Studies on the Phase Transition and Thermal Stability of Xydar and Zenite Series Liquid Crystalline Polymers.

TAI-SHUNG CHUNG [1, 2][*]

XING JIN [2]

Four commercially available main-chain LCPs, Xydar and Zenite series, are characterized by FTIR spectroscopy, DSC. polarized light microscopy (PLM), TGA, TGAFTIR, and elemental analysis. The FTIR spectrum of Zenite 8000B is found to be more similar to that of Vectra A950 than those of the other three LCPs. Xydar SRT900 and 1000 have almost the same FTIR spectra. Zenite LCPs have broader crystal-mesophase transition ([T.sub.CM]) and less clear liquid crystalline textures at ([T.sub.CM]) than those of Xydar LCPs. Xydar SRT1000 and Zenite 8000B have lower ([T.sub.CM]) than Xydar SRT900 and Zenite 6000, respectively. For all the four LCPs, the glass transitions can not be easily observed by DSC without the aid of annealing and the isotropic phase does not appear before thermal degradation. Xydar SRT1000 and Zenite 8000B have lower thermal stabilities as well as lower [T.sub.[CM.sup.S]] than Xydar SRT900 and Zenite 6000, respectively. It is interesting to find that [E.sub.a] (Apparent activation energy of thermal degrad ation) curves of Xydar SRT900 and SRT1000 have the similar shapes as those of Zenite 6000 and 8000B, respectively. For the first time, we have found that there is a minor degradation maximum for Zenite 8000 at about 2.5% weight loss in [N.sub.2]. For all the four LCPs, the [E.sub.a] values in an air atmosphere begin to decrease at a temperature close to the deflection points on the first derivative curves. The first stage of the thermal degradation in an air atmosphere has similar mechanisms as that in [N.sub.2].

INTRODUCTION

Because of the unique rigid-rod structure of liquid crystalline polymers (LCPs), they are endowed with many distinguished properties such as the high thermal stability, the ease of processing in their liquid crystalline mesophases and the ability to be used to yield high-strength and high-precision products. LCPs have been extensively used as molded components in the electronic industry. The excellent thermal, mechanical, and electrical properties of LCPs have promised a highly prospective potential market for them. Several companies have devoted much effort to develop new LCP products. Wholly aromatic LCPs are among the most important LCPs, of which Kevlar produced by DuPont is the earliest commercialized brand. Kevlar is a lyotropic LCP, while in recent decades the researches of wholly aromatic LCPs are mainly on thermotropic LCPs, such as Vectra of Hoechst Celanese, Xydar of Amoco, and X7G of Eastman [1-11]. DuPont has also commercialized its own brand of wholly aromatic thermotropic LCP, named as Zenite. These commercial LCPs provide us with a valuable resource to characterize and compare their thermal and chemical properties as a function of monomer structure. Here we will discuss the phase transition and thermal decomposition behavior of Xydar SRT900, SRT1000 and Zenite 6000, 8000B LCPs.

In this paper, we used FTIR to investigate the chemical structures of the four wholly aromatic thermotropic LCPs. DSC and PLM were employed to understand the crystal-mesophase transitions of these LCPs. We attempted to link up the chemical structure with the phase transition behavior of these LCPs.

In fact, fundamental studies of the behavior of thermotropic LCPs at elevated temperatures are urgently needed for their applications in microelectronics, defense, composites and aerospace, where thermal oxidation properties become critical. In addition to our previous work [12], there have been some publications of such studies. For example, Crossland et al. [13] used TGA and pyrolysis-gas chromatography/mass spectrometry (Py-GC/MS) to study the thermal stability and thermal degradation mechanisms of the homopolymer of p-hydroxybenzoic acid (p-HBA) and its copolymer with terephthalic acid (TA), and biphenol (BP). They also calculated the overall activation energy for degradation by regression analysis of the isothermal TGA data. Dufour et al. [14] analyzed the thermal decomposition products of Vectra A950 and Xydar SRT-500 by TGA-MS. Hummel et al. [15] utilized MS and FTIR spectroscopy to analyze the decomposition products from linear-temperature programmed pyrolysis and they discussed the degradation mecha nisms including Fries-analogue rearrangements. Sueoka et al. [16] studied the thermal degradation mechanisms of the LCP with the composition of Xydar SRT-900. The origins of the main pyrolysis products from the corresponding comonomer units were estimated. Sato et al. [17] studied the flame resistant and self-extinguishing mechanisms of Vectra A950, Xydar SRT-500 and X7G by directly coupled thermal analysis MS. Li et al. [18] have reported some kinetic data of the thermal decomposition of some special LCPs.

In our previous work [12], we studied the thermal degradation behavior of Vectra A950, B950, and Xydar SRT900 with TGA and TGA-FTIR. In this paper, we will continue the thermal decomposition studies on Xydar SRT1000 and two Zenite LCPs. By using non-isothermal TGA methods, the kinetic parameters can be obtained over an entire temperature range. In isothermal TGA experiments, some degradation reactions inevitably occur before reaching the target temperature. Therefore, we conducted dynamic TGA experiments with constant heating rates, and used Ozawa-Flynn [19, 20] and Kissinger [21] methods to calculate the kinetic parameters. Since TGA-FTIR has its own advantages in identifying the volatile products of thermal degradation, we use TGA-FTIR to explore the degradation mechanisms of the LCPs.

1. EXPERIMENTAL

1.1 Materials

Four thermotropic LCP resins were studied, namely, Xydar SRT-900 and Xydar SRT-1000 of Amoco, Zenite 6000 and Zenite 8000B of Du Pont. All four LCPs are beige in color. The Xydar SRT-1000 was in the form of chunks, while the other three LCPs were in the pellet state. Xydar SRT-900 is synthesized from p-hydroxybenzoic acid (HBA), biphenol (BP), and terephthalic acid (TA) with a HBA/ BP/TA molar ratio of 2:1:1. It is believed that Zenite LCPs are composed of TA, HBA, hydroquinone, and other aromatic diols, aromatic dicarboxylic acids and aromatic hydroxy carboxylic acids [22-24]. Therefore, all four LCPs can be classified as polyesters.

1.2 FTIR (Fourier Transform Infrared)

LCP samples were scrubbed into powder, then ground with potassium bromide in a mortar, and finally compressed into small disks for FTIR analysis on a Perkin-Elmer Spectrum 2000 FTIR spectrometer. The scan range was 4000 to 400 [cm.sup.-1] with a resolution of 4 [cm.sup.-1].

1.3 DSC (Differential Scanning Calorimetry)

A Perkin-Elmer Pyris-1 DSC was used. The system was calibrated with high purity indium and zinc before tests. During tests, the samples were kept under a [N.sub.2] atmosphere of 20 psi, and on-line baseline subtraction was used. The samples were heated at 10[degrees]C/min from 80[degrees]C to 400[degrees]C and kept isothermal for 2 minutes, then cooled to 80[degrees]C and kept isothermal for 2 minutes. Afterwards, two other cycles were carried out following the same procedure.

1.4 PLM (Polarized Light Microscope)

The sample was first sandwiched between two glass cover slips and placed on the heating stage (Linkam THMS-600). After the target temperature was reached with a heating rate of 1 10[degrees]C/min and isothermal for more than 5 minutes, the sample was subjected to a slight pressure and shear force. The sample was observed in situ by a polarized light microscope (Olympus BX5O) with crossed polarizers. A red plate with a retardation of 530 nm was inserted between the polarizers. The microscopic data was analyzed with computer software (Image-Pro Plus 3.0).

1.5 TGA (Thermogravimetric Analysis) and TGA-FTIR

The thermal degradation behavior of samples was investigated under the same conditions as stated elsewhere (12). Dynamic TGA experiments were conducted at 5, 10, 20, and 40[degrees]C/min heating rates in both air and [N.sub.2] for Xydar SRT900, while at 5, 10, 15, 25[degrees]C/min for the other samples. The TGA part of the TGA-FTIR experiment was conducted at 10[degrees]C/min. The sample weight was about 5-7 mg.

1.6 Elemental Analysis

The elemental analysis was done on a Perkin-Elmer 2400 CHN elemental analyzer. This instrument uses the classical combustion method to convert the sample elements to simple gases of [CO.sub.2], [H.sub.2]O, and [N.sub.2]. The result gases are then controlled to exact conditions of pressure, temperature, and volume. Finally, they are separated and measured as a function of thermal conductivity. All the content other than C, H, N is regarded as oxygen element. The sample weight is around 1 mg.

2. RESULTS AND DISCUSSION

2.1 FTIR Spectra

The FTIR spectra of the four LCPs are shown in Fig. 1. For comparison, the FTIR spectrum of Vectra A950 is also included. Vectra A950 is a well-known main-chain thermotropic polyester LCP synthesized from 73% HBA and 27% hydroxynaphthoic acid (HNA). All of the five polyesters have a strong band at 1750 [cm.sup.-1], which is ascribed to the stretching of the ester carbonyl. They also have a broad band at 3076 [cm.sup.-1] associated with the stretching of C-H bonds on aromatic rings.

It is very interesting to note that the spectrum of Zenite 8000B is apparently more similar to that of Vectra A950 than those of the other three LOPs. In both spectra there exist bands at 1634, 1473, 1383, 1341, 1183, 920, and 475 [cm.sup.-1]. This similarity is probably caused by the naphthalene units in the polyesters. Normally there are ring vibration bands around 1600 and 1500 [cm.sup.-1] for aromatic homocyclic rings as shown for Xydar LCPs. The naphthalene ring produces two more ring vibration bands at 1634 and 1473 [cm.sup.-1] for both Vectra A950 and Zenite 8000B. The 1383 and 1341 [cm.sup.-1] bands arise from the stretching of C-O bonds connected to the naphthalene ring and the C-O bonds between the naphthalene ring and the phenolic oxygen. The 1183 and 920 [cm.sup.-1] bands can be assigned to the in-plane and out-of-plane bending of naphthalene C-H, respectively. Because normally only the bands above 600 [cm.sup.-1] are discussed in the literature, the 475 [cm.sup.-1] band is difficult to assign. T he above conclusion is in accordance with Li's analysis that the bands at 1605, 1340 and 478 [cm.sup.-1] indicate the presence of naphthalene units (25). The spectrum of Zenite 6000 also has a somewhat similar character, which suggests that probably there is also a small amount of naphthalene units in Zenite 6000. The fact that the spectra of Xydar SRT900 and 1000 are almost the same suggests that these two LOPs have almost the same units.

2.2 Phase-Transition Study by DSC

Figure 2 shows the second cycle DSC data of the four LCPs. The curves of the third cycle are almost the same as those of the second cycles, except the exotherms during cooling decrease about 2[degrees]C. These endotherms and exotherms are associated with the crystal-liquid crystalline mesophase transition. The melting endotherms (crystal [right arrow] mesophase transition, [T.sub.CM]) on the heating curves of Xydar SRT900 and SRT1000 occur at 346[degrees]C and 290[degrees]C, which correspond to the crystallization exotherms (mesophase [right arrow] crystal transition) of 302[degrees]C and 254[degrees]C during the cooling, respectively. Since the FTIR spectra of these two Xydar LCPs are almost the same, the difference of the peak temperatures is most probably caused by the change of the component ratio. For Zenite 6000 and 8000B there are only unclear and broad [T.sub.CM] endotherms on the heating curves at around 326[degrees]C (for Zenite 6000) and 267[degrees]C (for Zenite 8000B), corresponding to the apparen t crystallization exotherms at 289[degrees]C and 228[degrees]C during the cooling, respectively. The heats of fusion of all four LCPs are all around 2 J/g. The broader melting transitions of Zenite LCPs suggest that they have broader distributions of molecular weight.

For all four LCPs, the degree of supercooling for [T.sub.CM] is about 40[degrees]C, the glass transition can not be observed with the unannealed samples. From the TGA results (discussed later), we know that all four LCPs begin to decompose around 420[degrees]C, while the maximum DSC temperature is 400[degrees]C, so that the liquid crystalline mesophase-isotropic phase transition may not occur before thermal decomposition.

2.3 Phase-Transition Study by PLM

Figure 3a shows the crystal-liquid crystalline mesophase transition morphology of the four LCPs at the [T.sub.CM] measured by DSC. We can see that the liquid crystalline textures at [T.sub.CM] of Zenite LCPs are less clear than those of Xydar LCPs. This is probably because there exist more than three kinds of monomer units with different combinations of aromatic and functional groups in Zenite LCPs. The different lengths of the units and the random arrangements of the ester groups cause mismatching in the position of ester linkages of neighboring polymer chains. Consequently it is difficult to form the regular structure of the liquid crystalline phase. An examination of the micrographs in Fig. 3b shows no apparent differences between the mesophases of these two series of LCPs at 40[degrees]C over their [T.sub.CM]. Figure 3c shows the PLM morphologies of the liquid crystalline phases of the four LCPs at 410[degrees]C. At this high temperature, all four LCPs flow more easily, and the liquid crystalline charact er becomes much clearer. However, the isotropic phase still does not appear. These PLM results are consistent with the DSC results. The liquid crystalline mesophase is still observable after the sample is cooled to room temperature. Here we do not assert whether these mesophases are nematic, cholesteric or smectic. However, most mesophases of the aromatic polyester LCPs reported are nematic.

2.4 TGA

Figures 4a and 4b show the weight remaining vs. time curves for Zenite 6000 thermal degradation in [N.sub.2] and air with different heating rates, and the first derivative curves of the 10[degrees]C/min TGA curves are also shown. The curves for Xydar SRT900 and 1000 are similar and therefore are not given here. From the first derivative curves we can see that there is only one major weight loss step during the thermal decomposition processes in [N.sub.2], while there are two steps in air. The TGA curves of Zenite 8000B in air are similar to those of the other three LCPs. However, Zenite 8000B has a strange minor degradation maximum before the major one in [N.sub.2] as shown in Fig. 4c. This phenomenon does not appear for Zenite 8000B in air and for the other three LCPs in both air and [N.sub.2]. To be cautious, we have repeated the experiment many times and this surprising phenomenon appeared over 80% of the time at different heating rates. This may imply the existence of a thermal decomposition of a minor c omponent or structural unit in Zenite 8000B. Although in Figs. 4a and 4b there is also a minor peak on the first derivative curves before the major degradation maximum, the peak is nearly unnoticeable and the reappearance possibility is no more than 60%. No residue is left after TGA experiments in an air atmosphere; however, 38.0%, 37.6%, 39% and 40.3% of the weight remains after experiments in [N.sub.2] for Xydar SRT900 and SRT1000, Zenite 6000 and 8000B, respectively.

Two methods were used to calculate the apparent activation energy of the thermal degradation reactions. One was the Kissinger [21] method, the other was the Ozawa-Flynn method, named after both Ozawa [19] and Flynn [20] because of their independent work. The theoretical backgrounds of the two methods have been explained elsewhere [12].

According to the Ozawa-Flynn (19, 20) method, there exists an equation to relate the heating rate, r, and the apparent activation energy, [E.sub.a] at any specific weight loss as follows:

ln r = C - 1.052 [frac {[E.sub.a]}{RT}] (1)

where C is a function of weight loss, but becomes a constant at any specific weight loss. R is the universal gas constant, and T is the reaction temperature. Thus the [E.sub.a] value can be calculated from the linear dependence of 1n r vs. 1/T at different heating rates. [E.sub.a] = R [cdotp] Slope/1.052.

It can be seen from Figs. 5a and 5b that the [E.sub.a] curves of Xydar SRT-900 and Zenite 6000 are similar in that the curves reach a maximum at around 45% weight loss for the two LCPs both in air and in [N.sub.2]. For Xydar SRT-1000 and Zenite 8000B, their [E.sub.a] values almost remain constant throughout the decomposition process in [N.sub.2]. In an air environment, their [E.sub.a] values do not change much before 40% weight loss, but then decrease dramatically. For all four LCPs, the [E.sub.a] value in an air atmosphere begins to decrease at a point after the first maximum of the first derivative of the degradation curve, and close to the deflection point. The deflection points are at 48%, 50%, 47%, 47% weight loss for Xydar SRT900, 1000, Zenite 6000 and 8000B respectively.

According to the Kissinger method [21], the relationship between r and [E.sub.a] at the temperature, [T.sub.m], (Temperature at the maximum degradation rate [([partial]W/[partial]T).sub.m]), obeys the following equation:

1n [frac {r}{[[T.sup.2].sub.m]}] = 1n [frac {nRA[(1 - [[alpha].sub.m]).sup.n-1]}{[E.sub.a]}] - [frac {[E.sub.a]}{[RT.sub.m]}] (2)

[alpha] = [frac {[W.sub.o] - [W.sub.a]}{[W.sub.o] - [W.sub.f]}] (3)

where n is the reaction order, A is the pre-exponential factor in the Arrhenius formula Aexp(-[E.sub.a]/RT), [alpha] is the degree of conversion and [[alpha].sub.m] is the degree of conversion at [T.sub.m], [W.sub.a], [W.sub.o], and [W.sub.f] are the actual, initial, and final weights of the samples respectively. If the [[alpha].sub.m] or weight loss percentage at [T.sub.m] is assumed to be constant, then the first term in the right-hand side Eq 2 becomes a constant. Thus, one can calculate the [E.sub.a] value from the linear dependence of ln(r/[[T.sub.m].sup.2]) on 1/[T.sub.m] at different heating rates following the relationship of [E.sub.a] = -R [cdotp] Slope. To check the validity of this assumption, we found that the weight loss percentages for Zenite 6000 at [T.sub.m] in [N.sub.2] atmosphere were 22.6%, 22.9%, 21.9%, and 24.3% at the heating rates of 5, 10, 15, 25[degrees]C/min, respectively. They were 21.7%, 21.0%, 19.6%, and 16.8%, respectively, at the first [T.sub.m] in air. The weight loss percentage s at [T.sub.m] are also very close at different heating rates for Xydar SRT-900, SRT-1000, and Zenite 8000B. Therefore, it may be reasonable to take [[alpha].sub.m] at the first [T.sub.m] as a constant. However, we did not use the [T.sub.m] of the second degradation stage in air as a reference point for data analysis because of scientific uncertainty.

As pointed out in the previous section, there is a minor but unique degradation maximum before the major one in [N.sub.2] for Zenite 8000 as illustrated in Fig. 4c. We also calculated the [E.sub.a] of this degradation using the Kissinger method. The weight loss percentages at this minor degradation maximum are 2.8%, 2.7%, 2.3% and 2.0% corresponding to the heating rates of 5, 10, 15, 25[degrees]C/min, respectively.

According to the method mentioned by Salin and Seferis [26], from Eq 2 we can get the intercept on the ordinate of a plot of ln (r/[[T.sub.m].sup.2]) vs. 1/[T.sub.m]:

I = 1n [frac {nRA[(1 - [[alpha].sub.m]).sup.-1]}{[E.sub.a]}] (4)

We can solve for A and substitute it into the equation describing the decomposition reaction 

[frac {rd[alpha]}{dT}] = Aexp[lgroup]- [frac {[E.sub.a]}{RT}][rgroup][(1 - [alpha]).sup.n] (5)

where r is the heating rate, r = [frac {dT}{dt}].

Then we can get the reaction order n as:

n = [[[lgroup][frac{d[alpha]}{dT}[rgroup]].sup.-1].sub.m] [frac{[E.sub.a][e.sup.1]}{rR}] (1 - [[alpha].sub.m] exp [lgroup][frac{-[E.sub.a]}{[RT.sub.m]}][rgroup] (6)

The values of [T.sub.m] for different experiments in air and [N.sub.2], the [R.sup.2] (correlation coefficient) of In (r/[[T.sup.2].sub.m]) vs. 1/[T.sub.m], [E.sub.a], n and lnr results are listed in Table 1. It shows that the [E.sub.a] values in Table 1 are in the same range with those at the same weight loss point from Ozawa-Flynn method, and the sequence of the [E.sub.a] values from the two methods are also the same. We can see that the [E.sub.a] of thermal degradation for the minor degradation maximum of Zenite 8000B in [N.sub.2] is apparently lower than all the other [E.sub.2]s at the major degradation maximum in [N.sub.2].

2.5 TGA-FTIR

The results of the solely TGA and TGA of TGA-FTIR experiments show that all four LCPs are more stable in [N.sub.2] than in air. Xydar SRT900 and Zenite 6000 are more stable than Xydar SRT1000 and Zenite 8000B, respectively, in [N.sub.2] and before about 46% weight remaining in air.

The general characteristics of the evolution of the degradation products from the TGA furnace can be observed from Figs. 6a and 6b, which show the stacked FTIR spectra of the exhausted products for Zenite 6000 from TGA in [N.sub.2] and air, respectively. The stacked FTIR spectra for the other three LCPs are similar to this and not shown here. By observing the [CO.sub.2] bands around 2350 [cm.sup.-1] and 670 [cm.sup.-1], we can see that [CO.sub.2] is the dominant product in all the experiments, which exists all through the degradation process, and the amount of [CO.sub.2] is proportional to the thermal decomposition rate. In order to see more details, several spectra were selected and superimposed as shown in Figs. 7 to 9 and are explained as follows.

2.5.1 TGA-FTIR in [N.sub.2] Atmosphere

Figure 7 shows the selected FTIR spectra of the degradation products for the four LCPs in N2 environments. We can see that the spectra for the four LCPs are very similar to one another. At 1% weight loss only the bands of [CO.sub.2] appear. At 5% weight loss for Xydar SRT900 and 2.5% weight loss for the other three LCPs, there are bands appearing at 1760, 1600, 1500, 1260, 1190, 1165 and 1068 [cm.sup.-1]. The 1760 [cm.sup.-1] band is ascribed to the stretching of the ester carbonyl. The 1600 and 1500 [cm.sup.-1] bands are assigned to the vibration of the aromatic rings. The 1260 and 1190 [cm.sup.-1] bands result from the C-O stretching. The 1165 and 1068 [cm.sup.-1] bands are associated with the C-H in-plane deformation on aromatic rings. These bands suggest the existence of aromatic esters and possibly ethers. These products indicate the initiation of the main chain breakdown. CO can be observed with its characteristic double bands at the right wing of the [CO.sub.2] 2350 [cm.sup.-1] band. CO does not becom e noticeable until around 10% weight loss for Xydar SRT900, and 5% weight loss for the other three LCPs. Almost at the same time as CO appears, other bands appear around 3650 and at 3056, 1340, 818, and 750 [cm.sup.-1]. These bands exist from 10% to 30% weight loss for Xydar SRT900 and from 5% to 40% weight loss for the other three LCPs. These bands suggest the formation of phenols, since the 3650 [cm.sup.-1] band represents the O-H stretching, and the 3056 [cm.sup.-1] band is associated with the C-H stretching of aromatic rings. The 1340 [cm.sup.-1] band is caused by the in-plane deformation of the phenol O-H, and the 818 and 750 [cm.sup.-1] bands represent the out of plane bending of C-H on aromatic rings. After 40% weight loss, the 3056, 1068, 818, 750 [cm.sup.-1] aromatic ring C-H bands become unnoticeable, while the 1600 and 1500 [cm.sup.-1] bands still remain noticeable. This implies that some aromatic rings with hydrogen atoms have been substituted.

There is a weak band at around 1700 [cm.sup.-1] from 20% to 40% weight loss for Xydar SRT900 and from 10% to 30% weight loss for the other three LCPs. This band results from the stretching of ketone carbonyls and suggests the formation of aromatic ketones. It is interesting to notice that in Fig. 7c the 1185 [cm.sup.-1] band has almost the same intensity as that of the 2350 [cm.sup.-1] [CO.sub.2] band till 40% weight loss for Zenite 6000. The above results are in accordance with Hummel's hypothesis [15] that the thermal degradation of LCPs is attributed to Fries-analogue rearrangement and main chain splitting with intra- and inter- molecular dehydrogenation mechanisms.

The fact that only the bands of [CO.sub.2] are noticeable at the beginning suggests that the degradation starts with the scission of the terminal carboxyl groups, forming [CO.sub.2]. Also, the evaporation of the residual monomers, solvent, and oligomers, as mentioned by Hummel [15], may occur; however, the amount of these products is too small to be observed by FTIR. The inter- or intra- molecular ester exchanges as mentioned by Dufour [14] may also happen at this stage, but it would not cause the apparent weight changes of samples. The [CO.sub.2] appearing at the beginning of the decomposition could result from the decarboxylation of the original terminal carboxyl groups. However, the molecular weights of these four LCPs are very high, so the original amount of terminal carboxyl units should be very small. Besides, the [CO.sub.2] amount remains higher than the amounts of other products during the whole degradation process. Therefore, [CO.sub.2] should be produced from the ester bond during the breakdown of the main chain. However, Sueoka et al. [16] suggested that in wholly aromatic polyesters with the structure [[phi].sub.1]-CO-O-[[phi].sub.2], the C-C bonds between [[phi].sub.1]-CO are preferentially cleaved over the other types of C-C bonds, and the C-O bonds between CO-O are preferentially cleaved over the C-O bonds between O-[[phi].sub.2] They also proposed that the phenyl benzoate is formed mainly by the recombination reactions, not directly from the original ester moiety. According to Sueoka's conclusion, CO should be easily formed if the [[phi].sub.1]-C and C-O bonds in the [[phi].sub.1]-CO-O-[[phi].sub.2] structure break at the same time, and [CO.sub.2] is difficult to produce with the difficulty of the breakage of the O-[[phi].sub.2] bond. But in fact the [CO.sub.2] amount is much higher than the CO amount throughout the degradation process. Therefore, in most of the cases those two bonds break separately. When C-O bond breaks first, the Fries-analogue rearrangement (15) and recombination reactions fo llow next, only a small amount of [[phi].sub.1]-CO[bullet] degrade further to form CO. Whilst if [[phi].sub.1]-C bond is cleaved first, [CO.sub.2] and [bullet][[phi].sub.2]- are produced next, instead of CO and [bullet]O[[phi].sub.2]-, because [bullet][[phi].sub.2]- radical is more stable than [bullet]O[[phi].sub.2]- radical.

2.5.2 TGA-FTIR in Air Atmosphere

The FTIR spectra of the products from the degradation process in an air environment are similar for the four LCPs. The FTIR spectra for Zenite 6000 is shown in Fig. 8 as an example. Because CO can easily be oxidized by [O.sub.2], the ratio of the [CO.sub.2] amount to the CO amount shown in Fig. 8 is much greater than that for the samples degraded in [N.sub.2] as shown in Fig. 7. Similarly, the ratios of the [CO.sub.2] amount to the amounts of other decomposition products in an air environment are also very high. Figure 9 displays the detailed FTIR bands in the range of 1000-2000 [cm.sup.-1] when the weight losses of these four LCPs are 20%. These FTIR bands have the same positions and shapes as those observed in [N.sub.2] atmosphere TGA studies as shown in Fig. 7, though the peaks between 1300 and 1800 [cm.sup.-1] overlap with water bands. These bands exist only before 40% weight loss in air for the four LCPs. This clearly suggests that though the [O.sub.2] oxidation reaction dominates all through the degrad ation process in an air atmosphere, in the first degradation stage there also exist the degradation mechanisms similar to those in [N.sub.2]. During the second degradation stage, [O.sub.2] breaks down the cross-links formed by intermolecular reactions in the first degradation stage (15) and the char begins to degrade by oxidation reactions. With the decrease in the extent of cross-linking and the amount of the char left, [E.sub.a] decreases gradually.

For Xydar SRT900, SRT1000 and Zenite 6000, CO bands are noticeable before and after [T.sub.d] (deflection point on the first derivative of the TGA curve), while becoming nearly unnoticeable around [T.sub.d]. This suggests that the formation of CO follows two different mechanisms in the two degradation stages. Probably during the first degradation stage CO is mainly generated from the terminal [phi]-CO[bullet] radicals, while in the second stage CO mainly results from the uncompleted oxidation of the chars. For Zenite 8000B, the CO amount develops gradually with the increase of weight loss before reaching 60% weight loss, after that it decreases rapidly to nearly zero. For all four LCPs, additional bands exist continuously in the range of 3500-3800 [cm.sup.-1] during the major part of the degradation process. These bands represent the stretching of free O-H but do not represent the existence of [H.sub.2]O because of the absence of the relatively stronger [H.sub.2]O bands at around 1300-1800 [cm.sup.-1], since u sually the intensity of the water bands at 1300-1800 [cm.sup.-1] should not be weaker than that of the water bands at 3500-4000 [cm.sup.-1]. This suggests that some phenols are produced.

2.6 Elemental Analysis

Table 2 lists the results of elemental analysis on fresh samples and the residues after TGA-FTIR experiment in [N.sub.2] with a 10[degrees]C/min heating rate. The data show that the nitrogen contents in the fresh samples are nearly equal to zero, which agrees with the fact that no elemental nitrogen exists in the four LCPs. Elemental nitrogen is thus not considered in the following steps.

The loss percentage of an element, as shown in Table

2, is calculated with the formula: [frac{100 X [P.sub.1] - W X [P.sub.2]}{100 X [P.sub.1]}]

where [P.sub.1] is the measured weight percentage of the specific element in the fresh sample, [P.sub.2] is the measured weight percentage of the specific element in the residue, and W is the weight remaining of the sample. The loss percentage of H atoms is much higher than that of C atoms for all the four LCPs, which suggests the occurrence of hydrogen substitution reactions on the aromatic C-Hs during the thermal decomposition. The presence of 20-25% of oxygen left in the sample supports the existence of ketone and ether structures in the residues of the four LCPs. Comparing these results with our previous results (12), we can say that the thermal degradation processes of Xydar and Zenite series are similar to that of the Vectra series from the TGA, TGA-FTIR and the elemental analysis results.

3. CONCLUSION

We have studied the chemical structures and phase transitions, and determined the thermal stabilities of four commercially available thermotropic LCPs. The FTIR spectrum of Zenite 8000B is apparently more similar to that of Vectra A950 than those of the other three LCPs. This similarity is probably caused by the presence of naphthalene units in these two polyester LCPs. The FTIR spectrum of Zenite 6000 also has a somewhat similar character, which suggests that probably there is also a small amount of naphthalene units in Zenite 6000. The FTIR spectra of Xydar SRT900 and 1000 are almost identical, suggesting that these two LCPs have almost the same units.

DSC results show that the supercooling of [T.sub.CM] for these LCPs is about 40[degrees]C. The glass transition can not be observed with the unannealed samples. The [T.sub.[CM.sup.S]] of Xydar SRT1000 and Zenite 8000B are lower than those of Xydar SRT900 and Zenite 6000, respectively. Zenite LCPs have broader melting transitions than Xydar LCPs.

According to the PLM results, the liquid crystalline textures at [T.sub.CM] of Zenite LCPs are less clear than those of Xydar LCPs. No apparent difference can be observed between the textures of the mesophases of these two series of LCPs at 40[degrees]C over their [T.sub.[CM.sup.S]]. At 410[degrees]C, all four LCPs exhibit increased ease of flow, and the liquid crystalline character becomes clearer. However, the isotropic phase still does not appear. The liquid crystalline morphology is still observable after the sample is cooled to room temperature.

TGA data suggest that these four LCPs are more stable in [N.sub.2] than in air. Xydar SRT900 and Zenite 6000 are more stable than Xydar SRT1000 and Zenite 8000B, respectively, in [N.sub.2] and before about 46% weight remaining in air. There is a minor degradation maximum for Zenite 8000 at about 2.5% weight loss in [N.sub.2]. Xydar SRT-900 and Xydar SRT-1000 have similar [E.sub.a] vs. temperature profiles to those of Zenite 6000 and Zenite 8000B, respectively. For all four LCPs, the [E.sub.a] values in an air atmosphere begin to decrease at a temperature close to the deflection point of the first derivative curves.

TGA-FTIR results show that [CO.sub.2] is the dominant degradation product and exists through the entire degradation process, and the amount of [CO.sub.2] is proportional to the thermal decomposition rate. The degradation in a [N.sub.2] atmosphere starts with the scission of the terminal carboxyl groups and forms [CO.sub.2]. Before reaching 5% weight loss, some aromatic esters and ethers are produced, and then some phenols and aromatic ketones are produced. Substitution reactions of aromatic C-H groups also happen during the degradation. In an air atmosphere the ratios of the [CO.sub.2] amount to the amounts of other decomposition products including CO are much higher than those in [N.sub.2]. Though the [O.sub.2] oxidation reaction dominates throughout the entire degradation process in an air atmosphere, in the first degradation stage there exist similar degradation mechanisms as found in [N.sub.2]. During the second degradation stage. [O.sub.2] breaks down the cross-linked components formed by intermolecular reactions in the first degradation stage and the char begins to degrade by oxidation reactions.

ACKNOWLEDGMENT

We thank gratefully Mr. S. L. Liu, Dr. Pramoda, Dr. J. J. Shieh, Dr. S. Mullick, Dr. Y. H. Lin, Ms. S. X. Cheng, and other members in our group for their useful suggestions and kind help during this research. Mr. X. Jin acknowledges the assistance of the research scholarship from NUS.

Institute [1] of Materials Research and Engineering (IMRE)

3 Research Link, Singapore 117602

Department [2] of Chemical Engineering, National University of Singapore 

10 Kent Ridge Crescent, Singapore 119260

(*.) To whom correspondence should be addressed.

REFERENCES

(1.) T. S. Chung, Polym. Eng. and Sci., 26, 901 (1986).

(2.) T. S. Chung, G. W. Calundann, and A. J. East, in Handbook of Polymer Science and Technology, 2, 625, N. P. Cheremisinoff, ed., Marcel Dekker, New York (1989).

(3.) R. A. Weiss and C. K. Ober, Liquid Crystalline Polymers, American Chemical Society, Washington, D.C. (1990).

(4.) W. Brostow, ed., Mechanical and Thermophysical Properties of Polymer Liquid Crystals, Chapman and Hall, New York (1998).

(5.) H. Han and P. K. Bhowmik, Prog. Polym. Sci., 22, 1431 (1997).

(6.) J. Liu, F. Rybnikar, and P. H. Geil, J. Macromol. Sci. Phys., B35, 375 (1996).

(7.) H. D. Kim and D. R. Paul, Journal of Applied Polymer Science, 40, 155 (1990).

(8.) Y. G. Lin and H. H. Winter, Macromolecules, 21, 2439 (1998).

(9.) J. Economy and K. Goranov, Advances in Polymer Science, 117, 221 (1994).

(10.) M. Gordon (editor) and N. A. Plate (guest editor), Liquid Crystal Polymers I, Springer-Verlag, Berlin (1984).

(11.) A. M. Donald and A. H. Windle, Liquid Crystalline Polymers, Cambridge University Press, Cambridge, New York (1992).

(12.) X. Jin and T. S. Chung, "Thermal Decomposition Behavior of Main-Chain Thermotropic Liquid Crystalline Polymers, Vectra A950, B950, and Xydar SRT900," accepted by Journal of Applied Polymer Science.

(13,) B. Crossland, G. J. Knight, and W. W. Wright, British Polymer Journal, 18, 371 (1986).

(14.) P. R. Dufour, K. G. H. Raemaekers, and J. C. J. Bart, Thermochimica Acta., 175, 263 (1991).

(15.) D. O. Hummel, U. Neuhoff, A. Bretz, and H.-J. D[ddot{u}]sse1, Makromol Chem., 194, 1545 (1993).

(16.) K. Sueoka, M. Nagata, H. Ohtani, N. Naga, and S. Tsuge, Journal of Polymer Science: Part A: Polymer Chemistry, 29, 1903 (1991).

(17.) H. Sato, T. Kituchi, N. Koide, and K. Furuya, Journal of Analytical and Applied Pyrolysis, 37, 173 (1996).

(18.) X. G. Li, M. R. Huang, and H. Bai, Die Angewandte Makromolekulare Chemie, 258, 9 (1998).

(19.) T. Ozawa, Bull. Chem. Soc. Japan, 38, 1881 (1965).

(20.) J. H. Flynn and L. A. Wall, Polymer Letters, 4, 323 (1966).

(21.) H. E. Kissinger, Anal. Chem., 29, 1702 (1957).

(22.) M. S. Connolly, U. S. Patent 4,664,972 (1987).

(23.) M. R. Samuels and M. G. Waggoner, U. S. Patent 5,466,773 (1995).

(24.) M. R. Samuels and M. G. Waggoner, U. S. Patent 5,525,700 (1996).

(25.) X. G. Li, Z. L. Thou, X. G. Wu, and T. Sun, Journal of Applied Polymer Science, 51, 1913 (1994).

(26.) I. M. Salin and J. C. Seferis, Journal of Applied Polymer Science, 47, 847 (1993).

               Thermal Degradation Data Calculated with the
                             Kissinger Method.
                                   Xydar SRT900          Xydar SRT1000
                                    [N.sub.2]    Air       [N.sub.2]
              5[degrees]C/min         548.0     519.0        529.0
[T.sub.m] [a] 10[degrees]C/min        564.0     535.0        537.2
(C[degrees])  15[degrees]C/min [b]    574.0     551.0        547.0
              25[degrees]C/min [c]    590.0     566.0        559.3
Correlation Coefficient, [R.sup.2]      0.9912    0.9989       0.9719
[E.sub.a] (kJ/mol)                    283.9     230.2        278.9
Weight loss at [T.sub.m] (%)           25.3      21.7         26.8
n                                       2.5       3.1          2.1
In A ([min.sup.-1])                    40.1      32.8         40.5
                                            Zenite 6000
                                    Air      [N.sub.2]   Air
              5[degrees]C/min      516.4       545.8    528.1
[T.sub.m] [a] 10[degrees]C/min     528.3       553.4    544.1
(C[degrees])  15[degrees]C/min [b] 536.0       567.6    555.1
              25[degrees]C/min [c] 546.0       572.0    569.5
Correlation Coefficient, [R.sup.2]   0.9999      0.9772   0.9979
[E.sub.a] (kJ/mol)                 265.5       297.7    206.4
Weight loss at [T.sub.m] (%)        21.6        22.9     19.8
n                                    4.0         2.1      2.3
In A ([min.sup.-1])                 38.5        42.3     28.8
                                   Zenite 8000B
                                   [N.sub.2] [d] [N.sub.2]   Air
              5[degrees]C/min          481.5       518.9    493.7
[T.sub.m] [a] 10[degrees]C/min         493.3       531.0    506.9
(C[degrees])  15[degrees]C/min [b]     499.7       544.0    516.0
              25[degrees]C/min [c]     512.2       552.5    527.5
Correlation Coefficient, [R.sup.2]       0.9903      0.9897   0.9987
[E.sub.a] (kJ/mol)                     249.3       251.2    229.4
Weight loss at [T.sub.m] (%)             2.5        24.6     14.4
n                                                    1.9      3.2
In A ([min.sup.-1])                                 36.5     33.7

(a.)[T.sub.m] is the temperature at the degradation maximum. In the case of TG in air, [T.sub.m] is the [T.sub.m] at the first [([partial]W/[partial]t).sub.m];

(b.)20[degrees]C/min for Xydar SRT900;

(c.)40[degrees]C/min for Xydar SRT900;

(d.)The minor degradation maximum point for Zenite 8000 in [N.sub.2] condition.

                    Elemental Analysis Results of Fresh
                       and Thermally Degraded LCPs.
              Element Percentage                  Element Percentage
               of Fresh Samples                  of the Residue After
                     (%)                           TGA in [N.sub.2]
                                                         (%)
                      C           H    N     O            C            H
Xydar SRT900        73.79        3.26 0.02 22.93        83.85         1.56
Xydar SRT1000       71.69        3.37 0.08 24.86        85.68         0.76
Zenite 6000         71.90        3.30 0.08 24.72        85.85         0.82
Zenite 8000B        72.48        3.30 0.06 24.16        86.49         1.08
                    Weight    Calculated Loss
                    Remain Percentage of elements
                     (%)            (%)
                O                    C              H     O
Xydar SRT900  14.59  38.0          56.82          81.82 75.82
Xydar SRT1000 13.56  37.6          55.06          91.52 79.49
Zenite 6000   13.33  39.0          53.43          90.31 78.97
Zenite 8000B  12.43  40.3          51.91          86.81 79.27