Formulate Product

1
Formulate Product
13
you need experiments in the lab
to get a recipe
don't do too many experiments
2
The Case
refrigerator door
powder coating
'orange peel'
from molten particles
3
Powder Coating versus Paint
no solvent
(toxic) solvent
recycles unused coating
overspray wasted
several layers required
single layer sufficient
conflicting flow requirements
viscosity adjusted separately
now dominates for coating of parts
4
Materials
di-ol
NPG
neo-pentyl-glycol
di-acid
TPA
terephthalic acid
di-acid 2
IPA
isophthalic acid
5
Polyester (1)
3 h, 240 oC SnCl4 cat
H2O
10 mol di-ol 9 mol di-acid
hydroxyl capped
(length distribution)
6
Polyester (2)
2 h, 240 oC SnCl4 cat 10 mm Hg
2 mol di-acid 2
O
O
C
C
OH
non-crystalline polymer
two acid end-groups
this is cooled, milled to powder
7
Cross-Linker (1)
TGIC
forms 3 links
tri-glycidyl-iso-cyanurate
tri-functional
a powder
carcinogenic
polyester and cross-linker mixed as powders
extruded and milled (120 oC)
TiO2 pigment
above TG, below cross linking temperature!
8
Coating Powder
sintered polyester and TGIC
NOT a molecular mix
DP
no cross-linking at low T
four parameters
poly-ester
TGIC
DP
DTGIC
Dpolyester
eTGIC
DTGIC
Dpolyester
9
Cross-Linker (2)
C
O
C
O
C
100 s at 220 oC
main reaction
10
Orienting ('Screening')
if not done earlier
'play' a bit with experiments
get your instruments working
try to get a 'workable' product
find experimental limits
11
The Great Number
type of di-ol (1) type of di-acid 1 (2) type of
di-acid 2 (3) fraction of di-acid 2 (4) chain
length polyester (5) amount of cross-linker
(6) particle size polyester (7) particle size
cross linker (8) particle size final powder
(9) curing temperature (10) curing time
(11) layer thickness (12)
investigate systematically?
5 points per variable
12 variables
512 2.4 108
experiments
1 experiment per day
requires 700 000 years
covering design space by experiment not possible
12
How to Reduce Experimenting
'understand' the process
1
split off 'independent' parts
2
find dependencies with a model
3
4
use a good search method
do experiments simultaneously
5
you learn to understand while working on the
problem
13
Our Understanding
1
cross-linked
cold substrate
substrate heated
electrical adhesion
polyester melts TGIC diffuses
solid film
14
Independent Parts
2
type of di-ol (1) type of di-acid 1 (2) type of
di-acid 2 (3) fraction of di-acid 2 (4) chain
length polyester (5) amount of cross-linker
(6) particle size polyester (7) particle size
cross linker (8) particle size final powder
(9) curing temperature (10) curing time
(11) layer thickness (12)
requirements of film known
polymer properties
literature may help
stoichiometric dependence
volume fraction TGIC 0.02..0.05
15
Particle Size Relations
2
particle size polyester (7) particle size cross
linker (8) particle size final powder (9) curing
temperature (10) curing time (11) layer thickness
(12)
largely determined by spraying requirements
small thin layer
large rough surface
30 mm
16
Spreading of a Drop
'dimensional analysis'
3
D0
flattening
s
h
t gt
flattened when
D
roughly
17
Diffusion of TGIC
3
dimensional analysis
R1
melting
time needed for diffusion
t 0
diffusion
t lt tD
diffused
t gt tD
roughly
18
Variables Left
type of di-ol (1) type of di-acid 1 (2) type of
di-acid 2 (3) fraction of di-acid 2 (4) chain
length polyester (5) amount of cross-linker
(6) particle size polyester (7) particle size
cross linker (8) particle size final powder
(9) curing temperature (10) curing time
(11) layer thickness (12)
choose via bulk properties of the polymer
one variable
not important
important, 1..4 mm?
spraying
30 mm
important, above Tmelt
take 'long enough'
spraying
70 mm
19
Search by Experiment (1)
4

5
dP particle size polyester
TC curing temperature
dP
measure smoothness 010
5
4
do three experiments
3
choose new point depending on result
TC
scientists don't do it this way
20
Search by Experiment (2)
4

5
search for an optimum
dP
dP
dP
4
4
6
4
6
5
4
5
4
5
4
6
3
3
3
TC
TC
TC
it is faster to change two or more variables at
the same time
21
Summary
concept
try to understand
lab techniques as in research
list variables
split off parts
but
use models
you must learn to understand many variables
improve
search with method
test in parallel
recipe
22
Spreading of a Drop (2)
disk model
100
0
0
100
200
300