34
The Glass Transition Temperature Measurement of Nitrocellulose by
Torsional Braid Analysis
Jia Zhanning
The Chemical Engineering Department, Beijing Institute of Technology, Beijing (China)
Messung der Glasubergangstemperatur von Nitrocellulose durch
Torsionsanalyse
Die Glasiibergangstemperatur von reiner Nitrocellulose wird
bestimmt durch Torsionsanalyse auf einem Substrat (braid). Die Repro-
duzierbarkeit und Zuverlassigkeit der experimentellen Ergebnisse wird
diskutiert.
Summary
The glass transition temperature of pure nitrocellulose is determined
by torsional braid analysis. The reproducibility and reliability of the
measured results are discussed.
1. Introduction
The primary purpose of this work is to determine the glass
transition temperature, Tg, of pure nitrocellulose (NC) by the
torsional braid analysis (TBA) technique. Since NC tends to
decompose with heat, reports had long been inconsistent on
its glass transition temperature from techniques like thermo-
gravimetric analysis (TGA) and differential scanning calori-
metry (DSC). Here, the glass transition temperature of NC
was obtained by TBA with good reproducibility and reliabili-
tY.
Torsional braid analysis is a dynamic mechanical method.
Dynamic tests are very sensitive to, and thus useful in deter-
mination of, the glass transition, crosslinking, phase separa-
tion and effects of fillers in polymers. However, in contrast to
classical methods of dynamic mechanical spectroscopy, for
which geometrical specimens are fabricated, the present tech-
nique permits the investigation of any type of polymer (or
chemical compound) which can be deposited or synthesized
on the braid, and thus much less quantities of materials are
needed. Furthermore, since its low frequency range is corre-
spondent to the relaxation time of polymers, TBA is rather
sensitive to movement of polymer chain segments and so it
can be used for the investigation of secondary transition and
molecular movement in polymers.
2. Principle
Perfectly elastic materials have no mechanical damping
and all the energy is stored as potential energy during defor-
mation. It is converted into kinetic energy when the applied
load is removed and the sample snaps back to its original
dimension. Thus, in the oscillations of such a material, the
amplitude will not decay as time goes by. This is called a
damping-free oscillation. In case of a viscous liquid, all the
energy used in deformations is dissipated into heat since it
0721-311S/92/0102-0034 $ 3 5 0 +.25/0
Propellants, Explosives, Pyrotechnics 17,34-37 ( 1 992)
Mesure de la temperature de transition vitreuse de la nitrocellulose
par analyse de torsion
La tempkrature de transition vitreuse de la nitrocellulose pure est
dkterminke par I’analyse de torsion sur un substrat (braid). La reproduc-
tibilitk et la fiabilitk des rksultats exptrimentaux font l’objet d’une dis-
cussion.
can not store potential energy at all. For a liquid inertia mem-
ber, the oscillations amplitude decays rapidly with time. The
properties of viscoelastic polymers are just between the two
extremes. When such materials are deformed, part of the
energy is stored as potential energy and part is dissipated as
heat. The oscillation amplitude decays with time.
This type of dynamic mechanical properties can be expres-
sed by the complex modulus G.
G = G’+iG” (1)
The real part of the modulus G’ is often called the storage
modulus because it defines the energy stored in the specimen
due to the applied strain. The imaginary part of the modulus
G” defines the dissipation of energy and is often called the
loss modulus. A very useful damping term is defined as
G”/G’=tan 6, which is called dissipation factor or loss tan-
gent. It is proportional to the ratio of energy dissipated per
cycle to the maximum potential energy stored during a cycle,
and thus obviously increases with dissipated energy.
In TBA test, the dynamic mechanical behavior is defined
in terms of the logarithmic decrement A and the relative
modulus G, (or relative rigidity l/p2). The logarithmic decre-
ment A is calculated from the logarithm of the ratio of the
peak amplitudes of two successive oscillations.
It is approximetely related to the loss tangent tan 6 as,
A = n: tan 6 = n G’/G’ (3)
Thus, the energy dissipation in terms of A or tan 6 can be
determined from the peak amplitudes of the oscillations.
The shear modulus G is defined,
8IL
G=- (4)
R2p2
where L, R , I are the length, radius, and moment of the inertia
disc, respectively, p is the period. Supposing Go is the shear
modulus of blank braid and G of the impregnated specimen,
they only differ in the periods, that is,
81L/R4p2 - 1/p2
G, = G/Go -
81L/R4pi
0VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1992
Propellants, Explosives, Pyrotechnics 17, 34-37 (1992) Glass Transition Temperature Measurement of NC by Torsional Braid 35

Since the period p for blank braid is constant, the relative loose single filaments. The braid is treated at 450°C for 2
modulus Gr only depends on the period p of the impregnated hours before experimentation.
specimen. Assuming that l/p;= 1, then Gr = ]/pi. Thus the 20-100mg of acetone solution ( 5 % ) of purified nitrocellu-
relative modulus determined by the experiment is merely lose is impregnated to such a braid. The braid is vacuumed
related to p. for 2-4 hours and then heated at 80°C for 30 min to remove
the residual solvent. The test is conducted in dry nitrogen or
3. Apparatus and Experimental in vacuum under constant warm-up speed of l-2"C/min. The
thermomechanical behavior diagrams of two types of NC
The schematic diagram of this TBA device is given in (N% = 12.6 and 12.2) are shown in Figures 2 and 3, respec-
Figure 1, the equipment is manually operating. It consists of tively.
braid specimen, fixed clamp, braid supporting rod, torsion
head assembly, inertial polarizer disc, light source and photo- 4. Results and Discussion
electric transducer. The specimen braid is a piece of glass
fibre band of about 200mm long, composed of about 3600 Two damping peaks of NC (N% = 12.6 and 12.2) for the
temperature range of 0°C-230°C can be seen in Figures 2 and
3(a). Both of these types show similar characteristics of the
torsion head assembly nitrocellulose. The maximum damping peak is at 174°C-
176°C at which their relative modulus decrease dramatically
fixed clamp by two to three orders of magnitude. This indicates that there
is a primary glass transition for NC at 174"C-l76"C. A shoul-
specimen
der-like peak is observed at about lOO"C, which is accompa-
'I nied by a not so dramatic but still obvious decrease of the
I
I relative modulus. It should be a secondary transition of NC.
braid supporting r o d
The glass transition is a transition of frozen segments into
mobile segments, and the energy needed for this transition is
termed as the activation energy E for glass transition. It can
be calculated from the width of the damping peak from TBA
test (see Appendix). The activation energy for glass transition
inertial polorizer disc of NC is 116.6 kcal/mol, and that for the secondary transition
is about 4.0 kcal/mol, which is in good agreement with the
LF/J-sJ--
'
polarizer

photoelectric transducer
Figure 1. Schematic diagram of torsional braid apparatus.
activation energies of general polymers for glass transition
(50-100 kcal/mol) and secondary transition (5- 15 kcal/mol).
It is also shown that NC begins to decompose at about
183°C and is fully decomposed at about 210"C, which is cor-

I I I I

20 $2
X
a
c
C
0)

10 E
L
0
W
-0
u
.I

E
I:
.-c
L
(3
cn
0
J

II I I I t 1
0 50 100
Temperature ["CI -
150 200 Figure 2. Thermomechanical behavior dia-
gram of nitrocellulose (N%=12.6).
36 Jia Zhanning Propellants, Explosives, Pyrotechnics 17,34-37 (1992)

Temperature I"C1- Temperature IOCI-

100 200 300 0 100 200 300 LOO
10

TBA
- 183"t

0
X
a
-1

,185"Td
lbl
\ -
TGA
~192OTd 0
Exothermic

DSC
~. 330OTd I
0 100 200
-
300 Exothermic c
Temperature ["CI

Figure 3. Diagrams of nitrocellulose (N%=12.2) : (a) torsional braid

analysis (TBA), (b) thermogravimetric analysis (TGA).
2
Endothermic
respondent to another damping peak in the diagram. During
this process, NC is crosslinked to form a hard block sticking
to the braid and this accounts for the rapid increase of modu-
0 100 200 300
Temperature I°Cl - 400

lus. After 210"C, the modulus tends to be constant again.
Figure 3(b) illustrates DSC and TGA diagrams of NC. From
the TGA curve we can see that NC begins to decompose and
loses weight at 185°C and at 192"C, a decomposition peak
can be observed in the DSC diagram. Therefore, the results of
the above techniques are in good agreement.
The glass transition of NC is close to the decomposition
temperature. There is no rubber elastic region and flow regi-
on but the decomposition and crosslinking is followed right
after the glass transition. This is different from other cellulose
polymers. For example, Figures 4(a) and (b) show the TBA
and DSC diagrams of triacetate cellulose. After the glass
transition at 195"C, a melting peak is observed at 290°C. It
begins to decompose and to crosslink at 328°C with the
modulus increasing to a maximum at 335°C .
There is an inconsistency in the reported values of the
glass transition temperature of NC as mentioned above.
Nielson(') gave a value of 53°C and another literature(*)
reported to be 120°C, while G.A. S ~ r o k i n (et ~ )al. proposed
that the glass transition temperature of NC is higher than its
decomposition temperature.
In our opinion, at least two factors account for the difficul-
ty in the Tg measurement of NC. Firstly since the glass tran-
sition temperature of NC is quite close to its decomposition
temperature, decomposition phenomena tend to cover the
glass transition in less sensitive experimental techniques. In
DSC test, since the exothermic peak during decomposition of
NC is rather large, but the glass transition should give a weak
endothermic peak and thus be covered by the decomposition.
Secondary, since pure NC is easy to decompose and thus is
difficult to be plasticized and molded into a pure NC sample.
Figure 4. Diagrams of cellulose triacetate: (a) torsional braid analysis
(TBA), (b) differential scanning calorimetry (DSC).
However, the solvent is difficult to be removed completely
from the relative large solution-molded samples. The bigger
the sample, the more difficult is the heat to be dispersed.
Thus the sample would decompose at relating low temperatu-
re and the glass transition would be covered.
In TBA test, it is unnecessary to fabricate geometrical spe-
cimens and only a tiny amount of sample (a few milligrams)
is impregnated on the braid. This TBA technique is rather
sensitive to distinguish Tg from T, of NC, though they are
quite close.
5. References
( I ) L.E. Nielson, "Mechanical Properties of Polymers", Reinhold
Publishing Corporation, New York, 1962.
( 2 ) DBP 2853388.
(3) G.A. Sorokin, I.V. Tigunin, and E.N. Fominych, Vysokomol. Soedin.
Ser. B , 11(7),522-525(1969).
Appendix
Calculation ofthe activation energy for glass transition
It is possible to calculate the activation energy for glass
transition from the half width of the damping peak obtained
from TBA test based on the standard linear solid dynamic
model.
Propellants, Explosives, Pyrotechnics 17, 34-37 (1992) Glass Transition Temperature Measurement of NC by Torsional Braid 37

The loss factor Q-I from this model is given by, Rearrangement of Eq. (8) gives,

Q-’ = tan d = G”/G’ = 24.’ [wt/(1+w2t2)] 1+w2t;,,- 4 Wt,, =0 (9)

(6)
Q-’ reflects the extent of energy dissipation. The solution of Eq. (9) is,
The relaxation process of polymers can be expressed by the Wto.5 = 2+ % (10)
activation energy of this relaxation process,
and the reasonable solution is taken as,
t = toeE/RT (7)
Wt,., = 2 + 6- (1 1)
substitution of Eq. (6) into Eq. (5) gives,
since wtmax= 1 at Tmax,the relaxation time at T, equals the
Q = 2 Q-lax[ ( ~ t , e ~ ~ ~ ) / ( l + w ~ t ~ e ~ ~ ~ ~ time
) ] of applied force. Thus, combining Eqs. (1 1 ) and (12)
Assuming that the temperature at the maximum damping is with Eq. (7), we obtain
Tmax(i.e. TJ, and that the temperature correspondenting to (2 + K)/(toeEmTo.s)= l/(toeEmTms) (13)
the half damping peak, is T,,, thus the loss factor at Tmaxis
Rearrangement of Eq. (13) gives
Q-kaX and at To,,, 1/2Q-kax.
E = R In ( 2 + K)/(ino,,
- irm,) (14)
When Q-km=2 Q-kaXthen
Therefore the activation energy can be calculated from this
5 )~/ (~ t ~ e ~ ~(8)
(1/2) Q-kax= 2 Q-kax[ ( ~ t , e ~ ~ ~ ol, + ~ o . equation.
S)]

(Received October 27, 1990; Ms 48/90)