Beyond the EU: DORA and NIS 2 Directive's Global Impact
SLOPE 3rd workshop - presentation 1
1. General presentation of NIR technology
in aspect of wood
by Jakub & Anna Sandak
CNR-IVALSA
NIR & wood: sounds good!
2° workshop
in memoriam of Federico Prandi
3. 3
Herschel directed sunlight through a glass
prism to create a spectrum and measured the
temperature of each color
the temperature of the colors increased from
the violet to the red part of the spectrum
after further studying, he concluded that there
must be an invisible form of light beyond the
visible spectrum
NIR spectroscopy
The history of NIR begins in 1800 with Frederick William Herschel
it was the first time that someone showed that there
were forms of light that we cannot see with our eyes
5. 5
Principle of molecular vibrations
• Spectroscopy provides information
about the vibrations of functional
groups in a molecule
• Every molecule has specific vibration
frequencies
• When polar molecule is exposed to
infrared light its starts vibrate since
certain frequency is absorbed
• Consequently a change in dipol
moment of the molecule occurs
8. 8
Band assignment - wood
code wavenumber
(cm
-1
)
band assignment
1 4198 CH deformation in holocellulose
2 4280 CH stretching + CH deformation in semi- and crystalline region in cellulose
3 4404 CH2 stretching + CH2 deformation of cellulose
4 4620 OH stretching + CH deformation of cellulose
5 4890 OH stretching + CH deformation of cellulose
6 5219 OH stretching + OH deformation of water
7 5464 OH stretching + CH stretching semi- or crystalline regions of cellulose
8 5587 CH stretching semi- or crystalline regions of cellulose
9 5800 CH stretching in furanose/pyranose due to hemicelluloses
10 5883 CH stretching in aliphatic chains
11 5935 CH stretching of aromatic skeletal in lignin
12 5980 CH stretching of aromatic skeletal in lignin
13 6287 OH stretching in crystalline region in cellulose
14 6450 OH stretching in crystalline region in cellulose
15 6722 OH stretching in semi-crystalline region in cellulose
16 6785 OH stretching in semi-crystalline region of cellulose
17 7008 OH stretching in amorphous region in cellulose
18 7309 CH stretching in aliphatic chains
19 7418 CH stretching in aliphatic chains
around 100 bands assigned to wood components (Schwanninger et al. 2011)
9. 9
NIR technique
No need special sample preparation
Non-destructive testing
Relatively fast measurement
No residues/solvents to waste
Possibility for determination of many
components simultaneously
High degree of precision and
accuracy
Direct measurement with very low
cost
Needs sophisticated statistics
methods for data analysis
Overlapping of spectral peaks
10. 10
NIR & wood: scale
Inspired by artwork by Mark Harrington (http://www.nzwood.co.nz)Inspired by artwork by Mark Harrington (http://www.nzwood.co.nz)
14. 14
Measuring NIR on wood:
CNR-IVALSA experiences
9. sample presentation
10. measurement
12. storage of results
5. sample conditioning 8. optimization of the set-up
13. exploratory analysis
15. optimal data set of spectra
14. spectra pre-processing
11. reference characterizations
15. 15
Measuring NIR on wood:
CNR-IVALSA experiences
16. calibration: qualitative analysis
17. validation of qualitative model
21. implementation of chemometric
models for identification
15. optimal data set of spectra
18. calibration: quantitative analysis
19. validation of quantitative model
22. implementation of chemometric
models for prediction
20. interpretation of spectra
23. generalization of knowledge
16. 16
What can NIR measure on wood?
absorbance + scatter
(not lignin, cellulose, density…)
17. 17
wood characteristics
influencing NIR measurement
• sample state: solid, chipped, milled …
• sample age: surface deactivation, oxidation and aging …
• material properties: density, porosity, moisture,
roughness …
• degradation due to biotic agents: fungal decay,
damage by insects, bacteria …
• degradation due to a-biotic agents: waterlogging,
weathering, ageing, chemical degradation …
• presence, position, quantity of wood defects: knots,
checks, slope of grain …
• mechanical damages and geometrical alterations:
cracks, delaminations, distortions, deformations …
19. 19
Second derivative
-0,00007
0
0,00007
40004500500055006000650070007500800085009000950010000
60°C 80°C 100°C 120°C 140°C 160°C 180°C 200°C
a
b
-0,0000125
0
0,0000125
550056005700580059006000610062006300640065006600670068006900700071007200730074007500
drugapochodnaabsorbancji
liczba falowa (cm
-1
)
a
-0,00007
0
0,00007
410042004300440045004600470048004900500051005200530054005500
liczba falowa (cm
-1
)
liczba falowa (cm
-1
)
b
20. 20
NIR: state-of-the-art instrumentation
• Fourier transform near infrared spectroscopy
• Dispersive fixed grating diode array (DA)
• NIR spectrometer with linear variable filter (LVF)
• Scanning grating or dispersive monochromator
(DM)
• Acoustic Optical Tunable Filters (AOTF)
• Micro-Electro-Mechanical-Systems (MEMS)
• Multispectral and hyperspectral cameras
• ???
21. 21
What NIR instrument for wood?
FT-NIR DA LVF DM AOTF MEMS cameras
Spectral range full limited limited full limited limited full
Scanning time (s) slow very fast very fast fast very fast very fast fast
resolution very high high limited high limited limited high
cost high middle low middle middle middle high
Signal/noise high limited limited high limited limited high
Calibrations transfer very good good good very good good limited limited
Shock resistance no yes yes no yes yes yes
31. but…
Data is not the same as information
Too much data - too little information
(Harald Martens)
32. 32
Chemometrics
the chemical discipline that uses
mathematical and statistical methods to
design or select optimal procedures and
experiments, and to provide maximum
chemical information by analyzing
chemical data
definition of the Chemometrics Society
33. 33
Multivariate data analysis
• Exploratory data analysis (data mining) – attempts to find
the hidden structure in large complex data sets
– Cluster analysis
– Principal Component Analysis
• Regression analysis and Predictive Models (developing
the models from available data and predict desired
response)
– Partial Least Squares Regression
– Multiplicative Linear Regression
• Classification Models (separation of group of object into
one or more classes based on distinguished characteristic)
– Cluster Analysis Test
– Identity Test
– SIMCA
34. 34
Cluster analysis: CA
analyze spectra for their similarity, divide most similar spectra into groups, which are
called clusters or classes
The spectral distance – heterogeneity explains the similarity between the spectra
The spectral distance 0 means that the spectra are identical
heterogeneity0
Sample1
Sample2
Sample3
Sample4
36. 36
Principal Component Analysis
It is used for de-correlation of highly correlated data to reduce multidimensional
data set to lower dimensions
it can separate set of input data into groups of peculiar similarities
x
y
PC1
PC2
37. 37
PCA example
PCA of NIR spectra of non
coated samples exposed to
South (4 years of natural
weathering);
Reggio Emilia
Roma
Loninghens
Macerata
Udine Trento
Lecce standard
a
38. 38
Identity test
Compare the unknown spectrum with all reference spectra
The result of comparison between two spectra is the spectral distance called hit quality
The better spectra match the smaller is spectral distance; HQ for identical spectra is 0
Model sample1
HQ1
> treshold1
Model sample3
HQ3
> treshold3
Model samplen
HQn
> tresholdn
Model sample2
HQ2
< treshold2
???
sample
39. 39
Ita
Pol S” Est
Fin
Pol S’
Pol N
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
Finlandia
Finlandia
Finlandia
Finlandia
Finlandia
Finlandia
Estonia
Estonia
Estonia
Estonia
Estonia
Estonia
PolskaN
PolskaN
PolskaN
PolskaN
PolskaN
PolskaN
PolskaS'
PolskaS'
PolskaS'
PolskaS''
PolskaS''
PolskaS''
Włochy
Włochy
Włochy
Włochy
Włochy
Włochy
dystansspektralny(HQ),próg(T)threshold(T),spectraldistance(HQ)
Finland
Finland
Finland
Finland
Finland
Finland
Estonia
Estonia
Estonia
Estonia
Estonia
Estonia
PolandS’
PolandS’
PolandS’
Italy
Italy
Italy
Italy
Italy
Italy
PolandS”
PolandS”
PolandS”
PolandN
PolandN
PolandN
PolandN
PolandN
PolandN
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
Finlandia
Finlandia
Finlandia
Finlandia
Finlandia
Finlandia
Estonia
Estonia
Estonia
Estonia
Estonia
Estonia
PolskaN
PolskaN
PolskaN
PolskaN
PolskaN
PolskaN
PolskaS'
PolskaS'
PolskaS'
PolskaS''
PolskaS''
PolskaS''
Włochy
Włochy
Włochy
Włochy
Włochy
Włochy
dystansspektralny(HQ),próg(T)
próg (T)
threshold(T),spectraldistance(HQ)
Finland
Finland
Finland
Finland
Finland
Finland
Estonia
Estonia
Estonia
Estonia
Estonia
Estonia
PolandS’
PolandS’
PolandS’
Italy
Italy
Italy
Italy
Italy
Italy
PolandS”
PolandS”
PolandS”
PolandN
PolandN
PolandN
PolandN
PolandN
PolandN
threshold (T)
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
Finlandia
Finlandia
Finlandia
Finlandia
Finlandia
Finlandia
Estonia
Estonia
Estonia
Estonia
Estonia
Estonia
PolskaN
PolskaN
PolskaN
PolskaN
PolskaN
PolskaN
PolskaS'
PolskaS'
PolskaS'
PolskaS''
PolskaS''
PolskaS''
Włochy
Włochy
Włochy
Włochy
Włochy
Włochy
dystansspektralny(HQ),próg(T)
próg (T)
dystans spektralny (HQ)
threshold (T)
spectral distance (HQ)threshold(T),spectraldistance(HQ)
Finland
Finland
Finland
Finland
Finland
Finland
Estonia
Estonia
Estonia
Estonia
Estonia
Estonia
PolandS’
PolandS’
PolandS’
Italy
Italy
Italy
Italy
Italy
Italy
PolandS”
PolandS”
PolandS”
PolandN
PolandN
PolandN
PolandN
PolandN
PolandN
Identity test (IT)
42. 42
• Spectroscopy is the study of the interaction between matter and
electromagnetic radiation
• Spectroscopic data is often represented by a spectrum, a plot of
the response of interest as a function of wavelength or frequency
• Absorption bands in the NIR are the result of combination and
overtone bands from the fundamental vibrations seen in the mid-IR
• The overtone and combination bands are 10 – 100 X less intense
than the fundamental bands in mid-IR, but are better fitting for
interpretation of wood
• Differences in NIR spectra are often very subtle, often overlapping
bands are present; difficult to interpret requiring training of
analysts to recognize these differences
Summarizing…
43. 43
+ to be continued on NIR&wood 2°
2. experimental design
1.problem definition
9. sample presentation
10. measurement
12. storage of results
16. calibration: qualitative analysis
17. validation of qualitative model
21. implementation of chemometric
models for identification
3. sample selection
4. sample preparation
5. sample conditioning
6. istrument selection
7. instrument set-up
8. optimization of the set-up
13. exploratory analysis
15. optimal data set of spectra
14. spectra pre-processing
18. calibration: quantitative analysis
19. validation of quantitative model
22. implementation of chemometric
models for prediction
20. interpretation of spectra
23. generalization of knowledge
11. reference characterizations