Photoelectrochemical evidence of nitrogen incorporation during anodizing sputtering – deposited Al–Ta alloys

Abstract

Anodic films were grown to 20 V on sputtering-deposited Al–Ta alloys in ammonium biborate and borate buffer solutions. According to glow discharge optical emission spectroscopy, anodizing in ammonium containing solution leads to the formation of N containing anodic layers. Impedance measurements did not evidence significant differences between the dielectric properties of the anodic films as a function of the anodizing electrolyte. Photoelectrochemical investigation allowed evidencing that N incorporation induces a red-shift in the light absorption threshold of the films due to the formation of allowed localized states inside their mobility gap. The estimated Fowler threshold for the internal photoemission processes of electrons resulted to be independent of the anodizing electrolyte confirming that N incorporation does not appreciably affect the density of states distribution close to the conduction band mobility edge. The transport of photogenerated carriers has been rationalized according to the Pai–Enck model of geminate recombination.

---

Introduction

The anodizing of aluminium and tantalum has been extensively studied due to the widespread use of the corresponding oxides as dielectric materials in electrolytic capacitors.^1,2 In the last decade, it has been proved that anodizing sputtering-deposited aluminium–tantalum alloys is a viable tool to grow Al–Ta mixed oxides, whose composition is a function of the Al/Ta ratio in the base alloy. In ref. 3–5 the structure, composition and physico-chemical properties of such mixed oxides have been widely investigated. Transmission electron micrographs of ultramicrotomed sections revealed that anodic oxides on Al–Ta alloys are amorphous and uniform in thickness, with flat metal/oxide and oxide/electrolyte interfaces. They show nanometric roughness as also confirmed by atomic force microscopy.⁵ The compositional depth profiles provided by Rutherford back scattering revealed the formation of oxides with uniform composition.^3,4 The oxide growth occurred due to simultaneous migrations of cations (Al³⁺ and Ta⁵⁺) outwards and anions (O²⁻) inwards proceeding in a cooperative manner, developing film materials both at the film/electrolyte and metal/film interfaces, respectively. However, the film structure and the composition are also dependent upon the electrolyte due to the possible incorporation of foreign species from the anodizing bath.^6–8 This is very important if we consider that the presence of foreign species in anodic films could affect the electronic properties of the films with possible detrimental effects on the dielectric properties of the oxides.

In ref. 9–11 it was shown that during anodizing in ammonium containing solutions of valve metals such as Nb, Ta and Ti, N incorporation occurs with significant effects on the solid state properties of the corresponding oxides. More interestingly, in the case of tantalum, such incorporation is reported to occur only when the electrolyte pH was higher than the pH of zero charge (pH_pzc) of Ta₂O₅, since the negatively charged oxide surface allowed ammonium cations reaching the film and being de-protonated by the imposed electric field.

In this work we want to study if N incorporation occurs during anodizing Al–Ta alloys of different compositions in an ammonium biborate solution. Glow discharge optical emission spectroscopy is used to directly assess the presence of N in the anodic films, while the possible effects on the solid state properties are studied by impedance measurements and photoelectrochemical experiments.

Experimental section

Metallic substrates and oxides

Aluminium, tantalum and Al–Ta alloys (10, 18, 20, 30, 42, 62, 81, and 91at% Ta) were deposited by dc magnetron sputtering on glass substrates. The deposition was carried out in a Shimadzu, SP-2C system, at a rate of 0.11 nm s⁻¹, by using an Al target 99.99% and a Ta target 99.9%; the chamber was initially evacuated to 5 × 10⁻⁷ mbar with subsequent sputtering using Ar at 3 × 10⁻³ mbar.

The anodizing was undertaken in a borate buffer (0.42 M H₃BO₃ and 0.08 M Na₂B₄O₇) (pH = 8) and 0.1 M ammonium biborate tetrahydrate (ABE, (NH₄)₂B₄O₇·4H₂O) (pH ∼9) at room temperature potentiodynamically at 10 mV s⁻¹. Alloys were anodized to 20 V vs. mercury/mercury oxide electrode (0 V vs. Hg/HgO = 0.098 V vs. SHE). The thicknesses of the anodic layers, estimated according to the anodizing ratio reported in previous studies,^3,4 are reported in Table 1.

Table 1 Anodic films thicknesses

Base alloy	Thickness/nm
Al	24.0
Al–10 at% Ta	24.4
Al–18 at% Ta	25.1
Al–20 at% Ta	25.3
Al–30 at% Ta	26.2
Al–42 at% Ta	27.2
Al–62 at% Ta	28.9
Al–81 at% Ta	30.5
Al–91 at% Ta	31.4
Ta	32.4

---

X-Ray reflectivity measurements

X-Ray reflectivity (XRR) spectra were collected using a Rigaku SmartLab X-ray diffractometer with a copper anode (Cu Kα radiation, λ = 0.15405 nm, 30 kV, 20 mA). XRR spectra were recorded over the 2θ angle range from 0 to 4 with a step size of 0.0004°. Rigaku GlobalFit was used to fit the spectra.

Glow discharge optical emission spectroscopy

Elemental depth profile analysis was conducted by glow discharge optical emission spectroscopy (GDOES) using a Jobin Yvon 5000 instrument in an argon atmosphere of 600 Pa by applying a RF of 13.56 MHz and power of 30 W with a pulse mode of 100 Hz and a duty cycle of 0.1 s. The wavelengths of the spectral lines used were 396.157, 302.017, 149.262, 249.678 and 130.217 nm for aluminium, tantalum, nitrogen, boron and oxygen respectively.

Impedance measurement setup

Impedance measurements were carried out in 0.25 M Na₂HPO₄ (pH ∼9) using a Parstat 2263 (PAR), connected to a computer for the data acquisition. A Pt net with a very high specific surface area was used as a counter-electrode, while mercury/mercury oxide was used as a reference electrode. The impedance spectra were generated by superimposing onto the continuous potential a sinusoidal signal of amplitude 10 mV over the frequency range 100 kHz–100 mHz, and the results were fitted using ZSimp software.

Photoelectrochemical measurement setup

The experimental set-up for photoelectrochemical investigations has been described elsewhere.¹² A 450 W UV-Vis xenon lamp, coupled with a monochromator, allows irradiation of the specimen through a quartz window. A two-phase lock-in amplifier, coupled with a mechanical chopper (with a frequency of 13 Hz) enables the separation of the photocurrent from the total current circulating in the cell. Photocurrent spectra reported are corrected for the relative photon flux of the light source at each wavelength, so that the photocurrent yield is represented in the y axis in arbitrary current units. All the experiments were performed in air at room temperature in 0.1 M ABE.

Results

Anodic film growth

Anodic films were grown to 20 V at 10 mV s⁻¹ on all investigated alloys. In Fig. 1a we report the growth curve in relation to anodizing the Al–91 at% Ta alloy. As typical for valve metals, after an abrupt increase, the current density reaches an almost constant value, which is a function of the metallic substrate composition. The current density is sustained by the oxidation of metals, according to the following half cell reaction:

Me → Me^z+ + ze⁻ (1)

where Me is Al and/or Ta depending on the alloy composition. According to Faraday's law in the case of high field growth of anodic films, it is possible to correlate the electric field strength, E, to the growth rate, dV/dt, according to the equation:in which i is the measured current density, M the molecular weight of the growing oxide, z the number of electrons circulating per mole of the formed oxide, F the Faraday constant, ρ the film density and η the growth efficiency. η is defined as:where i_form is the current density effectively employed for the film formation, i_diss is the current density fraction due to dissolution phenomena, expected to be negligible for all the investigated samples according to the Pourbaix diagrams in relation to Al and Ta,¹³ and i_el is the electronic current which is negligible due to the blocking character of the oxides, as confirmed by the very low current circulating during the reverse scan (see Fig. 1a).

Fig. 1 (a) Current density vs. potential curves recorded during the potentiodynamic growth of anodic films on the Al–91 at% Ta alloy at 10 mV s⁻¹ in 0.1 M ABE. (b) X-Ray reflectivity spectrum in relation to the 20 V anodic film grown on Al–91 at% Ta. Fitting parameters: density = 8.5 g cm⁻³, thickness = 37 nm, roughness = 0.33 nm.

In Fig. 1b we report the XRR spectrum in relation to the anodic film of Fig. 1a. The resulting angle-dependent curve, which is correlated to the changes in the electron density in the sample, was fitted with a triple layer model (SiO₂/alloy/oxide). The fitting procedure allowed estimating a film thickness very close to that estimated from the anodizing ratio (see Table 1), and a density which is very close to that reported for pure anodic Ta₂O₅³ with roughness in the order of few Å. The roughness factor resulted between 0.3 and 0.5 nm for the investigated samples.

The oxide thickness increases due to the chemical reaction between Ta⁵⁺, Al³⁺, and O²⁻. Mobile cationic species are produced at the metal/oxide interface and migrate outward toward the oxide/electrolyte interface. Mobile anionic species are produced at the oxide/electrolyte interface and migrate inward. In aqueous solutions, where water molecules might be expected to adhere to the film with negatively charged oxygen atoms adjacent to the oxide surface, mobile anionic species are mainly O²⁻ and OH⁻ ions derived from water de-protonation.^14,15 The OH⁻ ions, once incorporated, are converted to O²⁻ ions and protons, with H⁺ migrating out of the film under the high electric field. If large oxyanions are present in the anodizing bath (e.g. biborate), hydrogen bonds can be formed between H of water molecules and O of such oxyanions (see Scheme 1) with a consequent weakening of O–H covalent bonds in H₂O and, thus, easier de-protonation. However, it has been proved that other anionic species, A^y−, of the anodizing bath can also be incorporated during the anodizing process, leading to the formation of oxides containing foreign species.¹⁶ Their distribution along the film thickness is a function of the relative migration rate of A^y− with respect to O²⁻/OH⁻ ions. More specifically, incorporation of several inorganic¹⁷ and organic^18,19 anions has been reported during anodizing Al, while incorporation of several ions has been proved during anodizing Ta.²⁰ Recently, it has also been demonstrated that N incorporation occurs during anodizing of tantalum in ammonium biborate solutions, provided that the electrolyte pH is higher than the pH of zero charge (pH_pzc) of the oxide. In the latter case the oxide surface is negatively charged, thus allowing NH₄⁺ ions to approach the oxide/electrolyte interface and be de-protonated generating NH_3−x^x− moieties that are incorporated into the anodic film with the consequent formation of an oxynitride.¹⁰ In Fig. 2 we report GDOES elemental depth profiles in relation to anodic films grown to 20 V in 0.1 M ABE on Al–10 at% Ta, Al–42 at% Ta and Al–81 at% Ta alloys. The N signal is weak but present across the whole film thickness, thus confirming that nitrogen incorporation also occurs during anodizing Al–Ta alloys. In ref. 21 and 22 pH_pzc values of ∼9.5 and 2.9 are reported for alumina and tantala, respectively. Therefore, we expect that the higher is the Al content in the oxide, the lower is the extent of the incorporation phenomena (see Fig. 2).

Scheme 1

Fig. 2 GDOES depth profiles of the anodic films grown to 20 V on (a) Al–10 at% Ta, (b) Al–42 at% Ta and (c) Al–81 at% Ta alloys in 0.1 M ABE.

The formation of NH_3−x^x− moieties is assisted by weakening the N–H bond as depicted in Scheme 1 since, due to the large electronegativity of nitrogen, the formation of hydrogen bonds between H of ammonia and O of oxyanions (i.e. –NH⋯O) can also occur. As it occurs for OH⁻, NH_3−x^x− can lose protons and finally be incorporated into the anodic films as N³⁻ ions. N–H dissociation energy in ammonia (388 kJ mol⁻¹) is lower than O–H dissociation energy in water (463 kJ mol⁻¹). Moreover, ionic radii of NH₂⁻ and N³⁻ (0.173 nm and 0.171, respectively, according to ref. 23 and 25) are close to those of O²⁻ and OH⁻ (0.140 nm and 0.153 nm, respectively), and even lower than those of oxyanions, whose incorporation has been proved in previous studies (e.g. 0.238 nm for PO₄³⁻ and 0.230 nm for SO₄²⁻, see ref. 20). Thus, it is likely that an oxynitride can be formed during the anodizing process without significant structural problems, considering that the films are amorphous and that according to the literature the only effect of nitridation of crystalline tantalum oxide is an expansion of the lattice, due to the smaller ionic radius of O²⁻ (0.140 nm) with respect to N³⁻ (0.171 nm).²⁴ Such an expansion is also reported to occur for N-doped Ta₂O₅ and in spite of this, a crystalline stable phase is formed.

Impedance measurements

In order to obtain information on the dielectric properties of the investigated oxides, we recorded electrochemical impedance spectra and differential capacitance curves. In Fig. 3a–c we report the EIS spectra, recorded in 0.25 M Na₂HPO₄ at U_E = 5 V (Hg/HgO) (i.e. at potentials more anodic than the corresponding flat band potential, see below), in a Modified Bode representation in relation to the anodic films grown on Al–42 at% Ta, Al–62 at% Ta and Al–91 at% Ta alloys, respectively, in 0.1 M ABE and borate buffer to 20 V. As shown in Fig. 3, no significant differences are found in EIS spectra due to the anodizing electrolyte. The high |Z| and a phase angle very close to 90° confirm the formation of blocking oxides. Spectra can be simulated by the electrical circuit reported in Fig. 3d, where R_el is the electrolyte resistance (employed to get the Modified Bode representation), R_ox is the anodic film resistance and Q_ox is a constant phase element (CPE) introduced to model the oxide capacitance.

Fig. 3 EIS spectra in relation to anodic films grown on (a) Al–42 at% Ta, (b) Al–62 at% Ta and (c) Al–91 at% Ta alloys in ABE and borate buffer, recorded by polarizing the film at 5 V vs. Hg/HgO. (d) Electrical equivalent circuit employed to model metal/oxide/electrolyte interfaces.

Another RQ parallel was inserted to model the electrochemical reaction, where R_ct is the charge-transfer resistance and Q_H the non-ideal Helmholtz double layer capacitance. This equivalent circuit very well fits the experimental data in relation to the all anodic films grown in 0.1 M ABE, with the fitting parameters reported in Table 2. The very high R_ox and α values very close to 1 suggest that the oxides behave like ideal capacitors. It is interesting to mention that very high R_ct values were estimated, as expected for blocking the interface, and Q_H allowed estimating the double layer capacitance very close to 20 μF cm⁻², the value reported in the literature for aqueous solutions.²⁶

Table 2 Fitting parameters in relation to EIS spectra of all the investigated anodic films using the equivalent electric circuit in Fig. 3d

Base alloy	R el/Ω cm^2	R ct/Ω cm^2	Q H/S s^α cm^−2	α	R ox /Ω cm^2	Q ox/S s^α cm^−2	α
Al	14	1 × 10^4	2.1 × 10^−5	0.82	1 × 10^5	4.3 × 10^−7	1
Al–10 at% Ta	15	1 × 10^6	2.3 × 10^−5	0.85	1 × 10^8	4.4 × 10^−7	0.99
Al–18 at% Ta	14	1 × 10^6	2.1 × 10^−5	0.80	1 × 10^8	5.4 × 10^−7	1
Al–20 at% Ta	39	1 × 10^6	2.4 × 10^−5	0.87	1 × 10^9	4.2 × 10^−7	0.99
Al–30 at% Ta	26	1 × 10^6	2.5 × 10^−5	0.88	9 × 10^9	4.4 × 10^−7	0.99
Al–42 at% Ta	20	1 × 10^6	2.2 × 10^−5	0.88	1 × 10^8	6.2 × 10^−7	0.99
Al–62 at% Ta	24	1 × 10^6	2.2 × 10^−5	0.85	1 × 10^8	5.4 × 10^−7	0.99
Al–81 at% Ta	23	1 × 10^6	2.0 × 10^−5	0.82	1 × 10^8	7.8 × 10^−7	1
Al–91 at% Ta	27	4 × 10^6	2.2 × 10^−5	0.95	1 × 10^9	6.9 × 10^−7	0.99
Ta	39	1 × 10^6	2.2 × 10^−5	0.88	1 × 10^8	8.5 × 10^−7	1

---

In Fig. 4 we report the differential capacitance measurements, recorded at f = 1 kHz, for all the investigated films. It is evident that the capacitance is almost potential independent of and slightly dependent on the frequency, as expected for amorphous insulating materials.²⁶ A comparison with the capacitance of 20 V anodic films on Al–Ta alloys grown in borate buffer solution shows that no appreciable differences are evident on the overall impedance (Fig. 5a–c). This suggests that the presence of N in the oxide causes the formation of deep localized states, thus at the energy level so far from the conduction band mobility edge they cannot contribute to the measured capacitance in the range of investigated frequencies. This is in agreement with previous experimental findings on the effect of nitrogen incorporation during anodizing Ta and Ti,^10,11 which is reported to induce the formation of localized states close to the valence band mobility edge.

Fig. 4 Measured series capacitance in relation to all oxides grown up to 20 V vs. Hg/HgO in ABE. A.c. signal frequency: 1 kHz.

Fig. 5 Measured series capacitance in relation to the anodic films grown on (a) Al–42 at% Ta, (b) Al–62 at% Ta and (c) Al–91 at% Ta alloys in ABE and borate buffer. A.c. signal frequency: 1 kHz.

Provided that the equivalent circuit of Fig. 3d simulates well the impedance of the overall metal/oxide/electrolyte interface, it is easy to extract the oxide capacitance from C_M. Assuming that C_ox can be described by a parallel plate capacitor model:

where ε₀ (8.85 × 10⁻¹⁴ F cm⁻¹) is the vacuum permittivity, ε_r the film relative permittivity and d_ox its thickness. Whereas there are no differences in the anodic film growth with respect to anodizing solutions, oxide thicknesses reported in Table 1 can be considered in calculating ε_r for the oxides grown in 0.1 M ABE. Dielectric constants are reported in Table 3 and the values are in very good agreement with those reported for the oxides grown in borate buffer.

Table 3 Dielectric constants of anodic films estimated from differential capacitance measurements

Base alloy	Dielectric constant
Al	9
Al–10 at% Ta	10
Al–18 at% Ta	14
Al–20 at% Ta	12
Al–30 at% Ta	13
Al–42 at% Ta	14
Al–62 at% Ta	16
Al–81 at% Ta	24
Al–91 at% Ta	23
Ta	29

---

Photoelectrochemical measurements

In order to study the influence of N incorporation on the solid state properties of the oxides, photoelectrochemical measurements were carried out. Anodic photocurrent spectra (normalized for the maximum value of I_ph), in relation to the anodic film grown to 20 V on the Al–62 at% Ta alloy in both solutions recorded by polarizing the electrodes at 5 V (Hg/HgO), are reported in Fig. 6a. An evident red-shift in the optical absorption threshold is present for the anodic film grown in 0.1 M ABE, while no photocurrent is detected for λ ≥ 310 nm in the spectrum in relation to the oxide grown in borate buffer. From photocurrent spectra, it is possible to estimate the optical band gap value of the investigated anodic films according to the following equation:

(I_phhν)ⁿ ∝ (hν − E_g) (5)

which is valid for photon energy in the vicinity of the band gap. I_ph is the photocurrent yield, assumed to be proportional to the light absorption coefficient, hν is the photon energy, E_g is the optical band gap and the n value equal to 0.5 has been assumed for non-direct optical transitions.¹²

Fig. 6 Anodic photocurrent spectra (a) in relation to anodic films grown on the Al–62 at% Ta alloy in ABE and borate buffer, recorded by polarizing the electrodes at 5 V vs. Hg/HgO. The band gap estimate by assuming non-direct optical transitions in relation to the same anodic films grown in ABE (b) and borate buffer (c).

It is important to mention that due to the amorphous structure of the investigated oxides, the presence of allowed localized states inside the gap is predicted according to several models of density of states distribution,¹² thus it is more appropriate to define E_g in eqn (5) as the mobility gap of the layer. An E_g of 4.2 eV has been estimated for the anodic film grown in buffer, as shown in Fig. 6c, which is slightly lower than that estimated for thinner oxides (i.e. formation voltage = 10 V).²⁷ For the oxide grown in the ammonium-containing solution (Fig. 6b) by extrapolating the (I_phhν)^0.5vs. hν plot to zero it was possible to estimate an E_g of ∼4.1 eV but a long photocurrent tail is also present ending at 3.4 eV.

The latter is due to optical transitions involving the allowed states in the mobility gap of the oxide, generated by N incorporation.

The same photoelectrochemical behaviour was shown by anodic films grown on Al–Ta alloys with a Ta content ≥42 at%, (see Fig. 7), while for the oxide grown on Al–20 at% Ta there is no evidence of photocurrent at energy lower than the E_g of the oxide. We would like to mention that no anodic photocurrent was measured with anodic films grown on Al–Ta alloys with a lower Ta content.

Fig. 7 Band gap estimation by assuming non-direct optical transitions in relation to anodic films grown to 20 V on (a) Al–91 at% Ta, (b) Al–81 at% Ta, (c) Al–42 at% Ta, and (d) Al–20 at% Ta alloys in ABE.

We have also studied the dependence of the measured photocurrent on the applied potential, i.e. on the electric field strength across the films. In Fig. 8a we show the I_phvs. U_E curves (photocharacteristics) for the anodic films grown on Al–62 at% Ta to 20 V in both ABE and borate buffer, recorded at λ = 270 nm (hν = 4.58 eV). As it will be discussed better in the next section, the shape of I_phvs. U_E curves suggests the occurrence of recombination phenomena for both films grown in ABE and borate buffer. As expected for insulating films, by scanning the electrode potential toward the cathodic direction, a clear inversion of the photocurrent sign was revealed, as confirmed by a sharp change in the photocurrent phase angle (not reported).

Fig. 8 (a) Photocurrent vs. potential curves in relation to 20 V anodic films grown on the Al–62 at% Ta alloy in ABE and borate buffer recorded in 0.1 M ABE at λ = 270 nm. (b) Photocurrent vs. potential curve in relation to the oxide grown on the Al–62 at% Ta alloy in ABE at λ = 330 nm. Potential scan rate: 10 mV s⁻¹.

In the case of the insulating material both anodic and cathodic photocurrents can be measured when the polarizing potential is more anodic and more cathodic than the flat band potential, U_FB, of the oxide, respectively.

In order to estimate U_FB for the investigated layers, current vs. time measurements were carried out under constant irradiating wavelengths and manually chopping the irradiation. Fig. 9 shows current transients in relation to the anodic film grown on the Al–91 at% Ta alloy in ABE at two different polarizing potentials. Under strong anodic polarization, i.e. at 8 V (Hg/HgO) (Fig. 9a), soon after irradiation the current reaches an almost stationary value for all the investigated wavelengths (up to λ = 360 nm).

Fig. 9 Total current circulating under irradiation (on) and in the dark (off) in the oxide grown on the Al–91 at% Ta alloy in ABE, by polarizing the electrode at 8 V (a) and −0.8 V (b) vs. Hg/HgO at different wavelengths.

At U_E = −0.8 V (Hg/HgO) (Fig. 9b), in spite of the presence of anodic photocurrent spikes soon after irradiation, stationary photocurrent is cathodic, thus suggesting U_FB > −0.8 V (Hg/HgO). At this electrode potential, the electric field is not high enough to avoid the photocarrier recombination preventing cathodic photocurrent detection at λ = 360 nm, i.e. at photon energy lower than the mobility gap.

U _FB values estimated from current–time transients for anodic films grown on alloys with the Ta content ≥62 at% are reported in Table 4. For anodic films grown in 0.1 M ABE with a Ta content ≥62 at%, it was possible to record photocharacteristics under illumination with photons corresponding to hν lower than their mobility gap. In Fig. 8b we show I_phvs. U_E at λ = 330 nm (hν = 3.75 eV) for the anodic film grown on Al–62 at% Ta to 20 V in ABE.

Table 4 Flat band potential values estimated by recording current transients under constant irradiating wavelengths in relation to the anodic oxides with a Ta content ≥62at%

Base alloy	U fb/V (vs. Hg/HgO)
Al–62 at% Ta	−0.70
Al–81 at% Ta	−0.75
Al–91 at% Ta	−0.70
Ta	−0.65

---

I _ph decreases during the potential scan toward the cathodic direction and for U_E < U_FB cathodic photocurrent is measured, as evidenced in the magnified area of the plot. Thus, photocurrent spectra were recorded at a potential lower than U_FB (i.e. at −1.5 V vs. Hg/HgO), as shown in Fig. 10a for oxides grown on the Al–62 at% Ta alloy in both ABE and borate buffer solutions. As already found for several insulating oxides^28–33 as well as for thinner films grown in an ammonium free electrolyte on Al–Ta alloys,²⁷ a long photocurrent tail appears that has been explained by electron injection processes from the alloy Fermi level to the oxide conduction band (see Fig. 10d).

Fig. 10 (a) Cathodic photocurrent spectra in relation to the oxides grown on the Al–62 at% Ta alloy in ABE and borate buffer recorded at −1.5 V vs. Hg/HgO as an electrode potential. (b) Long wavelength region of the photocurrent spectra, recorded by using a UV filter. Fowler plots in relation to the oxide grown in buffer (c) and ABE (d). Inset: Internal photoemission of electrons from the metal Fermi level.

The threshold energy, E_th, associated with this process can be estimated according to Fowler's law:²⁸

(I_ph)^0.5 ∝ (hν − E_th) (6)

from the long wavelength region of the photocurrent spectra, corresponding to λ ≥ 400 nm, which has been recorded using a 400 nm cut-off filter to avoid the doubling effect (see Fig. 10b). The same internal photoemission process occurs for the anodic film grown in ABE with a Fowler threshold almost coincident with that estimated for N-free oxide. This experimental finding further supports that N incorporation induces the formation of allowed states close to the valence band mobility edge of the anodic films without significant effects on the density of states close to the conduction band mobility edge.

Discussion

In the framework of Butler–Gartner model, for crystalline insulating films, a linear dependence of the measured photocurrent on the applied potential is expected in the absence of trapping phenomena, which can modify the electric field distribution across the layer.³⁴

However, for both anodic films grown in ABE and buffer, such dependence is not linear thereby supporting the presence of recombination phenomena. For amorphous materials the lack of a long range order induces the formation of allowed localized states close to the mobility edge of the oxide, in agreement with the Mott and Davis model.³⁵ Thus, together with surface and bulk recombination phenomena expected for both crystalline and amorphous materials, we have to account for the occurrence of geminate recombination phenomena which reduce the efficiency of free carrier generation for amorphous anodic films on Al–Ta alloys. Geminate recombination occurs generally in any material where the photogenerated carriers display very low mobility.

The mobility of carriers in the localized states (below the conduction band and above the valence band edges) is much lower than in the extended states, so the probability of initial recombination effects in amorphous materials is quite high.

In fact, during the thermalization time, the electron–hole pairs do not cover a distance long enough to prevent recombination due to their mutual coulombic attraction. Owing to this insufficient separation, a certain fraction of the photogenerated carriers recombine before the transport process can separate them permanently.^36,37

To explain the dependence of photocurrent on both the applied potential and the irradiating wavelength, we recall Onsager's theory,³⁸ originally developed for weak electrolytes, and then applied by Pai and Enck to the initial recombination process (geminate recombination) of electron–hole pairs in amorphous selenium.³⁹ This was achieved by deriving a mathematical expression for the efficiency of photogeneration, η_g, taking into account the tri-dimensional aspect of the generation and the effect of the electric field on the initial separation of the photocarriers:

where E is the electric field, k the Boltzmann constant, e the electronic charge, and A = e²/4πεε₀kTr₀ with ε being the dielectric constant of the oxide and by considering that every absorbed photon creates a pair of thermalized carriers bound by their mutual coulombic attraction. r₀ is the thermalization length, i.e. the distance travelled by the photocarriers before their effective separation by the electric field. In eqn (7) the only parameter that changes with irradiating photon wavelength is r₀, which is a function of the excess energy ΔE = (hν − E_trans), i.e. the extra energy with respect to that necessary for the optical transition. During the thermalization process, this excess energy is dissipated over the local potential by phonon emission.³⁹ According to eqn (7), higher generation efficiencies can be obtained with high electric fields and high thermalization lengths. Since r₀ directly depends on photon energy, high efficiency values are achieved at short irradiation wavelengths, while strong geminate recombination effects are characteristic of photogeneration at high irradiation wavelengths. η_g can be introduced in the general expression of photocurrent on the applied potential, i.e. the applied electric field:where Φ₀ is the incident photon flux on the oxide surface, R is the reflection coefficient, α is the absorption coefficient, d_ox is the anodic film thickness, and L_D is the sum of drift lengths of injected photocarriers, expressed as:

L_D = (μ_hτ_h + μ_eτ_e)E (9)

with μ being the photocarrier mobility and τ the photocarrier lifetime.¹²Eqn (8) has been used to fit the photocharacteristics in relation to anodic films grown on Al–62 at% Ta, Al–81 at% Ta, Al–91 at% Ta alloys and on pure Ta to 20 V in 0.1 M ABE. In Fig. 11 fitting curves are overlapped to the photocharacteristics in relation to the anodic films grown on Al–62 at% Ta, Al–81 at% Ta and Al–91 at% Ta alloys, recorded at two different wavelengths, 240 nm and 330 nm, corresponding to irradiating photon energy higher and lower than the oxide band gap.

Fig. 11 Photocurrent vs. band bending in relation to anodic films grown to 20 V on Al–91 at% Ta, Al–81 at% Ta, and Al–62 at% Ta alloys in ABE, recorded at λ = 240 nm (a) and λ = 330 nm (b). Continuous lines are plotted according to eqn (8) with the fitting parameters of Table 5.

The reflection coefficient has been calculated by using a three-layer model (i.e. solution/oxide/metal) under normal light incidence and by including the effects of multiple reflections,⁴⁰ with optical constants for solution, metal and oxide taken from ref. 41 and 42. During the fitting procedure we used the flat band potentials estimated from current time transients (see Table 4) and oxides' thickness calculated from the anodizing ratio estimated in ref. 3 and 4 by direct inspection of transmission electron micrographs of the ultramicrotomed section (see Table 1).

The best fitting parameters are summarized in Table 5. It is interesting to mention that r₀ is inversely proportional to the wavelength, and that very low μτ values were derived from the best fitting procedure. This suggests that for λ corresponding to both supra- and sub-band gap values, photocarriers are generated in localized states. Due to the amorphous nature of the investigated anodic films, we expect that close to the conduction and valence band mobility edges, lattice disorder and/or the presence of not stoichiometric oxides induce the formation of localized states, which can extend up to 0.2 eV below CB and above VB edges.³⁰ For hν < E_g we can conclude that the allowed localized states are also present inside the mobility gap and μτ even lower than that estimated for the higher irradiating wavelength suggests that they are strongly localized.

Table 5 Fitting parameters in relation to photocharacteristics, recorded at different wavelengths in 0.1 M ABE, for anodic films grown to 20 V on Al–Ta alloys with a Ta content ≥62at%

Base alloy	Wavelength/nm	μτ/cm^2 V^−1	r 0/Å
Al–62 at% Ta	240	6.7 × 10^−15	10
	270	3.7 × 10^−15	7.5
	300	9.2 × 10^−16	7
	330	4.8 × 10^−16	6
Al–81 at% Ta	240	1.3 × 10^−14	10
	270	1.0 × 10^−14	6
	300	3.7 × 10^−15	4.5
	330	1.9 × 10^−15	4
Al–91 at% Ta	240	2.3 × 10^−14	14
	270	2.1 × 10^−14	7
	300	1.0 × 10^−14	5
	330	9.0 × 10^−15	4
Ta	240	2.4 × 10^−14	17.5
	270	1.9 × 10^−14	6
	300	9.0 × 10^−15	4
	330	7.7 × 10^−15	3

---

If we recall that Fowler threshold energies for films grown in borate buffer and films grown in ABE are almost coincident and that N incorporation does not contribute significantly to the measured impedance, we can suggest that anodizing in ABE leads to the formation of anodic films whose density of states can be described according to the sketch of Fig. 12.

Fig. 12 Electronic structure of amorphous semiconducting oxides following the Mott–Davis model with allowed localized states in the mobility gap. LS: localized states.

N incorporation induces the formation of allowed localized states close to the valence band mobility edge of the oxides, and that the energy distance of such states from the conduction band mobility edge is roughly correlated to the red-shift of the light absorption threshold of the anodic spectra.

Conclusions

Anodic films were grown to 20 V on sputtering-deposited Al–Ta alloys of several compositions in 0.1 M ABE and borate buffer solutions. According to GDOES compositional depth profiles, anodizing in ABE leads to the formation of N containing oxide layers with nitrogen distributed across the whole film thickness. The presence of N does not change appreciably the impedance of the oxides, as suggested by the comparison between the impedance spectra and the differential capacitance curves in relation to anodic films grown in ammonium-free and ammonium-containing solutions. In contrast, N incorporation significantly changes the photoelectrochemical behaviour of the anodic films grown on Al–Ta alloys with the Al content ≥20 at%. A red-shift of the light absorption threshold of the investigated layers was evidenced, which is attributed to the optical transition involving allowed states inside the mobility gap of the films. The dependence of photocurrent on photon energy and the electric field strength in the framework of Pai–Enck model for geminate recombination suggested that incorporation of nitrogen induces the formation of allowed localized states. Such states are close to the valence band mobility edge, since they do not contribute to the measured impedance up to a very low frequency of the a.c. signal (0.1 Hz) and since they do not affect the Fowler threshold for the internal electron photoemission phenomena.

Notes and references

1. G. Scaduto, M. Santamaria, P. Bocchetta and F. Di Quarto, Thin Solid Films, 2014, 550, 128.
2. S. Ezhilvalavan and T. Y. Tseng, J. Mater. Sci.: Mater. Electron., 1999, 10, 9.
3. H. Habazaki, K. Shimizu, P. Skeldon, G. E. Thompson and G. C. Wood, Proc. R. Soc. A, 1997, 453, 1593.
4. G. Alcalá, S. Mato, P. Skeldon, G. E. Thompson, P. Bailey, T. C. Q. Noakes, H. Habazaki and K. Shimizu, Corros. Sci., 2003, 45, 1803.
5. G. Alcalá, P. Skeldon, G. E. Thompson, A. B. Mann, H. Habazaki and K. Shimizu, Nanotechnology, 2002, 13, 451.
6. S. J. Garcia-Vergara, I. S. Molchan, F. Zhou, H. Habazaki, D. Kowalski, P. Skeldon and G. E. Thompson, Surf. Interface Anal., 2011, 43, 893.
7. F. Zhou, D. J. Leclere, S. J. Garcia-Vergara, T. Hashimoto, I. S. Molchan, H. Habazaki, P. Skeldon and G. E. Thompson, J. Electrochem. Soc., 2010, 157, C437.
8. H. Habazaki, K. Shimizu, S. Nagata, P. Skeldon, G. E. Thompson and G. C. Wood, Corros. Sci., 2002, 44, 1047.
9. S. Ono, K. Kuramochi and H. Asoh, Corros. Sci., 2009, 51, 1513.
10. F. Di Franco, M. Santamaria, F. Di Quarto, E. Tsuji and H. Habazaki, Electrochim. Acta, 2012, 59, 382.
11. S. Miraghaei, M. Santamaria and F. Di Quarto, Electrochim. Acta, 2014, 134, 150.
12. (a) F. Di Quarto, F. La Mantia and M. Santamaria, in Modern Aspects of Electrochemistry, No. 46: Progress in Corrosion Science and Engineering I, ed. S.-I. Pyun and J.-W. Lee, Springer, New York, 2009, ch. 4, pp. 231–316; (b) F. La Mantia, M. Santamaria, F. Di Quarto and H. Habazaki, J. Electrochem. Soc., 2010, 157, C258.
13. M. Pourbaix, Atlas of electrochemical equilibria in aqueous solutions, Pergamon Press, Oxford, UK, 1966.
14. Oxides and Oxide Films, ed. J. W. Diggle, Marcel Dekker, Inc., New York, 1973.
15. A. J. Brock and G. C. Wood, Electrochim. Acta, 1967, 12, 395.
16. H. Habazaki, K. Fushimi, K. Shimizu, P. Skeldon and G. E. Thompson, Electrochem. Commun., 2007, 9, 1222.
17. G. C. Wood, P. Skeldon, G. E. Thompson and K. Shimizu, J. Electrochem. Soc., 1996, 143, 74.
18. F. Di Quarto, S. Piazza, A. Splendore and C. Sunseri, Proceedings of the Symposium on Oxide Films on Metals and Alloys, The Electrochemical Society, Inc., Pennington, 1992.
19. K. Shimizu, H. Habazaki, P. Skeldon, G. E. Thompson and G. C. Wood, Electrochim. Acta, 2001, 46, 4379.
20. K. Shimizu, K. Kobayashi, G. E. Thompson, P. Skeldon and G. C. Wood, Philos. Mag. B, 1996, 73, 461.
21. E. McCafferty, Electrochim. Acta, 2010, 55, 1630.
22. E. McCafferty, J. Electrochem. Soc., 1999, 146, 2863.
23. A. F. Wells, Structural Inorganic Chemistry, Clarendon Press, Oxford, UK, 1975.
24. A. Dabirian and R. Van de Krol, Chem. Mater., 2015, 27, 708.
25. P. W. Atkins, The Elements of Physical Chemistry, Oxford University Press, Oxford, UK, 1996.
26. (a) F. Di Franco, M. Santamaria, F. Di Quarto, F. La Mantia, C. M. Rangel and A. I. De Sá, ECS J. Solid State Sci. Technol., 2013, 11, N205; (b) F. Di Franco, G. Zampardi, M. Santamaria, F. Di Quarto and H. Habazaki, J. Electrochem. Soc., 2012, 159, C33; (c) F. La Mantia, H. Habazaki, M. Santamaria and F. Di Quarto, Russ. J. Electrochem., 2010, 46, 1306.
27. A. Zaffora, F. Di Franco, M. Santamaria, H. Habazaki and F. Di Quarto, Electrochim. Acta, 2015, 180, 666.
28. M. Santamaria, F. Di Quarto and H. Habazaki, Electrochim. Acta, 2008, 53, 2272.
29. F. Di Quarto, M. Santamaria, P. Skeldon and G. E. Thompson, Electrochim. Acta, 2003, 48, 1143.
30. M. Santamaria, F. Di Quarto, P. Skeldon and G. E. Thompson, J. Electrochem. Soc., 2006, 153, B518.
31. M. Santamaria, F. Di Quarto, S. Zanna and P. Marcus, Electrochim. Acta, 2011, 56, 10533.
32. M. Santamaria, F. Di Quarto, S. Zanna and P. Marcus, Electrochim. Acta, 2007, 53, 1315.
33. M. Santamaria, F. Di Franco, F. Di Quarto, P. Skeldon and G. E. Thompson, J. Phys. Chem. C, 2013, 117, 4201.
34. F. Di Quarto, F. Di Franco, C. Monarca, M. Santamaria and H. Habazaki, Electrochim. Acta, 2013, 110, 517.
35. N. F. Mott and E. A. Davis, Electronic Processes in Non-crystalline Materials, Clarendon Press, Oxford, UK, 2nd edn, 1979.
36. M. Santamaria, F. Di Quarto and H. Habazaki, Corros. Sci., 2008, 50, 2012.
37. F. Di Quarto, S. Piazza, M. Santamaria and C. Sunseri, in Handbook of Thin Film Materials, ed. H. S. Nalwa, Academic Press, New York, 2002, vol. 2, ch. 8, pp. 373–414.
38. (a) L. Onsager, J. Chem. Phys., 1934, 2, 599; (b) L. Onsager, Phys. Rev., 1938, 54, 554.
39. D. M. Pai and R. C. Enck, Phys. Rev. B: Solid State, 1975, 11, 5163.
40. O. S. Heavens, Optical Properties of Thin Solid Films, Dover Publications, New York, 1965.
41. CRC Handbook of Chemistry and Physics, ed. D. R. Lide, CRC Press/Taylor and Francis, Boca Raton, 90th edn, 2010.
42. S. Adachi, The Handbook on Optical Constants of Metals, World Scientific, 2012.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5cp04347f

---