Additional issues with SEI formation - Electrolytes, SEI Formation, and Binders: A Review of No

While surface engineering can enhance CE@1st, and additives can improve long-term cycling, there are a few things to remember about SEI going forward. First off, SEI

formation, especially during the first cycle, is dependent on cell chemistry. Additionally, the pervasive belief that inefficiencies are solely due to SEI is not entirely true: carbon and insertion anode seem to be very adept at irreversibly trapping sodium ions. Lastly, the use of current organic solvents will result in SEI formation. This leaves two possible solutions:

either find new solvents that are more cathodically stable, and do not degrade as readily, or find additives that limit the degradation of the current solvents. The former would be more ideal, but it is also the less realistic, though a recent study by Nohira et al. did find 90%

capacity retention of 1000 cycles using an ionic liquid electrolyte.^[50]

The latter solution would be using additives such as FEC in ways to create a more stable SEI as to improve the long-term cycling prospects. Because there is so many unknowns about the current state of the SEI both experimentally and computationally, this latter method does not have a very clear research direction and thus a high throughput experimental screening of electrolyte additives might prove to be appropriate in this instance.

An additional wrinkle to consider regarding the SEI formation is whether the electrochemical cycling is conducted in a full cell, where an actual cathode is the source of sodium ions, or in a half cell, where the source of the sodium ions is merely a sodium counter electrode. The

half-cell setup is typically a favorite of initial research, as it allows the research to focus solely on the properties of the material at hand. In the half-cell approach, it is assumed that the sodium metal is at best not reactive, and at worse negligibly reactive, with the electrolyte.

However, this was shown to be untrue in studies by both Ponrouch et al.^[104] and Ji et al.^[105], as they showed that conventional carbonate-based solvents react with sodium metal to a quite significant extent. The study by Ji et al. went a step further to demonstrate that electrolyte degradation was in fact the main source behind the fading of the carbon anode in half-cells.

Furthermore, it was also shown that this fading was able to be reversed if a new piece of sodium metal, along with fresh electrolyte, was added to couple the ‘faded’ carbon anode.

This work shows NIB researchers two things: that the electrolyte, and its conditions, play a large role in determining the electrochemical properties of a battery, and that testing the long-term cycling viability of a material in a half-cell is inherently short-sighted due to the

problem with electrolyte degradation.

5. BINDERS

In addition to the electrolyte, which makes the electrochemical reaction possible, another important external component to the operation of a battery is the binder used to hold the electrode material onto the carbon collector. Unless the electrode material is a naturally self-standing material and can be used as a monolith anode, something will need to fix the electrochemically active mass to the current collector. Should the electrochemical material ever become detached from the current collector, it is rendered inactive, and can no longer access the electrons needed to make the storage reaction proceed.

While the selection of the binder may seem trivial, it is anything but. As we will see later on, the wrong binder selection can make all the difference between a failed anode material and a successful one—even if the electrochemically active material remains exactly the same. The roughly 5-10 mass % of the anode that is allocated to the binder can play a disproportionately large role in the overall battery performance.^[106]

Figure 9. a) Types of binders used in anode materials for NIBs. b) Histogram of the mass%

of binders used. c) Binder type per anode chemistry. d) Mass % of binders per anode chemistry.

As we can see from the Figure 9, the most commonly used binder material in the literature is polyvinylidene fluoride (PVDF), which is usually included into the electrode slurry in 10 wt.% by mass. This is not too surprising, as the most commonly used binder for commercial LIB electrodes has been PVDF, though it typically comprises 5% or less of the total mass.

Using 10 wt% is common amongst academic labs, but not industry. However, there are subgroupings within the literature, which bear more exploration.

(a) (b)

One of the noticeable subgroupings occurs with alloying, and conversion type anodes. While the use of binders is proportionally the same for carbon and insertion type anodes, alloying and conversion anodes show to have a much higher use of the carboxymethyl cellulose (CMC) binder. This highlights the fact that electrochemical needs for different types of battery chemistry differ, and as such, the binder needed for the optimal electrode changes as well. Furthermore, the lower prevalence of PVDF for alloying and conversion anode should come as no surprise. In the case of LIBs, it is well known that PVDF binder is not conducive to the performance of alloying type anodes.^[107]

Additionally, as for the mass percentage of the binder used, it is much different for alloy based anodes. A greater number of papers reported higher binder mass percentages for the alloy based anodes. This is due to the fact that, since alloys have a greater volume change upon sodiation than the other types of anodes, more binder is needed to prevent the

pulverization of the electrodes. Looking at the recent research, it is suggested that, like the case of the alloy and conversion materials, switching away from PVDF binder and towards ones, such as CMC or polyacrylic acid (PAA) might be promising.

5.1. Reactions with electrolyte to form SEI

First it is well known from the LIB literature that PVDF partially decomposes by reactions during electrochemical cycling, which negatively affects the CE@1^st and long-term cycling of the battery.^[108] This was the subject of a study by Villevieille et al. who found that the addition of FEC slows the rate of decomposition of the PVDF binder (Figure 9), which forms an NaF based SEI,^[109] thus giving researchers every incentive to pursue alternative binder choices.

Figure 10. Schematic representation of the breakdown of SEI formation due to the presence of PVDF binder, with either no FEC present, or FEC present. Reproduced with permission.

One such early example was reported by Komaba et al. who demonstrated that the use of CMC binder leads to a much more stable long-term cycling as compared to PVDF.^[110]

Additionally, CMC has the advantage of being water soluble, which makes it easier and safer to process, as opposed to PVDF that is processed in more toxic and volatile N-Methyl-2-pyrrolidone (NMP). Komaba et al. ^[111] and Huang et al.^[112] also made a similar observation through the use of a PAA binder with a Sn based anode, which also showed superior

performance, as compared to PVDF. In both papers, it was hypothesized that the PAA binder led to improved cycling due to the formation of a more stable passivation layer, as well as a lowered internal resistance. However, it should be noted that any improvements afforded by the use of a PAA binder are entirely dependent on the type of anode used. In the case of co-intercalating anode, Maibach et al. showed that PAA binders led poorer performances than expected.^[113]

5.2. Importance of binders for alloying anodes

While it becomes clear that the impact of the binder choice is insignificant on cycle life for hard carbon materials, the case is different for alloying type anodes. In alloying type anodes, it has been shown that the choice in the binder material has an outsized effect on the

performance of the battery (Figure 11). This is logical: during the alloying process, the growth of particles stresses the binder and one of two things can happen. Either the binder is too weak to control the swelling of the particles, which then loses contacts with the current

collector, or the binder is so strong that it becomes restrictive of the particle growth, leading to cracking or an incomplete alloying process. Furthermore, given the poor electrical conductivity of the alloying particles, it is also critical that the binder be somewhat

conductive. The effect of the binder on the electrochemical performance of an alloying type anode was demonstrated by Liu et al. who show the drastic differences in capacity of an Sn based anode when using a PVdF, CMC and Poly(9,9-dioctylfluorene-co-fluorenone-co-methylbenzoic ester) (PFM) binder.^[114] The PFM binder, which is more conductive than the other two, shows the best electrochemical performance with the Sn based anode. Meanwhile the CMC binder shows an average performance, while the PVdF binder displays very little capacity when directly compared to the other two binders. Similar results were also

demonstrated by Sun et al. who showed the improved performance in conversion-type Fe2O3

and NiO based conversion anodes, both of which displayed enhanced cycling performance through the use of CMC binder.^[115]

Figure 11. a) Cycling data with a Sn based anode comparing different binder types, where the PFM binder in clearly superior. Reproduced with permission.^[114] Copyright 2014, Elsevier. b) Comparisons of different binder types with Co3O4 and NiO based anodes.

Goodenough et al. also reported the use of a cross-linked chitosan binder for use with a Sb based alloy.^[116] Instead of using CMC, the authors relied on chitin and its deacetylated derivative of chitosan, as the basis for the binder. Using a glutaraldehyde additive, they were able to cross link the linear carbohydrate polymer. Comparing the electrochemical

performance, the cross-linked binder showed a capacity retention of 96.5% at a current rate of 660 mA g^-1 over 100 cycles. Meanwhile the non-cross-linked binder only showed a retention of 81.3 % over the same number of cycles. Upon further investigation, it was concluded that

(a) (b)

the cross-linking led to a stronger mechanical integrity of the electrode during the volume changes, while the polymer material itself led to low polarization, thus enabling the formation of a stable SEI layer.

5.3. Quantitative Analysis

For the quantitative analysis regarding the binders, we focused on the relationship between the binder type, and the CE@1st, the same analysis, but also sectioned off by anode

chemistry, and lastly, the IC of the individual anode chemistries as a function of the binder type.

Figure 12. Stacked bar graphs grouped by a) CE@1^st with binder type as a factor. b) Binder type with CE@1^st as a factor.

From Figure 12, we can see good evidence that the selection of binder material matters for CE@1^st. Papers that utilized either CMC, or no binder material were the most likely to display CE@1^st greater than 0.75. By contrast, PVDF binders reported the lower percentage of CE@1^st of greater than 0.50. Thus, while PVDF may be the most commonly used anode, it appears that it is not the most useful in order to achieve a high CE@1^st.

(a) (b)

Figure 13. Stacked bar graphs grouped by binder type with CE@1^st as a factor for a) Carbon anodes. b) Alloying anodes. c) Conversion anodes. d) Insertion anodes.

Further breaking down the CE@1^st by binder type, and anode chemistry, we can gather a few additional observations. First off, for carbon materials, there is no clear evidence that the use of PVDF is better, or worse, than the use of CMC. However, self-standing carbon anodes do show a high percentage of CE@1^st in excess of 0.50. In the case of alloy and conversion materials, there seems to be a slight advantage for self-standing anodes. The biggest apparent difference for alloy and conversion anodes is the inferiority of PDVF compared to all the other types of binders. Lastly, the insertion type of anodes show a surprisingly better

performance with PAA, though such an observation could be a mirage, as the PAA data only comes from a limited number of papers. The other types of binders for insertion anodes did not show too many differences.

Figure 14. Capacity retention during long-term cycling versus IC as a function of binder material for a) Carbon anodes. b) Alloying anodes. c) Conversion anodes. d) Insertion anodes.

Analyzing the IC as a function of both anode chemistry, and binder type, we can make some tentative conclusions. In the case of the carbon anodes, the PVDF binder seems to work the best, though there is not as much data on other binders for a strong comparison. In the case of the alloy, conversion, and insertion anodes, the best binder for long-term cycling appears to be CMC. This is especially true with conversion and insertion anodes, where the PVDF is substantially worse. For the alloy anodes, there are a lot of examples of PVDF working well, though the densest cluster at the higher capacity retentions is still that of CMC. Self-standing anodes perform well for carbon materials, but do not show as clear of benefits for the three other types of anode chemistry.

When comparing these results with those from Figure 14, we can see somewhat of a pattern emerging. For carbon anodes, the binder with the best CE@1^st was the self-standing anode – those with no binder. The self-standing anodes also seem to have a slight advantage in

(a) (b)

(d) (c)

capacity retention. In the case of the alloy and conversion anode, the best binder for a high CE@1^st was the CMC, which also happens to be the best binder for the long-term cycling.

Curiously, while the self-standing binder was also beneficial for the CE@1^st in alloy and conversion anodes, it was not as useful for long-term cycling, most likely due to the increased structural stresses due to the large volume changes. Thus, solely looking at carbon, alloy, and conversion anodes, we can hypothesize that a binder favorable to a high CE@1^st is likely to favor more stable long-term cycling – except in the case of self-standing alloying and conversion anodes. Unfortunately, in the case of insertion anode, no relationships between binder type and long-term cycling and CE@1^st could be established. The CMC binder, which performs well for long-term cycling was not the best for CE@1^st, while there was not enough data to make a meaningful comparison regarding the PAA binder.

5.4. Additional considerations

Concluding this section on binders, we would like to stress that binder should be considered as more than something with which to hold the electrode particles on the current collector. In the case of amorphous carbon anode, the use of binders such as PVdF can cause

inefficiencies in the 1^st cycle, and lead to poor cycle life. Alternatives, such as CMC, polyacrylic acid (PAA) or sodium alginate should be considered. The world of polymer chemistry is rich with plenty of different polymers, some even conductive, which have yet to be tested as a binder for electrode materials. Additionally, the role of the binder material for alloying and conversion type anodes becomes even more important, as it directly plays in to the capacity of the anode material and can make all the difference between a failed anode and a functional one, even if the active material stays exactly the same.

6. CONCLUSIONS

In this review article, we hope to have demonstrated some of the additional complexities imbued in the development of NIB anode materials. While most of the community focuses directly on the materials, their structure-property relationships, and their electrochemical performances, a host of outside factors, such as electrolyte choice, SEI formation, and binder selection, can strongly impact the observed results.

By highlighting some of the literature, as well as performing our own quantitative analysis, we hope to leave the reader with a few conclusions. On the issue of electrolyte selection, we have shown that the selection of the solvent plays a greater role in the SEI formation/cycling inefficiencies than the selection of the salt. In short, the differences between NaClO4 and NaPF6 are negligible, but those between EC:DEC and EC:PC are less so. Additionally, our meta-analysis has suggested that the use of EC:DMC as an electrolyte solvent should be looked at more closely, as it can potentially lead to high efficiencies on the first cycle.

Additionally, we included a substantial amount of information regarding the use of co-intercalating, ether-based solvents, and ionic liquids, which could play a large role in the development of next generation electrolyte solvents for NIBs.

For SEI formation, the amount of literature available on Na-based SEI formation is much less than what is available in the LIB literature, which limits our understanding. Furthermore, we have also revealed that carbon-based, and insertion-based anodes, show considerably lower CE@1^st than alloy, and conversion based anodes. While this could be construed as evidence that the SEI formation chemistry is different in carbon, and insertion anodes, when compared to alloying, and conversion ones, it can also be taken as a sign that carbon and insertion anodes are more conducive to irreversibly trapping Na ions. This would have the same effect lowering the efficiency on the first cycle, while not forming SEI in the form of electrolyte breakdown products.

On the issue of FEC and SEI mitigation, we have shown that FEC helps little in increasing CE@1^st for alloy and conversion anodes, while making it worse for carbon and insertion anodes. However, the inclusion of FEC strongly helps long-term cycling retention, as

measured through our devised metric of ‘integrated capacity’ in the case of alloy, conversion, and insertion anodes. These two observations taken together support the common hypothesis that FEC induces greater SEI formation, thus lowering CE@1^st, but also forms a more stable SEI, which is beneficial for long-term cycling. Unfortunately, the inclusion of FEC as an electrolyte additive does little to improve the CE@1^st or long-term cycling performance of carbon based anodes.

Regarding binder materials, we included literature that shows some of the deleterious effects of poor binder selection. Binders, while seemingly an innocuous part of the electrode, can

In document Electrolytes, SEI Formation, and Binders: A Review of Non-Electrode Factors for Sodium-Ion Battery Anodes (Page 24-43)