1. Scratch Resistance of PP as a Function of MFR and fiber content
Jungsub Lee1, Ilhyun Kim1 and Byoung-Ho Choi1*
1School of Mechanical Engineering, Korea University, Seoul, Korea
*Corresponding Author :

2. Ⅰ Ⅱ Ⅲ Ⅳ Ⅴ Contents Introduction Methods Materials and Experiments
Results & Discussion Ⅳ
Conclusion Ⅴ

3. Introduction Ⅰ

4. 1. Introduction
Recently, polymeric materials are widely used in many industrial fields.
Unlike metal, material property data of polymers are insufficient.
Especially, polymers have many differences in material properties with metals.
Ref)

5. 1. Introduction
Polymer materials have different structures and properties according to its monomer or additives
Understanding microstructures and their properties are closely related.
For example, melt flow index (MFI), number average molecular weight (Mn­) and weight average molecular weight (Mw) can be used to find the intrinsic properties of material.
For these reasons, researches have been focused on evaluating intrinsic properties by using structural properties.
Ref) Tim A. Osswald and Georg Menges, Material Science of Polymers for Engineers, In Material Science of Polymers for Engineers (Third Edition), Hanser, 2012

6. 1. Introduction
The relationship between mechanical properties and structural properties can be different as per the type of polymers
Young’s modulus of semi-crystalline polymers (PP and HDPE) are sensitive to the change in density while that of amorphous polymers are not.
Processing is also an important factor determining mechanical properties of polymers

7. 1. Introduction Scratch damages of polymer
Scratch refers a surface damage caused by dynamic indentation on polymers and are closely related to the emotional quality of the product.
It is mainly understood in aesthetical point of view but it can also cause functional failures.
Increasing use of polymeric materials results in scratch damage to be observed more common.
Various factors such as working environment, loading condition, material composition and mechanical properties should be considered to fully understand scratch behaviors.
Scratch resistance can be evaluated by visibility criterion or morphology analysis

8. 1. Introduction Objective
Understand the structure of Polypropylene (PP) using various material characterization methods and construct the relationship with scratch properties
Design of experiment (DOE)
Construct statistical estimations
Evaluate intrinsic properties of materials (DSC, GPC, Rheometer)
Measure mechanical properties
Compare the scratch properties to structural property parameters and draw major parameters affect mechanical properties
Construct statistical empirical equations for major structural parameters and scratch properties using multiple regression and neural network models

9. Methods Ⅱ

10. Experimental design process
2. Methods
Design of Experiment, DOE
Experimental design process
Ref)

11. 2. Methods Design of Experiment, DOE Principles of DOE
1. Randomization
Conduct tests in random order to exclude unwanted effects except experimental factors
2. Replication & Repetition
Repeat tests for multiple times to minimize R-square and increase reliability
3. Confounding (Fractional factorial design)
Confound less important interaction effect and block effect for efficient experiment
4. Blocking
Obtain uniform experimental environment by making blocks for entire tests
5. Orthogonality
Array factors orthogonally to obtain higher resolution and more reliable approximation in same runs

12. 2. Methods Multiple Regression Estimated (or predicted) value of Y
Regression model with 2 or more independent variables
Estimated
(or predicted)
value of Y
Estimated slope coefficients
intercept
Random Error
Evaluation of goodness-of-fit of regression model
- R2 : coefficient of determination, 0 ≤ R2 ≤ 1
Percentage of response variable variation that is explained by its relationship with one or more predictor variables
Evaluate how well the regression model explains the data
- Adjusted R2 , 0 ≤ R2 ≤ 1
R2 for any model will always increase when a new term is added
Adjusted R2 determines how well the model fits the data when adjusting for the number of predictors in the model
It is always same or less than R2

13. 2. Methods Neural networks
Mimicking biological activities of neurons in brain (interacting with each other and learn from experience)
High reliability
Complex relationship between input parameter and output parameter
Figure. Schematics of Neural network model and actual example of tensile properties estimation

14. 2. Methods Scratch test Designed to* ;
ASTM D7027 Standard Test Method for Evaluation of Scratch Resistance of Polymeric Coatings and Plastics Using an Instrumented Scratch Machine
Designed to* ;
Evaluate the scratch resistance of a particular material
Compare the relative scratch resistance of different materials
Determine the scratch coefficient of friction of materials
Test mode* ;
Applying increasing normal load from 2 to 50 N over a distance of 0.1 m at a constant scratch rate of 0.1 m/s
Applying constant normal load of 30 N over a distance of 0.1 m at a constant scratch rate of 0.1 m/s
*ASTM D , Standard Test Method for Evaluation of Scratch Resistance of Polymeric Coatings and Plastics Using an Instrumented Scratch Machine, ASTM International, West Conshohocken, PA, 2013,

15. 2. Methods
Evaluation of scratch resistances

16. Materials and ExperimentsⅢ

17. 3. Materials & Experiment
Material(resin) : PP copolymer (MFI 5, 27.5, 50)
DOE(using JMP)
2 factor 3 level : PP copolymer (MFI 5, 27.5, 50) + glass fiber(GF, 0, 25, 50%)
Sample
No.
PP co. MI
(%) GF
No.1 5
100
No.2 75 25
No.3 50
No.4
27.5
No.5
No.6
No.7

18. 3. Materials & Experiment
Tensile test condition (ISO527)
Sample No.
Run order
MFI
(g/10 min)
Glass fiber
(%)
Crosshead speed
(mm/min)
Repetitions
No. 1 4 5
5, 50, 500
No. 2 3 25
No. 3 7 50
No. 4 2
27.5
No. 5
No. 6 1
No. 7 6

19. 3. Materials & Experiment
DSC
Test machine : TA Q20
N2 condition
10℃/min(heating and cooling)
1st Heating (30℃ to 200℃)  5min isothermal  Cooling (200℃ to 30℃)  5min isothermal  2nd Heating (30℃ to 200℃)
GPC
Test machine : HT-GPC agilent PL-220
Rheology
Rheometer TA ARES-G2
Frequency Sweep at 210℃
Cross equation :

20. 3. Materials & Experiment
Modeling parameter
MFI
Talc content
Scratch speed
DSC
Heat of fusion
Crystallinity
GPC Mn Mw Mz PD
Rheology η0 η∞ K m

21. Figure. Schematics of scratch tester Kato tech. KK01 and scratch test.
3. Materials & Experiment
Scratch test
Parameter
Value
Scratch test
Scratch speed
100
mm/s
Scratch length mm
Initial normal load 1 N
Final normal load 40
Tip diameter
(unit : mm)
Figure. ISO 527 tensile test specimen.
Figure. Schematics of scratch tester Kato tech. KK01 and scratch test.

22. Results & Discussion Ⅳ

23. 4. Results & Discussion Tensile modulus – Regression model Speed-GF-Mw
Speed-GF-Mn
Speed-GF-Mw

24. 4. Results & Discussion Tensile strength – Regression model
Speed-GF-PD-K
Speed-GF-PD

25. 4. Results & Discussion Tensile strain – Regression model Speed-GF-MI
Speed-GF-MI-PD-K
Speed-GF-MI

26. 4. Results & Discussion Tensile toughness – Regression model
Speed-GF-MI
Speed-GF-Mw

27. 4. Results & Discussion Scratch critical loads – Regression model
1st critical load
2nd critical load
GF-Crystallinity-Mn
Mw-Heat of fusion

28. 4. Results & Discussion 1st critical loads – Neural network model
Speed-GF-DSC(Heat of fusion- Crystallinity

29. 4. Results & Discussion 1st critical loads – Neural network model
Speed-GF-GPC(Mn-Mw-Mz-PD)

30. 4. Results & Discussion 1st critical loads – Neural network model
Speed-GF-Rheology(η0, K, m)

31. 4. Results & Discussion 1st critical loads – Neural network model
Speed-GF- DSC-GPC-Rheology

32. 4. Results & Discussion 2nd critical loads – Neural network model
Speed-GF-DSC(Heat of fusion- Crystallinity

33. 4. Results & Discussion 2nd critical loads – Neural network model
Speed-GF-GPC(Mn-Mw-Mz-PD)

34. 4. Results & Discussion 2nd critical loads – Neural network model
Speed-GF-Rheology(η0, K, m)

35. 4. Results & Discussion 2nd critical loads – Neural network model
Speed-GF- DSC-GPC-Rheology

36. 4. Results & Discussion
1st critical load (Neural network model, R2= )
Speed-GF-GPC(Mn-Mw-Mz-PD)
( ) * :H1_ * :H1_ * :H1_3 2
H1_1
TanH(0.5 * ( * :GF * :Mn * :Mw * :Mz * :PD))
H1_2
TanH(0.5 * ( * :GF * :Mn * :Mw * :Mz * :PD))
H1_3
TanH(0.5 * (( ) * :GF * :Mn * :Mw * :Mz * :PD))
2nd critical load (Neural network model, R2= )
Speed-GF-Rheology(η0, K, m)
* :H1_ * :H1_ * :H1_3 3
H1_1
TanH(0.5 * (( ) * :GF e-10 * :Name("zero shear viscosity,η0") * :K * :m))
H1_2
TanH(0.5 * (( ) * :GF e-10 * :Name("zero shear viscosity,η0") * :K * :m))
H1_3
TanH(0.5 * ( * :GF * :Name("zero shear viscosity,η0") * :K * :m))

37. Conclusion Ⅴ

38. 5. Conclusion
Scratch resistances of PP were obtained from ASTM standard test
Critical loads and structural parameters (DSC, GPC, Rheometer) were statistically analyzed to evaluate correlation
Empirical equations on scratch resistances in terms of major structural parameters were constructed using statistical modeling (Multiple regression and neural network model )
- Modeling input factors were GF content, scratch speed, DSC (crystallinity, heat of fusion), GPC (Mn, Mw, Mz, PD) and Rheology (zero shear viscosity, K, m).
Based on R2, neural network models showed better reliability than regression models.
It can be understood in that neural network model has advantages in non-linear over regression model and shows better results.

39. Thank you for your attention!