Thin Solid Films 451–452(2004) 37–42

0040-6090/04/$ - see front matter� 2003 Elsevier B.V. All rights reserved.doi:10.1016/j.tsf.2003.10.151

Microstructure and electrical properties of plasma sprayed porous TiO2

coatings containing anatase

X.Y. Wang *, Z. Liu , H. Liao , D. Klein , C. Coddeta,b, b a a a

Laboratoire d’Etudes et de Recherches sur les Materiaux, les Plasmas et les Surfaces, Universite de Technologie de Belfort-Montbeliard,a ´ ´ ´90 010 Belfort Cedex, France

Corrosion and Protection Centre, University of Manchester Institute of Science and Technology (UMIST), P.O. Box 88, Manchester M60 1QD,b

UK

Abstract

Electrical properties of TiO are very sensitive to microstructure and phase composition. In the present study, plasma sprayed2

porous TiO coatings containing anatase were prepared using anatase powder and Ar–He–HyN plasma gases. The microstructure2 2 2

and phase composition of the coatings were characterized by opticalyscanning electron microscopy and X-ray diffraction. Theelectrical properties of the coatings were investigated by impedance spectroscopy. The prepared coatings were porous, havingporosities from 32 to 43% and pores from 1 to 80mm in diameter. Their phases were mainly rutile, plus 10.0–22.5% anatase byweight. It was evident that the deoxidisation occurred for both the anatase and rutile phases in the coatings. The electricalproperties were clearly correlated with the microstructure and phase contents of the coatings.� 2003 Elsevier B.V. All rights reserved.

Keywords: Plasma spraying; TiO ; Microstructure; Phase; Electrical property2

1. Introduction

Compared with other coating techniques, such as thesol–gel method, plasma spraying is a cost-effective andhighly flexible means for depositing coatings, fromseveral micrometers to millimeters in thickness on sub-strates of complicated shapes and with a large areaw1x.Recently, plasma spraying using anatase particles asstarting powder has been investigated for producingTiO coatings containing anatase phasew2x. The result-2

ing coating contained 15% anatase phase by weight andappeared promising for photocatalystic applications.In the previous studyw2x, attention was paid to

photocatalytic activity and photo-absorption of the coat-ings. The microstructures of the coatings were givenbriefly and the effects of process parameters on thecharacterization of the coatings were not taken intoaccount. The photocatalytic performance of anatase isgenerally considered to be superior to that of rutilew2x.However, as a result that molten TiO tends to form2

*Corresponding author. Tel.:q44-161-200-4129; fax:q44-161-200-4865.

E-mail address: x.wang-4@umist.ac.uk(X.Y. Wang).

rutile during solidification, plasma sprayed TiO coat-2

ings containing anatase unavoidably contain certainamounts of another polymorphic phase of TiO , rutile.2

On the other hand, because of the cooling rate of themolten or partially molten droplets in excess of 10 Ky6

s and low oxygen partial pressure during plasma spray-ing, deoxidisation of TiO in the sprayed coatings has2

been often observedw3–5x. The deoxidisation of TiO2modifies the valence state of atoms, and hence mayaffect the properties of the coatings, including photoca-talystic activity. However, the deoxidisation of TiO in2

plasma sprayed TiO coatings has not been characterized2

in detail so far; consequently, no attention has been paidto the effect of the deoxidisation of TiO in the coatings2

on the photocatalystic activity. Therefore, it is importantto further characterize the microstructure of plasmasprayed TiO coatings containing anatase, considering2

the effects of process parameters on phase content andthe deoxidisation of TiO .2

In this paper, porous TiO coatings containing anatase2

were prepared by plasma spraying using anatase powderand several sets of process parameters. Microstructuresof the coatings were characterized using optical andscanning electron microscopes(SEM) in terms of mor-

38 X.Y. Wang et al. / Thin Solid Films 451 –452 (2004) 37–42

Fig. 1. Secondary electron SEM image showing the size and mor-phology of the starting anatase TiO powder.2

Table 1Coating sample codes and the corresponding plasma spray parameters

Code Plasma power Plasma gas(lymin) Spray distance Powder feed rate(kW)

Ar He N2 H2(mm) (gymin)

AH00a 33 19 128 – – 100 30AHN1 33 19 80 1.6 – 100 30AHN2 33 19 60 3.2 – 100 30AHH1 33 19 90 – 1.0 100 30AHH2 33 19 80 – 2.0 100 30

No coating could be produced with this set of parameters.a

phology, porosity and pore size distribution. The phasesof the coatings were analysed using X-ray Rietveldmethod to determine the phase contents and degrees ofdeoxidisation of TiO w6x. As electrical properties of2

TiO are very sensitive to microstructure and phases2

w5,7x, the electrical properties of the coatings were alsoinvestigated by impedance spectroscopy to provide fur-ther evidence on the microstructure and crystallinephases of the coatings.

2. Experimental procedures

TiO coatings were applied onto stainless steel plates2

of 100=30=2 mm by an atmospheric plasma system,3

consisting of a Sulzer-Metco F4-MB plasma gun(Sulz-er-Metco, Switzerland) mounted on an ABB robot(ABB, Sweden). Main process parameters are listed inTable 1. As molten TiO tends to form rutile during2

solidification, helium gas was deliberatively employedto reduce the heat intensities of the plasma jet, and thusto increase opportunities of partially molten particles inthe form of anatase phase embedded within the coatings.The minor amounts of hydrogen and nitrogen gaseswere added into the plasma gases for adjusting theintensities of the plasma jetw8x. The starting powderconsisted of typically spherical aggregations of anataseparticles fabricated by spray drying(Fig. 1). Prior tothe spraying, the stainless steel plates were grit-blastedwith alumina abrasives. During spraying, the traversevelocity of the plasma gun relative to the stainless steelplates was 500 mmys.The microstructures of the coatings were observed on

both the fractured and polished cross-sections usingSEM. Meanwhile, the porosities and pore size distribu-tions of the coatings were measured on the polishedcross-sections using optical microscopy, in conjunctionwith a digital camera and image analysis methodw9x.For preparation of the polished cross-sections, the sam-ples were mounted with epoxy resin cast at pressurebelow 30 mbar. The deoxidisation of TiO and phase2

contents of the coatings were analysed by the X-rayRietveld method using the programRIETICA for Win-dows, version 1.7.7w6x. The related X-ray diffraction

data collection and analysis procedures have beendescribed in detail elsewherew10x. The electrical prop-erties of the coatings were studied using an impedancespectroscopy. The stainless steel substrates were polishedto act as electrodes. The coating surfaces were cleanedin an ultrasonic bath with ethanol and then painted witha silver paint acting as the counter electrode. Theelectrode areas painted were 3=3 mm . Prior to the2

measurements, the silver paints were fired at 4508C for30 min to increase the adhesion of the paint to thecoatings. Impedance measurements were conductedusing a Solartron SI 1255 HF frequency response ana-lyser coupled with a Solartron 1296 dielectric interface.An alternating current(AC) voltage with a 0.1 Vamplitude was applied to the specimen with the ACfrequencies of 1–1=10 Hz. The temperature was kept6

at 400 8C so as to eliminate the effect of moisture onthe electrical properties of the coatings. The spectrawere analysed using a software Zview impedance anal-ysis (Scribner Associates Inc., Southern Pines, NC) toextract the electrical properties of the TiO coatings. For2

each coating sample, three measurements were repeatedto obtain an average value.

39X.Y. Wang et al. / Thin Solid Films 451 –452 (2004) 37–42

Fig. 2. Secondary electron SEM images of the polished cross-sections of coatings AHN2(a) and AHH2 (b), showing the porous and lamellarstructure with the smaller pores within individual lamellae and larger pores along the interlamellar boundaries.

Fig. 3. Secondary electron SEM micrographs of the fractured cross-sections of coatings AHN1(a) and AHH1(b), showing the lamellar structurewith partially molten particles embedded between the lamellae. Note the arrows pointing at the lamellae, molten particles.

Fig. 4. Plots of the cumulative probability vs. the apparent diameterof pore of the four coatings. Note that these plots are drawn on thenormal probability co-ordinate.

3. Results and discussion

3.1. Results

No coatings could be produced using the set ofprocess parameters AH00 listed in Table 1, because the

plasma jet was too weak to melt the starting powderparticles w8x. The four coatings, produced using theother four sets of process parameters, are porous andhave the lamellar structure of plasma sprayed coatings.Smaller pores are located within individual lamellae andlarger pores are found along the interlamellar boundaries.There are also a number of partially molten powderparticles embedded between the lamellae(Figs. 2 and3). During the plasma spraying, a lower nitrogen orhydrogen flow rate obviously led to a lower depositionefficiency of the starting powder particles. Thus, variousspraying processes have been applied on the four coatingsamples to ensure comparable coating thicknesses in therange of 200–250mm.The porosities of the four coatings AHN1, AHN2,

AHH1 and AHH2 are 42.7, 32.3, 43.2 and 31.3%,respectively. A lower nitrogen or hydrogen flow ratealso leads to a higher porosity. The pore size is distrib-uted from 1 to 80mm for coatings AHN1 and AHH1,and from 1 to 40mm for coatings AHN2 and AHH2.Apparent transitions appear at the apparent pore diameterapproximately 4mm, for all the plots of the cumulativeprobability vs. the apparent pore diameter of the fourcoatings(Fig. 4). A cumulative probability represents a

40 X.Y. Wang et al. / Thin Solid Films 451 –452 (2004) 37–42

Table 2Refined parameters of the atom occupancies and phase contents for the starting powder and the four coatings by the X-ray Rietveld method

Sample Powder AHN1 AHN2 AHH1 AHH2

Anatase Occ.(O )y2 0.9776(40) 0.9476(96) 0.9496(184) 0.9392(88) 0.9508(116)Occ. (Ti )q4 0.9104(76) 0.7896(384) 0.7984(736) 0.7576(344) 0.8032(464)Occ. (Ti )q3 0.0896(76) 0.2104(384) 0.2016(736) 0.2424(344) 0.1968(464)TiO2yx TiO2y0.045 TiO2y0.105 TiO2y0.101 TiO2y0.122 TiO2y0.098

Wt.% 100 20.51(44) 10.05(0.36) 22.46(0.44) 17.58(0.44)

Rutile Occ.(O )y2 0.9588(32) 0.9620(36) 0.9600(64) 0.9600(81)Occ. (Ti )q4 0.8344(136) 0.8480(136) 0.8408(128) 0.8408(136)Occ. (Ti )q3 0.1656(136) 0.1520(136) 0.1592(128) 0.1592(136)TiO2yx TiO2y0.082 TiO2y0.076 TiO2y0.080 TiO2y0.080

Wt.% 79.49(96) 89.95(1.20) 77.54(0.91) 82.42(1.06)

Fig. 5. Representative complex impedance diagrams of the four coat-ings, measured at 4008C.

Fig. 6. Plots of apparent porosity and anatase content vs. resistivityfor the four coatings. The error bars of both the porosity and theanatase content stand for the 95% confidence intervals, and those ofthe resistivity mark the ranges of the measured data.

percentage of the number of pores whose apparentdiameters are less than a certain value, relative to thetotal number of the pores within a coating. The proba-bility distributions of pore size for the four coatings areapproximately identical at the apparent pore diametersless than 4mm. Significant differences of probabilitydistributions of the pore size for the four coatings areexhibited at the apparent pore diameters larger than 4mm. The coatings AHN1 and AHH1 with larger poros-ities have wider pore size distributions than the coatingsAHN2 and AHH2 with smaller porosities.The deoxidisation of both the anatase and rutile

phases in the coatings was evident from the reducedoxygen occupancy compared with that of stoichiometricTiO . The degrees of the deoxidisation and the other2

crystalline parameters of the anatase and rutile phasesin the coatings, obtained by the Rietveld method,appeared independent of the process parametersw10x.The phase contents of the coatings are clearly influencedby the process parameters, as seen in Table 2. Thecoatings mainly consisted of rutile, plus 10.0–22.5%anatase by weight, as determined from the scale factors.The anatase contents increase with an increase in thehelium flow rate.

The Nyquist plots measured for the four coatings allexhibit only one semicircle(Fig. 5). Their resistancesRare thus directly determined from the intercepts of theplots with the realZ-axis, shown by the diameter ofsemicirclesw7x. Then the resistivities of the coatings arederived from the resistances, being multiplied by theelectrode area and divided by the coating thickness. Theresistivities of the four coatings are in good linearcorrelation with the anatase contents, but in poor linearcorrelation with porosities of the coatings(Fig. 6).

3.2. Discussion

The helium flow rate in the present study could beapproximately used as indication of the intensity of thecorresponding plasma jet(Table 1). It can hence beconcluded that the anatase contents of the coatingsincrease with decreasing the intensity of the plasma jet,and are not significantly influenced by the type of theminor constitutes of the plasma gases, nitrogen and

41X.Y. Wang et al. / Thin Solid Films 451 –452 (2004) 37–42

hydrogen(Table 2). However, as observed during thespraying, the anatase content of 22.5% is close to themaximum content. It is impossible to produce satisfac-tory coating if the plasma jet is too weak, i.e. the heliumflow rate is further increased.A larger porosity can improve the photocatalytic

properties of the coatings because it can increase react-ing area. Using Ar–He–N plasma gas could produce2

coatings having relatively larger porosities compared tocoating made using Ar–He–H plasma gas, provided2

that other process conditions are approximately identical.The probability distributions of pore size of the fourcoatings (Fig. 4) suggest that larger pores along theinterlamellar boundaries are the most important contri-bution for the effects of the process parameters on theporosities of the coatings.The deoxidisation and formation of low-valence tita-

nium cations during the plasma spraying transforms therutile TiO in the coatings into similar n-type electronic2

semiconductors with low resistancesw11,12x. On theother hand, as molten TiO tends to form rutile during2

solidification, the presence of anatase TiO in the coat-2

ings is in the form of partially molten powder particlesembedded mainly between the rutile lamellae(Fig. 3).Such a form of anatase TiO has weak bonding with the2

rutile TiO , and hence shows similar response as the2

interlamellar poresyboundaries in the coatings whensubjected to the resulting current for the impedancemeasurement. Therefore, the Nyquist plots measured forthe four coatings all exhibit only one semicircle, corre-sponding to an equivalent circuit system that has onlyone resistance and capacitance parameters, associatedwith interlamellar poresyboundaries and anatase phasein the coatings. Both the interlamellar poresyboundariesand the anatase phase in the coatings would increasethe resistivities of the coatings. As the measured anatasecontent appeared to be proportional to the sum of theanatase content and apparent porosity, the resistivitiesare in good linear correlation with the anatase contents,but in poor linear correlation with porosities of thecoatings(Fig. 6). It can be thus seen that the impedancemeasurement can provide further information on themicrostructure features and crystalline phases of thecoatings.Compared with the previous result of plasma sprayed

TiO coating using anatase feedstockw2x, the present2

work has further improved the anatase content approxi-mately 7% in a coating, AHH1. The previous coating,containing 15% anatase phase by weight, showed excel-lent photocatalystic performancew2x. In addition to theanatase phase incorporated within the coating, porousmicrostructure and the formation of low-valence titani-um cations, due to the deoxidisation during the plasmaspraying, are probably also important factors to promotephotocatalystic performance. However, these need to befurther investigated. Compared to the results of sinteredTiO samplesw7,13x, the resistances andyor resistivities2

of the present coatings are apparently smaller. This canbe attributed to the fact that the n-type electronicsemiconductors rutile lamellae with low resistances inthe coatings are more or less interconnected despite thatthere exist a number of interlamellar poresyboundariesand anatase phases in the coatings. The relatively lowerresistivities of the present coatings may be used as asensing material having unique properties such as highsensitivity and selectivity.

4. Conclusion

Plasma sprayed porous TiO coatings containing ana-2

tase phase have been prepared using the anatase powderand different process parameters. The porosities of thecoatings ranged from 31.3 to 43.2% and the pore sizefrom 1 to 80mm. A lower nitrogen or hydrogen flowrate tended to produce coating with a higher porosity.Using Ar–He–N plasma gas, the produced coatings2

had relatively larger porosities than the coatings preparedusing Ar–He–H plasma gas.2

The degrees of the deoxidisation and the other crys-talline parameters of the anatase and rutile phases in thecoatings refined by the Rietveld method appeared inde-pendent of process parameters. The anatase contents ofthe coatings were in the range of 10.0–22.5% by weight,and increased with increasing helium flow rate in theplasma gases.The Nyquist plots measured for the four coatings all

exhibited only one semicircle, associated with the inter-lamellar poresyboundaries and anatase phase in thecoatings, corresponding to an equivalent circuit systemthat has only one resistance and capacitance parameters.The resistivities were in good linear correlation with theanatase contents, but in poor linear correlation with theporosities of the coatings.

Acknowledgments

The authors are grateful to Dr P. Xiao of MaterialsScience Centre of the University of Manchester andUMIST for his help in impedance measurements.

References

w1x R. McPherson, Thin Solid Films 83(1981) 279.w2x F. Ye, A. Ohmori, Surf. Coat. Technol. 160(2002) 62.w3x R.W. Smith, R. Novak, Powder Metall. Int. 3(1991) 147.w4x J.F. Li, C.X. Ding, J. Mater. Sci. Lett. 17(1998) 1747.w5x Y. Zhu, C. Ding, J. Eur. Ceram. Soc. 20(2000) 127.w6x B.A. Hunter, C.J. Howard, A Computer Program for Rietveld

Analysis of X-ray and Neutron Powder Diffraction Patterns,Lucas Heights Research Laboratories of Australian NuclearScience and Technology Organization,(2000) anonymous ftpsite: ftp.ansto.gov.auypubyphysicsyneutronyrietveld.

42 X.Y. Wang et al. / Thin Solid Films 451 –452 (2004) 37–42

w7x S.-H. Song, X. Wang, P. Xiao, Mater. Sci. Eng. B 94(2002)40.

w8x L. Powlowski, The Science and Engineering of Thermal SprayCoatings, Wiley, New York, 1995.

w9x J. Li, H. Liao, X. Wang, C. Coddet, J. Am. Ceram. Soc. 86(11) (2003) 1906.

w10x X.Y. Wang, Z. Liu, H. Liao, D. Klein, C. Coddet, Deoxidisationand phase analysis of plasma sprayed TiO by X-ray Rietveld2

method, Thin Solid Films, submitted for publication.w11x T. Oyama, Y. Iimura, K. Takeuchi, J. Mater. Sci. 34(1999)

439.w12x P. Knauth, H.L. Tuller, J. Appl. Phys. 85(1999) 897.w13x A.M. Azad, L.B. Younkman, S.A. Akbar, J. Am. Ceram. Soc.

77 (1994) 481.