• Cross-linking processes

The cross-linking of cable insulation is frequently accomplished by means of a process in which an organic peroxide is decomposed, forming radicals which abstract hydrogen atoms from the polymer molecules. The polymer radical sites then react together to form crosslinks. It will be clear that the cross-linking must be adequate to ensure that the insulation does not deform, despite the cable operating within the crystalline melting envelope. Thus control of the cross-linking process is necessary and DSC has been employed in several ways to assist in this control.

Figure 1 shows a DSC curve obtained for a precompounded commercially available polyethene containing a suitable peroxide. The sample (approximately 6 mg, weighed accurately) was held in a hermetically sealed aluminium pan, the instrument was purged with nitrogen and the heating rate was 10 K/min. After the first heating, the sample was allowed to cool and reheated. As for the thermal history work the data are baseline subtracted. The data also show the thermal history impressed on the polymer by the manufacturing process of the granules. A similar exothermto that observed in Figure 1 (1) was found from a scan of the pure peroxide, and for the purpose of this work, this exotherm was assumed to be due to the decomposition of the peroxide.

Figure 1 DSC curves for initial heating of polyethylene granule with peroxide incorporated (1) and subsequent reheating (2).

It has been found necessary to use hermetically sealed pans for this type of work. The decomposition products of the peroxide are appreciably volatile and if not contained, will volatilise, resulting in erroneously small peak integrations for the reaction. These types of data have proved useful in several ways.

First, the peroxide type can be determined using peak onset and peak position data for the exothermic peroxide decomposition. This information is useful for the prediction of decomposition products to be removed from the cable in subsequent process and also in the continuous vulcanisation plant used to effect the heating under pressure and to carry out cross-linking in the production process. Build-up of these decomposition products in the plant may otherwise cause a number of safety hazards.

Second, integration of the peroxide decomposition peak can be used to make aquantitative measurement of peroxide present. It is necessary to obtain curves with different amounts of peroxide and a suitable quantity of peroxide-free polyethene, ideally the base polymer, in order to produce a suitable calibration curve. This information can be used for material verification purposes but also for determination of residual peroxide where the cross-linking process is incomplete.

The investigation of incomplete reactions has also been employed to study dwell time requirements in the vulcanisation plant. Samples (10 mg) were heated isothermally in hermetically sealed pans at a temperature determined as being that of the coolest part of the cable during vulcanisation and for different short times of the order of minutes. After the short isothermal heating the DSC furnace was rapidly cooled, using liquid nitrogen, and then an experiment of the type shown in Figure 1 was run to measure the residual peroxide so that the extent of reaction for each time/temperature combination could be measured. Data for a series of experiments at 155◦C are given in Table 1.

Table 1 Extent of reaction for peroxide decomposition under isothermal conditions

Peroxide content measurement proved very useful when investigating alternative peroxides, some of which would not cross-link the polyethylene used in the experimental mixtures. It was possible to demonstrate that the peroxide was present in the mixtures but that it was insufficiently reactive to achieve adequate cross-linking.

Finally, the onset of peroxide decomposition is of considerable interest during the manufacturing process. The temperature of the polyethylene in the extrusion plant and dies must remain below the decomposition onset of the peroxide or else cross-linking will begin to occur there. This process results in inhomogeneities in the extruded insulation, which present a risk of electrical failure in service and are therefore unacceptable.

DSC measurements were also found useful when investigating the extent of cure of large epoxy castings used in the insulation of high-voltage cable joints and terminations. These are manufactured with a two-stage cure cycle and it proved possible to monitor cure at the end of each stage during some investigative work. Both residual reaction exotherm and thermodynamic glass transition temperature were monitored since the glass transition temperature of the resin increased as cure progressed. Figure 2 gives the glass transitions of two examples.

Figure 2 Thermodynamic glass transition of two cast epoxy samples showing the increase in Tg (lower curve) due to increased cure.

Interestingly, it was found that the thermodynamic glass transition matched not only the change in coefficient of expansion which accompanies the glass transition found by thermomechanical analysis (TMA), but also the heat distortion temperature as measured on macroscopic test bars. This finding permitted the use of DSC data with confidence for cure checks, in preference to the more complex and time-consuming preparation of heat distortion test bars.

• Investigation of unknowns

From time to time, it happens that problems arise in most manufacturing processes and cable manufacture is no exception. Faults in cables are not acceptable, owing to the long service life and reliability requirements, and therefore these must be investigated so that the cause of the fault can be remedied. Various analytical strategies are employed in such investigations, often in combination, since this gives better confidence in interpretation. DSC is a valuable tool for this type of work, either alone or in conjunction with other techniques, particularly where the fault is caused by a contaminant.

One such example involved a product with a thin insulation for low-voltage applications; this product is tested online for electrical integrity at a voltage considerably in excess of the service voltage and failures are cut out and rejected. At the time of the incident, online faults were at an unacceptable level, leading to costs in locating faults and removing them. The extrusion filters used to screen out foreign bodies were examined and a number of small pieces of plastic sheeting, which were identified as polyethene, using Fourier transform infrared (FTIR) spectroscopy, were observed. Samples were therefore subjected to DSC scans, at 10 K/min, in closed aluminium pans and under nitrogen purging between 30 and 200◦C. In addition, samples of candidate sources of this contamination were gathered together and were also analysed, by both DSC and FTIR spectroscopy. The DSC data are given in Figure 3.

Figure 3 DSC data for crystalline melting point of polyethene extrusion contaminant (1) and candidate sources.

The thermograms given in Figure 3 permitted the identification of the extrusion material, trace 1, and the similar trace just below it to be identified as the source of contaminant. This correlated well with FTIR data, which would have been insufficient in itself for identification because the actual analytical samples were themselves contaminated with matrix and so the FTIR spectra contained matrix peaks in addition to sample peaks.

Taking together the two data sets DSC and FTIR gave good confidence in the identification, whereas neither was fully conclusive alone. The practice of using more than one analytical techniques is very helpful when analyzing unknowns because it permits the testing of the interpretation of one data set against that from other techniques until a consistent and supportable interpretation of all data is achieved.

• Rapid scanning with DSC

High-speed calorimetry, as reported by Mathot and co-workers, has also proved extremely useful on a number of occasions for identification of very small inclusions. Figure 4 shows the melting point of a sample cut out of an extrusion filter.

Figure 4 Melting point of inclusion on extrusion filter. Sample mass 89 ug.

Experimental details were as follows: The sample was contained in a crimped standard pan; the instrument used with water cooling and nitrogen purging, which had been calibrated to run at the scan rate of 250 K/min. The sample had been removed from the filter using dissection tools whilst observing with a light microscope. It had first been placed in a diamond anvil cell for FTIR microspectroscopy, and the infrared (IR)-absorption spectrum, which permitted identification of a polyamide, was obtained. The sample was then recovered from the diamond anvil cell, again with the aid of the microscope and transferred to a pre-weighed sample pan for the DSC experiment, which gives confidence in the IR spectroscopy and suggests, on the grounds of melting point, that this is a nylon-6 grade.

This analytical strategy has also been applied to very small polyethene inclusions of the same material as those referred to in Figure 3, with some success. In both cases the additional confidence of having consistent data and interpretations from two techniques proved invaluable in supporting the production engineering investigations which were under way to eliminate the fault.

A different opportunity to make use of higher scan rates arose when some very small defects in an extruded cable screen were presented for analysis. The screen material is difficult to analyse by DSC because it contains a high (>30%) concentration of carbon for electrical purposes and exhibits very weak DSC responses to any transitions. Data were acquired in a DSC instrument, scanning at 50 K/min with helium purging on samples of about 0.35 mg. The initial data (Figure 5) gave no particular suggestion of any contaminant, and derivative curves were obtained for matrix and defect. It was concluded that the defect was pre-cross-linked gel because of the similarity of the trace from the defect to that of the pre-cross-linked gel.

Figure 5 First derivative DSC curves for matrix (1) and defect (2) in extruded cable screen.

Interestingly, a second fault of this type also appeared and the analysis was repeated. This fault was, however, shown to be due to a contaminant as the difference between the data for matrix and defect show in Figure 6. In both investigations the data obtained was of considerable importance to the resolution of the defect-producing mechanism.

Figure 6 DSC data, with derivatives, for matrix (1) and defect position (2) for extruded cable screen with contaminant.

These samples were also subjected to some TMA, which reinforced the interpretation of the DSC data and is described below. This experience, as well as demonstrating the value of at least attempting a rapid scan approach with difficult samples, also shows that investigation of derivative curves can be of great assistance in interpreting data, particularly from difficult samples.

Measurement of the glass transition sometimes causes considerable difficulties if attempted by the conventional heating rate approach to DSC, particularly in the case of formulations carrying heavy loads of filler as either reinforcement or fire retardant. It was desired to determine if a highly filled cable-jacketing material was suitable for low-temperature service. In this case the DSC analyzer, with liquid nitrogen cooling and helium purging, was used from −100 to 25◦C at a heating rate of 100 K/min for 0.5-mg samples. An example of the data is given in Figure 7, together with an expansion curve obtained by TMA using a 3.7-mm-diameter probe and 10-mN load at 10 K/min. It is observed that all three curves presented show a transition between −40 and −30◦C and that whilst the data from either experiment might be regarded with some caution, the combination of data is more persuasive of the presence of a transition in that region, permitting advice to be provided about minimum permissible temperatures of service once a suitable factor of safety had also been considered.

Figure 7 TMA signal (1) for cable sheath compared with DSC signal (2) and derivative (3).

It is arguable that none of the applications of rapid-scanning DSC given above would have been accessible by a more conventional approach. In the case of the identifications described, the technique was used to corroborate FTIR spectroscopic analysis on very small samples, whilst in the case of the other applications TMA was used to corroborate the DSC investigation of difficult samples. These corroborations were deemed necessary as part of the learning cycle in use of a novel application but are also desirable when dealing with unknowns, particularly those contaminated with the matrix. At the time of writing, the strategy appears to present an opportunity to extend the DSC experiment to samples which would have been previously considered too small or otherwise intractable.