- Email: [email protected]
Electrochemical properties of porous bismuth electrodes
Electrochimica Acta 55 (2010) 5746–5752
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Electrochemical properties of porous bismuth electrodes T. Romann ∗,1 , E. Lust 1 Institute of Chemistry, University of Tartu, Ravila 14A, 50411 Tartu, Estonia
a r t i c l e
i n f o
Article history: Received 5 February 2010 Received in revised form 3 May 2010 Accepted 4 May 2010 Available online 12 May 2010 Keywords: Bismuth electrode Microelectrode Rough surface Porous metal Frequency dispersion of capacitance
a b s t r a c t The properties of Bi surfaces with different roughnesses were characterized by electron microscopy, cyclic voltammetry, and impedance spectroscopy. Two different strategies were used for preparation of porous bismuth layers onto Bi microelectrode surface in aqueous 0.1 M LiClO4 solution. Firstly, treatment at potential E < −2 V (vs. Ag|AgCl in sat. KCl) has been applied, resulting in bismuth hydride formation and decomposition into Bi nanoparticles which deposit at the electrode surface. Secondly, porous Bi layer was prepared by anodic dissolution (E = 1 V) of bismuth electrode followed by fast electroreduction of formed Bi3+ ions at cathodic potentials E = −2 V. The nanostructured porous bismuth electrode, with surface roughness factor up to 220, has negligible frequency dispersion of capacitance and higher hydrogen evolution overvoltage than observed for smooth Bi electrodes. © 2010 Elsevier Ltd. All rights reserved.
1. Introduction Deposition of porous bismuth by fast electroreduction of the Bi3+ ions has been described in different studies [1–3]. The electrode surface roughening has been detected during anodic oxide layer formation and reduction of Bi2 O3 to a metallic Bi [4–7]. Bi surface roughening has also been established at high cathodic current densities under intensive hydrogen evolution conditions [6–10], explained as a result of decomposition of volatile bismuth hydride BiH3 , formed together with hydrogen, and deposition of Bi nanoclusters [10]. The differential capacitance and chronocoulonometry measurements can be used to determine the surface area formed [6,11–14]. The ideally polarizable electrode behaves as an ideal capacitor because there is no faradaic charge transfer across the solution|electrode phase boundary. In this case, the equivalent electrical circuit (EC) consists only of the solution resistance, Rs , in series with the double-layer capacitance, Cdl [14]. The electropolished Bi single crystal electrode has been found to behave as a nearly ideally polarizable electrode [12]. Frequency dispersion of capacitance is often observed in the case of solid electrodes, which have been attributed to the atomic scale roughness, crystallographic heterogeneity, and chemical inhomogeneities on the solid surface [13–21]. Frequency dispersion can be represented with a constant
∗ Corresponding author. Tel.: +372 7 375175. E-mail address: [email protected] (T. Romann). 1 ISE member. 0013-4686/$ – see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2010.05.012
phase element CPE with an impedance ZCPE = A−1 (jω)−˛ , where A is a CPE coefficient, j = (−1)1/2 , ω is angular frequency of the ac signal, and ˛ is fractional exponent, equal to 1 for an ideal capacitor, 0.5 for Warburg semi-infinite diffusion and 0 for an ideal resistor [13,14]. Porous systems usually show larger frequency dispersion than smooth ones [14,15]. Surprisingly, it was found that porous pure metal or nanoporous carbon electrodes may have only small frequency dispersion effect in the absence of specific adsorption [15–18]. Frequency dispersion occurs for the Au and Pt electrodes in halide solutions within the potential region, where adsorption and superficial species rearrangement take place [22–24]. The properties of a solid substance depend on its size, and the quantum confinement effects occur when the dimensions of a solid reach a level of molecular dimensions. The electrical properties of semimetallic Bi demonstrate strong dependence on the nanowire diameter or nanofilm thickness, and the semimetal-tosemiconductor transition is expected to take place at the film with thicknesses from 23 to 32 nm [25,26]. Oscillating dependence of resistivity on the film thickness, and large magnetoresistance effects appear for the Bi nanostructures [26]. The nanoporous Bi thin films exhibit an order-of-magnitude reduction in thermal conductivity, compared to that of solid films, due to the reduction in the phonon mean free path [27]. The main aim of this work is to compare the electrochemical properties of the smooth and porous Bi electrodes prepared using 50 m diameter bismuth cleaved capillary electrode (BiCCE) and simple in situ roughening techniques.
T. Romann, E. Lust / Electrochimica Acta 55 (2010) 5746–5752
5747
Fig. 1. Photograph (18 cm × 9 cm) of the BiCCE measurement system. Inset: BiCCE cleaving procedure (1.6 mm × 1.2 mm) under optical microscope.
2. Experimental 2.1. Preparation of Bi macro- and microelectrodes and construction of measurement cells The Bi(0 0 1) and Bi(1 1 1) (purity 99.9999%) single crystal electrodes (rhombohedral symmetry indices are used throughout this paper) of 4 mm in diameter were glued into a glass tube with transparent epoxy glue, polished to a mirror finish by 50 nm alumina suspension, and purified in an ultrasound bath filled with pure water [12]. Additionally, the Bi electrodes were electrochemically polished in a HCl + KI solution immediately before each experiment [12,28]. Electrochemical measurements with Bi macroelectrodes were carried out in a single-compartment glass cell including large platinum gauze as a counter electrode and Luggin capillary for the reference electrode connection. Ag|AgCl in saturated KCl aqueous solution was used as a reference electrode in all experiments, and temperature was kept at 25 ± 1 ◦ C. The Bi microelectrodes of 50 ± 10 m diameter were prepared by filling the soda-lime glass capillaries with bismuth (Mateck; purity > 99.9999%) in vacuum according to the procedure described in our previous work [11]. The cleaved surface of BiCCE can be repeatedly prepared in situ under an electrolyte solution by breaking the small cylinders from the top of the electrode (inset in Fig. 1). It has been found that the BiCCE surface is mainly atomically flat with (1 1 1) orientation, however, sometimes there are also some (0 0 1) steps [11]. The solution flow cell (Fig. 1) has four connections: the working electrode (BiCCE) moves to the right before each cleavage; the counter electrode (CE) moves downward, breaking the electrode, and thereafter it moves back; the right channel is for incoming the fresh solution; the downward channel is an outlet for the solution and broken cylinders, also accommodating the Ag|AgCl reference electrode. The BiCCE glass cell was filled with 0.1 M LiClO4 aqueous solution with the help of the syringe pump Aladdin-1000. A small flow rate of 0.1 mL h−1 of the electrolyte solution (satu-
rated with hydrogen) was maintained during the electrochemical measurements in order to keep dissolved oxygen, diffusing slowly through the connections of the measurement cell, away from the electrode surface. The BiCCE with in situ renewable surface in a combination with the solution flow cell enable to prepare repeatedly the new cleaved electrode surface, and thereafter dissolve and deposit a new porous electrode and, after rinsing the measurement compartment with clean electrolyte, to study the electrochemical properties of the deposited Bi in the Bi3+ cation free solution. Due to spherical diffusion for the small electrode, the iR-drop is small even at the high current densities applied. It was found by optical microscope that the hydrogen bubbles do not adsorb onto the Bi microelectrode surface. 2.2. Reagents and other apparatus The solutions for electrochemical measurements were prepared from ultra pure water (MilliQ+), LiClO4 (Sigma Aldrich 99.99%), and were saturated with electrolytic hydrogen (Barken BALSTON hydrogen generator, purity > 99.9999%) for 30 min prior to the measurements in order to remove dissolved oxygen. Electrochemical measurements were carried out using Autolab PGSTAT 30 potentiostat with a FRA 2 impedance measurement system. Frequency range was varied from 10 mHz to 1 MHz with 5 mV ac amplitude. The BiCCE|electrolyte interface was monitored through the glass cell using a Meiji optical reflectance microscope MX8530. The scanning electron microscope (SEM) images were taken with the Leo 1430 apparatus. 3. Results and discussion 3.1. Single crystal Bi macroelectrodes Cyclic voltammetry measurements in 0.1 M LiClO4 aqueous solution indicate low-current density (j) within the potential
5748
T. Romann, E. Lust / Electrochimica Acta 55 (2010) 5746–5752
Fig. 2. (a) Cyclic voltammogram (potential sweep rate 2 mV s−1 ) for Bi(0 0 1) in 0.1 M LiClO4 . (b) Phase angle vs. logarithm of frequency (with solution resistance subtracted) for Bi(0 0 1) in 0.1 M LiClO4 aqueous solution at −0.6 V. Solid and broken lines in (b) have been obtained by fitting of measured spectra with equivalent circuits I and II, respectively. Rel is electrolyte resistance, CPE is constant phase element, Rct is charge transfer resistance and Cdl is double-layer capacitance.
regions from −1.2 to −0.40 V for Bi(0 0 1) (Fig. 2a) and from −1.15 to −0.4 V for Bi(1 1 1) electrode. At the same time the phase angle obtained from electrochemical impedance reached −90◦ (Fig. 2b), and the slope of a log(−Z ) vs. log f plot at −0.6 V (not shown) was −1.000 for Bi(0 0 1) and −0.997 for Bi(1 1 1), obtaining the values of a CPE exponent, ˛, equal to 1.00 and 0.997 for Bi(0 0 1) and Bi(1 1 1) planes, respectively. The equivalent circuit (EC) I in Fig. 2b, that is often used to represent the impedance behaviour of the solid electrodes, was used for fitting the calculated impedance spectra to experimental ones [14]. The parameters for the Bi(0 0 1)|0.1 M LiClO4 aqueous solution interface at E = −0.6 V, calculated according to EC I, have the following values: Rel = 20 cm2 (±0.7%), CPE coefficient A = 21.9 F cm−2 s˛−1 (±0.3%), ˛ = 1.00 (±0.06%) and Rct = 1.9 × 106 cm2 (±18%), square of the standard deviation between the original data and calculated spectrum (each data point weight is normalized by its magnitude), 2 , is equal to 0.003, and weighted sum of squares 2 = 0.22 (40 measurement points within the frequency region from 500 to 0.1 Hz were used in fitting). As the value of ˛ was close to 1 and the error in the value of Rct was large, the spectra for electropolished Bi(0 0 1) were calculated also by simpler EC II in Fig. 2b, and the following values were obtained: Rel = 20 cm2 , Cdl = 22.0 F cm−2 , 2 = 0.008 and 2 = 0.61. Therefore, the conditions of ideal polarizability are nearly satisfied. The value of ˛ equal or higher than 0.997 is the best result ever measured for the bismuth single crystal electrodes, whereas ˛ = 0.997 has been obtained for the Au(1 1 1) electrode in 0.1 M HClO4 solution [29]. The dependence of the differential doublelayer capacitance on potential is cup-shaped and with almost the same value for both Bi(0 0 1) [12] and Bi(1 1 1) electrodes without any large differences. The average values of these capacitances were used for calculation of the surface area in the following experiments. 3.2. BiCCE The potential of BiCCE was fixed at −1.2 V, and some cylinders from the top of BiCCE were broken until a good flat surface
Fig. 3. Series differential capacitance (50 Hz) vs. potential plots (a) and cyclic voltammograms (b) at potential scan rate of 10 mV s−1 in 0.1 M LiClO4 aqueous solution for freshly cleaved and differently roughened BiCCE, as noted in figure.
was achieved [11]. The shape of the Cs vs. E as well as j vs. E-curves for 50 m BiCCE in Fig. 3a and b are similar, but the uncorrected values of the capacitance are three orders of magnitude smaller than those for Bi(0 0 1) or Bi(1 1 1) electrodes with 4 mm diameter. The values of surface roughness from 1.1 to 1.8 were achieved due to the non-ideal cleavage: there were some (0 0 1) steps on the mainly (1 1 1) oriented surface [11], and some bending of the capillary electrode during cleavage process probably caused the separation of bismuth from the glass capillary causing the leakage of an electrolyte solution between the glass walls and lateral sides of the Bi surface. Therefore, the electrochemical impedance spectrum for BiCCE (Fig. 4a) was not ideal: the slope of the log(−Z ) vs. log f dependence (Fig. 4b) calculated was −0.93 (thus, ˛ = 0.93), and phase angles reached only −83◦ (Fig. 4c). The following best-fit estimates of the model parameters have been obtained for the BiCCE|0.1 M LiClO4 aqueous solution interface at E = −0.6 V according to EC I: Rel = 0.89 cm2 , A = 24.2 F cm−2 s˛−1 , ˛ = 0.92, Rct = 7.8 × 104 cm2 , 2 = 0.015, and 2 = 1.1 (40 measurement points from 5000 to 0.1 Hz).
T. Romann, E. Lust / Electrochimica Acta 55 (2010) 5746–5752
5749
3.3. Roughening of the Bi electrode by anodic dissolution and following fast reduction of Bi3+ For preparation of the porous bismuth electrode, BiCCE was held at a potential of +1 V for 1 s followed by immediate potential switch to −2 V for 0.5 s. Then the potential was fixed at −1.2 V and the syringe pump was regulated to the higher flow rate of 5 mL of the electrolyte solution per hour for 10 s in order to remove the unreacted Bi3+ ions from the solution. A small part of the Bi electrode dissolved at +1 V and formed an oversaturated solution of the Bi3+ (and BiO+ ) cations near the electrode surface. By applying the potential step to −2 V, the reduction of the Bi3+ ions occurred, and the formation of black sponge was visible under optical microscope, having much larger volume than the volume of Bi dissolved during the oxidation step. SEM data in Fig. 5a, b and c reveal that different deposit morphologies coexist inside the porous Bi. There are pyramidal compact Bi crystals with a linear dimension up to 800 nm, as well as smaller dendrites consisting of the Bi nanowires with a diameter of about 60 nm. During a few minutes after preparation, the volume of freshly formed bismuth sponge collapses twofold (detected by visual inspection through microscope) and the surface area reduces about 50% based on the values of the capacitance measured. It was observed that the surface area of porous Bi decreases further during some hours, although to a smaller extent. The aging of porous Bi has also been detected previously by other authors [4,6]. This effect can be explained by low melting point of bismuth (271.5 ◦ C). The melting point of a substance reduces noticeably as the dimensions of the solid particles (i.e. Bi dendrites-nanowires in bismuth sponge) reach a molecular lever [30]. However, it has been found experimentally by Olson et al. that the Bi nanoparticles of 5 nm diameter are still in a solid state at room temperature [31]. Thus, the collapsing of the Bi sponge may be caused by the Ostwald ripening of some initially formed tiny structures. Another explanation is that some small Bi structures have more positive surface charge due to the low conductivity of the bismuth nanostructures, and therefore dissolve in an aqueous solution. Our results show that without protective cathodic potential (−1.3 V < E < −0.4 V) the black Bi sponge dissolves completely in a few hours. Electrochemical measurements were carried out half an hour later when the surface area decrease in time was fallen below 8% per hour. The identical shape of the capacitance–potential curve for the Bi sponge in Fig. 3a, compared to that for the smooth Bi macroelectrode [11] or BiCCE, means that the formed Bi surface is clean (no trace of organic compounds). Also the quantum confinement effects influencing the electrical properties of the Bi nanostructures [26] seem to have a negligible effect on the electrochemical properties of porous Bi. The probable explanation is that the Bi surface properties determine the electrochemical properties because the Bi surface is better conductor than the bulk [25], and the metal–semiconductor transition inside the electrode nanostructures does not influence the capacitance of the Bi electrode|electrolyte interface. The surface area of deposited Bi, estimated from the ratio of the capacitances for porous and smooth Bi, was high, and the roughness factor R = 220 was calculated. From the black sponge volume (obtained from the optical microscopy data) and measured capacitance, a specific capacitance of 2 F mL−1 can be estimated for a freshly deposited porous Bi electrode. Taking also into account the charge value cor-
Fig. 4. Imaginary part (−Z ) vs. real part (Z ) of the impedance plot (inset shows the high frequency part) (a) and log(−Z ), log f plot (original data for electrodes) (b) at −0.6 V for cleaved BiCCE (1), treated cathodically (2), or anodically dissolved
and thereafter cathodically reduced BiCCE (3). Dependences of the phase angle and impedance magnitude log|Z| vs. log f (c) for BiCCE (1) and porous Bi (3). Solid lines in plots (a) and (c) have been obtained by fitting of the spectra, calculated by EC I, to the experimental spectra.
5750
T. Romann, E. Lust / Electrochimica Acta 55 (2010) 5746–5752
Fig. 5. SEM images for BiCCE after dissolution and electrodeposition steps.
responding to the reduction of Bi3+ cations, a volume of compact Bi can be calculated. Thus, porosity of about 93% can be estimated from the ratio of the volume of porous Bi and the volume calculated for compact Bi. Analysis of the j, E-curves (Fig. 3b) indicates that the current density at −1.6 V is nearly 5 times lower (or from another aspect the hydrogen evolution overvoltage is 140 mV higher) for deposited Bi compared to BiCCE. The hydrogen evolution reaction at a Bi electrode is believed to be limited by the rate of step (1), followed by the fast electrochemical hydrogen desorption step (2) [32]: H2 O + e− ↔ Hads + OH− (slow)
(1)
−
(2)
−
Hads + H2 O + e → OH + H2 (fast)
Slower hydrogen evolution at the porous Bi sponge might be caused by accumulation of the reaction products, hydrogen bubbles and
hydroxide anions inside the cavities of porous Bi. Although, the negligible hysteresis in the j vs. E curve for porous Bi, and the hemispherical shape of the −Z vs. Z plot (not shown in figures) excludes the diffusion of OH− ions as a limiting step. There are also other explanations, such as stronger adsorption of hydrogen atoms on the porous Bi structure or different electrical properties of the Bi nanostructures. The low-current density region from −1.3 to −0.45 V was established for porous Bi sponge in 0.1 M LiClO4 aqueous solution. Based on the experimental impedance data for porous Bi (given in Fig. 4a), a perfect capacitor behaviour at moderate frequencies (from 0.1 to 60 Hz) has been established, but there are deviations at higher and very low frequencies. For porous electrodes, a part of the surface area is outside of the pores and has impedance similar to a flat surface [16]. In such case, instead of a straight line at 45◦ , observed for cylindrical pores at high frequencies, the angles between 45◦ and 90◦ or a fragment of a semicircle have been obtained [14,16]. According to the high frequency impedance data in Fig. 4a, the deposited Bi sponge formed on the BiCCE has a line angle of 78◦ , and is behaving electrochemically closer to a flat surface than to a porous one [16]. Fig. 4b and c shows that the slope of the log(−Z ) vs. log f dependence for porous Bi is −1.000 and the phase angle reaches −90◦ . The elements in EC I, fitting the deposited porous Bi|0.1 M LiClO4 aqueous solution interface at E = −0.6 V (from 0.1 to 60 Hz), have the following values: Rel = 67 cm2 , A = 21.9 F cm−2 s˛−1 , ˛ = 0.999, and Rct = 9.9·105 cm2 , 2 = 0.0025 and 2 = 0.11. Fitting with EC II gives Rel = 67 cm2 , Cdl = 21.9 F cm−2 , 2 = 0.014 and 2 = 0.68. The fractional exponent value as high as 0.999 for the bismuth sponge is surprising as the porous electrodes, reported in the literature, have usually much larger frequency-dependence of capacitance [14]. Although, porous Au with a roughness factor of 770 was characterized with the value of the CPE exponent ˛ ∼ 0.977 [16], and ˛ up to 0.99 was calculated for the nanoporous carbon cloth electrode (specific surface area of 1500 m2 g−1 ) in the 1 M (C2 H5 )3 CH3 NBF4 solution in acetonitrile [17]. Earlier studies with the lead and tin electrodes [15,18] showed that the frequency dispersion is not caused by surface roughness or porosity of an electrode, but rather by energetic and adsorption heterogeneities on a surface. Nevertheless, large discrepancies between electrochemical properties of the gold single crystal interfaces (zero charge potential difference is up to 450 mV [13]) suggest some frequency-dependence for porous gold. Low melting points and smaller differences in the electrochemical properties of different Sn, Pb and Bi single crystal electrodes [13] suggest lower atomic level roughness and smaller frequency dispersion of the capacitance for porous Sn, Pb, or Bi electrodes compared with porous Au. The roughness factor for porous Pb, prepared using in situ oxidation and reduction steps, is about 50 [15] whereas the roughness factor of porous Au varies from 20 to 2500, depending on the preparation method used [16]. 3.4. Bi electrode behaviour at high cathodic potentials BiCCE was held for 5 min at E = −2.2 V in 0.1 M LiClO4 solution, and the SEM images of the resulting surfaces (Fig. 6) indicate the presence of the porous Bi layers on smooth Bi surface. The structured area under dispersed Bi (Fig. 6b) is not characteristic of a cleaved bismuth surface [11,33], but rather of a chemically or electrochemically etched Bi surface. The magnitude of the C vs. E-curves in Fig. 3a also demonstrates a noticeable roughening of the Bi electrode surface during holding it at E = −2.2 V. The calculated electrode roughness R, obtained by comparison of the capacitance values for freshly cleaved BiCCE and cathodically treated electrode, reaches maximum at −2.1 V (R = 10.4) and at more cathodic potentials surface
T. Romann, E. Lust / Electrochimica Acta 55 (2010) 5746–5752
5751
Fig. 7. Z , Z plot for BiCCE in 0.002 M LiClO4 aqueous solution at different extremely cathodic potentials, given in figure. Solid lines have been obtained by fitting of measured spectra with equivalent circuit III, where Rel is electrolyte resistance, Cdl is double-layer capacitance, Rct is charge transfer resistance, Ca is pseudocapacitance and Ra is adsorption resistance.
Fig. 6. SEM images for BiCCE after 5 min holding at −2.2 V in 0.1 M LiClO4 aqueous solution.
roughness decreases slightly. The shape of the C vs. E curve in Fig. 3a indicates a bismuth surface free from the specifically adsorbed compounds. The impedance properties for this surface (Fig. 4a and b) are intermediate between the properties of BiCCE and highly porous bismuth. The current density values indicate the similar hydrogen evolution overpotential compared to smooth BiCCE. There is a semicircle in the −Z , Z -plot for BiCCE at −1.6 V (not shown in figures) indicating the occurrence of a reaction with a single charge transfer limiting step. The impedance spectra at more cathodic potentials have an additional effect – an inductive loop at the lower frequency region of the spectrum. However, due to the fast hydrogen evolution reaction from concentrated electrolyte solutions, there are large noise effects in the impedance spectra. Therefore, the measurements for the solutions with lower salt con-
centration are presented here. All the impedance spectra in Fig. 7 have the inductive loops, and the spectrum measured at −2.2 V even has a nearly round shape. The inductive loops are often associated to the phase change of a solid electrode, i.e. to corrosion or dissolving of the interfaces [34,35] or electrodeposition of the metals [36]. The Monte-Carlo simulations of metal dissolution show that the adsorbate relaxation can effectively initiate the inductive loops in an electrochemical impedance spectrum [37,38]. Equivalent circuit III in Fig. 7 corresponds to the impedance behaviour of a faradaic reaction involving one adsorbed species [14]. The impedance curves in Fig. 7 can be fitted with EC III, if Ra and Ca have both negative values, or with EC containing an inductance element [14]. In both cases, the forward and reversible rate constants for the electrochemical desorption step (2) have to be larger than the rate constant for step (1) [14]. The parameters for the Bi|0.002 M LiClO4 aqueous solution interface at E = −2.1 V calculated using EC III have the following values: Rel = 11 cm2 , Cdl = 36.0 F cm−2 , Rct = 31 cm2 , Ca = −0.014 F cm−2 and Ra = −13 cm2 , 2 = 0.007, and 2 = 0.37. It should be noted that replacing Ca with CPE would increase the fitting quality noticeably (2 and 2 decrease 4 times), but Cdl was found to be independent of ac frequency. According to the E-pH diagram for the bismuth-water system [39], BiH3 is thermodynamically stable if the electrode potential E is lower than that calculated from the following equation: E(V vs. Ag|AgCl) < −1.00 − 0.0591 pH − 0.0197 log[p(BiH3 )],
(3)
where p(BiH3 ) is the partial pressure of bismuth hydride. The infrared spectra for BiH, BiH2 , and BiH3 in solid matrices have been measured at 4 K [40]. The high-resolution infrared spectrum also indicates the existence of BiH3 at −50 ◦ C, being stable for up to half an hour [41]. Hydride generation with NaBH4 can be used for the detection of bismuth in an aqueous solution [42]. The formation of BiH3 was proved by mass spectroscopy [43] and it was reported to be stable up to 40 s at room temperature [42]. Therefore, the Bi surface roughening can be a result of decomposition of volatile bismuth hydride BiH3 , formed together with molecular hydrogen
5752
T. Romann, E. Lust / Electrochimica Acta 55 (2010) 5746–5752
[10]: Bi + 3Hads ↔ BiH3 ↑ (sidereaction)
Acknowledgements (4)
Reaction (4) leads to the bismuth dissolution that occurs preferably at the Bi surface defects or inner corners of the Bi crystal faces, resulting in the formation of stepped Bi surface like demonstrated in Fig. 6b. The black spots at the crystal corners are probably the pores. Gaseous bismuth hydride exists in these conditions for about a second, decomposing quickly to the Bi nanoparticles. It has been found that the presence of the Bi nanoparticles increases the decomposition speed of BiH3 [42]: 2BiH3 → 2Bi ↓ + 3H2
(5)
The bismuth nanoparticles deposited at the electrode surface or onto the glass capillary locating near the electrode survive under protective cathodic potential, but the particles locating far from the electrode surface dissolve quickly in the 0.1 M LiClO4 aqueous solution. Chen et al. [10] showed that the Bi nanoparticles are stable in a basic solution, but dissolve quickly in the neutral or acidic solutions. This explains the existence of the porous Bi layers on the BiCCE glass capillary (Fig. 6a) where there is a basic solution due to the production of OH− ions as a result of the hydrogen evolution reaction. 4. Conclusions The surface of the bismuth cleaved capillary electrode (BiCCE) can be roughened using high cathodic or anodic potential pulses. Treatment of Bi at potential E = −2.1 V (vs. Ag|AgCl in sat. KCl) created the flat surfaces covered with Bi nanoparticles, characterized with a low frequency inductive loop in the impedance spectrum measured. Anodic dissolution at E = 1 V and following fast electroreduction at E = −2 V yielded a deposition of a porous bismuth sponge (with a surface roughness factor up to 220, but collapsed twofold in a few minutes after deposition) that contained small Bi crystals as well as dendrites composed of 60 nm bismuth nanowires. Despite the rougher surface structure, the porous Bi behaves as an electrochemically less active material. It was demonstrated that porous Bi has higher hydrogen evolution overpotential than that observed for BiCCE or Bi single crystal macroelectrodes. The impedance behaviour for prepared porous bismuth electrode|electrolyte interface is close to a pure capacitor, and only at higher frequencies there are small deviations toward the behaviour characteristic of a porous electrode. The value of constant phase element exponent ˛ equal to 1.00 for bismuth single crystal macroelectrodes as well as for porous bismuth indicates that the frequency dispersion of the capacitance does not directly depend on the surface roughness or porosity. However, it exists for the Bi microelectrodes having isolation defects and enhanced sensitivity to the traces of electroactive dissolved oxygen in the solution. The prepared rough Bi electrodes are useful in electrochemical analysis due to the low background current and enhanced sensitivity for the analyte collection into the porous structure.
This work was supported in part by the Estonian Science Foundation under Projects Nos. 7791 and 6970. References [1] S. Jiang, Y.-H. Huang, F. Luo, N. Du, C.-H. Yan, Inorg. Chem. Commun. 6 (2003) 781. [2] B. O’Brien, M. Plaza, L.Y. Zhu, L. Perez, C.L. Chien, P.C. Searson, J. Phys. Chem. C 112 (2008) 12018. [3] M. Motoyama, Y. Fukunaka, S. Kikuchi, Electrochim. Acta 51 (2005) 897. [4] M.A. Pérez, M. López Teijelo, J. Electroanal. Chem. 583 (2005) 212. [5] M.A. Pérez, O.E. Linarez Pérez, M. López Teijelo, J. Electroanal. Chem. 596 (2006) 149. [6] D.E. Williams, Ph.D. Thesis, University of Auckland, New Zealand, 1974. [7] X. Chen, S. Chen, W. Huang, Z. Li, Cailiao Gongcheng 10 (2008) 243. [8] U.V. Palm, V.E. Past, Russ. J. Phys. Chem. 38 (1964) 447. [9] T.T. Tenno, U.V. Palm, Sov. Electrochem. 9 (1973) 385. [10] X. Chen, S. Chen, W. Huang, J. Zheng, Z. Li, Electrochim. Acta 54 (2009) 7370. [11] T. Romann, S. Kallip, V. Sammelselg, E. Lust, Electrochem. Commun. 10 (2008) 1008. [12] T. Romann, V. Grozovski, E. Lust, Electrochem. Commun. 9 (2007) 2507. [13] E. Lust, in: A.J. Bard, M. Stratmann (Eds.), Encyclopedia of Electrochemistry, vol. 1, Wiley, Weinheim, 2002 (chapter 2). [14] A. Lasia, in: B.E. Conway (Ed.), Modern Aspects of Electrochemistry, vol. 32, 2002, p. 143. [15] N.A. Hampson, C. Lazarides, A.C. Tyrrell, R. Leek, J.P.G. Farr, Surf. Technol. 15 (1982) 167. [16] R. Jurczakowski, C. Hitz, A. Lasia, J. Electroanal. Chem. 572 (2004) 355. [17] A. Jänes, E. Lust, J. Electroanal. Chem. 588 (2006) 285. [18] S.A. Cotgreave, N.A. Hampson, P.C. Morgan, M. Welsh, R. Leek, Surf. Technol. 13 (1981) 107. [19] T. Pajkossy, Solid State Ionics 176 (2005) 1997. [20] Z. Kerner, T. Pajkossy, J. Electroanal. Chem. 448 (1998) 139. [21] Z. Kerner, T. Pajkossy, Electrochim. Acta 46 (2000) 207. [22] T. Pajkossy, J. Electroanal. Chem. 364 (1994) 111. [23] R.S. Neves, E. De Robertis, A.J. Motheo, Electrochim. Acta 51 (2006) 1215. [24] T. Pajkossy, T. Wandlowski, D.M. Kolb, J. Electroanal. Chem. 414 (1996) 209. [25] Ph. Hofmann, Prog. Surf. Sci. 81 (2006) 191. [26] F.Y. Yang, K. Liu, K. Hong, D.H. Reich, P.C. Searson, C.L. Chien, Science 248 (1999) 1335. [27] D.W. Song, W.-N. Shen, B. Dunn, C.D. Moore, M.S. Goorsky, T. Radetic, R. Gronsky, G. Chen, Appl. Phys. Lett. 84 (2004) 1883. [28] E.J. Lust, U.V. Palm, Soviet Electrochem. Engl. Tr. 21 (1985) 1304. [29] Z. Kerner, T. Pajkossy, Electrochim. Acta 47 (2002) 2055. [30] A.P. Chernyshev, Mater. Lett. 63 (2009) 1525. [31] E.A. Olson, M.Yu. Efremov, M. Zhang, Z. Zhang, L.H. Allen, J. Appl. Phys. 97 (2005) 034304–34311. [32] K. Lust, E. Perkson, E. Lust, Russ. J. Electrochem. 36 (2000) 1257. [33] S. Kallip, E. Lust, Electrochem. Commun. 7 (2005) 863. [34] T. Zhang, C. Chen, Y. Shao, G. Meng, F. Wang, X. Li, C. Dong, Electrochim. Acta 53 (2008) 7921. [35] F.-H. Cao, Z. Zhang, J.-X. Su, J.-Q. Zhang, Mater. Corros. 56 (2005) 318. [36] C. Cachet, R. Wiart, I. Ivanov, Y. Stefanov, R. Rashkov, J. Appl. Electrochem. 24 (1994) 713. [37] P. Córdoba-Torres, M. Keddam, R.P. Nogueira, Electrochim. Acta 54 (2008) 518. [38] P. Córdoba-Torres, M. Keddam, R.P. Nogueira, Electrochim. Acta 54 (2009) 6779. [39] B.P. Nikolskii, Spravochnik khimika 3, Moskva, 1964, p. 770. [40] X. Wang, P.F. Souter, L. Andrews, J. Phys. Chem. A 107 (2003) 4244. [41] W. Jerzembeck, H. Bürger, L. Constantin, L. Margulès, J. Demaison, J. Breidung, W. Thiel, Angew. Chem. Int. Ed. 41 (2002) 2550. [42] S.Y. Chen, Z.F. Zhang, H.M. Yu, Anal. Bioanal. Chem. 374 (2002) 126. [43] A. D’Ulivo, Z. Mester, R.E. Sturgeon, Spectrochim. Acta Part B 60 (2005) 423.
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Electrochemical properties of porous bismuth electrodes T. Romann ∗,1 , E. Lust 1 Institute of Chemistry, University of Tartu, Ravila 14A, 50411 Tartu, Estonia
a r t i c l e
i n f o
Article history: Received 5 February 2010 Received in revised form 3 May 2010 Accepted 4 May 2010 Available online 12 May 2010 Keywords: Bismuth electrode Microelectrode Rough surface Porous metal Frequency dispersion of capacitance
a b s t r a c t The properties of Bi surfaces with different roughnesses were characterized by electron microscopy, cyclic voltammetry, and impedance spectroscopy. Two different strategies were used for preparation of porous bismuth layers onto Bi microelectrode surface in aqueous 0.1 M LiClO4 solution. Firstly, treatment at potential E < −2 V (vs. Ag|AgCl in sat. KCl) has been applied, resulting in bismuth hydride formation and decomposition into Bi nanoparticles which deposit at the electrode surface. Secondly, porous Bi layer was prepared by anodic dissolution (E = 1 V) of bismuth electrode followed by fast electroreduction of formed Bi3+ ions at cathodic potentials E = −2 V. The nanostructured porous bismuth electrode, with surface roughness factor up to 220, has negligible frequency dispersion of capacitance and higher hydrogen evolution overvoltage than observed for smooth Bi electrodes. © 2010 Elsevier Ltd. All rights reserved.
1. Introduction Deposition of porous bismuth by fast electroreduction of the Bi3+ ions has been described in different studies [1–3]. The electrode surface roughening has been detected during anodic oxide layer formation and reduction of Bi2 O3 to a metallic Bi [4–7]. Bi surface roughening has also been established at high cathodic current densities under intensive hydrogen evolution conditions [6–10], explained as a result of decomposition of volatile bismuth hydride BiH3 , formed together with hydrogen, and deposition of Bi nanoclusters [10]. The differential capacitance and chronocoulonometry measurements can be used to determine the surface area formed [6,11–14]. The ideally polarizable electrode behaves as an ideal capacitor because there is no faradaic charge transfer across the solution|electrode phase boundary. In this case, the equivalent electrical circuit (EC) consists only of the solution resistance, Rs , in series with the double-layer capacitance, Cdl [14]. The electropolished Bi single crystal electrode has been found to behave as a nearly ideally polarizable electrode [12]. Frequency dispersion of capacitance is often observed in the case of solid electrodes, which have been attributed to the atomic scale roughness, crystallographic heterogeneity, and chemical inhomogeneities on the solid surface [13–21]. Frequency dispersion can be represented with a constant
∗ Corresponding author. Tel.: +372 7 375175. E-mail address: [email protected] (T. Romann). 1 ISE member. 0013-4686/$ – see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2010.05.012
phase element CPE with an impedance ZCPE = A−1 (jω)−˛ , where A is a CPE coefficient, j = (−1)1/2 , ω is angular frequency of the ac signal, and ˛ is fractional exponent, equal to 1 for an ideal capacitor, 0.5 for Warburg semi-infinite diffusion and 0 for an ideal resistor [13,14]. Porous systems usually show larger frequency dispersion than smooth ones [14,15]. Surprisingly, it was found that porous pure metal or nanoporous carbon electrodes may have only small frequency dispersion effect in the absence of specific adsorption [15–18]. Frequency dispersion occurs for the Au and Pt electrodes in halide solutions within the potential region, where adsorption and superficial species rearrangement take place [22–24]. The properties of a solid substance depend on its size, and the quantum confinement effects occur when the dimensions of a solid reach a level of molecular dimensions. The electrical properties of semimetallic Bi demonstrate strong dependence on the nanowire diameter or nanofilm thickness, and the semimetal-tosemiconductor transition is expected to take place at the film with thicknesses from 23 to 32 nm [25,26]. Oscillating dependence of resistivity on the film thickness, and large magnetoresistance effects appear for the Bi nanostructures [26]. The nanoporous Bi thin films exhibit an order-of-magnitude reduction in thermal conductivity, compared to that of solid films, due to the reduction in the phonon mean free path [27]. The main aim of this work is to compare the electrochemical properties of the smooth and porous Bi electrodes prepared using 50 m diameter bismuth cleaved capillary electrode (BiCCE) and simple in situ roughening techniques.
T. Romann, E. Lust / Electrochimica Acta 55 (2010) 5746–5752
5747
Fig. 1. Photograph (18 cm × 9 cm) of the BiCCE measurement system. Inset: BiCCE cleaving procedure (1.6 mm × 1.2 mm) under optical microscope.
2. Experimental 2.1. Preparation of Bi macro- and microelectrodes and construction of measurement cells The Bi(0 0 1) and Bi(1 1 1) (purity 99.9999%) single crystal electrodes (rhombohedral symmetry indices are used throughout this paper) of 4 mm in diameter were glued into a glass tube with transparent epoxy glue, polished to a mirror finish by 50 nm alumina suspension, and purified in an ultrasound bath filled with pure water [12]. Additionally, the Bi electrodes were electrochemically polished in a HCl + KI solution immediately before each experiment [12,28]. Electrochemical measurements with Bi macroelectrodes were carried out in a single-compartment glass cell including large platinum gauze as a counter electrode and Luggin capillary for the reference electrode connection. Ag|AgCl in saturated KCl aqueous solution was used as a reference electrode in all experiments, and temperature was kept at 25 ± 1 ◦ C. The Bi microelectrodes of 50 ± 10 m diameter were prepared by filling the soda-lime glass capillaries with bismuth (Mateck; purity > 99.9999%) in vacuum according to the procedure described in our previous work [11]. The cleaved surface of BiCCE can be repeatedly prepared in situ under an electrolyte solution by breaking the small cylinders from the top of the electrode (inset in Fig. 1). It has been found that the BiCCE surface is mainly atomically flat with (1 1 1) orientation, however, sometimes there are also some (0 0 1) steps [11]. The solution flow cell (Fig. 1) has four connections: the working electrode (BiCCE) moves to the right before each cleavage; the counter electrode (CE) moves downward, breaking the electrode, and thereafter it moves back; the right channel is for incoming the fresh solution; the downward channel is an outlet for the solution and broken cylinders, also accommodating the Ag|AgCl reference electrode. The BiCCE glass cell was filled with 0.1 M LiClO4 aqueous solution with the help of the syringe pump Aladdin-1000. A small flow rate of 0.1 mL h−1 of the electrolyte solution (satu-
rated with hydrogen) was maintained during the electrochemical measurements in order to keep dissolved oxygen, diffusing slowly through the connections of the measurement cell, away from the electrode surface. The BiCCE with in situ renewable surface in a combination with the solution flow cell enable to prepare repeatedly the new cleaved electrode surface, and thereafter dissolve and deposit a new porous electrode and, after rinsing the measurement compartment with clean electrolyte, to study the electrochemical properties of the deposited Bi in the Bi3+ cation free solution. Due to spherical diffusion for the small electrode, the iR-drop is small even at the high current densities applied. It was found by optical microscope that the hydrogen bubbles do not adsorb onto the Bi microelectrode surface. 2.2. Reagents and other apparatus The solutions for electrochemical measurements were prepared from ultra pure water (MilliQ+), LiClO4 (Sigma Aldrich 99.99%), and were saturated with electrolytic hydrogen (Barken BALSTON hydrogen generator, purity > 99.9999%) for 30 min prior to the measurements in order to remove dissolved oxygen. Electrochemical measurements were carried out using Autolab PGSTAT 30 potentiostat with a FRA 2 impedance measurement system. Frequency range was varied from 10 mHz to 1 MHz with 5 mV ac amplitude. The BiCCE|electrolyte interface was monitored through the glass cell using a Meiji optical reflectance microscope MX8530. The scanning electron microscope (SEM) images were taken with the Leo 1430 apparatus. 3. Results and discussion 3.1. Single crystal Bi macroelectrodes Cyclic voltammetry measurements in 0.1 M LiClO4 aqueous solution indicate low-current density (j) within the potential
5748
T. Romann, E. Lust / Electrochimica Acta 55 (2010) 5746–5752
Fig. 2. (a) Cyclic voltammogram (potential sweep rate 2 mV s−1 ) for Bi(0 0 1) in 0.1 M LiClO4 . (b) Phase angle vs. logarithm of frequency (with solution resistance subtracted) for Bi(0 0 1) in 0.1 M LiClO4 aqueous solution at −0.6 V. Solid and broken lines in (b) have been obtained by fitting of measured spectra with equivalent circuits I and II, respectively. Rel is electrolyte resistance, CPE is constant phase element, Rct is charge transfer resistance and Cdl is double-layer capacitance.
regions from −1.2 to −0.40 V for Bi(0 0 1) (Fig. 2a) and from −1.15 to −0.4 V for Bi(1 1 1) electrode. At the same time the phase angle obtained from electrochemical impedance reached −90◦ (Fig. 2b), and the slope of a log(−Z ) vs. log f plot at −0.6 V (not shown) was −1.000 for Bi(0 0 1) and −0.997 for Bi(1 1 1), obtaining the values of a CPE exponent, ˛, equal to 1.00 and 0.997 for Bi(0 0 1) and Bi(1 1 1) planes, respectively. The equivalent circuit (EC) I in Fig. 2b, that is often used to represent the impedance behaviour of the solid electrodes, was used for fitting the calculated impedance spectra to experimental ones [14]. The parameters for the Bi(0 0 1)|0.1 M LiClO4 aqueous solution interface at E = −0.6 V, calculated according to EC I, have the following values: Rel = 20 cm2 (±0.7%), CPE coefficient A = 21.9 F cm−2 s˛−1 (±0.3%), ˛ = 1.00 (±0.06%) and Rct = 1.9 × 106 cm2 (±18%), square of the standard deviation between the original data and calculated spectrum (each data point weight is normalized by its magnitude), 2 , is equal to 0.003, and weighted sum of squares 2 = 0.22 (40 measurement points within the frequency region from 500 to 0.1 Hz were used in fitting). As the value of ˛ was close to 1 and the error in the value of Rct was large, the spectra for electropolished Bi(0 0 1) were calculated also by simpler EC II in Fig. 2b, and the following values were obtained: Rel = 20 cm2 , Cdl = 22.0 F cm−2 , 2 = 0.008 and 2 = 0.61. Therefore, the conditions of ideal polarizability are nearly satisfied. The value of ˛ equal or higher than 0.997 is the best result ever measured for the bismuth single crystal electrodes, whereas ˛ = 0.997 has been obtained for the Au(1 1 1) electrode in 0.1 M HClO4 solution [29]. The dependence of the differential doublelayer capacitance on potential is cup-shaped and with almost the same value for both Bi(0 0 1) [12] and Bi(1 1 1) electrodes without any large differences. The average values of these capacitances were used for calculation of the surface area in the following experiments. 3.2. BiCCE The potential of BiCCE was fixed at −1.2 V, and some cylinders from the top of BiCCE were broken until a good flat surface
Fig. 3. Series differential capacitance (50 Hz) vs. potential plots (a) and cyclic voltammograms (b) at potential scan rate of 10 mV s−1 in 0.1 M LiClO4 aqueous solution for freshly cleaved and differently roughened BiCCE, as noted in figure.
was achieved [11]. The shape of the Cs vs. E as well as j vs. E-curves for 50 m BiCCE in Fig. 3a and b are similar, but the uncorrected values of the capacitance are three orders of magnitude smaller than those for Bi(0 0 1) or Bi(1 1 1) electrodes with 4 mm diameter. The values of surface roughness from 1.1 to 1.8 were achieved due to the non-ideal cleavage: there were some (0 0 1) steps on the mainly (1 1 1) oriented surface [11], and some bending of the capillary electrode during cleavage process probably caused the separation of bismuth from the glass capillary causing the leakage of an electrolyte solution between the glass walls and lateral sides of the Bi surface. Therefore, the electrochemical impedance spectrum for BiCCE (Fig. 4a) was not ideal: the slope of the log(−Z ) vs. log f dependence (Fig. 4b) calculated was −0.93 (thus, ˛ = 0.93), and phase angles reached only −83◦ (Fig. 4c). The following best-fit estimates of the model parameters have been obtained for the BiCCE|0.1 M LiClO4 aqueous solution interface at E = −0.6 V according to EC I: Rel = 0.89 cm2 , A = 24.2 F cm−2 s˛−1 , ˛ = 0.92, Rct = 7.8 × 104 cm2 , 2 = 0.015, and 2 = 1.1 (40 measurement points from 5000 to 0.1 Hz).
T. Romann, E. Lust / Electrochimica Acta 55 (2010) 5746–5752
5749
3.3. Roughening of the Bi electrode by anodic dissolution and following fast reduction of Bi3+ For preparation of the porous bismuth electrode, BiCCE was held at a potential of +1 V for 1 s followed by immediate potential switch to −2 V for 0.5 s. Then the potential was fixed at −1.2 V and the syringe pump was regulated to the higher flow rate of 5 mL of the electrolyte solution per hour for 10 s in order to remove the unreacted Bi3+ ions from the solution. A small part of the Bi electrode dissolved at +1 V and formed an oversaturated solution of the Bi3+ (and BiO+ ) cations near the electrode surface. By applying the potential step to −2 V, the reduction of the Bi3+ ions occurred, and the formation of black sponge was visible under optical microscope, having much larger volume than the volume of Bi dissolved during the oxidation step. SEM data in Fig. 5a, b and c reveal that different deposit morphologies coexist inside the porous Bi. There are pyramidal compact Bi crystals with a linear dimension up to 800 nm, as well as smaller dendrites consisting of the Bi nanowires with a diameter of about 60 nm. During a few minutes after preparation, the volume of freshly formed bismuth sponge collapses twofold (detected by visual inspection through microscope) and the surface area reduces about 50% based on the values of the capacitance measured. It was observed that the surface area of porous Bi decreases further during some hours, although to a smaller extent. The aging of porous Bi has also been detected previously by other authors [4,6]. This effect can be explained by low melting point of bismuth (271.5 ◦ C). The melting point of a substance reduces noticeably as the dimensions of the solid particles (i.e. Bi dendrites-nanowires in bismuth sponge) reach a molecular lever [30]. However, it has been found experimentally by Olson et al. that the Bi nanoparticles of 5 nm diameter are still in a solid state at room temperature [31]. Thus, the collapsing of the Bi sponge may be caused by the Ostwald ripening of some initially formed tiny structures. Another explanation is that some small Bi structures have more positive surface charge due to the low conductivity of the bismuth nanostructures, and therefore dissolve in an aqueous solution. Our results show that without protective cathodic potential (−1.3 V < E < −0.4 V) the black Bi sponge dissolves completely in a few hours. Electrochemical measurements were carried out half an hour later when the surface area decrease in time was fallen below 8% per hour. The identical shape of the capacitance–potential curve for the Bi sponge in Fig. 3a, compared to that for the smooth Bi macroelectrode [11] or BiCCE, means that the formed Bi surface is clean (no trace of organic compounds). Also the quantum confinement effects influencing the electrical properties of the Bi nanostructures [26] seem to have a negligible effect on the electrochemical properties of porous Bi. The probable explanation is that the Bi surface properties determine the electrochemical properties because the Bi surface is better conductor than the bulk [25], and the metal–semiconductor transition inside the electrode nanostructures does not influence the capacitance of the Bi electrode|electrolyte interface. The surface area of deposited Bi, estimated from the ratio of the capacitances for porous and smooth Bi, was high, and the roughness factor R = 220 was calculated. From the black sponge volume (obtained from the optical microscopy data) and measured capacitance, a specific capacitance of 2 F mL−1 can be estimated for a freshly deposited porous Bi electrode. Taking also into account the charge value cor-
Fig. 4. Imaginary part (−Z ) vs. real part (Z ) of the impedance plot (inset shows the high frequency part) (a) and log(−Z ), log f plot (original data for electrodes) (b) at −0.6 V for cleaved BiCCE (1), treated cathodically (2), or anodically dissolved
and thereafter cathodically reduced BiCCE (3). Dependences of the phase angle and impedance magnitude log|Z| vs. log f (c) for BiCCE (1) and porous Bi (3). Solid lines in plots (a) and (c) have been obtained by fitting of the spectra, calculated by EC I, to the experimental spectra.
5750
T. Romann, E. Lust / Electrochimica Acta 55 (2010) 5746–5752
Fig. 5. SEM images for BiCCE after dissolution and electrodeposition steps.
responding to the reduction of Bi3+ cations, a volume of compact Bi can be calculated. Thus, porosity of about 93% can be estimated from the ratio of the volume of porous Bi and the volume calculated for compact Bi. Analysis of the j, E-curves (Fig. 3b) indicates that the current density at −1.6 V is nearly 5 times lower (or from another aspect the hydrogen evolution overvoltage is 140 mV higher) for deposited Bi compared to BiCCE. The hydrogen evolution reaction at a Bi electrode is believed to be limited by the rate of step (1), followed by the fast electrochemical hydrogen desorption step (2) [32]: H2 O + e− ↔ Hads + OH− (slow)
(1)
−
(2)
−
Hads + H2 O + e → OH + H2 (fast)
Slower hydrogen evolution at the porous Bi sponge might be caused by accumulation of the reaction products, hydrogen bubbles and
hydroxide anions inside the cavities of porous Bi. Although, the negligible hysteresis in the j vs. E curve for porous Bi, and the hemispherical shape of the −Z vs. Z plot (not shown in figures) excludes the diffusion of OH− ions as a limiting step. There are also other explanations, such as stronger adsorption of hydrogen atoms on the porous Bi structure or different electrical properties of the Bi nanostructures. The low-current density region from −1.3 to −0.45 V was established for porous Bi sponge in 0.1 M LiClO4 aqueous solution. Based on the experimental impedance data for porous Bi (given in Fig. 4a), a perfect capacitor behaviour at moderate frequencies (from 0.1 to 60 Hz) has been established, but there are deviations at higher and very low frequencies. For porous electrodes, a part of the surface area is outside of the pores and has impedance similar to a flat surface [16]. In such case, instead of a straight line at 45◦ , observed for cylindrical pores at high frequencies, the angles between 45◦ and 90◦ or a fragment of a semicircle have been obtained [14,16]. According to the high frequency impedance data in Fig. 4a, the deposited Bi sponge formed on the BiCCE has a line angle of 78◦ , and is behaving electrochemically closer to a flat surface than to a porous one [16]. Fig. 4b and c shows that the slope of the log(−Z ) vs. log f dependence for porous Bi is −1.000 and the phase angle reaches −90◦ . The elements in EC I, fitting the deposited porous Bi|0.1 M LiClO4 aqueous solution interface at E = −0.6 V (from 0.1 to 60 Hz), have the following values: Rel = 67 cm2 , A = 21.9 F cm−2 s˛−1 , ˛ = 0.999, and Rct = 9.9·105 cm2 , 2 = 0.0025 and 2 = 0.11. Fitting with EC II gives Rel = 67 cm2 , Cdl = 21.9 F cm−2 , 2 = 0.014 and 2 = 0.68. The fractional exponent value as high as 0.999 for the bismuth sponge is surprising as the porous electrodes, reported in the literature, have usually much larger frequency-dependence of capacitance [14]. Although, porous Au with a roughness factor of 770 was characterized with the value of the CPE exponent ˛ ∼ 0.977 [16], and ˛ up to 0.99 was calculated for the nanoporous carbon cloth electrode (specific surface area of 1500 m2 g−1 ) in the 1 M (C2 H5 )3 CH3 NBF4 solution in acetonitrile [17]. Earlier studies with the lead and tin electrodes [15,18] showed that the frequency dispersion is not caused by surface roughness or porosity of an electrode, but rather by energetic and adsorption heterogeneities on a surface. Nevertheless, large discrepancies between electrochemical properties of the gold single crystal interfaces (zero charge potential difference is up to 450 mV [13]) suggest some frequency-dependence for porous gold. Low melting points and smaller differences in the electrochemical properties of different Sn, Pb and Bi single crystal electrodes [13] suggest lower atomic level roughness and smaller frequency dispersion of the capacitance for porous Sn, Pb, or Bi electrodes compared with porous Au. The roughness factor for porous Pb, prepared using in situ oxidation and reduction steps, is about 50 [15] whereas the roughness factor of porous Au varies from 20 to 2500, depending on the preparation method used [16]. 3.4. Bi electrode behaviour at high cathodic potentials BiCCE was held for 5 min at E = −2.2 V in 0.1 M LiClO4 solution, and the SEM images of the resulting surfaces (Fig. 6) indicate the presence of the porous Bi layers on smooth Bi surface. The structured area under dispersed Bi (Fig. 6b) is not characteristic of a cleaved bismuth surface [11,33], but rather of a chemically or electrochemically etched Bi surface. The magnitude of the C vs. E-curves in Fig. 3a also demonstrates a noticeable roughening of the Bi electrode surface during holding it at E = −2.2 V. The calculated electrode roughness R, obtained by comparison of the capacitance values for freshly cleaved BiCCE and cathodically treated electrode, reaches maximum at −2.1 V (R = 10.4) and at more cathodic potentials surface
T. Romann, E. Lust / Electrochimica Acta 55 (2010) 5746–5752
5751
Fig. 7. Z , Z plot for BiCCE in 0.002 M LiClO4 aqueous solution at different extremely cathodic potentials, given in figure. Solid lines have been obtained by fitting of measured spectra with equivalent circuit III, where Rel is electrolyte resistance, Cdl is double-layer capacitance, Rct is charge transfer resistance, Ca is pseudocapacitance and Ra is adsorption resistance.
Fig. 6. SEM images for BiCCE after 5 min holding at −2.2 V in 0.1 M LiClO4 aqueous solution.
roughness decreases slightly. The shape of the C vs. E curve in Fig. 3a indicates a bismuth surface free from the specifically adsorbed compounds. The impedance properties for this surface (Fig. 4a and b) are intermediate between the properties of BiCCE and highly porous bismuth. The current density values indicate the similar hydrogen evolution overpotential compared to smooth BiCCE. There is a semicircle in the −Z , Z -plot for BiCCE at −1.6 V (not shown in figures) indicating the occurrence of a reaction with a single charge transfer limiting step. The impedance spectra at more cathodic potentials have an additional effect – an inductive loop at the lower frequency region of the spectrum. However, due to the fast hydrogen evolution reaction from concentrated electrolyte solutions, there are large noise effects in the impedance spectra. Therefore, the measurements for the solutions with lower salt con-
centration are presented here. All the impedance spectra in Fig. 7 have the inductive loops, and the spectrum measured at −2.2 V even has a nearly round shape. The inductive loops are often associated to the phase change of a solid electrode, i.e. to corrosion or dissolving of the interfaces [34,35] or electrodeposition of the metals [36]. The Monte-Carlo simulations of metal dissolution show that the adsorbate relaxation can effectively initiate the inductive loops in an electrochemical impedance spectrum [37,38]. Equivalent circuit III in Fig. 7 corresponds to the impedance behaviour of a faradaic reaction involving one adsorbed species [14]. The impedance curves in Fig. 7 can be fitted with EC III, if Ra and Ca have both negative values, or with EC containing an inductance element [14]. In both cases, the forward and reversible rate constants for the electrochemical desorption step (2) have to be larger than the rate constant for step (1) [14]. The parameters for the Bi|0.002 M LiClO4 aqueous solution interface at E = −2.1 V calculated using EC III have the following values: Rel = 11 cm2 , Cdl = 36.0 F cm−2 , Rct = 31 cm2 , Ca = −0.014 F cm−2 and Ra = −13 cm2 , 2 = 0.007, and 2 = 0.37. It should be noted that replacing Ca with CPE would increase the fitting quality noticeably (2 and 2 decrease 4 times), but Cdl was found to be independent of ac frequency. According to the E-pH diagram for the bismuth-water system [39], BiH3 is thermodynamically stable if the electrode potential E is lower than that calculated from the following equation: E(V vs. Ag|AgCl) < −1.00 − 0.0591 pH − 0.0197 log[p(BiH3 )],
(3)
where p(BiH3 ) is the partial pressure of bismuth hydride. The infrared spectra for BiH, BiH2 , and BiH3 in solid matrices have been measured at 4 K [40]. The high-resolution infrared spectrum also indicates the existence of BiH3 at −50 ◦ C, being stable for up to half an hour [41]. Hydride generation with NaBH4 can be used for the detection of bismuth in an aqueous solution [42]. The formation of BiH3 was proved by mass spectroscopy [43] and it was reported to be stable up to 40 s at room temperature [42]. Therefore, the Bi surface roughening can be a result of decomposition of volatile bismuth hydride BiH3 , formed together with molecular hydrogen
5752
T. Romann, E. Lust / Electrochimica Acta 55 (2010) 5746–5752
[10]: Bi + 3Hads ↔ BiH3 ↑ (sidereaction)
Acknowledgements (4)
Reaction (4) leads to the bismuth dissolution that occurs preferably at the Bi surface defects or inner corners of the Bi crystal faces, resulting in the formation of stepped Bi surface like demonstrated in Fig. 6b. The black spots at the crystal corners are probably the pores. Gaseous bismuth hydride exists in these conditions for about a second, decomposing quickly to the Bi nanoparticles. It has been found that the presence of the Bi nanoparticles increases the decomposition speed of BiH3 [42]: 2BiH3 → 2Bi ↓ + 3H2
(5)
The bismuth nanoparticles deposited at the electrode surface or onto the glass capillary locating near the electrode survive under protective cathodic potential, but the particles locating far from the electrode surface dissolve quickly in the 0.1 M LiClO4 aqueous solution. Chen et al. [10] showed that the Bi nanoparticles are stable in a basic solution, but dissolve quickly in the neutral or acidic solutions. This explains the existence of the porous Bi layers on the BiCCE glass capillary (Fig. 6a) where there is a basic solution due to the production of OH− ions as a result of the hydrogen evolution reaction. 4. Conclusions The surface of the bismuth cleaved capillary electrode (BiCCE) can be roughened using high cathodic or anodic potential pulses. Treatment of Bi at potential E = −2.1 V (vs. Ag|AgCl in sat. KCl) created the flat surfaces covered with Bi nanoparticles, characterized with a low frequency inductive loop in the impedance spectrum measured. Anodic dissolution at E = 1 V and following fast electroreduction at E = −2 V yielded a deposition of a porous bismuth sponge (with a surface roughness factor up to 220, but collapsed twofold in a few minutes after deposition) that contained small Bi crystals as well as dendrites composed of 60 nm bismuth nanowires. Despite the rougher surface structure, the porous Bi behaves as an electrochemically less active material. It was demonstrated that porous Bi has higher hydrogen evolution overpotential than that observed for BiCCE or Bi single crystal macroelectrodes. The impedance behaviour for prepared porous bismuth electrode|electrolyte interface is close to a pure capacitor, and only at higher frequencies there are small deviations toward the behaviour characteristic of a porous electrode. The value of constant phase element exponent ˛ equal to 1.00 for bismuth single crystal macroelectrodes as well as for porous bismuth indicates that the frequency dispersion of the capacitance does not directly depend on the surface roughness or porosity. However, it exists for the Bi microelectrodes having isolation defects and enhanced sensitivity to the traces of electroactive dissolved oxygen in the solution. The prepared rough Bi electrodes are useful in electrochemical analysis due to the low background current and enhanced sensitivity for the analyte collection into the porous structure.
This work was supported in part by the Estonian Science Foundation under Projects Nos. 7791 and 6970. References [1] S. Jiang, Y.-H. Huang, F. Luo, N. Du, C.-H. Yan, Inorg. Chem. Commun. 6 (2003) 781. [2] B. O’Brien, M. Plaza, L.Y. Zhu, L. Perez, C.L. Chien, P.C. Searson, J. Phys. Chem. C 112 (2008) 12018. [3] M. Motoyama, Y. Fukunaka, S. Kikuchi, Electrochim. Acta 51 (2005) 897. [4] M.A. Pérez, M. López Teijelo, J. Electroanal. Chem. 583 (2005) 212. [5] M.A. Pérez, O.E. Linarez Pérez, M. López Teijelo, J. Electroanal. Chem. 596 (2006) 149. [6] D.E. Williams, Ph.D. Thesis, University of Auckland, New Zealand, 1974. [7] X. Chen, S. Chen, W. Huang, Z. Li, Cailiao Gongcheng 10 (2008) 243. [8] U.V. Palm, V.E. Past, Russ. J. Phys. Chem. 38 (1964) 447. [9] T.T. Tenno, U.V. Palm, Sov. Electrochem. 9 (1973) 385. [10] X. Chen, S. Chen, W. Huang, J. Zheng, Z. Li, Electrochim. Acta 54 (2009) 7370. [11] T. Romann, S. Kallip, V. Sammelselg, E. Lust, Electrochem. Commun. 10 (2008) 1008. [12] T. Romann, V. Grozovski, E. Lust, Electrochem. Commun. 9 (2007) 2507. [13] E. Lust, in: A.J. Bard, M. Stratmann (Eds.), Encyclopedia of Electrochemistry, vol. 1, Wiley, Weinheim, 2002 (chapter 2). [14] A. Lasia, in: B.E. Conway (Ed.), Modern Aspects of Electrochemistry, vol. 32, 2002, p. 143. [15] N.A. Hampson, C. Lazarides, A.C. Tyrrell, R. Leek, J.P.G. Farr, Surf. Technol. 15 (1982) 167. [16] R. Jurczakowski, C. Hitz, A. Lasia, J. Electroanal. Chem. 572 (2004) 355. [17] A. Jänes, E. Lust, J. Electroanal. Chem. 588 (2006) 285. [18] S.A. Cotgreave, N.A. Hampson, P.C. Morgan, M. Welsh, R. Leek, Surf. Technol. 13 (1981) 107. [19] T. Pajkossy, Solid State Ionics 176 (2005) 1997. [20] Z. Kerner, T. Pajkossy, J. Electroanal. Chem. 448 (1998) 139. [21] Z. Kerner, T. Pajkossy, Electrochim. Acta 46 (2000) 207. [22] T. Pajkossy, J. Electroanal. Chem. 364 (1994) 111. [23] R.S. Neves, E. De Robertis, A.J. Motheo, Electrochim. Acta 51 (2006) 1215. [24] T. Pajkossy, T. Wandlowski, D.M. Kolb, J. Electroanal. Chem. 414 (1996) 209. [25] Ph. Hofmann, Prog. Surf. Sci. 81 (2006) 191. [26] F.Y. Yang, K. Liu, K. Hong, D.H. Reich, P.C. Searson, C.L. Chien, Science 248 (1999) 1335. [27] D.W. Song, W.-N. Shen, B. Dunn, C.D. Moore, M.S. Goorsky, T. Radetic, R. Gronsky, G. Chen, Appl. Phys. Lett. 84 (2004) 1883. [28] E.J. Lust, U.V. Palm, Soviet Electrochem. Engl. Tr. 21 (1985) 1304. [29] Z. Kerner, T. Pajkossy, Electrochim. Acta 47 (2002) 2055. [30] A.P. Chernyshev, Mater. Lett. 63 (2009) 1525. [31] E.A. Olson, M.Yu. Efremov, M. Zhang, Z. Zhang, L.H. Allen, J. Appl. Phys. 97 (2005) 034304–34311. [32] K. Lust, E. Perkson, E. Lust, Russ. J. Electrochem. 36 (2000) 1257. [33] S. Kallip, E. Lust, Electrochem. Commun. 7 (2005) 863. [34] T. Zhang, C. Chen, Y. Shao, G. Meng, F. Wang, X. Li, C. Dong, Electrochim. Acta 53 (2008) 7921. [35] F.-H. Cao, Z. Zhang, J.-X. Su, J.-Q. Zhang, Mater. Corros. 56 (2005) 318. [36] C. Cachet, R. Wiart, I. Ivanov, Y. Stefanov, R. Rashkov, J. Appl. Electrochem. 24 (1994) 713. [37] P. Córdoba-Torres, M. Keddam, R.P. Nogueira, Electrochim. Acta 54 (2008) 518. [38] P. Córdoba-Torres, M. Keddam, R.P. Nogueira, Electrochim. Acta 54 (2009) 6779. [39] B.P. Nikolskii, Spravochnik khimika 3, Moskva, 1964, p. 770. [40] X. Wang, P.F. Souter, L. Andrews, J. Phys. Chem. A 107 (2003) 4244. [41] W. Jerzembeck, H. Bürger, L. Constantin, L. Margulès, J. Demaison, J. Breidung, W. Thiel, Angew. Chem. Int. Ed. 41 (2002) 2550. [42] S.Y. Chen, Z.F. Zhang, H.M. Yu, Anal. Bioanal. Chem. 374 (2002) 126. [43] A. D’Ulivo, Z. Mester, R.E. Sturgeon, Spectrochim. Acta Part B 60 (2005) 423.