You are on page 1of 12

Journal of New Materials for Electrochemical Systems 9, 221-232 (2006)

J. New. Mat. Electrochem. Systems

Behaviour of Titanium in Sulphuric Acid


Application to DSAs


D. Devilliers, M.T. Dinh, E. Mah, D. Krulic, N. Larabi and N. Fatouros

Universit Pierre et Marie CurieParis 6, Laboratoire LI2C-Electrochimie, CNRS UMR 7612


4 Place Jussieu - 75252 Paris Cedex 05 - France

Received: January 04, 2006, Accepted: March 02, 2006

Abstract: Titanium-based Dimensionally Stable Anodes (DSAs) are commonly used in industrial cells, even in corrosive solu-
tions. It is necessary to control the corrosion of their titanium substrates. In this paper, the behaviour of titanium electrodes in
sulphuric acid solutions is studied by means of potential decay curves at open circuit potential and impedance spectroscopy measure-
ments. In addition, an electroanalytical determination of soluble Ti(IV) species in the acid electrolytic solutions is proposed, giving
a convenient tool for detecting in situ the dissolution of TiO2 and the corrosion of Ti, which can lead to the deactivation of DSAs.

Keywords: Titanium, Dimensionally Stable Anodes, deactivation process, titanium corrosion, titanium electroanalysis

1. INTRODUCTION acid [4-7]. At room temperature, titanium is rapidly cor-


roded in concentrated acid solutions: H2 SO4 , HCl, H3 PO4
The great technological interest in titanium is due to its and some organic acids (oxalic acid) [8,9]. The corrosion
versatility, leading to many dierent applications in spa- rate depends on the nature of the acid, its concentration,
cial, aeronautic, naval and electrochemical industries. In- the temperature of the solution, and the presence of a pas-
deed, titanium is used as a substrate material for Dimen- sivating layer on the metal. It is reported that the cor-
sionally Stable Anodes (DSAs). They constitute particular rosion current intensity decreases if a thick oxide layer is
modified electrodes with remarkable electrochemical prop- present. The anodic transfer coecient also decreases with
erties [1].Before deposition of metallic mixed oxides on the layers of increasing thickness. In H2 SO4 , HCl and H3 PO4 ,
surface of the substrate, the latter must be etched in acid generalised corrosion occurs. Robin et al. have compared
medium; concentrated HCl or H2 SO4 are commonly used the behaviour of titanium and titanium alloys in sulphuric
for that purpose [2]. acid [10].
In order to be able to predict the service life of DSAs, the The metal dissolution is slow in oxidant-containing aque-
behaviour of titanium in concentrated acid solutions must ous solutions without complexing agents, due to the pres-
be studied, especially in sulphuric acid solutions which are ence of the native TiO2 film on the substrate [11-13]. The
commonly used in industry. As other valve metals such presence of oxidizing species such as oxygen and chromic
as tantalum, titanium is thermodynamically unstable in ions is reported to improve the performance of titanium
aqueous media and is oxidized by oxygen present in the sur- [14]. In hot concentrated HCl or H2 SO4 , general corro-
rounding environment. Electrogenerated oxide films may sion of titanium occurs, but, in that kind of environments,
also be formed on that metals [3]. The metal is covered the presence of oxidizing agents and certain multivalent
by a stable oxide film of TiO2 when the metal is exposed metal ions have the ability to passivate Ti. For that rea-
to air or anodised. Its corrosion resistance is excellent in son, the experiments must be carried out with high purity
many media, but its corrosion rate may be significant in H2 SO4 in order to get reproducible results [15]. Titanium
concentrated acid solutions. For example, the protecting is severely corroded in fluorhydric acid, due to the forma-
oxide film is dissolved gradually in concentrated sulfuric tion of TiF6 2 complexes [16].
Many authors have studied the corrosion of titanium by
To whom corresponding to: E-mail: devill@ccr.jussieu.fr, Phone: dierent techniques; most of them are rapidly reviewed
33-1 44 27 36 77, Fax: 33-1 44 27 38 56

221
222 D. Devilliers et al. / J. New Mat. Electrochem. Systems

below. At last, the determination of the amount of Ti(IV) room temperature is about 5nm. Blackwood et al. [25]
in solution by electroanalytical techniques is described. have studied the formation of anodic TiO2 films poten-
tiodynamically grown in 3 M H2 SO4 . After several hours
1.1. Open Circuit Potential (OCP) studies and po- at OCP in the same solution, partial dissolution has oc-
larisation curves curred; the amount of dissolved material was deduced from
As a titanium oxide layer is always formed in air, freshly the amount of electricity used during the re-growth of the
polished titanium dipped in a corrosive medium takes at film. It was found that the dissolution is uniform and slow,
least 24h to attain a reproducible value of the stationary but the stability of the film depends on its mode of prepara-
state current and OCP. The stable value expected i.e tion. Mller et al. [27] found that anodisation of titanium
0.66 V is obtained in strong acids after dissolution of the in acid solutions is accompanied by a potential indepen-
air-formed oxide layer [17]. dent chemical dissolution.
The evolution of the rest potential of titanium in acidic Frayret et al. [28,29] proposed a model to explain dis-
media has been studied by several authors [18-20]. Its solution and passivation of titanium. In nitric acid, the
value strongly depends on the concentration of the acid. OCP shifts to higher values for long exposure times, in re-
In concentrated solutions, titanium rapidly attains the ac- lation to thickening of the TiO2 film formed at the surface
tive state and corrosion occurs rapidly. of the electrode (oxidizing character of HNO3 ). The corro-
K. Azumi et al. [21] have studied the behaviour of tita- sion potential Ecor is stable when the rate of film formation
nium electrodes during long-term immersion experiments. equals the rate of film dissolution [10].
The corrosion resistance was evaluated by monitoring the Titanium is used in desalination plants in Middle East
OCP of the electrodes, some samples being previously po- countries (heat exchanger tubes) due to its excellent
larized at 10 V in order to form an anodic oxide film. A erosion-corrosion resistance in sea water. However, in these
potential shift in the positive direction usually means the plants, descaling is a common maintenance operation (re-
growth of the oxide film. Such a phenomenon was observed moving of CaCO3 and Mg(OH)2 ). This operation is car-
even in deaerated neutral solutions. ried out using an acid with or without inhibitors [30]. The
Thomas and Nobe [22] have studied the corrosion of Ti authors have studied the behaviour of titanium in 0.5%
in 0.5 M H2 SO4 . They have found that the OCP of the H2 SO4 solution (a sucient concentration for clearing Ti
metal was in the passive range (400-700 mV vs SCE). How- tubes in desalination plants). The corrosion rate at 50 C
ever, if the electrode was first pretreated in 1M HF, a rapid is very low: 0.0025 mm/y in aerated solution and 0.0020
decrease of the OCP was observed and the titanium disso- mm/y in deaerated-inhibited solution. Other authors have
lution occurred as the active state was reached (corrosion reported highest rates in more concentrated sulphuric acid:
potential: Ecor = -680 mV vs SCE). That rapid decrease 1.9 mm/y in 4% H2 SO4 at 50 C [31].
of the OCP is called self-activation. The same behaviour The corrosion current density is more important with
was observed by Brauer and Nann [23] who noticed that concentrated HNO3 solutions (about 2A/cm2 corre-
the time necessary for the appearance of self-activation was sponding to 25 m/year)
decreasing with increasing molar concentration of H2 SO4 . In H2 SO4 (20-80 % solutions), the OCP shifts towards
If the titanium electrode is passivated by anodic po- lower values.
larization before being in contact with the corrosive solu- The stabilized values are in the active region of the Pour-
tion of H2 SO4 , the self activation does not appear rapidly. baix diagram [10] (-680 to -750 mV/SCE) and the corrosion
Blackwood et al. [24,25] concluded that that phenomenon current density increases with acid concentration. (0.17
was related to the regular dissolution of the oxide film. It mA cm2 for 20% to 6.86 mA cm2 for 80%).
was shown that thinning of the oxide film in 3 M H2 SO4 at
OCP occurs by uniform dissolution at constant rate rather 1.2. EQCM
than by localized attacks [25]. These results were obtained Recently, a new technique for the study of corrosion of
by in situ ellipsometry. titanium in sulphuric acid has been proposed [32]. The ti-
J.Krysa et al. have studied the corrosion rate of tita- tanium foils were glued onto the crystal of an electrochem-
nium in 1-4 M sulphuric acid at temperatures of 25-60 C ical quartz crystal microbalance (EQCM); the dissolution
[19,26]. The OCP of titanium suddenly decreases; thus the of the active metal and the growth of the passive film at
metal is in the active state. The duration after which that higher anodic potentials were studied from the measure-
decrease is obtained depends strongly on the pre-treatment ment of mass changes. Experiments were made in 5M
of the sample: for a given concentration of H2 SO4 an elec- H2 SO4 . The air-formed oxide layer was removed by ap-
trode pre-treated in HF gives rise to an immediate sharp plying a cathodic constant potential (-500 mV). Thus, an
decrease; electrodes without any pretreatment give rise to active OCP of -673 5 mV was obtained and dissolution of
a long potential plateau before the sharp decrease of po- the metal occurred, at a rate of 0.8 g min1 cm2 . When
tential. applying increasing potentials, the anodic current, related
The rest potential of Ti in 10 N H2 SO4 is reported to be to dissolution of Ti, gave a decrease in mass in the EQCM
-1.14V vs Hg, Hg2 SO4 , 10 N H2 SO4 reference electrode, experiments. After the passivation potential Ep (-470mV
i.e. in the active corrosion region of the metal [15]. vs SCE), the current decreased (oxide formation); the mass
The thickness of protecting oxide films formed in air at continued to decrease in the active-passive region because
Behaviour of Titanium in Sulphuric Acid
-Application to DSAs- / J. New Mat. Electrochem. Systems 223

the passive layer was not intact. Above Ep , Ti4+ started was treated at 220 C in an autoclave in presence of wa-
to form, in addition to Ti3+ . ter. The authors have studied the corrosion resistance of
In 0.1 M H2 SO4 solutions, the oxygen concentration was their samples in simulated body fluid by impedance spec-
sucient to keep the surface passive. The OCP measured troscopy. For oxidized samples, an (RQ)Q equivalent cir-
was in the range -50 to 150 mV/ SCE, due to dierent cuit is correct; however, for the oxidized and hydrother-
oxide layer thicknesses [32]. mally heated samples, an R(RQ)(RQ) circuit is necessary
because the Nyquist diagrams exhibit two loops. For the
1.3. Impedance spectroscopy oxidized samples, after several hours of exposure in the
Azumi and Seo have studied the impedance response of corrosive medium, the parameter n of the CPE in-
anodic TiO2 films during a potential sweep in neutral so- creases, suggesting that the layers become tighter. The
lutions [33]. Impedance was measured at a fixed frequency hydrothermal treatment leads to a modification of the ox-
of 10 Hz in order to determine the capacity of the interface, ide film: it is supposed to be composed of a compact inner
related to the thickness of the film and to its composition. layer and a porous outer layer. During exposure in the
Usually, the authors prefer to plot the impedance dia- corrosive medium, the inner layer is rebuilt (the values of
grams in a large frequency domain. According to dierent R and n decrease) whereas for the outer layer R and n
authors, the Nyquist diagrams of a Ti electrode at the rest increase, suggesting that the outer layer becomes tighter.
potential exhibit one or two capacitive loops. Nevertheless, the real surface of the samples is not known
The impedance diagram at the rest potential exhibits and it is dicult to deduce precise information from im-
one semicircle at low frequencies: the maximum of the pedance measurements. The authors suggested that more
imaginary part Z is obtained for f=2.5 Hz. No semicircle precise results should be obtained by measuring the con-
is obtained at high frequencies. The corrosion resistance is centration of titanium ions in the corrosive medium after
estimated to be ca. 100 cm2 [15]. a long time of exposure.
The Surface Charge Approach (SCA) may be used for
giving interpretation of the impedance spectra. In that 1.4. Reactions
model, a capacitive behaviour is observed at high fre- Among the studies reported above, some of them have
quency. That loop is the parallel combination of the barrier shown the evidence of coupling between film growth and
film capacitance Cb and the resistance of charge carriers metal dissolution in the passivation process of titanium.
migration Rb . The thickness of the film varies linearly with That phenomenon is also observed for other metals such
the applied potential E; consequently, 1/Cb varies linearly as Nb and Mo [34]. One must take into account several
with E [34]. Below the passivation potential, the loop is at- reactions, including dissolution of the metal oxide and ox-
tributed to the double layer capacitance and charge trans- idation of the metal.
fer resistance. The anodic dissolution of titanium in sulphuric acid
For other authors, the impedance diagrams at OCP ex- leads to Ti3+ ions, according to Armstrong et al. [37].
hibit two capacitive loops. The first loop (CL , RL ) in the In sulphuric acid, the dissolution of the TiO2 layer leads
low frequency region is attributed to the space charge layer to colourless Ti(IV), but when the bared metal is reached,
formed in the oxide, and to the charge transfer resistance the oxidation of the Ti substrate leads to Ti(III), as violet
through the oxide film [21]. The second loop (CH , RH ), Ti3+ ions, which can be further oxidised into Ti(IV) by
observed in the high frequency region is attributed to ab- dissolved oxygen.
sorption / desorption of solution species [21], although such The following reactions occur:
phenomena are usually observed at low frequencies. Dissolution of the TiO2 films in acid media:
RL and CL 1 increased with immersion time, corre-
sponding to the thickening of the oxide in the neutral so- TiO2 +4H+ Ti4+ +2H2 O
lution at 298 K [21].
Colorless soluble Ti(IV) species are formed.
In 2 M sulphuric acid, the Nyquist plot of a titanium
When the protecting TiO2 layer is completely removed
electrode recorded at the OCP exhibits two capacitive
from some parts of the electrode, titanium is in the active
loops [35]; the high frequency loop is attributed to charge
state and Ti3+ ions are formed:
transfer; the diameter of the first semi-circle gives the value
of the charge transfer resistance, Rct . Ti+3H+ Ti3+ +3Hads
Evaluation of the corrosion resistance of several tita-
nium materials is performed, comparing the values of Rct : Evolution of hydrogen occurs.
Higher values indicate a higher corrosion resistance [35]. In aerated solutions, Ti3+ is oxidised:
Baszkiewicz et al. [36] have improved the corrosion resis-
1
tance of titanium with the modification of the surface by 2H+ +2Ti3+ + O2 2Ti4+ +H2 O
2
plasma electrolytic oxidation and subsequent hydrother-
mal treatment. As a very large voltage was applied to 1.5. Determination of the concentration of Ti(IV)
the electrode (400 V), the breakdown of the native TiO2 At the mercury drop electrode, the reduction of titanic
films occurred, accompanied by sparks. Thus a porous ox- salts in acid solutions proceeds to the titanous state. The
ide coating was formed. After that operation, the sample kinetics of the reduction of Ti(IV) in 0.4-9 M H2 SO4 was
224 D. Devilliers et al. / J. New Mat. Electrochem. Systems

systematically investigated. Experiments in dierential cm2 ) until the final potential was 10 V vs SCE. With such
pulse voltammetry (DPV) show a sequence of electrochem- conditions, the thickness of the oxide layer (35 nm) is very
ical reactions which occur in succession or even simultane- reproducible [43]. For verifying the reproducibility, up to
ously [38]. 5 samples were tested for one kind of experiments.
In DC polarography, whatever the concentration of
H2 SO4 , the limiting current is diusion-controlled and di- 2.2. Electrochemical devices
rectly proportional to Ti(IV) concentration [38-40]. The For long-time experiments (measurement of OCP), sev-
determination of the diusion coecient of Ti(IV) species eral sealed electrochemical cells were built, in order to
at 25 C was performed as a function of H2 SO4 concentra- avoid evaporation. In these cells, a reference electrode was
tion up to 9 M. From the limiting current, concentrations immersed periodically for the measurement of the OCP of
up to about 104 M can then be directly determined [38]. the working electrode. For short time experiments (mea-
In H2 SO4 with C > 3.5 M, a well-defined peak appears in surement of o.c.p. in 5 M H2 SO4 ), the reference electrode
DPV. The peak current is directly proportional to the ana- was immersed continuously in the electrolytic solution.
lytical concentration of Ti(IV) and depends on the H2 SO4 For long-time experiments in sulphuric acid solutions,
concentration. For practical analytical purposes, the ra- the best choice of reference electrode is a mercury sulphate
tio of the peak current density to Ti(IV) concentration, electrode in order to prevent the diusion of Cl ions from
Ip /S[Ti(IV)], where S is the electrode area, was tabulated SCE electrodes.
in the concentration range from 3 to 9 M in H2 SO4 .[41]. However, the use of commercial sulphate electrodes
As was said above, titanium is commonly used as a sub- Hg/Hg2 SO4 /K2 SO4 is no more correct because K+ cations
strate material for DSAs in concentrated acid solution. may diuse from the reference electrode compartment to
In addition, recently, Ti/TiO2 /PbO2 electrodes have been the electrochemical cell and H+ cations from the electro-
proposed for the oxidation of Cr(III) to Cr(VI) in sulphuric chemical cell to the reference electrode compartment; these
acid solutions [42]. It is important to evaluate the corrosion variations of composition of the filling solution and the
resistance of that material in order to predict the service electrolyte of the cell may alter the results of the exper-
life of industrial electrodes. iments. For that reason we have prepared two reference
In this paper, the behaviour of titanium is studied in electrodes without liquid junction potential and adapted
sulphuric acid solutions by the mean of potential decay to our media:
curves at open circuit, and impedance measurements. The
i. Hg/Hg2 SO4 /0.5 M H2 SO4 electrode for experiments
influence of the purity of the metal and the presence of a
performed with 0.5 M H2 SO4 solutions in the cell.
TiO2 film, anodically built, are evidenced.
It is denoted ESAM (for Electrode with Sulphuric
In addition, the determination of the amount of Ti(IV)
Acid). Its potential is 0.669 V vs SHE.
in solution was performed by a pulse voltammetric tech-
nique, allowing to attain the direct quantitative evaluation ii. Hg/Hg2 SO4 /5 M H2 SO4 electrode for experiments in
of the dissolution of the protecting TiO2 films and the cor- the cell containing 5 M H2 SO4 ; it is denoted ESA5M.
rosion of the metal. Its potential is 0.564 V vs SHE.

2. EXPERIMENTAL All the potentials mentioned in this paper are given with
respect to those two electrodes.
2.1. Materials The auxiliary electrodes were platinum wires.
Pure titanium disks (diameter: 10 mm, Goodfellow) An Autolab PGSTAT 30 potentiostat was used for the
were used as working electrodes. Two kinds of purity were polarization resistance determination. The polarization
tested: (i) catalogue number 410; purity = 99.6 %; main curves were plotted, applying a slow potential scan (v =
impurities (in ppm): Al = 300; Fe = 1500; O = 2000; Sn 1 mV s1 ) at the working electrode, in the vicinity of the
= 200; (ii) catalogue number 432; purity = 99.99 %; main corrosion potential Ecor (Ecor 20mV).
impurities (in ppm): Al = 1.3; Cr = 1.3; Fe = 3.65; Ta < Some impedance measurements were carried out using a
5; V = 0.38; Zr = 0.76. frequency response analyzer SOLARTRON 1255 coupled
All the samples were mechanically polished with to a EG&G 273 A potentiostat; other experiments were
Buehler-Met grinder paper up to grid size 1200, cleaned performed with the Autolab PGSTAT 30 equipped with
in methanol in an ultrasonic bath, and rinsed with dis- the FRA2 frequency response analyzer module. The am-
tilled water. This procedure results in the formation of a plitude of the applied sinusoidal signal was 10mV; the im-
native oxide layer on the titanium substrates. Its thickness pedance experiments were performed in the range 100 kHz
and composition have been evaluated [43]. 2.6 mHz. Fitting and simulations were performed with
In addition, some samples were electrochemically pol- the Equivalent Circuit software (Boukamp) or the FRA2
ished in non aqueous solvent, according to a procedure Autolab software.
described elsewhere [44]. The disks were mounted on a
special holder, adapted on a Tacussel rotating disk elec-
trode system. Thus, they were anodised in a 0.5 M H2 SO4
solution according to a galvanostatic procedure (j = 5 mA
Behaviour of Titanium in Sulphuric Acid
-Application to DSAs- / J. New Mat. Electrochem. Systems 225

Table 1. Variation of the OCP and values of the impedance parameters for a non-anodised titanium electrode immersed
in 0.5 M H2 SO4 .
Immersion time (h) Er (mV/ESAM) Rs (cm2 ) Rox (kcm2 ) Ctot (Fcm2 ) n
0.25 378 3.8 3950 21.7 0.966
2 411 3.8 2210 24.1 0.967
3 454 3.8 1750 26.2 0.967
5 510 6.2 1180 30.4 0.963
72 514 3.1 856 55.5 0.965
168 518 3.0 511 65.9 0.968
336 521 3.3 498 70.9 0.966
504 524 3.1 434 78.4 0.972
672 530 3.4 413 89.8 0.973
1008 541 3.6 331 102 0.974
1344 545 3.2 276 105.4 0.974
2352 550 3.0 29 113 0.94

-100

-200

-300

-400

-500

-600
0 500 1000 1500 2000 2500

Time (h)

Figure 1. Variation of the OCP of an anodised titanium


electrode (purity 99.6 %) in 0.5 M H2 SO4 . Initial thickness
Figure 2. Nyquist diagram for an anodised titanium elec-
of the anodic oxide layer: 35 nm.
trode at the OCP (E = -520 mV/ESAM) after t = 624 h
of immersion in 0.5 M H2 SO4 .

3. RESULT AND DISCUSSION


3.1. Behaviour of titanium in 0.5 M H2 SO4 gle of that branch, relative to the imaginary axis is linked
For this section, only 99.6 % purity titanium has been to the dimensionless parameter, n, of the constant phase
tested. element (CPE), representing the behaviour of the inter-
face. For all these experiments, the equivalent circuit is
3.1.1. Anodised titanium only composed of a resistance Rs (resistance of the elec-
The OCP of the samples decreases sharply from -190 to trolyte) in series with a CPE. As the value of n lies between
-480 mV vs ESAM during the first twenty four hours of 0.94 and 0.97 during the first 800 hours of immersion, the
immersion. After that period, the variation is not so im- CPE is almost a pure capacitance. Its value continually in-
portant. After 2200 hours, the values of the rest potential creases from 6.7 F cm2 after 0.25 hour, to 51 F cm2
is about -520 mV and does not correspond to a corrosion after 768 hours. For very long durations, the value of n
situation (see Figure 1) decreases to 0.9. This capacitive term is composed of two
The impedance diagram, plotted at the OCP in the terms in series: the double layer capacitance and the ca-
Nyquist representation, exhibits only one infinite branch, pacitance of the protecting oxide layer.
even after t = 624 h of immersion (see Figure 2). The an- The increase of C is related to the decrease of the thick-
226 D. Devilliers et al. / J. New Mat. Electrochem. Systems

Table 2. Variation of the OCP and values of the impedance parameters for a anodised titanium electrode (purity 99.6
%) immersed in 5 M H2 SO4 .
Immersion time (h) Er (mV/ESA5M) Rs (cm2 ) Rox (kcm2 ) Cox (Fcm-2) n1 Rtc (cm2 ) Cdl (Fcm2 ) n2
0 40 8.92 - 5.33 0.96 - - -
24 230 5.66 - 11.3 0.97 - - -
48 360 6.32 - 17.8 0.97 - - -
120 982 6.61 197 258 0.90 - - -
144 935 7.36 65 509 0.92 36.2 0.18 1.0
168 947 5.76 46 616 0.92 26.8 0.244 1.0

Figure 3. Evolution of the Nyquist diagrams for a non-


anodised titanium electrode in 0.5 M H2 SO4 at the OCP, Figure 4. Evolution of the Nyquist diagrams for an an-
for three durations of immersion. odised titanium (purity 99.6 %) electrode at the OCP in
5 M H2 SO4 . The OCP and the duration of immersion are
indicated in the inset.

ness of the protecting TiO2 layer. The interface does not


behave as a pure capacitance for long time experiments;
this phenomenon could be attributed to the increase of 1.
roughness. These experiments show that the oxide layer (anodic ox-
ide or native oxide) initially present on the titanium sub-
3.1.2. Non-anodised titanium strate is not completely removed, even after 2350 h of im-
For these experiments, the samples are only covered by a mersion i.e. corrosion is not observed, although the thick-
native oxide layer after wet polishing with abrasive paper. ness of that protecting layer decreases with time.
Nevertheless, as for the anodised samples the OCP values The situation is dierent in concentrated sulphuric acid
continuously decrease and a value of -550mV vs ESAM is solutions, as reported below.
attained after 2350 hours of immersion.
The impedance diagrams do not exhibit one infinite 3.2. Behaviour of titanium in 5 M H2 SO4
branch, but an arc of a semi-circle (see Figure 3). The 3.2.1. Anodised titanium
equivalent circuit is now Rs (Ct , Rox ).Where Rs is the re- In this section, two kinds of samples were studied, with
sistance of the electrolyte, Rox the resistance of the oxide dierent purity: 99.6 or 99.99 %.
layer, and Ct the capacitance of the oxide layer). A CPE For the 99.6 % sample, the OCP decreases rapidly when
may be used instead of a pure capacitance, for a better fit- it is immersed in 5 M H2 SO4 . After 120 hours, Er = -982
ting of the curve. Thus, a value of n0.97 is obtained, even mV vs ESA5M. This very low value indicates that titanium
for very long time experiments (t = 1350 h). As previously, is in the active state and that corrosion occurs. The initial
the value of Ct increases for long duration of immersion, yellow colour of TiO2 disappears rapidly; in addition, the
although Rox decreases. All the data are reported in Table solution becomes violet, indicating the presence of Ti3+
Behaviour of Titanium in Sulphuric Acid
-Application to DSAs- / J. New Mat. Electrochem. Systems 227

Figure 5. SEM image of a titanium sample (purity 99.6 Figure 6. Details of a corroded grain of the same titanium
%) electropolished and anodised, after 120 h of immersion sample.
in 5 M H2 SO4 . General view, showing the grains at the
surface

ions. value of n is near 0.97, i.e. the capacitive term is almost


The evolution of the Nyquist diagram obtained with ideal. After 120 h of immersion, the first loop (high fre-
those samples is presented in Figure 4. At the beginning quency loop) leads to the value of the capacitive term Cox
of the experiment (t = 0; Er = 40 mV vs ESA5M) the dia- which does not behave as an ideal term (the value of n is
gram is limited to one infinite branch, as in the case of the 0.9 or 0.92).
0.5 M H2 SO4 solution (see previous section). However, for The charge transfer resistance is extracted from the sec-
more than 100 h of immersion, the diagrams exhibit one ond loop (low frequency loop); it is estimated to be Rct
capacitive loop, which diameter decreases. After 144 h, a 30 cm2 , while a very large capacitive term is obtained,
second capacitive loop is also present. which is usually attributed to the double layer capacitance.
The equivalent circuit passes progressively from the sim- However, the evaluation of these two terms is not easy be-
ple one Rs ,Cox , to Rs (Cox Rox ) and at least Rs (Cox Rox ) cause the area of the sample is strongly modified during
(Cdl Rct ). The low-frequency loop is attributed to the the corrosion process. As can be seen with a microscope,
contribution of the double layer capacitance Cdl in paral- the roughness is very important after long time immer-
lel with the charge transfer resistance Rct due to corro- sion experiments. The surface of the sample after 120 h
sion of the metal. The same equivalent circuit is proposed of immersion in 5 M H2 SO4 is presented in Figure 5. On
by other authors [21,36]. However, Baszkiewicz et al. do this SEM micrograph, it can be seen that the corrosion
not present the Nyquist diagrams; they only give a table rate is not uniform on all the grains of the samples. This
of data. These data indicate that the time constants of observation has two important consequences:
the two loops are not very dierent; in such case, the two
loops are probably not clearly distinguished on the dia- i. in the impedance diagrams, two loops are observed.
grams; in addition, very low values of the dimensionless One of them (the high frequency loop) may be at-
parameter, n, of the CPE are sometimes obtained by the tributed to the part of the electrode which is not
authors (0.68). In our case, the lowest value of n was only attacked so strongly.
0.9 (see Table 2). The non-ideal capacitive behaviour in
the experiments of Baszkiewicz et al. is probably due to ii. the increase of the capacitive terms is related to the
the dierent method of preparation of their samples. In bare part of the electrode, characterised by a large
our experiments, electropolishing followed by careful an- area, due to a considerable increase of the roughness
odisation leads to compact oxide layers. (see Figure 6).
The values of Rox are not given in Table 2 for the three
first experiments because the equivalent circuit is just com- From these experiments, it is concluded that corrosion of
posed of a resistance, Rs , and Cox in series. the titanium samples occurs relatively rapidly. As shown
The value of the capacitance of the interface increases below, the corrosion rate may decrease considerably if a
as previously observed with 0.5 M H2 SO4 solutions. The high purity metal is used instead of a 99.6 % sample.
228 D. Devilliers et al. / J. New Mat. Electrochem. Systems

-500

E0 (mV/ESA5M)
-1000

-1500
0 20 40 60 80 100
Immersion time (h)

Figure 9. Variation of the OCP of a non-anodised titanium


Figure 7. Variation of the OCP of an anodised titanium electrode (purity 99.6 %) in 5 M H2 SO4 .
electrode (purity 99.99 %) in 5 M H2 SO4 . Initial thickness
of the anodic oxide layer: 35 nm.

Figure 10. Evolution of the Nyquist diagrams for a non-


Figure 8. Nyquist diagram for an anodised titanium (pu- anodised titanium electrode (purity 99.6 %) in 5 M H2 SO4 ,
rity 99.99 %) electrode at the OCP (E = -300 mV vs MSE at the OCP. Short durations of immersion.
5M) after t = 5280 h of immersion in 5 M H2 SO4 .
Behaviour of Titanium in Sulphuric Acid
-Application to DSAs- / J. New Mat. Electrochem. Systems 229

Ti(III) as Ti2 (SO4 )3 , according to the literature [45]. If


that solution is in contact with air, the violet colour van-
ishes, due to the oxidation of Ti(III) into Ti(IV) species
by oxygen.
The impedance diagrams are presented in Figure 10 for
short time of immersion, only one capacitive loop is ob-
served. Its diameter decreases rapidly. When corrosion
arises, a second capacitive loop appears (Figure 11). The
equivalent circuit has already been described in the previ-
ous section. All the data are reported in Table 3.
It can be thought that during the corrosion process, the
high frequency loop (Rox Cox ) may be attributed to the
contribution of the part of the electrode which is still cov-
ered by a thin layer of oxide, or by insoluble corrosion
products. We have also measured the polarization resis-
tance of the titanium sample in 5 M H2 SO4 after 5 hours
of immersion i.e. when the metal is still covered by a pro-
Figure 11. Evolution of the Nyquist diagrams for a non- tecting oxide layer. The value of Rp was found to be 4.5
anodised titanium electrode (purity 99.6 %) in 5 M H2 SO4 , 105 cm2 . This value is very similar to the value of
at the OCP. Long durations of immersion. Rox extracted from the impedance diagrams (see Table 3).
After 70 h of immersion, the Rp value is 80 cm2 ; it is the
same order of magnitude that the value extracted from the
impedance diagram: Rox + Rct 110 cm2 . This small
Another set of experiments was performed with high value of Rp means that titanium is strongly corroded: jcor
purity titanium(99.99 %). With an anodised sample, the = 2A cm2 , corresponding to vcor = 0.7 mm/y.
OCP tends to a constant value of -300 mV vs ESA5M, It must be noticed that the determination of Rp from po-
instead of -960 mV as seen previously (see Figure 7). larisation curve at the vicinity of Er is not accurate if that
After 400 h, we have observed the sample by AFM and by value is not constant. That determination is only correct
optical microscopy. No alteration of the surface was visible. when Er remains almost constant, i.e. for the experiment
Impedance diagrams were plotted after a very long time after 70 h of immersion in our case.
of immersion (5280 h, corresponding to 7 months). The The measurement of the OCP allows to know if titanium
Nyquist diagram, presented in Figure 8, indicates that the is in the active state or not; the impedance diagrams also
sample is not yet corroded. give information on the behaviour of the protecting oxide
From these experiments, we deduce that the rate of cor- layer present on the surface of the electrodes. However, it
rosion strongly depends on the purity of the metal. How- is useful to have an independent method of investigation in
ever, the cost of high purity titanium is too high and that order to know if there is a permanent competition between
quality of metal cannot be used in industrial Electrochem- formation and dissolution of titanium oxide.
istry for preparing DSAs. For that reason, the behaviour For that reason, we have coupled our experiments with
of non- anodised samples in 5 M H2 SO4 hereafter was only an electroanalytical method which allows the in situ deter-
performed with 99.6 % purity samples. mination of the Ti(IV) content in the H2 SO4 solution.
The DPV technique has been carried out for the direct
3.2.2. Nonanodised titanium determination of soluble Ti(IV) in sulphuric acid. First,
These experiments were performed only with 99.6 % pu- we give hereafter the characteristic results obtained with 3
rity samples, after wet polishing with abrasive paper. As samples:
it was expected, the OCP decreases rapidly from its initial
value (-364 mV vs ESA5M) to about 1200 mV after 14 Sample 1: High purity titanium (99.99 %), electropo-
hours of immersion (see Figure 9). During these short lished, and anodised immersed in 5 M H2 SO4 during
time experiments, the potential of the working electrode 8760 h. At the end of immersion time, the rest po-
was measured continuously vs a ESA5M reference elec- tential was -191 mV vs ESA5M.
trode. The evolution of hydrogen bubbles was visible at
the end of the experiments. Sample 2: Titanium (purity 99.6 %) electropol-
After the minimum at -1200 mV, the OCP increases very ished and anodised, after 6 days of immersion in 5
slowly and tends to -1100 mV after 100 hours of immer- M H2 SO4 . The rest potential was: -1023 mV vs
sion. This value is in perfect agreement with the rest po- ESA5M.
tential of titanium in 10 N H2 SO4 (-1.14 V vs Hg/Hg2 SO4 ,
10 N H2 SO4 reference electrode) found by other authors Sample 3: Titanium (purity 99.6 %) mechanically
[15]. polished; non anodised, immersed 14 h in 5 M
The solution becomes violet, due to the formation of H2 SO4 . Its rest potential was -1027 mV vs ESA5M.
230 D. Devilliers et al. / J. New Mat. Electrochem. Systems

Table 3. Variation of the OCP and values of the impedance parameters for a non-anodised titanium electrode (purity
99.6 %) immersed in 5 M H2 SO4 .
Immersion time (h) Er (mV/ESA5M) Rs (cm2 ) Rox (kcm2 ) Cox (Fcm2 ) n1 Rtc (cm2 ) Cdl (Fcm2 ) n2
0.25 200 11.21 3660 16.7 0.96 - - -
3 360 11.96 1460 24.3 0.96 - - -
5 490 11.42 427 30.7 0.94 - - -
7 553 11.13 131 37.8 0.93 - - -
23 1110 9.77 0.156 86.2 0.92 92 0.05 0.88
70 1006 10.85 0.063 160 0.92 45 0.04 0.8

Table 4. Determination of the soluble Ti(IV) species in concentrated sulphuric acid


after several hours of immersion, by DPV.
Experimental conditions: E = 40 mV; t = 40ms; S=2 mm2
In column 5, S is expressed in mm2 and Ip in mA.
Immersion Rest potential I p Ip
Sample [H2 SO4 ]/M S[Ti(IV)] [Ti(IV)]/M
Time/h (mV vs MSE5M) /A
1 5.95 8760 191 0.85 2.9 1.7 103
2 5.3 144 1023 0.68 9.1 6.7 103
3 5.05 72 1027 0.61 12.2 1.0 102

The results obtained with these three samples are given protecting TiO2 layer is scarcely dissolved in 0.5 M H2 SO4 .
in Table 4. For each determination, it was necessary to These results are in agreement with those of J. Krysa et al.
check the actual concentration of sulphuric acid because, [26] who found that the self-activation time (time necessary
for long-time experiments, some evaporation of water may for obtaining the steady state corrosion rate) depends on
occur. the thickness of the TiO2 layer present on the electrode
Another result will also be discussed below. It was ob- before its immersion in 4 M H2 SO4 . These authors give an
tained with a diluted H2 SO4 solution (0.5 M). Experiments expression of the steady state corrosion rate of the metal,
were performed in 0.5 M H2 SO4 in the presence of 0.2 M rcor : rcor increases linearly with the cube of the H2 SO4 con-
H2 C2 O4 . In this medium, a well-defined diusion peak is centration. Our results give an additional information: the
obtained in DPV. From this peak, Ti(IV) concentrations rate of dissolution of the TiO2 layer initially present on the
about 106 M can be accurately determined [46]. electrodes also increases with the H2 SO4 concentration.
- Sample 4: Titanium (purity 99.6 %) electropolished Low purity titanium (99.6%) is strongly corroded in
and anodised, immersed in 0.5 M H2 SO4 during 4080 h. 5 M H2 SO4 and the active state may be reached rapidly,
The rest potential was -410 mV vs ESAM. The Ti(IV) especially if the metal is not anodised; on the contrary,
concentration was found to be: 2.2 105 M. the active state was not reached with high purity tita-
nium (99.99 %) even after 7 months of immersion in 5
M H2 SO4 . However, it does not mean that the dissolution
3.3. Discussion
of the TiO2 protecting layer did not occur for that ma-
In concentrated sulphuric acid, the high value of the
terial, since Ti(IV) soluble species were evidenced in the
rest potential of the ultra pure titanium sample (sample 1)
resulting solution. It only means that the dissolution rate
indicates that it is not in the active state, even after a very
is very slow, so that the active state is not reached even
long duration of immersion. This means that a protecting
after 7 months. Indeed, when a Ti sample is immersed in
TiO2 film is still present on the electrode. However, the
sulphuric acid, two competing reactions occur: formation
large amount of Ti(IV) species present in solution mean
of TiO2 (oxidation of titanium by water or by dissolved
that dissolution of the protecting TiO2 film did occur dur-
oxygen if the solution is not deaerated) and dissolution of
ing the immersion, but not enough rapidly for giving a
TiO2 , leading to Ti(IV) soluble species.
completely bared metal.
Anodisation of a titanium electrode delays the appear-
The two low purity samples (2 and 3) were both in
ance of the active state (and thus corrosion of the sample);
the active state when the immersion experiments were
however for applications to DSA, it is not recommended
stopped. It is thus correct to find large amounts of Ti(IV)
to build a thick TiO2 before modifying the surface by a
in solution.
RuO2 -TiO2 electrocatalytic layer since thick layers lead to
Low amounts of Ti(IV) were found for the titanium elec-
the passivation of the electrodes. Another solution may
trode immersed in 0.5 M H2 SO4 (Sample 4). Even after
consist in choosing high purity titanium substrates; how-
4080 h of immersion, the active state was not reached. The
Behaviour of Titanium in Sulphuric Acid
-Application to DSAs- / J. New Mat. Electrochem. Systems 231

is not in direct contact with the corrosive medium if the


PbO2 is suciently compact.

4. CONCLUSION
In this paper, confident values of OCP were obtained be-
cause reliable reference electrodes without junction were
used for the experiments. These results allow to under-
stand the deactivation process of DSAs. Indeed, it is
sometimes observed that the electrocatalytic layer onto
the surface of the titanium substrate progressively disap-
pear, falling down at the bottom of the electrolysis cell.
In acid solutions, TiO2 is dissolved at the interface sur-
face/electrolyte. The electrocatalytic XO2 particles (X=
Ru or Ir) are progressively isolated and are no more firmly
attached to the surface of the electrode. That phenom-
enon occurs, even if the metallic substrate is never in di-
rect contact with the acid medium because titanium oxide
is progressively reformed from the reaction of Ti with water
Figure 12. Schematic deactivation of a DSA in acid (passivation reaction).
medium. The electrocatalytic XO2 particle (X = Ir or It is not always possible to evidence the dissolution of
Ru) is progressively isolated and is finally detached from TiO2 from usual electrochemical methods (polarisation re-
the electrode surface. sistance determination; impedance measurements) because
that reaction is counter-balanced by the passivation of the
electrode (formation of a TiO2 layer).
ever, their cost is too high. Our results show that DSAs When the dissolution rate is greater than the forma-
operating in concentrated acid solutions may suer from tion rate in sulphuric acid, some parts of the electrode are
TiO2 dissolution continuously. Of course, when the elec- naked. This situation is easy to detect by dierent tech-
trodes are polarized anodically, the TiO2 protecting film is niques:
rebuilt permanently, but the dissolution of TiO2 may in-
duce a detachment of the XO2 electrocatalytic oxide par- 1. from the OCP curves: sharp decrease of the poten-
ticles (IrO2 or RuO2 ), and contribute to the loss of per- tial, leading to very negative values.
formance. The composition of the layer is about 30 % of
precious metal oxide and 70 % of TiO2 . The presence of 2. from impedance diagrams: decrease of the real part
TiO2 is necessary for stabilization of the layer. Among of impedance, especially in the high frequency loop.
the dierent deactivation mechanisms of DSAs, one finds:
3. from the measurement of the polarisation resistance
substrate passivation, coating consumption or detachment,
Rp : its value is very small when the electrode is no
and mechanical damages [47,48]. The macroscopic deacti-
more covered by a TiO2 layer.
vation of electrodes by formation of a passivating layer may
be understood by a phenomenological model [49], which al- However, if the concentration of the acid is not very im-
lows to understand the hysteresis in the j-U characteristics; portant (i.e. 0.5 M H2 SO4 ), the dissolution rate is smaller
that model may probably be applied to the deactivation of than the passivation rate; thus, the metal is never naked.
DSAs resulting from the growth of a passive TiO2 layer Ageing in diluted sulphuric acid solutions do not lead to
under the electrocatalytic layer. dramatic variations of the OCP, the impedance or the Rp
Here, we consider the second cause of deactivation: coat- values.
ing consumption or detachment. In concentrated sulphuric The dissolution of TiO2 may lead to the deactivation of
acid, the dissolution of TiO2 is a purely chemical reaction, DSAs; however, the phenomenon is not easy to prove by
which occurs independently of the applied potential. When classical electrochemical techniques, except by the deter-
a DSA is used in such a medium, TiO2 will be dissolved; mination of the concentration of Ti(IV) in the solution,
the anodic potential applied to the electrode allows TiO2 for example by the electroanalytical technique described
to be formed again on the surface, but this phenomenon above. It is a convenient tool for detecting in situ the dis-
may probably be responsible for the detachment of elec- solution of TiO2 and the corrosion of Ti, which can lead
trocatalytic particles from the surface of the electrode, as to the deactivation of DSAs.
shown in Figure 12. Consequently, industrial DSA must
not be used in too much acidic media. On the contrary, REFERENCES
low cost DSAs Ti/TiO2 /PbO2 electrodes, which have
been proposed for the oxidation of Cr(III) into Cr(VI) in [1] D. Devilliers, E. Mah in Trends in Electrochemistry
concentrated sulphuric acid [42] may be used because TiO2 research, Ed., M. Nunez, Nova Science Publishers
232 D. Devilliers et al. / J. New Mat. Electrochem. Systems

Inc., New York, USA (ISBN: 1-59454-457-3), under sion, 41, 656 (1985).
press. [30] I.N. Andijani, S. Ahmad, A.U. Malik, Desalination,
[2] G.P. Vercesi, J. Rolewicz, C. Comninellis, Ther- 129, 45 (2000).
mochim. Acta, 176, 31 (1991). [31] H. Satoh, Desalination, 97, 45 (1994).
[3] O. Kerrec, D. Devilliers, H. Groult, M. Chemla, Elec- [32] M. Herranen, J.-O. Carlsson, Corros. Sci., 43, 365
trochim. Acta, 40, 719 (1995). (2001).
[4] T. Ohtsuka, M. Masuda, N. Sato, J. Electrochem. Soc., [33] K. Azumi, M. Seo, Corros. Sci., 43, 533 (2001).
132, 787 (1985). [34] M. Bojinov, S. Cattarin, M. Musiani, B. Tribollet,
[5] D.J. Blackwood, L.M. Peter, Electrochim. Acta, 34, Electrochim. Acta, 48, 4107 (2003).
1505 (1989). [35] F. Contu, B. Elsener, H. Bhni, Corros. Sci., 46, 2241
[6] D.J. Blackwood, L.M. Peter, D.E. Williams, Elec- (2004).
trochim. Acta, 33, 1143 (1988). [36] J. Baszkiewicz, D. Krupa, J. Mizera, J.W. Sobczak, A.
[7] D.J. Blackwood, R. Greef, L.M. Peter, Electrochim. Bilinski, Vacuum, 78, 143 (2005).
Acta, 34, 875 (1989). [37] R.D. Armstrong, J.A. Harrison, H.R. Thirsk, R. Whit-
[8] R. Schutz, D.E. Thomas, in Metals Handbook, Ed., field, J. Electrochem. Soc., 117, 1003 (1970).
J.R. Davis, 9th ed., Vol. 13, ASM, Metals Park, OH, [38] N. Fatouros, D. Krulic, N. Larabi, J. Electroanal.
1987, p. 669. Chem., 568, 55 (2004).
[9] B.D. Craig, D.S. Anderson, Handbook of Corrosion [39] J.J. Lingane, J. H. Kennedy, Anal. Chim. Acta 15, 294
Data, ASM International, Materials Park, OH, 1995. (1956).
[10] A. Robin, J.L. Rosa, H.R.Z. Sandim, J. Appl. Elec- [40] R.L. Pecsok, J. Am. Chem. Soc., 73, 1304 (1951).
trochem., 31, 455 (2001). [41] N. Larabi, Thse de lUniversit Pierre et Marie Curie,
[11] R.D. Armstrong, J.A. Harrison, H.R. Thirsk, R. Whit- Paris, 2005.
field, J. Electrochem. Soc., 117, 1007 (1970). [42] D. Devilliers, M. T. Dinh Thi, E. Mah, Q. Le Xuan,
[12] M. Pankuch, R. Bell, C.A. Melendres, Electrochim. Electrochim. Acta, 48, 4301 (2003).
Acta, 38, 2777 (1993). [43] J. Pouilleau, D. Devilliers, F. Garrido, S. Durand-
[13] L.D. Arsov, C. Kormann, W. Plieth, J. Electrochem. Vidal, E. Mah, Mat. Sci. Eng. B, 47, 235 (1997).
Soc., 138, 2964 (1991). [44] E. Mah, D. Devilliers, Electrochim. Acta, 46, 629
[14] T. Fijii, H. Baba, Corros. Sci., 31, 275 (1990). (2000).
[15] Z. Jiang, J. Wang, Q. Hu, S. Huang, Corros. Sci., 37, [45] G. Svehla Vogels, Qualitative Inorganic Analysis,
1245 (1995). 6th edition, Longman, Harlow, UK, 1987, p. 268.
[16] J.L. Delplancke, Thse dagrgation, Universit libre [46] D. Krulic, N. Larabi, N. Fatouros, J. Electroanal.
de Bruxelles, 1997. Chem., 579, 239 (2005).
[17] E.J. Kelly in Modern Aspects of Electrochemistry, [47] D. Devilliers, H. Groult, Can. J. Chem. Eng., 76, 991
Vol.14, Eds., J.OM. Bockris, B.E. Conway, R.E. (1998).
White, Plenum Press, New York, 1982, Chapter 5. [48] G.N. Martelli, R. Ornelas, G. Faita, Electrochim. Acta,
[18] W.M. James, M.E. Straumanis, in Encyclopedia of 39, 1551 (1994).
Electrochemistry of the Elements, Vol. V, Ed., A. J. [49] R. Wthrich, E.A. Baranova, H. Bleuler, C. Comninel-
Bard, M. Dekker, New York, 1976, Chapter 7. lis, Electrochem. Comm., 6, 1199 (2004).
[19] J. Krysa, R. Mraz, I. Rousar, Materials Chemistry and
Physics, 48, 64 (1997).
[20] D. Schlain, J.S. Smatko, J. Electrochem. Soc., 99, 417
(1952).
[21] K. Azumi, N. Yasui, M. Seo, Corros. Sci., 42, 885
(2000).
[22] N.T. Thomas, K. Nobe, J. Electrochem. Soc., 116, 1748
(1969).
[23] E. Brauer, E. Nann, Werkst. Korros., 20, 676 (1969).
[24] D.J. Blackwood, R. Greef, L.M. Peter, Electrochim.
Acta, 34, 875 (1989).
[25] D.J. Blackwood, L.M. Peter, D.E. Williams, Elec-
trochim. Acta, 33, 1143 (1988).
[26] J. Krysa, R. Mraz, Electrochim. Acta, 40, 1997 (1995)
[27] A. Mller, G.S. Popkirov, R.N. Schindler, Ber. Bun-
senges. Phys. Chem., 96, 1432 (1992).
[28] J.P. Frayret, A. Caprani, H. Luc, F. Priem, Corrosion,
40, 14 (1984).
[29] J.P. Frayret, A. Caprani, T. Jaszay, F. Priem, Corro-

You might also like