Photoacoustic Spectroscopy (PAS) Technology Overview

Introduction:The unique aspect of PAS technology, direct detection oflight absorption by the sample, allows quick andnondestructive measurement of absorbance spectrawithout sample preparation or contact, and alsoadvanced measurements such as depth probing to reveallayered or gradient composition.

The basics of how PAS works and some of its applicationsare discussed in this overview.

1

Topics• Early History, pages 3-4.

• Direct measurement of Absorbance, page 5.

• The FTIR Connection, pages 6-7.

• PAS Signal Generation, pages 8-11.

• PAS Sampling Depth and Applications, pages 12-20.

• Spectral Subtraction of Depth Varying Spectra, pages 21-22.

• Phase Analysis with Depth Analysis Applications, pages 22-30.

• Chemometric Analysis of Photoacoustic Spectra

• Process Study Applications

• Reference Samples for PAS, page 56.

• Wood Chemistry, pages 31-36. • Polymer Chemistry, pages 37-39.

• Aging of Composites, pages 40-47. • Wood to Biochar Conversion, pages 48-55.

2

AG Bell’s Spectrophonefrom 1881 marks the beginning of photoacoustic spectroscopy

from A. G. Bell, Philosophical

Magazine, 1881, Ser. 5, 11, 510-528.

3

Bell also sponsored Michelson’s development of the

interferometer that lead to FTIR spectrometers which have

played a crucial role in PAS.

Heidelberg, Baden, Germany

April 17th, 1881

My dear Mr. Bell,

The experiments concerning the relative motion of the earth with respect to the ether have just been

brought to a successful termination. The result was however negative. . . .

At this season of the year the supposed motion of the solar system coincides approximately with the

motion of the earth around the sun, so that the effect to be observed was at its maximum, and accordingly if the

ether were at rest, the motion of the earth through it should produce a displacement of the interference fringes, of

at least one tenth the distance between the fringes; a quantity easily measurable. The actual displacement was

about one one hundredth, and this, assignable to the errors of experiment.

Thus the question is solved in the negative, showing that the ether in the vicinity of the earth is moving

with the earth; a result in direct variance with the generally received theory of aberration. . . .

N.B. Thanks for your pamphlet on the photophone,

Very truly yours,

Albert A. Michelson

Master, U.S. Navy

4

How does PAS relate to other sampling

techniques?

Thermal Emission

Sample

IRBeam

PhotoacousticResponse

ReflectedTransmitted

Absorbed

Only PAS measures the actual absorption of light in a sample bysensing the absorption driven heating and thermal expansionthat results in a photoacoustic pressure signal. 5

FT-IR PAS Instrumentation

6

FTIR-PAS Spectrometer SchematicFixed Mirror

Source

Beamsplitter

Sample Compartment

Detector

Laser Detector

Las

er

Moving

Mirror

Infrared

00

Retardation

Lig

ht

Inte

nsi

ty

PA Detector

7

PAS Signal Generation

Heat Out

Mic

rop

ho

ne

Sample

L1/

Modulated Infrared In

Gas

= absorption coefficient L = thermal diffusion length

Sound Waves

Absorption of intensity-modulated lightwithin the sample results in temperatureoscillations and generation of thermalwaves resulting in thermal expansion drivenpressure oscillations in the gas above thesample. A microphone senses theseoscillations as the PAS signal. L and 1/α arethe decay lengths for thermal and lightwaves propagating within the sample.

The direct detection of light absorption inthe sample by the heat produced andsubsequent acoustic signal results in theabsence of interference effects often seenin transmission and reflection spectra ofthin films. The spectra plotted abovedemonstrate this where the photoacousticspectrum has been converted to atransmission spectrum for comparison.

8

Thermal Only

Thermal and Acoustic

Thermal and Acoustic

with Background

Full

Saturation

Onset of

Saturation

0.0001 0.001 0.01 0.1 1 10 100

1E-8

1E-7

1E-6

1E-5

1E-4

1E-3

Absorption Coefficient (1/cm)

Thousands

Signal Magnitude

Absorption Coefficient (cm )-1

1000001000010001001010.1

Sig

nal

Mag

nit

ud

e

(lo

gar

ith

mic

sca

le)

As the absorption coefficient increases,the amount of heat deposited withinthe layer defined by L increases causingthe photoacoustic signal to increasewith α thus allowing absorption spectrato be measured. At high values of α, allof the heat is deposited just below thesurface of the sample resulting in signalsaturation and truncation of strongabsorption peaks.

PAS Signal Generation

These spectra illustrate the truncation ofstrong absorbance bands and how it isreduced when the modulation frequenciesare increased. The top low frequencyspectrum and the bottom high frequencyspectrum have been scaled so that the weakunsaturated peak heights near 2200 cm-1

are equal. The stronger bands of the topspectrum are all truncated and even whenthe top spectrum is rescaled for the middlespectrum truncation is present. 9

At very high values of α, the PAS response isdelayed in phase by 90° relative to the IRbeam waveform whereas at very low α thewaveform is delayed by a total of 135° dueto the longer propagation time required forthermal waves generated deeper within thesample.

The PAS Signal Has Magnitude and Phase Components

0.001 0.01 0.1 1 10 100 1000

0.001

0.01

0.1

1

Sig

na

l M

ag

nit

ud

e (

arb

. u

nit

s)

90

100

110

120

130

Sig

na

l P

ha

se L

ag

(d

eg

ree

s)

Absorption Coefficient X Thermal Diffusion Length

SaturationOnset

This plot shows how the PAS signalmagnitude and phase vary with a sample’sabsorption coefficient, α. Note that thephase reaches full saturation at higherabsorption values than the magnitude.Both of these components can becombined using a theoretical PAS signalgeneration model to produce “linearized”spectra that move the onset of saturationto higher values of α (seehttp://tiny.cc/pqgt9).

FullSaturation

10

Phase delay

Photoacoustic Signal Linearization

Sample Thermal Diffusivity = 0.001 cm2/s

(lo

gar

ith

mic

sca

le)

Sig

nal

Mag

nit

ud

e

1 10 100 1000 10 4 10 5 10 6

Absorption Coefficient (cm )-1

Linearized

Non-Linearized

Full Saturation

Linearization Range

10 Hz

100 Hz

10000 Hz

1000 Hz

The phase of the photoacoustic signal saturates at higher values of the absorptioncoefficient as shown in the last figure and this allows calculation of a “linearized”spectrum that has a linear dependence extending to higher values of theabsorption coefficient. This reduces the truncation of strong absorbance peakswhich are troubling to classical spectroscopists but turn out to actually have littleinfluence when modern chemometrics are use in both quantitative and qualitativeanalyses. Linearization, however, is very useful in reducing sampling depth andincreasing surface specificity as illustrated later in the case of a siloxane treatedfabric.

11

PAS Sampling Depth

• With rapid scan FTIRs sampling depth, L, depends on the interferometer’sOPD mirror velocity, v, the sample’s thermal diffusivity, D, and the

wavenumber, ѵ, at a particular point in a spectrum.

• L is defined as the reciprocal of the thermal wave decay coefficient, as=(πf/D)1/2 because thermal oscillations in the gas, where the PA signal isgenerated, are not efficiently driven, due to thermal wave decay, from deeper

than L =(D/πv ѵ)1/2, where the IR beam modulation frequency, f = vѵ.

• With step-scan FTIRs sampling depth is constant across the spectrum andcontrolled by the phase modulation frequency, fm, and given by L=(D/π fm)1/2.

• L only defines the sampling depth if the sample’s absorption coefficient, α, inthe spectral region of interest, is low enough to allow light to penetratedeeper than L. If this is not the case, the sampling depth is determined by thedecay of light waves rather than of thermal waves and is given by 1/α .

12

0.1 3001 10 100

0.001

25

0.01

0.1

1

10

Re

tard

ati

on

Ve

loc

ity

(c

m/s

)

15.8

400000

25

50

100

200

400

800

2500

5000

10000

20000

40000

80000

200000

La

ser-

Fri

ng

e F

req

ue

nc

y (

Hz

)

Thermal Diffusion Length (µm)

Based on 0.001 cm²/s

thermal diffusivity.

Phase Modulation Frequency (Hz)

400200100 50 25 10 51k2k

PAS Sampling Depth

13

An example of the importance PAS sampling

depth in the case of a 0.57 mm thick

polypropylene sample

• A 0.57 mm sample is too thick for a transmission measurement.

• PAS samples a thickness less than the sample’s physical thickness and produces excellent spectra without sample preparation.

• This is the a key advantage of PAS over transmission spectroscopy.

14

Polypropylene Sample 0.57 mm Thick

Transmission

Photoacoustic

Absorbance

10002000300040005000

Wavenumber

-80

-60

-40

-20

0

20

40

60

80

Abso

rban

ceP

erce

nt

Tra

nsm

issi

on

15

How Do PAS and Transmission Spectra Compare When Sampling Depths Are Equal?Photoacoustic versus Transmission Acquisition

Formvar Membrane (approximately 50 nm)

50010001500200025003000350040000

0.01

0.02

0.03

0.04

0.05

Wavenumbers (cm )

Abso

rbance U

nit

s

Photoacoustic

Transmission(converted to absorbance)

-1

Thse Formvar polymerthin film spectra measured bytransmission and photoacousticspectroscopies show that thetwo spectra are identical whenthe sample is very thin, which inthis case was approximately 30nm. Sampling depths areessentially equal with very thinfilms.

16

Varying Sampling Depth

to Observe Surface Layers

• Example of surface treatment of a fabric with Siloxane.

• At what concentration does a layer form on the fabric surface?

• Varying the sampling depth of the PAS spectra allows the layer and its depth variation to be observed.

• Note how linearization enhances surface specificity.

17

750800850900950

Wavenumber

Pho

toac

oust

ic S

igna

l 40 kHz Linearized

40 kHz

5 kHz

400 Hz

2%

2%

2%

2%

0%

0%

0%

0%

Surface Layer Present at 2% ConcentrationPA Spectra Untreated and 2% Siloxane-Treated Fabric

Sampling

Depths:

1 micron

4 microns

10 microns

30 microns

Siloxane Surface Layer Signature

Linearization of the PASsignal allows thesiloxane surface layer tobe observed and one tosee that it is primarilyconcentrated in a layerapproximately onemicrometer in thickness.

18

Surface Layer Not Present at 0.2% Concentration

750800850900950

Wavenumber

Pho

toac

oust

ic S

igna

l 40 kHz Linearized

40 kHz

5 kHz

400 Hz

0.2%

0.2%

0.2%

0.2%

0%

0%

0%

0%

PA Spectra Untreated and 0.2% Siloxane-Treated Fabric

1 micron

Sampling

Depths:

4 microns

10

microns

30

microns

19

Polyethylene Pellets with a Surface Additive

500100015002000250030003500

Wavenumber

Linearized

Standard

Surface segregatingadditive peak

20

Bulk additivepeak

Linerarization can also be used to distinguish between surface segregating andbulk additive peaks. The deeper sampling blue spectrum’s surface segregatingpeak is smaller than the bulk additive peak but the reverse is true for theshallower sampling red linearized spectrum. Its greater surface specificityincreases the height of the surface segregating peak.

Spectral Subtraction of Depth Varying Spectra

• Spectra of thin surface layers of samples can be obtained byspectral subtraction of a deeper (lower mirror velocity) from ashallower (higher mirror velocity) spectrum.

• Before subtracting, however, it is necessary to mathematicallyreduce the degree of saturation in the low frequencyspectrum so that is equal to that in the higher frequencyspectrum (see: http://tinyurl.com/3vm7plo).

• The next figure illustrates how the spectrum of a surfacesegregating polymer additive on polyethylene pellets can beisolated by this method. Note that the saturation equalizationdifference spectrum agrees well with the extracted surfaceadditive spectrum while the bulk additive peak near 1000cm-1 appears in neither the extract nor the differencespectrum plotted above.

21

Isolating the Spectrum of an Additive

5001000150020002500300035004000

Wavenumber

Without Saturation Equalization

5001000150020002500300035004000

Wavenumber

With Saturation Equalization

10 kHzScan Speed

2.5 kHzScan Speed

Difference

Extract

10 kHzScan Speed

2.5 kHzScan Speed

Converted to10 kHz

Difference

Extract

22

Phase Analysis

23

As mentioned in the signal generation discussion, the photoacoustic signal has both amagnitude and a phase. At low absorption values, heat is developed deep within thesample and its transit time to the gas is longer than for high absorption which heatsnear the surface. This results in a phase delay in the signal. Phase delays in thephotoacoustic signal of homogeneous samples are due solely to the sample’sabsorption coefficient variations whereas gradient or layered samples can also havephase differences due to absorption depth differences related to their geometry. Theplot on the right shows the signal magnitude vs. phase behavior for three points inthe spectrum of polyethylene pellets which has surface segregating and bulkadditives. The minimum of the bulk additive band (B) lags the minima of the verystrong C-H and the surface segregating (S) bands as expected.

Phase delay

Phase Delay (Degrees)

Chemically Modified Polymer Surface

Wavenumber (cm )-1

50020003000 2500 15003500 10004000

SB

Surface

WBOH

Mag

nit

ude

24

Phase Analysis for a Surface Treated Polymer

Phase Angle0° 30° 60° 90° 120° 150° 180°

Mag

nit

ud

e

SB(X 1/3)

42°

Surface30°

OH15°

WB69°

25

The photoacoustic signal vs. phase plot reveals that the OHspecies have formed a layer on top of the surface treatmentlayer.

Phase angle selected magnitude spectra of PET/PC assembly show features of the two components more clearly.

2000 1500 1000 500

Wavenumbers (cm )-1

PETMagnitude

PolycarbonateMagnitude

36o

75o

26

Magnitude spectra of a PET/PC assembly with sampling depth varied by modulation frequency.

5001000150020000

1,000

2,000

3,000

Wavenumber (cm )

Polycar-bonate

25 Hz

200 Hz

2.5 kHz

20 kHz

PET

-1

Comparison of Varying Sampling Depth by Modulation Frequency and Phase Angle Selection

Phase Modulation Spectra of PET/PC Layer Assembly

180

135

90

45

0

Phase A

ngle (d

egrees)

2000 1500 1000 500

Wavenumber (cm )-1

400 Hz phase

modulation

27

PC PET

PCPET

Magnitude vs. Phase Plot for PET/PC Assembly

0° 60° 120° 180°

Phase Angle

Pea

k H

eig

ht

(Arb

itra

ry U

nit

s)

0

1597 cm Polycarbonate(Unsaturated)

-1

1578 cm PET(Unsaturated)

-1

1778 cm Polycarbonate(Saturated)

-1

1730 cm PET (Saturated)X 1/2

-1

28

The minima of both the saturated and unsaturated PETpeak magnitudes lead those of the PC peaks which indicatesthat PET is the top layer of the assembly.

Phase Dependence on Depth for Varying Thicknesses of PET on a PC Substrate

Increasing thicknesses of PET on PC progressively delays the phase of thesignal from the 1778 cm-1 PC peak relative to the phase of the 1730 cm-1 PETpeak which remains constant when the PET thickness is changed. 400 HzPhase Modulation

2.5 µm

3.6 µm

6 µm

0

Mag

nit

ude

131°

PET (1730 cm )

0

148°

153°

165°

135°

Phase Angle180°120°60°0°

Polycarbonate (1778 cm )

Bare2.5 µm

3.6 µm

6 µm

Phase Angle180°120°60°0°

-1-1

Mag

nit

ude

29

PET Layer Thickness from Photoacoustic Phase

D = 0.0010 cm/s²

= 400 Hz x 2

0

2

4

6

8

10

Phase Lag

Fil

m T

hic

knes

s (

m)

60°40°20°0°

( )Phase Shift (radians) =

2D

1/2 x L

= x

= film thickness

= thermal diffusion length

= thermal diffusivity= modulation frequency (rad/s)

x

LD

1778 cm-1 Polycarbonate Peak

30

340029001730166016001505146014251370132512751230116011101050

895870830810680

123456789

1011121314151617181920

Bonded O–H stretchingC–H stretchingC=O stretching in xylanKeto-carbonyl conjugated with benzene ringBenzene ring stretching in ligninBenzene ring stretching in ligninCH deformation in lignin and CH bending in xylanCH scissor vibration in celluloseCH bending in cellulose and hemicelluloseCH wagging vibration in celluloseGuaiacyl nuclei in ligninSyringyl nuclei in lignin and C=O in xylanC–O–C asymmetric band in cellulose & hemicelluloseO–H association band in cellulose and hemicelluloseC–O stretching band in cellulose and hemicelluloseC1 group frequency in cellulose and hemicellulose1,3,4-substituted benzene ring in softwood lignin1,3,4,5-substituted benzene ring in hardwood lignin1,3,4-substituted benzene ring in softwood ligninCOH out-of-plane bending in cellulose

3

2

2

2

2

1000150020000

5

10

15

20

25

30

300040000

5

10

15

20

25

30

1

1

2

2

3

3

5

5

4

4

6

6

8

8

7

7

10

10

9

9

11

12

11

12

14

14

13

1315

15

16

16 19

17

18

20

20

Mixed

Southern

Hardwoods

Ponderosa

PineChemometricAnalysis of FTIR Photoacoustic

Spectra of Wood

31

Peak Assignments for Woods

22 24 26 28 30 3222

24

26

28

30

32

Lignin by Wet Chemistry (weight %)

Lig

nin

by P

AS

(w

eight

%)

Acacia

Douglas Fir

Eucalyptus

Loblolly Pine

Ponderosa Pine

Mixed So. Hardwoods

Southern Yellow Pine

White FirRMSEP = 0.70 wt %

r² = 0.93

FTIR Spectra + Chemometrics Lignin Composition

32

14 16 18 20 22 24 26 2814

16

18

20

22

24

26

28

Hemicellulose by Wet Chemistry (weight %)

Hem

icel

lulo

se b

y P

AS

(w

eight

%)

Acacia

Douglas Fir

Eucalyptus

Loblolly Pine

Ponderosa Pine

Mixed So. Hardwoods

Southern Yellow Pine

White FirRMSEP = 0.81 wt %

r² = 0.95

FTIR Spectra + Chemometrics Hemicellulose

33

35 40 45 50 5535

40

45

50

55

Glucan by Wet Chemistry (weight %)

Glu

can b

y P

AS

(w

eight

%)

Acacia

Douglas Fir

Eucalyptus

Loblolly Pine

Ponderosa Pine

Mixed So. Hardwoods

Southern Yellow Pine

White FirRMSEP = 1.48 wt %

r² = 0.98

FTIR Spectra + Chemometrics Glucan

34

RMSEP = 0.87 wt %

r² = 0.94

0 2 4 6 8 10 120

2

4

6

8

10

12

Extractives by Wet Chemistry (weight %)

Ex

trac

tiv

es b

y P

AS

(w

eig

ht

%)

Acacia

Douglas Fir

Eucalyptus

Loblolly Pine

Ponderosa Pine

Mixed So. Hardwoods

Southern Yellow Pine

White Fir

FTIR Spectra + Chemometrics Extractives

35

Qualitative Wood Analyses

Classifying Hardwoods from Softwoods

Classifying Species of Softwoods

36

Comparision of PAS and ATR for Quantitative Analysis of Polymer Blends in Pellet Form

Blend/Component PAS/SEP ATR/SEP

2/A 0.56 0.70

2/B 0.20 0.30

2/C 0.55 0.61

3/A 0.31 1.24

3/B 0.31 1.24

Average SEP 0.39 0.82

Standard Error of Prediction (SEP) (wt %) Results

The SEP for the PAS measurements is a factor of two better than for ATRbecause PAS does not have the ATR crystal contact reproducibility issue andPAS spectra enhance the weak spectral features as demonstrated in the nexttwo figures.

37

Wavenumber

Wavenumber

Polymer Blend No. 2

Photoacoustic Spectra

ATR Spectra

38

Wavenumber

Wavenumber

Polymer Blend No. 3

Photoacoustic Spectra

ATR Spectra

39

FTIR-PAS Analysis of Aging Processes in

Carbon Fiber/Epoxy Composites

• Increasing use of composites in aircraft make characterizing the aging of prepregs and finished parts important.

• Aging changes in epoxy chemistry causes degradation of mechanical properties.

• FTIR-PAS detected changes in chemistry near the surface can predict degradation of bulkmechanical properties.

40

• 25-ply composite panel

• Aging: Baked in air for 4 hours at 11 equally spaced temperatures from 149 to 288 °C (300 to 550 °F). Ten samples at each temperature.

• Interlaminar Shear Strength (ILSS) Test of Mechanical Property Degradation

• Chemometrics relate spectra to ILSS

Artificial Aging of IM7/977-3 Carbon

Fiber/Epoxy Composite

41

PAS Spectra of Thermally Degraded Composites

• PAS spectra of the heat-stressed panels show clear differences at every temperature step.

• The composite has a service limit of 177 °C, so PAS sees changes after four hours of heating even at temperatures well below the service limit.

500100015002000

Wavenumber

288 °C

260 °C

232 °C

204 °C

177 °C

149 °C

274 °C

218 °C

246 °C

191 °C

163 °C

25 °C

42

PAS/PLS Model Predicting Laminate Degradation

• The PAS spectra and interlaminar shear strengths were measured for ten samples at each temperature, and the ILSS measurements were averaged (horizontal bars indicated standard deviations).

• A partial least squares model was built to predict ILSS from the spectra.

2 0 3 0 4 0 5 0 6 0 7 0 8 0

Averaged Inte rlaminar Shear Strength (M Pa)

2 0

3 0

4 0

5 0

6 0

7 0

8 0

Pre

dic

ted

In

terl

am

ina

r S

he

ar

Str

en

gth

(M

Pa

)

204 °C

177 °C

149 °C

191 °C

163 °C

25 °C

288 °C

260 °C

232 °C

274 °C

218 °C

246 °C

standard error = 1.6 MPa

Cross Validation of PLS Model

43

Analysis of Ambient Temperature Aging of

Prepreg Sheets to Determine if

Out Lifetime Has Expired

• Carbon fiber/epoxy prepreg sheets are kept in a freezer prior to laying up of sheets to form a part.

• Lay up process can take days, meanwhile cure advances at room temperature.

• In this experiment, prepregs kept at room temperature in dry air up to 60 days.

• PA spectra and other properties measured every few days. On each occasion; 1 sample spectrum acquired for PLS training set and 2 samples as unknowns.

• Laminates made and tested for ILSS on half of the testing occasions.

44

PA Spectra of Aging Prepreg

Wavenumber (cm-1

)

1000120014001600180020002200

Photo

acou

stic

Sig

nal

(ar

b.

unit

s)

0

20

40

60

80

100

120

140

160

2 Days

12 Days

24 Days

42 Days

21

87

18

00

16

70

15

20

× 5

17

50

90

8

97

6

10

35

11

07

11

53

12

50

13

00

16

15

45

PLS Model for Prepreg Age

from PA Spectra

Actual Prepreg Age (days)

-10 0 10 20 30 40 50 60 70

Pre

dic

ted P

repre

g A

ge

(day

s)

-10

0

10

20

30

40

50

60

70

SECV = 1.47 days

Actual Prepreg Age (days)

-10 0 10 20 30 40 50 60 70

Pre

dic

ted P

repre

g A

ge

(day

s)

-10

0

10

20

30

40

50

60

70

rms error = 2.19 days

Cross Validation of Training Set Analysis of Unknowns

46

Actual ILSS (ksi)

9.0 9.5 10.0 10.5 11.0

Pre

dic

ted I

LS

S (

ksi

)

9.0

9.5

10.0

10.5

11.0

rms error = 0.31 ksi

PLS for ILSS of Aged-Prepreg Laminates

Actual ILSS (ksi)

9.0 9.5 10.0 10.5 11.0

Pre

dic

ted I

LS

S (

ksi

)

9.0

9.5

10.0

10.5

11.0

SECV = 0.21 ksiAge (days) marked

3960

2145

51

27

2

15

3

10

5

33

Cross Validation of Training Set Analysis of Unknowns

47

Processing Wood into Biochar

• FTIR Photoacoustic Spectroscopy allows thedepletion and generation of chemicalcompounds to be observed during thepyrolysis and charring of biomass.

• This capability can be used to characterize therole of biochar chemistry, due to differentcharring conditions and biomass feedstocktypes, in soil fertility, water quality, andcarbon sequestration studies.

48

Charring Apparatus

Wood biomass wasexposed to a series of90s long temperaturesteps starting at a 200°Cnitrogen gas temperatureand proceeding in 25°Csteps to 300°C and on to500°C in 10°C steps.

49

Wavenumber (cm-1

)

1000200030004000

Sig

nal (a

ll sp

ectr

a o

n s

am

e s

ca

le)

0

50

100

150

200

Fresh

200 oC

300 oC

320 oC

340 oC

360 oC

380 oC

400 oC

450 oC

500 oC

Infrared Spectra of Poplar: 25°C to 500°C

Every absorption feature in the spectrum is altered by pyrolysis. Most of themany absorption peaks of the fresh wood disappear and are replaced by a fewbroader absorptions from the char. In addition a background absorption beginsto push the baseline of the spectra upward starting at roughly 340°C. This risingbaseline indicates the formation of biochar, which absorbs all wavelengths ofinfrared light. 50

Pyrolysis Temperature (oC)

200 250 300 350 400 450 500

Heig

ht (s

cale

d s

o m

ax =

100)

0

20

40

60

80

100

Char Formation vs. Temperature

Poplar Wood

51

Wavenumber (cm-1

)

600800100012001400160018002000

Sig

nal (a

ll sp

ectr

a o

n s

am

e s

ca

le)

0

50

100

150

200

Fresh

200 oC

340 oC

380 oC

500 oC

1740

1504

1373

1327

1238

1061

1701

1597

Wave-

number

(cm-1)

Source

3522Bond stretch of O-H hydrogen-

bonded to aromatic network

3356O-H bond stretch in cellulose,

hemicellulose, and lignin

3055C-H bond stretch in aromatic

hydrocarbons

2932C-H bond stretch in aliphatic

(non-aromatic) hydrocarbons

1740C=O bond stretch in xylan (part of

hemicellulose)

1701Carbonyl band (tentative

assignment)

1597Aromatic (benzene) ring stretch

(only from lignin in fresh wood)

1504Aromatic (benzene) ring stretch

in lignin

1373CH2 bending in both cellulose

and hemicellulose

1327CH2 wagging vibration in

cellulose

1238

Mixed source; both syringyl

nuclei in lignin and C=O in

hemicellulose

1061C=O bond stretch in both

cellulose and hemicellulose

Sources of the absorptionbands whose peak heightswere monitored as afunction of pyrolysistemperature

52

Pyrolysis Temperature (oC)

200 250 300 350 400 450 500

He

igh

t (s

ca

led s

o m

ax =

10

0)

0

20

40

60

80

100

1740 cm-1

; C=O stretchin xylan (hemicellulose)

1504 cm-1

; aromaticstretch in lignin

1373 cm-1

; CH2 bend in

cellulose & hemicellulose

1327 cm-1

; CH2 wag in cellulose

1238 cm-1

; syringyl in lignin &C=O in hemicellulose

1061 cm-1

; C=O stretch incellulose & hemicellulose

Species Being DepletedDuring Charring

53

Pyrolysis Temperature (oC)

200 250 300 350 400 450 500

He

igh

t (s

ca

led s

o m

ax =

10

0)

0

20

40

60

80

100

1701 cm-1

; C=O band (tentative)

1597 cm-1

; aromatic ring stretch in lignin

Species Being FormedDuring Charring

54

Pyrolysis Temperature (oC)

200 250 300 350 400 450 500

He

igh

t (s

ca

led t

o m

ax =

10

0)

0

20

40

60

80

100

3522 cm-1

; Stretch of O-H H-bondedto aromatics

3356 cm-1

; O-H stretch in cellulose,hemicellulose & lignin

3055 cm-1

; C-H stretch in aromatics

2932 cm-1

; C-H stretch in aliphatics

Formation- Depletion of Species During Charring

55

PAS Reference SamplesPAS magnitude spectra of samples must be divided by a spectrumof a black absorber to remove spectral features due to the lightsource and the spectrometer’s optics. PAS phase spectra must bereferenced to a specific phase angle in order to be used inlinearizing spectra and in other data processing calculations .MTEC has developed special reference samples for thesepurposes.

Magnitude Reference Phase Reference56