~4~od Science

Wood Sci. Technol. 19:139-149 (1985) and Technology

9 Springer-Verlag 1985

The significance of the glass transition of

lignin in thermomechanical pulping*
G. M. Irvine, Clayton, Victoria, Australia

Summary. Changing the temperature at which the thermomechanical pulping of wood is

carried out may result in si~ificant changes in pulp properties. Such changes are widely
attributed to the effect of the glass transition of the li~cmin. Pulping literature references to this
effect frequently appear to imply that this transition is manifested as an abrupt softening of the
lignin either at a single temperature (the glass transition "point") or over a relatively narrow
temperature range. Application of the time temperature correspondence principle of polymer
science and the Wflliams-Landel-Ferry equation to laboratory measurements of the transition
suggests otherwise: in terms of the response of the lignin material to the very rapid cyclic
deformation processes occurring within the refiner, the glass transition is predicted to occur
over a temperature interval spanning the approximate temperature range 100-170 ~ within
which most thermomechaaical pulping is carried out. The consequences of this prediction are
discussed in terms of the likely fracture processes occurring during the refining process.

Introduction

When wood is pulped by the thermomechanical process, that is, by the mechanical
refining of wood chips at nominal temperatures of 100 ~ or higher, the properties
of the pulp are remarkably dependent upon the temperature attained within the
refiner (Atack 1972; t-Iiggins et al. 1978). For softwoods an optimum temperature
for defibration and refining has been identified at about 130-135 ~ and is thought
to correspond to a softe~aJ_ng of the wood structure resulting from the glass transition
of the lignin (Atack 1972; Higgins et al. 1978; Sinkey 1979). Recent studies by
differential thermal analysis (DTA) of eucalypt and pine samples, on the other
hand, showed that the glass transition of lignin in water-saturated wood occurred
over the approximate temperature interval 6 0 - 9 0 ~ (Irvine 1984). This apparent
discrepancy m a y be superficially explained in terms of the recognized dependence
of such viscoelastic p h e n o m e n a upon the time scale of the experiment. In general,
however, the relationship between the glass transition of lignin and the processes of
wood fracture which occur in the refiner has not been well understood. The
objective of this paper is to show how some known factors concerning the
viscoelastic behaviour of polymers can be used to assist our understanding of this
relationship. At the same time, an apparently c o m m o n conception that the

* The author thanks Mr. V. Balodis and Dr. H. G. Higgins for valuable comments on the
manuscript
140 G.M.Irvine

transition occurs as a sudden "melting" of the material at a glass transition "point"

will be shown to have tittle basis.

Nature of the glass transition

The glass transition, a characteristic of the viscoelastic behaviour of amorphous
polymers and other glass-forming liquids, may be manifested in many ways, for
example, as a change in the tensile or shear modulus (or stiffness), the heat
capacity, or the coefficient of thermal expansion. At temperatures below the
transition a viscoelastic material is stiff (the modulus is high) and glassy, and
probably brittle. In the transition region the stiffness decreases by, typically, a
factor of 103, and at higher temperatures, if the molecular weight is above a critical
value (normally about 104-10~), the material will exhibit rubber-like elasticity as a
result of chain entanglements (Kaelble 1971; Aklonis et al. 1972). If the material is
not crosslinked, and providing thermal degradation does not occur, further increase
in temperature will eventually result in rubbery flow as the entanglements begin to
slip. The length of the rubbery "plateau" in the stiffness-temperature curve is
dependent upon the number of entanglements per molecule, and hence upon the
molecular weight. If crosslinking is present, such flow cannot occur. Since the
probable form of lignin in situ is that of a crosslinked polymer of high molecular
weight (Goring 1971), flow is unlikely to occur in this material unless numerous
primary chemical bonds are first broken by, for example, chemical action or milling
of the wood.
The glass transition phenomenon is a controversial subject, and several different
theories of its mechanism exist (Andrews 1966; Kaelble 1971; Aklonis et al. 1972).
Probably the most common interpretation is that rotational or translational
motions of segments of the molecules become progressively activated as thermal
expansion provides the amount of "free volume" (the volume not occupied by
molecules) necessary to allow these motions to take place. The addition of low
molecular weight diluents (plasticizers) to a polymer may shift its glass transition to
lower temperatures. This effect may be explained in terms of an increase in the free
volume of the system. In cases of plasticization of polar molecules b y water,
however, this explanation is not entirely satisfactory (Andrews 1966; Aklonis et al.
1972). In particular, the plasticization of llgnin by water is thought to result from
the replacement of intermolecular hydrogen bonding within the lignin by lignin-
water linkages (Sakata, Senju 1975; Atack, Heitner 1979).

Experimental measurements of the lignin #ass transition

Few measurements of glass transitions in lignin have been reported which are
directly comparable with the above-mentioned results of DTA studies of water-
plasticized in situ llgnin (Irvine 1984). Thermally induced shrinking and coalescence
of powdered materials such as isolated lignins (Goring 1963; Takamura 1968;
Sakata, Senju 1975) may be associated with rubbery flow of the polymer molecules
rather than with the glass transition - the procedures by which these materials are
obtained probably cause structure modifications of sufficient magnitude for such
flow to become possible. Transitions detected in-isolated lignins at much lower
temperatures (Icelovich et al. 1977) presumably relate to secondary relaxations, i.e.
Glass transition of lJgnin in thermomechanical pulping 141

transitions in the glassy region which result from motions which are more localized
than those concerned in the glass transition (Kaelble 1971; Aldonis etal. 1972).
Differential scanning calorimetric studies of isolated llgnins by Hatakeyama et al.
(1972) were of dry samples only. The usefulness of such measurements on isolated
lignins is also limited by the dependence of the softening and/or glass transition
temperatures of these materials on the particular wood species concerned, the
isolation procedures employed and the thermal histories of the samples (Goring
1963; Sakata, Senju 1975; Irvine 1984).
Most published studies of transitions in in situ llgnins relate to torsional
pendulum experiments, in which wood samples of appropriate geometry are
initially torsionally strained, and then released, and the resulting damped oscilla-
tions observed. From such experiments, values can be obtained for the torsional
modulus and the internal friction, the latter defining the proportion of the
deformation energy Which is dissipated as heat. Experiments of this type conducted
by Atack (1972) and by H6glund et al. (1976) have shown that differences between
transition temperatures in various hardwood and softwood species are small. The
"softening temperature" (defined as the temperature at which the internal friction is
at a maximum) of water-saturated black spruce was found by Atack (1972) to be
about 100 ~ at a measuring frequency of about 10 Hz, equivalent to a deformation
time scale of about 10-1 s. The temperature interval over which the transition
occurred, obtainable from the torsional modulus vs temperature curve in a manner
analogous to that adopted for DTA curves (Irvine 1984), was approximately
80-110~ H6glund etal. (1976) obtained for several wood species softening
temperatures of approximately 90 ~ at about 1 Hz. Becker et al. (1977), using a
related electrodynamic method in which the measuring frequency was about
10~ Hz, found the softening temperature to have shifted to about 120~ at this
higher frequency the transition appeared to span the approximate interval
90-150~
Since the time scale of the DTA experiment (Irvine 1984) is about 10~ s, it is
apparent that the glass transition of the llgnin in wood shifts progressively to higher
temperatures as the time interval of the measurement procedure becomes shorter.
This shift, a well-known characteristic of the glass transition, is a consequence of the
kinetic nature of the transition (Aklonis et al. 1972). These results also illustrate the
less widely appreciated characteristic that, as the measuring time becomes shorter,
the transition also becomes broader. In the disc refiner, the time scale of the
deformation of the material is even shorter than that of the experiments of Becker
et al. (1977). Estimation of the temperature range of the lignin glass transition
under such circumstances is possible, though not as simple a matter as is sometimes
suggested.
The low temperature limit of the temperature range of the glass transition as
obtained by any of several possible measurement techniques in which temperature
is the experimental variable is commonly referred to as the glass transition
temperature Tg. As noted above, however, the value obtained for this temperature
~1l depend on the time scale of the measurement procedure. In this paper, Tg will
be used to refer specifically to the transition temperature measured by some slow
technique such as dilatometry, where the time interval of measurement is of the
order of 104 seconds (Cooper 1966). Consequently, Tg will be treated as a constant
142 G.M. Irvine

for the material concerned. Variable quantities T1 and T2 will be Used to refer,
respectively, to the low and high temperature limits of the transition as the time
interval of measurement varies from 104 s to smaller values.

Time-temperature correspondence and the WLF equation

Mechanical properties of viscoelastic materials, such as the tensile and shear

moduli, are functions of time as well as of temperature. In this respect there is an
interrelationship between time and temperature such that the viscoelastic behaviour
of a polymer over a short time interval at one temperature is equivalent to its
behaviour over a longer interval at a lower temperature (Aklonis et al. 1972). For
example, in the measurement of the stress relaxation behaviour (the time-
dependent change in stress at a constant deformation and temperature) of a
polymer, the curve obtainable as a function of log time at any temperature T is
similar to that obtainable at any other chosen reference temperature, but is shifted
along the log time axis by an amount log aT, where aT is known as the shift factor.
For many amorphous polymers the shift factor may be determined by use of the
Williams-Landel-Ferry (WLF) equation, the so-called "universal" form o f which is
log aT = - 17.44 ( Y - Tg)/(51.6 + T - Tg) (1)
where the reference temperature is the dilatometric or equivalent glass transition
temperature Tg (Williams et al. 1955). The WLF equation is illustrated graphically
in Fig. 1. f f t h e time interval for this measurement of Tg is taken to be 104 s, the log
aT scale can be replaced by the logarithmic time scale shown at the fight hand side
of Fig. 1.
The WLF equation therefore enables the transformation of data obtained as a
function of time at a particular temperature into a corresponding function of time

0 4
-I Goring's "powder collapse" method 5
-2 ~ E l Differential
e Dilatometry
thermal analysis 2
-3 I
-4 0
Torsional pendulum -~
-5
o~-6
-7 ctrodyTmic mefhod
-8 -4
-9 -5
-I0 -6
-II
-12
I I 1 I I I
0 20 40 6(3 80 I00
T-Tg
Fig. 1. WLF shift factor axas a function of temperature. Ttme intervals corresponding to
various experimental methods and to the disc refiner are also indicated
Glass transition of lignin in thermomechanical pulping 143

170

L••
150

~40

130
TMP optimum temperature

120

~l IlO i Rubbery
.... \ region
I-- /ransition
I00 region

9O

8O

70
l I I regt~ ~TI
6O I regi~ I I I

-6 -5 -4 -3 -2 -I 0 I 2 3 4
log time,s
Fig. 2. Calculated variation with time of the lower (T~) and upper (T~) temperature limits of
the glass transition oflignin in water-saturated Pinus radiata wood

at a different temperature by means of a straightforward displacement along the

log time axis. However, DTA and torsional pendulum experiments yield essentially
functions of temperature at a particular time (or frequency), and not the reverse.
Such a simple time-temperature data conversion step is not, therefore, possible,
since Fig. 1 shows that the relationship between time and temperature is not linear.
Previous attempts at transformation of such experimental results into the time
scales appropriate to wood defibrafion, refining and grinding operations by means
of assumed linear time-temperature relationships (Goring 1963; Atack 1972;
H6glund et al. 1976; Becket et al. 1977) must therefore be regarded as unsound, or
at best as rough approximations.
]-he transformation of a complete temperature function, such as a DTA curve
over the region of the glass transition, would be a complex procedure. However, a
reasonable approximation can be derived from just two points representing the
limits of the transition range, e.g. the points of intersection of the tangents to the
DTA curve before, during and after the transition (Irvine 1984). The temperature of
the lower (glassy) limit of the transition for a time interval other than 102 s (the
DTA time scale) is then simply calculated from Eq. (1): Tg at l(P s is first found,
then T1 for the required time interval. Calculation of temperatures for the upper
(rubbery) limit, T2, by means of Eq. (1) is more complicated, since the point
required is not only time-shifted from Tg at 104 s, but is also further shifted by the
width of the transition.
Figure 2 shows the results obtained from application of the WLF equation t o
the transition temperature range (65-87~ obtained for water-saturated P i n u s
144 G.M. Irvine

radiata wood by DTA at 102 s (Irvine 1984). The calculations are given in the
Appendix. The predicted temperature ranges of the transition at 10 -1 and 10-3s
(79-116 ~ and 93-152 ~ respectively) are in good agreement with the previously
mentioned results obtained by Atack (1972) at 10 Hz and by Becker et al. (1977) at
l0 s Hz. The torsional pendulum experiments of H6glund et al. (1976) did not give a
clear indication of the temperature range over which the transition occurred.
However, a value of about 90 ~ was obtained for the temperature at which the
internal friction was at a maximum. I-Iris point will normally occur close to the
centre of the transition region, which is predicted from Fig. 2 to the about 90 ~ for
a measuring time of 1 second. These comparisons indicate agreement with the WLF
relationship over a range of time scales and temperatures. It is therefore reasonable
to expect the predicted results to hold at even shorter time scales, providing that the
accepted upper limit of validity of the WLF equation (about 100 ~ above Tg
(Williams et al. 1955) is not exceeded.
The time scale range of 10-3"~ to 10~'5 s shown as a hatched area in Fig. 2
corresponds to the deformation frequency range (103'5 to 104-s Hz) appropriate to a
typical disc refining operation. This range is readily determined from the refiner
plate surface geometry and the motor speed. Also shown hatched in Fig. 2 is the
temperature range 130-135 ~ which is regarded as the optimum for the thermo-
mechanical pulping of softwoods. The zone of intersection of these hatched areas is
close to the centre of the predicted region of the glass transition.

Fracture processes in the refiner

HOglund et al. (1976) suggested that the optimum temperature for the TMP process
is that at which the maximum quantity of energy which makes the fibres "flexible
by internal bond breaking and heat generation" is absorbed. They further
suggested that this "useful" energy can be identified with the energy dissipated
within the viscoelastic material over the stress cycle as a hysteresis loss. The
optimum temperature should therefore fall somewhere between the low-tem-
perature limit and the centre of the transition (H6glund et al. 1976; Becker et al.
1977). This approach is questionable, however, since permanent structure modifica-
tion does not necessarily result from such hysteretic energy dissipation. Indeed, in
torsional pendulum experiments the imposed stresses and resulting strains are
normally well below those levels at which significant fracture of the material will
occur. In the refiner, on the other hand, hysteretic energy dissipation will certainly
occur as a necessary consequence of the processes of bond-breaking and fracture
(see below), but there is no a priori reason to associate a maximum in the hysteresis
effect with the optimum temperature for the refining process.
The process of disintegrating chips into fibre bundles, separating the fibres and
refining them to achieve the desired properties involves fracture through fatigue
mechanisms which operate parallel to and across various interfaces within the
complex structure of the wood. Parallel fracture is generally more desirable and
transverse fracture less so. In considering the defibration and refining processes, we
should perhaps be concerned more with the ultimate properties (such as tensile
strength and ultimate strain) of wood and its components than with the moduli at
Glass transition of fignin in thermomechanical pulping 145

sub-fracture levels. However, for polymeric materials these ultimate properties have
been shown to depend on both temperature and time in much the same way as the
moduli (Smith 1958; Migiiaresi et al. 1981). Therefore, for example, as the tensile
and shear moduli oflignin decrease over the course of the transition, the strength in
tension or shear should also decrease markedly. Thus, as the lignin softens, it will
also become weaker.
The processes that occur when composite materials are subjected to both.
monotonic and cyclic loading have been outlined in a recent monograph by Hull
(1981). If a composite consisting of elastic fibres in a viscoelastic matrix is loaded in
a direction transverse to the longitudinal fibre axis, a strain magnification, and
hence a non-uniform stress distribution, will occur in the regions of the matrix
between the fibres. Stress concentrations will also occur in the vicinity of defects or
voids in the material. Debonding at the fibre-matrix interface and cracking of the
matrix by brittle fracture or, where possible, by molecular flow may commence at
stress and strain levels well below those at which ultimate failure of the material
occurs. Under conditions of cyclic (fatigue) loading, such debonding and matrix
cracking will occur at even lower stress levels, and the method and rate of crack
propagation will depend markedly on the viscoelastic properties of the matrix
material.
The defibration and refining of wood chips can be considered to be a particular,
although exceedingly complex, example of this fracture mechanism, with fracture
occurring on several levels from separation of the fibres at or near the middle
lameUa to fracture between and within the various elements of the fibre at the
microfibfiIlar level. Let us therefore consider the changes that are likely to occur as
the temperature is increased from the glassy to the rubbery end of the glass
transition. For these purposes it is helpful to refer to the interrupted lamella model
for the ultrastructure of the wood cell wall, developed from experimental obser-
vations by Goring and co-workers, in which cellulosic microfibrils and lignin-
hemicellulose matrix elements are preferentially oriented in discontinuous lamellae
parallel to the middle lamella (Kerr, Goring 1975). As an ultimate description of
cell ultrastructure, the model may not be fully acceptable (Ruel, Barnoud 1981), but
at the level of detail which we require it is unlikely to be grossly misleading.
Under the conditions of water-saturation that apply during defibration and
refining operations, plasticization of the hydrophilic amorphous cellulose and
hemiceiluloses by water will result in the reduction of their glass transition
temperatures to or below room temperature (Back, Salmen 1982). At temperatures
below the lignin glass transition, the cell wall can therefore be considered as
consisting of largely tangentially oriented lamellae of two zones, each of which has
both crystalline or glassy (relatively hard and strong) and rubbery or plastic
(relatively soft and weak) components: the crystalline (hard) and amorphous (soft)
regions of the microfibfils, and the lignin (hard) and hemicellulose (soft) forming
the matrix. Each zone is incompletely softened, and the overall modulus of the
wood remains high. The middle lamella, consisting largely (about 70%) of lignin,
will be hard and glassy. Fracture of the fibres will take place predominantly along
surfaces parallel to the fibre direction where the bonding between lamellae is
weakest. There will also be a tendency for cracks propagating within the cell wall to
travel in a transverse direction. The net effect will be that the desirable fibre surface
146 G.M. Irvine
development will be companied by a certain incidence of fracture of fibres across
the direction of the fibre axis (fibre damage). The fracture will be more of a brittle
nature, and energy losses through viscoelastic hysteresis mechanisms less marked,
than at higher temperatures.
As the temperature is increased through the region of the lignin glass transition,
the remaining (lignin) portion of the matrix regions will begin to soften, as will the
middle lamella. Because of the preferential tangential orientation of the matrix
regions (Kerr, Goring 1975), a greater proportion of cracks will occur along the
tangential and fibre directions creating new surfaces within the matrix and along
the interfaces with the cellulosic fibrils. The desired surface development will thus
be obtained with a lesser incidence of fibre damage. Crack propagation will,
however, be accompanied by a significantly greater degree of viscoelastic deforma-
tion at the crack tip. Consequently, higher hysteresis losses will be incurred in
addition to the greater useful consumption of energy.
By analogy with the temperature dependence of the ultimate mechanical
properties of synthetic polymers and composites (Smith 1958; Migliaresi et al.
1981), the response of the liwnin-fich regions to the refining forces will approach
that of a cross-linked rubber as the refining temperature nears the upper limit of
the lignin glass transition. The failure mode will then again tend towards brittle
fracture, but at a higher ultimate strain than for the glassy state. Because of the
three-dimensional structure of the lignin, it is unlikely that significant rubbery flow
of the material will take place. Under these conditions the fibres will tend to
separate within the middle lamella or at the interface with the primary wall, rather
than through the relatively narrow interconnected zones of matrix within the cell
wall. The fibres are then likely to be completely encased in lignin which will
reharden on cooling.
Such a dependence upon refining temperature of the nature of the pulp
produced in a thermomechanical process was observed by Higgins et al. (1978) in
the laboratory-scale production of TMP from Pinus radiata at refining temperatures
between 100~ and 170 ~ Maximum development of fibre surfaces occurred at
about 130 ~ less fibrillation and a greater incidence of fibre damage occurred at
lower temperatures, and at higher temperatures the fibres produced were smooth,
lignin-encased, had poor bonding properties, and were resistant to further refining
under atmospheric conditions. Maximum energy consumption also occurred near
130~ however, the improvement in strength properties (breaking length, burst
index) was proportionately greater than the corresponding increase in energy
consumption (Higglns, Puff 1978).

Concluding remarks

Figure 2 indicates that the glass transition of lignin wilhin wood at the time scales
pertinent to commercial refining processes may span a temperature interval
approaching 100 ~ In these circumstances, therefore, the frequently encountered
reference to the glass transition "point" ofli-nin is misleading. On the other hand,
if the change in ~4scoelastic properties of lignin over the transition region were
similar to those of other un- or lightly-crosslinked polymeric materials, the decrease
Glass transition of lignin in thermomechanical pulping 147

in modulus over this interval would be approximately three orders of magnitude,

and the decrease in tensile strength perhaps two orders of magnitude (Smith 1958;
Aklonis etal. 1972; Migharesi etal. 1981). Since lignin is a three-dimensional
polymer, the changes in these and other mechanical properties over the transition
may be smaller, but still sufficiently large for si,p-nificant changes to occur in the
modes of fracture of the material at the various levels of structure discussed earlier.
In either self- or externally-pressurized commercial refiners, considerable
temperature variation normally occurs between the inlet and discharge. For
example, in a commercial single-stage open-discharge refiner operating under
certain conditions, measured temperatures ranged from 100 ~ at the inlet and
discharge to 155 ~ near the centre of the refining zone (Stationwala et al. 1979).
From Fig. 2 it may be deduced that such a temperature interval might correspond
to up to a hundred-fold change in the effective modulus of the lignin in response to
the applied forces of deformation. In a recent analysis of the different energy
requirements of wood grinding, chip refining, and related processes, Atack (1981)
concluded that substantial improvements in energy-quality relationships in these
processes were unlikely. In relation to chip refining, however, no mention was
made of the effect of temperature on energy consumption and pulp quality. It may
be that developments in technology and engineering may allow greater control over
the temperature profile within the refining zone; if so, further gains in energy-
quality relationships may indeed be possible.

Apl~endix

Examples of use of WLF equation (Eq. 1)

1. Calculation, for untreated water-saturated Pinus radiata, of T1, the lower tem-
perature limit of the transition for a general measuring time of 10Y seconds, given
that T1=65~ when y = 2 (Irvine 1984). Since Tg has been defined for a time
interval of 104 s, Eq. (1) becomes
log a T = y - 4 = - 17.44 (T~ - Tg)/(51.6 +T~ - T g ) (2)
The given data may be substituted into Eq. (2) to obtain Tg= 58.3 ~ (Since these
equations involve temperature differences only, it is permissable to retain tem-
peratures in ~ rather than converting to Kelvin.) Eq. (2) then becomes
y = (990 - 13.44 T0/(T1 - 6.7) (3)
or

T1 = (990 + 6.7 y)/(y + 13.44) (4)

2. Calculation, for the same material, of Ta, the upper temperature limit of the
transition for a general time of 10Y s, given that T2 = 87 ~ when y = 2 (Ixvine 1984).
As could be seen, T2 diverges further from T1 as the measuring time becomes
smaller. However, the time-temperature correspondence principle shows that, for a
series of isothermal experiments, the separation between the be~nning and end
points of the transition on the logarithmic time scale is constant. Consequently, the
variation of Ta with time differs from that ofT1 with time by that same constant log
148 G.M. irvine

time difference. If T1 and T2 are both known for a given time, that difference can be
calculated, thus:
Substitution of 87 ~ for T1 in Eq. (3) yields y = -2.23. Since this value for T2 is
observed at y = 2, the constant time factor difference between the T1 and T2 curves
is 104~ s. The T2 relationship can then be obtained directly from Eq. (4) by
replacing y by y-4.23, giving
T2 = (962 + 6.7 y/(y + 9.21) (5)
These relationships are illustrated in Fig. 2.

References

Aklonis, J. J.; MacKnight, W. J.; Shen, M. 1972: Introduction to polymer viscoelasticity. New
York: Wfley-interscience
Andrews, R. D. 1966: Transition phenomena and solid-state structure in glassy polymers. J.
Polymer Sci. C14:261-265
Atack, D. 1972: On the characterization of pressurized refiner mechanical pulps. Svensk
Papperstid. 75 (3): 89-94
Atack, D. 1981: Fundamental differences in energy requirement between the mechanical
pulping processes. Svensk Papperstid. 84 (14): 22-26
Atack, D.; Heitner, C. 1979: Dynamic mechanical properties of sulphonated eastern black
spruce. Pulp Paper Can. Trans. Tech. Sec. 5 (4): TR 99-108
Back, E. L.; Salmen, N. L. 1982: Glass transitions of wood components hold implications for
molding and pulping processes. Tappi 65 (7): 107-110
Becker, H.; Hrglund, H.; Ttstad, G. 1977: Frequency and temperature in chip refining. Paperi
ja Puu 59 (3): 123-130
Cooper, W. 1%6: Elastomers, synthetic. In: Encyclopedia of polymer science and technology.
Vok 5, pp. 406-482. New York: Wlley-interscience
Goring, D. A. I. 1963: Thermal softening of lignin~ hemicellulose and cellulose. Pulp Paper
Mag, Can. 64 (12): T 517-527
Goring, D. A. I. 1971: Polymer properties oflignin and lj~nin derivatives. In: Sarkanen, K. V.;
Ludwig, C. I-L (Eds.): Lignins: Occurrence, formation, structure and reactions, pp.
695-~768. New York: Wlley-interscience
Hatakeyama, H.; Kubota, K.; Nakano, J. 1972: Thermal analysis of lignin by differential
scanning calorimetry. Cellulose Chem. Technol. 6:521-529
l-figgins, H. G.; Irvine, G. M.; Purl, V.; Wardrop, A~ B. 1978: Conditions for obtaining
optimum properties of radiata and slash pine thermomechanieal and chemithermo-
mechanical pulps. Appita 32 (1): 23-33
Higgins, H. G.; Puri, V. 1978: Personal communication
HSglund, IL; Sohlin, U.; Tlstad, G. 1976: Physical properties of wood in relation to chip
refining. Tappi 59 (6): 144-147
Hull, D. 1981: An introduction to composite materials. Cambridge: Cambridge University
Press
Ioelovich, M. Ya.; Kaimin', I. F.; Kreitsberg, Z. N.; Aronchik, B. M.; Dobele, G. V. 1977:
Transition temperatures of lignin and the effect of low-molecular weight compounds on
them. Khim. Drev. (Riga) 3:31-35
lrvine, G. M. 1984: The glass transitions oflignin and hemicellulose and their measurement by
differential thermal analysis. Tappi 67 (5): 118-121
Kaelble, D. H. 1971: Physical chemistry of adhesion. New York: WJley-interscience
Kerr, A. J.; Goring, D. A. I. 1975: The ultrastructural arrangement of the wood cell wall.
Cellulose Chem. Technok 9:563-573
Migliaresi, C.; Nicolais, L.; Nicodemo, L.; Di Benedetto, A~ T. 1981: Mechanical properties of
glass microsphere-filled composite materials. Polymer 2 (1): 29-35
Glass transition of lignin in thermomechanical pulping 149

Rue], I~; Bamoud, F. 1981: Supramolecular aspects ofcomp, wood constituents as seen by elec-
tronmicroscope investigations_ Preprints International Symposium on Wood and Pulping
Chemistry, Stockholm, Vol. 1:11-15. Stockholm: SPCI
Sakata, I.; Senju, R~ 1975: Thermoplastic behaviour of lignin with various synthetic
plasticizers. J. Appl. Polymer Sci. 19:2799-2810
Sinkey, J. D. 1979: The heating of wood chips in thermomechanical pulping. Pulp Paper Can.
80 (3): T75-79
Smith~ T. L 1958: Dependence of the ultimate properties o f a GR-S-rubber on strain rate and
temperature. J. Polymer Sci 32: 99-113
Stationwala, M. I.; Atack, D.; Wood, J. 1~; W'fld, D. J.; Karnis, A~ 1979: 3-he effect ofcoutroI
'variables on refining zone conditions and pulp properties. Preprints International
Mechanical Pulping Conference, Toronto, pp. 93-109. Montreal: Canad. Pulp Paper
Assoc.
Takamura, N. 1968: Studies on hot pressing and drying process in the production of
fibreboard. III. Softening of fibre components in hot pressing of fibre mat. J. Japan Wood
Res. Soc. 14 (2): 75-79
Wdliams, M. L.; Landel, R~ F.; Fe W , J. D. 1955: The temperature dependence of relaxation
mechanisms in amorphous polymers and other glass-forming liquids. J. Am. Chem. SOc.
77 (14): 3701-3707

(Received September 12, 1983)

Dr. G. M. Irvine,
CSIRO Division of Chemical and Wood Technology,
Private Bag 10, Clayton, Victoria,
Australia, 3168