Introduction

Alkaline fuel cells (AFCs) and proton exchange membrane fuel cells (PEMFCs) have been extensively studied amongst other fuel cell types due to their potential applications in portable electronic and automobile industry. AFCs also have several advantages over PEMFCs on both faster cathode kinetics and Ohmic polarisation. Moreover, the alkaline environment, which is less corrosive, ensures a longer durability and the use of non-noble cheap oxygen reduction reaction electro-catalysts cuts down the costs of the fuel cell drastically. Therefore, AFCs can theoretically outperform PEMFCs. A major limitation of AFCs however is the progressive carbonation of the alkaline electrolyte due to CO2 formation from air or the oxidation product of the fuel. This problem was addressed mainly by the application of anion exchange membranes (AEMs) [1,2,3,4,5,6]. Over the past decade, alkaline direct alcohol fuel cells (ADAFCs) have attracted increasing interest due to the favourable reaction kinetics in alkaline media, easy handling of the liquid fuels and the higher energy densities achievable. ADAFCs using AEMs have several advantages over conventional alkaline fuel cells, such as (1) no electrolyte leaking, (2) since there is no mobile cation, there is no precipitated carbonate, (3) reduced alcohol crossover, (4) potential simplified water management due to the fact that water is produced at the anode and consumed at the cathode and (5) potentially reduced corrosion [1, 7].

The oxidation of alcohols is necessary for energy production and this requires breaking the C-C bonds which is difficult to do [8, 9]. The main catalyst used in the electro-oxidation of alcohols is Pt catalysts, and this has been done in acidic media. The extensive use of Pt catalysts is not economically viable due to Pt cost, the slow kinetics of the reaction, and also the CO poisoning of the catalyst [10, 11]. Pd-based catalysts in basic media are great alternatives. Pd and Pt have similar chemistries (same group elements). The electro-oxidation reaction of ethanol using Pd in alkaline media is limited by slow kinetics. The rate is limited by the removal of adsorbed acetyl species by reacting with hydroxyl species forming acetates [8, 10]. A minor quantity of the acetyl species goes towards decomposition to form hydrocarbons and carbon monoxide further oxidised to carbon dioxide.

Pd-based catalysts have been investigated for the alcohol oxidation reaction (AOR) in alkaline media, and further research still needs to be done to improve activity, selectivity and stability of the catalysts [6, 11,12,13,14,15,16]. This can be done by formation of nanosized catalysts for high surface area, and the addition of other elements to Pd to improve catalytic activity. Electrochemical atomic layer deposition (E-ALD) allows the formation of well ordered, tailor made, and thin films material using monolayer formation on well-ordered surfaces using underpotential deposition and surface-limited redox reactions [17]. The formation of Pd nanostructures on Au substrates is well covered in literature [18,19,20,21,22,23].

Ethanol is attracting great attention as fuel since it is less toxic than methanol and can be massively produced from agricultural products or biomass, in addition to the advantage of high-specific energy [24]. Hence, tremendous efforts have been made to the development of alkaline direct ethanol fuel cells (ADEFCs). Ethanol has also been proven by research to have a lower crossover rate and affects the cathode performance less severely than methanol [25]. This study aims to fabricate PdSn, PdBi and PdBiSn thin films on Au substrate using E-ALD technique. Pd was prepared via surface limited replacement reaction (SLRR) of underpotential (UPD) deposited Cu on Au electrode. Pd containing Sn and or Bi was deposited using SLRR of Cu UPD followed by direct UPD of Sn or Bi. The deposition cycle used was developed from the study of the reduction and oxidation of each element using cyclic voltammetry. The formed material was characterised using field emission scanning electron microscope (FESEM) for morphology equipped with energy dispersive detector (EDS) for the elemental analysis. The activity of each material towards the oxidation of ethanol and glycerol was investigated using cyclic voltammetry.

Experimental Procedure

Materials and Reagents

High purity (≤ 1 μS/cm) deionised water and ACS Reagent grade chemicals (Sigma-Aldrich) were used in the preparation of all solutions. The chemical solutions prepared were as follows: 0.1 mM PdCl2 in 0.5 M HCl, 1 mM CuSO4 in 0.1 M H2SO4, 0.2 mM SnO in 0.1 M H2SO4 and 0.2 mM Bi2O3 in 0.1 M H2SO4.

Substrate Treatment

Au-coated glass slides containing 100 nm Au film on 5 nm Ti on glass (evaporated metal films, Ithaca, NY) were used as substrate. Prior use, they were cleaned by sonication in acetone for 5 min, deionised water rinse, dipped in concentrated HNO3 for 2 min, deionised water rinse and finally dried with nitrogen.

E-ALD Deposition

The substrate was placed into the electrochemical flow cell and immediately further cleaned by flushing with 0.1 M H2SO4 while the potential of the cell was alternated three times between 1400 mV and − 200 mV. A three-electrode electrochemical flow cell was used during the deposition. Au wire and Ag/AgCl (3 M KCl) (MF-2021, BASI) served as auxiliary and reference electrodes, respectively. Copper tape was used as mode of contact for the working electrode (Au-coated glass). The solutions were degassed with nitrogen gas for at least 60 min and then pumped through master flex tubes from their reservoirs using a peristaltic pump. A rubber gasket was used to limit the surface of the Au electrode to 2.1 cm2. The deposition conditions were studied and optimised using cyclic voltammetry and analysis of current-potential time plots.

Electrodeposition of Pd-Based Materials on Au Substrate

The electrodeposition of Pd-based materials was done using Cu UPD on Au substrate was adapted from the work done by Mkwizu and co-workers [26], and Modibedi et al. [12], as well as other sources [27, 28]. It was carried out as follows: (i) pump the Cu solution into the cell at an applied potential (Eapp), (ii) keep the Cu solution in the flow cell to deposit Cu at Eapp, (iii) pump Pd solution into the cell at open circuit potential (Eocp), (iv) keep Pd solution in the cell for deposit Pd (SLRR) at Eocp and (v) rinse with electrolyte at Eocp. This cycle was repeated as many times as required to form the deposit. Pd materials containing Sn and Bi were formed by introducing a second element in the following steps and depositing it at UPD as described below.

Electrodeposition of Sn and Bi on Au Substrate

The deposition cycle that was used to form a layer of Bi or Sn is described. A deposit of Bi involved these steps: (i) flow Bi3+ solution at − 50 mV for 10 s, (ii) deposit Bi at − 150 mV for 5 s and (iii) rinse in 0.1 M H2SO4 at − 150 mV. A similar cycle was used to deposit Sn as follows: (i) flow Sn2+ solution at − 50 mV for 15 s, (ii) deposit Sn at − 425 mV for 5 s and (iii) rinse in 0.1 M H2SO4 at − 425 mV for 15 s. Sn or Bi was deposited first or followed a layer of electrodeposited Pd, and the formed deposits are listed in Table 1.

Table 1 List of Pd-based materials catalyst deposition description charge for the prepared catalysts in 0.1 M KOH
Full size table

Characterisation

The electrochemical activity of the prepared Pd-based catalysts was evaluated using an electrochemical flow cell with three electrode system as described in literature [12, 27, 29]. Cyclic and linear sweep voltammetry in N2 saturated 0.1 M KOH, and 0.1 M EtOH on a potentiostat Metrohm PGSTAT101 controlled by NOVA 1.8 software was conducted. The morphology of the materials was evaluated using a scanning electron microscope performed on a LEO 1525 field emission scanning microscope (FE-SEM) with the acceleration voltage of 2.00 kV, and equipped with dispersive detector (EDS) for elemental mapping.

Results and Discussion

Electrodeposition of Various Pd on Au Catalysts

The characteristic cyclic voltammogram (CV) of 1 mM CuSO4 in 0.1 M H2SO4 on Au electrode is shown in Fig. 1a. The scan began at 0.35 V negatively until a lower limit of − 0.35 V and reversed. The negative scan shows a bulk deposition peak onset at − 0.025 V and the corresponding bulk stripping peak at 0.05 V. The insert graph A shows the UPD region where it was found to occur between 0.2 and 0 V. This is consistent to the values reported in literature [27, 28, 30]. The obtained Cu UPD (Fig. 1b) was used to form a Pd deposition cycle, where Pd was deposited at a set potential, followed by replacement with Pd at open circuit potential. The deposition cycle was made up of the following steps: (i) pump Cu2+ solution for 15 s at OCP; (ii) deposit Cu at 50 mV for 10 s at 150 mV; (iii) pump Pd2+ solution for 15 s at OCP; (v) hold the solution at OCP for 10 s and (iv) rinse in blank for 15 s at OCP.

Fig. 1
figure 1

a Cyclic voltammogram of 1 mM CuSO4 in 0.1 M H2SO4 on Au at a scan rate of 10 mV/s, and an insert of the window opening. b Cyclic voltammogram of 1 mM CuSO4 in 0.1 M H2SO4 on Bi covered Au, and Sn covered Au, at a scan rate of 10 mV/s. c Cyclic voltammograms of Au electrode in 0.2 mM Bi2O3 vs Ag/AgCl, and an insert of Au electrode in 0.1 M H2SO4, at a scan rate of 10 mV/s. d Cyclic voltammograms of Au electrode in 0.2 mM SnO vs Ag/AgCl, at a scan rate of 10 mV/s

Full size image

The cycle was repeated 25 and 50 times for film growth. The formation of Pd-based catalysts containing Sn and Bi will be done by building from the above deposition cycle. The cyclic voltammograms of Bi3+ and Sn2+ are shown in Fig. 1c, d respectively with an insert of Au electrode CV, where the UPD of Bi3+ and Sn2+ was found to be at 50 mV and − 325 mV respectively. A characteristic cyclic voltammogram of the Au electrode in 0.1 M H2SO4 negative scan shows reduction peak corresponding to the reduction of Au2O3 to Au (peak A) [31] and turned at − 200 mV before hydrogen evolution peak could appear. The reverse scan continued until a potential of 1400 mV was reached. This results in the appearance of the oxidation peaks, indicating the oxidation of surface Au atoms from different crystallographic planes (peaks B). The window opening cyclic voltammetry study of Bi deposition on Au in 0.2 mM Bi3+ solution is shown in Fig. 1c. All CV curves were scanned negatively and then reversed. There are three oxidation peaks observed at approximately 0 mV, 150 mV and 275 mV (peaks c, d and e). The area of peak at 0 mV increased when scanned more negatively, indicating bulk Bi deposition. The latter two peaks had the same peak height regardless of increasing the negative potential. These peaks are representative of the UPD oxidation peaks. UPD processes are not dependent on potential change because they are surface limited reactions. The study of Sn deposition on Au in 0.2 mM Sn2+ solution is shown in the window opening CV shown in Fig. 1d. The potential was scanned from open circuit potential and held at different negative limits for 30 s to force the stripping peak of Sn to appear. As a result, no reductive regions were observed on the scan, only two oxidative features were visible at − 200 mV and 0 mV (peaks A and B). The onset UPD potential was found to be at − 350 mV which slowly grew to bulk deposition at − 425 mV and beyond.

The information obtained from the CV studies was used to deduce a deposition cycle for catalysts materials containing Pd, Sn and Bi. Materials made up of PdSn, PdBi and PdBiSn were made using the above deposition cycle followed by Sn or Bi UPD or both. The binary catalyst PdSn deposition cycle steps were as follows: (i) pump Cu2+ solution for 15 s at OCP; (ii) deposit Cu for 10 s at 50 mV; (iii) pump Pd2+ solution for 15 s at OCP; (v) hold the solution at OCP for 10 s; (iv) rinse in blank for 15 s at OCP; (vi) pump Sn2+ solution for 15 s at OCP; (vii) deposit Sn at for 10 s at − 325 mV and (viii) rinse in blank for 15 s at OCP. A similar deposition cycle was used for PdBi by replacing Sn UPD with Bi UPD of 50 mV. The formation of a material by alternating Pd SLRR followed by Bi and Sn at UPD was unable to form a deposit, due to Sn and Bi UPD being too far apart. The summary of the deposition sequence for each material is given in Table 1.

This resulted in the ternary material deposition cycles being designed to avoid this by depositing Pd and Bi first, followed by depositing Pd and Sn, using the PdBi cycle and then the PdSn cycle. Two materials were made namely PdBiSn 25 and PdBiSn 30. The ternary PdBiSn 30 was deposited using a sequence made up of the PdBi cycle repeated 15 times followed by the PdSn cycle repeated 15 times. The PdBiSn 25 was made by depositing PdBi 13 time and PdSn 12 times. The possibility of depositing Sn or Bi on top of the Pd layer to form PdSn or PdBi was explored. A CV of Cu2+ UPD on Bi and Sn deposited on Au monolayer (Fig. 1b) showed that the Cu2+ UPD has not changed. This resulted in the deposition cycle for SnPd and BiPd being the reverse of PdSn and PdBi. A visible deposit for BiPd was not able to form.

Structural Characterisation

The morphology of the formed deposits was investigated using SEM. The Pd deposits were found to have rounded particles clustered all over the surface; in some areas, they were found to agglomerate as illustrated in Fig. 2a, b, and the deposit was found to contain Pd distributed all over the surface as shown on the map (Fig. 2c–e) and spectra (Fig. 2f). Pd deposits form larger structures on the surface and this is prominent when the number of cycles was increased or for thicker deposits as shown in Fig. 3a, b. PdSn and SnPd deposits (Fig. 3c, d) were found to have similar characteristics to each other, where smaller particles are dominant on the surface and few larger randomly distributed particles. Elemental map did not reveal any differences in particle composition. The PdBi deposits (Fig. 3d) were found to contain well-rounded larger particles on the surface distributed evenly as well were smaller particles.

Fig. 2
figure 2

SEM images of a, b Pd/Au 25 at different magnification, c Pd catalysts overlaid elemental map, d Pd elemental map, e Au elemental map and f EDS spectrum

Full size image
Fig. 3
figure 3

SEM images of a Pd 25, b Pd 50, c PdSn, d SnPd, e PdBi and f PdBiSn 25

Full size image

The elemental map indicated that the surface was evenly covered with Pd and Bi. PdBiSn 25 was also similar structurally to the other catalysts (Fig. 3f), and the elemental map of PdBiSn 25 in Fig. 4a–d and spectra (Fig. 4e) shows the presence of Pd, Sn and Bi on the surface directly on top of each other all over the surface. When comparing all the materials, only the PdBi materials exhibited a different morphology to the rest of the prepared materials.

Fig. 4
figure 4

Elemental map and EDS spectra of PdBiSn 25 catalyst showing: a Overlaid image; b Pd; c Sn; d Bi

Full size image

Electrochemical Characterisation

The cyclic voltammograms of the Pd and Pd-modified catalysts in 0.1 M KOH solution are presented in Fig. 5a, b. The peaks found below − 0.8 V onwards are due hydrogen absorption and adsorption on Pd surface. Au is not prone to hydrogen evolution reactions [11] hence no peaks were observed for Au. The oxidation peaks found at 0 V and above is due to the oxidation of Au oxides. The reduction peak observed in the region of 0 V towards − 0.3 V is caused by the reduction of the Pd oxides. This peak is different for all the catalysts, indicating that the modified catalysts have different electrochemical activity. The Pd on Au catalyst exhibited the most pronounced peaks in the hydrogen evolution region (− 0.8 V and below), and a higher charging current. The CV shows the signature of Pd in alkaline electrolyte [32]. The SnPd and PdBiSn 25 catalyst had the exact same interaction with KOH, as indicated by their cyclic voltammograms matching. The reduction peaks for all the catalysts occurred at different potentials implying that all the material has a different Pd dispersion on the surface. The current associated with this reduction peak is directly proportional to the charge required to reduce Pd oxides as shown in Eq. 1 used to calculate the electrochemical active surface (EAS) of the catalyst [11, 32]. The charge obtained by integrating the area under the Pd reduction peak is shown in Table 1. The results shows that BiPd has the highest charge, followed by PdBi, the other catalysts had charge values lower than that of the Pd catalyst. Therefore, the BiPd and PdBi catalysts will have a higher EAS compared to the other catalysts; the presence of Bi improved the exposed surface area of the Pd catalysts.

$$ \mathrm{EAS}\left({\mathrm{m}}^2{\mathrm{g}}_{\mathrm{Pd}}^{-1}\right)=\frac{{\mathrm{Q}}_{\mathrm{r}}}{\left(4.24\ {\mathrm{cm}}^{-2}\times {\mathrm{W}}_{\mathrm{Pd}}\right)\mathrm{Ag}\left({\mathrm{cm}}^2\right)} $$
(1)
Fig.5
figure 5

a Cyclic voltammogram of Pd, PdBi, PdSn and PdBiSn catalysts, b cyclic voltammogram of Pd, BiPd, SnPd and PdBiSn 25 catalysts in 0.1 M KOH at a scan rate of 10 mV/s

Full size image
QR:

Reduction charge

WPd:

Pd loading mass in grams

Ag:

Geometry of the surface area

Ethanol Electro-Oxidation

The activity of the prepared catalysts towards ethanol electro-oxidation reaction (EOR) was investigated using cyclic voltammetry in 0.1 M KOH solution and 0.1 M EtOH and is shown in Fig. 6 and the summary of the results shown in Table 2. The forward scan of the CV curves for all the catalysts had lower peak current than the reverse scans except for the PdBi catalysts. The BiPd catalysts showed no activity towards ethanol oxidation and therefore excluded for further studies due to Bi removal on the surface during Pd deposition. Pd catalyst exhibited the lowest peak current and broader peaks compared to all the other catalysts for the forward reaction, indicating a slow consumption of EtOH on the catalyst [11, 33]. The highest peak currents were observed from the SnPd and SnBiPd 25 catalysts for the reverse CV scan, these two materials showed the same activity for EOR, and also had the same interaction with KOH. The sharpness and rapid increase in peak currents is due to the fast removal of the adsorbed acetate intermediate species that were partially oxidised during the forward reaction. This is the rate-limiting step for EOR as shown in Eqs. 2–5 where Eq. 5 shows the rate-limiting step [8, 11].

$$ \mathrm{Pd}+{\mathrm{OH}}^{-}\to \mathrm{Pd}-{\mathrm{OH}}_{\mathrm{ads}}+{e}^{-} $$
(2)
$$ \mathrm{Pd}+{\mathrm{CH}}_3{\mathrm{CH}}_2\mathrm{OH}\to \mathrm{Pd}-{\left({\mathrm{CH}}_3{\mathrm{CH}}_2\mathrm{OH}\right)}_{\mathrm{ads}} $$
(3)
$$ \mathrm{Pd}-{\left({\mathrm{CH}}_3{\mathrm{CH}}_2\mathrm{O}\mathrm{H}\right)}_{\mathrm{ads}}+3{\mathrm{OH}}^{-}\to \mathrm{Pd}-{\left({\mathrm{CH}}_3\mathrm{CO}\right)}_{\mathrm{ads}}+3{\mathrm{H}}_2\mathrm{O}+3{\mathrm{e}}^{-} $$
(4)
$$ \mathrm{Pd}-{\left({\mathrm{CH}}_3\mathrm{CO}\right)}_{\mathrm{ads}}+\mathrm{Pd}-{\mathrm{OH}}_{\mathrm{ads}}+{\mathrm{OH}}^{-}\to {\mathrm{CH}}_3{\mathrm{COO}}^{-}+2\mathrm{Pd}+{\mathrm{H}}_2\mathrm{O} $$
(5)
Fig.6
figure 6

a Cyclic voltammogram of Pd, PdBi, PdSn and PdBiSn catalysts; b cyclic voltammogram of Pd, SnPd and PdBiSn 25 catalysts; c cyclic voltammograms for the ethanol comparison of PdSn and SnPd catalysts; d chronoamperometric curves at − 0.2 V of ethanol electro-oxidation in 0.1 M KOH + 0.1 M EtOH. (Scan rate of 10 mV/s was used)

Full size image
Table 2 Summary of ethanol electro-oxidation parameters for the different catalysts
Full size table

The CVs in Fig. 6a, b show the onset potential for the EOR reaction for the different catalysts. The onset potential for Pd, PdBi, PdSn, SnPd, PdBiSn 25 and PdBiSn 30 catalysts is − 0.37 V, − 0.46 V, − 0.46 V, − 0.50 V, − 0.50 V and − 0.53 V respectively. These results show that modifying Pd with either Sn or Bi improved catalytic activity due to the more negative onset of EOR on the binary and ternary catalysts compared to Pd catalyst; this is in agreement with literature [34, 35]. The similarities between PdBi and PdSn activities could be due to the similar chemical properties of Sn and Bi. While the PdBiSn 25 catalyst showed higher activity due to the presence of both Sn and Bi. The earliest onset potential was found on PbBiSn 30 material as expected because this material contains more Pd (30× deposition cycles) than the others (25× deposition cycles).

The PdSn and SnPd catalysts in Fig. 6c showed that SnPd catalysts had the highest Iforward and Ireverse compared to PdSn, as well as the sharper Ireverse peak, indicating that it removes the adsorbed carbonate species much faster than PdSn does as discussed in literature [36]. PdBiSn 30 catalyst had a slower ethanol oxidation due to broader peak for forward scan. The rapid drop of current observed for PdBiSn 25 indicates a faster EOR, accompanied by a sharp increase for the reverse scan indicating faster by product removal. The ratio of the reverse current and forward current (Iforward/Ireverse) is shown in Table 2,with the highest obtained from PdBi catalyst compared to all the other catalysts, and this was the better catalyst because it was able to achieve good activity and the least susceptible to poisoning. The highest value of 1.25 was obtained for PdBi catalyst followed by PdBiSn 30 with 0.74.

Chronoamperometry scans illustrating the stability of the catalysts in ethanol at − 0.2 V is shown in Fig. 6d. The initial high current decreased rapidly within the first 60 s, followed by a gradual decrease over time as the catalysts get poisoned. The highest current was obtained for PdBiSn 30 catalyst over 60 min. The order of the catalysts stability based on current was found to be PdBiSn 30>PdSn>PdBiSn 25>SnPd>PdBi>Pd. Pd catalyst was found to be the most susceptible towards poisoning, and the addition of both Sn and Bi improved the stability of the catalyst, which supports the bifunctional mechanism stated in literature [36, 37].

Conclusion

Pd-based binary and ternary catalysts were prepared using underpotential deposition method and tested for the electro-oxidation of ethanol in alkaline electrolyte. Sn and Bi were used as second or/and third metal to form binary and ternary Pd-based catalysts. Pd catalysts were deposited using Cu SLRR method while Sn and Bi were directly deposited at UPD. These materials were found to exhibit similar morphology with the exception of PdBi material. The addition of Sn and Bi to Pd was found to improve activity of the catalysts for ethanol oxidation, and also makes these catalysts least susceptible to poisoning. The addition of both Sn and Bi improved the catalysts stability in ethanol electro-oxidation. This work demonstrated that the addition of Sn and Bi on Pd catalyst can effectively lower the cost of fuel cell catalysts without compromising catalyst performance.