Complex Impedance and Dielectric Dispersion in Carbon Fiber Reinforced Cement Matrices William J. McCarter, w Gerry Starrs, Thomas M. Chrisp, and Philip F. G. Banﬁll School of the Built Environment, Heriot-Watt University, Edinburgh EH14 4AS, Scotland, U.K. The electrical response of carbon ﬁber reinforced cement mortar over the frequency range 1 Hz–10MHz has been presented. The low frequency conductivity is shown to be directly inﬂuenced by increasing ﬁber dosage, with conductivity percolation occurring at dosages in the range 0.35%–1.0% (by volume). When viewed across the entire frequency spectrum, the conductivity and dielectric constant frequency dispersions reveal two regions of relaxation. The deﬁnition of these regions becomes more pro- nounced as the ﬁber dosage is increased. It is proposed that the low frequency relaxation process (1 Hz–1 kHz) is related to the enhanced charge transfer characteristics at the electrode– material interface produced by the presence of the carbon ﬁbers in the vicinity of the electrodes. The high frequency relaxation process (5 kHz–10 MHz) results from a Maxwell–Wagner effect produced by the presence of highly conducting inclusions (the carbon ﬁbers) in the low conductivity mortar matrix. I. Introduction C EMENTITIOUS materials form an important group of struc- tural materials. They are, however, poor conductors of electricity and the addition of short ﬁbers, such as carbon or steel, can make these materials electrically conductive. This could open up a diverse range of nonstructural applications; e.g., conductive ﬁber cement matrices could ﬁnd general appli- cation in electrical grounding, static charge dissipation, electrical resistance heating, and cathodic protection systems in reinforced concrete structures 1–3 ; by utilizing its piezoresistive properties, these materials also possess self-monitoring capabilities with respect to deformation and damage under static and dynamic loading. 4–8 Work on conductive ﬁber cement-based systems has, in the main, utilized the d.c. or low-frequency electrical conductivity/ resistivity of the material. Studies on the a.c. electrical properties of conductive ﬁber cements are more limited 9–11 and tend to be conﬁned to cement pastes with data normally presented in the form of a complex impedance (Nyquist) plot which allows an equivalent electrical circuit to be developed to mimic the ob- served response in terms of combinations of resistors and ca- pacitors. However, this particular formalism is not necessarily the best method for revealing the underlying effects responsible for the observed impedance response which is dependent upon conduction and polarization processes operative within the ma- terial. In order to develop an understanding of the mechanisms responsible for the impedance behavior, it is informative to study the frequency-domain dispersion characteristics of both dielectric constant and conductivity. This communication pre- sents a systematic study on the a.c. electrical response of mortars containing short carbon ﬁbers at dosages in the range 0%–1.5% (by volume) and offers a phenomenological explanation for the response. New data are presented in this respect. II. Experimental Procedure (1) Sample Preparation Mortar specimens were made with ASTM Type I Portland ce- ment (CEM I 42.5N) and a siliceous sand of maximum particle size 2 mm (Table I). A water reducing plasticizer (Fosroc Conplast P515, Fosroc Ltd., Tamworth, U.K.) was used in all mixes at a dosage of 0.6% (by weight of cement). Isotropic, pitch-based carbon ﬁbers (Sigraﬁl C s , SGL Technik GmbH, Meitingen, Germany) were used throughout the experimental program. The ﬁbers were coated with a water-soluble sizing at approximately 2% by mass of ﬁber; for the current work, a ﬁber length of 3 mm with a ﬁlament diameter of 7.5 mm was used throughout. The properties of the ﬁbers are presented in Table II with ﬁber dosages incrementally varied over the range 0%– 1.5% (by volume of specimen). A mortar with a sand/cement ratio of 0.5 and water/cement ratio of 0.5 (both by mass) was used. The sand and cement were initially dry-mixed in a 10 dm 3 Hobart planetary motion mixer; the ﬁbers and plasticizer were added to the gauging water and thoroughly mixed to disperse the ﬁbers before adding to the mixer. Specimens were cast in plexiglass molds which had in- ternal dimensions 40 mm  40 mm  160 mm (long); three spec- imens were cast for each mixture. To ensure intimate contact with the sample, two, stainless-steel electrodes, 35 mm  45 mm  2 mm (thick), were embedded within the samples at the time of casting and placed 150 mm apart. This represents a ‘‘two-point’’ measurement conﬁguration. The specimens were demolded after 24 h and cured under saturated conditions at 201C for 28-days at which point they were tested. (2) Measurements The real (resistance) and imaginary (reactance) components of the impedance were acquired using a Solartron 1260 frequency response analyser (Solartron Ltd., Hampshire, U.K.) in voltage drive mode employing a logarithmic sweep over the frequency range 1 Hz–10 MHz. The signal voltage was 100 mV and the measurements were recorded at 20 frequencies points per decade within this frequency range. Connection to the Solartron FRA was by means of short, individually screened coaxial cables to the voltage high/low and current output/input terminals. Cable impedance was nulled from the measurements by undertaking an open-circuit, short-circuit, and load calibration at all test fre- quencies. The load calibration entailed measuring the impedance of a parallel combination of a high precision resistor and capacitor (0.1% tolerance level). Speciﬁc values were chosen to provide an impedance behavior similar to that of the speci- mens being investigated. From this set of three measurements, together with the impedance measured with the sample between the electrodes, the sample impedance can then be deembedded T. Gur—contributing editor w Author to whom correspondence should be addressed. e-mail: w.j.mccarter@hw.ac.uk Manuscript No. 25378. Received October 21, 2008; approved February 25, 2009. J ournal J. Am. Ceram. Soc., 92 [7] 1617–1620 (2009) DOI: 10.1111/j.1551-2916.2009.03057.x r 2009 The American Ceramic Society 1617