08 0.13 0.06 0.20 RBC30

Average 5.5 12.2 23.6 29.0 31.0 12.1 COV 0.29 0.09 0.07 0.09 0.07 0.29

TABLE 4.4 – WAVE-SCAN VALUES FOR LT2 CONDITION (N = 36)

T = 163°C, t = 90 min

du Wa Wb Wc Wd We

RCP30 Average 11.4 22.9 39.8 39.4 36.0 16.0

COV 0.13 0.07 0.04 0.06 0.05 0.22 BBC30

Average 9.0 22.8 38.8 35.9 36.5 15.5 COV 0.18 0.09 0.05 0.14 0.07 0.21

RBC30 Average 7.8 19.6 32.4 28.0 29.9 12.8

COV 0.23 0.14 0.09 0.15 0.07 0.23

TABLE 4.5 – WAVE-SCAN VALUES FOR CNOMINAL CONDITION (N = 36)

T = 171°C, t = 20 min

du Wa Wb Wc Wd We

RCP30

Average 10.6 19.7 36.6 37.3 35.5 16.2 COV 0.12 0.09 0.05 0.07 0.06 0.29

BBC30 Average 8.0 17.8 30.2 37.8 36.8 15.8

COV 0.25 0.11 0.08 0.13 0.06 0.20 RBC30

Average 6.8 16.6 28.3 28.3 30.8 11.7 COV 0.21 0.09 0.06 0.10 0.06 0.26

44

TABLE 4.6 – WAVE-SCAN VALUES FOR HT1 CONDITION (N = 36)

T = 193°C, t = 9 min

du Wa Wb Wc Wd We

RCP30 Average 13.4 28.0 44.6 37.2 34.4 15.7

COV 0.11 0.07 0.04 0.06 0.05 0.17 BBC30

Average 14.2 32.6 46.5 35.4 36.1 18.5 COV 0.14 0.08 0.05 0.13 0.07 0.15

RBC30 Average 13.8 32.7 45.9 29.9 31.1 15.3

COV 0.11 0.07 0.04 0.07 0.05 0.15

TABLE 4.7 – WAVE-SCAN VALUES FOR HT2 CONDITION (N = 36)

T = 193°C, t = 35 min

du Wa Wb Wc Wd We

RCP30 Average 22.8 43.3 54.1 40.8 36.8 20.8

COV 0.10 0.04 0.02 0.08 0.07 0.16 BBC30

Average 23.3 45.6 54.0 40.3 38.1 22.1 COV 0.12 0.06 0.03 0.09 0.06 0.16

RBC30 Average 21.4 43.1 53.1 32.2 31.8 19.7

COV 0.08 0.04 0.03 0.06 0.05 0.18

A summary of du to We contrast values is shown in Table 4.8. The dash in Table 4.8 indicates no change in appearance value with either time or temperature. The up/down arrows in the third and fifth columns indicate the increase/decrease of the appearance value with time as the temperature was maintained constant, while up/down arrow in the last column (Temperature Effect) shows the increase/decrease of the appearance parameter with the increase in temperature. The appearance value comparisons were made with Minitab Statistical Software, using the following method. It should be noted that for comparisons, the whole range of raw data points (36) was used and not just the one average point. An F-test was performed to check the variance of data at a 95% confidence limit. For example, to check if the du values for RCP30 at LT1 and LT2 had equal or unequal variances, all du measurements for Run 10, 11, 12 (cured at LT1) were compared with all du measurements for Run 13, 14, 15 (cured at LT2) (see Appendix 10 for raw data). Depending on the results of the F-test, Student’s t-test for equal or unequal variances was performed at a 95% confidence limit to check if the means of each appearance value changed

45

significantly. Results of the t-test are represented by the arrows or lines in Table 4.8

It can be seen from Table 4.8 that the temperature and time effects are very clear for du, Wa, and Wb contrast values for RCP30. The contrast values increased with increased time as the temperature was kept constant and also increased with the increase in temperature. This behavior can be seen in Figure 4.6. The graph includes also the values at the “nominal” cure condition (20 minutes @ 171°C), which also confirms the finding that the du, Wa and Wb increased with increased baking temperature. However, for the nominal temperature there was only one baking time considered (20 min) so the change in contrast values with time for that temperature was not calculated, but the change in contrast with temperature for 20 minutes of curing can be determined from a vertical line through Cnominal. These findings are in agreement with the ones reported by Barletta et al. (2007) that better surface appearance was achieved at low temperatures. Also, Hannon et al. (1976) reported that an increase in temperature decreased gloss (worsened appearance). Similar trends were observed for RBC30 and BBC30 (Figure 4.7 & 4.8). A graphical illustration of all appearance elements for all three paint types can be found in Appendix 11. The results for the time effect at the high temperature for Wc, We and Wd are similar to du, Wa, and Wb, in that there is an increase in contrast value with increase in time at a constant temperature. The only exception is Wd for RBC30. For this paint there is no significant change in Wd contrast value with an increase in time if the high temperature (193°C) maintained constant. Also, the temperature effect is very clear (last column in Table 4.8). All contrast values for Wc, Wd, and We increase with increased temperature with the exception of Wd from RCP30 which shows no significant increase (36.4 to 36.8). At the lower temperature, We remains unchanged statistically as the time increases. The results for Wc and Wd are not as consistent. Wc seems to increase for RCP30 and remain unchanged for RBC30 and BBC30; while Wd seems not to be affected by time increase for RCP30 and BBC30 but decreases for RBC30. The values for Wc, Wd and We at LT1 and LT2 are very close, even in the cases where statistically it was shown that there was a slight increase or decrease. This could mean that the long-wave structures are not greatly influenced by time changes at the low temperatures. A better understanding of the time effect will be gained when isothermal experiments are analyzed (Chapter 5).

46

TABLE 4.8 – SUMMARY OF APPEARANCE PARAMETER CHANGES WITH TIME AND TEMPERATURE (30µm)

Cure Conditions

Temperature Effect

LT1 LT2 HT1 HT2 Red Colorkey Primer – RCP30

du 8.5 11.4 ↑ 13.4 22.8 ↑ ↑ Wa 15.1 22.9 ↑ 28.0 43.3↑ ↑ Wb 31.2 39.8 ↑ 44.6 54.1 ↑ ↑ Wc 37.4 39.4 ↑ 37.2 40.8 ↑ ↑ Wd 35.4 36.0 — 34.4 36.8 ↑ — We 15.2 16.0 — 15.7 20.8 ↑ ↑

Red Basecoat - RBC30 du 5.5 7.8 ↑ 13.8 21.4 ↑ ↑ Wa 12.2 19.6 ↑ 32.7 43.1↑ ↑ Wb 23.6 32.4 ↑ 45.9 53.1↑ ↑ Wc 29.0 28.0 — 29.9 32.2↑ ↑ Wd 31.0 29.9 ↓ 31.1 31.8 — ↑ We 12.1 12.8 — 15.3 19.7↑ ↑

Black Basecoat - BBC30 du 7.0 9.0 ↑ 14.2 23.3 ↑ ↑ Wa 16.7 22.8 ↑ 32.6 45.6 ↑ ↑ Wb 30.0 38.8 ↑ 46.5 54.0 ↑ ↑ Wc 34.0 35.9 — 35.4 40.3 ↑ ↑ Wd 35.6 36.5 — 36.1 38.1 ↑ ↑ We 14.2 15.5 — 18.5 22.1 ↑ ↑

Legend

Cure

Condition LT1 LT2 CNominal HT1 HT2 Temperature

(⁰C) 163 (325 F) 171 (340 F) 193 (380 F)

Time (min) 14 90 20 9 35

47

FIGURE 4.6 - APPEARANCE VS. CURE CONDITION FOR RCP3 0

20

15

10

40

30

20

50

40

3041

39

37 37

36

35

9080706050403020100

20.0

17.5

15.0

du

Wa 0,1-0,3mm

Wb 0,3-1,0mm

Wc 1-3mm

Wd 3-10mm

Cure Time [min]We 10-30mm

163

171

193

(°C)

Temperature

Cure

22.8

13.410.5 11.4

8.543.2

28.0

19.8 22.9

15.154.1

44.6

36.739.8

31.240.8

37.2 37.3

39.4

37.436.9

34.4

35.536.0

35.4

20.8

15.7 16.1 16.015.2

48

FIGURE 4.7 - APPEARANCE VS. CURE CONDITION FOR RBC3 0

18

12

640

30

20

50

40

30

32

30

28 32

31

30

9080706050403020100

20

16

12

du

Wa 0,1-0,3mm

Wb 0,3-1,0mm

Wc 1-3mm

Wd 3-10mm


163

171

193

(°C)

Temperature

Cure

21.4

13.8

6.8 7.85.5

43.1

32.7

16.6 19.612.2

53.145.9

28.332.4

23.632.2

29.9

28.3 28.029.0

31.8

31.130.8

29.9

31.0

19.7

15.3

11.712.812.1

49

FIGURE 4.8 - APPEARANCE VS. CURE CONDITION FOR BBC3 0

20

15

10

40

30

20

50

40

3040.0

37.5

35.0

38

37

36

9080706050403020100

21

18

15

du

Wa 0,1-0,3mm

Wb 0,3-1,0mm

Wc 1-3mm

Wd 3-10mm


163

171

193

(°C)

Temperature

Cure

23.3

14.2

8.0 9.07.0

45.6

32.6

17.822.8

16.754.0

46.5

30.2

38.8

30.040.3

35.4

37.8

35.9

34.038.1

36.1

36.8 36.5

35.622.1

18.5

15.8 15.514.2

50

Comparing Wc contrast values for the three paint types as illustrated in Figure 4.9, it was found that RBC had the lowest contrast values for all cure conditions (LT1, LT2, Cnominal, HT1 and HT2). The same was true for Wd and We (Appendix 12). It was interesting to see that RBC30 still had the lowest contrast for du, Wa and Wb but only at LT1, LT2 and Cnominal. At the high temperature all three paints had comparable values except for Wa for RCP30 at 193°C (9min) which had the lowest contrast value. For all appearance elements and under all conditions RBC30 had lower contrast values than BBC, while RCP30 values fluctuate above and below BBC30. Graphs similar to Figure 4.9 for all appearance elements can be found in Appendix 12. The ratio between short-waves (du, Wa, Wb) and long-waves (Wc, Wd) is important to the way appearance is perceived by the human eye (BYK, 2008). In a so called “ideal” spectrum there is a higher peak at Wb, a minimum at Wc and another peak at Wd which is lower than the Wb peak. If short-wave values are 1.5 times higher than long-wave values, then appearance will be smoother due to the masking effect (Giroux, 2006). Structure spectra developed for LT1 for all three paint types are graphically illustrated in Figure 4.10. The rest of the graphs can be found in Appendix 13. It can be seen from Figure 4.10 that RBC30 yields the lowest contrast values but the shape of the structure spectrum is the same for all three paints. Also, it can be seen that the contrast values for RCP30 and BBC30 are almost identical for these conditions with a slight variation in Wc which is lower for BBC30. However, the shape of the spectrum at LT1 is very different from what is accepted as a “good” spectrum. There is obviously little coverage of long-waves from short-waves since the Wb values are less than Wd. From Figure 4.11 it can be seen that the shape of the structure spectrum at LT2 is similar for all three paints and it is close to that for LT1. At the nominal cure conditions, RBC30 still has the best appearance. However, du, Wa and Wb values are similar to BBC30; while du and Wa are very close to RCP30 (Figure 4.12. From Wc to We BBC30 and RBC30 have identical values while RBC30 is much lower. Also the shape of the spectrum seems to have a slight minimum at Wc for RBC30 but it is almost trapezoidal in shape for RCP30 and BBC30 (Figure 4.12). For the upper process temperature (193°C) at both s hort and long temperature intervals (9 min and 35 min) all three paints seem to perform the same for du, Wa, and Wb, Wd and We with a slightly lower value of Wa for RCP30 at HT1 (9min @ 193°C). However, Wc and Wd, as discussed in the previous paragraph, are much lower for RBC30 at both conditions (HT1 and HT2). Development of the “good” spectrum is seen at HT1 and HT2 with the Wb value being greater than Wd (Figure 4.13 & 4.14). However, contrast values are above 30 (Wb = 45.9 at HT1 and for Wa = 43.1 and Wb = 53.1 at HT2). A contrast value of 30 or below is considered in industry as indicator of smooth appearance (Giroux, 2006).

51

FIGURE 4.9 - WC VS. CURE CONDITIONS FOR ALL THREE P AINTS

FIGURE 4.10 - STRUCTURE SPECTRUM AT LT1

75

50

30

35

160 25

40

170180 0

190

Wc 1-3mm

Cure T ime [min]

Cure Temperature (°C)

BBC 30 µm

RBC 30 µm

RCP 30µ

Paint Type

52

FIGURE 4.11 - STRUCTURE SPECTRUM AT LT2

FIGURE 4.12 - STRUCTURE SPECTRUM AT CNOMINAL

FIGURE 4.13 - STRUCTURE SPECTRUM AT HT1

53

FIGURE 4.14 - STRUCTURE SPECTRUM AT HT2

The low contrast values for RBC30 indicate a better smoothness of the final film. This could be attributed to the pigment type and formulation. However, even though RBC30 has the lowest contrast values, the shape of the spectrum is not ideal. The finding that RCP30 has comparable appearance qualities with BBC30 and RBC30 supports the hypothesis that colorkey primers can achieve a similar appearance to basecoats and as a result can replace them in certain areas of the vehicle.

RBC25 and BBC25 Red basecoat and black basecoat powder paints of 25µm particle size will be discussed in this section. Similar to the previous section for the 30 µm paints, the film build was maintained within a tight range (50.4 to 60.7 µm) to limit the impact on appearance. The process of calculating averages and overall averages are the same as for 30 µm paints. Summaries of film build averages for each run, together with the COVs are included in Appendix 14. Table 4.9 summarizes overall averages for film build.

TABLE 4.9 - SUMMARY OF FILM BUILD OVERALL AVERAGES

RBC25 BBC25

Run

Cure Time [min]

Process Temperature

[°C] FB

[µm] COV FB

[µm] COV

7,8,9 9 193 58.4 0.04 50.4 0.037

10,11,12 14 163 59.4 0.031 53.2 0.023

13,14,15 20 171 59.5 0.028 52.9 0.032

4,5,6, 35 193 58.9 0.036 50.5 0.015

1,2,3 90 163 60.7 0.030 52.1 0.029

0.0

10.0

20.0

30.0

40.0

50.0

60.0

du Wa Wb Wc Wd We

Co

ntr

ast

Structure Spectrum

Upper Cure Temperature (193°C @ 35 min)

RCP30

BBC30

RBC30

54

Appearance measurements and average calculations were similar to the ones discussed previously for 30 micron paints. Tables 4.10 to 4.14 show overall averages and COVs for all panels baked under the same conditions (4 panels x 3 replicates x 3 readings/panel = 36 readings). It can be seen that COVs for all structure spectrum elements are below 0.27 (27%), with We still having the consistently highest COV; the only exception is the higher COV of Wa for RBC25 at LT2 and Cnominal. Tables with the raw data for all other paints and cure conditions are included in Appendix 10. A summary of du to We contrast values are shown in Table 4.15. The dash in Table 4.15 indicates no change in appearance value with either time or temperature. The up/down arrows indicate the increase/decrease of the appearance value with time as the temperature was maintained constant, while the up/down arrow in the last column (Temperature Effect) shows the increase/decrease of the appearance parameter with the increase in temperature. These appearance values were compared to create Table 4.15 and plotted using Minitab Statistical Software. It should be noted that for comparisons, the whole range of 36 raw data points was used and not just the one average point. An F-test was performed to check the variance of data at the 95% confidence limit. For example, to check if the du values for RCP at LT1 and LT2 had equal or unequal variances, all du measurements for Runs 10, 11, and 12 (cured at LT1) were compared with all du measurements for Run 13, 14, 15 (cured at LT2). Depending on the results of the F-test, the Student t-test for equal or unequal variances was performed at 95% confidence limit to check the difference in means for each structure size. Results of the t-test are represented by the arrows or lines in Table 4.15. It can be seen from Table 4.15 that the temperature and time effects are very consistent for du, Wa, Wb and We contrast values for RBC25 and BBC25. Their contrast values increased with increased time as the temperature was kept constant and also increased with the increase in temperature. This behavior can be seen in Figure 4.15 and Figure 4.16. The graph includes also the values at the “nominal” cure condition (CNominal). It can be seen that for RBC25 all Cnominal

appearance values were slightly lower than the values for LT1 except for Wa which did not show much of a change (12.2 and 12.4). For BBC at nominal conditions (Figure 4.16) similar trends are observed, but in this case the Cnominal

du value is the same as LT1. As mentioned in the previous section, for the nominal temperature there was only one baking time considered (20 min) so the change in contrast values with time for that temperature was not considered. A graphical illustration of all appearance elements for two paint types can be found in Appendix 15. There is no significant change in Wc and Wd for RBC25 as the time increases at the low temperature (163°C). Under the same conditi ons for BBC there is still no significant change for Wd, but there is a slight decrease for Wc. At the high temperature end (193°C) both Wc and Wd increase wit h increase in time for RBC25, while for BBC25 the same effect as at the low temperature is observed – a slight decrease for Wc and no significant change for Wd. This finding again could mean that the long-wave structures are not greatly influenced by time

55

changes at the low temperatures. A better understanding of the time effect will be gained when isothermal experiments are analyzed (Chapter 5). Similar to 30 micron paints, when comparing Wc contrast values for RBC25 and BBC25 as illustrated in Figure 4.17, it can be seen that RBC has the lowest contrast values for all cure conditions (LT1, LT2, Cnominal, HT1 and HT2). In contrast to 30 micron paints however, this trend was followed throughout the structure spectrum. Graphs similar to Figure 27 for all appearance elements can be found in Appendix 16.

TABLE 4.10 - STRUCTURE SPECTRUM FOR LT1 CONDITION

T = 163°C, t = 14 min

du Wa Wb Wc Wd We

BBC25 Average 7.6 18.6 32.9 33.5 31.2 12.6 STDev 1.35 1.91 2.82 3.03 1.81 2.75

COV 0.18 0.10 0.09 0.09 0.06 0.22

RBC25 Average 4.3 12.2 24.5 21.8 25.8 9.4 STDev 0.99 1.34 3.62 2.02 1.47 2.55 COV 0.23 0.11 0.15 0.09 0.06 0.27

TABLE 4.11 - STRUCTURE SPECTRUM FOR LT2 CONDITION

T = 163°C, t = 90 min

du Wa Wb Wc Wd We

BBC25 Average 8.9 22.4 34.9 31.4 31.4 13.6

STDev 1.20 1.89 2.26 2.75 1.68 3.13 COV 0.14 0.08 0.06 0.09 0.05 0.23

RBC25 Average 5.6 16.0 28.6 21.5 25.4 10.9 STDev 1.44 2.82 4.10 2.17 1.56 2.67 COV 0.26 0.18 0.14 0.10 0.06 0.24

56

TABLE 4.12 - STRUCTURE SPECTRUM FOR C NOMINAL

T = 171°C, t = 20 min

du Wa Wb Wc Wd We

BBC25 Average 7.6 17.1 29.3 31.8 30.2 12.4 STDev 1.21 1.46 1.40 2.85 1.64 3.36

COV 0.16 0.09 0.05 0.09 0.05 0.27 RBC25

Average 3.9 12.4 22.5 19.5 24.3 9.0 STDev 1.03 1.29 1.70 2.31 1.52 1.92 COV 0.27 0.10 0.08 0.12 0.06 0.21

TABLE 4.13 - STRUCTURE SPECTRUM FOR HT1 CONDITION

T = 193°C, t = 9 min

du Wa Wb Wc Wd We

BBC25


RBC25 Average 11.7 30.1 46.2 25.6 26.5 13.3 STDev 1.31 1.93 2.32 1.55 1.56 2.48

COV 0.11 0.06 0.05 0.06 0.06 0.19

TABLE 4.14 - STRUCTURE SPECTRUM FOR HT2 CONDITION

T = 193°C, t = 35 min

du Wa Wb Wc Wd We

BBC25


RBC25 Average 18.5 40.9 53.8 28.7 28.4 17.7 STDev 2.00 2.12 1.83 1.63 1.35 2.69

COV 0.11 0.05 0.03 0.06 0.05 0.15

57

TABLE 4.15 - SUMMARY OF APPEARANCE PARAMETERS (25µM )

Cure Conditions

Temperature Effect

LT1 LT2 HT1 HT2 Red Basecoat – RBC25

du 4.3 5.6 ↑ 11.7 18.5 ↑ ↑ Wa 12.2 16.0 ↑ 30.1 40.9 ↑ ↑ Wb 24.5 28.6 ↑ 46.2 53.8 ↑ ↑ Wc 21.8 21.5 — 25.6 28.7 ↑ ↑ Wd 25.8 25.4 — 26.5 28.4 ↑ ↑ We 9.4 10.9 ↑ 13.3 17.7 ↑ ↑

Black Basecoat – BBC25 du 7.6 8.9 ↑ 17.4 23.9 ↑ ↑ Wa 18.6 22.4 ↑ 39.6 48.5 ↑ ↑ Wb 32.9 34.9 ↑ 51.7 54.6 ↑ ↑ Wc 33.5 31.4 ↓ 37.3 36.7 ↓ ↑ Wd 31.2 31.4 — 32.7 33.2 — ↑ We 12.6 13.6 ↑ 16.4 21.1 ↑ ↑

58

FIGURE 4.15 - APPEARANCE VS. CURE CONDITION FOR RBC 25

15

10

540

30

20

50

40

30

30

25

2028

26

24

9080706050403020100

16

12

8

du

Wa 0,1-0,3mm

Wb 0,3-1,0mm

Wc 1-3mm

Wd 3-10mm


163

171

193

(°C)

Temperature

Cure

5.6

18.5

3.94.3

11.7

16.0

40.9

12.412.2

30.1

28.6

53.8

22.524.5

46.2

21.5

28.7

19.521.8

25.6

25.4

28.4

24.3

25.826.5

10.9

17.7

9.09.4

13.3

Red Basecoat (RBC 25)

59

FIGURE 4.16 - APPEARANCE VS. CURE CONDITION FOR BBC 25

20

15

10

45

35

25

50

40

30

36

34

32

33.0

31.5

30.0

9080706050403020100

20

16

12

du

Wa 0,1-0,3mm

Wb 0,3-1,0mm

Wc 1-3mm

Wd 3-10mm


163

171

193

(°C)

Temperature

Cure

8.9

23.9

7.67.6

17.4

22.4

48.5

17.118.6

39.6

34.9

54.6

29.332.9

51.7

31.4

36.7

31.8

33.5

37.3

31.4

33.2

30.2

31.2

32.7

13.6

21.1

12.412.6

16.4

Black Basecoat (BBC) 25 micron

60

Structure spectrum developments for LT1 for RBC25 and BBC25 are graphically illustrated in Figure 4.18. The rest of the graphs can be found in Appendix 17. It can be seen from Figure 4.18 that RBC yields the lowest contrast values with a local minimum at Wc, but the peak at Wd is slightly higher than Wb. From the graphs in Appendix 17 it can be seen that for all conditions, RBC25 has lower contrast values. A close to “ideal” shape is seen for HT1 and HT2, but again contrast values are above 30. A value of 30 is considered in industry as indicator of smooth appearance, and values above it are an indication of worsening of appearance (Giroux, 2006). Consistent to the results for 30 micron paints RBC was the paint that had lower contrast values.

FIGURE 4.17 - WC VS. CURE CONDITIONS FOR RBC25 AND BBC25

FIGURE 4.18 - DEVELOPMENT OF STRUCTURE SPECTRUM LT1 (RBC25 AND BBC25)

7520

50

25

30

160

35

25170

180 0190

Wc 1-3mm

Cure T ime [min]

Cure T emperature (°C)

BBC 25 µm

RBC 25 µm

Paint Type

0.0

10.0

20.0

30.0

40.0

du Wa Wb Wc Wd We

Co

ntr

ast

Structure Spectrum

Lower Cure Temperature (163°C @ 14 min)

BBC25

RBC25

61

4.2.2. Constant Formulation

This section will focus on comparison of the paints with the same formulation but different particle size. The first group to be discussed will be black basecoat of 20, 25, and 30 µm, followed by red basecoat of 25 and 30 µm will be discussed. BBC30, BBC25, BBC20 Film build overall averages for BBC20 are shown in Table 4.16. Film builds for BBC30 and BBC25 were shown in Tables 4.2 and 4.9, respectively. Film build averages for Runs 1-15 are found in Appendix 14. Appearance values for BBC20 are illustrated in Figure 4.19. For BBC30 and BBC25, appearance values are illustrated in Figure 4.8 and 4.16, respectively. Table 4.17 summarizes time and temperature effects for BBC20 paint. Table 4.18 summarizes overall averages for BBC20. The effects for BBC25 and BBC30 were discussed in the previous sections. For BBC20 all appearance elements are increased by an increase in time and temperature. A summary of all appearance elements for all three particle sizes of BBC is included in Appendix 18.

TABLE 4.16 - FILM BUILD OVERALL AVERAGES FOR BBC20

Run Cure Time

[min]

Process Temperature

[°C] FB [µm]

BBC20

Average of 7,8,9 9 193 53.1

Average of 10,11,12 14 163 54.8

Average of 13,14,15 20 171 55.1

Average of 4, 5, 6, 35 193 51.4

Average of1,2,3 90 163 49.9

Table 4.17 - SUMMARY OF APPEARANCE PARAMETERS BBC20

Cure Conditions

Temperature Effect

LT1 LT2 HT1 HT2 Black Basecoat - BBC20

du 5.8 8.4 ↑ 14.9 21.9 ↑ ↑ Wa 14.9 20.3 ↑ 34.8 44.8 ↑ ↑ Wb 25.9 32.1 ↑ 47.8 54.2 ↑ ↑ Wc 24.5 28.8 ↑ 29.3 32.8 ↑ ↑ Wd 24.9 27.1 ↑ 27.2 29.1 ↑ ↑ We 10.5 11.8 ↑ 16.3 19.1 ↑ ↑

62

In agreement with literature findings, it was found that the smaller particle size yields smoother appearance. It is obvious from Figure 4.20 that BBC20 yields the lowest contrast values for Wd when compared to BBC30 and BBC25. For Wc, Wd, and We there is a clear increase in contrast values with increase in particle size – BBC20 lowest and BBC30 highest (Appendix 19). For du, Wa, and Wb the trend is not as clear. BBC20 still has the lowest value, but when BBC25 and BBC30 are considered, the increase in contrast value does not always follow the increase in particle size. For example, from the du 3-D graph in Appendix 19 it can be seen that at LT1, HT1 and HT2, BBC25 has higher du contrast values than BBC30. The same was observed for Wa and Wb. This could mean that at short-wave ranges the effect of particle size is not as pronounced as in long-wave ranges. From the structure spectrum (Figure 4.21) it can be seen that BBC20 has the lowest contrast values compared to BBC25 and BBC30 although values for du, Wa and We are very similar. The rest of the graphs can be found in Appendix 20. Also, from the graphs in Appendix 20 it can be seen again that the contrast values for du, Wa, and Wb especially at higher temperatures are almost identical and they start separating after Wb. Only at high process temperature (HT) does Wb exceed Wd enough to provide long-wave coverage.

63

TABLE 4.18 - STRUCTURE SPECTRUM FOR BBC20

BBC20, T = 163 °C, t = 90min

Run # Du Wa Wb Wc Wd Wd Average 8.4 20.3 32.1 28.8 27.1 11.8 STDev 1.12 1.66 2.24 2.84 1.44 2.50 C.V. 0.13 0.08 0.07 0.10 0.05 0.21

BBC20, T = 193 °C, t = 35min Average 21.9 44.8 54.2 32.8 29.1 19.1 STDev 1.86 1.57 1.42 1.79 1.79 3.77 C.V. 0.08 0.03 0.03 0.05 0.06 0.20




64

FIGURE 4.19 - APPEARANCE VS. CURE CONDITION FOR BBC 20

24

16

8

40

30

20

50

40

30

32

28

2429

27

25

9080706050403020100

20

15

10

du

Wa 0,1-0,3mm

Wb 0,3-1,0mm

Wc 1-3mm

Wd 3-10mm


163

171

193

(°C)

Temperature

Cure

8.4

21.9

6.65.8

14.9

20.3

44.8

16.514.9

34.8

32.1

54.2

28.525.9

47.8

28.8

32.8

24.424.5

29.3

27.1

29.1

25.024.9

27.2

11.8

19.1

10.810.5

16.3

65

FIGURE 4.20 - WD VS. CURE CONDITIONS FOR BBC20, BBC 25 AND BBC30

FIGURE 4.21 - STRUCTURE SPECTRUM FOR BBC ALL SIZES AT LT1

75

25 50

30

35

160 25

40

170180 0

190

Wd 3-10mm

Cure T ime [min]


BBC 20 µm

BBC 25 µm

BBC 30 µm

Paint Type

0.0

10.0

20.0

30.0

40.0

du Wa Wb Wc Wd We

Co

ntr

ast

Structure Spectrum


BBC20

BBC25

BBC30

66

RBC30 and RBC25 A summary of all appearance elements for both RBC30 and RBC25 can be found in Appendix 21. A 3D plot of Wc for both paints (Figure 4.22) once again shows that the smaller particle size has lower contrast values for Wc. The same can be said for Wd and We (Appendix 22). Dullness (du), Wa, and Wb for both paints have similar values especially at LT1 and HT1 and HT2. It seems from the 3D graphs that the highest difference between two paints is in Wc and Wd. For both RBC25 and RBC30 the structure spectrum at LT1 is shown in Figure 4.23. It can be seen that for RBC25 the shape of the spectrum shows a minimum at Wc. However, only for HT1 and HT2, the ratio between Wb and Wd is close to 1.5. A comparison between RBC25 and RBC30 structure spectrum for all conditions can be found in Appendix 23. Appendix 24 shows structure spectrum development for all 6 paints on the same graph. It can be seen from graphs in that appendix that RBC25 micron is the paint that yields the lowest contrast values and the best spectrum shape. This reinforces the significance of formulation since if only particle size was considered it was anticipated that BBC20 would have the lowest overall contrast values.

FIGURE 4.22 - WC VS. CURE CONDITIONS FOR RBC25 AND RBC30

7520

50

24

28

160 25

32

170180 0

190

Wc 1-3mm

Cure T ime [min]


RBC 25 µm

RBC 30 µm

Paint Type

67

FIGURE 4.23 - STRUCTURE SPECTRUM FOR RBC ALL SIZES AT LT1

4.2.3. Abrasion resistance and Adhesion

The Scribe and Gravel tests for paint adhesion to the substrate and chip resistance were performed on 3 replicates for each cure condition, totaling 15 panels for each paint type. It was found that all panels passed both the adhesion test and Gravel test, regardless of the film build, paint type, or cure condition. Table 4.19 shows a summary of scribe and gravel test as well as the film build and its COV. Gravelometer rating is shown in Appendix 1 and detailed tables for other paint types can be found in Appendix 25. In contrast to the findings reported by Barletta et al. (2007), which indicated that mechanical properties were not fully developed for films baked at low process temperature and time. However, it was found that adhesion properties of the all three powder coatings considered in this research did not change with the process temperature and time. This may be due to the fact that panels were e-coated prior to powdercoating, so the adhesion required is not to the substrate directly. These findings have a significant meaning for the application of powder coatings in the automotive industry. The first and most important finding is that a good finish quality and mechanical properties can be achieved at low temperatures and cure times. This would mean energy savings for the automotive assembly plant. The second result of importance is that colorkey primer yields comparable results with powder basecoats. This means that these primers can be used instead of the basecoat and still get the desirable finish quality. This would translate in cost and material savings for the automotive industry. The third result of interest is that these powder basecoats yielded good finish quality without the presence of defects such as popping, pinhole, or sagging. In addition, all six powder coatings investigated passed the scribe and gravel tests, indicating that not only the appearance, but also the adhesion and abrasion resistence

0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

du Wa Wb Wc Wd We

Co

ntr

ast

Structure Spectrum


RBC25

RBC30

68

properties are acceptable. This would mean that they can successfully replace the waterborne counterparts. The fourth result of importance is the identification of the coating that yields the lowest contrast values (RBC25). Knowing the formulation of a coating that yields the best appearance, would aid in creating coatings with similar formulation but different colors in order to get similar appearance results.

TABLE 4.19 - SUMMARY OF SCRIBE AND GRAVEL TESTS FOR RCP30

RCP30 Run Scribe Gravel E+P [µm] COV

1 pass pass 81.4 3.0 2 pass pass 80.0 2.9 3 pass pass 80.9 3.9 4 pass pass 79.1 3.7 5 pass pass 79.4 1.8 6 pass pass 81.5 2.0 7 pass pass 78.7 1.6 8 pass pass 76.2 2.5 9 pass pass 82.2 2.6

10 pass pass 82.2 4.0 11 pass pass 83.1 1.2 12 pass pass 78.9 2.9 13 pass pass 79.0 2.4 14 pass pass 81.7 3.0 15 pass pass 81.5 2.8

Note: E+P is the film build of ecoat and powder coat. The powder coat film build is not reported individually since the film build before spray was not measured for these panels. However, the ecoat film build was around 22 µm based on the measurements of the 25 X 25 cm panels.

4.3. Conclusion

It can be concluded from these experiments that cure time and temperature have a significant effect on appearance qualities. At lower temperatures contrast values are low indicating a smooth finish; however the shape of structure spectrum is not ideal. As the temperature increases the shape improves, however, contrast values in general also increase. The values of du, Wa and Wb always increase with time and temperature. In general, the smoothness decreases as the time and temperature increase. The next chapter will shed more light on the effect of time as experiments are conducted under isothermal conditions at short time increments (2 minutes and 5 minutes) in order to better understand the effect. Contrast values of We were unchanged as a result of adding powdercoating, which means that waviness of the substrate “telegraphs” through powdercoat.

69

Particle size also is an important factor. When different particle sizes of the same paint formulation were compared, the smaller particle size always led to smaller contrast values of the long-waves (Wc, Wd, and Wd) for all cure conditions. This was true for both red and black basecoats. The results for short-waves were not as consistent. For black basecoat, the smaller particle size gave better appearance at all cure conditions except HT1; while the red basecoat only at LT2 and Cnominal. The colorkey primer had appearance qualities that were comparable with the powder basecoats and so could easily replace the latter. This would eliminate the need to apply the basecoat over the primer for areas of the vehicle that the appearance is not of high importance. After all the research phases are completed, the results will aid in optimizing the application parameters in order to obtain a better finish quality.

70

CHAPTER 5 - PHASE 2: ISOTHERMAL AND DSC POINTS EXPERIMENTS

5.1. Stage A – Isothermal Experiments

The intent of these experiments is to measure how the surface appearance develops during cure at a constant temperature. Three powder coatings were used for these experiments, RBC25, BBC25, and BBC20. After powder coating was applied and panels baked for the time and temperature settings described in Chapter 3 (Materials and Methods), the film build and appearance elements (du – We) were measured and documented. A summary of the film build values and the coefficients of variation (COV) for 163°C i sothermal runs is included in Table 5.1. FB summaries for 171°C and 193°C isother ms can be found in Appendix 26. Since there were three panels per cure condition, FB values represented in Table 5.1 are averages of measurements taken on three panels for each time of the isothermal runs. An example of the data used for calculating FB at each cure time is given in Table 5.2.

Given the large number of panels used for each isothermal run and the large number of contrast measurements, the original spreadsheets are not included here. Instead summary tables are included. These summary tables also show all variables considered in isothermal experiments such as time, Datapaq value (DPV), complex viscosity (η*), film build (FB) and degree of conversion (α). Table 5.3 shows an overall summary of all variables (either measured or calculated) for RBC25 at 163°C. Tables for other isothermal experim ents and powder coatings can be found in Appendix 27. The uncertainty for the ramp time, ideal time and target temperature (Table 5.3) were determined from the Datapaq profile using the procedure that follows. For example for the 163°C/95 min condition (Figure 5.1), it took 8.75 minutes ( target was 7.5 min) to reach 161°C (target 163°C) and after this ramp was completed then the panels stayed in the oven for an additional 93.6 min (target 95 min). The difference between target time/temperature and actual time/temperature yielded the uncertainty values.

It was found that when appearance values were plotted against ideal time above (target time at target temperature after ramp time was completed), real time above (as indicated by Datapaq measurements) or total time (ramp time + time at target temperature), there was no difference in appearance versus time. Appendix 28 illustrates this for RBC25 at 171°C; th e same was true for other powder coatings and cure conditions. As a result, from this point on in the dissertation only ideal time will be mentioned. It should be noted that appearance values (du to We) presented on summary tables are the averages of all

71

measurements for a specified time/temperature combinations. Table 5.4 illustrates the calculation of the coefficient of variation (COV) for each average contrast value; and a summary is included in Appendix 29. It should be noted that Wa and Wb are known as short-waves while Wc, Wd and We are classified as long-waves.

TABLE 5.1 - FB SUMMARY FOR ISOTHERMAL RUNS AT 163°C

T = 163 °C BBC20 BBC25 RBC25

Cure time [min]

FB [µm] COV

FB [µm] COV

FB [µm] COV

0 52.0 0.04 52.9 0.04 51.8 0.08 2 51.6 0.05 51.4 0.06 52.2 0.05 4 53.0 0.03 50.0 0.04 53.6 0.09 6 51.1 0.03 51.4 0.03 54.2 0.04 8 49.5 0.06 49.8 0.02 55.1 0.04 10 51.4 0.05 52.7 0.08 55.5 0.00 12 51.6 0.09 53.1 0.05 55.8 0.05 14 50.0 0.02 53.3 0.05 54.7 0.02 20 50.3 0.01 54.9 0.05 54.6 0.03 25 50.8 0.07 54.4 0.12 54.1 0.04 30 50.9 0.05 52.0 0.03 54.4 0.04 35 50.3 0.03 52.8 0.02 53.0 0.04 40 49.1 0.03 53.5 0.03 52.8 0.08 45 50.4 0.04 54.4 0.07 52.5 0.01 50 52.0 0.04 54.2 0.05 53.3 0.07 55 52.0 0.01 53.0 0.04 51.5 0.04 60 50.9 0.04 51.1 0.07 52.3 0.07 65 53.0 0.03 52.1 0.06 51.0 0.03 70 51.0 0.04 50.0 0.07 51.7 0.06 75 51.6 0.03 50.8 0.03 51.6 0.06 80 48.0 0.03 50.7 0.01 49.9 0.06 85 49.8 0.04 51.3 0.07 51.7 0.05 90 50.3 0.04 50.1 0.01 51.2 0.03 95 48.8 0.03 51.6 0.10 54.4 0.08

TABLE 5.2 - EXAMPLE OF RAW DATA USED TO CALCULATE A VERAGE FILM

THICKNESS AND COV

Cure time [min] Panel # FBa [µm] 2 1 48.5 2 2 52.6 2 3 53.7

Average b 51.6 COV 0.05

a) value in this column is the average of measurements at 9 locations in one panel. b) FB overall average of three panels.

72

FIGURE 5.1 - DATAPAQ PROFILE FOR RBC25, 163 °C AT 95 MINUTES

73

TABLE 5.3 - SUMMARY OF STRUCTURE SPECTRUM FOR 163°C ISOTHERM

Red Basecoat 25 µm (RBC25) 163 ± 4°C Isotherm, Ramp time 7.5 ± 2 min

Real Time a [min]

Ideal Time [±2.5min]

η* [Pa·s] DPV

FB [µm] α du Wa Wb Wc Wd We

7.7 0 27 6 51.8 64.5 2.9 4.9 11.7 27.8 28.3 9.8 9.8 2 65.9 26 52.2 73.2 5.5 11.4 20.0 27.4 28.5 9.8 11.7 4 221 40 53.6 79.3 5.6 11.7 21.4 26.9 27.9 11.1 13.7 6 748 46 54.2 84.2 5.2 11.8 24.8 26.6 28.2 11.0 15.8 8 2,150 68 55.1 88.2 4.7 11.6 24.8 26.4 28.7 8.4 17.8 10 4,960 69 55.5 90.9 5.3 12.3 24.6 26.3 27.5 9.9 19.8 12 9,270 63 55.8 93.0 4.9 10.8 19.5 23.9 27.8 9.3 21.7 14 14,500 79 54.7 94.6 5.1 11.4 19.9 24.7 27.8 8.2 28.2 20 32,100 145 54.6 97.8 3.9 11.5 20.0 21.6 25.2 9.0 33.2 25 45,600 182 54.1 98.9 5.9 16.5 31.1 26.6 27.8 10.3 32.7 30 56,900 166 54.4 98.8 5.6 15.5 29.2 25.8 27.4 11.3 37.8 35 66,200 206 53.0 99.4 6.6 16.6 30.5 28.7 29.2 10.6 47.8 40 73,700 287 52.8 99.8 7.0 17.1 31.0 30.2 29.7 11.2 52.8 45 79,800 320 52.5 99.9 6.5 17.9 32.5 29.3 30.1 11.2 57.8 50 85,000 344 53.3 100.0 7.0 17.7 31.9 29.0 29.5 10.9 62.7 55 89,100 400 51.5 100.0 7.4 17.4 31.7 30.2 30.4 12.3 67.6 60 92,400 416 52.3 100.0 7.2 17.5 31.5 29.5 29.7 11.5 72.2 65 95,300 452 51.0 100.0 6.9 16.5 28.4 29.8 30.1 9.6 77.7 70 97,400 504 51.7 100.0 7.7 18.6 32.8 29.7 29.5 13.3 82.6 75 99,400 528 51.6 100.0 7.1 19.1 34.0 30.2 29.3 13.3 87.8 80 101,000 553 49.9 100.0 9.2 20.1 34.7 31.7 29.7 14.8

74

TABLE 5.4 - CONTINUES

Red Basecoat 25 µm (RBC25) 163 ± 4°C Isotherm, Ramp time 7.5 ± 2 min

Real Time a [min]

Ideal Time [±2.5min]

η* [Pa·s] DPV

FB [µm] α du Wa Wb Wc Wd We

92.9 85 102,000 594 51.7 100.0 8.8 20.9 35.0 30.5 30.7 11.0 98.0 90 103,000 629 51.2 100.0 8.5 22.3 36.6 31.2 30.0 11.1

102.8 95 105,000 677 54.4 100.0 8.3 21.9 36.6 29.0 27.9 11.9 Note: Times reported are after the fast ramp that was 7.5 ± 2min a) Real time was extracted from oven tracer software and is the ramp time plus time at target temperature

75

TABLE 5.5 – EXAMPLE OF RAW DATA USED FOR AVERAGE AN D COV CALCULATION

RBC25, 163°C Isotherm Time Panel Reading du Wa Wb Wc Wd We 8 min Panel 1 Reading 1 2.8 11.2 25.1 24.1 27.4 7.1 8 min Panel 1 Reading 2 5 10.8 23.9 22.7 27.5 6.6 8 min Panel 1 Reading 3 5.9 12.1 25 26.6 30.4 9.8 8 min Panel 2 Reading 1 3.9 11.5 24.4 26.5 29.1 7.6 8 min Panel 2 Reading 2 4.3 10.2 22.6 27.1 27.1 9.7 8 min Panel 2 Reading 3 4.8 11.8 24.1 28.5 31.3 7.3 8 min Panel 3 Reading 1 4.9 12.5 24.6 26.1 29.6 8.9 8 min Panel 3 Reading 2 5.4 13 28 27.4 26.7 11.9 8 min Panel 3 Reading 3 5 11.5 25.6 28.3 29.6 6.3

Average 4.7 11.6 24.8 26.4 28.7 8.4 COV 0.18 0.07 0.06 0.07 0.05 0.29

Actual bake time and temperature for isothermal runs were extracted from the Datapaq oven tracer. Panels that were not cured within the specified ramp time, target time and temperature were discarded from analysis.

5.1.1. DPV and Appearance Datapaq Oven tracer was used to determine the time needed for panels to reach target temperature. Several trial runs were performed until the results were repeatable. To get the real time/temperature profile and calculate practical Datapaq Values (DPVs), probes were attached to the panel after they were sprayed, during cure under isothermal conditions at specified times (Table 3.5 in Chapter 3). DPV is used in the powder coatings industry to assess and optimize cure conditions. DPV is directly related to a paint’s degree of cure (Datapaq, 2009). In order to calculate DPV three points from an exponential constant cure curve, which is unique for a specific paint, are needed. These points are combinations of time and temperature that correspond to the same degree of cure (C0). So the degree of cure for each point on the constant cure curve is equal to the product of cure time and cure rate [C = τR(T)]. In calculating DPVs the major assumption made is that the curing rate (R) is a function of temperature. The relationship between cure time, �, and temperature, T, is expressed by Equation 5.1.

� = ��(��) , � < 0 �� > 0

�� < �� (Equation 5.1)

where A is a constant and calculated from the following expression:

76

A = A = A = A = ��(��)(��) (Equation 5.2 )

Total cure for the whole duration of the process would be evaluated considering the change in cure with the change in time throughout the curing process. In mathematical terms this can be expressed by the integral of the cure rate from zero to the final cure time. Considering the above factors, an expression for total cure can be derived (Eq. 5.3).

�� !� = �� " ��(��)�

� #� (Equation 5.3 )

Where A is a constant, C0 is the standard cure, and the time/temperature pair (τ$,T0) is a point on the constant cure curve. In order to calculate DPV, standard cure is assigned a value of 100. So DPV is calculated using Equation 5.4 (Datapaq, 2009).

&'( = �� " ��(��)�

� #� (Equation 5.4) DPVs were calculated by the Datapaq Insight Oven Tracer software. The three time/temperature pairs selected from the constant cure curve for Standard cure calculations were provided by the paint manufacturer (Table 5.5).

TABLE 5.6 - CURE SCHEDULE FOR DPV CALCULATIONS

Cure Schedule Low Middle High Temperature [ °C] 163 179 193

Time [min] 9 12 14 Minimum temperature = 160°C, Maximum temperature = 193°C Once the parameters shown in Table 5.5 are input in the software, a DPV is calculated for each specific cure scenario (recorded temperature, time profile). Appendix 30 shows the point entry in the software. If the DPV range is 100 to 400 then it is considered good process cure. Values below 100 indicate under-cure and values above 400 indicate over-cure (Schwark, 2009 personal correspondence). These DPVs do not necessarily correspond to the edges of the cure window for these particular paints. It should be noted that under-cure and over-cure are defined here based on process time and temperature specified by the paint manufacturer, and not using kinetic analysis such as DSC, TMA etc. Figures 5.2 to 5.4 show the progress of DPVs with time at isothermal cure conditions for all three paints. It can be seen from these figures that DPVs for the

77

three coatings are the same indicating excellent repeatability of experiments. DPVs for the three paints and the three temperatures are tabulated in Appendix 31.

FIGURE 5.2 - DPV VS TIME AT 163°C ISOTHERM


0

100

200

300

400

500

600

700

800

0 10 20 30 40 50 60 70 80 90 100

DV

P

Time [min]

BBC20 @ 163C BBC25 @ 163C RBC25 @ 163C

0

100

200

300

400

500

600

0 10 20 30 40 50 60 70 80

DV

P

Time [min]

BBC20 @ 171C BBC25 @ 171C RBC25 @ 171C

78


Figure 5.5 to 5.7 show the difference in DPVs at different temperatures for the same paint formulations. It is clear from these figures that different temperatures yield different DPVs for the same cure time.

FIGURE 5.5 - DPV VS TIME AT FOR RBC25

0

100

200

300

400

500

600

0 10 20 30 40 50

DV

P

Time [min]

BBC20 @ 193C BBC25 @ 193C RBC25 @ 193C

0

100

200

300

400

500

600

700

800

0 10 20 30 40 50 60 70 80 90 100

DP

V

Time [min]

RBC25 @ 163C RBC25 @ 171C RBC25 @ 193C

79

FIGURE 5.6 - DPV VS TIME AT FOR BBC20

FIGURE 5.7 - DPV VS TIME AT FOR BBC25

The Datapaq value is based on the measured time-temperature profile during curing. The formula used for DPV calculation includes both time and temperature and as a result could potentially be used as one variable instead of two when predicting appearance values. Minitab statistical software was used for data analysis. Figure 5.8 shows the contrast values (du-We) versus DPV for RBC25.

0

100

200

300

400

500

600

700

800

0 10 20 30 40 50 60 70 80 90 100

DV

P

Time [min]

BBC20 @ 163C BBC20 @ 171C BBC20 @ 193C

0

100

200

300

400

500

600

700

800

0 10 20 30 40 50 60 70 80 90 100

DP

V

Time [min]

BBC25 @ 163C BBC25 @ 171C BBC25 @ 193C

80

Table 5.6 summarizes the results. Both Figure 5.8 and Table 5.6 were used to interpret the behavior of structure spectrum elements with the increase in DPV for RBC25.

TABLE 5.7 - SUMMARY OF APPEARANCE CHANGES WITH TIME AND TEMPERATURE

163°C - 171°C 163°C - 193°C 171°C - 193°C

RBC25

du

Wa

Wb

Wc

Wd

We

BBC25

du

Wa

Wb

Wc

Wd

We

BBC20

du

Wa

Wb

Wc

Wd

We

difference in appearance

no difference in appearance

81

FIGURE 5.8 - STRUCTURE SPECTRUM ELEMENTS VS. DPV FOR RBC25

It can be seen from Figure 5.8 that du, Wa and Wb increase with the DPV for the three temperatures. The increase in contrast values (du, Wa, Wb) was faster for the 193°C isotherm. To see if the difference betwee n contrast values at these temperatures was significant, the F-test for variance and Student’s t-test for the difference in mean was performed. It was found from Student’s t-test, comparing means, that du, Wa, Wb were significantly different for 163°C - 193°C and for 171°C - 193°C isotherms respectively (Table 5.6). W hen 163°C and 171°C isotherms were compared it was found that there was no significant difference in the short-wave values.

For long-wave appearance parameters between 1 mm to 10 mm (Wc and Wd) it seems that the increase in DPV has little effect on the contrast values for all three isothermal temperatures. As the DPV increases there is a slight decrease in Wc contrast values at 163°C and 171°C up to a DP V of 145 (@163°C) and 156 (@171°C), respectively. At 193°C Wc contrast va lues increase steadily until DPV equals 262. After the DPVs indicated above are reached, then as the DPV increases further there is not much change in Wc contrast values at all three temperatures. Wd contrast values are much closer to each other for all three isotherms.

Contrast values for We show a different picture. Values for all three temperatures are similar up to a DPV around 220 and then there is a noticeable increase in We for 193°C, followed by 163°C.

20

15

10

5

0

5002500

40

30

20

10

0

50

40

30

20

10

0

5002500

40

30

20

10

0

40

30

20

10

0

5002500

20

15

10

5

0

du

DPV

Wa Wb

Wc Wd We

163

171

193

T [C]

Target

82

As indicated in the previous paragraph, DPVs below 100 and above 400 yielded panels that were under and over cured, respectively. Considering only cure conditions that yielded DPVs between 100 and 400 (Figure 5.9) it was found that the behavior of appearance parameters was similar to what was described above and illustrated in Figure 5.8. In selecting the optimum time/temperature combination it should be kept in mind that even though contrast values might be low (meaning better appearance) mechanical properties might not be fully developed. That is why using the DPVs as a reference is advantageous in selecting cure conditions that satisfy both criteria.

FIGURE 5.9 - STRUCTURE SPECTRUM VS DPV BETWEEN 100 AND 400 (RBC25)

Figure 5.10 illustrates appearance development with increase in DPV for the three isothermal temperatures for BBC25. Results are summarized in Table 5.6. It can be seen that at 193°C there is a steady incr ease of du, Wa and Wb contrast values. A similar behavior is seen for 163°C and 171°C isotherms, but with less slope. For these two isotherms (163°C and 171°C) contrast values of all spectrum elements are not different. It is observed from Figure 5.10 that Wb has larger variation, than,du and Wa, at 163°C, however , the outcome of a t-test with equal variances (found from the F-test) is that these values are not significantly different from those at 171°C and should be accepte d with caution. It is reported in the literature that the F-test is sensitive to non-normal distributions and outliers and this could be affecting the results. Other results for short-wave values are similar to the ones discussed for RBC25 for all isothermal comparisons.

18

15

12

9

6

450300150

40

30

20

10

50

40

30

20

450300150

40

30

20

10

0

40

30

20

10

0

450300150

20

15

10

5

0

du

DPV

Wa Wb

Wc Wd We

163

171

193

T [C]

Target

83

In terms of long-wave elements, similar to RBC25, they remain almost unaffected by the increase in DPV. Results of comparisons for Wc, Wd, and We values between isothermal temperatures are similar to the ones reported for RBC25 (Table 5.6). Wc and Wd remain unchanged at all isothermal temperatures. We is the same for 163°C and 171°C comparison, but values increase significantly at 193°C.

FIGURE 5.10 - STRUCTURE SPECTRUM ELEMENTS VS. DPV FOR BBC25

20

15

10

5

5002500

50

40

30

20

10

50

40

30

20

5002500

50

45

40

35

30

40.0

37.5

35.0

32.5

30.0

5002500

25

20

15

10

du

DPV

Wa Wb

Wc Wd We

163

171

193

T [C]

Target

84

BBC20 results are illustrated in Figure 5.11 and it can be seen that the trends are similar to the ones discussed for RBC25 and BBC25.

FIGURE 5.11 - STRUCTURE SPECTRUM ELEMENTS VS. DPV FOR BBC20

Figures 5.12 to 5.17 compare du to We values of the three coatings. It can be seen that the three paints give similar contrast values for each isothermal temperature. It was noticed that slightly lower contrast values were achieved with RBC25, which is in agreement with the findings reported in Chapter 4.

FIGURE 5.12 – DULLNESS (DU) VS. DPV FOR ALL THREE P AINTS

25

20

15

10

5

5002500

50

40

30

20

10

60

50

40

30

20

5002500

40

30

20

10

0

40

30

20

10

0

5002500

20

15

10

5

0

du

DPV

Wa Wb

Wc Wd We

163

171

193

T [C]

Target

8006004002000

25

20

15

10

5

8006004002000

25

20

15

10

5

163

DPV

du

171

193

20B163

20B171

20B193

25B163

25B171

25B193

25R163

25R171

25R193

Paint

Panel variable: Target T

85

FIGURE 5.13 - WA VS. DPV FOR ALL THREE PAINTS

FIGURE 5.14 - WB VS. DPV FOR ALL THREE PAINTS

8006004002000

48

36

24

12

0

8006004002000

48

36

24

12

0

163

DPV

Wa

171

193

20B163

20B171

20B193

25B163

25B171

25B193

25R163

25R171

25R193

Paint

Scatterplot of Wa vs DPV


8006004002000

50

40

30

20

10

8006004002000

50

40

30

20

10

163

DPV

Wb

171

193

20B163

20B171

20B193

25B163

25B171

25B193

25R163

25R171

25R193

Paint


86

FIGURE 5.15 - WC VS. DPV FOR ALL THREE PAINTS

FIGURE 5.16 - WD VS. DPV FOR ALL THREE PAINTS

8006004002000

50

40

30

20

8006004002000

50

40

30

20

163

DPV

Wc

171

193

20B163

20B171

20B193

25B163

25B171

25B193

25R163

25R171

25R193

Paint

Scatterplot of Wc vs DPV


8006004002000

40

35

30

25

8006004002000

40

35

30

25

163

DPV

Wd

171

193

20B163

20B171

20B193

25B163

25B171

25B193

25R163

25R171

25R193

Paint


87

FIGURE 5.17 - WE VS. DPV FOR ALL THREE PAINTS

It is clear that there is a direct relationship between DPV and short-wave spectrum elements at 193°C. As DPV increased so did the contrast values for du, Wa and Wb regardless of paint type or particle size. The same was observed for 163°C and 171°C, but at a lower slope. Contrast values for Wc and Wd remained unaffected by the increase in DPV. Actually, an initial decrease in Wc and Wd for RBC25 and BBC20 was observed with the increase in DPV at 163°C and 171°C. However, this decrease occurred before D PV reached a value of 100, which was considered as a minimum value for cure. Those points corresponded to the first few minutes of cure and it could be that the paint was still under the influence of levelling forces.

5.1.2. Viscosity and Appearance

Viscosity is an important characteristic of powder coatings that has been correlated to mechanical and appearance properties of the cured film. Viscosity was measured for RBC25 and BBC25 isothermally at the same temperatures and time intervals as for the isothermal experiments. It has been reported that complex viscosity (η*) can be used to explain appearance and mechanical properties of powder coatings instead of viscosity (Hannon et al.1976). Complex viscosity was calculated and used in this research and from this point on in this dissertation viscosity will refer to complex viscosity. Table 5.7 summarizes viscosity measurements at each temperature and time interval. The reason to measure viscosity only for RBC25 and BBC25 was to see the difference in viscosity behavior between different paint formulations and to rule out the effect of particle size. Also, by measuring the viscosity isothermally it would be possible

8006004002000

25

20

15

10

8006004002000

25

20

15

10

163

DPV

We

171

193

20B163

20B171

20B193

25B163

25B171

25B193

25R163

25R171

25R193

Paint


88

to investigate any correlations with appearance elements cured under the same conditions. The viscosity – temperature profile as provided by the PPG lab can be found in Appendix 32. The viscosities of the coatings studied follow the same pattern when heated. They decrease with increasing temperature, reach a minimum and then increase again due to crosslinking. Figure 5.18 shows calculated complex viscosity development for 163°C, 171°C, and 193°C, respectively for both coatings under isothermal conditions. The decreasing part of viscosity with temperature is not fully shown in Figure 5.18, because this figure represents the viscosity values at the times used for isothermal experiments and for which there is an actual panel baked. It can be seen from Figure 5.18 that for the first 12 minutes both coatings have similar viscosity values at 163°C and 171°C. As the time progresses, RBC25 has a higher viscosity than BBC25. However, viscosities of RBC25 at 163°C and 171°C are almost the same up to 40 minutes. After that time, RBC25 at 163°C has a higher viscosity. As for BBC25, higher viscosity values are observed at 171°C than at 163°C. Viscosity increases at a higher rate at 193°C than at 163°C a nd 171°C for both coatings. It was noticed that at all three temperatures Red Basecoat (RBC25) had the highest viscosity values.

FIGURE 5.18 – COMPLEX VISCOSITY VS. TIME AT THREE ISOTHERMAL TEMPERATURES

0.0E+00

2.0E+04

4.0E+04

6.0E+04

8.0E+04

1.0E+05

1.2E+05

1.4E+05

0 20 40 60 80 100

Vis

cosi

ty [

Pa

.s]

Time [min]

RBC25 @ 163C RBC25 @ 171C RBC25 @ 193C

BBC25 @ 163C BBC25 @ 171C BBC25 @ 193C

89

TABLE 5.8 - ISOTHERMAL VISCOSITY MEASUREMENTS

T = 163 RBC2

5 BBC25

Time η* η*

[min] [Pa·s] [Pa·s]

0.2 27 24.7

2 65.9 44.8

4 221 98.7

6 748 223

8 2,150 466

10 4,960 927

12 9,270 1,710

14 14,500 2,880

20 32,100 8,410

25 45,600 14,500

30 56,900 20,900

35 66,200 26,900

40 73,700 32,200

45 79,800 37,000

50 85,000 4 ,200

55 89,100 44,800

60 92,400 48,000

65 95,300 50,800

70 97,400 53,200

75 99,400 55,400

80 101,00

0 57,100

85 102,00

0 58,700

90 103,00

0 60,100

95 105,00

0 61,400

T = 171

RBC25 BBC25

Time η* η*


0.2 20.5 21

2 64.1 54.2

4 261 139

6 1,050 388

8 3,220 1,020

10 7,270 2,390

12 12,700 4,660

14 19,000 7,720

16 25,400 11,400

18 31,400 15,400

20 37,000 19,500

25 49,000 29,400

30 58,200 38,200

35 65,400 45,600

40 70,900 51,900

45 75,200 57,000

50 78,500 61,300

55 81,100 64,900

60 83,200 67,800

65 84,800 70,200

70 86,200 72,200 Table 5.7 (b)

T = 193°C

RBC25 BBC25

Time η* η*


0.2 38.9 24.5

2 2,060 413

4 22,100 6,890

6 45,200 21,100

8 64,600 35,600

10 78,900 48,800

15 101,000 72,200

20 111,000 86,100

25 117,000 94,400

30 120,000 99,500

35 121,000 103,00

0

40 122,000 105,00

0 Table 5.7 (c)

Table 5.7 (a)

90

There are several parameters that can be determined from rheological measurements such as induction time, which is the time required to reach minimum viscosity, gelation time, which is the time to where G’ = G’’, vitrification time, which is the time when viscosity levels off after the dramatic increase during cure, and finally the minimum viscosity value itself. These parameters for both paints are summarized in Table 5.8. They will help to better interpret the behavior of the appearance parameters in relation to viscosity. It was noticed that the minimum viscosity decreased with the increase in temperature for BBC25.

TABLE 5.9 - SUMMARY OF RHEOLOGICAL PARAMETERS

Paint Type

Temperature [°C]

Gelation Time [min]

Induction Time [min]

Min Viscosity

[Pa.s]

Vitrification Time [min]

Vitrification Viscosity

[Pa.s]

RBC25 163 8.2 7 26.1 60 92,400

RBC25 171 7.4 7 19.2 50 78,500

RBC25 193 1.8 6.2 29.3 25 117,000

BBC25 163 12.4 7 24 75 55,400

BBC25 171 9.4 7 20.1 55 64,900

BBC25 193 2.6 6.4 19.5 30 99,500

In order to conserve energy and material, the viscosity effect was only studied using RBC25 and BBC25 coatings. Figure 5.19 shows wave-scan contrast vs. viscosity for RBC25 and Figure 5.20 for BBC25. Even though there was an increase in contrast values (du, Wa and Wb) since the beginning of viscosity increase, the most noticeable increase in contrast was observed after vitrification (Table 5.8). Viscosity at vitrification times for 163°C, 171°C and 193°C were found to be 92,400Pa.s, 78,500Pa.s and 117,000Pa.s, respectively. These viscosities corresponded to DPVs of 416, 409, and 319 which, except for the last value, were above the 400 upper limit. This indicated that the coating was over-cured. Similar trends were found for BBC25 but at different viscosity values (Figure 5.20).

In terms of Wc and Wd it can be seen from Figure 5.19 that the viscosity does not affect these contrast values for RBC25; similar values are achieved at all three temperatures. For We, contrast values are similar for all three temperatures up to the vitrification viscosity, and after that there is an increase in contrast values of We at 163°C and 193°C. From Figure 5.20 it can be seen that du, Wa and Wb for BBC25 behave similar to RBC25. However, for Wc, Wb and We it seems that there is an oscillating variation of contrast values with viscosity. This could be due to the fact that BBC25 was the paint that had the highest number of data points that did not fall within the target time for cure (ramp and at isothermal temperature). These points make the trends more “choppy”.

It is hypothesized that after vitrification it is not possible for the melted powder coating to level any further and as cure continues, the paint starts to wrinkle and

91

so the appearance worsens. Wrinkling as a phenomenon has been reported in the literature as an outcome of over-curing (Basu, 2005).

FIGURE 5.19 - APPEARANCE VS. VISCOSITY FOR RBC25

FIGURE 5.20 - APPEARANCE VS. VISCOSITY FOR BBC25

20

15

10

5

100000500000

40

30

20

10

0

50

40

30

20

10

100000500000

40

30

20

10

0

40

30

20

10

0

100000500000

20

15

10

5

0

du

Complex Viscosity [Pa.s]

Wa Wb

Wc Wd We

163

171

193

T [C]

Target

20

15

10

5

100000500000

50

40

30

20

10

50

40

30

20

100000500000

50

45

40

35

30

40.0

37.5

35.0

32.5

30.0

100000500000

25

20

15

10

du

Complex Viscosity [Pa.s]

Wa Wb

Wc Wd We

163

171

193

T [C]

Target

92

When appearance vs. viscosity data were compared between the two coatings, it was found that trends for du-We were similar for both paints. However, it was noted that the contrast values for RBC25 were lower, which is surprising, given the generally high viscosities (Figures 5.21 to 5.26). As indicated previously, RBC25 had a higher melt viscosity than BBC25 at all three temperatures.

FIGURE 5.21 – DULLNESS (DU) VS. VISCOSITY FOR RBC25 AND BBC25

FIGURE 5.22 - WA VS. VISCOSITY FOR RBC25 AND BBC25

1200009000060000300000

25

20

15

10

5

1200009000060000300000

25

20

15

10

5

163

Viscosity

du

171

193

25B163

25B171

25B193

25R163

25R171

25R193

Paint


1200009000060000300000

40

30

20

10

0

1200009000060000300000

40

30

20

10

0

163

Viscosity

Wa

171

193

25B163

25B171

25B193

25R163

25R171

25R193

Paint


93

FIGURE 5.23 - WB VS. VISCOSITY FOR RBC25 AND BBC25

FIGURE 5.24 - WC VS. VISCOSITY FOR RBC25 AND BBC25

1200009000060000300000

50

40

30

20

10

1200009000060000300000

50

40

30

20

10

163

Viscosity

Wb

171

193

25B163

25B171

25B193

25R163

25R171

25R193

Paint


1200009000060000300000

50

40

30

20

1200009000060000300000

50

40

30

20

163

Viscosity

Wc

171

193

25B163

25B171

25B193

25R163

25R171

25R193

Paint


94

FIGURE 5.25 - WD VS. VISCOSITY FOR RBC25 AND BBC25

FIGURE 5.26 - WE VS. VISCOSITY FOR RBC25 AND BBC25

1200009000060000300000

40

35

30

25

1200009000060000300000

40

35

30

25

163

Viscosity

Wd

171

193

25B163

25B171

25B193

25R163

25R171

25R193

Paint


1200009000060000300000

25

20

15

10

1200009000060000300000

25

20

15

10

163

Viscosity

We

171

193

25B163

25B171

25B193

25R163

25R171

25R193

Paint


95

5.1.3. Degree of Conversion and Appearance

Differential Scanning Calorimetry (DSC) has been widely used to characterize cure kinetics of powder coatings. DSC analyses were performed for the 6 powder coatings used in this study. It was found that each thermogram demonstrated a glass transition temperature (Tg ) with a strong enthalpic relaxation endotherm, a weak melting peak and a broad curing exotherm. Some of the melting peaks were weak and the Tm data could not be determined. A summary of data is presented in Table 5.9.

TABLE 5.10 - SUMMARY OF TG, TM AND RELAXATION PEAK TEMPERATURE AT

EACH HEATING RATE

Sample ID Heating Rate [°C/min]

Tg [°C]

Relaxation Peak Temp.

[°C]

Tm, Peak Temp

[°C]

Black basecoat B20

5 50 55 84

10 51 57 -

20 54 60 86

Black basecoat B25

5 48 53 76

10 52 58 -

20 54 60 -

Black basecoat B30

5 50 55 -

10 53 57 86

20 54 60 87

Red basecoat R25

5 51 55 -

10 52 58 -

20 55 60 87

Red basecoat R30

5 49 55 -

10 52 57 83

20 54 59 85

Red Primer

5 55 60 -

10 58 62 -

20 59 64 90

Tg = Glass Transition Temperature, Tm = Melting Temperature

A typical thermogram of BBC25 (5°C/min heating rate ) indicating Tg and melting peak is shown in Figure 5.27. Overlays of DSC thermograms at the three heating rates are shown in Figure 5.28. The rest of the DSC overlays at the three heating rates for other powder coatings are included in Appendix 33.

96

FIGURE 5.27 - DSC TYPICAL THERMOGRAM FOR BBC25

76.52°C

71.60°C0.2172J/g

53.38°C

48.26°C(H)

47.25°C

49.54°C

-0.5

-0.4

-0.3

-0.2

-0.1

0.0

Hea

t Flo

w (

W/g

)

0 50 100 150 200 250 300

Temperature (°C)

Sample: Black Basecoat B25Size: 5.7800 mgMethod: TG EVALUATION

DSCFile: H:...\5°C\dsc-Lindi-060409.004Operator: cawRun Date: 04-Jun-2009 10:52Instrument: 2920 MDSC V2.6A

Exo Up Universal V4.2E TA Instruments

97

FIGURE 5.28 - DSC THERMOGRAM WITH OVERLAYS AT THREE HEATING RATES FOR BBC25

20°C/min

10°C/min

5°C/min

-1.5

-1.0

-0.5

0.0

Hea

t Flo

w (

W/g

)

0 50 100 150 200 250 300

Temperature (°C)

Black Basecoat B25––––––– Black Basecoat B25– – – – Black BC B25––––– ·

Exo Up Universal V4.2E TA Instruments

98

The kinetics analysis was performed using the TA instrument Thermal Stability Kinetics software which was based on the ASTM E698 method (Appendix 34).This model assumed the reaction order (n) to be 1. Results of kinetic analysis are summarized in Table 5.10. It can be seen that the onset of cure increases with the heating rate, but there is small variation in between different powder coatings and different particle sizes. Thermograms at different heating rates indicating specific temperatures in each exotherm are shown in Appendix 35. Figure 5.29 shows Arrhenius plots at different conversions for BBC25. Figure 5.30 shows how rate constant (k) changes with temperature at 99% conversion. Graphs for all other coatings are included in Appendix 35.

TABLE 5.11 - SUMMARY OF KINETIC ANALYSIS

Sample ID

Heating Rate

[°C/min]

Onset of

Cure [°C]

End of

Cure [°C]

a)Conversion at peak temperature 99% Conversion

Activation Energy [kJ/mol]

b)LogZ [1/min]

Activation Energy [kJ/mol]

Log Z (1/min)

BBC20 5 110 238

65.9 6.98 60.1 6.21 10 128 260

20 139 290

BBC25 5 109 241

60.6 6.37 63.8 6.58 10 130 264

20 144 290

BBC30 5 115 238

65.9 7.00 73.2 7.61 10 126 257

20 130 278

RBC25 5 116 240

62.4 6.67 72.1 7.65 10 121 253

20 138 284

RBC30 5 108 242

64.4 6.93 76.2 8.08 10 121 257

20 120 277

RCP30 5 114 228

60.7 6.47 70.2 7.50 10 125 257

20 137 268

a) Conversion at the temperature of the exothermal peak. b) Z is the pre-exponential factor

99

FIGURE 5.29 - ARRHENIUS PLOT AT DIFFERENT CONVERSIO NS OF BBC25

0.4

0.6

0.8

1.0

1.2

1.4

Log

[Hea

ting

Rat

e (°

C/m

in)]

1.8 2.0 2.2 2.4 2.6

DSCSample: Black Basecoat B25Operator: cawRun Date: 4-Jun-09 10:52


P(a)

1.0%

10.0%

20.0%

30.0%

60.0%

70.0%

80.0%

99.0%

100

FIGURE 5.30 - RATE CONSTANTS AT 99% CONVERSION OF B BC25

0

1

2

3

4

5

Rate

Const

ant

(1/

min

)

100 150 200 250 300 Temperature (°C)

DSCSample: Black Basecoat B25Operator: cawRun Date: 4-Jun-09 10:52


E: 63.8 kJ/mole +/- 1.51%Log Z: 6.58 1/min +/- 1.51%60 min half-life: 118.1°CEnthalpy: 28.6 J/g +/- 4.12%Conversion: 99.0 %

101

The degree of conversion (α) was calculated using the k values obtained from DSC experiments. Equation 5.4 was used to calculate α at the specified time at each isothermal temperature. Results are summarized in Appendix 27. It should be noted that the time used to calculate α included ramp time since conversion starts before the panel has reached the target isothermal temperatures. So the time used for calculation is the time interval indicated on the “Real Time” column in Appendix 27.

-ln(1-α) = kt (Equation 5.5)

This discussion will focus on any possible relationship between the degree of conversion and the appearance only for RBC25, BBC25 and BBC20, since these were the coatings considered for the isothermal experiments. Figure 5.31 shows degree of conversion for the three isotherms for BBC20. It can be seen that conversion occurs very fast at 193ºC and at around 10 min the coating has reached 99% conversion. At 171ºC it takes about 20 minutes to reach 99% conversion and at 163ºC it takes about 30 minutes. It should be noted that 100% conversion (α = 1) is a theoretical concept and it is never achieved in practice since it requires infinite time. Similar conversion progress trends were observed for RBC25 and BBC25.

FIGURE 5.31 - DEGREE OF CONVERSION (α) AT THREE ISOTHERMS FOR BBC20

0

20

40

60

80

100

120

0 10 20 30 40 50 60 70

Co

nv

ers

ion

Time (min)

163C

171C

193 C

102

Figure 5.32 shows structure spectrum contrast values versus the degree of conversion for RBC25 at the three isotherms. For the 193ºC isotherm it is difficult to interpret the results since 99% conversion is reached very fast and so it seems from the plot (Figure 5.32) that there is an increase in contrast values (du – We) with conversion. However, this could be because conversion is not actually changing and most of the contrast values are plotted against an almost constant conversion value of 99%. In the 163ºC isotherm it can be seen that du, Wa and Wb contrast values show a slight decreasing tendency until a conversion of about 98% then there is an increase. However, as explained above for 193ºC, for many points the increase occurs at a high degree of conversion. In terms of Wc, Wd and We, contrast values decrease up to a conversion of about 95% and then increase.

FIGURE 5.32 - STRUCTURE SPECTRUM VS. CONVERSION FOR RBC25

For BBC25 (Figure 5.33), trends for all contrast values (du – We) after 98% conversion at all three isotherms are similar to RBC25 – increase in contrast with increase in conversion. However, for a degree of conversion between 60% and 98%, all contrast values seem to change very little with the increase in degree of conversion.

For BBC20 (Figure 5.34), results for du, Wa and Wb are similar to BBC25. This was expected given the constant formulation of the paint. However, based on the results for BBC25, it was expected that Wc would almost remain constant with increasing conversion up to values around 98%. In contrast, it can be seen from Figure 5.34 that Wc and Wd values show a slight increase with conversion.

20

15

10

5

1008060

40

30

20

10

0

50

40

30

20

10

1008060

33

30

27

24

21

30.0

27.5

25.0

1008060

18

15

12

9

du

Conversion

Wa Wb

Wc Wd We

163

171

193

T [C]

Target

103

Results for 171ºC again are different from BBC25, however, they are similar to RBC25. Values of Wc and Wd start very high and decrease with increasing degree of conversion up to around 98% and then they decrease sharply with further increase in conversion. There is no particular change for We up to a conversion of 98% and then after that values increase.

FIGURE 5.33 - STRUCTURE SPECTRUM VS. CONVERSION FOR BBC25

FIGURE 5.34 - STRUCTURE SPECTRUM VS. CONVERSION FOR BBC20

20

15

10

5

1008060

50

40

30

20

10

50

40

30

20

1008060

50

45

40

35

30

40.0

37.5

35.0

32.5

30.0

1008060

25

20

15

10

du

Conversion

Wa Wb

Wc Wd We

163

171

193

T [C]

Target

25

20

15

10

5

1007550

50

40

30

20

10

60

50

40

30

20

1007550

35.0

32.5

30.0

27.5

25.0

32

30

28

26

24

1007550

18

16

14

12

10

du

Conversion

Wa Wb

Wc Wd We

163

171

193

T [C]

Target

104

One thing to remember is that the values discussed in the previous paragraphs are determined after the paint has gone through the ramp time of about 8 minutes. Specific ramp times for each isotherm can be found in the corresponding summary tables in Appendix 27. As a result, powder coatings have reached degrees of conversion between 50-60% (163ºC and 171ºC) and 98% (193ºC) before the “time zero” of isothermal runs, which makes it difficult to establish a clear trend for isotherms.

Contrast values of appearance parameters increase even after α=99%. This could be an indication of appearance of wrinkling due to over-cure. Another possibility is that the degree of conversion values are overestimated by calculation, which is reported in literature for DSC kinetic calculations (Neag & Prime, xxxx; Bartletta et al., 2007). So it could be that α has not reached 99% and is still increasing, which could be related to the increase in contrast values.

5.1.4. Before Ramp Experiments and First Few Minutes Isotherms

In order to get a clearer understanding of what happens to appearance values during the first few minutes of ramp, while the panel is reaching the target temperature for isothermal runs, as well as to establish a relationship between low conversion and appearance, experiments were conducted by allowing panels to bake at 2 minutes time intervals up to a total of 8 minutes which corresponds to the ramp time before isotherms (See Materials and Methods section). Appendix 36 shows a summary of all parameters measured or calculated for all three powder coatings that were used in this section (RBC25, BBC25 and BBC20). Since these experiments correspond to the ramp times for each isothermal cure, data in Appendix 36 is grouped based on the target isothermal temperature. DPVs will not be used for this part of the analysis since time is short for the ramp and the cure has not started so DPVs are equal to zero. It takes a few minutes before the DPV starts accumulating (see Appendix 27 & 36). It should be noted that the actual process temperature is much lower than the target isothermal temperature since the panels were left in the oven for shorter amount of time (see “Real Temperature” column in Appendix 36, which is the real temperature at the end of run). For simplicity of discussion 163ºC, 171ºC and 193ºC will be used to distinguish between plots. However, it should be kept in mind that these are not the actual temperatures at which the panels are baked.

In analyzing the results for before-ramp points, data for the first few minutes of isothermal runs after the ramp time (up to 10 min @ 163ºC, up to 8 min @ 171ºC and up to 6 min @ 193ºC) were included as well. This was done to get a continuum of cure from time zero (no ramp), where the degree of conversion is zero, up to about 90% conversion. Figures 5.35 to 5.37 show the progress of structure size values with cure time for all three paints. It can be seen from

105

Figure 5.35 that du, Wa and Wb for all three coatings are initially very high. The contrast values go through a minimum at about 5 minutes and then increase up to about 10 minutes. Between 10 to 15 minutes contrast values remain constant. In terms of Wc, Wd and We, contrast values are very high initially and then they start decreasing until about 7 minutes. After that, contrast values remain constant up to about 16 minutes. Similar trends (du-We) observed at 171ºC as well (Figure 5.36). However for 171ºC, the increase in contrast values for Wa and Wb occur with a steeper slope than at 163°C.

FIGURE 5.35 - STRUCTURE SPECTRUM VS. REAL TIME AT 1 63ºC


40

30

20

10

0

15105

40

30

20

10

0

60

40

20

15105

70

60

50

40

30

60

50

40

30

20

15105

40

30

20

10

du

Real Time [min]

Wa Wb

Wc Wd We

20B163

25B163

25R163

Paint

40

30

20

10

0

15105

40

30

20

10

0

60

40

20

15105

60

50

40

30

20

60

50

40

30

15105

40

30

20

10

du

Real Time [min]

Wa Wb

Wc Wd We

20B171

25B171

25R171

Paint

106

Figure 5.37 shows contrast values for 193°. Trends are similar to 163ºC and 171ºC. It can be seen from the three figures (Figure 5.35 – 5.37) that three paints behave the same with RBC25 showing slightly lower contrast values. Also, another thing to note is that the shape of trends in Figures 5.35 to 5.37 resembles that of viscosity change with time at the beginning of cure. However, it was found that there was no observable trend of structure elements with viscosity for the first few minutes of cure.


Figure 5.38 to 5.40 show structure spectrum contrast values versus the degree of conversion. It can be seen from Figure 5.38 that at 163°C up to α around 25%, contrast values for du decrease, then remain unchanged up to about 60% conversion and then there is a tendency to increase. The same was observed for du at 171°C (Figure 5.39). For Wa, trends are simil ar for 163°C and 171°C – decrease up to α=25%, constant up to 50%, then increase up to 70% and then level off again. At 193°C the trend for du and Wa i s similar. Contrast values reach a minimum at α=45% and then increase (Figure 5.40). For Wb, trends are similar for all three temperatures investigated (163°C, 171 °C and 193°C); values reach a minimum and then increase. However, the minimum is reached at different degrees of conversion, 25% for 163°C and 171°C and 45% for 193°C.

Contrast values for Wc, Wd and We show similar behavior at 163°C and 171°C – they decrease up to about 50% degree of conversion and then level off. For 193°C the lowest value is achieved at α=45% and then the values are almost

20

15

10

5

0

1284

30

20

10

0

60

40

20

1284

60

50

40

30

20

45

40

35

30

25

1284

30

25

20

15

10

du

Real Time [min]

Wa Wb

Wc Wd We

20B193

25B193

25R193

Paint

107

constant up until α=98%. After 98% conversion, contrast values for Wc, Wd and We increase again. The results are similar for all three powder coatings.

FIGURE 5.38 - STRUCTURE SPECTRUM AT 163ºC


40

30

20

10

0

100500

40

30

20

10

0

60

40

20

100500

70

60

50

40

30

60

50

40

30

20

100500

40

30

20

10

du

Conversion

Wa Wb

Wc Wd We

20B163

25B163

25R163

Paint

40

30

20

10

0

100500

40

30

20

10

0

60

40

20

100500

60

50

40

30

20

60

50

40

30

100500

40

30

20

10

du

Conversion

Wa Wb

Wc Wd We

20B171

25B171

25R171

Paint

108


Table 5.11 shows a summary of contrast values from this research, another study conducted with waterborne basecoat sprayed on 25 x 25 cm panels, and values collected in plant on a full scale vehicle. It can be seen from Table 5.11 that appearance values (du to We) achieved with powder basecoats are comparable with the values yielded by waterborne counterparts. This finding would be valuable in terms of future use of powdercoatings instead of solventborne and waterborne basecoats.

TABLE 5.12 – COMPARISON OF APPEARANCE WITH WATERBOR NE BASECOATS

Present Study Powder Basecoat

BBC20, BBC25, RBC25

DOE Study Blue

Water-borne Basecoat

In plant Study Blue

water-borne Basecoat

163°C 171°C 193°C a b du 4.7 - 13 4.4 - 29 6.1 - 24.4 5.8 - 28.9 14.1 - 61 Wa 4.2 - 23 12.6 - 29.4 15.5 - 48.1 12 - 47.6 5.4 - 39.5 Wb 18 - 37 21.1 - 39.5 25.8 - 55.1 28.9 - 63.6 3.9 - 59.2 Wc 21 - 34 22.0 - 34.9 21.5 - 37.1 21.3 - 42.2 8.3 - 49.6 Wd 24 - 40 25.1 - 31.4 25.2 - 32.3 19.6 - 36.8 11.4 - 41.7 We 8.1 - 14.4 7.9 - 14.3 7.9 - 17.9 8 - 17.7 4.4 - 26.9

a) Hemashankar (2008), cure schedule not reported b) Ruvinova (2006), cure schedule not reported

20

15

10

5

0

100500

30

20

10

0

60

40

20

100500

60

50

40

30

20

45

40

35

30

25

100500

30

25

20

15

10

du

Conversion

Wa Wb

Wc Wd We

20B193

25B193

25R193

Paint

109

5.2. Stage B – DSC Significant Points Experiments

Initially six temperatures were selected from the DSC exotherm for BBC25: 40°C, 50°C, 100°C, 145°C, 187°C, and 287°C. It was found that at 40°C and 50°C powder was uncured and it could be blown off the panel. Two panels were allowed to reach 50°C and then left in the oven at that temperature for 15 minutes and 35 minutes, respectively to see if time would affect powder behavior. It was found that still the coating did not melt and was in the same state as when sprayed. This was expected since those temperatures are very close to the glass transition temperature (Tg =51°C, the average of T g for the three heating rates) for BBC25 and about 25°C less than the melting temp erature.

At a temperature of 100°C and 1 minute, the applied powder was cured to the extent that was possible to wave-scan the panel. This temperature is about 24 degrees higher than the melting peak temperature (Tm = 76°C). It can be said that the panel temperature is between 76°C and 100° C that the melted powder coating starts to cure and form its appearance characteristics. Since powder was uncured at 50°C and cured at 100°C it was decided t o run experiments at a temperature in between: 75°C (167°F). This temperat ure is close to Tm. One panel was put in the oven and allowed to reach 75°C and then removed from the oven after 1 minute. The panel was cured and paint was not tacky. However, the appearance was very rough and could not be measured with the wave-scan instrument. At temperatures around Tm, powder coating melts and begins to cure but the levelling is not complete and the surface roughness is high. Temperatures 145°C and 187°C mark the beginning and end, respect ively, of the exotherm revealed in DSC. Panels cured at those temperatures for 1 minute were fully cured and it was possible to wave-scan them. The oven could not reach 287°C, the last temperature selected from the DSC chart, so that point was removed from the test matrix.

Based on the results with the BBC25 coating, the DSC significant points experiments for BBC20 did not include runs at 40°C and 50°C since the powder was not able to melt. Experiments were conducted at 75°C for 15 minutes, 75°C for 35 minutes, 125°C and 215°C for 1 minute. Film build was measured for all panels and for BBC20 was maintained in the range of 46-56µm, while for BBC25 the range was 41-51 µm. A summary of all parameters measured for these experiments is included in Appendix 37.

Figure 5.41 shows development of appearance elements for specific points for BBC25. It can be seen that at 100°C du, Wa and Wb c ontrast values are very high, which is expected since the coating is still levelling. They decrease at 145°C which is the beginning of the exotherm and th en increase again toward the end of the exotherm (187°C). The beginning of t he exotherm indicates the

110

start of the crosslinking reaction between powder coating ingredients. The reaction reaches a maximum which is represented by the peak of the DSC exotherm and then the reaction rate starts to decrease with further curing until the crosslinking is complete. At 145°C, the coatin g has achieved less than 10% conversion while at 187°C about 40% of the coating has reacted. This reinforces the finding that short-wave values increase with the degree of conversion and temperature. However, the initial decrease of contrast values means that while levelling forces are in action the appearance elements are still changing. For Wc, Wd and We it can be seen that after the initial drop in contrast, the values remain almost constant at 145°C and 187°C.

Figure 5.42 illustrates the results for BBC20. The effect of time is clear from this figure. Panels baked at 75°C for 15 minutes show th e worst appearance. However at the same temperature, panels baked at 35 minutes show lower contrast values. This could mean that as the paint starts to melt and flow, the longer the time (the more levelling) results in a better appearance. Then at higher temperatures (125°C and 215°C) results are similar to BBC25.

FIGURE 5.41 - STRUCTURE SPECTRUM FOR BBC25 AT DSC P OINTS (PANELS CURED FOR 1 MINUTE AFTER REACHING THE DESIR ED

TEMPERATURE)

48

36

24

12

0

200150100

48

36

24

12

0

80

60

40

20

200150100

70

60

50

40

30

60

50

40

30

200150100

50

40

30

20

10

du dullness

Temperature [C]

Wa 0,1-0,3mm Wb 0,3-1,0mm

Wc 1-3mm Wd 3-10mm We 10-30mm

100

145

187

[C]

Temperature

111

FIGURE 5.42 - STRUCTURE SPECTRUM FOR BBC20 AT DSC P OINTS (PANELS CURED FOR 1 MINUTE AFTER REACHING THE DESIR ED

TEMPERATURE)

The results of this phase have significant value for the automotive paint process applications. First of all the DPV can replace the time and temperature in predicting the appearance of final finish. This is important since the same DPV can be achieved at different temperature/time combinations and finding the DPV that yields the best appearance at the lowest temperature and least amount of time would mean time, energy and cost savings. Second, the long-wave contrast values are not affected by factors considered. As a result the focus of the basecoat stage application can be directed on choosing the best conditions that optimize short-wave values; since long-waves are affected only by clearcoat application. Third, a particle size of 25 micron seems to yield better appearance than 20 micron. This would mean saving in terms of manufacturing since for some formulations (i.e. RBC) it was difficult to manufacture powder coatings with lower than 25 micron particle size. Also, the larger particle size flows better through equipment. Lastly, knowing the viscosity and conversion where wrinkling starts to appear would enable automotive powder coating users to adjust application parameters and cure conditions to avoid reaching that point and thereby achieve better finish quality.

60

45

30

15

0

200150100

60

45

30

15

0

60

45

30

15

0

200150100

60

45

30

15

0

60

40

20

200150100

50

40

30

20

10

du dullness

Temperature [C]

Wa 0,1-0,3mm Wb 0,3-1,0mm

Wc 1-3mm Wd 3-10mm We 10-30mm

125_1min

215_1min

75_15min

75_35min

Cure Condition

Structure Spectrum vs Temperature at DSC points (BBC 20)

112

5.3 Conclusion

Isothermal experiments give insight on the relationship between structure spectrum elements and Datapaq value (DPV), complex viscosity (η*), degree of conversion (α), time and temperature. It was found that du, Wa and Wb (Wa and Wb are short-wave) steadily increased with the DPV for all three temperatures (163°C, 171°C, 193°C) for the three powder coatings investigated (RBC25, BBC25 and BBC20). Values of du, Wa and Wb at 163°C and 171°C isotherms were very close and in most cases there was no statistically significant difference. However, the values at 193°C were much higher.

In terms of Wc and Wd (long-wave) contrast values were very close at all three isotherms for the three coatings and were slightly related to the change in DPVs. For We contrast values, there was a noticeable increase at 193°C for the three coatings although there is a lot of scatter in these plots. This could mean that cure time and temperature have little effect on long-waves created by basecoats.

Viscosity was found to influence appearance elements as well. The strongest effect was found on du, Wa, Wb and We. While for Wc and Wd there was a slight change in contrast values with an increase in viscosity. It was found that the increase in short-wave contrast values was more pronounced when the viscosity exceeded the value at vitrification point. It was hypothesized that wrinkling appears at this point.

The relationship with the degree of conversion was less obvious than with DPV and viscosity. The reason is that at the temperatures selected for isothermal experiments the degree of conversion reached about 60% within a short amount of time. So the majority of appearance values were measured at 99% of conversion. Even though the degree of conversion was at that seemingly constant value, the appearance contrast values (du – We) continued to increase. This is proof that after the 99% conversion wrinkling appears due to overcure.

Experiments conducted for the first few minutes (0 – 15) of cure indicated that initial contrast values (du-We) were very high, decreased and reached a minimum and then levelled off as the time progressed. Results were similar for all three coatings (RBC25, BBC25 and BBC20). At the low temperatures (163°C and 171°C) short-waves showed an increase after the first level off and then levelled off again. At 193°C contrast values reach a minimum and steadily increase and no range where values remain constant was observed. For long-waves it was found that values decreased with conversion, reached a minimum and then levelled off. This was true for all three temperatures and all three coatings.

113

CHAPTER 6 - PHASE 3: VARYING HEATING RATE EXPERIMENTS

6.1. Comparison between 5 and 10°C/min

In order to investigate the effect ramp speed has on finish quality, two ramp speeds were selected (5°C/min and 10°C/min). The pa rticle size effect was controlled by using powder coatings of the same particle size (25 µm). Two different paints (RBC25, BBC 25) were used to see if paint formulation would make a difference in finish quality at controlled ramp speeds. Figure 6.1 shows that the film build is constant over time and paint type. Raw data for film builds are summarized in Appendix 38. All other application parameters were maintained constant (see Materials and Methods section).

Tables 6.1 and 6.2 summarize values of appearance elements for BBC25 and RBC25, respectively. Each structure spectrum value presented in those tables is the average of several values. The coefficient of variation of each calculated average is shown in Appendix 39. Data analyses were performed using Minitab Statistical Software.

FIGURE 6.1 - FILM BUILD TREND OVER TIME

0

10

20

30

40

50

60

Ru

n 1

Ru

n 2

Ru

n 3

Ru

n 4

Ru

n 5

Ru

n 6

Ru

n 7

Ru

n 8

Ru

n 9

Ru

n 1

0

Ru

n 1

1

Ru

n 1

2

Ru

n 1

3

Ru

n 1

4

Ru

n 1

5

Ru

n 1

6

Ru

n 1

7

Ru

n 1

8

FB

[µ

m]

Film Build Trends

RBC 25

BBC 25

114

TABLE 6.1 - APPEARANCE DATA FOR BBC25

BBC25

Cure Condition

Temperature [°C]

Ramp Speed

[°C/min] *Time [min] du Wa Wb Wc Wd We

LT1_R10 163 10 14 11.1 19.1 29.4 41.1 35.9 14.6 LT2_R10 163 10 25 11.7 20.7 30.6 40.5 33.8 15.9 LT3_R10 163 10 90 16.5 30.4 39.9 44.2 36.4 17.2 MT1_R10 171 10 11 12.8 22.4 32.1 43.0 35.5 16.4 MT2_R10 171 10 20 11.9 22.7 31.5 39.2 35.2 14.5 MT3_R10 171 10 56 15.5 30.3 39.0 43.5 36.1 16.5 HT1_R10 193 10 9 18.2 33.4 42.5 44.6 37.2 17.0 HT2_R10 193 10 15 18.8 36.4 44.0 43.2 35.9 16.5 HT3_R10 193 10 35 26.1 45.4 51.1 45.3 38.3 18.8 LT1_R5 163 5 14 8.3 14.9 25.8 39.7 35.6 17.1 LT2_R5 163 5 25 12.0 21.3 31.9 45.2 36.9 17.6 LT3_R5 163 5 90 11.7 22.5 33.9 45.7 37.4 17.2 MT1_R5 171 5 11 10.4 17.5 28.7 41.1 36.0 14.8 MT2_R5 171 5 20 10.7 19.5 29.9 43.2 38.3 15.0 MT3_R5 171 5 56 12.5 25.7 36.2 40.0 36.3 16.0 HT1_R5 193 5 9 15.7 29.2 39.0 45.7 36.9 16.8 HT2_R5 193 5 15 16.9 32.2 42.6 44.9 37.5 16.9 HT3_R5 193 5 35 24.1 44.6 50.8 45.0 37.1 18.8

*Time = is the time panels were cured at target temperature after the ramp time and set time

were completed.

TABLE 6.2 - APPEARANCE DATA FOR RBC25

RBC25

Cure Condition

Temperature [°C]

Ramp Speed

[°C/min] *Time [min] du Wa Wb Wc Wd Wd

LT1_R10 163 10 14 10.8 18.4 29.8 47.1 39.8 19.4 LT2_R10 163 10 25 10.6 19.4 30.0 44.3 37.2 17.1 LT3_R10 163 10 90 14.7 27.5 38.7 48.2 38.1 17.6 MT1_R10 171 10 11 12.1 20.6 32.4 46.6 37.3 19.8 MT2_R10 171 10 20 11.7 19.9 31.7 45.8 37.2 17.3 MT3_R10 171 10 56 14.8 26.2 38.0 45.3 35.9 19.4 HT1_R10 193 10 15 19.1 35.7 43.9 45.4 38.2 18.4 HT2_R10 193 10 35 24.9 42.9 47.7 42.4 36.3 19.8 HT3_R10 193 10 9 15.0 28.4 38.3 41.5 36.2 16.3 LT1_R5 163 5 14 11.4 15.7 29.9 45.5 37.7 17.5 LT2_R5 163 5 25 10.9 17.5 29.8 44.0 37.5 18.2 LT3_R5 163 5 90 13.5 24.7 36.1 45.0 35.8 15.8 MT1_R5 171 5 11 10.1 16.9 28.9 42.7 36.5 17.3 MT2_R5 171 5 20 12.6 22.2 33.9 47.2 37.9 18.6 MT3_R5 171 5 56 17.1 29.9 40.5 47.3 37.7 17.3 HT1_R5 193 5 15 19.8 35.9 46.3 45.5 38.2 17.3 HT2_R5 193 5 35 27.5 45.3 51.5 47.6 37.8 20.3 HT3_R5 193 5 9 16.5 29.9 40.2 47.2 37.6 20.3

115

Figure 6.2 shows appearance elements at the two ramp rates for BBC25 163°C at 14 minutes. It can be seen that between ramp speeds: du, Wa and Wb are slightly different with the 10°C/min ramp giving hi gher values. The Wc, Wd and We contrast values however, were not affected by the varying ramp rate. At the same temperature but 25minute bake time (Figure 6.3) it can be seen that Wc and Wd are slightly different at the two ramps with the 5°C/min ramp giving the higher values, while the rest of the appearance parameters are not affected. Increasing the cure time to 90 minutes (163°C) resu lted in different du, Wa and Wb values (10°C/min ramp higher values) and no chan ge in Wc, Wd and We contrast values (Figure 6.4). This result was similar to the 14 minute bake time at the same temperature; however the difference between values was higher.

Similar results as for 163°C@14minutes and 163°C@90 minutes were seen for panels cured at 171°C and 11 minutes (Figure 6.5). For 171°C and 20 minutes which is considered the nominal cure condition for these specific paints, it was found that du, Wb and We were not affected by ramp speed while Wa, Wc and Wd were slightly different. The 10°C/min ramp gave higher Wa values while the 5°C/min ramp gave higher Wc and Wd values (Figure 6.6). At 171°C and 56 minutes again very slight difference on du and Wa was observed between ramps, with the 10°C/min leading to higher values for elements that were different (Figure 6.7). The shape of the spectrum however, was more curved than for previous cure conditions.

FIGURE 6.2 - STRUCTURE SPECTRUM FOR BBC25 AT LT1_R5 AND LT1_10

0.0

10.0

20.0

30.0

40.0

50.0

du Wa Wb Wc Wd We

Co

ntr

ast

RR5_163C_14min RR10_163C_14min

116



0.0

10.0

20.0

30.0

40.0

50.0

du Wa Wb Wc Wd We

Co

ntr

ast


0.0

10.0

20.0

30.0

40.0

50.0

du Wa Wb Wc Wd We

Co

ntr

ast


117

FIGURE 6.5 - STRUCTURE SPECTRUM FOR BBC25 AT MT1_R5 AND MT1_10

FIGURE 6.6 - STRUCTURE SPECTRUM FOR BBC25 AT MT2_R5 AND MT2_10 (NOMINAL CONDITIONS)

0.0

10.0

20.0

30.0

40.0

50.0

du Wa Wb Wc Wd We

Co

ntr

ast


0.0

10.0

20.0

30.0

40.0

50.0

du Wa Wb Wc Wd We

Co

ntr

ast


118

FIGURE 6.7 - STRUCTURE SPECTRUM FOR BBC25 AT MT3_R5 AND MT3_10

Panels cured at 193°C at three different times (9, 15 and 35 minutes) gave the same contrast values for both ramps (Figures 6.8 to 6.10).

It seems that at each temperature at the lowest cure time only the shortwave parameters are slightly affected by the ramp speed variation. With the increase in cure time the differences become smaller and smaller. The difference between contrast values however, is very small and could be due to variation between

experimental parameters.

FIGURE 6.8 - STRUCTURE SPECTRUM FOR BBC25 AT HT1_R5 AND HT1_10

0.0

10.0

20.0

30.0

40.0

50.0

du Wa Wb Wc Wd We

Co

ntr

ast


0.0

10.0

20.0

30.0

40.0

50.0

du Wa Wb Wc Wd We

Co

ntr

ast


119

FIGURE 6.9 - STRUCTURE SPECTRUM FOR BBC25 AT HT2_R5 AND HT2_10

FIGURE 6.10 - STRUCTURE SPECTRUM FOR BBC25 AT HT3_R 5 AND HT3_10

Figures 6.11 – 6.13 shows appearance elements for RBC25 cured at 163°C for 14, 25 and 90 minutes. It can be seen that in general contrast values of all elements are the same for both ramps.

At 171°C and 11minutes Wa, Wb and Wc are different at the two ramps while du, Wd and We are the same (Figure 6.14). It is the 10°C/min ramp that again yields the highest values. At the same temperature but increased cure time (20 minutes) the results showed no difference in any of the appearance elements for both ramp rates (Figure 6.15). Figure 6.16 indicates that by increasing cure time to 56 minutes the Wa and Wb were slightly different while the rest of the appearance parameters remained the same. In this case it was the 5°C/min ramp that gave the highest contrast values.

0.0

10.0

20.0

30.0

40.0

50.0

du Wa Wb Wc Wd We

Co

ntr

ast


0.0

10.0

20.0

30.0

40.0

50.0

60.0

du Wa Wb Wc Wd We

Co

ntr

ast


120

Figure 6.17 shows that for 193°C and 9 minute cure time the only values that differed at the two ramp speeds were Wc and We with the 5°C/min ramp giving the higher values. Increasing the time to 15 minutes showed only a slight difference at Wb (Figure 6.18). As for the 9 minute case, it was the 5°C/min ramp that gave the higher value. While increasing the cure time further to 35 minutes Wb and Wc were slightly different; the 5°C/min gave the highest values. (Figure 6.19). As with BBC25 in cases where the appearance elements had different values at the two ramp rates, the difference was less than 5 units.

FIGURE 6.11 - STRUCTURE SPECTRUM FOR RBC25 AT LT1_R 5 AND LT1_10


0.0

10.0

20.0

30.0

40.0

50.0

du Wa Wb Wc Wd Wd

Co

ntr

ast


0.0

10.0

20.0

30.0

40.0

50.0

du Wa Wb Wc Wd We

Co

ntr

ast


121


FIGURE 6.14 - STRUCTURE SPECTRUM FOR RBC25 AT MT1_R5 AND MT1_10


0.0

10.0

20.0

30.0

40.0

50.0

60.0

du Wa Wb Wc Wd We

Co

ntr

ast


0.0

10.0

20.0

30.0

40.0

50.0

du Wa Wb Wc Wd We

Co

ntr

ast


0.0

10.0

20.0

30.0

40.0

50.0

du Wa Wb Wc Wd We

Co

ntr

ast


122


FIGURE 6.17 - STRUCTURE SPECTRUM FOR RBC25 AT HT1_R5 AND HT1_10


0.0

10.0

20.0

30.0

40.0

50.0

du Wa Wb Wc Wd We

Co

ntr

ast


0.0

10.0

20.0

30.0

40.0

50.0

du Wa Wb Wc Wd We

Co

ntr

ast


0.0

10.0

20.0

30.0

40.0

50.0

du Wa Wb Wc Wd We

Co

ntr

ast


123


From the figures discussed in the previous paragraphs it can be seen that the structure spectrum development is very similar for all time/temperature pairs and for both paints. This could mean that a difference of 5 degrees in heating rates is not sufficient to produce a significant difference in appearance values. In addition, an increase in process temperature (from 163°C to 193°C) seems to dampen the effect of ramp speed even further especially for BBC25. None of the cure scenarios or heating rates generated the ideal spectrum with a local minimum at Wc.

6.2. Comparison of 5°C/min and 10°C/min with a Very Fast Ramp

A better looking spectrum was achieved in previous runs (Chapter 4) where process temperature and time were the same as most of the cure scenarios discussed in this chapter. However, the heating rates used in Chapter 4 were very fast (average 65°C/min). Figures 6.20 and 6.21 show the structure spectra for panels baked at 163°C and 14 minutes at three r amp speeds for BBC25 and RBC25 respectively. It can be seen from Figure 6.20 that the fast ramp yields lower Wc, Wd and We values, but the du, Wa and Wb are close to that for 10°C/min. It is interesting to see however, that th e shape of the spectrum for the fast ramp leans toward the ideal spectrum. From Figure 6.21 it is obvious that the fast ramp gives a minimum at Wc and the values of all elements are lower than at 5 and 10°C/min. However, as discussed in Chapter 4, this time and temperature pair does not give the long-wave coverage which is a desired appearance phenomenon.

0.0

10.0

20.0

30.0

40.0

50.0

60.0

du Wa Wb Wc Wd We

Co

ntr

ast


124

FIGURE 6.20 - STRUCTURE SPECTRUM FOR BBC25 AT THREE RAMPS (163°C@14 MIN)

FIGURE 6.21 - STRUCTURE SPECTRUM FOR RBC25 AT THREE RAMPS (163°C@14 MIN)

Figures 6.22 and 6.23 show the structure spectra for panels cured at the three heating rates at the high temperature/time condition for BBC25 and RBC25, respectively. It can be seen that in both figures the values are not much different except for Wc and Wd. However, the shape of the spectrum is very close to ideal – minimum at Wc and short-waves are about 1.5 times higher than long-waves (Wb is about 1.5 times Wd). This indicates that fast heating rates can yield better appearance in terms of the shortwave coverage and spectrum shape. In terms of the magnitude of contrast values, excluding Wc and Wd, there is little difference between the three rates for both paints especially at high temperature and longer cure times.

0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

40.0

45.0

du Wa Wb Wc Wd We

Co

ntr

ast

high ramp 10C/min 5C/min

0.0

10.0

20.0

30.0

40.0

50.0

du Wa Wb Wc Wd We

Co

ntr

ast


125

FIGURE 6.22 - STRUCTURE SPECTRUM FOR BBC25 AT THREE RAMPS (193°C@35 MIN)

FIGURE 6.23 - STRUCTURE SPECTRUM FOR RBC25 AT THREE RAMPS (193°C@35 MIN)

0.0

10.0

20.0

30.0

40.0

50.0

60.0

du Wa Wb Wc Wd We

Co

ntr

ast


0.0

10.0

20.0

30.0

40.0

50.0

60.0

du Wa Wb Wc Wd We

Co

ntr

ast


126

6.3. Conclusion

The appearance values of panels baked at 5 and 10°C /min were similar. In cases where the values were different, the difference was subtle and less than 5 units. The difference was more pronounced for BBC25 than for RBC25. The results could mean that a difference of 5 degrees in heating rates is not sufficient to produce a statistically significant difference in appearance values.

It was noticed that in none of these low ramp cure scenarios was the ideal spectrum with a minimum value at Wc achieved. The long-wave coverage was not observed as well.

When results at these two ramps were compared with the results of experiments conducted at the same temperature/time pairs but at a much faster ramp (average 65C/min) it was found that for both paints the faster ramp yielded generally lower contrast values for Wc and Wd. The structure spectrum for the very fast ramp has a closer to ideal shape, which is more obvious at the 193°C/35min condition, and there is better long-wav e coverage. Based on these results the ramp speed will not be considered in the model generation. The reason is that experiments with these ramp speeds are only conducted for selected points of the cure window and there are not enough experiments “in between” low ramp speeds (5, 10°C/min) and high (65°C/min).

127

CHAPTER 7 – PHASE 4: MODEL GENERATION

7.1 Levelling Model

Levelling of powder coatings is known to affect the appearance of the final film. However, there are only a few studies reported in the literature that have investigated the mechanism of levelling of powder coatings. The basis for all levelling models presented in the literature has been the theory by Smith et al. (1961), Orchard (1962) and Rhodes (1968). Orchard’s equation (1962) gives an expression that describes the reduction of undulations in the paint film. A simplified equation (Equation 7.1) has been derived from Orchard’s equation (1962), and verified by Rhodes (1968), which represents the levelling process in the case where the film thickness of coatings is smaller than the wavelength of surface undulations. This has been proven to be true during levelling of powder coatings (Spitz, 1973).

�� = )� �*+,-./0

/1,2 (Equation 7.1)

Where: A = amplitude of undulations A0 = initial amplitude

3 = viscosity 4 = surface tension 5 = wavelength of undulations X = mean film thickness t = time Rearranging Equation 7.1 to solve for lambda (wavelength) it can be shown that the wavelength of undulations increases as their amplitude decreases with time. This increase in wavelength, as the levelling process proceeds, has been shown to describe the formation of a smoother film (a decrease in amplitude gives a decrease in surface roughness) and it is dependent on film build, viscosity and cure time (Spitz, 1973). In this research it was decided to use Equation 7.1 as a starting model to describe the levelling process and calculate the wavelength (λ*) of surface undulations. However, since the amplitude of the appearance parameters considered in this study (Wa – We) was not available, the contrast values calculated by wave-scan DOI were used instead. Also, the value of viscosity was not kept constant (as it was the case of the original equation); complex viscosities measured at each isothermal temperature were used in these calculations. This way to some extent, the change in viscosity with time was

128

taken into account. Film build measurements were used instead of the mean film thickness. The wavelength was calculated with Equation 7.2.

1∗ = 7 (89/02∗ )�/,

;< (=>)�/, (Equation 7.2)

Where:

λ* = calculated wavelength of undulations [µm] FB = film build [µm]

η* = complex viscosity [?

@A B]

Wi = contrast value of appearance element (Wa, Wb, Wc, Wd, We)

C = constant; C = CD EFGH/IJH/I , C1 = constant from units conversion.

σ = surface tension = 0.035 N/m Using Equation 7.2, lambda values for each appearance element were calculated. Since it has been reported that these wavelengths are dependent on time, viscosity and film build, the calculated values were plotted against the “FB

term” (KLMN

O∗ ) as shown in Figure 7.1. The FB term was plotted in a logarithmic

scale. Lambda* were calculated by pooling together the Wi values at the three temperatures (163°C, 171°C and 193°C). It can be seen that there is a high correlation between the calculated values and the FB term, which is in good agreement with the findings reported in the literature (Spitz, 1973). The data were fitted to an equation of the form presented in Equation 3.

1∗ = P(89/02∗ )Q ( Equation 7.3)

where: a and b are constants. It is interesting to see that the calculated lambda* values for each appearance element overlap. This was expected since the calculated lambda* is in fact an average wavelength of the superposition of several periodic waves. As a result, even though each appearance element has its own wavelength range, they can be represented by an overall wavelength which ranges from 0.1 – 30 mm. This range is inclusive of the individual wavelength ranges from Wa to We. The same trend was observed for BBC25 (Figure 7.2). Plotting the logarithm of lambda* values and the logarithm of the FB term values, it was found that a linear relationship exists (Figure 7.3). The data were fitted into Equation 4.

129

log(λ*) = A + B*log(89/02∗ ) (Equation 7.4)

Where A and B are constants. The value for constant A was 1.683 and for B was 0.25 which are in excellent agreement with the values reported by Spitz, (1973) (1.674 and 0.243 respectively).

FIGURE 7.1 - LAMBDA* FOR APPEARANCE ELEMENTS OF RBC 25

FIGURE 7.2 - LAMBDA* FOR APPEARANCE ELEMENTS OF BBC 25

y = 52.102x0.252

R² = 0.9983

0

500

1000

1500

2000

2500

3000

1 10 100 1000 10000 100000 1000000 10000000

λ*

[µm

]

FB*t/η

Undulations for RBC25

lambda*(Wa)

lambda*(Wb)

lambda*(Wc)

lambda*(Wd)

lambda*(We)

Power (lambda*(We))

y = 50.958x0.2523

R² = 0.9983

0

500

1000

1500

2000

2500

1 10 100 1000 10000 100000 1000000 10000000

λ*

[µ

m]

FB^3*t/η

Undulation Progress for BBC25

lambd *(Wa)

lambda*(Wb)

lambda*(Wc)

lambda*(Wd)

lambda*(We)

Power (lambda*(We))

130

FIGURE 7.3 - LOG(LAMBDA*) VS. LOG(FB TERM) FOR BBC2 5

Utilization of this model to describe the levelling process in terms of increasing λ* is deemed adequate. However, the model cannot forecast the behavior of appearance elements (Wa to We). As a result, other model generation scenarios were considered.

7.2 Regression Model

Regression analysis was performed to find a relationship between dependant (du, Wa, Wb, Wc, Wd and We) and independent variables. Finding such a relationship will enable development of a predictive model for the appearance parameters. Three types of multiple regression models were investigated: stepwise forward, stepwise backward and the best subset. Multiple regressions were used since several variables were identified to influence the appearance elements. The adjusted R-Sq (Adj R-Sq) value was used to determine the adequacy of any regression model since this value compensates for the number of independent variables, as is not the case with the coefficient of determination (R-sq).

The forward stepwise regression starts with selection of the independent variable that has the lowest probability value (p-value). This procedure is followed by adding one variable at a time and checking the p-value. The process stops when the last variable with the p-value lower than the confidence limit (α) value specified is added to the model. The p-value is the probability of rejecting the null hypothesis. The smaller the p-value the smaller is the probability that rejecting the null hypothesis is a mistake at the specified confidence limit. In this study the confidence limit for all regression models was set to 95% (α = 0.05).

y = 0.2502x + 1.6846

R² = 0.9998

0

0.5

1

1.5

2

2.5

3

3.5

4

0 1 2 3 4 5 6 7

log

(λ*

)

log(FB^3*t/η)

BBC25

lambda*(Wa)

lambda*(Wb)

lambda*(Wc)

lambda*(Wd)

Linear (lambda*(Wd))

131

The backward regression starts with all independent variables and on each step it eliminates the variable with the highest p-value. The process stops when all the variables left in the model have p-values lower than the alpha value, which means that they are statistically significant.

The best subset regression identifies the set of variables that best predict the response. There are several combination of independent variables considered in the set; and the set with the highest adjusted R-sq value , and the smallest Mallow’s Cp value is the model considered to fit the response the best (Minitab, 2010).

Any combination of variables that gave adjusted R-sq values higher than 64% was considered as strong correlation and a good predictor of the response. Regression analysis was performed individually for each temperature (163, 171, 193°C) and for each powder coating (RBC25, BBC25, and BBC20) used in isothermal tests. It should be noted that since viscosity data were not available for isothermal runs with BBC20, that variable was not considered in the regression model for this coating. Individual outcomes of all-variable, forward, backward and best subset regressions, for each coating and each isothermal temperature, are included in Appendix 40. It was found that some variables were common for all regressions and some appeared only in a specific regression model. Table 7.1 summarizes the common and uncommon predictor variables for all three paints at the three temperatures.

132

TABLE 7.1 – SUMMARY OF COMMON PREDICTORS

du Adj R-Sq Wa Adj R-Sq Wb Adj R-Sq Wc Adj R-Sq Wd Adj R-Sq W e Adj R-SqCommon DPV 80.6% DPV, α 89% 80.9% FB 75% FB <63% <63%

Uncommon η DPV, α, η DPV, α, η η DPV, FB, η

Added in sub settingCommon DPV, FB, α 93% DPV, FB, η 95.6% DPV, FB, α 95.6% FB 84% FB <63%Uncommon α DPV, η DPV, η

Added in sub setting η 95.6% α 83.50%Common DPV 97.00% DPV, η 95.9% DPV, η 95% η 96% η <63% DPV 93.50%Uncommon FB, DPV FB FB

Added in sub settingCommon at all temperatures DPV DPV, η DPV DPV, FB, η FB, η DPV, FB

du Adj R-Sq Wa Adj R-Sq Wb Adj R-Sq Wc Adj R-Sq Wd Adj R-Sq W e Adj R-SqCommon DPV <63% α, η <63%

Uncommon α, η DPV

Added in sub settingCommon DPV, α, η 73.3% DPV, α, η 96.3% DPV, η 96.5% DPV, FB 74.5% FB <63% <63%Uncommon α DPV, η

Added in sub setting FB 73.3% FB 96.3%Common DPV 95.4% DPV 92.1% DPV 85.7% DPV <63% DPV <63% DPV <63%Uncommon FB α η η

Added in sub setting

Common at three temperatures DPV, FB DPV DPV, α DPV DPV

du Adj R-Sq Wa Adj R-Sq Wb Adj R-Sq Wc Adj R-Sq Wd Adj R-Sq W e Adj R-SqCommon DPV, FB <63% DPV 74% DPV, α 87% DPV, FB 84% FB, α <63%Uncommon FB

Added in sub settingCommon DPV, FB, α 79.9% DPV 83.3% DPV <63% FB, α 68.9% DPV, α 70.8% DPV <63%Uncommon FB, α

Added in sub settingCommon DPV 97.7% DPV 96% DPV 93.60% DPV, FB 96.1 DPV 73.60%Uncommon FB, α FB FB

Added in sub settingCommon at three temperatures DPV, FB DPV, FB DPV FB DPVshown only on two temperatures either in common or uncommon factorsshown in all either as common or uncomon

Adj R-Sq is the highest of forward or backward regre ssion

171°C

193°C

163C°

163C°

171°C

193°C

BBC20

163C°

171°C

193°C

RBC25

BBC25

133

To see the individual contribution of each predictor variable in the overall regression model, each variable was regressed individually. Results are summarized in Table 7.2.

Analyzing the results of different regression models (Table 7.1) and regressing each dependent variable against each independent variable individually (Table 7.2), predictive factors were selected (Table 7.3). Models were only generated for RBC25 and BBC25. The reason BBC20 was not included was because of the lacking of information about viscosity. Since the viscosity was found to be a factor of influence for a few appearance elements of RBC25 and BBC25 films, it was thought that constructing a model for BBC20 without including the viscosity would not be adequate. Predictor models are summarized in Tables 7.4 to 7.6.

It was found that du, Wa and Wb at 193°C can be predicted adequately only using DPV. The only exception is Wa for RBC25 which included viscosity as well. However, if viscosity was removed the Adj R-Sq was still above 80%.

For Wb (RBC25@193°C) similar to Wa, viscosity appeared as one of the important variables. Adding viscosity to the Wb model for RBC25 at 193°C only increased the regression coefficient by 3%. This was not considered a significant increase and as a result the model with DPV only was considered. For Wb for BBC25 at 193°C, DPV was responsible for 85.7% (Adj R-Sq) of the Wb behavior. However, regression analysis showed that conversion was one of uncommon factors that could predict Wb values. When conversion was added to the regression model along with DPV, the Adj R-Sq coefficient only increased 1.8%. In this case, again only the model with DPV as a predictor was considered. Adjusted R-sq values for du, Wa and Wb at 193°C for both coatings, were high (83.9% to 96.7%), which indicates a high correlation. The model for Wc included DPV, Viscosity and FB for RBC25. However, for BBC25 there was no variable that predicted the behavior of Wc. For this coating there were no variables to predict Wd and We. For RBC25, similar to BBC25, there were no predictors for Wd. However, DPV was found to be significant predictor (Adj R-Sq = 92.3%) for RBC25.

As the temperature was decreased from 193°C to 171°C, the DPV was still one of the predictive factors. However, film build, viscosity and conversion appeared as significant for du, Wa, Wb and Wc. For both coatings there were no predictive factors for Wd and We. As the temperature was lowered further (163°C) it was found that none of the appearance parameters could be predicted for BBC25 and nor could Wd and We for RBC25. For RBC25, the results for du, Wa, Wb and Wc were similar to RBC25 at 193°C. The only exceptions were Wb @163°C and Wc at 193°C that had viscosity as additional predictor and Wa@163°C that did not have viscosity.

Note: Only variables that contributed more than 25% on predicting the response are included (Adj R-Sq = 25% or higher) 134

TABLE 7.2 – SUMMARY OF REGRESSION ANALYSES USING IN DIVIDUAL INDEPENDENT VARIABLES

163°C 171°C 193°C RBC 25

Predictor R-sq (adj) Predictor R-sq (adj) Predictor R-sq (adj) du DPV, η 80.6, 74.7 DPV, η 77.1, 65.0 DPV, η 96.7, 77.5

Wa DPV, η, α 83.4, 83.3, 62.1 DPV, η 86.7, 69.6 DPV, η 94.2, 85.2

Wb η, DPV, α 76.8, 72.9, 61.9 DPV, η 89.0, 65.0 DPV, η 92.7, 87.1

Wc FB, DPV, η 64.8, 54.2, 54.3 FB 71.7% η, DPV 88.9, 79.9

Wd FB, DPV, η 50.5, 32.9, 37.8 FB 28.9 η, α, FB, DPV 46.9, 43.5, 34.8, 28.8

We η, DPV, FB 42.2, 41.6,35.6 none DPV, η 92.3, 70.1

BBC25 Predictor R-sq (adj) Predictor R-sq (adj) Predictor R-sq (adj)

du none η, DPV 50.4, 57.6 DPV, η 94.7, 76.8,

Wa DPV, α 29.5, 26.9 DPV, η, 91.1, 85.1 DPV, η 93.5, 84.5

Wb α 32.1 DPV, η 94.2, 86.8 DPV, η, α 83.9, 83.6, 50.7

Wc none FB 64.9 DPV 29.3

Wd none FB 61.3 η, DPV 30.3, 50.3

We none none η, DPV, α 40.3, 43.5, 28.9

BBC20 Predictor R-sq (adj) Predictor R-sq (adj) Predictor R-sq (adj)

du DPV, FB 44.7, 36.9 DPV 47.4, DPV 97.2

Wa DPV, α, FB 70.7, 52.6, 28.7 DPV 79.8 DPV 95.1

Wb DPV, α, FB 79.9, 70.4, 26.2 DPV 34 DPV, α 92.1, 45.3

Wc FB, DPV, α 80.7, 33.2, 28.4 DPV, α 42.7, 63.2 DPV, FB, α 80.2, 48.5, 45.3

Wd none DPV, α 52.0, 66.3 none

We none none DPV 73.6

135

The fact that two of the long-wave constituents (Wd and Wd) could not be predicted, was in line with the results of Chapter 5, which indicated that these elements do not change significantly as the cure proceeds.

TABLE 7.3 – SUMMARY OF SIGNIFICANT PREDICTOR VARIAB LES

163°C 171°C 193°C RBC25 BBC25 RBC25 BBC25 RBC25 BBC25

du DPV None DPV,FB,α DPV,η,α DPV DPV Wa DPV None DPV,FB,η DPV,η,α DPV,η DPV Wb DPV,η None DPV,FB,α DPV,η DPV DPV Wc DPV,FB None DPV,FB,η DPV, FB DPV, FB,η None Wd None None None None None None We None None None None DPV None

TABLE 7.4 - FINAL PREDICTOR MODELS FOR RBC25 AND BB C25 AT 163°C

163°C RBC25 Appearance Equation Adj R -Sq

du du = 4.54 + 0.00646 DPV 80.6 Wa Wa = 10.5 + 0.0179 DPV 83.4% Wb Wb = 20.3 + 0.00718 DPV + 0.000103 η* 76.4% Wc Wc = 73.7+ 0.00457 DPV - 0.886 FB 74.3% Wd None We None

163°C BBC25 Appearance Equation Adj R -Sq

du None Wa None Wb None Wc None Wd None We None

136



du du = 24.3+ 0.00896 DPV - 0.256 FB - 0.0691 α 93.0% Wa Wa = 33.6 + 0.0251 DPV - 0.425 FB - 0.000046 η* 94.8% Wb Wb = 49.6 + 0.0233 DPV - 0.301 - 0.149 α 95.1% Wc Wc = 69.5 + 0.00992 DPV - 0.852 FB - 0.000043 η* 82.0% Wd None We None


du du = 11.5 + 0.0152 DPV - 0.000053 viscosity - 0.0579 α 71.5% Wa Wa = 22.7 + 0.0335 DPV - 0.000080 η* - 0.104 α 96.0% Wb Wb = 22.4 + 0.0365 DPV - 0.000100 η* 95.9% Wc Wc = 66.5 + 0.00388 DPV - 0.687 FB 74.6% Wd None We None



du du = 4.61 + 0.0326 DPV 96.7% Wa Wa = 13.6 + 0.0444 DPV + 0.000066 η* 96.2% Wb Wb = 25.1 + 0.0640 DPV 92.7% Wc Wc = 50.9 + 0.00855 DPV + 0.000026 η* - 0.547 FB 96.4% Wd None We We = 8.05 + 0.0183 DPV 92.3%


du du = 7.46 + 0.0286 DPV 94.7% Wa Wa = 17.9 + 0.0587 DPV 93.5% Wb Wb = 30.0 + 0.0567 DPV 83.9% Wc None Wd None We None

The model adequacy was checked by performing residual analysis (Appendix 41). It was found from the normal probability plot of residuals, that values were close to the straight line indicating that residuals follow a probability distribution. Also, from the residual vs. fits plot it was found that there was no specific pattern from the plotted residuals which meant that the models were adequate. The high Adj R-sq value indicates that the model is a good predictor of the behavior of the response. Also, it should be noted that on all models selected to predict the appearance variables, the predictive factors had a p-value lower than 0.05 which means that they are statistically significant at the 95% confidence level.

137

These results mean that by adjusting the cure conditions to achieve a specific DPV, it will be possible to get the desired response in short-wave parameters. This is true for all temperatures and powder coatings. The appearance of viscosity as an important factor, especially at lower temperatures, could be explained with the fact that the film is still under the influence of levelling forces, and as indicated by the levelling model (Section 7.1), viscosity is an important factor. However, multicollinearity might be responsible for these results as well. Multicollinearity makes it difficult to determine which of the variables is truly contributing to the response, if they are correlated. A value of 0.9 of greater for the Pearson correlation indicates the existence of high multicollinearity. For values of 0.8 and below the variables are not considered as colliner. A correlation matrix of independent variables was calculated to determine if they were correlated (Table 7.7). The cells highlighted in green show variables that are correlated. It was found that DPV and viscosity are highly correlated. This was expected since DPV has time as part of the mathematical calculation and viscosity changes with time as the coating is heated at a specific temperature. The correlation between DPV and viscosity could be the reason for both being included in the regression models. Also, it should be kept in mind that these findings are for basecoats only and normally for the outer surfaces there is a clearcoat layer that is applied as well. The du should improve after the clearcoat is added.

TABLE 7.7 – CORRELATION MATRIX FOR BBC25 AT 193°C

DPV Viscosity FB Conversion DPV 1

Viscosity 0.936 1 FB -0.214 -0.020 1

Conversion 0.665 0.795 0.186 1

7.3 Conclusion

The modified Orchard’s model is a good descriptor of levelling process. It indicated that time, viscosity and film build can be used to illustrate the levelling process in terms of the lengthening of the orange peel amplitude, as predicted by Spitz (1973). Using the contrast values instead of the amplitudes in the Orchard’s model, still yielded plausible results. However, the model cannot be used to predict individual behavior of each appearance element (Wa to We).

The regression model seemed to be the best predictive model for short-waves. It was found that DPV was the common factor of all temperatures and coatings considered. The next most common variable was viscosity. At 171°C, film build and conversion appeared as important factors for both RBC25 and BBC25. With

138

the exception of We for RBC25 at 193°C, where DPV was found as important predictive factor, no model was found to predict the behavior of Wd and We for both coatings and the three temperatures. This concludes that DPV is the most important factor in forecasting the behavior of short-wave appearance elements. This is of great importance, since by controlling the short-wave elements it is possible to control the overall appearance through long-wave coverage.

139

CHAPTER 8 – CONCLUSIONS AND RECOMMENDATIONS

8.1 Conclusions

This research investigated the effect of particle size, cure time and temperature, Datapaq value (DPV), degree of conversion (α), viscosity (η), ramp speed, film build (FB) and formulation on appearance quality of two automotive powder basecoats and one automotive colorkey primer. Colors considered were red and black for the basecoats and dark red for the colorkey primer. The particle sizes considered were 20, 25 and 30 µm for the black basecoat (BBC20, BBC25, BBC30), 25 and 30 µm for the red basecoat (RBC25, RBC30) and 30 µm for the colorkey primer (RCP30). The appearance was quantified using the contrast values of the structure spectrum elements (du, Wa, Wb, Wc, and Wd).

It was found that the contrast values of du to Wc for the e-coated panels, before powder coating was applied, were constant and showed a small variation. We values however, showed a high variation. As expected, after application of any of the powder coatings investigated, the contrast values of du-Wc increased when compared to those pre-spray. It was surprising to find that for We however, there was no difference in pre-spray and after-spray contrast values. This is a demonstration of the substrate waviness “telegraphing” through the coating.

For all cure conditions short-wave contrast values increased with time, temperature and DPV. Curing panels at 163°C and 171 °C, resulted in similar short-wave contrasts values. For 193°C, however the re was a dramatic increase in short-waves. In terms of long-waves, there was no appreciable change with time, temperature or DPV.

Viscosity was found to affect short-waves as well – increase in viscosity resulted in an increase in short-wave values. However, in contrast to the trends observed with DPV, the trend lines in short-wave values did not separate based on temperature. Instead, the increase followed a continuous polynomial curve. The increase was more pronounced after vitrification point suggesting that wrinkling appears after this point. Again, long-waves remained unaffected by the viscosity. All appearance values (du to We) for the first 10 minutes of cure without ramp followed the same trend as the melt viscosity. They started high, went through a minimum and then increased again. This reflects the effect of the melt viscosity at the beginning of cure. Both short-waves and long-waves showed an increase at conversion values equal to or greater than 99%. It was hypothesized that the increase in contrast values after the chemical reaction is completed, is due to appearance of wrinkling. These findings were true for all particle sizes and powder basecoats investigated.

There was no difference in contrast values of structure spectrum elements cured at ramp speeds of 5 and 10°C/min. The ideal spectru m was not achieved for either of the time/temperature pairs and either of these ramps. In general, faster ramps (> 55°C/min) yielded lower contrast values an d smoother appearance. Also, the fast ramp and 35 min/193°C gave the ideal spectrum shape and good

140

long-wave coverage (a maximum at Wb and a minimum at Wc and the short-wave, specifically Wb values, 1.5 times greater than long-waves). This suggests that with powder coatings faster ramps yield better appearance, which is in contradiction with findings from other studies which report that longer cure at low viscosity yields better levelling. There was no appreciable difference in appearance for 5 and 10°C/min ramps. The results co uld mean that a difference of 5 degrees in heating rates is not sufficient to produce a significant difference in appearance values.

When different particle sizes of the same paint formulation were compared, as expected from published literature, the smaller particle size always led to smoother appearance. Smaller contrast values of the long-waves (Wc, Wd, and Wd) were obtained for all cure conditions. This was true for both red and black basecoats. The results for short-waves were not as consistent. For black basecoat, the smaller particle size gave better appearance at all cure conditions except HT1; while the red basecoat only at LT2 and Cnominal.

In terms of formulation, RBC25 was the powder coating that gave the smoothest finish. Given that the general formulations of black and red basecoat were similar, the difference could be attributed to the pigment type. Even though RBC25 had the smallest contrast values, the values for all powder basecoats and colorkey primer were comparable to values generated from panels treated with waterborne basecoat and solventborne clearcoat. The importance of this finding is two-fold. First, the colorkey primer can easily substitute the basecoat layer in certain areas of the vehicle. Second, the powder basecoats can replace waterborne counterparts.

The relationship with the degree of conversion was less obvious than with DPV and viscosity. The reason is that, at the temperatures selected for isothermal experiments, the degree of conversion reached about 60% by the time of the first measurement. So the majority of appearance values were related to 99% conversion. Even though the degree of conversion was at that seemingly constant value, the appearance contrast values (du – We) continued to increase. This is evidence that after the 99% conversion wrinkling appears due to over-cure (longer process time and temperature).

The modified Orchard’s model is a good descriptor of the levelling process. It indicated that time, viscosity and film build can be used to illustrate the progress of levelling. Using the contrast values instead of the amplitudes in the Orchard’s model, still yielded plausible results. However, the model cannot be used to predict individual behavior of each appearance element (Wa to We).

The regression model seemed to be the best predictive model for short-waves. It was found that DPV was the most common statistically significant, independent factor for all temperatures and coatings considered. The next most common variable was viscosity. No predictive model was found for long-waves. This leads

141

to the conclusion that DPV is the most important factor in forecasting the behavior of short-wave appearance elements. This is of great importance, since by controlling the short-wave elements it is possible to control the overall appearance through long-wave coverage.

In general, this study suggests that powder basecoats and colorkey primers can yield appearance qualities comparable to waterborne counterparts. A reduction in cost, energy and waste could be possible, if powder coatings are implemented. This study came to the conclusion that DPV is the most important factor controlling the appearance of powder basecoats. In order to achieve better appearance of powder coatings, DPVs should be controlled to a range between 100 and 400 and combined with fast heating ramps would make it possible to achieve the desired appearance. The extended and high temperature cures lead to higher contrast values – low appearance quality. At low process temperature and time adhesion is still acceptable based on the scribe and gravel tests. Viscosity was found important in many cases however, there is a strong correlation between DPV and viscosity, so it is difficult to draw any clear conclusions about the effect of viscosity.

8.2 Significance of this Research

The processes involved in painting a vehicle produce the most emissions to the environment of the entire vehicle manufacturing process. End-of-pipe solutions have been implemented in the past to control manufacturing emissions. The next quantum leap in reducing coating emissions will come from materials substitution or elimination, such as the use of powder paints. In order to be implemented on a widespread basis, a more scientific understanding of powder coating processes and the environmental impacts associated with them will be needed.

Even though many researchers have studied the effect of spray application devices on paint particle size, very few studies have been conducted to investigate the relationship between the finish appearance to the paint particle size. Even less has been done to understand the effect that the paint levelling near the surface has on the product appearance.

The results of this research will aid in optimizing the cure parameters in order to obtain a better finish quality. A high quality finish minimizes the need for the vehicle to be repainted and as a result decreases the material and energy use as well as emissions to water, land and air.

8.3 Recommendations

• This study only investigated one colorkey primer. In order to draw more generalized conclusions, it would be necessary to repeat the study with different sizes and colors of colorkey primers.

142

• Isothermal experiments were conducted only with basecoats. Repeating these experiments with colorkey primer will shed more light on the performance of these coatings.

• Black basecoat was used that had a smallest size of 20 microns, while the smallest size for red basecoat was 25 microns. It would be interesting to run the experiments with 20 micron red basecoat and see if the 25 micron is still superior or not. Also, it would be beneficial if different particle sizes were investigated, to see which particle size gives an optimum appearance.

• The powder coatings used were non-metallic in formulation. Investigating metallic powder coatings would widen the range of their application.

• Experiments were conducted on 25 x 25 cm panels mounted vertically on the holding rack. It would give more insight on appearance of powder coatings if these experiments were conducted considering the horizontal orientation as well. Also, repeating experiments in a whole vehicle would be beneficial to determine if the curvature of the surface would have any effect on appearance.

• The study suggested that faster ramp leads to improved appearance. However, experiments with varying ramp speeds are only conducted for selected points of the cure window and there were not enough experiments “in between” cure conditions for 5, 10 and 65°C/min to identify the ramp speed where the change occurs to get the maximum at Wb and the minimum at Wc while at the same time maintaining low contrast values.

• The powder coatings used were high gloss formulations. The contrast values achieved were within the range of waterborne basecoat + clearcoat finishes. It would be interesting to see the results of running these experiments with powder basecoat or colorkey primer and clearcoat.

• Film thickness was maintained within as tight a range as possible. However, studies have shown that thickness of applied coating affects the appearance of the final finish. It would be beneficial to repeat the experiments by changing the film thickness.

• Since DPV was found to be the most important factor, it would be beneficial to narrow the range to an optimum that would consistently yield

143

better appearance. Low process temperature and time would be desirable in terms of appearance and energy savings. However, there is need to find the lowest temperature/time pair that would save energy and lead to acceptable appearance without compromising the adhesion and abrasion resistance of the final finish. To determine these conditions, weathering tests would be useful.

• In terms of predictive models, there is the need to consider models of higher than the first order and different arrangements of parameters. This could be done using specific software that generates curve fitting equations.

144

REFERENCES IN ALPHABETICAL ORDER

Adams, J. S. (1989). E.L.P.O. and powder coat – new paint development and implementations. Finishing ’89 Conference Proceedings, Cincinnati, Ohio.

Anastasopoulos, Angelos;Prendi, Lindita; Tam, Edwin (2009). Life Cycle Inventory and Data Application Protocols for the Automotive Paint Processes. JCT CoatingsTech. V. 6, No. 6, pp. 26-35. Andrei, D. C., Hay, J. N., Keddie J. L., Sear, R. P. and Yeates, S. G. (2000). Surface levelling of thermosetting powder coatings: theory and experiment. J. Phys. D: Appl. Phys. V. 33, pp.1975-1981.

Ansdell, D.A. (1999). Automotive Paints. In: R. Lambourne and T.A. Strivens (Eds) Paint and Surface Coatings - Theory and Practice (2nd Ed.) (Abington, Cambridge, England: Woodhead Publishing Ltd), pp. 411-491.

Arjona, M.M. (2000). Powder coating takes the eco-challenge. Modern Metals, V. 56, No. 4, pp. 52- 54.

Barletta, M., Lusvarghi, L., Mantini, F. P. and Rubino, G. (2007). Epoxy-based thermosetting powder coatings: surface appearance, scratch adhesion and wear resistance. Surface & Coating Tech., V. 201, pp. 7479-7504.

Basu, K. Soumendra, Scriven, E. L., Francis, F. L., McCormick, V. A., Reichert, R. Veronica (2005). Wrinkling of Epoxy Powder Coatings, Journal of Applied Polymer Science, V. 98, pp. 116-129.

Bate, D. A. (1990). The Science of Powder Coatings, V. 1, SITA Technology: London.

Biris, A. S., Mazumder, M.K., Yurteri, C. U., Sims, R. A., Snodgrass, J. and De, S. (2001). Gloss and texture control of powder coated films. Particulate Sci. & Tech., V. 19, No. 3, pp. 199-217.

BYK – Gardner (2008). QC Solutions for Coatings and Plastics, Columbia, USA.

Challener, C. (2005). The VOC factor, part 1: resins and paints. Journal of Coatings Technology, V. 2, No. 12, pp. 36-43.

Chiou, J. (1999). Volatile organic compounds emission control study of surface coating operation on Taiwan automobile industry. Proceedings of the 2nd Pacific Cleaner Production Roundtable (Brisbane, Australia: Australian Cleaner Production Association Inc).

145

Challener, C. (2006). Environmentally friendly paints and coatings: making the world a greener place. JCT CoatingsTech, V. 3, No. 7, pp. 28-34.

Conesa, C., Saleh, K., Thomas, A., Guigon, P. and Guillot, N. (2004). Characterization of flow properties of powder coatings used in the automotive industry. KONA, V. 22, pp. 94–106.

Crea, P. (1996). Automotive powder coating. Products Finishing, pp.52–56.

Davis, D. (2000). Improved coating processes hold promise for automotive finishers. Industrial Paint & Powder, V. 76, No. 4, pp. 24-26.

Despatch Industries (2007). Protocol Plus Control (Heat only Version 4.0 Instruction Manual). Minneapolis, MN: Despatch Industries.

Dobson, D.I. (1996). Life cycle assessment for painting processes: putting the VOC issue in prespective. Progress in Organic Coatings, V. 27, pp. 55-58.

Gherlone, Luigi; Rossini, Tullio; Stula, Vladimir (1998). Powder coatings and differential scanning calorimetry: the perfect fit. Progress in Organic Coatings, V. 34, pp. 57-63.

Grandhee, S. (2003). Applications for automotive coatings. In: D. Urban and K. Takamura (Eds) Polymer Dispersions and Their Industrial Applications (Wiley – VCH: Weinheim, Germany), pp. 163-187.

Hannon, M. J., Rhum, D. and Wissbrun, K. F. (1976). Model for the correlation of rheological measurements with the levelling and cure of thermosetting powder coatings. J. Coatings Tech., V. 48, No. 621, pp. 42-54.

Haldankar, S. G., Shalati, D. M., DeGooyer, J. W., Gessner, A.M., Bosma, M.,

Brinkhuis, H.C.R. and Vijerberg, A.M.C. (2008). Novel rheology control agents. JCT CoatingsTech, V. 5. No. 6, pp. 38-43.

Hazel, J.N. (1996). Using LCA to select coatings for optimum environmental and cost performance. Journal of Coatings Technology, V. 68, No. 861, pp. 117-121.

Hazel, J.N. (1997). A life-cycle assessment study of automotive basecoats. Design for Environmentally Safe Automotive Products and Processes, SAE Special Publications, 1263(970699), pp. 111–115.

ITWGema (2010). Powder Applicator [accessed online on February 2010] http://main.itwransburg.com/Products/Product/ID:47 Joesel, H.K. (2001). Automotive applications for UV-cure. Industrial Paint & Powder V. 4, pp. 26-28.

146

Josefsson, E. (1997). Powder painting of truck cabs-helping to preserve the environment. Industrial Robot, V. 24, No. 2, pp. 144-148.

Jones, N.F. (2001). Outlook for zero-VOC resins. J. Coatings Tech., V. 73, No. 916, pp. 63–71.

Kenny, J.C., Ueno, T. and Tsutsui, K. (1996). Analytical approach for high quality appearance powder coatings. J. Coatings Tech., V. 68, No. 855, pp. 35–43.

Kia, F.S., Rai, N.D. and Simmer, C.J. (1997) Capture of overspray powder paint in a wet booth system. Journal of Coatings Technology, V. 69. No. 875, pp. 23-31.

Kim, B.R., Kalis, E.M. and Adams, J.A. (2001). Integrated emissions management for automotive painting operations. Pure & Applied Chem., V. 73, No. 8, pp. 1277–1280.

Lee, Sun Sang; Han, Z. Y. Hank; Hilborn, G. Jons (1999). Surface structure build-up in thermosetting powder coatings during curing. Progress in Organic Coatings, V. 36, pp. 79-88.

Mage, D. and Zali, O. (1992). Motor Vehicle Air Pollution (Geneva, Switzerland World Health Organization).

Nix, V. G. and Dodge, J. S. (1973). Rheology of powder coatings. J. Paint Tech., V. 45, No. 568, pp. 59.

Orchard, S. E. (1962). On surface levelling in viscous liquids. Journal of Applied Scientific Research, Section A, V. 11, pp. 451 – 464.

Osterhold, Michael (1996). Characterization of surface structures by mechanical and optical Fourier spectra. Progress in Organic Coatings, V. 27, pp. 195-200.

Osterhold, Michael and Niggemann, Frank (1998). Viscosity – temperature behavior of powder coatings. Progress in Organic Coatings, V. 33, pp. 55-60.

Osterhold, Michael and Armbuster, Kornelia (2009). Characterizing the appearance of medium glossy surfaces. Progress in Organic Coatings, V. 65, pp. 440 – 443.

Papasavva, S., Kia, S., Claya, J. and Gunther, R. (2001a). Life cycle analysis of automotive powder primers. European Coatings Journal, V. 1, pp. 92-95.

147

Papasavva, S., Kia, S., Claya, J. and Gunther, R. (2001b). Characterization of automotive paints: an environmental impact analysis. Prog. Organic Coatings, V. 43, pp. 193–206.

Papasavva, S., Kia, S., Claya, J. and Gunther, R. (2002). Life cycle environmental assessment of paint processes. Journal of Coatings Technology, V. 74, No. 925, pp. 65-76.

Prendi, L., Henshaw, P. and Tam, E. K.L. (2006). Automotive coatings with improved environmental performance. Int. J. Environmental Studies, V. 63, No. 4, pp. 463-471.

Prendi, Lindita; Anastasopoulos, Angelos; Tam, Edwin (2008). Life Cycle Inventory of the Automotive Paint Processes. JCT CoatingsTech, V. 5, No. 6. Rhodes,J. R., Ph.D. Thesis (1968). Rheology of Vitreous Coatings, Ohio State University, Columbus, Ohio, USA.

Sims, R. A., Mazumder, M.K., Chok, W., Wankum, D. L., Guo, W., Tebbetts, G. and Chasser, T. (1999). Effect of particle size distribution on transfer efficiency and appearance in powder coatings. Particulate Sci. & Tech., V. 16, No. 3, pp. 239-248.

Smith, N. D. P., Orchard, S. E., and Rhind-Tutt, A.J. (1961). J. Oil and Color Chemists' Association. V. 44, pp.618.

SpecialChem (2010). Powder Coating Center. [Accessed on line on March 2010] http://www.specialchem4coatings.com/tc/powder-coatings/index.aspx?id=market

Spitz, G. T., (1973). Levelling or Powder Coatings. ACS Div. Org. Coatings and Plastics Chem. Papers. V. 33, No. 2, pp. 502-507.

Szirtes, T. (1998). Creating Efficient Graphs with dimensionless variables. Machine Design. http://www.machinedesign.com.

Tighe, C., Giroux, J. and Henshaw, P. (2008). Effect of clearcoat thickness on surface roughness. JCT CoatingsTech, V. 5, No. 6, pp. 24-28.

TA Instruments (2008). A review of DSC kinetics methods, TA-073. TA instruments, Thermal Analysis & Rheology, A subsidiary of Waters Corporation.

Uhlmann, P. and Grundke, K. (2001). Influence of additives on interfacial phenomena during film formation of powder coatings. J. Coatings Tech., V. 73, No. 915, pp. 59–65.

148

Wallace, I. (2002). A Geography of the Canadian Economy. Oxford University Press, Toronto, Canada.

Yanagida , K., Yamamoto, M. and Tsuitsui, K. (1998). J. Powder Coatings, V. 3, pp. 28.

Yasuhara, K. (1995). Surface pretreatment of aluminium for automotive applications. In: G. Fettis (Ed) Automobile Paints and Coatings (VCH Verlagsgesellschaft: Weinheim, Germany), pp. 9-27.

149

APPENDICES

Appendices can be found in the CD accompanying this Dissertation

150

VITA AUCTORIS

Lindita Prendi (Poloska) was born in 1973 in Tirana, Albania. She graduated from the University of Tirana in 1996 where she obtained a bachelor’s degree in Chemical Engineering. In 2000 she graduated from St. Clair College of Applied Arts and Technology and received a Diploma of Technology in Chemical Engineering Technology (Industrial / Environmental). She then studied Environmental Engineering at the University of Windsor where she completed her undergraduate, master’s and doctoral work.

AUTOMOTIVE POWDER COATINGS: AN INVESTIGATION OF …

Documents

Transcript of AUTOMOTIVE POWDER COATINGS: AN INVESTIGATION OF …