The influence of composition on band gap and dielectric constant of anodic Al-Ta mixed oxides

Electrochimica Acta 180 (2015) 666–678

Contents lists available at ScienceDirect

Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta

The influence of composition on band gap and dielectric constant of anodic Al-Ta mixed oxides Andrea Zafforaa , Francesco Di Francoa,1, Monica Santamariaa,* ,1, Hiroki Habazakib,1, Francesco Di Quartoa,1 a b

Electrochemical Materials Science Laboratory, DICAM, Università di Palermo, Viale delle Scienze, Ed. 6, Palermo, Italy Graduate School of Chemical Sciences and Engineering, Hokkaido University, Sapporo, Hokkaido, Japan

A R T I C L E I N F O

A B S T R A C T

Article history: Received 14 April 2015 Received in revised form 13 August 2015 Accepted 14 August 2015 Available online 20 August 2015

Al-Ta mixed oxides were grown by anodizing sputter-deposited Al-Ta alloys of different composition. Photocurrent spectra revealed a band gap, Eg, slightly independent on Ta content and very close to that of anodic Ta2O5 (
4.3 eV) with the exception of the anodic film on Al-10at% Ta, which resulted to be not photoactive under strong anodic polarization. The photoelectrochemical characterization allowed to estimate also the oxides flat band potential and to get the necessary information to sketch the energetic of the metal/oxide/electrolyte interfaces. Impedance measurements allowed to confirm the formation of insulating material and to estimate the dielectric constant of the oxides, which resulted to be monotonically increasing with increasing Ta content (from 9 for pure Al2O3 to 30 for pure Ta2O5). ã 2015 Elsevier Ltd. All rights reserved.

Keywords: Anodizing Band gap Flat band potential Dielectric constant Al-Ta mixed oxides

1. Introduction Investigation of high-k materials as alternative gate dielectrics for the technology advancement of complementary metal oxide semiconductor (CMOS) applications has started gradually to shift its focus from single metal oxides (HfO2, Ta2O5, ZrO2, TiO2 or Al2O3) to doped or mixed oxides [1–4]. The idea behind this choice is that by mixing of high-k dielectrics or by their doping with appropriate elements it would be possible to engineer the electrical properties of the materials, combining the favourable properties of the starting dielectrics while suppressing their individual disadvantages. Since a large number of such materials are valve metals oxides, anodizing has been proposed as a simple and low cost process for preparing both single metal and mixed oxides, whose structure, thickness, composition and morphology can be efficiently and easily tailored controlling the metal or alloy composition and the oxidation conditions [5–14]. One of the most promising candidates for storage capacitors in nanoscale dynamic random access memories (DRAMs) [15–17] is Ta2O5 due to its high storage ability and low leakage current. One of the main drawback of this oxide is its band gap which is reported to be low with respect to the value necessary to assure good performances of the devices.

* Corresponding author. ISE member

1

http://dx.doi.org/10.1016/j.electacta.2015.08.068 0013-4686/ ã 2015 Elsevier Ltd. All rights reserved.

In previous works [18–20] it was demonstrated that mixing WO3, HfO2 or TiO2 to Ta2O5 improved the dielectric and insulating properties of thin tantalum oxide films. A promising oxide partner for Ta2O5 is Al2O3, not only due to its very high band gap, but also due its glass former character [21], which reduces the possibility of crystallization. Moreover, Al can be incorporated as substitutional atoms into Ta2O5, acting as acceptor and compensating oxygen vacancies with consequent minimization of leakage current. In spite of this very encouraging perspectives, previous study on Al doped Ta2O5 prepared by radio frequency reactive sputtering and subsequent high temperature annealing has shown that permittivity of the doped tantalum oxide is lower with respect to the corresponding pure oxide, and that some caution must be used in the selection of the metal gate, which can react with Al doped oxide, building up other oxide layers between metal and dielectric with detrimental effect on the performance of CMOS devices [22]. Nevertheless, the variation of electrical properties of mixed oxides strongly depends on the amount of added partner oxide, on the method of incorporation and on the structure of the films, thus, in this work we focused on the preparation and characterization of Al-Ta mixed oxides by anodizing in aqueous solutions sputterdeposited Al-Ta alloys. Photoelectrochemical measurements were carried out in order to estimate the band gap and flat band potential as a function of the oxide composition. Impedance measurements were performed in order to study the electrical properties of the oxides and to estimate the permittivity as a function of their composition. The experimental findings were

A. Zaffora et al. / Electrochimica Acta 180 (2015) 666–678

667

system, at a rate of 0.11 nm s1, by using targets of 99.99wt% Al and 99.9wt% Ta; the chamber was initially evacuated to 5  105 Pa with subsequent sputtering using Ar at 3  101 Pa. Anodizing was performed potentiodynamically at 10 mV s1 at room temperature in a borate buffer (0.42 M H3BO3, 0.08 M Na2B4O7) (pH = 8), in which both Al2O3 and Ta2O5 are thermodynamically stable, according to Pourbaix's diagrams relative to AlH2O and Ta-H2O at room temperature [23]. Alloys were anodized to

used to sketch the energetics of the metal/oxide interface and thus to estimate the band offset. 2. Experimental Aluminium, tantalum and Al-Ta alloys (10, 18, 20, 30, 42, 62, 81, 91 at% Ta) were deposited by dc magnetron sputtering on glass substrates. The deposition was carried out in a Shimadzu, SP-2C

3000

0.6 Ta

2500

0.5 Al-81at.%Ta

2000

0.4

Al-62at.%Ta Al-42at.%Ta

1500

0.3

Al-20at.%Ta 1000

0.2

Al-18at.%Ta

Photocurrent/nA

Photocurrent/nA

Al-91at.%Ta

0.1

500

0

0 200

250

300

350

400

Wavelength/nm Fig. 1. Raw photocurrent spectra relating to anodic films grown up to 10 V vs Hg/HgO on Al-18at% Ta, Al-20at% Ta, Al-42at% Ta, Al-62at% Ta, Al-81at% Ta, Al-91at% Ta alloys and on pure Ta, recorded by polarizing the electrodes at 5 V in 0.1 M ABE.

Fig. 2. Band gap estimate by assuming non direct optical transitions relating to anodic films grown to 10 V on (a) Al-18at% Ta, (b) Al-42at% Ta, (c) Al-62at% Ta, and (d) Al-91at% Ta alloys, recorded by polarizing the electrodes at 5 V vs Hg/HgO in 0.1 M ABE.

668

A. Zaffora et al. / Electrochimica Acta 180 (2015) 666–678

3. Results Anodic films were grown to 10 V at 10 mV s1 on all investigated materials. As typical of valve metals, after an initial overshoot, the current density reaches an almost constant value, which is a function of the metallic substrate composition, and rapidly decays to zero in the backward scan as expected to the blocking character of the oxides (See SI). With the exception of pure Al, the surface of the anodized alloy appeared yellow due to light interference phenomena (See SI).

-150

0

4.2 0

1

2

3

4

ðIph hnÞn / ðhn  Eg Þ

-60

300

-90

200

-120

100

-150 -180 1

2

3

4

5

100 -150 0

-180 5

Photocurrent/nA

150

0

d)

-30

120

-60

90

-90

60

-120

30

-150

0

Phase/°

Photocurrent/nA

180

-30

-120

E vs (Hg/HgO)/V

-30

400

0

-90

4

0

b)

-2 -1 0

Phase/°

200

3

ð1Þ

0

-60

2

1

E vs (Hg/HgO)/V

c)

1

0.8

With the exception of the anodic films grown to 10 V (vs Hg/ HgO) on pure Al and Al-10at% Ta alloy, all other investigated oxides proved to be photoactive under anodic polarization, as shown in Fig. 1, where the photocurrent spectra relating to anodic oxides grown on several Al-Ta alloys, recorded by polarizing the electrodes at 5 V vs Hg/HgO in 0.1 M ABE. The photocurrent intensity is very poor for the oxides on Al18at% Ta and Al-20at% Ta, and increases by increasing Ta content. For photon energy in the vicinity of the band gap, the following equation holds:

5

400

-2 -1 0

0.6

3.1. Photoelectrochemical measurements

E vs (Hg/HgO)/V

300

0.4

Fig. 3. Band gap as a function of Ta atomic content: theoretical prediction (—) according to eq (8a) and best fitting curve (- - -).

-180 -2 -1 0

0.2

Ta atomic fraction

500

-120

200

4.35

Phase/°

400

4.4

4.25

600

-90

y = 0.0113x2 - 0.2355x + 4.4906 R² = 0.9791

4.45

4.3

-30 -60

600

4.5

Photocurrent/nA

800

a)

4.55

0

Phase/°

Photocurrent/nA

1000

4.6

Band gap/eV

a formation potential of 5 V vs mercury/mercury oxide electrode (0 V vs Hg/HgO = 0.098 V vs SHE) and then, after checking the blocking character of the anodic films, up to 10 V (see Fig. S1). The experimental set-up for photoelectrochemical investigations has been described elsewhere [24]: the arrangement is composed by a 450W UV–vis xenon lamp coupled with a monochromator, which allows a monochromatic 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) allows to get photocurrent, by separating it 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 ammonium biborate tetrahydrate (ABE, (NH4)2B4O74H2O) (pH
9). Impedance measurements were carried out in 0.25 M Na2HPO4 (pH
9) through a Parstat 2263 (PAR), connected to a computer for the data acquisition. A Pt net with a very high specific surface was used as counter electrode, while a mercury/mercury oxide as reference electrode. The impedance spectra were generated by superimposing to the continuous potential a sinusoidal signal of amplitude 10 mV over the frequency range 100 kHz–100 mHz. and the results were fitted with ZSimp software.

-180 -2 -1 0

1

2

3

4

5

E vs (Hg/HgO)/V

Fig. 4. Photocurrent and phase vs potential curves relating to 10 V anodic films grown on (a) pure Ta, (b) Al-91at% Ta, (c) Al-81at% Ta, and (d) Al-62at% Ta. Irradiating wavelength: 260 nm, 250 nm for oxide on Al-62at% Ta alloy. Potential scan rate: 10 mV/s.

A. Zaffora et al. / Electrochimica Acta 180 (2015) 666–678

669

Fig. 5. Total current circulating under irradiation (on) and in the dark (off) in the oxide grown on Al-91at% Ta alloy at two different wavelengths, 260 and 280 nm, by polarizing the electrode at 5 V (a), – 0.8 V (b) and – 1 V (c) vs Hg/HgO in a 0.1 M ABE solution.

in which hn is the photon energy, Eg is the optical band gap, and the exponent, n, is 0.5 for indirect (non direct for amorphous materials) optical transitions [25]. The optical band gap can be estimated by extrapolating to zero the (Iphhn)1/2 vs hn plot, as shown in Fig. 2 for 10 V oxides grown on Al-18at% Ta, Al-42at% Ta, Al-62at% Ta and Al-91at% Ta alloys. According to the values reported in Fig. 3, mixed oxides band gap is very close to that estimated for pure anodic Ta2O5 and, thus, significantly lower with respect to the reported band gap values for anodic alumina Al2O3 [26–29]. However, it is important to stress that we were not able to estimate the band gap of the anodic oxide with the lowest tantalum content, i.e. for the anodic film on Al-10at % Ta, due to the absence of anodic photocurrent.Photocurrent vs electrode potential curves under constant irradiating wavelength (photocharacteristics) were recorded by scanning the polarizing voltage toward the cathodic direction at 10 mV s1 in a 0.1 M ABE solution. An inversion of the photocurrent sign was evidenced for anodic films on Ta and on Al-Ta alloys with Ta content 62 at.%, as suggested by the change in photocurrent phase angle (see Fig. 4). This occurs with insulating materials, where both anodic and cathodic photocurrent can be generated depending on the direction of the electric field across the layer and, thus, on the applied potential with respect to the flat band potential, UFB, of the oxide. The photocurrent sign inversion potential, Uinv, is usually assumed as a proxy of the flat band potential. However, a more reliable estimate of UFB can be obtained by recording current transients under constant irradiating wavelength and manually chopping the irradiation, as shown in Fig. 5, where we report the current vs time curves relating to anodic film grown on Al-91at% Ta at three electrode potentials under two different irradiating wavelength, 260 and 280 nm. The ratio between photocurrent and current was so low for Al rich oxides that it was not possible to record current time transients as those of Fig. 5. It is evident that under high anodic potential (i.e. 5 V vs Hg/HgO, see Fig. 5a) the current increases almost instantaneously soon after irradiation in the anodic direction (anodic photocurrent), while at potential slightly anodic with respect to the inversion potential (i.e. –0.8 V vs Hg/HgO, see Fig. 5b) a strong recombination between the photogenerated charge carriers occurs, with consequent reduction

of the photocurrent after few seconds. Interestingly, at –1 V vs Hg/HgO (see Fig. 5c), which is more anodic than the inversion potential (see Fig. 4b), anodic photocurrent spikes are followed by stationary cathodic photocurrent, thus suggesting that UFB must be higher than the Uinv. From the inspection of all the recorded current transients it is possible to set the flat band potential of the 10 V anodic film on Al91at% Ta at
– 0.90 V (Hg/HgO). The same procedure allowed to estimate the flat band potentials reported in Table 1, showing that the presence of Al into the mixed oxides induces a shift toward the cathodic direction of UFB with respect to that estimated for pure Ta2O5, thus reducing the energy distance between conduction band

Table 1 Flat band potential values estimated by recording current transients under constant irradiating wavelength relating to the anodic oxides with a Ta content 62at%. Base alloy

Ufb vs (Hg/HgO)/V

Al-62at% Ta Al-81at% Ta Al-91at% Ta Ta

1 1.20 0.90 0.80

Table 2 Parameters obtained by fitting according to power law, Iphn / UE, the experimental photocharacteristics recorded for anodic films grown on Al-Ta alloys with Ta content 62at%, at 10 mV s1 in 0.1 M ABE. Base alloy

Wavelength/nm

n

V* vs (Hg/HgO)/V

Al-62at% Ta

230 250 270 240 260 280 240 260 280 240 260 280

0.80 0.70 0.60 1 1 0.90 1 0.90 0.80 1 1 1

0.94 1.11 1.24 1.26 1.02 1.03 1.60 1.28 1.38 1.60 1.37 1.14

Al-81at% Ta

Al-91at% Ta

Ta

670

A. Zaffora et al. / Electrochimica Acta 180 (2015) 666–678

1.5 1.5

(Iph)1/2 /a.u.

Photocurrent/nA

1.2

0.9

1.2 0.9 0.6 0.3 0

0.6

1.5

2

3

2.5

3.5

hν/eV 0.3

0 400

450

500

550

600

650

700

Wavelength/nm Fig. 6. Cathodic photocurrent spectrum relating to the oxide grown on Al-91at% Ta alloy recorded at –1.5 V vs Hg/HgO as electrode potential in 0.1 M ABE, using a UV filter. Inset: internal photoemission threshold energy estimate according to Fowler’s law.

This is confirmed by fitting the Iph vs potential curves recorded at different wavelength according to power law

Table 3 Internal photoemission threshold energy values estimated according to Fowler's law relating to several anodic films grown on Al-Ta alloys. Base alloy

Eth/eV

Al-10at% Ta Al-20at% Ta Al-30at% Ta Al-62at% Ta Al-91at% Ta Ta

2.13 2.21 2.23 1.53 1.50 1.41

Iph n / UE

edge and oxide Fermi level. Moreover, it is evident that the reported values are more anodic than the inversion potential. Such difference can be explained by the presence of localized states inside the mobility gap of the oxide, which can behave as traps for the injected photocarriers modifying the electric field distribution across the oxide [30].

ð2Þ

The best fitting exponent n and the potential corresponding to Iphn = 0, V*, are reported in Table 2. It is noteworthy to mention that for the mixed oxides the photocharacteristics displayed a supralinear (n < 1) behaviour with n decreasing with increasing l. For insulating crystalline films a linear dependence of the measured photocurrent on the applied potential is expected in absence of trapping phenomena which can modify the electric field distribution across the layer [31–33]. The supralinear behaviour can be explained by the presence of surface and bulk recombination phenomena involving the photogenerated carriers, as well as to geminate recombination effects generally occurring in any material where the photogenerated carriers display very low mobility (e.g. amorphous films). In presence of geminate recombination effects an influence of the

0.8 1 0.7

(Iph)1/2/a.u.

0.8

Photocurrent/nA

0.6 0.5

0.6 0.4 0.2

0.4

0

0.3

2

2.5

3

3.5

hν/eV

0.2 0.1 0 400

450

500

550

600

650

700

Wavelength/nm Fig. 7. Cathodic photocurrent spectrum relating to the oxide grown on Al-10at% Ta alloy recorded at –1.5 V vs Hg/HgO as electrode potential in 0.1 M ABE, using a UV filter. Inset: internal photoemission threshold energy estimate according to Fowler’s law.

A. Zaffora et al. / Electrochimica Acta 180 (2015) 666–678

671

Fig. 8. EIS spectra relating to all the investigated anodic films, recorded by polarizing the film at 1 V vs Hg/HgO in 0.25 M Na2HPO4. (a) Modulus and (b) Phase angle. Inset: equivalent circuit.

Table 4 Fitting parameters relating to EIS spectra of the all investigated anodic films using equivalent electric circuit of Fig. 8a. Base alloy

Rohmic/V cm2

Rct/V cm2

QH/S sa cm2

a

Rox/V cm2

Qox/S sa cm2

a

Al Al-10at% Ta Al-18at% Ta Al-20at% Ta Al-30at% Ta Al-42at% Ta Al-62at% Ta Al-81at% Ta Al-91at% Ta Ta

10 14 32 21 17 31 30 40 28 57

6  105 1 105 2  104 2  104 1 104 1 105 1 104 1 105 1 105 3  105

2.1 105 2.1 105 2.4  105 2.1 105 2.8  105 2.2  105 2.4  105 2.2  105 2.3  105 2.4  105

0.90 0.80 0.87 0.88 0.90 0.86 0.91 0.90 0.90 0.90

1 108 1 108 2  108 1 108 1 108 1 108 1 108 1 108 1 108 1 109

8.3  107 8.8  107 9.8  107 9.2  107 9.4  107 1.0  106 1.4  106 1.4  106 1.6  106 1.7  106

0.98 1 0.99 1 0.99 1 0.99 1 1 1

electric field and of the photon energy through the thermalization distance of the photocarriers [33] is expected on the efficiency of photocarriers generation, hg, which can explain the supralinear behaviour of the photocharacteristics. The best fit exponent

n = 1 for anodic film on Ta suggests an independence of Iph on

hg, owing to the almost constant efficiency of charge separation in

the exploited potential (i.e. electric field) range [33]. The presence of cathodic photocurrent allowed recording photocurrent spectra at potentials more cathodic than the inversion potential, as shown in Fig. 6 for the anodic film on Al91at% Ta. It is evident the presence of photocurrent at hn < Eg, thus it was possible to record photocurrent even for l  400 nm using an UV filter (lcutoff = 400 nm) to avoid doubling effect on the measured photocurrent. The measured photocurrent is due to an electrons injection process from the alloy Fermi level to the oxide conduction band. The threshold energy, Eth, associated to this process can be estimated according to Fowler’s law [9]: ðIph Þ1=2 / ðhn  Eth Þ

ð3Þ

as shown in Fig. 6. All the estimated Eth values are reported in Table 3 as a function of the oxides composition. Notably, in spite the absence of anodic photocurrent in the case of anodic film

672

A. Zaffora et al. / Electrochimica Acta 180 (2015) 666–678

oxide capacitance. The impedance of the Constant Phase Element is

grown on Al-10at% Ta, cathodic photocurrent was revealed allowing the estimation of the Fowler threshold for this oxide. In Fig. 7 we report the cathodic photocurrent spectrum in the long wavelength region relating to Al-10at% Ta recorded at UE = –1.5 V (Hg/HgO), from which a Fowler threshold of
2.1 eV can be estimated.

ZðvÞ ¼

1 ðjvÞa Q

ð4Þ

where v is the frequency of the a.c. perturbation signal and a is a fitting parameter ranging from 0 to 1, with a = 1 corresponding to an ideal capacitance [34]. Another RQ parallel was inserted to model the electrochemical reaction, where Rct is the chargetransfer resistance and QH the non ideal Helmholtz double layer capacitance. This equivalent circuit very well fits the experimental data, with the fitting parameters reported in Table 4. It is noteworthy to mention that QH
20 mF cm2 (a is very close to 1) as expected in highly concentrated aqueous solution [35]. Moreover, the very high Rox values and a very close to 1 suggest that the oxides behave like ideal capacitors. It is interesting to mention that EIS spectra were also recorded at potential more cathodic than UFB, as shown in Fig. 9. For anodic oxides grown on alloys with a Ta content 62at% the cathodic EIS spectra are similar

3.2. Impedance measurements In order to get information on the dielectric properties of the investigated oxides, we recorded electrochemical impedance spectra and differential capacitance curves. In Fig. 8 we report the EIS spectra in Bode representation relating to all the investigated oxides recorded in 0.25 M Na2HPO4 at UE = 1 V (Hg/HgO). As shown in the figures, they can be simulated by the electrical circuit reported in the inset, where Rohmic is the ohmic resistance (accounting for contact and electrolyte resistances), Rox is the anodic film resistance and Qox is a Constant Phase Element (CPE) introduced to model the

1.00E+05

a)

Magnitude/Ω cm2

1.00E+04

1.00E+03 Ta Al-91at.%Ta 1.00E+02 Al-81at.%Ta

1.00E+01 0.1

1

10

100

1000

10000

100000

f/Hz 90

b) 80 70

Phase/°

60 50 40 Ta

30

Al-91at.%Ta

20

Al-81at.%Ta

10 0 0.1

1

10

100

1000

10000

100000

f/Hz Fig. 9. EIS spectra relating to anodic films grown on Al-81at% Ta and Al-91at% Ta alloys and on pure Ta2O5, recorded by polarizing the film at – 1.2 V vs Hg/HgO in 0.25 M Na2HPO4. (a) Modulus and (b) Phase angle.

A. Zaffora et al. / Electrochimica Acta 180 (2015) 666–678

673

4 Ta Al-91at.%Ta Al-81at.%Ta Al-62at.%Ta Al-42at.%Ta Al-30at.%Ta Al-20at.%Ta Al-18at.%Ta Al-10at.%Ta Al

a)

C/µF cm-2

3

2

1

0 -2

-1

0

1

2

3

4

5

4 Ta Al-91at.%Ta Al-81at.%Ta Al-62at.%Ta Al-42at.%Ta Al-30at.%Ta Al-20at.%Ta Al-18at.%Ta Al-10at.%Ta Al

b)

C/µF cm-2

3

2

1

0 -2

-1

0

1

2

3

4

5

E vs (Hg/HgO)/V Fig. 10. Measured series capacitance relating to all oxides grown up to 10 V vs Hg/HgO in borate buffer. a.c. signal frequencies: (a) 1 kHz, (b) 100 Hz. Characterization solution: 0.25 M Na2HPO4.

to those recorded under anodic potential, as also confirmed by the best fitting parameters (not shown). In contrast, for anodic films with higher tantalum content a significant reduction of Rox is evidenced. In Fig. 10 we report the differential measured capacitance, CM, curves relating to all the investigated oxides recorded at two constant a.c. signal frequencies. At 1 kHz CM is almost potential independent for all the investigated oxides (see Fig. 10a), as expected for insulating amorphous oxides [13], while a careful inspection of Fig. 10b shows that, at 100 Hz, CM slightly increases under strong cathodic polarization (UE < UFB) for Ta2O5 and for mixed oxides with a tantalum content  81at%. This finding is in agreement with the EIS spectra of Fig. 9, which show that there is a change in the impedance of these oxides with respect to that measured under anodic polarization. Provided that the equivalent circuit of Fig. 8a well simulates the impedance of the overall metal/oxide/electrolyte interface, it is easy to extract the oxide capacitance from CM. Assuming that Cox

can be described by a parallel plate capacitor model: C ox ¼

er e0 d

ð5Þ

where e0 (8.85  1014 F cm1) is the vacuum permittivity, er the film relative permittivity and d its thickness. By knowing d, it is possible to estimate er as a function of the oxide composition. The thickness values (reported in Fig. 11) inserted in eq. (5) were estimated from the anodizing ratios reported in Refs. [36,37], which in turn were estimated by a direct inspection of the film thickness by transmission electron microscopy of the oxides’ ultramicrotomed cross sections. The dielectric constants derived from CM measured at 1 kHz and under strong anodic polarization are reported in Fig. 11, where it is evident that er varies monotonically from 9 (pure Al2O3, in accordance to Ref. [38]) to 30 (pure Ta2O5). Thus the addition of small amount of Al2O3 into Ta2O5 induces a reduction in the pure tantalum oxide dielectric constant, in agreement with the results reported in Ref. [39] and in contrast to what reported in Ref. [40].

674

A. Zaffora et al. / Electrochimica Acta 180 (2015) 666–678

35

specifically, Eg was found to depend linearly on the square of the difference of electronegativity (Pauling’s scale) between oxygen (xO) and metal (xM), according to the following equation:

18 16

30

14 12 10

15

8

εr

20

6

10

Thickness/nm

25

4 5

Eg  DEam ¼ AðxO  xM Þ2 þ B

ð6Þ

where A and B are determined by best-fitting of the experimental Eg vs (xO  xM)2 data. With the exception of Ni, which followed the sp correlation, the following two equations were found for sp metal and d metal oxides respectively: spmetalÞ

Eg  DEam ðeVÞ ¼ 2:17ðxM  xO Þ2  2:71

ð6aÞ

2 0

0 0

20

40

60

80

100

%at Ta Fig. 11. Dielectric constant and thickness values estimated for all the investigated oxides as a function of the Ta content into the base alloy.

4. Discussion The study of Al2O3 and Ta2O5 mixed oxides presents some interesting issues both for their possible application as high-k materials in microelectronics and as a model system to test some correlation between optical band gap and mixed oxide composition. Both oxides have high dielectric constant and are wide band gap materials. Several polymorphs are reported to exist for aluminium oxide and their band gap changes significantly, according to values reported in the literature experimentally estimated or theoretically calculated by quantum-chemical density functional theory [41]: Eg = 4.4 eV is reported for h-alumina, while Eg = 8.5 eV is reported for a-Al2O3. Moreover, Eg = 6.2 eV is reported for anodic oxide as well as for physical evaporated amorphous Al2O3 [42,43] in very good agreement with the value estimated for the g-Al2O3 phase, which has been suggested to display a very low crystallinity and possibly a missing long-range order [41]. No meaningful differences are reported for different Ta2O5 polymorphs: Eg between 3.90 eV and 4.30 eV is reported [44] with the highest values measured for thin anodic films (see also above) usually displaying a microcrystalline structure with a short range order quite similar to the crystalline monoclinic phase [45,46].The experimental findings arising from photoelectrochemical investigation and from impedance measurements on anodic films on sputtering-deposited Al-Ta alloys of different compositions provided evidence that by changing the Ta content in the base alloy the optical band gap does not change significantly, while the dielectric constant increases on going from pure Al2O3 (i.e. e = 9), to that estimated for Ta2O5 (i.e. e = 30). The mixed oxide composition is directly related to the base alloy composition and to the migration rates of the metal ions. Al3+ and Ta5 + have transport numbers t > 0 in Al2O3 and Ta2O5 respectively (tAl3þ ¼ 0:4 and tTa5þ ¼ 0:24 [36,37]). During anodizing of Al-Ta alloys, Al3+ and Ta5+ ions migrate toward the oxide/electrolyte interface at similar rates [36,37], with simultaneous migration of O2 toward the metal/oxide interface. According to previous results reported in Refs. [36,37], oxides composition is almost coincident with base alloys composition, provided that Ta content is >15at%. In previous work [47], a correlation between the optical band gap values of crystalline oxides, MxOy, and the square of the electronegativity difference of their constituents was proposed. Such a correlation was derived by assuming a direct relation between the optical band gap and the single M-O bond energy, using the Pauling equation for the single bond energy. More

d  metalÞEg  DEam ðeVÞ ¼ 1:35ðxM  xO Þ2  1:49

ð6bÞ

where DEam ¼ 0 for crystalline oxides, whilst increasing DEam values have been suggested as the degree of crystallinity decreases. The recent Eg theoretical estimates of Peitinger et al. above reported for g-Al2O3 lends further support to the validity of the proposed correlations (eq. (6)) derived by assuming a direct relationship between the optical band gap and the single M-O bond energy, with this last parameter calculated by using the Pauling’s equation for the single bond energy. As recently noted by Walsh and Butler, the chemical approach reflected in eq. (6) is appealing for its conceptual simplicity as well as for its ability to provide reliable estimates of the optical band gap of many binary oxides, even if it has some limitations since such an approach “cannot account for the effects of bonding or crystal structures” [48]. However, in the case of mixed oxides with metals belonging to the same d-d or sp-sp groups, it has been shown that eqs. (6a) and (6b) are able to provide reliable estimates of Eg, provided that the difference of electronegativity (EN) of cationic species is 0.6. In these cases we suggested that the average single bond energy could be expressed as weighted contribution of cations involved in the network formation. Thus, the same correlation has been extended to the case of mixed oxides, by assuming an average cationic electronegativity defined as:

xM;av ¼ Xi xi þ Xj xj

ð7Þ

where i and j refer to the two metals in the “mixed” oxide, and xi and xj represents their cationic atomic fraction. We have shown that such a procedure is able to provide in a straightforward way a quadratic dependence of Eg on the oxide composition [49] as well as the so-called bowing parameter traditionally invoked to fit the compositional dependence of Eg of random semiconducting alloys [50]. The extension of eq. (6) to sp-d mixed oxides poses the problem of which best-fitting line should be used. In a recent work on a very similar system we have shown that the anodic oxide films grown on Al-Nb alloys display optical band gap steadily growing with increasing Al content in the alloy (and then in the anodic oxide) and that Eg values could be nicely fitted as a function of the difference of EN between oxygen (xO) and cation (xM,av), according to the following relationship: EgðAlNbÞ ðeVÞ ¼ 2:1451ðxO  xM;av Þ2  4:2767

ð8aÞ

where the average cationic electronegativity has been calculated by means of eq. (7) above reported. A rapid check of the validity of eq. (8a) in fitting the band gap values of anodic films on Al-Ta alloys can be carried out by assuming for Ta and Al metal the common value of Pauling's electronegativity parameter (x; = 1.5) reported by Pauling [51]. After a trivial calculation we get a band gap value of 4.30 eV in very good agreement with the Eg value measured for oxide film grown on Al-81at% Ta.

A. Zaffora et al. / Electrochimica Acta 180 (2015) 666–678

675

Fig. 12. Approximate sketch of the energetic levels of metal/oxide/electrolyte interface for anodic films grown on (a) Al-20at% Ta (shaded area represents strongly localized states due to the very low Ta content) and (b) Al-91at% Ta alloys.

676

A. Zaffora et al. / Electrochimica Acta 180 (2015) 666–678

However, a closer inspection of Fig. 3 evidences a systematic dependence of Eg on film composition: band gap of the oxides slightly increases by increasing aluminium content in the base alloy. In previous papers on the photoelectrochemical behaviour of anodic films on Al-valve metals alloys (Al-W, Al-Nb, Al-Ti) [7], [12], we suggested that DEam (see eqs. (6a) and (6b)) can be influenced by oxides composition, and thus can explain small changes in the measured band gap values. Although such a possibility cannot be completely dismissed, the results of the previous investigations on Al-Nb mixed oxides encouraged us to propose another physically sound interpretation of such dependence, based on eq. (8a), taking into account a recent theoretical study by M.F. Peintinger et al. on the band gap of stable and metastable alumina polymorphs. According to these authors, an indirect band gap of 6.2 eV can be calculated by DFT for g-Al2O3, a metastable phase of alumina characterized by a poor crystallinity. This value is in the range (6.0–6.3 eV) proposed for thin amorphous anodic films grown on pure aluminium reported in Ref. [26]. According to this, by using as trial parameter for anodic alumina an average band gap value of 6.15 eV, an EN parameter xAl = 1.478 can now be derived by means of eq. (6a) (with DEam = 0) in the range foreseen by Pauling (1.5  0.05) [51]. As discussed in Refs. [12], [44] for Al-Nb mixed oxides, eq. (8a) predicts a quadratic a dependence of Eg on composition, provided that xav is expressed according to eq. (7): Eg ðxÞðeVÞ ¼ Eg;i þ 2Axj ðxi  xj Þðxan  xi Þ þ Axj 2 ðxi  xj Þ2

ð8bÞ

where A = 2.1451 and xi and xj are the EN parameters of the two cations present in the oxide film. Such equation can be extended to Al-Ta mixed oxides, by substituting xAl = 1.478 and by deriving xTa through an educated guess, which allows to get a good agreement between the experimental Eg values and the expectations according to eq. (8b). We have also kept as constrain for Al2O3 a band gap of 4.50 eV which is very close to the value calculated for h-Al2O3, a defective spinel structure, studied in Ref. [41]. It is tempting to hypothesize that in the h-Al2O3 phase the Al-O bond strength is similar to that existing in anodic Al-Ta mixed oxides containing at least 20at.% of d-metal and that for Ta content below such threshold, a change toward the more stable g-Al2O3 structure, like in pure anodic alumina, occurs. The higher Eg value (6.2 eV) estimated for g-Al2O3, but very similar to that reported for anodic alumina, could be interpreted as an indication that in such structure the Al-O bond presents a larger strength.It is finally interesting to compare the quadratic dependence predicted by eq. (8b), based on the above described chemical approach, with the usual bowing equation, assumed for random semiconducting alloys: Eg ðxÞ ¼ Eg1 ð1  xÞ þ Eg2 x  bxð1  xÞ ¼ Eg1 þ ax þ bx

2

ð9Þ

where Eg1 and Eg2 are the band gap values of the pure alloys constituents. If we compare eq. (8b) with eq. (9), b = A (xTa  xAl)2
0 and a = 2A(xi-xj)(xan  xi) = 0.234 are expected. These values are in agreement with those estimated by fitting the band gap values with a second order polynomial (see Fig. 3), further supporting the consistence of the proposed equation. Knowledge of the band gap and flat band potential enables considerations on the energetic of the alloy/oxide/electrolyte interfaces, which is of key importance for the possible practical application of these films in CMOS based devices [21]. In order to have a complete image of the energy levels for the investigated oxides, we need to estimate the band edges position as a function of the Ta content. The first step is to set EM F (metal Fermi level) for the alloys at –4.25 eV vs vacuum independently on their composition, due to the very close work functions reported for Al and Ta [52]. The second step is to locate the conduction band mobility edge, ECB, which is relatively easy for the investigated

oxides since the energy distance between EM F and ECB is coincident with Eth estimated from the cathodic photocurrent spectra (see section 3.1). Knowledge of the band gap allows to locate the valence band mobility edge, EVB, while starting from the flat band potential, it was possible to set the oxide Fermi level, Eox F , according to the following relationship [12]: Eox F ¼ jejUFB þ jejUref

ð10Þ

where e is the electron charge and Uref is the potential of the employed reference electrode with respect to the vacuum scale, set to –4.6 eV [53]. Fig. 12 shows the approximate energetic sketches for anodic films on sputter deposited Al-20at% Ta and Al-91at% Ta. In representing the energetic levels of metal/oxide/electrolyte interface, we considered an ideal representation of the junction without stressing the behaviour of the interface under working conditions. Thus, we assumed the oxide at the corresponding flat band potential, neglecting the effect of charge exchange at the metal/oxide interface and possible change in band bending due to charge trapping phenomena that can occur with amorphous materials [7,8], [11,12], [27–29], [54–57]. It is worthy to mention that the presence of Al2O3 into anodic Ta2O5 for Ta rich oxides cause a reduction of the energy distance between the ECB and the oxide Fermi level due to the cathodic shift of the flat band potential. This can explain why under cathodic polarization a reduction of the Ta rich oxides' resistance is revealed. Moreover, we need to split the energy levels corresponding to the Al 3s-3p and Ta 5d orbitals to account for the estimated Eg and Eth for anodic oxide on Al-20at.% Ta. According to the internal photoemission threshold estimated for such oxide we should locate EVB at energy far with respect to the expected value of
–7.2 eV, as already found for Al-Nb mixed oxides [12]. Only when Ta content is sufficiently high the band structure of the oxide becomes similar to that of Ta2O5. This explains also why when Ta content is even lower (i.e. anodic film on Al-10at% Ta) we are able to measure Eth, while we do not record any anodic photocurrent. Ta content is not enough to induce the formation of the states that allow to measure a band gap of 4.44 eV for the anodic film on Al-20at% Ta and oxide valence band edge becomes closer to EVB of g-Al2O3 located at energy (
–8.2 eV according to Ref. [12]) lower than EVB of Al-Ta mixed oxides (see Fig. 12). 5. Summary and Conclusions Al-Ta mixed oxides of different composition were grown by anodizing sputter-deposited Al-Ta alloys in slightly alkaline solution. The photoelectrochemical characterization allowed to estimate the band gap of the oxides as a function of their composition, with the exception of the oxide with the lowest Ta content, i.e. for the anodic film on Al-10at% Ta alloy. Eg resulted to be slightly dependent on the oxide composition and this dependence was rationalized in the frame of a correlation between the band gap of mixed oxides and the electronegativity of their constituents. More specifically, such chemical approach foresees a quadratic dependence, which recalls the bowing equation, usually employed to fit the band gap values of semiconducting alloys as a function of their composition. According to this interpretation, the bowing parameter has been directly related to the difference of electronegativity of the partner metals in the mixed oxides. A band gap of 4.5 eV is used for alumina which is very close to the value calculated by DFT for h-Al2O3 polymorph which is much lower than the band gap usually reported for anodic Al2O3, thus suggesting that the presence of Ta in the mixed oxides has a strong influence on the strength of Al-O bonds. The dependence of Iph on the electrode potential, i.e. on the electric field, suggests that all the investigated anodic films behave

A. Zaffora et al. / Electrochimica Acta 180 (2015) 666–678

like insulating amorphous materials, while the current vs time transients recorded at constant electrode potential by manually chopping irradiation allowed to get a reliable estimate of the oxide flat band potential. From the long wavelength region of cathodic photocurrent spectra it was possible to estimate the Fowler threshold for internal electron photoemission processes and, thus, the energy distance between the conduction band mobility edge and the alloy Fermi level, even for the anodic film on Al-10at% Ta, which was used to sketch the energetics of the metal/oxide/electrolyte interfaces. The impedance measurements allowed to further support the insulating behaviour of the investigated oxides and to estimate their dielectric constant as a function of the composition. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. electacta.2015.08.068. References [1] L. Manchanda, M.D. Morris, M.L. Green, R.B. van Dover, F. Klemens, T.W. Sorsch, P.J. Silverman, G. Wilk, B. Busch, S. Aravamudhan, Multi-component high-k gate dielectrics for the silicon industry, Microelectron. Eng 59 (2001) 351. [2] M. Koyama, A. Kaneko, T. Ino, M. Koike, Y. Kamata, R. Iijima, Y. Kamimuta, A. Takashima, M. Suzuki, C. Hongo, S. Inumiya, M. Takayanagi, A. Nishiyama, Effects of nitrogen in HfSiON gate dielectric on the electrical and thermal characteristics, IEEE Technical Digest of International Electron Devices Meeting, San Francisco, CA, 2002 p. 849. [3] A. Paskaleva, A.J. Bauer, M. Lemberger, S. Zurcher, Different current conduction mechanisms through thin high-k HfxTiySizO films due to the varying Hf to Ti ratio, J. Appl. Phys. 95 (2004) 1124. [4] J. Petry, O. Richard, W. Vandervost, T. Conard, J. Chen, V. Cosnier, Effect of N2 annealing on AlZrO oxide, J. Vac. Sci. Technol. A 21 (2003) 1482. [5] M. Santamaria, D. Huerta, S. Piazza, C. Sunseri, F. Di Quarto, The Influence of the Electronic Properties of Passive Films on the Corrosion Resistance of Mo-Ta Alloys A Photoelectrochemical Study, J. Electrochem. Soc. 147 (2000) 1366. [6] F. Di Quarto, M. Santamaria, P. Skeldon, G.E. Thompson, Photocurrent Spectroscopy Study of Passive Films on Hafnium and Hafnium-Tungsten Sputtered Alloys, Electrochim. Acta 48 (2003) 1143. [7] M. Santamaria, F. Di Quarto, P. Skeldon, G.E. Thompson, Effect of Composition on the Photoelectrochemical Behavior of Anodic Oxides on Binary Aluminum Alloys, J. Electrochem. Soc. 153 (2006) B518. [8] M. Santamaria, F. Di Quarto, H. Habazaki, Influences of structure and composition on the photoelectrochemical behaviour of anodic films on Zr and Zr-20at.%Ti, Electrochim. Acta 53 (2008) 2272. [9] M. Santamaria, F. Di Quarto, H. Habazaki, Photocurrent Spectroscopy Applied to the Characterization of Passive Films on Sputter-Deposited Ti–Zr Alloys, Corros. Sci. 50 (2008) 2012. [10] M. Fogazza, M. Santamaria, F. Di Quarto, S.J. Garcia-Vergara, I. Molchan, P. Skeldon, G.E. Thompson, H. Habazaki, Formation of anodic films on sputteringdeposited Al-Hf alloys, Electrochim. Acta 54 (2009) 1070. [11] F. Di Franco, G. Zampardi, M. Santamaria, F. Di Quarto, H. Habazaki, Characterization of the solid state properties of anodic oxides on magnetron sputtered Ta, Nb and Ta-Nb alloys, J. Electrochem. Soc. 159 (2012) C33. [12] M. Santamaria, F. Di Franco, F. Di Quarto, P. Skeldon, G.E. Thompson, Tailoring of the solid state properties of Al-Nb mixed oxides: a photoelectrochemical study, J. Phys. Chem. C 117 (2013) 4201. [13] F. Di Franco, M. Sántamaria, F. Di Quarto, F. La Mantia, C.M. Rangel, A.I. De Sá, Dielectric Properties of Al-Nb amorphous mixed oxides, ECS J. Solid State Sci. Technol. 11 (2013) N205. [14] F. Di Franco, M. Santamaria, F. Di Quarto, E. Tsuji, H. Habazaki, The influence of nitrogen incorporation on the optical properties of anodic Ta2O5, Electrochim. Acta 59 (2012) 382. [15] International Technology Roadmap for Semiconductors (ITRS). http://public. itrs.net (accessed Apr 14, 2015). [16] G.D. Wilk, R.M. Wallace, J.M. Anthony, High-k gate dielectrics: Current status and materials properties considerations, J. Appl. Phys. 89 (2001) 5243. [17] E. Atanassova, T. Dimitrova, Thin Ta2O5 Layers on Si as an Alternative to SiO2 for High-Density DRAM Applications, in: H.S. Nalwa (Ed.), Handbook of Surfaces and Interfaces of Materials, Vol. 4, Academic Press, San Diego, CA, 2001 Ch. 9, p. 439. [18] K.M.A. Salam, H. Fukuda, S. Nomura, Effects of additive elements on improvement of the dielectric properties of Ta2O5 films formed by metalorganic decomposition, J. Appl. Phys. 93 (2003) 1169.

677

[19] W.S. Lau, T.S. Tan, P. Babu, N.P. Sandler, Mechanism of leakage current reduction of tantalum oxide capacitors by titanium doping, Appl. Phys. Lett. 90 (2007) 112903. [20] J. Lu, Y. Kuo, J.Y. Tewg, Hafnium-doped tantalum oxide high-k gate dielectrics, J. Electrochem. Soc. 153 (2006) G410. [21] J. Robertson, R.M. Wallace, High-k materials and metal gates for CMOS applications, Mater. Sci. Eng. R 88 (2015) 1. [22] A. Skeparovski, N. Novkovski, E. Atanassova, A. Paskaleva, V.K. Lazarov, Effect of Al gate on the electrical behaviour of Al-doped Ta2O5 stacks, J. Phys. D: Appl. Phys. 44 (2011) 235103. [23] M. Pourbaix, Atlas of electrochemical equilibria in aqueous solutions, Pergamon Press, Oxford, 1966. [24] F. Di Quarto, F. La Mantia, M. Santamaria, Physicochemical Characterization of Passive Films and Corrosion Layers by Differential Admittance and Photocurrent Spectroscopy, in: S.-I. Pyun, J.-W. Lee (Eds.), Modern Aspects of Electrochemistry, No. 46: Progress in Corrosion Science and Engineering I, Ch. 4, Springer, New York, 2009 p. 231. [25] N.F. Mott, E.A. Davis, Electronic Processes in Non-crystalline Materials, Clarendon Press, Oxford, 1979. [26] F. Di Quarto, C. Gentile, S. Piazza, C. Sunseri, Photoelectrochemical Study on Anodic Aluminum Oxide Films Internal Photoemission Processes at the MetalOxide Interface, J. Electrochem. Soc. 138 (1991) 1856. [27] N.V. Nguyen, O.A. Kirillov, W. Jiang, W. Wang, J.S. Suehle, P.D. Ye, Y. Xuan, N. Goel, K.-W. Choi, W. Tsai, S. Sayan, Band Offsets of Atomic-Layer-Deposited Al2O3 on GaAs and the Effects of Surface Treatment, Appl. Phys. Lett. 93 (2008) 082105-1. [28] V.V. Afanas’ev, M. Houssa, A. Stesmans, Band Alignments in Metal–Oxide– Silicon Structures with Atomic-Layer Deposited Al2O3 and ZrO2, J. Appl. Phys. 91 (2002) 3079. [29] Y.-C. Yeo, T.-J. King, C.J. Hu, Metal-Dielectric Band Alignment and Its Implications for Metal Gate Complementary Metal-Oxide-Semiconductor Technology, J. Appl. Phys. 92 (2002) 7266. [30] H. Gerischer, On the Interpretation of Photoelectrochemical Experiments with Passive Layers on Metals, Corros. Sci. 31 (1990) 81. [31] R. Crandall, Modeling of thin film solar cells: Uniform field approximation, J. Appl. Phys. 54 (1983) 7176. [32] F. Di Quarto, S. Piazza, R. D’Agostino, C. Sunseri, Amorphous semiconductorelectrolyte junction. Influence of the wavelength and oxide thickness on the photocharacteristics of a-Nb2O5 anodic films, J. Electroanal. Chem. 228 (1987) 119. [33] F. Di Quarto, F. Di Franco, C. Monarca, M. Santamaria, H. Habazaki, Photoelectrochemical characterization of amorphous anodic films on Ti-6at.% Si, Electrochim. Acta 110 (2013) 517. [34] M.E. Orazem, B. Tribollet, Electrochemical Impedance Spectroscopy, J. Wiley, Hoboken, 2008. [35] H. Gerischer, Neglected problems in the pH dependence of the flatband potential of semiconducting oxides and semiconductors covered with oxide layers, Electrochim. Acta 34 (1989) 1005. [36] H. Habazaki, K. Shimizu, P. Skeldon, G.E. Thompson, G.C. Wood, Interrelationships between ionic transport and composition in amorphous anodic oxides, Proc. Roy. Soc. A 453 (1997) 1593. [37] G. Alcalá, S. Mato, P. Skeldon, G.E. Thompson, P. Bailey, T.C.Q. Noakes, H. Habazaki, K. Shimizu, Anodic film growth in the Al-Ta alloy system, Corros. Sci. 45 (2003) 1803. [38] G. Scaduto, M. Santamaria, P. Bocchetta, F. Di Quarto, The effect of hydration layers on the anodic growth and on the dielectric properties of Al2O3 for electrolytic capacitors, Thin Solid Films 550 (2014) 128. [39] S. Luby, Dielectric properties of anodic oxides formed on sputtered Ta-Al alloy films, Thin Solid Films 32 (1976) 61. [40] D. Spassov, E. Atanassova, A. Paskaleva, Lightly Al-doped Ta2O5: Electrical properties and mechanisms of conductivity, Microelectron. Reliab. 51 (2011) 2102. [41] M.F. Peintinger, M.J. Kratz, T. Bredow, Quantum-chemical study of stable, meta-stable and high-pressure alumina polymorphs and aluminum hydroxides, J. Mater. Chem. A 2 (2014) 13143. [42] Y.-N. Xu, W.Y. Ching, Self-Consistent Band Structures, Charge Distributions, and Optical-Absorption Spectra in MgO, a-Al2O3, and MgAl2O4, Phys. Rev. B 43 (1991) 4461. [43] T.T. Jiang, Q.Q. Sun, Y. Li, J.J. Guo, P. Zhou, S.-J. Ding, D.W. Zhang, Towards the Accurate Electronic Structure Descriptions of Typical High-Constant Dielectrics, J. Phys. D: Appl. Phys. 44 (2011) 185402. [44] E. Franke, C.L. Trimble, M.J. DeVries, J.A. Woollam, M. Schubert, F. Frost, Dielectric function of amorphous tantalum oxide from the far infrared to the deep ultraviolet spectral region measured by spectroscopic ellipsometry, J. Appl. Phys. 88 (2000) 5166. [45] L.A. Aleshina, V.P. Malinenko, A.D. Phouphanov, N.M. Jakovleva, The shortrange order of anodic amorphous oxide films of Ta and Nb, J. Non-Cryst. Solids 87 (1986) 350. [46] V.A. Shvets, V. Sh. Aliev, D.V. Gritsenko, S.S. Shaimeev, E.V. Fedosenko, S.V. Rykhlitski, V.V. Atuchin, V.A. Gritsenko, V.M. Tapilin, H. Wong, Electronic structure and charge transport properties of amorphous Ta2O5 films, J. NonCryst. Solids 354 (2008) 3025. [47] F. Di Quarto, C. Sunseri, S. Piazza, M.C. Romano, Semiempirical Correlation between Optical Band Gap Values of Oxides and the Difference of Electronegativity of the Elements. Its Importance for a Quantitative Use of

678

[48] [49]

[50] [51] [52]

A. Zaffora et al. / Electrochimica Acta 180 (2015) 666–678 Photocurrent Spectroscopy in Corrosion Studies, J. Phys. Chem. B 101 (1997) 2519. A. Walsh, K.T. Butler, Prediction of Electron Energies in Metal Oxides, Acc. Chem. Res. 47 (2014) 364. F. Di Quarto, F. Di Franco, M. Santamaria, A chemical approach to the estimate of the optical band gap and bowing parameter in mixed d-d metal oxides, in: J. Heber, D. Schlom, Y. Tokura, R. Waser, M. Wuttig (Eds.), Technical Digest of Frontiers in Electronic Materials, Aachen, Germany, 2012 17-20 June, p. 648.. J.E. Bernard, A. Zunger, Electronic Structure of ZnS, ZnSe, ZnTe, and Their Pseudobinary Alloys, Phys. Rev. B 36 (1987) 3199. L. Pauling, The Nature of Chemical Bond, Ch. 3, Cornell University Press, Ithaca, 1960. D.R. Lide (Ed.), CRC Handbook of Chemistry and Physics, 90th Edition, CRC Press/Taylor and Francis, Boca Raton, 2010.

[53] R. Memming, Semiconductor Electrochemistry, J. Wiley-VCH, Weinheim, 2001. [54] N. Sato, Electrochemistry at Metal and Semiconductor Electrodes, Elsevier, Amsterdam, 1998. [55] B.C.-M. Lai, N.-H. Kung, J.Y.-M. Lee, A study on the capacitance–voltage characteristics of metal-Ta2O5-silicon capacitors for very large scale integration metal-oxide-semiconductor gate oxide applications, J. Appl. Phys. 85 (1999) 4087. [56] V.V. Afanas’ev, Electron Band Alignment at Interfaces of Semiconductors with Insulating Oxides: An Internal Photoemission Study, Adv. Condens. Matter Phys 2014 (2014) 1–30 301302. [57] F. Di Franco, P. Bocchetta, M. Santamaria, F. Di Quarto, Light induced electropolymerization of poly(3,4-ethylenedioxythiophene) on niobium oxide, Electrochim. Acta 56 (2010) 737.