Tim

The melting behaviour of heat crystallized poly(ethylene terephthalate) P. J. HOLDSWORTHand A. TURNER-JONES The crystallization of amorphous poly(ethylene terephthalate) has been investigated using differential scanning calorimetry (DSC) and x-ray techniques. In particular, the occurrence of double melting peaks in the DSC thermograms has been examined, It is shown that the high temperature peak is due to a recrystallization and melting process taking place while the material is being scanned. Hence the DSC peaks are not a direct reflection of the state of the material at room temperature prior to the scan. As a result, some proposals in the literature for the structural changes accompanying the crystallization of poly(ethylene terephthalate) must be re-examined. Some interesting differences between the effect of temperature and time on the crystallization of poly(ethylene terephthalate) are also reported, INTRODUCTION POLY(ETHYLENE TEREPHTHALATE) (PET) can be quenched from the melt to produce material which is amorphous at room temperature. This amorphous PET can be crystallized by heating to above its glass transition temperature. The structural changes accompanying this crystallization process have often been investigated using differential scanning calorimetry (DSC), or the similar technique of differential thermal analysis. It has been found that PET which has been crystallized under a wide range of conditions frequently exhibits two endothermic fusion peaks 1 6. One of these peaks, peak I, remains at a constant temperature. The second peak, peak I[, is only observed after heat treatment above a certain temperature. Initially peak II is small and occurs at much lower temperatures than peak I. However, with increasing crystallization temperature, or crystallization time, peak I I increases in size and moves to higher temperatures. At the same time peak I decreases in size and eventually disappears completely. At present there are conflicting views about the origin of these peaks. Bell and Murayama ~ have proposed that peak [ is associated with chain folded crystals and peak II with crystals containing partially extended chains. Roberts 6, on the other hand, has interpreted peak I as being due to bundlelike crystals and peak II to chain folded crystals. Both authors tacitly assumed that no structural changes took place during scanning in the DSC. On this basis, the melting peaks were assumed to be directly related to the structure of the material at room temperature prior to the scan. It is our contention that these assumptions are incorrect. The isothermal crystallization of amorphous PET has been examined in detail by various authors 7-1°. Zachmann and Stuart 1° found, in particular, that only imperfect crystallites were formed at low crystallization temperatures. Upon heating to higher temperatures, the perfection of the crystallites 195

P. J. H O L D S W O R T H A N D A. T U R N E R - J O N E S

increased, often within very short times. They concluded that this was due to partial melting and recrystallization. Hughes and Sheldon 11 carried out a OTA study of amorphous PET. They found that the apparent area under the melting peak was significantly larger than the apparent area under the crystallization peak. They suggested that the crystallization peak was followed by a continuous crystallization process which was not detectable in the thermogram baseline. Bair 12 et al have shown conclusively that multiple melting peaks observed in DSC thermograms of polyethylene single crystals are the result of annealing during scanning. Considering the above results, and many of our own unpublished results, we were lead to the following working hypothesis: When PET is scanned on a osc, crystallites formed at low temperatures undergo a continuous perfection process as a result of partial melting and recrystallization during the scan. This leads to an increase in the overall crystallinity. We would like to present experimental evidence to show that this is a reasonable hypothesis. This is based on complementary DSC and x-ray studies of heat crystallized PET. We will show that on the basis of this hypothesis we can readily account for the occurrence of double endotherm peaks in DSC scans of PET. Some interesting differences between the temperature and time dependence of the crystallization process will also be reported.

EXPERIMENTAL

The starting material was a 0.25 mm thick cast film of PET. The polymer was made from dimethyl terephthalate and ethylene glycol by an ester interchange reaction. It contained less than 2 ~ copolymerized diethylene glycol. Samples of film were heat crystallized by heating in a Perkin Elmer DSC-1 calorimeter from room temperature to the required temperature, holding at this temperature for a given time, and then cooling to room temperature. In all cases heating and cooling rates of 16°C/min were used. X-ray, DSC and density measurements were carried out on the resulting samples. The melting behaviour of the samples was examined on the DSC using 10 mg of sample and at a scanning rate of 16°C/min. Scale readings were calibrated by scanning standard melting point substances at the same rate. Wide angle x-ray patterns of the materials were recorded in transmission using a Philips Diffractometer equipped with a proportional counter and pulse height analyser. Cu-Ka radiation was used. Density measurements were made at 23°C using a density gradient column made up from a calcium nitrate-water mixture. From the measured density values (p), values of the volume crystalline fraction o f the materials were calculated using the relation w~ =

-p - p~ Pe-- pa

196

(1)

THE MELTING BEHAVIOUR OF PET

It was assumed that pc -= 1-455 g/cm 3 (ref. 13). The measured density of the cast amorphous film was 1.338 g/cm z. This was assumed as the value ofpa. These crystallinity values should not be regarded as absolute. However, particularly since all the materials used in this study were unoriented it is thought that the use of equation (1) will give a good guide to relative changes in crystallinity. The spherulitic texture of some samples was investigated by observing thin microtomed sections between crossed polars.

R ES U LTS

Figure 1 shows DSC traces of amorphous samples which had first been heated in the DSC to various temperatures and cooled immediately. A scan of the

E 'E

A

i

t

I

150

200

250

_

Temperature (°C)

Figure l DSC traces of samples of originally amorphous PETwhich had been heated to various temperatures and cooled immediately: (A) amorphous, (B) 177.5°C, (C) 201 °C, (D) 220~C, (E) 240°C original amorphous film is also included. Figure 2 shows x-ray scans of the principal reflections of these materials. The intensity scale should be regarded as arbitrary since no attempt was made to standardize the volume of material irradiated. Density crystallinity values are shown in Figure 3. The samples prepared in the above manner were at elevated temperatures for different times, that is the total time spent in the DSC during heating and cooling. Taking an arbitrary highest temperature of 248.5°C, a second series of samples was prepared by heating the amorphous film to a given temperature, and then holding for a time, t = (2/16)(248.5 -- T) min. In this way all samples were treated in the DSC for equal times. They will be referred to as 197 P--N

P. J. HOLDSWORTHAND A. TURNER-JONES

v

D c

-

C

B

A

I

I

10

20 30 2e (degrees) Figure 2 X-ray diffractometer scans of PET samples which had been crystallized at various temperatures: (A) amorphous, (B) 177.5°C, (C) 201 °C, (D) 220°C, (E) 240°C

0.6

o

I:1

O

E 0.2

i--I O0

r-I

r-i

I

~

100

II

i

150

2 0

Temperature (*C) Figure 3 Variation of the volume crystalline fraction (Wv) with crystallization temperature of originally amorphous PET samples [] immediately cooled series; O isochronous series 198

THE MELTING BEHAVIOUROF PET

an isochronous series. Their density crystallinity values are shown in Figure 3. The x-ray patterns of the samples were indistinguishable from those of the materials cooled immediately from the same temperatures. DSC traces of a third series of samples, originally amorphous, which had been heated to temperatures close to the original melting point of the material, and then cooled immediately, are shown in Figure 4. Also shown is the scan of a sample which had been heated to 280°C, which is well above the melting point, and then cooled immediately to room temperature at 16°C/rain.

.,..,

g

.<

& D

A

.I

200

I

250 Temperature ( ° C )

Figure 4 DSC traces of originally amorphous PEa- which had been heated to the following temperatures (and cooled immediately) (A) amorphous, (B) 224°C, (C) 236°C, (D) 244.5°C, (E) 248'5°C, (F) 252°C, (G) 280°C

The effect of different holding times was also investigated. Amorphous samples were heated to 220°C or 240°C and then either cooled immediately (holding time 0 rain) or held for 5, 30 or 960 min before cooling. The resulting Dsc scans and some of the x-ray traces are shown in Figures 5 and 6 respectively. Density and density crystallinity values for these samples are given in

Table 1. 199

P. J. HOLDSWORTH AND A. TURNER-JONES Table 1 Density and crystallinity values of PET following crystallization at 220°C and 240°C Temperature (°C)

Time (min)

p (g/cm 3)

Wv

220

0 5 30 960

1 '396 1'397 1.399 1.404

0.50 0.505 0.52 0-56

240

0 5 30 960

1,406 1"407 1 "408 1.409

0.58 0.59 0.60 0.61

One amorphous sample was heated to 220°C, held for 5 min and then scanned to higher temperatures. The same procedure was also carried out using a holding temperature of 240°C. The resulting scans are shown in Figure 7. Also shown for comparison are scans of samples which had been similarly prepared but cooled to room temperature prior to scanning. Four amorphous samples were heat crystallized at 240°C for 5 min. They were subsequently scanned at 2 °, 4 °, 8 ° and 16°C/min respectively. The results are shown in Figure 8.

O .O

t'~

¢..)

220

2t, O

260 220 2/-,0 Temperature (*C)

260

Figure 5 DSC traces of PET samples crystallized at 220°C for: (A) 0 min, (B) 5 min, (C) 30 min, (D) 960 min, and at 240°C for (E) 0 min, (F) 5 min, (G) 30 rain, (H) 960 min

200

THE MELTING BEHAVIOUR OF PET

5

C

I

10

|

20

J

30

2 e (degrees)

Figure 6 X-ray diffractometer scans of ?ET crystallized at 220°C for: (A) 0 rain, (B) 960 rain, and at 240°C for (C) 0 rain, (D) 960 rain

201

P. J. HOLDSWORTH AND A. TURNER-JONES

I I ! t I I I I |

! ! ! I I

II D ,.......,../

(::

s/

I

/ B

. s / ~.. s

240

n- \ /

/ %~*"'--

250 260 Temperature ( *C )

Figure 7 DSC traces of materials crystallized for 5 min at (A) (B) 220°C (C) (D) 240°C. (A) and (C) were recorded without cooling to room temperature, (B) and (D) were recorded after cooling

202

THE MELTING BEHAVIOUR OF PET

tn

C

t-

.t)

D. ¢..)

I

1

1

240

25O

26O

Temperature ('C)

Figure 8 DSC traces of PET which had been crystallized at 240°C for 5 rain and recorded at heating rates of (A) 16°C/min, (B) 8°C/min, (C) 4°C/rain, (D) 2°C/min

203

P. J. HOLDSWORTH AND A. TURNER-JONES DISCUSSION

The DSC scan of amorphous PET shows a 'cold' crystallization peak and a melting peak (Figure 1). Between these peaks there is a wide intermediate temperature range over which nothing appears to happen. From measurements made on several samples, however, we found that the area under the crystallization peak was only about 70-80 ~o of the area under the melting peak. Similar differences have been reported previously4,11. The simplest explanation is that there is a continuous increase in crystallinity during the intermediate temperature range which is not detectable in the DSC baseline. The results in Figure 3 confirm that there is an increase in crystallinity in the intermediate temperature range. Points from the isochronous and the immediately cooled series fall on roughly the same curve in Figure 3. Hence the increase in crystallinity is not merely because samples which had been treated at higher temperatures had also been in the calorimeter for longer times. The correspondence between the x-ray traces of these two series supports this conclusion. All samples which had been heated to temperatures in the intermediate temperature range showed the same, space-filling, spherulitic texture, irrespective of the holding temperature, or time. Hence, crystallinity changes within this range must occur within the spherulites. This is in agreement with the work of Zachmann and Stuart 1°. The increase in crystallinity might occur in two ways. Either the total number of crystallites or the average perfection of the existing crystallites, may increase. The x-ray scans in Figure 2 suggest strongly that the second alternative is more important. As the temperature of treatment increases, there is a continuous increase in the sharpness and degree of resolution of the reflections. The broadening of x-ray reflections is usually attributed to two effects14. If the crystal size is small in a direction perpendicular to a set of planes, the x-ray reflection from those planes will be broadened. Broadening will also result if a lattice is strained. With polymers it is very difficult to separate size and strain effects1~. We have made no attempt to do so and will use the general term 'perfection' to cover both effects. On this basis, the x-ray results show that the overall perfection of the crystallites increases continuously in the DSC over the intermediate temperature range. An indication of the nature of the perfection process is given by the work of Zachmann and StuarO °. They isothermally crystallized samples of amorphous PET as fully as possible, raised the temperature, and then quenched after different times. They found that the density initially decreased, but then increased to exceed the starting value. They concluded that upon raising the temperature more perfect crystals are produced as a result of partial melting followed by recrystallization. These results applied to isothermal experiments. In a DSC scan the temperature is changing continuously. It is reasonable to suppose, therefore, that the increase in perfection of the crystallites occurring during a DSC run is the result of a continuous melting and recrystallization process. On this basis, consider what is happening at any particular temperature in the intermediate range during a scan. The least perfect crystals will be melting. These will recrystallize to form more perfect crystals at some higher temperature (i.e. later in the scan). Crystals which had 204

THE MELTINGBEHAVIOUROF PET previously melted will be recrystallizing. Some crystals will be stable and will not melt until a higher temperature is reached. The net result is a small increase in the average perfection of the crystals. When we talk about the ~melting of crystals' in the present context, we do not wish to imply that the crystals necessarily melt completely. It could be that only partial melting takes place before recrystallization. On the basis of the foregoing considerations, the thermograms of amorphous PET can now be interpreted as follows (1) When amorphous PET is heated in a scanning calorimeter, there is a crystallization process which converts the material into an assembly of imperfect crystallites. This is revealed by a large exothermic peak. The peak width suggests that crystallites with differing degrees of perfection are formed. (2) In the intermediate temperature range there is a continuous increase in crystallinity. This is due, at least primarily, to an increase in the average perfection of the crystallites by a continuous melting and recrystallization process. This produces a gradual increase in crystallinity which takes place over a wide temperature range and, therefore, is not detectable in the DSC baseline. (3) When the temperature is sufficiently high, crystallites which are melting, can no longer recrystallize. The result is a broad endothermic melting peak. Consider now the origin of double melting peaks. If we stop scanning at some temperature above the crystallization peak, the continuous melting and recrystallization process will be interrupted, producing a crystallite perfection distribution characteristic of the holding temperature. Upon reseanning we can await an associated melting process at some temperature above the holding temperature. If the holding temperature were low enough, we could expect the original continuous melting and recrystallization process to be completely re-established, as scanning proceeds. The complete thermogram should then display a melting process corresponding to the holding temperature and a main melting peak indistinguishable from that given by amorphous PET. This is in complete agreement with the present experimental results (see Figure 1) and those of the literaturO 6. I r a high holding temperature is used, the crystals attain a high perfection and hence a high stability. The associated melting process may then occur at such high temperatures that the continuous melting and recrystallization process is only partially re-established or not at all. This is because melted crystals are unable to recrystallize before they are scanned to temperatures so high that recrystallization is no longer possible. As the holding temperature is raised, the melting peak arising from the continuous melting and recrystallization process should diminish and finally vanish altogether, leaving only a melting peak characteristic of the original crystallization conditions. This is in complete agreement with the present experimental results (Figures 1 and 4) and with those in the literaturO 6. 205 P

O

P. J. HOLDSWORTH AND A. TURNER-JONES

Consider now the effect of scanning rate on thermograms of material which had been crystallized at 240°C for 5 rain (Figure 8). At a rate of 16°C/min, only one melting peak is evident but as the rate is reduced, a second peak appears at higher temperature and grows in size. This effect is perfectly explicable on the basis of the foregoing arguments. The peak in the 16°C/min run is associated with the holding temperature. Because the scanning rate is fast there is no chance for recrystallization to occur. As the scanning rate is reduced however, there will be increasing recrystallization, producing material which melts at a higher temperature. As the holding time is increased at any particular temperature the crystallite perfection distribution should become more characteristic of the holding temperature and the associated melting process greater in magnitude. This is in agreement with present results (see Figure 5) and those in the literaturO-6. An increase in temperature is also observed. Some of the foregoing effects might occur because material remains molten at the holding temperature and then recrystallizes upon subsequent cooling. Two sets of results speak against this. Firstly, none of the subsequent Dsc traces resembles that of material which had definitely been melted by heating to 280°C prior to cooling (see Figure 4). Secondly, essentially identical thermograms are observed irrespective of whether or not the materials are cooled to room temperature prior to scanning (see Figure 8). From the results of similar experiments Bell and Murayama 5 also concluded that the observed effects were not the results of recrystallization during cooling. In the foregoing paragraphs we have shown that a continuous perfection process can take place during scanning of PET in a scanning calorimeter. We have shown further that when account is taken of this process the occurrence of double melting peaks in heat crystallized PEa" can be easily explained. It should be noticed that it has not been necessary to postulate that the peaks represent crystallites with basically distinct morphologies. Neither do we have to postulate that crystals of high melting point are converted into crystals with initially lower melting points (which would seem to be very unlikely on thermodynamic grounds). Previous explanations of the effect5, 6 contained both postulates. Similar double melting peak thermograms have been observed in nylon 6616, polystyrene16, and drawn PETxT. In these cases too continuous melting and recrystallization during scanning with the DSC may provide the correct explanation. To conclude, we wish briefly to point out some interesting differences between the effect of time and temperature upon the crystallization of amorphous PET. The breadths of the x-ray reflections of PET samples, and hence their x-ray perfections, are highly dependent upon crystallization temperature, but relatively insensitive to time (see Figures 2 and 6). Similar results can be found in the literature 17-19. The density crystallinity values show a similar trend (see Table 1). Contrary to this, the melting behaviour is highly sensitive to crystallization time (see Figure 5). With increasing time the melting peaks associated with the holding temperatures of 220°C or 240°C become more pronounced and move to higher temperatures. This must be due to structural changes which have little or no effect upon the x-ray perfection or the density. 206

THE MELTING BEHAVIOUR OF PET

The simplest expression for the melting point of a polymer is

Tm - - Tm oc,

( 1_

A HzL

!1

(2)

Where Tmoc is the melting point of an infinite perfect crystal, ~e is the surface energy of the crystals, AHy their heat of fusion and L their thickness. The x-ray long period of most materials increases with holding time at a given crystallization (annealing) temperature. It is generally thought that this is due, at least partly, to an increase in the thickness of the crystalline regions. As can be seen from equation (2), such an increase would produce an increase in melting point. Zachmann and Schmidt 19 have reported however, that the long period of PET decreases with holding time. It seems very unlikely that the crystalline regions would decrease in thickness, but if this happened the melting point would be reduced and not increased. The increase in melting point is also unlikely to be due entirely to an increase in AHI since this would mean a reduction in lattice strain and hence an increase in x-ray perfection. As mentioned previously, no appreciable increase in x-ray perfection is observed. The melting point would increase if oe were reduced. This would happen if the surfaces became smoother in the course of time. Very tentatively, it is possible to suggest one process that might produce a more regular surface at the same time as thinner crystals. Suppose the original crystals have irregular surfaces through which the molecules pass roughly at right angles and suppose that with increasing time the surface tries to regularize in such a way that the crystals become bounded by a crystallographic plane. If this were a plane such as (001), which is not perpendicular to the molecular axis, a reduction in crystal thickness would result. It is clear that more work is required before the differing effects of time and temperature on the crystallization of PEr can be properly understood.

CONCLUSIONS

When heat crystallized PET is run on a DSC, tWO melting peaks are often observed. Whilst the peak at the lower temperature is associated with the crystallization conditions, the peak at the higher temperature is the result of a melting and recrystallization process which takes place during the scan. Consequently, the DSC scans are not a direct reflection of the state of the material at room temperature prior to the scan. There are significant differences between the effects of temperature and time on the crystallization process. When the temperature is increased there is an increase in x-ray perfection. At a given temperature, further structural changes take place in the course of time which result in an increased melting point but no large change in x-ray perfection. The present work also illustrates the pitfalls which can be awaited if Dsc is used for morphological studies in isolation from other techniques. 207

P. J. H O L D S W O R T H A N D A. T U R N E R - J O N E S

ACKNOWLEDGEMENTS The a u t h o r s would like to t h a n k W. J. Price for carrying out much of the practical work, a n d D. A. Hemsley for m a k i n g the optical e x a m i n a t i o n of the materials.

I C I Plastics Division, Welwyn Garden City, England

(Received 11 September 1970)

REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

Kanetsuna, H. and Maeda, K. Kogyo Kagaku Zasshi 1966, 69, 1784 Mitsuishi, Y. and Ikeda, M. Kobunshi Kagaku 1966, 23, 319 Mitsuishi, Y. and Ikeda, M. Kobunshi Kagaku 1966, 23, 310 Lawton, E. L. and Cates, D. M. Amer. Chem. Soc. Reprints, San Francisco Meeting, 1968, p 851 Bell, J. P. and Murayama, T. J. Polym. Sei. (A-2) 1969, 7, 1059 Roberts, R. C. Polymer, Lond. 1969, 10, 117 Keller, A., Lester, G. and Morgan, L. B. Phil. Trans. Roy. Soc. 1954, A247, 23 Cobbs, W. H. and Burton, R. L. J. Polym. Sci. 1953, 10, 275 Zachmann, H. G. and Stuart, H. A. Makromol. Chem. 1960, 41, 131 Zachmann, H. G. and Stuart, H. A. Makromol. Chem. 1960, 41, 148 Hughes, M. A. and Sheldon, R. P. J. applPolym. Sci. 1964, 8, 1541 Bair, H. E., Salovey, R. and Huseby, T. W. Polymer, Lond. 1967, 8, 9 Daubeny, R. De P., Bunn, C. W. and Brown, C. J. Proc. Roy. Soe. (A) 1954, 226, 531 Klugg, H. P. and Alexander, L. E., 'X-ray diffraction procedures', John Wiley, New York, 1954, Chapter 9 Buchanan, D. R., Mc.Cullough, R. L. and Miller, R. L. Aeta eryst. 1966, 20, 922 Bell, J. P. and Dumbleton, J. H. J. Polym. Sei. (A-2) 1969, 7, 1033 Yabayashi, T., Orito, Y. and Yamada, N. Kogyo Kagaku Zasshi, 1966, 69, 9 Kilian, H. G., Halboth, H. and Jenckel, E. KolloidZ. 1960, 172, 166 Zachmann, H. G. and Schmidt, G. F. Makromol. Chem. 1962, 52, 23

208