CAN DATA ANALYSIS BE THE
ANSWER TO SPEED-UP YOUR
PYROLYSIS?
Aschwin van der Horst
[email protected]
Py-GC-FID/MS
• Py-GC-FID/MS of polymers reveal the monomer composition
of the polymers
 Time
U
Concentration 
Pyrolysis Gaschromatography
U
Py-GC-FID/MS
• Py-GC-FID/MS of polymers reveal the monomer composition
of the polymers
Ferromagnetic
Curie-Point
Paramagnetic
Curie point analysis
Ferromagnetic
Curie-Point
Paramagnetic
Pierre Curie studied ferromagnetism and paramagnetism for his doctoral thesis, and discovered the effect of
temperature on paramagnetism which is now known as Curie's law. The material constant in Curie's law is known as
the Curie constant. He also discovered that ferromagnetic substances exhibited a critical temperature transition,
above which the substances lost their ferromagnetic behavior. This is now known as the Curie point.
Curie point analysis
Ferromagnetic
Curie-Point
Paramagnetic
 the magnetic susceptibility; the influence of an applied magnetic
field on a material
M the magnetic moments per unit volume
H the macroscopic magnetic field
B the magnetic field
C the material-specific Curie constant
µ0 the permeability of free space
g the Landé g-factor
J(J+1) the eigenvalue for eigenstate J2 for the stationary states
within the incomplete atoms shells (electrons unpaired)
µB the Bohr Magneton (Planck)
kB Boltzmann's constant
Synthesis of 37 homopolymers
Solvent
Monomer
AMBN
Optimal Curie point determination
Multivariate data analysis
In only very few cases pyrolysis GC can be evaluated by
isolated chromatographic peaks or signals
 univariate method (Linear Regression)
In most cases the information is contained in larger
chromatographic peak regions and has to be extracted by
 multivariate methods (Multiple Linear Regression)
Calibration  Validation  Prediction
reference analysis
calibration samples
100% Homopolymer
pyrolysis
pyrolysis
unknown samples
Copolymers
determine parameters
of the calibration
model
pyrolysis
reference analysis
validation samples
100% Homopolymer
100% 100%
100% 100%
M
calibration model
100% 100%
100% 100%
validate
calibration model
M
M
prediction with
calibration model
44 %
38 %
28 %
22 %
...
46 %
37 %
26 %
25 %
...
Building chromatographic data Matrix (M)
Validation of Pyrolysis Model (M)
M-1
Homopolymer
Scores
u
%
Residuals
R
Validation Py-Model after Zeroing (M0)
Contribution of Styrene = 90%
Contribution of t-ButylAcrylate = 15%
Validation Py-Model after Zeroing (M0)
Contribution of n-BMA = 40%
Contribution of OctadecylAcrylate = 40%
Validation of Pyrolysis Model after Zeroing (M0)
•
•
•
Validation of the pyrolysis Model after zeroing (M0) shows that the Loadings have about
100% Scores in the prepared model. Furthermore, the validation samples have better
fit compared to the non-zeroing Model (M), therefore M0 is used for validation.
For Prediction the Calibration Model M0 is used.
Copolymers are made by:
– Making a calculated mix (t) of 25/75, 50/50 and 75/25 Polymer A/B and the composition (St) is
calculated with Model M0

M 01  t  St
– Copolymer standards with known composition are pyrolysed (p) and the composition (Sp) is
calculated with Model M0

M 01  p  S p
Theoretical Prediction of iBOA-iBOMA with Model M0
No dissimilarity is found for iBOA-iBOMA copolymers
Model will NOT work when both monomers are present

M 01  t  St
S
MMA
S
100
0
MMA
0
100
MMA25S75
69
31
MMA50S50
42
58
Theoretical Prediction of S-MMA with Model M
Styrene = 96%
MMA= 85%
Model works well for theoretical copolymers of S-MMA because
of the presence of enough dissimilarity between Styrene and MMA.

M 01  t  St
MMA75S25
20
80
0
S
S
MMA
MS A1
MS A2
MS 100
100
0
70
77
81
Determination
of Synthesized
S-MMA
with
0
100
30
23 Model M
19 0
MMA
Styrene = 96%
MMA= 85%

M 01  p  S p
Why is M0 not working well?
•
•
•
•
•
Focus is on individual monomer
peaks in chromatography.
Every homo-polymer has different
optimal Curie point resulting in
dimers, trimers, etc, but also
fragmentation of the monomer in
pyrolysis.
Layer thickness and concentration
of the polymer.
Breakdown of MMA, BMA, iBMA,
EHA, etc.
Not all acrylic monomers can be
determined by Py-GC-FID/MS, e.g.
AA and MAA.
Alternative Py-LC-PDA Analysis
A. van der Horst et al., J. Chem. Chem. Eng. 8, 2014
A. van der Horst et al. LCGC-Europe. 9, 2016
F. Cheng-Yu Wang et al., Anal. Chem, 1995
Off-Line Py-UPLC-PDA Analysis
A. van der Horst et al., J. Chem. Chem. Eng. 8, 2014
A. van der Horst et al. LCGC-Europe. 9, 2016
Online Py-HPLC-PDA Analysis
A. van der Horst et al., J. Chem. Chem. Eng. 8, 2014
A. van der Horst et al. LCGC-Europe. 9, 2016
HEMA-HEA Py-LC-PDA Quantification
40
35
30
%
25
20
VHEA
15
VHEM
10
5
0
100/0
80/20
60/40 50/50 40/60
VHEM / VMEA (%)
20/80
0/100
A. van der Horst et al., J. Chem. Chem. Eng. 8, 2014
A. van der Horst et al. LCGC-Europe. 9, 2016
Conclusion
• Data-analysis can be used to determine the monomer ratio in polymers if:
–
–
–
–
–
–
Enough dissimilarities
No fragmentation of monomers occur
Model polymers are used
Monomer is volatile enough for GC-analysis
Sample repeatability is good
Standards can be made to calibrate the model
• Not all acrylic monomers can be determined by GC-FID, acid and hydroxyl
functional monomers can better be determined by Py-(U)HPLC-PDA which
is a very good alternative to perform comprehensive composition analysis
of all acrylic monomers in a polymer
Acknowledgements
•
•
Staff:
Karen Huiskes
Jacqueline Slaakweg
Afke Kroes
Peter van den Berg
Stephan Ploegaert
Ton Cleiren
Rutger van Geel
Sven Dissel
Jason Nefs
Students:
Stephan Tromper
Ramon van de Bilt
Metine van Dam
Jenny Kools
Tammy Zandvliet
Eldert Valk
Yannick van Hooijdonk
Robbie Westerneng
Kevin Faro
Antal Biesheuvel
Luc van ‘t Hof
disclaimer
Notice: Trademarks indicated with the ®, ™ or * are registered, unregistered or pending trademarks of Allnex Belgium SA or its directly or indirectly affiliated
Allnex Group companies.
Disclaimer: Allnex Group companies (“Allnex”) decline any liability with respect to the use made by anyone of the information contained herein. The
information contained herein represents Allnex's best knowledge thereon without constituting any express or implied guarantee or warranty of any kind
(including, but not limited to, regarding the accuracy, the completeness or relevance of the data set out herein). Nothing contained herein shall be construed
as conferring any license or right under any patent or other intellectual property rights of Allnex or of any third party. The information relating to the products
is given for information purposes only. No guarantee or warranty is provided that the product and/or information is adapted for any specific use, performance
or result and that product and/or information do not infringe any Allnex and/or third party intellectual property rights. The user should perform its own tests
to determine the suitability for a particular purpose. The final choice of use of a product and/or information as well as the investigation of any possible
violation of intellectual property rights of Allnex and/or third parties remains the sole responsibility of the user.
© 2016 Allnex Belgium SA. All Rights Reserved