A

udrrra©

K

1

tmiotflcericr

i1

0d

Available online at www.sciencedirect.com

Synthetic Metals 157 (2007) 1054–1063

Dielectric characterization of conducting textiles using free spacetransmission measurements: Accuracy and methods for improvement

Eva Hakansson a, Andrew Amiet b, Akif Kaynak c,∗a Center for Materials and Fibre Innovation, Geelong Technology Precinct, Deakin University, Geelong, Victoria 3217, Australia

b Defence Science and Technology Organisation, Maritime Platforms Division, P.O. Box 4331, Melbourne, Victoria 3001, Australiac School of Engineering and Information Technology, Deakin University, Geelong, Victoria 3217, Australia

Received 19 September 2007; received in revised form 31 October 2007; accepted 31 October 2007Available online 21 December 2007

bstract

The dielectric behaviour of in-situ polymerized thin polypyrrole (PPy) films on synthetic textile substrates were obtained in the 1–18 GHz regionsing free space transmission and reflection methods. The PPy/para-toluene-2-sulphonic acid (pTSA) coated fabrics exhibited an absorptionominated total shielding effectiveness (SE) of up to −7.34 dB, which corresponds to more than 80% of incident radiation. The permittivity

esponse is significantly influenced by the changes in ambient conditions, sample size and diffraction around the sample. Mathematical diffractionemoval, time-gating tools and high gain horns were utilized to improve the permittivity response. A narrow time-gate of 0.15 ns produced accurateesponse for frequencies above 6.7 GHz and the high gain horns further improved the response in the 7.5–18 GHz range. Errors between calculatednd measured values of reflection were most commonly within 2%, indicating good accuracy of the method.

2007 Elsevier B.V. All rights reserved.

rmitti

fnTamahars

awsf

eywords: Polypyrrole; Thin films; Electromagnetic interference shielding; Pe

. Introduction

Thin coatings of intrinsically conducting polymers (ICP) onextile substrates, also referred to as conducting textiles, are

ost commonly produced by chemical in-situ polymerizationn the presence of a textile substrate. Due to the tuneable naturef electrical properties as well as attractive physical proper-ies such as low weight, access to a wide range of structures,exibility, drape and low cost. ICP-coated textiles are goodandidate materials for use in thin electromagnetic interfer-nce (EMI) shielding materials. Further, ICPs have both sheetesistivity and capacitance at microwave frequencies, facilitat-ng the creation of microwave absorbing materials [1], whichan reduce the total amount of interference in this frequencyegime.

In our investigations, the dielectric characteristics of conduct-ng polypyrrole (PPy) coated textiles were determined in the–18 GHz frequency range using a non-destructive broadband

∗ Corresponding author. Tel.: +61 3 5227 2909; fax: +61 3 5227 2167.E-mail address: akaynak@deakin.edu.au (A. Kaynak).

rp

vcsb

379-6779/$ – see front matter © 2007 Elsevier B.V. All rights reserved.oi:10.1016/j.synthmet.2007.11.001

vity; Free space method

ree space method [2] utilizing unlensed microwave horn anten-as and a mathematical method for the removal of diffraction.he free space method is suitable for flexible thin samples whichre difficult to measure using conventional microwave measure-ent techniques, such as waveguide methods, dielectric probes

nd coaxial transmission lines. The measurement technique isighly reproducible and a very large number of measurementsre done on each sample, thus giving rise to statistically reliableesults. The only drawback of the method is the requirement ofophisticated and expensive equipment [3].

In the free space transmission measurements of magnitudend phase of S21, there is a certain degree of inaccuracy, whichill be evident as an error in the permittivity, reflection, transmis-

ion and absorption percentages. The errors principally originaterom the variations in experimental conditions, diffraction ofadiation around sample, stray reflections and variations in sam-le geometry.

This paper aims to present an analysis of the accuracy and

alidity of the free space transmission results obtained for PPy-oated textiles. The effects of calibration, removal of diffraction,ample placement and geometry are discussed. A comparisonetween the calculated and the measured reflection magnitudes

E. Hakansson et al. / Synthetic Metals 157 (2007) 1054–1063 1055

Table 1Sample, constituents, thicknesses and resulting surface resistivity of PPy-pTSA coated conducting textiles

Sample Structure Pristine thickness(mm)

Thickness after polymerization(mm)

Surface resistivity(�/sq)

Nylon-Lycra (90% Nylon, 10% Lycra) Double-sided basket-weave 0.53 0.54 180–1300Velvet (100% polyester) Flat one side, short fleece other 1.18 1.19 985Q

fscs

amto

2

2

vtw

A0fihamstfllcsfitts

2

2

tt9mt

2

mmCatmrra

2

pmws(l1icca

prccw(fi(

22sastt

uilt (100% polyester) Plain weave one side, non-wovenlayer, open mesh one side

rom transmission and reflection methods, respectively, are pre-ented. Possibilities for further reduction of errors and variationsaused by the diffraction of radiation around samples of smallerizes (≤305 mm × 305 mm) are discussed.

Accuracy of the measurements of reflection, transmission,bsorption and dielectric properties is important for evaluatingaterials for electromagnetic interference shielding applica-

ions. In this context, free space transmission methods and waysf improving the accuracy are explored.

. Experimental

.1. Materials and reagents

Thin films of polypyrrole were formed on textile substratesia in-situ polymerization [4–6] in an aqueous solution at roomemperature. Textiles of different composition and structure usedere as specified in Table 1.Para-toluene-2-sulfonic acid monohydrate (pTSA) (Sigma–

ldrich was used as dopant) in concentrations of up to.036 mol/l. The pyrrole (Aldrich) monomer concentration wasxed at 0.045 mol/l and the concentration of ferric chlorideexahydrate (FeCl3) (Fluka) acting as the oxidant was keptt 0.1 mol/l, suggested as optimized concentrations [6]. Poly-erization times from 60 to 300 min were used. After coating

amples were dried overnight at 25 ◦C in a drying cabinet, cuto size (305 mm × 305 mm or 500 mm × 500 mm) and storedat at 20 ± 2 ◦C at 65 ± 2% RH. In one of our previous pub-

ications, scanning electron microscopy (SEM) study of theonducting polymer coated fabrics showed a homogenous andmooth coating of conducting polypyrrole on each individualbre in the textile [7]. Bulk polymerized polypyrrole deposi-

ions were seen in the form of clusters and nodular particles onhe coating surface. These depositions were not adherent to theurface.

.2. Instrumentation

.2.1. Thickness measurementsThickness measurements were made on preconditioned tex-

ile samples in a standard atmosphere using a textile thickness

ester (DGTW01B, Mitutoyo, Japan) in accordance with ISO073-2 standard (0.5 kPa). The average thickness value from 20easurements on each sample was used in permittivity calcula-

ions.

trfs

3.80 3.89 165

.2.2. Conductivity measurementsThe surface resistivity of the conducting fabrics was

easured using a digital multimeter in a controlled environ-ent according to AATCC (American Association of Textilehemists and Colorists) test method 76–1995, where two rect-ngular copper electrodes (20 mm × 30 mm) are pressed ontohe fabric surface with a 10 N weight. Ten resistance measure-

ents were recorded and averaged for each sample. The surfaceesistivity RS is given by RS = R(l/w), where R is the measuredesistance of the fabric, l is the distance between the electrodesnd w is the width of each electrode.

.2.3. Dielectric characterisationA free space method was used to determine the dielectric

roperties of the PPy-coated conducting textiles in the 1–18 GHzicrowave region. An Agilent Technologies 8510C vector net-ork analyser (VNA) connected to an 8517A S-parameter test

et (Agilent) with an 83651B synthesized frequency sourceAgilent) was used to perform all the measurements and col-ect S21 and S11 data. The VNA has a dynamic range of over00 dB and resolution of at least 0.01 dB in magnitude and 0.01◦n phase. The frequency range of the complete system is able toover from 45 MHz to 50 GHz. An IBM compatible computerontrols the system, with the software written by the seconduthor [2].

It is possible to extract both complex permittivity and com-lex permeability of the sample under test using a combinedeflection/transmission measurement since both S11 and S21an be detected. However, due to the non-ferrous nature ofonductive fabrics investigated, only the complex permittivityas calculated. This was achieved by collection of either S21

‘transmission only’ method) or S11 (‘reflection only’) over therequency range. The approximation that the relative permeabil-ty of the conducting textile sample is equal to that of free spaceμ∗

r = 1 + 0i) is used throughout this work.

.2.4. Methodology for dielectric measurements

.2.4.1. Calibration. The free space ‘transmission only’ mea-urements were calibrated using a response calibration withoutny sample in the line between the two horns. The transmittedignal corresponds to the total response from the sample andhe diffraction around the sample. The calibration plane was athe top of the foam support level, where the fabric was posi-

ioned. The magnitude and phase of S21 with no sample wasecorded. Besides the response calibration, the calibration usedor ‘reflection only’ measurements also included an isolationtep. The isolation step involved collection of magnitude and

1 tic Me

pbpemaTaTp

2thsmtboeasoBpdtsw

stcfarsoma(acdtw6imv

2ptwtT

wtRwt(otTmi

a7SwcitacEhpcacct

2etstD[icmf

ε

wpow

t(aa

056 E. Hakansson et al. / Synthe

hase of S11 of a perfect reflector at the calibration plane backedy two 120 mm thick convoluted absorptive carbon black loadedolyurethane (PU) foams, used to reduce stray reflections. Tonsure that the reflected signal came from the conducting poly-er textile sample only, all samples to be tested were backed by

dditional absorptive foams during the reflection measurements.he reflection measurements are more sensitive to changes inmbient conditions, sample size, sample placement and air-gaps.herefore, the majority of the measurements presented here wereerformed using the transmission technique.

.2.4.2. Diffraction removal and time-gating. Multiple reflec-ions in the sample and stray reflections between the sample,orns and surrounding equipment during the free space mea-urements are likely to occur. This may cause error in theeasurements. Smith et al. have previously discussed calibra-

ion for free space methods with antennae mounted on an archut did not suggest any method of diffraction removal [8,9]. Inur work, the diffraction signal has been removed using a math-matical method involving two fast Fourier transforms (FFT)nd one inverse FFT. A time-gate was applied to the responseignal in the time domain to include only signals that reach theutput port within a predetermined amount of time. A Kaiser-essel window with alpha = 1.92 [10] is used in the time domainrocess. The time-gate is based upon the Kaiser-Bessel win-ow, with extra points added to the tails of the data to smoothhe transformation (Amiet [2]). This transformation was cho-en to provide low sidelobes with a small increase in peakidth.Only the radiation that reached the detection horn within a

pecified time was detected and used for permittivity extrac-ion. The time-gate was chosen to allow enough time to ensureollection of the complete signal from the sample at any givenrequency, including a sufficient amount of internal reflectionsnd to avoid errors in the gated signal due to unwanted strayeflections from e.g. surrounding equipment. The values pre-ented for the 1–18 GHz frequency range have a gate spanf 1.0 ns, which is the minimum allowed gate appropriate foreasurements at 1 GHz [2]. A reduced gate of 0.15 ns was

pplied to the response from selected samples of small size305 mm × 305 mm) with an additional 120 calculation pointsdded at each frequency end to improve the accuracy. The gateenter was kept at 0.0 ns. The 0.15 ns gate corresponds to aistance travelled by the wave equivalent to 0.045 m, which isrue at a frequency of 6.6667 GHz. Therefore, the re-gated dataith a 0.15 ns gate can only be relied upon at frequencies of.7 GHz and over. The resulting signal after diffraction removal,ncluding appropriate time-gating, was used to calculate the per-

ittivity and subsequent reflection, transmission and absorptionalues.

.2.4.3. Data collection. The conducting textile sample waslaced flat on a 300 mm thick polystyrene foam support in

he line of radiation between two microwave horn antennae,hich were connected to the radiation output system. The dis-

ance between the emitting horn and the sample was 0.31 m.he send horn, which was surrounded by absorptive foams,

tcs(

tals 157 (2007) 1054–1063

as positioned below the sample and the receive antenna abovehe sample. Two different kinds of horns, DRG-118A (Antennaesearch) and LHAO-750 (Continental Microwave and Tool)ere used. The broadband horns (DRG-118A) were used in

he majority of the experiments, while the high gain hornsLHAO-750) were used to improve the response by a reductionf diffraction. The high gain horns had a significantly higherypical gain of 15–22 dBi than the broadband horns (6–16 dBi).he higher gain decreases the beam-width of the radiation trans-itted, hence facilitating a reduction in sample size without an

ncrease in resulting diffraction.The radiation output system generated a swept signal

cross the 1–18 GHz frequency range (broadband horns) or the.5–18 GHz range (high gain horns). The scattering parameter21 or S11 was recorded at 401 frequency points across the band,ith 500 readings averaged at each frequency point. The data

ollection for one sweep across the frequency range took approx-mately 40 s. The obtained complex quantities S21 or S11 wereransformed into absolute magnitude and phase of the reflectionnd transmission. Subsequently, the permittivity of the materialould be extracted when the appropriate formulae were used.xtensive testing by the second author in a previous work [2]as shown that there is no measurable difference between testingerformed with plane waves or spherical waves. Since the sameonditions are used in the calibration, any effects are minor. Theuthor had investigated whether some focusing of the spheri-al beam would occur when the sample was in place, but usingomputer modelling of the effect proved it did not show up inesting.

.2.4.4. Calculations. The relative permittivity is the ratio oflectric field strength in vacuum relative to that of an encoun-ered medium. It consists of a real part associated with the energytoring capacity of the material and an imaginary part relatedo the electrically dissipative, or lossy, nature of the material.ipole polarization and charge migration have been mentioned

11] however, main contribution to dielectric losses in conduct-ng polymers at microwave frequencies may be attributed to freeharge rather than dipole interaction. The relative complex per-ittivity εr of a material, i.e. the permittivity relative to that of

ree space (ε0 = 8.854 × 10−12 F/m), is equal to

r = ε′r + iε′′

r (1)

here, ε′r is the real part and ε′′

r is the imaginary part ofermittivity. The negative values used for the imaginary partf permittivity are due to a sign convention adopted in thisork.As an electromagnetic wave travelling in free space encoun-

ers a material, some of the incident radiation enters the materiali.e. transmitted or absorbed) and some is reflected. Consideringmaterial/air interface, the reflection coefficient Γ is defined

s the fraction of the incident radiation that is reflected from

he front surface of a material, while the transmission coeffi-ient, T, is the ratio of transmitted to the incident electric fieldtrength. Assuming that the impedance of air is that of free spaceZ1 = 376.7 �), the reflection and transmission coefficients can

tic Me

ba

Γ

a

T

w

t

S

a

S

t[nDt

t

f

r

x

rcUsap

t(t

a

f

a

f

w

L

td

st

R

a

T

c

R

a

T

A

3

3u

mLeo3

E. Hakansson et al. / Synthe

e expressed in terms of relative permittivity and permeabilitys

= Z1 − Z2

Z2 + Z1=

√(μr/εr) − 1√(μr/εr) + 1

(2)

nd

= e−i(ω/c)d√

εμr (3)

here, d is the thickness of the material.Nicolson and Ross showed that for electrically thin materials

he scattering parameters S11 and S21 can be described as [12]

21(ω) = (1 − Γ 2)T

1 − T 2Γ 2 (4)

nd

11(ω) = (1 − T 2)Γ

1 − T 2Γ 2 (5)

The Newton method has been used to solve the implicit equa-ion for permittivity from free space transmission measurements2]. The approximation method is used due to the fact that it isot possible to express the permittivity in terms of S21 explicitly.uring the calculations, the permeability was fixed to be equal

o that of free space (μ = 1 + 0i).For the ‘transmission only’ method Eq. (4) was re-arranged

o the form f(εr) = 0 as

(x) = f1(εr) ≡ S21

{εr cos

[dω

√εr

c

]

+i

√εr

4(1 + εr) sin

[dω

√εr

c

]}− εr = 0 (6)

Using Mathematica this function was differentiated withespect to εr and can be expressed as

f ′1(εr) ≡ S21

4c

{(4c + idω(1 + εr)) cos

[dω

√εr

c

]

+(

ic(1 + 3εr)√εr

− 2dω√

εr

)sin

[dω

√εr

c

]}− 1 (7)

It is then possible to find the roots iteratively using

n = xn−1f (xn−1)

f ′(xn−1)(8)

The iteration was stopped when the difference between theoots xn and xn−1 was less than 10−7. The same calculation wasarried out at 401 frequencies across the frequency span tested.sing equations valid for normal incidence on a single dielectric

lab in free space [13], the coefficients of total incident reflectionnd transmission were then calculated based upon the calculatedermittivity from the measured S21 data for all 401 frequencies.

Using the ‘reflection only’ method, it is possible to expresshe reflection signal S11 in terms of ε and μ when using equations2) and (3) in (5). Rearrangement of this expression produceshe equations used to calculate permittivity from reflection [2]

astt

tals 157 (2007) 1054–1063 1057

s

(x) = f2(εr) ≡ S11

+ (εr − 1)(L − 1)

2√

εr(L + 1) + L(εr + 1) − (εr + 1)= 0 (9)

nd

′(x) = f ′2(εr)

≡

(1/

√εr

) ((L − 1)

[(√εr − 1

)2

+L(√

εr + 1)2

])+ 4idL(ω/c)(εr − 1)((√

εr − 1)2 − L

(√εr + 1

)2)2 = 0

(10)

here,

= ei2d(ω/c)√

εr (11)

The permittivity calculations from reflection were done usinghe iterative Newton method in a similar fashion as previouslyescribed for transmission.

Using the reflection and transmission coefficients, it is pos-ible to calculate the magnitudes of reflection (R[dB]) andransmission (T[dB]) from

[dB] = 20 log |S11| (12)

nd

[dB] = 20 log |S21| (13)

The percentages of reflection (R[%]) and transmission (T[%])an easily be calculated using the relationships

[%] = 100 × 10(R[dB]/10) (14)

nd

[%] = 100 × 10(T [dB]/10) (15)

The absorption percentage A [%] was calculated by using

[%] = 100 − T [%] − R[%] (16)

. Results

.1. Permittivity response for reference materials andncoated samples

Polytetrafluoroetylne (PTFE) was chosen as the referenceaterial as it Teflon is commonly used as a test standard.iquids have well known real and imaginary permittivity, how-ver, they are difficult to measure. The result of permittivitybtained for the reference PTFE slab with the dimensions00 mm × 300 mm and a thickness of 5.4 ± 0.08 mm as well

s the uncoated Nylon-Lycra, velvet and quilt textiles can beeen in Fig. 1. Slight variations can be seen for the significantlyhick PTFE slab probably due to edge effects, which is knowno increase the uncertainty in calibration of measurements [9].

1058 E. Hakansson et al. / Synthetic Metals 157 (2007) 1054–1063

Fs

TttsTwa

r1cm

3

Ldq(sfisiBp

cst[bf[alws

FL

co

diisic

ε

w

titrwaal

ig. 1. Permittivity responses for polytetrafluoroethylene (PTFE) and uncoatedubstrate textiles.

he response for the PTFE follows the value of relative permit-ivity of ε = 2.04 + 0i found in literature [2,14–16]. Similarly,he permittivity value for a polymethylmetacrylate (PMMA)heet is close to the known value of ε = 2.6 + 0i [2,14,15,17].he transmission measurement data show good correspondenceith tabulated values for well-known reference materials, hence

n ability to obtain accurate measurements is assumed.The permittivities for the uncoated fabrics show smooth

esponse with low values of complex permittivity between.0 + 0i and 1.5 + 0i. The uncoated substrate materials are notonducting hence resulting in almost 100% transmission oficrowave radiation [18].

.2. Permittivity response for polypyrrole coated textiles

The permittivity for 305 mm× 305 mm conducting Nylon-ycra textiles with different polymerization times, but sameopant concentration (0.018 mol/l pTSA), as a function of fre-uency are shown in Fig. 2. Both the real (ε′) and imaginaryε′′) parts of permittivity decrease as the frequency increases andhow a smooth response. As the polymerization time is increasedrom 5 to 120 min, ε′ for a sample doped with 0.018 mol/l pTSAncreases. The rate of increase in magnitude of ε′ is higher athort polymerization times. The imaginary part of permittivityncreases with increase in polymerization time up to 180 min.eyond this time, no significant increase in imaginary part ofermittivity is recorded.

The real part of complex permittivity undergoes a smallhange with an extension of the polymerization time (lowerurface resistivity), while the imaginary part increases substan-ially. This is in agreement with the results of Kuhn and Child4]. The real part of complex permittivity has been reported toe influenced by the topographic features of the coating sur-ace due to interfacial polarisation occurring in the material19–21]. However, the interfacial polarisation mechanism has

n average polarisation time of 10–2 s [17] and would mostikely not be influential at the high frequencies tested in thisork. The permittivity changes would more likely be a con-

equence of the effect of dopant, dopant–polymer interaction,

dalr

ig. 2. Permittivity response for 305 mm × 305 mm PPy-pTSA coated Nylon-ycra with different polymerization times. Concentration: 0.018 mol/l pTSA.

hain length, spatial organization of these molecules as a resultf the experimental parameters.

As the frequency of radiation increases, both ε′ and ε′′ecrease irrespective of the polymerization times. The responses frequency dependent. The resulting decrease with an increasen frequency gives the characteristic shape of the permittivitypectra that is displayed in Fig. 2. The values of relative imag-nary permittivity are directly proportional to the total (ac + dc)onductivity of the material and may be expressed as

′′ = iσtot

ω(17)

here, σtot is the total conductivity of the material.Similarly, both ε′ and ε′′ increase with the dopant concentra-

ion when the polymerization time is kept fixed at 180 min. Thencrease in �′ with dopant concentration may be indicative ofhe added dopant taking part in the charge storage in the mate-ial. A significant increase in the imaginary part of permittivityas observed in the Nylon-Lycra samples even when very small

mounts of pTSA dopant were added compared to only FeCl3cting as oxidant and dopant. The permittivity values trans-ate to percentages of reflection, transmission and absorption as

escribed above. The average values of reflection, transmissionnd absorption as well as total conductivity, total transmissionoss and shielding effectiveness across the 1–18 GHz frequencyange can be found for different polymerization times (at fixed

E.H

akanssonetal./Synthetic

Metals

157(2007)

1054–10631059

Table 2Reflection, transmission and absorption at different polymerization times for PPy-pTSA coated Nylon-Lycra

Polymerization time (min) Average reflection (%) Average transmission (%) Average absorption (%) Total conductivity (S/m) Total transmission loss (%) Shielding effectiveness (dB)

Uncoated 0.086 99.68 0.23 0 <1 05 1.62 78.26 20.12 1.27 21.74 −1.0615 8.30 53.77 37.93 3.56 46.23 −2.6930 14.99 39.84 45.18 5.76 60.16 −4.0060 19.11 33.31 47.59 7.21 66.69 −4.77120 26.98 24.23 48.78 10.19 75.77 −6.16180 32.65 19.49 47.85 10.25 75.95 −6.19300 33.72 18.46 47.82 13.10 81.54 −7.34

Concentration: 0.018 mol/l.

Table 3Reflection, transmission and absorption for different dopant concentrations for PPy-pTSA coated Nylon-Lycra

Dopant concentration (mol/l pTSA) Average reflection (%) Average transmission (%) Average absorption (%) Total conductivity (S/m) Total transmission loss (%) Shielding effectiveness (dB)

No dopant 4.93 63.49 31.6 2.49 36.51 −1.970.004 15.02 39.65 45.33 5.76 60.35 −4.020.009 20.88 31.28 47.84 7.75 68.72 −5.050.018 27.36 24.05 48.59 10.25 75.95 −6.190.027 31.07 20.66 48.27 11.87 79.34 −6.850.036 31.37 20.49 48.14 11.92 79.51 −6.89

Polymerization time: 180 min.

1060 E. Hakansson et al. / Synthetic Metals 157 (2007) 1054–1063

FmP

dfir

hld

3

3

tdddtstpitihi(

tFsp

3S

im

F

arspsaiasmrttfm5cimambient conditions did not significantly influence the measure-ments.

ig. 3. Raw diffraction and resultant data from time domain transmissioneasurement on 305 mm × 305 mm sample PPy-pTSA coated Nylon-Lycra.olymerization time 180 min, 0.027 mol/l.

opant concentration) and different dopant concentrations (atxed polymerization time) can be found in Tables 2 and 3,espectively.

Although some of the conducting polymer samples testedave relatively high conductivity they still have significantlyower conductivity than metallic conductors, which have con-uctivity in the order 60 × 106 S/m.

.3. Evaluation of the accuracy of the free space method

.3.1. Diffraction patternsThe raw data from the send horn, the diffraction data, and

he resultant data in the time domain for a small sample withimensions 305 mm × 305 mm are presented in Fig. 3. The rawata was obtained from the sample measurement, the diffractionata was obtained when a perfect reflector with the same size ashe sample under test blocked the radiation path and the resultantignal was obtained by the removal of the diffraction data fromhe raw data. It is obvious that the diffraction signal at small sam-le size will influence the permittivity calculations significantlyf it is not removed from the raw signal. It can also be concludedhat an adequate portion of the response signal from the samples detected within the time-gate span used, since the signal isarmonic at both ends of the time span. The diffraction patterns similar, but much less pronounced, for the larger sample size500 mm × 500 mm).

The diffraction trace obtained with the smaller perfect reflec-or is significantly higher than for the larger one, as seen inig. 4. Slight variations in the diffraction traces from differentize reflectors may be caused by currents induced in the metallates being re-radiated to the receive horn.

.3.2. The effect of variations in experimental conditions on

21 response

The calibration made for the free space transmission methods only valid for a short time since the values of phase and

agnitude of the sample change as a result of changes in the

Fec

ig. 4. Diffraction signal in time domain from two sizes of perfect reflectors.

mbient conditions [2]. The differences in slip in magnitudeesponse of S21 for a 305 mm square PPy-coated Nylon-Lycraample are displayed in Fig. 5. The ‘initial’ measurement waserformed immediately after calibration, while the other mea-urements were performed after set time delays of up to 15 minfter calibration (y = 0 corresponds to initial measurement). Its obvious that there is an effect of time between calibrationnd measurement. The data at the low frequency end are moreensitive to change in ambient conditions since the change inagnitude is largest in this range. The difference in magnitude

emains within 0.02 dB of the measurement just after calibra-ion. The slip in phase of S21 is generally more significant thanhe slip in magnitude but is within ±0.6◦ for a typical sampleor delays of up to 15 min between calibration and measure-ent. The differences are small at short delay times (less thanmin) after calibration, while an increase in the time betweenalibration and data collection gives larger deviations. The cal-bration was repeated in 5 min intervals during transmission

easurements. This ensured that the errors due to changes in

ig. 5. Slip in S21 magnitude for PPy-pTSA coated Nylon-Lycra at differ-nt time delays after initial calibration (y = 0). Polymerization time 180 min,oncentration 0.027 mol/l.

E. Hakansson et al. / Synthetic Me

FN

3

titivtpisoil

2mtpa

dp

FN

hitpfo

3u

tattatdvimibRcpuaniav(r

ig. 6. Sample size effect on the relative permittivity for PPy-pTSA coatedylon-Lycra. Polymerization time 120 min, 0.018 mol/l.

.3.3. The effect of sample sizeThe accuracy of the free space transmission measurement in

he 1–18 GHz frequency range increases as the sample size isncreased, due to a smaller total amount of diffraction bypassinghe sample. This can be confirmed by looking at the permittiv-ty of the different sized samples; smaller samples show moreariability in values of both real and imaginary parts of permit-ivity compared with the larger ones. A typical example of theermittivity response for these two different size samples fromdentical polymerization conditions can be seen in Fig. 6. Themoother response for the larger samples highlights the effectsf sample size on the amount of noise in the response. Signif-cantly less residual influence from diffraction is present in thearger sample.

The average absorption in 1 GHz frequency spans (1–2 GHz;–3 GHz, etc.) are displayed in Fig. 7. Smaller sample showsore scatter in the absorption data and larger standard devia-

ion compared to the larger one. Batch variations from identicalroduction settings result in the small differences present in

bsorption levels between the samples.

The thickness measurement is a source of error in allielectric measurements, particularly in the case of conductingolymer coated fabrics. Since the fabric is compressible it is

ig. 7. Sample size effect on variation in absorption levels for PPy-pTSA coatedylon-Lycra. Polymerization time 120 min, 0.018 mol/l.

3m

n

Fri

tals 157 (2007) 1054–1063 1061

ard to determine the true thickness. Moreover, since the coat-ng is on the surface of the fabric and complete penetration ofhe conducting polymer into the inner layers of the fabric is notresent, thickness determination is approximate. However, theabric thickness measurement was averaged over a large numberf measurements and considered to be sufficiently accurate.

.4. Validation of free space transmission method resultssing the ‘reflection only’ method

The transmission only technique can be used to calculatehe reflection magnitude via the formulae suggested by Bal-nis [13]. The accuracy of the reflection from calculations inhe transmission method has been compared with the reflec-ion obtained using the ‘reflection only’ free space techniquend evaluated in terms of similarity and deviations. The ‘reflec-ion only’ method requires additional calibration and utilizesifferent extraction formulae. The reflection measurement pro-ides a good check of the transmission measurement becauset is an entirely different method. Reflection measurements are

ore difficult to perform accurately because of the phase shiftn reflection that can occur in free space measurements due toending of the sample holder or non-flatness of the sample.eflection and transmission measurements are required for cal-ulating permittivity and permeability, but since the (relative)ermeability is 1, it is not necessary. A different algorithm issed to calculate permittivity from reflection measurements; ifdifferent result came out of that calculation then the tech-

ique would need to be looked at. The transmission techniques generally the best method to use as it offers high accuracynd is relatively simple to perform. Unless the sample size isery small (less than a wavelength) or extremely absorptivetransmission <−30 dB), the technique usually provides goodesults.

.4.1. Calculated and measured values of reflectionagnitudeFig. 8 shows the measured and calculated reflection mag-

itudes for two PPy-coated Nylon-Lycra samples with two

ig. 8. Measured (from reflection) and calculated (from transmission) values ofeflection magnitude for PPy-pTSA coated Nylon-Lycra at different polymer-zation times. Concentration: 0.021 mol/l.

1062 E. Hakansson et al. / Synthetic Me

Fnt

drTtttp

ehwfrasca

3r

ri0aTfrt

ant

4

pi

trw1nipifrTibt

fmaic

cTSemew0

ifle0Trgrhedowe

utctaof

ig. 9. Percentage difference between calculated and measured reflection mag-itudes for PPy-pTSA coated Nylon-Lycra with 60 and 180 min polymerizationime. Concentration: 0.021 mol/l.

ifferent polymerization times; 60 and 180 min. Values foreflection at the low and high frequency ends are indicated.he comparison shows that the values from calculated reflec-

ion magnitudes (from transmission) show less variation thanhe actual measurements of reflection magnitudes. This is dueo the higher sensitivity in reflection measurements to samplelacements, air-gaps, etc.

At a short polymerization time of 60 min the reflection lev-ls are very low. The longer polymerization times of 180 minave higher reflection. The reflection values for the sampleith 180 min polymerization time increases from 8.2% at low

requencies to 17.5% at high frequencies. The correspondingeflection values for a 60 min polymerization time are 2.1%nd 4.5%. In general it can be said that the calculated valueslightly overestimate the amount of reflection at higher frequen-ies whereas the reflection is well estimated by the calculationst lower frequencies.

.4.2. Difference between calculated and measuredeflection magnitudes

The percentage differences between calculated and measuredeflection magnitudes for sample with 60 and 180 min polymer-zation times are presented in Fig. 9. The error was less than.4 dB throughout the frequency range for the 180 min samplend only exceeded 0.5 dB for the 60 min sample above 13 GHz.he accuracy was slightly better for the sample polymerized

or 60 min. The calculated values underestimated the amount ofeflection up to approximately 10 GHz, where a change in trendo an over-estimation of the reflection occurred.

It can be concluded that the error between the calculatednd measured reflection magnitudes is usually below 2% andever exceeds 8%, which confirms the accuracy of the free spaceransmission test method used.

. Discussion and conclusions

Both the polymerization time and the dopant concentrationlay major roles in the dielectric properties of the conduct-ng textiles. By monitoring these parameters, it is possible to

R

tals 157 (2007) 1054–1063

une the dielectric characteristics to a certain extent. Both theeal and imaginary parts of the complex permittivity increaseith polymerization time, with the real part stabilizing beyond20 min. Additionally, the real part of permittivity does not sig-ificantly change with an increase in frequency beyond 12 GHzrrespective of the polymerization time. The imaginary part ofermittivity, on the other hand, continues to change with increasen the polymerization time. It also changes throughout the wholerequency range. The values of the real part of permittivityemain stable above dopant concentrations of 0.018 mol/l pTSA.he imaginary part of permittivity increases with an increase

n dopant concentration up to concentrations of 0.018 mol/l,eyond which further changes contributes little to an increase inhe imaginary part of permittivity.

The permittivity levels for conducting textiles vary withrequency. The real part of permittivity increases with poly-erization time and dopant concentration, reaching a plateau

t certain time-dopant concentration combinations whereas themaginary part of permittivity exhibits a frequency dependenthange throughout the tested range.

The response from the free space measurement systemhanges continuously due to changes in ambient conditions.he drift in S21 phase is more significant than the drift in21magnitude. Calibration of transmission free space set-upvery 5 min is sufficient to avoid large deviations in measure-ents. Magnitude and phase data for S21 at the low frequency

nd is more sensitive to change in ambient conditions comparedith the data at the high frequency end. Deviation is within.02 dB for S21 magnitude and ±0.6◦ for S21 phase.

The sample size will affect the amount of diffraction bypass-ng the sample, which in turn determines the amount ofuctuation in permittivity values. It is possible to reduce theffects of diffraction by applying a narrow time-gate span of.15 ns instead of 1.0 ns for the 305 mm × 305 mm samples.he data obtained with the narrower gate displays a smoother

esponse with less diffraction noise. The permittivity values forate spans of 0.15 ns are only valid in the range 6.7–18 GHz. Theesponse from transmission measurements carried out using theigh gain horns, operating in the 7.5–18 GHz frequency range,ntails almost no fluctuation. This is indicative of very smalliffraction errors. The re-gating of the signal is successful alson textured fabrics. No significant problems seem to appearith layered structure fabrics with sheets of lower conductivity

mbedded between outer, more conductive layers.Good correlation between the calculated and measured val-

es of reflection magnitude demonstrates the effectiveness of theransmission measurement method. The deviations between thealculated and measured values are within 8%. This is indica-ive of the free space transmission test method having adequateccuracy for thin flexible samples. Due to the imprecise methodf measuring reflection on what is essentially a lightweight,oldable fabric, the estimate of less than 8% is realistic.

eferences

[1] B. Chambers, Electron. Lett. 30 (1994) 1626.[2] A. Amiet, PhD Thesis, Monash University, Australia (2003).

tic Me

[[

[[

[

[

[

[

[

[

UK, 1993, p. 934.[20] T.C.P. Wong, B. Chambers, A.P. Anderson, P.V. Wright, Electron. Lett. 28

E. Hakansson et al. / Synthe

[3] L. Rupprecht, Cond. Polym. Plast. Ind. Appl. (1999) 143.[4] H.H. Kuhn, A.D. Child, in: T.A. Skotheim, R.L. Elsenbaumer, J.R.

Reynolds (Eds.), Handbook of Conducting Polymers, Marcel Dekker, NewYork, 1998, p. 993.

[5] H.H. Kuhn, A.D. Child, W.C. Kimbrell, Synth. Met. 71 (1995) 2139.[6] A. Kaynak, R. Beltran, Polym. Int. 52 (2003) 1021.[7] E. Hakansson, A. Amiet, A. Kaynak, Synth. Met. 156 (2006) 917.[8] F.C. Smith, B. Chambers, J.C. Bennett, IEE Proc A: Sci, Meas. Technol.

139 (1992) 247.[9] F.C. Smith, B. Chambers, J.C. Bennett, IEE Proc.: Sci. Meas. Technol. 141

(1994) 538.10] F.J. Harris, Proc IEEE 66 (1978) 51.11] A.C. Metaxas, R.J. Meredith, Industrial Microwave Heating, Peter Pere-

grinus Ltd., London, 1983.

12] A.M. Nicolson, G.F. Ross, IEEE T Instrum. Meas. 19 (1970) 377.13] C.A. Balanis, Advanced Engineering Electromagnetics, John Wiley and

Sons, 1989.14] J. Baker-Jarvis, E.J. Vanzura, W.A. Kissick, IEEE T Microw. Theory 38

(1990) 1096.

[

tals 157 (2007) 1054–1063 1063

15] J. Baker-Jarvis, R.G. Geyer, J.H. Grosvenor, M.D. Janezic, C.A.Jones, B. Riddle, C.M. Weil, IEEE T Dielect. Insul. 5 (1998)571.

16] Z. Abbas, R.D. Pollard, R.W. Kelsall, IEEE Trans. Instrum. Meas. 50 (2001)1334.

17] H. Senko, ‘Investigation into and control of microwave induced thermalrun-away’, PhD Thesis, Faculty of science and technology, Deakin Uni-versity (1997).

18] E. Hakansson, A. Amiet, S. Nahavandi, A. Kaynak, Eur. Polym. J. 43 (2007)205.

19] T.C.P. Wong, B. Chambers, A.P. Anderson, P.V. Wright, Proceedings ofthe 8th International Conference on Antennas and Propagation, Edinburgh

(1992) 1651.21] P.V. Wright, T.C.P. Wong, B. Chambers, A.P. Anderson, Adv. Mater. Opt.

Electron. 4 (1994) 253.