Quantification of Electronic Activity Inside Photo-Activated TiO₂ Layers through a New Electrical Model Supported by Electrokinetic Data

Zeta potential is one of the most important magnitudes to determine stability of suspensions in different industries, such as ceramic casting, cements and plaster, brickmaking and pottery, paper coatings, catalyst supports, water treatment and food processing, amongst others. Rani et al.¹ have recently reported, in environmental sciences, that photocatalytic degradation of toxic phenols from water using bimetallic metal oxides increases as the negative zeta potential of the oxides is greater. In the food industry, zeta potential also plays a fundamental role in the quality control of new nanoparticles designed for the encapsulation of active substances² or in molecular studies involving proteins such as ovomucin.³ In marine science, in the absence of macromolecular interactions, adhesion forces between sea-water bacteria and titanium surfaces are governed by (unspecific) electrostatic interactions,⁴ which are stronger in the areas of the titanium surface with a higher amount of microstructures.

It is also known that electrical surface charge has an important influence on bioadhesion processes on the surface of medical devices. In particular, electrical interactions can promote osseointegration and prevent bacterial colonization of an implant surface. Lee et al.⁵ have recently demonstrated that grafting of different polymers on PMMA (Poly(methyl methacrylate)) was effective against bacterial adhesion because these coatings increased the negative surface charge of PMMA and so caused coulomb repulsion between the negatively charged bacteria and the grafted surfaces. Teranaka et al.⁶ have also reported that zeta potential was involved in the generation of polymicrobial biofilms on mirror-ground oral materials. With regard to osseointegration, Aminian et al.⁷ have shown that modification of alumina and zirconia surfaces with alkaline phosphatase made the surfaces more negatively charged and more hydrophilic, which resulted in greater cell adhesion, more pronounced cell spreading and a higher number of focal contact points. Electrical interactions, or more specifically small surface currents, are also likely to be behind bactericidal effect and improved osseointegration presented by the surface of the biomaterial Ti6Al4V after irradiation with ultraviolet light. This electronic activity inside the UV-excited TiO₂ passivation layer on the Ti6Al4V has not yet been quantified and therefore cannot be specifically attributed to the bactericidal effects of the titanium alloy after excitation.⁸

In most applications, electrical characterization of surfaces is performed on samples in colloidal form, using electrophoresis to obtain the electrical interaction potential or zeta potential of the samples. However, electrical charge is also of importance in other fields where it works on a large surface area. This is the case, for example, of membranes used in ultrafiltration processes or biomaterials for the manufacture of medical devices. A recent publication has reported in this respect that zeta potential measurements are widely used for the characterization of colloidal suspensions, but have scarcely received attention in the study of large solid surfaces.⁹ It is not only the composition of the bulk which influences the electrical properties of the surface, but also the geometry of the material. Small spherical-like particles and extended surfaces made from the same material can have different electrical surface properties,¹⁰ and for this reason, electrical characterization of a material must be carried out with samples whose geometry is as close as possible to the real situation of its final use.

Bearing this in mind, it would appear from the literature that there are two aspects which need to be improved for characterized surfaces to better reflect the system they represent:

• (i)  
  The electrical information of large surfaces, through zeta potential, is often assumed to be similar to that obtained from the same materials but in colloidal format. Schulz et al.,¹¹ for instance, assumed that the electrical information of flat films (13 mm) was the same as that of 50 μm particles made from the same material when studying the influence of the surface charge of large poly(amidoamine)-alginate hydrogels in adhesion and proliferation of human mesenchymal stem cells. Dwivedi et al.¹² also performed an interesting study in which they first electrically characterized through zeta potential vancomycin-loaded niosomes and then coated the surface of large medical devices (bone plates) with these nanoparticles to control bacterial colonization and biofilm formation, finding that the electrical information obtained for the niosomes was close to that of the large coated surfaces.
• (ii)  
  The extended-large surfaces characterized in the literature are usually dielectric surfaces: polymers, glasses, ceramics, etc.^13–15 with minimal information on conductive surfaces. In particular, it is difficult to find information on the surfaces widely used in the manufacture of orthopaedic devices such as titanium or its alloys. Electrical characterization of large metallic surfaces through zeta potential is not well understood, probably because of the particularities of streaming potential and streaming current techniques.¹⁶ As reported in the work of Roessler et al.,¹⁷ electronic conduction of metal had an influence on the overall value of the measured streaming potential while operating with an electrokinetic analyser, which resulted in different zeta potential values from those obtained from streaming current. Underestimation of the zeta potential from streaming potential was then directly related to the conductivity of the metal in another study carried out by our group,¹⁶ leading us to recommend determination of the zeta potential through streaming current, rather than streaming potential measurements.

Continuing the above research, we explored how electrical information about the bulk metal could be obtained from the underestimated value of the streaming potential, and within this framework, the objectives of the present paper were established.

The first objective was to quantify, for the first time, the electrical contribution of the bulk metal to the measured value of the streaming potential. We then proposed a new model of electrical analysis which would permit the determination of magnitudes such as surface charge density, displaced charge, volumetric density of displaced charge and current density inside the metallic samples. The model has been validated using Ti6Al4V and aluminum and can be used to detect, for the first time, the huge electrical change that the TiO₂ passivation layer of the alloy undergoes after excitation with UV light.

Three dielectrics were employed: Poly(methyl methacrylate) (PMMA), borosilicate glass and polydimethylsiloxane (PDMS). Two disks of 25 mm diameter were used in each experiment to configure the electrokinetic channel.

We used the dielectrics to obtain the geometrical factor of the cell. Briefly, the geometrical factor is defined as the ratio between the channel length (L) and the cross section of the electrokinetic channel (Q): L/Q. It provides information on the compression of the channel and, since the electrokinetic cell is adjustable, it should be maintained constant to achieve a better comparison between the results from the different samples. We monitored channel compression level by plotting pump voltage versus pressure (p) between both ends of the electrokinetic channel. Equal plots guaranteed the same compression inside the electrokinetic channel, and therefore the same flow path. In the electrokinetic analyser software (EKA, Anton Paar), this process is called "flow check". Calculations made with dielectrics were transferred to the conductor samples under the same flow check conditions.

To obtain the value of L/Q, the electrokinetic cell resistance (R_cell) was obtained through the ratio between V_str (streaming potential) and I_str (streaming current),¹⁶ which accounts for the conductivity of the flow path (K_cell).

The value of K_cell was taken as that of the electrolyte (liquid conductivity). Theoretically, an electrolyte with high ionic strength should be used to obtain a liquid conductivity higher than the surface conductivity of the sample (in the case of dielectrics). This is guaranteed in 10⁻¹ M KCl but the limit for neglecting surface conductivity versus liquid conductivity can be established at 10⁻³ M.¹⁶ We made estimations of L/Q by using both ionic strengths: 10⁻¹ M and 10⁻³ M. The values of L/Q were similar but in the case of 10⁻¹ M the dispersions of streaming potential (V_str) and streaming current (I_str) were very high, and so it was necessary to analyze the plots of V_str-p and I_str-p in sections, by considering only the slope of the section where ΔV_str or ΔI_str had a clear linear relationship with Δp.

With this information, the average L/Q value used in this work is 5369 m⁻¹. The liquid conductivity values, K_cell, ranged between 17–19 mS/m.

We performed the electrical analysis with three different metallic materials: Ti6Al4V (disks of 25 mm diameter, 2 mm thickness), aluminum (disks of 25 mm diameter and 1.36 mm thickness) and UV-irradiated Ti6Al4V (UV-Ti6Al4V) (disks of 25 mm diameter, 2 mm thickness). In the case of Ti6Al4V, the spontaneous passivation layer on its surface, mainly composed of amorphous TiO₂,¹⁸ was taken into account. This layer, with about 5–10 nm thickness, has semiconductor properties and can be activated under irradiation with UV-C light. In the present work, we carried out activation using a Philips UV lamp of 257 nm wavelength for 15 h, as previously described.⁸

To clarify the presentation of results, we analyzed the metallic samples with two comparison groups:

Ti6Al4V versus aluminum (Group 1)

Ti6Al4V versus UV-Ti6Al4V (Group 2)

Ti6Al4V was used as reference in both comparisons. The first group (Ti6Al4V versus aluminum) was used to validate the electrical model. The second group was used to evidence the electrical changes occurring on the Ti6Al4V surface after irradiation with UV light.

As previously stated, we measured the streaming current to obtain the real zeta potential of the metallic samples from the following expression:¹⁶

ɛ_r, ɛ₀ and η being relative permittivity of the electrolyte solution, vacuum electric permittivity and solution viscosity. Since the compression of the electrokinetic channel was equal to that of dielectrics (same flow check), L/Q value was taken as the value for dielectrics.

dI_str/dp was the slope of the streaming current (I_str) vs. pressure (p) plot. Pressure was applied in a differential form (ramp form) from 0 to 600 mbar.

The value of ζ from Equation 2 was used to obtain the real value of the streaming potential (V_str-real) and this was compared to the value measured with the EKA system (V_str).¹⁶ Thus,

In order to obtain absolute values of the measured streaming potential (V_str) and to relate them with the values of V_str-real, curves of V_str vs. p were displaced to the zero origin after being fitted to a straight line (R² always close to 1).

The difference between V_str-real and V_str was the potential difference between the extremes of the metallic disk as a consequence of the electronic displacement inside the metal disks during the streaming potential measurement process in the EKA (V_metal). That is to say:

V_metal has not been previously quantified and its value can be related to the electric field generated by the electronic density accumulated at both extremes of the metallic disks. This information was the starting point for the electric field model that we propose.

Electric field model

Fig. 1a shows a schematic representation of the charge distribution proposal inside the metallic samples. EKA operation under the streaming potential option (open circuit) generated electronic displacements inside the two metal disks which made up the electrokinetic channel. These electron movements provoked the accumulation of charge at both ends of each disk and, consequently, a new potential difference appeared, V_metal. The measured potential difference between the electrodes (V_str) was affected by this electronic distribution and so the real streaming potential (V_str-real) was expressed as the sum of V_str and V_metal (Equation 4).

Zoom In Zoom Out Reset image size

As can be seen, both ends of the disks closest to the electrodes become charged with the opposite charge due to the electron movement inside the metal, that is, each metal disk becomes polarized and a surface charge distribution appears at each end of the disk.

In the approximation used in this study, we assume that each surface charge distribution has the height of the disk thickness, represented by "d", and a small width, represented by "L", as in the scheme in Fig. 1b.

If L is small, the curvature is also negligible and can be approached to a thin rectangular strip with dimensions d^*L. This strip can also be considered to consist of many segments of L lengths.

The electric field caused by any charged segment along the bisector line (Fig. 1c) is given by:

λ being the linear charge density along the segment, x the minimal distance from the segment to the point where the electric field is calculated, and θ the angle represented in Fig. 1c.

This electric field is directly related, through the gradient function, with the potential difference between two points.

By relating x with θ:

Thus setting the reference potential far from the segment where θ ≈ 0°, then

In the case of a strip of length L and width d, the strip can be assumed to be composed of many segments of the same length distributed in a continuous form. If d is small in comparison to the distance x, where the electric field is calculated, the electric field can be said to be the sum of the electric fields due to the infinite segments inside the strip, so:

σ being the surface charge density of the strip.

Consequently, the potential can be expressed by:

The potential difference between the two ends of the disk is the difference between the potential at the positive end (V₊) and the potential at the negative end (V₋) and both potentials are obtained through the Equation 10. Consequently, V_metal is equal to this potential difference,

Then if V_metal is known from Equation 4, it would be possible to obtain the surface charge density (σ) accumulated at both ends and, consequently, the charge (electrons) displaced through the metal (Q), the volumetric displaced electrons density (e_vol) and the current density inside the conductor (J).

t being the time and S the cross section of the conductor.

Although exact quantification of the above magnitudes: σ, Q, e_vol, J, is impossible because they are based on the electrical model and exact quantification of parameters such as L, comparisons between samples provide valuable information. Deviations from this assumption would introduce certain variations to the specific value of the electric field in any conductor, although comparisons between samples would not be affected.

In particular, θ (directly related with L) in the calculations of V₊ and V₋ was taken as 5°, which gave an L value of about 4 mm (x = 25 mm, diameter of the disks). Calculations with higher angles such as 10°, or lower such as 2°, did not produce significant changes in the relationships obtained (data not shown).

Table I shows the absolute values of I_str and V_str from EKA of Group 1 at 300 mbar and 500 mbar. The calculated values of V_str-real and V_metal from Equations 3 and 4 are also given.

All obtained values were negative, thus indicating the negative charge of the surface inside the electrolyte. Since the sign of the zeta potential is not relevant in the future analysis, because it is based on the changes in the magnitudes, absolute values are easier to work with, and for this reason all the data in the tables are positive.

Results demonstrate the higher values of V_str-real versus V_str regardless of the conductor or the pressure. It is of interest to note that the potential difference generated at both ends of the conductors due to electronic conduction was always higher than the measured streaming potential ΔV_metal > V_str, which reveals the important contribution of electron movements inside the metals during the streaming potential measurements.

As expected, streaming currents and streaming potentials increased with pressure, and consequently so did V_str-real and V_metal. This means that electronic currents inside the metals were higher as the pressure increased.

Regardless of the pressure selected, currents and potentials were higher for aluminum than for Ti6Al4V. In particular, the potential difference between the two ends of the metallic disks (V_metal) was higher for aluminum than for Ti6Al4V and the changes between both metals appeared to be greater at 300 mbar (8 mV) than at 500 mbar (6 mV). In other words, V_metal for Al was 1.6 and 1.3 (±0.2) times higher than for Ti6Al4V at 300 mbar and 500 mbar, respectively. This result is in agreement with the higher conductivity of aluminum versus Ti6Al4V, K_Al = 3.78·10⁷ S/m; K_Ti6Al4V = 6·10⁵ S/m,¹⁹ implying greater capability of aluminum, compared to Ti6Al4V, to internal electronic displacements.

Table II summarizes the magnitudes obtained with the electrical model from the value of V_metal. Surface charge density at both ends of the disks (σ), displaced charge (Q) and electron density, associated to surface charge density, over the conductor volume (e_vol) also increased with pressure and were higher for Al than for Ti6Al4V, regardless of the pressure. The differences between Al and Ti6Al4V were similar at both pressures: Δσ≈3 C/m²; ΔQ≈1 C; Δe_vol≈1.7 e⁻/m³. The final column of Table II compares the current density between both metals, indicating that electrical activity for Al was double that for Ti6Al4V.

Table III presents the absolute values of I_str and V_str from EKA and the calculated V_str-real and V_metal at 300 mbar and 500 mbar for Ti6Al4V and UV-Ti6Al4V.

Similar to what is seen in Table I, the magnitudes in UV-Ti6Al4V increased with pressure, V_str-real being always higher than V_str measured from EKA and V_metal also higher than V_str.

On comparing aluminum, Ti6Al4V and UV-Ti6Al4V, the highest absolute V_str-real was for UV-Ti6Al4V, with a value of 69 mV at 500 mbar. This was also associated to the highest V_metal for this sample, regardless of the pressure.

The differences between Ti6Al4V and UV-Ti6Al4V must be associated to the electrical changes occurring on the TiO₂ layer after irradiation. The difference between V_metal(Ti6Al4V) and V_metal(UV-Ti6Al4V) informs about the contribution of the electronic activity inside the excited TiO₂ layer after irradiation with ultraviolet light since the irradiation only affects the passivation layer of TiO₂. The potential difference between both ends of the excited TiO₂ layer is expressed in the final column of Table III and is denoted as V_excited-TiO2.

The electrical model developed in Materials and Methods to obtain information on the electronic displacements inside the metal from the data of V_metal can be also applied to the activated TiO₂ layer, taking as input value that of V_excited-TiO2. In this case, the charge distribution at both ends of the TiO₂ layer, closest to the electrodes, can be placed close to a strip with a thickness of about 5 nm. Results are collected in Table IV.

The amount of charge (Q) displaced between both ends of the excited TiO₂ layer was similar to that for Ti6Al4V. This charge moves through a much more reduced area than the whole disk of Ti6Al4V and, for this reason, the surface charge densities at both ends of the layer must be much higher than those of the Ti6Al4V disk and so, consequently, must the electronic displacements. As shown in the first column of Table IV, σ(excited-TiO₂) was five orders of magnitude higher than σ(Ti6Al4V), regardless of the pressure selected for analyzing the results. These large differences were also evident in the volumetric electronic density e_vol, responsible for the appearance of the surface charge at both ends of the layer and, similarly, in the ratio between the current density inside the irradiated TiO₂ layer as compared to that inside the Ti6Al4V disk (J_excited-TiO2/J_Ti6Al4V).

The results of Table IV show, for the first time, the huge electronic activity inside a thin layer of TiO₂ after irradiation with UV-C. Conductivity of the TiO₂ layer before irradiation was that of a dielectric, in the range between 10⁻¹¹ and 10⁻¹⁶ S/m (TiO₂ material information from Goodfellow Catalogue). However, after irradiation, conductivity of the TiO₂ layer changed drastically to that of conductor values. As far as we know, these conductivity values of the excited semiconductor have not been reported in the literature, but we can estimate their values by taking into account the relationship between the current density and the electric field:

at 300 mbar:

and at 500 mbar:

Despite the fact that the above magnitudes are affected by experimental uncertainties (Tables III and IV), it is of interest to highlight the practically identical values obtained in the equations even with input values that depended on pressure. This is what was expected, since the ratio between the conductivities should be intrinsic to the materials and not dependent on surface charge distribution due to the electric field model or on the V_metal obtained from a specific pressure. This is an indication, therefore, of the reliability of the model developed in this research and of how the data should be analyzed. Consequently,

Conductivity of the passivated layer of TiO₂ on Ti6Al4V, after irradiation, became five times higher than that of the Ti6Al4V bulk. This result reinforces the idea that the semiconductor was still activated after the ultraviolet lamp was disconnected, and that this electronic activity must be behind the bactericidal effects described for the irradiated Ti6Al4V biomaterial.²⁰ The electrical changes are also in agreement with the previous electrochemical impedance spectroscopic (EIS) results where EIS and Mott-Schottky plots showed that UV treatment caused an initial increase on the surface electrical conduction of this material. In addition, EIS and polarization plots indicated higher corrosion currents at the UV treated surfaces, compared to the non-irradiated samples.²¹ In line with these results, Deng et al.²² have reported recently that UV-illumination stimulated the nano-TiO₂ semiconductor with elevated conductivity in favor of passing corrosion current.

The study of the properties of TiO₂ is an issue of great interest in the research community. It is also true that most research focuses on the study of TiO₂ nanoparticles with direct application to the manufacture of batteries, solar cells, biosensors, water treatments, cosmetics, etc.^23–25 Other applications include neurosciences and the textile industry.^26,27

Excitation of TiO₂ is key in many of these applications and therefore we consider that it is of great importance to obtain maximum information from the excited state of TiO₂. It has been confirmed both by previous works and by this research that excitation of TiO₂ is not something instantaneous that disappears when the irradiation ceases but that it is maintained in time and the electrical changes associated with this fact are measurable. For the first time, electrical properties are quantified and the conductivity value of this excited state of TiO₂ is provided.