Electrochemical Performance and in Operando Charge Efﬁciency Measurements of Cu/Sn-Doped Nano Iron Electrodes

: Fe-air or Ni-Fe cells can offer low-cost and large-scale sustainable energy storage. At present, they are limited by low coulombic efﬁciency, low active material use, and poor rate capability. To overcome these challenges, two types of nanostructured doped iron materials were investigated: (1) copper and tin doped iron (CuSn); and (2) tin doped iron (Sn). Single-wall carbon nanotube (SWCNT) was added to the electrode and LiOH to the electrolyte. In the 2 wt. % Cu + 2 wt. % Sn sample, the addition of SWCNT increased the discharge capacity from 430 to 475 mAh g − 1 , and charge efﬁciency increased from 83% to 93.5%. With the addition of both SWCNT and LiOH, the charge efﬁciency and discharge capacity improved to 91% and 603 mAh g − 1 , respectively. Meanwhile, the 4 wt. % Sn substituted sample performance is not on par with the 2 wt. % Cu + 2 wt. % Sn sample. The dopant elements (Cu and Sn) and additives (SWCNT and LiOH) have a major impact on the electrode performance. To understand the relation between hydrogen evolution and charge current density, we have used in operando charging measurements combined with mass spectrometry to quantify the evolved hydrogen. The electrodes that were subjected to prolonged overcharge upon hydrogen evolution failed rapidly. This insight could help in the development of better charging schemes for the iron electrodes.


Introduction
Consumption of energy produced through fossil fuels is considered to be one of the major contributors to global warming [1]. To achieve a sustainable energy system, many nations have initiated subsidies to add wind and solar power generation to the electricity grid. The intermittent character of these renewable energy sources requires, however, large-scale energy storage systems to handle the load fluctuations. At present, only 1% of the total energy produced is stored by the pumped hydroelectric storage plants, which offers the lowest cost among the available storage solutions today ($180-200/kWh), and accounts 98% of the installed storage systems [2,3]. On the other hand, Fe + 2OH -↔ Fe(OH) 2  As a consequence of the self-discharge and hydrogen evolution, the iron electrode presents poor charge efficiency, low charge retention and low discharge capacity with decreased life cycle [12,13]. Additionally, the passive oxide and hydroxide phases gradually build up over time [12]. To overcome hydrogen evolution, sulfur additives are either added to the electrode formulation (e.g., FeS + Bi 2 O 3 , Bi 2 S 3 ) or to the electrolyte by addition of Na 2 S, K 2 S, and organosulfur compounds [14][15][16][17]. The adsorbed sulfur depassivates the iron electrode and thereby it increases the electrode conductivity, rate capability and further poisons the active catalytic surfaces for the hydrogen evolution. Previous studies have shown that bismuth as an additive suppresses the hydrogen evolution by increasing the overpotential [18][19][20].
To circumvent the low active usage of the material as well as passivation inducing capacity decay, nano iron, nano iron-carbon composites and FeS materials have been used as active materials [21][22][23][24][25][26][27][28][29]. Notably, iron with graphene, MWCNT, amorphous carbon composites have shown to be producing increased redox activity and high capacity. However, these materials have presented rapid capacity decay, which is caused by nanoparticle agglomeration and by increased rate of hydrogen evolution [29][30][31][32][33]. Sundar Rajan et al. [18] demonstrated that carbon grafted iron delivered 400 mAh g −1 of specific discharge capacity (discharge rate 100 mA g −1 ). C.-Y. Kao et al. [10] reported that in the presence of Cu a high discharge capacity (800 mAh g −1 ) at 3200 mA g −1 was achieved. The results confirm that Cu and amorphous carbon enhances the iron discharge performance. Our recent work on copper substituted core-shell structured iron/iron carbide/carbon active material has also shown to maintain stable discharge capacity (350 mAh g −1 ) at higher discharge rate 100-350 mA g −1 [34]. Chamoun et al. used stannate electrolyte additive, which resulted in high capacity at 400 mAh g −1 with 85% charge efficiency [35]. Here the thin layer of Sn deposits on the iron electrode is believed to suppress the H 2 evolution. Therefore, amorphous carbon, Cu, and Sn have the capability to enhance the iron electrode redox activity. In this work, we have used and studied nano/iron/carbon/Cu/Sn and nano/iron/carbon/Sn as two type of active materials for the iron electrode. We also investigated the effect of single-wall carbon nanotube (SWCNT) and LiOH on the iron electrode to maintain the performance of the charge and discharge capacities of these materials. This study aims to show that the particle agglomeration is prevented by the carbon layer, while Cu maintains the stable conductivity, whereas Sn suppresses the hydrogen evolution. We have achieved capacity as high as 600 mAh g −1 for the SWCNT with LiOH addition to the electrolyte. Furthermore, we investigated the effect of charge current density on the hydrogen evolution rate for different electrode formulations coupled with mass spectrometry. From this study, we have observed that high charge current density reduces the hydrogen evolution rate.
The powder samples were analyzed by X-ray diffraction (XRD) (Siemens Bruker D5000), equipped with a Co K α source (30 mA, 40 kV). The phase identification was made on the EVA software, and the powder diffraction file (PDF 2) was used to identify the plane reflection. The morphologies of the powder samples were analyzed by Zeiss Ultra-55 scanning electron microscope (SEM), and transmission electron microscope (TEM) image was obtained by JEOL JEM-2100F. X-ray photoelectron spectroscopy (XPS) was investigated in a PHI Quantum 2000 scanning XPS microprobe with Al Kα radiation as a source. The spot size was 0.1 × 0.1 µm 2 , and the energy resolution was set to 0.3 eV. The intensity of the ratio of the peaks were in the integrated area over the background. The fit-XPS software was used for peak positioning. The Raman measurements were carried out using an imaging spectrometer (iHR550, Horiba Jobin Yvon, Longjumeau cedex, France) in confocal geometry, using 532 nm laser light for excitation. Notch filter at ±50 cm −1 around 532 nm was used to filter the scattered light.
The electrodes were prepared by mixing the doped nano iron powder 80 wt. % with high surface area carbon black 5 wt. % (Ketjenblack EC-300J AkzoNobel), Bi 2 S 3 (5 wt. %) and PTFE 10 wt. % (60 wt. % water suspension). In the other two-electrode formulations, 0.1 wt. % SWCNT (TUBALL ™ SWCNT 0.2 wt. % in N-Methyl-2-pyrrolidone (NMP) solution) was added. The composite electrode in the form of paste was prepared by mixing the materials as mentioned above with D-70 Shellsol ® (Shell Chemicals, Europe) using an agitation rate of 6000 RPM (Waring ® laboratory blender LB20ES, Clarkson Laboratory & Supply Inc., CA, USA). Then, the filtered wet mass was rolled over to a defined thickness of 0.7 mm on a-nickel wire (100 mesh). The electrode was compressed (375 kg cm −2 ) and dried in N2 atmosphere and sintered at 325 • C for 30 min [36,37]. The tested electrodes are specified as shown in Table 1. The amount of iron in the electrodes was found to vary between 22-29 mg cm −2 . The electrodes in Table 1 were studied in a three-electrode cell compartment, in which a conventional sintered nickel electrode (Ni(OH) 2 /NiOOH) was used as a counter electrode, Hg/HgO as a reference electrode and the iron composite as a working electrode. In all the experiment 6 M KOH was used as electrolyte. In Li experiment, 6 M KOH + 0.65 M LiOH was used as electrolyte.
All the reported potential values were against Hg/HgO reference electrode immersed in the same electrolyte solution.
All the repeated charge (reduction to iron) and discharge (oxidation to iron hydroxide) measurements were done on the Land battery testing system (CT2001-5V2A BTS System). Charge and discharge currents were fixed at 200 and 100 mA g −1 , respectively, and the discharge potential was limited to -0.800 V vs. Hg/HgO. All the capacity and current details are based on the amount of iron in the composite. Hydrogen evolution and iron reduction potentials were measured at the end of each charging with the initial hump in the charging plateau. Then this was subtracted from the corresponding equilibrium potential value to get the overpotential.
The in operando charge efficiency measurements have been described in earlier work, where electrochemical water splitting catalysis coupled with mass spectrometry allowed the detection of gaseous products [38]. The cell consisted of two chambers and was separated by a nonwoven cellulose membrane (Freudenberg, 700/18F) and sealed with rubber gaskets (Kuntze, ESO2 425-010 EPDM). Each chamber with a volume of 48 cm 2 was constantly purged with argon gas through the inlets at the bottom of the chambers. From each chamber the gas outlets were joined in a single exhaust, which served as a gas sampling point for MS measurements (Pfeiffer, Thermostar GSD320-QMG220). The exhaust was then passed through a gas flow meter (Rota Yokogawa, RAK-D41). The iron electrode samples were evaluated with a potentiostat (BioLogic, SP-50) in a two-electrode setup with an oversized conventional nickel electrode as the counter electrode. The geometrical surface areas of the iron and nickel electrodes were 2.25 and 9 cm 2 , respectively. The electrodes were formed (30-35 cycles) prior to this experiment. All samples for in operando charge efficiencies were cycled three times with 5, 10 and 15 mA cm −2 of charge and 5 mA cm −2 for discharge. Figure 1 illustrates the crystal phases for the XRD patterns of the two samples: CuSn (a) and Sn (b). In both samples, α-Fe was the dominant phase. In the case of the Sn sample, Fe 3 SnC was also observed as a minor phase along with the α-Fe phase. However, when 2 wt. % Cu was introduced with 2 wt. % Sn, Fe 3 Sn phases were observed with traces of Fe 3 SnC. Metallic Cu and Sn phases were not detected due to the low concentrations of the respective elements. Even though both samples have more than 20 wt. % carbon, no graphitic phases are observed, which suggest that these carbons are of amorphous in nature, which may improve the electrode conductivity, whereas the porous network will also increase the ionic conductivity [39,40]. A further advantage of the carbon is that it will buffer the volume expansion and maintain stable electrical conductivity of the iron electrode during the electrochemical cycling [41]. was limited to -0.800 V vs. Hg/HgO. All the capacity and current details are based on the amount of iron in the composite. Hydrogen evolution and iron reduction potentials were measured at the end of each charging with the initial hump in the charging plateau. Then this was subtracted from the corresponding equilibrium potential value to get the overpotential. The in operando charge efficiency measurements have been described in earlier work, where electrochemical water splitting catalysis coupled with mass spectrometry allowed the detection of gaseous products [38]. The cell consisted of two chambers and was separated by a nonwoven cellulose membrane (Freudenberg, 700/18F) and sealed with rubber gaskets (Kuntze, ESO2 425-010 EPDM). Each chamber with a volume of 48 cm 2 was constantly purged with argon gas through the inlets at the bottom of the chambers. From each chamber the gas outlets were joined in a single exhaust, which served as a gas sampling point for MS measurements (Pfeiffer, Thermostar GSD320-QMG220). The exhaust was then passed through a gas flow meter (Rota Yokogawa, RAK-D41). The iron electrode samples were evaluated with a potentiostat (BioLogic, SP-50) in a two-electrode setup with an oversized conventional nickel electrode as the counter electrode. The geometrical surface areas of the iron and nickel electrodes were 2.25 and 9 cm 2 , respectively. The electrodes were formed (30-35 cycles) prior to this experiment. All samples for in operando charge efficiencies were cycled three times with 5, 10 and 15 mA cm −2 of charge and 5 mA cm −2 for discharge. Figure 1 illustrates the crystal phases for the XRD patterns of the two samples: CuSn (a) and Sn (b). In both samples, α-Fe was the dominant phase. In the case of the Sn sample, Fe3SnC was also observed as a minor phase along with the α-Fe phase. However, when 2 wt. % Cu was introduced with 2 wt. % Sn, Fe Sn phases were observed with traces of Fe SnC. Metallic Cu and Sn phases were not detected due to the low concentrations of the respective elements. Even though both samples have more than 20 wt. % carbon, no graphitic phases are observed, which suggest that these carbons are of amorphous in nature, which may improve the electrode conductivity, whereas the porous network will also increase the ionic conductivity [39,40]. A further advantage of the carbon is that it will buffer the volume expansion and maintain stable electrical conductivity of the iron electrode during the electrochemical cycling [41].     Moreover, both samples were analyzed by TEM with the results as shown in Figure 3. Figure 3a presents the morphology of the CuSn sample showing uniform particle sizes (100-300 nm) embedded in the filamentous carbon layer. In the Sn sample shown in Figure 3b, the particles are smaller (30-150 nm), unevenly sized with irregular shapes. From the TEM mass contrast, darker and lighter platelets shaped iron particle are observed. Thus, closely fused interconnected particles tend to decrease the electrode resistance and enhance the electrochemical performance of the CuSn sample [42].  Moreover, both samples were analyzed by TEM with the results as shown in Figure 3. Figure 3a presents the morphology of the CuSn sample showing uniform particle sizes (100-300 nm) embedded in the filamentous carbon layer. In the Sn sample shown in Figure 3b, the particles are smaller (30-150 nm), unevenly sized with irregular shapes. From the TEM mass contrast, darker and lighter platelets shaped iron particle are observed. Thus, closely fused interconnected particles tend to decrease the electrode resistance and enhance the electrochemical performance of the CuSn sample [42]. Figure 4 displays the XPS spectra of the CuSn sample. Figure 4a shows the Fe 2p spectrum with characteristic peaks at 706.7 eV of metallic Fe(0) 2p 2/3 , 711 eV of Fe(II) 2p 2/3 and 724.5 eV 2p 1/2 . Figure 4b shows the Cu 2p spectrum, with peaks observed at 932.6 and 933.6 eV corresponding to metallic Cu(0), Cu(I) and Cu(II). The Sn 3d 2/5 spectrum (Figure 4c) has one sharp peak at 486.9 eV. This peak corresponds to SnO or SnO 2 . Moreover, a weak shoulder is observed at 485 eV for metallic Sn(0). Figure 4d illustrates the O 1s spectrum. There are two peaks: one at ca 530 eV corresponding to the metal oxides Fe 2 O 3 , Fe 3 O 4 , FeO and SnO 2 , and the second peak at 532.6 eV related to OH [43]. The C 1s spectrum (Figure 4e) shows one strong peak at 284.8 eV confirming the presence of carbon in the sample. The XPS results further confirm that the copper, iron, and tin species exist either in metallic or oxide form on the sample surface. The intensity ratios with corrected atomic sensitivity factor (ASF) (excluding C) are presented in Table 2. We have set the Cu intensity to 1. The intensity ratio corroborates that the tin species are enriched, and copper is depleted on the surface. Almost all iron at the surface is oxidized in the form of Fe 2 O 3 and Fe 3 O 4 , while the core is maintained in metallic form. There is a weak signal from Fe 3 SnC in the XRD data. Given the rather small relative amount we can neither confirm nor exclude carbide from our XPS-data. Moreover, both samples were analyzed by TEM with the results as shown in Figure 3. Figure 3a presents the morphology of the CuSn sample showing uniform particle sizes (100-300 nm) embedded in the filamentous carbon layer. In the Sn sample shown in Figure 3b, the particles are smaller (30-150 nm), unevenly sized with irregular shapes. From the TEM mass contrast, darker and lighter platelets shaped iron particle are observed. Thus, closely fused interconnected particles tend to decrease the electrode resistance and enhance the electrochemical performance of the CuSn sample [42].   Figure 4a shows the Fe 2p spectrum with characteristic peaks at 706.7 eV of metallic Fe(0) 2p / , 711 eV of Fe(II) 2p / and 724.5 eV 2p / . Figure  4b shows the Cu 2p spectrum, with peaks observed at 932.6 and 933.6 eV corresponding to metallic Cu(0), Cu(I) and Cu(II). The Sn 3d / spectrum (Figure 4c) has one sharp peak at 486.9 eV. This peak corresponds to SnO or SnO . Moreover, a weak shoulder is observed at 485 eV for metallic Sn(0). Figure 4d illustrates the O 1s spectrum. There are two peaks: one at ca 530 eV corresponding to the metal oxides Fe O , Fe O , FeO and SnO , and the second peak at 532.6 eV related to OH [43]. The C 1s spectrum (Figure 4e) shows one strong peak at 284.8 eV confirming the presence of carbon in the sample. The XPS results further confirm that the copper, iron, and tin species exist either in metallic or oxide form on the sample surface. The intensity ratios with corrected atomic sensitivity factor (ASF) (excluding C) are presented in Table 2. We have set the Cu intensity to 1. The intensity ratio corroborates that the tin species are enriched, and copper is depleted on the surface. Almost all iron at the surface is oxidized in the form of Fe O and Fe O , while the core is maintained in metallic form. There is a weak signal from Fe3SnC in the XRD data. Given the rather small relative amount we can neither confirm nor exclude carbide from our XPS-data.  To discern the type of carbon in the CuSn sample, Raman spectroscopy was carried out as shown in Figure 4f. Several prominent peaks are observed at 216, 277, 401, 1340, 1607 cm −1 . The first three of these peaks correspond to Fe O [44], while the remaining two peaks represent the D and G bands in carbon materials. Moreover, another diffuse band around 2700 cm −1 is characteristic for the 2D band. Both the D and G bands originate from the in-plane optical phonons in the graphene layer [45]. The D band is not observed in defect-free Highly Ordered Pyrolytic Graphite (HOPG), but becomes Raman active due to defects within the graphene planes [45]. The deconvoluted intensity ratio (ID/IG) of the D and G bands gives the degree of graphitization in the sample and an ID/IG value of 0.64 was obtained. This (ID/IG) ratio and the weak and broad 2D band suggest that the sample is conductive and partially graphitized [46]. The Raman results confirm that amorphous carbon is present in the  To discern the type of carbon in the CuSn sample, Raman spectroscopy was carried out as shown in Figure 4f. Several prominent peaks are observed at 216, 277, 401, 1340, 1607 cm −1 . The first three of these peaks correspond to Fe 2 O 3 [44], while the remaining two peaks represent the D and G bands in carbon materials. Moreover, another diffuse band around 2700 cm −1 is characteristic for the 2D band. Both the D and G bands originate from the in-plane optical phonons in the graphene layer [45]. The D band is not observed in defect-free Highly Ordered Pyrolytic Graphite (HOPG), but becomes Raman active due to defects within the graphene planes [45]. The deconvoluted intensity ratio (I D /I G ) of the D and G bands gives the degree of graphitization in the sample and an I D /I G value of 0.64 was obtained. This (I D /I G ) ratio and the weak and broad 2D band suggest that the sample is conductive and partially graphitized [46]. The Raman results confirm that amorphous carbon is present in the CuSn sample. Furthermore, the presence of FeO X peak suggests that carbon layer does not completely coated over the particles. Figure 5 illustrates the discharge profiles of the electrodes at the constant current discharge of 100 mA g −1 . All four samples had 5 wt. % Bi 2 S 3 , 10 wt. % PTFE and 5 wt. % Ketjenblack EC-300J. Two of the samples had 0.1 wt. % SWCNT (TUBALL ™ OCSIAL EUROPE, Leudelange, Grand-Duché de Luxembourg). Their initial discharge capacities are shown in Table 3. Among them, sample CuSn with 0.1 wt. % CNT and 0.65 M LiOH demonstrate the highest initial discharge capacity at 603 mAh g −1 with 91 % charge efficiency, while the CuSn sample with CNT and without CNT exhibit 470 and 434 mAh g −1 with 93 and 82% charge efficiency, respectively. This performance increase is due to the enhanced electrode conductivity and porosity created by CNT, Cu, amorphous carbon network, which acts as an electron transport and ion transport pathways. Thus, provides a better interface and the ideal reaction site for charging reaction [47,48]. While the addition of LiOH increase the iron dissolution and form a porous hydroxide, which enhancing the electrolyte access to the active material [29]. In contrast to CuSn containing samples, the Sn sample with CNT shows low discharge capacity of 378 mAh g −1 and 76.7 % efficiency, which indicates that 4 wt. % Sn only could not improve the electrochemical performance as with the CuSn sample. Further observations of discharge plateaus ( Figure 5) show that all the electrodes have an initial stable discharge plateau at −0.93 V and when the cycle number increases the discharge plateau potentials decrease to −0.9 V. This potential decay implies that the electrode resistance is increased during cycling, caused by accumulated non-conducting discharge products preventing further oxidation of iron. This effect is more significant on the SnCNT sample, which showed lower performance with the highest overpotential and capacity decay among the four electrodes. In our previous study, 2% Cu doped iron had delivered stable performance at 367 mAh g −1 of discharge capacity and with 80 % of charge efficiency [34]. Comparing the bare Cu and bare Sn-doped samples, Sn performance shows poor performance in terms of capacity and cyclability. Possible reasons for this decay could be tin's lower conductivity without the presence of Cu and non-uniform nanoparticle size distribution as shown in Figures 2d and 3b.

Electrochemical Performance of the Electrodes
The charge profiles of the electrodes at 200 mA g −1 are given in Figure 6. The iron hydroxide reduction voltage plateaus appear between −1.1 V to −1.15 V. During the initial cycles all the electrodes have a single charge plateau and after increasing cycle number two distinct plateaus are observed, one for the iron reduction and another for the hydrogen evolution at −1.2 V. The length of these two plateaus varies with cycle numbers. It can be seen from the CNT containing CuSn electrodes that the iron reduction potential is much lower (about 10 mV) than the CuSn sample without CNT. This confirmed that the charge efficiency is improved by the addition of 0.1 wt. % CNT in the CuSn electrode [21]. The other noticeable difference among the electrodes is the potential overshoot at the start of the charging (inset Figure 6). This primarily arises due to the resistance of iron hydroxide nucleation to Fe transformation [11]. In the CNT containing electrodes this overshoot gradually disappeared over cycles. Increase in the overshoot potential was observed for the CuSn as well indicating that the CNT addition favored nucleation. Two overshoot regions are observed with the Sn sample. This could be due to dissolved tin being reduced back to form Sn or Fe-Sn alloys [35,49]. the SnCNT sample, which showed lower performance with the highest overpotential and capacity decay among the four electrodes. In our previous study, 2% Cu doped iron had delivered stable performance at 367 mAh g −1 of discharge capacity and with 80 % of charge efficiency [34]. Comparing the bare Cu and bare Sn-doped samples, Sn performance shows poor performance in terms of capacity and cyclability. Possible reasons for this decay could be tin's lower conductivity without the presence of Cu and non-uniform nanoparticle size distribution as shown in Figure 2d and Figure 3b.    The capacity of the CuSnCNTLi electrode drops to 50% from 600 (at 1st cycle) to 300 mAh g −1 at 225th cycle (loss of 1.3 mAh g −1 /cycle). The CuSnCNT electrode is more stable with a capacity of 277 mAh g −1 with an 55% charge efficiency after 450 cycles (loss of 0.41 mAh g −1 /cycle). At 450 cycles, the capacity of the CuSn sample drops (loss of 0.47 mAh g −1 /cycle) to 250 mAh g −1 with an efficiency of 45 %. Even though LiOH addition remarkably increases the capacity up to 62% of its theoretical value; its long-term stability is found to be poor. The prolonged cycling increases the H2 evolution and the mechanical stability of the electrode worsens because of too high porosity resulting in loss of active material from the electrode [29]. Similarly, the SnCNT (loss of 0.96  Figure 7 shows the overpotential trends of iron reduction and hydrogen evolution during cycling. The first plateau corresponds to iron reduction as shown in Figure 7a, and the second plateau to hydrogen evolution as shown in Figure 7b. Long iron reduction plateau and a low overpotential results in a higher charge acceptance. The iron reduction overpotential is more positive for the CNT electrodes compared to the CuSn electrode. Without CNT addition the iron reduction overpotential is found to be around −0.17 V while with CNT, the overpotential was reduced to −0.14 V, while the overpotential dropped additionally with LiOH to −0.11 V. On the contrary, the 4 wt. % Sn shows an increasing overpotential (Figure 6d) despite the addition of CNT. Thus, CNT and LiOH addition decrease the iron reduction overpotential only in the CuSn sample. The hydrogen evolution region takes place at lower overpotentials during the initial cycles and increases over subsequent cycles. However, this overpotential increment does not translate into improved performance. Upon the charge/discharge cycles, the iron reduction plateau length gradually decreases while the hydrogen evolution plateau slowly increases to higher potentials. This could be ascribed to the depletion of sulfur and subsequent Bi enrichment on the iron electrode [20,34]. Differential electrochemical mass spectrometry is a useful technique to study the aging of batteries by coupling mass spectrometry with electrochemical techniques [50]. Here, in operando charge efficiency studies were performed when charging the iron electrodes at different current densities and simultaneously quantifying the amount of H2 gas collected in the mass spectrometer. In this study, it was assumed that deviation from 100 % charge efficiency is solely caused by H2 gas evolution originating from charging the iron electrodes. The measurements are based on Faraday's law: where n = number of moles (mol), I = current (A), t = time (s), z = electrons transferred per ion (z = 2 for H2) and F = 96,485 C mol −1 . Stainless steel (SS) was chosen as a standard to quantify the accumulated H2 gas upon assuming 100% efficiency. A linear potential sweep from 0.3 to 0.55 V vs. Ag/AgCl with a sweeping rate of 0.2 mV s −1 was done to calibrate the accumulated H2 gas on SS and the charge passed was correlated with the recorded counts of H2 from the mass spectrometer. Calibration was done prior to each sample study and quantification factors were calculated according  The capacity of the CuSnCNTLi electrode drops to 50% from 600 (at 1st cycle) to 300 mAh g −1 at 225th cycle (loss of 1.3 mAh g −1 /cycle). The CuSnCNT electrode is more stable with a capacity of 277 mAh g −1 with an 55% charge efficiency after 450 cycles (loss of 0.41 mAh g −1 /cycle). At 450 cycles, the capacity of the CuSn sample drops (loss of 0.47 mAh g −1 /cycle) to 250 mAh g −1 with an efficiency of 45 %. Even though LiOH addition remarkably increases the capacity up to 62% of its theoretical value; its long-term stability is found to be poor. The prolonged cycling increases the H 2 evolution and the mechanical stability of the electrode worsens because of too high porosity resulting in loss of active material from the electrode [29]. Similarly, the SnCNT (loss of 0.96 mAh g −1 /cycle) sample also presents poor long-term stability.
As shown in the TEM images in Figure 3, the large particle size distribution and excessive H 2 evolution is detrimental to the long cycle performance of the electrode. The high capacities could be realized by using the nano CuSn sample with different additives (LiOH and CNT), which also imply that the CNT addition increases the porosity. The porosities of the electrodes lied in between 55 to 60%. Highly porous electrodes have higher interparticle resistance, compared to a metallic conductor. During cycling, increase in the interparticle contact resistance affects the charge acceptance of the iron reduction and consequently accelerates the capacity decay.
Differential electrochemical mass spectrometry is a useful technique to study the aging of batteries by coupling mass spectrometry with electrochemical techniques [50]. Here, in operando charge efficiency studies were performed when charging the iron electrodes at different current densities and simultaneously quantifying the amount of H 2 gas collected in the mass spectrometer. In this study, it was assumed that deviation from 100 % charge efficiency is solely caused by H 2 gas evolution originating from charging the iron electrodes. The measurements are based on Faraday's law: where n = number of moles (mol), I = current (A), t = time (s), z = electrons transferred per ion (z = 2 for H 2 ) and F = 96,485 C mol −1 . Stainless steel (SS) was chosen as a standard to quantify the accumulated H 2 gas upon assuming 100% efficiency. A linear potential sweep from 0.3 to 0.55 V vs. Ag/AgCl with a sweeping rate of 0.2 mV s −1 was done to calibrate the accumulated H 2 gas on SS and the charge passed was correlated with the recorded counts of H 2 from the mass spectrometer. Calibration was done prior to each sample study and quantification factors were calculated according to: where I EC = current recorded from the potentiostat (A) and I MS = extracted ion signal from the MS (A). The quantification factors were determined to (12.18 ± 1.32) × 10 6 . The number of H 2 (g) moles per second evolved during charge of the iron electrodes was calculated by: where n H 2 (g) = number of H 2 (g) moles evolved (mol). This is further derived to the cumulative H 2 (g) capacity: where Q H 2 (g) = charge passed corresponding to H 2 (g) (C). Lastly, the charge efficiency was defined as: where Q tot = total capacity of the iron electrode obtained from the potentiostat. In Figure 8 the charge efficiency at 5, 10 and 15 mA cm −2 is shown for the four samples. For each sample, the top plot shows the cell potential and number of H 2 (g) moles (Equation (6)) vs. time while the bottom plot shows the charge efficiency (Equation (8)) and cumulative H 2 (g) capacity (Equation (7)) vs. time. Each charge curve was taken at the last cycle of the specific current density. Also, the H 2 (g) counts from the MS was extended to ensure that all accumulated gas in the system was considered. The highest efficiencies achieved were in the following order: CuSnCNT > CuSn > CuSnCNTLi > SnCNT. These results are comparable to Table 3, except that the order between CuSn and CuSnCNTLi are on the opposite. The lower efficiency obtained for CuSnCNTLi is due to the prolonged charging at potentials above 1.65 V, evident from Figure 8c. The efficiency evolution during charge in Figure 8 shows three distinguished slopes at potentials: (1) below 1.5 V, (2) between 1.5-1.6 V and (3) above 1.6 V. In the first region, the least H 2 (g) amount was evolved while nucleation of Fe(OH) 2 to Fe took place. In the second, constant rate of H 2 (g) was released upon iron reduction. Lastly, H 2 (g) evolution dominated above 1.6 V and led to the overcharging the iron electrode. The highest efficiency of charging iron electrodes occurred at higher current densities, and the efficiency results showed that the electrodes benefited from charging at higher current densities than 10 mA cm −2 . Samples with enhanced conductivity, for instance added Cu and CNT, revealed higher charge acceptance of iron reduction, which ultimately suppressed the H 2 (g) evolution. The efficiencies acquired from the mass spectrometer were compared with the coulombic efficiencies from the potentiostat and are validated with the results as shown in Table 4 with small deviations as a result.  In summary, we have studied two type of nano iron doped electrode materials: one with CuSn and the other one with Sn. Among them, CuSn-doped nano iron samples have delivered high capacities and efficiencies ranging from 434 to 603 mAh g −1 and 83 to 93%, respectively, depending on the additives (CNT, or CNT and LiOH). These performance values are even higher from the  In summary, we have studied two type of nano iron doped electrode materials: one with CuSn and the other one with Sn. Among them, CuSn-doped nano iron samples have delivered high capacities and efficiencies ranging from 434 to 603 mAh g −1 and 83 to 93%, respectively, depending on the additives (CNT, or CNT and LiOH). These performance values are even higher from the reported values in literature [18,23,34]. Based on coupled mass spectrometry and charging iron electrodes, and the galvanostatic cycling profile, CNT containing electrodes have a stable hydrogen evolution plateaus at 1.65 V vs. NiOOH or −1.2 V vs. Hg/HgO. This high rate of hydrogen evolution is not beneficial to the electrode structure. It might destroy the electrode pore structure and interconnectivity of the particles. On the contrary to this, the CuSn does not have pronounced hydrogen evolution plateaus. In the meanwhile, the hydrogen evolution is continuing throughout the charging, not limited to the end charge. MS data (Figure 8a) also confirms this trend. Therefore, inhibiting the high hydrogen evolution could improve the electrode mechanical stability and life cycle, which could be done by constant voltage charging or potential limited constant current charging. High current charging (10 and 15 mA cm −2 ) improved the electrode efficiency as shown in Table 4.
This study shows that addition of both CuSn, partly amorphous carbon and CNT improve the electrode performance. The Cu, carbon and CNT give a stable interfacial conductivity irrespective of the state of charge. Furthermore, Sn could slow down the hydrogen evolution kinetics [35]. These dopant elements and CNT additive enhance the nano iron cyclability.

Conclusions
In our study, CuSn and Sn-doped Nano iron/carbon composites were investigated as negative electrode material for alkaline electrolyte. The best performing CuSn electrode has shown interconnected iron particles, uniformly distributed Cu, Sn and carbon network for effective charge and discharge operation. As a result, of this, CuSn-based electrodes have shown excellent performance. This work also demonstrates the effectiveness of SWCNT and LiOH addition to the electrode on the performance. The CNT addition improves the charge efficiency and capacity to 93% and 470 mAh g −1 , respectively. Even though the CNT with LiOH addition increased the performance to 91% and 603 mAh g −1 , its cycle stability is compromised. In operando MS analysis quantifies the amount of evolved hydrogen and shows the three different hydrogen evolution rate domains. The higher hydrogen evolution at the end of the charge made the electrodes begin to fail rapidly. The different charge current density vs. performance shows that at higher current density the amount of evolved hydrogen is reduced, and the performance is also improved dramatically. These findings will help to develop suitable charging programs for the iron electrodes. This CuSn nano iron is a suitable negative electrode for the alkaline Ni-Fe or Fe-air battery application.