In-line Analysis of EVA Copolymers using Vibrational Spectroscopy

1
In-line Analysis of EVA Copolymers using
Vibrational Spectroscopy
S.E.Barnes, M.G.Sibley, H.G.M.Edwards,
I.J.Scowen and P.D.Coates IRC in Polymer
Science Technology, School of Engineering,
Dept. of Chemical and Forensic Sciences,
University of Bradford
2
Outline

• Context
• In-line Instrumentation
• Aims and objectives of current research
• Results
• Data manipulation
• Conclusions
• Future work
• Acknowledgements

3
Context
High level of interest into on-line / in-line
analysis of polymer processing
increased demands on quality of polymeric
materials / products Immediate detection of
problems during processing optimises operating
conditions saves energy reduces waste reduces
need for off-line testing
4
Process spectroscopy

• Enhanced monitoring
• molecular specificity
• towards real time
• non-invasive
• non-destructive

• Multi-probe measurements
• NIR ultrasound
• Raman ultrasound
• Raman NIR

• Process analysis
• blends
• additives
• decomposition
• reactive extrusion

5
Applications

• Continuous monitoring for
• polymer characterisation
• analysis of polymer blends / co-polymer
  composition
• additive identification / quantification
• shear induced orientation studies (polarised
  Raman)
• degradation / process induced change
• reactive extrusion

6
In-line techniques
7
In-line Transmission NIR

• fibre-optic transmission probes - samples across
  melt section
• variable path length from 1 to 11 mm
• wavenumber range 8500 - 4000 cm -1 (1200-2500
  nm)
• resolution 2-16 cm -1

8
Raman Spectroscopy
Hololab RXN-2 analyser, Kaiser Optical Systems
300 mW , 785 nm laser CCD detector 3 channels
5 cm-1 resolution
In-line fibre-optic probe with sapphire
window focal distance - 2.5 mm - point
measurement Dynisco type fitting
9
Current Research
Extrusion of a series of ethylene vinyl acetate
(EVA) random co-polymers Comparison of the
response and sensitivity of in-line Raman and NIR
as well as ultrasound to alterations in Vinyl
acetate (VA) concentration
10
Experimental setup

• Materials
• EVA copolymers with varying VA content between 2
  44 wt
• VA content of each copolymer determined by TGA
  analysis

Sample VA content (wt )
1 2
2 7.3
3 9
4 17.1
5 27.8
6 34.2
7 43.1
11
Experimental setup

• Process parameters
• Betol BC38 single screw extruder
• screw speed 15 rpm
• extrusion temperature 180C
• Instrument set-up
• NIR - 3 mm path length, 4 cm-1 resolution, 8000
  to 4000 cm-1
• Raman 1 spectrum per minute 28 second
  exposure,1 accumulation
• Real-time monitoring during extrusion
• NIR relative intensity of IR active spectral
  features
• Raman integrated area of Raman active spectral
  features
• ultrasound transit time
• Melt temperature and pressure measured

12
Sensor arrangement
Single screw extruder with Raman, NIR,
ultrasound, P and T sensors
13
Raman results
In-line Raman spectra of PEVA (3010 - 450 cm-1)
Spectral region 1800 to 550 cm-1 chosen for
multivariate analysis
14
Raman results
Change in Raman spectrum (2000 cm-1 500 cm -1)
during extrusion of EVA copolymer samples (2 -
43.1 wt VA)
15
Raman results
Change in integrated peak area of O-CO def.
feature at 629 cm-1 during extrusion
16
NIR results
NIR absorbance spectra showing first overtone and
combination band region of EVA copolymer melts
Spectral region 6200 to 5080 cm-1 chosen for
multivariate analysis
17
NIR results
C-H stretch of VA
3D plot showing alterations in NIR spectral
features during extrusion of EVA copolymers
18
Ultrasound Results
Alteration in Ultrasonic transit time and melt
pressure during sequential extrusion of EVA
copolymers
Melt pressure in the die varied throughout the
extrusion process, due to variation in viscosity
(MI values) Alteration in transit time is not
linear with pressure variation
19
Ultrasound vs Raman
Response of transit time and integrated area of
feature at 629cm-1 to alteration in VA content
20
Multivariate analysis Grams PLSIQ

• multivariate techniques
• partial least squares regression (PLS)
• principle component analysis (PCR)
• model spectral variation in a calibration data
  set
• calibration for all constituents in
  multi-component systems
• whole spectrum / selected spectral areas used
  for calibration
• no pre-treatment of spectra is necessary

21
Data manipulation

• 15 spectra of each copolymer in PLS-IQ
  calibration
• first derivatives of the spectra to eliminate
• NIR baseline shifts
• Raman fluorescence and background noise
• data mean centred to enhance subtle differences
  between spectra
• specific regions of the spectra were chosen for
  the model
• NIR - 6030-5080 cm-1
• Raman - 1800 to 550 cm-1
• PLS1 used to produce a calibration

22
Raman Results
PLS1 results showing comparison of predicted
against actual VA percentage (one factor model)
Standard error of calibration /- 0.56 VA
(1s) R2 0.99
23
Raman Results
Predicted against actual VA percentage for
independent spectral data set
Standard error of prediction /- 0.67 VA R2
0.991
24
Raman Results
Sample True wt VA Predicted wt VA Error
1 2 2.2 9.19
2 7.3 6.87 5.9
3 9 8.9 1.11
4 17.1 17.06 0.23
5 27.8 28.02 0.79
6 34.2 34.0 0.58
7 43.1 42.72 0.88
True and predicted values for VA content in EVA
copolymers
25
NIR Results
PLS results showing comparison of predicted
against actual VA percentage (one factor model)
Standard error of calibration /-0.604 VA
(1s) R2 0.9981
26
NIR Results
Predicted against actual VA percentage for
independent spectral data set
Standard error of prediction /-0.631 VA R2
0.998
27
NIR Results
Sample True wt VA Predicted wt VA Error
1 2 1.92 3.88
2 7.3 7.39 1.27
3 9 8.96 0.43
4 17.1 17.24 0.84
5 27.8 27.81 0.03
6 34.2 34.74 1.58
7 43.1 42.64 1.07
True and predicted values for VA content in EVA
copolymers
28
Ultrasound
Change in ultrasonic transit time with VA
content comparison of three sets of experimental
data

• Extruder operated at 15 rpm during Tests 1 and 3
  10 rpm during Test 2
• Ultrasonic transducers removed and repositioned
  after test 1 difference transit time
  between data in test 1 and tests 2 and 3

29
Conclusions

• In-line prediction of wt VA content has been
  successfully conducted using in-line Raman NIR
  and Ultrasound
• PLS analysis has been applied to Raman and NIR
  data to build multivariate successful calibration
  models.
• A successful PLS model for the Raman region 1800
  550 cm-1 was produced with an SEP value of 0.67
  wt VA (one principal component).
• The NIR region 6030-5080cm-1 was used to
  construct a PLS model with an SEP of 0.63 wt VA
  for a one-factor model.

30
Conclusions

• Ultrasonic transit time is dependent upon melt
  density and bulk modulus, which change with wt
  VA.
• Ultrasound is highly sensitive to changes in VA
  content of the copolymer resins. The
  repeatability of the ultrasonic data is shown to
  be excellent.

31
Future work Raman / NIR

• further analysis of EVA copolymers using various
  NIR path lengths
• in-line Raman and NIR to evaluate MI during
  polymer extrusion
• qualitative and quantitative analysis of polymer
  additives
• Raman and NIR to monitor reactive extrusion
  processes
• melt orientation studies polarised Raman

32
Future workFluorescence

• implementation of new in-line, variable focus
  fluorescence probe
• fluorescence for the quantitative analysis of
  polymer additives.
• melt temperature measurement using fluorescent
  probes
• temperature dependant dyes

Lens tube and fibre optics inserted into outer
casing of in-line fluorescence probe
33
Acknowledgements

• Dr Elaine Brown / IRC colleagues
• IRC and EPSRC for support of research
  studentships
• AT Plastics inc. for the kind donation of the
  copolymers
• RAPRA for TGA analysis
• CPACT for the opportunity to present