Journal of Colloid and Interface Science 286 (2005) 526–535 www.elsevier.com/locate/jcis Colloid stability of synthetic titania and the influence of surface roughness David R.E. Snoswell, Jinming Duan, Daniel Fornasiero, John Ralston ∗ Ian Wark Research Institute, University of South Australia, Mawson Lakes, Adelaide, SA 5095, Australia Received 3 August 2004; accepted 21 January 2005 Available online 10 March 2005 Abstract The colloid stability of synthetic titania particles was studied as a function of KCl concentration at pH values of 6.3, 6.7, and 8.4, using static light scattering to obtain stability ratios. Standard DLVO theory was then used to calculate the stability ratios as a function of salt concentration. Reasonable agreement between theory and experiment could only be obtained if an effective interaction radius, corresponding to surface asperities on the titania particles, was used in the calculation. High-resolution TEM images suggest that the effective interaction radius corresponds to the size of surface crystallites formed during synthesis.  2005 Elsevier Inc. All rights reserved. Keywords: Titania; Surface roughness; Colloid stability; Zeta potential 1. Introduction Deviations from classic DLVO behavior have been at- tributed to a wide selection of causes including discreteness of surface charge [1], steric and relaxation effects [2,3], the presence of nanobubbles on particles [4–6], and the inherent surface roughness of the particles themselves [7–10]. The is- sue of surface roughness is of particular interest here, since it is highly relevant when describing real particles. Theoretical stability ratios calculated using classical DLVO theory can be plotted on a log–log graph of sta- bility ratio versus electrolyte concentration to generate a “stability curve.” These curves comprise two sections; a di- agonal section at low electrolyte concentration below the ccc (critical coagulation concentration), where the stability ratio varies, and a horizontal section above the ccc. DLVO theory often overestimates the experimental rate of aggrega- tion [11], and consequently stability curves generated from DLVO theory have a much steeper gradient in the slow aggregation regime below the ccc, compared with experi- mental data. Classical DLVO theory assumes that colloidal particles are perfect spheres; hence numerous theoretical * Corresponding author. E-mail address: john.ralston@unisa.edu.au (J. Ralston). studies have demonstrated ways in which DLVO theory can be modified to model surface asperities found on real parti- cle surfaces [8–10,12]. In particular, some of these studies show that the gradient of the theoretical stability curve is re- duced markedly when surface roughness is considered in the calculations [12,13]. DLVO theory predicts that the gradient of the stability curve will depend strongly on the particle radius, a predic- tion which was not observed in the early pioneering experi- ments of Ottewill and Shaw [14]. The latex particles used in their study [14] were thought to be ideal for the verification of DLVO predictions of colloidal stability, since they were monodisperse and spherical and could be manufactured in various sizes. However, more recent analyses of similar latex particles reveal they have “hairy” surfaces due to extended polymer chains [15]. This leads to changes in dimension with changes in pH, electrolyte concentration, and temper- ature, contributing to small scale surface roughness [15]. These results suggest that the effect of macroscopic particle size may be masked by the presence of surface roughness at a much smaller scale. The effect of surface roughness on DLVO calculations has also been noted in AFM force measurements. For ex- ample, Drummond and Senden discuss this issue while in- vestigating the geometry of AFM silicon nitride cantilever 0021-9797/$ – see front matter  2005 Elsevier Inc. All rights reserved. doi:10.1016/j.jcis.2005.01.056