The Thermodynamic Properties of

Polytetrafluoroethylene
SUK FA1 LAU,* HIDEMATSU SUZUKI,? and BERNHARD
WUNDERLICH, Department of Chemistry, Rensselaer Polytechnic
Institute, Troy,New York 12181

Synopsis
Polytetrafluoroethylenes of different crystallinity were analyzed between 220 and 700 K by dif-
ferential scanning calorimetry. A new computer coupling of the standard DSC is described. The
measured heat capacity data were combined with all literature data into a recommended set of
thermodynamic properties for the crystalline polymer and a preliminary set for the amorphous
polymer (heat capacity, enthalpy, entropy, and Gibbs energy; range 0-700 K). The crystal heat
capacities have been linked to the vibrational spectrum with a 0 3 of 54 K, and 01 of 250 K, and a
full set of group vibrations. C , to C, conversion was possible with a Nernst-Lindemann constant
of A = 1.6 X 10-3 mol K/J. The glass transition was identified as a broad transition between 160
and 240 K with a AC, of 9.4 J/K mol. The room-temperature transitions at 292 and 303 K have
a combined heat of transition of 850 J/mol and an entropy of transition of 2.90 J/K mol. The equi-
librium melting temperature is 605 K with transition enthalpy and entropy of 4.10 kJ/mol and 6.78
J/K mol, respectively. The high-temperature crystal form I is shown t o be a condis crystal (con-
formationally disordered), and for the samples discussed, the crystallinity model holds.

INTRODUCTION
Polytetrafluoroethylene (PTFE) is one of the three simplest macromolecular
model compounds, polyethylene and selenium being the other two. To provide
information on thermodynamic properties for linear macromolecules ATHAS,
an advanced thermal analysis laboratory, was estab1ished.l The first result of
the ATHAS effort was a data bank on heat capacities, updated to the year 1980.2
The next step is to establish full sets of macroscopic thermodynamic properties
and link these, as much as possible, to the microscopicproperties of the molecules.
This work has been completed for the other two simplest macromolecules
polyethylene3 and selenium: and work is under way on polypropylene5 and
several polyoxides.6 In this paper a report on PTFE is given.
When collecting information for the data bank it became clear that, although
several measurements of heat capacity over a wide temperature range had been
published for PTFE, the heat capacity curve was not understood well enough
to arrive at a recommended set of heat capacities at that time. In this paper new
measurements by differential scanning calorimetry (DSC) on a series of PTFE
sample of widely varying crystallinity between 220 and 700 K are reported to
clarify the heat capacity problem. In addition, the heat of fusion, about which
, ~ been redetermined (4.10 kJ/mol). Simi-
some doubt was raised r e ~ e n t l yhas

* Present address: Research Center, Hercules Inc., Wilmington, DE 19899.

On leave from Kyoto University.

Journal of Polymer Science: Polymer Physics Edition, Vol. 22,379405 (1984)

0 1984 John Wiley & Sons, Inc. CCC 0098-1273/84/030379-27$04.00
380 LAU, SUZUKI, AND WUNDERLICH

larly, the equilibrium melting temperature of as-polymerized PTFE, which is

obscured by superheating? has been measured (605 K). The heat of transition
at room temperature (0.85 kJ/mol) will be shown to be strictly proportional to
crystallinity for high-crystallinity PTFE, a fact which could only be established
by careful baseline evaluation based on heat capacity calculation. The glass
transition region (160-240 K, AC, = 9.4 J/K mol at 200 K) was described earlier
using special low-temperature DSC.g The glass transition data are in full accord
with the information presented here.
After the recommended data for the heat capacity of crystalline PTFE had
been established, they were linked to the vibrational frequency spectrum. The
two low-frequency (skeletal) modes are fitted to low-temperature heat capacity
data in a two-parameter ( 0 3 and 01)Tarasov fit.1° The seven high-frequency
(group) vibrations are taken from normal-mode calculations based on Raman
and infrared data. Earlier interpretations from our laboratorylo and by Choy
et al.ll in the low-temperature region from 0 to 200 K are largely in agreement
with the present analysis, which covers the much expanded temperature range
from 0 to 700 K and is based on more extensive experimental information. The
calculated data, which represent the heat capacity C , at constant volume, are
linked to the measured heat capacity C, at constant pressure, and all thermo-
dynamic data are calculated. For amorphous PTFE, melt data and crystallinity
extrapolations are used to establish an estimate of the thermodynamic data, also
from 0 to 700 K.
A final discussion of the thermodynamic state of crystalline PTFE between
the room-temperature transition and the melting temperature (crystal form I)
leads to the conclusion that this crystal is best called a condis (conformationally
disordered) crystal. A full discussion of the condis crystal as one of six possible
mesophases is given in ref. 12.
The work presented here is the central part of the doctoral thesis of one of us
(S.F.L.).l3

EXPERIMENTAL

Samples
Eleven different PTFE samples have been kindly donated at various times
by researchers from the DuPont Research Laboratories (Polymer Products and
Central Research Departments). They are characterized as follows.
Sample HV is a virgin (as-polymerized) granular PTFE of high molecular
weight (estimated to be >lo7) (No. B-9707). This sample was also analyzed
earlier as to its superheating on melting (sample H).8 In addition, four similar
virgin PTFE samples (Nos. E-30845-72-1 to 4) selected for their high heat of
fusion were kindly supplied by Dr. H. W. Starkweather (DuPont Central Re-
search Dept.). They were used only for a check on the heat-of-fusion determi-
nation. In this paper these latter four samples are labeled HV-A, -B, -C, and
-D. The samples were used as received and are taken as our reference polymers.
They are assumed to be 100%crystalline. This assumption is based on the almost
complete absence of amorphous x-ray diffraction14and on infrared15and NMR
evidence.16 Often quoted crystallinity estimates range from 95 to 100%(but see
also recent data discussed below which suggest crystallinities as low as 84%). For
heat capacity measurements above 10 K, variations in crystallinity by 5% are
 THERMODYNAMIC PROPERTIES OF PTFE 381

outside the error limit, which overall is about 3%. The difference between
amorphous and crystalline PTFE heat capacity is, except for the lowest tem-
peratures, less than 20%, so a 5%error in crystallinity results in a 1.0%error in
heat capacity. The heat of fusion data derived here are, on the basis of this as-
sumption, probably too small by several percent. Again, this hardly reaches the
usual error limit for polymeric materials. The recently proposed high heat of
fusion of PTFE7 is larger by 24%, and an often used handbook value17 is lower
by 23% than the value determined here.
Sample HQ is chemically identical to sample HV, but was first kept molten
for 15 min a t 625 K and then cooled at 320 K/min to the temperature of mea-
surement. The morphology oficrystals grown from the melt of this sample was
discussed earlier.18 From the heat of fusion, a crystallinity of 33.2% can be de-
rived for this sample.
Sample MQ is a dispersion polymer of about 106 molecular weight (No. B-
9708). The sample was also studied earlier.8J8 Its pretreatment was 15 min
in the molten state at 625 K followed by cooling at 320 k/min to the temperature
of measurement. The crystallinity from the heat of fusion is 43.0%. A virgin
sample of the same polymer (MV) was also used for heat capacity measurement
in the temperature region 390-480 K, but showed little difference from sample
HV except in the melting range.
Sample LQ is similar to sample MQ, but of lower molecular weight (estimated
to be about 5 X lo5) (No. B-9709). The crystallinity from its heat of fusion is
66.2%. Again, a virgin sample LV was also used for the heat capacity measure-
ment in the temperature region 390-480 K, but showed little difference from
sample HV except in the melting range.
Sample TlOOV is a poly(tetrafluoroethy1ene-co-hexafluoropropylene) (Teflon
100 FEP resin). It contains 12 f 2 w t % hexafluoropropylene (or 4 CF3 branches
per 100 carbon backbone atoms). The sample was used as received. Its crys-
tallinity from the heat of fusion measurement is 17.1%. In the temperature re-
gion of interest (220-700 K) the minor difference in chemical structure is insig-
nificant for heat capacity calculations when judged by comparison with poly-
ethylene and polypropylene (crystalline polypropylene has a t 220 and 400 K
specific heat capacities different from the values for polyethylene by +3 and -4%,
and amorphous polypropylene at 500 K has the same specific heat capacity as
p ~ l y e t h y l e n e ) . ~Samples
,~ of identical composition (T11OV and T160V) but
lower molecular weight had practically identical heat capacities up to the melting
range. Between 220 and 470 K the heat capacities of these two additional
samples were averaged with those of sample T100V.
Sample 9704V is branched PTFE with a perfluoroether group of 2-5 carbon
atoms per branch (Teflon PFA TE-9704). The number of branches is about 1.4
per 100 backbone carbon atoms. The sample was'used as received. The mea-
surements were used only for qualitative discussion of the room-temperature
transition because of the uncertainty in the chemical structure. Its heat capacity
data are shown in Figure 3 (curve 111).

Instrumentation
A Perkin-Elmer DSC-2 was used for all measurements. The HewletcPackard
calculator previously used with this instrument, described in refs. 19 and 20, was
replaced with a LMS-1000-DSC data station.21 A block diagram of this instal-
382 LAU, SUZUKI, AND WUNDERLICH

Perkin-Elmer
Perkin-Elmer Perkin-Elmer
Refrigeration j 3
DSC-2 Recorder-56
Unit

LMS-1000
Computer System
1 n <

Digital CRT
6 Dual Disc Drive

Tektronix 4010-1 Digitalflecwriter IV

Printer
Regent 25 Graphics Terminal (LA 34-AA)
I

Houston Instrument
> Digital Plotter c
DMP-7

lation is shown in Figure 1. The data station consists of a Z-80-based micro-

computer with 64 kbytes of random access memory and dual 8-in. floppy disk
drives. A multichannel 16-bit analog to digital converter with differential inputs
is used for data assimilation. A Tektronix 4010 graphic terminal provides screen
graphics. The digital plotter gives hard-copy graphics. An ADD-CRT terminal
is available for the input of constants and run parameters. The DigitalDecwriter
provides tabular data output.
Programs for operation and computation are written in FORTRAN IV, while
the data collection program is in 2-80 assembly language. Data collection is
interrupt driven, using the DSC programming pulse. The computer starts the
run and data collection simultaneously,to make the sequence number of the data
input agree with the temperature. The digital input is of 16-bit resolution with
programmable gains of 1,10,100, and 500. The input is usually set to correspond
to full-scale displacement of the DSC recorder. At most, there are at present
30 data points recorded per kelvin. The overall scheme of data generation is
shown in Figure 2 and is largely self-explanatory.
 THERMODYNAMIC PROPERTIES OF PTFE 383

0 Start

.L
Data- * < DS C COLL>
Collecting
Notes :
* Datafile Generated
<xxx> Name i f Coaputer
Program t o be
used t o Generate
Result

(Averaged)
k
Calculation

w
I Data i n Heat1

TI*- -*ion

MELTING
m.p. POINT
I 1 - I

0 (a)
Fig. 2. Scheme of the generation of final data from the ATHAS computer-interfaced differential
scanning calorimeter. (Continued on next 3 pages.)

Amplitude Measurements
The heat capacity and heat of fusion measurements were done on samples of
10-30 mg size to give maximum amplitudes. The samples were sealed in stan-
dard aluminum pans of comparable mass to the pans used for baseline and ref-
erence scans ( f O . l mg). Despite the fact that differences in sample pan weight
can be corrected for with the known heat capacity of aluminum, it was found
earlier that better precision is made possible by matching masses instead of
calculation of corrections. For a typical 30-mg PTFE sample, a 1-mg difference
384 LAU, SUZUKI, AND WUNDERLICH

0
Print
Degree of

Use

I <UCALC>
Use option:
PRINT

,
yes PARTIAL

Prln t

+
P a r t i a l Area
I
I yes : < s P H E ~ ; ~ ~ ~
Use
Plot F i l e option:

Cp Data a t

Point i n

Locate
Abnormal Point
<MANDATA>

Point

Fig. 2 (Continued on next 2 pages.)

in A1 mass causes a calculated error of 2.8% at 450 K. The overall precision of

the instrument is, sample permitting, as high as f0.5%.19
All runs were performed at 10 K/min heating rate in a nitrogen atmosphere.
The temperature ranges were covered in steps of no more than 100 K to minimize
errors due to baseline curvature. The differential power amplitude was cali-
brated with aluminum oxide (sapphire). Data were sampled every 0.2 s and the
recording sensitivity was 2 mJ/s. A t least two measurements over the whole
temperature range were made for each sample. Average data points are used
for the discussion. For the computation sequence see Figure 2.
The heats of transition were sometimes found to cover rather broad temper-
 c
THERMODYNAMIC PROPERTIES OF PTFE 385

-+
no

-, Smothin

Smoothed Av
Cp data
*
option:
SMOOTH

yes <DSCLIST>

Av Cp Table
A

(C)

Fig. 2 (Continued on next page.)

ature ranges. In these cases the accuracy of the measurement rests with the
assignment of the proper baseline. From Figures 4 and 6 below it is obvious that
the room-temperature transition of HV reaches, for example, from 172 to 396
K and its melting range from about 470 to 645 K. In such cases a rather careful
baseline separation is necessary. The equation for the separate amplitudes to
be summed to give the heat of fusion is
hn = Cp, + ( m ~ f , / m ~ f ) ( C p-c ,C p a , ) - Cpc, (1)
where h, is the heat of fusion amplitude at n, corrected for the heat capacity
386 LAU, SUZUKI, AND WUNDERLICH

no
<DSCPLOT>
Use
option: 3
yes <DSCPLOT>
Use o p t i o n : 1
V V I

Hard Visual
COPY Graphic
output

V &

baseline (J/K mol); n is the running number, the width of the integration interval
being defined by the temperature difference Tn - Tn-l;C,, is the overall ex-
perimental heat capacity at n, including heat capacity and heat of fusion effects;
AHf, is the sum of all h up to n; and AHf is the total heat of fusion, evaluated
from
I f
Mf = - C (hn + h n - d ( T n - T n - 1 ) (2)
2i
where i and f are the initial and final values of n. AHf,IAHf represents the
relative crystallinity decrease up to the point in question, so that at the beginning
of melting the subtracted baseline is C,, and at the melt end it is C,, and the
baseline is properly apportioned between the two limits. Since AHf is not known
initially, several iterations until a constant value is obtained are necessary, using
initially for AHf the area over a linear baseline. For samples partially crystalline
below the melting range, the appropriate semicrystalline heat capacity replaces
C,,. Between three and nine iterations were needed to reach constant vaues
of AHfi The integration intervals were 0.2 K.
The room-temperature transition was analyzed under the assumption that
the heat capacity of the crystal changes only insignificantly from the calculated
value based on the vibrational frequency spectrum (curve I in Fig. 4). This leads
to a transition enthalpy

mt = 21 n=c
f
E,[ ( C p , - Cpc,) + CCp,-, - C p c n - l ) l ( T n - T n - 1 ) (3)

where all heat capacities are defined as for eq. (1). For HV this analysis decreases
AHt by 20% over a straight point-to-point baseline (see Fig. 4,dotted line). A
baseline chosen without concern about the crystalline heat capacity may lead,
in contrast, to a substantially smaller AHt.
 THERMODYNAMIC PROPERTIES OF PTFE 387

15.

75.

35.

5.0

310 410 510 610

Temperature (K)

Fig. 3. Heat capacities of 4 PTFEs over the temperature range 220-700 K: (I) sample HV (100%
crystalline), (11) Sample HQ (33.2%crystalline), (111) sample 9704V, (IV) sample TlOOV (17.1%
crystalline).

Temperature Measurements
The temperature scale of the DSC was calibrated with five sharp melting
standards, with lags corrected if necessary.22 Quoted temperatures should be
accurate to f0.5 K.
Besides instrument lags, there are considerable lags in the sample due to su-
perheating.s The melting peak of sample HV reaches the steady state owing
to the heat capacity of the melt only at about 650 K. This is far beyond the
388 LAU, SUZUKI, AND WUNDERLICH

x
U
d
0
4
0
V
50.0
U

Temperature (K)

Fig. 4. Heat capacities of three PTFEs in the low-temperature range 160-410K (I) recommended
heat capacity for crystalline PTFE, (11) sample HV (100% crystalline), (111) sample HQ (33.2%
crystalline), (IV) sample TlOOV (17.1%crystalline, but no room-temperature transition).

equilibrium melting temperature and the expected instrument lag. To get an

estimate of the equilibrium melting temperature, a series of HV samples were
kept for 12 h a t a predetermined temperature in the melting range. Each sample
was then quickly cooled to 595 K, a temperature which is too high to induce
crystallization, but low enough to enable the steady state to be established for
DSC measurements from 600 K on. Crystals not melted after the 12-h heating
gave melting peaks in the usual high-temperature range (peak a t about 620 K).
Keeping the sample above 604.7 f 0.1 K completely eliminated any fusion on
subsequent heating from 595 K. Thus we use 605 K for all calculations involving
 THERMODYNAMIC PROPERTIES OF PTFE 389

t 1

0.2 0.4 0.6 0.8

C r y s t a l l i n i t y (wc)

Fig. 5. Heat capacity of PTFE as a function of crystallinity between 360 and 480 K. Sample HV
100, LQ 66.2, MQ 43.0, HQ 33.2, Tl00V 17.1%crystalline. The data at 0% crystallinityrefer to data
extrapolated linearly from the melt, indicating the need of correction below 480 K [see eq. (4)].

the equilibrium melting temperature. This temperature is most likely within

5 K of the true equilibrium melting temperature.

RESULTS
The heat capacities of four of the six samples are displayed in Figure 3. The
marked points are the properly averaged data points a t the data bank standard
intervals (10 K). The curves are fitted to these points. For all computations,
however, the actual full set of data was used. Figure 4 shows some of the same
390 LAU, SUZUKI, AND WUNDERLICH

70.0

60.0

h
4
I
u
rl
I
4
0
-
9

h
U
50.0
.-I
u
4
n
V

40.0

30.0

2 40 36 0 480 600

Temperature (K)

Fig. 6. Heat capacity of PTFE in the liquid state: (I) sample T100,17.1% crystalline; (11) sample
HQ, 33.2%crystalline; (111) recommended crystalline heat capacity, for reference; (IV) linearly ex-
trapolated melt data; (V) modified melt data, eq. (4’); (VI) adjusted data of ref. 9 on sample
T100V.

data amplified for discussion of the room-temperature transition. Below 240

K the heat capacities of the less crystalline samples start to approach the heat
capacity of the crystalline sample (and of the recommended data curve also
shown). This is the upper end of the glass transition, as verified by low-tem-
perature DSC9 (see also Fig. 6).
The size of the two endothermic transitions is proportional to the weight-
fraction crystallinity, except for the two copolymers. Sample TlOOV shows no
low-temperature transition (Figs. 3 and 4), while sample 9704V shows a small
transition at lower temperature (Fig. 3).
 THERMODYNAMIC PROPERTIES OF PTFE 391

TABLE I
Heat of Fusion of Polytetrafluoroethylene (Samples of High Molecular Weight, Considered Close
To Completely Crystalline)
Tbeginninga Tpeaka Tenda Number of
Sample (K) (K) (K) (kJ/mol) measurements
HV-A 602 612 618 620 4.21 4
HV-R 603 611 ... 617 4.11 5
HV-C 601 610 617 620 4.09 7
HV-D 614 619 ... 620 4.12 6
HV 605 613 617 623 3.89 3
Averageb 4.10 f 0.09
a These temperatures approximately characterize the melting peak (or double peak). Heats of
fusion were obtained by integration from 480 to 645 K.
Weighted average according to number of measurements.

The heat capacities between the two endotherms decrease with increasing
crystallinity. Figure 5 shows the heat capacities as functions of crystallinity,
calculated from the heats of fusion.
All samples show practically the same heat capacity after fusion. This can
be seen from Figures 3 and 6. Curve IV in Fig. 6 is drawn after a linear fit of all
measured melt data between 560 and 700 K. The equation is
C,, = 49.45 + 0.0332T (4)
The heat capacity is given in joules per kelvin-mole. The standard deviation
of eq. (4) is 0.2%.
The heat of fusion data on the samples HV are listed in Table I.

RECOMMENDED DATA
The data survey of PTFE heat capacity measurements (ref. 2, Part VII) was
extended to the end of 1982. Apart from the present data only one new mea-
surement of heat capacity of PTFE has been reported,23 but the results do not
meet the standards4 of acceptable data since no sample characterization was
given. (Data from 3 to 20 K, 10%higher a t 10 K, 9% lower at 15 K than the rec-
ommended data below.)
Table I1 shows the full set of recommended data for crystalline PTFE. It was
derived as follows. From 0.3 to 4 K, data by Salinger and Cieloszyk (sample 98
in ref. 2) are recommended. From 5 to 100 K data by Choy et al. (sample 99 in
ref. 2) are chosen.' To these are added from 30 to 100 K the data of Furukawa
et al. (Sample 90 in ref. 2). The two data sets agree in the region of overlap within
a standard deviation of 0.7%. The data of Furukawa et al., are used alone up to
170 K. Above 170 K none of the data of ref. 2 are recommended without modi-
fication because of the complications from the two endotherms. The data listed
above 170 K are those derived using the calculations outlined in the section on
computation below.
The heat capacity of sample HV, which is considered 100%crystalline, shows
an internal standard deviation of 0.3%(average of two to four measurements)
for the entire temperature range. Between 390 and 480 K, eight measurements,
including also samples MV and LV, show a standard deviation of 0.1%. The data
392 LAU, SUZUKI, AND WUNDERLICH

TABLE I1
Recommended Thermodynamic Data of Crystalline Polytetrafluoroethylene
CP H$ - Hg sc, -(G$ - H $
T (K) (J mol-l K-l) (J mol-') (J mol-l K-l) (J mol-')
0.0 0.0 0.0 0.0 0.0
0.3 0.0000649 0.00000974 0.00003246 0.0
0.4 0.0001249 0.00001923 0.00005889 0.0000043
0.5 0.0002226 0.00003660 0.00009676 0.00001 18
0.6 0.0003704 0.00006625 0.0001499 0.0000237
0.7 0.0005816 0.0001139 0.0002223 0.0000418
0.8 0.0008707 0.0001865 0.0003183 0.0000681
0.9 0.001252 0.0002926 0.0004422 0.0001054
1.0 0.001740 0.0004422 0.0005988 0.0001566
1.2 0.003094 0.0009256 0.0010306 0.0003111
1.4 0.005041 0.001739 0.001649 0.0005688
1.6 0.007687 0.003012 0.002489 0.0009706
1.8 0.01111 0.004892 0.003587 0.001564
2.0 0.01538 0.007541 0.004973 0.002405
3.0 0.05071 0.0 4059 0.01727 0.01122
4.0 0.1086 0.1202 0.03930 0.03694
5.0 0.2284 0.2887 0.0757 0.08982
10.0 1.228 3.930 0.4969 1.039
15.0 2.565 13.41 1.231 5.059
20.0 3.915 29.61 2.148 13.35
25.0 5.155 52.29 3.153 26.54
30.0 6.301 80.93 4.194 44.89
40.0 8.441 154.6 6.299 97.32
50.0 10.31 248.4 8.385 170.9
60.0 12.27 361.3 10.44 265.0
70.0 14.17 493.5 12.47 379.6
80.0 16.02 644.4 14.49 514.5
90.0 17.75 813.3 16.47 669.4
100.0 19.37 998.9 18.43 844.0
110.0 21.20 1202 20.36 1038
120.0 22.90 1422 22.28 1251
130.0 24.53 1659 24.18 1484
140.0 26.13 1913 26.05 1735
150.0 27.65 2182 27.91 2005
160.0 29.13 2465 29.74 2293
170.0 30.62 2764 31.55 2599
180.0 31.98 3077 33.34 2924
190.0 33.24 3403 35.10 3266
200.0 34.47 3742 36.84 3626
210.0 35.67 4093 38.55 4003
220.0 36.82 4455 40.24 4397
230.0 37.94 4829 41.90 4808
240.0 39.03 5214 43.54 5235
250.0 40.10 5609 45.15 5678
260.0 41.13 6015 46.74 6138
270.0 42.15 6432 48.32 6613
273.15 42.47 6565 , 48.81 6766
280.0 43.15 6858 49.87 7104
290.0 44.13 7295 51.40 7611
292.0 (T,) 44.32 7383 51.70 7714
292.0 44.32 8097 54.15 7714
298.16 44.91 8372 55.08 8050
300.0 45.09 8455 55.36 8152
303.0 (T,) 45.37 8591 55.81 8318
 THERMODYNAMIC PROPERTIES OF PTFE 393

TABLE I1 (continued)
303.0 45.37 8727 56.25 8318
310.0 46.02 9047 57.30 8716
320.0 46.95 9511 58.77 9296
330.0 47.86 9985 60.23 9891
340.0 48.76 10,469 61.68 10,501
350.0 49.63 10,960 63.10 11,125
360.0 50.49 11,461 64.51 11,763
370.0 51.35 11,970 65.91 12,415
380.0 52.17 12,488 67.29 13,081
390.0 52.99 13,014 68.65 13,761
400.0 53.81 13,548 70.00 14,454
410.0 54.60 14,090 71.34 15,161
420.0 55.39 14,640 72.67 15,881
430.0 56.15 15,197 73.98 16,614
440.0 56.91 15,763 75.28 17,361
450.0 57.65 16,335 76.57 18,120
460.0 58.39 16,916 77.84 18,892
470.0 59.10 17,503 79.11 19,677
480.0 59.82 18,098 80.36 20,474
490.0 60.51 18,699 81.60 21,284
500.0 61.20 19,308 82.83 22,106
510.0 61.88 19,923 84.05 22,940
520.0 62.55 20,545 85.25 23,787
530.0 63.20 21,174 86.45 24,646
540.0 63.86 21,810 87.64 25,516
550.0 64.48 22,451 88.82 26,398
560.0 65.12 23,099 89.99 27,292
570.0 65.75 23,754 91.14 28,198
580.0 66.36 24,414 92.29 29,115
590.0 66.96 25,081 93.43 30,044
600.0 67.58 25,753 94.56 30,984
605.0 (T,) 67.88 26,092 95.12 31,458
610.0 68.18 26,432 95.68 31,935
620.0 68.77 27,117 96.80 32,898
630.0 69.35 27,808 97.90 33,871
640.0 69.93 28,504 99.00 34,856
650.0 70.50 29,206 100.1 35,851
660.0 71.06 29,914 101.2 36,857
670.0 71.64 30,627 102.2 37,875
680.0 72.19 31,347 103.3 38,902
690.0 72.75 32,073 104.4 39,941
700.0 73.30 32.802 105.4 40,990

are also in good agreement with the data of Furukawa et al. (sample 90 in ref. 2)
+
(maximum discrepancy 0.7%),while the data from 340 to 560 K agree with the
results of Douglas and Harman (sample 94 in ref. 2) (maximum discrepancy
-0.9%).
The thermodynamic functions in Table I1 have been calculated from the listed
heat capacities by point-to-point integration (trapezoidal rule). A t 292 K, 714.1
J/mol of the room-temperature transition is added to the enthalpy, and 2.45 J/K
mol is added to the entropy. The subdivision of the room-temperature transition
was done using the percentages established by Quinn et al.24from dilatometry
and by Marx and Dole25from adiabatic calorimetry (ratio of the transition heats
and volumes 84:16). A t 303 K, 136.0 J/mol and 0.45 J/K mol are accordingly
added to the enthalpy and entropy, respectively. The melting enthalpy and
394 LAU, SUZUKI, AND WUNDERLICH

entropy are added at the newly established melting temperature of 605 K (4100
J/mol and 6.78 J/K mol, respectively).
The present data on molten PTFE are more extensive than the literature data
of Douglas and Harman (sample 96 in ref. 2). They are 2.5% higher at 640 K and
1.4% lower at 700 K [see Figs. 3 and 6 and eq. (4)]. It is quite clear that an ex-
trapolation to lower temperature is not possible. Figure 5 shows the extrapolated
points of the melt heat capacity at the 0% crystallinity ordinate. Obviously the
heat capacity of the supercooled melt must decrease. The crystallinity-ex-
trapolated points are then used, instead, to derive the following heat capacity
equation (J/K mol):
C,, = 29.51 + 6.917 - 10-2T + 1.292 - 10-5T2 (4’)
Table I11 contains the measured melt heat capacities from 700 K down to 480

TABLE 111
Thermodynamic Data of Amorphous PTFE
CP H$ - Hf, S$ G $ - Hk
T (K) (J/K mol) (J/mol) (J/K mol) (J/mol)
0.0 0.0 2809 3.38 2809
0.3 (0.00006491) 2809 3.38 2808
0.4 (0.0001249) 2809 3.38 2808
0.5 (0.0002226) 2809 3.38 2807
0.6 (0.0003704) 2809 3.38 2807
0.7 (0.0005816) 2809 3.38 2807
0.8 (0.0008707) 2809 3.38 2806
0.9 (0.001252) 2809 3.38 2806
1.0 (0.001740) 2809 3.38 2806
1.2 (0.003094) 2809 3.38 2805
1.4 (0.005041) 2809 3.38 2804
1.6 (0.007687) 2809 3.39 2804
1.8 (0.01111) 2809 3.39 2803
2.0 (0.01538) 2809 3.39 2802
3.0 (0.05071) 2809 3.40 2799
4.0 (0.1086) 2809 3.42 2795
5.0 (0.2284) 2809 3.46 2792
10.0 (1.228) 2813 3.88 2774
15.0 (2.565) 2822 4.61 2753
20.0 (3.915) 2839 5.53 2728
25.0 (5.155) 2861 6.54 2698
30.0 (6.301) 2890 7.58 2663
40.0 (8.441) 2964 9.68 2576
50.0 (10.31) 3057 11.77 2469
60.0 12.27 3170 13.82 2341
70.0 14.17 3303 15.86 2193
80.0 16.02 3453 17.87 2024
90.0 17.75 3622 19.86 1835
100.0 19.37 3808 21.81 1627
110.0 21.20 4011 23.74 1399
120.0 22.90 4231 25.66 1152
130.0 24.53 4468 27.56 886
140.0 26.13 4722 29.44 601
150.0 27.65 4991 31.29 297
160.0 29.13 5275 33.12 -25
 THERMODYNAMIC PROPERTIES OF PTFE 395

TABLE I11 (continued)

170.0 30.62 5573 34.93 -365
180.0 (32.96) 5891 36.75 -724
190.0 (35.28) 6232 38.59 -1100
200.0 (37.59) 6597 40.46 -1496
210.0 (39.91) 6984 42.35 -1910
220.0 (42.22) 7395 44.26 -2343
230.0 (44.58) 7829 46.19 -2795
240.0 46.85 8286 48.14 -3267
250.0 47.61 8758 50.06 -3758
260.0 48.36 9238 51.95 -4268
270.0 49.12 9726 53.79 -4797
273.15 49.36 9881 54.36 -4967
280.0 49.89 10,221 55.59 -5344
290.0 50.65 10,723 57.35 -5908
298.15 51.28 11,139 58.76 -6382
300.0 51.42 11,234 59.08 -6496
310.0 52.19 11,752 60.78 -7090
320.0 52.96 12,277 62.45 -7706
330.0 53.74 12,811 64.09 -8339
340.0 54.52 13,352 65.71 -8988
350.0 55.30 13,901 67.30 -9653
360.0 56.08 14,458 68.87 -10,334
370.0 56.87 15,023 70.41 -1 1,030
380.0 57.66 15,596 71.94 -11,742
390.0 58.45 16,176 73.45 -12,469
400.0 59.24 16,765 74.94 -13,211
410.0 60.04 17,361 76.41 -13,968
420.0 60.84 17,966 77.87 -14,739
430.0 61.64 18,578 79.31 -15,525
440.0 62.44 19,198 80.74 -16,326
450.0 63.25 19,827 82.15 -17,140
460.0 64.06 20,463 83.55 -17,969
470.0 64.87 21,108 84.93 -18,811
480.0 65.39 21,759 86.31 -19,667
490.0 65.72 22,415 87.66 -20,537
500.0 66.05 23,074 88.99 -21,420
510.0 66.38 23,736 90.30 -22,317
520.0 66.72 24,401 91.59 -23,226
530.0 67.05 25,070 92.87 -24,149
540.0 67.38 25,742 94.12 -25,084
550.0 67.71 26,418 95.36 -26,031
560.0 68.04 27,096 96.58 -26,991
570.0 68.38 27,779 97.79 -27,963
580.0 68.71 28,464 98.98 -28,947
590.0 69.04 29,153 100.16 -29,943
600.0 69.37 29,845 101.3 -30,950
605.0 (T,) 69.54 30,192 101.9 -31,458
610.0 69.70 30.540 102.5 -31,969
620.0 70.04 31,239 103.6 -33,000
630.0 70.37 31,941 104.7 -34,041
640.0 70.70 32,646 105.8 -35,094
650.0 71.03 33,355 106.9 -36,158
660.0 71.36 34,067 108.0 -37,233
670.0 71.70 34,782 109.1 -38,319
680.0 72.03 35,501 110.2 -39,415
690.0 72.36 36,223 111.2 -40,522
700.0 72.69 36,948 112.3 -41,640
396 LAU, SUZUKI, AND WUNDERLICH

400

300

E(K) 200

100

40 80 120 160

Temperature (K)

Fig. 7. Characteristic 8 temperatures of crystalline PTFE: (I) Debye three-dimensional 8~

Temperature using full C, data; (11) Tarasov 81 temperature, group vibrations subtracted; (111)
Tarasov 8 3 temperature, group vibrations subtracted; (IV) Tarasov 81 temperature, using full C,
data; (V) Tarasov 8 s temperature, using full C, data.

K from eq. (4)and data calculated with eq. (4’)down to 240 K. Below 240 K the
glass transition startsg and a practically linear connection is made to the crys-
talline heat capacity a t 160 K. Curve VI of Figure 6 shows that this is a rea-
sonable estimate of the amorphous heat capacity. The lower-temperature data
are assumed to be identical to those for the crystal. For practically all linear
macromolecules2glassy and crystalline heat capacities are almost identical except
for temperatures below 50 K, where crystalline samples have lower heat capac-
ities. The low-temperature heat capacities are, however, usually so small that
they change the other thermodynamic quantities only little. The uncertain data
in the glass transition region and the too high C, of the crystalline PTFE below
50 K are placed in parentheses in Table 111. By integration and fitting of the
enthalpy, entropy, and Gibbs free energy a t 605 K, the preliminary data of
thermodynamic properties listed in Table I11were derived. Of particular interest
is the entropy S i of glassy PTFE a t 0 K, namely 3.38 J/K mol. This value is to
be compared to S t of polyethylene (2.59 J/K mo1),3 and S t of selenium (3.63 J/K
m01).4

COMPUTATIONS
With the newly generated recommended heat capacity data for the crystal,
it is possible to try to interpret the heat capacities in terms of the frequency
spectrum. Recently a full set of inversion routines of heat capacities to Debye
and Tarasov 8 temperatures has been programmed for the computer, making
use of the Clenshaw-Curtis double self-adapting integration strategy.26
 THERMODYNAMIC PROPERTIES OF PTFE 397

TABLE IV
Group Vibration Frequencies of Polytetrafluoroethylene from Dispersion Curvea
Einstein Box distribution
Normal mode N E ~ YE (cm-l) N B ~ VL (cm-') vu (cm-1)
v1 asymmetric CF2 0.20 1449 0.67 1250 1440
stretching 0.13 1256
v2 C-C stretching -- ... 0.53 1161 1379
0.47 1161 1229
v3 symmetric CF2 0.17 1149 0.83 723 1154
stretching
14 CF2 wagging 0.20 576 0.20 630 672
0.60 587 672
vg CF2 rocking 0.27 755 0.73 307 519
v6 CF2 bending 0.13 385 0.87 283 385
v7 CF2 twisting 0.23 294 0.77 190 289
a Vibrational frequencies retrieved from ref. 27; 1 cm-l is equivalent to 1.4388 K in 8 tempera-
ture.
b Number of vibrators.

Before using these programs, it was necessary to establish the number of

skeletal modes of vibration contributing to the low-temperature heat capacities.
For polyethylene there are two skeletal modes per CH2.3J0j26 The carbon<arbon
stretching vibrations, which might also be chosen as skeletal vibrations, are
closely spaced in the frequency region of the group vibrations. Group vibrations
are roughly those vibrations which involve the atoms connected to the backbone
chain. Only the vibrations which can approximately be described as backbone
torsion and bending are thus coupled sufficiently along the whole chain to be
called skeletal vibrations. These, in turn, also are coupled to a small degree
intermolecularly. One can assume that for PTFE there is little difference from
polyethylene for the assignment of vibrational modes, but one expects frequency
shifts.
Analyzing first the lowest heat capacities with a three-dimensional Debye
frequency spectrum of two vibrators per mole of CF2 leads to curve I in Figure
7. Only at the lowest temperatures does the constancy of 80 = 90 f 4 K signify
a fit to a three-dimensional Debye function. This value agrees well with earlier
data on a smaller database (83 K),'O and with data derived from sound velocity
(98 K).ll
To extend the description of heat capacities in terms of approximate frequency
spectra to linear macromolecules,a partial one-dimensional Debye function was
coupled by Tarasov with a one-dimensional Debye function.10>26The total vi-
brations still remain at two per mole of CF2. Using this more advanced inversion
routine leads to curves IV and V. The parameter 81is connected with the upper
frequency limit of the one-dimensional continuum and 8 3 with the upper limit
of the three-dimensional continuum. The ratio 83/@1 is often used to judge the
ratio of intermolecular to intramolecular vibrations. The data bank heat ca-
pacities give reasonable initial constancy for 81 = 250 and 8 3 = 54 f 1K. These
two parameters describe the heat capacities up to about 50 K. The sharp decline
of 81and increase in Q3 beyond 50 K is caused by the influence of the group vi-
brations. For polyethylene,these become important only above 150K; for PTFE
the fluorine mass is, however, 19 times that of hydrogen, reducing the vibrational
398 LAU, SUZUKI, AND WUNDERLICH

80.0

-,
h

'7 60.0
0

9
v

x
*
II

8P 40.0
u
u
(Y
I

20.0

140 280 420 560

Temperature (K)

Fig. 8. Calculated heat capacities of crystalline PTFE (81= 250 K, 0 3 = 54 K, group vibrations
as listed in Table IV): (I) experimental data on sample HV for reference, (11) calculated heat capacity
(at constant volume), (111) heat capacity contribution of the group vibrations, (IV) heat capacity
contribution of the skeletal vibrations only.

frequencies of the group vibrations so that they interfere somewhat with the
calculation of 0 1 and 0 3 .
The third approximation involves the subtraction of the group vibrations,
based on the calculations of Hannon et al.,27which are listed in Table IV. The
frequency distributions are broken up into Einstein terms for rather sharp
portions of the spectrum and box distributions for linear portions of the spec-
trum. Subtracting first all heat capacity contributions due to the group vibra-
tions as listed in Table IV from the recommended data of Table I before inversion
leads to curves I1 and 111. This removes the deviatioii of 0 3 completely, but
overcorrects 0 1 somewhat. The ultimate decrease of 01 beyond 150 K is due
to the difference between the heat capacities a t constant volume and constant
pressure, and is corrected below. Since the new values of 01 extrapolate to the
same minimum of 250 K, this value w;s used for future calculations. The slow
change in 81 a t higher temperature does not affect the later heat capacity cal-
culations seriously because its error is caused by subtracting a somewhat too large
group vibration contribution, which is added again In the calculation of the heat
capacity.
For a system of known 0 1 , 0 3 , and group vibration spectrum, a full heat ca-
pacity calculation routine has been programmed in our ATHAS computation
enter.'^,^^ The results of this calculation are shown as curve I1 in Figure 8 to-
gether with the experimental data on sample HV and data-bank data for lower
temperatures (curve I). The skeletal contributions alone are shown by curve
IV, the group vibrations by curve 111. The fit up to 170 K is well within the error
limit. Below 5 K the maximum error relative to the experimental data is -7%
 THERMODYNAMIC PROPERTIES OF PTFE 399

200 300 400 500

Temperature (K)

Fig. 9. Values of the Nernst-Lindemann constant A [eq. (6)]: (I) computation by inserting C,,
from calculation (curve I1 of Fig. 8) and C, from measurement (curve I of Fig. 8) into eq. ( 6 ) ,(11)
computation by equating of eqs. (5) and (6) with the experimentaldata of ref. 30, (111) as for (11) with
data of ref. 31.

in this calculation; above 5 K the maximum error is f 1.5%. The discrepancies

compare to -27 and *lo%, respectively, for the calculations in ref. 11.
We also used the full frequency spectrum of ref. 27, which was calculated for
isolated chains for the heat capacity calculation, instead of substituting the
low-frequency fit to a Tarasov function for the two skeletal vibrations. We broke
the two skeletal vibration distributions of Hannon et al. into 44 different averaged
Einstein terms. No fit to the experimental data could be obtained. A t 0.3 K
the discrepancy between calculated and measured data is a factor of 1000, at 5
K a factor of 10. Skeletal vibrations calculated for isolated chains are not ap-
plicable to crystals.
At higher temperature one finds a deviation due to the room-temperature
transition and later on the C p - C, difference becomes noticeable. Finally, the
fusion endotherm causes major deviations of the experimental data from the
calculation.
The final step of the calculation of the heat capacity must thus be the addition
of C, - C, to get the heat capacity a t constant pressure. For this calculation
we used the thermodynamic relation

Cp - C , = (a2TV)/p (5)
where a is the thermal expansivity, V the molar volume, and 0the compress-
ibility. Also, the Nernst-Lindemann a p p r o ~ i m a t i o n
was
~ ~used:
Cp - C, = AC:T/T,
400 LAU, SUZUKI, AND WUNDERLICH

200 300 400 500 600

Temperature (K)

Fig. 10. Comparison of measured and calculated heat capacities of crystalline PTFE: (I) ex-
perimental data of sample HV (100% crystalline), (11) calculated C, from C, and eq. (6) with A =
1.62 X mol K/J, (111) C, calculated, as for (11) in Fig. 8.

where A is a constant and T, is the equilibrium melting temperature. The

parameter A does not vary much from sample to sample. For elements and ionic
solids Nernst and Lindemann29found values for A between 2.4 X 10-3 and 9.6
X K mol/J.
In a first attempt to calculate C, - C , we took the differences between curves
I and I1 of Figure 8 and converted them into values of A with the help of eq. (6).
The results are shown as curve I in Figure 9. Also shown in Figure 9 are values
of A from a comparison of independent measurements of expansivity and com-
pressibility by Zoller30 (curve 11) and Weir31(curve 111). The main peak at 300
K is clearly due to the room-temperature transition. The deviations of curve
I1 are partially due to the sample, which was only semicrystalline. As a com-
promise we chose a value of 1.62 X 10-3 mol K/J for A . This value is similar to
A for p~lypropylene.~
With the A value of 1.62 X mol K/J, heat capacities at constant pressure
were calculated from the constant-volume heat capacities. These calculated
C, are shown as curve I1 in Figure 10together with the measured heat capacities
(curve I) and the constant-volume heat capacity (curve 111). Below 170 K, C,
and C, are practically identical. These calculated heat capacities are free of any
influence of transitions. They indicate that the experimental heat capacity of
crystalline PTFE is only free of transition contributions between 400 and 500
K. In this fact lies the main reason for the earlier difficulty in arriving at a set
of recommended data. The data of curve I1 above 170 K are now listed in Table
I1 as recommended heat capacities for completely crystalline PTFE.
 THERMODYNAMIC PROPERTIES OF P T F E 401

TABLE V
ComDarison of '32 Temperaturesa
Mass EC0h(298) 03 '33(ma~s/E,,h)~/~
Polymer (8) (J/cm3) (K) (Same units)
Polytetrafluoroethylene 100 6695 54 6.6
Poly(viny1idene fluoride) 64 7530 73 6.7
Poly(viny1 fluoride) 46 9625 105 7.3
a '33 data from ref. 26; E,,h(298) is the cohesive energy a t 298 calculated from data of ref. 32.

DISCUSSION
The new data here presented show that differential scanning calorimetry can
achieve the accuracy needed to derive equilibrium thermodynamic functions
for linear macromolecules. DSC is ideally suited for the multiple measurements
needed to study crystallinity and thermal history effects. Since heat capacities
are structure insensitive, equilibrium properties can be arrived a t by elimination
of the nonequilibrium transition effects. Heat capacities are then independent
of heating rate. Using DSC coupled with adiabatic calorimetry for low-tem-
perature data one can cover the total temperature range of interest.
The computation ability displayed in Figures 1and 2 permits large-volume
data handling, so that present limitations on accuracy are pushed back to the
calorimeter design. It is of interest to note that thousands of such calorimeters
have been put into use over the last 20 years, but that only a limited amount of
quantitative data has reached the scientific literature. Also, it should be re-
marked that little progress in actual calorimeter design has been made in the
last 10 years: instead effort has been directed toward design of the proper com-
putation facilities. We feel that with the present generation of minicomputers
the limit of data improvement due to statistical analysis has been reached at the
level of about f0.5% of heat capacity. The next step must be an improvement
in calorimeter design.
The connection of the heat capacity of the fully crystalline polymer to the vi-
brational spectrum reaches the accuracy of the experimental data (see Figs. 8
and 10). It is not possible to use the skeletal vibrations calculated for the isolated
chain for calculations of low-temperature heat capacities, but a fit of a two-
parameter Tarasov function seems adequate. Of the two parameters, Q3 is re-
lated to the intermolecular vibrations. The recently calculated list of (33 values
for 35 linear macromoleculesz6shows relatively small variations from polymer
to polymer. The average 0 3 is 83.2 f 24.4 K. Clearly molecules of low mass and
more densely packed crystals have the higher Q3, while larger-mass molecules
and more open crystals have the lower 0 3 .
On the basis of mass and cohesive energy density32it is possible to rationalize
the O3 of the three fluorocarbon macromolecules that have crystal structures
sufficiently similar to show isomorphism or isodim~rphism.~~ Table V displays
the numerical comparison, assuming the changes in 0 3 frequency are proportional
to the square root of the ratio of cohesive energy density over mass. The im-
portance of this observation lies in the predictive capability of this relationship
for the heat capacities of the series of all fluorocarbon copolymers.
402 LAU, SUZUKI, AND WUNDERLICH

The other Tarasov parmeter, 81,is related to intramolecular vibrations. It

was found earlier10326that all carbon backbone macromolecules have, because
of similar backbone bond structure, only mass-dependent 8 1 temperatures.
Indeed, 81of PTFE (250 K) is close to 81of polyethylene (520 K) multiplied by
the square root of the inverse ratio of masses [520(28/100)1/2= 2751. With a
well-established list of group vibrations for hydrocarbon polymers and fluoro-
carbon polymers, we have documented for this group of macromolecules a pos-
sibility to predict heat capacities of crystalline homopolymers and copolymers
based on the microscopic vibrational properties.
Two questions remaining in the calculation of heat capacities are the univer-
sality of the constant A in eq. (6) for the conversion of constant-volume data to
those at constant pressure and the ultimate effect of anharmonicity on the heat
capacity calculation. The problem of anharmonicity becomes particularly ob-
vious if one remembers that the model of harmonic oscillators on which all fre-
quency spectrum calculations are based does not account for volume expan-
sion.
With the calculated heat capacity of crystalline PTFE one can now understand
the earlier difficulties we had in arriving at a set of recommended data on ex-
perimental data alone. The room-temperature transition is broadened, par-
ticularly, for highly crystalline PTFE, to such a degree as to obscure the heat
capacity over a wide temperature range. Only after the removal of this non-
equilibrium contribution can the heat capacity be established.
While the data for crystalline PTFE are felt to be of similar accuracy to the
recommended data for other macromolecule,2 the data for amorphous PTFE
must still be considered preliminary. Better low-temperature data as a function
of crystallinity by adiabatic calorimetry are needed for improvement. PTFE
is one of the few linear macromolecules which shows large deviations of the heat
capacity from linearity above the glass transition. This observation also points
toward the need for better theoretical interpretation of heat capacities of liq-
uids.
Turning to the transitions of PTFE, it is interesting to note the similarity of
the glass transitions of polyethylene and polytetrafluoroethylene. The glass
transition of polyethylene starts at 120 K, with the major increase in heat capacity
occurring between 200 and 250 K.34 For PTFE we find the range of the glass
transition between 160 and 240 K,9 also a broad range, but with less of a two-step
character than in polyethylene. The increases in heat capacity are 10.5 J/K mol
at 237 K for polyethylene and 9.4 J/K mol at 200 K for PTFE, quite close to 11.3
f 2.3 J/K mol, the average of many AC,, for macromolecules with small mobile
units (beads).35
The heat of transition at room temperature (850 J/mol) is considerably larger
than was reported earlier [383,36539,25377,37and 400,38J/mol]. This must be
attributed to the better baseline separation in the present investigation. The
new value of the transition enthalpy for completely crystalline PTFE is also
supported by the transition enthalpy for sample HQ (285 J mol-l). From the
heat of fusion, the crystallinity of HQ is 33.2%. From the heat of transition, the
same crystallinity results (33.5%). The room-temperature transition of melt-
crystallized samples is much sharper than that of the virgin material (see Fig.
4). This has led in the past to discrepancies in comparisons of transition
enthalpies with crystallinities.
It is also interesting to note in Figure 4 that the copolymer with short side
 THERMODYNAMIC PROPERTIES OF PTFE 403

chains (T100V) has practically no room-temperature transition, while the

longer-side-chain copolymer (9704V) has a reasonable transition. An obvious
explanation is that the short CF3 branches may be partially included in the crystal
on high-temperature crystallization and prevent the setting up of the precise
1*15/7 or 1*13/6helices. The longer branches are, however, most likely rejected
more completely by the growing crystal, permitting better crystallization.33
The entropy of transition at room temperature (for the combined transitions)
is 2.90 J/K mol, a sizeable amount. Adding the transition entropy to the melting
entropy (6.78 J/K mol) leads to a total entropy change of 9.68 J/K mol for the
change from the rigid crystal to the melt. This compares favorably with poly-
ethylene (9.91 J/K mol) and the average entropy of fusion of 9.4 f 1.4 J/K mol
found per mole of rigid backbone groups for 30 well-documented macromole-
~ ~ high melting temperature of PTFE thus rests mainly with the lower
c u l e ~ .The
entropy of the crystal relative to the melt and not the reverse, as is occasionally
stated. Should PTFE undergo a single-step melting from the triclinic crystal
form 11, one would expect, based on the overall enthalpy and entropy of transi-
tion, a melting temperature closer to the equilibrium melting temperature of
polyethylene (414.6 K).39
We recognize the importance of the high-temperature crystal form of PTFE,
which is stable above the room-temperature transition (crystal form I), by calling
it a “condis” crystal,12 a crystal which is conformationally disordered. The
special mesophase character of this crystal form has already been suggested by
Starkweatherto who compared the crystal to the low-molecular-weightsmectic
liquid crystals. The entropy of transition is not sufficient to account for full
randomization of the chain conformation among two trans minima [Rln2 = 5.8
J/K mol)]; a larger portion of the transition entropy would, in addition, have to
be assigned to the volume expansion at the transition.
Melting represents the most important nonequilibrium process in thermal
analysis of macromolecules. The PTFE samples show mainly superheating.8
The newly established experimental equilibrium melting temperature of 605
K is in reasonable agreement with the onset of melting listed in Table I. These
temperatures of the beginning of melting are obtained by back-extrapolation
of the melting peak to the crystal baseline and should be largely independent
of heating rate if superheating is the only nonequilibrium effect.39 Only a small
part (less than 5%)of the melting occurs at lower temperature (down to about
500 K; see Fig. lo), an indication that there is only a small percentage of poor
crystals or partially ordered intercrystalline material. The heat of fusion is in-
dependent of equilibrium considerations and also does not change much with
temperature between 600 and 625 K because of similar crystal and melt heat
capacities (compare Tables I1 and 111). As a result, there should be little error
in assigning the total (nonequilibrium) heat of fusion to the approximate equi-
librium melting temperature for the calculation of the entropy of fusion.
It remains to comment on the recently found higher heats of fusion (5.10
kJ/mol). They are based on the measurement of the change of melting tem-
perature with pressure via the Clausius-Clapeyron equation7 and have found
some support from extrapolation of room-temperature specific volume versus
heat of fusion.41 It was suggested41that the indication of complete crystallinity
for the samples used in this research obtained by NMR may result from inter-
facial “immobile amorphous” portions of the molecules, i.e., may be an indication
of the breakdown of the crystallinity model. If this were so, heat capacities of
404 LAU, SUZUKI, AND WUNDERLICH

this interfacial material would also be close to crystalline and our data in Table
I1 would be little affected, but the heats of fusion would be seriously affected.
This effect on the heat of fusion should, most likely, even be more than propor-
tional to the weight of interfacial material since strained, uncrystallized molecules
usually show an exotherm on relaxation which would be expected to be super-
imposed on the melting endotherm. Although these speculations may be con-
sistent, one should remember that a 5.1-kJ/mol heat of fusion would lead to an
overall entropy of fusion of 11.3J/K mol. This would be higher than the entropy
of fusion of any other comparable macromolecule. However, calculations of the
conformational entropies of fusion for PTFE based on the general knowledge
of the stiffer PTFE molecule have been consistently l o ~ e r . Also, ~ ~ >any
~ ~non-
linearity of the crystalline specific volume above the last measured point of 593
K or changes in the superheating affecting dT,/dp may alter the Clausius-
Clapeyron result. At present this problem cannot be resolved. It should have
little influence on the heat capacities presented here, but it may require read-
justment of the preliminary amorphous H , s,and G in the future.

CONCLUSION
PTFE has been added to the list of thermodynamically better understood
macromolecules. For the crystal, a recommended set of thermodynamic func-
tions (C,, H , S ) has been presented. For the amorphous sample, a preliminary
set of thermodynamic functions has been derived. The crystal heat capacities
have been linked to the vibrational spectrum and the Nernst-Lindemann con-
stant has been estimated. The three transitions, glass transition, room tem-
perature transition, and melting transition, have been identified, characterized
thermodynamically, and compared with those of other macromolecules. The
special properties of PTFE (high mobility in the crystalline state and high
melting temperature) have been linked to the large existence range of a condis
crystal form (crystal form I). Without the condis state PTFE would be closer
to polyethylene. Differential scanning calorimetry was shown to be able to give
sufficiently precise information to support the crystallinity model for PTFE for
the structure-insensitive thermodynamic properties (i.e., C,, H , AC,, A H f , and
AHt are approximated by a linear addition of amorphous and crystalline con-
tributions). It is known that amorphous and crystalline areas are dependent
(i.e., connected to tie molecules) and may also be of less than macroscopic size
(<1pm in one or more dimension). One would expect the crystallinity model
to break down should these structure limitations become aggravated, for example,
in drawn or quenched samples. Perhaps recent high heats of fusion derived from
dT,/dp data are an indication of this breakdown.
This work was supported by the Polymers Program of the National Science Foundation under
Grant No. DMR78-15279. The authors would also like to thank Dr. H. W. Starkweather, Jr., for
extensive discussions, review of the paper, and supplying high-crystallinity PTFE samples.

References
1. See, for example, U. Gaur and B. Wunderlich, Am. Chem. SOC.Symp. Ser. 197, T. Provder,
Ed.,American Chemical Society, Washington,DC, 1982, p. 355; Am. Chern. SOC.Diu.Org. Coat.Plast.
Prepr. 45,576 (1981); B. Wunderlich, Am. Lab., 14,28 (1982).
 THERMODYNAMIC PROPERTIES OF PTFE 405

2. U. Gaur, S.-F. Lau, A. Mehta, H.-C. Shu, B. B. Wunderlich, and B. Wunderlich, J. Phys. Chem.
Ref. Data, 10,89,119,1001,1051 (1981); 11,313,1065 (1982); 12,29,65,91 (1983).
3. U. Gaur and B. Wunderlich, J . Phys. Chem. Ref. Data, 10, 119 (1981);see also refs. 26, 13,
and 10.
4. U. Gaur, H.-C. Shu, A. Mehta, and B. Wunderlich, J . Phys. Chem. Ref. Data, 10.89 (1981);
see also P. H.X. Shu and B. Wunderlich, in Thermal Analysis in Polymer Characterization, E. Turi,
Ed., Heyden-Wiley, New York, 1981, p. 124 and ref. 26.
5. J. Grebowicz, S.-F. Lau, and B. Wunderlich, J. Polym. Sci., to appear; U. Gaur and B. Wun-
derlich, J. Phys. Chem. Ref. Data, 10,1051 (1981).
6. H. Suzuki and B. Wunderlich, unpublished, U. Gaur and B. Wunderlich, J. Phys. Chem. Ref.
Data, 10,1001 (1981).
7. H. W. Starkweather, Jr., P. Zoller, G. A. Jones, and A. J. Vega, J . Polym. Sci. Polym. Phys.
Ed., 20,751 (1982).
8. E. Hellmuth, B. Wunderlich, and J. M. Rankin, Appl. Pvlym. Symp., 2,101, (1966).
9. S.-F. Lau, J. P. Wesson, and B. Wunderlich, Macromolecules, to appear.
10. B. Wunderlich and H. Baur, Adu. Polym. Sci., 7,151 (1970).
11. C. L. Choy, W. Y. Leung, and F. C. Chen, J . Polym. Sci. Polym. Phys. Ed., 17,87 (1979).
12. J. Grebowicz and B. Wunderlich, Adu. Polym. Sci., 60/61,1 (1983).
13. S.-F. Lau, Ph.D. Thesis, Rensselaer Polytechnic Institute, 1982.
14. C. A. Sperati and H. W. Starkweather, Jr., Adu. Polym. Sci., 2,465 (1961).
15. R. E. Moynihan, J . Am. Chem. Soc., 81,1045 (1959).
16. A. J. Vega and A. D. English, Macromolecules, 13,1635 (1980).
17. C. A. Sperati, in J. Brandrup and E. H. Immergut, Eds. Polymer Handbook, Interscience,
New York, 1975, p. V-29.
18. L. Melillo and B. Wunderlich, Kolloid Z. 2. Polyn., 250,417 (1972).
19. U. Gaur, A. Mehta, and B. Wunderlich, J. Therm. Andy., 13,71 (1978).
20. A. Mehta and B. Wunderlich, Am Chem. SOC.Diu. Org. Coat. Plast. Prepr., 35,393 (1975).
21. Laboratory MicroSystems, Inc., 106 Third Street, P.O. Box 336, Troy, NY 12181.
22. B. Wunderlich and R. C. Bopp, J. Therm. Anal., 6,335 (1974).
23. B. Terziiska, H. Madge, and V. Lovchinov, J. Therm. Anal. 20,33 (1981).
24. F. A. Quinn, Jr., D. E. Roberts, and R. N. Work, J. Appl. Phys., 22,1085 (1951).
25. P. Marx and M. Dole, J . Am. Chem. Soc., 17,4771 (1955).
26. Yu. V. Cheban, S.-F. Lau, and B. Wunderlich, Colloid Pvlym. Sci., 260,9 (1982).
27. M. J. Hannon, F. J. Boerio, and J. L. Koenig, J . Chem. Phys., 50,2829 (1969).
28. S. F. Lau and B. Wunderlich, J. Therm. Anal., 28,59 (1983).
29. W. Nernst and F. A. Lindemann, 2. Electrochem., 17,817 (1911).
30. P. Zoller, J . Appl. Pvlym. Sci., 22,633 (1978) (numerical data courtesy Dr. Zoller).
31. C. N. Weir, J. Res. Natl. Bur. Stand., 53,245 (1954).
32. D. W. Van Krevelen and P. J. Hoftyzer, Properties of Polymers, Elsevier, Amsterdam,
1972.
33. B. Wunderlich, Macromolecular Physics, Vol. I. Crystal Structure, Morphology, Defects,
Academic, New York, 1973.
34. U. Gaur and B. Wunderlich, Macromolecules, 13,445 (1980).
35. B. Wunderlich, J . Phys. Chem., 64,1052 (1960).
36. C. E. Weir, J . Res. Natl. Bur. Stand., 50,95 (1953).
37. S. R. Urzendowski, A. H. Guenter, and J. R. Asay, Polym. Prepr. Am. Chem. SOC.Diu. Polym.
Chem., 9,879 (1968).
38. G. T. Furukawa, R. E. McCoskey, and G. J. King, J. Res. Natl. Bur. Stand., 49,273 (1952).
39. B. Wunderlich, Macromolecular Physics, Vol. 3, Crystal Melting, Academic, New York,
1980.
40. H. W. Starkweather, J. Polym. Sci. Polym. Phys. Ed., 17,73 (1979).
41. H. W. Starkweather,J. Polym. Sci. Polym. Phys. Ed., 20,2159 (1982).
42. P. J. Sundarajan, J. Appl. Polym. Sci., 22, 1391 (1978).
43. A. E. Tonelli, J. Chem. Phys., 52,4749 (1970).

Received February 8,1983

Accepted September 28,1983