auricap - a Chemist's View
On Capacitor Dielectric Materials - A Chemist's View
Karl A. Weber, Ph.D.
The fundamental function of a capacitor is to store opposite polarity electrostatic charges on a pair of electrically isolated (insulated) conductive surfaces. The quantity of charge stored on each of these surfaces is ideally directly proportional to their surface areas and inversely proportional to the distance between the surfaces. Such a simple understanding needs to be refined to take into account the effect of the electric field set up between the charged plates on the insulating material. In theory this concern could be avoided by constructing capacitors without any material between: a vacuum.
Unfortunately, such high vacuum devices would be impractical. The next best insulating material is air since it provides limited interacting material and a very high resistance. Air is only practical for the lowest capacitances. Other insulating materials have included: paper/oil, minerals, ceramics, glasses, ceramic corrosion layers on metals and plastic films. Any insulating material used in capacitors of identical dimensions will increase the capacitance with respect to that of a vacuum.
The proportionality constant relating each material's capacitance enhancement over that of a vacuum is known as its "dielectric constant." The dielectric constant is a measure of the extent to which the insulating material's surface interacts with the electric field set up between the charged plates. The constant is dependent on two molecular level properties: the permanent "dipole moment" and the "polarizability" or the induced change in dipole moment due to the presence of an electric field. The permanent dipole moment is the average over the various dipole moments given rise to by structural charge density differences over intramolecular distances. The charge density differences result from the electronegativity differences between the various atoms which comprise the molecular structure of the insulator. Polarizability is the property which arises from changes in the molecular electron distribution induced by the applied electric field. Both of these properties contribute to a net field, of opposite orientation to the inducing electric field between the charged plates. The larger the dielectric constant, the greater the induced field on the surface of the insulating material or "dielectric."
In A.C. applications, where signal handling is involved, factors which affect the rates of both charging/discharging become key issues. Even though dielectrics with larger constants allow smaller size/capacitance devices, the properties of such dielectrics contribute deleteriously to audio signal processing. Where dielectrics with larger constants are employed, their larger dipole moments/polarizabilities interact more strongly with the inter-plate field, resulting in a stronger induced opposing field on the dielectric. When a capacitor is discharged across a load the polarized dipoles thermally relax in a statistical manner, exhibiting a time decay, observed as a tailing decay of residual current as complete discharge is approached. If the capacitor is suddenly discharged, allowed time to set and then shunted across a load it will discharge a residual current (of the same polarity as the initial charge). Upon dissipation of the bulk of the charge, the polarized dipoles on the dielectric thermally relax, which results in a residual charge on the plates. The residual charge from dielectric relaxation is known as the "dielectric absorption." When an audio signal is passed through a capacitor the dielectric absorption prevents full charging and discharging of the capacitor at the frequency of the alternating current signal. When the signal reverses the charging on the plates the dielectric absorption presents a lagging current of the former polarity, a hysteresis effect results. This effect becomes more acute with increasing frequency.
Obviously now, not all dielectrics are equal. In audio applications it is desirable to seek the insulating material with the lowest practical dielectric absorption; hence, lowest dielectric constant, barring size and economics. Dielectric materials can be classified based on their relative polarity/polarizability properties, which the dielectric constants and dielectric absorptivities parallel.
What follows is a qualitative categorization of dielectric materials in decreasing polarity/polarizability based on chemical structure considerations (Dielectric constant data "K" given when available):
I. Metal oxide corrosion layers (electrolytic capacitors):
1) Tantalum oxide (K = 11)
2) Aluminum oxide (K = 7)
Both consist of polar metal oxide bonds possessing large permanent dipole moments, polarizability factors are negligible.
II. Ceramics and Glasses:
1) Ceramics - typically alumina or aluminosilicates (K = 4.5 - thousands)
2) Glasses - typically borosilicate (K = 4-8.5)
Similarly, the polar inorganic oxide bonds in these materials have large permanent dipole moments.
1) Mica (most common) - an alkali metal aluminosilicate, hydrate (K = 6.5 - 8.7)
Same as II.
A. Polymer films - functionally linked - ranked in order of decreasing functional linkage polarity (brackets "[ ]" indicate guess based on functional group polarity):
1) Polyesters (ex. Mylar) - ester (K = 3.2 - 4.3)
2) [Kapton - ether and imide]
3) Polyamides (ex. Nylon) - amide (K = 3.14 -3.75)
4) Polycarbonate - carbonate (K = 2.9)
5) [PEEK - ether and ketone]
6) [Poly(phenylene oxide) - PPO - ether]
7) [Poly(phenylene sulfide) - PPS- thioether]
The members of the above list can essentially be ranked based on polarity considerations alone, though polarizability considerations are significant for the latter members of the list.
B. Polymer films - carbon chain backbone - ranked in order of decreasing attached-group polarity/polarizability:
1) Poly(vinyl chloride) - PVC - chloro-substituted (K = 3.3 - 4.55)
2) Poly(chlorotrifluoroethylene) - chloro- and fluoro-substituted (K = 2.48 - 2.76)
3) Poly(p-phenyleneethylene) - Parylene - exception to list phenyl ring in backbone (K = 2.65)
4) Polystyrene - phenyl-substituted (K = 2.54 - 2.56)
5) Polyethylene - essentially unsubstituted carbon chain (K = 2.3 - 2.37)
6) Polypropylene - methyl-substituted (K = 2.1)
7) Poly(tetrafluoroethylene) (ex. Teflon) - perfluoro-substituted (K = 2.0 - 2.1)
To rank the first two members of this list consideration must be given to both, polarity and polarizability considerations. Polymer 2) is adequately fluorinated to cancel C-F bond polarities, the C-Cl bonds are the prime contributors to its polarity. Since C-F bonds are not verypolarizable, polymer 1) has a higher polarizability than polymer 2) and a correspondingly higher dielectric constant. Polymers 3) and 4) can be ranked primarily on their polarizabilities, which are significantly higher due to the pi-electrons in their phenyl moieties. Polymers 5 and 6 differ mainly in that the methyl substituted chains are less prone to wrap against themselves due to steric methyl-methyl interactions. In Teflon the C-F bond polarities essentially cancel since it is completely fluorinated, and given that the C-F bonds are not very polarizable it exhibits overall less polarizability than its unsubstituted carbon chain analogue, 5), polyethylene.
Based on polarization/dielectric constant considerations for minimization of dielectric absorption, the best films for audio applications are teflon and polypropylene. Runners up would be polyethylene and polystyrene, based on these considerations alone. Throughout this discussion, I have assumed that dielectric absorption and the dielectric constant are directly correlated.
Apparently, when polarizability factors predominate, the time constant for relaxation of the field induced dipole is critical. Otherwise, one would expect polypropylene to have a lower dielectric absorption than polystyrene, which is not the observed result. This can be reasoned by re-examining what is being polarized by the field in each. In the case of the polystyrene, the pi-electrons in the aromatic rings (which have been modeled, in the past, as a "free electron gas") can orient electronically, with less mechanical change in the polymer structure. Hence, it can relax faster. In contrast, the polarization of polypropylene involves more mechanical change of the structure, and hence a slower relaxation rate.
Up to this point, I have only mentioned in passing paper/oil (paper-in-oil) capacitors. These classic devices from "days of yore" are making a comeback in some audio circles - especially among tube connoisseurs. Certainly, they are of interest. Unfortunately, they employ a composite dielectric that consists of a paper spacer/absorbent saturated with an oil; therefore, they do not readily lend themselves to the present simple analysis. Since the dielectric polarization primarily occurs on the surface of the dielectric material, in the vicinity of the plates, the dielectric constant in such a capacitor would consist of a weighted average of two dielectric constants: the most significant weighting attributed to the oil and the lesser weighting attributed to the surface area of the fibrules of paper in contact with the metal plates. The weightings for each dielectric component are not readily measurable. Hence, all that we know is that the contribution of the paper cannot be neglected, since it acts as a supporting spacer for the tightly rolled foil plates and makes intimate contact with them. One should measure the dielectric constant for each type of paper/oil combination under consideration.
The most common combination is that of a petroleum derived mineral oil absorbed into kraft paper. Common foils include aluminum and tin. More "exotic" variants on this theme, are those capacitors distributed by a certain manufacturer/distributor in England (who’s name shall remain omitted here), which consist of a vegetable oil/unspecified type paper dielectric and copper or silver foil plates. [It should be noted that the same British manufacturer also sells a series of mylar film (K = 4)/foil capacitors for signal handling: a pecular dielectric material for high cost/performance audio signal carrying applications.] The nature of the plate metal is not nearly as important as that of the oil and the paper. Mineral oils, consist mainly of saturated hydrocarbon oils and exhibit very low dielectric constants near K = 2. In contrast, papers such as Kraft paper exhibit dielectric constants on the order of K = 4. If we assume a mere 10% paper-plate contact, the composite dielectric constant would be near K = 2.2; which is similar to that of polypropylene. The 10% figure is merely an arbitrary suggestion. Unfortunately, one would expect two superimposed polarization thermal relaxation rates: that associated with the polar solid cellulosic paper would be significantly slower than that for the non-polar liquid mineral oil. In contrast, the relaxation rate for a non-polar polymer film such as polypropylene would be nearer that of mineral oil than that of paper; and its relaxation characteristics would be more uniform due to its homogeneous nature. The use of vegetable oil in place of the mineral oil only makes matters worse for the paper/oil composite dielectric; since vegetable oil, as a fatty acid ester, would exhibit a dielectric constant in the vicinity of K = 3. If we again assume 10% paper-foil contact, the composite dielectric constant would be near K = 3.1. (Perhaps such capacitors might be better suited for the culinary arts!)