Electrodeposition of Cobalt Selenide Thin Films: An Electrochemical Quartz Crystal Microgravimetry Study

The mechanism of electrodeposition of cobalt selenide (CoSe) thin ﬁlms was investigated by the combined application of linear sweep voltammetry (LSV) and electrochemical quartz crystal microgravimetry (EQCM) on Pt-coated quartz electrodes. Cobalt selenide ﬁlms were electrodeposited on the Pt surface from 0.1 M Na 2 SO 4 electrolyte solution containing 5 mM SeO 2 and 5 mM Co(CH 3 COO) 2 by linear sweep voltammetry. Four cathodic waves were observed during the linear scans and the reactions corresponding to these waves were investigated with LSV and EQCM. Combined stripping voltammetry and EQCM showed that CoSe was electrodeposited via two routes: (1) Underpotential deposition of Se followed by deposition of cobalt as CoSe; and (2) Reaction of Co(II) with electrogenerated Se(-II) to result in CoSe. Compositional analyses revealed that the electrodeposited ﬁlms contained CoSe and free Se, depending on the deposition potential. However, no cobalt was found in these ﬁlms because of chemical (galvanic) instability of the cobalt ﬁlm in the deposition bath.

Transition metal chalcogenides, comprised of earth-abundant elements, have served as effective alternatives for expensive, noble metal catalysts for the hydrogen evolution reaction (HER), and the oxygen evolution reaction (OER). 1 Especially, cobalt selenides have attracted much attention due to their favorable attributes such as good electrical conductivity, optimal bandgap (∼1.5 eV) in terms of match with the solar spectrum, and a high optical absorption coefficient. Thus they have been used as counterelectrodes in photoelectrochemical cells and in dye-sensitized solar cells, 2,3 as electrocatalysts for HER [4][5][6][7][8] and OER, [5][6][7][8] as anode material in lithium-ion batteries, 9 and as a visiblelight absorber in photovoltaic solar cells. 10 The phase diagram of the Co-Se system 11 reveals two homogeneous and stable compounds at room temperature: CoSe 2 and CoSe. Two other possible compositions (Co 3 O 4 and Co 2 Se 3 ) have not yet been conclusively established. 11 Cobalt selenides have been synthesized by several routes including chemical bath deposition, 12,13 solvothermal method, 14 mechanical alloying, 15 and electrodeposition. [2][3][4]6,7,10,16 Of these variant synthetic options, electrodeposition offers many advantages 17 such as cost-effectiveness, simplicity, easy scale-up, the use of relatively mild conditions, and the fact that (volatile) organometallic chemicals are not needed (as in techniques such as molecular beam epitaxy and atomic layer deposition). In addition, the composition and structure of electrodeposited films can be controlled by controlling pH, electrolyte composition, or deposition potential.
In this work, we present a detailed study of the electrodeposition mechanism of cobalt selenide (CoSe) using combined linear sweep voltammetry and electrochemical quartz crystal microgravimetry (EQCM). We are only aware of one prior study on the electrodeposition mechanism for CoSe on fluorine-doped tin oxide (FTO) electrode using cyclic voltammetry. 16 On the other hand, the addition of a complementary mass-change probe such as EQCM to a voltammetrybased study, offers considerable advantage in a mechanistic sense as we demonstrate below. Compositional analysis of the electrodeposited films using combined stripping voltammetry and EQCM, is finally demonstrated.

Experimental
Cobalt acetate tetrahydrate, selenium dioxide, and sodium sulfate were from Sigma-Aldrich, and used without further purification. For voltammetry and EQCM, an EG&G Princeton Applied Research 263A instrument equipped with Power Suite electrochemistry software, a Seiko EG&G model QCA 922 instrument and an oscillator module (QCA 922-10), was used. A single compartment, threeelectrode cell setup was used for electrochemical experiments at room temperature and comprised of an AT-cut, Pt-coated quartz crystal (geometric area, 0.2 cm 2 ) working electrode, a Pt counter electrode, and a Ag/AgCl/3 M NaCl reference electrode. All potentials below are quoted with respect to this reference electrode. For voltammetry, the potential scan rate was 20 mV/s. The cleanliness of the Pt working electrode surface was checked by cyclic voltammetry in 0.5 M H 2 SO 4 by cycling potential between 0.8 V and −0.6 V until the voltammetric and EQCM frequency signals were stable. The electrolytes were degassed with high-purity nitrogen prior to the electrochemical measurements and nitrogen blanket was used during the measurements.
X-ray diffraction (XRD) patterns were recorded on a Bruker D8 DISCOVER diffractometer with Cu Kα radiation source. Film morphology and composition were obtained on a field emission scanning electron microscope (JEOL Model 6700F) equipped with an energydispersive X-ray emission analysis (EDX) probe. 89 V and substantial current flow beyond −1.3 V due to proton reduction (i.e., HER). During the cathodic scan, frequency decrease (mass increase) started at ∼−0.4 V, coinciding with the first cathodic wave and continuing until ∼−1.5 V. No frequency (i.e., mass) change was observed at the linear scans after ∼−1.5 V. Unlike the cyclic voltammograms reported for the FTO electrode in LiCl solutions containing SeO 2 and Co(CH 3 COO) 2 by previous workers, 16 Figure 1 shows a more complex wave morphology, which will be assigned and explained in detail below.

Results and Discussion
To probe reactions occurring at each cathodic wave in Figure 1, experiments were performed using combined linear sweep voltammetry and EQCM at the Pt electrode in electrolytes containing only selenium or cobalt species. Figure 2A shows an LSV scan and corresponding frequency changes for a Pt-EQCM electrode in 0.1 M Na 2 SO 4 electrolyte containing 5 mM Co(CH 3 COO) 2 . The cathodic wave at −0.87 V can be assigned to cobalt deposition, which is consistent with the frequency decrease (mass gain). Comparison of this cathodic wave with Wave d in Figure 1 clearly revealed that Wave  d stemmed from cobalt reduction. As shown in Figure 1, no further decrease in frequency was observed again after ∼ −1.5 V.
It is well known that the electrochemistry of selenium is complicated due to several factors including underpotential deposition, coupled chemical reactions, stepwise reduction of Se(IV) to Se(0) and to Se(-II) as well as the 6-electron pathway of Se(IV) to Se(-II). 18 Since Wave d in Figure 1 was found to be due to cobalt reduction, it is reasonable to attribute the other three Waves a∼c to the electrochemistry of Se(IV). To assign those waves, linear sweep cathodic voltammetry was combined with EQCM at a Pt electrode using 0.1 M Na 2 SO 4 containing only 10 mM SeO 2 . In Figure 2B, the first wave at ∼−0.57 was associated with mass gain and the wave at ∼−0.82 V was accompanied with mass decrease. As discussed in the literature, the wave at ∼−0.57 V is due to a combination of Reactions 1 and 2: 18 Reaction 2 can be combined with the chemical Reaction 3 to result in Se(0) deposition, which is manifest from the frequency decrease during the wave at −0.57 V.
Wave a in Figure 1 is now assigned to the deposition of Se(0) by Reaction 1 as well as Reaction 2 combined with Reaction 3. Another wave at ∼−0.82 V wave is due to the reduction of Se(0) to selenide via Reaction 4, which is manifest from the mass decrease and cathodic current: A slight mass increase was observed at ∼−1.2 V due to the Se(0) re-deposition via Reaction 3 and Reaction 4. At the end of a cathodic scan, the EQCM frequency does not return to the initial value, which implies incomplete stripping of Se(0). This trend is ascribed to the presence of Se(IV) in the electrolyte. When the electrolyte contained no Se(IV), complete stripping of Se was achieved at the end of a cathodic scan (see below). From the potential and frequency change, Wave c in Figure 1 is now assigned to the reduction of Se(0) to selenide. Another evidence that Wave c in Figure 1 is due to reduction of Se(0) is contained in Figure 2C.
The stripping voltammogram along with frequency change was obtained using a Se-modified electrode in 0.1 M Na 2 SO 4 blank electrolyte. The Se-modified electrode was prepared at the potential of −0.6 V in 0.1 M Na 2 SO 4 containing 10 mM SeO 2 . During the electrodeposition, a potential of −0.6 V was held until the EQCM fre- quency changed to ∼700 Hz. The stripping peak at ∼−0.80 accompanied with monotonic mass decrease can be assigned to the reduction of Se(0) to selenide. Unlike the scan in Figure 2B, the frequency returned to the initial value at the scan termination, diagnosing complete stripping of the electrodeposited selenium layers. [This contrast with the trend above.] Therefore, the waves at ∼−0.8 V in Figures 1 (wave  c), 2B and 2C all originated from the electroreduction of selenium to selenide (Reaction 4). While Waves a, c and d in Figure 1 are properly assigned from the data in Figure 2, Wave b was not observed in voltammograms obtained using a bare Pt electrode in electrolytes containing cobalt or selenium species as well as for a Se-modified electrode in a blank electrolyte. From previous experiences with the electrodeposition and stripping analysis of CdSe, 19,20 CdTe 21 and Bi 2 Te 3 , 22 it can be speculated that electrodeposited cobalt selenide films may contain free Se, free Co and CoSe. In addition, the reduction of Se to Se(-II) and CoSe to Co + Se(-II) can be assumed to occur at different potentials. To find out the possibility of Wave b in Figure 1 stemming from the reduction of CoSe to Co + Se 2− , cobalt selenide films were prepared at −0.7 V in 0.1 M Na 2 SO 4 electrolyte containing 5 mM Co(CH 3 COO) 2 and 5 mM SeO 2 . Figure 3A shows a LSV scan accompanied with mass change for as-prepared cobalt selenide films in 0.1 M Na 2 SO 4 . The first peak at ∼−0.73 V is again assignable to the reduction of Se(0) to Se 2− ; therefore, the second peak at ∼−1.03 V can be assigned to the reduction of CoSe to Co + Se 2− . Co-deposited free selenium can be cathodically stripped at −0.8 V using a blank electrolyte. A stripping potential of −0.8 V was selected based on the data in Figure 2C. After complete stripping of free selenium, the peak at −0.73 V (observed in Figure 3A and attributable to Se reduction) now disappeared from the linear sweep voltammogram ( Figure 3B). Thus Figure 3B clearly shows that the voltammteric signature from free Se is absent.
Further evidence to support the above assignment of the ∼−1.03 V peak to the reduction of CoSe to Co + Se 2− can be obtained from measuring the number of electrons transferred (n). The n value is calculated from the slope of the charge-frequency plot by combining the Sauerbrey equation and Faraday's law. 22,23 The n value was found to be ∼2, which corroborates the notion that the peak at ∼−1.03 V is due to the electroreduction of CoSe to Co + Se 2− . From the data in Figure 3, it is confirmed that Wave b in Figure 1 is not due to the reduction of CoSe to Co + Se 2− .
In order to probe the origin of Wave b in Figure 1, a Pt electrode was modified with a CoSe film by scanning the electrode potential to −0.6 V in 0.1 M Na 2 SO 4 electrolyte containing 5 mM Co(CH 3 COO) 2 and 5 mM SeO 2 . Then, a LSV scan with EQCM was obtained at the Pt-EQCM electrode modified with cobalt selenide film in 0.1 M Na 2 SO 4 containing 5 mM Co(CH 3 COO) 2 ( Figure 4A). As shown in the figure, a single major cathodic wave at ∼−0.89 V was accompanied with monotonic mass increase, which pointed to cobalt deposition.
An interesting feature was observed in the voltammogram obtained at the same electrode in 0.1 M Na 2 SO 4 containing 5 mM SeO 2 ( Figure  4B). The voltammogram showed current increase with mass increase, unlike the voltammogram obtained with bare Pt electrode in 0.1 M Na 2 SO 4 containing 10 mM SeO 2 ( Figure 2B). The cathodic wave at ∼−0.68 V with mass decrease is clearly due to the reduction of newlydeposited selenium on the surface to selenide during the cathodic scan. This is manifest from the difference in stripping potential of selenium in Figure 2B (∼−0.82 V, selenium to selenide) and in Figure 3A (∼−1.03 V, cobalt selenide to cobalt and selenide). To enhance the wave prominence, selenium layers were deposited by applying −0.1 V at the cobalt selenide film-modified electrode in 0.1 M Na 2 SO 4 containing 5 mM SeO 2 . Selenium stripping in 0.1 M Na 2 SO 4 blank electrolyte now resulted in an amplified wave signal at ∼−0.63 V with mass decrease ( Figure 4C). In other words, the waves at ∼−0.68 V in Figure 4B and at ∼−0.63 V in Figure 4C are both due to the reduction of selenium to selenide on the cobalt selenide film surface. The other waves at more negative potentials in Figure 4C are due to the electroreduction of free selenium to selenide and CoSe to Co + Se 2− , respectively.
To further examine whether Wave b was due to the reduction of selenium on the Se or CoSe surface, a Se-modified Pt electrode was prepared at −0.5 V in 0.1 M Na 2 SO 4 containing 5 mM SeO 2 . Since the cobalt selenide films used in Figure 4 can be composed of free Se and CoSe, a CoSe-modified electrode was also prepared at −0.7 V in 0.1 M Na 2 SO 4 electrolyte containing 5 mM Co(CH 3 COO) 2 and 5 mM SeO 2 , followed by complete stripping of selenium at −0.8 V in 0.1 M Na 2 SO 4 blank electrolyte. As shown in Figure 5A, Se reduction on the Se layers revealed two cathodic waves at ∼−0.54 and ∼−0.88 V due to Reactions 1 and 4, respectively, which are slightly different from Wave b in Figure 1.
The LSV scan morphology and frequency changes in Figure 5A are very similar to those in Figure 2B. Finally, another voltammogram accompanied with mass changes was obtained for the CoSe-modified electrode in 0.1 M Na 2 SO 4 containing 5 mM SeO 2 ( Figure 5B). Current and mass increase from the beginning of the cathodic scan are very similar to the features in Figure 4B. In addition, one distinct wave at ∼−0.64 V is well matched with Wave b in Figure 1, which clearly proved that this wave is assignable to reduction of newly-deposited selenium (to selenide) on the CoSe surface. Summarizing the data in Figures 2-5, Waves a, c, d are attributable to reduction of Se(IV) to Se(0), Se(0) to Se(-II), and Co 2+ to Co, respectively on the Pt surface, while Wave b arises from the further reduction of deposited Se(0) to Se (-II) on the CoSe surface.

Electrodeposition Mechanism and Film Compositional Aspects
As discussed earlier, Wave a in Figure 1 involves the underpotential deposition of selenium (Reaction 1). However, as in the case of many metal chalcogenides (e.g., CdSe), induced reduction of the metal ion, i.e., cobalt ions in this case, is also involved here resulting in the formation of cobalt selenide (Reaction 5): The reduction current in 0.1 M Na 2 SO 4 containing 5 mM Co(CH 3 COO) 2 and 5 mM SeO 2 was larger than that in 0.1 M Na 2 SO 4 electrolyte containing 5 mM SeO 2 , pointing to the involvement of cobalt reduction via Reaction 5 and/or Reaction 6: 16 Electrodeposition at −0.5 V in 0.1 M Na 2 SO 4 solutions containing 5 mM Co(CH 3 COO) 2 and 5 mM SeO 2 (peak potential of wave a) resulted in films containing free selenium as well as cobalt selenide, even though the deposition potential was too positive for cobalt deposition. Clearly, underpotential deposition of Co is involved here, and is driven by the free energy of compound formation with the pre-deposited Se. Figure 6A shows the composition of cobalt selenide films, electrodeposited at −0.5 V using 0.1 M Na 2 SO 4 containing 5 mM Co(CH 3 COO) 2 and 5 mM SeO 2 . The combined LSV-EQCM data for the 0.1 M Na 2 SO 4 blank electrolyte clearly demonstrated that electrodeposited CoSe films at −0.5 V contained free Se (wave at ∼−0.78 V) and CoSe (wave at ∼−1.04 V. At a deposition potential of −0.7 V (a potential beyond Waves b and c in Figure 1), CoSe was electrosynthesized with free selenium via Reactions 3 and 6. This was confirmed by reduction waves in the voltammogram with mass changes for the electrodeposited films in 0.1 M Na 2 SO 4 blank electrolyte ( Figure 6B). Two reduction waves at ∼−0.72 V and ∼−1.03 V are due to reductions of selenium to selenide and CoSe to Co + Se 2− , respectively.
Interestingly enough, these electrodeposited cobalt selenide films contained no free cobalt regardless of the deposition potential as long as films were in touch with the deposition electrolyte for enough time. This trend is attributable to the instability of cobalt films in acidic electrolytes and the spontaneous corrosion of cobalt by more noble species such as Se(IV). Figure 7A shows the stability of cobalt films in contact with 0.1 M Na 2 SO 4 electrolyte whose pH was adjusted to that of the electrodeposition solution case. As shown in the figure, the electrodeposited cobalt film gradually dissolved and frequency returned to the original value within 5 min. On the other hand, when the cobalt film was in the electroplating solution, cobalt dissolution was complete in less than 3 min demonstrating the effect of Se(IV) species. Considering standard reduction potentials, Se(IV) can oxidize cobalt to Co 2+ : 24 Therefore, cobalt films are more easily oxidized in the deposition solution than in a blank solution with the same pH.
In situ composition analysis of electrodeposited cobalt selenide films revealed that free cobalt was removed immediately after electrodeposition as shown in Figure 7B. In the figure, the initial mass gain was due to film deposition at −0.7 V. After deposition, the film was in the deposition solution without applied bias potential (i.e., at open circuit) and spontaneous mass loss was observed for the reasons discussed above (see also Figure 7A). Next, the solution was changed to 0.1 M Na 2 SO 4 blank electrolyte and free selenium was cathodically stripped at −0.8 V, followed by the stripping of Se in CoSe and Co in CoSe at −1.3 V and +2.0 V, respectively. The content of free Se in the electrosynthesized film was determined by the frequency changes at −0.8 V. The EQCM frequency changes at −1.3 V (Se in CoSe, 900 Hz) and +2.0 V (700 Hz, Co in CoSe) were in good agreement with the atomic mass ratio (Co: Se = 1: 1.3). Clearly, these electrodeposited films contained no free cobalt and the compound stoichiometry was close to 1: 1 as expected for CoSe.  Figure 6B. Between the electrodeposition and stripping steps, the electrodeposited films were left in the deposition bath under no applied bias potential. Stripping was performed in 0.1 M Na 2 SO 4 electrolyte at different potentials. Figure 8 displays the XRD pattern obtained from the electrodeposited CoSe films on a SnO 2 coated glass substrate. One major peak at 31.94 • clearly confirms the presence of CoSe. Other peaks are from the underlying FTO electrode support. No other peaks from metallic cobalt and selenium were observed. Figure 9 contains a scanning electron micrograph of the electrodeposited CoSe film. The films were prepared by the procedure described in Figure 8. The as-deposited films showed a smooth and uniform surface with well-defined grain boundaries and the electrode surface was well covered with cobalt selenide films. The side view clearly showed well-distinguished layers comprising of the top CoSe layer, the (∼700 nm), SnO 2 substrate layer, and the bottom glass support.

Conclusions
This study demonstrated the application of combined linear sweep voltammetry and electrochemical quartz crystal microgravimetry for studying the CoSe electrodeposition system. Electrodeposition involved selenium deposition first followed by the induced (underpotential) deposition of cobalt to result in the formation of CoSe. On the other hand, CoSe can also be electrodeposited by the reaction  of Co 2+ with electroreduced Se 2− at more negative potentials. The electrochemical behavior of selenium on the Pt surface was different from that at the pre-deposited CoSe film surface. The electrodeposited cobalt selenide films contained no free cobalt due to its instability in acidic electrolytes and due to galvanic corrosion instigated by the more noble Se(IV) species. A method for in situ compositional anal-ysis of electrosynthesized cobalt selenide films using combined stripping voltammetry and EQCM was finally demonstrated. A newly developed method for compositional analysis was elaborated.