Sai Abhishek Palaparty‡
,
Rajankumar L. Patel‡ and
Xinhua Liang*
Department of Chemical and Biochemical Engineering, Missouri University of Science and Technology, Rolla, MO 65409, USA. E-mail: liangxin@mst.edu
First published on 26th February 2016
Tin oxide (SnO2) has a high theoretical capacity (∼782 mA h g−1), but it experiences large volume changes during charge and discharge cycles that cause rapid capacity fade, which limits its practical use as an anode material. In an attempt to solve this, we coated these particles with ultrathin electrochemically active iron oxide (FeOx) films that act as an artificial solid electrolyte interphase layer, thus stabilizing the SnO2 particles for better longevity of significantly improved performance at high current densities in a practical voltage window. Since there exists a tradeoff between species transport and protection of particles (expecting long life), a film with an optimum thickness was achieved by atomic layer deposition (ALD) of FeOx on SnO2 particles. With an optimum thickness of about 0.24 nm after 20 cycles of iron oxide ALD (20Fe), an initial capacity of ∼658 mA h g−1 was achieved at a high current density of 1250 mA g−1. After 1000 cycles of charge/discharge at 1250 mA g−1, the 20Fe sample showed a capacity retention of 94% as compared to 52% of the uncoated sample when cycled at room temperature; at 55 °C, the capacity retention of the 20Fe sample was 93% compared to 33% of the uncoated sample.
Atomic layer deposition (ALD) is the process of choice for such ultrathin film growth as it enables conformal, pin-hole free, and high aspect ratio film formation.26 These ultrathin ALD films increased the cycle life and capacity retention; however, normally there was a decrease in the initial discharge capacity of the ALD coated samples, as compared to the uncoated sample. In these studies, an ultra-thin film was generally used and the charge/discharge cycle were limited to small cycle numbers (<200 cycles). This can be due to the fact that these films (e.g., Al2O3, ZnO and ZrO2) used in these studies were insulating and thus increased mass transfer resistance for Li+ transfer. There was a trade-off between capacity and cycling life due to the insulating properties of ALD films. Recently, we demonstrated that this dilemma could be solved by using conductive ultrathin cerium oxide films.27 The initial capacity of the optimal ∼3 nm CeO2-coated LiMn2O4 particles showed an initial discharge capacity increase of 24% compared to the pristine one, and the capacity retention significantly improved to 96% and 95% after 1000 cycles at room temperature and 55 °C, respectively, when cycled at 1C rate. This study showed that both high capacity and high cycling stability can be achieved by using suitable conductive thin film coating with an optimal thickness.
In this study, ultrathin iron oxide film is considered as a candidate to improve the performance of SnO2 nanoparticles, since iron oxide is electrochemically active and has been used as an anode with a theoretical capacity higher than that of SnO2.28–30 Also, iron oxide is abundant in nature and is environmentally benign. Iron oxide thin film coating on SnO2 has been studied previously and it was found out that the synergy between iron oxide and SnO2 could stabilize its structure and improve its electrochemical performance.17,20,31–33 However, in these studies, the iron oxide films were prepared by liquid phase methods, and they were either too thick or not conformally coated on the SnO2 particles surface. In addition, these studies were limited to low cycle numbers of charge/discharge testing or tested at low current densities for capacity retention and/or cycle life. To the best of our knowledge, there has been no electrochemical study of ultrathin film of iron oxide coated on LIB electrode particles by ALD. In this study, we report the initial discharge capacity increase and long cycling life of the iron oxide ALD coated commercial SnO2 nanoparticles at both room temperature and elevated temperature when tested in a practical voltage window.
The particles were subjected to X-ray powder diffraction analysis in Philips X-Pert multi-purpose diffractometer (MPD) using Cu Kα radiation (λ = 1.54056 Å) with 2θ ranging from 5 to 90° at a scanning rate of 1.4° min−1. The coated particles were visualized using FEI Tecnai F20 TEM/STEM supported with an energy dispersive X-ray spectrometer system. The loading of Fe on coated samples was determined using inductively coupled plasma atomic emission spectrometer (ICP-AES). The thermogravimetric analysis (TGA) of the uncoated and the coated particles was conducted using a Q50 TGA/DSC (TA instruments) with a flowing oxygen atmosphere (40 mL min−1) at a heating rate of 10 °C min−1 up to 1000 °C.
The anode for the coin cell was prepared using 5 wt% of PVDF binder dissolved in NMP, which was added to a mixture of 85 wt% of anode material and 10 wt% carbon black to form a slurry. Using a razor blade, this slurry was spread on a Cu foil (>99.9%, MTI corporation) uniformly mounted on a glass plate. This electrode composite was placed in an oven and dried under vacuum at 120 °C to evaporate the solvent for 8 hours. After that, disks of approximately 8–13 mm in diameter were punched and cold pressed. Two electrode CR 2032 coin cells were fabricated in an Ar glovebox using punched disks as active anode electrode and with Li metal (99.99%, Aldrich) as counter and reference electrode. The two electrodes were separated using a porous Celgard-2320 separator composed of a 20 μm thick polypropylene (PP)/polyethylene/PP tri-layer film. Commercial electrolyte (LiPF6 in 1:1:1 volume ratio of EC:DMC:DEC) was used as purchased from MTI corporation.
The prepared CR 2032 coin cells were subjected to galvanostatic charge–discharge capacity testing at different current densities, capacity retention testing for 1000 charge–discharge cycles and ac impedance analysis at room temperature as well as at 55 °C. An 8-channel battery analyzer (Neware Corporation) was used to measure charge/discharge capacity ranging from 0.5 to 3 V. Electrochemical impedance spectroscopy (EIS) of the prepared cells were performed using a BioLogic SP-150 potentiostat and impedance analyzer. The impedance was measured over a frequency range of 0.01 Hz to 1 MHz and at a perturbation of 5 mV. Equivalent circuit models for the impedance spectra were fitted using EC-Lab software.
Fig. 2 depicts the galvanostatic discharge capacities of coin cells assembled from the UC, 10Fe, 15Fe, 20Fe, and 30Fe samples obtained using different current densities at both room temperature and 55 °C. The measurements were carried out at different current densities of 50 mA g−1, 125 mA g−1, 250 mA g−1, 500 mA g−1, 1250 mA g−1, and 2500 mA g−1 and each for five cycles. In this study, we set the discharge cutoff voltage to 0.5 V as a practical limit and the charge cutoff voltage to 3 V for testing the cycle life and capacity retention of our coated and uncoated samples. Cycling at room temperature and a low current density of 50 mA g−1, the UC sample showed a discharge capacity of ∼744 mA h g−1, which is close to the theoretical capacity of SnO2. The 10Fe, 15Fe and 20Fe exhibited higher initial discharge capacities as compared to the UC and the 30Fe sample at the current densities of 50 mA g−1, 125 mA g−1, and 250 mA g−1. At high current densities (i.e., 1250 mA g−1), a clear distinction is seen in the performance of various samples. The UC sample performed poorest due to formation of solid electrolyte interface (SEI) at high current densities. The 10Fe, 15Fe, and 20Fe performed significantly well in terms of capacity retention as the ultrathin coating served as an artificial SEI layer at higher current densities. However, the 30Fe sample performed the poorest among the coated samples. A similar trend was observed at 55 °C; however, the initial capacities of the 10Fe, 15Fe, and 20Fe samples at the low current density of 50 mA g−1 were higher than those tested at room temperature. This could be resulted from high electrical and ionic conductivity, and low ohmic potential at high temperature for tin based electrodes.36
The SEI is very crucial for the cycle life of the LIB.37 This is because it forms a protective layer that prevents undesirable reactions with the electrolyte, but it grows thicker while cycling, which eventually slows the Li+ diffusion. Therefore, the ultrathin ALD film in our case can act as an artificial SEI layer thereby passivating the entire surface of the active material, and potentially stopping the formation of the growing organic SEI layer. This could be the reason for better performance of the coated samples as compared to the UC sample during cycling at high current densities of 1250 mA g−1 and 2500 mA g−1. Most of the previous ALD studies on battery electrode particles demonstrated that this ultrathin coating could delay the transport of Li+, thus decreasing the specific capacity.21,27 The reason is that the ALD films were insulating metal oxides in the previous studies. In this study, iron oxide was used as the coating materials. Iron oxide is electrochemically active and has been used as anode materials. In addition, iron oxide has a higher theoretical specific capacity as compared to tin oxide.29,30 At a low current density of 50 mA g−1, the 10Fe, 15Fe and 20Fe samples showed an improvement in initial capacity both at room temperature and at 55 °C, when compared to the UC sample. This trend of increase in initial discharge capacity at higher temperature is in agreement with other studies on iron oxide/SnO2 nanocomposites.38,39 The voltage testing window falls inside the electrochemical active region for iron oxide. This means elemental Fe was formed due to irreversible initial reaction between iron oxide and Li+. The reaction between SnO2 and Li+ yielded Sn and Li2O. This means there would be formation of Sn/Fe/Li2O matrix. The formation of Sn/Fe/Li2O matrix further increased the electronic conductivity.17 In one previous report of graphene/SnO2/Fe2O3 nanocomposites, the Sn acted as an inactive matrix for the iron oxide, which contributed most towards capacity.39 However, in our case, the film is very thin and the amount of iron oxide is very less than that of the SnO2. Hence, it is very difficult to determine individual contribution of SnO2 and iron oxide towards the capacity of the coated samples.
The exact reason for the poor performance of the 30Fe sample is not clear. This could be due to the increased mass transfer resistance of Li+ due to “excessive” thickness of the film. Similar thickness effect was observed for the case of LiMn2O4 particles coated with CeO2 films.27 However, 30 cycles of iron oxide is only about 0.36 nm thick (growth rate of 0.012 nm per cycle), which may seem too thin to be said excessively thick. For example, in our recent studies, the optimum thickness of CeO2 films on 8 μm LiMn2O4 particles was 3 nm.27 Studies by Lee et al. showed that a 1.1 nm thick alumina coating on LiCoO2 (∼400 nm) itself was too thick and increased Li+ diffusion resistance.40 Studies by Guan and Wang of alumina ALD coated on micron sized LiMn2O4 powders showed that a film thickness of ∼1.2 nm was too thick.41 Sun et al. showed that a ∼1 nm thick film of ZrO2 on Li4Ti5O12 anode had an optimal performance.42 1.2 nm alumina is overly thick, while 3 nm of CeO2 is not, because the alumina film is insulating whereas CeO2 has high ionic conductivity and has been used as a solid electrolyte.43,44 Hence, it can be said that the definition of “overly” thick film depends upon the nature of the film and the substrate.
Herein, the FeOx film of ∼0.36 nm thickness may itself be too thick for the nano-SnO2 particles in the 30Fe sample. Recent study on lithium ion diffusion mechanism in iron oxide electrode suggested that there existed three regions in the diffusion profile of intercalated iron oxide.45 The region consisting of Fe/Li2O was found to be the slowest for Li+ diffusion. In our case, the Li+ ion first reacted with the FeOx film before entering the SnO2 lattice and hence, Fe/Sn/Li2O matrix would form. With increase in the number of ALD coating cycles, the FeOx film got thicker, which in turn lead to more Fe/Li2O formation. Since, this is the region for the slowest diffusion for Li+, and with limited lithiation, it increased the mass transfer resistance in case of 30Fe sample that led to poorer performance as compared to the other coated samples at higher current densities. Hence, it can be said that the 30Fe sample has an “overly” thick ALD film. Studies to understand this mechanism in detail is being pursued.
To see whether this ALD coating enhances the electrochemical performance in terms of capacity retention and cycle life of SnO2 at higher current densities, the samples were cycled at a high current density of 1250 mA g−1 at both room temperature and at 55 °C (Fig. 3). The FeOx coated samples (10Fe, 15Fe, and 20Fe) performed far better in terms of capacity retention. At room temperature, after 1000 cycles of charge/discharge, the coated samples showed high capacity retention, ∼91% capacity retention of the 10Fe sample, ∼92% of the 15Fe sample, and ∼94% of the 20Fe sample. This is because the ultrathin film was sufficiently thick and provided artificial SEI. Also, the formation of the electronic conductive in situ Sn/Fe/Li2O matrix leads to a better capacity as compared to the UC sample even after a large number of charge–discharge cycles. In contrast, the 30Fe, similar as the UC, experienced a severe capacity fade after only 350 cycles as compared to the other coated samples at room temperature. At room temperature after 1000 cycles, the capacity retention of the UC sample was about ∼52%. The performance of 30Fe sample is even worse as it showed a capacity retention of only ∼11% after 1000 cycles at this current density. This is because of the formation of thick SEI on the UC sample and the presence of “overly” thick FeOx film on the 30Fe sample.
Fig. 3 Galvanostatic discharge capacities of SnO2 particles coated with various thicknesses of iron oxide ALD films at 1250 mA g−1 between 0.5–3 V at (a) room temperature (b) 55 °C. |
At 55 °C, a similar trend was observed but there was an increase in initial discharge capacity of the coated samples when compared to the testing at room temperature. The 10Fe sample showed a capacity retention of ∼89%, compared to ∼90% of the 15Fe sample and ∼93% of the 20Fe sample after 1000 charge/discharge cycles. The very high capacity retention over long cycling of these coated samples can be attributed to the formation of the stable artificial SEI layer provided by the ultrathin film. At higher temperature, the performance of the cell degrades faster.46 This is also evidenced in form of reduced capacity retention during testing at higher temperature. The capacity retention of the UC and 30Fe sample at this temperature is ∼33% and ∼10%, respectively.
To further understand the effect of this ultrathin film on the performance of the SnO2 particles, testing was performed at lower current densities at both room temperature and 55 °C. Fig. 4 shows the results at a current density of 500 mA g−1. The UC sample showed an initial discharge capacity of ∼680 mA g−1 at room temperature. The 10Fe, 15Fe and 20Fe samples showed an initial discharge capacity of ∼689 mA h g−1, ∼706 mA h g−1 and ∼720 mA h g−1, respectively, at room temperature. The 10Fe, 15Fe and 20Fe samples showed excellent capacity retention of ∼93%, ∼93%, ∼94%, respectively, even after 1000 cycles. With decrease in current density, there was more capacity retention for the UC and the 30Fe sample. The UC sample retained a capacity of ∼66%, whereas the 30Fe sample retained a capacity of ∼62% after 1000 cycles. As in the previous case, the initial capacity of the samples further improved and the capacity retention decreased when tested at a higher temperature of 55 °C.
Fig. 4 Galvanostatic discharge capacities of SnO2 particles coated with various thicknesses of iron oxide ALD films at 500 mA g−1 between 0.5–3 V at (a) room temperature (b) 55 °C. |
The cycling performance of the cells were also tested at a current density of 250 mA g−1 at both room temperature and 55 °C (see Fig. S5 in ESI†). Compared to the testing at 500 mA g−1, all the samples showed a further improvement in capacity retention at both room temperature and 55 °C. These results suggest that the effect of the ultrathin film is more significant for the samples tested at high current densities as it passivated the particle surface by serving as an artificial conformal SEI layer and thus providing protection from undesirable reactions with the electrolyte. It is also interesting to note that during initial cycles, the 30Fe sample performed better than the UC sample. However, with cycling the 30Fe sample performs poorer as compared to the UC sample (Fig. 3 and 4). This could be due to the electrochemical active nature of the ultrathin film. As discussed earlier, when Li+ reacts with iron oxide, volume expansions occur.28–30 This could increase the stress on the ultrathin film. This increased stress could be the reason for poorer performance of 30Fe sample than that of the UC sample.
This series of testing indicates that the 10Fe, 15Fe, and 20Fe samples showed a very high capacity retention even after 1000 cycles of charge/discharge at a high current density of 1250 mA g−1. This is a significant achievement as compared to the previous studies. For example, FeOx/SnO2 composite synthesized by El-Shinawi et al. showed a capacity retention of only ∼20% after 100 cycles when discharged at a current density of 400 mA g−1.17 Heterostructures of iron oxide and SnO2 produced by Zhou et al. also showed significant capacity fade of about 75% only after 30 charge/discharge cycles at 1000 mA g−1.32 In contrast, our 20Fe sample showed a significantly higher capacity retention of 94% even at a high current density of 1250 mA g−1 after 1000 cycles at room temperature. Ultra-small SnO2 nanocomposites have demonstrated enhanced performance in terms of capacity retention at high current densities.7,16 However, the average size of the synthesized particles in these studies was less than <10 nm. Our study is unique as we demonstrated how capacity retention can be improved for much larger commercial SnO2 nanoparticles (<100 nm) using optimal ultrathin film coating by ALD.
To understand the kinetics change due to the ultrathin film, EIS analysis was performed for the coated samples as well as the uncoated samples at room temperature and 55 °C. Fig. 5 presents the results tested at room temperature. The equivalent circuit (Fig. 5c), comprised of three resistances. Rohm refers to the uncompensated ohmic resistance between the working electrode and the reference electrode, i.e., Rf (the resistance for lithium ion mobility in the surface layer including SEI layer and/or surface modification layer), Cct (the ideal capacitance of the surface layer and the double layer), and Rct (the charge transfer resistance). W represents the Warburg impedance that outlines the lithium ion diffusion in the bulk material. The Warburg impedance and the lithium ion diffusion coefficient of the working electrode are inversely proportional. Though these values of resistances have no physical significance, it can be used to compare the kinetics of the coated and the uncoated samples.
At room temperature, in the EIS of the fresh cells, two significant semicircles for the UC sample, whereas only one major semicircle was observed for the coated samples. In the case of the coated samples, there was overlap between contribution of the charge transfer resistance at mid-high frequencies and the SEI layer/ultrathin coating contribution at high frequencies,47 which could be the reason for appearance of only one major semicircle in the EIS. From the impedance parameters tested at room temperature (Table 1), the 20Fe sample has the least charge transfer and Warburg impedance values, when compared to other samples before and after 1000 charge/discharge cycles. The coated samples showed low film resistance even after 1000 cycles of charge/discharge, when compared to the UC sample. Out of all the tested samples, the 30Fe sample shows the highest diffusion resistance, which could be the reason for its poor performance when compared to the other coated samples. As compared to the 30Fe sample, the decreasing values of the charge transfer resistance and lower Warburg impedance values of the 10Fe, 15Fe and 20Fe samples served as evidence to say that the 20Fe sample had an optimal film thickness.
RT | Rohm (ohm) | Rct (ohm) | Rf (ohm) | Cct (μF) | Cf (μF) | W (ohm s−1/2) | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Sample | 0th | 1000th | 0th | 1000th | 0th | 1000th | 0th | 1000th | 0th | 1000th | 0th | 1000th |
UC | 20 | 26.1 | 150 | 180.3 | 50 | 77.6 | 0.2 | 0.16 | 0.02 | 0.015 | 70 | 103.9 |
10Fe | 5 | 6.8 | 70 | 87.6 | 10 | 18.9 | 2 | 1.5 | 0.5 | 0.4 | 60 | 74.4 |
15Fe | 5 | 5.6 | 60 | 71.3 | 14.3 | 21.8 | 2 | 1.7 | 0.5 | 0.5 | 94 | 106.9 |
20Fe | 6 | 6.8 | 40 | 59.3 | 17.5 | 27.8 | 10 | 9 | 10 | 9.1 | 50 | 64.7 |
30Fe | 4 | 5.3 | 50 | 77 | 21 | 31.6 | 2 | 1.7 | 1 | 0.5 | 111 | 136.2 |
At 55 °C, a similar trend was seen when compared to the EIS analysis at room temperature. For the coated samples only one semicircle was observed, whereas for the uncoated sample, two semicircles were present in the EIS analysis (Fig. 6). Out of all the samples, the 20Fe sample showed the lowest charge transfer resistance and Warburg resistance, as compared to the other samples after 1000 cycles of charge/discharge (Table 2). These values are also lower when compared to the results tested at room temperature. This could be due to the increase in electric and ionic conductivity of SnO2 particles at higher temperatures. The 30Fe sample experiences the highest Warburg resistance of all the samples. When compared to the room temperature testing, the charge transfer and the Warburg resistance values decreased during the testing at 55 °C. This supplements the improved performance of the coated samples at higher temperatures when compared to the samples tested at room temperature in terms of initial capacity.
55 °C | Rohm (ohm) | Rct (ohm) | Rf (ohm) | Cct (μF) | Cf (μF) | W (ohm s−1/2) | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Sample | 0th | 1000th | 0th | 1000th | 0th | 1000th | 0th | 1000th | 0th | 1000th | 0th | 1000th |
UC | 15 | 19.2 | 130 | 172.7 | 13 | 47.3 | 0.3 | 0.21 | 0.022 | 0.013 | 84.3 | 121.7 |
10Fe | 5 | 6.3 | 63.2 | 73.6 | 7.8 | 15.3 | 2.1 | 0.18 | 0.5 | 0.4 | 55.7 | 69.3 |
15Fe | 4.8 | 6 | 51.2 | 64.2 | 10 | 18.3 | 2.3 | 2.1 | 0.6 | 0.5 | 84 | 94.7 |
20Fe | 5.5 | 7 | 32 | 60.1 | 12.1 | 22.8 | 12 | 9 | 11.5 | 10 | 42.3 | 67.8 |
30Fe | 4.3 | 6 | 62 | 88 | 15.3 | 27.1 | 2.1 | 1.9 | 1.2 | 1 | 105.3 | 122.9 |
From the EIS analysis at both room temperature and at 55 °C, it is found that the 20Fe sample has the optimal ALD film thickness. Iron oxide is conductive (see Fig. S6 in ESI†). The lower values of the film resistance of the coated samples, as compared to the uncoated sample, after 1000 cycles of charge/discharge is indicative of the ultrathin film serving as an artificial SEI layer. The lower charge transfer, film and Warburg resistances of the 10Fe, 15Fe and 20Fe samples, as compared to the UC sample, could probably explain improved performance at high current densities.
Footnotes |
† Electronic supplementary information (ESI) available: Iron oxide film characterization, charge/discharge cycling of coated and uncoated SnO2 particles at 250 mA g−1, and conductivity measurements. See DOI: 10.1039/c6ra00083e |
‡ These authors contributed equally. |
This journal is © The Royal Society of Chemistry 2016 |