Advanced Analytical techniques for Monitoring Composition Fluctuations of a Dielectric Elastomer Processed by Extrusion

·Home
·Table of Contents
·Materials Characterization and testing

Advanced Analytical techniques for Monitoring Composition Fluctuations of a Dielectric Elastomer Processed by Extrusion

Julio Roberto Bartoli
Depto.Eng.Metalúrgica e Materiais - Universidade de S.Paulo/USP, Brasil;
Fernando Galembeck
Instituto de Química da Universidade Estadual de Campinas/UNICAMP;
Helion Vargas
Centro de Ciencia e Tecnologia (LCFIS)-Universidade Estadual do Norte Fluminense, R.J.
Geraldo R. de Almeida
Faculdade de Engenharia Industrial/FEI, S.Bernardo do Campo.
Contact

ABSTRACT

point-to-point changes in the composition of the dielectric compound;
point-to-point changes in the thermophysical properties of the compounds.

INTRODUCTION

₃

₄

Dielectric polymer materials such as the EPDM or EPR rubbers are normally used for the electric insulation of high voltage cables[7,8,9,10] due to their good chemical, mechanical and electrical properties, as cable dielectrics for up to 230 kV. The dielectric strength of these materials is one of the most significant properties for the evaluation of their performance, as a function of the operation conditions (electric stress and temperature) and of the environment[11,12]. Previous work[13] demonstrated that cable failure under high electrical fields, as characterised by dielectric breakdown, does not occur at random: perforations are concentrated in the so called welding regions of the insulation described further and shown in Figure 1. This suggested that there may be some non-uniformity in the insulation, perhaps in its chemical composition. Several physico-chemical techniques, such as: photoacoustic spectroscopy (PAS), x-ray fluorescence (XRF), mass spectrometry (MS and GC/MS) and thermogravimetry (TGA), were used to detect and to quantify some of the compound rubber constituents and to verify the uniformity throughout the finished cable insulation.

EXPERIMENTAL

Materials and sample preparation:
Figure 1 shows an exploded view of the manufactured cables investigated in this study, a high voltage cable (138 kV - 400 mm²) insulated with an EPDM (Nordel 2722 by Du Pont) rubber compound. The main rubber compound additives are: kaolin (Al₂O₃·2SiO₄), Pb₃O₄, ZnO, a quinoline derivative antioxidant and paraffin. Cable insulation is applied over the conductor by extrusion of the compound, through entry ports orthogonal to the cable axis (extruder cross-head). There are two entry ports, at 180° from each other, as shown in Figure 1-A. These two regions are here named upper and lower inlet regions (UIR and LIR, respectively). Starting from these regions the polymer will flow, surrounding the electrical conductor and as the fluxes meet, two regions of junction are characterised: the external (EWR) and the internal (IWR) welding regions with respect to the extrusion machine. The procedure to obtain samples for all analysis is illustrated in sequence through Figures 1-b to 1-d. The sector enclosed between the UIR and the EWR (0° to 90°) is the region of the cable wherein additive concentrations were determined. The cable was cut in longitudinal slices along its axis, with a thickness of 1 mm as shown in Figure 1-D. On the whole, 12 distinct areas were sampled referring to the three radial distances: external, central and internal layers, at four angles (0°, 30°, 60°, 90°) as illustrated in Figure 1-E.


Fig 1: Exploded view of the high-voltage cable for measurement mapping. The procedure to obtain samples (A, B, C and D) and the regions analysed (E) are illustrated.

Binary mixtures of EPDM with the antioxidant or lead oxide were prepared in a mixer, in the laboratory. Calibration standards were obtained by mixing appropriate amounts of the antioxidant or lead oxide and EPDM.

Methods and instrumentation:
The physical and chemical techniques: PAS, XRF, MS, GC/MS and TGA, were used for the monitoring of the EPDM compound constituents. PAS and XRF are non destructive sample techniques.

PAS spectrometry: the photoacoustic spectra of the samples in the 260-800 nm (UV/visible) region were measured using a commercial (EDT, model OAS-400) photoacoustic spectrometer. This instrument consists of a 300 W high pressure xenon-arc lamp, monochromator, mechanical rotating sector, PAS cell and a detection system consisting of two lock-in amplifiers. All spectra were scanned at 50 nm/min at a spectral half-band pass of 8 nm and with a time constant of 1 sec and 20 Hz chopping frequency. Rubber samples used in this analysis were cut to 13 x 4 mm, 1 mm thick from pressed rubber mats (calibration) or from finished cable insulation. The wavelength ranges of the absorption bands for Pb₃O₄ is 540-550 nm, ZnO is 380-390 nm and for the antioxidant are 300-310 nm and 350-360 nm.

XRF spectrometry: the X-ray fluorescence spectra of the samples were measured using a Philips spectrometer (model Pm 9901) for Pb and Zn, to detect Pb₃O₄ and ZnO from the rubber compound, respectively. The measurements conditions were: 25 kV and 7 mA, W source and LiF crystal monochromator, 10 seconds cumulative counting. The detection angles used were: Pb at 33,92° (La 1) and background at 35,0°; Zn at 41,74° (Ka 1) and background at 40,24°. A Rigaku spectrometer was used for the Si and Al measurements to detect kaolin from the rubber compound. The measurements conditions were: 50 kV and 20 mA, Cr source, DTED monochromator, 20 second cumulative counting. The detection angles used were: Si at 108,16° and Al at 142,77°. The rubber samples diameter were 38 mm (25 mm effective) and thickness of 2 mm (thickness for signal saturation).

Mass and GC/MS spectrometry: the mass spectra were measured using Varian-MAT spectrometer (model 311-A). A probe for direct sample insertion was used to volatilise paraffin from the rubber samples at temperature of 200° C. The HP 5995 Gas Chromatograph coupled to Mass Spectrometer was also used to detect paraffin from previous rubber compound distillation (6 h at 180 °C, dissolved in CHCl₃).

Thermogravimetrical analysis (TGA): the rubber samples were pyrolised at 800°C (N₂ atmosphere) using a Du Pont TGA equipment (model 1090) in order to determine the total amount of inorganic material of the rubber compound samples (kaolin + Pb₃O₄ + ZnO).

Statistical analysis of the measurements data were done using Analysis of Variance (95% of confidence), two or three classifications, with replication. Signal intensities of the various samples were used to calculate X_l , defined in the four-parameter case of the Beta distribution[14]:

X_l = (I-I_min) / (I_max-I_min)

_min

_max

_min

_max

Thus, X_l is the normalised measured signal intensity variation for a given sample region. This is calculated as the ratio between I_max-I_min (I is the signal intensity at a given area and I_min is the minimum intensity for the set under examination) and I_max-I_min(a scale parameter, where I_max is the maximum intensity for the same set). X_l may vary between 0 and 1 (or 100%),depending on the specific sample analysed being closer to the minimum or maximum values found.

RESULTS AND DISCUSSION

PAS spectroscopy analysis.
Pb₃O₄and antioxidant determination in EPDM compound:
For several binary mixtures of EPDM with lead oxide (Pb₃O₄) or antioxidant it was measured the Photoacoustic (PA) signal as a function of concentration. Figure 2 shows the PA spectra of Pb₃O₄in EPDM. In the upper part of the figure, the PA spectrum of Pb₃O₄ is also shown. As can be seen in the insert of Figure 2, there is a simple linear correlation between the extrapolated intersection points at l = 543 nm and concentration of Pb₃O₄.

Fig 2: PAS of Pb₃O₄ in EPDM compound: 1.2% (A), 2.4% (B), 3.4% (C), 4.8% (D), 5.2% (E), 7.0% (F), 9.1% (G) and in powder form (H). Plot (insert) of signal amplitude versus Pb₃O₄ concentration in EPDM compound.

The spectra of the binary (Antioxidant-EPDM) mixtures were obtained at twelve different concentrations, from 0 to 2.0%, as shown in Figure 3. There is a strong absorption which peaks at l < 260 nm and two well defined shoulders at 303 and 355 nm. Signal intensity was measured at those two wavelengths, starting from the base line of each spectrum, and plotted as a function of antioxidant concentration, as shown in the insert of Figure 3. These figures show that the PA signal intensity increases (not linearly) over the range of antioxidant content from 0 to about 1.8% w/w (see Figure 3). When the concentrations exceeds 1.8% w/w, the PA signal amplitude decreases with concentration. Spectral shape changes were not detected in the concentration range studied, and thus the intensity maximum between 1.0 and 2.0% w/w cannot be attributed exclusively to saturation effects. This behaviour is still to be clarified. At this point we suggest that the solubility of antioxidant in the matrix has a maximum around 1.0% w/w at room temperature; the signal intensity decrease would be a result of phase separation. The range of antioxidant content between 0.5 and 1.0 is normally used in EPDM-based elastomeric mixtures in high-voltage cable insulation.

Fig 3: PAS of antioxidant in EPDM compound: 0%(A), 0.2%(B), 0.4%(C), 0.6%(D), 0.8%(E), 0.9%(F), 1%(G), 1.1%(H), 1.3%(I), 1.5%(J), 1.7%(K), 2%(L). Plot (insert) of signal amplitude vs. antioxidant concentration in EPDM

It may be readily appreciated from Figures 2 and `3 that this procedure is able to determine antioxidant concentration in elastomeric mixtures at the 1.0% level, with a relative standard deviation of 0.04. This indicates thus that binary mixtures may be analysed using the photoacoustic method.

Topoanalysis of rubber additives in finished cables:
Photoacoustic spectra were taken from selected areas of a finished, manufactured cable insulation, as described in Figure 1. The spectra are given in Figure 4 and show two "knees": one at 540-550 nm, associated with the presence of Pb₃O₄; the other, at 380 nm, corresponds to absorption by ZnO, as can be seen in the same figure, from the spectrum of ZnO in powder from at 380 nm; absorption shoulders assigned to the antioxidant which should occur at about 320 and 360 nm are not clearly discernible.

Fig 4: PA spectra for different areas in the cable: (CEN, 0°), (INT, 0°), (EXT, 0°). The spectrum of ZnO in powder form is also shown.

Spectra from the different areas sampled have similar shapes. They show small but definite intensity differences. These intensity differences were analysed as follows: signal intensity was measured from each spectrum, at 303 and 355 nm; moreover, the height of the 380 and 540-550 nm "knees" was measured, using the intersection of the extrapolated lines, as in Figure 2. Figures 5 and 6 give a visual description of the measured intensity changes calculated using the Beta distribution, already defined.

Measurements at 380 and 540-550 nm are dependent on ZnO and Pb₃O₄ concentration, respectively. Readings at 303 and 355 nm are associated with other components, including any contribution from antioxidant. The X_l changes at the various l do not follow a single, definite pattern. However, there are some irregularities. First, the graphs (Figures 5 and 6) obtained at 303 and 355 nm do superimpose very well; there is also appreciable overlap among the four curves for the internal samples ant the three curves (303, 355 and 380 nm) corresponding to external samples. Least overlap is obtained at the central region.

Fig 5: Signal intensity changes at 303 and 355 nm in the External, Central and Internal layers of the finished cable.

Fig 6: Intensity changes of Pb₃O₄ and ZnO in the External, Central and Internal layers of the finished cable.

There are significant statistical differences (95% confidence) between the minimum and maximum amounts of concentration measured for all elements in the interior of the cable dielectric compound. ZnO and Pb₃O₄ showed relative concentrations variations of 13.8% and 8.7%, respectively. Variations until 20% were found in the PAS signal for the antioxidant at different points of the measured cable section. However, the signal intensity attributed to the antioxidant is reduced after the sequential steps of compound processing.

XRF spectroscopy analysis.
Pb₃O₄and ZnO determination in EPDM compound:

Fig 7: XRF Signal intensity of Zn (top) and Pb (low) versus their respective concentrations in EPDM compound. Plot (insert) of XRF intensity of Zn and Pb versus specimen thickness.

For several ternary mixtures of EPDM with Pb₃O₄ and ZnO it was measured the XRF signal as a function of concentration. As can be seen in Figure 7, there is a linear correlation between the Zn and Pb concentrations and their emission intensity of X-ray fluorescence. The range of concentration analysed was C_f x 0.5 and C_f x 1.1, here C_f is the compound formula concentration of the metallic oxides. It was followed the internal standard criteria, one of the compensation methods for comparative or partial XRF analysis[4]. The intensity ratio between the fluorescent element and the internal standard is non matrix dependent and it is equivalent to the fluorescent element concentration. The internal standards were applied according the following rules[4]: no element should be present with an absorption edge wavelength between the wavelengths of the two emission lines under comparison and, also, no element with a strong emission line between the absorption edges of the two elements under comparison. So, for the quantitative analysis of Zn, Pb was used as internal standard and vice-versa; that is for all specimens having different concentrations of Zn or Pb it was kept invariant the Pb and Zn concentrations, respectively. The emission intensity of XRF were also normalised by the specimen linear density (g/mm), being the exposed surface constant (471.4 mm²).

The appropriate specimen thickness was determined considering the emission intensity ratio for a given thickness (I_h) and an infinite thickness (I_oo). It is suggested[4] an "infinite" specimen thickness (I_h/I_oo > 0.999) for XRF quantitative analysis to avoid the minimum effect of this parameter in the measurements. The plot insert in Figure 7 shows the ratio I_h/I_oo as a function of the thickness (h), either for Zn or Pb. It is observed the signal saturation for h ³ 1.2 mm, where I_h/I_oo > 0.999. Therefore a standard thickness of 2 mm was used for all specimens cut from the finished cables.

Topoanalysis of rubber additives in finished cables:
XRF measurements were taken from selected areas of a finished, manufactured cable insulation, as described in Figure 1, at three radial lengths but only two angular regions: 0° (UIR) and 90° (EWR). Figure 8 shows the statistical results (Beta distribution) of the XRF measurements for the elements: Pb (as Pb₃O₄) and Zn (as ZnO) of the EPDM dielectric compound (finished cable), Si and Al are not presented here but they showed similar concentration variations. The reduced variables of the Beta distribution were calculated for the XRF emission intensities averages (n = 6), as defined in the Experimental Section. The maxima amounts, related to signal intensity or concentration, are found in the central or intermediate regions of the cable dielectric section. The Pb₃O₄ and ZnO presented the same variation in the EWR, welding region, but not in the UIR, inlet region of the polymer melt.

Fig 8: Variations of Pb and Zn concentrations (Beta distrbution) observed in the finished cable at two angular (0° and 90°) and three radial (external, central and internal) regions.

There are significant statistical differences (95% confidence) between the minima and maximum amounts of concentration measured for all elements, including Si and Al (as kaolin), in the interior of the cable dielectric compound. Zn and Pb showed relative concentrations variations of 3.6% and 3.1%, respectively. Si and Al showed relative variations of 5.7% and 4.9%, respectively. This difference between Si and Al is not statistically significant as should be expected because the variation of kaolin must be the same either using Si or Al as control. The discrepancy between the relative variations of the XRF emissions and PAS signals attributed to the metallic oxides (ZnO and Pb₃O₄) could be due to the fact that the set and samples dimensions were not the same for each analysis. Furthermore, XRF is a more absolute method than PAS. The large differences observed using PAS could be related to variations on the thermal properties of the material, including the polymeric matrix. The thermal properties of the matrix could affect the PAS signal but not the XRF emission. The rubber samples from the dielectric cable used for PAS had 1/10 of the surface used for XRF, thus reducing the sample surface the material heterogeneities are more evident.

The variations or concentrations fluctuations of the inorganic constituents of the rubber compound could be associated to the compound preparation and processing. Compound preparation is characterised by a dispersive mixing ("banbury" mixer, before extrusion) and a distributive mixing (single screw extruder). The reasons for these fluctuations are:

inadequate dispersion of the inorganic constituents, due to: geometrical factors (shape and dimensions), density, rubber anisotropy, polymer specific interactions (chemical and physics);
rheological effects of the rubber extrusion processing: differences on the shear stresses, shear rates and orientation of the polymeric flows and temperature, changing the forces at the polymer-inorganic constituent interfaces;
the degree and direction of orientation of the non spherical constituents and the angle of orientation related to shear stress axis;
constituents fractionating after mixing steps and polymer processing (flow constrictions, melt fracture, elongation flow at the die exit).
different patterns of the melt flows modify the distribution of the constituents, smooth flow streams come from the two inlet regions (UIR and LIR) and break, reform and joint the flows at the two welding regions (EWR and IWR). In the welding regions the rebuilding of the laminar flow streams is no longer possible and distorted flows take places.

The inhomogeneities observed in the extruded dielectric rubber from its inorganic constituents could cause local differences on the material structure, physical or chemical, and consequently on its dielectric properties. The electrical permitivity of the inorganic constituents (ZnO~10, Pb₃O₄~20, kaolin~5) are higher than the EPDM based rubber (~2.5). Some points of the rubber could have an effective electrical field which could be increased from 1.3 to 2.5 times the average electrical field.

TGA and GC/MS analysis of the finished cable.

₃

₄

Mass spectrometry and GC/MS analysis were also taken from the selected areas of the cable in order to determine the distribution of the low molar mass constituents, paraffin and antioxidant, of the rubber compound. Lower amounts of paraffin than its original concentration were found in all cable dielectric regions and gradients of concentrations were also observed from the inner (-40%) to the outer (-20%) regions. The mass spectrometry showed that the molar mass of the paraffin vary in all regions of the dielectric, being the paraffin of lower mass (< 480 a.m.u.) more incident at the outer regions of the cable. Rheological effects on extrusion of the rubber compound could cause the fractionation of the paraffin chains, as the inlet melt regions present higher shear rates than the welding regions of the dielectric. The diffusion of paraffin inside the dielectric could be related to differences on the mass diffusion coefficient of paraffin, due to the variation of melt viscosity with the shear rate (non Newtonian flow). Other aspects that could affect the paraffin diffusivity is the degree of crosslinking (vulcanisation) of the rubber compound.

The loss of paraffin mass could increase the porosity or decrease the apparent density of the rubber compound. This could create microvoids on the rubber and could reduce the local permitivity (voids ~1.0). The effective electrical field could vary from 1.3 (spherical voids) to 1.5 (cylindrical voids) the average electrical field and it could cause partial electrical discharges which starts the breakdown mechanism. The loss of paraffin mass in the cable dielectric was found more significant at the inner regions, close to the conductor, were the electrical gradient is higher.

CONCLUSIONS

Composition fluctuations were found for inorganic and low mass organic constituents of a EPDM based rubber (dielectric of high voltage cable) using photoacoustic spectroscopy (PAS), x-ray fluorescence (XRF), mass spectrometry (MS and GC/MS) and thermogravimetry (TGA). Some relative variations were: kaolin 5.7% (using XRF and TGA), lead oxide 8.7% (PAS, XRF), zinc oxide 13.8% (PAS, XRF), antioxidant 20% (PAS) and paraffin is lost from 20% to 40% of weight (MS, GC/MS). These variations were assigned to: (i) point-to-point changes in the composition of the dielectric compound; (ii) point-to-point changes in the thermophysical properties of the compounds. Changes in chemical composition may arise from: imperfect dispersion of the rubber components, in the admixture and compounding fabrication steps; regions of smooth flow streams converging to regions of distorted flows; component fractionation, during extrusion and vulcanisation; rubber component degradation, during the fabrication process. The reasons for component fractionation after rubber extrusion and vulcanisation are: the rubber flows through channels and ports of various dimensions - field flow fractionation may ensue; the rubber compound is subjected to shear and to temperature gradients - thermal diffusion and convection may ensue. The fluctuations in composition and thermal properties of the compound caused dielectric constant fluctuations from point-to-point. As a consequence, some insulation points are subject to higher electric stresses than other points, and are more susceptible to dielectric breakdown. It was demonstrated that the heterogeneity of polymeric compounds and distribution of the antioxidant, paraffin and mineral charges dispersed in high voltage cable insulation can be analysed successfully using the PAS, XRF, MS, GC/MS and TGA techniques.

Acknowledgements:

The authors are grateful to: Taralli, C., R&D Chief Engineering (in memoriam); Kawano, Y., (XRF analysis) Institute of Chemistry - USP; the company laboratory staff: Hattori, R.S.; da Costa, R.A.; Saran, A. C.; Albuquerque M. R.; Rabelo, C.A.S. and also the financial support of Fapesp.

REFERENCES

ROSENCWAIG, A., Photoacoustics and Photoacoustic Spectroscopy, J. Wiley & Sons, N.Y., 1980, p. 307.
GALEMBECK, F., et al., Characterisation of a new composite material: Fe2O3 impgregnated PTFE: Photoacoustic spectroscopy and particle size determination, J. Appl. Polym. Sci. 25, pp.1427-1430 (1980).
VARGAS, H., et al., Optical absorption measurements in a new composite materials by combined photoacoustic and beam transmission tecnhiques, J. Appl. Phys. 57 (9) pp.4431-4436 (1985).
TERTIAN, R. & CLAISSE, F., Principles of quantitative X-ray fluorescence analysis, London, Heyden & Son. Ltd., 1982, p.382.
RUDENWICZ, P. & MUNSON, B., Determination of additives in polypropylene by selective chemical ionization mass spectrometry, Analytical Chem., 58 (2); pp. 358-361, Feb. 1986.
BARTOLI, J. R., Flutuações de composição em um elastômero dielétrico processado por extrusão, M.Sc. Dissertation, Unicamp-Unversidade Estadual de Campinas, Mar. 1988.
OCCHINI, E., et al., Thermal, mechanical and electrical properties of EPR insulations in power cables, IEEE Trans. Power Appl. Sys., PAS 102 (7), pp. 1942-52, July 1983.
TANAKA, T., & GREENWOOD, A., Advanced power cable technology, present and future, Boca Raton. CRC Press Inc., v.2.,p. 1-53.
BROWN, M., Compounding of ethylene propylene polymers for electrical applications, IEEE Electrical Insulation Magazine, v. 10, n.1, p.16, Jan/Feb., 1994.
BROWN, M. and HEMPHILL, J.J., Advances in metallocene polymers for wire and cable applications, IEEE Electrical Insulation Magazine, v. 14, n.1, pp.5-8, Jan/Feb., 1998.
BARTOLI, J.R., et al., Thermal aging of an insulating EPR-rubber compound, IEEE-Int. Symp. On Electrical Insulation, Cambridge (MA), pp. 225-228, June 1988.
MONTANARI, G.C., et al., Electrical endurance of EPR insulated cable models, IEEE-Int. Symp. On Electrical Insulation, Cambridge (MA), pp. 196-199, June 1988.
HATTORI, R.S., et al. Effect of molecular structure on dielectric properties of EPR, IEEE-1987 CEIDP, Annual Report, Pittsburgh, Oct.,1987.
BURY, K.V., Statistical models in Applied Science, J. Wiley & Sons, N.Y., 1975, chapter 10.